Figurative Thought and Language

Relevance Theory,
Figuration,
and Continuity
in Pragmatics

edited by
Agnieszka Piskorska

John Benjamins Publishing Company

Relevance Theory, Figuration, and Continuity in Pragmatics
Figurative Thought and Language (FTL)
issn 2405-6944
The aim of the series is to publish theoretical and empirical research on Figuration broadly
construed. Contributions to the study of metaphor, metonymy, irony, hyperbole, understate-
ment, idioms, proverbs and other understudied figures as well as figurative blends will be
considered. Works on figuration in gesture and multi-modal expression, embodiment and
figuration, pragmatic effects of figurativity and other topics relevant to the production, use,
processing, comprehension, scope, underpinnings and theoretical accounts involving figura-
tion, will also be considered.
Volumes in the series may be collective works, monographs and reference books, in the English
language.
For an overview of all books published in this series, please see
benjamins.com/catalog/ftl

Editors
Angeliki Athanasiadou Herbert L. Colston
Aristotle University University of Alberta

Editorial Board
Salvatore Attardo Sam Glucksberg Günter Radden
Texas A&M University, Commerce Princeton University University of Hamburg
John A. Barnden Albert Katz Francisco José Ruiz
University of Birmingham Western University, Canada de Mendoza Ibáñez
Benjamin K. Bergen Sandrine Le Sourn-Bissaoui University of La Rioja
University of California, San Diego Université Rennes 2 Maria Sifianou
Daniel Casasanto Jeannette Littlemore National and Kapodistrian
University of Chicago University of Birmingham University of Athens

Eva Filippova Marilyn A. Nippold Gerard J. Steen

Charles University Prague University of Oregon University of Amsterdam

Raymond W. Gibbs, Jr. Klaus-Uwe Panther Linda L. Thornburg

University of California, Santa Cruz University of Hamburg Nanjing Normal University

Rachel Giora Penny M. Pexman

Tel Aviv University University of Calgary

Volume 8
Relevance Theory, Figuration, and Continuity in Pragmatics
Edited by Agnieszka Piskorska
Relevance Theory, Figuration,
and Continuity in Pragmatics

Edited by

Agnieszka Piskorska
University of Warsaw

John Benjamins Publishing Company

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/ftl.8
Cataloging-in-Publication Data available from Library of Congress:
lccn 2020006062 (print) / 2020006063 (e-book)
isbn 978 90 272 0554 4 (Hb)
isbn 978 90 272 6119 9 (e-book)

© 2020 – 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

Acknowledgements vii

Introduction: The literal-figurative language continuum and

optimally relevant interpretations 1
Agnieszka Piskorska

Part 1. Continua in non-literalness

Chapter 1
Category extension as a variety of loose use 25
Ewa Wałaszewska

Chapter 2
Metonymic relations – from determinacy to indeterminacy 45
Maria Jodłowiec and Agnieszka Piskorska

Part 2. Concepts, procedures and discourse effects

Chapter 3
Evidential participles and epistemic vigilance 69
Manuel Padilla Cruz

Chapter 4
The Greek connective gar: Different genres, different effects? 95
Sarah Casson

Chapter 5
Metarepresentation markers in Indus Kohistani: A study with
special reference to the marker of desirable utterances loo 121
Beate Lubberger
vi Relevance Theory, Figuration and Continuity in Pragmatics

Part 3. Multimodality and style

Chapter 6
When EVERYTHING STANDS OUT, Nothing Does:
Typography, expectations and procedures 167
Kate Scott and Rebecca Jackson

Chapter 7
Relevance, style and multimodality: Typographical features as
stylistic devices 193
Ryoko Sasamoto and Minako O’Hagan

Part 4. Pragmatic effects and emotions

Chapter 8
Towards a relevance-theoretic account of hate speech 229
Jadwiga Linde-Usiekniewicz

Chapter 9
Tropes of ill repute: Puns and (often thwarted) expectations of relevance 259
Agnieszka Solska

Part 5. Stylistic effects in literary works

Chapter 10
Another look at “Cat in the rain”: A cognitive pragmatic approach
to text analysis 291
Seiji Uchida

Chapter 11
Echoic irony in Philip Larkin’s poetry and its preservation in
Polish translations 309
Agnieszka Walczak

Chapter 12
Humour and irony in George Mikes’ How to be a Brit 327
Maria Angeles Ruiz-Moneva

Name index 351

Subject index 355
 Acknowledgements

I would like to express my gratitude to Prof. Angeliki Athanasiadou and Prof.

Herbert L. Colston for offering me an opportunity to publish this volume in their
series Figurative Thought and Language. My sincere thanks are due to all the
contributors, many of whom have also acted as internal reviewers, reading and
commenting on the first drafts of chapters written by other authors. I am very
grateful to two external anonymous reviewers for their insightful remarks and
suggestions on the content of all the chapters. I would also like to thank Esther
Roth (Acquisition Editor), Ineke Elskamp, and the whole team at John Benjamins
for their generous help with handling the production process of the volume.
 Introduction
The literal-figurative language continuum and
optimally relevant interpretations

Agnieszka Piskorska
University of Warsaw, Poland

1. Non-literalness in inferential communication

The observation attributed to Lakoff and Johnson (1980) that non-literalness

permeates everyday communication is nowadays treated as commonplace. Since
the early 1980s issues related to figurativeness have dominated research conducted
within cognitive linguistics,1 and occupied a prominent place in the framework of
relevance theory – approaches that should be seen as “complementary rather than
contradictory” (Wilson, 2011a, p. 192). The two approaches share the fundamental
belief that communication cannot be treated in isolation from cognition and other
psychological processes; the main difference between them being that cognitive
linguistics sees the sources of figurativeness in human thought, investigating its re-
lation to language and grammar (Athanasiadou, 2017), whereas relevance theory
is focused on the inferential aspects of understanding figures of speech.
Although many questions regarding figurativeness remain to be addressed,
thanks to the scholarly works carried out in the two traditions, these questions
are now much more subtle and specific than they used to be. For example, having
discarded the view that figurative language involves a departure from a norm, we
are no longer concerned with explaining why rational language users would break
the norms of communication on a massive scale. Instead, we try to improve our
understanding of the relationship that figurative language bears to encoded mean-
ings on the one hand, and to mental representations of speakers and addressees of
figurative utterances on the other. As Colston (2015, p. 3) aptly puts it, research on
figurativeness should address the question “[h]‍ow can language that demonstrably
disconnects with speaker-intended meaning somehow achieve more meaning?”

1. The term cognitive linguistics is used here as a proper name referring to the school of thought
founded by George Lakoff and Ronald Langacker.

https://‍doi.org/10.1075/ftl.8.00pis
© 2020 John Benjamins Publishing Company
2 Agnieszka Piskorska

If we were to choose one most important realization resulting from the many
years of theoretical and experimental pursuits concerning the nature of figurative
language, it might be that there are no clear-cut boundaries between literalness
and figurativeness, or between various forms of figurativeness. In other words,
although prototypical cases of figures such as metaphor, metonymy or irony stand
out from literalness, these figures are not natural kinds and should be seen as
forming a continuum with other uses of language, typically considered literal.
Within the relevance-theoretic framework this standpoint was put forward
in the very first complete formulation of the theory (Sperber & Wilson, 1986),
and later developed into what is known as a continuity or deflationary account
of metaphor (Sperber & Wilson, 2008; cf. also Carston, 2002; Wilson & Carston,
2007), and echoic account of irony (Sperber & Wilson 1998a, Wilson, 2006, 2013;
Wilson & Sperber, 2012). On the continuity view of non-literalness it is postulated
that figurative uses employ essentially the same mechanisms as some kinds of
literal uses of language, and that the difference between the former and the latter
is that of a degree: typically, the further the hearer moves from literalness, the
more processing effort is expended and more positive cognitive effects (possibly
also different kinds of effects) can be expected to be gained. This is so because,
according to one of the fundamental claims put forward by Sperber & Wilson
(1995), every act of communication creates expectations of relevance,2 which
triggers the operation of inferential mechanisms in an individual to whom an act
of communication is addressed. Geared towards producing an optimally relevant
interpretation, i.e. one providing the maximum of cognitive gains at a justifiable
processing cost, these mechanisms guide the inferential comprehension processes
that fill the gap between the schematic linguistic meaning of a sentence, and the
context-specific, speaker-intended meaning of an utterance.
To provide a somewhat more detailed explanation of how relevance theory
views this process in the case of two most extensively studied figures, i.e. meta-
phor and irony, brief overviews with some illustrative examples will be presented
below, in order to prepare the ground for introducing topics discussed further
in this volume.
Let us focus on the deflationary view of metaphor first. As can be observed,
speakers may encode their thoughts in such a way that linguistic representations
resemble their mental representations to a greater or lesser degree. Consider
Examples (1)–(3):
(1) My cat is a European shorthair.

2. This is known as the Communicative Principle of Relevance, the exact formulation of which
is: “Every act of ostensive communication communicates a presumption of its own optimal
relevance” (Sperber & Wilson, 1995, p. 260).
 Introduction 3

(2) My cat is my best friend.

(3) My cat is a dog.

Utterance (1) bears full resemblance to the speaker’s thought (on the standard
assumption that it is a true assertion), and the concept encoded by the expression
a European shorthair is communicated literally. Utterance (2) seems to be used
somewhat loosely, since it is disputable whether the meaning encoded by the noun
friend is literally applicable to animals. Even animal lovers who do not draw a
sharp divide between animals and humans may admit that the prototypical con-
cept of friendship involves such elements as mutuality of feelings and symmetry of
actions, and animals cannot fully reciprocate what humans do for them. Though
not a metaphor, then, the concept FRIEND*3 communicated in (2) can be said
to be used more loosely than the concept European shorthair in (1): the denota-
tion of FRIEND* is broader than that of the corresponding encoded concept, as it
includes non-human friends. Example (3) is a typical case of metaphor, in which
the source and target concepts belong to distinct categories. The meaning com-
municated by DOG* in this particular utterance diverges from the literal meaning
of dog and becomes (possibly) something like ‘a feline creature having some ste-
reotypical properties of dogs’, becoming in some respects broader than the literal
denotation of dog, and in some respects narrower, as it excludes those properties
of dogs that are not intended to be part of the interpretation, e.g. barking. It has to
be noted that despite the differences in the use of predicate concepts in (1)–(3), the
three utterances are statements about the speaker’s cat and can be subject to true/
false judgments. Utterance (3) is not different in this respect from (1) and (2) in
that it can be considered true if the feline referent displays some relevant features
characteristic of the Canis familiaris species, and false otherwise. Relevance theory
captures this observation by analyzing metaphorically understood concepts, like
DOG* in (3), as part of the proposition expressed by the utterance, and therefore
contributing to truth-conditional explicit content rather than feeding into impli-
catures. Equating the potential of literal, loose and metaphorical statements to
express assertions, this postulate constitutes one of the main tenets upon which
the continuity view rests.

3. Following the terminology introduced by Barsalou (1983), such concepts are referred to as
‘ad hoc’, since they result from occasion- and context-sensitive modifications (i.e. narrowings
and/or broadenings) of encoded concepts. In the relevance-theoretic literature, they are conven-
tionally marked by capitalization and asterisks, e.g. FRIEND* or DOG**. The name ‘ad hoc’ may
be misleading, however, as not all such context-specific meanings are worked out on the spot
every time an utterance is processed. It can be for instance argued, that the concept FRIEND*
referring to an animal has a stable place in an individual’s conceptual system (cf. Sperber &
Wilson, 1998b; Hall, 2017).
4 Agnieszka Piskorska

Having briefly characterized the standard relevance-theoretic account of

metaphor in terms of a relatively large discrepancy between a communicated
concept and the corresponding encoded concept, let us now turn to the account
of irony. Unlike metaphorical utterances, ironic ones are believed not to commu-
nicate the proposition corresponding to the utterance produced by the speaker.
Such propositions are in various ways ludicrously inadequate, as illustrated below:
(4) Holidays of my dreams! (on discovering that the holiday apartment
overlooks a rubbish dump)
(5) I love people who answer phone calls (when the person called fails to answer
repeatedly).

While (4) is blatantly false and (5), even if true, blatantly irrelevant, they both
allude to some other representations, in this case to commonly held expectations
that holiday homes should be worth the money we pay for them and that people
should answer phone calls. In both cases, the expectations have clearly been
frustrated, hence the speaker expresses the attitude of dissociation and possibly
disappointment. On this account, irony is technically defined as echoic use of
language, involving a tacit expression of an attitude to representations resem-
bling and alluding to other representations (Sperber & Wilson 1995, p. 328). To
demonstrate that irony – like metaphor – displays continuity with other forms
of expression, let us note that using representations in virtue of resemblance to
other representations, i.e. as metarepresentations (Wilson, 2000), and expressing
an attitude to them is commonly exploited in many communicative situations.
For instance, the speaker may report someone else’s utterances in such a way that
her formulations resemble the original wording, at the same time expressing an
attitude to the content reported. If Inez and John promised themselves that they
would go on holidays of their dreams, and the apartment they had rented turned
out to be amazing, one of them could utter (4) echoing their desires and expressing
an attitude of happiness towards their fulfillment. It can be easily imagined that –
depending on the (mis)‍match between their expectations and the actual state of
affairs they encounter – Inez or John could say (4) expressing a variety of attitudes
to the content echoed by this utterance, from utter disgust, through mockery
and criticism (especially if the holiday home was booked by the other person),
through disbelief, to sheer enthusiasm and delight. Some of these attitudes – those
that can be subsumed under the term “dissociation” – would make utterance (4)
ironic, but as Wilson (2006) points out, attitudes are not clearly delineated, and
one typically shades off into another. Therefore, there may not be a clear cut-off
point between irony and other echoic uses of language, which indicates that irony
is not a natural class either.
 Introduction 5

The picture that emerges from the above-presented sketchy description of

metaphor and irony is that each of them draws on the same interpretive mecha-
nisms as some other forms of expression: metaphor arises out of the narrowing and
broadening of the encoded concept (as other non-metaphorical ad hoc concepts),
and irony exploits resemblance and tacit expression of an attitude, similarly to
other echoic utterances.4 The picture becomes more complicated when we take
into account two additional observations made about various figurative uses. First,
as pointed out by Carston (2010, 2012), the interpretive mechanism that can be
plausibly postulated for metaphor of the A is B type, as described above for (3), may
not be applicable to extended metaphors, often comprising several lines of a text
or stanzas in a poem, which may jointly form a metaphorical scenario including a
number of non-literally intended concepts.5 In such cases, a metarepresentational
route of processing (a la irony) might be effectuated by the comprehension system
in the human mind, which is seen as more economical than creating a chain of ad
hoc concepts. Although Wilson (2018) questions the premise on which Carston’s
(2012) postulate is based, namely that creating ad hoc concepts should be seen
as more effortful that processing literal meaning, she also admits that extended
metaphors have a special property which she describes as “lingering of linguistic
form” (Wilson, 2018, p. 195). It follows then that the decoded linguistic meaning
does play a role in the interpretation of some metaphors, and it may do so to
various degrees. Second, blends of two or three figures, e.g. irony and hyperbole,
irony, hyperbole and rhetorical question, metaphor and metonymy, etc. are very
frequently attested (Colston, 2015, pp. 2–3). As pointed out by Colston (2017),
the ultimate interpretation of such complex figures may not be equal to the sum
of effects attributable to particular mechanisms, to the extent that it is possible
to separate and measure these effects experimentally.6 A discussion on hyperbole

4. It has been pointed out by one of the reviewers that to prove complete continuity between
literalness and figurative language it would have to be demonstrated that “there is no mechanism
at all that is exclusive to any of the figures.” I believe that, while the discussion in this chapter
may not prove this, it indicates that both metaphor and irony are characterized by a unique
combination of otherwise non-exclusive mechanisms.

5. One of the examples of such extended metaphors discussed by Carston (2012) is the well-
known poem “A valediction: Forbidding mourning” by John Donne representing lovers’ souls
as twin compasses.

6. The synergic effect of combining hyperbole and irony was discussed by Colston and O’Brien
(2000), who observed that hyperbolic ironies are richer in pragmatic effects than non-hyperbolic
ironies. Within the relevance-theoretic model the combined effect of metaphor and irony is
discussed by Popa (2010). Jodłowiec and Piskorska (2015) touch upon metaphoric metonymies,
such as They have employed a pretty face.
6 Agnieszka Piskorska

offered by Carston and Wearing (2015) points toward similar conclusions – hyper-
bole combined with metaphor (as in John is a saint) makes a different contribution
to the meaning of an utterance than hyperbole combined with irony (as in What
a huge dessert to someone who treated me to one raspberry). What these find-
ings suggest is that not only do encoded meanings of words become fine-tuned in
contexts, but also that the very mechanisms that derive these contextual meanings
are fine-tuned in accordance with the pursuit to optimize relevance. Thus, when
the pragmatic processes identified for isolated central uses of particular figures,
such as metaphor, hyperbole or irony interact with one another in the course of
processing, it results in a synergic effect.
This kind of synergy is straightforwardly accommodated in the model of utter-
ance comprehension known as mutual parallel adjustment, proposed by Sperber &
Wilson (1998b). On this model, inferential operations are not executed serially but
run in parallel, so that one operation can be fed or inhibited by another. This is so
because the comprehension system in the mind does not apply a fixed algorithm,
but is geared towards the maximization of cognitive benefits, which are relative
to an individual’s interests in a specific situation. In other words, expectations of
relevance that guide the comprehension process are always occasion-specific, and
so are the resultant effects of combining various elements of linguistic form with
contextual factors.
Much as this synergic perspective on figurativeness is natural for relevance
theory, together with the continuity view it justifies the generalization of the label
“deflationary,” originally posited for metaphor, over the accounts of all figures of
speech proposed within this framework. It also motivates what can be called an
inductive approach to data, i.e. investigating – in specific contexts – pragmatic
effects of a continuum of communicative devices ranging from figurative to literal,
where a pragmatic effect is explained by Colston (2015, p. 6) as “a reference to
mental/internal activity taking place in a person traceable to his or her encounter-
ing figurative as well as other language.”
Adoption of the deflationary approach to figurativeness has consequences on
how the notion of style is understood. In his seminal work applying relevance
theory to translation, Gutt (2000) describes style as “not only what the [author]
intended to convey but also how he conveyed it,” adding that “style is, in some
sense, the way the writer or speaker expresses himself, resulting, for example,
from the words he chooses, or the way he constructs his sentences” (2000, p.130).
Wilson (2018) underscores that “[a] long-standing aim of relevance theory has
been to show that stylistic and poetic effects traditionally associated with figura-
tive utterances arise naturally in the pursuit of relevance” (2018, p. 192), noting
further, in a similar vein to Gutt’s, that “the presence of a certain word or phrase,
a certain syntactic structure or prosodic feature, may contribute directly to
 Introduction 7

relevance by reducing processing effort or increasing cognitive effects.” Wilson

(2018) further acknowledges that it is not only linguistic properties of an utterance
that contribute to relevance: its visual, auditory or other elements may affect the
path towards achieving relevance too. As these statements indicate, stylistic effects
are expected to arise out of the processing of all facets of an utterance, whether
they are figurative or literal; whether they are related to semantic, syntactic, pro-
sodic or yet another level on which the speaker exercises freedom to select a form
optimally suited to expresses her intention, in accordance with her “abilities and
preferences.” Sometimes such choices are relatively unmarked and their impact
on pragmatic effects is inconspicuous, but in other cases – to use Deirdre Wilson’s
words again – “a certain aspect of linguistic form may seem relevant enough to
attract attention and lead on to further reflection” (2018, p. 192).

2. Contributions to this volume

The deflationary approach to figurativeness and style is reflected in the papers col-
lected in this volume, discussing prototypical figures of speech, such as metonymy
and irony, side by side with other aspects of communicative stimuli which may
impact on the balance of cognitive gains and processing cost in various ways,
including supporting strong and weak implicatures of the kind envisaged by the
speaker, playing on lexical ambiguity, exploiting typographical features of a mes-
sage, as well as utilizing the potential of linguistic items which do not contribute
to the content of the message expressed. Apart from offering relevance-theoretic
analyses of these issues, the chapters refer to other models of the phenomena
under discussion, in this way providing a background for reflection on what rel-
evance theory contributes to understanding them. The approaches that relevance
theory is contrasted with include various descriptive accounts, Gricean analyses,
the pretence theory of irony and cognitive linguistics.
The contents are organized into five parts. Part 1 Continua in non-literalness
includes two contributions directly pertaining to figurative language and pragmatic
continua. First, Ewa Wałaszewska’s chapter Category extension as a variety of loose
use revisits the literalness-metaphor continuum by investigating category exten-
sion – a phenomenon neighbouring on approximation, hyperbole and metaphor.
Prototypical cases of category extension involve the broadening of lexically encoded
concepts denoting well-known brand names used to refer to other members of the
category to which the original encoded concept belongs (e.g. Hoover for any vac-
uum cleaner). Apart from such plain instances, category extension includes more
creative ones, in which it can be combined with metaphor. Taking into account that
it can also be blended with approximation or hyperbole, Wałaszewska recognizes a
8 Agnieszka Piskorska

complex, non-linear and multi-dimensional structure of the literalness-metaphor

continuum. Apart from making an attempt to situate category extension on this
continuum, the chapter links it to such phenomena as ‘lexical warfare’, ‘paragons’,
‘snowclones’, ‘schemata’ and ‘pragmatic routines’. For instance, the widespread
practice of coining expressions such as Putin is the new/next Hitler, according to
Wałaszewska, is reminiscent of creating category models organized around what
Lakoff (1987) calls ‘a paragon’. Since this pattern has gained enormous popularity,
especially in the media, its productivity can be described by means of the somewhat
pejorative term ‘snowclone’. Wałaszewska thus concludes that many of the category
extensions that started off as creative processes, have in fact become routinized.
The other chapter in this section, Maria Jodłowiec and Agnieszka Piskorska’s
Metonymic relations – from determinacy to indeterminacy, deals with me-
tonymy – a figure which has attracted considerable attention within cognitive
linguistics but has been relatively underresearched by relevance theorists, the
notable exceptions being Papafragou (1996), Falkum (2011), Wilson (2015) and
also Jodłowiec and Piskorska (2015). Having presented a short overview of the
cognitive linguistics analyses of metonymy, the authors go on to sketch an account
of the relations between the metonym and the communicated concept couched in
inferential terms, on which the typically exploited patterns such as part for whole,
author for work etc. are explained as generalizations resulting from repeated use
rather than mechanisms underlying metonymic mappings. Basing on the observa-
tion that metonymically expressed concepts that contribute to the explicitly com-
municated content of an utterance are often indeterminate, the authors posit that
the pragmatic mechanism responsible for identifying the referents of such concepts
is contextual cognitive fix, introduced by Jodłowiec (2015) as an alternative to free
enrichment. On the cognitive fix account of metonymic reference assignment, the
hearer may be satisfied with an underspecified representation of a referent as long
as such a representation is sufficient for warranting implicatures constituting the
main communicative import of an utterance. For example, when one concert goer
says to another Anderszewski plays Bach at Warsaw Philharmonic, the addressee
may immediately infer the implicature that there will be a huge interest in this con-
cert and rush to get tickets without entertaining a complete mental representation
of the referent of Bach as ‘Bach’s music’ or ‘Bach’s pieces that can be performed on
the piano’. As is argued in the chapter, the indeterminacy of metonymic concepts
may be a matter of degree, which is another manifestation of continuity, and may
be exploited in various ways. Sometimes an indeterminate concept makes a simi-
lar contribution to relevance on the effect side as a fully saturated concept would,
but is less costly to process. On other occasions, indeterminacy of metonymically
communicated concepts itself may be a source of stylistic effects, which brings to
the fore a similarity between metonymy and metaphor.
 Introduction 9

The three chapters included in Part 2 Concepts, procedures and discourse ef-
fects provide a counterpoint to Part 1 in that the linguistic items discussed in them
would not be standardly thought of as belonging to figurative language. Instead,
their primary function is to indicate how the propositional content expressed
by an utterance is to be represented by the hearer in terms of epistemic stance
or in terms of its role in discourse. Such a function is often executed by means
of a procedure7 encoded by an item, but it may also be realized via conceptual
encoding. As the three contributions evince, choosing an expression marking an
epistemic attitude or discourse role may also produce pragmatic effects by exerting
impact on the kind and strength of positive cognitive effects communicated by an
utterance. This is not accounted for by traditional descriptive approaches to con-
nectives and discourse markers; nor is a viable explanation offered how different
(albeit related) functions of these expressions are inferentially derived from their
encoded core meanings.
Manuel Padilla Cruz’s chapter Evidential participles and epistemic vigilance
focuses on participial adjectives of the type alleged, suspected etc., as illustrated
by the phrase the alleged killer. Padilla Cruz treats these expressions as encoding
conceptual information on the communicator’s epistemic stance towards the noun
phrase including them. More specifically, the role of these participial adjectives is
to indicate to the hearer that there is not enough evidence to attribute a quality
or fact predicated by the head noun to its referent, e.g. s/‍he who is referred to
as the alleged killer may still turn out not to have committed the act of killing.
Interestingly, as Padilla Cruz observes, these participles also indicate, possibly in
the form of implicatures, that despite the existing doubts, the hearer might expect
that decisive evidence confirming the fact or activity in question will become
available in the future. In this way, the communicator encourages the addressee
to entertain certain assumptions with moderate strength, or, possibly suspend
their belief in these assumptions until more evidence comes along. According
to Padilla Cruz, similarly to evidential adverbials (e.g. allegedly), the participles
under scrutiny target the suite of cognitive functions known as epistemic vigilance
mechanisms (Sperber et al., 2010). The crucial difference between the evidential
adverbials and participles is that the former concern the epistemic status of the

7. The term ‘procedural meaning’ was originally introduced into relevance theory by Diane
Blakemore (1987) to account for the non-truth conditional meaning of conjunctions (e.g. but),
and discourse particles (e.g. after all). The notion has been since applied to pronouns (Wilson &
Sperber, 1993), evidentials (Ifantidou, 2001), inference patterns (Unger, 2011) as well as intona-
tion, tone of voice and other prosodic features, (Wilson & Wharton, 2006; Wharton, 2009, 2012;
Scott 2017). Another important elaboration of the notion concerns its role in argumentation
(Wilson, 2011c, 2016).
10 Agnieszka Piskorska

whole proposition expressed by an utterance, whereas the scope of the latter is

limited to the noun phrase in which they occur.
As evidenced by the title of Sarah Casson’s contribution The Greek connective
gar: different genres, different effects?, the author hypothesizes that the functions
of the connective may be linked to the type of discourse in which it appears, i.e.
in argumentation and in the narrative. Casson treats gar as a procedural marker
targeted at epistemic vigilance mechanisms triggering the inferential procedure
of strengthening, and investigates its use in the corpus of Koine Greek New
Testament texts. The results support the view that the core function of gar is
to indicate that the following utterance, when combined with some contextual
assumptions, is intended to increase the relevance of the preceding assumption
through strengthening. While the core function manifests itself in both types of
discourse considered in the chapter, the specific outcome of its operation depends
on the context. In argumentation, gar typically introduces an assertion providing
evidence which strengthens the believability of a claim. In narrative, the role of gar
can be best seen as introducing a clause providing explanation, thereby making
a previous assertion more understandable and plausible. Casson observes that
the explanatory function of gar can also be utilized in argumentation in cases
where providing further evidence for the believability of a previous assertion is
not necessary because the hearer accepts it as true anyway. In some contexts both
the explanatory and confirmatory functions of gar are present, in which case the
contextual assumptions accessed in the comprehension process are also used by
the epistemic vigilance mechanism to help to convince the hearer of the truth of
the claim (cf. Sperber et al., 2010).
Casson concludes by stating that it is not the use of gar by itself that creates
varying stylistic effects in different genres but that these effects can be attributed
to the interaction of the core meaning of gar with the characteristics of a type of
communication in which it is used, such as its content and other stylistic features.
Beate Lubberger’s chapter Metarepresentation markers in Indus Kohistani:
a study with special reference to the marker of desirable utterances loo deals
with procedural markers from the Pattan variety of Indus Kohistani, an Indo-
Aryan language spoken in Northern Pakistan, focusing on the particle loo. Having
scrutinized two corpora: one consisting of folk stories, narratives of personal
experiences, expository and procedural texts, and the other including recorded
conversations spoken in the family of the author’s language consultant, Lubberger
teased out two functions of the particle loo. The first function is to mark utter-
ances that a speaker wishes the hearer to convey to a third person, which could
be paraphrased as tell X that…. The second function of loo is to mark the third
person imperative (he should do Y). Drawing on the notions of procedural mean-
ing (Blakemore, 1987) and metarepresentation (Wilson, 2000), Lubberger argues
 Introduction 11

that these two functions have a common denominator: in both of them loo acts as
a procedural marker that indicates a specific type of metarepresentation, namely
that of desirable utterances. The second of the two uses of loo mentioned above can
therefore be also explicated as an imperative utterance that the speaker desires to
be addressed to a third person.
The function of indicating metarepresentation is not unique to loo as there
are four such markers in Indus Kohistani, each of them specializing in marking a
different kind of metarepresentation. As the author supposes, the marker loo may
be the first procedural indicator of metarepresentations of desirable utterances
described to this day.
Lubberber’s chapter makes a generalization regarding the speaker’s choice of
how a metarepresentation of a desirable utterance can be expressed: tacitly, by
means of a procedural marker, such as loo, or by means of an expression encoding
this idea conceptually. Since not all the options are afforded by all languages, this
introduces an interesting cross-linguistic perspective on stylistic choices available
to speakers with respect to desirable utterances.
The two contributions constituting Part 3 Multimodality and style, Kate Scott
and Rebecca Jackson’s When EVERYTHING STANDS OUT, Nothing does: typog-
raphy, expectations and procedures and Ryoko Sasamoto and Minako O’Hagan’s
Relevance, style and multimodality: typographical features as stylistic devices
apply relevance theory to analyses of typographical features of a written text, such
as typeface, font, colour and other special effects. Both chapters approach typo-
graphical features as choices of form (in this case paralinguistic rather than strictly
linguistic), with Scott and Jackson’s analysis capitalizing directly on the notion of
procedural meaning exploited in the previous part. The two chapters draw on the
observation that when a speaker chooses a more effort consuming formulation,
as compared to some existing alternative, the hearer’s expectations of relevance
tend to be affected, again with respect to the less costly alternative, in that more
effects and different kinds of effects are expected. Thus, the explanation of the con-
sequences of exploiting typographical features, both in written messages and in
multimodal communication follows naturally from general pragmatic principles.
Scott and Jackson’s chapter focuses on two typographical features: the use of
capitals and italics, exploring their role in text interpretation against the back-
ground of general observations on what typography can contribute to meaning.
First, the authors scrutinize parallels and differences between prosody and ty-
pography, concluding that even though spoken language has richer paralinguistic
means to interact with the encoded message than written language, typography
can play some of the functions carried by prosody in spoken messages, and that
these functions can be related to various levels of meaning, i.e. propositional
content, propositional attitude or implicatures. Even though typography cannot
12 Agnieszka Piskorska

be seen as a direct analogue of prosody, Scott and Jackson propose that it should
likewise be analyzed in procedural terms, arguing that in a way typical of many
other procedural items, typography guides the inferential process rather than
directly contributes cognitive effects.
It is important to note that unlike many procedural items discussed in the
literature (e.g. conjunctions), typographical features should be seen as procedural
devices and not expressions as they do not encode specific procedures. Instead,
they should be treated as having a capacity to trigger procedures (cf. Wilson, 2016)
and to constrain and guide pragmatic processes (cf. Carston, 2016). Because of
that, the effect of typography is relative to the default convention that holds in
a language – a departure from such a convention, e.g. using capitals in place of
minuscule increases effort and modifies expectations of relevance, triggering a
search for extra effects to offset the effort ostensively claimed by using the unusual
typography. As Scott and Jackson point out, the extra benefit offered by the use
of capitals and italics lies in highlighting the locus of relevance – an effect con-
nected with the content of the assumptions communicated. In this way the chapter
demonstrates that the observable affinity of typographical highlighting devices to
contrastive stress can be explained by the fact that they are both procedural devices.
Sasamoto and O’Hagan’s chapter Relevance, style and multimodality: typo-
graphical features as stylistic devices analyzes the role of another highlighting
device – impact captions, referred to as telops, in Japanese television programmes.
Widely used in Japanese and other Asian cultures, impact captions are a special
kind of open captions (or subtitles) which either transcribe verbatim an utterance
spoken in a TV show, or display its edited version in order to highlight some parts
of the discourse.
As Sasamoto and O’Hagan observe, even though the functions and mecha-
nisms of typography employed in traditional writing and in digital communica-
tion are similar, the latter offers more opportunities to exploit visual properties of
written messages than the former. This is often taken advantage of by producers,
who use a variety of fonts, colours, sizes and other special effects as a means of
increasing a programme’s appeal to the audience.
Like Scott and Jackson, Sasamoto and O’Hagan underscore the pragmatic
character and context-dependent function of typographical features, arguing
against semiotic approaches on which each typographical feature is seen as a car-
rier of specific encoded meaning. Such approaches cannot explain all the cases
of exploiting visual stimuli for the sake of guiding the comprehension process,
whereas an inferential approach rooted in general pragmatic principles offers an
adequate account of the whole variety of subtle stylistic effects resulting from the
use of such stimuli.
 Introduction 13

In their analyses of impact captions, Sasamoto and O’Hagan draw on

Blakemore’s (2008) account of appositive phrases used in order to evoke stylistic
effects in the form of a range of weak implicatures or non-propositional affec-
tive responses. Sasamoto and O’Hagan adopt one of Blakemore’s mechanisms,
in which the two phrases in apposition form a hybrid representation, to explain
how ostensively used typographical features of impact captions interact with other
elements of the message, thereby giving rise to a hybrid representation too, in
the sense that its ultimate content is affected by all these multimodal elements.
The authors mention two specific examples how typographical features can af-
fect the message – on the level of implicatures, where weak, non-propositional
effects reside, and on the level of higher-level explicature, where propositional
attitude is constructed. The authors’ theoretical considerations are supported by
the results of an eye-tracking study, offering qualitative empirical evidence that
viewers’ gaze direction and intensity are indeed strongly affected by special effects
imposed on captions.
Part 4 entitled Pragmatic effects and emotions includes texts addressing the
issue of pragmatic effects intended mostly to evoke affective responses of nega-
tive valence. Jadwiga Linde-Usiekniewicz’s chapter Towards a relevance theory
account of hate speech attempts to capture the nature of hate speech – a phenom-
enon that is not as easily amenable to systematic investigation as it might seem.
Related to many other recognized forms of offensive language, such as slurs, hate
speech cannot be equated with any of them. As Linde-Usiekniewicz notes, any
attempts to establish a set of linguistic criteria for identifying all instances of hate
speech (and only of hate speech) are doomed to failure since it is a heterogeneous
phenomenon, and the hate component may be expressed overtly or covertly, using
literal or figurative language, or even under a guise of a compliment. The appar-
ently elusive or subjective nature of hate speech is reflected in numerous highly
controversial rulings of courts, discussed in the chapter, such as a failure to proffer
charges against authors of the slogan We will liberate Poland from euro-traitors,
Jews, masons and the government mafia on the grounds that the verb liberate is
used in the future indicative (rather than imperative) form, and is therefore not an
incitement to any emotion or action. Commenting on this and similar examples,
Linde-Usiekniewicz rightly observes that the legal interpretations are sometimes
“in total contravention of principles that govern human communication.”
Despite the fact that some seemingly promising theories proposed for the
analyses of hate speech have been misused in judiciary practice, (eg. Critical
Discourse Analysis or Speech Act Theory), and fell short of grasping the nature
of hate speech, Linde-Usiekniewicz is optimistic about employing the relevance-
theoretic apparatus to tease out the characteristics of the phenomenon in question.
As predicted by relevance-theoretic comprehension heuristics, the use of certain
14 Agnieszka Piskorska

words and associated concepts automatically triggers a processing path leading

to a certain explicature, implicated premises and implicated conclusions that
make the utterance relevant. In this process, posits the author, for an utterance
to count as an instance of hate speech, two kinds of assumptions have to appear:
one concerning the targeted group’s inferiority and the other, dubbed “the hate
component” communicating a need to do something negative about the targeted
group. It appears that these two kinds of assumptions can be jointly considered as
necessary conditions for distinguishing hate speech from related phenomena and
that they therefore provide not only interesting theoretical observations but also
useful criteria applicable to judiciary practice.
Agnieszka Solska’s chapter Tropes of ill repute: puns and (often thwarted)
expectations of relevance develops the relevance-theoretic account of puns (cf.
also Yus, 2003; Padilla Cruz, 2015), discussing their widely acknowledged yet
underresearched aspect, namely bad reputation, which they have earned, deserv-
edly or not, by eliciting ambivalent emotions in recipients. Although the negative
attitude to puns can be attributed to massive criticism levelled against this trope
at the rationalistic and seriousness-promoting Age of Reason, this is not the only
reason (pardon the lame pun, which seems to fit the argument, though) why puns
are treated as “wayward stylistic devices” or “tropes of lesser value”. Solska offers
two explanations of this phenomenon: one drawing on the Gricean theory of
conversation, on which punning utterances can be seen as violating one or more
conversational maxims without giving rise to any conversational implicatures, and
the other, being the main point expounded in the chapter, couched in terms of
the relevance-theoretic notion of expectations of relevance that every utterance
creates. According to Solska, such expectations can be broken down into six
components: that the utterance communicates one identifiable explicature, that
it has some informative value, that the positive cognitive effects resulting from
its processing justify the effort, that the processing does not require reapplying
the relevance-theoretic comprehension procedure, that the context necessary
for comprehension is easily available, and that the form of an utterance does not
require special attention. As the author argues, many puns violate some or all of
the above-mentioned components jointly constituting the hearer’s expectations
of relevance, which justifies why they may be considered as less indispensable in
communication than, for instance, metaphors.
The last part of the volume, Stylistic effects in literary works, includes three
chapters that draw on standard relevance-theoretic concepts in order to analyze
various aspects of literary texts. Despite the fact that processing information on
imaginary events, personae and their psychological states created by writers and
poets does not impact directly on the reader’s representation of the real world, as
we all know, such states and events can (and are often expected to) make a strong
 Introduction 15

appeal to our intellect and emotions. This is so because information about ficti-
tious characters as well as utterances produced by them are processed against the
inventory of assumptions included in the reader’s cognitive environment (Wilson,
2011b), so what we learn from literature is (or is not) relevant to us in the same way
as assumptions communicated by real people, i.e. by giving rise to positive cogni-
tive effects. Moreover, despite the fact that pragmatic mechanisms applied in the
comprehension and interpretation of literary texts are the same as in the discourse
of everyday conversational exchanges, the form of devices deployed in literature
tends to be “relevant enough to attract attention and lead on to further reflection,”
to repeat the quote from Wilson (2018) mentioned in a previous passage.
This part opens with Seiji Uchida’s chapter Another look at “Cat in the rain”:
A cognitive pragmatic approach to text analysis, employing the notion of strong
and weak implicatures to show how writers of literary texts can affect readers’
interpretations by predicting which contextual assumptions will be used in pro-
cessing. To this end, interplay between linguistic encoding and inference is often
exploited – for instance, by choosing certain grammatical structures and lexical
items, writers invite some inferences on the part of the reader. In this way, strongly
implicated assumptions can be exploited for the sake of creating twists in the plot
(Uchida, 1998). In Hemingway’s short story “Cat in the rain” there are conflicting
inferences concerning the identity of a cat – or two cats. According to the majority
of critics and linguists who have analyzed the short story, the lexical devices used
by Hemingway support the inference that there are two cats – one that appears
under a table at the beginning, and another that is brought by a hotel maid at the
end. These lexical devices include the noun phrases a kitty and a big tortoise-shell
cat to refer to what is alleged to be two different cats, or the very presence of the
indefinite article a in the second NP. There are also supporters of the view that the
difference in the ways of referring to the cat(s) can be explained by a change of the
perspective, which moves from one protagonist to the other. Uchida’s chapter does
not aim to univocally solve the dilemma, but proposes to analyze the conflicting
interpretations in terms of strong and weak implicatures, with the more likely
interpretation being strongly implicated and the less likely one weakly implicated.
An interesting point is that in the case of literary works, the strongly implicated
interpretation does not necessarily override the weakly implicated one, as they
may coexist (“Schrödinger’s cat in the rain?”) and be pondered upon by readers
and critics. A Gricean analysis in terms of implicature cancellation that the chapter
discusses as a possible alternative to the proposed relevance-theoretic solution
does not seem to allow the uncertainty to linger.
Poetry is expected to appeal to the reader through a condensation of sty-
listic devices, contributing a range of effects which are often indeterminate and
hearer-specific (hence the name “poetic effects,” which do not, however, have to
16 Agnieszka Piskorska

be confined to poetry – cf. Sperber & Wilson, 1995, ch. 4.6). A contrast between
everyday metaphor and poetic one is relatively easy to grasp and has been ap-
proached in theoretic terms. To my knowledge, no parallel distinction has been
given systematic attention in the case of irony, although there is clearly a continu-
um of cases between highly conventionalized ironies and creative ones. Agnieszka
Walczak’s chapter Echoic irony in Larkin’s poetry and its preservation in Polish
translations analyses how poetic irony is exploited in two poems by Philip Larkin,
proclaimed ironist by many critics, confronting the impressions that gave rise to
this proclamation with the relevance-theoretic notions of irony and parody. One
of the points made by this chapter is that if irony is understood as saying one thing
and meaning the opposite, it leads to a simplistic interpretation of the poems dis-
cussed. Similarly, identifying irony with pretence and ignoring its echoic character
falls short of capturing all the intricacies of Larkin’s ironies as well. To argue the
point, the author first analyses “Church going,” where the speaker/narrator enacts
church rituals and reads a line from the book of prayer in a parodic way, but then,
when echo in the church repeats the line, his attitude of distancing himself from
religion is questioned. In this way, ironic dissociation becomes a multi-layered
process as each level of metarepresentation can be embedded under another one –
a view which is, needless to say, impossible to develop on the ‘opposite meaning’
approach to irony. It can be added at this point, somewhat on an informal note,
that the metaphorical reading of echo in the church as ironic echoing may have a
particular allure for a relevance-theorist.
The implications of adopting the relevance-theoretic approach to irony for
interpreting Larkin are also discussed on the basis of two translations of his poem
Water. According to Walczak, one of them recognizes the echoic character of irony
while the other follows the ‘opposite meaning’ rhetorical tradition. This suggests
that the translator’s grasp of the mechanism of ironic communication may be
reflected in the ultimate form the target text receives.
The role of irony discussed by Maria Angeles Ruiz Moneva in her chapter
Humour and irony in George Mikes’ How to be a Brit is quite different from that
discerned by Walczak in her analyses of Larkin’s poems. Ruiz Moneva discusses
how these tropes arise in the intercultural clash between the perspective of an
immigrant and that of native Brits. Typically of newcomers, Mikes is amused and
seemingly surprised by what is (or was?) stereotypically represented as the British
way of life, including the sense of uniqueness that, in his opinion, the British as-
cribe to themselves. The mocking attitude to views attributed to the British results
in irony, the main point of which is to distance oneself from the way the British
metarepresent themselves. Humour, on the other hand, can be traced to a number
of inferential mechanisms, such as reference assignment, disambiguation, enrich-
ment, etc. exploited by the narrator to create incongruity between his own and the
 Introduction 17

reader’s expectations of relevance, and the way relevance is actually achieved when
a British way of doing things is ultimately presented. Ruiz Moneva claims that
humour and irony reinforce each other because they both capitalize on augment-
ing the real or apparent contrast between a foreigner’s lifestyle and that of Britons.
Interestingly, the contrast in many cases turns out to be illusory, which makes for
Mikes’ inclusion in the group targeted as the object of mockery and satire. As Ruiz
Moneva observes, the distinction between humour and irony may not be sharp
and the interpretation of particular fragments as either ironic, humorous or both
may vary among readers. Along the mutual reinforcement of humour and irony,
this can be seen as an example of the synergic effect mentioned before, enabled by
the continuity of literalness, humorousness and irony postulated within relevance
theory and contrasting with formalist or Gricean approaches.

3. Some questions for the future

In this section I would like to offer a few remarks concerning possible questions or
problems that could be addressed in future relevance-theoretic research in order
to better elucidate factors that impact on the derivation of pragmatic effects in
various ways. They will pertain to the issues discussed in the consecutive parts of
the volume and will be presented in the same order.
As regards Part 1 of the volume, it can be stated that despite the abundant body
of relevance-theoretic research on the literalness-metaphor continuum, some of
its aspects require further investigation, especially the processing and outcome of
blending various figures from the continuum. Metonymy, on the other hand, has
attracted relatively little attention on the part of relevance theorists until recently,
and many of its aspects deserve further investigation, e.g. its relation to polysemy
or the question of what constitutes the relevance-theoretic analogue of ‘the same
domain’ to which the encoded concept and the communicated concept are sup-
posed to belong. Experimental research on developmental aspects of figurative
language, including metonymy (Falkum, 2019) will definitely throw new light on
the so far less understood facets of this figure.
Linguistic items discussed in Part 2, i.e. evidential participles and procedural
markers raise questions about the relationship between understanding and be-
lieving, i.e. between the inferential comprehension mechanism and epistemic
vigilance mechanisms that verify the credibility of information obtained in
communication. Sperber et al. (2010) advocate the view that activating a set of
contextual assumptions for the sake of understanding a message makes these as-
sumptions natural candidates to be used in the process of verifying whether the
message should be believed or not. On the other hand, the massive modularity
18 Agnieszka Piskorska

view espoused by Sperber (2002, 2005) holds that cognitive mechanisms, referred
to as modules, are specialized for processing domain-specific inputs and as such
they do not exchange information between one another. Addressing the ques-
tion of possible interdependence between modules would also have important
implications for understanding the relationship between linguistic decoding and
inference. All in all, despite the fact that the massive modularity view is widely
accepted in relevance theory, its consequences for the interplay of linguistic de-
coding, inferential comprehension, epistemic vigilance and affective mechanisms
have yet to be fully explored.
The issues discussed in Part 3, i.e. the contribution of typographical features
to meaning, are related to the development of technology, which offers new ways
of exploiting multimodality in communication. Because of that it is essential to
understand the mechanisms underpinning the interaction of visual elements with
the linguistic form of the message, contextual assumptions used in the compre-
hension process and other factors affecting the hearer’s expectations of relevance.
One can be certain that more technological innovations will be utilized in digital
communication in the future, so the theory will have to meet the challenge of
accounting for them.
With respect to the theme of Part 4, i.e. the interplay of cognitive effects and
affective responses to cognitively processed stimuli, I would like to start with
stating the obvious, namely that relevance theory is a theory of communication
and cognition. In the light of this, various attempts to modify or extend relevance
theory so that it would include a component accounting for emotions or social
aspects of communicative interactions should be made with caution. As I argue
elsewhere (Piskorska, 2017), the findings of relevance theory can be fruitfully
combined with theories explaining societal factors of communication without
any necessity or need to incorporate tools belonging to these theories into the
relevance-theoretic apparatus. The same applies to psychological theories of emo-
tions, especially those couched in terms of appraisal (cf. Moors, Ellsworth, Scherer
& Frijda, 2013). Having said this, I would like to emphasize that, on the other
hand, it is natural for researchers dealing with communication intended to evoke
affective responses to pose questions about the relationship of these responses to
comprehension. Verbal humour is, after all, primarily intended to amuse the audi-
ence and hate speech is intended to induce hate, rather than inform the audience
about the speaker’s hateful attitude. Keeping in mind the above-stated disclaimer
concerning the integrity of relevance theory, it is therefore interesting to examine
how and to what extent psychological theories of emotions and relevance theory
can be mutually informed. Highly promising in this respect is recent work by
Wharton and Strey (2019).
 Introduction 19

As far as the interpretation of literary works is concerned, despite the fact that
a number of researchers have convincingly shown how inferential mechanisms
are employed in the interpretation of literary discourse (e.g. Uchida, 1998; Clark,
1996, 2009; Wilson, 2011b, 2018), the fundamental question of how informa-
tion included in fiction leads to relevant implications for the recipient should be
given more attention, especially in the case of some genres that have recently won
considerable popularity. As Wilson points out, literary texts create two types of
expectations of relevance, one that she calls “internal” (2011b, p. 76), i.e. the reader
expects to be able to make sense of a plot or a poem, and the other that can be
referred to as external, i.e. the reader works out some implications and incorpo-
rates them into his cognitive environment as true representations of the world.
As I mentioned (following Wilson, 2011b, 2018) in the presentation of Part 5, the
expectations of external relevance can be satisfied due to the fact that the reader
processes the content of literary works in the context of assumptions retrieved
from his cognitive environment, which is a representation of reality. At this point
the question arises about the interpretation of literature that is ostensibly discon-
nected from reality, such as the fantasy genre, typically set in imaginary universes.
Wilson (2018, p. 203) remarks that “some fiction (e.g. genre novels, romantic or
detective fiction) may be written largely to entertain, and have mostly the ‘internal’
type of relevance.” In my opinion, the issue of the type of expectations of relevance
raised by fantasy and related genres deserves close scrutiny in the light of interest
and popularity they enjoy.

References

Athanasiadou, A. (2017). Introduction. Figurative thought, figurative language, figurative gram-

mar? In A. Athanasiadou (Ed.), Studies in figurative thought and language (pp. 1–14). John
Benjamins, Amsterdam.  ​https://‍doi.org/10.1075/hcp.56.int
Barsalou, L. (1983). Ad hoc categories. Memory and Cognition, 11(3), 211–227. ​
https://‍doi.org/10.3758/BF03196968
Blakemore, D. (1987). Semantic constraints on relevance. Oxford: Blackwell.
Blakemore, D. (2008). Apposition and affective communication. Language and Literature, 17(1),
37–57.  ​https://‍doi.org/10.1177/0963947007085054
Carston, R. (2002). Thoughts and utterances: The pragmatics of explicit communication. Oxford:
Blackwell.  ​https://‍doi.org/10.1002/9780470754603
Carston, R. (2010). Metaphor: Ad hoc concepts, literal meaning and mental images. Proceedings
of the Aristotelian Society, 110 (3), 297–323.
Carston, R. (2012). Metaphor and the literal/nonliteral distinction. In K. Allan, & K. M. Jaszc-
zolt (Eds.), The Cambridge handbook of pragmatics (pp. 469–92). Cambridge: Cambridge
University Press.  ​https://‍doi.org/10.1017/CBO9781139022453.025
20 Agnieszka Piskorska

Carston, R. (2016). The heterogeneity of procedural meaning. Lingua, 175–176, 154–166. ​

https://‍doi.org/10.1016/j.lingua.2015.12.010
Carston, R., & Wearing, C. (2015). Hyperbolic language and its relation to metaphor and irony.
Journal of Pragmatics, 79, 79–92. .  ​https://‍doi.org/10.1016/j.pragma.2015.01.011
Clark, B. (1996). Stylistic analysis and relevance theory. Language and Literature, 5, 163–178. ​
https://‍doi.org/10.1177/096394709600500302
Clark, B. (2009). Salient inferences: Pragmatics and the inheritors. Language and Literature, 18,
173–213.  ​https://‍doi.org/10.1177/0963947009105343
Colston, H. L. (2015). Using figurative language. Cambridge: Cambridge University Press. ​
https://‍doi.org/10.1017/CBO9781316226414
Colston, H. L. (2017). Pragmatic effects in blended figures. The case of metaphtonymy. In A.
Athanasiadou (Ed.), Studies in figurative thought and language (pp. 273–294). John Ben-
jamins, Amsterdam.  ​https://‍doi.org/10.1075/hcp.56.11col
Colston, H. L., & O’Brien, J. (2000). Contrast of kind versus contrast of magnitude: The prag-
matic accomplishment of irony and hyperbole. Discourse Processes, 30 (2), 179–199. ​
https://‍doi.org/10.1207/S15326950DP3002_05
Falkum, I. (2011). The semantics and pragmatics of polysemy: A relevance-theoretic account. PhD
Thesis. London: UCL.
Falkum, I. L. (2019). Metaphor and metonymy in acquisition. A relevance-theoretic perspec-
tive. In K. Scott, R. Carston, & B. Clark (Eds.), Relevance: Pragmatics and interpretation
(pp. 205–217). Cambridge: Cambridge University Press. ​
https://‍doi.org/10.1017/9781108290593.018
Gutt, E.-A. (2000). Translation and relevance. Manchester: St. Jerome.
Hall, A. (2017). Lexical pragmatics, explicature and ad hoc concepts. In I. Depraetere & R.
Salkie (Eds.), Semantics and pragmatics: Drawing a line (pp. 85–100). Cham, Switzerland:
Springer.  ​https://‍doi.org/10.1007/978-3-319-32247-6_6
Ifantidou, E. (2001). Evidentials and relevance. Amsterdam: John Benjamins. ​
https://‍doi.org/10.1075/pbns.86
Jodłowiec, M. (2015). The challenges of explicit and implicit communication. A relevance-theoretic
approach. Frankfurt: Peter Lang.
Jodłowiec, M., & Piskorska, A. (2015). Metonymy revisited: Towards a new relevance-theoretic
account. Intercultural Pragmatics, 12 (2), 161–187.  ​https://‍doi.org/10.1515/ip-2015-0009
Lakoff, G. (1987). Women, fire, and dangerous things: What categories reveal about the mind.
Chicago: The University of Chicago Press. ​
https://‍doi.org/10.7208/chicago/9780226471013.001.0001
Lakoff G., & Johnson, M. (1980). Metaphors we live by. Chicago: The University of Chicago
Press.  ​https://‍doi.org/10.7208/chicago/9780226470993.001.0001
Moors, A., Ellsworth, P. C., Scherern, K., & Frijda, N. H. (2013). Appraisal Theories of Emotion:
State of the Art and Future Development. Emotion Review, 5(2), 119–124. ​
https://‍doi.org/10.1177/1754073912468165
Padilla Cruz, M. (2015). On the role of vigilance in the interpretation of puns. Humor, 28(3),
469–90.  ​https://‍doi.org/10.1515/humor-2015-0068
Papafragou, A. (1996). On metonymy. Lingua, 99, 169–195. ​
https://‍doi.org/10.1016/0024-3841(96)‍00016-2
Piskorska, A. (2017). Editorial: relevance theory and intercultural communication problems.
Research in Language, 15(1), 1–9.  ​https://‍doi.org/10.1515/rela-2017-0005
 Introduction 21

Popa, M. (2010). Ironic metaphor: A case for metaphor’s contribution to truth-conditions In

E. Wałaszewska, M. Kisielewska-Krysiuk & A. Piskorska (Eds.) In the mind and across
minds: A relevance-theoretic perspective on communication and translation (pp. 224–245).
Newcastle: Cambridge Scholars Publishing.
Scott, K. (2017). Prosody, procedures and pragmatics. In I. Depraetere & R. Salkie (Eds.), Se-
mantics and pragmatics: Drawing a line (pp. 323–341). Berlin: Springer. ​
https://‍doi.org/10.1007/978-3-319-32247-6_18
Sperber, D. (2002). In defense of massive modularity. In E. Dupoux (Ed.), Language, brain and
cognitive development: essays in honor of Jacques Mehler (pp. 47–57). Cambridge Mass: MIT
Press.
Sperber, D. (2005). Modularity and relevance: How can a massively modular mind be flexible
and context-sensitive? In P. Carruthers, S. Laurence, & S. Stich (Eds.), The innate mind:
structure and content (pp. 53–68). Cambridge: Cambridge University Press. ​
https://‍doi.org/10.1093/acprof:oso/9780195179675.003.0004
Sperber, D., & Wilson, D. (1986). Relevance: Communication and cognition (1st ed.). Oxford:
Blackwell.
Sperber, D., & Wilson, D. (1995). Relevance: Communication and cognition (2nd ed.) Oxford:
Blackwell.
Sperber, D., & Wilson, D. (1998a). Irony and relevance: A reply to Seto, Hamamoto and Ya-
manashi. In R. Carston, & S. Uchida (Eds.), Relevance theory. Applications and implications
(pp. 283–293). Amsterdam: John Benjamins.  ​https://‍doi.org/10.1075/pbns.37.16spe
Sperber, D., & Wilson, D. (1998b). The mapping between the mental and the public lexicon.
In P. Carruthers, & J. Boucher (Eds.), Thought and language (pp. 184–200). Cambridge:
Cambridge University Press.  ​https://‍doi.org/10.1017/CBO9780511597909.012
Sperber, D., & Wilson, D. (2008). A Deflationary account of metaphors. In R. W. Gibbs (Ed.),
The Cambridge handbook of metaphor and thought (pp. 84–105). Cambridge: Cambridge
University Press.  ​https://‍doi.org/10.1017/CBO9780511816802.007
Sperber, D., Clément, F., Heintz, C., Mascaro, O., Mercier, H., Origgi, G., & Wilson, D. (2010).
Epistemic Vigilance. Mind & Language, 25/4, 359–393. ​
https://‍doi.org/10.1111/j.1468-0017.2010.01394.x
Uchida, S. (1998). Text and relevance. In R. Carston, & S. Uchida (Eds.), Relevance theory. Ap-
plications and implications (pp. 161–178). Amsterdam: John Benjamins. ​
https://‍doi.org/10.1075/pbns.37.10uch
Unger, C. (2011). Exploring the borderline between procedural encoding and pragmatic infer-
ence. In V. Escandell-Vidal, M. Leonetti & A. Ahern (Eds.), Procedural meaning: Problems
and perspectives (pp. 103–127). Emerald Group Publishing.
Wharton, T. (2009). Pragmatics and non-verbal communication. Cambridge: Cambridge Univer-
sity Press.  ​https://‍doi.org/10.1017/CBO9780511635649
Wharton, T. (2012). Prosody and meaning: Theory and practice. In J. Romero-Trillo (Ed.),
Pragmatics and prosody in English language teaching (pp. 97–116). Dordrecht: Springer. ​
https://‍doi.org/10.1007/978-94-007-3883-6_7
Wharton, T., & Strey, C. (2019). Slave of the passions. Making emotions relevant. In K. Scott,
R. Carston, & B. Clark (Eds.), Relevance: Pragmatics and interpretation (pp. 253–266).
Cambridge: Cambridge University Press.  ​https://‍doi.org/10.1017/9781108290593.022.
Wilson, D. (2000). Metarepresentation in linguistic communication. In D. Sperber (Ed.), Meta-
representations (pp. 411–448). Oxford: Oxford University Press.
22 Agnieszka Piskorska

Wilson, D. (2006). The pragmatics of verbal irony: Echo or pretence? Lingua, 116, 1722–1743. ​
https://‍doi.org/10.1016/j.lingua.2006.05.001
Wilson, D. (2011a). Parallels and differences in the treatment of metaphor in relevance theory
and cognitive linguistics. Intercultural Pragmatics, 8(2), 177–196. ​
https://‍doi.org/10.1515/iprg.2011.009
Wilson, D. (2011b). Relevance theory and the interpretation of literary works. UCL Working
Papers in Linguistics, 23, 47–68.
Wilson, Deirdre. (2011c). The conceptual-procedural distinction: Past, present and future. In V.
Escandell-Vidal, M. Leonetti & A. Ahern (Eds.), Procedural meaning: Problems and perspec-
tives (pp. 3–31). Emerald Group Publishing. ​
https://‍doi.org/10.1108/S1472-7870(2011)‍0000025005
Wilson, D. (2013). Irony comprehension: A developmental perspective. Journal of Pragmatics
59(A): 40–56.  ​https://‍doi.org/10.1016/j.pragma.2012.09.016
Wilson, D. (2015). Explaining metonymy. Paper delivered at Relevance Round Table Meeting 4.
Institute of English Studies, Jagiellonian University in Kraków.
Wilson, D. (2016). Reassessing the conceptual-procedural distinction. Lingua, 175–176, 5–19. ​
https://‍doi.org/10.1016/j.lingua.2015.12.005
Wilson, D. (2018). Relevance theory and literary interpretation. In T. Cave, & D. Wilson (Eds.),
Reading beyond the code: Literature and relevance theory (pp. 185–204). Oxford: Oxford
University Press.
Wilson, D., & Carston, R. (2007). A unitary approach to lexical pragmatics: Relevance, inference
and ad hoc concepts. In N. Burton-Roberts (Ed.), Pragmatics (pp. 230–259). Basingstoke:
Palgrave Macmillan.  ​https://‍doi.org/10.1057/978-1-349-73908-0_12
Wilson, D., & Sperber, D. (1993). Linguistic form and relevance. Lingua, 90, 1–25. ​
https://‍doi.org/10.1016/0024-3841(93)‍90058-5
Wilson, D., & Sperber, D. (2012). Explaining Irony. In Wilson, D., & Sperber, D. Relevance and
meaning (pp. 123–145). Cambridge: Cambridge University Press. ​
https://‍doi.org/10.1017/CBO9781139028370.008
Wilson, D., & Wharton, T. (2006). Relevance and prosody. Journal of Pragmatics, 38(10),
1559–1579.  ​https://‍doi.org/10.1016/j.pragma.2005.04.012
Yus, F. (2003). Humor and the search for relevance. Journal of Pragmatics, 35, 1295–1331. ​
https://‍doi.org/10.1016/S0378-2166(02)‍00179-0
Part 1

Continua in non-literalness
 Chapter 1

Category extension as a variety of loose use

Ewa Wałaszewska
University of Warsaw, Poland

The chapter focuses on category extension, typically illustrated by the use of sa-
lient brand names such as Hoover for any vacuum cleaner, and seeks to provide a
comprehensive discussion of this relevance-theoretic notion. Category extension
is regarded as part of the continuum of loose uses, along with approximation,
hyperbole and metaphor, but, compared with the other varieties, it has not been
given due attention. The chapter seeks to clarify the theoretical status of category
extension, by discussing its relation to the other varieties of loose use and by
analysing different cases of the phenomenon falling into two groups: limited
and creative category extensions. The chapter also shows how concepts such as
lexical warfare, paragons, snowclones, schemata and pragmatic routines may
improve the understanding of category extension.

Keywords: category extension, lexical modulation, broadening, loose use,

relevance theory, snowclones, schemata, pragmatic routines

1. Introduction

Category extension is a term used in relevance theory to refer to a specific broad-

ening of lexical meaning, which can be illustrated by the use of salient brand
names such as Xerox for any photocopier, and similar uses of common nouns and
personal names (Wilson, 2003; Wilson & Carston, 2007; Sperber & Wilson, 2008).
For example, in the sentence He is another Shakespeare, the proper noun may be
used as a name of the category of immensely talented writers1 since Shakespeare
seems to be viewed as an ideal member for this category. This brings to mind the

1. This interpretation of the use of Shakespeare as a category name is most likely to be accepted
as the default one, but it is, by no means, the only plausible reading. Under appropriate contex-
tual conditions, the proper noun could also serve as a name of different categories whose central
member (and point of reference) is Shakespeare, for example, the category of playwrights
contemporary to Shakespeare and having the same social prominence, etc.

https://‍doi.org/10.1075/ftl.8.01wal
© 2020 John Benjamins Publishing Company
26 Ewa Wałaszewska

use of the notion of ‘category extension’ as inextricably connected with that of cat-
egory prototype. This understanding of category extension can be found in cogni-
tive linguistic approaches (e.g. Lakoff, 1987; Taylor, 1989/1995) or in Kronenfeld’s
(1988, 1996; see also Brown, 1990) anthropological semantics. A category may
develop its membership by extending the boundary from a prototype to related
non-prototypical entities. Category extension may be based on relations between
a prototype and non-prototypical entities motivated by similarity (e.g. perceptual),
metaphor and metonymy. In conceptual blending theory (e.g. Fauconnier &
Turner, 1998), on the other hand, category extension is listed as one of the many
functions of blends: a conventional category is extended to new material. Unlike
permanent category changes, category extensions are provisional, determined by
the local purposes of expression or naming.
In relevance theory, category extension is regarded as a case of broadening,
belonging to the continuum of loose uses, along with approximation, hyperbole
and metaphor. This continuum is typically presented by relevance theorists in such
a way that only the three phenomena are described in relation to one another,
while category extension is usually ignored. Even though it is clearly declared
to belong to the continuum, its status and relation to the other varieties of loose
use are somewhat vague. Therefore, the aim of this chapter is to describe this
phenomenon in greater detail to throw some light on its nature and place in the
postulated continuum.
The chapter also discusses the assumed division of category extensions into
two types by analysing various examples of both types. The first type, so-called
‘limited category extensions,’ is described in terms of changes to lexical properties
and is shown to be a possible cause of ‘lexical warfare,’ understood by Ludlow
(2014, p. 7) as “battles over how a term is to be understood”. The second type,
that of ‘creative category extensions’ is confronted with a range of phenomena dis-
cussed by other authors and on different occasions such as paragons, snowclones,
schemata and pragmatic routines, which suggests that category extensions of this
type may be not as creative as they seem.

2. Lexical modulation

It is now generally accepted in pragmatics that lexically-encoded meanings may

get modified in use. Such contextual modifications and adjustments of word
meanings have been extensively investigated within the field of lexical pragmat-
ics, a relatively new and dynamic area of linguistic research, from at least two
broad rival perspectives: neo-Gricean (e.g. Blutner, 1998, 2011; Horn, 2007, 2008;
Huang, 2009) and relevance-theoretic (e.g. Carston, 2002; Wilson & Carston,
 Chapter 1. Category extension as a variety of loose use 27

2007; Sperber & Wilson, 2008; Wilson & Sperber, 2012; Wałaszewska, 2015).
Often-quoted examples of such meaning modulations include the noun money or
the verb drink (Examples (1)–(2), both used to show that word meanings may be
narrowed, and the adjectives raw and empty (Examples (3)–(4)), typically used to
illustrate how word meanings may by loosened (or broadened).
(1) You need money to buy a Porsche.
(2) It is very important not to drink early in the pregnancy.
(3) This meat is raw.
(4) After the party, he dumped empty bottles into the recycling bin.

In (1) and (2), the noun money and the verb drink are used to communicate some-
thing more specific (narrower) than their encoded meanings; a large amount of
money and drink (a lot of) alcohol, respectively. In (3) and (4), the adjectives will
be interpreted as communicating meanings that are less specific (looser) than the
encoded ones. If uttered of an undercooked meat, raw is used loosely to mean not
that it has not been cooked at all but that it has not received sufficient thermal
processing to be edible or to suit the speaker’s taste. The bottles described as empty
may not be literally empty; they may still contain small amounts of wine, beer, etc.
It is worth noting that neo-Gricean approaches do not really address the
problem of how broadened meanings are generated, which may be associated with
the apparent lack of the tools needed to cope with the issue of broadening in a
satisfactory theory-consistent way. They tend to focus on narrowed (default or
stereotypical) interpretations, which might be related to the assumption, generally
endorsed by such approaches, that lexical meaning is underspecified and hence
has to be enriched (narrowed) in the process of interpretation (see e.g. Huang,
2009). On the other hand, relevance-theoretic research, though oriented to both
types of pragmatic meaning modulation, seems to be more extensive in the case of
broadened meanings (Wałaszewska, 2015, pp. 18–19).

2.1 Ad hoc concept construction: Narrowing and broadening

In relevance theory, lexical modulation is assumed to occur “when the meaning

contributed by a word to the utterance meaning differs from the stable lexically
encoded meaning of the word” (Allott, 2010, p. 109). Lexical modulation is also
known as ad hoc concept formation since, in relevance theory, (in most cases)
word meanings come from the associated concepts, and pragmatically modulated
28 Ewa Wałaszewska

meanings require the construction of unlexicalised ad hoc concepts.2 These are

constructed in the process of utterance interpretation in response to expectations
of relevance and derived from the interaction of information stored in the logical
and encyclopaedic entries3 for lexicalised concepts with the context and the prin-
ciples of relevance (Carston, 1997, 2002). In particular, a given lexicalised concept
activated by the use of the encoding word may be employed as the input for the
construction of a non-lexicalised concept during the utterance interpretation. An
ad hoc concept constructed via the process of narrowing picks out only “a subset
of the items that fall under the encoded concept,” which means that the word is
used “with a more restricted denotation” (Wilson & Carston, 2007, p. 232). An ad
hoc concept yielded on the basis of broadening involves the extension of the set
of items that fall under the encoded concept, and the word is used “with a con-
sequent [marginal or substantial – E.W.] expansion of the linguistically-specified
denotation” (Wilson & Carston, 2007, p. 234; see also Wałaszewska, 2015).
Unlike narrowing, broadening is necessarily related to the suspension
of literalness since in the construction of an ad hoc concept some of the logi-
cal information associated with the lexically-encoded concept may be dropped
and the resulting ad hoc concept may traverse the boundaries of the lexicalised
concept. Cases of broadening, as generally assumed in relevance theory, include
approximation, category extension, hyperbole and metaphor. Since relevance
theory postulates the existence of a continuum between literal and non-literal uses
of language, it is not surprising that loose (non-literal) uses of language form a
continuum too, with no easily identifiable cut-off points between them (Wilson &
Carston, 2007; Sperber & Wilson, 2008; Wilson & Sperber, 2012; Wilson, 2014).
Hence, on the relevance-theoretic approach, approximation, category extension,
hyperbole and metaphor are not distinct types; “they merely occupy different
points on a continuum of degrees of broadening” (Carston, 2012, p. 479). On the
other hand, the distinction between these varieties is motivated by the amount
of looseness involved: there is a substantial broadening of the lexically encoded
concept in the case of hyperbole, whereas the broadening is really slight in the case
of approximation. When compared with hyperbole, metaphor is viewed “as a still
more radical variety of broadening (…), involving a greater departure from the

2. As observed by Hall (2017, p. 98, n.11), the name ad hoc concept may be misleading as it
suggests the construction of something totally novel. Actually, in relevance theory, the term is
intended to mean a non-lexicalised concept, be it a stable or a one-off entity.

3. In general, the difference between the two types of entries boils down to the kind of informa-
tion made accessible. The logical entry for a concept provides access to computational informa-
tion, whereas the encyclopaedic entry gives access to representational information (Sperber
& Wilson 1995, p. 89).
 Chapter 1. Category extension as a variety of loose use 29

encoded meaning” (Wilson & Carston, 2007, p. 235). Let us illustrate these three
cases of broadening with the following uses of the word marathon (taken from
Carston, 2016, pp. 203–204):
(5) I’m going to run a marathon for charity.
(6) My morning jog is a marathon.
(7) The presidential election campaign is a marathon.

On the assumption that the speaker’s intention is not to cover a distance of

precisely 42.195 kilometres (26.2 miles), but simply to run a slightly shorter race
(say, of 40 kilometres), the lexically-encoded concept marathon (‘a long-distance
running race with a distance of 42.195 kilometres’) activated by the use of the
word marathon in (5) is broadened to yield an ad hoc concept marathon*.4 Apart
from literal marathons, marathon* includes in its denotation also long-distance
races of slightly less than 42 kilometres, which are experienced as equally long and
demanding as marathons. Hence, (5) is an example of approximation: the differ-
ences between marathon and marathon* are inconsequential for the speaker’s
purpose at hand.
In (6), assumed to be uttered by a person who has started jogging in her local
park only recently, the word marathon is used to communicate a broader ad hoc
concept (marathon**) that includes both literal marathons and “runs that are
considerably shorter, but which, like actual marathons, are difficult and fatigu-
ing for the person who undertakes them” (Rubio Fernandez et al., 2013, p. 47).
Undoubtedly, marathon** is more radically broadened than marathon* and
the use of marathon in (6) illustrates hyperbole.
In (7), the communicated ad hoc concept marathon***, which includes not
only literal marathons, long-distance races, and difficult and fatiguing shorter
runs but also events or (both physical and psychological) activities characterized
by great length and concentrated effort, is more radically broadened than mara-
thon**. The example is a clear case of metaphorical use with the information
about the physical distance and the activity of running associated with the concept
marathon not being accessed. To sum up, “for each of these uses, a different
subset of the mass of information associated with the lexical concept marathon is
the most highly activated one and plays an instrumental role in determining which
particular ad hoc concept the speaker intended” (Carston, 2016, pp. 203–204).
What the examples above show is that the discussed three varieties of loose
use are clearly ordered with respect to the extent of broadening: each successive

4. The standard procedure is to use small caps to indicate context-independent lexicalised

concepts and small caps followed by an asterisk to indicate ad hoc concepts.
30 Ewa Wałaszewska

variety in the continuum is more radically broadened than the preceding one. A
question arises why category extension has been left out of the picture so far.

3. Category extension

As mentioned above, in relevance theory, category extension is one of the four

varieties of loose use that form a continuum between less radical and more radi-
cal broadenings. It is understood as involving the application of “a word with a
relatively precise sense to a range of items that clearly fall outside its linguisti-
cally specified denotation” (Sperber & Wilson, 2008, p. 91). Such an extension is
possible if the items inside and outside the word’s denotation share a number of
contextually relevant properties (Sperber & Wilson, 2008, p. 91). This description
of category extension is quite general and does not really help understand the na-
ture and mechanism of this instance of broadening. The assumption that it forms
a continuum with the other varieties of loose use is not helpful either since, while
discussing the continuum, relevance theorists typically focus on approximations,
hyperboles and metaphors (as illustrated by the above discussion of marathon):
they discuss the relationships between the three varieties, illustrate them with ex-
amples and pinpoint differences. Category extension is only mentioned in passing
and ignored in such analyses.
A notable exception comes from Sperber and Wilson (2008, p. 94), who not
only emphasise the existence of “a continuum of cases between category exten-
sion and metaphor” but also suggest the way to tease them apart. They argue that
“category extension involves the projection of defining, or at least characteristic,
properties of the encoded concept onto a broader category, (…) whereas the type
of broadening involved in metaphor is based on relatively peripheral or, at least,
contingent properties” (Sperber & Wilson, 2008, p. 94). Nevertheless, even if
this observation reveals the essential difference between category extension and
metaphor and as such may be used as a diagnostic to distinguish cases of one
variety from cases of the other, it does not imply much about the place of category
extension in the continuum or its relations with approximation and hyperbole.
At this point, it might be a good idea to supplement Carston’s (2016) set of
loose uses of the word marathon (see Examples (5)–(7) above) with two more
examples, both of which involve what is arguably felt to be a case of category
extension, to see how it might fit in the continuum.
 Chapter 1. Category extension as a variety of loose use 31

(8) He has swum a marathon. Yes, swam 26.2 miles. Incredible.5

(9) She swam a marathon for charity and raised £220.

As can be seen from (8) and (9), the property of the lexically-encoded concept
marathon which is not projected onto the broader category is that of running.
This property is still associated with the resulting ad hoc concept, but as one of a
number of properties specifying possible ways of moving over a distance, which
allows the ad hoc concept to include in its denotation not only running but also
swimming. As a matter of fact, it is possible to extend the category over various
ways of covering a distance such as running, swimming, but also inline skating or
walking (as there are inline skating marathons and walk marathons). On the other
hand, what is projected onto the broader category are the properties of physical
distance and activity of racing over a distance, which seem to be defining or, at least,
characteristic properties associated with the lexicalised concept. Hence, based on
Sperber and Wilson’s diagnostic, it might be concluded that the ad hoc concept
communicated by the uses of the word marathon in (8) and (9) is not a metaphor
but a case of category extension. Interestingly, cases of category extension may also
involve approximation or hyperbole; for instance, the use of the word marathon
in (9), in the same vein as (5) above, may be interpreted as an approximation. It
is also easy to imagine a newbie swimmer to proudly say I swam a marathon to
marvel at their achievement of covering just a few miles, which would illustrate a
hyperbolic use of category extension.
The above discussion shows that it is by no means obvious how to accom-
modate category extension into the relevance-theoretic continuum of loose uses.
This uncertain status of category extension can be demonstrated by the fact that
relevance-theoretic descriptions of the continuum slightly differ. The presently
accepted view is that, as discussed above, there is a continuum of loose uses or-
dered, from least to most radical, in the following way: approximation, category
extension, hyperbole and metaphor. In the earlier approach (e.g. Wilson, 2003),
approximation and category extension were considered to be the two major types
of broadening, with metaphor and hyperbole being described as more radical
varieties of category extension. On the other hand, as follows from a recent com-
ment made by Carston and Wearing (2015, p. 87, n.19) about the continuum of
loose uses, category extensions are apparently regarded as falling on the side of
“mundane non-figurative loose uses,” together with approximations, apparently
in opposition to hyperboles and metaphors. This view, however, is based on the
assumption that a typical example of category extension is the use of hoover for

5. Example taken from <http://‍www.beginnertriathlete.com/discussion/forums/thread-view.

asp?tid=59367>
32 Ewa Wałaszewska

any vacuum cleaner, not only for vacuum cleaners manufactured by the Hoover
company. It turns out, though, that the group of phenomena referred to as category
extensions is more varied and heterogeneous.

4. Types of category extension

As observed by Sperber and Wilson (2008), cases of category extension form a

continuum extending from “limited category extensions” to “creative category
extensions.”

4.1 Limited category extensions

Limited category extensions can be illustrated by Examples (10)–(12) taken from

Sperber and Wilson (2008, p. 91):
(10) [On a picnic, pointing to a flattish rock]: That’s a table!
(11) [Handing someone a paper tissue]: Here’s a Kleenex.
(12) [Handing someone a paper napkin]: Here’s a Kleenex.

Example (10) shows that common nouns, here table, can be interpreted as cases
of category extension. Certainly, a flattish rock is not the same as a table so it
cannot be categorized as one. On the other hand, the rock’s flat surface makes it
possible for picnickers to use it as a table on which food and drink can be placed.
The property of having a flat top, which is one of the most characteristic properties
associated with the concept table, gets projected onto the broadened category.
While this property gains salience, the (logical) property of being a piece of fur-
niture is dropped, which makes it possible for flattish rocks to be admitted into
the category. In (11), the salient brand name Kleenex is used to denote a broader
category of disposable tissues which also contains items from less salient brands.
In (12), Kleenex is used to denote a broader category of disposable tissues of any
brand and paper napkins, which can be used in the same way as tissues. The dif-
ferences between a table and a flattish rock, on the one hand, and between paper
tissues made by the Kleenex company and those that are not, or between paper
tissues and paper napkins, on the other, are inconsequential for a picnicker and
a tissue user, respectively. The ad hoc concepts table* (‘a flat surface at which
one can eat their food’), kleenex* (‘any disposable paper tissue’) and kleenex**
(‘any paper product which can be used instead of paper tissue’) are referred to as
“limited category extensions” (Sperber & Wilson, 2008; Wałaszewska, 2015).
 Chapter 1. Category extension as a variety of loose use 33

Sperber and Wilson do not explicitly motivate the term “limited category ex-
tension.” However, it may be related to the implicit assumption that there is a limit
beyond which a particular category cannot be plausibly extended. At this point
let us mention an attested example of what seems to be a case of limited category
extension used by Peter Ludlow (2014) in his discussion of meaning modulations.
[S]‍everal years ago Sports Illustrated produced a list of the top athletes of the 20th
century, to which they added the famous racehorse Secretariat. I was listening to
a sports talk radio show when viewers called in to complain that horses can’t be
athletes. (Ludlow, 2014, p. 85)

This example shows that extending the denotation of the concept athlete (‘a per-
son who is very good at sports or physical exercise, especially one who competes
in organized events’) so as to include racehorses (or this particular racehorse in
Ludlow’s example) was not found acceptable by some people.6 The controversy
aroused by using the word athlete to refer not only to human athletes but also
to racehorses shows how category extension (or meaning modulation in general)
may result in what Ludlow calls ‘lexical warfare,’ understood here as disagree-
ments between people about the limits of meaning modulation lexical items may
undergo in use.7 According to Ludlow (2014, p. 7), in a number of cases, “meaning
modulation is automatic and to some degree cooperative.” But there are also cases
of meaning modulation which make people not only aware of the ongoing process
but also actively involved in a dispute, or even a fight, over a word’s meaning.
This shows that some meaning modulations may be interpreted in the intended
way, but not accepted and sometimes actively rejected. This may, in turn, prevent
them from stabilisation and conventionalisation, which would ultimately lead to a
change in the meaning of the employed word.
If, as Sperber and Wilson (2008, p. 94) suggest, category extension involves
the projection of defining, or at least typical, properties of the lexically encoded
concept onto a broader category, what remains to be seen is what properties are
taken to be defining/typical in the case of the concept ATHLETE. Based on the
answers given to the question “Are racehorses athletes?” on the question-and-
answer website “Yahoo! Answers”8 and on the Internet forum “Tigerdroppings,”9

6. As suggested by one of the reviewers, this may be related to contextual expectations: readers
of a sports magazine may expect relatively fine category distinctions in the area of sport, while
they may be more liberal in a more general context.

7. It is worth noting that, for Ludlow (2014), lexical warefare is typical of political discourse.

8. <https://‍answers.yahoo.com/question/index?qid=20070320174457AAmLVCq>

9. <http://‍www.tigerdroppings.com/rant/more-sports/are-race-horses-athletes/7248206/>
34 Ewa Wałaszewska

it is possible to identify some of those properties that people think justify that
racehorses should be included in, or excluded from, the category of athletes. Some
of the inclusion-view supporters notice that racehorses “have intense training
schedules and carefully regimented exercises. They are supervised by trainers
and grooms in the same way human athletes have coaches and personal trainers”
(app1188); “they have fitness levels, training, competition, divisions and rankings
for winners…” (doingitright44). The exclusion-view proponents point out that
“horses aren’t human” (Cdawg) or that horses cannot make “a conscious decision
to be an athlete” (Lisa) in the way that humans do. The above shows that if, for
example, properties such as ‘being a human’ or ‘being able to take a conscious
decision to be an athlete’ are taken to be defining in the case of ATHLETE, then
the broadening of the concept so as to include non-human creatures incapable of
conscious decision making is not likely to be (easily) accepted and the category
extension may be rejected, even as a one-off use. On the other hand, if properties
such as ‘having coaches, special diets, trainings, competitions, etc.’ are regarded
as typical of ATHLETE, they gain salience over other properties and license
category extension.

4.2 Creative category extensions

Creative category extensions can be illustrated by Examples (13)–(15).

(13) [Discussing tanning on sunbeds]: Sunbeds are the new cigarettes. 
 (Sperber & Wilson, 2008, p. 91)
(14) [Discussing Russia’s foreign policy in the near abroad]:
Putin is the new/next Hitler.  (Wałaszewska, 2015, p. 119)
(15) [Discussing Russia’s foreign policy in the near abroad]:
Putin is another Hitler.

What the above examples communicate is not that the differences between sunbeds
and cigarettes, as in (13), or Putin and Hitler, as in (14)–(15), are insignificant.
Rather, (13) shows that in the category of dangerous or destructive addictions,
sunbeds are put in the place occupied so far by cigarettes. Thus, the word cigarettes
is used to refer to a broader category of dangerous addictions of which literal
cigarettes are the most salient member. In (14), Hitler is used as a common noun
to represent a broader category of dictators and sociopaths, dreaming of world
domination. Similarly, in (15), Hitler is the most salient member of the broadened
category and hence his name is used as a common noun to denote it. There is an
important difference between (14) and (15): in (14), Putin occupies the place in
the broader category previously taken by Hitler (he is suggested as the most salient
 Chapter 1. Category extension as a variety of loose use 35

example), whereas in (15), Putin is just a member of the broader category, whose
most salient example is still Hitler.
The examples of category extension in (14) and (15) bring to mind Lakoff ’s
(1987, p. 87) discussion of category models organised around a ‘paragon.’ A
paragon is defined as an individual member of a category that “represent[s] ei-
ther an ideal or its opposite” (Lakoff, 1987, p. 87; see also Lakoff, 1999, p. 404). A
paragon performs a dual function: first, it represents the ideal of the category to
which it belongs and second, it stands for the whole category the ideal of which
it represents. Hence, it serves as a cognitive ‘reference point’ (Langacker, 1993),
and establishes norms and raises expectations against which other members of the
category are evaluated, giving rise to typicality effects (there are better and worse
members) (Evans & Green, 2006, pp. 272–273). According to Lakoff, paragons are
typically used in constructions such as a regular X, another X, the X of Ys (e.g. the
Cadillac of vacuum cleaners).
It is worth noting that, not all of these constructions will be interpreted, from
a relevance-theoretic perspective, as cases of category extensions. For example,
Leonard Cohen is the Lord Byron of rock music, which involves referring to the
popular singer with the name of the famous romantic poet, has the effect of el-
evating him and his songs to a higher level of culture (Davis, 1996, p. 445) and
appears to involve a metaphor. In relevance-theoretic terms, broadenings which
cause the lexicalised concept to lose some of its central or typical properties, or
in other words, broadenings based on relatively peripheral or, at least, contingent
properties, are better described as metaphors than category extensions, though
there is no clear demarcation between the two. On the other hand, if we compare
(16) and (17), the former is a clear case of metaphor involving the broadening
based on the relatively peripheral properties of being proud and arrogant and
having psychological problems, whereas the latter is more likely to be a category
extension since it involves the projection of characteristic properties such as being
a dictator restricted to the context of Africa.
(16) [attributed to veteran Brazilian footballer Giovanni as his opinion about the
former coach Louis Van Gaal]:
Van Gaal is the Hitler of the Brazilian players; he is arrogant, proud and has
a problem.10

10. Example taken from <http://‍www.goal.com/en/news/12/spanish-football/2010/​06/​01/​

1952469/​louis-van-gaal-is-the-hitler-of-brazilian-players-brazilian>
36 Ewa Wałaszewska

(17) [a comment about the president of Rwanda, Paul Kagame]: Kagame is the
Hitler of Africa.11

Particularly interesting are examples of creative category extensions that involve

the pattern ‘X is the new Y’. Most of the examples used by relevance theorists
to illustrate creative category extensions are based on this pattern (Wilson,
2004, p. 345):
(18) [in fashion journalism] Brown is the new black.
(19) [in cookery writing] Mint is the new basil.
(20) [in interior design writing] Is oak the new pine?
(21) Syria must not become the new Bosnia.  (Hall, 2017, p. 86)
(22) A hundred-km trail run is the new marathon.12

In (18), black evokes the category of staple colours in one’s wardrobe; in (19),
basil evokes the category of fashionable herbs; in (20), pine evokes the category
of trendy furniture wood (Wilson, 2004, p. 345). The name Bosnia in (21) is used
not to refer to the country or even to the armed conflict that took place in Bosnia
and Herzegovina in the early 1990s, but to “a broader category, paraphrasable
as something like conflicts where earlier foreign intervention might have averted
descent into all-out war” (Hall, 2017, p. 86) of which the Bosnian war is still the
most salient example (though the Syrian civil war is gaining more salience in this
category). It is worth noting that Hall’s analysis in terms of category extension
has been criticised on the grounds that the use of the name of a certain place for
the event that took place there is typical of a metonymical relation and as such it
is “a more systematic kind of extension than [the] ad hoc paraphrase suggests”
(Lemmens, 2017, p. 111). This critical comment seems to be off-target since the
presence of a metonymical relation has no direct bearing on the constructed ad
hoc concept. Actually, what the example shows is that category extensions may
exploit metonymical relations by broadening metonymic concepts. A connection
between category extension and metonymy certainly deserves further investiga-
tion (for an interesting discussion of metonymy in relevance theory, see Jodłowiec
& Piskorska, 2015).
Finally, in (22), the use of the word marathon communicates the ad hoc concept
of demanding long running races that require a lot of stamina and preparation,

11. Example taken from <http://‍rwandarwabanyarwanda.over-blog.com/article-rwanda-

kagame-has-referred-himself-as-hitler-48215025.html>

12. Example taken from <http://‍paleo-runner.blogspot.com/2013/08/steep-is-new-flat.html>

 Chapter 1. Category extension as a variety of loose use 37

which includes in its denotation both marathons involving a distance of 42.195

kilometres run on a flat road surface and runs over a longer distance on hiking
trails in mountainous terrain.
Sentences following the pattern ‘X is the new Y’ were recognized by Glen
Whitman (2004) as typical examples of what he called ‘snowclones,’ referring to
the phenomenon earlier described by Pullum (2003) as “a multi-use, customizable,
instantly recognizable, time-worn, quoted or misquoted phrase or sentence that
can be used in an entirely open array of different jokey variants by lazy journalists
or writers.” The name ‘snowclone’ comes from the formula If Eskimos have N words
for snow, X surely have Y words for Z, which is quite often encountered in journal-
istic writing even though the basic assumption about the language of the Inuit is at
least suspicious. Other snowclones include In space, no one can hear you X or Keep
calm and X (Pullum, 2004). To borrow an example from Mark Liberman (2005),
in the snowclone X is the new communism, as of 2005, values of X included Islam,
Frenchness, capitalism, ultra-liberalism, libertarianism, trashy-chic fashion, ter-
ror, terrorism, and organized crime, to which it is now possible to add corruption,
multiculturalism, environmentalism or populism.
Zwicky (2006) provides a reasonable reconstruction of how snowclones come
into being, by suggesting four stages:
a. At the pre-formula stage, a certain idea is expressed in various ways, e.g. ‘what
one person likes, another person detests,’ ‘things that please some people repel
others,’ which are all understood literally, require no special knowledge to
understand, and may be readily created.
b. At the stage of first fixing, a memorable catchy expression is produced for con-
veying the idea. The expression in its fixed form spreads, and becomes popular
as a cliché, proverb, quotation, or well-known formula, e.g. ‘One man’s meat is
another man’s poison.’
c. At the stage of variation, the fixed expression may become extended with the
development of open slots, or via puns or other changes to it, e.g. ‘One man’s
trash is another man’s treasure’ or ‘One man’s Mede is another man’s Persian.’
d. At the stage of second fixing or snowcloning, such variants become (relatively)
fixed as schemas with open slots in them, as in ‘One man’s X is another man’s
Y.’
To put it in other words, in a snowclone, a specific expression which is fixed
becomes less fixed as a result of introducing a variable, while the original
meaning of the construction generalizes. The increase in schematicity results
in increased productivity, at least, over a short period of time (Traugott, 2014,
pp. 97–99; see also Traugott & Trousdale, 2013). Within the schema ‘X is the
new Y,’ the subschema ‘X is the new black’ has proved particularly attractive and
38 Ewa Wałaszewska

hence productive.13 Following the 2012 Wikipedia article, Traugott (2014, p. 99)
describes it in the following way:
‘X is the new black’ is an expression used to indicate the sudden popularity or
versatility of an idea at the expense of the popularity of a second idea. It is the
originator of the phrasal template ‘X is the new Y.’ The phrase (…) has often been
used for ironic or humorous purposes.

In general, the pattern ‘X is the new black’ or ‘X is the new Y’ and, actually, most
of the snowclones can be described as schemata. Schemata are similar to formu-
laic expressions (e.g. idioms, proverbs) since both types share certain form- and
meaning-related properties, for example, they have canonical form, a fixed order
of certain lexical items or stereotyped intonation and they trigger particular con-
notations or social meanings (Van Lancker Sidtis, 2004). Moreover, these proper-
ties make them easily identifiable by language users. On the other hand, unlike
formulaic expressions, whose flexibility is limited, schemata are surprisingly
versatile since they have at least one open free slot, which has to be necessarily
filled with some (creative) new lexical material. The open slot allows for various
lexical items and constitutes “the thematic crux of the utterance” (Van Lancker
Sidtis et al., 2015, p. 40). As an illustration, Van Lancker Sidtis et al. (2015, p. 40)
use the schema “the end of (the) X as we know it,” which is said to express “resigna-
tion, superior knowledge, and a bit of doom, all of which will color the meaning
of X, which can be any word or phrase.” This schema can be seen in expressions
such as the end of the world as we know it or the end of globalisation/art/Europe as
we know it. Van Lancker Sidtis et al. (2015, p. 40) make an insightful observation
about the nature of schemata: “they provide the ability to communicate highly
specialized nuances, while allowing for this meaning constellation to be applied to
very disparate phenomena – the chosen novel words.”
From a relevance-theoretic perspective, in the schema ‘X is the new Y,’ the
variable Y involves a broadened ad hoc concept based on the most salient member
of the created category which at the same time lends its name to that category.
What is communicated by utterances derived from the schema is that the most
salient member of the category (‘Y’) is replaced by another (‘X’), thus affecting the
broadened concept, for example, in Yellow is the new black this season, the most
salient member of the extended category, black (‘Y’), is replaced by another one,
here yellow (‘X’). Interestingly, it seems that the variable Y does not have to be

13. According to Allan (2009, p. 627), the snowclone X is the new black is likely to have origi-
nated from the fact that a simple elegant black dress is so popular that it has become “a must for
every woman’s wardrobe.”
 Chapter 1. Category extension as a variety of loose use 39

filled with a category extension, as in the example above. For instance, in (23), it
appears to be a metaphor.
(23) Risk is the new black in world turned upside down.14

These schematic patterns, through repeated use, may incline hearers to follow
certain well-trodden inferential paths for processing the encoded concepts and
activated assumptions, or what Vega Moreno (2007), while discussing idioms and
standardised metaphors, calls ‘pragmatic routines.’15 According to Vega Moreno
(2007, p. 3), “repeated derivation of the same sort of implications in processing a
familiar stimulus result in the development of a special type of cognitive procedure,
a pragmatic routine, for the processing of this stimulus.” In the case of snowclones,
the familiar stimulus is the schema itself, which guides the reader/hearer along
a certain inferential route. For example, the schema ‘X is the new black’ will be
typically interpreted as having a negative connotation and “refer[ring] to (passing)
hypes where ‘old stuff ’ is presented as ‘new stuff ’ but is basically ‘same old, same
old’” (Lemmens, 2017, p. 112). Therefore, even though it is impossible to deny
their productivity, such schemata turn out to be more limited than creative.

5. Conclusions

In relevance theory, category extension has been given a relatively general defini-
tion and illustrated by a handful of examples. Construed as involving broadening
of a lexically encoded concept, usually associated with a brand name, proper name
or even a common noun, category extension is viewed as part of the continuum
of non-literal (loose) uses, whose other members are approximation, hyperbole
and metaphor. It has been shown that of the four phenomena included in this
continuum, category extension has received the least attention.
In particular, while presenting the continuum of broadened uses, relevance
theorists typically exemplify and discuss three of them: approximation, hyper-
bole and metaphor, in this particular order based on the increasing extent of
broadening. The question remains where to place category extension within the

14. Example from <http://‍www.smh.com.au/business/risk-is-the-new-black-in-world-turned-

upside-down-20100803-116pe.html>

15. Pragmatic routines have been discussed or at least mentioned in a number of relevance-
theoretic studies, e.g. Unger (2011) shows how they can be applied to the analysis of procedural
markers, Ifantidou (2014) refers to them while discussing the role of cognitive routines in L2
learning, Falkum (2015) sees a connection between them and cases of systematic polysemy,
Solska (2017) suggests that they can be used for analysing conventionalized similes.
40 Ewa Wałaszewska

continuum. The examples discussed in the chapter show that category extension
can be distinguished from metaphor, though the distinction between them may be
easily blurred. Moreover, category extension may combine with approximation or
hyperbole. This may suggest a complex, non-linear, multi-dimensional structure
of the continuum.
Category extensions have been classified into two types: limited and creative
category extensions. In the chapter, an attempt has been made to extend the
range of examples and to reanalyse the available data. As shown in the chapter,
speakers may disagree about an attempted lexical modulation and even wage a
lexical warfare when a certain category extension seems unacceptable to them.
Interestingly, it seems that category extensions are more likely than metaphors
to lead to disagreements or even battles over how words are to be interpreted.
Calling a racehorse an athlete in a metaphorical way may be criticised for aesthetic
reasons; extending the category of athletes to admit racehorses may be considered
as linguistically abusive.
As regards creative category extensions, these refer to the most salient mem-
bers of the extended categories, sometimes reordering category members with
respect to their salience, importance or representativeness. Common nouns are
encountered here, but typical examples involve proper names such as those of po-
litical and military leaders. These examples bring to mind the notion of ‘paragon’ as
discussed by Lakoff, even though the two terms are not fully congruent. Not only
do some category extensions fail to invoke any paragons, but also some paragons
are used metaphorically. On the other hand, a number of typical creative category
extensions follow schematic patterns, such as ‘X is the new/next Y,’ which has been
discussed as an example of snowclones, or frequently used, sometimes overused,
schemata with some empty slots to fill. It is shown that the use of such schemata
may involve ‘pragmatic routines.’ These findings reveal the true nature of creative
category extensions, showing that their alleged creativity is actually limited.

Acknowledgements

I would like to thank Agnieszka Piskorska for her insightful comments on the earlier version
of this chapter. I am also indebted to the anonymous reviewers for their helpful suggestions for
revision.
 Chapter 1. Category extension as a variety of loose use 41

References

Allan, K. (2009). The connotations of English colour terms: Colour-based x-phemisms. Journal
of Pragmatics, 41, 626–637.  ​https://‍doi.org/10.1016/j.pragma.2008.06.004
Allott, N. (2010). Key terms in pragmatics. London: Continuum.
Blutner, R. (1998). Lexical pragmatics. Journal of Semantics, 15, 115–162. ​
https://‍doi.org/10.1093/jos/15.2.115
Blutner, R. (2011). Some perspectives on lexical pragmatics. In D. Archer & P. Grundy (Eds.),
The pragmatics reader (pp. 99–114). London: Routledge.
Brown, C. H. (1990). A survey of category types in natural language. In S. Tsohatzidis (Ed.),
Meanings and prototypes: Studies in linguistic categorization (pp. 17–47). London & New
York: Routledge.
Carston, R. (1997). Enrichment and loosening: Complementary processes in deriving the
proposition expressed? Linguistische Berichte, 8, 103–127. ​
https://‍doi.org/10.1007/978-3-663-11116-0_7
Carston, R. (2002). Thoughts and utterances: The pragmatics of explicit communication. Oxford:
Blackwell.  ​https://‍doi.org/10.1002/9780470754603
Carston, R. (2012). Metaphor and the literal/non-literal distinction. In K. Allan & K. M. Jaszczolt
(Eds.), The Cambridge handbook of pragmatics (pp. 469–492). Cambridge: Cambridge
University Press.  ​https://‍doi.org/10.1017/CBO9781139022453.025
Carston, R. (2016). Contextual adjustment of meaning. In N. Riemer (Ed.), The Routledge hand-
book of semantics (pp. 195–210). London & New York: Routledge. ​
https://‍doi.org/10.4324/9781315685533.ch11
Carston, R., & Wearing, C. (2015). Hyperbolic language and its relation to metaphor and irony.
Journal of Pragmatics, 79, 79–92.  ​https://‍doi.org/10.1016/j.pragma.2015.01.011
Davis, S. (1996). Metonymy. In T. Enos (Ed.), Encyclopedia of rhetoric and composition. Com-
munication from ancient times to the information age (pp. 444–446). New York & London:
Routledge.
Evans, V., & Green, M. (2006). Cognitive linguistics: An introduction. Edinburgh: Edinburgh
University Press.
Falkum, I. L. (2015). The how and why of polysemy: A pragmatic account. Lingua, 157, 93–99. ​
https://‍doi.org/10.1016/j.lingua.2014.11.004
Fauconnier, G., & Turner, M. (1998). Conceptual integration network. Cognitive Science, 22,
133–187.  ​https://‍doi.org/10.1207/s15516709cog2202_1
Hall, A. (2017). Lexical pragmatics, explicature and ad hoc concepts. In I. Depraetere & R.
Salkie (Eds.), Semantics and pragmatics: Drawing a line (pp. 85–100). Cham, Switzerland:
Springer.  ​https://‍doi.org/10.1007/978-3-319-32247-6_6
Horn, L. (2007). Neo-Gricean pragmatics: A Manichaean manifesto. In N. Burton-Roberts
(Ed.), Pragmatics (pp. 158–183). Houndmills, Basingstoke: Palgrave Macmillan. ​
https://‍doi.org/10.1057/978-1-349-73908-0_9
Horn, L. (2008). Pragmatics and the lexicon. In P. van Sterkenburg (Ed.), Unity and diversity of
languages (pp. 29–41). Amsterdam: John Benjamins.  ​https://‍doi.org/10.1075/z.141.05hor
Huang, Y. (2009). Neo-Gricean pragmatics and the lexicon. International Review of Pragmatics,
1, 118–153.  ​https://‍doi.org/10.1163/187731009X455866
Ifantidou, E. (2014). Pragmatic competence and relevance. Amsterdam & Philadelphia: John
Benjamins.  ​https://‍doi.org/10.1075/pbns.245
42 Ewa Wałaszewska

Jodłowiec, M., & Piskorska, A. (2015). Metonymy revisited: Towards a new relevance-theoretic
account. Intercultural Pragmatics, 12, 161–187.  ​https://‍doi.org/10.1515/ip-2015-0009
Kronenfeld, D. (1988). Full bloods and protestants: Semantic extension in complex domains.
Unpublished manuscript, University of California, Riverside.
Kronenfeld, D. (1996). Plastic glasses and church fathers: Semantic extension from the ethnosci-
ence tradition. Oxford & New York: Oxford University Press.
Lakoff, G. (1987). Women, fire, and dangerous things: What categories reveal about the mind.
Chicago: The University of Chicago Press. ​
https://‍doi.org/10.7208/chicago/9780226471013.001.0001
Lakoff, G. (1999). Cognitive models and prototype theory. In E. Margolis & S. Laurence (Eds.),
Concepts: Core readings (pp. 391–421). Cambridge, MA: MIT Press.
Langacker, R. W. (1993). Reference-point constructions. Cognitive Linguistics, 4, 1–38. ​
https://‍doi.org/10.1515/cogl.1993.4.1.1
Lemmens, M. (2017). A cognitive, usage-based view on lexical pragmatics: Response to Hall. In
I. Depraetere & R. Salkie (Eds.), Semantics and pragmatics: Drawing a line (pp. 101–114).
Cham, Switzerland: Springer.  ​https://‍doi.org/10.1007/978-3-319-32247-6_7
Liberman, M. (2005, March 27). Liberalism is the new communism. [Web log post]. http://‍itre.
cis.upenn.edu/~myl/languagelog/archives/002010.html
Ludlow, P. (2014). Living words: Meaning underdetermination and the dynamic lexicon. Oxford:
Oxford University Press.  ​https://‍doi.org/10.1093/acprof:oso/9780198712053.001.0001
Pullum, G. (2003, October 27). Phrases for lazy writers in kit form. [Web log post]. http://‍itre.
cis.upenn.edu/~myl/languagelog/archives/000061.html
Pullum, G. (2004, January 16). Snowclones: Lexicographical dating to the second. [Web log
post]. http://‍itre.cis.upenn.edu/~myl/languagelog/archives/000350.html
Rubio-Fernández, P., Wearing, C., & Carston, R. (2013). How metaphor and hyperbole differ:
An empirical investigation of the relevance-theor etic account of loose use. UCL Working
Papers in Linguistics, 25, 46–65.
Solska, A. (2017). Degrees of ‘punniness’? A relevance-theoretic account of puns and pun-
like utterances. In A. Piskorska & E. Wałaszewska (Eds.), Applications of relevance
theory: From discourse to morphemes (pp. 198–222). Newcastle upon Tyne: Cambridge
Scholars Publishing.
Sperber, D., & Wilson, D. (1995). Relevance: Communication and cognition. (2nd ed.). Oxford:
Blackwell.
Sperber, D., & Wilson, D. (2008). A deflationary account of metaphors. In R. W. Gibbs (Ed.),
The Cambridge handbook of metaphor and thought (pp. 84–105). Cambridge: Cambridge
University Press.  ​https://‍doi.org/10.1017/CBO9780511816802.007
Taylor, J. R. (1989/1995). Linguistic categorization: Prototypes in linguistic theory. (2nd ed.).
Oxford: Clarendon Press.
Traugott, E. C. (2014). Toward a constructional framework for research on language change. In
S. Hancil & E. König (Eds.), Grammaticalization – theory and data (pp. 87–105). Amster-
dam & Philadelphia: Benjamins.
Traugott, E. C., & Trousdale, G. (2013). Constructionalization and constructional changes. Oxford:
Oxford University Press.  ​https://‍doi.org/10.1093/acprof:oso/9780199679898.001.0001
Unger, C. (2011). Exploring the borderline between procedural encoding and pragmatic infer-
ence. In V. Escandell-Vidal, M. Leonetti, & A. Ahern (Eds.), Procedural meaning: Problems
and perspectives (pp. 103–127). Bingley: Emerald Group Publishing. ​
https://‍doi.org/10.1163/9780857240941_006
 Chapter 1. Category extension as a variety of loose use 43

Van Lancker Sidtis, D. (2004). When novel sentences spoken or heard for the first time in the
history of the universe are not enough: Toward a dual-process model of language. Interna-
tional Journal of Language and Communication Disorders, 39, 1–44. ​
https://‍doi.org/10.1080/13682820310001601080
Van Lancker Sidtis, D., Cameron, K., Bridges, K., & Sidtis, J. J. (2015). The formulaic schema in
the minds of two generations of native speakers. Ampersand, 2, 39–48. ​
https://‍doi.org/10.1016/j.amper.2015.02.001
Vega Moreno, R. E. (2007). Creativity and convention: The pragmatics of everyday figurative
speech. Amsterdam & Philadelphia: John Benjamins.  ​https://‍doi.org/10.1075/pbns.156
Wałaszewska, E. (2015). Relevance-theoretic lexical pragmatics: Theory and applications. New-
castle upon Tyne: Cambridge Scholars Publishing.
Wilson, D. (2003). Relevance and lexical pragmatics. Rivista di Linguistica, 12, 273–291.
Wilson, D. (2004). Relevance and lexical pragmatics. UCL Working Papers in Linguistics, 16,
343–360.
Wilson, D. (2014). Relevance theory. UCL Working Papers in Linguistics, 26, 1–20.
Wilson, D., & Carston, R. (2007). A unitary approach to lexical pragmatics: Relevance, inference
and ad hoc concepts. In N. Burton-Roberts (Ed.), Pragmatics (pp. 230–259). Houndmills,
Basingstoke: Palgrave Macmillan.  ​https://‍doi.org/10.1057/978-1-349-73908-0_12
Wilson, D., & Sperber, D. (2012). Introduction: Pragmatics. In Meaning and relevance (pp. 1–27).
Cambridge: Cambridge University Press.  ​https://‍doi.org/10.1017/CBO9781139028370
Zwicky, A. (2006, March 13). Snowclone mountain? [Web log post]. http://‍itre.cis.upenn.
edu/~myl/languagelog/archives/002924.html
 Chapter 2

Metonymic relations – from determinacy

to indeterminacy

Maria Jodłowiec and Agnieszka Piskorska

Jagiellonian University in Kraków, Poland / University of Warsaw, Poland

The chapter offers an account of metonymy and discusses some stylistic effects
which may be provided by metonymic expressions. We treat metonymically
communicated concepts as part of the inferentially established proposition of an
utterance and argue that many such concepts may be indeterminate. We posit
that the reference of metonymic expressions is assigned through the operation
of a pragmatic mechanism called contextual cognitive fix, which can be seen as
an alternative to free enrichment. In line with other relevance-theoretic works,
we see indeterminacy as an asset rather than deficit of communication, showing
that it can be a source of stylistic effects also in the case of metonymy.

Keywords: comprehension heuristic, free enrichment, inferential processes,

metonymy, mutual parallel adjustment, non-lexicalized concepts, non-
paraphrasability, semantic indeterminacy, shallow processing, vagueness

1. Introduction

As a starting point for our considerations on metonymy, we would like to recall

two fundamental assumptions underlying relevance-theoretic pragmatics: first,
that understanding any communicated message involves inference (Sperber &
Wilson, 1995) and that the vast majority of concepts communicated in utterances
depart from their literal, or encoded sense, in a variety of subtle and nuanced ways
(Sperber & Wilson, 1998a; Carston, 2002; Wilson, 2003; Wilson & Carston, 2007).
These observations have a bearing on the analysis which we would like to present
in this paper, namely, of relations between encoded concepts and their commu-
nicated metonymic readings. We believe that such relations, and ultimately the
actual meaning of metonymic concepts are established in the inferential process of
assigning a referent to an encoded concept, and that this operation is not confined

https://‍doi.org/10.1075/ftl.8.02jod
© 2020 John Benjamins Publishing Company
46 Maria Jodłowiec and Agnieszka Piskorska

to a mere replacement of one concept by another. Instead it should be analysed via

positive cognitive effects it contributes to utterance interpretation.
Generally acknowledged to be pervasive in communication, metonymy comes
in various kinds. The most widely explored parameter along which the relation-
ship between the encoded concept and its intended metonymic reading can vary
is the conventional/original continuum, which on a relevance-theoretic approach
corresponds with the contribution of inference to interpretation. At one of its
extremes, there are metonymic meanings which have become conventionalised
to the point that they are themselves encoded in the lexicon. Arguably, when
asked to give someone a ring, the hearer retrieves the phone call meaning of ring as
one of its conventionally sanctioned senses, at least when used in this or related
idiomatic expression (cf. Vega Moreno, 2007). Although the metonymic relation
between ring and phone call can still be perceived, it may in fact play little role in
the comprehension process, in which the role of inference is confined to resolving
the lexical ambiguity of ring.
The conventional character of some metonymies does not reside in the expres-
sion itself but is instead reflected in a general pattern which can be abstracted from
it, as in I love Mussorgsky, instantiating the ‘artist for work’ pattern. This is what
makes such metonymies similar to the phenomenon known as regular or system-
atic polysemy (Apresjan, 1974; Peters & Peters, 2000; Pustejovsky, 1995)1 and is
sometimes treated as conceptual mechanisms underpinning the use of metonymy.
On the inferential approach to metonymy we assume here, it can be acknowledged
that such patterns play a facilitating role in inferential comprehension (as anal-
ogy does in reasoning) but they are seen as mere regularities resulting from the
repeated usage of expressions and iterative character of human experience.
There are also metonymic expressions whose interpretations arise on the fly
exploiting a perceivable contingency of entities in the world, as in The blue vest
is a cousin of mine, whispered proudly by an opera viewer referring to a singer
on the stage. Interestingly, if contingency of this kind is more permanent, as in
the case of the attributes of one’s looks, the metonymic expression can start to
function as a nickname, such as Duckface or Bluetooth. This, once again, leads to
the conventionalisation of the metonymic relation, at least initially in the private
lexicon of a group of language users (cf. Wilson, 2015).

1. Unlike prototypical systematic polysemy (e.g., rabbit as ‘rabbit fur’, ‘rabbit meat’ or a living
animal), which in relevance-theoretic terms can be accounted for as enrichment of underspeci-
fied encoded meaning into specific contextual meaning (Falkum, 2015, cf. also Falkum, 2011
and Wałaszewska, 2008), metonymy involving regular patterns does not seem to be amenable
to such treatment: it would be implausible to posit that the encoded meaning of London is
underspecified and can be contextually enriched into ‘the city’ or ‘the government of the UK’.
 Chapter 2. Metonymic relations – from determinacy to indeterminacy 47

Another parameter, which has received much less attention than conven-
tionality/originality of the metonymic relation, is the degree of its determinacy,
depending in turn on how precise is the sense evoked by the metonymic concept.
In the examples mentioned above, ring and the blue vest point to precisely de-
lineated referents, i.e., to an act of making a phone call and a specific individual,
respectively. The intended reading of Mussorgsky, however, seems to be much
less precise, possibly including the pieces composed by Mussorgsky (at least the
ones the speaker knows), his style, or some features specific to his style, such as
its ‘unpolished’ and somewhat abrupt quality. Below we shall argue that target
concepts in metonymic relations can be communicated with varying degrees of
determinacy and that this property of metonymy can be exploited in a variety of
ways related to the basic relevance-theoretic notions of positive cognitive effects
balanced against processing costs.
Since the greatest bulk of research on metonymy has been carried out in
the cognitive linguistics paradigm, we devote the next section to an overview of
selected ideas offered within this framework. Then, in Section 3 we shall present
a relevance-theoretic account of metonymy based on the pragmatic mechanism
called contextual cognitive fix (Jodłowiec, 2015), pointing out how, in our view,
it enhances our understanding of the figure under scrutiny by allowing for the
possibility of a metonymic referent being identified more or less precisely, depend-
ing on its intended contribution to the expected cognitive gains. In Section 4, we
analyse a number of attested examples of metonymic uses, by means of which we
demonstrate how metonymy is exploited for the sake of reaching an optimally
relevant interpretation. These examples range from mundane uses to ones offering
more imaginative stylistic effects. We conclude by rounding up what we see as the
main strengths of the approach advanced in this paper.

2. Cognitive linguistics on metonymy

The literature on metonymy offered within the cognitive linguistics framework

is vast, with the most important contributions by Barcelona (2000, 2011), Brdar-
Szabo and Brdar (2011), Fauconnier (1997), Fauconnier and Turner (2002),
Kövecses and Radden (1998), Lakoff and Johnson (2003), Littlemore (2015),
Panther and Thornburg (2004, 2005), Ruiz Mendoza Ibáñez (2011), Turner and
Fauconnier (1995, 2000). Due to the limitations of space, we shall present a brief
and subjective summary of these ideas in order to show how they converge on or
diverge from the relevance-theoretic view expounded in the next section.
In Lakoff and Johnson’s (2003) formulation, metonymy is a conceptual phe-
nomenon, involving a mapping based on the relationship of contiguity between
48 Maria Jodłowiec and Agnieszka Piskorska

the source and target domain. On the conceptual level, metonymic mappings are
represented by patterns such as part for whole, author for work, object for
user, etc., and on the linguistic level they can be instantiated by such expressions
as good heads in (1):
(1) There are a lot of good heads in the university.

According to Barcelona (2000, p. 5), the link between source and target domains is
firmly rooted in human experience since when a human being interacts with a ‘part’
(author, object, etc.), they necessarily interact with ‘the whole’ (work, user, etc.).
As regards, the function of metonymy in communication, for Lakoff and
Johnson it is “primarily […] referential […], that is, it allows us to use one entity
to stand for another” (2003, p. 36), yet it often highlights relevant features of the
target concept, as good heads in (1) does by drawing attention to the university
staff ’s intellectual abilities. Brdar-Szabo and Brdar underscore that metonymy
affords a way of “saying two things at the price of one” (2011, p. 236), in this way
contributing to the economy of expression.
The nature of the metonymic mapping, i.e. a relationship between the source
and target concept has received various treatments. Kövecses and Radden (1998)
suggested that it should be defined in terms of providing mental access to one entity
(target) via another (source), whereas Croft (1993) saw this process as highlighting
a domain within a complex domain matrix. Other researchers, e.g. Brdar-Szabo
and Brdar (2011) and Ruiz Mendoza Ibáñez (2011), claimed that mapping should
be seen as a domain expansion or reduction, depending on whether the source
contains the target concept or the other way around. Barcelona (2011) opted for
treating the metonymic mapping as a way of imposing a perspective (cf. Langacker
1993) from which the target concept is viewed.
Some space has been devoted as well to delineating the boundaries of the
phenomenon. Despite the consensus that metonymy is a category structured
around prototypical examples, shading off into other zone-activation phenomena
(Barcelona, 2011), there are divergent views on what should be included in the
set of prototypical features. According to Panther and Thornburg (2005, p. 42),
one such feature is “the degree of conceptual prominence of the target concept,”
whereas for Barcelona (2011) it is referential character and the distinctness of the
target domain from the source domain. The boundaries of metonymy understood
as a referring device are significantly extended in the works by Panther and
Thornburg (2004, 2005) and Radden et al. (2007), who posit that various kinds
of inferences are metonymic in nature (e.g. an explicitly expressed statement of a
wish can stand metonymically for the speech act of making a request).
We consider many of the above-presented findings compelling – especially the
idea that human experience provides a key to understanding metonymy, or that
 Chapter 2. Metonymic relations – from determinacy to indeterminacy 49

the source concept can be seen as a point of access to the target concept. Much as
we agree with the observation that all inferences are metonymic in a broadly con-
ceived sense (because the encoded meaning stands for the communicated mean-
ing), we believe that this approach to metonymy obfuscates its nature. To shed
light on it, we propose that only prototypical cases of referential metonymy should
be taken into account and that the fact to be explained is how inference plays a
role in metonymy comprehension rather than how metonymy explains inference.

3. A relevance-theoretic treatment of metonymy involving contextual

cognitive fix

To start our considerations on the role of inference in comprehension, let us

note that it goes far beyond constructing implicatures of an utterance (Sperber &
Wilson, 1995; Recanati, 1993; Carston, 2002; Wilson, 2003). The idea is empirically
grounded in the observation that many inferences treated on the Gricean account
as implicatures make a contribution to the truth-conditions of the main proposi-
tion expressed by an utterance, i.e., to what is commonly referred to as the expli-
cature (Carston, 2002; Sperber & Wilson, 1995; Wilson & Sperber, 2004; Recanati,
2004) or impliciture (Bach, 1994, 2010), and it is therefore counterintuitive and
theoretically unsound to treat them as implicatures. This can be illustrated by (2)
(2) If someone leaves a manhole uncovered and you break your leg, you can sue
 (Wilson & Sperber, 1998b, p. 3)

On Grice’s view, the sequential reading of and as ‘and then’ is due to a particular-
ized conversational implicature based on the maxim of manner. If this reasoning
is extended to the causal interpretation of and as ‘and because of that’, the assump-
tion that one can sue if they have broken a leg falling into an uncovered manhole
would be merely implicated and therefore cancellable. Similarly, the metaphoric
reading of a sniffer dog in (3) would be seen as a quality-based implicature, with no
explicit assumption communicated at all. On the relevance-theoretic treatment,
such inferences as those in (2) and (3) are treated as a pragmatic contribution
to the explicitly communicated proposition. Since this treatment of inferential
processes now holds as standard in relevance theory, we take it as a departure
point for our account of metonymy, which we also aim to explain in terms of an
inferential contribution to the explicature of an utterance.
(3) Be careful with her, she is a sniffer dog

Besides, our proposal draws on the observations put forward by Sperber & Wilson
(1998a), namely, that the mental lexicon is richer and more complex than the
50 Maria Jodłowiec and Agnieszka Piskorska

public lexicon: there are numerous stable mental concepts which do not map
directly onto words. Yet, as Sperber & Wilson (1998a, 2008) claim, such concepts
are effable, since speakers can apply various strategies to communicate them by
means of existing words. We believe that on a general level, using metonymy is an
example of such a strategy aimed at communicating a concept represented in the
language of thought but either not having a counterpart in a natural language or
not having it available at a reasonable processing cost. This can be illustrated by the
concept referred to by Buckingham Palace below:
(4) Buckingham Palace has been forced to defend Prince Andrew over sex abuse
allegations.  http://‍rt.com/uk/219835-prince-andrew-scandal-defence/

The frequent use of Buckingham Palace and similar expressions with metonymic
readings, which receive fairly similar interpretations across contexts, suggests that
such concepts are fairly stable. Yet, they are non-lexicalised, since their mean-
ings cannot be expressed by means of single lexical items or compact phrases. In
Jodłowiec and Piskorska (2015) we proposed to represent such concepts as mental
entities tagged by the metonymic expression, which in the case of the concept
communicated by Buckingham Palace in (4) would be
(5) [ENTITY A]Buckingham Palace

The notation itself is arbitrary but it is intended to capture the following features
of the concept in question: in the language of thought there is a conceptualisation
of ENTITY A, being the actual referent of the metonymic phrase Buckingham
Palace, linked to its encoded meaning by a contiguity relationship. The question
that naturally arises at this point is what or who exactly ENTITY A is. There is no
one univocally accepted answer to this question. As we have mentioned before, we
discard the view that there will invariably be a specific ENTITY which will replace
the concept encoded by the metonymic expression; we believe that the degree to
which the communicated concept can be specifically individuated varies from
utterance to utterance and is conditioned by the need to optimise relevance. For
instance, the main relevance of (4) above can be assumed to lie in communicating
the implicature that the risk of Prince Andrew’s reputation being damaged was
judged as serious (knowing that Buckingham Palace’s customary standpoint on
personal affairs of the royal family members is ‘no comment’). Note that this as-
sumption will be easily recoverable even if the concept communicated as ENITITY
A remains highly underspecified – an average communicator and an average
addressee need not entertain a conceptualisation which would include specific
information on the possibly complex network of people and institutions behind
issuing statements on behalf of the British monarch. Constructing such a specific
conceptualisation would not contribute any additional inferences increasing the
 Chapter 2. Metonymic relations – from determinacy to indeterminacy 51

overall relevance of (4), therefore, a recipient of newspaper reports will have his
expectations of relevance satisfied with the referent of Buckingham Palace repre-
sented schematically as some ENTITY contiguous with the encoded meaning of
the phrase Buckingham Palace, i.e., the official residence of the British monarch.
In suggesting the interpretation of utterance (4) we have referred to ‘an average
recipient’ for whom such an interpretation based on entertaining an underspeci-
fied concept would be plausible. Needless to say, there are individuals who may
entertain a much more precisely delineated conceptualisation of Buckingham
Palace – one would expect, for instance, the British Prime Minister to have one.
The core of our argument presented above is that in principle, metonymic expres-
sions can refer to mental concepts whose content may not be fully determinate
and will remain schematic.2 Such concepts can be fairly stable, yet not lexicalised,
as explained in more detail below.
Consider Example (6) illustrating a different use of the same metonymic ex-
pression, uttered by a waiter in a café, with the phrase Buckingham Palace referring
to a customer wearing a T-shirt with a picture of Buckingham Palace:
(6) Buckingham Palace wants a cappuccino.

Here the phrase Buckingham Palace is used as an individuating expression point-

ing to a specific person. It is clear that for this utterance to achieve relevance, the
addressee has to identify the referent as the individual the speaker has in mind.
Yet, although the referent in the real world is precisely individuated, and hence the
conceptualisation is in this sense more precise than the one corresponding to the
phrase Buckingham Palace in (4), in other respects it also remains underspecified.
It is so because, again, neither the speaker nor the addressee would gain any cogni-
tive benefits from entertaining a conceptualisation of the referent which would
be rich in detail. Thus, if the mental representation of the referent in question is
something like (7) below it does not have to include any information about the
referent beyond what is necessary to identify them. It does not therefore appear
necessary to conceptualise (7) as ‘A person/man/young lad/elderly gentleman
etc. wearing a Buckingham Palace T-shirt’. Thus, much as the usual practice of
paraphrasing metonymic expressions by replacing them by their descriptive near-
equivalents (e.g. Ham sandwich by ‘Ham sandwich orderer/customer’ or ‘a lady/
gentleman who ordered a ham sandwich’) is justifiable for expository purposes,

2. Underspecification of concepts referred to here occurs at the level of communicated import

(i.e. the level of the language of thought), so it is distinct from pro-concepts, which are taken
to be typically encoded by certain words (e.g. pronouns, gradable adjectives, etc.), and whose
semantic incompleteness (schematicity) is an essential property of their lexical content (cf.
Sperber & Wilson, 1998a; Wilson, 2011).
52 Maria Jodłowiec and Agnieszka Piskorska

it misses out on a very important point why we use metonymy at all: we do so

to refer to concepts which may be schematic rather than fully determinate, so a
paraphrase including a characterisation of the referent is in fact inadequate. The
observation that figurative meanings are non-paraphrasable seems to have been
more appreciated with reference to metaphor than metonymy, to which, we be-
lieve, it also applies with full strength.
(7) [ENTITY B]Buckingham Palace

Following the line of argumentation that one of us has recently presented to pos-
tulate an inferential procedure called “contextual cognitive fix” (Jodłowiec, 2015,
Chapter 2), we see the inferential step (or steps) that render the metonymic expres-
sion interpretable in the way described above as a process of the cognitive fix type.
The rationale for positing this pragmatic mechanism (Jodłowiec, 2015)3 stems
from the observation that the procedure of free enrichment, standardly called
upon in relevance theory (Carston, 1998, 2002, 2004) to account for the recovery
the explicitly communicated speaker-intended contextual meaning by adding
some conceptual content to a minimally truth-evaluable proposition, appears not
to represent adequately what is going on in the hearer’s mind. For illustration,
consider (8), uttered by Mary to Peter as they are about to go out for dinner
(8) I am ready.

Free enrichment, even if constrained by considerations of optimal relevance, com-

pels the hearer to construct a specific and enhanced conceptualisation of ready
in the course of interpreting (8). Such a conceptualisation can be ‘ready to leave
the house’, ‘ready to go out for dinner’ or yet another thought of similar content.
In real communication, though, neither the speaker nor the hearer need, or are
in fact likely to entertain such a specifically enriched proposition. Most probably,
the main relevance of (8) will lie in an implicature, such as ‘Peter should put on
his coat immediately’, and to arrive at this implicature it does not seem necessary
for Peter to enrich ready in any of the ways enumerated above. This argument
extends to a host of other examples discussed in the literature, such as It is raining,
which, in contrast to what is often argued by proponents of free enrichment, need
not include any information on the precisely identified location of the rain, at
least not for a hearer who has to decide whether to take an umbrella with him or
not. In contradistinction to free enrichment, which is quantitative, cognitive fix is
qualitative in nature (Jodłowiec, 2015, p. 69). It enables the speaker and hearer to
identify the relevant denotation of the concept used in the explicature, at the same

3. Arguments against free enrichment can also be found, for instance, in Borg (2016) Buchanan
(2010), Corazza & Dokic (2012), and Viscente & Martínez-Manrique (2005).
 Chapter 2. Metonymic relations – from determinacy to indeterminacy 53

time allowing for the possibility that such a concept may be underspecified. As we
have demonstrated, the nature of the inferential process identifying the referents
of metonymic expression is such that it appears to be an instantiation of the work-
ings of cognitive fix. On the other hand, the frequently attested underspecification
of metonymic readings is not restricted to this type of language use only, since
pragmatic indeterminacy is of common occurrence in human communication (cf.
Sperber & Wilson, 2015).
Returning for a while to Examples (4)–(6), let us note that apart from the de-
gree and kind of indeterminacy, there are two more important differences between
(5) and (7), which represent the metonymic subjects of (4) and (6), respectively:
the mental concept in (5) seems to be, as has been mentioned, fairly stable, whereas
the one in (7) is genuinely one-off. At the same time, the use of Buckingham Palace
in (4) is typical, as is generally the practice of referring to institutions by the names
of locations or buildings where they reside, whereas the one in (6) is much less
sanctioned. It seems plausible to assume that the productivity of some metonyms
and their susceptibility to conventionalisation has to do with the stability of con-
cepts they represent: if a non-lexicalised mental concept appears to be stable, it
is likely to be verbalised by a metonymic expression, which is also accepted as
natural. For example, people may have a fairly stable, yet non-lexicalised concepts
which can be approximately described as ‘whoever issues statements on behalf of
the British monarch’, or ‘whoever is responsible for making decisions on behalf
of the state’, which can be expressed metonymically as Buckingham Palace and
London (in the case of the UK), respectively. On the other hand, people are un-
likely to have a stable mental representation of ‘a person wearing a (certain) outfit
(with a certain pattern on it)’, so referring to a person as a blue vest or Buckingham
Palace can be perceived as somewhat less natural, or requiring a special context
for interpretation. As pointed out by Wilson (2015), metonymy has a potential to
sort objects or people and the sorting criterion is the property explicitly invoked
by the metonymic expression: doctors sort patients by their diseases, waiters sort
customers by their orders, etc. It is only natural that doctors have stable repre-
sentations of patients with a certain disease, waiters of customers ordering some
dishes, conductors of orchestra members playing a certain instrument, etc. Other
sorting criteria, such as sorting roommates by potentially useful objects they have
(irons, coffee machines – cf. Wilson, 2015) are also possible but since concepts
such as ‘roommate owning a household implement’ are unlikely to be stable, even
in room renting circles, metonymies based on them are, again, felt to be somewhat
unusual. If this argument is right, then the explanation why many metonymies
correspond to established patterns such as ‘disease for patient’, ‘author for work’
etc., lies in the fact that such patterns owe their existence to stable unlexicalised
concepts of the ‘patient with a (certain) disease’, ‘work by a (certain) author’ type.
54 Maria Jodłowiec and Agnieszka Piskorska

Unlike cognitive linguists then, we see those patterns not as underlying schemas
that give rise to metonymic uses (e.g. Lakoff & Johnson, 2003; Kövecses & Radden,
1998), but on the contrary, as generalisations resulting from the existence of stable
unlexicalised concepts, as postulated in relevance theory on independent grounds
(Sperber & Wilson, 1998a). At the same time, we shed some light on the vexing
question of why, although in principle it is possible to exploit every contiguity
relation as a basis for metonymy, some metonymies are felt to be more natural: it is
much more plausible to use a saxophone to refer to a saxophonist than to a person
carrying a saxophone. This is so, we believe, since the concept ‘a person playing
an instrument’ is likely to be stable in a mental repertoire of a speaker, whereas a
concept ‘a person carrying a musical instrument’ is not.

4. Metonymy and cognitive effects

As indicated above, the account of metonymy expounded here is rooted in the

assumption that the concept communicated by a metonymic expression, while
sometimes fully determinate, may remain underspecified, or even indeterminate,
and that this indeterminacy does not undermine the success of communication,
but in fact may be instrumental in making verbal communication highly effective.
The key idea underpinning this approach is that the major goal of commu-
nication has to do with cognitive effects: the speaker’s aim in saying something
is to produce certain cognitive effects in the hearer, and, reciprocally, the hearer’s
aim is to recover the cognitive effects as intended by the speaker. Following the
relevance-theoretic model of utterance comprehension, we assume that human
cognition is programmed to maximise relevance, and human communication,
being a cognitive activity, automatically seeks to optimize relevance. In simple
(and somewhat simplified) terms, this means that the hearer’s mind automatically
seeks to get a satisfying return for the effort expended, ensuring that the effects
arrived at are congruent with the manifest intentions of the speaker, as only these
count as communicated.
Given that a verbal (or non-verbal, as the case may be) stimulus is just a clue
provided by the communicator for the addressee to recover the intended message,
relevance theory views the comprehension process as inferential through and
through, with the interpreter working out what is being communicated by fol-
lowing the shortest route to the communicator-intended cognitive effects. This is
modelled within this pragmatic framework as the relevance-oriented comprehen-
sion heuristic, in accordance with which the recipient is taken to
 Chapter 2. Metonymic relations – from determinacy to indeterminacy 55

a. Follow a path of least effort in constructing an interpretation of the utterance

(and in particular in resolving ambiguities and referential indeterminacies,
in going beyond linguistic meaning, in supplying contextual assumptions,
computing implicatures, etc.); [and to]
b. Stop when [his] expectations of relevance are satisfied. (Sperber & Wilson,
2005, p. 360)
Thus it is posited that utterance interpretation, both at the level of fine-tuning the
meaning of its components to get the explicitly communicated content and at the
level of accessing the intended contextual implications and drawing conclusions
to recover its implicit import, involves the addressee taking the easiest route to ob-
taining a satisfactory range of cognitive effects (Wilson & Sperber, 2004). All these
interlocking steps, whether local, for instance concerning lexical adjustments,
or more global, like accessing relevant background assumptions and deriving
implicatures, are inferential and run in parallel. This means that the hypotheses
about the sense of the different components of the utterance being processed and
those about the overall interpretation are being developed simultaneously and
their instant verification in view of seeking an adequate cognitive return leads to
appropriate amendments as the on-line interpretation unfolds. On this approach
then, verbal comprehension is by definition a process of mutual parallel adjust-
ment “in which tentative hypotheses about contextual assumptions, explicatures
and contextual implications are incrementally modified so as to yield an overall
interpretation which satisfies the hearer’s expectations of relevance” (Wilson &
Carston 2007, p. 248).4
We believe it is crucial to emphasize at this juncture that communication in
general, and utterance production in particular, is essentially about achieving a
certain cognitive outcome in the hearer. In other words, it is not just the ultimate
but the primary goal of the interpreter to get the pragmatic effect that the speaker
intends him to achieve by saying what she does (cf. Gibbs & Colston, 2012, p. 6).
This means that utterance comprehension should be thought of mainly as geared
to affecting the recipient’s cognitive state. As Sperber & Wilson, (2015, p. 140)
aptly put it, “[w]hat people do when they communicate is precisely to overtly
reveal something of their own mind in order to bring about such changes of mind
in their audience.” To be more specific, the speaker’s utterance is triggered by a cer-
tain thought she entertains and wants to share with her audience, and her purpose
in producing a verbal stimulus is to create a similar thought in the recipient. Her
communicative act will be successful if the speaker’s original thought and the one

4. For more details on mutual parallel adjustment, see e.g., Carston (2002), Noveck & Sperber
(2007), Sperber & Wilson (1998a, 2005, 2008), Wilson & Sperber (2002, 2004).
56 Maria Jodłowiec and Agnieszka Piskorska

surfacing in the interlocutor’s mind as a result of the utterance processing contrib-

ute their shared implications to the cognitive environments of the communicator
and the audience. It is the similarity of the mental states of the speaker and the
hearer then rather than thought duplication that is emphasized on this construal.5
This kind of approach provides some interesting insights into how metonymic
expressions function in communication. First, as indicated above, since the mean-
ing that a given expression conveys in context is not encoded by this expression,
but the expression provides evidence about the intended meaning which the
interpreter needs to infer (Sperber & Wilson, 2008, p. 86), how its sense is settled
depends on the cognitive effects manifestly intended by the communicator and on
satisfying the interpreter’s expectations. This suggests that the metonymic expres-
sion provides a cue to getting the speaker-intended effects and its interpretation
will depend directly on what these effects are and what the shortest route to reach-
ing them is, which has important consequences for how it is processed.
What this means in practical terms can best be demonstrated through ex-
amples, so let us turn to these now. We have chosen some examples featuring one
particular metonymic expression, namely Stella Artois, which as will be shown,
will have different interpretations depending on the pragmatic effects the utter-
ance it features in is designed to give rise to. These interpretations do not only vary
in denotations that are picked out in different communicative contexts, which is
typical of metonymical polysemy (Sweep, 2012), but, as will be emphasized below,
may give rise to schematic, not fully determinate interpretations whenever these
will be sufficient to get the manifestly intended and satisfying range of cognitive
effects. Consider (9)–(12):
(9) It was nine o’clock at work and for the last half hour a Greyhound and a
Stella Artois had been dishing about men. 
 (My Old Man by Amy Sohn, p. 106)
(10) Stella Artois has been using mobile to send messages to consumers about its
Cider product on days of key marketing activity.
 (https://‍www.marketingweek.com/2014/08/20/
mobile-research-under-threat-as-consumers-go-dark)

5. This idea is elaborated on in Sperber & Wilson (1995) and Carston (2002, p. 47), who under-
scores that “the thought(s) that the speaker seeks to communicate are seldom, if ever, perfectly
replicated in the mind of the audience; communication is deemed successful (that is, good
enough) when the interpretation derived by the addressee sufficiently resembles the thoughts
the speaker intended to communicate”.
 Chapter 2. Metonymic relations – from determinacy to indeterminacy 57

(11) Drinkers had switched from bottles to multi-pack cans at lower-price points,
hitting Stella Artois and Beck’s in particular.
 (http://‍www.foodcareers.net/news/ab-inbev-
admits-stella-artois-losing-market-share-96.htm)
(12) Stella Artois is licensed to third parties in many countries, like Australia,
Bulgaria, Croatia, Hungary, Romania and many others.
 (adapted from http://‍www.ab-inbev.com/content/dam/universaltemplate/
ab-inbev/investors/reports-and-filings/sec-filings/SECfilings20-F1004.pdf)

What the phrase Stella Artois conveys in each of the examples appears markedly
different. Being part of a story in which the narrator, the bartender named Roxy,
reflects on what happened in the bar one day, in (9) a Greyhound and a Stella
Artois refer to two female customers drinking Greyhound and Stella Artois beer. It
is much less obvious what Stella Artois stands for in (10) and (11). While it can be
assumed that it is the advertising bureau that is responsible for sending messages
to prospective buyers, and that customers’ preference for lower price products
negatively affects financial profits of the producer, it would be a misconception
to suggest that processing (10) automatically leads to interpreting Stella Artois as
‘Stella Artois advertising bureau’, and that in (11) Stella Artois should be fleshed
out as ‘Stella Artois financial profits’. An individual well acquainted with how the
company is organized may know that there is a marketing department in Stella
Artois and will spontaneously read this meaning into the metonymic expression,
but most readers will not find it consequential to identify precisely which section in
the company or who in fact is responsible for targeting mobile phone users in the
promotion campaign. It is not difficult to imagine a situation in which an alarmed
parent might be commenting on the fact that their child was sent a text about cider
products by Stella Artois: the reflection that it is appalling that alcoholic beverages
are advertised via mobile devices comes to the fore and is easily inferable without
pinning down exactly who is accountable for what is going on. In a similar vein,
the message that the two large beer producers experienced a decline in profit,
recovered as the meaning communicated by (11), does not require the reader to
seek a precise sense of the expression Stella Artois.
The idea that successful verbal comprehension may be rather shallow and not
move beyond a schematic representation of a metonymic expression used by the
speaker is probably best illustrated by an utterance very similar to the attested (12).
Let us assume that in a beer-drinking situation in which a person from Belgium
has just remarked that trying to be patriotic, he always orders Stella Artois, to
which his female companion responds with: Mind you, Stella Artois is licensed
to third parties in many countries, for instance, in Austria and Hungary. In this
context, the main import of the utterance under scrutiny is the implicature that
58 Maria Jodłowiec and Agnieszka Piskorska

what the patriotic beer-drinker thinks of as Belgian beer may not necessarily have
been brewed in Belgium. When processing the utterance the addressee will most
probably not bother to represent Stella Artois as a fully-fledged conceptualization
‘brewing of Stella Artois beer’, because the intended implicature can be generated
without that. In effect, the representation of Stella Artois remains quite schematic,
which – in accordance with what has been postulated above – is captured by the
language of thought formula [Entity X]Stella Artois. The metonymic expression is the
tag on the intended entity, providing indexing that gives access to a potentially
fuller conceptualization, should one be required, or simply to a range of inferences
crucial in working out the optimally relevant cognitive effects.
More often than not what metonymies communicate is quite vague, as the
short passage below exemplifies: it causes no comprehension problems for the
reader, but proves a challenge at the level of identifying accurately the concepts
linguistically encoded by the expression Stella Artois:
(13) [T]he Home Office also plans to amend the Sexual Offences Act of 2003 so
that a drunk girl can’t say yes. Though she swoons as she staggers, and begs
a man to stay the night, if a woman is utterly pie-eyed, the new law will say
her consent to sex means nothing. And if she then takes the man to court,
she’ll win. The man can swear she wanted him, produce friends to testify to
her drunken attempts to unbutton his shirt, but if she claims to have been
blotto, he’ll be banged up for rape. (…) Why this peculiar disparity between
how the genders are judged? If a woman, when drunk, isn’t responsible for
her actions, then why should a man be? If Stella Artois can force a girl to
assent to sex against her will, then why can’t a man claim it was the Stella
that removed his trousers too?
 (Mary Wakefield, “If she’s blotto, he’s a rapist. How absurd.” The Independent,
13.03.2006; (http://‍www.telegraph.co.uk/comment/personal-view/3623606/
If-shes-blotto-hes-a-rapist.-How-absurd.html)

The comprehension of the last sentence poses no problem, even though it is de-
batable what Stella Artois actually communicates in this context. Unlike in the
previous examples in which we were able to explicate the denotation of the refer-
ent as ‘advertising bureau’, ‘profits’ and ‘act of brewing’, respectively, even though
such specified denotations are unlikely to figure in the explicatures entertained by
hearers, the meaning of Stella Artois in (13) is much vaguer. The idea that shallow
processing is relied on explains why this is so. The underlying claim is that verbal
comprehension does not always involve the recovery of a fully-fledged, truth-
evaluable explicature: a less than complete explicature will do when the intended
cognitive effects can be achieved on the basis of a schematic representation of
certain components of the utterance, the case in point being here metonymic
 Chapter 2. Metonymic relations – from determinacy to indeterminacy 59

expressions. This kind of approach finds support in abundant empirical evidence

which reveals that comprehenders often content themselves with good enough
interpretations (Ferreira et al., 2002). There have been a number of experiments
revealing that people frequently take shortcuts and do not engage in full syntactic
or semantic analyses of verbal inputs they are processing (cf., among others, Barton
& Sanford, 1993; Bredart & Modolo, 1988; Ferreira et al., 2002; Natsopoulos, 1985;
Park & Reder, 2004; Reder & Kusbit, 1991; Sanford, 2002; Sanford & Graesser,
2006; Van Oostendorp & de Mul, 1990). This happens because, as hinted at above,
the true aim of the human comprehension system is not to represent the speaker’s
utterance in great detail and accuracy, but to form representations that are most
suitable for the hearer’s current purposes and are compatible with the speaker’s
intentions (Ferreira & Patson, 2007).6 The approach in which metonymic expres-
sions are treated as conceptual blueprints, which may and often do remain partly
vague, captures this.
This kind of construal is close in spirit to the raison d’être of relevance theory,
focused on elucidating the extraordinary efficiency of human verbal communica-
tion vis-à-vis the complexity of the whole process. It appears reasonable to assume
that whenever the speaker-intended, optimally relevant cognitive effects can be
reached without working out the fully truth-evaluable explicature, the interpre-
tation will be short-circuited in this way. In other words, the comprehension
module of the hearer following the path of least effort will deliver a schematic
explicature as long as it ensures a return of a contextually satisfying interpretation.
Shallow processing of this kind is frequently followed in the case of metonymic
expressions, which as argued in this paper, provide access to mental schemata
to be left unelaborated when no effects beyond those guaranteed by a schematic
representation at the level of explicature are to be sought. Thus, more often than
not metonymies function as conceptual blueprints, whose unequivocal conceptual
content may be hard to identify, and whose function in verbal communication
may have to do with setting the interpreter’s mind on the relevant inferential track.
The major part of the argument advanced above has concentrated on the
comprehension side of the communication coin, so it seems in place to add
a remark or two on the speaker’s perspective. The speaker will use a metonym
when she conceptualizes a certain entity in terms of a relation this entity bears
to the concept literally denoted by the metonymic expression, provided that she
can trust the recipient to easily recover the intended meaning by establishing the
relation between the two. In practical terms, a metonym provides the basis for
inferring the kind of meaning the speaker wants to convey, at the same time being
a shortcut to achieving the desired cognitive effects. That is why we can refer to

6. Cf. also Maitra’s (2007) remarks on partially successful communication.

60 Maria Jodłowiec and Agnieszka Piskorska

total strangers and expect that our interlocutor will easily identify who we mean,
as illustrated by The Gucci handbag used to be my student. It may also happen
that the speaker herself has a vague idea of who (or what) exactly is involved in
a certain action but still wants to communicate something important about this
action or its consequences, as exemplified by (14), in which it is quite vague what
the Academia actually stands for:
(14) The Academia does not really encourage interdisciplinary research.

Metonymy mirrors the way speakers represent the reality around and provides
a handy tool to be used to talk about things that they sometimes have only a
vague notion about. This indicates that incompleteness of meaning may have a
role to play not only at the level of utterance interpretation, but also at the level
of verbal production.
There is one more attribute of metonymy reflected in the examples discussed
above: apart from providing a useful and economical means of pointing to entities
and affording access to assumptions crucial for comprehension, metonymy creates
certain stylistic effects. These can probably hardly be noticed when highly conven-
tionalized metonymic patterns are employed, but are vivid in less standard uses,
in which a kind of wordplay may be traced. Pinker (2007, p. 119) aptly observes
that “[w]hen a waitress refers to a customer as a sandwich, she is not just saving
breath. She is exercising a dry wit, reducing the patron she otherwise fawns over
into the undignified commodity that is their only real common concern.” This
kind of effect can be detected in a number of examples discussed here, but seems
particularly striking in (13). The potency of the closing rhetorical question: “If
Stella Artois can force a girl to assent to sex against her will, then why can’t a man
claim it was the Stella that removed his trousers too?” resides in the strong imagery
created by anthropomorphising Stella Artois, which reinforces the communicative
impact of what is verbally conveyed.
Since, as indicated above, metonymic expressions are naturally polysemous,
they easily lend themselves to punning. Let us stipulate that (15) below is uttered
in the context in which John is employed in the Stella Artois company as an ac-
countant, and having to do a lot of overtime, recently he has taken to drinking
heavily, which is mutually manifest to the speaker and the hearer.
(15) Stella Artois is killing John.

In this communicative situation, a punning effect is brought forth: both meanings

of Stella Artois, i.e, ‘the Stella Artois company’ and ‘excessive consumption of Stella
Artois beer’, may be taken as the speaker-intended optimally relevant interpreta-
tion. In this way a peculiar cognitive upshot is accomplished: the recipient is left
with two competing interpretations, and unable to choose between them, he is
 Chapter 2. Metonymic relations – from determinacy to indeterminacy 61

left with the double entendre. This demonstrates well the very nature of puns,
being “the product of a context deliberately constructed to enforce an ambiguity,
to render impossible the choice between meanings, to leave the reader or hearer
endlessly oscillating in semantic space” (Attridge, 1988, p. 141; original emphasis).

5. Conclusion

Unlike in some earlier approaches to metonymy within relevance theory (Falkum,

2011; Papafragou, 1996), on the relevance-theoretic model presented in this paper,
understanding metonymy is treated as a properly inferential process, in which the
hearer infers the speaker-intended meaning by following the relevance-oriented
comprehension heuristic. It is argued that the recovered meaning of metonymic
expressions may not necessarily surface as full-fledged complete thoughts, but
may range from fully-determinate and unequivocally designated meanings to
incomplete mental representations in which certain constituents will be only
schematically rendered: what is decisive in this respect is how a satisfying range of
cognitive effects is obtained. An important strength of this kind of analysis is that
it accounts for vaguer aspects of communication, which while commonly ideal-
ized away by many pragmaticists, are by no means marginal in verbal exchanges
and should be given due attention (Wilson & Sperber, 2012, p. 16).
As we have demonstrated, such vague concepts can also be communicated
by metonymic expressions. Many of them, such as (12) “Stella Artois is licensed
to third parties” and similar are not themselves a source of rich pragmatic effects.
Instead, the indeterminacy of the referent concept plays a role in utterance com-
prehension as an effort saving factor, which facilitates the derivation of relevant
implicatures. On other occasions, such as (13) “Stella Artois can force a girl…” it
seems to be more directly responsible for the stylistic effects of the interpretation,
in this case the personified representation of beer which becomes an active agent
manipulating human actions. Besides, there exist metonymies used especially for
their punning effect resulting from the activation of two different, possibly quite
determinate readings, jointly constituting a speaker-intended interpretation like
in other instances of deliberately exploited ambiguity. In any case, the metonymic
mechanism does not involve mere replacing of one concept by another.
While the main point of this chapter was to argue for the model of metonymy
in which the linguistically encoded concept is often assigned an indeterminate
reading by the contextual cognitive fix mechanism, and to highlight that inde-
terminacy can be a means rather than an obstacle to successful communication
(cf. Sperber & Wilson, 2015, p. 141), we also addressed in passing the widespread
intuition that some metonymies are plausible and catch on easily, whereas others
62 Maria Jodłowiec and Agnieszka Piskorska

are less likely to be productive. The explanation we offered is that communicators

use metonymic expressions to refer to concepts that they already have in their
mental lexicons but for which there are no compact expressions in the public
lexicon. This idea is supported by the fact that many metonymies are confined to
professional jargons (as ham sandwich to waiters’ jargon) because the existence of
such unlexicalised concepts is motivated by their frequent use in the conversations
related to professional topics.
Applying the apparatus of relevance theory to metonymy shows that this
pragmatic framework is very productive and affords insights into a number of dif-
ferent linguistic phenomena. What is more, a wide range of themes pertaining to
verbal production and comprehension find a viable explanation when scrutinized
through the lens of relevance and are described in “a conceptually unified way”
(Sperber & Wilson, 2015, p. 138).

References

Apresjan, J. D. (1974). Regular polysemy. Linguistics, 14 (2), 5–32.

Attridge, D. (1988). Unpacking the portmanteau, or who’s afraid of Finnegans Wake? In J. Culler
(Ed.), On puns: The foundation of letters (pp.140–155). Oxford: Blackwell.
Bach, K. (1994). Conversational impliciture. Mind & Language, 9, 124–162. ​
https://‍doi.org/10.1111/j.1468-0017.1994.tb00220.x
Bach, K. (2010). Impliciture vs. explicature: what’s the difference? In E. Romero, & B. Soria
(Eds.) Explicit communication: Robyn Carston’s pragmatics (pp. 126–137). Basingstoke:
Palgrave–Macmillan.  ​https://‍doi.org/10.1057/9780230292352_8
Barcelona, A. (2000). Introduction: The cognitive theory of metaphor and metonymy. In A. Bar-
celona (Ed.), Metaphor and metonymy at the crossroads. A cognitive perspective (pp.1–28).
Berlin: Mouton de Gruyter.
Barcelona, A. (2011). Reviewing the properties and prototype structure of metonymy. In R.
Benczes, A. Barcelona, & F. J. Ruiz de Mendoza Ibáñez (Eds.), Defining metonymy in cogni-
tive linguistics: Towards a consensus view (pp.7–58). Amsterdam: John Benjamins. ​
https://‍doi.org/10.1075/hcp.28.02bar
Barton S., & Sanford A. (1993). A case study of anomaly detection: Shallow semantic processing
and cohesion establishment. Memory & Cognition, 21(4), 477–487. ​
https://‍doi.org/10.3758/BF03197179
Borg, E. (2016). Exploding explicatures. Mind & Language, 31(3), 335–355. ​
https://‍doi.org/10.1111/mila.12109.
Brdar-Szabo, R., & Brdar, M. (2011). What do metonymic chains reveal about the nature of
metonymy. In R. Benczes, A. Barcelona, & F. J. Ruiz de Mendoza Ibáñez (Eds.), Defining
metonymy in cognitive linguistics: Towards a consensus view (217–248). Amsterdam: John
Benjamins.  ​https://‍doi.org/10.1075/hcp.28.12brd
Bredart S., & Modolo, K. (1988). Moses strikes again: Focalization effect on a semantic illusion.
Acta Psychologica, 67, 135–144.  ​https://‍doi.org/10.1016/0001-6918(88)90009-1
 Chapter 2. Metonymic relations – from determinacy to indeterminacy 63

Buchanan, R. (2010). A puzzle about meaning and communication. Noûs, 44(2), 340–371. ​
https://‍doi.org/10.1111/j.1468-0068.2010.00743.x
Carston, R. (1998). Postcript. In A. Kasher (Ed.), Pragmatics. Critical concepts (pp. 464–477).
London: Routledge.
Carston, R. (2002). Thoughts and utterances. The pragmatics of explicit communication. Malden,
MA: Blackwell.  ​https://‍doi.org/10.1002/9780470754603
Carston, R. (2004). Explicature and semantics. In S. Davis, & B. S. Gillon (Eds.), Semantics: A
reader (pp. 1–44). Oxford: Oxford University Press.
Corazza, E., & Dokic, J. 2007. Sense and insensibility or where minimalism meets contextualism.
In G. Preyer, & G. Peter (Eds.), Context-sensitivity and semantic minimalism: New essays on
semantics and pragmatics (pp. 169–193). Oxford: Oxford University Press.
Croft, W. (1993). The role of domains in the interpretation of metaphors and metonymies.
Cognitive Linguistics, 4, 335–370.  ​https://‍doi.org/10.1515/cogl.1993.4.4.335
Falkum, I. (2011). The semantics and pragmatics of polysemy: A relevance-theoretic account.
London: University College London Ph.D. dissertation.
Falkum, I. (2015). The how and why of polysemy: A linguistic account. Lingua, 157, 83–99. ​
https://‍doi.org/10.1016/j.lingua.2014.11.004.
Fauconnier, G. (1997). Mappings in thought and language. Cambridge: Cambridge University
Press.  ​https://‍doi.org/10.1017/CBO9781139174220
Fauconnier, G., & Turner, M. (2002). The way we think: Conceptual blending and the mind’s
hidden complexities. New York: Basic Books.
Ferreira, F., & Patson, N. (2007). The “good enough” approach to language comprehension.
Language and Linguistics Compass, 1(1–2), 71–83. ​
https://‍doi.org/10.1111/j.1749-818X.2007.00007.x
Ferreira, F., Bailey K. G. & Ferraro, V. (2002). Good-enough representations in language com-
prehension. Current Directions in Psychological Science, 11(1), 1–15. ​
https://‍doi.org/10.1111/1467-8721.00158
Gibbs, R. W., & Colston, H. L. (2012). Interpreting figurative meaning. Cambridge: Cambridge
University Press.  ​https://‍doi.org/10.1017/CBO9781139168779
Jodłowiec, M. (2015). The challenges of explicit and implicit communication: A relevance-theoretic
approach. Frankfurt am Main: Peter Lang.  ​https://‍doi.org/10.3726/978-3-653-05190-2
Jodłowiec, M., & Piskorska, A. (2015). Metonymy revisited: Towards a new relevance-theoretic
account. Intercultural Pragmatics, 12(2), 161–187.  ​https://‍doi.org/10.1515/ip-2015-0009
Kövecses, Z., & Radden, G. (1998). Metonymy: Developing a cognitive linguistic view. Cognitive
Linguistics, 9(7), 37–77.  ​https://‍doi.org/10.1515/cogl.1998.9.1.37
Lakoff, G., & Johnson, M. (2003). Metaphors we live by. (2nd ed.) Chicago: The University of
Chicago Press.  ​https://‍doi.org/10.7208/chicago/9780226470993.001.0001
Langacker, R. W. (1993). Reference-point constructions. Cognitive Linguistics, 4, 1–38. ​
https://‍doi.org/10.1515/cogl.1993.4.1.1
Littlemore, J. (2015). Metonymy: Hidden shortcuts in language, thought and communication.
Cambridge: Cambridge University Press.  ​https://‍doi.org/10.1017/CBO9781107338814
Maitra, I. (2007). How and why to be a moderate contextualist. In G. Preyer, & G. Peter (Eds.),
Context-sensitivity and semantic minimalism: New essays on semantics and pragmatics
(pp. 112–132). Oxford: Oxford University Press.
Natsopoulos, D. (1985). A verbal illusion in two languages. Journal of Psycholinguistic Research,
14(4), 385–398.  ​https://‍doi.org/10.1007/BF01067882
64 Maria Jodłowiec and Agnieszka Piskorska

Noveck, I., & Sperber D. (2007). The why and how of experimental pragmatics: The case of
“scalar inferences”. In N. Burton-Roberts (Ed.), Pragmatics (pp. 184–212). Basingstoke:
Palgrave Macmillan.  ​https://‍doi.org/10.1057/978-1-349-73908-0_10
Panther, K-U., & Thornburg L. (2004). The role of conceptual metonymy in meaning construc-
tion. Metaphoric.de, 6, 91–116. http://‍www.metaphorik.de/sites/www.metaphorik.de/files/
journal-pdf/062004pantherthornburg.pdf
Panther, K-U., & Thornburg L. (2005). Inference in the construction of meaning: The role of
conceptual metonymy. In E. Górska, & G. Radden (Eds.), Metonymy-metaphor collage
(pp. 37–57). Warsaw: Warsaw University Press.
Papafragou, A. (1996). On Metonymy. Lingua, 99, 169–195. ​
https://‍doi.org/10.1016/0024-3841(96)00016-2
Park, H., & Reder, L. M. (2004). Moses illusion: Implication for human cognition. In R. Phol
(Ed.), Cognitive illusions: A handbook on fallacies and biases in thinking, judgement and
memory (pp. 275–291). Hove: Psychology Press.
Peters, W., & Peters I. (2000). Lexicalised systematic polysemy in WordNet. Proceedings of LREC
2000, Athens. http://‍www.lrec-conf.org/proceedings/lrec2000/pdf/148.pdf
Pinker, S. (2007). The stuff of thought: Language as a window into human nature. London: Pen-
guin.
Pustejovsky, J. (1995). The generative lexicon. Cambridge, MA: MIT Press.
Recanati, F. (1993). Direct reference: From language to thought. Oxford: Blackwell.
Recanati, F. (2004). Literal meaning. Cambridge: Cambridge University Press.
Reder, L., & Kusbit, G. (1991). Locus of the Moses illusion: Imperfect encoding, retrieval or
match? Journal of Memory and Language, 30, 385–406. ​
https://‍doi.org/10.1016/0749-596X(91)90013-A
Ruiz de Mendoza Ibáñez, F. J. (2011). Metonymy and cognitive operations. In R. Benczes, A.
Barcelona, & F. J. Ruiz de Mendoza Ibáñez (Eds.), Defining metonymy in cognitive linguis-
tics: Towards a consensus view (pp. 103–124). Amsterdam: John Benjamins. ​
https://‍doi.org/10.1075/hcp.28.06rui
Sanford, A. (2002). Context, attention and depth of processing during interpretation. Mind &
Language 17(1–2). 188–206.  ​https://‍doi.org/10.1111/1468-0017.00195
Sanford, A., & Graesser, A. (2006). Shallow processing and underspecification. Discourse Pro-
cesses, 42(2), 99–108.  ​https://‍doi.org/10.1207/s15326950dp4202_1
Sperber, D., & Wilson D. (1995). Relevance: Communication and cognition, 2nd ed. Oxford:
Blackwell.
Sperber, D., & Wilson D. (1998a). The mapping between the mental and the public lexicon.
In P. Carruthers, & J. Boucher (Eds.), Language and thought: Interdisciplinary themes
(pp. 184–200). Cambridge, UK: Cambridge University Press. ​
https://‍doi.org/10.1017/CBO9780511597909.012
Sperber, D., & Wilson D. (1998b). Pragmatics and time. In R. Carston, & S. Uchida (Eds.),
Relevance theory: Applications and implications (pp. 1–22). Amsterdam: John Benjamins.
Sperber, D., & Wilson D. (2005). Pragmatics. UCL Working Papers in Linguistics, 17, 353–388.
Reprinted in F. Jackson, & M. Smith (Eds.), 2005. Oxford handbook of contemporary phi-
losophy (pp. 468–501). Oxford: Oxford University Press.
Sperber, D., & Wilson D. (2008). A deflationary account of metaphors. In R. Gibbs (Ed.), The
Cambridge handbook of metaphor and thought (pp. 84–105). Cambridge: Cambridge Uni-
versity Press.  ​https://‍doi.org/10.1017/CBO9780511816802.007
 Chapter 2. Metonymic relations – from determinacy to indeterminacy 65

Sperber, D., & Wilson D. (2015). Beyond speaker’s meaning. Croatian Journal of Philosophy,
XV(44), 117–149.
Sweep, J. (2012). Metonymical object changes: A corpus-oriented study on Dutch and German.
Amsterdam: University of Amsterdam Ph.D. dissertation.
Turner, M., & Fauconnier, G. (1995). Conceptual integration and formal expression. Metaphor
and Symbolic Activity, 10(3), 183–203.  ​https://‍doi.org/10.1207/s15327868ms1003_3
Turner, M., & Fauconnier, G. (2000). Metaphor, metonymy and binding. In A. Barcelona (Ed.),
Metaphor and metonymy at the crossroads. A cognitive perspective (pp. 133–145). Berlin:
Mouton de Gruyter.
Van Oostendorp, H., & de Mul, S. (1990). Moses beats Adam: A semantic relatedness effect on a
semantic illusion. Acta Psychologica, 74(1), 35–46. ​
https://‍doi.org/10.1016/0001-6918(90)90033-C
Vega Moreno, R. (2007). Creativity and convention: The pragmatics of everyday figurative speech.
Amsterdam: John Benjamins.  ​https://‍doi.org/10.1075/pbns.156
Viscente, A., & Martínez-Manrique, F. (2005). Semantic underdetermination and the cognitive
uses of language. Mind & Language, 20(5), 537–558. ​
https://‍doi.org/10.1111/j.0268-1064.2005.00299.x
Wałaszewska, E. (2008). Polysemy in relevance theory. In E. Mioduszewska, & A. Piskorska
(Eds.), Relevance round table I, (pp. 123–134). Warsaw: Warsaw University Press.
Wilson, D. (2003). Relevance and lexical pragmatics. Rivista di Linguistica, 15(2), 273–291.
Wilson, D. (2015). Explaining metonymy. Paper delivered at the Relevance Round Table Meeting
4. Institute of English Studies, Jagiellonian University in Kraków.
Wilson, D., & Carston R. (2007). A unitary approach to lexical pragmatics: Relevance, inference
and ad hoc concepts. In N. Burton-Roberts (Ed.), Pragmatics (pp. 230–259). Basingstoke:
Palgrave Macmillan.  ​https://‍doi.org/10.1057/978-1-349-73908-0_12
Wilson, D., & Sperber, D. (2002). Truthfulness and relevance. Mind, 111, 583–632. ​
https://‍doi.org/10.1093/mind/111.443.583
Wilson, D., & Sperber, D. (2004). Relevance theory. In L. Horn, & G, Ward (Eds.), The handbook
of pragmatics (pp. 607–632). Oxford: Blackwell.
Wilson, D., & Sperber, D. (2012). Meaning and relevance. Cambridge: Cambridge University
Press.  ​https://‍doi.org/10.1017/CBO9781139028370
Part 2

Concepts, procedures and discourse effects

 Chapter 3

Evidential participles and epistemic vigilance

Manuel Padilla Cruz

Universidad de Sevilla

For communicated contents to be accepted by the audience, they have to pass

the filters of epistemic vigilance mechanisms, which check the credibility and
reliability of communicators and the information provided. Communicators
may lack adequate evidence about the information they dispense. One of the
ways to indicate to the audience that they are uncertain about some information
(rather than to put their reputation as reliable speakers at risk) is to use parti-
cipial adjectives, such as alleged or suspected. The chapter discusses the features
of such adjectives and argues that they specialise for marking the speaker’s
epistemic stance towards the information communicated – a function they
share with other evidentials. Unlike many other expressions denoting epistemic
stance, however, they appear to be confined in their scope to the noun phrase
in which they occur.

Keywords: evidential participles, evidentials, evidentiality, epistemic

vigilance, epistemic stance, (dis)belief, stylistic choices, propositional attitude,
comprehension

1. Introduction

This chapter focuses on a group of the so-called participial adjectives, which seem
to modify heads of noun phrases in attributive position (Greenbaum & Quirk,
1993; Huddleston, 1993). As opposed to prototypical adjectives, participles like the
English alleged or suspected, or the Spanish supuesto or presunto, do not contribute
to the propositional content of assertions or claims by providing information about
the properties or states of the head nouns they co-occur with. Instead, they work
as indicators of the quality of the information communicated about the referent of
that noun and the communicator’s epistemic stance towards what she says.
In contrast to a variety of linguistic expressions signalling epistemic stance
towards a whole proposition, the participles under analysis indicate whether a
fragment of a proposition can be (dis)credited due to lack of adequate evidence

https://‍doi.org/10.1075/ftl.8.03crz
© 2020 John Benjamins Publishing Company
70 Manuel Padilla Cruz

at the time of speaking or writing. More specifically, these participles prevent

the audience from definitively attributing the agency of or the responsibility for
some event, action or state of affairs to the referent of the noun with which they
co-occur, as well as from assigning a certain quality or condition to the referent
of that noun. Consequently, the audience are entitled to assume that further evi-
dence is required in order to attribute agency, responsibility or a condition to the
nominal referent in the (near) future. To put it differently, the participial adjectives
examined here caution the audience that they should not think that, when making
a claim, the referent of a noun has carried out a certain action or that the state of
affairs denoted by the noun actually holds true, even if responsibility for such ac-
tion may subsequently be proved or the denoted state of affairs may subsequently
be found out to hold true. Thus, informers seek to avoid derivation of erroneous
implications and formation of unwarranted beliefs.
Many of the actions performed in verbal communication are made evident
by means of a variety of markers, particles, lexical items or formulaic or ritualised
expressions. Comprehension of dispensed information may be facilitated when
communicators indicate the relationships between different propositions and
guide the audience to the expected cognitive effects (Sperber & Wilson, 1995;
Wilson & Sperber, 1993, 2002a, 2004). Expressions exhibiting the communicator’s
attitude(s) or feeling(s) towards what she says and expressions unveiling, or simply
hinting, that the communicator (firmly) believes, does not believe or questions
some information, or that she possesses or lacks sound evidence for it also aid
interpretation. Stylistic choices, then, end up being fundamental for ensuring that
the audience arrive at intended meaning and what is said achieves the expected
effects. As part and parcel of the stylistic decisions that communicators constantly
make, the participial adjectives under analysis are added to assertions or claims in
an attempt to achieve optimal relevance and make the communicator’s informa-
tive intention unambiguously manifest.
This chapter begins by briefly illustrating some of the actions performed ver-
bally in Section 2. Next, Section 3 exemplifies the variety of linguistic expressions
guiding and constraining comprehension. Since among those expressions are the
past participles alluded to, Section 4 will compare their features to those of pro-
totypical participial adjectives, from which they differ in some morpho-syntactic
aspects. Such differences suggest that those past participles make up a group that
is specialised for a very specific function: assisting the audience in their interpreta-
tive task by indicating whether they can believe, put into question or discredit
the information imparted on the basis of available evidence. By so doing, these
participles share functions with other evidentials (Ifantidou, 2001), so here they
will be labelled evidential participles. Finally, Section 5 discusses their functions
and contribution to communication.
 Chapter 3. Evidential participles and epistemic vigilance 71

2. Things we do with words

Verbal communication is an incredibly complex activity wherewith and wherein a

wide array of social and cognitive goals are pursued and fulfilled. The foundational
works of Speech Act Theory (Austin, 1962; Searle, 1969) made it clear that ut-
terances may not simply describe the world, but also perform actions in more
or less direct or transparent ways. Among them are requesting goods or services
(1), congratulating others on their achievements (2), complimenting them on, for
instance, their appearance (3), or apologising for offences or trespasses (4):1
(1) Can/Could you (please) take this parcel to the post-office for me and send it?
(2) Excellent essay!
(3) I like your new hairdo.
(4) Sorry I am late!

Satisfactory accomplishment of those actions requires determining social

distance, which interlocutor holds more power because of qualifications, skills,
personal qualities or position within a particular institution, or if an action may
involve some cost or benefit to the interlocutors, put them in a difficult position or
contravene their sociality rights and sociality obligations (Brown & Levinson, 1987;
Spencer-Oatey, 2008).2 As a result, interlocutors constantly make choices regarding
what to say and how to say it, and formulate utterances in such a way that they give
sufficient clues about their desire to be considerate or to create, maintain, re-define
or destroy social relationships in a particular situation (Arundale, 2006; Locher,
2006). The request in (1) could be rendered more indirectly if the speaker thought
that going to the post-office might take the requestee a good while or if she wanted
to avoid sounding too bossy because she had previously asked him other favours:
(5) a.
Do you think you could take this parcel to the post-office and send it
for me?
b. I was wondering whether the post-office might still be open.

1. See Norrick (1980) and Marandin (1987) for differences between compliments and congratu-
lations.

2. Sociality rights and obligations are those social or personal expectancies or entitlements
that individuals claim for themselves. Some of them are constantly negotiated, while others
are culturally or situationally determined beforehand. Individuals expect those rights to be
respected, so they have expectations which, if unsatisfied, may affect their social relationships
(Spencer-Oatey, 2008).
72 Manuel Padilla Cruz

Likewise, if a straightforward congratulation on an unequalled essay (2) was per-

ceived as detached, the speaker could elaborate on the reasons why she praises it as
a way of empathising with the hearer and making it clear that she intends to praise
his work (Norrick, 1980; Marandin, 1987):
(6) Excellent essay! It addresses most of the current challenges in relevance-
theoretic pragmatics.

Not knowing another person well may render complimenting risky, as the com-
plimentee may think that the complimenter simply seeks to comply with rules
of etiquette, is flattering or even making fun of him (Wolfson, 1983; Holmes &
Brown, 1987). That riskiness is avoided by means of a more indirect formulation
which evidences that the complimenter has noticed a change in the complimentee
but does not unveil her real opinion about it:
(7) It seems that you’ve been to the hairdresser’s.

If a recurrent formula to apologise (4) is thought to sound insincere, the speaker

may want to show true regret by adding the reasons for the apology:
(8) There was a huge traffic jam and I could not make it on time for class.

Dispensing information other people might need or have an interest in impacts

their set of beliefs. These may be altered by adducing evidence that certain facts
or states of affairs hold or that some of them should be abandoned. Thus, people
attempt to convince others of facts or states of affairs, to persuade them to do
certain things or believe specific issues, or even to manipulate them.
Since information may be true or false, when imparting it individuals are
normally interested in appearing competent – i.e. skilled communicators and
knowledgeable about specific issues – and benevolent – i.e. reliable or trustworthy
informers (Sperber, 1994). Grice (1975) himself captured the intuition that com-
municators are usually expected to tell the truth or that telling it is beneficial for
communication in his Maxim of Quality.3 Benevolence causes speakers to indicate
certainty or lack of evidence about information by means of various stylistic choic-
es. For example, if an individual wants to communicate that it actually is or may be
cloudy in Seville at a specific time, she can opt for any of the formulations below
in order to present her degree of certainty or the kind of evidence she relies on:
(9) a. It is cloudy in Seville now.
b. It must be cloudy in Seville now.
c. It may be cloudy in Seville now.

3. See Sperber & Wilson (1995), Wilson (1995) and Wilson & Sperber (2002b) for arguments
against this maxim and the cooperative principle.
 Chapter 3. Evidential participles and epistemic vigilance 73

d. It might be cloudy in Seville now.

e. They say that it could be cloudy in Seville now.
f. I have (just) heard that it is cloudy in Seville now.

3. Style and relevance

Comprehension normally follows the path of least cognitive effort and maximum
cognitive benefit. The human mind carries out a process of mutual parallel adjust-
ment that mobilises a series of mechanisms or modules that perform a number of
simultaneous tasks (Carston, 2002). Among them are decoding, inferring, mind-
reading, emotion-reading or assessing the veracity and reliability of information
(Wilson, 2012).4 Decoding yields an organised set of conceptual representations
(Sperber & Wilson, 1995, p. 72). Inference enables segmentation of sounds and
identification of words, parsing and disambiguation of syntactic constituents, as-
signment of reference, adjustment of the conceptual content of some words, and
recovery of ellipsed material. The output of these tasks is the lower-level explica-
ture of an utterance, or a fully propositional form whose truthfulness or falsity
can be verified.
Inference, mindreading and emotion-reading work jointly in order to
construct a description of the speech act that the speaker is thought to accom-
plish, the attitude that she is considered to have towards what she says or her
degree of certainty about it: the higher-level explicatures (Sperber & Wilson, 1995,
pp. 181–182). Finally, inference makes it possible to relate the content of an ut-
terance to assumptions that the audience have evidence to think that the speaker
expected them to supply – implicated premises – so as to arrive at an intended
implicated content – implicated conclusions (Sperber & Wilson, 1986/1995; Wilson
& Sperber, 2002a, 2004).
The fast pace at which these tasks are performed and psychological and physi-
ological factors – e.g. tiredness, absent-mindedness, multi-tasking, drowsiness,
etc. – often negatively affect their output. In fact, on many occasions the audi-
ence experience problems at disambiguating constituents, assigning reference or
constructing higher-level explicatures, miss implicatures or arrive at unintended
ones (Yus Ramos, 1999a, 1999b; Padilla Cruz, 2013). In order to ensure correct
understanding, communicators may guide the audience by means of linguistic or
expressive choices, i.e. style.

4. As opposed to cognitivists, relevance theory endorses the massive modularity thesis, which
conceives of the human mind as a complex set of specialised and mandatory mechanisms.
74 Manuel Padilla Cruz

Within relevance-theoretic pragmatics stylistic choices are seen as result-

ing from the communicator’s willingness to be optimally relevant and assist
comprehension (Sperber & Wilson, 1986/1995, p. 219; Trotter, 1992, p. 11). The
presumption of optimal relevance that every utterance communicates entails that
the utterance will be worth the audience’s effort to process it and that it is worded
in the most effort-saving way, depending on the speaker’s abilities and preferences
(Sperber & Wilson, 1995, p. 270).5 Indeed, speakers always make decisions on
issues such as the structure or level of formality of their contributions and take
advantage of the repertoire of their language in order to assist the audience in their
interpretative tasks.
In (1) above, interrogative syntax, a conventionalised formula (can/could you
+ verb) and the illocutionary force indicating device (Levinson, 1983) please favour
a requestive interpretation and rule out an interpretation as a question about the
hearer’s physical abilities. Exclamative syntax and the structure “adjective + noun”
facilitate an interpretation of (2) as a congratulation – provided the appropriate
paralanguage accompanies it. Quite similarly, a structure such as I like followed by
a direct object contributes to the interpretation of (3) as a compliment (Wolfson &
Manes, 1980; Manes & Wolfson, 1981). In the case of (4), the occurrence of sorry
renders the apology interpretation almost automatic. Had the speaker wanted to
avoid sounding insincere or felt that a more indirect formulation like (8) could be
misinterpreted, she could also have resorted to an explicit performative:
(10) I apologise for being late to class.

Although probably costlier in terms of processing effort, alternative, more verbose

formulations like (5)–(8) often give the audience evidence of the speakers’ inten-
tion to communicate weak implicatures about, for example, how she treats them
or their social relationship (Sperber & Wilson, 1995). Derivation of those impli-
catures involves additional, beneficial cognitive effects (Escandell Vidal, 1998) of a
social or behavioural nature (Jary, 2013; see also Haugh, 2014), which could not be
obtained by means of other stylistic choices. Indeed, more elaborate and verbose
formulations sometimes get marked as opposed to expected, default ones – which
would be unmarked – and give the audience reason to suspect the speaker’s inten-
tion to make (more) manifest assumptions about her considerateness or politeness
(Jary, 1998, p. 9).

5. The notion of ‘abilities’ refers to the cognitive skills and capabilities underlying linguistic
performance, while that of ‘preferences’ alludes to a variety of goals, such as complying with
norms of politeness or norms dictating the type or amount of information to impart, to whom,
how, when and where to present it, etc. (Mazzarella, 2013, pp. 33–35).
 Chapter 3. Evidential participles and epistemic vigilance 75

Relevance theorists have extensively shown that speakers assist the construc-
tion of interpretative hypotheses by means of a rich variety of linguistic resources.
For instance, discourse markers indicate the relationships between specific
propositions; their procedural meaning steers mental computations in one direc-
tion or another by imposing constraints on the type of inferential process needed
(Blakemore, 1987, 2002; Jucker, 1993; Moeschler, 1993; Wilson & Sperber, 1993;
Rouchota, 1995). Accordingly, a marker like so encodes the instruction that two
propositions must be connected by a relation of cause and effect:6
(11) Mark came to Seville. So he visited the cathedral and the Alcázar.

Attitudinal adverbials (e.g. happily, sadly, unfortunately, etc.), interjections (e.g.

oh!, wow!, etc.), intonation and paralanguage help infer the speaker’s attitude
towards the proposition expressed. While intonation and paralanguage have been
argued to be purely procedural elements (Imai, 1998; Wilson & Wharton, 2006;
Wharton, 2009), attitudinal adverbials have been claimed to encode a conceptual
content, even if it does not contribute to the lower-level explicature of an utterance,
but to its higher-level explicature (Ifantidou, 1992). Interjections would make up
a hybrid category that includes elements that may be placed along a cline ranging
from purely procedural elements to more conceptual ones (Wharton, 2003, 2009):7
(12) Unfortunately, Mark could not come to Seville this year.
(13) Wow! Mark has come to Seville.

Expressive possibilities enable speakers to reformulate utterances with a view to

ensuring correct understanding. Reformulations have been argued to constrain
interpretative routes and enable exploration of the encyclopaedic entries of lexical
items (Blakemore, 1992, 1993, 1994):8
(14) Mark visited the Alcázar in Seville. In other words, he visited the oldest
inhabited royal residence in Europe.

6. According to Carston (2016), the various contributions on connectives or discourse markers

made after Blakemore’s (1987, 2002) influential work may be considered the first stage in the
development of the notion of procedural meaning.

7. See Padilla Cruz (2009) for comments on Wharton’s (2003, 2009) relevance-theoretic
analysis of interjections. The relevance-theoretic analyses of elements contributing to attitudinal
descriptions are part of the second stage in the development of the notion of procedural mean-
ing (Carston 2016).

8. For a different view and criticism, see Culpeper (1994).

76 Manuel Padilla Cruz

Concerning presentation of information, relevance theorists have also argued

that the so-called illocutionary adverbials (e.g. frankly, seriously, etc.), evidential
adverbials (e.g. obviously, evidently, clearly, etc.), hearsay adverbials (e.g. allegedly,
reportedly, etc.) and some parenthetical expressions (e.g. they say, I hear, etc.)
indicate possession or lack of supportive evidence for information. These ele-
ments show the informer’s different degrees of commitment to the proposition
expressed or whether the information dispensed is considered true, false, reliable
or in need of further evidence (Ifantidou, 1992, 1993, 2001; Wilson & Sperber,
1993; Wilson, 1999):
(15) Frankly, Mark visited the Alcázar in Seville.
(16) Clearly, Mark enjoyed the Alcázar.
(17) Allegedly, Mark came to Seville and visited the Alcázar.
(18) a. I hear that Mark did not enjoy the Alcázar.
b. Mark did not enjoy the Alcázar, I hear.
c. Mark, I hear, did not enjoy the Alcázar.

Similar functions are fulfilled by the indicative mood (9a), the various modal
verbs (9b)–(9d) and the reporting verbs (9e)–(9f). In Japanese, the utterance-final
hearsay particle -tte is employed with the same purposes (Itani, 1994, 1998), while
in Sissala ré is normally inserted in reported speech or thought, or after speech-
act and propositional-attitude verbs corresponding to think, believe or know
(Blass, 1989, 1990).
English and Spanish also resort to past participles that seem to work similarly.
In a genre like journalism, where professional ethics encourages objective pre-
sentation of facts, clear indication of the source(s) of evidence and unambiguous
reference to the veracity of information (Stovall, 2004; Meyers, 2010; Rich, 2015),
headlines often attempt to draw readers’ attention with texts like these:
(19) Brother of alleged Holly Bobo killer arrested for disposing of evidence.
 (www.mydailynews.com 19/09/2014)
(20) Cops release images of waitress’s alleged killer. 
 (www.kaieteurnewsonline.com 7/‍09/2014)
(21) Boy suspected kidnapper dead after Colorado hostage standoff. 
 (www.reuters.com 5/‍08/2014)
(22) Update: Suspected kidnapper arrested after crashing car. 
 (www.racinecontyeye.com 22/07/2014)
 Chapter 3. Evidential participles and epistemic vigilance 77

(23) Detenido el supuesto homicida de una mujer en un coto en Coria. (Sevilla)

 (www.lavanguardia.com 22/09/2014)
Suspected killer of woman in reserve in Coria (Seville) arrested.
(24) El supuesto homicida de El Atazar ahogó y acuchilló a su exmujer. 
 (www.elmundo.com 9/‍08/2011)
Suspected El Atazar killer choked and stabbed ex-wife.
(25) Alaya cifra en 855 millones el presunto fraude de los ERE en Andalucía
 (www.lasexta.com 15/08/2014)
Judge Alaya values presumed ERE fraud in Andalusia in 855 millions.

The English past participles alleged and suspected and the Spanish equivalents
supuesto and presunto in these examples do not denote any temporary, accidental
or permanent feature or state of the referent of the nouns they accompany, so they
cannot be regarded as proper modifying adjectives. Rather, they suggest that the
individuals referred to as killers or kidnapper, and the action labelled fraud, are not
yet to be definitely considered or believed as such at a particular time, as each of
them might still be under judicial investigation or in need of such investigation. In
fact, in countries like Spain, laws seek to guarantee, protect and respect individuals’
right to a presumption of innocence, so potential criminals and crimes must not
be properly regarded as such until enough evidence is found, adduced or provided
by witnesses or investigation, and the court or judge announces a sentence. These
participles hint that the referents of the pre-modified nouns could be referred to in
a particular way or attributed specific properties – being a killer or kidnapper, or
fraudulence – in the (near) future as a result of discovery of subsequent supporting
evidence which, at the time of writing, is still non-existent or unconfirmed.
Participial adjectives are past participles that appear as adnominal modifiers
of nouns (Huddleston, 1988, 1993; Greenbaum and Quirk, 1993). Prototypical
participial adjectives denote qualities or states of the modified nouns, so they are
included as members of the category of adjectives. Like other grammatical catego-
ries incorporating lexical items from other categories, that of adjectives is an open
one which, in addition to some adverbs and nouns modifying nominal heads, also
groups de-verbal items like present and past participles.
Clearly, the past participles above (19)–(25) do not work as proper participial
adjectives. Not only do they not denote qualities or states of a noun, but also they
do not exhibit some of the features characterising adjectives and participial
adjectives.
78 Manuel Padilla Cruz

4. Properties of prototypical participial adjectives

Many past participles in English or Spanish have adjectival functions and

share some of the general properties characteristic of central or core adjectives
(Huddleston, 1988, pp. 108–110; Haegeman & Guéron, 1999, pp. 56–57, 71–72;
Collins & Hollo, 2000, pp. 80–81; Börjars & Burridge, 2001, pp. 64–65):
a. Occurrence in adnominal positions either before the head noun and after
the determiner (attributive position), as in English, or after the head noun
(postpositive position), as in Spanish:9
(26) a. A broken chair.
b. Una silla rota.
(27) a. The worried person.
b. La persona preocupada.
(28) a. The greatest record broken.
b. El mayor record (jamás) batido.

b. Occurrence in predicative position as subject complement in attributive sen-

tences (29) or as object complement (30):
(29) a. The chair is broken.
b. La silla está rota.
(30) a. He considered the man worried.
b. Consideró al hombre preocupado.

c. Pre-modification by intensifiers:
(31) a. He was very worried.
b. (Él) Estaba muy preocupado.
c. The chair was utterly destroyed.
d. La silla estaba completamente destrozada.

d. Gradability, i.e. comparative and superlative forms:

(32) a. He was more shocked than surprised.
b. (Él) Estaba más impactado que sorprendido.
(33) a. He was the most surprised person in the room.
b. (Él) Era la persona más sorprendida de la sala.

9. While in Spanish this seems to be the average or default position for adjectives, in English
this position is restricted to adjectives ending in -able or -ible when the head noun is modified
by another adjective in the superlative or other modifiers.
 Chapter 3. Evidential participles and epistemic vigilance 79

When used attributively, participial adjectives usually have a passive meaning that
can be captured by an equivalent defining relative clause:
(34) a. The offended man > The man who/that was offended.
b. El hombre ofendido > El hombre que ha sido ofendido.
c. Lost property > The property that has been lost.
d. Propiedad perdida > La propiedad que ha sido perdida.

The passive reading, nevertheless, is ruled out when the corresponding verb is
intransitive:
(35) a. The departed train > The train that (has/had) departed.
b. El difunto marido > El marido que ha fallecido.
c. The escaped prisoner > The prisoner who (has/had) escaped.
d. El prisionero huido > El prisionero que ha huido.

Predicative position, in contrast, is only permitted for some participles:

(36) a. The curtains are faded > The curtains have faded.
b. Las cortinas están desteñidas > Las cortinas han desteñido.
c. He is now retired > He has now retired.
d. (Él) Ahora está jubilado > (Él) Ahora se ha jubilado.

Like prototypical adjectives, participial adjectives contribute to the proposition

expressed. Replacement with another participle or adjective changes sentence
meaning. Replacement with a defining relative clause does not alter the propo-
sitional content, although the resulting sentence differs in terms of syntactic
structure (34). In relevance-theoretic terms, participial adjectives can therefore be
said to encode a conceptual content that contributes to the lower-level explicatures
of utterances and affects their truth conditions (Sperber & Wilson, 1986/1995;
Wilson & Sperber, 2002a, 2004).
Despite their attributive position, the adjectival status of the participial adjec-
tives in the headlines in the previous section (19)–(25) is dubious. Neither can
they be intensified (37), nor do they seem to have comparative or superlative
forms (38)–(39):
(37) a. *The brother of very alleged Holly Bobo killer.
b. *Very suspected kidnapper arrested.10
c. *Detenido el muy supuesto homicida.
d. *El muy presunto fraude de los ERE.

10. Note, however, that these participial adjectives can be modified by adverbs such as widely.
80 Manuel Padilla Cruz

(38) a. *He was more alleged than suspected.

b. *(Él) Era más presunto que supuesto
(39) a. *He was the most alleged killer (in the room).
b. *(Él) era el homicida más supuesto (de la sala).

Also, placing those participles in predicative position yields a bizarre sentence:

(40) a. *The killer is alleged.
b. *El homicida es presunto.

Furthermore, although the corresponding verbs of the English participles are

transitive and, consequently, would allow for a passive transformation, substitu-
tion with a passive form results in an odd sentence:
(41) a. *The killer who/that is/was alleged.
b. *El homicida que es/fue presunto.

Even if allege is transitive, the person alleged to have done something is not its
direct object, so a sentence like (42) would be ungrammatical:
(42) *The police alleged him.

Passivisation could only be licensed if the participles were followed or comple-

mented by an infinitival clause:
(43) The man who/that is/was alleged/suspected to have committed three crimes.

Substitution of the participles of the Spanish verbs suponer or presumir with

a passive form would not be very common or natural either. Despite a certain
preference for impersonal structures with se as a subject slot-filler – se supone, se
presume – if the Spanish participles of these verbs were substituted with imper-
sonal forms, they would subcategorise a finite complement clause introduced by
the complementiser que, which functions as their direct object:
(44) El hombre que se supone/presume que ha cometido tres crímenes.

Failure to share the properties of prototypical participial adjectives suggests that

the past participles analysed in this work may constitute a sub-type in their own
right specialised for fulfilling a specific function: informing about the communi-
cator’s evidence when making an assertion. In this respect, these past participles
work as evidentials and their insertion in utterances is intended to achieve specific
perlocutionary effects: whether the audience should take with a pinch of salt, ques-
tion or not yet completely believe a claim or assertion made because of lack of the
pertinent evidence (Wilson, 2012; Piskorska, 2016).
 Chapter 3. Evidential participles and epistemic vigilance 81

5. On the evidential function of some past participles

Some modals verbs (e.g. may, must, should), adjectives (e.g. able, possible) and
clausal elements (e.g. I think, they say) have often been considered devices to ex-
press modality (Palmer, 1986, 2001). Modality must be distinguished from verbal
features such as mood, tense and aspect, which are realised by inflections in many
languages, because “[…] it does not refer directly to any characteristic of the event
[…]” presented in a proposition, “[…] but simply to the status of the proposition”
(Palmer, 2001, p. 1). As Huddleston (1993, pp. 165–166) states, modality is “[…] a
rather broad term for […] a category of meaning.”
Traditionally, grammarians have differentiated two kinds of modality:
a. Epistemic – from the Ancient Greek word for knowledge (Huddleston, 1993,
p. 166) – which is related to the status of a proposition as true, false, probable,
possible, necessary, etc., depending on what the speaker knows. This kind of
modality is also known as extrinsic because there is a “[…] human judgement
of what is or is not likely to happen” and seen as more objective (Greenbaum
& Quirk, 1993, p. 60).
b. Deontic – from the Ancient Greek word alluding to the notion of binding –
which has to do with how the speaker presents an action, i.e. as obligatory,
permitted, advisable, etc. (Huddleston, 1993, pp. 167–168). This kind of mo-
dality is also labelled intrinsic because some “[…] human control over events”
is involved and is seen as more subjective (Greenbaum & Quirk, 1993, p. 60).
The so-called epistemic modals have more recently been re-analysed as convey-
ing information about the speaker’s attitude(s) towards the propositional content
communicated. Among those attitudes are, for instance, strong (dis)belief in, (un)
certainty about or (non-)commitment to the truth of the propositional content
of an assertion.
Modality must be distinguished from evidentiality, a linguistic category “[…]
whose primary meaning is source of information” (Aikhenvald, 2004, p. 3) and re-
fers to the speaker’s indication of her degree of commitment to a claim depending
on available evidence (Crystal, 1991, p. 127). Such indication is motivated by the
source(s) of the information taken into account when communicating (Dendale &
Tasmowski, 2001), which may be perceptual or epistemological (Cornillie, 2007,
p. 45): visual, non-visual but directly perceived, informed by perceptual clues,
assumed via testimony, etc. (Aikhenvald, 2004). Since the source(s) of informa-
tion may be more or less reliable or trustworthy (Matthews, 2007), evidentiality
is the indication of the origin(s) of the knowledge on which the speaker bases her
assertion(s) about (a) particular state(s) of affairs, or of the compatibility of that/
those state(s) of affairs with her own universe of beliefs (Nuyts, 2006, p. 10).
82 Manuel Padilla Cruz

Stylistic choices lead to optimally relevant interpretations, but those interpre-

tations need not be reliable or credible. In effect, informers do not always behave
benevolently because they may attempt to deceive or misguide the audience, or
competently because they may lack adequate evidence when imparting informa-
tion. In relevance-theoretic pragmatics, linguistic expressions like evidential and
hearsay adverbials, parenthetical elements or hearsay particles have been analysed
as indicators of the sort of evidence communicators rely on and are subsumed
under the umbrella term evidentials (Ifantidou, 2001). Evidentials have been
claimed to indicate whether the audience should believe, discredit or be cautious
about dispensed information. They have been argued to trigger the construction
of (more) sophisticated higher-level explicatures that include some indication
of the speaker’s degree of (un)certainty about the communicated content or the
evidence she relies on when making an assertion:
(45) a. The speaker (strongly/firmly) believes that p.
b. The speaker is (very) certain/uncertain that p.

Relevance theorists posit that the human mind has developed a complex set of
specialised mechanisms for monitoring the believability and reliability of inform-
ers – i.e. the source(s) of information – and the information that they supply –
i.e. the content itself. Those mechanisms take into account a variety of internal
and external factors that determine whether a person and the information she
provides should be trusted (Origgi, 2013) and trigger an attitude of epistemic
vigilance (Mascaro & Sperber, 2009; Sperber et al., 2010). Following Wilson (2012,
2016), the past participles on which this chapter focuses could be considered,
like other evidentials, to fulfil an important function: activating and assisting
epistemic vigilance mechanisms in their assessments of the trustworthiness of
informers and information. Accordingly, the participles in question could have
become specialised for targeting epistemic vigilance mechanisms by alerting them
to the likely veracity or falsity of claims due to lack of evidence.

5.1 Evidentials and epistemic vigilance

Informers and information deserve trust to a greater or lesser extent. People have
epistemic confidence in other individuals and assign or deprive them of epistemic
trustworthiness (Fricker, 2007). Research in developmental psychology has revealed
that between the ages of two and three, children develop a sensibility towards
individuals and what they say. As a result, children can determine the veracity
or falsity of information, can contradict or correct assertions they consider false
or questionable, and seem to prefer individuals whom they regard as benevolent
and competent on the basis of past personal experiences and what other people
 Chapter 3. Evidential participles and epistemic vigilance 83

tell them about those individuals (Clément et al., 2004; Koenig & Harris, 2007;
Heyman, 2008; Corriveau & Harris, 2009). These findings lend support to the idea
that the human mind comprises a cluster of varied mechanisms specialised for
discriminating the trustworthiness or reliability of informants and of information.
Epistemic vigilance mechanisms take into account a variety of sources of trust
(Origgi, 2013, pp. 227–233) which cause an individual to think of others and what
they claim or assert in a particular way:
a. Beliefs and prejudices about an informer’s reliability accrued from prior ex-
changes.
b. The relevance of what is said.
c. Internalised social norms of complying with authorities or experts in some
domain or issues.
d. The informer’s socially distributed reputation as an informant.
e. Signals that unveil knowledge/ignorance about or possession/lack of evidence
for specific issues: hesitation, stuttering, odd syntax, rephrasing, difficulties at
finding appropriate words, or particles, adverbials and clausal elements like
those presented above.
f. Emotional reactions biasing the conclusions derived about the informer: af-
fect, anger, wrath, etc.
g. Moral commitments determining whether the hearer should actually think of
the informer in a particular way.
These sources cause epistemic vigilance mechanisms to trigger a certain alertness
or critical stance towards informers and information (Sperber et al., 2010, p. 363).
These mechanisms do not automatically generate distrust, but a critical attitude
that differs from blind, naïve and uncritical trust (Sperber et al. 2010; Mercier &
Sperber, 2011; Sperber & Mercier, 2012). Such an attitude empowers individuals
to move from a position of indiscriminate trust, in which credibility is almost au-
tomatically given to beliefs or states of affairs, or another position of gullible trust,
where information is believed even if it contradicts previous personal observations,
to one of sceptical trust (Clément et al., 2004, pp. 361–363). In this last position,
information and the implications following from it are not uncritically believed,
above all if the informer has proved unreliable beforehand or the information is
perceived not to have been duly supported by pertinent evidence. In so doing,
vigilance mechanisms safeguard individuals against some of the possible risks of
communication: deception and misinformation.
Epistemic vigilance mechanisms may not always be effective, as they may be
weakly activated (Michaelian, 2013, p. 42). Their default state, according to Sperber
(2013, p. 64), is one of moderate activation. However, individuals raise their acti-
vation “[…] by a closer inspection of data, sometimes interrogating [themselves]
84 Manuel Padilla Cruz

about the sources of [their] trust or distrust, and sometimes by refining [their]
cognitive heuristics” (Origgi, 2013, p. 224). When vigilance is raised, individuals
become actively vigilant and their alertness to the quality of information increases.
Active or strong vigilance involves (Origgi, 2013, p. 226–227):
i. External vigilance or ‘looking outward’, so to say, in order to become aware
of the operating cultural norms and contextual elements (e.g. preceding
discourse, paralanguage, elements in the communicative situation, etc.) that
determine allocation or deprival of trust to information.
ii. Internal vigilance or ‘looking inward’, so to say, by scrutinising the interpreta-
tive steps taken, the cognitive tasks performed, the beliefs used when contex-
tualising information and the conclusions reached.
Raised vigilance is fundamental for maintaining a necessary critical stance on
the biases, social pressures and prejudices that might affect thinking. Indeed, an
actively vigilant attitude facilitates awareness of the reasons why trust is allocated
to interlocutors and information.
Informants may also contribute to the activation of vigilance mechanisms or
assist them in their decisions about whether to believe or discredit information
and/or informers through stylistic decisions. In addition to the elements exempli-
fied in Section 3, relevance theorists have also recently paid attention to some
quotatives in languages like Estonian or Sissala, which are exploited in narratives
and argumentation in order to aid vigilance mechanisms in their tasks (Unger,
2016). In fact, in the case of argumentation more specifically, convincing the
audience of some fallacies may be a matter of bypassing the filters of vigilance
mechanisms (Oswald, 2011, 2016).11
The past participles discussed in this chapter seem to fulfil a similar assistive
function for epistemic vigilance mechanisms. In a similar way to hearsay particles,
hearsay and evidential adverbials, parenthetical expressions and some main verbs
(Ifantidou, 1992, 1993, 2001; Itani, 1994, 1998; Wilson & Sperber, 1993; Wilson,
1999), those participles indicate whether an informer has (sound) evidence to
consider that a part of a claim she makes actually holds true at the moment of
speaking. In other words, those participles indicate that the way in which an event,
state of affairs or individual are labelled, characterised or alluded to in an assertion
should not necessarily be taken for granted. Thus, those participles additionally

11. In some types of jokes, vigilance mechanisms would detect if the comprehension module
has been fooled into granting plausibility to an interpretative hypothesis that appears optimally
relevant but is inadequate (Padilla Cruz, 2012), while in puns vigilance mechanisms would
detect that ambiguous words or fragments are not correctly disambiguated (Padilla Cruz, 2015).
 Chapter 3. Evidential participles and epistemic vigilance 85

prevent the audience from giving credibility to implications likely to be derived

from the claim made at the time of speaking.
Consider assertions or claims like (46) and (48), from which implications like
(47) and (49) may respectively be drawn:
(46) Suspected woman kidnapper arrested after car crash.
(47) a. Someone will be sentenced to prison.
b. Someone might have to pay compensation.
(48) Judge values presumed fraud in 855 millions.
(49) a. Someone is a robber.
b. Someone will be sent to prison.
c. Someone will have to return stolen money.
d. Someone should not have public responsibilities.

The participles in (46) and (48) unveil that the informer has not accrued the requi-
site evidence to be certain at the time of speaking about the veracity of a likely state
of affairs mentioned in those assertions or claims – namely, that a particular person
has actually kidnapped a woman or that someone has really committed monetary
fraud – or that the evidence for regarding someone or some events in a particular
way is not completely reliable. Thus, these participles caution the audience against
deriving and crediting certain implications (47), (49). Those participles alert vigi-
lance mechanisms to the potential (im)plausibility of a state of affairs mentioned
in a claim or assertion, as well as to its likelihood to be (un)trustworthy.
The morpho-syntactic peculiarities of those participles suggest that, in con-
trast to standard pre-modifying past participles, languages like English or Spanish
might have dedicated them to activate epistemic vigilance mechanisms or raise
the activation of these mechanisms in case they are weakly activated or “dormant,”
so to say. This is essential for these mechanisms to detect informers’ trustworthi-
ness or certainty about claims or assertions they make. Raised activation of these
mechanisms results in external vigilance, which prompts an audience to search
for additional evidence that backs them up in believing or discrediting a state(s)
of affairs alluded to in a claim or assertion, or wait for the necessary evidence
to be adduced.
Due to these functions, past participles like English alleged or suspected
and Spanish presunto or supuesto could be termed evidential participles and be
described as alerters of epistemic vigilance mechanisms to the credibility or likeli-
hood of a state of affairs mentioned in a claim or assertion. Evidential participles
indicate whether the audience are entitled to believe, discredit or take with a grain
of salt a part or fragment of what an informant says, and to draw and believe
86 Manuel Padilla Cruz

possible implications ensuing from it. Even though evidential participles share the
functions described above with other evidential expressions, their pre-modifying
position right before the modified head noun seems to suggest that they work
slightly differently.

5.2 What evidential participles communicate

One of the most significant contributions of relevance-theoretic pragmatics has

been its distinction between conceptual and procedural expressions (Blakemore,
1987, 1992, 2002; Wilson & Sperber, 1993, 2002a, 2004; Carston, 2016; Wilson,
2016). While the former encode representations that may become constituents
of lower-level explicatures and, therefore, affect the truth-conditional content
of utterances, the latter encode instructions steering processing and compre-
hension. Typical examples of purely conceptual expressions are nouns, verbs,
adjectives or adverbs, while typical examples of purely procedural expressions
are discourse markers.
The conceptual-procedural distinction is not a clear-cut one, nonetheless: not
all conceptual expressions contribute to lower-level explicatures and, therefore, to
the truth-conditional content of utterances, and some procedures may be encoded
by expressions with a conceptual nature (Wilson & Sperber, 1993). There are
conceptual expressions that contribute to higher-level explicatures by constraining
their construction; for instance, illocutionary, attitudinal and evidential adverbi-
als.12 On the other hand, some procedural expressions may encode some sort of
conceptual content, even if schematic; for example, personal pronouns.
Evidential participles seem to be conceptual and to contribute to the proposi-
tion expressed. If they are removed or replaced with an adjective or another past
participle, the meaning of the resulting assertion and its truth conditions change:
(50) a. Alleged/Suspected killer arrested.
b. Killer arrested.
c. Handsome killer arrested.
(51) a. Presunto/supuesto homicida arrestado.
b. Homicida arrestado.
c. Apuesto homicida arrestado.

In the first sentence of these two examples, the speaker claims that someone who
must (still) be suspected to be a killer due to lack of contrary evidence has been ar-
rested. In the second sentence, the speaker asserts that a person who definitely is a

12. See Ifantidou (1992, 1993) for a discussion of different tests for truth-conditionality to these
expressions, such as insertion in conditional or disjunctive structures.
 Chapter 3. Evidential participles and epistemic vigilance 87

killer has been arrested. In the third sentence, the speaker informs about the arrest
of a killer who is characterised by a particular feature. The evidential participle,
therefore, contributes to the lower-level explicature of the utterance.
Evidential participles have hearsay adverbial counterparts like English alleg-
edly and supposedly (52) and Spanish supuestamente and presuntamente (53). These
adverbials, which are respectively derived from the transitive verbs to allege and to
suppose in the case of English, and from suponer and presumir in that of Spanish,
may be placed in distinct positions. Note that, if the corresponding English verb is
passivised, it subcategorises an infinitival clause (54). Spanish, in contrast, prefers
an impersonal form followed by a finite complement clause that is introduced by
the complementiser que and functions as the direct object (55):
(52) a. Allegedly, the killer was sent to prison.
b. The killer was, allegedly, sent to prison.
c. The killer was sent to prison, allegedly.
(53) a. Supuestamente, el homicida fue enviado a prisión.
b. El homicida fue, supuestamente, enviado a prisión.
c. El homicida fue enviado a prisión, supuestamente.
(54) The killer is alleged to have been sent to prison.
(55) Su supone que el homicida fue enviado a prisión.

However, what the speaker reports by means of assertions containing hearsay

adverbials (52), (53) differs from what she claims with assertions containing
evidential participles:
(56) a. The alleged killer was sent to prison.
b. El supuesto homicida fue enviado a prisión.

In (52) and (53) the speaker communicates her uncertainty about or lack of strong
evidence concerning the fact that someone, who can definitely be regarded as a
killer, was actually sent to prison. In (56) the speaker asserts that someone, who
cannot yet be considered a killer because of lack of backing evidence, was cer-
tainly imprisoned. Therefore, while hearsay adverbials indicate the weakness of
the informer’s belief in, certainty about or commitment to a whole proposition,
evidential participles only unveil the weakness of her belief in, certainty about or
commitment to a state of affairs alluded to in an assertion by means of one of its
constituents: a noun phrase. Obviously, the hearer of an utterance like (56) may
think that the speaker, upon using the evidential participle, also intends to subtly
invite an implicature to the effect that she hopes or expects – or that it is hoped or
expected – that confirming evidence will come to light.
88 Manuel Padilla Cruz

The fact that hearsay adverbials and evidential participles affect what an in-
former claims in differing ways is attested by distinct tests. Firstly, hearsay adverbi-
als are not properly integrated into the syntax of an assertion or claim. They are
separated from it by a pause in speech and a comma in writing, thus constituting
an independent tone-unit. Moreover, they may be freely placed before or after the
assertion/claim, or in the middle of it. However, evidential participles occupy a
fixed position as pre-modifiers of a nominal head.
Secondly, substitution of hearsay adverbials by their corresponding verbs
followed by a clause acting as direct object is possible. For evidential participles
to be replaced by their corresponding verbs, it would be necessary to use a defin-
ing relative clause. In English, that clause would have a passive verb that in turn
subcategorises an infinitival clause. In Spanish, the verb of that clause would be
an impersonal form that subcategorises a finite complement clause introduced
by the complementiser que. Additionally, for the defining relative clause to be
licensed in Spanish, the initial noun – e.g. killer – needs substituting with another
one – e.g. person, man:
(57) a.
The man who is alleged/supposed to be the/a killer has been sent to
prison.
b. El hombre que se supone/cree que es un homicida ha sido enviado a
prisión.

These formal and transformational differences reveal that the scope of hearsay
adverbials is the whole asserted proposition, whereas that of evidential participles
is only a part of the asserted proposition. Therefore, hearsay adverbials instruct
epistemic vigilance mechanisms to be cautious about the veracity of a whole
proposition, while evidential participles alert those mechanisms to the likely
untruthfulness of a state of affairs alluded to in an assertion/claim.

6. Conclusion

Comprehension is assisted by a number of linguistic and paralinguistic elements

that include discourse markers, attitudinal adverbials, interjections or intonation.
Languages also count on other devices that enable informers to warn about the
believability of information: hearsay particles and adverbials, evidential adverbi-
als, some main verbs and parenthetical expressions. Generally labelled eviden-
tials, these devices indicate informers’ benevolence and level of competence in
some domain or issue.
This chapter has looked into a group of past participles which, unlike proto-
typical ones, do not have adjectival functions but work as evidentials. Evidential
 Chapter 3. Evidential participles and epistemic vigilance 89

past participles, as they are termed in this chapter, encode a conceptual content
that contributes to the truth-conditional content of the proposition where they ap-
pear, but they also encode procedural meaning. Like hearsay adverbials, evidential
participles have been argued to target the cluster of mechanisms responsible for
an attitude of epistemic vigilance. They enact the activation of those mechanisms
or increase their activation, which results in external vigilance of preceding or
upcoming discourse. Active vigilance is essential for avoiding (indiscriminate)
gullibility and adopting the sceptical trust indispensable for the formation and
fixation of accurate beliefs.
While hearsay adverbials take within their scope the whole proposition they
are appended to, the scope of evidential participles is more limited. It only is a
fragment of an asserted proposition: a nominal head. This means that evidential
participles alert vigilance mechanisms to the fact that the referent of a nominal ex-
pression alluded to in the asserted proposition should not be taken as responsible
for a particular state of affairs or to the fact that the state of affairs referred to by
means of a noun may not actually hold as true at the moment when an assertion is
made because of lack of adequate evidence or unreliability of available evidence. As
a result, the audience is not entitled to derive and regard as true some implications
that may follow from the assertion made. Thus, evidential participles suggest that it
would only be in the (near) future, and provided that reliable evidence is adduced,
that a nominal referent could definitely be considered responsible for a particular
state of affairs, or a state of affairs alluded to could be believed to actually hold.
This function could also intertwine with another one: marking interpretive
use. Evidential participles would somehow show that the noun with which they
occur is not descriptively used or should not be thus used at the moment of speak-
ing or writing. If it was, its referent could in effect be considered responsible for
the state of affairs in question, or the state of affairs alluded to through that noun
would actually hold. What evidential participles indicate is that the nominal refer-
ent is or will be considered in a particular manner by certain people if certain
circumstances eventually or ultimately applied. To put it differently, evidential
participles indicate that the referent is considered or believed by certain people
to actually have certain characteristics or to likely have them in the (near) future
provided certain circumstances applied. The activation of vigilance mechanisms
surely depends on this marking of interpretive use, a dependence that future
research should certainly study in detail.
90 Manuel Padilla Cruz

References

Aikhenvald, A. (2004). Evidentiality. Oxford: Oxford University Press.

Arundale, R. B. (2006). Face as relational and interactional: A communication framework for
research on face, facework and politeness. Journal of Politeness Research. Language, Behav-
iour, Culture, 2(2), 193–216. ​https://‍doi.org/https://‍doi.org/10.1515/PR.2006.011
Austin, J. (1962). How to Do Things with Words. Oxford: Oxford University Press.
Blakemore, D. (1987). Semantic Constraints on Relevance. Oxford: Blackwell.
Blakemore, D. (1992). Understanding Utterances. Oxford: Blackwell.
Blakemore, D. (1993). The relevance of reformulations. Language and Literature, 2, 101–120. ​
https://‍doi.org/10.1177/096394709300200202
Blakemore, D. (1994). Relevance, poetic effects and social goals: A reply to Culpeper. Language
and Literature, 3, 49–59.  ​https://‍doi.org/10.1177/096394709400300104
Blakemore, D. (2002). Relevance and Linguistic Meaning. The Semantics and Pragmatics of Dis-
course Markers. Cambridge: Cambridge University Press. ​
https://‍doi.org/10.1017/CBO9780511486456
Blass, R. (1989). Grammaticalisation of interpretive use: The case of ré in Sissala. Lingua, 79,
299–326.  ​https://‍doi.org/10.1016/0024-3841(89)90060-0
Blass, R. (1990). Relevance Relations in Discourse. Cambridge: Cambridge University Press. ​
https://‍doi.org/10.1017/CBO9780511586293
Börjars, K., & Burridge, K. (2001). Introducing English Grammar. London: Arnold.
Brown, P., & Levinson, S. C. (1987). Politeness. Some Universals in Language Usage. Cambridge:
Cambridge University Press.  ​https://‍doi.org/10.1017/CBO9780511813085
Carston, R. (2002). Thoughts and Utterances. The Pragmatics of Explicit Communication. Oxford:
Blackwell.  ​https://‍doi.org/10.1002/9780470754603
Carston, R. (2016). The heterogeneity of procedural meaning. Lingua, 175–176, 154–166. ​
https://‍doi.org/10.1016/j.lingua.2015.12.010
Clément, F., Koenig, M., & Harris, P. (2004). The ontogenesis of trust. Mind & Language, 19(4),
360–379.  ​https://‍doi.org/10.1111/j.0268-1064.2004.00263.x
Collins, P., & Hollo, C. (2000). English Grammar: An Introduction. Basingstoke: Palgrave Mac-
millan.
Cornillie, B. (2007). Evidentiality and Epistemic Modality in Spanish (Semi-)Auxiliaries: A
Cognitive-functional Approach. Berlin: Mouton de Gruyter.
Corriveau, K., & Harris, P. (2009). Preschoolers continue to trust a more accurate informant 1
week after exposure to accuracy information. Developmental Science, 12(1), 188–193. ​
https://‍doi.org/10.1111/j.1467-7687.2008.00763.x
Crystal, D. (1991). A Dictionary of Linguistics and Phonetics. Oxford: Blackwell.
Culpeper, J. (1994). Why relevance theory does not explain ‘The relevance of reformulations’.
Language and Literature, 3, 43–48.  ​https://‍doi.org/10.1177/096394709400300103
Dendale, P., & Tasmowski, L. (2001). Introduction: Evidentiality and related notions. Journal of
Pragmatics, 33(3), 339–348.  ​https://‍doi.org/10.1016/S0378-2166(00)00005-9
Escandell Vidal, V. (1998). Politeness: A relevant issue for relevance theory. Revista Alicantina de
Estudios Ingleses, 11, 45–57.  ​https://‍doi.org/10.14198/raei.1998.11.05
Fricker, M. (2007). Epistemic Injustice. Power & the Ethics of Knowing. Oxford: Oxford University
Press.  ​https://‍doi.org/10.1093/acprof:oso/9780198237907.001.0001
 Chapter 3. Evidential participles and epistemic vigilance 91

Greenbaum, S., & Quirk, R. (1993). A Student’s Grammar of the English Language. London:
Longman.
Grice, H. P. (1975). Logic and conversation. In P. Cole, & J. L. Morgan (Eds.), Syntax and Seman-
tics. Vol. 3: Speech Acts (pp. 41–59). New York: Academic Press.
Haegeman, L. M. V., & Guéron, J. (1999). English Grammar: A Generative Perspective. Oxford:
Blackwell.
Haugh, M. (2014). Im/Politeness Implicatures. Berlin: Mouton de Gruyter.
Heyman, G. D. (2008). Children’s critical thinking when learning from others. Current Direc-
tions in Psychological Science, 17(5), 344–347. ​
https://‍doi.org/10.1111/j.1467-8721.2008.00603.x
Holmes, J., & Brown, D. F. (1987). Teachers and students learning about compliments. TESOL
Quarterly, 21, 523–546.  ​https://‍doi.org/10.2307/3586501
Huddleston, R. (1988). English Grammar: An Outline. Cambridge: Cambridge University Press. ​
https://‍doi.org/10.1017/CBO9781139166003
Huddleston, R. (1993). Introduction to the Grammar of English. Cambridge: Cambridge Univer-
sity Press.
Ifantidou, E. (1992). Sentential adverbs and relevance. UCL Working Papers in Linguistics, 4,
193–214.
Ifantidou, E. (1993). Parentheticals and relevance. UCL Working Papers in Linguistics, 5, 193–210.
Ifantidou, E. (2001). Evidentials and Relevance. Amsterdam: John Benjamins. ​
https://‍doi.org/10.1075/pbns.86
Imai, K. (1998). Intonation and relevance. In R. Carston, & S. Uchida (Eds.), Relevance Theory.
Applications and Implications (pp. 69–86). Amsterdam: John Benjamins. ​
https://‍doi.org/10.1075/pbns.37.06ima
Itani, R. (1994). A relevance-based analysis of hearsay particles: Japanese utterance-final tte.
UCL Working Papers in Linguistics, 6, 379–400.
Itani, R. (1998). A relevance-based analysis of hearsay particles: With special reference to
Japanese sentence-final particle tte. In R. Carston, & S. Uchida (Eds.), Relevance Theory.
Applications and Implications (pp. 47–68). Amsterdam: John Benjamins. ​
https://‍doi.org/10.1075/pbns.37.05ita
Jary, M. (1998). Relevance theory and the communication of politeness. Journal of Pragmatics,
30(1), 1–19.  ​https://‍doi.org/10.1016/S0378-2166(98)80005-2
Jary, M. (2013). Two types of implicatures: Material and behavioural. Mind & Language, 28(5),
638–660.  ​https://‍doi.org/10.1111/mila.12037
Jucker, A. H. (1993). The discourse marker well: A relevance-theoretical account. Journal of
Pragmatics 19(5), 435-452.  ​https://‍doi.org/https://‍doi.org/10.1016/0378-2166(93)90004-9
Koenig, M., & Harris, P. (2007). The basis of epistemic truth: Reliable testimony or reliable
sources? Episteme, 4(3), 264–284.  ​https://‍doi.org/10.3366/E1742360007000081
Levinson, S. C. (1983). Pragmatics. Cambridge: Cambridge University Press. ​
https://‍doi.org/10.1017/CBO9780511813313
Locher, M. A. (2006). Polite behaviour within relational work: The discursive approach to
politeness. Multilingua, Journal of Cross-Cultural and Interlanguage Communication, 25(3),
249–267.  ​https://‍doi.org/10.1515/MULTI.2006.015
Manes, J., & Wolfson, N. (1981). The compliment formula. In F. Coulmas (Ed.), Conversational
Routine. Explorations in Standardized Communication Situations and Prepatterned Speech
(pp. 115–132). The Hague: Mouton.
92 Manuel Padilla Cruz

Marandin, J.-M. (1987). Des mots et des actions : ‘Compliment’, ‘complimenter’ et l’action de
complimenter. Lexique, 5, 65–99.
Mascaro, O., & Sperber, D. (2009). The moral, epistemic, and mindreading components of
children’s vigilance towards deception. Cognition, 112(3), 367–380. ​
https://‍doi.org/10.1016/j.cognition.2009.05.012
Matthews, P. H. (2007). Oxford Concise Dictionary of Linguistics. 2nd edition. Oxford: Oxford
University Press.
Mazzarella, D. (2013). ‘Optimal relevance’ as a pragmatic criterion: The role of epistemic vigi-
lance. UCL Working Papers in Linguistics, 25, 20–45.
Mercier, H., & Sperber, D. (2011). Why do humans reason? Arguments for an argumentative
theory. Behavioral and Brain Sciences, 34(2), 57–111. ​
https://‍doi.org/10.1017/S0140525X10000968
Meyers, C. (Ed.). (2010). Journalism Ethics: A Philosophical Approach. Oxford: Oxford Univer-
sity Press.  ​https://‍doi.org/10.1093/acprof:oso/9780195370805.001.0001
Michaelian, K. (2013). The evolution of testimony: Receiver vigilance, speaker honesty and the
reliability of communication. Episteme, 10(1), 37–59.  ​https://‍doi.org/10.1017/epi.2013.2
Moeschler, J. (1993). Relevance and conversation. Lingua, 90(1-2), 149-171.
Norrick, N. (1980). The speech act of complimenting. In E. Hovdhaugen (Ed.), The Nordic
Languages and Modern Linguistics (pp. 296–304). Oslo: Universitetsforlaget.
Nuyts, J. (2006). Modality: Overview and linguistic issues. In W. Frawley (Ed.), The Expression
of Modality (pp. 1–26). Berlin: Mouton de Gruyter.
Origgi, G. (2013). Epistemic injustice and epistemic trust. Social Epistemology: A Journal of
Knowledge, Culture and Policy, 26(2), 221–235. ​
https://‍doi.org/10.1080/02691728.2011.652213
Oswald, S. (2011). From interpretation to consent: Arguments, beliefs and meaning. Discourse
Studies, 13(6), 806–814.  ​https://‍doi.org/10.1177/1461445611421360e
Oswald, S. (2016). Rhetoric and cognition: Pragmatic constraints on argument processing.
In M. Padilla Cruz (Ed.), Relevance Theory. Recent Development, Current Challenges and
Future Directions (pp. 261–285). Amsterdam: John Benjamins. ​
https://‍doi.org/10.1075/pbns.268.10osw
Padilla Cruz, M. (2009). Towards an alternative relevance-theoretic approach to interjections.
International Review of Pragmatics, 1(1), 182–206. ​
https://‍doi.org/10.1163/187731009X455884
Padilla Cruz, M. (2012). Epistemic vigilance, cautious optimism and sophisticated understand-
ing. Research in Language, 10(4), 365–386.  ​https://‍doi.org/10.2478/v10015-011-0040-y
Padilla Cruz, M. (2013). Understanding and overcoming pragmatic failure in intercultural com-
munication: From focus on speakers to focus on hearers. IRAL, International Review of
Applied Linguistics in Language Teaching, 51(1), 23–54. ​
https://‍doi.org/10.1515/iral-2013-0002
Padilla Cruz, M. (2015). On the role of vigilance in the interpretation of puns. Humor. Interna-
tional Journal of Humor Research, 28(3), 469–490. ​
https://‍doi.org/10.1515/humor-2015-0068
Palmer, F. (1986). Mood and modality. Cambridge: Cambridge University Press.
Palmer, F. (2001). Mood and Modality. 2nd edition. Cambridge: Cambridge University Press. ​
https://‍doi.org/10.1017/CBO9781139167178
 Chapter 3. Evidential participles and epistemic vigilance 93

Piskorska, A. (2016). Perlocutionary effects and relevance theory. In M. Padilla Cruz (Ed.),
Relevance Theory. Recent Developments, Current Challenges and Future Directions
(pp. 287–305). Amsterdam: John Benjamins.  ​https://‍doi.org/10.1075/pbns.268.11pis
Rich, C. (2015). Writing and Reporting News: A Coaching Method. Boston: Cengage Learning.
Rouchota, V. (1995). Discourse connectives: What do they link? UCL Working Papers in Lin-
guistics, 7, 199-212.
Searle, J. (1969). Speech Acts: An Essay in the Philosophy of Language. Cambridge: Cambridge
University Press.  ​https://‍doi.org/10.1017/CBO9781139173438
Spencer-Oatey, H. (Ed.) (2008). Culturally Speaking. Culture, Communication and Politeness.
London: Continuum.
Sperber, D. (1994). Understanding verbal understanding. In J. Khalfa (Ed.), What Is Intelligence?
(pp. 179-198). Cambridge: Cambridge University Press.
Sperber, D. (2013). Speakers are honest because hearers are vigilant. Reply to Kourken Michae-
lian. Episteme, 10(1), 61–71.  ​https://‍doi.org/10.1017/epi.2013.7
Sperber, D., & Mercier, H. (2012). Reasoning as a social competence. In H. Landemore, & J.
Elster (Eds.), Collective Wisdom: Principles and Mechanisms (pp. 368–392). Cambridge:
Cambridge University Press.  ​https://‍doi.org/10.1017/CBO9780511846427.016
Sperber, D., Clément, F., Heintz, C., Mascaro, O., Mercier, H., Origgi, O., & Wilson, D. (2010).
Epistemic vigilance. Mind and Language, 25(4), 359–393. ​
https://‍doi.org/10.1111/j.1468-0017.2010.01394.x
Sperber, D. & Wilson, D. (1995). Relevance. Communication and Cognition. 2nd ed. Oxford:
Blackwell.
Stovall, J. G. (2004). Journalism: Who, What, When, Where, Why, and How. Boston: Pearson.
Trotter, D. (1992). Analysing literary prose: The relevance of relevance theory. Lingua, 87(1-2),
11-27.  ​https://‍doi.org/https://‍doi.org/10.1016/0024-3841(92)90023-C
Unger, C. (2016). Evidentials, genre and epistemic vigilance. In M. Padilla Cruz (Ed.), Relevance
Theory. Recent Developments, Current Challenges and Future Directions (pp. 239–258).
Amsterdam: John Benjamins.  ​https://‍doi.org/10.1075/pbns.268.09ung
Wharton, T. (2003). Interjections, language and the ‘showing-saying’ continuum. Pragmatics
and Cognition, 11(1), 39–91.  ​https://‍doi.org/10.1075/pc.11.1.04wha
Wharton, T. (2009). Pragmatics and Non-verbal Communication. Cambridge: Cambridge Uni-
versity Press.  ​https://‍doi.org/10.1017/CBO9780511635649
Wilson, D. (1995). Is there a maxim of truthfulness? UCL Working Papers in Linguistics, 7,
197–212.
Wilson, D. (1999). Metarepresentation in linguistic communication. UCL Working Papers in
Linguistics, 11, 127-161.
Wilson, D. (2012). Modality and the conceptual-procedural distinction. In E. Wałaszewska, &
A. Piskorska (Eds.), Relevance Theory: More than Understanding, (pp. 23–44). Newcastle:
Cambridge Scholars Publishing.
Wilson, D. (2016). Reassessing the conceptual-procedural distinction. Lingua, 175–176, 5–19.​
https://‍doi.org/10.1016/j.lingua.2015.12.005
Wilson, D., & Sperber, D. (1993). Linguistic form and relevance. Lingua, 90(1–2), 1–25. ​
https://‍doi.org/10.1016/0024-3841(93)90058-5
Wilson, D., & Sperber, D. (2002a). Relevance theory. UCL Working Papers in Linguistics, 14,
249–287.
Wilson, D., & Sperber, D. (2002b). Truthfulness and relevance. Mind, 111(443), 583–632. ​
https://‍doi.org/10.1093/mind/111.443.583
94 Manuel Padilla Cruz

Wilson, D., & Sperber, D. (2004). Relevance theory. In L. Horn, & G. Ward (Eds.), The Handbook
of Pragmatics (pp. 607–632). Oxford: Blackwell.
Wilson, D., & Wharton, T. (2006). Relevance and prosody. Journal of Pragmatics, 38(10),
1559–1579.  ​https://‍doi.org/10.1016/j.pragma.2005.04.012
Wolfson, N. (1983). An empirically based analysis of complimenting in American English. In N.
Wolfson, & E. Judd (Eds.), Sociolinguistics and Language Acquisition (pp. 82–95). Rowley:
Newbury House.
Wolfson, N., & Manes, J. (1980). The compliment as a social strategy. Papers in Linguistics, 13(3),
391–410.  ​https://‍doi.org/10.1080/08351818009370503
Yus Ramos, F. (1999a). Towards a pragmatic taxonomy of misunderstandings. Revista Canaria
de Estudios Ingleses, 38, 217–239.
Yus Ramos, F. (1999b). Misunderstandings and explicit/implicit communication. Pragmatics,
9(4), 487–517.  ​https://‍doi.org/10.1075/prag.9.4.01yus
 Chapter 4

The Greek connective gar

Different genres, different effects?

Sarah Casson1,2
1St. Mellitus College, UK / 2SIL International

Traditional accounts of the Greek connective gar offer an inadequate causal

definition, while multiplying descriptive categories. Linguistic accounts have
paid little attention to the role of gar in different kinds of communication. This
article proposes a relevance-theoretic procedural explanation which provides a
unifying cognitive explanation for gar’s communicative role. Gar indicates that
the premise it introduces is relevant in relation to a previously communicated
claim, which, when combined with implicit assumptions, it strengthens. This
strengthening may work itself out differently in diverse kinds of communication:
typically, as confirmation in argumentation, and explanation in narrative.
Perceived differences in stylistic effects sometimes associated with gar are
attributable not to multiple functions, but to recognition of differing communi-
cator goals in narrative and argumentation.

Keywords: cognitive effect of strengthening, persuasion, connectives, gar,

Koine Greek, argumentation, narrative, procedural marker, New Testament,
epistemic vigilance

1. Introduction

This paper will consider the question of whether the Greek connective gar con-
tributes to different stylistic effects in different kinds of communication, such as
argumentation and narrative. Gar is found in both Classical and Koine Greek with
a similar range of uses, but I will concentrate on its role in the Koine Greek of New
Testament texts. I will build on Blass’s (1998) tentative analysis of gar as a proce-
dural marker indicating the inferential procedure of “strengthening” and consider
whether this is adequate to explain its varied uses in argumentation and narrative
material. I will explore the suggestion that gar may trigger the proposed argu-
mentative module and target epistemic vigilance mechanisms. I will suggest an

https://‍doi.org/10.1075/ftl.8.04cas
© 2020 John Benjamins Publishing Company
96 Sarah Casson

understanding of the connective’s varied role in terms of differing communicator

goals and the expectations of relevance raised by different types of communication.

2. Traditional and linguistic accounts of gar

The meaning or function of gar has long preoccupied interpreters of both Classical
Greek texts and the New Testament. Typically it has traditionally been glossed in
English as for, and analysed by Greek scholars such as Denniston (1934, p. 58)
as having primarily an “explanatory” and a “causal” function, but with various
other uses in diverse contexts (Denniston, 1934, p. xv). The third edition of Bauer,
Danker, Arndt and Gingrich (BDAG) (2000, p. 189–190), a standard English
reference lexicon for New Testament Greek, lists the following definitions:
γάρ …conjuction used to express cause, clarification or inference…1. marker of
cause or reason, for…. b. used with other particles and conjunctions…. c. γάρ
is sometimes repeated. It occurs twice either to introduce several arguments for
the same assertion or to have one clause confirm the other…. d. the general is
confirmed by the specific…. e. often the thought to be supported is not expressed,
but must be supplied from the context…. f. often in questions…. 2. marker of
clarification, for, you see…3. marker of inference…

Such traditional explanations do not adequately account for the function of gar,
tending to multiply descriptive categories. Some, however, such as BDAG’s defini-
tions 2e. and 3. above, do recognise that information introduced by the connective
seems to be associated with implicitly communicated assumptions.
In recent years, some linguists have proposed explanations of gar which seek
to identify a single function for the connective. Some scholars, taking a discourse
analytic approach to Classical Greek, regard gar as a “push” particle which “pushes
[or subordinates] the utterance [that it introduces] to a lower level than the pre-
ceding discourse” (Slings, 1997, cited by Bakker, 2009, p. 41). Similarly, Levinsohn
and Runge’s discourse analytic approach to New Testament Greek explains gar
as introducing “background material which strengthens some aspect of what
has just been presented” (Levinsohn, 2000, p. 90), rather than advancing the
discourse (Runge, 2010, p. 52). These explanations focus on gar’s role in informa-
tion structure but do not consider its possible relationship to cognitive processes.
Meanwhile, Black (2002), and most recently Zakowski (2016), combine discourse
analytic and functional perspectives with the notion of procedural meaning in
their exploration of gar in certain narrative passages in the New Testament. This
allows both, in different ways, to consider not only the discourse function of gar,
but also to touch on its cognitive role.
 Chapter 4. The Greek connective gar 97

3. Argumentation and narrative

3.1 Defining argumentation and narrative

Neither traditional accounts of gar nor more recent linguistic analyses pay par-
ticular attention to differences in the connective’s use in argumentation in contrast
to its use in narrative. The notions of argumentation and narrative are, of course,
very general and understood in diverse ways by scholars in different disciplines.
They touch on the much debated issue of literary genre, which is beyond the scope
of this paper. Sperber (2001, p. 401) discusses Goldman’s (1999, p. vii) distinction
between the social practices of “testimony, i.e. the transmission of observed (or al-
legedly observed) information from one person to others, and argumentation, i.e.
the defense of some conclusion by appeal to a set of premises that provide support
for it.” Blass (2005, p. 171) explains argumentation in terms of a communicator’s
aim of gaining the trust of her addressee: “Both communicator and addressee
know that the acceptance of testimony depends on trust. In order to gain the trust
of the addressee the communicator often engages in argumentation.”
I will use this generic distinction between testimony and argumentation to
inform my definition of narrative and of argumentation. By narrative, I mean
communication in which the communicator intends to inform the addressee of
“observed information” in the form of a story. By story, I mean a series of causally
connected events, with participants, a problem, a development and an outcome. I
would like to suggest that in the case of narrative there is typically an assumption
that the addressee basically trusts the communicator and accepts her testimony:
its trustworthiness is not an issue. In other words, in narrative, the addressee tends
to “accept the authority” of the communicator: see Sperber et al.’s, claim (2010,
p. 366) that “If the addressee accepts the (epistemic or practical) authority of the
communicator, recognition of the informative intention will lead to its fulfilment,
and hence to the production of the appropriate cognitive or behavioural response.”
As a result, in narrative we typically find little evidence of persuasive techniques.
By argumentation, on the other hand, I mean communication in which the
communicator intends to defend certain conclusions or claims via premises that
show the rationality and thus the believability of those claims. In argumentation,
the speaker’s overt aim is to persuade the addressee of the truth or rationality of
her claims by using logical argument and associated rhetorical forms. She aims to
gain the addressee’s trust in relation to her own trustworthiness as a source and in
relation to the rationality and coherence of the content of her communication (see
Sperber et al., 2010, 369f and 374f), since the addressee may not be convinced of
these at the outset. In order to do so she may employ persuasive rhetorical displays.
I suggest that argumentation is thus an example of communication in which the
98 Sarah Casson

speaker is aware that there is a real possibility that “the communicative intention
[may] be fulfilled without the corresponding informative intention… in other
words, an audience [may] correctly understand an utterance without accepting or
complying with what they have understood.” (Sperber et al., 2010, p. 366).
In this discussion I will also refer to hortatory material found in the New
Testament. By this, I mean communication in which the communicator aims to
persuade the addressee of the desirability of certain actions/states of affairs by us-
ing exhortations and explanatory or illustrative material, and associated rhetorical
forms. I understand hortatory material to be an example of the broader type of
argumentation as defined above. In the case of hortatory material, the addressee’s
confidence in the communicator’s trustworthiness may or may not be high, but
there will be other reasons why the addressee may initially be reluctant to ac-
cept what is proposed as desirable. As a result, the communicator chooses to use
persuasive techniques.
From the point of view of the addressee, there are important differences in the
expectations of relevance raised by argumentation and hortatory texts on the one
hand, and by narrative on the other. In the case of argumentation and hortatory
discourse, the addressee expects the communication to be relevant by persuading
him of the truth of certain assertions, or of the desirability of certain actions or
states of affairs. In the case of narrative, on the other hand, the addressee expects
the communication to be relevant as a compelling story, evoking emotion and
satisfying curiosity, a story which is plausible, but not necessarily true in the real
world. In other words, claims of truth and rationality are not primarily at stake in
narrative. I would like to suggest that this means that the addressee’s epistemic
vigilance mechanisms are on a lower level of alert and less likely to be activated
when processing narrative (see the fuller discussion of epistemic vigilance below).

3.2 Gar in argumentation and narrative

Despite the lack of interest traditionally shown in the relationship between types
of communication and the varied use of gar, the apparently differing function of
the connective in these two types of texts is noteworthy. In argumentation, gar
very often introduces evidence which serves as proof or a rationale for a claim,
contributing to the persuasiveness of the communication. It is sometimes used
in a chain of successive justificatory or explanatory premises in tight logical se-
quences of argumentation, each clause introduced by gar serving to strengthen the
preceding assertion. In narrative, on the other hand, gar very often introduces an
explanation giving a reason for an event or action. It can also be used to introduce
a detailed narrative after an initial global summary, orientation or question. De
Jong (1997, p.177), who considers gar in Classical texts from a discourse analytic
 Chapter 4. The Greek connective gar 99

perspective, claims that on occasion the connective is employed in “the narrative

device of the epic regression,” introducing a flashback or successive steps back in
time anterior to the principal time of reference of the narrative. And certain New
Testament scholars such as Thrall (1962, p. 47–48) have sometimes associated the
use of gar in the narrative of Mark’s Gospel with a style that is “unsophisticated,”
or “less logical” than others.

4. A procedural account of gar

4.1 Blass’s proposed account of gar

Blass (1998, p.7), applying a relevance-theoretic analysis to the use of the connec-
tive in the Pauline epistles of the New Testament, puts forward the hypothesis that
gar functions as a procedural marker, encoding procedural instructions which
impose constraints on relevance. In her view, gar constrains the addressee’s search
for relevance by indicating a premise which strengthens, or “backwards confirms,”
an existing assumption. While this explanation of the connective offers an insight-
ful, unified account of uses of gar in New Testament argumentation, it does not
consider the role of the connective in New Testament narrative. We need to ask
whether this strengthening account of gar is adequate to explain occurrences of
the connective in narrative, or whether a procedural analysis of the connective
needs to be adjusted to take them into account.

4.2 Examples of gar in argumentation and hortatory material

Let us look more closely at some varied examples of gar in the New Testament. Gar
is found considerably more frequently in non-narrative texts, such as Paul’s letter
to the Romans (principally a closely argued theological exposition containing
some hortatory passages also), than in narrative. According to Runge’s statistical
analysis (2010, p. 52), of 1,041 occurrences of gar in the New Testament, only 10%
are to be found in narrative proper. In addition to expository argumentation, the
connective is also more commonly found in hortatory material (such as Jesus’
Sermon on the Mount), in which, as in argumentation, the communicator seeks
to persuade the addressee of the truth of assertions that will be useful or desirable.
Typically, as Blass (1998) has shown, in New Testament argumentation and
hortatory texts, gar introduces a premise or supporting evidence that strengthens
an assertion or exhortation that has just been communicated, by allowing it to be
deduced independently in combination with other implicit background assump-
tions held by the addressee.
100 Sarah Casson

(1) Romans 2:1

v.1a ASSERTION:
Dio anapologētos ei ō anthrōpe pas ho krinōn
Therefore inexcusable you.are O person every the.one judging
‘Therefore you are without excuse, you, whoever you are, who judge.’

When further disambiguated, this can be put something like: “Therefore you are
guilty, you…who judge others.”
v.1bPREMISE:
en ō gar krineis ton heteron seauton katakrineis
in which gar you.judge the other yourself you.condemn
‘gar you condemn yourself for whatever you judge others for.’
Gar in v.1b introduces a premise which represents part of the evidence for the truth
of the claim in v.1a that the person who judges others is without excuse. This prem-
ise, when combined with accessible background assumptions from the addressee’s
encyclopaedic knowledge in a series of inferences, yields the assertion in v.1a as a
conclusion. We might represent the inferences involved in the following way:
EXPLICIT ASSUMPTION stated in v.1b (premise 1):
You, person who judges, condemn yourself for whatever you judge others for.
CONTEXTUAL ASSUMPTION from background knowledge of the world
(premise 2):
If a judge is condemned for what he judges others for, he is as guilty as them, i.e.
without excuse.
CONCLUSION (confirming assertion in v.1a):
Therefore you, person who judges, are without excuse.
The assertion in v.1a is thus independently confirmed as true and shown to be ra-
tional by a series of inferences derived from the assertion in v.1b, plus background
assumptions already held as true by the addressee. As a result, the assertion made
in v.1a is more strongly accepted by the addressee, that is, the cognitive effect of
strengthening takes place.
(2) Romans 13:8
v.8a-b EXHORTATION:
v.8a Mēdeni mēden ofeilete
to.no-one nothing owe
v.8b ei mē to allēlous agapan
if not the one.another love
‘Owe no-one anything except (the debt of) love for one another.’
 Chapter 4. The Greek connective gar 101

We might express the disambiguated and enriched form of the assertion in v.8a-b
as: “It is desirable not to have any debts but to love one another.”
v.8c PREMISE:
ho gar agapōn ton hereton nomon peplērōken.
the.one gar loving the other law has.fulfilled
‘gar anyone who loves other people fulfils (the requirements of) the
(Jewish) law.’
Gar in v.8c introduces a premise which represents part of the evidence for the
truth that the immediately preceding exhortation is desirable. This premise, when
taken together with contextual assumptions from the audience’s encyclopaedic
knowledge, backs up the exhortation, increasing the strength with which it, as
a desirable state of affairs, is accepted by the addressee. Gar gives procedural in-
structions that what follows in v.8c should be taken as a premise in the inferential
procedure of strengthening a preceding claim. The procedure follows a relevance-
driven heuristic to arrive at an interpretation which leads to the strengthening of
the most relevant previous assumption, which in this case is the immediately pre-
ceding utterance: It is desirable…to love one another. The immediately preceding
co-text (Romans 13: 7) exhorts addressees to acquit themselves of all obligations,
taxes, respect and honour, and the first part of the exhortation, v.8a, “Owe no-one
anything,” is then a summary of this. Given this co-text, it is the second part of the
exhortation, v.8b, “love one another,” which is salient as a new piece of information
which needs further strengthening. We can represent the sequence of inferences
something like this:
EXPLICIT ASSUMPTION stated in v.8c: (premise 1): Anyone who loves other
people fulfils the (Jewish) law (by loving other people).
CONTEXTUAL ASSUMPTION (premise 2) from background knowledge: If
someone fulfils the Jewish law by doing an action, it is desirable to do that action.
CONCLUSION (confirmation of desirability of exhortation in v.8b): It is
desirable to love other people (one another).
The result of the inferential procedure is that the preceding communicated as-
sumption “It is desirable… to love one another” is held more strongly by the
addressee (that is, he is more persuaded of its truth) because the addressee has
independently verified its reasonableness and logical coherence, accessing exist-
ing assumptions that help to confirm it. He has checked it against his existing
knowledge of the world.
The subsequent utterance in the next verse is also introduced by gar and like-
wise leads to the confirmation or strengthening of the preceding assertion.
102 Sarah Casson

(3) Romans 13:9

to gar Ou moicheuseis Ou foneuseis Ou klepseis
the gar not you.will.commit.adultery not you.will.murder not you.will.steal
Ouk epithymēseis kai ei tis hetera entolē en tō logō toutō
not you.will.covet and if some other commandment in the saying this
anakefalaioutai en tō Agapēseis ton plēsion sou hōs seauton.
is.summed.up in the You.will.love the neighbour of.you as yourself.
‘Gar the commandments ‘Do not commit adultery’, ‘Do not murder’, ‘Do not
steal’, ‘Do not covet’, and any other commandment there may be, is summed
up in this one command: ‘Love your neighbour as yourself.’’

When the addressees hear v.9, gar triggers the inferential procedure of strengthen-
ing. This procedure, following a relevance-guided heuristic, takes the complex ex-
plicit assertion in v.9 as a premise which, in combination with existing background
assumptions, confirms the first preceding assertion it finds which is compatible
with the principle of relevance. The procedure searches for accessible background
assumptions to combine in an inferential series that independently confirms the
truth of the preceding utterance, “Love is the fulfilling of the law.” The inferential
procedure might look something like this:
EXPLICIT ASSUMPTION stated in v.9 (premise 1): All the commandments
(of the Jewish law) are summed up in the command: “Love your neighbour as
yourself.”
CONTEXTUAL ASSUMPTION from background knowledge (premise 2): If
all the commandments of the law are summed up in the command: “Love your
neighbour as yourself,” then if a person loves her neighbour (other people),
she fulfils the law.
CONCLUSION (confirmation of truth of assertion in v.8b): Love (of other
people/one’s neighbour) is the fulfilling of the law.
As a result of the procedure triggered by gar, the addressee is persuaded to hold
more strongly as true the claim that “Love is the fulfilling of the law.”
In all the above examples, then, gar encodes instructions to follow a procedure
that combines the utterance that gar introduces with existing background assump-
tions to increase the strength with which a previously communicated assumption
is held by the addressee. In this way, gar guides the addressee to find increased rel-
evance, or additional cognitive effects, from a preceding utterance. As Blass puts it,
the connective is a device which guides and constrains the addressees’ inferential
processes (1998, p. 2) by indicating “how the proposition it prefaces contributes
to relevance” (1998, p. 9). It indicates that the addressee should expect that the ut-
terance it introduces contributes to the relevance of the wider communication not
 Chapter 4. The Greek connective gar 103

in its own right by leading to the derivation of a fresh implication or conclusion,

but rather in relation to a previously communicated assumption, increasing the
strength with which the latter is held. We can say that gar thus reduces processing
costs by guiding the addressee to process the utterance it introduces in relation to
the preceding utterance and in the context created by it, to “look backwards,” as
it were, rather than searching for the relevance of the utterance introduced by gar
as a new implication with a fresh set of contextual assumptions. The information
gar introduces contributes indirectly to relevance by making further contextual
assumptions available within the context already created by the previous utter-
ance. These lead to the derivation of further cognitive effects from this preceding
assertion; more precisely, to the cognitive effect of strengthening its believability.
In this sense, gar might be understood as introducing background rather than
foreground information (cf. Sperber & Wilson, 1995, p. 217 and Unger, 2002, p. 7).
We might also say, perhaps, that gar is an indication that there are more cogni-
tive effects to be derived from a preceding assertion, and that the communicator
suspects that the addressee’s expectations of relevance from that assertion are not
yet fully satisfied and the assertion is consequently not yet firmly or fully held as
true by the addressee. This would be the case if an addressee is initially sceptical
about the believability or truth of a particular assertion, or cannot fully make sense
of how the assertion fits with his existing knowledge of the world, before the com-
munication of the premise introduced by gar.

4.3 Gar in narrative texts

When we examine gar in narrative in the New Testament, we find some overlap
with its use in argumentation and persuasive communication, but also examples
which at first glance seem to differ markedly from this. While considerably less
common than in argumentation, gar is sometimes used in narrative to introduce
explanatory detail which helps make a previous assertion (usually a description of
an action or event) more understandable within the narrative.
(4) Matthew 19:22
v.22a assertion:
akousas de ho neaniskos ton logon apēlthen
having.heard PM1 the young.man the word he.went.away
lypoumenos
grieving
‘When he heard this the young man went away sad’

1. procedural marker.
104 Sarah Casson

v.22b explanation:
ēn gar echōn ktēmata polla
was gar having possessions many.
‘Gar he had lots of possessions.’

The literary context of this example is an encounter between Jesus and a young
man who asks a question about how to obtain eternal life. In the preceding text in
v.21a Jesus has answered the question by saying that the young man should sell all
his possessions and then follow him. In v. 22a we are presented with the assertion
that the young man went away sad, an assertion which on its own might initially
be difficult to make sense of, given the fact that the young man had achieved his
goal of meeting Jesus and had received an answer from him and an invitation to
join his disciples. This assertion may thus raise questions in the addressee’s mind
such as “Why did the young man go away sad?” In order for this assertion to be
optimally relevant, connecting to the addressee’s existing understanding of the
world and leading to adequate cognitive effects, a further explanation is needed.
This is provided by the assertion following gar in v.22b: “He had lots of posses-
sions.” The series of inferences involved here can be represented as follows:
EXPLICIT ASSERTION stated in v.22a: The young man went away sad.
IMPLICIT QUESTION raised by v.22a: Why did the young man go away sad?
EXPLICIT ASSUMPTION stated in v.22b: Gar the young man had lots of pos-
sessions.
CONTEXTUAL ASSUMPTION from background knowledge: If people have
lots of possessions, they do not want to give them away.
CONTEXTUAL ASSUMPTION from background knowledge:
If someone tells someone else with lots of possessions to give them away, the
person with lots of possessions will be unhappy.
CONTEXTUAL ASSUMPTION from preceding text (v.21):
Jesus told the young man to give his possessions away.
CONCLUSION (confirmation of plausibility of assertion in v.22a):
(This is why) The young man went away sad.
We can explain the elements involved in the inferential process here in the follow-
ing way. The explicit assertion in v.22a raises a question in the addressee’s mind:
“Why did the man go away sad?” Gar then gives the procedural instruction that
what follows in v.22b should be taken as part of an explanation of a preceding
assertion which increases its relevance. Following a relevance-guided heuristic,
the addressee’s comprehension module identifies the assertion in v.22a as the one
for which v.22b is to provide an explanation. The module searches for additional
accessible contextual assumptions to combine with the assertion in v.22b, for
 Chapter 4. The Greek connective gar 105

example, “If people have lots of possessions, they do not want to give them away”
and “If someone tells someone else with lots of possessions to give them away, the
person with lots of possessions will be unhappy.” When these contextual assump-
tions are combined inferentially with the explicit assumption in v.22b, they answer
the implicit question, that is, they satisfy the expectations of relevance raised by
the assertion in v.22a, “The young man went away sad.” In this way the cognitive
effects derived from this assertion are increased: it becomes more understandable,
that is, better connected to the addressee’s existing knowledge of the world, and
thus more strongly accepted as plausible within the world of the narrative. This in
turn makes the story more interesting and memorable, in short, more relevant. We
have seen how a premise introduced by gar that confirms a preceding assertion
might be viewed by some as background information. In this narrative example,
too, gar might be understood as introducing background information which sup-
plies explanatory detail for the events of the narrative, grounding the events of the
narrative in a plausible narrative world, making sense of them by connecting them
to existing assumptions about the world.
From this initial example of gar in a narrative text, we can suggest that, just
as in argumentation, gar encodes procedural instructions. These trigger an in-
ferential procedure which combines the utterance following gar with contextual
assumptions so as to increase the cognitive effects of a preceding assertion by
increasing the strength with which it is held. Whereas in argumentation, gar often
introduces a premise that serves as evidence or proof for a preceding assertion, in
this narrative example gar introduces information that provides an explanation
for a preceding assertion that perhaps raises questions. In both cases, the result
of the procedure triggered by gar is to increase the cognitive effects derived from
a preceding assertion; the assertion introduced by gar is not relevant in its own
right, but in relation to something that precedes it. As Blakemore suggests in her
discussion of the connective “after all” (1987, p. 90), “the relevance of a proposi-
tion introduced by [the connective] is created by the presentation of the preceding
proposition.” In the examples considered so far, whether in argumentation or
narrative material, gar reduces processing costs by guiding to where relevance
is to be found, and indicating what kind of cognitive effects are to be expected,
namely strengthening.
The example discussed above of gar in a narrative text seems compatible with
a procedural strengthening account of the connective such as Blass’s. Other nar-
rative uses of gar are at first sight less easily explained in terms of strengthening,
however. In her article “ΓAΡ Introducing Embedded Narratives” (1997), De Jong
discusses the way the connective may be used in certain contexts in Classical
Greek to introduce a span of narration after an initial announcement or summary
of events, or a summons to listen. She shows how it may also be employed in a
106 Sarah Casson

series of successive steps further back into the past within a narrative. She cites an
example from the beginning of Homer’s Iliad (1997, p.177) to illustrate this use,
which she claims is part of the narrative device of the epic regression. Interestingly,
De Jong sees gar in such cases as introducing “background information.” In her
view, such a use of the connective originated in a “typically archaic (oral?) form of
narration, viz. announcing an event and then going back in time and filling in the
details of how this event came about.” (1997, p.179).
In the narrative of Mark’s Gospel we also find certain occurrences of gar where
the connective apparently introduces successive steps back in time. Several ex-
amples are found in the account of John the Baptist’s death at the hands of Herod.
(5) Mark 6:16–29
v.16 akouas de ho Hērōdēs elegen Hon egō apekefalisa
having.heard PM the Herod he.was.saying The.one I beheaded
Iōannēn houtos ēgerthē.
John this.one was.raised
‘When he heard this [that Jesus and his disciples were doing miracles],
Herod said, ‘John, the one I beheaded, has been raised from the dead.’’
v.17 Autos gar ho Hērōdēs aposteilas ekratēsen ton Iōannēn kai edēsen
Himself gar the Herod having.sent arrested the John and bound
auton en fylakē dia Hērōdiada tēn gynaika Filippou tou
him in prison on.account.of Herodias the wife of.Philip the
adelfou autou hoti autēn egamēsen.
brother of.him because her he.married
‘Gar Herod himself had given orders that John should be arrested,
bound and put in prison on account of Herodias, the wife of his brother
Philip, because he, Herod, had married her.’
v.18 elegen gar ho Iōannēs tō Hērōdē hoti Ouk exestin soi
he.was.saying gar the John to.the Herod that not it.is.lawful for.you
echein tēn gynaika tou adelfou sou.
to.have the wife of.the brother your
‘Gar John had been saying to Herod, ‘It is not lawful for you to have
your brother’s wife.’’

In v.16 an event is described as part of the wider narrative of Mark’s gospel: in

response to the reports of Jesus’ miracles, Herod declares that John the Baptist has
been raised from the dead. An explanatory narrative then follows in verses 17–29,
which steps or flashes back to a time anterior to the main narrative in vv.14–16 and
the preceding wider narrative. At first sight one might be tempted to analyse gar as
a signal indicating the switch in time, since the verb forms of the anterior events in
 Chapter 4. The Greek connective gar 107

vv.17–18 do not indicate anterior time: they are in the aorist and imperfect tenses,
as are the verbs of the main events of the narrative in vv.14, 15 etc.
If we look more closely at this use of gar, however, we can see that, as in
Example (3) above, where gar introduces an explanation, the connective intro-
duces information which serves as explanatory detail, encouraging access to ad-
ditional contextual assumptions which, together with the explicit information in
v.17, increase the relevance of the preceding assertion. Herod’s declaration in v.16
raises important implicit questions for the addressee, firstly “How/when/why did
Herod behead John?” and secondly “Why did Herod declare that John, whom he
had beheaded, was raised from the dead?” This sudden reference to John’s behead-
ing is the first indication that has been given in the wider narrative that John is
dead. Previously there have been direct references to John in Chapter 1, where he
preaches in the desert and baptises Jesus, and an indirect reference in Chapter 2
v.18 when John’s disciples are referred to. The implication there is that John is still
alive. The information given in Chapter 6 v.16 thus comes as something of a shock.
The questions raised by this information must be answered in order for Herod’s
declaration to achieve optimal relevance and to be accepted as comprehensible
and plausible within the narrative.
The first step back in time in v.17 (“gar Herod had given orders that John
should be … put in prison on account of Herodias… because he had married her”)
makes accessible certain related contextual assumptions, such as:
Herod had had a connection with John the Baptist.
Herod had imprisoned (and killed – v.16) John for personal reasons.
Herod had treated John unjustly.
Herod was afraid of John the Baptist, etc.

But although these or similar contextual assumptions are made accessible by the
information in v.17, they do not allow the deduction of a complete series of infer-
ences which yields the assertion in v. 16 as a conclusion, increasing its cognitive
effects so that the expectations of relevance raised by this assertion are satisfied.
The implicit questions raised by v.16, “How/when/why did Herod kill John?” and
“Why did Herod declare that John the Baptist was raised from the dead?” remain
unanswered, although an addressee might hazard some guesses regarding their an-
swers. This, might, however, require unreasonable processing costs. Consequently,
even after v.17 has been processed, there remains a gap in comprehension in
relation to v.16: the addressee is, as yet, not fully convinced of the plausibility of
the assertions in that verse, and it is not yet adequately relevant. As a result, as
the story continues to unfold in the following verses, the addressee continues to
search for assumptions that will answer the implicit questions raised by v.16 and
for ways in which the expectation of relevance raised by that verse will be satisfied.
108 Sarah Casson

This continuing search for relevance contributes to narrative suspense and makes
the story more compelling. In this example, then, gar appears to guide towards an
explanation that will increase the relevance of the preceding assertion, but no fully
satisfactory, adequately relevant explanation is derived. Nevertheless, the effect of
the procedure triggered by gar in v.17 seems to be to guide the addressee to search
for relevance from the assertions in v.17 in relation to the preceding assertion in
v.16, namely in relation to Herod and his attitudes and reactions, rather than to
take these assertions as new implications to be processed in a fresh context.
The assertions in v.17 in turn raise an implicit question: “Why did Herod throw
John into prison etc. on account of his wife Herodias?” The further step back in
time in v.18, “John had been saying to Herod, ‘It is not lawful for you to have
your brother’s wife,’” which is again introduced by gar, then answers this further
question. It does this by giving details of events anterior to those described in v.17,
which make other contextual assumptions accessible. The latter can be combined
in an inferential series to arrive satisfactorily at an explanation of the assertion in
v.17. These contextual assumptions might include, for example:
If a subject criticises a king’s marriage, the king will be angry with the subject.
Herod was angry with John because of John’s criticism of his marriage to
Herodias.
Herod wanted to stop John criticising his behaviour/to punish John.
If a king puts a subject in prison, the subject will be stopped from criticising the
king/ the subject will be punished.
EXPLANATION OF ASSERTION in v.17: (Therefore) Herod had ordered
John to be put in prison.

The assertion in v.18 thus increases the relevance of the assertion in v.17 by
making new contextual assumptions accessible from encyclopaedic knowledge,
which lead to increased cognitive effects derived from the information in v.17.
In particular, the assumptions communicated in v.17 are more comprehensible,
and more strongly accepted as plausible, because of the information in v.18. But
questions remain about the plausibility of the information communicated in v.16,
and its comprehensibility within the wider narrative. The question of how/when/
where Herod had killed John has still not been fully answered and the addressee
is still looking for an answer to this, expecting the subsequent narrative to be rel-
evant in relation to this. V.19 describes Herodias’ attitude towards John: “Herodias
hated him and wanted to kill him, but she was not able to.” This information is not
introduced by gar but by the connective de. We can suggest that this is because the
assertion about Herodias is relevant as a new implication, rather than providing
additional information which increases the relevance of the preceding assump-
tion. This new implication needs to be processed in the context of a new set of
 Chapter 4. The Greek connective gar 109

contextual assumptions i.e. assumptions relating to Herodias and her motivations

rather than to the previous context relating to Herod and his motivations.
In v.20, on the other hand, gar again introduces an assertion: “Herod was afraid
of John, knowing that he was a righteous and holy man, and protected him, and
was greatly disturbed when he heard John, but he liked to listen to him.” Here, gar
indicates that what follows, namely the assertions “Herod was afraid of John…and
protected him” should be processed in combination with additional contextual
assumptions which together yield an explanation of the preceding assertion in
v.19b: “Herodias was not able to kill John.” This explanation helps to increase the
relevance of this assertion, answering an implicit question that might be raised in
the addressee’s mind, namely, “Why was Herodias not able to kill John?” Once
again, the procedural instructions given by gar have the effect of increasing the
cognitive effects of the preceding assertion, making better sense of it by explaining
the reason for it, thus increasing the strength with which it is held by the addressee
as plausible in the narrative world.
In vv.21–28 the addressee is presented with the sequence of events which led
to John the Baptist’s beheading. This sequence is not introduced by gar, presum-
ably because these events are to be interpreted as relevant in their own right,
each yielding new implications of their own, rather than increasing the relevance
of a preceding assertion by explaining it further. Nevertheless, as the addressee
processes this sequence of events, he still has in his mind the unanswered implicit
questions of v.16. The expectation that the story will provide an answer to these
questions, thus satisfying the expectation of relevance raised by v.16, creates
suspense and interest on the part of the addressee, and encourages him to access
contextual assumptions related to these questions. For example, the information in
v.20 that Herod was afraid of John because he was a righteous and holy man, and
the information in v.26 that Herod was greatly distressed, encourage addressees
to access contextual assumptions relating to Herod’s fear of John’s supernatural
power, to his religious scruples and to his troubled conscience, all of which might
imply a susceptibility to believing that John had been raised from the dead. The
information in vv.27 and 28 describing how John was beheaded finally fully satis-
fies the addressee’s expectation that the narrative will answer the questions “How/
when/why did Herod kill John?” raised in v.16.
The story reaches its climax as follows:
110 Sarah Casson

v.25 kai eiselthousa euthys meta spoudēs pros ton basilea ētēsato
and having.entered immediately with haste to the king she.asked
legousa Thelō hina exautēs dōs moi epi pinaki tēn kefalēn
saying I.want PM at.once you.may.give to.me on platter the head
Iōannou tou baptistou.
of.John the Baptist.
‘And immediately she (the daughter of Herodias) rushed to the king and
asked him, ‘I want you right now to give me the head of John the Baptist
on a platter.’’

v.26 kai perilypos genomenos ho basileus dia tous horkous kai

and very.sad having.become the king on.account.of the oaths and
tous anakeimenou ouk ethelēsen athetēsai autēn.
the guests not he.wanted to.refuse her.
‘And although greatly distressed, the king did not want to refuse her
because of his oaths and his dinner guests.’

v.27 kai euthys aposteilas ho basileus spekoulatora epetaxen enenkai

and immediately having.sent the king executioner he.ordered to.bring
tēn kefalēn autou. Kai apelthōn apekefalisen auton en tē fylakē.
the head his. And having.gone he.beheaded him in the prison.
‘So he immediately sent an executioner with orders to bring his (John’s)
head, who went and beheaded him in the prison.’

v.28 kai ēnenken tēn kefalēn autou epi pinaki kai edōken autēn tō
and he.brought the head of.him on platter and he.gave it to.the
korasiw kai to korasion edōken autēn te mētri autēs.
girl and the girl gave it to.the mother of.her.
And he brought his head on a platter and gave it to the girl, and the girl
gave it to her mother.’

Although the assertions in vv.27 and 28 do increase the cognitive effects to be de-
rived from the declaration in v.16, increasing its comprehensibility and plausibil-
ity, these assertions are not introduced by gar. Why not? Firstly, I suggest, because
the assertions in these verses are not be processed in the context used for v.16, but
in a different context of their own, with many additional contextual assumptions
accessible (for example, assumptions activated by the intervening events of the
narrative.) Secondly, the assertions in vv.27 and 28, which represent the climax of
the story, are relevant in their own right, communicating many new implications
within the wider narrative (e.g. “Herod was weak and easily manipulated,” “Herod
killed an innocent man,” “John died a prophet’s death after speaking out against
unrighteousness,” “The political authorities in Palestine were corrupt/unjust,”
 Chapter 4. The Greek connective gar 111

“The political authorities in Palestine were likely to treat unjustly other prophets
who spoke out against unrighteousness”, “John’s death implies that Jesus would
suffer a similar fate” etc.). The assertions in vv.27 and 28 are not only relevant as an
explanation of Herod’s declaration in v.16.
In summary, we can say that although the examples of gar apparently introduc-
ing steps back into the past in v.17 and v.18 at first seem different from other more
straightforward explanatory occurrences, in fact this use of the connective can also
be explained by a procedural strengthening account. In these examples, too, gar
encodes instructions to search for contextual assumptions which help to increase
the relevance of a preceding assertion, by increasing the strength with which it is
held as an assumption by the addressee. In other words, gar guides towards the
cognitive effect of strengthening a previously communicated assumption, in this
case by making better sense of it in the narrative, increasing its plausibility in the
world of the narrative. In the particular context of this example, the assumptions
introduced by gar as part of the explanation for the preceding assertion happen to
refer to events that occurred prior to the assertion being explained. This is easily
explicable in terms of a natural temporal sequence of cause and effect: ordinarily
the cause of an event or action will be chronologically prior to it. Thus when gar
occurs in narrative material introducing an explanation we should expect that it
may often introduce an anterior event. Its function, however, is not to indicate the
narrative device of a flashback.
In the light of this examination of examples of gar in both argumentation and
narrative, I suggest that the connective has a core function of encoding procedural
instructions which reduce processing costs by guiding the addressee to find rel-
evance in a particular way. It indicates that the assertion it introduces should be
combined with accessible contextual assumptions in an inferential process, the
outcome of which increases the cognitive effects that are derived from a previ-
ously communicated assertion. More specifically, it contributes to the cognitive
effect of strengthening: as a result of the procedure triggered by gar, the preceding
assertion is more strongly held as true in the case of argumentation, or as plausible
within the world of the story in the case of narrative. In logical argumentation, a
premise strengthens a preceding assertion by guiding towards additional evidence
for its truth. In a narrative, on the other hand, an explanation strengthens to the
extent that it causes an assumption to be better understood and thus to be more
strongly accepted as a plausible part of the world of the narrative. Similarly, in
cases where gar introduces an explanation within argumentation, the preceding
claim is likewise becomes more comprehensible and is consequently adhered to
more strongly as a plausible assertion in the argument.
Thus we might say that gar signals that the information it introduces is relevant
not in its own right but rather it contributes to the relevance of an assertion that
112 Sarah Casson

precedes it. Depending on the context of the communication and the expectations
of relevance it raises, gar may guide the addressee’s comprehension module to pro-
cess what it introduces as part of an explanation, or as evidence which confirms
a previous claim, or perhaps, on occasion, as both. Such an analysis accounts for
the connective’s varied uses in argumentation and narrative in terms of a core,
strengthening function.

5. Different stylistic effects?

If, as I have suggested, gar has a core function across argumentative and narrative
texts, how do we account for the perception on the part of some interpreters that
it contributes to varied styles and stylistic effects in different kinds of communica-
tion? Does gar not only give procedural instructions, but also contribute to extra
cognitive effects? We have already noted De Jong’s assertion that gar is to be as-
sociated with “a typical archaic (oral?) form of narration” in Classical Greek (1997,
p. 179). Meanwhile, in relation to New Testament Greek, scholars such as Thrall
(1962, pp. 47–48) have claimed that the use of gar in Mark’s Gospel contributes to
a certain style that, in Thrall’s judgement, is not the most “logical.” She bases this
judgement on the fact that “the narrator first mentions the “striking points in the
story and then fit[s] in the explanatory details…by using gar.” Thus “The sequence
of events is related backwards…and the fact which in strict logic should have been
mentioned at the beginning…is left until the end.”
In certain examples of argumentation, on the other hand, interpreters see gar
as characteristic of tightly argued logic. In the epistle to the Romans Chapter 1,
verses 16 to 20, for example, six propositions providing supporting evidence
introduced by gar occur in a consecutive series. According to New Testament
scholar N.T.Wright (2013, p. 765) this successive use of premises introduced by
gar indicates a tight sequence of thought in which each premise becomes a conclu-
sion which is justified by a following premise. This contributes to a forceful and
compelling persuasive style, created in part by the explicit guidance to strengthen-
ing that gar gives, as well as by the content of the series of confirming premises.
This contrasts sharply with the apparently “less than logical” style which Thrall
finds in the narrative of Mark’s Gospel.
Sperber’s notion of a display of coherence (2001, p. 410) is of relevance to the
question of possible stylistic effects created by the repeated use of gar in quick
succession in argumentation. Sperber suggests that the use of certain inferential
and logical connectives in argumentation contributes to such a display which
helps to counter the addressee’s epistemic vigilance mechanisms which check the
 Chapter 4. The Greek connective gar 113

reliability and coherence of content.2 If this were the case with gar, then the use
of the connective, combined with the argumentative pattern of claim backed up
by premise, would have the effect of drawing attention to the accumulation of a
string of conclusions with their confirming premises. This would then produce a
rhetorical or persuasive effect on the addressee, countering his epistemic vigilance
by increasing his awareness of the logic of the argument, and of the compelling
weight of evidence backing it up.

6. Epistemic vigilance and stylistic effects?

Can we account for the apparently differing stylistic effects associated with gar’s
use in narrative as opposed to argumentation in terms of the interaction between
the connective and the so-called argumentative module and its epistemic vigilance
mechanisms? Wilson (2011, 2012) and Unger (2012), among others, have sug-
gested that some procedural indicators such as connectives may trigger this pro-
posed argumentative module of the addressee’s mind and play a role in countering
epistemic vigilance. They argue that connectives may counter those mechanisms
which assess the reliability of the content of the communication, thus helping to
persuade the addressee of the believability of the communication and its rational
coherence with the addressee’s existing beliefs. Unger (2012, p. 57) further argues
that certain connectives in Behdini-Kurdish can give rise to either “argumentative
or non-argumentative readings.” He suggests that such connectives, which have
the potential to give rise to both argumentative and non-argumentative readings,
do not themselves trigger the addressee’s argumentative module. Instead, they
trigger a procedure in the inferential comprehension module by giving instruc-
tions which guide the inferential process (as we have suggested gar does). It is the
resulting inference which may or may not trigger epistemic vigilance mechanisms
in the argumentative module, depending on whether this inference is relevant “as
an argument for a claim or as merely establishing textual (narrative) connectiv-
ity.” (Unger, 2012, p. 57). In other words, the inference which is the result of the
procedure triggered by gar may or may not be assessed for rational coherence
and reliability of content, depending on whether it is interpreted as argumentative
i.e. seeking to persuade the addressee to accept the rationality of a truth claim,

2. In his discussion of this subject, Sperber (2001, p. 409) uses the term coherence to refer
specifically to the kinds of “logical relationships and evidential relationships of support” that
are presented in argumentation to convince the addressee of its reasonableness and rationality,
rather than using it with a more general sense of any kind of discourse relations. I will follow
Sperber’s use of the term.
114 Sarah Casson

or as narrative testimony, in which case its believability as a rational claim is

not likely be in view.
Could Unger’s proposal that some connectives give rise to both “argumenta-
tive” and “non-argumentative” readings help to explain some of the perceived
stylistic differences, noted above, created by gar’s use in argumentation in contrast
to narrative? More specifically, can we say that gar achieves different stylistic
effects depending on whether the argumentative module is triggered as part of
the inferential process towards which gar guides? Applying Unger’s analysis of
conclusion-indicating connectives to gar, a premise-indicating rather than a
conclusion-indicating connective, the suggestion would be that gar itself does
not trigger the argumentative module or epistemic vigilance. Instead, it triggers a
procedure in the inferential comprehension module. It is the resulting inference,
and whether it is relevant as an explanation or as confirming proof, which may or
may not trigger epistemic vigilance. More importantly, the preceding assertion
would also trigger epistemic vigilance in the case of an argumentative claim, but
would not in the case of a non-argumentative assertion in a narrative (i.e. the
presentation of an event or action in the story). On this account, explanatory, non-
argumentative assertions introduced by gar would not be assessed by epistemic
vigilance. Confirmatory, argumentative assertions introduced by the connective,
on the other hand, would be assessed by this mechanism.

7. Communicator goals, expectations of relevance and stylistic effects

While this may be true, it seems unlikely that it is the activation of epistemic vigi-
lance mechanisms which creates the impression of a persuasive style for the ad-
dressee. It seems more likely that the explanation for any such stylistic effects is to
be sought in the goals of the communicator, and in the addressee’s corresponding
expectations of relevance, than in the activation of the addressee’s interpretative
mechanisms. As noted above, communicator goals, and the addressee’s expecta-
tions of relevance, vary according to different types of communication. Wilson
(2011, pp. 20–21) summarises the twin goals of ostensive communication, from
the addressee’s point of view, as comprehension on the one hand, and accepting
or believing the information communicated on the other. The addressee’s compre-
hension module is involved in achieving the first goal. The addressee’s epistemic
vigilance mechanisms are engaged in relation to the second.
As already discussed, one of the fundamental characteristics of argumentation
in contrast to narrative is the communicator’s intention to persuade the addressee
to accept certain claims as rational and believable. Sperber (2001, p. 410) suggests
that argumentation involves providing evidence that shows the reasonableness of
 Chapter 4. The Greek connective gar 115

a communication, displaying its internal and external coherence, so as to persuade

the addressee. (By “external coherence” is meant logical and rational compatibility
with existing assumptions which the addressee already holds.) Specific linguistic
forms and rhetorical devices associated with argumentation, as well as reasoning
and evidence at the level of content, are all employed by the communicator to
convince the addressee. At the same time, for the addressee, these elements are
an indication that the communicator has a persuasive intention. It seems intuitive
that as soon as an addressee detects that a communicator has a persuasive inten-
tion, he will have particular expectations of relevance (i.e. that the text will be
relevant to him by persuading him of the truth of claims). As a result, his epistemic
vigilance mechanisms checking for reliability of source and content will be put on
high alert, ready to check the communication for coherence and rationality.
In the case of occurrences of gar in argumentation which appear to contribute
to a tightly argued rhetorical style, the communicator’s aim is to convince the ad-
dressee of a previous claim, by means of the presentation of evidence, which serves
as proof, in conjunction with existing assumptions the addressee has about the
world. The impression that a chain of successive occurrences of gar in argumenta-
tion contributes to a persuasive stylistic effect is surely attributable primarily to
the accumulation of compelling claims and evidence, which work together to con-
vince, rather than to the repeated use of gar alone. At the same time, in such cases
although gar primarily guides towards the inferential process of strengthening, we
can suggest that it may also secondarily display the coherence of the communica-
tion (cf. Sperber, 2001, p. 410) in order to counter epistemic vigilance mechanisms
tasked with assessing the reliability and coherence of content. The intended effect,
both of the confirming premises, and of the use of gar itself, is to convince the
addressee to hold the previous claim more strongly as true. Gar reduces processing
effort by guiding to a strengthening interpretation, increasing the relevance of the
strengthened claim, but it also perhaps counters epistemic vigilance by flagging up
evidence. It may be that gar comes to be associated in the addressee’s mind with
evidence, so that whenever the addressee encounters it the initial impression is of
the presence of convincing evidence.
The expectation, raised by argumentative communication, of a particluar kind
of relevance in terms of truth claims, presumably predisposes the addressee to
interpret gar in argumentation as guiding towards a persuasive interpretation,
towards the strengthening of claims, rather than towards explanatory detail that
increases comprehensibility. It is my suggestion that the stylistic effect of forceful
persuasion sometimes associated with successive occurrences of gar comes from
the fact that within argumentation gar is used repeatedly to indicate an accumula-
tion of evidence, drawing attention to it and thus to the internal and external co-
herence of the argument. In such a context, successive occurrences of gar, together
116 Sarah Casson

with the evidence itself, contribute to the impression of a forceful, clearly directed,
coherent and well supported argument.
In narrative, on the other hand, the communicator’s aim is primarily to inform
with a view to awakening the addressee’s interest and identification with the world
of the story, and engaging his emotions. My suggestion is that the information
is typically readily accepted once understood, since the rationality and truth of
claims are not primarily at stake. The addressee expects the communication to be
relevant as a compelling story which makes sense in the narrative world. I suggest
that his epistemic vigilance is not automatically put on high alert by the detection
of such a kind of communication. Thus, in narrative texts, because of expectations
of relevance raised by the type of communication and the communicator’s aim,
the addressee is poised to interpret such an assertion as an explanation, not as evi-
dence. In narrative, an assertion preceding gar does not trigger epistemic vigilance
and questions about its truth in the way that a claim in argumentation would, but
it may raise an implicit question of comprehension. In this case, the utterance in-
troduced by gar is processed by the addressee’s inferential comprehension module
but does not ordinarily need to be assessed for coherence by epistemic vigilance.
If this is so, then in such a non-persuasive text type, repeated uses of gar will not
be interpreted as displaying coherence (in Sperber’s sense of the term) because the
rationality of the communication is not primarily under scrutiny.
Finally, we may want to question the perception that in certain instances in
narrative texts gar contributes to an “oral” or unsophisticated, “less than logical”
style (see De Jong and Thrall). Such a perception can perhaps, however, be ex-
plained in cognitive terms by the fact that in initially describing an action or event,
and then explaining its causes or the participants’ motivations in more detail, the
communicator is making extra assumptions available to the addressee in what
could be viewed as an afterthought (Blakemore, 1987, p. 90). Blakemore argues
that “the need to provide either explanation or justification may not be anticipated
in advance, and in this respect utterances prefaced by you see and after all may be
regarded as afterthoughts or repairs.”3 In the oral communication of a story, once
the event line is underway, a communicator may typically present first what is
most relevant and salient in terms of events or actions, either because she has less
opportunity to anticipate in advance what may be needed by way of explanation
(in contrast to written communication), or as a chosen strategy in order to capture
the attention of her audience. She then subsequently provides additional contex-
tual information to help addressees to connect these events more satisfactorily

3. Blakemore argues that you see introduces an explanation in English, while after all introduces
a ‘justifying’ premise or evidence for a previous claim. Thus these two connectives in English
roughly cover the function of gar in Koine Greek.
 Chapter 4. The Greek connective gar 117

with their existing understanding of the world of the narrative. Any perceived
“oral” style that were to result from this would not be attributable to the use of gar
itself, but to the order in which the events/actions and explanations are presented.

8. Conclusions

We can conclude that in both argumentation and narrative gar has the same core
function of indicating that the following utterance, in combination with addi-
tional contextual assumptions, should be interpreted as increasing the relevance
of a preceding assumption through strengthening. The information introduced
by gar achieves relevance in relation to the preceding assumption which it helps
to strengthen and may thus be perceived as relatively less salient, or background
information. While gar has a core function, this is worked out in different ways in
different kinds of communication: the specifics of the inferences that result from
its use will depend on the context. In argumentation, gar most often introduces an
assertion which is to be interpreted as evidence which strengthens the believability
of a claim. In narrative, gar typically introduces material which is to be interpreted
as an explanation, making a previous assertion more understandable and thus
compelling and plausible in the narrative world. Gar introducing explanations
may, however, sometimes be found in sections of argumentation in cases where
the addressee is prepared to accept the truth of an assertion on trust without
persuasion. In addition, certain premises introduced by gar are potentially both
explanatory, and confirmatory. In this case, they provide access to additional con-
textual assumptions which aid the comprehension of a previous claim. Once these
assumptions are accessed, they also help to convince the addressee of the truth of
the claim, by providing evidence which serves as proof.
Finally, the use of gar does not by itself create varying stylistic effects in dif-
ferent kinds of communication. The impression that such effects exist may be
attributable to the fact that gar contributes to the stylistic effects of particular kinds
of communication as part of a package, that is, in combination with certain kinds
of content and with other stylistic features that characterise that type of communi-
cation. These are all determined by the communicator’s goals and the addressee’s
corresponding expectations of relevance.
118 Sarah Casson

References

Bakker, S. (2009). On the curious combination of the particles γάρ and οὖν. In S. Bakker, & G.
Wakker (Eds.), Discourse cohesion in ancient Greek (pp. 40–61). Leiden: E. J. Brill. ​
https://‍doi.org/10.1163/ej.9789004174726.i-284.20
Bauer, W., Danker, F. W., Arndt, W. F., & Wilbur Gingrich, F. (Eds.). (2000). A Greek-English lexi-
con of the New Testament and other early Christian literature, 3rd edn. Chicago: University
of Chicago Press.  ​https://‍doi.org/10.7208/chicago/9780226028958.001.0001
Black, S. (2002). Sentence conjunctions in the Gospel of Matthew. Sheffield: Sheffield Academic
Press.
Blakemore, D. (1987). Semantic constraints on relevance. Oxford: Blackwell.
Blass, R. (1998). Constraints on relevance in Koine Greek in the Pauline letters. Unpublished
paper presented at SIL Exegetical Seminar, Nairobi, 29th May to 19th June, 1993.
Blass, R. (2005). Manipulation in the speeches and writings of Hitler and the NSDAP. In L.
De Saussure, & P. Schulz (Eds.), Manipulation and ideologies in the twentieth century
(pp. 169–190). Amsterdam: John Benjamins.  ​https://‍doi.org/10.1075/dapsac.17.09bla
De Jong, I. (1997). Γάρ introducing embedded narratives. In A. Rijksbaron (Ed.), New approaches
to Greek particles, ASCP 7 (pp. 175–185). Amsterdam: J.C. Gieben.
Denniston, J. (1934). The Greek particles. Oxford: Clarendon Press.
Goldman, A. (1999). Knowledge in a social world. Oxford: Oxford University Press. ​
https://‍doi.org/10.1093/0198238207.001.0001
Levinsohn, S. (2000). Discourse features of New Testament Greek. A coursebook on the informa-
tion structure of New Testament Greek. 2nd ed. Dallas: SIL International.
Runge, S. (2010). A discourse grammar of New Testament Greek. Peabody, Massachusetts: Hen-
drikson Publishers.
Slings, S. (1997). Adversative relators between PUSH and POP. In A. Rijksbaron (Ed.), New
approaches to Greek particles (pp.101–129). Amsterdam J. C. Gieben.
Sperber, D. (2001). An evolutionary perspective on testimony and argumentation. Philosophical
Topics 29:401–413.  ​https://‍doi.org/10.5840/philtopics2001291/215
Sperber, D., Clément, F., Heintz, C., Mascaro, O., Mercier, H., Origgi, G. & Wilson, D. (2010).
Epistemic Vigilance. Mind & Language, 25, 359–393. ​
https://‍doi.org/10.1111/j.1468-0017.2010.01394.x
Sperber, D., & Wilson, D. (1995). Relevance: Communication and Cognition, 2nd ed. Oxford:
Blackwell.
Thrall, M. (1962). Greek particles in the New Testament. New Testament tools and studies III.
Leiden: E. J. Brill.
Unger, C. (2002). Global coherence, narrative structure and expectations of relevance. Paper
presented at the conference Relevance Theory and Literature, University of Huddersfield,
12th-13th September, 2002.
Unger, C. (2012). Epistemic vigilance and the function of procedural indicators in communica-
tion and comprehension. In E. Wałaszewska, & A. Piskorska (Eds.), Relevance theory: More
than understanding (pp. 45–73). Newcastle-upon-Tyne: Cambridge Scholars Publishing.
Wilson, D. (2011). The conceptual-procedural distinction: Past, present and future. In V. Escan-
dell-Vidal, M. Leonetti, & A. Ahern (Eds.), Procedural meaning: Problems and perspectives
(pp. 3–26). Emerald Group Publishing. ​
https://‍doi.org/10.1108/S1472-7870(2011)‍0000025005
 Chapter 4. The Greek connective gar 119

Wilson, D. (2012). Modality and the conceptual-procedural distinction. In E. Wałaszewska, &

A. Piskorska (Eds.), Relevance theory: More than understanding (pp. 23–43). Newcastle-
upon-Tyne: Cambridge Scholars Publishing.
Wright, N. T. (2013). Paul and the faithfulness of God. Parts I-IV, vol. 2. Christian Origins and the
Question of God 4. London: SPCK.
Zakowski, S. (2016). She was twelve years old: on γάρ and Mark 5:42. University of Ghent.
https://‍biblio.ugent.be/publication/5917339/file/7203193.pdf.
 Chapter 5

Metarepresentation markers in
Indus Kohistani
A study with special reference to the marker of
desirable utterances loo

Beate Lubberger
Independent researcher

The marker loo, a discourse particle of Indus Kohistani, spoken in Northern

Pakistan, marks utterances that a speaker wishes her audience to convey to a
third party and furthermore indicates third-person imperative. The proposed
analysis uses relevance theory as framework and includes a brief look at three
other Indus Kohistani discourse particles: the “hearsay” evidential lee, the quota-
tive karee, and the complementizer če. The distinction made within relevance
theory between utterances that represent states of affairs and utterances that
metarepresent other representations (speech, thought) on one hand, and
between conceptual and procedural meaning on the other hand, allows for an
analysis of all four markers as procedural indicators of metarepresentation. What
distinguishes them is the type of metarepresentation they mark: the particle loo
specifically indicates non-attributive metarepresentations of desirable utterances.

Keywords: discourse marker, procedural encoding, non-attributive

metarepresentation, desirable utterances, third-person imperative, Indus
Kohistani

1. Introduction

This paper describes and analyses four discourse particles in Indus Kohistani,1 an
Indo-Aryan language spoken in the mountainous region of Northern Pakistan:2

1. ‘ISO 639–3 code [mvy]’

2. In the literature Indus Kohistani has been described as belonging to the group of Dardic
languages; these are spoken in an area stretching from the Hindukush over the western part of

https://‍doi.org/10.1075/ftl.8.05lub
© 2020 John Benjamins Publishing Company
122 Beate Lubberger

the “hearsay” evidential lee, the quotative karee, and the complementizer če are
briefly discussed to provide the context for the marker loo, the paper’s main topic.
Like the other three particles, it indicates quotation and has a seemingly twofold
function. Firstly, it marks utterances that a speaker wishes her hearer to convey to
a third person. As an illustration of this use, suppose I say to my sister, “If you see
mother, tell her that I will come on Sunday.” The underlined utterance is the one
that I ask my sister to say to my mother. The Indus Kohistani version of this utter-
ance using the marker loo would be “If you see mother, I will come on Sunday loo.”
Secondly, loo occurs in clauses that have a third person subject and as predicate
a verb marked for second person imperative, as in kar-a=loo ‘he do-imp.2sg=loo’
or ‘he should/may do/let him do’. In such instances, loo seems to encode a third
person imperative meaning.
The proposed analysis of the markers uses relevance theory (Sperber &
Wilson, 1995), an inferential theory of communication, as framework. Two
particular notions within relevance theory bear on this analysis: the first is the
distinction between conceptual and procedural encoding (Blakemore, 1987,
2002), the second is the use of different types of metarepresentation in linguistic
communication (Wilson, 2012a). This paper argues that all four markers indicate
metarepresentation. Lee, karee, and če mark metarepresentations of attributed or
self-attributed quotations. Loo, on the other hand, indicates one specific type of
non-attributive metarepresentation, namely that of desirable utterances, thereby
helping the hearer to construct an appropriate higher-level explicature of the
kind “the speaker wishes me to tell mother that [desirable utterance],” to take the
example from above.
The data used in this paper consist of a large corpus of oral texts that include
folk stories, narratives of personal experiences, expository and procedural texts;
these have been recorded, transcribed and translated between 2001 and 2009 by
the author.3 A second corpus of recorded conversations has been collected within
the extended family of my main language consultant in the years 2011 to 2014;
the passages relevant for this study have again been transcribed and translated by

the Himalayas. However, the use of the term ‘Dardic’ is disputed as the languages labelled by it
are not a separate branch of Indo-Aryan.

3. The corpus includes one hundred and five texts of varying length: the shortest ones have
twenty sentences, the longest ones five hundred and thirty sentences. Average length is between
one hundred and two hundred sentences. The author lives in Pakistan and started collecting
texts for learning and analysing the language.
 Chapter 5. Metarepresentation markers in Indus Kohistani 123

the author. The data represent the Pattan4 variety of Indus Kohistani, the dialect
spoken in settlements near the river Indus.5
The paper is structured as follows. In Section 2, I will look at some analyses
of evidentials and quotation markers. Then, after a brief introduction of the key
elements of relevance theory, the distinction between conceptual and procedural
meaning and the concept of linguistic metarepresentations will be reviewed and
four Indus Kohistani discourse markers briefly introduced. Section 3 will describe
in more detail the uses of the Indus Kohistani marker loo and then present its
analysis as a procedural marker of one specific type of metarepresentation, namely
that of desirable utterances. Section 4 will summarize the findings.

2. Theoretical background

2.1 Analyses of evidential and quotation markers

All four Indus Kohistani particles mentioned here – loo, the main topic of this
paper as well as the other three that will be briefly described – have in common
their use as markers of quotation in the widest sense: lee is a reported-speech
evidential, karee indicates quotations and self-quotations of speech and thought,
če is a multi-functional complementizer with uses similar to those of karee, and
loo indicates utterances that a speaker wishes her audience to say to an absent
third party. In this section I will briefly look at some analyses of evidentials and
markers of quotations.

2.1.1 Evidentials
Are evidentials modals or are they a distinct category? Palmer (2001), for instance,
assumes that both epistemic and evidential systems are part of modality. Aikhenvald
(2004), on the other hand, views evidentiality as a grammatical category (this
excludes lexical expressions of information source) distinct from modality. In her
seminal typological study, she notes as core meaning of evidential markers “source
of information” although evidential systems may develop epistemic extensions.
Other secondary/additional meanings as described by her include mirative exten-
sions, uses of “reported” markers as tokens of narrative genres, the marking of
ironic and sarcastic utterances, and indicating complements of verbs of cognition.

4. Pattan [páṭã] is the name of a settlement in the main Indus valley.

5. For more information about the language Indus Kohistani see Hallberg & Hallberg (1999),
Hallberg (1992), Lubberger (2014), Zoller (2005).
124 Beate Lubberger

The question as to how these different meanings/extensions may be explained

is left open.
Within Turkic languages, Johanson (2000) offers a different analysis of gram-
matical markers that indicate reported speech, knowledge gained by inference,
and mirativity.6 In Turkic languages, both “reported” and “inferred” evidentiality
as well as mirativity may be marked grammatically. However whereas Aikhenvald
sees mirativity as related to evidentiality but nevertheless as a distinct category, for
Johanson both evidential and mirative meaning are instances of the notion of indi-
rectivity. An utterance marked by either one of the markers has thus a second layer
that, in Johanson’s words, “expresses the speaker’s attitude towards the proposition
expressed in the sentence” (2000, p. 70). This analysis covers all the above men-
tioned uses: reported, inferential and mirative (called receptive by Johanson) and
seems to be confirmed by the fact that some Turkic languages employ different
markers for the three notions whereas others (modern Turkish for instance) have
one marker that covers all three uses subsumed under “indirectivity.”
Faller (2002, 2006) studies two evidential markers of Cuzco Quechua, using as
theoretical framework speech act theory. In her analysis, both the “direct” and the
“reported” marker, are not part of the propositional content of a speech act; they
function as illocutionary operators, modifying the illocutionary force level of the
speech act in that they “provide clues to the speaker’s attitudes toward the content
of a proposition.” For instance the use of the “reported” marker in an utterance
indicates to a hearer the speaker’s attitude of not necessarily believing the proposi-
tion expressed, provided there is another person (the one quoted) who has said
something with the same propositional content.
As Unger (2016) notes, these approaches shed light on how evidentials guide
a hearer’s comprehension process by indicating the type of evidence but do not
explain the whole range of meaning extensions mentioned above. In Section 2.2.1
we will see that, within relevance theory, facilitating a hearer’s comprehension is
not the only goal of using evidential markers. Rather, they are employed in order to
influence a hearer toward believing what is being communicated. In other words, a
speaker uses evidentials to overcome a hearer’s epistemic vigilance (Sperber et al.,
2010; Wilson, 2011). This insight can, for instance, be used to explain the meaning
extension of evidentials as tokens of genre (Unger, 2016).

6. The notion of mirativity includes “speaker’s unprepared mind, unexpected new information
(new for the speaker) and concomitant surprise” and is seen as related to but distinct from
evidentiality (Aikhenvald, 2004, p.195).
 Chapter 5. Metarepresentation markers in Indus Kohistani 125

2.1.2 Grammaticalized quotation markers and complementizers

The grammaticalization of a number of different word categories (for instance
verbs of saying, of action, of motion, of inchoativity, of similarity and manner, and
pronominal items among others) into quotation markers and further into comple-
mentizers has been described for many languages (for example: Lord, 1976, 1993
for African languages, Bashir, 1996 and Saxena, 1995 for South-Asian languages,
Chappell, 2008 for Sinitic languages). Lord was the first to describe this particular
grammaticalization path; her work has influenced further studies of this particular
phenomenon. It is now assumed that the most typical grammaticalization pathway
looks as follows:
Original form > quotation marker > complementizer > purpose-clause linker >
reason-clause linker > conditional-clause linker > naming linker > marker used
with onomatopoeic words/ideophones > comparative marker > mirative marker

Grammaticalized “quotative indexes”7 (a term used by Güldemann) may further

function as listing construction markers, topic markers, clause-final discourse
particles expressing self-evident assertions, occur in warnings, in echo questions,
and as clause-initial markers for exclamation (Chappell, 2008, p. 5; see also Bashir,
1996; Güldemann, 2008; Lord, 1993).
In his extensive survey of quotative indexes in African languages, Güldemann
makes several interesting points. Firstly he treats all complements of verbs of speech,
cognition and perception as quotes in a very general sense as defined by him:
Reported discourse is the representation of a spoken or mental text from which
the reporter distances him-/‍herself by indicating that it is produced by a source of
consciousness in a pragmatic and deictic setting that is different from that of the
immediate discourse (2008, p. 244)

Secondly, he shows that grammaticalized quotative indexes not only appear

in all the constructions mentioned above but are also used in connection with
other types of expression such as non-linguistic sound imitations (for instance
the imitation of an animal sound), with ideophones8 and similar linguistic signs,
and with representational gestures. Güldemann then notes that, like quotes, such
expressions/gestures have to be “inserted” into discourse; often the same means,

7. Güldemann (2008) uses this term to describe what is also referred to as matrix clause in a
complex sentence of complementation. An example of a quotative index is ‘X said to Y’, followed
by a speech complement, or it may consist of just a quotation marker or a complementizer.

8. Definition by Doke (1935, p. 118): “A vivid representation of an idea in sound. A word, often
onomatopoeic, which describes a predicate, qualificative or adverb in respect to manner, colour,
sound, smell, action, state or intensity”
126 Beate Lubberger

namely grammaticalized complementizers, are used to mark them. In the relevant

literature, such uses of quotative indexes are seen as secondary, due to semantic
bleaching or the result of expansion into the non-speech domain. Güldemann,
however, analyzes all four instances of quotation marker use (that is: indicating
quotations, non-linguistic sound imitations, ideophones and representational
gestures) as instances of mimesis, a term meaning “imitation, representation,
mimicry.” He argues that mimesis is the common ground for all four uses men-
tioned above. In his own words,
The behavior of the speaker is such that (s)he PERFORMS the communicated
state of affairs rather than linguistically describing it, as if the unmarked employ-
ment of the signs of language failed to achieve the particular communicative goal,
The speaker attempts to demonstrate, to re-instantiate, to imitate, to replay the
event as close to the purported original as is desired in the context and as human
means of expression allow him/her to do so. (2008, p. 286)

This analysis covers all four uses: quotation proper and “quotation” of non-linguis-
tic sounds, ideophones and representational gestures. Furthermore, Güldemann
claims that grammaticalization does not necessarily have to start with a speech
verb and then develop along a unidirectional cline. A linguistic form might as
well start out as a marker of, for example, ideophones, and later on develop into
a quotation marker. In Güldemann’s analysis, a marker in the grammaticalization
process expands/develops within the same domain, namely that of mimesis. Other
accounts, on the other hand, describe the typical process and direction of gram-
maticalization as well as what motivates change/grammaticalization in general but
do not offer an explanation as to why quotation markers take this particular path
(for instance Hopper & Traugott, 1993).
The literature reviewed above covers markers like the Indus Kohistani lee,
karee and če described in Section 2.3. As for the main topic of this paper, so far I
have not been able to find descriptions, let alone analyses of, markers similar to
Indus Kohistani loo. This marker is distinct from other quotation markers in that
it indicates a very special kind of “quotation,” namely that of utterances that a
speaker wishes her audience to say to a third person not present. Furthermore, loo
functions as indicator of third-person imperative.
The analysis of the marker loo in this paper is based on relevance theory. Within
this framework, not only loo but also the markers lee, karee and če are analyzed as
markers of metarepresentation. Relevance theory is able to explain the common
ground that unites all four markers – they are all indicators of metarepresentation
– as well as the differences between the markers: they indicate different types of
metarepresentation. The marker loo, I claim, specifically indicates metarepresenta-
tions of “desirable utterances,” a term coined by Wilson and Sperber (2012). In the
 Chapter 5. Metarepresentation markers in Indus Kohistani 127

next section, I will present those elements of relevance theory that are particularly
relevant for my analysis.

2.2 Relevance theory

Relevance theory, the framework used in this paper, explains human communica-
tion as mind-reading, i.e. as mainly inferential. It is based on two principles: the
“Cognitive principle of relevance: Human cognition is geared towards the maxi-
misation of relevance” and the “Communicative principle of relevance: Every act
of ostensive communication communicates the presumption of its own optimal
relevance” (Sperber & Wilson, 1995, p. 260). The first principle states that of all
inputs (auditory, visual, sensory etc.) to our cognitive system, we tend to pay at-
tention to those that seem to be the most relevant to us. According to the second
principle a speaker who addresses someone communicates by this very act that
“her utterance is relevant enough to be worth being processed, and that it is the
most relevant one compatible with her abilities and preferences” (Wilson, 2012a,
p. 238). When we process input we expect cognitive effects: the new information
processed within existing assumptions may confirm and strengthen these, it may
lead to new conclusions, thereby increasing our knowledge of the world, or it may
contradict existing assumptions and cause us to eliminate those and so improve
our knowledge of the world. Relevance is defined as a quality of cognitive inputs,
determined by cognitive effects to be gained on the one hand, and by processing
effort to be invested on the other hand. An input’s relevance is in direct relation
to its cognitive effects: other things being equal, the greater the cognitive effects,
the greater the relevance of the input. The opposite is true in relation to the mental
effort: the smaller the processing effort, the greater the relevance of the input.
Decoding an utterance is only the first step of the comprehension process and
results in a quite incomplete logical form. This form then needs to be enriched
and completed by accessing contextual information, using inference, until an
interpretation is reached that satisfies the hearer’s expectations of relevance. This
inferential comprehension procedure has been summarized as follows:
Follow a path of least effort in computing cognitive effects.
a. Consider interpretations in order of accessibility (and in particular in
resolving ambiguities and referential indeterminacies, in going beyond
linguistic meaning, in supplying contextual assumptions, computing
implicatures, etc.).
b. Stop when your expectation of relevance is satisfied.
(Sperber, Cara and Girotto 1995, p. 51; see also Wilson, 2012a, p. 238).
128 Beate Lubberger

One other basic idea developed within relevance theory is the distinction between
two types of word meaning: words that encode concepts or that are descriptive and
words that encode procedural instructions.

2.2.1 Conceptual vs. procedural encoding

Most of the words we use, for instance house, car or go, encode a concept, and as
long as the concept is known, it is not difficult to explain the meaning. Defining
the meaning of words like but, so, or also proves to be more of a challenge. Their
meaning cannot easily be explained in conceptual terms. It was Blakemore who
first analysed such words as encoding not concepts but procedural or computa-
tional meaning. Procedural indicators, as they are termed, are meant to constrain
a hearer’s inferential comprehension procedure, thereby reducing processing costs
(Blakemore, 1987). They act on different levels of the comprehension process:
for instance personal pronouns encode procedural information that guides the
hearer towards recovering the appropriate referent from a given context; they are
constraints on the explicature of an utterance. Discourse connectors such as but or
so, on the other hand, constrain the way in which appropriate implicatures may be
found; and evidentials for instance constrain the hearer towards the construction
of a higher-level explicature such as “speaker X has heard that [utterance].” In fact,
any kind of grammatical marking might be analysed as encoding procedural or
computational instructions aimed at constraining the comprehension procedure
and at the same time decreasing the processing effort (LaPolla, 1997, p. 13). The
distinction between conceptual and procedural meaning and the insights gained
from it have since been successfully applied to the analysis of discourse connec-
tors, discourse markers and other grammatical words (Blakemore, 1987, 2002;
Blass 1990; Ifantidou 2001; Noh 1998; Unger 2012a, 2012b).
More recently, the understanding of how words encoding procedural meaning
work has been expanded. The function of procedural indicators is not limited to
constraining the inferential comprehension process and saving hearer processing
effort and time. It has been argued (Sperber et al., 2010; Wilson, 2011, 2012b) that
in addition to the comprehension module necessary for the comprehension of overt
communication, there are other cognitive procedures, located in distinct modules.
One such a cognitive epistemic vigilance mechanism (Sperber et al., 2010) helps
in assessing the trustworthiness of the source as well as of the content of com-
municated information and may be activated by procedural indicators as part of
their function. But what exactly is epistemic vigilance, and what role does it play in
successful communication? A speaker wants her addressee not only to understand
but also to believe what she says; the hearer, on the other hand, does not only want
to make sense of what is communicated but also wants to protect himself from ac-
cidental or intentional misinformation (Wilson, 2011, p.20). To achieve the latter,
 Chapter 5. Metarepresentation markers in Indus Kohistani 129

the hearer has to assess (i) the reliability of the information source, that is, the
speaker and (ii) the reliability of the communicated information. In short, he has
to exercise epistemic vigilance. So what can a speaker do to achieve her goals of be-
ing understood and believed? She will word her utterances in such a way that they
pass the hearer’s epistemic vigilance mechanisms. Specifically, by using procedural
indicators that show logical connections within the communicated content or
with background information, she attempts to convince the hearer to accept the
content; likewise, the use of evidential markers for instance is aimed at presenting
herself as a trustworthy source of information. It follows that procedural indicators
can function in two ways: (i) as constraints on relevance within the comprehen-
sion procedure with the aim of making less costly the hearer’s understanding, and
(ii) as activators of the hearer’s epistemic vigilance module in order to make him
accept and believe what is communicated. These insights have been successfully
applied to the analysis of evidential markers and other grammatical indicators
(Unger, 2012a, 2012b, 2016) and, as already mentioned in Section 2.1.1, have been
used to explain for instance the use of evidentials as tokens of genre.

2.2.2 Linguistic metarepresentations

The second concept of importance for the analysis proposed here is that of
representation and metarepresentation; indeed, these are basic notions of cogni-
tion and Theory of Mind in general. The photo of a tree represents a real tree;
likewise, the word tree is the representation of a real tree. Knowledge about the
world is stored in our minds in the form of concepts or representations of the
world. An utterance such as “It was raining this morning,” represents a state of
affairs in the world if it is true that it was raining this morning. Such utterances
are instances of the descriptive use of representations (Sperber & Wilson, 1995,
p. 228). Linguistic representations may, however, represent not only concepts
but other representations; for instance, an utterance may represent another ut-
terance or thought. The speaker of such a reported utterance or thought is then
metarepresenting the original speaker’s utterance or thought; this is an instance of
the metarepresentational use of an utterance. The relation between representation
and metarepresentation is one of resemblance rather than identity as long as both
share logical and contextual implications. So, whether I say, “Susan has asked me,
‘would you come for lunch on Sunday?’,” or “Susan has asked if I want to come for
lunch Sunday,” or, “Susan has invited me for lunch on Sunday,” or, “I am invited
to lunch,” all four of my utterances metarepresent Susan’s question although they
differ in the degree of resemblance.
Reported speech is a case of public metarepresentation whereas reported
thoughts are instances of mental metarepresentation. Linguistic metarepresen-
tations are, however, by no way limited to quotations of others’ utterances and
130 Beate Lubberger

thoughts. Such attributive metarepresentations also include self-attributive meta-

representations such as quotes of a speaker’s previous utterances and thoughts,
introduced for instance by “As I said …” or “I think that…” A special case of
attributive metarepresentation are echoic utterances: the speaker echoes an at-
tributed utterance or thought and at the same time expresses her attitude towards
it. Irony has been analysed as such an instance where the speaker quotes or echoes
an attributed utterance or thought and at the same time dissociates herself tacitly
from this quote (Sperber & Wilson, 1995; Wilson, 2012a).
There are more kinds of linguistic metarepresentation of representations that
are not attributed to someone, for example mention of sentence types, utterance
types or proposition types. Furthermore, within relevance theory regular questions
and exclamations have been analysed as metarepresenting desirable thoughts, and
regular negations and disjunctions as metarepresenting possible thoughts. Other
instances of metarepresentation are utterances that a speaker wants her audience
to say to someone; this is a case of metarepresenting a desirable utterance, whereas
drafts and rehearsals of future conversations may be analysed as metarepresen-
tations of possible utterances (Wilson & Sperber, 2012; Clark, 1991; Noh, 1998,
Wilson, 2012a). In Section 3.2, such non-attributive metarepresentational use will
be discussed in more detail.
How does a hearer recognize whether an utterance is the representation of
a state of affairs or of another representation? Quotations, for instance, may be
introduced by “he said,” or “I have heard”; the fact that this is an attributive meta-
representation is encoded conceptually. Grammatical markers such as hearsay
particles, question markers or interrogative clause word order are other means
of identifying metarepresentations, as are intonation, inverted commas and other
paralinguistic resources. Such metarepresentational devices encode procedural
information. There may be no overt marking at all in which case the interpretation
of such an utterance as a metarepresentation has to be wholly inferred (Wilson,
2012a, p. 244). Languages also differ in what types of metarepresentation they
explicitly encode and whether the encoding is conceptual or procedural.
The distinction between representational and metarepresentational use of ut-
terances, based on works of Gutt (1991), Sperber & Wilson (1995), Wilson (2012a)
has had impact on the analysis of a number of linguistic items such as evidentials
(for instance Blass, 1990; Ifantidou, 2001; Itani, 1994; Papafragou, 2000; Papafragou
et al., 2007), attitudinal adverbs and particles (see for instance Ifantidou-Trouki,
1993; Wharton, 2009), quotation (see Gutt 1991; Noh 1998), and other discourse
particles (Unger, 2012a, 2012b). Some of these items, take for instance evidentials,
may have an array of seemingly unrelated functions; an analysis as indicators of
metarepresentation is able to explain these uses. The four Indus Kohistani dis-
course markers mentioned in the next sections are a case in question: what unites
 Chapter 5. Metarepresentation markers in Indus Kohistani 131

them is the fact that they are all indicators of metarepresentational use; what
distinguishes them from one another is the type of metarepresentation they mark.

2.3 Indus Kohistani markers of metarepresentation

The marker loo discussed in this paper is one of several particles that indicate
that the utterance marked by it does not represent a state of affairs but another
representation, be it speech or thought. In other words, all of them are markers
of metarepresentation. What distinguishes them is what type of metarepresenta-
tion they indicate. Before discussing loo let us therefore have a brief look at the
other three markers:9
i. The Indus Kohistani metarepresentation marker karee is the grammaticalised
converb10 form of the verb kar- “do.” Both the converb kareé “having done”
and the marker karee occur frequently in discourse. It is assumed that karee
started out as a quotation marker; its grammaticalisation path is similar to that
of grammaticalised SAY11 verbs found in many African languages and also in
languages of the Indian Subcontinent. Nowadays, karee is used as indicator of
reported (‘she said’), self-reported (‘as I said’) and hypothetical or not realized
speech but mostly as marker of reported and self-reported thoughts, as the
following examples illustrate. Example (1) is an instance of a speech not to be
realized. The clause marked by karee is in square brackets.
(1) wál talá bazií ṣeé nií man-ìi [če zõṍ c̣ aal-áthe karee]
but there go.cvb such neg say-imp.2pl comp 1pl.erg find-prs.pfv mrm
‘But when you go there don’t say that you have found (the baby)’
[lit.: ‘… do not say, “We have found the baby”’] (How they found the baby
# 57)

This example also exemplifies that any quotation in Indus Kohistani – be it of

speech or thought – is in direct speech form.
In (2) the clause containing the thought of the women (the first clause in
square brackets) is embedded within the main clause and only recognizable as a
thought by the presence of the marker karee.

9. For a more detailed analysis of these discourse markers see Lubberger (2014).

10. Haspelmath defines the converb as “the nonfinite verb form whose main function is to
mark adverbial subordination” (Haspelmath, 1995). The Indus Kohistani converb is extensively
used in clause chaining; such constructions consist of several subordinate clauses containing a
converb and followed by the sentence-final finite verb form of the main clause.

11. SAY verb stands for the semantic field of verbs of saying.
132 Beate Lubberger

(2) khẽ sãĩ́ ɡharimaaṣ-õṍ zúno-zuno [maasúm mar-èel

devm 3pl.dist woman-pl.obl quickly~redupl child die-ads.m
thú karee] [karápu này kar-eé] maasúm piiruú
be.prs.m.sg mrm cutting umbilical.cord do-cvb child over.there
ɡal-áɡil
put-pfv2
‘Then the women, thinking that the baby is dead, would quickly cut the
umbilical cord and put the baby aside’
[lit.: ‘Then the women, the baby is dead karee, would quickly cut the
umbilical cord and put the baby aside’] (Conception, birth #229).

Note that the “thought” clause is immediately followed by another subordinate

clause containing the converb kareé.
Indus Kohistani clauses expressing purpose and reason are also marked by
karee because they, too, express thought in the widest sense. Example (3) below is
a purpose clause.
(3) muná way-ɡal-eé só maṭoó tal sẽẽ̀ [čuúṭ
down go.down.cvb-throw-cvb 3sg.dist boy.obl on 3sg.erg slap
dàm karee] hàa buí kìir
give.sbjv.1sg mrm hand upwards do.pfv1
‘Having climbed down (from the tree) he raised his hand to hit the boy’
[lit.: ‘Having climbed down he “I will hit the boy” karee raised his hand’]
(Hair parting story #21)

The purpose “to hit the boy” is being expressed as the direct-speech quotation of
a thought.
The next Example (4) contains a reason clause marked by karee. Again, the
reason why the man dismissed his employee is presented in the form of a thought
quotation in direct speech form.
(4) màaṣ-e eýt kar-i-aáthe [miiɡeé pèes nií
man-erg abandoninɡ do-caus-prs.pfv 1sg.dat money neg
thé karee]
be.prs.m.pl mrm
‘The man made him lose (his job) because he had no money (to pay an
employee)’
[lit.:‘The man dismissed him, I do not have money karee’] 
 (conversation 19.11.2013)

All clauses marked by karee seen so far contain the representation of a quotation
or thought – someone else’s or one’s own – rather than that of a state of affairs.
Karee may also occur in constructions of similarity and of naming. An instance of
 Chapter 5. Metarepresentation markers in Indus Kohistani 133

the latter would be the following sentence, “He is a Gujjar, X (name) karee” which
can be translated as “He is a Gujjar named X.” The name here is an instance of
mention and, in relevance-theoretic terms, a non-attributive metarepresentation.
The grammaticalized marker karee has been analysed as an indicator of
attributive and self-attributive public and mental metarepresentations; its oc-
currence in naming and similarity constructions suggests that its use has been
further extended to include not only the domain of attributive but also part of the
domain of non-attributive metarepresentations. This may also shed some light on
the grammaticalization path of karee (and other similar markers) as described in
Section 2.1.2. I assume that a SAY verb, for instance, starts out as marker of one
particular type of metarepresentation, namely attributive public ones, and then
gradually extends its uses to include other types of attributive, and later on, of
non-attributive metarepresentations.
ii. The more recently borrowed12 complementizer and multifunctional marker
če is on its way to gradually take over functions described for karee: če is the
default complementizer of speech, perception and cognition complements
and of other sentential complements that express cognitive functions in the
widest sense; like karee it also may occur in purpose and reason clauses.
Where če replaces karee, it too may be analysed as a marker of attributed and
self-attributed public and mental metarepresentations. The examples below
show not-so-typical complement constructions where the verb of the main
clause is not a speech verb, nevertheless the complement is a direct speech
utterance or thought.
(5) ás wáxt-a man só deeqàan intezàar kar-àant [če
3sg.prox.obl time-obl in 3sg.dist tenant waiting do-prs.m comp
hĩ̀ĩ ṣṹ makày mĩ̀ĩ ɡhãĩ́ ho-ṣaýt]
now 3sg.dem corn 1sg.poss big.inch become-fut.f
‘At that time the tenant is waiting for his corn to grow’
[lit.: ‘… is waiting that “now my corn will become big”’] (About deqani #32)

The clause introduced by če is the complement of the verb “wait.” This verb does
not always have a sentential complement; take for instance the sentence in (6),
where a possessive construction has been used.
(6) ṹ tã̀ã xawànd-ãã intezàar kar-àynt
3sg.prox refl.poss.m husband.obl-gen.m waiting.m do-prs.f
‘She is waiting for her husband’ (elicited)

12. The marker is most probably borrowed from Pashto če, Pashto being one of the languages of
wider communications in North Pakistan (see also Lubberger 2014, p. 123).
134 Beate Lubberger

In Example (5), however, the purpose of waiting has been expressed as a thought
of the one doing the waiting, hence the marking by če indicating an utterance or
thought.
In Example (7), the clause in square brackets expresses a purpose in the form
of a thought, recognizable as such by the use of direct speech. Consequently it is
marked by če.
(7) ás wáxt-a man ɡát taawìiz kar-i-aánt-ø [če
3sg.prox.obl time-obl in again charm do-caus-prs.m-pl.m comp
miiɡeé mazduurìi hòo miiɡeé pèes hòo]
1sg.dat labour become.sbjv.3sg 1sg.dat money become.sbjv.3sg
‘At that time again (men) get made a charm so that they may get work, may
get money’
[lit.: ‘… so that I may get work, I may get money’] (The evil eye #168)

In summary, where če replaces karee, it too may be analysed as a marker of attribu-

tive and self-attributive public and mental metarepresentations.
iii. The Indus Kohistani ‘reported’ marker lee is an evidential marker;13 it indicates
hearsay or reported information, that is, quoted speech only. In relevance-
theoretic terms, quotations are public attributive metarepresentations,
consequently, lee is an indicator of public attributive metarepresentations. Its
use is illustrated by Examples (8a) to (8d), taken from a narrative about the
earthquake in 2005. The speaker is quoting one of her sons talking to her when
he came home after the earthquake.
khẽ ẽ̀ẽ
(8) a. màayn ãã oó kuú=lee
devm 3sg.prox.erg say.pfv1 yes voc older.sister=rep
‘Then he said, “Yes, oh Older Sister’
b. má ḍi-íthu=lee
1sg flee-prs.pfv.m.sg=rep
‘I have fled’
c. iskul-ã̀ĩ kùṛ-muṛ bazíthe=lee
school-gen.f wall.f.pl-things.like.that.echo go.prs.pfv.m.pl.=rep
‘The walls of the school went (down)’
d. hãã maasmá búṭ báč hu-úthe=lee
and child.pl all saved become-prs.pfv.m.pl=rep
‘But the children all escaped’
(The earthquake #188, 190–1)

13. At present, the origin of lee is unknown. There is a similar ‘Reportativ’ particle in Waigali,
one of the languages of Eastern Afghanistan, see Buddruss 1987, Degener 1998, Lubberger 2014.
 Chapter 5. Metarepresentation markers in Indus Kohistani 135

In this example, the utterance is the complement of the matrix clause14 “Then he
said.” The marker lee follows each clause of the quotation. However it is not neces-
sary for utterances marked by lee that they be the complement of a matrix clause.
The presence of lee alone is sufficient to mark a clause as reported information, as
the next example illustrates. In (9), the speaker is talking about a house that she
herself had not yet seen but her husband and her sons had told her about.
(9) c̣ àa kamrií ɡhẽ̀ẽ ɡhẽ̀ẽ baá-ø thé=lee
three room.pl big.m.pl big.m.pl house/room-pl be.prs.m.pl=rep
‘There are three rooms, big, big rooms, they say’
(conversation 10.9.2012)

In this example, only the presence of lee indicates that this is not the representa-
tion of a state of affairs that the speaker herself has seen, but that she is reporting
someone else’s speech.
A look at all narrative text within the data corpus shows that lee does not
mark all reported information. As mentioned above, the complementizer če is the
default quotation marker. So, apart from indicating “source of information,” lee
must have some additional meaning. Recalling the two-fold function of at least
some procedural indicators, namely constraining the search for relevance and
triggering the hearer’s epistemic vigilance mechanism, it is argued that lee is an
evidential marker that functions in both ways. By using lee the speaker signals that
she wants the hearer to take over some of the evaluation responsibility rather than
committing herself to the truth of the proposition, by explicitly laying open the
source of her information (Unger, 2012a, 2012b; Wilson, 2011). In Example (8)
above, the immediate results of a strong earthquake are told; reason enough for
the speaker to explicitly mark each clause of reported information as of someone
who was an eye witness.
All three discourse markers introduced above indicate someone else’s or self-
reported speech or thought, that is: attributive metarepresentations. The fourth
marker, loo, the main topic of this paper, is different in that it does not mark meta-
representations of utterances or thoughts attributed to someone. Rather it indicates
a type of utterance that is the metarepresentation of another utterance not yet been
uttered but desirable from the point of view of the speaker. Such metarepresenta-
tions are said to be non-attributive and will be the topic of Section 3.2.

14. A matrix clause is a clause with a dependent clause embedded within; in the cases discussed
here the dependent clause is a speech complement.
136 Beate Lubberger

3. The Indus Kohistani marker loo

This discourse marker has, like the ones introduced in the previous section, not
been described before. There is, however, one mention of loo in Buddruss (1959,
p.21); he takes note of verb forms ending in “-lō̃,” for instance “hara-lō̃” which he
translates as “er möge nehmen” [‘he may take/let him take’].15 At the same time, he
remarks that he is not sure as to how to classify such forms. At this point, I am not
able to offer any insight into the particle’s etymology.
The marker loo is an enclitic that follows its host, the clause-final verb. Unlike
Indus Kohistani independent words, loo has no pitch accent; it forms a phonologi-
cal unit with the preceding verb. The marker may precede or follow other clause-
final clitics such as question and negation markers.

3.1 Uses of the marker loo

The Indus Kohistani marker loo occurs in two different settings: (i) it follows
clauses that a speaker wants her hearer to say to a third party. To exemplify this
use, the English example given in the introduction is repeated here. Suppose I see
my sister and say to her, “If you see mother tell her that I will come on Sunday,”
or, “If you see mother, I will come on Sunday.” The underlined clauses are the ones
that in Indus Kohistani would be marked by loo. Following Wilson I will call such
utterances desirable utterances: it is desirable from the point of view of the speaker
that such an utterance be said by someone to a third person. (ii) The discourse
marker is also used to indicate third-person imperative. In the following sections,
both uses will be illustrated in more detail.

3.1.1 loo as indicator of desirable utterances

The English example above already indicates that in a discourse situation where
loo is used, there are three parties involved: a speaker A, a hearer or addressee B,
and a third person or recipient C often not present. The speaker asks the hearer to
convey a message to the third person as Example (10) illustrates. A girl had been
sent by her mother to find out if my language consultant was at home; and upon
finding her there she got the following answer.
(10) tú bá tãĩ̀ yàa-i man-á če [má baá-ø
2sg go.imp.2sg refl.poss.f mother-dat say-imp.2sg comp 1sg house-obl
man thí=loo]
in be.prs.f=dum
‘Go; tell your mother that I am at home’ (conversation 20.2.2012)

15. Author’s translation

 Chapter 5. Metarepresentation markers in Indus Kohistani 137

The message “I am at home,” that the speaker wants the girl to pass on to her
mother, is the complement of a speech verb; the clause marked by loo is an asser-
tion.
In Example (11), the utterance marked by loo contains a command. Again, the
speaker explicitly tells the hearer to pass it on by making it the complement of the
speech verb say.
(11) pií baažìi dé man-á [tú khá=loo]
over.there older.sister.dat give.imp.2sg say-imp.2sg 2sg eat.imp.2sg=dum
‘Go over and give (the cookie) to Bajii (Older Sister); tell her to eat it’
[lit.: ‘Tell her, “Eat loo!”’]  (conversation 4.6.2012)

Similarly, in (12) loo is used to mark a question that the speaker wants her ad-
dressee to ask someone else.
(12) beaatée-ø man-á [oó beaatée bi-ínt=aa=loo]
Beate.dat say-imp.2sg voc Beate go-prs.f=q=dum
‘Say to Beate, “Beate, are you leaving?”’
(conversation 25.6.2012)

In the examples seen so far, loo marks complements of speech verbs; such comple-
ments may consist of assertions, questions, negated propositions, and directives.
The absence of loo would not make any difference in the meaning of the utterances
because the speaker makes her wish to convey the desirable utterance explicit by
embedding it within a matrix clause such as “tell person X [utterance].”
The following examples show that desirable utterances do not have to be com-
plements of speech verbs; the presence of loo in an utterance is enough to indicate
that this is an utterance the speaker wishes the hearer to pass on to someone else.
Example (13) is taken from a conversation between a mother and her toddler son.
I had greeted the little one and asked how he was. The mother then said to him,
(13) [rùuɣ thú=loo]. tĩ̀ĩ ɡí hàal thí
well be.prs.m.sg=dum 2sg.poss what condition be.prs.f
‘(Say), “I am fine. How are you?” ’
[lit.: ‘(I) am fine loo. How are you?’]  (conversation 8.6.2012)

Here, the only indication that this is a clause that the speaker wants her son to
say to me is the marker loo (which, given the context, also includes the following
clause “How are you?”).
Example (14) is a negated clause. Again, loo is the only indicator of this being
an utterance that the speaker wants her addressee to repeat to a third person.
138 Beate Lubberger

(14) [ãĩ́ ṣa-áthe naíi=loo]

3pl.prox wear-prs.pfv neg=dum
‘Tell G16 that (the children) have not worn them (=the clothes are clean)’
[lit.: ‘(The children) have not worn them (the clothes) loo’]
 (conversation 24.8.2012)

Example (15) was uttered while I was working with my language consultant. In
another room of the house, a TV was blaring away. My language consultant called
one of her daughters-in-law who was in sight and told her (15),
(15) oó D khún kãã́ thú [ṣás awàaz kám
voc name.f inside who be.prs.m.sg 3sg.dem.dom voice less
kar-ìi=loo]
do-imp.2pl=dum
‘Oh D, who is inside (the room with the TV)? Tell them to turn down the
volume’
[lit.: ‘Oh D, who is inside (the room with the TV)? Turn down the volume
loo’]  (conversation 29.10.2012)

Here, the clause marked by loo contains a directive; the verb has a second-person
plural imperative suffix. Again, nothing besides the presence of loo indicates that
the speaker wishes her addressee to say this utterance to a third person.
Like clauses marked by loo that are embedded within a matrix clause with
a speech verb, independent loo-marked clauses may contain assertions, nega-
tions, questions, and directives. Example (16) is an illustration of the use of loo in
questions. It is an exchange between two speakers that happened when someone
outside was rattling the street side door of the house.
AK: maasmá thé
(16) a. báṣ hoó ta C
child.pl be.prs.m appearing become.imp.2sg dm name.f
‘There are children. Come, C!’
b. CR: maasmá thé oó kuú
child.pl be.prs.m.pl voc older.sister
‘There are children (rattling the door), Older Sister’
c. AK: [ɡií kar-àant-ø=loo] alá
what do-prs.m-pl.m=dum there
‘Ask them what they are doing there’
[lit.: ‘What are (you) doing loo there?’]  (conversation 4.2.2013)

16. Names of speakers and persons mentioned in the data have been replaced by capital letters
so as not to reveal their identity.
 Chapter 5. Metarepresentation markers in Indus Kohistani 139

The utterance within the scope of loo is the question “what are you doing?” The
presence of the marker loo indicates the only possible interpretation of utterance
(16c), namely that CR should ask the children “what are you doing?” The adverb
alá “here” in this clause is following the marker loo as an afterthought.
The next examples are of utterances marked by loo that are being reported. In
(17), an addressee B is reporting the desirable utterance of speaker A meant for
recipient C.
(17) asĩ̀ĩ yàa-e bilaál miiɡeé minát kar-álaas só
3sg.prox.poss mother-erg yesterday 1sg.dat request do-pst.pfv 3sg.dist
man-àynt mĩ̀ĩ ṣṹ surát pats-íthi [miiɡeé ɡuulìi-
say-prs.f 1sg.poss 3sg.dem private.parts be.sore-prs.pfv.f 1sg.dat pill.pl
waal-á=loo]
bring.down-imp.2sg=dum
‘Yesterday her mother asked me for a favour; she said, “I am sore; tell Beate
to bring some pills for me”’
[lit.: ‘… I am sore, bring some pills for me loo’] (conversation 20.4.2013)

In this example, “her mother” is speaker A, who said the desirable utterance
“bring pills for me”; hearer B is reporting to me, recipient C, what speaker A had
requested her to ask me. Like in the previous examples, loo seems to indicate that
the utterance marked by it should be passed on to someone else. Recipient C is not
explicitly named here because hearer B is able to infer from the context to whom
to pass on the request.
Example (18) is a report of utterances marked by loo, where the person report-
ing is neither speaker A nor hearer B nor recipient C but a fourth party.
(18) ṹ telfun-á kar-àynt khuná ɡhayàa-ø haát če
3sg.prox phone-pl do-prs.f up.valley grandmother.obl through comp
[abàa wá=loo] [mií har-á=loo] [abàa
father come.down.imp.2sg=dum 1sg.dom take-imp.2sg=dum father
wá=loo] [mií har-á=loo]
come.down.imp.2sg=dum 1sg.dom take-imp.2sg=dum
‘She is making phone calls (to the place) up-valley, via her grandmother,
saying “Tell father to come down. Tell him to take me home. Tell father to
come down. Tell him to take me home”’
[lit.: ‘She is making phone calls (to the place) up-valley, via her grandmother,
“Father, come down loo. Take me loo. Father, come down loo. Take me loo”’]
 (conversation 19.10.2012)

Here a speaker D is quoting speaker A’s utterances that she told addressee B, the
grandmother, so that addressee B would then pass on the request to recipient C,
140 Beate Lubberger

speaker A’s father. There is no speech verb, no explicit request or command to tell
C the desirable utterance; the presence of the marker loo is sufficient to make clear
the meaning of the reported speech.
In Example (19), again, there are four parties involved: a speaker A, who
gives instructions as to how to take a certain medicine to addressee B, who passes
this on to addressee C who in turn is asked to tell it to recipient D. The actual
speaker of the utterance is addressee B. I (speaker A) had given her instructions
about how to take the medicine, and she passed the instruction on to C who was
supposed to tell D.
(19) ṣás dawaí U-i dé tú man-á [če
3sg.dem.dom medicine name.f-dat give.imp.2sg 2sg say-imp2sg comp
beaatèe man-áthe=loo [ṣás dawaí dís-ø man ék
Beate.erg say-prs.pfv=dum 3sg.dem.dom medicine day.obl in one
wàar kha-ṍ thú=loo]]
time eat-inf be.prs.m.sg=dum
‘Give this medicine to U. Tell her that Beate has said to take this medicine
once a day’
[lit.: ‘…, say, “Beate said loo, ‘This medicine has to be taken once a day loo’”]
(elicited 4.5.2013)

In this construction, loo occurs twice: first it follows the speech verb in “Beate
has said” and then again at the end of the speech complement. Both “Beate has
said” and the complement containing the construction are marked by loo, and
both have to be passed on.
So far we have seen that loo marks clauses that are complements of speech verbs
as well as independent clauses, where the request to pass on the utterance is not
made explicit otherwise. In the latter kind of clauses, loo might best be translated
as “Tell person X the preceding utterance.” Looking at all these uses, loo might be
considered a marker that indicates utterances that someone wants someone else to
tell to a third or fourth person. Whoever wishes a particular utterance to be said
to someone else – this may be a speaker A, a person B or C – may use loo to mark
it. However, the person actually passing on the utterance in question will use a
reported speech marker such as karee instead of loo. Recall Example (10) where
my language helper said to a girl to tell her mother, “I am at home.” The girl will
deliver this message to her mother as illustrated in (20).
(20) koó ṣeé man-áthe če [má baá thí karee]
older.sister.erg like.this say-prs.pfv comp 1sg house be.prs.f mrm
‘The older sister has said that she is at home’
[lit.: ‘The older sister has said, “I am at home karee”’] (elicited)
 Chapter 5. Metarepresentation markers in Indus Kohistani 141

This is a typical quotation; the quoted part is embedded within a matrix clause of
the kind “person X said.” Here the actual speaker, the girl, is not requesting her
audience to pass on a desirable utterance; she is just quoting what someone else
told her, hence the utterance is not marked by loo but by karee.
Finally, what about utterances that a speaker asks her audience not to say to
someone else? This would be a case of a metarepresentation of an utterance that
is not desirable but a metarepresentation (of a thought) nonetheless. In Indus
Kohistani such an utterance is not marked by loo but by the metarepresentation
marker karee, used to indicate reported and self-reported utterances and thoughts,
as Example (21) illustrates. The context of the utterance is as follows: During one
of my visits, the family of one of my language consultant’s daughters-in-law called
and wanted to meet me at her house. As they did not live nearby my language con-
sultant decided that it would be too late that day. She instructed her daughter-in-
law to call her family and tell them that they should come next week “when Beate
would be here,” giving the impression that I was not present at the time of their call.
(21) man-á [wáxt-ø tal nika-í=loo] šóo=aa man-á naíi
say-imp.2sg time-obl on come.out-imp.2pl=dum right=q say-imp.2sg neg
[ṹ uka-íliis karee]
3sg.prox go.up-pst.pfv.f mrm
‘Tell (them) to set off early (next week), right? Do not tell them that she
(Beate) was here (today)’
(conversation 7.1.2013)

The first clause in square brackets is the metarepresentation of a desirable utter-

ance (“tell them”); the second bracketed clause is one that the speaker asks not to
tell, hence it is not marked by loo but by karee.

3.1.2 loo as indicator of third person imperative

The marker loo also occurs in utterances that express directives aimed at persons
not present; in other words, loo may indicate third person imperative. Indus
Kohistani has a morphologically marked imperative for second person singular
and plural only. To indicate second person singular imperative, the suffix -á is
added to the root of a (regularly inflected) verb, for instance the singular impera-
tive form of kar- ‘do’ is kar-á ‘(you.sg) do!’. Likewise the suffix -ìi is added to a
verb root to indicate second person plural imperative as in kar-ìi ‘(you.pl) do!’.
Quite often the singular imperative suffix is dropped, leaving the bare verb root as
in kár ‘(you.sg) do!’
In Indus Kohistani, imperatives aimed at third persons may express several
notions which are all contained in the term jussive (van der Auwera, Dobrushina
& Goussev, 2013): (i) they may express a command, (ii) they may signal the
142 Beate Lubberger

speaker’s permission to, or (iii) the speaker’s agreement with something a third
person is going to do. Syntactically, a third person imperative can be embedded
within a matrix clause such as “tell person X that [directive]”; an instance of this
would be Example (11), repeated here, where the directive is the complement of
a speech verb and is followed by loo. This form is most often used to express the
strongest notion of jussive, a command.
(22) pií baažìi dé man-á [tú khá=loo]
over.there older.sister.dat give.imp.2sg say-imp.2sg 2sg eat.imp.2sg=dum
‘Go over and give (the cookie) to Bajii (Older Sister); tell her to eat it’
[lit.: ‘Tell her, “Eat loo!”’]  (conversation 4.6.2012)

The second pattern to be observed is illustrated in (23). Again, the third person
directive is followed by loo; but here the directive is not embedded within a matrix
clause. The example is taken from a recorded text about tuberculosis, a talk given
to people waiting in a village clinic. The speaker exhorts his addressees as follows.
(23) ɡí wáxt khàaŋ iil-uú [só muúṭyõõ tã̀ã hàa
what time cough come.pfv-cond 3sg.dist in.front refl.poss.m hand
tshá=loo yaá bíɡi zòṛ dhaý=loo] če só
place.imp.2sg=dum or some clothes hold.imp.2sg=dum comp 3sg.dist.
khàaŋ pií mút màaṣ-ãã kira nií ríŋɡ-ee (TB text #32)
cough over.there other man-gen.m to neg attach-sbjv.3sg  
‘If he (the TB patient) has to cough then he should cover his mouth with his
hand or with some piece of cloth so that the cough (the droplets) does not
reach others’ (TB text #32)

Note that in this example, there is no explicit “Tell persons X and Y” matrix clause;
furthermore the subject of the clauses marked by loo is a third-person singular
pronoun whereas the verb has second-person imperative marking.
Example (24) is from the same talk about tuberculosis; again, the clauses
marked by loo contain commands.
(24) ɡhèn marìiz ṭiibĩ̀ĩ dawaí istemàal kar-àant tás
who patient TB.gen.f medicine use do-prs.m 3sg.dist.obl
bã̀ã mút xálk.õõ kira húm ṣṹ pakàar thí
house.gen.m other people.pl.gen for also 3sg.dem necessary be.prs.f
[če sãĩ́ ḍaakṭár-ø ɡee baí=loo] [yaá marìiz ḍaakṭár-ø
comp 3pl.dist doctor-obl to go.imp.2pl=dum or patient doctor-obl
ɡee har-í=loo] [ao tã̀ã ṭiibì-ãã muainá
to take-imp.2pl=dum and refl.poss.m tb-gen.m examination.m
kar-í=loo]
do-imp.2pl=dum
 Chapter 5. Metarepresentation markers in Indus Kohistani 143

‘For a patient who is taking TB medicine it is necessary that the other people
in his house should go to the doctor, or other ill persons in his house should
be taken to the doctor and should be checked for TB’  (TB text #34)

Here, the clauses marked by loo are complements not of a speech verb but of the
matrix clause “it is necessary that […].” Again, the clauses in question have a
third person (plural) subject whereas the verb is marked for second person plural
imperative, followed by loo. Although in both (23) and (24) there is no mention
of “telling someone to do something,” it is implied that people listening to the talk
pass on the commands to such persons that are ill.
This type of construction, where the imperative clause is not the complement
of a speech verb, is most often used for the weaker forms of a third-person impera-
tive, namely for permission and agreement, as the next examples illustrate. In (25)
the notion expressed by the imperative clause is one of permission. The example
is taken from a folk story about a man considered stupid by his fellow villagers
because he had burned down their forest. One day they told him that they would
kill his cow because she had eaten up their crop. The man called stupid by the oth-
ers said, “Go ahead and kill my cow, but give me the hide.” The villagers’ reaction
was as follows.
(25) xálkõõ man-áɡil ṭhiík thí ɡínče masùu béetus
people.pl.erg say-pfv2 right be.prs.f because meat 1pl.incl
kha-ṣát-ø ao tsàam-ø hin ṹ ɡí kar-àant [amãĩ́
eat-fut.m-pl.m and skin-obl with 3sg.prox what do-prs.m refl
har-á=loo]
take-imp.2sg=dum
‘The people said, “That is alright, because we will eat the meat; what can he
do with the hide! Let him take it.”’
[lit.: ‘… He you.take it loo’] (A Kohistani story #44)

The subject in the last clause that is marked by loo has been omitted but from the
preceding clause it is clear that it is ṹ ‘he’, a third person subject. Here, a translation
of the last clause marked by loo as “tell him to take the hide” might still be possible
but is less felicitous than “let him take the hide.”
The following utterance in (26) is said about a man who has caused an en-
mity and is now residing with a friend for a prescribed amount of time, during
which a mediator will try to solve the enmity peacefully. During this period of
time both parties agree that they won’t kill each other. Hence, we find the per-
missive in Example (26), uttered by the party who was grieved about the one
who caused the enmity.
144 Beate Lubberger

(26) khẽ ṹ khulàa til-á=loo

devm 3sg.prox open move-imp.2sg=dum
‘So let him move freely/he may move freely’
[lit.: ‘So he openly you.move loo’] (A way to end a feud #110)

The notion expressed in this example is again one of permission; the clause con-
tains an explicit third person subject.
The weakest form of a third person imperative, namely agreement, can be seen
in the next examples. The context of (27) is a description of what people do during
winter to keep themselves warm. The speaker tells that in the cold season, people
build an open fire place inside the house, quoting them as “Let the (inside of) the
house become blackened, but our children are kept warm.”
(27) baá kiṣĩĩ́ hó=loo
house black.inch become.imp.2sg=dum
‘Let the house become black’
[lit.: ‘House you.become black loo’] (Prep. for winter #111)

Example (28) is another instance of this use. It is part of a conversation that hap-
pened when I brought some wool that the women of A’s family had requested me
to buy for them.
(28) C: ɡõ̀õ sáx suɡ-àa a-áthe
yarn.m very fine-m.sg bring-prs.pfv
‘(She) has brought very nice wool’
A: má man-àynt heleék tìiz thú
1sg say-prs.f a.bit bright be.prs.m.sg
‘I say, it (the colour) is a bit bright’
C: ɡõ̀õ=aa
yarn=q
‘The yarn?’
clicks with tongue, indicating she does not agree
A: [hó=loo]=aa

become.imp.2sg=dum=q
‘Let it be then? [lit.: ‘You.become loo ?’]’ or a more free translation,
‘Should we take/accept it then?’
C: ɡõ̀õ sáx suɡàa thú
yarn.m very fine.m.sg be.prs.m.sg
‘The yarn is very nice’  (conversation 7.9.2012)

In fact, the expression hó=loo ‘become.2sg.imp=loo’ or more freely translated as

“let it be” or “it doesn’t matter” is used very frequently in discourse, indicating
agreement to all kinds of actions/events or states. In Example (28) above, the
 Chapter 5. Metarepresentation markers in Indus Kohistani 145

speaker has still some control in that she can refuse to accept the wool. However,
an imperative verb form followed by loo may also be used to indicate agreement
to or acceptance of circumstances one cannot change. Example (29) is taken from
a folk story about an evil fairy who has snatched away a princess’ husband. The
princess goes on a long search for her husband and finally manages to find the
fairy, get her imprisoned and set her husband free. In (29) the fairy resigns herself
to the loss of the prince.
(29) sẽ̀ẽ man-áthe hĩ̀ĩ tsõṍ mií bánd kar-áɡil. xawàand
3sg.dist.erg say-prs.pfv now 2pl.erg 1sg.dom closed do-pfv2 husband
mìn bazíɡaa. [waleé har-á=loo]. hĩ̀ĩ ɡí kar-àm
1sg.abl go.pfv2.m.sg but take-imp.2sg=dum now what do-sbjv.1sg
‘She said, “Now you have imprisoned me. The husband went away from me.
But let (the princess) take (him). What can I do now”’
[lit.: ‘She said, “Now you have imprisoned me. The husband went away from
me. But take him loo”’] (Prince and fairy #202–4)

There is nothing left to command or permit for the speaker of this utterance, only
to accept what has happened. Note that here too, the verb of the clause marked
by loo has second person imperative marking; the subject of this clause has been
omitted. From the context we know that it must be the princess who was not pres-
ent when this was uttered, that is: if the subject was to be mentioned explicitly it
would be a third person subject.

3.1.3 Summary: Uses of the marker loo

The Indus Kohistani marker loo is a discourse marker that is put to the following
uses:
1. loo follows utterances that a speaker explicitly asks her audience to pass on to
a third party. Such utterances are the complements of speech verbs. They may
contain assertions, questions, negations, and commands.
2. loo follows utterances that are not complements of a speech verb but inde-
pendent clauses. These, too, are utterances that a speaker wishes the hearer
to tell a third person. The presence of loo alone, it seems, is enough to convey
this wish. The recipient of the marked utterance has to be inferred from the
context. Again, such utterances may contain assertions, questions, negations,
and commands.
3. loo is used to indicate third person imperative meaning. As Indus Kohistani
does not have a morphological third person imperative marker, there are two
strategies for forming third person directives; both of them employ the marker
loo. Firstly, a directive aimed at a third person may be the complement of a
146 Beate Lubberger

matrix clause such as “tell person X” and is followed by loo as in Examples (11)
and (21) above. The complement itself contains a command, the addressee of
the command is an (implied) second person subject, and the verb has second
person imperative marking. The second type of third person imperative con-
struction consists of an independent clause (or of a complement other than of
a speech verb) with a third person subject, explicit or implied, but a second
person imperative suffix on the verb that is followed by loo. This latter kind of
third person imperative clause may express the notion of command and, more
often, of permission and of agreement.
The question an analysis has to answer is: does loo have two different meanings,
namely (i) indicating utterances that a speaker wants her addressee to say to
someone else and (ii) indicating third-person imperative, or is there an analysis
that can account for both uses? In the following section I will propose that loo
be analysed as a procedural indicator of one specific type of metarepresentation,
namely that of desirable utterances. In both uses, a speaker is metarepresentating
an utterance that she wishes to be said to someone else. The presence of loo makes
this wish explicit regardless of whether such an utterance is the complement of a
speech verb or not. I will further argue that loo does not encode “tell person X,”
that is, the request to pass the utterance on to a third person. This is an implica-
tion that the hearer has to infer in instances where loo is the only indicator of
a desirable utterance.

3.2 Analysis of the marker loo

The three Indus Kohistani discourse markers mentioned in Section 2.3 are all in-
dicators of attributive or self-attributive metarepresentation; an utterance marked
by any of them is the metarepresentation of someone else’s or one’s own previously
entertained and mentioned thought or utterance. The marker loo, on the other
hand, is used to indicate a different kind of metarepresentation that is not the quo-
tation of someone’s previous utterance or thought, hence that is non-attributive.
The next section will review examples of non-attributive metarepresentational use
before we will look in more detail at the one type that is indicated by the marker loo.

3.2.1 Non-attributive metarepresentations

As Wilson (2012a) points out, metarepresentational use is by no way restricted
to quotations of attributed utterances and thoughts. Example (30), (Example (6)
in Wilson, 2012a, p. 232) illustrates instances of abstract non-attributive
metarepresentation such as mentions of sentence types (30a), utterance types
 Chapter 5. Metarepresentation markers in Indus Kohistani 147

(30b), proposition types (30c)–(d), and mentions of names (30e), words (30f),
and concepts (30g).
(30) a. Dragonflies are beautiful is a sentence in English.
b. Shut up is rude.
c. It’s true that tulips are flowers.
d. Roses and daisies are flowers entails that roses are flowers.
e. I like the name Petronella.
f. Abeille is not a word of English.
g. Tulip implies flowers.

Furthermore, within relevance theory, regular (that is: non-echoic) interrogatives

and exclamations as well as regular negations and disjunctions are seen as non-
attributive metarepresentations. Within speech-act theory, regular questions have
been analysed as a subtype of directives, namely as requests for information (Bach
& Harnish, 1979, p. 40); problematic for such an analysis are those types of ques-
tions that are not uttered to request information, for instance rhetorical questions,
exam questions, self-addressed, and indirect questions. The relevance-theoretic
analysis treats interrogatives not as requests for information but as metarepresen-
tations of desirable thoughts; a speaker uttering a question is thinking about and
metarepresenting the answer that is desirable or relevant for someone (Sperber
& Wilson, 1995; Clark, 1991; Wilson 2012a). If the interrogative is a request for
information then the answer is desirable or relevant for the speaker. Rhetorical
and expository questions, on the other hand, are relevant for the hearer; a question
such as “What are the symptoms of measles: first, raised temperature, second, …”
is the metarepresentation of an answer that is desirable or relevant for the hearer.
So, the analysis of interrogatives as metarepresentations of desirable or relevant
thoughts/answers that are desirable for someone can accommodate all kinds of
questions. In the same way, exclamatives have been analysed along the same lines as
metarepresentations of desirable thoughts (Sperber & Wilson, 1995; Clark, 1991).
A further type of non-attributive metarepresentation are regular (that is,
non-attributive) negations and disjunctions. They have been analysed as meta-
representing possible thoughts (Wilson, 2012a). Likewise, there are non-attributive
metarepresentations of desirable and possible utterances. Wilson (2012a, p. 256)
mentions as examples of the latter drafts, essay plans, and also the mention of
utterance types as in (30b).
Now let us look at metarepresentations of desirable utterances in more detail;
it is this type of metarepresentation that, I claim, is indicated by the marker loo.
The English Example (31) from Noh (Example (65) in Noh, 1998, p. 273) is an
instance of a desirable utterance.
148 Beate Lubberger

(31) MARY [to Peter, as door bell rings]: If that’s John, I’m not here.

Here, Mary wants Peter to say, “Mary is not here” in case it is John who has rung
the doorbell; in her utterance, she is metarepresenting the utterance “Mary is not
here” that she wishes Peter to say. Another instance of metarepresentations of
desirable utterances is illustrated in (32) (Example (37) in Wilson, 2012a, p. 255).
(32) VICAR [to bride]: I, Amanda, take you, Bertrand, to be my lawful wedded
husband.
BRIDE: I, Amanda, take you, Bertrand, to be my lawful, wedded husband.

Here, too, the vicar is metarepresenting the utterance that he wishes the bride to
say and that is then repeated by the bride. In both examples, the desirable utterance
is not explicitly marked as such. In (31), Peter has to infer that Mary wants him
to say “Mary is not here” in case John is the one that rang. In relevance-theoretic
terms, the explicature “I am not here” has to be embedded within a higher-level
explicature such as “Mary wishes me to say to John ‘Mary is not here.’” On the
other hand, Mary could have said to Peter, “If that’s John then tell him that I am
not here,” thereby making the desirable utterance explicit in conceptual terms,
namely by saying “tell him.”
Let us look at another example. Suppose that Peter is telling Mary that he is
going to visit his mother who has been admitted to the hospital. Mary, who is on
her way out to work, replies, “I will go to see her after work.” In this context, Mary’s
utterance does not just have the aim to inform Peter that she intends to visit her
mother-in-law after work. Rather, she wishes Peter to tell his mother that Mary
will come to visit her later on after work. So here, too, Mary is metarepresenting
an utterance that she wishes Peter to say to someone else; this utterance is desirable
from Mary’s point of view. Her wish that Peter pass on the desirable utterance is
not encoded conceptually, in fact, not encoded at all. Peter has to infer from the
context that this utterance is not merely for his own information but that with it
goes the request to pass it on to his mother. Again, as in the examples above, Mary
could instead have said to Peter, “Would you tell your mom that I will come and
see her after work?” In this case, Mary would have made explicit that she wishes
Peter to pass on the desirable utterance “I will go to see her after work.”

3.2.2 Indus Kohistani loo: A procedural indicator of metarepresentations of

desirable utterances
So far, we have seen that in the English examples of metarepresentations of desir-
able utterances above, the hearer has to infer the presence and also the type of
metarepresentation. I have also shown, that such metarepresentations may be
conceptually encoded by embedding them within a matrix clause such as “tell
 Chapter 5. Metarepresentation markers in Indus Kohistani 149

person X.” In Indus Kohistani, I claim, metarepresentations of desirable utterances

have to be indicated as such, not necessarily conceptually (although they may), but
by using the procedural indicator loo. So there seem to be three ways of handling
metarepresentations of desirable utterances: (i) they may not be marked at all; the
hearer has to infer that the utterance is the metarepresentation of another, a desir-
able one, and that the speaker wishes him to repeat the utterance to someone, as in
Examples (31) and (32), (ii) the speaker’s wish to have the hearer say the desirable
utterance to someone else may be encoded conceptually by embedding the desir-
able utterance as the complement of a speech verb, after the pattern of “tell person
X [desirable utterance],” and (iii) a procedural indicator such as Indus Kohistani
loo is used to indicate metarepresentations of desirable utterances; for the time
being let us say that this marker guides the hearer towards the construction of a
higher-level explicature such as “speaker A wishes someone to say the utterance
marked by loo to someone else.” In other words, in Indus Kohistani metarepre-
sentations of desirable utterances have to be procedurally encoded.17 Consider
Example (33), uttered by my language consultant to one of her daughters-in-law
sitting with us.
(33) [čèey kar-ìi=loo]
tea do-imp.2pl=dum
‘Tell them to make tea’,
[literally: ‘Make tea loo’]  (conversation 4.6.2012)

The basic explicature of this utterance is a simple command “make tea!” The marker
loo does not encode a concept and therefore does not contribute to the explicature.
But it indicates that the utterance “make tea” is not just a regular command (which
in relevance-theoretic terms is the representation of a desirable state of affairs) but
the metarepresentation of another representation, specifically of a desirable ut-
terance that the speaker wishes someone to say to a third person. In our example,
the presence of loo guides the addressee towards the construction of a higher-level
explicature such as “the speaker wishes someone to tell someone else ‘make tea’.”
The speaker does not make explicit whom to tell to make tea or who is supposed
to do the telling and the hearer follows the relevance-theoretic comprehension

17. One of the reviewers pointed out that some of the English as well as the Indus Kohistani
examples that are not complements of SAY verbs might be ambiguous in that they might also
be interpreted as factual statements/claims. In English, if I say to someone, “if you see mother,
I will come on Sunday,” leaving out the phrase “tell her,” both meanings, “tell her I will come
on Sunday” (metarepresentation) as well as “I will come on Sunday” (statement/claim) are, at
least theoretically, possible. Considerations of relevance will guide the hearer towards the most
meaningful interpretation. In Indus Kohistani, on the other hand, the presence of loo leaves
open only one option: to treat the utterance marked by it as a metarepresentation.
150 Beate Lubberger

procedure (see Section 2.2) in order to eliminate indeterminacy (whom to tell to

make tea; who is supposed to do the telling) by accessing contextual information
along the following lines, “tea is made in the kitchen” and “my sisters-in-law are in
the kitchen” and arrives at the conclusion “my mother-in-law wishes me to tell my
sisters-in-law to make tea.” Thus the referential indeterminacies are solved and the
addressee’s expectations of relevance satisfied.
Similarly, in Example (34) below only the presence of the marker loo indicates
that the utterance preceding it is not a simple question that the speaker asks her
audience, but the metarepresentation of an utterance that she wishes the hearer to
say to someone. The context of (34) is as follows: a toddler boy was greeting me by
shaking my hand. His mother, watching us, then said to him:
[tú rùuɣ thí=aa =loo]
(34) a.
2sg well be.prs.f=q=dum
‘Say “Are you well?”’
[lit.: ‘Are you well loo?’]
Whereupon the boy said to me (b):
b. tú rùuɣ thí=aa
2sg well be.prs.f=q
‘Are you well?’  (conversation 3.9.2012)

The explicature of (34a) is “are you well?,” a simple question that the mother might
ask her son; his appropriate answer would most probably be, “yes, I am well.” The
marker loo, however, indicates that the question marked by it is not just that, but
the metarepresentation of an utterance that is desirable from the point of view of
the speaker, that she wishes the hearer to say to someone. Again, the search for
relevance is being constrained by loo towards the construction of a higher-level
explicature such as “my mother wishes me to say to someone ‘Are you well?.’”
Contextual information such as “I am shaking hands with Beate” help the little
boy to come to the conclusion “my mother wishes me to ask Beate, ‘Are you well?’”.
Therefore, his answer in (34b) is his asking me “Are you well?”
In Example (35) the desirable utterance is the complement of a speech verb.
Several people were around when my language consultant said the following utter-
ance to one of the boys present.
(35) oó V pií E-i man-á [zã̀ã kira čèey
voc name.m over.there name.f-dat say-imp.2sg 1pl.poss.m for tea
aá=loo]
bring.imp.2sg=dum
‘Oh V, tell E over there to bring tea for us’  (conversation 29.6.2012)
 Chapter 5. Metarepresentation markers in Indus Kohistani 151

Why did the speaker make explicit her wish that V tell the desirable utterance to
E by saying, “tell E” in addition to the marker loo instead of using just loo? In this
particular instance, several people were around so that it would have been difficult
for the addressee to figure out whom among them to tell to bring tea. By making
explicit in conceptual terms the recipient of the desirable utterance, the speaker
helped her addressee to save undue processing effort.
Let us look at a few more examples from Section 3.1.1. (16), repeated here,
is an exchange between two speakers that happened when someone outside was
rattling the street side door of the house.
AK: maasmá thé
(16) a. báṣ hoó ta C
child.pl be.prs.m appearing become.imp.2sg dm name.f
‘There are children. Come, C!’
b. CR: maasmá thé oó kuú
child.pl be.prs.m.pl voc older.sister
‘These are children (rattling the door), Older Sister’
c. AK: [ɡií kar-àant-ø=loo] alá
what do-prs.m-pl.m=dum there
‘Ask them what they are doing there’
[lit.: ‘What are (you) doing loo there?’]  (conversation 4.2.2013)

In (16c), the basic explicature of AK’s utterance is “what are you doing there?”
Without the marker loo, it might be very difficult or even impossible for the hearer
of this utterance to infer that this is the metarepresentation of a desirable utter-
ance. The presence of loo makes this explicit, but the hearer still has to infer both
whom the desirable utterance is intended for, and who is supposed to pass it on.
Contextual information and considerations of relevance will guide the hearer in
recovering the speaker’s meaning, and he will stop processing as soon as he has
arrived at a conclusion that seems relevant enough to him. In this example, the
hearer concludes that he should pass on the desirable utterance to the children who
are present in the immediate context.
In Example (36) the situation is a bit more complicated. The context of this
utterance is as follows: I had brought medicine for one of my language consultant’s
sons at a previous visit. She then asked me,
(36) [ãĩ́ khaá baari-aá=loo]=aa
3pl.prox eat.cvb finish.off-imp.2sg=dum=q
‘Should I tell him to finish them (pills)?’
[lit.: ‘“Finish them (pills)” loo ?’]  (conversation 14.12.2012)

The explicature of this utterance is the command “finish/take all the pills” followed
by loo, followed by a question marker. The marker loo indicates that this command
152 Beate Lubberger

is a desirable utterance causing the hearer to construct the higher-level explicature

“someone wishes someone else to say to someone, ‘finish the medicine’”. The ques-
tion marker =aa, having the whole desirable utterance within its scope, further
constrains the hearer to the explicature “does someone wish someone else to say
to someone, ‘finish the medicine’”? The rest is left to inference. The ultimate ad-
dressee of the desirable utterance must be the speaker’s son; and obviously the
speaker, my language consultant, does not want to say, “Do I wish you to tell my
son to finish the medicine?” The only interpretation of the utterance that makes
sense i.e. which is relevant to the hearer is the one “do you wish me to tell my son
to take and finish the medicine?” In fact, this utterance contains more than one
metarepresentation: (i) we have the metarepresentation of the desirable utterance
“tell X to finish the medicine,” (ii) this metarepresentation of a desirable utterance
is one that the speaker attributes to her hearer (“you wish someone to tell X to
finish the medicine”): by quoting it a second level of metarepresentation is added,
and (iii) finally by asking “do you wish someone to tell X to finish the medicine”
the speaker is metarepresenting a desirable or relevant answer i.e. we have a third
layer of metarepresentation.
The next Example (17), repeated here, is an instance of loo in a reported ut-
terance. My language consultant told me what someone else had requested from
her. This is a case of addressee B reporting to addressee C what speaker A had said.
(17) asĩ̀ĩ yàa-e bilaál miiɡeé minát kar-álaas só
3sg.prox.poss mother-erg yesterday 1sg.dat request do-pst.pfv 3sg.dist
man-àynt mĩ̀ĩ ṣṹ surát pats-íthi [miiɡeé ɡuulìi-
say-prs.f 1sg.poss 3sg.dem private.parts be.sore-prs.pfv.f 1sg.dat pill.pl
waal-á=loo]
bring.down-imp.2sg=dum
‘Yesterday her mother asked me for a favour; she said, “I am sore; tell Beate
to bring some pills for me”’[lit.: …“I am sore; bring some pills for me loo”]
  (conversation 20.4.2013)

The explicature of the quoted utterance of speaker A is “I am sore; bring some pills
for me.” loo indicates that the clause marked by it is not just a command/request
but one that the speaker wishes someone to say to someone else. Considerations of
relevance lead addressee B to the interpretation that speaker A wishes her to pass
on the request to me, addressee C.
So far I have argued that loo is a procedural marker of one specific type of non-
attributive metarepresentation, namely that of desirable utterances. The data used
in this paper indicate that the use of loo is obligatory with desirable utterances and
sufficient to constrain the hearer towards an interpretation of the concerned utter-
ance as an utterance that is not the description of a state of affairs, or a command,
 Chapter 5. Metarepresentation markers in Indus Kohistani 153

or a question, but the metarepresentation of an utterance that the speaker wishes

someone else to say to someone. Indus Kohistani allows both: marking of meta-
representations of desirable utterances by loo only, and conceptual encoding such
as “you, addressee B, tell person C…” in addition to using the procedural indica-
tor loo. It is not by chance whether a speaker uses one or the other way; rather,
considerations of relevance determine how explicit an utterance is. A speaker, in
order to make her utterance as relevant as possible for her audience, may make
more information explicit when it would be too difficult and costly to infer it from
the context. Or the speaker makes her utterance less explicit because the necessary
information can easily be recovered from the context; if made explicit it would be
redundant and would cause unnecessary processing effort. In any case, the marker
loo reduces processing effort by making explicit or more explicit that the marked
utterance is the metarepresentation of a desirable utterance, thereby helping the
hearer to construct the appropriate higher-level explicature.

3.3 The marker loo in third person imperative constructions

Next let us look at instances of what I have called third-person imperative use.
In the examples in Section 3.1.2, it seems that loo functions as a third person
imperative marker. The question is: can the analysis of loo as procedural indica-
tor of metarepresentations of desirable utterances also account for this use, or
are there two different markers loo, one discourse marker and one third person
imperative marker?
There are several reasons for disfavouring an analysis of loo as a genuine third-
person imperative marker:
i. As we have seen in Section 3.1.2, Indus Kohistani has morphologically
marked second person imperative, that is, suffixes. One would expect that in
a language, where second person imperative is indicated by a suffix, the third
person imperative would be expressed in the same way. However, we have
seen that loo is a clitic, behaving quite differently from a suffix.
ii. Languages that employ morphological marking for indicating second and
third person imperatives have distinct suffixes for each of the two. Indus
Kohistani, however, has just the second person imperative suffixes; in case of a
third person imperative, the second person imperative suffix is not replaced by
a third person imperative suffix but “topped” or augmented with the clitic loo.
iii. There is no clear-cut distinction between the use of loo as a desirable utter-
ance marker (DUM) and loo as an indicator of third person imperative, as the
examples in the next section will illustrate.
154 Beate Lubberger

3.3.1 Comparison between “desirable utterances” metarepresentations and

third person imperative constructions
As a reminder a few typical examples of both uses are repeated here. Example (14),
repeated here, contains the metarepresentation of a desirable utterance. The con-
text is such that the speaker was handing back some children’s clothes that she had
borrowed from G, telling her addressee (who was not G) the following:
(14) [ãĩ́ ṣa-áthe naíi=loo]
3pl.prox wear-prs.pfv neg=dum
‘Tell G. that (the children) have not worn them (=the clothes are clean)’
[lit.: ‘(The children) have not worn them (=the clothes) loo’]
 (conversation 24.8.2012)

The desirable utterance in this example is a negated description of a state of affairs.

The presence of loo indicates to the addressee that this utterance is not intended
for her but that the speaker wishes someone to pass on the utterance to someone
else. The rest has to be inferred (who or what is the referent of ãĩ́ ‘they’, who did not
wear the clothes, and whom to tell the utterance).
Next let us look at instances where the desirable utterance consists of a com-
mand. Example (11), repeated here, is again an instance of a metarepresentation
of a desirable utterance; here the desirable utterance consists of a command. In
Example (11), repeated here, a mother had given her toddler son a cookie for me
and then told him what to say.
(11) pií baažìi dé man-á [tú khá=loo]
over.there older.sister.dat give.imp.2sg say-imp.2sg 2sg eat.imp.2sg=dum
‘Go over and give (the cookie) to Bajii (Older Sister); tell her to eat it’
[lit.: ‘Tell her, “Eat!”’]  (conversation 4.6.2012)

Here the desirable utterance is not only indicated by loo but the speaker has made
explicit that she wants her son to say the utterance to me, by uttering “say (to her).”
The next Example (26), repeated here, is an instance of a third person impera-
tive, conveying permission.
(26) khẽ ṹ khulàa til-á=loo
devm 3sg.prox open move-imp.2sg=dum
‘Let him move freely/he may move freely’
[lit.: ‘He open you.move.imp2sg loo’] (A way to end a feud #110)

As described in Section 3.1.2, the verb of the concerned utterance is marked for
second person imperative followed by loo whereas the subject of the clause is a
third person personal pronoun. Although it would be possible to treat the utter-
ance as a desirable utterance that someone is requested to pass on to someone
 Chapter 5. Metarepresentation markers in Indus Kohistani 155

else, i.e. “tell him, ‘Move freely,’” it is less felicitous that “let him move freely” or
“he may move freely.”
Example (37) is similar in that it contains a third person imperative construc-
tion, conveying agreement, the weakest form of a third person directive. The ut-
terances are from a conversation about blood transfusion. Some people are afraid
that when they give blood to someone they themselves may die of weakness. The
speaker explained this concept to me.
(37) bé man-àant-ø raát nií dìi
1pl.excl say-prs.m-pl.m blood neg give.imp.2pl
‘We say “Do not give blood.”’
[saatãĩ́ ék màaṣ mar-á=loo]
in.the.end one man die-imp.2sg=dum
‘In the end one person (the patient in need of blood) may die/let him die.’
[lit.: ‘In the end one person “you.die” loo’]
mút màaṣ-ãã raát nheel-eé pií nií dìi
other man-gen.m blood.m take.out-cvb over.there neg give.imp.2pl
‘Do not take another man’s blood and give it (to the patient) over there’
ɡínče tèe so-ĩ́ mar-áṣat
because then 3sg.dist-also die-fut.m
‘Because then he too (the one who gave blood) will die’
  (Conception, birth #485–8)

Note that here, too, the subject of the third person imperative clause is the full
noun phrase one man whereas the verb is marked for second person imperative
and followed by loo. The only possible interpretation of the marked utterance is
“he may die”; the alternative “tell him, ‘You die’” can be ruled out because no-one
would do such a thing.
Let us first note the similarities: both kinds of construction express a third
person imperative, that is, in both instances, a speaker utters a directive meant
for a person not present. An utterance containing a directive for the person it is
addressed to is, in relevance-theoretic terms, the representation of a desirable state
of affairs, whereby the state of affairs may be desirable from the point of view of
the speaker, or from that of the addressee. So for instance the command “make
tea” represents a desirable state of affairs where the addressee makes tea. In other
words, an utterance containing a second person imperative is not a metarepresen-
tation. The Indus Kohistani third person imperative constructions on the other
hand, whether they are the complement of a speech verb or are indicated by loo
only, both metarepresent the utterance that the speaker, because the addressee of
the command is not present, wishes someone else to say.
156 Beate Lubberger

Now let us look at the differences. Syntactically, there are few differences
between the two types of utterances we are comparing here: (i) third person im-
perative clauses always have a third person subject, explicit or implied,18 whereas
desirable utterances may have an explicit second person subject, (ii) third per-
son imperatives are never complements of a speech verb such as “tell person X
[complement].” In other words, there is no explicit request to an addressee B to
reproduce the marked utterance to a recipient C. Desirable utterances as discussed
in Section 3.1.1 on the other hand may or may not be complements of a speech verb.
The main difference between the two uses, I argue, is in the interpretation
of the concerned clauses. Desirable utterances marked by loo are always meant
to be passed on by a second person to a third person. With third person im-
perative clauses marked by loo this is possible in some cases but more often not.
Example (37) below might be interpreted both ways. It is taken from a conversa-
tion between my language consultant and me. I was expecting two guests, and one
of them had asked if she might come with me to visit her. I passed on the request
and she, my consultant, answered,
(38) [é=loo] [é=loo] bhaizdá waal-á
come.imp.2sg=dum come.imp.2sg=dum both bring.down-imp.2sg
‘She may come, she may come. Bring both (of them) down (to us)’ or
‘Tell her, “Come”; tell her, “Come”; bring both (of them) down (to us)’
  (conversation 10.9.2012)

This utterance may be interpreted both ways, as a desirable utterance, to be passed

on by the addressee and as a third person imperative. However, even if the utter-
ance is interpreted as a third person imperative construction, it is still implied that
I, the addressee of this utterance, will inform my guest that she may come.
The next Example (23), repeated here, is somewhat similar to (37) in that it
allows two ways of interpretation.
(23) ɡí wáxt khàaŋ iil-uú [só muúṭyõõ tã̀ã hàa
what time cough come.pfv-cond 3sg.dist in.front refl.poss.m hand
tshá=loo yaá bíɡi zòṛ dhaý=loo] če só
place.imp.2sg=dum or some clothes hold.imp.2sg=dum comp 3sg.dist.
khàaŋ pií mút màaṣ-ãã kira nií ríŋɡ-ee
cough over.there other man-gen.m to neg attach-sbjv.3sg 

18. Indus Kohistani is a pro-drop language, that is, the subject of a clause is often omitted.
 Chapter 5. Metarepresentation markers in Indus Kohistani 157

‘If he (the TB patient) has to cough then he should cover his mouth with his
hand or with some piece of cloth so that the cough (the droplets) does not
reach others’, or ‘Tell the TB patient that if he has to cough then he should
cover his mouth with his hand or with some piece of cloth so that the cough
(the droplets) does not reach others.’  (TB text #32)
Although in this utterance, the third person imperative meaning is the first that
comes to mind, it is still possible to interpret the utterance as one that the hearer is
supposed to re-tell a TB patient.
Example (25), repeated below, is best interpreted as a third person directive,
expressing permission.
(25) xálkõõ man-áɡil ṭhiík thí ɡínče masùu béetus
people.pl.erg say-pfv2 right be.prs.f because meat 1pl.incl
kha-ṣát-ø ao tsàam-ø hin ṹ ɡí kar-àant [amãĩ́
eat-fut.m-pl.m and skin-obl with 3sg.prox what do-prs.m refl
har-á=loo]
take-imp.2sg=dum
‘The people said, “That is alright, because we will eat the meat; what can he
do with the hide! Let him take it.”’
[lit.: ‘… He you.take it loo’] (A Kohistani story #44)

Here, village people are talking about a man who himself is absent. It would still
be possible but not felicitous to say, “Someone tell him that he may take the hide.”
Now recall Example (37) from above (blood donating): this is an instance of
a third person imperative where it is not possible at all to say, “in the end, tell
that person, ‘You die’”. What these examples show us is that there is no clear-cut
distinction between the two uses, rather a gradual waning of the notion that the
utterance should be passed on by a specific person to a third person.
To summarize, there are differences between the two uses of loo: the obligatory
third person subject (explicit or implied) in clauses with third person imperative
meaning, and the absence of a (conceptually encoded or inferred) request to the
addressee to pass on the utterance to a third party. On the other hand, it is not
possible to clearly distinguish between the two different uses of loo, as a gradual
change from one to the other can be observed.

3.4 Indus Kohistani third person imperative: A special case of desirable

utterance

The question remains to be answered: is the use of loo as a third person imperative
indicator different from that as an indicator of desirable utterances? I think not.
158 Beate Lubberger

In both uses, a speaker A utters an utterance that is intended for someone not
present, to an addressee B. In both uses, the speaker metarepresents an utterance
that she wishes someone to say to the person not present. So the utterance marked
by loo is the metarepresentation of a desirable utterance in both instances. What
distinguishes uses of loo described in Section 3.1.1 from its use as a third person
imperative is that in the former, the procedural indicator seems to also encode
“tell addressee C”; whereas in the latter construction, this interpretation is pos-
sible in some but not in all cases. In light of this, the analysis of loo as presented
in Section 3.2.2 has to be slightly rephrased: I propose that loo is an indicator
of metarepresentations of desirable utterances, thereby constraining the hearer
towards the construction of a higher level explicature of the kind “the speaker
wishes someone to tell someone else [desirable utterance].” It is, however, not part
of the meaning of loo to indicate that a speaker actually expects her addressee to
fulfill her wish, that is: do the actual telling. Where a speaker expects her addressee
to tell the desirable utterance to a third person she has two options: (i) she will
embed the desirable utterance within a matrix clause such as “tell person X that
[desirable utterance],” or (ii) she will not use an explicit command but will phrase
the desirable utterance in such a way that the hearer can inferentially recover the
intended meaning, i.e. that he is supposed to pass on the desirable utterance. In
other cases, the speaker may just expect her addressee to acknowledge her wish
to say, “[desirable utterance].” Here again, it is up to the hearer to work out the
intended meaning, led by contextual information and considerations of relevance.
Let me illustrate this point with the final Example (33), repeated here. My lan-
guage consultant gave one of her daughters-in-law, who was sitting with us, the
following directive.
(33) [čèey kar-ìi=loo]
tea do-imp.2pl=dum
‘Tell them (the sisters-in-law) to make tea’,
[literally: ‘Make tea loo’]  (conversation 4.6.2012)

The utterance consists of a command, followed by loo. The marker loo indicates
that this utterance is the metarepresentation of an utterance that the speaker
wishes to be said to someone. The interpretation “tell your sisters-in-law to make
tea” is the most relevant in this specific context: a guest is present; guests have to
be served tea; tea is made in the kitchen; the sisters-in-law-are in the kitchen. The
same utterance might be said in different circumstances. Suppose I am visiting
my language consultant’s family. After some time my language consultant tells her
daughters-in-law to make tea and prepare a meal for me. I explain to her, that I
have to leave shortly and that her daughters-in-law should not take the trouble
to cook a meal, however, “čèey karìi=loo,” ‘let them make tea’, thereby indicating
 Chapter 5. Metarepresentation markers in Indus Kohistani 159

agreement to their making tea. In both scenarios the utterance marked by loo is
the same, but different contexts lead the hearer to different interpretations.
In this section, I have argued that the analysis of loo as a procedural indicator
of metarepresentations of desirable utterances covers both uses: such utterances
that a speaker wants a person B to tell a person C as well as utterances containing a
third person imperative. In both cases loo indicates the metarepresentation of a de-
sirable utterance. The hearer has to work out by inference whether just to acknowl-
edge the speaker’s wish (to pass on the desirable utterance), or to fulfil it as well.
Some final remarks are due: how did the marker loo originate, as a marker
of desirable utterances in general or as a third person imperative marker? Apart
from a dictionary (Zoller, 2005) and a few earlier data collections with only one
mention of loo (Leitner, 1893; Buddruss, 1959; Fitch & Cooper, 1985; Hallberg,
1992; Hallberg & Hallberg, 1999) there are no documented Indus Kohistani texts,
so one can only speculate. I assume that indicating desirable utterances is the basic
function of loo; its use to indicate third-person imperatives is an extension of this
basic function, presumably developed because Indus Kohistani does not have a
morphological third-person imperative marker. The use of an already available
desirable utterance marker as a jussive marker seems to be an economical choice
for this function.
Indus Kohistani has metarepresentation markers for reported speech (lee), for
reported and self-reported speech and thought (karee, če) and for desirable utter-
ances (loo). Other metarepresentational devices in Indus Kohistani are verb forms
such as perfective 2 (a marked perfective that complements the simple perfective)
and the imperfective past verb form; these indicate other types of non-attributive
metarepresentation. It seems that in Indus Kohistani discourse, the use of these
devices is not a matter of choice but is obligatory. In other words, procedural
encoding of metarepresentation is a must whereas conceptual encoding may or
may not be added, dependent on considerations of relevance. An Indus Kohistani
speaker will use conceptually encoded metarepresentational “pointers” in addition
to procedural indicators only in such cases where the recovery of the speaker’s
meaning would otherwise cost the hearer undue processing effort, would result in
ambiguity, or would not be possible at all. And as we have seen, encoding metarep-
resentations by procedural indicators still leaves a lot to be inferred.

4. Conclusion

In this paper, I have described the Indus Kohistani discourse marker loo, one of
four markers that indicate metarepresentation. The enclitic loo marks utterances
that a speaker wishes someone to say to someone else; furthermore, loo is used
160 Beate Lubberger

in third person imperative constructions. Within the framework of relevance

theory, loo has been analysed as a procedural indicator of one specific type of
non-attributive metarepresentation, namely that of desirable utterances. In both
uses, the utterance marked by loo is the metarepresentation of another utterance
that the speaker wishes someone to say to a third person. I have argued that the
so-called third person imperative use of the marker loo is just another instance of
its use as a metarepresentation marker of desirable utterances. As Indus Kohistani
grammar has no genuine third person imperative marker, the use of loo is the
standard way of expressing third person imperative meanings.
Indus Kohistani metarepresentations of desirable utterances may be the
complement of speech verbs; in such cases the speaker makes explicit her wish
for someone to say the utterance to a third person. When a desirable utterance is
indicated as such by loo only, it is left to the hearer to infer whether the speaker ac-
tually expects her addressee to say the desirable utterance to someone, or whether
she just wants her wish to be acknowledged.
To my knowledge, the marker loo is the first procedural indicator of metarep-
resentations of desirable utterances that has been described and analysed as such.
To complement the picture, I have also briefly introduced and described the
uses of the Indus Kohistani “reported” marker lee, the quotation marker karee and
the complementizer če. Central to the analysis of all of them is the notion of, and
the distinction between, different types of metarepresentation within relevance
theory. The four markers can thus be analyzed as having one common meaning,
namely that of indicating metarepresentation. What distinguishes them is the kind
of metarepresentation they mark. Relevance theory can furthermore explain the
twofold function of the evidential marker lee: indicating information source and
aiding a hearer’s epistemic assessment. Also, the analysis of the grammaticalized
marker karee as an indicator of metarepresentation may explain why karee (and
similar grammaticalized quotatives) develop in a particular direction, namely
along a cline within the domain of metarepresentation: a quotative starts out as
indicator of one type of metarepresentation (usually of public attributive ones),
then expands to include more and more other types of (self-) attributive, and later
on, also non-attributive metarepresentations.

Acknowledgements

This paper is based on Lubberger 2014 (M.A. thesis).

 Chapter 5. Metarepresentation markers in Indus Kohistani 161

Abbreviations

1 first person imp imperative

2 second person inf infinitive
3 third person inch inchoative
abl ablative incl inclusive
ads adjectival derivative suffix m masculine
caus causative mrm metarepresentation marker
comp complementizer neg negation, negative
cond conditional obl oblique
cvb converb pfv1 perfective
dat dative pfv2 marked perfective
dem demonstrative pl plural
devm development marker poss possessive
dist distal prox proximal/proximate
dm discourse marker prs present
dom differential object marking q question marker
dum desirable utterance marker redupl reduplication
echo echo formation refl reflexive
erg ergative rep reported
excl exclusive sbjv subjunctive
f feminine sg singular
fut future voc vocative
gen genitive

References

Aikhenvald, A. Y. (2004). Evidentiality. New York: Oxford University Press.

Bach, K., & Harnish, R. M. (1979). Linguistic communication and speech acts. Cambridge, MA:
MIT Press.
Bashir, E. (1996). Mosaic of tongues: Quotatives and complementizers in North-West Indo-Ary-
an, Burushaski and Balti.” In W. L. Hataway, & W. Heston (Eds.), .Studies in Pakistani Popu-
lar Culture. Lahore (Pakistan): Sang-e-Meel Publikations and Lok Virsa Publishing House.
Blakemore, D. (1987). Semantic constraints on relevance. Oxford: Basil Blackwell.
Blakemore, D. (2002). Relevance and linguistic meaning: The semantics and pragmatics of dis-
course markers. Cambridge: Cambridge University Press. ​
https://‍doi.org/10.1017/CBO9780511486456
Blass, R. (1990). Relevance relations in discourse: A study with special reference to Sissala. Cam-
bridge: Cambridge University Press.  ​https://‍doi.org/10.1017/CBO9780511586293
Buddruss, G. (1959). Kanyawali: Proben eines Maiyā̃-Dialektes aus Tangir (Hindukusch).
Münchner Studien zur Sprachwissenschaft, Beiheft B.
162 Beate Lubberger

Buddruss, G. (1987). Ein Ordal der Waigal-Kafiren des Hindukush. Cahiers Ferdinand de Sau-
ssure, No. 41, Cahier deédieé à Georges Redard, 31–43.
Chappell, H. (2008). Variation in the grammaticalization of complementizers from verba dicendi
in Sinitic languages.” Linguistic Typology Vol. 12, 1, 45–98.
Clark, W. (1991). Relevance theory and the semantics of non-declarative sentences. Ph.D. thesis
University College London.
Degener, A. (1998). Die Sprache von Nisheygram im afghanischen Hindukusch. Wiesbaden:
Harrassowitz Verlag.
Doke, C. M. (1935). Bantu linguistic terminology. London: Longmans & Green.
Faller, M. (2002). Semantics and pragmatics of evidentials in Cuzco Quechua. Ph.D. dissertation.
Stanford University.
Faller, M. (2006). Evidentiality and epistemic modality at the semantics/pragmatics interface.
http://‍web.eecs.umich.edu/~rthomaso/lpw06/fallerpaper.pdf.
Fitch, M., & Cooper, G. (1985). Report on a language and dialect survey in Kohistan district.
Journal of Central Asia Vol. 8 No. 1, 139–149.
Güldemann, T. (2008). Quotative indexes in African languages: A synchronic and diachronic
survey. Berlin, New York: Mouton de Gruyter.  ​https://‍doi.org/10.1515/9783110211450
Gutt, E.-A. (1991). Translation and relevance. 1st edn. Oxford: Blackwell.
Hallberg, D. G. (1992). The languages of Indus Kohistan. Sociolinguistic survey of Northern
Pakistan.” In C. R. Rensch, S. J. Decker, & D. G. Hallberg (Eds.), Languages of Kohistan. Vol.
1. Islamabad: National Institute of Pakistan Studies, Qaid-i-Azam University and Summer
Institute of Linguistics.
Hallberg, D. G., & Hallberg, C. E. (1999). Indus Kohistani: A preliminary phonological and
morphological analysis. Islamabad: National Institute of Pakistan Studies, Qaid-i-Azam
University and Summer Institute of Linguistics.
Haspelmath, M. (1995). The converb as a cross-linguistically valid category. In M. Haspelmath,
& E. König (Eds.), Converbs in cross-linguistic perspective. Empirical approaches to language
typology 13. Berlin, New York: Mouton de Gruyter.
Hopper, P. J., & Closs Traugott, E. (1993). Grammaticalization. Cambridge: Cambridge Univer-
sity Press.
Ifantidou, E. (2001). Evidentials and relevance. Amsterdam: John Benjamins. ​
https://‍doi.org/10.1075/pbns.86
Ifantidou-Trouki, E. (1993). Sentence adverbials and relevance. Lingua, 90, 69–90. ​
https://‍doi.org/10.1016/0024-3841(93)90061-Z
Itani, R. (1994). A relevance-based analysis of hearsay particles: Japanese utterance-final tte.
UCL Working Papers in Linguistics, 4, 215–237. http://‍www.ucl.ac.uk/psychlangsci/re-
search/linguistics/publications/wpl/94papers/ITANI.
Johanson, L. (2000). Turkic indirectives. In L. Johanson, & B. Utas (Eds.), Evidentials: Turkic,
Iranian and neighbouring languages. Berlin, New York: Mouton de Gruyter. ​
https://‍doi.org/10.1515/9783110805284.61
LaPolla, R. J. (1997). Grammaticalization as the fossilization of constraints on interpretation: To-
wards a single theory of communication, cognition, and the development of language. Randy
J. LaPolla Publications. http://‍tibeto_burman.net/rjlapolla/papers/LaPolla_1997_Gram-
maticalization_as_the_Fossilization_of_Constraints_on_Interpretation.pdf
Leitner, G. W. (2001[1893]). Dardistan in 1866, 1886 and 1893: With several appendices. New
Delhi: Bhavana Books & Prints.
 Chapter 5. Metarepresentation markers in Indus Kohistani 163

Lord, C. (1976). Evidence for syntactic reanalysis: From verb to complementizer in Kwa. In S. B.
Steever, C. A. Walker, & S. S. Mufwene (Eds.), Papers from the parasession on diachronic
syntax, Chicago Linguistic Society, 12, 179–191.
Lord, C. (1993). Historical change in serial verb construction. Amsterdam, Philadelphia: John
Benjamins.  ​https://‍doi.org/10.1075/tsl.26
Lubberger, B. (2014). A description and analysis of four metarepresentation markers of the
language Indus Kohistani. M.A. theses in linguistics at the University of North Dakota.
http://‍arts-sciences.und.edu/summer-institute-of-linguistics/theses/2014-lubberger-beate.
cfm
Noh, E-J. (1998). The semantics and pragmatics of metarepresentation in English: A relevance-
theoretic approach. P.h.D. thesis University College London.
Palmer, F. R. (2001). Mood and modality. Cambridge: Cambridge University Press. ​
https://‍doi.org/10.1017/CBO9781139167178
Papafragou, A. (2000). Modality: Issues in the semantics-pragmatics interface. Oxford: Elsevier
Science.
Papafragou, A., Li, P., Choi, Y., & Han, Ch. (2007). Evidentiality in language and cognition.
Cognition, 103 (2), 253–299.  ​https://‍doi.org/10.1016/j.cognition.2006.04.001
Saxena, A. (1995). Unidirectional grammaticalization: Diachronic and cross-linguistic evidence.
Sprachtypologie und Universalienforschung, 48.4, 350–372.
Sperber, D., Cara, F., & Girotto, V. (1995). Relevance theory explains the selection task. Cogni-
tion, 57, 31–95.  ​https://‍doi.org/10.1016/0010-0277(95)00666-M
Sperber, D., Clément, F. Heintz, Ch., Mascaro, O., Mercier, H., Origgi, G., & Wilson, D. (2010).
Epistemic vigilance. Mind & Language, 25, 359–393. ​
https://‍doi.org/10.1111/j.1468-0017.2010.01394.x
Sperber, D., & Wilson, D. (1995). Relevance, Communication and Cognition. 2nd ed. Oxford:
Blackwell Publishing.
Unger, Ch. (2012a). Epistemic vigilance and the function of procedural indicators in commu-
nication and comprehension. In E. Wałaszewska, & A. Piskorska (Eds.), Relevance theory:
More than understanding. Newcastle upon Tyne: Cambridge Scholars Publishing.
Unger, Ch. (2012b). Procedural semantics, metarepresentation, and some particles in Behdini
Kurdish.” Lingua, 122, 1613–1635.  ​https://‍doi.org/10.1016/j.lingua.2012.08.009
Unger, Ch. (2016). Evidentials, genre and epistemic vigilance. In M. Padilla Cruz (Ed.), Rele-
vance Theory. Recent Developments, Current Challenges and Future Directions. Amsterdam;
Philadelphia: John Benjamins.  ​https://‍doi.org/10.1075/pbns.268.09ung
Van der Auwera, J., Dobrushina, N., & Goussev, V. (2013). Imperative-Hortative Systems. In
M. S. Dryer, & M. Haspelmath (Eds.), The world atlas of language structures online. Leipzig:
Max Planck Institute for Evolutionary Anthropology. http://‍wals.info/chapter/72
Wharton, T. (2009). Pragmatics and non-verbal communication. Cambridge: Cambridge Univer-
sity Press.  ​https://‍doi.org/10.1017/CBO9780511635649
Wilson, D. (2011). The conceptual-procedural distinction: Past, present and future. In V. Escan-
dell-Vidal, M. Leonetti, & A. Ahern (Eds.), Procedural Meaning: Problems and Perspectives
(pp. 3–31). Bingley, UK: Emerald.  ​https://‍doi.org/10.1108/S1472-7870(2011)0000025005
Wilson, D. (2012a). Metarepresentation in linguistic communication. In D. Wilson, & D. Sper-
ber Meaning and Relevance (pp. 230–258). New York: Cambridge University Press. ​
https://‍doi.org/10.1017/CBO9781139028370.014
164 Beate Lubberger

Wilson, D. (2012b). Modality and the conceptual-procedural distinction. In Relevance Theory:

More than Understanding. In E. Wałaszewska, & A. Piskorska (Eds.), Relevance theory: More
than understanding. (pp. 23–43). Newcastle upon Tyne: Cambridge Scholars Publishing.
Wilson, D., & D. Sperber. (2012). Mood and the analysis of non-declarative sentences. In D.
Wilson, & D. Sperber, Meaning and Relevance (pp. 230–258). New York: Cambridge Uni-
versity Press.
Zoller, C. P. (2005). A grammar and dictionary of Indus Kohistan. Vol. 1: Dictionary. Berlin, New
York: Mouton de Gruyter.  ​https://‍doi.org/10.1515/9783110197303
Part 3

Multimodality and style

 Chapter 6

When EVERYTHING STANDS OUT,

Nothing Does
Typography, expectations and procedures

Kate Scott and Rebecca Jackson

Kingston University London, UK / Independent researcher

Writers often use typographical features such as capitalization, boldface or

underlining to draw attention to particular words or phrases in a written text.
In this article we use ideas from the relevance-theoretic pragmatic framework
to discuss how use of typographical features may convey meaning and/or
produce stylistic or literary effects in written texts. We show how typography,
like prosody in spoken language, can be used intentionally to guide a reader’s in-
terpretation. We discuss a range of examples to show how typographical features
are interpreted relative to the rest of the text and relative to the cultural norms
and expectations associated with a particular language and writing system.

Keywords: typography, orthography, natural highlighting, prosody,

paralinguistic information, punctuation, procedural meaning, contrast, hearer
expectations

1. Introduction

Whenever we produce a written utterance we make decisions about how that ut-
terance will look. Will the letters be uppercase or lowercase? Should any of the
characters be underlined or presented in bold? If we are typing or using computer
software, which font and typeface should we use?1 Walker (2001, p. 10) defines
typography as “the visual organisation of written language,” whether handwritten
or produced via mechanical or digital means. As Butterick (2010) explains “every
time you put words on a page, you’ve made typography happen,” and crucial to our

1. Typeface being the style of text, for example, Times New Roman or Garamond, and font being
the specific size and weight of a character within a typeface, for example, italicized Arial at 12
point.

https://‍doi.org/10.1075/ftl.8.06sco
© 2020 John Benjamins Publishing Company
168 Kate Scott and Rebecca Jackson

discussions in this chapter is that, as Crystal (1998, p. 11) puts it, “it is possible for
us to use typographical features to change the linguistic meaning of what we write”.
Consider, for example, the use of italics in Examples (1)–(4):
(1) Andrew threw to Patrick and then he threw to Matthew.
(2) …duly rehearsing a small ceremony known in high ecclesiastical life as the
nolo episcopari…  (George Eliot, Silas Marner, p. 73).
(3) Oh – you’re Jordan Baker  (F. Scott Fitzgerald, The Great Gatsby, p. 23).
(4) …there was a private sale at the boutique on Madison… two weeks ago! and
though I figured out that one of the doormen probably withheld the card to
piss me off, it still doesn’t erase the fact that I missed the fucking sale… 
 (Bret Easton Ellis, American Psycho p. 162).

In Example (1) the italicisation of the pronoun he changes the way in which refer-
ence is resolved, paralleling the effect of contrastive stress in spoken utterances
(Wilson & Wharton, 2006; Scott, 2017), in (2), italics are used to present a Latin
term in an English text, and in (3) and (4), the use of italics highlights particular
parts of the utterance, creating a sense of contrast and emphasis respectively, and
communicating the speaker’s attitude or emotion.
In this chapter we start from the assumption that the typographic choices
made by writers are not random, but are driven by communicative intentions
and, more specifically, by considerations of relevance. We consider the interaction
that these choices have with the meaning conveyed by the utterance, focusing in
particular on the sense of emphasis associated with capitalisation and italicisation
of the letterform. We apply ideas and principles from relevance theory (Sperber
& Wilson, 1995; Blakemore, 2002; Carston, 2002; Wilson & Sperber, 2012) to try
to understand the cognitive processes that underlie the relation between features
of the text and the meaning that the text conveys, and as such we focus on the
pragmatics of typography. We will show that not only can the typographic features
change the linguistic meaning of what is written, but they can also affect what is
implicitly conveyed by the text.
We start in Section 2 by discussing the relation between typography and
linguistics, and drawing parallels with the work done by prosody and intonation2
in spoken utterances. In Section 3 we introduce key ideas from the relevance-
theoretic pragmatic framework, which are then used in Section 4 to suggest an

2. In this chapter we use prosody as a general term for the suprasegmental phonological features
of spoken language including intonation, tone, stress and rhythm. Some of the works we cite
refer specifically to intonation, but we are not concerned with the distinction here, and no
significant distinction is intended in terms of our arguments.
 Chapter 6. Typography, expectations and procedures 169

analysis of typography used for contrastive effect. We demonstrate that variation

in the representation of the letterform and specifically in the use of capitalisation
or italicisation of phrases, words or parts of words can affect the interpretation of
the utterance by interacting with context specific expectations of the reader. We
conclude that typography is a broadly procedural means of guiding a hearer to
an intended interpretation. As we acknowledge in the final section, this approach
might then be used in the future to analyse further communicative effects associ-
ated with typeface, use of colour and other visual features.

2. Typography and language

2.1 Typography and linguistics

Style guides and manuals of good practice such as Butterick’s Practical Typography
(Butterick, 2010) provide rules and guidelines for effective use of typographical
resources. However, very little work has been done on applying what we know
about language and communication in general to the realm of typography. In this
chapter we seek to begin redressing this balance by considering some of the ways
in which the visual representation of text on a page can influence what that text
communicates. Specifically, we consider the inferential processes that a reader
undertakes, and how a writer can use typographical features and qualities to guide
the reader to a particular interpretation.
The relation and interaction between typography and linguistics is often re-
marked upon in the literature but is, as yet, relatively unexamined. Both linguists and
typographers have suggested that each discipline might learn from the other. As van
Leeuwen (2006, p. 138) notes “[l]‍inguists and semioticians are lagging behind…in
their recognition of the communicative role of typography”. Twyman (1982, p. 16)
suggests that there is a need to “engage in serious study of verbal graphic language3
in much the same way as linguistic scientists have studied spoken language”. Walker
(2001, p. 15) attempts to redress the balance as she examines the “social, cultural
and technological factors that affect the decisions we make about the visual or-
ganisation of our writing”. However, as she notes, unlike linguistics, typography
does not have “a formal established, descriptive tradition”. As we will argue, insights
from cognitive pragmatics are ripe to fill this gap. More specifically, we suggest that
relevance theory offers a cognitive framework in which we can understand how the
ostensive forms chosen by writers interact with the meaning conveyed.

3. Twyman uses the phrase “verbal graphic language” to distinguish graphic representations of
words from pictorial or schematic graphics.
170 Kate Scott and Rebecca Jackson

The role that typography plays in the communication of meaning is also much
discussed in existing literature on typography. In a classic typographical work
Beatrice Warde (1955, p. 11) describes typography as the “invisible carrier of con-
tent”. She uses the analogy of a crystal goblet which carries a fine wine and allows
the inherent virtues of the wine to shine through without distraction. However,
Butterick (2016) sees this analogy as problematic. Rather, he suggests that typog-
raphy does not just carry the text, it “gives shape and visibility to something that
otherwise cannot be seen”. He sees typography and text as being in a complemen-
tary relationship since typography “reinforces the meaning” in the text.
Twyman (1982, p. 2) discusses the “relationship between information content
and its visual presentation” and makes a distinction between what he calls the
“intrinsic” and “extrinsic” features of written language. “Intrinsic features” are
those which “reside in the characters themselves” (Twyman, 1982, p. 11), includ-
ing “italics, bolds, small capitals, size of letterforms and style of letterforms”. These
features are part of a given font or character set. “Extrinsic features” are those
relating to the layout of the text on the page, the configuration of characters and
their colour. In Twyman’s terms, this means “what can be done to those characters
or sets of characters” (p.11). In this chapter we focus on the intrinsic features of
written language. There is undoubtedly much to be said about the corresponding
extrinsic features.4 However, since typography generally has been understudied
in cognitive pragmatics, we first consider so-called intrinsic features, turning to
extrinsic features in future work.
Sadokierski (2011) discusses the use of non-standard typographical devices in
literary fiction. For her, as for Warde, in literary fiction “effective typography…is
essentially invisible” (2011, p. 103). This invisibility is maintained by conforming
to typographical conventions, and it encourages the reader to become immersed
in the fictional world of the text. However, as she discusses, this invisibility may
be breached, either subtly to signify “shifts in time, place, narrative voice or plot”
(2011, p. 104), or more disruptively to draw the reader’s attention “back to the
printed page, back to the material surface of the book” (ibid). Focusing mostly on
extrinsic features, Sadokierski discusses examples to illustrate the range of uses
to which typographical devices might be put. These include indicating a shift in

4. An anonymous reviewer questioned why colour should be classified as an extrinsic, rather

than intrinsic feature, given that bolding could be seen as an intensification of colour. We agree
that there is likely to be some overlap between the feature categories here, not least due to the
increased control that digital technology now affords individual writers. We agree that our
analysis could easily be extended to incorporate changes in colour, and we hope to do this in
future work. Therefore, while we restrict our focus here to just those features that fall under
Twyman’s classification, we acknowledge that this distinction should be revisited in future work,
and in particular in light of technological advances in printing and word-processing.
 Chapter 6. Typography, expectations and procedures 171

narrative voice or pace, and inclusion of implied ephemera items such as letters,
bill posters or text messages. However, whatever the effect, as Sadokierski explains,
“[i]‍f an author consciously disturbs typographical conventions, we must assume
it is meaningful”. We agree with this intuition and will formalize this in relevance-
theoretic terms to provide a cognitive explanation for why this should be.
We are not, of course, the first to apply cognitive pragmatic principles to
the issue of how the graphical representation of text can interact with meaning.
Sasamoto (2014) and Sasamoto and O’Hagan (This volume) apply relevance-
theoretic ideas to impact captions (telops) used on television shows, and they
conclude that typographical features of these captions can “contribute to relevance
by functioning as a highlighting device to guide viewer’s interpretation processes”
(Sasamoto & O’Hagan, This volume, p. 217). A key assumption that we share with
Sasamoto et al. is that “the choice of typography and the consequences it has for
communication, does not require a treatment as a special case of communica-
tion” (Sasamoto & O’Hagan, This volume, p. 204) but rather falls out from general
pragmatic principles and interpretation heuristics.
We, like Crystal (1998), van Leeuwen (2006) and Twyman (1982) believe
that there is much that typographers can learn from work in linguistics and vice
versa. For the purposes of this paper we are concerned with what we might learn
about how intrinsic typographical variation interacts with meaning, by applying
principles from the relevance-theoretic pragmatic framework.
A key difference between written and spoken language is the richness of the
social, contextual and prosodic cues available to guide the hearer’s interpretation.
In face-to-face spoken interaction a speaker may use gesture and facial expression
to communicate their attitudes or emotions as they speak, and the prosodic quali-
ties of a spoken utterance can affect what is communicated at various levels and in
a variety of ways (House, 2006; Wilson & Wharton, 2006; Wharton, 2009). These
clues to speaker meaning are unavailable in written texts, and typographical fea-
tures including letterform and punctuation go some way in performing the same
functions. In the following section, we consider the parallels between prosody in
spoken language and typography in written language, and ask how ideas from the
analysis of prosodic features can help us understand how typography interacts
with speaker meaning.

2.2 Typography and prosody

The parallels between prosody and typography are discussed by several typogra-
phers. Walker (2001, p. 12) notes that, like prosody, graphic devices can be used
for “a number of communicative functions depending on context”. Both Warde
172 Kate Scott and Rebecca Jackson

(1955) and van Leeuwen (2006) draw parallels between the form of the text and
the spoken voice. For van Leeuwen (2006, p. 140) letterforms:
can only be more or less readable and more or less aesthetically pleasing, just as,
to the traditional linguist, speech sounds may well be more or less clearly and/or
beautifully pronounced, but cannot carry meaning on their own

For Warde, “[t]‍ype well used is invisible as type, just as the perfect talking voice is
the unnoticed vehicle for the transmission of words, ideas”. We feel that this char-
acterisation of speech sounds is over-simplistic. Speech is not just a concatenate
string of words with encoded meanings. A speaker’s voice may convey much more
than this. Our voices can convey social information about our regional origins and
age, for example. Our voices can be used to generate novel onomatopoeiac coinages
whose effects are not semantically encoded, but are derived by exploiting resem-
blances of sound in an attempt to recreate and communicate sensory experiences
which cannot be linguistically encoded (Sasamoto & Jackson, 2016). Nevertheless,
there are fruitful parallels to draw between prosody and typography. The shared
quality picked up by van Leeuwen and Warde in their discussions is central to our
analysis. Both prosody and typography are suprasegmental in that they layer on
top of the encoded semantic meaning of word forms. It is impossible to produce a
spoken utterance without using pitch, tone and stress patterns, and it is impossible
to produce a written utterance without giving the letters a physical form.
Another assumption that is key to our analysis is that prosodic and typo-
graphical elements share similarities in terms of how they contribute to the
interpretation process. Fodor (2002) hypothesised that readers apply an internal
voice to what they read. Moreover, it seems that typography and other written
elements such as punctuation contribute to how this “inner prosody” is applied,
and also to the prosodic patterns that are used when reading aloud. Pekkola et al.
(2005), Yao et al. (2011) and Perrone-Bertolloti et al. (2012) present evidence
that the auditory cortex is activated when we engage in silent reading, suggest-
ing that some experience of sound is associated with the reading process. In a
study of punctuation and prosody in silent reading, Drury et al. (2016) found that
silent readers appear to activate prosodic patterns which affect how information
is processed “downstream”. Their Event Related Potential (ERP) study suggested
that elements of written utterances such as commas have the potential to resolve
incongruities or promote particular readings, and are comparable in their effect
to prosodic features of spoken utterances such as intonation boundaries. Berger
et al. (2016) studied the effect of different typefaces on utterances that were read
aloud by speakers of German, and they determined that different fonts influence
prosodic features such as fundamental frequency, intensity and tempo, while
Gross et al. (2014) found that presenting utterance constituents in capitals induced
 Chapter 6. Typography, expectations and procedures 173

focus effects in participants’ internal prosody when reading. The picture that
emerges here is one where we apply internal or external prosody to what we read,
and where borderline and non-linguistic formal elements of written speech both
directly influence and behave similarly to prosodic cues in spoken language.
A further connection between prosody and typography lies in the observation
that both can be used to perform a range of similar functions. Both House (2006)
and Wharton (2009; 2012) discuss the different roles that prosody can perform.
House (2006, p. 1544) identifies what she calls three “functional orientations” that
intonation may have in a given utterance. First, there are speaker-oriented prop-
erties which include identity indicators and paralinguistic information related
to attitude, emotion and affect. Second, there are linguistic-oriented properties
where “intonation is used as an integral part of the linguistic form to contribute to
the propositional content in some way”. Finally, House discusses the interactional
orientation in which intonation may play a role in constructing and managing the
discourse. Wharton (2012, p. 98) also discusses the variety of ways in which “the
way we say what we say is capable of conferring on those words entirely new layers
of meaning”. He identifies a continuum of functions from the properly linguistic,
such as “tonal lexical contrasts in languages such as Burmese and Thai,” through
uses which convey “information about constituency relations and grammatical
structure,” to those which communicate information about the speaker’s attitude
or emotions. What these various functions have in common is that they “guide the
utterance interpretation process by altering the salience of possible interpretations
of utterances” (Wharton, 2009, p. 141). Clearly, while such prosodic guidance
is not available for use in written language, we might predict that writers would
employ analogous written means for guiding readers towards intended meanings,
and this is the prediction that we explore here.
Typography, punctuation and the language specific conventions associated with
them go some way to performing the same range of functions in written language
that prosody performs in spoken utterances. Consider the examples in (5)–(9):
(5) You’re going to Alicante.
(6) You’re going to Alicante?
(7) You’re going to Alicante!
(8) The kittens were black and white.
(9) The kittens were black, and white.

Whereas the difference between the statement in (5) and the question in (6) may
be indicated via the tone used when spoken, this is indicated by punctuation
when written. Likewise, while the speaker’s attitude of surprise or excitement in
174 Kate Scott and Rebecca Jackson

(7) may be conveyed by use of a wide pitch range when spoken, it is indicated by
the use of an exclamation mark when written. A speaker can guide the hearer to
the intended interpretation in (8) and (9) by how she chunks the utterance into
intonation phrases (IPs). An IP boundary between black and and indicates that
there are two types of kitten: black and white, whereas a single IP indicates that
all of the kittens are part-black and part-white. The same distinction is achieved
via use of a comma in the written version. In Examples (1) and (3) above, italics
are used to indicate that a contrast is intended, and this would be achieved via
contrastive stress placement when spoken (Wilson & Wharton, 2006; Scott, 2017).
Finally, the italics used to convey the narrator’s attitude in Example (4), would, we
assume, be paralleled by an emphatic intonation pattern if spoken.
While there are clear parallels between the two systems, they do not line up
exactly. Neither punctuation nor typography is a direct encoding of intonation or
prosody. Wharton (2009) following Sebeok (1972) discusses the contrast between
digital and analogue communicative systems and the role they play in conveying
meaning. Punctuation is, for the most part, digital. It is either present or absent.
For example, a written utterance in English either ends with a question mark or
it does not. However, when a speaker utters a spoken question in English there
are a range of prosodic patterns she might use (Wells, 2006; Clark, 2012), and she
can vary her pitch by degrees to communicate information about her attitudes,
emotions and expectations.
Some communicative aspects of prosody have no direct counterpart in tradi-
tional written representations of language. For example, there is no conventional
written indicator for the tone of voice often associated with irony (Bryant & Fox
Tree, 2005).5 There is, however, some evidence that conventions for indicating
irony may be emerging within computer-mediated-communication (CMC). Work
by Shovholt, Grønning and Kankaanranta (2014), Filik et al. (2015), Thompson
& Filik (2016) has found that certain emoticons including those representing
acts of winking and tongue poking were consistently, although not exclusively,
used to mark sarcasm. Thompson and Filik (2016, p. 177) suggest that this usage
“may be partly conventionalised” (p. 177). It is perhaps not surprising to find such
conventions emerging in mediated discourse where users are stripped of many
contextual cues, the audience is often “imagined” rather than known (boyd, 2010;
Marwick & boyd, 2010; Litt, 2012; Tagg & Seargeant 2014), and in which contexts
of interpretation are “collapsed” (Marwich & boyd 2010; Wesch 2009).

5. In recent history, new punctuation marks such as the ‘snark mark’ have been proposed, but
none of these has ever caught on.
 Chapter 6. Typography, expectations and procedures 175

Finally, as discussed in Scott (2018), instances of punctuation and typography

entering spoken language, as in the Examples (10)–(12), suggest that they perform
a communicative role not directly available via purely prosodic means.
(10) She means trouble with a capital T  (BYU-BNC B20)6
(11) I don’t like walking full stop  (BYU-BNC KD1)
(12) …I just happened to find this woman and she was The Person That Knew, in,
you know, in initial caps…  (Daniels, 2016)

Crystal (1998) discusses cases where the capitalisation of a particular noun

makes a difference to the concept that is being communicated, as is the case,
for example with western Australia versus Western Australia. He notes how we
sometimes express this contrast in spoken language by describing the typography
out loud, as in (13):
(13) Bible with a capital B  (Crystal, 1998, p. 15)

Finally, with the case of so-called “air quotes,” we find an orthographical conven-
tion represented in face-to-face communication via an iconic gesture (Lampert
2013; Cirillo 2019). This gesture prototypically involves the speaker “raising and
flexing the index and middle finger of both hands twice,” and it is, Cirillo (2019,
p. 12) suggests, a “kinesic representation of a punctuation mark”. Bolinger (1983),
points out that prosody may be viewed as part of the “gestural complex” with
its origin in iconicity and working in tandem with facial expressions and bodily
gestures. It is therefore perhaps not surprising that we find gestures used alongside
prosody to perform these functions in spoken interaction. Given these parallels
between prosody and typography, we ask if we can learn anything about how
typography interacts with meaning from existing work and ideas on prosody and
interpretation. Wilson and Wharton (2006), Wharton (2009; 2012), Clark (2012;
2013) and Scott (2017) are amongst those who offer analyses of prosody which
use ideas from the relevance-theoretic pragmatic framework. In the next section,
we introduce this framework and the associated notion of procedural meaning,
before going on to discuss how these same ideas might be insightfully applied to
the typographical data.

6. Examples of usage taken from the British National Corpus (BNC) were obtained under the
terms of the BNC End User Licence. Copyright in the individual texts cited resides with the
original IPR holders. For information and licensing conditions relating to the BNC, please see
the web site at http://‍www.natcorp.ox.ac.uk
176 Kate Scott and Rebecca Jackson

3. Relevance theory and procedural meaning

Relevance theory is a cognitive pragmatic framework. It is concerned with the

relationship between form and meaning in context, and the cognitive mechanisms
and principles which underpin the interpretation process. Although applied in
the analysis of linguistic phenomena, it is not a linguistic framework. Relevance
theory recognises the communicative potential of both verbal and non-verbal
acts, and considers both to be rich and worthy of pragmatic study. As such, it
lends itself well to an exploration of how typography communicates meaning. The
use of relevance theory is particularly appropriate when parallels are to be drawn
between typographic elements and aspects of prosody that have already been
treated as non-linguistic within the framework (Wharton, 2009; Wharton, 2012).
In relevance theory, communication is seen as ostensive and inferential.
Communicators provide deliberate and attention-attracting evidence for their
communicative intentions, and audiences use these ostensive acts to infer com-
municator meanings in context. Key to this chapter is the idea that audiences have
particular expectations in communication. On the one hand, they have expecta-
tions about aspects of the context itself, and, on the other, they have expectations
about the nature of effects available for recovery. Within relevance theory, expec-
tations about effects fall out from the way that human cognition is considered
to proceed generally. An ostensive stimulus is relevant to an individual when
its processing yields a positive cognitive effect and the positive cognitive effect
justifies the cognitive effort expended in processing that stimulus. Relevance, as
a property, is not fixed in advance, and different stimuli are relevant to different
people in different contexts.
A central assumption in relevance theory is the Cognitive Principle of
Relevance. According to this principle, human cognition is geared towards the
maximisation of relevance. Humans have a general cognitive goal of maximis-
ing the amount of information they derive from processing the world around
them, and they will invest attentional resources into inputs and stimuli that are
likely to result in positive cognitive effects. However, in the specific domain of
ostensive-inferential communication, humans are not entitled to expect that all
inputs will be maximally relevant. When a communicator produces an ostensive
stimulus, there are constraining factors which interact with potential cognitive
effects and with the effort involved in their recovery. On the one hand, a speaker
may be constrained by her ability; she may, for example be tired, ill, drunk or less
eloquent than others, or she may simply not have all of the relevant information.
On the other hand, a speaker may be constrained by her own preferences. She may
be unwilling to provide a more relevant utterance as it would not be in her best
interests to do so. The abilities and preferences of our co-communicators mean
 Chapter 6. Typography, expectations and procedures 177

that we are not entitled to expect them to be maximally relevant. Nevertheless,

since individuals make overt demands on our processing resources by virtue of
ostensively attracting our attention, we are entitled to expect that the stimulus will
be at least worth our while to process. Sperber and Wilson (1995, p. 270) formu-
lated these expectations into their definition of optimal relevance, given in (14):
(14) An utterance or other ostensive stimulus will be optimally relevant if:
a. it is relevant enough for it to be worth the addressee’s effort to process it,
and
b. it is the most relevant one compatible with the communicator’s abilities
and preferences.

According to the second, or Communicative, Principle of Relevance, “[e]‍very act

of ostensive communication communicates a presumption of its own optimal
relevance” (Sperber & Wilson, 1995, p. 158). This second principle has impor-
tant consequences that will be key for our account of how typography interacts
with meaning.
First, if humans seek to optimise the relevance of ostensive stimuli as outlined
above, this suggests the existence of a heuristic for efficient interpretation, and, in-
deed, relevance theory proposes such a heuristic: the two-part relevance-theoretic
comprehension procedure, given in (15):
(15) a.
Follow a path of least effort in computing cognitive effects: Test
interpretive hypotheses (disambiguations, reference resolutions and so
on) in order of accessibility.
b. Stop when your expectations of (optimal) relevance are satisfied.

The principles of relevance and the resulting comprehension procedure have two
important consequences which will play a key role in our account of typography
and meaning, and which we consider in turn. First, if a hearer is put to extra ef-
fort, he can expect to derive extra cognitive effects to offset and justify this effort.
Sperber and Wilson (1995, pp. 219–224) discuss how this balance between effect
and effort can explain some poetic and stylistic effects, including those which arise
from repetition in examples like (16) and (17).
(16) I shall never, never smoke again.
(17) My childhood days are gone, gone.

As Sperber and Wilson (1995, p. 221) explain, “the task of the hearer faced with
these utterances is to reconcile the fact that a certain expression has been repeated
with the assumption that optimal relevance has been aimed at”. The increase in
linguistic processing effort incurred by the repetition must be offset by an increase
178 Kate Scott and Rebecca Jackson

in cognitive effects. How this increase is realised will be context dependent, and
again, the hearer will follow a path of least effort and test out the most acces-
sible interpretations first. The speaker of (16) may be interpreted as conveying
a greater commitment to the assumption expressed than if she had not repeated
the word never, whereas the extra effects in (17) relate to communication of the
speaker’s attitude and emotions. As Sperber and Wilson (1995, p. 222) suggest,
(17) might be interpreted as conveying that the “speaker is experiencing a torrent
of memories”. See Jackson (2016) for a detailed discussion of repetition from a
relevance-theoretic perspective.
The second consequence of the principles of relevance is that we might expect
to find expressions and devices that contribute to relevance, not by increasing the
cognitive effects, but by guiding the hearer’s interpretation processes, and thus
reducing both the risk of misunderstanding and the processing effort required.
First recognised and analysed by Blakemore (1987; 2002), so-called procedural
expressions perform this function. Most content words (cat, sing, purple) are
generally considered to encode concepts which figure as representations in the
language of thought. However, the purpose of procedural terms seems to be to
provide instructions or guidance to the hearer on how he should process those
representations. A typical example of procedurally encoded meaning would be the
discourse connectives so and after all in Examples (18) and (19):
(18) Kelly can’t come to the party tonight, so she’s going to stay in Manchester.
(19) Kelly can’t come to the party tonight. After all, she’s going to stay in
Manchester.

The choice of discourse connective guides the hearer as to whether she should
interpret Kelly’s staying in Manchester as a cause or consequence of her not being
able to go to the party. Since the idea of procedural meaning was first developed,
it has been used to analyse a range of expressions including pronouns (Wilson
& Sperber, 1993; Hedley, 2005; Scott, 2016), mood indicators, discourse particles
and, significantly for our purposes here, prosody (Imai, 1998; Escandell-Vidal,
1998; Fretheim, 2002; Clark, 2012; Clark, 2013). Wilson and Wharton (2006) and
Wharton (Wharton, 2009; Wharton, 2012) make the case for procedural analyses
of both properly linguistic and natural, affective functions of prosody. Scott (2017)
argues for a procedural analysis of contrastive prosodic stress, and demonstrates
that the communicative effects of the contrast must be worked out relative to the
discourse context. For example, in (20), contrastive stress on the pronoun he af-
fects the propositional, truth-conditional content of the utterance by encouraging
reference to be resolved, not on Andrew, but on another salient male. In (21),
 Chapter 6. Typography, expectations and procedures 179

however, the stress on she does not necessarily affect how reference is resolved, but
can, instead, be used to communicate the speaker’s attitude towards the referent.
(20) Andrew hit Ben and then HE hit Farooq  (Scott, 2017, p. 332).
(21) Nobody told me SHE was coming  (Scott, 2017, p. 336).

Given the parallels between prosody and typography, we suggest that the notion
of procedural meaning might also be insightfully applied to cases where the typo-
graphical representation of a written utterance affects its interpretation, and we
attempt to do this in the next section.

4. Typography, procedures and relevance

4.1 Typography and interpretation

In this section, we make the case for analysing typography in written language
as a procedural device. According to Wilson (2011, p. 20) “[m]‍ost languages…
have a cluster of procedural items (e.g. punctuation, prosody and various types
of discourse particle)…whose function is to guide the comprehension process in
one direction or another”. Drawing on the parallel work on prosody, we use ideas
from the relevance-theoretic framework to analyse cases of emphatic contrastive
typography.
While the understanding of the role and scope of procedural meaning in
utterance interpretation has grown significantly since it was first developed by
Blakemore, recent work by Wilson (2016) and Carston (2016) has identified
two characteristics shared by those devices standardly analysed as procedural.
First, according to Wilson (2016, p. 11) “the function of procedural expressions
is to activate domain-specific procedures which may be exploited in inferential
communication”. That is, procedural expressions have a triggering effect. Second,
according to Carston (2016, p. 159), “what they do is constrain and guide prag-
matic processes which are essential in deriving the intended interpretation”. In
the remainder of this chapter, we argue that typography can be used by a writer to
both trigger inferential processes and to guide those processes. As we noted above,
typography is not an expression in its own right, but rather, like prosody, it layers
on top of linguistic expressions. Therefore, we might characterise typography as
a procedural device rather than a procedural expression. Typography does not
encode a procedure, but rather interacts with general pragmatic principles and
with the relevance-theoretic comprehension procedure to both trigger and guide
inferential interpretation processes.
180 Kate Scott and Rebecca Jackson

For the purposes of this chapter, we focus on cases where typography is used to
create a contrastive effect akin to contrastive stress in spoken language. Consider
the examples in (22)–(25), adapted from House (2006). House uses these examples
as typographical representations of contrastive prosodic stress in spoken language.
However, written versions of the utterances follow the same interpretative patterns
in their own right.
(22) Jenny teaches violin
(23) JENNY teaches violin
(24) Jenny TEACHES violin
(25) Jenny teaches VIOLIN

In each of these utterances, the explicit, truth-conditional meaning is the same.

Each is true if and only if Jenny teaches violin. However, in Examples (22)–(25),
the reader is encouraged to infer that there is some relevant contrast between
the described state of affairs and another possible state of affairs. For example,
in (23) the reader is encouraged to search for relevance in the fact that it is Jenny
and not someone else who teaches violin. The relation between the typographical
representation of (23) and its contrastive interpretation is twofold. First, the search
for extra effects is triggered by the unexpected typographical representation, and,
then, second, the typography itself naturally highlights where the reader should
expect to find those effects. We now consider each stage in turn.

4.2 Expectations and effects

To demonstrate (a) how the contrastive typography interacts with utterance inter-
pretation and (b) what role expectations play in this, we consider the examples in
(22)–(25) more closely. According to relevance theory, all ostensive stimuli carry
with them an assumption of optimal relevance, and, as a result, they trigger the
relevance-theoretic comprehension procedure. Written utterances are ostensive
stimuli and so when processing (22)–(25), the reader will follow the path of least
effort in testing interpretations, and will only stop when the cognitive effects
derived are adequate to justify the processing effort that has been invested. This
is fairly straightforward in Example (22), as the proposition that jenny teaches
violin is assumed to be relevant enough in its own right. However, in (23)–(25)
the typographical representation of the utterance is different, and we claim, unex-
pected. When discussing contrastive prosodic stress Wilson and Wharton (2006,
p. 1567) argue that:
 Chapter 6. Typography, expectations and procedures 181

It follows from the Communicative Principle of Relevance that if two stress pat-
terns differ in the amounts of processing effort required, the costlier pattern should
be used more sparingly, and only in order to create extra, or different, effects.

We suggest that the same applies to typographical patterns, again following directly
from the Communicative Principle of Relevance. If two typographical patterns
differ in the amounts of processing effort required, then the costlier pattern should
only be used to create extra effects. This leads us to a related question which will
be central to our account of typographical variation. Why should the patterns in
(23)–(25) be more costly in terms of processing effort than the pattern in (22)? The
answer to this lies, we suggest, in the role played by expectations.
As Clark (2013, p. 104) explains “[a] number of factors are now known to
affect processing effort. These include recency of use, frequency of use, perceptual
salience, ease of retrieval from memory, linguistic or logical complexity and size of
the context.” Barsalou (2014) also notes that recency, frequency and, in the case of
strings of linguistic material, length, influence ease of access. For length, Barsalou
(2014, pp. 226–227) notes that furniture takes longer to access than chair, for
example. He too reports that higher frequency words are accessed more quickly
than lower frequency ones (e.g., cat vs. feline) and that recently activated linguistic
material is more readily (re)‍accessed than something not recently drawn upon.
Linguistic and pragmatic processing happens online and in real time (see
Breheny, et al., 2013 and Sedivy, et al., 1999 for discussions of reference assign-
ment and incremental processing). Furthermore, according to relevance theory,
any ostensive stimulus triggers the addressee to seek out an optimally relevant
interpretation, and to do this he will make predictions and hypotheses about
what the speaker might be intending to communicate. Crucially, relevance theory
claims that we follow a path of least effort in forming and testing these online
interpretative hypotheses. As recency and frequency of use affect processing ef-
fort by facilitating ease of access, typographical patterns that are frequently and
consistently used will demand less effort from the reader. Thus, a written utterance
which conforms to expected typographical conventions necessarily demands less
processing effort from the reader than one which deviates from those conven-
tions. The examples in (23)–(25) require more processing effort precisely because
they deviate from the reader’s expectations. We do not usually expect an entire
word to be capitalised in an otherwise conventionally presented written utterance.
This deviation from what is expected puts the reader to more effort and triggers a
search for more effects to justify this extra effort.
The role played by confirmation and disconfirmation of expectations in inter-
pretative effects has been discussed and explored elsewhere in relation to music
(Huron, 2006; Steinbeis, et al., 2006; Pearce, et al., 2010; Egermann, et al., 2013),
182 Kate Scott and Rebecca Jackson

language processing generally (Van Berkum, et al., 2005; Levy, 2008) and prosody
specifically (Calhoun, 2009; Scott, 2017; Wharton & Clark, personal communica-
tion). As Huron explains:
Expectation is a constant part of mental life. A cook expects a broth to taste a
certain way. A pedestrian expects traffic to move when the light turns green…
If my text were abruptly to change topics, or if the prose suddenly switched to a
foreign language, you would probably be dismayed. Nor do the changes need to
be dramatic in order to have an effect (Huron, 2006, p. 3)

Integrating this notion of expectation with the relevance-theoretic pragmatic

framework allows us to be slightly more optimistic about the disconfirmation
of expectations than Huron appears to be here. Relevance theory predicts that
rather than simply being dismayed, a hearer or reader assuming optimal relevance
will seek out cognitive effects that justify the extra effort inherent in processing
something unexpected. Calhoun (2009, p. 61) picks up on a similar idea in her
discussion of contrastive prosody:
Pragmatic effects of contrastiveness arise because speakers use probabilistic
expectations about the prosody/information structure interface precisely in order
to get those pragmatic effects. Speakers indicate through the manipulation of
expected prominence that their utterance means more [than] its propositional
content/information structure suggests.

While disconfirmation of a reader’s expectations triggers the search for extra ef-
fects, typographical expectations are, of course, context specific at the levels of
the language, discourse context and text itself. Different languages have different
conventions for how text is represented on the page. In German, the first letter of
all nouns is capitalised, whereas in English this is only the case for proper nouns.
In English the first person singular pronoun I is capitalised wherever it occurs in
a sentence, while the other pronouns are only capitalised at the start of a sentence.
There are also genre specific conventions, such as capitalising the initial letter
of each content word in a book title or using italics for Latin terms, as seen in
Example (2) above. Twyman (1982, p. 21) notes that typographical “conventions
change over the years” and we find examples of usage changing, developing and
standardising over time. As Lass (1999) notes, in the writing of Daniel Defoe,
“capitals appear to be used at random – and not always even after a period” as
illustrated in (26) from the novel Roxana:
(26) I have given you the whole Detail of the Story, to lay it down as a black
Scheme of the Way how Unhappy Women are ruin’d by Great Men 
 (Daniel Defoe, Roxana, p. 64).
 Chapter 6. Typography, expectations and procedures 183

While the distribution of capitals in (26) appears random when compared against
current literary conventions, our approach predicts that it is not in fact so, but
rather that these choices are a result of the general pragmatic considerations we
discuss here interacting with the text at a time when typographical conventions
were unestablished. If it were truly random we would expect function words to
occasionally receive initial capitalization, but this does not seem to be the case. The
words that are represented with the initial letter capitalized are nouns, verbs and
adjectives, and we can assume Defoe, or more likely the typesetter, intended some
sort of salience to be afforded to them. We see a similar pattern and motivation in
(27), which recreates the text and typography from a sign placed in a pedestrian
area of London. The writer has disregarded normal conventions in order to draw
special attention to the words pigeons and rats.
(27) Do not feed the Pigeons as this attracts Rats.

As Crystal points out “once a typographic convention is established, contrasts

could be made” (p.16).
Expectations are also relative to the individual text itself. Consider the example
in (28), seen on both the website for a London college and printed on a banner
outside the college building itself:
(28) THE ACADEMIC SIXTH FORM FOR INDEPENDENT THINKERS.

A style decision has been taken to present the whole utterance in upper case
letters. A consequence of this is that capitalization can no longer to be used to
highlight any one specific part of the utterance, as it would not be unexpected,
text-internally. Instead, the determiner the is presented in upper case italics. This
use of italics is unexpected and encourages the reader to look for additional effects
related to the determiner, parallel to those resulting from contrastive stress in a
comparative spoken version. Walker (2001, p. 71) provides similar examples of
word-processed public notices which are produced entirely in upper case. In these
examples, underlining is used to add emphasis to particular words, as in (29):
(29) ALL CARS MUST BE MOVED OFF THE ESTATE ON BOTH DAYS
 (Walker, 2001, p. 71).

As these examples demonstrate, salience is not an objective quality of one particu-

lar letterform, but is relative to the discourse and textual context. Consider again
the use of italics in (1)–(4). Italics are neither bigger not bolder than the non-italic
font. They stand out because they are otherwise unexpected, and this drives the
reader to seek out effects connected to them.
184 Kate Scott and Rebecca Jackson

4.3 Typography as a natural highlighting device

We have suggested above that, as part of an ostensively communicated written ut-

terance, an unexpected change in letterform such as capitalization or use of italics
contributes to interpretation in two ways. It (a) triggers a search for more effects to
justify the extra effort inherent in processing an unexpected stimulus and (b) high-
lights the locus of relevance, or, in other words, suggests where the extra and dif-
ferent effects can be found which justify the use of a marked typographical pattern.
In Section 4.2 we discussed the triggering role played by expectations in this
process, and in this section, we move on to look at the highlighting function
of contrastive typographic devices in further detail. Previous work within the
relevance-theoretic framework has used the notion of “natural highlighting” to
account for the effects of contrastive prosody (Wilson & Wharton, 2006; Wharton,
2009; Scott, 2017). Given the parallels between prosody and typography discussed
above, we suggest that a similar approach to contrastive graphical features could
prove fruitful. Furthermore, Wharton (2009, p. 142) notes that ostensive gazing,
pointing and contrastive stress are used to draw attention to particular utterance
constituents which are then optimally processed and achieve effects in line with
“the automatic working of the relevance-theoretic comprehension heuristic”. If,
as we suggest, contrastive typography is a natural highlighting device, then it too
produces its effects via interaction with these general pragmatic principles, and no
additional theoretical machinery is required.
A key assumption from the relevance-theoretic framework which underpins
our account is the idea that pragmatic mechanisms operate not just on linguistic
utterances but on any ostensive behaviour. Relevance theory has long acknowl-
edged that ostensive behaviour falls along a continuum from showing to saying
(Sperber & Wilson, 1995; Wharton, 2009), and while linguistic utterances are
examples of saying, a communicator may convey her meaning by ostensively
showing something to an addressee. Developing Gricean insights into speaker
meaning, Wharton (2009) notes that when we engage in ostensive-inferential
communication, there are two layers of information involved. There is the first
layer of information that the communicator intends to communicate, and there
is a second layer of evidence on the basis of which the first layer of information
can be worked out. This working out is constrained by the search for optimal rel-
evance, as outlined above. The evidence provided for the recovery of the first layer
of information can take two forms. A communicator can “say” what they intend to
communicate, or they can “show” it.7

7. Or some blend of the two. As Wharton (2009) and Sasamato and Jackson (2016) have argued,
interjections and onomatopoeia which are moving towards lexicalisation retain characteristics
 Chapter 6. Typography, expectations and procedures 185

Saying obtains when the evidence provided for the recovery of the first layer
involves linguistic decoding and inference, and when the evidence provided for
recovery of the first layer is relatively more “indirect”. By “indirect,” Wharton
means that there are relatively more inferential steps involved in moving from the
second layer of information to recovering the speaker’s meaning. The paradigm
case of saying involves utterances of linguistic material. However, it is sometimes
not optimally relevant for communicators to rely on linguistic decoding and infer-
ence for communicating their speaker meaning. There are things that we wish to
communicate that do not lend themselves well to linguistic encoding, such as, for
example, impressions and other non-propositional effects. Therefore, communi-
cators also have the option of producing an act of showing as evidence for the
recovery of the first layer. Showing obtains when evidence is presented which does
not involve linguistic decoding, and conclusions are drawn purely inferentially.
The inferential steps involved in processing acts of showing are generally fewer
than saying because there are no steps required to flesh out a proposition schema
on the basis of linguistic decoding. As such, showing is generally described as
more direct. Consider the exchanges in (30) and (31):
(30) Kelly: What do we have for dinner?
Rosie: Beans and bread.
(31) Kelly: What do we have for dinner?
Rosie: [demonstratively opens cupboard to reveal a can of baked beans and
a loaf of bread]

Rosie’s reply in (30) is a case of saying. Her fragmentary utterance must be

completed, understood as a response, and pragmatically enriched to yield an
explicature. On the basis of this explicature, Kelly may recover an implicature that
the meal they will eat is beans on toast. In (31), however, no utterance must be
completed. There is no need for pragmatic enrichment of explicit content, as there
is no decoding. Constrained by expectations of relevance and the relevance-the-
oretic comprehension procedure, Kelly must simply understand Rosie’s showing
as a response, and draw optimally relevant conclusions from the intentional act of
showing the ingredients. Rosie’s showing is relatively more direct than her saying.
When we examine phenomena which have been analysed as natural highlight-
ing devices in relevance theory (e.g., emotive prosodic features, ostensive pointing,
facial expressions, impact captions (telops), choice of writing system, etc.), they can
all be seen as providing evidence for the first layer of speaker’s meaning to some
extent. In other words, highlighting can be understood as a degree of showing.

of both showing and saying. However, such cases are not the focus of this chapter.
186 Kate Scott and Rebecca Jackson

The message that a communicator conveys may also be more or less determi-
nate (Sperber & Wilson, 2015). Compare (32) and (33):
(32) The flight leaves at 11:25.
(33) Juliet is the sun.

Whereas an utterance such as (32) communicates a determinate proposition about

when the flight leaves, a metaphor such as (33), communicates something much
less determinate. Sperber and Wilson (2015) argue that acts of showing can be
both determine and indeterminate. We see this reflected in the variety of functions
to which italicisation can be put. In the example in (1), it affects how reference is
resolved, making a difference to the truth conditions of the proposition expressed.
In (4), it is used to convey the narrator’s emotions and attitudes, which are less
determinate. The use in Example (3) lies somewhere in the middle. A contrast is
clearly implied, but the reader has to work out why this contrast is relevant.
Finally, we want to consider why these variations in typography have a
highlighting effect, and we will once again consider the role that expectations and
conventions play in this. However, this also brings us to the role of figuration and
metaphor in prosody and typography. According to Gussenhoven (2002, p.1),
those aspects of prosody which are not language-specific derive their meaning
via “metaphors of biological conditions”. There are natural links, it is claimed,
between a speaker’s physicality and the sounds that that speaker produces. Males,
for example, generally have longer vocal tracts than females, and so produce ut-
terances at a lower fundamental frequency. Similarly, a speaker who puts more
effort into an articulation will produce a sound that is likely to be more precise and
to have “more numerous pitch movements” (Guseenhoven 2002, p.1) than one
produced with less care and attention. Thus, the prosodic qualities of utterances
become metaphorically associated with characteristics, attitudes and affect. Lower
tones are associated with stereotypically masculine traits such as dominance
or confidence, and precisely articulated utterances with greater than normal
pitch variation signal that the speaker is particularly engaged or invested in the
discourse. While these associations stem from a natural relationship between
the speaker’s physicality and the quality of the sounds produced, a speaker may
mimic or exploit these associations metaphorically to evoke the same effects. For
example, while a speaker cannot physically change the size of her larynx, she can
raise or lower her register to make herself sound more or less feminine or mascu-
line. We see a similar metaphorical relationship inherent in the link between visual
salience and conceptual salience in written text. It is, of course, not coincidental
that the natural highlighting devices used in typography tend to make the text big-
ger, darker or more salient. It is not by pure change that, as Crystal (2010, p. 188)
 Chapter 6. Typography, expectations and procedures 187

notes, capitalization is used to convey “loudness and special significance”. There is

a natural link between size and likely relevance. Animals are wary of other animals
with larger body sizes, and young children will shy away from particular large,
dark shapes. In empirical work examining the cognitive consequences of boldface
for word recognition, Macaya and Perea (2004) suggest that low frequency words
are recognised faster as legitimate words if they are rendered in bold. Pelli et al.
(2006) showed that a Bookface bold font outperformed Bookface normal font
for readability, while Perea and Rosa (2002) found that lower frequency words
are processed faster when presented in uppercase than when not. Adams and
Edworthy’s (1995) work on signal changes in warning labels found a relationship,
among others, between font size and perceived warning urgency.
However, this cannot be the complete story, as once again, language, culture
and text specific conventions and expectations play an important role. For example,
we do not pay extra attention to initial capitalisation where it is expected at the
start of a sentence of English. Neither do we attribute any particular highlighting
intention to a newspaper headline that is larger and bolder than the main body
text below it, other than its identification as a headline. Central to our approach,
is the claim that while upper case letters are larger than their lower case cousins,
it is only when they deviate from conventions that the search for extra meaning
is triggered. Words in italics are neither larger nor darker than their counterparts
in standard font, and yet italics are frequently used as a highlighting device in
literary fiction. This is, we suggest, because the key lies in the disconfirmation of
expectations and the corresponding search for extra effects. When something is
unexpected it draws the reader’s attention and, as Sperber and Wilson (1995: p. 49)
explain “[t]‍here is…no point in drawing someone’s attention to a phenomenon
unless it will seem relevant enough to him to be worth his attention”.

5. Summary, conclusion and next steps

While the very act of producing a written utterance is an ostensive act which trig-
gers an initial search for relevance and sets the reader off down the path of least
effort, an unexpected typographical pattern will distract the reader from this path.
In order to justify the distraction they must assume a new path on which the extra
effort is justified by some cognitive effects that would not otherwise have been
derived. Typographical variation such as the italics in (1)–(4) or the capitalisation
in (23)–(25) not only triggers the search for extra effects, but also guides the reader
as to where they might expect to find those effects. The change in typography
draws the reader’s attention, and so, following the path of least effort, he will be
likely to test interpretative hypotheses in which the extra effects are connected to
188 Kate Scott and Rebecca Jackson

the highlighted word. The suprasegmental quality of typography means that the
triggering of the search for more effects and the highlighting of where those effects
might be found can be achieved without the need to include additional words or to
change the syntactic structure of the sentence.8
Thus contrastive typography, like contrastive prosody, shares both a triggering
and a guiding function with various communicative devices and expressions that
have been analysed as procedural. However, we do not want to say that typographi-
cal features encode a procedure directly. This is because the effect that the features
have on interpretation is context specific and both the triggering process and the
effects that are generated are relative to the discourse context of the utterance,
widely construed. It is not the presence or absence of a feature that guides the
interpretation but the confirmation or disconfirmation of expectation about how
an utterance will be produced.
We believe that our cognitive and procedural approach to typography can
be developed in various ways in future work. Extrinsic typographical features
including page layout, spacing and use of colour perhaps rely on natural and cul-
tural codes and interact with textual and literary conventions and genres. Again,
disconfirmation of expectations and natural highlighting may prove insightful
notions in analysing the resulting effects.
Although writing long before the development of relevance theory and of
cognitive pragmatics more generally, Warde (1955) seems to be taking a broadly
cognitive approach to communication in her discussion of typography when she
notes that “[t]‍alking, broadcasting, writing and printing are all quite literally forms
of thought transference”. According to relevance theory, an utterance is no more
than a clue to the speaker’s intended meaning, and thus the process of interpreting
an utterance is the process of working out what the speaker intended to com-
municate. The substance of a written text, that is, the words and structure that
make up the sentences, stays the same regardless of how it is rendered on the
page. However, as we have shown, the typographical presentation can affect how
a written utterance is interpreted in various different ways. A writer can use the
typographical resources at her disposal to help guide the reader towards her in-
tended meaning. In this sense we agree with Warde when she says that typography
is “a means of doing something,” and so, in relevance-theoretic terms, we suggest
that it functions procedurally.

8. These are, of course, other options available to the speaker. In (23)- (25), for example, she
could have used a cleft sentence to draw attention to a particular part of the utterance: (i) It was
Jenny who played the violin. (ii) It was the violin that Jenny played, etc.
 Chapter 6. Typography, expectations and procedures 189

References

Adams, A. & Edworthy, J. (1995). Quantifying and predicting the effects of basic text display
parameters on the perceived urgency of warning labels: Trade offs involving font size,
border width and colour. Ergonomics, 38(11), 2221–2237. ​
https://‍doi.org/10.1080/00140139508925264
Barsalou, L. W. (2014). Cogntive psychology: An overview for cognitive scientists. New York:
Psychology Press.
Berger, S., Marquard, C. & Neibuhr, O. (2016). How different typefaces affect speech prosody.
Proc. 8th International Conference of Speech Prosody, 513–517.
Blakemore, D. (1987). Semantic constraints on relevance. Oxford: Blackwell.
Blakemore, D. (2002). Relevance and linguistic meaning: The semantics and pragmatics of dis-
course markers. Cambridge: Cambridge University Press. ​
https://‍doi.org/10.1017/CBO9780511486456
Bolinger, D. (1983). The inherent iconism of intonation. In J. Haiman (Ed.), Iconicity in syntax
(pp. 97–109). Amsterdam: John Benjamins.
boyd, d. (2010). Social network sites as networked publics: Affordances, dynamics and implica-
tions. In Z. Papacharissi (Ed.), Networked self: identity, community and culture on social
networking sites (pp. 39–58). Abingdon: Routledge.
Breheny, R., Ferguson, H. J. & Katsos, N. (2013). Taking the epistemic step: Toward a model of
on-line access to conversational implicatures. Cognition, 126, 423–440. ​
https://‍doi.org/10.1016/j.cognition.2012.11.012
Bryant, G. A. & Fox Tree, J. E. (2005). Is there an ironic tone of voice?. Language and Speech,
48(3), 257–277.  ​https://‍doi.org/10.1177/00238309050480030101
Butterick, M. (2010). Butterick’s Practical Typography. [Online] Available at: http://‍practical­
typography.com
Butterick, M. (2016). Drowning the Crystal Goblet. [Online] Available at: http://‍practical­
typography.com/drowning-the-crystal-goblet.html
Calhoun, S. (2009). What makes a word contrastive? Prosodic, semantic and pragmatic per-
spectives. In D. Barth-Weingarten, N. Dehé & A. Wichmann (Eds.), Where prosody meets
pragmatics: Research at the interface (pp. 53–78). Bingley: Emerald. ​
https://‍doi.org/10.1163/9789004253223_004
Carston, R. (2002). Thoughts and utterances: The pragmatics of explicit communication. Oxford:
Blackwell.  ​https://‍doi.org/10.1002/9780470754603
Carston, R. (2016). The heterogeneity of procedural meaning. Lingua, 175–176, 154–166. ​
https://‍doi.org/10.1016/j.lingua.2015.12.010
Cirillo, L. (2019). The pragmatics of air quotes in English academic presentations. Journal of
Pragmatics, 142, 1–15.
Clark, B. (2012). The relevance of tones: prosodic meanings in utterance interpretation and in
relevance theory. The Linguistic Review, 29(4), 643–661. ​
https://‍doi.org/10.1515/tlr-2012-0024
Clark, B. (2013). Procedures and prosody: weak encoding and weak communication. In: F.
Liedtke & C. Schulze (Eds.), Beyond words: Content, context and inference (pp. 151–181).
Berlin: De Gruyter Mouton.  ​https://‍doi.org/10.1515/9781614512776.151
Clark, B. (2013). Relevance theory. Cambridge: Cambridge University Press. ​
https://‍doi.org/10.1017/CBO9781139034104
190 Kate Scott and Rebecca Jackson

Crystal, D. (1998). Toward a typographical linguistics. Type, 2(1), 7–23.

Crystal, D. (2010). The Cambridge encyclopedia of language. 3rd ed. Cambridge: Cambridge
University Press.
Daniels, P. (2016). Drive all night to watch the blues. [Online] Available at: http://‍playbackmedia.
co.uk/the-chels-the-chelsea-podcast-archive/
Defoe, D. (1964 [1724]). Roxana. Oxford: Oxford University Press.
Drury, J. E., Baum, S. R., Valeriote, H. & Steinhauer, K. (2016). Punctuation and implicit prosody
in silent reading: an ERP study investigating English garden-path sentences. Frontiers in
Psychology, Volume 7, p. 1375.  ​https://‍doi.org/10.3389/fpsyg.2016.01375
Egermann, H., Pearce, M. T., Wiggins, G. A. & McAdams, S. (2013). Probabilistic models of
expectation violation predict psychophysiological emotional responses to live concert
music. Cognitive, Affective and Behavioural Neuroscience, 13, 533–553. ​
https://‍doi.org/10.3758/s13415-013-0161-y
Eliot, G. (1861). Silas Marner. London: Penguin.
Ellis, B. E. (1991). American Psycho. London: Picador.
Escandell-Vidal, V. (1998). Intonation and procedural meaning: the case of Spanish interroga-
tives. In V. Rouchota & A. Jucker (Eds.), Current issues in relevance theory (pp. 169–203).
Amsterdam: John Benjamins.  ​https://‍doi.org/10.1075/pbns.58.09esc
Filik, R., Țurcan, A., Thompson, D., Harvey, N., Davies, H. & Turner, A. (2016). Sarcasm and
emoticons: Comprehension and emotional impact, The Quarterly Journal of Experimental
Psychology, 69(11), 2130–2146.  ​https://‍doi.org/10.1080/17470218.2015.1106566
Fitzgerald, F. S. (1926). The Great Gatsby. London: Penguin.
Fodor, J. D. (2002). Psycholinguistics cannot escape prosody. SP-2002, 83–90.
Fretheim, T. (2002). Intonation as a constrainst on inferential processing. s.l., s.n., pp. 59–64.
Gross, J. et al. (2014). Evidence for prosody in silent reading. Reading Research Quarterly, 49(2),
189–208.  ​https://‍doi.org/10.1002/rrq.67
Gussenhoven, C. (2002). Intonation and interpretation: phonetics and phonology. In B. Bel
& I. Marlien (Eds.), Proceedings of the 1st International Conference on Speech Prosody
(pp.47–57). Aix-en-Provence, France, 11–13 April.
Hedley, P. (2005). Pronouns, procedures and relevance theory. Durham Working Papers in
Linguistics, 11, 41–55.
House, J. (2006). Constructing a context with intonation. Journal of Pragmatics, 38(10),
1542–1558.  ​https://‍doi.org/10.1016/j.pragma.2005.07.005
Huron, D. (2006). Sweet anticipation: Music and the psychology of expectation. Cambridge, MA.:
MIT Press.
Imai, K. (1998). Intonation and relevance. In R. Carston & S. Uchida (Eds.), Relevance theory:
Applications and implications (pp. 69–86). Amsterdam: John Benjamins. ​
https://‍doi.org/10.1075/pbns.37.06ima
Jackson, R. (2016). The pragmatics of repetition, emphasis and intensification. University of
Salford: PhD Thesis.
Lampert, M. (2013). Say, be like, quote (unquote), and the air-quotes: interactive quotatives and
their multimodal implications: The ‘new’ quotatives remind us of the vocal, verbal, and
gestural dimensions of speech. English Today, 29(4), 45–56. ​
https://‍doi.org/10.1017/S026607841300045X
Lass, R. (1999). The Cambridge history of the English language volume III 1478–1776. Cambridge:
Cambridge University Press.
 Chapter 6. Typography, expectations and procedures 191

Levy, R. (2008). Expectation-based syntactic comprehension. Cognition, 16, 1126–1177. ​

https://‍doi.org/10.1016/j.cognition.2007.05.006
Litt, E. (2012). Knock, knock. Who’s there? The imagined audience. Journal of Broadcasting &
Electronic Media, 56(3), 330–345.  ​https://‍doi.org/10.1080/08838151.2012.705195
Macaya, M. & Perea, M. (2004). Does bold emphasis facilitate the process of visual word recog-
nition. Spanish Journal of Psychology, 17(e2), 1–5.
Marwick, A. E., Boyd, D. (2010). I tweet honestly, I tweet passionately: Twitter users, context
collapse, and the imagined audience. New Media and Society, 13, 96–113.
Pearce, M. T., Müllensiefen, D. & Wiggins, G. (2010). The role of expectation and probabilistic
learning in auditory boundary perception. Perception, 39, 1367–1391. ​
https://‍doi.org/10.1068/p6507
Pekkola, J. et al. (2005). Primary auditory cortex activation by visual speech: an fMRI study at
3T. NeuroReport, 16, 125–128.  ​https://‍doi.org/10.1097/00001756-200502080-00010
Pelli, D. G., Burns, C. W., Farell, B. & Moore-Page, D. C. (2006). Feature detection and letter
identification. Vision Research, 46(28), 4646–4674. ​
https://‍doi.org/10.1016/j.visres.2006.04.023
Perea, M. & Rosa, E. (2002). Does “whole-word shape” play a role in visual word recognition.
Perception and Psychphysics, 64(5), 785–794.  ​https://‍doi.org/10.3758/BF03194745
Perrone-Bertolotti, M. et al. (2012). How silent is silent reading? Intracerebral evidence for top-
down activation of temporal voice areas during reading. Journal of Neuroscience, 32(49),
17554–17562.  ​https://‍doi.org/10.1523/JNEUROSCI.2982-12.2012
Sadokierski, Z. (2011). Disturbing the text: typography devices in literary fiction. Book 2.0, 1(2),
101–135.  ​https://‍doi.org/10.1386/btwo.1.2.101_1
Sasamoto, R. (2014). Impact caption as a highlighting device: attempts at viewer manipulation
on TV. Discourse, Context and Media, 6, 1–10.  ​https://‍doi.org/10.1016/j.dcm.2014.03.003
Sasamoto, R. & Jackson, R. (2016). Onomatopoeia – showing-word or saying-word? Relevance
theory, lexis, and the communication of impressions. Lingua, 175–176, 36–53.
Sasamoto, R. & O’Hagan, M. (2020). Relevance, style and multimodality: typographical features
as stylistic devices. In A. Piskorska (Ed.), Relevance theory, figuration and continuity in
pragmatics (pp. 193–226). Amsterdam: John Benjamins.
Scott, K., (2016). Pronouns and procedures: Reference and beyond. Lingua, 175–176, 69–82.
Scott, K. (2017). Prosody, procedures and pragmatics. In I. Depraetere & R. Salkie (Eds.), Se-
mantics and pragmatics: Drawing a line (pp. 323–341). Berlin: Springer. ​
https://‍doi.org/10.1007/978-3-319-32247-6_18
Scott, K. (2018). Hashtags work everywhere: The pragmatic functions of spoken hashtags.
Discourse, Context and Media, 22, 57–64.
Sebeok, T. (1972). Perspectives in zoosemiotics. The Hague: Mouton.
Sedivy, J. C., Tanenhaus, M. K., Chambers, C. G. & Carlson, G. N. (1999). Achieving incremen-
tal semantic interpretation through contextual representation. Cognition, 71, 109–147. ​
https://‍doi.org/10.1016/S0010-0277(99)‍00025-6
Skovholt, K. Grønning, A. & Kankaanranta, A. (2014). The communicative function of emoti-
cons in workplace emails::-). Journal of Computer-Mediated Communication, 19, 780–797. ​
https://‍doi.org/10.1111/jcc4.12063
Sperber, D. & Wilson, D. (1995). Relevance: Communciation and Cognition, 2nd ed. Oxford:
Blackwell.
Sperber, D. & Wilson, D. (2015). Beyond speaker’s meaning. Croatian Journal of Philosophy,
15(44), 117–149.
192 Kate Scott and Rebecca Jackson

Steinbeis, N., Koelsch, S. & Sloboda, J. A. (2006). The role of harmonic expectancy violations in
musical emotions: evidence from subjective, physiological and neural responses. Journal of
Cognitive Neuroscience, 188, 1380–1393.  ​https://‍doi.org/10.1162/jocn.2006.18.8.1380
Tagg, C. & Seargeant, P. (2014). Audience design and language choice in the construction and
maintenance of translocal communities on social network sites. In P. Seargeant, & C. Tagg,
(Eds.), The language of social media: Identity and community on the Internet (pp. 161–185).
Palgrave MacMillan, Basingstoke.  ​https://‍doi.org/10.1057/9781137029317_8
Thompson, D. & Filik, R. (2016). Sarcasm in written communication: Emoticons are efficient
markers of intention. Journal of Computer-Mediated Communication, 21, 105–120. ​
https://‍doi.org/10.1111/jcc4.12156
Twyman, M. (1982). The graphic presentation of language. Information Design Journal, 1(2),
2–22.  ​https://‍doi.org/10.1075/idj.3.1.01twy
Van Berkum, J. et al. (2005). Anticipating upcoming words in discourse: evidence from ERPs
and reading times. Journal of Experimental Psychology: Learning, Memory and Cognition,
31(3), 443–467.
van Leeuwen, T. (2006). Towards a semiotics of typography. Information Design Journal + Docu-
ment Design, 14(2), 139–155.  ​https://‍doi.org/10.1075/idj.14.2.06lee
Walker, S. (2001). Typography and language in everyday life: Prescriptions and practices. Harlow:
Pearson.
Warde, B. (1955). The crystal goblet. London: The Sylvan Press.
Wells, J. C. (2006). English intonation: An introduction. Cambridge: Cambridge University Press.
Wesch, M. (2009). Youtube and you: experiences of self-awareness in the context collapse of the
recording webcam. Explorations in Media Ecology 8(2), 19–34.
Wharton, T. (2009). Pragmatics and non-verbal communication. Cambridge: Cambridge Univer-
sity Press.  ​https://‍doi.org/10.1017/CBO9780511635649
Wharton, T. (2012). Prosody and meaning: theory and practice. In J. Romero-Trillo (Ed). Prag-
matics and prosody in English language teaching (pp. 97–116). Dordrecht: Springer. ​
https://‍doi.org/10.1007/978-94-007-3883-6_7
Wilson, D. (2011). The conceptual-procedural distinction: past, present and future. In V. Escan-
dell-Vidal, M. Leonetti & A. Ahern (Eds.), Procedural meaning: Problems and perspectives
(pp. 3–31). Bingley: Emerald.  ​https://‍doi.org/10.1108/S1472-7870(2011)‍0000025005
Wilson, D. (2016). Reassessing the conceptual-procedural distinction. Lingua 175–176, 5–19. ​
https://‍doi.org/10.1016/j.lingua.2015.12.005
Wilson, D. & Sperber, D. (1993). Linguistic form and relevance. Lingua, 90, 1–25. ​
https://‍doi.org/10.1016/0024-3841(93)‍90058-5
Wilson, D. & Sperber, D. (2012). Meaning and relevance. Cambridge: Cambridge University
Press.  ​https://‍doi.org/10.1017/CBO9781139028370
Wilson, D. & Wharton, T. (2006). Relevance and prosody. Journal of Pragmatics, 38(10),
1559–1579.  ​https://‍doi.org/10.1016/j.pragma.2005.04.012
Yao, B., Belin, P. & Scheepers, C. (2011). Silent reading of direct versus indirect speech activates
voice-selective areas in the auditory cortex. Journal of Cognitive Neuroscience, 23(10),
3146–3152.  ​https://‍doi.org/10.1162/jocn_a_00022
 Chapter 7

Relevance, style and multimodality

Typographical features as stylistic devices

Ryoko Sasamoto and Minako O’Hagan

Dublin City University, Ireland / The University of Auckland, New Zealand

Brightly coloured textual inserts, which often occupy a sizable part of the TV
screen, have become a key feature in Japanese TV. This paper unpacks the
contribution of such multimodal stimuli to inference, and the consequences this
has for the interpretation process. Using data derived from a mixed-methods
approach (i.e. eye-tracking and a multimodal content analysis), we evaluate
the relationship between cognitive processing and communicative stimuli. We
demonstrate how typographical features (colours and fonts) are used as high-
lighting stylistic devices by TV producers to manipulate the viewer comprehen-
sion process by guiding the audience to an intended interpretation. The results
suggest how editorial choices regarding typographical features to trigger certain
effects might be subsumed under the current view of style in relevance theory.

Keywords: typography, multimodality, Japanese TV, telop, style, viewer

behaviour, pragmatic effects, eyetracking, propositional attitudes, affect

1. Introduction

With the increasing use of digital devices, communication is now taking place
more and more in a digital space. This shift in communication modes comes with
an interesting use of written communication via text on screen. Digital devices,
compared with traditional writing devices such as pen and paper, are the driving
force for an enhanced form of written communication and people are engaging
in more visually appealing written communication, exploiting different fonts,1
colours, textual effects such as animations, and, in some cases, emoticons. Indeed,

1. Strictly speaking, font and typeface should be distinguished. However, for the purpose of
this paper and for simplicity, we use the term font to refer to both the design of letters and a
particular presentation (i.e. size and weight) of the letters.

https://‍doi.org/10.1075/ftl.8.07sas
© 2020 John Benjamins Publishing Company
194 Ryoko Sasamoto and Minako O’Hagan

as van Leeuwen (2006) points out, typographical features are playing a more
prominent role in modern day communication than it has previously been the
case. This paper seeks to address the role of typographical features in the interpre-
tation of multimodal contents, namely, how features such as font and colour affect
the viewer interpretation process. We use Japanese TV programmes as our source
of stimuli, as such programmes often consist of a complex layer of multimodal
representations. In particular, we focus on telops (pop-up captions) which contain
the outlandish use of typographical features. Our main goal is to demonstrate how
a relevance-theoretic framework can be usefully applied in shedding light on the
role played by such features used as stylistic devices. This paper also briefly refers
to our preliminary study using eye-tracker to gauge the impact of these features
on the recipients based on their gaze pattern. The paper is structured with the
current Section 1 providing a background on telops used on Japanese TV, followed
by a discussion in Section 2 on stylistic features from a perspective of multimodal
analysis. Section 3 covers the treatment of typographical features as pragmatic ef-
fects within the framework of relevance theory. Next, a case study using a Japanese
TV programme is presented to demonstrate the application of relevance theory
and multimodal analysis of telops in Section 4. We will see how typographical
features could give rise to enhanced pragmatic effects such as communication
of weak implicatures and propositional attitudes. Section 5 briefly introduces a
preliminary eye-tracking study in an attempt to present some empirical supposed
based on viewer data. Section 6 provides a summary of the findings in relation to
our research questions.

1.1 Background: Telops on Japanese TV

In the last decade or so, the excessive use of written props in the form of captions
has become a key feature of certain TV programmes in Japan as well as in other
parts of Asia. Compared to other types of captions, such textual inserts (or Telops
as known in Japan) are immediately conspicuous and dominate the TV screen
(see Figure 1).2
Figure 1 shows a typical scene from a Japanese entertainment programme,
belonging to a genre called a variety show. The scene contains three clusters of cap-
tions: the caption at the top left shows the title of this particular episode, while the
caption at the top right shows the topic of the discussion taking place in the scene.
The caption at the bottom, which is an example of telops, is a verbatim rendering

2. Throughout this paper, parts of figures have been pixelated for copyright and portrait rights.
 Chapter 7. Relevance, style and multimodality 195

Figure 1. Multi-coloured captions used in a typical Japanese programme Honmadekka!?

(2013)

of an utterance produced by one of the studio panelists.3 Telops are usually added
during the post-production editing process and presented in multiple colours. In
this scene, captions at the top of the screen are written in red, white, and black,
while a telop appearing at the bottom of the screen is written in red, green, yellow,
and blue with a white ‘glow’ effect.
Although telops are often (but not always) verbatim representations of ut-
terances produced by the speakers in the programme, they are essentially not
intended as an aid for the hearing-impaired, unlike ‘closed captions’ designed
specifically for accessibility.4 Colours and fonts used in telops are therefore not
in accordance with norms established for such subtitles prioritising legibility and
often contain more than one colour within one caption. These captions may not
even be faithful representations of the spoken elements of the programme. The
added stylistic features are expected to give rise to enhanced pragmatic effects.
This raises the questions of what such effects are, and how they arise.
A few attempts have been made to shed light on the use of telops, and scholars
such as Shiota (2003) and Kimura et al. (2000) present taxonomies of types and
uses of telops. While there is no agreed taxonomy for telops, it is often claimed that
telops are either verbatim representations of utterances, or an interpretation or
summary of discourse, and that they are used to add extra effects as well as facili-
tating viewer comprehension. Another common acknowledgement is the presence

3. This does not mean telop are always verbatim representations of utterances produced in a
programme. They can be a verbatim production of an utterance, a summary, or a TV producer’s
interpretation of a certain utterance/thought.

4. These are intralingual subtitles known as subtitles for the deaf and the hard-of-hearing
(SDH) which the viewers can turn on and off, hence ‘closed’ as opposed to open captions which
the viewers cannot turn off.
196 Ryoko Sasamoto and Minako O’Hagan

of mediators who come between the speaker and the viewers (e.g. Sasamoto, 2014;
O’Hagan, 2010; Shiota, 2003). O’Hagan (2010) and Sasamoto (2014) highlight that
telops are inserted in order to guide the viewers’ cognitive interpretation processes
to suit the programme directors’ needs for enhanced viewing experiences, rather
than merely being a comprehension aid for viewers.
Departing from descriptive typologies of telops, O’Hagan (2010) presents
a case study of a popular game show in Japan and concludes that telops have a
humour-framing function which a director can exploit in order to emphasise
a humorous effect. This can be done, for example, by “dramatizing the trivial”
(O’Hagan, 2010, p. 85). Drawing upon this study, Sasamoto (2014) has investi-
gated the relationship between telop framing function (or “highlighting function”)
and the viewer interpretation process. Sasamoto argues that telop are a highlight-
ing device that a programme director can use to draw the viewers’ attention to
chosen elements for a specified purpose such as to induce laughter. In so doing,
programme directors can ensure that there is a common platform that enables
TV directors to guide viewers in processing programmes in ways that suit their –
not the viewers’ – needs. The relevance-theoretic notions of the Cognitive and
Communicative principles of relevance (Sperber & Wilson, 1995) also justify this
analysis, as these principles predict that the use of telop contributes to relevance
by raising cognitive and affective mutuality by way of shared assumptions and
affective responses. That is, ostensively-used telops demand the viewers’ attention.
This is empirically shown by an eyetracking study which demonstrated that the
viewers cannot help but pay attention to telops (Sasamoto & Doherty, 2013).5 As
a result, TV programme directors can assume that viewers are extremely likely to
process what the directors want them to. This ensures viewers access assumptions
related to the highlighted (i.e. captioned) elements. In other words, the purpose of
using telops is to manipulate the interpretation process by urging viewers to search
for relevance in a way that suits the TV producer.
Let us now see in further detail how telops are used – see Figure 2.
Figure 2 is a scene taken from a special episode of a popular TV series, the Himitsu
no Kenmin Show. The theme of this TV show is regional culture, and most often
concerns food. In this particular episode, they were reporting on the comedy-
loving, humorous nature of people from Osaka. The TV crew asked members
of the public to mimic the action of playing golf on the street but gave them a
selection of objects which were not golf clubs. Figure 2 shows a passer-by who was
asked to participate with the TV crew saying to him: “Can you swing this putter?”

5. Sasamoto and Doherty (2013) conducted an eye-tracking experiment with 16 native speakers
of Japanese and demonstrate that viewers do pay attention to captions regardless of information
communicated by such captions’ content
 Chapter 7. Relevance, style and multimodality 197

Theme of the episode (above)

Programme title (below)

[Asked to swing a golfclub]

“Using this putter…I’m
holding butter!”

Transliteration of ‘putter’ in Transliteration of ‘butter’ in

katakana katakana

Figure 2. A typical telop use in an entertainment programme (Kenmin Show, Shinshun

special, 2013)

The only problem is that none of the objects offered to him is a putter. Just before
the moment in the scene, which features the punchline, the man is given a block
of butter, which the crew had suggested he should swing. He pretends to swing
the butter while saying “using this putter…” and stops halfway through his swing,
declaring “I am holding butter!” The verbatim representation of this utterance is
given in the telop at the bottom of the screen. This telop highlights this particular
utterance, by focusing on the word play contrasting between the word “putter” in
white lettering and the word “butter” in butter-like golden-yellow.6 By highlight-
ing this particular element with the telop, the TV director ensures that the viewers’
attention is drawn to the same points as s/‍he wants to stress, triggering an affective
response by the viewer to the obviously humorous situation. This increases the
likelihood that the at-home viewers will react in the same way as the studio guests,
who are also watching this clip and finding it funny. In other words, in some cases,
telops are associated with affective communication (c.f. Blakemore, 2008). We will
return to this point in Section 3.

6. Note that Japanese uses transliterations of the English words butter (バター, batā)‍and putter
(パター, patā, a type of a golf club) as loan words, which leads to a pun in Japanese in the same
way it does in English.
198 Ryoko Sasamoto and Minako O’Hagan

2. Typographical features, stylistics and multimodal analysis

2.1 Suitability of typographical features

The nature of TV medium means that telops are used simultaneously with other
stimuli delivered to the audience via different communicative channels. Every
communicative stimulus, such as an utterance, sound effect and even the choice of
font, seems to interact with the others, forming an overall multimodal content. On
the one hand, a telop can be treated as a single unit (mode) while it, on the other
hand, is itself multimodal, if counting typographical features such as colours and
fonts that are embedded in a caption just like a matryoshka doll.
As shown in Figures 1 and 2, telops are often displayed with various typo-
graphical features, availing of the use of different fonts, colours and backgrounds.
It is generally acknowledged that such typographical choices influence the way a
message is interpreted by a recipient, as metaphorical associations in the recipi-
ent’s mind play a significant role in visual communication (van Leeuven, 2005).
Indeed, Meier et al. (2007, 2008) explored links between font and affect, and found
that people often associate a positive attitude with brighter colour (Meier et al.,
2007), and a positive attitude with a bigger font (Meier et al., 2008). Lakens et al.
(2012) also established a link between negative words and the colour black. Such
links, or embodied affect-space links (Bornstein & Becker-Matero, 2012), are based
on the assumption that “there are inborn, pre-existing connections between affect
and bodily experience” (p. 95). This would suggest that humans are capable of
exploiting such metaphorical associations to create similar effects (such as the
brighter colour evoking a positive attitude) by presenting a particular visual cue.
Indeed, many scholars report the metaphorical association between font
and affect and suitable uses of various fonts in specific contexts (c.f. Doyle &
Bottomley, 2009; Henderson et al., 2004). For example, the use of the playful font
comic sans in a legal letter is likely to look incongruous with the formal nature
of the intended communication. At the same time the use of such a font is likely
to be perceived suitable for a child’s birthday party invitation. On television, a
producer typically takes into account the nature of a message and matches it to
a selection of fonts. Indeed, there seems to have emerged a convention in which
certain font types are associated with certain types of text and communication.
The typographical selection of a telop could therefore reveal a link to the intended
affective responses which certain captions are expected to trigger in the recipients
(as indicated above).
Issues with font selection have been discussed in various disciplines, includ-
ing Audiovisual Translation (AVT), Web studies, and aesthetics. In AVT, it is
generally agreed that sans-serif fonts such as Arial and Helvetica are preferred to
 Chapter 7. Relevance, style and multimodality 199

serifed fonts for subtitles because of their better legibility (Díaz Cintas & Remael,
2007, p. 84). On the other hand, in a more general font design context, serifed
typefaces (serif) are preferred for longer sentences and, hence, for the main body
of a text, while sans-serif is preferred for headings (c.f. Takahashi & Katayama,
2012). Takahashi and Katayama (2012) explain that serifed typeface contributes
to readability while sans-serif typeface promotes visibility. Indeed, as we will see
in Section 4, serifed fonts (MS Mincho being a typical example of serifed font for
Japanese) are used for a longer telop employed to add experts’ explanations, while
san-serif (MS Gothic being a typical example of san-serif for Japanese) is used for
a telop added to enhance humour, such as those in Figure 1 and Figure 2.7

2.2 Social semiotic approach to typography: Multimodal analysis of fonts

In the previous section, we saw how studies in AVT and web-studies focused
on the suitability of fonts in various contexts. In contrast, scholars such as van
Leeuwen (2005, 2006) describe typographical features as semiotic resources for
meaning-making from a multimodal perspective. Scholars with a special interest
in multimodality and multimodal analysis (e.g.van Leeuwen, 2005, 2006; Stökl,
2005; Norgaard, 2009) see typography as a mode, which has its own grammar
and lexis, and often take a descriptive approach to gain insight into the anatomy
of typography. Van Leeuwen (2006, p. 151) provides the system of a network of
distinctive features such as weight, expansion, and curvature of letters which have
the potential to be developed into “grammar.” Similarly, Stökl (2005) presents his
grammar of typography, again, based on the distinctive features of each letter.
The main aim of these multimodality scholars is to provide a systematic theory
of “meaning making,” and they argue that typography, which can be presented as
a mode that has a system, or grammar, has three “metafunctions” as devised by
Halliday (1976). For example, using the image of the words DEAD MAN written
in a font that resembles bones, van Leeuwen (2006) argues that the bone-like font
plays an ideational metafunction, which Halliday (2007, p. 183) defines as “content
function of language.” He also demonstrates that a change from standard to italics,
or standard to bold could have an interpersonal metafunction, which is under-
stood as the function of language which expresses the relationship between the
speaker and hearer. For example, “ice” written in bold and upper case, as in “ICE,”
may be taken as communicating a demand, while it may be taken as a question if
written in italics (with a question mark), as in “ice?” Finally, van Leeuwen (2006)
argues that the use of bold and different colours, which is often seen in magazine

7. Note that each TV station typically has an in-house font for telops and rarely uses fonts
widely available on Windows such as MS Gothic or MS Mincho.
200 Ryoko Sasamoto and Minako O’Hagan

articles, demonstrates a textual metafunction, which is the last of three meta-

functions Halliday proposes. Textual metafunction manages how ideational and
interpersonal functions are linked and presented in text (cf. Halliday 2007, p. 184).

2.3 Issues with the multimodal approach

As we have seen, multimodal approaches provide a description of the set of tools

available to the communicator. And indeed, a framework based on multimodal
analysis of typography provides us with an insight into how the communicator
can choose semiotic resources for meaning making (using bold for emphasis, for
example). However, multimodal analysis does not explain how readers recover the
intended meaning, or how the use of these tools on screen gives rise to certain
stylistic effects for viewers. In addition, the “semiotics” of each typographical fea-
ture is not always guaranteed as it can vary from context to context. For example,
italics may be used for an emphasis in one context but as a marker of a quotation
in another. For this reason, it is not a matter for semiotics or semantics to provide
an explanation. That is, it is not entirely clear how multimodal description or clas-
sification could enable us to explain the context-specific nature of typographical
features.
Furthermore, it is not clear whether we need such artificial levels of, or even
psychological motivation for such “meaning” as ideational, interpersonal and
textual metafunctions in order to explain how we interpret communicative stimuli
such as utterances, let alone to account for the features of typography, which are
part of written stimuli. For example, while the use of certain fonts is likely to influ-
ence a reader’s or viewer’s interpretation, it is not clear how each kind of influence
can be distinctly explained. For example, if, as van Leeuwen (2006) argues, a
bone-like font plays an “ideational” function and leads the reader to recover a
sense of death, while the use of bold plays “interpersonal” function which leads
the reader to recover a sense of “demand,” how is it that typographical features play
a particular role at different levels of communication, and how should the reader
or viewer ascertain which role and level to select? Or, is it the case that the role of
typographical features can be explained independently of these so-called meta-
functions? As it stands, current multimodal analysis does not provide answers to
these questions. In the next section, we will see how relevance theory, a cognitively
grounded theory of communication, can provide a theoretical architecture in
which the relationship between “meaning making” tools and pragmatic effects
can be accounted for.
 Chapter 7. Relevance, style and multimodality 201

3. Relevance theory and pragmatic effects

So far, we have seen how typography has been accounted for in various approaches.
We have also seen that there is little agreement as to how typographical features
affect overall meaning despite the consensus that typographical features do af-
fect overall meaning. However, if we look at phenomena other than typography,
problems with overall stylistic effects are nothing new. In the traditional, descrip-
tive literature, intuitions about the relationship between style and linguistic form
were often accounted for by distinctions such as given-new and theme-rheme
(c.f. Halliday & Hasan, 1976; Green, 1980; Reinhart, 1981). In relevance theory,
the so-called “stylistic effects” are explained as pragmatic effects, which are com-
municated as a result of the use of a particular linguistic form. One of such cases
which are often discussed is emphasis8 (cf. Sperber & Wilson 1995; Clark, 2013).
Consider Example (1):
(1) a. Alfie plays the drums.
b. What Alfie plays is the drums.
c. It is the drums that Alfie plays.

In (1), all three examples communicate the fact that Alfie plays the drums.
However, the use of different linguistic structures puts an emphasis on a different
constituent. In contrast with (1a), which seems most “neutral” (i.e. without any
particular emphasis), the emphasis is on Alfie in (1b), while in (1c), the emphasis
is on the drums. In contrast with the traditional approaches to stylistic effects,
which often rely on artificial levels such as theme-rheme, topic-comment, given-
new, presupposition-focus, and background-foreground distinctions, in relevance
theory, stylistic effects emerge from a natural linkage between linguistic form and
pragmatic effects. Sperber and Wilson (1995) explain that these stylistic effects are
the result of the speaker exploiting structural features of utterances to convey the
intended interpretation at a minimum (justifiable) cost. They argue that linkages
between linguistic form and the overall interpretation such as in (1) above can be
explained in their cognitively-grounded relevance-theoretic framework, as “given
that utterances have constituent structure […] and given that they are processed
over time, the most cost-effective way of exploiting these structural features will
give rise to a variety of pragmatic effects” (Sperber and Wilson 1995, p. 217). In
other words, the speaker exploits whatever structures are available to him, in order
to communicate at a minimum cost possible. Similarly, Wilson and Wharton (2006)
demonstrate that so-called stylistic devices such as stress patterns can be used as

8. This is not to say that all cases of stylistic effects are linked to word order.
202 Ryoko Sasamoto and Minako O’Hagan

natural highlighting devices to draw attention to particular utterance constituents,

guiding the hearer to an optimally relevant interpretation in those contexts.
While relevance-theoretic stylistics enables us to account for how “particular
linguistic formulations give rise to particular kinds of effects” (Clark 2013, p. 352),
the focus has largely been on verbal communication. The linkage between non-
verbal features of communication and stylistic effects is under-researched, except,
perhaps, for Wharton’s (2009) work on non-verbal communication, Yus’s (2009)
work on multimodal metaphors and Scott and Jackson’s (this volume) work on
typography. Scott and Jackson’s work in this volume is particularly relevant to the
current study. Taking a relevance-theoretic approach, they examine a range of
typographical features, capitalisation and italicisation in particular, and argue that
the typographical choices are driven by the author’s communicative intensions
and considerations for relevance. Like Scott and Jackson (this volume), we also
consider the role of typography as the guide for the reader to reach the intended
interpretation. However, with increased interest in multimodality, it is imperative
that we take into consideration the fact that communication more often than not
comes in more than one mode. It is thus important to examine how ostensively
used multi-modal stimuli, such as telops on Japanese TV programmes, interact
with other programme elements, and influence the viewer interpretation process.
Blakemore (2008) examines a range of apposition cases from a variety of texts
and explains how apposition gives rise to non-propositional effects. These effects
are ineffable and will be lost under paraphrase. Blakemore (2008, p. 5) defines af-
fective communication as “communication of [non-paraphrasable] impressions, or
non propositional effects.” According to Blakemore, while cases of appositions can
be analysed in terms of “an impression of amplification or heightened vividness”
that they give rise to, the effects are not always derived in the same manner in each
case. She demonstrates two ways in which the amplifying effects can be derived:
(2) He felt depressed, flattened. 
 (SEU w.1.16.6.239–40. Cited originally by Meyer, 1992, p. 67)
(3) He made a complete mental retreat; went far away. 
 (Maurice Gee, In My Father’s Den, p.171).
 (examples from Blakemore 2008: 19)

According to Blakemore (2008), Example (2) is a case of intensification, where

the first and second segments of the apposition both provide the audience with
access to distinct sets of assumptions. The audience is hence guided to compare
the effects that each segment gives rise to. For example, in (2), the first segment
depressed yields a representation of a state of mind in a general sense, while the
second segment flattened, communicates a similar state of mind, but perhaps a
 Chapter 7. Relevance, style and multimodality 203

more intense feeling. Notice that in (2), flattened is used figuratively, i.e. we see
the use of language that has gone through the process of lexical adjustment (in
this case, broadening) and “its encyclopaedic entry includes representations of a
state of mind” (Blakemore, 2008, p. 24). The second segment thus encourages the
audience to access assumptions related not only for a person’s mental state but
also those related to the encoded meaning of flat. It is the audience’s responsibil-
ity to decide what these properties are, based on the comparison between two
concepts derived from the two segments the speaker produces. In so doing, the
speaker can provide the basis for deriving an ad-hoc concept, which would yield
a range of weak implicatures, giving a more vivid representation (and hence a
more intensified representation). In other words, according to Blakemore (2008),
intensification type appositions are a case of the speaker guiding the audience to
access two sets of assumptions, where the first might yield cognitive effects yet
the intended interpretation should be derived from the second set. The relevance
of such cases lies in the way that these appositions enable the speaker to draw
attention to the differences in the range of weak assumptions that can be derived
from the first and second segments.
In contrast, Example (3), according to Blakemore, is analysed as a case of hy-
brid representation. Here, unlike the case of intensification where two concepts
are processed separately and compared with each other in order to give rise to an
intensified interpretation, the two segments are combined for the communication
of a single intensified concept rather than two individual concepts. In such cases,
the range of implicatures that are derived from the combined concept cannot be
recovered from each segment processed independently. According to Blakemore
(2008), the sequence in (3) is intended as a description of someone accused of a
terrible crime. Here, both a complete mental retreat and went far away, due to the
word mental in the first segment, will result in interpretations where the character
is understood to seek mental refuge instead of physically going away. However,
according to Blakemore (2008), by providing the second segment, the writer can
guide the audience to further explore the context to recover a wider range of
contextual assumptions. In the case of (3), the second segment would encourage
the inclusion of assumptions about distance. Blakemore (2008, p. 27) explains that
the first segment will lead readers to access concepts such as withdraw, retire,
refuge, go back, while the second segment would lead readers to extend the
context on the basis of contextual assumptions associated with far away, which
then creates the sense of distance between the character and the narrator. As the
audience is given access to these two distinct sets of assumptions, they will be
able to explore these assumptions which will then result in the recovery of the
wider array of weakly communicated assumptions. This wider range of assump-
tions cannot be retrieved on the basis of each segment taken independently, hence
204 Ryoko Sasamoto and Minako O’Hagan

Blakemore’s term hybrid representation. The combined representation will

enable the audience to derive a wider range of weak implicatures and can thus lead
to more vivid representations.
Blakemore’s (2008) analysis is focused on apposition cases and their affective
interpretation and thus may seem irrelevant to cases of typography and their effect
on viewers. However, the fact that both the apposition cases and the typography
cases involve the communication of intangible feeling (and hence affective com-
munication) makes it worthwhile to explore the similarities between the phenom-
ena. Blakemore’s (2008) analysis of apposition and affective communication dem-
onstrates how such stylistic meanings can be explained in general terms without
the need for being treated as a special case of communication. The analysis thus
suggests that the choice of typography and the consequences it has for communi-
cation, does not require a treatment as a special case of communication either. It
is, rather, a case of exploiting whatever structure (or “semiotic resources” in multi-
modality terms) the communicator has at hand to achieve the intended effect at a
minimum justifiable cost; in a similar way that a writer takes advantage of apposi-
tion. Of course, there is a fundamental difference: while Blakemore’s apposition
cases are single-modal and involve appositions of two linguistic representations,
the typographical choice, which is the focus of this study, is multimodal and will
not have a concrete linguistic representation as such. However, a typographical
choice can easily be treated as a communicative stimulus, which will evoke a
certain interpretation, as it gives quick-and-dirty access (or a shortcut) to a set
of assumptions that are associated with the particular typographical feature.9 For
example, in some cases, typographical features may provide an extra access point
to a wider set of weak implicatures in a similar way to hybrid representations,
i.e. when combined with the linguistic input delivered by the base caption. That
is, it may be possible to argue that the very nature of being multimodal is what
enables telops to deliver a hybrid representation. This point will be illustrated later
in the next section.

9. In some cases, it may be a set of assumptions that are extremely weak which amounts only to
communication of impressions (c.f. Sperber & Wilson, 2015)
 Chapter 7. Relevance, style and multimodality 205

4. Case study: Theoretical explanation and reception study of

viewer behaviour

4.1 Research question and research design

As stated in Section 1, the aim of the current study is to analyse the effects that
typological features have on viewers’ interpretation processes, namely, how os-
tensively used, multi-modal stimuli such as telop interact with other programme
elements, and influence the viewer interpretation process. In order to answer this
question, we will investigate how typographical features of telops contribute to the
interpretation of multiple stimuli and also how typographical features of telops
influence interpretations.
We begin with an in-depth observation of multimodal data from a selected
Japanese TV programme. Drawing upon our observations and the relevance-
theoretic explanation of affective communication (Blakemore 2008), we propose
a cognitively-grounded explanation of the role that typographical features play
in the interpretation process. We then combine our explanation with some
empirical data gained from a preliminary eye-tracking study in Section 5 where
our relevance-theoretic explanation will underpin objective physiological data
from the eye-tracker.

4.2 Telops and typographical features

As illustrated in Figure 3, the use of more than one font and one colour is common
in most telops in certain types of entertainment programmes in Japan:
In Figures (3a) and (3b), sans-serif is used, while serifed font is used in
Figures (3c) and (3d). In Figure (3a), a keyword labelling the background is used,
while a text-box background is used in Figure (3d). Figure (3b) does not use any
background, while, in Figure (3c), the whole screen is blackened and used as a
background. It is not unusual to see the screen filled with nothing but writing
like this; such uses of captions are often seen as a summary panel, or as a cohesive
device to facilitate a narrative flow such as a change of topics.
In contrast with conventional subtitles, these telops are presented using vivid
colours while varying in font size. Except for Figure (3a), where the caption is writ-
ten in white, all other captions are presented in more than one colour. Moreover,
these captions may use multiple colours as in Figure 1 where four colours were used
in a single caption. Furthermore, telops often deploy animations (or text effects):
206 Ryoko Sasamoto and Minako O’Hagan

Caption in sans-serif font in

keyword-labelling background

a.

Telop in sans-serif, in two

colours. A half of the telop has a
purple ‘glow’ in the background

b.

Caption on blackened screen,

presented in red and white

c.

Captions in serifed font on

text-box background, letters are
given in red (left half ) and black
(right half )

d.

Figure 3. Captions with various fonts and backgrounds (Honmadekka!? 2013)

 Chapter 7. Relevance, style and multimodality 207

“I need to see him mentally

suffer…”

Figure 4. Telop with various colours and pulse effect (Honmadekka!? 2013)

In Figure 4 above, the female presenter is discussing how she would need to see
her cheating lover mentally suffer before she can forgive him. The telop, which is
written in black with the ‘glow’ effect in purple, pulsates as she speaks. Similarly,
in Figure 5 below, the telop at the bottom (in red) expands and then returns to its
original size (to the size of the first letter, which remains unchanged during the
effect) as if to emphasise the anger or irritation.

[As a response to a
panellist’s off-the-cuff
comment]
“Why are you offering that
[cold drink] to me!?”

Figure 5. Telop with enlargement effect (Honmadekka!? 2013)

In this programme we counted over 50 occurrences of font changes with over 100
cases of different effects being used within a 20-minute sequence and this seems
to be a typical frequency across other similar entertainment shows. As we can see
from these examples, there does not seem to be any limit to the use of ‘semiotic
resources’, and indeed, the TV directors seem to exploit typographical features
to the maximum of their potential. As Yamamoto (personal communication)
explains, the TV directors are determined to take advantage of any editorial prop
in order to create a programme that draws viewers’ attention, so that they can
stay ahead of the game with different programmes broadcast at the same time. If
they did not keep providing new stimuli, captions might lose their effects through
familiarity. Bombarding viewers with telops is one of the methds the TV directors
employ to keep the programme engaging. This raises the question of what effects
208 Ryoko Sasamoto and Minako O’Hagan

the typographical features give rise to and how they are conveyed. In the next
section, we will draw on the relevance-theoretic notion of pragmatic effects and
non-propositional effects and outline how they can be applied in an analysis of the
role of typographical features.

4.3 Typographical features and interpretation – telops and relevance theory

4.3.1 “Affective” interpretation and typographical features

This section provides an explanation on how such uses and typographical features
of telops can be accounted for in terms of relevance-theoretic stylistics. Writers
often use italics or bold to put an emphasis on their meaning. It is therefore not
surprising to see similar devices being used to create the effect of emphasis on
telops. The difference is, unlike with two-dimensional writing, TV directors can
exploit time and space for such a “form” and add animations (textual effects). Such
additional features as vibrations and enlargements of telops can simply be treated
as a choice of form that a programme director can exploit in order to give rise to a
variety of pragmatic effects.
What is interesting is that such stylistic effects as emphasis, communicated
by the ‘formal’ device such as added animation, are sometimes accompanied by
further pragmatic (or poetic even) effects, such as weak implicatures as we see
below. (cf. Sperber & Wilson, 1995; Pilkington, 2000; Wharton, 2009). There
are also cases where typographical features contribute to the communication of
propositional attitudes as we will see in 4.3.2. But let us first address the case of
weak implicatures. Recall Figure 4, in which a telop is used to display the pre-
senter’s desire to see her lover suffer. This telop is presented in a font that is typical
of the Japanese horror manga genre. It also uses a darker colour, compared to
other captions. It is in black and has a purple glow around it. In contrast with
typical stylistic effects derived from the use of particular linguistic forms (c.f.
Halliday & Hasan, 1976; Green, 1980; Reinhart, 1981), the use of such fonts and
colours does not communicate a specific assumption such as “emphasis.” Instead,
it communicates a more intangible feeling, or an impression,10 such as those often
associated with such emotions as jealousy, grudge or revenge, just like the
effects the hybrid representation cases discussed in Blakemore (2008) give rise
to. The presence of weak implicatures is not limited to verbal communication.
Wharton (2009), for example, shows how gestures and facial expressions can be
used to convey a range of weaker assumptions. In a very similar way, the use of
a particular font colour enables the TV director to guide the audience to recover
a range of assumptions associated with horror, or, in this case, a bunny boiler. In

10. See Sperber & Wilson (2015) for a fuller account of the communication of impressions.
 Chapter 7. Relevance, style and multimodality 209

addition, this telop has the added animated effect of vibration/pulsing. The use of
pulse echoes the feeling of “horror” conveyed by the use of this particular font,
resembling a bodily reaction often expected in a tense situation evoking fear. Such
exploitation of resemblance in communication has been discussed extensively in
relevance theory (cf. Sperber & Wilson, 1995; Noh, 2000; Wilson, 2000; Wilson &
Sperber, 2012) and indeed, what such added animated effect does is to exploit the
resemblance in form to communicate some other representations. In other words,
as with Blakemore’s (2008) hybrid representation cases, the use of multiple visual
effects creates a combined representation which facilitates the communication of
a wide range of weak implicatures.11
Similarly, in Figure 5, we can see the use of an animated effect (enlargement)
and a red font colour. The utterance which the telop represents is produced by the
host of this programme, who is one of the most popular comedians of all time in
Japan. When an expert panellist on this programme explains how cold drinks stop
people from being selfish, one of the studio (celebrity) panellists says “it is difficult
to make a selfish person drink,” while gesturing towards the host. The host, tak-
ing the cue for generating laughter, counters the panellist, saying “Why are you
offering that (drink) to me!?”, as if to say that the panellist implied the host was a
selfish person. Here the director exploits the structure of enlargement of the telop
to emphasise the show host’s “indignation,” and as a result, this emphasis encour-
ages the recovery of certain effects via highlighting devices. The pragmatic effect
of emphasis is communicated. In addition to this, the use of a blood-red colour
and a dominating typeface communicates a range of ineffable feelings associated
with emotions such as anger or danger (that awaits the panellist who angered the
powerful host). The font enlargement might also be taken as “shouting,” echoing
the feelings of anger communicated by the use of this particular colour.
In both cases typographical features provide quick access to a set of assump-
tions that are often associated with certain typographical elements. Such as-
sumptions are combined with the representation delivered by the linguistic input
provided by the encoded element of the telop. The combined representation would
then evolve into a set of assumptions that the audience is expected to recover.
However, this is not to say that these feelings or impressions are always communi-
cated by the use of particular fonts and colours (or coded by these typographical
features). In fact, there are cases where the chosen colour does not seem to be
associated with any particular emotion, unlike certain colours that are more or less
conventionalised (or grammaticalised). Instead, the use of certain typographical

11. As mentioned in Section 3, unlike Blakemore’s hybrid representation cases, interpreting

telop as ‘hybrid representations’ involves comparison of representations between different
modes due to its multimodal nature.
210 Ryoko Sasamoto and Minako O’Hagan

features in a specific context would allow the viewers to access a range of assump-
tions, not just one assigned ‘meaning’, which will help them recover the intended
effects. Recall Figure 2, where a member of the public was asked to play golf using
objects other than a golf club. In Figure 2, the second half of the telop is coloured
in yellow. Yellow is often associated with happiness, creativity, and sunshine in
many cultures (Colourmatters, 2012). However, in this case it is obvious that the
colour of the telop is chosen to match the colour of butter. The use of yellow in
this example highlights the fact that it is the butter, not a putter, which the speaker
is holding. Unlike the previous examples where typographical features provided
quick-and-dirty access to a ready-made set of assumptions about feelings of hor-
ror (in the case of Figure 4) or anger (in the case of Figure 5), the use of yellow in
this example coincides with one of the encyclopaedic entries for the concept but-
ter. In so doing, the TV director can emphasise humour arising from a feigned
surprise expressed by the speaker who contrasts between “butter” and “putter.”
Figure 2 was the case where the colour was used not to highlight a certain
emotional state, but to highlight a “gap” between an existing assumption (that it
is a putter that the speaker is holding) and a new assumption (realisation that it is
actually butter that he is holding), and, of course, the wordplay that links “butter”
and “putter.” By highlighting this gap, the TV director could emphasise the feeling
of “surprise” arising from incongruity. This suggests that we cannot assign “mean-
ing” to a typographical feature. For example, the colour red may invoke the pas-
sion and dedication of an athlete as opposed to the anger we saw in Figure 5, while
the horror font may invoke, in this case, an impression of mocking. Furthermore,
cross-modal associations, such as those between colours and emotions, might
be influenced by culture-specific assumptions. What we are suggesting here is
that viewers will have access to a range of assumptions which are associated with
such typographical features and recover a range of feelings/impressions that suit
this particular context. It is important to stress that we cannot simply assign the
role each typographical feature plays without a specific context. Such features, as
multimodality scholars seem to claim, are tools for the communicator to exploit
(or “semiotic resources”). However, it seems fair to suggest that typographical
features can influence the viewer interpretation process at least in two ways: as a
structure (through animation/text effect) that gives rise to a pragmatic effect, and
as a pointer for the recovery of a range of intangible feelings or impressions. Our
data shows that the interpretations are context-specific and recovered in line with
expectation of relevance.

4.3.2 Propositional attitudes and typographical features

In contrast to cases we have seen above, where the intention is not chiefly to affect
explicit content of the utterance, there are cases in which the change of the font can
 Chapter 7. Relevance, style and multimodality 211

affect the recovery of a higher-level explicature12 by influencing the interpretation

of a propositional attitude. Figure 6 demonstrates this:

“Obviously you think you are

the funniest comedian.”

a.

“I am so grateful for all the

job opportunities given to
me.”

b.

[Above the caption in (b)]

“Liar”

c.
Figure 6. Telop with font change (Honmadekka!? 2013)

In Figure 6 (a), celebrity guests tease the host for his self-assuredness, producing
remarks such as “obviously you think you are the funniest comedian (captioned).”

12. See, for example, Wilson and Sperber (1993) for a fuller discussion of a higher-level expli-
cature.
212 Ryoko Sasamoto and Minako O’Hagan

The host responds, in Figure 6 (b), denying the allegation of being so full of him-
self. He says “I am ever so grateful for all the wonderful job opportunities given to
me,” which is captioned. Responding to the host’s utterance, other panellists say
“liar” which is captioned in Figure 6 (c) just above the host’s line shown in Figure 6
(b) that is retained in Figure 6 (c).
The typographical features used in this sequence are particularly notewor-
thy, as not only does the font change from sans-serif to serifed, but the line is
now italicized. Although there are no set rules or guidelines as far as telops are
concerned, serifed fonts such as Times New Roman are often recommended for
official documents such as legal documents or business letters for their “clean and
authoritative” nature (Butterick, 2008). By swapping from a sans-serif font to a
more official looking font, the TV director can be taken as indicating that this
is what the host is “required” to state publicly rather than how he actually feels
privately. Although this is highly speculative, the change of typography may argu-
ably be providing an additional prompt for the viewer to doubt the host is being
sincere, and to conclude that it is his “official” stance. As a result, viewers would
recover a higher-level explicature such as the one in (4):
(4) The speaker (and the mediator) do not believe [what’s represented by the
telop]

This suggests that the typographical features of a written stimulus can sometimes
affect the explicit content of an utterance, and in this case, the change of font
affects the interpretation of a higher-level explicature, guiding the viewers to a
certain propositional attitude.

5. An empirical approach to telops

So far, we have argued that typographical features influence the way the viewer
would interpret the telop, and that the use of typographical features gives rise to a
range of pragmatic effects. The question now concerns how we could develop this
theoretical analysis further to demonstrate the influence of non-verbal stimuli in
multimodal contents which the audience is exposed to. In this section we provide
initial empirical support for this relevance-theoretic analysis of typographical fea-
tures. In essence we sought more empirical evidence to answer the following main
research question: how do ostensively used multi-modal stimuli such as Japanese
telops interact with other elements in the same programme, and influence the
viewer interpretation process? To answer this question, we would first need to sub-
stantiate the assumption that typographical features do affect the audience. To this
end, we conducted eye-tracking experiments, albeit on a pilot basis. Eye-tracking
 Chapter 7. Relevance, style and multimodality 213

technology enables us to evaluate audience reception of multimodal content in an

objective manner, and eye-tracking studies commonly involve measuring of the
cognitive effort13 of the subject, i.e. how much visual attention the subject pays
to the stimuli. However, in our study the eyetracking data was sought to provide
some empirical support for our theoretical discussion. So, the data was treated as
supplementary evidence and analysed in a qualitative manner rather than provid-
ing answers quantitatively.

5.1 Methodology: Eye-tracking

We conducted pilot experiments using the Tobii Eye-Glass, which is a 30 Hz,

monocular eye tracker. Sixteen participants (consisting of fourteen female and
two male subjects), all native speakers of Japanese, took part in this experiment.14
The mean age of the participants was 20.6 years with all of them having normal
or corrected-to-normal vision. No motor or neurological abnormalities were
reported. The experiment was approved by, and conducted in accordance with the
regulations of Dublin City University’s Research Ethics Committee. A short clip
(duration = 22 minutes: 99 seconds, including 90 seconds of TV commercials) was
prepared with a Japanese primetime programme called Honma dekka!? TV (Fuji
TV, July 13th, 2013), which is a popular variety show with a panel of experts and
celebrity guests. The clip was displayed on a 32 inch TV screen. The experiment
was conducted in a simulated living room setup with environmental controls. The
particular type of eyetracker device was used for its portability advantage over a
fixed-desktop type eye-tracker to ensure ecological validity.

5.2 Results: Eye-tracking data

Gaze data from each participant was visualised and compared. Of the total data
collected we had to discard data from 4 subjects due to data corruption and tech-
nical problems. Areas of Interest (AOIs) were set for a telop with added effects and
font changes. Visualisations were created for the full recording (i.e. the visualisa-
tion of the gaze data for the whole duration of the clip) from which we isolated
each particular segment of interest (i.e. the visualisation of the gaze data from the
duration that the chosen telop were being displayed on screen). Out of 438 telops,
17 were found to have added effects such as vibration, enlargement, or entrance

13. It is true that no agreement has been reached over a universal measure of cognitive process-
ing effort (see Just & Carpenter, 1993; Schultheis & Jameson, 2004)

14. It is true that it is a small-scale study. However, eye-tracking studies of this sort typically has
high attrition rate (O’Brien, 2009).
214 Ryoko Sasamoto and Minako O’Hagan

with special effects, such as the ones shown in Figures 4 and 5. We counted only
3 occasions where an italicised serifed font rather than the usual sans-serif, such
as the one in Figure 6(b) were used and we classify them as cases of “font change.”
We therefore categorised the data from the full recording into 4 groups and cre-
ated separate visualisations for each of the following scenarios: (a) Full recording
minus the above 3 cases of font change. This was done by removing the gaze data
for the duration of time that the 3 captions which were written in italicised serifed
font appeared on screen. This first category therefore contained 435 captions; (b)
Recording that contains only the 3 cases of font change; (c) Full recording minus
text effects. This was done by removing the gaze data for the duration of the 17 cap-
tions that were found to have added animated effects (i.e. vibration, enlargement,
and special entrance effect). This category therefore contained 421 captions; and
(d) the recorded data for the duration of the 17 captions that had added animated
effects. These were then compared for each participant. A ‘heat map’ was chosen
for visualisation as it would represent the intensity of the subjects’ visual attention
most clearly. In all recordings, there were no deviations (no extra effect, sound or
visual) observed in these particular scenes except for the textual effects and font
changes that appeared with the telop at the bottom of the screen.
In most cases, it was observed that the audiences’ gaze patterns changed in the
segments which contained telops with both font change and animated effect cases
in comparison to recordings of segments that do not contain such typographical
changes. In Figure 7, we can see the comparison of gaze patterns for 3 selected
subjects.15 In Figure 7, we compared segments of recordings that do and do not
contain changes of fonts in telops:
A heatmap visualization such as Figure 7 illustrates where participants’ fixa-
tion was concentrated. The degree of fixation is shown as hotspots – the colour
red shows the most fixations followed by yellow and then green. The label starting
with P is an identifier for each participant. In Figure 7, the images on the left
show the heat maps for the recording of segments that do not contain font changes
(category a), while the images on the right present heat maps for the recording
which contains font changes (category b). In all 3 cases, gaze patterns in the image
on the left (recording with no font change) show relatively intense attentions to
the upper part of the TV monitor with general coverage in the region where telops
are often inserted. In contrast, on the heat maps of the duration with font changes
(images on the right), all subjects show more erratic movement of their gazes,
which are more widely distributed (and so is the gaze intensity).

15. Images of all subjects (except for the discarded data from 4 subjects) are given in the
Appendix. Due to the colour restrictions we present the gaze visualisation only.
 Chapter 7. Relevance, style and multimodality 215

P01 Segments with no font change P01 Segments with font change

P07 Segments with no font change P07 Segments with font change

P10 Segments with no font change P10 Segments with font change
Figure 7. Font change and visualisation of eye-tracking data

Similar changes in gaze patterns are seen in comparisons between the seg-
ments which contained telop with animated effects such as vibration, enlargement,
and entrance with special effects (category c), and recordings of segments that do
not contain such text effect (category d).
Here, the gaze pattern is even more erratic, and it seems to move downwards
compared with segments with no text effect. The visualisation also shows that
the intensity of gaze is more widely distributed, rather than staying in one part
216 Ryoko Sasamoto and Minako O’Hagan

P01 Segments with no text effect P01 Segments with text effect

P07 Segments with no text effect P07 Segments with text effect

P10 Segments with no text effect P10 Segments with text effect
Figure 8. Text Effects and Visualisation of Eye-tracking data

of the screen as it does in the images on the left (the heat maps for segments
with no text effect).
While these visualisations do not provide supporting evidence in a quantitative
sense, they indicate that it may be worthwhile to explore the change in participants’
viewing behaviour. In both Figure 7 and Figure 8, gaze data for the full recording
seems to gather around one big cluster around the upper-centre position of the
screen, often with much less-intense gaze distribution towards the lower end of
 Chapter 7. Relevance, style and multimodality 217

the TV screen. Interestingly, the heat map for the full recording, which shows the
intensity of gaze throughout the recording, seems to focus on the upper side of the
screen rather than the bottom region where telops appear constantly. However,
in both cases, heat maps for text effects and font changes seem to show that gaze
moves in a more erratic manner than for the full recording, across the screen.
These findings suggest that typographical features could be the factor that in-
fluenced the participants’ attention at least on the basis of qualitative observations.
The result demonstrates how audiences’ visual attention is influenced somewhat
by typographical features and hence supports our analysis that typographical
features do contribute to the viewer interpretation process. The findings from this
study support what relevance theory predicts: these more ostensive stimuli, thanks
for the layering of typographical features such as font and colour, attract more
attention, so viewers pay more attention to them in pursuit of cognitive effects.

6. Conclusion

This paper is concerned with the growing importance of typographical features in

the interpretation process in the context of shifts in written communication due to
the use of digital devices. While the “core” of written communication may still rest
with verbal inputs, the contribution of non-verbal stimuli, such as typographical
features including fonts and colours to communication is not insignificant. With
the advances of technology, it has become easier to use various typographical
tools such as fonts and colours to help communicate ineffable feelings, which may
not be expressible in a concrete verbal representation. We demonstrated how TV
directors use typographical features in order to communicate intangible feelings
or impression using telops on Japanese TV programmes as a case study. We also
demonstrated how the viewers gaze pattern seems to be somewhat affected by
such stimuli. It was our aim to answer the question: how do typographical features
such as font and colour affect TV viewers’ interpretation processes? Japanese TV
programmes were ideal as our data source, as these programmes involve complex
layers of multimodal representations including the use of outlandish telops ex-
ploiting various typographical features.
Drawing upon Blakemore’s (2008) analysis of appositions and affective com-
munication, we have argued that typographical choices contribute to affective
communication in particular, helping to communicate intangible impressions or
feelings rather than one strong (or concrete) thought representation. Typographical
features contribute to relevance by functioning as a highlighting device to guide
viewer’s interpretation process. In some cases, they guide the viewers to recover a
range of pragmatic effects at an affective level (i.e. communication of layers of weak
218 Ryoko Sasamoto and Minako O’Hagan

assumptions) by contributing to the recovery of hybrid representations. In other

(probably rare) cases, the choice of a particular typographical feature leads the
viewer to recover a specific propositional attitude. We thus showed the important
role of ostensively-used non-verbal stimuli in communication and closed a gap in
the relevance theoretic literature by providing an account of an under-researched
stylistic device. The observation from our eye-tracking study, albeit on a pilot basis,
seems to indicate that viewers are influenced by the use of certain typographical
features, although further research is necessary to ascertain the degree of influence
typographical features may have on the viewers’ interpretation process.

Acknowledgements

We would like to thank DCU doctoral researcher in Computing, Feiyun Hu, for his continued
assistance with the Tobii Glasses, The Research Support office and the Faculty of Humanities
and Social Science at Dublin City University for their funding of this project under the
Enhancing Performance scheme in 2013, and Dr. Stephen Doherty for collaborating from
Sydney for this project.

References

Blakemore, D. (2008). Apposition and affective communication. Language and Literature, 17

(1), 37–57.  ​https://‍doi.org/10.1177/0963947007085054
Bornstein, R. F., & Becker-Matero, N. (2012). Metaphor and psychodynamic research. In K. S.
Montana (Ed.), Metaphor and fields: Common ground, common language, and the future of
psychoanalysis (pp. 92–103). London: Routledge
Butterick, M. (2008). Font recommendations. Available at: http://‍typographyforlawyers.com/
about.html
Clark, B. (2013). Relevance theory. Cambridge: Cambridge University Press. ​
https://‍doi.org/10.1017/CBO9781139034104
Color Matters. (2012). The Meanings of yellow. http://‍www.colormatters.com/yellow
Díaz-Cintas J., & Remael A. (2007). Audiovisual translation: Subtitling. Manchester: St. Jerome.
Doyle, J. R., & Bottomley, P. A. (2009). The massage in the medium: Transfer of connotative
meaning from typeface to names and products. Applied Cognitive Psychology 23(3),
396–409.  ​https://‍doi.org/10.1002/acp.1468
Green, G. (1980). Some wherefores of English inversions. Language, 56, 582–601. ​
https://‍doi.org/10.2307/414451
Halliday, M. A. K., & Hasan, R. (1976). Cohesion in English. London: Longman.
Halliday, M. A. K . (2007). Language and education. London: Continuum.
Henderson, P. W., Giese, J. L., & Cote, J. A. (2004). Impression management using typeface
design. Journal of Marketing, 68(4), 60–72.  ​https://‍doi.org/10.1509/jmkg.68.4.60.42736
 Chapter 7. Relevance, style and multimodality 219

Just, M. A., & Carpenter, P. A. (1993). The intensity dimension of thought: Pupillometric indices
of sentence processing. Canadian Journal of Experimental Psychology, 47 (2), 310–339. ​
https://‍doi.org/10.1037/h0078820
Kimura, T., Hosoi, A., Honda, N., Kato, Y., Kawamura, F., Koizumi, A., Oosawa, Y., Suzuki, E.
and Watabe, K. (2000). “テレビ画面に踊る文字たちの生態学” [Physiology of Letters
Dancing on TV screen], Galac, 366.
Lakens, D., Semin Gün R., & Foroni, F. (2012). But for the bad, there would not be good:
Grounding valence in brightness through shared relational structures. Journal of Experi-
mental Psychology: General, 141(3), 584–594.  ​https://‍doi.org/10.1037/a0026468
Meyer, Ch. F. 1992. Apposition in contemporary English. Cambridge: Cambridge University
Press.
Meier, B. P., Robinson, M. D., & Caven, A. J. (2008). Why a Big Mac is a good Mac: Associations
between affect and size. Basic and Applied Social Psychology, 30, 46–5. ​
https://‍doi.org/10.1080/01973530701866516
Meier, B. P., Robinson, M. D., Crawford, L. E., & Ahlvers, W. J. (2007). When “light” and “dark”
thoughts become light and dark responses: Affect biases brightness judgments. Emotion 7,
366–376.  ​https://‍doi.org/10.1037/1528-3542.7.2.366
Noh, E.- J. (2000). Metarepresentation: A relevance-theoretic approach. Amsterdam: John Ben-
jamins.  ​https://‍doi.org/10.1075/pbns.69
Nørgaard, N. (2009). The semiotics of typography in literary texts. A multimodal approach.
Orbis Litterarum, 64(2), 141–160.  ​https://‍doi.org/10.1111/j.1600-0730.2008.00949.x
O’Brien, S. (2009). Eye tracking in translation process research: Methodological challenges
and solutions. In I. Mees, F. Alves, & S. Gopferich (Eds.), Copenhagen studies in language:
Methodology, technology and innovation in translation process research (pp. 251–266).
Copenhagen: Samfundsitteratur.
O’Hagan, M. (2010). Japanese TV entertainment: Framing humour with open caption telop. In
D. Chiaro (Ed.), Translation, humour and the media: Translation and humour, Volume 2
(pp. 70–88). London: Continuum.
Reinhard, T. (1981). Pragmatics and linguistics: An analysis of sentence topics. Philosophica,
27, 53–94.
Sasamoto, R. (2014). Impact caption as a highlighting device: Attempts at viewer manipulation
on TV. Discourse, Context and Media, 6, 1–10.  ​https://‍doi.org/10.1016/j.dcm.2014.03.003
Sasamoto, R., & Doherty, S. (2013). Manipulating the mind of others – the use of subtitles as
viewer manipulation. Cognitive Futures of the Humanities, Bangor, UK , 04–APR–13–06–
APR–13.
Schultheis, H., & Jameson, A. (2004). Assessing cognitive load in adaptive hypermedia systems:
Physiological and behavioural methods. In W. Nejdl, & P. de Bra (Eds.), Adaptive hyperme-
dia and adaptive web-based systems (pp. 18–24). Eindhoven: Springer Verlag. ​
https://‍doi.org/10.1007/978-3-540-27780-4_26
Shiota, E. (2003). “ 関連性理論とテロップの理解” [Relevance Theory and Understanding of
Telop]. Journal of Ryukoku University, 63–91.
Sperber, D & Wilson, D. (1995). Relevance: Communication and cognition. 2nd ed. Oxford:
Blackwell.
Stökl, H. (2005). Typography: Body and dress of a text – a signing mode between language and
image. Visual Communication, 4 (2), 204–214.  ​https://‍doi.org/10.1177/1470357205053403
220 Ryoko Sasamoto and Minako O’Hagan

Takahashi T., & Katayama N. (2012). ‘‍伝わるデザイン：研究発表のユニバーサルデザイ

ン[Design to Communicate: Universal Design for Research Presentation]’ [Online]. Avail-
able at http://‍tsutawarudesign.web.fc2.com/seisaku.html
van Leeuwen, T. (2005). Typographic meaning. Visual Communication, 4( 2), 137–142. ​
https://‍doi.org/10.1177/1470357205053749
van Leeuwen, T. (2006). Towards a semiotics of typography. Information Design Journal, 14 (2),
139–155.  ​https://‍doi.org/10.1075/idj.14.2.06lee
Wharton, T. (2009). Pragmatics and non-verbal communication. Cambridge: Cambridge Univer-
sity Press.  ​https://‍doi.org/10.1017/CBO9780511635649
Wilson, D. (2000). Metarepresentation in linguistic communication. In D. Sperber (Ed.),
Metarepresentations: A multidisplinary perspective (pp. 411–48). Oxford: Oxford University
Press.
Wilson, D., & Sperber, D. (1993). Linguistic form and relevance. Lingua, 90(1), 1–25. ​
https://‍doi.org/10.1016/0024-3841(93)‍90058-5
Wilson, D., & Sperber, D. (2012). Meaning and relevance. Cambridge: Cambridge University
Press.  ​https://‍doi.org/10.1017/CBO9781139028370
Wilson, D. & Wharton, T. (2006). Relevance and prosody. Journal of Pragmatics, 38, 1559–1579. ​
https://‍doi.org/10.1016/j.pragma.2005.04.012
Yus, F. (2009). Visual metaphor versus verbal metaphor: A unified account. In Ch. Forceville, &
E. Uriós-Aparisi (Eds.), Multimodal Metaphor (pp. 145–172). Berlin: Mouton de Gruyter.

Sources of visual material

Himitsu no Kenmin Show. 2007–2015. Nihon Television Network. Produced by Shinichi Nose,
Minoru Sato, and Kie Shimizu. Tokyo.
Honmadekka!? 2009–2015. Fuji Television Network. Produced by Michiko Kametaka and
Masayuki Harashima. Tokyo.

Appendix. Visualisations of all participants

1. Comparison between segments with no text effects and segments with text
effects

P01 Segments with no text effects P01 Segments with text effects
 Chapter 7. Relevance, style and multimodality 221

P02 Segments with no text effects P02 Segments with text effects

P04 Segments with no text effects P04 Segments with text effects

P05 Segments with no text effects P05 Segments with text effects

P06 Segments with no text effects P06 Segments with text effects
222 Ryoko Sasamoto and Minako O’Hagan

P07 Segments with no text effects P07 Segments with text effects

P08 Segments with no text effects P08 Segments with text effects

P10 Segments with no text effects P10 Segments with text effects

P12 Segments with no text effects P12 Segments with text effects
 Chapter 7. Relevance, style and multimodality 223

P14 Segments with no text effects P14 Segments with text effects

P15 Segments with no text effects P15 Segments with text effects

P16 Segments with no text effects P16 Segments with text effects

2. Comparison between segments with font change and segments with font
change

P01 Segments with no font change P01 Segments with font change
224 Ryoko Sasamoto and Minako O’Hagan

P02 Segments with no font change P02 Segments with font change

P04 Segments with no font change P04 Segments with font change

P05 Segments with no font change P05 Segments with font change

P06 Segments with no font change P06 Segments with font change
 Chapter 7. Relevance, style and multimodality 225

P07 Segments with no font change P07 Segments with font change

P08 Segments with no font change P08 Segments with font change

P10 Segments with no font change P10 Segments with font change

P12 Segments with no font change P12 Segments with font change
226 Ryoko Sasamoto and Minako O’Hagan

P14 Segments with no font change P14 Segments with font change

P15 Segments with no font change P15 Segments with font change

P16 Segments with no font change P16 Segments with font change
Part 4

Pragmatic effects and emotions

 Chapter 8

Towards a relevance-theoretic account

of hate speech

Jadwiga Linde-Usiekniewicz
University of Warsaw, Poland

Hate speech has been studied as a social, psychological and legal phenomenon
in various frameworks offered by relevant disciplines, with most famous account
being Judith Butler’s analysis of hate speech in terms of Austinian speech act
theory. This chapter explores the possibility of applying relevance-theoretic
analysis to hate speech. It argues that all kinds of hate speech, ranging from
most direct to most covert, are instances of ostensive behavior that requires
being processed together with some mental representations, corresponding to
either intuitive or reflected beliefs about generalized inferiority of a group of
people by virtue of their race, ethnicity, faith/atheism, gender, sexual orientation,
sexual identity, or etc. or, their corresponding meta-representations, for greatest
(tentatively optimal) relevance.

Keywords: hate speech, covert and overt hate speech, direct and indirect hate
speech, hate speech and cognitive vigilance, slurs and quasi-slurs

1. Introduction

The purpose of the paper is to argue that ‘hate speech’ can be adequately de-
fined and described as a communicative phenomenon within the framework of
relevance theory, and additionally, that some of the issues associated with being
targeted by hate speech can also be satisfactorily described by relevance theory.
In particular, it will be argued that all the mechanisms discussed when the issue
of identifying ‘hate speech’ as such is raised, and specifically those that involve its
being understood, are the same as those analyzed when ordinary communication
is discussed, e.g. Wilson (1994).
Generally, and outside law, hate speech is defined as “speech that attacks a per-
son or group on the basis of race, gender, or sexual orientation” (Random House
Dictionary). Some hate-speech phenomena involve slurs, i.e. “derogatory terms

https://‍doi.org/10.1075/ftl.8.08lin
© 2020 John Benjamins Publishing Company
230 Jadwiga Linde-Usiekniewicz

targeting individuals and groups of individuals on the basis of race, nationality,

religion, gender or sexual orientation” (Bianchi 2014, p. 35), but hate speech is not
restricted to the use of such terms. “In legal literature, hate speech commonly re-
fers to incitement to hatred or discrimination against persons because they belong
to a certain group, such as race/ethnicity, religion, sexual orientation or gender”
(Noorloss, 2011, p. 4), from which it follows that different kinds of utterances may
qualify as hate speech.
The very term ‘hate speech’ is neither straightforward nor easy to define,
despite the quotes above. Two separate issues are involved: one concerns a seman-
tic, or rather an intensional definition of the term, and the other proposing an
extensional one, that would provide criteria for recognizing each and every item
of alleged hate speech for judiciary purposes (Brown, 2017a, b). Yet it should be
noted that several of the studies I cite rely on their subjects being able to recognize
instances of hate speech without either intensional or extensional definition.
The purpose of the analysis presented here is therefore two-fold. Firstly, it
aims to explain in relevance-theoretic terms some of the phenomena involved in
understanding instances of hate speech. Secondly, it proposes an approximation to
an intensional definition of hate speech within the relevance-theoretic framework.
Relevance theory seems an unsuitable framework for analyzing utterances that
carry the negative emotional load of hatred or incitement to hatred. First of all,
early relevance theory has focused on informative aspect of verbal communication
and cognition, with emotions being “treated as external to cognition” (Piskorska,
2012, p. 107). Although it appears that there is growing body of opinion that the
emotional component should be somehow factored into relevance-theoretic anal-
yses, this paper attempts to present hate speech as a purely cognitive phenomenon,
reformulating the ‘hate’ component, or possibly the HATE* component involved
as an issue of attitude, and not emotion, in line with Brown (2017a).

2. Background

What has given me the idea to try to analyze it in terms of relevance theory was
some instances of hate speech in Polish, where in most cases the Prosecution
initially declined to charge, and the legal reason they gave in each instance for
not prosecuting. Some of the reasons the two examples of alleged hate speech
became causes celèbres was that the Prosecution’s decisions raised a lot of criti-
cisms among general public: they have been widely commented at least in more
liberal press (with some misrepresentations). In addition, they have spurred
the Procurator General to issue a set of guidelines concerning prosecuting hate
crimes and instances of hate speech indictable by Polish law. Importantly, as far as
 Chapter 8. Towards a relevance-theoretic account of hate speech 231

I could ascertain, in some of these cases the decision not to indict was eventually
reversed, and at least in one case the guilty verdict and the sentencing was upheld
in the Appellate Court.
In general, Polish law does not criminalize generic hate speech, but some
instances of hate speech targeted at groups defined by nation, race, ethnicity, re-
ligious affiliation or lack of religious affiliation are indictable, as seen in following
quotes from the Penal Code:
Article 256. Whoever publicly promotes a fascist or other totalitarian system of
state or incites hatred based on national, ethnic, race or religious differences or for
reason of lack of any religious denomination shall be subject to a fine, the penalty
of restriction of liberty or the penalty of deprivation of liberty for up to 2 years.

Article 257: Whoever publicly insults a group within the population or a particu-
lar person because of his national, ethnic, race or religious affiliation or because of
his lack of any religious denomination or for these reasons breaches the personal
inviolability of another individual shall be subject to the penalty of deprivation of
liberty for up to 3 years.

The majority of instances in question, which eventually became causes celèbres

and will be further examined here, concerned anti-Semitic, or at least in part anti-
Semitic hate speech, though as will be seen from other examples used in the text,
Polish hate speech targets all kinds of groups, including blacks, Arabs and LGBT
persons (Bilewicz et al., 2014; Czarnecki, 2009a).
The first example is a comment posted by somebody, allegedly wronged by a
local businessman in Kielce:
(1) […] żydowskie ścierwo. Przestrzegam was przed tą firmą. To są typowi rasowi
Żydzi
‘Jewish scum (lit. carrion). I warn you against this company. These are
typical racial Jews’

The businessman so assaulted filed a complaint. The Prosecutor declined to indict

and justified their decision by saying that calling someone a Jew (whether they
are Jewish or not) is not an insult, and calling somebody scum may be “irony or
satire,” thus there were no grounds for charging the writer under Article 257. The
statement that the word ścierwo ‘scum’ can be used ironically actually comes from
an expert opinion the Prosecutor had sought.
There were several instances where at a football match against Widzew the
supporters of the opposite team chanted
(2) Hamas, Hamas, Juden auf den Gas
‘Hamas, Hamas, Jews to the gas (ovens)’
232 Jadwiga Linde-Usiekniewicz

or other chants of that ilk, e.g.

(3) Jazda z Żydami
‘Clear the Jews out’

The Prosecution declined to indict (2) under Article 256 on the basis that the re-
marks were not ‘incitement to hatred’ against Jews because they were not targeted
against ‘real’ Jews, but against the opposing Widzew team (who are traditionally
nicknamed Jews in football parlance). In some instances the court dismissed the
charges on the basis that the football hooligans had no intention of verbally attack-
ing Jews. In one instance the Appellate Prosecutor’s Office proffered charges, and
the court found the defendants guilty. The sentence included a fine and a court
order to view Cud Purymowy “The Purim Miracle”: a movie in which a family of
rampant anti-Semites and football fans learns that they are of Jewish origin. The
verdict was appealed against, but it was upheld by the Appellate Court.
In yet another instance, a Prosecutor refused to proffer charges against people
who had painted swastikas on the walls of buildings, and he argued that paint-
ing swastikas did not necessary promote fascist system of state or incite to hatred
(cf. Article 256), since swastikas are also Hindu symbols of luck. This particular
Procurator did not deem it necessary to seek an expert opinion (there is no such
obligation under law) and came by his arguments all by himself. He was later
dismissed from his post.
Interestingly, there is yet another example of the judiciary failing to find an
instance of hate speech an incitement to hatred. The case made it to the Supreme
Court, which ruled that the banner featuring:
(4) Wyzwolimy Polskę od euro – zdrajców, Żydów, masonów i rządowej mafii
‘We will liberate Poland from euro-traitors, Jews, masons and the
government mafia’

was not an instance of incitement to hatred, because among other things, the verb
form appearing on the banner was the Future Indicative and not Imperative. (SN
2007)
While there might be and are specific legal issues involving differences in what
is and what is not indictable under Article 256 and Article 257 and what constitutes
a crime under each of them (the same Supreme Court ruling defines “incitement
to hatred,” SN 2007), there is no doubt that outside the jurisprudence, or when
common reason is applied, there is something basically wrong in maintaining that
saying Jewish scum is not a verbal attack and in maintaining that Hamas, Hamas,
Juden auf den Gas is not a verbal attack against Jews, even when directed against a
football team, i.e. a group not overtly recognized as Jewish. This means that these
examples are understood by the general public as instances of hate speech and the
 Chapter 8. Towards a relevance-theoretic account of hate speech 233

failure to prosecute them can be explained in terms of failing to attach to them

the communicative import they do have in real life. In other words, what emerges
from the justifications given is that the law has analyzed these instances in total
contravention of principles that govern human communication, or, to be more
specific, as if relevance theory did not apply to human communication when ex-
amined in judiciary settings. Moreover, my initial observation was that all the in-
stances quoted are easily analyzed as instances of hate speech if relevance-theoretic
tools are applied to them. This observation was further confirmed by the analyses
themselves, and in addition relevance theory has proved a tool that enables a theo-
retically uniform account not only of the linguistic and communicative nature of
hate speech but also of various extra-linguistic phenomena associated with it.

3. Hate speech: Some legal, social and pragmatic issues

Some types of hate speech constitute indictable offences under specific national
laws, e.g. the two articles of Polish Penal Code quoted in the previous section
(Also see for example Noorloss, 2011, for a comparative review of laws in several
European countries). The most extreme kind of hate speech, called “dangerous
speech” (Benesch, 2012a, b) i.e. one that constitutes “incitement to genocide”
(Benesch, 2008) is punishable under Article 25 of the Rome Statute. Nevertheless,
there is a general consensus among scholars who study hate speech phenomena
that in most cases hate speakers couch their speech in such a way that it does not
infringe the existing laws (Mason, 2007, p. 34).
On the other hand, different international and national bodies do monitor
press, media and internet in order to document instances of hate speech. However,
in no case have I found any mention of valid linguistic criteria for identifying such
instances. This is not surprising because even if they existed, they would serve to
identify only some instances, and would result in falsely identifying some other
texts as hate speech. Thus, the instances of hate speech discussed in most research
have been identified by individuals (representatives of monitoring bodies, as in
Leets, 2002; Czarnecki, 2009a) or expert judges (Bilewicz et al, 2014; Warner &
Hirschberg, 2012). In most cases it can be assumed, or is explicitly stated, that the
monitoring or judging individuals are members of the targeted groups or sympa-
thetic to their cause, with the exception of Kwok & Wong (2013), who state that
their human evaluators, of the same age and gender, but of different races, diverged
in their judgments about the degree of offensiveness of the studied remarks.
Interestingly enough, both those scholars who seek to establish automatic
tools for hate speech recognition and those who manually analyze internet sites
of organizations known for their supremacist views tend to agree that the lexical
234 Jadwiga Linde-Usiekniewicz

contents of hate speech utterances is not predictive enough for hate speech. It is
also recognized that the nature of the verbal attack ranges from blatantly overt
to covert, allusive, but nevertheless strong. The overt attacks may include obvi-
ous racial slurs or offensive terms, as in the examples of anti-gay material used in
Leets (2002, p. 357):
(5) a. Queer!
b. Fag!
c. You fucking faggot!while covert hate speech may be quite convoluted as
in:
(6) […] tym popisem logiki pan Gebert skutecznie rozbija etniczny stereotyp
przypisujący Żydom szczególną inteligencję 
 (Kowalski & Tulli, 2003, quoted by Nijałkowski, 2008)
‘by this feat of logic Mr. Gebert successfully explodes (lit. smashes) the
ethnic stereotype that attributes special intelligence to Jews’.1

or actually appear as a part of an apparent praise, as in some of the stimulus mate-

rial used in Leets (2001, p. 685):
(7) After leaving a class discussion on affirmative action, a European American
student said to an [African American/Asian American/Hispanic American]
classmate:
a. You must have some white blood in you, because most blacks can’t make
it at a white university. They just don’t have what it takes.
b. You must have some white blood in you, because most Orientals aren’t
as articulate and bold in expressing their ideas. They just don’t have the
same confidence.
c. You must have some white blood in you, because most wetbacks can’t
make it at a white university. They just lack the sophistication.2

At the same time, expert judges who identify hate speech instances for the pur-
pose of research and members of monitoring bodies do recognize hate speech in
most instances. In some cases the perception or evaluation may differ from one

1. The example was culled from Nowy czas, a right wing Polish language magazine published in
the UK. Konstanty Gebert is a Polish Jew, a renowned journalist, columnist and writer. Prefixing
the name with pan (Mr) in a newspaper polemics in Polish is a mark of disrespect. It should be
noted that both Kowalski & Tulli (2003) and Nijałkowski (2008) classify it an instance of hate
speech, maybe because its blatant anti-Semitic character, and attacking the writer not on the
basis of what he had written but on the basis of his ethnicity.

2. In the Hispanic American version of the material a racially offensive term appears, absent
in the two other versions. There is anecdotal evidence of similar kinds of back-handed compli-
ments, offered apparently in good faith, toward Israelis, praising them for financial acumen.
 Chapter 8. Towards a relevance-theoretic account of hate speech 235

individual to another or from a member of one minority to another (Kwok & Wong
2013). Yet the way hate speech is defined in the literature confirms the impression
that it somehow defies linguistic definition. The researchers underline that “hate
speech can hurt and/or undermine human dignity” (Leets 2002), however, the
question how to explain the way in which this occurs in linguistic terms remains
unresolved. Specifically, while several frameworks may be helpful in identifying
a particular instance of hate speech, at the same time it can be easily argued that
the frameworks that suggest themselves for the analysis would apply different
mechanism to each particular instance. Though since van Dijk (1987) it has been
advocated that Critical Discourse Analysis may be a tool of choice (Nijakowski
2008), I have argued (Linde-Usiekniewicz 2015a) that its reliance on context, if
misapplied, may bring about mistaken analyses. In the case of (2) the immediate
context, i.e. of a football match and football rant, could be and indeed was used
as an argument against (2) being an instance of incitement to hatred and possibly
against its being hate speech, as apparently no ‘real’ Jews were thus attacked. It
was only against a larger context, i.e. the historical fact of Holocaust and Nazi
gas ovens, that the hate component, and even the incitement to hatred, indictable
under Article 256 of Polish Penal Code, was identified. Additionally, expert judges
and research subjects in psychological and sociological research (e.g. Leets, 2001,
2002; Bilewicz et al., 2014) are not presented with real-life contexts, yet are capable
of identifying hate speech and even judging how harmful it is.
Surprisingly enough, as I have argued elsewhere (Linde-Usiekniewicz,
2015a), Speech Act Theory is not as promising a tool as it would appear at first
glance, in spite of Butler’s (1997) claim. The very distinction she draws between
‘illocutionary’ and ‘perlocutionary’ performatives throughout her book shows
that hate speech comes in many guises and can be traced to all aspects of speech
acts. When Austin’s concepts are applied to hate speech in a more orthodox way,
it can be shown that in some instances hate speech can be perfectly produced
and adequately captured as an element of the locutionary aspect of a speech act
(Austin, 1962, p. 102), as in:
(8) Pedalski pomiot bezcześci buciorami pamięć po polskich bohaterach.
‘A faggot spawn defiles the memory of Polish [fallen] heroes with [her]
clumpers [i.e. heavy shoes].’

(quoted in Czarnecki 2009b, p. 18), or stimulus material from hate letters to Gay
and Lesbian Alliance Against Defamation, used by Leets (2002, p. 347)
(9) a.
It’s time for you and your kind (devil’s spawn) to run back into the
closet and slam the door behind
b. To hell with you and your way of life you butt screwers and pussy lickers
236 Jadwiga Linde-Usiekniewicz

Both in (8) and (9) the hate component is generated by the phatic act (Austin 162,
p. 92), i.e. the use of such phrases as pedalski pomiot, devil’s spawn and butt screw-
ers and pussy lickers, and by the rhetic act (Austin 162, p. 114), i.e. the reference to
specific individuals.
Hate speech can be traced to illocutionary acts, as in yet another piece of
experimental stimulus material used by Leets (2001, p. 685): “A female European
American student was walking with her [African American/ Asian American/
Hispanic American] boyfriend to class. Two male European American students
walked by and warned her to:
(10) get your arm off that [jungle bunny/ slope/beaner]

Again, if the issue of ‘real Jews’ vs. ‘opposing football team’ is set aside, (3) and (4)
can be qualified as illocutionary acts. Yet the harm perpetrated by hate speech,
discussed in psychological and sociological works quoted, can be equated in some
cases with the perlocutionary act and in most cases not as an act of perlocution,
but as a perlocutionary sequel (Austin, 1962, p. 117). Moreover, understanding
an utterance as an instance of hate speech does not require the recognition of
the speech act (Nicolle, 2000, p. 235). Just the opposite, (4) is easily judged as
instance of hate speech independently of its alleged or dismissed perlocutionary
consequence of inciting racial hatred.3
The speech act theory analysis of hate speech instances, briefly presented
above, can be seen as just the evidence for hate speech phenomena not constituting
a homogenous class, deploying locutionary, illocutionary or perlocutionary strat-
egies. This in itself is not a weakness,4 but may turn into one in judiciary situations,
where complaints are wrongly dismissed, because Prosecutors or Courts use the
absence of hate speech phenomenon on one level (be it locutionary, illocutionary
or perlocutionary) to decree that law has not been infringed (Linde-Usiekniewicz,
2015b). Specifically, in the swastika example the Prosecutor focused on the phatic
and completely missed the rhetic aspect, while in the case of (4) the Supreme Court
gave weigh to the locutionary act, and completely dismissed the illocutionary act.
By contrast, in (2) the initial refusal to prosecute can be seen as a result of focusing
on the rhetic (reference to football team) and the illocutionary (football rant),
while ignoring the phatic aspect.

3. Actually, when I first presented this example to a Polish audience quite well versed in theo-
retical pragmatics, their outraged reaction was to wonder whether the Supreme Court seriously
had considered it a promise and not a threat, as it in fact is (see below for a detailed analysis).

4. This has been rightly pointed out to me by an anonymous reviewer.

 Chapter 8. Towards a relevance-theoretic account of hate speech 237

4. Some terminological refinements

There is indeed reasonable ground for saying that (4) can be understood as a prom-
ise or as a threat, depending on who the audience is. And this leads to the necessity
of drawing some distinctions within the complex phenomenon of hate speech.
In her interview with Timothy Garton Ash, Susan Benesch (2012a) made
a distinction between ‘direct hate speech’ and ‘indirect hate speech’. Direct hate
speech occurs when the attacked minority, or the group targeted by the attack are
at the same time the (intended) audience. Indirect hate speech occurs when the
members of the targeted group are not the (intended) audience. Thus incitement
to hatred or to genocide is an instance of indirect hate speech, while slandering or
abusing some minority groups constitute direct hate speech. Among the examples
quoted so far (5), (7), and (9) are obvious instances of direct hate speech, while (1),
(4), (6) and (8) are equally obvious examples of indirect hate speech. Example (10)
does not fall easily into either group: the European American female can be con-
strued either as a member of the targeted group (by virtue of associating with a
member of the minority), and in such case the example could be qualified as direct,
or as a non-member: under this interpretation the hate speech becomes indirect.
Examples (2) and (3) are more complex: as they are football rants they may be
construed as either rooting for the speaker’ team, in which case the addressee is
the same team and co-fans, or as ranting against the rival team. Yet the distinction
is immaterial as far as their being an instance of hate speech is concerned.
Interestingly enough, different studies discussed so far seem to concentrate
on either one or the other type. Thus Leets (2001, 2002) focuses on direct hate
speech to evaluate the response. Van Dijk (1987), Mason (2007), and the Polish
monitoring publications (Kowalski & Tulli, 2003; Czarnecki, 2009a) studied the
indirect hate speech. Yet, although the claim that hate speech hurts people is made
without distinguishing the two kinds, it can be argued that the indirect hate speech
hurts the targeted groups in at least two ways. One is through perpetuating their
image as inferior, possessing all the negative characteristics attributed to them,
and the other is through making them witness hate speech. Thus, obviously (4),
can be seen as a promise (to the construed audience of those who would share the
view of the banner-carrier that Poland is oppressed by euro-terrorists, masons,
Jews and the government mafia) but as an equally obvious threat to a Jewish per-
son (and possibly to anyone identifying themselves as a member of other groups
listed on the banner).
For the purpose of analyzing hate speech yet another distinction has to be
made. Some instances of hate speech are overt enough: slurs, fighting words, of-
fensive terms and threats are used. Such is the case of żydowskie ścierwo in (1), the
Prosecutor and their expert’s opinion notwithstanding, and of (5a)–(c), as well
238 Jadwiga Linde-Usiekniewicz

as of the racial slur in (7c), and of pedalski pomiot ‘faggot spawn’ in (8) or the
insults in (9). Other interesting examples come again from Leets (2002, p. 347)
stimulus material.
(11) a.
I don’t give a damn what you say about me, you bagel-eating, hook-
nose, lox-eating . . . Jew
b. To the “white Jews” in the audience, I say: It’s gonna be a rough ride,
buddy . . . . Buckle your seat
belts . . . because I didn’t come to pin the tail on the donkey, I came to
pin the tail on the honkey

Yet in other instances the derogatory character of a remark, the one that leads to
its being classified as an instance of hate speech is covert or partly covert. Such is
the case of (6), and of (7a)–(b), and also of many of the examples analyzed by van
Dijk (1987) and by Mason (2007). The allusive nature of hate speech, at least of its
Polish variety, was noted by Nijałkowski (2008), who highlights its manipulative
character and reliance on implicatures and presuppositions.5
It is widely recognized that ‘hate speech’ does not need to be linguistic. It in-
cludes the use of symbols, either negatively established, i.e. with a long and painful
history of use, such as swastikas or burning crosses (as in the famous RAV case
discussed in depth in Butler (1997)), or less conventionalized, such as throwing
a banana at Dani Alvez at the Spanish League match, April 2014, or defacing
mosque walls with drawings of pigs.6 If these and similar instances are regarded
as hate speech, it is no longer speech in the ordinary sense, but a type of ostensive
communicative behavior, as defined by Sperber & Wilson (1986) and Wilson
& Sperber (2004) among others. This, coupled with the varieties of hate speech
established above, and the observed fact that the element that triggers the recogni-
tion of an utterance as belonging to the hate speech genre can be a part of its literal
meaning to be decoded, or more often an allusion or implicature to be inferred
(here in the relevance-theoretic sense of these terms), strongly suggests that hate
speech can be adequately and more importantly, uniformly analyzed in relevance-
theoretic terms. Indeed, the initial failures to prosecute in cases illustrated in (1)
and (2) and Supreme Court ruling concerning (4) show that the judiciary failed

5. Nijałkowski uses the term ‘implicature’ in the Gricean sense. In line with many discourse
scholars he considers implicatures and presuppositions as manipulative, i.e. ethically wrong,
elements of language use.

6. In the latter case, as in RAV, there is a legal issue involved in qualifying them as ‘hate speech’
crimes or simply ‘hate crimes’. If they are considered means of expressions, and not simple acts
of vandalism and attacks against property, there are freedom of expression issues involved (as
can be seen in Justice Scalia ruling in RAV vs. St. Paul)
 Chapter 8. Towards a relevance-theoretic account of hate speech 239

to apply ordinary rules of utterance understanding. Psychological and sociological

studies into hate speech, e.g. Leets (2001, 2002), as well as monitoring all kinds of
media (e.g. Kowalski & Tulli, 2003; Czarnecki, 2009a; Bilewicz et al., 2014) prove
that people are capable of understanding hate speech in all its variations, and the
type is immaterial for the recognition and the effect to take place.

5. Understanding covert hate speech in relevance-theoretic terms

In the closing part of the previous section I have given some reasons for applying
relevance theory to the analysis of hate speech. In this section I will focus on its
most elusive form, i.e. the covert kind, be it direct or indirect. I will try to show
that instances of this subgenre of hate speech are understood thorough the same
mechanisms as happens with non-assaulting utterances (Wilson, 1994).
This is a (somehow) novel approach, since relevance theory is seldom applied
to hostile linguistic behavior and related areas: notable exceptions have been Mateo
& Yus’s (2000) analysis of slurs and their translation (but not of those specific slurs
that are elements of hate speech) and an attempt at their taxonomy (Mateo & Yus
2013), and an attempt by Bianchi (2014) to analyze the phenomenon of appropria-
tion of hate speech elements (slurs) as an instance of their echoic use, followed
by Blakemore (2014) paper on the difference between slurs and expletives. Some
research into jokes (e.g. Solska, 2012), or more generalized pragmatic issues, e.g.
Zegarac (2012), analyze hostile linguistic behavior, yet the instances they focus on
are far removed from hate speech itself.
One of the reasons might be the fact that relevance theory generally assumes
that the speaker is more often collaborative than not in terms of her having the
informative intention and being “anxious to avoid misunderstanding, and [..] ac-
tively helping the hearer to recognize the intended interpretation” (Wilson, 1994).
The only instance she is seen differently is when the notion of epistemic vigilance
is invoked (Sperber et al., 2010; Mazzarella, 2013). However, in these accounts the
possible alternative to the competence and benevolence of the speaker is her com-
municative incompetence and malevolence: the latter understood not as hostility
but as intention to misinform.
Yet another reason is that relevance theory focuses strongly on the com-
municative intention of the speaker, which in turn is built upon her informative
intention. Interestingly, even when relevance theory is compared to speech act
theories, the examples discussed are of speech acts based on informative inten-
tions and socially important information (Nicolle, 2000; Jakobsen, 2010). The
general consensus is to consider emotions as something external to cognition
(Piskorska, 2012, p. 107), though the author argues convincingly that emotions
240 Jadwiga Linde-Usiekniewicz

may influence the way utterances are processed. Nevertheless, I will try to show
that the relevance-theoretic account of hate speech does not need to factor in emo-
tions into the model of comprehension.7
In my analysis of hate speech I will be relying on several basic relevance theory
tenets, quoted below:
Relevance-guided comprehension procedure:

a. Follow a path of least effort in computing cognitive effects: Test interpretative

hypotheses (disambiguations, reference resolutions, implicatures, etc.) in
order of accessibility.
b. Stop when your expectations of relevance are satisfied. 
 (Wilson & Sperber, 2004, p. 613)

Optimal relevance:
An ostensive stimulus is optimally relevant to an audience iff

i. It is relevant enough to be worth the audience’s processing effort

ii. Is the most relevant one compatible with the communicator abilities and
preferences. (Wilson & Sperber 2004, p. 613)

Optimal relevance
An utterance, on a given interpretation, is optimally relevant if and only if:

a. it achieves enough contextual effects to be worth the hearer’s attention;

b. it puts the hearer to no gratuitous processing effort in achieving those effects.
 (Wilson, 1994, p. 50)

These have to be seen against a larger backdrop of relevance theory, in which the
cognitive effects and the processing effort depend on the assumptions already
present in the audience’s cognitive environment and, specifically, on intended
contextual assumptions, i.e. the implicated premises (Wilson & Sperber, 2004,
pp. 608–611, 615–616). In particular, the assumptions already present in the
audience’s cognitive environment need to take form of mental representations
(Wilson & Sperber, 2004), corresponding to either intuitive or reflective beliefs
(Sperber, 1997).
For hate speech these assumptions have to concern a generalized inferiority
of a group of people by virtue of their race, ethnicity, faith/atheism, gender, sexual
orientation, sexual identity, etc. and it is against them that the utterances in question
need to be processed for greatest (tentatively optimal) relevance. In addition, these

7. In the initial version of this paper (Linde-Usiekniewicz, 2014) I argued that there is an
emotional component involved in the way the targeted groups process hate speech. Here I will
argue (in the next section) that it can be analyzed without referring to emotional impact. I am
indebted to Deidre Wilson’s remarks for directing my attention to this possibility.
 Chapter 8. Towards a relevance-theoretic account of hate speech 241

assumptions must be widely accessible to the audience. In other words, hate speech
only works if there is an appropriate set of assumptions, available or manifest.8
Some of these assumptions might be associated directly with the stimulus itself:
such is the case of the swastika or other forms of pictorial defacement. Obviously,
if the audience fails to recognize the association, or fails to identify the ostensive
character of the stimulus, it would also fail to process the utterance adequately.9
In evidence for my claim that covert hate speech delivery and understanding
follow the basic principles of relevance theory and can be adequately analyzed in its
terms I will give abbreviated and tentative analyses of three examples quoted above.
The first one is (4), a banner carried by a participant of the ‘Independence
March’, i.e. a march organized by ultranationalist movement to celebrate
the Independence Day (the anniversary of re-establishment of the Republic
of Poland in 1918).
When the utterance is decoded and the reference is resolved (there is no am-
biguity involved) and the presupposition that Poland is not a free state (resulting
from the use of the verb to liberate) is made manifest, one arrives at the implicated
premises, intuitively quite strong.
(12) a.
The reason Poland is not free has to do with the presence of certain
groups of peoplec.These groups include people referred to as ‘euro-
traitors’, Jews, masons and a governmental mafia.
b. These groups include people referred to as “euro-traitors”, Jews, masons
and a governmental mafia.

Moreover, the implicated premises, which are activated by the terms euro-traitors
and governmental mafia include:
(13) a.
One of the reasons Poland is not a free state is the fact that it accessed
the European Union.
b. Those who allowed it to happen/made it happen are traitors to Polish
sovereignty.

8. These assumptions usually would correspond to what is otherwise considered prejudices.

Nevertheless, at this stage of the argument they have to be treated as any bona fide assump-
tions. I will return to the issue at a later stage. For Polish the presence of these assumptions is
documented in Bilewicz et al., (2014). The notion of generalized inferiority introduced here
dispenses with the problem of specificity raised in Blakemore’s criticism of Hom (2008). The
issue is further discussed in the section devoted to slurs and is illustrated by the discussion
surrounding Example (18).

9. That actually happened with a Polish audience presented with the RAV example: some of
them did not understand the significance of the burning cross and believed that the issue really
was the fire hazard.
242 Jadwiga Linde-Usiekniewicz

c. The current Polish government is a mafia, i.e. it is dishonest and not a

valid government.

Coupling the two resident evils with Jews and masons (by virtue of conjoining all
four) gives rise to the following strong implicatures:
(14) a.
Euro-traitors and the government mafia are not the only enemies of the
state.
b. Masons and Jews resident in Poland are enemies as well.

These implicatures are made strong because they are built upon two other impli-
cated premises, traditionally upheld by Polish nationalist movement:
(15) a.
Jews residing in Poland are an internal enemy, as they are not ethnic
Poles.
b. Jews should be blamed for all evil that befalls Poland.
c. Masons are an internal enemy of Poland because they represent an
occult and international society, with its own goals that are detrimental
to the interests of Poland and specifically to the ethnically construed
Polish nation.

The mention of Jews, i.e. an ethnic group, results in the utterance no longer being
a political harangue but an instance of hate speech. Moreover, for greater relevance
it relies on the audience processing it together with the assumption of generalized
Jews’ culpability for all evils and at the same time strengthens the assumption
of their culpability.10
The processing here is quite straightforward, though I would argue that to
make the implicature that Jews living in Poland are an internal enemy of the state
a strong one, invoking a previous mental representation of their being just that,
is required. The presence of Wyzwolimy ‘we shall free’ and the presupposition it
carries alone does not give raise to these strong implicatures, but to much weaker
ones. The understanding procedure involved is similar to that proposed by Wilson
(1994, p. 40) for her example:
(16) a. Does Viv play cricket well?
b. He plays for the West Indies.

10. This generalized assumption is best presented and made ridicule of in an old joke: A: (com-
menting on the allegedly sorry state of Poland ‘All of this is the fault of Jews and cyclists’. B: ‘Of
cyclists? Why cyclists?’. This example also shows that only quite specific assumptions are being
recovered as implicated premises: (7) does not rely on other negative assumptions about Jews,
e.g. that they are care excessively about money, to mention just one, possibly the least offensive.
 Chapter 8. Towards a relevance-theoretic account of hate speech 243

which gives rise to a strong implicature that Viv is a good cricketer only “on the as-
sumption that anyone who plays for the West Indies is a good cricketer” (Wilson,
1994, p. 40).
In the processing sketched above the mutual-adjustment inferential mecha-
nism accounts for the relative strength of premises and implicatures (Wilson &
Sperber, 2004). Strong implicatures are necessarily activated when the utterance
is processed. By the same token processing them involves the least cognitive ef-
fort and they inevitably lead to the greater cognitive effect. If they are held by
the addressee to be true (i.e. they have been entertained by them before) they
are thus confirmed. If not, they need to be assumed as representations held by
the speaker, because otherwise they would not be (optimally) relevant in the
relevance-theoretic sense.
Indeed, (4) is quite complex when compared to (7a)–(c), in which the assump-
tions about blacks/Asians/Hispanics intellectual and/or social inferiority as a part
of what is decoded from the final part of the utterance. In that instance the strong
implicature is limited to:
(17) If you were 100% black/Asian/Hispanic you would not have been able to say
these smart/articulate things that you said.

On the other end of the scale of complexity there is (2), where in order to process
it, let alone understand it, the audience needs to possess, first of all, a specific
mental representation of the Holocaust in order to process the to the gas element
in Jews to the gas, including Nazis being German speakers, otherwise the echoic
use of language falls flat here. It is also an interesting example of interpretative use
of language, since it is not embedded in an utterance communicating the actual
speaker’s attitude towards the interpretatively used utterance. Here the endorse-
ment is tacit, and communicated by the very use of German-sounding expression.
Secondly, the audience needs to have a mental representation of the political situ-
ation in the Middle East, and specifically of Hamas and its anti-Israeli militancy.
Moreover, they need a mental representation allowing them to draw connections
between Israel and Jews.
Again, all the premises necessary to process (2) and the implicatures it gives
rise to (some of them have been very briefly presented in the previous paragraph)
rely on the deployment of mutual adjustment inferential mechanism. Since
processing the German-sounding phrase already involves considerable effort to
arrive at the representation of Jews being send to gas ovens, and it is necessarily
activated when the utterance is processed (otherwise the rant makes no sense), it
strengthens the explicature to the degree that it cannot be overlooked when the
utterance is examined in judiciary setting to assess if it constituted incitement to
hatred and possibly even to violence against real Jews.
244 Jadwiga Linde-Usiekniewicz

This example also provides an illustration of two other rules offered by Wilson
(1994), namely “The first acceptable interpretation is the only acceptable interpre-
tation” (Wilson, 1994, p. 49) and “Extra effort implies extra effect” (Wilson, 1994,
p. 52). Namely: additional cognitive effort was required to understand the rant
as targeted against the opposing team (in the form of a leap from Jews as such to
‘Jews’ as in Widzew team), but it provided the required or intended extra effect, on
top of the first acceptable interpretation.11
In addition, the analysis presented so far shows that in order to process hate
speech utterances one does not need to entertain these assumptions oneself: they
may constitute mental representations or mental meta-representations of other
people’s beliefs (Wilson & Sperber, 2004).

6. The ‘thin skin’ phenomenon vs. cognitive vigilance. Understanding of

hate speech by targeted audiences

One would expect that the targeted groups would tend to be more prone to
identifying utterances, and particularly those derogating them, as representing
hate speech. This expectation may lead to deeming their opinions oversensitive
(Nijałkowski, 2008). Though not couched in those terms, the comparison between
what the targeted groups consider hate speech and what is considered hate speech
by general public, presented in Bilewicz et al. (2014), seems to support that claim.
Yet, other studies, i.e. Leets & Giles (1997); Leets & Giles (1999); and Leets, Giles,
& Noels (1999) yielded “robust counterintuitive findings that European Americans
perceive direct racist slurs as more harmful to the recipient than do the actual
targeted ethnic minority group members” (Leets, 2001, p. 677). The difference be-
tween Polish and American findings may be due to profound differences between
the Polish and American studies: different minorities, different methodology,
different study design, different subject groups, different study goals and different
stimulus material used. Leets (2001) talks about slurs, but (7a)–(b) that should be
considered covert hate speech is actually part of her material; yet all the other ex-
amples contain offensive nouns. Bilewicz et al. (2014) used actual quotes deemed
hate speech by expert judges. The sample examples they present contain overt
and covert hate speech. Yet it is also possible that the differences noted by Leech

11. The Prosecutor who initially declined to prosecute failed to take into account a common
sense equivalent of this rule. The Appellate Prosecutor and the two Courts apparently con-
formed to this rule of analysis, which resulted in the eventual trial, and the guilty verdict being
upheld by the Appellate Court.
 Chapter 8. Towards a relevance-theoretic account of hate speech 245

(2001) are related to the ways adopted by European Americans and the members
of targeted groups to preserve positive social identity (the black sheep effects):
European Americans were attempting to save face by dissociating from the
blatant prejudicial statement of a deviant in-group member. By condemning the
derogatory racial slur, they were renouncing the message (or at least avoiding the
appearance of being prejudiced) and trying to preserve a positive social identity.
 (Leets 2001, p. 699)

Bilewicz et al.’s findings about the correlation between being in favor of hate speech
ban and authoritarian views and specifically their relating this correlation to the
existence of a norm that forbids such utterances (Bilewicz et al. 2014, p. 108) may
be seen as meshing with Leets’s findings.
Yet, the fact that covert hate speech affects the targeted group, together with
the claim that alleged hate speech needs to be processed together with specific
assumptions about the groups’ denigrating characteristics (intellectual or social
inferiority, threat to the society, etc.) raises yet another question: If most these
assumptions would most likely be deemed false by members of targeted groups,
why when they are faced with hate speech, they still invest processing effort in
something that they know is blatantly untrue? The positive cognitive effect (Wilson
& Sperber, 2004, p. 608) is nil.
As already mentioned, initially (Linde-Usiekniewicz, 2014) I thought that the
difference between the way the intended audience of hate speech, i.e. those who
share the views of the hate-speaker and are likely to accept all the premises as true,
and the targeted audience process hate speech is linked to emotions. I assumed that
both types of audience follow the path of least effort in computing cognitive effects
as applied to disambiguation and reference resolutions (since these are necessary
in order to comprehend the utterance and possibly reject it). At this stage there
may be little divergence between the two groups in the order in which the assump-
tions are accessed. According to Sperber et al. (2010, p. 362) “our mental systems
start by automatically accepting communicated information, before examining it
and possibly rejecting it.” Yet, at the stage of arriving at a satisfying interpretation
there is an important difference between the computing paths followed by the two
types of audiences. For the targeted group the interpretive hypothesis of offensive
contents is formed early enough because it corresponds to the assumptions about
offensive character of the utterance, which are for them contextually most avail-
able. Once these assumptions are identified and processed, the expectations of
relevance are satisfied and the processing stops here: the specific contents of the
offence is not worth further processing effort. This would have also accounted
for the apparent thin-skin phenomenon, or the alleged oversensitivity of targeted
audiences to hate speech.
246 Jadwiga Linde-Usiekniewicz

In other words, I believed that a Jew, faced with (6), which really involves a huge
processing effort even to be decoded, would identify it as a piece of hate speech
and refuse to process it further to arrive at the possibly weak implicature missed by
Nijałkowski (2008). (He had rightly recognized the strong implicature that Gebert
is stupid, but missed the weak one that there is no need to pay attention to what
he says as a columnist, writer and activist). Yet targeted individuals do process and
understand such utterances. Would that mean that they approach them in a two
distinct steps: identifying it as hate speech without apprehending their meaning
in its complexity, and later overcoming their indignation to process it further?
This solution is at odds with basic ideas of relevance theory, which is based on the
balance between cognitive effort invested and (positive) cognitive effect.
Actually, what I would posit is that the difference between the way the two
types of audiences interpret covert hate speech lies elsewhere. The intended audi-
ence of indirect hate speech share the assumptions necessary for the utterance
to be optimally relevant. In the case of hate speech these assumptions may often
have “no clearly identified source to be accepted and transmitted purely on the
ground that it is widely accepted and transmitted” (Sperber et al. 2010, p. 380) As
mentioned, they would correspond to actually held beliefs, mostly reflective, due
to their overgeneralized character.
When the targeted audience is faced by covert hate speech, the moment it is
recognized as such, for the purpose of understanding it (both in terms of decod-
ing and inferring) they substitute the assumptions necessary for the utterance to
be optimally relevant by their meta-representations (Wilson & Sperber, 2004) or
by assumptions that other people hold these assumptions. (And obviously they
would be well aware of the fact that people do hold such beliefs.) In addition, the
same analysis applies to audiences who, although not members of the targeted
groups, find those denigrating assumptions reprehensible. That is why in the case
of (1) the general public, or at least its more liberal wing, found the Prosecutors’
initial decision absurd. Moreover, when the targeted groups or those sympathetic
to them, or at least opposed to hate speech as such, react more strongly to in-
stances of covert hate speech, they do it because they are aware that hate speech
utterances, by replicating these assumptions, give them circulation. Importantly,
in some cases they are aware of the fact that the views they correspond to have
already led to violence or even genocide.

7. Slurs and quasi-slurs

The overlap between slurs, and specifically slurs targeting people on the basis of
their ethnicity, religion, gender, gender identification or sexual orientation, and
 Chapter 8. Towards a relevance-theoretic account of hate speech 247

hate speech is obvious and intuitive. Yet there is no one-to-one mapping between
the two concepts, in spite of both phenomena targeting the same groups. I have
amply argued above that what I called covert hate speech need not rely on slurs.
On the other hand, whether all instances of slur use would qualify as hate speech
remains an open question.
Recent research about slurs, e.g. Bianchi (2014), Blakemore (2014), Croom
(2014), and the literature cited therein, take into account the fact that slurs tend
to be appropriated, i.e. used within the targeted group as neutral terms or even
instruments of positive self-identification. In some cases the community allows
non-members to use these words in their non-derogatory sense (Bianchi 2014)
The word queer, initially a slur targeting a male homosexual, has made it into the
academia, with the launching of Queer Studies at many prestigious universities.
The non-derogatory use of slurs has given renewed force to the debate whether
the derogatory effect is a part of their semantics or depends on pragmatic ele-
ments, such as context of use. Both Blakemore (2014) and Croom (2014) convinc-
ingly show that slurs cannot be analyzed as purely expressive linguistic items; they
agree that they contain a descriptive or conceptual content that ordinarily enables
the recognition of their referents, i.e. the targeted group (with some exceptions
discussed and rightly discarded). Blakemore additionally enters into a debate with
Hom (2008) and proposes an account according to which the referential or de-
scriptive element of slurs is a part of their conceptual meaning, while “the offence
caused by the use of a slur is not derived from its semantically encoded content. It
is derived from the meta-linguistic knowledge that the word is an offensive means
of predicating and referring” (Blakemore, 2014, p. 13).
As already mentioned, Blakemore (2014, p. 12) is right when she argues that
the derogatory contents Hom associates with the word chink, i.e.:
ought to be subject to higher college admission standards, and ought to be subject
to exclusion from advancement to managerial positions, and, because of being
slanty-eyed, and devious, and good at laundering, and .. all because of being
Chinese. (Hom, 2008, p. 431)

cannot constitute its semantics. But she fails to discuss that his proposal follows a
dismissal of trying to define the meaning of chink along the lines of “Chinese and
despicable because of it” (Hom, 2008, p. 416). Hom argues that the “despicable
because of this” component would not be appropriate for the word nigger, as it
“simply fails to explain the strength of this word’s negative, derogatory content
relative to others in comparison (e.g. “chink”).” (Hom, 2008). Yet he seems to be
248 Jadwiga Linde-Usiekniewicz

confusing the linguistic meaning of a slur with the history of its use and societal
attitudes and violence meted to those referred to thus.12
As I have argued elsewhere (Linde-Usiekniewicz, 2012, p. 64), a word can en-
code (i.e. have it as a part of its lexical (standard, conventional) meaning) not only
the descriptive characteristics of its referent, but also valorizations and attitudes.
Not all of these negative attitudes need to be offensive: the negative valorization
is borderline offensive in brat, since it refers to human being, and not offensive at
all in jalopy. The problem with the analysis of slurs is that due to the high degree
of offensiveness they carry and their taboo character, their early analysis has been
patterned on the analysis of other taboo words, such as cuss-words, the represen-
tational meaning of which has been overlooked.13
The claim about the semantic character of valorization and/or attitudinal
component made in (Linde-Usiekniewicz, 2012) did not take into account the dif-
ference between conceptual meaning and procedural meaning (Wilson, 2011a).
In relevance-theoretic literature the procedural character of the expressive or
emotional component is recognized (Blakemore, 2014 and the literature quoted
therein.) However, Blakemore links the negative valorization associated with slurs
with pragmatic adjustments. She writes:
[T]‍he hearer’s metalinguistic knowledge about the word hori will lead to the as-
sumption that the relevance of an utterance which contains it lies in the negative
attitude it communicates. In order to satisfy this expectation of relevance, the
hearer must pragmatically adjust the concept of MAORI PERSON so that the
concept taken to be communicated is one which gives access to a range of negative
culturally stereotypical assumptions which are attributed to the user of the slur.
 (Blakemore, 2014, p. 13).

Yet I think it is arguable that the procedural meaning of group-targeted slurs

should be seen somehow differently: as the instruction to process, possibly for
optimal relevance, the reference (to some minority group or an individual) against
some mental representation about the group. Moreover, it can be argued that in the
case of standard, i.e. derogatory use of slurs “the linguistic system [emphasis mine]
is linked to the rest of the cognitive system in such a way that activating the word
will systematically activate the associated concept or procedure and vice versa”
(Wilson, 2011a, p. 9). By contrast, the pragmatic adjustment is necessary when

12. This is not to say that I disagree in principle with Critical Race Theorists quoted by Hom.
What I am arguing is that in some branches of linguistics it is valid to separate the meaning of
words from the atrocities performed by those who use them.

13. See Croom (2014) for arguments against purely expressive contents of slurs, including non-
racial ones.
 Chapter 8. Towards a relevance-theoretic account of hate speech 249

a slur is used in a non-derogatory way, be it appropriated or not (Hom, 2008).

Thus the ban Blakemore refers to can be metalinguistic, but geared to the speakers’
benevolence and competence, as defined by Sperber et al. (2010). Speakers don’t
say kike/hori/chink/wetback when they mean ‘Jew/Maori/Chinese American/
Hispanic’ respectively in the same way that they would avoid using a conjunction
they deem inappropriate. The difference between slurs and words typically encod-
ing procedural meaning can be seen as consisting (1) in the reprehensible char-
acter of the assumptions invoked by slurs, and (2) in the fact that the procedural
instructions encoded in slurs do not uniformly point to the same assumptions if
there are several derogatory representations involved. The procedural instruction
thus further specifies that the representation accessed should be the most relevant
given the context. It is immaterial that some of these assumptions can be also
accessed if a non-derogatory term is used. Just the opposite, it is in the latter case
when pragmatic adjustments of concepts are necessary. The contrast between B’s
response in (18a) and their response in (18b), taken from Blakemore (2014):
(18) a. A: George just sits around on the veranda all day.
B: What d’you expect? He’s a hori.
b. A: George just sits around on the veranda all day.
B: What d’you expect? He’s a Maori.

would in my analysis, rely on there being a contextually created concept of

MAORI* in (18b).14
Nevertheless, most likely not all the instances of the use of slurs would qualify
as hate speech, be it direct or indirect. Blakemore (2014, p. 4) gives two more ex-
amples of the use of the word hori, which is an offensive term for a Maori, quoted
here as (19) and (20) respectively
(19) A: Karen married a hori.
B: Well, I don’t care if she married a Martian. She is still coming to my
party.
(20) Sue believes that that hori in charge of accounts should be fired.

Since I have been arguing that hate speech operates upon available assumptions
about the targeted group’s inferiority that are culture specific, I cannot say if they
would be judged as instances of hate speech by New Zealanders or by English
native speakers in general. However, when I substitute a Polish truly offensive term
for blacks/Orientals in (19) the resulting utterances are offensive and derogatory,

14. My proposal calls for alternative analysis of both appropriated use of slurs and their non-
derogatory, non-appropriated use. As the issue is somewhat tangential here, for the lack of space
I will not discuss it further.
250 Jadwiga Linde-Usiekniewicz

but would not constitute instances of hate speech. This may be due to the fact
that I am not aware of any negative assumptions concerning representatives of
these groups as being bad spouse material. By contrast, (20) could be hate speech
if among the prejudices against Maoris/blacks/Orientals there were one of their
being dishonest. Thus substituting Polish language racial slurs for blacks/Arabs/
Orientals does not make a hate speech utterance of it, but substituting even a
non-offensive Polish word for Jew might, though with rather weak implicatures of
general untrustworthiness around money.
Yet it is the overlap between slurs and cuss-words that gives rise to some
instances of racist issues being involved in an abuse, even without a racial slur
featuring in the utterance. Blakemore (2014, p. 2) writes:
the use of a slur is not necessary for the attribution of offence. On the one hand,
offence may be attributed to a speaker who uses an otherwise acceptable term in
the context of an utterance containing an expletive, pejorative epithet or prosodic
device which has pejorative effects.

and she illustrates it with an example of a footballer allegedly saying to another:

(21) You fucking black cunt.

Blakemore comments:
Since then the footballer has been tried and found not guilty of racially abus-
ing another player by a Magistrates’ Court on the grounds that although he did
not deny producing the words in ([21]]), he did not intend racial abuse. At the
time of writing, however, he has been found guilty of racial abuse by the Football
Association the grounds that he did utter the words […]. The point here is that the
utterance is offensive because the non pejorative word black is used in an insult
where the intention to insult is communicated by the use of expletives fucking and
cunt in an utterance of the form You Pred P. (Blakemore, 2014, p.2)

The similarity to (1) is striking, though I would argue that the insulting effect is not
conditional on the use of You Pred form. Saying about a black person, and maybe
particularly about a male black person:
(22) This fucking black cunt.

would be equally offensive.

The instances of (1), (20) and (21) constitute what I have called ‘quasi-slurs’:
words referring to race, religion, gender, gender orientation and sexual orientation,
which are not offensive in themselves, but are nevertheless used in direct insults and
in hate speech utterances. In (1) a neutral word meaning ‘Jewish’ is linked with a
noun ścierwo that in its first sense means ‘carrion’ and is a neutral term. Dictionaries
of Polish also give two more senses: (1) informal term for a piece of meat that has
 Chapter 8. Towards a relevance-theoretic account of hate speech 251

gone bad or is otherwise unappetizing; (2) informal derogatory terms for a person
of no moral worth. As already mentioned, the justification for not prosecuting
given was that calling someone a Jew, irrespectively of their being Jewish or not is
not an insult and that the word ścierwo can be used playfully. However, what the
alleged expert failed to recognize is that the playful usage of the word is restricted to
its sense ‘piece of meat’. Yet, even though żydowskie ścierwo can be in theory under-
stood as referring to a piece of meat, presumably kosher, the utterance is necessarily
disambiguated here as referring to human beings. And just as calling someone black
cunt is offensive, and a racially offensive term for a black person, calling somebody
Jewish scum is offensive, and racially offensive, even if the person is a non-Jew. Just
as for (18b) the concept of MAORI* was postulated, a pragmatically and contextu-
ally derived JEW* and BLACK* can be proposed for (1) and (20), (21) respectively.

8. The hate component

As has been argued above, utterances containing mere representations of some

groups’ inferiority or despicability, retrievable as a part of the explicature, as in
(7a)–(b) and (8), though reprehensible, cannot be regarded as instances of hate
speech. Neither can those utterances where the assumptions corresponding to
these representation are retrieved as an implicature, as in (2) and (4). Similarly,
some instances of utterances featuring slurs in their standard, i.e. derogatory use,
though offensive and equally reprehensible, do not constitute hate speech. In
those instances, while the derogatory component is present, the hate component
is somehow absent.
On the other hand, non-derogatory words, such as black or Jewish combined
with highly expressive offensive words (1, 21, 22), thus producing the starred
concept, are clear examples of hate speech.
This leads to conclusion that the hate component needs to be a part of what is
decoded or inferred from an utterance, at least partly independently of the nega-
tive representations invoked by the use of slurs or by otherwise communicated
assumptions.
The very term ‘hate speech’ and the notion of ‘hate component’ suggest, since
hate is normally understood as an emotion, that the ‘hate component’ should be
described as related to emotions. Yet whatever the nature of the ‘hate component’
in ‘hate speech’, it cannot be understood as if it were just a means “[to] encode
procedures for the identification of the speaker’s attitude or emotional state”
(Blakemore, 2014, p. 4). Such understanding would give unwarranted weight to
‘protected speech’ arguments and, more importantly, would be contrary to psy-
chological findings discussed in the previous sections.
252 Jadwiga Linde-Usiekniewicz

What is proposed here instead is to try to trace the hate component to the
notion of attack, present in non-legal definitions of hate speech, as quoted above.
This attack actually takes the form of an implicature, in some instances weak, and
in some instances strong, along the lines of ‘something ought to be done about the
targeted individual or the targeted group’. The range of possible actions envisaged
might be decodable or inferable from the utterance and might range from relatively
mild, e.g. fining the person who stood on fallen heroes’ monument to deliver her
speech at a pro-LGBT event (cf. (8)) to less mild, ranging from “subject[ing them]
to higher college admission standards, and […] subject[ing them] to exclusion
from advancement to managerial positions,” to use the quote from Hom 2008,
through making the targeted group disappear from the public view, as in (9a)–(b),
or from the country, as in (4), to actual extermination, as in (2) – but not because of
what the individual did (stood on the monument, cheated somebody in business)
but because they allegedly belong to the group for which the negative representa-
tion is otherwise an implicated premise, explicature or another implicature.

9. Instances of non-literalness of hate speech

Since I have argued that hate speech phenomena are perfectly analyzable within
the relevance-theoretic framework, it should not come as a surprise that non-
literal phenomena can be observed in some instances of hate speech, though they
have received marginal attention in my analyses presented so far.
First of all hate speech can recruit echoic use of language, as clearly seen in
(2). Secondly, many racial slurs are metaphoric or metonymic, although their
figurative character may be etymological and not pragmatic. A telling example is
provided by the use of the word expression żydowskie ścierwo‘’ (in 1), though the
derogatory sense in reference to human beings is attested in dictionaries and need
not to be established as an ad hoc concept. Two of the non-verbal instances of hate
speech, i.e. defacing the mosque with drawing of pigs and throwing a banana at
Dani Alvez are clear examples of metonymy. Since they are easily analyzable as
such within relevance theory (Eizaga Rebollar, 2015) no special attention has been
paid to their metonymic character in this paper. The material presented here con-
tained no instance of metaphors, yet shocking examples can be found in studies
dealing with incitement to genocide. These consist – among others – of calling the
potential victims cockroaches, vermin, germ or pests (Benesch, 2008). Specifically,
the incitement to Tutsi genocide in Rwanda contained such phrase as calling to cut
 Chapter 8. Towards a relevance-theoretic account of hate speech 253

down the tall trees (and the reference of the concept of TALL TREES* to Tutsi was
mediated by their being taller than Hutus).15
Moreover, skilled hate-speakers tend to couch their speech in such a way that
it does not infringe the specific laws of their respective countries (Mason, 2007,
p. 34), or at least they can mount a successful defense claiming that they were
speaking literally and not metaphorically, or vice versa, or they were speaking in
jest, or they were not in fact referring to the protected minority (e.g. legal issues in
persecuting the football supporters in (2)) – whatever suits their defense best. In
some cases the case need not to end in court (cf. Linde-Usiekniewicz, 2015b for
some of the instances)

10. Conclusions

It has been argued in this chapter that identifying and understanding hate speech
follows the same comprehension procedures that are present in understanding
ordinary speech. Two specific components of hate speech have been identified,
or rather re-formulated within relevance-theoretic perspective. One is the as-
sumptions about the targeted group’s inferiority. It was claimed that in covert hate
speech these assumptions may appear either as implicated premises, implicated
conclusions or as part of the explicature. The same assumptions are connected to
slurs, though contrarily to Blakemore (2014) they are presented here as something
that the procedural meaning of slurs instructs the audience to access. It has been
also argued that non-derogatory words, when combined with derogatory words,
pragmatically create derogatory concepts, similar to semantically derogatory terms.
Two distinctions of hate speech types were introduced: direct hate speech, i.e.
the one for which the targeted group is the intended audience, and indirect, for
which the intended audience may already entertain assumptions on the targeted
group’s inferiority, or may be likely to accept them. It was argued that despite
the fact that the targeted group rejects premises on which hate speech relies,
they still arrive at all or most of the intended offensive implicatures. This is so
because the background assumptions about inferiority are metarepresented in the
comprehension process.
The other component, i.e. the hate component of hate speech was identified
as, again, explicated or inferable reference to some action targeted against the tar-
geted group or the targeted individual. For an utterance to qualify as hate speech,
either direct or indirect, it is necessary for the audience to be able to retrieve both

15. The expression is amply documented in publications dealing with Rwanda genocide. It can
be easily analyzed within the framework proposed by Wilson (2011b).
254 Jadwiga Linde-Usiekniewicz

components: the one that has been referred to as the assumption about inferiority
and the hate component referring to the assumption about the need of an ac-
tion. In addition, in accordance to the principle of relevance, larger positive (in
relevance-theoretic terms) cognitive effect is obtained when the two assumptions
are processed together, in addition, mutual-adjustment inferential mechanism
assures that they are thus processed. The relative violence of hate speech utter-
ances depends on several factors combined together: the contents of the assump-
tions themselves, the violence of the action, and the cognitive effort necessary to
obtain the cognitive effect. The cognitive effect produced by direct and indirect
hate speech actually tends to be polar. The effect produced in a witness of hate
speech (Croom, 2014) would vary accordingly to whether they would side with
the targeted group or not.
Thus Hom’s (2008) possibly true claim about the nigger being “the most nox-
ious racial epithet in the contemporary American lexicon” is linked to the strength
of racial prejudices in the US and the effect it has had in terms of violence. The
burning cross in RAV is violent hate speech because it relies on the representa-
tions of KKK practices. For similar reasons, it is not the noxious character of
anti-Semitic assumptions themselves that gives rise to the perceived violence of
anti-Semitic hate speech, but representations about Holocaust. Łaziński (2014)
equates the force of anti-black hate speech in the US with anti-Semitic hate speech
in what he calls ‘the West’.
The difference between (2) and (4) clearly illustrates the impact of the action
alluded to or presented in hate speech. In (4) no violence is mentioned, while
in (2) the echoic use of German and the descriptive mental representation (i.e.
Jews being burned in gas ovens) not only make a direct reference to Holocaust,
but also to physical extermination, on a par with the burning cross. The weak-
ening effect of the amount of cognitive effort involved is illustrated by apparent
innocuousness of (6).
This is not to say that the weaker variants of hate speech are less reprehensible
than the stronger variants; they may only be more difficult to adjudicate as such by
different types of courts and court-like bodies.

References

Austin, J. L. (1962). How to do things with words. Oxford: Oxford University Press.
Benesch, S. (2008). Vile crime or inalienable right. Defining incitement to genocide. Virginia
Journal of. International Law, 48/3, 485–428.
Benesch, S. (2012a). Susan Benesch interview with Timothy Garton Ash. http://‍voicesthatpoison.
org/2012/04/20/free-speech-debate.
 Chapter 8. Towards a relevance-theoretic account of hate speech 255

Benesch, S. (2012b). Dangerous speech: A proposal to prevent group violence. http://‍www.

worldpolicy.org/sites/default/files/Dangerous%20Speech%20Guidelines%20Benesch%20
January%202012.pdf
Bianchi, C. (2014). Slurs and appropriation: An echoic account. Journal of Pragmatics, 66, 35–44. ​
https://‍doi.org/10.1016/j.pragma.2014.02.009
Bilewicz, M., Marchlewska, M., Soral, W. & Wiśniewski M. (2014). Mowa nienawiści. Raport z
badań sondażowych [Hate speech. A survey report] Warsaw: Fundacja im. Stefana Batorego.
Blakemore, D. (2014). Slurs and expletives: a case against a general account of expressive mean-
ing. Language Sciences, 41, 1–14.  ​https://‍doi.org/10.1016/j.langsci.2014.06.018
Brown, A. (2017a). What is hate speech? Part 1: The myth of hate. Law and Philosophy, 36 (4),
419–468.  ​https://‍doi.org/10.1007/s10982_017-9297-1
Brown, A. (2017b). What is hate speech? Part 2: Family resemblances. Law and Philosophy. 36
(5), 561–613.  ​https://‍doi.org/10.1007/s10982_017-9300-1x
Butler, J. (1997). Excitable speech. A politics of the performative. New York: Routledge.
Croom, A. M. (2014). The semantics of slurs: A refutation of pure Expressivism. Language Sci-
ences, 41, 227–242.  ​https://‍doi.org/10.1016/j.langsci.2013.07.003
Czarnecki, G. (Ed.). (2009a). Raport o homofobicznej mowie nienawiści w Polsce [The report on
homophobic hate speech in Poland], Warsaw: Kampania przeciw Homofobii.
Czarnecki, G. (2009b). Chronologia homofobicznych wydarzeń w roku 2008 [The chronology
of homophobic incidents in 2008] . In G. Czarnecki (Ed.), Raport o homofobicznej mowie
nienawiści w Polsce (pp. 17–24). Warsaw: Kampania przeciw Homofobii.
van Dijk, T. (1987). Communicating racism: Ethnic prejudice in thought and talk. Thousand
Oaks, CA: Sage Publications.
Eizaga Rebollar, B. (2015). A relevance-theoretic perspective on metonymy. Procedia–Social and
Behavioral Sciences, 173, 191–198.  ​https://‍doi.org/10.1016/j.sbspro.2015.02.052
Hom, Ch. (2008). The semantics of racial epithets. Journal of Philosophy, 105 (8), 416–440. ​
https://‍doi.org/10.5840/jphil2008105834
Jakobsen, R. (2010). The interpretation of indirect speech acts in relevance theory. Journal of
Linguistic Intercultural Education, 3, 7–23.
Kowalski, S. & Tulli, M. (2003). Zamiast procesu. Raport o mowie nienawiści [Instead of a trial. A
report on hate speech]. Warsaw: WAB.
Kwok I. & Wong, Y. (2013). Locate the hate: detecting tweets against Blacks. In M. desJardins,
& M. Littman (Eds.), Proceedings of the Twenty-Seventh AAAI Conference on Artificial
Intelligence (pp. 1621–1622). Bellevue, Washington: Association for the Advancement of
Artificial Intelligence.
Leets, L. (2001). Explaining perceptions of racist speech. Communication Research, 28 (5),
676–706.  ​https://‍doi.org/10.1177/009365001028005005
Leets, L. (2002). Experiencing hate speech: Perceptions and responses to anti-Semitism and
antigay speech. Journal of Social Issues, 58 (2), 341–361. ​
https://‍doi.org/10.1111/1540-4560.00264
Leets, L. & Giles, H. (1997). Words as weapons: When do they wound? Investigations of harmful
speech. Human Communication Research, 24, 260–301. ​
https://‍doi.org/10.1111/j.1468-2958.1997.tb00415.x
Leets, L. & and Giles, H. (1999). Harmful speech in intergroup encounters: An organizational
framework for communication research. In M. Roloff (Ed.), Communication Yearbook vol.
22 (pp. 91–137). Thousand Oaks, CA: Sage.
256 Jadwiga Linde-Usiekniewicz

Leets, L., Giles H. & Noels K. (1999). Attributing harm to racist speech. Journal of Multilingual
and Multicultural Development, 20, 209–215.  ​https://‍doi.org/10.1080/01434639908666377
Linde-Usiekniewicz, J. (2012). From conflict through compromise to collaboration: Semantics,
syntax and information structure in natural languages. Warsaw: Nakładem Wydziału Po-
lonistyki.
Linde-Usiekniewicz, J. (2014). Towards a relevance-theory account of hate speech. Keynote
lecture presented at Interpreting for Relevance 2014, Warsaw, September 29–30.
Linde-Usiekniewicz, J. (2015a). Teoria relewancji jako narzędzie opisu „mowy nienawiści”
[Relevance theory as a tool for describing ‘hate speech’] Studia pragmalingwistyczne, VII,
53–68.
Linde-Usiekniewicz, J. (2015b). Wybrane problemy rozpoznawania językowych cech „nawoływania
do nienawiści” [Some problems in recognitions of linguistic features of “incitement to hatred”],
ms. http://‍www.zglosnienawisc.otwarta.org/incydent/244.
Łaziński, M. (2014). Jeszcze o słowie Murzyn i o stereotypach. Po lekturze artykułu Margaret
Ohii Mechanizmy dyskryminacji rasowej w systemie języka polskiego [More on the word
Murzyn and stereotypes. Upon reading Margaret Ohias’ Mechanisms of racial discrimina-
tion in the system of Polish language]. Przegląd Humanistyczny, 5, 127–141.
Mason, G. (2007). The reconstruction of hate language. In K. Gelber & A. S. Ackary Stone (Eds.),
Hate speech and freedom of speech in Australia (pp. 34–81). Leichhardt: Federation Press.
Mateo, J. & Yus F. (2000). Insults: A relevance-theoretic taxonomical approach to their transla-
tion. International Journal of Translation, 12 (1), 97–130.
Mateo, J. & Yus F. (2013). Towards an intercultural pragmatic taxonomy of insults. Journal of
Language Aggression and Conflict, 1 (1), 87–114.  ​https://‍doi.org/10.1075/jlac.1.1.05mat
Mazzarella, D. (2013). ‘Optimal relevance’ as a pragmatic criterion: The role of epistemic vigi-
lance. UCL Working Papers in Linguistics, 35, 20–45.
Nicolle, S. (2000). Communicated and non-communicated acts in relevance theory. Pragmatics,
10 (2), 33–245.  ​https://‍doi.org/10.1075/prag.10.2.04nic
Nijałkowski, L. M. (2008). Mowa nienawiści w świetle teorii dyskursu [Hate speech in the
light of the theory of discourse]. In A. Horolets (Ed.), Analiza dyskursu w socjologii i dla
socjologii [Discourse analysis in and for sociology] (pp. 113–133). Toruń: Wydawnictwo
Adam Marszałek.
van Noorloss, M. (2011). Hate speech revisited. A comparative and historical perspective on
hate speech law in the Netherlands and England & Wales. Cambridge UK – Portland (Or.):
Intersentia.
Piskorska, A. (2012). Cognition and emotions – a joint effort at obtaining positive cognitive
effects?. In A. Piskorska (Ed.), Relevance studies in Poland. Volume 4. Essays on language
and communication (pp. 102–111). Warsaw: Wydawnictwa Uniwersytetu Warszawskiego.
Random House Dictionary. (2014). Dictionary.com Unabridged. http://‍dictionary.reference.
com/browse/hate speech.
SN. (2007). Postanowienie Sądu Najwyższego–Izba Karna z dnia 5 lutego 2007 r. IV KK 406/06.
Solska, A. (2012). Appealing and appalling: Puns and degrees of relevance. Paper delivered at
Interpreting for Relevance: Discourse and Translation. Warsaw, September 25–26, 2012.
Sperber, D. (1997). Intuitive and reflective beliefs. Mind & Language, 12 (1), 67–83. ​
https://‍doi.org/10.1111/1468-0017.00036
Sperber, D., Clément, F., Heintz, Ch., Mascaro, O., Mercier, H., Origgi, G. & Wilson, D. (2010).
Epistemic Vigilance. Mind & Language, 24 (4), 359–393. ​
https://‍doi.org/10.1111/j.1468-0017.2010.01394.x
 Chapter 8. Towards a relevance-theoretic account of hate speech 257

Sperber, D. & Wilson, D. (1986). Relevance. Communication and cognition. Oxford: Oxford
University Press.
Warner, W. & Hirschberg, J. (2012). Detecting hate speech on the WorldWideWeb. In A. Farz-
indar & D. Inkpen (Eds.) Proceedings of the workshop on semantic analysis in social media
(pp. 19–26). Avignon: Association for Computational Linguistics. http://‍www.aclweb.org/
anthology/W12-06
Wilson, D. (1994). Relevance and understanding. In G. Brown, K. Malmkjaer, A.Pollitt & J. Wil-
liams (Eds.), Language and understanding (pp. 35–58). Oxford: Oxford University Press.
Wilson, D. (2006). The pragmatics of verbal irony: Echo or pretense?. Lingua, 116, 1722–1743. ​
https://‍doi.org/10.1016/j.lingua.2006.05.001
Wilson, D. (2011a). The conceptual-procedural distinction: Past, present and future. In V.
Escandell-Vidal, M. Leonetti & A. Ahern (Eds.), Procedural meaning: problems and perspec-
tives (pp. 3–32). Bingley: Emerald Group Publishing. ​
https://‍doi.org/10.1108/S1472-7870(2011)‍0000025005
Wilson, D. (2011b). Parallels and differences in the treatment of metaphor in relevance theory
and cognitive linguistics. Studia Linguistica Universitatis Iagellonicae Cracoviensis, 128,
195–2013.  ​https://‍doi.org/10.2478/v10148-011-0025-1
Wilson, D. & Sperber, D. (2004). Relevance theory. In L. Horn, & G. Ward (Eds.), Handbook of
pragmatics (pp. 607–632). Oxford: Blackwell.
Zegarac, V. (2012). U-Pragmatics and E-Pragmatics: Making a case for I-Pragmatics. Paper deliv-
ered at Interpreting for Relevance: Discourse and Translation. Warsaw, September 25–26,
2012.
 Chapter 9

Tropes of ill repute

Puns and (often thwarted) expectations of relevance

Agnieszka Solska
University of Silesia, Poland

Puns, a form of figurative language exploiting ambiguity in its many guises, tend
to be regarded as a trope of lesser value. The present article explores this tainted
reputation of puns in the light of the relevance-theoretic model of utterance
interpretation and attributes it to the fact that like no other figure of speech
they often thwart our expectations of relevance. Moreover, many of them do
so boldly, conspicuously, thus making us particularly likely to take note of the
emotional reactions we experience and to articulate them. The article identifies
six ways in which puns may go against our expectations about verbal inputs and
attempts to specify the possible negative consequences this may have for our
perception of puns.

Keywords: puns, wordplay, ambiguity, tropes, figurative language, paronomasia,

utterance interpretation, pun perception

1. Introduction

In discussions of metaphorical language the title of Ortony’s (1975) article is

sometimes evoked. Promising to spell out “Why metaphors are necessary and not
just nice,” it sums up the views on metaphor which, after decades, if not centuries,
of extensive research, now prevail in the scholarly world. If the commonly held
attitudes towards puns were to be expressed in a similar rubric, its more charitable
version might be “Why puns are not necessary, though they can be nice,” but
its more likely variant would probably be “Why puns are neither necessary nor
particularly nice.” Though puns are widely employed in works of literature and in
diverse everyday situations, though they have their apologists, though they are an
object of study within the realms of literary stylistics, theory of literature, as well
as linguistics, the fact remains that to many language users, especially speakers of
English, they are somewhat suspect. These figures of style, exploiting an ambiguity

https://‍doi.org/10.1075/ftl.8.09sol
© 2020 John Benjamins Publishing Company
260 Agnieszka Solska

between identically or similar-sounding words for a humorous or rhetorical ef-

fect, are often perceived as inappropriate, as something to be ashamed of. This is
reflected in set expressions used in English to acknowledge that a pun has been
made. While no speaker ever apologizes for speaking metaphorically or for resort-
ing to irony, phrases such as Pardon the pun or (often insincere) assurances No pun
intended or are routinely uttered when a pun has been produced.
The widespread perception of punning as a verbal art of low standing has re-
ceived only scant attention in linguistic literature, whose relatively recent interest
in puns has revolved mainly around classifying puns, settling terminological is-
sues and investigating the mechanisms underlying the punning effect. Such issues
seem solidly grounded within the purview of linguistic inquiry unlike the issue of
the tainted reputation of puns, which seems bound up with stylistic preferences or
cultural norms. In this article all of these domains will be touched upon, yet the
main impetus will be directed at exploring the odium puns seem to carry in the
light of the relevance-theoretic model of utterance interpretation. A close look will
thus be taken at puns as ostensive verbal stimuli creating in the comprehender an
expectation that the effort he has exerted while processing a pun will be rewarded
by sufficient cognitive benefits to justify that effort. A number of puns found in
different communicative settings will be examined in an attempt to demonstrate
that the kind of contextual effects offered by specific punning utterances are not
uniform and may often be disappointing to some hearers. I will argue that the
problematic status of puns in the minds of many languages users is to a large ex-
tent the consequence of the fact that like no other stylistic device puns often thwart
the audience’s expectations of optimal relevance. Moreover, many of them do so
boldly, conspicuously and this reflects on the way all puns are perceived.
The discussion is organized as follows. Section 2 takes note of the ambivalent
status of puns among literary devices. Section 3 documents the many ways in
which puns exploit ambiguity and Section 4 links the distrust of puns, prevalent
in contemporary world, to the attitudes to ambiguity that emerged in Europe as
a consequence of the historical and philosophical changes taking place during
the Age of Reason. Section 5 explores the notion that puns are deviations from
linguistic norms. Section 6 presents the relevant-theoretic view on puns and iden-
tifies six specific ways in which punning utterances may thwart the expectations
language users have about verbal communication. Section 7 attempts to specify
the possible negative consequences each of the departures from the expected may
have on our perception of puns.
 Chapter 9. Tropes of ill repute: puns 261

2. Puns as ‘wayward’ stylistic devices

Long before puns drew the attention of linguists, they had come under the scrutiny
of masters of rhetoric and literature, who identified a whole family of rhetorical
figures which are either pun-based or which have punning versions. Its members,
none of whose names is a household term, include paronomasia, syllepsis, antanac-
lasis, polyptoton and asteismus.1 What sets puns off from other figures of speech is
the particularly strong and emotional reactions they tend to elicit in both expert
and casual readers of literary works as well as in authors and literature scholars.
These emotions, which fluctuate between appreciation and scorn, have been
recorded in many comments on puns, such as Dryden’s often quoted quip dismiss-
ing puns as “the lowest and most groveling form of wit” (1672, p. 223; reprinted
in Craig, 2005, p. 295) or Boswell’s more charitable opinion conceding that “a
good pun may be admitted among the smaller excellencies of lively conversation”
(Boswell, 1791, entry for June 19, 1784; reprinted in Chapman, 1980, p. 1309).
Considering how prevalent the derisive views on punning are, it is not surprising
that in his Linguistic Guide to English Poetry (1969) Leech offered no arguments
defending the use of metaphors, alliterations or other rhetorical devices, but he
thought it necessary to write an entire section “In Defence of the Pun” (Leech,
1969, 212–214).
It can be argued that the ambivalent feelings inspired by puns in practitioners
of literature and literature scholars are a reflection of their unclear status as literary
devices. In the title of this article puns are referred to as tropes because this is how
they are treated in the British rhetorical tradition. The tradition goes back to mid
16th century, when Richard Sherry (1550) divided all figures of speech into tropes
and schemes. He defined the former as involving using words, phrases, or images
in a way “not intended by their normal signification,” and the latter as representing
“a change in standard word order or pattern.” Puns are counted among tropes
because they play on double (sometimes triple) meanings of words. The problem
is that numerous examples can be found of puns that refuse to be neatly slotted
into one of the two rhetorical categories.
Multiple meanings are definitely at play in Examples (1)–(3), taken from works
of British and American literature spanning more than 300 years. In (1), which is
an example of antanaclasis, two senses of the key expression dissemble are brought
into play, as on its first appearance the verb indicates putting on a disguise, and

1. Paronomasia is the most general term used in rhetoric to refer to punning wordplay. In
antanaclasis a word is repeated in a different sense. Polyptoton is a figure of speech that involves
repetition of words derived from the same root while asteismus is a mocking or humorous reply
that plays on a word. The two latter figures do not have to be pun-based but they often are.
262 Agnieszka Solska

on the second, acting hypocritically. In Dryden’s description of the welkin (i.e.

the vault of heavens), given in (2), the pivotal word pitched simultaneously carries
two unrelated senses: ‘covered with pitch, made black’ and ‘erected, set up (like a
tent)’, both used metaphorically. The sylleptic pun in (3) contrasts the literal and
the metaphorical sense of the predicate covered themselves: the former applying to
the noun dust, the latter to the noun glory.
(1) I will dissemble myself in’t; and I would I were the first that ever dissembled
in such a gown. (Shakespeare, Twelfth Night)
(2) but soon he found
The welkin pitched with sullen clouds around
 (Dryden, “On the Death of Amyntas. A Pastoral elegy”)
(3) they […] covered themselves with dust and glory.
 (Mark Twain, The Adventures of Tom Sawyer)

There can be no doubt that the ingenious puns quoted above are tropes.2 However,
in other, equally ingenious puns the dividing line before tropes and schemes is
blurred. Consider the triple pun in (4), involving the word still, which charac-
terizes the eponymous urn in Keats’s “Ode to a Grecian Urn.” The pun is rather
easy to miss since none of the three senses it carries, i.e. (1) ‘as of now’ (yet), (2)
‘unmoving’, and (3) ‘forever’, can be regarded as going beyond its ‘normal significa-
tion’. There is also nothing out of the ordinary about the way the word, in any of
its senses, fits into the structure of the sentence in which it appears. In brief, it is
neither trope-like nor scheme-like.
(4) Thou still unravish’d bride of quietness,
Thou foster-child of silence and slow time,
 (Keats, “Ode to a Grecian Urn”)

On the other hand, the often quoted Shakespearean pun in (5) in which Mercutio
makes an allusion to his imminent death, fits the description of both a trope and
a scheme. What makes it trope-like is the extra sense of ‘pertaining to the grave’
or ‘laid in a grave’, with which Mercucio manages to imbue the adjective grave.
What makes it scheme-like is the fact that the extra meaning emerges from the
unexpected use of the noun grave in the pre-modification of another noun.
(5) Ask for me tomorrow, and you shall find me a grave man.
 (Shakespeare Romeo and Juliet Act 3 Scene 1)

2. Arguably, antanaclasis is somewhat scheme-like since repeating a fragment of a message does

break the most commonly found pattern of speech.
 Chapter 9. Tropes of ill repute: puns 263

3. Vocalizations that cash in on ambiguity

In “False Wit: Punning,” a cavilling pamphlet against puns, published in The

Spectator on May 10, 1711, Joseph Addison proclaimed that “[i]‍t may be said of a
pun, […] that it is vox et praeterea nihil; a sound and nothing but a sound (second
highlighting mine).” As was demonstrated in Section 2, his criticism misses the
point. Puns are much more than sonic stylistic devices, as their effect is tied up
with both sound and meaning.
What makes this key property of puns possible is ambiguity, a pervasive feature
of natural languages, embodied in such phenomena as polysemy, homonymy, ho-
mophony, paraphony (i.e. imperfect homonymy) and paronymy (the use of words
which are derived from the common root or which sound similar).3 In punning
utterances, a potentially ambiguous linguistic form (referred to in literature as the
punning element, the pivot or the connector), appears in a context which forces
the comprehender to activate two or more of its meanings or which forces him to
consider the meaning of another similar sounding word or phrase. Examples of
the resulting pairings of meanings4 are presented below, first in a vertical pun, i.e.
one in whose structure the connector appears once, then in a horizontal pun, i.e.
one in whose structure the connector is repeated.

homonymy:
(6) Being in politics is just like playing golf: you are trapped in one bad lie after
another. [untruth/position of a ball]
(7) She was always game for any game. [‘having the required spirit’/’pastime’]

polysemy:
(8) When it rains it pours. 
 (advertising slogan for Morton salt) [‘flows freely’/’rains heavily’]
(9) We must, indeed, all hang together or, most assuredly, we shall all hang
separately.  (Benjamin Franklin) [‘be united’/’be punished’]

3. Homography is sometimes added to this list, though it can be argued that homographic puns,
such as (i) and (ii), can be treated as cases imperfect homophony.

i. Poland Polishes Off U.S. Volleyball Team. [‘pɒlɪʃɪz/’pǝʋlɪʃɪz]

ii. Polish up your Polish apples. (sign on a fruit stand in a Polish shop in Britain) [‘pɒlɪʃ/’pǝʋlɪʃ]

4. I discussed these parings of meanings in Solska (2012a, pp. 172–173; 2012b, pp. 387–388).
264 Agnieszka Solska

homophony:
(10) Everybody kneads it.  (Ad for Pillsbury flour) [‘kneads’/’needs’]
(11) Literacy Hour gives pupils the right to write.

imperfect homophony:
(12) Merchant of Tennis.  (name of a sports shop) [‘Tennis’/’Venice’]
(13) The ballot is better than a bullet.5

paronymy:
(14) I used to be a doctor but then I lost patience.  [‘patience’/’patients’]
(15) Some men are wise and some are otherwise.

Note that it is mainly vertical puns that seem to get bad press. Investigation con-
ducted by Solska and Rojczyk (2015) into the appeal pun-based city promotional
slogans have for the audience provides some backing for this observation. One of
the findings of this empirical study was that slogans containing vertical puns were
rated as less pleasing than the ones containing horizontal puns. In fact, with the
exception of one slogan, all of them fell on the ‘not-pleasing’ side of the attitude
scale, unlike horizontal puns, whose ratings placed them firmly on the ‘pleasing’
side of the scale. In what follows, the discussion will revolve mainly around the less
favourably perceived vertical puns.

4. Puns as victims of the western flight from ambiguity

The hostile or at least ambivalent attitudes towards puns are a relatively new
phenomenon and are quite different from the sentiments which predominated in
the pre-modern Western world. There is ample evidence for the fascination an-
cient Greeks and Romans held for ambiguity and ambiguity-based wordplay. The
equivocal pronouncements of the Delphic oracle often made use of lexical and/or
structural ambiguity. Philosophers such as Aristotle and Plato did not shy away
from using puns in their writings and neither did the most acclaimed poets of the
ancient world, including Homer and Ovid. Rather than discourage orators from
using puns in their speeches, Quintillian in his Institutio Oratoria offered them

5. Not all scholars would regard this particular kind of utterance as a pun.
 Chapter 9. Tropes of ill repute: puns 265

advice on how to use them well. The famed orator Cicero (in Guthrie, 1840: 174)
noted in De Oratore that “[e]‍quivocal sayings are esteemed as being of the wittiest
kind.” Medieval times brought forth an abundance of erotic literature in which,
as noted by McDonald (2006), punning double entendre was quite common. In
the times of Renaissance, the Italian courtier and diplomat Castiglione devoted
an entire section of his Book of the Courtier to the topic of how to pun in good
taste. The honor of being the most prolific punster in Elizabethan England goes to
William Shakespeare, who is estimated to have used over 3000 of them in his plays
and sonnets (Adamczyk, 2006, p. 302). Shakespeare’s proclivity for punning was
to garner a fair share of criticism in later centuries (cf. Mahood, 1968). However,
in the Bard’s times the practice of punning was so widespread, that, as observed by
Mathews (1988, p.134), “[c]‍lergymen punned in the pulpit, judges upon the bench,
statesmen at the council-board, and even criminals in their dying speeches.”
As noted by Pollack (2011), the turning point in the perceptions of puns in
the West, especially in English-speaking countries was the arrival of the Age of
Reason, whose leading intellectual figures began to see all instances of equivocal
language as a problem, and set off to “to sweep away all the fogginess of words, to
reduce the language to its simplest terms, to make it as accurate, concrete and clear
an image of the material world as was possible” (Jones, 1932, p. 319–320, quoted
in Pollack, 2011, p. 76). Philosophers such as Descartes, Leibnitz, Locke or Hume
eloquently expressed their scorn for ambiguous forms of language. In his Essay
Concerning Human Understanding, Locke insisted that
those who pretend seriously to search after or maintain truth, should think
themselves obliged to study, how they might deliver themselves without obscurity,
doubtfulness, or equivocation, to which men’s words are naturally liable, if care be
not taken. (Locke [1690] in Niddittch 1975, p. 504)

This “flight from ambiguity,” which is how Levine summed it up in the title of his
1985 book, affected the way ambiguity-based wordplay was perceived. As reported
by Alderson (1996), numerous satirical pamphlets against puns appeared, such as
“God’s Revenge Against Punning,” in which an anonymous author drew parallels
between punning, the Black Death and the Great Fire of London. Defenders of
puns ran to their rescue: in 1716 Swift (also anonymously) published his Modest
Defence of Punning (reprinted in Davis, 1939), and in 1719 Sheridan’s Ars Punnica
(reprinted in 2010) appeared in print, extolling the virtues of quips and laying out
the rules for successful punning. However the damage was done. The distrust of
puns continued throughout the Age of Reason, with negative attitudes solidifying
by the end of the 18th century. Though the practice of punning has never been
abandoned, in the popular mind puns and ambiguity have become stigmatized. In
mid 19th century Holmes (1858, p. 17) described the pun as “primâ facie an insult
266 Agnieszka Solska

to the person you are speaking to” and as late as mid 20th century, Kaplan (1955,
p. 39) characterized ambiguity as “the common cold of the pathology of language.”
However important the historical and philosophical factors might be for shap-
ing the present-day attitudes towards puns, I hope to show that there more fun-
damental reasons for the ambivalent attitudes towards punning, reasons hinted at
in Chovanec’s (2005, p. 62) observation that “punning works counter to the basic
principles of effective communication.”

5. Puns as deviations from linguistic and communicative norms

In linguistics no systematic research has yet been conducted to explore the stigma
attaching to puns. Pun-related issues explored in some detail include attempts at
explaining how the punning effect arises (Yus, 2003, 2016; Giora, 2003; Solska,
2012a, 2012b, 2012c, 2017), establishing pun taxonomies (Tanaka, 1992, 1994; Yus,
2003, 2016; Dynel, 2010), investigating their connection with humour (Attardo,
1994; Yus 2003, 2016), as well as analysing their role in specific communicative
settings, such as advertising slogans (Tanaka, 1992, 1994), newspaper headlines
(McArthur, 1992; Goddard, 1998; Dor, 2003; Chovanec, 2005), and their role in
interpersonal discourse (Norrick, 1984; Solska, 2013). Empirical studies on pun
appreciation and their perceived humorousness, conducted by van Mulken et. al.
(2005) as well as Solska and Rojczyk (2015; 2016) come closest to touching upon
the issues that are under discussion in the present article.
A common theme in many linguistic accounts of puns, albeit one that tends
to be mentioned, rather than expounded in any way, is that as utterances they
behave in a way that is extraneous to the default functioning of language. Puns
are often described as anomalies, as deviations from linguistic or communicative
norms or, at best, as mistakes. The reservations many language users have for puns
would then represent the distrust people often feel for anything that diverts from
a perceived standard.

5.1 Puns as mistakes or anomalies

The idea that the punning effect arises by mistake can be found in Norrick’s ob-
servation (Norrick, 1984, p. 202) that “[p]‍uns can be represented in propositional
networks as parallel propositions mistakenly sharing an apparently common term”
(emphasis mine). Norrick does not make it clear at whose door the mistake he
speaks of should be laid. Treating them as a mistake on the part of the addressee
could apply only to situations where another strand of meaning is identified,
though none was intended. This is the fate that befell the promotional slogan for
 Chapter 9. Tropes of ill repute: puns 267

the city of Hong Kong, which became an unintentional pun, when the outbreak of
SARS epidemic in 2002 heightened the salience of its literal meaning.
(16) Hong Kong Will Take Your Breath Away.

Treating puns as a mistake on the part of the speaker would rule out the possibility
of the intentional production of puns. This obviously cannot be the case, as many
puns are deliberately produced and well thought out. There’s no reason to suspect
that producers of conversational witticisms, such as (6) and (17) were ever dis-
satisfied with what they said or with how they said it. Likewise, there is no reason
to assume that punning book titles, such as (18), or pun-based titles of scholarly
articles, such as (19) appear in print without first undergoing careful consideration.
(17) Atheism is a non-prophet institution. 
 (George Carlin) [‘non-prophet’/’non-profit’]
(18) Sound Foundations. 
 (title of Adrian Underhill’s book on learning and teaching pronunciation)
(19) The case for case.  (title of a research article by Charles Fillmore)

Some puns, that is homonymy- and polysemy-based vertical puns, certainly go

against the expectation that one linguistic form should communicate one concep-
tual meaning, the fact that will be explored in Section 6. That is why zeugmatic
puns, such as (3) or (20), are regarded as a semantic anomaly. In puns of this sort
one word or expression is understood differently in relation to two (or more) other
words, which it modifies or governs. For example, in (20), lay oscillates between
two senses: the literal sense ‘put away’, which applies to hats and the metonymi-
cally extended sense ‘have sex with’, applicable to friends.
(20) All I need is room to lay a hat and a few friends.  (Dorothy Parker)

5.2 Puns as violations of gricean maxims of conversation

As for the communicative principles puns fail to follow, the ones usually cited are
Grice’s (1975) maxims of conversation. For instance, Chovanec (2005) identifies
two Gricean maxims which are supposedly violated by pun-based newspaper
headlines such as (21) and (22). These and other headlines which Chovanec dis-
cusses in his article, all of them reporting on soccer matches between the Czech
Republic and other countries, exploit the homophony of the word Czech and other
identical sounding words or morphemes. Chovanec remarks that by evoking the
game of chess and the domain of banking and finance, which have little to do with
the issues reported, headlines (21) and (22) violate the maxim of relation. He also
268 Agnieszka Solska

interprets the fact that they are metaphorical as providing more information than
is necessary and thus violating the quantity maxim.
(21) Czechmate as Germany crash out. 
 [‘Czechmate’/‘checkmate’] (Chovanec, 2005, p. 63)
(The Telegraph. June 24, 2004, on a match between Germany and the Czech
Republic)
(22) Cashing in on the Czechs.  [‘Czechs’/‘checks’] (Chovanec, 2005, p. 64)
(The Guardian July 1, 2004, on betting on a semi-final match between Greece
and the Czech Republic)

It is easy enough to find puns that on one or both of their meanings flout specific
Gricean maxims or submaxims, often more than one at the same time. By forcing
the addressee to associate two distinct meanings with the same linguistic expres-
sion all puns seem to run afoul of the maxim of manner. However, many of them
adhere to rather than depart from the “Be brief ” submaxim: after all, expressing
more than one meaning via one form seems to be an ultimate manifestation of
brevity. Moreover, despite capitalizing on ambiguity, not all puns flout the “Avoid
ambiguity of expression” submaxim. In fact, only homonymy- and polysemy-
based vertical puns are ambiguous. As noted in Section 3, horizontal puns are
unambiguous since they explicitly convey only one propositional meaning. So
do homophony- and paraphony-based vertical puns, which force the audience to
consider another identically- or similar-sounding expression. However, of the two
meanings puns like these induce the addressee to entertain, only one is actually
conveyed via an explicitly expressed connector.
Can the fact that many puns violate Gricean maxims be the reason for the
strong and less than favourable emotions puns evoke? Can it also be the reason for
the apologies so often heard when a pun has been produced? This is doubtful. After
all, other tropes, i.e. ironies, metaphors or hyperboles, which in Grice’s account
arise as a result of departing from specific maxims, do not carry the stigma with
which puns are burdened and, as pointed out in the first paragraph of this chapter,
they do not generate apologies. However, the Gricean model of communication
with its notion of maxim flout does offer some insight into the bad press puns
tend to get. The insight springs from the idea that is central to Gricean pragmat-
ics, which is that a blatant failure to fulfill a maxim, i.e. a maxim flout, results in
generating an implicature. This notion mandates the prediction that an utterance
which flouts some maxim(s) yet fails to generate implicatures should be perceived
as pragmatically odd. This prediction captures pun-related facts quite well.
While many maxim-flouting punning utterances do generate implied mean-
ings, there are many that fail to trigger implied meanings or trigger none that
 Chapter 9. Tropes of ill repute: puns 269

are easy to identify. This is particularly obvious in punning jokes, such as (23),
and riddles, such as (24) and (25), which flout various maxims for what might
be perceived as totally spurious reasons. First of all, they all flout the maxim of
manner: the joke in (23) incorporates the non-existent and nonsensical expres-
sion on porpoise, which the hearer has to process in order to identify another,
similar-sounding unexpressed expression on purpose. In the riddle in (24), the
pun-based answer is spelled in an unconventional way so as to enable the ad-
dressee to spot the connector hidden in a longer word. In (25), the verb planet,
which the riddle evokes, cannot be meaningfully incorporated into any kind of
proposition. Moreover, utterances (23) and (25) flout the maxim of truthfulness,
as on one or other of their readings nobody could take them for truthful represen-
tations of some state of affairs: having a whale of a time is something that can be
experienced, not something that any creature, including dolphins in the zoo, can
do on purpose. It is also very unlikely that punsters always come on time. As for
(25), it flouts the maxim of quantity, since announcing that to organize a party one
has to plan it amounts to stating the obvious.
(23) The dolphins in the zoo are having a whale of a time. Not accidentally but on
porpoise.  [‘porpoise’/’purpose’]
(24) Why are punsters always on time? Because they are pun-ctual.
(25) How do you throw a party in space? You plan it.  [‘plan it’/’planet’]

One might point out that utterances (23)–(25) belong in the domain of verbal
humour, and thus do not represent bona fide cooperative communication Gricean
model was supposed to account for. However, puns violating various maxims
without triggering implied meanings permeate all types of discourse. Consider
Examples (26)–(28). The punning blend in (26) comes from a memo sent out on
July 13, 2015 by Donald Tusk, president of the European Council of the European
Union, after marathon negotiations about the Greek crisis. Example (27) has been
chosen as a name for a gift shop selling Native American artifacts. Example (28)
comes from an online article describing squat toilets in China. The first two
utterances contain nonce formations and thus depart from the maxim of man-
ner. However, in (26) the sense of the blend aGreekment (perhaps we might
call it a puntmanteau) is rather transparent (‘agreement with or concerning the
Greeks’) and thus the meaning conveyed by the utterance is explicit, rather than
implicit in nature.6

6. Any implicit meanings the blend aGreekment, may generate are due to its visual rather than
linguistic qualities. Incorporating the word Greek into the innards of the word agreement, the
270 Agnieszka Solska

Similarly, it is hard to identify any implied meanings the eye-catching shop

sign in (27) many convey. As for the parenthetical fragment in Example (28),
which incorporates the punning connector flush, it flouts the maxim of quantity
since it says more than is necessary. After all, the reader has already learnt from the
preceding text that in some Chinese restrooms toilets are level with the ground.
The pun highlights this fact but, again, no obvious implicature can be identified.
(26) Someone can say that we have “aGreekment” (from a memo following
negotiations on the Greek crisis negotiations)  [‘Greek’/’agreement’]
(27) Sioux Venirs (sign on a Native American gift shop) 
 [‘Sioux Venirs’/’Souvenirs’]
(28) Squat toilets are not seen in most of the Western world, but in China, it’s
more likely than not that a person will walk into a restroom and find a toilet
that is level (or, pardon the pun, flush) with the floor. 
 [‘level with a surface’/’wash out (a toilet)’] (Christopher, 2017)

There is no denying that many puns arouse the feeling that the speaker is not play-
ing by the rules of the linguistic game, however ‘the game’ is defined. Doubtless it
is this perceived oddity of many punning utterances that underlies the often felt
impression that an apology is in order, or the need to react to a pun with a laugh
or a groan. In the next section I will attempt to reinterpret this oddity in the light
of the relevance theoretic theory as a failure on the part of the speaker to produce
an utterance that is optimally relevant to some hearers.

6. Relevance-theoretic model and the perception of puns

6.1 Reservations about the model’s adequacy for tackling the emotional
reactions to puns

Can Sperber and Wilson’s (1986/1995) relevance theory shed any light on why puns
are perceived as tropes of ill repute? At first glance, it appears to be ill-equipped
to account even for the very existence of verbalizations of this sort. The heuristic
it posits for processing utterances, dubbed the relevance-theoretic comprehension
procedure, seems to rule out situations when more than one meaning of a single
linguistic form ends up being used in the interpretation process. Its claim that
in interpreting verbal inputs the comprehender follows principles that cannot be

new formation is essentially a pictorial icon, symbolically representing the involvement of

Greece within the agreement.
 Chapter 9. Tropes of ill repute: puns 271

violated seems difficult to reconcile with the fact that the retention of ambiguity in
vertical puns and the resulting oscillating effect depart from the default outcome
of deriving meanings. Finally, as a cognitively-based model of utterance interpre-
tation, Relevance Theory seems to be the wrong kind of framework to choose for
exploring the emotional reactions puns evoke in language users.
Nonetheless, the analytic tools of relevance-theoretic framework have been
used to investigate a number of pun-related issues, both those connected with pun
processing and those that go beyond it. These include the extensive use of puns
in advertising slogans (Tanaka, 1992, 1994; Alrasheedi, 2014), their potential for
evoking humor (Yus, 2003, 2016), the role of epistemic vigilance in pun interpre-
tation (Padilla Cruz, 2015), and the factors determining pun appreciation (van
Mulken et al., 2005; Solska & Rojczyk, 2015). A relevance-theoretic account of
various aspects of pun processing including the ‘connecting’ role of the connector
and of the oscillating effect can be found in Solska (2012a), where an argument was
made that puns do not arise in violation of any communicative rules or maxims
but as a consequence of the hearer following the inviolable principles which drive
utterance comprehension. In what follows I hope to demonstrate that Sperber and
Wilson’s model can also prove useful for explaining at least some aspects of why
puns tend to stir up strong and predominantly negative feelings in language users.

6.2 A tainted concept: Hanging a dog that’s been given a bad name

In relevance-theoretic pragmatics the interpretation of all kinds of utterances,

including punning ones, is treated as following the same underlying principles,
which are grounded in human cognition and applied automatically whenever a
language user is confronted with a verbal input, and which cannot be violated.
These are that cognition is governed by the search for maximal relevance while
communication is driven by the search for optimal relevance. In other words,
when processing inputs to the cognitive systems we pay the most attention to
those which give us the greatest possible cognitive gain for the least mental ef-
fort. However, when confronted with inputs which qualify as ostensive signals, i.e.
inputs which have been deliberately produced (e.g., utterances and gestures), we
assume that they have been produced because their producer wants us to inform
us of something and we expect that the cognitive effects we will gain from pro-
cessing such inputs will justify the effort we have invested into processing them.
This expectation, dubbed the presumption of optimal relevance, can be fully
realized or it may be thwarted, if the balance of cost and gain turns out to be too
precarious. Among the utterances we produce, most of which are non-punning,
there is no shortage of those which fail to yield sufficient cognitive effects for a
specific addressee or which turn out to be too difficult to process. The resulting
272 Agnieszka Solska

misunderstandings, communication breakdowns as well as situations where we

end up giving up attempts to make sense of a verbal input are part and parcel of
everyday verbal exchanges, yet no other type of utterance gets singled out as being
notoriously recalcitrant.
Part of the reason why no specific stigma is likely to attach to those other
utterances is because there is nothing it can attach to. No convenient label can
possibly exist to identify the concept of ‘non-punning utterances that fail to be
optimally relevant’. Non-punning utterances assume a myriad of shapes and guises
and they perform countless discourse functions. Moreover, while they are some-
times problematic, most of the time they do meet our expectations of relevance.
As for puns, they are obviously also extremely heterogeneous. Pun classifications,
such as the one painstakingly devised by Dynel (2010), reveal how diverse struc-
turally, conceptually and functionally they are. However in the popular mind, at
least in the mind of an English speaker, they are all perceived as one distinct type
of utterance, subsumed under an all-embracing label ‘pun’.
In the relevance theory model content words, such as the word pun, are treated
as encoding mentally-represented concepts which provide access to the lexical,
logical and encyclopedic entries associated with them. In the case of the concept
PUN, it does not matter that its logical entry, specifying links to other concepts,
is somewhat fuzzy and imprecise. What matters is that for many language users
its encyclopedic entry is tainted with negative connotations, doubtless the legacy
of centuries of high profile pun-bashing by disgruntled men of letters. As shown
by the italicized fragments in Table 1, it is there, in the encyclopedic entry of the
concept PUN, that the stigma we are talking about is lurking. Puns are like the
proverbial dog that we hang because it has already been given a bad name.

Table 1. A possible (obviously incomplete) representation of the word pun in a speaker’s

mental lexicon
Conceptual Pun
address
linguistic entry: information about the lexical and grammatical properties of the word: N, V
information about the phonetic form of the word: /‍pʌn/
logical entry: inferential links to concepts: utterance, wordplay
encyclopedic the objects instantiating the concept: examples of specific punning
entry: utterances,
folk assumptions and cultural beliefs: puns are humorous, infantile,
frivolous, puns are ‘the lowest form of wit’, puns are sometimes inappropri-
ate
personal experiences of situations when the use of a pun was accompanied
by an apology, the production of a pun met with guffaws and/or groan
 Chapter 9. Tropes of ill repute: puns 273

The encyclopedic entry of a concept is specific to the individual and there are
doubtless many people whose beliefs about puns include positive assumptions
such as that puns are witty, that they are evidence of creativity and of high com-
mand of language, and whose experience of puns includes enjoyment of wordplay
or satisfaction at creating a particularly poignant quip. However even the greatest
pun lovers cannot help perceiving punning utterances through the lens of the
preconceived ideas associated with the concept PUN and cannot help but judge
such utterances accordingly. As we will see, the specific features of puns, to be
presented below, are only likely to reinforce these ideas.

6.3 Key properties of punning utterances

If the utterances we are likely to call puns are anything but homogenous, why
would we have just one label for them? Arguably this is because, diverse though
they are, they all have unique properties that set them off from other kinds of
utterances. Reformulating the three conditions on ‘puniness’, which I proposed in
Solska (2017), the utterances which meet these conditions and are therefore likely
to be perceived as puns, exhibit the following three characteristics:
1. They all contain a linguistic form (the connector) that can link two, and possi-
bly more, meanings via the phonetic part of the lexical entries of the concepts
which that linguistic form makes available.
2. The context against which the connector is processed contains at least two dis-
junctive elements, which bring to bear assumptions inducing the interpreter
to derive more than one meaning of the connector.
3. The two meanings the interpreter is induced to derive are clearly distinct from
each other, i.e. the concepts invoked may overlap, but they are not approxima-
tions of each other.

6.4 Language users’ expectations about verbal inputs and the ways they are
subverted in puns

Now, utterances which exhibit these characteristics make up only a small per-
centage of all utterances. They stand out because their properties make them go
against what we have come to expect from verbal communication based on our
experience with the verbal inputs we most commonly encounter, i.e. non-punning
ones. This includes
– Expectation 1, which is that in verbal inputs only one stable meaning is at play
at word and utterance level,
274 Agnieszka Solska

– Expectation 2, which is that people produce utterances because they want to

inform the audience of something, i.e. to convey some kind of message,
– Expectation 3, which is that what the utterance producers want to inform their
audience about may be of value to the latter,
– Expectation 4, which is that the addressee’s task is to identify the one meaning
of individual words, utterance fragments and whole utterances serving as the
vehicle for conveying that meaning,
– Expectation 5, which is that utterance producers provide clues guiding the
hearer towards that one meaning,
– Expectation 6, which is that the process of deriving meanings will not engage
the addressee’s conscious thought.
These expectations are not something language users are likely to be consciously
aware of. After all, most humans do not spend much (or any) time reflecting on
what is involved in utterance comprehension. Moreover, except for the convenient
label ‘pun’ (at least in English), they lack the vocabulary allowing them to dress
whatever observations they may have in words. Many punning utterances they
produce or encounter merely create in them a general impression that something
about these utterances “is not quite right,” the result being that the negative con-
notations which are already part of the concept PUN can only get confirmed
and strengthened.
Unlike ordinary users of language, the practitioners of linguistics have at their
disposal both the theoretical expertise and the terminology allowing them to
articulate pun-related observations. For relevance theorists the six language users’
expectations about verbal inputs are essentially expectations of their relevance.
The first three are connected with the cognitive effects they evoke (specifically,
the quality and the quantity of these effects), and the last three are connected
with the way the cognitive effects are derived, i.e. with the processing effort. The
analytical toolkit developed within the relevance-theoretic framework makes
it possible to reformulate the first five expectations in terms of such notions as
explicature, informative intention, positive cognitive effects, relevance-theoretic
comprehension procedure, context. The sixth one can be handled by the notion of
the depth of processing.
In relevance theory, explicature is defined as a communicated assumption
which is a “development of (a) a linguistically encoded logical form of the ut-
terance, or of (b) a sentential subpart of a logical form” (Carston, 2002, p. 124,
elaborating on Sperber and Wilson, 1995, p. 182). Note that all puns whose con-
nector fragments surface only once, i.e. all vertical puns, resist attempts to derive
only one explicature. This goes against expectation 1.
 Chapter 9. Tropes of ill repute: puns 275

The speaker’s informative intention, i.e. the speaker’s intention to induce a

belief in the audience, is seen in relevance theory as one of the two levels of inten-
tions involved in ostensive communication. The other one, the communicative
intention, is the intention to make the addressee aware of the speaker’s informative
intention. Many puns do not fail in their promise that they will provide evidence of
speaker intended meaning. If fact, they exceed this promise by providing evidence
of two clearly identifiable intentions. There are many, however, that at least on one
of their readings defy attempts at identifying the speaker’s informative intention.
In doing so they thwart expectation 2.
Relevance Theory applies the term positive cognitive effect to the result a
freshly encountered stimulus brings about by interacting with some of the as-
sumptions already in the cognitive system. Three types of positive cognitive effects
are distinguished: (1) those which confirm the assumptions already making up the
hearer’s knowledge, (2) those which contradict erroneous assumptions, and (3)
those which add new assumptions to the existing ones. Note that many (again not
all) puns fail to yield satisfactory cognitive effects on at least one of their readings,
thus coming short of expectation 3.
The so-called relevance-theoretic comprehension procedure is the procedure
posited for deriving meanings at word and utterance level. It involves following a
path of least effort in computing cognitive effects, testing interpretive hypotheses
in order of accessibility and stopping the derivation process as soon as we have
derived a good enough meaning or have given up the hope of doing so (cf. Wilson
& Sperber, 2004, p. 613). As I argued in Solska (2012a), all puns require the hearer
to reprocess a verbal string which has already yielded a meaning that cannot or
cannot instantly be rejected. In other words, they induce the hearer to apply the
procedure and then to reapply it. This is a clear departure from expectation 4.
In the relevance theory model, context is seen as the hearer’s cognitive envi-
ronment, that is the information available for processing an utterance, gleaned
from diverse extra- and intra-linguistic sources including assumptions stored un-
der the encyclopedic entries of the concepts made accessible by the material being
processed, the discourse surrounding that material, the physical setting in which
communication is taking place, and the “expectations about the future, scientific
hypotheses or religious beliefs, anecdotal memories, general cultural assumptions,
beliefs about the mental state of the speaker” (Sperber & Wilson, 1986/1995,
p. 15–16). Note that while in pun-free discourse the addressee is guided by the
contextual clues towards selecting only one interpretation, in puns the context has
(at least) two disjunctive elements leading him to establish (at least) two meanings
of the connector fragment. This subverts expectation 5.
The notion of the depth of processing does not appear in Sperber and Wilson’s
seminal work (1986/1995), which merely acknowledges the fact that in online
276 Agnieszka Solska

processing of utterances we decode and infer meanings on the fly, and are often
satisfied with good enough outcomes of the interpretation process. Obviously,
occasionally some of us do engage in reflective processing, where we keep trying
to ‘unpack’ layers of meaning we suspect can be lurking in the text that are not
instantly obvious to us. On encountering most puns we are forced to do the ‘un-
packing’ and to ponder, or at least to notice, the mechanics of verbal processing.
This goes against expectation 6.

7. Consequences of puns falling short of expectations of relevance

In any sphere of life departures from the expected may evoke emotional reactions
which can be negative, positive or ambivalent. I believe that this is precisely why
puns so often stir up all manner of strong responses. In what follows I will attempt
to identify the possible negative consequences each of the six departures from the
expected may have on our perception of puns.

7.1 The ‘Necker cube effect’: Unease at being unable to disambiguate

The language users’ expectation that it should be possible to home in on one

stable meaning of each utterance is unshaken by the fact that utterances may
and often do yield implicit meanings, both strongly and weakly communicated.
This expectation underlies the major difference between non-puns and puns: the
explicit univocality at word and utterance level, which characterizes the former,
and the explicit bivocality in the case of puns. As I demonstrated in Solska (2012a),
depending on the conceptual configuration of a specific utterance, the success in
detecting a pun is contingent on
1. in homonymy-based puns: identifying more than one stable sense of a lexi-
cally ambiguous word, e.g. the two meanings of lie in (6),
2. in polysemy-based puns: constructing two different modifications of one lexi-
cally given concept, e.g. the literal and metaphorical meaning of covered in (3),
3. in puns based on homophony and paraphony: identifying another, unex-
pressed expression identical or similar in sound to the connector fragment,
and thus bringing into play the target concept(s) it yields, e.g. kneads as the
trigger for ‘needs’ in (10), or Tennis as the trigger for ‘Venice’ in (12).
The process of interpreting a pun is thus effectively an exercise in ‘ambiguation’.
We end up with two meanings, or to be more precise, with two explicatures, each
incorporating a different key concept, thus creating in our mind what might be
 Chapter 9. Tropes of ill repute: puns 277

described a verbal equivalent of a bistable visual figure, such as the Necker cube7
seen in Figure 1, which alternates between two images or other similar perceptu-
ally ambiguous shapes.8

Figure 1. The Necker cube: Is the dot located at the front or at the rear?

The result is the most noticeable feature of vertical puns, sometimes described as
the oscillating effect. The hearer’s attention swings back and forth between two
interpretations, unable to let go of either. Curiously, this effect is present even in
single retention puns, whose less viable or downright invalid meaning lingers on
before being rejected. Some interpreters may relish the sensation of oscillating
between meanings. Others may find it unsettling, much as they may feel uneasy
when confronted with ambiguous visual stimuli, which make them realize that the
senses we depend on for survival can be in fact unreliable. The resulting negative
emotions may include:
– the frustration at being unable to home in on just one meaning,
– the unease at the language ‘not behaving as it should’,
– irritation at the punster who went out of his way to make this happen.
Note that the oscillating effect does not occur in puns in which the connector is re-
peated, i.e. in horizontal puns, whose comprehender ends up constructing a single
explicature for the entire utterance. The fact that these puns meet rather than de-
part from expectation 1 might explain the findings of Solska and Rojczyk’s (2015)
study, in which vertical puns were found to be less pleasing than horizontal ones.

7.2 Nothing but sound? Irritation at the punster’s sometimes dubious

informative intention

Punning utterances are no different from other ostensive signals in that they
too are treated as evidence of the communicative and the informative intention
of the speaker. Most puns do not fail to meet the expectation that an utterance
has been made to inform us of something, i.e. to bring to our attention a certain

7. So-called after the scholar who first described it (Necker, 1832).

8. For instance, the well-known old/young woman drawing (Boring, 1930).

278 Agnieszka Solska

assumption. A vertical pun with two equally valid meanings provides evidence for
two informative intentions. For instance, the producer of (29), a political slogan
which helped the British Conservative Party win the general elections in 1978,
intended to inform the audience both that Labour Party is ineffective and that the
British workforce is jobless. Puns with only one valid meaning seem to come with
only one informative intention. The writer of the newspaper headline in (30) must
have wished to inform the readers that investors were discouraged by the high
cost of an iceberg towing project and not that the prospect of spending a lot of
money made their body temperatures drop. Note that the literal meaning of chilled
in (30) is inadmissible but it is invoked for a good reason, to focus the hearer’s
attention on some aspect of the conceptual domain connected with the topic at
hand. After all big floating mountains of ice do have a chilling effect on whatever
they come close to.
(29) Labour Isn’t Working.
(30) [A] trial tow of a 6.5 million-ton iceberg would cost about $ 10 million – a
sum that chilled investors. (The Time July 31, 2011)

However, there is no shortage of puns whose informative intentions are difficult to

identify or which are questionable. These are exactly the puns which, in Addison’s
parlance, are “nothing but a sound.” Consider (31)–(33), captions on some of the
countless orange-you-glad memes one can come across on the Internet:
(31) Orange you glad you’re here?
(32) Orange you glad to see me?
(33) Orange you glad to be my Valentine?

In each of these the ungrammatical and downright absurd supra-lexemic connec-

tor orange you is inadmissible and thus serves as a prime sending the hearer in
search for a valid alternative. The legitimate Aren’t you target string gets evoked
because it shares part of the phonetic form with the prime. Indeed that seems to
be the point of these puns: not to ask if the audience is glad to see the producer
of the pun but to draw the hearer’s attention to the fact that orange you and aren’t
you sound similar. The producers of such puns have clearly gone to some trouble
to make sure this meta-linguistic and unrequested information is not lost on their
audience. Not only are the captions phrased in a peculiar way, but the salience
of the spurious meaning is raised by orange-colored backgrounds on which the
captions appear or by pictures of orange fruit.
Two facts about puns of this sort should not escape our notice. The first one
is that they clamor for our attention and cannot be ignored. As a result, they may
 Chapter 9. Tropes of ill repute: puns 279

induce us to form generalizations about all puns: that they are of questionable
communicative value. Obviously, though meaningless, these puns are not point-
less. As noted by Padilla Cruz (2005), utterances with low information content
tend to serve the phatic function. The speaker’s intention is thus not to commu-
nicate any assumptions, but to build or strengthen rapport with his audience. The
fact remains, however, that this is achieved by a vocalization whose information
value is nil. Second, they are the kind of utterances people, at least native speakers
of English, are most likely to encounter early on in their lives. Many de dicto jokes
(e.g. knock-knock jokes) and riddles exploiting the totally accidental homophony
or near homophony between two strings of sounds are part of many people’s
childhood memories. This may be the basis for yet another generalization about
puns: that they are something childish, something to be left behind as one grows
up, something inappropriate in ‘serious’ discourse.

7.3 Offering ‘fool’s gold’: Annoyance at the often problematic

cognitive effects

Commenting on what is gained by resorting to puns, Redfern (1984, p. 9) remarked

that they can be a “way of striking gold, albeit often of the fool’s variety.” If we
equate the metaphorical gold he speaks of with the cognitive effects to be gained,
some punning utterances are veritable goldmines. It is puns like these that perme-
ate poetry, turn some of our utterances into witticisms and create memorable book
titles. The fact remains however, that many puns offer cognitive effects which are
questionable and thus go against Expectation 3, which is that whatever a speaker
or writer wants to inform us of will actually have some informative value.
Relevance theory recognizes only positive cognitive effects but if anyone
insisted on identifying utterances that yield ‘negative’ ones, they would have no
trouble finding them among puns. There are spectacularly many which, instead
of adding to our store of knowledge, confirming, or contradicting what we know,
transgress in at least one of the following ways:
1. By yielding meanings which are too fragmented, effectively a handful of loose
concepts that cannot be integrated into any kind of logical form and thus com-
municate a specific belief.
2. By yielding meanings that amount to communicating the obvious,
3. By yielding meanings which are difficult to accept because they are unlikely to
be true or are blatantly false,
4. By yielding meanings which have no bearing on the matter at hand.
On their explicitly stated readings the humorous puns (31)–(33) are guilty of
transgression 1. The same can be said of (25) on its evoked reading: the sentence
280 Agnieszka Solska

You planet is meaningless. On its b-meanings (34) states the obvious and thus
commits transgression 2.
(34) When it pours it reigns. (advertising slogan for Michelin tyres)
a. When it pours with rain, Michelin reigns.
b. When it pours with rain, it rains.

As for the puns guilty of transgression 3, on its extraneous, b-meaning, advertising

slogan (10), repeated here as (35), is not true and (8), repeated here as (36), is
only sometimes true. Dynel (2010) distinguished a whole category of ‘zero mean-
ing’ puns. Her example, given in (37), yields two meanings serving as poise and
counterpoise for oscillation. However neither of them brings effects that could
be integrated to the body of assumptions we have about the world: there are no
undercurrents in bowls of muesli and currant fruits are incapable of performing
any kind of action.
(35) Everybody kneads it. (advertising slogan for Pillsbury flour)
a. Everybody needs Pillsbury flour.
b. Everybody kneads Pillsbury flour.
(36) When it rains it pours. (advertising slogan for Morton salt)
a. When it rains Morton, salt flows freely.
b. When it rains, it rains heavily.
(37) My friend drowned in a bowl of muesli: a strong currant pulled him in.
 (Dynel, 2010, p. 121).

True or not, the b-meanings of (34) and (36) commit transgression 4, as do many
paraphonic puns in names of businesses. For instance, the name of the carpet
cleaning firm in (38) puns on the name of a well-known American singer and
song-writer. The reference to Bruce Springsteen doubtless makes the business
name memorable and amusing yet identifying this reference seems to add no
information of relevance to the prospective customer who would like to have
his carpet cleaned.
(38) Spruce Springclean (name of a carpet cleaning firm)

Puns yielding meanings of questionable value tend to capture the attention of who-
ever encounters them and to stay in his memory. This is in fact what makes them
desirable in advertising slogans or shop signs. Some addressees doubtless enjoy
identifying the allusions lurking in such puns and appreciate the punster’s ingenu-
ity or even erudition. However, even the most appreciative addressees cannot help
being disappointed with the quality of the cognitive effects such puns have to offer.
 Chapter 9. Tropes of ill repute: puns 281

7.4 ‘Here we go again’: Resentment at being made to do the same

mental work twice

Perhaps the most striking way in which puns differ from non-punning utterances
is that they go counter expectation 4. As I argued in Solska (2012a), in an at-
tempt to obtain the optimal range of cognitive effects, the interpreter of a pun
processes its key fragment, then processes it again since the conceptual informa-
tion provided by the context gives him reason to believe that the meaning it has
yielded, though relevant, is not relevant enough. The repeated application of
the relevance theoretic comprehension procedure is particularly easy to note in
zeugmatic puns and in puns exploiting the garden path effect, such as (39), where
the idiomatic sense of the connector turns out to be inappropriate and has to be
replaced by the literal one.
(39) I made a killing on Wall Street a few years ago… I shot my broker.

However, inducing the interpreter to perform the same mental work more than
once is a characteristic feature of all vertical puns. The interpreter who has suc-
cessfully resolved a pun may find the pay-offs adequate and may end up feeling
pleased at solving a linguistic puzzle and appreciative of the punster’s wit. At the
same time he may feel a little disgruntled at having been made to exert twice as
much effort than is normally required. If the cognitive rewards are not sufficient,
he has every reason to feel frustrated and resentful.
The frustration and resentment can be particularly high if the interpreter has
problems resolving a pun and keeps mulling over possible interpretations in vain.
A participant of a study I once conducted confessed that for years he had been
baffled by utterance (36), the slogan inscribed on cans of Morton salt, suspecting
that a pun might be involved. He kept interpreting the word it as an expletive sub-
ject of the verb pour, unable to see that it might refer to the salt. It was not until he
read the two meanings of the slogan, spelled out in the study questionnaire, that he
was jerked free from the processing loop he had been struck in. This experience,
doubtless familiar to many language users, can only have confirmed rather than
dispelled the ingrained prejudice he is likely to have as a native speaker of English.

7.5 I seem to be missing something: The embarrassment at failing to

construct an adequate context

Attridge (1988, p. 141) describes puns as “product(s) of a context deliberately con-

structed to enforce an ambiguity.” In some punning utterances the two disjunctors,
i.e. ambiguity-enforcing elements of the interpreter’s cognitive environment, are
impossible to miss. For instance, one cannot make sense of a zeugmatic utterance
282 Agnieszka Solska

without realizing that its pivotal fragment must be understood in two different
ways, each applicable to one of its subject or object nouns. Many non-zeugmatic
punning utterances also contain equally obvious disjunctors, which act as two
forces pulling the comprehender in two opposite directions. For example, the pun
in (6) explicitly mentions politics and golf thus drawing the hearer’s attention to
the two disparate conceptual domains.
However, in many cases the resolution of a pun may depend on extralinguistic
factors. To detect the pun in the promotional slogan for the Canadian city of
Thunder Bay in (39) one has to know that it makes a reference to its location
on the banks of Lake Superior. The ‘flow freely’ meaning of pour in (36) is made
salient by the fact that the girl depicted on the cans of Morton salt, walking in the
rain under a big umbrella, is leaving behind her a trail of salt flowing freely from
an open spout of the container. This fragment of the picture must have escaped the
notice of the unsuccessful interpreter of (36) described in Section 7.4.
(40) Superior By Nature.

Evocative puns, such as (12), (37) or (38), are particularly likely to frustrate some
hearers since their comprehension depends on the hearer’s ability to retrieve a
similar sounding expression from memory. In puns of this sort, the connector
itself is one of the disjunctors, and is placed in a co-text which allows it to serve
as a prime pointing to the unexpressed target, to which it is phonetically similar.
Puns like these exploit fragments of idioms or set phrases, proverbs, titles of books
and films and can only be detected if the addressee has the requisite knowledge
to identify the target phrase. Success or failure in detecting such puns will depend
first of all on whether these are part of the hearer’s memory and second on how
quickly they can be retrieved.
What will happen if one of the disjuncting elements is not present, or at least
it is insufficiently prominent in the comprehendor’s mind? Two scenarios are pos-
sible. On the one hand, lacking the necessary resources, the hearer will stop after
deriving one conceptual meaning and the pun will go unnoticed. On the other
hand, and this is particularly pertinent to our discussion, the hearer might suspect
that a pun is involved yet be unable to resolve it and may end up feeling that ‘he
is missing some reference.’ The latter effect on the hearer is much more damaging
to his ego: possible consequences may include the feeling of annoyance at being
linguistically or culturally uninitiated, embarrassment or resentment at being
made to look stupid or uneducated. On the other hand, some evocative puns, al-
luding as they do to well-known proverbs, titles or sayings, may be seen as stale,
clichéd and uncreative, essentially rhetorical devices of lesser value. The resulting
impression may result in the contempt for the perceived unoriginality, laziness or
even insincerity of the creator of such forms, or of punsters in general.
 Chapter 9. Tropes of ill repute: puns 283

7.6 The indignation at being made to ponder the inner workings of verbal
processing

In non-punning discourse, the comprehension process tends to take place below

the level of conscious thought, with addressees decoding and inferring meanings
as they move along. In puns, the smooth running of verbal processing is disrupted.
The interpreter is forced to pause momentarily and revise the hypothesis he has
formed about the meaning of some fragment without entirely letting go of the first
one. Puns stop us in our tracks, give us a mental jolt, they are not only “a pistol
let off at the ear” (Lamb, 1838, p. 285) or a yapping dog “which jumps out of its
setting […] and bites the Master on the ankles” (Empson, 1953, p. i), they are
brain-teasers. Capitalizing as they do on both meaning and form, they draw the
comprehender’s the attention to the mechanics of language processing. In a way
they are like mini-lectures on linguistic issues, lectures which are unrequested and
titillating since limited to providing examples for issues that do not get spelled
out. It is this forced, uncalled-for awareness of verbal processes that makes puns
stand out and can produce a negative emotional reaction, for instance, a feeling
of unease at realizing that the connection between form and meaning is in fact
arbitrary but also displeasure at being made to do something the language user,
at least modern language user, is not normally interested in doing: pondering the
inner workings of language processing. As noted by Read (2007, p. 83):
The modern sensibility prefers the mechanics of a rhetorical effect to be hidden
from view; anything which smacks of contrivance or artifice, any construction
which leaves the scaffolding in place, is regarded with some suspicion […]. In
other words, the more obvious the pun to the reader (regardless of what feats of
ingenuity went into its fabrication), the less pleasure there is to be derived from it.

8. Concluding remarks

Marveling on the expressive powers of metaphor José Ortega y Gasset (1925, p. 35)
noted that it “is perhaps one of man’s most fruitful potentialities. Its efficacy verges
on magic.” By contrast, the lowly pun “remains an embarrassment to be excluded
from ‘serious’ discourse, a linguistic anomaly to be controlled by relegation to the
realms of the infantile, the jocular, the literary” (Attridge, 1988, p. 189). Puns stub-
bornly refuse to be confined to the three domains to which Attridge would have
them relegated. They keep popping up in witticisms, in advertisements, in political
slogans, in book titles and titles of scholarly articles. Their widespread use is the
best testament to their effectiveness and an indication that they too may offer a
range of “fruitful potentialities.”
284 Agnieszka Solska

Scholars investigating punning utterances have tended to focus on these

potentialities. In her (relevance-based) discussion of puns used in advertising
slogans, Tanaka (1992, 1994) identified a number of positive effects pun may
evoke, among them the pleasure and satisfaction the audience may feel at having
solved the pun, the enjoyment of humor many puns trigger, and the appreciation
of the addresser’s wittiness. She also noted the potential puns have for attracting
the audience’s attention, for increasing the memorability of the message, and for
offering a wealth of additional contextual effects. In this article I have focused on
the often mentioned but under-researched ‘dark side’ of puns. Having considered
stylistic, philosophical and linguistic reasons for their tainted reputation, especially
the ideas that follow from Gricean pragmatics, I attempted to use the relevance-
theoretic framework to demonstrate that puns particularly often and particularly
conspicuously thwart the audience’s expectations of relevance. I argued that it is
these subverted expectations that tend to generate negative responses, responses
that may end up being generalized and give all puns a bad name.
I am aware that I have not identified the full range of reasons for the ambiva-
lent and mostly negative reactions to puns. It would be worthwhile to contrast my
ideas with those of scholars working within different other analytic frameworks.
Though the volume of research puns have generated is quite impressive, to the best
of my knowledge, no attempts have yet been made to outline a systematic account
of their tainted reputation, grounded in specific models of utterance comprehen-
sion. I hope the present attempt to provide such an account based on the analytic
toolkit of relevance theory will inspire others to explore the explanatory potential
of other models. Gibbs and Colston (2012, p. 214) perceive “appreciation as the
next or final step in the interpretation of figurative language comprehension.” The
study of puns, including the study of how they affect their addressees, can reveal
much not only about the way we process verbal inputs but also about the predic-
tive powers of the theories with which we try to account for these processes.

References

Adamczyk, M. (2006). The formal composition of puns in Shakespeare’s Love’s Labour’s Lost: A
corpus-based study. Studia Anglica Posnaniensia, 42, 301–321.
Addison, J. ([1711]). False wit: Punning. The Spectator, 10 May 1711, no. 61. https://‍www.guten-
berg.org/files/12030/12030-h/‍SV1/Spectator1.html
Alderson, S. J. (1996). The Augustan attack on the pun. Eighteenth century life 20(3), 1–19.
Alrasheedi, E. S. (2014). The use of covert communication, irony and puns in print and online
English advertising: A relevance-theoretic account. International Journal of Humanities and
Social Science, 4(12), 70–79.
Attardo, S. (1994). Linguistic theories of humor. New York: Mouton.
 Chapter 9. Tropes of ill repute: puns 285

Attridge, D. (1988). Upacking the portmanteau; Or who’s afraid of Finnegan’s Wake? In Culler,
J. (Ed.), On puns. The foundation of letters. Oxford/New York: Basil Blackwell.
Boswell, J. ([1791]/‍1980). Life of Johnson. In R. W. Chapman (Ed.), James Boswell: Life of Johnson.
Unabridged. Oxford: Oxford University Press.
Boring, E. G. (1930). A new ambiguous figure. American Journal of Psychology, 42, pp. 444–445. ​
https://‍doi.org/10.2307/1415447
Carston, R. (2002). Thoughts and utterances: The pragmatics of explicit communication. Oxford:
Blackwell Publishing.  ​https://‍doi.org/10.1002/9780470754603
Chovanec, J. (2005). Czeching out puns and clichés in football reporting. In J. Chovanec (Ed.),
Theory and practice in English studies 3 (pp. 61–67). Brno: Masarykova Univerzita.
Christopher. (2017). How to use a squat toilet. China Highlights, DOA June 1, 2017. www.
chinahighlights.com/travelguide/article-squat-toilet-in-china.htm
Craig, D. H. (Ed.). (2005). Ben Johnson: A critical heritage. London and New York: Routledge
Dor, D. (2003). On newspaper headlines as relevance optimizers. Journal of Pragmatics, 35,
675–721.  ​https://‍doi.org/10.1016/S0378-2166(02)‍00134-0
Dryden, J. (1672/2005). Defence of the epilogue; or, an essay on the dramatic poetry of the last
age. In D. H. Craig (Ed.) Ben Johnson: A critical heritage. London and New York: Routledge.
Dynel, M. (2010). How do puns bear relevance. In M. Kisielewska-Krysiuk, A. Piskorska, &
E. Wałaszewska (Eds.), Relevance studies In Poland. Volume 3 (pp. 105–124). Warsaw:
University of Warsaw Press.
Empson, W. (1953). Seven types of ambiguity. New York: New Directions Publishing.
Gibbs, R. W. Jr., & Colston, H. L. (2012). Interpreting figurative meaning. New York: Cambridge
University Press.  ​https://‍doi.org/10.1017/CBO9781139168779
Giora, R. (2003). On our mind: Salience, context, and figurative language. Oxford: Oxford Uni-
versity Press.  ​https://‍doi.org/10.1093/acprof:oso/9780195136166.001.0001
Goddard, A. (1998). The language of advertising. London: Routledge.
Grice, H. P. (1975). Logic and conversation. In P. Cole & J. L. Morgan (Eds.), Syntax and seman-
tics. Volume 3: Speech acts (pp. 41–58). New York: Academic Press.
Guthrie, W. 1840. M. T. Cicero De oratore. Or, his three dialogues upon the character and quali-
fications of an orator. Translated into English with notes historical and explanatory, and an
introductory preface. Oxford: Henry Slatter. https://‍babel.hathitrust.org/cgi/pt?id=hvd.320
44018723890&view=1up&seq=5
Holmes, O. W. Sr. (1858/1955). The autocrat of the breakfast table. New York: Heritage Press.
Jones, R. F. (1932). Science and language in England of the mid-seventeenth century. The Journal
of English and Germanic Philology, 31, 315–331.
Kaplan, A. (1955). An experimental study of ambiguity and context. Mechanical Translation,
2(2), 39–46.
Lamb, C. (1838). The works of Charles Lamb to which are prefixed his letters. New York: Harper
and Brothers.
Leech, G. N. (1969). A linguistic guide to English poetry. London: Longman.
Levine, D. N. (1985). The flight from ambiguity: Essays in social and cultural theory. Chicago:
University of Chicago Press.  ​https://‍doi.org/10.7208/chicago/9780226056210.001.0001
Necker, L. A. (1832). Observations on some remarkable optical phenomena seen in Switzerland;
and on an optical phenomenon which occurs on viewing a figure of a crystal or geo-
metrical solid. The London and Edinburgh Philosophical Magazine and Journal of Science,
1(5), pp. 329 – 337
286 Agnieszka Solska

Nidditch, P. (Ed.). (1975). An essay concerning human understanding in four books: Clarendon
edition of the works of John Locke (p. 10). Oxford: Clarendon Press.
Mahood, M. M. (1968). Shakespeare’s wordplay. London: Routledge.
Mathews, W. (1988). Wit and humor: Their use and abuse. Chicago: SC. Griggs and Company.
McArthur, T. (Ed.). (1992). The Oxford companion to the English language. Oxford: Oxford
University Press.
McDonald, N. (Ed.) (2006). Medieval obscenities. York: York Medieval Press.
Norrick, N. R. (1984). Stock conversational witticisms. Journal of Pragmatics, 8, 195–209. ​
https://‍doi.org/10.1016/0378-2166(84)‍90049-3
Ortega y Gasset, J. (1925/1948). The dehumanization of art and other essays on art, culture
and literature [La deshumanización del arte e ideas sobre la novela]. Princeton: Princeton
University Press.
Ortony, A. (1975). Why metaphors are necessary and not just nice. Educational Theory, 25,
45–53.  ​https://‍doi.org/10.1111/j.1741-5446.1975.tb00666.x
Padilla Cruz, M. (2005). On the phatic interpretation of utterances: A complementary relevance-
theoretic approach. Revista Alicantina de Estudios Ingleses, 18, 227–246. ​
https://‍doi.org/10.14198/raei.2005.18.11
Padilla Cruz, M. (2015). On the role of vigilance in the interpretation of puns. Humor, 28(3),
469–90.
Pollack, J. (2011). The pun also rises: How the humble pun revolutionized language, changed his-
tory, and made wordplay more than some antics. New York: Gotham Books.
Read, S. (2007). Puns: Serious wordplay. In S. Adamson, G. Alexander & K. Ettenhuber (Eds.),
Renaissance Figures of Speech (pp. 81–94). Cambridge: Cambridge University Press. ​
https://‍doi.org/10.1017/CBO9780511988806.011
Redfern, W. (1984). Puns. Oxford: Basil Blackwell.
Sheridan, T. (1719/2010). Ars Punnica. Farmington Hills, Mich.: Gale Ecco.
Sherry, R. (1550/2012). A treatise of schemes and tropes very profytable for the better vnder-
standing of good authors. London: Tredition Classics. http://‍www.gutenberg.org/
files/28447/28447-h/‍28447-h.htm
Solska, A. (2012a). Relevance-theoretic comprehension procedure and processing multiple
meanings in paradigmatic puns. In E. Wałaszewska, & A. Piskorska (Eds.), Relevance
theory: More than understanding (pp. 167–182). New Castle upon Tyne: Cambridge
Scholars Publishing.
Solska, A. (2012b). The relevance-based model of context in processing puns. Research in Lan-
guage, 10(4), 387–404.  ​https://‍doi.org/10.2478/v10015-012-0001-0
Solska, A. (2012c). The relevance of the juxtaposition of meanings in puns. In W. Skrzypczak, T.
Fojt, & S. Wacewicz (Eds.), Exploring language through contrast (pp. 184–200). Newcastle-
upon-Tyne: Cambridge Scholars Publishing.
Solska, A. (2013). Between politeness and impoliteness: the face-work of punning. Paper
presented at the 1st Sosnowiec Symposium on Communication Across Cultures: Face and
Interactions, 26–27 April 2013, University of Silesia, Poland.
Solska, A. (2017). Degrees of ‘punniness’? A relevance-theoretic account of puns and pun-
like utterances. In A. Piskorska, & E. Wałaszewska (Eds.), Applications of relevance
theory: From discourse to morphemes (pp. 198–222). New Castle upon Tyne: Cambridge
Scholars Publishing.
 Chapter 9. Tropes of ill repute: puns 287

Solska, A & Rojczyk. A. (2015). Appreciation of purposive ambiguity: the relevance of puns
in city promotional slogans. In D. Gabryś-Barker (Ed.), Topics in applied psycholinguistics
(pp. 61–80). Katowice: Wydawnictwo Uniwersytetu Śląskiego.
Solska, A. & Rojczyk. A. (2016). "Experience Our Sense of Yuma”: puns and humor in city
promotional slogans. Paper presented at the Linguistic Approaches to Funniness, Amusement
and Laughter (LAFAL) 4th International Symposium: Theoretical Issues in Humour: Building
Bridges across Disciplines, 17–18 March 2016, University of Łódź, Poland.
Sperber, D & Wilson, D. (1986/1995). Relevance: Communication and cognition, 2nd ed. Oxford:
Blackwell.
Swift, J. (1716/1939). Modest defence of punning. In H. Davis (Ed.), Prose Works. Volume 4
(pp. 205–210). Oxford Blackwell.
Tanaka, K. (1992). The pun in advertising. Lingua, 87, 91–103. ​
https://‍doi.org/10.1016/0024-3841(92)‍90027-G
Tanaka, K. (1994). Advertising language: A pragmatic approach to advertisements in Britain and
Japan. London: Routledge.
van Mulken, M., van Enschot-van Dijk, R., & Hoeken, H. (2005). Puns, relevance and apprecia-
tion in advertisements. Journal of Pragmatics, 37, 707–721. ​
https://‍doi.org/10.1016/j.pragma.2004.09.008
Wilson, D., & Sperber D. (2004). Relevance theory. In G. Ward, & L. Horn (Eds.), Handbook of
pragmatics (pp. 607–632). Oxford: Blackwell.
Yus, F. (2003). Humor and the search for relevance. Journal of Pragmatics, 35, 1295–1331. ​
https://‍doi.org/10.1016/S0378-2166(02)‍00179-0
Yus, F. (2016). Humor and relevance. Amsterdam and Philadelphia: John Benjamins.
Part 5

Stylistic effects in literary works

 Chapter 10

Another look at “Cat in the rain”

A cognitive pragmatic approach to text analysis

Seiji Uchida
Nara University, Japan

Hemingway’s “Cat in the rain” has been discussed from various viewpoints,
both by literary critics and by linguists. Among the points in dispute is the
issue of whether the cat the American wife saw under the table in the rain was
identical with the big tortoise-shell cat introduced in the final paragraph of the
story. The chapter reconsiders the issue in terms of strong/weak implicatures in
relevance theory. I argue that the interpretation that the cat the hotel owner told
the maid to bring to the American wife is different from the one the wife saw
is obtained as a strong implicature and that the identity of the two cats can be
derived from a weak implicature. I also argue that the latter interpretation leads
to an interesting implication: both cats could be the same type of a tortoise-shell
cat but not identical.

Keywords: strong/weak implicature, Gricean implicature, reference assignment,

cognitive effect, ‘twist’ in literary text, tortoise shell cat

1. Introduction

Relevance theory advocated by Sperber and Wilson (1995) is a promising cog-

nitive pragmatic theory, which tries to explain various kinds of communication
such as linguistic/non-linguistic communication, spontaneous/non-spontaneous
conversation, and literary/non-literary texts.
It is generally assumed in relevance theory that relevance-theoretic notions
based on cognitive and communicative principles of relevance should apply to
literary texts as well as to ordinary face-to-face utterances. (Sperber & Wilson,
1995; Clark, 1996; Uchida, 1998, 2011, 2013; Pilkington, 2000; Wilson 2012).
In literary texts we sometimes find grammatically ambivalent or deviant
expressions. Look at the following text, for example (Uchida 1998).

https://‍doi.org/10.1075/ftl.8.10uch
© 2020 John Benjamins Publishing Company
292 Seiji Uchida

(1) Holly, swinging her legs from the kitchen stool, lectures her mother on
natural foods. Holly is ten. Waldeen says, ‘I’ll have to give your teacher
a talking to. She’s put notions in your head. You’ve got to have meat to
grow.’ Waldeen is tenderizing liver, beating it with the edge of a saucer. Her
daughter insists that she is a vegetarian. – J. Mason, “Graveyard Day”

“Graveyard Day” begins with this passage. A proper noun Waldeen is introduced
in the third sentence. Although this Waldeen is not grammatically guaranteed to
refer to ‘her mother’ above, we may go on reading assuming that Waldeen is Holly’s
mother, and we find that this assumption is right when we reach the sentence Her
daughter insists that she is a vegetarian. This assumption is not grammatically, but
pragmatically derived: to use a relevance-theoretic term, it is the most relevant
interpretation. If the writer wanted to stick to a grammatically ‘helpful’ expression,
the first sentence would be something like (2).
(2) Holly, swinging her legs from the kitchen stool, lectures her mother,
Waldeen, on natural foods.

Let me take another example from deictic time reference.1 The following sentences
are regarded as ungrammatical.
(3) a. *Last night Andrea had left Joan’s house early.
b. *Tomorrow he had a lunch date with Enid Armstrong.

However, when these sentences appear in the passage below, we don’t see anything
wrong there.
Ellie was sure that last night Andrea had left Joan’s house early because
(4) a.
she was planning to meet Rob Westerfield. 
 (M. H. Clark, Daddy’s Little Girl)
b. Tomorrow he had a lunch date with Enid Armstrong. He couldn’t wait to
get his hands on her ring. – M. H. Clark, Loves Music, Loves to Dance

Deictic terms, last night and tomorrow, are usually used to refer to the time of
one day before and one day after the utterance time respectively. Strictly speaking,
last night in (4a) and tomorrow in (4b) do not follow the grammatical rules, but
we go on reading without any doubt. The first relevant interpretation of (4a), for

1. Concerning time reference in fiction, see Adamson (1995), Nikiforidou (2012), Uchida
(2011, 2013) among others.
 Chapter 10. Another look at “Cat in the rain” 293

example, would be that last night is the previous night before Ellie felt sure. In (4b)
it is assumed that he was to have a lunch date with Enid next day.2
In this way, pragmatic interpretations sometimes overrule grammatical rules
in literary texts. I have argued that those grammatically deviant phenomena in
literary texts can be explained in terms of relevance theory. (Uchida, 1998, 2011,
2013) The present paper especially focuses on reference assignment, which plays a
key role in interpreting a given literary text.
Following this introduction, in Section 2, I introduce the concept of implica-
tures in relevance theory, which is used as the theoretical background of the present
analysis. In Section 3, I present a case study of personal pronouns. Section 4 is wholly
devoted to reanalyzing Hemingway’s “Cat in the rain”; Section 4.1 reviews a number
of main approaches to the identity of the two cats that appear in the story, and in
Section 4.2 the issue is to be considered in view of Gricean implicature. Finally, in
Section 4.3, I present a solution to the identification puzzle based on one of the rel-
evance-theoretic notions of strong and weak implicatures. Section 5 is a conclusion.

2. Implicature in relevance theory

‘Implicature’ is a concept introduced by Grice (1975). The basic idea is that im-
plicature is ‘what is not said’ as opposed to ‘what is said’. It is derived through
inference in terms of maxims such as Quantity, Quality, Relation, and Manner.
Suppose Peter and Mary are discussing who should be the next president of
their company, as in (5).
(5) Peter: I recommend Billy for the next president.
Mary: He is too young.
Implicatures of Mary’s utterance would be something like (6a), (6b) or (6c).
(6) a. I don’t agree with you.
b. I don’t recommend Billy.
c. He is not qualified.

What Mary really says in (5) is just ‘He is too young.’ But ‘he is too young’ for
what? Of course, he is too young to be ‘the next president’ of the company, not ‘a

2. The grammatical version of (4a), for instance, would be:

(i) Ellie was sure that the previous night Andrea had left Joan’s house early because she was
planning to meet Rob Westerfield.
Uchida (2013) calls (4a) an example of ‘livelification’ compared to an ordinary grammatical
version of (i).
294 Seiji Uchida

professor’ or ‘the Pope.’ We are expected to recover what is not explicitly said in
(5). Relevance theory calls sentence (7) the ‘explicature’ of Mary’s utterance.
(7) Billy is too young to be the next president of the company.

This explicature is clearly distinguished from Grice’s ‘what is said’.

Relevance theory also posits two kinds of implicature: implicated premise and
implicated conclusion. Look at the following exchange in (8):
(8) Peter: Can you go shopping in the afternoon?
Mary: I’m a bit busy today.
We can get ‘Mary can’t go shopping’ from Mary’s utterance I’m a bit busy today
based on the general assumption ‘If you are busy, you won’t go shopping’. To use
relevance-theoretic terms, the explicature (8) ‘I’m a bit busy today’, together with
the implicated premise ‘If you are busy, you won’t go shopping’, will lead to the
recovery of the implicated conclusion that Mary cannot go shopping on that
particular afternoon.
Another interesting concept concerning implicature in relevance theory is
that there is degree in implicature. Suppose A says to B, Will you marry me? and
B replies as in (9):
(9) a. I’m married.
b. I’ve got a fiancé.
c. I’m not thinking of getting married right now.
d. Give me some time.
e. Oh, wow. I’m really thrilled.

Utterance (9a) strongly implicates that B cannot accept A’s proposal, while A may
not feel rejected entirely if B says (9b). Furthermore, the possibility of getting
accepted by B will be increased from (9c) to (9e). In other words, the degree of
implicatures varies from a stronger one to a less strong or weaker one in each of
the utterances in (9).
So implicatures have the following properties (Sperber & Wilson, 1995,
pp. 199–200):
a. The implicatures of an utterance may vary in their strength.
b. Strong implicatures are those premises and conclusions which the hearer is
strongly encouraged but not actually forced to supply.
c. The weaker the encouragement and the wider the range of possibilities, the
weaker the implicatures.
d. The weaker the implicatures, the less confidence the hearer can have.
 Chapter 10. Another look at “Cat in the rain” 295

That is to say, strong implicatures are implicatures a majority of hearers will derive
and weaker implicatures are implicatures a smaller number of hearers may arrive
at. Story tellers can covertly use these properties of implicatures to guide the read-
ers to a certain goal and may betray their expectations at the final stage in order to
obtain poetic effects such as suspense and twists.3
In this paper I will show that a cognitive pragmatic perspective is a promis-
ing approach to text analysis, explaining that strong and weak implicatures can
be a key notion to contribute to the discussion of the identity of the two cats
in “Cat in the rain.”

3. Strong and weak implicatures and reference assignments in

literary texts

Uchida (1998) calls a ‘twist’ such a kind of dramatic effect that authors of fictional
prose can get when they try to orient the readers’ attention toward certain direc-
tions and at some point they deceive their expectation. Uchida (1998) discusses
how twists are created and interpreted in literary texts, exploiting a relevance-
theoretic notion of strong and weak implicatures. In this section, we’ll see a case
where personal pronouns play a vital role in reaching the interpretation intended
by the author.
A short story, “You’ll never live to regret it,”4 by Jeffery Archer begins with the
following passage.
(10) And so it was agreed: David would leave everything to Pat. If one of them
had to die, at least the other would be financially secure for the rest of their
life. David felt it was the least he could do for someone who’d stood by him
for so many years, especially as he was the one who had been unfaithful.

The readers may assume that the story is about a couple, David and Pat, and that
David has decided to leave what he has to Pat when David dies. This assumption
will be supported by passage (11), which immediately follows (10).
(11) They had known each other almost all their lives, because their parents
had been close friends for as long as either of them could remember.
Both families had hoped David might end up marrying Pat’s sister Ruth,
and they were unable to hide their surprise – and in Pat’s father’s case his

3. See Section 3 for a definition of ‘twist’.

4. J. Archer, Twelve Red Herrings (1997, HarperCollins).

296 Seiji Uchida

disapproval – when the two of them started living together, especially as Pat
was three years older than David.

Then the story continues and Pat is to receive $1,000,000 as a beneficiary when
David dies. The amount is quite large, so David is required to go through a medical
check. In (12) below, David goes to the surgery in New York together with Pat.
(12) The two of them traveled down to New York together the following Monday
to keep the appointment David had made with Geneva Life’s senior medical
consultant. They parted a block away from the insurance company’s head
office, as they didn’t want to run the risk of being seen together. They hugged
each other once again, but as they parted David was still worried about
whether Pat would be able to go through with it.
A couple of minutes before twelve, he arrived at the surgery. A young
woman in a long white coat smiled up at him behind her desk.
‘Good morning,’ he said. ‘My name is David Kravits. I have an appointment
with Dr Royston.’

Take a close look at the underlined he. What does this he refer to? The first inter-
pretation we get might be that he refers to David. From the sentences both above
and below he, it is strongly implicated that he refers to David.
But the reader may begin to feel uneasy about this interpretation when he or
she reads through to the following conversation between David and Pat after the
medical check.
(13) ‘Everything go all right?’
‘I think so,’ he said. ‘They told me I’d be hearing from them in the next few
days, once they’ve had the results of all their tests.’
‘Thank God it hasn’t been a problem for you.’
‘I only wish it wasn’t for you.’
‘Don’t let’s even think about it,’ said David, holding tightly onto the one
person he loved.

Notice that the fifth utterance is provided by David, which means that the first ut-
terance in (13) should also belong to the person who had been waiting for ‘David’
who underwent the check. That ‘David’ is Pat in fact. The truth is that David, who
had been suffering from AIDS and would die after the seventh installment of pay-
ment to the life insurance company, persuaded Pat to get through the medical
check in his place. Yes, Pat is also a man. Pat is a name which can be used for both
men and women. That is, Pat stands not only for Patricia, but also for Patrick.5

5. There is another twist towards the end of the story.

 Chapter 10. Another look at “Cat in the rain” 297

How has the reader been deceived? It has been strongly implicated by those
statements of (14) that Pat is a female, and Archer has carefully avoided using a
male pronoun to refer to Pat until the use of he in line 2 (13).
(14) a.
David felt it was the least he could do for someone who’d stood by him
for so many years, especially as he was the one who had been unfaithful.
b. … the two of them started living together, … .
c. When they eventually went to bed that night, Pat desperately wanted to
make love to David, … .
d. They hugged each other once again.

The statements of (14) are not applied only to male-female partners. It is not
unusual that men are unfaithful to their male partners and live together with male
friends or partners. Pat avoided making love with David because David had suf-
fered from AIDS. And of course hugging does take place between males.
Once we know the truth and read the story again, we find that Archer does
not say overtly that Pat is male and that it is not explicitly stated either that Pat is
female. Even those statements in (14) can be true if Pat is correctly interpreted
as a male person.
We can also find a clue that ‘David’ is in fact Pat. On the fifth page Pat is
described as a sufferer of asthma as in (15).
(15) As they climbed into bed, Pat felt an attack of asthma coming on, … .

Four pages later, when ‘David’ was asked to list diseases he had suffered, he didn’t
mention asthma. This scene strongly suggests that ‘David’ is not really David, but
perhaps few readers may notice this clue.
As has been discussed, personal pronouns can be manipulated depending
on who they refer to. Reference assignment is mainly a matter of grammar, but
pragmatics can intervene there. Especially in text, referential relations may be
determined in terms of implicatures based on the principle of relevance. Strong
and weak implicatures are one of the key notions to explain those aspects of
textual peculiarity.

4. Approaches to the identity of the two cats

“Cat in the rain” is a story about an American couple, who are staying at a rather
small hotel in Italy. From the hotel room window, the American wife sees a cat
outside in the rain seeking shelter under a table. Feeling sorry for the cat, she goes
downstairs to get it, but it is gone when she gets there. Disappointed, she returns
to their room and finds her husband George still reading a book on the bed. She
298 Seiji Uchida

tells George what she wants, but he does not show any interest. At the end of the
story, the maid in charge of their room brings a big tortoise-shell cat for the wife,
saying it’s from the hotel owner.
As Bennett (1990: 245) says, “Cat in the rain” is one of the best short stories
by Ernest Hemingway. The story is just around 1,400 words long, but it has been
inviting intriguing discussions on such topics as ‘What kind of relation is there
between the American wife and her husband?’, ‘What do the wife’s favorable
descriptions towards the hotel owner imply?’, ‘What does the ‘cat’ symbolize?’, ‘Is
the cat the maid brought to the wife the same cat as the one she saw from the hotel
room?’, and so on. The concern in this paper is with the last one, an issue of the
identity of the two cats.

4.1 Interpretations of the identity of the cats

The beginning of the story tells us that an American couple is staying on the second
floor at an Italian hotel facing the sea, the public garden, and the war monument. It
is raining and the American wife looks out of their window. The story goes:
(16) The American wife stood at the window looking out. Outside right under
their window a cat was crouched under one of the dripping green tables. The
cat was trying to make herself so compact that she would not be dripped on.
‘I’m going down and get that kitty,’ the American wife said.
‘I’ll do it,’ her husband offered from the bed.
‘No, I’ll get it. The poor kitty out trying to keep dry under a table.’
The husband went on reading, lying propped up with the two pillows at the
foot of the bed.
‘Don’t get wet,’ he said.

But the wife fails to find the cat and returns to their room. On the way she meets
the owner of the hotel, suggesting that she likes him because he might have every-
thing her husband does not possess. The story ends with the following passage:
(17) Someone knocked at the door.
‘Avanti,’ George said. He looked up from his book.
In the doorway stood the maid. She held a big tortoise-shell cat pressed tight
against her and swung down against her body.
‘Excuse me,’ she said, ‘the padrone asked me to bring this for the Signora.’

The issue in dispute is whether the cat mentioned in (16) is the same cat or not
as the one introduced in (17). The following three interpretations have been pro-
posed by linguists as well as literary critics:
 Chapter 10. Another look at “Cat in the rain” 299

(18) a. The two cats are identical. : Baker (1952)

b. The two cats are different. : Hagopian (1962), Stubbs (1983), Carter
(2008), Bennett (1990), Ohnuma (1991).
c. It is not possible to decide whether the two cats are identical or not. :
Lodge (1981), Holmesland (1990)6

Baker (1952) does not say the reason explicitly, but from the fragment quoted
below we can conclude that he maintains the first interpretation because It in the
last sentence refers to ‘the cat’ in the preceding passage.
The poor girl is the referee in a face-off between the actual and the possible. The
actual is made of rain, boredom, a preoccupied husband, and irrational yearn-
ings. The possible is made of silver, spring, fun, a new coiffure, and new dresses.
Between actual and possible stands the cat. It is finally sent up to her by the kindly
old inn-keeper, whose sympathetic deference is greater than that of the young
husband. (Baker, 1952, p. 136)

The second interpretation (18b) is advocated by the majority of critics and lin-
guists such as Hagopian (1962), Stubbs (1983), Bennett (1990), Ohnuma (1991)
and Carter (2008), mainly noting that the noun kitty, which is repeatedly used
throughout the story, cannot be used in reference to a ‘big’ tortoise-shell cat.7
Hagopian (1962) is not so positive, but the big tortoise-shell cat is too big for the
wife to embrace as a baby:
It is not clear whether this is exactly the same cat that the wife had seen from the
window – probably not; in any case, it will most certainly not do. The girl is willing
to settle for a child-surrogate, but the big tortoise-shell cat obviously cannot serve
that purpose. (Hagopian, 1962, p. 222)

Stubbs (1983) and Carter (2008) claim, respectively, that a kitty cannot in any way
be a big tortoise-shell cat:
We have already seen that Hemingway fails to assert either that it is the same cat,
or a different one, and this already weakens his commitment to the truth of both
these propositions. My interpretation is therefore that Hemingway implicates that
it is not the same cat. He does this by inserting information, which is otherwise

6. Stubbs, Carter, and Ohnuma are linguists, while the others are literary critics.

7. A kitty may not necessarily imply a ‘young cat.’ Ohnuma (1987, 1991) mentions that in this
story the noun may be used instead of cat as a kind of ‘baby talk’ in that kitty frequently appears
especially when the wife talks to her husband.
One of the reviewers also says that it is possible to use kitty as an endearment expression for a
large cat. If we consider this point from the perspective of the present paper, we can say that kitty
carries a strong implicature that the reference is to a rather small cat, as to be discussed below.
300 Seiji Uchida

irrelevant: that the maid brings a big tortoise-shell cat. Informally, we might say
that there is no reason to mention what kind of cat it is, unless this is significant,
and unless we are expected to draw our own conclusions. (Stubbs, 1983, p. 209)

The linguistic texture of the story would lead us to conclude that the kitty is not
the same as the cat described at the end. (Carter, 2008, pp. 106–107)

Both of them came to the conclusion noticing linguistic features.

Bennett (1990)’s claim is based not on the content of the story, but on the
extratextual information that Hemingway mentions two cats in his note of “Cat
in the rain”:
Extratextual evidence supports the view that there are two different cats. …. Since
both the garden and the green table appear in the published story, it is reasonable
to conclude that the ‘pair of cats’ also appears. The ‘small cat’ in Hemingway’s note
becomes the female ‘kitty’ the wife sees, and the ‘big’ male cat becomes the ‘big
tortoise-shell’, which the padrone sends to the room. (Bennett, 1990, p. 247)

On the other hand, Lodge (1981) advocates a different interpretation:

We can now fully understand why the ending of the story is so ambiguous: it is
primarily because the narration adopts the husband’s perspective at this crucial
point. Since he did not rise from the bed to look out of the window at the cat
sheltering from the rain, he has no way of knowing whether the cat brought by the
maid is the same one – hence the non-committal indefinite article ‘a big tortoise-
shell cat.’ (Lodge, 1981, p. 15)

Lodge claims that a ‘kitty’ is described from the perspective of the American wife,
while ‘a big tortoise-shell cat’ is presented from her husband’s perspective, so it is
not possible to judge the identity of the two cats. This explanation sounds plausible
and is surely supported, for example, by looking at the way of introducing a proper
noun George, which appears in the following context toward the end of the story.
(19) She went on up the stairs. She opened the door of the room. George was on
the bed, reading.

Compare (19) to (20), where ‘George’ is replaced by ‘her husband.’

(20) She went on up the stairs. She opened the door of the room. Her husband
was on the bed, reading.

It is natural that the wife sees ‘George’, not ‘her husband’, when she opens their
hotel room.8 This explains that (19) above is depicted from the perspective of

8. This is another instance of ‘livelification’.

 Chapter 10. Another look at “Cat in the rain” 301

the American wife and that perspectives of the characters play a crucial role
in this short story.
Holmesland (1990) approves of Lodge’s claim, saying:
The ambiguity of the ending, Lodge suggests, hinges on the mystery of the tortoise-
shell cat’s identity. We do not know whether it is the ‘kitty’ the wife spotted outside
and so do not know whether she will be pleased to get it.
 (Holmesland, 1990, p. 60)

Ohnuma (1987) also approves of Lodge’s interpretation, but Ohnuma (1991) clearly
says that the two cats were different because the description is still made by the
author, Hemingway, even if it is narrated from the wife or husband’s perspective.
As we have discussed, the same cat hypothesis is practically unwarranted,
while the different cat hypothesis gains a high score, although the ‘unattainable’
approach cannot be dismissed. Which one should we stand on?

4.2 Gricean implicature analysis

There is an approach to this issue in terms of implicature in the Gricean sense.

Cancellability is one of the features of Gricean implicatures: “a generalized conver-
sational implicature can be canceled in a particular case” (Grice, 1989, p. 39). But
we have to notice that this is done “by the addition of a clause that states or implies
that the speaker has opted out.” (Ibid.) Let me give an example in (21).
(21) A: Can you go shopping in the afternoon?
B: I’m a bit busy today, but I can spare a few minutes if you wish.
The first part of B’s utterance implicates that B cannot go shopping, but B cancels
the implicature in the second part of the utterance. Notice that it is the speaker
who can cancel. It will be clearer when we compare this phenomenon with a case of
denial or negation, which is freely denied or negated by the other party as in (22).
(22) A: Bill’s father is a professor.
B: No, he’s a lawyer.
Stubbs’ conclusion that they are different cats is derived from the fact that the maid
brings a big tortoise-shell cat. He also mentions cancellability of implicatures in
the Gricean sense, using (23).
(23) The big tortoise-shell cat didn’t look like the kitty the wife had seen – but it
was.9

9. Stubbs’ interpretation is, as mentioned above, that the big tortoise-shell cat is not the cat the
wife had seen.
302 Seiji Uchida

According to Stubbs, we get the following implicature in (24) from the first part of
the sentence in (23).
(24) The tortoise-shell cat is not the same cat as the wife saw in the rain.

Stubbs says this implicature is cancelled by the latter part of (23). In other words,
the reason why the identity of the two cats is ambiguous is that various implica-
tures obtained can be cancelled any way (Stubbs, 1983, pp. 209–210, cf. Ohnuma,
1987) But notice that (24) is not explicitly stated in the text of “Cat in the rain.”
This is just one of the implicatures we can derive from the story. The point here is
that the person who can cancel an implicature is the person himself or herself who
is responsible for what she or he said.
One more crucial question about this implicature analysis: How should we
treat the other two interpretations once we take one of those? Should we just cancel
or ignore those interpretations? What kind of criteria should we stand on if we can?
It would not be possible to reply to these questions in terms of Gricean implicature
analysis. Instead, I propose a strong and weak implicature approach to this issue.

4.3 Strong and weak implicature analysis

On a lexical level it is clear that there is a distinctive contrast between a kitty and
a big tortoise-shell cat. From this we will assume that the two cats are presented as
different ones. To use our term, it is strongly implicated that the cat the American
wife saw in the rain is not the same cat as the maid brought at the end of the story.
If this is an implicature obtained from the text, it cannot be cancelled unless the
same author says so.
(25) strong implicature: The cat the American wife saw in the rain is not the same
cat as the maid brought at the end of the story.

Before discussing the other two interpretations in (18), let us focus on the use of
female pronouns10 in the underlined sentences in the Appendix, which are quoted
here as (26):
(26) The American wife stood at the window looking out. Outside right under
their window a cat was crouched under one of the dripping green tables. The
cat was trying to make herself so compact that she would not be dripped on.

10. There is another use of a female pronoun in the utterance by the American wife toward the
end of the story, where her refers to an imaginative kitty:
‘I want to pull my hair back tight and smooth and make a big knot at the back that I can feel,’
she said. ‘I want to have a kitty to sit on my lap and purr when I stroke her.’
This use of her will be ignored here.
 Chapter 10. Another look at “Cat in the rain” 303

Look at the female personal pronouns herself and she, which undoubtedly refer
to ‘a cat’ in the previous sentence. Where do these female pronouns come from?
Bennett (1990) notices this particular anaphoric reference and claims that
the American wife doesn’t know that the cat is female, but she subconsciously
transfers her feeling of homelessness to the cat:11
The effect is a sense of homelessness, similar to the condition of a homeless cat
in the rain, and the authorial voice identifies the cat “crouched under one of the
dripping green tables” as a female, “trying to make herself so compact that she
would not be dripped on.” When the wife sees the cat in the rain she has no way of
knowing that it is female, yet she immediately makes a subconscious transference
of her own sense of homelessness to the cat, and she wants to do for the cat what
George will not do for her, provide a place of acceptance and comfort.
 (Bennett, 1990, p. 252)

I’d like to take a quite different approach to this issue, noting the biological
peculiarities of tortoise-shell cats.12 The colors of tortoise-shell cats are usually white,
black and brown or orange13 and are easily identified as such. But it may not be known
to many people that tortoise-shell cats are mostly female. Female cats have two X
chromosomes, while males have one as well as a Y chromosome. That is, females
have an XX pattern and males an XY pattern, which is the same as we human beings:
Patterns of chromosomes
Females: XX
Males: XY

The black and brown colors are determined by the X chromosome, while the
color of white is not carried by an X or Y chromosome. Now consider the possible
combinations of black and brown.
(27) Females: XX
a. Black Brown
b. Black Black
c. Brown Brown

11. Ohnuma (1987, p. 92; cf. Ohnuma, 1991, pp. 55–56) also discusses the use of female pro-
nouns. He claims that the passage in (26) is depicted by Hemingway from the perspective of the
American wife and adds that American men often use female pronouns to refer to nouns which
are normally referred to by neutral pronouns.

12. Bennett (1990, pp. 255–256, 487) also notices the following peculiarities, but he doesn’t
mention the relationship between those peculiarities and the uses of female pronouns in (26).

13. Incidentally, we say ‘Mike-neko’ in Japanese for a tortoise-shell cat. It literally means ‘a cat
with three colors’.
304 Seiji Uchida

(28) Males: X
a. Black
b. Brown

Suppose that the color white happens to emerge in those cases in (27) and (28).
Then it is only in the case of (27a) that we get a cat with three colors. That is, when
these two basic colors are combined with the color white, those cats will be three-
tone color cats or tortoise-shell cats. Since male cats have only one X chromosome,
the likelihood that male cats become tortoise-shell cats is exceptionally rare. (Some
say that the likelihood might be one in 30,000.)
From the evidence we’ve shown so far, we can provide a possible reason why
female pronouns are used in (26) above. The wife sees the cat in the rain from their
hotel room on the second floor before it gets dark. Under those circumstances it’s
not difficult at all to identify the cat as a tortoise-shell cat with a distinctive feature
of three colors. This is why female pronouns are used.14
If this assumption is right, a fourth interpretation concerning the identity of
the cats will arise. The cat the American wife saw in the rain was also a tortoise-
shell cat, and the maid brought her the same kind of cat. Yes, the kind of cat is
significant as Stubbs (1983, p. 209) states, but he only focuses on the size of the cat.
Note the underlined part in the quote which is repeated below.
We have already seen that Hemingway fails to assert either that it is the same cat,
or a different one, and this already weakens his commitment to the truth of both
these propositions. My interpretation is therefore that Hemingway implicates that
it is not the same cat. He does this by inserting information, which is otherwise
irrelevant: that the maid brings a big tortoise-shell cat. Informally, we might say
that there is no reason to mention what kind of cat it is, unless this is significant,
and unless we are expected to draw our own conclusions. (Stubbs, 1983, p. 209)

Now we should add another interpretation to (18) above:

(18) d. The two cats are tortoise-shell cats.

It is sometimes claimed that if the two cats are the same, the American wife’s wish
is fulfilled and that if they are different cats, it leads to an unhappy ending for
the wife. This fourth interpretation may invite further ‘literary’ implications. The
following might be a candidate: the hotel owner’s consideration for the American
wife can be detected there, but it also shows a limitation since he was not able to
provide the same cat as the wife saw in the rain.

14. According to the following web site, Hemingway shared his Key West home with more
than 30 cats. (http://‍user.xmission.com/~emailbox/cat_lovers.htm as of August 2012) He must
have been a cat lover and most probably knew that tortoise-shell cats are mainly female. Cf.
http://‍www.hemingwayhome.com/
 Chapter 10. Another look at “Cat in the rain” 305

We can explain those four interpretations in terms of strong and weak implica-
tures. The contrast of a kitty with a big tortoise-shell cat strongly implicates that the
two cats are different. As Lodge correctly pointed out, it is impossible to judge the
identity of the cats since the end of the story is described from the viewpoint of the
husband who didn’t see the cat in the rain. This implies that there still remains a
possibility that the two cats are identical. We could say this interpretation is rather
weakly implicated, definitely less strong than the ‘not identical’ interpretation.
Perhaps our fourth interpretation is most weakly implicated since a biologically
peculiar feature of tortoise-shell cats is not widely known and the female personal
pronouns herself and she are tacitly used.

5. Conclusion

Whichever interpretation we take, Hemingway is successful in leading the read-

ers to various possible interpretations. Those interpretations vary from the most
accessible to the least accessible. Stubbs (1983) challenges the issue in terms of
Gricean implicature analysis, but as I discussed in Section 4.2, it is not clear who is
responsible for cancelling an implicature. Furthermore, the questions remain how
a certain implicature is selected among several and how the others are dismissed. In
the present paper I took a relevance-theoretic strong/weak implicature approach.
I have argued that the interpretation that the cat the hotel maid brought to
the American wife is different from the one the wife saw is presented as a strong
implicature, and that the possibility that the two cats are identical can be explained
as a weak implicature. I also argued that there may be a much weaker implicature
that both cats can be the same kind of cat, i.e. a tortoise-shell cat. The ‘identical’
interpretation entails that those two cats are tortoise-shell cats, but the hypoth-
esis here leads to another possibility: the two cats belong to the same species
but are not identical.
The weaker an implicature is, the more contextual effects can be obtained if
the implicature is the one the author implicitly tries to convey. When it turns out,
especially towards the end of a story, that the least unexpected interpretation is
the very interpretation intended by the author, the reader’s expectation, which
has been accumulated through strong implicatures, is finally betrayed. The effect
is an example of the revision of previously derived assumptions, one of the three
contextual or cognitive effects in relevance theory. This is also a typical instance of
‘twists’ in literary texts Uchida (1998) discusses.
I believe that the present paper strongly supports Wilson’s claim that “the no-
tion of weak communication is particularly useful in analysing the vaguer aspects
of literary interpretation” (Wilson 2012).
306 Seiji Uchida

References

Adamson, S. (1995). From empathetic deixis to empathetic narrative: Stylization and (de-
)‍subjectivisation as processes of language change. In D. Stein, & S. Wright (Eds.), Subjectiv-
ity and subjectivisation (pp. 195–224). Cambridge: Cambridge University Press. ​
https://‍doi.org/10.1017/CBO9780511554469.010
Baker, C. (1952). Hemingway: The writer as artist. Princeton University Press.
Bennett, W. (1990). The poor kitty and the padrone and the tortoise-shell cat in “Cat in the
rain”. In J. J. Benson (Ed.), New critical approaches to the short stories of Ernest Hemingway
(pp. 245–256). Duke University Press.  ​https://‍doi.org/10.1215/9780822382348-020
Carter, R. (2008). Style and interpretation in Hemingway’s “Cat in the rain”. In R. Carter, & P.
Stockwell (Eds.), The language and literature reader (pp. 96–108). London & New York:
Routledge.
Clark, B. (1996). Stylistic analysis and relevance theory. Language and Literature, 5, 163–178. ​
https://‍doi.org/10.1177/096394709600500302
Grice, P. (1989). Logic and conversation. In P. Cole, & J. L. Morgan (Eds.), Speech Acts (Syntax
and semantics 3) (pp. 41–58). New York: Academic Press.
Hagopian, J. V. (1962). Symmetry in “Cat in the rain”. College English 24, 220–222. ​
https://‍doi.org/10.2307/373291
Holmesland, O. (1990). Structuralism and interpretation: Ernest Hemingway’s “Cat in the
rain”. In J. J. Benson (Ed.), New critical approaches to the short stories of Ernest Hemingway
(pp. 58–72). Duke University Press.  ​https://‍doi.org/10.1215/9780822382348-005
Lodge, D. (1981). Analysis and interpretation of the realist text: A pluralistic approach to Ernest
Hemingway’s “Cat in the rain”. Poetics Today 1–4, 5–22.
Nikiforidou, K. (2012). The past+now in language and literature. In B. Dancygier, & E. Sweetser
(Eds.), Viewpoint in language: A multimodal perspective (pp.177–197). Cambridge: Cam-
bridge University Press.  ​https://‍doi.org/10.1017/CBO9781139084727.014
Ohnuma, M. (1987). “Cat in the rain” no bumpo no ichimen [A grammar of “Cat in the rain”].
Japanese Linguistics, 6–11, 83–92.
Ohnuma, M. (1991). “Cat in the rain” no bumpo sairon [A grammar of “Cat in the rain” (2)]
Annual Report of Studies in Humanities and Social Sciences, Nara Women’s University, 34,
48–67.
Pilkington, A. (2000). Poetic effects: A relevance theory perspective. Amsterdam: John Benjamins. ​
https://‍doi.org/10.1075/pbns.75
Sperber, D., & Wilson, D. (1995). Relevance: Communication and cognition. 2nd ed. Oxford:
Blackwell.
Stubbs, M. (1983). Discourse analysis: The sociolinguistic analysis natural language. Oxford: Basil
Blackwell.
Uchida, S. (1998). Text and relevance. In R. Carston, & S. Uchida (Eds.), Relevance theory: Ap-
plications and implications (pp. 161–178). Amsterdam: John Benjamins. ​
https://‍doi.org/10.1075/pbns.37.10uch
Uchida, S. (2011). Goyoron no shatei [The scope of pragmatics]. Tokyo: Kenkyusha.
Uchida, S. (2013). Kotoba wo yomu, kokoro wo yomu [Reading words, reading mind]. Tokyo:
Kaitakusha.
Wilson, D. (2012). Relevance and the interpretation of literary works. In A. Yoshimura, A. Suga,
& N. Yamamoto (Eds.), Kotoba-o Mitsumete [Observing linguistic phenomena] (pp. 3–19).
 Chapter 10. Another look at “Cat in the rain” 307

Appendix. CAT IN THE RAIN

THERE were only two Americans stopping at the hotel. They did not know any of the people
they passed on the stairs on their way to and from their room. Their room was on the second
floor facing the sea. It also faced the public garden and the war monument. There were big palms
and green benches in the public garden. In the good weather there was always an artist with his
easel. Artists liked the way the palms grew and the bright colors of the hotels facing the gardens
and the sea. Italians came from a long way off to look up at the war monument. It was made of
bronze and glistened in the rain. It was raining. The rain dripped from the palm trees. Water
stood in pools on the gravel paths. The sea broke in a long line in the rain and slipped back down
the beach to come up and break again in a long line in the rain. The motor cars were gone from
the square by the war monument. Across the square in the doorway of the café a waiter stood
looking out at the empty square.
The American wife stood at the window looking out. Outside right under their window a
cat was crouched under one of the dripping green tables. The cat was trying to make herself so
compact that she would not be dripped on.
‘I’m going down and get that kitty,’ the American wife said.
‘I’ll do it,’ her husband offered from the bed.
‘No, I’ll get it. The poor kitty out trying to keep dry under a table.’
The husband went on reading, lying propped up with the two pillows at the foot of the bed.
‘Don’t get wet,’ he said.
The wife went downstairs and the hotel owner stood up and bowed to her as she passed the
office. His desk was at the far end of the office. He was an old man and very tall.
‘Il piove,’ the wife said. She liked the hotel-keeper.
‘Si, si, Signora, brutto tempo. It’s very bad weather.’
He stood behind his desk in the far end of the dim room. The wife liked him. She liked
the deadly serious way he received any complaints. She liked his dignity. She liked the way he
wanted to serve her. She liked the way he felt about being a hotel-keeper. She liked his old, heavy
face and big hands.
Liking him she opened the door and looked out. It was raining harder. A man in a rubber
cape was crossing the empty square to the café. The cat would be around to the right. Perhaps
she could go along under the eaves. As she stood in the doorway an umbrella opened behind her.
It was the maid who looked after their room.
‘You must not get wet,’ she smiled, speaking Italian. Of course, the hotel-keeper had sent her.
With the maid holding the umbrella over her, she walked along the gravel path until she was
under their window. The table was there, washed bright green in the rain, but the cat was gone.
She was suddenly disappointed. The maid looked up at her.
‘Ha perduto qualche cosa, Signora?’
‘There was a cat,’ said the American girl.
‘A cat?’
‘Si, il gatto.’
‘A cat?’ the maid laughed. ‘A cat in the rain?’
‘Yes,’ she said, ‘under the table.’ Then, ‘Oh, I wanted it so much. I wanted a kitty.’
When she talked English the maid’s face tightened.
‘Come, Signora,’ she said. ‘We must get back inside. You will be wet.’
‘I suppose so,’ said the American girl.
308 Seiji Uchida

They went back along the gravel path and passed in the door. The maid stayed outside to
close the umbrella. As the American girl passed the office, the padrone bowed from his desk.
Something felt very small and tight inside the girl. The padrone made her feel very small and at
the same time really important. She had a momentary feeling of being of supreme importance.
She went on up the stairs. She opened the door of the room. George was on the bed, reading.
‘Did you get the cat?’ he asked, putting the book down.
‘It was gone.’
‘Wonder where it went to,’ he said, resting his eyes from reading.
She sat down on the bed.
‘I wanted it so much,’ she said. ‘I don’t know why I wanted it so much. I wanted that poor
kitty. It isn’t any fun to be a poor kitty out in the rain.’
George was reading again.
She went over and sat in front of the mirror of the dressing table looking at herself with the
hand glass. She studied her profile, first one side and then the other. Then she studied the back
of her head and her neck.
‘Don’t you think it would be a good idea if I let my hair grow out?’ she asked, looking at her
profile again.
George looked up and saw the back of her neck, clipped close like a boy’s.
‘I like it the way it is.’
‘I get so tired of it,’ she said. ‘I get so tired of looking like a boy.’
George shifted his position in the bed. He hadn’t looked away from her since she started to
speak.
‘You look pretty darn nice,’ he said.
She laid the mirror down on the dresser and went over to the window and looked out. It was
getting dark.
‘I want to pull my hair back tight and smooth and make a big knot at the back that I can feel,’
she said. ‘I want to have a kitty to sit on my lap and purr when I stroke her.’
‘Yeah?’ George said from the bed.
‘And I want to eat at a table with my own silver and I want candles. And I want it to be spring
and I want to brush my hair out in front of a mirror and I want a kitty and I want some new
clothes.’
‘Oh, shut up and get something to read,’ George said. He was reading again.
His wife was looking out of the window. It was quite dark now and still raining in the palm
trees.
‘Anyway, I want a cat,’ she said. ‘I want a cat. I want a cat now. If I can’t have long hair or any
fun, I can have a cat.’
George was not listening. He was reading his book. His wife looked out of the window where
the light had come on in the square.
Someone knocked at the door.
‘Avanti,’ George said. He looked up from his book.
In the doorway stood the maid. She held a big tortoise-shell cat pressed tight against her and
swung down against her body.
‘Excuse me,’ she said, ‘the padrone asked me to bring this for the Signora.’
 Chapter 11

Echoic irony in Philip Larkin’s poetry

and its preservation in Polish translations

Agnieszka Walczak
University of Warsaw, Poland

This chapter deals with the implications that the relevance-theoretic account of
irony might have for the interpretation of Philip Larkin’s poetry, acknowledged
to be ironic by many literary critics. This is exemplified by poems Church Going
and Water. In reference to the latter, the chapter also considers the impact
of adopting this model on assessing translations. To illustrate this point, two
Polish translations of this poem are discussed. It is argued that incorporating an
analysis of echoic irony into the critique of the translations can shed light on the
issue of their faithfulness to the original.

Keywords: echoic irony, meaning reversal, Philip Larkin, translation,

comparative study

1. Introduction

While Larkin’s work is generally labeled ‘ironic’ by many literary scholars and
readers, it should not be taken for granted that a pragmatic theory of verbal irony
is straightforwardly applicable to the irony permeating Larkin’s poetry. In this
chapter I argue that the relevance-theoretic model of irony is suitable for that
purpose, substantiating this claim with the analysis of the poem “Church-Going.”
Then I apply similar reasoning to the comparative analysis of two translations of
“Water,” arguing further that recognizing the attributive-echoic character of irony
in poetic texts affects their interpretation and can reveal itself in translation. One
of the translations under scrutiny, by Jacek Dehnel, recognizes the echoic character
of irony while the other, by Stanisław Barańczak, follows the rhetorical tradition
in which irony is explained as meaning the opposite of what is said.1 The two

1. A popular definition describes irony as “the use of words that mean the opposite of what you
really think, especially in order to be funny” (Merriam-Webster’s Dictionary: 872).

https://‍doi.org/10.1075/ftl.8.11wal
© 2020 John Benjamins Publishing Company
310 Agnieszka Walczak

readings of irony by the two translators lead to different interpretations of the same
original text, which suggests that the translator’s intuitive grasp of irony as echo
or as oppositeness may be reflected in the ultimate form the target text receives.
The paper is structured into four parts: Section 2 offers a brief introduction
to the relevance-theoretic approach to irony. While it may happen that an ironic
utterance implies the opposite of its literal content, according to relevance theory,
irony “plays on the relationship between the speaker’s thought and a thought of
someone other than the speaker” (Sperber & Wilson, 1995, p. 243), and is ex-
plained as echoic type of speech. Section 3 analyses an example of echoic irony in
Philip Larkin’s poem “Church-Going,” where I argue that dismissing irony as ‘the
opposite of what the speaker means’ leads to a simplistic interpretation. Section 4
focuses on a close reading of two Polish translations of “Water,” proving that the
magnitude of irony in the translated texts largely depends on how many ‘utterances’
in the poem are attributed to someone else. Section 5 offers concluding remarks.

2. Relevance theory on irony

The relevance-theoretic account of irony was put forward by Sperber and Wilson
(1995, 1998) and subsequently developed by Wilson (2006), Wilson and Sperber
(2012), Curco (2000), and Yus (2000, 2016). On this approach irony is seen as
a variety of interpretive (in contradistinction to descriptive) use of language, in
which an utterance achieves relevance in virtue of resemblance to another rep-
resentation. More specifically, irony is conceptualized as attributive echoic use,
where echo is technically defined as the speaker’s alluding to another representa-
tion, such as an utterance or a thought, and simultaneously expressing attitude to
it. If the attitude communicated by an utterance is one of dissociation from the
representation echoed, then this utterance may be considered an instance of irony.
The representation targeted by irony is tacitly attributed to someone other than the
speaker at the time of the utterance (Curco, 2000, p. 261; Wilson, 2006, p. 1729).
This need not be a specific individual, as representations can be attributed to
groups of people, types of people or even people in general (Sperber & Wilson,
1995, p. 238) Therefore, an ironic comment such as Lovely weather uttered during
a downpour need not be immediately preceded by an overtly verbalized hope that
the weather be nice that day, as it echoes a commonly held expectation that the
weather should be nice in general. Since the salience of thoughts and attitudes
held by groups of people, types of people, and people in general varies between
individuals, the degree to which an utterance is recognized as ironic may vary
between the listeners. Besides, irony as such is not a sharply delineated class: there
is a whole range of dissociative attitudes, from highly critical to playfully mocking,
 Chapter 11. Echoic irony in Philip Larkin’s poetry 311

which naturally shade off into non-ironic attitudes. Also echoic uses are related
to other tacitly attributive uses of language, such as indirect speech or thought,
(Wilson, 2006, p. 1733), with all these phenomena forming continua rather than
distinct categories.
Understanding echoic irony involves essentially two elements: recognizing
attribution and dissociation (Curco, 2000, p. 268–273, Piskorska, 2016, p. 51,
Sperber & Wilson, 1995, p. 240). Firstly, recognizing attribution means that the
hearer realizes that the propositional content of the utterance resembles another
representation, such as a thought or an utterance produced by or attributed to
another party (or the speaker herself prior to the time of utterance – this is the
case of self-ironic comments in which speakers mock their own previous naivety).
Secondly, understanding dissociation involves a realization that the utterance does
not correspond to the speaker’s own beliefs and that the speaker, in a tacit way,
distances herself from the representation echoed. Both elements are essential for
an ironic interpretation to arise. In recognizing an ironic character of an utterance,
the hearer applies the relevance-theoretic comprehension heuristic (Sperber &
Wilson, 1995), guided by the communicative principle of relevance.
To provide a more straightforward pattern of irony comprehension, a schema
is provided below based on Curco (2000). A literally-intended utterance that as-
serts p could be described as follows:
the speaker intends
the hearer to believe that
the speaker thinks that
p.

The comprehension of an ironic utterance, on the other hand, may be represented

as in the following schema:
the speaker intends,
the hearer to believe that
the speaker thinks that
someone thinks that/has said that
p
and the speaker does not believe that p.

As seen in the above-presented schema, an ironic utterance involves a metarep-

resentation, i.e. a speaker’s thought about someone else’s thought or utterance,
to which it bears resemblance (Wilson, 2000). Resemblance relations in ironic
utterances come in two varieties: interpretive resemblance in content, which
can be assessed in terms of shared implications, and metalinguistic resemblance,
which manifests itself in terms of linguistic form. The former is present in the
pure type of verbal irony targeted mostly at someone’s views, whereas the latter
312 Agnieszka Walczak

is exploited in parodic irony. The two types of resemblance can co-exist, which
means that irony can be tinged with parodic elements to various extents. It should
however be noted that it is not an objective resemblance between representations
that determines the kind of irony but the speaker’s intention to exploit one or the
other type of resemblance as a means of achieving the desired effect (Piskorska,
2016, Wilson, 2006).
The notion of pretence forms the key concept in approaches to irony which
can be seen as alternative to the relevance-theoretic echoic account. In one of the
versions of pretence theories, Clark and Gerrig (1984), drawing on the etymology
of the word ‘irony’ and Grice’s remarks, argue that: “In being ironic […] a speaker
is pretending to be an injudicious person speaking to an uninitiated audience; the
speaker intends the addressees of the irony to discover the pretense and thereby
see his or her attitude toward the speaker, the audience, and the utterance” (1984,
p. 12). Clark and Gerrig support their argument with an observation that the
ironist often assumes (or impersonates) the tone of voice and other features of
the target of mockery (cf. Wilson, 2006). As Wilson (2006) concedes, pretence
may be exploited in parodic irony but need not play any role in prototypical cases
of verbal irony. For instance, as Wilson explains, it is possible to ironically echo a
thought even if the person to whom this thought is attributed would never express
it. Therefore, if no speech act is involved, there is nothing to imitate or imperson-
ate. This argument is supported by my analysis: it is only by reference to abstract
thoughts or attitudes rather than specific utterances or behaviours that irony in
Larkin’s work is achieved.

3. Philip Larkin’s echoes

Philip Larkin’s work is universally considered to be ironic. “Everything about Larkin

rests on irony, that English specialty and vice” wrote Martin Amis (Diedrick n.d.).
Jacek Dehnel commented: “Larkin very often made use of irony, wrote things that
were absolutely contrary to his own beliefs” (translation mine) (Winiarski, 2010).
However, irony in Larkin’s poetry cannot be accounted for by the traditional
rhetoric tradition since it is not based on saying the opposite of what is meant; the
meaning is constituted by the interplay between what is uttered and meant by the
speaker and understood by the hearer. While irony is often grasped intuitively by
readers and critics, the echoic theory of irony allows the hearer/reader to explain
clearly where irony lies without shedding any of its subtlety or without coarsely
simplifying an ironic utterance to the ‘opposite of what the speaker meant.’ What
follows in this section spells out how dissociative-echoic understanding of irony
may be applied to an instance of very subtle irony in Larkin’s Church-Going. The
 Chapter 11. Echoic irony in Philip Larkin’s poetry 313

logic of this analysis is then applied to a comparative analysis of two translations

of Water in section four.
For illustration, let us consider the first two stanzas of Church-Going where
irony is – quite literally – echoic.
Once I am sure there’s nothing going on
I step inside, letting the door thud shut.
Another church: matting, seats, and stone,
And little books; sprawlings of flowers, cut
For Sunday, brownish now; some brass and stuff
Up at the holy end; the small neat organ;
And a tense, musty, unignorable silence,
Brewed God knows how long. Hatless, I take off
My cycle-clips in awkward reverence.

Move forward, run my hand around the font.

From where I stand, the roof looks almost new -
Cleaned, or restored? Someone would know: I don’t.
Mounting the lectern, I peruse a few
Hectoring large-scale verses, and pronounce
‘Here endeth’ much more loudly than I’d meant.
The echoes snigger briefly. Back at the door
I sign the book, donate an Irish sixpence,
Reflect the place was not worth stopping for. (…) (Larkin, 1990, pp. 97-98)

Philip Larkin creates in the poem “an ironic and self-critical persona” (Motion,
1993, p. 30) who is a “disillusioned pragmatist” tempted to give in to “hopeful ro-
mantic yearning” (Motion, 1993, p. 38). An atheist cyclist enters a modest church
and feels compelled to half-heartedly enact the ritual gestures. He pronounces a
phrase used by the reader to mark the end of the reading of a Lesson according
to the Book of Common Prayer with the intention to mock its apparent antiquity
and hollowness. When the words are echoed in the empty space of the building,
the ambiance briefly becomes menacing. Although the lyrical I retains his offhand
manner, he does nevertheless feel obliged to follow the rituals, as if he were being
watched. Upon leaving, he regrets “stopping for” the church.
It is interesting in what way the lines: “I peruse a few / Hectoring large-scale
verses, and pronounce / “Here endeth” much more loudly than I’d meant. / The
echoes snigger briefly.” may be seen as ironic. As defined by Sperber and Wilson
(1995, p. 239), irony is a dissociative attitude to an echoed content. For an utter-
ance to trigger an ironic interpretative path, it needs to be recognized as echoic, i.e.
bearing resemblance to some other representation, and contextually inadequate,
i.e. being patently false, ridiculous, hyperbolic or irrelevant. The salience of these
314 Agnieszka Walczak

features varies – sometimes the echo is immediately striking and sometimes it

is the inadequacy that is first noticed by the hearer (Piskorska, 2016, p. 61).
Additionally, identifying an ironic attitude of dissociation can clue the hearer in
to the speaker-intended interpretation. All three elements are present in these few
lines, as will be shown below.
As already mentioned, the general attitude of the speaker towards the church
and what it stands for is one of dissociation. The speaker is a non-believer who off-
handedly visits a randomly chosen church as a break from a different activity. His
gestures and phrases reflect an attitude of dissociation: “Hatless, I take off / My cy-
cle-clips in awkward reverence.” As he is already “hatless,” he kneels upon entering
the church to show respect. However, his “reverence” is “damp[ed] down” (Lodge,
1991, p. 218) by the stylistically awkward “cycle-clips.” We can thus see the paradox
that is present in many of Larkin’s poems: the serious or metaphysical undertones
of reverence are overturned by reality (Jarniewicz, 1994, p. 18), which hints at the
attitude of dissociation on the part of the speaker. Examples include: “letting the
door thud shut.” (which is impolite), “stuff / Up the holy end” (foregrounding the
preposition renders the phrase disrespectful), “God knows how long” (which is
a colloquial phrase flouting a very straightforward understanding of the “Thou
shalt not take the name of the Lord thy God in vain” Commandment), “Someone
would know: I don’t.” (the short and offhand “I don’t [know]” implies: ‘I don’t care’
rather than ‘I admire whoever knows that’), “donate an Irish sixpence” (worthless
currency in England), “[I] Reflect the place was not worth stopping for.” (a tacit
expression of the speaker’s dissociation from the church and what it stands for).
The phrase “Here endeth,” being a quotation, is overtly attributed to the reader
of a Lesson according to the Book of Common Prayer, and, by extension, serves as
a representation of religious attitudes (rather than a religion’s theological intrica-
cies). It also points to the attitude of an unshaken conviction as to the character of
the relations between the earthly existence and the hereafter.
While the general attitude of the speaker would suffice to suggest an ironic
interpretation of the speaker’s reading of “Here endeth,” the inadequacy of the
phrase is also signalled in three other ways. Firstly, the description of the verses
as “hectoring large-scale” is incompatible with the provincial atmosphere in the
church. Secondly, as there is no grammatical subject or linguistic context pro-
vided, we do not know what ends, which makes the phrase sounds incomplete and
somewhat comical. Finally, this dramatically final statement is uttered to the total
emptiness of the temple, which creates a possibly humorous clash between the
desolation of the place and the dramatic pronouncements in it that ought to attract
people hungry for metaphysical answers. Given the dissociative attitude hinted
at in the whole stanza and the fact that the line alludes to religious discourse, the
reader is invited to, though not overtly guided, to read the line ironically.
 Chapter 11. Echoic irony in Philip Larkin’s poetry 315

The reading of the verse may also be considered parodic. More precisely, it is
targeted at a speech act which, as elaborated on above, tacitly expresses a view of the
world and the speaker’s utterance bears metalinguistic resemblance to it. The speaker
does not change the linguistic form of the phrase, and his manner draws the reader’s
attention to the “discrepancy between a description of the world someone is appar-
ently putting forward and the way things actually were” (Wilson, 2006, p. 1723).
As the echoes respond, the sense of security of the speaker in his fixed attitudes
is shaken. What was once uttered as parody now ‘comes to life’ in that it responds.
For a split second and half a line, the echoes become a subject (of a sentence)
capable of an answer. In their response, they echo the attitudes of the speaker that
he ironically included in the first reading of “Here endeth,” that is, the attitude
of dissociation from the views held by believers. The clues for dissociation are
twofold. Firstly, the echoes ‘snigger,’ a literary equivalent of a smirk one could use
in spoken dialogue. Secondly, their response also points to the ludicrousness of the
speaker’s dissociative attitude in that: one, it proves that the church is not empty,
invalidating the speaker’s assumption; two, if saying the words out loud brings to
life some sort of being capable of an answer, then the style of “hectoring large-scale
verse” is possibly not as unreasonable and tasteless as the speaker assumed before;
third, the proposition expressed by the utterance may actually be not hollow.
Startled by a new presence in the church, the speaker instantly understands the
attitude of dissociation of the echoes and recognizes that they are not attributed to
people in church but to himself. The echoes briefly parody the speaker’s attitude of
dissociation from the parodied words, assume an ironic attitude to his sense of se-
curity in being a non-believer capable of distancing himself from whatever draws
people to churches. The irony of the echoes consists in their attitude towards the
attitude expressed by the ironic utterance of the speaker.
Returning to the schema presented in Section 2.1, the utterance in quotation
marks “Here endeth” would, at a representational level, look like this:
the reader in church intends
everyone to believe that
the reader in church thinks that
Q.

However, as was argued above, when read out loud by the speaker, the sentence
becomes ironic. It is overtly attributed to someone else via the quotation marks
e.g., to a reader in church. Therefore, it echoes attitudes which are associated with
an assertion of the existence of a supernatural being. Secondly, it can be said that
the speaker dissociates himself from such an assertion as how he prepares to say
“Here endeth” out loud and what does not follow in the poem (that is, an assertion
of faith) is proof of his dissociation. Therefore, we can recognize the attribution:
316 Agnieszka Walczak

the speaker in the poem intends

the reader/himself to believe that
the speaker in the poem thinks that
the reader thinks that
Q
and the speaker does not believe that
Q.

However, as the words are uttered “much more loudly than [he]’d meant,” the
speaker’s irony is itself ridiculed and repeated in a dissociative manner by the
echoes:
the echoes intend
the speaker in the poem to believe that
the echoes think that
the speaker of the poem thinks that, although people in the church
think that Q, he does not believe that Q
and the echoes do not believe that not Q.

Now it is echo that attributes to Larkin an informative intention that is itself ironic.
The identity of the first addressee, then the addressee that responds (the echoes
sniggering) changes in the perception of the speaker. First, the irony is aimed at
believers, but then it is not the believers who retort, but undefined, plural ‘echoes.’
They may be interpreted in several ways, for instance: as history, as God (with a
rather grand gesture), or, more likely, as something which the speaker, as an athe-
ist, does not believe in, e.g.: ‘the other world’, a spiritual being capable of response
and willing to engage in the lives of human beings. By repeating the words, the
echoes parody the attitudes expressed by the speaker. The ironic interpretation of
one phrase is not fixed but is relative to who is saying what to whom and where.
Moreover, if we remember that the whole poem is – typically of Larkin – similar
in structure to a short story and its ‘narrator’ is looking back at a past event, then
the characters multiply and therefore irony ‘shoots’ in even more directions. After
all, it could be – especially given the fact that this is an extremely brief interlude
that is not followed by any sort of assertion of faith in a spiritual being – that the
speaker is now recounting that brief moment to dissociate himself from his past
attitudes. The speaker may attribute the echoed thought to themselves, as long
as the thought was entertained at a different time than the ironic utterance was
produced (Curco, 2000, p. 262). Every time a new representation of that one ut-
terance appears, it can be echoed in a dissociative manner by the reader, following
the communicative principle of relevance. The recurrence is potentially endless, as
long as the reader considers it relevant.
If, on the other hand, the meta-ironic reading of “The echoes snigger briefly.” is
dismissed and echoes are taken to laugh with the speaker rather than at him, then
 Chapter 11. Echoic irony in Philip Larkin’s poetry 317

the whole stanza loses its impetus. The beginning of the poem would then amount
to a description of a parochial church which may evoke both quaint nostalgia and
rational criticism, but in which there is no room for the speaker’s personal invest-
ment. Without such deeply disconcerting tension in which the speaker doubts
his own logic and points of reference, the poem would simply coolly replay an
old theme of the clash between the real and the serious, which would amount to
a “shortcut to producing works that by virtue of their literariness will pass for art”
(Jarniewicz, 1994, p. 42). Without this ultimately self-referential irony, the poem
would comment on a cultural motif, but would lose credibility as an expression of
real-life experience. If the echoes sniggered only to reinforce the negative attitude,
they would no longer problematize the position of the speaker and their role
would be no more than ornamental. That which is the domain of pure aesthetics
as opposed to life was for Larkin the domain of illusion rather than reality and had
no place in his approach to poetry (Jarniewicz, 1994).
Of course, when describing the visit to the church, the speaker does not aban-
don the elements of parody and the ironic description leads us to believe that the
startled reaction of the speaker deserves a quiet snigger on the part of the reader,
too. In that sense, perhaps, the echoes snigger with the speaker. In the end, he talks
of his accidental doubt, momentary premonition in ironic terms. However, only
by recognizing the self-referential character of irony, which is made possible by the
notion of echo and which could not be grasped in terms of the meaning reversal
approach to irony, does the “ironic and self-critical persona” that Larkin is taken
to be appear in full. Depriving the text of that aspect simplifies the meaning and
allows for comfortable deception by fully dissociating the speaker from what is
echoed, without allowing for extra layers of irony. The amount of irony is not fixed
and it is the sensitivity of the reader that will decide to what extent she dares to
trust the speaker at words’ value.
In conclusion, the speaker of the poem realises that the echoes are being ironic
about the speaker’s being ironic. This example shows how intricately woven irony
in Larkin’s work may be; not merely in the form but also essential to the content.
However subtle it seems, it is generally recognized by scholars and the popular
audience. It is also only at this level of subtlety that the differences between transla-
tions of “Water” may be pinpointed.

4. Larkin’s irony in translation

The perception of Larkin’s work and Larkin himself in Poland had long been
shaped by Czesław Miłosz’s Przeciwko poezji Filipa Larkina (Against the Poetry
of Philip Larkin), whose opinion, interpreted as critical, was offset by the
318 Agnieszka Walczak

publication of Barańczak’s 44 wiersze (44 poems, 1991). Today, Jerzy Jarniewicz

figures as the leading specialist in the Polish “Larkinland” (Szymańska, 2012).
Zebrane (Collected, 2008), a translation by Jacek Dehnel, was published in 2008.
Both Barańczak and Dehnel translated Larkin’s “Water” (presented below), but
with starkly different results.
If I were called in
To construct a religion
I should make use of water.

Going to church
Would entail a fording
To dry, different clothes;

My liturgy would employ

Images of sousing,
A furious devout drench,

And I should raise in the east

A glass of water
Where any-angled light
Would congregate endlessly. (Larkin, 1990, p. 93)

Even solely at the lexical level, the seemingly simple wordings bear resemblance
to phrases connected with religion, and too many of them seem to echo – in the
technical sense – religious discourse for us to dismiss them as coincidental echoes.
The conditional structure of the poem suggests an element of playacting as the
speaker elaborates on what he would do if he were in a different context.
In the short text of just 53 words, around 35% of the phrases are connected
with religious topics. These phrases were put in bold in the text above. The speaker’s
attitude to the religious discourse alluded to, however, is difficult to pinpoint. The
wordings appear to represent a highly organized way of being religious, as if the
phrases in bold were a list of the prerequisites a movement or an organization
needs to fulfill to be considered a religion.
The dissociation is extremely subtle, but its tacit pronouncement may be
observed in several aspects. Firstly, the juxtaposition of the logical, almost sci-
entific language (“if,” “construct,” “make use of,” “entail,” “employ”) with religious
discourse (“called in,” “congregate,” “furious devout”) creates a stylistic dissonance.
The attitude is also hinted at by the conditional structure of the text: if he were
“called in,” the speaker would do this and that, which implies that he in fact will
not. Moreover, as already mentioned above, the conditional at the beginning
suggests an element of “make believe” speech in front of an audience. Finally, by
including so many echoic elements in such a peculiar context, the speaker makes
 Chapter 11. Echoic irony in Philip Larkin’s poetry 319

the listener/reader suspicious of his sincerity. The resemblance to elements of

religious discourse and recreating rituals which have earmarks of liturgy allow
the reader to infer the speaker’s dissociative attitude – or perhaps even a mocking
one, if the reader recognizes the similarity between “rais[ing] (…) a glass of water”
and drinking alcohol in a pub. One cannot construct a religion using water, so the
lyrical I is probably not serious about it. The images are not merely absurd (hence
the dissociation), but ultimately also ironic in that the view of being serious about
religious rituals can be attributed to those who hold such views, i.e. church goers.
While the blatant falsity of an utterance does not unequivocally point to irony,
it may be a strong suggestion of its presence (Wilson, 2006, p. 1724). One could
imagine this text being read out as a parody. One could even venture the claim that
it would be much harder to read it fully literally, since many of its phrases would
have to be treated as devoid of any echo.
As language users are not omniscient, it is always possible to overlook an
echo, or to hear an echo where none was intended. This makes the process of
translation even more difficult. The importance of not dismissing the possible
echoes in Larkin’s Water is elaborated upon in the comparative analysis of the two
translations of the text, one of which recognizes the echoic character of irony and
the other fails to retain the echoes and the attitude of dissociation in the text. Both
texts are presented below, together with their literal back-translations into English:
320 Agnieszka Walczak

Woda, tr. Jacek Dehnel Water (back-translation)

Gdyby mnie powołano If I were called in
Do stworzenia religii, To create a religion,
Uczyniłbym użytek z wody. I would make use of water.
Chodzenie do kościoła Going to church
Wymagałoby przeprawienia się przez bród would require crossing a ford
Ku innym, suchym ubraniom; towards different, dry clothes.
Moja liturgia posługiwałaby się My liturgy would employ
Obrazami przemakania, Images od sousing
Gwałtownej, nabożnej ulewy. of a brutal, devout downpour.
I wzniósłbym na wschodzie And I would raise in the east
Szklankę wody, gdzie światło A glass of water, where light
Wpadałoby pod dowolnym kątem Would fall in at any angle
I gromadziło się bez końca. and gather endlessly.
 (Larkin, 2008, p. 65)

Woda, trans. Stanisław Barańczak Water (back-translation)

Gdyby mnie powołano If I were called in
Do wymyślenia religii To come up with a religion
Jej podstawą uczyniłbym wodę. I would make water its basis.
Zamiast chodzić do kościoła Instead of going to church
Przeprawiano by się w bród przez rzekę, [People] would ford a river
Po czym przebierano w suchą odzież; And afterwards change into dry clothing;
Oparłbym całą liturgię I would rest the whole liturgy
Na symbolice przemoknięcia do nitki On the symbolism of getting wet to [the last]
thread
W dzikim akcie strzelistym ulewy, In the wild towering act of the downpour;
A na wschodzie postawiłbym wzniosłą And in the east I would put a lofty
Szklankę wody, w której snop światła Glass of water, in which the beam of light
Mógłby w swoich zmiennych załamaniach Could in its changing refractions swarm
Mrowić się bez końca jak tłum wiernych. endlessly like a crowd of believers.
 (Larkin, 1991, p. 67)

In Dehnel’s translation, the famous “construct[ing] a religion” becomes “creating

a religion.” The word “construct” suggests artificial work, science, which is usually
considered religion’s opposite. One may construct a church – but not an idea,
unless the idea boils down to its packaging. ‘Creating’, however, implies that the
speaker exists prior to religion – a religion which would probably (once created) be
described as eternal, prior to everything. This word in Polish is strongly associated
with religious discourse and cannot be read innocently in the context of religion,
 Chapter 11. Echoic irony in Philip Larkin’s poetry 321

thereby leading the reader to suspect an echo of beliefs related to cosmogony.

Referring to religious discourse is not that strong in the original.
Barańczak’s translation of the word is more neutral than in the original, but
already in the following line the translator makes up for this and the idea of con-
structing becomes the conceit of the poem. Water is the “foundation, basis” of the
new religion; “downpour” is “towering,” reminding the reader of the architecture
of Gothic churches.
The words from the next stanza both in the original and in Dehnel’s transla-
tion allude to religious discourse, and the attitude of the speaker can be seen as
ironic: “going to church” (and all the implied mindlessness it contains), “a fording”
(“the fords of the Jordan”), “to different…” (the motif of a journey, reminiscent of,
e.g. Moses and the Red Sea). Barańczak, on the other hand, makes the sentences
easier to understand by using phrases such as “instead of ” and “after,” which is
typical of his translation style in general (Szymańska 2012, 71). This, however,
renders the wordings less conspicuous, and the reader is not cued in to doubt the
sincerity of the speaker. In Barańczak’s text, the fording through the water takes
place “instead of ” “church-going.” In this way, church-going is depreciated even
more overtly than in the original while the idea of fording is elevated and not at-
tributed to someone else, instead becoming an allusion and gaining in seriousness
as a substitute ritual.
Larkin’s “image” remains an “image” in Dehnel’s translation; Barańczak, on
the other hand, talks about a “symbol,” again elevating the discussion. “My liturgy
would employ / Images of sousing, / A furious devout drench” of the original sug-
gests that the image of the sousing would show a “furious devout drench.” For
Dehnel, similarly, the “brutal devout downpour” is a description of the images
of “sousing.” In Barańczak’s version, however, “being wet to [the last] thread /
In the wild towering act of the downpour” changes the meaning by not treating
the “wild towering act of the downpour” as an elaboration on the “sousing,” but,
rather, an element added to it. Together, the lines create a new vision. Moreover,
the line in the original and in Dehnel’s translation is definitely negative, while in
Barańczak’s version it cannot be classified as either positive or negative. Finally,
Dehnel “raises” while Barańczak “puts” the glass: Larkin’s and Dehnel’s gestures
are reminiscent of elevation. Barańczak stays with his conceit of constructing –
one can (in Polish) “put” a house or a temple, which is a different vision, and does
not allude to religious discourse.
The somewhat fragmentary character of the sentences in the original and
in Dehnel’s translation, which becomes starker in comparison with Barańczak’s
text, guides the reader to stop at every turn of phrase with suspicion. Barańczak’s
version reads much too smoothly to suggest to the reader a possibility of an
ironic attitude. Barańczak creates images in the poem and makes us look for their
322 Agnieszka Walczak

meaning. Dehnel remains more faithful to the original and simply enumerates the
features of the “drench.” While Barańczak decides to use what is considered poetic
language, Dehnel incorporates everyday phrases into the text (“I would make
use”), which is characteristic of Larkin’s work in general.
It can be hypothesised that each of the translators’ interpretation of the text
was influenced by their ability (or willingness) to identify echoic elements in the
text and the extent to which these elements can be attributed to a source. While
both translators clearly recognize some allusions to religious discourse in the text,
it is only Dehnel that remains more suspicious of the words employed, toys with
their meaning in a fashion similar to Larkin’s, and remains more focused on the
form rather than the image. For Barańczak, on the other hand, words serve as a
means of conveying new meanings and images rather than being used as chan-
nels through which associations are prompted, from which the speaker may be
inclined to distance himself.

4.1 Larkin’s irony in translation. Conclusion

Water is not a poem which is easy to interpret definitely and it is my contention

that to a large extent our interpretation of it depends on how many of its phrases
the reader is able to associate with various styles and registers. Jerzy Jarniewicz
(2006) in his analysis of Water suggests that in fact the last gesture of raising a glass
could be either serious or ironic: “glass” may both refer to the beautiful substance,
or a beverage glass, rendering the gesture a parodic blasphemy (the words “glass”
and “sousing” could allude to a more pub-like type of situation: glass to a “whiskey
glass” and “soused” to “drunk”). The translator’s failure to clue the reader in to
recognize the resemblance to this pub gesture is the main reason why Barańczak’s
interpretation strikes as less dramatic, or more optimistic. While Barańczak to
some extent certainly expressed Larkin’s idea, Larkin’s poetry is not lavish with
final statements and Barańczak’s version leaves few to none alternative interpreta-
tions. Dehnel, by being more faithful, saves them. The light that “congregate[s]
endlessly” similarly “congregates with no end.” Barańczak decides to be more
specific and adds: “like a crowd of believers,” narrowing the possible ways of un-
derstanding the line, adding a new image and in a way, again, elevating the activity
of the light by contrasting it with the ordinary. The main difference between the
translations is that Barańczak trusts the words he uses or, if he decides not to, this
is explicitly stated in the text (e.g. by the phrase “instead of ”). In this way, “water”
and “religion” retain their meaning. Dehnel, on the other hand, following Larkin
closely, seems, by focusing on the language and thereby dissociating himself from
an innocent manner of speaking, to parody the phrases used to describe “religion”
and to question their validity. However, while at the linguistic level the attitude
 Chapter 11. Echoic irony in Philip Larkin’s poetry 323

towards metaphysical concepts is withdrawn, the twist may consist in that the final
image, i.e. the light congregating in the glass of water, may represent a metaphysi-
cal experience which is unattainable via ordinary language and blurred by high
registers. Dehnel reveals the vision to us only at the end, while Barańczak’s series
of images anticipate the final one throughout the poem. Dehnel guides the reader
towards the final thought by putting preconceived assumptions and attitudes in
doubt (rather than denying them.) Barańczak on the other hand fails to cue the
reader in to search for relevance on the ironic reading of the utterance, which leads
the reader to develop an interpretation embedded under an endorsing attitude. In
this case, the result is a simplified and distorted text. Pretence, although present
in both texts, does not exhaust the intricacy of irony, which is gradual. While the
irony in Church Going to a large extent may be understood as pretence (the speaker
enacts ritual gestures while expecting the reader to see through the pretence), the
irony in Water is more complex in that it is the echoes of popular wordings and
attitudes behind them combined with a generally negative attitude that trigger an
ironic reading of the text. The attitudes however are extremely subtle as it is, and
it is difficult to imagine what kind of parody they could involve, save for the last
glass-raising gesture. It seems that it is because of the echo that the text’s irony is
so subtle: it is up to the sensitivity of the reader to recognize any “ventriloquisms”
(Swarbrick, 1995) in the text. If the echoes are disregarded, the poem loses its
ironic character, whatever the gesture in the final lines may be, as the analysis
of Barańczak’s translation proves. In short, I contend that although the examples
quoted in the essay bear the characteristics of pretence, they are essentially echoic
and the pretence is an “extension, not an alternative” (Wilson, 2006, p. 1740)
to the echoicity of the utterances. The understanding of irony presented by the
echoic approach is a promising tool for analyzing Larkin’s irony without shedding
any of its subtlety.

5. Conclusion

Depriving Larkin’s poetry of the richness of his ironies may come at a cost to the
depth of the text. The dilemmas his poetry considers are expressed by means of
irony; and irony understood as echoicity rather than saying “the opposite of what
the speaker meant” or pure pretence. What Motion calls a “self-critical persona”
that gained a voice thanks to the democratizing spirit after World War Two is not
a persona with clean-cut answers that calls “things as they are,” but a figure whose
disillusionment lies in that he is able to question and doubt what they think. What
Barańczak does more openly than Dehnel – that is, gives more artistic freedom
to his own voice – prompts the question to what extent a translator has the right
324 Agnieszka Walczak

to change and re-interpret a translated piece. This, however, is a topic for another
paper. Mine ends with the reiterated statement that what we do understand in-
tuitively while reading Larkin’s ironies, namely, the distant echoes of dissociation,
can be fully accounted for by the echoic theory of irony.

Acknowledgements

I would like to thank Professor Agnieszka Piskorska, without whose help, advice and generosity
this article never would have been completed.

References

Clark H. H., & Gerrig, R. J. (1984). On the pretense theory of irony. Journal of Experimental
Psychology: General, 113(1), 121–126.  ​https://‍doi.org/10.1037/0096-3445.113.1.121
Curco, C. (2000). Irony: Negation, echo and metarepresentation. Lingua 110, 257–280. ​
https://‍doi.org/10.1016/S0024-3841(99)‍00041-8
Diedrick, J. (n.d.) Martin Amis on Philip Larkin. http://‍www.martinamisweb.com/pre_2006/
larkin.htm
Jarniewicz, J. (1994). The uses of the commonplace in contemporary British poetry: Larkin, Dunn
and Raine. Łódź: Wydawnictwo Uniwersytetu Łódzkiego.
Jarniewicz, J. (2006). Larkin. Odsłuchiwanie wierszy [Larkin: Listening to the poems]. Kraków:
Wydawnictwo Znak.
Larkin, P. (1990) Collected poems. London & Boston: Faber and Faber.
Larkin, P. (1991). 44 wiersze [44 Poems]. Translated by S. Barańczak. Kraków: ARKA.
Larkin, P. (2008). Zebrane. Mniej oszukani, wesela w Zielone Świątki i wysokie okna [Collected: The
less deceived, Whitsun weddings and high windows]. Translated by Jacek Dehnel. Wrocław:
Biuro Literackie.
Lodge, D. (1991). The modes of modern writing. Great Britain: Edward Arnold.
Motion, A. (1993). Philip Larkin. A writer’s life. England: Faber and Faber.
Piskorska, A. (2016). Echo and inadequacy in ironic utterances. Journal of Pragmatics, 101,
54–65.  ​https://‍doi.org/10.1016/j.pragma.2016.05.007
Sperber, D. & Wilson, D. (1995). Relevance. Communication and cognition. 2nd ed. Oxford UK
& Cambridge USA: Blackwell.
Sperber, D. & Wilson, D. (1998). Irony and relevance: A reply to Drs Seto, Hamamoto and Ya-
manashi. In R. Carston, & S. Uchida (Eds.), Relevance theory: Applications and implications
(pp. 283–293). Amsterdam: John Benjamins.  ​https://‍doi.org/10.1075/pbns.37.16spe
Swarbrick, A. (1995). Out of Reach. The Poetry of Philip Larkin. Malaysia: Macmillan Press Ltd.
Szymańska, K. (2012). Larkina portret zwielokrotniony. O dwóch polskich przekładach bry-
tyjskiego poety [Larkin’s portrait enlarged. On two Polish translations of the British Poet].
Warszawa: Wydawnictwo Naukowe Semper.
Wilson, D. (2000). Metarepresentations in linguistic communication. In D. Sperber (Ed.), Meta-
representations (pp. 411–448). Oxford: Oxford University Press.
 Chapter 11. Echoic irony in Philip Larkin’s poetry 325

Wilson, D. (2006). The pragmatics of verbal irony: Echo or pretence?. Lingua, 116, 1722–1743. ​
https://‍doi.org/10.1016/j.lingua.2006.05.001
Wilson, D & Sperber, D. (2012). Explaining irony. In D. Wilson, & D. Sperber (Eds.), Meaning
and relevance (pp. 123–145). Cambridge: Cambridge University Press. ​
https://‍doi.org/10.1017/CBO9781139028370.008
Winiarski, J. (2010) Ten stary szelma, Larkin (Rozmowa z Jackiem Dehnelem i Jerzym Jarnie-
wiczem). http://‍www.literaturajestsexy.pl/ten-stary-szelma-larkin-rozmowa-z-jackiem-
dehnelem-i-jerzym-jarniewiczem
Yus, F. (2000). On reaching the intended ironic interpretation. International Journal of Com-
munication, 10(1–2), 27–78.
Yus, F. (2016). Relevance theory and contextual sources-centred analysis of irony. Current re-
search and compatibility. In M. Padilla Cruz (Ed.), Relevance theory: Recent developments,
current challenges and future directions (pp. 147–171). Amsterdam and Philadelphia: John
Benjamins.  ​https://‍doi.org/10.1075/pbns.268.06yus
 Chapter 12

Humour and irony in George Mikes’

How to be a Brit

Maria Angeles Ruiz-Moneva

University of Zaragoza, Spain

Within relevance theory, and contrarily to classical formalist models, it is

claimed that figurative language does not require any special processing route,
other than the steps followed in the addressee’s inferential recognition of the
speaker’s communicative intention made manifest through the message encoded
in a certain context. The present paper sets out to analyse humour and irony
as used by Mikes in How to be a Brit as manifestations of figurative language
following a relevance-theoretic approach. It will be shown that the inferential
steps followed by the addressee in coping with the message communicated by
the speaker with different degrees of strength can be exploited for the expression
of irony and humour.

Keywords: George Mikes, How to be a Brit, figurative language, humour, irony

1. Introduction

Humour and irony have often been approached as manifestations of figurative

language, which exploit different mechanisms in order to provoke a certain effect
upon the addressee. Within relevance theory, both call for inferential processes
so that the addressee will derive the meaning intended to be communicated by
the speaker. For this framework, inference lies at the core of the processing of any
form of language in communication, whether literal or figurative.
Neither when using humour nor irony will the speaker expect her target
audience to believe that she endorses what she has expressed. This is one sense in
which ironic or humorous utterances are taken to be figurative. But the interpreta-
tion of figurative utterances also means that the speaker seeks to be understood
“as less than fully literal” (Clark, 2013, p. 25). This amounts to admitting that the
difference between literal and figurative utterances is, for relevance theory, based
on a scale of different degrees.

https://‍doi.org/10.1075/ftl.8.12cas
© 2020 John Benjamins Publishing Company
328 Maria Angeles Ruiz-Moneva

Within this approach, the contrast between literal and non-literal uses of lan-
guage does not rest upon any qualitative difference whatsoever, in contrast to both
Classical rhetoric approaches and also to Grice. As Clark notes, “deciding how
literally a speaker intends her utterance to be taken is another thing that hearers
need to infer” (2013, p. 25). As consistently shown by this author, there are sev-
eral reasons which underlie the relevance-theoretic approach and its criticism of
traditional and Gricean views: first, if non-literal utterances were just a departure
from their literal counterparts, there would be no point in using them; second,
neither do traditional nor Gricean approaches explain what the addressee is sup-
posed to infer in coping with the meaning of non-literal uses of language; third,
the deviation from a norm (as formalist theories put forward) fails to be intuitively
grounded; or fourth, experimental research has demonstrated that figurative ut-
terances do not necessarily take longer or demand more effort than their literal
counterparts (Gibbs, 2001; 1994).
Humour and irony have extensively been covered within the relevance-theo-
retic framework: the former, in works by authors such as Jodłowiec (1991), Curcó
(1998; 1997a; 1997b; 1996; 1995), Torres Sánchez (1999), Yus (2017a, 2017b, 2016a,
2004, 2003, 1995–96), or Biegajlo (2014); the latter was addressed even before the
theory was explicitly fleshed out (Sperber and Wilson 1986/95), by Sperber and
Wilson (1981, 1978), or Carston (1981, 1980) and proposals for its analysis extend
up to the present (Yus 2016b, c, d, 2009, 2001–2000, 2000, 1997–98; Piskorska
2016; Wilson 2013, 2009, 2006; Wilson & Sperber, 2012). Likewise, the possible
connections between irony and humour from a relevance-theoretic perspective
have also been explored (most significantly, by Yus, 2016a, or Piskorska, 2014).
In this paper, the roles played by both irony and humour in the creation of style
and as manifestations of figurative uses of language will be explored. Following the
relevance-theoretic framework, it will be assumed that style and stylistic effects
guide the addressee’s search for optimal relevance.
In How to be a Brit. A George Mikes Minibus, formed by “How to be an alien”,
“How to be inimitable” and “How to be decadent,” George Mikes depicts what are
reflected as the strange habits of the British as seen by an outsider. The author, a
Hungarian journalist who arrived in England at the time of the Second World
War and who remained there until his death in 1978, tells about his impressions
regarding the English people after having been living there for many years. He
initially places himself as an outsider to that society, as an ‘alien’, as he depicts
himself, and therefore, introduces a central element of narrative perspective.
Such narrative perspective is essentially characterised by the distance that the
speaker establishes between himself and the English. It allows him to reflect and
draw on what he regards as his maladjustment to the world knowledge and to
 Chapter 12. Humour and irony in George Mikes’ How to be a Brit 329

the cosmovision of the British. This narrative point of view is also likely to result
in the exploitation of resources such as humour or irony. In his work How to be
a Brit, Mikes offers, therefore, highly satiric, humorous and ironic views of the
experiences that the author had as a foreigner among the British.
Even so, he progressively identifies with the British and eventually becomes
one of them. As a result, even though his irony initially adopts the perspective of
an outsider, then his targets are depicted affectionately as well as critically. This
is reflected in his “Envoi” which closes “How to be decadent”: “Let me say one
more thing in conclusion. When I wrote that other little book, thirty years ago, I
admired the English enormously but did not like them very much; today I admire
them much less but love them much more” (Mikes, 1977, p. 263). It seems, there-
fore, that Mikes’ How to be a Brit may be suitable for the study of what might be
called interfaces between irony and humour.
On the whole, Mikes makes fun of some of the most typical habits of the
English – what might be called the ‘British way of life’. However, he ultimately
laughs at what he conceives of as the sense of uniqueness which, in his view, the
English seem to feel about themselves.
Irony and humour become central aspects in the interpretation of How to be
a Brit. I shall claim and try to demonstrate that the two resources are closely con-
nected in the work. This is so because on the one hand, the author sets out to mock
the British, in particular, what he regards as the British distinctive traits, some of
which, nevertheless, are also shared by most of the ‘aliens’; and, on the other hand,
he does so by distancing himself from the target of his attacks, and therefore, by
being ironic. In the paper, these aspects will be analysed from a relevance-theoretic
perspective.

2. Theoretic framework

The connections between irony and humour have been widely explored through-
out history and the discussion comes down to very recent times (Ruiz Gurillo &
Alvarado Ortega, 2013; Attardo, 2001b; Curcó 1996). For Attardo, “irony may con-
tribute to the perception of humour in a text” (2001b, p. 122). Moreover, both irony
and humour may be based on some incongruity or contextual inappropriateness
between a certain situation and the expectations raised about it (Attardo, 2001a,
2000). Incongruity has been re-interpreted by Yus (2016a, 2004, 2003, 1995–96)
in relevance-theoretic terms as a clash between possible interpretations: the one
which is more likely and contextually favoured, and the one which is less likely in
relevance theoretic terms, but which is ultimately intended to be communicated
330 Maria Angeles Ruiz-Moneva

by the speaker and which gives rise to humorous effects.1 The extra processing
effort is therefore rewarding for the addressee, and the more it is so, the greater his
responsibility in deriving such an interpretation. Similarly, Curcó (1995) claims
that the interpretation of humorous utterances relies upon an interaction between
two aspects: namely, the perception and manipulation of the incongruous, and the
search for relevance.
Attardo (2001b) points at extra processing effort in dealing with irony and hu-
mour, and therefore opens the way for a relevance-theoretic explanation of both.
In fact, within the scope of relevance theory, Curcó (1996) has claimed that both
irony and humour are echoic, even though the sources of echo in either are dif-
ferent. Thus, irony entails negative inferences and may be understood as indirect
negation. In contrast, humour basically relies upon two or more semantic scripts
which mutually oppose and replace one another. The most basic, first-level kinds
of oppositions between scripts, within the General Theory of Verbal Humour
(GTVH), according to Attardo and Raskin (1991), are real vs. unreal, normal vs.
abnormal and possible vs. impossible.
While both irony and humour may echo communicated assumptions, which
may have been made manifest with different degrees of strength and within
different explicit-implicit continua, the source of echo may also be found among
assumptions accessible within the most immediate context or among societal
norms, principles, cultural assumptions, etc. The latter, that is, echoing cultural
differences and their exploitation for the creation of irony and humour will be
paramount in Mikes’ How to be a Brit.
Some of the most important functions of irony include the following
(Rodríguez Rosique, 2013): it may strengthen complicity between participants in
communication – in particular, whenever they come to share a context or cognitive
environment whose access may be hindered to some other participants –; it may
be used in argumentative speech as a strategy for persuasion; or it may intensify a
particular aspect of criticism. The latter is probably the most important function
fulfilled by irony in Mikes’ How to be a Brit.
Within the General Theory of Verbal Humour, proposed by Attardo (2001a);
and Attardo and Raskin (1991), the cognitive construction of scripts, as inter-
nalised and structured areas of information, becomes paramount. The General
Theory of Verbal Humour points at six knowledge resources, hierarchically organ-
ised, whose presence or absence in texts will determine whether these are humor-
ous or not. These knowledge resources are script opposition, logical mechanism

1. Also within the relevance theoretic framework, Curcó (1998, 1997a, 1997b, 1996, 1995) refers
to these as key assumption and target assumption, respectively, being the latter, therefore, the one
which leads to humorous effects.
 Chapter 12. Humour and irony in George Mikes’ How to be a Brit 331

(which corresponds to the resolution phase in incongruity-resolution models

of humour), situation, target, narrative strategy and language. The cognitive ap-
proach to humour put forward by General Theory of Verbal Humour makes this
theory compatible, in our view, with a relevance-theoretic approach to humour
and also to irony.2
Attardo (2001a) sets out to generalise the application of these theories to all
kinds of texts, regardless of their length, as is the case of humorous narratives, or
of their genre. The application of the General Theory of Verbal Humour to hu-
morous texts larger and more complex than the joke presents two different cases,
depending on whether they are structurally similar to or different from a joke. The
contributions of pragmatics concern, crucially, the importance of approaching
meaning in context. This also means considering the speaker’s communicative in-
tention, and the fact that she deliberately chooses to leave certain aspects implicit.
Those may closely be related to humour, whose meaning is ultimately left to the
addressee to infer. Inference, implicatures or presuppositions become, thus, part
of the common core of the analytic tools of humour, just as they are also applicable
to the analysis of irony.
In an elaboration of the General Theory of Verbal Humor put forward by
Attardo (2001), the basic framework of the theory is maintained, but it is also
expanded by incorporating such concepts as the distinction between punch
lines (occurring at the end of a joke) and jab lines (which can be found in any
part of a humorous text other than its ending), and notions accounting for their
possible distribution in a text, namely, strands and stacks. This is so because the
humorous line of a text may present a much wider complexity than a joke. Those
texts may have different jab lines, scattered throughout the text, as in the case of
diffuse disjunctors, such as those resulting from exploiting register for the sake
of humour and irony.
Mikes’ How to be a Brit may be said to achieve optimal relevance by showing
how the British metarepresent themselves, and also how other foreigners or ‘aliens’
metarepresent the British. As pointed out by Wilson (1999), metarepresentations
may be of four main kinds: thoughts about thoughts; thoughts about utterances;
utterances about thoughts; and utterances about utterances. A clash between
the two kinds of metarepresentations entertained respectively by the aliens or

2. As noted by Yus, “the incongruity-resolution schema fits relevance-theoretic ideas perfectly

and this theory provides a cognitively rooted foundation for why certain senses of words are
selected or why certain situations (make-sense frames) are constructed while interpreting the
joke: hearers cannot help doing that because they have this biologically rooted aim to maximise
relevance” (2016a, p. 94). In my view, this compatibility between the two theories also applies to
humorous texts other than the joke.
332 Maria Angeles Ruiz-Moneva

foreigners and by the British, becomes the basis for the opposition of scripts that
creates humour in the work.
Be that as it may, both irony and humour rely on context for their interpretation
and call for inferential processes. The approach followed in the present paper takes
into consideration the following assumptions for the analysis. The processing of
either humour or irony as manifestations of figurative language fails to represent
any qualitative departure from the processing of everyday speech. Humour and
irony may be used by speakers aiming at optimal relevance3 and represent creative
uses of language. In this sense, humour and irony may play an important role in
the creation of style in a text. Within relevance theory, it has been assumed that
style guides or constrains the addressee’s search for relevance (Sperber & Wilson
1995; Furlong, 1996; Pilkington, 2000a; Carston, 2002a; Wilson, 2011; Clark, 2013,
to name but a few representative authors). As for the relationship between irony
and humour in the work, our central claim is that in Mikes’ How to be a Brit the
ironic attitude of distance adopted by the narrator reinforces the comic effects
intended to be communicated.
Within relevance theory, humour has been addressed by authors such as
Jodłowiec (1991), Curcó (1998, 1997a, 1997b, 1996, 1995), or Yus (2017a, 2017b,
2016a, 2003). These authors agree that humorous utterances neither depart from
any supposed ‘literalness’ nor need any special processing mechanism. In coping
with humorous and also ironic utterances, addressees are supposed to abide by
the relevance-guided comprehension heuristic, derived from the presumption of
optimal relevance (Wilson & Sperber, 2004), which is worded as follows:
a. Follow a path of least effort in driving cognitive effects: test interpretations
(e.g. disambiguations, reference resolutions, implicatures, etc.) in order of
accessibility.
b. Stop when your expectations of relevance are satisfied.
(Clark, 2013, p. 119).
This means that any extra processing effort that may be required by certain utter-
ances such as irony or humour will be surpassed by contextual and cognitive effects
attained by the addressee. Moreover, in the case of the interpretation of figurative
language the addressee is supposed to have a much greater amount of freedom
and responsibility in reaching the interpretation intended to be communicated
by the speaker. This is so because, in utterances where humour or irony plays a

3. Optimal relevance occurs whenever an utterance or any other stimulus has, on any given
interpretation, on the one hand, enough contextual or cognitive effects to be worth the ad-
dressee’s attention, and also, on the other hand, it puts her to no gratuitous processing effort in
achieving such effects.
 Chapter 12. Humour and irony in George Mikes’ How to be a Brit 333

significant part, the implicatures conveyed by the utterance are communicated

with lesser degrees of strength, in the form of weak implicatures (Pilkington, 2000;
Blakemore, 1992; Sperber & Wilson, 1995; Carston 2002a, 2002b).
More specifically, in an application of relevance theory to the interpretation of
humorous utterances, Curcó (1995) draws a distinction between foreground and
background implications, depending on whether they are relevant – and therefore
have contextual effects – on their own or not. These two kinds of implications
may be manipulated for the sake of the creation of humorous effects. In fact, this
is a key aspect to cope with humour and irony in Mikes’ How to be a Brit, as the
analysis will show.
Curcó also dwells upon the similarities in the processing of humour and irony
within relevance theory. Thus, both involve the conveyance of a dissociative at-
titude, expressed towards some of the conflicting propositions appearing in the
course of processing an utterance. The clash between such conflicting propositions
which are presented to the audience with different degrees of focalisation – i.e.,
background vs. foreground assumptions – and the implications of this for humor-
ous utterances are aspects thoroughly worked upon and further elaborated by Yus
(most notably, 2016a, b, c, 2013, 2004, 2003, 1995–96).
For Yus, and within the general relevance theoretic framework, the principle
of relevance enables addressees to test assumptions in order of accessibility,4 and
also to select the interpretation intended to be communicated by the speaker. This
principle is valid and applicable for any form of ostensive-inferential communica-
tion. A further and key implication is that resources such as humour or irony do
not require any special processing mechanism.
Moreover, Yus (2016a, 2003) deals with the relevance-theoretic account of
humour in depth. In humorous utterances, the speaker may manipulate the ex-
pectations of relevance entertained by the addressee. Thus, the first interpretation
congruent with the principle of relevance may initially shortcut the humorous inter-
pretation. No matter if the addressee may be requested to invest further processing
effort, his eventually reaching the humorous intended interpretation, which had
initially remained as a background assumption, will produce additional contextual
and cognitive effects. Moreover, the addressee will have selected the humorous
interpretation on the sole grounds of his own interpretive process, which is likely
to arouse further satisfaction because the cognitive effects outweigh the efforts.

4. Within relevance theory, accessibility is a property of context. Context is formed by a set of

assumptions. These will be accessed if they give way to a sufficient number of contextual impli-
cations. Depending on the communicator or the addressee’s accessibility to the context, they
may be in a better or worse position to cope with the meaning intended by their interlocutor.
334 Maria Angeles Ruiz-Moneva

Another assumption underlying relevance theory which is applied to the

processing of humour has to do with the fact that context is taken to be a cognitive
entity which is selected by the addressee on the basis of accessibility of contextual
assumptions, in accordance with the relevance-guided comprehension heuristic
presented above. As Yus notes, “What is inevitable (…) is a readjustment of the
addressee’s relevance-seeking extensions of context and of his expectations of
relevance (mental effort versus cognitive interest) when humorous strategies are
detected” (2003, p. 1300). Furthermore, an important aspect introduced by Yus
(2003) concerns the fact that if the addressee is in the mood to be entertained or
amused, and therefore, ready to participate in communicative interchanges where
irony or humour are an important aspect of the message intended to be commu-
nicated, he may find humour far more enjoyable: “Humorous effects such as the
enjoyment in the resolution of incongruity are worth this extra cognitive effort, es-
pecially if the hearer is ready to join in the joking game” (2003, p. 1300; my italics).
What is more, Yus argues that the different phases of the twofold relevance
theoretic interpretive process of decoding and inference may be exploited for
humorous purposes (2003, p. 1304-ff.). This is so mainly because resources such
as irony or humour do not require any special processing mechanisms. These dif-
ferent phases will be explained below and will be applied to the analysis of humour
in Mikes’ How to be a Brit.
Yus (2009, 1995–96, p. 505) refers to the saturation5 of information and of
cognitive clues as the main aspects shared by the processing of both irony and
humour. In the case of irony, there is a criterion of incompatibility among those
different contextual sources which allow the addressee to infer the speaker’s ironi-
cally intended interpretation. On processing a humorous or an ironic utterance,
the addressee does not follow any peculiar path. The information obtained from
the speaker’s utterance is processed in a certain context and will yield cognitive

5. Saturation of contextual sources has become a central feature of irony in the most recent
contributions to the study of irony and humour within the relevance-theoretical framework
made by Yus (2009, 2016b). This author refers to the existence of certain contextual sources
among which a distinction is made between, on the one hand, a leading contextual source and
several supportive contextual sources, on the other hand. The leading contextual source is essen-
tial for the successful ironic interpretation of an utterance, whilst the processing of supportive
contextual sources may help the addressee to reach the communicative intention made manifest
by the speaker. Following the relevance-theoretical comprehension heuristic, it is stated that
the prioritisation in the access to these contextual sources may help the reader to achieve a
better balance between processing effort and contextual effects. In particular, those supportive
contextual sources are meant to saturate that contextual information available to the addressee
in his processing of the ironic utterance. This means that they help the addressee in reaching the
ironic interpretation of the utterance, while the leading contextual source points at the speaker’s
dissociative attitude that characterises echoic irony.
 Chapter 12. Humour and irony in George Mikes’ How to be a Brit 335

effects. The speaker’s utterance may, however, communicate certain implicatures

which are conveyed with various degrees of strength. As this author notes, irony
may be closely associated with humour. Moreover, the combination of the two,
irony and humour, may reinforce certain existing bonds between communicators,
such as friendship, comradeship or even solidarity (as has also been experimen-
tally tested and shown by Gibbs 2005 or Gibbs and Colston 2002).
In the following sections, I will demonstrate that in Mikes’ How to be a Brit
irony is one more resource at the disposal of humour: it basically sets the context
and reinforces the contextual effects to be communicated. In the work, humour
allows the writer to point at and satirise the customs, habits, attitudes, charac-
teristics, feelings, etc., of the British; it is connected, therefore, with the author’s
communicative intention. If this is so, it is mainly because, as relevance theory
claims, both irony and humour are echoic, so that the speaker tends to express her
attitude – usually of scorn, distance, criticism, or the like – towards her utterance
or towards a concrete state of affairs. Within a relevance-theoretic perspective, in
both irony and humour there are certain incongruities or incompatibilities be-
tween the contextual sources accessed by the speaker. Consequently, the addressee
will be requested to trace the interpretation that he finds most suitable. The more
autonomy and responsibility in the derivation of the implicatures communicated
through these contextual incompatibilities or incongruities is likely to arouse
the more satisfaction in the addressee. Moreover, in Mikes’ How to be a Brit the
relationship established between participants – on the one hand, the speaker; on
the other hand, the British – may be said to be characterised by ambivalence and
shifting throughout time.

3. Analysis: Strategies and purposes for the conveyance of irony and

humour in George Mikes’ How to be a Brit

Generally, irony and humour may result in misunderstandings if the addressee

fails to grasp the speaker’s communicative intention, or, to put it otherwise, if he
fails to access the cognitive environment pointed at by the speaker. However, in
Mikes’ How to be a Brit it is the narrator’s misunderstanding of situations he is
describing that lead to irony and humour. The point is, therefore, to analyse the
exploitation of humour and irony in the work, and shed light upon the kinds of
humorous strategies used by Mikes to convey his puzzled attitude of ironic dis-
tance towards what he perceives or interprets as misunderstandings. These are
often interculturally grounded. It will be assumed that culture, and also intercul-
tural differences, are a part of the context or cognitive environment. Therefore, the
336 Maria Angeles Ruiz-Moneva

speaker’s lack of accessibility to the context envisaged by interlocutors is likely to

result in misunderstandings.
In Mikes’ How to be a Brit, both irony and humour become essential devices
for the creation of style and tend to complement and counterbalance each other.
The author can be expected to laugh at certain aspects of the British, as seen by an
outsider who, nevertheless, progressively integrates himself with that society and
at least to some extent comes to feel certain affinity or sympathy with its members.
Next it will be shown that there are several ways in which Mikes’ How to be a
Brit can be said to be echoic. First, the analysis will focus on “Prefaces,” in so far
as they show the speaker’s attitude, which underlies the expression of irony and of
humour in the work. This will be followed by the analysis of humorous strategies
employed in the work and the ways in which these are affected by the attitudes
of ironic distance adopted by the speaker. Finally, the conclusions of the study
will be presented.

3.1 The prefaces

How to be a Brit is formed by three different parts. The author himself refers to
his intentions in writing the book in the “Preface” and also in the “Preface to the
24th impression.” These pieces, therefore, become essential to analyse the author’s
perspective or standpoint to what he is narrating.
The first of part of the series is “How to be an alien” (Mikes, 1946), which, in
contrast to the expectations and reactions of its readership, the author claims not
to have been intended to be funny or humorous. It is structured into two different
sections, “How to be a general alien” and “How to be a particular alien.” As the
author notes, it is meant to show his impressions regarding his living among those
whom he initially regarded as “strange people” (Preface: 9). This preface also offers
instances of humour based on contradictions and on consequences which do not
logically follow from the premises or assumptions entertained. Therefore, they
involve crucial aspects of incongruity and refer to scripts which contradict each
other, since mutually excluding propositions result in the same outcome: “Study
these rules, and imitate the English. There can only be one result: if you don’t
succeed in imitating them you become ridiculous; if you do, you become even
more ridiculous” (“How to be an alien,” Preface, p. 18).
Significantly enough, in his references in the “Preface” to “How to be inimi-
table: Coming of age in England” (Mikes, 1960), he shifts and alternates between
they and we as the grammatical subject of the clause where he explains the purpose
of this part: “In these past twenty-one years England has gained me and lost an
Empire” (p. 97, my italics); “After the war it seemed that we would hardly survive
the blow of victory; nonetheless today we are nearly as well off as the Germans
 Chapter 12. Humour and irony in George Mikes’ How to be a Brit 337

themselves” (p. 98–99, my italics). He also addresses his readership directly: “Oh,
yes, if you want to be a modern Briton – a Briton of the sixties – you have to follow
an entirely a new set of rules. Here they follow” (p. 99, my italics).
A significantly peculiar aspect about Mikes’ How to be a Brit in connection
with the target audience has to do with the fact that, in contrast to what was
indicated in classical treatises – such as Muecke (1982; 1970; 1969) or Booth
(1974) – there is no clear dichotomy between those who can cope with irony, on
the one hand, and those who fail to do so, on the other hand. This is an important
feature of irony which in relevance-theoretic terms has been accounted for in
terms of context accessibility – by authors such as Yus (2016d, 2009, 2000–01, 2000,
1997–98). However, what appears to be the case with the audience or potential
target of irony in How to be a Brit is that it may be regarded as transversal, in the
sense that both the British and other kinds of readership may have access to the
meaning intended by the speaker. In a sense, this may be due to the blending of
humour and irony in the work.
On the whole, the “Prefaces” set the context and provide the basis for the script
oppositions leading to humour that will be projected in the rest of each of the three
parts of Mikes’ How to be a Brit. They also offer clear indices of the attitude of the
speaker and of his communicative intention to distance himself from what he is
narrating and from the targets of humour and irony in the work.

3.2 A relevance-theoretic analysis of humour and irony in George Mikes’

How to be a Brit

A particularly interesting aspect of Mikes’ How to be a Brit in the analysis of irony

and humour concerns the ways in which the same topic is covered all through the
different parts that make it up (namely, “How to be an alien,” “How to be inimi-
table” and “How to be decadent”). This may result in echoic irony and in humour,
by making manifest how later on the author reacts towards this topic as addressed
in the earlier pieces of the other books. As for irony, those may be regarded as
instances of what Hatim (1997) termed as intertextual echoes, whose source refers
to other works or texts and which, in my view, apply not only to translation.6 In

6. Even though a reference is made to the classification of echoes into three different kinds,
namely, intratextual, intertextual and contratextual echoes, as put forward by Hatim (1997), this
is only suggested for the purposes of establishing certain relationships between different texts.
In any case, the concept of echo followed in this essay seeks to abide by the relevance-theoretical
approach to it as an essential or definitory trait of irony. Therefore, echoic irony means that the
speaker distances herself from the propositional content of an utterance attributed to somebody
else than herself at the time of speaking: she fails, therefore, to endorse such propositional
content, which is moreover regarded by her as ludicrously false, inappropriate or irrelevant.
338 Maria Angeles Ruiz-Moneva

this sense, a further refinement or distinction between intertextual sources and

intertextual echoes can be proposed. The difference between the two lies in that
intertextual sources have to do with connections that may be found between dif-
ferent texts, but which, in contrast to intertextual echoes, need not be linked to the
conveyance of irony. These intertextual sources may be related to some extent to a
manifestation of the principle of confirmation (or reinforcement) of assumptions.
In relevance-theoretic terms, this is a cognitive process whereby new information
strengthens and provides evidence for old assumptions. Hatim’s concept of inter-
textual echoes (1997) could be then equated with intertextual sources, whereas
intertextual echoes would be taken to refer to those manifestations of irony whose
source is traced in other different texts. The two parts which follow “How to be
an Alien” also let the reader become acquainted with the ways how the book was
received, no matter if the author’s attitude is clearly humorous and ironic.
Thus, in the different parts of the work these intertextual sources and echoes
have to do with such aspects of British life as queuing, (not) being clever, tea, views
on language, or on shopping. To begin with, this is what the author remarks about
queuing in both “How to be an alien” and “How to be decadent”:
(1) An Englishman, even if he is alone, forms an orderly queue of one.
 (“How to be an Alien”: The National passion, p. 54)
Quite a few people told me that I was mistaken when I made fun of the English
queuing habit. It was simply a war-time expediency, it was explained to me, and it
would disappear in no time.
It is still with us and will remain with us forever because it corresponds to an
inner need, it is a way of self-expression. Other nations need occasional outbursts
of madness and violence; the English need occasional excesses of self-discipline.
Other nations, under unbearable stress, shout, howl, get into brawls, run amok;
the English queue up for a cup of tea.
Demonstrations in other countries are violent affairs, with baton charges and
mass arrests. Such things have occurred here, too, in the past. Today, if you are
bored, you arrange a demo.
 (“How to be decadent”: On the national passion, p. 223).

These are instances which show that irony and humour in How to be a Brit rely
on exaggerations and hyperboles which take everyday actions and customs to the
utmost and absurd consequences. In my view, this is where incongruity rests upon,

Irony is, therefore, connected with the expression of a dissociative attitude; and this is what it
essentially communicates, namely, an attitude to a certain utterance or state of affairs and also to
those who may have endorsed them. The point is that the source of echo may be present in the
text itself (in the case of intratextual echoes) or may be associated to other different texts (in the
case of intertextual echoes).
 Chapter 12. Humour and irony in George Mikes’ How to be a Brit 339

because of the absurdity of the script oppositions introduced. There is also an ele-
ment of cultural background which underlies this one and many of the humorous
situations developed in the work: thus, many of the aspects that are depicted as
idiosyncratically English are in fact also common, trivial and recurrent in many
other parts of the world. The assumptions initially shown as likely to be abandoned
are however strengthened and reinforced. What is more, the consequences are
taken to the extreme of absurd, unlikely situations. These are presented in what
relevance-theory has referred to as ad-hoc concepts7 and which Yus (2016a, 2004)
has applied to the account of humour. This would hold here in the connection
between being bored and arranging a demo, which in habitual contexts would
bear no resemblance or logical connection whatsoever.
The author’s views on how (not) to be clever also stand for instances of echoic
irony and humour appearing in two different parts of How to be a Brit, namely,
“How to be an alien” and “How to be inimitable.” In this case, the author holds
similar views, so that irony is based on the confirmation of assumptions, or on
addressing the same kind of target, with essentially the same kind of attitude.
Therefore, irony would not rest on intertextual echoes, but rather on intertextual
sources. The recurrence is grounded on the target of irony and on the speaker’s
attitude towards it. Thus, irony and humour rely on similar views and attitudes:
(2) In England it is bad manners to be clever, to assert something confidently. It
may be your own personal view that two and two make four, but you must
not state it in a self-assured way, because this is a democratic country and
others may be of a different opinion.
 (“How to be an alien”: ‘How not to be clever’, p. 42)
One thing you must learn in England is that you must never really learn anything.
You may hold opinions – as long as you are not too dogmatic about them – but it is
just bad form to know something. You may think that two and two make four; you
may rather ‘suspect it’; but you must not go further than that. Yes and no are about
the two rudest words in the language.
 (“How to be inimitable”: ‘On not knowing anything’, p. 163; my italics)

In the case of the fragment taken from “How to be an alien,” humour starts with
the line of a satiric utterance that contradicts common sense and world knowledge,

7. Within relevance-theory, ad hoc concepts are regarded as constituents of the explicature of

the speaker’s utterance, and give access to the set of encyclopaedic assumptions that enable the
derivation of implicatures. The important point is that the properties associated with the ad
hoc concept could have been retrieved neither from the concept originally encoded nor from
the vehicle alone; rather, they spring from the association of the two (Carston 2002b; 1997). As
noted by Clark (2013, p. 249), concepts are always adjusted and specified when they are accessed
in a certain context, so that they allow addressees to derive certain implicatures.
340 Maria Angeles Ruiz-Moneva

which may be the conclusion ad absurdum of particular aspects that the author
has observed about the British. Humour and also irony can likewise be linked to
the attitude of echoic distance adopted by the speaker towards the facts that he
interprets – rather than just presenting them – about these groups of people. In
a sense, he is conveying what are but his opinions as if they were facts or factual
assumptions. The reasons provided by the author cannot be said to spring logically
from the premises or assumptions. The incongruity also focuses upon the contrast
between facts and opinions.
This contrast between facts and opinions, or rather, between knowledge or
cognition and opinions, is the aspect that is retaken in “How to be inimitable”: as
in the previous case, there is an absurd statement which is presented as if it were
logical and commonsensical enough. Most importantly, the expression of attitude
that yields humorous and also ironic effects is grounded on what relevance-theory
terms as higher-level explicatures. These involve “embedding the proposition ex-
pressed in a higher-level description” (Carston & Uchida, 1998, p. 297), which may
convey, refer to or describe, amongst other aspects, the speaker’s propositional
attitude. Therefore, they are relevant by guiding and constraining the reader’s deci-
sions to be made on the interpretation (Clark, 2013; Blakemore, 1992).
As noted above, one of the ways in which humour may be conveyed and
accounted for (Yus, 2016a; 2003) has to do with the exploitation for humorous
purposes of the different phases involved in the coding/decoding and inferential
aspects of verbal communication. Such phases are the following: extraction of a
logical form; ambiguity resolution; reference assignment; enrichment (or filling of
a semantic gap, which can be exploited humorously); or the derivation of impli-
catures. Next, instances of the exploitation of inferential processes in the different
parts of the How to be a Brit will be analysed. Regarding the first case, that is,
humour based on the extraction of a logical form, it should be noted that it calls
for the existence of a certain syntactic ambiguity. No instances of this kind have
been found in Mikes’ How to be a Brit. The analysis of all other possible sources of
humour and irony, as exemplified in the work, follows.

a. Resolution of ambiguities or disambiguation

Ambiguity resolution or disambiguation is another of the inferential steps taken
by the addressee when deducing or trying to find out his interlocutor’s meaning.
It has to do with both the recovery of the explicatures of an utterance and with the
selection of the propositional form intended to be communicated by the speaker.
Within relevance theory, it is shown that this is a process guided by the expecta-
tions of relevance entertained by the addressee, on the basis of his access to the
contextual evidence available. In Mikes’ How to be a Brit this can be illustrated in
passages such as the following:
 Chapter 12. Humour and irony in George Mikes’ How to be a Brit 341

(3) It is so much nicer to ask, when someone speaks of Barbados, Banska

Bystrica or Fiji: ‘Oh those little islands … Are they British?’ (They usually
are). (“How to be an alien”: ‘How Not to be clever’, p. 44).

The text shows the speaker making fun of the British cosmovision. Humour is
reinforced through the final understatement, which makes it clear the speaker’s
point of view as well as her interlocutor’s attitude. Either a positive or a negative
answer might have been possible, and therefore, the addressee is requested to
entertain two possible scenarios. In the context, it is only the positive answer that
enhances and reinforces certain stereotypical assumptions generally entertained
about the British. In the end, it is this one that is recovered in the context and that
gives way to humorous effects.
Therefore, the British individual is shown to find it necessary to see his as-
sumptions confirmed, regarding whether those islands are British or not. He is
depicted as taking it for granted that they are indeed British. His own cosmovision,
and having it confirmed, seems to be the only thing that matters to him.
The question put forward is in this sense ultimately rhetorical, since a positive
answer was expected (at least from the point of view of the British, in accordance
with the author’s assumptions). The possible ambiguity concerning whether “those
little islands” were British or not is thus eventually disentangled and spelled out.

b. Reference assignment
Within relevance theory, reference assignment is one of the subtasks involved
in the identification of a propositional form. Its function is to make it possible
that pronouns and other deictic expressions are associated with the conceptual
entity they refer to.
(4) Bargaining is a repulsive habit; compromise is one of the highest human
virtues – the difference between the two being that the first is practised on
the Continent; the latter in Great Britain.
 (“How to be an alien”: ‘How to compromise’, p. 48, my italics).

As is well known, the contrast between Britain and the rest of the European
Continent is the major theme of Mikes’ How to be a Brit. No matter if the underly-
ing message seems to be that what is being criticised about the British can so be
done about anybody else, the presumed differences between the British and the
rest of the world are enhanced. These contrasts are exaggerated and often give way
to humour and irony. It becomes, therefore, very important to determine which is
the referent in each case.
The important thing here is to make it clear what the following terms point
at: namely, the two (corresponding to bargaining and to compromise), the first
342 Maria Angeles Ruiz-Moneva

(alluding to bargaining) and the latter (associated with compromise). Moreover,

this is exploited for humorous and also for ironic purposes, since it is quite unclear
that there may be any difference between these two concepts, despite the contrasts
claimed to exist between them, on the one hand, and between Britain and the rest
of Europe, on the other hand.

c. Enrichment or filling of a semantic gap

In natural language, utterances are always semantically incomplete. Depending
on the context envisaged by the speaker, and most importantly, on the context ac-
cessed by the addressee, and on whether these two contexts match, in interpreting
the message the addressee will tend to enrich the logical form of the utterance and
reduce its indeterminacies. This is done in the process of searching for a relevant
interpretation of the utterance, so that it produces optimal contextual effects with-
out demanding too large processing effort. As noted by Clark (2013, p. 179), with a
view to satisfying expectations of relevance, the addressee may set out to flesh out
the explicit content of an utterance even beyond the level of the full proposition
which can be either true or false:
(5) I heard of a distinguished, pure-minded English publisher who adapted John
Steinbeck’s novel The Grapes of Wrath, so skilfully that it became a charming
little family book on grapes and other fruits, with many illustrations.
 (“How to be an alien”: ‘A word on some publishers’, p. 36, my italics).

The process of enrichment may lead to the creation or entertainment of assump-

tions which may have been unexpected, and the result may be eventually humor-
ous. This can be the case of what happens in the example above, where the title of
the book by Steinbeck has been further amplified wittily to criticise the editorial
tasks performed by some publishers. This aspect is moreover the main target of
criticism intended to be made manifest by the author in ironic and satirical terms.

d. Derivation of implicatures
In relevance-theoretic terms, implicatures are defined as assumptions ostensively
communicated which are reached exclusively through processes of pragmatic
inference. Implications are conclusions that can logically be extracted from certain
premises or assumptions. In turn, implicatures are intentionally communicated
implications (Clark, 2013: 216). There are two basic types of implicatures, namely,
implicated premises and implicated conclusions. In relevance theory, it is consid-
ered that the addressee must derive such implicatures so as to infer the meaning
intended by the speaker. These may be exploited for humorous effects, so that
absurd or at least unexpected conclusions may follow from the evidence provided.
Let us illustrate how this works in Mikes’ How to be a Brit:
 Chapter 12. Humour and irony in George Mikes’ How to be a Brit 343

(6) Q: Why don’t they [the British] work harder?

A: They just don’t like hard work. The Germans have a reputation for hard
work, so they like to keep it up. The British find it boring. Then, apart
from a tiny and despicable minority, the British dislike the idea of taking
part in the rat race. They will give up certain advantages – knowingly
and with their eyes open – in order to be able to stick to certain values
and a way of life.  (“How to be decadent”: ‘How to panic quietly’, p. 237).

The passage above is inserted in a chapter entitled “How to be decadent” dealing

with some studies that were undertaken by foreign media which, according to
Mikes, sought to identify the reasons for what he terms as the English disease,
namely, the lack of any worries, concern or even despair or panic on the part of
the British after the loss of the Empire (and which contradicted the expectations
of those foreign newspapers or magazines). In it, therefore, Q refers to the inter-
viewer, who is a journalist working in any these foreign media, and A would reflect
the perspective of Mikes or the protagonist of How to be a Brit.
In this fragment, the speaker (that is, Mikes or the protagonist) follows a rea-
soning process in which the encyclopaedic entries conventionally attached to the
British and the German ways of life, respectively, are shown. While it might have
seemed that it was the German style that was supposed to be praised, the British
are depicted as being reluctant to abandon their customs and habits.
The passage can be read in terms of a premise-conclusion syllogism: thus, the
British (so does Mikes seem to suggest) fail to work harder because they do not
like to do so. While Germans want to live up to the expectations that have been
created and are commonly entertained regarding them, it seems that in contrast
the British do not seem to care about what other people may think about them.
They just seem to stick to what might have been called as ‘the British way of life’. In
the end, probably the two are just stereotypes, which may be quite contradicted by
individuals of either nationality.

e. Irony and humour resulting from a contradiction of expectations

Some of the main sources for humour in the work have to do with the perspective
or point of view adopted. It is often reflected as a disruption of expectations, which
may be expressed in different ways, such as the conditionals whose fulfilment is
impossible, but which are worded as only improbable or unlikely. Some rhetorical
figures also enhance and strengthen the intended interpretation, thus guiding the
reader towards it and its relevance.
In relevance-theoretic terms, the disruption of expectations leading to hu-
mour is inscribed in the two-stage processing of utterances, following a process
344 Maria Angeles Ruiz-Moneva

of mutual parallel adjustment (Clark, 2013, p. 144).8 In the case of humorous

utterances, this process of mutual parallel adjustment is explained in this way:
“If the utterance does not satisfy these expectations, a search for a more relevant
(and humour-connoted) interpretation worth being processed may be activated,
despite this supplementary mental effort” (Yus, 2004 p. 324).9 This author also
explains how humourists tend to manipulate information which is accessible
or salient for the audience, so that certain implications of (humorous) utter-
ances may be backgrounded and even irrelevant in certain contexts, but surely not
so in different ones.
In the example that follows this is the case with the prosopopeias or personi-
fications:
(7) The British meteorologists forecast the right weather – as it really should
be – and then these impertinent little anti-cyclones interfere and mess up
everything.
That again proves that if the British kept to themselves and did not mix with
foreign things liked Polar and Azores anti-cyclones they would be much
better off. (“How to be an alien”: ‘The Weather’, p. 29).

In our view, humour in the passage above, in the first place, may be said to rely
on the contrast between, on the one hand, the right weather, which actually does
not occur in real life, because of the existence of anti-cyclones, and, on the other
hand, what the weather is actually like. Therefore, the ‘right’ weather is not found,
because such a potential state of affairs is somehow contradicted, or not actualised
in real life. But at the same time, a further antagonism seems to be established
between the British and foreign things, here exemplified by anty-cyclones. It is
true that the author refers to the latter as foreign things, but yet he seems to place
these on an equal foot with the British.
Not only does the author mock the British and the ways in which they perceive
themselves, or the ways how others (the foreigners, including the author) regard
them. Instances of self-addressed irony or humour can also be traced. They may
show the author’s (hence, the speaker’s) perspective towards what he is narrating.
They may be expressed in the form of understatements and may reflect ambigu-
ity, contradiction, or even absurdities. This is so because the conclusions cannot

8. Within the relevance-theoretic framework, the process of mutual parallel adjustment accounts
for the tasks undertaken by the addressee when trying to understand the message communi-
cated by the speaker. These tasks have to do with the working out of explicatures, of implicated
premises and also of implicated conclusions. These processes are worked out in parallel.

9. Similar claims had previously been put forward by Jodłowiec (1991).

 Chapter 12. Humour and irony in George Mikes’ How to be a Brit 345

be logically derived from the assumptions entertained, which are therefore what
Curcó (1998, 1997b) refers to as the key assumptions:
(8) I don’t quite know what anti-cyclones are, but this is not important; I hate
cyclones and I am very anti-cyclone myself.
 (“How to be an alien”: ‘The weather’, p. 29).

Thus, it seems quite contradictory to claim, as the author does, that on the one
hand, he does not really know what an anticyclone is, and then to identify himself
as one, on the other hand. Curcó introduces the term key assumptions in order to
refer to a strongly implicated premise in humorous speech or propositions. This
key assumption is generally retrieved from background knowledge (in the ex-
ample above, it would correspond to the concept of anticyclone). This proposition
clashes with another assumption, which is made in the closest or most immediate
context where the interpretation is supposed to take place. The latter assumption
is termed by Curcó as the target assumption, which is the one supposed to have
been intended to be communicated by the speaker and which results in humorous
interpretations (above, the fact that the speaker regards himself as an anti-cyclone,
… even though he does not know what anticyclones are, … and even though they
are hardly attributable to human beings, as a matter of fact).

4. Conclusions

George Mikes’ How to be a Brit is a work where irony and humour play an essential
role in the creation not only of meaning, but also of style. Therefore, they have
been found to guide the reader in the search for relevance: thus, the analysis has
shown that the different phases followed by him in the interpretive process have
been exploited and manipulated for the creation of humour and irony. Moreover,
neither humour nor irony represents any substantial or qualitative departure from
the interpretation of any utterance.
On the one hand, irony is connected to the speaker’s feelings, intention and at-
titude. This is so because his attitude may be characterised by remarkable distance
that he adopts from the target of his criticism. The protagonist of Mikes’ How to
be a Brit depicts himself as an outsider, as an alien, and adopts a perspective of
distance – and sometimes of a certain superiority, or more precisely, detachment
– from his intended targets of criticism.
On the other hand, humour can be related to the writer’s central purpose of
mocking or satirising those aspects of British life that he has found remarkable in
some sense. On the whole, irony may be seen as being at the disposal of humour,
as a form of reinforcement of the conveyance of the author’s communicative
346 Maria Angeles Ruiz-Moneva

purposes. Thus, irony and humour may be seen as complementary and reinforcing
each other in the author’s creation of style.
The analysis of Mikes’ How to be a Brit, based on the relevance-theoretic ap-
proach to communication, shows that irony is essentially echoic. This means that
irony has two fundamental traits: first, it tacitly expresses the speaker’s attitude to
the beliefs, thoughts or ideas being represented; and second, the attitude involved
in the echo corresponds to the speaker’s dissociation from the thoughts echoed.
In the book, this is seen in the standpoint adopted by the protagonist who, be-
ing characterised as an ‘alien’, as a foreigner, amongst strange people, manages to
unmask their contradictions. No matter if his attitude is of distance or dissocia-
tion, some of the aspects criticised may be applicable to many other people and
even to himself, so that he gradually becomes identified with the object or target
of his criticism.
As for humour, it has been shown that in Mikes’ How to be a Brit, it is often
based on the contradiction or twist of expectations that lead the reader to revise
an initially more accessible interpretation. That interpretation is reached follow-
ing the inferential procedures which are applicable to any kind of discourse. This
indicates, therefore, that, as predicted by relevance theory, the same mechanisms
are involved in the processing of both literal and of figurative meaning (the lat-
ter being exemplified in this essay by humour and irony). In this way, the reader
tends to select the interpretation that is most accessible in a certain context. Yet,
the broadening or extension of such context often tends to favour or foreground
an alternative interpretation, which may have been initially undermined or
backgrounded, and which results in humour. The analysis has revealed that, as
claimed by Yus (2016a, 2004, 2003, 1995–96), the same inferential strategies at
work in the processing of everyday speech can be applied to the interpretation
of humorous discourse.
Recurrent traits in the analysis of irony and humour have also been found,
which can be accounted for from a relevance-theoretic perspective. Thus, the
reading of Mikes’ How to be a Brit also provides evidence for the consideration of
both irony and humour as echoic uses of language. In the work, the speaker – who
may be said to represent the ‘real’ or ‘historical’ author Mikes – distances himself
from the British, and does so by offering the reader a somehow distant, detached
picture of such target of criticism. Yet, as shown above, the ambiguity characteristic
of irony and humour allows Mikes to hold a much more complex and ambivalent
attitude towards the British, which embraces both ironic distance and affection.10

10. Moreover, this is congruent with the most recent tendencies in the study of irony within a
relevance-theoretic perspective (most notably, Yus 2018, 2016a, 2016b, 2016c). This author has
highlighted that irony involves not only a propositional attitude of echoic distance (as has been
 Chapter 12. Humour and irony in George Mikes’ How to be a Brit 347

The analysis carried out also confirms the essential connections between
irony and humour from a relevance-theoretic perspective pointed at by Piskorska
(2014), namely, the exploitation of incongruity; their being targeted at a victim;
the exploitation of creativity by both irony and humour; an element of surprise
springing from the resolution of the words or situation that has led to irony,
humour or both; the addressee’s need to rely on metarepresentational abilities to
cope with the meaning expressed by irony, humour or both; or the establishment
of a certain bond between the speaker and the target addressee.
Irony has been found to be used as one of the vehicles or instruments enabling
the author to convey humour as intended to be communicated by the speaker. All
in all, the analysis has shown that humour and irony do not demand any specific
processing mechanisms or routes. The two can be approached as aspects of style,
and as such they guide the reader’s search for relevance. As a result, their un-
derstanding follows the same processes as any other utterance. These interpretive
processes are then manipulated for the creation of irony and humour.
Neither irony nor humour is therefore a property of a certain text. Rather,
they represent aspects of the addresser’s communicative intention; they are also
reflected in the style of the text, and, as noted above, as such guide or constrain
the addressee’s search for relevance and the recovery of the message intended
to be communicated. It is ultimately left to the addressee to interpret a piece
of discourse as containing irony or humour, or both, and it is precisely the ad-
dressee’s engagement and search for his own interpretation that will make irony
and humour most enjoyable.

Acknowledgements

The author wants to express her gratitude for the comments, suggestions and indications put
forward by several anonymous referees on a previous draft of this paper.

References

Attardo, S. (2000). Irony as relevant inappropriateness. Journal of Pragmatics, 32.6, 793–826. ​

https://‍doi.org/10.1016/S0378-2166(99)‍00070-3
Attardo, S. (2001a). Humorous texts: A semantic and pragmatic analysis. Berlin: Mouton de
Gruyter.  ​https://‍doi.org/10.1515/9783110887969

recurrently claimed within relevance theory). It may also involve a second-order affective attitude
that the speaker may show towards the source of his irony, or towards his intended addressees.
348 Maria Angeles Ruiz-Moneva

Attardo, S. (2001b). Humor and irony in interaction: From mode adoption to failure of detection.
In L. Anolli, R. Ciceri, & G. Riva (Eds.), Say not Say: New perspectives on miscommunication
(pp.165–185). Amsterdam: IOS Press.
Attardo, S. (2013). Intentionality and irony. In L. Ruiz Gurillo & B. Alvarado Ortega (Eds.), Irony
and humor. From pragmatics to discourse (pp. 39–57). Amsterdam: Benjamins. ​
https://‍doi.org/10.1075/pbns.231.04att
Attardo, S., & Raskin, V. (1991). Script theory revisited: joke similarity and joke representation
model. Humor, 4.3–4, 293–347.
Biegajlo, M. (2014). From which position should I get this joke?! A relevance-driven joke inter-
pretation: Naive optimism, cautious optimism, sophisticated understanding. International
Studies in Humour, 3(1), 2–14.
Blakemore, D. (1992). Understanding Utterances. Oxford: Blackwell.
Booth, W. (1974). A rhetoric of irony. Chicago and London: The University of Chicago Press.
Carston, R. (1980). Approaches to verbal irony (MA Thesis). London: University College.
Carston, R. (1981). Irony and parody and the use-mention distinction. Nottingham Linguistic
Circular, 10, 24–35.
Carston, R. (1997). Relevance-theoretic pragmatics and modularity. UCL Working Papers in
Linguistics 9: 29-53.
Carston, R. (2002a). Thoughts and utterances. Oxford: Blackwell. ​
https://‍doi.org/10.1002/9780470754603
Carston, R. (2002b). Metaphor, ad hoc concepts and word meaning – More questions than
answers. UCL Working Papers in Linguistics 14, 83–105.
Carston, R. & Uchida, S. (Eds.) (1998). Relevance Theory. Applications and Implications. Am-
sterdam: Benjamins.
Clark, B. (2013). Relevance theory. Cambridge: Cambridge University Press. ​
https://‍doi.org/10.1017/CBO9781139034104
Curcó, C. (1995). Some observations on the pragmatics of humorous interpretations. A
relevance-theoretic approach. UCL Working Papers in Linguistics 7, 27–47.
Curcó, C. (1996). The implicit expression of attitudes, mutual manifestness and verbal humour.
UCL Working Papers in Linguistics 8: 89–99.
Curcó, C. (1997a). The Pragmatics of Humorous Interpretations: A Relevance-Theoretic Account.
Unpublished PhD. Dissertation. University College London.
Curcó, C. (1997b). Relevance and the manipulation of the incongruous: Some explorations
of verbal humour. In M. Groefsema (Ed.), Proceedings of the University of Hertfordshire
relevance theory workshop (pp. 68–72). Chelmford: Peter Thomas and Associates.
Curcó, C. (1998). Indirect echoes and verbal humour. In V. Rouchota, & A. H. Jucker (Eds.),
Current issues in relevance theory (pp. 304–325). Amsterdam: Benjamins. ​
https://‍doi.org/10.1075/pbns.58.13cur
Furlong, A. (1996). Relevance theory and literary interpretation. PhD. Thesis. University of
London.
Gibbs, R. W. (1994). The poetics of the mind: Figurative language, thought and understanding.
Cambridge: Cambridge University Press.
Gibbs, R. W. (2001). Evaluating contemporary models of figurative language understanding.
Metaphor and Symbol, 16, 317–333.  ​https://‍doi.org/10.1080/10926488.2001.9678900
Gibbs, R. W. (2005). Irony as persuasive communication. In H. L. Colston, & A. N. Katz (Eds.),
Figurative Language Comprehension. Social and Cultural Influences (pp.131–151). Mahwah:
L.E.A.
 Chapter 12. Humour and irony in George Mikes’ How to be a Brit 349

Gibbs, R. W., & Colston, H. L. (2002). The risks and rewards of ironic communication. In
L. Anolli, R. Ciceri, & G. Riva, Say not to Say: New Perspectives on Miscommunication
(pp. 181–194). Amsterdam, Netherlands: IOS Press.
Hatim, B. (1997). Communication across cultures. Translation theory and contrastive text linguis-
tics. Exeter: University of Exeter Press.
Jodłowiec, M. (1991). What makes jokes tick. UCL Working Papers in Linguistics, 3, 241–253.
Mikes, G. (1986 [1946–1977]). How to be a Brit. A George Mikes minibus. Harmondsworth:
Penguin.
Muecke, D. C. (1969). The Compass of irony. London: Methuen.
Muecke, D. C. (1970). Irony. London: Methuen.
Muecke, D. C. (1982). Irony and the ironic. London: Methuen.
Pilkington, A. (2000). Poetic effects. Amsterdam & Philadelphia: John Benjamins. ​
https://‍doi.org/10.1075/pbns.75
Piskorska, A. (2014). A relevance-theoretic perspective on humorous irony and its failure.
Humor 274, 661–685.
Piskorska, A. (2016). Echo and inadequacy in ironic utterances. Journal of Pragmatics 101,
54–65.  ​https://‍doi.org/10.1016/j.pragma.2016.05.007
Raskin, V. (1985). Semantic mechanisms of humor. Dordrecht: Reidel.
Rodríguez R. S. (2013). “The power of inversion. Irony, from utterance to discourse.” In L. Ruiz
Gurillo, & B. Alvarado Ortega (Eds.), Irony and humour. From pragmatics to discourse
(pp. 17–38). Amsterdam & Philadelphia: Benjamins. ​
https://‍doi.org/10.1075/pbns.231.03rod
Ruiz Gurillo, L., & B. Alvarado Ortega (Eds.) (2013). Irony and Humour. From pragmatics to
discourse. Amsterdam/Philadelphia: Benjamins.
Sperber, D. (1996). Explaining culture. A naturalistic approach. Oxford: Blackwell.
Sperber, D., & Claidière, N. (2006). Defining and explaining culture. (comments on Richerson
and Boyd Not by genes alone). Reading available at Cultural Evolution Workshop. Washing-
ton University in St. Louis.
Sperber, D., & Wilson, D. (1978). Les ironies comme mentions. Poétique, 36, 399–412.
Sperber, D., & Wilson, D. (1981). Irony and the use-mention distinction. In P. Cole (Ed.), Radi-
cal Pragmatics (pp. 295–318). New York: Academic Press.
Sperber, D., & Wilson, D. (1995). Relevance. Communication and cognition, 2nd ed. Oxford:
Blackwell
Torres S. (1999). Estudio pragmático del humor verbal. M.A. thesis. Cadiz: University of Cadiz.
Wilson, D. (1999). Metarepresentation in linguistic communication. UCL Working Papers in
Linguistics, 11, 127–162.
Wilson, D. (2006). The pragmatics of verbal irony: Echo or pretence? Lingua, 116, 1722–1743. ​
https://‍doi.org/10.1016/j.lingua.2006.05.001
Wilson, D. (2009). Irony and metarepresentation. UCL Working Papers in Linguistics, 21,
183–226.
Wilson, D. (2011). Relevance theory and the interpretation of literary works. UCL Working
Papers in Linguistics, 23, 69–80.
Wilson, D. (2013). Irony comprehension: A developmental perspective. Journal of Pragmatics,
59, 40–56.  ​https://‍doi.org/10.1016/j.pragma.2012.09.016
Wilson D., & Sperber, D. (2004). Relevance theory. In L. R. Horn, & G. Ward (Eds.), The hand-
book of pragmatics (pp. 607–632). Oxford: Blackwell.
350 Maria Angeles Ruiz-Moneva

Wilson D., & Sperber, D. (2012). Explaining irony. In D. Wilson, & D. Sperber Meaning and
relevance (pp. 123–145). Cambridge: Cambridge University Press. ​
https://‍doi.org/10.1017/CBO9781139028370.008
Yus, F. 1995–96. La teoría de la relevancia y la estrategia humorística de la incongruencia-
resolución. Pragmalingüística, 3–4, 497–508.
Yus, F. (1997–98). Irony: context accessibility and processing effort. Pragmalingüística, 5–6,
391–410.
Yus, F. (1999a). Misunderstandings and explicit/implicit communication. Pragmatics, 9 (4),
487–517.  ​https://‍doi.org/10.1075/prag.9.4.01yus
Yus, F. (1999b). Towards a pragmatic taxonomy of misunderstandings. Revista Canaria de
Estudios Ingleses, 38, 217–239.
Yus, F. (2000). On reaching the intended ironic interpretation. International Journal of Com-
munication, 10 (1–2), 27–78.
Yus, F. (2000–01). Literal/ non-literal and the processing of verbal irony. Pragmalingüística, 8–9,
349–74.
Yus, F. (2003). Humor and the search for relevance. Journal of Pragmatics, 359, 1295–1331. ​
https://‍doi.org/10.1016/S0378-2166(02)‍00179-0
Yus, F. 2004. Pragmatics of humorous strategies in El club de la comedia. In R. Márquez-Reiter, &
M. E. Placencia (Eds.), Current trends in the pragmatics of Spanish (pp. 320–344). Amster-
dam: Benjamins.  ​https://‍doi.org/10.1075/pbns.123.25yus
Yus, F. (2009). Saturación contextual en la comprensión de la ironía. In L. Ruiz Gurillo, & X.
Padilla García (Eds.), Dime cómo ironizas y te diré quién eres. Una aproximación pragmática
a la ironía (pp. 309–331). Frankfurt: Peter Lang.
Yus, F. (2013). An inference-centered analysis of jokes. The intersecting circles model of humor-
ous communication. In L. Ruiz Gurillo, & B. Alvarado Ortega (Eds.), Irony and humor.
From pragmatics to discourse (pp. 59–82). Amsterdam: Benjamins. ​
https://‍doi.org/10.1075/pbns.231.05yus
Yus, F. (2016a). Humour and relevance. Amsterdam and Philadelphia: John Benjamins. ​
https://‍doi.org/10.1075/thr.4
Yus, F. (2016b). Relevance theory and contextual sources-centred analysis of irony. Current
research and compatibility. In M. Padilla Cruz (Ed.), Relevance theory: Recent developments,
current challenges and future directions (pp. 147–171). Amsterdam & Philadelphia: John
Benjamins.  ​https://‍doi.org/10.1075/pbns.268.06yus
Yus, F. (2016c). Propositional attitude, affective attitude and irony comprehension. Pragmatics &
Cognition, 231, 92–116.  ​https://‍doi.org/10.1075/pc.23.1.05yus
Yus, F. (2016d). Incongruity-resolution and the explicit-implicated-implied interface. Paper
delivered at Linguistic Approaches to Funniness, Amusement and Laughter. 4th International
Symposium: Theoretical Issues in Humour, Building Bridges across Disciplines. University
of Lodz (Poland), March.
Yus, F. (2017a). Relevance-theoretic treatments of humor. In S. Attardo (Ed.), Routledge hand-
book of language and humor (pp. 189–203). Abingdon: Routledge.
Yus, F. (2017b). Incongruity-resolution cases in jokes. Lingua, 197, 103–122. ​
https://‍doi.org/10.1016/j.lingua.2017.02.002
Yus, F. (2018). Attaching feelings and emotions to propositions. Some insights on irony and
Internet communication. Russian Journal of Linguistics 22(1), 94–107. ​
https://‍doi.org/10.22363/2312-9182-2018-22-1-94-107
 Name index

A Blutner, R. 26 Clark, H. H. 312

Adamczyk, M. 265 Bolinger, D. 175 Clément, F. 83
Adams, A. 187 Booth, W. 337 Collins, P. 78
Adamson, S. 292 Borg, E. 52 Colston, H. L. 1, 5–6, 55,
Addison, J. 263 Boring, E. G. 277 284, 335
Aikhenvald, A. 81, 123–124 Börjars, K. 78 Cooper, G. 159
Alderson, S. J. 265 Bornstein, R. F. 198 Corazza, E. 52
Allan, K. 38 Boswell, J. 261 Cornillie, B. 81
Allott, N. 27 Bottomley, P. A. 198 Corriveau, K. 83
Alrasheedi, E. S. 271 boyd, d. 174 Croft, W. 48
Alvarado Ortega, B. 329 Brdar, M. 47–48 Croom, A. M. 247, 254
Apresjan, J. D. 46 Brdar-Szabo, R. 47–48 Crystal, D. 81, 168, 171, 175,
Arndt, W. F. 96 Bredart S. 59 183, 186
Arundale, R. B. 71 Breheny, R. 181 Culpeper, J. 75
Athanasiadou, A. 1 Brown, A. 230 Curco, C. 310–311, 316,
Attardo, S. 266, 229–331 Brown, C. H. 26 328–330, 332–333, 345
Attridge, D. 61, 281, 283 Brown, D. F. 72 Czarnecki, G. 231, 233, 235,
Austin, J. L. 71, 235 Brown, P. 71 237, 239
Bryant, G. A. 174
B Buchanan, R. 52 D
Bach, K. 49, 147 Buddruss, G. 134, 136, 159 Danker, F.W. 96
Baker, C. 299 Burridge, K. 78 Davis, S. 35
Bakker, S. 96 Butler, J. 235, 238 De Jong, I. 98, 105–106, 112,
Barcelona, A. 47, 48 Butterick, M. 167, 169–170, 116
Barsalou, L. 3, 181 212 de Mul, S. 59
Barton, S. 59 Degener, A. 134
Bashir, E. 125 C Dendale, P. 81
Bauer, W. 96 Calhoun, S. 182 Denniston, J. 96
Becker-Matero, N. 198 Cara, F. 127 Díaz-Cintas J. 199
Benesch, S. 233, 237, 252 Carpenter, P. A. 213 Diedrick, J. 312
Bennett, W. 298–300, 303 Carston, R. 2, 5–6, 12, 25, 26, Dobrushina, N. 141
Berger, S. 172 28, 29, 31, 45, 49, 52, 55–56, Doherty, S. 196
Bianchi, C. 230, 239, 247 73, 75, 86, 168, 179, 274, 328, Doke, C.M. 125
Biegajlo, M. 328 332–333, 339–340 Dokic, J. 52
Bilewicz, M. 231, 233, 235, 239, Carter, R. 299–300 Dor, D. 266
241, 244–245 Chappell, H. 125 Doyle, J. R. 198
Black, S. 96 Chovanec, J. 266–268 Drury, J. E. 172
Blakemore, D. 9–10, 75, 86, 105, Cirillo, L. 175 Dryden, J. 262
116, 122, 128, 168, 178–179, Clark, B. 19, 130, 147, Dynel, M. 266, 272, 280
197, 202–203, 205, 208, 239, 174–175, 178, 181–182, 201–
247–251, 253, 333, 340 202, 291, 327–328, 332, E
Blass, R. 76, 97, 99, 102 339–340, 342, 344 Edworthy, J. 187
352 Relevance Theory, Figuration and Continuity in Pragmatics

Egermann, H. 181 Hasan, R. 201, 208 L

Eizaga Rebollar, B. 252 Haspelmath, M. 131 Lakens, D. 194
Ellsworth, P. C. 18 Hatim, B. 337–338 Lakoff, G. 1, 26, 35, 40
Empson, W. 283 Haugh, M. 74 Lamb, C. 283
Escandell Vidal, V. 74, 178 Hedley, P. 178 Lampert, M. 175
Evans, V. 35 Henderson, P. W. 198 Langacker, R. W. 1, 35, 48
Heyman, G. D. 83 LaPolla, R. J. 128
F Hirschberg, J. 233 Lass, R. 182
Falkum, I. L. 8, 17, 39, 46, 61 Hollo, C. 78 Leech, G. N. 244, 261
Faller, M. 124 Holmes, J. 72 Leets, L. 233–239, 244–245
Fauconnier, G. 26, 47 Holmes, O. W. Sr. 265 Leitner, G. W. 159
Ferreira, F. 59 Holmesland, O. 299, 301 Lemmens, M. 36, 39
Filik, R. 174 Hom, Ch. 237, 247–249, 252, Levine, D. N. 265
Fitch, M. 159 254 Levinsohn, S. 96
Fodor, J. D. 172 Hopper, P. J. 126 Levinson, S. C. 71, 74
Fox Tree, J. E. 174 Horn, L. 26 Levy, R. 182
Fretheim, T. 178 House, J. 171, 173, 180 Liberman, M. 37
Fricker, M. 82 Huang, Y. 26–27 Linde-Usiekniewicz, J.
Frijda, N. H. 18 Huddleston, R. 69, 77–78, 81 235–236, 240, 245, 248, 253
Furlong, A. 332 Huron, D. 181–182 Litt, E. 174
Littlemore, J. 47
G I Locher, M. A. 71
Gerrig, R. J. 312 Ifantidou, E. 9, 39, 70, 75–76, Lodge, D. 299–301, 305, 314
Gibbs, R. W. 55, 284 82, 84, 86, 128, 130 Lord, C. 125
Giles, H. 244 Imai, K. 75, 178 Lubberger, B. 123, 131, 133–134
Gingrich Wilbur, F. 96 Itani, R. 76, 84, 130 Ludlow, P. 26, 33
Giora, R. 266
Girotto, V. 127 J Ł
Goddard, A. 266 Jackson, R. 172, 178, 184, 202 Łaziński, M. 254
Goldman, A. 97 Jakobsen, R. 239
Goussev, V. 141 Jameson, A. 213 M
Graesser, A. 59 Jarniewicz, J. 314, 317–318, 322 Macaya, M. 187
Green, G. 201 Jary, M. 74 Mahood, M. M. 265
Green, M. 35 Jodłowiec, M. 5, 8, 36, 47, 50, Maitra, I. 59
Greenbaum, S. 69, 77, 81 52, 328, 332, 344 Manes, J. 74
Grice, H. P. 49, 72, 267–268, Johanson, L. 124 Marandin, J.-M. 71–72
293–294, 301, 312, 328 Johnson, M. 1, 47–48, 54 Martínez-Manrique, F. 52
Gross, J. 172 Jones, R. F. 265 Marwick, A. E. 174
Guéron, J. 78 Jucker, A. 75 Mascaro, O. 82
Güldemann, T. 125–126 Just, M. A. 213 Mason, G. 233, 237–238, 253
Gussenhoven, C. 186 Mateo, J. 239
Gutt, E.-A. 6, 130 K Mathews, W. 265
Kankaanranta, A. 174 Matthews, P. H. 81
H Kaplan, A. 266 Mazzarella, D. 74, 239
Haegeman, L. M. V. 78 Katayama, N. 199 McArthur, T. 266
Hagopian, J. V. 299 Kimura, T. 195 McDonald, N. 265
Hall, A. 3, 28, 36 Koenig, M. 83 Meier, B. P. 198
Hallberg, C.E. 123, 159 Kövecses, Z. 47–48, 54 Mercier, H. 83
Hallberg, D. G. 123, 159 Kowalski, S. 234, 237, 239 Meyer, Ch. F. 202
Halliday, M.A.K. 199–201, Kronenfeld, D. 26 Meyers, C. 76
208 Kusbit, G. 59 Michaelian, K. 83
Harnish, R.M. 147 Kwok I. 233, 235 Modolo, K. 59
Harris, P. 83 Moeschler, J. 75
 Name index 353

Moors, A. 18 Q Steinbeis, N. 181

Motion, A. 313, 323 Quirk, R. 69, 77, 81 Stökl, H. 199
Muecke, D. C. 337 Stovall, J. G. 76
R Strey, C. 18
N Radden, G. 47–48, 54 Stubbs, M. 299–302, 304–305
Natsopoulos, D. 59 Raskin, V. 330 Swarbrick, A. 323
Necker, L. A. 276–277 Read, S. 283 Sweep, J. 56
Nicolle, S. 236, 239 Recanati, F. 49 Swift, J. 265
Nijałkowski, L. M. 234, 238, Reder, L. M. 59 Szymańska, K. 318, 321
244, 246 Redfern, W. 279
Nikiforidou, K. 292 Reinhart, T. 201, 208 T
Noels, K. 244 Remael, A. 199 Tagg, C. 174
Noh, E-J. 128, 130, 147, 209 Rich, C. 76 Takahashi T. 199
Nørgaard, N. 199 Rodríguez R. S. 330 Tanaka, K. 266, 271, 284
Norrick, N. R. 266 Rojczyk, A. 264, 266, 271, 273 Tasmowski, L. 81
Noveck, I. 55 Rosa, E. 187 Taylor, J. R. 26
Nuyts, J. 81 Rouchota, V. 75 Thompson, D. 174
Rubio-Fernández, P. 29 Thornburg, L. 47–48
O Ruiz Gurillo, L. 329 Thrall, M. 99, 112, 116
O’Brien, J. 5 Runge, S. 96, 99 Torres, S. 328
O’Brien, S. 213 Traugott, E. C. 37–38, 126
O’Hagan, M. 171, 196 S Trotter, D. 74
Ohnuma, M. 299, 301–303 Sadokierski, Z. 170–171 Trousdale, G. 37
Origgi, G. 82–84 Sanford, A. 59 Tulli, M. 234, 234, 237, 239
Ortega y Gasset, J. 283 Sasamoto, R. 171–172, 196 Turner, M. 26, 47
Ortony, A. 259 Saxena, A. 125 Twyman, M. 169, 169,
Oswald, S. 84 Scherer, K. 18 170–171, 182
Schultheis, H. 213
P Scott, K. 9, 168, 174–175, U
Padilla Cruz, M. 14, 73, 75, 178–179, 182, 184, 202 Uchida, S. 15, 19, 291–293,
84, 271, 279 Seargeant, P. 174 295, 305, 340
Palmer, F. R. 81, 123 Searle, J. 71 Unger, C. 9, 39, 84, 103, 113,
Panther, K-U. 47–48 Sebeok, T. 174 124, 128–130, 135
Papafragou, A. 8, 61, 130 Sedivy, J. C. 181
Park, H. 59 Sheridan, T. 265 V
Patson, N. 59 Sherry, R. 261 Van Berkum, J. 182
Pearce, M. T. 181 Shiota, E. 195–196 Van der Auwera, J. 141
Pekkola, J. 172 Skovholt, K. 174 van Dijk, T. 225, 238
Pelli, D. G. 187 Slings, S. 96 Van Lancker Sidtis, D. 38
Perea, M. 187 Solska, A. 39, 239, 263–264, van Leeuwen, T. 169, 171–172,
Perrone-Bertolotti, M. 172 266, 271, 273, 275–277, 281 194, 199–200
Peters I. 46 Spencer-Oatey, H. 71 van Mulken, M. 266, 271
Peters, W. 46 Sperber, D. 2–4, 6, 9–10, van Noorloss, M. 230, 233
Pilkington, A. 291, 332–333 16–18, 25, 27–28, 30–34, Van Oostendorp, H. 59
Pinker, S. 60 45, 49–5153–56, 61–62, 70, Vega Moreno, R. E. 39, 46
Piskorska, A. 5, 8, 18, 36, 50, 72–76, 79, 82–84, 86, 97–98, Viscente, A. 52
80, 230, 239, 311–312, 314, 103, 112–115, 122, 124, 126,
328, 347 127–130, 147, 168, 177–178, W
Pollack, J. 265 184, 186–187, 196, 201, 204, Wałaszewska, E. 27–28, 32,
Popa, M. 5 208–209, 211, 238–240, 34, 46
Pullum, G. 37 243–246, 249, 270–271, Walker, S. 167, 169, 171, 183
Pustejovsky, J. 46 274–275, 291, 294, 310–311, Warde, B. 170–172, 188
313, 328, 332–333 Warner, W. 233
354 Relevance Theory, Figuration and Continuity in Pragmatics

Wearing, C. 6, 31 171, 174–175, 177–180, 184, Y

Wells, J. C. 174 186–187, 196, 201–202, Yao, B. 172
Wesch, M. 174 208–210, 229, 238–240, Yus F.; Yus Ramos, F. 14, 73,
Wharton, T. 9, 18, 75, 130, 242–246, 248, 253, 274–275, 202, 239, 266–267, 310,
168, 171, 173–176, 178, 180, 291, 294, 305, 310–313, 315, 328–329, 331–334, 337,
182, 184–185, 201, 208 319, 323, 328, 331–333 339–340, 344, 346
Wilson, D. 1–10, 12, 15–16, Winiarski, J. 312
19, 25–34, 36, 45–46, 49–51, Wolfson, N. 72, 74 Z
53–56, 61–62, 70, 72, 73–76, Wong, Y. 233, 235 Zakowski, S. 96
79–80, 82, 84, 86, 103, Wright, N.T. 112 Zegarac, V. 239
113–114, 122, 124–130, Zoller, C. P. 123, 159
135–136, 146–148, 168, Zwicky, A. 37
 Subject index

A see also argumentation cognitive 54–61, 127, 177,

adverbials communicator goals 95–96, 182, 240, 271, 274, 279
hearsay 76, 82, 87–89 114 cognitive, of strengthening
illocutionary 76, 86 complementizer 121–123, 100, 103, 111, 305, 333,
affect 83, 173, 186, 198, 125, 133 see also strengthening
see also emotions comprehension 6, 18, 49, humorous 329–330, 333
affective interpretation 54–55, 57–59, 70, 73, 88, 114, of contrastiveness 182
208–210 124, 127–128, 179, 274, 284 positive cognitive 176, 243,
affective response 18, 196 irony 311 245–246, 254, 275, 291
ambiguity 61, 259–260, comprehension heuristic/ pragmatic 6, 55, 201–204
263–266, 268, 281, 340–341 procedure 54–55, 127, 177, stylistic 115, 332
appropriation 239, 255 240, 275, 332 emotions 18, 208–210, 230,
argumentation 9, 84, 95, concept 239–240, 251, 261, 277,
97–103, 105, 111–118 ad hoc 3, 5, 27–29, 31–32, 36, see also affect
see also argumentative 38, 339 encyclopaedic entries 28
communication broadening 5, 25–31, 36, enrichment 45–46, 52, 340,
assumption 39, 203 342
contextual 17, 240 narrowing 5, 27–30 free 52
key 330, 345 non-lexicalized 28, 50, 53 epistemic stance 69
target 330, 345 connective 75, 95–96, 98–99, epistemic vigilance 69,
attitude 102, 105–108, 111–114, 116 82–86, 112–116, 124, 128–129
affective 346–347, 350 context 275, 281–282, 333–335 evidence 69–70, 76–77, 81–82
dissociative 4, 310–311, 313, accessibility 337 see also evidentiality,
333, 338 continuity 2–5 evidentials
propositional 70, 73, 124, contrast 168, 174–175, 178, evidential participles 69–70,
210–212, 310, 340, 346 180, 186 81–88
Critical Discourse Analysis evidentiality 81, 123–124
B 235 evidentials 9, 70, 82, 123–124,
belief 70, 72, 81, 87, 275, 279 128–130
intuitive 240 D expectations
reflective 240, 256 deflationary approach to hearer 176–178, 180–83,
believability 82, 97, 103, figurative language 2, 6–7 186–88, 273–274
113–114, 117 desirable utterances 122, 126, of relevance 2, 6, 19, 98,
130, 136–141, 146–160 114–117, 273–274, 276–283,
C disambiguation 340 295, 344
category extension 25–26, 28, explicature 49, 58–59, 73, 158,
30–40 E 274, 277, 294
coherence 97, 101, 112–113, echo/echoic use 2, 4–5, 130, higher–level 73, 86, 211–212,
115–116 239, 252, 254, 310–314, 319, 340
communication 330, 337, 346 eye-tracking 196, 213–217
affective 202, 204, 217 effect
argumentative 115, 330, contextual 305, 333
356 Relevance Theory, Figuration and Continuity in Pragmatics

F irony 2, 4–6, 16, 130, 174, N

figurative language 1–7, 9, 13, 309, 310–317, 327–335, narrative 95–99, 103–112,
17, 19, 20, 259, 284, 285, 327, 337–338 345–347 116–117
332, 348 functions of 330 (non)‍literalness 1–7, 28, 252

G J O
gar 95–117 jab lines 331 offensive terms 234, 237
General Theory of Verbal orthography 175
Humour 330–31 K
Gricean maxims 267–270 Koine Greek 95, 116, 118 P
paralinguistic information
H L 167–175
hate speech 229–254 lexical modulation parentheticals 76
covert 234, 239–247, 253 26–3033–34 paronomasia 261
direct 237, 253 literary texts, interpretation of perlocutionary act 80,
indirect 237, 254 14–15, 19, 291–293, 295–297 235–236
overt 229, 234, 237 locutionary act 235–236 personal pronouns 86, 128,
highlighting loose use 26, 28–32, 42 295–297
function of telops 196 persuasion 97–98, 112, 114–115
natural 184–188, 201–202 M polysemy 46, 263
hortatory material 98–99 marker pragmatic routines 39–40
humour 327–335 discourse 75, 86, 128, premises
and irony 329–335 130–135 explanatory 98
framing by telops 196, 199 of metarepresentation implicated 73, 240, 294, 342
131–135 procedural device 179–180
I procedural 95, 99–112 procedural encoding 122,
Illocutionary act 235–36 quotation 123–127 128–129, 249
implicated conclusion 294 meaning procedural expressions 86,
Implication conceptual 248 178
background 333 figurative 63, 285, 346 procedural instructions 86,
foreground 333 literal 2–3, 28, 327–328, 99, 111, 249
implicature 49, 268, 291, see also (non)‍literalness prosody 167–168, 171–175,
293–295, 301–302, 305 procedural 9, 75, 123, 128, 178–179, 182, 184, 186
Gricean 301–302 167, 176–179, 249 pun 61, 84, 197, 259–284
strong 243, 246, 293–297, reversal of 317 punch line 331
302–305 metaphor 2–6, 28–31, 35, punctuation 171–175
weak 203–204, 208–209, 186–187, 259, 283
293–297, 302–305 metarepresentation 122, Q
incitement 129–131, 331 quasi-slurs 246–251
to genocide 233, 252–253 non–attributive 146–148
to hatred 230–33 235–237 metonymy 8, 17, 45–62 R
incongruity 329, 334 modality 81, 123 rationality 97–98, 113
resolution of 331 deontic 81 reference assignment/
Indus Kohistani 121–160 epistemic 81 resolution 168, 186, 181,
inferential processes 2, 6, module 292, 295–297, 341–342
52–54, 179, 327, 340–345 argumentative 95, 113–114 relevance
intention comprehension 84, 112, 113, constraints on 99, 129
communicative 98, 176, 239, 114, 116, 128 optimal 74, 127, 177, 180, 182,
275, 334–335 multimodality 193–204 240, 271, 332
informative 97–98, 239, 275, mutual parallel adjustment 6, reliability of information 73,
277–278 55, 73, 343–344 82–83, 112–113, 128–129
interjections 75, 184 rhetorical displays 97
interpretive use 89 rhetorical forms 97–98
 Subject index 357

S style 6, 73–74, 167, 201, 328, twist in literary texts 295–297,

saturation of contextual 332 346
sources 334 of St Mark’s Gospel 99, 112 typography 167–188, 199–202,
schemata 38–40 204
semantic indeterminacy 51 T
shallow processing 58–59, 64 telops 193–199, 204–218 V
slurs 229, 234, 237, 239, 244, testimony 97 vagueness 58, 60–61
246, 247–253 third-person imperative verbal attack 232–234
snowclones 37–40 153–159
Speech Act Theory 71, 124, tortoise shell cat 291 W
235 translation wordplay 60, 264–266, 273
strengthening 95, 99–103, of irony 317–323
105, 111–112, 115, 117 audiovisual 198–199
see also cognitive effect of tropes 261–262, 268
strengthening
The chapters in this volume apply the methodology of relevance theory
to develop accounts of various pragmatic phenomena which can be
associated with the broadly conceived notion of style. Some of them are
devoted to central cases of figurative language (metaphor, metonymy,
puns, irony) while others deal with issues not readily associated with
figurativeness (from multimodal communicative stimuli through
strong and weak implicatures to discourse functions of connectives,
particles and participles). Other chapters shed light on the use of
specific communicative styles, ranging from hate speech to humour and
humorous irony.
Using the relevance-theoretic toolkit to analyse a spectrum of style-related
issues, this volume makes a case for the model of pragmatics founded
upon inference and continuity, understood as the non-existence of sharply
delineated boundaries between classes of communicative phenomena.

“This original and entertaining series of essays uses the framework of

relevance theory to offer striking new insights into the workings of figurative
language and style. Why do puns have such a bad reputation? What is the
meaning of “sunbeds are the new cigarettes”? How does choice of typography
affect understanding? These and many more questions are illustrated
with original examples from everyday conversation and literary texts,
and illuminated from a variety of theoretical perspectives.”

Deirdre Wilson

isbn 978 90 272 0554 4

John Benjamins Publishing Company