Implicitness

Implicitness
From lexis to discourse
edi t ed by
Piotr Cap
Marta Dynel
John Benjamins Publishing Company

Implicitness
Pragmatics & Beyond New Series (P&bns)
issn 0922-842X
Pragmatics & Beyond New Series is a continuation of Pragmatics & Beyond and
its Companion Series. The New Series offers a selection of high quality work
covering the full richness of Pragmatics as an interdisciplinary field, within
language sciences.
For an overview of all books published in this series, please see
http://benjamins.com/catalog/pbns
Editor Associate Editor

Anita Fetzer Andreas H. Jucker
University of Augsburg University of Zurich
Founding Editors
Jacob L. Mey Herman Parret Jef Verschueren
University of Southern Belgian National Science Belgian National Science
Denmark Foundation, Universities of Foundation,
Louvain and Antwerp University of Antwerp
Editorial Board
Robyn Carston Kuniyoshi Kataoka Paul Osamu Takahara
University College London Aichi University Kobe City University of
Miriam A. Locher Foreign Studies
Thorstein Fretheim
University of Trondheim Universität Basel Sandra A. Thompson
Sophia S.A. Marmaridou University of California at
John C. Heritage
University of Athens Santa Barbara
University of California at Los
Angeles
Srikant Sarangi Teun A. van Dijk
Universitat Pompeu Fabra,
Susan C. Herring Aalborg University
Barcelona
Indiana University
Marina Sbisà
Masako K. Hiraga University of Trieste Chaoqun Xie
Fujian Normal University
St. Paul’s (Rikkyo) University
Deborah Schiffrin
Sachiko Ide Georgetown University Yunxia Zhu
The University of Queensland
Japan Women’s University
Volume 276
Implicitness. From lexis to discourse
Edited by Piotr Cap and Marta Dynel
Implicitness
From lexis to discourse
Edited by
Piotr Cap
Marta Dynel
University of Łódź

Amsterdam / Philadelphia
TM The paper used in this publication meets the minimum requirements of
8
the American National Standard for Information Sciences – Permanence

of Paper for Printed Library Materials, ansi z39.48-1984.
doi 10.1075/pbns.276
Cataloging-in-Publication Data available from Library of Congress:
lccn 2017013253 (print) / 2017027371 (e-book)
isbn 978 90 272 5681 2 (Hb)
isbn 978 90 272 6548 7 (e-book)
© 2017 – John Benjamins B.V.

No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any
other means, without written permission from the publisher.
John Benjamins Publishing Company · https://benjamins.com
Table of contents
chapter 1
Implicitness: Familiar terra incognita in pragmatics 1
Marta Dynel and Piotr Cap
Part I. Word and phrase
chapter 2
What’s a reading? 15
Mira Ariel
chapter 3
Pronouns and implicature 37
Wayne A. Davis
chapter 4
Implicitness in the lexis: Lexical narrowing and neo-Gricean pragmatics 67
Yan Huang
chapter 5
Zero subject anaphors and extralinguistically motivated subject
pro-drop in Hungarian language use 95
Enikő Németh T.
Part II. Sentence and utterance
chapter 6
Implicitness via overt untruthfulness: Grice on Quality-based figures
of speech 121
Marta Dynel
chapter 7
Lexical pragmatics and implicit communication 147
Deirdre Wilson and Patricia Kolaiti
chapter 8
Indirect ritual offence: A study on elusive impoliteness 177
Dániel Z. Kádár
vi Implicitness: From lexis to discourse
chapter 9
Implicitness in the use of situation-bound utterances 201
Istvan Kecskes
chapter 10
Thematic silence as a speech act 217
Dennis Kurzon
Part III. Text and discourse
chapter 11
The dynamics of discourse: Quantity meets quality 235
Anita Fetzer
chapter 12
Why don’t you tell it explicitly? Personal/subpersonal accounts
of implicitness 259
Marco Mazzone
chapter 13
Implicature and the inferential substrate 281
Michael Haugh
Index 305
Chapter 1
Implicitness
Familiar terra incognita in pragmatics
Marta Dynel and Piotr Cap

The aim of this chapter is to shed light on the concept behind the label “implicit-
ness”, which is ubiquitous in the pragmatic scholarship but has rarely constituted
the focus of attention per se. In the teeth of the intuitive use of the label, which
seems synonymous with “indirectness” used in some research traditions, an
organizing conception of implicitness phenomena is provided and advanced.
Implicitness is presented as encompassing both the use and the system side, and,
importantly, the cases where different-size phrasal, sentential and textual carriers
of indirect meaning can be identified. This view is reflected in the content of the
present volume.
Keywords: implicitness, indirectness, implicit meaning, research

in implicitness, levels of analysis
1. Explicating implicitness
Although implicitness is central to pragmatics and the term is ubiquitous in the

literature, it does not appear to enjoy a widely accepted definition. It is more of a
blanket term that researchers use intuitively with regard to a variety of phenomena
associated, broadly speaking, with the “hidden aspects of meaning” (Bertuccelli
Papi 2009: 140). These include, among others, different types of presuppositions,
entailments, implicatures and implicitures, explicatures, as well as a (growing) num-
ber of more specialized concepts denoting specific inferential schemata in different
languages (see Bertuccelli Papi 2000, 2009; Brandon 1998, 2015). The term “im-
plicitness” is rarely defined or explained, even though it is very frequently used; the
familiarity with it is presupposed and constitutes a pre-requisite to consider other
“more significant” and “more problematic” concepts or fields. For example, Östman
(1986) uses “implicitness” instrumentally to define the scope of the discipline of
pragmatics. The object of analysis and the defining task of pragmatics are, he says,
doi 10.1075/pbns.276.01dyn
© 2017 John Benjamins Publishing Company
2 Marta Dynel and Piotr Cap
to “explicate implicitness” (1986: 16). In this and a plethora of other claims and

definitions found in the pragmatic scholarship, “implicitness” is intuitively taken
to refer to inexplicit or nonliteral language use, a mismatch between the utterance
per se and the conveyed meaning.
In this vein, implicitness is taken for granted in countless nominal phrases
denoting areas of “focus-on” studies in specific forms of grammar and discourse.
There are, as contributions in this volume will confirm, “implicit arguments” in
semantic-pragmatic anaphora studies, “implicit topics” in turn-taking and con-
versation analysis, “implicit information” in the research on ellipsis, and, above all,
“implicit meanings” in neo-Gricean analyses of what is said/implicated, as well as in
post-Gricean analyses of the explicature vs. implicature distinction. “Implicit mean-
ings” transgress empirical boundaries and are used almost completely indiscrimi-
nately across the field of pragmatics, except perhaps in sub-fields such as speech act
theory and politeness studies, where the terms “indirect” and “indirectness” have
long been standardized (Haugh 2015). In (im)politeness research, the label “implic-
itness” is used with reluctance and largely as a “dispreferred” lexical substitute for
“indirectness”, even though the focal field of investigation is known to have been
partly inspired by Grice’s Cooperative Principle and his notion of “implicating” (see
Dynel 2013). For instance, what Haugh (2015) observes with regard to the concep-
tualizations of indirectness seems readily applicable to the conceptualizations of
implicitness: they all “involve some kind of inference on the part of the hearer that
either results in some kind of ‘mismatch’ between linguistic form/structure used by
the speaker and its (inferred) communicative function, or ‘mitigation’ of the force
of the communicated meaning (or sometimes both)” (2015: 21). The same kind of
correspondence exists in other politeness and politeness-related studies as well.
A notable exception is Kadar’s work in the present volume, where the concept of
implicitness is connected specifically and purposefully to the frequency with which
an utterance occurs (see the chapter-by-chapter description in Section 3).
In speech act theory, the parlance of Austin, Searle and their followers has ef-
fectively barred the notion of implicitness from practically all theoretical accounts
of language and language use, where the term “indirectness” has been the prime
choice and continues to reign supreme. Consequently, “implicitness” is treated as
a mostly extralinguistic concept, associated with the underlying patterns of world
knowledge and background assumptions. This tradition sanctions the use of such
terms as “implicit norms” or “implicit conditions”, with standard reference to the
extralinguistic contextual embedding of the utterance and the multiple (and often
vague in their own right) conditions for its successful, or “felicitous”, production.
One deviation from that trend is the use of “implicit” as a descriptor of acts reveal-
ing markedly extensive ranges of indirectness, that is producing a broad spectrum
of different-level interpretations along the explicature – implicature scale (Tanck
Chapter 1. Implicitness 3
2002; Kurzon 2007). This interesting deviation is again acknowledged in the present
volume, in Kurzon’s chapter. Drawing on the relevance-theoretic mindset, it does
not go any further than relevance theory in defining the boundaries separating the
graded “implicit” acts.
The lack of a solid conceptual handle on implicitness, both as a phenomenon
and as a label, may be a recognized fact, but apparently not a welcome one. There are
scholars, in the field of pragmatics and beyond, voicing their concern about possible
long-term methodological and empirical consequences. In a representative com-
mentary, Brandon (2015) observes that, as the number of domain-specific studies
involving implicitness is growing, the chances that the nature of implicitness as such
is investigated at general anthropological, psychological and psycholinguistic levels
become smaller and smaller. Indeed, each year sees published research, including
book-length work, devoted “explicitly” to particular manifestations of implicitness
in different utterance/text types and discourse domains. For example, the words
“implicitness” and “implicit” occur in the titles on information structure (Behrens
and Fabricius-Hansen 2009), indeterminacy and ellipsis (Shopen 1973; Sbisà 2007;
Bazzanella 2011), reference and discourse processing (Kirsner et al. 1998), cogni-
tive aspects of argumentation (Morency, Oswald and de Saussure 2008), as well as
organizational discourse (Hoogervorst, van der Flier and Koopman 2004), public
policy communication (van Mulken, van Enschot and Hoeken 2005) and discourse
of electronic genres (Xia, Huang, Duan and Whinston 2007), to name but a few.
While there is certainly no doubt about the contribution these works make to the
individual theories and empirical areas, it is quite unclear how they respond to the
many ontological and epistemological issues surrounding implicitness, including
the urgent question of personal, as well as sub-personal, reasons for implicit com-
munication (this question is taken up in Mazzone’s chapter in the present volume).
In a laudable (and unique) attempt to organize the discussion of implicitness
into a research agenda, Bertuccelli Papi (2000, 2009: 160) lists what she calls “the
hottest questions that the research on implicitness has to answer”. These include
questions about the types of inferences (used in the recovery of implicit meanings),
types of knowledge (underlying the inferences), levels and degrees of implicitness,
as well as limits of implicitness, that is “the possible extent to which we can omit say-
ing something without a communication failure” (Bertuccelli Papi 2009: 160). The
agenda defined by Bertuccelli Papi reflects the problematic issues encountered in
her account of several well-known concepts and sub-fields of semantics and prag-
matics, such as entailment, presupposition, implicature, and so on. Interestingly
enough, even though the account is itself comprehensive and thematically balanced,
the conclusions seem to be drawn from selected characterizations of the particular
areas and concepts. While Bertuccelli Papi is reluctant to acknowledge this discrep-
ancy, her agenda derives apparently more from issues of conventional implicature,
semantic presupposition and explicature, than from the “genuinely pragmatic” is-
sues of conversational implicature, impliciture and pragmatic presupposition. This
has to do, we believe, with the recurring question concerning lexical material that
lies at the core of any communication, and thus also any implicit communication.
It seems difficult to speculate about the levels and degrees of implicitness without
turning back, at one point or another, to the basic lexical properties of a phrase,
utterance, text, which are necessary and sufficient for that phrase, etc., to be the
carrier of implicit meaning in the first place. In that sense, a discussion of, for in-
stance, different presupposition triggers and their “potential for implicitness” may
be seen as more productive and more analytically sound (let alone more concrete)
than a similarly oriented discussion of particularized implicatures.
The stance of Bertuccelli Papi’s (2000, 2009) work is compatible with a much
earlier postulate by Östman (1986) that implicitness be studied with respect to the
interrelated points of reference – linguistic material and contextualization – with
the most promising results expected at the intersection of the two areas. Although
such a proposal presupposes a rather broad conception of implicitness, it simul-
taneously provides a solid conceptual platform on which to distinguish between
implicitness and indirectness. While indirectness involves, exclusively, the use
of language, implicitness extends over both the use and the system side, and
especially over the cases where different-size phrasal, sentential and textual car-
riers of indirect meaning can be identified in the lexico-grammatical arsenal of
a language. This is precisely the conception of implicitness to which we subscribe
in the present book. From an empirical standpoint, this step leads to two impor-
tant advantages. First, it encourages a global all-inclusive perspective on all acts of
implicit communication at different levels of linguistic organization. This seems a
much needed change since, traditionally, research in smaller-size units of commu-
nication, such as a word or phrase, has been reluctant to use the term “implicit-
ness”, saving it for more extensive, and thus more context-dependent forms (Bhat
2004; Davis 2005). As a result, implicitness phenomena observed at phraseological
and syntactic levels (such as, for instance, uses of pronominal forms) would often
receive inadequate attention or, at best, improper terms would be used to mark
them. The second advantage follows up on the first: the extended perspective al-
lows a much more comprehensive account of the strategic and rhetorical aspects of
implicitness (Ducrot 1972), involving both intentional and unintentional acts, as
well as (mostly) conscious and (mostly) automatic inferences. This is because one
can associate the particular levels of linguistic organization with different modes
of pragmatic inferences (Lepore and Stone 2015).
2. This collection: Aims and general structure
The theoretical aim of this volume is to respond to voices such as Bertuccelli Papi’s
(2000, 2009) and Brandon’s (2015), and to redress a certain methodological imbal-
ance that underlies analyses that work with the concept of implicitness. As has been
noted, research involving implicitness tends to benefit the empirical fields from
which data are collected, rather than the understanding of implicitness in global
terms. In contrast, our collection takes an integrated perspective and proposes a
research format that structures the individual studies of different implicitness phe-
nomena into a working conceptual framework which signposts well-demarcated
areas for further exploration. It adopts a unit-based approach, offering a panorama
of cutting-edge studies involving implicitness at different levels of linguistic forma-
tion. Specifically, the contributors consider the most problematic questions at three
distinct organization levels: “word and phrase” (Part One), “sentence and utterance”
(Part Two), and “text and discourse” (Part Three). As a result, the focal implicitness
phenomena and the main conceptual categories that define them (implicature, ex-
plicature, presupposition, implicature, etc.) undergo a continual reconsideration
and revision as the analytic scope widens. This approach, we hope, not only is lucid
and analytically friendly, but also adds explanatory power and critical insight to
some of the traditionally difficult issues central to the semantic-pragmatic interface.
For instance, the scrutiny of indexicals at the phrase level (Davis) and its partial
revision in the discussion of local context markers at the utterance level (Kecskes)
sheds light on how the typically “semantic” carriers of meaning (deixis, semantic
presupposition, etc.) behave in (sequences of) utterances and then in discourse,
and how they contribute to complex inferences.
A few comments and disclaimers are in order here. The view of implicitness
which we aim to promote in this collection is not identifiable within any individ-
ual contribution included in the volume. Rather, it is supposed to emerge from
the entire volume and the multiplicity of stances characterizing the different texts.
The problems undertaken in this volume have been tackled by authors coming
from very different theoretical camps – literalists vs. contextualists, minimalists
vs. relevance theorists, and so on. It would be unreasonable, if not naïve, to ex-
pect the research generated within these, often competing camps, to demonstrate
a simple consensus on what counts as “implicitness”; nor, in fact, would this kind
of aspiration be desirable. A more realistic ambition is, we believe, to define im-
plicitness by specifying the range of phenomena that fall under the umbrella
term “implicitness”, suggesting their general boundaries, and providing levels or
perspectives for their description. This is precisely why we speak of the use side
and the system side, and why we distinguish between the word, sentence and text
levels (even though these often intertwine when specific empirical or theoretical
research questions are addressed).
In addition to playing a theoretical role in elucidating the notion of implicit-
ness, taken together, the chapters serve a central methodological goal: they demon-
strate what background characteristics of different schools make it particularly
difficult for scholars to reach consensus on implicitness. For instance, Kecskes’s
text argues for the category of situation-bound utterances (SBU’s) to be associated
with implicitness, thus challenging a (more) popular position that implicit mean-
ing is derived on the basis of context-dependent inferential processing and does
not apply to conventionalized constructs such as SBUs. Differences of opinion
are also visible with regard to the processes/concepts of pragmatic enrichment
(Huang) and sentence implicature (Davis), the central query being the extent to
which they fall on the “implicit” side. While these “bones of contention” are iden-
tifiable based on the analytic methods followed in the different chapters, they also
emerge from the meta-comments by some of the authors (especially Huang, as
well as Wilson and Kolaiti).
Moving on to the empirical domain, our main aim is to test the implicitness
potential of the “semantic” vs. “pragmatic” properties of language and to capture the
relation between different levels of implicitness by determining what (types of) lin-
guistic expressions reside at which levels. As noted by Bertuccelli Papi (2009: 160),
there is “a gradient of explicitness which goes from silences, the extreme where the
speaker does not say anything but means a lot, through the half said, where explicit
hints are given of what the speaker means but does not want to say, to the other ex-
treme where the speaker says a lot but does not mean anything”. It is thus interesting
to determine how stable and recurrent are the speaker’s lexico-grammatical choices
at various levels of implicitness and generally, what amount of linguistic expres-
sion is associated with the particular level. This naturally invites further questions
concerning the limits of implicitness. We have acknowledged earlier Bertuccelli
Papi’s (2009) position which connects cases of communication failure to omissions
of linguistic material. However, for instance Kurzon’s work, including his chapter
in the present volume, puts a question mark over such a straightforward equation
(see also Kurzon 2007).
The general structure and theoretical characteristics of the present volume are
as follows. Part I deals with specific lexicogrammatical phrases and categories (in-
dexicals, demonstratives, anaphors, quantifiers, etc.) which trigger different implicit
meanings in different formal (syntactic) and functional (macro-contextual) embed-
dings. A variety of processes are described which engage these items in the pro-
duction of implicitness: the well-known processes of narrowing, enrichment and
anaphoric reference (in chapters by Huang, Ariel, and Németh T.), but also some less
explored phenomena, such as the input of deixis in creating implicatures (Davis).
The main point of Part I is to demonstrate that the phrase- and sentence-level lexical
material is not just a formal ingredient of communication but actually the necessary
condition for its well-formedness and interpretability. In fact, much of this mate-
rial contributes to complex implicit messaging, and thus to inferential processes
extending “upward” onto the utterance and discourse levels. Part II elaborates on
this point from the perspective of the utterance, adding important insights from the
study of figurative language (Dynel, as well as Wilson and Kolaiti), as well as ritual
and situation bound acts of communication (Kadar, and Kecskes). In investigating
the latter, Part II raises a necessarily related question, namely how the observed
recurrence of same/similar lexical form affects the distinction between implicitness
and indirectness. It is argued that due to its recurrent nature, ritualized implicitness
is not equal to indirectness, which invites revision of Speech Act Theory to give a
finer account of “explicit” / “direct” vs. “implicit” / “indirect” content and acts. This
revision is also necessary given the rather problematic speech act status of thematic
silence (Kurzon), that is a kind of silence that implicitly “codes” a specific topic. In
Part III, the most debated issues of implicitness are (re)addressed at the textual/
discourse level, thus forming a comprehensive catalogue of questions pervading
all the three levels, such as the size/length of information units, the complexity of
mental associative processes and the resulting caliber of inferences. Specifically,
Part III contains a discussion of social reasons and motivations for implicitness
(Haugh), set against cognitive conditions for production and reception of implicit
meanings (Mazzone), as well as relational properties of discourse conveying these
meanings (Fetzer).
3. Chapter-by-chapter overview
To make a collective contribution to the aims of the volume, the consecutive chap-
ters address issues of implicitness in diverse empirical areas, involving a variety of
data. The opening chapter by Mira Ariel problematizes the concept of a reading
associated with specific forms within semantics/pragmatics research. It contrasts
the speaker-intended messages with other legitimate inferences, which are, how-
ever, not speaker-intended. As is shown, there are, on the one hand, explicated
and implicated inferences, which are both speaker-intended. On the other, there
are unintended inferences such as Searle’s (1980) Background Assumptions and
Ariel’s (2004, 2016) Truth-compatible Inferences. Investigating forms such as and,
scalar quantifiers and or, Ariel shows that, contrary to traditional research, ana-
lysis of the unintended inferences as if they were inherent part of the reading is
methodologically erroneous. The next chapter, by Wayne A. Davis, discusses dif-
ferences among pronouns as related to differences in implicature. Similar to Ariel,
it critically assesses the mainstream research, concluding that its treatment of the
implicitness potential of pronouns needs much revision and elaboration. While it is
widely accepted that pronouns, and indexicals in general, have their stable deictic,
demonstrative and anaphoric uses, far less has been revealed with regard to their
pragmatic role in generating conventional and, more importantly, conversational,
implicatures. According to Davis, pronoun use creates a wide range of pragmat-
ic implicatures, through both non-semantic convention and specific contextual
factors. This, in turn, clearly extends the perception of utterance-level indexical
expressions as carriers of speaker-intended inferences performed by the hearer.
Next, Yan Huang’s chapter presents a neo-Gricean pragmatic analysis of lexical
narrowing, the phenomenon whereby the use of a lexical item implicitly conveys
a meaning that is more specific than the lexical expression’s lexically determined
meaning. The account is ontological in character and aims to establish which of
the well-recognized concepts of implicitness – explicature, pragmatically enriched
said, implicature – best reflects, or accommodates, the mechanism of enrichment
involving lexical narrowing. Discussing a host of approaches, from relevance the-
ory to Levinson’s three-principled model of pragmatics, Huang concludes that the
pragmatic enrichment involving lexical narrowing is essentially a neo-Gricean,
“pre-semantic” conversational implicature. It is, in other words, a conversational
implicature that intrudes into, or encroaches on, the semantics of the lexical items
(the weaker/general and the stronger/specific) involved. In the last chapter of Part
One, Enikő Németh T. analyses the pragmatics of implicit subject arguments in
Hungarian, concentrating on the zero subject anaphors and extralinguistically mo-
tivated subject pro-drop cases. Nemeth T.’s findings are twofold, casting light on
the lexical-grammatical-functional interface at phraseological and utterance levels.
First, it is empirically verifiable that the use or interpretation of zero subject an-
aphors and extralinguistically motivated subject pro-drop phenomena predicted by
grammar can be considered only a default use and interpretation that emerge due
to the lack of any pieces of information from the encyclopaedic knowledge, general
pragmatic knowledge, and/or specific context. Second, grammatical constraints and
pragmatic knowledge interact intensively in the licensing and recovering of zero
subject anaphors and extralinguistically motivated subject pro-drops phenomena,
i.e. kinds of implicit arguments with their own syntactic position in the structure
of utterances.
In its (partial) focus on utterance-level phenomena, Németh T.’s chapter paves
the way for texts included in Part Two of this volume. The first chapter there, by
Marta Dynel, addresses Grice’s views of Quality-based figures of speech (metaphor,
irony, hyperbole and meiosis) in the context of other aspects of his model of con-
versational logic in order to deliver an account of implicitness produced through
overt untruthfulness. Based on a careful overview of Grice’s original writings and
pertinent neo-Gricean scholarship, a reply is also provided to post-Gricean criti-

cism concerning the status of the category of Quality and the related figures. These
figures in their various forms and guises, which Grice’s framework is shown to be
able to capture, invariably display implicitness, according to Gricean thought. A
competitive account of non-literal metaphoric and hyperbolic uses of language
continues in Deirdre Wilson and Patricia Kolaiti’s chapter, which includes a lexical
pragmatic analysis of changes in the encoded linguistically-specified meaning of
expressions, following their use in the context of utterance. Engaging corpus tools
to study the ubiquitous processes of lexical narrowing, approximation and met-
aphoric extension, it arrives at two interrelated, empirical and theoretical points
that define the agenda of future implicitness research. As is demonstrated, there is
a continuum of cases between literal use, approximation, hyperbole and metaphor,
which favors search for a unitary account that applies across the whole continuum.
What is implicitly communicated is not always a specific cognitive content that the
speaker wants to convey but rather a whole range of cases with varying degrees
of specificity; thus, there ought to be an adequate theory of communication to
explain the entire range. The next chapter, by Dániel Z. Kádár shows how implic-
itness operates in impoliteness. The author explores the phenomenon of indirect
ritual offence, which includes cases of recurrent offences that are indirect and thus
make it difficult for the targeted person to respond. On top of its empirical merits,
Kadar’s analysis makes a contribution to the pragmatic examination of the differ-
ence between indirectness and implicitness. As the author argues, indirect ritual
offence earns language users’ metapragmatic “implicit” evaluations if the frequen-
cy of “indirect” attacks decreases but the attacks, nevertheless, continue. Relating
implicitness to the frequency of occurrence represents, apparently, one of the ways
in which to explain the ontology of implicitness. Another, more anthropologically
and culturally minded way is suggested in Istvan Kecskes’s text, which examines the
nature of implicit knowledge encoded in situation-bound utterances (SBUs). SBUs
are conventionalized, prefabricated pragmatic units whose occurrences are tied to
standardized communicative situations. This is because SBUs serve as interaction-
al patterns that mean the same to all speakers of a particular speech community.
The paradox of the use of SBUs, argues Kecskes, is that although most of them are
characterized by a high level of implicitness, still they may represent the most direct
way to express some social function. This conclusion not only adds to the number
of grounds on which to distinguish between implicitness and indirectness, but also
effectively predicates a scale of implicitness which is inversely proportional to the
scale of indirectness. The last chapter in Part Two, by Dennis Kurzon, discusses
the speech act status of thematic silence, i.e. a kind of silence that makes implicit,
or “codes”, specific topics. This chapter demonstrates that in most cases of thematic
silence, the non-mentioned implicit topic is still easily identifiable. This further
confirms conclusions from Kecskes’s analysis that degrees of implicitness are not
at all synonymous with degrees of indirectness. Kurzon’s chapter includes data
from political communication, allowing a smooth transition to broad-context and
multi-audience discursive issues depicted in Part Three.
Part Three consists of three chapters and begins with a theoretically oriented
essay by Anita Fetzer, who argues for a relational conception of discourse, involv-
ing clausal, as well as extra-clausal constituents. The former present proposition-
al meaning contributing to local and not-so-local argument and storyline, and
the latter encode procedural meaning in meta-discursive linkages, such as com-
ment clauses and discourse markers. On this dynamic view of discourse, implicit
meanings reside, and can be studied, in different and context-dependent patterns
of overtness for the linguistic realization of discourse units and the signalling of
discourse relations and other coherence strands. Among these potentially infinite
patterns, generically constrained forms constitute, naturally, the most promising
methodological and empirical terrain. In the next chapter, Marco Mazzone defines
the agenda for implicitness studies in integrated socio-psychological terms, pro-
posing that implicitness phenomena be approached at personal and sub-personal
levels. This involves exploring cases of implicitness in the light of communicative
predispositions anchored in the structure of human memory, as well as linguistic
behaviors reflecting human ability to construct contextually shared chains of goals.
As is demonstrated, the two research paths are not incompatible; they are presumed
to work together to explain cooperation between automatic and controlled associ-
ative processes, and thus to explain reasons for implicitness in general. Finally, in
the last chapter of the collection, Michael Haugh brings together cognitive, social
and cultural issues of conversational and discoursal practice, making a proposal
for a new model of implicature. The pre-proposed model can be described as in-
teractional, based on the understanding that implicatures are not just cognitive
constructs, but essentially social actions which should be analyzed with respect to
the broader inferential substrate from which they arise. The main argument is that
while the inferables that make up this inferential substrate generally remain em-
bedded, they may, on occasion, be exposed by interaction participants as the focus
of their exchange. As is shown in a number of natural language examples, there
exists a whole range of practices that license participants to expose an inferable or
set of inferables, including instances of “prompting”, where the speaker positions
another participant to make a pre-emptive offer through reporting possible troubles
or needs. This again echoes the points made by Kurzon, Kecskes and others, that
implicitness is conceptually distinct from indirectness, and that sometimes it is
the very “implicit” premise or chunk of knowledge that lies at the core of the most
“direct” communicative act and successful communication.
Altogether, the texts in this volume should be taken as a collective voice of ap-
preciation of the analytic richness of implicitness (Bazzanella 2011), which reflects in
the apparent multiplicity of theoretical handles on its variegated facets. The different
“favorite” concepts on which the chapters are hinged, from single-word quantifiers
to discursively constructed implicatures, can in turn be taken as indicators of the
most interesting and productive venues for further research. Implicitness is often
deployed in everyday language use according to the saying “the less you say the
more you say”, which echoes Levinson’s (1987: 68) Maxim of Minimization: “say as
little as necessary”. While, as has been noted, it would be difficult to propose any
obvious lines of agreement between the particular stances, the texts in the pres-
ent volume do heed a common methodological message: to draw from maximally
indeterminate forms in order to accomplish maximally robust and consequential
conclusions. As such, they prescribe a bottom-up, data-driven agenda in future
implicitness studies.
References
Ariel, Mira. 2004. “Most.” Language. 80: 658−706. doi: 10.1353/lan.2004.0162

Ariel, Mira. 2016. “Revisiting the Typology of Pragmatic Interpretations.” Intercultural Pragmatics
13: 1–35. doi: 10.1515/ip-2016-0001
Bazzanella, Carla. 2011. “Indeterminacy in Dialogue.” Language and Dialogue 1: 21−43.
doi: 10.1075/ld.1.1.04baz
Behrens, Bergljot, and Cathrine Fabricius-Hansen. 2009. “Introduction to Contributions in
‘Structuring Information in Discourse: The Explicit/Implicit Dimension’.” Oslo Studies in
Language 1: 2−16.
Bertuccelli Papi, Marcella. 2000. Implicitness in Text and Discourse. Pisa: Edizoni ETS.
Bertuccelli Papi, Marcella. 2009. “Implicitness.” In Handbook of Pragmatics Highlights vol. I: Key
Notions in Pragmatics, ed. by Jef Verschueren and Jan-Ola Östman, 139−162. Amsterdam:
John Benjamins. doi: 10.1075/hoph.1
Bhat, Darbhe Narayana Shankara. 2004. Pronouns. Oxford: Oxford University Press.
Brandon, Robert. 1998. Making It Explicit: Reasoning, Representing, and Discursive Commitment.
Cambridge, MA: Harvard University Press.
Brandon, Robert. 2015. From Empiricism to Expressivism. Cambridge, MA: Harvard University
Press.
Davis, Wayne A. 2005. Nondescriptive Meaning and Reference. Oxford: Oxford University Press.
doi: 10.1093/0199261652.001.0001
Ducrot, Oswald. 1972. Dire et ne pas dire. Principes de sémantique linguistiques. Paris: Hermann.
Dynel, Marta. 2013. “Being Cooperatively Impolite: Grice’s Model in the Context of (Im)Politeness
Theories.” In Research Trends in Intercultural Pragmatics, ed. by Istvan Kecskes and Jesus
Romero-Trillo, 55−83. Berlin: Mouton de Gruyter. doi: 10.1515/9781614513735.55
Haugh, Michael. 2015. Im/Politeness Implicatures. Berlin: Mouton de Gruyter.
doi: 10.1515/9783110240078
Hoogervorst, Jan, Henk van der Flier, and Paul Koopman. 2004. “Implicit Communication in
Organisations: The Impact of Culture, Structure and Management Practices on Employee
Behaviour.” Journal of Managerial Psychology 19: 288−311. doi: 10.1108/02683940410527766
Kirsner, Kim, Craig Speelman, Murray Maybery, Angela O’Brien-Malone, Mike Anderson, and
Colin MacLeod (eds). 1998. Implicit and Explicit Mental Processes. Mahwah, NJ: Lawrence
Erlbaum.
Kurzon, Dennis. 2007. “Towards a Typology of Silence.” Journal of Pragmatics 39: 1673–1688.
doi: 10.1016/j.pragma.2007.07.003
Lepore, Ernie, and Matthew Stone. 2015. Imagination and Convention: Distinguishing Grammar
and Inference in Language. Oxford: Oxford University Press.
Levinson, Stephen C. 1987. “Minimization and Conversational Inference.” In The Pragmatic
Perspective: Selected Papers from the 1985 International Pragmatics Conference, ed. by
Marcella Bertuccelli Papi and Jef Verschueren, 61−129. Amsterdam: John Benjamins.
doi: 10.1075/pbcs.5.10lev
Morency, Patrick, Steve Oswald, and Louis de Saussure. 2008. “Explicitness, Implicitness and
Commitment Attribution: A Cognitive Pragmatic Approach.” Belgian Journal of Linguistics
22: 197–219. doi: 10.1075/bjl.22.10mor
Östman, Jan-Ola. 1986. Pragmatics as Implicitness. PhD dissertation, University of Michigan.
Sbisà, Marina. 2007. Detto non detto. Le forme della comunicazione implicita. Roma-Bari: Laterza.
Searle, John R. 1980. “The Background of Meaning.” In Speech Act Theory and Pragmatics, ed. by
John RJohn R. Searle, Ferenc Kiefer, and Manfred Bierwisch, 11–41. Boston: Reidel.
doi: 10.1007/978-94-009-8964-1
Shopen, Tim. 1973. “Ellipsis as Grammatical Indeterminacy.” Foundations of Language 10: 65−77.
Tanck, Sharyl. 2002. “Speech Acts Sets of Refusal and Complaints: A Comparison of Native and
Non-native English Speakers’ Production.” TESOL Working Papers (http://auislandora.wrlc.
org/islandora/object/tesolworkingpapers%3A26; accessed September 1, 2015).
Van Mulken, Margot, Renske Van Enschot, and Hans Hoeken. 2005. “Levels of Implicitness
in Magazine Advertisements.” Document Design 13: 155−164. doi: 10.1075/idjdd.13.2.09mul
Xia, Mu, Yun Huang, Wenjing Duan, and Andrew Whinston. 2007. “Implicit Many-to-One
Communication in Online Communities.” Communities and Technologies 27: 265−274.
Part I
Word and phrase

Chapter 2
What’s a reading?
Mira Ariel
Tel Aviv University
Linguists sometimes assume that the readings associated with linguistic ut-
terance, both explicit and implicit, are self-evident. I here problematize the
concept of a reading associated with linguistic expressions, and restrict it to
interpretations systematically intended by the speaker using the utterance.
This is a stronger condition than is sometimes adopted, namely, meshing the
utterance with the objective reality that must lie behind the utterance (accord-
ing to the speaker). I reanalyze assumptions commonly considered part of
the reading associated with and, or and scalar quantifiers as Background or as
Truth-Compatible inferences. On this account, these assumptions do not fall
under the speaker’s communicative intention, and therefore do not constitute
readings.
Keywords: what is said, Truth-Compatible inferences, Background assumptions,

scalar quantifiers, explicature
1. Introduction
The goal of this chapter is to problematize the concept of a “reading” associated

with specific forms within semantics/pragmatics research. As is now commonly as-
sumed, the Gricean “What is said” representation, as well as the Relevance-theoretic
explicature, involve more than just the bare linguistic meanings of those utter-
ances (see the overviews in Huang (this volume) and Dynel (this volume)). More
often than not, these representations must combine the compositional linguistic
meanings with pragmatically derived inferences, and it is the combined whole that
counts as the speaker’s directly communicated message. While there are significant
differences between “what is said” representations and explicatures (explicatures
incorporate many more pragmatic inferences), neither include Particularized
Conversational Implicatures. Clearly, not all implicit meanings are equal. Inferences
doi 10.1075/pbns.276.02ari
16 Mira Ariel
that contribute to the development of a full proposition equal more. 1 Considerable

work (and debates) deal with how to best distinguish implicatures from “said” or
explicated inferences (inferences that form part of “what is said” and explicatures)
(Ariel 2008: Chapter 7; Carston 2002; Davis, this volume; Huang, this volume).
But the distinction I focus on in this chapter is a different one, one not usually
problematized in the literature. It contrasts (all) speaker-intended messages with
other, legitimate enough inferences, which are, however, not speaker-intended. On
this division, both explicated and implicated inferences fall on the same, speaker-
intended side of the fence. Two other types of inferences, Searle’s (1980) Background
assumptions and my own Truth-Compatible inferences (Ariel 2004, 2016) (see be-
low) are not speaker-intended, and hence fall on the other side of the fence. Now,
seemingly, this distinction has already been drawn. Grice’s ‘what is conveyed’, which
adds the implicated onto the “said” (and let’s assume we can extend this to the
more “liberal” explicature), supposedly represents all and only speaker-intended
meanings. Obviously, any other inferences fall outside of “what is conveyed”, and
are therefore of no interest to the linguist.
My point is, however, that unintended inferences have sometimes been errone-
ously analyzed as if they were part of the reading (the relevant “what is conveyed”).
Surprisingly, the reanalysis I’m proposing pertains not only to assumptions taken
as pragmatic inferences, whether implicated or explicated, but also to components
assumed to be linguistic meanings. Consider lexical items such as scalar quantifiers
some and most and constructions such as X and Y and X or Y. Are the readings
routinely associated with these expressions, as analyzed by linguists, restricted to
their explicatures and implicatures? What about Background assumptions crucial
for these interpretations (Searle 1980)? What about what I have called (Ariel 2004,
2016) Truth-Compatible inferences, inspired by Koenig (1991)? Truth-Compatible
inferences (TCIs) are assumptions about how reality must be, or most likely is ac-
cording to the speaker, given her utterance.
There isn’t really any defense in the literature of the assumed “reading” status
of lower or upper-bounded interpretations for the scalar quantifiers (some = “some
and possibly all” and some = “some but not all”), nor of enriched and conjunctions
(where e.g., and = “and then” or “and as a result” etc.), nor for inclusive and exclu-
sive interpretations as constituting or readings. Linguists simply assume they know
what the “appropriately enriched” level is that constitutes a reading for these ex-
pressions. Consider the following quote from Chierchia (2013: 15) about the read-
ings associated with or constructions: “’Interpretation’ is used here rather loosely
as referring to the basic compositional meaning enriched by whatever (possibly
1. The research here reported was funded by the Israel Science Foundation (grant 431/15).
Chapter 2. What’s a reading? 17
pragmatic) inference/ implicature may be tacked onto that’” (emphasis added). So,
the enriched, but still “basic”-level interpretation is not usually problematized. It
seems obvious that when analyzing a specific linguistic form (be it lexical or con-
structional), we need to offer not only a bare linguistic layer, but also (where ap-
propriate) an integrated semantic + pragmatic layer. But how is the reading to be
delimited? Which inferences should be incorporated into it?
Readings, as here used, are defined as recurrent, speaker-intended interpre-
tations consistently associated with specific linguistic forms. Special uses made
with such expressions are thus excluded. I propose that once we apply a battery of
“Faithful Report” tests proposed in Ariel (2016), inspired by Bach’s (1994) IQ test,
as well as a new test here proposed, based on the function of X, including Y and X,
Y excluded, we will be in for some surprises regarding what is a speaker-intended
reading for scalar quantifiers, for X and Y and for X or Y constructions. The most
important finding is that what are considered readings in the linguistic literature
often comprise of Background assumptions or mere Truth-Compatible inferences
about how the reality behind the utterance must be. Crucially, these inferences
are not at all speaker-intended, and hence should not be analyzed as part of the
expressions’ readings.
2. Tests identifying interpretation status
2.1 (Un)faithful report tests
Ariel (2016) revisits the typology of pragmatic inferences, and proposes to distin-
guish between different inferences, using a battery of tests, each aimed at identifying
a specific type of inference. The idea is that a different speaker reporting the orig-
inal speaker’s utterance must represent different pragmatic inferences differently.
Inferences that constitute part of the explicature (the truth-bearing proposition
intended by the speaker) pass the That is (to say) test, and conversational implica-
tures pass the Indirect Addition test (see below). Other pragmatic inferences are
speaker-intended, but their derivation is only ad hoc. Ironical interpretations are
a case in point. While part and parcel of the speaker’s intended reading on specific
occasions, unless conventionalized, these inferences are not recurrent readings of
specific linguistic expressions, because the association between the form and the
special, nonliteral reading is more often than not quite ad hoc. Those don’t count
as constituents of the reading associated with specific forms, then. Yet other prag-
matic inferences, while consistent with the speaker’s utterance, and plausibly even
subscribed to by her, do not fall under her communicative intention. These are
18 Mira Ariel
Searle’s (1980) Background assumptions and what I have called Truth-Compatible

inferences, which should not be analyzed as part neither of the ad hoc reading nor
the recurrent one. Readings must be restricted to speaker-intended interpretations.
In order to argue my point, I resort to Faithful Report tests, the results of which will
show that we need to reconsider well-established assumptions about the readings
associated with and, scalar quantifiers and or. Section 2 briefly introduces the rel-
evant tests, applying them to ad hoc interpretations. I apply the tests to and, scalar
quantifiers and or in Sections 3–5.
Consider the status of the pragmatic inferences involved in interpreting the
various somethings in the following examples:
(1) a. Let’s go down.
It’s not SOMETHING,
but it’s SOMETHING (Originally Hebrew, Feb. 10, 2008)
b. If you see something say something
(Sign at New York Subway station, spotted 10.07.2007)
c. If you see something suspicious or unusual, say something! Tell your
conductor or call Amtrak (Pamphlet on a train, spotted 6.02.2016)
As is, (1a) is contradictory, while (1b) is vague to the point of being uninforma-
tive. But this is not how we actually interpret these examples. (1a), addressed to
the speaker’s dog, was meant to convey something like “this going down is not
something wonderful, but it’s better than nothing, so it’s worth the dog’s while”.
(1b) is interpreted along the lines of (1c), where the first something is interpreted
as “something suspicious or unusual”. The second something in (1b) and (1c) is
interpreted as “say something about the suspicious thing”.
The contextual adjustments associated with the somethings above are certainly
speaker-intended, for the message directly expressed by the speaker relies on the
inferences involved. And what about the inference in (1a) that it’s worth it to the
dog to go down with the speaker? Again, this is a speaker-intended message, the
speaker trying to encourage the dog to come down with her, but it is not a directly
communicated assumption, but rather, a Particularized conversational implicature
(PCI). In Ariel (2016) I have proposed two tests aimed at helping researchers in
identifying speaker-intended inferences, and distinguishing between explicated in-
ferences (inferences that constitute part of the explicature) and PCIs. The idea is that
(i) Speaker-intended inferences can be explicitly cited by another speaker, as part
of a faithful report of the speaker’s utterance; (ii) But, at the same time, explicated
and implicated inferences can be so cited under quite different framings, ones that
reflect the difference in their statuses.
Consider explicated inferences first. Here the faithful reporter can explicate the
inferences, prefacing them by that is (to say). Indeed, (2) is a faithful report of (1a): 2
(2) S said that it’s not something, but it’s something, that is (to say) that this going
down is not something wonderful, but it’s better than nothing.
And (3) is a faithful report of (1b):
(3) S said that if you see something say something, that is (to say) that if you see
something suspicious say something about it.
What about the inference that “it’s worth the dog’s while” in (1a)? Like explicated
inferences, implicated inferences are speaker-intended. But they do not count as
directly communicated, as explicated inferences are. Hence, they are identified by
a different Faithful Report test: The Indirect Addition test. The failure of the That is
(to say) test in (4a) shows that “it’s worth the dog’s while” is not explicated in (1a).
The success of the Indirect Addition test in (b) shows that it is indeed implicated:
(4) a. ??S said that it’s not something but it’s something, that is (to say) that it’s
worth the dog’s while to go down.
b. S said that it’s not something but it’s something, and in addition, she
indirectly conveyed that it’s worth the dog’s while to go down.
Thus, because they are speaker-intended messages, both explicated and implicated
inferences can occur as explicit components in a faithful report by another speaker
(although under different framings).
Special, ad hoc interpretations, such as ironies, require yet another test.
Although speaker-intended just like regular PCIs are, such second-tier interpreta-
tions come to replace the original explicatures. The proper test for these cases is the
Replacement test, where the implicated inference replaces the original explicature
and is not simply an added inference, as other PCIs are. Consider:
(5) MARILYN: You threw a green pepper down my shirt.
ROY: … (SNIFF) … I thought it was funny.
MARILYN: … Hilarious. (SBC: 003)
Funny and hilarious are quite similar in meaning, but of course, Marilyn is actually
disagreeing with Roy’s assessment that throwing a green pepper down her shirt
is funny. Her hilarious is ironical. Note how another speaker can (6c) and cannot
(6a, b) faithfully report what Marilyn said:
2. S stands for the original speaker hereon.

20 Mira Ariel
(6) a. ??Marilyn said that it was hilarious, that is (to say) not funny at all.
b. ??Marilyn said that it was hilarious, and in addition, she indirectly conveyed
that it was not funny at all.
c. Marilyn said that it was hilarious, but actually, she indirectly conveyed that
it was not funny at all.
So far, we’ve seen three types of pragmatic inferences, all speaker-intended, and
hence reportable as such. But other potential inferences cannot be mentioned ex-
plicitly when faithfully reporting another’s utterance, because they are not part
of the speaker’s communicated intention, even if she actually subscribes to them.
Consider let’s go down in (1a). Surely, the speaker is suggesting that she and the
dog go downstairs in their routine way, using the elevator. She is not proposing that
the two of them use the stairs, nor certainly, climb down a rope from the window.
In addition, the going down will enable the dog to “go to the bathroom”, which is
a positive event for the dog, and these two inferences serve as necessary assump-
tions (implicated premises for Relevance theoreticians) for deriving the PCI (im-
plicated conclusion for Relevance theory) that it’s worth it for the dog to go down.
Background assumptions (Searle 1980) are defined as implicit aspects derived from
our world knowledge about how things must be for the states of affairs depicted
by the speaker’s utterance. Although the speaker is most likely committed to these
assumptions she need not and does not assume responsibility for communicating
them. According to Carston (2002: 3.7), Background assumptions are meant, but
they are not conveyed (see also Bach 2006). I have incorporated implicated premises
into Searle’s Background assumptions (Ariel 2016).
Indeed, any attempt to report the speaker’s utterance in this case by explicating
the Background assumptions is deemed unfaithful: 3
(7) a. ??S suggested to go down, that is (to say) to go down by the elevator.
b. ??S suggested to go down, and in addition, she indirectly conveyed that they
go down by the elevator.
c. ??S suggested to go down, but actually, she indirectly conveyed that they go
down by the elevator.
(8) a. ??S suggested to go down, that is (to say) to “go to the bathroom”.
b. ??S suggested to go down, and in addition, she indirectly conveyed that the
dog will go to the bathroom.
c. ??S suggested to go down, but actually, she indirectly conveyed that the dog
will go to the bathroom.
3. Note that a faithful report of nonassertions must include a speech act verb.
As we can see, none of the paraphrases offered in (7) and (8) are faithful reports of
the original utterance. Indeed, Background assumptions do not constitute speak-
er-intended messages, and hence, should not be included as part of the reading of
linguistic expressions.
There is another type of pragmatic inference potentially subscribed to by the
speaker, but again, not one she intends to in fact communicate. Truth Compatible
inferences are inferences (pragmatic ones, as well as entailments) which may very
well be true of the reality behind the speaker’s utterance; They may correspond to
assumptions the speaker herself entertains, but crucially, they do not fall under
her communicative intentions. The first to draw attention to such a phenomenon
was Koenig (1991) with respect to the meaning of numerals, and I have developed
the concept, as well as its implications in Ariel (2004 and onwards). Going back to
(1a) it’s absolutely compatible with what the speaker is suggesting to her dog that
the offer is to go down to the nearby dog park, which they often enough do. In fact,
it is the location the speaker had in mind for this going down. But crucially, let’s
go down is only compatible with this inference. No specific location is included in
the speaker’s intended message, although going down perforce involves a target
location. The point is that speakers may remain silent about aspects of the relevant
reality they discuss, even if they’re fully committed to their truth. “Going to the
dog park” is then merely a Truth-Compatible inference of “let’s go down” in (1a).
Consider similarly Phil’s if you have any questions… in (9):
(9) PHIL: I love to share that with everyone that I meet,
(H) if you have any questions about this,
or = … any other subject,
come on up and ask. (SBC: 027)
Phil has just he delivered a lecture at a science museum. It’s pretty obvious that if the
audience has not only questions, but also comments, additional support or criticism
of his lecture points they’re invited to come up and talk to him about these, despite
the fact that comments, supporting evidence and criticism don’t count as questions.
Indeed, Phil’s utterance is compatible with the audience coming up to him to ask
him questions and/or to offer comments, support or criticism of his lecture. But
again, although he may subscribe to such options, Phil’s utterance does not convey
them, neither directly nor indirectly: 4
(10) a. ??Phil said that if you have any questions about this you should come up and
ask, that is (to say) if you have any questions or comments or additional
support or criticism, you should come up and ask.
4. I here simplify the original utterance to avoid too cumbersome formulations.

22 Mira Ariel
b. ??Phil said that if you have any questions about this you should come up and
ask, and in addition, he indirectly conveyed that if you have comments,
additional support or criticism you should come up and ask.
c. ??Phil said that if you have any questions about this you should come up and
ask, but actually, he indirectly conveyed that if you have any questions
or comments, or additional support or criticism you should come up and
ask.
Discussing one’s comments, additional support or criticism of a lecture with Phil are
compatible with asking her “any questions”. This is why a speaker approaching Phil
with criticism, for example, will not be ignored by him because he did not invite the
audience to criticize him. Still, critical comments are not speaker-intended referents
in messages such as (9). The Faithful Report tests reflect this. In Sections 3–5 I will
apply these tests and argue that some interpretation currently analyzed as (part
of) the readings associated with and, or and the scalar quantifiers fail these tests,
because they are not in fact speaker-intended.
2.2 The including/excluded test
Section 2.1 offered tests to help distinguish between speaker-intended inferences

(explicated inferences, PCIs, and second-tier implicatures) and inferences which
are not speaker-intended (Background assumptions and Truth-Compatible infer-
ences). But of course, readings first and foremost pertain to linguistically encoded
meanings. Section 2.2 introduces two variants of a test that can help us identi-
fy (a subset of) linguistic meanings. The test allows us to check the viability of
possible instantiations of the purely linguistic meaning assumed by researchers.
Legitimate instantiations should pass the test, illegitimate ones should fail it. The
test thus complements the previous tests in that interpretations they identify as
Truth-Compatible inferences should also fail the Including/excluded test, since
TCIs are not explicated.
Consider the following examples, which contain an [X, includingstressed Y] string:
(11) a. JIM: … an=d,
… to me that,
… that always symbolized,
kind of rather clearly,
(H) the limitations of the scientific method.
(H) … Whi---
if it doesn’t take into account,
MICHAEL: … well,
JIM: … the … to---

… the the the whole human experience,
(H) and,
… including,
(H) … the … unbelievable parts of it.
… the parts that are just legend.
and myth. (SBC: 017)
b. Well she got committed or something and then your dad took care of all
you guys including the baby.(LSAC)
c. A: Every street in the US, including
B: [Yeah]
A: like gravel [roads]
C: [Dirt roads,] [yeah].(LSAC)
The construction under discussion consists of a category X (indicated by italics in

11), which is said to include a certain instantiation or instantiations of it, Y (indicat-
ed in bold in 11). 5 Indeed, according to Jim (a), the category of “the whole human
experience” includes as instantiations “the unbelievable parts of it”, “the parts that
are just legend”, and “myth”. Similarly, “all you guys” includes “the baby” (b), and
“every street” includes “gravel roads” and “dirt roads”. Still, not any instantiation
of the relevant category can appropriately occur in this construction. Contrast the
following pair:
(12) a. ??Weapons are not allowed beyond this point, including Uzis.
b. √Weapons are not allowed beyond this point, including replica guns/tear gas.
Uzis are no doubt instantiations of the category of weapons. The problem is that
they are “too good” exemplars of the category. While semantically reasonable, (12a)
is pragmatically anomalous, because there is no reason to specify that Uzis are not
allowed, given that weapons are not allowed. The only reason speakers have for
using this construction is if the Y instantiations are either marginal or questionable
members of the category X, as in (12b) (as well as 11a and c), or, even if not margin-
al, at least noteworthy, as in (11b). So, all in all, X, including[stressed]Y is used when
Y is a member of category X, which is worthy of singling out for explicit mention,
mostly because it is a marginal or nonsalient or unexpected member of the category.
Here’s a case where Y is actually a prototypical member of X, but since it’s not here
expected the string is acceptable:
5. Note that including must be stressible. When including is not stressed the construction is less
constrained.
24 Mira Ariel
(13) ?: This is a concert in Miami.

?: That’s Jonesy.
?: Is it Jonesy?
?: Looks like him.
?: The guy who’s translating is Peruvian.
?: He is? ((1 LINE OMITTED))
?: It sounds like Francisco, Francisco.
?: This is on all channels including regular channels, too, huh? (LSAC)
The last speaker is surprised that regular channels broadcast that kind of concert.
The crucial point for this chapter is that Y must be a member of the X category,
which means that we can use this construction to test which instantiations count
as members of some category and which don’t. Note that (14) is unacceptable,
even though, just like weapons, water bottles are not allowed beyond the security
check point at airports, because they are not (marginal) members of the category
of weapons:
(14) ??Weapons are not allowed beyond this point, including bottles of water.
So, the construction is useful for drawing the boundaries around categories.
A variant of the X, including Y construction can be similarly used for the same
purpose. The X, Y excluded construction is used to exclude Y from the X category
being predicated on, although Y is a bona fide member of X. Here Y needn’t be a
noteworthy member. Consider the following:
(15) Think of the lawyers we all know and can’t stand – present company excluded,
of course. (Massachusetts Lawyers Weekly, June 1996)
Sections 3–5 apply the Faithful Report and the Including/Excluded Y tests, and
challenge the reading status currently assumed for certain interpretations with re-
spect to specific linguistic expression. Specifically: (i) The coherence inferences
associated with and do often enough constitute (explicated) readings, but at other
times, I argue, they are mere Truth-Compatible inferences. Hence, once a reading
does not mean always a reading (Section 3). (ii) The compatibility of the scalar
quantifiers with “all”, currently assumed to be lexically specified (and hence, part of
the reading) should be reanalyzed as a Truth-Compatible inference too (Section 4).
Finally, (iii) the “not both” interpretation associated with what is considered an
exclusive reading of or should mostly count as a Background assumption. If I’m
correct, then not only pragmatic inferences (explicated and implicated inferences)
have been erroneously analyzed as inherent aspects of the reading associated with
linguistic expressions. The same is true even for linguistic meanings.
3. And-coherence inferences
Sentences conjoined by and often give rise to coherence inferences pertaining to

temporal, causal etc. relations between the states of affairs depicted. While orig-
inally analyzed as Generalized Conversational Implicatures by (neo-) Griceans,
Relevance theoreticians (Carston 1988; Wilson and Sperber 1998) have forcefully
argued for an explicated status for e.g., the causal inference in (16) (see also Ariel
2012; Carston 2002). Whether implicated or explicated, this causal inference cer-
tainly counts as part of the reading of (16):
(16) KEN: (H) So I eat the local food,
and get deathly ill. (SBC: 015)
Indeed, the causal connection between Ken’s eating the food and getting ill contrib-
utes to the truth-conditional content of the proposition expressed, because it is felt
to have been directly communicated. This is why Ken’s interlocutor can coherently
disagree with him, by saying: 6
(17) ~Joanne: That’s not true. The food’s got nothing to do with your getting ill.
Since the causal connection is taken to be part and parcel of Ken’s assertion, falsi-
fying it is a legitimate basis for declaring the conjunction as a whole as false.
Note the following writer, who explicitly says that a similar and conjunction
indeed directly communicates a causal connection:
(18) … Maariv came out with the banner headline “The literature teacher was fired
and committed suicide”… A direct causal connection.
(Originally Hebrew, Haaretz 5.30.2003)
I’ve defended this relevance-theoretic position in Ariel (2012). What I can here
add in support of the explicature analysis is results of tests designed to identify the
explicated/”said”, which distinguish it from assumptions that do not constitute part
of the speaker’s intended reading. These were introduced in Ariel (2016) and briefly
described in Section 2.1 above. To recapitulate, the tests distinguish between faithful
and unfaithful reports potentially made by someone having heard the speaker’s ut-
terance in its context. The idea is that while some pragmatic inferences count as part
of the reading intended by the speaker others do not. Hence, faithfully reporting
the speaker’s utterance may incorporate some pragmatic inferences, but not others.
6. ~ indicates an unattested example.

26 Mira Ariel
Note that the report in (19) is faithful, thus corroborating the explicature anal-
ysis of and conjunctions, whereby the causal connection (in this case) counts as
part of the reading:
(19) That is (to say) test: Ken said that he eats the local food and gets deathly ill,
that is (to say) that he eats the local food and as a result he gets deathly ill.
However, as I have argued in Ariel (2012), inferences associated with some construc-
tion need not necessarily carry the same status on different occurrences. Indeed the
coherence inferences associated with and are not invariably speaker-intended, and
are therefore not invariably part of the reading intended by the speaker. They may
be merely Truth-Compatible inferences.
Consider (20), where the conjunction introduces two events which are tempo-
rally ordered, such that the watching of the Akaba conference (on TV) preceded
the knocking on the door: 7
(20) We watched the Akaba conference ((talks about an Israeli-Palestinian peace
agreement)), we hoped for the best, and yesterday they knocked on our door
((to say that our son was killed by a Palestinian terrorist – MA)).
Although absolutely true about the reality depicted in (20), my claim is that the tem-
poral ordering between the two events here (first “hoping…” then “knocking…”) is
not a speaker-communicated assumption, and should not count as part of the read-
ing. Instead, what the speaker wishes to communicate is the contrast between the
optimistic feelings they had about peace and the killing of their son by a Palestinian.
This and only this is the relevant coherence inference which forms part of the reading.
Indeed, a faithful report of (20) is the following, where the coherence relation
is explicated with “but”:
(21) That is (to say) test: S said that… they hoped for the best, and yesterday they
knocked on their door, that is (to say) that they hoped for the best, but yesterday
they knocked on their door…
Since the temporal ordering is true enough, but crucially, not speaker-intended,
(20), augmented by the temporal inference, fails both the That Is and the Indirect
Faithful Report tests: 8
7. Material enclosed in double parentheses was left implicit in the addressor’s utterance be-
cause the writer felt it was self-evident in the original context. I’ve here added it for ease of
comprehension.
8. The natural paraphrase often used in temporally ordered cases is then, but since then is no
longer restricted to temporal relations it would be quite acceptable in the paraphrase above, but
it doesn’t mean that the inference is one of temporality.
(22) a. That is (to say) test: ?? S said that… they hoped for the best, and yesterday
they knocked on their door, that is (to say) they knocked on their door
later on, yesterday.
b. ??S said that… they hoped for the best, and yesterday they knocked on their
door, and in addition, he indirectly conveyed that they knocked on their
door later on, yesterday.
The paraphrases which incorporate the temporal assumption into the “said”/
explicated (22a) or the implicated (22b) are not considered faithful, because indeed,
the speaker had no intention of conveying it, not even implicitly. In other words,
not every legitimate inference (as the temporal inference here is) forms part of the
speaker-intended reading. Meshing utterances with the objective reality they refer
to is the wrong way to go about analyzing readings. Once we look to speakers’ inten-
tions, we can see that the same content (e.g. temporal or causal relations) may count
as part of the reading on some occasion, but not on another. It’s certainly not the
case that “once explicated, always explicated”. If so, coherence inferences about ob-
jectively true relations holding between the states of affairs depicted by and conjuncts
must not be seen as part of the speaker-intended reading in every case. The unified
account for and coherence inferences must be given up. Only a subset of these form
part of the reading, because only in some contexts does the speaker intend them.
4. Scalar quantifiers
Next, consider many and most in the following examples:
(23) a. WALT: He was afraid many times,

(H) but he p=ushed through his fears.
… In order (H) … to= … succeed and have v=ictory. (SBC: 021)
b. I mean if you guys are giving most of the money then that’s all right
(LSAC)
Common wisdom in linguistics has it that scalar quantifiers are only lower-bounded
at the core. Thus, many means “at least many (and possibly all)” and most means
“more than half and possibly all”. The idea behind this hypothesis was that should
it turn out that the “he” in (23) (Martin Luther King) was afraid at all times Walt’s
statement is not false. Similarly, if the addressees in (b) are giving all the money,
the assumption is that the speaker commits to the relevant state of affairs being all
right. But can we really attribute to Walt an intention to convey “at least many and
possibly all times”? Probably not. The same is true for most in (b). The explanation
normally given is that an upper-bound (“not all”) implicature is routinely derived,
28 Mira Ariel
creating a “more than half but less than all” interpretation for e.g., most. My analysis
is quite different, and crucially involves a Truth-Compatible inference component.
I maintain that the “possibly all” assumption (rarely) associated with scalar quanti-
fiers is not part of the quantifiers’ linguistic meaning. It is, rather, a TCI.
I have argued that scalar quantifiers (specifically, some and most) encode a
“circumbounded” meaning, that is, they are both lower- and upper-bounded (Ariel
2004, 2008, 2014, 2015). For example, the meaning of Most Xs are Ys specifies the
existence of a proper subset of Xs (“less than all”), larger than half, that are Ys.
Crucially, nothing is predicated about the complement set of Xs. 9 No implica-
ture is then needed to account for the prevalent, circumbounded reading of scalar
quantifiers: “What you say (the linguistic meaning) is what you get”. Most of the
arguments in favor of this analysis are not relevant to our point here, but the one
pertaining to TCIs is. What’s at stake is the status of “possibly all” assumptions with
respect to truth value judgments. Under the classical analysis, “possibly all” is part
of the lexical meaning, and should therefore count as part of the quantifier’s reading
(unless cancelled by an implicature). According to my analysis it is a pragmatic
inference, specifically, a TCI.
In a questionnaire I administered to Hebrew speakers, subjects were asked to
determine whether an utterance on most Xs are Ys… is true when reality is that “all
Xs are Ys” (Ariel 2004, 2006). If the meaning of most is only lower-bounded (“more
than half, and possibly all”) subjects should not hesitate in determining that the
proposition is indeed true. This is because the upper-bounding interpretation (as-
sumed to be “not all”) is a (Generalized) Conversational Implicature, and as such,
is not supposed to influence the truth conditional content of the triggering utter-
ance. The results of this experiment, however, were quite far from it. Depending
on the type of question and the context presented, the rates of subjects confirming
the truth of the most utterance when “all” was the case varied between 5.9% and
83.3%. In other words, in some cases a respectable majority (but nothing close to an
overwhelming majority) of the subjects determined that most… is true when “all” is
the case, but in others they absolutely thought it wasn’t. In fact, I’m not the only one
to have received such nonabsolute results. Ever since Noveck (2001), researchers
have repeatedly reported that subjects vary on whether they judge sentences such
as Some elephants are mammals as true or as false. 10
9. This lexical meaning is therefore not identical to the strengthened interpretation “more than
half but not all” commonly assumed. The latter does refer to the complement set in order to deny
that the predicate is true for it (see Ariel 2015, for discussion and a formal definition).
10. Note, however, that these experimental results have been interpreted as pointing to what
researchers call “logical” versus “pragmatic” judgers. “Logical” subjects judge such sentences
true, “pragmatic” subjects judge them false. I on the other hand, have found that the difference
lies not with the type of subject, but with the type of question.
What can account for this variability in truth judgments? It’s certainly not
expected if “possibly all” is a linguistic meaning. I have proposed that once we
assume that “possibly all” is not part of the linguistic meaning of some, many and
most, the variability can be straightforwardly accounted for. Linguistic meanings
are uncancelable, but pragmatic inferences, in this case, Truth-Compatible infer-
ences, are quite optional, since they are not speaker-intended, in fact. Hence, as-
suming a circumbounded analysis for the scalars, subjects faced with a reality in
which “all” is the case need to determine whether the partial quantity denoted by
the quantifier (say “many times”) is compatible with a state of affairs in which “all
(times)” is true. In other words, subjects here compare between the encoded (or
explicated) linguistic meaning (a partial quantity) and reality (“all”). Under such
circumstances judgments may indeed vary, depending on the TCI subjects adopt.
People may judge that the “part is compatible with the whole”, in which case they
will judge the sentence true. But they may refuse to derive this TCI, and then judge
the sentence as false.
In other words, TCIs, if derived, bridge the gap between aspects of reality (e.g.,
the number of times Martin Luther King was afraid, the number of elephants that
are mammals) and linguistic meanings, which semantically do not stand for those
facts. Such inferences are not guaranteed, of course. Subjects are less likely to bridge
this gap when the difference between “most” and “all” is discourse-relevant. In
other contexts, the difference is less relevant, in which case subjects might derive
the “possibly all” TCI, and judge the scalar quantifier sentence true if “all” is the
case (Ariel 2014, 2015). On my analysis, linguists treated a potential “compatible
with all” TCI (all TCIs are potential, in fact) as if it were a necessary component (“at
least…”) that they therefore imposed on scalar quantifiers (and other scalar items)
as a core linguistic meaning.
Once we analyze “compatibility with all” as a TCI for scalar quantifiers we
can understand why this interpretation would fail the That Is (a) and the Indirect
Addition (b) Faithful Report tests (I here test only one relevant part of 23):
(24) a. That is (to say) test: ?? S said that if you guys are giving most of the money
then that’s all right, that is (to say) if you guys are giving most and possibly
all/at least most the money then that’s all right.
b. Indirect Addition test: ?? S said that if you guys are giving most of the money
then that’s all right, and in addition, s/he indirectly conveyed that if you
guys are giving all the money then that’s all right.
According to these Faithful Report tests, “possibly all” is neither explicated (24a)
nor implicated (24b). However, just as we saw for and, once (not) explicated does
not mean always (not) explicated. Indeed, as argued in Ariel (2004), there are cases
where “possibly all” is explicated for scalar quantifiers (rarely, where we interpret
30 Mira Ariel
most as “at least most”). But in the majority of the cases, I’m suggesting that a TCI
status is the appropriate status for what is normally taken as a lexical semantic
meaning, and hence, as part of most’s reading when not cancelled.
Next, let’s apply the “X, including Y” test. We have seen above (Section 2.2)
that e.g., the category of weapons can be stretched to include replica weapons and
tear gas, but not bottles of water. We can then examine the boundaries around the
categories of scalar quantifiers. We don’t expect “50%” to be a legitimate member of
the category, since it’s not a linguistically viable value for most (25a). Neither should
a prototypical instantiation such as “85%” properly occur as the Y element in a con-
struction where most is the X category, because the construction is restricted to a
noteworthy instantiations (25b). “51%” (25c), however, may very well be noteworthy
here, since it is a marginal member of the “most” category. Indeed, this is the case:
(25) a. ??You can be admitted to the Linguistics department provided you get most
of the exam answers correct, including 50% of them.
b. ??You can be admitted to the Linguistics department provided you get most
c. √You can be admitted to the Linguistics department provided you get most
Now, what about the maximal value of most? According to common analyses, most
is only lower-bounded. Hence, 100% is its highest legitimate member. If my analysis
is correct, however, only a value that’s minimally less than all is most’s highest cat-
egory member, say “99%”. If so, we should see an acceptability difference between
99% and 100% as Ys in the X, including Y construction. Preliminary questionnaire
results show precisely this. Fifteen informants judged the acceptability of either
(a) or (b) in (26), on the assumption that the speaker in both utterances is the
principal of a very prestigious high school, interviewing a candidate for a teaching
job at that school. The informants were asked to rate the acceptability of (a) or (b)
on a 7-point scale: 11
(26) a. You should know that if most, including 99% of the students, are failing
your course, you will have to tutor each one of them individually.
b. You should know that if most, including 100% of the students, are failing
your course, you will have to tutor each one of them individually.
Note that the Downward entailing context here (a conditional antecedent) is

said to be strongly associated with a lower-bound-only reading for the quantifier
11. My informants were University of Pennsylvania male students. I thank Iddo Toledano for
eliciting these judgements for me.
(Chierchia 2013). Hence, there should have been no problem in accepting the (b)
version. Still, (26b) was not judged acceptable (receiving a mean rate of 3.75, which
is lower than the minimal 4, given a 7-point scale). (26a), on the other hand, was
judged acceptable, even if not considered stellar, with a mean rate of 4.7.
The contrast between (26a) and (26b) shows that most’s meaning does not
include “possibly all”. This is so, despite the fact that most can be compatible with
“all”, as has indeed been determined by the very same informants. Thus, I also
asked my informants to choose a “yes”, “no” or “can’t decide” answer to one of the
following questions:
(27) a. If most of the partners can make it, the meeting will be held on Feb. 17 in
Los Angeles.
It turned out that all the partners can make it. Will the meeting be held
on Feb. 17 in Los Angeles?
b. A senior associate in a law firm addressing the 25 junior associates of the
firm: If most of you are interested in free tickets to the Jazz concert tonight
we’ll be happy to give them out.
It turned out that everybody was interested. Will the law firm give them
all tickets?
Combining the responses to both questions, a solid majority chose “yes” (11/14)
(2 chose “no” and 1 chose “can’t decide”). In other words, most of my informants
thought that most is compatible with a reality in which “all” is the case. At the same
time, they did not find most, including 100% acceptable. Once again, we see a differ-
ence between the reading associated with most (upper-bounded) and the “possibly
all” Truth-Compatible inference potentially associated with most. “Possibly all” is
then not a reading of most, but rather, a Truth-Compatible inference potentially
associated with it.
Thus, the X, including Y construction helps us tease apart bona fide readings,
not only from interpretations that are clearly outside the linguistic category (e.g.,
“50%” for most), but also from interpretations that might be compatible with
the speaker’s intended reading (e.g., “100%” for most). Both equally fail the test,
which shows that neither are part of the speaker’s intended message. All in all, the
Including test supports the results from the Faithful Report tests in testifying that
“possibly all” is not part of the reading of the scalar quantifiers, because it’s a TCI.
32 Mira Ariel
5. So-called exclusive or interpretations
My final case for removing the reading status from an assumption currently seen
as part of a reading concerns so-called exclusive or cases, as in (28). This putative
reading of or constructions, whereby X or Y is interpreted as “either X, or Y but
not as both X and Y” is said to be derived from the linguistic “inclusive” meaning
(“possibly X, possibly Y, or possibly both X and Y”) by what is still called a “scalar
implicature” of “not both”:
(28) A group of social activists met at Noam’s offices last week to look for solutions.
They debated whether protest action was needed, or long-term activity in
schools and mosques. (Haaretz, 2.11.2014)
(28) is said to receive an exclusive reading because the speaker is not entertaining
the possibility that the activists will engage both in protest action and in long-term
activity in schools and mosques. That much is certainly true. But is the exclusion of
the conjunction of the alternatives a speaker-intended reading, or does it actually
have a different, nonreading type of source? I will suggest that the fact that the
conjunction of the alternatives is not here considered is not part of the speaker’s
communicated message. Rather, it is a Searle (1980) Background assumption.
Let’s see what the Faithful Report tests show about this purported “not both”
reading:
(29) a. That is (to say) test: ?? S said that they debated whether protest action was
needed, or long-term activity in schools and mosques, that is (to say)
whether protest action was needed, or long-term activity in schools and
mosques, but not both.
b. Indirect Addition test: ?? S said that they debated whether protest action
was needed, or long-term activity in schools and mosques, and in addition,
she indirectly conveyed that not both actions were needed.
c. Replacement test: ?? S said that they debated whether protest action was
needed, or long-term activity in schools and mosques, but actually, she
indirectly conveyed that protest action was needed, or long-term activity
in schools and mosques, but not both.
“Not both” is neither explicated here (29a) nor is it implicated (29b). It is also not
a strong implicature in a Two-Tier “special” use (29c). (28) describes a choice the
activists needed to make between alternatives (protest versus activity in schools
and mosques). The alternative of initiating both protest and school and mosque
activity, is not a viable alternative in the discourse here (although it may certainly be
in other contexts). Hence, denying it is not a speaker-intended message. According
to Ariel and Mauri (2015), the Explicature of the relevant portion of (28) is (this is
a Choice or construction):
(30) That is (to say) test: S said that the activists debated whether protest action
was needed, or long-term activity in schools and mosques, that is (to say) the
activists debated a choice between protest action and long-term activity in
schools and mosques.
The status of “not both alternatives”, considered an essential part of the exclusive
reading, is a Background assumption, accessed from our general knowledge store,
on our analysis. 12 It is not speaker-intended, although the speaker does expect
the addressee to take this assumption into consideration. If it’s a background as-
sumption, no wonder “not both” fails all the speaker-intended tests. It is not part
of the reading.
As argued in Ariel and Mauri (2015), this Background status for the “not
both” assumption is true for virtually all cases that would be classified ‘exclu-
sive’ readings under current assumptions. 13 What we see then is that “not both”
which is analyzed as implicated by Griceans, as explicated by Relevance theore-
ticians and as a grammatically derived implicature under recent analyses, such
as Chierchia (2013), should be reanalyzed as a Background assumption. In other
words, what’s considered an essential part of an or reading is not actually a reading
component. 14
Let’s now apply the additional X, Y excluded test, which identifies Y as an in-
stantiation of the X category, the linguistic meaning of X or Y here. The whole or
construction then stands for the X in the X, Y excluded construction, and “both
alternatives” instantiates the Y member, which should be excludable, if it is a legit-
imate instantiation of “X or Y”. (31) shows that this paraphrase fails for (28):
(31) ??Ssaid that the activists debated whether protest action was needed, or long-
term activity in schools and mosques, both protest action and long-term
activity in schools and mosques excluded.
12. The reason why my account does not need to somehow cancel a “possibly all” alternative is
that we do not assume a linguistic inclusive meaning for or constructions. But this is not crucial to
my point here, which is that “not both” is not explicated nor implicated in what linguists classify
as exclusive readings.
13. But see Ariel and Mauri (2015) for rare cases where “not both” happens to be implicated (as
a special case of “other alternatives are ruled out”).
14. We have argued that the relevant reading in cases such as (28) is “Unresolved Choice”.
34 Mira Ariel
(31) does not reflect the speaker’s intended message. Hence, the exclusive inter-
pretation, as it is currently defined (by specifically ruling out “both…”) is not the
speaker-intended reading of so-called exclusive cases. Once again, linguists have
been following objective facts about the reality behind the speaker’s assertion (in-
deed, the conjunction of the two alternatives is not considered) and imposed them
as a speaker-intended reading, when in fact they are Background assumptions.
6. Summary and conclusions
The point of this chapter is that putative interpretations long considered part of the
reading associated with certain linguistic expressions may not actually enjoy such a
prominent status. I consider an interpretation as part of the reading of some linguis-
tic expression provided it is systematically speaker-intended when the linguistic ex-
pression is used. Speaker-intended aspects of meaning include linguistic meanings,
explicated inferences, implicated inferences, as well as second-tier implicatures,
which come to replace the first-tier explicature. While all of these representations
are speaker-intended, only the first three are potentially recurrent interpretations.
Second-tier interpretations are typically special, ad hoc inferences. Hence, we’re left
with linguistic meanings, explicated and implicated inferences. My main argument
has been that for the cases here considered what are taken as linguistic meanings,
explicated inferences, or implicated inferences are not necessarily part and parcel
of the speaker-intended reading. The reason is that assumptions that have been
analyzed as bearing any one of these statuses are not actually speaker-intended, but
rather, Truth-Compatible inferences or Background assumptions.
With respect to and-related inferences, my point was that on some occasions
such inferences are speaker-intended, and hence part of the reading. However, on
other occasions, the very same coherence inferences are only Truth-Compatible
inferences. Such coherence inferences should not count as part of the and reading.
This means that the very same inference, associated with the very same linguistic
expression, may sometimes count as part of the reading, while on other occasions
it doesn’t, because it’s only a Truth-Compatible inference or a Background assump-
tion. Next, the so-called exclusive reading of X or Y constructions, whereby the con-
junction of the alternatives is ruled out (“not both”) is again not a speaker-intended
message. Rather, it is a Background assumption, and hence, not a reading. Finally,
while the “possibly all” interpretation associated with scalar quantifiers is considered
part of its lexical meaning, and hence, unless cancelled, part of its reading, it is more
often than not merely a Truth-Compatible inference. This is so, even in Downward-
entailing contexts, where “possibly all” is quite compatible with the quantifier. The
point is that compatibility with reality does not guarantee some assumption the sta-
tus of a speaker-intended message. Readings are about speaker-intended messages,
not about meshing the utterance with the objective reality that must lie behind it.
Recognizing the role of Background and Truth-Compatible inferences is important
for defining recurrent readings for constructions at the center of current debates
(and, scalar quantifiers, and or), and surprisingly enough, the implications are not
restricted to interpretations currently analyzed as pragmatic inferences. I propose
that contents that are currently analyzed as core linguistic meanings may in fact
not be so.
References
Ariel, Mira. 2004. “Most.” Language 80: 658–706. doi: 10.1353/lan.2004.0162

Ariel, Mira. 2006. “A ‘Just That’ Lexical Meaning for Most.” In Where Semantics Meets Pragmatics,
ed. by Klaus von Heusinger and Ken Turner, 49–91. London: Elsevier.
Ariel, Mira. 2008. Pragmatics and Grammar. Cambridge: Cambridge University Press.
doi: 10.1017/CBO9780511791314
Ariel, Mira. 2012. “Relational and Independent and Conjunctions.” Lingua 7: 1682–1715.
Ariel, Mira. 2014. “What Discourse Can(not) Teach Us.” International Review of Pragmatics 6:
181–210. doi: 10.1163/18773109-00602001
Ariel, Mira. 2015. “Doubling Up: Two Upper Bounds for Scalars.” Linguistics 53: 34–50.
doi: 10.1515/ling-2015-0013
13: 150–172. doi: 10.1515/ip-2016-0001
Ariel, Mira, and Caterina Mauri. in press. Why use ‘or’? Linguistics.
Bach, Kent. 1994. “Semantic Slack: What is Said and More.” In Foundations of Speech Act Theory:
Philosophical and Linguistic Perspectives, ed. by Savas L. Tsohatzidis, 267–291. London:
Routledge.
Bach, Kent. 2006. “The Top 10 Misconceptions about Implicature.” In Drawing the Boundaries of
Meaning: Neo-Gricean Studies in Pragmatics and Semantics in Honor of Laurence R. Horn,
ed. by Betty J. Birner and Gregory L. Ward, 21–30. Amsterdam: John Benjamins.
doi: 10.1075/slcs.80.03bac
Carston, Robyn. 1988. “Implicature, Explicature and Truth-theoretic Semantics.” In Mental
Representations: The Interface between Language and Reality, ed. by Ruth M. Kempson,
155–181. Cambridge: Cambridge University Press.
Carston, Robyn. 2002. Thoughts and Utterances: The Pragmatics of Explicit Communication.
Oxford: Blackwell. doi: 10.1002/9780470754603
Chierchia, Gennaro. 2013. Logic in Grammar. Oxford: Oxford University Press.
doi: 10.1093/acprof:oso/9780199697977.001.0001
Koenig, Jean-Pierre. 1991. “Scalar Predicates and Negation: Punctual Semantics and Interval
interpretations.” Chicago Linguistic Society 27: 140–155.
Noveck, Ira A. 2001. “When Children Are More Logical than Adults: Experimental Investigations
of Scalar Implicature.” Cognition 78: 165–188. doi: 10.1016/S0010-0277(00)00114-1
36 Mira Ariel
Searle, John R. 1980. “The Background of Meaning.” In Speech Act Theory and Pragmatics, ed.
by John RJohn R. Searle, Ferenc Kiefer, and Manfred Bierwisch, 221–232. Boston: Reidel.
doi: 10.1007/978-94-009-8964-1_10
Wilson, Deirdre, and Dan Sperber. 1998. “Pragmatics and Time.” In Relevance Theory: Applications
and Implications, ed. by Robyn Carston and Seiji Uchida, 1–22. Amsterdam: John Benjamins.
doi: 10.1075/pbns.37.03wil
Chapter 3
Pronouns and implicature
Wayne A. Davis
Georgetown University
In this chapter I investigate how differences among pronouns are related to dif-
ferences in implicature. I characterize pronouns within my foundational theory
of meaning, according to which words are conventional signs of mental states,
principally thoughts and concepts, and meaning consists in their expression.
Indexicals express concepts that are distinctive in the way they link to other con-
cepts or presentations. Indexical concepts are individuated by their sortal and
determiner components. Indexicals have deictic, demonstrative and anaphoric
uses. Pronouns are indexical words that can be used anaphorically with nouns
as antecedents. After reviewing previous findings that the distinction between
reflexive and non-reflexive pronouns cannot be explained in terms of impli-
cature or pragmatic principles, I describe a wide range of implicatures that are
generated by pronoun use. Some are semantic – what Grice called “conventional
implicatures”. Sentences have these uncancelable implicatures because of the
specific concepts expressed by the pronouns. Pronoun use also generates a wide
range of conversational or pragmatic implicatures, through both non-semantic
convention and specific contextual factors.
Keywords: pronouns, conventional implicature, conversational implicature,

indexicals, foundational theory of meaning
1. The expression theory and non-descriptive meaning
This chapter reflects the view that words are conventional signs of mental states,
principally thoughts and ideas, and that meaning consists in their expression. In
Davis (2003), I explain what it is for words to have meaning and express ideas in
terms of speaker meaning and expression, and what it is for a speaker to mean or
express something in terms of intention.
I focus on thinking the thought that p as a propositional attitude distinct from
believing that p. One can think the thought that the moon is made of green cheese
without believing it, and one can believe that bats fly without thinking that thought
doi 10.1075/pbns.276.03dav
38 Wayne A. Davis
at the moment. Thinking in this sense differs from believing in being an event in
the narrow sense of an occurrence or activity rather than a dispositional state. We
retain our beliefs when we are asleep or unconscious, but thoughts are something
actively going on. Thoughts, on my view, are structured events, and a particular
kind of mental representation. They are similar in many ways to sentences, but
fundamentally different. We think when thoughts occur to us. For S to think a
thought T is for T to occur to S. I define propositions as thoughts with a declarative
structure, or equivalently, as objects of belief and desire. Thought plays a role in
the explanation and prediction of action and emotion different from belief and
desire, but equally important. For example, no matter how much a man wants beer,
and how certain he is that he will get beer if he goes to the supermarket, he will
not actually get any beer (except perhaps by accident) unless he thinks about beer
at the right moment in the store. I argue at length that thoughts have constituent
structure – specifically, a phrase-structure syntax. Unlike the familiar “language of
thought” hypothesis, what is structured on my view are not hypothetical “vehicles”
of thought, but thoughts themselves.
I define ideas (or concepts) as thoughts or parts of thoughts, and distinguish
them carefully from conceptions (belief systems) and sensory images (structures
of sensations). Conceptions and images are important forms of mental representa-
tion, but meaning cannot be defined in terms of them. In addition to occurring
and being parts of thoughts, concepts can be acquired and possessed, and may
become associated with each other. We customarily refer to ideas using what I call
“ideo-reflexive reference”, whereby noun phrases containing an expression refer
to the idea it expresses. Thus we use the idea “vixen”, or equivalently, the idea of a
vixen, to refer to the idea of a female fox when we use ‘vixen’ therein to express that
idea. We similarly use the thought “Bush was president” or the thought that Bush was
president to refer to the thought expressed by “Bush was president” on that occasion.
Referring to an idea is different from expressing it.
Unlike Grice (1989), who assumed that meaning entails attempting to commu-
nicate and produce a belief in an audience 1 (false when talking to our pets, writing
individual words, or telling stories), I take speaker meaning to entail expressing
a thought, belief, or other mental state, which involves performing an observable
action as an indication or natural sign that the state is occurrent. Thus S expressed
the idea and meant female fox by uttering ‘vixen’ only if S uttered ‘vixen’ as an in-
dication that the idea female fox is occurring to him.
Grice focused on meaning that p (“cognitive” speaker meaning), which can
be defined as expressing the belief that p. Implying involves expressing one belief
1. See the overviews in Huang (this volume) and Dynel (this volume).
Chapter 3. Pronouns and implicature 39
by expressing another. If one writes ‘Mars exploded’ in a work of fiction, however,

one is expressing the thought but not the belief that Mars exploded. Similarly, one
expresses the idea of Mars by writing ‘Mars’, not a belief. Meaning “Mars exploded”
by a sentence, or “Mars” by a word or phrase, is a distinct kind of speaker meaning,
which I call “cogitative”. It can be defined as the direct expression of thoughts or
thought parts. As Schiffer (1972: 2–3) observed, when Mark says “Bush is brilliant”
ironically, Mark means “Bush is brilliant” by the sentence he uttered, but did not
mean that Bush is brilliant by uttering it. While this distinction is easy to miss
given that quotation and ‘that’ subordination are normally just alternative ways of
referring to a proposition, it is easy to see that what Mark did was express the belief
that Bush is not brilliant by expressing the thought that he is brilliant.
On my definition, the meaning of individual words, as well as the non-compo-
sitional meaning of idioms, is given in terms of what ideas they are conventionally
used to directly express. The meaning of compositional compounds is provided by
a recursion clause, based on conventions to use certain expression structures to
express certain idea structures. Convention here means “custom” or “practice estab-
lished by general consent or common usage”. Conventions in this sense are common
practices that perpetuate themselves through precedent, social acceptance, habit
and association, traditional transmission, and contributing to common goals, yet
are arbitrary. One goal served by language is communication. Conventions are
arbitrary in that other regularities could have served the same purposes and per-
petuated themselves in the same ways.
Defining meaning as idea expression rather than reference enables natural solu-
tions to Frege’s and Russell’s problems. People do think about Santa even though
Santa does not exist, and such thoughts have a part conventionally expressed by the
name Santa. So Santa has a meaning even though it has no referent. The thought
“ammonia is poisonous” is distinct from the thought “NH3 is poisonous” even
though ammonia is NH3. Since ammonia and NH3 express different thought parts,
they have different meanings, even though their extensions are identical. I reject
the Fregean assumption, though, that senses and concepts must be descriptive (see
especially Davis 2005). We clearly use Mars to express the common component
of the thought that Mars is a planet, the thought that Mars is smaller than Jupiter,
and so on. Kripkean arguments show that a name cannot be defined exclusively
in descriptive terms. Mars does not have a meaning like that of the fourth planet
from the sun. Other terms with nondescriptive meanings are the logical constants,
prepositions – and indexicals.
40 Wayne A. Davis
2. Indexical meaning and concepts
The paradigm indexicals include the personal pronouns I, you, he, she, and it; the
demonstrative pronouns this and that, plus noun phrases with them as determiners;
and the locative pronouns now, here, there, and then. These expressions contrast
markedly with proper names like Mars and definite descriptions like the fourth
planet from the sun in the way their reference is determined by an element of the
context of use. Indexicals have different referents in different contexts even when
used in the same sense and evaluated with respect to the same circumstances.
Indexicals not only have contextually variable referents, but different ways of
being used. Imagine that Thomas Jefferson utters sentence (1) while in a room with
George Washington, Alexander Hamilton, and Benjamin Franklin.
(1) Washington became president after he led the Continental Army to victory.
a. Anaphoric: ‘Washington’ is the antecedent of ‘he’.
b. Demonstrative: The speaker is pointing at Hamilton.
c. Deictic: The speaker is visually focusing on Franklin.
On the most natural interpretation of (1), the name ‘Washington’ is the antecedent
of ‘he’, and the pronoun refers to George Washington. This use is anaphoric. But it
is also possible that the speaker is using ‘he’ while pointing at Alexander Hamilton
with the result that the pronoun refers to Hamilton rather than Washington. This
use is demonstrative. Finally, it is possible that the speaker is using ‘he’ without
using it anaphorically, and without pointing at anything. The speaker might simply
be visually focusing on Benjamin Franklin, with the result that the pronoun refers
to Franklin rather than Hamilton or Washington. I call this the deictic use. 2Lyons
(1977: 660) observed that when he is anaphoric in sentences like (1), it is typically
uttered with normal stress. When deictic or demonstrative, it has heavier contras-
tive stress. He is used in the same sense (linguistic meaning) in all three uses. But
the referent is determined in different ways. The same three uses can be observed
with nearly all indexicals. For example, (1) can be redone with this patriot in place
of he. There are many distinct species of anaphoric, deictic, and demonstrative uses,
and other uses besides these.
An ideational theory may seem patently absurd for pronouns. Whereas it seems
tautological to say that Mars expresses the idea of Mars, and that the fourth planet
from the Sun expresses the concept of the fourth planet from the Sun, we cannot
say in general that he expresses the idea of he or of him. But this is due, I believe,
2. The terms ‘anaphoric’, ‘demonstrative’, and ‘deictic’ are common in linguistics, but there is
little consensus on their usage (see Dynel and Cap, Section 1, this volume). There are many ways
of classifying the great variety of indexical uses.
to the fact that when ‘φ’ is a pronoun in a phrase of the form the idea of φ, it must
have the objective case (him vs. he) and a particular referent. Being indexical, he
does not express a concept with a fixed referent. So ‘He’ expresses the idea of him is
either used improperly without a referent for him, or else it is not generally true.
When we use instead the quotational form the idea ‘he,’ an ideo-reflexive reading
is forced. ‘He’ expresses the idea ‘he’ is not obviously incorrect, but does presuppose
that he expresses a thought part.
Another possible reason for rejecting an ideational theory of indexical mean-
ing is that pronouns do not have enough descriptive content to determine their
referents or to discriminate between indexicals like it and that. But this argument
falsely assumes that all concepts are descriptive.
In the plus column, it is evident that sentences like (1) do express complete
thoughts in any given context of use, and that the pronouns express therein com-
ponents of those thoughts. It is not obvious, though, that they express the same
thought part on every occasion. It is conceivable that he expresses the concept of
Washington on one occasion, and the concept of Jefferson on another, without
any one concept being expressed on both occasions. However, he has its English
pronoun meaning but does not express the concept of any particular person in
contexts like (2):
(2) The French word il means “he”.
My main argument is an abduction: the hypothesis that pronouns express the same
thought part on different occasions, suitably developed, enables us to provide the
best explanation of their behavior, and to solve the problems confronting other
theories. This hypothesis receives analogical support from the fact that terms other
than pronouns express the same thought part when used with the same linguistic
meaning, even indexical terms like ‘enemy’ and ‘neighbor’. The hypothesis thus
enables a highly uniform account of word meaning. 3
3. Indexical determinants
I argue that the personal, demonstrative, and locative pronouns express primary
indexical concepts, ones that do not contain other indexical concepts. They are like
the concepts expressed by proper names and definite descriptions in serving as sub-
ject concepts in propositions. Primary indexical concepts are like syncategorematic
3. I present parts of my theory in Davis 2013, 2016a, and forthcoming. The whole theory is
presented in Indexical Meaning and Concepts (ms).
42 Wayne A. Davis
concepts, however, and markedly unlike proper name and definite description con-
cepts, in that they do not themselves represent particular objects and therefore
have no reference of their own. Something external to the concepts gives them an
extension and intension on particular occasions. I hypothesize that one of the dis-
tinctive features of primary indexical concepts is that their occurrences are typically
connected in a particular way with other representational mental events, includ-
ing sensory experiences and other subject-concepts, whose objects become their
referents. Primary indexical concepts are capable of being linked to a determinant.
Consider a sentence like (3), which has at least two different interpretations
and logical forms because either Jefferson or Adams can be the antecedent of he.
(3) Jefferson opposed Adams because he was a federalist.
What makes one noun the antecedent rather than another on a given occasion of
use? I hypothesize that he expresses the same thought part whenever it is used with
its third-person sense. Let us use ‘c(he)’ to refer to that concept (i.e., as short for the
concept ‘he’). Let ‘c(Jefferson)’ and ‘c(Adams)’ refer in the same way to the concepts
expressed by Jefferson and Adams. It seems evident that (3) is ambiguous because
it can express either a thought in which c(he) is linked to c(Jefferson) or one with
c(he) linked to c(Adams). Which thought is expressed depends on the speaker’s
intentions. The antecedent of he could also be a noun phrase used before (or after)
(3) is uttered, perhaps Hamilton. In that case, (3) expresses an unlikely thought
in which c(he) is linked to c(Hamilton). The speaker S cannot think that thought
unless c(he) and c(Hamilton) both occur to S and do so in the right relationship,
which I call indexical linkage.
Let '℞’ be a variable for any primary indexical concept and ‘δ’ for any deter-
minant. On my view, concepts are event types that can occur at different times as
parts of different thoughts. Let '℞δ’ stand for the subtype of ℞ that consists of its
occurrence linked to δ. Hence:
(4) ℞δ occurs to S iff ℞ and δ occur to S and their occurrences are linked.
℞δ is thus a more specific event type than ℞. ℞δ is to ℞ as driving with friends is to

driving, and striking in anger to striking. For a neural model, suppose ℞ is a circuit
whose activation is the occurrence of ℞. Let ℞ be capable of being activated by other
neural circuits, including δ. Then ℞δ might represent the activation of ℞ by δ over
a specific type of neural pathway.
From a semantic standpoint, the most important rule governing ℞δ is the de-
rived reference rule:
(5) ex{℞δ} = ex{δ} when ℞δ has an extension.
Thus the extension (referent) of c(he)c(Jefferson) is Jefferson, and the extension of

c(he)c(Adams) is Adams. So when he has Jefferson as its antecedent, it refers to Jefferson.
For Jefferson is the antecedent of he when the indexical concept expressed by he has
the concept expressed by Jefferson as its determinant. The speaker uses he to express
the generic concept c(he), thus meaning “he” by he. The speaker also expresses the
more specific concept c(he)c(Jefferson), thereby referring to Jefferson. Indexical terms
are characterized by a “double triangle of signification”: the reference of a term is
in general that of the idea it expresses, and the reference of an indexical idea is that
of the determinant it is linked to.
Returning to example (1), the pronoun he is used anaphorically to refer to
George Washington, on my theory, when he is used to express an occurrence of
c(he) linked to the concept c(Washington) expressed by the proper name Washington
on that occasion. When the pronoun is used demonstratively, the determinant is
related in a particular way to the speaker’s pointing gesture. I believe the determi-
nant in this case is a concept that refers in a particular way to the gesture. In (1),
it would be something like c(the man I am pointing at). In the deictic usage, there
is no antecedent or demonstration. The indexical concept is instead linked to a
perceptual, introspective, memory, or even hallucinatory experience – what I call
a presentation. When Jefferson used he deictically, he was referring to Franklin be-
cause he was visually attending to Franklin, and c(he) was linked to that perception.
Had Jefferson instead been using he to express an occurrence of c(he) linked to his
perception of Hamilton, he would have referred deictically to Hamilton. Note that
Jefferson could also have been using he to express an occurrence of c(he) linked
to his visual or auditory perception of Washington. In that case, Jefferson would
have referred deictically rather than anaphorically to Washington. And he might
not have recognized that he is Washington. We are going to focus on the anaphoric
use in this chapter, so I will not say more about the other uses.
4. Sortal and determiner components
How does c(he) differ from c(she), c(it), c(I), and other primary indexical concepts?
Are they atomic or do they have components? I hypothesize that each primary in-
dexical concept is composed of a very general sortal concept plus a nondescriptive
determiner. I present evidence that the sortal component of c(he) is c(male), whereas
that of c(I) is the more general concept c(animate). The sortal component of c(this),
expressed by ‘this’ as a pronoun, is c(thing); that of a phrase of the form ‘this N,’ in
which ‘this’ is a determiner, is c(N). The sortal component places a condition on
the derived reference rule (5):
44 Wayne A. Davis
(6) ex{℞δ} = ex{δ} provided ex{δ} is in the extension of the sortal component of
℞; otherwise ℞δhas no extension.
Suppose Terry points at someone he mistakenly believes to be male, and says He is

tall. Even though Terry is referring to a woman, he cannot refer to a woman because
of its sortal component. His statement suffers from the same presupposition failure
as This/The male is tall.
The determiner component of a primary indexical concept does a number of
things. It combines with the sortal component to form an singular subject-concept
that is indexical and therefore capable of linking to determinants. It furthermore
determines the specific range of determinants the concept can link to. No two pri-
mary indexical concepts can occur with the same range of determinants. Each ℞ has
a unique set of determinant constraints, which specify the determinants ℞ cannot
link to. Some pronouns, such as he and himself differ only in their determinant
constraints (Section 6); there is no difference in their sortal components. Others,
like he and I, have differences in determinant constraints that are independent of
their sortal components. Some are illustrated by (7).
(7) a. Bush believes “I was president”.
b. Bush believes that I was president.
c. Bush believes that he was president.
I can refer to Bush if Bush used (7b). But Bush cannot be the antecedent of I as
it typically is in (7a) or as he is in (7c). In marked contrast, the he in (7c) can be
interpreted anaphorically with Bush as its antecedent, but cannot be interpreted as
referring deictically to the speaker the way I typically is in (7b). Another distinctive
characteristic of English I arising from determinant constraints is illustrated by (8):
(8) a. Caesar met Vercingetorix and thought “I will be victorious”.
b. Caesar met Vercingetorix and thought “He will be victorious”.
In (8a), the pronoun can only have Caesar as its antecedent because that is the
subject of the propositional attitude verb thought. In (8b) the pronoun can be in-
terpreted as referring back to Caesar (especially given Caesar’s unusual practice
of referring to himself in the third-person), but is most naturally interpreted as
referring to Vercingetorix. It is easy to imagine languages in which the occurrence
constraints characterizing the personal pronouns are different.
5. Pronouns
The central feature in definitions of pronouns since Dionysius (c. 180 bc) is that
pronouns “substitute” for nouns, which they “replace” or “stand (in) for” – a notion
that is reflected in the etymology of the English word ‘pronoun’. This definition
can still be found in current linguistic encyclopedias and dictionary entries. 4 We
have already observed that the central pronouns occupy the syntactic positions of
proper nouns and noun phrases. For example, if you start with the grammatical
sentence The tallest mountain is in Nepal, and replace the subject with the pronoun
it, the result is a grammatical sentence: It is in Nepal. This feature, however, does
not distinguish pronouns from indexical phrases like this mountain, nor obviously
from proper nouns and noun phrases. This mountain and the mountain can both
substitute for the tallest mountain in the economist’s sense of being used instead
of it for a particular purpose. Moreover, some personal pronouns cannot occur as
subjects of predication, such as my and your.
It is often added that the function of pronouns is to abbreviate nouns and avoid
repetition, an idea dating back to Isodore of Seville (c. 570–636).
(9) a. The United States is a democracy, and the United States has a president.
b. The United States is a democracy, and it has a president.
c. The United States is a democracy, and the US has a president.
(9b) and (c) are both shorter than and avoid the repetition of (a). But only the US
is an abbreviation, and it is not a pronoun. In conformity with the notion that
pronouns are abbreviations, it is sometimes added that pronouns have the same
meaning as the noun they substitute for (Kullavanijaya 2005: 878). But unlike ‘the
US’, ‘it’ is never synonymous with ‘the United States’, even when that is its antecedent.
Although its reference would change if its antecedent were ‘France’, the meaning
of ‘it’ would not change. On the grounds that pronouns can “abbreviate” any of
a wide range of nouns, authors like Kullavanijaya (2005: 878) say that pronouns
have “no meaning in themselves”. Being indexical, pronouns have no referents “in
themselves” (that is, independently of a particular context of use), but they do have
distinct meanings that remain invariant from context to context. Suppose someone
points at a point of light in the sky and asks “What is that?”
(10) a. It is the morning star.
b. Venus is the morning star.
4. See for example Sportiche 2003: 407; Kullavanijaya 2005: 877; Saxena 2006: 131; Harley
2006: 194; Quirk et al. 1985: 76, 335, 858–9, 864–5. For histories, see Dinneen 1967: 101; Michael
1970: 69.
46 Wayne A. Davis
The sentences in (10) differ markedly in meaning. We can believe (a) without be-
lieving (b). Accounting for the difference is the “Problem of the Essential Indexical”
(Perry 1979), which is pressing for many theories. It in (a) is not an abbreviation of
Venus even if they have the same referent.
Pronouns can be defined as indexical words that can be used anaphorically with
nouns (or noun phrases) as their antecedents. Indexicals like this nation can be used
the same way, but are not pronouns because they are phrases. Proper names like
Venus, definite descriptions like the morning star, and abbreviations like the US
cannot be used anaphorically because they are not indexicals. Ditto, do so, thus, and
such can be used anaphorically, but they do not count as pronouns because their
antecedents cannot be nouns. Their antecedents instead must be sentences, verbs,
adverbs, and adjectives, respectively. They have been called “pro-sentences”, “pro-
verbs”, “pro-adverbs”, and “pro-adjectives”, which, together with pronouns, comprise
the general class of “pro-forms”. Note that even though pronouns are defined by the
fact that they can be used anaphorically, they need not be, as we illustrated with (1).
The only pronouns that must be used anaphorically are the reflexive pronouns, like
herself, and possibly the “logophoric” pronouns in certain African and Australian
languages (Bhat 2004: Chapter 3). The anaphoric use of pronouns is no more basic
or fundamental than the others.
6. The binding rules
Different pronouns have different constraints on their interpretations. In (11), for

example, the reflexive himself must have Hemingway as its antecedent, but the non-
reflexive him cannot.
(11) a. Hemingway shot himself.
b. Hemingway shot him.
In contrast to (11), Hemingway can be the antecedent of him in (12b), but not of
himself in (12a).
(12) a. John admired Hemingway. *Mary admired himself too.
b. John admired Hemingway. Mary admired him too.
Chomsky (1995: Section 1.4.2) accounted for such facts in terms of syntactic rules:
Rule A: A reflexive pronoun must be bound in its local domain.

Rule B: A nonreflexive pronoun must be unbound in its local domain. 5
5. Chomsky used ‘anaphor’ for reflexive pronouns, and ‘pronominal’ or ‘pronoun’ for
nonreflexives.
The local domain will here be characterized as the minimal governing clause (in-
cluding adjuncts) or nominal phrase.
As Chomsky (1995: 93, 96) defines binding, α binds ß only if they are “coindexed”
or “coreferential”. The binding conditions are thus sometimes formulated as saying:
a reflexive pronoun must corefer with an NP in its local domain, while a nonreflexive
pronoun must not. In its most common sense, adopted by Chomsky, ‘coreference’
means having the same referent, which is a semantic notion. In this sense, however,
Hemingway and him may perfectly well be coreferential in (11b), if only by accident.
This sentence would be fully grammatical if the speaker used him demonstratively
while pointing at a picture of Hemingway (knowingly or unknowingly), or if him
were anaphoric on the author of For Whom the Bell Tolls in another sentence.
Chomsky (1995: 96) initially defined coindexation as meaning anaphoric link-
age. A common index would thus indicate that one of the terms is the antecedent
of the other. If two singular terms are anaphorically linked, then they have the same
referent; but the converse fails. While Hemingway and him may have the same ref-
erent in (11b), they cannot be anaphorically linked. Him can refer to Hemingway
(the man) in (11b), but it cannot refer back to Hemingway (the name). The binding
conditions must therefore be formulated as saying: a reflexive pronoun must be an-
aphorically linked to an NP in its local domain, while the nonreflexive pronoun must
not be. It will be useful to indicate anaphoric links by superscripts, to distinguish
them from the referential identities indicated by subscripts. Then the grammatical
and ungrammatical readings of (11b) can be represented by (13).
(13) a. 1Hemingway1 shot 2him1.
b. 1Hemingway1 shot 2him2.
c. *1Hemingway1 shot 1him1.
The distinction between coreference in general (having the same referent) and
anaphoric coreference (having the same referent because of an anaphoric link) will
be critical in what follows.
Rules A and B hold for a wide variety of syntactic structures in English and
other languages. They do not hold for all personal pronouns, structures, or languag-
es, however. For example, some dialects of English allow nonreflexive pronouns to
have local antecedents where standard dialects do not. Thus 1Buck got 1him some
cigarettes is grammatical in Appalachian dialects, though not in standard British
or American dialects (Montgomery and Hall 2004: Section 2.3.1, Section 12.1).
And in standard English dialects, reflexives are permissible in sentences like Mary
ignored the blood on her, in which the local domain is a nominal phrase and the
reflexive is bound by the subject of the local clause. While Rules A and B do not
hold universally; they hold for a wide variety of structures in English and many
other languages. The non-universality of the rules casts no doubt on the fact that
they hold for structures like Hemingway shot __ (contrast Huang 2004: 292).
48 Wayne A. Davis
On my theory, Rules A and B hold where they do because reflexive and non-
reflexive pronouns have different determinant constraints, which is true because
the concepts they express have different determiner components. Levinson and
Huang propose instead to account for the data pragmatically, in terms of gener-
alized conversational implicatures and Neo-Gricean principles. We will examine
their account after introducing implicature.
7. Implicature
Implicature denotes a type of meaning. Contrast the following sentences, both of

which, let us assume, are true.
(14) John has a temperature means that John’s temperature is above normal. (Sentence
Meaning)
(15) By (saying) John has a temperature, Steve meant that John has a thermometer.
(Speaker Meaning, Cognitive)
Sentence meaning is a property of expressions rather than of speakers. It depends

on conventional usage and the syntax and semantics of the language, not on any
particular speaker or utterance. Consequently (15) may be true even though (14) is:
Steve may have used temperature either by mistake or purposely with an unconven-
tional meaning. Whether Steve meant that John has an above normal temperature
or a thermometer depends on what Steve intended. On my view, to mean that p is
to express the belief that p, which is to do something as an indication or natural sign
that one has that belief. The fact that natural signs and indications can exist even
when they are not recognized by anyone allows people to mean something even
when speaking to foreigners and animals. That indications need not be completely
reliable allows people to express beliefs they do not have, and thereby lie or mislead.
Speaker and sentence meaning may diverge even when the speaker makes no
mistake and is using words with their conventional meanings. Imagine the follow-
ing dialogue.
(16) Sue: Can John play?
Steve: John has a temperature.
If this was a typical exchange, Steve meant that John cannot play. But the sentence
he uttered means something very different. Hence Steve did not say that John can-
not play, he implied it. Speaker implication involves meaning that one thing is the
case by meaning that something else is. Grice (1989: 24) introduced the technical
term implicate for a closely related speech act: meaning or implying one thing by
saying another. Thus Steve implicated that John cannot play. If we know only what
Steve has said, we would not realize that he had answered Sue’s question. Implying
and implicating are what Searle (1975: 265–6) called indirect speech acts. If we do
not know whether Steve answered directly or indirectly, we may not know how to
evaluate him. For example, if Steve insincerely said John could not play, he lied. If
he insincerely implicated the same thing, he misled Sue without lying.
In the definition of implicature, saying means not the mere utterance of words,
but saying that something is the case, an illocutionary speech act like stating and
affirming but more general. What Steve said is that John has a temperature, some-
thing he could have said by uttering different words, perhaps in French. Say is to be
interpreted strictly in the definition of implicature, requiring that what a speaker
says be something that the sentence uttered conventionally means (modulo indexi-
cals and ellipsis). Stating or asserting something entails both saying and meaning it.
Steve’s implicature is conversational or pragmatic. It is not part of the meaning of
the sentence uttered, but depends on features of the conversational context. Had Sue
asked “Is John well?”, Steve could have implicated something completely different
(that he is not well) by saying the same thing. A semantic implicature is part of the
meaning of the sentence used. 6
(17) a. Washington, a Virginian, was the first president.
b. Washington was a Virginian.
Speakers implicate (17b) when they use (17a) literally. They mean, but do not say,
that Washington was a Virginian. Hence the literal use of (17a) while disbelieving
(17b) would be misleading, but not lying. Steve’s sentence in (16) can be used lit-
erally with its conventional meaning without his implicature. Thus Steve could co-
herently have added but he is well enough to play. In contrast, (17a) cannot be used
literally with its conventional meaning without implicating (17b); but Washington
was not a Virginian cannot be added. Grice (1989: 25) used the term ‘conventional’
for implicatures that are semantic in this way. This was unfortunate, because in
the sense characterized in Section 1 (“customary practice”), some nonsemantic
implicatures are also conventional.
Like mean, imply and implicate also have senses in which they apply to sentenc-
es. All semantic implicatures are sentence implicatures. Washington, a Virginian,
was the first president implicates that Washington was a Virginian. The implicature
is carried by the appositive construction. The implicature of p but q is carried by the
meaning of but. As a first approximation, a sentence has an implicature when speak-
ers conventionally use sentences of that form with the corresponding implicature. 7
6. Grice 1989: 25; Neale 1992: 523–9; Potts 2005, 2007; Huang 2014: Section 2.5; Davis 2014:
Section 2.
7. See Davis (2016c: Section 4) for a more precise characterization.
50 Wayne A. Davis
The conversational implicature we have examined is not a sentence implicature.

John has a temperature does not itself imply or implicate that John cannot play. It is
conventional to use sentences to answer questions by implicature in this way. But
sentences with the particular form N has a temperature are not customarily used
to implicate that N cannot play. Neither the particular words in the sentence, nor
the way they are composed, carries the implicature. This is what Grice (1989: 37–8)
called a particularized implicature. Other nonsemantic implicatures are different.
Consider Dan’s reply in (18):
(18) Cal: Who ate all the cookies?
Dan: Mickey ate some of the cookies.
In a typical context, Dan would have implicated Mickey did not eat all the cookies.
This is what Grice (1989: 37–8) called a generalized conversational implicature. The
use of sentences containing some to implicate not all is common in a wide range
of contexts. As a result, Mickey ate some of the cookies itself implicates Mickey did
not eat all of the cookies. In particular contexts, sentences of this form can be used
with other implicatures. Dan could have meant that Mickey ate none of the cookies
(irony), or even that he ate all of them (understatement). But those implicatures
are not associated with a feature of the sentence, and are not generally conveyed
when the form is used.
This relation between some and all is special. Contrast several. Because Mickey
ate several of the cookies is stronger, speakers could use Mickey ate some of the
cookies to implicate that Mickey did not eat several cookies. (Imagine the sentence
uttered in response to “Did Mickey eat several of the cookies?”) But Mickey ate
some does not itself implicate that Mickey did not eat several. Sentences containing
some are not customarily used to implicate not several. Generalized implicatures
like Dan’s are cancelable, unlike implicatures that are conventional in Grice’s sense.
Nonetheless, it is conventional (customary) to use an affirmative sentence ‘Σ(some)’
containing some with its usual meaning to implicate the proposition expressed by
‘Σ(not all)’, the sentence obtained from ‘Σ(some)’ by replacing some with not all.
And the use of ‘Mickey ate some of the cookies’ to implicate that Mickey did not eat
all of the cookies is an instance of that practice. So this implicature is, in an ordinary
but non-Gricean sense, a conventional conversational implicature. Steve’s particu-
larized implicature is conventional in neither sense. The arbitrariness characteristic
of conventions is shown by the fact that some implicates not all and not many, but
not not several or not only some.
8. Neo-Gricean explanations of the binding rules
Levinson and Huang hypothesize that the data summarized by Rule B is explained
by a version of Grice’s maxim of Quantity, “Make your contribution as informative
as is required for the purposes of the conversation.”
Given this principle, the use of a semantically weaker pronoun where a seman-
tically stronger reflexive could occur gives rise to a conversational implicature
which conveys the negation of the more informative, coreferential interpretation
associated with the use of the reflexive…. (Huang 2006: 233–4)
The alternates 〈himself, him〉 form a clear contrast set, one member more in-
formative than the other, as in a Horn scale, such that the use of the weaker him
Q-implicates the inapplicability of the stronger himself. (Levinson 2000: 287)
The following is a typical derivation of Dan’s implicature.

Dan has said that Mickey ate some of the cookies. If Dan was in a position to make
the stronger statement that Mickey ate all of the cookies, Cal’s question required
him to do so. Otherwise Dan would have violated the maxim of Quantity. Dan
therefore must have believed and meant that Mickey did not eat all the cookies.
In the same way, Levinson and Huang hypothesize that Hemingway shot him im-
plicates Hemingway did not shoot himself.
Neo-Gricean accounts of the binding rules face insuperable problems. 8 For
starters, implicatures are not derivable from conversational principles for a vari-
ety of reasons (Davis 1998; Lepore and Stone 2015). If the above derivation were
sound, for example, we could just as well predict that Mickey ate some of the cookies
implicates Mickey did not eat several of the cookies or Mickey did not eat only some
of the cookies.
Dan has said that Mickey ate some of the cookies. If Dan was in a position to
make the stronger statement that Mickey ate several of the cookies, Cal’s question
required him to do so. Otherwise Dan would have violated the maxim of Quantity.
Dan therefore must have believed and meant that Mickey did not eat several the
cookies.
But some does not implicate not several or not only some the way it implicates not
all or not many. Similarly, if the Q-principle accounted for why him cannot have
Hemingway as its antecedent in Hemingway shot him, it should also predict that her
8. I provide fuller account in Davis (2016a).

52 Wayne A. Davis
cannot have Mary as its antecedent in Mary ignored the blood on her. And it should
predict that Hemingway shot him implicates Hemingway did not shot someone else.
The claim that a sentence of the form Σ(him) implicates –Σ(himself) is also
incorrect. An utterance of Hemingway shot him would not generally implicate, or
be interpreted as implicating, that Hemingway did not shoot himself. In different
utterance contexts, him refers to different people. So suppose Mary used him to refer
to Fitzgerald. Then by uttering (11b), Mary said that Hemingway shot Fitzgerald. By
doing so, she would not imply or suggest that Hemingway didn’t shoot himself. This
is particularly clear if Mary, like the rest of us, knows that Hemingway committed
suicide. Suppose, on the other hand, that Jane used him pointing at a picture of
Hemingway. Then Jane has said that Hemingway shot Hemingway. Unless she were
being ironic, Jane could not thereby implicate that Hemingway didn’t shoot himself.
Furthermore, Rule B holds not only for affirmative sentences like (11b), but
also for negative sentences like (19a).
(19) a. Hemingway did not shoot him. Σ(him)
b. Hemingway did not not shoot himself. −Σ(himself)
When (19a) is Σ(him), it is completely obvious that it does not implicate –Σ(himself),
which is equivalent by double negation to Hemingway did shoot himself. If (19a)
did have that implicature, it would not be a Q-implicature. So the fact that him
cannot have Hemingway as its antecedent in (19a) clearly has nothing to do with a
Q-implicature that arises from the fact that the speaker used him rather than himself.
In order for −Σ(himself) to be a Q-implicature of Σ(him), and 〈himself, him〉
to be a Horn scale, the proposition expressed by Σ(himself) must be stronger than
the proposition expressed by Σ(him). That is, Σ(himself) must entail Σ(him) but not
conversely. Σ(himself) and Σ(him) are never related in this way. When himself and
him have different referents, the propositions expressed are independent. When the
pronouns have the same referent, the propositions are equal in strength.
The proposition that Hemingway shot himself is stronger than the proposition
expressed by (20a) when him has some man as its antecedent:
(20) a. Hemingway shot himself.
b. 1Some man is such that Hemingway shot 1him.
(20b) illustrates the “bound variable” use of pronouns. On the indicated interpre-
tation, the proposition expressed by (20b) is equivalent to the proposition that
Hemingway shot some man, and is clearly entailed by (20a). On this interpretation,
(20b) is like (20a) and unlike Hemingway shot him in expressing the same propo-
sition in every context of use.
Another reason why Rule B cannot be explained pragmatically is that conver-

sational implicatures are cancelable. A sentence with a generalized conversational
implicature can be used without that implicature. For example, Some passengers
survived lacks the typical scalar implicature when it is followed by Indeed, all sur-
vived. But the fact with which we began is that ‘him’ cannot have Hemingway as its
anaphoric antecedent in Hemingway shot him, whereas ‘himself ’ must have that
antecedent in Hemingway shot himself. A cancelable implicature generated by him
is incapable of explaining why him cannot occur in certain contexts.
We noted in Section 6 that Rules A and B do not account for the behavior
of himself and him in all syntactic contexts, not to mention all pronouns in all
languages. But nothing has cast doubt on the rule that himself must have a local
antecedent in (11a) while him cannot have one in (11b). The same is true for
countless other instances of Σ(himself) and Σ(him). Whether Rules A and B hold
for a structure Σ(_) depends entirely on the syntax and semantics of Σ(_), not on
the non-linguistic context of utterance. That the rules are not linguistic univer-
sals does not make them cancelable the way implicatures and other pragmatic
features are.
9. Pronoun implicatures arising from their sortals
In Section 8, I summarized my argument that the binding rules cannot be derived

from general conversational principles or attributed to differences in implicature.
Nonetheless, pronouns do generate a variety of implicatures. Some are due in part
to the sortal components of the concepts they express, and some to the determiner
components. Other implicatures are independent.
Many languages have gender as a grammatical feature in addition to number
and case. Pronouns in these languages must agree in gender with their noun ante-
cedents. In French, il (“he”) and bras (“arm”) are masculine, while elle (“she”) and
main (“hand”) are feminine. So il can refer back to mon bas but not mon main,
while elle can refer back to mon main but not mon bras. English pronouns appear
to be similar. Consider (21).
(21) a. The woman came and he conquered.
b. Alice came and he conquered.
c. The robot came and he conquered.
Sentence (21a) is perfectly grammatical if he is used demonstratively. The sentence

is defective in some way, however, if the woman is the anaphoric antecedent of he.
The sentences in (21) all appear similar to the ungrammatical sentences in (22).
54 Wayne A. Davis
(22) a. The woman came and they conquered.

b. The woman are tall.
c. Her is tall.
(22a) is ungrammatical because of the lack of number agreement between the plural
pronoun and the singular antecedent, (22b) because the plural are does not agree
with the singular woman, (22c) because the subject has the wrong case. It may
appear that the sentences in (21) are similarly ungrammatical because the gender
of the pronoun does not agree with the gender of its antecedent. But English does
not have this kind of grammatical gender, even on pronouns.
Sentence (21b) would typically share the same defect as (21a). For Alice is a
woman’s name. As a result, it may be hard to think of a context in which (21b)
could be properly used. But (21b) would be perfectly proper if Alice referred to
Alice Cooper, the (male) rock star. Indeed, it is improper to refer to Alice Cooper
as she. Similarly, even if the pronoun is anaphoric on the name, the conditional If
Alice is a father, he has a child is grammatical. Finally, while no amount of pretense
will make (22a) the right thing to say, it is proper to use he to refer to a woman if we
are pretending that she is a man. So the sentence (21b) is not ungrammatical even
if the masculine pronoun he is anaphoric on the woman’s name Alice.
Consider next (21c). An electro-mechanical robot without even animalian
form cannot be male or female. This impossibility is not enough to make (21c)
ungrammatical when The robot is the antecedent of he. A speaker who believed
that the robot was male, or was pretending it was, could properly use he to express
beliefs about it. And a conditional like If the robot is a father, he has children is
grammatical and true if he is refers back to the robot. These facts show that English
does not have a “natural” gender system either, one in which the grammaticality of
a pronoun depends on the sex of the referent of its antecedent. We will see further
problems for the natural gender thesis in §11.
The impropriety illustrated by (21) is semantic and pragmatic rather than syn-
tactic. (21b) presupposes Alice is male in the same sense that it presupposes Alice
exists. If either presupposition is false, (21b) is not true. (23b) has the same defect
if the man has Alice as its antecedent.
(23) a. The woman came and the man conquered.
b. Alice came and the man conquered.
c. The robot came and the man conquered.
On my view, sentences containing pronouns have presuppositions because of the

sortal components of the concepts expressed by the pronouns. By rule (6), ℞δ has a
referent only if δ has a referent satisfying the sortal concept in ℞. He has a referent
on any given occasion only if the indexical concept it expresses there does. So if
the antecedent of he on a given occasion is Alice, and the referent of Alice there is
female, then he has no referent. Since (21b) has a presupposition that is true only
if he has a referent, it presupposes that Alice is male.
He and she differ semantically in a way that Alice and Alan do not. The con-
ventions regulating which names are given to males and females are naming
conventions, not the usage conventions that determine lexical meaning. Naming
conventions are like word-formation conventions in this respect. The fact that Alice
is conventionally given to females does not mean that Alice expresses a concept
containing the sortal c(female). Indeed, Kripkean arguments against description
theories of proper names rule out the possibility that the concepts expressed by
proper names contain any but the most general sortals, such as c(individual object)
or c(concrete object) (Davis 2005: Chapter 13). So Alice came does not presuppose
that Alice is female even if Alice is the name of a woman.
The two presuppositions of sentence (21b) (repeated as (24)) are implicatures
(§7). The presuppositions are things meant but not said, implied but not asserted.
Let “⊐” symbolize implicature.
(24) Alice came and he conquered ⊐
a. Alice exists.
b. Alice is male.
c. He exists.
d. He is male.
e. ‘Alice’ has a male referent on this occasion.
f. ‘He’ has a male referent on this occasion.
A man who utters (24) does not say that Alice is male any more than he says that
Alice exists. Yet by saying that Alice came and he conquered, the man expresses
the belief and means that Alice is a man just as he expresses the belief and means
that Alice exists. Moreover, these two implicatures of (24) are generalized impli-
catures – implicatures of the sentence. Sentences of the form ‘N1 VP1 and he1 VP2’
are conventionally used to implicate both that N exists and that N is male. This is
true in part because of the meaning of he. The (b) implicature differs from the (a),
however, in being cancelable. Since it is carried by the pronoun he in virtue of its
meaning, the (b) implicature may appear to be purely semantic – like (a), another
classic case of what Grice called conventional implicature. Yet it is permissible to
use (24) with its conventional meaning but without implicating that Alice is male.
The cancellation would typically be implicit, as when the speaker points at Alan
when uttering he, using the pronoun demonstratively rather than anaphorically.
It would typically have contrastive stress. The implicature could also be canceled
explicitly, as in (25):
56 Wayne A. Davis
(25) Alice came and he conquered – Alan conquered Alice, that is.
The sequent in (25) clarifies that he refers to Alan, not Alice. The speaker used Alice
came and he conquered with its conventional meaning, but without implicating that
Alice was a male.
Since the (b) implicature depends on features of the context of use, it is a con-
versational or pragmatic implicature. However, it differs markedly in one respect
from the generalized implicatures extensively studied by linguists and philosophers.
It is uncancelable in typical cases. The most typical use of the third-person pronoun
in a sentence like (24) is the anaphoric, with Alice the antecedent. Alice is male is
presupposed when, and only when, Alice is the anaphoric antecedent of he in (24).
And when it is presupposed, the implicature is uncancelable. The clauses of Alice
came and he conquered, but Alice is not male are inconsistent in that case. If he were
anaphoric on Alice in (25), its sequent would be confusing rather than clarifying.
That is would be totally out of place, whereas too or however would make sense.
While the existence of a use that makes a generalized conversational implicature
uncancelable is a special case, it is not special to the pronoun he. It is true of all
pronouns, on my theory, because they express indexical concepts that (i) have sortal
components and (ii) are able to have concepts expressed by other referring terms
as their determinants.
Sentence (24) also has two implicatures that are conventional in Grice’s sense:
(c) and (d). (24) implicates that he (the object the speaker used he to refer to) exists
for the same reason it implicates (a): because he is the subject of a conjunct in (24),
not because of anything specific to its meaning. (24) implicates that he is male,
on the other hand, because of the specific meaning of he – on my view, because
he expresses a concept whose sortal constituent is c(male). Implicatures (c) and
(d) are uncancelable: A speaker cannot use (24) with its English meaning without
implicating that he (the conquerer) exists and is male.
Implicatures (a)–(d) of (24) are objectual: implicatures about the objects the
speaker said something about. Implicatures (e) and (f) are metalinguistic: implica-
tures about the words the speaker used. In addition to expressing the belief that the
person he is talking about (Alice) is male, a speaker using (24) will also commonly
express the further belief that the name ‘Alice’ and the pronoun ‘he’ have a male
referent. (24b) and (e) are different, though related, implicatures, as are (d) and (f).
10. Pronoun implicatures arising from their determiners
Example (11b) (repeated as (26)) illustrates another type of generalized implica-

ture carried by third-person non-reflexive pronouns. Identical subscripts represent
coreference as in Section 6.
(26) Hemingway shot him1.
a. ⊐ He1 is not Hemingway.
b. ⊐ He1 is not me (the speaker).
c. ⊐ He1 is not you (the addressee).
Because (26) contains a non-reflexive third-person pronoun, it would typically be

used with the three implicatures indicated because of the semantic contrast between
non-reflexive and reflexive pronouns, and between third-person pronouns and
first- and second-person pronouns. This if I say (26) to you, I will implicate “He is
not Hemingway”, referring to the referent of him in (26). I will also implicate “He is
not me” and “He is not you”. These are implicatures of the sentence Hemingway shot
him because sentences of the form ‘N V him’ are conventionally used to implicate
“He is not N”, “He is not me”, and “He is not you”. These implicatures are not part
of the meaning of (26), however, so they are not Gricean conventional implicatures.
They can be canceled. The (a) implicature could be canceled implicitly by pointing
at an easily recognized picture of Hemingway while uttering him. The (b) implica-
ture could be canceled explicitly by adding Unfortunately, he was me immediately
after (26). The (c) implicature could be canceled in the same ways.
The (b) and (c) implicatures of (26) exist because speakers customarily use
first- rather than third-person pronouns to refer to themselves (Caesar was a noted
exception), and second- rather than third-person pronouns to refer to their ad-
dressees. The (a) implicature holds because third-person non-reflexive pronouns
are generally not used to refer to the same person that other singular terms in the
local clause refer to. This is in part because of Rule B (Section 6), which reflects
a constraint on the indexical concepts expressed by non-reflexive pronouns. The
determinants expressed by non-reflexive pronouns cannot be other concepts in
the proposition expressed by the local clause. (26a) is also due to the fact that
speakers would generally use the reflexive himself rather than him in a sentence
whose subject is Hemingway if they want to refer to Hemingway. This choice fa-
cilitates understanding, for ‘himself ’ has to refer to Hemingway (Rule A). ‘Him’
could refer to Hemingway, but the hearer would have to figure that out from the
nonlinguistic context.
On my account, it, this, and that have the same sortal component, c(thing), the
concept expressed by thing in its most general sense, in which everything is a thing.
Entity, being, and object are synonyms. It, this, and that differ in meaning because
58 Wayne A. Davis
they express concepts with different determiners, hence different determinant con-
straints. Contrast (27a) with (b):
(27) a. The old man caught this and ate it.
b. The old man caught this and ate that.
On the most natural interpretation of (27a), it is used anaphorically with this as its
antecedent. (27b) cannot be interpreted the same way. That cannot have this as its
antecedent in (b). The pronouns in (27b) are most naturally used demonstratively
to refer to different fish. It in (27a) cannot be used demonstratively. So c(it) and
c(this) have different determinant constraints. Their determiner components allow
them a different range of determinants. Note that ‘this’ and ‘that’ can be used an-
aphorically, as in Napoleon met Wellington at Waterloo; this is where he was finally
defeated, with Waterloo the antecedent of this. Only certain anaphoric uses are
excluded. Note too that (27b) does not entail that the referents are different objects.
The speaker might have used this while pointing directly at the fish, while using that
while pointing at its reflection in a mirror. This and that themselves have different
determinant constraints. For example, a speaker can use This is me when focusing
visually on his body, but not when focusing visually on his reflection in a mirror.
The possibilities are reversed with That is me, unless the speaker were having an
out of body experience.
Because of this difference between it and this, (27a) and (b) have different
implicatures, both objectual and metalinguistic.
(28) a. The old man caught this and ate it.
⋣ It (what he ate) is not this (what he caught).
⋣ “This” and “it” have different referents on this occasion.
b. The old man caught this and ate that.
⊐ That is not this.
⊐ “This” and “that” have different referents on this occasion.
(28b) has these implicatures because speakers cannot use that to refer back to this,
and would generally use it anaphorically to refer to the same object. A speaker
could use that demonstratively to refer to the same object, but that would require
an independent demonstrative act. Only in special contexts are hearers likely to
interpret this and that in (b) has referring to the same object. (28a) lacks the parallel
implicature because it is most naturally used with this as its antecedent, so that both
have the same referent.
11. Independent pronoun implicatures
The English pronoun it is often classified as neuter because of the way it contrasts
with he and she. Compare:
(29) a. The centurion came and he conquered.
b. The centurion came and it conquered.
(29a) is appropriate if the centurion is the commander of a Roman army, but not if
it is a woman or a battle tank. (29b) is acceptable if the centurion is a tank but not
if it is a woman or man. The statements prompted by (30) are similarly improper:
(30) The pitcher is a twelve year old child.
a. It is a girl.
b. It is named Mary.
c. It is left-handed.
These examples show that just as he and she differ from il and elle in not being gram-
matically masculine or feminine, so it is not like the Latin id in being grammatically
neuter. The English nouns centurion and pitcher have no grammatical gender.
Whereas ‘him’ in (26) cannot have ‘Hemingway’ as its antecedent, ‘it’ in (30) can
have ‘the pitcher’ as its antecedent. In this respect, it again differs from this and that.
(31) The pitcher is a twelve-year old child.
a. This is left-handed.
b. This child is left-handed.
This in (31) can be used to refer to the pitcher, but only demonstratively, and only
with the impropriety of (30c). This cannot have the pitcher as its anaphoric an-
tecedent. This child differs markedly, for it can have the pitcher as its anaphoric
antecedent, with no impropriety.
The impropriety of (30) is not ungrammaticality, and is not due to a determi-
nant constraint. To what then is it due? A natural first hypothesis is that the sortal
component of c(it) excludes those of c(he) and c(she), and so applies only to things
that are neither male nor female – perhaps c(inanimate). But it can refer to male or
female animals, as in (32):
(32) A bird got into the attic. It made a nest and laid eggs.
Only female birds lay eggs. So if the sortal component excluded c(female), the sec-
ond sentence in (32) would have a false presupposition. There is no inconsistency
or impropriety in (32). The possible truth and propriety of When the rooster woke
up, it woke everyone else up similarly shows that the sortal component of c(it) cannot
exclude c(male).
60 Wayne A. Davis
Another semantic hypothesis is that the sortal component of c(it) is c(non-

person), making ‘it’ impersonal rather than neuter. 9 c(nonperson) is preferable to
c(nonhuman) because the use of it to refer to God or Superman is for most people
equally inappropriate. God and Superman are assumed to be persons, but nonhu-
man, indeed superhuman. Nonhuman animals are in contrast subhuman. However,
‘it’ can be used to refer to persons, as can ‘this’ and ‘that’. The questions in (33) are
appropriate and natural.
(33) Kim is a newborn baby.
a. Is it a boy or a girl?
b. What is it named?
We can see that the sortal component of c(it) has to be the most general concept
c(thing) whose extension includes everything by observing that the question in (34)
can be answered by filling in the blank with any common-noun phrase without
ungrammaticality or presupposition failure.
(34) What is it?
It is a/an ___.
The answer might be It is a man in armor, It is a girl, It is an elephant, It is an apple,

or It is a prime number. So the impropriety of (30a)–(c) is not due to the sortal com-
ponent of c(it) the way the impropriety of (21c) is due to the sortal component of
c(he). The propriety of the uses of it in (33) and (34) shows that the determiner com-
ponent of c(it) similarly fails to account for the impropriety of (29b) and (30a)–(c).
An alternative hypothesis is suggested by the impropriety of (35a) and (b):
(35) The violinist plays in the New York Philharmonic.
a. The thing is old.
b. The primate is not old.
We find the same impropriety when the is replaced by this or that in (35a) and (b).
(35a) and (b) have no false presuppositions. Violinists are things (in the general
sense), and all are primates. So the sentences have no semantic defect in the state-
ments, and one is sure to be true. I believe the defect is pragmatic. Both sentences
implicate that the referent is a nonperson. Someone who uttered (35a) would not
say that the thing (the violinist) is a nonperson, he would imply that by saying that
9. Cf. Wiese (1983: 392), who hypothesizes that the meaning of ‘it’ is “non-personal”; Quirk et
al. (1985: 314), who say that it expresses the “natural gender” [nonpersonal]; Bosch (1988: 216),
who claims that it refers to non-persons; and Corbett (1991: 12), who says that “male humans are
masculine (he), female humans are feminine (she) and anything else is neuter (it)”.
the thing is old. The choice of a singular term that could refer to inanimate beings
rather than one restricted to animate beings implicates that the referent is inan-
imate, and therefore not a person. Similarly, using primate rather than the more
specific human implicates that the referent is nonhuman, and presupposes that it
is a primate and thus not superhuman.
Generalization (36) holds for wide variety of general common nouns and com-
mon-noun phrases that apply to nonpersons as well as persons.
(36) The G+ is P ⊐ The G+ is a nonperson, provided ‘P’ does not exclude “nonperson”.
‘G+’ includes hominid, primate, mammal, animal, organism, living being, ma-
terial object, and thing, plus more specific common-noun phrases like bipedal
animal.
Not all common nouns applying to both nonpersons and persons carry the same
implicature, including referent and subject.
(37) The G– is P ⋣ The G– is a nonperson, even if ‘P’ does not exclude “nonperson”.
‘G–’ includes referent, subject, element, and common-noun phrases like first
referent and second element of a particular ordered pair.
Consider: Does ‘The current president of Brazil’ have a referent? Yes, the referent is
very popular; or: The experiment showed that dementia could be cured. The subject
was sixty-five years old. I have not attempted to give a complete enumeration of
the common nouns for which (36) and (37) hold. The proviso on (36) is necessary
because the predicate will often cancel the implicature carried by the subject. The
thing is a person does not implicate that the thing is a nonperson. It would be an odd
response to (35), to be sure. The subject suggests something false but the predicate
overrides the suggestion by entailing something already implied by (35). But the
response does not seem improper. Knowledge of the facts represented by (36) and
(37) is essential for full mastery of the terms represented by G+ and G−.
The arbitrariness of the nonperson implicature has an analog in scalar impli-
catures. As discussed in Section 8, ‘Some S are P’ implicates “Not Q S are P” when
‘Q’ is all or many, for example, but not when ‘Q’ is several or four.
(38) Some S are P ⊐ Not Q+ S are P.
‘Q+’ includes all and many.
(39) Some S are P ⋣ Not Q− S are P.
‘Q−’ includes several and n (for all n > 1).
Another analog is provided by the phenomenon commonly called “neg-raising”.

John does not believe God exists can be used to mean something stronger than the
negation It is not the case that John believes God exists, namely the contrary John
62 Wayne A. Davis
disbelieves that God exists, which entails John believes God does not exist. Many verbs
V+ can be used similarly, including desires, intends, is willing, and is likely. But many
syntactically similar verbs V− do not allow neg-raising, including knows, asserts, is
certain, and is afraid. 10
As we have seen, it behaves like the thing in carrying the implicature nonperson.
But it differs in carrying a second implicature, baby, illustrated by (33).
(40) It is P ⊐ It is a nonperson, provided ‘P’ does not exclude “nonperson”.
It is P ⊐ It is a baby, provided ‘P’ does not exclude “baby”.
The nonperson implicature would be more likely with (29b), while the baby impli-
cature would be more likely with (30a) or (b). Both implicatures are very common,
but in no context are both implicated. A speaker who knows the referent is a baby is
not in a position to say or implicate that it is a nonperson, and vice versa. In contrast
to he and she, the implicatures carried by it are not due to the sortal component of
the concept it expresses. And in contrast to this and that, they are not due to the
determiner component.
It may seem impossible for one sentence to have two implicatures that differ
in this way. But in fact, it is not at all unusual. For example, sentences containing
some with the well-known scalar implicature also have an ignorance implicature.
(41) Some S are P ⊐ Not all S are P.
Some S are P ⊐ It is unknown whether all S are P.
Sentences of the form ‘Some S are P’ are used in countless contexts. They are used
with the scalar implicature in many contexts, and with the ignorance implicature
in many others. But in no context is a sentence containing some used with both
the scalar implicature and the ignorance implicature. Someone who does not know
whether all passengers are dead is not in a position to say or implicate that not
all passengers are dead, and vice versa. One difference between the implicatures
represented by (41) and those represented by (36) and (40) is that the use of it
or the G+ is considered not just potentially misleading but improper when the
implicature is false.
If the connections between it and nonperson and baby are nonsemantic implica-
tures, they should be cancelable, and indeed they are canceled in a variety of cases.
The nonperson and baby implicatures are canceled if the predicate ‘P’ excludes them.
Thus It is a cat does not implicate It is a baby, and It is a six ounce baby girl does not
10. For further discussion of scalar implicature and neg-raising, plus references to the volumi-
nous literature, see Horn (1989) and Davis (2016d), as well as Ariel (this volume). I argue in Davis
(2016d) that the neg-raising interpretations differ from scalar implicatures in being senses, but
both are conventional.
implicate It is a nonperson. It is a senior citizen implicates neither. The implicatures

are also canceled in the identification contexts represented by (34), and even more
clearly in (42), given that who presupposes the referent is a person.
(42) Who is it?
It is __.
Both the baby and nonperson implicatures are canceled for a different reason when
the antecedent of it is the referent or the subject.
(43) The referent of Kim is a primate, so it is in the extension of primate.
The implicatures would also be canceled if (44) were used as an insult.
(44) It asked me to the prom.
When used as an insult, (44) resembles a metaphor, in that the speaker does not
mean what the sentence implicates. That is, the speaker is not expressing the be-
lief, and thus not implicating, that the referent is a nonperson or that the referent
is a baby. The speaker would most likely be expressing the belief and implicating
that the referent is in some way like either a baby or a nonperson, which would
be insulting to a teenager who invited the speaker to the prom. The thing and the
primate could be used to insult in the same way. The not-infrequent use of it to
insult may account for why the uses of it in (30) are considered not only potentially
misleading but inappropriate.
Wales (1996: 159) suggests that it “connotes” the unknown. This suggests that
‘It is P’ has the ignorance implicature It is unknown whether it is a person in addi-
tion to the two indicated in (40). The use of it in the common question What is it?
supports this suggestion. Asking that question generally indicates that the speaker
does not know what the referent is. However, What is it? can always be answered
using it. The person asking the question may have been ignorant, but the person
answering should not be.
Webster’s Encyclopedic Unabridged Dictionary (2001) specifically claims that
it may be used whenever the sex of the referent is “unknown or disregarded”. It
is true that if we do not know whether an animal or baby is male or female, we
will use it. But if we are referring back to the pitcher in Example (30), we cannot
properly say It is left-handed even if we do not know or care about the pitcher’s sex.
(30c) also shows that it does not generally mean or implicate sexless (pace Corbett
1991: 245ff), although (30c) could be used with that implicature as an insult.
64 Wayne A. Davis
12. Interrogative and imperative implicatures
As customary, we have defined implicature as meaning that something else is the

case by saying that something is the case. Thus defined, imperative and interrogative
sentences cannot be used to implicate anything because they cannot be used to say
that anything is the case. But imperative and interrogative sentences also can be
used to mean and express beliefs. Speakers can mean that the answer is yes by ask-
ing “Is the Pope Catholic?” in response to another question. A speaker can express
the belief that he has heard enough by ordering “Don’t say anything more”. So there
is a more general act of meaning that something is the case by saying, asking, or
ordering something else. We can use “p? ⊐ q” to mean that by asking “p?” speakers
conventionally mean that p, and “p! ⊐ q” to mean that by ordering “p!,” speakers
conventionally mean that p. All the pronoun implicatures we have discussed in
this chapter have interrogative and imperative counterparts. Paralleling (24b), we
have (45):
(45) Alice came. Did he conquer? ⊐ Alice is male.
Alice came. Conquer him! ⊐ Alice is male.
Paralleling (26) we have:
(46) Did Hemingway shoot him? ⊐ He is not Hemingway.

Hemingway, shoot him! ⊐ He is not Hemingway.
And paralleling (40), we have (47):
(47) Is it big?/Let it be big!

⊐ It is a nonperson.
⊐ It is a baby.
Finally, interrogatives and imperatives also have presuppositions, as “loaded ques-

tions” make vivid. “Does ___ have an elliptical orbit” has an existential presupposi-
tion whether the blank is filled by Vulcan or it. This defines another great territory
to explore the role that pronouns play in implicit communication.
References
Bhat, Darbhe N. S. 2004. Pronouns. Oxford: Oxford University Press.

Bosch, Peter. 1988. “Representing and Accessing Focussed Referents.” Language and Cognitive
Processes 3: 207–232. doi: 10.1080/01690968808402088
Chomsky, Noam. 1995. The Minimalist Program. Cambridge, MA: MIT Press.
Corbett, Greville G. 1991. Gender. Cambridge: Cambridge University Press.
doi: 10.1017/CBO9781139166119
Davis, W. A. 1998. Implicature. Intention, Convention, and Principle in the Failure of Gricean
Theory. Cambridge: Cambridge University Press. doi: 10.1017/CBO9780511663796
Davis, Wayne A. 2003. Meaning, Expression, and Thought. New York: Cambridge University Press.
Davis, Wayne A. 2005. Nondescriptive Meaning and Reference. Oxford: Oxford University Press.
doi: 10.1093/0199261652.001.0001
Davis, Wayne A. 2016a. “Pronouns and Neo-Gricean Pragmatics.” In Interdisciplinary Studies in
Pragmatics, Culture and Society, ed. by Alessandro Capone and Jacob Mey, 137–78. Cham:
Springer. doi: 10.1007/978-3-319-12616-6_5
Davis, Wayne A. 2016b. “A Theory of Saying Reports.” In Indirect Reports and Pragmatics:
Interdisciplinary Studies, ed. by Alessandro Capone, Ferenc Kiefer, and Franco Lo Piparo,
36–50. Cham: Springer doi: 10.1007/978-3-319-21395-8_15
Davis, W. A. 2013. Indexicals and de Se Attitudes. In Attitudes ‘de Se’: Linguistics, Epistemology,
Metaphysics. ed. by. N. Feit and A. Capone, 29–58. . Palo Alto, CA: CSLI Publications.
Davis, W. A. 2014. Implicature. In The Stanford Encyclopedia of Philosophy. ed. by. E. N. Zalta, Palo
Alto, CA: CSLI Publications. https://benjamins.com/#catalog/journals/ps.8.2.02hau/details
Davis, Wayne A. (2016c) “Implicature.” In The Oxford Handbook of Pragmatics, ed. by Sanford
Goldberg. Oxford: Oxford University Press. doi: 10.1017/CBO9780511663796
Davis, Wayne A. (2016d) Irregular Negatives, Implicatures, and Idioms. Cham: Springer.
doi: 10.1007/978-94-017-7546-5
Davis, Wayne A. (forthcoming) “The Property Theory and de se Attitudes.” In Reference and
Representation in Thought and Language, ed. by Maria de Ponte. Oxford: Oxford University
Press.
Dinneen, Francis P. 1967. An Introduction to General Linguistics. New York: Holt, Rinehart, and
Winston.
Grice, Herbert P. 1989. Studies in the Way of Words. Cambridge, MA: Harvard University Press.
Harley, Heidi. 2006. English Words: A Linguistic Introduction. Oxford: Blackwell Publishing.
Horn, Laurence R. 1989. A Natural History of Negation. Chicago: University of Chicago Press.
Huang, Yan. 2004. “Anaphora and the Pragmatics-Syntax Interface.” In The Handbook of
Pragmatics, ed. by Laurence R. Horn, and Gregory Ward, 288–314. Oxford: Blackwell.
Huang, Yan. 2006. “Anaphora, Cataphora, Exophora, Logophoricity.” In Encyclopedia of Language
and Linguistics, 2nd Ed., ed. by Keith Brown, 231–240. Amsterdam: Elsevier.
doi: 10.1016/B0-08-044854-2/01084-1
Huang, Yan. 2014. Pragmatics (2nd Edition). Oxford: Oxford University Press.
Kullavanijaya, Pranee. 2005. “Pro-forms.” In Encyclopedia of Linguistics, ed. by Philipp Strazny,
877–878. New York: Fitzroy Dearborn.
Levinson, Stephen C. 2000. Presumptive Meanings: The Theory of Generalized Conversational
Implicature. Cambridge, MA: MIT Press.
Lyons, John. 1977. Semantics. Cambridge: Cambridge University Press.
Michael, Ian. 1970. English Grammatical Categories and the Tradition to 1800. Cambridge:
Cambridge University Press.
Montgomery, Michael, and Joseph S. Hall. 2004. Dictionary of Smoky Mountain English. Knoxville,
TN: University of Tennessee Press.
Neale, Stephen. 1992. “Paul Grice and the Philosophy of Language.” Linguistics and Philosophy
15: 509–559. doi: 10.1007/BF00630629
66 Wayne A. Davis
Perry, John. 1979. The Problem of the Essential Indexical. Noûs 13:3–21. Reprinted in Meaning
and Truth, ed. by John Garfield and Michael Kiteley, 613–627. New York: Paragon House,
1991.
Potts, Christopher. 2004. The Logic of Conventional Implicatures. Oxford: Oxford University Press.
Potts, Christopher. 2007. “Into the Conventional-Implicature Dimension.” Philosophy Compass
2: 665–679. doi: 10.1111/j.1747-9991.2007.00089.x
Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik. 1985. A Comprehensive
Grammar of the English Language. London: Longman.
Saxena, Anju. 2006. “Pronouns.” In Encyclopedia of Language and Linguistics (2nd ed.), ed. by
Keith Brown, 131–133. Amsterdam: Elsevier. doi: 10.1016/B0-08-044854-2/00284-4
Schiffer, Stephen. 1972. Meaning. Oxford: Clarendon Press.
Searle, John. 1975. “Indirect Speech Acts.” In Syntax and Semantics 3: Speech Acts, ed. by Peter
Cole, and Jerry L. Morgan, 59–82. New York: Academic Press.
Sportiche, Dominique. 2003. “Pronouns: Pronominals.” In International Encyclopedia of
Linguistics (2nd ed.), ed. by William J. Frawley, 407–409. Oxford: Oxford University Press.
Wales, Katie. 1996. Personal Pronouns in Present-Day English. Cambridge: Cambridge University
Press.
Wiese, Bernd. 1983. “Anaphora by Pronouns.” Linguistics 21: 373–417. doi: 10.1515/ling.1983.21.2.373
Chapter 4
Implicitness in the lexis

Lexical narrowing and neo-Gricean pragmatics
Yan Huang
University of Auckland
This article is dedicated to the memory of Associate Professor Frantisek Lichtenberk

(1945–2015), my Auckland colleague and friend, and a distinguished scholar in the
field of Oceanic languages and linguistics.
In this chapter, I present a neo-Gricean pragmatic analysis of lexical narrowing.

By lexical narrowing is meant the phenomenon of implicitness whereby the use
of a lexical item conveys a meaning that is more specific than the lexical expres-
sion’s lexically encoded meaning. Lexical narrowing can be grouped into two
types: (i) Q-based and (ii) R/I-based. In the first type, the use of a semantically
weaker lexical item in a set of contrastive semantic alternates Q-implicates the
negation of the meaning associated with that of the semantically stronger lexical
item in the same set. In the second, the use of a semantically general lexical ex-
pression I-implicates a semantically more specific sense. Finally, I argue that the
pragmatic enrichment involving lexical narrowing is nothing but a neo-Gricean,
‘pre-semantic’ conversational implicature.
Keywords: lexical narrowing, lexical/semantic underspecification, pragmatic

enrichment, neo-Gricean lexical pragmatics, pre-semantic conversational
1. Introduction
The linguistically encoded meaning of a lexical item can sometimes be narrowed

down in the use of that lexical item. In other words, the “literal” meaning of a lexical
expression can sometimes be strengthened in use in a context. Consider (1) and (2).
(1) John broke his finger.
(2) Mary had a glass of milk for breakfast this morning.
doi 10.1075/pbns.276.04hua
68 Yan Huang
Here, (1) is narrowed to (3) and (2) is strengthened to (4).
(3) John didn’t break his thumb.

(4) Mary had a grass of cow’s milk for breakfast this morning.
In this chapter, based on Huang (1998, 2009, 2015), I advance my neo-Gricean

pragmatic analysis of lexical narrowing – the phenomenon whereby the use of
a lexical item implicitly conveys a meaning that is more specific than the lexical
expression’s lexically specified meaning (see also Wilson and Kolaiti, this volume).
Within the framework of neo-Gricean pragmatics (e.g., Levinson 1983, 2000, Horn
1984, 2012a, b, Huang 1991, 1994, 2000, 2007, 2014, 2015), lexical narrowing can
be grouped into two types. First, there is the Q-based lexical narrowing. In this
type, the use of a semantically weaker lexical item in a set of contrastive semantic
alternates (such as a Q- or Horn-scale) Q-implicates the negation of the meaning
associated with that of the semantically stronger lexical item in the same set. This
is the case for (1). From the use of the semantically weaker finger, we obtain the
pragmatically narrowed meaning “not thumb”, as in (3). The second type of lexical
narrowing is R/I-based. The basic idea here is that the use of a semantically gen-
eral lexical item is R/I-implicated to a semantically more specific sense. This is the
case for (2), where the semantically general lexical item milk is I-narrowed to the
semantically more specific interpretation “cow’s milk”, as in (4).
The setup of this essay is as follows. In the next section, I discuss semantic or
lexical underspecification. In Section 3, I outline, first Grice’s classical pragmatic
theory, then Horn’s two-principled, and finally Levinson’s three-principled, model
of neo-Gricean pragmatics. Section 4 proceeds to provide a neo-Gricean lexical
pragmatic analysis of the two types of lexical narrowing. Finally, in Section 5, I
argue that the pragmatic enrichment involving lexical narrowing is nothing but a
neo-Gricean, “pre-semantic” conversational implicature, that is, a conversational
implicature that intrudes into or encroaches on the lexical semantics of these
lexical items.
2. Semantic underspecification
2.1 Linguistic underdeterminacy thesis
When we say something, we usually mean more than what we have actually said can
mean literally. Put in a different way, there is a (huge) gap between linguistic meaning
(roughly, what is said) and speaker meaning (roughly, what is meant or commu-
nicated). This is embodied in the linguistic underdeterminacy thesis in (5) (e.g.,
Chapter 4. Implicitness in the lexis 69
Austin 1962; Searle 1980, 1983, Sperber and Wilson 1986; Levinson 2000; Carston
2002; Bach 2004; Horn 2004; Recanati 2004; Atlas 2005; Huang 2007, 2014, 2016).
(5) The linguistic underdeterminacy thesis
The linguistically encoded meaning of a sentence radically underdetermines
the proposition a speaker expresses when he or she utters that sentence. 1
Suppose Mary says (6a). On most occasions, the message she intends to convey is
likely to be something along the line of (6b), which is more than what can be fully
determined by the linguistic meaning of (6a).
(6) a. I have absolutely nothing to wear.
b. I have absolutely nothing [appropriate] to wear [for John’s party]. 2
Furthermore, the gap between what is said and what is conveyed is usually filled in
by something that is essentially pragmatic in nature. Consider first the assignment
of anaphoric reference in (7).
(7) a. The city authorities barred the anti-immigration demonstrators
because they advocated violence.
b. The city authorities barred the anti-immigration demonstrators
because they feared violence.
In (7), the determination of reference for the anaphoric pronoun they is dependent
crucially on our background assumption about who would most likely be advo-
cating or fearing violence. This extra-linguistic information is responsible for the
two opposing meanings/readings, namely, they refers back to the anti-immigration
demonstrators in (7a) but is linked to the city authorities in (7b). Finally, take (8).
(8) a. Peter drove down a street in a Rolls-Royce.
b. Peter drove down a street in London.
The two sentences in (8) have the same surface syntactic structure. But given our
knowledge about the world, the street is pragmatically enriched not to be in the car
for (8a) but, to be in London for (8b).
1. But Bach (2012) called the linguistic underdeterminacy thesis a “contextual platitude”.
2. What is put in the brackets here and in Example (13) below is considered to be a pragmatically
filled unarticulated constituent (UC) (see e.g., Huang 2013b).
70 Yan Huang
2.2 Semantic or lexical underspecification
The same story can be told of the production and comprehension of lexical items.
As our first example, witness the use of take back in (9), adapted from Blutner
(2004).
(9) a. The zoo keepers took the panda back to the zoo.
b. The zoo keepers took the tram back to the zoo.
The meanings of take back are different in (9a) and (9b). In (9a), the panda is the
object that the zoo keepers took back to the zoo, whereas in (9b), the tram is the
instrument that took the zoo keepers back. But, from a semantic or lexical point
of view, take back is in general not taken as ambiguous. In other words, take back
is treated as semantically or lexically underspecified, and is then pragmatically en-
riched when it is used in (9). The same is true of other common, polysemous verbs
in English such as cut, make, and open.
(10) (Adapted from Searle 1980, 1983 and Oxford Advanced Learner’s Dictionary)
a. John cut his finger/wrist/chin/face/skin.
b. John cut his hair/nails.
c. John cut the grass/lawn/hedge.
d. John cut a loaf/cake/pie.
(11) (Adapted from Searle 1980, 1983 and Oxford Advanced Learner’s Dictionary)
a. John opened the door/window/gate.
b. John opened the bottle/can/box.
c. John opened a drawer.
d. John opened his book/magazine/dictionary at/to page 36.
e. John opened the curtains.
f. John opened his bag/briefcase.
g. John opened his eyes/mouth.
h. John opened his palms/arms.
i. John opened the newspaper/map.
j. John opened the lid.
k. John opened the road.
l. John opened the meeting/conference/the negotiations/the investigation
into the accident.
m. John opened a bank account.
n. The government opened a new office/school/hospital.
o. The firemen opened the wall.
p. The Queen opened parliament/the Commonwealth Games.
q. The surgeon opened the abscess/wound.
As pointed out by Searle (1980, 1983), while cut has a common semantic or lex-
ical content in (10a)–(10d), (10a)–(10d) have to be understood differently. This
is because the contribution made by cut to the truth conditions of (10a)–(10d) is
different, given the set of different background assumptions or real-world knowl-
edge involved. Clearly, what constitutes cutting one’s wrist is different from what
constitutes cutting one’s nails, which is also different from what constitutes cutting
a loaf. Essentially the same analysis can be applied to the use of open in (11a)–(11q).
We move next to adjectives. Consider first (12).
(12) John is tall.
A gradable adjective such as tall in (12) is graded against different standards of

comparison according to the class of objects that are discussed or the referents to
which the adjective is applied. 3 In other words, the values tall denotes vary from
class of objects to class of objects or context to context. Notice next that a very
small elephant is still much bigger than a very big mouse. This is because what is
big for a mouse is not big for an elephant. When we say a small elephant, we mean
“small as elephants go”, and when we say a big mouse, we mean “big as far as mice
are concerned”. That is to say, small and big take a fixed denotation only with regard
the norm of size for the class of mice and the class of elephants, respectively. Putting
it in more technical terms, the meaning of a gradable adjective is definable only
in terms of a measure function that maps objects to norms of standards/scales in
which the adjective appears. Returning now to (12), when we utter the sentence, tall
maps the height of John onto the degrees, i.e., points or intervals that are ordered
along the dimension of the property of height. More specifically, when we say (12),
we use tall to express a relation between the height John possesses and the stand-
ard of comparison that is dependent on a particular class of objects in a particular
context (Kennedy and McNally 2005; Kennedy 2007). The relation might be that
John is (significantly) taller than an average American white man or an average
African-American basketball player, depending on context. These two contextually
strengthen meanings can be given in (13a) and (13b), respectively. 4
3. Occasionally, a gradable adjective is called a “scalar” adjective, because it can form a point
or interval on a scale – a set of ordered degrees or graduations of some variable property or di-
mension such as age, length, width, size, height, speed, weight, accuracy, cost, and temperature.
4. In the philosophy of language, there is a school of thought that is called “non-indexical
contextualism” or “semantic relativism”. According to this approach (e.g., MacFarlane 2007),
contrary to indexicalists, context sensitivity called for by contextulists is not caused by the se-
mantic content or truth condition of a sentence but by a variation in its circumstances of evalu-
ation. In other words, a sentence like (13) is context-sensitive not because it expresses different
72 Yan Huang
(13) a. John is tall [for an average American white man].

b. John is tall [for an average African-American basketball player].
How, then, about absolute adjectives? They, too, are object/context-dependent. A

particularly well-known case involves adjective-noun collocations. As observed by
Quine (1960), there is a contrast between the meaning of the color adjective-noun
combination red apple (red on the outside) and that of the color adjective-noun
construction pink grapefruit (pink on the inside). Furthermore, the combined
meanings of a color adjective and its head noun are frequently different even if
the same color adjective is used, as in (14), adapted from Quine (1960) and Cruse
(2000). These are Quine’s riddles relating to the pragmatics of adjectives.
(14) a. a red apple
b. a red watermelon
c. a red flag/scarf/tablecloth
d. a red book/car/house
e. a red eye
f. a red pencil
Clearly, the color adjective red, though containing an element of the “literal” meaning
that is common to all its uses, nevertheless does not make a simple and fixed semantic
propositions in different contexts, but because the truth or falsity of its occurrences depends on
the circumstances in which it is evaluated. A circumstance of evaluation has two parameters:
(i) one for the possible world(s) and (ii) one for “counts-as”. The “counts-as” parameter is a func-
tion from properties (such as tallness) to intensions. It is so-called because “it fixes what things
have to be like in order to count as having the property of tallness (or any other property) at a
circumstance of evaluation” (MacFarlane 2007: 246). Thus, on this view, the different utterances
of (13) on different occasions of use may have different truth values because they have different
circumstances of evaluation, even if they are situated at or reside in the same possible world, and
express the same proposition.
As already mentioned, MacFarlane’s approach has been (re-)labelled “semantic relativism”.
This is because the truth of a sentence has to be relevant, or relativized, to both a parameter of a
possible world(s) and a parameter of “counts-as”. While acknowledging that varying standards,
interests, knowledge etc. have a semantic role to play, proponents of semantic relativism reject
the contextualist claim that the role in question is relevant to the determination of what is said.
Rather, the role played by varying standards etc. is relevant to determining whether what is said
is true or false. Some semantic relativists distinguish a context of use from a context of assess-
ment, and insist that epistemic standards, for example, are features of the context of assessment.
For instance, on a semantic relativist view, the proposition expressed by the sentence John knows
that there was a network outage yesterday does not vary across different contexts (specifically in
relation to the meaning of know). Rather, its truth value is relative to, or varies with, a standard of
knowledge. See e.g., Huang (2013a, 2014, 2016) for discussion about the ongoing debate between
contextualism and semantic minimalism in the philosophy of language.
(or truth-conditional) contribution to any of the combinations in (14). Put another

way, the various uses of red in (14) have to be computed differently, as in (15).
(15) a.
The apple is red on its peel.
b.
The watermelon is red inside its flesh.
c.
The whole flag/scarf/tablecloth is red.
d.
The book is red on its cover.
The car is red on a large part of its exterior.
The house is red on its outside.
e. The ‘white’ of the eye is red.
f. The pencil is red on its outside. Or,
The pencil is with red lead.
Essentially the same account can be given to other color adjectives such as black,
green, and brown (about which more in Section 5). 5
Finally, we come to nouns.
(16) a. Mary likes wearing rabbit.

b. Mary likes eating rabbit.
5. Note the following sentences where an absolute adjective is used.

i. (Austin 1962)
France is hexagonal.
ii. John is bald.
iii. The fridge is empty.
On many occasions, sentences of this type can be used in a broad, “loose” way, as pointed out by
Austin (1962). France is not, strictly speaking, hexagonal. John still has quite a few wispy strands
of hair on his head. There can still be some odd groceries left in the fridge. These sentences have
to be pragmatically broadened to (iv) to (vi).
iv. France is roughly hexagonal.
v. John is roughly bald.
vi. The fridge is roughly empty
The same can be said of geometrical nouns such as circle and rectangle, as in (vii) and (viii), which
need to be pragmatically loosened to (ix) and (x)
vii. The children are standing in a circle.
viii. There is a rectangle of lawn in front of the house.
ix. The children are standing in an approximate circle.
x. There is an approximate rectangle of lawn in front of the house.
All this shows that (i) absolute adjectives such as empty does not need to satisfy the maximal value
as the norm of comparison, contra Kennedy (2007) and Kennedy and McNally (2005), and (ii)
the borderline between gradable and absolute adjectives is not as clear-cut as has been expected.
Cruse (2000: 121) called contextual (sense) modulation of this type “impoverishment”. In rele-
vance-theoretical lexical pragmatics, lexical broadening of this type is treated as approximation
(e.g., Wilson and Carston 2017).
74 Yan Huang
In (16a), rabbit tends to be interpreted as rabbit fur/skin, as in (17a); whereas in

(16b) it tends to be read as rabbit meat, as in (17b). 6 Other cases include more
common, polysemous nouns like concert, government, and opera.
(17) a. Mary likes wearing rabbit fur/skin.
b. Mary likes eating rabbit meat.
On a par with adjective-noun combinations, there are noun-noun compounds. Like

that of adjective-noun combinations, the semantic property of noun-noun com-
pounds is sometimes notoriously difficult to work out. For example, the relation
between the modifying noun and the head noun in (18) has to be pragmatically
enriched to (19) by selecting the pragmatically “appropriate” specification from the
set of semantically or lexically “possible” ones. Both adjective-noun combinations
and noun-noun compounds present challenges for the principle of compositional-
ity, namely, the view that the meaning of a complex expression is a compositional
function (i.e., combination) of the meaning of its parts and their syntactic mode
of combination.
(18) a. a bread/cheese/steak/meat knife
b. a kitchen/garden knife
c. a pocket knife
d. a steel knife
e. a carving knife
f. a toy knife
(19) a. a knife for slicing/cutting bread/cheese/steak/meat
b. a knife that is used in a kitchen/garden
c. a knife that can be carried in a pocket 7
d. a knife that is made of steel
e. a knife for cutting cooked meat
f. a knife that is used as a toy
6. Notice that pig, for example, cannot be used in this way. This is because there is also a cor-
responding food-denoting noun, namely, pork, the use of which usually blocks the conceptual
grinding mechanism with respect to the use of its corresponding animal-denoting noun.
i. Mary likes eating pork/?pig.
But lexical blocking of this kind can be cancelled under certain conditions, resulting in what
Blutner (2004) called de-blocking. This is illustrated in (ii), where the use of cow rather than beef
is more appropriated.
ii. Hindus are forbidden to eat cow/?beef.
7. Pocket knife can also mean “penknife” – a small knife with one or more blades that fold down
into the handle – especially in American English (Oxford Advanced Learner’s Dictionary).
From all this, we can conclude that there is also a (huge) gap between the linguis-
tically coded meaning of a lexical item and what a speaker actually means when he
or she uses the lexical expression on a particular occasion. Furthermore, the gap
can be filled in only by pragmatic means utilizing extra-linguistic information such
as context, real-world knowledge, and implication/inference.
How can a semanticist tackle the examples we have discussed so far? He or she
has to treat the lexical items under discussion as either semantically or lexically
ambiguous. However, there is a serious problem at the very heart of this ambigu-
ity analysis, namely, the account is methodologically unattractive: it runs directly
against the spirit of a metatheoretical principle known as “Occam’s razor”, which
dictates that entities are not to be multiplied beyond necessity. A particular ver-
sion of Occam’s razor, dubbed “modified Occam’s razor” by Grice (1989: 47) is that
senses or dictionary entries must not proliferate. This has the consequence that all
things being equal, an analysis which has to posit two or more lexical items is to be
rejected in favor of an account which does not. Therefore, if we adopt the semantic
analysis, we have to take most lexical items in natural language as ambiguous –
clearly not an economic analysis.
An alternative is for us to take the lexical underspecification thesis, namely,
the view that lexical items determine an unspecified representation, and combine
it with a theory of pragmatic enrichment (see also Blutner 1998). In other words,
whenever we are faced with a lexical item that is apparently semantically or lexically
ambiguous, we should try to analyze it as containing a single univocal, semantical-
ly broad sense with a set of defeasible pragmatic enrichments (see also Levinson
2000: 20–21). A desirable consequence of such an account is that an interaction
and division of labor between lexical semantics and pragmatics will allow us to
avoid unnecessary semantic ambiguity and preserve semantic parsimony, thereby
substantially reducing the size of the lexicon in a language (Levinson 1983: 37–38,
Huang 2004). One theory that can be utilized to provide a pragmatic analysis of
lexical narrowing is the neo-Gricean pragmatic theory, to which I now turn. 8
8. As I pointed out in Huang (2009), the other two pragmatic theories that can be applied
to the study of the lexicon are (i) neo-Gricean oriented Optimality-theoretic pragmatics (e.g.,
Blutner 1998, 2004, 2010) and (ii) (post-Gricean) relevance theory (e.g., Wilson and Carston
2007; Walaszewska 2015).
76 Yan Huang
3. Classical and neo-Gricean pragmatics
3.1 Classical Gricean pragmatics
On a general Gricean account of meaning and communication, there are two theo-
ries: a theory of meaning-n[on]n[atural] and a theory of conversational implicature. In
his theory of meaning-nn, Grice emphasized the conceptual relation between natural
meaning in the external world and non-natural, linguistic meaning of utterances.
He developed a reductive analysis of meaning-nn in terms of the speaker’s reflexive
intention, the essence of which is that meaning-nn or speaker meaning is a matter
of expressing and recognizing intention.
In his theory of conversational implicature, Grice suggested that there is an
underlying principle that determines the way in which language is used effectively
and effectively to achieve rational interaction in communication. He called this
overarching dictum the co-operative principle and subdivided it into nine max-
ims of conversation classified into four categories: Quality, Quantity, Relation, and
Manner. The names of the four categories are taken from the German philosopher
Immanuel Kant (Grice 1989: 26). The co-operative principle and its component
maxims ensure that in an exchange of conversation, the right amount of informa-
tion is provided and that the interaction is conducted in a truthful, relevant, and
perspicuous manner.
(20) Grice’s theory of conversational implicature
a. The co-operative principle
Make your conversational contribution such as is required, at the stage at
which it occurs, by the accepted purpose or direction of the talk exchange
in which you are engaged.
b. The maxims of conversation
Quality: Try to make your contribution one that is true.
i. Do not say what you believe to be false.
ii. Do not say that for which you lack adequate evidence.
Quantity:
i. Make your contribution as informative as is required (for the current
purposes of the exchange).
ii. Do not make your contribution more informative than is required.
Relation: Be relevant.
Manner: Be perspicuous.
i. Avoid obscurity of expression.
ii. Avoid ambiguity.
iii. Be brief (avoid unnecessary prolixity).
iv. Be orderly.
The Gricean co-operative principle and its attendant maxims in (20) can be sim-
plified in (21) (e.g., Huang 2000, 2007, 2014).
(21) Grice’s theory of conversational implicature (simplified)
a. The co-operative principle
Be co-operative.
b. The maxims of conversation
Quality: Be truthful.
i. Don’t say what is false.
ii. Don’t say what lacks evidence.
Quantity:
i. Don’t say less than is required.
ii. Don’t say more than is required.
Relation: Be relevant.
Manner: Be perspicuous.
i. Avoid obscurity.
ii. Avoid ambiguity.
iii. Be brief.
iv. Be orderly.
Assuming that the co-operative principle and its associated maxims are normally
adhered to by both a speaker and the addressee in a conversational interaction,
Grice suggested that a conversational implicature – roughly, any meaning or prop-
osition expressed implicitly by a speaker in his or her utterance of a sentence which
is meant without being part of what is said in the strict sense 9 – can arise from ei-
ther strictly observing or ostentatiously flouting the maxims. In Huang (e.g., 2007,
2014), I called conversational implicatures that are engendered by way of directly
observing the maxims conversational implicatureO, and conversational implica-
tures that are generated by way of the speaker’s deliberately flouting the maxims
conversational implicatureF.
A second Gricean dichotomy, independent of the conversational implicatureO/
conversational implicatureF one, is between those conversational implicatures
which arise without requiring any particular contextual conditions and those which
do require such conditions. Grice (1989: 31–38) called the first kind generalized
conversational implicatures (GCIs), and the second kind particularized conversa-
tional implicatures (PCIs).
9. Defined thus, conversational implicature is a component of speaker meaning rather than a

pragmatic inference (e.g., Saul 2002; Bach 2012; Horn 2004, 2012a, b). By contrast, Sperber and
Wilson (1995), Levinson (2000), Atlas (2005), and Chierchia (2013) are still treating conversa-
tional implicature as a pragmatic inference.
78 Yan Huang
Finally, Grice designed a battery of test to facilitate the identification of con-

versational implicature. First, defeasibility or cancellability – conversational im-
plicatures can disappear in certain linguistic or non-linguistic contexts. Secondly,
non-detachability – any linguistic expression with the same semantic content tends
to carry the same conversational implicature. (A principled exception is those con-
versational implicatures that arise via the maxim of Manner.) Thirdly, calculabili-
ty – conversational implicatures can transparently be derived via the co-operative
principle and its attendant maxims. Fourthly, non-conventionality – conversational
implicatures, though dependent on the saying of what is coded, are non-coded in
nature. Fifthly, reinforceability – conversational implicatures can be made explicit
without producing too much sense of redundancy. Sixthly, some conversational
implicatures may be indeterminate. They can be taken as conveying an open-ended
range of implicitly-expressed meanings relating to matters in hand. Finally, we have
universality – conversational implicatures tend to be universal, being rationally mo-
tivated rather than arbitrary (e.g., Grice 1989; Levinson 2000; Huang 2014: 39–43,
2015b, 2017 a,b). 10
3.2 Horn’s bipartite neo-Gricean model
Within the Gricean paradigm, the two most influential developments are the
neo-Gricean pragmatic models advance by Horn and by Levinson.
Horn (1984, 2012a, b) put forward a bipartite model. On Horn’s view, all of
Grice’s maxims (except the maxim of Quality) can be replaced with two funda-
mental and counterpoising principles: the Q[uantity]- and R[elation]-principles.
(22) Horn’s Q- and R-principles
a. The Q-principle
Make your contribution sufficient;
Say as much as you can (modulo the R-principle).
b. The R-principle
Make your contribution necessary;
Say no more than you must (modulo the Q-principle).
10. Needless to say, the Gricean doctrine has a long lineage. Some proto-Gricean notions can go
back at least as far as the first-century BC rhetorician Dionysius and the fourth-century rhetori-
cians Servius and Donatus. These ideas were later reiterated by the nineteenth-century English
philosophers John Stuart Mill and Augustus De Morgan. Much more recently in the 1950s,
Grice was also influenced by similar concepts put forward by his colleagues within the tradition
of ordinary language philosophy in the University of Oxford (e.g., Horn 2012a, b, Huang 2014).
See also the overview in Dynel (this volume)
In terms of information structure, Horn’s Q-principle, which collects Grice’s first

sub-maxim of Quantity and his first two sub-maxims of Manner, is a lower-bound-
ing pragmatic principle which may be (and characteristically is) exploited to en-
gender upper-bounding conversational implicatures: a speaker, in saying “…p…”,
ceteris paribus conversationally implicates that (for all he or she knows) “…at most
p…”. The locus classicus here is those conversational implicatures that arise from a
prototype Horn-scale.
On the other hand, the counterbalancing R-principle, which subsumes Grice’s
second sub-maxim of Quantity, his maxim of Relation, and his last two sub-maxims
of Manner, and which is based on Atlas and Levinson’s (1981) principle of infor-
mativeness, is an upper-bounding pragmatic law which may be (and systematical-
ly is) exploited to invite low-bounding conversational implicatures: a speaker, in
saying “…p…”, conversationally implicates that (for all he or she knows) “… more
than p…”. However, more recently Horn (2012a,b) has been of the view that the
R-principle is not in itself subsumable under Grice’s co-operative principle, but
under rationality.
Viewing the Q- and R-principles as a mere instantiation of Zipfian econo-
my, Horn (1984, 2012a, b) explicitly equated the Q-principle (“a hearer-oriented
economy for the maximization of informational content”) with Zipf ’s Auditor’s
Economy (the Force of Diversification, which tends towards a vocabulary of m
different words with one distinct meaning for each word) and the R-principle
(“a speaker-oriented economy for the minimization of linguistic form”) with
Zipf ’s Speaker’s Economy (the Force of Unification, which tends towards a vo-
cabulary of one word which will refer to all the m distinct meanings). The no-
tion of Speaker’s Economy is further distinguishable between mental inertia or
paradigmatic economy (économie mémorielle) and articulatory/physical inertia
or syntagmatic economy (économie discursive), hence internally dialectic in its
operation. The former is concerned with the reduction in the inventory of men-
tal lexicon; the latter, with the reduction in the number of linguistic units (e.g.,
Horn 2007: 173–174). While the Auditor’s Economy places a lower bound on the
informational content of the message, the Speaker’s Economy places an upper
bound on its form. Furthermore, Horn argued that the whole Gricean mechanism
for pragmatically contributed meaning can be derived from the dialectic inter-
action (in the classical Hegelian sense) between the two mutually constraining
mirror-image forces in the following way.
(23) Horn’s division of pragmatic labor
The use of a marked (relatively complex and/or prolix) expression when a
corresponding unmarked (simpler, less “effortful”) alternate expression is avail-
able tends to be interpreted as conveying a marked message (one which the
unmarked alternative would not or could not have conveyed).
80 Yan Huang
In effect, what the communicative equilibrium in (23) basically says is this: the
R-principle generally takes precedence until the use of a contrastive linguistic form
induces a Q-implicature to the non-applicability of the pertinent R-implicature.
3.3 Levinson’s trinitarian neo-Gricean pragmatic model
Considering that the Horn model fails to draw a distinction between semantic
and expression minimizations, Levinson (e.g., 2000) argued for a clear separation
between pragmatic principles that govern an utterance’s semantic or informational
content and pragmatic principles that govern its surface form. Consequently, he
proposed that the original Gricean programme (the maxim of Quality apart) be re-
duced to three neo-Gricean pragmatic principles: what he dubbed the Q[uantity]-,
I[nformativeness]-, and M[anner]-principles. Each of the three principles has two
sides: a speaker’s maxim, which specifies what the principle enjoins a speaker to
say and to implicate, and a recipient’s corollary, which dictates what it allows the
addressee to infer.
(24) Levinson’s Q-principle
Speaker’s maxim:
Do not provide a statement that is informationally weaker than your knowl-
edge of the world allows, unless providing a stronger statement would contra-
vene the I-principle.
Recipient’s corollary:
Take it that the speaker made the strongest statement consistent with what he
knows, and therefore that:
i. if the speaker asserted A(W), where A is a sentence frame and W an infor-
mationally weaker expression than S, and the contrastive expressions <S,
W> form a Horn scale (in the prototype case, such that A(S) entails A(W)),
then one can infer that the speaker knows that the stronger statement A(S)
(with S substituted for W) would be false (or K~ (A(S)));
ii. if the speaker asserted A(W) and A(W) fails to entail an embedded sen-
tence Q, which a stronger statement A(S) would entail, and <S, W> form
a contrast set, then one can infer the speaker does not know whether Q
obtains or not (i.e., ~K(Q) or equally {P (Q), P~(Q)}.
The Q-principle can be simplified as follows (e.g., Levinson 2000; Huang 2007, 2014).
(25) Levinson’s Q-principle (simplified)

Speaker: Do not say less than is required (bearing the I-principle in mind).
Addressee: What is not said is not the case.
The basic idea of the metalinguistic Q-principle is that the use of an expression (es-
pecially a semantically weaker one) in a set of contrastive semantic alternates (such
as a Q- or Horn-scale) Q-implicates the negation of the meaning/interpretation as-
sociated with the use of another expression (especially a semantically stronger one)
in the same set. Seen the other way round, from the absence of an informationally
stronger expression, we infer that the interpretation associated with the use of that
expression does not hold. Hence, the Q-principle is essentially negative in nature.
(I use “+>” to stand for “conversationally implicate”.)
(26) Q: <x, y>
y +>Q ~ x
(27) <all, some>
Some of John’s students can speak French.
+> Not all of John’s students can speak French.
Next, there is Levinson’s I-principle.
(28) Levinson’s I-principle

Speaker’s maxim: the maxim of minimization
“Say as little as necessary”, that is, produce the minimal linguistic information
sufficient to achieve your communicational ends, (bearing the Q-principle in
mind).
Recipient’s corollary: the rule of enrichment
Amplify the informational content of the speaker’s utterance, by finding the
most specific interpretation, up to what you judge to be the speaker’s m-in-
tended point, unless the speaker has broken the maxim of minimization by
using a marked or prolix expression.
Specifically:
i. Assume the richest temporal, causal and referential connections between
described situations or events, consistent with what is taken for granted.
ii. Assume that stereotypical relations obtain between referents or events,
unless this is inconsistent with (a).
iii. Avoid interpretations that multiply entities referred to (assume referen-
tial parsimony); specifically, prefer coreferential readings of reduced NPs
(pronouns or zeros).
iv. Assume the existence or actuality of what a sentence is about if that is
consistent with what is taken for granted.
Ignoring its four instantiations, the I-principle can be simplified as follows (e.g.,
Levinson 2000; Huang 2007, 2014).
82 Yan Huang
(29) Levinson’s I-principle (simplified)

Speaker: Do not say more than is required (bearing the Q-principle in mind).
Addressee: What is generally said is stereotypically and specifically exemplified.
Mirroring the effects of the Q-principle, the central tenet of the I-principle is that
the use of a semantically general expression I-implicates a semantically more spe-
cific meaning/interpretation. More accurately, the conversational implicature en-
gendered by the I-principle is one that accords best with the most stereotypical and
explanatory expectation given our knowledge about the world.
(30) I-scale: [x, y]
y + >I x
(31) (Conjunction buttressing)
p and q +> p and then q
+> p therefore q
+> p in order to cause q
John pressed the spring and the drawer opened.
+> John pressed the spring and then the drawer opened
+> John pressed the spring and thereby caused the drawer to open
+> John pressed the spring in order to make the drawer open
Finally, we come to Levinson’s M-principle.
(32) Levinson’s M-principle

Speaker’s maxim:
Indicate an abnormal, non-stereotypical situation by using marked expressions
that contrast with those you would use to describe the corresponding normal,
stereotypical situation.
Recipient’s corollary:
What is said in an abnormal way indicates an abnormal situation, or marked
messages indicate marked situations.
Specifically:
Where S has said p containing marked expression M, and there is an unmarked
alternate expression U with the same denotation D which the speaker might
have employed in the same sentence frame instead, then where U would have
I-implicated the stereotypical or more specific subset d of D, the marked ex-
pression M will implicate the complement of the denotation d’, namely d’ of D.
The M-principle can be simplified as follows (e.g., Levinson 2000; Huang 2007, 2014).
(33) Levinson’s M-principle (simplified)

Speaker: Do not use a marked expression without reason.
Addressee: What is said in a marked way conveys a marked message.
Unlike the Q- and I-principles, which operate primarily in terms of semantic infor-
mativeness, the metalinguistic M-principle operates primarily in terms of a set of
alternates that contrast in linguistic form. The fundamental axiom upon which this
principle rests is that the use of a marked linguistic expression M-implicates the
negation of the interpretation associated with the use of an alternative, unmarked
expression in the same set. Putting it another way, from the use of a marked lin-
guistic expression, we infer that the stereotypical interpretation associated with
the use of an alternative, unmarked linguistic expression does not obtain.
(34) M-scale: {x, y}
y +>M ~ x
(35) a. The timetable is reliable.
+> The timetable is reliable to degree n.
b. The timetable is not unreliable.
+> The timetable is reliable to degree less than n.
Given the above tripartite classification of neo-Gricean pragmatic principles, the

question that arises next is how inconsistencies arising from these potentially con-
flicting implicature apparatuses can be resolved. According to Levinson (2000),
they can be resolved by an ordered set of precedence, which encapsulates in part
Horn’s division of pragmatic labor in (23).
(36) Levinson’s resolution schema for the interaction of the Q-, I-, and M-principles
a. Level of genus: Q > M > I
b. Level of species: e.g., Q-clausal > Q-scalar
This is tantamount to saying that genuine Q-implicatures (where Q-clausal cancels

rival Q-scalar) precede inconsistent I-implicatures, but otherwise I-implicatures take
precedence until the use of a marked linguistic expression triggers a complementary
M-implicature to the negation of the applicability of the pertinent I-implicature (see
e.g., Huang 2014, 2017 a, b for further discussion).
4. Two types of lexical narrowing
In Section 3, I outlined, first classical Gricean pragmatics, secondly Horn’s, and

thirdly and finally, Levinson’s neo-Gricean pragmatic models. In this section, let
me provide a neo-Gricean pragmatic analysis of lexical narrowing.
84 Yan Huang
4.1 Q-based lexical narrowing
As already mentioned, lexical narrowing can be divided into two categories. In

the first, the use of a semantically weaker linguistic expression in a set of contrast
semantic alternates Q-implicates the negation of the meaning associated with the
use of another semantically stronger expression in the same set, hence narrowing
its meaning. Consider first (37).
(37) The tea is warm.
+> The tea is not hot
The gradable adjective warm has two readings: (i) at least warm, and (ii) warm but
not hot. In addition, it forms a Horn-scale with another gradable adjective hot. The
use of warm in (37) pragmatically excludes the meaning “at least warm”, and retains
only the meaning “warm but not hot”, hence strengthening the lexical item’s sense.
Warm is an “ordinary” scalar expression. The same lexical narrowing effect can be
achieved in the use of scalar logical operators and cardinal numbers.
(38) a. some (at least some, some but not all>
+> not all
b. 9 (at least 9, 9 but not larger than 9)
+> not larger than or only 9
A subtype of scalar expressions that displays Q-based lexical narrowing is con-

cerned with what is called “autohyponymy”. By autohyponmy is meant the lexical
phenomenon, whereby a lexical item is employed both as a superordinate term
with a default general meaning of a hyponymic taxonomy and as one of its own
(co-)hyponyms with a contextually more restricted or specific sense in the taxon-
omy (e.g., Cruse 2000; Levinson 2000). This is the case for finger, which has both
a default general sense (one of the five parts that stick out from a hand) and a con-
text-dependent, more specific meaning (one of the four long thin parts that stick
out from a hand). When the lexical item is used in its general sense, it functions as
a superordinate term, and when it occurs with its contextually more specific sense,
it acts as a (co-)hyponym of itself.
(39) finger
thumb finger
(40) John broke his finger.

+> John didn’t break his thumb
In (40), the use of finger as a superordinate term, where its semantically stronger
alternate thumb could have been employed, but was not actually used, implicates
the inapplicability of the sense associated with thumb. This has the consequence
that finger narrowly denotes the complement of its extension of finger as a co-
hyponym of its own, ruling out the meaning of thumb. In addition to (40), further
exemplification of lexical narrowing involving autohyponymy is provided in (41).
(41) a. dog +> not bitch
b. animal +> not human 11
c. gay +> not lesbian
d. actor +> not actress
e. lion +> not lioness
f. rectangle +> not square
Lexical narrowing of the type displayed in (37), (40), and (41) follows directly from
Horn’s and Levinson’s Q-principle. Notice that hot and warm form a Horn-scale in
(37). Since the speaker has chosen the semantically weaker expression warm, where a
semantically stronger one of equal brevity hot is available, he or she would contradict
the Q-principle if the semantically stronger expression held. Consequently, he or she
believes that the semantically stronger statement does not hold. Hence, from the use
of the semantically weaker warm, we obtain the pragmatically narrowed meaning
“not hot”. Exactly the same analysis can be made of (40) and (41), where already
mentioned, the Q-based reduction of meaning typically gives rise to autohyponymy.
Note that these Q-narrowed meanings are not part of the lexical semantics of
the linguistic items under consideration, because they can be cancelled, as in (42)
and (43). (I use “~ +>” to signify “do not conversationally implicate”.) Nevertheless,
they are part of the lexical implicature/inference of the linguistic expression.
(42) The tea is warm, if not hot.
~ +> The tea is not hot
11. Note that animal is polysemic, and consequently in one of its senses, it may act as a superor-
dinate to itself in another sense (Palmer 1981).
living non-living
vegetable animal in contrast with vegetable to cover birds,

fishes, insects as well as mammals
bird fish insect animal in the sense of ‘mammal’ to contrast

with birds, fishes, and insects to cover
both humans and beasts
human animal in the sense of ‘beast’ to contrast

with humans
86 Yan Huang
(43) John broke his finger, and in fact his thumb.

~ +> John didn’t break his thumb
4.2 R/I-based lexical narrowing
Secondly and more interestingly, there is the R/I-based lexical narrowing. The ba-
sic idea here is that the use of a semantically general lexical item is R/I-implicated
to a semantically more specific meaning/interpretation. This is the case for (2),
repeated here as (44), where the semantically general term milk is R/I-narrowed
to denote its culturally salient subset “cow’s milk” (cf. goat’s milk, soy milk, almond
milk, coconut milk, etc.).
(44) John had a glass of milk for breakfast this morning.
+> John had a glass of cow’s milk for breakfast this morning
More examples of this type are given in (a) with their I-narrowed versions in (b)
of (45)–(64).
(45) a. I usually have an egg for breakfast. (cf. egg used in an Easter egg)
b. +> I usually have a hen egg for breakfast
(46) (Advertisement against binge drinking on TVNZ).
a. It’s not the drinking; it’s the way we drink.
b. It’s not the alcoholic drinking; it’s the way we drink alcohol
(47) a. I don’t smoke.
b. +> I don’t smoke cigarettes 12
(48) a. Mike’s cousin is a nurse/secretary/prostitute.
b. Mike’s cousin is a female nurse/secretary/prostitute
(49) a. Fernando Lugo [the President of Paraguay] acknowledged he had a rela-
tionship with Viviana Carillo [while he was still a Catholic bishop].
(The New Zealand Herald 15 April 2009)
b. +> Fernando Lugo acknowledged he had a sexual/romantic relationship
with Viviana Carillo [while he was still a Catholic bishop]
(50) a. I don’t know if Mary has already had a partner.
b. +> I don’t know if Mary has already had a sexual/romantic partner.
(51) a. Something smells here!
b. +> Something smells bad/stinks here!
12. Note that as noted by Blutner (2010), in Amsterdam, the uttering of smoke especially in Please
smoke inside can R/I-imply “smoke joints (cigarettes containing marijuana)”. This is not unex-
pected given that the R/I-implicated narrowing “smoking cigarettes”, as in (47b), is pragmatic in
nature, which can be defeated by the background assumption about drug-taking in Amsterdam.
(52) a. The little girl has a temperature.

b. +> The little girl has a high temperature
(53) a. Some people are a bit surprised when they found out that I’ve got a brain.
(Catherine McQueen) 13
b. +> Some people are a bit surprised when they found out that I’ve got a
good brain
(54) a. Please have a heart for animals this Valentine’s Day! 14
b. +> Please have a kind heart for animals this Valentine’s Day!
(55) a. Clinton made a lot of noise and had a very big impact. He has the shoulders.
If he comes, the world is interested, and it brings attention to Lesotho.
When I climb on the foothills of Clinton, perhaps people will notice more.
(FT Magazine 19/20 August 2006)
b. +> Clinton made a lot of noise and had a very big impact. He has the strong
shoulders. If he comes, the world is interested, and it brings attention to
Lesotho. When I climb on the foothills of Clinton, perhaps people will
notice more.
(56) a. The Buddhist temple is some distance away from the city centre.
b. +> The Buddhist temple is some considerable distance away from the city
centre.
(57) a. It took time for Yan to write The Oxford Dictionary of Pragmatics.
b. +> It took a lot of time for Yan to write The Oxford Dictionary of Pragmatics..
(58) a. You need money to buy a property in Hong Kong.
b. +> You need a lot of money to buy a property in Hong Kong. 15
(59) a. John wants to marry an Asian. (British English)
b. +> John wants to marry a sub-Indian continental Asian
13. This use of brain has been found in other languages including Arabic, Chinese, and Modern
Greek.
14. Even if have a heart for is a set phrase. Here is another attested example, which comes from
the New Zealand Red Cross Refugee Services’ appeal for furniture donation.
(i) It takes hands to make a house, but it takes heart to make a home.
15. Notice that the proposition determined by the linguistically given meaning of each of the
sentences in (52a)–(58a) is trivially true, that is, every (living) human being has temperature,
brain, heart etc. In a sense, these sentences are superficially uninformative. Confronted with this
blatant infringement of Grice’s maxim of Quantity, the addressee assumes that the speaker is
actually co-operative, and has to work out why he or she has used such an apparently uninform-
ative sentence. The only way to do it is to interpret it as highly informative. By the R/I-principle,
not only the literal meaning, but something more than the literal meaning m-intended by the
speaker, namely, the strengthened meaning as well is computed.
88 Yan Huang
(60) a. John is reading two modern languages at Oxford. (British English)

b. +> John is reading two modern European languages other than modern
English at Oxford University
(61) a. That country is testing a bomb.
b. +> That country is testing an atom bomb
(62) a. Mary dislikes blood.
b. +> Mary dislikes human blood
(63) a. Something has happened in the hotel.
b. +> Something unusual has happened in the hotel
(64) a. Have you been to Shakespeare’s birthplace?
b. +>Have you been to the birthplace of William Shakespeare, the English
dramatist and poet
The class of R/I-implicated lexical narrowing is heterogeneous, but the strength-

ened readings in (b) of (45)–(64) share a number of properties. In the first place,
they are semantically more specific than the lexical items that engender them. For
example, in (46), from the use of the semantically general drinking/drink, we ob-
tain the semantically more specific meaning/interpretation “alcoholic drinking/
drink alcohol”. In (48), the speaker uses the semantically general nurse/secretary/
prostitute and narrows it to, and the addressee infers it as “female nurse/secretary/
prostitute”, which is semantically more specific. The difference between drink
and secretary, according to Horn (1984) and Levinson (2000), is that while drink
is an autohyponym, given that it expresses both a general (drink liquid) and
a specific (drink alcohol) meaning, secretary is not. The latter is sociologically
motivated, i.e., due to the social division of labor). In (64), from Shakespeare,
which may be the surname of a potentially large number of people in the English-
speaking world, we get the semantically more specific interpretation “William
Shakespeare, the English dramatist and poet”. The same can be said of all the
other examples. Secondly, unlike Q-based lexical narrowing, R/I-based lexical
narrowing is positive in nature. Thirdly, in some cases, it is characteristically
guided by stereotypical assumptions. This is the case, for instance, for (44) and
(45)–(48). Fourthly, it is generally non-metalinguistic, in the sense that it makes
no reference to something that might have been said but was not (e.g. Huang
1998, 2005, 2009, Levinson 2000).
Again, one point that needs to be emphasized is that the pragmatically
I-narrowed meanings are not part of the lexical semantics of the lexical items un-
der discussion. This can be seen by the fact that these lexical items can co-occur
with lexical items such as male, goat, and dog without producing any semantic or
lexical contradiction.
(65) a. A male nurse/secretary/prostitute was arrested yesterday.

b. John particularly likes goats’ milk.
c. Mary doesn’t drink tap water.
d. As soon as she saw her dog’s blood, Susan fainted.
e. Modern Standard Arabic is a modern language.
f. Something smells delicious in the dining hall!
g. Mr Key and Premier Wen Jiabao toasted the health of the relationship
and the desire to build on the one-year-old trade deal between the two
countries. (The New Zealand Herald 17 April 2009)
Finally, I-based lexical narrowing plays a similar role in the interpretation of adjec-
tive-noun combinations (66), noun-noun compounds (67), possessive construc-
tions (68), systematic ambiguity (69), and perhaps deferred reference (70). 16
(66) (Adapted from Lahav 1993 and Oxford Advanced Learner’s Dictionary)
a. brown cow
+> most of its body surface is brown
b. a brown book
+> most of its cover is brown
c. a brown newspaper
+> all its pages are brown
d. a brown paper bag
+> the whole paper bag is brown
e. a brown crystal
+> both its inside and outside are brown
f. a brown potato
+> a significant portion of its outer skin is brown
g. a brown bread
+> only the outside crust is brown
h. a brown house
+> only the outside of the house is brown
i. a brown eye
+> only the iris is brown
(67) a. the government’s drugs campaign
+> the government’s campaign against drugs
b. the government’s safe-sex campaign
+> the government’s campaign for safe-sex
16. For a detailed discussion of deferred reference, see Huang (2014: 252–259). See also Huang
(2009) and Horn (2017) for discussion of other topics within the framework of neo-Gricean
lexical pragmatics.
90 Yan Huang
(68) a. John’s father +> the one to whom he is son

b. John’s office +> the one in which he works
c. John’s class +> the one he attends
d. Newton’s ideas +> the ones originated from Newton
e. Halley’s comet +> the one named after Halley
(69) a. The whole nursery burnt down.
+> The buildings of the whole nursery burnt down
b. The whole nursery went on a country outing.
+> The staff and children of the whole nursery went on a country outing
(70) (A hotel chambermaid to another chambermaid)
Room 336 wants another blanket.
+> The hotel guest who is staying in Room 336 wants another blanket
5. Pragmatic enrichment involving lexical narrowing: Explicature,

the pragmatically enriched said, conversational impliciture
or conversational implicature?
In the last section, I presented a neo-Gricean pragmatic analysis of lexical narrow-

ing. In this section, I consider the question of what the pragmatic enrichment under
consideration is. Currently, within pragmatics and the philosophy of language, two
approaches can be identified: (i) the non-conversational-implicature approach and
(ii) the conversational implicature approach. Within the first camp, three analyses
are of particular interest. First, in relevance theory, pragmatic enrichment of this
kind is analyzed as explicature – an inferential development of one of the linguis-
tically-given incomplete conceptual representations or logical forms of a sentence
uttered (e.g., Sperber and Wilson 1995; Wilson and Kolaiti, this volume). Defined
thus, it is a pragmatically inferred component of the Gricean notion of what is said
(though what is said is abandoned in relevance theory). Secondly, somewhat similar
to the relevance-theoretic view is the position taken by Recanati (e.g., 2004, 2010).
According to Recanati, the pragmatic enrichment under discussion is part of the
pragmatically enriched said. Finally, a third approach is due to Bach (e.g., 2004,
2012). On Bach’s view, certain communicative content do not need to be recognized
as either part of what is said or part of what is conversationally implicated. Rather,
they constitute a middle ground between what is said and what is conversationally
implicated. Bach dubbed this middle level of speaker-meaning conversational im-
pliciture or impliciture for short.
On the other hand, within the neo-Gricean pragmatic framework, Levinson
(2000: 195–196) is of the view that the pragmatic enrichment under consideration
is neither an explicature; nor the pragmatically enriched said; nor an impliciture.
Rather, it is the same beast as a neo-Gricean conversational implicature. 17 In my

view (e.g., Huang 2007, 2013a, 2014), the reason why it is a conversational impli-
cature is threefold. Firstly, so-called explicature/the pragmatically enriched said/
impliciture is engendered largely by the same Gricean pragmatic mechanism that
yields a conversational implicature. Secondly, Recanati (1993) put forward two tests,
i.e. the availability principle and the scope principle to differentiate e xplicature/
the pragmatically enriched said/impliciture from conversational implicature. But
as I argued in Huang (e.g., 2007), neither of Recanati’s tests seems to work from a
theoretical point of view. This is also the case with work carried out in experimen-
tal pragmatics. I do not think that there is any experiment that can differentiate
explicature/the pragmatically enriched said/impliciture from conversational im-
plicature. Therefore, currently there is no failsafe test (both conceptual and experi-
mental) that can be employed to distinguish alleged explicature/the pragmatically
enriched said/impliciture from conversational implicature on a principled basis.
Thirdly, other things being equal, given the “Occam’s razor”, mentioned above,
the conversational implicature analysis is theoretically and methodologically pref-
erable, because it postulates less theoretical categories than the explicature/the
pragmatically enriched said/impliciture accounts. If neo-Gricean conversational
implicature can intrude on to truth-conditional content of an utterance, then a
problem known as Grice’s circle arises, namely, how what is conversationally im-
plicated can be defined in contrast to, and calculated on the basis of what is said,
given that what is said seems to both determine and to be determined by what
is conversationally implicated. Levinson’s proposal was that we should reject the
“received” view of the pragmatics-semantics interface, namely, the view that the
output of semantics is the input to pragmatics, and allow implicatures to play a
systematic role in “pre”-semantics, that is, to help determine the truth-conditional,
propositional or lexical content of an utterance (e.g., Levinson 2000, Huang 2007).
Putting it slightly differently, in order to avoid Grice’s circle, we need both “pre”-
and “post-semantic” pragmatics or what Korta and Perry (2011) called near- and
far-side pragmatics. Pragmatic enrichment involving lexical narrowing, discussed
in this essay, is therefore a neo-Gricean, pre-semantic conversational implicature. 18
17. See my neo-Gricean and revised neo-Gricean analyses of anaphora in e.g. Huang (1991,
1994/2007, 2000a, b, 2004, 2007 and 2014), which in effect argue that the pragmatic enrich-
ment involved in the determination of anaphora is a neo-Gricean, “pre-semantic” conversational
implicature.
18. Even if the dispute were entirely of a terminological rather than a substantive nature, the
force of my argument seems to remain. This is because to have fewer technical terms is better
than having more.
92 Yan Huang
Acknowledgements
Part of the material contained in this article was first presented to generations of linguistics post-
graduate students who attended my course on lexical pragmatics at the University of Oxford and
the University of Reading. I am grateful to David Cram for inviting me to teach a joint course
on lexical pragmatics with him at Oxford in 1998 and 2000. Earlier versions were subsequent-
ly given at a number of international conferences, workshops, and colloquia/seminars held in
China, Hong Kong, Taiwan, Europe, the UK, and the USA. I have benefited from the comments
received on all these occasions.
References
Atlas, Jay D. 2005. Logic, Meaning, and Conversation: Semantic Underdeterminacy, Implicature,
and Their Interface. Oxford: Oxford University Press.
doi: 10.1093/acprof:oso/9780195133004.001.0001
Austin, John L. 1962. How to Do Things with Words. Oxford: Oxford University Press.
Bach, Kent. 2004. “Pragmatics and the Philosophy of Language”. In The Handbook of Pragmatics,
ed. by Laurence R. Horn and Gregory Ward, 463–487. Oxford: Blackwell.
Bach, Kent. 2012. “Saying, Meaning, and Implicating”. In The Cambridge Handbook of Pragmatics,
ed. by Keith Allan and Katarzyna M. Jaszczolt, 47–68. Cambridge: Cambridge University
Press. doi: 10.1017/CBO9781139022453.004
Blutner, Reinhard. 1998. “Lexical Pragmatics.” Journal of Semantics 15: 115–162.
doi: 10.1093/jos/15.2.115
Blutner, Reinhard. 2004. “Pragmatics and the Lexicon.” In The Handbook of Pragmatics, ed. by
Laurence R. Horn and Gregory Ward, 488–514. Oxford: Blackwell.
Blutner, Reinhard. 2010. “Lexical Pragmatics.” In The Pragmatic Encyclopedia, ed. by Louise
Cummings, 247–251. London: Routledge.
Chierchia, Gennaro. 2013. Logic in Grammar: Polarity, Free Choice, and Intervention. Oxford:
Oxford University Press. doi: 10.1093/acprof:oso/9780199697977.001.0001
Cruse, Alan. 2000. Meaning in Language. Oxford: Oxford University Press.
Grice, Herbert P. 1989. Studies in the Way of Words. Cambridge, MA: Harvard University Press.
Horn, Laurence R. 1984. “Toward a New Taxonomy for Pragmatic Inference: Q-based and
R-based Implicature.” In Meaning, Form, and Use in Context: Linguistic Applications, ed. by
Deborah Schiffrin, 11–42. Washington, DC: Georgetown University Press.
Horn, Laurence R. 2004. “Implicature.” In The Handbook of Pragmatics, ed. by Laurence R. Horn
and Gregory Ward, 3–28. Oxford: Blackwell.
Horn, Laurence R. 2007. “Neo-Gricean Pragmatics: A Manichaean Manifesto.” In Pragmatics,
ed. by Noel Burton-Roberts, 158–183. London: Palgrave Macmillan.
Horn, Laurence R. 2012a. “Implying and Inferring.” In The Cambridge Handbook of Pragmatics,
ed. by Keith Allan and Katarzyna M. Jaszczolt, 69–86. Cambridge: Cambridge University
Press. doi: 10.1017/CBO9781139022453.005
Horn, Laurence R. 2012b. “Implicature.” In The Routledge Companion to Philosophy of Language,
ed. by G. Russell and D. Fara, 53–66. New York: Routledge.
Horn, Laurence, R. (forthcoming). “Lexical Pragmatics.” In The Oxford Handbook of Pragmatics,

ed. by Yan Huang. Oxford: Oxford University Press.
Horn, Laurence R., and Gregory Ward, Gregory (eds). 2004. The Handbook of Pragmatics. Oxford:
Blackwell.
Huang, Yan. 1991. “A Neo-Gricean Pragmatic Theory of Anaphora.” Journal of Linguistics 27:
301–335. doi: 10.1017/S0022226700012706
Huang, Yan. 1994. The Syntax and Pragmatics of Anaphora. Cambridge: Cambridge University
Press. doi: 10.1017/CBO9780511554292
Huang, Yan. 1998. Lecture notes on Lexical Pragmatics. Unpublished ms, University of Oxford
and University of Reading.
Huang, Yan. 2000. Anaphora: A Cross-linguistic Study. Oxford: Oxford University Press.
Huang, Yan. 2004. “Anaphora and the Pragmatics-Syntax Interface.” In The Handbook of
Pragmatics, ed. by Laurence R. Horn and Gregory Ward, 288–314. Oxford: Blackwell.
Huang, Yan. 2007. Pragmatics. Oxford: Oxford University Press.
Huang, Yan. 2009. “Neo-Gricean Pragmatics and the Lexicon.” International Review of Pragmatics
1: 118–53. doi: 10.1163/187731009X455866
Huang, Yan. 2013a. “Micro- and Macro-pragmatics: Remapping Their Terrains.” International
Review of Pragmatics 5: 129–162. doi: 10.1163/18773109-13050106
Huang, Yan 2013b. “Unarticulated Constituents in Neo-Gricean Pragmatics.” Paper presented at
the 1st International Pragmatic Conference of the Americas.
Huang, Yan. 2014. Pragmatics (2nd edition). Oxford: Oxford University Press.
Huang, Yan. 2015a. “Lexical Cloning in English: A Neo-Gricean Pragmatic Analysis.” Journal of
Pragmatics 86: 80–85. doi: 10.1016/j.pragma.2015.06.005
Huang, Yan. 2015b. “Neo-Gricean Pragmatic Theory of Conversational Implicature.” In The
Oxford Handbook of Linguistic Analysis, ed. by Bernd Heine and H. Narrog (2nd edition),
615–639. Oxford: Oxford University Press.
Huang, Yan. 2016. “Pragmatics: Language in Context.” In The Routledge Handbook of Linguistics,
ed. by Keith Allen, 205–220. London: Routledge.
Huang, Yan. (2017a). “Neo-Gricean Pragmatics.” In The Oxford Handbook of Pragmatics, ed. by
Yan Huang. Oxford: Oxford University Press, 47–78.
doi: 10.1093/oxfordhb/9780199697960.001.0001
Huang, Yan. (2017b). “Implicature.” In The Oxford Handbook of Pragmatics, ed. by Yan Huang.
155–179. Oxford: Oxford University Press. doi: 10.1093/oxfordhb/9780199697960.001.0001
Kennedy, Chris. 2007. “Vagueness and Grammar: The Semantics of Relative and Absolute
Gradable Adjectives.” Linguistics and Philosophy 30: 1–45. doi: 10.1007/s10988-006-9008-0
Kennedy, Chris, and Louise McNally. 2005. “Scale Structure, Degree Modification and the
Semantics of Gradable Predicates.” Language 81: 345–381. doi: 10.1353/lan.2005.0071
Korta, Kepa, and John Perry. 2011. Critical Pragmatics. Cambridge: Cambridge University Press.
doi: 10.1017/CBO9780511994869
Lahav, Ran. 1993. “The Combinatorial–connectionist Debate and the Pragmatics of Adjectives.”
Pragmatics and Cognition 1: 71–88. doi: 10.1075/pc.1.1.06lah
Levinson, Stephen C. 1983. Pragmatics. Cambridge: Cambridge University Press.
Implicature. Cambridge, MA: MIT Press.
MacFarlane, John. 2007. “Semantic Minimalism and Nonindexical Contextualism.” In Context-
Sensitivity and Semantic Minimalism: New Essays on Semantics and Pragmatics, ed. by
Gerhard Preyer and Georg Peter, 240–250. Oxford: Oxford University Press.
94 Yan Huang
Palmer, Frank. 1981. Semantics (2nd edition). Cambridge: Cambridge University Press.
Quine, Willard V. 1960. Word and Object. Cambridge, MA: MIT Press.
Recanati, François. 2004. Literal Meaning. Cambridge: Cambridge University Press.
Recanati, François. 2010. Truth-conditional Pragmatics. Oxford: Oxford University Press.
doi: 10.1093/acprof:oso/9780199226993.001.0001
Saul, Jennifer. 2002. “What Is Said and Psychological Reality: Grice’s Project and Relevance
Theorists’ Criticisms.” Linguistics and Philosophy 25: 347–372. doi: 10.1023/A:1015221313887
Searle, John R. 1980. “The Background of Meaning.” In Speech Act Theory and Pragmatics, ed. by
John R. Searle, et al., 221–232. Dordrecht: Reidel. doi: 10.1007/978-94-009-8964-1_10
Searle, John R. 1983. Intentionality: An Essay in the Philosophy of Mind. Cambridge: Cambridge
University Press. doi: 10.1017/CBO9781139173452
Sperber, Dan and Wilson, Deirdre. 1986. Relevance: Communication and Cognition. Oxford:
Blackwell.
Sperber, Dan and Wilson, Deirdre. 1995. Relevance: Communication and Cognition. 2nd edition.
Oxford: Blackwell.
Wałaszewska, Ewa. 2015. Relevance-theoretic Lexical Pragmatics: Theory and Applications.
Newcastle upon Tyne: Cambridge Scholars Publishing.
Wilson, Deirdre, and Robyn Carston. 2007. “A Unitary Approach to Lexical Pragmatics: Relevance,
Inference and ad hoc Concepts.” In Pragmatics, ed. by Noel Burton-Roberts, 230–259.
London: Palgrave Macmillan. doi: 10.1093/oxfordhb/9780199234769.003.0018
Chapter 5
Zero subject anaphors and extralinguistically

motivated subject pro-drop in Hungarian
language use
Enikő Németh T.
University of Szeged
The present chapter aims to analyse implicit subject arguments in Hungarian

language use, focusing especially on the zero subject anaphors and extralin-
guistically motivated subject pro-drop phenomena in utterance and discourse
contexts. Zero anaphors are types of implicit arguments which have their own
position in the syntactic structure of utterances and antecedents with which
they are coreferential. When an implicit argument obtains its interpretation ex-
tralinguistically, for instance, from the physical context, it is a manifestation of
exophoric reference. Exophoric implicit arguments do not have any antecedents;
instead, they refer to items in the external world. On the basis of a detailed anal-
ysis of these kinds of implicit arguments, the chapter has two interrelated con-
clusions. Firstly, the use or interpretation of zero subject anaphors and exophoric
subject pro-drop phenomena in Hungarian language use predicted by grammar
can be considered only a typical, default use and interpretation that emerge due
to the lack of any pieces of information from the encyclopaedic knowledge, gen-
eral pragmatic knowledge and/or specific context. And secondly, grammar and
pragmatics intensively interact in the course of use and interpretation of implicit
arguments.
Keywords: implicit arguments, zero subject anaphors, exophoric reference,

subject pro-drop, interaction between grammar and pragmatics, principle
of relevance
1. Introduction
In the past two decades the investigation of the lexically unrealised verbal argu-
ments in various languages has led to a conclusion that the occurrence and iden-
tification of implicit arguments are guided by grammatical and pragmatic factors
doi 10.1075/pbns.276.05nem
96 Enikő Németh T.
(cf. e.g. Cote 1996; García Velasco and Portero Muñoz 2002; Goldberg 2005, 2013;
Cummins and Roberge 2005; Iten et al. 2005; Scott 2006). Although the different
approaches to implicit arguments are rival explanations allocating different roles
to the various grammatical and pragmatic factors, they agree that an intensive
interaction between grammar and pragmatics has to be acknowledged.
The recent studies on lexically unrealised arguments in Hungarian (cf. e.g.
Dankovics 2005; Pethő and Kardos 2009; Németh T. 2010; 2014a, b, Németh T.
and Bibok 2010) have resulted in the same conclusion. Relying on this conclusion,
implicit arguments can be defined in a complex manner. Implicit arguments are
arguments involved in the lexical-semantic representations of verbs which, how-
ever, are lexically unrealised, and whose implicit presence in utterances is attested
by lexical-semantic, grammatical and pragmatic evidence (Németh T. 2012, 2014a,
b). Although the concept of grammar in its broader sense also includes lexical-
semantic information, it is worth mentioning lexical-semantic information sepa-
rately in the definition because of the orderliness of the licensing and interpretation
mechanisms of implicit arguments. Grammatical factors can include phonological,
morphological, syntactic and semantic constraints, while pragmatic factors refer to
the information from the general encyclopaedic knowledge and general pragmatic
knowledge as well as particular contextual information. The separate mention of
the lexical-semantic, grammatical and pragmatic factors in the definition does not
mean that these factors license the use of implicit arguments independently of each
other; on the contrary, their intensive interaction must be emphasised.
Defining implicit arguments from such an integrative, complex point of view
requires some essential theoretical and methodological decisions (Németh T. 2014a,
b, 2017). First, if we acknowledge an interaction between lexical-semantic, gram-
matical and pragmatic licensing factors, we have to accept that grammar (including
lexical-semantic properties) and pragmatics are not independent of each other but
they are two interacting components. Second, implicit arguments must be studied
not only in the sentential environment of language but in the utterances of lan-
guage use in particular contexts. By investigating lexically unrealised arguments in
utterances of language use those occurrences become describable and explainable
which are excluded from the research because of their grammatical ill-formedness
in sentence oriented frameworks. And third, the application of such a complex defi-
nition of implicit arguments also influences the spectrum of data in the research.
In addition to data of sentence oriented approaches gained from the researcher’s
or other native speakers’ intuitions, we can rely on data from various other direct
sources such as physical observation, spoken and written corpora, and thought
experiments. The integration of data from these different sources also widens the
scope of the research.
Chapter 5. Zero subject anaphors and extralinguistically motivated subject pro-drop 97
In Hungarian, as in other pro-drop languages, verbs occur with lexically unreal-

ised arguments more frequently and in ways controlled differently than in non-pro-
drop languages (Németh T. 2000, 2001). In Hungarian, verbs vary as to the manners
and contexts in which they can be used with implicit arguments. In Hungarian
language use, verbs can occur with implicit arguments in the following three ways
(Németh T. 2012, 2014a, b, 2017):
A. If some element of the lexical-semantic representation of a verb licenses the
lexically unrealised occurrence of an argument, according to the principle of
relevance.
B. If the rest of the utterance, i.e. immediate context with its contextual factors
including encyclopaedic pieces of information and grammatical requirements
provides a relevant, typical interpretation.
C. If extending the immediate utterance context of the argument results in an
interpretation consistent with the principle of relevance.
These three manners of occurrence of Hungarian verbs with lexically unrealised

arguments can be applied to subject, direct object, and adverbial arguments as well.
It is obvious from the above formulation of the three possibilities of occurrence of
Hungarian verbs with implicit arguments that they are based on the kind and size
of the context (Németh T. 2010: 306). In case (A) the lexical semantic-representa-
tion of the verbs, in case (B) the immediate utterance contexts in which verbs are
involved, and in case (C) the extended utterance contexts have to be considered
in addition to other factors to license the use of implicit arguments. Finally, it is
worth emphasising the role of the principle of relevance. The cognitive principle of
relevance (Sperber and Wilson 1986/1995: 260) underlies the use and interpreta-
tion of implicit arguments (Németh T. 2010: 308–309). As to their usage, with the
use of lexically unrealised arguments, highly accessible for both the speaker and
the hearer, the speaker can achieve the same effects as with explicit forms but with
less production effort. As to the interpretation of implicit arguments, the hearer
assumes that the speaker’s utterance involving an implicit argument is relevant
to him/her (see Wilson and Kolaiti, this volume). Therefore he/she attempts to
interpret it starting from the lexical-semantic representation of the verb. If this
does not yield a relevant interpretation, then he/she turns to the second and after
that to the third manner. In other words, the hearer takes into consideration the
lexical-semantic representation of the verbs, the immediate and extended utterance
contexts only to a necessary extent.
The present chapter aims to analyse implicit subject arguments in Hungarian
language use. A special focus will be given to the zero subject anaphors and extra-
linguistically motivated subject pro-drop phenomena in utterance and discourse
contexts. Thus, implicit subject arguments which occur in the second (B) and the
third (C) manners will be studied taking into consideration the interaction between
grammatical and pragmatic licensing factors.
The organisation of the chapter is as follows. After this introductory Section 1,
in Section 2, I will briefly discuss how the term anaphor can be interpreted in
various rival approaches and clarify how I define zero anaphors. In addition, I will
also explain what I mean by extralinguistically motivated pro-drop phenomena. In
Section 3, I will examine zero subject anaphors and extralinguistically motivated
subject pro-drop in Hungarian language use. Finally, in Section 4, I will summarise
the conclusions.
2. Anaphors, zero anaphors and extralinguistically motivated pro-drop

phenomena
In Hungarian language use, there are various types of implicit arguments, and their
presence in utterances can be demonstrated by various kinds of evidence. The sepa-
rate mention of the various licensing factors in the definition of implicit arguments
provided above in Section 1 indicates the diversity of the different types of implicit
arguments. The various sorts of evidence point to the presence of an implicit argu-
ment and attest that the particular utterance does not show any performance error.
Zero anaphors are types of implicit arguments which have their own position in the
syntactic structure of utterances and antecedents with which they are coreferential
and coindexed (cf. Németh T. 2007; Németh T. and Bibok 2010).
Zero anaphors differ from implicit arguments occurring in the first (A) manner,
which are not projected into the syntactic structure of utterances but are available
through the lexical-semantic representations of verbs as background information.
For instance, the verbs of natural phenomena such as havazik ‘snow’, villámlik ‘[for
lightning to] strike, hajnalodik ‘[for day to] break’, and esik ‘rain’ as well as verbs
of work such as vet ‘sow’, arat ‘harvest’, takarít ‘clean’, and gyógyít ‘treat, cure’ can
occur with implicit subject arguments in Hungarian language use because they
constrain their subject arguments in the lexical-semantic representation by means
of selection restrictions (Németh T. 2012, 2014a). If these verbs occur without a
lexically realised subject argument, their subject can be identified with the pieces
of information available with the help of selection restrictions. The lexical-semantic
representation of these verbs can be interpreted in two ways. Either only the events
denoted by the verbs are in the focus of attention or their subject as well. When only
the events are in the focus of attention, the subject arguments can be left implicit
and identified with the information provided by selection restrictions. When the
subject arguments are also in the focus of attention then they must be lexicalised.
In the case of verbs of natural phenomena and verbs of work the lexically unre-
alised subject arguments can be identified with the information available through
the selection restriction imposed on the subject arguments in the lexical-semantic
representation. Consider the lexical-semantic representation of the verbs havazik
‘snow’ and vet ‘sow’.
(1) havazik ‘snow’: ‘x which is snow falls in y which is place’
[[x FALL_IN y] : [[SNOW x] & [PLACE y]]]
(2) vet ‘sow’: ‘action of x who is a peasant causes that y which is seed goes
into z which is field’
[[[ACT x] : [PEASANT x]] CAUSE [[GO y] : [[SEED y] &
[[FIN [LOC y] ⊂ LOC z] : [FIELD z]]]]]
(Németh T. 2012: 468)
The verb havazik ‘snow’ has a cognate subject hó ‘snow’ which is involved in its lex-
ical-semantic representation. We can see in (1) that the verb havazik ‘snow’ places
a unique selection restriction on its subject argument x: ‘x is snow’. Since the verb
havazik ‘snow’ is morphologically derived from the noun hó ‘snow’, predicted by
the selection restriction to be the unique subject argument of the verb in question,
the verbal morphological structure is absolutely transparent. This transparent mor-
phological structure, as well as the phonetic form of the verb havazik ‘snow’, easily
recalls the subject argument hó ‘snow’, also predicted by the selection restriction.
Therefore, the explicit occurrence of hó ‘snow’ is disturbingly redundant with the
verb havazik ‘snow’, with which the noun hó ‘snow’ can henceforth appear only as
an implicit subject (Németh T. 2014a: 128). Consider also (3). 1
(3) Tegnap óta havazik [a hó] 1i. Már ötven
yesterday.nom since snows.indef the snow.nom already fifty.nom
centiméter [hó]i esett.
centimetre.nom snow.nom fell.indef.3sg
‘[Snow] has been snowing since yesterday. Already fifty centimetres [of snow]
have fallen.’ 2
1. Implicit arguments are provided in square brackets throughout the chapter.

2. The abbreviations used in the glosses throughout this chapter are the following: 1 – first
person, 2 – second person, 3 – third person, sg – singular, pl – plural, imp – imperative, indef –
indefinite (conjugation), def – definite (conjugation), inf – infinitive, pvb – preverb, nom – nom-
inative, acc – accusative, all – allative, dat – dative, del – delative, ill – illative, sub – sublative,
and sup – supperessive.
The possibility that in the second utterance in (3) one can refer to a hó ‘the snow’
by a zero anaphor supports the plausibility of the analysis based on an implicit
subject argument occurring with the verb havazik ‘snow’ as in the first utterance
of (3). At the same time it weakens the plausibility of the widely accepted analysis
in the Hungarian grammatical tradition according to which the noun hó ‘snow’ is
incorporated into the verb havazik ‘snow’ (Németh T. 2014a: 128).
As for the verb vet ‘sow’, it also puts a selection restriction on its subject ar-
gument which must be PEASANT. Verbs of work strictly and, in some cases even
uniquely (cf. e.g. Gyóntatnak ‘[Priests] confess [penitents]’), constrain what can fill
the subject position, since the general encyclopaedic knowledge about persons who
perform various kinds of work is built into their lexical-semantic representation
via selection restrictions (Németh T. 2012: 469). When the verbs of work are used
with implicit subject arguments with general readings, the subject arguments are
available through the background information provided by the selection restric-
tions. They can be identified with the typical persons who are used to performing
the particular work. The verbs of work occurring with implicit subject arguments
in such a way are conjugated for the indefinite form 3pl, cf. (4). If a verb of work
is conjugated for the 3sg form as in (5), the implicit subject argument cannot be
identified with the background information provided by the selection restriction;
instead, it requires a contextual interpretation, i.e. it can be identified not in the
first (A) but in the third (C) manner.
(4) Ősszel és tavasszal is vetnek [PARASZTOK] gabonát
in.autumn and in.spring also sow.indef.3pl peasants grain.acc
‘In fall and also in spring [the peasants] sow grain’. (Németh T. 2012: 466)
(5) (A father is driving and his son is looking out the window. The son notices a
man who is sowing.)
– Nézd apa! Vet.
look.imp.def.2sg daddy.nom sows.indef
While in (4) one can interpret the implicit subject argument relying only on the lex-
ical-semantic representation of the verb vet ‘sow’ and the verbal inflection indefinite
3pl form, in (5) the father has to extend the utterance context from the observable
physical environment to identify the implicit subject argument.
Let us return back to zero anaphors. Zero anaphors are not only types of im-
plicit arguments but lexically unexpressed types of anaphors. The vast majority
of the literature on anaphors, e.g. in text linguistics, discourse analysis, pragmat-
ics, and computational linguistics, agrees that anaphors refer back to explicitly
expressed words or phrases (cf. e.g. Huang 2000; Tolcsvai Nagy 2001; Lejtovicz
and Kardkovács 2007) and these words or phrases serve as antecedents for the
anaphors. From this, it follows that the antecedents of zero anaphors should also
occur in the previous parts of the utterances in which the zero anaphors are used,
or in places in the previous discourse outside the scope of the utterances containing
the zero anaphors, i.e. they must be identifiable in the preceding discourse context.
This characterisation of zero anaphors shows that they are not considered purely
sentential phenomena in this chapter but as phenomena that should be analysed in
utterance contexts and the discourse contexts of language use. Consider (6) and (7).
(6) (The father arrives home late in the evening, looks around and cannot see his
son. The mother says:)
Andrási fáradt volt, [Ø]i lefeküdt aludni.
András.nom tired was.3sg pvb.lay.indef.3sg to.sleep
‘András was tired, he went to bed.’
(7) (At a party two friends are chatting.)
– Új szomszédunki van. Tegnap átjött [Ø]i
new our.neighbour.nom is yesterday pvb.came.indef.3sg
bemutatkozni, kedves embernek látszik [Ø]i
pvb.introduce.inf kind man.dat seems.indef
– Megkínáltad [Ø]i?
pvb.offered.def.2sg
– Igen, de [Ø]i nem fogadta el.
yes but not accepted.def.3sg pvb
‘– We have a new neighbour, he came over yesterday, and he seems to be a
nice person.
– Did you offer him something [to drink or eat]?
– Yes, but he did not accept it.’
In (6) the subject argument of the verb lefeküdt ‘pvb.lay’ in the second part of the
utterance is left implicit as a zero anaphor [Ø]i which is coreferential with Andrási,
the subject argument of the predicate phrase fáradt volt ‘was tired’ in the first part
of the utterance. In (7) the zero subject argument [Ø]iof the verbs átjött ‘pvb.came’
and látszik ‘seems’ in the second utterance, the zero object argument [Ø]i of the
verb megkínáltad ‘pvb.offered’ in the third utterance, and the zero subject argu-
ment of the verb elfogadta ‘pvb.accepted’ in the fourth utterance are coreferential
with the explicitly expressed subject argument szomszédunki‘our neighbour’ in the
first utterance. In (7) zero anaphors operate at the discourse level across utterance
boundaries. In both (6) and (7) the zero anaphors have an antecedent in the previ-
ous part of the utterance or in previous utterances in the discourse.
However, it must be noted that it can happen that the antecedent of a zero
anaphor is not explicitly expressed in the utterance or discourse context but is
itself an implicit argument which can be identified with the help of a piece of
information from the observable physical context or encyclopaedic knowledge.
Consider (8) and (9):
(8) (Péter, András, and Jakab are distributing bread to children.)
Péter – Mindenki kapott [kenyeret]i phys?
everybody.nom got.indef.3sg bread.acc
András – Nem. Pálnak nincsen [Ø]i.disc
no Pál.dat not.is.indef
Jakab – Péter adott Pálnakj [Ø]i. disc Biztosan már
Péter.nom give.past.indef.3sg Pál.dat certainly already
megette [Ø]j disc [Ø]i. disc
pvb.ate.indef.3sg
‘Péter: – Has everybody got bread?
András: – No. Pál does not have any.
Jakab: – Péter gave [bread] to Pál. He must have eaten [it] already.’ 3
(9) (A customer would like to pay with a bank card in a supermarket. The shop
assistant turns the card reader to the customer and asks him/her:)
– Legyen szíves, üsse be [a Pin-kódját]i enc!
be.imp.indef.3sg please punch.imp.def.3sg in your.Pin-code.acc
– Jaj, nem jut eszembe [Ø]i
ouch not gets.indef my.mind.ill
‘– Please, punch [your PIN] in!
– Oh no, [it] doesn’t come to mind.’
In the interlinear rendering, the abbreviation disc marks the fact that the identifi-
cation of implicit arguments is possible by means of pieces of information coming
from the preceding discourse, including the previous parts of the same utterance
in which the implicit argument in question is included. The abbreviation phys
marks pieces of information originating in the observable physical environment,
and the abbreviation enc signals that the assumption by means of which the implic-
it argument is identified is encyclopaedic. In (8) the direct object argument of the
verb kapott ‘got’ is left lexically unrealised. It is easily identifiable by extending the
utterance context with the information from the physical context: [kenyeret]i phys
‘bread’. Although the direct object argument of the verb kapott ‘got’ has a zero form,
it cannot be categorised as a zero anaphor, since it does not have any antecedent
3. This conversation was also analysed in Németh T. (2001) in order to exemplify the various
ways of context extension. The focus of analysis here is not on the context extension but on the
zero anaphors.
in the utterance or previous discourse with which it can be coreferential and coin-
dexed. The implicit arguments indicated as [Ø]i in the subsequent part of the con-
versation are coreferential with the lexically unrealised direct object argument in
the first utterance [kenyeret]i phys. Thus, the implicit subject argument in the second
utterance of András’s turn Pálnak nincsen [Ø]i.disc. can be considered a zero subject
anaphor with its antecedent in the first utterance of the conversation.
A similar analysis can be proposed for (9), with the difference that the im-
plicit direct object argument [Ø]i in the second utterance as a zero anaphor has
its antecedent in the first utterance identifiable not from the observable physical
environment as in (8) but from encyclopaedic knowledge. In a situation where a
shopping event takes place in a supermarket, a customer who would like to pay with
a bank card is expected to know that her/his PIN is required to use the card reader.
Since this piece of encyclopaedic information is assumed to be highly available
to card owner customers, the shop assistant can leave the direct object argument
[a Pin-kódját]i enc ‘your PIN’ of the verb üsse be ‘punch in’ lexically unrealised.
Similarly to the lexically unrealised direct object argument in the first utterance
[kenyeret]i phys in (8), the encyclopaedically licensed implicit occurrence of the
direct object argument of the verb üsse be ‘punch in’ in (9) cannot be categorised
as a zero anaphor either. At the same time it can serve as an antecedent of the [Ø]i
zero subject anaphor of the second utterance.
Above when the use of verbs of natural phenomena with implicit subject ar-
guments was discussed, it was argued that the implicit subject arguments of verbs
of natural phenomena identifiable by means of selection restrictions can serve the
antecedents for zero anaphors or, to put it the other way round, the zero anaphors
occurring in the same or subsequent utterances attest that verbs of natural phe-
nomena are used with implicit subject arguments with which the zero anaphors are
coreferential. In (3) the zero subject argument of the verb esett ‘fell’ was analysed
in such a way. Let us take another example.
(10) (The weather forecast predicts snow. The daughter living in Szeged calls her
mother to find out what the weather is like in Celldömölk where the mother
lives.)
– Szegeden szállingózik [a hó]i. Celldömölkön
Szeged.sup softly.snows.indef the snow.nom Celldömölk.sup
esik [Ø]i már?
falls.indef already
‘In Szeged it [= the snow] is snowing softly.’ Does [= the snow] fall yet in Celldömölk?
– Már szakad [Ø]i.
already pours.indef
‘It [= the snow] is pouring with snow already.’
In the first utterance in (10), the verb szállingózik ‘snow softly’ occurs with an
implicit subject argument in the first (A) manner by means of the selection re-
striction. The lexical-semantic representation of szállingózik ‘snow softly’ puts a
selection restriction on its subject argument; it must be hó ‘snow’, i.e. x: hó ‘snow’.
This unique selection restriction makes it possible to use the verb szállingózik
‘snow softly’ without an explicit subject argument. This implicit subject argument
available as background information by means of the unique selection restriction
in the lexical-semantic representation of szállingózik ‘snow softly’ can serve as an
antecedent for the lexically unrealised subject argument of the verb esik ‘fall’ in
the daughter’s question. The implicit subject argument of esik ‘fall’ is a zero subject
anaphor [Ø]i. Similarly, in the mother’s answer, the verb szakad ‘pour’ also occurs
with the implicit subject argument [Ø]i in the third (C) manner. Both the implicit
subjects of esik ‘fall’ and szakad ‘pour’ are zero discourse anaphors.
When an implicit argument has a postcedent, it is called a zero cataphor. In
other terms, the expression with which the cataphor is coreferential follows the
cataphor later in the subsequent utterance or discourse. Consider (11), where the
zero subject in the first part of the utterance gets its interpretation from the second
part of the utterance, since [Ø]i is coreferential with the subject of the second clause
Andrási. Thus [Ø]i can be considered a cataphor.
(11) (The father arrives home late evening, looks around and does not see his son.
The mother says:)
Mivel [Ø]i fáradt volt, Andrási lefeküdt aludni.
Since tired was.indef.3sg András.nom pvb.lay.indef.3sg to.sleep
‘Because he was tired, András went to bed.’
In the literature one can also find a broad definition of anaphors, according to which
anaphors are uses of expressions the identification of which is dependent upon
other expressions in context, either their antecedents in the previous discourse or
postcedents in the subsequent discourse. Thus, zero forms categorised as cataphors
elsewhere are also covered by the term anaphor in a wider sense (Lyons 1977: 659;
Ehlich 1982; Huang 2012). Cataphors are sometimes called backward anaphors in
the generative linguistics tradition as well (cf. e.g. Mittwoch 1983; Reinhart 1985). 4
However, the term anaphor is only applied in the present paper to expressions which
have antecedents. Expressions with postcedents are called cataphors in accordance
with the majority of the literature (cf. e.g. Tolcsvai Nagy 2001a, b; Renkema 2004;
Huang 2009) and will not be discussed in more detail in the subsequent part of this
4. The other terms used for the phenomenon of cataphor are as follows: counter-unidirectional
anaphor, anticipatory anaphor, forward-looking anaphor, prospective anaphor, and cataphoric
reference (Huang 2012: 46).
paper, since they have coindexing identification principles similar to those which
govern anaphors. Both anaphors and cataphors are endophors, since they obtain
their antecedents and postcedents respectively from the previous or subsequent
part of the utterance or discourse (Gillon 2012).
When an implicit argument obtains its interpretation extralinguistically, for
instance, from the physical context, it is a manifestation of exophoric reference.
Exophoric implicit arguments do not have any antecedents or postcedents in the
utterance or discourse context they occur in; instead, they refer to items in the
external world (cf. Peral and Fernández 2000; Gillon 2012). Such implicit argu-
ments include zero personal pronouns which refer to the discourse participants,
i.e. the speaker and the hearer, and can be classified as zero exophoric pronouns
or zero deictic referential pronouns (Laczkó 2003; Kim et al. 2010; Mitkov 2013).
Consider (12).
(12) (Two students are discussing how to solve a problem. Both have their own ideas.
Finally, one of them suggests a solution which seems to be plausible and asks
the other:)
Megtaláltam a megoldást. Egyetértesz?
pvb.found.def.1sg the solution.acc agree.indef.2sg
‘I have found the solution. Do you agree with me?’
In (12) the verb megtaláltam ‘found’ occurs with a lexically unrealised 1st person
subject argument which refers extralinguistically to the speaker, as is indicated
by the inflectional morpheme -m. The verb egyetértesz ‘agree’ also occurs with a
zero subject exophorically; it refers to the hearer, as is indicated by the inflectional
morpheme -sz. Zero exophoric implicit arguments including zero exophoric pro-
nouns are not considered zero anaphors in the present study. Hungarian verbs can
occur with them in the third (C) manner by taking into account the inflectional
morphemes of the indefinite/definite conjugation paradigm on the verbs and ex-
tending the immediate utterance context in which the verbs in question are used
with the information from the observable physical environment or with encyclo-
paedic knowledge.
It must be emphasised that an implicit argument can obtain its interpretation
extralinguistically not only from the physical context through exophoric reference
as is widely assumed in the literature, but also from the encyclopaedic context as we
have seen above in (9), where the implicit direct object argument of the verb üsse
be ‘punch in’ can be identified with the PIN from encyclopaedic knowledge. This
type of implicit argument cannot be categorised as a zero anaphor either; it can be
identified by extending the immediate utterance context in which the verb occurs
with the information from the encyclopaedic knowledge in the third (C) manner.
It is also worth mentioning that the term anaphor differs from the conception
of anaphors in the generative grammatical tradition. In the Chomskyan types of
generative grammars anaphors have a narrower scope and are defined as expres-
sions which have dependent reference taken from an antecedent within the same
phrase or sentence (Radford 1997: 492). Anaphors include reflexives (cf. e.g. the
English reflexive pronouns myself/yourself/herself/themselves, etc.) and reciprocals
(cf. e.g. English reciprocal pronoun each other) which must be bound by an ante-
cedent in the same phrase or sentence (Chomsky 1981; Radford 1997: 114–116).
The same analysis can be provided for the behaviour of Hungarian reflexive pro-
nouns (magam ‘myself ’, magad ‘yourself ’, maga ‘herself/himself ’, etc.) as well as
the reciprocal pronoun egymás ‘each other’ (cf. É. Kiss 1985, 1998). Unlike in the
generative grammatical tradition, in the present chapter the term anaphor is applied
in harmony with the line of thinking presented previously in this Section, similarly
to text linguistics, discourse analysis, pragmatics, and computational linguistics, as
has been referred to above.
After discussing how the term anaphor is interpreted in various rival approach-
es and clarifying what I mean by zero anaphors, I wish to examine zero subject
anaphors in Hungarian language use.
3. Zero subject anaphors and extralinguistically motivated subject

pro-drop in Hungarian language use
Since Hungarian is a pro-drop language, there is a wide range of lexically un-

realised subjects in Hungarian language use. 5 However, not all kinds of subject
pro-drop phenomena can be considered zero subject anaphors. Zero subject an-
aphors are only those subject pro-drop phenomena which have an antecedent in
the previous parts of utterances or discourses in which they occur. As we have
seen above, zero anaphors in utterances and discourses which act similarly to
overt anaphoric pronouns refer to referents that have already been introduced
into the utterance or discourse, i.e. they have an antecedent with which they are
coreferential. Those subject pro-drop phenomena which do not have an antecedent
in the previous part of the utterance or discourse, however, obtain their interpre-
tation extralinguistically from the speech situation in the physical context cannot
5. Analysing occurrences of verbs of natural phenomena and verbs of work with implicit subject
arguments in Section 2, we have seen that not all kinds of implicit subjects are pro-drop phenom-
ena. These verbs can occur with implicit subject arguments in the first (A) manner, i.e. they are
not projected into the syntactic structure of utterances but are available through lexical-semantic
representations of the verbs as background information.
be categorised as zero subject anaphors. They have been termed exophoric zero
pronouns, as we have seen above. Thus, zero first and second personal subject
pronouns which refer to discourse participants such as the speaker and hearer
are also manifestations of subject pro-drop. Dropped subjects are indicated by the
first and second personal inflectional morphemes on verbs but they are not zero
anaphors (cf. (12) above). Third personal subject pronouns can also be dropped
on the basis of extralinguistical information, when they can be identified from the
physical or encyclopaedic contexts.
Third person zero subject anaphors which can be considered manifestations of
subject pro-drop are also indicated by means of the inflectional morphemes on the
verbs. Previous research into zero subject anaphors in Hungarian has demonstrated
that the licensing and interpretation of zero subject anaphors are influenced not
only by grammatical requirements but by pragmatic factors as well (cf. e.g. Pléh
1994, 1998; Tolcsvai Nagy 2001; Dankovics 2001, 2005; Németh T. 2007; Németh
T. and Bibok 2010). Results of Pléh’s and his colleagues (Pléh 1994, 1998: 164–194;
Pléh and Radics 1978; Pléh and McWhinney 1987) as well as of Dankovics’s (2001,
2005) psycholinguistics experiments have also supported the hypothesis that, on the
one hand, the grammar cannot account for all pro-form relations, and, on the other
hand, encyclopaedic information and particular contextual factors can override the
use or interpretation predicted by grammar. Testing the influence of constituency
constraints on interpretations of coreference in Spanish, Blackwell (2001) also con-
cludes that anaphor interpretations are constrained not only by grammatical (syntac-
tic) factors. Instead, they are strongly constrained by semantic entailments, general
semantic predictions, background knowledge, antecedent salience, and choice of
linguistic alternates. Similarly, studying null appositives, Capone (2008) comes to
the conclusion that often the semantics of sentences does not by itself result in a
complete and adequate interpretation, and, instead, pragmatics must contribute to
the identification of implicit constituents (Németh T. and Bibok 2010: 510).
Pléh (1994) also emphasises that when the antecedent of an anaphor is highly
accessible in the context this means that it can be identified straightforwardly, and
consequently the anaphor can have a zero form. A similar claim is made by Ariel
(1991), when she places referring expressions on an accessibility scale on the basis
of their cognitive accessibility and linguistic cues. Relevance theory also takes this
view (cf. Sperber and Wilson 1986/1995). Thus, on the basis of these economy
(relevance) considerations, we can assume that when the referent of an anaphor
can be easily and unambiguously identified in the context, there is no need to use
an explicit lexical form in Hungarian language use, i.e. the anaphor can be left im-
plicit (Pléh 1998: 167; Németh T. and Bibok 2010). The distribution of the forms
of anaphors az/azok ‘that/those, ő/ők ‘s/he / they’, and [Ø] supports this relevance
consideration (cf. also above this Section as well as Pléh 1998: 164–194).
Tolcsvai Nagy (2001: 289–290) comes to similar conclusions. He overviews the

possible realisations of third person subject anaphors in Hungarian and emphasis-
es that when there is one potential antecedent in the position of subject, topic and
agent or experiencer, and the anaphor itself is in an unstressed subject and topic
position, the anaphor is manifested by the zero form and inflectional morpheme
on the verb. If the anaphor in a stressed subject or focus position is human, it is
manifested as a third person personal pronoun ő ‘s/he’ or ők ‘they’. If the anteced-
ent is inanimate and the anaphor is in a stressed subject position, it is realised by
the third person demonstrative pronoun az ‘that’ or azok ‘those’ (Tolcsvai Nagy
2001: 289–290). Since the aim of the present study is to investigate implicit subject
arguments including zero subject anaphors in Hungarian language use, I do not
intend to deal with ő/ők ‘s/he/they’ and az/azok ‘that/those’ anaphors here. The
choice between the various possible forms of anaphors in Hungarian language
use is motivated by the accessibility of their referents. Zero realisations of subject
anaphors are supported by the high accessibility of the referents of anaphors.
Consider (13) and (14).
(13) Péteri hazament. Már [Ø]i befejezte
Péter.nom home.went.indef.3sg already pvb.finished.def.3sg
a munkáját.
the his.work.acc
‘Péter went home. He had already finished his work.’
(Tolcsvai Nagy 2001: 288)
(14) Az eurói árfolyama tovább emelkedik. [Ø]I
the euro.nom its.exchange.rate.nom further rises.indef
Elérheti a 315 forintot.
may.reach.def.3sg the 315 forint.acc
‘The exchange rate of the euro is still rising. It may reach 315 HUF.’
In (13) the antecedent of [Ø]i is human (Péteri) while in (14) it is inanimate (euroi),
i.e. the zero anaphor can be used in both animate and inanimate cases when the
anaphor itself is not stressed. The zero subjects of the second utterance in (13) and
(14) refer to the subject of the first utterance. This is what has been called in the
generative grammatical tradition the subject continuation principle originating
in the pro-drop property of Hungarian. According to this principle, the subject
of the second clause in utterances with complex and compound sentence struc-
tures can be realised as a zero anaphor if it is coreferential with the subject of the
first clause (Pléh 1994, 1998; Németh T. and Bibok 2010; Németh T. 2014b). The
zero anaphor in the second clause has its own position in the syntactic structure
of the utterance as in all cases of overt anaphors. Thus, they differ from the kind
of implicit subject arguments occurrences of which are licensed in the first (A)
manner, i.e. by the lexical-semantic representation of verbs (cf. the analysis of the
occurrence of verbs of natural phenomena as well as verbs of work with implicit
subject arguments in Section 2).
In other words, the basic syntactic rule is that when subjects are repeated they
can be dropped. However, if a noun in the previous clause is selected as an ante-
cedent of the subject in the second clause which was not a subject in that previous
clause, then the subject in the second clause must be pronominalised by the de-
monstrative pronoun az ‘that’ (Pléh and Radics 1978; Pléh and McWhinney 1987;
Németh T. 2012).
Pléh’s (1998: 164–194) and Dankovics’s (2001, 2005) psycholinguistic experi-
ments show that when there is more than one potential antecedent in a previous
clause of the same utterance or discourse, the interpretation of sentential and
discourse anaphors in Hungarian language use are constrained not only by the
syntactic rule of subject continuation but also by various other grammatical fac-
tors, such as sentence-grammatical typicality effects, thematic roles of verbs, types
of conjugation, word order, and pragmatic considerations. However, not all these
factors do in fact influence the occurrence of zero subject anaphors. For instance,
the word order of the previous clause does not have an impact on the antecedent
selection in the case of a zero subject anaphor (Pléh 1998: 177). Tolcsvai Nagy
(2001: 293) criticises these psycholinguistic approaches since they do not reveal
the underlying cognitive basis of the difference between the use of zero pronouns
and the additive pronouns ő/ők ‘s/he/they’ and az/azok ‘that/those’ as anaphors.
He proposes taking into consideration the notion of perspective addressed in
cognitive linguistics (cf. Sanders and Spooren 1997: 85–112) in order to arrive at
an explanation. The use of zero pronouns and additive pronouns ő/ők ‘s/he/they’
and az/azok ‘that/those’ indicates different perspectives (viewpoints) or changes
in perspectives. When the speaker uses a zero subject pronoun, this indicates that
there is no change in the perspective of the two clauses. The clause in which the
anaphor occurs should be interpreted from the same perspective as the previous
clause of the utterance in which the full nouns occur. Thus, the syntactic rule
of subject continuation formulated in the generative grammatical tradition and
applied in Pléh and colleagues’ psycholinguistic experiments can be considered
a syntactic manifestation of the identity of the perspectives of the two clauses.
It must be admitted that Pléh (1998: 175) also refers to the role of the change in
the perspective and indicates that there are cases when the syntactic rules or the
change in the perspective govern the pattern of subject anaphors in Hungarian
language use.
According to the results of the sentence oriented psycholinguistic experiments

mentioned above, when there is more than one potential antecedent in a previous
clause of the same utterance, the subject continuation rule of grammar strongly
predicts that the zero anaphor must be coreferential with the subject of the previous
clause, although there are other factors to be taken into account. However, it must
be highlighted that these results are valid only in a default case, i.e. in the absence
of specific, particular contextual information.
If we consider utterances in particular contexts of language use, it can be ar-
gued that grammar and pragmatics interact intensively and in multiple ways in
the licensing and interpretation of zero subject anaphors (Németh T. and Bibok
2010; Németh T. 2012, 2014b). 6 Pragmatic factors can support, modify or change
the reading predicted by grammar. Consider the case of the zero subject anaphor
in the utterance environment in (15). 7
(15) Az apai elküldte a lányátj a
the father.nom pvb.sent.def.3sg the his.daughter.acc the
nagymamáhozz, mert [Øi/j/z] rosszul érezte magát.
grandmother.all because sick felt.def.3sg himselfi/herselfj,z
‘The father sent his daughter to the grandmother, because [s/he] felt sick.’
Following from the syntactic rule of subject continuation, the zero subject of the
second clause in (15), i.e. [Øi] has to be coreferential with the subject phrase az apai
‘the father.nom’ in the first clause. There is no change in perspective in this inter-
pretation. According to the generalisations made by Pléh (1994, 1998), Dankovics
(2001, 2005), and Tolcsvai (2001), if a speaker wants to refer to the noun phrase a
lányát ‘his daughter.acc’ or a nagymamához ‘the grandmother.all’ in the second
clause, then s/he must use the demonstrative pronoun az ‘that’ as in (16). In both
interpretation possibilities of az ‘that’ the speaker changes the perspective of the
first clause. However, it is worth highlighting that although the reference of the
demonstrative pronoun az is ambiguous from the syntactic and semantic points
of view, since it can be coreferential with both noun phrases a lányát ‘his daughter.
acc’ and a nagymamához ‘the grandmother.all’, the identification of az ‘that’ with
the daughter is preferred pragmatically in the lack of a particular context.
6. Not only in the case of zero anaphors do the grammatical constraints and pragmatic factors
interact. They also interact in all cases of anaphors. For a complex analysis of anaphors in various
languages, see Huang 2004.
7. For a similar kind of treatment of the interaction between grammar and pragmatics in the
case of implicit subject arguments, see the analysis of the utterance A férji elkísérte a feleségétj az
orvoshozz, mert [Øi/j/z]nagyon izgult ‘The husband accompanied his wife to the doctor, because
[s/he] was very nervous.’ in Németh T. and Bibok 2010: 510; Németh T. 2014b: 687.

nagymamáhozz, mert azj/z rosszul érezte magát.
grandmother.all because shej,z sick felt.def.3sg himselfi/herselfj,z
‘The father sent his daughterj to the grandmotherz, because [shej,z] felt sick.’
Let us slightly modify the utterance in (16) and change the noun phrase a nagy-
mamához ‘the grandmother.all’ with the noun phrase a nagymamájához ‘her/his
grandmother’. Consider (17).
nagymamájáhozz, mert azz rosszul érezte magát.
her/his.grandmother.all because shez sick felt.def.3sg himselfi/herselfj,z
‘The father sent his daughterj to her/his grandmotherz, because [shej,z] felt sick.’
In (17) the interpretation according to which the noun phrase a nagymamájához

‘her/his.grandmother.all’ with the possessive suffix -ja preferably refers to the
daughter’s grandmother and the demonstrative pronoun az ‘that’ is coreferential
with the daughter’s grandmother can be considered a default pragmatic reading
based on the encyclopaedic knowledge. Usually granddaughters visit their grand-
mothers because of the latter’s illnesses. Although it must be noted that it is also pos-
sible to identify the reference of the demonstrative pronoun az ‘that’ in (17) with the
daughter as well, in this case a particular context is needed. For instance, the daugh-
ter became sick and her grandmother is a doctor who can treat her granddaughter.
Now, let us return to the examination of the zero subject anaphor in (15).
Pragmatic factors, namely, the particular speech situation, can support the inter-
pretation of the zero subject anaphor in (15) predicted by grammar. For instance,
if in a particular context the father felt sick and sent his daughter to his daughter’s
grandmother, i.e. his mother, to invite her to help him, the pragmatic factors sup-
port the interpretation constrained by grammar, i.e. the antecedent of the zero
anaphor [Øi] is the noun phrase az apai ‘the father.nom’.
However, particular pragmatic factors can override the interpretation of the
zero anaphor predicted by grammar. The speaker can refer to the noun phrase a
lányátj ‘his.daughter.acc’ in the first clause with a zero anaphoric pronoun [Øj] in
the second clause. If there is a particular speech situation in which it is known that
the daughter became sick and her grandmother being a doctor probably could help,
the antecedent of the zero anaphor [Øj] can be the direct object phrase a lányátj
‘his.daughter.acc’ in contrast to the interpretation forced by the grammatical rule
of subject continuation and the explanation by Tolcsvai Nagy (2001) based on the
identity of underlying perspectives. Since in the particular context, the daughter’s
illness is highly accessible, the daughter can be referred to by means of an implicit

form, i.e. a zero anaphor. Because of the high accessibility, there is no need to use
the overt demonstrative pronoun az ‘that’, in spite of the fact that there is a change
in perspective. Consequently, the high accessibility can motivate and license the
use of zero anaphors in particular contexts not only in cases when the perspectives
are the same in both clauses but also when a change in perspective takes place.
Finally, there is a third interpretation possibility. The zero subject anaphor in
the second clause of the utterance in (15) can refer to the grandmother as well. This
third interpretation possibility is more natural then the first two interpretations
take into consideration the general encyclopaedic knowledge. It is quite usual that
grandmothers have health problems and their children and grandchildren attempt
to help them. On the basis of this general background knowledge, the grandmother
can be considered the default interpretation of the zero subject anaphor in (15). A
particular speech situation where a particular grandmother became sick supports
this interpretation. Other particular contexts in which the grandmother was not
sick override this default interpretation and support either the identification of the
implicit subject argument [Ø] with the father predicted by the grammatical rule of
subject continuation or with the daughter. As we have seen above, in this latter case,
the particular speech situation overrides the interpretation predicted by grammar.
Let us take another occurrence of zero subject anaphor as an example where, in
addition to the grammatical requirements, general background world knowledge
and particular contextual information, another kind of pragmatic information,
namely, general pragmatic knowledge should be taken into account to identify the
antecedent of the zero subject anaphor. Consider (18). 8
(18) a. A bácsii emlékezett a fiúrajʹ [Øi/j]
the old.man.nom remembered.indef.3sg the boy.sub
tanácsot adott nekij/i.
advice.acc gave.indef.3sg him.dat
‘The old mani remembered the boyj, [hei/j] had given himj/i advice.’
b. A fiúi emlékezett a bácsirajʹ [Øj/i]
the boy.nom remembered.indef.3sg the old.man.sub
tanácsot adott nekii/j.
advice.acc gave. indef.3sg him.dat
‘The boyi remembered the old manj, [hej/i] had given himi/j advice.’
8. (16) is Pléh’s (1998: 178) example which was analysed by Németh T. and Bibok (2010: 510–
511). Here I add some new insights into the previous analysis providing details about the nature
of the general pragmatic knowledge relying on Searle’s (1969) speech act theory.
The utterances in (18a) and (18b) have the same grammatical structure. In (18a) the
zero subject anaphor [Øi] can be coreferential, therefore co-indexed with the subject
phrase a bácsii ‘the old.man.nom’ according to the grammatical requirement of the
subject continuation principle and underlying cognitive principle of the identity
of perspective. In spite of the structural coincidence between (18a) and (18b), in
(18b) the zero subject cannot be identified typically with the referent of the subject
phrase a fiú ‘the boy.nom’. The preferred reading is that the zero subject anaphor
[Øj] is coreferential with the sublative noun phrase a bácsiraj ‘the old.man.sub’.
Giving advice is a socially governed speech act. Searle (1969: 67) adds a comment
to the conditions for achieving success in giving advice that advice is not a form of
requesting. Advising is more likely to involve telling another person what is best
for him/her. Therefore, the preparatory conditions of the speech act of advising can
be extended with the condition that the adviser can only be a person with a con-
siderable advantage in knowledge, experience, age or social status; otherwise s/he
cannot tell what is best for another person. This piece of information is involved in
the general pragmatic knowledge concerning the speech act of giving advice, and
it can override the reading predicted by grammar in (18b). In (18b) the grammat-
ical principle of subject continuation predicts that the zero subject anaphor in the
second clause is coreferential with the subject phrase a fiú ‘the boy.nom’ in the first
clause, but the general pragmatic knowledge regarding advising requires that the
antecedent of the zero subject anaphor in the second clause is the sublative noun
phrase a bácsira ‘the old.man.sub’ in the first clause. Since in a typical situation, an
old man can be assumed to have more experience and knowledge than a boy has,
the old man fulfils the requirement placed by general pragmatic knowledge on the
successful performance of the speech act of advising. Therefore, general pragmatic
knowledge can override the interpretation predicted by grammar and select the
sublative (English equivalent: onto) noun phrase a bács-iraj ‘the old.man.sub’ as
an antecedent of the zero subject anaphor [Øj] in (18b).
However, a particular speech situation in which the boy is better informed than
the old man (for instance, in how to use the internet or a smart phone) overrides the
reading provided by general pragmatic knowledge and supports the grammatical
reading. It is worth noting that, similarly to the analysis of (18b), in (18a) particu-
lar contextual factors, namely, the particular speech situation according to which
the boy is better informed than the old man can also override the interpretation
predicted by grammar. In this latter situation the antecedent of the zero anaphor in
(18a) must be identified with the referent of the phrase a fiúra ‘the boy.sub’.
In Hungarian language use, zero subject anaphors can also occur in such a
way that their antecedents are not posited in the same utterance. In these cases the
immediate utterance context of zero subject anaphors must be extended with the
information from the preceding discourse to identify the referent of the implicit
argument, i.e. the antecedents of zero subject anaphors are found in the previous
utterances in discourse. Thus, while zero subject anaphors in utterance contexts
are used and interpreted in the second (B) manner of the occurrence of implicit
arguments, zero subject anaphors in discourses are licensed and identified in the
third (C) manner (cf. Section 1.1). Consider (19):
(19) A bohócdoktori az ölébe vette a kisfiútj
the clown.doctor.nom the his.lap.ill took.def.3sg the little.boy.acc
Énekelt [Øi/j] nekii,j.
sang.indef.3sg   him.dat
‘The clown doctori took the little boyj in his lap. [Hei,j] sang to himi,j.’
In (19) one can find a zero discourse anaphor whose antecedents are not posited
in the same utterance. Resolving this type of zero anaphor is also governed by the
interaction between grammatical and pragmatic constraints including background
encyclopaedic knowledge and pieces of information from the particular speech
situation. According to the grammatical requirements, in (19) the zero subject of
the second utterance [Øi] is coreferential with the subject of the first utterance,
i.e. with the noun phrase a bohócdoktori ‘the clown.doctor.nom’. The interpreta-
tion predicted by the grammatical rule of subject continuation is supported by
the encyclopaedic expectation that a clown doctor is a person who visits seriously
ill children in hospitals and attempts to make them forget about their illness for
a while and make them laugh. In order to fulfil this aim, a clown doctor sings to
children, plays with them etc.
However, let us imagine a particular situation where the clown doctor and the
little boy have known each other for some time. The little boy has a sore throat
and, therefore, he can hardly communicate. But he likes the clown doctor very
much and becomes very happy when he visits him. They have a favourite song
which the little boy tries to sing. In this situation, the zero subject of the second
utterance [Øj] can be coreferential with the direct object of the first utterance, i.e.
with the noun phrase a kisfiúti ‘the little.boy.acc’. Thus, the particular contextual
information does not support the reading predicted by the grammatical rule of
subject continuation.
Finally, it is worth mentioning the role of the subsequent discourse. In the
previous analyses, we have relied on the grammatical requirements, encyclopaedic
knowledge, general pragmatic knowledge and particular contextual factors. As to
these latter factors, the physical environment, the encyclopaedic context and the
previous utterance or discourse context were considered. Now let us see the mod-
ified version of (19) in (20) and (21).
(20) A bohócdoktori az ölébe vette a kisfiútj.

Énekelt [Øj] nekiiʹ megfeledkezve a fájdalomról.
sang.indef.3sg   him.dat forgetting the pain.del
‘The clown doctori took the little boyj in his lap. [Hej] sang to himi forgetting
about the pain.’
(21) A bohócdoktori az ölébe vette a kisfiútj.
Énekelt [Øi] nekijʹ majd álomba ringatta [Øi] [Øj]
sang.indef.3sg him.dat then dream.ill rocked.def.3sg
‘The clown doctori took the little boyj in his lap. [Hei,] sang to himj then rocked
[himj] into sleep.’
In (20) and (21) the subsequent part of the second utterance disambiguates the in-
terpretation of zero subject anaphors and in (21) also the identification of the zero
object anaphor. On the basis of this role the subsequent part of the utterance plays,
it is reasonable to add the discourse continuation to the factors which license the
use and interpretation of zero subject (and other kinds of) anaphors in discourses.
4. Conclusions
On the basis of the analysis in (3), (6)–(15), and (18)–(21), it can be seen, on the
one hand, that the use or interpretation of zero subject anaphors and extralinguis-
tically motivated subject pro-drop phenomena in Hungarian language use pre-
dicted by grammar can be considered only a typical, default use and interpretation
that emerge due to the lack of any pieces of information from the encyclopaedic
knowledge, general pragmatic knowledge and/or specific context, and, on the other
hand, that grammar (e.g. syntactic, morphosyntactic, and semantic constraints)
and pragmatics (general pragmatic knowledge, and particular contextual informa-
tion from the observable physical context, form the encyclopaedic knowledge, as
well as previous and subsequent parts of discourse) interact intensively (perhaps,
in multiple ways) in the licensing and recovering of zero subject anaphors and
extralinguistically motivated subject pro-drops phenomena, i.e. kinds of implicit
argument with its own syntactic position in the structure of utterances.
Acknowledgements
The research reported on in the present chapter was supported by the Hungarian Scientific Re-
search Fund OTKA NK 100804 Comprehensive grammar resources: Hungarian as well as by the
MTA-DE-SZTE Research Group for Theoretical Linguistics. This research was also supported
by the EU-funded Hungarian grant EFOP-3.6.1-16-2016-00014. I wish to thank Károly Bibok
and Marta Dynel for their remarks and suggestions which helped me to clarify some issues.
References
Ariel, Mira. 1991. “The Function of Accessibility in a Theory of Grammar.” Journal of Pragmatics
16: 443–463. doi: 10.1016/0378-2166(91)90136-L
Blackwell, Sarah. 2001. “Testing the Neo-Gricean Pragmatic Theory of Anaphora: The Influence of
Consistency Constraints on Interpretation of Coreference in Spanish.” Journal of Pragmatics
33: 901–941. doi: 10.1016/S0378-2166(01)80034-5
Capone, Alessandro. 2008. “Belief Reports and Pragmatic Intrusion (the case of null appositives).”
Journal of Pragmatics 40: 1019–1040. doi: 10.1016/j.pragma.2008.02.008
Chomsky, Noam. 1981. Lectures on Government and Binding. The Pisa Lectures. Dordrecht: Foris.
Cote, Sharon A. 1996. Grammatical and Discourse Properties of Null Arguments in English. PhD
dissertation, University of Pennsylvania.
Cummins, Sarah, and Yves Roberge. 2005. “A Modular Account of Null Objects in French.” Syntax
8: 44–64. doi: 10.1111/j.1467-9612.2005.00074.x
Dankovics, Natália. 2001. “Anaforikus viszonyok finn, észt és magyar összetett mondatokban
pszicholingvisztikai szempontból” [Anaphoric relations in Finnish, Estonian, and Hungarian
compound sentences from the psycholinguistic point of view]. Nyelvtudományi Közlemények
98: 120–142.
Dankovics, Natália. 2005. “A szórend szerepe az anafora-feldolgozásban” [The role of word order
in anaphor resolution]. In Az ezerarcú elme. Tanulmányok Pléh Csaba 60. születésnapjára
[The Thousand Face Mind. Studies on the Occasion of Csaba Pléh’s 60th Birthday], ed. by
Judit Gervain, Kristóf Kovács, and Mihály Racsmány, 117–134. Budapest: Akadémiai Kiadó.
Ehlich, Konrad. 1982. “Anaphora and Deixis: Same, Similar, or Different?” In Speech, Place and
Action. Studies in Deixis and Related Topics, ed. by Robert J. Jarvella and Wolfgang Klein,
315–338. Chichester: John Wiley.
É. Kiss, Katalin. 1985. “Az anaforikus névmások értelmezéséről.” [On the interpretation of ana-
phoric pronouns]. Általános Nyelvészeti Tanulmányok 16: 155–187.
É. Kiss, Katalin. 1998. “Mondattan” [The syntax]. In Új magyar nyelvtan [New Hungarian
Grammar], ed. by Katalin É. Kiss, Ferenc Kiefer, and Péter Siptár, 151–184. Budapest: Osiris
Kiadó.
García Velasco, Daniel, and Carmen Portero Muñoz. 2002. Understood Objects in Functional
Grammar. Amsterdam: University of Amsterdam Press.
Gillon, Brendan S. 2012. “Implicit Complements: A Dilemma for Model Theoretic Semantics.”
Linguistics and Philosophy 35: 313–359. doi: 10.1007/s10988-012-9120-2
Goldberg, Adele. 2005. “Argument Realization. The Role of Constructions, Lexical Semantics
and Discourse Factors.” In Construction Grammars: Cognitive Grounding and Theoretical
Extensions, ed. by Jan-Ola Östman and Mirjam Fried, 17–43. Amsterdam: John Benjamins.
doi: 10.1075/cal.3.03gol
Goldberg, Adele. 2013. “Arguments Structure Constructions versus Lexical Rules or Derivational
Verb Templates.” Mind and Language 28: 435–465. doi: 10.1111/mila.12026
Huang, Yan. 2000. “Discourse Anaphora: Four Theoretical Models.” Journal of Pragmatics 32:
151–176. doi: 10.1016/S0378-2166(99)00041-7
Huang, Yan. 2004. “Anaphora and the pragmatics-syntax interface.” In The Handbook of Pragmatics,
ed. by Laurence Robert Horn and Gregory Ward, 288–314. Oxford: Blackwell.
Huang, Yan. 2009. “Anaphora, Cataphora, Exophora, Logophoricity.” In Concise Encyclopaedia
of Semantics, ed. by Keith Allan, 19–31. Amsterdam: Elsevier.
Huang, Yan (ed.). 2012. The Oxford Dictionary of Pragmatics. Oxford: Oxford University Press.
Iten, Corinne, Marie-Odile Junker, Aryn Pyke, Robert Stainton, and Chaterine Wearing. 2005. “Null
Complements: Licensed by Syntax or by Semantics-Pragmatics?” CLA Annual Conference
Proceedings 2005: 1–15.
Kim, Kye-Sung, Park, Seong-Bae, Song, Hyu-Je, Park, Se Young, and Lee, Sang Jo. 2010.
“Identification of Non-referential Zero Pronouns for Korean-English Machine Translation.”
In Proceedings of PRICAI 2010: Trends in Artificial Intelligence. 11th Pacific Rim International
Conference on Artificial Intelligence, Daegu, Korea, August 30–September 2, 2010, 112–122.
Berlin: Springer.
Laczkó, Krisztina. 2003. “A mutató névmások funkcionális vizsgálata.” [The functional examina-
tion of demonstrative pronouns]. Magyar Nyelvőr 127: 314–325.
Lejtovicz, Katalin E., and Zsolt T. Kardkovács. 2007. “Anaphora Resolution.” Magyar Kutatók
8. Nemzetközi Szimpóziuma. 8th International Symposium of Hungarian Researchers on
Computational Intelligence and Informatics, 227–237. Budapest: BMF.
Lyons, John. 1977. Semantics. Cambridge: Cambridge University Press.
Mitkov, Ruslan. 2013. Anaphora Resolution. London: Routledge.
Mittwoch, Anita. 1983. “Backward Anaphora and Discourse Structure.” Journal of Pragmatics 7:
129–139. doi: 10.1016/0378-2166(83)90048-6
Németh T., Enikő. 2000. “Occurrence and Identification of Implicit Arguments in Hungarian.”
Journal of Pragmatics 32: 1657–1683. doi: 10.1016/S0378-2166(99)00114-9
Németh T., Enikő. 2001. “Implicit Arguments in Hungarian: Manners of Their Occurrence and
Possibilities of Their Identification.” In Argument Structure in Hungarian, ed. by István
Kenesei, 113–156. Budapest: Akadémiai Kiadó.
Németh T., Enikő. 2007. “Grammatika és pragmatika viszonya az implicit argumentumok tükré-
ben.” [The relationship between grammar and pragmatics in case of implicit arguments]. In
Nyelvelmélet − nyelvhasználat [Theory of Languge and Language Use], ed. by Tamás Gecső
and Csilla Sárdi, 188–197. Székesfehérvár: Kodolányi János Főiskola.
Németh T., Enikő. 2010. “How Lexical-Semantic Factors Influence the Verbs’ Occurrence with
Implicit Direct Object Arguments in Hungarian.” In The Role of Data at the Semantics–
Pragmatics Interface, ed. by Enikő T. Németh and Károly Bibok, 305–348. Berlin: Mouton
de Gruyter. doi: 10.1515/9783110240276
Németh T., Enikő. 2012. “Lexical-Semantic Properties and Contextual Factors in the Use of Verbs
of Work with Implicit Subject Arguments in Hungarian.” Intercultural Pragmatics 9: 453–477.
Németh T., Enikő. 2014a. “Hungarian Verbs of Natural Phenomena with Explicit and Implicit
Subject Arguments: Their Use and Occurrence in the Light of Data.” In The Evidential Basis
of Linguistic Argumentation, ed. by András Kertész and Csilla Rákosi, 103–132. Amsterdam:
John Benjamins. doi: 10.1075/slcs.153.05nem
Németh T., Enikő. 2014b. “Implicit Arguments at the Grammar-Pragmatics Interface: Some
Methodological Considerations.” Argumentum 10: 679–694.
Németh T., Enikő. 2017. “Theoretical and methodological issues in the research into implicit
arguments in Hungarian.” In Pragmatics at Its Interfaces, ed. by Stavros Assimakopoulos,
149-174. Berlin: Mouton de Gruyter.
Németh T., Enikő, and Károly Bibok. 2010. “Interaction between Grammar and Pragmatics: The
Case of Implicit Arguments, Implicit Predicates and Co-composition in Hungarian.” Journal
of Pragmatics 42: 501–524. doi: 10.1016/j.pragma.2009.07.001
Peral, Jesús, and Antonio Fernández. 2000. “Generation of Spanish Zero Pronouns into English.”
In Natural Language Processing − NLP 2000. Second International Conference. Patras, Greece,
June 2000. Proceedings, ed. by Dimitris N. Christodoulakis, 252–260. Berlin: Springer.
Pethő, Gergely, and Éva Kardos. 2009. “Cross-Linguistic Evidence and the Licensing of Implicit
Arguments.” Oslo Studies in Language 1: 33–61.
Pléh, Csaba. 1994. “Mondatközi viszonyok feldolgozása: az anaphora megértése a magyar-
ban.” [Processing intersentential relations: Interpreting anaphors in Hungarian]. Magyar
Pszichológiai Szemle 50: 287–320.
Pléh, Csaba. 1998. A mondatmegértés a magyar nyelvben [Sentence Comprehension in Hungarian].
Budapest: Osiris Kiadó.
Pléh, Csaba, and Katalin Radics. 1978. “Truncated Sentence, Pronominalization and the Text.”
Acta Linguistica Academiae Scientiarum Hungaricae 28: 91–113.
Pléh, Csaba, and Brian McWhinney. 1987. “Anaphora Resolution in Hungarian.” Acta Linguistica
Academiae Scientiarum Hungaricae 37: 103–124.
Radford, Andrew. 1997. Syntactic Theories and the Structure of English. A Minimalist Approach.
Cambridge: Cambridge University Press. doi: 10.1017/CBO9781139166706
Reinhart, Tanya. 1985. Anaphora and Semantic Interpretation. London: Croom Helm.
Renkema, Jan. 2004. Introduction to Discourse Studies. Amsterdam: Benjamins. doi: 10.1075/z.124
Sanders, José, and Wilbert Spooren. 1997. “Perspective, Subjectivity, and Modality from a
Cognitive Linguistic Point of View.” In Discourse and Perspective in Cognitive Linguistics,
ed. by Wolf-Andreas Liebert, Gisela Redeker, and Linda Waugh, 85–112. Amsterdam: John
Benjamins. doi: 10.1075/cilt.151.08san
Scott, Kate. 2006. “When Less Is More: Implicit Arguments and Relevance Theory.” UCL Working
Papers in Linguistics 18: 139–170.
Searle, John R. 1969. Speech Acts. An Essay in the Philosophy of Language. Cambridge: Cambridge
Sperber, Dan and Wilson, Deirdre. 1986/1995. Relevance: Communication and Cognition. Oxford:
Blackwell.
Tolcsvai Nagy, Gábor. 2001. “3rd Person Anaphora in Hungarian.” Hungarian Studies 15: 287–298.
Part II
Sentence and utterance

Chapter 6
Implicitness via overt untruthfulness

Grice on Quality-based figures of speech
Marta Dynel
Grice’s (1989a [1975], 1989b [1978]) conversational implicature is a salient mani-

festation of implicitness, whilst the four figures of speech contingent on flouting
the first maxim of Quality constitute an important group of phenomena that
promote conversational implicature. Implicitness is then central to the figures of:
metaphor, irony, hyperbole and meiosis. This chapter gives an exegesis of Grice’s
view of these Quality-based figures of speech, which enjoy a special status in
his framework of conversational logic. The critical analysis is performed against
the backdrop of the pertinent neo-Gricean and post-Gricean scholarship. This
chapter thus sheds new light on selected vexing issues in Grice’s philosophy, with
some being (unjustly) taken for granted or oversimplified, and other ones being
the subject of ongoing debates.
Keywords: implicature, saying, Quality, truthfulness, overt untruthfulness,

flouting, Grice, irony, metaphor, meiosis, hyperbole
1. Introduction
In spite of the passage of time, Grice’s lectures compiled in the 1989 volume never
cease to generate scholarly interest, serving a departure point for philosophical
and theoretical pragmatic debates on a wide spectrum of issues, such as inten-
tionality and speaker meaning, the Cooperative Principle and maxims, and the
levels of speaker meaning. Many of such debates concern the way Grice’s postu-
lates should be interpreted. This is hardly surprising as Grice’s discussions are not
unequivocal in many respects. On the other hand, despite (or perhaps because of)
its prominence and prevalence in the pragmatic scholarship, in academic research
and handbooks (representing not only pragmatics but also other linguistic disci-
plines), the Gricean philosophy of communication tends to be oversimplified and
doi 10.1075/pbns.276.06dyn
122 Marta Dynel
misinterpreted in many respects. 1 Some of these oversimplifications and misinter-

pretations concern the issue of implicating and implicature, notably the problem-
atic type related to four Quality-based figures of speech/rhetorical figures: metaphor,
irony, hyperbole and meiosis.
This chapter aims to contribute to the pragma-philosophical neo-Gricean stud-
ies of implicitness by addressing a selection of issues critical to Grice’s views of
Quality-based rhetorical figures promoting conversational implicatures. Section 2
revisits the sources of implicature. The focus of Section 3 is on the category of
Quality and its relationship with truthfulness. In Subsection 3.1, some attention is
paid to the status of saying in this context. Section 4 explores Grice’s remarks on
the four figures, whilst Section 5 addresses the criticism that Grice’s discussion of
these figures of speech has received.
2. Maxim nonfulfilment as the source of implicature
Grice’s (1989a [1975], 1989b [1978], 1989c) model of conversational or, general-
ly, communicative logic (which holds for all forms of communication, e.g. a let-
ter of recommendation or an informal conversation) is famously anchored in the
Cooperative Principle, which is to be understood as the principle of rationality
(see Davies 2000, 2007; Dynel 2008, 2013a). Together with the subordinate max-
ims, which fall into four categories borrowed from Kant, it serves as the basis for
intentional and rational communicative exchanges, and thus the communication of
(inherently intentional) speaker meaning (see Dynel 2011a and references therein).
In this framework, “one may distinguish, within the total signification, between
what is said (in a favored sense) and what is implicated” (Grice 1989a [1975]: 41).
In Grice’s original terminology, the label implicature is synonymous with the act
of implicating leading to implicatum, which stands for the meaning the speaker
implies (Grice 1989a [1975]: 24). Whilst the latter term is practically non-existent
in the neo-Gricean scholarship, “implicature” tends to be used in reference to the
meaning the hearer infers, or at least the meaning that emerges from an utterance
in accordance with the speaker’s intention (see Dynel 2011b).
The maxims, which Grice (1989c: 370) refers to as “conversational imperatives”
(possibly echoing Kantian philosophy, cf. Vincent Marrelli 2004), serve as “mor-
al commandments” and their “observance promotes and their violation dispro-
motes conversational rationality”. Therefore, following the maxims is something
that speakers ought to do, and may be expected to do, for the sake of rational
1. Some of the notorious claims reverberating across the interdisciplinary literature are that
there are four maxims, or that the Cooperative Principle is violated when an implicature arises.
Chapter 6. Implicitness via overt untruthfulness 123
and harmonious communication. However, speakers’ overt departures from the

maxims defy hearers’ expectations, triggering their inferential work to arrive at an
explanation, in the form of implicit meaning, for why the speakers have chosen
not to fulfil the maxims.
Generally, Grice (1989a [1975]: 30) lists four ways of “failing to fulfil” (see also
Vincent Marrelli 2004) a maxim: flouting, unostentatious violation, as well as clash
and opting out. The most important forms of maxim nonfulfilment (Mooney 2004),
the umbrella term preferred here, are unostentatious violations and floutings, dif-
ferentiated by their covertness and overtness respectively. This is clear when Grice
(1989a [1975]: 30) first makes the distinction. The speaker may “quietly and unos-
tentatiously violate a maxim; if so, in some cases he will be liable to mislead” 2 (Grice
1989a [1975]: 30, italics in the original). Covert nonfulfilment, which is intention-
ally made unavailable to the hearer, falls outside Grice’s model of communicative
rationality (see Vincent Marrelli 2004; Meibauer 2005; Dynel 2011b). On the other
hand, the speaker “may flout a maxim; that is, he may blatantly fail to fulfill it. […]
This situation is one that characteristically gives rise to a conversational implicature;
and when a conversational implicature is generated in this way, I shall say that a
maxim is being exploited (1989a [1975]: 30, italics in the original). However, in the
course of the seminal lecture, the distinction between covert nonfulfilment and
overt nonfulfilment resulting in implicatures becomes blurry.
Firstly, perhaps by an oversight, Grice is inconsistent in his use of the term
“violation”, introducing it as a technical term covering covert maxim nonfulfilment
(cf. the italics in the relevant quotation above) but using it also as a superordinate
label in reference to all forms of nonfulfilment, also overt/ostentatious nonful-
filment which promote implicatures, as his examples bear out (cf. Grice 1989a
[1975]: 31, 32, 33). This shows also in the headings depicting the sources of im-
plicature (Grice 1989a [1975]: 32–33, see the paragraph below), where “violation”
does not appear to stand for covert maxim nonfulfilment. Additionally, apart from
presenting “flouting” as synonymous with “exploiting” a maxim (1989a [1975]: 30)
or infringing a maxim (1989a [1975]: 33), Grice calls floutings “apparent violations”
(1989a [1975]: 31, 35), as well as “overt violations” (1989a [1975]: 36), as evidenced
by the presence of these terms in the section devoted to floutings. Also, Grice
seems to suggest that a flouting is “a justifiable violation” or “only a seeming, not
a real, violation,” whereby “the spirit, though perhaps not the letter, of the maxim
is respected” (Grice 1989c: 370). Irrespective of such ambivalent parlance, Grice’s
underlying distinction between (overt) flouting vs. (covert) violation is reflected
in the examples he provides.
2. This intentional act is typically dubbed “deception” not “misleading” (see e.g. Dynel 2011b,
Mahon 2015).
124 Marta Dynel
Secondly, next to floutings, Grice (1989a [1975]: 31) lists two other sources of
conversational implicature (duly classified as particularised conversational implica-
tures). 3 The three groups are: “Examples in which no maxim is violated, or at least
in which it is not clear that any maxim is violated”, “Examples in which a maxim
is violated, but its violation is to be explained by the supposition of a clash with
another maxim” (Grice 1989a [1975]: 32) and “Examples that involve exploitation,
that is, a procedure by which a maxim is flouted for the purpose of getting in a con-
versational implicature by means of something of the nature of a figure of speech”
(Grice 1989a [1975]: 33). The third group is the most amply discussed by Grice and
revisited in the neo- and post-Gricean scholarship. However, there is some doubt
concerning the first two groups, scantly discussed by Grice and hardly commented
on in the literature.
Under the first category (no maxim violation), Grice presents two two-turn
conversational exchanges in each of which the reply, at a glance, is not immediately
relevant to the preceding turn, but Grice (1989a [1975]: 32) suggests that the “Be
relevant” maxim is not infringed and the connection between the two turns is “ob-
vious”. “In both examples, the speaker implicates that which he must be assumed
to believe in order to preserve the assumption that he is observing the maxim of
Relation” (Grice 1989a [1975]: 32). On an alternative reading, both these examples
epitomise flouting the Relation maxim. The speakers produce overtly irrelevant
replies, and it is the recognition of this fact that guides the hearers towards seek-
ing the pivotal implicature, where this maxim is observed (see Dynel 2013b). The
second category, the violation explained by a clash, seems to be a very narrow one,
illustrated with an example of the speaker implicating his/her lack of knowledge by
“infringing” the first maxim of Quantity in order not to “infringe” the second max-
im of Quality (Grice 1989a [1975]: 32). The first maxim of Quantity is infringed,
i.e. overtly flouted, so that the second maxim of Quality should remain intact, so
that it should not be (covertly) violated. Therefore, the examples in this section
may be deemed a subtype of the third category, that is floutings. Incidentally, this
explanation reveals Grice’s tendency to use some words interchangeably as tech-
nical terms or non-technical labels. Grice uses “infringement” also with regard to
flouting. The thrust of all this is that the two sources of implicatures may be reduced
to maxim floutings.
3. Having presented the three groups of examples of implicature, Grice states “I have so far
considered only cases of what I might call “particularized conversational implicature” – that is
to say, cases in which an implicature is carried by saying that p on a particular occasion in virtue
of special features of the context, cases in which there is no room for the idea that an implicature
of this sort is normally carried by saying that p.” (1989a [1975]: 37).
In the third, most robust, group of examples, Grice (1989a [1975]: 33–37) fo-
cuses explicitly on floutings/exploitations of all maxims falling into the famous
four categories. Grice also dubs flouting “infringement” (1989a [1975]: 33) or “real
violation” (1989a [1975]: 35) in two of the sub-headers concerning particular max-
ims. Interestingly, in the retrospective epilogue, Grice (1989c: 370) provides two
sources of implicatures: “a violation on his [the speaker’s] part of a conversational
maxim is in the circumstances justifiable, at least in his eyes” or “what appears to be
a violation by him of a conversational maxim is only a seeming, not a real, violation;
the spirit, though perhaps not the letter, of the maxim is respected.” It is difficult to
tell what difference Grice saw between these two types of overt violation, but this
may have to do with the peculiar case of flouting the Relation maxim, as opposed
to unquestionable flouting, which the case of the Quality-based tropes, for instance.
Here, the focus will be only on the first 4 maxim of Quality as the source of
implicature. Before the figures of speech can be examined, the status of the Quality
supermaxim and two maxims needs to be established.
3. Quality and truthfulness
The Quality supermaxim states: “Try to make your contribution one that is true”
(Grice 1989a [1975]: 27). Therefore, the supermaxim necessitates trying to contrib-
ute meanings that are objectively “true”. This may, at first blush, suggest that Grice
is concerned with “objective” truth, rather than what the speaker believes to be the
truth, i.e. “truthfulness”. 5 However, the first maxim of Quality reads: “Do not say
what you believe to be false” (Grice 1989a [1975]: 27), being clearly centred on the
speaker’s expression of true beliefs, and hence truthfulness. The second maxim
of Quality, “Do not say that for which you lack adequate evidence” (Grice 1989a
[1975]: 27), which Grice (1989a [1975]: 30) paraphrases as “Have adequate evidence
for what you say”, brings to focus the speaker’s perspective, but it is motivated by
external evidence which the speaker believes to be true, and (ideally) what is the
truth. Taken together, the Quality supermaxim and maxims do not necessitate “tell-
ing the truth”. The second maxim recommends that the speaker base his/her utter-
ances on the available evidence, the first maxim is orientated towards the speaker’s
avoiding the expression of false beliefs, whilst the supermaxim indicates that one
should only try to make true contributions. By this, Grice seems to suggest that the
4. As Grice (1989a [1975]: 34) observes, examples of flouting the second maxim of Quality are
only intermittent, so the maxim has limited capacity to promote implicatures.
5. Truth vs. truthfulness is the classical distinction first proposed by Aquinas and Augustine
(see Vincent Marrelli 2004).
126 Marta Dynel
speaker may be wrong while having good intentions to communicate only what he/
she believes to be true. Also, the available evidence may not conform to the truth
(e.g. having been contrived by someone), of which the speaker is not aware. Thus,
in non-prototypical situations, the speaker may unwittingly say something false,
whilst still observing the supermaxim and the maxims.
Wilson (1995) and Wilson and Sperber (2000 [2002, 2012]) regard the two
Quality maxims as being related to what is said, whilst only the supermaxim to the
totality of speaker meaning, i.e. both what is said and what is implicated. They base
this interpretation on Grice’s parlance: “contribution” in the supermaxim and “say”
in the two maxims. Vincent Marrelli (2004) does not state that this interpretation
is misguided, but she does indicate that there is a terminological confusion about
this issue. This confusion stems partly from Grice’s use of the word “say” in each
of the maxims, which may be interpreted to mean “utter” or “verbalise” or, in the
technical sense, “convey speaker meaning” (see Section 5). Incidentally, like the
supermaxim of Quality, the two maxims of Quantity use the word “contribution”
(Grice 1989a [1975]: 26). It would be wrong to conclude that the maxims of Quality
and the maxims of Quantity enjoy markedly different statuses and pertain to dif-
ferent scopes of meaning, with only the latter maxims being all-encompassing like
the supermaxim of Quality.
Irrespective of Grice’s verbal formulation of the maxims of Quality, these two
maxims are similar to all the other ones in whole set (but see Dynel 2016a on the
special status of the first maxim of Quality) in that they pertain primarily to how the
speaker’s utterance is formulated with a view to communicating speaker meaning.
Therefore, the maxims ultimately apply to speaker meaning in its entirety, whether
what is said or implicature, which arises when a maxim is flouted. Most important-
ly, introducing (all) maxims’ floutings/exploitations, Grice specifies that “though
some maxim is violated [i.e. overtly violated/flouted] at the level of what is said,
the hearer is entitled to assume that that maxim, or at least the overall Cooperative
Principle, is observed at the level of what is implicated” (1989a [1975]: 33). Also,
Grice (1989d [1968]: 86) states: “what is implicated is what it is required that one
assume a speaker to think in order to preserve the assumption that he is observing
the Cooperative Principle (and perhaps some conversational maxims as well), if not
at the level of what is said, at least at the level of what is implicated.” These quota-
tions indicate that all maxims concern both levels of speaker meaning (even though,
as will be shown below, both levels of meaning are not mandatory, see Section 5).
Moreover, as a matter of an overarching moral rule, the first maxim pertains also
to the totality of speaker meaning communicated under the Cooperative Principle.
This is because Grice (1989c) does not allow for mendacity in his model of harmo-
nious communication, whose cornerstone is the speaker’s truthfulness (see Vincent
Marrelli 2004; Meibauer 2005; Dynel 2011b).
Overall, the category of Quality concerns truthfulness in speaker meaning

taken as a whole, that is the speaker’s expression of his/her genuine beliefs and
the speaker’s avoidance of expression of what he/she believes to be false validated
in the light of the available evidence. Importantly, Grice’s philosophy resides in
the speaker’s intention and rationality, as well as speaker meaning (see e.g. Dynel
2011a), whilst (objective) truth is not discussed. The supermaxim is a generalisa-
tion, whilst the two subordinate imperatives help put it into practice. The second
maxim supports the first one, suggesting how the speaker should present and devel-
op the beliefs he/she communicates to others in order to avoid accidental untruths.
It is, nonetheless, the first maxim of Quality that seems to be the pivotal one in
the model. The first maxim of Quality is the departure point for the discussion of
significant communicative phenomena: four stylistic figures, as well as lying (see
Dynel 2011b, 2016a, Meibauer 2014), being (overtly) flouted or (covertly) violated
respectively (cf. Bhaya Nair 1985). 6
The first maxim of Quality has been dubbed the maxim of truthfulness (Wilson
1995; Wilson and Sperber 2000 [2002, 2012]; 7 Livnat 2011; cf. Gibbs 1994; Livnat
2004). By contrast, referring to Akmajian et al. (2001), Vincent Marrelli (2003,
2004: 80) regards Grice’s Quality supermaxim as coinciding with “truthfulness”,
which she interprets as the speaker’ attempting to say what is true, whilst she per-
ceives the first maxim of Quality as involving “sincerity,” which captures speakers’
believing what they are saying. This tallies with a perspective that truthfulness
appertains to the speaker’s communicative intention and concern to communicate
what he/she believes to be the truth, whereas sincerity represents the compatibility
between the speaker’s utterance and his/her beliefs (cf. Ozar 2008). However, on
an alternative view, believing to be true what one is communicating is the essence
of truthfulness. Truthfulness cannot be equated with truth, as typically endorsed
in the studies on lying and deception, taken as a whole (e.g. Vincent Marrelli 2003,
2004; Dynel 2011b, 2016a; Mahon 2015). Although, as shown above, the second
maxim and the supermaxim are also embedded in the notion of truthfulness 8 (cf.
6. However, she also lists other phenomena, notably euphemisms, which are not tropes per
se and need not involve implicatures, as well as paradoxes, which do not seem to be based on
flouting the first maxim of Quality.
7. Wilson and Sperber (2000: 215, 2012: 127), reduce it to the maxim of “literal truthfulness”,
i.e. truthfulness that concerns only the level of what is said.
8. If “truthfulness” is understood very broadly, for instance as in the “oath of truthfulness” (see
Vincent Marrelli 2004), it will depend also on the other maxims. This is related to the fact that
deception may arise from the violation of maxims other than the first maxim of Quality (see
Dynel 2011b).
128 Marta Dynel
Wilson and Sperber’s (2000 [2002, 2012]: 215, 216) “maxims of truthfulness”, see

also Gibbs 1994), as opposed to objective truth, it is the first maxim of Quality that
should perhaps be dubbed the maxim of truthfulness. This is because it is the most
crucial maxim (responsible for the generation of rhetorical figures) and because it
corresponds directly to the prevalent definition of truthfulness as communicating
“what one believes to be true”, even though it is formulated by negation, and obliges
the speaker not to communicate what he/she believes to be false. The first maxim of
Quality can be translated into “Say what you believe to be true”, which is compatible
with Grice’s (1989a [1975]: 30) paraphrase of the second maxim of Quality. Grice’s
truthfulness does not concern only assertions, as will be argued below.
3.1 Truthfulness and saying
A problem that looms large in the context of the category of Quality and (lack of)
truthfulness is the type of utterance to which this label can pertain, and thus the
definition of what is said, as well as implicatures piggybacked on it.
In Grice’s view, meanings emerge not merely from what words convention-
ally mean, because “what words mean is a matter of what people mean by them”
(1989c: 340). However, conventional meanings do feed into the speaker’s mean-
ings, both what is said, and (in some cases) what is implicated. As Grice explains,
“In the sense in which I am using the word say, I intend what someone has said to
be closely related to the conventional meaning of the words (the sentence) he has
uttered” (Grice 1989a [1975]: 25, see also 1989b [1978]: 49). Grice uses “say” in
his “favored sense” (1989a [1975]: 25, 33; 1989b [1978]: 41), namely with reference
to what the speaker intends to communicate, next to what is implicated (Grice
1989a [1975]: 25, 27, 33). What is said is based on the speaker’s specific intention
in a given context, for example in terms of what particular referents his/her words
have or which of the alternative conventional meanings of a phrase the speaker
intends to communicate “on the particular occasion of utterance” (Grice 1989a
[1975]: 25). Importantly, Grice stresses that the speaker’s truthfulness (i.e. belief
what he/she says is true) while saying is a matter of an underlying commitment.
The speaker, who “has said that p” has “committed himself, in a certain way, to its
being the case that he believes that p, and while this commitment is not a case of
saying that he believes that p, it is bound up, in a special way, with saying that p”
(Grice 1989b [1978]: 42).
An assertion is traditionally understood as an act of presenting a statement/
proposition that one is producing as true (Peirce 1934: 384; see also Brandom 1983;
Jary 2010; Pagin 2015; Meibauer 2014).
Very frequently, saying is taken for granted as being synonymous with asserting,
and a few authors make this claim explicitly (e.g. Wilson and Sperber 2000 [2002,
2012]; Soames 2008; Stokke 2013; Saul 2012; Benton 2014). Wilson and Sperber
(2000 [2002, 2012]: 221) propose that on the technical/stronger interpretation (not
“expressing”), Grice’s “saying is not merely expressing a proposition but asserting
it, i.e. committing oneself to its truth.” Similarly, in the light of the fact that Grice
(1989a [1975]) uses “say” only for Quality maxims, Benton (2014) concludes that
Grice must mean “assert”. As Benton (2014: fn. 4) reports:
in unpublished notes (Grice Papers, 1947–1989), Grice consistently clarifies ‘says’
and ‘said’ as denoting assertive utterances. For example, in handwritten notes from
1966–75 (carton 1, folder 23) entitled ‘Saying’: Week I, he distinguishes “between
that which is actually said (‘asserted’) and that which is implied or otherwise
conveyed or got across” (p. 1). Moreover, in earlier notes from that file, Grice
contrasts the terms imply, suggest, convey, indicate, get across with say, state,
assert: the latter are, he says, “not right” for the implicature idea he is trying to
isolate (pp. 6–7).
Firstly, as Benton (2014) reports, Grice’s key goal in using “say” to mean “assert”
in these notes is to juxtapose it with meaning that is implicated. Hence, the “syno-
nyms” of “say” (i.e. “assert” or “state) should be taken as explanatory terms, rather
than as synonyms per se. Secondly, although the term “assert” occurs intermittently
in the earlier versions of some of the papers (see Baptista 2014), the formulations
equating “saying” with “asserting” were not included in the lectures published in
the seminal 1989 collection. This suggests that Grice must have realised that “as-
serting” had too narrow a meaning and showed various limitations, which is why
he purposefully eradicated this term from his volume.
Consequently, Grice does not address the notion of assertion in his collected
lectures, and his account of saying that emerges in the course of his papers does
not centre on this idea. Nowhere does he state that “saying” is only “asserting”.
What Grice (1989b [1978]: 51) does, though, is differentiate between “saying” and
“asserting” when he talks about a speaker who aims to “assert (or otherwise say)”.
Hence, he suggests that asserting is just one type of saying. Even if most of the
examples that Grice discusses in the context of maxim observance and flouting,
which is conducive to conversational implicature, are indeed assertions (with the
notable exception of the floutings of the first maxim of Quality), the Cooperative
Principle and the maxims are claimed to hold for entire conversations, which
must include utterances of different types, notably questions and imperatives.
Communication cannot possibly be comprised only of assertions. Some of the
examples that Grice (1989a [1975], 1989b [1978]) provides to elucidate his claims
rely on conversational exchanges comprising questions and answers, whilst it is
130 Marta Dynel
only the latter that are analysed. In the light of the two seminal lectures on logic
and conversation, it is evident that Grice aims to give a holistic picture of com-
munication based on communicative rationality and a set of maxims that facili-
tate conversation. Importantly, as Baptista (2014: 7) emphasises, saying cannot be
equated with asserting also because Grice (1989d [1968]) “extends his account
to imperatives and leaves the door open to a conception applicable to different
‘moods’; though what he calls ‘mood’ is much closer to (indicated) force.” The
thrust of all this is that there is no reason to believe that Grice’s “what is said”
cannot stem from utterances other than assertions.
Needless to say, implicatures are never assertions since these meanings are
always implicated, rather than being communicated literally. Importantly, they
need not take the form of statements, either. As Grice (1989b: 370) underscores,
implicature may be factual or imperatival: “an implicatum (factual or imperatival) is
the content of that psychological state or attitude which needs to be attributed to a
speaker”. This “factuality” covers statements about not only facts but also subjective
psychological states (e.g. emotions). Essentially, implicatures may constitute state-
ments corresponding to the speaker’s belief system (cf. the first maxim of Quality)
concerning verifiable facts or, much more elusive, mental states. On the other hand,
imperatival implicatures capture the speaker’s requests for action, which are in tune
with the speaker’s psychological state (i.e. the speaker genuinely intends the actions
to be performed). It should be observed that implicated requests can be issued via
different utterance forms (statements, imperatives and questions). This lends fur-
ther evidence to the claim that Grice’s saying cannot be reduced to asserting (and
making as if to say to making as if to assert, see Section 4).
In conclusion, the category of Quality is pertinent not only to assertions but
also to other utterance types captured by “saying”, notably non-assertive declar-
atives, questions, imperatives, or exclamations, which, in Grice’s view, should be
produced truthfully (or sincerely, in Speech Act Theory), with the speaker being
committed to them. As Carston (2002: 210) concludes “Gricean ‘saying’ is a generic
term for the three central speech acts of stating that p, asking whether p, and en-
joining someone to make it the case that p, and it does entail speaker commitment
(‘speaker meaning’, in his terms).” The content of the speaker’s belief, as expressed
by the first maxim of Quality, need not involve the propositional content of an as-
sertion (believe that the statement is true), but the communicative force of a given
utterance, which is representative of the speaker’s beliefs and other mental states.
Thus, the “what” in the first maxim of Quality that the speaker should believe to
be true represents any utterance type and its import or illocutionary force. In other
words, the speaker’s belief may concern the illocutionary force of any utterance,
not necessarily its propositional content, which may simply be absent. This view of
truthfulness is in tune with Habermas’s (1984). Revising his validity claims, he states
that any utterance is amenable to truthfulness evaluation, regardless of whether it
involves a cognitive act (an assertion), a regulative act (a request or promise), or
an expressive act.
In Grice’s view, it seems, the speaker should then believe (the import of) a
question or imperative he/she is uttering to be true. In other words, he/she should
pose the question or imperative truthfully by committing himself/herself to it. Even
an exclamation “Ouch” may be judged on its truthfulness, based on whether the
speaker believes himself/herself to be in pain or otherwise suffering. On the other
hand, the truthfulness of a question consists in that fact that the speaker tacitly
expresses a belief that he/she does not have the requisite knowledge and thereby
sincerely wants to get a reply from the hearer. Whilst the prevailing classical view
is that only propositions can have truth value, Grice seems to be concerned with
a different matter and to give a broader picture of truthfulness in communication
(both as what is said and as what is implicated, see Section 5) when he proposes
the category of Quality. These observations will have a bearing on the discussion
of Grice’s view of irony and the other Quality-based figures of speech.
4. Grice’s view of metaphor, irony, hyperbole and meiosis

and criticism thereof
Addressing the case of flouting the first maxim of Quality, Grice (1989a [1975]: 34)
briefly discusses irony, metaphor, meiosis 9 and hyperbole. These four rhetorical fig-
ures, known in classical rhetoric as tropes, are claimed to flout the first Quality
maxim: “Do not say what you believe to be false” (1989a [1975]: 27). The figures
involve the speaker’s expression of something he/she believes to be false, which
he/she means to be overt to the hearer. Metonymy is another salient figure that is
rooted in flouting the first maxim of truthfulness (cf. Gibbs 1994), but Grice disre-
gards it. 10 Although Grice explicates this with respect to irony and metaphor only,
he appears to consider all the four figures to be hinged on making as if to say (Grice
1989a [1975]: 30, 31, 34, 1989b [1978]: 40, 53). “Making as if to say” indicates that
9. Grice differentiates between meiosis and hyperbole. On an alternative traditional view, mei-
osis, i.e. exaggerated attenuation, is regarded as a type of hyperbole, next to auxesis (e.g. Smith
1657).
10. Another phenomenon that fits this class of implicatures is bald-faced lying (see Dynel 2011b,
2015; Meibauer 2015).
132 Marta Dynel
there is no speaker meaning coinciding with what is said. Implicature is the only
type of speaker meaning that arises from the four figures. 11
Grice (1989a [1975]) introduces meiosis and hyperbole merely by presenting
(but not discussing) two examples: “Every nice girl loves a sailor”, and “Of a man
known to have broken up all the furniture, one says He was a little intoxicated.”
(1989a [1975]: 34). Even though Grice does not acknowledge this, this example dis-
plays not only meiosis but also irony. Thus, Grice unwittingly accounts for meiotic
irony (see Dynel 2016b). It is not “a little” but “a lot” that the person in question was
intoxicated. The reversal of meaning central to this example makes the untruthful-
ness underlying meiosis more transparent. It may be extrapolated from Grice’s very
brief presentation of meiosis and hyperbole that, thanks to the flouting of the first
maxim of Quality, no what is said is present in them, and implicatures arise from
them as the sole level of speaker meaning.
Nevertheless, many authors endorse a different opinion about these figures. For
instance, Fogelin (1988: 13–14) argues that whilst irony contradicts/negates real-
ity, “hyperbole” and “meiosis”/“understatement” show degrees, only downscaling
or upscaling reality, and necessitating “strengthening” or “weakening correction”.
Moreover, some researchers advocate this approach, but resort to Grice’s termi-
nology as if completely disregarding Grice’s proposal of how the chosen figures
should be addressed (but see Bhaya Nair 1985; Vincent Marrelli 2004). For example,
Haverkate (1990: 102–103) concedes that “litotes” [a type of meiosis] and “hyper-
bole” depict the world “in terms of disproportionate dimensions,” implying that
they do not involve “empirical falsehood”. Haverkate (1990: 103) then states that
“the use of litotes and hyperbole implies a violation of the following [i.e. the first]
maxim of quantity” as they withhold “a certain amount of information” or convey
“an excessive amount of information” respectively. Similarly, Norrick (2004: 1737)
states that whilst “extreme case formulations” (Pomerantz 1986) flout the relevant
Quality maxim, “non-extreme hyperboles” “violate only the quantity maxim or
are heard as approximations to the speaker’s beliefs” (Norrick 2004: 1737). Colston
(2000) sees understatement as flouting “the maxim of Quantity”. Similarly, Brown
and Levinson (1987: 217, 219) consider understatements and overstatements to
be cases of “violating the Quantity Maxim”. For his part, Gibbs (1994: 392) avers,
without providing any explanation for this, that hyperbole “violates the maxim of
Quality”, whereas understatement “violates the maxim of Quantity”. Interestingly,
Livnat (2011) claims that in understatements, the flouting of the (first) maxim of
11. This view does not encompass conventionalised figures with lexicalised meanings, notably
metaphors, that cannot be seen as inviting implicatures. Creative metaphors cannot be con-
sidered, as many relevance theoreticians have argued, part of what is said (see Camp 2006 for
excellent criticism).
Quantity gives rise to, or uncovers, the flouting of the (first) maxim of Quality. As
these claims indicate, meiosis and hyperbole tend to be conceptualised as partly
contingent on the truth, and thus on flouting Quantity maxims. In both meiosis and
hyperbole, “the situation is described in terms that fall between the opposite and
the reality of the situation” (Colston and O’Brien 2000: 1563). What is noteworthy
is that hyperbole and meiosis show degrees, involving smaller or greater departures
from “reality”, which should be properly seen as what the speaker believes to be the
truth. As Fraser notes, “We talk of ‘slightly exaggerated’ and ‘greatly exaggerated’
but never the somewhat synecdochic use of language” (1983: 34). This must be the
reason why the authors refer to the category of Quantity (frequently, not specifying
which maxim they have in mind, though).
However, the first maxim of Quantity, viz. “Make your contribution as in-
formative as is required (for the current purposes of the exchange)” (Grice
1989a [1975]: 26), paraphrased as “Be as informative as is required” (Grice 1989a
[1975]: 30), seems to operate differently, not residing in the meanings of mitigat-
ing words. Flouting the first maxim of Quantity manifests itself in some meaning
being entirely absent from an utterance, a meaning that has to be inferred in the
form of an implicature. 12 Grice (1989a [1975]) illustrates flouting the first maxim
of Quantity with his canonical example of a recommendation letter in which some
information is not provided at all, but which the author wishes to impart. The letter’s
author is only implicating the relevant negative information about a candidate (his
inaptness for the post), while he is explicating other information. Similar workings
of the first Quantity maxim can be observed in the case of a maxim clash, where the
speaker chooses to be less informative than required. In a reply to a question where
a person (C) lives, the speaker says “Somewhere in the South of France,” where-
by the speaker “implicates that he does not know in which town C lives” (Grice
1989a [1975]: 32–33). 13 Grice (1989a [1975]: 33) also adds that tautologies (such
as “Women are women”) are extreme cases of flouting the first maxim of Quantity.
In these cases, contextually relevant meanings must be inferred from what is not
stated, but communicated “in-between the words”.
By the same token, the second maxim of Quantity, “Do not make your contri-
bution more informative than is required” (Grice 1989a [1975]: 27), does not advise
speakers against exaggeration, which is the essence of hyperbole. Grice (1989a
12. Grice’s original view of this issue does not encompass the so-called scalar implicatures ad-
dressed in post- and neo-Gricean scholarship.
13. Grice (1989a [1975]: 32) provides this example under the somewhat ambivalent header
“Examples in which a maxim is violated, but its violation is to be explained by the supposition of
a clash with another maxim.” This example seems to correspond to overt violation, i.e. flouting,
of the first maxim of Quantity.
134 Marta Dynel
[1975]: 26–27) himself is sceptical about the need for this maxim, the reason for
which is twofold. Firstly, saying too much does not constitute a major transgression;
it does not often fly in the face of the Cooperative Principle. Secondly, the effect of
this maxim is secured by the maxim of Relation, which guarantees the relevance
of the information provided. However, he hypothesises that “overinformativeness
may be confusing in that it is liable to raise side issues; and there may also be an
indirect effect, in that the hearers may be misled as a result of thinking that there is
some particular point in the provision of the excess of information” (Grice 1989a
[1975]: 26–27). Grice tentatively presents the case of this maxim being flouted if the
speaker provides excessive information to serve as evidence for some claim, as an
“oblique way of conveying that it is to some degree controversial whether or not”
something is the case (Grice 1989a [1975]: 34). 14
By contrast, in Gricean terms, meiosis and hyperbole rely not on conveying in-
sufficient or surplus information, but on “false or inaccurate information” (Nemesi
2010: 401). Also Meibauer (2014: 170) supports the original Gricean account, stat-
ing that the first “maxim of Quality is still the best candidate for a maxim playing
a role in the derivation of conversational implicatures related to over- [hyperbole]
and understatement [meiosis]”. In meiosis or hyperbole, whatever their forms may
be, the speaker does not increase or decrease the amount of information more than
necessary, whether in terms of the length of the verbal contribution or its meaning
(informativeness). What the speaker does, though, is misrepresent the reality as
he/she believes it to be, and is thereby untruthful. It does not matter how big a
departure from the speaker-believed reality a given meiosis or hyperbole displays.
Truthfulness, as reflected by Grice’s first Quality maxim, is not a matter of degrees
but rests on a binary opposition. In Grice’s radical (and correct) view, a meiotic/
hyperbolic utterance either is or is not truthful, and it cannot be something in-
between. Therefore, although not all instances of hyperbole and meiosis will strike
language users as being extreme, technically, they all do depend on untruthfulness
without exception. For instance, calling someone “upset,” whilst he/she was “abso-
lutely devastated” is untruthful in a technical sense, just as saying “I ate three cook-
ies” if I actually ate five is untruthful. This conceptualisation may indeed provoke
misgivings, especially when extremities are not used.
Grice (1989a [1975]: 34) presents metaphor as a figure that involves “catego-
rial falsity”. He illustrates this point by dint of an example: “You are the cream in
my coffee”, whereby “the speaker is attributing to his audience some feature or
features in respect of which the audience resembles (more or less fancifully) the
mentioned substance.” The meaning that Grice sees the speaker to be communicat-
ing is “You are my pride and joy” (1989a [1975]: 34). Interestingly, Grice conceives
14. Grice also allows for the fact that the speaker’s long-windedness may be considered not to be
designed. However, this falls outside Grice’s model of intentional communication of meanings.
of intertwining this metaphor with irony. He suggests that the interpretation of

metaphor precedes the unravelling of the ironic implicature. The understanding
of the irony consists in having deciphered an implicature based on the flouting of
the first maxim of Quality in the subordinate implicature, which shows a distinct
level of untruthfulness and Quality flouting (see Dynel 2016b for a discussion).
From among the four figures, irony is the one to which Grice devotes most at-
tention, which amounts to a few remarks, though. These few remarks have stimulat-
ed a lot of scholarly interest and controversy in post- and neo-Gricean scholarship.
To illustrate irony, Grice (1989a [1975]: 34) presents the following example: “X is
a fine friend” said by A about a friend who “has betrayed a secret of A’s to a business
rival.” The speaker’s utterance should then be interpreted as “the most obviously
related proposition”, the one “contradictory of the one he purports to be putting for-
ward” (Grice 1989a [1975]: 34). Thus, “X is not a fine friend” or perhaps “X is a poor
friend” is the meaning the speaker implicates, based on propositional contradiction.
Being ironic, the speaker hence makes as if to say “something he does not believe”
(Grice 1989a [1975]: 34), while the hearer supposes the speaker “to mean the nega-
tion of what he has made as if to say” (1989b [1978]: 53). However, as Grice’s (1989b
[1978]: 53) example “Look, the car has all its windows intact” (said in reference to
a car with a shattered window) indicates, the sole presence of meaning reversal is
insufficient as the sine qua non for irony. The speaker’s expression of judgement/
evaluation is the second necessary definitional component. Grice (1989b [1978])
also adds that an ironic utterance cannot be preceded by an explicit signal of its
presence, suggesting that it involves pretence, understood in accordance with the
word’s etymology (Grice 1989b [1978]: 54). This necessarily overt but unannounced
pretence, which may be regarded as associated with making as if to say and overt
untruthfulness, is to be recognised by the hearer. Otherwise, irony would fall flat.
This conceptualisation of irony has come in for heavy criticism, a large propor-
tion of which has been voiced by the supporters of relevance theory and the other
alternative accounts. Some of this criticism concerns the lack of cognitive validity
in Grice’s proposal, as evidenced by experimental tests (but see Reimer 2013 for
critical evaluation thereof). Notwithstanding such debates, in no way does Grice
consider the actual mental processes involved in the interpretation of irony, which
is the focus of cognitive experimental studies, giving “only” a descriptive philo-
sophical account of communication, and making a few remarks about irony. This
should be borne in mind when his work is criticised.
Many authors have rightly criticised Grice’s brief discussion of irony for being
non-exhaustive and inapplicable to all irony (see Garmendia 2015; Dynel 2011b,
and references therein). In Grice’s defence, however, he cannot have aspired to
define irony, in all its forms and guises, but to show that it may be a vehicle for
implicature based on the flouting of the first maxim of Quality. It is then hardly
surprising that irony should be much more diversified than Grice’s few notes appear
136 Marta Dynel
to indicate. Although Grice may not have conceived of different forms of irony, he
cannot be said to have purposefully excluded them, contrary to what some authors
tend to suggest. He simply failed to recognise the heterogeneous nature of the fig-
ure, whilst his general proposal, if adequately modified, may be applicable to the
various forms irony manifests itself in (see Dynel 2013b).
Although the key example Grice deploys while explaining the mechanics of
irony relies on the contradiction of a proposition (Grice 1989a [1975]: 34), the con-
cept of a proposition is not critical to his presentation of irony. His account based
on flouting the first maxim of Quality and the emerging evaluative implicature will
easily accommodate irony that centres on the sub-propositional contradiction of
semantic/lexical meaning of an untruthful lexical element (cf. Wilson 2006) in an
utterance, which does not span the entire propositional content (e.g. “I was happily
weeping in the corner for an hour, and nobody even bothered to ask me what the
problem was”). Since saying does not need not coincide with asserting, and making
as if to say need not coincide with making as if to assert, flouting the first maxim
of Quality manages to encompass ironic exclamations, imperatives and questions,
contrary to what critics may claim. In the same vein, Grice’s framework will also
encompass the negation of pragmatic meaning of an utterance (e.g. uttering “Thank
you” when someone has shut the door in your face to convey disappointment and
annoyance) (see Dynel 2011b). In a nutshell, untruthfulness may manifest itself at
the level of a whole proposition, in its parts, as well as at the level pragmatic force
(for a similar conceptualisation, yet involving SAT, see Haverkate 1990).
However, cases of irony based on what is said (dubbed verisimilar irony, e.g. “I
appreciate the silence of this library” addressed to someone who keeps talking on
the phone while sitting there) may be more problematic (see Dynel 2017, and refer-
ences therein). This type of irony does involve what is said, besides communicating
the pivotal evaluative implicature. At first blush, it does not flout the first maxim
of Quality or necessitate meaning negation, either. It may be postulated, however,
that it does display both the features, yet at the level of as if implicature, for it flouts
the maxim of Relation at the level of what is said (cf. Dynel 2017).
Another issue subject to criticism is that Grice fails to differentiate irony from
the other figures capitalising on Quality floutings (e.g. Wilson 2006; Wilson and
Sperber 2000 [2002, 2012]; cf. Garmendia 2015). 15 Indeed, Grice does not seem
to be concerned with why the figures are applied or how they differ, because he
15. Another tenuous counterargument against the Gricean view is that it does not explain the
rationale underlying irony or its puzzling features, such as the normative bias or the tone of voice
(Wilson and Sperber 2012), which are accounted for by the relevance-theoretic model. Grice must
have had no aspirations to capture these characteristics of irony, which should not be regarded
as a major drawback of his approach.
merely gives them as an illustration of the flouting of the first maxim of Quality.
However, in his second lecture on logic and conversation, Grice (1989b [1978])
refines his approach and explicitly states that irony must necessarily convey an
attitude towards its object. Hence, he highlights irony’s distinctive feature: “Irony
is intimately connected with the expression of feeling, attitude, or evaluation. I
cannot say something ironically unless what I say is intended to reflect a hostile or
derogatory judgement or a feeling such as indignation or contempt” (Grice 1989b
[1978]: 53–54; see Garmendia 2015 for discussion). Wilson (2006) does acknowl-
edge Grice’s addition but indicates that it is insufficient, because Grice does not
specify the object of evaluation, or the connection between the utterance and the
evaluation. Again, these are the issues to which Grice’s philosophical account does
not pay any heed. Admittedly, the referent must be chosen in a commonsensical
manner for each ironic utterance. For instance, ironic criticism may concern a
previous utterance (produced much earlier or in the preceding turn), the hearer’s
or someone else’s action, the state of affairs, or a combination thereof.
Overall, from Grice’s perspective, implicated (negative) evaluation based on
flouting the first maxim of Quality forms the sine qua non for irony, which involves
unannounced overt pretence and frequently necessitates some form of meaning
negation. This is not the case with the other Quality-based figures, which are based
on overt explicit 16 untruthfulness but do not evince the necessary combination
of the conditions met by irony. Meiosis and hyperbole may also be considered to
carry evaluation, but this evaluation is of a special type, for it concerns attenuation
or augmentation and may be positive or negative. Importantly, this evaluation is
not conducive to meaning reversal, but meaning substitution with a lexical item
conveying greater (for meiosis) or lesser (for hyperbole) intensity. Whilst not all
ironic implicatures pivot merely on meaning reversal, those which are hinged on
surface-level evaluation most typically do. It is this evaluative expression that is
amenable to meaning opposition (as in Grice’s canonical example), which invites
truthful evaluative implicature. Meiotic/hyperbolic evaluation will show an iron-
ic character only on condition that meaning reversal is involved in it (see Dynel
2016b). On the other hand, metaphor can hardly be mistaken for irony. Firstly, no
meaning contradiction comes into play, insofar as metaphors are translated into
literal expressions in the light of the relevant features of the vehicle attributed to
the tenor. Secondly, although metaphor can also convey criticism or praise via im-
plicature, evaluation is by no means inherent in its workings (cf. Garmendia 2010).
16. Verisimilar irony, based on truthful utterances, does not exhibit overt explicit untruthfulness
(but may be regarded as displaying overt implicit untruthfulness, cf. Dynel 2013b, 2017), which
is why it can hardly be mistaken for the other figures.
138 Marta Dynel
In conclusion, irony is a figure independent and distinguishable from the other

Quality-based figures, but it may co-occur with any of them (see Dynel subm b).
These second-order debatable issues aside, Grice’s approach to irony and the
other figures is overburdened with several fundamental problems.
5. Quality-based figures as an alleged flaw in Grice’s proposal

of implicature
There is some criticism of the Gricean model in respect of the Quality-based figures,
mainly with reference to irony. Such criticism is not necessarily well-founded, as
will be argued here (see also Garmendia 2011, 2015).
Wilson (2006: 1725) states that in tropes, such as irony, the recovery of the
speaker’s implicature neither restores the Cooperative Principle and maxims, nor
explains “why a maxim has been violated”, since “if the speaker has said something
she believes to be false, the situation cannot be remedied by the recovery of an
implicature”. Similarly, Wilson and Sperber (2000 [2002, 2012]: 221) insist that in
the Quality-based figures of speech, “the maxim of truthfulness is irretrievably
violated, and the implicature provides no circumstantial justification whatsoever.”
In other words, the tropes undermine Grice’s proposal that maxim floutings con-
ducive to conversational implicature are “in the circumstances justifiable, at least
in his [the speaker’s] eyes” (Grice 1989c: 370). Also, Wilson (2006: 1724) criticises
Grice for failing to account for why meaning should be conveyed ironically via
“blatant falsehood” (technically, this should be deemed what the speaker believes
to be false) if the same meaning might as well be “literally expressed”. Furthermore,
from the critics’ perspective, implicatures are typically added to what is said. Indeed,
conversational implicatures are frequently seen as meanings that arise in addition to
what is said (e.g. Carston 2002; Haugh 2015 and references therein). However, this
does not apply to irony, as well as the other three tropes, which is unproblematic for
Grice’s model, which the critics fail to acknowledge (see Garmendia 2011, 2015).
All this leads them to a conclusion that the Quality-based figures fall outside Grice’s
original proposal of how implicatures are generated. Wilson (2006: 1725) suggests
that “Grice had to extend both his notion of implicature and his account of how
implicatures are derived.” These claims appear to be specious on various grounds.
Firstly, Grice seems not to have been preoccupied with the reasons for commu-
nicating implicatures. As he concludes his discussion of conversational implicature,
he merely states that “there may be various possible specific explanations, a list of
which may be open,” for why conversational implicature comes into being and
“the conversational implicatum in such cases will be disjunction of such specific
explanations” (1989a [1975]: 40). He takes for granted the fact that implicatures
may be variously motivated, whether arising from tropes or other forms of implicit
language use. For instance, the speaker may intend to create a rhetorical effect (e.g.
emphasise the evaluation in irony), to be polite or impolite, to invite a humorous
reaction, or to come over as being creative or witty, all of which may underlie the
use of tropes. However, Grice did not aspire to explain why irony (or the other three
figures) should be used. He merely sketchily depicted them by placing them among
other communicative phenomena in his framework of communication.
Secondly, as reported in the previous section, in his presentation of the le-
gitimate sources of conversational implicature under the Cooperative Principle,
according to “a procedure by which a maxim is flouted” (Grice 1989a [1975]: 33),
Grice (1989a [1975]: 34) lists irony and metaphor, as well as meiosis and hyperbole,
as the cases in which the first maxim of Quality is flouted. These implicatures do
not differ from any other implicatures consequent upon floutings/exploitations
of the other maxims, whilst the Cooperative Principle holds, and the maxims are
observed at the level of implicature (cf. Grice 1989a [1975]: 33). It is difficult to tell
why the first maxim of Quality should be “irretrievably violated.”
It is indeed the case that most examples of conversational implicatures that
Grice (1989a [1975]: 32–35) provides are piggy-backed onto what is said, whose
import is communicated, even though it is secondary to the key implicature. The
implicature is the central meaning the speaker intends to communicate, and which
vindicates maxim flouting. For instance, addressing tautology as a case of flouting
the first maxim of Quantity, Grice concedes that “at the level of what is said, in my
favored sense, such remarks are totally noninformative and so, at that level, cannot
but infringe the first maxim of Quantity” (1989a [1975]: 33). What is said is subsid-
iary to the implicature that the maxim flouting invites. Still, some may extrapolate
from this a conclusion that Grice does not allow for speaker meaning coinciding
solely with implicature, however insignificant what is said may be.
Nonetheless, even before presenting the whole list of maxim floutings, Grice
suggests two sources of implicature: “A man saying (or making as if to say) that p
has implicated that q” (1989a [1975]: 30). Grice (1989b [1978]: 41) clearly indicates
the possibility of lack of what is said when addressing the problem of distinct layers
of speaker meaning: “in a given case one or more of these elements may be lacking.
For example, nothing may be said, though there is something which a speaker
makes as if to say.” Importantly, Grice also stresses that “the implicature is not
carried by what is said, but only by the saying of what is said, or by ‘putting it that
way’” (1989a [1975]: 39). This undermines the assumption that what is said is the
direct, and hence indispensable, source of implicature (see also Dinges 2015). In
the light of these quotations, it is evident that implicature may capitalise on either
what is said or making as if to say. In the latter case, implicature exhausts speaker
meaning. The figures dependent on flouting the first maxim of Quality represent
140 Marta Dynel
this category, which may indeed be quite restricted in scope (but wider that Grice
originally thought, cf. metonymy and bald-faced lies). It is not the case, however,
that this makes such implicatures different or less important, compared to those
which arise from what is said. Grice’s solution to capture these figures (which are
prominent in language use) does not sound forced. The fact that, on occasion, Grice
does not explicitly take into account the cases where what is said is not present as
part of speaker meaning (by listing making as if to say next to it) may be deemed an
inadvertent omission on his part, rather than a purposeful statement. The bottom
line is that making as if to say suffices as the bedrock for conversational implicature,
and this is by no means a marginal or exceptional situation.
The proponents of the relevance-theoretic approach, however, have another
misgiving about Grice’s framework. Equating what is said with asserting, Wilson
(1995) and Wilson and Sperber (2000 [2002, 2012]: 221), question the status of the
first maxim of truthfulness:
it is hard to see why a maxim of truthfulness is needed at all. It seems to follow
from the very notion of an assertion as a commitment to truth […] that your as-
sertions should be truthful. In fact, the only pragmatic function of the maxim of
truthfulness, on this interpretation, is to be violated in metaphor and irony, thus
triggering the search for an implicature.
Indeed, the first maxim of Quality is Grice’s tool to account for the tropes (not
only metaphor and irony, but also hyperbole and meiosis). However, this maxim,
like all the other ones, necessarily underlies all utterance types, not only assertions
(cf. the argument that saying is not asserting in Section 3.1). This is why it is by no
means otiose. Additionally, assertions may be false (being tantamount to lying), just
as other utterance types (promoting other deception), but this maxim obliges the
speaker not to be covertly mendacious, which is the primary reason for its exist-
ence (see Vincent Marrelli 2003, 2004; Meibauer 2005; Dynel 2011b). Finally, it is
dubious whether the word “say” in the formulation of the maxim is used in Grice’s
favoured sense. This is the most vexing issue in Grice’s discussion of the four tropes.
Grice’s view of irony (and the other tropes as well) is considered to involve an
internal contradiction, and thus an error. It is caused by the notion of “making as
if to say” and the formulation of the first maxim of Quality as “do not say” (Wilson
1995; Wilson and Sperber 2000 [2002, 2012]). The critics claim that it is the fa-
voured meaning of “say” (i.e. “assert”) 17 that Grice uses in the formulation of the
first Quality maxim. Consequently, if the speaker only makes as if to say in irony
and the other figures, then the maxim is not overtly violated (Wilson 1995; Wilson
17. As argued earlier (see Section 3.1), there are no reasons to believe that saying means asserting,
even though it depends on the speaker’s commitment.
and Sperber 2000 [2002, 2012]; cf. Garmendia 2015). In other words, if nothing
is said but only made as if to say, the first maxim of Quality is not flouted, for the
speaker does not say, strictly speaking. As a result, no implicature can come into
being. By contrast, Grice’s pivotal assumption is that irony and the other three
figure centre on the first Quality maxim’s flouting and implicatures. A solution to
this internal problem is not easy to find.
Garmendia and Korta (2007: 192), supportive of Grice’s approach, propose that
the first maxim of Quality should be reformulated as “Don’t say or make as if to say
what you believe to be false” so that the maxim can be flouted and the implicature
can arise. This is, they suggest, how Grice’s proposal be salvaged. However, this
significant modification seems to deprive the maxim of its original formal appeal
and comprehensibility, and makes it stand out among the other maxims, especially
when juxtaposed with the second maxim of Quality. There may be a simpler solu-
tion to the puzzle.
It might be claimed that proposing the Quality maxims, Grice uses the word
“say” in the sense of “utter” or “verbalise”, rather than using it in the “favoured”
technical sense to capture on level of speaker meaning. This technical use of “say”
would thus (wrongly) suggest that implicature is not amenable to Quality. It is
interesting to notice that the two maxims of Quality are the only ones which rest
on the word “say”. The other ones are based on commonsensical/folk notions, be-
ing devoid of any technical terminology. Nowhere does Grice state or even imply
that this “say” in the two maxims should be taken in his “favored” sense, which
he tends to do in his writings where he is not talking about what is said vs. what
is implicated. Moreover, Grice tends to vacillate in his use of “say” between two
different meanings, one technical (his favoured sense) and the other conventional,
typical of ordinary language use. This is evidenced, for instance by the following
quotations, where Grice uses “say” in both ways: “how what is said is to be said”
(1989a [1975]: 27), “the saying of what is said, or by ‘putting it that way’” (1989a
[1975]: 39) “the manner in which what is said has been said” (1989b [1978]: 41), or
“saying what is said” (1989b [1978]: 43, 51). Additionally, he frequently uses “say”
according to the folk/conventional understanding of the word when he presents
his examples, for instance the canonical ironic utterance: “A says X is a fine friend”
(1989a [1975]: 34, 1989b [1978]: 53, emphasis mine), or meiosis: “one says He was
a little intoxicated” (1989a [1975]: 34, emphasis mine).
Taking all this into consideration, what Grice must have meant under the for-
mulation of the first maxim is simply “Do not utter/verbalise what you believe to
be false”, and this concerns any utterance type (i.e. a statement, question or im-
perative). Consequently, when the speaker does utter what he believes to be false,
making this overt, the hearer needs to seek a reason for the speaker’s departure
from the maxim, thereby finding the relevant implicature, the speaker meaning.
142 Marta Dynel
The notion of making as if to say does not suggest an alternative version of realising
the first maxim of Quality. It only places emphasis on the fact that in irony (and the
other figures), no what is said is communicated and truthful implicature arises. 18
Epilogue
This chapter has revisited Grice’s postulates concerning irony, metaphor, hyperbole
and meiosis, the four figures of speech dependent on flouting the first maxim of
Quality. These constitute a salient group of implicatures in Grice’s model of con-
versational logic based on the Cooperative Principle. In the course of the chapter,
several arguments were proposed in defence of Grice’s seminal postulates. Whether
these were interpreted plausibly remains open to future discussion. Whether they
were interpreted in line with Grice’s thought is never to be known.
Acknowledgements
This work was supported by the Ministry of Science and Higher Education (Project number
IP2015 012874, Decision 0128/IP3/2016/74).
References
Akmajian, Adrian, Richard Demers, Ann Farmer, and Robert M. Harnish. 2001. Linguistics: An
Introduction to Language and Communication. Cambridge, MA: The MIT Press.
Baptista, Luca. 2014. “Say What? On Grice on What Is Said.” European Journal of Philosophy 22
(1): 1–19. doi: 10.1111/j.1468-0378.2011.00467.x
Benton, Matthew. 2014. “Gricean Quality.” Noûs. doi: 10.1111/nous.12065
Bhaya Nair, Rukimi. 1985. “Telling Lies: Some Literary and Other Violations of Grice’s Maxim
of Quality.” Nottingham Linguistic Circular 14: 53–71.
Brandom, Robert. 1983. “Asserting.” Noûs 17: 637–650. doi: 10.2307/2215086
Brown, Penelope, and Stephen Levinson. 1987. Politeness: Some Universals in Language Usage.
Cambridge: Cambridge University Press.
Camp, Elisabeth. 2006. “Contextualism, Metaphor and What Is Said.” Mind & Language 21(3):
280–309. doi: 10.1111/j.1468-0017.2006.00279.x
18. However, in practice, such implicatures may be covertly untruthful, and hence conducive to
deception (Dynel 2011b, 2015).
Colston, Herbert. 2000. “On Necessary Conditions for Verbal Irony Comprehension.” Pragmatics
and Cognition 8: 277–324. doi: 10.1075/pc.8.2.02col
Colston, Herbert, and Jennifer O’Brien. 2000. “Contrast and Pragmatics in Figurative Language:
Anything Understatement Can Do, Irony Can Do Better.” Journal of Pragmatics 32: 1557–1583.
doi: 10.1016/S0378-2166(99)00110-1
Davies, Bethan. 2000. “Grice’s Cooperative Principle: Getting the Meaning Across.” Leeds Working
Papers in Linguistics and Phonetics 8: 1–26.
Davies, Bethan. 2007. “Grice’s Cooperative Principle: Meaning and Rationality.” Journal of
Dinges, Alexander. 2015. “Innocent Implicatures.” Journal of Pragmatics 87: 54–63.
Dynel, Marta. 2008. “There Is Method in the Humorous Speaker’s Madness: Humour and Grice’s
model.” Lodz Papers in Pragmatics 4(1): 159–185. doi: 10.2478/v10016-008-0011-5
Dynel, Marta. 2011a. “Turning Speaker Meaning on Its Head: Non-Verbal Communication and
Intended Meanings.” Pragmatics and Cognition 3: 422–447. doi: 10.1075/pc.19.3.03dyn
Dynel, Marta. 2011b. “A Web of Deceit: A neo-Gricean View on Types of Verbal Deception.”
International Review of Pragmatics 3(2): 137–165. doi: 10.1163/187731011X610996
Dynel, Marta. 2013a. “Being Cooperatively Impolite: Grice’s Model in the Context of (Im)Politeness
Theories.” In Research Trends in Intercultural Pragmatics, ed. by Istvan Kecskes and Jesus
Romero-Trillo, 55–83. Berlin: Mouton de Gruyter. doi: 10.1515/9781614513735.55
Dynel, Marta. 2013b. “Irony from a Neo-Gricean Perspective: On Untruthfulness and Evaluative
Implicature.” Intercultural Pragmatics 10: 403–431. doi: 10.1515/ip-2013-0018
Dynel, Marta. 2015. “Intention to Deceive, Bald-Faced Lies, and Deceptive Implicature: Insights
into Lying at the Semantics-Pragmatics Interface.” Intercultural Pragmatics 12: 309–332.
doi: 10.1515/ip-2015-0016
Dynel, Marta. 2016a. “Contrasting and Combining Covert and Overt Untruthfulness: On Lying,
Deception, Irony and Metaphor.” Pragmatics & Cognition 23: 175–209.
Dynel, Marta. 2016b. “Two Layers of Overt Untruthfulness: When Irony Meets Metaphor,
Hyperbole or Meiosis.” Pragmatics & Cognition.
Dynel, Marta. 2017. “The irony of irony: Irony based on truthfulness.” Corpus Pragmatics 1: 3-36.
Fogelin, R. 1988. Figuratively Speaking. New Haven: Yale University Press.
Fraser, Bruce. 1983. “The Domain of Pragmatics.” In Language and Communication, ed. by Jack
Richards, and Richard Schmidt, 29–59. London: Longman.
Garmendia, Joana. 2010. “Irony Is Critical.” Pragmatics & Cognition 18: 397–421.
doi: 10.1075/pc.18.2.07gar
Garmendia, Joana. 2011. “She’s (Not) a Fine Friend: “Saying” and Criticism in Irony.” Intercultural
Pragmatics 8: 41–65. doi: 10.1515/IPRG.2011.002
Garmendia, Joana. 2015. “A (Neo)Gricean Account of Irony: An Answer to Relevance Theory.”
International Review of Pragmatics 7: 40–79. doi: 10.1163/18773109-00701003
Garmendia, Joana, and Kepa Korta. 2007. “The Point of Irony.” In Language, Representation and
Reasoning, ed. by Mixel Aurnague, Kepa Korta, and Jesus M. Larrazabal, 189–200. Bilbao:
upv-ehu.
Gibbs, Raymond. 1994. The Poetics of Mind: Figurative Thought, Language and Understanding.
Grice, Herbert Paul. 1989a [1975]. “Logic and Conversation.” In Studies in the Way of Words,
Herbert Paul Grice, 22–40. Cambridge, Mass.: Harvard University Press.
144 Marta Dynel
Grice, Herbert Paul. 1989b [1978]. “Further Notes on Logic and Conversation.” Studies in the
Way of Words, Herbert Paul Grice, 41–57. Cambridge, Mass.: Harvard University Press.
Grice, Herbert Paul. 1989c. “Retrospective Epilogue.” In Studies in the Way of Words, Herbert
Paul Grice, 339–386. Cambridge, Mass.: Harvard University Press.
Grice, H.P. 1989d [1968]. “Utterer’s Meaning and Intentions”. In Studies in the Way of Words.
Cambridge, MA: Harvard University Press, 86–116.
Habermas, Jürgen. 1984. The Theory of Communicative Action I: Reason and the Rationalization
of Society, Transl. by Thomas McCarthy. Boston: Beacon Press.
Haugh, Michael. 2015. Im/Politeness Implicatures. Berlin: Mouton de Gruyter.
doi: 10.1515/9783110240078
Haverkate, Henk. 1990. “A Speech Act Analysis of Irony.” Journal of Pragmatics 14: 77–109.
doi: 10.1016/0378-2166(90)90065-L
Jary, Mark. 2010. Assertion. Basingstoke: Palgrave Macmillan. doi: 10.1057/9780230274617
Livnat, Zohar. 2004. “On Verbal Irony, Meta-Linguistic Knowledge and Echoic Interpretation.”
Pragmatics & Cognition 12: 57–70. doi: 10.1075/pc.12.1.05liv
Livnat, Zohar. 2011. “Quantity, Truthfulness and Ironic Effect.” Language Sciences 33(2): 305–315.
doi: 10.1016/j.langsci.2010.10.009
Mahon, James. 2015. “The Definition of Lying and Deception.” In The Stanford Encyclopedia of
Philosophy (Fall 2008 Edition), ed. by Edward N. Zalta, URL = <http://plato.stanford.edu/
archives/fall2015/entries/lying-definition/>
Meibauer, Jörg. 2005. “Lying and Falsely Implicating.” Journal of Pragmatics 37: 1373–1399.
doi: 10.1016/j.pragma.2004.12.007
Meibauer, Jörg. 2014. Lying at the Semantics-Pragmatics Interface. Boston/Berlin: Mouton de Gruyter.
doi: 10.1515/9781614510840
Mooney, Annabelle. 2004. “Co-operation, Violation and Making Sense.” Journal of Pragmatics
36: 899–920. doi: 10.1016/j.pragma.2003.10.006
Nemesi, Attila. 2010. “Data-Gathering Methods in Research on Hyperbole Production and
Interpretation.” In The Role of Data at the Semantic-Pragmatic Interface, ed. by Eniko T.
Nemeth and Karoly Bibok, 381–417. Berlin, New York: De Gruyter Mouton.
doi: 10.1515/9783110240276.381
Norrick, Neal. 2004. “Hyperbole, Extreme Case Formulations.” Journal of Pragmatics 36: 1727–
1739. doi: 10.1016/j.pragma.2004.06.006
Ozar, Anne. 2008. The Moral Significance of Sincerity. PhD Thesis. Fordham University.
Pagin, Peter. 2015. “Assertion.” In The Stanford Encyclopedia of Philosophy (Spring 2015 Edition),
ed. by Edward N. Zalta (ed.), URL = <http://plato.stanford.edu/archives/spr2015/entries/
assertion/>
Peirce, Charles. 1934. “Belief and Judgment.” In Collected Papers, vol. 5, ed. by Charles Hartshorne
and Paul Weiss, 376–387. Cambridge: Harvard.
Pomerantz, Anita. 1986. “Extreme Case Formulations: A Way of Legitimizing Claims.” Human
Studies 9: 219–229. doi: 10.1007/BF00148128
Reimer, Marga. 2013. “Grice on Irony and Metaphor: Discredited by the Experimental Evidence?”
International Review of Pragmatics 5: 1–33. doi: 10.1163/18773109-13050101
Saul, Jennifer. 2012. Lying, Misleading, and the Role of What Is Said. Oxford: Oxford University
Press.
Smith, John. 1657. Mysterie of Rhetorique Unvail’d. Facsimile reprint of 1st ed., London: George
Everfden.
Soames, Scott. 2008. “Drawing the line between meaning and implicature—and relating both to
assertion.” Noûs 42: 440–4655. doi: 10.1111/j.1468-0068.2008.00691.x
Stokke, Andreas. 2013. “Lying and Asserting.” Journal of Philosophy 110: 33–60.
doi: 10.5840/jphil2013110144
Vincent Marrelli, Jocelyne. 2003. “Truthfulness.” In Handbook of Pragmatics, ed. by Jef
Verschueren, Jan-Ola Östman, Jan Blommaert, and Chris Bulcaen, 1–48. Amsterdam/
Philadelphia: John Benjamins.
Vincent Marrelli, Jocelyne. 2004. Words in the Way of Truth. Truthfulness, Deception, Lying across
Cultures and Disciplines. Napoli: Edizione Scientifiche Italiane.
Wilson, Deirdre. 1995. “Is There a Maxim of Truthfulness?” UCL Working Papers in Linguistics
7: 197–212.
Wilson, Deirdre. 2006. “The Pragmatics of Verbal Irony: Echo or Pretence?” Lingua 116: 1722–
1743. doi: 10.1016/j.lingua.2006.05.001
Wilson, Deirdre, and Dan Sperber. 2000 [2002, 2012]. “Truthfulness and Relevance.” UCL
Working Papers in Linguistics: 215–254. [Reprinted as Wilson, Deirdre, and Dan Sperber.
2002. “Truthfulness and Relevance.” Mind 111: 583–632; “Truthfulness and Relevance.”
In D. Wilson, and D. Sperber. Meaning and Relevance, 47–83. Cambridge: Cambridge
University Press.]
Wilson, Deirdre, and Dan Sperber. 2012. “Explaining Irony.” In Meaning and Relevance, Deirdre
Wilson, and Dan Sperber. Cambridge: Cambridge University Press, 123–14.
doi: 10.1017/CBO9781139028370.008
Chapter 7
Lexical pragmatics
and implicit communication
Deirdre Wilson and Patricia Kolaiti

University College London and New York College Athens
In this paper, we consider what is implicitly communicated when the linguisti-

cally-specified (encoded) meaning of a word or phrase is modified in use. Well-
studied examples of such modification include lexical narrowing, approximation
and metaphorical extension. A striking feature of much research in this area is
that narrowing, approximation and metaphorical extension tend to be seen as
distinct processes which lack a common explanation; indeed, the philosopher
Donald Davidson famously argued that since metaphor does not always deter-
mine “a cognitive content that its author wishes to convey and that the inter-
preter must grasp if he is to get the message”, it does not fall within the scope of
a theory of communication at all. Using corpus data to shed light on some of
the issues raised by this research, we will make two main points. First, there is a
continuum of cases between literal use, approximation, hyperbole and metaphor,
which favours the search for a unitary account that applies across the whole con-
tinuum. Second, what is implicitly communicated is not always a specific “cog-
nitive content that the speaker wishes to convey”: rather, there is a continuum of
cases with varying degrees of specificity, and an adequate theory of communica-
tion should explain the whole range.
Keywords: lexical narrowing, broadening, approximation, metaphor,

communication, speaker’s meaning
1. Introduction
Recent research on lexical pragmatics takes as its starting point the fact that words
are often used in ways that depart (sometimes a little, sometimes a lot) from their
encoded meanings, the ones assigned them by the grammar. We invent new words,
and people understand us. We blend two words together, and people understand us.
We use nouns, adjectives or prepositions as verbs, and people understand us. We
borrow words from other languages; we use words approximately, hyperbolically
doi 10.1075/pbns.276.07wil
148 Deirdre Wilson and Patricia Kolaiti
or metaphorically. As children or adults, we pick up the meanings of unfamiliar

words without being taught, just by hearing them uttered in context. We see words
come into fashion and vanish; we see them acquire new meanings and lose old ones.
While sociolinguists, historical linguists, philosophers and psychologists have long
been interested in different aspects of these phenomena, it is only in the last twenty
years or so that pragmaticists have begun to look systematically at how pragmatic
principles and mechanisms apply at the level of the word or phrase rather than the
whole utterance, and to talk of a separate domain of “lexical pragmatics”. 1
A central goal of lexical pragmatics is to investigate the processes by which
linguistically specified (encoded) word meanings are modified in use. Well-studied
examples of such processes include lexical narrowing, approximation and meta-
phorical extension, illustrated in (1)–(3):
(1) Susan has a temper.
(2) Norfolk is flat.
(3) That politician is a mouse.
In (1), temper is naturally interpreted as meaning not just “temper” but “bad
temper”, in (2), flat is interpreted as meaning not “strictly and literally flat” but
“flattish”, and in (3), mouse is interpreted as denoting not a small rodent but a
timid person. The challenge for lexical pragmatic theories is to explain how such
interpretations arise.
A striking feature of much existing research in this area is that narrowing, approx-
imation and metaphorical extension tend to be seen as distinct processes which lack
a common explanation. Thus, narrowing is often treated as a case of I-implicature,
involving a default inference to a stereotypical interpretation (Levinson 2000; see
also Blutner 1998, 2004). Approximation is often treated as a case of pragmatic
vagueness involving different contextually-determined standards of precision (Lewis
1979; Lasersohn 1999). In the Gricean literature, metaphor is widely seen as involv-
ing blatant violation of a pragmatic maxim of literal truthfulness in order to convey
a related implicature (Grice 1967/1989; Levinson 1983). Typically, such accounts do
not generalise: metaphors are not analysable as rough approximations, narrowings
are not analysable as blatant violations of a maxim of literal truthfulness, and so
on. Thus, narrowing, approximation and metaphor tend to be seen as sui generis
processes involving distinct pragmatic mechanisms or principles.
In fact, many philosophers of language see metaphor as not falling within the
scope of a theory of communication at all. In their view, communication necessarily
1. See e.g. Carston (1997, 2002); Blutner (1998, 2004); Lascarides and Copestake (1998); Sperber
and Wilson (1998, 2008); Glucksberg (2001, 2003); Fauconnier and Turner (2002); Wilson and
Sperber (2002, 2012); Horn (2004, 2012); Recanati (2004, 2010); Wilson and Carston (2007).
Chapter 7. Lexical pragmatics and implicit communication 149
involves the intention to convey a specific cognitive content – e.g. a Gricean speak-
er’s meaning – which can be rendered as a proposition or a small set of propositions,
whereas in metaphor, there is typically no such specific content that the speaker
intends to convey. As Donald Davidson (1978: 48–49) puts it,
When we try to say what a metaphor ‘means’, we soon realise that there is no end to
what we want to mention. If someone draws his finger along a coastline on a map
…, how many things are drawn to your attention? You might list a great many, but
you could not finish since the idea of finishing has no clear application. How many
facts or propositions are conveyed by a photograph?
For Davidson, the problem is not just that what is communicated by use of a met-
aphor does not fit the definition of a Gricean speaker’s meaning; it is that nothing
is communicated at all:
The central error about metaphor is most easily attacked when it takes the form of a
theory of metaphorical meaning, but behind that theory, and statable independent-
ly, is the thesis that associated with a metaphor is a cognitive content that its author
wishes to convey and that the interpreter must grasp if he is to get the message. This
theory is false, whether or not we call the purported cognitive content a meaning.
(Davidson 1978: 46)
This line of argument has been developed by Lepore and Stone (2010, 2015), who
argue explicitly that metaphors cannot be handled using the standard machinery
of semantics and pragmatics:
The commonplace view about metaphorical interpretation is that it can be charac-
terised in traditional semantic and pragmatic terms, thereby assimilating metaphor
to other familiar uses of language. We will reject this view, and propose in its place
the view that, though metaphors can issue in distinctive cognitive and discourse
effects, they do so without issuing in metaphorical meaning and truth, and so
without metaphorical communication. (Lepore and Stone 2010: 1)
In contrast with these non-unitary approaches, relevance theorists have been trying
to develop a unitary account of lexical-pragmatic processes based on three main
claims. First, there is no pragmatic maxim of literal truthfulness of the type proposed
by Grice, so that what is implicitly communicated by use of a metaphor or hyperbole
must arise through some other route than blatant violation of this maxim (Wilson
and Sperber 2002). Second, encoded word meanings are no more than a clue to the
speaker’s meaning, and are typically adjusted (e.g. broadened or narrowed) in the
course of the comprehension process, using information accessible in the discourse
context (Carston 1997, 2002; Sperber and Wilson 1998). Third, there is a continuum
of cases of broadening, from approximation through various types of loose use to
“figurative” uses such as hyperbole and metaphor, which all involve the same in-
terpretive mechanisms and can be explained in the same way (Sperber and Wilson
1985/6, 2008; Wilson and Carston 2006, 2007; Vega Moreno 2007). On this unitary
approach, metaphor and hyperbole are not seen as rhetorical figures involving dis-
tinct interpretive mechanisms or principles. Instead, they fall squarely within the
domain of a theory of overt intentional (“ostensive”) communication designed to
handle a continuum of cases of varying degrees of specificity, ranging from those
where there is indeed a specific cognitive content a speaker wants to get across, to
those where the speaker’s intention is much vaguer, and what is implicitly conveyed
might be best described as an “impression” rather than a speaker’s meaning (Sperber
and Wilson 1986/1995: 46–54; Sperber and Wilson 2015).
How might the use of corpus data shed light on these differing approaches to
lexical pragmatics? In what follows, we will illustrate how we used corpus-based
evidence to test two main hypotheses put forward by the relevance-based unitary
account. 2 The first, discussed in Section 2, is that lexical narrowing is a highly flex-
ible, creative and context-sensitive process, which cannot be satisfactorily handled
in terms of a notion of default inference. The second, discussed in Section 3, is that
there is no sharp theoretical distinction between literal, loose and metaphorical
uses, but a continuum of cases with no clear cut-off point between them, which
all involve the same interpretive mechanisms and are understood in the same way.
In Section 4, we argue that lexical narrowing and broadening can combine in the
interpretation of a single word, which strongly favours a unitary account of the full
range of lexical-pragmatic processes. In Section 5, we return to the issue of whether
a theory of communication should deal only with cases where there is a specific
cognitive content that the speaker wishes to convey. Section 6 is a brief conclusion.
2. Lexical narrowing
Lexical narrowing involves the use of a word or phrase to convey a more specific
concept (with a narrower denotation) than the linguistically encoded meaning (see
Huang, this volume), as when temper in (1) above is interpreted as meaning “bad
temper”, or reputation in (4) is interpreted (on different occasions) as meaning
“good reputation” or “bad reputation”:
(4) Bill has a reputation.
2. The corpus analyses described were carried out as part of the AHRC project “A Unified
Theory of Lexical Pragmatics” (AR16356), on which Deirdre Wilson was Principal Investigator
and Robyn Carston Co-Investigator. We would like to acknowledge the support of the AHRC
and thank Robyn Carston, Tim Wharton and Rosa Vega-Moreno for valuable comments and
discussion. The corpus used was the 56 million word Bank of English; for further details, see
Kolaiti and Wilson 2014.
One of the goals of lexical pragmatics is to explain how hearers infer the appro-
priate narrowed interpretation on the basis of the linguistically encoded meaning,
together with contextual information.
A common approach to narrowing, put forward by Stephen Levinson (2000: 37–
8, 112–34), treats it as a type of default inference governed by an Informativeness
heuristic (“What is expressed simply is stereotypically exemplified”), itself backed
by a more general I-principle instructing the hearer to
Amplify the informational content of the speaker’s utterance, by finding the most
specific interpretation, up to what you judge to be the speaker’s m-intended point
(…). (Levinson 2000: 114)
The I-heuristic might be seen as dealing with stereotypical narrowings, as when bach-
elor is understood to mean a stereotypical bachelor, while the I-Principle might deal
with less stereotypical cases such as (1) or (4), where the degree of bad temper or the
type of reputation indicated may vary from occasion to occasion. 3 On this approach,
hearers are seen as automatically constructing a stereotypical (or otherwise enriched)
interpretation and accepting it in the absence of contextual counter-indications. For
Levinson, who interprets the notions of implicitness and implicature in a very broad
sense, the resulting enriched interpretation is seen as implicitly communicated even
though it contributes to the truth-conditional content of the utterance.
The alternative relevance-theoretic view, which we favour, is that lexical nar-
rowing is a far more creative and flexible process, involving the construction of
ad hoc, occasion-specific concepts influenced by a much wider range of cogni-
tive and contextual factors than default approaches take into account (e.g. Carston
1997, 2002, Sperber and Wilson 1998, 2008, Wilson and Sperber 2002; Wilson
and Carston 2007). Thus, in order to satisfy the expectations of relevance raised
by (1) or (4), the concept of a temper, or a reputation, might be narrowed to dif-
ferent degrees, and in different directions, in different contexts, yielding a range
of occasion-specific (“ad hoc”) concepts, e.g. temper*, temper**, reputation*,
reputation**, reputation***, and so on. 4 For relevance theorists, the resulting
3. Notice, though, that the I-Principle does not explain how the hearer identifies the speaker’s
intended meaning, but presupposes that he has some independent means of judging what this
is. To put it slightly unkindly, the I-Principle says “Choose a more specific interpretation if you
think this is what the speaker intended.” But the goal of a pragmatic theory is to explain how
hearers decide that a certain meaning was intended, and given that lexical broadening is just as
common as lexical narrowing, the I-Principle – which deals only with narrowing – does not get
us any closer to this goal.
4. We will follow the usual practice in lexical pragmatics of representing linguistically specified
meanings (“lexical concepts”) in small capitals (temper) and occasion-specific meanings (“ad
hoc concepts”) in small capitals followed by one or more asterisks (temper*, temper**…).
narrowed interpretation consists, on the one hand, of an explicature containing

an ad hoc concept (e.g. temper*, reputation**) which contributes to the truth
conditional content of the utterance, and on the other hand, of an array of potential
implicatures. On this approach, the array of implicatures is part of what is implicitly
communicated, while the ad hoc concept is part of what is explicitly communicated.
In what follows, we will use implicitness and implicature, explicitness and explicature
in the narrower relevance-theoretic sense unless otherwise indicated (for discus-
sion, see Carston 2002; Wilson 2016a).
As a starting point for examining the substantive differences between Levinson’s
default-inference approach to narrowing and relevance theory’s more flexible, con-
text-dependent approach, we took a standard problem in lexical pragmatics that
does not seem obviously to favour either approach: the fact that the adjective red
is typically narrowed in different directions in common adjective-noun combina-
tions such as red eyes, red hair, red apple, red pencil, red stamp, etc. (picking out
a different shade, distributed in different ways across the surface of the object, in
different combinations). A default-based approach might handle this by assigning
red a different default interpretation for each common adjective-noun combination,
and predict that this will be automatically preferred in the absence of contextual
counter-indications. Our hypothesis was that, although there is probably a range
of fairly standard narrowings of red in the context of eyes, hair, apple, pencil, stamp,
etc., the interpretations revealed by our corpus data would still be diverse and cre-
ative enough to raise questions about the default approach. We will illustrate using
the common adjective-noun combination red eyes.
In fact, we found considerable evidence of the creativity and flexibility of nar-
rowing even in this common combination. In each case, the adjective red was plau-
sibly understood as communicating a slightly narrower concept (e.g. red*, red**)
appropriate to the wider discourse context, picking out a particular shade other
than focal red, differently distributed over the surface of the eyes. Here are some
illustrations:
(5) (…) red eyes denote strain and fatigue.
Here red is naturally interpreted as picking out a reddish-pink shade ranging

around the edges of the eye, on the bags under the eye and perhaps on part of the
cornea too. The exact shade and distribution the speaker is taken to convey would
vary depending on further contextual clues about the colour of the skin and the
degree of strain or fatigue involved.
(6) In a photography session: [This flashing light is] to stop you getting red eyes.
Here red is naturally interpreted as picking out a luminous, rusty red on the iris only.
(7) In a conversation about demons: (…) two burning red eyes she recalled (…).
Here red picks out a fiery and luminous red, distributed over both the cornea and
the iris or the iris alone.
There are also metaphorical uses, as in (8):
(8) (…) eyes red with resentment (…).
Out of a total of 54 occurrences of red eyes and its variants (e.g. “eyes+red” and
“red+intervening items+red”) in the corpus, our search identified 26 different such
“discourse contexts”. The results are summarised below, along with an indication
of the frequency of occurrence of the combination red eyes in each such context:
Table 1. Discourse contexts of red eyes and its variants

Context Lines Number of occurrences
Context of crying 4, 6, 18, 35, 38, 41, 42, 43, 14
44, 50, 57, 58, 59, 64
Context of hardship and/or stress 21, 40, 45, 53, 55, 60 6
and/or fatigue
Context of eyes gleaming in the dark 29, 30, 31 3
Context of flu/cold 51, 65 2
Context of a gorilla mask 1 1
Context of eye damage 2 1
Context of midgets 3 1
Context of dizziness 7 1
Context of sheep-like eyes 8 1
Context of Albinos 17 1
Context of red-eye effect 19, 20 2
Context of rage* 23, 54, 68 3
Context of resentment* 39 1
Context of demons 26 1
Context of drunkenness 27 1
Context of Caymans 28 1
Context of koels /cuckoos 34 1
Context of eczema 36 1
Context of heat and sand 54 1
Context of sore eyes 70 1
Context of colour of one’s eyes 11 1
Inconclusive cases: (context of fiction)
Terrestrials with long ears 10 1
Fictional insects 22 1
Unknown entity 24, 25, 47 3
“Giants” and “heroes” 32, 33 2
Unknown context 69, 9 2
Notice that 17 of the 26 discourse contexts occur only once. The significant propor-
tion of one-off uses suggests a level of creativity that poses problems for the default
account and argues for a more flexible, context-dependent approach.
These results provide some evidence for the view that a hearer interpreting the
phrase red eyes on different occasions draws on a wide range of contextual informa-
tion in constructing an overall interpretation. Relevant contextual factors include
the type of entity the eyes belong to (e.g. humans, animals, insects, demons or “a
terrifying [gorilla] mask with little red eyes that blinked”), the cause of redness (e.g.
eczema, drunkenness, crying, flu/cold, fatigue, exposure to heat, sand, light, etc.),
and the severity of the cause (affecting the degree and distribution of redness). The
degree and direction of narrowing seem to vary considerably from one discourse
context to another, and it is not obvious that any unique default analysis would
provide a better starting point for constructing the full range of interpretations than
the encoded “literal” meaning (which we assume simply specifies that the eyes in
question must be red in some respects). This case contrasts markedly with those
standardly discussed in the literature – for instance, Levinson’s secretary narrowed
to “female secretary” (Levinson 2000: 117) or Blutner’s red apple narrowed to “apple
with red skin” (Blutner 1998) – where a single “normal” or “stereotypical” interpre-
tation seems to hold across most or all discourse contexts.
The results also raise several questions for default-based approaches. For in-
stance, should the same default interpretation be seen as assigned in every case
(e.g. to every occurrence of red eyes in our table above, regardless of the discourse
context), or could there be several “default” interpretations, each appropriate to a
different discourse context? To account for the flexibility in interpretation revealed
by our search, there would either have to be a very large number of “default” inter-
pretations (raising the question of how hearers choose among them), or else the
default interpretation would have to be seen as overridden by contextual factors
in a very wide range of cases. A simpler alternative might be to assume (as on the
relevance-theoretic account we favour) that narrowing is directly affected by ency-
clopaedic knowledge and pragmatic principles, without passing through an initial
“default interpretation” stage.
A further question for default-based approaches is about how they handle cases
where the interpretation remains vague or open. In the absence of adequate contex-
tual clues, for instance, a hearer may narrow the interpretation only to some extent
(e.g. “red in a way that would be appropriate to the eyes of an imaginary insect”) or
leave the interpretation open and not make the effort to narrow at all. According to
relevance theory, narrowing should not apply automatically to every occurrence of
red eyes, but is triggered by pragmatic factors (in particular, the goal of finding an
interpretation that satisfies expectations of relevance). This account predicts that
hearers will only narrow to a point where the utterance becomes relevant enough
(i.e. to a point where it yields enough implications, for a low enough processing cost,
to satisfy the particular expectations of relevance raised in that discourse context).
In the absence of such triggering factors, it is predicted that narrowing will not take
place, and the resulting interpretation may be relatively vague.
In our search, we encountered 9 inconclusive cases where the entities described
as having “red eyes” were either not specified in the immediate linguistic context 5 or
were invented or non-existent living kinds (“fictional insects”, “terrestrials with long
ears”, etc.). Why assume that hearers construct a concrete mental representation of
the shade and distribution of redness over the eyes of a “terrestrial with long ears”
at all? It is a genuine problem for default-based approaches to explain what happens
to the automatic assignment of a default interpretation in such cases.
The notion of a default inference has been developed in several different ways
(see e.g. Levinson 2000: Chapter 1.5; Geurts 2009; Jaszczolt 2014). Here we will
consider how Levinson’s default-based account, which has had considerable influ-
ence in pragmatics, might deal with the corpus data above. According to Levinson
(2000), default narrowings are generalised conversational implicatures, to be dealt
with in a theory of utterance-type meaning designed to explain how sentences are
systematically paired with preferred interpretations irrespective of the contexts
in which they occur. For Levinson, a theory of utterance-type meaning contrasts
with a theory of utterance-token meaning, or speaker’s meaning, such as relevance
theory, which is designed to take context and speakers’ intentions into account. It
should follow that on Levinson’s approach, information about the wider discourse
context cannot be taken into account in the course of lexical narrowing, and the
same default interpretation (specifying a certain shade and degree of redness, dis-
tributed over certain parts of the eye) must be automatically assigned to every
occurrence of red eyes, regardless of any available contextual information about the
speaker, audience, preceding discourse, topic of conversation, observable physical
environment, and so on.
As Noveck and Sperber (2007) point out, on the assumption that communica-
tive systems tend to favour least-effort principles and to evolve in the direction of
increasing efficiency, the value of a default-based approach will depend heavily on
the distributional frequencies of interpretations on which the default interpreta-
tion proves acceptable and those in which it has to be overridden or cancelled for
contextual reasons. By our fairly generous estimate, a default approach of the type
Levinson proposes would guide the hearer in the right direction – and therefore
5. We restricted the discourse context to a default of 6 lines before and after the search term. If
the default context did not provide enough clues, we expanded the search to a further 10 lines
before and after the search term and, if the context was still insufficient, we marked the case as
open/ inconclusive.
help with processing costs – in roughly 50% of the cases in our table above (i.e. those
involving crying, fatigue, flu/cold, eye damage, eczema, heat/sand and sore eyes,
although each result would have to be contextually fine-tuned in the light of more
detailed contextual information about the speaker, addressee, person described,
cause of the condition, etc.), but would be positively misleading and incur the costs
of cancellation in the remaining 50% of cases. A more flexible inferential approach
such as relevance theory would involve context-sensitive – and therefore relatively
costly – fine tuning of the encoded lexical meaning in the full range of cases, but
without the costs of default derivation followed by cancellation and reinterpretation
in 50% of the cases. It is far from obvious that the statistical tendencies revealed by
our corpus justify a default rather than an inferential account of lexical narrowing
on grounds of economy of processing, yet this was the main rationale for the default
approach proposed in Levinson (2000: Chapter 1.3).
A further claimed advantage of default-based approaches to narrowing is that
they explain the ready accessibility of “normal”, or “stereotypical”, narrowings in the
absence of special contextual factors. However, there are other ways of explaining
this ready accessibility without appeal to defaults, as in Horn’s approach based on
his R principle (“Say no more than you must”, Horn 2004: 13) or relevance the-
ory’s approach based on the Presumption of Optimal Relevance, which predicts
that “normal” or “stereotypical” interpretations, being more frequently used and
therefore typically less costly to construct, will be favoured by the relevance-guided
comprehension heuristic 6 as long as they yield enough implications to satisfy the
audience’s expectations of relevance. Moreover, neo-Griceans such as Levinson,
Horn and Blutner have been primarily concerned with Grice’s category of general-
ised conversational implicatures – those that go through in the absence of special
contextual features – and have said little or nothing about how they would treat
loose, hyperbolic or metaphorical uses of language, which are heavily context de-
pendent and in Grice’s framework violate his first Quality maxim (“Do not say what
you believe to be false”). Relevance theorists have consistently argued against this
maxim and defended the view that there is a continuum between literal, loose and
metaphorical uses rather than a set of clearly definable theoretical categories which
play distinct roles in communication and comprehension (Wilson and Sperber
2002). In the next section, we will consider what light the corpus data can shed on
this debate.
6. “Follow a path of least effort in deriving implications: test interpretive hypotheses in order
of accessibility, and stop when you have enough implications to satisfy your expectations of
relevance” (Sperber and Wilson 2002/2012: 276).
3. The continuum of literal, loose and metaphorical uses
Lexical broadening involves the use of a word or phrase to convey a more general
concept (with a broader denotation) than the encoded meaning, as in (9), where
hexagonal is naturally interpreted as meaning “roughly hexagonal”, or (10), where
forever is interpreted as meaning “much longer than expected or desired”:
(9) France is hexagonal.
(10) It took forever to finish that paper
A striking feature of much research in this area is that different interpretive pro-
cedures have been proposed for a range of phenomena which could all be seen as
varieties of broadening. Thus, approximation is often treated as a case of pragmat-
ic vagueness involving different contextually-determined standards of precision
(Lewis 1979; Lasersohn 1999). Metaphor and hyperbole are still widely seen as
involving blatant violation of Grice’s first Quality maxim, with the use of metaphor
implicating a related simile or comparison and the use of hyperbole implicating
a related weaker proposition (Grice 1967/1989). Typically, these accounts do not
generalise: metaphors are not analysable as rough approximations, approximations,
which generally pass unnoticed, are not analysable as blatant violations of the first
Quality maxim, and so on.
Relevance theorists, by contrast, have been exploring the hypothesis that there
is no clear cut-off point between literal use, approximation, hyperbole and meta-
phor, but merely a continuum of cases of broadening, which are understood in the
same way, using the same relevance-guided comprehension heuristic described
above (footnote 6): that is, following a path of least effort in looking for impli-
cations that will make the utterance relevant in the expected way (Carston 1997,
2002, Wilson and Sperber 2002; Wilson and Carston 2006, 2007, 2008, Sperber
and Wilson 2008; Carston and Wearing 2011; Wilson 2011a). On this approach,
approximation, metaphor and hyperbole are not natural kinds, to be dealt with
by distinct pragmatic mechanisms, and there is no fact of the matter about what
is “really” a metaphor, hyperbole or approximation and what is not. In classifying
our corpus data, then, we used “approximation”, “hyperbole” and “metaphor” not
as theoretical terms but as handy descriptive labels to pick out a range of more or
less prototypical examples, in line with standard rhetorical practice. To illustrate,
consider the (invented) utterance in (11):
(11) The sea is boiling.
This might be intended and understood literally (as indicating that the sea is at or
above boiling point), as an approximation (indicating that the sea is close to boiling
point), a hyperbole (indicating that the sea is hotter than expected or desired) or
a metaphor (indicating that the sea, while not necessarily hot, is bubbling, churn-
ing, emitting vapour, etc.). The issue is whether these are theoretically distinct
interpretations involving different interpretive procedures, or whether they merely
occupy different points on a continuum, and are all understood in the same way,
by adjusting the encoded meaning so that it yields enough implications to satisfy
expectations of relevance.
To provide some evidence that might help to choose between these approaches,
we focused on the adjectives boiling, raw, and painless, all of which are strictly de-
fined but often loosely or metaphorically used. The results showed that broadening
is not rare in language use. In fact, in the cases of boiling and painless, loose uses
predominate:
Table 2. Relative Frequency of Literal and Loose uses of boiling

Type of use Frequency in % terms
Literal 49.4%
Literal or Approximate 14.2%
Loose (i.e. non-literal) 36.1%
Table 3. Relative Frequency of Literal and Loose uses of painless

Type of use Frequency in % terms
Literal 20.2%
Literal or Approximate 15.7%
Loose (i.e. non-literal) 64.0%
The results for painless illuminate the relation between literal use and approxima-
tion in unexpected ways. Consider (12):
(12) In a discussion of euthanasia: I would want something clean and painless: no
botch-ups. It would be the doctor or no one.
Here, the denotation of painless is plausibly understood as including not only cases
where the procedure was strictly and literally painless, but also those involving a
small amount of physical pain, which would still be insignificant compared to the
distress the patient would have to go through if allowed to die naturally. In other
words, the encoded concept painless is broadened to painless*, whose denotation
includes, but goes beyond, instances that are strictly and literally painless. Around
16% of all uses of painless fell into this category, with strictly literal uses making
up around 20%.
From a theoretical point of view, “approximations” are sometimes seen as ex-

cluding the possibility of a literal interpretation (as, for instance, describing an
object as squarish would generally be understood as excluding the possibility that it
is strictly and literally square). However, the frequency of cases such as (12), which
are indeterminate between literal and “approximate” interpretations, suggests that
many loose or approximate uses of words involve a type of broadening from which
the denotation of the linguistically encoded concept is not automatically excluded. 7
Our findings for boiling show in more detail the form that the continuum of
literal and loose uses of the same word might take. In a total of 332 occurrences,
we found 164 cases which could only be understood literally (to mean “at or above
boiling point”), as in (13):
(13) Poached eggs come out well in a small dish using boiling water.
There were a further 47 cases in which either a literal or an approximate interpre-

tation would be appropriate, as in (14):
(14) Cover the cake with the icing, smoothing with a knife dipped in boiling water.
By contrast, there were only 4 cases where an approximate interpretation would be

appropriate and a literal interpretation would not, as in (15):
(15) For sauce, melt chocolate (…) over boiling water, then beat until smooth.
(Those of you who have tried to melt chocolate in a bain-marie might already know
that if the water in the bain-marie is literally boiling, chocolate will not melt but
crumble.)
Towards the figurative end of the continuum, we found 80 cases where boiling
was metaphorically used, as in (16):
(16) (…) several small boats disappeared in boiling seas (…).
There were 4 clear cases of hyperbole, as in (17), and 13 cases in which metaphor
and hyperbole were combined, as in (18) (where boiling indicates a higher-than-
desired temperature, but is loosely applied to something that is not a liquid):
(17) Bring some more ice, this whisky is boiling hot (…).
(18) This summer is promising to be long and boiling (…).
7. On a relevance-theoretic approach, cases where approximation is understood as excluding

the possibility of a literal interpretation would result from a combination of narrowing and
broadening, providing further evidence for a unitary account; see Section 4 for discussion.
Finally, there were 19 cases that would be traditionally classified as synecdoche, as

in (19) (we will not consider the theoretical analysis of synecdoche here):
(19) You’re changing small things like boiling a kettle (…).
Note also that the metaphorical uses of boiling were quite varied. More specifically,
we found metaphors indicating:
–– anger and emotional frustration, as in (20): 8

(20) The brothers, seemingly stable, are absolutely boiling inside with various
frustrations (…)
–– excessive heat, as in (21):
(21) This summer is promising to be long and boiling (…).
–– tension, as in (22):
(22) Cup final, against anyone Pakistan relations almost at boiling point (…).
–– and finally, movement or appearance (typically of water or clouds), as in (16)
(repeated here for convenience):
(16) (…) several small boats disappeared in boiling seas (…).
These results provide some evidence for our view that there is a continuum of
cases of broadening, and that the degree and direction of broadening are heavily
context-dependent.
Our corpus data highlight two important differences between the Gricean and
relevance-theoretic approaches to broadening. First, Grice retains a sharp distinction
between literal and figurative uses inherited from classical rhetoric, and like many
philosophers of language (e.g. Lewis 1979), he treats loose talk and rough approx-
imations as falling on the literal rather than the figurative side (to be analysed as
involving variations in contextually-determined standards of precision rather than
blatant violation of a maxim of truthfulness). Second, he sees figurative uses such as
metaphor and hyperbole as not contributing to explicit truth-conditional content or
“what is said”, but merely to what is implicitly communicated, or implicated. On this
approach, the speaker of the metaphorical (16) above would have made no assertion,
but merely implicated that several small boats disappeared in seas that resembled
boiling liquid. By contrast, relevance theorists deny that there is a clear theoretical
distinction between literal and figurative uses, and treat the ad hoc concepts derived
via lexical-pragmatic processes (e.g. boiling*, painless*) as contributing directly
to explicit truth-conditional content (explicatures), and indirectly to implicit import
(implicatures), across the whole “literal-figurative” continuum.
8. Cognitive linguists would treat this as a case of conceptual metaphor. On the relation between
cognitive linguistic and relevance-theoretic treatments of metaphor, see Wilson (2011a).
In the light of this, consider the use of painless in (12) (repeated here for con-
venience), or in the simpler (invented) example in (23):
(12) In a discussion of euthanasia: I would want something clean and painless: no
botch-ups. It would be the doctor or no one.
(23) Dentist to patient: The injection will be painless.
We have analysed painless in (12) as conveying either its literal meaning painless
(“with no pain”) or an approximation, painless* (“with almost no pain”). But the
presence of the small amount of pain that would justify classifying painless in (12)
or (23) as an approximation shades off imperceptibly into the amount of pain that
would justify classifying it as a hyperbole, painless**, (“with less pain than expect-
ed or feared”). Grice’s framework predicts that this imperceptible shading off gives
rise to a dramatic difference in processing on either side of the approximation/
hyperbole divide: on the one side, the speaker is making a genuine assertion, albe-
it under reduced standards of precision, whereas on the other side, she is merely
implicating that she wants a death that would not hurt too much or is offering an
injection that will not hurt too much. To our knowledge, there is no experimental
evidence whatsoever of such a dramatic processing difference between different
degrees of broadening. In the relevance-theoretic framework, by contrast, where
both approximation and hyperbole contribute to explicit truth-conditional content
or explicatures (as well as carrying implicatures), this imperceptible shading off
between approximation and hyperbole is both predicted and explained.
4. When narrowing and broadening combine
In Section 2, we have tried to show that lexical narrowing is a much more flexible
and creative process than default-based accounts would suggest; in Section 3, we
have tried to show that lexical broadening covers a continuum of cases ranging from
the mildest approximation at one end to the most poetic metaphors at the other. It
is worth emphasising that we do not see narrowing and broadening themselves as
distinct pragmatic processes with their own developmental trajectories and break-
down patterns. Rather, narrowed interpretations, broadened interpretations and
interpretations that are narrowed in some respects and broadened in others are
simply outcomes of applying a single relevance-oriented comprehension heuristic:
“Follow a path of least effort in looking for implications; test interpretive hypotheses
in order of accessibility, and stop when you have enough implications to satisfy your
expectations of relevance.” 9
9. For a recent outline of the assumptions of relevance theory, see Wilson (2017).
To illustrate, consider how bird would be interpreted in the (invented) examples

(24) and (25):
(24) At Christmas, the bird was delicious.
(25) Concorde had finished its last flight and the big bird came in to land.
Let us assume that bird encodes the concept bird, which denotes members of the
class Aves. We all have a huge variety of encyclopaedic information about different
subsets of birds, which provides access to a huge range of contextual assumptions
that might be used in interpreting (24) or (25). These assumptions will be activat-
ed to different degrees in different discourse contexts, making some contextual
implications more salient (and hence easier to derive) than others. For a hearer
interpreting (24), the encyclopaedic information likely to be most highly activat-
ed, and the implications that are therefore likely to be most salient, will concern
the subset of birds traditionally eaten at Christmas. By assuming that the bird the
speaker has in mind belongs to this subset, the hearer has easy access to a range of
implications (about the meal itself, what happened next, etc.) that might satisfy his
expectations of relevance. Among these implications will be some that a speaker
aiming at relevance must manifestly have intended the hearer to derive; these will
be not only implications but implicatures, and hence part of what is implicitly
communicated by the utterance. Thus, the encoded concept bird is adjusted in the
course of the comprehension process so as to warrant a range of implications and
potential implicatures that make the utterance relevant in the expected way. Here,
the outcome of the adjustment process is the ad hoc concept bird*, which is part
of what is explicitly communicated and denotes the subset of birds traditionally
eaten at Christmas. Since the communicated ad hoc concept bird* is more specific
than the encoded concept bird, the effect is a narrowing of the encoded meaning.
For a hearer interpreting (25), the encyclopaedic information about birds most
likely to be highly activated, and the implications that are therefore likely to be made
most salient by the prior mention of Concorde, will concern the subset of birds that
can fly, their manners of flight, and in particular, the manner of flight appropriate
to a large, slender gliding bird. By assuming that the concept the speaker intends
to communicate denotes a category which includes not only large, slender gliding
birds but other objects that share this manner of flight, the hearer has easy access
to a range of implications about the sight of Concorde coming in to land that might
satisfy his expectations of relevance. Thus, the encoded concept bird is adjusted
in the course of the comprehension process so as to warrant a range of implica-
tions and potential implicatures that make the utterance relevant in the expected
way. In this case, the outcome of the adjustment process is the ad hoc concept
bird**, which is narrower in some respects than the encoded concept bird (since
it includes only birds with the appropriate manner of flight) and broader in others
(since it includes some objects that are not members of the class Aves).
Our proposal is, then, that the interpretation of (24) and (25) crucially involves
a search for enough implications and potential implicatures to satisfy expecta-
tions of relevance, with the encoded concept being broadened or narrowed so as
to warrant these implications. To see what light corpus evidence might shed on this
proposal, we investigated the adjective empty, which we analysed as encoding the
concept empty, an absolute concept denoting the set of items that contain nothing
all. The adjective empty occurs in all the subcorpora of the Bank of English in a total
of 2336 concordances. To make the search manageable, we decided to focus on the
89 relevant examples in the subcorpus Ukephem (which consists of ephemera –
leaflets, adverts, etc.), since we were particularly interested in colloquial/spoken
language oriented samples.
Our hypotheses were, first, that encoded word meanings typically undergo
narrowing or broadening in the course of comprehension, and second, that these
departures from encoded meaning take place in different directions and to differ-
ent degrees, with the departures being governed by the search for relevance. Thus,
we expected to find that the encoded concept empty was consistently adjusted to
denote a more fine-tuned type and degree of emptiness (empty*, empty**, etc.).
The sorts of variations we expected to find were (a) variations in the type of content
that the item is understood to be empty of (e.g. empty of wine, empty of water,
etc.) and (b) variations in the degree to which that content is understood as lacking.
Our aim was to illustrate the great diversity of ways in which one particular adjec-
tive was used, and to show that narrowing and broadening are flexible enough to
present challenges to any default account.
Findings for empty
a. Word meanings are narrowed in different directions and to different extents
Our investigation of empty illustrates all three points discussed above. In all
the utterances we investigated, the lexical meaning was narrowed in different
directions and to different extents. In each case, the encoded concept empty
was adjusted to represent a more fine-tuned kind of emptiness. Compare, for
instance, the following utterances:
(26) Later in the year, when the granaries are empty, families have to return
to the market to buy grain.
(27) But whatever you do, don’t play sport on an empty stomach or after a
heavy meal.
Neither (26) nor (27) involves a strict use of empty. If nothing else, the empty
granaries must at least contain air, and the empty stomach gastric fluids. It
therefore seems plausible to assume that in the first case the communicated
concept is the narrower one empty of grain and in the second the narrower
one empty of recently received food. Such fine-tunings occur repeatedly
throughout our search. In each case, the audience brings to bear different
contextual assumptions in specifying the type of contents to be understood
as lacking.
b. Variation across discourse contexts
It follows from our arguments of the last two sections that the concept com-
municated by use of empty may vary considerably across contexts. The sample
of discourse contexts available for empty in the Bank of English is very diverse.
Unlike with red eyes, where certain contexts (e.g. fatigue, crying, etc.) tend to
recur rather frequently, there is much greater contextual variation in the uses
of empty. With the exception of just a few recurring contexts (empty stomach,
property empty of tenants, bus empty of passengers and a few others) all the
discourse contexts we examined are one-off occurrences. The relatively high
proportion of one-off uses favours a highly context-sensitive inferential ap-
proach to lexical narrowing, such as the one proposed in relevance theory,
rather than a neo-Gricean approach based on an appeal to generalised conver-
sational implicatures, as discussed in Section 2 above.
c. Variation within discourse contexts
Our investigation of the word-set red eyes revealed a potential problem for our
hypothesis about the creativity and context-dependence of word use: there was
a noticeable relative constancy in the direction of narrowing within a given type
of discourse context. Although the degree and direction of narrowing regularly
varies across contexts, within a given context narrowing seems to go in roughly
the same direction, and to roughly the same degree. In all 14 cases of crying and
all 6 cases of fatigue, for example, the shade of red the speaker would be taken
to have conveyed, and its distribution over the surface of the eye, is roughly the
same. Our sample did not reveal even a single case in which internal variation
within a given discourse context could be observed. Our hypothesis at that
point was that the lack of variation related to the sample available in the Bank
of English for red eyes, or the limited range of respects in which eyes can be
plausibly described as red, rather than to some general fact about the behaviour
of narrowing or the behaviour of discourse contexts themselves.
This was confirmed by our search on empty, which revealed at least one
case where significant variation occurs in the direction of narrowing within
the same broader discourse context (a discussion of empty property). Compare
(28) and (29):
(28) (…) opening up to homeless people, the thousands of empty properties
we know they have on their books.
(29) She was eventually housed, but in a completely empty flat.

Although in (28) empty property is understood as conveying empty of ten-
ants, in (29) it would clearly be understood as conveying empty of furni-
ture. We take this as evidence that variation in the direction of narrowing
occurs not only across but also within types of discourse context.
d. Interaction between broadening and narrowing
An interesting issue for lexical-pragmatic theories is how broadening and nar-
rowing interact. As noted above, it seems reasonable to assume that both broad-
ening and narrowing can apply in the interpretation of a single monosemous
word (e.g. bird). According to Wilson and Carston (2006, 2008), for instance,
the metaphorical use of princess in a certain discourse context to convey prin-
cess* (denoting the set of spoiled, indulged people unwilling to undertake
menial chores) involves not only a broadening of the denotation of princess to
include some people who are not princesses, but also a narrowing to include
only that subset of princesses who are spoiled, indulged etc. For frameworks in
which narrowing and broadening are treated as distinct pragmatic processes,
this raises the question of whether they apply sequentially, and if so, in what
order. In the case of empty, it seems that first the encoded concept has to be
narrowed by specifying the relevant type of content, and only then can it be
positioned at an appropriate point on the loose-literal-metaphorical continu-
um. Thus, consider (30):
(30) remember to take with you any empty tablet bottles or containers to
show what has been (…).
Here, the discourse context in theory permits either a literal or an approximate
interpretation. So we might suppose that empty is first narrowed to mean emp-
ty of tablets. Then, if the utterance would have enough implications to satisfy
the audience’s expectations of relevance even with a tablet or two still left in the
bottle, empty will be understood as an approximation; but if relevance enough
is achievable only on the assumption that there are no tablets left at all, empty
will be strictly understood.
Now consider the following examples:
(31) [clues] that your child has been sniffing include: finding empty butane,
aerosol or glue cans (…).
(32) (…) the three of us are sharing a room with pizza remnants, empty wine
bottles and flagging concentration (…)
(33) (…) opening up to homeless people the thousands of empty properties
we know they have on their books.
Again, it might seem plausible to assume that empty is first narrowed to mean,
say, empty of wine, empty of aerosol, etc., and then broadened in contex-
tually appropriate ways. However, given our knowledge of the world, it is hard
to imagine a wine bottle or an aerosol can being completely and utterly empty
of wine or aerosol (assuming it has been used to store wine or aerosol at all).
Typically, even an “empty” wine bottle or aerosol can will show traces of their
original contents, which rules out a strictly literal interpretation. In such cases,
only approximate interpretations seem acceptable.
Example (33) raises just the opposite problem. In normal circumstances,
for a property to be appropriately described as empty in the sense of empty of
tenants, it is imperative that not even a single tenant remains. Whereas the
presence of minute traces of wine or aerosol would generally be inconsequential
enough for (31) or (32) to be regarded as true, or true enough, the presence of
even a single tenant in an otherwise empty property has significant social and
legal consequences; so (33) would be regarded as false and misleading, rather
than “true enough”, if a single tenant remained. This type of context leaves no
room for lexical broadening, and approximate uses of empty in this sense are
generally ruled out.
These examples show that the extent to which the contextually relevant
contents must be present or absent for something to be appropriately de-
scribed as “empty” of it is itself heavily context-dependent. Relevance theory
helps to explain the appropriateness judgements involved: while the presence
of a tablet or two in an otherwise empty bottle would falsify very few of the
implications and potential implicatures on which the relevance of the descrip-
tion “empty tablet bottle” depends, and the description is therefore relevant
enough, the presence of a tenant or two in an otherwise empty property would
falsify most or all of the implications on which the relevance of the descrip-
tion “empty property” depends, making the description irrelevant. These cases
provide some support for the view that lexical narrowing and broadening are
not distinct processes, but merely outcomes of a unitary process of mutually
adjusting explicit content, context and implications/implicatures in order to
satisfy expectations of relevance (see also Sperber and Wilson 1998; Wilson
and Sperber 2002).
To summarise, our 89-line sample of the adjective empty contained:
a. 26 lines (29.2%) in which empty could be interpreted either literally or
approximately,
b. 38 lines (42.7%) in which only a literal interpretation would be plausible,
c. 14 lines (15.7%) in which only an approximate interpretation would be
plausible, and
d. 11 lines (12.4%) where empty is interpreted metaphorically.
e. Metaphorical uses of empty

As with all the other terms that we looked at in the corpus, the metaphorical
uses we found for empty (as either an adjective or a verb) were quite varied:
(34) a. Sugar gives you empty calories.
b. Shelter is hard to find and empty days are spent wandering the streets.
c. [When] a smoker is deprived of a cigarette he or she will feel empty
and restless at first.
d. Otto Ritter, a German archaeologist working in the Empty Quarter.
e. (…) at a price to stock your wardrobe and not empty your pocket.
f. (…) photo The Empty Raincoat CHARLES HANDY.
g. Sit back, empty your head of foolish thoughts and just close your eyes.
h. Try to empty your mind of anxious or guilty thoughts.
i. Abasio turned aside to empty his stomach, noisily and messily.
j. Law has become a target of efforts to empty it of intrinsic meaning.
k. Life without letters from you would be much colder and emptier
than it is.
f. The meaning of empty: absolute or underspecified?
The data briefly presented above raise several theoretical questions about the
encoded meaning of empty. One possible approach would be to treat empty (as
we have done) as meaning “containing nothing at all”, and thus as encoding an
absolute concept that is rarely strictly satisfied in nature. It would then follow
that on almost all occasions of use, some broadening of the denotation of the
encoded concept takes place.
Another possible approach would be to treat empty as encoding an un-
derspecified concept, or “pro-concept”, such as empty of x, where x must be
pragmatically supplied (Sperber and Wilson 1998; Carston 2002; Wilson 2011b,
Wilson 2016). It would then follow that in at least some cases – those where
the pragmatically inferred contents may indeed be entirely lacking (e.g. a class-
room completely empty of pupils) – the under-specified term empty may well
be strictly and literally used. It is a genuine question for lexical semantics how
to choose between these two analyses – if we need to, of course, given that in
the framework of relevance theory, it is quite conceivable that two individuals
with different encoded meanings for empty might converge on the same com-
municated concept via different routes. 10
10. Recanati (2010: Chapter 2) treats empty as polysemous, with two conventional senses: an
absolute sense (empty) meaning “containing nothing at all”, and an approximate sense which
accounts for loose uses and makes it possible to describe items as more or less “empty”. However,
he does not consider the type of cases we have discussed here, where both narrowing and broad-
ening apply to the same item, and it is not clear how they would fit into his framework.
5. The scope of a theory of communication
Many linguists and philosophers of language equate a theory of communication

with a theory of Gricean non-natural meaning (meaningnn), or speaker’s meaning. 11
According to Levinson (2000: 12–13), for instance,
… a theory of communication has as its target the full scope of Grice’s (1957)
non-natural meaning … meaningnn (or something of the sort) draws an outer
boundary on the communicational effects that a theory of communication is re-
sponsible for.
For Grice, a speaker’s meaning is a complex mental state involving a series of nest-
ed intentions. To mean something by a declarative utterance, the speaker must (at
least) intend the addressee (a) to form a certain belief, and (b) to recognise that
the speaker intends him to form that belief. Thus, a theory of speaker’s meaning is
a theory of a certain type of overt intentional communication.
However, as Grice pointed out, if meaning is used in its ordinary-language
sense, not all cases of overt intentional communication are appropriately described
as cases of meaning. For instance, when Mary displays her bandaged leg in response
to Peter’s invitation to play squash, she overtly intends Peter to believe both that her
leg is bandaged and that she cannot play squash; but while it seems appropriate to
describe her as meaning that she cannot play squash, it does not seem appropriate
to describe her as meaning that her leg is bandaged (Grice 1967/1989: 109). For
Grice, this difference was important, and he added a third clause to his definition of
speaker’s meaning designed to exclude the type of case where talk of meaning seems
inappropriate. According to this definition, for a speaker to mean something by an
utterance (or other communicative act), she must not only (a) intend the addressee
to form a certain belief, and (b) intend him to recognise that she intends him to
form that belief, but must also (c) intend him to form that belief at least partly on
the basis of his recognition that she intends him to form it. Clause (c) guarantees
that when Mary shows Peter her bandaged leg in response to his invitation to play
squash, she does not mean that her leg is bandaged (since Peter can see this for
himself, irrespective of her intentions in the matter), but she does mean that she
cannot play squash (since the bandage may be deceptive and at least part of Peter’s
reason for believing that she cannot play squash must be his recognition that she
intends him to form that belief).
While Sperber and Wilson (1986/1995: 53–4) share Grice’s intuition that use of
the ordinary-language term meaning is inappropriate in certain cases, they argue
11. Grice’s term was utterer’s meaning, which was intended to cover acts of both verbal and
non-verbal communication; we will use speaker’s meaning here in a similar way.
that the resulting definition of speaker’s meaning does not pick out a natural class of
phenomena, since the cases Grice wants to exclude fall under exactly the same gen-
eralisations as those he wants to include. Instead, they propose a broader definition
of overt intentional (ostensive) communication, which covers many instances of
“showing” that Grice wanted to exclude, and subsumes prototypical Gricean speak-
er’s meanings as a special case. This broader theory applies not only to non-verbal
communication but to many verbal cases where “showing” and “telling” combine,
or where the intended effect is not easily rendered as a single proposition or small
set of propositions that the speaker specifically intends to convey. Thus, the show-
ing of photographs, the use of affective intonation and other expressive elements,
the use of “figurative” utterances and the creation of attitudinal, stylistic or poetic
effects are seen as falling squarely within the scope of a theory of ostensive com-
munication, whereas Davidson’s framework entirely excludes them and in Grice’s
framework they receive only partial treatment at best. 12
As Sperber and Wilson (2015) point out, Grice’s definition of speaker’s meaning
conflicts with intuitions about the appropriate use of the ordinary-language term
meaning in several types of case. To take just one, consider (35) and (36), where the
speaker simultaneously performs an action and comments on it:
(35) TV cook (pouring sauce over a dish): Now I’m pouring the sauce.
(36) Mary (opening a bottle of champagne): This is how you can open a bottle of
champagne.
In (35), the audience can see for themselves that the cook is pouring the sauce,
and in (36) they can see for themselves that this is how they can open a bottle of
champagne, irrespective of the speaker’s intentions in the matter. It follows from
Grice’s definition that in (35), the cook does not mean that she’s pouring the sauce,
and in (36), Mary does not mean that this is how the audience can open a bottle of
champagne, which seems absurd. 13 Now compare (36) with (37):
(37) Mary (opening a bottle of champagne): This is how you should open a bottle of
champagne.
12. See Sperber and Wilson (1986/1995: Chapter 1). For insightful discussion of the issues raised
in this paragraph, see Wharton (2003, 2009).
13. Intuitively, what is going on in these two cases is that the cook both means and shows that
she is pouring the sauce, and that Mary both means and shows that this is how the audience can
open a bottle of champagne. For Grice, though, meaning that P and showing that P (in the sense
being discussed here) are mutually exclusive, so a speaker cannot simultaneously mean and show
the same thing.
It follows from Grice’s definition that (37) is a genuine case of meaning, since the
audience cannot see for themselves, just by watching Mary open a bottle, that this is
how they should open a bottle of champagne. But a definition of speaker’s meaning
which applies to (37) but excludes (36) seems implausible at best.
Grice (1967/1989: 39–40) acknowledges in passing that what is implicitly com-
municated may exhibit a degree of indeterminacy or open-endedness, and he takes
for granted that metaphors fall within the scope of a theory of speaker’s meaning
despite the element of indeterminacy involved. Lepore and Stone (2010) follow
Davidson (1978) in taking a much more restrictive view.
Metaphorical meaning … requires an audience to recognise the specific content a
speaker wants to get across, and to use the signal of the metaphor as a basis for the
uptake of that content. Since we deny this must happen in normal confrontations
with metaphors, we therefore reject metaphorical meaning.
(Lepore and Stone 2010: 44)
Indeed, Lepore and Stone (2015: 5) argue that the effects not only of metaphor but
of irony, humour and various other creative uses of language are too heterogene-
ous and open-ended to be publicly shared, and therefore cannot be dealt with in a
standard pragmatic theory:
The effects we consider cover a wide range of figurative and evocative language,
including metaphor, sarcasm, irony, humor, and hinting. Many theorists have
suggested that general interpretive principles fully explain these creative uses of
language. However, we argue that such utterances are better characterized as in-
vitations to the audience to follow a specific direction of thought in exploring
the contributions of the utterance … The thinking involved is heterogeneous and
diverse – it’s not just a circumscribed or uniform application of principles of ra-
tionality. And so, the insights interlocutors get by pursuing and appreciating this
thinking also fall outside the scope of pragmatics as traditionally conceived.
(Lepore and Stone 2015: 5)
But if metaphors fall outside the scope of pragmatic theories, how do speaker and
hearer co-ordinate on their interpretation?
Consider (38):
(38) Because of Waco and the Branch Davidians, the Federal government was
walking on eggshells to not repeat that experience.
(Interview with Jeff Nicholls, April 2016)
Grice saw the speaker of (38) as blatantly violating the first Quality maxim in order
to implicate a related simile (that the Federal government acted as if they were walk-
ing on eggshells), leading to an indefinite range of further potential implicatures
(e.g. that the Federal government acted with extreme caution, delicacy, alertness
to potential dangers, and so on). Relevance theorists see the hearer of (38) as con-
structing an ad hoc concept, walking on eggshells*, which is part of the propo-
sition the speaker is taken to assert and which warrants an indeterminate array of
potential implicatures (that the Federal government acted with extreme caution,
delicacy, alertness to potential dangers, and so on). For both Grice and relevance
theory, then, metaphors contribute to what is implicitly communicated. For Lepore
and Stone (2015: 6), by contrast, metaphors communicate nothing; and indeed,
since implicatures typically exhibit an element of indeterminacy, they reject the
notion of conversational implicature itself. How, then, do they account for the fact
that the speaker and hearer generally achieve a satisfactory degree of co-ordination
in interpreting a metaphor such as (38)?
According to Lepore and Stone (2015: 171), metaphor involves a special type of
thinking – “metaphorical thinking” – which is analogical in nature. As they put it, the
appeal to “metaphorical thinking” seems to offer “just as good an explanation … as
would some kind of implicated meaning”. More generally, they suggest that each of
the creative uses of language they discuss involves a distinctive kind of “imaginative
engagement” which may lead to similar effects in similar audiences, but not as the
result of any act of overt intentional communication (Lepore and Stone 2015: 173).
In the light of this proposal, consider (39) (discussed in Sperber and Wilson
2015: 121):
(39) I could kill for a glass of water.
In normal circumstances, none of the italicised words in (39) would be strictly and
literally used: kill is a hyperbole and the encoded concept kill would be broadened
in the course of the comprehension process; glass is a loose use and the encoded
concept glass might be broadened to denote cups and plastic containers as well
as glasses, while with water, the encoded concept water would typically be nar-
rowed to denote water that is cold and pure enough to be drinkable. As our corpus
analyses in Sections 2–4 have shown, broadening and narrowing are flexible, cre-
ative, highly context-dependent processes even in apparently prosaic expressions
such as red eyes. For Lepore and Stone, lexical narrowing and broadening would
therefore have to lie outside the scope of pragmatics, and the only way to account
for them would be to appeal to a proliferating range of distinctive “types of think-
ing” – “approximate”, “hyperbolic”, “enriching” and so on. But our corpus analyses
have also shown that there is a continuum of cases of broadening, with no clear
cut-off point between approximation, hyperbole and metaphor, and this presents
a challenge to the claim that each involves a distinct “type of thinking”. We believe
that the unitary relevance-theoretic approach sheds more light on the continuum
of cases of broadening than any appeal to distinct lexical-pragmatic processes or
distinct “types of thinking”.
In this section, we have tried to show that a theory of communication should

aim to explain a much wider range of phenomena than are typically discussed in
linguistics and philosophy of language. Relevance theorists have assumed from the
start that what speakers intend to communicate should be thought of not as a single
proposition or a small set of propositions, but as an array of propositions, which
may be characterised in several different ways. At one extreme, this array might
consist of a single proposition or a small set of propositions that can be individually
enumerated; typically, one of these propositions would be explicitly communicated
and the others implicitly communicated. This is the only type of case normally
considered in pragmatics. At the other extreme, typified by showing a photograph
or pointing to the view, the array might consist of a huge range of propositions
that may be characterised in terms of a description rather than by listing its mem-
bers; in such non-verbal cases, the whole array would be implicitly communicated;
and there is a continuum of cases between the two extremes. Relevance theorists
have also assumed from the start that the speaker’s goal is not to induce a specific
belief in the hearer, but to make an array of propositions more manifest to him. A
proposition is manifest to an individual to the extent that he is likely to entertain
it and accept it as true. When a huge array of propositions is made manifest to an
individual, they will be manifest to different degrees, and contribute to relevance in
different ways. Thus, in interpreting (38) or (39), different individuals may enter-
tain and accept different members of the same array of propositions, and this may
be just the degree of co-ordination needed for successful communication to take
place (for further discussion, see Sperber and Wilson 1986/1995: 50–60, 217–224;
Sperber and Wilson 2015).
6. Concluding remarks
Our corpus studies (limited though they are) provide some support for a unitary
account of lexical-pragmatic processes. They confirm that lexical narrowing and
broadening are highly flexible and context-dependent processes which can combine
in the interpretation of a single word, and support the view that there is a continuum
of cases between literal, approximate, hyperbolic and metaphorical use (cf. Huang
(this volume) and Dynel (this volume)). In Section 2, we have tried to show that
Levinson’s neo-Gricean approach to stereotypical narrowing is not obviously more
cost-effective than the unitary relevance-theoretic approach, and offers no clear
explanation for more flexible, context-sensitive cases at all. In Section 3, we have
tried to show that the standard Gricean framework (which, incidentally, offers no
treatment of lexical narrowing) predicts a dramatic processing difference between
a metaphorical or hyperbolic interpretation of a certain expression (e.g. painless)
and a literal or approximate interpretation, whereas the unitary relevance-theoretic

account predicts an imperceptible shading off between approximate and hyperbolic
interpretations, and this prediction is borne out by our data. In Section 4, we have
argued that narrowing and broadening frequently combine in the interpretation
of a single word (e.g. empty), and that some general account of the contextual
factors that trigger broadening, narrowing or both broadening and narrowing is
needed; and in Section 5, we have tried to show that there is no good reason to
exclude creative uses of language from the scope of a theory of overt intentional
communication.
All this suggests that the goal of an adequate pragmatic theory should be
to provide a unitary account of the full range of lexical-pragmatic processes.
However, largely as a result of historical accident (perhaps combined with differ-
ences in intellectual taste), the only explicit attempts so far at developing such an
account have been made within the relevance-theoretic framework. Neo-Griceans
working on lexical narrowing have shown little interest in extending their account
to cover metaphor or hyperbole; philosophers and literary scholars working on
metaphor and hyperbole have shown no interest in extending their account to
approximation or narrowing, and so on, and semanticists and logicians working
on approximation have shown little interest in metaphor or hyperbole. Our claim
is not that relevance theory offers the only possible unitary account: the challenge
is to propose a better one.
Acknowledgements
We are grateful to Marta Dynel and Piotr Cap for helpful comments on an earlier version.
References
Blutner, Reinhard. 1998. “Lexical Pragmatics.” Journal of Semantics 15: 115–162.

doi: 10.1093/jos/15.2.115
Blutner, Reinhard. 2004. “Pragmatics and the Lexicon.” In The Handbook of Pragmatics, ed. by
Laurence Horn and Gregory Ward, 488–514. Oxford: Blackwell
Carston, Robyn. 1997. “Enrichment and Loosening: Complementary Processes in Deriving the
Proposition Expressed?” Linguistische Berichte 8: 103–127.
Carston, Robyn, and Catherine Wearing. 2011. “Metaphor, Hyperbole and Simile: A Pragmatic
Approach.” Language and Cognition 3: 283–312. doi: 10.1515/langcog.2011.010
Davidson, Donald. 1978. “What Metaphors Mean.” Critical Inquiry 5: 31–47. doi: 10.1086/447971
Fauconnier, Gilles, and Mark Turner. 2002: The Way We Think: Conceptual Blending and the
Mind’s Hidden Complexities. New York: Basic Books.
Geurts, Bart. 2009. “Scalar Implicature and Local Pragmatics.” Mind and Language 24: 51–79.
doi: 10.1111/j.1468-0017.2008.01353.x
Glucksberg, Sam. 2001. Understanding Figurative Language: From Metaphors to Idioms. Oxford:
Oxford University Press. doi: 10.1093/acprof:oso/9780195111095.001.0001
Glucksberg, Sam. 2003. “The Psycholinguistics of Metaphor.” Trends in Cognitive Science 7: 92–96.
doi: 10.1016/S1364-6613(02)00040-2
Grice, H. Paul. 1957. “Meaning.” Philosophical Review 66: 377–388. doi: 10.2307/2182440
Grice, H. Paul. 1967. “Logic and Conversation” (William James Lectures), Harvard University.
Grice, H. Paul. 1989. Studies in the Way of Words. Cambridge, MA: Harvard University Press.
Horn, Laurence. 2004. “Implicature.” In The Handbook of Pragmatics, ed. by Laurence Horn and
Gregory Ward, 3–28. Oxford: Blackwell.
Horn, Laurence. 2012. “Histoire d’*O: Lexical Pragmatics and the Geometry of Opposition.” In
New Perspectives on the Square of Opposition, ed. by J.-Y. Béziau and G. Payette, 393–426.
Bern: Peter Lang.
Jaszczolt, Kasia. 2014. “Defaults in Semantics and Pragmatics.” Stanford Encyclopedia of
Philosophy (Spring 2014 Edition). http://plato.stanford.edu/archives/spr2014/entries/
defaults-semantics-pragmatics/
Kolaiti, Patricia, and Deirdre Wilson. 2014. “Corpus Analysis and Lexical Pragmatics: An
Overview.” International Review of Pragmatics 6: 211–239. doi: 10.1163/18773109-00602002
Lepore, Ernie, and Matthew Stone. 2010. “Against Metaphorical Meaning.” Topoi 29: 165–180.
doi: 10.1007/s11245-009-9076-1
Lascarides, Alex, and Ann Copestake. 1998. “Pragmatics and Word Meaning.” Journal of
Linguistics 34: 55–105. doi: 10.1017/S0022226798007087
Lasersohn, Peter. 1999. “Pragmatic Halos.” Language 75: 522–551. doi: 10.2307/417059
Lewis, David. 1979. “Scorekeeping in a Language Game.” Journal of Philosophical Logic 8: 339–359.
doi: 10.1007/BF00258436
Levinson, Stephen. 1983. Pragmatics. Cambridge: Cambridge University Press.
Levinson, Stephen. 2000. Presumptive Meanings: The Theory of Generalized Conversational
Implicature. Cambridge: Cambridge University Press
Noveck, Ira, and Dan Sperber. 2007. “The Why and How of Experimental Pragmatics.” In
Pragmatics, ed. by Noel Burton-Roberts, 184–212. London: Palgrave.
Recanati, François. 2010. Truth-Conditional Pragmatics. Oxford: Oxford University Press.
doi: 10.1093/acprof:oso/9780199226993.001.0001
Sperber, Dan, and Deirdre Wilson. 1985. “Loose Talk.” Proceedings of the Aristotelian Society
86: 153–171.
Sperber, Dan, and Deirdre Wilson. 1986/1995. Relevance: Communication and Cognition. Oxford:
Blackwell.
Sperber, Dan, and Deirdre Wilson. 1998. “The Mapping between the Mental and the Public
Lexicon.” In Language and Thought, ed. by P. Carruthers and J. Boucher, 184–200. Cambridge:
Cambridge University Press. doi: 10.1017/CBO9780511597909.012
Sperber, Dan, and Deirdre Wilson. 2002. “Pragmatics, Modularity and Mindreading.” Mind and
Language 17: 3–23. doi: 10.1111/1468-0017.00186
Sperber, Dan, and Deirdre Wilson. 2008. “A Deflationary Account of Metaphors.” In The
Cambridge Handbook of Metaphor in Language and Thought, ed. by Raymond Gibbs, 84–105.
Cambridge: Cambridge University Press. doi: 10.1017/CBO9780511816802.007
Sperber, Dan, and Deirdre Wilson. 2015. “Beyond speaker’s meaning.” Croatian Journal of
Philosophy vol. XV, no. 44: 117–149.
Steen, Gerard. 2007. Finding Metaphor in Grammar and Usage: A Methodological Analysis of
Theory and Research. Amsterdam: John Benjamins. doi: 10.1075/celcr.10
Steen, Gerard, Aletta Dorst, Berenike Herrman, Anna J. Kaal, Tina Krennmayr, and Trijntje
Pasma. 2010. A Method for Linguistic Metaphor Identification. Amsterdam: John Benjamins.
doi: 10.1075/celcr.14
Vega Moreno, Rosa. 2007. Creativity and Convention: The Pragmatics of Everyday Figurative
Speech. Amsterdam: John Benjamins. doi: 10.1075/pbns.156
Wharton, Tim. 2003. “Natural Pragmatics and Natural Codes.” Mind and Language 18: 447–477.
doi: 10.1111/1468-0017.00237
Wharton, Tim. 2009. Pragmatics and Non-Verbal Communication. Cambridge: Cambridge
Wilson, Deirdre. 2011a. “Parallels and Differences in the Treatment of Metaphor in Relevance
Theory and Cognitive Linguistics.” Intercultural Pragmatics 8: 177–196.
doi: 10.1515/iprg.2011.009
Wilson, Deirdre. 2011b. “The Conceptual-Procedural Distinction: Past, Present and Future.” In
Procedural Meaning: Problems and Perspectives, ed. by Victoria Escandell-Vidal, Manuel
Leonetti and A. Ahern, 66–84. Leiden: Brill.
Wilson, Deirdre. 2016. “Reassessing the Conceptual–Procedural Distinction.” Lingua 175–176:
5–19. doi: 10.1016/j.lingua.2015.12.005
Wilson, Deirdre. 2017. “Relevance Theory.” In Oxford Handbook of Pragmatics, ed. by Yan Huang,
79-100.Oxford: Oxford University Press.
Wilson, Deirdre, and Robyn Carston. 2006. “Metaphor, Relevance and the ‘Emergent Property’
Issue.” Mind and Language 21: 404–433. doi: 10.1111/j.1468-0017.2006.00284.x
Wilson, Deirdre, and Robyn Carston. 2007. “A Unitary Approach to Lexical Pragmatics:
Relevance, Inference and ad hoc Concepts.” In Pragmatics, ed. by Noel Burton-Roberts,
230–259. London: Palgrave. doi: 10.1093/oxfordhb/9780199234769.003.0018
Wilson, Deirdre, and Robyn Carston. 2008. “Metaphor and the ‘Emergent Property’ Problem: A
Relevance-theoretic Treatment.” The Baltic International Yearbook of Cognition, Logic and
Communication 3: 1–40.
Wilson, Deirdre, and Dan Sperber. 2002. “Truthfulness and Relevance.” Mind 111: 583–632.
doi: 10.1093/mind/111.443.583
Wilson, Deirdre, and Dan Sperber. 2012. Meaning and Relevance. Cambridge: Cambridge
Wilson, Deirdre. 2016a (forthcoming). “Relevance Theory.” In Oxford Handbook of Pragmatics,
ed. by Yan Huang. Oxford: Oxford University Press.
Wilson, Deirdre. 2016b. “Reassessing the Conceptual–Procedural Distinction.” Lingua 175-176:
5–19. doi: 10.1016/j.lingua.2015.12.005
Chapter 8
Indirect ritual offence

A study on elusive impoliteness
Dániel Z. Kádár
University of Huddersfield
The present chapter examines the phenomenon of indirect ritual offence, which
includes cases of recurrent offences that are indirect and as such make it dif-
ficult for the targeted person to respond to them. Manifestations of indirect
ritual offence include a series of indirect attacks that recurrently target a person,
and cases in which the targeted person is recurrently ostracised. While this
behaviour has received significant attention is social psychology, it has not been
studied in pragmatics and (im)politeness research, in spite of the fact that it is
one of the most widely discussed forms of impoliteness behaviour in public dis-
courses. The examination of this phenomenon also contributes to the pragmatic
examination of the difference between indirectness and implicitness; as I point
out in this chapter, indirect ritual offence becomes implicit from both academic
and lay points of view if the frequency of such attacks decreases but the attacks
nevertheless continue. In such cases, the targeted person may reinterpret pre-
vious attacks as “harmless” and speculate about the nature of new attacks, and
even more importantly the abuser can easily claim that they have not intended
to offend the other at all – thus, in such cases indirect ritual offence gains an
implicit nature. In order to illustrate this point, I examine ways in which “im-
plicit” as an evaluator tends to be metapragmatically used in accounts on indi-
rect ritual offence.
Keywords: impoliteness and offence, indirectness, implicitness, frequency,

metapragmatics
Nem hívtak meg. Amikor szóba hoztam a dolgot, nem reagáltak semmit, de láttam
a fejükön, hogy magukban röhögnek a szemetek. Aztán újra kihagytak, és ez már
magáért beszélt.
They didn’t invite me. When I mentioned this matter, they didn’t react, but I saw
that they laughed amongst themselves, the shitbags. And then they left me out
again, and it spoke for itself.
(Excerpt from a post event interview in Hungarian, cited from Kádár 2013: 154)
doi 10.1075/pbns.276.08kad
178 Dániel Z. Kádár
1. Introduction
Say I am driving in an orderly way but I upset the driver of the car behind me for some
reason, and I find myself in a road rage incident: the other driver jumps out as I arrive
at a red light and starts yelling at me. I may experience a range of emotions, spanning
fear, through upset, to anger, and I have plenty of ways to react, such as yelling back,
taking a menacing bodily posture, or threatening the other that the police will know
about this incident by taking out my phone and starting to video record him. By all
accounts, I will have many discursive resources (Thornborrow 2002) and practical
means to react, and a clear case to make if I want to take the road raging driver to
court: for example, in the UK, the Public Order Act (1986) prohibits such acts, and
there is a relatively clear legal stance of the consequences of such behaviour:
A recent survey conducted by Zurich Insurance brought some surprising road
rage statistics. Of the 1,161 UK drivers polled, 66% admitted using threatening
behaviour to another road user. Around 77% of those polled also felt that this type
of threatening behaviour was acceptable. But in the eyes of the law this type of
threatening behaviour can lead to the driver being charged under Section 4 and 5
of the Public Order Act 1986. Using threatening behaviour or using verbal insults
and abuse are offences in the UK.
Retrieved from: http://www.nopenaltypoints.co.uk/
road-rage-driving-offences.html)
But let us now imagine that I am an employee of a company in which it is tacit-

ly agreed – without anybody saying this explicitly – that promotion depends on
whether or not I get involved in after-work parties in staff members’ flats where
important information is being exchanged and networking is being done. Let us
suppose that I regard myself as a hardworking person and I do my best to get on
with my colleagues, but I nevertheless feel that my peers dislike me and they inten-
tionally avoid involving me in such events. Even worse, what happens if I realise
that my peers actually hate me, and want to hurt me, by recurrently inviting each
other in public in front of me without inviting me? If I gather the courage, I may
get into one of the parties uninvited, but my colleagues’ facial expressions will be
contemptuous as I “nominate myself ” to get invited, and they may make me feel
that I am unwanted by recurrently not inviting me. If I have an outspoken person-
ality, I may attempt to accuse them of being abusive, as the person in the post-event
interview that opens this chapter did, but it is very likely that in such a case I will
not get a straight response, or will be mocked for “taking things too seriously”, and
the abusive behaviour will continue. I may perceive that what is happening to me
is not only impolite but also immoral, 1 and I may feel anger, pain, and frustration.
1. Although social psychological research (e.g. Glasø et al. 2007) shows that victims of such
behaviour may also accuse themselves.
Chapter 8. Indirect ritual offence 179
However, I will have very limited opportunities to defend myself: the law does not
stipulate that people have to be invited to informal parties, nor does it say anything
about the contemptuous facial expressions of my colleagues.
These brief anecdotes show the power of a form of interpersonal behaviour,
which I define as indirect ritual offence 2 (see Kádár 2013), that is a form of recurrent
in-group behaviour, in the case of which other group insiders either intentionally
neglect, or even make a series of subtle attacks on a stigmatised individual (see
Goffman 1986). 3 For example, indirect ritual offence can include cases in which
other group members recurrently refuse to respond to the stigmatised person’s
e-mails, do not allow him/her to be involved in workplace activities in which this
person is entitled to get involved, and so on, or make a series of seemingly harm-
less jokes. Thus, in terms of indirectness, this form of behaviour has two – more
and less indirect – types, including cases (a) in which the attackers make a series
of subtle attacks on a targeted person, and (b) in which they recurrently fail to
involve the targeted person in an activity in which they would be entitled to get
involved. However, the nature of these behaviour types coincide in that neither of
them provide any explicit cues, such as impoliteness formulae (see Culpeper 2010),
by means of which the victim can undoubtedly claim that (s)he has been subjected
to abuse. On the other hand, both of these behavioural forms are meant to make
the individual realise that what is happening is certainly not a coincidence: in this
respect, they have to be clearly distinguished from unintentional offence (Culpeper
2011), e.g. cases in which someone is neglected by accident but no intended attack
on the person takes place. The aim of indirect ritual offence is to hurt the victim
as much as possible, while minimising the individual’s chance to defend him or
herself. Studying this issue thus provides insight into a key aspect of indirectness
(cf. Kecskes, this volume), namely, its interrelationship with impoliteness and in-
terpersonal abuse.
I define such indirect offences as ritual due to two interrelated reasons. First,
from a technical point of view, they fit into my etic definition (Kádár 2016) of ritual
behaviour, which is the following:
Ritual is an action with formalised features that embodies a practice, and which is
relationship forcing, i.e. by operating, it reinforces/transforms interpersonal rela-
tionships. Ritual is realised as an embedded liminal (mini-)performance, and this
performance is bound to relational history (and related moral order), or historicity
in general (and related moral order).
2. In the present chapter I do not distinguish between “rudeness”, “impoliteness”, and “offence”,
hence following Culpeper’s (2011) uptake.
3. It is worth noting that the English language has the idiom “to send someone to Coventry” to
popularly describe this phenomenon; I am grateful to Liz Marsden for drawing my attention to this.
If we take the above-discussed example of indirect in-group offence, it is clear that

it operates with formalised features: in order to make the victim feel that (s)he is
undesired, those who stigmatise him/her need to provide appropriate linguistic
and non-linguistic cues, such as salient shifts from personal to professional top-
ics once his or her presence is acknowledged in the interaction, a colder tone of
voice when others talk to him or her, and so on. It is partly due to the valenced
nature of these cues (see Kádár and Haugh 2013) that the stigmatised individual
will ultimately realise that what is happening is intentional, and partly that these
actions embody a practice as they recur. 4 Thus, the raison d’être of such ritual
offences is to make it impossible for the targeted individual to reinterpret them
as unintentional (which is a typical feature of humiliating behaviour in general,
as social psychologists such as Silver et al. 1986 argue). Also, these formalised
features are supposed not to be associated with impoliteness in a direct sense.
This form of behaviour is also relationship forcing in that it inherently deterio-
rates the relationship between the attacker(s) and the targeted person although
the individual may intend to integrate into the group, as far as the others uphold
excluding him or her by means of this recurrent practice of attacks, (s)he will
have little chance of regaining a fully ratified (non-stigmatised) participation sta-
tus within the community (Goffman 1981). This form of in-group behaviour is
also a liminal performance, as is the case with any ritual; in the anthropological
literature (see van Gennep 1909[1960]), “liminal” – which comes from the Latin
limen “threshold” – means that participants of a ritual action cross a certain border
between what is perceived as “ordinary” and “extraordinary” behaviour, and this
act of crossing benchmarks a potential relational turning point (see Hopper and
Drummond 1990). I use “liminal” in a broader sense than anthropologists, who
often apply this notion to capture ritual events of celebration (see e.g. Metcalf and
Huntington 1991); in my ritual framework (Kádár 2016), “liminality” describes
ritual as a relational target-oriented action, which departs from what is regarded
as the “ordinary flow” of interaction as “salient” from the participants’ point of
view. In the case of the example discussed here, the indirect offence sets into op-
eration as other participants theatrically ignore the stigmatised individual. These
“ordeals” depart from the ordinary flow of events: they are saliently different from
target-oriented workplace interaction, and they trigger negative emotions and the
perception that something immoral or “unfair” is happening to the individual.
4. Note that not every ritual practice recurs within a particular interpersonal relationship, as
there are one-off rituals, such as rites of passage, which embody practices in a broader social sense.
However, the raison-d’être of implicit ritual offence is to recur within a particular interpersonal
relationship, and so it is inherently recurrent.
This brings us to the second reason why such forms of behaviour can and
should be defined as ritual. It can be argued that ritual is an action by means of
which the moral order of a community or broader society is maintained (Whutnow
1989). In the case of the above example, there are two moral orders involved. On
the one hand, the individual may perceive that what is happening is unfair (and it is
also regarded as unfair by outsiders of the community) and may metapragmatically
voice this perception (see Kádár and Marquez Reiter 2015). On the other hand,
from the group’s perspective, stigmatising the individual may actually be proper, as
it is very likely that a person becomes targeted because their persona or behaviour
contradicts what is regarded as acceptable/desired within the given community.
Thus, while those who perform the abusive behaviour may be aware that in other
circumstances and by group outsiders, their behaviour may be regarded as unac-
ceptable, they may regard their punitive ostracising behaviour as appropriate, as it
reanimates the ethos of their group and reinstates its moral order.
2. Indirect ritual offence
The phenomenon of indirect ritual offence is important, in my view, due to two

reasons.
2.1 Contribution to research on implicitness
First, the notion of indirect ritual offence contributes to research on implicitness,

which is the key topic of the present volume, from the perspective of (im)polite-
ness (in particular: impoliteness) research. Implicitness has been in the focus of
(im)politeness studies, to some extent, since Brown and Levinson’s (1987) seminal
study. While it is beyond the scope of the present chapter to provide a detailed
overview of the history of research on this theme, it is worth pointing out here a
number of studies that can be regarded as noteworthy contributions to research on
implicitness and (im)politeness:
–– Spencer-Oatey’s (2005) ground-breaking research, which has a strong element

of intercultural communication, has incorporated “implicitness v. explicitness”
as a (cultural) variable according to which people evaluate (im)politeness
behaviour;
–– Holtgraves (2005) examined implicitness and performatives, by focusing on
speech acts that do not contain performative verbs – this research provided
insight into the power of implicitness in a range of (im)politeness-related be-
havioural forms, such as blaming;
–– In a somewhat similar way, Limberg’s (2009) study convincingly pointed out

that when it comes to certain phenomena such as threats, which are closely
associated with understandings and perceptions of (im)politeness, implicitness
can be a key factor;
–– Lorenzo-Dus et al. (2011) examined the concept of implicitness as a potential
aggravator of (im)polite messages.
Along with such theoretical examinations of the interrelation between implicitness

and (im)politeness, the concept of implicitness has been deployed in a practical way
in cross-cultural and variational pragmatics research on (im)politeness; the perhaps
most representative research in this area was provided by Ren (2013).
Yet, in (im)politeness research implicitness has been significantly less impor-
tant than the topic of indirectness, which is not only at the centre of the Brown and
Levinsonian (1987) framework and speech act-based approaches to (im)politeness
(see e.g. Blum-Kulka 1987; Byon 2006), but also it plays an important role in more
recent, interaction-based explorations of (im)politeness behaviour (see e.g. Grainger
2011). What has remained somewhat unclear is the relationship between implic-
itness and indirectness; as Dynel (2009: 25) points out, quite convincingly to me:
in literature on (im)politeness the term “indirectness” is used in reference to/in-
stead of “implicitness”, the latter appearing to be a more appropriate term, insofar as
the theories are anchored in the Gricean account. Indirectness is one manifestation
of implicitness […]
I believe that a fundamental advantage of studying indirect ritual offence is that

the examination of this phenomenon reveals that implicitness and indirectness
should be rigorously distinguished in the field. There is no doubt that the offensive
behaviour studied is indirect to some extent, as it does not include swearing or other
lexemes or forms of behaviour conventionally associated with impoliteness; it fits
into Culpeper’s (2011: 194) argument on impoliteness, according to which – unlike
what was claimed in previous research such as Leech (1983) – it cannot be taken
for granted that the more indirect an utterance is, the less offensive it becomes. As
Culpeper notes,
Generally, with respect to directness, it is departures from the convention, whether
through using directness or non-conventional indirectness, that led to higher eval-
uations of impoliteness. Regarding low-power speakers commanding high-power
addressees, we found that the differences between the different degrees of direct-
ness with respect to impoliteness judgments were relatively small. We suggested
that the mere fact of commanding someone of relatively high power in a context
where they clearly have no special right to do so is enough to lead to the evaluation
of strong impoliteness, and at these higher levels of impoliteness the finer linguistic
differences amongst our command items gets lost in the ‘white noise’ of offence. We
suggested that the fact that the conventionally indirect ‘could you be quiet’ uttered
in this context does not exacerbate the impoliteness more than the other direct-
ness categories may be because of possible leakage from the conventional polite
meaning associated with ‘could you X’ structures and the fact that the mismatch
between the conventional meaning and the context is not strong.
(Culpeper 2011: 194)
There is no doubt that the ritual phenomenon studied is indirect on the formal
level (and, indeed, “indirectness” tends to describe the productional format of an
utterance in the field): it is an effective tool to attack someone exactly because, while
there is a continuous “white noise” of offence, due to its highly indirect nature it
does not provide any opportunity for the victim to make a complaint or counter-
attack, without finding him or herself in an even worse position (see Section 4). In
Culpeper’s example above, giving a command in a highly indirect way is a notewor-
thy case, as the person in power who receives such a command may find it difficult
to respond without losing his/her temper and portraying him or herself as rude.
Now, let us try to think of cases in which literally nothing happens, i.e. non-action
is an action (as in the case of the opening anecdote), or if such actions are so subtle
that the targeted person can hardly complain without appearing “unreasonable”; in
such cases, indirectness is the very essence of the operation of impoliteness.
It is worth arguing that not every indirect phenomenon is necessarily implicit
on the level of meaning making. As a ritually offensive action recurs as a practice,
it is clear that its meaning becomes explicit from the recipient’s perspective, in that
recurrence itself conveys the clear message that the participants of this ritual prac-
tice stigmatise the victim. According to Cambridge Dictionaries Online, 5 “implicit”
means “suggested but not communicated directly”; this definition is confirmed by
academic ones – for example, Carston (2009: 36) argues that the main difference
between implicit and explicit meanings is that
the ‘explicit’ one is a pragmatically inferred development of the linguistically en-
coded content while the implicit one(s) are wholly pragmatically inferred.
If we accept this definition, the examination of indirect ritual offence raises a note-
worthy issue: if something recurs e.g. in the form of non-action being an action, it
is clear that the offensive message is communicated directly, unlike in the case of
implicit offences. That is, by default in indirect ritual offence the content is “linguis-
tically encoded”, as Carston (2009) puts it, because even non-action follows linguis-
tic patterns as it ritualistically recurs. Impoliteness is implicit insofar as it provides
5. See: http://dictionary.cambridge.org/dictionary/english/implicit .
retractability of meanings (even though not all implicit meanings are cancellable;
see Jaszczolt 2009): consequently, implicit forms of impoliteness leave the door
open to the abuser to argue that they have not been abusive at all (see Dunham et
al. 2006). As I argue in Section 4.2, in ”ordinary” cases of indirect ritual offence, the
offenders have no intention at all of retracting their offensive meanings.
I intend to make a contribution to the examination of implicitness and indi-
rectness, by demonstrating that indirect ritual offence can become “implicit” in lay
metapragmatic accounts if a certain amount of time elapses between instances of
offence. As indirect ritual offence lacks implicitness only if it does not leave doubt as
to the offender’s meaning, it needs to recur relatively frequently in order to make the
target feel that (s)he is under attack. It is then interesting to examine cases in which
attacks become infrequent and the victim begins to speculate about the potential
harmlessness of the abuser(s)’s behaviour. In such cases the offensive nature of the
offences not only becomes ambiguous but also implicit: the impolite meaning is
suggested but not communicated directly due to the low frequency of the attacks.
As I will point out in Section 4, such forms of abusive behaviour operate as an ef-
fective tool to target a victim exactly due to their implicit nature. Interestingly this
lay understanding, which I will examine by focusing on the way in which people
metapragmatically reflect on such cases of abuse by using the word “implicit”, also
fits into academic definitions of this notion, that is, there is an overlap between lay
and academic understandings of implicitness. These involve decreasing frequency
that increases the above-discussed retractibility of meanings, and consequently the
implicitness of abusive behaviour in an academic sense. Also, with the decreasing
frequency of such attacks, their pattern tends to become less obvious from a par-
ticipant’s point of view, i.e. the “linguistically encoded” nature (Carston 2009) of
the offence weakens.
Note that in the present metapragmatic approach, I examine retrospective ac-
counts on indirect ritual offence in which “implicit” occurs (see Section 4.2). As
such accounts are metadiscursive by nature (see Haugh and Kádár, forthcoming),
i.e. they engage in post-event metapragmatic activity, they reflect an intermix of
etic observer and emic participant perspectives: the narrators themselves were pres-
ent in the abusive events discussed, with some of them being the targets in these
events; however, due to the time lag involved in narratives people tend to distance
themselves from the narrated event, and so these accounts unavoidably gain an etic
element. Thus, the lay understandings of “implicitness” in these narratives are, in
a sense, etic by nature, similar to academic observations.
2.2 An understudied but important phenomenon
Indirect ritual offence is worth studying because it has been understudied in prag-
matics (but see Kádár 2013), while it has retained significant attention in other
fields, in particular, social psychology (see Section 3). Thus, there is an interdisci-
plinary gap, which the present paper attempts to bridge to some extent, even though
it is evident that covering such a theme is beyond the scope of a single paper, and
so further research is due in this area. The importance of this phenomenon might
be illustrated by recent research dedicated to workplace bullying; I cite a fragment
of this research here:
According to a 2007 Zogby / Workplace Bullying Institute poll:

–– 37% of American workers have been bullied at work
–– 13% are currently being bullied
–– 24% have been bullied in the past
[…]
[Workplace bullying] frequently includes the following:
–– criticism, fault-finding
–– undermining
–– marginalizing, ostracizing
–– isolation, exclusion
–– singling out, different treatment
–– belittling, demeaning, degrading, ridiculing
(Retrieved from: https://www.linkedin.com/pulse/20140922230816-108561273-
workplace-bullying-a-growing-problem-with-invisible-injuries)
Apart from the surprising frequency of bullying, which is also valid to other do-
mains of life such as schools, care homes, the army, and so on, 6 a noteworthy fact is
that in the list above (which I do not cite it in its full length), forms of behaviour that
may come into existence as indirect ritual offence occur in top places (from point
number 2 to 5). “Undermining”, “marginalizing, ostracizing”, “isolation, exclusion”,
and “singling out, different treatment” may manifest themselves in recurrent and
indirect ways, i.e. in the form of indirect ritual offence, and research on workplace
discourse shows that this is often the case (see an overview in Kádár 2013). In sum,
indirect ritual offence is worth academic exploration, as it is an elusive phenomenon
that lurks in many people’s daily lives.
As a related point, due to its elusive nature, indirect ritual offence is of inter-
est to the pragmatician from a metapragmatic point of view – in particular, from
6. See e.g. Sharp and Smilth (2002); Monks et al. (2009).

the perspective of metalexis, i.e. lexical items that describe certain interpersonal
interactional phenomena. In order for indirect ritual offence to emerge, several con-
textual and interactional factors need to operate, as the following figure illustrates:
stigma
in-group indirect
practice behaviour
ostracising
Figure 1. The operation of indirect ritual offence
The lined part in the interface of bubbles illustrates indirect ritual offence.
Due to this elusive characteristic of indirect ritual offence, in various languages
there seem to be an abundance of lexemes to describe it, and metadiscourses show
that people are aware of its existence as a practice. That is, following Haugh’s (2015)
argument, it is part of the emic practices of certain groups, and broader societies
tend to be aware of the social problems that such practices trigger, and due to
this, there are many abstract emic concepts to directly describe these practices.
Table 1 illustrates metalexical emic “equivalents” for this phenomenon in English
and Japanese:
The reason why I cite Japanese language here is that in Japan bullying more fre-
quently emerges in social discourses than in many other cultures, and also it tends
to appear in indirect forms (e.g. Kanetsuna et al. 2006); 7 consequently, in Japanese
there are popular metalexemes for indirect bullying with very specific and narrow
meanings. For example, while the English word “ignore” only refers to indirect abuse
if it occurs in contexts of abusive behaviour, the Japanese mushi-suru exclusively
7. For a noteworthy popular discussion of this theme see: http://www.nippon.com/en/currents/

d00054/
describes indirect abusive behaviour. This does not imply that in English-speaking
cultures indirect ritual abuse is not a “popular” topic: my goal here is simply to
illustrate that in some languages and cultures the phenomenon studied in this
chapter is part of social cognition even beyond its understandings in English and
other European languages/cultures (cf. Verschueren 2000).
Table 1. English and Japanese metalexemes for phenomena that can come into existence
in the form of indirect ritual abuse 8
English Japanese
ignore: to disregard someone (as a group mushi-suru 無視むる: to pretend that
member) someone does not exist
to send someone to Coventry (idiomatic murahachibu ni sareru 村八分にされる:
expression) to implicitly or explicitly ignore someone
(this term originates in historical Japan) 8
exclude: to disallow a person to exercise haijo-suru 排除する: exclude from a group
group membership
ostracise: to exclude from a group or society tsuihō-suru 追放する: to exclude someone
from a group
indirectly bully: collective term for indirect kansetsu-tekina ijime 間接的ないじめ:
abuse indirect bullying
Despite such differences existing in terms of metalexemes across languages and cul-
tures, in both English and Japanese there are only relatively complex descriptors of
an academic nature for indirect ritual offence as a general phenomenon, including
the English “indirectly bully” and the Japanese kanetsu-tekina ijime. 9 This showcas-
es the difficulty of examining indirect ritual offence, namely, that as it takes many
forms, participants of such bullying events themselves tend to refer to particular
practices of this phenomenon. Thus, collective descriptors – which tend to be used,
for example, in the media – unavoidably have an abstract etic (and potentially ac-
ademic) nature. Also, as the metalexemes shown in Table 1 often occur in popular
8. Although the examination of this phenomenon is beyond the scope of the present paper,
examining abusive phenomena by looking into modern metalexemes of historical origin, as well
as historical discourses on such behaviour, is a noteworthy theme. See a discussion on this issue
in Kádár (2013).
9. Note that in this sense it is worth distinguishing popular media from individual narratives
of events of abuse studied in this paper: while such narratives tend to influenced by media, and
so popular academic metalexemes occur in them, individuals prefer, in my experience, non-
academic metalexemes to describe their life experiences. For example, though English language
blogs on indirect ritual offence do, occasionally, use the metalexeme “ostracise”, it is perhaps not
a metalexeme that many individual language users would prefer.
discourses, they tend to be used with different meanings by various authors and
media outlets. 10 In addition, certain types of this behaviour can only be described
rather than expressed – although descriptors may include metaphors, which have
strong expressive value – supposedly in any language: in particular, there are no sin-
gle metalexemes to describe active forms of indirect ritual offence, i.e. cases when
members of a group make a series of subtle attacks on the individual; in English,
the closest metalexeme is perhaps “ostracise” and in Japanese tsuihō-suru, but none
of these lexemes mean exactly this phenomenon.
The definitional complexity of indirect ritual offence does not cause a problem
in various fields, such as social psychology, in which researchers pursue particular
interests in individual subtypes of this phenomenon, such as stigmatisation and
ostracisation (see e.g. Smart Richman and Leary 2009; Twyman et al. 2010), iso-
lation and intentional exclusion (see e.g. Olweus 1994), and in different contexts
such as schools (e.g. Sharp and Smith 2002), workplaces (see e.g. Rayner 1997),
and disadvantaged groups (e.g. Curtner-Smith et al. 2006). While the notion of
directness v. indirectness has been addressed by some social psychologists (see
Garandeau and Cilessen 2006; Goethem et al. 2010), to the best of my knowledge
no comprehensive research has been dedicated to the way in which it works in in-
teraction. This is why pragmatics-based research on indirect ritual offence has the
potential to fill a knowledge gap: pragmatics, due to its focus on modus operandi,
allows us to examine this phenomenon beyond individual manifestations or context
types of this phenomenon.
3. Data
The data studied in this chapter consist mainly of anecdotes and post-event ac-
counts of in-group ritual abuse, as well as post-event discussions of these ritual
practices, rather than authentic recordings. This is due to the hidden nature of
indirect ritual offence, and abusive ritual practices in general: usually such ritual
practices are inaccessible to the etic observer at the place and time of their occur-
rence. Unlike, for example, a corporate ritual of welcoming, there would be very
few people who would allow any group outsider ethnographers to join the group’s
practice, even on a temporary basis. This is due to the shame value attached to these
practices in normative public moralising discourse: as such emic ritual practices
contradict the public’s etic moral order, those who perform them prefer secrecy (ir-
respective of the fact that they may evaluate their ritual positively, see Kádár 2017).
10. See an overview of this issue in Haugh and Kádár (forthcoming).

Thus, apart from analysing films, such as Bully (2001), which display interactions
that are ritualistic, and reality shows and “roasting” shows (i.e. when well-known
people are invited to ridicule a celebrity), it is difficult to record spoken data that
represents abusive ritual practices, and this is even more the case when it comes
to indirect ritual offence. The situation is somewhat similar in the case of written
interaction even though there are open threads that represent abusive ritual in com-
munities: 11 unfortunately, it is more often than not difficult to access community
websites which belong to closed communities with a relational history, and also
reference materials on cyberbullying rarely cite abusive threads (see e.g. Kowalski
et al. 2008). The present research relies on a database of 81 anecdotal descriptions
of indirect rituals, as well as a set of 19 short (and slightly edited) interviews which
I conducted in Hungary.
4. Analysis
In the present section, I first illustrate the operation of indirect ritual offence
(Section 4.1), by overviewing the various types of this behaviour. Following this
(Section 4.2), I examine cases in which indirect ritual offence becomes implicit by
nature, by focusing on the criterion of time lapse between offences, and the oper-
ation/interpersonal effect of such offences.
4.1 Indirect ritual offence in operation
As mentioned at the beginning of this chapter, indirect ritual offence has two types,
including cases in which the targeted person is subjected to recurrent gentle stabs,
and others in which seemingly no action is taken, but as the occurrence of such
actions is part of the situated moral order, non-action becomes action.
It is pertinent to note that there is no clear borderline between these types of
behaviour, as the following Example (1) illustrates:
(1) I am an Occupational therapist in the NHS [the National Health Service of
the UK] and I love my work. I am conscientious and dedicated with expertise
in Brain injury. I have been qualified for 18 years and live for my work. Over
the last 5 years I have been subjected to consistant bullying where a group of
11. The most typical ones are chatrooms of online game websites. Importantly, online virtual
identities are real ones for netizens (see e.g. Thomas 2007), and so this multimodal online data
type (cf. e.g. emoticons) seems to be suitable to represent the destructive effect of ritual practices
in interaction.
managers have talked about me in their lunchbreak, joked about ‘smashing my

face in’, made a secret agreement not to answer my questions in regard to a new
computer system, making gestures behind my back and many many more nasty
things [my emphasis]. Eventually, I walked out of the building and was on sick
leave for 3 months. A meeting was arranged with HR but my angry manager
used it to critise me stating things like ‘look at the state of you, bursting out of
your uniform. No wonder people don’t respect you’. I was advised to complete
bullying forms, then one of the bullies filled in a bullying form against me
saying she felt harrassed that I had accused her of bullying. I was subjected to
3 hours of questions. 5 months later I have heard nothing from HR and the
bullying continues. I have gained 2 stone in weight in 5 months and had a chest
infection for 3 months. I despair. My wonderful career has died. I put a brave
face on for the patients and nurses on the ward. I am considering sick leave
again and seeking legal advice.
(Retrieved from: http://www.nhs.uk/Livewell/workplacehealth/
Pages/bullyingatwork.aspx)
A noteworthy characteristic of this anecdote is that it provides a glimpse into the

social psychology of the behaviour studied. That is, the victim of such abusive be-
haviour becomes stigmatised as a result of an unspoken agreement, as members of
a group perceive the tension between the group’s moral order and the individual’s
behaviour/characteristics. In terms of interactional behaviour, this example illus-
trates that “passive” (non-action as an action) and “active” forms of indirect ritual
abuse can co-occur. For example, recurrently failing to answer the targeted person’s
questions is a seemingly passive form of behaviour, while making gestures behind
the person’s back represents active abuse.
Yet, in various cases of my data these two types of behaviour appear separately
from each other. For instance, the following Example (2) illustrates the operation
of recurrent non-action as a form of abuse in its full power:
(2) Isolated – staff involved would never sit with me during morning tea, lunches,
meetings, courses, etc. My name was omitted from birthday acknowledgements.
All other staff names on whiteboard in staff room and on work trays were in
black, only mine was in red. When we were asked to bring a plate for morning
teas or special lunches, no one ate any of mine. I volunteered to help on many
projects only to find later that the projects had been completed without my help.
(Retrieved from: http://www.sheilafreemanconsulting.biz/case-studies.htm)
A noteworthy characteristic of this behaviour is that, while it generates a high-level

of awareness on the victim’s side, as Example (4) below illustrates, there is actually
very little that the targeted person can do, as attempts to explicitly resist generate
claims that the person is making undue complaints. In the course of collecting data
for the present study, I could find just one case in which the victim of such a form
of behaviour was able to successfully counter the abusive behaviour of the group.
The second type of this ritual practice is more “visible” than the first type, as
it manifests itself in the form of active action, but this visible action is disguised as
harmless, as the following example illustrates:
(3) at a staff night out and after dinner we went to a bar and the Principal said,
‘Come on, Trace, let’s find you a man.’ (Needless to say I am single)
(Retrieved from: http://www.sheilafreemanconsulting.biz/case-studies.htm)
Example (3) is drawn from an Australian school teacher’s description of how she
was bullied in her workplace. The Principal’s utterance could represent “collegial”
teasing workplace humour, or jocular mockery as Haugh (2010) defines it. However,
as it is part of the recurrent exclusion of the target from the group, she perceives the
destructive function, and she contextualises this utterance accordingly. Yet, there is
little that this person could have done, not only because the Principal holds institu-
tional power in an educational institution, but also because this ritual is disguised
as humour and protesting against it would further weaken the victim’s position. As
Bloch et al. (2011) note about the act of abusive humour, jokes which disguise the
hostile and punitive nature of the interaction make it difficult for the scapegoat to
respond without demonstrating his or her “lack” of humour.
The difficulty of resisting indirect ritual offence is illustrated by the following
thread on an internet discussion board, which I studied in Kádár (2013). As this
thread illustrates, it is often the case, at least in the data I studied, that an attempt to
openly resist such forms of abusive behaviour creates an opportunity for further abuse:
(4) Everyone Plzz Stop Ignoring Me!!!!!
1. AIRFLKISAAC (07-30-2011)
Mr.Bean ignores me pasqualina ignores me nmeade ignores me what am
i not “pro” enough for them everytime i say hi they dont say anything and
the next they say hi to some random person again am i not a “pro”mkwii
racer do you guys want me to be 9999 vr or hack the game well first of all
im all legit racing second i HATE when people ignore me..... and dont put
smart alec responses on this thread it will get me even more mad and dont
put this thread is in wrong section because I DONT CARE IM TRYING
TO PROVE A POINT AND ITS NOT JUSt pasqualina and mr.bean its
everyone!!!!!!!!!
2. Daphne (07-30-2011)
Actually you are. Most say that people that can keep over 9000 VR are pro.
And anyway, I’m ignoring you because playing with legit people is making
me mad as of late (Except for Miller).
3. Wolfy (07-30-2011)
It’s because they don’t feel like talking to you. Now stop bitching.
4. ξvePorta (07-30-2011)
Who cares? They don’t have to talk to the random fanboys. Just kidding,
it’s cause you’re Canadian.
5. Wolfy (07-30-2011)
Or it’s probably the Canadian thing.
6. Vices (07-30-2011)
*Pats head*
7. 13Stark37 (07-30-2011)
you suck soul
8. tortuga (07-30-2011)
*ignored*
9. AIRFLKISAAC (07-30-2011)
[…]I know I would ignore him after that..
10. Flying Mint Bunny (07-30-2011)
ok im sorry… i just feel bad now
11. Γ&#Megalodon24 (07-30-2011)
xD
12. Ryan1191 (07-30-2011)
*Ignores* This is the wrong place to put this thread. Now everyone’s ig-
noring you. (Retrieved from: http://www.mariokartwii.com/f18/
everyone-plzz-stop-ignoring-me-80655.html)
This interaction was retrieved from an online car racing game fun site forum.
AIRFLKISAAC post this message upon perceiving the higher-order intentionality
behind the behaviour of other members of the online community. That is, he feels
that he is being (ritually) ignored, as other community members do not respond
to his postings. If one analyses the responses to his appeal to the community, it
becomes clear that it is taken up by the others to further stigmatise him. In line 2
Daphne makes a non-committal response (see more on non-committal respons-
es in e.g. Clift 2012), by stating that he or she does not avoid AIRFLKISAAC for
what the latter claimed as a potential reason for being ignored – that is, not being
a professional player – but exactly because he or she is not interested in playing
with professionals. Through this response Daphne manages to diplomatically dis-
tance herself from AIRFLKISAAC, i.e. this is a strategy of “disaffiliation” (cf. Drew
and Walker 2009), and the stigmatised person thus remains isolated. In line 3
Wolfy animates the voice of “the normal”, as he or she refers to AIRFLKISAAC’s
stigmatised status in an explicit manner (“It’s because they don’t feel like talking
to you.”). Furthermore, he downgrades AIRFLKISAAC’s appeal to the group by
designating it as “bitching”, which is a frequently used word in in-group ritual
abusive behaviour to label somebody’s behaviour as deviant, i.e. different from
what members of the group would expect (cf. Zurcher 1985). In line 4 ξvePorta
makes use of what seems to be humour, by posting “Who cares? They don’t have
to talk to the random fanboys.” While in the following sentence he designates
this utterance as “kidding”, this is not “harmless” humour in the sense that (a) it
represents AIRFLKISAAC’s appeal as insignificant (“Who cares?”), i.e. it con-
firms that AIRFLKISAAC is excluded from the group, and (b) it reinforces the
divide between the group and the victim by portraying the stigmatised person
as a “random fanboy”. The ritual abuse then continues, and the next line of spe-
cific interest is 8 in which tortuga makes an explicit reference to “the normal’s”
stance to ignore the stigmatised by posting “*ignored*”. In turn 9 AIRFLKISAAC
makes a last attempt to restore his status among the others by apologising for
posting his enquiry and referring to his negative emotions (“ok im sorry… i just
feel bad now”). Apology could work in many other settings to restore membership;
as Tavuchis (1993: 8) argues, “apology expresses itself as the exigency of painful
re-membering”. However, AIRFLKISAAC remains stigmatised and the destructive
ritual continues: in line 10 Flying Mint Bunny turns towards other members of
the community as he refers to AIRFLKISAAC in the third person, as one he will
ignore, and in line 11 Γ&#Megalodon24 supports this uptake by posting “xD”, an
emoticon for laughter. The ritual continues to unfold to line 18, in which Ryan1191
addresses the victim directly again, but not with the goal to “acknowledge” his
existence. Instead of this, he animates the group’s voice in the form of a verdict, by
posting the utterance “*Ignores* This is the wrong place to put this thread. Now
everyone’s ignoring you”.
The present section has illustrated the way in which indirect ritual offence
operates. I believe that this section has shown that the form of recurrent impolite
behaviour studied here is very different from one-off instances of impoliteness,
which is more frequently studied in the field. In what follows, let us examine cases
in which indirect ritual offence becomes implicit.
4.2 When indirect ritual offence becomes implicit
In my dataset of 81 anecdotes of indirect and recurrent ritual offence there are 9

cases in which the narrator of the event of abuse uses the label “implicit”. 12 I have
no reason to assume that in these metadiscursive accounts the English word “im-
plicit” means exactly the same thing: following the uptake of Haugh and Kádár
(forthcoming), I argue that there is possibly a large gap between the meanings of the
same lexeme in various lay metadiscurses. However, there is a noteworthy similarity
between the cases in my data: in all the instances in which the narrator describes a
certain indirect ritual offence as implicit, there is a considerable time lag between
occurrences of abuse. This indicates that indirect ritual offence becomes implicit
from a ‘lay’ point of view if its frequency of occurrence decreases. Furthermore, as I
have argued in Section 2.1, this lay understanding of implicitness as a phenomenon
that occurs with the decreasing frequency of indirect offence coincides with the
academic understanding of implicitness as an interactional form that decreases the
retractibility of meanings. It is clear that the more infrequent these offences become,
the easier it becomes for the abusive person(s) to retract the offensive meaning.
Furthermore, with the decreasing frequency of such attacks, their pattern tends to
become less obvious from a participant point of view, i.e. the “linguistically encod-
ed” nature (Carston 2009) of the offence weakens.
If one examines Examples (1)–(3) in the previous section, it becomes clear that
all these cases of abuse took place frequently, and it was this frequency which made
the offensive nature of the abusers’ behaviour explicit from both the victim’s and
the abusers’ perspectives. For example, in extract (1) the managers who attack the
NHS therapist do not attempts to make the attack implicit, and they are outspoken
about their negative views of this person during the HR meeting. In Extract (4),
following the abusive ignorance of the targeted person, the group members make
it clear that they do actually ignore him, and that they will continue doing this. In
both cases the targeted person has no doubt about the attackers’ intention. While,
as Example (4) has demonstrated, this is a powerful form of abuse, the examination
of the 9 “implicit” cases in my dataset reveals that attacks, which are perceived as
implicit, are an equally effective form of rudeness. In these cases, the offence takes
place on an infrequent basis; due to the anecdotal nature of my data I am unable to
specify the actual time rate that “infrequency” means, but these anecdotes, such as
Example (5) below, make it clear that attacks in such cases are infrequent enough
to make the targeted person feel that the danger is over. The following Example (5)
illustrates the operation of this form of offensive behaviour:
12. Note that in the present research I have not studied cases in which “implicit(ness)” is used
with regard to one-off events, as such events are not ritual by nature.
(5) My name is Emmeline (pronounced like Em-ah-lyn). My story isn’t all that
bad, but I want to show a new form of bullying most teachers in my school
don’t recognize and my story that was mentioned very subtly to my friends.
[…]
Today, people in school don’t bullying [sic.] you in a very blunt way. They do
it more manipulatively, but I know I’m not fooled. (I even made myself my
own quote) They would ask you to come over and hang out with your group,
slowly making you feel comfortable with them and behind your back – laugh-
ter about what ridiculous things you do. If you did something weird, they
would “encourage” you to do more just to laugh at you. They would even try
to make fun of you by asking them to be your friend. This girl is (calling her
weird is rude) sorta... Naive about the things around her. Boys pretend they
have a crush on her; following her, saying I love you, hugging her. I wish she
would just be able to figure out that, they’re trying to bully her by letting her
comfort get out.
They say to tell on people who bully you, but why? In that form of bullying (I
call it “manipulative bullying”) would they believe you? The people will just
continue on and on and on and they would probably believe no one.
I know this wasn’t a bad bullying story, but I just wanna say, be careful with the
people you are with. Because people are cruel these days in the most implicit
and quiet way possible.
(Retrieved from: http://www.noplace4hate.org/real-bullying-stories/)
As this anecdote indicates, recurrent forms of indirect ritual offence become im-
plicit from a lay observer perspective if the abusers let someone’s “comfort get out”:
by being nice to a person who perceives themselves as unpopular, then offending
this person, just to continue comforting them is the way in which implicit ritual
offence operates. Similar to indirect cases, this form of ritual offence is difficult to
defend against, as people with authority “don’t recognize” it, as the author of this
anecdote makes clear.
Frequency seems to be a central issue in anecdotal accounts of implicit ritual
offence; some narrators, as in the following Example (6), are clear about the fact
that such forms of ritual offence work well exactly because they are occasional:
(6) It’s not explicit bullying; they’re not putting fireworks in his bag or doing
anything horrific, but sometimes [my emphasis] they make him the joke of a
few pranks and it’s more implicit bullying where he’s left out and has no one.
(Retrieved from: http://nineteen-yearslater.tumblr.com/page/31)
This anecdote – which narrates the problems of a freshman in a high school –

is noteworthy because it illustrates that implicit ritual offence can co-occur with
other forms of abusive behaviour. Here the victim is generally ostracised, which
is a “passive” form of abuse (cf. Section 4.1), but at the same time he is subject to
occasional pranks, which remain “implicit”, supposedly due to their relatively gentle
and infrequent nature.
5. Conclusion
The present chapter has attempted to fill a knowledge gap in pragmatics by ex-
amining the phenomenon of indirect ritual offence. The examples studied have
illustrated that it is fruitful for the impoliteness researcher to examine recurrent
ritual behaviour, as such forms of impoliteness operate in a significantly different
way from more widely studied one-off offences. In addition, this chapter has aimed
to contribute to research on implicitness, which is the focus of the present volume,
by examining cases in which indirect ritual offence becomes implicit from a lay
observer point of view, and being metapragmatically voiced.
Connecting implicitness with frequency of occurrence is, of course, only a
situated interpretation of this phenomenon. It reflects a lay understanding of this
phenomenon, which however coincides with academic definitions of how implic-
itness operates in impoliteness. It is subject to further research to attest whether
implicitness in other forms of interpersonal interactional behaviour can be ap-
proached on this basis or not. Yet, it is perhaps safe to argue that, due to the fact
that the formation and maintenance of interpersonal relationships is attuned to
regularities (see Terkourafi and Kádár 2017), and also peoples’ understandings of
the moral order within a given community is influenced by regularities (see Kádár
2017), irregular behaviour increases the sense of ambiguity.
References
Bloch, Sidney, Sally Browning, and Graham McGrath. 2011. “Humour in Group Psychotherapy.”
British Journal of Medical Psychology 56: 89–97. doi: 10.1111/j.2044-8341.1983.tb01535.x
Blum-Kulka, Shoshana. 1987. “Indirectness and Politeness in Requests: Same or Different?”
Journal of Pragmatics 11: 131–146. doi: 10.1016/0378-2166(87)90192-5
Brown, Penelope, and Stephen C. Levinson 1987. Politeness: Some Universals in Language Usage.
Byon, Andrew Sangspil. 2006. “The Role of Linguistic Indirectness and Honorifics in Achieving
Linguistic Politeness in Korean Requests.” Journal of Politeness Research 2: 247–276.
doi: 10.1515/PR.2006.013
Carston, Robyn. 2009. “The Explicit/Implicit Distinction in Pragmatics and the Limits of Explicit
Communication.” International Review of Pragmatics 1: 35–62. doi: 10.1163/187731009X455839
Clift, Rebecca. 2012. “Identifying Action: Laughter in Non-humorous Reported Speech.” Journal
Culpeper, Jonathan. 2010. “Conventionalised Impoliteness Formulae.” Journal of Pragmatics 42:

3232–3245. doi: 10.1016/j.pragma.2010.05.007
Culpeper, Jonathan. 2011. Impoliteness: Using Language to Cause Offence. Cambridge: Cambridge
Curtner-Smith, Mary E., Anne M. Culp, Rex Culp, Carrie Scheib, Kelly Owen, Angela Tilley,
Molly Murphy, Lauren Parkman, and Peter W. Coleman. 2006. “Mothers’ Parenting and
Young Economically Disadvantaged Children’s Relational and Overt Bullying.” Journal of
Child and Family Studies 15: 177–189. doi: 10.1007/s10826-005-9016-7
Drew, Paul, and Tracy Walker. 2009. “Going Too Far: Complaining, Escalating and Disaffiliation.”
Dunham, Yarrow, Andrew Scott Barron, and Mahzarin R. Banaji. 2006. “From American
City to Japanese Village: A Cross-cultural Investigation of Implicit Race Attitudes.” Child
Development 77: 1268–1281. doi: 10.1111/j.1467-8624.2006.00933.x
Dynel, Marta. 2009. “Where Cooperation Meets Politeness: Revisiting Politeness Models in View
of the Gricean Framework.” Brno Studies in English 35: 25–43.
Garandeau, Claire F., and Antonius H. N. Cilessen. 2006. “From Indirect Aggression to Invisible
Aggression: A Conceptual View on Bullying and Peer Group Manipulation.” Aggression and
Violent Behaviour 11: 612–625. doi: 10.1016/j.avb.2005.08.005
Gennep, van, Arnold. 1909 [1960]. Rite de Passage [The rites of passage]. New York and London:
Routledge.
Glasø, Lars, Stig Berge Matthiesen, Morten Birkeland Neilsen, and Ståle Einarksen. 2007. “Do
Targets of Workplace Bullying Portray a General Victim Personality Profile?” Scandinavian
Journal of Social Psychology 48: 313–319. doi: 10.1111/j.1467-9450.2007.00554.x
Goffman, Erving. 1981. Forms of Talk. Philadelphia: University of Pennsylvania Press.
Goffman, Erving. 1986. Stigma: Notes on the Management of Spoiled Identity. New York: Simon
& Schuster.
Grainger, Karen. 2011. “Indirectness in Zimbabwean English: A Study of Intercultural
Communication in the UK.” In Politeness Across Cultures, ed. by Franncesca Bargiela-
Chiappini and Daniel Kádár, 171–193. Basingstoke: Palgrave Macmillan.
doi: 10.1057/9780230305939_9
Haugh, Michael. 2010. “Jocular Mockery, (Dis)affiliation, and Face.” Journal of Pragmatics 42:
295–317. doi: 10.1016/j.pragma.2009.12.018
Haugh, Michael. 2015. “The role of English as a Scientific Metalanguage for Research in Pragmatics:
Reflections on the Metapragmatics of ‘Politeness’ in Japanese.” East Asian Pragmatics 1: 41–
74. doi: 10.1558/eap.v1i1.27610
Haugh, Michael, and Dániel Z. Kádár. (forthcoming). The Metapragmatics of Politeness and
Impoliteness.
Holtgraves, Thomas. 2005. “The Production and Perception of Implicit Performatives.” Journal
Hopper, Robert, and Kenneth Drummond. 1990. “Emergent Goals at a Relational Turning
Point: The Case of Gordon and Denise.” Journal of Language and Social Psychology 9:
39–65. doi: 10.1177/0261927X9091003
Jaszczolt, Katarzyna 2009. “Cancellability and the Primary/Secondary Meaning Distinction.”
Intercultural Pragmatics 6: 259–289. doi: 10.1515/IPRG.2009.015
Kádár, Dániel Z. 2013. Relational Rituals and Communication: Ritual Interaction in Groups.
Basingstoke: Palgrave Macmillan. doi: 10.1057/9780230393059
Kádár, Dániel Z. 2017. Politeness, Impoliteness and Ritual: Maintaining the Moral Order in
Interpersonal Interaction. Cambridge: Cambridge University Press.
Kádár, Dániel Z., and Michael Haugh. 2013. Understanding Politeness. Cambridge: Cambridge
Kádár, Dániel Z., and Rosina Márquez Reiter. 2015. “(Im)politeness and (Im)morality: Insights
from Intervention.” Journal of Politeness Research 11: 139–260. doi: 10.1515/pr-2015-0010
Kanetsuna, Tomoyuki, Peter K. Smith, and Yohji Morita. 2006. “Copying with Bullying at School:
Children’s Recommended Strategies and Attitudes to School-based Interventions in England
and Japan.” Aggressive Behaviour 32: 570–580. doi: 10.1002/ab.20156
Kowalski, Robin M., Susan P. Limber, and Patricia W. Agatston. 2008. Cyberbullying: Bullying in
the Digital Age. London: Blackwell Publishing.
Leech, Geoffrey. 1983. Principles of Pragmatics. Harlow: Longman.
Limberg, Holger. 2009. “Impoliteness and Threat Responses.” Journal of Pragmatics 41: 1376–1394.
doi: 10.1016/j.pragma.2009.02.003
Lorenzo-Dus, Nuria, Pilar G. Blitvich, and Particia Bou-Franch. 2011. “On-line Polylogues and
Impoliteness: The Case of Postings Sent in Response to the Obama Reggaeton YouTube
Video.” Journal of Pragmatics 43: 2578–2593. doi: 10.1016/j.pragma.2011.03.005
Monks, Claire P., Peter K. Smith, Paul Naylor, Christine Barter, Jane L. Ireland, and Iain Coyne.
2009. “Bullying in Different Contexts: Commonalities, Differences and the Role of Theory.”
Aggression and Violent Behaviour 14: 146–156. doi: 10.1016/j.avb.2009.01.004
Olweus, Dan. 1994. “Bullying at School: Basic Facts and Effects of a School Based Intervention
Program.” Journal of Child Psychology and Psychiatry 35: 1171–1190.
doi: 10.1111/j.1469-7610.1994.tb01229.x
Peter Metcalf and Richard Huntington 1991. Celebrations of Death: The Anthropology of Mortuary
Ritual. Cambridge: Cambridge University Press.
Rayner, Charlotte. 1997. “The Incidence of Workplace Bullying.” Journal of Community & Applied Social
Psychology 7: 199–208. doi: 10.1002/(SICI)1099-1298(199706)7:3<199::AID-CASP418>3.0.CO;2-H
Ren, Wei, Ching Yin Lin, and Helen Woodfield. 2013. “Variational Pragmatics in Chinese: Some
Insights from an Empirical Study.” In Research Trends in Intercultural Pragmatics, ed. by
Istvan Kecskes, and Jesus Romero-Trillo, 283–314. Berlin: Mouton de Gruyter.
doi: 10.1515/9781614513735.283
Sharp, Sonia, and Peter K. Smith (eds) 2002. School Bullying: Insights and Perspectives. London:
Routledge.
Silver, Maury, Rosaria Conte, Maria Miceli, and Isabella Poggi. 1986. “Humiliation: Feeling,
Social Control and the Construction of Identity.” Journal for the Theory of Social Behaviour
16: 269–283. doi: 10.1111/j.1468-5914.1986.tb00080.x
Smart Richman, Laura, and Mark Leary. 2009. “Reactions to Discrimination, Stigmatization,
Ostracism, and Other Forms of Interpersonal Rejection: A Multimotive Model.” Psychological
Review 116: 365–383. doi: 10.1037/a0015250
Spencer-Oatey, Helen. 2005. “(Im)politeness, Face and Perceptions of Rapport: Unpackaging
Their Bases and Interrelationships.” Journal of Politeness Research 1: 95–119.
doi: 10.1515/jplr.2005.1.1.95
Tavuchis, Nicholas. 1993. Mea Culpa: A Sociology of Apology and Reconciliation. Palo Alto, CA:
Stanford University Press.
Terkourafi, Marina, and Dániel Z. Kádár. 2017. “Convention and Ritual.” In The Palgrave Handbook
of Linguistic Politeness, ed. by Jonathan Culpeper, Michael Haugh, and Daniel Kádár.
Basingstoke: Palgrave Macmillan.
Thomas, Angela. 2007. Youth Online: Identity and Literacy in the Digital Age. Bern: Peter Lang.
Thornborrow, Joanna. 2002. Power Talk: Language and Interaction in Institutional discourse.
Harlow: Longman.
Twyman, Kimberly A., Conway F. Saylor, Danielle Saia, Michelle M. Macias, Lloyd A. Taylor, and
Eve Spratt. 2010. “Bullying and Ostracism Experiences in Children with Special Health Care
Needs.” Journal of Developmental & Behavioural Pediatrics, 31: 1–8.
doi: 10.1097/DBP.0b013e3181c828c8
van Goethem, Anne A. J., Ron H. J. Scholte, and Reinout W. Wiers. 2010. “Explicit and Implicit
Bullying Attitudes in Relation to Bullying Behaviour.” Journal of Abnormal Child Psychology
38: 629–642. doi: 10.1007/s10802-010-9405-2
Verschueren, Jef. 2000. “Notes on the Role of Metapragmatic Awareness in Language Use.”
Pragmatics 10: 439–456. doi: 10.1075/prag.10.4.02ver
Whutnow, Robert. 1989. Meaning and Moral Order: Explorations in Cultural Analysis. Berkeley,
CA: University of California Press.
Zurcher, Louis A. 1985. “The War Game: Organizational Scripting and the Expression of Emotion.”
Symbolic Interaction 8: 191–206. doi: 10.1525/si.1985.8.2.191
Chapter 9
Implicitness in the use

of situation-bound utterances
Istvan Kecskes
State University of New York, Albany
This chapter discusses the nature of implicit knowledge encoded in situa-

tion-bound utterances (SBUs) that are defined as highly conventionalized,
prefabricated pragmatic units whose occurrences are tied to standardized com-
municative situations because they serve as interactional patterns and rituals
that usually mean the same to all speakers of a particular speech community. It
will be argued that there is a strong connection between implicitness and con-
ventions of usage in language use. Conventional routine expressions like SBUs
encode information that is equally available for all members of the given speech
community. This information is usually implicit because the functional meaning
of expressions is rarely reflected in their compositional meaning. The paradox of
the use of SBUs is that although most of them are characterized by a high level of
implicitness, they may still represent the most direct way to express some social
function.
Keywords: situation-bound utterances, conventions of usage, tacit knowledge,

situational context, experience-based context
1. Introduction
The main argument of this chapter is that there is a direct relationship between
conventions and implicitness in human languages, which can be well demonstrated
through the use of SBUs. Conventions are developed to assure regular conformity
within a speech community which is important to its effectiveness. Relying on
Lewis (1969) Eemeren and Grootendorst (1984) argued that a language usage con-
vention exists in the usage of the members of a speech community if
A. the language of the members of the community displays a certain regularity
which occurs in strictly delineated cases,
doi 10.1075/pbns.276.09kec
202 Istvan Kecskes
B. the members of the community expect these regularities to occur in those cases,
C. the members of the community prefer the regularity to occur in those cases
because it solves a problem of communication or interaction.
Condition A expresses the factual aspect of conventions, condition C the normative

aspect, and condition B the social nature of conventions.
Searle (1979) made a distinction between conventions of language and conven-
tions of usage: “It is, by now, I hope, uncontroversial that there is a distinction to be
made between meaning and use, but what is less generally recognized is that there
can be conventions of usage that are not meaning conventions (Searle 1979: 49).”
Morgan basically talked about the same distinction:
In sum, then, I am proposing that there are at least two distinct kinds of convention
involved in speech acts: conventions of language …and conventions in a culture
of usage of language in certain cases…The former, conventions of language, are
what make up the language, at least in part. The latter, conventions of usage, are a
matter of culture (manners, religion, law…). (Morgan 1978: 69)
This chapter focuses on conventions of usage that is represented in our case by

situation-bound utterances. SBUs function like frozen implicatures and are a part
of a kind of collectively shared social background knowledge. People acquire them
through participation in social practices in our primary process of socialization
through imitation, training, and repetition, as Wittgenstein (1989) says, in the sense
of drilling. SBUs serve as a cohesive and unifying force in a language providing
speakers with the feeling of group-inclusiveness. SBUs usually communicate an im-
plicit norm, namely the correct participation that is distinguished from an incorrect
participation: one does it this way and not another in the given speech community.
As Millikan (2008) says “… To speak a language “idiomatically” is merely to use
mostly patterns that are well worn, that is, familiar (trite), patterns that place few
demands on a hearer.”
These idiomatic units deserve special attention for two reasons. First, they are
highly conventionalized. They mean a peace of mind for the speaker: her/his mes-
sage will not be misunderstood because SBUs usually have the same functional
meaning for all members of a speech community. Second, they have implicit infor-
mation that is not directly stated, still members of the speech community will take
it for granted so SBUs are usually not misunderstood in L1. Examples:
(1) Bob: You three visit the museum without me. Looking at fussy old paintings
is not my cup of tea.
(2) Going to church, Mary said, was not her cup of tea.
(source: McGraw-Hill Dictionary of American Idioms and Phrasal Verbs.
© 2002 by The McGraw-Hill Companies, Inc.)
Chapter 9. Implicitness in the use of situation-bound utterances 203
As Examples (1) and (2) demonstrate, the expression “it’s not my cup of tea” is a
frequently used SBU in situations when the speaker wants to refer to the fact that
the issue at hand is not his/her business. No L1 speaker of American-English will
understand this expression literally. Its most salient meaning for them is its figura-
tive meaning. 1 The socio-cultural load that is encoded in this SBU is so strong that
L1 speakers of English will interpret this expression figuratively even without actual
situational context (cf. Kecskes 2013). The implied functional meaning of the ex-
pression is like a tacit agreement between the members of the speech community. It
is an important element of group inclusiveness. However, this works differently for
non-native speakers who are not members of the target language culture (English)
and represent different cultures. They may not understand the socio-cultural load
that is attached to this expression, which can lead to misunderstanding.
2. Characteristics of situation-bound utterances
As mentioned above SBUs represent a relatively frequent group in formulaic lan-

guage use. The category of formulaic language is constituted by multi-word expres-
sions that tend to convey holistic meanings that are either more than the sum of the
individual parts, or else diverge significantly from a literal, or word-for-word mean-
ing and operate as a single semantic unit (Gairns and Redman 1986). There are
serious debates about what exactly can be considered formulaic language. Fillmore
found that “an enormously large amount of natural language is formulaic, automatic
and rehearsed, rather than propositional, creative or freely generated” (Fillmore
1976: 24). Altenberg (1998) claimed that 80% of our language production can be
considered formulaic. Wray’s definition of formula is very broad: “a formulaic se-
quence [is] a sequence, continuous or discontinuous, of words or other elements,
which is, or appears to be, prefabricated: that is, stored and retrieved whole from
memory at the time of use, rather than being subject to generation or analysis by
the language grammar” (Wray 2002: 9). Based on this definition much of human
language is formulaic rather than freely generated.
I argued (Kecskes 2007, 2013) that prefabricated sequences can be considered
formulaic only if they have psychological saliency as coherent units for the L1
speakers and are usually motivated and allow relatively few structural changes such
as fixed semantic units, speech formulas, phrasal verbs, idioms and situation-bound
utterances, for instance, “to be frank”, “as a matter of fact”, “got you”, “be my guest”,
“you are all set”, “no strings attached”, “kick the bucket”, etc. I ignored sequences
1. According to Giora’s graded salience hypothesis a lexical item or expression can have either
its literal meaning or its figurative meaning or both as most salient (Giora 1997, 2003).
204 Istvan Kecskes
such as “if you say…”, “this is good…”, “I have been…”, etc., which frequently occur
in any corpora but hardly have any psychological saliency in the mind of a L1
speaker. According to this approach SBUs can certainly be considered formulaic.
But it is still important to clarify the relation of SBUs to “conversational routines”
(cf. Coulmas 1981; Aijmer 1996) on the one hand, and to idioms on the other hand.
Semantic idioms (“make both ends meet”, “kick the bucket”) do have psychological
reality as coherent units. They are stored as unanalyzed chunks in memory just
like words, and are retrieved as a whole. They are usually not tied to particular
situations and can occur in any phase of a conversation where speakers find their
use appropriate. Pragmatic idioms are different. They can be split into two groups:
conversational routines and situation-bound utterances. The difference between
them is socio-cultural rather than linguistic.
Conversational routines (CR) have an inclusive relation to SBUs. CRs consti-
tute a much broader category than SBUs. Conversational routines include speech
formulas (“you know”, “I see”, “no problem”), discourse markers (see Fraser 1999)
and SBUs. All SBUs are conversational routines, but this is not so conversely because
not all expressions labeled as conversational routines are SBUs. For instance, “you
know”, “I see”, and “no problem” can be considered conversational routines but
they are not SBUs. Aijmer (1996: 11) argued that conversational routines are ex-
pressions which, as a result of recurrence, have become specialized or “entrenched”
for a discourse function that predominates over or replaces the literal referential
meaning. It is not easy to draw the dividing line between conversational routines
and SBUs but there are some features that distinguish them. Conversational routines
are function-bound rather than situation-bound. They can express one and the same
particular function in any situation while SBUs frequently receive their charge from
the situation itself. For instance, “after all” or “to tell you the truth” are conversa-
tional routines rather than SBUs. They can be uttered in any situation where they
sound appropriate. However, expressions such as “how do you do?” used upon
acquaintance, or “welcome aboard” as a greeting to a new employee make sense
only in particular well-definable situations (see Kecskes 2013).
The tie of an SBU to a particular situation that charges the particular meaning
of the given SBU may become so dominant that the functional-situational meaning
may take over as the most salient meaning of the expression, for instance, “piece of
cake”, “help yourself ”, “give me a hand”, or “it’s not my cup of tea”. Conversational
routines tend to have discourse functions rather than a situation-bound function, for
instance, “as a matter of fact”, “suffice it to say”, “to tell the truth”, and others. Discourse
functions are not necessarily tied to particular situations. They can be expressed by
conversational routines including not only SBUs but also expressions of turn-taking,
internal and external modifiers, discourse markers, connectors, and others.
SBUs differ from semantic idioms in origin, purpose and use. The likelihood
of occurrence of lexico-semantic idioms is usually unpredictable while the use of
situation-bound utterances is generally tied to particular social contexts. Semantic
idioms just like metaphors originate from creative linguistic acts. They are used
to represent complex content in a tangible way that can hardly be analyzed con-
ceptually. Situation-bound utterances are repetitive expressions whose use saves
mental energy. Semantic idioms are like lexemes while SBUs are more like prag-
matic markers. SBUs fulfill social needs. People know if they use these prefabricated
expressions they are safe: nobody will misunderstand them because these phrases
usually mean the same to most speakers of a speech community. However, there is
a price for repetitiveness. SBUs often lose their composition meaning and become
pure functional units denoting greetings, addressing, opening, etc. This is where
we can draw the dividing line between semantic idioms (“spill the beans”, “kick the
bucket”, “pull one’s leg”, etc.) and SBUs (“see you later”, “it’s been a pleasure meeting
you”, “say hello to”, etc.). While semantic idioms are not transparent at all, pragmatic
idioms (SBUs) generally remain transparent and usually have a freely generated
counterpart (for instance: “get out of here”, “welcome aboard”). In contrast to idi-
oms SBUs do not mean anything different from the corresponding free sentences:
they simply mean less. Their meaning is functional rather than compositional.
However, there are some cases when semantic idioms (mainly frozen metaphors)
function like SBUs such as “help yourself ”; “piece of cake” and the like. In fact,
some of those semantic idioms can become SBUs when they begin to be used very
often in one or more typical social situation such as “help yourself ”. This process
will be discussed later.
The loss of compositionality is a matter of degree. When SBUs are frequently
used in a particular meaning they will encode that meaning, and develop a particu-
lar pragmatic function. This pragmatic property is getting conventionalized when it
starts to mean the same thing for most native speakers. That is to say, when native
speakers are asked what comes into their mind first when they hear a given expres-
sion, and their response will be very similar, we can say that the SBU has already
encoded a specific pragmatic property. For instance, in Chinese nobody will think
of the literal meaning of “nĭ chī le ma?” [have you eaten?], when it is used as an SBU.
SBUs are both selective and completive. They are selective because they are
preferred to be used to a number of utterances, both freely generated and idiomatic,
which equally could be used in the given situation. SBUs are completive because
they evoke a particular situation, which freely generated utterances usually do not
do. For instance, “let me tell you something” generally creates a negative expec-
tation by the hearer, or “step out of the car, please” is something that most people
identify with police stops. In freely generated utterances, the sense of the utterance
206 Istvan Kecskes
is defined by the interplay of linguistic meaning and context, situation, background

knowledge. In SBUs, however, the communicative meaning, the sense of the utter-
ances, is encoded, and fixed by pragmatic conventions. Consequently, prior context
encapsulated in them can create actual situational context, for instance, “license and
registration, please”, “can I help you?”, “do you want to talk?”. All these expressions
can create their own situation (based on the “history” of their prior use) without
being used in an actual situational context.
3. Tacit knowledge
In language use, interlocutors need to rely on tacit knowledge that is available for
all members of the speech community. It is needed because however complex a
language system is, it would be very difficult to verbalize everything (cf. Kurzon,
this volume). So, to some extent, it is true that linguistic signs, and partly, utter-
ances are underdetermined, and both speakers and hearers take this for granted.
They use their tacit knowledge and actual situational context to put sense into their
utterances (speakers) and to make sense of what they hear (hearers). The following
example shows different degrees of explicitness: 2
(3) Bob: Do you think Jim was drunk at the party yesterday?
1. Tell me about it
2. Of course, he was
3. Well, he could hardly find the bathroom
4. Nobody knows
5. You may be right
The first response (1) appears to be the vaguest semantically. However, at the
pragmatic level of analysis the expression “tell me about it” is the clearest of all to
express agreement with Bob. It is a strong implicature. 3 It is strongly implicated
that the speaker does not want Bob to tell about how drunk Jim was because he
knows that. In this context, Bob will interpret this expression (“Tell me about it”)
only one way: his interlocutor agrees with him. And it does not matter whoever
the L1 speaker is, Bob or someone else, the interpretation will be the same. How
is it possible that the expression that shows the vaguest relevance to the question
semantically will convey the strongest reply? The reason is that L1 language us-
ers have what we call “preferred ways of saying things” (cf. Wray 2002; Kecskes
2. “Explicitness” is used here as an opposition to “implicitness”.

3. About “strong implicatures” see, for instance, Ariel (2016)
2013) and “preferred ways of organizing thoughts” (Kecskes 2013). As members

of a speech community, we share tacit knowledge with other members, which is
collectively salient. In the example above “tell me about it” can be considered a
strong implicature because the implied functional meaning has priority over most
(even direct) utterances.
Preferred ways of saying things are generally reflected in the use of formulaic
language and figurative language while preferred ways of organizing thoughts can
be detected through analyzing, for instance, the use of subordinate conjunctions,
clauses and discourse markers. Selecting the right words and expressions, and
formulating utterances in ways preferred by the native speakers of that language
(“nativelike selection”) is more important than syntax. SBUs can be considered
preferred ways of expressing social and/or situational functions. For instance, when
the nurse says in the waiting room at the doctor’s office: “The doctor will be in in a
few minutes”. Everybody knows that this will not happen: “a few minutes” could be
half an hour or more. The function of the expression is to indicate to the patient that
his/her presence is acknowledged, and s/he will be taken care of soon. This shared,
tacit knowledge of participants is very important here. What Putnam (1973) called
“the division of linguistic labor” makes it possible that one can be understood as
sharing a language with other members of one’s speech community because one
looks for their help when deciding how one’s terms are correctly applied, and the
other participants do the same with one. It is in this cooperative sense of sharing
that someone can share a language with his/her fellow speakers in more than just
a “loose” sense.
When SBUs are frequently used in a particular meaning they will encode that
meaning, and develop it into a particular pragmatic function. As mentioned above
SBUs are both selective and completive. They are selective because they are pre-
ferred to be used in a number of utterances, both freely generated and idiomatic,
which equally could be used in the given situation. SBUs are completive because
they evoke a particular situation, which freely generated utterances usually do not
do. For instance, “let me tell you something” generally creates a negative expectation
by the hearer, or “step out of the car, please” is something that most people identify
with police stops. In freely generated utterances the sense of the utterance is defined
by the interplay of linguistic meaning and context, situation, background knowl-
edge. In SBUs, however, the communicative meaning, the sense of the utterances is
encoded, and fixed by pragmatic conventions to some extent. Consequently, prior
context encapsulated in them can create actual situational context. For instance, “be
my guest”, “can I help you?”, “you are all set”. All these expressions can create their
own context without being used in an actual situational context.
It is, however, very important that we understand that these preferred ways of
saying things, these conventions serve only as guidelines, models. Models are used
208 Istvan Kecskes
as points of reference and guidance. We decide to approximate them more or less

according to the demands of a specific situation. So they are not requirements but
are something whose impact we cannot ignore. They are there in L1 and will always
affect what we say or how we interpret words and utterances.
4. Context-dependence of implicitly conveyed information
The functional meaning of SBUs is usually less context-dependent than other types
of expressions because they are tied to certain social situations. However, some
SBUs are more context-sensitive than others. Before we discuss what context-
sensitiveness depends on in SBUs, we will need to look at how context is under-
stood in this chapter.
4.1 Understanding context
According to the dominant view in linguistics, context-sensitivity (in various

forms) is a pervasive feature of natural language. Carston claimed that “…linguisti-
cally encoded meaning never fully determines the intended proposition expressed”
(Carston 2002: 49). Consequently, linguistic data must be supplemented by non-lin-
guistic, contextual interpretation processes. This contextualism requires a discus-
sion of what is understood by “context” in pragmatics research.
In linguistics, context usually refers to any factor – linguistic, epistemic, phys-
ical, social – that affects the actual interpretation of signs and expressions. The
notion that meanings are context-dependent has informed some of the most pow-
erful views in current linguistic and philosophical theory, all the way from Frege to
Wittgenstein and beyond. Frege’s (1884) Context Principle asserts that a word has
meaning only in the context of a sentence. Wittgenstein (1921) basically formulated
the same idea, saying that an expression has meaning only in a proposition; every
variable can be conceived as a propositional variable. Such external perspectives
on context hold that context modifies and/or specifies word meanings in one way
or another. Context is seen as a selector of lexical features because it activates some
of those features while leaving others in the background.
Opposite to the externalist view on context is the internalist perspective. It
considers lexical units as creators of context (e.g. Gee 1999; Violi 2000). Violi (2000)
claimed that our experience is developed through a regularity of recurrent and
similar situations that we tend to identify with given contexts. Standard (prior
recurring) context can be defined as a regular situation that we have repeatedly
experienced, about which we have expectations as to what will or will not happen,
and on which we rely to understand and predict how the world around us works.
It is exactly these standard contexts that are very important for SBUs. This is why I
mentioned the completive nature of SBUs above: they are able to evoke a particular
situation. For instance, the SBU “do you want to talk?”, without any actual situa-
tional context, may evoke the situation in which it is used: Someone appears to be
in trouble and his friend/wife/colleague/etc. turns to him with this question. The
tacit knowledge that is shared is that the person offers help, consolation, etc. with
uttering the SBU. Gumperz (1982) said that utterances somehow carry with them
their own context or project that context. Similarly, Levinson (2003), referring to
Gumperz’s work, claimed that the message-versus-context opposition is misleading,
because the message can carry with it, or forecast its context.
Referring to the socio-cognitive approach to pragmatics, Kecskes (2013) argued
that the main problem with both the externalist and internalist views of context is
that they are one-sided inasmuch as they emphasize either the selective (external-
ist) or the constitutive role (internalist) of context. However, the dynamic nature
of human speech communication requires a model that recognizes both regularity
and variability in meaning construction and comprehension, and takes into account
both the selective and constitutive roles of context at the same time. Millikan (1998)
claimed that the conventional sign (the lexical unit) is reproduced (or “copied”
as she said), not discovered or invented anew by each producer–processor pair.
This can only happen if the linguistic unit has some kind of regular reference to
certain contexts in which it has been used. Leibniz (1976 [1679]) said: “… si nihil
per se concipitur, nihil omnino concipietur” (“… if nothing is understood by itself,
nothing at all will ever be understood”). Words of a particular language can create
context because they encapsulate prior situational contexts in which a speaker has
used them (e.g. Violi 2000; Kecskes 2008). Violi argued that “it is not the existence
of a given context that makes the use of the word possible, but the use of the word
that initiates a mental process in the listener which seeks to construct a context
in which its present use could be most appropriate” (Violi 2000: 117). When “get
out of here” or “license and registration, please” are uttered without any actual
situational context, these expressions will create a situational context in the mind
of hearers because of their prior experience with these lexical units. This relative
regularity attached to lexical units of a language changes diachronically while varia-
bility (variety) changes synchronically. That is to say standardized lexical meanings
change diachronically while variations in the context of their utterance change syn-
chronically. What we need is an approach to communication that recognizes both
the selective and constitutive role of context. It is necessary to make a difference
between actual situational context, which theories of pragmatics mean when they
use the term “context”, and prior experience-based context that is the main condition
210 Istvan Kecskes
for standardization and conventionalization. Both types of context are equally im-
portant in meaning construction and comprehension. The constitutive role of prior
experience-based context is well demonstrated in the SBUs above. The tacit knowl-
edge built in SBUs is shared by most members of an L1 speech community.
4.2 SBUs are context sensitive in different degrees
SBUs are either transparent or not, and are usually motivated in varying degrees.
They are “idiomatized” in the sense that the words in them are taken as a whole and
constitute a pragmatic unit with a particular function. Nattinger and DeCarrico
(1992: 128) referred to them as “idioms with a pragmatic point”. The less an SBU is
motivated, the more it becomes similar to what we called “semantic idioms” above.
According to the degree of motivation, and partly their context sensitivity we can
distinguish three types of SBUs: plain, loaded and charged (Kecskes 2003, 2010).
Plain SBUs have a compositional structure and are semantically transparent. Their
situational meaning may only differ slightly from their propositional meaning be-
cause their pragmatic extension is minimal if any. Their meaning can usually be
computed from their compositional structure so they are not context sensitive.
For instance:
(4) Assistant: Can I help you, Sir?
Customer: Thank you. I’m just looking.
In this conversation “Can I help you?” and “I’m just looking” function as plain SBUs
while “thank you” is a speech formula.
On the other end of the continuum we find loaded SBUs that are the closest
to semantic idioms because they may lose their compositionality and are usually
not transparent semantically any more. They may be considered semantic idioms
with a pragmatic function. Their pragmatic function is more important than their
original literal meaning that is difficult to recall if needed. These SBUs are “loaded”
with their pragmatic function that remains there, and usually cannot be cancelled
by the actual situational context because it is encoded in the expression as a whole.
Consequently, loaded SBUs are not as context sensitive as charged SBUs. We think
of a particular situation even if we hear the following expressions without their
routine context: “It’s not my cup of tea”, “tell me about it”, “is that it?”, etc., because
their most salient meaning is the one that is extended pragmatically.
Charged SBUs come in between plain and loaded SBUs and they are actual sit-
uational context sensitive. An SBU may exhibit pragmatic ambiguity, in the sense
that its basic function is extended pragmatically to cover other referents or mean-
ings (Sweetser 1990: 1). For instance, this is the case with a phrase such as “see you
soon”, which retains its original sense but can also be conventionally (situationally)
interpreted as a closing, a way to say good-bye to one’s partner. So this expression
has two interpretations: a literal one and a situation-bound one. However, the situ-
ation-bound function (“closing”) is charged by the actual situational context only. If
the expression “see you soon” is given without a particular actual situational context
it may be ambiguous because it can create one of two situations in the mind of a
hearer: (1) closing, a way to say good-bye, and (2) what its compositional meaning
says: the speaker will see the interlocutor soon.
(5) Sara: OK, Bill, I must go now. See you soon.
(6) Gail: When can we continue discussing this plan?
Mary: I think after lunch, say, at 3:00. See you soon.
The difference between (5) and (6) is that in (6) “see you soon” has a concrete ref-
erence. They will meet again at 3:00, which will be soon.
Here is another example with the expression “welcome aboard”.
(7) Jenny to the flight attendant: Where is seat number 17?
Flight attendant: To the left. Welcome aboard.
(8) Jill: I am glad that I was offered this job.
Bob: We appreciate that you accepted our offer. Welcome aboard.
In (7) “welcome aboard” is transparent. It functions as a SBU while in (8) the ex-
pression is an SBU that also serves as a greeting but in a figurative sense: greeting
a new employee.
5. Can SBUs be underspecified?
Langacker (2000: 152) argued that the component structures of complex expres-

sions usually motivate, but do not predict the composite structure. Language thus
exhibits partial rather than full compositionality, i.e. the meaning of a composite
structure is underspecified and must be constructed. This seems definitely true in
most cases. But does this work with SBUs? Can SBUs be underspecified?
Yus (1999) argued that due to the higher context-dependence of implicitly
conveyed information, the more implicit the intended interpretation is (i.e., the
greater the number of contextual assumptions required for the listener to interpret
the message) the more likely it is that the interpretive process will end up in some
kind of speaker-hearer mismatch. In other words, the more hearer-dependent and
the less speaker-controlled interpretation is, the greater the risk of eventual mis-
understanding. Yus is right if we relate what he says to general language use. But
212 Istvan Kecskes
SBUs are unique expressions. As discussed above, the compositional meaning of

situation-bound utterances often becomes of secondary importance and the func-
tional aspect begins to dominate. Frequency and familiarity that are of key impor-
tance to implicitness correlate in a unique way in SBUs. Frequency can be general
or attached to a particular register or situation. For instance, the utterance “Hello,
how are you?” is very frequently used in everyday interaction. This is true because
the situation (meeting and greeting others) requiring the use of this (or a similar)
expression occurs very often. There seems to be a difference between word frequen-
cy and utterance frequency. Word frequency refers to the general use of words in
any kind of situation. Utterance frequency, however, is more register-oriented and/
or situation-bound. It is especially true if we take SBUs. It does not make much
sense to speak about the general frequency of utterances when they are usually
register-oriented and/or situation-bound. Consequently, the frequency of an SBU
depends on the frequency of a given register or a situation the SBU is attached to.
In L1 SBUs are never underspecified and usually not misunderstood. This is
because they are direct reflections of what is considered appropriate language use
in a speech community. But because of their tacit function and implicit directness
SBUs mean particular challenge to non-native speakers of a language. Here is a
conversation that took place between Roberts Sanders and a Pizza Hut associate. 4
(9) He was ordering a pizza on the phone. The woman who answered was fluent
in English but had an accent.
Sanders: I’d like to order a medium pizza.
Woman: Is that pickup or delivery.
Sanders: Pickup.
Woman: Is that it?
Sanders: What?
Woman: Is that it?
Sanders: Is that what?
Woman: (No response. Silence)
Sanders: We want three toppings: pepperoni, mushroom, cheese
Woman: OK, you want pepperoni, mushroom and cheese
Sanders: Right.
Woman: Okay, about 20 minutes.
Although the woman at Pizza Hut was fluent in English, her inappropriate use
of the SBU “is that it” caused a slight breakdown in the interaction. Normally,
“is that it?” is a formula used to close this part of the transaction and move on to
something new or closing. But for Sanders, the transaction was not expected to
4. I am obliged to Robert Sanders for providing me with this example.

have been moving on, because they were still in the middle of the ordering process.
He had not told the assistant yet what toppings he wanted. The woman repeated,
“is that it?”, and Sanders said “is that what?”. This was followed by silence from
the woman who must have been confused. Then Sanders told her what toppings
he wanted, which she understood perfectly, and then they closed the transaction
properly.
6. Conclusion
Implicit communication may be more demanding, but it is far from unusual, and
satisfies a high number of communicative purposes in everyday interactions, which
can range from the communication of information connoted with contextual im-
plications, to making information mutually manifest, or to the performance of
conversational strategies with the help of situation-bound utterances that assure
appropriateness in an actual situational context.
As far as implicitness is concerned, SBUs are unique pragmatic units because
in them tacit functional meaning does not hamper communication rather makes
interaction smoother (cf. Kurzon, this volume; Dynel and Cap, this volume).
Interlocutors can be sure that no misunderstand occurs. It was argued and demon-
strated that although SBUs are characterized by a high level of implicitness, still
they may represent the most direct way to express some social functions. They are
considered strong implicatures although sometimes their compositional meaning
does not seem to have much relevance in the actual situational context.
SBUs serve as a cohesive and unifying force in a language providing speakers
with the feeling of group-inclusiveness. This is because they usually communicate
an implicit norm, which is considered important for the speech community mem-
bers to follow. Tacit knowledge that is encoded in them is connected with their
functional meaning rather than their compositional structure and meaning.
SBUs are probably the only pragmatic units where recipient design works prop-
erly: speaker control works at its highest level and hearer-dependency is decreased
to its minimum level. The result is mutual understanding. Interlocutors can only
wish that they have access to more SBUs.
214 Istvan Kecskes
References
Aijmer, Karin. 1996. Conversational Routines in English: Convention and Creativity. London:
Longman.
Altenberg, Bengt. 1998. “On the Phraseology of Spoken English: The Evidence of Recurrent
Word-Combinations.” In Phraseology: Theory, Analysis, and Applications, ed. by Paul Cowie
Anthony, 101–122. Oxford: Clarendon Press.
13: 1–35. doi: 10.1515/ip-2016-0001
Coulmas, Florian. 1981. Conversational Routine: Explorations in Standardized Communicative
Situations and Prepatterned Speech. The Hague: Mouton de Gruyter.
Eemeren, van Frans H., and Rob Grootendorst, 1984. Speech Acts in Argumentative Discussions.
A Theoretical Model for the Analysis of Discussions Directed Towards Solving Conflicts of
Opinion. Dordrecht: Foris Publications. doi: 10.1515/9783110846089
Fillmore, Charles J. 1976. “The Need for a Frame Semantics within Linguistics.” Statistical
Methods in Linguistics 12: 5–29.
Frege, Gottlob. 1884/1980. The Foundations of Arithmetic. Trans. by John L. Austin (2nd ed.).
Evanston, IL: Northwestern University Press.
Gairns, Ruth, and Stuart Redman. 1986. Working with Words: A Guide to Teaching and Learning
Vocabulary. Cambridge: Cambridge University Press.
Giora, Rachel. 1997. “Understanding Figurative and Literal Language: The Graded Salience
Hypothesis.” Cognitive Linguistics 8: 183–206. doi: 10.1515/cogl.1997.8.3.183
Giora, Rachel. 2003. On Our Mind: Salience Context and Figurative Language. New York: Oxford
University Press. doi: 10.1093/acprof:oso/9780195136166.001.0001
Gumperz, John J. 1982. Discourse Strategies. Cambridge: Cambridge University Press.
doi: 10.1017/CBO9780511611834
Kecskes, Istvan, 2003. Situation-Bound Utterances in L1 and L2. Berlin: Mouton de Gruyter.
doi: 10.1515/9783110894035
Kecskes, Istvan. 2007. “Formulaic Language in English Lingua Franca.” In Explorations in Pragmatics:
Linguistic, Cognitive and Intercultural Aspects, ed. by István Kecskés, and Laurence R. Horn,
191–219. Berlin: Mouton de Gruyter.
Kecskes, Istvan. 2010. “Situation-bound Utterances as Pragmatic Acts.” Journal of Pragmatics 42:
2889–2897. doi: 10.1016/j.pragma.2010.06.008
Kecskes, Istvan. 2013. Intercultural Pragmatics. Oxford: Oxford University Press.
doi: 10.1093/acprof:oso/9780199892655.001.0001
Langacker, Ronald. 2000. Grammar and Conceptualization. Berlin: Mouton de Gruyter.
Leibniz, Gottfried Wilhelm. 1976/1679. Philosophical Papers and Letters. Trans. and ed. by Leroy
E. Loemker. Dordrecht/Boston: D. Reidel. doi: 10.1007/978-94-010-1426-7
Levinson, Stephen C. 2003. “Language and Mind: Let’s Get the Issues Straight!” In Language in
Mind: Advances in the Study of Language and Cognition, ed. by Gentner Dedre, and Susan
Goldin-Meadow, 25–46. Cambridge, MA: MIT Press.
Millikan, Ruth. 2008. “A Difference of Some Consequence between Conventions and Rules.” Topoi
27: 87–99. doi: 10.1007/s11245-008-9026-3
Morgan, Jerry L. 1978. “Two Types of Convention in Indirect Speech Acts.” In Pragmatics (Syntax
and Semantics 9), ed. by Peter Cole, 261–280. New York: Academic Press.
Nattinger, James R., and Jeanette S. DeCarrico. 1992. Lexical Phrases and Language Teaching.
New York: Oxford University Press.
Putnam Hilary. 1973. “Meaning and Reference.” The Journal of Philosophy 70: 699–711.
doi: 10.2307/2025079
Searle, John. 1979. Expression and Meaning: Studies in the Theory of Speech Acts. Cambridge:
Cambridge University Press. doi: 10.1017/CBO9780511609213
Violi, Patrizia. 2000. “Prototypicality, Typicality, and Context.” In Meaning and Cognition: A
Multidisciplinary Approach, ed. by Liliana Albertazzi, 103–123. Amsterdam: John Benjamins.
doi: 10.1075/celcr.2.06vio
Wittgenstein, Ludwig. 1921/1922. Tractatus Logico-Philosophicus. Trans. by Charles Ogden. London:
Routledge and Kegan Paul.
Wray, Alison. 2002. Formulaic Language and the Lexicon. Cambridge: Cambridge University
Press. doi: 10.1017/CBO9780511519772
Yus, Francisco. 1999. “Misunderstandings and Explicit/Implicit Communication.” Pragmatics 9:
487–517. doi: 10.1075/prag.9.4.01yus
Chapter 10
Thematic silence as a speech act
Dennis Kurzon
University of Haifa
This chapter will discuss the speech act status of thematic silence, one of the four
types of silence presented in Kurzon (2007). This type of silence is not strictly
silence in the sense that someone is not talking where talk is expected. It refers
to the non-mention of a topic by the speaker (“s/he is silent about” in contrast
to “s/he is silent”). Despite its non-mention, the topic is often implicit and iden-
tifiable, since it is salient in one way or another. For example, in a speech or in
a press interview, a politician does not mention, or refuses to refer to, a topic
which may embarrass him/her, and this topic has recently been in the news.
Assuming that intentional silence is as communicative as the linguistic ways
of conveying implicit meaning, and has therefore some illocutionary force, the
question asked is what kind of speech act the omitted discourse (utterance / sen-
tence) may be. It will be argued that in discourses such as speeches, interviews,
or narratives in general, the silence usually replaces an assertive speech act,
though there may be instances in which the omitted speech act may be a direc-
tive or commissive, etc. speech act.
Keywords: silence, typology of silence, thematic silence, speech acts,

conversation analysis, political speech
1. Introduction
In Kurzon (2007) I set up a typology of silence consisting of four types: conversa-

tional, textual, situational and thematic. The first type – conversational silence – is
the usual type of silence described and analyzed in the literature on silence in lin-
guistics and pragmatics, especially in the field of conversation analysis. This silence
may occur when a participant in a conversation does not respond to a question or
comment addressed to him/her, as in
doi 10.1075/pbns.276.10kur
218 Dennis Kurzon
(1) Claire:
So then we were worse o- ‘n she an’ she went down
four,
(0.5)
Claire: But uhm
(1.5)
Claire: | Uh
Chloe: | Well then it was her fault |Claire,
Claire: |Yeah she said one
no trump, and I said
Two, an’ then she went back t’ two…
(Sacks et al. 1974: 704)
where Chloe finally responds after Claire utters three utterances separated by two
instances of silence, one half a second long, and the other, one and half seconds
long. A further example of conversational silence may occur when a person who
is present in the conversation, that is, one of the apparent participants, does not
contribute anything (verbal or non-verbal) to the conversation. A third example is
the right of silence in criminal law when a suspect refuses to answer questions put
to him/her by the police investigator or by counsel in court.
Textual silence and situational silence occur in fairly well-defined circum-
stances. Textual silence refers to the silence that occurs in a situation in which one
or more persons are silent while reading or looking at a specific text. This may
occur in a library where silence tends to be the rule (perhaps more broken now-
adays than obeyed). Library users may be reading a book or article that interests
them, or looking at a microfiche, or may simply be looking at the online catalogue
on a computer screen. Textual silence may also occur in the classroom when the
teacher asks the class to read silently to themselves a passage or a page of a book.
A further example is saying a prayer in silence. Textual silence may also occur
at home when members of the family are sitting together in the living room, but
each of them is reading something else – in silence without disturbing anyone else.
These examples may be somewhat ideal. Comments such as “Did you see this in
the newspaper?” may be uttered though all the family members are reading their
individual reading material. Children at the back of the classroom may be talking
quietly among themselves, despite textual silence, without the teacher noticing.
All of these instances of textual silence may and do occur, perhaps with less fre-
quency than formerly.
Situational silence is similar to textual silence in that a person or a group of peo-
ple are silent in a specific situation, but no particular text is being read or recited by
heart. This occurs often at institutionalized events, for example war remembrance
ceremonies with its one or two-minute silence, or the somewhat controversial mo-
ment of silence at the start of the school day in many of the states of the USA (see,
e.g., Kurzon 2011). The silence of soldiers or police waiting in an ambush is another
Chapter 10. Thematic silence as a speech act 219
example of situational silence, while far less formal is the silence of guests waiting
for the guest of honour at a surprise party.
These three types of silence may all be timed. In conversational silence, even
a pause of half a second or so, which is usually considered a pause (see Example 1
above), may be perceived as silence given the circumstances. If there is a silent
participant physically present in a conversation, his or her silence could last for
quite some time. Reading a page in silence can be timed, and teachers asking their
students to do just that would judge how much time is needed. In institutional-
ized instances of situational silence, the length of silence is laid down by law or by
convention.
The fourth type of silence in my typology (Kurzon 2007) – thematic silence – is
in effect not silence, though it is referred to in many languages, including English,
by this word. 1 This type of silence may be glossed as “to be silent about” as opposed
to “to be silent” as in the other three types. Typical instances of this type of silence
may be found in the speech of diplomats, who avoid mentioning something that
may disclose, for example, secret negotiations, or in the speech of politicians who
avoid saying something to their detriment. Omitting something – deliberately or
not – may also, of course, occur in everyday conversation, as we shall see.
The question I am asking in this chapter is whether we may identify what kind
of speech act the silence replaces. This question is too broad for the other types of
silence delineated above. A text that is being read in silence could be a mere shop-
ping list, or an editorial in a newspaper in which assertions, questions, suggestions
may occur, or a statute passed by parliament, which consists mainly of directives,
or something on an internet website with statements, questions, commands, sug-
gestions, acknowledgements, encouragements, congratulations, etc. In such texts
the speech act status of the text as a whole or of individual sentences or utterances
may be identified. The thematic silence of a speaker, on the other hand, is highly
contextualized in the co-text of what the speaker is saying and/or in the situation
in which the speaker is speaking.
In the following section, I will discuss intentional and unintentional silence in
terms of what is called “mnemonic silence” in psychology, followed in Section 3 by
an analysis of utterances that are speech acts with various degrees of indirectness,
which I have labeled implicit speech acts. In Section 4 a political speech, that of
Ed Miliband, the ex-leader of the British Labour Party at the party conference
in September 2014, will be analyzed in terms of the silence on two issues he was
expected to talk about – immigration and the budget deficit. Section 5 concludes
the chapter.
1. But not in, for example, in languages such as French, Basque, Polish or Hausa in which one
would have to say something like “S/he did not speak about X”.
220 Dennis Kurzon
2. Intentional and unintentional silence
A person who is silent in a social interaction may intentionally decide not to speak as
in the case of a suspect exercising the right of silence in a police investigation, or his/
her silence is unintentional in that it may be psychologically motivated: the person is
embarrassed or too shy to respond. Another possible reason for a person being silent
in a social interaction is that s/he has forgotten what to say. But in the case of thematic
silence, it is not because s/he has forgotten what to say – this may lead to conversation-
al silence – but s/he has forgotten to mention a specific thing while speaking. From
a psychological perspective, Berger (2004) distinguishes between strategic reasons
for silence, i.e. silence which the potential speaker decides to “perform”, and what he
calls “involuntary speechlessness”. Based on two studies in which he asked groups of
students to recall the most recent instance of silence and describe the circumstances,
he found that the main causes of silence are unexpected information, stress, extreme
emotions or nervousness, and lack of information and knowledge about the topic,
though the latter was shown not to be a major cause of silence.
Stone et al. (2012: 40), also from a psychological perspective, but here specifically
in the field of memory, discuss “mnemonic silence”, which is similar to what I have
called thematic silence. They define mnemonic silence as “the refusal or failure to
remember”, that is “a mnemonic silence occurs when a person fails or refuses to ex-
press in a conversation a memory that, under other circumstances, could and would
be remembered and expressed.” They set up four types of mnemonic silence based
on whether the memory is remembered overtly or covertly, and whether the silence
is intentional (refusal to speak) or unintentional (failure to speak). So we have:
Refusing to remember overtly while remembering covertly: this “often involves
deception” (Stone et al. 2012: 41). The silence can be expressed through commis-
sion such as by lying or through omission by either not saying anything at all or by
talking about something totally unrelated to what is being omitted.
Refusing to remember overtly and covertly: this is the case where people deliberate-
ly attempt not to remember. People, too, “can effectively prevent themselves from
thinking a thought or even covertly remembering a memory” (: 41).
Failure to remember overtly while remembering covertly: where the circumstances

make overt remembering impossible. For example, in a conversation in which there
is rapid turn taking, a person may not “get a word in edgewise” (: 41).
Failure to remember overtly and covertly: “the thrust of the conversation prevents
the details from ever coming to mind” (: 41), which may happen unintentionally.
Thematic silence, then, can be deliberate in that the person remembers something
covertly but does not express it, or unintentional when the person does not mention
something because s/he has forgotten about the matter.
3. Implicit speech acts
In this chapter, I am examining whether the illocutionary force of speech acts that
would have been uttered but for thematic silence may be identified, and, if so,
whether the topic of the omitted item may be identified. Before analyzing a polit-
ical speech in which the speaker intentionally or unintentionally omitted certain
issues, let us look at several cases of speech acts whose illocutionary force may be
unclear, in that the speech act that may be intended is not explicitly uttered but may
be implied according to the circumstances. These cases may be considered more
“indirect” than Searle’s indirect speech acts (Searle 1975), since it may not be clear
at all which speech act is being uttered. Moreover, there may be several degrees
of implicitness (cf. Kecskes, this volume), as in the first example (Section 3.1), in
which we have only a report of the conversation and no quotation from the con-
versation itself.
3.1 Promise
Our first utterance is
(2) She didn’t promise him to go to the party.
The person referred to by the subject of the sentence (“she”) may have said a num-
ber of things, all of which may have pertained to the party. What she did not do
is promise. She was not silent about the promise, since there was no promise. The
non-mention of something may be called silence, but it seems to relate to assertive
speech acts (constatives) only, and not to performatives in general. That is to say,
if she did not mention the party at all, but the issue of the party was salient in the
social relationship between the two people involved, then we can say “She was
silent about the party”, i.e. non-mention of a specific assertive speech act. But, as I
have said, it is to be assumed that the subject did talk about the party, but did not
promise anything even implicitly.
3.2 Gratitude
The second utterance is
(3) It was a very useful lift you gave me yesterday. I got to the meeting on time.
The question that may be asked is whether we have an instance of thematic si-
lence in this utterance, since the speaker does not explicitly thank the addressee,
or has omitted an expression of gratitude. The speaker does not say “thank you” or
222 Dennis Kurzon
something equivalent. There is talk, but no explicit speech act of thanking. The ut-
terance, however, would probably be interpreted as an indirect speech act of thank-
ing since the tenor of the utterance, especially the word useful, may imply gratitude.
Moreover, an observer – a “ratified participant” (in Goffman 1981’s terms) or even
a bystander – would not say “He was silent about thanking her”, but rather some-
thing like “he didn’t thank her”, or if an indirect speech act is understood, then “He
sorta thanked her” (or “he sort of thanked her”). However, as Austin (1962) himself
argued, performatives may have non-verbal equivalents:
[…] it is possible to perform an act of exactly the same kind not by uttering words,
whether written or spoken, but in some other way. For example, I may in some
places effect marriage by cohabiting or I may bet with a totalisator machine by
putting a coin in the slot, instead of uttering the words “I do” or “I bet …”.
(1962: 8)
Performatives, then, may be performed in silence. But the performance of such

non-verbal performatives may be executed only by means of a conventional act or
series of acts, as in Austin’s examples (see, too, 1962: 79–80). There is no conven-
tional way of promising or thanking in silence, and for that matter swearing an oath
in silence. A witness in court putting his/her hand on the Bible is not swearing an
oath; it is only a preparatory condition, as it were, to a verbal utterance.
3.3 Silent about
The third utterance is
(4) “He is silent about the file.”
If the person referred to by the subject pronoun (“he”) has forgotten to bring a file
to a meeting, he may want to hide the fact and hope that no one will remember that
he is supposed to bring the file, or that no one will notice that he has not brought
it. In such a case, he will, of course, avoid uttering an assertive such as “I’ve left
the file at home”, but may, nevertheless, talk of many other matters. An observer
(without a doubt, a ratified participant in the context of a meeting) who knows that
he is supposed to bring the file may then say to him/herself, or quietly to someone
sitting close by: “He is silent about the file”. Therefore, if we accept the notion of
thematic silence – and this holds for English and other, but far from all, languages
(see note 1), the failure of the speaker to state that he has forgotten to bring the file
is an instance of thematic silence.
Of these three cases, it may be concluded that thematic silence occurs only in
(3), since in (1) no promise was made, nor even hinted at, according to what is said
about the conversation, and the utterance in (2) may be interpreted as an indirect
speech act of thanking. The lack of an expression of a promise as in (1) is not the-
matic silence, and neither is an implicit expression of gratitude as in (2). If such is
the case, may we say that thematic silence is limited only to assertives, as in (3)?
On the other hand, a spoken or written text may function, as a whole, as a
promise or as a refusal, for example, what van Dijk (1977) has termed a “macro
speech act”. In such a case, the explicit performative is frequently not uttered; it is
replaced by thematic silence. To illustrate this phenomenon, I shall examine in the
following subsection the speech act of refusal.
3.4 Refusals
An explicit utterance of refusal as in “I refuse to do X / say X” is a performative;

one may after all say “I hereby refuse…”, where the occurrence of hereby indicates
a performative (Kurzon 1986: 6). But what kind of performative is it? It may be
argued that a refusal is an exercitive speech act. As Marina Sbisà has explained
(personal communication), the person who refuses to do something exercises his/
her authority over his/her own conduct, thus denying the interlocutor any right
to have certain things done by him/her. It may have a weak commissive aspect,
too, because it commits the speaker not to do what s/he has refused to do. But this
is not the main feature of the act, since if someone who refused to do something
later decides to do it, this is usually regarded as less of an inconsistency than when
someone who promised to perform later decides to refrain from performing.
For Tanck (2002: 2, citing Chen 1996), however, the speech act refusing is a
speech act set, consisting of more one than one speech act. The reason for this is
that refusals are often discussed in the context of face-threatening acts (Campbell
1990, Aliakbari and Changizi 2012). They are frequently toned down to reduce
such a threat, so direct refusals such as “No!” are rarely uttered on their own (Tanck
2002). A refusal, for example, may consist of three separate speech acts:
a person who refuses may utter “an expression of regret”,
a direct refusal (e.g. “I can’t…”), followed by
an excuse.
A direct refusal, such as “No” or “I refuse”, may be considered an assertive speech

act expressing one’s intention not to do what the following infinitive clause refers
to; so, in “I refuse to come”, the speaker expresses or asserts his/her intention not
to come. Hence, to avoid a verbal act that could be interpreted as face-threatening,
not only is a direct refusal not uttered, but we may find that nothing is uttered. That
is, the refusal may be replaced by silence. This may occur in turn-taking: A asks B
224 Dennis Kurzon
a question, or requests something from B, or A says something and expects B to

respond, but in each of these cases B is silent. So, in the following example, A utters:
(5) What do you think of inviting X for dinner?
in which A is asking B (“you”) directly – A is selecting B, and the preferred an-

swer (Pomerantz 1984) would probably be some token of agreement. However, if
B does not want X to be invited, s/he would have to give a dispreferred response,
which would be in the nature of a lack of agreement with the suggestion. Some
sort of explanation might be necessary in such a case; an argument may even
ensue. Silence, then, constitutes another type of dispreferred response, which has
the added advantage for B that s/he does not have to give an explanation, at least
not at first. B’s silence has to be interpreted by A, and the interpretation would
probably be that instead of the silence, B would have uttered a speech act of refusal.
But A cannot know the reasons for the refusal because of B’s silence, unless s/he
knows B very well.
Likewise, the person who asks for a favour, and is met by silence or who invites
someone to dinner but is met by the silence of the invitee, has now to interpret the
meaning of that silence. One possibility is that the silent person did not hear, in
which case the speaker may repeat the request / invitation or whatever. Another
possibility is that the silent person wants to save face and not explicitly refuse
the request / invitation. The omission of the explicit refusal – and the subsequent
silence – may be considered to be an instance of thematic silence. In the case of re-
fusals, as discussed here, we may argue that this is a further example of an assertive
speech act being replaced by silence, since a refusal, it is argued, is an expression of
one’s intention not to do something.
Thematic silence in most cases seems to replace an assertive speech act, unless
there is a good reason to assume otherwise. In the following Section 4, a political
speech is analyzed in which the speaker has been criticized for being thematically
silent about matters central to the (then) current political debate.
4. Thematic silence in a political speech
In this section I will analyze thematic silence in one specific political speech at a
party conference in Britain. However, before we take a look at the speech, we have
to examine what kinds of speech acts are expected in a political speech – in this
particular example, a speech laying out the policy of the party some months before
the general elections.
4.1 Election pledges
A reader or hearer of a speech may have the same set of expectations from such
a political speech as one has from a party manifesto, the document produced by
parties before an election, setting out their future policies if elected to form the
government. It is popularly believed that a politician or a political party in their
pre-election manifesto would utter a series of speech acts which are interpreted as
having the illocutionary force of promise; and, it is further believed, these promises
are eventually broken if the party wins the elections and forms the government.
These promises or pledges, as they are often called, are uttered in order to attract
votes in the elections.
This cynical approach to the politician’s promises does not seem to be em-
pirically justified, however. Bara (2005) argues in her study of party pledges in
the United Kingdom between 1987 and 2001 that the situation is not as is usually
depicted. First, it is necessary to say what is mean by “pledge”. A pledge is defined
as “a specific commitment on behalf of a party to act in a certain area following a
strategy also mentioned” (Bara 2005: 587, citing Rallings 1987), and she dismisses
as pledges general statements “which fail to make any suggestion concerning out-
comes” (Bara 2005: 588). The following, from the Conservative Party manifesto of
1992, is a general statement and not a pledge: “We will increase the use of private
sector management skills”. What count are specific and detailed pledges, for exam-
ple, the following from the same Conservative Party manifesto:
We will transfer the core responsibilities of the Department of Energy to the
Department of Trade and Industry and responsibilities for energy efficiency to
the Department of the Environment, ending the need for a separate department.
This pledge not only outlines “precisely what the intention is and what action is to
be taken” (Bara 2005: 589), but gives precise information as to how this policy is to
be carried out. “Pledge specificity,” then, is “the primary basis for considering how
we might go about checking implementation” (ibid.). Bara then presents statistical
data to show that these detailed pledges are implemented or at least put on the
government agenda. In her study, of the 43 specific pledges in the period under
discussion 88% either were declared to be part of the new government’s policy
and steps would be taken to implement them or, more importantly, were actually
implemented “within a reasonable time – in this case, the life of the Parliament”
(Bara 2005: 594). These specific pledges, therefore, may be considered commissives
that tend to be fulfilled, and not vague statements of intention.
There seems, therefore, to be a large gap between the expression “election prom-
ises” or “pledges”, on the one hand, if we follow Bara’s and Rallings’ definition,
226 Dennis Kurzon
and the type of speech act that frequently occurs in political speeches, on the oth-
er. These speech acts are in the main statements of intention, but we also find a
closely connected speech act – predictions, which are also assertives, having the
same direction of fit from word to world (Searle 1976). The general message of
these statements of intention and predictions is that if the particular party wins the
elections, they will carry out what they have stated to do in these general pledges
(though, as we have seen, such is not the case). The most frequent speech acts in a
political speech, then, would have the illocutionary force of assertion: statements
of intention and predictions. However, there is obviously a link between statements
of intention and promises; both, for example, relate to a future event (or events)
and both commit the speaker to fulfill what s/he says, but in different ways. The
assertive speech act commits “the speaker (in varying degrees) to something’s be-
ing the case, to the truth of the expressed proposition” (Searle 1976: 10), while the
promise, a commissive speech act, commits the speaker “to some future course of
action” (1976: 11). In other words, a statement of intention expresses the speaker’s
current position vis-à-vis something s/he will do later on, while a promise puts the
speaker under some sense of obligation to perform a specific act or series of acts
in the future. A government may have intended to carry out a certain policy, as
expressed in their manifesto and in political speeches preceding an election, but
the current situation, such as a global financial crisis, may prevent it, and if such is
the case, not fulfilling election pledges may be generally accepted among the public.
Of course, many people simply do not see in election pledges anything that may be
interpreted as promises – despite what Bara (2005) has said.
4.2 Miliband’s party conference speech 2014
The speech I have selected in this analysis is that given by Ed Miliband, the then lead-
er of the British Labour Party at the party conference in Manchester on September
23, 2014. 2 The Labour Party subsequently lost the general elections in May 2015,
and Miliband resigned from the party leadership. Firstly, we will look at the types
of speech acts that occur explicitly in this political speech. A straightforward asser-
tion – an utterance which, it is claimed, constitutes the truth – is Miliband’s
She wasn’t just speaking for herself, she was speaking for millions of people across
our country. Millions of people who have lost faith in the future.
2. The text is found on a number of sites. Here is the one I used – from the site of the Labour
Party: http://www.labour.org.uk/blog/entry/2014-labour-conference-speech (accessed October
13, 2015).
while an assertion of intention (and not a pledge or a promise in Bara’s sense) is

seen in
this Labour Party will show you over the coming years you made the right choice.
We may often find other forms of assertions such as rhetorical questions, e.g.
Can the Tories be the answer?
I’ll tell you why they can’t be the answer, because …
in which the speaker not only asks the question but, as in this instance, may answer
it, too. A rhetorical question may be asked without it being answered, which does
not prevent the hearer from interpreting the utterance as an assertive speech act.
Here is an example from Miliband’s speech:
No wonder people have lost faith in the future. That’s why so many people voted
to break up our country. Is it any wonder? The deck is stacked. The game is rigged
in favour of those who have all the power.
in which the rhetorical question “Is it any wonder?” is a stylistic variant of the
assertive “It is no wonder”.
Apart from assertives, there are other speech acts that may be found in political
speeches. Miliband, for example, expressed gratitude to people who have helped
the speaker (and/or the party) in a number of ways. In the following, he thanks the
Scottish members of the Labour Party:
I want to start by thanking all of Labour’s Team Scotland for the part they played
in keeping our country together.
which refers to the referendum that had taken place in Scotland on September 18,
2014, one week before the party conference, in which the Scottish population voted
against independence from Britain by a vote of 55.3% against 44.7%, who voted
for independence. Ironically, it was the almost complete transfer of support among
Scottish voters from the Labour Party – despite the part they played in keeping the
country together – to the Scottish National Party that led to the disastrous results
for the Party in the elections in the following May. 3
At the centre of Miliband’s speech were the six national goals he spelled out to
the Party and to the British voters in the coming elections. It is here where I will
focus on thematic silence. These national goals constitute, he claims, “a plan for the
3. The switch of votes did not lead to Milliband’s defeat – he would not have been able to get a
majority of votes in the country as a whole in any case – but to the loss of 40 seats (out of 41 in
the previous elections in 2010) from Scottish constituencies. The party ended up with an overall
loss of 26 seats in Parliament.
228 Dennis Kurzon
next ten years. Britain 2025”. In terms of the criteria set down by Rallings (1987)
and Bara (2005), as discussed above, these goals are rather vague statements of
intention (assertive speech acts) that may be implemented if the Labour Party win
the following elections. The use of the modal may here first of all foreshadows the
following conditional clause; secondly, even if Labour had won the elections the
goals do not contain any hint of how the policy would be implemented. This can
be seen clearly in the wording of each of the goals.
Miliband’s first goal is
that we halve the number of people in low pay by 2025. Transforming the lives of
two million people in our country.
That this is not a pledge may be seen from the lack of information as to how the
goal will be achieved. How will a government under Miliband halve the number of
people on low pay? In terms of basic definitions, he does not mention how much
is “low pay”; nor does he spell out how many people are on low pay: are there
two million or four million, i.e. Miliband said that the government will halve the
number – is that from four to two million? Without going into details, the same
vagueness may be seen in the other five goals.
The second national goal is that
all working people should share fairly in the growing wealth of the country. That
means, as the economy grows, the wages of everyday working people grow at the
same rate.
And the third national goal:

by 2025, Britain becomes truly a world leader in the green economy, creating one
million new jobs as we do. Under this government, we’re falling behind Germany,
Japan, the United States and even India and China when it comes to green tech-
nologies and services.
Fourthly,
by 2025 as many young people will be leaving school or college to go on to an
apprenticeship as currently go to university.
Fifthly,
by 2025, for the first time in fifty years, this country will be building as many homes
as we need. Doubling the number of first time buyers in our country.
Lastly, the Labour Party in office will create – by 2025 – “a truly world-class 21st
century health and care service”. It may be assumed that if the Labour Party had
won the elections, they would have tried to pass and implement laws which would
hopefully lead to a situation in which the above-mentioned goals be achieved.

However, as discussed above, these goals are in effect general statements that can-
not be considered pledges (Rallings 1987). These statements are, then, assertives
and not commissives. Furthermore, had the Labour Party won, they could have
guaranteed appropriate policies only until 2020, for the life of a Parliament is a
maximum of five years (since the Fixed Terms Parliaments Act of 2011), and not
till 2025. To sum up, political parties put forward manifestos declaring what they
hope to achieve – these are assertives, and not commissives.
4.3 Thematic silence in Miliband’s speech
But it was noted by the press after Miliband’s speech that two important issues
had been omitted. He was silent about immigration and about the budget deficit.
Carole Walker of the BBC, reporting from Manchester, where the conference took
place, said that
Labour leader Ed Miliband failed to mention immigration or the deficit in his
speech because he forgot… Mr Miliband delivered the speech mainly from mem-
ory, without the help of an autocue and using only basic notes.
(BBC, Sept. 24, 2014)
The Guardian, a day later, quotes Nick Clegg, at that time the deputy prime minister
from the Liberal-Democratic Party, the junior coalition partners in Prime Minister
David Cameron’s government, who accuses
Ed Miliband of ‘choosing to forget’ the deficit in his conference speech as he belit-
tled Labour’s offer of an extra £2.5 billion annually to the National Health Service.
(Guardian, Sept. 25, 2014)
Using Stone et al. (2012)’s model (see Section 2 above), we may say that Clegg
saw in this omission a case of “refusing to remember overtly while remembering
covertly”, which, they add, “often involves deception” (2012: 41). However, in this
case, if we follow Clegg, it is not deception that is involved here but a lack of policy
on these issues.
The New Statesman of September 2014 also spoke of what the opponents to the
Labour Party (“the Tories”, i.e. the Conservative Party – the major coalition partner)
had said about the speech:
After Ed Miliband had finished his conference speech, he was immediately at-
tacked by the Tories for failing to mention the deficit.… [T]he original text … did
feature a section on the deficit, as well as a longer one on immigration (which was
mentioned just once).
230 Dennis Kurzon
The reporter defends Miliband, for in “a 65-minute, no-notes performance it is

perhaps inevitable that Miliband will forget some passages.” Hence, in Stone et al.’s
terms, this is a “failure to remember overtly and covertly”, which may be uninten-
tional, and not an intentional omission. The speech as given at the conference did
mention immigration, as the New Statesman suggests:
We need to reform Europe. We need to reform Europe on the economy, on immi-
gration, on benefits, on all of these big issues. (my emphasis)
The magazine then quotes what Miliband had forgotten when he gave his speech
without any notes. The first two sentences he remembered. It is the subsequent
three sentences including one question which he answered by a rhetorical question
(which is an assertive speech act; see above) that he had forgotten:
We need to reform Europe. We need to reform Europe on the economy, on immi-
gration, on benefits, on all of these big issues. But here is the question for Britain.
How do we reform Europe? Do we reform Europe by building alliances or by
burning alliances? (New Statesman, September 2014)
In the speech as given and in the intended speech, Miliband mentioned immi-
gration in the context of Europe and the European Union. Its relevance to Britain
seems marginal, a position which is questionable today following the vast wave of
African and Middle East migrants banging on Europe’s doors.
Broadly speaking, we may say that Miliband was silent about the two issues of
immigration and of the budget deficit he had proposed was needed to implement
the party’s policies. This is thematic silence, and like the six goals, we are talk-
ing about statements of intention – assertive speech acts. This is the case whether
Miliband, in Stone et al. (2012)’s model, failed “to remember overtly while remem-
bering covertly”, or failed “to remember overtly and covertly”. In the first case, what
is implied is that he was silent about these issues, since he did not have any solution
to the pressing problems. The second case, on the other hand, implies that he had
simply forgotten what he had prepared to say. Government spokespersons would
interpret Miliband’s thematic silence in terms of the first case, while Miliband sup-
porters would accept the second case. Observers of a case of thematic silence who
are neutral in their attitude to the speaker (e.g. perhaps foreign correspondents)
face an ambiguity, which may be considered a frequent and natural response to
a person’s silence about a matter. It may be only those observers who have their
own agenda, such as attitudes held by opposing political groups in our case, who
would publicly suggest a deliberate omission, and interpret the silence as intentional
thematic silence.
5. Conclusion
On the basis of the analysis of Ed Miliband’s speech in Section 4, we may say that the
typical speech acts found in political speeches are statements of intention and pre-
dictions (assertives). Promises (usually called pledges) also occur, but only if they
are specific and detailed, not vague statements of intention, as are the six national
goals in Miliband’s party speech. We may also find expressions of gratitude. Other
types of speech acts are of course possible, according to the circumstances, but the
primary illocutionary force is that of assertion. Since expressions of gratitude and
promises may be implicit, as shown in Section 3 above, we find that appropriate
speech act verbs or performatives are omitted, but if it is clear from the context
what speech act is being performed (often, in such cases, as indirect speech acts),
we are not talking about thematic silence.
Thematic silence in a political speech, then, is usually a substitute for statements
of intention and predictions – that is to say assertive speech acts – about topics that
the speaker either has forgotten, that is a case of “failing to remember overtly and
covertly” (Stone et al. 2012: 41), or, secondly, has deliberately left out in order not to
embarrass him-/herself, or, thirdly, has nothing to say on the matter (both these last
two are cases of “refusing to remember overtly while remembering covertly”; ibid.).
It may be further argued that not only things that are said implicitly but also
topics that are deliberately omitted may secure uptake, in Austin’s terms, among the
hearers. These cases of thematic silence are, then, substitutes for speech acts, and
it is the text-type, in our case a political speech, that determines to a large extent
which speech acts are expected to be found.
References
Aliakbari, Mohammad, and Mahsa Changizi. 2012. “On the Realization of Refusal Strategies
by Persian and Kurdish Speakers”. International Journal of Intercultural Relations 36: 659–
668. doi: 10.1016/j.ijintrel.2012.04.009
Austin, John L. 1962. How to Do Things with Words. Oxford: Oxford University Press.
Bara, Judith. 2005. “A Question of Trust: Implementing Party Manifestos.” Parliamentary Affairs
58: 585–599. doi: 10.1093/pa/gsi053
Berger, Charles R. 2004. “Speechlessness: Causal Attributions, Emotional Features and Social
Consequences.” Journal of Language and Social Psychology 23: 147–179.
doi: 10.1177/0261927X04263821
Campbell, Kim. 1990. “Explanations in Negative Messages. More Insight from Speech Act
Theory.” Journal of Business Communication 27: 357–375. doi: 10.1177/002194369002700403
Chen, Hongyin J. 1996. “Cross-cultural Comparison of English and Chinese Metapragmatics
in refusal.” Indiana University Manuscript, ERIC Document Reproduction Service No. ED
408 860.
232 Dennis Kurzon
Goffman, Erving, 1981. Forms of Talk. Philadelphia: University of Pennsylvania Press.

Kurzon, Dennis. 1986. It Is Hereby Performed…: Explorations in Legal Speech Acts. Amsterdam:
John Benjamins. doi: 10.1075/pb.vii.6
Kurzon, Dennis. 2007. “Towards a Typology of Silence.” Journal of Pragmatics 39: 1673–1688.
doi: 10.1016/j.pragma.2007.07.003
Kurzon, Dennis. 2011. “Moment of Silence: Constitutional Transparency and Judicial Control.”
International Journal for the Semiotics of Law 29: 195–209. doi: 10.1007/s11196-010-9205-6
Pomerantz, Anita. 1984. “Pursuing a Response.” In Structure of Social Action, ed. by J. Maxwell
Atkinson, and John Heritage, 152–163. Cambridge: Cambridge University Press.
Rallings, Colin. 1987. “The Influence of Election Programmes: Britain and Canada 1956–1979.”
In Ideology, Strategy and Party Changes: Spatial Analyses of Post-War Election Programmes
in 19 Democracies, ed. by Ian Budge, David Robertson, and David Hearl, 1–14. Cambridge:
Sacks, Harvey, Emanuel A. Schegloff, and Gail Jefferson. 1974. “A Simplest Systematic for the
Organization of Turn-taking for Conversation.” Language, 50: 696–735.
doi: 10.1353/lan.1974.0010
Searle, John R. 1975. “Indirect Speech Acts.” In Syntax and Semantics 3: Speech Acts, ed. by Peter
Cole, and Jerry L. Morgan, 59–82. New York: Academic Press.
Searle, John R. 1976. “A Classification of Illocutionary Acts.” Language in Society, 5: 1–23.
doi: 10.1017/S0047404500006837
Stone, Charles B., Alin Coman, Adam D. Brown, Jonathan Koppel, and William Hirst. 2012.
“Toward a Science of Silence: The Consequences of Leaving a Memory Unsaid.” Perspectives
on Psychological Science 7: 39–53. doi: 10.1177/1745691611427303
Tanck, Sharyl. 2002. “Speech Acts Sets of Refusal and Complaints: A Comparison of Native and
Non-native English Speakers’ Production.” TESOL Working Papers (http://auislandora.wrlc.
org/islandora/object/tesolworkingpapers%3A26; accessed September 1, 2015)
Van Dijk, Teun. 1977. Text and Context. London: Longman.
Part III
Text and discourse

Chapter 11
The dynamics of discourse

Quantity meets quality
Anita Fetzer
University of Augsburg
This chapter argues for a dynamic perspective on discourse, delimiting dis-

course from context on the one hand, and from arbitrarily concatenated
discourse units on the other. Addressing the question of granularity in an
explicit manner, it argues for a relational conception of the constitutive parts
of discourse, analogously to the functional-grammar-based differentiation be-
tween clausal and extra-clausal constituents. The former present propositional
meaning contributing to local and not-so-local argument and storyline, and
the latter encode procedural meaning, in particular metadiscursive linkages.
A relational conception of discourse unit comprises ordinary discursive units,
e.g. single and complex clauses and utterances, or speech acts, conversation-
al contributions, but also larger units, e.g. paragraph / episode or opening,
closing and topical sections, or speech act complexes. Metadiscursive units
comprise prototypical extra-clausal constituents, e.g. comment clauses and
discourse markers. The delimiting frame of discourse genre does not only
constrain the structuring of discourse as regards the patterned linearization of
discourse units, but also the degrees of implicitness/overtness for the linguistic
realization of discourse units and the signalling of discourse relations and oth-
er coherence strands.
Keywords: context, discourse, discourse pragmatics, discourse unit, discourse

relations, parts-whole perspective
1. Introduction
Context has played a prominent role in the analysis of meaning, accounting for the
differentiation between context-dependent meaning in pragmatics and context-
independent meaning in semantics. Context-dependent meaning may further be
distinguished with respect to more generalized and more particularized types of
meaning, with the former anchored in default contexts and assigned the status of
doi 10.1075/pbns.276.11fet
236 Anita Fetzer
being more direct, immediate or salient, and the latter anchored in particularized
contexts and assigned the status of being more indirect and less immediate, as is
captured by the differentiation between different types of generalized conversational
implicatures and particularized conversational implicatures (e.g., Grice 1975; Horn
1984; Levinson 2000), or by relevance-theoretic approaches to meaning as well as by
lexical pragmatics (e.g., Carston 2002; Clark 2013; Wilson and Sperber 2012). More
recently, pragmatic approaches to discourse have argued for a more explicit differenti-
ation between context on the one hand, and discourse on the other (e.g. Fetzer 2013).
Discourse is – like context – used in diverging frameworks, where it refers to
different theoretical constructs defined in accordance with frame-work-specific
premises. For instance, discourse is used synonymously with text, a linguistic
surface phenomenon, denoting longer stretches of written or spoken language.
Discourse is also frequently used to refer to both a theoretical construct and its
instantiation in context, i.e. type and token. While there is general agreement
about the quantity of discourse, that is “language patterns above the sentence“
(Widdowson 2004: 3), there is still quite some controversy about its qualitative
status, as is reflected in discourse semantics and discourse pragmatics. However,
the nature of the connectedness between the quantity and quality of discourse as
regards the expression and interpretation of discursive meaning contained in “lan-
guage patterns above the sentence”, as well as the size of “language patterns above
the sentence”, in particular the basic discourse unit and the question of whether
it is fuzzy or discrete have not yet been addressed comprehensively. Widdowson
(2004) himself qualifies his rather general definition of discourse cited above, mak-
ing explicit possible implications and arguing that the definition “would seem to
imply that discourse is sentence writ large: quantitatively different but qualitatively
the same phenomenon. It would follow, too, of course, that you cannot have dis-
course below the sentence” (Widdowson 2004: 3; original emphasis). And there is
yet another fallacy in the purely quantitative definition: if “the difference between
sentence and discourse is not a matter of kind but only of degree, then they are
presumably assumed to signal the same kind of meaning. If sentence meaning is
intrinsically encoded, that is to say, a semantic property of the language itself, then
so is discourse meaning” (ibid.). To accommodate not only quantity and quality, but
also their interdependencies and interactions, a felicitous analysis of discourse and
discourse meaning needs to go beyond the code model of language and accommo-
date the pragmatic premises of rationality, intentionality of communicative action
and cooperation, as well as the non-trivial truism that the whole, that is discourse
delimited by discourse genre, is more than the sum of its constitutive parts, that is
discourse units with force, content and metadiscursive meaning. This is apparent
in the meaning of the whole, which is more than the sum of the meanings of its
separate parts. A pragmatic analysis of discourse thus “has to do not with what texts
Chapter 11. The dynamics of discourse 237
mean, but with what might be meant by them, and what they are taken to mean”
(Widdowson 2004: 35).
This chapter examines the dynamics of discourse considering (1) the structur-
ing of discourse as captured by the granularity of discourse units and their lineariza-
tion, (2) the linguistic realization and interpretation of discourse units, and (3) the
connectedness between discourse unit and meaning constrained by adjacency and
dovetailedness, contributing to the construal of discourse coherence. It is organized
as follows: Section two will deal with the discourse unit, which is understood as
relational and doubly contextual. It will also approach granularity by comparing
discourse units in text linguistics and discourse analysis. Section three focuses on
discourse relations which are conceived of as dovetailed and multiply discursive,
and it examines the theory and practice of discourse relations, focussing in par-
ticular on their linguistic realization in context with respect to varying degrees of
explicitness. Section four examines pragmatic discourse and discourse pragmatics,
and concentrates on the interdependencies of quantity and quality, relating the lin-
guistic realization of discourse relations with the cueing of coherence strands and
the construal of discourse coherence. Section five explores discourse pragmatics
in action, and illustrates the underlying frame of reference with respect to micro,
meso and macro discourse units. Section six serves as a conclusion, summarizing
the results obtained and weighing the results.
2. Discourse unit: Relational and doubly contextual
The analysis of discourse, also referred to as longer stretches of talk, has utilized
various units of investigation, such as simple and complex clause or sentence, which
have been adopted from diverging grammatical frameworks; proposition from se-
mantics; turn-constructional unit, turn or sequence from ethnomethodological
conversation analysis, or simply utterance from usage-based frameworks.
In text linguistics (e.g., de Beaugrande and Dressler 1981) the syntactic unit of
sentence counts as a discourse unit, while discourse is delimited by text-types and
classified accordingly. In functional discourse grammar (e.g., Givón 1993; Halliday
1994; Martin and Rose 2008) the syntactic unit of clause is the unit of investiga-
tion and discourse is delimited and framed by episodes or genre, for instance.
Discourse semantics considers the semantic unit of proposition as its unit of anal-
ysis, while a concatenated sequence of propositions is seen as a delimiting frame.
More dynamic models also integrate illocutionary force (e.g., Asher and Lascarides
2003; Moeschler 2002; Roulet 1991, 2006) and use speech act, proposition and
utterance as their unit of analysis, and concatenated units as delimiting frames.
In spite of methodological differences, approaches to discourse do not only share
238 Anita Fetzer
a quantitative definition of quantifiable discourse units as is captured by clause,

sentence, utterance, turn-constructional unit and turn, but also some kind of de-
limiting frame, that is a discourse genre or adjacent units, and thus the implied
premise that discourse is a parts-whole configuration in which the whole is more
than the sum of its constitutive parts (cf. Fetzer 2014).
Quantitatively oriented studies tend to focus on the linearization of discourse
units as well as on the quality of their connectedness, while qualitatively oriented
discourse studies share the assumption that discourse as a linearized whole of con-
catenated units comes in with the presumption of being coherent (cf. Bublitz, Lenk
and Ventola 1999; Gernsbacher and Givón 1995; Chafe 1994). In qualitative studies
it is not “language patterns above the sentence” and their semantic wellformedness,
which make them cohere but rather participants who negotiate the meaning of
discourse units and of discourse-as-a-whole, thereby construing and negotiating
discourse coherence. Hence, discourse coherence does not lie in the discourse itself
but rather in participants’ minds and therefore is a socio-cognitive construct. This
view is also implicit in cohesion-based analyses of texture (e.g., Hasan and Halliday
1987), in which discourse coherence is connected intrinsically with cohesion and
cohesive ties, that is linguistic items which signal, if not encode, the nature of the
connectedness between the constitutive parts and the whole.
The structuring of discourse, or of conversation, has been been examined in
ethnomethodological conversation analysis and in interactional sociolinguistics.
Both subscribe to the premise of indexicality of communicative action, and thus
are appropriate frames of reference for examining the connectedness between
discourse units and discourse-as-a-whole: “Sequential organization refers to that
property of interaction by virtue of which what is said at any time sets up expec-
tations about what is to follow either immediately afterwards or later in the inter-
action” (Gumperz 1992: 304). The linearization of discourse, that is the sequential
organization of its constitutive parts, is connected intrinsically with the premise
that discourse-as-a-whole is coherent and that the connectedness between the
constitutive parts contributes to the construal of both local and global coherence.
Linearization is thus constrained by adjacency, i.e. adjacency position, adjacency
relation and adjacency expectation, and by dovetailedness (Fetzer and Speyer 2012;
Speyer and Fetzer 2014), as well as by the delimiting frame of discourse genre. In
functional grammar, the construal of discourse coherence is based on coherence
strands, that is referential continuity, temporal continuity, spatial continuity and
action continuity (Givón 1993), and the explicit accommodation of discourse (or
coherence) relations adds more discursive glue. Discourse relations may be sig-
nalled explicitly with discourse markers 1 and other metadiscursive devices, they
1. Discourse marker is used as an umbrella term including pragmatic markers and discourse
connectives, to name but the most prominent ones (cf. Fetzer 2012).
may be signalled implicitly with indexical references to discourse-relation-specific

coherence strands, and they may be signalled in a combined manner (cf. Maier,
Hofmockel and Fetzer, 2016).
Discourse units are relational from both discourse-structuring and dis-
course-meaning perspectives. Adapting the conversation-analytic concept of dou-
bly contextual (Heritage 1984) to monologic, dialogic and multi-party discourse,
adjacently positioned discourse units are doubly contextual in as far as they provide
linguistic context for the production and interpretation of neighbouring discourse
units. But it is not only linguistic context, which embeds a discourse unit. Linguistic
context is functionally equivalent to the linguistic realization of the participants’
communicative intentions and therefore contains references to both the partici-
pants’ cognitive contexts, i.e. mental representations and common ground, and to
the social and sociocultural contexts of the discourse, which are also indexed in
the linguistic realization of a discourse unit.
The dynamics of discourse can only be captured if the fundamental pragmatic
premises of rationality, intentionality of communicative action and cooperation as
well as felicity conditions are adapted to discursive linearization. This is because
the sequencing of discourse makes manifest the discourse units’ (in Sbisà’s terms
“moves”) perlocutionary effects: “When considering a sequence of moves, it is rea-
sonable to view the output of one move as coinciding with the input for the next”
(2002a: 72). Bach goes further by connecting micro, meso and macro domains – in
structuring-of-discourse terms: micro discourse units, meso discourse units and
macro discourse units – with respect to different types of intention: “communicative
(illocutionary) intentions generally are accompanied by perlocutionary intentions,
and individual utterances [micro discourse units, A. F.] are usually parts of larger
plans [meso and macro discourse units, A. F.]. So it is plausible to suppose that iden-
tifying a speaker’s perlocutionary intentions and broader plans [meso and macro
discourse units, A. F.] is often relevant to identifying his communicative intention”
(Bach 1992: 397). Against this background, a relational conceptualization of dis-
course unit does not only need to accommodate content, but also force and meta-
discursive meaning. The metadiscursive meaning expressed by micro discourse
units may comprise “continue” with further particularizations, such as “elaborate”,
“exemplify”, or “contrast” with further particularizations, e.g., “reformulate”, “ac-
cept partially” or “change”. Meso discourse units refer to sequences with force,
content, and metadiscursive meaning, for example “continuative sequence with
more evidence”, “expansion of argumentative sequence” or “contrastive sequence
with counter arguments”, “counter-argumentative sequence with counter-counter
arguments”. Macro discourse units refer to discourse genre with force, content and
metadiscursive meaning.
Perlocutionary intentions are also inherent in Austin’s conception of per-
locutionary act, which manifests itself in the “achievement of a perlocutionary
240 Anita Fetzer
object (convince, persuade) or the production of a perlocutionary sequel” (Austin

1976: 181). In discourse-unit-based terminology, the Austinian conception of
speech act does not only have force and content, but also metadiscursive meaning,
which is reflected in the reference to some kind of continuation, connectedness,
series or sequence. This can be interpreted as a requirement to connect a discourse
unit with adjacent discourse units, and possibly with other more remote ones,
bringing about the understanding of the content, force and metadiscursive meaning
of the discourse unit, thus contributing to the construal of discourse coherence, as
is made explicit in the coherence principle (Mey 2001), which goes beyond textual
coherence, including coherence with respect to pragmatic presuppositions, illocu-
tionary intentions and perlocutionary intentions.
The effects of discourse units thus need to be considered explicitly with re-
spect to cognitive effects, i.e. the recipient’s recognition of meaning and force, the
construal of discourse common ground 2 Fetzer (2007c) and the construal of in-
tersubjective reality, and with respect to social effects, i.e. discourse expectations,
and rights and obligations of particular discourse units and their felicity condi-
tions. However, it is not only discourse units that are situated in context, but also
the context itself situates and conditions discourse units. This is particularly true
for discursively implicated meaning, which is what the context makes it to be.
Conversely, a discourse unit may create the context for which it is appropriate (cf.
Mey 2011), as is also argued for by Levinson (1983:293):
What makes some utterances after a question constitute an answer is not only the
nature of the utterance itself but also the fact that it occurs after a question with a
particular content – ‘answerhood’ is a complex property composed of sequential
location and topical coherence across two utterances, amongst other things; sig-
nificantly there is no proposed illocutionary force of answering.
The linearization of monologic and dialogic discourse is thus a multilayered, com-

plex endeavour. It is based on communicative intentionality, on the strategic use
of language constrained by the linguistic system, and on participants acting in
2. Discourse common ground is an interlocutor-, context- and genre-dependent variant of com-

mon ground. It is anchored in a network structure and connected with other types of discourse
common ground. The network structure is functionally equivalent to Background (Searle 2010).
Discourse common ground is composed of mental representations, propositions, and factual and
contextual assumptions, which may vary in strength. It undergoes continuous updating and con-
tinuous re-organisation as assumptions are read, written and deleted, and contextual implications
are raised in strength, lowered in strength or erased (cf. Fetzer 2004, 2007c). Changes resulting
from the administration of emergent discourse common ground may result in changes of other,
higher-level discourse common grounds.
accordance – and they may locally also act in disaccordance – with the contextual
constraints and requirements of discourse genre. The sequential organization and
linearization of discourse is not only a linguistic-surface phenomenon, but rather
depends on the socio-cognitive construct of discourse common ground, which is
updated and administered continuously. Discourse common ground is – like dis-
course – a dynamic construct, which is negotiated and updated continuously, i.e.
confirmed, modified or restructured, by storing new information and by updating
already stored information, which may require the restructuring of the partici-
pants’ individual and collective discourse common grounds. Individual discourse
common ground administers an individual’s personal administration of discourse
common ground, while collective discourse common ground administers negoti-
ated and ratified discourse common grounds; both may diverge to varying degrees
(Fetzer 2007c). Against this background, discourse unit and discourse are collective
concepts, accommodating speaker and hearer, the set of speaker and hearer, as well
as all other potential participants. The production and interpretation of discourse
units as well as their sequential organization and linearization are always goal- and
participant-directed, and thus a collective endeavour.
3. Discourse relations: Dovetailed and multiply discursive
The structuring and linearization of discourse is connected intrinsically with the

question of granularity, i.e. size and conceptualization of discourse units, and with
the semantics and pragmatics of their connectedness. The former has been ad-
dressed in the previous section in which discourse units have been described as
containing force, content and metadiscursive meaning. Furthermore, discourse
units may vary in size and be constituted of minimal discourse units, such as com-
ment clauses, discourse markers or elliptical constructions, micro discourse units,
such as clauses, utterances or conversational contributions, or meso discourse
units, such as clause complexes, paragraphs or larger units. Discourse is delimited
by discourse genre, the macro discourse unit. Discourse units are thus relational
constructs. Discourse units have also been described as doubly contextual, as is
reflected in linearization and structural adjacency as well as in adjacency relation
and adjacency expectations, as is captured by dovetailedness.
The discursive constraint of dovetailedness, as put forward in logic and conver-
sation (Grice 1975) provides the necessary tool to account for the connectedness
between discourse units and discourse-as-a-whole. Grice specifies the constraint for
the unit of conversational contribution as “such as is required, at the stage at which
it occurs, by the accepted purpose or direction of the talk exchange [the lineariza-
tion of discourse, A. F.] in which you are engaged” (Grice 1975: 45), implying that
242 Anita Fetzer
conversational contributions are linked by one or more common goals manifest in

prior and succeeding contributions. In discourse, conversational contributions have
the status of a discourse unit, which may be composed of smaller discourse units,
such as minimal discourse units. The discursive constraint of dovetailedness, this
chapter argues, also holds for minimal discourse units as well as for more complex
discourse units, such as sequences, as well as for discourse-genre-as-a-whole.
Dovetailedness is both semantic and pragmatic. It is implicit in the conversa-
tion-analytic conditional relevance and in the two-part sequence of adjacency pair,
which, following Mey, “is a case of coherent sequencing, but not all sequencing
needs to be defined strictly in terms of adjacency” (Mey 2001: 249). Dovetailedness
is fundamental to the construal of discourse coherence, which does not mean that
it is meaning-based only. Dovetailedness refers to two sides of a coin, metaphori-
cally speaking. On the one hand, it refers to structural adjacency, and on the other
to adjacency relation and adjacency expectation. Should adjacency relation and
adjacency expectation conflate, structural adjacency also holds, as is the case with
adjacency pairs, that is patterned co-occurrences of two social actions produced by
different speakers, such as greeting and greeting; request and compliance/non-com-
pliance; offer or invite and acceptance/refusal; assessment and agreement /disa-
greement; and question and expected answer / unexpected answer or non-answer
(cf. Levinson 1983: 336). The second parts of the adjacency pairs just listed are not
of equal standing, as one of them is preferred, and the other is dispreferred, as has
been examined in the framework of preference organization. The classification as
preferred and dispreferred second is not based on the interlocutors’ psychological
disposition, but rather on structural and distributional features and hence closely
connected with the linguistic concept of markedness (cf. Levinson 1983: 307).
Adjacency relation goes beyond structure-based positioning. It is a pragmatic
concept, which may be encoded in discourse and thus made explicit, or it may be
assigned a presuppositional status and thus needs to be inferred. Adjacency relation
may have a narrow scope and be assigned the status of a local constraint, as is the
case with adjacency pairs and their preferred and dispreferred seconds, or it may
have a wider scope and be assigned the status of a less-local constraint, as is the
case with insertion sequences and topical digression, and pre- and post-sequences
in conversation. Closely related to the concept of adjacency relation is the notion
of adjacency expectation.
The cognitive concept of adjacency expectation is a discourse notion par ex-
cellence. It is the foundation against which two adjacent discourse units may be
classified as a particular adjacency pair with preferred and dispreferred seconds,
or against which the second social action may be assigned the status of the first
unit of an insertion sequence. For instance, in the activity type of interview so-
cial actions performed by the interviewer tend to count as questions and social
actions performed by the interviewee tend to count as answers. Should an inter-

viewee opt for the social action of asking a question, he or she needs to refer to the
social-action format of question in an explicit manner, e.g. by saying “may I ask
you a question” (cf. Fetzer 2000). Adjacency does not only comprise the conver-
sation-analytic conception of adjacency holding between turns, that is adjacency
pair / position / relation / expectation. It may also refer to the turn-internal organ-
ization or discourse-internal sequential organization of discourse units (or parts),
and of discourse units and discourse as a whole. In discourse, adjacency position
may conflate with adjacency relation and with adjacency expectation, as has been
shown above. However, it is also possible that it neither conflates with adjacency
relation nor with adjacency expectation. In that case, a conversational implicature
is triggered and the nature of the connectedness between the adjacently positioned
discourse units is inferred.
A particular type of adjacency relation is the discourse (or coherence) relation,
which holds between two discourse units. Discourse relations have been defined
in the discourse semantic framework of Segmented Discourse Representation
Theory (Asher and Lascarides 2003), which analyzes the logical relation between
two discourse segments, which refers to one particular type of discourse unit, i.e.
a complex linguistic unit with propositional content and illocutionary force of its
own. Any discourse segment p2 usually stands in a logical relation to at least one
other preceding segment p1 (or rather: the addressee construes a logical relation
between them, in order to vouchsafe coherence). The propositions p1 and p2 are
in the discourse relation R if the inferences the addressee makes and the logical
connection s/he draws between p1 and p2 are in accordance with the ones de-
fined for R. As discourse is not a purely linear phenomenon, but is hierarchically
structured, Segmented Discourse Representation Theory distinguishes between
two kinds of discourse relations: coordinating relations that keep the discourse
on the same level, and subordinating relations that introduce a lower level in
the discourse hierarchy. Any discourse relation can hold between two adjacently
positioned discourse segments, or between two discourse segments that are not
adjacently positioned, i.e. where adjacency position, adjacency relation and adja-
cency expectation do not conflate.
Asher and Lascarides (2003) define a number of discourse relations. Examples
of coordinating relations are Continuation (p2 shares a common topic with p1),
Narration (Continuation plus temporal sequence between p1 and p2), and Contrast
(p2 offers an alternative to p1 or a part of p1), for instance. Examples of subordi-
nating relations are Elaboration (p2 offers additional information about one of the
referents in p1), Explanation (p2 offers a reason for p1 or parts of p1) and Comment
(p2 takes p1 as topic). The definitions of discourse relations in Segmented Discourse
Representation Theory are based on the unit of discourse segment, and need to be
244 Anita Fetzer
adapted to the relational definition of discourse units promoted in this chapter, as

it excludes minimal discourse units, such as comment clauses and discourse mark-
ers, 3 and does not explicitly include meso discourse units either. The inclusion of
larger discourse units, which may be constructed in the processing of discourse, and
assigned the status of one expanded unit with its own recontextualized illocutionary
force, would, however, be possible.
To adapt the definitions of Segmented Discourse Representation Theory to a
dynamic theory of discourse and to its linguistic realization, the coherence strands
of referential continuity, temporal continuity, spatial continuity and action con-
tinuity are supplemented by lexical coherence. This makes it possible to draw a
connection between Segmented Discourse Representation Theory and its defining
conditions for discourse relations on one hand, and their linguistic realization in
discourse in line with Givón’s definition of coherence strands as particularized
features on the other:
These strands are clearly the most concrete, salient, observable links between claus-
es in coherent discourse. But the phenomenon of discourse coherence is richer
yet. First, coherence strands may connect – or ground [original emphasis] – the
clause either to the current text, to the current speech situation, or to generic-lexical
knowledge. Second, coherence strands may extend either locally, between adjacent
clauses, or globally, across larger text-structures. Third, coherence strands may be
either semantic or pragmatic in nature. Finally, the strands may ground the clause
in either an anaphoric or a cataphoric direction” (Givón 1993: 287)
Coherence strands account explicitly for the coherence-construing function of

clausal constituents and clauses and are, for this reason, indispensible for a gram-
mar of discourse. Furthermore, they also allow for the context-dependent analysis
of multiply discursive discourse units or of their constitutive parts, for instance
of the 1st-person-singular cognitive-verb based constructions I think and I mean.
If used as parentheticals and thus as discourse markers constituting a minimal
discourse unit, the former grounds the speaker’s attitude and signals subjectifi-
cation, while the latter signals that a more appropriately formulated content or
illocutionary force is forthcoming, thus expressing both anaphoric and cataphoric
reference. If used as matrix clause, they ground the speaker’s cognitive process for
I think, and express some change in one or more already grounded clauses or text
for I mean (Fetzer 2014).
3. Metadiscursive comments have been reanalyzed as the speech act type of expositive.
Expositives have illocutionary force and a propositional content with a metarepresentational
status (cf. Fetzer 2013, Oishi and Fetzer 2016).
Coherence strands are made manifest through (a) topic continuity, i.e. topic
identity or specification, (b) tense and aspectual coherence (including modality),
(c) lexical coherence, in particular in lexical relations (antonymy etc.) and meta-
comments, and (d) default grammatical word order vs. pragmatic word order. A
systematic analysis of coherence strands may not only explain higher or lower
degrees of glueyness and thus of discourse coherence, but also predict syntactic
formatting, which is relevant to the linguistic realization of discourse units: “The
more thematically connected a conjoined clause is with an adjacent clause – the
more strands of thematic coherence it shares with that adjacent clause – the more
likely it is to appear reduced, less finite, syntactically integrated with that other
clause” (Givón 1993: 318).
The instantiation of discourse relations and the number and kind of indexical
references to defining conditions and particularized features depends on context,
and not all discourse relations may be fully specified in a discourse. Those defin-
ing conditions and particularized features indexically referred to in discourse are
assigned the status of discursive glue. Since full specification of discourse relations
in a discourse is not a necessary condition (cf. Maier, Hofmockel and Fetzer, 2016),
their linguistic realization allows for multiply discursive functions. For instance,
the coordinating discourse relations of Continuation and Narration share almost
the same particularized features and one defining condition, but Narration re-
quires further temporal constraints, 4 which may not always be made explicit in
their linguistic realization at a particular stage in a discourse. A similar scenario
holds for the subordinating discourse relations of Elaboration and Explanation with
the latter requiring the accommodation of further constraints, providing reasons
and temporal consequences. If their linguistic realizations do not explicitly refer
to “reason”, for instance with a causal discourse marker, their linguistic realization
may allow for multiply discursive assignments.
Analogously to the degree of explicitness in the linguistic realization of com-
municative actions in context, the linguistic realization of discourse relations allows
for varying degrees of explicitness and thus for multiply discursive assignments.
The degree of explicitness for the realization of discourse relations does not only
depend on context and discourse genre, but also on different languages, as has been
shown for German and English for the discourse units of sentence and clause (cf.
Speyer and Fetzer 2014).
The indexing of discourse relations and their linguistic realization as regards
the degree of “glueyness” depends on the semantics of the discourse relation, the
4. Continuation may compete with Narration as well as with Result, if they are not sufficiently
indexed, corroborating Levinson’s (2000: 122) claim that “when events are conjoined, they tend
to be read as temporally successive, and, if at all possible, as causally connected”.
246 Anita Fetzer
local context, and the discourse genre. Discourse relations may be fully specified
by indexical reference to all of their defining condition(s) and to all of their par-
ticularized features and thus have a higher degree of “glueyness”, and they may
even be overspecified if a discourse connective is added to their fully specified
linguistic realization. They may also be underspecified by indexical reference to
their defining condition(s) and/or to some particularized features and thus have a
lower degree of “glueyness”. In the case of underspecification, discourse relations
holding between discursive units permit multiple interpretations. To preempt pos-
sible overlaps, underspecified discourse relations may be signaled with discourse
markers, and to ensure speaker-intended interpretation, fully specified discourse
relations can be embellished with discourse markers. Depending on the number
of features and conditions indexed, discourse relations can be overspecified, fully
specified or they can be underspecified to various degrees and thus allow multiply
discursive functions.
The analysis of discourse is fundamentally concerned with the nature of the
connectedness between parts and wholes, and for this reason discourse is a rela-
tional construct par excellence, relating separate parts locally as well as globally with
regard to their connectedness to discourse-as-a-whole. Discourse is thus not only
quantity, as is captured by the number of its constitutive parts, but also quality, as
is reflected in the force and nature of connectedness of its constitutive parts, which
are relational by definition. Sentence, clause, utterance and proposition have been
assigned the status of discourse units across discourse grammar and discourse se-
mantics. Discourse pragmatics, however, requires a relational discourse unit which
does not only account for content and force, but allows for the accommodation of
the dynamics of discourse and thus for varying quantities, i.e., discourse marker,
utterance, paragraph(s) or sequence(s), accommodating not only the duality of
form and function, but also their instantiations in context.
4. Pragmatic discourse and discourse pragmatics
Pragmatics has been defined as the study of context-dependent meaning as well as

the study of speaker-intended meaning, presupposing the existence of language,
language user and context on the one hand, and context-independent meaning on
the other. It has been referred to as the study of invisible meaning (Yule 1996: 127),
the science of the unsaid (Mey 2001: 194), the study of meaning as it “emerges
in language use” (Marmaridou 2000: 1), and the study of linguistic acts, commu-
nicative action and their appropriateness (van Dijk 2008; Fetzer 2004). Pragmatics
has thus been anchored firmly to the performance of communicative action in
context. But is the reference to context in the definition of pragmatics sufficient?
Doesn’t pragmatics rather investigate the performance of communicative action

in discourse embedded in context, if not communicative action performed with
discourse in context?
It is impossible to conceptualize communicative action without the explicit
accommodation of context, in which it is embedded and to which it refers, and
it seems impossible to conceptualize communicative action in context without
the explicit accommodation of discourse, which constrains its production and
interpretation, and delimits context. Discourse is composed of linguistic context
(including other semiotic codes), and it needs cognitive context to account for
discourse production, discourse processing, grounding and discourse coherence.
Discourse is embedded in social context and at the same time contains social
context. Against this background, defining pragmatics as the study of discourse-
dependent meaning seems more appropriate, as would be captured by the concept
of pragmatic discourse. However, this would narrow down the frame of investiga-
tion to discourse-as-a-whole (or to communicative-action-as-a-whole), ignoring
the constitutive parts of discourse, which may be speech acts, pragmatic acts,
communicative acts or conversational contributions, for instance. Discourse is
thus both: discourse-as-a-whole and linearized discourse units, which constitute
discourse-as-a-whole.
In the framework of pragmatic discourse, discourse and its constitutive parts
need to be relational concepts, relating discourse and context, discourse and com-
municative action, communicative action and interlocutors, and interlocutors with
the things they do with words in discourse in context, and the things they do with
discourse in context. Only a relational frame of reference can capture the dynamics
of pragmatic discourse, i.e. the unfolding of discourse-as-whole on the one hand,
and of variation of linearized sequences and within linearized sequences on the
other, and thus the connectedness between parts and wholes, transcending clear-
ly delimited frames of investigation. Discourse pragmatics, however, needs to go
beyond a pragmatics-based conception of discourse. Adopting the fundamental
premises of pragmatics, i.e. rationality, intentionality of communicative action, con-
textualization and cooperation, it considers discourse as dynamic, i.e. both process
and product, and multilayered, and adapts them to the quantity-based definition of
discourse by accounting for granularity in an explicit manner. This allows for a dy-
namic and relational analysis of the structuring of discourse as regards granularity,
which is reflected in the sequential organization of discourse, and linearization, as
well as of the nature of the connectedness between discourse units made manifest
in discourse relations, cohesion and coherence.
A discourse-pragmatics-based discourse unit contains force, content and
metadiscursive meaning, with varying degrees of explicitness. What is more, the
discourse-pragmatics-based discourse unit is also indexical, expressing exophoric
248 Anita Fetzer
and endophoric reference. Because of their relational conceptualization, discourse

units are doubly contextual. By contextualizing prior discourse units they pave the
ground for the production and interpretation of upcoming discourse units thus
indicating how the discourse is to proceed, i.e. whether there is some change in the
intended direction as is signalled by contrastive discourse markers or other devices,
or whether there is no intended change and the discourse is to proceed as intended,
as is signalled by continuative discourse markers, for instance.
In discourse pragmatics, granularity does not only refer to micro, but also to
meso and macro units of investigation (cf. Fetzer 2004, 2013). A relational con-
ceptualization of discourse unit, this chapter argues, would allow to account for
the extension of frame from micro, i.e. minimal units, such as discourse markers,
which contain force and metadiscursive meaning, but only presupposed content,
or conversational contributions, which contain force, content and metadiscursive
meaning, to meso, i.e. sequences or episodes, to macro, i.e. discourse-genre-as-a-
whole. Meso and macro discourse units also contain force, content and metadiscur-
sive meaning. A very broad notion of discourse, as is reflected, for instance, in the
discourse on context or the discourse on political correctness, could be captured
by a unit “beyond ‘macro”.
Thibault’s definition of genre (2003: 44) may accommodate the distinction be-
tween micro discourse unit, meso discourse unit, and macro discourse unit:
genres are types. But they are types in a rather peculiar way. Genres do not spec-
ify the lexicogrammatical resources of word, phrase, clause, and so on. Instead,
they specify the typical [original emphasis, A. F.] ways in which these are com-
bined and deployed so as to enact the typical semiotic action formations of a given
community.
Since genres “specify typical [original emphasis, A. F.] ways in which these [mi-
cro discourse units and their realizations, A. F.] are combined and deployed [to
constitute meso discourse units and their linguistic realizations, i.e. sequences or
episodes, A. F.] so as to enact the typical semiotic action formations of a given
community”, they can be assigned the status of macro discourse units.
Connected intrinsically with the “typical ways” of doing things with words
in a discourse genre – or in an activity type, in Levinson’s terms – are inferential
schemata (Levinson (1979:370):
… there is another important and related fact, in many ways the mirror image of
the constraints on contributions, namely the fact that for each and every clearly
demarcated activity there is a set of inferential schemata [original emphasis, A. F.].
These schemata are tied to (derived from, if one likes) the structural properties of
the activity in question.
The communicative value of micro, meso and macro discourse units is expressed
in these “typical ways” of doing things with words in discourse genres, and the cor-
responding “inferential schemata” constrain the production and interpretation of
discourse units in discourse in context, as will be illustrated in the following section.
5. Discourse pragmatics in action
Discourse pragmatics is anchored to a dynamic frame of reference, which requires

the explicit accommodation of relational units of investigation, i.e. discourse units
with force, content and metadiscursive meaning. It departs from the premise that
discourse units are produced and interpreted in accordance with the key premises
of pragmatics, rationality, intentionality of communicative action, contextualiza-
tion and cooperation, and that the linguistic realization of discourse units makes
these key premises explicit in and through discourse. Not all discourse units are of
equal standing. Meso and macro discourse units need to have force, content and
metadiscursive meaning, while micro discourse units may have force, content and
metadiscursive meaning, but do not need to. Minimal discourse units, such as
discourse markers express procedural meaning, and thus have force and metadis-
cursive meaning, but presupposed content only.
The analysis of discourse from a structure-based viewpoint and a type-based
analysis of discourse units as its constitutive parts seems to be a less complex en-
deavour because there is no immediate need to account for the linguistic realization
of discourse units, in particular for context- and discourse-dependent variation.
However, if discourse is approached from a micro-unit-based, bottom-up perspec-
tive the linguistic realizations of discourse units need to be delimited from embed-
ding linguistic context as well as from surrounding discourse units, in order to be
related to and mapped with the discourse-unit-types they realize. The mapping
processes are generally constrained by local and not-so-local genre-specific con-
straints and by the participants’ local and not so-local communicative intentions.
Possible form-function and function-form mappings are examined in the following
from micro, meso and macro perspectives.
The micro discourse unit with the force of request can be realized by one of
the following tokens: “X!” with explicit force, content but implicit metadiscursive
meaning; “let’s X” with indirect conventional force, explicit content and implicit
metadiscursive meaning; “would you be good enough to X” with indirect conven-
tional force, content and metadiscursive meaning; or by a conventionalized refer-
ence to its preparatory conditions “could you X” or “can you X” both with indirect
conventional force, content and metadiscursive meaning. The micro discourse unit
with requestive force can also be a constitutive part of the meso discourse units
250 Anita Fetzer
“asking for directions” or of the “opening section of fundraising letter”. Analogously

to the Gricean differentiation between generalized and particularized implicatures
a reference to preparatory conditions may count as a generalized discourse unit
with requestive force.
Meso discourse units go beyond individual discourse units and capture more or
less bounded sequences with particular force, content and metadiscursive meaning,
for instance the opening section or closing section of a macro discourse unit, or
topical sequences, which are also referred to as episodes. As is the case with the
linguistic realization of micro discourse units as generalized discourse units, meso
discourse units, such as opening and closing sections, can also be realized as gen-
eralized meso discourse units, for instance “how are you?”, “how are you doing?”,
“how nice to see you”, or “it was so good to see you, we must meet again some time”,
“give us a call when you’re in town” or “keep in touch” for mundane spoken interac-
tion. Generalized meso discourse units are also frequent in the context of academic
discourse and its conventional introductory and conclusive sections. However, it
is not generally the whole meso discourse unit, which is realized as a generalized
meso discourse unit, but only a number of its constitutive micro discourse units,
for instance “in this paper the research results from X have been corroborated” or
“the following section summarizes the results obtained”.
Unlike conventionalized indirect speech acts, generalized discourse units ad-
ditionally make explicit metadiscursive meaning with respect to sequential posi-
tioning and dovetailedness. For instance, the conventionalized indirect speech act
“how are you?” does not intend to elicit information about the addressee’s physical
or psychological condition but rather counts as a common greeting with the appro-
priate second part “Fine, and how are you?”. The generalized discourse unit “how
are you?” has the generalized force “request for dovetailed information” with the
content “addressee’s condition” and the metadiscursive meaning “initial discourse
unit in an interaction inviting dovetailed response in order to function as a com-
mon greeting”. It differs from the not-generalized discourse unit “how are you?”
expressing a request for dovetailed information with the content “addressee’s condi-
tion” and the metadiscursive meaning “non-initial discourse unit in an interaction,
but initial discourse unit of the first meso discourse unit from the topical part of a
medical encounter”. In that medical interview, there are generally two “how-are-
you”-discourse units, a generalized discourse unit and a non-generalized one (cf.
Coupland, Coupland and Robinson 1992).
Discourse units contain force, content and metadiscursive meaning, which
are constrained by felicity conditions on the one hand, and appropriateness condi-
tions on the other. While felicity conditions specify constraints on the force of the
discourse unit, such as the preparatory conditions for requestive force as well as
on the content of the discourse unit, i.e. what is requested for the discourse unit to
be felicitous in generalized (or default) contexts (cf. Fetzer 2012), appropriateness

conditions particularize felicity conditions by specifying their conditions of use as
regards their degree of explicitness for metadiscursive meaning as well as for their
genre- and context-specific production and interpretation. As such, appropriate-
ness conditions are anchored to sociocultural context and to genre conventions (cf.
Fetzer 2007a, 2007b), and they are based on the Gricean Cooperative Principle, its
maxims and implicatures, which are pillars for the production and interpretation
of genre- and context-dependent meaning. This is particularly true for the discur-
sive constraint of dovetailedness (cf. Fetzer 2013, 2014), i.e. make your discourse
unit “such as is required, at the stage at which it occurs, by the accepted purpose
or direction of the talk exchange in which you are engaged” (Grice 1975: 45).
Appropriateness conditions specify the conditions of use for macro discourse units
as regards production and interpretation, and they may be particularized for dis-
course units for particularized purposes. Macro discourse units, especially if their
quantity is huge, allow for a much lower degree of generalization. There may, how-
ever, be some instances with a high degree of generalization, such as institutional
warnings in the written format.
Discourse units are produced and interpreted in accordance with basic prag-
matic premises and thus express communicative intentionality. Macro discourse
units may express various communicative intentions on the micro and meso lev-
els, and for this reason, macro discourse units are assigned discourse purpose.
Discourse purpose is a pragmatic concept and is connected with intentionality of
communicative action (Cohen, Morgan and Pollack 1992; Levinson 1995; Searle
1983). It is made manifest in discourse and discursive action analogously to the
speech-act-theoretic operationalization “X counts as Y in context C” with felicity
conditions as context categories (Sbisà 2002a, 2002b), and their particularization as
appropriateness conditions, that is “X counts as Y in discourse D in context C”. For
instance, the macro discourse unit of interview, whose purpose is to elicit informa-
tion, is composed of micro discourse units with the force “request for information”
and the metadiscursive meaning “provide dovetailed response”, if produced by the
interviewer, and discourse units with the force of “assertion” and the metadiscursive
meaning “dovetailed response” and discourse units with the force “rejection” and
the metadiscursive meaning “contrastive dovetailed response” or one of its par-
ticularizations, if produced by the interviewee: discourse units with assertive force
provide the information requested and rejections do not comply with the request
for some particular information. Micro discourse units are thus different from
ordinary communicative acts because they additionally express dovetailedness, se-
quentiality and participant format in their felicity conditions and appropriateness
conditions. Meso discourse units in interviews are bounded sequences in which the
communicative status of micro discourse units is negotiated, either by follow-up
252 Anita Fetzer
sequences with respect to obtaining more precise information or by clarification

sequences about the content, force or metadiscursive meaning.
Discourse purpose and the relational nature of discourse units with force, con-
tent and metadiscursive meaning require the accommodation of context within
discourse units, and the accommodation of the context embedding discourse units.
What is more, the intentionality of communicative action in discourse and with
discourse presupposes its iterability and thus the differentiation between type, i.e.
typical ways of how discourse units are done with words (and other semiotic codes),
and their actual linguistic realization in discourse and context.
The macro discourse unit of interview may undergo discourse-purpose-specific
particularization, according to the kind of information elicited. Political interviews
are used strategically to elicit and systematize political information, oral exam-
inations are used in educational contexts to assess the examinee’s expertise, job
interviews are used to evaluate a candidate’s suitability and expertise, and health
interviews are used to elicit information about patients’ conditions. The macro
discourse unit of interview is also used to elicit and systematize citizenship-ori-
ented information about relevant criteria for the (non)qualification for income
support, housing benefit or political asylum, and it may also be used for various
other purposes. Depending on their discourse-specific purpose, particularized in-
terviews are composed of discourse-specific micro discourse units constrained by
linguistic and social style (e.g., lexicon, syntactic complexity, discursive complexity,
non-verbal code of conduct) and the participants’ face-wants and face-needs (cf.
Brown and Levinson 1987), for instance formal style with negative politeness or
informal style with positive politeness (cf. Fetzer 2000 for the particularized dis-
course unit of political interviews). Discourse-specific purpose may also constrain
the sequential organization of meso discourse units, such as elaborate opening or
closing sections, ad-hoc pre- or side sequences, reformulation sequences or devia-
tions from the participant-specific employment of discourse units with requestive
force, for instance interviewees asking questions to perform requests, which are,
in general, accounted for.
The particularization of macro discourse unit manifests itself in the particu-
larization of content, e.g., political ideologies, expertise, health condition, and of
discourse purpose, e.g. communicate and promote political ideologies, perform
student’s assessment, identify appropriate cure, while their force, i.e. request for
dovetailed information and assertion / rejection, remains stable. Meso discourse
units are composed of – more or less bounded – sequences or episodes, and thus
contain content, which may undergo meso-discourse-unit specific particulariza-
tion. For instance, the opening section of the macro discourse unit of medical
interview or oral examination generally displays particularized opening sequences,
which may not be of immediate relevance to the topical sections but rather have a
phatic function; bureaucratic interviews will display less positive politeness and aim
for a more neutral stance. The meso discourse unit of closing section will also show
particularization for the particularized interviews and usually display a lower de-
gree of positive and negative politeness. The meso discourse unit of topical sequence
of the particularized macro discourse units is more content-oriented and varies
accordingly. Thus, the particularization of macro discourse units is inherited to the
particularization of meso discourse units. But is the particularization of discourse
purpose also inherited to the particularization of meso discourse units? Discourse
purpose is a macro concept and therefore anchored to the macro discourse unit by
definition. To undergo meso-discourse-unit-specific particularization and at the
same time contribute to the interactional organization of the macro discourse-unit-
as-a-whole, possible particularizations need to be related to the discourse-purpose-
as-a-whole, and should meso discourse units deviate from the overall discursive
purpose, they would need to be accounted for, as is the case with side sequences,
clarification sequences or critical incidents (cf. Fetzer 2006 for political interviews).
Discourse allows for systematic discourse-internal and discourse-external vari-
ation and thus for particularization of macro, meso and micro discourse units with
respect to linearization and style, accommodating discourse purpose as well as local
and not-so-local communicative intentionality.
6. Conclusion
Discourse has been described as a multifarious and multilayered construct, which

seems almost impossible to delimit (cf. Fetzer 2013, 2014). It has been defined as
quantitatively larger than one discourse unit, and is thus – in discourse pragmat-
ics – composed of a number of concatenated micro and meso discourse units. In
discourse pragmatics, discourse units are pragmatic concepts and constrained by
felicity conditions on the micro level and by appropriateness conditions on the meso
and macro level. The linearization of the constitutive units of discourse allows for
multiple combinations, whose ordering is constrained by discourse genre and dis-
cursive purpose as well as by the participants’ communicative goals. While the con-
stitutive units of discourse can be analysed as grammatical or ungrammatical, true
or false, felicitous or infelicitous, or appropriate or inappropriate, their ordering
cannot be classified along those lines only. This is because discourse is a parts-whole
configuration in which the meaning of the whole is more than the sum of its separate
parts. If the ordering of the parts changes, so does the meaning of the whole.
Micro, meso and macro discourse units and their linguistic realizations are
related dialectically in discourse. This holds for micro discourse units and their
linguistic realizations, for meso discourse units and their linguistic realizations,
254 Anita Fetzer
and for macro discourse units and their linguistic realizations. Analogously to mul-
tilayered context, macro discourse units and their constitutive parts of meso and
micro discourse units are multilayered and doubly, if not multiply contextual, and
their order of inclusion, i.e. micro, meso and macro, corresponds to their order of
accessibility (cf. Fetzer 2012).
To count as a discourse unit in discourse pragmatics, the unit needs to have
(1) illocutionary force, (2) content, and (3) metadiscursive meaning. Classical dis-
course units have all three, and minimal discourse units, such as discourse markers,
have illocutionary force and metadiscursive meaning but only presupposed content.
From a quantitative perspective, the macro discourse unit is the largest unit. It is
composed of meso and micro discourse units and therefore allows for variation
with respect to the local and not-so-local linearization of meso discourse units, and
of micro discourse units, thus displaying discourse-internal and discourse-external
variation. In the case of systematic variation, macro discourse units undergo a pro-
cess of particularization, as has been shown for different types of interviews. Meso
discourse units are quantitatively smaller units and may allow for particularization,
but usually only for particularized meso discourse units, such as opening or closing
meso discourse units.
The goal of this chapter has been to examine the constitutive parts of dis-
course in discourse pragmatics. It proposes a relational conception of discourse
unit carrying force, content and metadiscursive meaning. The chapter has shown
that a relational conception of discourse unit provides necessary tools to capture
the dynamics of discourse, and to describe and analyse both the constitutive parts
of discourse as well as the entire discourse from a parts-whole perspective. On
this view of discourse and discourse units, implicit meanings reside and can be
studied in different and context-dependent patterns of overtness for the linguis-
tic realization of discourse units and the signalling of discourse relations. Among
these apparently infinite patterns, generically constrained forms emerge as the most
promising methodological and empirical terrain.
References
Asher, Nicolas and Alex Lascarides. 2003. Logics of Conversation. Cambridge: Cambridge
University Press.
Austin, John L. 1976. How to do Things with Words. Cambridge: Cambridge University Press.
Bach, Kent. 1992. “Communicative Intentions, Plan Recognition, and Pragmatics: Comments on
Thomason and Littman and Allen.” In Intentions in Communication, ed. by Philip Cohen,
Jerry Morgan, and Martha E. Pollack, 389–400. Cambridge: MIT Press.
Brown, Penelope, and Stephen Levinson. 1987. Politeness. Some Universals in Language Usage.
Bublitz, Wolfram, Uta Lenk, and Eija Ventola (eds). 1999. Coherence in Spoken and Written
Discourse. Amsterdam: John Benjamins. doi: 10.1075/pbns.63
Carston, Robyn. 2002. Thoughts and Utterances. The Pragmatics of Explicit Communication.
Chafe, Wallace. 1994. Discourse, Consciousness and Time. Chicago: Chicago University Press.
Clark, Billy. 2013. Relevance Theory. Cambridge: Cambridge University Press.
doi: 10.1017/CBO9781139034104
Cohen, Philip, Jerry Morgan and Martha E. Pollack (eds). 1992. Intentions in Communication.
Cambridge: MIT Press.
Coupland, Justin, Nik Coupland, and Jeffrey D. Robinson.1992. ‘“How Are You?”: Negotiating
Phatic Communion.” Language in Society 21: 207–230. doi: 10.1017/S0047404500015268
de Beaugrande, Robert, and Wolfgang Dressler. 1981. Einführung in die Textlinguistik. Tübingen:
Niemeyer. doi: 10.1515/9783111349305
Fetzer, Anita. 2000. “Negotiating Validity Claims in Political Interviews.” Text 20: 1–46.
Fetzer, Anita. 2004. Recontextualizing Context: Grammaticality Meets Appropriateness.
Amsterdam: John Benjamins. doi: 10.1075/pbns.121
Fetzer, Anita. 2006. ‘“Minister, We Will See How the Public Judges You’. Media References in
Political Interviews.” Journal of Pragmatics 38: 180–195. doi: 10.1016/j.pragma.2005.06.017
Fetzer, Anita. 2007a. “Context, Contexts and Appropriateness.” In Context and Appropriateness:
Micro Meets Macro, ed. by Anita Fetzer, 3–27. Amsterdam: John Benjamins.
doi: 10.1075/pbns.162.02fet
Fetzer, Anita. 2007b. ‘“If I May Say So’: Indexing Appropriateness in Dialogue.” In Context
and Appropriateness: Micro Meets Macro, ed. by Anita Fetzer, 115–146. Amsterdam: John
Benjamins. doi: 10.1075/pbns.162.08fet
Fetzer, Anita. 2007c. “Reformulation and Common Grounds.” In Lexical Markers of Common
Grounds, ed. by Anita Fetzer and Kerstin Fischer, 157-179. London: Elsevier.
Fetzer, Anita. 2012. “Contexts in Interaction: Relating Pragmatic Wastebaskets.” In What is a
Context? Linguistic Approaches and Challenges, ed. by Rita Finkbeiner, Jörg Meibauer, and
Petra Schumacher, 105–127. Amsterdam: John Benjamins. doi: 10.1075/la.196.08fet
Fetzer, Anita. 2013. “The Pragmatics of Discourse.” Topics in Linguistics 11: 5–12.
Fetzer, Anita. 2014. “Conceptualizing Discourse.” In Handbooks of Pragmatics. The Pragmatics
of Discourse. Vol. 3, ed. by Klaus Schneider, and Anne Barron, 35–61. Berlin: Mouton de
Gruyter. doi: 10.1515/9783110214406-003
Fetzer, Anita, and Augustin Speyer. 2012. “Discourse Relations in English and German Discourse:
Local and Not-So-Local Constraints.” Intercultural Pragmatics 9: 413–452.
doi: 10.1515/ip-2012-0025
256 Anita Fetzer
Gernsbacher, Morton-Ann, and Talmy Givón (eds). 1995. Coherence in Spontaneous Text.
Amsterdam: John Benjamins. doi: 10.1075/tsl.31
Givón, Talmy. 1993. English Grammar: A Function-Based Introduction. Amsterdam: John
Benjamins. doi: 10.1075/z.engram1
Grice, Herbert Paul. 1975. “Logic and Conversation.” In Syntax and Semantics. Vol. III, ed. by
Peter Cole, and Jerry L. Morgan, 41–58. New York: Academic Press.
Gumperz, John J.1992. “Contextualization and Understanding.” In Rethinking Context: Language
as an Interactive Phenomenon, ed. by Alessandro Duranti and Charles Goodwin, 229–252.
Halliday, Michael A. K. 1994. Introduction to Functional Grammar. London: Arnold.
Halliday, Michael, A. K. and Ruqaiya Hasan. 1987. Cohesion in English. London: Longman.
Heritage, John. 1984. Garfinkel and Ethnomethodology. Cambridge: Polity Press.
Horn, Laurence R. 1984. “Toward a New Taxonomy for Pragmatic Inference: Q-based and
R-Based Implicature.” In Meaning, Form and Use in Context: Linguistic Applications, ed. by
Deborah Schiffrin, 11–42. Washington: Washington University Press.
Levinson, Stephen C. 1979. “Activity Types and Language.” Linguistics 17: 365–399.
doi: 10.1515/ling.1979.17.5-6.365
Levinson, Stephen C. 1983. Pragmatics. Cambridge: Cambridge University Press.
Levinson, Stephen C. 1995. “Interactional Bias in Human Thinking.” In Social Intelligence and
Interaction, ed. by Esther Goody, 221–260. Cambridge: Cambridge University Press.
doi: 10.1017/CBO9780511621710.014
Implicature. Cambridge: MIT Press.
Maier, Robert M., Carolin Hofmockel and Anita Fetzer. 2016. “The Negotiation of Discourse
Relations in Context: Co-Constructing Degrees of Overtness.” Intercultural Pragmatics
13(1): 71–105.
Marmaridou, Sophia. 2000. Pragmatic Meaning and Cognition. Amsterdam: John Benjamins.
doi: 10.1075/pbns.72
Martin, Jim R., and David Rose. 2008. Genre Relations. Mapping Culture. London: Equinox.
Mey, Jacob. 2001. Pragmatics. An Introduction. Oxford: Blackwell.
Mey, Jacob. 2011. “Speech Acts in Context.” In Context and Contexts: Parts Meet Whole?, ed. by
Anita Fetzer and Etsuko Oishi, 171–180. Amsterdam: John Benjamins.
doi: 10.1075/pbns.209.11mey
Moeschler, Jacques. 2002. “Speech Act Theory and the Analysis of Conversations.” In Essays in
Speech Act Theory, ed. by Daniel Vanderveken, and Susumu Kubo, 239–261. Amsterdam:
John Benjamins. doi: 10.1075/pbns.77
Oishi, Etsuko and Anita Fetzer. 2016. “Expositives in Discourse”. Journal of Pragmatics 96: 49–59.
Roulet, Eddy. 1991. “On the Structure of Conversation as Negotiation.” In (On) Searle on
Conversation, ed. by John R. Searle, Herman Parret, and Jef Verschueren, 91–100. Amsterdam:
John Benjamins. doi: 10.1075/pbns.21
Roulet, Eddy. 2006. “The Description of Text Relation Markers in the Geneva Model of Discourse
Organization.” In Approaches to Discourse Particles, ed. by Kerstin Fischer, 115–132. Oxford:
Elsevier.
Sbisà, Marina. 2002a. “Cognition and Narrativity in Speech Act Sequences.” In Rethinking
Sequentiality: Linguistics Meets Conversational Interaction, ed. by Anita Fetzer, and Christiane
Meierkord, 71–97. Amsterdam: John Benjamins. doi: 10.1075/pbns.103.04sbi
Sbisà, Marina. 2002b. “Speech Acts in Context.” Language and Communication 22: 421–436.
doi: 10.1016/S0271-5309(02)00018-6
Speyer, Augustin, and Anita Fetzer. 2014. “The Coding of Discourse Relations in English and
German Argumentative Discourse.” In The Pragmatics of Discourse Coherence, ed. by Helmut
Gruber, and Gisela Redeker, 87–119. Amsterdam: John Benjamins. doi: 10.1075/pbns.254
Searle, John R. 1983. Intentionality. Cambridge: Cambridge University Press.
doi: 10.1017/CBO9781139173452
Searle, John. 2010. Making the Social World: The Structure of Human Civilization. Oxford: Oxford
University Press.
Thibault, Paul J. 2003. “Contextualization and Social Meaning-making Practices.” In Language
and Interaction. Discussions with John J. Gumperz, ed. by Susan L. Eerdmans, et al., 41–62.
Amsterdam: John Benjamins. doi: 10.1075/z.117.05thi
Van Dijk, Teun. 2008. Discourse and Context. A Sociocognitive Approach. Cambridge: Cambridge
Widdowson, Henry. 2004. Text, Context, and Pretext. Critical Issues in Discourse Analysis. Oxford:
Blackwell. doi: 10.1002/9780470758427
Wilson, Deidre, and Dan Sperber. 2012. Meaning and Relevance. Cambridge: Cambridge University
Press. doi: 10.1017/CBO9781139028370
Yule, George. 1996. The Study of Language. Cambridge: Cambridge University Press.
Chapter 12
Why don’t you tell it explicitly?

Personal/subpersonal accounts of implicitness
Marco Mazzone
University of Catania
Why do we use implicitness as a communicative strategy? In this chapter I

suggest that the reason has to do – among other things – with the structure
of memory, on the one hand, and the human ability to construct contextually
shared chains of goals, on the other. I also propose that by framing the issue of
implicitness in these terms we may both gain a vantage point for the evaluation
of pragmatic theories and have important insights about practical reasoning and
its relation with utterance understanding.
Implicitness phenomena may be approached at personal and sub-personal
levels. On the one hand, they are explained by communicative predispositions
anchored in the structure of human memory, on the other they reflect human
ability to construct contextually shared chains of goals. These two research paths
are not incompatible: personal and sub-personal components work together and
I argue that this is rooted in the cooperation between automatic and controlled
associative processes.
Keywords: motives for implicitness, levels of implicitness, personal level,

subpersonal level, politeness, relevance
1. Overview
Why do we use implicitness as a communicative strategy? An answer to this

question may be provided at the personal level of commonsense psychology, in
terms of reasons that agents may have for not conveying explicitly the intended
propositions. Research on politeness (starting from Lakoff 1973; Leech 1983;
Brown and Levinson 1987) is an especially well-studied case in point, based on
the assumption that protecting the face of interlocutors in conversation is an im-
portant reason for speakers to be indirect. But the personal-level approach is not
the only way to address the issue. Another possibility is that language production
and comprehension in context are accounted for by mechanisms operating at the
doi 10.1075/pbns.276.12maz
260 Marco Mazzone
subpersonal level, and that this is the level at which an explanation of implicitness
has to be looked for. An influential representative of this approach is Relevance
Theory (from now on, RT; see Sperber and Wilson 1986/1995), according to
which key to inferential communication is a subpersonal mechanism guided by
expectations of relevance (for an explicit characterization of RT as a subpersonal
approach, see Carston 2000 and Wilson and Kolaiti, this volume). Although RT is
essentially a theory of comprehension and has devoted less attention to language
production and to politeness in particular, its general approach to why commu-
nication is mostly implicit is nevertheless rather clear. Besides, Jary (1998) has
provided a convincing account of politeness in relevance theoretic terms (see
also Escandell-Vidal 1998).
My main aim in this chapter is to examine the two approaches to implicitness,
the personal and the subpersonal, and also to compare them with a pragmatic
framework I have been developing in the last decade, one important feature of
which is a strong continuity between the personal and the subpersonal descrip-
tions of cognition in general, and of pragmatics in particular. According to that
view, pragmatic understanding is accounted for in terms of associative activation
of representations, a domain-general mechanism constrained by the schematic and
hierarchical organization of memory (Mazzone 2011, 2015a). Goals are conceived,
specifically, as high-level schemata coordinating lower-level representations in both
conceptual and sensorimotor domains (Mazzone 2009, 2011, 2015a), and the fur-
ther assumptions are made that communication involves the formation of chains of
goals encompassing both communicative and non-communicative intentions, and
that in joint activities, including communication, shared intentional contexts are
construed in which sub-plans of different agents mesh with each other (Mazzone
2015b). Another important assumption of that view is that schematic representa-
tions are equally recruited into controlled and automatic processing, so that the
latter is subject to the same rational constraints imposed on the former by the
non-arbitrary structure of memory. This is a first sense in which there is no radical
discontinuity between the personal and the subpersonal level: both controlled and
automatic processes are rational (inferential) in the sense specified above (Mazzone
2014a). Moreover, those processes vary along a continuum (Mazzone and Campisi
2013; Mazzone 2013a, 2015a). An important consequence for pragmatics is that
what constitutes a genuine case of implicitness may be less clear-cut than is usually
thought, since an implicature can be short-circuited and therefore lexicalized at
different degrees, thus becoming more and more automatic, that is activated by a
more and more direct association (Mazzone 2014b).
On the basis of this background, I intend to propose that a variety of subper-
sonal and personal factors might contribute to making human communication
largely indirect.
Chapter 12. Why don’t you tell it explicitly? 261
On the side of subpersonal factors, I will consider the role of informativeness

and what might be called a “point of arrest” factor. The general idea is that com-
municative situations usually put constraints on the communicative intentions of
the speaker, so that the implicit content, in its general lines, is often more expected
by the interlocutors than the explicit one: as a consequence, the latter is more in-
formative than the former and it is actually the point at which the speaker stops
the search for what is relevant to communicate. As we will see, these factors are
largely compatible with RT’s account, at least when expectations of relevance are
conceived in terms of expected types of cognitive effects.
On the other hand, as far as personal-level factors are concerned, I will consider
the role of what I will call the “affiliation” and the “dominance” effects. Affiliation
has to do with the traditional phenomena of politeness, which I find useful and
informative to redescribe in terms of goal sharing (cf. Kadar, this volume). In prac-
tice, if communicating implies construing shared intentional contexts where ours
and others’ goals are represented together and adopted by each of the interlocutors,
indirect communicative strategies can be used to promote a sense of affiliation in
the service of mitigation either by providing (and thus “putting in common”) a
super-arching goal by reason of which we cannot adopt a goal proposed by the
interlocutor, or by presenting a request by way of showing our interest toward the
conditions that might enable the interlocutor to satisfy the goal expressed by our
request. But affiliation can also be promoted by gratuitously sharing emotions and
interests towards something, even when there are no action goals that one inter-
locutor expects the other to adopt. In such cases, the communicative behavior and
its referent (the object of interest) constitute a minimal means-end structure whose
sharing promotes affiliation. Humor can be considered a form of indirectness that
exploits this kind of emotion-based or interest-based affiliation at a higher level:
by being humorous speakers express a content in an indirect and even surprising
manner and invite addressees to share appreciation for the wit of such indirectness.
As is well known, humor can also be recruited into politeness strategies, since the
sense of affiliation that it promotes can be exploited for mitigation effects. On the
contrary, humor can be used to convey a sense of control and superiority (the
“dominance effect”), as in some cases of parental humor, with cases of ambiguity
between the two (Rundquist 1990; Brumark 2006).
In the framework I propose, the personal and subpersonal factors described
above are not incompatible with each other, on the contrary they are presumed to
work together. In contrast with RT, pragmatic processing is not explained in terms
of a specialized subpersonal mechanism, it is explained instead in terms of coop-
eration between automatic and controlled domain-general processes, conceived of
as different dynamics of activation within a common associative network. Not only
do I propose that the presumed opposition between automatic and controlled is in
262 Marco Mazzone
fact a matter of degrees, but also that there is an intermediate condition between
pure automatic processing and personal-level conscious reasoning, a condition
characterized by largely automatic activation sustained and monitored by conscious
attention (Mazzone 2013a, 2013b) – with conscious attention focusing on contents
that may change from one situation, and from one moment, to another. In sum, my
claim is that automatic activation and conscious attention are both constitutive-
ly present with varying proportions in any episode of pragmatic processing (and
strategic reasoning as well), both on the side of the addresser and the addressee.
In order to better appreciate how this proposal works, I also intend to analyze
the way in which Jary (1998) exploits, in the service of his relevance theoretic
account, the distinction between genuine cases of strategic politeness involving
implicatures, on the one hand, and cases of politeness as social indexing on the
other. His general argument seems to be that, in most cases, the linguistic forms
associated to politeness are neither very informative nor strategically designed to
obtain certain effects, they are just the way things get done in order to conform to
sociocultural constraints. As a consequence, one should not be surprised if most
uses of polite forms have low relevance: in those cases, politeness is not part of
what is pragmatically communicated, in other words, it is anticipated rather than
implicated. In accordance with RT, the argument goes, relevance increases only
when forms of politeness are strategically manipulated in such a way as to contra-
dict some sociocultural expectations, in which cases politeness is communicated,
and specifically it is implicated.
My analysis will focus on two points. First, Jary characterizes the distinction
between politeness as social indexing and strategic politeness (or between antici-
pated and implicated politeness) also in terms of “unselfconscious compliance with”
versus “selfconscious exploitation of sociocultural constraints on language usage”
(Jary 1998: 7; emphasis mine). It should be noted, however, that while this is entirely
compatible with my proposal that conscious attention is constitutive of pragmatic
processing, it is less obviously compatible with RT, according to which pragmatic
processing is accounted for by an automatic, subpersonal mechanism, with con-
sciousness having no special role to play in it. Second, if my approach is correct,
the distinction between cases in which there is implicit communication of polite-
ness and cases in which there is only politeness as social indexing is not clear-cut:
politeness phenomena may be consciously attended at varying degrees. I intend to
provide a more specific analysis of some of the factors possibly involved, along the
lines of Mazzone (2014b).
In what follows, I will first summarize personal- and subpersonal-level views
about why human communication happens to be implicit; then I will sketch my
general approach to pragmatic processing; finally, in the light of that approach, I
will turn back to the possible explanations of implicitness.
2. Personal/subpersonal explanations of implicitness and politeness
Grice (1989) laid the foundations for a pragmatic theory of implicit meaning. In
his view, there is quite often a gap between what is (explicitly) said and what is
(implicitly) meant by a speaker with an utterance: while the former is determined
(except for peripheral aspects) by linguistic information, the latter is derived from
the former thanks to rational inferences also taking into account context and
background knowledge (together with the Cooperative Principle and the max-
ims). However, Grice characterized his own descriptions of those inferences as
“rational reconstructions”, implying that he was not committed to the view that
utterance understanding actually involved complex reasoning at the personal level
(see Geurts and Rubio-Fernández 2015 for a recent discussion of this issue; cf. also
Dynel, this volume).
Research on politeness in particular (starting from Lakoff 1973; Leech 1983;
Brown and Levinson 1987) and accounts of the reasons for implicitness in gen-
eral have mostly followed Grice’s style of explanation: they have focused on
personal-level reasons for speakers to be indirect, while paying little attention to
the actual cognitive mechanisms presumed to account for that capacity. Thomas
(1995/1998: 143), for instance, has famously summarized the possible reasons to
be indirect as follows: (1) the desire to make one’s language more/less interesting;
(2) the desire to increase the force of one’s message; (3) competing goals; and (4)
politeness/regard for face. As it may be seen, she makes an explicit appeal to com-
monsense psychological notions such as desires on the part of the speaker and goals
of both speakers and hearers (points 1 to 3). As for politeness (point 4), reference
to personal-level goals, purposes or desires is ubiquitous throughout the literature.
For an influential example, in the introduction to the second edition of Politeness
Brown and Levinson (1987: 1) claim that “the phenomena we review below seem
to require an enormously complex kind of reflexive reasoning about other agents’
desires”, and then they quote Goody (1978: 12) according to which this reasoning,
with its roots in interpersonal ritual, “may be fundamental in an evolutionary sense
to social life and human intelligence”. Although the complex reasoning they make
an appeal to is characterized as “reflexive”, nevertheless there is a clear commitment
to the commonsense psychological view according to which agents engage in prac-
tical reasoning about their own and others’ goals and desires.
While, as I said, the literature on politeness is not focused on the precise cog-
nitive mechanisms through which implicit meaning is conveyed, one possibility is
that the above personal-level descriptions based on rational reconstruction have to
be taken seriously. Recanati (2004), in particular, has proposed a cognitive account
according to which the inference from explicit to implicit content is performed
by processes that – in Carston’s (2007: 19) words – “are to be understood as part
264 Marco Mazzone
of a more general theory of human action interpretation and so as having the

philosophically central property of being rational, personal-level (as opposed to
subpersonal) processes”. Recanati characterizes such processes as conscious, but
only in a dispositional sense: they need not actually be conscious, it is enough that
the speaker has the ability to access consciously the whole inference from explicit
to implicit meaning whenever required. This move is intended to reduce the cost of
being committed to a personal-level explanation, in a way that is similar to Brown
and Levinson’s appeal to a reflexive kind of practical reasoning. Since pragmatic
processing is performed with a fluency and rapidity that is hardly compatible with
fully conscious reasoning, it would not be wise to commit oneself to any such strong
assumption. Unfortunately, speaking in terms of reflexive reasoning is not very
informative from a cognitive point of view, and Recanati’s proposal of a disposition-
ally conscious form of practical reasoning does not make things better: if a process
is only dispositionally conscious, it is not actually conscious at all (Carston 2007;
Mazzone 2013a), so it is not clear how it differs from straightforward automatic
(subpersonal) processing. However, I submit, such proposals correctly point to the
need for something intermediate between purely automatic and fully conscious
processing, as I will make clear below.
Given the difficulty of tailoring a plausible personal-level account, one might be
tempted to choose the subpersonal alternative. This is in fact the approach taken by
RT. In its view, the process at play in pragmatic understanding is automatic rather
than conscious. Interestingly, this is not the only aspect for which the theory qual-
ifies as subpersonal: its key notion of relevance is not defined, as it would be more
obvious, in terms of being relevant with respect to goals. As Sperber and Wilson
(1987: 742) put it:
Some commentators […] think our definition of relevance fails to do justice to
pretheoretical intuitions. Utterances are relevant, they feel, to purposes, goals, top-
ics, questions, interests, or matters in hand. We define relevance in a context and
to an individual. We say what a context is, how it is constructed and how, once
constructed, it affects cognition and comprehension. One reason we did not set
out to define relevance to a purpose, goal, and so on, is that we had no idea how
to answer the analogous questions for any of these terms.
Whatever the reasons, relevance theorists do not characterize utterance under-

standing as a special case of practical reasoning about goals at the personal level.
Rather, they appeal to a subpersonal mechanism operating as follows. First, the
mechanism produces interpretations, that is, inferences from explicit meaning
(together with accessible contextual assumptions) to implicit meaning (together
with other accessible contextual conclusions). Second, those interpretations are
assessed against expectations of relevance, with relevance defined in terms of the
balance between cognitive effects and efforts: relevance varies as a direct function
of the number of cognitive effects and an inverse function of the processing effort
expended. When an interpretation satisfies the expectations of relevance raised by
the utterance, that interpretation is accepted by the pragmatic module as the one
(presumably) intended by the speaker. In practice, considerations of processing
effort are intended to ensure that the process is cognitively sustainable, but what
really matters for an interpretation to be accepted as relevant is the amount of its
cognitive effects.
Incidentally, starting at least from Sperber and Wilson (1998), a different route
has been explored according to which expectations of relevance concern the type
instead of (or in addition to) the amount of cognitive effects (for a wider discus-
sion of this point, see Mazzone 2015a: 5). This is strictly related to the proposal
that pragmatic processing does not operate sequentially, by means of only forward
inferences from the explicit meaning to the intended cognitive effects (passing
through the selection of appropriate contextual assumptions). On the contrary,
there is a parallel process based on both forward and backward inferences, thanks
to which explicit content, contextual assumptions and cognitive effects are mutually
adjusted to each other:
Mutual adjustment is seen as taking place in parallel rather than in sequence. The
hearer does not first identify the proposition expressed, then access an appropriate
set of contextual assumptions and then derive a set of cognitive effects. In many
cases […], he is just as likely to reason backward from an expected cognitive effect
to the context and content that would warrant it.
(Wilson 2004: 353, emphasis mine)
In other words, the idea is that specific cognitive effects may be suggested by the
communicative situation and then constrain interpretations backwards, in that they
provide expectations about what the speaker intends to convey with her utterance.
We will put to use these considerations later. For the time being, it should be
clear that expectations about types of cognitive effects are quite different from ex-
pectations about their amount and, according to RT’s definition of relevance, it is
the latter that are at the core of comprehension. In other words, the line of thought
prevailing in RT is that comprehension is driven not by expectations of specific
effects, but rather by generic expectations that utterances are as informative as
possible. To summarize, in RT pragmatic comprehension is not conceived as a case
of practical reasoning, it is conceived instead as a subpersonal process guided by
expectations of informativeness.
The relevance theoretic account of politeness proposed by Jary (1998) is per-
fectly in line with this view. His main point is that Brown and Levinson (1987) are
wrong in proposing a Gricean account of politeness, according to which “whenever
266 Marco Mazzone
the so-called polite forms/strategies are used then an additional layer of meaning
is necessarily communicated” (Jary 1998: 2). Brown and Levinson’s main assump-
tion is that polite forms/strategies provoke a deviation from rational efficiency, as
defined by Grice’s norms (the Cooperative principle and maxims), and such devi-
ation has the effect of producing an implicature which communicates politeness.
Jary rejects, in general, the view that communicative behavior is based on rational
norms whose violation triggers further attempts to rationalize it: as emphasized by
RT, rational reconstructions of this sort should not be taken literally in a cognitive
perspective. Specifically, Jary’s (1998: 2) objection to Brown and Levinson (1987)
is that “polite forms often go unnoticed by the participants”, which implies that
they mostly involve processes of social indexing – that is, the automatic selection
of forms that are appropriate to social contexts – and not of strategic politeness (a
distinction proposed by Kasper 1990). In other words, politeness mostly involves
“unselfconscious adherence to sociocultural constraints” instead of selfconscious
and “strategic manipulation of these towards egocentric ends” (Jary 1998: 18). So,
in the end, as Fraser (1990) put it, politeness is often anticipated instead of (infer-
entially) implicated.
This line of reasoning, however, applies to most but not to all cases of polite-
ness. What about the cases in which, according to Jary himself, there is strategic
manipulation of sociocultural constraints? His answer is that these are cases in
which the speaker “provides evidence for the hearer that she holds him in higher
[or lower] regard than he had assumed to be mutually manifest” (Jary 1998: 8). As a
consequence of such violations of expectations, certain aspects of the speaker’s ver-
bal behavior “are relevant enough to be worth the hearer’s attention” (Jary 1998: 9,
Figure 1). In sum, according to Jary, in those cases politeness is indeed implicated,
but this is because polite forms/strategies are used in an informative way: they are
relevant in RT’s sense, and therefore they can be processed by the subpersonal
mechanism described by RT, which is allegedly guided by expectations of relevance
(informativeness).
Thus, in neither case would interpretation of politeness require personal-
level conscious reasoning: either it would be performed by automatic processes
of social indexing, or by the subpersonal mechanism for utterance interpretation
described by RT.
One final remark is in order. Jary (1998) repeatedly characterizes non-strategic
politeness as unselfconscious compliance with sociocultural constraints, and once
in the paper he characterizes, in contrast, strategic politeness as “selfconscious
exploitation of sociocultural constraints” (Jary 1998: 7, emphasis mine). But this
mention of consciousness is hardly compatible with his conclusion that even the
understanding of strategic politeness is ensured by a subpersonal mechanism, not
by personal-level conscious reasoning. 1 To be sure, relevance theorists have in-

sistently claimed that utterances are designed so as to worth, and in fact attract,
the hearer’s attention. However, what they really meant is that utterances are de-
signed so as to be processed by the appropriate subpersonal mechanism based on
relevance. Despite the huge evidence that utterances attract conscious attention
proper, and the quite reasonable hypothesis that this actually has a role to play in
their processing, RT always describes pragmatic understanding as a wholly subper-
sonal, automatic affair (the entire issue is addressed in detail in Mazzone 2013b).
Nevertheless, the whole distinction between strategic and non-strategic politeness,
which is crucial for Jary’s (1998) line of thought, seems built on the intuition that
the former (but not the latter) involves conscious appreciation of the implicature.
In other words, politeness appears to be implicated (instead of merely anticipated)
when it is consciously noticed.
Let us take stock. Some researchers have proposed, in line with Grice’s ap-
proach, personal-level explanations for implicitness, but then they typically tend to
minimize their commitment to full-fledged conscious processing – for instance, by
characterizing the process as only dispositionally conscious, or as a case of reflexive
reasoning. On the other hand, relevance theorists and others have proposed sub-
personal explanations of implicitness, although our analysis of Jary (1998) shows
how difficult it is to completely dismiss the intuition of a strategic, conscious ap-
preciation of implicitness. As we will see in a moment, a way out of this dilemma is
the hypothesis that the production/understanding of implicit contents is based on
mechanisms that are intermediate between purely automatic and fully conscious
processing.
3. An automatic/controlled continuum of processing
In my view, recent accounts of pragmatic processing have run into difficulties

because of the lack of an intermediate notion between presence and absence of
consciousness (Mazzone 2013a, 2013b, 2016). There is in fact a crucial difference,
which should not be disregarded, between the conscious execution of each step of
1. One possibility is that while strategic politeness is processed subpersonally by the hearer, the
speaker produces it instead through selfconscious reasoning. However, Jary (1998) insistently
characterizes non-strategic politeness in terms of something that goes unnoticed by the “interact-
ants” (the “participants” to the conversational exchange) – that is, by both speakers and hearers.
This seems to mean that, conversely, strategic politeness is noticed (and it is strategic in that it is
noticed) by speakers and hearers as well. And this in turn suggests that the processing of strategic
politeness is selfconscious on the part of the hearer as well.
268 Marco Mazzone
a complex cognitive task (be it verbal reasoning or whatever) and cases in which
the automatic activation of representations is only sustained consciously. Let us call
the former “conscious reasoning” and the latter “conscious attending”. Conscious
reasoning is an especially slow and effortful mode of processing, while conscious
attending is the “low-effort mode” – in Kahneman’s (2011: 24) words – that ac-
companies, sustains and monitors most of our cognitive activities. As Satpute and
Lieberman (2006: 91) put it, after all “we live in a supraliminal world” for the most
part of our lives, while we are (and can be) engaged in effortful conscious reasoning
only occasionally.
In this perspective, automatic and conscious processes mostly occur togeth-
er and cooperate with each other (more details in Mazzone and Campisi 2013),
whereas fully automatic processing on the one hand and fully conscious reasoning
on the other are only extreme cases. According to a widely accepted model, auto-
matic and controlled (conscious) processes can be conceived as different dynamics
through which representations in associative memory are activated. While auto-
matic processes consist in weak and local spreading activation with fast rise time
and rapid decay time, conscious processes consist instead in strong activation of
distant areas in the cortex (typically, both frontal and posterior sensorimotor ar-
eas), which form long-distance loops that are self-sustaining and thus enduring
in time (see for instance Dehaene et al. 2006; Dehaene 2014; see also Mazzone
2015c, Section 2). In other words, consciousness is essentially a mechanism for the
maintenance of activation, which makes it possible even for representations with
very indirect connections to act on each other. Barsalou (1982) distinguished three
cases with regard to in/directness of connections. When (first case) connections
between representations are sufficiently strong and direct, the association alone
can make the representations accessible to each other. For less direct or weaker
associations (second case), representations are reciprocally accessible only if con-
texts contribute further activation. Finally (third case), when the associations are
significantly indirect or weak, simple spreading activation may not be sufficient, and
this is when conscious attention is required in order to ensure sustained activation.
Importantly, this third case does not require that each of the related representations
and their connections be consciously attended, which would make this a case of
conscious reasoning. It is sufficient that conscious attention focuses on some of the
related representations, so that their activation becomes high enough to sustain the
non-conscious activation of the others. Conscious reasoning can be considered as
an extreme of this continuum (a fourth case, as it were), which is required when
indirectness is such that not even conscious attention may suffice.
According to the picture sketched above, different dynamics of activation may
run through the same network of associative memory, enabling representations at
different distances to interact with each other. The four dynamics described – simple
spreading activation; contextual spreading activation; conscious attention; con-

scious reasoning – form a one-directional scale with regard to their range of ac-
tivation: any dynamic has a shorter range than the ones at its right, and a longer
range than the ones at its left. Thus, representations whose distance to each other
is very short can interact by means of any of those dynamics, while representa-
tions at longer distances can only interact through progressively more powerful
(longer-range) ones. For instance, conscious reasoning allows the interaction of
representations that are very indirectly related, but also of representations at any
shorter distance, while pure spreading activation allows the interaction only of
representations at very short distance. All this said, as we noted above, any cognitive
process of reasonable complexity can be described as a case of conscious attention,
since it is likely to be a cooperation of pure spreading activation, contextual spread-
ing activation and consciousness in the low-effort mode, while conscious reasoning
is a rather costly and rare activity. Another important qualification is that what we
treat as four clear-cut cases might be better conceived as points of reference in a
continuum of possibilities.
A crucial implication of this approach is that by changing the dynamics through
which representations get activated, the rational structure is left unchanged, in
that it depends on the organization of associative memory. Generally speaking,
associative memory is presumed to be so organized as to enable inferences by way
of pattern completion (Barsalou 2005: 628). Mazzone (2014a, Section 3) shows
that pattern completion can be described as a form of generalized modus ponens,
provided that patterns have a schematic, hierarchical structure. Schemata can in
fact be described as higher-level, complex representations that prescribe the or-
ganization of their components at a lower level, so that the activation of one of
such components allows the prediction of the others through the activation of the
schema. In practice, the schematic organization of associative memory makes it
possible, in principle, to draw inferences in both automatic and conscious modes
of activation. 2
In the light of the above considerations, both purely personal and purely sub-
personal accounts of implicitness are hardly plausible. It is reasonable to assume
that pragmatic inferences are rarely if ever processed in a fully automatic or fully
conscious mode: the normal case is that they are consciously attended, meaning
that only some of the representations involved – depending on the circumstanc-
es – are focused consciously, while others are not attended. Moreover, although
2. The picture is complicated by the fact that, in the cognitive literature, the term “inferential”
is sometimes restricted to mean conscious inferential processing. The consequences of this ambi-
guity are discussed at length in Mazzone 2015c; 2016). However, this will do no harm for present
purposes.
270 Marco Mazzone
according to this view what is consciously noticed may change from one circum-
stance to another (see Mazzone and Campisi 2013), this is not entirely uncon-
strained. As we said, the distance between representations may be so great that only
long-range modes of processing, involving some degree of consciousness, allow
them to interact. Conversely, when representations are (more) directly connected
there is probably no need of conscious processing and therefore, ceteris paribus, a
(partly) automatic mode of processing is presumably preferred.
This may explain the intuition appealed to by Jary (1998) with regard to the
distinction between strategic politeness and politeness as social indexing. Since
polite forms are highly standardized, their original implicatures are to some extent
short-circuited (in the sense of Morgan 1978): they are no more calculated starting
from what used to be the literal meaning. Instead, polite forms acquire a new –
social-indexed – literal meaning. For instance, “Can you pass me the salt?” acquire
the new meaning of a request to pass the salt in relatively formal situations. This
amounts to saying that the polite form and the implicated content become directly
associated with each other and therefore the former can activate the latter also by
means of automatic processing. At the same time, polite forms are open to be used
strategically, not only in principle (in that any transition between strongly/directly
associated representations can also be attended consciously), but also because the
indirecteness of polite forms is made noticeable by certain linguistic facts. Let me
exemplify with a different case than politeness.
In a discussion of figurative meanings, Mazzone (2014b: 49) points out that
“the frequency with which a word form conveys a meaning is just one factor among
many in its conventionalisation”. Among the other factors involved, mention is
made of the frequency of previously lexicalized meanings. For instance, in met-
aphorical sentences such as “Juliet is the sun”, the phrase “the sun” is not likely
to be strongly associated – let us say – to the concept BEAUTIFUL, not even if
it were frequently used in that sense: the phrase continues to be basically used to
express the concept SUN, a fact that contributes (together with other factors) to
prevent complete lexicalization of a new meaning in favor of the more indirect
route. Something similar seems to hold for polite forms. For instance, although
the polite use of “can” seems to be highly lexicalized, the primary use is no less
frequent. This implies that the primary meaning remains strongly associated to the
word and competes for activation with the secondary one, thus making the former
easily accessible and salient, and its relationship with the latter easy to be noticed.
The general point is that, although short-circuiting of polite forms makes available
a more direct route for the activation of the secondary meaning, the less direct
route is also strong (though possibly to a lesser extent), and thus easily accessible
in any circumstance.
On the basis of the above considerations, Jary (1998) seems right in pointing
out that politeness is not always strategic, and that it is strategic to the extent that
it is consciously noticed. However, I have also provided reasons to think that there
is no clear-cut boundary between strategic and non-strategic uses of politeness.
Moreover, it is a quite reasonable assumption that utterance understanding always
requires conscious attention, that is, a mode of processing that involves both per-
sonal and subpersonal components – although in non-strategic politeness attention
is not focused (by definition, as it were) on polite forms.
Before turning to motives for implicitness, there is a final point that needs to be
addressed. Associative memory, as we saw, is organized in hierarchies of schemata,
which are responsible for the rational structure of both automatic and conscious
processing. This consideration also applies to goals, since (as argued in Mazzone
2015a: 8–9) they can be conceived as components of schemata of a specific kind:
associative structures composed of ends, means and related situations, thanks to
which the activation of any of these components is transmitted – as a form of
pattern completion – to the others. In Mazzone (2015b) I gave reasons for believ-
ing that those schemata are involved in the associative construction of means-end
chains that cross the boundary between perception and action, and in particular
between observed and executed actions and therefore between the self and the
others. In practice, in joint activities agents spontaneously construe and update
on a continuous basis shared intentional contexts, in which sub-plans of different
agents mesh with each other. In a sense, each agent integrates the others’ goals into
her own action planning. As we will see in a moment, this is of crucial importance
for the emergence of implicitness in human communication.
4. Motives for implicitness
As Grice (1957: 387) observed, we easily understand that “a man who calls for a
‘pump’ at a fire would not want a bycicle pump”. Although this is an example of ex-
plicit (not implicit) meaning construction through disambiguation, a more general
lesson can be drawn from it. The reason why we easily understand what a speaker
means in such a situation is that, in Grice’s words, his communicative intention has
to “fit with some purposes he obviously has”. In the example, mention of a fire is
sufficient to trigger an entire scene, complete with the (most frequent) goals that
an agent in that situation may be expected to have.
The general fact is that we are continuously, not to say compulsively, focused
on recognizing the activities that people perform around us and on attributing
goals/intentions to those people. This is part of what Tomasello (2009: xvi) called
272 Marco Mazzone
“a special kind of […] intelligence, comprising species-unique social-cognitive

skills and motivations for collaboration, communication, social learning, and oth-
er forms of shared intentionality”. The distinction between social-cognitive skills
and motivations is of some interest to us. Not only are we endowed with skills such
as the ability to form shared intentional contexts meshing our and others’ goals.
We are also concerned about others and specifically about their needs and goals.
According to Tomasello, this “empathetic concern” (Tomasello 2009: 12) is at the
root of “the especially cooperative structure of human, as opposed to other primate,
social interaction” and also of the “fundamentally cooperative structure of human
communication, as initially discovered by Grice” (Tomasello 2008: xi).
In contrast with Grice’s Cooperative Principle, Relevance Theory has often
emphasized that in human communication selfish strategies are no less important
than cooperative ones. To some extent this is certainly true. Interestingly, however,
there is strong evidence that infants behave altruistically before they learn to behave
selfishly, and they definitely show a strong empathetic concern for the others’ goals
(and expect a similar concern in others) at the age in which they learn to speak (a
review of the evidence is in Tomasello 2009).
We will turn to motivations and empathetic concerns later. Let us first address
the cognitive side of the issue by turning back to Grice’s example. As a general rule,
it is safe to assume that the presence of a fire is extraordinarily salient for, and in
fact consciously attended by, anyone who is present. Then, the strong activation of
the concept of fire can be expected to activate the goal of extinguishing it, which in
turn can be expected to activate, though to a minor extent, the means frequently
used to pursue it (thanks to means-end schemata). Thus, when the hearer listens
to the word “pump” and, as a consequence, the activation spreads from the lexical
entry to the related concepts, one of these happens to be already activated as a
possible means for extinguishing the fire. We can generalize this analysis in the
following way. Goals are often made manifest by the observed situations (as in the
fire case) via our knowledge of human activities and needs, based on schematic
representations variously described as scripts, activity types (Levinson 1992) and so
on. As a consequence, representations of goals are often the starting point both of
utterance production and understanding, by means of backward inferences of the
kind described by RT. For this reason, the speaker usually does not even notice (in
context) the ambiguity of “pump”, since the right meaning is present to her mind
first (as an appropriate means for the salient goal), and the word is recovered from
memory only after. Luckily, however, the hearer is no more likely to be misled by
the ambiguity, since he too, at least in part, follows the backward route from the
goal to the means and then to the word.
What does this imply with regard to implicit meaning? Let us consider the
following exchange (from Wilson and Carston 2007, modified):
Peter: Will Sally look after the children if we get ill?

Mary: Sally is kind and caring.
Mary’s reply clearly involves an implicature, since an affirmative answer to Peter’s

question is implicitly conveyed by the thought that SALLY IS KIND AND CARING.
Now, Mary can be thought to produce the utterance by a backward route starting
from the implicit meaning, or, more precisely, from a more abstract representation
of the goal at stake. In fact, the previous utterance addressed to Mary by Peter makes
manifest a goal that Peter has, which in the normal case is consequently adopted by
Mary as well: the goal to receive a yes/no answer to his question. In other words,
this goal can be expected to become part of a shared intentional context, due to our
general ability and tendency to share goals.
Generally speaking, in most cases the communicative goal of the speaker is
already established (or at least strongly constrained) either by previous interac-
tions with the hearer or by the shared non-linguistic situation, and the speaker
reasons backwards from that goal. This is far from surprising: in theory of action
it has been proposed that the normal route by which agents plan their actions is
by backward inferences from goals to the means which have proved apt to pursue
them in the past (Hommel 2003; Hommel et al. 2001; Mazzone 2015b). In the
above example, the case is a little different, in that the goal – the implicit meaning
to be communicated – is not entirely determined until the explicit meaning is
conceived. Specifically, in order to determine whether the appropriate answer to
Peter is either a yes or a no, Mary has to look backwards for reasons to conclude
in favor of one of them. The point, however, is that the general goal to be pursued
(giving a yes/no answer) is already established and it is part of the shared inten-
tional context: in this sense it is more expected (less informative) than the reasons
for deciding between the affirmative and the negative answer. Thus, when the
speaker finds a convincing reason whose relationship with either an affirmative
or a negative answer is easily accessible, she does not need to make explicit which
is the intended meaning.
In sum, my suggestion is that, in most situations, communicative goals are
already salient and manifest to both speakers and hearers, though only in an ab-
stract form: consequently, they are less informative than the information needed
in order for them to be more precisely determined. Finding this information then
acts as the “point of arrest” of the inferential process by which the communicative
intention is formed. As a matter of fact, that information is – by construction, as
it were – relevant with respect to the mutually manifest communicative goal to be
pursued: it is in fact the information needed to pursue that goal.
For another example, we can safely assume that a telephone ringing in a house
is salient and readily activates in anyone listening the goal to answer it. Thus, if
274 Marco Mazzone
Peter and Mary are both at home, any of them can be expected either to answer
or to acknowledge, by backward reasoning, the factors preventing him/her from
doing it. This backward reasoning is in the service of action, in the first place.
However, due to our spontaneous tendency to think in a “we-mode”, both Peter
and Mary presumably construe a shared intentional context, in which the goal to
answer the phone is represented as a common goal. As a consequence, whenever
Peter (for example) happens to access reasons why he is not able to pursue that goal,
he immediately recognizes those reasons not only as something that prevents him
from taking action, but also as information that is relevant to Mary. In sum, since
the goal of answering the phone is already part of the shared intentional context,
it is as if both Mary and Peter were considering (and asking to each other) the yes/
no question “Will you answer the phone?”, which makes a yes/no answer strongly
expected. This is why this answer can be left implicit, while giving instead reasons
why one cannot answer the phone (“I’m in the shower!”).
A couple of remarks are in order.
First, the present proposal has important points in common with RT. This is
true, however, only under a certain interpretation of the notions of “relevance” and
“expectations of relevance”. I propose that an interpretation is relevant with respect
to goals, meaning that the former needs to have the appropriate rational relation-
ship with the latter. This is quite different from conceiving relevance in terms of
the amount of cognitive effects. Besides, in my account interlocutors are guided
by expectations of relevance in the sense of expectations about the communicative
goals to be pursued, while no mention is made of expectations about the amount
of cognitive effects (implications) to be obtained. In sum, my account is entirely
compatible with RT’s proposals that (i) utterances cause expectations about the
type (but not the amount) of conclusions to be drawn, and (ii) those expectations
consequently trigger backward inferences, which in my proposal are based on ra-
tional relationships with goals.
Second, I conceive the mechanism described above as a subpersonal expla-
nation of implicitness. This is not in the sense that the relationship between the
explicit meanings and the communicative goals they pursue in context cannot be
consciously noticed. As I propose instead, any inference that can be drawn automat-
ically can also be attended consciously in the appropriate circumstances (depending
on the dynamic of activation). The suggestion, however, is that explicit meaning acts
as a point of arrest for backward inferences, and this goes unnoticed by people: it
is simply the way in which the mechanism works. In other words, speakers hardly
notice that they have been implicit, and definitely they do not acknowledge that
they have been implicit in order to select a more relevant formulation.
While in this sense speakers are not aware of informativeness as a reason for
implicitness, in another sense there are circumstances in which speakers seem to
choose more indirect and informative formulations deliberately. In particular, this
seems to be the case with humor and certain figurative uses of language. They can
be thought of as manifestations of wit that are usually received with signs of social
approval. In the light of our previous considerations, communication is based on
goal sharing and the construction of shared intentional contexts. Now, when the
speaker shows her wit by choosing an indirect and interesting way to pursue her
communicative goal, this is readily recognized by the hearer as an invitation to
share the speaker’s exercise of intelligence and enjoy it. In other words, the basic
sense of affiliation (the “empathetic concern”) that is present in most exchanges to
a low degree, as a sort of standard mode, can be increased by sharing a social game
of exhibition of intelligence.
To be sure, since the construction of shared intentional contexts can always
be exploited to pursue selfish goals, linguistic exhibitions of intelligence can also
be used for dominance (versus affiliation) purposes. In her study of family con-
versations at dinner, whose general conclusion is that – contrary to a popular be-
lief – women’s speech are not more indirect than men’s speech, Rundquist (1990)
observes that most cases of men’s indirectness are humorous. Based on the analysis
of her material, she suggests that fathers’ humor might often be motivated by the
fact that men “tend to posture, to want a dominant role in the conversation, to want
to be center-stage” (Rundquist 1990: 513). In other words, fathers tend to assume
a more authoritative and dominant role, and this is marked by an exhibition of
intelligence conveyed by humor. In other cases, fathers seem to resort to humor as
a strategy for appeasement and mitigation (Rundquist 1990: 514), which obviously
is not in conflict with the observation that they aim to lead the way.
In sum, both in affiliation and dominance cases, one important motivation for
humor seems to be the deliberate exploitation of indirectness for purposes of social
adjustment. If this is correct, there is here a common ground between humor and
strategic politeness 3 whose strict relationship has been noticed but little analyzed.
My suggestion is that both are rooted – in Tomasello’s (2009: xvi) words – in our
“special kind of intelligence”, that is, in the tendency to share means-end structures
and on the related concern for the goals of others.
3. In line with our previous considerations, for “strategic politeness” I intend deliberate use
of polite forms, irrespective of whether there is violation of expectations about sociocultural
constraints (as proposed instead by Jary 1998; for a criticism of his proposal, see Escandell-Vidal
1998).
276 Marco Mazzone
How, specifically, does the notion of goal sharing help us in explaining strategic
politeness (leaving aside the fact that strategic politeness may be short-circuited
in the course of time)? It is common to explain indirect refusals in terms of the
desire to avoid face-threatening acts, and indirect requests in terms of providing an
opportunity for a negative answer – again, in order to avoid being directive. But, to
start with, why should a direct refusal appear threatening and an indirect one less
so? One possibility is that by expressing an overarching goal that provides a reason
for the refusal, the speaker increases the degree of affiliation with the hearer. In
practice, the speaker exploits the human tendency to share goals by focusing on a
goal (e.g., “I have an exam tomorrow”) that is incompatible with the one proposed
by the interlocutor (e.g., “Will we go to the movies tonight?”). So, in a sense the
impossibility to pursue the latter becomes a matter of mutual evidence and mutual
concern, instead of a choice on the part of the speaker. On the contrary, a direct
refusal would show indifference toward the spontaneous tendency of the interloc-
utor to take part in the decision-making process. On the other hand, providing an
opportunity for a negative answer is a way to mitigate a request in that it amounts
to showing interest for the interlocutor and his goals. When the speaker says “Can
you tell me what time it is?” she is not so much providing opportunities for a refusal
as showing interest for possible goals of the hearer that are incompatible with the
one she proposed. This is a way to show affiliation, so as to communicate closeness
to the interlocutor.
5. Conclusions
I do not presume to have covered all the possible cases of implicitness. The bottom
line, however, should be clear. If unity and coherence in science are virtues, I hope
to have shown that there might be, if not a single motive for implicitness, at least
a single ground for the different motives described above. Implicitness may be the
result both of subpersonal factors – such as the fact that communicative intentions
are often formed backward, starting from expectations about goals to be pursued
in context – and of personal factors – such as the fact that individual goals are
shared by the participants to communicative interactions, with effects of affiliation
that can be increased (or decreased) by manipulating the means-end structure. A
plausible hypothesis is that those different motives have a common ground, which
is a mechanism of practical reasoning based on associative activation of means-end
structures that crosses the boundaries between self and others.
References
Barsalou, Lawrence W. 1982. “Context-independent and Context-dependent Information in

Concepts.” Memory & Cognition 10: 82–93. doi: 10.3758/BF03197629
Barsalou, Lawrence W. 2005. “Situated Conceptualization.” In Handbook of Categorization
in Cognitive Science, ed. by Henry Cohen, and Claire Lefebvre, 619–650. St. Louis, MO:
Elsevier. doi: 10.1016/B978-008044612-7/50083-4
Brown, Penelope, and Stephen Levinson. 1987. Politeness: Some Universals in Language Usage.
Brumark, Åsa. 2006. “Non-observance of Gricean Maxims in Family Dinner Table Conversation.”
Carston, Robyn. 2000. “The Relationship between Generative Grammar and (Relevance-theoretic)
Pragmatics.” Language & Communication 20: 87–103. doi: 10.1016/S0271-5309(99)00020-8
Carston, Robyn. 2007. “How Many Pragmatic Systems Are There?” In Saying, Meaning, Referring.
Essays on the Philosophy of François Recanati, ed. by Maria Frapolli, 1–17. New York: Palgrave.
Dehaene, Stanislas. 2014. Consciousness and the Brain. Deciphering How the Brain Codes Our
Thoughts. New York: Viking Press.
Dehaene, Stanislas, Jean-Pierre Changeux, Lionel Naccache, Jérôme Sackur, and Claire Sergent.
2006. “Conscious, Preconscious, and Subliminal Processing: A Testable Taxonomy.” Trends
in Cognitive Sciences 10: 204–211. doi: 10.1016/j.tics.2006.03.007
Escandell-Vidal, Victoria. 1998. “Politeness. A Relevant Issue for Relevance Theory.” Revista
Alicantina de Estudios Ingleses 11: 45–57. doi: 10.14198/raei.1998.11.05
Fraser, Bruce. 1990. “Perspectives on Politeness.” Journal of Pragmatics 14: 219–236.
doi: 10.1016/0378-2166(90)90081-N
Geurts Bart, and Paula Rubio-Fernández. 2015. “Pragmatics and Processing.” Ratio 28: 446–469.
doi: 10.1111/rati.12113
Goody, Esther (ed.). 1978. Questions and Politeness: Strategies in Social Interaction. Cambridge:
Grice, Paul. 1957. “Meaning.” Philosophical Review 66: 377–388. doi: 10.2307/2182440
Grice, Paul. 1989. Studies in the Way of Words. Cambridge, MA: Harvard University Press.
Hommel, Bernhard. 2003. “Planning and Representing Intentional action.” The Scientific World
Journal 3: 593–608. doi: 10.1100/tsw.2003.46
Hommel, Bernhard, Jochen Müsseler, Gisa Aschersleben, and Wolfgang Prinz. 2001. “The Theory
of Event Coding (TEC): A Framework for Perception and Action Planning.” Behavioral and
Brain Sciences 24: 849–937. doi: 10.1017/S0140525X01000103
Jary, Mark. 1998. “Relevance Theory and the Communication of Politeness.” Journal of Pragmatics
30: 1–19. doi: 10.1016/S0378-2166(98)80005-2
Kahneman, Daniel. 2011. Thinking, Fast and Slow. New York: Farrar, Straus and Giroux.
Kasper, Gabriele. 1990. “Linguistic Politeness: Current Research Issues.” Journal of Pragmatics
14: 193–218. doi: 10.1016/0378-2166(90)90080-W
Lakoff, Robin. 1973. “The Logic of Politeness: Or Minding Your p’s and q’s.” In Papers from the
Ninth Regional Meeting of the Chicago Linguistics Society, 292–305. Chicago: University of
Chicago Press.
Leech, Geoffrey. 1983. Principles of Pragmatics. London: Longman.
Levinson, Stephen C. 1992. “Activity Types and Language.” In Talk at Work: Interaction in
Institutional Settings, ed. by Paul Drew, and John Heritage, 66–100. Cambridge: Cambridge
University Press.
278 Marco Mazzone
Mazzone, Marco. 2009. “Pragmatics and Cognition: Intentions and Pattern Recognition in Context.”
International Review of Pragmatics 1: 321–347. doi: 10.1163/187730909X12535267111615
Mazzone, Marco. 2011. “Schemata and Associative Processes in Pragmatics.” Journal of Pragmatics
43: 2148–2159. doi: 10.1016/j.pragma.2011.01.009
Mazzone, Marco. 2013a. “Automatic and Controlled Processes in Pragmatics.” In Perspectives on
Linguistic Pragmatics, ed. by Alessandro Capone, Franco Lo Piparo, and Marco Carapezza,
443–467. New York: Springer. doi: 10.1007/978-3-319-01014-4_18
Mazzone, Marco. 2013b. “Attention to the Speaker. The Conscious Assessment of Utterance
Interpretations in Working Memory.” Language & Communication 33: 106–114.
doi: 10.1016/j.langcom.2013.01.001
Mazzone, Marco. 2014a. “Crossing the Associative/Inferential Divide: ad hoc Concepts and the
Inferential Power of Schemata.” Review of Philosophy and Psychology 5: 583–599.
doi: 10.1007/s13164-014-0201-8
Mazzone, Marco. 2014b. “The Continuum Problem: Modified Occam’s Razor and the
Conventionalisation of Meaning.” International Review of Pragmatics 6: 29–58.
doi: 10.1163/18773109-00601003
Mazzone, Marco. 2015a. “Constructing the Context through Goals and Schemata: Top-down
Processes in Comprehension and Beyond.” Frontiers in Psychology 6: 1–13.
doi: 10.3389/fpsyg.2015.00651
Mazzone, Marco. 2015b. “Pragmatics and Mindreading. Forward and Backward Inferences in
Shared Intentional Contexts.” Intercultural Pragmatics 12: 289–307. doi: 10.1515/ip-2015-0015
Mazzone, Marco. 2016. “What Kind of Associative and Inferential Processes? A Response to
Rubio-Fernández (2013).” International Review of Pragmatics 8: 143–161.
doi: 10.1163/18773109-00801006
Mazzone, Marco. In press. “Trivializing Modularity. A Radical Associative-Representational
Account of Cognition.” Epistemologia.
Mazzone, Marco, and Emanuela Campisi. 2013. “Distributed Intentionality: A Model of
Intentional Behaviour in Humans.” Philosophical Psychology 26: 267–290.
doi: 10.1080/09515089.2011.641743
Morgan, Jerry. 1978. “Two Types of Convention in Indirect Speech Acts..” In Syntax and Semantics,
vol. 9, Pragmatics, ed. by Peter Cole, 261–280. New York: Academic Press.
Rundquist, Suellen. 1990. “Indirectness in Conversation: Flouting Grice’s Maxims at Dinner.”
Proceedings of the Sixteenth Annual Meeting of the Berkeley Linguistics Society, 22–46.
Linguistic Society of America.
Satpute, Ayai B., and Matthew D. Lieberman. 2006. “Integrating Automatic and Controlled
Processing into Neurocognitive Models of Social Cognition.” Brain Research 107: 86–97.
doi: 10.1016/j.brainres.2006.01.005
Sperber, Dan, and Deirdre Wilson. 1986/1995. Relevance. Communication and Cognition. Oxford:
Blackwell.
Sperber, Dan, and Deirdre Wilson. 1987. “Precis of Relevance: Communication and Cognition.”
Behavioral and Brain Sciences 10: 697–754. doi: 10.1017/S0140525X00055345
Sperber, Dan, and Deirdre Wilson. 1998. “The Mapping between the Mental and the Public
Lexicon.” In Language and Thought, ed. by P. Carruthers and J. Boucher, 184–200. Cambridge:
Thomas, Jenny. 1995/1998. Meaning in Interaction. An Introduction to Pragmatics. London and
New York: Longman.
Tomasello, Michael. 2008. Origins of Human Communication. Cambridge, MA: MIT Press.
Tomasello, Michael. 2009. Why We Cooperate. Cambridge, MA: MIT Press.
Wilson, Deirdre. 2004. “Relevance and Lexical Pragmatics.” UCL Working Papers in Linguistics
16: 343–360.
Wilson, Deirdre, and Robyn Carston. 2007. “A Unitary Approach to Lexical Pragmatics:
Relevance, Inference and ad hoc Concepts.” In Pragmatics, ed. by Noel Burton-Roberts,
230–260. New York: Palgrave. doi: 10.1093/oxfordhb/9780199234769.003.0018
chapter 13
Implicature and the inferential substrate
Michael Haugh
University of Queensland, Brisbane
What is said is always silhouetted

by the penumbra of the unsaid
(Welch 2005)
Implicatures are generally conceptualised as additional thoughts, beliefs, intentions

and desires that are intended by the speaker to be recognised by the hearer as
intended by the speaker. However, implicatures are not simply cognitive con-
structs, but in being accomplished by persons in interaction, are arguably social
actions in their own right. In this chapter, it is proposed that a proper account of
implicature needs to be developed with respect to the broader inferential sub-
strate from which implicatures arise. It is suggested that while the inferables that
make up this inferential substrate generally remain embedded, that is, where
talk progresses without participants orienting to the action that is accomplished
through the inferable as an object of interactional business, they may, on occa-
sion, be exposed by those participants as the focus of interactional business in
that sequence. It is then proposed that a range of practices license participants
to expose an inferable (or set of inferables), including instances of “prompting”,
where a speaker positions another participant to make a pre-emptive offer through
reporting (possible) troubles, difficulties or needs. It is concluded that given
through implicating participants can implicitly orient to both moral and relational
concerns, our understanding of implicatures should not be divorced from the
inferential substrate of interaction in which they are invariably locally situated.
Keywords: implicature, inference, intention, social action, interactional, pragmatics
1. Implicature and inference
The notion of implicature was first introduced by the philosopher Paul Grice in
order to discuss instances in which what a speaker means goes beyond the meaning
literally expressed by a particular utterance in communication (Grice [1975]1989).
doi 10.1075/pbns.276.13hau
282 Michael Haugh
The term ‘implicature’ was specifically coined by Grice to exclude logical impli-
cations and focus on the ordinary language sense of implying as “expressing indi-
rectly”, “insinuating” and “hinting at” and so on. Many scholars have combined this
work on implicature with Grice’s (1957, 1969) seminal work on speaker meaning
as attained through the recognition of communicative intentions, leading to the
common view that implicatures pertain to additional thoughts, beliefs, intentions,
desires and so on that are intended by the speaker to be recognised (by the hearer)
as intended by the speaker (Bach 2012; Bianchi 2013; Horn 2004; Levinson 2000),
a view advocated in passing by Grice ([1978]1989: 49) himself, as Dynel (2014)
points out. In some cases, what is modelled is the hearer’s inferences about what the
speaker intended to implicate (e.g. Sperber and Wilson 1995). However, implica-
tures themselves have generally been modelled as speaker-intended, and debates in
relation to the theorisation of implicature have been dominated by instances where
what is being implicated is relatively determinate.
One issue with the received view of implicature in pragmatics, however, is
that Grice’s ([1975]1989, [1978]1989) original account of implicature was arguably
sketched in much broader terms than how it has subsequently been taken up in the
field. For a start, in introducing the notion of implicature, Grice ([1975]1989: 24)
made an important distinction between “implicate” (cf. imply, mean), “implica-
ture” (cf. implying) and “implicatum” (cf. what is implied). In other words, Grice
([1975]1989) carefully distinguished between implicature as the process of impli-
cating something in addition to what is said, and implicatum as what is taken as
meant by the speaker. Saul (2002a) further advocates drawing a further distinc-
tion between “audience implicature”, that is, what the hearer thinks the speaker
implicated, and “utterer implicature”, that is, what the speaker thinks he or she
implicated. In Grice’s original terms, this amounts to making a distinction between
speaker implicatum and hearer implicatum. Without drawing such distinctions,
the understandings of speakers and hearers as to what has been implicated remain
conflated (Kecskes 2013; Saul 2002a).
Grice ([1975]1989) also acknowledged that what is implicated may involve a
“disjunction of specific explanations” (: 40); in other words, it is not certain which
of the possible things that may have been implicated were actually meant by the
speaker. A similar point has been made by Terkourafi (2013: 201–202) who notes
cases where a “conjunction of inferences” may arise, for instance, in communi-
cation amongst intimates. This passing observation from Grice was significantly
extended in the Relevance theoretic claim that implicatures lie on a continuum
from “strong” through to “weak”. While speakers may be held primarily responsi-
ble for inferences that are “strongly implicated”, there are a range of “weak impli-
catures” for which the hearer is primarily responsible for deriving (Sperber and
Wilson 1995: 56–57).
Chapter 13. Implicature and the inferential substrate 283
It has thus been argued by some that inferences about the intentions of speak-
ers are not necessary for implicatures to arise (Gauker 2001; Saul 2002a, 2002b),
or at the very least tracing such putative intentions in conversational interaction is
no straightforward matter (Haugh 2008, 2012). 1 Indeed, it has become increasingly
apparent that implicatures can vary in their degree of pragmatic explicitness with
respect to the transparency of their illocutionary point, target and semantic content
(Culpeper and Haugh 2014), as well as with respect to the degree to which speakers
are held to be committed to or accountable for the implicature in question (Haugh
2015a). The latter is not, however, simply a matter of the (perceived) cognitive real-
ity of that speaker, but a moral matter as well, given we are held accountable for the
real-world consequences of what we are taken to be meaning (Haugh 2013a). For
that reason, implicatures are not simply cognitive constructs, but in being accom-
plished by persons in interaction, are arguably social actions in their own right.
In this chapter, it is proposed that a proper account of implicature, whether
one elects to remain with a more tightly focused, formal account, or one takes a
more liberal view of the range of phenomena involved, needs to be developed with
respect to the broader inferential substrate from which implicatures arise. The
chapter begins, in the following section, by first briefly outlining what is encom-
passed by the inferential substrate of conversational interaction. We then move
to consider, in section three, how we might start tapping into this inferential sub-
strate, drawing from interactions in which participants are getting acquainted. It
is proposed that the inferences made available in the course of such interactions
generally remain “embedded” within other action trajectories, although they may
be “exposed” by those participants as the primary focus of interactional business
in some cases. It is then suggested, in section four, that our understanding of
the relationship between implicature and the broader inferential substrate can
be interrogated through careful analyses of the various interactional practices by
which implicatures are accomplished by participants. The ways in which speakers
can “prompt” pre-emptive offers from others – what are commonly treated as
“requestive hints” in pragmatics – through reporting (possible) troubles, diffi-
culties or needs is taken as a case in point. The chapter concludes by considering
the implications for pragmatics of admitting that the inferential substrate from
which instances of implicature emerge constitutes a legitimate object of study in
its own right.
1. It is worth noting that some scholars have also suggested that “generalised implicatures” or
“conventional implicatures” are not necessarily speaker-intended in the same way as “particula-
rised implicatures” (e.g. Davis 1998; Haugh 2015a; Terkourafi 2005). However, the focus in this
chapter is primarily on the latter.
284 Michael Haugh
2. The inferential substrate of interaction
There are various ways in which we as ordinary speakers demonstrate awareness

of the richer inferential substrate that underpins many of our communications
with others. We talk of the “subtext” of a conversation, complain that someone is
speaking in “code”, or speculate about what a person might have really “meant”
by a seemingly innocuous compliment or observation. The results of early work
by Garfinkel (1964), in which he asked his students to note down what was really
meant in apparently innocuous exchanges with family members, are quite reveal-
ing in that respect. A single utterance was interpreted as meaning dozens of differ-
ent things by those participants. It is clear that we often take others to be meaning
much more than what they might appear to be meaning at first glance.
Consider the following dialogue taken from the television comedy, Big Bang
Theory. Penny has just witnessed Lesley, a work colleague of Leonard’s, ask him to
join the physics department string quartet.
(1) Penny: So hey, your friend’s really cute, anything going on with you two?
Leonard: Lesley? No, no-oh, what are you kidding?
Sheldon: He asked her out once, it was an embarrassing failure.
Leonard: Thank you Sheldon
Sheldon: Oh, I’m sorry, was that supposed to be a secret?
Penny: Oh, that’s too bad, you guys would make a cute couple.
(“The hamburger postulate”, Big Bang Theory, Season 1, Episode 5, 22 October
2007, director: Andrew Weyman, writers: Dave Goetsch and Steven Molaro)
Penny’s comment that Leonard and Lesley “would make a cute couple” subse-
quently sets Leonard off thinking about what she might have “meant” by that.
(2) Leonard: What did Penny mean, “you’d make a cute couple”?
Sheldon: Well, I assume she meant that the two of you together would constitute
a couple that others might consider cute. An alternate, though some-
what less likely interpretation is that you could manufacture one. As in,
“Oh, look, Leonard and Leslie made Mr. and Mrs. Goldfarb! Aren’t they
adorable?”
Leonard: If Penny didn’t know that Leslie had already turned me down, then
that would unambiguously mean that she, Penny, thought I should ask
her, Leslie, out, indicating that she, Penny, had no interest in me asking
her, Penny, out; but because she did know that I had asked Leslie out,
and that she, Leslie, had turned me down, then she, Penny, could be
offering me consolation – “That’s too bad, you would have made a cute
couple…” – while thinking, “good, Leonard remains available.”
Sheldon: You’re a lucky man, Leonard.
Leonard: How so?
Sheldon: You’re talking to one of the three men in the Western Hemisphere capa-
ble of following that train of thought.
Leonard: Well, what do you think?
Sheldon: I said I could follow it, I didn’t say I cared.
(“The hamburger postulate”, Big Bang Theory, Season 1, 2007)
While Sheldon offers a fairly straightforward, albeit unlikely, account, Leonard

develops a seemingly complex set of hypotheses about what Penny may have meant.
Notably, on his own admission, he cannot be sure what Penny might have intended,
but nevertheless speculates about what can be inferred from her comment with
regards to her potential (romantic) interest in him. This kind of example does not
sit well within most formal accounts of implicature given there is indeterminacy
with respect to Penny’s intentions here. Yet clearly what Penny has said motivates
Leonard to draw his own inferences. This is not uncommon. In daily interaction, it
is clear that multiple inferences can often be drawn from what is said (or not), how
it is said, and when it is said (or not). In some cases, like the one above, we can be
left wondering whether or not something was implied by another speaker, or what
exactly they might have been intimating.
The question, then, is what counts as a licensed inference (or set of inferences),
that is, the inferences for which we can legitimately hold the speaker in question
accountable. The standard response in many accounts of implicature that we can
only hold speakers accountable for those inferences that they intended us to draw is
arguably too narrowly focused. It restricts us to inferences that are (unequivocally)
speaker-intended, leaving cases like the one above without proper explanation.
There is evidently a much richer inferential substrate underpinning conversational
interaction than many standard accounts of implicature would generally admit. 2
The challenge for us as ordinary participants in conversational interaction is thus
to figure out what can be legitimately inferred. This process is not just a matter of
“reading off ” the intentions of others, but is arguably invested with some degree of
agency on the part of participants. We can sometimes choose which inferences we
draw, and perhaps more importantly, we can choose which inferences we indicate
we have drawn to others. There is also inevitably variance amongst individuals in
regard to the degree that they attend to the inferential substrate from which implica-
tures emerge, depending on a whole range of factors, ranging from our own individ-
ual personality and values, through to our relationship with the person in question.
2. I am not claiming that scholars developing such accounts are unaware that these other sorts
of inferences exist, but simply that such accounts only admit a formally constrained set of data
to be explained by the theory in question. Relevance theory does come closer in proposing the
notion of mutual manifest assumptions (Sperber and Wilson 1995). However, it does not admit
these as part of what is communicated.
286 Michael Haugh
Yet despite awareness of the rich set of inferences available to participants in

communicating with each other on the part of ordinary speakers, there have only
been a limited number of attempts in pragmatics to theorise this broader inferen-
tial substrate. Bertuccelli Papi’s (2000) proposal that what is left “unsaid” can be
divided into that which is “implicated” and that which is “subplicit” is perhaps one
of the most significant to date. While she defines implicatures, following the stan-
dard view, as inferences that are “recognised by the hearer as intentionally meant
by the speaker”, subplicit meanings are defined as those inferences that “glide into
the mind of the hearer as side effects of what is said or not said” (Bertuccelli Papi
2000: 147). Examples of the latter are argued to include “alluding”, “insinuating”,
“veiled implying”, and so on. More recently, Ariel (2016) has proposed that along-
side speaker-intended “strong implicatures” and “particularised implicatures”, there
are also “background assumptions” and “truth-compatible inferences”, which are
not necessarily speaker-intended. In some respects, then, it is becoming increas-
ingly apparent that we need to go beyond the time-honoured notion of implica-
ture and expand our analytical lens to consider what Raclaw (2013) refers to as
“inferables”, namely, that which can be inferred from what is said (or not said),
how it is said and when it said (or not) in conversational interaction, as a broader
phenomenon of interest in pragmatics.
One of the challenges facing the analyst in attempting to tap into this rich
inferential substrate, however, is that these sorts of inferables very often rely on
particular background knowledge, experience or understanding on the part of
those participants to arise. Take, for instance, the case of “implicitude”, a practice
where participants reach “an implicit agreement on what can remain unsaid and
on recognition and acceptance of the intention to leave it unsaid as a mutually
shared premise for the construction of what is meant” (Bertuccelli Papi 2000: 153).
It is obvious that for the casual observer, this kind of subtext may well remain
completely impenetrable. Other forms of subplicit meaning, such as “insinuations”
or “allusions”, can remain equally as opaque to observers (and no doubt many
participants on occasion). It is for this reason, perhaps, that researchers in prag-
matics have retreated from the study of the inferential substrate in favour of that
which is seemingly more readily tractable to the observer-analyst. As Levinson
(2013: 115–116) points out, “researching the unsaid is not an easy task: one needs
to show that both participants are oriented to something not occurring and delib-
erately kept off-stage as it were”. The challenge facing the analyst is how to study
this rich inferential substrate in a systematic and ordered manner that is ultimately
amenable to formal theorisation.
Aside from judicious use of participant observation (including making record-
ings of interactions in which one, as the analyst, is also a participant-observer),
another way in which to address the seeming intractability of the inferential
substrate is to begin by studying interactions in which the background or history

between participants is relatively controlled and knowable to the analyst. In the
following section, we examine the ways in which the inferential substrate can be
observed to emerge in the course of interactions in which participants are getting
acquainted (Haugh forthcoming; Svennevig 1999).
3. Embedded and exposed inferables in initial interactions
Borrowing from Jefferson’s (1987) distinction between “embedded” and “exposed”

repair, we can talk of inferences that remain “embedded” in talk-in-interaction as
those where the talk in progress continues without the participants orienting to the
action that is accomplished through that inferable as an object of interactional busi-
ness. Inferences that are “exposed”, in contrast, are those inferables that are treated
by the participants as the focus of interactional business in that sequence. While the
latter are relatively tractable to the analyst, given next speakers can display uptake of
an inferable through their response (Sacks, Schegloff and Jefferson 1974), thereby
exposing it, embedded inferences remain are somewhat less analytically tractable as
they often rely on particular background assumptions, experiences and so on that
may only be known or knowable to those participants in question.
Given the important role that background assumptions play in affording infer-
ences from what is said (or not), how it is said, and when it is said (or not), initial
interactions between participants who have not previously met arguably constitutes
a useful site of analysis in attempting to tap into inferables that remain embedded.
This is because the analyst is on a relatively equal footing with the participants in
relation to the background assumptions and interpersonal histories that inform the
inferential substrate. One can thus observe this inferential substrate being co-con-
structed “online” as it were as the participants get acquainted over the course of the
interaction (Haugh forthcoming).
We can see the way in which inferables are embedded in back-and-forth of
interaction in the following set of excerpts from an initial interaction between Peter
(an American) and Sally (an Australian) who are both in their early twenties. In
this first excerpt, Sally is following up her question about where Peter comes from.
The interpolated aspiration in uttering “N(h)ew (York City)” (Potter and Hepburn
2010) makes available the inference that comparing “Brisbane” and “New York
City” is potentially problematic. 3
3. In this chapter, excerpts from recordings of spontaneous conversational interaction (as

opposed to scripted talk from film or television shows) are transcribed according to standard
conventions in conversation analysis (Jefferson 2004).
288 Michael Haugh
(3) CAAT: AmAus02: 0:44

30 S: an:d how are you finding Brisbane then,
31 compared to N(hh)ew York City.
32 P: ah, (0.2) it’s a lot slower, a lot more
33 laid back, bu::t (1.0) it’s it’s more like
34 a place to live whereas New York is more
35 a place to vacation.
36 S: o:h yeah.
37 P: yeah (.) [°I think°]
38 S: [°h m m° ]
39 P: it’s nice, there’s beaches everywhere::,
40 people are ni:ce, .hh the cars aren’t trying
41 to hit you [they actually stop at the c]ross = 
42 S: [huh:: he he he he ]
43 P:  = [walks? .hh]
44 S:    [.hh .hh] .hh
45 P: that’s something I’(h)m n(hh)ot used
46 t(hh)o. .hh.hh
47 S: yeah.
48 P: ye(h)ah.
What is notable here is Peter’s characterisation of Brisbane as a place to “live” as

opposed to New York which is a place one is better to only visit (lines 33–35). He
goes on to offer an account for his characterisation of Brisbane compared to New
York, namely that there are lots of beaches (line 39) and “people are ni:ce” (line 40).
Offering positive assessments of the place one is evidently not originally from
with a person who is clearly a local is not atypical of these sorts of exchanges.
However, Peter then goes on to reveal that not only is Brisbane a good place to live
in general, but also that he has plans to come to live in Australia himself, as we can
see in the following excerpt.
(4) CAAT: AmAus02: 1:46

72 S: yeah. so why’d you come ↑here?
73 P: um:: (0.5) I was thinking about moving to
74 Australia, just in general.
75 S: [↑oh yeah.]
76 P: [cos uh I ] like the culture and everything
77 I’ve done a few like research projects on it
78 for school. ·hh but I was like I should
79 pro::bly visit there befo(hh)re I(hh):
80 decide I just wanna go there.
81 S: yeah.
82 P: but? yeah. (0.2) still mi:ght. I figur’ this
83 would b-be a place I’d wanna settle down at.
84 (0.7)
85 P: but I gotta travel a little bit more.
86 S: ↑oh ye[ah.
87 P: [like. (0.2) not done yet.
88 S: yea:h
Notably, Peter first outlines his possible intentions (lines 73–74), then offers an
account for why he is thinking about moving to Australia (lines 76–77), before
finally addressing Sally’s original question (line 72) as to why he is now currently
in Australia (lines 78–80). Peter then repeats that he is thinking about settling in
Australia (line 82–83). Following this self-disclosure, there is a hearable gap at a
point where some form of affiliative response might be due from Sally but is not
forthcoming (line 84). Peter then goes on to qualify his intentions about moving
to Australia by claiming he intends to travel more first (lines 85, 88). In this way,
Peter characterises himself as a young American in Australia who has a potential
interest in staying in Australia long-term. This makes available a number of infer-
ences given these are two young university students who may be single and open
to a relationship (although neither participants knows about the relationship status
of the other just yet).
Indeed, it subsequently emerges that Sally has had an American boyfriend
previously. Given Peter has revealed his possible desire to live in Australia this
again makes available a number of inferences about what their overall project here
in getting acquainted might turn out to be.
(5) CAAT: AmAus02: 2:59

121 S: I had a: (.) American boyfriend,
122 like- (0.7) ah:: two years ago,
However, this unprompted self-disclosure (Haugh and Carbaugh 2015) is embed-

ded in a response to a prior complaint sequence in which Peter has claimed partic-
ular expertise with respect to knowing how to use the terms “ketchup” and “tomato
sauce” properly. This prior sequence is initiated when Peter complains about the
way in which Australian speakers of English use the term “tomato sauce” instead
of “ketchup” (lines 88–89, 91, 93).
(6) CAAT: AmAus02: 2:13

88 P: you know you guys say tomato sauce instead
89 of ketchup but
90 S: hhh [ahuh huh huh huh] [.hh huh huh
91 P: [it kinda gets on] my nerves [a lot but-
92 (0.2)
93 P: [it’s alright it’s] alright hmhem
94 S: [huh huh . hh ]
95 S: you know what, I was in Woolie’s the other day,
96 and I saw um:(0.7) tomato sauce, and
97 ketchup (0.5) next to each other. and I was
98 trying to read the bottles, to see, like- what
99 is the difference in ingredients? = is it actually
100 different? but- .hh
101 P: .hh ketchup’s ketchup, tomato sauce is like
290 Michael Haugh
102 pasta sauce, (.) or pizza sauce. (0.2) to me.

103 S: ah. [we call that (0.2) [tomato paste.] yeah.
104 P: [( ) (0.2) [tomato paste.]
105 P: and tomato paste is that without anything in it.
106 jus- just pure tomato puree and everything,
107 [and that’s
108 S: [hhhehehehehe .hh
109 P: and that’s like a whole ‘nother realm, that’s what
110 you use to make °the tomato sauce.°
111 S: ah::
112 P: hehhehheh
113 S: [yeah]
114 P: [.hhh] yeah, just a little th- um, I own a few
115 pizza places, my family does? back in the
116 [States.] so (0.5) y- yeah, = 
117 S: [↑okay. ]
118 P:  = so hem we kinda (0.7) we’re particular about our
119 sauces.
120 S: mm. (0.2) [>I can imagine.<]
121 P: [mheh heh heh ]
Sally passes up on the opportunity to affiliate with this complaint (Drew and Walker
2009), and instead offers an account for this possible “misuse” by Australians,
namely, there is no clear distinction drawn between “ketchup” and “tomato sauce”
by speakers of Australian English (lines 95–100). However, rather than taking this
to be an indication of a possible “dialectical” difference (i.e. between different va-
rieties of English), Peter takes this to be an opportunity to “tell” Sally how they
should be differentiated (lines 101–110), despite an intervening attempt to establish
her own independent knowledge of differences (line 103). In so doing he implicitly
takes an epistemic stance that he knows something about which she is evidently
unclear, tipping the epistemic gradient in his favour (Heritage 2012). Notably,
while Sally’s response, “ah:: yeah” (lines 111 and 113) acknowledges Peter’s claim,
it does so in a minimal way (Jefferson 1993), and what is hearably withheld is an
acknowledgement of the correctness of Peter’s claim (Gardner 2007). This occasions
an account that appears designed to explain Peter’s original complaint, alongside
a claim to particular “expertise”, namely, his involvement in a family pizza busi-
ness (lines 114–119). Once again, Sally only minimally acknowledges this claim
to epistemic authority in line 120 with “mm” followed by a conversational aside,
“I can imagine”, where there is equivocality as to whether this is being framed as
a form of soliloquy or “self talk” (Goffman 1981), whereby she is taking an ironic
footing (Clift 1999).
Sally subsequently reports on her prior experiences with an American boy-
friend following Peter’s complaint about the misuse of these terms.
(7) CAAT: AmAus02: 2:33

121 S: I had a: (.) American boyfriend, like- (0.7) ah::
122 two years ago, and he would always pick on (.)
123 everything I said, and say you’re saying it wrong,
124 you’re saying it wrong [cuz I’m a linguistics = 
125 P: [hmhmhm heh heh
126 S:  =student studying (0.2) .hh you know = 
127 [prescriptive and descriptive language = 
128 P: [hmh
129 S:  = and>blah blah blah<  I was like-
130 P: hm
131 (0.2)
132 S:  =>No I’m not saying it wrong.<  hh:haha
133 shut up ha s(hh)ort of thing.
134 P: if you understand, who cares? hmh
135 S: yeah exactly. yeah, so.
136 P: hem hem hm
Notably, embedded within this telling is a complaint is directed at her ex-boy-

friend, an American, who used to “pick on everything” she said (lines 121–124),
and also used to claim epistemic primacy based on his claimed expertise as a stu-
dent of linguistics (lines 124–127). Sally indicates the annoyance she felt about
this in characterising this talk from her ex-boyfriend in a dismissive manner, as
“blah blah blah” (line 129), and then animating prior occasions when she told him
to “shut up” (line 133). What is inferable from Sally’s complaint, given it follows
Peter’s complaints about “incorrect” usage of terms, is that although the target is
ostensibly her ex-boyfriend, Peter himself could also be a target (Haugh 2015b).
Indeed, he initially responds with an open-mouthed facial expression indicating
surprise (Sendra, Kaland, Swerts and Prieto 2013). However, the inference that
this complaint is also directed at Peter is defeasible, and in subsequently affiliating
with it (line 134), Peter treats it as such, that is, as directed only at the ex-boyfriend.
Up until this point in the interaction, then, a complex set of potentially con-
flicting inferences have been made available. On the one hand, Peter has expressed
interest in remaining in Australia, thus making available a range of inferences
around future relationships. Sally has also indicated possible interest in Americans
in disclosing that she has had an American boyfriend. However, the latter self-dis-
closure is embedded in a complaint sequence about her previous boyfriend’s
behaviour, which seems remarkably similar to what Peter has just been evidently
doing in the prior sequence. We can see here, then, the beginnings of a complex
inferential substrate that underpins what follows in this interaction.
What is important to note, however, is that these inferences have remained
“embedded” (Jefferson 1987). That is, the interaction-thus-far has progressed with-
out the participants orienting to the possible actions that are accomplished through
292 Michael Haugh
these inferables as objects of interactional business in their own right. In that sense,
they do not formally constitute implicatures per se. Yet this set of inferables never-
theless forms part of the inferential substrate that has been co-constituted through
the course of this interaction thus far.
What then marks out an inferable as open to formal characterisation as an
implicature is arguably that it has been “exposed” (Jefferson 1987) as the focus of
interactional business in that sequence. We can observe just such a case of this
in the following excerpt from the same initial interaction as above. Prior to this
excerpt Peter and Sally have been talking about foods found in Australia but in
the United States. Peter has indicated he does not generally like these “Australian”
foods, but then claims he likes to eat kangaroo.
(8) CAAT: AmAus02: 4:01

166 P: I like kangaroo though.
167 you [guys ] don’t eat it here though for some reason.
168 S: [yeah?]
169 (.)
170 S: ↑n[o::.]
171 P: [don ]’t know why. [it’s ] wild there’s no hormones
172 S: [(I::- ]
173 P: there’s ºnoº, there’s nothing i:n it?
174 it’s the healthiest meat you can eat? [bu:t?]
175 S: [yeah?]
176 (.)
177 S: I guess [because] (0.2) it’s a bit of a
178 P: [he hem?]
179 cultural thing? like where I’m from people
180 feed kangaroo meat to dogs? so: (0.2) ye:ah.
181 P: ºI guess I’m a dog.º
182 S: [I- no: I don’t mean dogs ] eat it
183 P: [I like dogs so that’s okay]
184 I just mean because [(I know )] that my fath-
185 P: [>that’s okay<]
186 S: you know like my stepdad’s a farmer? and that’s
187 what he does? it’s just (0.2) I wouldn’t
188 ºevenº (0.2) like I would eat it, but I just
189 wouldn’t go out of my way to ºeat itº
190 sort of thing
Peter is here evidently seeking an account as to why Australians do not eat kan-
garoo meat (line 167). While Sally appears to be initiating just such an account
(line 172), Peter goes on to list reasons why it would be good for Australians to do
so using a recurrently observed three-part structure for listing (Jefferson 1991),
that is, (1) it’s wild, (2) there are no hormones in it, there’s nothing in it, and (so)
(3) it’s the healthiest meat you can eat. In so doing, Peter attends to the potentially
offensive inference that arises from soliciting an account through a soliloquy that
is why-interrogative implicative. As Bolden and Robinson (2011: 115) point out,
“why-interrogatives index a measure of epistemic access to the accountable event
and communicate a stance that it does not accord with common sense and thus
is possibly inappropriate or unwarranted”. It follows that co-implicated with this
account solicitation is an implicit criticism directed at Australians, of which Sally
is evidently one, for not eating kangaroo meat (when evidently they should do so),
or at least such a criticism is inferable.
Sally then proceeds to offer just such an account, couching it as a “cultural
thing” (lines 177, 179), before explaining that kangaroo meat is seen as something
to be fed to dogs (lines 179–180). The upshot of this account that is projected by
“so” is hearably attenuated (Schegloff 2007: 83) and thus remains unstated, leaving
it to Peter to respond in a way that demonstrates his understanding of what has
been left unsaid (Raymond 2004). In this case, it can be readily inferred that if
kangaroo meat is fed to dogs, then it is likely to be considered not fit for human
consumption. In that sense, it constitutes a mild rebuke of Peter’s prior implicated
criticism. This viewpoint is qualified, as previously noted, however, as one held
amongst people in the area from which Sally comes (line 179), a “small country
town” (line 62, data not shown), and by scalar implicature, not necessarily a view
held by all Australians.
Peter responds to Sally’s account by making as if he thinks she has just implied
that he must be a dog if he likes to eat kangaroo meat (line 181). This is delivered
sotto voce to indicate the potential delicacy of this inference (Lerner 2013). In so
doing, Peter construes Sally has having implicated something (ostensibly) offen-
sive. However, the marked change in prosodic contour also offers a cue to the
interpretation of this ostensibly serious taking of offence as a form of teasing, spe-
cifically, jocular pretence (Haugh 2016a). Specifically, he appears to be launching
an opportunistic tease given this “insult” is inferable, although in reality does not
seem to be licensed by her account.
Sally, however, does not treat it as a jocular tease, but rather responds seriously,
disavowing an intent on her part to imply just that, as well as displaying signs of
embarrassment through stuttering speech and a somewhat incoherent claim that
dogs don’t eat it (line 182) (Goffman 1956). Overlapping with this disavowal is the
delivery of an absolution, “that’s okay”, by Peter for this offence (Robinson 2004),
which pre-empts the incipient apology from Sally, along with an account, namely,
that he likes dogs and so apparently doesn’t mind being labelled as one (line 183).
The latter, once again, suggests this absolution is being delivered somewhat tongue-
in-cheek, thereby introducing ambiguity as to whether he is once again mildly
teasing her. However, as Sally goes on to elaborate her disavowal (line 184) through
attributing that view on kangaroo meat as one held by her father and not herself
294 Michael Haugh
(lines 186–190), Peter repeats this absolution (line 185), thereby shifting back into
an unambiguously serious frame.
We would, of course, be hard pressed to claim that Sally intended to imply
that she thinks of Peter as a dog since he has admitted to liking to eat kangaroo
meat. She strenuously denies any such intention. Peter’s response thus appears to
constitute an instance of quite deliberate misconstrual of what she meant (Clark
1997). However, Peter is nevertheless able to make as if he is taking this as what she
has implied because it is inferable from what she has said, and in light of what has
just been said. In (mis)construing what she has implied in this way, Peter thereby
“exposes” an inferable that would (and probably should) have otherwise remained
buried in the inferential substrate.
In this section, it has been proposed that in the course of interaction, an infer-
ential substrate is cumulatively co-constituted through the layering of inferables
that can be drawn from what is said (or not), how it is said, and when it is said (or
not). For the most part these inferences remain embedded, that is, participants do
not orient to the action(s) accomplished through that inferable as an object of the
interactional business of the talk-in-progress. In some cases, however, an embed-
ded inference may be exposed by the participants, thereby becoming the focus of
interactional business in the sequence in question.
In the following section, the ways in which implicatures can emerge from this
inferential substrate are discussed.
4. Implicature as social action
Implicatures have generally been theorised in pragmatics as additional thoughts,

beliefs, intentions, desires and so on that are intended by the speaker to be rec-
ognised (by the hearer) as intended by the speaker (Bach 2012; Bianchi 2013; Horn
2004; Levinson 2000), or what the hearer thinks the speaker intended to implicate
(e.g. Sperber and Wilson 1995), as we previously noted. One consequence of this
focus on implicature as a form of (speaker) meaning is Grice’s original insight that
implicatures arise through implicating, and thus arguably constitute a form of social
action (Haugh 2015a), has been largely neglected.
In this section, it is proposed that implicatures arise through participants
exposing an inferable, that is, by treating an inference as a focus of interactional
business. On this view, it follows that not responding to that which has been impli-
cated constitutes an accountable absence, and is thus an action in and of itself. To
disattend an embedded inferable, in contrast, does not constitute an “official” or
“accountable” absence (Sacks 1992: 294; Schegloff 1968). Thus, despite its decid-
edly “unofficial” status, the inferential substrate is arguably the engine room from
which implicatures emerge. It is suggested that by studying implicatures as a form
of social action, that is, as an interactional accomplishment on the part of partici-

pants, we are able to eschew the ontological circularity that a definition of implica-
tures as speaker-intended can create for the observer-analyst (Haugh 2008).
To characterise implicature as inferables from the inferential substrate that are
exposed by participants raises questions, however, about what licenses participants
to expose them in the first place. Rather than identifying different species of impli-
cature based on the original typology proposed by Grice ([1975]1989), and debating
the theoretical value of these different types, the aim of the interactional approach
to implicature is to study the various practices by which implicatures arise. Haugh
(2015a), for instance, proposes these practices include, although are not limited
to, attenuating, withholding, soliciting/prompting, pre-empting, and disattending.
In the remainder of this section, we will focus on just one of those practices,
“prompting”, where a speaker positions another participant to pre-emptively ini-
tiate an action sequence, such as an offer, invitation, proposal, account and so on,
thereby locating agency, and thus responsibility for that action with that other per-
son rather than the speaker (Haugh 2016b, 2017; cf. Haugh 2015a: 262). We focus,
in particular, on instances where through reporting (possible) troubles, difficulties
or needs hearers may be prompted to make pre-emptive offers.
As Drew (1984: 136) notes, reportings can make a particular social action
“relevant, but not obligatory” next. By reporting (possible) troubles, difficulties or
needs a speaker is inviting but not necessarily mobilising occasion “an opportunity
for a recipient to offer assistance” (Kendrick and Drew 2014: 111). Verbal requests
or pre-requests, in contrast, mobilise responses that orient to the prior turn as
a request or request-implicative (Stivers and Rossano 2010). This difference is
important because it follows that “requesting” does not have the same moral impli-
cations for participants as “prompting an offer”. The latter allows the addressed
recipient to display a greater degree of agency (Enfield 2014), because he or she is
given the opportunity to volunteer the object or assistance in question (Kendrick
and Drew 2016). Prompting offers in this manner also displays a claim on the part
of the speaker to a lower degree of entitlement on their part, with respect to the
recipient in question, to be making a request in the first place (Lindström 2005;
Stevanovic and Peräkylä 2012). Haugh (2015a: 310) suggests that the interpersonal
implications of this form of prompting, including whether specific instances might
be evaluated as ‘polite’, ‘impolite’ and so on, derives from these locally accomplished
deontic stances (Haugh 2015a: 310), thereby accounting for the consistent finding
that such “hints” are not always evaluated positively by participants (Blum-Kulka
1987; Terkourafi 2011; Sifianou 1993, 1997).
In the following excerpt taken from a telephone conversation between Lesley
and her mother, they are talking about the mother’s impending visit. Lesley’s sug-
gestion that her mother not bring too much luggage has emerged from their prior
discussion about her mother taking the bus.
296 Michael Haugh
(9) Holt: X(C):1:2:7: 8:18

472 Les: ↑don’tbring↑ too much
473 (0.2)
474 Mum: ↑Oh that’s w’I wz g’n t’tell↓you I’m:- (.) I’m
475 ↑not bringing an (.) I’m not bring any big
476 j:ump- (.) big ca:rdigans
477 (0.2)
478 Les: No: y[ou c’n
479 Mum: [or a dressing gow:n I- .a (0.2) I [c’n ba-
480 Les: [No-
481 (.)
482 Les: You c’n[↑borrow mi↓:ne
483 Mum: [( )-
484 Mum: Yeh I thought w’l I’d brin:g the small case (‘n
485 [just)
486 Les: [eeYes do.
487 Mum: ( ).
488 Les: .hh
489 (0.2)
490 Mum: Okay love
Lesley’s mother responds to her daughter’s suggestion (line 472) by reporting her
intention to not bring any heavy clothing (lines 474–476, 479). This intention is
couched as one that had already been formulated prior to Lesley’s advice (“I was
going to tell you”, line 474). This frames it as something that was already “on her
mind”, and thus makes available the inference that the advice is unnecessary (i.e.
she already knows that it is better to keep her luggage light given she will be trav-
elling on the bus). In addition, in framing her intention in this way, the mother
also makes available the inference that she had spent some time considering what
clothes to bring, and thus that there is the potential for trouble to arise if she does
not bring appropriate clothing along with her.
Lesley exposes this inferable through a subsequent (pre-emptive) offer that
she can lend her mother such clothing (lines 478, 480–482). In so doing, she con-
strues her mother as possibly “hinting” that she wanted just such an offer from
her daughter. The mother downplays this possible inference, however, through an
equivocal response to the offer. She repeats her intention to use a “small case” (line
484), thereby indicating she is likely to take up the offer if necessary as she con-
firms she will not bring any heavy clothing. In responding in that way, it remains
equivocal whether she intended to prompt her daughter to make this pre-emptive
offer (Haugh 2017).
What emerges here, then, is that an assumption that likely forms part of the
inferential substrate, namely, that the mother and daughter sometimes lend clothes
to each other, motivates an inferable, that is, the mother may need to borrow some
heavy clothing if it is cold during her visit, which is subsequently exposed through
the daughter’s offer. Whether the mother was “hinting” that she was wanting just
such an offer to be made by daughter remains ambivalent (Ogiermann 2015). In
Brown and Levinson’s (1987) terms, any possible intention that could be imputed
to the mother by Lesley remains “off record”. For an action to be accomplished
off-record means that it is accomplished in such a way that “it is not possible to
attribute only one clear communicative intention to the act…the actor leaves him-
self an ‘out’ by providing himself with a number of defensible interpretations; he
cannot be held to have committed himself to just one particular interpretation of
his act” (Brown and Levinson 1987: 211). Yet despite the equivocality about the
mother’s possible intentions, the mother and daughter have nevertheless inter-
actionally accomplished this pre-emptive offer as one that is touched off by the
mother’s prior reporting, and it is in that sense that we can say that the offer from
Lesley is prompted by her mother’s reporting.
The way in which offers of assistance can be prompted by participants report-
ing troubles is also apparent in the following excerpt taken from an interaction
between two American study abroad students, Linda and Jim, who are getting
acquainted. The sequence begins with Jim reporting that he has been offered a
job at Surfers Paradise (which is at the Gold Coast), but that he is undecided as to
whether he will actually take up the offer (lines 550–552), because the hours are
difficult (line 554).
(10) CAAT: AmAm01: 14:12

550 J: I:- I (0.2) ↑kinda got a job in [Sur↑fers] bu:t = 
551 L: [↑ni:ce. ]
552 J:  =I’m not ↑su:re if I gonna <ta:ke ↓it>
553 (0.2)
554 J: [.hh >cause] like it’s like <really bad hou:↓rs = 
555 L: [↓ni:ce:. ]
556 J:  =[>like it’s<]
557 L:  =[doing what ]
558 (.)
559 J: ah: i- (.) pizza.
560 (.)
561 L: ↑oh:
562 J: yeah it’s ah: (.) I work like the night
563 shifts on Frida- and [Saturday night?] it’s = 
564 L: [o h y e a h ]
565 J:  =like ten: P M to five A M = 
566 L:  =oh shi(hh)t [haha ]
567 J: [yea:h.] so hh I even need to
568 find someone n Surfers I can sta:y with = 
569 J:  =[like ] before the bu(hh)ses start £go:[ing£]
570 L: [yea:h] [well] = 
571 L:  =you can ↑stay with ↑me: I [live in] ↓Sur[fers.]
572 J: [°oh° ] [oh ]
298 Michael Haugh
573 sweet [that’s] awesome but= 

574 L: [(but) ]
575 J:  =↑a:lso like (0.4) if I worked every week↑en:d
576 that like completely kills >my travel time↑< 
577 and [I’m only] here ‘til July:. right?
578 L: [ yea:h ]
It subsequently emerges that the trouble with the hours is there are not any buses
that return to Brisbane (where he lives) when the job finishes, which means he
would need to find someone to stay with at Surfers Paradise until the buses start
running again (lines 562–569). Following this reporting of the troubles he faces if
he takes up the job, Linda proposes that Jim can stay with her as she has a place
in Surfers (line 571). This offer is formatted as a modal through which she gives
Jim permission to stay at her place (i.e. You can X), thereby foregrounding her
agency in initiating the offer (Curl 2006: 1257). As the offer is positioned immedi-
ately following Jim reporting that he needs to find somewhere to stay at the Gold
Coast if he is going to take up the job offer, it is apparently designed to be heard as
prompted by this reporting.
There is little reason to suspect, however, that Linda’s offer is made on the basis
of her inferences about Jim’s putative intentions in reporting these troubles, as
Jim evidently does not know that Linda lives near Surfers Paradise, or at least she
believes he does not know this (line 571). Moreover, Jim responds to this prompted
offer in a non-straightforward manner. Although he indicates appreciation of the
offer (lines 572–573), he goes on to offer additional reasons for not taking up the
job offer (lines 575–577). Thus, while his response is officially equivocal in that he
leaves it open to being taken up if he accepts the job offer, it appears to be rejec-
tion-implicative in that he indicates he may not be taking up the job, and so will
not need to stay at Linda’s place.
Linda’s pre-emptive offer is thus not grounded in inferences about putative
intentions, but rather in an observable interactional practice where reporting (pos-
sible) troubles, difficulties or needs recurrently prompts pre-emptive offers (Drew
1984; Haugh 2015a; Kendrick and Drew 2014, 2016), which themselves recurrently
occasion non-straightforward responses (Haugh 2017). These non-straightforward
responses constitute a means by which participants treat the possibility that the
first speaker intended to prompt this offer as an interactionally sensitive manner. In
other words, the way in which an inferable is exposed as part of the ongoing inter-
actional business of the talk-in-progress, thereby affording the emergence of an
implicature from the inferential substrate, is not so much a matter of the putative
intentions of those participants, but rather is grounded in recurrent interactional
practices for accomplishing implicatures in ways that enable such inferences to be
exposed and yet remain officially off-record.
It is commonly thought that the inferences underpinning implicatures are, by

definition, defeasible (Grice [1975]1989). However, subsequent work has indicated
that what is implicated is not always necessarily straightforwardly “cancellable”
(Weiner 2006). Implicatures may, of course, be the object of repair, that is, they
may be qualified or withdrawn. However, that which has been implicated by a
prior speaker nevertheless remains implicated (Haugh 2013b). The raison d’être
of implicature is thus not that they can be “cancelled”, but rather that they remain
“off record”, and so open to these kinds of repairs or negotiations by participants.
In this section, it has been proposed that implicatures are kept “off record” by the
very practices by which they are interactionally accomplished. Specifically, rather
than assuming implicatures are speaker-intended inferences, it has been suggested
that they constitute a form of social action that is tied through recurrent interac-
tional practices, such as prompting, to the inferential substrate in which they are
grounded. This allows us to remain cautious, as analysts, about attributing partic-
ular intentions to speakers given doing so in the case of implicatures is arguably
no straightforward matter.
5. Conclusion
It has been suggested here that any theory of implicature needs to be grounded
with respect to the broader inferential substrate that is co-constituted through con-
versational interaction, both at particular moments in time, and over the course
of multiple interactions. On the view of some scholars in pragmatics, these sorts
of inferables may fall outside of what they regard as the proper remit of a theory
of meaning or communication. However, this just goes to prove the limits of such
theories, as in constraining the object of study to speaker-intended inferences, such
theories are not able to readily account for the layers of inferables that evidently
interpenetrate many of our conversational interactions with others. It has thus been
suggested here that researchers in pragmatics need to study what is demonstrably
of importance to participants in conversational interaction, not only that which
seems theoretically tractable. What can be inferred from what is said (or not said),
or how it is said, and the ways in which participants position these inferables with
respect to the conversational record is an important site at which the “micro pol-
itics” of everyday, mundane interaction are accomplished. Ultimately, then, if the
notion of implicature is to remain a vibrant point for analysis and theorisation in
pragmatics, our understanding of implicatures should be not be divorced from the
inferential substrate of interaction in which they are invariably locally situated. It
follows that we need to admit a wider range of data that reflects how language is
actually used by participants in co-constituting the inferential substrate.
300 Michael Haugh
The approach sketched in this paper is, of course, only preliminary, and admit-
tedly leaves open more questions than it really answers. There remains further
work to be done to develop methods that allow us to dig further into the inferen-
tial substrate as it is interactionally accomplished in the course of conversational
interaction. While we clearly carry memories of what has been said, and not said,
in prior interactions into current interactions, it would be a mistake to assume
that this inferential substrate can be treated as mere static background knowledge
carried around in the minds of participants. There also remains work to better
understand the workings of this inferential substrate not only in interactions in
English but across other languages. Finally, there is work to be done to tease out
the various practices by which we license others to draw inferences, or attempt
to legitimise the inferences we have drawn ourselves. The extent to which such
work can align with extant, more formal approaches to implicature remains open
to question, although, in principle at least, there is no reason to assume that such
alignment cannot eventually follow. What needs to precede such attempts, how-
ever, is gaining a better understanding of how participants themselves accomplish
implicatures in interaction, and how implicatures relate to the broader inferential
substrate from which they emerge.
Acknowledgements
I would like to acknowledge the support of a Discovery grant from the Australian Research
Council (DP120100516) that has enabled the research reported in this paper to be undertak-
en. I would also like to thank Bob Sanders for our ongoing conversations and his work on the
AmAus02 data with others at the State University of New York at Albany that has prompted
many of the ideas developed in this paper. I take, of course, full responsibility for the appropri-
ateness or otherwise of these ideas.
References
Ariel, Mira. 2016. “Revisiting the Typology of Pragmatic Interpretations.” Intercultural Pragmat-
ics 13 (1): 1–35. doi: 10.1515/ip-2016-0001
Bach, Kent. 2012. “Saying, Meaning, and Implicating.” In The Cambridge Handbook of Pragmat-
ics, ed. by Keith Allan and Kasia M Jaszczolt, 47–67. Cambridge: Cambridge University
Press. doi: 10.1017/CBO9781139022453.004
Bertuccelli-Papi, M. 2000. Implicitness in Text and Discourse. Pisa: Edizoni ETS.
Bianchi, Claudia. 2013. “Implicating.” In The Pragmatics of Speech Actions, ed. by Maria Sbisà and
Ken Turner. Berlin: Mouton de Gruyter. doi: 10.1515/9783110214383.107
Blum-Kulka, Shoshana. 1987. “Indirectness and Politeness in Requests: Same or Different?” Jour-
nal of Pragmatics 11: 131–146. doi: 10.1016/0378-2166(87)90192-5
Bolden, Galina B., and Jeffrey D Robinson. 2011. “Soliciting Accounts with Why-Interrogatives in
Conversation.” Journal of Communication 61 (1): 94–119. doi: 10.1111/j.1460-2466.2010.01528.x
Brown, Penelope, and Stephen Levinson. 1987. Politeness. Some Universals in Language Usage.
Clark, Herbert. 1997. “Dogmas of Understanding.” Discourse Processes 23: 567–598.
doi: 10.1080/01638539709545003
Clift, Rebecca. 1999. “Irony in Conversation.” Language in Society 28 (4): 523–553.
doi: 10.1017/S0047404599004029
Culpeper, Jonathan, and Michael Haugh. 2014. Pragmatics and the English Language. Bas-
ingstoke: Palgrave Macmillan.
Curl, Traci S. 2006. “Offers of Assistance: Constraints on Syntactic Design.” Journal of Pragmatics
38 (8): 1257–1280. doi: 10.1016/j.pragma.2005.09.004
Davis, Wayne. 1998. Implicature. Intention, Convention, and Principle in the Failure of Gricean
Theory. Cambridge: Cambridge University Press. doi: 10.1017/CBO9780511663796
Drew, Paul. 1984. “Speakers’ Reportings in Invitation Sequences.” In Structures of Social Action,
ed. by Maxwell J. Atkinson and John Heritage, 102–128. Cambridge: Cambridge University
Press.
Drew, Paul, and Traci Walker. 2009. “Going Too Far: Complaining, Escalating and Dissafiliation.”
Journal of Pragmatics 41 (12): 2400–2414. doi: 10.1016/j.pragma.2008.09.046
Dynel, Marta. 2014. “A Survey of ‘Intercultural pragmatics’ and its Outlook on the Gricean Phi-
losophy of Communication.” International Review of Pragmatics 6: 307–319.
doi: 10.1163/18773109-00602006
Enfield, N. J. 2014. “Human Agency and the Infrastructure for Requests.” In Requesting in Social
Interaction, ed. by Paul Drew and Elizabeth Couper-Kuhlen, 35–53. Amsterdam: John
Benjamins. doi: 10.1075/slsi.26
Gardner, Rod. 2007. “The Right Connections: Acknowledging Epistemic Progression in Talk.”
Language in Society 36: 319–341. doi: 10.1017/S0047404507070169
Garfinkel, Harold. 1964. “Studies of the Routine Grounds of Everyday Activities.” Social Problems
11 (3): 225–250. doi: 10.2307/798722
Gauker, Christopher. 2001. “Situated Inference Versus Conversational Implicature.” Nous 35
(2):163–189. doi: 10.1111/0029-4624.00292
Goffman, Erving. 1956. “Embarassment and Social Organisation.” American Journal of Sociology
62: 264–271. doi: 10.1086/222003
Goffman, Erving. 1981. Forms of Talk. Philadelphia: University of Pennsylvania Press.
Grice, H. P. 1957. “Meaning.” Philosophical Review 66: 377–388. doi: 10.2307/2182440
Grice, H. P. 1969. “Utterer’s Meaning and Intention.” The Philosophical Review 78 (2): 147–177.
doi: 10.2307/2184179
Grice, H. P. 1975. “Logic and Conversation.” In Syntax and Semantics, Volume 3. Speech Acts, ed.
by Peter Cole and Jerry Morgan, 41–58. New York: Academic Press.
Grice, Paul. 1978. “Further Notes on Logic and Conversation”. In Syntax and Semantics Volume
9. Pragmatics, ed. by Peter Cole, 113–127. NY: Academic Press.
Grice, Paul. 1989. Studies in the Way of Words. Cambridge, Massachusetts: Harvard University
Press.
302 Michael Haugh
Haugh, Michael. 2008. “The Place of Intention in the Interactional Achievement of Implicature.”
In Intention, Common Ground and the Egocentric Speaker-Hearer, ed. by Istvan Kecskes and
Jacob Mey, 45–86. Berlin: Mouton de Gruyter.
Haugh, Michael. 2012. “On Understandings of Intention: A Response to Wedgwood.” Intercul-
tural Pragmatics 9 (2): 161–194. doi: 10.1515/ip-2012-0011
Haugh, Michael. 2013a. “Speaker Meaning and Accountability in Interaction.” Journal of Prag-
matics 48: 41–56. doi: 10.1016/j.pragma.2012.11.009
Haugh, Michael. 2013b. “Implicature, Inference and Cancellability.” In Perspectives on Pragmatics
and Philosophy, ed. by Alessandro Capone, Franco Lo Piparo and Marco Carapezza. New
York: Springer. doi: 10.1007/978-3-319-01011-3_6
Haugh, Michael. 2015a. Im/politeness Implicatures. Berlin: Mouton de Gruyter.
doi: 10.1515/9783110240078
Haugh, Michael. 2015b. “Impoliteness and Taking Offence in Initial Interactions.” Journal of
Haugh, Michael. 2016a. “‘Just Kidding’: Teasing and Claims to Non-Serious Intent.” Journal of
Haugh, Michael. 2016b. “Prompting as a Higher-Order Pragmatic Act.” In Pragmemes and The-
ories of Language Use, ed. by Keith Allan, Alessandro Capone and Istvan Kecskes, 167–190.
New York: Springer.
Haugh, Michael. 2017. “Prompting Offers of Assistance in Interaction.” Pragmatics and Society
8(2): 183–207.
Haugh, Michael. Forthcoming. Getting Acquainted in Interaction. Cambridge: Cambridge Uni-
versity Press.
Haugh, Michael, and Donal Carbaugh. 2015. “Self-Disclosure in Initial Interactions Amongst
Speakers of American and Australian English.” Multilingua 34 (4): 461–493.
doi: 10.1515/multi-2014-0104
Heritage, John. 2012. “Epistemics in Action: Action Formation and Territories of Knowledge.”
Research on Language and Social Interaction 45 (1): 1–29. doi: 10.1080/08351813.2012.646684
Horn, Laurence. 2004. “Implicature.” In Handbook of Pragmatics, ed. by Laurence Horn and
Gregory Ward, 3–28. Oxford: Blackwell.
Jefferson, Gail. 1987. “On Exposed and Embedded Correction in Conversation.” In Talk and
Social Organisation, ed. by Graham Button and John R E Lee, 81–100. Clevedon: Mutlilin-
gual Matters.
Jefferson, Gail. 1991. “List Construction as a Task and Resource.” In Interaction Competence, ed.
by George Psathas, 63–92. New York: Irvington Publishers.
Jefferson, Gail. 1993. “Caveat Speaker: Preliminary Notes on Recipient Topic-Shift Implicature.”
Research on Language and Social Interaction 26 (1): 1–30. doi: 10.1207/s15327973rlsi2601_1
Jefferson, Gail. 2004. “Glossary of Transcript Symbols With an Introduction.” In Conversation
Analysis: Studies from the First Generation, ed. by Gene Lerner, 13–23. Amsterdam: John
Benjamins. doi: 10.1075/pbns.125.02jef
Kecskes, Istvan. 2013. “Why Do We Say What We Say the Way We Say It.” Journal of Pragmatics
48: 71–84. doi: 10.1016/j.pragma.2012.11.010
Kendrick, Kobin H, and Paul Drew. 2014. “The Putative Preference for Offers over Requests.” In
Requesting in Social Interaction, ed. by Paul Drew and Elizabeth Couper-Kuhlen, 87–113.
Amsterdam: John Benjamins. doi: 10.1075/slsi.26
Kendrick, Kobin H, and Paul Drew. 2016. “Recruitments: Offers, Requests, and the Organization
of Assistance in Interaction”. Research on Language and Social Interaction 49(1): 1–19.
doi: 10.1080/08351813.2016.1126436
Lerner, Gene. 2013. “On the Place of Hesitating in Delicate Formulations: A Turn-Constructional
Infrastructure for Collaborative Indiscretion”. In Conversational Repair and Human Under-
standing, ed. by Makoto Hayashi, Geoffrey Raymond, and Jack Sidnell, 95–134. Cambridge:
Levinson, Stephen. 2000. Presumptive Meanings. The Theory of Generalised Conversational Impli-
cature. Cambridge, Mass: MIT Press.
Levinson, Stephen C. 2013. “Action Formation and Ascription.” In Handbook of Conversation
Analysis, ed. by Tanya Stivers and Jack Sidnell, 103–130. Malden, MA: Wiley-Blackwell.
Lindström, Anna. 2005. “Language as Social Action. A Study of How Senior Citizens Request
Assistance with Practical Tasks in the Swedish Home Help Service.” In Syntax and Lexis
in Conversation, ed. by Auli Hakulinen and Margaret Selting, 209–230. Amsterdam: John
Benjamins. doi: 10.1075/sidag.17.11lin
Ogiermann, Eva. 2015. “Direct off-record requests? – ‘Hinting’ in family interactions.” Journal of
Potter, Jonathan, and Alexa Hepburn. 2010. “Putting Aspiration into Words: ‘Laugh Particles’,
Managing Descriptive Trouble and Modulating Action.” Journal of Pragmatics 42 (6): 1543–
1555. doi: 10.1016/j.pragma.2009.10.003
Raclaw, Joshua. 2013. Indexing Inferables and Organizational Shifts: ‘No’-Prefaces in English Con-
versation. Unpublished PhD dissertation, University of Colorado.
Raymond, Geoffrey. 2004. “Prompting Action: the Stand-Alone ‘so’ in Ordinary Conversation.”
Research on Language and Social Interaction 37 (2): 185–218. doi: 10.1207/s15327973rlsi3702_4
Robinson, Jeffrey D. 2004. “The Sequential Organisation of ‘Explicit’ Apologies in Naturally
Occurring English.” Research on Language and Social Interaction 37 (3): 291–330.
doi: 10.1207/s15327973rlsi3703_2
Sacks, Harvey. 1992. Lectures on Conversation. Volume 1. Oxford: Blackwell.
Sacks, Harvey, Emanuel Schegloff, and Gail Jefferson. 1974. “A Simplest Systematics for the
Organisation of Turn-Taking for Conversation.” Language 50: 696–735.
doi: 10.1353/lan.1974.0010
Saul, Jennifer. 2002a. “Speaker Meaning, What Is Said, and What Is Implicated.” Nous 36 (2):
228–248. doi: 10.1111/1468-0068.00369
Saul, Jennifer. 2002b. “What Is Said and Psychological Reality: Grice’s Project and Relevance The-
orists Criticisms.” Linguistics and Philosophy 25 (3): 347–372. doi: 10.1023/A:1015221313887
Schegloff, Emanuel. 1968. “Sequencing in Conversational Openings.” American Anthropologist
70: 1075–1095. doi: 10.1525/aa.1968.70.6.02a00030
Schegloff, Emanuel. 2007. Sequence Organization in Interaction. Cambridge: Cambridge Univer-
sity Press. doi: 10.1017/CBO9780511791208
Sendra, Veronica Crespo, Constanjin Kaland, Marc Swerts, and Pilar Prieto. 2013. “Perceiving
Incredulity: The Role of Intonation and Face Gestures.” Journal of Pragmatics 47: 1–13.
doi: 10.1016/j.pragma.2012.08.008
Sifianou, Maria. 1993. “Off-Record Indirectness and the Notion of Imposition.” Multilingua 12
(1): 69–79. doi: 10.1515/mult.1993.12.1.69
Sifianou, Maria. 1997. “Politeness and Off-Record Indirectness.” International Journal of the
Sociology of Language 126: 163–179. doi: 10.1515/ijsl.1997.126.163
304 Michael Haugh
Sperber, Dan, and Deirdre Wilson. 1995. Relevance. Communication and Cognition. Second ed.
Oxford: Blackwell.
Stevanovic, Melisa, and Anssi Peräkylä. 2012. “Deontic Authority in Interaction: The Right
to Announce, Propose, and Decide.” Research on Language and Social Interaction 45 (3):
297–321. doi: 10.1080/08351813.2012.699260
Stivers, Tanya, and Federico Rossano. 2010. “Mobilising Response.” Research on Language and
Social Interaction 43 (1): 3–31. doi: 10.1080/08351810903471258
Svennevig, Jan. 1999. Getting Acquainted in Conversation. Amsterdam: John Benjamins.
doi: 10.1075/pbns.64
Terkourafi, Marina. 2005a. “Beyond the Micro-Level in Politeness Research.” Journal of Politeness
Research 1 (2): 237–262. doi: 10.1515/jplr.2005.1.2.237
Terkourafi, Marina. 2011. “The Puzzle of Indirect Speech.” Journal of Pragmatics 43 (11): 2861–
2865. doi: 10.1016/j.pragma.2011.05.003
Terkourafi, Marina. 2013. “Re-Assessing the Speech Act Schema: Twenty-First Century Reflec-
tions.” International Review of Pragmatics 5 (2): 197–216. doi: 10.1163/18773109-13050203
Weiner, Matthew. 2006. “Are all conversational implicatures cancellable?” Analysis 66 (2): 127–130.
doi: 10.1007/s12136-012-0177-x
Welch, Denis. 2005. “The Representative.” NZ Listener. 1 October (Issue 3412). Available at
http://www.listener.co.nz/uncategorized/the-representative/.
Index
A context-dependent 4, 6, 10, 72, Grice 8–9, 38, 48–51, 67, 76–83,

affiliation (effect) 261, 276, 84, 152, 154, 160, 166, 171–172, 121–142, 149, 161, 168–173, 241,
281–282 208, 235, 244, 246, 251, 254, 263, 267, 271–272, 282–284,
approximation 9, 73, 147–149, 277 294–296
157–161, 165, 171, 173 contextualist (approach) 5, 72
automatic processing 260, 262, conventional implicature H
268, 270 see implicature humor 139, 170, 261, 275–276
conventions of usage 201–202 hyperbole 8–9, 121–122, 131–
B conversation analysis 2, 217, 134, 137, 139–140, 142, 144, 147,
background assumptions 2, 7, 237–238, 287, 302–303 149–150, 157, 159–161, 171, 173
15–18, 20–22, 34, 71, 286–287 conversational implicature
broadening 73, 147, 149–151, see implicature I
157–161, 163, 165–167, 171–173 implicature
D conventional ~ 4, 37, 55, 57,
C discourse 66, 283
communication ~ context see context conversational ~ 4, 8, 15,
explicit ~ 35, 92, 148, 173, ~ pragmatics 235–237, 17–18, 25, 28, 37, 48, 51, 53,
196, 215, 255 246–249, 253–254 57, 67, 76–79, 82, 90–91,
harmonious ~ 123, 126 ~ quality 236–238, 246 121–124, 129, 134, 138–139,
implicit ~ 3, 12, 64, 147, 213, ~ relations 10, 235, 237–238, 155–156, 164, 171, 236, 243,
215, 262 241, 243–247, 254–257 281–300
intentional ~ 134, 168, 171, ~ unit 10, 235–254 particularized
173 dominance (effect) 261 conversational ~ 4, 15,
theory of ~ 15, 153–154, 156, 18, 50, 56, 77, 124, 236, 250
168, 172 E implicit
communicative exophoric reference 95, 105 ~ arguments 2, 8, 95–98,
~ goal 39, 197, 241–242, 253, explicated inference 16, 18–19, 100, 102–103, 105–106,
259–261, 264, 271–276, 278 22, 34 114–115, 117–118
~ intention 15, 17, 21, 127, explicature 1–5, 8, 15–19, ~ meaning 1–4, 6–7, 10, 15,
239, 249, 251, 253, 261, 271, 25–26, 33–35, 90–91, 152, 123, 184, 217, 254, 263–264,
273, 276, 282, 297 160–161 271–273
consciousness 255, 262, implicitness (phenomenon)
266–270, 277 F 1–13, 67, 121–122, 152, 177,
context flouting 77, 121, 123–127, 129, 181–182, 184, 194, 196, 201,
discourse ~ 95, 101, 105, 114, 131–133, 135–142, 278 206, 212–213, 221, 235, 259–
149, 152–155, 162, 164–165 foundational theory of meaning 260, 262–263, 267, 269, 271,
experience-based ~ 37 274, 276
201, 209–210 (im)politeness 9, 11, 144,
linguistic ~ 155, 239, G 177, 179–184, 193, 196–199,
247, 249 goal see communicative goal 302 see also politeness
situational ~ 201, 203, grammar and pragmatics indexicals 5–6, 8, 37, 39–41,
206–207, 209–211, 213 95–96, 110, 117–118 46, 49, 65
social ~ 205, 247, 266
306 Implicitness: From lexis to discourse
indirectness 1–2, 4, 7, 9–10, N Relevance Theory (RT)

177, 179, 182–184, 188, 196–197, neo-Gricean lexical pragmatics 3, 8, 20, 75, 90, 107, 135, 152,
219, 261, 268, 275, 278, 301, 67, 89 154–156, 161, 164, 166, 167, 171,
303 173, 260–265, 272, 291
inference 2–5, 7–8, 15–35, 75, O
85, 148, 150–152, 155, 243, offence 9, 177, 179–189, 191, S
263–265, 269, 272–274, 193–196, 293 scalar quantifiers 7, 15–18, 22,
281–287, 289, 291–300 overt untruthfulness 8, 121, 135 24, 27–31, 34–35
see also explicated inference, schemata (of inference)
Truth-Compatible Inference, P 1, 248–249, 260, 269, 271–272
schemata of inference particularized conversational silence
intention implicature see implicature thematic ~ 7, 9, 217, 219,
see communicative intention parts-whole perspective 224, 227, 229, 231
interactional pragmatics 281 235, 254 typology of ~ 217
irony 8, 50, 121–122, 131–132, politeness 2, 252–253, 259–263, situation-bound utterances
135–139, 141–145, 170, 301 265–267, 270–271, 275–276 6, 9, 201–205, 212, 219
see also (im)politeness social action 10, 242–243, 281,
L political speech 217, 219, 221, 283, 294–295, 299
lexical narrowing 8–9, 67–68, 224–227, 231 speaker meaning 37–39, 48, 68,
75, 83–86, 88–91, 147–148, (practical) reasoning 76–77, 90, 121–122, 126–127,
150–151, 155–156, 161, 164, 259, 263–265, 276 130, 132, 139–142, 282, 300
166,171–173 pragmatic enrichment subject pro-drop 8, 95, 97–98,
lexical/semantic underspecification 6, 8, 67–68, 75, 90–91 106–107, 115
67–68, 70, 75, 246 pre-semantic conversational
implicature 8, 67–68, 93 T
M see also implicature tacit knowledge 201, 206–207,
maxim (of conversation) principle of relevance 95, 97 209–210, 213
11, 51, 73, 76–82, 87, 121–142, pronouns 7–8, 37, 40–41, Truth-Compatible Inference
148–149, 156–157, 160, 170, 44–48, 52–54, 56–58, 64, 81, (TCI) 7, 15–18, 21–22, 24,
229, 251, 257, 266 105–107, 109 26, 28–29, 31, 34–35, 286
meiosis 8, 121–122, 131–134, 137, quality (of discourse) truthfulness 121–122, 125–128,
139–142 see discourse 131, 134, 138, 148–149, 160
memory 10, 43, 203–204, 220,
259–260, 268–269, 271–272 Q W
metaphor 8–9, 63, 121–122, Quality (maxim) 8–9, 76, what is said (saying) 2, 15–16,
131–132, 134–135, 137, 139–140, 77, 78, 90, 121, 122, 124–142, 68–69, 72, 77, 90–91, 122, 124,
142, 147–150, 153, 156–161, 165, 157, 170 126–132, 134, 136, 138, 139–142,
167, 170–173, 188, 205, 272 160, 282, 286–287, 294, 299
metapragmatic(s) 177, 181, 184, R
185, 196 reading 7, 15–18, 21–22, 24–28, Z
30–34, 41, 47, 69, 81, 84, 88, zero subject anaphors 8, 95,
100, 110–111, 113–114, 124 97–98, 106–110, 113–115
Although the term implicitness is ubiquitous in the pragmatic scholarship,
it has rarely constituted the focus of attention per se. This book aims
to help crystallize the concept of implicitness by deining its linguistic
boundaries, as well as specifying and exploring its diferent communicative
manifestations. The contributions by leading specialists scrutinize the
main conceptualizations, forms and occurrences of implicitness (such as
implicature, impliciture, explicature, entailment, presupposition, etc.) at
diferent levels of linguistic organization. The volume focuses on phrasal,
sentential, and discursive phenomena, showcasing the richness and
variety of implicit forms of communication, systematizing (where possible)
the existing analytic perspectives, and identifying the most productive
procedures for further exploration. Taken together, the chapters exhibit
theoretical diferences that hinder a consensus on the nature of implicitness,
but they simultaneously reveal methodological points of contact and raise
common questions, thereby signposting a future analytic agenda. The book
will appeal to both theoretically and empirically minded scholars working
within and across the disciplines of Pragmatics, Semantics, Language
Philosophy, Discourse Analysis, and Communication Studies.
isbn 978 90 272 5681 2

Implicitness - From Lexis To Discourse PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Implicitness - From Lexis To Discourse PDF

Uploaded by

Copyright:

Available Formats

From lexis to discourse

John Benjamins Publishing Company

Editor Associate Editor

John Benjamins Publishing Company

the American National Standard for Information Sciences – Permanence

© 2017 – John Benjamins B.V.

Part I. Word and phrase

Part II. Sentence and utterance

Part III. Text and discourse

Marta Dynel and Piotr Cap

Keywords: implicitness, indirectness, implicit meaning, research

Although implicitness is central to pragmatics and the term is ubiquitous in the

to “explicate implicitness” (1986: 16). In this and a plethora of other claims and

2. This collection: Aims and general structure

pertinent neo-Gricean scholarship, a reply is also provided to post-Gricean criti-

Ariel, Mira. 2004. “Most.” Language. 80: 658−706. doi: 10.1353/lan.2004.0162

Word and phrase

Keywords: what is said, Truth-Compatible inferences, Background assumptions,

The goal of this chapter is to problematize the concept of a “reading” associated

that contribute to the development of a full proposition equal more. 1 Considerable

2. Tests identifying interpretation status

2.1 (Un)faithful report tests

Searle’s (1980) Background assumptions and what I have called Truth-Compatible

And (3) is a faithful report of (1b):

2. S stands for the original speaker hereon.

4. I here simplify the original utterance to avoid too cumbersome formulations.

2.2 The including/excluded test

Section 2.1 offered tests to help distinguish between speaker-intended inferences

JIM: … the … to---

The construction under discussion consists of a category X (indicated by italics in

(13) ?: This is a concert in Miami.

Sentences conjoined by and often give rise to coherence inferences pertaining to

6. ~ indicates an unattested example.

Next, consider many and most in the following examples:

(23) a. WALT: He was afraid many times,

Note that the Downward entailing context here (a conditional antecedent) is

5. So-called exclusive or interpretations

6. Summary and conclusions

Ariel, Mira. 2004. “Most.” Language 80: 658–706. doi: 10.1353/lan.2004.0162

Pronouns and implicature

Keywords: pronouns, conventional implicature, conversational implicature,

1. The expression theory and non-descriptive meaning

by expressing another. If one writes ‘Mars exploded’ in a work of fiction, however,

2. Indexical meaning and concepts

℞δ is thus a more specific event type than ℞. ℞δ is to ℞ as driving with friends is to

Thus the extension (referent) of c(he)c(Jefferson) is Jefferson, and the extension of

4. Sortal and determiner components

Suppose Terry points at someone he mistakenly believes to be male, and says He is

6. The binding rules

Different pronouns have different constraints on their interpretations. In (11), for

Chomsky (1995: Section 1.4.2) accounted for such facts in terms of syntactic rules:

Rule A: A reflexive pronoun must be bound in its local domain.

Implicature denotes a type of meaning. Contrast the following sentences, both of

Sentence meaning is a property of expressions rather than of speakers. It depends

The conversational implicature we have examined is not a sentence implicature.

8. Neo-Gricean explanations of the binding rules

The following is a typical derivation of Dan’s implicature.

8. I provide fuller account in Davis (2016a).

Another reason why Rule B cannot be explained pragmatically is that conver-

9. Pronoun implicatures arising from their sortals

In Section 8, I summarized my argument that the binding rules cannot be derived

Sentence (21a) is perfectly grammatical if he is used demonstratively. The sentence

the American National Standard for Information Sciences – Permanence