Mixing Metaphor

Metaphor in Language, Cognition,

and Communication (MiLCC)
issn 2210-4836
The aim of the series is to publish theoretical and empirical interdisciplinary research on the
effective use of metaphor in language and other modalities (including, for instance, visuals) for
general or specific cognitive and communicative purposes. The aim of the series is to offer both
fundamental and applied contributions to the state of the art. The series also invites proposals
for inter-cultural and cross-cultural studies of metaphor in language, cognition, and commu-
nication. Room will be given as well to publications on related phenomena, such as analogy,
metonymy, irony, and humor, as long as they are approached from a comparable perspective.
The scope of the series comprises approaches from the humanities and the social and cognitive
sciences, including philosophy, cultural studies, linguistics, cognitive science, communication
science, media studies, and discourse analysis. More focused attention may be paid to the role
of metaphor in the domains of religion, literature and the arts, the media, politics, organization
and management, law, economics, health, education, and science.
For an overview of all books published in this series, please see
http://benjamins.com/catalog/milcc

Editor
Gerard J. Steen
University of Amsterdam

Editorial Board
Frank Boers
Victoria University of
Wellington
Tony Berber Sardinha
Pontifical University of São
Paulo
Lynne Cameron
The Open University
Daniel Casasanto
University of Chicago
Alan Cienki Thomas Fuyin Li
VU University Amsterdam & Beihang University (BUAA)
Moscow State Linguistic University
Cornelia Müller
Joep Cornelissen Europa Universität Viadrina,
VU University Amsterdam Frankfurt/Oder
Alice Deignan Francisco José Ruiz de
University of Leeds Mendoza Ibáñez
University of La Rioja, Logroño
Raymond W. Gibbs, Jr.
University of California, Santa Cruz Elena Semino
Lancaster University
Joe Grady
Cultural Logic, Providence

Volume 6
Mixing Metaphor
Edited by Raymond W. Gibbs, Jr.
Mixing Metaphor
A descriptive and prescriptive analysis

Edited by

Raymond W. Gibbs, Jr.

University of California, Santa Cruz

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.

In collaboration with the Metaphor Lab Amsterdam.

doi 10.1075/milcc.6
Cataloging-in-Publication Data available from Library of Congress:
lccn 2015043290 (print) / 2015050863 (e-book)
isbn 978 90 272 0210 9 (Hb)
isbn 978 90 272 6750 4 (e-book)

© 2016 – 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

Introduction vii
Raymond W. Gibbs, Jr.

Part I.  Is Mixed Metaphor a Problem?

chapter 1
A view of “mixed metaphor” within a conceptual metaphor theory framework 3
Zoltán Kövecses
chapter 2
Mixed metaphors from a discourse dynamics perspective: A non-issue? 17
Lynne Cameron
chapter 3
Why mixed metaphors make sense 31
Cornelia Müller
chapter 4
Tackling mixed metaphors in discourse: New corpus and
psychological evidence 57
Julia E. Lonergan & Raymond W. Gibbs, Jr.

Part II.  Reasons for Mixing Metaphor

chapter 5
Mixed metaphor: Its depth, its breadth, and a pretence-based approach 75
John Barnden
chapter 6
Mixed metaphor is a question of deliberateness 113
Gerard Steen
chapter 7
When languages and cultures meet: Mixed metaphors in the discourse of
­Spanish speakers of English 133
Fiona MacArthur
 Mixing Metaphor: The State of the Art

chapter 8
The ‘dull roar’ and the ‘burning barbed wire pantyhose’: Complex
­metaphor in accounts of chronic pain 155
Jonathan Charteris-Black

Part III.  Effects of Mixing Metaphor

chapter 9
We drink with our eyes first: The web of sensory perceptions, aesthetic
­experiences and mixed imagery in wine reviews 179
Carita Paradis & Charlotte Hommerberg
chapter 10
A corpus-based study of ‘mixed metaphor’ as a metalinguistic comment 203
Elena Semino
chapter 11
Mixing in pictorial and multimodal metaphors? 223
Charles Forceville
chapter 12
Extended metaphor in the web of discourse 241
Anita Naciscione
Index 267
 Introduction

Raymond W. Gibbs, Jr.

University of California

1.  Mixing metaphor in perspective

Several years ago I asked a group of university students, studying psychology, to

write down whatever they knew about “the topic of metaphor.” My general aim
was to capture something about people’s folk beliefs about metaphor without
defining “metaphor” or providing linguistic examples. The survey did not reveal
very much, which is not surprising perhaps given the vague, open-ended ques-
tion. Some people noted that metaphor involved talking about one topic in terms
of another, or was prominent in poetry, while a few students had nothing to say.
The most frequent comment, however, made by 40% of all respondents, was that
“one should not mix up your metaphors” (or similar wording). When I later asked
students to explain this remark, several stated that the prohibition against mixed
metaphor was something they were explicitly taught in high school. Indeed, if
you look up the phrase “mixed metaphor” on the internet, you will immediately
encounter a whole host of negative comments about the topic, mostly from books
or essays on writing style and composition. Consider a few of these definitions of
mixed metaphor and warnings about its use:
“A succession of incongruous or ludicrous comparisons.”
“Mixed metaphors often, but not always, result in a conflict of concepts.”
“When you use metaphor, do not mix it up. That is, don’t start by calling something
a swordfish and end by calling it an hourglass.”
“Even if a mixed metaphor sings, it should be derailed.”

One older English composition textbook admonishes:

“Sometimes writers go so far as to forget the metaphorical significance of their
words as to combine distinctly incongruous metaphors, producing what are called
mixed metaphors. … All such forms of confusion manifest a certain insincerity;
they show that one is writing without having one’s mind on what he is really
saying.” (Lathrop, 1920: 200)

As these remarks make clear, mixed metaphor does not have a good reputation,
which is why people are often instructed to avoid it. Examples of mixed m
­ etaphor

doi 10.1075/milcc.6.001int
© 2016 John Benjamins Publishing Company
 Mixing Metaphor: The State of the Art

are also posted on the internet to alert people of the bad results that arise when
we get sloppy with our metaphors. Consider a recent example of a statement,
published in the Boston Globe, in which a labor arbitrator commented on a city
tax proposal: “I conclude that the city’s proposal to skim the frosting, pocket the
cake, and avoid paying the fair, reasonable, and affordable value of the meal is a
hound that will not hunt.” The speaker’s remark is representative of “a succession
of incongruous and ludicrous comparisons” that most of us find amusing, even if
we still typically see this as poor, mixed-up metaphor.
The reality, though, is that mixed metaphors are everywhere in both speech
and writing as is made evident by the examples discussed in the present chapters.
There is great irony, for me, in that the most notable belief about metaphor, namely
that we should not mix metaphors together, is a topic which is rarely studied in the
interdisciplinary world of metaphor scholarship. We have explored the vast ways
that people use metaphor in language, gesture, art, and other multimodal media.
Metaphor is now widely recognized as a fundamental scheme of thought, which
gets manifested at many levels of human experience, ranging from culture and his-
tory, through everyday thinking and language, down to neural firings. Still, little
attention is given to the fact that mixed metaphor is quite common.
Are mixed metaphors simply cognitive errors, bad writing, or failed attempts
at humor? One could argue that people possess a rich set of conceptual metaphors,
which enables them to understand a wide variety of abstract concepts. Enduring
conceptual metaphors provide coherence for many aspects of experience. None-
theless, we all occasionally slip up and produce bizarre mixtures of metaphorical
statements. How does this happen?
A different possibility is that mixing metaphors actually demonstrates people’s
cognitive flexibility to think of abstract topics in a myriad of metaphorical ways.
For example, the labor arbitrator talked about a series of ideas using colorful lan-
guage that may reflect his varied metaphorical understandings of the city tax pro-
posal and its implications.
I have always taken great delight in mixed metaphors and noted a slow
­emergence of interest in mixed metaphor within the metaphor community. This
book aims to bring the topic of mixing metaphor center stage within the world of
metaphor studies. The contributors to this volume are all established metaphor
scholars, some of whom have written briefly on the topic of mixed metaphor or
done work that is clearly relevant to both how and why people sometimes combine
metaphors in their use of language, gesture, and multimodal media. Authors were
encouraged to write anything they wanted in connection with mixed metaphor,
and to make use of any empirical findings and theories that they felt best explained
the existence and use of mixed metaphor. The end result is a diverse collection of
articles, appealing to varying empirical findings and theories of metaphor.
 Introduction 

Still, the most notable, consistent theme of this volume is that mixed meta-
phors do not reflect cognitive errors or necessarily impede our understanding of
what other people mean to communicate. People may rapidly shift their attention
between different source domains when speaking of some topic, or engage in elab-
orative reasoning about a single source domain to reveal some of its less known or
hidden meanings. Mixing metaphor in discourse does not necessarily cause listen-
ers or readers undue cognitive effort to interpret speakers’ messages in context.
In many cases, source domain information may only be partly activated, which
enables people to easily attend to other possible sources for expressing metaphori-
cal meanings. In fact, the existence of mixed metaphor, both within and outside of
language, offers testimony to the cognitive flexibility that is the hallmark of human
intelligence and creativity.

2.  Summary of the chapters

The first group of chapters examines whether mixed metaphor represents errors in
thinking and causes difficulties in understanding.
Zoltán Kövecses’s chapter is titled “A view of mixed metaphor within a con-
ceptual metaphor theory framework.” Conceptual metaphor theory (CMT) has
sometimes been criticized for being unable to explain mixed metaphor. Some
scholars have argued that any assortment of linguistic metaphors from different
source domains should be prohibited within CMT because people presumably
think and speak about abstract topics from the perspective of specific concep-
tual metaphors. Once a person conceives of a topic in a metaphorical manner, the
individual should use metaphorical language that is consistent with this specific
metaphorical scheme. Kövecses outlines several reasons why this view of CMT
is incorrect. CMT actually predicts the emergence of mixed metaphors in dis-
course given the fact that many abstract targets are structured by multiple source
domains. In most cases, people are not aware that they are mixing their metaphors
because the varied source domains are activated only to a small degree. People’s
multiple, metaphorical understandings of single topics enable them to easily,
mostly unconsciously shift from one to another, depending on the specifics of the
discourse situation.
Lynne Cameron’s chapter “Mixed metaphor from a discourse dynamics per-
spective: A non-issue?” focuses on how mixed metaphor unfolds in spontaneous
talk. Using her discourse dynamics approach, Cameron argues that there is little
linguistic evidence (e.g., pauses, hesitations, explicit comments from speakers) to
suggest that people experience difficulty understanding mixed metaphorical mes-
sages in context. A crucial part of Cameron’s explanation of mixed metaphor is
 Mixing Metaphor: The State of the Art

that speakers manage their interactions “through metaphor,” which makes even
mixed figures a natural extension of what people are doing when talking. Read-
ers may consciously recognize mixed metaphors in texts given the slower pace
of reading and the visual presence of incongruent figures on the printed page.
But even with written language, the perils of using mixed metaphors are vastly
overstated, especially given the perspective of those individuals immersed in the
flow of discourse. As Cameron argues, most of the prohibitions against mixed
metaphor come from scholars who embrace an asocial view of metaphor, one that
fails to properly acknowledge what people are really doing with metaphor in both
conversation and writing.
Cornelia Müller’s chapter in this section is titled “Why mixed metaphors
make sense.” She argues that mixed metaphor is a perfectly reasonable way of
speaking, writing, and acting (e.g., using gestures), given people’s cognitive abili-
ties to respond to moment-by-moment affordances within different communica-
tive situations. People are able to flexibly shift their attention to various aspects of
metaphorical mappings, and sometimes foreground ideas that are not standard
readings of conventional metaphors. For example, consider the statement “The
butter mountain has been in the pipeline for some time,” which refers to how over-
production of butter in the EU has led to huge amounts of butter awaiting distri-
bution for some time. Although this mixed metaphor presents a clash of domains
on the surface, the two metaphors (e.g., butter mountain and pipeline) nonetheless
work together semantically (e.g., huge amounts of something and awaiting distri-
bution). Bringing forth background knowledge when speaking and gesturing acti-
vates different, mostly hidden, aspects of metaphorical meaning that frequently
makes good sense for the discourse participants. This dynamic view of metaphor
emphasizes how conventional metaphors range along a continuum from sleeping
to waking depending on speakers’ and listeners’ specific degree of cognitive activa-
tion for source domains as discourse unfolds in interpersonal interactions.
Julia Lonergan and Raymond Gibbs’s chapter is titled “Tackling mixed meta-
phors in discourse: Corpus and psychological studies.” This article first presents
results of a corpus examination of a set of mixed metaphor excerpts originally
published in The New Yorker magazine in their “Block That Metaphor!” column.
Most of the metaphorical phrases seen in these excerpts have been previously
employed with similar meanings in other contexts. But the mixing of these dif-
ferent metaphorical expressions is novel. A psychological study, where people
were asked to write out their interpretations of the different parts of each mixed
metaphorical narrative, showed that participants gave very consistent readings
of these phrases. More importantly, people appear to be readily able to integrate
these diverse ­metaphorical phrases into a coherent whole, mostly because of the
ancillary assumptions they make about the source domains in these metaphors.
 Introduction 

The second group of chapters primarily explores how and why people employ
mixed metaphors in discourse.
John Barnden’s chapter, “Mixed metaphor: Its depth, its breadth, and a
­pretense-based approach,” presents an overview of his computer model, ATT-
Meta, which is dedicated to reasoning about metaphor and has been extended to
handle various kinds of mixed metaphor. A key feature of ATT-Meta is its adapta-
tion of a fictionalist/pretense world in which metaphor users pretend that what is
described is literally true. This pretense enables people to draw rich metaphorical
inferences using their extensive knowledge of the source topic. Barnden illustrates
many of the advantages of ATT-Meta for dealing with mixed metaphor, and he
strongly argues against the traditional idea that metaphor understanding primar-
ily depends on constructing parallel mappings from a source to a target. Instead,
much metaphor reasoning involves people drawing on ancillary assumptions
about the source topic to create meaningful interpretations of metaphors, even
ones that are complexly mixed.
Gerard Steen’s contribution is titled “Mixed metaphor is a question of delib-
erateness.” Although many observers claim that mixed metaphors are the prod-
ucts of illogical thinking or simple errors in language production, Steen suggests
that mixed metaphors are typically deliberately produced for specific rhetorical
purposes. In some cases, admittedly, the deliberate use of mixed metaphors leads
to a clash of images, which some may consider to be poor writing (e.g., “From
November 1958 through the summer of 1966 the crisis over Berlin simmered, dip-
lomatic nerves frayed, and the exodus of East Germans grew to a flood.”). Still, the
deliberate mixing of metaphors may lead listeners and readers to infer rich meta-
phorical messages precisely when people recognize their deliberate composition
and use. Steen explores different cases where metaphors are deliberately and non-
deliberately mixed and generally argues how paying attention to the deliberateness
of mixed metaphors is an important feature of a three-dimensional approach to
metaphor focusing on metaphor in language, thought, and communication.
Fiona MacArthur’s chapter, “When language and cultures meet: Mixed
metaphors in the discourse of Spanish speakers of English,” examines how non-
native speakers often graft metaphorical conceptualizations and wordings from
their native language when speaking in a second language. She calls these unique
constructions “hybrid” rather than “mixed” metaphors because speakers are
often successful in communicating these messages in face-to-face interactions.
Of course, even native speakers borrow words and metaphors from other lan-
guages, but non-native speakers do this quite often, as when one person, talk-
ing about her anticipated difficulty in crafting her essays for an upcoming exam
said, “My ­problem is that they are not developed so they have just in squares so
I have to joint at the ideas.” The statement “joint at the ideas” in several places
 Mixing Metaphor: The State of the Art

refers to her wanting to write coherent exam answers that integrate several ideas
and pieces of information. MacArthur claims that hybrid metaphors constitute a
kind of shared understanding, or “conceptual pacts,” between speakers and listen-
ers through their continued use in discourse. Listeners typically recognize the ad
hoc nature of these hybrid metaphors and focus more on the content, as opposed
to the form, of what non-native speakers say. One implication of this analysis is
that training non-native speakers to become proficient in a second language does
not require strict adherence to conventional linguistic forms. There are important
communicative benefits to be gained from people’s use of hybrid metaphors that
are often “condemned under the rubric of mixed metaphor.”
Charles Charteris-Black’s chapter is titled “The ‘dull roar’ and the ‘burning
barbed wire pantyhose’: Complex metaphor in chronic accounts of pain.” From an
examination of interviews with people who are in chronic pain, Charteris-Black
argues that mixed metaphor serves the important rhetorical function of making
a speaker’s claims quite credible to listeners. For example, when people describe
their pain, they employ repeated, mixed images to convey very strong emotions
that are otherwise difficult to articulate given the private, subjective nature of their
pain and discomfort (e.g., “I used to feel that I was wearing a burning barbed wire
pantyhose.”). Using mixed metaphors provides speakers with a sense of greater
control over the chaos of their experience of being ill. Pain suffers use repeated
and mixed metaphors, along with other figures, to emphasize the sincerity of their
beliefs when making claims or arguments to audiences. Mixed metaphors are often
used purposefully for their unique appeal without necessarily being conscious.
The final collection of chapters focuses on ways that mixed metaphors are
realized in speech, writing, and multimodal discourse.
Carita Paradis and Charlotte Hommerberg’s chapter is titled “We drank with
our eyes first: The web of sensory perception, aesthetic experiences, and mixed
mappings in wine reviews.” Wine reviews, authored by famous wine connoisseurs,
often present a stream of consciousness reporting a person’s mixed sensory expe-
riences of picking up, smelling, tasting, and relishing a glass of wine. Consider
a part of one review: “Bright crimson. Extremely sweet and ripe – almost New
World – with some floral aspects. This one is lively and flirtatious with some pretty
dry, sandy tannins underneath. Rather unusual. Could do with just a tad more
acidity to life it. Just a bit confected? Very brutal finish.” The narrator employs
various metaphorical source domains to describe his different visual, olfactory,
gustatory, and tactile sensations (e.g., “lively,” “flirtatious,” “underneath,” and “lift
it”). P
­ aradis and Hommerberg’s analysis highlights the importance of metaphor,
among other things, in writers’ transitions from sensory perception (tasting the
wine) to conception (thinking about the wine) and then into language (talking
about the experience). They note how the pervasive mixing of metaphor is almost
 Introduction 

expected within wine reviews as a discursive practice as writers try to evoke some
of their rich, yet mixed sensory experiences of wine that readers can then read
about and imagine for themselves.
Elena Semino’s chapter is titled “A corpus-based study of ‘mixed metaphor’
as a metalinguistic comment.” She explores a group of 141 occurrences of the
phrase “mixed metaphor” from the Oxford English Corpus (e.g., “This is funny
for me, because I rely on the words always being at my finger tips and on the tip of
my tongue [with apologies for the mixed metaphor] but currently, there’re not”).
Semino’s analysis describes how “mixed metaphor” is used in different genres,
grammatical forms, and for different rhetorical purposes. Most uses of “mixed
metaphor” offer a negative assessment of some topic, although many others are
employed to show off speakers’ and writers’ humor and creativity. Semino con-
cludes by questioning whether “mixed metaphor” necessarily represents a “viable
and operationable technical term,” given the diversity of ways people apply the
phrase in characterizing their own and other’s discourse.
Charles Forceville’s chapter “Mixing in pictorial and multimodal metaphors?”
explores whether it is proper to conceive of mixed metaphor in nonlinguistic
domains. Almost all discussions of mixed metaphor focus on natural language.
Forceville demonstrates through several case studies that several pictures, film
segments, and multimedia products simultaneously mix their metaphors. For
example, one ad for the VW Golf shows an art gallery exhibit of an engine with a
long pipe attachment that ends with the horn of a musical instrument. The merg-
ing of car engine with music instrument, both of which are part of an art exhibit,
implies that the VW Golf is a powerful artistic object, producing beautiful music.
The quote at the bottom of the ad reads “Golf R32. Perfectly tuned.” Forceville
claims that examples like these are instances of metaphorical blends, which may
be best explained within conceptual blending theory. Mixed metaphors in pictures
and films may be more consciously created than what is seen in everyday speech.
Metaphors may be readily mixed within nonlinguistic domains that serve impor-
tant functional and aesthetic purposes and can quickly capture people’s attention
and appreciation.
Anita Naciscione’s chapter is titled “Extended metaphor in the web of
­discourse.” She analyses the stylistic use of lexical metaphor (e.g., “fire” mean-
ing passion from LOVE IS FIRE) and metaphorical phraseological units (e.g.,
“The road to hell is paved with good intentions.”) in discourse, ranging from
Old ­English to Modern English literary texts. One conclusion she draws is that
extended ­metaphor, sometimes involving an interrelationship of metaphor and
metonymy, “is not a slip of the mind as a ‘mixture’ of thought but a regular element
in each case of use.” Extended metaphor has been observed in texts throughout
history and is sometimes recreated in novel forms within contemporary discourse.
 Mixing Metaphor: The State of the Art

One of the beauties of simultaneous metaphorical images, even those involving

several topics, is that it provides semantic and stylistic coherence in the texts,
which extend to multiple uses of metaphor as well. Naciscione concludes that the
traditional views of mixed metaphor as contaminated or impermissible forms are
simply incorrect and that the study of extended metaphor is critical to under-
standing the interaction of stylistic patterns in figurative meaning construction.
One issue over which scholars differ concerns whether mixed metaphors
arise automatically or from deliberate discourse strategies. Many authors here
have argued that people mix metaphors quite unconsciously within the flow of
discourse, while others emphasize the role of purposeful, deliberate, and perhaps
quite conscious thought processes. Mixed metaphors may, indeed, emerge from
various combinations of fast-acting unconscious and slower-developing conscious
thinking. Still, it is clear that the existence of mixed metaphor is not a reflection of
cognitive confusion or disorganized linguistic forms.
There is certainly much more that needs to be explored empirically on how
and why people employ mixed metaphors in language and other domains. My
hope is that this volume will facilitate new research and discussion on mixing
metaphors. The traditional divide between “metaphor” and “mixed metaphor”
may ultimately be untenable given the various flexible ways that people employ
metaphor in language and elsewhere. Rather than being a deviation from proper
metaphor use, mixed metaphors may be ideal reflections of people’s typical meta-
phorical experiences in language, thought, and communication. In this manner,
the study of mixing metaphor may offer significant insights into contrasting theo-
ries of metaphor. Several of the chapters here outline theoretical frameworks that
may explain some of the various ways that metaphors are mixed in language and
other forms of expression. I boldly suggest that one important test for any com-
prehensive theory of metaphor is its ability to explain both how mixed metaphors
come into being and are ultimately interpreted by others. This “mixing metaphor
test” should become one of the essential ways by which scholars evaluate the rigor
and breadth of competing metaphor theories.
part i

Is Mixed Metaphor a Problem?

 chapter 1

A view of “mixed metaphor” within a

conceptual metaphor theory framework

Zoltán Kövecses
Eötvös Loránd University

How does conceptual metaphor theory handle mixed metaphors? Several

metaphor scholars argue that mixed metaphor is a phenomenon that conceptual
metaphor theory cannot handle. Their argument is that, given the claims of
conceptual metaphor theory, mixed metaphors should not occur at all. This is
because once a conceptual metaphor is activated in discourse by means of a
linguistic metaphor, that conceptual metaphor should lead to and support the
use of further linguistic examples of the same conceptual metaphor. However,
in real discourse, the argument goes, most metaphors are mixed, which
indicates that conceptual metaphors are not activated and thus do not lead
to further consistent linguistic metaphors of the same conceptual metaphor.
In the paper, I will argue that the idea of the production of consistent and
homogeneous linguistic examples does not necessarily follow from conceptual
metaphor theory and that, as a matter of fact, the opposite is the case: given
conceptual metaphor theory, we should expect the use of mixed metaphors in
natural discourse.

1.1  Introduction

Most of the metaphors used in natural discourse are of the mixed type. When
examining the metaphors found in natural discourse, we find that the metaphors
in each other’s (near-)vicinity in discourse derive from very different source
domains. But we also find that this does not really present any difficulty for us
to understand the metaphors used. A version of the CMT view is provided by
­Kimmel (2010) to explain this curious fact about mixed metaphors. Kimmel
argues that mixed metaphors are easily understood in discourse largely due to
various properties of discourse.

doi 10.1075/milcc.6.01kov
© 2016 John Benjamins Publishing Company
 Zoltán Kövecses

Let us now return to the specific question of why mixed metaphors are cognitively
successful. In my view, many mixed metaphors are processed unproblematically
because the grammatical structure of the passage exercises little pressure to
integrate them conceptually, and simply lets us interpret them as referring
to different ontological levels. I claim that what influences metaphor cluster
processing most directly are the relations of the clause units in which they occur.
We can answer major parts of the mixed metaphor riddle by paying attention
to how adjacent metaphors are distributed over clauses and what degree of
grammatical integration these clauses show. Three major possibilities can be
distinguished. In some cases metaphors co-specify each other within a single
clause; sometimes they occur in rather tightly connected clauses, and sometimes
they only belong to a larger rhetoric structure spanning more loosely connected
clauses.(Kimmel, 2010: 110)

Kimmel demonstrates these points with the example below:

Tony Blair’s criticism of EU regulations […] would be laughable if it were not
so two-faced (A). While preaching the pro-business gospel (B), he has done
nothing to stop the tide of EU rules (C) and red tape (D) from choking
Britain (E).

Kimmel suggests that the issue and problem of mixing metaphors, that is, con-
ceptually integrating them, only comes up when the different metaphors occur
within a single clause, as in the case of “to stop the tide of EU rules (C) and
red tape (D) from choking Britain (E).” Whereas in the case of the relationship
between (A) and (BCDE) and (B) and (CDE) there is no pressure to resolve
the incompatibility in imagery because of the grammatical structure of the
discourse, in the case of (CDE), which is a relationship between images in a
single clause, there is a pressure to resolve imagistic incompatibility between
“tide,” “red tape,” and “choking Britain.” Given such examples, Kimmel draws
the conclusion:
Only the close syntactic integration of two metaphors within a clause can enforce
or foster a close integration of their semantic content qua imagery. Where mixed
metaphors occur across clauses no ontological clashes will be felt to begin with
and secondary mechanisms to keep the clash at bay are dispensable. If this is
correct, making sense of mixed metaphors is a natural byproduct of default clause
processing.(Kimmel, 2010: 110)

This provides a potential explanation of why most mixed metaphors are not taken
to be cases of incompatible images and are, thus, not problematic for metaphor
interpretation.
However, Kimmel’s account does not explain a number of issues relating to
mixed metaphors. These include:
 Chapter 1.  A view of “mixed metaphor” within a conceptual metaphor theory framework 

1. Why are imagistically incompatible metaphors selected at a particular point

in discourse?
2. Why are mixed metaphors so common?
3. Why do we have cases of metaphorically entirely homogeneous discourse?
4. Why do we often have widely divergent source domains inserted into the
discourse?
5. How do we comprehend mixed metaphors?

Let us briefly discuss each of the questions above.

1.2  Some questions about mixed metaphors

1.2.1  W
 hy are imagistically incongruent metaphors selected at a particular
point in discourse?
Given a target domain, or frame, that was introduced into the discourse, certain
elements of the domain need to be employed and linguistically expressed. These
elements are the meanings that a speaker wants to express in the course of produc-
ing the discourse. When expressing a target domain meaning, the speaker needs
to employ an element that either comes from the target domain directly or from a
source domain that is systematically linked to that target by means of a set of map-
pings. In the former case, the speaker speaks literally (i.e., directly), in the latter, he
speaks metaphorically (i.e., indirectly).
The target domain meanings form a part of aspects of the target domain,
such as progress, functioning, control, stability of structure, and so on. There are,
in many cases, conventional source domains whose main function is to meta-
phorically express such aspects of target domain concepts (see Kövecses, 2010a).
In other words, there are source concepts that profile these aspects of the target
domain. For instance, the source domain of journey typically profiles the notion
of “progress,” that of machine profiles “functioning,” that of war and fight-
ing profile “control,” that of building profiles “stability of structure,” that of the
human body profiles “(hierarchical) structure” and “appropriate condition (of
some structure),” and so on. In some of my publications I refer to these as the
“meaning foci” of a source domain (see, e.g., Kövecses, 2000, 2002/2010a, 2005).
Given these meaning foci, we get generic metaphors like progress is a j­ ourney,
functioning is (the working of) a machine, control is war/fighting,
stability of structure is a building, and several others. These are meta-
phors that apply to a large number of targets; that is, they have a wide scope
­(Kövecses, 2000, 2002/2010a).
 Zoltán Kövecses

If this is the case, we should not expect discourses about a particular target
domain topic to be expressed by large sets of homogeneous metaphorical expres-
sions (i.e., expressions that belong to the same source domain). In contrast, given
a particular target domain and its various aspects, we should expect metaphorical
linguistic expressions to occur in the discourse that capture and are based on the
typical source domains that are conventionally employed to express and capture
those aspects.
To use Kimmel’s example, the writer of the passage above needs to talk about
“unnecessary paperwork, which is disliked by most people” at a certain point of
the discourse. He could use the phrase in quotation marks or some other phrase
(like “irritatingly redundant bureaucracy”), but, instead, he chooses a metaphori-
cal expression, red tape, which renders this meaning appropriately in a convention-
alized manner. (Actually, it is debatable whether red tape is a metaphor. It could
be argued it is a metonymy.) As a matter of fact, the application of this phrase is
just as natural as that of tide in the same clause in the sense of “a huge amount of
something.” In other words, no matter how incompatible the images evoked by
tide and red tape are, they are selected because they can render (indirectly) the
elements required by the different aspects of the target domain at a certain point
in the discourse.

1.2.2  Why are mixed metaphors so common?

Kimmel (2010) also observes that in the newspaper coverage of the 2004/2005
European Union referenda most of the metaphors were of the mixed type (76%).
Mixed metaphors are common even within the same clause (see Kimmel, 2010)
because the target domains, or frames, we are developing in the course of produc-
ing and understanding (metaphorical) discourse have many different aspects to
them, and these aspects normally require different source domains, or frames, for
their conceptualization, as was noted above.
Assume, for example, that the topic of discussion is the family. The notion
(domain, frame) of family comes with a large number of different aspects (consti-
tuted by conceptual elements), including parents, children, the raising of children,
the cost of raising children, the family as a cohesive group of people, and a host of
others. Below is a sentence I found discussing some of these aspects, as cited by
David Brooks in the International Herald Tribune (November 17–18, 2012): “The
surest way people bind themselves is through the family.” What does way have
to do with bind? The word way belongs to the motion frame, while bind to the
physical joining frame (inside the frame a group of physically connected
entities). These are incongruent images (i.e., metaphorical source domains) that
create mixed metaphors.
 Chapter 1.  A view of “mixed metaphor” within a conceptual metaphor theory framework 

To account for why these and many other incongruent images are so common,
I propose, in line with the argument in the previous section, that all (target domain)
concepts consist of a number of different aspects, and that each of these aspects can
be conceptualized metaphorically by means of different source domains – domains
that may yield incongruent images when juxtaposed. To take the example above,
as the discussion of the family domain is unfolding, various aspects of it are men-
tioned; for example, the issue of what kind of social unity is provided by the fam-
ily and the issue of how a certain state can be achieved (“The surest way people
bind themselves is through the family.”). As regards the second metaphor of the
sentence (bind), it is based on the conceptual metaphor cohesive social groups
are groups of physically connected entities. Given this metaphor, one of
its mappings “the physical joining of entities → the formation of social groups”
accounts for the use of bind. Social groups constitute more or less stable states of
interconnected entities. The issue of how more stable states are achieved is another
aspect of the family/social group domain. And since the issue of how certain states
are achieved in general is metaphorically conceptualized as manner (of action)
is path (of motion), which is a submetaphor of (deliberate) change of state
is (forced) motion, we can account for the use of the word way in the sentence
above. The resulting images are fairly incongruent when juxtaposed (way – path/
motion vs. bind – the physical joining of entities/a group of physically
connected entities), but their incongruence is natural since we use different
source domains to conceptualize different aspects of the target domain. As pointed
out by Kimmel (2010), this is very common in natural discourse; it seems to be the
dominant pattern – the rule, rather than the exception.

1.2.3  W
 hy do we have cases of metaphorically entirely homogeneous
discourse?
If the account of metaphor in discourse as presented above is valid, it is highly
unlikely that we find metaphorically entirely homogeneous discourses. If it is the
case that different target domains have different aspects and the different aspects are
conceptualized by means of different source domains, discourse about a particular
target domain will be characterized by a variety of different conceptual metaphors,
as we have seen above. However, it is clear that there are such cases of metaphori-
cally homogenous discourse, though their occurrence is not very common.
Consider one example of such a text:
Performance targets are identical to the puissance at the Horse of the Year Show.
You know the one – the high-jump competition, where the poor, dumb horse is
brought into the ring, asked to clear a massive red wall, and as a reward for its
heroic effort is promptly brought back and asked to do it all over again, only higher.
 Zoltán Kövecses

I’ve never felt anything but admiration for those puissance horses which, not so
dumb at all, swiftly realize that the game is a bogey. Why on earth should they
bother straining heart, sinew and bone to leap higher than their own heads, only
to be required to jump even higher? And then possibly higher still.

Hard work and willingness, ponders the clever horse as he chomps in the stable
that night, clearly bring only punishment. And so next time he’s asked to canter
up to the big red wall, he plants his front feet in the ground and shakes his head.
And says, what do you take me for – an idiot?
 (Melanie Reid, The Times, Monday, February 4, 2008).

In this case, the same conceptual metaphor, life is a horse show, lends coher-
ence to a single discourse. The novel conceptual metaphor that structures the
text is characterized by a number of mappings: puissance horses are compared to
people, riders to managers, the red walls as obstacles to the targets people have to
achieve, having to jump over the obstacles to being subject to assessment, clearing
the obstacles to achieving the targets, raising the obstacles to giving more difficult
targets, the Horse Show to life, and so on and so forth. This elaborate metaphorical
analogy provides a great deal of structure for the text.
However, it may be even more important than the coherent organization of
the discourse that the author wishes to make a point emphatically and vividly by
the use of the homogeneous metaphor. This expressive function of the discourse
can often provide an explanation for the use of a single coherent metaphorical
image in discourse.
In addition to its organizational and expressive function, metaphorically
homogeneous discourses can also be used for didactic purposes. Parables are com-
monly used this way. One of the best-known parables is Jesus’ parable of the sower
in Matthew 13 (Biblegateway, http://www.biblegateway.com/passage/?search=Ma
tthew+13&version=NIV):
1 That same day Jesus went out of the house and sat by the lake. 2 Such large
crowds gathered around him that he got into a boat and sat in it, while all the
people stood on the shore. 3 Then he told them many things in parables, saying:
“A farmer went out to sow his seed. 4 As he was scattering the seed, some fell along
the path, and the birds came and ate it up. 5 Some fell on rocky places, where it did
not have much soil. It sprang up quickly, because the soil was shallow. 6 But when
the sun came up, the plants were scorched, and they withered because they had
no root. 7 Other seed fell among thorns, which grew up and choked the plants.
8 Still other seed fell on good soil, where it produced a crop – a hundred, sixty or

thirty times what was sown. 9 Whoever has ears, let them hear.”

Jesus spells out the mappings that constitute the conceptual metaphor god’s
words are seeds in the following way:
 Chapter 1.  A view of “mixed metaphor” within a conceptual metaphor theory framework 

18 “Listen then to what the parable of the sower means: 19 When anyone hears the

message about the kingdom and does not understand it, the evil one comes and
snatches away what was sown in their heart. This is the seed sown along the path.
20 The seed falling on rocky ground refers to someone who hears the word and at

once receives it with joy. 21 But since they have no root, they last only a short time.
When trouble or persecution comes because of the word, they quickly fall away.
22 The seed falling among the thorns refers to someone who hears the word, but

the worries of this life and the deceitfulness of wealth choke the word, making it
unfruitful. 23 But the seed falling on good soil refers to someone who hears the
word and understands it. This is the one who produces a crop, yielding a hundred,
sixty or thirty times what was sown.”

Characteristic of the discourses structured by the life is a horse show and god’s
words are seeds metaphors is their conceptual homogeneity. This is provided for
by consistently applying images (elements: ideas, meanings) from the same source
to structure the various aspects of the target. In such cases, as a result of the consis-
tent structuring (based on a single source image), no or very few mixed metaphors
are found in discourse. What discourses such as the above show is that certain
discourse functions (vividly making a point, teaching people) may provide fairly
strong motivation for the use of a single consistent metaphorical image (unlike
the typical case of metaphorically varied imagery, given less marked functions of
discourse).

1.2.4  W
 hy are often widely divergent source domains inserted into
discourse?
While in the case of metaphorically homogeneous discourse we experience a cer-
tain degree of surprise as a result of the unexpectedly homogeneous character of
the discourse, the opposite can also occur: In certain other kinds of discourse, we
can be surprised by some of the widely divergent metaphors that are used in con-
nection with a target domain. This element of surprise comes, I suggest, from the
fact that, first, instead of the expected (conventionalized) metaphors associated
with a target (see Section  2) and, second, deliberately homogeneous metaphors
chosen for a target (see Section 3), entirely divergent metaphors are used for the
target domain. The question is where such unexpectedly divergent metaphors
come from. (It should be noticed that here the question word “why” means some-
thing different than in the previous section titles; instead of the meanings “what
causes the use of mixed metaphors?” and “for what purpose do we use mixed
metaphors?”, here it means “what allows us to use mixed metaphors?”)
I argued in some recent publications (Kövecses, 2005, 2010b, in press) that
the choice of metaphors in particular discourses is often influenced by the local
and global context. The local context involves, among other things, the ­immediate
 Zoltán Kövecses

cultural, social, physical, linguistic, etc. contexts of the discourse. Let us take
an example that I first analyzed in previous publications (Kövecses, 2010a, b).
In it, Bill Whalen, a professor of political science in Stanford and an advisor to
Arnold Schwarzenegger in his campaign to become governor of California, says
the following about Schwarzenegger:
“Arnold Schwarzenegger is not the second Jesse Ventura or the second Ronald
Reagan, but the first Arnold Schwarzenegger,” said Bill Whalen, a Hoover
Institution scholar who worked with Schwarzenegger on his successful ballot
initiative last year and supports the actor’s campaign for governor.

“He’s a unique commodity – unless there happens to be a whole sea of immigrant

body builders who are coming here to run for office. This is ‘Rise of the Machine,’
not ‘Attack of the Clones.’” (San Francisco Chronicle, A16, August 17, 2003)

Focusing on the second paragraph, we may note that the target domain that is dis-
cussed by Whalen is Schwarzenegger’s uniqueness as an individual. The statement
“He’s a unique commodity” clearly indicates this. The linguistic metaphor unique
commodity is based on the people are commodities conceptual metaphor. Next,
we find the highly conventional metaphorical expression “a whole sea of immi-
grant body builders,” where sea indicates any large quantity, in this case body
builders. This is an instance of the conventional conceptual metaphor a large
number of people is a large mass of substance. Equally conventional is the
linguistic metaphor “run for office,” an instance of the presidential election is
a race conceptual metaphor. The commodity and large mass of substance
metaphors emphasize Schwarzenegger’s unique character.
Against the background of this “sea” of highly conventional metaphorical
expressions and conceptual metaphors, two nonliteral phrases at the end of the
paragraph, “This is ‘Rise of the Machine,’ not ‘Attack of the Clones,’” stand out
as unconventional and novel. In all probability, Whalen produces them and we
understand them because Schwarzenegger played in the first of these movies,
and we share this knowledge with the conceptualizer (Whalen) about the topic of
the discourse (Schwarzenegger). Moreover, these are movies that everyone knew
about in California and the US in 2003; that is, they were part and parcel of the
immediate cultural context at the time.
The movie title Rise of the Machine is a widely divergent metaphor in that it
does not come from the available set of conventional conceptual metaphors to
capture an aspect of the target domain. It derives from and is sanctioned by the
immediate context of the discourse in which the metaphor is created. Such meta-
phors often provide images that may not fit the conventionally available source
domains of a given target; nevertheless, they are easily understood through the
supportive role of context.
 Chapter 1.  A view of “mixed metaphor” within a conceptual metaphor theory framework 

1.2.5  How do we comprehend mixed metaphors?

The issue of mixed metaphors is largely an issue in metaphor processing. We can
think of metaphor comprehension in two ways: We can suggest that metaphorical
expressions in discourse activate domains, or frames, representing source domains
that are linked to target domains. This is the more or less accepted view among
those who embrace conceptual metaphor theory.
The other view of the comprehension of metaphorical expressions in discourse
maintains that there is no activation of source domains in most cases. Instead, the
metaphorical expressions used in discourse are associated with a conventionalized
metaphorical sense in the target domain, and it is this target meaning that gets
activated in the course of processing metaphorical discourse.
The two views have different implications for the study of mixed metaphors.
On the CMT view, whenever we come across mixed metaphors, we have to some-
how resolve the incompatibility in imagery between two source domains that get
juxtaposed. On the non-CMT view, such incompatibilities do not arise because
linguistic metaphors are processed directly – namely, in terms of their target
domain meanings. It follows that if we have no difficulties understanding discourse
involving mixed metaphors, it is because either we eliminate the incompatibilities
in the course of processing (for the CMT view) or because we apply “shallow pro-
cessing” (Gibbs, 1999), that is, we comprehend target meanings directly (for the
non-CMT view).
Given the two ways of metaphor comprehension briefly described above, it
is the non-CMT view that appears to be the more adequate way to account for
the processing of mixed metaphors: If people can process the majority of mixed
metaphors effortlessly, this is because they do not run into any incompatibilities as
a result of shallow processing, where they comprehend the metaphors of discourse
in a direct manner.
Can we nonetheless save the CMT view of metaphor comprehension for, or
in the case of, mixed metaphors? I would say yes. My suggestion would be that the
various source domains (syntactically close to each other) that are associated with
various aspects of target domains get activated to varying degrees in the course of
metaphor comprehension. In the case of comprehending mixed metaphors (the
majority of metaphor examples in natural discourse), the various (near-)adjacent
source domains are activated to a low degree. This way, the low level of activation
for a given source does not interfere with the low level of activation of another
source. Consequently, if the activation level of adjacent sources is low, it does not
produce incompatibilities or conceptual clashes between the sources. This assumes
a threshold above which source activation can produce interference or incompat-
ibility between two (near-)adjacent source domains. The observation that there is
 Zoltán Kövecses

no such felt incompatibility between the sources indicates that the level of activa-
tion stays below the threshold level in most cases of mixed metaphors in natural
discourse. At the same time, for this model of comprehension to work, the low
level of source activation should be sufficient to activate source-to-target map-
pings on which target meanings are based. It is this mechanism that can function
as the substitute for the “direct target meaning” of the shallow processing view.
In other words, in this account I rely on and assume the validity of two propo-
sitions. One is that in the case of metaphor the various source domains and the
metaphorical linguistic expressions that make them manifest are characterized by
differing degrees of conventionality. Second, the various degrees of conventional-
ization for the source domains and source expressions are correlated with differing
degrees of neural activation. Lower levels of conventionality and, hence, activation
do not produce consciously recognized incompatibility between source domains
and their metaphorical expressions, while higher levels do. In my view, this is what
we find in most cases of the use of mixed metaphors, as discussed above. This is my
commonsense assumption that I offer merely as a hypothesis for how we compre-
hend mixed metaphors in discourse.
Consider now a case where the level of activation, in my judgment at least,
does go beyond the threshold. Take the following examples of mixed metaphors
from one of the many websites devoted to the topic (http://therussler.tripod.com/
dtps/mixed_metaphors.html):
If they do that, they might as well take the open door policy and throw it right
out the window!

The example is about policy making. This involves the kinds of policy one pursues
and whether the policies used are successful or appropriate or not. The phrase
open door policy is based on the source domain of container (a government
is a container) and seeing (knowing is seeing). The phrase throw it out of
the window implies the source domain of discarded objects (in the metaphor
unnecessary ideas are discarded objects. How do we know that the activa-
tion level goes beyond the threshold? We know it because although we under-
stand the target domain meaning of the sentence easily, the sentence also produces
a humorous effect. As is well-known, humor often involves some incongruence
between two images (such as source and target or, in the case of mixed meta-
phors, two adjacent source domains). The humorous effect can only be produced
by a very high level of activation for the sources as well (and not just the target).
Without the high level of activation of the two sources, we would easily gloss over
the obvious incompatibilities. In cases like this, the humorous effect produced by
many mixed metaphors can help us determine the level of activation of the two
adjacent source domains.
 Chapter 1.  A view of “mixed metaphor” within a conceptual metaphor theory framework 

Some opponents of (parts of) conceptual metaphor theory (e.g., Shen and
Balaban, 1999) hold that it is only the widespread application of metaphorically
homogeneous discourse that could prove source domain activation in the use of
conceptual metaphors. Since, however, such metaphorically homogeneous dis-
courses are rare, they suggest that one of the major claims of conceptual metaphor
theory (i.e., simultaneous activation of source and target) is called into question.
Why are metaphorically homogeneous discourses so rare? In all probability,
this is because, in line with the proposal offered above (Sections 1 and 2), most
metaphorical source domains profile a particular aspect (or aspects) of that source
domain. But if they do occur, this indicates that a single source domain is used to
characterize a target (or a large part of it). In this case, we find the full (or rather,
nearly full) activation of that source for the target in question (cf. the horse show
example). In terms of the activation model I am suggesting, this would mean that
the activation level of the source is above the threshold and remains high through-
out the relevant parts of the discourse. This can be achieved under two conditions.
One is that the source domain is conceptually complex enough to be mapped
onto a variety of different aspects of the target. For instance, the conceptual com-
plexity of the horse show source domain seems to be sufficient to present several
rather different aspects of life that could also be rendered by means of several
different source domains with the appropriate meaning focus.
The other condition is that for some communicative or stylistic reason it is
advantageous for the speaker/conceptualizer to conceptualize (and talk about) the
target via the same source domain. Making a point vividly and forcefully, as in the
case of the horse-show example, and accomplishing a didactic purpose, as we saw
in the case of Jesus’ parable of the sower, may be one such reason. This second
condition is clearly needed: Most cases of metaphorically homogeneous discourse
have an added communicative or stylistic component (such as convincing, irony,
humor, esthetic effect) in a way that is not present in discourses that fit the domi-
nant pattern of metaphorically non-homogeneous discourse.

1.3  Mixed metaphors and the issue of deliberateness

Does it follow that the metaphorical expressions in metaphorically homogeneous

discourses are used deliberately? After all, it may seem that the metaphorical
expressions that are based on a consistent image (source) and are thus homoge-
neous are the result of a deliberate choice. To some extent this may be so, but
the question is whether the source domain itself emerges in a deliberate way to
begin with. Is it the case that the author of life is a horse show came to this
metaphorical source domain by way of a conscious or deliberate act? Did she ask
 Zoltán Kövecses

what would be a good source image for the particular conceptualization of life she
wanted to demonstrate? Not denying the possibility of this happening in certain
cases, the chances are that she had the source image in mind first spontaneously
and intuitively, and then she expanded that image in a more or less deliberate
manner to cover as much of the target as possible. This also seems to happen in
the case of “context-induced” metaphors discussed above (Section  4). In them,
the various contextual factors may unconsciously and intuitively prime the use of
source domains.
If this suggestion is more or less on the right track, we can see the use of
metaphorical source domains involving at least two steps: first, the unconscious
and intuitive emergence of the source, and, second, a more or less deliberate con-
ceptual and linguistic elaboration of the source in relation to the target. As we saw,
the latter is especially likely to occur when particular communicative and stylistic
goals drive the metaphorical process.

1.4  Conclusions

In this chapter, I argued that conceptual metaphor theory can handle mixed met-
aphors in a natural way. Its naturalness arises from the fact that general target
domains have a variety of different aspects to them that can be, and often are,
conceptualized in terms of a variety of potentially very different source domains.
The use of these different source domains leads naturally to the use of mixed meta-
phors. In a way, most communicatively and stylistically neutral cases of discourse
can only be metaphorically non-homogeneous discourse that consists of a variety
of different source images, that is, mixed metaphors.
We do not have any difficulty processing such discourses. My suggestion here
was that this is possible because, in most cases, source domains are activated at
a low-level. This low level of activation for a particular domain does not inter-
fere with the activation of an adjacent source domain in the discourse. However,
the low level of activation should be sufficient enough to activate source-to-target
mappings that can result in the required meanings needed to understand and pro-
duce discourse.
Clearly, this model of comprehending mixed metaphors is just a hypothesis.
The model involving varying degrees of activation of source domains seems to
provide a reasonable framework for the comprehension of mixed metaphors and
at the same time to be compatible with conceptual metaphor theory. My hope
is that the ideas laid out in the paper will serve as potential hypotheses that
can be tested experimentally using psycholinguistic or cognitive neuroscience
methods.
 Chapter 1.  A view of “mixed metaphor” within a conceptual metaphor theory framework 

References

Gibbs, R. (1999). Researching metaphor. In L. Cameron & G. Low (Eds.), Researching and apply-
ing metaphor (pp. 29–47). Cambridge: Cambridge University Press.
Kimmel, M. (2010). Why we mix metaphors (and mix them well): Discourse coherence, concep-
tual metaphor, and beyond. Journal of Pragmatics, 42, 97–115.
doi: 10.1016/j.pragma.2009.05.017
Kövecses, Z. (2000). The scope of metaphor. In A. Barcelona (Ed.), Metaphor and metonymy at
the crossroads (pp. 79–92). Berlin: Gruyter.
Kövecses, Z. (2002). Metaphor. A practical introduction. New York: Oxford University
Press.  doi: 10.1017/CBO9780511614408
Kövecses, Z. (2005). Metaphor in culture. Universality and variation. New York: Cambridge
­University Press.
Kövecses, Z. (2010a). Metaphor. A practical introduction (2nd ed.). New York: Oxford University
Press.
Kövecses, Z. (2010b). A new look at metaphorical creativity in cognitive linguistics. Cognitive
Linguistics, 21(4), 663–697.  doi: 10.1515/cogl.2010.021
Kövecses, Z. (In press). Where metaphors come from. Reconsidering context in metaphor. New
York: Oxford University Press.
Shen, Y., & Balaban, N. (1999). Metaphorical (in)coherence in discourse. Discourse Processes,
28(2), 139–153.  doi: 10.1080/01638539909545077
Steen, G. (2011). The contemporary theory of metaphor – now new and improved! Review of
Cognitive Linguistics, 9(1), 26–64.  doi: 10.1075/rcl.9.1.03ste
 chapter 2

Mixed metaphors from a discourse dynamics

perspective
A non-issue?

Lynne Cameron
The Open University

The ‘mixed metaphor problem’ is considered from the discourse dynamics

perspective, examining instances from metaphor clusters in dialogue and
interaction. Several types of multiple juxtaposed verbal metaphors are found but
they rarely demonstrate the semantic dissonance or stylistic undesirability held
characteristic of problematic mixed metaphors. Multiple verbal metaphors are not
stylistically-tricky additions to the flow of talk but are constitutive of it, discursive
resources that contribute to the flow of jointly-constructed meaning. Multiple
metaphors are shown to result from and find coherence in: shifting discourse
topics; anaphoric reference and lexico-conceptual pacts; being combined in
coherent metaphorical scenarios; layering of conventionalised and systematic
metaphors.
The second part of the chapter demonstrates how multiple metaphors were
selected and combined to form the basis of the new model of empathy~dyspathy
dynamics in social science research, and reflects on the inevitability of mixing
metaphors.

2.1  Introduction

I approach the issue of mixed metaphors from a discourse dynamics perspec-

tive, i.e., by examining how people use them in the flow of situated text and talk
(Cameron, 2010; Cameron et al., 2009). Across my data sets of people engaged in
spontaneous talk, it is not at all unusual to find multiple metaphor vehicles1 from

.  I use ‘vehicle’ to refer to words or phrases used metaphorically in the flow of talk and text
(see Cameron et al., 2009 for details on how these are identified. Metaphor ‘topic’ is used to refer
to the contextual meaning of the vehicle; in talk, metaphor topics are very often not explicit.

doi 10.1075/milcc.6.02cam
© 2016 John Benjamins Publishing Company
 Lynne Cameron

different semantic fields occurring close to each other. It is, however, unusual to
find people struggling to make sense of them, commenting on their semantic dis-
sonance, or producing any paralinguistic feature, such as pausing, hesitation, or
laughter, to suggest that they have even noticed the juxtaposition. Mixed meta-
phors appear to be a non-issue.
In this chapter, I consider this phenomenon and suggest that the most likely
explanation is that participants are constructing discourse meanings ‘through’
the metaphors and so do not notice dissonance between the basic, more physical
senses of juxtaposed metaphor vehicles. Things may be different in composing text
because (a) text is visible and (b) reading is a slower process, allowing more time
and attention to notice semantic dissonance between metaphor vehicles on page
or screen. And it is in the context of writing that advice to avoid mixed metaphors
is often heard. Semantic risks of mixed metaphors appear exaggerated; such advice
is more about preferred or conventionalised stylistic preferences.
In Extract (1), from a detective novel, a tabloid newspaper editor is talking to
a junior reporter, Zed, about the need to find a good story:
Extract (1)
…death’s nicely suspicious, and suspicion is our bread and butter. Metaphor,
by the way, in case you’re thinking otherwise. Our purpose is to fan the fire –
another metaphor, I think I’m on a roll here – and see what comes crawling out
of the woodwork.’
‘Mixed,’ Zed muttered.
‘What?’
‘Never mind…’
(Believing the Lie by E. George, 2012, p. 67, London: Hodder & Stoughton)

Zed’s muttered comment about the editor’s mixed metaphors is thrown into the
dialogue but not further developed. It contributes to the characterisation of Zed
the author as a young man more suited to poetry than tabloid journalism. The
single word comment, ‘Mixed’, leaves readers to work out that Zed is critiquing the
editor’s successive metaphors, and assumes shared knowledge/attitude that mix-
ing metaphors, while characteristic of tabloid writing, is stylistically undesirable.
The ‘mixed’ comment presumably refers to the editor’s juxtaposition of the
bread and butter metaphor (with contextual meaning ‘how we make our living’)
and the fan the fire metaphor (with contextual meaning ‘create trouble in the
possible crime situation’) have dissonant basic meanings in the physical world.2

.  The terms ‘basic meaning’ and ‘contextual meaning’ are taken from the ‘pragglejaz’ meta-
phor identification procedure, in which they must both contrast and connect if metaphor is to
be identified (pragglejaz group, 2007).
 Chapter 2.  Mixed metaphors from a discourse dynamics perspective 

‘Mixed metaphors’, in this conventional, stylistic meaning of the phrase, are jux-
taposed metaphor vehicles whose basic meanings are incompatible and create a
semantic ‘dissonance’, while their contextual meanings are not.
The following metaphor what comes crawling out of the woodwork is not
incompatible with fan the fire, but works with it in a coherent metaphorical
­scenario in which fires lead to rats, cockroaches and other pests running out of
burning buildings. The contextual meaning of the running pests is those people
whose crimes are revealed by the newspaper’s actions. These metaphors would
not count as ‘mixed’ since any dissonance in basic meanings can be resolved at a
scenario level.3
In the rest of the chapter, I consider ways in which multiple metaphors work
in the dynamics of discourse and why they seldom, if ever, can be labelled ‘mixed
metaphors’. Secondly, I reflect on my own experience as a writer deliberately mak-
ing use of metaphors to explain ideas from my research, and how multiple meta-
phors in text result ineluctably from the limits of metaphor. I conclude with a
discussion of implications for theory.

2.2  Multiple metaphors in talk

The obvious place to look for multiple metaphors is in ‘metaphor clusters’, stretches
of talk with significantly more metaphors than surrounding talk (­Cameron  &
Stelma, 2004). A majority of clusters include verbal metaphors from several
­distinct semantic fields.

2.2.1  The production of multiple metaphors

In a cluster, as in any stretch of talking-and-thinking,4 speakers explore and
develop ideas as they speak about them. Even if they begin with a clear idea of
what they want to express, the process of putting into words, hearing them spo-
ken, and seeing the effect on one’s interlocutor may well produce adjustments and
adaptations ‘in the moment’ of talking-and-thinking. As conversation around a

.  ‘Metaphorical scenarios’ differ, theoretically, from Musolff ’s ‘metaphor scenarios’ (e.g.,
Musolff, 2004), which are large-scale conceptual notions rather than phenomena relating to
discourse activity (Cameron et al., 2010).
.  ‘Talking-and-thinking’ is the term I use to capture the inseparability of cognitive pro-
cesses and using language in interaction during spontaneous conversation (Cameron, 2003).
It builds on Slobin’s studies of ‘thinking for speaking’ (Slobin, 1996).
 Lynne Cameron

discourse topic5 proceeds, metaphorically-used words and phrases, in the same

way as for any lexical item (McCarthy, 1988) are repeated, relexicalised, developed,
and contrasted with antonyms of various types (Cameron, 2008). These shifts give
rise to successive verbal metaphors.
Chafe (1996) describes the flow of talk as a stream of idea units, each the
focus of participants’ attention for the second or two in which an intonation unit
is produced and interpreted, and then receding from attention into some level of
on-going awareness. In this flow of talking-and-thinking, the researcher identi-
fies metaphor vehicles as words and phrases that contrast with the on-going dis-
course topic and contribute to its meaning (Cameron, 2003; pragglejaz group,
2007). Metaphor topics are seldom explicitly presented as in Extract (1): suspicion
is…; our purpose is…. The analyst constructs metaphor topics by extrapolating
from the talk some explicit formulation of contextual meaning, but discourse par-
ticipants interpret metaphor vehicles in the flow of discourse. When we consider
verbal metaphors from this discourse dynamic perspective, it becomes obvious
how potentially dissonant basic meanings, that might produce a feeling of mixed
metaphors, would be backgrounded in favour of ‘making sense’ of the on-going
flow of talking-and-thinking.
In this section, I use previously analysed clusters to discuss different ways in
which multiple metaphors are present in talk. The data are reconciliation conver-
sations between a perpetrator of violence and the daughter of one of his victims.
The data have been used to investigate the development of empathy (Cameron,
2007, 2011) and to develop metaphor cluster methodology ­(Cameron & Stelma,
2004). Readers are referred to those publications for details of data and the
method of metaphor analysis, including metaphor identification and grouping
of individual verbal metaphors into larger ‘systematic metaphors’, i.e. trajectories
of semantically-connected verbal metaphors across the talk (also Cameron  &
Maslen, 2010).
The conversations feature Patrick Magee, who in 1984 planted a bomb in
the UK hotel where the Conservative party conference was taking place, killing
five people and injuring many others, and Jo Berry, whose father, Sir Anthony
Berry, was one of those killed and who wanted to meet Magee in order to try to
understand the roots of the violence. Their talk contains a high density of verbal
metaphor in comparison with other discourse contexts I have studied, partly
because of the intense emotions involved and partly because of the abstract
nature of much of what they talk about: grief, political motivation, empathy. The

.  ‘Discourse topic’ refers to what is being talked about, and is different from ‘metaphor
topic’.
 Chapter 2.  Mixed metaphors from a discourse dynamics perspective 

following set of ways in which multiple metaphors are used is not intended as an
exhaustive t­ axonomy; analysis of other genres of talk will no doubt add further
types. Extracts are included as typical examples of each from the many available
in the data.

2.2.2  Multiple metaphors arising from shifting discourse topics

Mere juxtaposition of metaphor vehicles from distinct semantic fields, such as
occurs in Extract (2), is not necessarily indicative of potential ‘mixed’ metaphors.
It may just be that a speaker moves to a different discourse topic, as in line 743:
Extract (2)
737 Pat .. sometimes you get a –
738 like a glimpse.
739 even in the midst of –
740 er,
741 a lot of struggle.
742 .. of the other person’s humanity.
743 but there’s also a mechanism there that,
744 … I think locks it out.

In this extract, Pat first talks about how another person’s humanity can be meta-
phorically lost from sight in a conflict (struggle), although getting a glimpse of that
humanity may still be possible (738–42). At line 743, he shifts the topic (marked
by falling intonation at the end of 742 and but), from talking about sometimes
seeing the other as human to talking about the denial of humanity during con-
flict. The new discourse (sub)topic is spoken of with different metaphor vehicles –
­mechanism, locks out – that construct a more machine-like scenario in contrast
with the human seeing metaphor of glimpse. Although the verbal metaphors
appear ‘mixed’, it is rather a case of a shift to a contrasting aspect of the discourse
topic, through a shift in vehicles that emphasises the particular contrastive aspect
of locking out the other’s humanity.
Extract (2) illustrates a feature of metaphor in discourse that is important to
this discussion: that the ‘opposite’ or antonym of a vehicle is not necessarily what
would work as antonym for the basic sense of the word or phrase. Instead the
contrasting vehicle highlights a particular aspect. Here, the opposite of glimpsing
is locking out, rather than not seeing. Although locking out does prevent seeing,
and so is not incompatible, it highlights agency and suggests the cold hardness
of metal. Deignan (2005) gives other examples of this phenomenon from her
corpus studies. Meaning construction in talking-and-thinking through meta-
phor is not constrained by logical relations in semantic fields, but is more fluid
and flexible.
 Lynne Cameron

2.2.3  M
 ultiple metaphors arising from anaphoric reference and
lexico‑conceptual pacts
Extract (3) initially appears more likely as a candidate for dissonant ‘mixed meta-
phors’ when breaking the cycle is juxtaposed with bitterness, and step with listening
to a point of view:
Extract (3)
329 Pat … and I think,
330 …(1.0) the only way you can bri- –
331 well,
332 …(2.0) an essential means of er –
333 breaking that –
334 er,
335 cycle,
336 … of bitterness,
337 …(1.0) is to take that step.
338 .. you –
339 you know,
340 that preparedness,
341 you know,
342 to listen.
343 Jo .. hmh
344 Pat .. to the other point of view.

Examination of the immediately previous talk (Extract (4)) explains why the first
apparent mixing does not produce a comprehension problem. A second or so
before, Pat had introduced the idea of bitterness into the conversation, with the
contextual meaning of feeling hatred for people from the other side in the conflict:
307 Pat .. I think it’s so easy to –
308 er,
309 … perpetuate.. bitterness,
310 by –

When he speaks of breaking that cycle of bitterness, re-use of the verbal metaphor
bitterness signals that he is continuing and extending the earlier discourse topic
into the idea of a cycle, and shifting to the contrasting notion of breaking the cycle.
This continuity produces discourse coherence across 309–336 through anaphoric
reference. In the flow of talking-and-thinking between Jo and Pat, bitterness works
as a ‘lexico-conceptual pact’.6 It becomes the shared way of referring to the idea of

.  An extension of Brennan & Clark’s (1996) notion of ‘conceptual pact’.

 Chapter 2.  Mixed metaphors from a discourse dynamics perspective 

‘feeling hatred for people from the other side in the conflict’ and in re-use, it car-
ries reminders of what has already been said. Metaphors are mixed in the chain
bitterness – breaking that cycle – of bitterness, but any dissonance is only superfi-
cial; once we look more deeply into the dynamics of the particular conversation,
it dissolves.
The verbal metaphor take that step (337) is used to describe how the cycle
might be broken, as a development of the discourse topic as in 2,1 above. It follows
a one second pause (long in conversation terms) that marks, and signals to the
listener, the start of a new focal idea. The pause separates the step metaphor suf-
ficiently from the cycle metaphor to avoid a problem of dissonance.

2.2.4  Multiple metaphors combining in a metaphorical scenario

Multiple metaphors can sometimes happily combine, as in Extract (1), as parts of
a coherent metaphorical scenario (Cameron et al., 2010). Among frequently-used,
conventionalised metaphors in the data, a felicitous mixing combined metaphors
of journeys and seeing, exemplified in Extract (5):
Extract (5)
882 …(2.0) but when you start losing sight of the –
883 .. t- the –
884 the fact that you’re also harming a human being.
885 …(1.0) you lose sight of that,
886 or ignore it,
887 or you find it easier to ignore it.
888 … that’s.. always had a price.
889 …(1.0) and some way,
890 well down the line.
891 …(1.0) you know,
892 you’re going to come face-to-face with that price.

In the systematic metaphorical scenario, which is constructed across the talk, not
only in these short extracts from it, seeing metaphors are used for ­understanding/
knowing, often in a negative or restricted sense, as with glimpse in Extract (2).
Seeing the other person stands for knowing and respecting their humanity. In lines
882 and 885 of Extract (5), losing sight of the humanity of the other has the contex-
tual meaning of dehumanising them; when something is out of sight it becomes
irrelevant.
Down the line is one of many journey metaphors across the data. In their con-
versations, Jo and Pat were on separate journeys, out of grief for Jo and towards
deeper understanding of the human consequences of violence for Pat. Their paths
physically and literally crossed when they met each other face-to-face for the first
 Lynne Cameron

time. As well as each other, they meet other things on their journeys: responsibil-
ity, obstacles, welcome from other victims. In 892, Pat refers to a meeting on his
journey as coming face-to-face with his own loss of humanity. Coming face-to-face
combines journey and seeing metaphors to create the idea of meeting. While
face-to-face is a body metaphor, face also works metonymically for eyes that see
what comes in front of them.
Although verbal metaphors from the fields of journeys and seeing might
appear dissonant, and thus as candidates for being ‘mixed’, they cohere without
any problem in embodied metaphorical scenarios of people moving across a
landscape.
We can note in passing that the price metaphors in lines 888 and 892, which
also appear dissonant with the other metaphors, are in fact another example of
anaphoric reference and drawing on a lexico-conceptual pact, as in 2.2 above. The
use of a price metaphor in 888 is not a first use but relates back in the flow of
talk to earlier instances where Pat metaphorically describes his turn to violence
as having a price or a cost, creating a debt etc. The utterance in 888 re-introduces
that topic with a kind of summary of what he had said earlier about ignoring the
human aspect of violence, and the lines that follow then add to the discourse topic,
using that price as the shared phrase that captures all that has been said before on
the topic.

2.2.5  Layered metaphors

A further case of multiple, but not mixed, metaphors occurs when a metaphor
vehicle is itself metaphorised, as in Extract (6):
Extract (6)
1768 Jo and they could see,
1769 … how from my healing journey,
1770 if I could build a bridge with you,
1771 that would
1772 …(1.0) help me.

As in the previous sub-section, seeing and journey metaphors work together

unproblematically. In the phrase my healing journey, Jo brings two ideas and dis-
course topics together, with a layering effect in which healing seems to be dis-
cursively prior. As well as being a lexico-conceptual pact agreed in the dialogue,
healing is a conventionalised metaphorical way of talking about recovering from
grief in terms of recovery from physical injury. Evidence that this metaphor
belonged in Jo’s language resources from before meeting Pat, comes from its use in
a poem she wrote several years before their meeting and reads aloud in their con-
versation. The process of healing is further metaphorised as a journey, producing a
 Chapter 2.  Mixed metaphors from a discourse dynamics perspective 

‘second-order’ metaphor layered over the first-order metaphor. Journey might be

seen as ‘mixed’ if juxtaposed with healing as two first-order metaphors but, I sug-
gest, is more adequately seen as layered or embedded.

2.2.6  Multiple metaphors in metaphor clusters: Summary

In this section, I have shown how unproblematic juxtaposition of metaphors can
occur in the dynamics of discourse, arising from: shifting discourse topics, ana-
phoric reference and lexico-conceptual pacts, coherence via metaphorical sce-
narios, and layering of conventionalised and systematic metaphors. Analysis of
metaphor as a discourse dynamic phenomenon shows the ‘mixed metaphor prob-
lem’ to be much exaggerated. Multiple verbal metaphors are not stylistically-tricky
additions to the flow of talk but are constitutive of it, discursive resources that
contribute to the flow of jointly-constructed meaning. As such, meaning-making
principles (e.g., relevance) apply to verbal metaphors as to the use of other lexical
items, and problematic mixing is likely to be avoided.
In the next section, I reflect on my own processes of constructing metaphors
to use in research, and some of the insights that provides on the (non-)issue of
mixed metaphors.

2.3  Multiple metaphors in theory-building

As an applied linguist, I draw on my expertise in metaphor to address social sci-

ence issues. That process in turn develops method and theory, not least because it
involves collecting and analysing relatively large amounts of spoken data, such as
focus group discussions or the series of conversations used here, that reveal how
metaphor works in talking-and-thinking, and frequently contradict the broader
generalisations of conceptual metaphor theory. In this section, I describe some
recent theory-building with metaphor and how it seems to lead, inescapably, to
multiple metaphors (also Spiro et al., 1989).
As a recent three year project researching empathy in talk, and using
­metaphor-led discourse analysis as primary tool, comes to an end, I am currently
writing up findings. In writing for social science journals, I usually tend to avoid
too much metaphor so as not to sound ‘woolly’ or ‘unscientific’. This time, I am
experimenting with deliberate and reflective use of metaphor, to build a model of
empathy dynamics and to describe the discourse dynamics of empathy, i.e., how
people ‘do’ or resist empathy in talk with other people.
The empirical starting point for the model of empathy in dialogue was the
series of the reconciliation conversations used above, complemented with seven
 Lynne Cameron

further discourse studies. At the same time as conducting empirical analysis, I was
reviewing the empathy literature from neuroscience and psychology. Jo Berry and
Patrick Magee began their on-going dialogue from a position of great separation
or ‘alterity’ (otherness – Bakhtin, 1986) and, through talking together, came to
understand more about ‘the Other’ as a complex human being – this is the central
process of empathy. Their reflections on their processes and their dialogic jour-
neys towards understanding each other, offer metaphors for model-building. In
Extract (7), Pat reflects on Jo’s metaphor for reconciliation as building bridges,7
and turns it around to apply to the political situation in Ireland that, he felt, moti-
vated IRA violence. He produces a chain of three verbal metaphors as antonyms
of building bridges: distances, barriers, exclusions. The extract exemplifies the kind
of metaphorical language that I drew on to construct a higher-level metaphorical
model of empathy.
Extract (7)
1633 Pat there’s an <X adverse/inverse X>,
1634 to that er,
1635 …(1.0) you know,
1636 er,
1637 …(2.0) figure of speech
1638 you know,
1639 bridges.
1640 … bridges can be built.
1641 … and that is if you,
1642 .. actively –
1643 er,
1644 .. create,
1645 er,
1646 .. distances.
1647 … barriers.
1648 … or what are they?
1649 they are exclusions
1650 …(1.0) and er,
1651 .. a thing I believe absolutely fundamentally,
1652 is that er,
1653 …(1.0) if you exclude anybody’s voice,
1654 …(1.0) you know,
1655 … you’re se- –
1656 you’re sowing the seed for later violence.

.  Pat’s reference to Jo’s ‘figure of speech’ is one of very few instances where the discourse
provides evidence of metaphoricity.
 Chapter 2.  Mixed metaphors from a discourse dynamics perspective 

For the positive process of coming to know the Other, I chose to use c ­ onnecting
as the key metaphor (Cameron, 2007a, 2011). Building bridges (1639–40) was just
one type of connecting used in the data alongside others such as shared and close-
ness. Empathy as connecting was also implied through metaphorical ways of
describing lack of empathy: detached; shut out; locked out, and by metaphorical
descriptions of actions that create empathy, such as breaking down barriers. While
the individual verbal metaphors were too specific to use in a model, the label
given to their systematic metaphor worked well as part of the model: e­ mpathy in
­dialogue is connecting.
The connecting metaphor was developed in line with empirical findings
from the full set of studies, and with published studies in other disciplines, to pro-
vide a further layer to the model that describes how people ‘do’ empathy in talk.
Close discourse analysis of how Jo and Pat managed their connecting produced
this descriptive part of the model, with three types of ‘empathy gestures’, two of
which were given labels linked to connecting: ‘Allowing connection’, discourse
patterns that allow the Other access to the Self ’s thoughts and feelings, e.g., by
explaining one’s decisions or feelings to the other; ‘Entering into the Other’s world’,
discourse patterns that show an attempt to understand those thoughts and feel-
ings, e.g., by using their words or speaking from their perspective.
The model also needed to describe how empathy is negotiated and resisted
(Cameron, 2012) and so the connecting metaphor was developed by add-
ing its contrast(s). The obvious candidate would be separating as antonym of
connecting. However, other studies in the project were showing that people in
dialogue produced three kinds of discourse patterns, rather than just one, that
separated themselves from others (Cameron et al., in press). Metaphors to accu-
rately describe these discourse patterns were suggested by data such as Extract (7):
­distancing, blocking and lumping.
Let me explain how these metaphors were chosen, and why multiple meta-
phors appear to be ‘mixed’. The process of selecting the most appropriate metaphor
to use in theory-building is recursive, moving between the published literature,
empirical findings, and systematic verbal metaphors across the data. ­Distancing
describes people avoiding empathy by positioning the Other as inaccessible through
being far away and thus beyond consideration. Blocking describes people avoid-
ing empathy by putting in place some kind of affective barrier, such as stereotyping
or prejudice, that removes the need to think of others as individual and complex
human beings lumping. I judged that the other metaphorical antonym for con-
necting that Pat produces in Extract (7), exclusions, was covered by blocking,
along with other verbal metaphors such as lock me in there and closed down. Hav-
ing added the theoretical construct of ‘dyspathy’ to describe processes that work
to prevent empathy (Cameron, 2012), I now had the m ­ etaphors of ­connecting,
 Lynne Cameron

­distancing and blocking as the basis of the new model of empathy~dyspathy

dynamics ­(Cameron, 2013).
In the process of model-building, I selected the four metaphors to describe, in
combination, at a carefully constructed superordinate level, the human dialogic,
psychological and neurological processes of empathy. Although the result is what
looks like a set of ‘mixed metaphors’, the four together have ecological and con-
struct validity. I link this paradox to the feature mentioned in Section 2.2. above
and confirmed by corpus studies: that the words that people use metaphorically in
negotiating their embodied worlds with others do not always match abstract logi-
cal relations of semantic fields.
The discourse dynamics theory of metaphor holds that large-scale system-
atic metaphors that work across socio-cultural groups (and that might coincide
with what other theories would call ‘conceptual metaphors’) emerge from repeated
instances of use in dialogue and interaction (Cameron, 2010). What emerges as
part of these super-ordinate metaphors are sub-ordinate metaphors that describe
relations such as antonymy and contrast, equivalence, gradedness, and meronymy
(whole-part relations). Because these discursively-constructed relations8 do not
always coincide with abstract logical relations, the sub-ordinate level may include
metaphors that not only appear ‘mixed’ but also contravene the ‘invariance prin-
ciple’ (see Lakoff, 1993).

2.4  Conclusions

As someone who works with a socio-cultural theoretic, dialogic and dynamic view
of language and thinking (and life), I have always had problems with the strongest
claims of conceptual metaphor theory. Not of the central ideas: that metaphors
are much more common in language and thought than has been acknowledged;
that metaphors contribute importantly to thinking and to communication; that
metaphors not only work individually but also interrelate in larger systems of
meaning and lexis; that our embodied physical experience contributes to under-
standing through analogy and thus to metaphor. These ideas I continue to find
exciting and intriguing. What I cannot accept are the more grandiose claims: that
we cannot think or speak without metaphor, when we clearly can and do; that
metaphors are hard-wired into our brains in some way and available always and to

.  Discursive psychology (Edwards, 1997) insists that our only reality is our discursive con-
structions. I would not go that far but there clearly is a need to distinguish between relations
constructed discursively and those imposed by logic.
 Chapter 2.  Mixed metaphors from a discourse dynamics perspective 

e­ veryone, rather than learnt alongside everything else social and cultural that we
learn, including categories and language; and, most pertinent here, that concep-
tual metaphors are fixed and invariant mappings across domains.
My basic objection to this claim of fixedness and invariance is that it leads
to an impoverished, asocial view of metaphor that denies important realities of
human noticing, reasoning and languaging, and that I find inadequate for describ-
ing and explaining what people do with metaphor and thus what metaphor ‘is’.
Furthermore, the scientific method of reduction(ism) that seems to under-
pin a cognitive linguistic drive to peel back, abstract away, and purify concepts in
order to theorise, can be inappropriate for studying human phenomena; at some
point in the peeling back, the flesh is damaged, the object of study changes nature
and construct validity disintegrates. It is crucial to stop abstracting and gener-
alising processes carried out in the name of theory-building before this point is
reached. Removing the dynamism of metaphor mappings and relations is, I argue,
a step too far.
Rather than working from the ‘generalization commitment’, that aims “to
characterise the general principles governing all aspects of human language”
(Lakoff  1990, p. 53), the ‘discourse dynamic commitment’ seeks to understand
how actual people use language resources in the shifts and flow of dialogue and
interaction. This chapter has illustrated how the ‘problem’ of mixed metaphors
disappears when we examine metaphor in dialogue as people talk-and-think
together. The use of multiple metaphors in human understanding is necessary and
inevitable; that our multiple metaphors sometimes seem to clash is mostly an illu-
sion perceived from outside of discourse.

References

Bakhtin, M. M. (1986). Speech genres and other late essays. Austin: University of Texas Press.
Brennan, S., & Clark, H. (1996). Conceptual pacts and lexical choices in conversation. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 22(6), 1482–1493.
doi: 10.1037/0278-7393.22.6.1482
Cameron, L. (2003). Metaphor in educational discourse. London: Continuum.
Cameron, L. (2007). Patterns of metaphor use in reconciliation talk. Discourse and Society,
18(2), 197–222.  doi: 10.1177/0957926507073376
Cameron, L. (2008). Metaphor shifting in the dynamics of talk. In M. S. Zanotto, L. C
­ ameron, &
M. Cavalcanti (Eds.), Confronting metaphor in use (pp. 45–62). Amsterdam: John
Benjamins.  doi: 10.1075/pbns.173.04cam
Cameron, L. (2010). The discourse dynamics framework for metaphor. In L. Cameron &
R. Maslen (Eds.), Metaphor analysis: Research practice in applied linguistics, social sciences
and the humanities (pp. 77–94). London: Equinox.
Cameron, L. (2011). Metaphor and reconciliation. New York: Routledge.
 Lynne Cameron

Cameron, L. (2012). Dyspathy: The dynamic complement of empathy. Living with Uncertainty
Working Paper, 5. Retrieved from http://www.open.ac.uk/researchprojects/livingwithuncer
tainty/p5.shtml
Cameron, L. (2013). A dynamic model of empathy and dyspathy. Living with ­Uncertainty Working
Paper, 6. Retrieved from: http://www.open.ac.uk/researchprojects/livingwithuncertainty/
sites/www.open.ac.uk.researchprojects.livingwithuncertainty/files/files/6%20Empathy%20
model.pdf
Cameron, L., & Maslen, R. (Eds.). (2010). Metaphor analysis: Research practice in applied
linguistics, social sciences and the humanities. London: Equinox.
Cameron, L., Maslen, R., & Low, G. (2010). Finding systematicity in metaphor use. In
L. ­Cameron & R. Maslen (Eds.), Metaphor analysis: Research practice in applied linguistics,
social sciences and the humanities (pp. 116–146). London: Equinox.
Cameron, L., Maslen, R., & Todd, Z. (In press). The dialogic construction of self and other after
terrorism. Journal of Peace and Conflict.
Cameron, L., Maslen, R., Todd, Z., Maule, J., Stratton, P., & Stanley, N. (2009). The discourse
dynamics approach to metaphor and metaphor-led discourse analysis. Metaphor & Sym-
bol, 24(2), 63–89.  doi: 10.1080/10926480902830821
Cameron, L., & Stelma, J. (2004). Metaphor clusters in discourse. Journal of Applied Linguistics,
1(2), 7–36.  doi: 10.1558/japl.2004.1.2.107
Chafe, W. (1996). Discourse, consciousness and time. Chicago: University of Chicago Press.
Edwards, D. (1997). Discourse and cognition. London: Sage Publications.
Lakoff, G. (1990). The invariance hypothesis: Is abstract reason based on image schemas? Cogni-
tive Linguistics, 1(1), 39–74.  doi: 10.1515/cogl.1990.1.1.39
Lakoff, G. (1993). The contemporary theory of metaphor. In A. Ortony (Ed.), Metaphor and
thought (2nd ed., pp. 202–251). New York: Cambridge University Press.
doi: 10.1017/CBO9781139173865.013
McCarthy, M. (1988). Some vocabulary patterns in conversation. In R. Carter & M. McCarthy
(Eds.), Vocabulary and language teaching (pp. 181–200). London: Longman.
pragglejaz group. (2007). MIP: A method for identifying metaphorically-used words in dis-
course. Metaphor and Symbol, 22(1), 1–40.  doi: 10.1080/10926480709336752.
Slobin, D. (1996). From “Thought and Language” to “Thinking for Speaking”. In J. Gumperz &
S. Levinson (Eds.), Rethinking linguistic relativity (pp. 70–96). New York: Cambridge
­University Press.
Spiro, R., Feltovitch, P., Coulson, R., & Anderson, D. (1989). Multiple analogies for complex con-
cepts: antidotes for analogy-induced misconception in advanced knowledge acquisition.
In S. Vosniadou & A. Ortony (Eds.), Similarity and analogical reasoning (pp. ­498–531).
­Cambridge: Cambridge University Pres.  doi: 10.1017/CBO9780511529863.023
 chapter 3

Why mixed metaphors make sense

Cornelia Müller
European University Viadrina

This paper explores why speakers and addressees seem to have no problem in
making sense of mixed metaphors. We will argue that the mixing of metaphors
reveals something about the nature of conventionalized metaphoric meaning that
is as interesting for cognitive linguists as speech errors are for psycholinguists.
First, it shows that so-called dead metaphors are alive for speakers, second it
reveals that people deal creatively with all the meaning facets of metaphoric
meaning – including the uncommon ones, and third we will argue that
the mixing of metaphors can be explained by assuming a dynamic view on
metaphoric meaning making. This view suggests that rather than being static and
fixed, metaphoric meaning is the product of a process of cognitively activating
selected facets of source and target, or vehicle and tenor. As a consequence the
mixing of metaphors is considered a result of a shifting focus of attention, or
of dynamically foregrounding facets of meaning that are backgrounded in the
common reading.

3.1  Introduction

Mixed metaphors usually make perfect sense for a speaker and an addressee at
a given moment in time, even if a speaker’s intentions run counter to conven-
tionalized metaphoric meanings and favor a non-conventional reading. Mixed
metaphors may make sense, because people attend to meaning in a dynamic and
flexible manner, responding to the moment-by-moment affordances of the com-
municative encounter they are immersed in. Mixing metaphors is a consequence
of shifting one’s attention to uncommon aspects of metaphoric meaning. By fore-
grounding what is being backgrounded in the standard reading of ­conventionalized

doi 10.1075/milcc.6.03mul
© 2016 John Benjamins Publishing Company
 Cornelia Müller

­ etaphors the mixing of metaphors changes the semantic salience structure, cre-
m
ating different versions and degrees of activated metaphoric meaning.1
Such a dynamic view on metaphoric meaning is not a position favored in the
literature on style and rhetoric. By describing the mixing of metaphors as a practice
that is “widely regarded as a stylistic flaw caused by unthinkingly mixing ‘clichés’”
(McArthur 1992: 663), The Oxford Companion to the English Language represents
in fact an opposed point of view. This commonplace stance considers the mixing
of metaphors to be a consequence of not being aware of the metaphoricity of an
expression. It is assumed that it occurs unwittingly and happens because the literal
meaning of a metaphoric expression is cognitively not activated during language
production. When producing a mixed metaphoric expression speakers just don’t
think about the ‘literal’ reading of the metaphor. The Oxford Companion offers the
following example:
ENGL: The butter mountain has been in the pipeline for some time.
FIG: ‘As a result of the overproduction of butter in the EU, huge amounts of
it have been awaiting distribution for some time.’
LIT: ‘A real mountain consisting of butter is stuck in a kind of oil pipeline.’

In the BBC News from 1987, the president of the Farmer’s Union mixes two con-
ventional metaphoric expressions: the compound “butter mountain” and the idi-
omatic expression “to be in the pipeline”. With regard to their figurative meaning
the two metaphors work well together but they do not combine on the literal level.
So, here we have a nice example for what has been considered the decisive seman-
tic structure of mixed metaphors. The analysis offered appears straightforward.
However, we believe that the mixing of metaphors calls for a more differentiated
account. The goal of this chapter is to lay out, how and in what sense mixing meta-
phors is more complex, and more interesting too, than to be dealt with as stylistic
error solely.
In support of this assumption cognitive-linguistic analyses of different kinds
of mixed metaphors will be presented. Based on these analyses, we will emphasize
that the mixing of metaphors is based on the foregrounding of uncommon aspects
of meaning. Beforehand, however, we will discuss, why Linguistic Metaphor
­Theories and Conceptual Metaphor Theory conceive of the mixing of metaphors

.  This paper is based on a book chapter in “Metaphors Dead and Alive, Sleeping and
Waking” by the author, published in (2008a). In dealing with mixed metaphors the paper also
provides a condensed presentation of a usage-based account to metaphor theory ‘a dynamic
view’, which has been developed in book-length detail in Müller (2008a). Over the course of
this paper we will now and then hint to more detailed analyses and argumentations provided
in the book.
 Chapter 3.  Why mixed metaphors make sense 

as problematic in some way. In a conclusion we will outline the theoretical con-

sequences of such a flexibility and creativeness in the online creation of emergent
metaphoric meaning and claim that the cognitive rationale for the production of
mixed metaphors is that metaphors are products of a cognitive process of activat-
ing metaphoricity.

3.2  Why should mixing metaphors be problematic?

Linguistic Metaphor Theories assume that metaphors require a dual seman-

tic structure and they locate metaphoric meaning on the level of the seman-
tics of language only. For them metaphor is matter of language, not of thought
­(Müller 2008a: 40–61). Linguistic Metaphor Theories (LMT) have often been char-
acterized as traditional theories of metaphor (Lakoff 1993, Lakoff & Turner 1989).
Indeed, this view of metaphor does have a long tradition going back to the actual
coining of the term metaphor by the Greek philosopher Aristotle. In a famous
passage of his Poetics Aristotle reflects about the different ways in which a word
(ónoma) can be applied, metaphor being one of them.
“Metaphor (metáphorá) is the imposition […] of a foreign name, a transfer either
from genus to species, or from the species to the species, or according to what is
analogous.” (Aristotle Poetics XXI.1457b, transl. by E.M. Cope 1970: 374)

Metaphor is conceived as the transfer of a word from one place in the language to
another one and the etymology of metaphor indicates this idea of transfer too: in
Ancient Greek metáphorá means transfer of something or somebody (Pape 1914).
The verb from which the Greek noun is derived expresses the idea of moving an
object from one place to another: metapherro signifies “to transfer something or
to move something to some other place” (Pape 1914). The basic sense of the word
remains vivid until today in Modern Greek: moving companies in Greece often
carry the noun metaphorein on their trucks.
The quotation and the etymology of the word indicate that the Aristotelian
understanding of metaphor is a Linguistic Theory of Metaphor: it addresses the
moving of words from one place in the language (e.g., from one kind of mean-
ing) to another one: Linguistic Metaphor theories consider metaphor as a matter
of words and how they can be applied. This has been the predominant view on
metaphor for most of the time since Aristotle’s treatment and it has influenced
many different strands of research, among them: in structural and formal seman-
tics (Cohen 1993; Abraham 1998; Davidson 1978; van Dijk 1975;), in pragmat-
ics (Cohen 1975; Grice 1975; Sadock 1993; Searle 1993; Sperber and Wilson
1986), in semiotics (Eco 1984; Köller 1975, 1986), and in stylistics and literature
 Cornelia Müller

­(Friedrich  1968; ­Kayser 1976; Bertau 1996; Brooke-Rose 1958; Coenen 2002;
Lausberg 1960; Kubczak 1978; Meier 1963).
For advocates of Linguistic Metaphor Theories metaphor is a specific
form of language or a particular linguistic phenomenon (for more detail, see
­Müller 2008a: 52–61). Linguistic metaphor theories contrast with Cognitive Meta-
phor Theories (CogMT), which consider metaphor either as a specific form of
thought, or more generally as a cognitive phenomenon. When it comes to assert-
ing the nature of metaphor Cognitive Metaphor Theories, in turn, fall into two big
strands: either they share basic LMT assumptions or they assume that metaphor is
a phenomenon of general cognition (Katz 1998 for a similar distinction).
Cognitive Metaphor Theories that adhere to the belief of a linguistic character
of metaphor are prominent in psycholinguistics and cognitive science more gen-
erally and call for an integration of linguistic metaphor processing in theories of
speech production and comprehension (cf. a selection of references: Glucksberg
and Keysar 1990, 1993; Keysar and Bly 1995; McGlone 1996; Miller 1993; ­Murphy
1996, 1997; Ortony 1988; Paivio and Walsh 1993; Stock, Slack, and Ortony 1993;
Winner and Gardener 1993). Proponents of this type of Cognitive Metaphor The-
ory tend to depart from the assumptions that metaphor is a matter of poetic lan-
guage primarily, that it is an aberration of the default case of literal language use
because it violates truth conditions and in doing so figurative language use distorts
reality. Therefore the cognitive processing of metaphor has to be modeled differ-
ently from literal language processing and is supposed to take more time than the
processing of figurative language (cf. Katz 1998).
Another type of Cognitive Theories with a bias towards linguistic metaphor
theory can be found in the philosophical, rhetorical, linguistic, and stylistic tra-
dition of metaphor research. Renowned proponents of this type of theory have
become very influential: Ivor A. Richards Philosophy of Rhetoric is probably the
first Anglo-Saxon formulation of metaphor as a form of thought (Richards 1936),
Max Black’s Models and Metaphors and his interaction theory of metaphor remains
a touchstone especially in literature studies (Black 1962), and in the ­German tradi-
tion Harald Weinrich publishes Sprache in Texten (Language in Texts) and develops
a notion of verbal metaphors that belong to image-offering and image-receiving
fields (Weinrich 1976). These scholars share the assumption that metaphor is
a matter of thought and a matter of language, they also believe that metaphor
belongs to the realm of poetic language (primarily), that it is an aberration from
the default, a violation of a proposition or a pragmatic rule, a tension between
focus and frame, or a contradictory predication. On the other hand they hold that
metaphor is an omnipresent principle of language, that metaphors are structured
in image fields (semantic fields within a language), or that metaphors are basically
mental models.
 Chapter 3.  Why mixed metaphors make sense 

In some parts of their work they even seem to be inclined to favor a Cogni-
tive Theory of Metaphor with a bias towards general cognition. So for Richards
metaphor is having two thoughts of different things together (Richards 1936), for
Black metaphors are mental models and may reveal how things are (Black 1962),
for Weinrich metaphors are coupled image fields which constitute a world-view,
guide inferences, host and stimulate verbal metaphors (Weinrich 1976).
Lakoff and Johnsons Conceptual Metaphor Theory (CMT) appears to push
this tradition further towards a Cognitive Theory that regards metaphor as a fun-
damental principle of human cognition (Lakoff and Johnson 1980, 1999). In a
nutshell CMT transforms metaphor theory into an experientially grounded the-
ory of human understanding. Metaphor is conceived as a primarily non-linguistic
form of organizing experience and a principle of understanding. Other Cognitive
Theories of Metaphor that consider metaphor a principle of general cognition can
be found in cognitive anthropology, sociology, and linguistics. Here it is assumed
that metaphors come with conceptualizations of experience and may influence
learning, reasoning, and understanding. Adherents to this tradition tend to think
of metaphor in terms of cultural models that shape collective ways of thinking
about the world. Examples are Reddy’s conduit metaphor of communication
(Reddy 1993), Schön’s analysis of social policy (Schön 1993), Quinn’s analysis of
concepts of marriage (Quinn 1999a, 1999b) or Shore’s analysis of baseball games
(Shore 1996).
However, the most influential current Cognitive Metaphor Theory is certainly
Lakoff and Johnson’s CMT. Scholars working with the CMT framework tend to
hold that metaphor is a general principle of conceptual organization, a fundamen-
tal form of thought and understanding and accordingly a pervasive and funda-
mental principle of ordinary language. CMT holds that language is fundamentally
figurative, i.e., metaphor is the default case, not literal meaning.
While Linguistic Metaphor Theories assume that metaphor is a phenomenon
of language and with Aristotle that metaphor involves the transfer of meaning
from one place in the language to another one, Conceptual Metaphor Theory
argues that metaphor is a principle of thought in which experiencing and under-
standing one thing in terms of another is held to be the primary nature of meta-
phor. For our discussion of mixed metaphors, and why this has been considered as
problematic for LMT and CogMT, we need to consider their differing assumptions
about the nature of metaphor.
For Linguistic Metaphor Theories the duality of metaphoric meaning is a
core characteristic of the nature of metaphor. It is what constitutes the vitality of
metaphors. A metaphor is ‘live’ or ‘vital’ if and if only literal and figurative mean-
ings are recognized by an idealized speaker/hearer, reader/writer – otherwise it
is regarded ‘dead’. In short, a vital metaphor is a metaphoric expression a speaker
 Cornelia Müller

(and/or ­listener) is (typically) consciously aware of. A logical prerequisite of this

is that the metaphoric expression is transparent. The third criterion for consider-
ing a metaphor a vital one is non-conventionality. This helps to distinguish new
and original metaphors from conventional ones. Vital metaphors differ from
waking metaphors with regard to conventionalization. A waking metaphor that
metaphoric expressing that would be transparent and conventional would be con-
sidered a waking metaphor. It could be consciously used or not.
Consider a classic source for this assumption:

“A so-called dead metaphor is not a metaphor at all, but merely an expression that
no longer has a pregnant metaphorical use. A competent reader is not expected to
recognize such a familiar expression as ‘falling in love’ as a metaphor, to be taken
au grand sérieux. Indeed it is doubtful whether that expression was ever more
than a case of catachresis”. (Black 1993: 25)

In this passage Black makes clear that “so-called dead metaphors” are not part of
what metaphor theories should deal with, since they have seized to be metaphors
altogether. Dead metaphors are no longer used and recognized as metaphors and
they may have never been understood as such. The notion of dead metaphor that
Black expresses here is characteristic for Linguistic Metaphor Theories and it is
crucial for their understanding of mixed metaphors. Dead metaphors are con-
ventionalized expressions that have lost (or have never had) a double meaning,
i.e., a meaning in which both the literal and the figurative sense were co-present.
In this perspective the conventionalization of metaphors leads automatically to a
decrease in the awareness of metaphoricity and often also to a loss of transparency.
And since they are dead, Linguistic Metaphor Theories consider them not to be
metaphors at all.
Mixed metaphors, however, for the most part, concern precisely those kinds
of metaphors that are considered dead in the LMT sense. This is an interesting
fact, because if Linguistic Metaphor Theories were right, then the mixing of meta-
phors should not happen at all. Also there is a contradiction in claiming on the one
hand that conventionalized metaphors are dead, because their metaphoricity is
not vital, while on the other hand characterizing the mixing of clichéd metaphors
as a thinking flaw. LMT assumptions are contradictory here. If, preventing speak-
ers and writers from a stylistic fault, requires awareness of the double meaning of
a metaphor, then proponents of traditional accounts of metaphor require a facet of
metaphoric meaning to be cognitively active that they claimed to be not available
when characterizing conventional metaphors as dead ones. We will argue, on the
contrary, that dead metaphors appear to have some life, at least in the less con-
scious regions of linguistic attention and we intend to show in this chapter that the
mixing of metaphors provides nice evidence for the claim that dead metaphors are
 Chapter 3.  Why mixed metaphors make sense 

in fact very much alive. Black’s position is representative for Linguistic Metaphor
Theories and his opinion continues to express a widely shared, almost canonical,
view on conventionalized metaphors. A view that Gibbs in his book ‘The Poetics
of Mind’ critically comments on:

“Even though many instances of contemporary speech have obvious figurative

roots, most scholars assume that idiomatic language may once have been
metaphorical but has lost its metaphoricity over time and now exists in the
mental lexicon as a set of stock formulas or as dead metaphors. Just as speakers
no longer view leg of table as metaphoric, few people recognize phrases such as
spill the beans, blow your stack, off the wall, in the pits, or a rolling stone gathers no
moss as being particularly creative or metaphoric. After all, metaphors are lively,
creative, and resistant to paraphrase, whereas idioms, chlichés, and proverbs are
hackneyed expressions that are equivalent in meaning to simple literal phrases.
To classify some utterance or phrase as “idiomatic,” “slang,” or “proverbal” is
tantamount to a theoretical explanation in itself, given the widely held view that
such phrases are dead metaphors that belong in the wastebasket of formulas and
phrases that are separate from the generative component of a grammar.”
(Gibbs 1994, 267–8; emphases in the original)

The traditional, linguistic perspective on metaphor has been challenged by Con-

ceptual Metaphor Theory proponents who turned the argument upside down:
metaphors that are conventionalized and omnipresent in ordinary language are
the ones which are deeply entrenched in the conceptual systems of language users.
They are the metaphors that guide thinking and experiencing of meaning in a
mundane and subtle way. In short, what are considered dead metaphors in Lin-
guistic Metaphor Theories is what is regarded as most ‘alive’ in Conceptual Meta-
phor Theory.

“The ‘dead metaphor’ account misses an important point; namely, that what is
deeply entrenched, hardly noticed, and thus effortlessly used is most active in our
thought. The metaphors above may be highly conventional and effortlessly used,
but this does not mean that they have lost their vigor in thought and that they are
dead. On the contrary, they are ‘alive’ in the most important sense–they govern
our thought–they are “metaphors we live by”. (Kövecses 2002, IX)

Kövecses addresses exactly the type of metaphors that Black characterizes as dead.
The controversy is about metaphoric idiomatic expressions as well as metaphori-
cally used lexemes. Examples are: “falling in love,” “driving someone crazy,” and
“She is hungry for knowledge.” or “The jacket I saw in the shop window pulled
me into the store” (Black 1993, 25; Kövecses 2002: 20–22). Conceptual Metaphor
Theory holds that language users conceive of conventionalized metaphors as
metaphors – but they do so on a different level of consciousness: “Our system of
 Cornelia Müller

conventional metaphor is ‘alive’ in the same sense that our system of grammati-
cal and phonological rules is alive; namely, it is constantly in use, automatically,
and below the level of consciousness” (Lakoff 1993: 245). Thus CMT assumes that
the metaphoricity of lexicalized metaphors is active on the level of non-conscious
processes. It is considered an entailment of activated conceptual metaphors, which
in turn are based on entrenched mappings between a source and a target domain.
Conceptual metaphors are processed as unconsciously and automatically as other
structural linguistic knowledge: “Our system of conventional metaphor is mostly
unconscious, automatic, and used with no noticeable effort, just like our linguistic
system and the rest of our conceptual system” (Lakoff 1993: 245). Active mappings
between a source and target domain for conventional metaphors would explain
why it does not make sense to combine metaphors using different and incompatible
source domains, because this would result in a contradictory conceptualization.
In Conceptual Metaphor Theory mixed metaphors are discussed in the context
of consistency and coherence across conceptual metaphors. They are described as
impermissible combinations of underlying conceptual metaphors and the reason
for being impermissible is their lack of shared entailments. Permissible combina-
tions would involve at least one shared entailment, e.g., at least one epistemic corre-
spondence between metaphors (Lakoff and Johnson 1980: 92; Lakoff and Kövecses
1987: 201). A permissible combination would create coherence across conceptual
metaphors but could involve inconsistency with regard to some aspects of the
source domains (Lakoff and Johnson 1980: 92–95; ­Müller 2008a: 138–143). On the
basis of the analysis of the conceptual metaphors an argument is a ­journey and
an argument is a container Lakoff and Johnson argue that the two are coher-
ent but not fully consistent, because “there is no single image that completely fits
both metaphors” (Lakoff and Johnson 1980: 94).
CMT thus provides an explanation why it could be problematic to mix meta-
phors, but it does not explain, why people nevertheless do it. If, indeed, the con-
ceptual system underlying verbal metaphoric expressions is active and alive all
the time during language production, it should hinder speakers from combining
metaphors with inconsistent aspects of semantic, syntactic and pragmatic mean-
ing. If all conventionalized metaphors guide the way we think in the same way,
then the combination of contradictory source domains should not happen. Yet it
does – and even quite frequently and in all kinds of natural discourses.
To conclude, both Linguistic as well as conceptual Metaphor Theories have
problems explaining the phenomenon of mixing metaphors appropriately. Both
do not explain how metaphors relate to the individual mind and how conscious-
ness and attention play out in the process of metaphor use. This is why they both
have difficulties seeing the rationale guiding this widespread linguistic phenom-
enon. We are suggesting that not the mixing of metaphors in problematic, but that
 Chapter 3.  Why mixed metaphors make sense 

the theories of metaphors cannot account for dynamically emerging metaphoric

meaning, because both of them focus on products of language use rather than the
processes which bring these individual and subjective forms of metaphoric mean-
ing about. Mixing metaphors appears problematic from a normative and collective
view on language use, not from a subjective point of the speaker.
This, however, is the point I would like to further develop in this chapter. An
interesting empirical study that indicates such a subjective perspective is Cienki &
Swan’s (2001) study on multi-part metaphoric expressions in naturalistic con-
versations, and which is the line of argument we will pursue. For this particular
research Cienki and Swan coded three conversations between students for all syn-
tactic phrases that contained a combination of at least two conceptual metaphors.
The conversations were elicited by a set of questions regarding the moral standards
of exam taking. Although Cienki and Swan’s focus in this paper was on the coher-
ence and consistency of conceptual metaphors (including the sharing of entail-
ments), their results are highly interesting for research on mixed metaphors. What
they found were three different types of multi-part metaphoric expressions; two of
them, we would suggest, are cases of mixed metaphors:

a. multi-part expressions in which cross-metaphorical coherence is most easily

achieved (Cienki & Swan 2001: 12–15)
b. multi-part expressions in which cross-metaphorical coherence is less straight-
forward (Cienki & Swan 2001: 15–17)
c. multi-part expressions in which cross-metaphorical coherence is problematic
(Cienki & Swan 2001: 17–28).

The expressions characterized as showing less or even problematic coherence

between conceptual metaphors can be considered cases of mixed metaphors.
In fact, the authors discuss one example that structurally resembles one of the
examples to be discussed later in this chapter: “So maybe I don’t uphold as strong
of a moral code y know” (Cienki and Swan 2001: 18, underlining for metaphoric
expressions in the original). Although the source domains do not overlap the
insertion of the adjective is perfect syntactically. Cienki and Swan argue that “we
can talk about upholding the law, a standard, a moral code, etc.” but this “is com-
monly elaborated on with an adjective like high, so, upholding high standards”
(Cienki and Swan 2001: 18, underlining for metaphoric expressions in the origi-
nal). Combining this construction with strong as adjective is uncommon and
brings in a conceptual metaphor whose entailments do not interweave nicely with
the conceptual metaphors on which “upholding a moral code” operates. Cienki
and Swan then provide a blending analysis to reconstruct the levels of metaphoric
meaning on which this combination of conceptual metaphors makes sense. They
 Cornelia Müller

also point out that over the three conversations they coded, there was no case in
which an interlocutor did react to the problematic combinations of metaphors. So,
apparently, mixing metaphors is not problematic for conversation partners.
“But we observed in the data examined here that multi-part metaphoric
expressions with non-coherent entailments did not appear problematic for either
the speakers or listeners. Even when the speakers in our data constructed blends
which did not cohere syntactically or conceptually, addressees did not let this
deter from the flow of conversation, but made sense out of the given phrase in
some way, as evidenced by their continuation of the topic rather than asking for
clarification.” (Cienki and Swan 2001, 28–29; emphasis CM)

As Cienki and Swan observe, speakers and addressees do not find it problematic
to make sense of mixed metaphors at all, they just assume that what has been
said is meaningful. This observation provides further grounds for proposing that
mixing of metaphors appears to be problematic not for the language users but
for metaphor theories: because those cannot account sufficiently for this dynamic
and uncommon way of metaphoric meaning making. In the following section,
we will suggest that people’s motivation to mix metaphors is a subjective online
process of activating metaphoricity in which uncommon aspects of meaning are
foregrounded.

3.3  Mixed metaphors foreground uncommon aspects of meaning

It is an intriguing fact that people seem to be able to make sense even of the strangest
combinations of metaphors. As Gibbs points out: “we understand what the original
speakers must have intended with each of these examples” (Gibbs 1994: 4). How is
it possible that we as analysts and as participants in conversations make so easily
sense of mixed metaphors? Maybe it is because “people have a strong, natural dispo-
sition to attribute intentionality to human language and action” (Gibbs 1999: 19)?
If so, can we spell out more explicitly, what kind of senses these are? Do we find
systematic digressions from a correct reading, or are those mixed metaphors ran-
dom errors? For Linguistic Metaphor Theories the answer is relatively simple:
mixed metaphors make sense on the level of figurative meaning, the semantic vio-
lation concerns the literal level only. In terms of Conceptual Metaphor Theory
the non-permissive combinations of two metaphors results from a combination
of non-coherent entailments of conceptual metaphors ­(Müller  2008: ­138–147).
A look at three exemplary cases of mixed metaphors intends to shed some light
on these questions (Müller 1998a: 134–177). Note, that the examples discussed
subsequently are all cases of mixed metaphors taken from English and German
 Chapter 3.  Why mixed metaphors make sense 

reference books on style and as such they document the intuition of experts on
style on what is considered to be a mixed metaphor.
In the case of the above mentioned butter mountain example, the mixing of
metaphors is realized by combining a noun phrase (a metaphoric compound)
with an attributional verb phrase (including a metaphoric prepositional attribute)
(cf. Müller 2008a: 143–147, 164–167). In the metaphoric compound butter moun-
tain, mountain is mapped onto the target domain of quantification and, hence,
serves to express a large stock of piled up butter. The metaphoric prepositional
attribute in the pipeline, in turn, maps onto the target domain of transport and
specifies a means of steady distribution, with the preposition in locating the sub-
stances. In the conventional reading of the idiomatic prepositional phrase, these
would typically be some kind of newsworthy information.

ENGL: The butter mountain has been in the pipeline for some time.
FIG: ‘As a result of the overproduction of butter in the EU, huge amounts of
it have been awaiting distribution for some time.’
LIT: ‘A real mountain consisting of butter is stuck in a kind of oil pipeline.’

What happens in the mixing of the two metaphors in this sentence is that the prep-
ositional phrase attributes impossible properties to the metaphoric compound–at
least if we depart from a ‘real’ world ontology of the metaphoric source domains.
Clearly, in the ‘real world’, a butter mountain does not fit into a pipeline. The mix-
ing of metaphors creates a semantic inconsistency on the level of the literal mean-
ing. If, however, source domain information is suppressed and the focus is solely
on the target domains, the combination of the two metaphors does actually not
interfere and could be paraphrased as follows: “As a result of the overproduction
of butter in the EU, huge amounts of it have been awaiting distribution for some
time”. The combination of the two metaphoric expressions is viable on the level of
figurative, but not on the level of literal meaning. Put differently, this is a canonical
case of a mixed metaphor, one that reference works on style regard as a stylistic
fault and describe as a consequence of a “thinking flaw”.
However, what, at first glance, might well look like a lack of linguistic com-
petence, when considered more carefully, shows how speakers use and combine
metaphoric meaning online as they speak. We suggest that the speaker’s attention
was focused only on the figurative meaning, while suppressing the literal meaning
and the source domain information. As a result, only the figurative meaning was
foregrounded. In addition to the particular attentional focus, what from a per-
spective of blending analysis (Grady, Oakeley and Coulson 1999; Fauconnier and
Turner 2002) might have semantically motivated this mix are the semantic roles
of the two blended metaphoric spaces which go together perfectly well: undergoer
(butter mountain – input space 1) and action operating upon undergoer (has been
 Cornelia Müller

in the pipeline – input space 2) (Figure 3.1). The metaphoric blend then operates
on the projection of those two semantic elements: a huge amount of something
(undergoer – input space 1) and something awaiting distribution (action operat-
ing upon undergoer – input space 2). The blend is based on the general idea that
there is some kind of entity that has been awaiting distribution for a long time.
What does not go into the blend are the specific properties of the entity and of the
medium of transportation. Regarded in this way, the uncommon combination of
metaphors appears to be motivated by a particular focus of the speaker’s attention.
The matching semantic structures reveal what might have motivated this particu-
lar form of combining metaphoric meaning.

GENERIC SPACE

- undergoer
- action operating
upon undergoer

INPUT SPACE 1 INPUT SPACE 2

‘butter mountain’ ‘in the pipeline’

a huge amount (a huge amount
of something of something)
(awaits distribution) awaits distribution

sthg = solid sthg = liquid

substance (butter) substance
(oil, water, gaz)

The buttermountain temporal and

has been aspectual
in the pipeline framing
for some time

BLENDED SPACE

Figure 3.1.  The ‘butter mountain example’ from a blending analysis point of view.
­(Müller1998a:  166)
 Chapter 3.  Why mixed metaphors make sense 

It is this kind of uncommon combination of different metaphors which reveals

that the salience of elements of meaning depends upon a specific choice a speaker
makes, be it a specific viewpoint he or she adopts or a specific intention fol-
lowed. The salience structures of those elements of meaning that go into meta-
phoric expressions and also into their mixings are apparently created ad hoc, in
the moment of speaking. Metaphoricity cannot be properly accounted for as a
fixed property of words or thoughts, rather we must think of it as being created
and activated on the spot. Sometimes in the flow of speech, attention digresses to
uncommon aspects of meaning and violates established and entrenched routes of
meaning construction. This is what we see in the mixing of metaphors and this is
what the next examples will illustrate.
In the following example, we find such a less common case: the mixing of met-
aphors in a conditional sentence. This example (we will term it ‘the rope example’)
differs from the canonical examples of mixed metaphors in that we find semantic
inconsistencies with regard to the literal as well as to the figurative level of mean-
ing (cf. Müller 2008a: 148–154, 168–169).
GERM: Wenn alle Stricke reißen, hänge ich mich auf.
FIG: If all else fails, I’ll give up.
LIT: If all ropes break, I’ll hang myself up.

The example is taken from a German reference book on linguistics (Bußmann

1990: 371), and it has circulated in German works on style for many years, becom-
ing a landmark case of a mixed metaphor.2 Among interlocutors however, it often
stimulates laughter rather than disapproval. But, however this sentence is assessed,
it is always treated as a formulation that is somehow deviant. It is clear, that the
sentence is semantically inconsistent regarding its literal meaning. It combines
two idioms that, when taken literally, produce a logical contradiction: with broken
ropes, it is hard to hang oneself. This incongruence is strengthened by the seman-
tic implications of the conditional construction. The conditional structure sets
up a temporal and often causal relation between two events or actions, such that
the first event comes first and provides the necessary conditions for the second
event to take place.3 In the rope example, the conditions set up by the first idiom
make the action referred to in the second idiom impossible. In other words, they
­interfere with the semantic implications of the conditional clause. Once the ropes
are broken, they can no longer be used for hanging anything up.

.  See, for instance, Braak (1974). It is one of the most referred to examples in German lin-
guistic reference literature, much as “Green ideas sleep furiously” or “The cat is on the mat” in
the Anglo-Saxon world.
.  This is characteristic of conditional clauses in general (Eisenberg 1986, Sweetser 1996).
 Cornelia Müller

This means, however, that on the literal level, the digression from an uncon-
ventional reading concerns not only and maybe even not primarily two incom-
patible experiential source domains. At the least, they operate on experiential
domains that are somehow related, both having something to do with ropes. The
incongruence is not just a matter of wrongly combining two verbal metaphors
from different experiential domains, which do not make sense in ‘real world’
ontology, as in the butter mountain example. On the contrary, what characterizes
the rope example is that the semantic inconsistency is reinforced, if not produced,
by the semantic implications of the conditional clause.
A further contrast to the rope example is that the sentence is also semanti-
cally inconsistent on the level of the figurative meaning. The protasis “Wenn alle
Stricke reißen” (‘if all else fails’) implies a positive turn in the dependent clause,
yet the apodosis brings in a negative one “hänge ich mich auf ” (‘I will give up’). In
addition, the second idiom carries presuppositions, which are violated too. Typi-
cally, ‘hänge ich mich auf ’ (‘I’ll hang myself up’, ‘giving up’) is preceded by repeated
unsuccessful trials of the same kind. Yet, the protasis ‘Wenn alle Stricke reißen’
(‘If all ropes break’, ‘If all else fails’) describes various possibly successful ways of
solving a problem, not the repetition of one activity. In short, the rope example
comes with semantic inconsistencies on the literal as well as on the figurative level,
including violations of the implications of the conditional sentence.
What might have prompted the speaker/writer, to create this particular mix of
two metaphors? We are suggesting that there is a sense in which these violations of
semantic structures may have been fully reasonable for a speaker/writer. Presum-
ably her attentional focus was on core aspects of the figurative meaning, disregard-
ing implications and presuppositions of the conditional clause, on the figurative
as well as on the literal level of meaning. A blending analysis shows further what
could have been the underlying logic she followed (Figure 3.2).
Notably, the protasis comes with a specific event structure that matches well
with the event structure of the apodosis. Input space (1) brings in the idea of ‘If all
activity options have been tried without success…’ that matches with the idea for-
mulated in the apodosis ‘… no further activities will be realized’ (input space 2).
What the speaker/writer presumably conceptually integrated, is the bare condi-
tional structure without the semantic implications of the idioms (figurative and
literal meaning) and two matching interrelated event structures. This is, at least,
what is foregrounded and the speaker’s focal attention on these elements of mean-
ing might have motivated this unconventional way of combining metaphors.
The last example (‘the molting river example’) that we would like to discuss
comes from a newspaper article and is discussed as a case of mixed metaphor in
a German work on stylistics (Schneider 1999: 237). The mixing is realized in yet
another type of syntactic structure: two metaphoric attributes specify a noun, with
 Chapter 3.  Why mixed metaphors make sense 

GENERIC SPACE

Conditional
If - then

INPUT SPACE 1 INPUT SPACE 2

Protasis Apodosis
Event structure (P) Event structure (A)
“If all activity options “…no further
have been tried activities will be
without success…“ realized“

Implications for Presupposition for

Apodosis: protasis:
“…a positive “repeated
solution will be unsuccessfull
performances of the
found“
same activity…“

Event structure (P) Event structure (A)

“Wenn alle Stricke reißen… …hänge ich mich auf“

“If all ese fails… …I‘ll giv e up“

BLENDED SPACE

Figure 3.2.  The ‘rope example from a blending analysis point of view. (cf. Müller 2008a: 169)

the second attribute being a contaminated idiomatic expression. Here we encoun-

ter a particularly interesting case, which is rather complex and stands in opposi-
tion to the first two examples mentioned for one major reason: it is semantically
perfectly consistent on the literal level. It is most likely that the activation of the
source domain information that comes with the first metaphor triggered the con-
tamination of the second one. As a result, we have a truly unconventional and
 Cornelia Müller

creative form of combining two idiomaticized metaphors – which make sense,

because they foreground uncommon aspects of the two matching source domains
(cf. Müller 2008a: 157–160, 172–175).
GERM: Die Isar soll sich wieder zu einem Wildwasser Fluß mausern also
offenbar mit neuen Federn schmücken.
FIG: The Isar is supposed to convert itself into a torrent hence apparently
adorn itself with a new outfit.
LIT: The Isar is supposed to be molting itself into a torrent hence apparently
adorn itself with new feathers.

This mixed metaphor makes perfect sense when focusing on the literal meaning
only: somebody is molting, which means that old plumage is being shed and a new
and nicer one is being put on. The experiential source domains (“birds changing
feathers”) of the two expressions are congruent. The writer has simply elaborated
the verbal metaphor ‘sich mausern’ (‘to take a turn for the better’, ‘to molt’) from a
source domain involving the changing of feathers by combining it with an idiom
using a similar source domain.
On the level of the figurative meaning, however, the combination of the two
metaphoric idioms does not appear to work. Indeed, we find here a semantic
inconsistency on the level of the figurative meaning, likely to result from the dif-
ferent figurative meanings of the combined metaphoric expressions. A closer look
at the second idiom “sich mit neuen Federn schmücken” (to adorn oneself with a
new outfit, to adorn oneself with new feathers) will make this clear:
GERM: sich mit fremden Federn schmücken
FIG: to deck oneself out in borrowed plumes
LIT: to adorn oneself with foreign feathers

The second metaphoric attribute is a variation of a fixed idiomatic expression,

which has been contaminated by a new lexical unit (a structurally similar case
to the one Cienki and Swan report on). The writer has replaced “fremde Federn”
(foreign feathers) with “neue Federn” (new feathers). Although this seems to be a
fairly minor semantic divergence on the literal level of meaning, the implications
of the change are quite far reaching for the figurative meaning. This is because
“sich mit fremden Federn schmücken” (to deck oneself out in borrowed plumes,
to adorn oneself with foreign feathers) comes with a clearly negative connotation
(instead of adorning oneself with ones own merits, one uses stolen ones). The
author has created a new metaphor, which carries a clearly positive connotation
“sich mit neuen Federn schmücken” (to adorn itself with a new outfit, to adorn
itself with new feathers). The mixed metaphor is part of a sentence that describes
the political plans for a river running through the major park in Munich. Politics
aims at a new outfit for the river, one that will give it a new and more interesting
 Chapter 3.  Why mixed metaphors make sense 

character as a wild-water river. For the author of this sentence, clearly the positive
connotation overrode the negative one of the original idiomatic expression “sich
mit fremden Federn schmücken” (to deck oneself out in borrowed plumes). In
fact, it appears that the activation of the source domain of the first metaphor (sich
mausern, to convert itself, to be molting) was so strong that not only did it trigger
a related source domain for the second metaphoric attribute but actually moti-
vated a contamination of an idiomatic expression and provoked a change in the
connotation from negative to positive. So, the combination of the two metaphoric
attributes does make sense, if we background the figurative meaning – including
the clash in connotations, and focus primarily on the literal meaning of the two
metaphoric attributes. This analysis gains further support when considering the
example from a blending point of view (Figure 3.3).
The combination is fairly straightforward, when it comes to the basic semantic
roles involved: an agent and self-reflexive actions upon agent with specific quali-
ties. What, on the basis of these semantic roles, is projected from both input spaces
is the basic idea that something is embellishing itself and that embellishing carries
a positive connotation. This is what both metaphoric attributes have in common.
We are now in a position to speculate that the writer’s flow of attention (Chafe
1994, 1996) online while working on the newspaper article actually foregrounded
the basic semantic structure of the literal meaning only and this is what presum-
ably has led him to produce this very sentence. This third example shows most
clearly that the traditional accounts have incorrectly reduced mixed metaphors
to a semantic inconsistency on the literal level. Instead, we are suggesting, that
what happens here is an interesting case of new emergent meaning, one in which
preceding salience structures (the positive connotation of the changing feathers,
a new outfit) override conventionalized salience structures (the negative connota-
tion of the borrowed plumes, a stolen outfit) and create new metaphoric meaning.
The example reveals that source domains of conventionalized metaphoric expres-
sions can be highly cognitively activated, meaning that they can be foregrounded
to such a degree that they even inspire elaborations of upcoming metaphors in the
discourse.
We have seen in the three examples discussed above that the mixing of meta-
phors occurs with different syntactic and semantic structures and usually operates
on few projections, which run counter to the common reading of the involved
metaphors. The projected elements may be literal (source domain) or figurative
(target domain) and can affect different facets of meaning construction, be they
semantic, syntactic or pragmatic.
As a conclusion of the analysis, we would like to suggest that mixed meta-
phors are much more varied than expected and that they offer important insights
into online processes of metaphoric meaning making. Although these linguistic
 Cornelia Müller

GENERIC SPACE

– agent
– self reflexive
action upon agent
– quality of action

INPUT SPACE  INPUT SPACE 

Isar sthing sthing Isar

is … itself is … itself is … itself is … itself
molting embellishing embellishing adorned with
borrowed
– positive – negative
plumes
connotations connotations
– change – change at some
(child – adult other person’s
grey – color) account

“Die Isar soll sich…”

‘The Isar is supposed’

“…sich…zu mausern”
‘to be…molting itself’
(’to convert itself’)

“mit
neuen
Federn schmücken”
‘adorn itself with new
feathers’

BLENDED SPACE

Figure 3.3.  The ‘molting river example’ from a blending analysis point of view.
­(Müller 2008:  173)

analyses are post-hoc explanations and can therefore never actually offer true
psychological evidence of those processes, they nevertheless might indicate that
metaphoric meaning is based on a process of meaning construction in which met-
aphoricity is established as part of a cognitive process and must be thought of as
intrinsically coupled with the flow of cognitive and interactive attention.
 Chapter 3.  Why mixed metaphors make sense 

3.4  Conclusion – A dynamic view on metaphors in language use

In conclusion, we propose that the mixing of metaphors results from dynamic

activations of uncommon aspects of metaphoric meaning. The cognitive rationale
for the production of mixed metaphors is a process of activating metaphoricity,
which follows the flow of attention in discourse and results in a selective fore-
grounding and backgrounding of aspects of metaphoric meaning. We believe that
these observations call for a dynamic view on metaphor, a view that regards con-
ventionalized metaphors ranging along a cline from sleeping to waking depending
on the cognitive activation of metaphoricity for a given speaker and interlocutor
at a given moment in the flow of a conversational interaction (Müller 2003, 2007,
2008a,b). If this holds true, the mixing of metaphors can be used as a source of evi-
dence for studying dynamic processes of metaphoric meaning making that are as
interesting for cognitive linguistics and probably also for psychologists as speech
errors for psycholinguists.
Furthermore, the cognitive-linguistic perspective on mixed metaphors offered
here refutes once more the notion of dead metaphors. Dead metaphors are not
dead at all, as long as they are transparent and their metaphoricity might be acti-
vated and creatively used by speakers and their interlocutors to make sense. Work
done on other forms of metaphor usage is underlining this claim. For instance
Wilson and Gibbs (2007) showed in an experimental study that real as well as
imagined body movement primed metaphor comprehension and Gibbs (1993b)
provides ample psychological evidence for the vitality of conventional metaphors
in idioms. Looking at the ways in which metaphors are used multimodally offers
other kinds of evidence for a dynamic online construction of metaphoric mean-
ing. Gestures may provide an experiential glimpse at a verbalized metaphoric
meaning (Cienki & Müller 2008a,b; Müller 2008a: 86–103, 178–209; Müller 2008b,
Müller  & Cienki 2009) in many ways similar to when headlines and pictures
are combined in newspapers. When complementing a headline with a picture
that instantiates some facet of metaphoric meaning, newspapers seek to attract
the attention of readers, by exploiting the rhetorical effect of activating conven-
tionalized metaphoric expressions (Müller 2008a: 103–113, 178–209). Research
on audio-visual media has documented, on the other hand, that the dynamics
of multimodal metaphoric meaning making involves the cognitive activation of
metaphoricity and the modulation of affects in the viewer’s process of reception.
Kappelhoff and Müller (2011) suggest that what we see in multimodal metaphors
in film as well as in gestures is that metaphoric meaning making is grounded in
embodied felt experiences and therefore is as much a cognitive as an affective phe-
nomenon ­(Kappelhoff & ­Müller 2011). We argue that in the process of perceiv-
ing metaphors in the c­ inema ­spectators construct metaphoric meaning as they
 Cornelia Müller

‘­follow’ the film and that this meaning making is grounded in the arrangement of
compositional units of what Kappelhoff terms ‘cinematic expressive movement’
(Kappelhoff 2004). A cinematic expressive movement is a unit of composition that
creates a movement Gestalt using all facets of cinematic techniques: from staging
to cutting, cadrage, and mis-en-scène, from camera to music, to acting and light.
It is a movement-image, not the image of a moving person, and it affects the spec-
tator emotionally and immediately as much as a gestural expression of anger, fear
or happiness affects an interlocutor. Cinematic metaphors emerge from and are
grounded in the sensory, felt experiences of perceiving those movement qualities.
Metaphors in film, as much as in gestures, always come with certain movement
qualities. So, a gesture depicting a particular time span (two flat hands facing,
represent two ends of a container) can be performed with a delightful and subtle
or with a sharp and rigid movement. The same holds for a cinematic metaphor,
which can be embedded in a lively, free and cheerful or in a rigid and sharp com-
position of cinematic movement.
Therefore, Kappelhoff and Müller (2011) suggest that, in metaphoric mean-
ing making, cognitive and affective experiences merge and that metaphors emerge
from a cognitive and affective process. Metaphors come to existence only in the
moment: In the moment of watching a film, of hearing and seeing somebody talk,
of reading a text. Following Ricoeur the reader of a text ‘decides’ whether a meta-
phor is created “like a spark that flashes when two semantic fields that were hith-
erto distant from each other collide. In this sense, a metaphor only exists in the
instant in which reading lends new life to the collision of the semantic fields and
generates the impertinent predication” (Ricoeur 1986: 6).
What Ricoeur holds for reading metaphors, we assume to be a fundamental
and general property of metaphors in use. Metaphor is a process of activating
metaphoricity. The nature of metaphoric meaning is not convincingly captured
in the traditional dichotomies of dead versus alive (Müller 2008b). Black is right
to assume that metaphors depend on “a pregnant use” (Black 1993: 25). More
specifically what characterizes metaphors in language use is that metaphoric-
ity oscillates between sleeping and waking, between a low degree of activation
(then it is backgrounded) and a high degree of activation (when foregrounded)
(Müller 2008a: 178–209). Cognitive-linguistic analyses of foregrounding strate-
gies have shown that this process can be empirically reconstructed by attending
to multimodal activation cues, such as a gesture, a picture or a verbal elabora-
tion of metaphoric meaning (Müller & Tag 2010). A future goal will be to relate
those degrees of metaphor activation further to different levels and forms of
consciousness ­(Müller 2011) and to the embodied and affective dimensions of
metaphor (Kappelhoff & Müller 2011; Koch et al. 2012; Kolter et al. 2012; Müller
in press).
 Chapter 3.  Why mixed metaphors make sense 

Studying those different forms of metaphor usage has led us to formulate a

dynamic view on metaphor (Müller 2003, 2007, 2008a, b). This dynamic view is
sympathetic with Cameron’s discourse dynamics framework, which considers
metaphor as emergent from discourse (Cameron 2009, 2010, 2011). However,
while Cameron’s discourse led metaphor analysis targets linguistic expressions
that are potentially metaphoric for the participants in a conversation, the dynamic
view advocated here proposes that conversational partners display for each other
whether metaphoricity is in the foreground of their interactive attention or not.
Mixed metaphors are a highly specific case of foregrounding aspects of meta-
phoric meaning that are normally backgrounded and they show that conventional
metaphors may be creatively used – awakened or revitalized. Here is where our
dynamic view relates with Steen’s concept of deliberate metaphor or a potential
paradox of metaphor (Steen 2008, 2011, 2013). Steen argues, that “Most metaphor
in language may be processed in non-metaphorical ways, raising a potential para-
dox (Steen 2008)” (Steen 2013: 193). Suggesting that only deliberately used meta-
phors are processed metaphorically, his model can be characterized as dynamic
too. In fact, his account of the metaphorical potential of a metaphor that may lie
dormant in discourse addresses similar phenomena as the activation of metapho-
ricity view proposed here.
To conclude, Linguistic as well as Conceptual Metaphor Theory cannot con-
vincingly account for the processes involved in metaphoric meaning making
because their focus is on metaphors as products not as processes (Gibbs 1993a).
As an alternative, we have suggested a dynamic theory of metaphor that starts out
from the process of use. The dynamic view, offers a usage-based account of meta-
phor, one that regards metaphors as based on the cognitive activity that does the
mappings or the projections between the experiential domains involved. In terms
of Bowdle and Gentner’s psycholinguistic career-of-metaphor theory, the dynamic
activation of metaphoricity in mixed metaphors, documented in this chapter,
could be considered as a shift back from processing as categorization to process-
ing as comparison, which they consider characteristic for novel metaphors. Obvi-
ously this type of revitalization of metaphoric meaning can only happen when the
source domain or the base of the metaphoric term is still transparent for a given
speaker-listener, i.e., this would be a case of type (1) of dead metaphors proposed
by Bowdle and Gentner (2005).
The mappings and projections in such activations of metaphoricity are multi-
directional, much as Black’s interactive theory on metaphor has suggested (Black
1962, 1993; Müller 2008a: 114–133) and as Blending Theory (Grady, Coulson,
Oakley 1999) holds. In this view the making of metaphors involves the creation
of particular salience structures (for a different account of salience in metaphors,
see Giora 2003) and a cognitive process of “seeing as” (Müller 2008a: 22–39).
 Cornelia Müller

Establishing metaphoricity between domains of experience is furthermore con-

ceived of as a general cognitive process (in line with the cognitive commitment)
that may apply to all kinds of metaphoric structures in all kinds of modalities. In
other words, it is modality independent and therefore is in accordance with the
generalization commitment of cognitive linguistics. It is also an embodied pro-
cess, both in regard to the basic cognitive-linguistic assumption that cognition
is grounded in embodied felt experiences (Gibbs 2005; Johnson 1987, 2007) and
in that it evokes embodied experiences, since, after all, what we do, when using
metaphors, is experiencing one domain of experience through another one. For
such a dynamic view of metaphor, mixed metaphors are an expected product of
creative language use. They show dynamic activations of non-common aspects
of meaning and unconventional ways of combining them. From this perspec-
tive, the mixing of metaphors makes perfect sense–it is a natural way of using
language.

References

Abraham, W. (1998). Linguistik der uneigentlichen Rede. Linguistische Analysen an den Rändern
der Sprache. Tübingen: Stauffenburg Verlag.
Bertau, M-C. (1996). Sprachspiel Metapher. Denkweisen und kommunikative Funktion einer
rhetorischen Figur. Opladen: Westdeutscher Verlag.
Black, M. (1962). Models and metaphors. Studies in language and philosophy. Ithaca: Cornell
University Press.
Black, M. (1993). More about metaphor. In A. Ortony (Ed.), Metaphor and thought (2nd ed.,
pp. 19–41). Cambridge: Cambridge University Press.  doi: 10.1017/CBO9781139173865.004
Bowdle, B. F., & Gentner, D. (2005). The career of metaphor. Psychological Review, 112(1),
193–216.  doi: 10.1037/0033-295X.112.1.193
Braak, I. (1969=1974). Poetik in Stichworten. Kiel: Hirt.
Brooke-Rose, C. (1958). A grammar of metaphor. London: Secker & Warburg.
Bußmann, H. (1990). Lexikon der Sprachwissenschaft (2nd ed., p. 371). Kröner.
Cameron, L. (2009). The discourse dynamics approach to metaphor and metaphor-led discourse
analysis.Metaphor and Symbol, 24(2), 63–89.  doi: 10.1080/10926480902830821
Cameron, L. (2010): The discourse dynamics framework for metaphor. In L. Cameron &
R. Maslen (Eds.), Metaphor analysis: Research practice in applied linguistics, social science
and the humanities (pp. 77–94). London: Equinox.
Cameron, L. (2011). Metaphor and reconciliation: The discourse dynamics of empathy in post-
conflict conversations. London: Routledge.
Cienki, A., & Müller, C. (2008a). Metaphor, gesture and thought. In R. Gibbs (Ed.), Cambridge
handbook of metaphor and thought (pp. 483–501). Cambridge: Cambridge University
Press.  doi: 10.1017/CBO9780511816802.029
Cienki, A., & Müller, C. (Eds.). (2008b). Metaphor and gesture. Amsterdam: John Benjamins.
doi: 10.1075/gs.3
Cienki, A., & Swan, D. (Ms.). Metaphors, coherence, and blending.
 Chapter 3.  Why mixed metaphors make sense 

Chafe, W. (1994). Discourse, consciousness, and time. The flow and displacement of conscious
experience in speaking and writing. Chicago/London: University of Chicago Press.
Chafe, W. (1996). How consciousness shapes language. Pragmatics & Cognition (Special issue
‘On language and consciousness’), 4(1), 55–64.  doi: 10.1075/pc.4.1.06all
Coenen, H-G. (2002). Analogie und Metapher. Grundlegung einer Theorie der bildlichen Rede.
Berlin/New York: de Gruyter.  doi: 10.1515/9783110894639
Cohen, T. (1975). Figurative speech and figurative acts. Journal of Philosophy, 72, 669–682.
doi: 10.2307/2024631
Cohen, L. J. (1993). The semantics of metaphor. In A. Ortony (Ed.), Metaphor and thought
(pp. 58–70). Cambridge: Cambridge University Press.  doi: 10.1017/CBO9781139173865.006
Davidson, D. (1978). What metaphors mean. Critical Inquiry, 5, 31–47.  doi: 10.1086/447971
van Dijk, T. (1975). Formal semantics of metaphorical discourse. Poetics, 4, 173–198.
doi: 10.1016/0304-422X(75)90081-9
Eco, U. (1984). Semiotics and the philosophy of language. Bloomington: Indiana University Press.
Eisenberg, P. (1986). Grundriß der deutschen Grammatik. Stuttgart: Metzler.
Fauconnier, G., & Turner, M. (2002). The way we think. Conceptual blending and the mind’s hid-
den complexities. New York: Basic Books.
Friedrich, H. (1968). Die Metapher (Bochumer Disk). Poetica, 2, 100–130.
Glucksberg, S., & Keysar, B. (1990). Understanding metaphorical comparisons: Beyond similar-
ity. Psychological Review, 97, 3–18.  doi: 10.1037/0033-295X.97.1.3
Glucksberg, S., & Keysar, B. (1993). How metaphors work. In A. Ortony (Ed.), Metaphor and
thought (2nd ed., pp. 401–424). Cambridge: Cambridge University Press.
doi: 10.1017/CBO9781139173865.020
Gibbs, R. (1993a). Process and products in making sense of tropes. In A. Ortony (Ed.), Meta-
phor and thought (2nd ed., pp. 252–276). Cambridge: Cambridge University Press.
doi: 10.1017/CBO9781139173865.014
Gibbs, R. (1993b). Why idioms are not dead metaphors. In C. Cacciari & P. Tabossi (Eds.), Idi-
oms, processing structure and interpretation (pp. 57–77). Hillsdale, N.J.: Lawrence Erlbaum.
Gibbs, R. (1994). The poetics of mind. Figurative thought, language, and understanding.
­Cambridge: Cambridge University Press.
Gibbs, R. (1999). Intentions in the experience of meaning. Cambridge: Cambridge University
Press.
Gibbs, R. (2005). Embodiment and cognitive science. Cambridge: Cambridge University Press.
doi: 10.1017/CBO9780511805844
Giora, R. (2003). On our mind: Salience, context, and figurative language. New York: Oxford
University Press.  doi: 10.1093/acprof:oso/9780195136166.001.0001
Grady, J., Oakley, T., & Coulson, S. (1999). Blending and metaphor. In R. Gibbs, Jr., & G. Steen
(Eds.), Metaphor in cognitive linguistics (pp. 101–124). Amsterdam: John Benjamins.
doi: 10.1075/cilt.175.07gra
Grice, H. P. (1975). Logic and Conversation. In P. Cole & J. L. Morgan (Eds.), Syntax and seman-
tics 3: Speech acts (pp. 41–58). New York: Academic Press.
Johnson, M. (1987). The body in the mind. The bodily basis of meaning, imagination, and reason.
Chicago/London: University of Chicago Press.
Johnson, M. (2007). The meaning of the body. Aesthetics of human understanding. Chicago:
­Chicago University Press. doi: 10.7208/chicago/9780226026992.001.0001
Kappelhoff, H. (2004). Matrix der Gefühle. Das Kino, das Melodrama und das Theater der Emp-
findsamkeit. Berlin: Vorwerk.
 Cornelia Müller

Kappelhoff, H., & Müller, C. (2011). Embodied meaning construction. Multimodal metaphor
and expressive movement in speech, gesture, and in feature film. Metaphor and the Social
World, 1, 121–153.  doi: 10.1075/msw.1.2.02kap
Katz, A. (1998). Figurative language and figurative thought: A review. In A. Katz, C. Cacciari,
R. Gibbs, Jr., & M. Turner (Eds.), Figurative language and thought (pp. 3–43). New York/
Oxford: Oxford University Press.
Kayser, W. (1976). Das sprachliche Kunstwerk. Eine Einführung in die Literaturwissenschaft.
(17th ed.). Bern, Munich: Francke.
Keysar, B., & Bly, B. (1995). Intuitions of the transparency of idioms: Can you keep a secret by
spilling the beans? Journal of Memory and Language, 34, 89–109.
doi: 10.1006/jmla.1995.1005
Koch, S., Fuchs, T, Summa, M., & Müller, C. (2012). Body memory, metaphor, and movement.
Amsterdam: John Benjamins. doi: 10.1075/aicr.84
Kolter, AsA., Ladewig, S., Summa, M., Koch, S., Müller, C., & Fuchs, T. (2012). Body memory
and the emergence of metaphor in movement and speech. An interdisciplinary case study.
In S. Koch, T. Fuchs, M. Summa, & C. Müller (Eds.), Body memory, metaphor and move-
ment. (pp. 201–226). Amsterdam: John Benjamins.  doi: 10.1075/aicr.84.16kol
Köller, W. (1975). Semiotik und Metapher. Untersuchungen zur grammatischen Struktur und kom-
munikativen Funktion von Metaphern. Stuttgart: J.B. Metzlersche Verlagsbuchhandlung.
Köller, W. (1986). Dimensionen des Metaphernproblems. Zeitschrift für Semiotik, 4, 379–410.
Kövecses, Z. (2002). Metaphor. A practical introduction. Oxford: Oxford University Press.
Kubczak, H. (1978). Die Metapher. Beiträge zur Interpretation und semantischen Struktur der
Metapher auf der Basis einer referentialen Bedeutungsdefinition. Heidelberg: Carl Winter.
Lakoff, G. (1993). The contemporary theory of metaphor. In A. Ortony (Ed.), Metaphor and
thought (pp. 202–251). Cambridge: Cambridge University Press.
doi: 10.1017/CBO9781139173865.013
Lakoff, G., & Johnson, M. (1980). Metaphors we live by. Chicago/London: Chicago University
Press.
Lakoff, G., & Kövecses, Z. (1987). The cognitive model of anger inherent in American English.
In D. Holland & N. Quinn (Eds.), Cultural models in language and thought (pp. 195–221).
Cambridge: Cambridge University Press.  doi: 10.1017/CBO9780511607660.009
Lakoff, G., & Turner, M. (1989). More than cool reason. A field guide to poetic metaphors. C
­ hicago:
Chicago University Press.  doi: 10.7208/chicago/9780226470986.001.0001
Lakoff, G., & Johnson, M. (1999). Philosophy in the flesh. New York: Basic Books.
Lausberg, H. (1960). Handbuch der literarischen Rhetorik. Eine Grundlegung der Literaturwis-
senschaft. München: Max Hueber.
McArthur, T. (Ed.). (1992). The Oxford companion to the English language. Oxford: Oxford Uni-
versity Press.
McGlone, M. (1996). Conceptual metaphors and figurative language interpretation: Food for
thought? Journal of Memory and Language, 35, 544–565.  doi: 10.1006/jmla.1996.0029
Meier, H. (1963). Die Metapher. Versuch einer zusammenfassenden Betrachtung ihrer linguist-
ischen Merkmale. Winterthur: Verlag P.G. Keller.
Miller, G. (1993). Images and models, similes and metaphors. In A. Ortony (Ed.), Metaphor and
thought (2nd ed., pp. 357–400). Cambridge: Cambridge University Press.
doi: 10.1017/CBO9781139173865.019
Müller, C.(2003). Gestik als Lebenszeichen ‘toter Metaphern’. Tote, schlafende und wache
­Metaphern. Zeitschrift für Semiotik, 1–2, 61–72.
 Chapter 3.  Why mixed metaphors make sense 

Müller, C. (2007). A dynamic view on gesture, language and thought. In S. Duncan, J. Cassell, &
E. Levy (Eds.), Gesture and the dynamic dimension of language. Essays in honor of David
McNeill (pp. 109–116). Amsterdam: John Benjamins.  doi: 10.1075/gs.1.12mul
Müller, C. (2008a). Metaphors dead and alive, sleeping and waking. A dynamic view. Chicago:
University of Chicago Press.  doi: 10.7208/chicago/9780226548265.001.0001
Müller, C. (2008b). What gestures reveal about the nature of metaphor. In A. Cienki & C. Müller
(Eds.), Metaphor and gesture (pp. 219–245). Amsterdam: John Benjamins.  doi: 10.1075/gs.3
Müller, C. (2011). Reaction paper. Are ‘Deliberate’ metaphors really deliberate. A Question of
human consciousness and action. Metaphor in the Social World, 1, 61–66.
doi: 10.1075/msw.1.1.06mul
Müller, C., & Cienki, A. (2009). Words, gesture and beyond: Forms of multimodal metaphor in
the use of spoken language. In C. Forceville & E. Urios-Aparisi (Eds.), Multimodal meta-
phor (pp. 329–366). Berlin/New York: Mouton de Gruyter.
Müller, C., & Ladewig, S. (2014). Metaphors for sensorimotor experiences: Gestures as embodied
and dynamic conceptualizations of balance in dance lessons. In M. Borkent, B. Danygier, &
J. Hinnel (Eds.), Language and the creative mind (pp. 295–324). Chicago: University of
Chicago Press.
Müller, C., & Tag, S. (2010). Attention, foregrounding and embodied activation. Multimodal
metaphors in spoken discourse. Cognitive Semiotics, 6, 85–119.  doi: 10.3726/81610_85
Murphy, G. L. (1996). On metaphoric representation. Cognition, 60, 173–204.
doi: 10.1016/0010-0277(96)00711-1
Murphy, G. L. (1997). Reasons to doubt the present evidence for metaphoric representation.
Cognition, 62, 99–108.  doi: 10.1016/S0010-0277(96)00725-1
Ortony, An. (1988). Are emotion metaphors conceptual or lexical? Cognition and Emotion, 2,
95–103.  doi: 10.1080/02699938808408066
Paivio, A., & Walsh, M. (1993). Psychological processes in metaphor comprehension and mem-
ory. In A. Ortony (Ed.), Metaphor and thought (2nd ed., pp. 307–328). Cambridge: Cam-
bridge University Press.  doi: 10.1017/CBO9781139173865.016
Pape, W. (1914). Handwörterbuch der griechischen Sprache. Braunschweig: Vieweg und Sohn.
Quinn, N. (1999a). Research on shared task solutions. In C. Strauss & N. Quinn (Eds.), A cogni-
tive theory of cultural meaning (pp. 137–188). Cambridge: Cambridge University Press.
Quinn, N. (1999b). Research on the psychodynamics of shared understandings (Ch 7). In C.
Strauss & N. Quinn (Eds.), A cognitive theory of cultural meaning (pp. 189–209) (first pub-
lished 1997). Cambridge: Cambridge University Press.
Reddy, M. (1993). The conduit metaphor: A case of a frame conflict in our language about
language. In A. Ortony (Ed.), Metaphor and thought (2nd ed., pp. 164–201). Cambridge:
Cambridge University Press.  doi: 10.1017/CBO9781139173865.012
Ricoeur, P. (1986). Die lebendige Metapher. München: Wilhelm Fink.
Richards, I. A. (1936). The philosophy of rhetoric. New York: Oxford University Press.
Sadock, J. (1993). Figurative speech and linguistics. In A. Ortony (Ed.), Metaphor and thought
(2nd ed., pp. 42–57). Cambridge: Cambridge University Press.
doi: 10.1017/CBO9781139173865.005
Schneider, W. (1999). Deutsch für Kenner. Die neue Stilkunde. Hamburg: Piper. (First published
1996)
Schön, D. (1993). Generative metaphor: A perspective on problem-setting in social policy. In
A. Ortony (Ed.), Metaphor and thought (2nd ed., pp. 137–173). Cambridge: Cambridge
University Press.  doi: 10.1017/CBO9781139173865.011
 Cornelia Müller

Searle, J. (1993). Metaphor. In A. Ortony (Ed.), Metaphor and thought (2nd ed., pp. 83–111).
Cambridge: Cambridge University Press.  doi: 10.1017/CBO9781139173865.008
Shore, B. (1996). Culture in mind: Meaning construction and cultural cognition. New York:
Oxford University Press.
Sperber, D., & Wilson, D. (1986). Relevance. Communication and cognition. Oxford, UK/­
Cambridge, Mass.: Blackwell.
Steen, G. (2008). The paradox of metaphor. Why we need a three-dimensional model of meta-
phor. Metaphor and Symbol, 23(4), 213–241.  doi: 10.1080/10926480802426753
Steen, G. (2011). What does ‘really deliberate’ really mean? More thoughts on metaphor and
consciousness. Metaphor and the Social World, 1(1), 53–56.  doi: 10.1075/msw.1.1.04ste
Steen, G. (2013). Deliberate metaphor affords conscious metaphorical cognition. Cognitive
Semiotics, V(1–2), 179–197.
Stock, O., Slack, J., & Ortony, A. (1993). Building castles in the air: Some computational and
theoretical issues in idiom comprehension. In C. Cacciari & P. Tabossi (Eds.), Idioms: Pro-
cessing, structure, and interpretation. Hillsdale, NJ: Lawrence Erlbaum.
Sweetser, E. (1996). Mental spaces and the grammar of conditional constructions. In G. Faucon-
nier & E. Sweetser (Eds.), Spaces, worlds, and grammar (pp. 318–333). Chicago/London:
University of Chicago Press.
Weinrich, H. (1976). Sprache in Texten. Stuttgart: Ernst Klett Verlag.
Wilson, N., & Gibbs, R. (2007). Real and imagined body movement primes metaphor compre-
hension. Cognitive Science, 31, 721–731.  doi: 10.1080/15326900701399962
Winner, E., & Gardner, H. (1993). Metaphor and irony: Two levels of understanding. In
A.  Ortony (Ed.), Metaphor and thought (2nd ed., pp. 425–443). Cambridge: Cambridge
University Pres.  doi: 10.1017/CBO9781139173865.021
 chapter 4

Tackling mixed metaphors in discourse

New corpus and psychological evidence

Julia E. Lonergan & Raymond W. Gibbs, Jr.

University of California

We offer an analysis of a small corpus of mixed metaphor excerpts taken from

“The New Yorker” magazine’s column titled “Block That Metaphor!” (BTM). Our
aim was to explore the main hypothesis that people interpret mixed metaphors
as being meaningful and coherent because of their abilities to engage in elaborate
reasoning about the source domains explicitly mentioned in these texts. A first
study investigated the different individual metaphorical expressions in this
series of mixed metaphorical narratives. We found that most of these have been
employed, and conveyed metaphorical meanings, in other kinds of discourse, and
that the metaphors within the BTM vignettes were, indeed, mixed, and done so
in a variety of ways. A second study asked university students to write out their
interpretations of the different phrases in these excerpts, one-by-one as they read
through each vignette. Analysis of participants’ protocols showed tremendous
consistency in how people understood the individual metaphors and that the
mixed metaphors in these narratives made good sense. The specificity of people’s
understandings of verbal metaphors in narratives is aided by their rich social
and cultural knowledge of the source domains referred to explicitly in language.
Even if the underlying conceptual metaphors for these verbal metaphors appear
to clash, people make use of their elaborate source domain information to create
specific concepts and images that often make mixed metaphors perfectly coherent
and, at times, delightful to hear and read.

4.1  Introduction

Consider the following metaphoric narrative, first appearing in The Times

­(October 12, 1998), and later reprinted in The New Yorker magazine as an item in
its long-running filler titled “Block that Metaphor!”:

doi 10.1075/milcc.6.04lon
© 2016 John Benjamins Publishing Company
 Julia E. Lonergan & Raymond W. Gibbs, Jr.

“As I look at it with broad brush, there are a lot of things going South at the
same time,” said Morris Goldstein, a former International Monetary Fund official
and senior fellow at the Institute for International Economics in Washington.
“Where’s the good news coming from? There’s no silver bullet out there!”

Do you understand what the speaker means by what he says here? Although high-
school English teachers often advise students against using mixed metaphors, the
mixing together of different figurative phrases litter speech and writing. Readers
encounter two problems when faced with mixed metaphor narratives. First, can
they understand the metaphoric meanings of the individual expressions, such as
“As I look at it with broad brush,” “a lot of things going South at the same time,”
and “There’s no silver bullet out there!”? Second, readers must find ways of creat-
ing an overall meaning of the different metaphoric phrases together. Of course,
these two challenges are relevant to understanding any instance of connected dis-
course, metaphorical or not. Still, the mixing of metaphors, with their very dif-
ferent source domains (e.g., a “broad brush,” “things going South,” and “no silver
bullet”) may pose a formidable task that readers and listeners encounter all too
often in everyday discourse.
Our studies examined three hypotheses on people’s use and understanding of
mixed metaphors in discourse, and not just within single sentences, particularly
those collected in the “Block That Metaphor” excerpts published in the New Yorker
over many decades. The first possibility, dubbed the “mixed metaphor is nonsense”
hypothesis, suggests that the mixing of metaphors is sloppy, and the product of a
speaker’s disorganized thoughts about some topic. Under this view, most mixed
metaphors are jumbled-up clusters of “dead metaphors” and understood as having
little metaphorical function. The mixing of multiple source and target domains,
and the shifting between multiple metaphorical concepts, creates a disruption that
is confusing and difficult to meaningfully interpret. Mixed metaphors are some-
times perceived as funny because we mock speakers for thinking and speaking in
this jumbled manner. But under this hypothesis, people should find mixed meta-
phorical narratives difficult to interpret.
Hypothesis 2, adopted from “conceptual metaphor theory,” states that mixed
metaphors reflect the relatively coherent combining of two active conceptual met-
aphors (Kovecses, 2010). Consider the following example of a mixed metaphor,
also taken from the “Block That Metaphor!” column: “I don’t want to say they lost
sight of the big picture, but they have marched to a different drummer,” Victor
Fortuno, the general counsel of Legal Services Corporation, said of the individual
lawyer’s challenges. “Whether it will upset the apple cart, I don’t know.”
This excerpt combines several linguistic expressions that are motivated by
different conceptual metaphors, according to standard conceptual metaphor
theory. “I don’t want to say they lost sight of the big picture” is motivated by
 Chapter 4.  Tackling mixed metaphors in discourse 

­ nderstanding is seeing, because the perception of a physical object is under-

u
stood as the ability to comprehend something. This entails that light and the lack
thereof are physical properties used to aid or impede vision and comprehension.
On the other hand, “They have marched to the beat of a different drummer” is
presumably motivated by life is a journey, where the physical-motion-event of
traveling along a path is understood as being conducted by someone who “is off
timing” in his footsteps, entailing that he may be overtaken by the sounds of a dif-
ferent drum beat. Drumming is an audio accompaniment to a person’s journey,
such that the physical inability to keep time is projected onto better understanding
the abstract ability to proceed in sync with others along a path toward some com-
mon goal. Finally, the expression “Whether it upsets the apple cart, I don’t know”
is linked to the conceptual metaphor the mind is a container and ideas are
individual food items. The apple cart is the mind that contains structured mat-
ter and if disturbed can have a calamitous and destabilizing effect on the mind’s
smooth operation.
Although conceptual metaphor theory provides a coherent framework for
understanding the conceptual, embodied motivation for both conventional
and novel metaphoric language (Gibbs, 1994; Gibbs & Colston, 2012; Lakoff &
­Johnson, 1999), it is not clear whether the theory explains why people sometimes
mix their metaphors (but see Kövecses, this volume). Of course, there is nothing
within conceptual metaphor theory that insists on the consistency of conceptual
metaphors in discourse, even when speaking of a particular topic. Indeed, con-
ceptual metaphor theory has provided abundant data showing how individual tar-
get domains can be conceived through numerous, sometimes conflicting, source
domains (Kövecses, 2010). For example, people may conceive of love in terms of
journeys, manufactured objects, magic, plants, and so on, each of which expresses
different entailments that do not add up to one monolithic, internally consistent
concept for love. If anything, conceptual metaphor theory suggests that people can
flexibly mix metaphors, perhaps at some cognitive cost, when speaking of some
topic, precisely because of the different, flexible, ways that we metaphorically con-
ceive of some abstract topics. Still, conceptual metaphor theory does not explain
exactly why people mix the specific metaphors they sometimes do, or why some
combinations of metaphors may be more consistent with one another than other
assortments of verbal metaphors.
Hypothesis 3, called the “source domain reasoning” idea, suggests that peo-
ple’s reasoning about source and target domain mappings extends beyond the
single conventional parallel mappings typically seen in conceptual metaphors the-
ory. There are other non-parallel systems operating concurrently with conceptual
metaphors, such as affective and socio-cultural factors, which critically shape how
people create and understand particular tropes. Under this view, many metaphors
 Julia E. Lonergan & Raymond W. Gibbs, Jr.

in discourse obtain their meanings less from systematic parallel mappings from
source to target domain, and more from compensatory within-source domain rea-
soning (Lee & Barnden, 2001). For example, consider the metaphoric expression
“We are driving in the fast lane on the freeway of love” (Lakoff, 1993). This novel
metaphor is partly motivated by the underlying conceptual metaphor love rela-
tionships are journeys, which gives rise to various source-to-target mappings
including the idea that the lovers are traveling in a vehicle together. But we typically
interpret this novel metaphor as suggesting that the lovers are in the vehicle alone
together and experiencing great emotional excitement during this part of the jour-
ney, possibly because of the speed in which they are moving on the fast lane of a
freeway. These ancillary assumptions about the nature of the love journey alluded
to by this novel metaphor are not specific to the particular conceptual metaphor
at work here, namely love relationships are journeys. For this reason, ancil-
lary assumptions, called “view-neutral mapping adjuncts” (VNMAs) (Barnden &
Lee, 2001), create lower-order submetaphors related to people’s reasoning about
the source domain. More specifically, VNMA are non-parallel adjuncts that relate
tacit social and cultural knowledge to conceptual metaphors.
Hypothesis 3 claims that people’s production and understanding of mixed
metaphors are motivated by their ability to engage in extensive source domain
reasoning. Consider the last expression in the opening mixed metaphor vignette-
“There’s no silver bullet out there!” Under hypothesis 3, people interpret this
expression not only in terms of the basic conceptual metaphor ideas are objects,
but via the more detailed submetaphor rare ideas are rare objects. The sub-
metaphor is created given people’s cultural knowledge about “silver bullets” as
being special, rare objects that are needed to deal with very special problems.
Thus, just as rarely found silver bullets can kill special, dangerous people, so too
can finding a silver bullet stand for the discovery of a special solution to the eco-
nomic problem alluded to by the speaker of the above excerpt. This additional
reasoning, including people’s emotional responses to the source domain, creates
a complex interpretation of the verbal metaphor, one that is far richer than that
given by the conceptual metaphor alone.
Most importantly, it is via this more elaborate source domain reasoning that
people can create links between what appears to be incompatible metaphoric
expressions. Talking about silver bullets makes perfect sense, then, in the context
of having a broad discussion of seeking solutions to an economy that has “gone
South.”
We now report the findings of one corpus analysis and one psychological
study examining people’s interpretations of some “Block That Metaphor” (BTM)
vignettes. These studies provide empirical evidence in support of hypothesis 3
(source domain reasoning) and suggest the inadequacy of hypotheses 1 and 2. Our
 Chapter 4.  Tackling mixed metaphors in discourse 

findings, more generally, point to an important revision of conceptual metaphor

theory for handling creative metaphor use, including that involving the mixing
of metaphors in discourse. Examining the validity of our preferred hypothesis 3
requires that we ask ordinary readers to spell out their interpretations of the dif-
ferent segments of the mixed metaphor narratives. We specifically hoped to find
evidence within people’s interpretations that they engaged in extensive source
domain reasoning, perhaps through access of different VNMAs.

4.2  Study 1

Our interest in Study 1 was threefold. First, we assessed whether the different figu-
rative phrases within the BTM excerpts really conveyed metaphorical meanings in
context. Second, we then explored whether other speakers typically used the same,
or very similar, phrases in other contexts to express metaphorical messages. This
second step was accomplished by searching for instances of the different phrases in
a large corpus of language, the “British National Corpus” (or BNC), a 100 million
word corpus of spoken British English from the late part of the 20th century, and
determining the extent to which each expression was employed with metaphorical
intent. Finally, we examined combinations of the different metaphorical phrases
to determine whether the BTM excerpts really consisted of “mixed metaphors.”

4.2.1  Method
We randomly selected nine BTM excerpts printed in “The New Yorker” from
1990 to 2007. Each vignette was then broken down into their component phrases,
resulting in a corpus of 62 different expressions (e.g., “Look at it with broad brush,”
“Things going South,” and “There is no silver bullet out there”). Appendix A pres-
ents a list of the 9 BTM excerpts.
We then analyzed each example for the presence of metaphorically used
words according to the criteria outlined by the “Metaphor Identification Proce-
dure” (MIP) (Pragglejaz Group, 2007). MIP essentially determined the contextual
meaning of each word in a text, established its “basic” meaning from dictionaries
(and other sources), and marked as “metaphor” any instances in which the contex-
tual and basic meanings were related to each other through a comparison process.
We were in initial agreement for 95% of our original, independent markings and
resolved our differences on the remaining items through follow-up discussion.
MIP only identifies metaphorically used words, so we then determined
whether the individual phrases overall expressed metaphoric meanings. Once
again, we made these judgments independently, and found ourselves in agreement
 Julia E. Lonergan & Raymond W. Gibbs, Jr.

98% of the time, and resolved our differences over just a few items during follow-
up discussions.
Finally, we searched for each phrase within the BNC, a 100 million word corpus
of spoken British English from the late part of the 20th century, and determined,
again using MIP, whether these were used in metaphorical ways in the BNC.

4.2.2  Results
Our first analysis revealed that 20% of all the lexical units within the BTM vignettes
expressed metaphorical meanings in their contexts. This proportion underes-
timates the salience of metaphor within the narratives given that many words
referred to proper names and the businesses these people worked for (e.g., “said,
Morris Goldstein, a former International Monetary Fund official and senior fellow
at the Institute for International Economics in Washington”). If proper names and
other identifying information are removed, the overall proportion of metaphori-
cally used words in the BTM excerpts rises to 25%. This proportion represents a
relatively high degree of metaphorically used words compared to other genres (see
Steen et al., 2010).
Not every phrase or utterance which contains a metaphorically used word
necessarily expresses a global metaphorical meaning. We therefore judged whether
each phrase (e.g., “Looking at it with broad brush”) could be seen as expressing
metaphorical meaning, based on the MIP analysis, and found that, indeed, 100% of
the 62 phrases in our collection conveyed metaphorical messages in their contexts.
We then searched the BNC to see if the BTM phrases were commonly used,
and done so with metaphorical meanings, within a broad corpus of English lan-
guage use. Our examination included both searching for the exact wording of the
62 phrases in the BTM excerpts, but also slight variations of these (“going South”
instead of “things going South”). This analysis revealed that 78% of the phrases
from our corpus were also evident within the BNC. By looking at the local contexts
in which these phrases were found, again using MIP as a guideline for determining
metaphoricity in meaning, we found that 100% of the phrases had been used pre-
viously within English to express metaphorical meanings. These latter two results
suggest that the BTM mixed metaphors contained conventional phrases that were,
indeed, conventional in the sense of being frequently used to talk about various
abstract topics in metaphorical ways. For this reason, the mixed metaphors in the
BTM excerpts were not particularly novel, with the exception of slight differences
in the wording of the phrases in the BTM passages and those in the BNC.
At this point, we could confidently claim that the BTM excerpts contained
metaphorical phrases which were also widely used in other contexts to express
metaphorical meanings. To what extent, though, did the 9 different BTM excerpts
contain mixed metaphors? To answer this question, we examined the explicit
 Chapter 4.  Tackling mixed metaphors in discourse 

source domains in the different metaphorical expressions (e.g., “broad brush”

when talking of examining some abstract idea or problem; “going South” when
talking about an abstract economic problem), and determined that each of the
9 excerpts employed several different source domains (e.g., “going South” refers to
traveling in a Southward direction). This result alone is not particularly meaning-
ful given that several BTM excerpts referred to several different target domains
(e.g., examining an abstract situation, judging that a problem existed and was get-
ting worse, realizing that there was no solution to the problem). It should not be
surprising that different target domains were discussed in terms of different source
domains.
But the mixing of metaphors mostly resides in the various combinations
between target and source domains within the 9 BTM excerpts. Consider one part
of a BTM excerpt in which a critic comments negatively about a column posted
on an Internet blog.
“This sort of journalistic tripe is poison, and yet, at the same time, grist for the
mill among the twisted jackals who make up Congress and who, it seems, have
no qualms about using the Internet as a personal whipping post whenever it suits
their fancy.”

This long statement refers to (a) the original Internet column, (b) members of
Congress who may read columns like this one, and (c) how these individuals may
use such columns on the Internet for political purposes. Thus, there are three
basic target domains explicitly noted, and each of these is described using differ-
ent source domains. For example, the column is poison tripe for some people, and
grist for the mill for others (2 source domains for this single target), the mem-
bers of Congress who read the column are twisted jackals (2 source domains for
this single target), who use the material on the Internet as a whipping post for
political ends. Overall, this one part of the BTM excerpt employs several different
source domains for metaphorically talking about several different topics or target
domains. The vividness of the BTM excerpt, however, stems from the diversity of
source domain experiences that are mentioned. These different source domains
have little or no pre-existing semantic relations when used to convey either meta-
phoric or non-metaphoric messages. Many of the BTM excerpts reflect this type of
diverse mixture of different target and source domains.
A different mixing of metaphors used a single source domain to talk of several
different target domains. Consider the following BTM vignette:
“At last, the lip service that education has been lathered in for the past couple
decades seems to have found the razor’s edge among people who are willing to do
something about improving it here, at the elementary level all the way up through
public universities. But as budgets continue to be shaved, will the arts once again
be trimmed from the programming?”
 Julia E. Lonergan & Raymond W. Gibbs, Jr.

Several target topics are referred to in this example, which include (a) talk about
education, (b) people who are willing to do something to improve education, and
(c) the education budget, particularly relating to the arts. This interrelated set of
targets are all metaphorically described in terms of the source domain of men’s
face shaving, as in talk of “lip service,” “lathered,” “razor’s edge,” “shaved,” and
“trimmed.” One may generally expect people to have less difficulty interpreting
this BTM excerpt precisely because it employs a single extended metaphor refer-
ring to men’s shaving, compared to mixed metaphors that shift between different
source domains. The humor in this case arises from the way the single source
domain is applied to several different targets.
The above examples demonstrate that the various BTM excerpts mix their
metaphors in rather different ways, even if they all are good examples of how
speakers and writers mix, or twist, their metaphors in discourse. Further corpus
and experimental studies should clearly tease apart the different ways that target
and source domains are referred to in mixed metaphors, both within and across
utterances or sentences. For now, the question remains whether people are capable
of inferring speakers’ communicative messages when reading the mixed meta-
phorical narratives in our small BTM collection.

4.3  Study 2

Study 2 investigated ordinary readers’ understandings of the BTM excerpts. We

asked a group of university students to read the different vignettes, one line at a
time, and write out their immediate interpretations of each phrase. This method
has long been employed in psycholinguistics especially in regard to how people
interpret longer texts expressing complex meanings (see Gibbs & Colston, 2012
for a review). Although this “write out after interpreting” method does not tap into
fast-acting, unconscious processing of linguistic meaning, it enables research to
more fully explore the products of complex figurative understanding.
Our general expectation was that people should not only give relatively con-
sistent figurative interpretations for the different conventional metaphors, but also
give evidence of inferring both the conceptual metaphors and submetaphors that
presumably motivated the production of these phrases. Thus, people should give
evidence in their protocols of both conceptual metaphorical understandings and
submetaphors arising from reasoning about the different source domains. This
reasoning process makes use of ancillary assumptions about the source topic,
including those relating to people’s social and cultural associations for the topic.
Finally, we predicted that people would also provide fairly consistent readings of
the excerpts, demonstrating not only similar readings of the different m
­ etaphorical
 Chapter 4.  Tackling mixed metaphors in discourse 

phrases, but also some sense of how these meanings could be integrated into a
global understanding for each BTM vignette.

4.3.1  Method
4.3.1.1  Participants
Eighteen Psychology students from the University of California, Santa Cruz par-
ticipated in this study and received course credit for their service.

4.3.1.2  Materials and procedure

Participants were presented with a booklet that contained all the BTM vignettes,
with each phrase being shown on its own page. The participants’ task was to read
each phrase and write down their understanding of “What is the speaker’s mean-
ing?” for each item. There were no restrictions on how much a participant could
write in response to each phrase, but they typically wrote about two to three lines
for each component of the BTM vignettes.

4.3.2  Results
Participants wrote down a variety of responses describing their understandings
of speakers’ messages for each metaphorical phrase in the BTM vignettes. For
instance, consider some parts of the participants’ responses to understanding the
first phrase, “As I look at it with a broad brush,” from the vignette presented at the
beginning of this chapter:
He is saying how broad the problem is.
I understand it to mean “generally”
The meaning has to do with the broad scope of things.
Have an open mind.
The speaker looks at it with an open mind and general understanding
As one looks at it generally and quickly.
The individual is looking very broadly.
He is looking without account for small things.
Looking at the big picture, in broad terms.
He is looking with bushy eyebrows intensely at something.
He is looking at it with an open mind.
As I look at it with a broad mind that is sharp.

For at least 10 of the 11 responses listed here, participants gave some indication
of understanding that “broad brush” refers to approaching an abstract topic in
a general, open manner. Only a single participant gave a literal reading of the
phrase, but one that was apparently based on a misunderstanding of “brush” as
being “bushy.” Not surprisingly, people’s responses often contained metaphorical
 Julia E. Lonergan & Raymond W. Gibbs, Jr.

words. Metaphors are notoriously challenging to paraphrase and earlier research

has shown that people often produce metaphor when asked to paraphrase meta-
phor (Gibbs, 1994). Our analysis suggested that participants gave metaphorical
responses to different metaphorical phrase 50% of the time (e.g., “an open mind,”
“looking very broadly” or “going down the drain” in response to “going South”). Of
course, people’s nonmetaphorical responses (e.g., “I understand it to mean ‘gener-
ally’) also can give indications of their metaphorical understandings of the differ-
ent phrases.
We reviewed each response for whether it indicated a metaphorical or non-
metaphorical interpretation of the specific BTM phrase. A specific response was
marked “metaphorical” if the participant gave evidence that the speaker’s message
refers to some abstract, non-physical situation. Overall, 96% of the participants
gave metaphoric interpretations for the phrases. The findings demonstrate that
people predominately understood the different parts of the BTM excerpts as con-
veying metaphorical meanings.
We next assessed whether participants’ responses gave any indication of the
possible conceptual metaphors underlying the different metaphorical phrases.
For example, one student responded to “As I look at it with a broad brush” with
the metaphorical answer “They mean look at it with an open mind.” This response
suggests that the reader understood the phrase as relating not to physical vision,
but to the more abstract idea of tryng to understand something is to try
and see that thing quite clearly, which is a more specific instantiation of the
primary metaphor knowing is seeing. Moreover, people appeared to be infer-
ring, in this case, that ideas are objects given that the object to be compre-
hended in this context was abstract, namely the economic crisis. Our examination
of the participants’ responses showed that 88% of these alluded to different con-
ceptual metaphors as partly motivating the meanings of the various metaphori-
cal phrases. This suggests that people’s understandings of different metaphorical
phrases in the BTM excerpts were motivated by various entrenched conceptual
metaphorical mappings.
The fact that people were highly consistent in their interpretations of the
different metaphorical idioms, and understood something about their underly-
ing metaphorical underpinnings, despite the highly conventional nature of these
phrases, also shows that people can readily infer different conceptual metaphors as
they read through the BTM mixed metaphor excerpts. This finding by itself is not
at all surprising given the vast psycholinguistic literature showing how concep-
tual metaphors are often automatically accessed during metaphorical language use
(Gibbs, 1994; Gibbs & Colston, 2012). However, the fact that people could readily
infer implicit conceptual metaphors while reading the mixed metaphor excerpts
shows, for the first time, the degree to which people are cognitively flexible in
 Chapter 4.  Tackling mixed metaphors in discourse 

switching between different conceptual metaphorical understandings of language.

People appear to possess metaphorical flexibility even when they interpret a series
of verbal metaphors based on very different source domains or metaphors (as was
the case for most of the BTM vignettes).
People’s responses, however, also provided excellent evidence of their under-
standing of the different phrases as being motivated by various submetaphors. The
submetaphors were identified through examination of the more specific adjuncts
mentioned in people’s responses. For example, when understanding “looked at it
with a broad brush,” people did not just understand this as referring to knowing
is seeing, but explicitly noted that the “broad brush” referred to seeing something
with an “open mind,” which may reflect the metonymic inference that a brush is
an instrument to clarify what one is seeing. Putting this altogether, this one
response reflects the idea that trying to understand something is perceptu-
ally examining it with a specific instrument of sight.
More specific analysis of participants’ protocols revealed an assortment of
associations they had to what the specific metaphorical terms showed about the
topic target. This included cases where participants commented on the speak-
er’s particular beliefs (e.g., strong or weak), wants or needs (e.g., the speaker is
“expressing displeasure toward how Congress uses the internet to punish peo-
ple”), attitudes (e.g., positive or negative), and certainty (e.g., how certain was the
speaker of what he/she was saying).
On average, people mentioned 3.5 of these associations, or specific adjuncts,
for their interpretations of each mixed metaphor vignette. Across all of the par-
ticipants’ responses to the different metaphorical phrases they read, 73% of them
contained references to these ancillary assumptions. We do not report this finding
to test a specific hypothesis in which we compared this percentage to some null
hypothesis. Nonetheless, the fact that such a strong majority of respondents gave
these elaborated interpretations shows that people clearly understood the verbal
metaphors in a far richer way than simple inferring their underlying conceptual
metaphors. Instead, people make use of ancillary assumptions and beliefs about
the specific source domains to create detailed interpretations of what these mixed
metaphors mean in context.
Finally, participants’ responses showed that their understandings of the dif-
ferent metaphorical phrases were not isolated, but that they were integrating these
interpretations across the various verbal metaphors within each vignette. Con-
sider again the “As I look at it with a broad brush” narrative. People gave evidence
in their protocols of the idea that ideas are objects and that being able to see
the objects from a certain perspective enabled them to understand the ideas. They
then inferred that “A lot of things going South at the same time” specifically related
to the abstract idea referring to the economic crisis. The next phrase, “Where is
 Julia E. Lonergan & Raymond W. Gibbs, Jr.

the good news coming from?” does not have metaphorical meaning, but people
unanimously understood that this referred to finding some solution to the eco-
nomic crisis. The final phrase “There is no silver bullet out there” refers to the idea
that no special, unique solution available to take care of the economic crisis.
Most generally, the participants showed good evidence of not just understand-
ing the verbal metaphors one-by-one, but understood each one as part of a larger
metaphorical whole referring in this case to stating the problem of the economic
crisis, noting its severity, and realizing that there may be no miracle solution avail-
able to take care of the problem. These findings support the main contention of
hypothesis 3, that merging of multiple and distinct objects in the source domain
does not inhibit linguistic comprehension. Instead, VNMA adjuncts broaden
and tighten the scope of the conceptual metaphor mapping by linking the stable
(transcendent and non-parallel) aspects of the target domain (affect and values)
to a dynamic source domain. People did not simply understand the different
metaphorical idioms in terms of basic conceptual metaphors, but used their local
cultural knowledge to further refine the precise metaphorical meanings of the dif-
ferent BTM phrases.

4.4  Conclusion

Mixing metaphors may often produce humorous narratives that still make good
sense to listeners and readers. Our examination of the construction and under-
standing of a specific collection of mixed metaphors, taken from The New Yorker
magazine’s “Block That Metaphor!” column, reveals that people produce mixed
metaphors using many conventional expressions employed to convey metaphori-
cal messages in other contexts. The novelty of these mixed metaphor narratives
comes from both the way different source domains are used to talk of a specific
abstract target topic and from the extended exploration of a single source domain
when talking of various target topics (e.g., “lip service,” “lathered,” “razor’s edge,”
“shaved,” and “trimmed” all in reference to discussion of funding for education). In
both cases, speakers are elaborating upon their understandings of various source
domains, with some of these extended metaphors perhaps resulting from a type of
within-domain lexical and conceptual priming.
Our most important empirical finding, though, is that readers of these mixed
metaphor vignettes appeared to have little difficulty interpreting these narratives
at either a local, single metaphor level or in figuring out a coherent meaning for
each passage as a whole. This is seen as evidence against Hypothesis 1. Of course,
readers’ possible previous familiarity with the metaphorical meanings of the indi-
vidual verbal metaphors in these narratives surely facilitated their understanding
 Chapter 4.  Tackling mixed metaphors in discourse 

of both these phrases and the overall vignettes. But it was readily apparent that
readers did not simply retrieve a frozen, stored conventional meaning for each
phrase. Instead, they understood something about what each individual metaphor
meant or implied in its specific context and then were able to integrate these into
more coherent metaphorical narratives as they continued reading each vignette.
The key here was people’s ability to engage in more than parallel source to target
domain mappings. People drew more extensive source domain inferences, based
on their social and cultural knowledge, about what a speaker or writer was refer-
ring to, or alluding to, through the use of different mixed metaphors. In this man-
ner, the present findings support proposals from Barden (this volume, and also
see Lee & Barnden, 2001), also given as Hypothesis 3, on the importance of non-
parallel systems, or value neutral mapping adjunct, in understanding metaphors,
mixed or not.
Our findings, on the surface, appear to show the inadequacy of conceptual
metaphor theory for explaining both the production and understanding of mixed
metaphor, particularly if it is assumed that general level conceptual metaphors
provide the primary foundation for people’s interpretation of mixed metaphors.
But this is not the conclusion we wish to draw. Conceptual metaphors may surely
be an important motivating force guiding people’s use and interpretation of meta-
phorical language, exactly as has been shown in extensive linguistic and psycho-
logical research. Still, conceptual metaphors alone do not provide the complete
basis upon which linguistic metaphors, mixed or not, are understood, even if they
are often critical to building a meaningful structure representing people’s under-
standing of metaphorical language. As the present research shows, people under-
stand mixed metaphors by inferring more detailed submetaphors, such as trying
to understand something is perceptually examining it with a specific
instrument of sight. This metaphorical understanding, then, is a more specific
instantiation of a basic conceptual metaphor, such as knowing is seeing. We
believe it appropriate to talk of this inferring of a specific conceptual metaphor as
something that people spontaneously create in the moment of understanding ver-
bal metaphors in context, rather than accessing a previously stored conventional
conceptual metaphor. One way of expressing this sort of detailed metaphorical
understanding is that it represents an emergent “structural metaphor,” as argued
for by Cameron (this volume, 2011).
For us, it ultimately does not matter whether one adheres to ideas about con-
ceptual metaphors, VNMAs, or structural metaphors. It is clear, though, that the
specificity of people’s understandings of verbal metaphors in narratives is aided by
their rich social and cultural knowledge of the source domains referred to explic-
itly in language. Even if the underlying conceptual metaphors appear to clash,
people makes use of their elaborate source domain information to create specific
 Julia E. Lonergan & Raymond W. Gibbs, Jr.

concepts and images that often make mixed metaphors perfectly coherent and, at
times, delightful to hear and read.

References

Cameron, L. (2011). Metaphor and reconciliation: The discourse dynamics of empathy in post-
conflict conversations. London: Routledge.
Gibbs, R. (1994). The poetics of mind: Figurative thought, language, and understanding. New
York: Cambridge University Press.
Gibbs, R., & Colston, H. (2012). Interpreting figurative meaning. New York: Cambridge Univer-
sity Press.  doi: 10.1017/CBO9781139168779
Kövecses, Z. (2010). Language, mind, and culture. Oxford: Oxford University Press.
Lakoff, G., & Johnson, M. (1999). Philosophy in the flesh. New York: Basic Books.
Lee, M., & Barnden, J. (2001). Reasoning about mixed metaphors with an implemented AI sys-
tem. Metaphor and Symbol, 16, 29–42.  doi: 10.1207/S15327868MS1601&2_3
Pragglejaz Group. (2007). MIP: A method for identifying metaphorically-used words in dis-
course. Metaphor and Symbol, 22, 1–40.  doi: 10.1080/10926480709336752
Steen, G., Dorst, A, Herrmann, B., Kaal, A., Krennmayr, T., & Pasma, T. (2010). A method for
linguistic metaphor identification. Amsterdam: John Benjamin.  doi: 10.1075/celcr.14

Appendix A

Block that Metaphor Vignettes The New Yorker Magazine, (1997–2009)

The column produced more flames than an oil field in Abu Dhabi. “The hysterical tone of
the column is astounding,” wrote cyberpundit Brock N. Meeks. “This sort of journalistic tripe
is poison and yet, at the same time, grist for the mill among the twisted jackals who make up
Congress and who, it seems, have no qualms about using the Internet as a a personal whipping
post whenever it suits their fancy.”
The criticism of Kenneth Starr is that he knows when to cut and run. Unless he drops
another shoe, he may be lucky to get out of another town without some tar and feathers from his
fans who had hoped he would bring down the Clintons with his 2 ½- year investigation.
“I don’t want to say they lost sight of the big picture, but they have marched to a different
drummer,” Victor Fortuno, the general counsel of Legal Services Corporation, said of the indi-
vidual lawyers’ challenges. “Whether it will upset the apple cart, I don’t know.”
Criticizing those who wanted more time to consider whether to hire personal assistants,
she said: “Those who don’t have the balls to go ahead and do what they need to do, and bite
whatever bullet or whatever ball they need to bite, then they ought to allow for some of us
to bite it.”
“As I look at it with a broad brush, there are a lot of things going south at the same time,”
said Morris Goldstein, a former International Monetary Fund official and senior fellow at the
Institute for International Economics in Washington. “Where’s the good news coming from?
There’s no silver bullet out there.”
 Chapter 4.  Tackling mixed metaphors in discourse 

“Until I feel comfortable I’m going to be like a burr under a saddle blanket because I want
to know exactly what kind of a pig in a poke we bought here,” Sen. Chuck Swysgood, R-Dillon,
told an informal gathering of lawmakers and state officials.
“I don’t care if it’s $100,” said Dick Sargent of Golden, an I-25 proponent. “It distracts from
the goal, and the goal is so big–we’re hunting elephants with a BB gun. If we don’t all get behind
fixing the hole in the hull, we’re all going down. And that team spirit isn’t there.”
“We’re not unconcerned,” said Mary Matalin, Vice President Dick Cheney’s chief political
aide. “We’re not so inflexible or blind that we’re like Stepford wives and husbands marching like
lemmings over a cliff. What we’re doing now is recalibrating.”
At last, the lip service that education has been lathered in for the past couple decades seems
to have found the razor’s edge among people who are willing to do something about improving
it here, at the elementary level all the way up through public universities. But as budgets con-
tinue to be shaved, will the arts once again be trimmed from the programming?
part ii

Reasons for Mixing Metaphor

 chapter 5

Mixed metaphor
Its depth, its breadth, and a pretence-based approach

John Barnden
University of Birmingham, UK

The article sketches how a particular approach to metaphor, the ATT-Meta

approach, which has been partially realized in an implemented AI program,
copes with various types of mixing. Mixing here is broadly construed as including
felicitous compounding, not only infelicitous mixes such as when there are
unintended comical effects. The structures of mixing considered included
chaining (called here serial mixing), parallel mixing (e.g., when different
metaphorical views are brought to bear on the target of the metaphor), and
combinations of serial and parallel mixing. ATT-Meta has specific technical
advantages as regards mixing. These include (i) a focus on individual mappings
as opposed to packages of mappings such as conceptual metaphors, (ii) the use of
generic mappings called view-neutral mapping adjuncts, which are not specific
to particular metaphorical views and cope with certain core types of information
that are commonly manipulated in metaphor, and (iii) the construal of mappings
as going from pretence spaces (fictional spaces) to surrounding reasoning spaces
as opposed to going from one domain to another.

5.1  Introduction

Lee & Barnden (2001) distinguished different types of mixing while seeking to
account for them within a common overall framework, namely the ATT-Meta
approach to metaphor. This approach has been realized in a working computer
program also called ATT-Meta (Agerri et al. 2007; Barnden, 2001a,b, 2006a, 2008,
2015). In this chapter I give a more extensive, updated account of how ATT-Meta
deals with mixed metaphor, and discuss distinctive features of ATT-Meta that help
with the processing of mixed metaphor. These features include: the use of pretence
spaces; nesting of them in different ways to account for different patterns of meta-
phor mixing; a flexible, opportunistic method for a­ pplication of mappings; and the

doi 10.1075/milcc.6.05bar
© 2016 John Benjamins Publishing Company
 John Barnden

use of very generic mappings that provide source-to-target transfer of information

of certain general sorts, irrespective of what the specific target and source are and
irrespective of the specific metaphorical connection between them is being used
in a discourse segment.
This chapter takes a liberal view of what mixing includes. “Mixing” is usually
taken to mean that the same target A is viewed both as B and as C more or less at
the same time in a piece of discourse, with B and C being distinctly different source
subject matters. However, I regard that phenomenon as parallel mixing. Mixing in
this chapter also includes metaphor “chaining” – when A is viewed as B where B is
in turn viewed as C. I refer to chaining as serial mixing. This terminology deviates
from the practice of authors such as White (1996) who explicitly exclude chaining
from the realm of mixing. But it is more fruitful to think of all the different ways of
combining metaphors within a discourse as mixing, especially because (as we will
see) there are demarcation disputes between the types, and fuzziness about whether
an utterance exhibits mixing at all. Of course, one could insist on using a word like
“combining” or “compounding” for the general phenomenon, and reserve “mixing”
for the parallel case, or more especially for parallel cases with comical or negative
stylistic effects, but it is undesirable to use terms that are quasi-synonyms in ordi-
nary discourse for importantly different technical purposes.
Some naturally occurring examples of mixed metaphor of sorts that are of
interest in this chapter are:
(1) “[T]he thought of her step-mother’s arrival … hung over her mind like a
dark angry cloud.”1
(2) “We do not have a chocolate army [that] fades away at the first sign of
trouble.”2
(3) “[The weather is] settling into a drier frame of mind[.]”3
(4) “But [Ireland] is also an island, divided, angry, full of old demons and old
hate.”4
(5) “This liberation of his spirit from the load of his weakness …”5

.  Jolly, S. Marigold Becomes a Brownie, p. 44. London, U.K.: Blackie & Son – The Anytime
Series (no date).
.  From Question Time programme on TV channel BBC1, UK, 16 July 2009.
.  From a weather report on BBC Radio 4, U.K., 7 am, 30 July 2003.
.  From Iris Murdoch, The Book and the Brotherhood, p. 82. London: Penguin Books, 1988.
.  From G.K. Chesterton, The Man Who Was Thursday: A Nightmare, p. 66. London: Penguin
Books, 1986.
 Chapter 5.  Mixed metaphor 

(6) “Sharon pulled herself out of her jeans, the words ‘How could he? How
could he?’ jumping about her wearied brain. Senseless, leaving her empty,
cold, helpless. Another voice, angry and vindictive, shouted in her ear,
‘Serves you right, you silly fool: play with fire and watch your life go up in
flames. It was all so predictable[.]’ ”6

I will be commenting on these and other examples below, explaining why they
involve mixing and what sorts of mixing they involve. As a brief indication, in
(1) we have serial mixing of a view of a thought as a cloud and a view of the cloud
as a person. Also, that thought-as-cloud leg is mixed in parallel with mind-as-ter-
rain. In (5) we have a parallel mixing of spirit-as-person with a view of weakness
as a weighty physical object.
The plan of the article is as follows. Section 5.2 describes the ATT-Meta
approach in general, without focussing on mixing, but concentrating on aspects
that are especially relevant to mixing. Section 5.3 discusses the ATT-Meta approach
to mixed metaphor, and points out how the use of a form of pretence, the particu-
lar nature of ATT-Meta’s mappings, and other aspects of the approach help with
the processing of mixed metaphor. Section 5.4 discusses the possibility that in gen-
eral there may be no objective matter of fact about whether a particular discourse
segment involves mixing of metaphor, or, if it does, about what particular sort and
structure of mixing it involves. Section 5.5 briefly concludes.
A matter we will have to leave aside is the important one of whether metaphor
can be rigorously distinguished from other figurative phenomena such as meton-
ymy. If it can’t, then important types of mix such as metaphor/metonymy mixes
(Goossens, 1990; Ruiz de Mendoza Ibáñez & Velasco, 2002) become indivisible in
principle from the case of metaphor/metaphor mixes. Elsewhere (Barnden 2010)
I have argued that the notions of metaphor and metonymy are just loose heuristic
ones, with scientific reality lying at a lower level consisting of certain fundamen-
tal dimensions, and allowing compromises and overlaps between metaphor and
metonymy. That work extends the idea of some other authors that there is a spec-
trum of phenomena on which typical metaphor and typical metonymy are merely
particular points (Dirven, 2002; Radden, 2002). However, it would be too complex
to involve metonymy in this chapter.
I will regard simile as one type of surface form that metaphorical expression
can take. However, this is mainly for brevity, and readers who think of simile as a
distinctly different phenomenon from metaphor can often take me to be using the
word “metaphor” metonymically to mean metaphor or simile, at the cost of taking
the chapter to cover mixtures of metaphor and simile!

.  From magazine My Story, May 1995, p. 17. Gibraltar: Editions Press Ltd.
 John Barnden

5.2  The ATT-Meta approach

Here I outline the ATT-Meta approach to metaphor, without specifically attending

to mixed cases, although a form of mixing will in fact arise naturally in examples.
ATT-Meta is first and foremost a theoretical processing account of aspects of meta-
phor understanding. It could inform psychological modelling of human metaphor
understanding or intelligent computer programs for metaphor understanding,
and has certain philosophical implications (not spelled out in this article) for the
nature of metaphor and metaphor understanding. Thus, in essence, the approach
was developed in order to investigate the fundamental nature of metaphor.
But for the purposes of ensuring that the theory is workable and conceptually
coherent, the approach has been partially implemented in a computer program,
called the ATT-Meta system or program (or just ATT-Meta, when this is clear
enough). The working system does not currently actually accept natural language
sentences or have a metaphor-identification aspect. Rather, it is a system for han-
dling just the reasoning and source/target mapping actions that the ATT-Meta
approach holds to be needed for handling a certain broad type of metaphor.7 Also,
the system is not intended to be definitive as to how the approach should be imple-
mented in computer software – many other implementations of the broad prin-
ciples of the ATT-Meta approach could be envisaged.

5.2.1  ATT-Meta’s orientation and a quick example

The ATT-Meta approach is primarily geared towards a point somewhere between
lexicalized metaphor, requiring just look-up of the metaphorical meaning, and
metaphor that puts subject matters together in an entirely unfamiliar way. The
approach mainly addresses metaphorical language that does rest on known map-
pings between subject matters, but where there are utterance elements that are not
directly mapped by those mappings. Such language goes open-endedly beyond
known mappings. The following example serves to convey a general idea of the
type of metaphor to which ATT-Meta is geared, and to illustrate some of ATT-
Meta’s principles.
(7) “One part of Mary was insisting that Mike was adorable.”8

.  However, there is an ongoing project Gargett & Barnden, 2015 to connect the ATT-Meta
system to an implementation of Embodied Construction Grammar (Feldman, 2010), which
has some metaphor-handling abilities. This will provide a natural-lanuage front-end for ATT-
Meta and provide part of a capability for metaphor identification.
.  This is a simplified version of a real-discourse example given in full later.
 Chapter 5.  Mixed metaphor 

I take this to rest on two very general metaphorical views that are often used about
the mind, First, there is the view of a person or a person’s mind as containing peo-
ple (that I will call “sub-persons”) with their own mental states. The sub-persons
are often portrayed as parts of the person as in (7). For brevity I call this view mind
as having parts that are persons. Secondly, the sub-persons may be portrayed
as communicating in natural language, again as in (7). In such a case, the utter-
ance also rests on a metaphorical view of ideas as internal utterances, so we
immediately have a case of mixed metaphor. I will concentrate on mind as having
parts that are persons for now.9
In the ATT-Meta approach, a metaphorical view only involves a small number
of very general, high level mappings. In the case of mind as having parts that
are persons, one mapping is as follows: a person having some reason or motive to
believe/desire/intend/fear/like/… something is metaphorically viewed as (meta-
phorically corresponds to, is metaphorical mapped to) at least one sub-person hav-
ing a reason or motive to believe/desire/intend/fear/like/… it. So, if we know from
an utterance such as (7) that some sub-person believes something, then a fortiori
that sub-person presumably has a reason/motive to believe it, and hence, via the
mapping, so does the overall, real person. (From now on I will talk just about
motives to believe, for brevity, rather than motives or reasons.)
The metaphorical view allows for different sub-persons to have differ-
ent beliefs (etc.) that conflict with each other, in which case the real person has
motives to believe various conflicting things, without believing any one of them.
Thus, the view caters for utterances like “One part of me believes that angels exist,
but another part believes they don’t.” The real person has both a motive to believe
that angels exist and a motive to believe that they don’t.
The only other mapping involved in mind as having parts that are p­ ersons
is between the overall person believing/desiring/… something and every meta-
phorical sub-person believing/desiring/… it. This is important, as follows.
In many utterances based on mind as having parts that are persons, only
one sub-person is explicitly mentioned, as for instance in “One part of Peter thinks
angels exist.” Here we just get the conclusion that the real person has a motive to
think that angels exist. There is no conclusion that he/she has a motive to think
that angels do not exist. Rather, there is a weaker effect, namely the ­conclusion

.  There are many examples of the use of the two views in the ATT-Meta databank
(Barnden,  n.d.). There and in previous papers I have called the first one mind parts as
persons. That label is briefer but more inaccurate than mind as having parts that are
persons. It is inaccurate because it looks as though it is assuming that the mind really has
identifiable parts. But in fact the analysis into parts is merely an aspect of the metaphorical
view itself.
 John Barnden

that it is not the case that Peter believes that angels exist. This arises because, prag-
matically, we can presume that there are other, non-mentioned, sub-persons that
lack the belief that angels exist (because otherwise why would the speaker have
mentioned just one sub-person as believing that angels exist?). But this does not
mean that these additional sub-persons positively believe that angels do not exist.
Rather, from the fact that not all sub-persons believe that angels exist, we can infer,
by means of the second mapping above, that it is not the case that Peter believes
that angels exist. If he/she did, then all the sub-persons would.
In the example just discussed, there is no reason to suppose that additional
sub-persons have beliefs contrary to the sub-person mentioned in the sentence.
However, in (7), although again only one part is explicitly mentioned, there is now
some extra information that does allow us to infer that some other sub-person
has a belief contrary to that of the mentioned sub-person. That is, we can infer
that there is another sub-person that believes that Mick is NOT adorable. This arises
because of the real-world nature of insisting. Typically, someone insists something
because there is a conversation with a person who denies it. Thus, the presence of
a sub-person who claims that Mick is not adorable can be inferred (as a default, or
working assumption). Given a general default that when someone claims some-
thing they believe it, this sub-person presumably believes that Mick is not ador-
able. Thus, we do get the strong effect that the person has motives both to believe
that Mick is adorable and to believe that he isn’t.
A key lesson from the above explanation is the subtle meaning effects that
can arise just from deploying a couple of very general mappings and from doing
some inferencing about a sketchily presented source scenario. That inferencing is
based usually on commonsense knowledge about the real-world subject matter on
which the source scenario is based. In our examples the source subject matter is
that of ordinary groups of people and conversations, together with people’s mental
states and utterance actions such as insisting. In particular, there is no need at all
to propose that the mentioned or implied sub-persons correspond to identifiable
parts or aspects of the real person, or to propose that there is some internal, real
mental action that can be clearly held to correspond to the action of insisting in
the sentence. Rather, the mentions of parts and of insisting in (7) are merely tools
towards constructing a rich source scenario, which in turn conveys in an economi-
cal, accessible and vivid manner the presence of a conflicted state of mind.
This also illustrates one principle of ATT-Meta. The approach, while mapping-
based, tries to avoid as far as possible the on-line creation, during the understand-
ing process, of new mappings to cater for source-scenario elements for which there
is no mapping. Inference within the source scenario is done in order to find things
that existing mappings can directly work upon. In our example these things are
the motives-to-believe that individual sub-persons have. These directly ­mappable
 Chapter 5.  Mixed metaphor 

things may not even be things mentioned in the sentence. Indeed, it could be that
very little in the sentence itself ends up being mapped to the target scenario. (The
target scenario is the real person’s mental states, in our example.) Neither the men-
tioned “part” nor the insisting is mapped to anything.
ATT-Meta is focussed on sentences like (7) that, arguably, contain meta-
phorical elements that are not lexicalized for the hearer (the hearer has no stored
­target-scenario meaning for them) and that, moreover, go beyond the metaphori-
cal mappings the hearer knows about. I call such sentences map-transcending. This
is of course a relative notion, being dependent on what is lexicalized for the hearer
and what mappings the hearer possesses. Map-transcending metaphor could be
said to be a form of “extended” metaphor, though this is a vaguer term.10 The fol-
lowing are some other examples of map-transcending metaphor – more precisely:
metaphor that is plausibly map-transcending for a typical hearer.
(8) “Even today, within the deepest recesses of our mind, lies a primordial fear
that will not allow us to enter the sea without thinking about the possibility
of being attacked by a shark.”11

This rests on prevalent metaphorical views of mind as physical space and ideas
and emotions as physical objects. It is unlikely that ordinary hearers have a
detailed enough non-metaphorical conception of the mind that supplies anything
to which the mentioned “recesses” could map, let alone the notion of them being
“deep” or what it means for a fear to “lie” somewhere. Thus, the recesses and their
depth, and the lying of the fear, are map-transcending aspects of the utterance.
They are there just to convey that we have a fear that comes to consciousness in
appropriate moments (entering the sea) even though it is normally not something
we are consciously experiencing.
(9) “The managers were getting cricks in their necks from talking up [to some
people in power over them] and down [to the managers’ subordinates].”12

Arguably the neck-cricks are a map-transcending element getting at the idea that,
in the source scenario, the managers constantly have to turn their heads to talk
to people physically above and physically below, and thereby acquire cricks. It is
common for abstract control relationships, especially in organizational settings,

.  Lexicalized metaphor in the sense intended corresponds roughly to conventional meta-
phor. However, my concern is with whether a word or phrase is lexicalized for some particular
understander, whereas conventionality is about a language, not specific understanders.
.  http://sharkresearchworldwide.org/interactions.htm, accessed 31 July 2013.
.  Goatly (1997: p. 162). The example is from the Daily Telegraph newspaper.
 John Barnden

to be metaphorically viewed in terms of relative vertical position of the people

concerned. However, this view does not address the question of someone having
a crick in their neck. Also, no conventional metaphorical sense for “crick” appears
in a consulted dictionary (Chambers). Only one example was found in the B ­ ritish
13
National Corpus (BNC) of metaphorical neck cricks, or other neck pains, being
used metaphorically to describe mental/emotional states in situations with no
actual or potential turning of real heads. This example was “The draught from
Microsoft’s increasingly popular Windows is giving rival software firms a crick
in the neck,” which exploits the fact that a draught of air can cause a neck-crick.
Now, annoying things and circumstances are often conventionally described as
being a “pain in the neck’” or just “a pain.” It may therefore be possible to analyse
(9) as resting on a metaphorical view underlying these idioms, for example a view
of annoying item as a pain. However, (9) shows some elaboration in that the
specific notion of a crick is introduced, and linked to the specific context-specific
circumstance of the constant turning of the managers’ heads in two opposite,
physical directions. These cricks cause pain, emotional stress, difficulty in con-
tinuing such head-turning, and unwillingness to continue it. Such feelings and
so forth are in the source scenario, but in the presence of suitable, rather generic
mappings (see below), the target-scenario effect is that the managers experience
emotional stress as a result of their conversations, difficulty in continuing them
and unwillingness to continue them.
Consider now
(10) “I don’t think strings are attached. If there are any they’re made of nylon.”14

There is a common metaphorical view of requirements as attached strings. How-

ever, the being-made-of-nylon is presumably a map-transcending element of the
second sentence. One piece of evidence of this is that no instance of “nylon” being
used metaphorically for any purpose was found in the BNC, whether or not in
conjunction with “strings.” In the context of the sentences, the nylon element
helps to convey that the requirements, if present at all, are not readily noticeable,
because of the translucency of nylon.15 These examples and/or similar ones have

.  Accessed via the tools at http://corpus.byu.edu/bnc/

.  From Newsnight programme on TV channel BBC2, U.K., 3 July 2007. Plausible punctua-
tion added. The speaker was an African politician being interviewed about a new investment
by China in mineral mining.
.  Some informants say that, out of context, the element conveys to them that the require-
ments are strong, because of the strength of nylon in contrast to ordinary string. This is a valid
alternative interpretation in general, but does not fit the actual context in which (10) lay.
 Chapter 5.  Mixed metaphor 

been analysed under the ATT-Meta approach (see for example Barnden, 2001b,
2006b, 2009, 2015). The analyses bear out the principles that map-transcending
elements of the sort above usually do not need to have mappings invented for
them, and that subtle meanings can be derived from small sets of highly general
mappings combined with commonsense inference within the scope of, and serv-
ing to flesh out, the source scenarios implied by the sentences.
As so far explained, ATT-Meta’s approach resonates with the notion of “meta-
phorical entailments” in much work on conceptual metaphor theory, based ini-
tially on the work of Lakoff & Johnson (1980). However, the work on ATT-Meta
has spelled out how entailments (essentially source-scenario inferences) work
in much more detail, while also getting away from the excessively strong notion
of entailment – the ATT-Meta approach emphasizes the uncertain nature of the
inferences involved. At a suitably high level the approach bears some strong simi-
larities to those of Hobbs and Narayanan (Hobbs 1992; Narayanan, 1999), but
the approach has distinctive features not yet mentioned, including: the use of
“pretence” spaces, “ancillary assumptions,” and “view-neutral mapping adjuncts.”
These are all very important for the application of ATT-Meta to mixed metaphor,
and will be explained in the following subsections.

5.2.2  Fictionalist/pretence-based approach

A distinctive feature of ATT-Meta, compared to other approaches in cognitive lin-
guistics, psychology and AI, is to handle (map-transcending) metaphor through
a pretence mechanism. I use a very broad notion of pretence here. It is akin and
even perhaps identical to that involved in thinking counterfactually.16 In think-
ing through what would have happened had Obama lost the 2012 US presiden-
tial election, one mentally pretends that Obama did lose the election and then
explores that pretend scenario. This broad notion of pretence in no way involves
deceiving oneself or others of anything, or of physically acting a role.
Under this weak notion of pretence, the metaphor understander pretends
that what the metaphorical sentence literally says is true, and draws consequences
from it using knowledge about the source subject matter, those consequences still
being regarded as part of the pretence. What was referred to above as a source sce-
nario is more precisely a pretended scenario. If a consequence derived within the
pretence (e.g., that there is an additional sub-person who denies that Mick is ador-
able, or that a fear in a recess is relatively inaccessible physically) can be handled
by a known metaphorical correspondence, then the correspondence can create a

.  Indeed, the ATT-Meta approach has also been shown to be able to achieve counterfactual
reasoning (Lee, 2010).
 John Barnden

corresponding claim about the target situation being addressed by the utterance.
For example, for (7) one consequence in the pretence is that that there is a sub-
person who has a motive to believe that Mick is not adorable; and the existence of
this sub-person is used by a correspondence mentioned above to derive that Mary
herself has this motive.
Mappings in ATT-Meta serve to bridge between aspects of the pretence and
reality. More precisely, since pretences can be nested within spaces other than real-
ity, including within other pretences (see Section 5.3.1), mappings serve to bridge
between a pretence space and the surrounding reasoning space – the space imme-
diately surrounding the pretence. In the remainder of Section 5.2 this surrounding
space will be a reasoning space concerning reality.
Another way of putting it is that the understander uses the literal meaning
of the utterance to construct a fictional scenario which he/she/it then elaborates,
selectively extracting information about the target through the application of
mappings. A pretended scenario is similar to a world as depicted by a fictional nar-
rative. The ATT-Meta approach is therefore akin to fictionalist approaches to met-
aphor in philosophy (e.g., Walton, 2004), and to the use of imaginary worlds for
poetry understanding (Levin, 1988). Recently, Carston and Wearing (2011) have
sketched in a preliminary way an extension to the Relevance Theory approach to
metaphor by adding what I call a pretence space. This is in order to extend Rel-
evance Theory to some of the types of phenomena ATT-Meta has been applied to.
Any pretence-based or fictionalist view subscribes to the notion that literal
meanings – or more precisely, source-domain meanings – of words or expressions
used metaphorically are active in the process of understanding and indeed are cen-
tral to it, at least under some conditions. The psychological evidence on this matter
is mixed but contains some supportive indications (for results and discussion see,
for instance: Bowdle & Gentner, 2005; Gibbs & Matlock, 2008; Rubio Fernández,
2007; Smolka, Rabanus & Rösler, 2007). A complication is that many studies look
at relative processing speed of literal and metaphorical sentences, but the idea that
literal meaning is used in computing metaphorical meaning does not have clear
implications for processing speed. For one thing it does not imply that the validity
of literal interpretation must first be discarded before metaphorical interpretation
is tried. Moreover, the time needed to resolve the difference between the iteral
and metaphorical meanings during understanding may be swamped by the time
needed to connect either sort of meaning inferentially to the unfolding context.

5.2.3  Metaphorical views and mappings in ATT-Meta

Metaphorical views in ATT-Meta are roughly similar to conceptual meta-
phors, but with a conceptual level of generality comparable to Grady’s primary
 Chapter 5.  Mixed metaphor 

­metaphors (Grady, 1997). The notion of mapping in ATT-Meta consists in the

fact that for a given metaphorical view, such as ideas as physical objects,
there is a small set of correspondence rules that can be used in order to relate
aspects of some source subject matter (being used in a pretence) to aspects of
some target subject matter (in the reasoning space surrounding the pretence).
For instance, the view of mind as physical space only currently involves two
correspondence rules. The most important one is a correspondence rule link-
ing, on the target side, an agent’s ability to mentally use an idea (in thinking, for
example) to, on the source side, the idea being physically located somewhere
within the agent’s mind metaphorically viewed as a physical space. In the ATT-
Meta system, this correspondence is actually encapsulated in a reasoning rule of
the following rough form:
(11) IF (in a pretence) a person P’s mind is a physical region
AND (in the surrounding reasoning space) J is an idea,
THEN
(in the surrounding space) P’s being able mentally to use J
CORRESPONDS TO
(in the pretence) J being physically located within that region.

This rule covers both unconscious and conscious mental use of idea J. We also
need a more specific rule, confined to conscious use:
(12) IF (in a metaphorical pretence) a person P’s mind is a physical region
AND (in the surrounding reasoning space) J is an idea
THEN
(in the surrounding space) P’s being able consciously to use J
CORRESPONDS TO
(in the pretence) J being physically located within the main part of that
region

(The main part of a mind region will be discussed below.) The IF part of such rules
acts as an appropriateness condition or guard. During processing of a metaphori-
cal utterance, it can become apparent that a person P’s mind is being viewed as
a physical region. This can happen, for instance, if the “recesses” of the person’s
mind are mentioned in the utterance, as in (8). Then, the rule above can fire for P
and any idea J that may be salient. What the rule does is create the correspondence
specified in the THEN part, for the particular person P and idea J at hand. Notice
therefore that the created correspondence is about the particular person and idea,
not all people and ideas. Also, it is dynamically constructed as just described
rather than being statically present. These are important points behind the open-
endedness and flexibility of the ATT-Meta approach, and will be shown below to
be helpful for mixed metaphor.
 John Barnden

The ATT-Meta approach itself does not ordain what metaphorical-view-­

specific correspondence rules such as the one above exist in minds or should be
used in AI systems. Rather, the approach is partly a theory of how such correspon-
dence rules can be used, in general. Naturally, in order to illustrate the application
of the ATT-Meta approach, or run the ATT-Meta system on specific examples,
particular correspondence rules need to be assumed.
There are general, informal expectations about what correspondence rules
are usually like, apart from the assumption that they include guards as above. An
important expectation is that they are at a very high level of description. They
are in this way similar in spirit to the mappings of Grady (1997), as opposed to
the original type of mappings used in conceptual metaphor theory (e.g., as in
Lakoff  & Johnson 1980). Those older mappings might belong to views such as
theories are buildings and therefore relate aspects of theories in particular to
aspects of buildings in particular. However, Grady’s mappings are illustrated by
­persisting is remaining erect and of organization is physical structure.
Grady claims that these are the key mappings needed to make sense of examples
that talk of theories as if they were buildings, and that that (usually) one does not
need to consider mappings that link theory-specific elements to building-specific
elements such as walls, windows and plumbing. The ATT-Meta approach concurs
with the general insight here, as the mappings discussed so far illustrate.

5.2.4  The pretence-based nature of mappings

In ATT-Meta, correspondences (mappings) are by definition between the contents
of a pretence and contents outside the pretence. They are not by definition between
different subject matters or domains. In Barnden (2010) I argue for scepticism,
shared with other authors such as Haser (2005: pp. 32ff), about the scientific utility
of the notion of “domain” in describing what metaphor fundamentally is or how it
works. This is despite the fact that heuristically and intuitively it can indeed often
be useful to talk of metaphor as mapping between domains.
Thus, in ATT-Meta, mappings are not tied to specific domains of life, except
in so far as may be intuitively implied (to the theoretician) by the use of particular
concepts within the correspondence-rule guards and the correspondences them-
selves. It is heuristically and presentationally useful to regard ATT-Meta’s map-
pings as relating “subject matters” to each other, and indeed the subject matters
that are linked are often qualitatively distinct in some intuitive way, such as in the
case of a mapping between mental usage of ideas and physical operation on those
ideas considered as physical objects. However, ATT-Meta makes no assumptions
whatever as to how close or distinct, or how disjoint or overlapping, the two sub-
ject matters are, and there is no use of subject matter distinctions anywhere in the
 Chapter 5.  Mixed metaphor 

approach. This point is embodied in the fact that ATT-Meta mappings technically
go between pretences and their surrounds, not between subject matters.
This point is well illustrated by the two mappings mentioned above for mind
as having parts that are persons, one of which is more precisely glossed as
follows. X is a variable standing for some proposition. Key differences between the
two rules are italicized.
(13) IF (in the reasoning space surrounding a pretence) P is a person
AND (within the pretence) P has one or more sub-persons (parts that are
persons)
THEN
(in the surrounding space) P’s having some motive to believe X
CORRESPONDS TO
(in the pretence) at least one sub-person having a motive to believe X.
(14) IF (in the reasoning space surrounding a pretence) P is a person
AND (within the pretence) P has one or more sub-persons (parts that are
persons)
THEN
(in the surrounding space) P’s believing that X
CORRESPONDS TO
(in the pretence) all the sub-persons believing that X.

Both sides of the correspondences in the THEN parts of these rules are about
people having motives to believe things. There is no useful domain distinction.

5.2.5  Detail in a sub-persons example

Here I give some additional detail of a simplification of (7):
(15) “One part of Mary was saying that Mike was adorable.”

Now, taking sentence (15) literally, the mentioned part of Mary says that Mike is
adorable. This fact about Mary is a premise used within the pretence. Given the
general default that when people claim things they believe them, the premise is
used to infer

A. the [mentioned] part of Mary (a sub-person) believes that Mike is adorable.

It follows a fortiori that

B. that sub-person has a motive to believe that Mike is adorable.

Since Mary does have a sub-person, correspondence rule (13) applies, creating a
correspondence between Mary’s having a motive to believe that Mike is adorable
 John Barnden

and the sub-person having such a motive. This correspondence can now be used to
create from (B) the proposition about reality that

C. Mary has some motive for believing that Mike is adorable.

Thus, overall, a few simple inference steps lead from a within-pretence premise
derived directly from (15), taken literally, to a within-pretence proposition (B) that
is mapped to become a within-reality proposition (C). See Figure 5.1 (where the
example is put into the present tense for simplicity).

LITERAL MEANING “One part of Mary says that

says(p, adorable(mick)) Mick is adorable”
is–part–of(p, Mary)

REALITY
SPACE

is–person(p)

believe(p, adorable(mick))

has–motive(p, believe(p, adorable(mick)))

has–motive(mary,
believe(mary, adorable(mick))
PRETENCE
SPACE

Figure 5.1.  Showing the processing for (a present tense version of) example (15). The large
box shows the pretence space. The circled arrow crossing the box boundary from inside shows
a mapping action, specifically one that arises from correspondence rule (13) in the text, as-
sociated with the mind as having parts that are persons view. The thin lines joining the
circled arrow show the dependence of the correspondence on the guard conditions in (13).
Other arrows show ordinary inference steps

Also, much as noted earlier, we can take it as a pragmatic inference from the utter-
ance of (15) that not all sub-persons of Mary say that Mick is adorable. It can then
be inferred that there is evidence that these other parts lack the belief that Mick is
 Chapter 5.  Mixed metaphor 

adorable. With the help of correspondence rule (14) it can be inferred that there is
evidence that Mary lacks the belief that Mick is adorable, since some sub-persons
may lack this belief.
As a further illustration of the open-endedness of metaphor and how ATT-
Meta can deal with it, consider the following variant of (7):
(16) “One voice inside Mary was insisting that Mike was adorable.”

This preserves the essential quality for present purposes of part of the following
real example:
(16a) “Suddenly I was having second thoughts. About us, I mean. Did I really
want to get married and spend the rest of my life with Mick? Of course you
do one small voice insisted.”17

(16) and (16a) do not mention any person-like part of Mary. But the existence of
the voice can be used to infer such a part by default, within the pretence. Moreover,
as with (7) the insistence can be used to infer by default that at least one other sub-
person has said that Mick is not adorable. That other sub-person can be inferred (by
default) to believe that Mick is not adorable. From then on the understanding pro-
cess is much as with (15), except that we now get the extra conclusion that Mary has
a motive to believe that Mick is not adorable. We therefore stiffen a conclusion that
was derived from (15) – that there is evidence that she lacks the belief that Mick is
adorable – to the default conclusion from (16) that she does indeed lack that belief.

5.2.6  Ancillary assumptions

A novel feature of ATT-Meta is the inclusion of ancillary assumptions. These are
important aspects of some metaphorical views, alongside correspondence rules
such as (13, 14). They serve to fill out a source scenario by various standard, default
expectations involved in the view. Ancillary assumptions provide an ability some-
what akin to the scenarios of Musolff (2004).
For example, a set of ancillary assumptions that I use for mind as physical
space amounts to saying that if a person’s mind is metaphorically a physical region
then, in the source scenario, that region has a (highly localized) main part, the
person has a conscious self that is a person, and this person is physically located in
the main part of the region (as opposed to subsidiary parts such as recesses or the
periphery of the region). Moreover, under the metaphorical view the conscious
self corresponds to the person (there is a metaphorical mapping between the con-
scious self and the person).

.  From magazine My Story, May 1995, p. 6/7. Gibraltar: Editions Press Ltd.
 John Barnden

The assumption of a conscious self is useful in many applications of mind as

physical space, including a mixed metaphor example below. There is no assump-
tion here that people objectively have person-like conscious selves (homunculi).
They are only being posited as an aspect of mind as physical space.
Consider this example:
(17) “The idea was in the far reaches of Anne’s mind”18

The location of the idea in the far reaches implies, within the source scenario, that
the idea is not immediately and easily operable upon physically by the conscious-
self-person, which is in the main part of Anne’s mind-region. This lack of physi-
cal operability by the conscious-self-person can be mapped, with the help of the
main correspondence associated with ideas as physical objects, to Anne’s lack
of ability, in reality, to use the idea consciously.
As another case, in using ATT-Meta on disease as possessed object
examples, I include an ancillary assumption that, if a disease is being viewed as
a possessed object, then that object is “copiable.” Copiability is a feature of some
real-world objects, for instance documents. Giving someone such an object does
not entail ceasing to possess it. This allows the approach to treat language such as
“Mary gave John a cold” as implying that John developed a cold without Mary’s
ceasing to have a cold. Crucially, just positing that a disease can be viewed as a
physical object does not of itself logically imply copiability of that object. Copiabil-
ity is an extra assumption forming part of long-term knowledge about the way the
view of disease as possessed object is used.
And an ancillary assumption is merely an assumption – a default. It is pos-
sible that a speaker could creatively talk about a disease as a non-copiable physical
object, in which case the assumption would be defeated. The ATT-Meta approach
allows this freedom. Thus, in understanding “John offloaded his cold onto Mary,”
the specific nature of offloading can be used to defeat the default that John keeps
the cold when giving it.

5.2.7  View-neutral mapping adjuncts

There was no treatment above of the effect of the “small” qualifier of the voice in
Example (16a). Suppose this qualification is added to (16). The effect of this on the
meaning of (16) is to convey the refinement that the motive to believe that Mick is
adorable is relatively unimportant in Mary’s current mental state.

.  Cf. “In the far reaches of her mind, Anne knew Kyle was having an affair, …” from article
“Facing up to the Dreadful Dangers of Denial” in Cosmopolitan, US ed., 216 (3) March 1994.
 Chapter 5.  Mixed metaphor 

How could this meaning refinement come about? Presumably because, in the
pretence scenario, the sub-person’s utterance is (by default) relatively unimport-
ant, through being relatively quiet. So, it can be inferred a fortiori that the sub-
person’s motive to believe that Mick is adorable is relatively unimportant in the
overall pretended scenario. The sub-person’s motivational state corresponds under
the mind as having parts that are persons view to Mary’s having a motive to
believe that Mick is adorable. Thus, as long as this correspondence can somehow
be used as a reason to transfer the lack of importance as well, we get the desired
effect about lack of importance of Mary’s motive to believe that Mick is adorable.
This is where view-neutral mapping adjuncts (VNMAs) come in. There are
general qualities about source scenarios that are very often transferred in meta-
phor to the target scenarios, no matter what the specific metaphorical view is. (See
related points made by Carbonell, 1982, and Narayanan, 1999.) Amongst such
qualities are evaluative properties and relationships such as goodness, value and
importance. I therefore include the following rule in the ATT-Meta approach:

(18) IF entity X in a pretence

CORRESPONDS TO
entity Y in the surrounding reasoning space
THEN
the importance of X in the context of the pretence scenario
CORRESPONDS TO
the importance of Y in the context of the surrounding space.

As qualitative degrees to which situations hold also correspond across a pretence

boundary in a situation-sensitive way – more on this below – a small degree of
importance in the pretence situation maps to small degree of importance in the
situation in the surrounding reasoning space.
The use of other VNMAs is illustrated by the following example. It is derived
from (1) and is a simplification of a real mixed-metaphor example to be treated in
Section 5.3.1.

(19) “The thought is a dark cloud hanging over her.”

The pronoun “her” refers to a young girl, Marigold, and the thought in question
is the thought of her stepmother’s arrival at the house. I analyse (19) as resting
on the metaphorical view of ideas as physical objects. The dark cloud and the
hanging-over constitute map-transcending aspects of the sentence. Now, I assume
the existence of some correspondence rules for ideas as physical objects. For
instance, one rule relates conscious usage of an idea in reality to the person’s con-
scious self physically operating upon the idea. (Conscious selves arise through
ancillary assumptions associated with ideas as physical objects, as explained
 John Barnden

in Section 5.2.6.) But it turns out that these correspondences are not relevant to
(19). The understanding of (19) is instead achieved through VNMAs. First, note
the following connotations of (19):
(20) Marigold is depressed by (or doesn’t like) the thought of her stepmother’s
arrival.
That mood is likely to persist for some time.

In a real outdoors situation, a dark cloud is depressing, or unlikable, or affectively

negative in some other way – I will stick to the depression option for simplicity of
illustration. It is depressing partly because of darkness itself but partly also because
of the indication of possible rain to come. Also, because it is “hanging over” Mari-
gold, it is static; and as clouds do not usually makes sudden movements or changes
of speed, it will probably stay hanging for some time. It will also presumably stay
dark – again, a given cloud is unlikely to change between light and dark with any
speed. A further subtlety is that it is a matter of inference that, in the pretend sce-
nario, Marigold is outdoors. (Of course, in reality she may not be.) If she weren’t,
the cloud would be hanging over her house (for example), not her. (19) is indeed
a good example of the richness and subtlety of metaphorical meaning, and how so
much can be gleaned from simple, familiar life situations.
But how are the connotations in (20) produced, more exactly? Given that the
pretence scenario has been enriched by conclusions about the cloud’s darkness
and persistence of hanging, and its depressing effect on Mary, we can then also
infer within the pretence that Mary’s depressed mood is itself persistent. This uses
a principle that if a cause persists then by default the effect persists as well. All we
need now is to be able to map the depressing effect and its persistence from pre-
tence into reality in order to get the two connotations listed above.
This mapping is done in ATT-Meta by two further VNMAs. One maps qualita-
tive temporal attributes of within-pretence situations to corresponding situations
outside the pretence. The other maps within-pretence affective states to affective
states outside the pretence, in certain circumstances. Expressed informally, these
VNMAs are:
(21) IF some situation P in a pretence
CORRESPONDS TO
some situation S in the surrounding reasoning space
THEN
(in the pretence) P’s having a specific qualitative temporal attribute
CORRESPONDS TO
(in the surrounding space) S’s having the same attribute.

A qualitative temporal attribute is an attribute such as immediacy, persistence,

intermittency, and gradualness.
 Chapter 5.  Mixed metaphor 

(22) IF something P in a pretence

CORRESPONDS TO
something S in the surrounding reasoning space
AND
some cognitive agent within the pretence
CORRESPONDS TO
some cognitive agent in the surrounding space
THEN
(in the pretence) the pretence agent’s bearing a particular mental/affective
attitude towards P
CORRESPONDS TO
(in the surrounding space) the corresponding surrounding-space agent’s
bearing the same attitude towards S.

These VNMAs will produce the connotations in (20). It is first found, through
VNMA (22), that Marigold’s being depressed by the cloud (in the pretence) cor-
responds to her being depressed by the thought (in reality). As she is indeed
depressed by the cloud in the pretence, she is depressed by the thought in reality.
Then, because of the correspondence between Marigold’s within-pretence
depressed mood and her inferred within-reality depressed mood, VNMA (21)
creates a correspondence between the persistence of the within-pretence mood
and the persistence of the within-reality mood.
This sequence illustrates the fact that one VNMA can act upon the results
of another, to create a sequence of more and more elaborate correspondences.
Schematically, with within-pretence aspects on the left and within-reality aspects
on the right:
Marigold ↔ Marigold
cloud ↔ thought
depressed-about(Marigold, cloud) ↔ depressed-about(Marigold, thought)
persistent(depressed-about(Mar., cloud)) ↔ persistent(depressed-about(Mar.,
thought))

The first line here reflects a general feature of ATT-Meta’s handling or pretence,
namely that entities can lie in more than one space, although they may change
their nature between spaces. Entities keep their nature on going into a pretence,
unless there is something about the pretence that change their nature. So Marigold,
who is in both the reality space and the pretence space, is just a person in both.
But the thought changes into a cloud in the pretence. The multiple presence of an
entity in different spaces can be construed as a matter of identity mappings across
space boundaries.
Other VNMAs include ones that handle the following: complementation
(e.g., converting a proposition about loving X, or believing that Y, to not loving X,
 John Barnden

or failing to believe that Y; or converting in the opposite direction);19 further tem-

poral information such as time order of events and (qualitative) rates of change;
causation/enablement/attempting relationships; ability to do something; tendency
to do something; normal functioning (i.e., something doing what it is designed
or evolved to do); degree to which a situation holds; and uncertainty with which
a situation holds. Notice that VNMAs do not rely on any specific metaphorical
view, and are generic in that sense. On the other hand, they are merely default
rules, so their implications can be defeated in specific circumstances by other
evidence.
Work on the ATT-Meta approach indicates that metaphorical utterances
often get much, and some cases almost all, of their effect via VNMAs rather than
directly from view-specific mappings. The latter often merely provide a scaffold
to allow VNMAs to handle the most important information. The treatment of
a wide variety of examples in (Barnden 2001b, 2006b) provides evidence for
this claim. To take one case, consider again Example (9). The emotional dis-
tress from the neck crick, and indirectly caused by the managers’ conversations,
transfers to become emotional distress in reality, caused by the conversations,
because of VNMA (22) handling emotional states and because of a causation
VNMA. (Note, however, that the causal chain in reality space is not assumed
to contain items that correspond to the neck-crick itself or the physical pain it
causes.) Equally, the within-pretence unwillingness of the managers to continue
with the conversations, and the difficulty in doing so, transfer to reality, because
of VNMAs handling temporal matters (the potential continuation itself), emo-
tional/mental states, ease, and degrees. In all this the only view-specific map-
ping used is the very basic one of relative vertical position corresponding to
control relationships.

5.2.8  Goal-directed reasoning

The ATT-Meta approach gives a major role to goal-directed reasoning. Although
the descriptions of reasoning above are couched as moving forward from prem-
ises towards conclusions, the process is actually typically assumed to proceed in
a goal-directed way.20 That is, there is some goal or issue that the system is trying
to address, and reasoning steps are attempted towards that end. For example, in
the case of (19), the actual context raises the question of Marigold’s mood and her

.  This VNMA has already been implicitly used several times in examples.
.  The overall ATT-Meta approach allows non-goal-directed as well as goal-directed
reasoning to be used. However, the implemented ATT-Meta system can at present only do
­goal-directed reasoning.
 Chapter 5.  Mixed metaphor 

dislike of her stepmother. So the goal of investigating Marigold’s affective states

concerning her stepmother is posted. Given the presence of VNMA (22) han-
dling affective states, this can be converted into the goal of investigating her affec-
tive states in the pretence. By a process of backwards chaining through inference
rules, it is discovered that it is known that a dark cloud is hanging over her, and
the necessary inferences can now be rolled forward to conclude that, in the pre-
tence, she is depressed about the cloud. This then rolls forward via VNMA (22)
to become the conclusion that she depressed by the thought of her stepmother’s
arrival.
Goal-directed reasoning is an extremely powerful tool for combatting the
notorious indeterminacy of metaphorical meaning (see, e.g., Stern 2000). Suitably
deployed it can guide metaphor understanding towards uncovering meaning that
is relevant to the context. See Barnden (2009) for more on this.

5.2.9  ATT-Meta and blending

ATT-Meta has some similarity to, and some differences from, the blending-
theory approach to metaphor (Turner & Fauconnier, 1995; Fauconnier &
Turner, 2008). Pretence spaces are very like blend spaces, especially because of
something not yet mentioned: namely that a pretence can opportunistically use
information from reality, much as a fictional story such as one about S­ herlock
Holmes can use real information about London. So a pretence can be a blend
between aspects of the surrounding reasoning space and pretended world
aspects, and consequences drawn in the pretence can depend on both types of
information. Moreover, reasoning within the pretence is like the elaboration of
a blend space.
However, in the blending-theory approach to metaphor, there are specific
input spaces, and it is between these that metaphorical mappings work, whereas
in ATT-Meta metaphorical mappings work directly between pretence contents
and contents outside the pretence. Blending theory does not have a correlate of
VNMAs and has not developed an extensive concern with details of gradedness
(the matters of degree above) or uncertainty. On the other hand, blending theory
has been applied to a much wider variety of linguistic issues than ATT-Meta has.

5.3  ATT-Meta and mixed metaphor

Lee and Barnden (2001) provide an early account of how the ATT-Meta approach
deals with mixed metaphor. The present account reflects major developments
since then.
 John Barnden

5.3.1  The marigold example: Mixed form

Consider now real example (1), repeated here:

(23) “[T]he thought of her step-mother’s arrival … hung over her mind like a
dark angry cloud.”

For present purposes I take this as meaning the same as

(24) “[T]he thought of her step-mother’s arrival was a dark angry cloud hanging
over her mind.”

That is, I do not address any special effect of the simile form in (24), centred on the
“like.” A more careful analysis might have it that the thought was metaphorically
some unknown object hanging over her mind, where that object was merely like
a dark angry cloud. The example is already complex enough without bringing in
this possibility.
The main point I wish to address is that now the cloud from (19) is itself being
metaphorically viewed as “angry” in (23, 24). We have here an example of serially
mixed metaphor: the thought as a cloud, the cloud in turn as an angry person.
The first link in this mixing rests on the metaphorical view of ideas as physical
objects. The second rests on personification, which broadly speaking is the view
of non-person as person. The non-person can be any sort of entity – concrete or
abstract, and if concrete then living object or not.
Metaphorically casting inanimate objects as “angry” is common, as in saying
that a part of one’s body is or looks angry to convey that it is inflamed. This appeals
to one meaning of “angry” listed in, for instance, Chambers’ dictionary. Another
meaning, when the word is applied to the sky, etc., is: “of threatening … aspect.”
However, “threatening” itself has a standard (metaphorical) meaning in Chambers
as “promising rain …” when applied to skies, clouds, etc. It is therefore reasonable
to think that a hearer of (24) has a lexicalized metaphorical meaning of “angry”
when applied to skies, clouds, etc. that is something like “indicative of rain etc.
coming shortly.”
If so, (24) could be treated by a slight enrichment of the way we treated (19)
above. We already said above that the “dark” nature of the cloud is depressing and
suggests rain. That suggestion is strengthened by the “angry” – i.e., the rain is more
likely and more imminent. This enrichment is within the scope of VNMAs: the
imminence is a matter of time-course, just as persistence is, as discussed above;
and uncertainty is also transferred out of a pretence by a VNMA. The more certain
something is in the pretence, then (other things being equal) the more certain a
corresponding situation, if any, is outside the pretence.
 Chapter 5.  Mixed metaphor 

Although “angry” is therefore plausibly a case of lexicalized metaphor, it is

instructive also to see how ATT-Meta could treat it if it were active, n ­ on-lexicalized
metaphor. The structure of the treatment is broadly the same as would occur with
other examples of serial mixing where both links are more clearly active. And in
any case an individual hearer may not in fact have an appropriate lexicalized meta-
phorical sense for the word, e.g., a child or a foreign learner of English. Even in
the case of a hearer who does know the above lexicalized metaphorical sense, the
metaphor may be reawakened because of other uses of “angry” or related terms in
the discourse context. There is also the evidence (see, e.g., Boroditsky, 2000) that
metaphor that is normally thought entirely conventional actually is unconsciously
active for us, so that it is possible that “angry cloud” etc. does make us uncon-
sciously entertain the idea of the cloud as a person.
Turning to another complication in (24), there is an added type of mixing
that may be less evident. Unlike (19), (24) says that the cloud is hanging over
Marigold’s mind, not over Marigold herself. So now Marigold’s mind is being cast
either (i) as a piece of physical terrain or (ii) as a physical object located on such a
terrain. Since mind-as-terrain is a commonly occurring special case of the com-
monly used metaphorical view of mind as physical space, I will assume here
that mind-as-terrain is what is used, as a more economical alternative than (ii). So,
if we lump Marigold’s mind and thoughts together as one subject matter, we have
one aspect of this subject matter being viewed as a physical space and another
aspect being viewed as a physical object.
Thus the overall structure of the mixing in (24) is the view of Marigold’s mind
as a physical terrain mixed in parallel with thought-as-cloud, where the latter is
serially mixed with cloud-as-person.
I now proceed to outline how the mixing is handled using the pretence space
mechanism of ATT-Meta.

5.3.2  Deployment of pretence spaces, VNMAs and inference

The treatment of (24) rests on nesting of pretences. There is an outer pretence in
which the thought of the stepmother’s arrival is a cloud; and within that pretence
there is an inner pretence in which that cloud is an angry person. See Figure 5.2.
As far as the inner pretence is concerned, reality consists of the outer pretence.
Effects flow between the inner pretence and outer pretence in just the same way as
effects flow between the outer pretence and reality.
As regards the parallel aspect of the mix, the outer pretence gets enriched by
the inference that Marigold’s mind is a piece of physical terrain. Thus in the outer
pretence the thought is a cloud hanging over that terrain.
 John Barnden

is–cloud(TSA)
is–person(CS) hanging–over(TSA, mind(M)) REALITY
SPACE
in–main–part(CS, mind(M))
is–terrain(mind–of(M))
is–near(TSA, CS)

PERSISTENCE
INFERENCE

is–angry(TSA)
persistently
(strongly by VNMAs
is–person(TSA)
(feels–threatnd(CS,TSA))

by VNMAs
strongly persistently
(feels–threatnd(CS, TSA)) (strongly
(feels–threatnd
INNER (M,TSA))
PRETENCE OUTER
SPACE PRETENCE
SPACE

Figure 5.2.  Showing the pretence structure and some major mapping and inferential links for
Example (24). M stands for Marigold. TSA stands for the thought of the stepmother’s a­ rrival.
CS stands for Marigold’s conscious self. The solid arrows with large heads show the action of
VNMAs, working in this diagram on the “strongly” and “persistently” qualifiers, and on the
feeling of threat in going from the inner to the outer pretence and from the outer pretence to
­reality. (Notice there are no view-specific mapping actions, and hence no circled thick arrows
like the one in Figure 5.1.) The circled dashed arrows show the effect of ancillary assumptions
­associated with mind as physical space. The other arrows show ordinary inference steps.
For simplicity of illustration, not all propositions and links are shown. In particular, the infer-
encing that ­derives is-near(TSA, CS) is not shown

Marigold’s mind and the thought are in all three spaces: reality, the outer pre-
tence and the inner pretence. As before, the thought changes into a cloud in the
outer pretence, and changes again into a person in the inner pretence, where
this ­personhood is inferred from the angriness. Marigold’s mind changes into a
physical terrain in the outer pretence, but stays as a physical terrain in the inner
pretence.
Recall from Section 5.2.6 that, by an ancillary assumption for mind as
­physical space, within the pretence a conscious self is inferred (by default) to
be present within the main part of the mind-space. Furthermore, this agent cor-
responds metaphorically to the real agent (Marigold) outside the pretence. Thus,
 Chapter 5.  Mixed metaphor 

by virtue of VNMA (22), which is concerned with mental and emotional states,
such states of the two agents (Marigold and her conscious self) also correspond.
Notice the slight difference here from Example (19), where Marigold herself,
rather than her conscious self, is an agent in the pretence. But the VNMA works
similarly both for (19) and our current example. For (19), it makes Marigold’s
mental/emotional states in reality correspond to her mental/emotional states in
the pretence.
The fact that, in the outer pretence, the conscious-self person is in the main
part of the mind-space can be used also in the inner pretence. This is because a
pretence can use information from surrounding spaces unless it is suppressed by
defeating information. In the inner pretence, the cloud is an angry person that I’ll
refer to as the cloud-person. Hence, Marigold’s conscious self is caused to experi-
ence a strong sense of threat from the cloud-person, in the inner pretence.
We need here the fact, in the inner pretence, that the cloud-person is near
to Marigold’s conscious self. This is also imported from the outer pretence. The
nearness is inferred in the outer pretence in the following way. Assuming that
Marigold’s thought about her stepmother is consciously entertained by Marigold,
we obtain with the help of rule (12) the result that the cloud is in the main part of
the mind terrain. Given that this main part is highly localized and that the con-
scious self is located within it, it can be assumed that the cloud-person is near to
the conscious-self person.
Now, VNMA (22), together with the identity mappings for the thought and
the conscious-self-person, lead to the creation of a correspondence between
(in the inner pretence) the conscious-self person feeling strongly threatened by
the cloud-person

and
(in the outer pretence) the conscious-self person feeling strongly threatened by
the cloud.

Thus, because in the outer pretence Marigold’s conscious self person feels threat-
ened by the cloud, and the cloud corresponds to the thought (of the stepmother’s
arrival) in reality, we get via the VNMA (22) again the result that in reality Mari-
gold feels threatened by that thought.
Further, much as for (19), the VNMA that deals with the temporal charac-
teristics of states can lead to a further important result. In the outer pretence, the
cloud is hanging, implying that this state of affairs is likely to last for considerable
time (in the context of the time scale of everyday weather events). Thus, by a com-
monsense inference concerning the fact that the feelings are about the cloud, the
 John Barnden

unpleasant feelings in the outer pretence are also likely to last for a c­ onsiderable
period. The VNMA mentioned now transfers this longevity or persistence to
become persistence of Marigold’s unpleasant feelings in the real world (on the
time scale of everyday dynamics of feelings).
One thing needs to be noted about intensities with which properties apply,
in the ATT-Meta approach – specifically the intensity of the fear in our current
example. In the outer pretence the negative feelings towards the cloud are reasoned
about in terms of the properties of real clouds and not in terms of the properties of
angry people, because in the outer pretence the cloud is not an angry person. Thus,
in the outer pretence it can be inferred that the cloud is causing negative reactions
in Marigold in the sense that it looks as though it is going to lead to rain (etc.).
So, in particular, the degree of intensity of unpleasant feeling in the outer pretence
should be inferred to be commensurate with what rain could lead to, not what an
angry person would lead to. So, it is not that the exact degree of unpleasantness is
carried over (by the degree VNMA) from inner pretence to outer pretence. Rather,
is a tendency towards high intensity in the context of interpersonal situations that
carries over to a tendency towards high intensity in the context of everyday weather
situations.

5.3.3  More on parallel mixing

In the treatment of (24), the parallel mix of ideas as physical objects and mind
as physical space naturally arose in one pretence. This is because in the outer
pretence the use of mind as physical space arose by default inference from the
use of ideas as physical objects. The view of ideas as physical objects was
inherent in the fact that the thought was viewed as a cloud. It was the fact that this
cloud was hanging over Marigold’s mind that led to the inference that her mind
was a physical space.
Mixing the two views in one pretence works well here because of the natu-
ral compatibility between the two views. It is nevertheless useful to regard mind
as physical space and ideas as physical objects as separate views. Even
though very many examples of mind as physical space do also involve ideas
as physical objects, as evidenced by the examples in the ATT-Meta databank
­(Barnden,  n.d.), there are also many cases when the views occur separately.

Ideas are often alternatively viewed as objects external to the person holding or
entertaining the ideas, as in “They kicked the idea round the room.” Conversely
the mind can be viewed as a physical space without taking ideas to be physical
objects in the normal sense of things that can move, be handled, etc. For instance,
under mind as physical space, thoughts can be portrayed as physically located
­linguistic expressions ­(spoken or written), as in:
 Chapter 5.  Mixed metaphor 

(25) “She hoped he wasn’t going to quiz her, and searched her mind for whatever
Cliffs Notes might be jotted there – just in case.”21

Felicitous parallel mixing also occurs in any case of mind as having parts that
are persons where a mind-part makes an utterance. We then have parallel mix-
ing with ideas as internal utterances. The following are some further real-
discourse examples of compatible parallel mixing. The first two are repeats of
(4) and (5) respectively.
(26) “But [Ireland] is also an island, divided, angry, full of old demons and old
hate.”

[Parallel mix of: Ireland as a physical container that can be full; hate as a physical
object that could be put into such a container; and of problems etc. as demons in
the container. There may also be parallel mixing with Ireland as an angry person
who is individually “divided” in a mental sense, but another, possibly more plau-
sible analysis is that there is a metonymy to the people of Ireland, who are literally
angry and divided from each other.]
(27) “This liberation of his spirit from the load of his weakness …”

[Parallel mix of spirit as physical person and an abstract personal quality (weak-
ness) as a heavy physical object.]
(28) “[The Dean’s] gaze went slowly up to the ceiling [where there were depic-
tions of the twelve signs of the zodiac], as if seeking comfort in his own
private astrological heaven. Comfort came to him in some measure as his
eye moved from Cancer to the taut form of Sagittarius. … At this moment
the Dean’s eye, voyaging still among his rafters, rested on Aquarius, …”22
[Parallel mix of the Dean’s eye as voyaging person and the room’s ceiling space as
outer space or heaven.]
But the views in a parallel mix may not be very compatible. In response to
this, ATT-Meta can when necessary handle parallel mixing by having separate
pretences that sit side-by-side within the same surrounding reasoning space. Pro-
visional contributions to the surrounding space are then drawn by mappings from
the two pretences, and these contributions can combine in whatever way is pos-
sible and discourse-relevant in the surrounding space. The question of whether

.  From Patti Davis, Bondage, p. 142. New York: Pocket Books (Simon & Schuster),
1994. Cliffs Notes are a popular series of educational booklets. “Cliffs Notes” in (25) refers,
­metonymically, to the sort of written notes in the booklets, not to the booklets as such.
.  From Michael Innes, Death at The President’s Lodging, p. 39. London: Penguin Books, 1988.
 John Barnden

more than one pretence is needed in a case of parallel mixing was mooted in
Lee and Barnden (2001) and is still in part an open problem in general. However,
it is possible now to be more specific than in the earlier work, as follows.
The multi-pretence approach is most natural when the pretence contents are
very different qualitatively, or when one metaphor has already been understood
and another metaphor is introduced rather separately afterwards. A good real-
discourse example of the latter case is a text about US foreign policy23 in which
we find “Afghanistan is Vietnam,” then some explanation of this view, and then
“But Afghanistan is not simply like Vietnam,” some more explanation, then
­“Afghanistan is Yugoslavia,” some explanation of this, then “But Afghanistan is not
simply like Yugoslavia,” and so forth, adding in Colombia and then Somalia in the
same cumulative way. (The text itself is too long to include here, and is a serious
piece of analysis, not a comical pot-pourri.) It is arguably most economical to set
up a new pretence to handle each new metaphor, and let the new inferences sug-
gested in the surrounding space be combined as appropriate with the inferences
from the previous metaphors. This “combining” does not preclude defeating the
previously proposed ones. Trying instead to insert the source material and the
associated mappings into an updated version of a single, old pretence could be
difficult and computationally expensive.
On the other hand, I conjecture that in cases of parallel mixing where the vari-
ous metaphorical views crop up close by in discourse (e.g., in the same fairly short
sentence), people tend first to try to use one pretence, for good or ill. Comical
effects of infelicitous parallel mixed metaphor suggest that one pretence is tried,
even if this turns out not actually to be the best approach.
It is actually a good strategy to try a one-pretence analysis early on in under-
standing, for several reasons. First and most obviously, there is an overhead in
deciding how many pretences should be used and into which pretences various
premises should be put. Indeed, secondly, it may be quite unclear until considerable
reasoning has been done that there is any mixing. Recall again the opportunistic
way in which ideas as physical objects and mind as physical space arise from
each other in examples above. Not much prior reasoning was needed to bring in the
mapping in these particular cases, but the amount needed is in general open-ended.
Thirdly, a roughly Gricean or Relevance Theoretic account of language
­(Sperber & Wilson, 1995) would predict that normally people are cooperative in
their mixing of metaphor so that in the case of mixing where the various views are
active, the mixing can normally be coherently achieved without having to worry

.  http://zioneocon.blogspot.co.uk/2004/06/bret-stephens-in-wsj-opinion-journal.html (ac-

cessed 21 March 2012).
 Chapter 5.  Mixed metaphor 

about using more than one pretence. Indeed, the hallmark of comical, unintended
effects is precisely that the speaker does not seem to have noticed a clash, and this
may be because the metaphorical senses involved are lexicalized for the speaker
(or all but one of the senses are), so the speaker does not notice the conflict at the
literal level.

5.3.4  Combining different types of mixing

We have already seen, with (24), that parallel and serial mixing acts can themselves
be combined. In (24) the combination amounted to having an outer pretence in
which two views (ideas as physical objects and mind as physical space) were
mixed in parallel, and also an inner pretence to provide a serial mix of one of those
views (ideas as physical objects) with personification of the cloud. But this is
just one shape that combination of mixing types can take. Any configuration of
nesting of pretences is in principle allowed in ATT-Meta. So, as an abstract illus-
tration, there could be three outer pretences, two of which contain an inner pre-
tence. Or any inner pretence could contain two or more pretences in parallel with
each other within it, and any of these can contain further pretences, and so on.
It is not yet clear how commonly such elaborate structures are needed in prac-
tice. However, (6), repeated here as (29), is an example of mixed metaphor that
appears to require a complex pretence structure. It is taken from a romantic-story
magazine that is presumably meant to be easily understandable by the average
person.
(29) “Sharon pulled herself out of her jeans, the words ‘How could he? How
could he?’ jumping about her wearied brain. Senseless, leaving her empty,
cold, helpless. Another voice, angry and vindictive, shouted in her ear,
‘Serves you right, you silly fool: play with fire and watch your life go up in
flames. It was all so predictable[.]’”24

I analyse this as instantiating the view of ideas as internal utterances. This

view is more basically instantiated in sentences like (25) and the following:
(30) “Some people go to bed at night thinking: ‘That was a good day.’ I am one of
those who worries and asks: ‘How did I screw up today?’”25

Sometimes in such examples there may be an implication that the person in

question had a thought episode that felt to that person like uttering/hearing
inner speech, or that felt like reading some writing. But this is not a necessary

.  From My Story, May 1995, p. 17. Gibraltar: Editions Press Ltd.
.  From interview with actor Tom Hanks, Saga Magazine, January 2006, pp. 80–81.
 John Barnden

i­mplication: a thought that is not clothed in felt inner speech or writing can nev-
ertheless be reported as if it were so. Notice that even if the person has the sensa-
tion of an inner utterance, it is still metaphorical to talk of the person’s state as
an utterance. There is no real utterance, only (at most) an inner representation
of one. If in fact the person is not being claimed to be having an inner-utterance
sensation, then there is an additional layer of metaphoricity. In fact, this is itself a
case of serial mixed metaphor: a thought episode is metaphorically portrayed as
the person having the sensation of experiencing something that is itself portrayed
metaphorically as an utterance.
Irrespective of this issue, in (29) the voice is metaphorically viewed as some-
thing that can jump around inside Sharon’s brain, so that the voice is some sort of
jumping physical object – an animate creature, by default. In particular, her brain
is being viewed as a container of such objects. But a further thought of Sharon’s is
metaphorically viewed as an external utterance, uttered by a voice shouting in her
ear. So already we have a parallel mix of ideas as external utterances, ideas
as internal utterances, and brain as container of ideas, with the ideas as
internal utterances component serially mixed with a view of a voice as an ani-
mate creature. And ideas as inner utterances is itself being inherently a matter
of serial mixing, as above, if the understander does not take Sharon to be actually
having the sensation of inner utterances.
On top of all this Sharon is metaphorically cold and empty, and either meta-
phorically or hyperbolically senseless, though the metaphors here may be lexical-
ized and therefore tending not to add to the complexity of pretence structure.
It is also almost embarrassing to have to mention that (29) involves yet another
metaphor, within the second voice’s utterance: “play with fire and watch your life
go up in flames.” This is not a case of mixing in the sense so far used: rather it is
embedding of metaphor within a reported thought. What is of special interest here,
though, is that metaphor can not only be embedded in real reported speech, as in
“Sharon said to Jamie, “My life will go up in flames””, and in thought reports such
as “Sharon thought her life would go up in flames”, but it can also be embedded
within a thought report that is itself metaphorically couched as a speech report by
means of ideas as internal utterances, as in (29).

5.3.5  Advantages that ATT-Meta brings to mixed metaphor

While ATT-Meta is not a complete account of non-mixed metaphor, let alone
mixed metaphor, it brings positive things to the table that are of great utility as
regards mixing. First, we have the already-noted freedom that ATT-Meta allows
as regards pretence structure (e.g., how pretences are nested), supporting com-
plex combinations of serial and parallel processing. This encompasses also a type
 Chapter 5.  Mixed metaphor 

of mixing that is not mentioned above and that Goatly (2002) calls multivalency.
This is where two or more different targets are both metaphorically addressed by
means of the same source subject matter at more or less the same time in dis-
course. This is a converse to the type of parallelism discussed above, where one
target is addressed by two or more different source subject matters. I call the latter
parallelism source-wise parallelism, and Goatly’s multivalency target-wise paral-
lelism. Target-wise parallelism can be realized in the ATT-Meta approach – the
two target subject matters simply occur together in the space surrounding the
pretence(s) – but as with source-wise parallelism there is the issue of how many
pretence spaces to use.
Note, however, that in the ATT-Meta approach both target-wise and source-
wise parallelism are a matter of viewpoint, and are not objective clear-cut mat-
ters. This is particularly so because of ATT-Meta not subscribing to clear divisions
between subject matters. Whether a particular stretch of discourse is about one
target subject matter or about several can be a matter of viewpoint, as can be
(though in practice to a lesser degree) the question of whether one source subject
matter is being used or more than one.
Secondly, there is an important issue about the status of metaphorical views
such as mind as physical space, etc. Although specific views are mentioned in
the exposition above, they actually have no separate reality under the ATT-Meta
approach. The theory does not propose that they have their own individual exis-
tence in the human mind, nor does it ordain that they do so in an implemented
computer system. In particular, views are not entities in the ATT-Meta system.
Rather, what is proposed as having mental or computational reality are the view-
specific correspondence rules such as (11–14) and view-neutral rules such as (18)
etc. The former type of rules can intuitively and roughly be regarded as belonging
to some particular metaphorical view, but this belonging is not explicitly recorded.
At best, the fact that two different correspondence rules belong to the same view
is implicit in the fact that they involve similar guard conditions (the IF parts of
correspondence rules). For example, a correspondence rule’s guard could contain
the condition that some idea in the reasoning space surrounding the pretence is a
physical object within the pretence. This is tantamount to checking that that idea is
being subjected to the view of ideas as physical objects. Two rules with such a
check can to that extent be regarded as sharing a metaphorical view. But the guards
could also have other, non-shared, elements, causing the rules to be used under
somewhat different circumstances.
Implicit in these points is the fact that, by virtue of their guards, correspon-
dence rules opportunistically decide from themselves whether they are relevant
or not (to put it metaphorically!). This gives great operational flexibility. No top-
down decision is needed that some metaphorical view, as such, should hold. It
 John Barnden

might happen within a pretence that something is inferred to be a physical object,

and this conclusion might be picked up by a correspondence rule’s guard. But the
mechanism performing the physical-object inference was just making an ordinary
inference about a scenario, not making a decision about some view or about the
applicability of some particular correspondence rule.
Also, insofar as the different views in a parallel mix might each involve
more than one correspondence rule, there is no question of the hearer trying to
bundle together these separate packages of rules as wholes, somehow creating
a new coherent combined package of rules. That would be a matter of selecting
correspondence rules as a result of first deciding what mix is involved. Rather,
in ATT-Meta it is the other way around as well as more implicit and more frag-
mentary. It is correspondence rules that are individually selected by the needs
of inference, and what views are thereby mixed together (in some theoretical
construal of the processing) is an implicit side-effect of what rules are chosen.
Moreover, it may well be that not all correspondence rules associated with a
given view are used.
Thirdly, what correspondence rules do is create specific correspondences, by
which I mean correspondences that apply to specific entities, not all entities of
a class. For example, rule (11), when it fires, creates a correspondence between
physical manipulation of a specific idea (such as the idea of Anne’s husband being
unfaithful) and a specific person’s mental usage of that idea. Nothing is said about
other ideas or people. These could in principle be subject to different metaphori-
cal views, or none. This specificity and diversity can be important in practice –
­witness (29), where some ideas are metaphorically viewed as internal utterances
and some as external utterances, and some ideas are viewed as jumping creatures
while o
­ thers are not.
The specificity is also key in analysing the nature of the following example:
(31) “My husband stands beside and behind me.”26

This casts human relationships, such as support between people and people/world
relationships, in terms of physical space. It could be analysed as a case of par-
allel mixing. One sort of personal relationship is regarded as physically-behind,
and another is regarded as physically-beside. “Behind” and “beside” both convey
a support relationship, but the types of support are subtly different. “Beside” has
more a quality of transparently working together on some task, whereas “behind”
has more a quality of giving background help such as moral, financial or electoral
support. So, we have parallel mixing at the level of quite specific mappings, where

.  Variant of an encountered example, but source not recorded.

 Chapter 5.  Mixed metaphor 

these mappings might have been (naively) thought to be just different aspects of
one metaphorical view of personal relationships as physical relationships. Now,
because in (31) the mappings both happen to be applied to the very same pair of
entities (the woman and her husband) we get a clash if we use a single pretence
space (the husband is somehow both physically beside and behind the woman),
making the mixing somewhat infelicitous and comical. But there is no general
clash between the mappings as such. There would be no sense of clash in saying
“My husband is beside me and my mother is behind me.” We still have parallel
mixing, but it is now felicitous.
Fourthly, it is not even the case that a correspondence rule has to be intui-
tively regarded as part of just one metaphorical view. The conditions in a guard
could intuitively reflect more than one view. A guard might for instance check
that one idea is being viewed as an animal (i.e., it is an animal in the pretence)
and a check that some other idea is being viewed as a cage, to cater for special
effects of cases where one idea is restrained by another idea. So both a view of an
idea as an animate object and a view of an[other] idea as an inanimate object are
mixed together. This feature of ATT-Meta is useful for handling familiar mixes
of particular metaphorical views. In particular, it can allow for the mix to involve
additional mappings not inherited from either view. Guards in ancillary assump-
tions can have similar benefits.
Fifthly, a reason for VNMAs being generally useful for mixing is as follows.
The more that the weight of metaphor understanding is on VNMAs and on the
type of information they deal with (emotions, abilities, causation, time, etc.) than
on mappings associated with particular metaphorical views, the easier it is (i) to
bypass any apparent clashes of subject matter between different view-specific map-
pings, in parallel cases, and (ii) to derive coherent and useful effects from a mixing
situation, in both parallel and serial cases. For a simple example of this, consider
the parallel mixing of ideas as internal utterances and ideas as external
utterances in (29). One voice seems to be inside her brain, another outside. But
the voices’ utterances are both very negative, leading via a VNMA to the conclu-
sion that Sharon is (in reality) having very negative thoughts. The inside/outside
distinction does not matter to this point.
In serial cases of mixing, notice that the types of information delivered by
VNMAs into an outer pretence from an inner pretence – information about
affect, causation, time, etc. – are of course the very types that can be acted upon
by VNMAs operating from the outer pretence into its surrounding space. This
is exemplified by the Marigold mixed-metaphor Example (24), where fear is
transmitted from inner pretence to outer pretence and thence in turn into real-
ity. Indeed, these points are particularly powerful for (24), as our analysis relied
very little on view-specific mappings. The only view-specific mapping needed
 John Barnden

was the correspondence between Marigold’s conscious-self-person and Marigold

­herself. In other respects the edifice of mapping rests simply on identity mappings
and VNMAs.
This possibility of view-specific mappings not being involved very much, or
at all, incidentally helps ATT-Meta deal with certain types of novel metaphor. For
instance, the thought-as-cloud view can be entirely novel for ATT-Meta as long as
adequate relevant information can be transferred by VNMAs.

5.4  Further discussion: Variability of analysis

We have already noted the question of how many pretences to use in a case of par-
allel mixing, and the point that whether a metaphorical utterance involves parallel
mixing or not in a one-pretence case is a matter of theoretical judgment, because
there is no objective criterion as to when more than one source subject matter is
involved.
But in addition, it is often the case that it is not clear whether a serial or a par-
allel analysis is appropriate. For instance, consider (2), repeated here:
(32) “We do not have a chocolate army [that] fades away at the first sign of
trouble.”

One analysis that could be suggested is that there is parallel mixing of (i) a meta-
phorical view of soldiers in the [British] army as chocolate soldiers and (ii) a meta-
phorical view of people leaving a physical or abstract situation as fading. The view
in (ii) is a familiar one – for example one can talk about a crowd of people “fading
away,” as if the crowd were an image or the outer colouring of something. Simi-
larly, in (i), assuming that the key point about the chocolate is its propensity to
melt, we have an implicit use of the metaphorical view that is involved in saying
“the crowd melted way.”
However, this parallel analysis disconnects the fading from the properties of
chocolate. It takes the sentence as making the same point about the army twice, in
effect: once via the implication of melting (from the chocolate), and again sepa-
rately from the explicit “fades.” Another possibility, which may be more plausible,
is that to take the fading to be connected to the chocolateness: there is serial mix-
ing of (i) the metaphorical view of the soldiers as chocolate ones, and (ii) a meta-
phorical view of chocolate’s melting as fading.
There may not be a single correct analysis. It may be more a matter of per-
sonal (unconscious) choice by the hearer. Or there may be advantages to one
or other possibility. An advantage of the serial analysis is that it explains why
the speaker would bother to mention both chocolate and fading. Also, a danger
 Chapter 5.  Mixed metaphor 

with a parallel analysis is that one might try to find an aspect of chocolate that
provides different information about the army than that provided by the fad-
ing, given the very fact that parallel mixing is often used to talk about different
aspects of the target.

5.5  Conclusions

The article has sketched how a particular approach to metaphor, the ATT-Meta
approach, can handle various aspects of mixed metaphor, and how various
advantages in this handling drop out of its handling of metaphor in general. The
scepticism about domain divisions, and the related focusing on individual map-
pings as opposed to whole metaphorical views, allows a finer grain of analysis
and computational processing for mixing purposes. The use of guards in map-
ping specifications (correspondence rules) allows great flexibility in the use of
mappings and particularly in the mixing of views, and allows special features of
familiar mixes of specific views to be handled efficiently. The use of view-neutral
mapping adjuncts not only provides great power in the handling of metaphor
in general but also, by downplaying (in some cases to zero) the significance of
view-specific mappings, it additionally facilitates the interactions inherent in
mixed metaphor.

Acknowledgments

The research in this article was supported in part by Research Project Grant F/00 094/BE from
the Leverhulme Trust in the UK. It derived from work supported by grant EP/C538943/1 from
the Engineering and Physical Sciences Research Council in the UK.

References

Agerri, R., Barnden, J. A., Lee, M. G., & Wallington, A. M. (2007). Default inferences in met-
aphor interpretation. In B. Kokinov, D. C. Richardson, T. R. Roth-Berghofer, & L. Vieu
(Eds.), Modelling and using context: 6th international and interdisciplinary conference
(CONTEXT 2007). Lecture Notes in Artificial Intelligence, (Vol. 4635, pp. 1–14). Springer.
Barnden, J. A. (2001a). Uncertainty and conflict handling in the ATT-Meta context-based sys-
tem for metaphorical reasoning. In V. Akman, P. Bouquet, R. Thomason, & R. A. Young
(Eds.), Modeling and using context: Third international and interdisciplinary conference
(CONTEXT 2001). Lecture Notes in Artificial Intelligence, (Vol. 2116, pp. 15–29). Berlin:
Springer.
 John Barnden

Barnden, J. A. (2001b). Application of the ATT-Meta metaphor-understanding approach to

selected examples from Goatly. Technical Report CSRP–01–01, School of Computer Sci-
ence, The University of Birmingham, UK.
Barnden, J. A. (2006a). Artificial intelligence, figurative language and cognitive linguistics. In
G. Kristiansen, M. Achard, R. Dirven, & F. J. Ruiz de Mendoza Ibáñez (Eds.), Cognitive linguis-
tics: Current applications and future perspectives (pp. 431–459). Berlin: Mouton de Gruyter.
Barnden, J. A. (2006b). Consequences for language learning of an AI approach to metaphor.
In J. Salazar, M. Amengual, & M. Juan (Eds.), Usos Sociales del Lenguaje y Aspectos Psico-
lingüísticos: Perspectivas Aplicadas (pp. 15–57). Palma, Mallorca: Universitat de les Illes
Baleares.
Barnden, J. A. (2008). Metaphor and artificial intelligence: Why they matter to each other. In
R. W. Gibbs, Jr. (Ed.), The Cambridge handbook of metaphor and thought (pp. 311–338).
­Cambridge, UK: Cambridge University Press.  doi: 10.1017/CBO9780511816802.020
Barnden, J. A. (2009). Metaphor and context: A perspective from artificial intelligence. In
A.  Musolff & J. Zinken (Eds.), Metaphor and discourse (pp. 79–94). Basingstoke, UK:
­Palgrave Macmillan.
Barnden, J. A. (2010). Metaphor and metonymy: Making their connections more slippery. Cog-
nitive Linguistics, 21(1), 1–34.  doi: 10.1515/cogl.2010.001
Barnden, J. A. (2015). Open-ended elaborations in creative metaphor. In T. R. Besold,
M. ­Schorlemmer, & A. Smaill (Eds.), Computational Creativity Research: Towards Creative
Machines (pp. 217–242). Atlantis Press (Springer).
Barnden, J. A. (n.d.). ATT-Meta Project Databank: Examples of usage of metaphors of mind.
http://www.cs.bham.ac.uk/ jab/ATT-Meta/Databank/
Boroditsky, L. (2000). Metaphoric structuring: understanding time through spatial metaphors.
Cognition, 75, 1–28.  doi: 10.1016/S0010-0277(99)00073–6
Bowdle, B. F., & Gentner, D. (2005). The career of metaphor. Psychological Review, 112(1),
193–216.  doi: 10.1037/0033-295X.112.1.193
Carbonell, J. G. (1982). Metaphor: An inescapable phenomenon in natural-language com-
prehension. In W. Lehnert & M. Ringle (Eds.), Strategies for natural language processing
(pp. 415–434). Hillsdale, NJ: Lawrence Erlbaum.
Carston, R., & Wearing, C. (2011). Metaphor, hyperbole and simile: A pragmatic approach.
Language and Cognition, 3(2), 283–312.  doi: 10.1515/langcog.2011.010
Dirven, R. (2002). Metonymy and metaphor: Different mental strategies of conceptualization.
In R. Dirven & R. Pörings (Eds), Metaphor and metonymy in comparison and contrast
­(pp. 75–111). Berlin, New York: Mouton de Gruyter.
Fauconnier, G., & Turner, M. (2008). Rethinking metaphor. In R. W. Gibbs, Jr. (Ed.), The
­Cambridge handbook of metaphor and thought (pp. 53–66). Cambridge, UK: Cambridge
University Press.  doi: 10.1017/CBO9780511816802.005
Feldman, J. (2010). Embodied language, best-fit analysis, and formal compositionality. Physics of
Life Reviews, 7, 385–410.  doi: 10.1016/j.plrev.2010.06.006
Gargett, A., & Barnden, J. (2015). Gen-Meta: Generating metaphors by combining AI and
corpus-based modeling. Web Intelligence 13(2), 103–114.
Gibbs, R. W., Jr., & Matlock, T. (2008). Metaphor, imagination, and simulation: Psycholinguis-
tic evidence. In R. W. Gibbs, Jr. (Ed.), The Cambridge handbook of metaphor and thought,
(pp. 161–176). Cambridge, UK: Cambridge University Press.
doi: 10.1017/CBO9780511816802.011
Goatly, A. (1997). The language of metaphors. London and New York: Routledge.
doi: 10.4324/9780203210000
 Chapter 5.  Mixed metaphor 

Goatly, A. (2002). Conflicting metaphors in the Hong Kong Special Administrative Region edu-
cational reform proposals. Metaphor and Symbol, 17(4), 263–294.
doi: 10.1207/S15327868MS1704_2
Goossens, L. (1990). Metaphtonymy: the interaction of metaphor and metonymy in expressions
for linguistic action. Cognitive Linguistics, 1, 323–340.  doi: 10.1515/cogl.1990.1.3.323
Grady, J. E. (1997). THEORIES ARE BUILDINGS revisited. Cognitive Linguistics, 8(4),
267–290.  doi: 10.1515/cogl.1997.8.4.267
Hobbs, J. R. (1992). Metaphor and abduction. In A. Ortony, J. Slack, & O. Stock (Eds.), Commu-
nication from an artificial intelligence perspective: Theoretical and applied issues (pp. 35–58).
Berlin: Springer-Verlag.  doi: 10.1007/978-3-642-58146-5_3
Lakoff, G., & Johnson, M. (1980). Metaphors we live by. Chicago: University of Chicago Press.
Lee, M. (2010). Truth, metaphor and counterfactual meaning. In A. Burkhardt & B. Nerlich
(Eds.), Tropical truth(s): The epistemology of metaphor and other tropes (pp. 123–135).
­Berlin/New York: De Gruyter.
Lee, M. G., & Barnden, J. A. (2001). Reasoning about mixed metaphors with an implemented AI
system. Metaphor and Symbol, 16(1&2), 29–42.  doi: 10.1207/S15327868MS1601&2_3
Levin, S. R. (1988). Metaphoric worlds. New Haven, CT and London, UK: Yale University
Press.
Musolff, A. (2004). Metaphor and political discourse: Analogical reasoning in debates about
Europe. Basingstoke, UK: Palgrave Macmillan.  doi: 10.1057/9780230504516
Narayanan, S. (1999). Moving right along: A computational model of metaphoric reasoning
about events. Proceedings of the National Conference on Artificial Intelligence (AAAI ‘99)
(pp. 121–128). AAAI Press.
Radden, G. (2002). How metonymic are metaphors? In R. Dirven & R. Pörings (Eds.), Metaphor
and metonymy in comparison and contrast (pp. 407–434). Berlin, New York: Mouton de
Gruyter.
Rubio Fernández, P. (2007). Suppression in metaphor interpretation: Differences between
meaning selection and meaning construction. Journal of Semantics, 24, 345–371.
doi: 10.1093/jos/ffm006
Ruiz de Mendoza Ibáñez, F. J., & Díez Velasco, O. I. (2002). Patterns of conceptual interaction.
In R. Dirven & R. Pörings (Eds.), Metaphor and metonymy in comparison and contrast
(pp. 489–532). Berlin, NY: Mouton de Gruyter.
Smolka, E., Rabanus, S., & Rösler, F. (2007). Processing verbs in German idioms: Evidence
against the configuration hypothesis. Metaphor and Symbol, 22(3), 213–231.
doi: 10.1080/10926480701357638
Sperber, D., & Wilson, D. (1995). Relevance: Communication and cognition. 2nd ed. Oxford:
Blackwell.
Stern, J. (2000). Metaphor in context. Cambridge, MA: Bradford Books, MIT Press.
Turner, M., & Fauconnier, G. (1995). Conceptual integration and formal expression. Metaphor
and Symbolic Activity, 10(3), 183–204.  doi: 10.1207/s15327868ms1003_3
Walton, K. (2004). Metaphor and prop oriented make-believe. In E. John & D. M. Lopes (Eds.),
Philosophy of literature – Contemporary and classic readings: An anthology (pp. 239–247).
Oxford: Blackwell.
White, R. M. (1996). The structure of metaphor: The way the language of metaphor works. Oxford,
UK: Blackwell.
 chapter 6

Mixed metaphor is a question of deliberateness

Gerard Steen
University of Amsterdam

This paper aims to explore the interaction between mixed metaphor and
deliberateness in order to throw some new light on the nature of mixed metaphor.
The basic claim is that the typical or strongest experience of mixed metaphor
arises when two metaphors conflict that are both used deliberately as metaphors.
It is likely that all other cases of conceptual clashes between adjacent metaphors
do not get recognized as mixed metaphor because their components are not
used deliberately as metaphors. Whether the clash between one deliberate and
one non-deliberate metaphor can elicit the experience of mixed metaphor is an
in-between case that is also discussed. The gist of the paper is, then, that research
on mixed metaphor needs to take into account the variable communicative status
of each of the presumably clashing metaphors, making a distinction between their
deliberate or non-deliberate use as metaphors.

6.1  Introduction

With the cognitive turn in metaphor studies (Gibbs, 2008), a wealth of research
has been produced that shows that metaphor is more than just a figure of speech
used as a rhetorical flourish in the language of poets, politicians and other kinds
of wordsmiths. Metaphor has been reconceptualized from the stylistic device it
used to be for over two millennia to a conceptual tool that helps us make sense
of everyday experience, in particular when everyday experience requires the use
of abstract categories that have to do with complex phenomena like organization
and management, government and politics, health and care, and so on. Research
has shown that our regular use of language is full of metaphor, examples includ-
ing talk about organizations as plants that can grow and be pruned, energy as a
liquid that can flow or stream, arguments as fights that can be won or lost, and
theories as buildings that can be strong or weak and need support or buttressing
(Lakoff  and Johnson, 1980). This ubiquity of metaphor in language is taken by
many as a reflection of underlying figurative conceptualizations that are pervasive

doi 10.1075/milcc.6.06ste
© 2016 John Benjamins Publishing Company
 Gerard Steen

and systematic, called ‘conceptual metaphors’ in cognitive linguistics. All of these

linguistic forms and conceptual structures are presumably used without much
awareness on the part of the language user, whether in production or reception,
and this would explain why we do not notice the use of metaphor very often. The
use of metaphorical meanings is just as automatic as the use of grammatical struc-
tures like subject and predicate.
Yet people can and do notice metaphor in special circumstances, one of
which is when one metaphor is inappropriately mixed with another metaphor.
The example in the Merriam Webster dictionary is ‘If we want to get ahead we
have to iron out the remaining bottlenecks.’ This is a special case of when meta-
phorical language use simply goes wrong, other cases involving, for instance, a
metaphorical comparison getting much too difficult or vague to be understood.
Metaphor consequently does get noticed sometimes and is then attended to as
metaphor by language users. This is again to be compared with other moments
when language use derails and draws attention to itself as language use, as when it
becomes a noticeable problem that an utterance lacks a subject or a predicate, or
that grammatical concord between subject and predicate is incorrect. Along these
lines, mixed metaphor is typically seen as a case of flawed and careless language
use on the part of the producer who is then accused of not paying sufficient atten-
tion to the potential effects of their utterance on the receiver. Mixed metaphor is
hence also commonly seen as something to be avoided.
This poses an intriguing problem. If metaphor use is automatic and uncon-
scious, how can so many language and communication advisers spend so much
time telling writers and speakers to avoid mixing metaphors? Such advice pre-
supposes that it is possible to monitor one’s metaphor use, which is an activity
that is anything but automatic and unconscious. Mixed metaphor may therefore
involve interesting questions about deliberate metaphor use, itself a controversial
issue in the theory of metaphor (Gibbs, 2011; Steen, 2008, 2011a, 2013, in press).
It is my aim in this paper to explore the interaction between mixed metaphor and
deliberateness in order to throw some new light on the nature of mixed metaphor.
My basic claim will be that the experience of mixed metaphor arises when two
metaphors conflict that are used deliberately as metaphors and that it is likely that
all other cases of conceptual clashes between adjacent metaphors do not get rec-
ognized as mixed metaphor because they are not used deliberately as metaphors.

6.2  Mixed metaphor and deliberateness

The internet abounds with sites presenting hilarious examples of mixed m

­ etaphor
that have caught people’s attention. On the website www.about.com, ­Richard
 Chapter 6.  Mixed metaphor is a question of deliberateness 

­ ordquist features the following instance produced by a British Member of

N
­Parliament: “Mr. Speaker, I smell a rat. I see him floating in the air. But mark me,
sir, I will nip him in the bud.” There is a separate page on www.JimCarlton.com
­listing his favorite mixed metaphors including “It’s time to step up to the plate
and lay your cards on the table.” Mignon Fogarty has an entry on http://­grammar.
quickanddirtytips.com/mixed-metaphors.aspx discussing a phrase by Obama, that
people thought that Obama was ‘green behind the ears’, which struck her as a
mixed metaphor. And we could go on.
What is interesting about these and many other examples is that they may
be explained by a specific special feature: they often involve the use of an idiom-
atic phrase that is motivated by metaphor, which is either not completed in the
expected way or not followed up in a consistent manner but by another, conflict-
ing idiom. Thus, ‘green behind the ears’ involves a mix-up of ‘wet behind the ears’
and ‘green’, and the replacement of ‘wet’ by ‘green’ is noticeable because it gener-
ates an incorrect idiomatic construction that is then correctly recognized as mixed
metaphor. The other two examples both involve the use of one figurative idiom fol-
lowed by another that is so different that it breaks the expected sense of cohesion
between two consecutive parts of discourse, two figurative constructions combin-
ing in such a way that their non-figurative meanings begin to stand out and clash.
These cases are therefore not just a matter of mixing metaphors but of mixing
idioms that happen to be metaphorical.
Mixed metaphors that involve one or more metaphorical idioms, like the ones
above, seem to be specially prominent cases of all mixed metaphor, which con-
cerns a slightly more subtle matter. Generally, mixed metaphor simply depends on
a noticeable clash between the non-metaphorical meanings of two metaphorical
uses of words within one relevant grammatical or discursive frame. Two good
illustrations of this phenomenon are the following:
(1) … – the economic cake grew fast enough in these years for most demands
to be accommodated without conflict (Judt, 2005: 266)
(2) The Italian welfare state in the 1960s was still a rather rough-and-ready
edifice that would not reach maturity until the following decade …
(Judt, 2005: 413)

Both have been taken from Tony Judt’s magnificent work on the history of
Europe after the Second World War, which I will use as my data source for
this chapter. Intuitively, Judt’s writing looks representative of its kind, history,
which is close enough to most storytelling and reflection on events in life to
be more generally interesting. With its 800-plus pages, it offers a wide variety
of metaphorical ­language use, including some possibly mixed metaphors like
(1) and (2).
 Gerard Steen

The reasons these two cases can be seen as mixed should be obvious. Cakes do
not grow but at most become bigger, which does help to explain the slip into ‘grow’;
and edifices do not reach maturity, but at most get completed, which may be seen
as close to reaching maturity. In both cases, the non-metaphorical meanings of
the second metaphorical images (‘grew’ and ‘reach maturity’) clash with the non-
metaphorical meanings of the first metaphorical images (‘cake’ and ­‘edifice’). The
crucial question about mixed metaphor is why these non-metaphorical ­meanings
become prominent whereas in most other cases of adjacent metaphor, the non-
metaphorical, basic meanings of metaphorically used words do not draw people’s
attention, making the ubiquitous use of metaphor so unobtrusive.
I will explore this phenomenon from a specific theoretical perspective that
taps into the potential prominence of metaphor. I have argued that it is true that
most metaphor is unconscious and automatic, but that we need another distinc-
tion to get a handle on some cases of metaphor use that are special (Steen, 2008).
This distinction has to do with the deliberate versus non-deliberate use of meta-
phors as metaphors: most metaphor use is non-deliberate, but some metaphor use
is deliberate. I would contend that the use of many metaphorical idioms qualifies
as potentially deliberate: when a Member of Parliament says ‘Mr Speaker, I smell
a rat’, this appears to involve the selection of a type of expression deliberately
introducing a different image of a situation. Deliberate metaphors are intention-
ally chosen perspective changers, invoking distinct conceptual domains as alien
referents in the on-going discourse. (This is a notion of deliberateness that does
not necessarily involve consciousness, but is based on the general goal directed-
ness of language use, which may also be and in fact typically is unconscious; see
the discussion between Gibbs 2011 and Steen 2011a). Because of this deliberate
metaphor use, the metaphorical, alien image may be represented as a distinct ref-
erent in people’s minds and be available long enough to produce a potential clash
with the next metaphorical image if it is nearby enough and incompatible enough
to be noted as invoking yet another distinct conceptual domain. If it is agreed that
metaphorical idioms have such a potentially deliberate metaphorical quality, this
might explain why mixed metaphors involving metaphorical idioms like the ones
at the beginning of this section are so prominent.
Is it possible that ‘the economic cake’ can also be qualified as deliberately
metaphorical, and can this explain the sense of clash with the next metaphori-
cally used word ‘grew’? And is it possible that ‘The Italian welfare state was a
rough-and-ready edifice’ is also deliberately metaphorical, again explaining the
sense of clash with ‘that would not reach maturity’? I believe that a good case
can be made for an affirmative answer to these questions. Thus, the use of the
domain indicator ‘economic’ suggests that the writer intentionally instructs the
reader to give an economic interpretation to the next word, ‘cake’, which by itself
 Chapter 6.  Mixed metaphor is a question of deliberateness 

sets up a culinary referent. This is a form of metaphor signaling which would

make ‘the economic cake’ a deliberate metaphor. As a deliberate metaphor, ‘cake’
can then be felt to clash with the next word that is metaphorical, ‘grow’. Whether
this is true even if that next metaphor related word is not used deliberately itself,
as is probably the case here, is another question, which we will address later. The
relation between mixed metaphor and deliberate metaphor use clearly merits
further analysis.
In fact, jumping ahead to the suggestions I will be making later, mixed meta-
phor and deliberate metaphor may be seen as two unexpected allies in research on
the activation of metaphor as metaphor in people’s attention and consciousness
(Steen, 2011b, 2014). If mixed metaphor is characterized by its ability to impinge
on people’s attention and consciousness, it is possible that this is largely based on
its relation to deliberate metaphor. The mixing of two deliberate metaphors could
then be noticed more often as mixed metaphor than the mixing of one deliber-
ate and one non-deliberate metaphor, which in turn should be more prominent
than the mixing of two non-deliberate metaphors. These are hypotheses that can
be tested experimentally, which is on our program for research. Vice versa, since
deliberate metaphor use is still a controversial affair, demonstrating that it has dif-
ferentiated effects in mixed metaphor would help in establishing its own nature
and function as well.
My plan for the rest of this chapter is as follows. I will first illustrate the dis-
tinction between deliberate versus non-deliberate metaphor use. Then I will take
a close look at some mixed metaphors to see whether and how they interact with
deliberate metaphor use. This will lead to new questions about mixed metaphor
and deliberate metaphor that can be put on the agenda for future research.

6.3  Deliberate versus non-deliberate metaphor

My basic argument is the following. As a rule, metaphors are not produced or

noticed as metaphors by language users. People do not typically go around think-
ing they need a metaphor to construct their next utterance, or recognizing one
metaphor in other people’s use after another. At the same time, metaphors abound
in our language use, which also means that they are often adjacent to each other
and consequently may cause occasional conceptual conflicts between them: not
every metaphor is consistent with or compatible with the next metaphor. How
often such conflicts occur in the structure of language is hard to say, but with
an estimated average of 13.6% metaphor-related words in natural discourse
(Steen et al., 2010), one in every 7 to 8 words is used metaphorically, which should
give some concrete idea.
 Gerard Steen

The established ubiquity of metaphor in language sharply contrasts with the

low frequency of spontaneous metaphor recognition, including mixed metaphor
recognition. There may hence be more conflicting metaphors in the structure of
language than experiences of mixed metaphor in language users’ processing. The
question therefore arises when, or which, adjacent and conflicting metaphors are
noted and experienced as mixed.
The answer to this question may have much to do with the distinction
between deliberate versus non-deliberate metaphor use: I will argue that at least
one deliberate metaphor may be needed for mixed metaphor to be recognized.
If a metaphor is not used deliberately as a metaphor, as generally is the case, it is
immaterial whether that metaphor conflicts with an adjacent metaphor. However,
if a metaphor is used deliberately as a metaphor, which happens in more specific
conditions, it does become relevant if the next metaphor is consistent or not. Per-
haps this alerts language users to non-deliberate metaphors, but it will most likely
trigger mixed metaphor recognition for two deliberately used metaphors in a row
that are also conflicting with each other. Deliberate metaphor may hence offer
the key to understanding why some adjacent metaphors are experienced as mixed
while most adjacent metaphors are not (cf. Kimmel, 2010).
Consider the following random paragraph from Tony Judt’s Postwar (p. 298),
with separate sentences numbered for convenient reference:
(3) (1) The first lesson of Suez was that Britain could no longer maintain
a global colonial presence. (2) The country lacked the military and
economic resources, as Suez had only too plainly shown, and in the wake
of so palpable a demonstration of British limitations the country was
likely now to be facing increased demands for independence. (3) After
a pause of nearly a decade, during which only the Sudan (in 1956) and
Malaya (in 1957) had severed their ties with Britain, the country thus
entered upon an accelerated phase of de-colonization, in Africa above all.
(4) The Gold Coast was granted its freedom in 1957 as the independent
state of Ghana, the first of many. (5) Between 1960 and 1964, seventeen
more British colonies held ceremonies of independence as British
dignitaries traveled the world, hauling down the Union Jack and setting up
new governments. (6) The Commonwealth, which had just eight members
in 1950, would have twenty-one by 1965, with more to come.

This is a regular stretch of historical writing, with no metaphors calling atten-

tion to themselves as metaphors, apart from one. There are plenty of metaphors to
find, all of them highly conventionalized, as a slightly more technical glance will
reveal (Pragglejaz Group, 2007; Steen et al., 2010), but they all do not seem to be
used deliberately as metaphors. Sentence (1) opens with the metaphorical use of
lesson, which is described in Macmillan sense 2 as ‘something you learn from life,
 Chapter 6.  Mixed metaphor is a question of deliberateness 

an event, or an experience’. Sentence (2) features in the wake of, a set phrase that
has been given its own sense description ‘happening after an event or as a result
of it’, deriving from wake, ‘the track that appears in the water behind a moving
boat’. In the same sentence, a case can be made for palpable and facing, both of
which have to do with embodied experiences in their basic senses whereas they
are here applied to the more abstract, complex concepts of ‘demonstration’ and
political negotiation. Sentence (3) contains clear examples like severed their ties
with, entered upon, and accelerated, all of which are also conventionalized in their
metaphorical use. This type of metaphorical conventionalization even extends to
the use of in in sentence (4) and between in sentence (5), where time is talked
about in terms of space. All of these are regular, non-deliberate metaphors in that
their use does not call attention to the metaphorical status of the word(s). What
is more, it is quite possible that all of these words are processed by lexical disam-
biguation, not on-line cross-domain mapping in conceptual structure, whether by
construction or retrieval (cf. Steen, 1994: 15–22; 2008).
Sentence (5) features an interesting variant of zeugma in its final part, ­British
dignitaries ‘hauling down the Union Jack and setting up new governments’: setting
up is used in contrast to hauling down, but setting up is not used in its concrete
motion sense, whereas hauling down is. The use of ‘setting up’ may have been the
only metaphor in the entire stretch that has caught the reader’s attention as meta-
phorical. I would contend that this is precisely because it has been used deliber-
ately as a metaphor. The reason why I think it is deliberate becomes clear once it is
acknowledged that the style is intentionally humorous here: the humor is based in
the contrast between the two motion words in their nonmetaphorical and meta-
phorical uses that is not accidental or fortuitous but intended for ironic effect. This
entails that ‘setting up new governments’ is a metaphor that is deliberately used as
a metaphor.
This example shows that conventional metaphor can be used both non-­
deliberately and deliberately. All of the previous examples in excerpt (3) are con-
ventional whereas none of them is deliberate, somehow drawing attention to itself
as being metaphorical, but ‘setting up new governments’ involves a convention-
ally metaphorical use of the verb to set up which is here deliberately used as a
metaphor to wittily contrast with the non-metaphorical hauling down. Deliberate
versus non-deliberate metaphor use involves an independent dimension of meta-
phor use, and it is to be crossed with the one distinguishing between conventional
versus novel conceptual structure.
I have labeled the dimension accommodating deliberateness the ‘communi-
cative’ dimension of metaphor: it deals with the communicative status of meta-
phor as a metaphor (or not). The dimension ordering conventional versus novel
conceptual structure is the scale familiar from cognitive linguistics dealing with
 Gerard Steen

­ etaphor in thought. What remains is the third dimension of metaphor in dis-

m
course, i.e. the one that distinguishes between different linguistic forms of meta-
phor, such as metaphor versus simile, something we will return to later on. We
need a three-dimensional model including metaphor in language, thought, and
communication to account for the role of metaphor in discourse (Steen, 2008).
The point here is that the communicative dimension of metaphor, including the
contrast between deliberate versus non-deliberate metaphor use, has been ignored
for a long time in contemporary metaphor theory and research. Its explanatory
potential has been underestimated and is here argued to extend to the phenom-
enon of mixed metaphor as well.
As will be appreciated, deliberate metaphor use can depend on very local sty-
listic considerations. In the present case, the prior phrasal context ‘hauling down
the Union Jack’ creates the first half of what turns out to be a semi-zeugmatic
construction that is completed by the metaphor ‘setting up new governments’ that
may therefore be seen as deliberate – the utterance and its humorous effect do not
come about accidentally. The experience of deliberate metaphor does not require
its full-blown presentation or recognition as a metaphor: the only thing that is
needed is that it should lead to some, however fleeting, moment of difference that
is dependent on the distinct representation of the source domain item, signaled
here by the humor that is caused by the non-fortuitous contrast between the non-
metaphorical hauling down of flags versus the metaphorical setting up of new
independent governments.
Having now set the stage for the difference between non-deliberate and delib-
erate metaphor, here are some more conspicuous examples of deliberate metaphor
from the same book. The most convincing and extreme case of deliberate meta-
phor is extended metaphor:
(4) West Germany had navigated safely between the Scylla of neo-Nazism and
the Charybdis of philo-Soviet neutralism, and was anchored securely within
the Western alliance, despite the misgivings of critics at home and abroad.
(p. 265)
(5) Moscow was the flattering mirror of their political illusions. In November
1956, the mirror shattered. (p. 322)

Extended metaphor involves the deliberate use of metaphor across independent

clauses or even sentences. Tony Judt does not use many of these, the first half of
his book only including the above two examples if I am not mistaken. What is
striking is that another six can be found in quotations from other speakers, offer-
ing somewhat more spectacular metaphorical comparisons across sentences that
are clearly meant to enliven the text. This almost suggests that extended metaphor
is too much of a good thing for decent historical writing. By contrast, it clearly is
 Chapter 6.  Mixed metaphor is a question of deliberateness 

not in media discourse, as is shown by the many examples of extended metaphor

examined for the presence of mixed metaphor by Kimmel (2010); he concludes
that mixed metaphor does not occur much between sentences. In the next section
we will consider one or two potential cases from Judt.
Deliberate metaphor that is not extended (and not mixed but ‘pure’) can be
found more often in Judt’s writing. One example is (6):
(6) The second Stalinist ice age was beginning. (p. 145)

All of the words in this sentence except domain indicator Stalinist come from
an alien semantic field: they ostensibly do not talk about Soviet history but
about geology. This is no accidental but deliberate metaphor use. That they are
supposed to be interpreted metaphorically is explicitly signaled by Stalinist,
another symptom that the producer is deliberately using the metaphor-related
words as metaphor, that is, as an expression involving a mapping from the
source domain of ice age to the target domain of Stalinist repression. It is not
just the signal, however, that makes this metaphor deliberate: in the broader
context, there is no doubt that the author deliberately intends to set up a func-
tional contrast between two domains of meaning that is to be mentally attended
to as such:
(7) It is significant that the attacks on Tito and his followers coincided with
the full flowering of the Stalinist personality cult and the purges and show
trials of the coming years. For there is little doubt that Stalin truly did see
in Tito a threat and a challenge, and feared his corrosive effect on the fealty
and obedience of other Communist regimes and parties. The Cominform’s
resistance, in its journals and publications, on the ‘aggravation of the class
struggle in the transition from capitalism to socialism’ and on the ‘leading
role’ of the Party risked reminding people that these had been precisely
the policies of the Yugoslav Party since 1945. Hence the accompanying
emphasis on loyalty to the Soviet Union and Stalin, the rejection of
all ‘national’ or ‘particular’ roads to Socialism and the demand for a
‘redoubling of vigilance’. The second Stalinist ice age was beginning. (p. 145)

The paragraph could also have ended ‘The second ice age was beginning’, with
no damage to referential comprehensibility or communicative effect (even though
there would have been a loss in referential specificity). This clearly is a deliberately
metaphorical utterance that does require the reader to heed to the fact that it is
metaphorical by giving separate attention to the source domain concepts of the
beginning ice age as distinct referents in the meaning of the text. It is fundamen-
tally different to the examples of non-deliberate metaphor discussed under (3), or
to the non-deliberate metaphors in this very excerpt itself, like attacks or flower-
ing in the first sentence. To be fully explicit, attacks does not require the reader
 Gerard Steen

to ­contrast the argumentative and war senses of the word – on the contrary, that
would be distracting in a way that does not hold for the final sentence of (7).
There is yet another reason why this particular metaphor is deliberate: its nov-
elty. In the dictionary, the expression ice age is described in geological terms only,
‘a period of time thousands of years ago when large areas of the Earth were cov-
ered in ice’ (Macmillan). Tony Judt has clearly looked for an appropriate rhetorical
device to close the paragraph and devoted some attention to crafting it. Indeed, the
last sentence is followed by a blank line in the chapter, which makes its function
of closing device even more effective. In its constructed novelty, the final utter-
ance is therefore no accidental use of metaphor. This illustrates how deliberate
metaphor use comprises both conventional as well as novel metaphorical struc-
tures in thought. (Not all novel metaphor is deliberate by definition, though, as for
instance happens when children or mental patients use language in ways that are
innovative to the general language user but themselves do not intend to construct
novel cross-domain mappings that are presented as such.)
The discursive ploy in (6) is that the author has condensed a metaphorical
comparison between an understood referent in the text (amounting to something
like ‘the second Stalinist period of repression’) and its metaphorical image (‘the
second ice age’) into a singular referential expression (‘the second Stalinist ice
age’). A textually more explicit version of this type of one-on-one metaphorical
comparison can be seen in (8):

(8) But the British saw the ECSC as the thin edge of a continental wedge in
British affairs. (p. 159)

To say that the British see ‘the ECSC’ as ‘the thin edge of a continental wedge’
involves an explicit comparison that crosses two conceptual domains, and is there-
fore metaphorical. This is as deliberate a metaphor as one can forge. In the present
case, the metaphorical expression is fully conventional, as can be checked in the
dictionary: the thin edge (or end) of a wedge is ‘something that is not important by
itself but will have serious, usually bad, effects in the future’ (Macmillan). That it is
metaphorical can be argued because its components ‘thin’, ‘edge’, and ‘wedge’ each
still have their own original concrete sense, and because the expression as a whole
has been broken up by the insertion of ‘continental’. We will see another example
of this same expression in a similarly free form under mixed metaphor below. The
signal that one thing is seen in terms of something else marks the author’s inten-
tion that the reader attend to the fact that this is a metaphorical construction.
Even though it may be clear that a metaphor is deliberate, it is not always clear
whether it is to be seen as novel or conventional:

(9) Religion, especially the Catholic religion, basked in a brief Indian summer
of restored authority. (p. 227)
 Chapter 6.  Mixed metaphor is a question of deliberateness 

People can literally bask in the sun while relaxing and enjoying themselves,
and they can metaphorically enjoy other people’s attention and approval, as if
they are sitting in the sun, especially upon success; for the abstract notion of
religion to bask in something positive, however, is an innovative application
of a conventional metaphor. Moreover, people usually bask in the sun, but to
bask in an Indian summer is another novel extension, with a strong hyperbolic
element. This new application of a conventional expression or idea is precisely
what will draw language users’ attention, however briefly or superficially, to
the communicative status of this metaphor as metaphorical. It does not make
sense to claim that the metaphorical meaning of this utterance would have
been constructed non-deliberately – the structure of the language suggests that
the author intends readers to pick up on the special (ultimately metaphorical)
nature of this construction in one way or another, even if it would just involve
a smile.
Apart from the question of novel versus conventional conceptual structure,
there are other structural aspects to the deliberate nature of example (9), which
can also be found in the following deliberately metaphorical constructions:
(10) The state thus lubricated the wheels of commerce, politics and society in
numerous ways. (p. 362)

(11) …–it was Korea, not Schumann, that sent the West German industrial
machine into high gear. (p. 159)

What is shared between (9) through (11) is their personified subjects combined
with a metaphorical action that explicates a predicate and an object and/or adver-
bial phrase in terms of the metaphorical source domain. Abstract agents ‘religion’,
‘state’ and ‘Korea’ are portrayed as people respectively basking in a brief Indian
summer, lubricating wheels, or sending a machine into high gear. Given the pre-
ceding contexts, which are all about directly designated historical referents and
processes, these are notable deviations from the dominant semantics of the text,
drawing attention to themselves as deliberately metaphorical. In particular, the
most important intended referents in the state of affairs designated by the sentence
are all involved in metaphor: the agent of the action is an abstract entity, the action
itself is a metaphorically expressed action, and the affected or other semantic roles
are also expressed metaphorically. More than one of these intended referents in the
projected state of affairs is expressed by means of a concept belonging to another
domain than the dominant topical one; this makes the metaphorical status of the
utterances quite deliberate.
Very close to this pattern is the following set of examples:

(12) …; in the USSR it was the events of 1956 that tore the veil from the eyes of
hitherto committed Communists like the young Leonid Pliushch. (p. 322)
 Gerard Steen

(13) But ironic or not, the reburial of Rajk provided the spark that was to ignite
the Hungarian revolution. (p. 314)

In (12) the construction suggests that ‘the events of 1956’ act as a person tear-
ing the veil from the eyes of committed communists. This is a heavy metaphori-
cal expression whose main referents (designated by verb, ‘tore’, object, ‘veil’, and
adverbial adjunct, ‘from the eyes’) are indirectly expressed as coming from some
other domain than the one of Hungarian history. Intuitively the personification of
‘the events of 1956’ is different and harder to conceptualize than the ones we saw in
(9) and (10), with states acting as persons, and even in (8), with religion acting as a
person. One important reason may be the fact that events is plural, not easily map-
ping onto one person doing the action of tearing the veil from somebody’s eyes.
In (11), provide is metaphorical and has effects on the role of ‘the reburial of Rajk’
as a grammatical agent; however, this does not necessarily produce personifica-
tion, making ‘the reburial of Rajk’ human: provide is a verb that displays a range of
conventional subjects, from people through institutions to events. As a result, the
deliberate quality of the metaphor in this sentence does not rest on personification
and may be limited to the combination of spark and ignite – a typically journalistic
way of metaphorically expressing this type of event, according to the Macmillan
dictionary. Some deliberate metaphors belong to the clichés of specific registers,
but this does not make their use any less deliberate (on the contrary).
Slightly more controversial may be the view that the following cases are also
deliberate:
(14) On the basis of the terms agreed at Evian de Gaulle called a referendum
on Sunday July 1st and the French people voted overwhelmingly to free
themselves of the Algerian shackle. (p. 288)
(15) Within a year it was clear that Paris and Algiers were on a collision course.
(p. 288)

(16) De Gaulle understood economic stabilization and modernization largely as

weapons in the struggle to restore national glory. (p. 290)

Their complete conventionality and the lack of signaling in (14) and (15) makes
them less typical candidates for deliberate metaphor use. However, there are still
other factors that are important to note. In particular, the use of shackle as an
abstract concept is a mainly literary device, according to the Macmillan diction-
ary, the concrete sense being the more regular meaning of this word; this heightens
the contrast between the abstract content of all of sentence (14), on the one hand,
and its unexpected ending in a concrete ‘literary’ image that requires metaphori-
cal interpretation. The final position of the word in the sentence may increase
this effect. This argument might also be applied to the next two sentences, but
 Chapter 6.  Mixed metaphor is a question of deliberateness 

the assumption of some prominence or salience of the concrete meanings of the

metaphor related words ‘on a collision course’ and ‘weapons in a struggle’ over the
abstract meanings seems more questionable there. It is interesting to note, though,
that the position of on a collision course at the end of the sentence in (15) may
also increase its prominence in comparison with the sentence-internal position of
weapons in a struggle in (16). For (16), on the other hand, there is the presence of
the signal that de Gaulle understands X as Y, where Y clearly comes from another
domain if it is taken in its basic concrete sense of ‘arms’. It is the next word ‘strug-
gle’ instead of ‘war’ or ‘battle’ that reduces this bias again, however, so that this
example becomes somewhat dubious. These are variables of deliberate metaphor
use the structures, functions and effects of which are currently studied in our lab.
Deliberate metaphor involves the use of metaphor as a metaphor. From a
structural-functionalist point of view, this can only be observed if the metaphor
producer has left traces of this intention in the language. This most clearly happens
when metaphors are extended across sentences or when they are accompanied by
metaphor signals. Other symptoms include salient deviations from the register of
a text, as when a literary meaning is inserted in a historical text, when a concrete
image is placed at the end of a sentence that is all abstract, or when a figurative
idiom comprising a number of source-domain referents is located in a text that is
otherwise non-figurative and colloquial at all. This is when metaphor becomes devi-
ant instead of regular, drawing attention to itself as a stylistic or rhetorical means to
change the reader’s perspective from inside the target domain to one positioned in
some other, alien source domain. It is this experience of deliberate metaphor which
I hold to be a precondition for the spontaneous recognition of mixed metaphor.

6.4  From deliberate to mixed metaphor

Let us now turn to some clear cases of mixed metaphor and examine whether they
can be analyzed as involving deliberate metaphor.
(17) Trials were but the visible tip of an archipelago of repression: prison, exile,
forced labor battalions. (p. 191)
(18) But whereas the Moscow Trials of the 1930s, particularly of the 1938 trial of
Nikolai Bikharin, had been sui generis, theatrical innovations whose shock
value lay in the grisly spectacle of the Revolution consuming not just its
own children but its very architects, the trials and purges of later decades
were shameless copies, deliberately modeled on past Soviet practice, as
though the satellite regimes hardly merited even an effort at verisimilitude.
And they came, after all, at the end of a long string of judicial purges.
(p. 178)
 Gerard Steen

(19) Just as Western Europe was about to enter an era of dramatic

transformation and unprecedented prosperity, eastern Europe was slipping
into a coma: a winter of inertia and resignation, punctured by cycles of
protest and subjugation, that would last for nearly four decades. (p. 195)

Example (17) sets up as a sentence where one would expect the visible tip of an
iceberg of repression, but continues in an unexpected turn, replacing iceberg with
archipelago, which entirely changes the metaphorical image from one floating ice-
berg that is a risk to a set of stable islands. The following case has an image of the
Revolution consuming its own children, which involves a cannibalistic parenting
scenario, coordinated with the Revolution consuming its very architects, which
by implication turns the revolution into a building so that the action of cannibal-
ism becomes non-sensical. In (19), Eastern Europe slips into a metaphorical coma
which is then equated with a metaphorical winter, a bad spell of a rather different
status and quality, that is moreover then punctured, which seems to turn the win-
ter into a material object such as a tire or a tank. There can be little hesitation that
these examples illustrate the essence of mixed metaphor.
I hold that each of these cases does not involve just a conflict between two
metaphors, but a conflict between two deliberate metaphors, and that this explains
their prominence as mixed metaphors. Example (17) begins with the deliberate
metaphor that trials are but the visible tip of repression, which by expectation
would be equated with an iceberg. Using the expression that something is the tip
of an iceberg is deliberately metaphorical, as it involves a multiword metaphorical
idiom that stands out from the rest of the concepts of the utterance. To then change
that expression into another, totally novel metaphor, by replacing iceberg by archi-
pelago, is also deliberate without any doubt. It is quite likely that the author was
inspired here by the topic of his text (cf. Semino, 2008) and reminded of A ­ lexander
Solzhenitzyn’s novel Gulag Archipelago.
This is clever, deliberately metaphorical writing that has been badly edited.
The archipelago metaphor does not make sense in the present grammatical con-
struction: if trials are the tip of an archipelago of repression, it follows that the
archipelago has only one tip, which is precisely not the point of the archipelago
metaphor (but in fact is the point of the iceberg metaphor). The sentence probably
meant to say that trials are the visible aspect of a system of repression (tip of the
iceberg) that has more manifestations than just trials, such as prison, exile, and
forced labor battalions (which can be compared to an archipelago of repression)
but then got reduced to a grammatically nonsensical structure. Its mixed nature
depends on the clash between the two images, while the visibility of this clash
depends on the fact that the two images are each clearly deliberate and therefore
separately prominent. Their grammatical intertwining within one flawed idiom-
atic construction may have increased this visibility.
 Chapter 6.  Mixed metaphor is a question of deliberateness 

Example (18) has also textually reduced two coordinated deliberate meta-
phors: the revolution consumes its own children and the revolution consumes its
own architects. Both metaphors are deliberate for the same reason: they involve
an abstract subject, the Revolution, functioning as a personified agent in the can-
nibalistic action of consuming people, in the one case its own children and in the
other case its own architects. These metaphors are deliberate for the high degree
of fantastic content that draws attention to itself as perhaps a form of hyperbole
which to some may be over the top. Their combination in a juxtaposition between
children and architects forms the basis of a contrast that turns out to be an irrecon-
cilable conflict, producing mixed metaphor. The prominence of the clash is again
due to the prominence of each of the two separate metaphors as deliberate.
Example (19) begins with one obviously deliberate metaphor, Eastern Europe
was slipping into a coma. We have personification plus two source domain refer-
ents again. The colon introduces an apposition to the notion of coma, but changes
it from a coma into a winter: we are shifting scales from personal misfortune to
the cycle of seasons here, which involves at least a form of aggrandizement that
can encompass an entire half continent – this may again be a case of topic-driven
metaphor (Semino, 2008). The notion of winter is obviously metaphorical with
respect to both coma as well as Eastern European life, and it is deliberately meta-
phorical in its construction of ‘a winter of inertia and resignation’. The logical clash
between the two notions of coma and winter is obvious and prominent because
both notions are deliberately positioned in the text as source domain terms in
their own right. The experience of mixed metaphor is explained again by the com-
bination of two incompatible deliberate metaphors. It is somewhat attenuated,
however, if the colon is read as introducing a reformulation of the previous meta-
phor, one that is seen as more apt in terms of scale as well as topic for the purpose.
What is interesting about these three cases is that they all seem to have a delib-
erate metaphor that is immediately magnified into another deliberate metaphor.
The second deliberate metaphor, however, also seems to be incompatible with the
first one. This produces a clear feeling of metaphor mixing, and eventually of the
possibility of bad writing.
Next is a case that involves two conflicting metaphors that may be less sponta-
neously recognizable as a mixed metaphor, which, if true, could be due to the fact
that only one of the two metaphors involved is deliberate:

(20) … – the economic cake grew fast enough in these years for most demands
to be accommodated without conflict (Judt, 2005: 266)

This example was discussed above. The text explicitly tells us that we are not talk-
ing about a regular cake but an economic cake. This is a signal for the metaphorical
status of ‘cake’ in its context, suggesting that it is to be taken as a culinary word
 Gerard Steen

here that requires reinterpretation in the domain of economics. To signal this very
operation makes the metaphor deliberately used as a metaphor. We are talking
about the economy in terms of cakes, and this is made explicit in the text: the cake
is to be seen and represented as a true cake in order to be mapped on to an aspect
of the economy.
However, this is not the whole story. The economy is to be represented in
terms of a cake that grows. This is where a conflict arises, because cakes do not
grow, but merely become bigger. We have moved outside the target domain of
economy to some source domain involving cakes which has been explicitly repre-
sented in the meaning of the utterance, but this source domain does not include
the concept of growing. From this perspective, we have a conceptual clash, which
would lead to the conclusion of mixed metaphor.
At the same time, however, both cakes and economies can conventionally
‘grow’. And the specific use of the verb grow in this particular context does not
draw attention to itself as a deliberately used metaphor in connection with either
cake or economy – there is nothing in its semantic scale, grammatical positioning,
or pragmatic properties that makes it stand out as a metaphor. This downgrades its
perceptibility as introducing a second source domain.
What is illustrated by (20), therefore, is the disjunction between the fact that
two adjacent metaphors in the structure of language may display a conceptual
conflict, on the one hand, and the fact that their communicative force may vary
between deliberate and non-deliberate use, on the other hand. If the communica-
tive force of one of the two metaphors involved in the conceptual clash is non-
deliberate, it becomes less prominent as a metaphor and it may, as a result, remain
‘invisible’ as a second alien source domain. If this happens, the experience of a
clash will be attenuated or disappear, so that a sentence like (20) would be less eas-
ily experienced spontaneously as a mixed metaphor (unless one is on the lookout
for conceptual clashes in order to detect mixed metaphors in bad writing). This is
a prediction based on structural-functional analysis of metaphorical language use
that can be experimentally tested.
Similar questions can be asked about (21), also discussed above:
(21) The Italian welfare state in the 1960s was still a rather rough-and-ready
edifice that would not reach maturity until the following decade … (p. 413)

For (21), the basic structure of the sentence says that the Italian welfare state is an
edifice, which involves an A is B metaphor that to many is the prototypical form
of metaphor as a rhetorical figure of speech. It involves the deliberate construc-
tion of a false identity or class-inclusion statement that is too prominent to miss
as a purposeful rhetorical device. If this is accepted, the temporary activation of
‘edifice’ as a concept and referent of its own in people’s attention is ineluctable,
 Chapter 6.  Mixed metaphor is a question of deliberateness 

and this requires integration of this concept and referent as an alien entity into the
dominant semantics of the target domain. This is when a cross-domain mapping
is needed.
Given this presence of ‘edifice’ as a source domain concept and referent in
attention, the elaboration of its nature by ‘reaching maturity’ can become problem-
atic. The text says that we are dealing with an edifice that would not reach maturity.
A quick check does not return the combination of these two terms as a regular col-
location in the domain of buildings or architecture, so that ‘maturity’ must be seen
as genuinely metaphorical with respect to ‘edifice’. It is clearly also metaphorical
with respect to the true target of the utterance, the Italian welfare state. However,
the metaphorical meaning in that connection becomes more conventional: the
online Longman dictionary has as an example ‘the era when the Republic came to
political maturity.’ This yields an interesting problem for analysis.
From the perspective of the grammatical connection with ‘edifice’, ‘reaching
maturity’ displays two crucial metaphorical referents (reach and maturity), which
in combination with edifice form a novel and visible metaphorical construction
that can qualify as possibly deliberate. This reading would make the clash with the
first deliberate metaphor, ‘the Italian welfare state was an edifice’ quite prominent,
explaining why this clash between the two metaphors may be experienced as a
typical case of mixed metaphor.
However, from the perspective of the Italian welfare state, ‘reaching matu-
rity’ might be seen as much less deliberate. This is a conventional expression
for complex systems to come to developmental completion, and it is only the
presence of two source-domain referents, ‘reach’ and ‘maturity’, that could count
as symptoms of deliberateness. But given the idiomatic nature of the metaphor-
ical phrase, it becomes doubtful whether it was deliberately constructed as a
metaphor. This possibly non-deliberate status of ‘reach maturity’ would make it
much less prominent as a metaphor, favoring referential representation in target
domain terms of completion only. This in turn could explain the lack of a visible
clash with ‘edifice’.
Depending on whether readers process ‘reaching maturity’ as related to ‘edi-
fice’ or to ‘welfare state’, the value of this clash might therefore become stronger
or weaker, yielding mixed metaphor or not. This is due to the ambivalent nature
of ‘reaching maturity’ as a deliberate metaphor, which depends on the way it is
semantically connected to its discursive context. This is again a prediction that can
be experimentally tested.
A comparable situation seems to hold for (22):

(22) Such a departure from Soviet practice was the thin edge of a democratic
wedge that would spell doom for Communist parties everywhere. (p. 320)
 Gerard Steen

Here we have another A is B structure for the first metaphor, turning ‘Such a
departure from Soviet practice was the thin edge of a democratic wedge’ into a
deliberate metaphor. The insertion of the domain indicator ‘democratic’ reinforces
this quality. Again, this would position the concepts and referents ‘thin edge’ and
‘wedge’ into the mental representation of the state of affairs designated by the text,
making them prominently alien entities that require integration by cross-domain
mapping in order to preserve textual coherence.
The elaboration of ‘the thin edge of the wedge’ by the next metaphor, ‘that
would spell doom’, also resembles what happens in (21). Thin edges of wedges
cannot really spell doom, so that spelling doom is metaphorical with respect to
wedge. A departure from some practice cannot literally spell doom either, so that
the second metaphor is doubly metaphorical again. However, spell doom for is
listed as a conventionally metaphorical expression, ‘cause end, death, or destruc-
tion’, and its typical subject would be some encompassing process or event, like
departures from some practice. This yields the same structural situation as with
(21): if ‘that would spell doom for’ is interpreted by readers in relation to ‘the
thin end of a democratic wedge’, it becomes quite deliberate as a metaphor, but if
it is interpreted in relation to ‘such a departure from Soviet practice’, it becomes
quite conventional and possibly non-deliberate. The same conclusion follows as
for (21): depending on the way readers represent the two metaphors in their situ-
ation model for the text, the value of the conceptual clash might become stronger
or weaker, consequently yielding mixed metaphor or not.
Is it possible to find an example with a semantic clash between two metaphors
that are both not deliberate? And can it then be argued that their prominence as
mixed metaphors is less than the above kinds of cases? Here is one possible case:
(23) The personality cult around the Soviet dictator, already well advanced
before the war, now rose to its apogee. (p. 174)

In other circumstances, the clash between horizontally advancing and vertically

rising might be noticeable as mixing metaphors, but since both metaphors are non-
deliberate, it takes some focused effort to dig this out. Another example is (24):
(24) In the palpably rigged Sofia show trial of …

One can metaphorically rig trials, but to call this rigging ‘palpable’ does not make
sense. However, since both metaphors are not deliberate, it is just the analyst look-
ing for mixed metaphors who will probably stumble over them.
There hence seems to be some ground for thinking that mixed metaphor is a
question of deliberate metaphor. Conceptual clashes between adjacent metaphors
appear to become prominent if the metaphors are used deliberately as metaphors.
If one deliberate metaphor clashes with a non-deliberate metaphor, it is possible
 Chapter 6.  Mixed metaphor is a question of deliberateness 

that there still is some experience of mixed metaphor, but this may be less strong.
Two non-deliberate metaphors clashing with each other may not produce any
recognition of mixed metaphor at all. These are claims on the basis of structural-­
functional language analysis which are currently turned into predictions for
behavioral research in our lab.

6.5  Epilogue

Let us finish by zooming out and considering some other examples.

(25) From November 1958 through the summer of 1961 the crisis over Berlin
simmered, diplomatic nerves frayed and the exodus of East Germans grew
to a flood. (p. 251)

There is a possibility that some people might object to the three distinct metaphors
in the series of three coordinated clauses in (25) as mixed, but if they do, that
depends on the fact that the three are clearly all deliberate metaphors, too.
(26) But for the peoples to the east of that barrier, thrust back as it seems into a
grimy, forgotten corner of their own continent, at the mercy of a semi‑alien
Great Power no better off than they and parasitic upon their shrinking
resources, history itself ground slowly to a halt. (p. 196)

There is a possibility that some people have trouble integrating the two metaphori-
cal images of peoples thrust back into a grimy, forgotten corner of their own con-
tinent on the one hand, and history itself grinding slowly to a halt on the other
hand, into one encompassing situation model. But if that happens, this is because
both of the metaphorical images are highly deliberate.
(27) Post-war Europe was still warmed by the fading embers of the nineteenth-
century economic revolution that had almost run its course, leaving behind
sedimentary evidence of cultural habits and social relations increasingly at
odds with the new age of airplanes and atomic weapons. If anything, the
war had set things in reverse. The modernizing fervor of the 1920s and even
the 1930s had drained away, leaving behind an older order of life. (p. 227)

There is a possibility that fading embers is seen as conflicting with running its
course and leaving behind sedimentary evidence, which might then be tran-
scended by constructing a volcanic eruption scenario that could account for all
this. This activity would make the metaphorical image in the first sentence quite
deliberate and vivid, which would then clash with the war setting things in reverse
in the second sentence, a clear case of mixed metaphor that also depends on all of
the components in the different sentences being deliberate metaphors. This might
 Gerard Steen

even revitalize ‘drained away’ as also deliberately metaphorical, adding to the met-
aphorical complexity of this passage for the reader. Mixed metaphor can clearly
also occur between sentences, but then it also depends on the deliberate nature of
the metaphors involved.

References

Gibbs, R. W., Jr., (Ed.). (2008). The Cambridge handbook of metaphor and thought. Cambridge
etc: Cambridge University Press.  doi: 10.1017/CBO9780511816802
Gibbs, R. W., Jr. (2011b). Are ‘deliberate’ metaphors really deliberate? A question of human con-
sciousness and action. Metaphor and the Social World, 1(1), 26–52.
doi: 10.1075/msw.1.1.03gib
Judt, T. (2005). Postwar: A history of Europe since 1945. London: Pimlico.
Kimmel, M. (2010). Why we mix metaphors (and mix them well): Discourse coherence, concep-
tual metaphor, and beyond. Journal of Pragmatics, 42, 97–115.
doi: 10.1016/j.pragma.2009.05.017
Lakoff, G., & Johnson, M. (1980). Metaphors we live by. Chicago: Chicago University Press.
Pragglejaz Group. (2007). MIP: A method for identifying metaphorically used words in dis-
course. Metaphor and Symbol, 22(1), 1–39.  doi: 10.1080/10926480709336752
Semino, E. (2008). Metaphor in discourse. Cambridge: Cambridge University Press.
Steen, G. J. (1994). Understanding metaphor in literature: An empirical approach. London:
Longman.
Steen, G. J. (2008). The paradox of metaphor: Why we need a three-dimensional model of meta-
phor. Metaphor and Symbol, 23(4), 213–241.  doi: 10.1080/10926480802426753
Steen, G. J. (2011a). What does ‘really deliberate’ really mean? More thoughts on metaphor and
consciousness. Metaphor and the Social World, 1(1), 53–56.  doi: 10.1075/msw.1.1.04ste
Steen, G. (2011b). From three dimensions to five steps: The value of deliberate metaphor. meta-
phorik.de, 21, 83–110.
Steen, G. J. (2013). Deliberate metaphor affords conscious metaphorical cognition. Cognitive
Semiotics, 8.
Steen, G. J. (In press). Developing, testing and interpreting Deliberate Metaphor Theory. Journal
of Pragmatics.
Steen, G. J., Dorst, L., Herrmann, B., Kaal, A., Krennmayr, T., & Pasma, T. (2010). A method
for linguistic metaphor identification: From MIP to MIPVU. Amsterdam: John Benjamins.
doi: 10.1075/celcr.14
 chapter 7

When languages and cultures meet

Mixed metaphors in the discourse of Spanish
speakers of English

Fiona MacArthur
Universidad de Extremadura

This chapter examines the metaphors used by speakers of English as a second

language (L2), showing how these are often the result of the mixing of two
linguistic and conceptual systems. The resulting “hybrid” metaphors may be
unconventional in English and therefore seen as problems in need of remedy.
However, the concept of native speaker norms as a model for metaphor
production may be unrealistic as a goal for learners. As this chapter shows,
hybrid metaphors are an almost inevitable outcome of language contact, and
emerge in the speech and writing of even highly proficient users of English as
a L2. I illustrate the type of hybrid metaphors that learners produce, and how
communicatively successful they may be in different contexts, suggesting that
proposed remedies for infelicitous metaphor use by L2 users of learners of English
can most usefully be framed from the perspective of discourse.

7.1  Introduction

Pesmen (1991) has argued that the condemnation – indeed prohibition – of mixed
metaphors arises from expectations that discourse (and the metaphors used in
creating discourse) will reveal the coherence in culture or world view that is char-
acteristic of autonomous closed systems. However, neither cultures nor languages
are autonomous closed systems – and nowhere is this more obvious than when we
consider the interlanguage systems of foreign language learners. As has long been
recognized (e.g., Weinreich, 1968), when people speak two or more languages, they
mix them. That is, despite the well-known but misleading metaphor, languages are
not “acquired”: the words, forms, and meanings of the discourse of others do not
survive unchanged when they become the “possession” of an individual, but rather

doi 10.1075/milcc.6.07mac
© 2016 John Benjamins Publishing Company
 Fiona MacArthur

undergo transformations as they are appropriated and assimilated. Learning a for-

eign language (L2) is not just a process of imitating and reproducing forms and
meanings according to target language norms; rather, learners go beyond these
norms, creating novel forms through analogy and the recombination of patterns
(Larsen Freeman, 1997). Bakhtin (1981) discussed this in the following way:
The word in language is half someone else’s. It becomes one’s “own” only when
the speaker populates it with his own intentions, his own accent, when he
appropriates the word, adapting it to his own semantic and expressive intention.
Prior to this moment of appropriation, the word does not exist in a neutral and
impersonal language … but rather it exists in other people’s mouths, in other
people’s contexts, serving other people’s intentions; it is from here that one must
take the word, and make it one’s own. (Bakhtin, 1981, pp. 293–294)

The way that learners adapt the words of English to their own semantic and expres-
sive intentions may, at times, strike the ear of the expert user of the language as
odd, unconventional, even incomprehensible on occasion. This is especially the
case when it comes to metaphor.
Studies of the metaphoric productions of learners of English as a second lan-
guage (L2) have shown that entrenched conceptual and linguistic routines familiar
from discourse in the first language tend to have a great influence on learners’
understanding of the meaning potential of the words they use to express their
ideas metaphorically. This often results in a kind of metaphor mixing that involves
the grafting of metaphorical conceptualizations and wordings onto target lan-
guage forms. I shall refer to these creations as “hybrid”1 metaphors, to distinguish
them from the standard use of the term “mixed” metaphor.
Researchers interested in examining such unconventional metaphorical
utterances have sought to discover the linguistic and conceptual processes that
motivate hybrid metaphors (e.g., MacArthur, 2010; Nacey, 2010; Philip, 2005,
2010), sometimes situating their enquiry within the framework of “error analysis”,
where the somewhat pejorative terms “interference” or “negative transfer” are a
commonplace, and native speaker norms the standard against which metaphor
performance is judged. In other words, this line of research may seek to identify

.  The term “hybrid metaphor” has been used in different ways by other researchers. For
example, it has been used to describe a particular type of visual metaphor, where the source
and target are fused in one object (Van Mulken et al., 2010), or to describe the conceptually in-
coherent metaphors that may emerge in a given language as a result of the similarity between
an autochnonous metaphor and one which has been calqued from another language (Oncins-
Martínez, 2014). Clearly, “hybrid metaphor” means something different here, and should be
regarded as a nonce-term used to distinguish the metaphor mixing described in this chapter
from the standard understanding of a “mixed metaphor”.
 Chapter 7.  When languages and cultures meet 

areas where non-native speakers’ discourse departs from target language norms
in order to be able to guide efforts at intervention (by language teachers or other
professionals involved in instructed second language acquisition) in remedying
the “problems” observed.
However, to regard the unusual metaphorical productions of non-native
speakers of English as errors may be unhelpful on at least two counts. In the first
place, the role of English as an international language or lingua franca, functioning
most often as a vehicle of communication in discourse events that do not involve
‘inner circle’ speakers of English (Kachru, 1985), makes the notion of ‘native-
speaker norm’ a questionable one (cf. Seidlhofer, 2005, 2008, 2009). The spread
of English as an international language calls for “a mindset appreciative rather
than fearful of diversity” (Clyne and Sharifian, 2008, p. 28.). In the second place,
applying the notion of error to metaphor studies in general is not a helpful way
of approaching the use of metaphor in discourse. In one sense, all metaphorical
utterances – whether they are novel or quite conventional among members of a
language-speaking group – can be regarded as erroneous, for metaphor regularly
violates the rules that hold for denotational reference, or the cooperative conver-
sational maxim of telling the truth. However, as Kennedy (1993, p. 218) put it,
although a metaphor violates normality, it does so in such a way that it makes
sense, “presenting violations that introduce relevant ideas”.
The relevance of the metaphorical idea depends, crucially, on the context in
which it is produced. As Barnden (2009) has shown, it is context that drives the
comprehension of metaphors, and certain aspects often considered necessary for
felicitous metaphorical language use, such as parallelism between source and tar-
get domains, may be overridden or made irrelevant by context. In a similar vein,
Kövecses (2009) has sought to account for variation in metaphor use, explaining
the “overriding factors that lead groups of people and individuals to employ non-
universal metaphors” (2009, p. 23) in terms of the “pressure of coherence”. In this
view, a speaker or writer’s awareness of context, which includes the cultural and
communicative context (2009, p. 18) will often lead to the choice of a specific, pos-
sibly novel metaphor, in lieu of a more widely known or used one. From the per-
spective of these researchers, then, unsuccessful or erroneous metaphors would be
those that fail to introduce a relevant idea in the discourse context in which they
are used, not those which, taken out of context, sound unusual or ‘foreign’.
One of the clearest cases of the infelicitous use of metaphor is when a writer or
speaker uses a vehicle term that clashes with one or more different vehicle terms
referring to the same topic in the same or adjacent clauses – or what is convention-
ally referred to as a mixed metaphor. This is usually taken to be problematic for the
reader or listener because, if the mental images triggered by the words in discourse
are incompatible with each other, they introduce irrelevant or distracting ideas in
 Fiona MacArthur

the same or adjacent clauses or utterances, presenting a challenge to coherence, or

the unity of thought that the reader/listener expects from a meaningful stretch of
discourse.
Using examples from a variety of sources, in this chapter I look at the inter-
play of metaphors produced by non-native speakers of English, illustrating the
type of hybrid metaphors that learners produce, and how communicatively suc-
cessful they may be in different contexts. Here, the standard understanding of
metaphor mixing may prove a useful yardstick by which to judge the relative
success of a metaphor in its context of use, as well as the ‘conceptual pacts’
(Brennan and Clark, 1996) established (or not) by speakers engaged in commu-
nicative interaction. I hope to show that efforts to guide teacher intervention or
proposed remedies for infelicitous metaphor use in the interlanguage of learn-
ers of English can most usefully be framed from the perspective of discourse,
rather than at the level of conceptual mapping or the linguistic form of a meta-
phorical utterance.

7.2  Metaphor and the learner of English

In second language acquisition research, it is a commonplace to state that the

developing language systems of language learners will be influenced by the learn-
er’s first language (L1). Weinreich (1968) referred to this as just one of three types
of language “mixing” that occurs when a speaker uses more than one language;
other scholars have preferred to use the terms ‘transfer’ or ‘interference’ to limit
their discussion to the impact of the L1 on the phonological, grammatical or lexi-
cal forms in the target language (e.g., Odlin, 1989). This substrate influence is seen
not only in the interlanguage of individuals but also in varieties of English spoken
in different parts of the world. It is thus not unusual to find that some metaphors
used conventionally by speakers of one variety of English should not be used or
well understood by speakers of another variety, an example being the idiom used
in Irish English: ‘to put something on the long finger’ (from Irish Gaelic Chuir ar
an méar fada é) (Odlin, 2003), a metaphor that is mostly unfamiliar to speakers of
English from other parts of the world.
Furthermore, since language is used to communicate shared conceptual-
izations of issues that may be of great importance locally, it is not surprising to
find that the shared cultural conceptualizations of a community in one part of
the English-speaking world might be different from another’s and be reflected in
the metaphors used in one variety but not in another. For example, Polzenhagen
and Wolf (2007) relate the widespread concern with issues of corruption in Sub-
Saharan countries to the vast stock of lexical items and expressions – including
 Chapter 7.  When languages and cultures meet 

metaphors – that denote corrupt practices in African English.2 Many of the meta-
phorical expressions (e.g. ‘eat money’ [to get rich] or ‘give kola’ [to bribe]) map
the domain of satisfying physical hunger onto the domain of acquiring material
wealth, reflecting culture-specific conceptualizations that are not shared by speak-
ers of English from other parts of the world.
It is therefore not surprising to find that researchers interested in the role
of metaphor in second language acquisition have found again and again that L1
conceptual and linguistic routines have a great influence on learners’ use and
understanding of metaphors (e.g., Johansson Falck, 2012; Littlemore, 2001, 2003;
Littlemore and Low, 2006; MacArthur, 2010; Nacey, 2010; Philip, 2005, 2010).
Indeed, although there has been a great deal more research into the ways that
learners of English understand the metaphors used by expert speakers of the lan-
guage than into the ways that such learners express themselves metaphorically
using target language words, recently there has been evidence of an increased
interest in the metaphors actually produced by learners of English as a second lan-
guage. For example, a recent study carried out by Littlemore et al. (forthcoming)
identified the metaphors used by Greek and German learners of English who took
Cambridge ESOL examinations at different proficiency levels. This study showed
that at elementary and pre-intermediate stages of language proficiency, learners
do not tend to use many metaphors, and those that they do use most commonly
involve function words such as prepositions rather than open class items such as
nouns or verbs. However, Littlemore et al. found that the number of metaphors
involving open class items increased significantly at B2 or upper-intermediate
level, in response to the communicative demands made of them. This radical qual-
itative and quantitative change in metaphor use suggests that, in response to the
kind of tasks requiring that they state their personal opinions on certain issues or
highlight their personal significance, learners need to use metaphorical language,
a finding that is very much in line with MacArthur’s (2010) small-scale study of
metaphor use by Spanish undergraduate students at upper-intermediate level.
These studies further show that, when needing to experiment with metaphor-
ical language to communicate on abstract topics, intermediate learners will tend
to draw on L1 metaphors when expressing themselves in English. For example, in
Littlemore et al. (forthcoming), these researchers point out that an unconventional
metaphor produced by a German student – “TV reporters have wrapped their
reports in dramatic pictures” – is a direct translation of the German die Fernseh-
reporter haben ihre Berichte in dramatische Bilder verpackt, which is very similar

.  These expressions were found to be significantly more frequent in the corpus of African
English the authors examined than in corpora of British and American English.
 Fiona MacArthur

to what a Spanish student did when she wrote that “you can only know the cul-
ture properly if you dominate a foreign language because language and culture
are the two faces of the same coin” (MacArthur 2010, p. 161), where ‘face’ (cara)
is the conventional metaphor in Spanish when referring to the ‘side’ of a coin, and
‘dominate’ more similar to the Spanish dominar (algo) than ‘master’, the verb con-
ventionally used by English speakers in this context.
Nacey’s (2010) study of the metaphors used by Norwegian learners of English
found that even at an advanced level, the influence of the L1was evident in some
of the novel metaphors produced by these learners. For example, Nacey explains
that when a student wrote “We have built a chaotic and pressured environment
by ourselves, and the only way to step out of this pattern is by changing our own
views” (Nacey, 2010, p. 188), this unusual expression was likely to be the student’s
rendition of skritte ut av, a phrase which, in Norwegian, can be used to refer both
to literal and metaphorical motion. The influence of mother tongue conceptual-
izations and wordings on the metaphors produced by L2 speakers of English is
therefore not found solely in the speech and writing of intermediate learners of
English, but also in those who are proficient in the language.
This is a conclusion echoed by Johanssen Falck (2012) in her study of ‘path’,
‘way’ and ‘road’ expressions used by Swedish speakers of English. When she com-
pared the way that these expressions were used in the British National Corpus
(BNC) and in a corpus of advanced Swedish students’ writing, Johanssen Falck
found that the latter used ‘way’ expressions much more frequently than native
speakers, but ‘path’ and ‘road’ expressions less frequently, concluding that “[t]
his tendency mirrors the pattern that Swedes tend to use vag (‘way/road’), but
not stig (‘path’) in metaphorical ways” and that “Swedish students do not use the
differences between real-world ‘paths’, ‘roads’, and ‘ways’ to express finer shades
of meaning” (Johansson Falck, 2012, p. 130). All in all, then, research into the
metaphors produced by L2 speakers of English has shown that, at all levels of pro-
ficiency, the influence of culture-specific conceptualizations and wordings may
often be discerned when we examine the way that such speakers express them-
selves metaphorically, often giving rise to what I have termed hybrid metaphors.
Philip (2010) has argued that it is not always a question of conceptual interfer-
ence but rather the influence of entrenched linguistic routines that contribute to
the production of anomalous metaphors in the writing of the Italian learners of
English in her classes. For example, she explains how when a learner talks about
experiencing a ‘heavy depression’, the metaphor is phrased in this way because
the equivalent word that typically collocates with Italian depresione is pesante
(‘heavy’), while in English the typical collocate is ‘deep’. The resulting expression
is “wrong”, according to this researcher, because it does not match the form con-
ventionally used by native speakers of English. It is therefore a problem in need
 Chapter 7.  When languages and cultures meet 

of remedy. In Philip’s view, attention to metaphor in English language classrooms

will aim “to prevent interlanguage from seeping into learners’ speech and writing”
(2010, p. 71). She further states that even when ‘meaning can be extracted” from
metaphorical phrases when considered in their full context of use, “by failing to
adhere to L2 phraseological norms, fluency is compromised” (201, p. 76). This is a
similar argument to Danesi’s (1995) view that it is necessary to encourage ‘concep-
tual fluency’ among foreign language learners, where ‘conceptual fluency’ seems to
be equivalent to adopting wholesale the conceptual metaphors used by inner circle
speakers of a language.
A number of objections could be made to Philip’s approach to the hybrid
metaphors produced by foreign language learners, among which is the fact that
the adoption of native speaker norms (as instantiated in the Bank of English or
other large corpora) condemns language learners to severely curtailing their abil-
ity to express themselves in ways that are coherent with their own social and cul-
tural identities. Furthermore, there is no reason to think that, just because it is
unconventional, a metaphor such as ‘heavy depression’ could not be used in a par-
ticular discourse context to communicate a rich understanding of depression as a
weight, as opposed to conceptualizing it as a pit or void. I shall return to this point
later, but here it is worth recalling that when speakers of any language appropriate
(or ‘borrow’) the words used in another language, they are at liberty to rephrase
or reconstrue their metaphorical senses to serve their own communicative pur-
poses, the only limitation being that they should be communicatively effective in
the discourse event in which they are being used. This, according to Widdowson
(2003, p. 42), is the real hallmark of language proficiency: “You are proficient in
a language to the extent that you possess it, bend it to your will, assert yourself
through it rather than simply submit to the dictates of its forms”.
Ultimately, though, whatever one’s objections to the kind of cultural imperial-
ism implied by the research that seeks to foster adoption of native speaker meta-
phors by non-native speakers of English, the whole effort would most likely prove
unsuccessful, as variation, rather than standardization is the hallmark of language,
including those languages which are used by speakers living far from the areas
where a particular language originated. Thus, to consider the hybrid metaphors
of speakers from other parts of the world simply as errors or problems in need
of remedy is to run up against exactly the same sort of problems experienced in
earlier times when Latin, not English, was the international vehicle of communi-
cation. Familiar to all scholars of the Romance languages is the painstaking but
ultimately useless work of an unknown Roman grammarian who, in his Appendix
Probi (the palimpsest attached to the Instituta Artium, a work written in the third
or fourth century AD) listed 227 errors in ‘X not Y’ form. The aim of this work
was to bring the use of Latin throughout the Roman world into line with C ­ lassical
 Fiona MacArthur

Latin, and focused on errors of pronunciation, grammar or lexis in the local vari-
eties that had evolved in various regions. The effort was, of course, fruitless, just
as attempting to prevent the emergence of interlanguage forms in the speech or
writing of learners of English may prove equally fruitless.
In this regard, it should not be forgotten that speakers of English have also
taken some great liberties with words and metaphors “borrowed” from other
­languages when adapting them to their own expressive ends. A clear case of this
is the use of the bound preposition ‘on’ following the verb ‘depend’, which is the
anglicized form of the Latin verb dependere. Image schemas of balance appear
to be associated by speakers of English with embodied experiences of ‘support’ or
upward pressure from below (resistance), metaphorically expressed in words
and phrases that construe this as the self having something below the body to keep
it from falling (e.g. lean/rely/build on someone or something) (Gibbs et al.1994,
Johnson 1987). However, individuals may also have embodied experiences of
how the danger of falling may be averted by attaching themselves to something
above the body, as a person in danger of falling from a tree or a steep slope might
grab something above him/her to keep him/herself in place. The metaphori-
cal sense of Latin dependere (lit. ‘hang from’) arises from this alternative image
schema.
As the Online Etymological Dictionary recalls, English ‘depend’ appears in
the language early in the 15th century, with the meaning: “to be attached to as
a condition or cause,” a figurative use, from M.Fr. dependre, lit. “to hang from,
hang down,” from L. dependere “to hang from, hang down; be dependent on, be
derived,” from de- “from, down” + pendere “to hang, be suspended” (emphasis
added). Early uses of the verb in English cited in the Oxford English Dictionary
(OED) show that usage varied as regards the following preposition: this could be
‘from’, ‘of ’ and even ‘in’ (for example, “ the werk that he werketh dependeth of
fortune and not of hym” [1413] or “from a right understanding of this, depends
the Knowledge of many Places in both sacred and profane Writers” [1730]) with
‘on’ becoming established as the preferred preposition only in the 19th century:
for example, “whether the bond should be enforced or not would depend on his
subsequent conduct” [1848] (all emphases added).
Of course, attachment and support may be conflated in English uses of
the preposition ‘on’ (compare, for example, ‘the picture on the wall/the apple on
the tree’ with ‘the book on the table/the ship on the sea’). However, Romance
languages such as Spanish, French or Portuguese maintain the coherence of the
image as attachment to something above by using the preposition de (‘of/from’)
after the verb (Fr. dépendre de, Sp. depender de, It. dipendere da, and so on). For the
speaker of a Romance language, then, speakers of English have created a mixed
(hybrid) metaphor when bringing ‘depend’ into line with support/­attachment
 Chapter 7.  When languages and cultures meet 

uses in ‘lean’ or ‘rely’ + on, expressing this simultaneously as support from some-
thing below and attachment to something above when using the combination
“depend + on.”
The possibility that this is a mixed or hybrid metaphor is most probably com-
pletely opaque to ordinary speakers of English. Interestingly, however, for Spanish
learners of English (as well as speakers of other Romance languages) it is not. Typi-
cally, speakers of these languages “misuse” the bound preposition after ‘depend’
in English, preferring ‘of ’ instead (Jiménez Catalán, 1996, p.11). The following
extract from the VOICE corpus shows this usage in the speech of a person whose
first language is Portuguese:

fifty per cent of the decision (.) of the board that will decide (.) if the budget will
be approve(d) or NOT (.) fifty per cent of (the decision) will depend of youth (.)
er s- structures (.) representative structures (.) and the other (.) the other half will
depend of the governmental members. (Source VOICE corpus
 [powgd 510: 389] http://voice.univie.ac.at/ Bold type added)

The use of ‘of ’ instead of ‘on’ in this and other contexts might be classed sim-
ply as a performance error, possibly the result of the choice of a preposition that
sounds similar to the one sought, or the result of the kind of probabilistic process-
ing (Chater & Manning, 2006) which will lead the learner to substitute a more
frequent, familiar form (in this case, the preposition ‘of ’) for the more idiomatic
one (in this case, ‘on’). However, Spanish undergraduate students I have taught
over the years vigorously defend their choice of ‘of ’ when corrected on this par-
ticularly recalcitrant error in their speech or writing. For them, the use of ‘on’
seems completely unnatural and even illogical, and they justify their choice of a
different preposition as being more congruent with the sense of ‘depend’, which
they identify with the virtually identical verb depender in Spanish. Whatever the
justification for the use of the preposition ‘of ’, this will likely sound impossible to
the expert speaker of English because it involves a closed class item or function
word, and thus violates the kind of “phraseological norm” (Philip 2010, p. 76) that
exerts a great pressure towards conformity in use. When learners insert ‘of ’ in the
slot always reserved for ‘on’ by expert speakers, this will be perceived as simply
the wrong choice – and hence teachers will discourage learners from using this
preposition.
In contrast, metaphorical creativity is likely to be better tolerated when it
involves the use of open class items or lexical words. Unconventional metaphori-
cal utterances involving open class items are more difficult to classify as being
communicatively ineffective or “wrong”, if only considered as decontextualised
utterances. For example, an undergraduate student at a Spanish university wrote
about an argument she had with her boyfriend, expressing her attempts to keep
 Fiona MacArthur

her ­temper as “I tried to hold back my nerves” (MacArthur, 2010). Here, for the
expert speaker of English, there appears to be a clash between ‘hold back’ and
‘nerves’. ‘Nerves’, like the equivalent noun nervios in Spanish, literally refers to the
bundles of fibres that form part of the system that conveys impulses of sensation,
motion, and so on, between the brain or spinal cord and other parts of the body.
Speakers of both Spanish and English, however, use the name of these fibres to
describe emotional states, particularly the nervous agitation that is caused by fear,
anxiety, or stress (‘an attack of nerves’/un ataque de nervios, for example). And in
both languages the invariable plural may stand for a state of anger or irritation:
Spanish poner a alguien de los nervios (Lit. ‘to put someone of the nerves’) is very
similar in meaning to English ‘get on someone’s nerves’. However, while English
‘nerves’ can be ‘soothed’ or ‘calmed’ (as though they were some kind of animate
being), they are not usually ‘held back’, for the evidence from the BNC shows that
among the significant collocates to the right of this phrasal verb we do not find
‘nerves’ but rather nouns that denote liquids (‘hold back the tide/tears/water’),
people (‘hold back rioters/crowds’), abstract processes (‘hold back progress’) or
even emotions (‘hold back impatience/vitriol/passion’).
It would appear overly simplistic, however, to suggest that the metaphor used
by this student does not work simply because of the mismatch between the con-
ventional collocates of ‘hold back’ or ‘nerves’, or that we have here an incom-
prehensible or unacceptable hybrid metaphor. On the face of it, there seems no
good reason to think that if we are able to think of ‘nerves’ as an animate entity
that can be soothed or calmed we could not conceive of their being ‘held back’ or
restrained. All that can be said is that it is unconventional among E­ nglish speakers
to express themselves in this way. And if the only objection to this ­metaphorical
production is that it is unconventional, and therefore it is simply an error, this
would be tantamount to suggesting that no novel metaphor could possibly be
interpreted nor a conventional mapping be elaborated (in the sense of Lakoff and
Turner [1989]). Rather, speakers and writers would be trapped in the conven-
tional forms of metaphorical reasoning and expression that have been handed
down from generation to generation of English speakers, without being able to
add anything to the pool of alternative metaphorical construals of experience,
events and situations.
Variation in metaphor use is well-attested, and responds to a number of dif-
ferent factors (cf. Kövecses’ [2011] discussion of context-induced variation for
example). It seems likely that the hybrid metaphors of L2 speakers arises from
the “pressure of coherence” (Kövecses, 2005, 2009), in this case the pressure to
be coherent with one’s own native metaphorical language and conceptualiza-
tions, which is bound to differ in significant ways from that of the L1 speaker
of English.
 Chapter 7.  When languages and cultures meet 

7.3  M
 ixed, extended, and repeated metaphors in language learner
discourse

In order to understand how discourse context may affect the success of a hybrid
metaphor, let us consider how two undergraduate students employ a calque of
the conventional Spanish metaphor abrir la mente (a algo) (‘to open one’s mind
to something’) when writing about the experience of learning a foreign language
(see MacArthur, 2010, for further details). Although English speakers may use a
somewhat similar metaphor to describe a person’s ability and willingness to con-
sider and understand different points of view from his/her own, this is commonly
seen as a static attribute of a person – ‘to be open-minded’ – while in Spanish it is
construed as a dynamic process.

1. Travel broadens the mind and learning a foreign language is very useful if you
want to go abroad. You can visit other countries and know new people, new
cultures, and this can open your mind.
2. The first two years of my university life helped me open my mind and see that
English was far away more complex than what I expected. It represented my
first experience with the culture that surrounded the language, a way to a bet-
ter understanding of English, a path to my original aim.

In Extract (1) the student explains the benefits of learning a foreign language by
appealing to the over-worn metaphor ‘travel broadens the mind’, a cliché in Eng-
lish that invites some kind of elaboration if any fresh life is to be breathed into
it. However, the student expands the notion by saying that this “can open your
mind”, thus producing a mixed metaphor. ‘Broaden’ implies conceptualizing the
mind as a malleable substance; ‘open’, in contrast, implies some kind of container,
which is not congruent with the image triggered by ‘broaden’. Goatly (1997) or
Müller (2008) have pointed out that readers and listeners may not notice a mixed
metaphor such as this if the images that motivate them are not triggered by the
words used. That is, if the metaphor is ‘dead’. However, the repetition of the word
‘mind’ in this text is likely to activate the link between the two expressions, and
bias the reading of both as metaphors. In this case, the clash will be apparent and
the use of the two metaphors in close adjacency infelicitous.
The second student, in contrast, uses the hybrid metaphor in a context
(Extract [2]) which supports the metaphoric conceptualization offered. Key to the
extended metaphor this student uses is the notion of distance (‘far away’) where
a ‘path’ or ‘way’ can be discerned (‘see’). In this context, the container metaphor
instantiated in ‘open my mind’ is perfectly congruent with the surrounding text
(suggesting, perhaps, that the container-mind has a closed window or door, which,
 Fiona MacArthur

when opened, allows the person to ‘see’ what is outside or ahead). In a sense, what
the writer has done here is tantamount to what Sharifian (2009) has suggested that
people need to do when engaged in intercultural communication: to spell out for
their interlocutors the background that informs their use of words:
[O]ften speakers build on the assumption that their cultural conceptualizations
are shared by their hearers. In the revised model of communication for English
as an International language, interlocutors would first need to minimize the
assumption of shared cultural conceptualizations. That is, participants in EIL
communicative events would need to constantly remind themselves that ‘other
interlocutors may not share the same schema, category or metaphor that I am
drawing on in my production and comprehension’ (2009, p. 247).

The difference between the use of the same metaphor in these two texts lies in
the fact that the second writer has supported and made comprehensible his
­culture-bound conceptualization through the use of an extended, coherent met-
aphor, while the first writer has relied on superficial markers of textual cohe-
sion through the repetition of a key word (‘mind’), a lexical tie which creates the
conceptual clash.
Repetition – whether of one’s own words or of another’s – is a characteris-
tic of spoken discourse that has received a great deal of attention. Metaphors in
spoken discourse may be repeated, reworded or challenged within and across
turns in the course of an interaction (Cameron, 2010), a feature that can also be
observed in conversations held in English that involve non-native speakers of Eng-
lish (MacArthur & Littlemore, 2011). In the following extract from a conversa-
tion between Jim (pseudonym for a lecturer at a British university) and Carmen
(pseudonym for a Spanish undergraduate student spending a period of time in
the UK as an Erasmus student), the interlocutors are discussing the kind of exam
questions Carmen will be asked and how she should put together and write up her
answers (for more details about the data reproduced here, see Littlemore et al.,
2012 or MacArthur, 2011). Here we find the metaphor ‘make a story (of it)’ (used
in response to Carmen’s difficulty in joining the ideas “My problem is that they are
not developed so they have just in squares, so I have to joint all the ideas”) re-used
by both interlocutors:
Jim: So what you have to do in the exam is you have to- you have to- so each
lecture is about a different topic and you have to explain one of those topics, um
so maybe language and gender, we talked about language and gender, that was
last week, or this week was speech, you know, so the features of speech compared
to the features of writing, um so yes, you do that

Carmen: My problem is that they are not developed so they have just in squares,
so I have to joint all the ideas, so the- what- this is what I have to do
 Chapter 7.  When languages and cultures meet 

Jim: That’s the difficulty, so you have a PowerPoint, and it tells you this, this, this,
this, and this, and you have to make sense of it, yes, it’s difficult. It is difficult
to make a story. Sometimes your PowerPoints they just tell you lots of- so we
did that lecture a few weeks ago on genre, remember that, and that was very
technical. You had a piece of information about this, a piece of information about
that, a piece, and it’s hard, I agree, it’s hard to make an essay out of it

Carmen: But in your lectures, I am very grateful of them because you can explain
us, you- you do an introduction before, so you joint all the ideas, you give us an
introduction, you joint all the ideas this PowerPoint with this one, who is this
author, yeah, it’s better.

Jim: Well I’m glad about that, okay, so that- I’m glad that helps, so you think
making a story out of it helps

Carmen: Yes

Jim: As opposed to just looking at the PowerPoint slides

Carmen: Or maybe my house when I read all the PowerPoints I try to do a story
in my mind and I do, like, an outline

Carmen repeats “joint all the ideas” three times, identifying what for her is the
main problem involved in turning the information gathered in a lecture into a
piece of writing, attributing to Jim the ability to link ideas: “I am very grateful
….you joint all the ideas”. Jim’s response to her difficulty is to focus on creat-
ing something, signalled through his metaphorical uses of ‘make’: “make sense”,
“make a story”, “make an essay out of it”. And he interprets Carmen’s expression
of gratitude for his ability to ‘joint all the ideas” as being co-referential with his
own formulation of the solution: “so you think making a story out of it helps”, an
interpretation ratified by Carmen’s re-use of the same metaphor (substituting ‘do’
for ‘make’) “when I read all the PowerPoints I try to do a story in my mind”.
Gibbs and Cameron (2008) discuss the re-use of metaphors in the reconcili-
ation dialogues studied by Cameron (2007). In their study, Gibbs and Cameron
relate the emergence and stabilisation of metaphor in discourse or the ‘meta-
phoreme’ (Cameron & Deignan, 2006) to Brennan and Clark’s (1996) notion of
‘conceptual pacts’. This also proves useful when considering the use of metaphor in
intercultural communication. Put briefly, Brennan and Clark describe how when a
concept needs to be labeled for the first time in discourse (e.g., a particular type of
shoe, the loafer, as opposed to shoes generally), one speaker may provide an ad hoc
label for this concept, which may be picked up and repeated by a conversational
partner. The use of the term (a single word or phrase) as the shared way of refer-
ring to the concept is termed a ‘conceptual pact’, an agreement that may be only
temporary and in use in one particular conversation, or which may be re-used and
 Fiona MacArthur

recycled in further discourse events. The use of the metaphorical ‘make/do a story
out of it’ seems to constitute a temporary conceptual pact between Jim and Car-
men, a shared way of understanding how to overcome the difficulties involved in
creating a coherent text from separate or previously unrelated ideas.
However, immediately after her re-use of this metaphor, Carmen appears to
break this joint understanding by using a very conventional metaphor which is
not fully coherent with the ‘story’ metaphor: “I try to do a story in my mind and I
do, like, an outline”. For the expert speaker of English, the basic meaning of ‘out-
line’ is “the contour or boundary, real or apparent, by which a figure is defined or
bounded in the plane of vision” (OED sense 1b), while its conventional figurative
extension signifies “a rough draft or general sketch in words; a description, giving
a general idea of the whole, but leaving details to be filled in” (OED sense 3.a.) An
‘outline’ therefore does not ‘joint’ (sic) ideas nor provide a narrative thread (‘make
a story’) and therefore appears incompatible with the foregoing discourse. How-
ever, Carmen marks the use of this word as relevant to ‘do[ing] a story’ twice, once
by the use of the coordinating conjunction ‘and’, and then by hedging the expres-
sion with a marker that may possibly introduce a non-literal comparison: ‘I do,
like, an outline’. A reasonable interpretation of Carmen’s intended meaning (which
does not, of course, match the conventional understanding of ‘outline’) is that of
another kind of ‘line’ or thread: one that joins rather than sketches the contours of
an argument. And her use of ‘like’ may be what makes it possible for Jim to under-
stand the connection that Carmen sees between ‘making/doing a story’ and ‘doing
an outline’. The conversation continues in the following way:
Jim: Right

Carmen: Do an outline about all the ideas, but later you have to write them down.
I haven’t done them yet but I do it. And also, I will help, I will ask for help to some
tandem partner because if I do summaries I will I then check it with

Jim: yeah I see ok

Jim’s use of ‘right’ and ‘I see’ are conventional signals of comprehension of what
another has said, and show that Jim’s intention is to communicate that he is under-
standing where Carmen is going with her development of the ‘story’ metaphor.
Whether he did in fact fully understand this or see its relevance is impossible
to ascertain for certain; what can be observed is that Carmen’s mixed metaphor
caused no immediate problem for the interaction in which it occurred. The fore-
going discourse appears to have afforded the sufficient contextual support for the
novel use of ‘outline’ to be successfully interpreted.
In contrast, a similar “misuse” of another compound involving ‘line’ did cre-
ate a problem in a conversation between a Spanish student José M and his lecturer
 Chapter 7.  When languages and cultures meet 

in English Language Teaching Methodology, Debbie. When Spanish learners of

English use compounds with ‘line’, some appear to focus on the second element,
rather than on the first (classifying) element of the compound (out/head/guide) as
the following extract shows:
Debbie: Does it help you when you have assignments to have very clear urm steps
to follow and guidelines?

José M.: Yes

Debbie: Or do you prefer when you’re freer to decide?

José M.: I prefer following the steps that I have in the headline

Debbie:In the guidelines

José M.: In the guidelines, cos I think I work better if I follow a pattern of the- the
guidelines

Debbie interprets José M’s use of ‘headline’ in turn 4 as an error (perhaps imply-
ing that he does not understand what is meant by ‘guidelines’), which leads to her
repetition of the term to prompt its exact replication by the student. However, it
does not seem as though José M has misunderstood, because he has picked up on
and expanded on Debbie’s metaphor of the usefulness of ‘having steps to follow’,
which would lead one to think that, for him, a ‘headline’ is very much the same
thing as a ‘guideline’, that is, what is salient for him (and for Carmen when using
‘outline’) is ‘line’.
In English and its Spanish equivalent línea this means ‘a stroke or mark …
traced with a pen or tool, etc., upon a surface’ (OED sense 7a) or “sucesión conti-
nua e indefinida de puntos en la sola dimensión de la longitud” (Diccionario de la
Real Academia Española [DRAE] sense 1). In Spanish línea can be used in sim-
ilar ways to English ‘outline’ (Forma, silueta o perfil. [DRAE sense 4]) but also
to describe ‘conducta o comportamiento en una determinada dirección’ (DRAE
sense 6), that is, action or behaviour carried out in one particular direction, which
the DRAE illustrates with the phrase [s]u línea fue muy coherente (lit: his line was
very coherent). The convergence and divergence of the extended senses of ‘line’ in
the two languages leads to these learners’ using the compounds in ways which are
not coherent with the senses they have for native speakers of English. However,
how a native speaking interlocutor may respond to this will also be different. Deb-
bie has been José M’s teacher of English for a number of years, and sees her role as
encouraging him to produce well-formed utterances, hence her correction of his
utterance. Jim’s role is different: in his conversation with Carmen his focus is on
the content of the discussion, rather than on the form of her utterances. Differ-
ences such as these are likely to prove important in conversations between native
 Fiona MacArthur

and non-native speakers and contribute to different understandings and evalua-

tions of non-native speakers’ use of hybrid metaphors.
According to Brennan and Clark, conceptual pacts are specific to the interac-
tions in which they occur:
[W]hen speakers ground a reference they are creating a conceptual pact, a
temporary agreement about how the referent is to be conceptualized. So when the
same speakers face new addressees, they have to establish new conceptual pacts,
and these may not be the same as those established with previous addressees.
(Brennan & Clark, 1996, p. 1484)

In Figure 7.1, which reproduces the transcript of a conversation between three

business students with different L1s, we can see how a cluster of meaning emerges
around the idea that a particular type of appearance can be likened to the (unde-
sirable) look of a puppy, at least as far as S1 – a young man from Venezuela – is
concerned. S2, a female (L1 Dutch) teases him about not wanting to go out on a
date because of his appearance, picking up on and re-using the figurative compari-
son. S3, the third participant in the conversation, is another male (L1 Indonesian).
He does not use the word ‘puppy’ but his contributions show he understands the
grounds of the comparison:

214 S2: And just have a good night’s sleep go on your DATE or whatever you want
to do tonight=
215 S1: =no: i’m not going to cos i gotta ma:n i’ve shaved i look like a puppy (.)
216 S2: <imitating> so maybe she likes pu:ppies </imitating>=
217 S2 =oh man no but this <1><fast> you’re a woman </fast> </1>
218 S3 <1> is that </1> what’s bothering you? <2> <@> your shaving </@> @
</2> @@@
219 S1 <2> @@@@</2>
220 S2 i don’t MIND a puppy look from time to time?
221 S1 <soft> from time to time </soft>
222 S3 it’s a ritual <@> man @ @ man’s shaving </@> (.)
223 S2: yeah (1)
224 S1: i haven’t shaved in the longest time (2) like ALL of it. like it’s all gone (2)
225 S2: puppy (2)
226 S1: and they’re probably thinking something else <to S4> no it’s my fa:ce we’re
talking about </to S4> (1)

Figure 7.1.  Puppy metaphor in VOICE corpus (http://voice.univie.ac.at). EDcon 496:214–226.

Bold type added

S1 compares his appearance to that of a puppy, to explain that he’s not keen to
go on a date because he has shaved. S2 picks up and repeats the word ‘puppy’
three times (‘maybe she likes puppies’ [216], ‘puppy look’ [220] and ‘puppy’
[225]). The fact that the ‘puppy-look’ is insider talk (and therefore not expected
 Chapter 7.  When languages and cultures meet 

to be ­comprehensible to people who have not taken part in the conversation) is

signalled by S1 in 226 when he addresses the researcher (S4) who is recording
the conversation and says “they’re probably thinking something else. No it’s my
face we’re talking about”. And indeed there is no evidence in the British national
Corpus (BNC) or the Corpus of Contemporary American English (COCA)
to show that comparing one’s appearance to that of a puppy will convention-
ally imply the absence of hair on the face. Rather, in these two corpora, ‘like
a puppy’ is used to describe behaviour (boisterousness, servility, and so on)
rather than appearance. Thus the metaphoreme that emerges in this discourse
context is an example of a temporary conceptual pact between these particu-
lar interlocutors, a phenomenon that Seidlhofer, in her analysis of English as
lingua franca conversations like this one, refers to as ‘local idiomaticity’ or the
dynamic online application of the idiom principle and “the use of local idiom-
atic coinages devised to meet an immediate communicative need” (­ Seidlhofer,
2009, pp. 197–8). When considered in the light of Brennan and Clark’s find-
ings on conceptual pacts, we notice that an important feature of such pacts is
evident here; the ad hoc labels used to name concepts in conversations tend
to become shorter with each use. Here, what begins as a ‘dramatic’ compari-
son (“I look like a puppy”), is shortened by S2 to ‘puppies’, ‘puppy-look’ and,
finally, ‘puppy’. That is, the metaphor s2 is a puppy emerges in the discourse,
to become part of the memory of the three interlocutors, and possibly part of
their shared repertoire of ways of referring to and construing a particular type
of appearance on future occasions.
This conversation demonstrates clearly that successful communication in EIL
does not require the use of conventional English metaphors. Speakers are well able
to coin their communicatively relevant metaphors when they are needed.

7.4  Discussion and conclusion

As has been seen, hybrid metaphors – or the metaphors that result from the mix-
ing of two linguistic and conceptual systems and their affordances in terms of
wordings or culture-specific conceptualizations – often emerge in the speech and
writing of those who use English as a second language. Similar to the term “mixed
metaphor”, which is conventionally used to refer the conceptual clash between
vehicle terms referring to the same topic in the same or adjacent clauses, these may
present a challenge to coherence, or the unity of thought that the reader/listener
expects from monolingual speakers and writers (Pesmen, 1991) when they are
attempting to produce a meaningful stretch of discourse. It is thus not surprising
to find that this type of metaphor mixing has been condemned, or regarded as a
 Fiona MacArthur

problem in need of remedy (e.g., Philip, 2010) when judged against native-speaker
norms of metaphor production.
However, as I have attempted to show, these hybrid metaphors are the almost
inevitable result of the contact between two linguistic and conceptual systems.
Furthermore, the hybrid metaphors produced by L2 speakers of English can be
viewed simply as another manifestation of the kind of culture-specific variation
that has been observed in cross-linguistic studies of metaphor use (e.g., Kövecses,
2005). Indeed, they may also be regarded as a particular type of response to the
‘pressure of coherence’ discussed by Kövecses (2005, 2009). As he points out, the
choice of a specific metaphor in a particular communicative context is often the
result of local factors (for example, the topic under discussion) and arises precisely
because individuals and groups of people have different experiences of all kinds –
of physical environments, of prior experience of discourse events, of concerns and
interests, among many others – that affect the kind of metaphors they employ in
different circumstances. These different experiences may themselves function as
attractors (Gibbs, 2012; Gibbs & Cameron, 2007) in the dynamics of metaphor use
in a L2 and result in the emergence of hybrid metaphors in the speech and writing
of non-native users of English.
The analysis offered here of some authentic examples of the use of hybrid
metaphors in discourse involving L1 and L2 users of English, as well as one con-
versation in which no L1 speaker participated, casts doubt on the accuracy and
usefulness of the notion of error when applied to this kind of mixed metaphor. On
the one hand, I have suggested that a difference can usefully be established between
those hybrid metaphors that violate the ‘Idiom Principle’ (Sinclair, 1991) when it
comes to closed class items like prepositions (e.g. ‘depend of/on’) and those that
involve open class items such as nouns or verbs (e.g. ‘open/broaden one’s mind’).
The latter type of hybrid or novel metaphor may be more or less communicatively
successful, depending on how each one is deployed, the mode of communication,
the relationship between the interlocutors, or the purpose of the communicative
event. The use of hybrid metaphors in written text, for example, may require maxi-
mal elaboration by the writer in order to ensure the reader’s comprehension. In
contrast, in conversations, the metaphors produced may be jointly constructed
and constitute temporary conceptual pacts among the speakers involved in the
interaction.
Only a small number of discourse events have been considered here. Never-
theless, it still seems possible to single out some features of the use of hybrid meta-
phors in discourse involving NNSs of English that could prove useful, because of
their possible application in instructed second language acquisition settings.
First and foremost, training learners to become proficient users of English as
an international language in intercultural communication (rather than ­training
 Chapter 7.  When languages and cultures meet 

them only to be able to communicate with L1 users of English) would seem to

imply paying equal attention to metaphor in the L1 as in the L2. Spontaneous,
unreflective use of mother tongue conceptualizations side by side with L2 meta-
phors (as when the student linked ‘broadening’ and ‘opening’ minds) may result
in the kind of conceptual clashes that have universally been condemned under
the rubric of “mixed metaphor”. Such clashes may arise not only because a learner
is largely unaware of the metaphoricity of the words and phrases s/he uses in the
first language, but also because, in attempting to establish cohesive ties between
sentences and utterances (for example, with the repetition of a previously used
word), s/he may fail to notice that conceptual coherence has been compromised.
When conscious attention is paid to language production, such as when crafting a
written text, it may be possible for learners to take advantage of guidance on how
to mitigate the possible problems of communication caused by the use of hybrid
metaphors.
For example, learning how to explicitly signal a metaphorical use of words (by
using hedges or tuning devices of the sort described by Goatly [1997] or Cameron
and Deignan [2003]) would help learners to alert the reader/listener to the fact
that a culture-specific metaphor is being used as metaphor and is not simply some
incomprehensible misuse of English, as Carmen did when she used ‘outline’ in
an unconventional way. Likewise, the effective use of such metaphors in mono-
logic discourse could be enhanced by expanding the image chosen to express a
concept figuratively. Extended, rather than one-off metaphors, are likely to serve
the speaker/writer best when attempting to communicate his/her own concep-
tualizations to others who do not share them (cf. Barnden 2009). After all, the
conceptual pacts established between interlocutors in such communicative events
are what make dialogue possible. However, such conscious control over meta-
phorical expression may be possible in only a limited number of circumstances
(cf Gibbs, 2011) and further research would be needed to determine to what extent
such communication strategies can be taught or indeed learned.
The well-known adage “think globally but act locally” implies two quite differ-
ent types of research activity on the part of applied metaphor researchers interested
in contributing to improving L2 speakers’ use of metaphor when communicating
in English. On the one hand, thinking globally means that the notion of “native-
speaker norms” (whatever they are taken to be) has largely to be abandoned as
inappropriate in a world where English is a vehicle of international communica-
tion, rather than the property of L1 users of the language. On the other hand,
it implies that efforts to intervene would more usefully look at ways of helping
learners to use metaphor more effectively, rather than simply encouraging them to
parrot the conventional metaphors of a foreign language and culture(s). To do so
would not simply severely curtail their ability to express themselves in ways that
 Fiona MacArthur

are faithful to their own identities, but also fail to recognize the fact that hybrid
metaphors – like other interlanguage forms – are the inevitable result of contact
between languages and cultures. However, metaphor, unlike rule-bound aspects
of a language system such as grammar, is a particularly appropriate vehicle for
expressing novel (or simply previously unencountered) conceptualizations. Its
power can be turned to good use, rather than being used to trap speakers into
reproducing the pre-fabricated patterns of speech and thought of people with
quite different social, cultural and personal identities.

References

Bakhtin, M. M. (1981). The dialogic imagination: Four essays. In M. Holquist (Ed.), (C. ­Emerson &
M. Holquist, Trans.). Texas: University of Texas Press.
Barnden, J. (2009). Metaphor and context: A perspective from artificial intelligence. In
A. ­Musolff & J. Zinken (Eds.), Metaphor and discourse (pp. 79–94). Basingstoke: Palgrave
macmillan.
Brennan, S. E., & Clark, H. H. (1996). Conceptual pacts and lexical choice in conversation.
Journal of Experimental Psychology: Learning, Memory, and Cognition, 22(6), 1482–1493.
doi: 10.1037/0278-7393.22.6.1482
Cameron, L. (2007). Patterns of metaphor use in reconciliation talk. Discourse in Society, 18(2),
197–222.  doi: 10.1177/0957926507073376
Cameron, L. (2010). The discourse dynamics framework for metaphor. In L. Cameron & R.
Maslen (Eds.), Metaphor analysis: Research practice in applied linguistics, social sciences and
the humanities (pp. 77–94). London: Equinox.
Cameron, L., & Deignan, A. (2003). Using large and small corpora to investigate tuning devices
around metaphor in spoken discourse. Metaphor and Symbol, 18(3), 149–160.
doi: 10.1207/S15327868MS1803_02
Cameron, L., & Deignan, A. (2006). The emergence of metaphor in discourse. Applied Linguis-
tics, 27(4), 671–690.  doi: 10.1093/applin/aml032
Chater, N., & Manning, C. D. (2006). Probabilistic models of language processing and acquisi-
tion. Trends in Cognitive Sciences, 10(7), 335–344.  doi: 10.1016/j.tics.2006.05.006
Clyne, M., & Sharifian, F. (2008). English as an international language: Challenges and possibili-
ties. Australian Review of Applied Linguistics, 31(3), 28.1–28.16.
Danesi, M. (1995). Learning and teaching languages: The role of “conceptual fluency”. Interna-
tional Journal of Applied Linguistics, 5(1), 3–20.  doi: 10.1111/j.1473–4192.1995.tb00069.x
Gibbs, R. W. (2011). Are deliberate metaphors really deliberate? A question of human conscious-
ness and action. Metaphor and the Social World, 1(1), 26–52.  doi: 10.1075/msw.1.1.03gib
Gibbs, R. W. (2012). Metaphor, snowflakes, and termite nests: How nature creates such beautiful
things. In F. MacArthur, J. L. Oncins-Martínez, M. Sánchez-García, & A. M. Piquer-Píriz
(Eds.), Metaphor in use: Context, culture, and communication (pp. 347–371). Amsterdam:
John Benjamins.  doi: 10.1075/hcp.38.25gib
Gibbs, R. W., Beitel, D. A., Harrington, M., & Sanders, P. (1994). Taking a stand on the mean-
ings of stand: Bodily experience as motivation for polysemy. Journal of Semantics, 11,
231–251.  doi: 10.1093/jos/11.4.231
 Chapter 7.  When languages and cultures meet 

Gibbs, R. W., & Cameron, L. (2008). The social-cognitive dynamics of metaphor performance.
Cognitive Systems Research, 9, 64–75.  doi: 10.1016/j.cogsys.2007.06.008
Goatly, A. (1997). The language of metaphors. London & New York: Routledge.
doi: 10.4324/9780203210000
Jiménez Catalán, M. R. (1996). Frequency and variability in errors in the use of English preposi-
tions. Miscelánea: A Journal of English and American Studies, 17, 171–187.
Johansson Falck, M. (2012). Metaphor variation across L1 and L2 speakers of English: Do differ-
ences at the level of linguistic metaphors matter? In F. MacArthur, J. L. Oncins-Martínez,
M. Sánchez-García, & A. M. Piquer-Píriz (Eds.), Metaphor in use: Context, culture, and
communication (pp. 109–133). Amsterdam: John Benjamins.  doi: 10.1075/hcp.38.10joh
Johnson, M. (1987). The body in the mind. Chicago: University of Chicago Press.
Kachru, B. (1985). Standards, codification and sociolinguistic realism: The English language in
the Outer Circle. In R. Quirk & H. G. Widdowson (Eds.), English in the World (pp. 11–30).
Cambridge: Cambridge University Press.
Kennedy, J. M. (1993). Drawing and the blind: Pictures to touch. New Haven, CT: Yale University
Press.
Kövecses, Z. (2005). Metaphor in culture: Universality and variation. Cambridge: Cambridge
University Press.  doi: 10.1017/CBO9780511614408
Kövecses, Z. (2009). Metaphor, culture, discourse: The pressure of coherence. In A. Musolff &
J. Zinken (Eds.), Metaphor and discourse (pp. 11–24). Basingstoke: Palgrave macmillan.
Kövecses, Z. (2010). A new look at metaphorical creativity in cognitive linguistics. Cognitive
Linguistics, 21(4), 663–697.  doi: 10.1515/cogl.2010.021
Lakoff, G., & Turner, M. (1989). More than cool reason: A field guide to poetic metaphor. Chicago:
University of Chicago Press.  doi: 10.7208/chicago/9780226470986.001.0001
Larsen-Freeman, D. (1997). Chaos/complexity science and second language acquisition. Applied
Linguistics, 18(2), 141–165.  doi: 10.1093/applin/18.2.141
Littlemore, J. (2001). The use of metaphor in university lectures and the problems that it
causes for overseas students. Teaching in Higher Education, 6, 331–351.
doi: 10.1080/13562510120061205
Littlemore, J. (2003). The effect of cultural background on metaphor interpretation. Metaphor
and Symbol, 18, 273–288.  doi: 10.1207/S15327868MS1804_4
Littlemore, J., Krennmayr, T., Turner, S., & Turner, J. (Forthcoming). An investigation into meta-
phor use at different levels of second language writing. Applied Linguistics.
Littlemore, J., & Low, G. (2006). Metaphoric competence and communicative language ability.
Applied Linguistics, 27(2), 268–294.  doi: 10.1093/applin/aml004
Littlemore, J., MacArthur, F., Cienki, A., & Holloway, J. (2012). How to make yourself under-
stood by international students: The role of metaphor in academic tutorials. ELT Research
Papers: British Council Publications, 12(6), 1–27.
MacArthur, F. (2010). Metaphorical competence in EFL: Where are we and where should we be
going? A view from the language classroom. AILA Review, 23, 155–173.
doi: 10.1075/aila.23.09mac
MacArthur, F. (2011). On the use of metaphor in office hours’ consultations carried out in Eng-
lish between lecturers and students with different first languages. International Journal of
Innovation and Leadership in the Teaching of Humanities, 1(1), 23–44.
MacArthur, F., & Littlemore, J. (2011). On the repetition of words with the potential for met-
aphoric extension in conversations between native and non-native speakers of English.
Metaphor and the Social World, 1(2), 201–238.  doi: 10.1075/msw.1.2.05mac
 Fiona MacArthur

Müller, C. (2008). Metaphors dead and alive, sleeping and waking: A dynamic view. Chicago:
University of Chicago Press.  doi: 10.7208/chicago/9780226548265.001.0001
Nacey, S. (2010). Comparing metaphors in L1 and L2 English. Unpublished Ph.D. Dissertation.
University of Oslo.
Odlin, T. (1989). Language transfer. Cambridge: Cambridge University Press.
doi: 10.1017/CBO9781139524537
Odlin, T. (2003). Looking for roots in the substrate: The cases of Ebonics and Anglo-Irish. OSU-
WPL, 57, 120–128.
Oncins-Martínez, J. L. (2014). English idioms borrowed and reshaped: The emergence of a
hybrid metaphor in Spanish. In A. Musolff, F. MacArthur, & G. Pagani (Eds.), Metaphor and
intercultural communication (pp. 149–165). London & New York: Bloomsbury Academic.
Pesmen, D. (1991). Reasonable and unreasonable worlds: Some expectations of coherence in
culture implied by the prohibition of mixed metaphor. In J. W. Fernandez (Ed.), Beyond
metaphor: The theory of tropes in anthropology (pp. 213–243). Stanford, CA: Stanford Uni-
versity Press.
Philip, G. (2005). From concept to wording and back again: Features of learners’ production of
figurative language. In A. Wallington, J. Barnden, S. Glasbey, M. Lee, & I. Zhang (Eds.),
Proceedings of the third interdisciplinary workshop on corpus-based approaches to figurative
language (pp. 46–53). Birmingham: University of Birmingham Press.
Philip, G. (2010). Drugs, traffic, and many other dirty interests: Metaphor and the language
learner. In G. Low, Z. Todd, A. Deignan, & L. Cameron (Eds.), Researching and applying
metaphor in the real world (pp. 63–79). Amsterdam: John Benjamins.
doi: 10.1075/hcp.26.05phi
Polzenhagen, F., & Wolf, H. G. (2007). Culture-specific conceptualizations of corruption
in African English: Linguistic analyses and pragmatic implications. In F. Sharifian
& G. B. Palmer (Eds.), Applied Cultural Linguistics (pp. 125–168). Amsterdam: John
Benjamins.  doi: 10.1075/celcr.7.09pol
Seidlhofer, B. (2005). English as a lingua franca. ELT Journal, 59(4), 339–341.
doi: 10.1093/elt/cci064
Seidlhofer, B. (2008). Of norms and mindsets. Australian Review of Applied Linguistics, 31(3),
33.1–33.7.  doi: 10.2104/aral0833
Seidlhofer, B. (2009). Accommodation and the idiom principle in English as a Lingua Franca.
Intercultural Pragmatics, 6(2), 195–215.  doi: 10.1515/IPRG.2009.011
Sharifian, F. (2009). Cultural conceptualisations in English as an International Language. In F.
Sharifian (Ed.), English as an international language: Perspectives and pedagogical issues (pp.
242–253). Bristol: Multilingual Matters.
Sinclair, J. (1991). Corpus, concordance, collocation. Oxford: Oxford University Press.
Van Mulken, M., Le Pair, R., & Forceville, C. (2010). The impact of perceived complexity, devia-
tion and comprehension on the appreciation of visual metaphor in advertising across three
European countries. Journal of Pragmatics, 42(12), 3418–3430.
doi: 10.1016/j.pragma.2010.04.030
Weinreich, U. (1968). Languages in contact: Findings and problems.The Hague: Mouton.
Widdowson, H. (2003). Defining issues in English language teaching. Oxford: Oxford University
Press.
 chapter 8

The ‘dull roar’ and the ‘burning barbed wire

pantyhose’
Complex metaphor in accounts of chronic pain

Jonathan Charteris-Black
University of West England

This chapter examines metaphors in a corpus of interviews with people

experiencing chronic pain. An important communicative purpose for people
experiencing chronic pain is to get others to acknowledge its reality and I
suggest that the mixing of metaphors contributes to the communication of
pain. Where the rhetorical purpose is to communicate a means of gaining
control over pain such as therapy or medication, speakers tend to use
repeated metaphors or elaborate and extended metaphors based on some
type of conceptual blending. Metaphor mixing occurs where the speaker’s
purpose is to emphasise the intensity of the embodied experience by
representing the pain as out of control. The greater the semantic divergence of
metaphor source domains, the more intense the embodied experience of pain,
and the greater the agency of the pain rather than the speaker. Conversely, when
a speaker is discussing aspects of pain that can be controlled – as when using
medication or in therapy – he or she uses metaphors that are more semantically
convergent such as repeated or extended metaphors. Therefore the greater the
semantic convergence of metaphor vehicles, the more the speaker represents
him or herself as in control of the pain. The emergence of a metaphor theme
through use of complex metaphors therefore enhances the credibility of the
lived experience of chronic pain and shows metaphor use to be purposeful.

8.1  Introduction

In this chapter I will examine complex use of metaphor in a corpus of interviews

with people who have experienced chronic pain. A complex use of metaphor is
where a speaker uses two or more metaphor vehicles in a stretch of discourse
that addresses a particular topic or aspect of the experience of pain – such as its

doi 10.1075/milcc.6.08cha
© 2016 John Benjamins Publishing Company
 Jonathan Charteris-Black

symptoms, or its effect on family relationships. Complex metaphors range from

simple repetition to ‘mixed metaphor’ where the same topic is described using
semantically divergent metaphor vehicles. Between these two extremes are other
complex uses such as extensions and elaborations where the implications of a
metaphor vehicle are exploited in a way that is often highly expressive. Extract 1
shows complex use of metaphor to provide a strong negative evaluation of medical
professionals:

Extract 1
I thought to them we were only a lump of flesh, you know what I mean, you
weren’t a person, you were just this bit of mutton, sitting in this chair, and they
could do what they liked with you, they could use you as a guinea pig if they
wanted and you would sit there and nod …. (CP04, female, 50, care officer)

This metaphor is complex as the three phrases in italics use metaphor vehicles
relating to the semantic field of ‘flesh’ to describe a single topic: the feeling of
alienation experienced by a person undergoing a medical examination. Initially
an inanimate frame is activated by ‘lump of flesh’ which is then elaborated with
reference to animal flesh; ‘mutton’ and ‘guinea pigs’ are semantically contrasting
(lamb versus guinea pig and dead versus alive), but both metaphors elaborate on
the source domain of ‘flesh’ to describe the feelings of passivity experienced by the
speaker in relation to their doctor.
I will identify some of the characteristics of complex use of metaphor in inter-
views with people who have experienced chronic pain. Since the interviews cover a
range of topics including the physical symptoms of pain, its psychological impact,
the experience of the medical health system, sources of support, medication
and therapy I will be interested in whether different types of complex metaphor
communicate different rhetorical perspectives. I will also consider how far these
metaphors can be described as ‘purposeful’ because they realise particular com-
municative needs. Textually complex uses of metaphor – such as when they are
repeated or elaborated – is an indication that metaphors are purposeful because
they contribute to a particular communicative goal.
For a discussion of the metaphors occurring in these accounts of chronic pain
I prefer the term ‘purposeful metaphor’ Charteris-Black (2012a) because the use
of metaphors to communicate such embodied experience has the purpose of con-
veying the experience of pain – even if this purpose is not necessarily one of which
speakers are conscious. Any actions involving the body, such as swerving to avoid
a head-on collision with an oncoming car, leaping to catch a fast moving ball or
clasping our stomach when in pain are instinctive: that is to say they are purpose-
ful without necessarily being actions of which we are fully conscious. This also
extends to language when talking about pain: if I drop a hammer on my foot and
say ‘ouch’, or swear when stubbing my toe, it is an instinctive verbal response to
 Chapter 8.  The ‘dull roar’ and the ‘burning barbed wire pantyhose’ 

the experience of pain. It has the purpose of communicating this visceral experi-
ence, but it is not necessarily premeditated or planned. I suggest that the complex
use of metaphor in interviews, as in Extract 1, may contribute purposefully to an
embodied simulation of the experience and even evoke an empathetic response
(Barsalou 2008, Gibbs 2006a and b, Semino 2010). However, I will avoid making
any claims as to whether or not metaphors that contribute to embodied simulation
of an experience are ‘deliberate’ because this concept is not the main focus of this
chapter, and has been extensively discussed elsewhere (see Charteris-Black 2012a;
Gibbs 2011 & 2015; Steen 2008, 2011 & 2013). By adding to the credibility of such
accounts, complex uses of metaphor can have the outcome of communicating the
authenticity of pain and readers of this chapter will be able to make their own
judgements as to how far they do this.
Charteris-Black (2012b) found that feelings of loss of control were a central
theme with people who had experienced depression. Here I take this argument
further by suggesting that metaphor ‘mixing’ occurs in the language of people
who have experienced chronic pain where the speaker’s purpose is to emphasise
the intensity of the embodied experience by representing the pain as out of con-
trol. The more forceful the semantic collision of semantically divergent metaphor
vehicles, the more intense the embodied experience of pain, and the greater the
agency of the pain rather than that of the speaker. Conversely, when a speaker is
discussing aspects of pain that can be controlled – for example through medica-
tion or therapy – he or she repeats or extends metaphors rather than ‘mixes’ them.
For example metaphors are extended when describing the effects of medication or
pain management techniques. The more there is semantic convergence of meta-
phor vehicles through extension of the same source domain, the more the speaker
represents himself or herself as in control of the pain. An overview of this argument
is provided in Figure 8.1 below.

Complex use of
metaphor

Repeated Extended Elaborated

Mixed metaphor
metaphor metaphor metaphor

Semantically Semantically
METAPHOR VEHICLES
Convergent Divergent

Pain as under Pain as beyond

control control

Figure 8.1.  Complex uses of metaphor

 Jonathan Charteris-Black

Examination of complex use of metaphor in transcribed interviews allows the

researcher to identify correspondences between the type of metaphor that is used
and the particular stage in the interview when it occurs. This in turn allows us to
infer the speaker’s rhetorical intentions and expressive purposes – perhaps rather
more than enabling us to know for sure whether they lead to embodied simulation
by hearers. In this respect it contributes to a theory of purposeful metaphor in
which ‘purpose’ can be conceptually represented as originating in a source (the
need to recount the experience of pain) that gives rise to a path (an articulated
response in the interview) to realise the goal of p­ roviding a rich account of the
lived experience of chronic pain (Charteris-Black 2012a).

8.2  Complex metaphor

Before describing my own study in the following section, in this section, draw-
ing on relevant literature such as Fauconnier & Turner 2002, Gibbs and Franks
2002, Goatly 2011, Semino 2010 and Shen & Balaban 1999, I will illustrate some
complex uses of metaphor. I will first illustrate repeated metaphor, then mixed
metaphor before considering extended and elaborate metaphors.

Repeated Metaphor
Extract 2
<I: Do you ever have any concerns about taking medication at all?
R:> Well taking medication for me keeps my pain down to a dull roar, I’m on the
minimum tablets that I can take, if I pushed it I probably could take more, but I
really don’t want to, so yes I have this feeling in my head saying, live with what
you’ve got as long as it keeps it down to a dull roar it can’t be that bad for your body
because it is still doing something, but I don’t fancy going any stronger than what
I’m on … (CP04, female, 50, care officer)

Here the phrase ‘down to a dull roar’ is repeated to describe the effect of tak-
ing medication: repetition emphasises its regularity and routine. There is a mild
semantic opposition in the oxymoron ‘dull’ and ‘roar’ since a roar from a large
beast would usually not be described as ‘dull’ – however, some form of semantic
tension is quite typical of complex use of metaphor in these interviews and the
metaphor is clearly repeated.

8.3  Mixed metaphor

A ‘mixed’ metaphor harnesses two different metaphor vehicles to refer to, or

to describe, a single metaphor target (or topic). As Goatly (2011: 287) notes,
 Chapter 8.  The ‘dull roar’ and the ‘burning barbed wire pantyhose’ 

t­raditionally, the term is pejorative; this is because in literary approaches mixed

metaphors are associated with chaotic or unclear thinking and a lack of planning.
They typically occur when a strong opinion is being voiced and speakers’ emo-
tions inhibit clear and effective communication – though in some types of writing,
such as dialogue, they may be used for poetic effect. However, in a critique of con-
ceptual metaphor’s claim that metaphors occur coherently in natural discourse,
Shen & Balaban (1999: 151) found that:
The use of metaphors in unplanned discourse appears more like free, uncontrolled
“navigation” between a large number of root metaphors than a consistent
elaboration of any unifying root metaphors. Indeed special planning seems to be
required to make discourse metaphorically coherent.

This implies that the mixing of metaphors is normal in naturally occurring dis-
course – in their case newspaper articles – and that the activation of a single root
or conceptual metaphor requires some ‘planning’ or effort on the part of the lis-
tener; this suggests that metaphors may occur according to a speaker’s purpose at
a particular point in the discourse.
One of the problems faced by previous researchers into metaphor in health
and psychotherapeutic contexts (e.g. Gibbs & Franks 2002; Levitt et al. 2000;
McMullen & Conway 2002; Plug et al. 2011) is that the semantic approach to the
classification and counting of metaphors entails that they are allocated to particu-
lar source domains. However, where two metaphor vehicles from different source
domains are used to describe a single topic – so-called ‘mixed metaphors′- as is
often the case in these contexts, it is not clear as to whether to classify the meta-
phor by one, or both, source domains. This has implications for the counting of
metaphor and it is partly for this reason that the approach in this chapter is quali-
tative and treats mixed metaphors as normal rather than exceptional.
Blending theory (Fauconnier and Turner 2002) and the discourse dynam-
ics approach (Cameron 2010) provide further evidence that metaphors are typi-
cally mixed in naturally occurring discourse. However, the mixing of metaphors,
although purposeful may not be something of which speakers are fully conscious.
As Gibbs & Franks (2002) found in a study of the use of metaphor by women
experiencing cancer:
Our analysis of the women’s conceptual metaphors in their narratives does not
assume that the speakers consciously chose to talk metaphorically, and nor do we
assume that the speakers were conscious of their metaphorical thought. As much
research shows, people are often unaware of the metaphorical nature of their talk
and of the metaphorical patterns of thought underlying this type of speech…
(Gibbs & Franks 2002, 161–162)

It seems that metaphors can be mixed without being used consciously, and
further empirical research has been helpful in exploring this possibility (see
 Jonathan Charteris-Black

Gibbs & ­Colson 2012). The notion of ‘mixing’ implies that two (or more) meta-
phor vehicles are syntactically integrated so that the syntax ensures they are
referring to the same entity. In Extract three, where the speaker is describing
the importance of choosing the right clothing to reduce the effect of pain, three
unrelated metaphor vehicles are combined to refer to the physical experience
of pain:
Extract 3
<I: …..How do you think you came to that decision that you know this is
something that you had to live with?

<R …… And so I then suddenly thought “Well why not wear long johns” so I got
long johns and that was a joy because my trousers weren’t rubbing against my
legs. With long johns compressed tight and didn’t rub, you know. So that was a
brilliant discovery and again I said it alleviated a lot of pain and inconvenience.
I mean I used to, I used to feel that I was wearing [um] burning barbed wire [er]
pantyhose. (CP48, male, 61, writer and driver)

‘Burning’ and ‘barbed wire’ are syntactically and conceptually integrated in a single
image based on the possible causes of physical harm that combines elements of the
effects of fire and of sharp material on human skin. ‘Mixed metaphors’ typically
display such linguistic and conceptual integration – and ‘blended’ may be a better
term than ‘mixed’ for such cases. I will therefore analyse the metaphor vehicles in
this extract using blending theory.
The metaphor vehicles are syntactically integrated because ‘burning’ and
‘barbed wire’ pre-modify ‘pantyhouse’ and ‘wear’ is the subject of a verb from the
semantic field of garments. The three metaphor vehicles describe the same topic of
‘experience of pain’ with reference to possible causes of physical damage (Semino
2010: 209ff).
The metaphor vehicles also show an interesting characteristic of mixed meta-
phors: as well as being syntactically integrated they are conceptually coherent and
are fully blended. It is easy to imagine the embodied experience of having hot or
sharp objects in contact with intimate and sensitive parts of the body. In their
account of conceptual blends Fauconnier and Turner (2002) distinguish between
single-scope networks where one of the inputs forms the organizing frame for a
blend and double-scope networks where there is a clash between two organizing
frames, each of which has its own emergent structures and contributes towards
the blend. As shown in figure two, Extract 3 illustrates such a double scope net-
work: at the generic level there is a contrast between entities that harm that those
protect because in input spaces one and two ‘burning’ and ‘barbed wire’ acti-
vate frames for harm while input space three ‘pantyhose’ activates a frame for
protection. In the conceptual blend there is a semantic clash between ‘burning’,
 Chapter 8.  The ‘dull roar’ and the ‘burning barbed wire pantyhose’ 

‘barbed wire’ and ­‘pantyhose’; such clashes have the potential to form blends that
are highly creative. The conceptual relations in this double scope network are
summarised in Figure 8.2:

GENERIC SPACE (1) GENERIC SPACE (2)

CAPACITY FOR CAPACITY FOR
HARM PROTECTION

INPUT SPACE (1): INPUT SPACE (2): INPUT SPACE (3):

FIRE – ‘burning’ SHARP OBJECTS – GARMENTS –
‘barbed wire’ ‘pantyhose’

BLENDED SPACE ‘burning barbed wire pantyhose’

Figure 8.2.  Double scope networks in mixed metaphor

An explanation of the effect of such metaphors in pain accounts that I will develop
in more detail in the next section is that it may evoke a form of embodied simula-
tion on the part of hearers – as if they had actually experienced harmful objects
near their skin (Semino 2010). In this respect ‘mixed’ or ‘blended’ metaphors
share the characteristics of successful literary metaphor: although metaphors may
appear to clash because they are semantically diverse, they share the effect of sim-
ulating embodied experience. Here the generic frames of harm and protection
potentially activate an embodied response.
In blending theory this is accounted for by the generic space that contains
what both inputs have in common in single scope networks – so in the example
above although ‘burning’ and ‘barbed wire’ are from two different input spaces,
they share the common property of having the potential to inflict physical
harm. However, since this conflicts with our prototypical frame for clothing –
that it should protect, keep warm etc. – a double scope network is activated.
The element of semantic violence involved in mixing a metaphor expresses
the tensions arising from the lived experience of pain. Creative expression of
these tensions by mixed or blended metaphors, may be therapeutic in so far as
it articulates the psychological and physical reality of chronic pain as well as
potentially having the positive social effect of evoking an empathetic response
from hearers.
 Jonathan Charteris-Black

8.4  Extended and elaborated metaphor

An extended metaphor is where a series of semantically related metaphor vehicles

describe the same metaphor topic. The vehicles will be consistent in so far as they
contribute to a single coherent image. In Extract 4 the speaker gives an account of
how an extended metaphor (my term not his) was used by medical professionals
to describe the condition of his backbone:
Extract 4
As they explained it to me it was like a sand castle, you build a sand castle on the
beach and when it dries off the wind takes the grit away gradually, you know, like
sand dunes, and that’s what my problem is [er] there is no known cure….
 (CP35, male, 56, logistics manager)

His backbone is described as a ‘sandcastle’ and this is extended by drawing on our

knowledge of the properties of sand – that when it dries it becomes lighter – to
provide a single image of wind blowing and forming sand dunes.
Extended metaphors are relatively static but in conceptual blending theory the
term ‘emergent structure’ refers to some form of development in the idea under-
lying a metaphor – in particular when new conceptual relations become avail-
able in the blend that were not present in the original separate inputs (Fauconnier
and Turner 2002: 42ff.). The term ‘elaborate metaphor’ may therefore be prefer-
able when referring to such emergent structure. Elaborate metaphors develop our
understanding of a particular topic in a creative and visual way that evokes images.
Consider Extract 5:
Extract 5
<I: And you also have the pain. Can you tell me a little bit about the pain?
R:> Yeah it starts off [er] round about the top of the shoulders, round about the
neck, where your neck meets your shoulders and it’s like hot pins. Somebody
pushing hot pins into you. It works its way down the back and then misses out
about 4” in the back. Then hits the lower back. By that time it’s like somebody has
put a red-hot piece of metal inside one of your bones. And from the hip to the knee
is really extremely painful. It’s [er] the worst, the worst of it is, it’s an extremely
internal pain. It’s not a surface pain. It’s as if somebody had this hot piece of metal
and it was inside your bone and they were running it up and down. You know, and
it’s, it’s not the first time that I’ve been in tears through it. And there’s not a lot you
can do about it. (CP32, male, 57, crime reporter)

Here the speaker commences with a metaphor based image of a cause of physi-
cal damage: ‘pushing hot pins into you’; these are then elaborated with refer-
ence to movement- ‘It works its way down the back’; force – ‘Then hits the lower
back’. Then the vehicle ‘hot pin’ is elaborated by enlargement into a ‘hot piece
 Chapter 8.  The ‘dull roar’ and the ‘burning barbed wire pantyhose’ 

of metal’ and the movement metaphor is elaborated in terms of direction ‘they

were r­ unning it up and down’. Although the metaphor vehicles are from distinct
­semantic fields – ­temperature, sharp objects and movement – this is not what is
most characteristic of their complexity. Nor are they syntactically compressed as in
mixed metaphor – it will be noted that the temperature and movement metaphors
occur in separate phrases that run parallel to each other. The complexity is in how
they are elaborated in terms of detail since a powerful visual image is evoked by
the references to temperature (‘red hot’), size (‘piece of metal’) and movement (‘up
and down’).
The emergent structure of such complex metaphors expresses the intensity
of the physical experience of pain in a way that we did not find with the simple
explanatory function of the extended metaphor to describe a medical condition
(Extract 4). Such creativity and expressivity in the use of elaborate metaphor
therefore has more potential for communicating the lived experience of pain and
evoking some form of embodied response.
Although Goatly (2011) considers some other types in an account of ‘meta-
phor interplay’, these are perhaps more relevant to a literary analysis of complex
metaphor and I have selected above those types of complexity that were most
characteristic of the accounts of chronic pain in the data I examined. When meta-
phor is used in a complex way there is typically interaction between metaphors
that are repeated, extended, elaborated and mixed. More importantly, they realise
the underlying expressive purpose of the speaker. In this respect – while ‘mixed’
at the surface linguistic level – they may be conceptually and rhetorically consis-
tent because (as predicted by the notion of ‘purposeful metaphor’) they realise
a coherent perspective on the condition of chronic pain. Complex use of meta-
phor therefore provides further empirical evidence of how metaphor is typically
purposeful  – even though it makes no claims as to how conscious speakers are.
Awareness of complex use of metaphor – including ‘mixed’ metaphors – may con-
tribute to pain therapy for the psychological and social impact of chronic pain.
With this in mind I will now consider some of the claims that have been made for
embodied simulation.

8.5  Embodied simulation, pain and systematic metaphor

Barsalou (2008) offers the following definition of simulation:

Simulation is the re-enactment of perceptual, motor, and introspective states

acquired during experience with the world, body, and mind. As an experience
occurs (e.g., easing into a chair), the brain captures states across the modalities
 Jonathan Charteris-Black

and integrates them with a multimodal representation stored in memory (e.g.,

how a chair looks and feels, the action of sitting, introspections of comfort and
relaxation). Later, when knowledge is needed to represent a category (e.g. chair),
multimodal representations captured during experiences with its instances
are reactivated to simulate how the brain represented perception, action, and
introspection associated with it. (Barsalou 2008: 618–9)

Building on this work and on Gibbs (2006a & b), Semino (2010) argues that the
use of metaphor facilitates an embodied simulation of pain by people who listen
to or read such metaphors. Drawing on a range of psycholinguistic and neuro-
scientific evidence, and using illustrations from uses of the word ‘pain’ in a cor-
pus of English, she proposes that richer and more complex metaphors trigger a
more intense empathic response because they evoke more detailed pain-inducing
scenarios. She offers the alluring suggestion that detailed, creative and textually
complex metaphors in linguistic descriptions of pain activate embodied simulated
responses. Citing a study by Osaka et al. (2004) of words that are associated with
pain and contain onomatopoeic elements, she suggests that:
…..an internal simulation involving the affective component of the neural system
for pain seems to be possible in response to linguistic descriptions of pain, as well
as in the response to the perception of pain in others. (Semino 2010: 212)

She cites Gallese (2009: 520) in developing the argument that such simulation
forms the basis of human empathy:
Following this perspective, empathy is to be conceived as the outcome of our
natural tendency to experience our interpersonal relations first and foremost at
the implicit level of intercorporeity, that is, the mutual resonance of intentionally
meaningful sensory-motor behaviors. (p. 523) (In Semino 2010: 213)

She accepts the tentative nature of these speculations, partly because they are
based on decontextualised metaphors: her evidence comprises searches of word
strings relating to ‘pain’ in the Bank of English and some quotations from sufferers
of chronic pain. The extracts presented here are examples of more contextualised
metaphors from interviews with people experiencing chronic pain; however, given
the size of dataset (829,540 words) it has not been possible as yet to examine, or
quantify every complex use of metaphor.
I will argue that the pain scenarios formed by complex metaphors need also to
be understood as arising from the speaker’s underlying communicative purposes
that emerge in the course of an interview. For example, when interviewees are rep-
resenting their pain as being beyond their control they are more likely to combine
metaphors from highly divergent semantic fields (‘mixed metaphors’). C ­ onversely,
when they are have the purpose of describing therapeutic ­interventions that
 Chapter 8.  The ‘dull roar’ and the ‘burning barbed wire pantyhose’ 

enabled them to gain psychological control over pain, metaphor vehicles are more
closely related semantically. For example, extended metaphors usually occur when
interviewees are describing therapeutic interventions that facilitate a degree of
control over the pain. Such findings can be partially accounted for Cameron’s ‘dis-
course dynamics framework’ (Cameron 2010: 91) in which she defines the notion
of systematic metaphor as follows:
A systematic metaphor is not a single metaphor but an emergent grouping of
closely connected metaphors. Within the discourse dynamics framework, a
systematic metaphor is a collecting together of related linguistic metaphors that
evolve and are adapted as the discourse proceeds….. As an emergent formulation,
systematic metaphor may come to constrain and influence how discourse
participants think and talk about topics.

As I illustrated in the analysis of Extract 3, this account is concordant with con-

ceptual blending theory and there is other evidence in these interviews of how,
for some individuals, metaphors that were introduced early on in the interview
prime further uses of related metaphors to communicate a rhetorically consis-
tent account of the pain experience. As a result metaphors that are ‘mixed’ or
‘blended’ – when analysed in terms of discourse dynamics and conceptual blend-
ing – might be interpreted as contributing to a purposeful expression of the lived
experience of pain.

8.6  The interviews

There were 48 interviews relating to chronic pain that were drawn from a large
sample of 1,036 qualitative interviews with people who had experienced a health
or illness condition (either as a patient or as a carer). These interviews were col-
lected for the development of a publicly available and widely used web site “Health
Talk On-Line”1 that offers insight into what these experiences are like for the peo-
ple who have had them. Between 2003 and 2004, 29 women and 19 men who had
experienced chronic pain were interviewed in their own homes. The interviews
were open-ended, and lasted from between 90 and 180 minutes.
The aim of the interviews was to elicit personal experiences of any issues that
were important to the interviewees. They cover topics such as life before chronic
pain; the initial encounter with the condition; its diagnosis and t­ reatment – ­clinical
and psychological; its social consequences such as its effect on relationships and

.  http://www.healthtalkonline.org/ was developed by Oxford University Medical Research

Centre.
 Jonathan Charteris-Black

sources of support. Interviews were recorded by audio and/or video with the
­consent of each participant and were professionally transcribed before being
returned to the interviewees for checking. The total word length of these inter-
views was 829,540 words (including interviewers’ speech) and the average length
of each interview was 17,000 words.

Procedure
A sample of the interview transcripts (n=12) were read in full to identify meta-
phors and the topics where they occurred; the interviewer’s questions were help-
ful in identifying these topics. For example, in Extract 5 above the question “Can
you tell me a little bit about the pain?” serves to identify the topic of describ-
ing the symptoms of pain. Although the remaining 36 interviews were not anal-
ysed in detail, analysis of these was supported by computer software (Wordsmith
­version 5) so that words and phrases that were classified as metaphor in the sample
could then be examined in these interviews – using Wordsmith’s concordancing
facility – to allow some generalisation. For example, Figure  8.3 shows the con-
cordances that were identified from the metaphor ‘downward spiral’ and relate to
topic of expressing the psychological effects of pain:

Figure 8.3.  Concordance lines for ‘downward spiral’ in chronic pain interviews

Identification of metaphor involved establishing whether a phrase had a more basic

meaning. A more basic meaning is one that is more concrete, one that is related
to bodily action or one that is historically earlier (Pragglejaz 2007). These criteria
were applied at the level of the phrase rather than at the level of individual words.
This divergence from the Pragglejaz procedure is necessary when identifying com-
plex use of metaphor because by definition it is phrasal. For example, ‘downward
spiral’ (Figure 8.2) is a metaphor that incorporates two words from two different
semantic fields (direction and movement) to convey the sense of loss of control
experienced in illness. Using the Pragglejaz approach this would be ­analysed as
two separate metaphors, I analyse this as a single complex use of metaphor because
 Chapter 8.  The ‘dull roar’ and the ‘burning barbed wire pantyhose’ 

although there are two metaphor vehicles, there is only a single metaphor target:
loss of control. This qualitative approach is considered appropriate for a research
method that is not quantitative because I am interested in identifying some of the
discourse characteristics of complex use of metaphor rather than in quantifying
individual metaphors.
Once topics and complex uses of metaphor had been identified the final stage
was to identify the communicative purpose of the metaphors with reference to
their particular context where they occurred in the interview. For example, taking
a fuller context of lines 7 and 8 in Figure 8.3 above:

Extract 6
And basically I went into a complete downward spiral of pain, depression, anxiety,
anger, fear, guilt, sleeplessness, [um] side effects of the painkillers that I was taking,
you know, didn’t really have any proper pain relief. [um] It was the classic sort
of downward spiral that people with chronic problems tend to go into and I was
getting absolutely no support at all from the medical profession, they were very
uncomfortable when I appeared and tried to shove me off onto another person
[laughs]. (CP07, female, 56, careers officer)

Here repetition of the complex metaphor ‘downward spiral’ occurs when the
speaker generalises from her personal situation to that of people in general to pro-
vide a negative evaluation of the medical profession. The rhetorical purpose of
the complex metaphor is therefore evaluation but in such as way that the speaker
represents herself – at least retrospectively – as being ‘in control’ of a negative situ-
ation. Of course, at the time when she was experiencing the ‘downward spiral’ her
use of metaphor may have been quite different, but in the context of a retrospective
account of chronic pain, the repetition of a metaphor has the discourse character-
istic of showing her as ‘in control’.

8.7  Analysis of complex use of metaphor

I will first present some of the complex uses of metaphors that were used to discuss
particular topics in the interview. This will lead me to identify some of the rhetori-
cal purposes of complex metaphors – and show how they can be used systemati-
cally throughout an interview. I will argue that while particular complex uses of
metaphor correspond with the communicative purpose of the stage in the inter-
view where they occur, there is an underlying coherence in the use of complex
metaphors that is purposeful and creates the potential for embodied simulation
on the part of the hearer. Complex uses of metaphor may assist in giving speakers
a feeling of control and therefore realise this communicative purpose.
 Jonathan Charteris-Black

There were numerous examples of complex uses of metaphors to describe the

major topics covered by the interviews. However, while other forms of complex
use of metaphor were pervasive in the interviews, clear examples of mixed meta-
phor were quite rare and a number of candidate mixed metaphors were, on closer
analysis, revealed to be extensions or elaborations as in Extract one above. The
metaphors that were most convincingly analysed as mixed were similar to those in
Extract 3 and provide an account of the overwhelming physical effects of chronic
pain as in Extract seven:
Extract 7
…..because if you get up tight, if you get angry the tension, stress all goes to you
so that when you get keyed up and people that suffer chronic pain when they get
into that state their pain runs riot that is it at full swing, so you’ve got to try and
cut out the tension and the stress…. (CP03, male, 56, fitter)

Here there seems to be a mixing of metaphor vehicles relating to tension (‘keyed

up’, ‘cut out’) and motion (‘runs riot’ and ‘at full swing’). The loss of control at the
linguistic level that is communicated by mixing of metaphor symbolises the loss of
psychological control that the speaker experiences over his condition. The mixing
of metaphors arises from the intensity of the experience that is to be expressed.
The clashing of source domains that occurs in a mixed metaphor symbolically
re-enacts the chaotic and disruptive experience that characterises intense chronic
pain. The loss of physical control that is expressed from a first person perspective
with an apparent incoherence may be shown to be highly coherent and purposeful
when analysed symbolically; it is similar to what Frank (1995) describes as ‘Chaos
narratives’.
Other types of complex use of metaphor were more common than mixed
metaphor and the following table provides an illustration of these and the topics
they were used to describe. The left hand column shows the metaphor topic and
the right hand column shows complex use of metaphor (in italics) and underneath
each extract I have noted the type of complexity.

Repetition
In Extract 8 (see Table 8.1) the metaphor ‘hammer’ is repeated and has the purpose
of expressing how the speaker gained control over pain. Although at first sight this
might seem like a mixed metaphor because there are two source domains (‘fire’
and ‘tools’), this is not supported by syntactic analysis because the subject of the
fire metaphor is the body part that is causing the pain, while the subject of ‘ham-
mer’ is the speaker who is discussing a possible scenario for dealing with the pain
(hitting the body part where it is experienced).
There are several other instances where repetition of a metaphor is used to
communicate gaining control. For example, in Extract two ‘dull roar’ was repeated
 Chapter 8.  The ‘dull roar’ and the ‘burning barbed wire pantyhose’ 

Table 8.1.  Complex use of metaphors and topics

Topic Complex use of metaphor

Extract 8
Describing …because sometimes even to walk was a pain because the joints in my toes
the physical were on fire you know sometimes I felt like getting a hammer and hammering
symptoms of my toes and hammering my hands because of the pain (CP01, female, 56,
chronic pain school secretary)
Repetition of ‘hammer’.
Extract 9
Describing So the pain patient finds himself in a situation where they are trying a therapy,
the they are normally hopeful about it [um] have a lot of hopes pinned on, on it
psychological being successful and it may work for them or unfortunately it may not and
impact of if it doesn’t they are plunged back again into this sort of cycle of [um] despair,
chronic pain disappointment, [um] depression and so forth [um] until they try the next
thing. So it becomes a sort of [um] a situation where you ricochet to one therapy
and if it doesn’t work for you ricochet to another one and there’s no rhyme or
reason to it or you can spend an awful lot of money and that’s even if you have
good therapists. (CP07, female, 56, careers officer)
Elaboration, extension and repetition of ‘movement’ metaphor
Extract 10
Describing Again no effect at all and finally [um] we tried one called [um] Maprotiline
the effects of which is not very well known and for me it was a revelation. It was like a, a
medication veil coming down between me and the pain. The pain was still there. I knew I
was still experiencing it, it just didn’t seem to matter so much. It seemed to be
blunted in some strange way. (CP07, female, 56, careers officer)
Elaboration of ‘separation’ metaphor
Extract 11
Describing this GP who came in on a Saturday morning to examine me was [phew], I
the mean he’d be better in a meat market, you know, he was absolutely horrendous,
experience he was, he had no manners, he was like a production worker, you know, like
of medical in a big giant car factory, “Oh here’s another car body, here’s another car body”
professionals (CP35, male, 56, logistics manager
Elaboration (butchering and manufacturing) also extension
Extract 12
Describing And she would really honestly say to her friends and family that I help her as
the effect on much as she helps me, so yes, which makes me feel less of a burden as well, I
relationships don’t feel a burden at all you know there’s no guilt asking for help, yes it works
brilliantly, we get on so well and yes become great, great friends. (CP02,
female, 29, voluntary worker)
Repetition of ‘burden’ metaphor
Extract 13
Describing …..when you’re in chronic pain everything seems to shoved into wee boxes,
experience of where as when you’re in chronic pain you’ve got to knock down wee walls so
support that everything becomes one, but its not a big jumble, its interlaced that you
have this, its like a safety net under somebody that’s a trapeze artist, but your
safety net is your family, your support group, your friends, your medical people
and all the rest of it, that’s your safety net. (CP04, female, 50, care officer)
Extension, elaboration and repetition of ‘container’ and ‘circus’ metaphors

(Continued)
 Jonathan Charteris-Black

Table 8.1. (Continued)  Complex use of metaphors and topics

Topic Complex use of metaphor

Extract 14
Describing So this is called visualisation and we were taking this journey through a
psychological meadow to a glass building and then we’re invited into the building and up to
pain this other glass wall where you were asked if you wanted to put all your pain
management behind the wall or some of your pain behind the wall and then retrace your steps
techniques back and I remember thinking now who in their right mind would say oh well
I’ll just put some of my pain behind the wall, and I’ll take the rest back with me.
(CP01, female, 56, school secretary)
Repetition and Extension of journey and container metaphors.

and in Extract 12 there is a repetition of ‘burden’ to convey the idea that the speaker
no longer experiences herself as a burden; this corresponds with L ­ evitt  et  al.
(2000: 29) who found that ‘The metaphors in the good outcome therapy were
found to evolve from an emphasis on the experience of ‘carrying a burden’ in the
initial sessions to the experience of ‘unloading the burden’ in later sessions’. So the
repetition of a metaphor can be used to express a positive psychological experi-
ence as well as to emphasise the negative experience of the physical symptoms of
pain. In Extract 13 the repetition of ‘safety net’ communicates how social rela-
tionships contribute to gaining control over pain. In Extract 14 the repetition of
‘behind the wall’ communicates a technique for gaining control over pain. There
is, then, a clear correspondence between the communicative purposes of express-
ing control and repetition of metaphors: in this respect linguistic repetition of the
same form signifies psychological control over the medical condition.

Extension & Elaboration

Extension occurs in Extracts 9, 11, 13 and 14; elaboration and conceptual blend-
ing occurs in 9, 10, 11, 13 and 14. In Extract 9 the metaphor of ‘ricochet’ is
extended to describe the apparently random experience of changing from one
therapy to another; the metaphor highlights the speaker’s lack of control over this
process. However, ‘ricochet’ can also be analysed as an elaboration of ‘plunged’
because – although the metaphors are semantically distinct (‘plunge’ usually has
an animate subject as in ‘he plunged into the lake’, while ‘ricochet’ usually has
an inanimate subject as in ‘the bullet ricocheted around the room’) – there are
some shared grounds such as motion and speed. In this case both metaphors high-
light the lack of agency of the people experiencing this condition. The inanimate
prosody of ‘ricochet’ corresponds with the use of the passive form of the verb
‘plunged’ to express a lack of control over pain. Similarly, in Extract 10 in the uses
of ‘veil’ and ‘blunting’ arise from a conceptual blending based at the generic level
on i­ nanimate process.
 Chapter 8.  The ‘dull roar’ and the ‘burning barbed wire pantyhose’ 

These findings are reminiscent of Gibbs and Franks (2002) identification of

a novel metaphor in which a woman talks about her experience with cancer as a
type of dance:

“Dance with me,” cancer commanded. “No,” I shrieked in a fusion of fear and
disbelief. I wanted nothing to do with this would-be suitor, and I surely couldn’t
comprehend why he had chosen me in the first place. Before I could make sense
of the insanity, I realized that this dance was not optional. Cancer’s clutch was
firm as he led me to the floor. Arm and arm we were clumsily stepping to the
awkward beat of chaos. (Gibbs & Franks 2002: 159)

As the authors go on to comment: ‘…the woman’s dance scenario still reflects her
metaphorical conceptualization of cancer in personified terms as a dancer with
her as cancer’s unwilling partner’ (ibid. 160). Here the image of a powerful dancer
is a way of expressing the sense of external agency when the speaker experiences
loss of control.
In Extract 11 ‘meat market’ and ‘big giant car factory’ are semantically dis-
tinct vehicles but for an identical target domain – the insensitive GP. Here the
GP is constructed as an external agent exerting a negative force on the speaker.
In a study of metaphors in descriptions of seizure experiences Plug et al. (2011)
identify a number of metaphors involving an external agent and relate these to the
conceptual metaphor a seizure involves actions performed by an exter-
nal agent; there is evidence of the speaker experiencing themselves as at the
mercy of an external agent in Extract 11. Although the metaphors express feelings
of the absence of control over treatment options, there is a sense of control in
being able to identify this as a problem. Plug et al. (ibid) also propose a conceptual
metaphor a seizure is a mechanical process in which there is some type of
­disconnection between mind and body – here there appears to be a disconnection
between patient and medical professional. As in Extract 1, metaphors deperson-
ify the patient by describing her/him as something inanimate (bit of ­mutton/car
body). Such metaphors satisfy the communicative purpose of expressing a strong
negative evaluation of a medical professional: while the speaker may not have felt
in control at the time, she/he is able to offer a purposeful and consistently nega-
tive evaluation of a medical practitioner in the discourse setting of a retrospective
interview.
In ‘shoving into wee boxes’ (Extract 13) there is a similar presence of an
external force; ‘shoving’ is extended to ‘knocking down wee walls’: ‘boxes’ and
‘walls’ are both containers, so there is some semantic relatedness of the vehi-
cles. However, the ‘container’ is then elaborated into a ‘circus’ with the image
of a trapeze artist’s safety net; this conceptual blending of two input domains
seems to be what is meant by the emergence of metaphor in Cameron’s discourse
 Jonathan Charteris-Black

­ ynamics approach. Conceptually, the blend works because when the containers
d
are knocked down they create disorder (‘a big jumble’). This then triggers a sense
of danger that in turn triggers an image of protection from the danger in the form
of a safety net. This type of conceptual blending characterises complex metaphor
elaboration.
A further example of this occurs in Extract 14 where the semantic fields of
journey and buildings are blended: ‘taking this journey through a meadow to a
glass building’: just as the speaker is trapped in her pain, so the pain itself can be
put away in a separate container. This use of conceptual blends in therapy by visu-
alisation is reminiscent of Gibbs and Franks analysis of novel metaphors. In an
account they use to illustrate the process of recovery there is a blending of emer-
gence from containment with vigorous motion: “I felt like my spirit was able to
sing again and that I had taken off the cloak of disease – that I had been carrying
this heavy cloak of disease for about six months and that in dancing I had taken
it off and my spirit was singing again” (Gibbs & Franks 2002:  160). ­Conceptual
blending is therefore a form of verbal control in which mixed metaphors have
the potential to be therapeutic and confirm their claim that ­‘Metaphors help
people make sense of illness by integrating the idea of illness into their own
lives and by offering an answer to the existential question, “Why me?” (Gibbs &
Franks 2002: 160).
For example, consider Extract 15:

Extract 15
….. you might have felt good in the part you went into get fixed but the pain
seemed to jump to somewhere else, well if I’m not getting in that door I’m going to
somebody’s else’s door to chap to get in, so it was a case of it was like chasing, it was
like a circle, it was like the wee hamster in the circle, you know, my elbow’s sore, fix
my elbow, my neck’s sore, fix my neck, my bum’s sore, fix my bum, my knees sore,
so you’re it seemed to move. It is like one of the advertisements I saw on the TV
about this wee thing that moved round about your body, well if am not going to get
there I’m going there, you know so. (CP04, female, 50, care officer)

Here the concept of pain as moving is first introduced with ‘jump’; this is then
extended to ‘chasing’, and elaborated to ‘the wee hamster in the circle’. The descrip-
tion of pain blends the concepts of ‘movement’ and ‘small animals’ on the grounds
that a small animal, such as a hamster or mouse, prototypically moves around
in something circular (a wheel). Interestingly, the speaker then refers to a visual
image of TV advertisement suggesting that emergence and blending can be moti-
vated by multi-modality (Forceville 2008). This is interesting because, as we saw
in Extract 14, it implies the potential for elaborate metaphor in the use of pain
 Chapter 8.  The ‘dull roar’ and the ‘burning barbed wire pantyhose’ 

management therapy. There are other instances in the interviews where complex
metaphor is used in descriptions of therapy:
Extract 16
if the pain feels really hot [um] sort of closing your eyes and thinking about
something nice and you know like a nice cool blue veil just sort of slipping over
you to kind of help to ease the pain or sort of seeing yourself in a room [um] in a
nice comfy chair and just sort of floating there and this lovely veil of relaxation and
comfort just kind of slipping over you. [um] Just things like that really. It’s quite
difficult to explain. (CP12, female, 47, secretary)

Here a ‘hot pain’ is elaborated into ‘a nice cool blue veil’ that is further elaborated
so that it is ‘floating’ and ‘slipping over you’. Various aspects of the ‘veil’ are elabo-
rated: first its colour ‘blue’; then various attributes of a veil such as its lightness
in weight and then its function of covering. Metaphor elaboration creates visual
images that fancifully stretch real world knowledge in an emergent metaphor that
may contribute to gaining control over pain. Creative and expressive linguistic
elaboration therefore has the potential for a therapeutic communicative purpose.
Extract 17 shows how extension and elaboration is typical of therapeutic
methods for pain management:
Extract 17
The process of talking about it, what used to happen to me was is that it was like
being in a room with a door in the corner, and all of a sudden this door would just
slightly crack open and you’d see the answer on the other side, but you couldn’t tell
somebody what the answer was, and you couldn’t describe what the answer was, but
it had given you that little tiny glimmer into what was going wrong and that made
you that little bit stronger to adapt your behaviour ….
 (CP27, male, 54, senior manager and civil engineer)

Here the image of a door opening is extended by reference to knowledge that when
a door is open, light can then come in (‘glimmer’) and things that were hidden on
the other side of the door become known.
In general, then, we have seen that metaphor extension and elaboration often
occurs at stages in the interviews where the speaker is offering a negative evalu-
ation of their treatment or by describing methods for actually overcoming pain.
As with Gibbs and Franks analysis of the novel ‘dancer’ metaphor both these pur-
poses reflect a degree of rhetorical control – even when describing what is was like
to feel out of control.
The identification of complex metaphors and communicative purposes at
particular stages in an interview also creates opportunities for the analysis of
­systematic metaphor that pervades a whole interview. The interview with CP04
 Jonathan Charteris-Black

was 19,239 words in length and various complex metaphors comprised of meta-
phor vehicles from the semantic field of animals contribute to a coherent rhetori-
cal purpose that permeates the interview. Some of these complex metaphors have
already been analysed above in terms of repetition (Extract 2) and elaboration
(Extracts 9,10,11 & 13). However, they are repeated here to demonstrate how they
are distributed throughout the interview:

Table 8.2.  Complex use of metaphor in CP04 (female, 50, care officer)
Word location Complex use of metaphor

5,265 Well taking medication for me keeps my pain down to a dull roar
8757 you know what I mean, you weren’t a person, you were just this bit of
mutton, sitting in this chair, and they could do what they liked with you,
they could use you as a guinea pig
9,466 so it was a case of it was like chasing, it was like a circle, it was like the wee
hamster in the circle,…It is like one of the advertisements I saw on the TV
about this wee thing that moved round about your body,
11,787 say to yourself right ok I’m not going to be a little mouse anymore,
12,674 so I was biting the head off him
17,587 you either put up with the sweats and keep your pain down to a roar

There are various explanations of the recurrence of metaphor vehicles relating to

animals: one possibility is that the first animal metaphor makes the introduction
of further animal metaphors more likely because of priming (see Hoey 2005 for
account of lexical priming). This explanation argues that once the frame of an
animal is introduced this primes the discourse in a certain way: the speaker as a
passive animal (‘bit of mutton’, ‘guinea pig’) but also potentially as an active one
(notice the transition from a ‘little mouse’ to ‘biting the head off him’). Building
on Cameron’s discourse dynamics approach, a related explanation is offered by
Ritchie (2010: 66):
Since the simulations activated by a particularly expressive metaphor may remain
activated for some time, if subsequent metaphors activate similar or compatible
simulations the cumulative effect may be distinct from what could be accomplished
by any one metaphor on its own, and may also be more enduring. Conversely, and
consistent with Cameron’s approach ….. the simulations activated by a previously
used metaphor may be expanded and connected with entirely different topics
through the artful repetition or transformation of a metaphor or narrative.

In the case here the emergence of a metaphor theme enhances the credibility of
the lived experience of chronic pain experience – one that incorporates, physical,
emotional, psychological and social aspects.
 Chapter 8.  The ‘dull roar’ and the ‘burning barbed wire pantyhose’ 

8.7  Conclusion

In retrospective interview accounts of chronic pain, metaphors are mixed when

the speaker has the rhetorical purpose of expressing feelings of loss of control
over the pain. The semantic clashing implied by mixed metaphors symbolises the
loss of psychological control over the condition. When informants are evaluating
aspects of their treatment or otherwise distancing themselves from the topic by
taking a more ‘objective’ view of pain with the purpose of expressing feelings of
control over it they use repeated, extended or elaborated metaphors. Similarly,
where the rhetorical purpose is to communicate other means of gaining control
over pain, such as a therapy or the effect of a medication, speakers tend to use
repeated metaphors or elaborate and extended metaphors based on some type of
conceptual blending. Such emergent metaphors symbolise the gaining of psycho-
logical control over the condition.
The framing of diverse narrative components with complex metaphors pro-
vides a linguistic explanation of what Frank (1995) refers to as chaos and res-
titution narratives – with familiar schema contributing to the coherence of the
account and its persuasive potential. Because complex uses of metaphor can
be interpreted as compliant with underlying rhetorical goals – irrespective of
whether they are repeated, elaborated or mixed – they can be described as ‘pur-
poseful’ metaphors.

References

Barsalou, L. W. (2008). Grounded Cognition. Annual Review of Psychology, 59, 617–645.

doi: 10.1146/annurev.psych.59.103006.093639
Cameron, L. (2010). The discourse dynamics framework for metaphor. In L. Cameron &
R. Maslen (Eds.), Metaphor analysis (pp. 77–94). London: Equinox.
Charteris-Black, J. (2004). Corpus approaches to critical metaphor analysis. Basingstoke & New
York: Palgrave-MacMillan.  doi: 10.1057/9780230000612
Charteris-Black, J. (2012a). Forensic deliberations on ‘purposeful metaphor. Metaphor and the
Social World, 2, 1–21.  doi: 10.1075/msw.2.1.01cha
Charteris-Black, J. (2012b). Shattering the bell jar: Metaphor, gender and depression. Meta-
phor & Symbol, 27, 199–216.  doi: 10.1080/10926488.2012.665796
Fauconnier, G., & Turner, M. (2002). The way we think: Conceptual blending and the mind’s hid-
den complexes. New York: Basic Books.
Forceville, C. (2008). Metaphor in pictures and multimodal representations. In R. W. Gibbs, Jr.
(Ed.), The Cambridge handbook of metaphor and thought (pp. 462–482). Cambridge:
­Cambridge University Press.  doi: 10.1017/CBO9780511816802.028
Frank, A. W. (1995). The wounded storyteller: Body, illness, and ethics. Chicago & London:
­University of Chicago Press.  doi: 10.7208/chicago/9780226260037.001.0001
 Jonathan Charteris-Black

Gallese, V. (2009). Mirror Neurons, embodied simulation, and the Neural basis of social identi-
fication. Psychoanalytic Dialogues, 19, 519–536.  doi: 10.1080/10481880903231910
Gibbs, R. W. J. (2006a). Embodiment and cognitive science. New York: Cambridge University
Press.
Gibbs, R. W. J. (2006b). Metaphor interpretation as embodied simulation. Mind & Language, 21,
434–458.  doi: 10.1111/j.1468–0017.2006.00285.x
Gibbs, R. W. J. (2011). Are deliberate metaphors really deliberate? A question of human con-
sciousness and action. Metaphor and the Social World, 1(2011), 26–52.
doi: 10.1075/msw.1.1.03gib
Gibbs, R. W. J. (2015). Does deliberate metaphor theory have a future? Journal of pragmat-
ics.  doi: 10.1016/j.pragma.2015.03.016
Gibbs R. W. J, & Franks, H. (2002). Embodied metaphor in women’s narratives about their experi-
ences with cancer. Health Communication, 14(2), 139–165.  doi: 10.1207/S15327027HC1402_1
Gibbs, R. W., & Colston, H. (2012). Interpreting figurative meaning. New York: Cambridge
­University Press.  doi: 10.1017/CBO9781139168779
Goatly, A. (2011). The language of metaphors. 2nd ed. London & New York: Routledge.
Hoey, M. (2005). Lexical priming: A new theory of words and language. Oxford & New York:
Routledge.
Levitt, H., Korman, Y., & Angus, L. (2000). A metaphor analysis in treatments of depression:
metaphor as a marker of change. Counselling Psychology Quarterly, 13, 23–35.
doi: 10.1080/09515070050011042
McMullen, L. M., & Conway, J. B. (2002). Conventional metaphors for depression. In S. R. Fus-
sell (Ed.), The verbal communication of emotions. Mahwah New Jersey: Lawrence Erlbaum.
Osaka, N., Osaka, M., Morishita, M., Kondo, H., & Fukuyama, H. A. (2004). A word expressing
affective pain activates the anterior cingulate cortex in the human brain: An fMRI study.
Behavioural Brain Research, 153, 123–127.  doi: 10.1016/j.bbr.2003.11.013
Plug, L., Sharrack, B., & Reuber, M. (2011). Metaphors in the description of seizure experiences:
Common expressions and differential diagnosis. Language and Cognition, 3, 209–233.
doi: 10.1515/LANGCOG.2011.008
Pragglejaz Group. (2007). MIP: A method for identifying metaphorically used words in dis-
course. Metaphor and Symbol, 22, 1–39.  doi: 10.1080/10926480709336752
Ritchie, L. D. (2010). Between mind and language: ‘A journey worth taking’. In L. Cameron &
R. Maslen (Eds.), Metaphor analysis (pp. 57–76). London: Equinox.
Semino, E. (2010). Descriptions of pain, metaphor, and embodied simulation. Metaphor and
Symbol, 25, 205–226.  doi: 10.1080/10926488.2010.510926
Shen, Y., & Balaban, N. (1999). Metaphorical (In)coherence in discourse. Discourse Processes,
28, 139–153.  doi: 10.1080/01638539909545077
Steen, G. (2008). The Paradox of metaphor: Why we need a three-dimensional model of meta-
phor. Metaphor and Symbol, 23, 213–241.  doi: 10.1080/10926480802426753
Steen, G. (2011). The contemporary theory of metaphor – now new and improved! Review of
Cognitive Linguistics, 9, 26–64.  doi: 10.1075/rcl.9.1.03ste
Steen, G. (2013). Deliberate metaphor affords conscious metaphorical cognition. Journal of
­Cognitive Semiotics, 5, 179–197.
part iii

Effects of Mixing Metaphor

 chapter 9

We drink with our eyes first

The web of sensory perceptions, aesthetic experiences
and mixed imagery in wine reviews

Carita Paradis & Charlotte Hommerberg

Lund University / Linnaeus University

This chapter analyzes the language resources that writers have at their disposal to
describe their experience of the web of sensory perceptions that are evoked in the
wine tasting practice. The task of the writer is to provide a mental understanding
of the sensations as well as a prehension of the experiences. We show that this
involves the weaving together of the senses, starting with the sight of the wine,
followed by a description that is iconic with the wine tasting procedure. The
descriptors are systematically used cross-modally both through ontological
cross‑overs and through longer stretches of mixed imagery. We also show how
the socio-cultural context of wine consumption correlates with the types of
imagery used in wine descriptions.

9.1  Introduction

The power of language to evoke vivid sensory imagery is an important prerequisite

for successful communication about topics dealing with both facts and fiction.
This power has been exploited and celebrated by poets, advertisers, politicians,
educators and many others. There are topics that are more challenging than others
when it comes to the means of expression that speakers have at their disposal to
make themselves understood. One such topic is the description and evaluation of
sensory perceptions. This chapter addresses the question of how descriptions and
assessments of the complexities of sensory perceptions are accomplished. While
we focus on the wine reviewing genre, occasional instances of wine descriptions

doi 10.1075/milcc.6.09par
© 2016 John Benjamins Publishing Company
 Carita Paradis & Charlotte Hommerberg

from literature as well as in marketing texts are also included.1 At first blush, such
descriptions come across as “wonderful, chaotic, creative, heroic, challenging”
(Gibbs 2010, p. 1), but on closer inspection such descriptions are also very system-
atic and orderly in their communication of multiple layers of meanings.
Wine reviews are short texts, primarily aimed at consumers, written by wine
journalists and/or connoisseurs. It is the reviewer’s professional task to be able to
convey the representation of the perceptual landscape in a way that is understand-
able and appealing to the reader’s sensory system. The reviewing task presupposes
considerable knowledge in several different domains, related to both the produc-
tion and the consumption side of wine. In addition, it also presupposes extraor-
dinary perceptive talents for the identification of the niceties of the four different
types of sensory perceptions involved in the wine tasting practice, i.e. vision,
smell, touch, taste, and the combined impression of the perceptions resulting
in an aesthetic response to the entire experience. Last but not least, the ability to
communicate this experience through language is of utmost importance for the
reviewers’ credibility among consumers and consequently their professional suc-
cess (Hommerberg 2011; Hommerberg & Paradis 2014).
The tasting practice is the sine qua non for all wine review writing, starting
from the reviewer’s inspection of the wine’s appearance in the glass, through the
nose and the mouth, and finally into the gullet. The appearance of the wine is
the only sensory perception that can possibly be identified in isolation. However,
visual appearance has been shown to be of crucial importance for descriptions
of olfactory and gustatory experiences of wines too (Morrot et al. 2001). Already
when the glass is agitated to release the smells of the wine, we are under the influ-
ence of its visual properties. After that we swill the wine around the mouth and
breathe in air so that we optimize the experience of the taste and the texture of the
wine. The situational context and the specific practices of wine tasting pave the way
for mixed descriptions both at the level of the individual senses and more holisti-
cally. For instance, in the case of individual sense descriptions, as in a sweet nose
of earth, a notion of taste, sweet, is used to describe smell, and in long taste the
notion of length, long, is used as a descriptor of taste. At the text and discourse
level a range of ontological domains combine and plait together the ­perceptions in
a stream of sensory experiences in the text as a whole, as shown in (1). Italics are
added to highlight the domain mixtures.

.  It should be noted that our definition of the notion of ‘wine review genre’ is delimited
to those instances of winespeak that have the purpose of providing consumption advice. In-
stances of winespeak from for instance the literature and marketing are thus not seen as being
part of the wine review genre. See Hommerberg (2011) for a discussion of the notions of
genre and register in relation to winespeak.
 Chapter 9.  We drink with our eyes first 

(1) There is something about these kitchen sink red blends that just gets me
excited. This offering from Ben Marco is terrific. It is opaque and pitch black
colored. It opens with a fragrant black raspberry and black licorice bouquet.
On the palate, this wine is full bodied, balanced, rich, and very fruit forward.
The flavor profile is a delicious boysenberry and vanilla oak blend with notes
of blackberry with a hint of blueberry. The finish is dry and its moderate
dusty tannins are nicely prolonged. This tasty fruit bomb is a wonderful wine.
Grill up a big stack of messy ribs and enjoy it with this gem. Enjoy – Ken2

In this example, the visual descriptors include notions of wine transparency, color
and pitch. Its smell is described through words denoting objects, raspberry and
licorice, and the color term black, and its taste through berries, spices and wood,
with notes of and hints of other berries, suggesting musical overtones. The finish,
or the aftertaste, is what is left on the palate after the wine has been swallowed. It
is described through a notion of touch (dry) and substances (tannins) that have
a tactile effect in the mouth, here also described to be both subdued by dust and
presented to be moving along an unbounded scalar dimension of length.
Two things make the description of wine particularly challenging. Firstly, as in
sensory descriptions in general, there is an alleged paucity of sensory vocabulary,
in particular in the domain of smell (Engen 1982; Sweetser 1990; Holz 2007;
Vanhove 2008; Burenhult & Majid 2011; Majid & Burenhult 2014), and although
all sensory perceptions are important in wine assessment, smell is of particular
importance, since if you block your nose, you cannot taste the wine. Secondly,
the very transformation of perceptions into cognition calls for exploitations of
form-meaning resources that can communicate the ineffable. Wine reviewers
have to make use of the conceptual domain matrices evoked by words denoting
concrete objects and situations in the world in order to be able to straddle the
gap between perception and cognition. The elusive mixture of sensations in wine
tasting takes the form of mixed metonymizations, metaphors and similes in wine
reviewing discourse. Such construals of meaning are taken to be motivated by
the fact that concrete word meanings elicit qualitatively different processing in
the form of mental imagery than abstract word meanings in that they evoke
rich sensory experiences which are intimately tied up with our experiences in
life (Huang et al. 2010). E ­ xpressions evoking a mixture of ontological sources,
construed through synaesthetic ­metonymizations (zone activations) of specific
conceptual ­dimensions or through metaphors and similes are essential for the
transfer of ­sensory ­perceptions into text and discourse (Caballero & Suárez Toste
2010; Paradis 2010; Caballero & Paradis 2013; 2015).

.  http://www.kenswineguide.com/ (accessed on 4 November 2011)

 Carita Paradis & Charlotte Hommerberg

The contention of this chapter is that when we experience comestibles or bev-

erages such as wine, our entire sensorium is activated at more or less the same
time along with our aesthetic and emotional responses. Portrayals of sensory per-
ceptions in wine reviews are mixed because they are affected by the fact that the
sensory perceptions themselves merge into a mixture. Section 2 provides a short
account of the tasting event in relation to the textual structure of wine reviews.
Section 3 relates how the stream of sensory experiences of the wines are described
using terminological descriptors through properties and objects, followed by a
description of metaphors and similes and their role in the representation of sen-
sory and aesthetic experiences in Section 4. In Section 5, the focus is on the discur-
sive role of mixed imagery in winespeak and how the socio-cultural dimension of
consumption also affects the way that mixed imagery is drawn on in wine reviews.
Section 6 concludes the chapter.

9.2  Describing and evaluating sensory experience

Wine reviews normally deal with one or more of three different events: the pro-
duction of the wine, the description/assessment of the wine tasting experience and
a reference to the future consumption, for instance the wine’s ideal drink time.
This is illustrated in (2) by means of a review from Robert Parker’s wine magazine
The Wine Advocate.3
(2) While this is a strong effort from a property that too often does not live
up to its pedigree, I had hoped the 2005 Beychevelle would merit an even
higher score. A deep ruby/purple hue is accompanied by a sweet perfume of
roasted herbs, black cherries, and even blacker fruits. The wine is medium
to full-bodied with sweet tannin, good acidity, and a fruitcake-like spiciness
and earthiness. Pure and long with a tannic clout that is neither intrusive
nor excessive, this elegant, powerful effort should be at its finest between
2017–2030.

The three different events that are referenced in (2) are fundamentally distinct
in terms of time and space frames, source of evidence and mode of knowing
­(Hommerberg & Paradis 2014). The first sentence provides information about
the background and the production of the wine, followed by the description of
the sensory experience of the actual tasting event, where the source of evidence
is first-hand information about the reviewer’s direct visual, olfactory, gustatory
and tactile perceptions. The very last piece of information offers an assessment

.  https://www.erobertparker.com/entrance.aspx.
 Chapter 9.  We drink with our eyes first 

and a recommendation of the future consumption event.4 The reviewer’s source

of evidence when referring to ideal future drinking time is an intricate mixture of
­inferences based on the reviewer’s background knowledge and wine tasting expe-
rience as well as on the perceptual experience of the wine reviewed.
The central part of the review text is the depiction of the tasting event, and
many reviews include only this part, as illustrated by Jancis Robinson’s text in (3): 5
(3) Bright crimson. Extremely sweet and ripe – almost New World – with some
floral aspects. This one is lively and flirtatious with some pretty dry sandy
tannins underneath. Rather unusual. Could do with just a tad more acidity
to lift it. Just a bit confected? Very brutal finish.

As noted in previous investigations of wine discourse, the presentations of the

tasting event are typically iconic descriptions of the tasting practice, i.e. wine
reviewers’ descriptions of the tasting event mirror the journey of the wine from
the glass through the nose and the mouth and finally into the gullet or the spit-
toon ­(Silverstein 2003; Herdenstam 2004; Caballero 2007; Hommerberg 2011;
­Caballero  & Paradis 2013). The iconic representation of the tasting procedure,
which is a typical feature of wine reviews, is shown in Table 9.1, where the reviews
written by Robert Parker and Jancis Robinson, represented as (2) and (3), have been
broken down in accordance with the sensory perceptions that the text refers to:

Table 9.1.  Breakdown of wine reviews (2) and (3) into three perceptual domain
­descriptions6
Robert Parker Jancis Robinson

vision A deep ruby/purple hue is Bright crimson.

accompanied by
smell a sweet perfume of roasted herbs, black
cherries, and even blacker fruits.
taste and The wine is medium to full‑bodied
touch with sweet tannin, good acidity, and a
(mouthfeel) fruitcake-like spiciness and earthiness.
Pure and long with a tannic clout that
is neither intrusive nor excessive,
Extremely sweet and ripe – almost
New World – with some floral aspects.
This one is lively and flirtatious
with some pretty dry sandy tannins
underneath. Rather unusual. Could do
with just a tad more acidity to lift it.
Just a bit confected? Very brutal finish.

.  For more details about recommendations in wine reviewing, see Paradis (2009a, 2009b)
and Hommerberg (2011).
.  http://www.jancisrobinson.com/.
.  We sincerely thank Ms Jancis Robinson and Mr Robert Parker for giving us access to their
wine reviews for the purpose of this study.
 Carita Paradis & Charlotte Hommerberg

While this representation may seem simple and straightforward when we read
the wine texts, the tasting procedure involves highly complex interactions of sen-
sory perceptions which are related to the requisites and limitations of the human
senses. In the wine tasting situation, the senses are ordered hierarchically so that
one can smell the wine without tasting and feeling it, but one cannot experience
the taste and mouth-feel without simultaneously smelling the wine. The visual
experience is in a superordinate position compared to all the other senses, since
the color of the wine can be observed without interference of other sensory input.
Physiologically, vision is also known to be our most consistent source of objective
data about the world. Herdenstam (2004: 60) points out that as much as one third
of the brain is occupied by the interpretation of visual information, while only 1%
of the capacity of the brain is dedicated to smell, and the senses of smell and also
of taste are associated with much more subjectivity than vision. Smell is known to
appeal to emotions, but to simultaneously be an elusive phenomenon from a cog-
nitive point of view (Classen et al. 1994: 2–3). Zucco (2007: 161) notes that com-
munication among humans about olfactory perception is complicated by the fact
that humans are conscious of smells only when they are present: it is not possible
to retrieve olfactory stimuli from memory, since olfactory representations are not
conceptual, merely perceptual. This characteristic of the sensory apparatus dates
from primate evolution, when humans began to exchange olfactory perspicacity
for enhanced color vision (Goode 2007: 81).
When interacting with the wine, tasters experience the wine through all their
senses more or less at the same time, not one at a time. The visual inspection is an
exception, since it takes place prior to the entirely individual internal bodily expe-
rience of the taster. The experiences are processed in the brain and eventually sup-
porting and contributing to the experience that has to be described through the
language in the reviews. As clarified by Table 9.1, the references to vision and smell
are more or less clearly distinguishable. In Parker’s as well as Robinson’s review, the
descriptions of the gustatory and tactile properties of the wine (taste and touch)
are however intertwined, presumably because at the third stage of the tasting pro-
cedure, when the wine is in the mouth, the different sensory inputs are difficult to
tease apart. The term palate is in fact often used to refer to both of those sensory
domains.

9.3  Sensory descriptors

The role of words and expressions in winespeak as in human communication

­generally is to trigger the activation of conceptual structures, constrain their
application in accordance with the current context and to activate sensory and
 Chapter 9.  We drink with our eyes first 

kinesthetic experiences. Whereas activations of such experiences are taken to be

of crucial importance for symbolization more generally (Oakley 2009: 125), they
play an even more central role in descriptions of vision, smell, taste and mouthfeel
in wine reviews. It has been shown in the linguistics literature that descriptions
of perceptions are characterized by synesthesia (Viberg 1984: 136; Sweetser 1990;
Shen 1997). Olfactory experiences are described in terms of things and events
that we perceive through our eyes (Lehrer 1975; Morrot et al. 2001; Popova 2003,
2005; Plümacher & Holz 2007; Paradis & Eeg Olofsson 2013; Paradis 2015b). In
the psychological literature, there is evidence that suggests that verbal descriptions
are not essential, or even necessarily activated, for successful odor-guided cogni-
tion. For instance, Parr, Heatherbell and White (2002) show that whereas olfactory
perceptual skill is critical to wine expertise, verbal skill such as forced naming of
a perceived odor and/or matching terms may in fact interfere with olfactory per-
formance in some situations. Similarly, Herdenstam (2004:79) observes that when
the wine taster concentrates on the component parts in order to give an analytic
(decomposed) description of the wine tasting experience, the synthetic aspect (the
unity) of the experience is eclipsed. Many descriptors of the smell of wine express
everyday things that most readers can relate to and have past experiences of, e.g.
fruit (apple, lemon), herbs and spices (vanilla, nutmeg), flowers and plants (violet,
cedar), sweets (chocolate, jam), beverages (coffee, tea) and minerals (chalk, earth).
Common descriptors also relate to body parts of human beings (body, backbone,
nose) and to people’s personalities and behavior, such as masculine, shy, intellectual
and voluptuous (Suárez 2007; Caballero 2007).
Paradis and Eeg Olofsson (2013) identify two main types of more termino-
logical descriptions of the sensory perceptions. The first type of description is
through words for properties along dimensions in the different sensory modali-
ties as well as properties of objects, in which case the objects are named. They
list some common and typical descriptors for the visual experiences, using col-
locates of color in their data base, for the olfactory experiences, using aroma, and
finally for gustatory and tactile experiences using palate. The results are shown
in Table 9.2.
As Table 9.2 shows, the descriptors of color are basically of two types: conven-
tionalized color terms, such as black and crimson, which are descriptors relating
to visual assessments, and descriptors such as dark, deep and soft, which are used
across the sensory perceptions. The descriptors that combine with aroma/s are
mainly objects of different kind, such as apricot, spicebox, apple, but also deriva-
tions such as animal-like and cassis-scented and again cross-modal property words
such as deep and thin. Interestingly, color descriptors can in fact also be found to
explicitly modify aromas, as in (3), although such combinations are very rare in
the database.
 Carita Paradis & Charlotte Hommerberg

Table 9.2.  List of examples of different descriptors of color, aroma/s and palate
color aroma/s palate
black, blue, amber, crimson, apricot, earthy, floral, game- austere, big, chewy, dense, dry,
garnet, deep-ruby, green, like, oaky, Oriental, musty, deep, fat, pure, rich, ripe, supple,
purple, plum, red, white spice-box, perfumed, almond, sweet, long
… apple, blackberry, rose, nut, …
dark, deep, soft, solid, peach textured, creamy-textured,
shallow, bright, dense, … silken-textured, concentrated,
brilliant, full, strong, weak, animal-like, caramel-infused, multi-dimensional, sustained,
young, thick chocolate-drenched, cassis- oily
… scented …
…
deep, dusty, focused, full,
huge, expansive, thin, tight
…

(3) The Abruzzo might seem rather far south for Chardonnay, but the
­mountains of the interior cool down temperatures during the evening
and night, and the 2001 Chardonnay Marina Cvetic, in addition to its ripe
lemon and white aromas and subtle oak spices, manages to combine a tonic
acidity to the volume and viscosity of the flavors.

Furthermore, among the descriptors of palate in Table 9.2, both cross-modal

terms such as dry and deep and modality specific terms such as textured and oily
are frequently used. It deserves to be pointed out that both color, aroma/s and pal-
ate are modified by descriptors that evoke properties; only aroma and aromas are
modified by object descriptors.7 However, although most of the object descriptors
are mainly employed to describe olfactory characteristics, it is important to note
that these objects also provide visual as well as gustatory and tactile information.
The object descriptors represent a mixture of different sources, from various dif-
ferent spheres such as fruit, flowers and plants, herbs and spices, sweets,
­beverages, and minerals. These concrete objects are used to evoke contingent
properties that the objects produce or properties that are typical of them. They
are also the kind of objects that form part of most wine descriptions and termi-
nologies, such as the Aroma Wheel (Noble et al. 1984), which was developed by
oenologists at the University of California at Davis for descriptions of smell. The

.  Paradis and Eeg-Olofsson (2013) and Paradis (2015b) offer a semantic analysis of the
cross-modal descriptors for properties and objects, construed through synesthetic meton-
ymizations. They argue for a monosemy view of the meanings of those descriptors, which
means that, while the type of construal is one of metonymization (a salience phenomenon),
they are not metonyms proper but rather zone activations within senses (Paradis 2004/2011).
 Chapter 9.  We drink with our eyes first 

Aroma Wheel has been further developed for both whites and reds and for taste as
well by the German Wine Institute.
As already mentioned, smell, taste and touch, unlike visual experience, are
perceived effects of the wine. They are more strongly tied to the experiencer and less
autonomous (Dubois 2007: 173–175). In a study concerned with the categorization
of odors, Dubois (2007) notes that there is no prior categorization to build on, and
what is more, olfactory sensations do not have names, at least not in Indo-European
languages. It should, however, be noted that the differences across cultures may
be greater than we think due to the fact that very little research on these things
has been carried out in cultures other than Western cultures (Classen 1993;
Howes 2004, 2011; Majid & Levinson 2011; Díaz Vera & Caballero 2013; Majid
& Burenhult 2014; Caballero & Paradis 2015; Paradis 2015a;). Dubois reports
on an identification experiment of 16 familiar odors across 40 participants. The
experiment shows that the majority of the responses to the olfactory test items by
the participants include the name of the source of the odor, such as lemon, orange
and apple. On a more specific level, the participants produce specifications such
as sweet lemon, green apple or use a name of another artifact such as lemon drops or
apple shampoo. This is also what we encounter in wine descriptions in wine reviews.

9.4  Sensory and kinesthetic imagery

The second type of description identified by Paradis and Eeg-Olofsson (2013) is

the use of imagery including both metaphorization and similes. These techniques
are secondary in the sense that, while properties and object descriptors are
always present, the descriptions construed as metaphorizations and similes are
not always present. Properties and objects contribute with stative descriptions,
while metaphorizations and similes tend to be more dynamic and kinesthetic.
The expressions of meanings that are dynamic may involve verbs/adjectives or
nouns construed through metaphors and similes in ontological domains such as
animate beings, buildings, machines, malleable entities, and e­ xplosive
artifacts.8 The main function of the dynamic expressions is to describe the
experience of taste and touch as well as to give a holistic evaluative description of
the wine. Unlike the object descriptors from the vegetal, chemical or geological
spheres, discussed in Section 3, the metaphors and similes are both more
dynamic and also more clearly associated with an evaluative element. They lack

.  Note that the vast majority of potentially dynamic meanings, i.e. verbs, do not at all
express dynamic meanings but stative meanings such as ‘being’ and ‘possessing’.
 Carita Paradis & Charlotte Hommerberg

the terminological basis that characterizes the previously described synthetic

metonymizations. An indication of this is that metaphorical descriptions are not
found in Aroma Wheels or other analysis schemas but are more idiosyncratic
across wine critics and perhaps they can be said to be one of the distinguishing-
marks of their writing styles.
Examples (4), (5), (6) and (7) from Wine Advocate all contain expressions of
imagery.
(4) In the past I have described certain wines as being like ballerinas, well-
known actresses, football players, etc. These are rugby players, strong, rough
(at first), but explosive, with considerable stamina (staying power)
(5) If tasting [X] was like swallowing and electric eel, this is like getting hooked
up to a generator.
(6) The 1996 Chateauneuf du Pape smells like an old hippy haven with its
­incense, smoky, roasted herbs, and fleshy, overripe black cherry fruit.
(7) … this blockbuster reminds me of Mohammed Ali – “It floats like a butterfly
and stings like a bee.” It is majestic, large-scaled, and undoubtedly a future
legend.

All the above examples are tokens of the critic’s efforts to be both crystal clear and
perhaps also entertaining. In (4) Robert Parker is making a metadiscursive con-
trast on his own writing, pointing out that ballerinas, actresses and football players
are distinct from rugby players. Examples (5) and (6) are both vivid descriptions of
the taste and smell respectively, and (7) gives a forceful description of the wine in
question in which the reviewer concludes with a general evaluative description in
the form of a mixture of metaphor, blockbuster, and three similes in a row, reminds
me of Mohammed Ali and floats like a butterfly and stings like a bee, mixing the
notion of blockbusters, boxers, with fragile and elegant creatures such a butterflies
and bees, which are clearly more potent, audible and punctual than the slow, sail-
ing movements of the butterfly. The critic, in this case Robert Parker, makes his
presence explicit by using the word me, adding a personal meditative, emotional
touch to the simile.
Caballero (2007) reports on the use of manner-of-motion verbs in wine
reviews when exploring them from the point of view of figurative language in wine
discourse. She reports that these descriptions are used to provide general evalua-
tions as well as information about the smell and the taste of wines in a dynamic,
rather than in a static way, as shown in examples (8), (9), (10), and (11) (examples
(11), (33), (35) and (20) respectively from Caballero (2007).
(8) [This wine] kicks off with the purest scent of smashed berries, and then
­offers up a supercharged black-cherry-laden palate.
 Chapter 9.  We drink with our eyes first 

(9) Here’s a wine that doesn’t slap you silly, but creeps up sideways, with
­seductively soft tannins that carry subtle flavors and blackberries and herbs.
(10) Sturdy, rich and detailed, with complex, earthy currant, sage, mineral
­tobacco and anise flavors that fan out and saturate the palate.
(11) [This wine] pours out beautifully focused pear, quince, honey and spice
­flavors, yet manages to fell elegant and restrained as the flavors sail on
and on.

Caballero notes that the manner-of motion verbs are either used to describe some-
thing that happens very abruptly or something that is durative indicating that the
wine has the potential of lingering for a long time. Examples (9), (10) and (11)
are all dynamic renderings of the experience of the wines; (8) is an example of the
abrupt initiator, while the others are durative. The descriptors that express force-
fulness and abruptness are often found early on in the wine review, while descrip-
tors that express duration and persistence are often used in the final evaluation of
the wine.
As suggested by the title of this chapter, vision is an extremely important
source of information in the context of wine description and evaluation. It is the
first stage of the tasting event which provides hints about the smell, taste and touch
and about the status of the wine in terms of quality and age.9 The importance of
vision in the choice of object descriptors for smell in the wine reviews in corpus
data can be explained with reference to an experimental investigation carried out
by Morrot et al. (2001). They set up an investigation of the interaction between
vision and smell assessments in wine description in two steps. The first part
is a lexical analysis of descriptors used in wine tasting comments by a French
wine maker and experts from one English and two French wine tasting guides,
which showed that when the smell of a wine was described, the descriptors
used denoted objects that have the same color as the wine, i.e., dark objects for
red wine and light-colored objects for white wine. The lexical analysis led them
to hypothesize that the existence of synaesthesia of smell and vision in wine
description is psychophysically grounded. The hypothesis was later confirmed by
an experiment, in which the smell of a white wine artificially colored red with an
odorless dye was described by means of descriptors used about red wines by a
panel of 54 professional tasters. Because of the visual disinformation, the olfactory
information went unnoticed by the tasters. According to Morrot et al. (2001),
humans have never developed a specific olfactory terminology to describe odors,

.  Even though the universality and primacy of vision may be questioned, or deserve to be
questioned, it is of particular importance for wine assessment (Howes 2013).
 Carita Paradis & Charlotte Hommerberg

which o­ bviously constitutes a serious challenge for wine critics. Corpus evidence
of wine descriptions from Paradis and Eeg-Olofsson’s investigation also reveals
that the smell descriptors pattern differently in the descriptions of red wines and
white wines, as shown in (12) and (13).
(12) The 2003 Chardonnay Sbragia Limited Release (2,000 cases) is surprisingly
restrained and delicate for this cuvee. Medium to full-bodied, with c­ opious
quantities of buttered popcorn, pineapple, orange blossom, and melon
­characteristics as well as outstanding depth, purity, and balance, it will last
for 2–3 years.

(13) … The dark, saturated ruby color is followed by an unevolved, super-rich

nose of roasted herbs, nuts, black fruits, and Asian spices. Spectacularly rich,
with an unctuous, multi-dimensional flavor profile, and a chewy, robust
­finish, this big wine (14.5% alcohol) exhibits marvelous balance, as well as
the potential to last for 10–15 years.

Example (12) is a description of a white wine and (13) of a red wine. What is
important here is the critic’s choice of descriptors for smell. In the case of smell
descriptors of the white wine (butter, popcorn, pineapple, orange blossom, melon),
the colors of the descriptors are light, while the reverse is true of the descriptors of
the red wine (roasted, nuts, black fruits). The sum of them together is more than
the individual descriptors. The clearly light objects and the clearly dark objects
have an effect of what may not be as clearly colored, e.g., Asian spice.
As has been shown, most of the descriptors span over more than one of the
sensory domains and their use in those different domains does not give rise to
ambiguities or infelicities in language use, which, had that been the case, would
be suggestive of substantial sense distinctions (Paradis & Eeg Olofsson 2013,
­Caballero & Paradis 2013 – also, including architecture). Many of the descriptors
are actually explicitly used for descriptions of more than one modality, e.g.,
soft color, soft smell, soft taste and soft textures and so are the properties of the
objects, e.g., lemon, vanilla, blackberry, which are primarily descriptors of smell,
but in that capacity they also range over the other modalities as shown in (12)
and (13). Most of the descriptors which, for instance, are used in the part of the
text describing smell, are clearly crucial for our understanding of the color, taste
and touch of the wine. Paradis and Eeg-Olofsson’s (2013) and Paradis (2015b)
argue that synesthetically flexible notions map onto the same primitive concepts
for the different sensory perceptions, or put differently, no conceptual primacy
exists in the realm of sensory perceptions. The contention is that it is not the
case that soft smell is primarily a notion of touch. Soft spans the experiences of
sharp of the sensory perceptions of vision, smell, taste and touch. They call
the conceptual preference hierarchy into question and thereby also the primacy
 Chapter 9.  We drink with our eyes first 

of earlier uses of the words as an argument for primacy. Instead, they argue that
the lexical syncretism is grounded in how the conceptualization of our sensorium
works, i.e., we cannot taste something without smelling something and we cannot
taste something without feeling something, and over and above everything, in
wine tasting, is the sight of the wine. Even though vision seems to have special
status in wine tasting, the investigations do not provide support for the notion of
conceptual primacy of one of the meanings as reflected in language. This does not
mean that we fail to acknowledge the physiological differences between the various
specific sensory modalities, but what we do acknowledge in the context of wine
assessment is that, at the conceptual level and at the level of the transformation
of sensory perceptions into conceptual structure and subsequently into language,
there is the flexibility of the uses of the descriptors. Conceptual structure in
the domain of sensory experiences appears to be a supramodal representation,
expressed through syncretic word forms, which do not pertain to a single type
of experience but to an overarching representation capable of capturing modal
convergences and similarity structures that define categories, such as properties of
objects and imagery (Binder & Desai 2011, Paradis 2015b).

9.5  The discursive role of imagery

The past few decades’ exploding interest in wine among new groups of consumers
worldwide has entailed an increasing demand of authoritative consumption advice
in this field. Deference for authority is a natural way of shaping our understanding
when we access any new domain of knowledge, and most of today’s international
consumer groups can be understood to enter the epistemic domain of wine as
adults without previous cultural background to influence their judgment of taste
(Orrigi 2007: 185–187). In response to the globalized wine trend, the writing of
wine reviews has developed into a profession in which people can earn money and
in some cases even make a living (Charters 2007: 157). The wine review is there-
fore an expansive field of discourse, where the talents of the critic are crucial for
successful communication with the readers, and hence for professional success.
The role of imagery in wine reviews is multifaceted, opening up for a mixture
of potential discursive impacts. Imagery has the discursive potential of selecting
and thereby simultaneously concealing parts of the communicated experience,
thus persuading the addressee to see the world in certain particular ways rather
than others. In addition, metaphors can function as covert ways of establishing
communal values without apparently imposing a value system on the addressee.
Simultaneously, the use of imagery can forge interpersonal bonds and contribute
to the construction of both the writer’s and the addressee’s identity, since it draws
 Carita Paradis & Charlotte Hommerberg

on shared associations with past experiences (Charteris-Black 2004: 11–13). An

additional function of imagery, especially in genres such as the wine review, which
are constrained by a limited format, is that it allows several layers of meaning to be
communicated in a condensed linguistic form. It is therefore economical.
Among wine lovers, it is common that the experience of wine is first and fore-
most seen as an aesthetic pleasure similar to the experience of art or music, and
for that reason wine reviews often contain mixtures of imagery that range over the
whole register of aesthetic and emotional responses in the reader. Although an
instance from literature rather than the wine reviewing genre, the following pas-
sage from Evelyn Waugh’s Brideshead Revisited can perhaps be seen as an extreme
illustration of mixed aesthetic imagery in wine description. Framed as the nar-
rator’s nostalgic memory of prewar carefreeness, it describes the two main char-
acters’ aesthetic/emotional response to their experience of tasting three different
wines from an ancestral British wine cellar (Waugh, 2003:81–82):

‘…It is a little, shy wine like a gazelle.’

‘Like a leprechaun.’
‘Dappled, in a tapestry meadow.’
‘Like a flute by still water.’
‘…And this is a wise old wine.’
‘A prophet in a cave.’
‘…And this is a necklace of pearls on a white neck.’
‘Like a swan.’
‘Like the last unicorn.’

The quotation above is taken from a passage in which the two main characters of
Waugh’s novel, Charles Ryder and Sebastian Flyte, are engaged in tasting wine in
Sebastian’s aristocratic home. Under the increasing influence of the wines’ alcohol,
they come up with more and more colorful, poetic depictions of the wines they are
tasting. Although the two gentlemen are thus portrayed as somewhat intoxicated,
it is not impossible to interpret the vivid imagery of their discourse as having some
kind of meaning in relation to the sensory experiences evoked by the wines they
are discussing. For instance, the expressions shy like a gazelle, pearls on a white
neck and the last unicorn may bring to mind characteristics such as reserve, youth,
purity, innocence and chastity. When associated with the target domain of wine,
these characteristics can be taken to refer to a young wine with restrained smell
and taste, which at its present stage of maturity does not overwhelm the senses,
but which holds promises of future sensory and aesthetic delights as it reaches
its full maturity. Leprechaun, when associated with wine, may be understood as
an ­unreliable wine whose smell is captivating, suggesting a hidden treasure, but
which ends up being a disappointment since the taste of the wine does not c­ onfirm
 Chapter 9.  We drink with our eyes first 

the quality that the smell promised. The wise old prophet in the cave conjures a
fully mature wine, which has developed all its olfactory and gustatory potential.
Pearls on a white neck as well as prophet in a cave are conceivably associated with
a positive value, while we can imagine leprechaun to carry negative evaluation,
although this value is only communicated covertly so that it has to be inferred by
the interlocutor. Hommerberg and Don (2015) offer a taxonomy for analyzing the
values expressed in winespeak based on the Appraisal model.
It is noteworthy that Charles and Sebastian employ a combination of meta-
phor and simile in their dialogue. Their strategy of using the explicit marker like
functions to signal that the comparison between source domain and target domain
is not self-evident, but rather incomplete or hypothetical (Low 2010: 295). In the
dialogue above, gazelle, leprechaun, flute, swan and unicorn are presented as mean-
ingful sources of comparison in order to enhance the recipient’s understanding of
the discussed entity, but the simile marker allows the focus of attention to remain
concentrated on the target domain, i.e., the wine. This contrasts with the depictions
of the wines as being shy, wise, a prophet in a cave and a necklace of pearls, where
the recipient’s capacity to construe the comparison is taken for granted rather than
explicated in the text. In this passage, it is interesting to note how metaphors and
similes work in combination to reinforce the interlocutors’ depictions of the wines’
qualities. For instance, the metaphorical description of the wine as being shy is sub-
sequently intensified by the simile which compares the wine’s shyness to that of a
gazelle, an animal which depends on its innate reserve for its survival.
While the mixtures of imagery used by Charles and Sebastian may have the
capacity to invoke certain aesthetic responses and evaluations with respect to the
wines’ qualities, they simultaneously construe the invoked source domains as
being accessible to and meaningful for the two participants, thus confirming their
group identity. The imagery used in the passage positions the novel’s characters
as members of a group for which references to tapestry meadow and pearls on a
white neck evoke comparable past experiences, essentially associated with British
upper-class upbringing, acculturation and education, where the capacity to enjoy
sensory pleasure from consumption experiences was elevated to an art associated
with extreme refinement in true Romanticism spirit. While it should be noted that
Waugh’s novel is written in the aftermaths of the Second World War and sheds
critical rather than idealizing light on the extravagance of the prewar period, the
passage nonetheless serves as an illustration of the socio-cultural role of imagery
in the register of winespeak.
Silverstein (2003), who has studied wine reviews written by the British expert
Michael Broadbent, takes particular note of occurrences of figurative expressions
such as well-bred and gentlemanly. Broadbent’s wine jargon positions the
audience as members of a social group identifying itself with respect to inherited
 Carita Paradis & Charlotte Hommerberg

­reeding and life-long acculturation, while simultaneously excluding those

b
that do not belong to this group. Shesgreen (2003) however, observes that “the
language of social class and gender”, which used to be popular in wine reviews
during the latter half of the 20th century, is more or less out of fashion among
today’s most influential wine critics. Instead, what he refers to as “the language
of fruit and vegetables” (extending into the domain of wine through a process of
metonymization, or more precisely zone activation as discussed by Paradis and
Eeg Olofsson (2013) and Paradis (2015b)) is currently more widespread, at least
among American wine writers.
While confirming Shesgreen’s observation concerning the strong preponder-
ance of fruit and vegetable descriptors, Hommerberg’s (2011) study of American
wine critic Robert Parker’s reviews also reveals patterns of mixed imagery that
seemingly reach beyond portraying the purely perceptual dimension of the wine
tasting experience, which is exemplified in (14) below:
Lush, medium-bodied, and sensual, it will benefit from 1–2 more years of
(14) 
bottle age, and should drink well for 12–14.

The use of lush and sensual in (14) invites association to somewhat imprecise source
domains, leaving the exact interpretation of these items up to the addressee. While
lush suggests abundance or opulence in general, the co-occurrence in this par-
ticular instance with sensual may tend to, from a male, heterosexual perspective,
inspire associations to the characteristics that the items lush and sensual would
refer to in the domain of woman and apply these to the domain of wine. The
potential discursive impact of such imagery is manifold: on the level of the sensory
perception of this particular wine, the items can be taken to communicate a view
of this wine as being unrestrained and accessible in its supply of olfactory and gus-
tatory qualities. Simultaneously, while this is not overtly articulated, the choice of
imagery in (14) also covertly invites the audience to share the evaluative position
that being unrestrained and accessible are desirable qualities in a wine. A further
understanding of the potential meaning-making that these items may give rise to
is to position the writer and audience in the same group as members of a male,
heterosexual discourse community.
For female readers, the corresponding stereotypical reading of the same
items might be that drinking this wine is like being a lush, sensual woman who
is attractive to heterosexual men. While the presentation may thus invoke the
source domain of woman, it is worth noting that the association is only drawn
on implicitly by means of the metaphorical invocation of the source domain.
The construction represents the comparison between source domain and target
domain as accessible to and unproblematic for readers. This makes the metaphor
different from the simile, which specifies the relation between source domain and
 Chapter 9.  We drink with our eyes first 

target domain as one of partial similarity rather than identity, as illustrated below
by the constructed example (14a):
(14a) This medium-bodied wine is like a lush, sensual woman. It will benefit from
1–2 more years of bottle age, and should drink well for 12–14.

The degree to which the expressions lush, sensual in 14 actually do invoke associa-
tions to the source domain of woman may of course differ across readers. For a
wine expert familiar with Parker’s writing, these items may instead be interpreted
as non-figurative, precise descriptors, designating particular sensory impressions
caused by chemical combinations in the wine’s molecules (see also Caballero &
Suarez Toste 2010). However, since it can be expected that the primary target audi-
ence of wine reviews is made up of consumers in general rather than wine experts,
it is probable that the source domain of these figurative expressions is still likely to
color readers’ interpretations to some extent.
Rather than resorting to myth, aristocracy or poetry to find appropriate
source domains for figurative expressions, the associative imagery used by today’s
wine critics frequently draws on the worlds of sports, architecture, business, sex
and personification in general. The mixture of imagery in (15), which is also taken
from a review written by Parker, illustrates how the source domains of sex and
sports car are brought together in order to characterize the aesthetic/emotional
impact of the wine’s smell:
(15) Smelling like a concoction whipped up by a deranged monk who spent
too much time in solitary confinement, it pushes the olfactory senses into
­overdrive with its array of earthy, jammy fruit, and herb scents.

Rather than stating it explicitly, the allusion to a monk in solitary confinement

implies that he thinks of nothing but sex, which leads to a transference of this
obsession to the concoction that he is producing. The expression smelling like a
concoction explicitly signals the partiality of the comparison relation. This con-
trasts with the corresponding metaphorical version, which is constructed for ana-
lytical purposes in (15a):
(15a) This is a concoction whipped up by a deranged monk who spent too much
time in solitary confinement.

The tentative similarity relation construed by the simile construction is distinct from
the association to the source domain of sports car in (15), where the metaphorical
expression pushes the olfactory senses into overdrive establishes a comparison
between the source domain and the target domain without explicitly stating that
the comparison relation is one of similarity. In other words, the comparison to
the source domain of sports car, one of the key symbols stereotypically related
 Carita Paradis & Charlotte Hommerberg

to male potency, is presented as unproblematic for readers to associate with the

domain of wine. It is therefore less noticeable that this comparison is being made.
While the evaluative orientation of this passage is not explicitly stated, both the
simile and the metaphor nonetheless invite a positive reading.
The macho lingo that can be found in some of Parker’s texts resurfaces in
reviews written by other contemporary critics. In (16), which is taken from a
review written by Robinson, the critic employs associations with body-building
drugs to invite the audience to join her in her negative assessment:
(16) …this is a wine on steroids. Where is the gentle refreshment value?

While communicating negative attitude, the reference to steroids simultaneously

identifies the audience as a group for which the connection between the domain
of body-building and the domain of wine is unproblematic. The recipient is
implicitly encouraged to imagine a (male) person with an unnaturally sturdy body
whose muscles are not, or not only, the result of physical activity. Interpretations
may differ among readers, depending on their familiarity with wine discourse. For
habitual members of the discourse community, the expression wine on steroids
may take on a specific meaning, referring to how the wine was produced, suggest-
ing that artificial techniques were employed. This type of imagery in wine reviews
can therefore give rise to a mixture of different possible readings. Again, the meta-
phorical construction suggests a well-established comparison relation, which does
not call for a detour through a simile construction.
Since wine critics’ careers depend on retaining the audience’s interest, mix-
tures of imagery are sometimes drawn on in wine reviews as attention-grabbing
devices to create an element of surprise for entertainment purposes. The following
depictions can be seen as examples of such exaggerated present-day imagery:
(17) …it rumbles like an 18-wheeler over the palate; the finish is like one of
those wild skies dotted with angry clouds but sunny with crepuscular
beams. Oh man, Rieslaner!10
(18) This dark wine…helicopters into the mouth with spinning blades of intense
fruit.11

It takes considerable effort on the part of the prospective reader to interpret the
imagery employed in (17) and (18) so that it becomes meaningful in relation to the

.  The extract presented in 17 is taken from a text written by Terry Theise, an American
importer of German wine, which means that the ultimate purpose is to promote rather than
review the wine. Nonetheless, Theise is also regarded as a reputed authority.
.  Example (18) is from a review written by Andrew Jefford for Financial Times of London.
 Chapter 9.  We drink with our eyes first 

context. While it is not easy to imagine what an 18-wheeler could have in common
with the taste/mouthfeel of a German white wine, it is perhaps possible that the
simile is intended to highlight the wine’s grip as it is perceived by the palate’s
tactile receptors. The potential meaning of the poetic simile involving a dramatic
skyline is even more obscure in relation to the qualities of a Rieslaner’s aftertaste
and evaporation. Perhaps the simile only relates to the emotions evoked by the
stunning scenery occasionally encountered in nature and draws on a comparison
between these emotions and the emotional response evoked by this wine’s finish.
The combination of the two similes in (17) also opens up for further interpre-
tations depending on the reader’s past experiences. Since this text is intended for
promotion on the American wine market, the combination of the two similes may
also conjure up familiar image of a huge truck on an otherwise deserted interstate
highway pictured against the backdrop of a spectacular skyline. The construction
with like in (17) invites readers to interpret the relation between source domains
and target domain as being not quite established. This contrasts with the meta-
phor in (18) which construes the comparison between source domain and target
domain as taken for granted. The helicopter metaphor drawn on in (18) presum-
ably purports to highlight and reinforce the impression of intensity that the wine
gives rise to against the palate (see Section 2).
Both (17) and (18), while perhaps unusually extravagant, can be said to be
typical of today’s wine discourse in that the mixtures of imagery drawn on do not
tend to invoke myth, inherited breeding and upper-class education. This type of
descriptions defines both wine writers and prospective readers as members of a
group that is prone to experience wine in terms of present-day phenomena acces-
sible to everyone rather than prestigious heritage suggesting inherited breeding
reserved for the few. In contrast to the metaphors and similes drawn on in Waugh’s
novel, the mixed imagery used in today’s wine discourse relates to a wider audi-
ence without aristocratic roots and upper-class education. This development can
be seen as a consequence of the exploding interest in wine as a status beverage
among ever-growing groups of new consumers world-wide. Importantly, the sug-
gestive, multilayered meaning potential of the mixed imagery used in examples
(14) – (18) above could not have been achieved by means of corresponding literal
expressions.

9.6  Summary

This chapter is concerned with the descriptions and evaluations of the web of
sensory perceptions that arise in the wine tasting practice and which subsequently
are written up and communicated through language to the readers of the
 Carita Paradis & Charlotte Hommerberg

discourse community. The transition from the tasting practice to the discourse
practice involves not only the transition from perception to conception but
also the transition into language. Our particular window into the study of
these complexities is through the language resources made use of in the wine
reviewing genre. We have described how the social context and the purpose of
wine reviewing serve as the motivating mechanisms for the mixture of ontological
domains to enable successful communication. This involves the weaving together
of the descriptions of the different individual senses to an experience that also
activates the whole sensorium.
What we note is that the mixing of ontological structure is pervasive in wine
reviewing. It may be regarded as the technique used in the descriptions at the
discourse level as well as at the level of the meanings of words and constructions.
The terminological analytical descriptions are rather well-established terms for
properties and objects from a mixture of ontological domains, and the holistic
and synthetic descriptions are often more idiosyncratic metaphors and similes
that by definition are construals of comparison across domains in which case
the source domain is a ‘concrete’ conceptual structure and the target domains an
‘abstract’ ditto. We have identified ontological mixtures (and lexical syncretism)
both in the case of terminological expressions of property descriptors such as
soft and sharp, and object descriptors referring to vegetal, chemical and geo-
logical matters that in spite of their being descriptors of smell, range over the
sensory perceptions involved in the tasting event, i.e. smell, taste and touch.
However, in contrast to the more stative and terminological descriptors, there
are different types of imagery where we more clearly deal with comparisons
across domains and cultures as is the case for metaphors and similes such as the
flavors that sail on and on, or the wine that is like an old hippy haven. In contrast
to the terminological descriptors, the source domains of the imagery and the
mixture of imagery used in the wine reviews reflect certain socio-cultural values
and thereby play an important role for the creation of a sense of belonging and
affinity.

References

Binder, J. R., & Desai, R. H. (2011). The neurobiology of semantic memory. Trends in Cognitive
Sciences, 15(11), 527–536.  doi: 10.1016/j.tics.2011.10.001
Burenhult, N., & Majid, A. (2011). Olfaction in Aslian ideology and language. Senses & Society,
6(1), 19–29.  doi: 10.2752/174589311X12893982233597
Gibbs, R. W. (2010). The wonderful, chaotic, creative, heroic, challenging world of researching
and applying metaphor: A celebration of the past and some peeks into the future. In
L. Cameron (Ed.), Metaphor in the real world. Amsterdam: John Benjamins.
doi: 10.1075/hcp.26.01gib
 Chapter 9.  We drink with our eyes first 

Caballero, R. (2007). Manner-of-motion verbs in wine description. Journal of Pragmatics, 39,

2095–2114.  doi: 10.1016/j.pragma.2007.07.005
Caballero, R., & Paradis, C. (2013). Perceptual landscapes from the perspective of cultures
and genres. In R. Caballero & J. Díaz Vera (Eds.), Sensuous cognition – Explorations into
human sentience – Imagination, (e)motion and perception (pp. 77–105). Berlin: de Gruyter
Mouton.  doi: 10.1515/9783110300772
Caballero, R., & Suárez Toste, E. (2010). A genre approach to imagery in winespeak. In G. Low,
Z. Todd, A. Deignan, & L. Cameron (Eds.), Researching and applying metaphor in the real
world (pp. 265–287). Amsterdam: John Benjamins.  doi: 10.1075/hcp.26.15cab
Caballero, R., & Paradis, C. (2015). Making sense of sensory perceptions across languages and
cultures. Functions of Language 22(1), 1–19.  doi: 10.1075/fol.22.1
Charteris-Black, J. (2004). Corpus approaches to critical metaphor analysis. Basingstoke: Palgrave
MacMillan.  doi: 10.1057/9780230000612
Charters, S. (2007). On the evaluation of wine quality. In B. C. Smith (Ed.), Questions of taste.
The philosophy of wine (pp. 157–182). Oxford: Signal Books.
Classen, C. (1993). Worlds of sense. Exploring the senses in history and across cultures. London:
Routledge.
Classen, C., Howes, D., & Synnott, A. (1994). Aroma. The cultural history of smell. London &
New York: Routledge.
Deutsches Weininstitut. Available at www.deutcshesweine.com.
Díaz Vera, J., & Caballero, R. (Eds.). (2013). Sensuous cognition – Explorations into human sen-
tience – Imagination, (e)motion and perception. Berlin: de Gruyter Mouton.
doi: 10.1515/9783110300772
Dubois, D. (2007). From psychophysics to semiophysics: categories as acts of meaning. A case
study from olfaction and audition back to colors. In M. Plümacher & P. Holz (Eds.), Speak-
ing of colors and odors (pp. 167–185). Amsterdam: John Benjamins.
doi: 10.1075/celcr.8.10dub
Engen, T. (1982). The perception of odurs. New York: Academic press.
Goode, J. (2007). Wine and the brain. In B. C. Smith (Ed.), Questions of taste. The philosophy of
wine (pp. 79–98). Oxford: Signal Books.
Herdenstam, A. (2004). Sinnesupplevelsens estetik. Vinprovaren, i gränslandet mellan konsten och
vetenskapen. Stockholm: Dialoger. English title: Experience of an aesthetic sensation. Wine
tasters in the field between art and science.
Holz, P. (2007). Cognition, olfaction and linguistic creativity: Linguistic synesthesia as poetic
device in cologne advertisement. In M. Plümacher & P. Holz (Eds.), Speaking of colors and
odors (pp. 185–202). Amsterdam: John Benjamins.  doi: 10.1075/celcr.8.11hol
Hommerberg, C. (2011). Persuasiveness in the discourse of wine: The rhetoric of Robert Parker.
Växjö: Linnaeus University Press.
Hommerberg, C., & Paradis, C. (2014). Constructing credibility in wine reviews. Evidentiality,
temporality and epistemic control. In D. Glynn & M. Sjölin (Eds.), Subjectivity and Epistemicity.
Corpus, discourse, and literary approaches to stance (pp. 211–238). Lund University Press.
Hommerberg, C., & Don, A. (2015). Appraisal and the language of wine appreciation: a critical
discussion of the potential of the Appraisal framework as a tool to analyse specialised
genres. Functions of Language, 22(2), 161–191.
Howes, D. (Ed.). (2004). Empire of the senses: The sensual culture reader. Oxford: Berg.
Howes, D. (2011). Cultural synaesthesia: Neurological vs. anthropological approaches to the
study of intersensoriality. Intellectica, 55, 156–187.
 Carita Paradis & Charlotte Hommerberg

Howes, D. (2013). Sensous cognition: The language of senses. In R. Caballero & J. Díaz Vera
(Eds.), Sensuous cognition – Explorations into human sentience – Imagination, (e)motion
and perception (pp. 293–298). Berlin: de Gruyter Mouton.
Huang, H-W., Lee, C-L., & Federmeier, K. D. (2010). Imagine that! ERPs provide evidence for
distinct hemispheric contributions to the processing of concrete and abstract concepts.
Neuroimage, 49, 1116–1123.  doi: 10.1016/j.neuroimage.2009.07.031
KensWineGuide. Available at [http://www.kenswineguide.com/]. Date of access 4 November
2011.
Lehrer, A. (1975). Talking about wine. Language, 51(4), 901–923.  doi: 10.2307/412700
Low, G. (2010). Wot no similes? In G. Low, Z. Todd, A. Deignan, & L. Cameron (Eds.), Research-
ing and applying metaphor in the real world. Amsterdam: John Benjamins.
doi: 10.1075/hcp.26
Majid, A., & Levinson, S. C. (2011). The senses in language and culture. Senses and Society, 6(1),
5–18.  doi: 10.2752/174589311X12893982233551
Majid, A., & Burenhult, N. (2014). Odors are expressible in language, as long as you speak the
right language. Cognition, 130(2), 266–270.  doi: 10.1016/j.cognition.2013.11.004
Morrot, G., Brochet, F., & Dubourdieu, D. (2001). The color of odors. Brain and Language, 79,
309–320.  doi: 10.1006/brln.2001.2493
Noble, A., Arnold, R., Buechsenstein, J., Leach, E., Schmidt, J., & Stern, P. (1984). Modification
of a standardized system of wine aroma terminology. American Journal of Enology and
Viticulture, 107–109.
Oakley, T. (2009). From attention to meaning: explorations in semiotics, linguistics, and rhetoric.
Berlin: Peter Lang.
Orrigi, G. (2007). Wine epistemology: The role of reputational and ranking systems in the
world of wine. In B. C. Smith (Ed.), Questions of taste. The philosophy of wine. Oxford
(pp. ­183–197). Signal Books.
Paradis, C. (2004/2011). Where does metonymy stop? Senses, facets and active zones. Metaphor
and symbol, 19(4), 245–264. (Reprinted in P. Hanks & R. Giora (Eds.), Metaphor and Figu-
rative Language: Critical Concepts in Linguistics, Amsterdam: Routledge).
doi: 10.1207/s15327868ms1904_1
Paradis, C. (2009a). This beauty should drink well for 10–12 years: A note on recommendations
as semantic middles. Text & Talk, 29(1), 53–73.  doi: 10.1515/TEXT.2009.003
Paradis, C. (2009b). Prime time: The middle construction in wine drinking recommendations.
In R. Bowen, M. Mobärg, & S. Ohlander, (Eds.), Corpus and discourse – and stuff. Papers
in honour of Karin Aijmer (pp. 243–251). Gothenburg Studies in English 96. Acta Univer-
sitatis Gothoburgensis.
Paradis, C. (2010). Touchdowns in winespeak: Ontologies and contruals in use and meaning-
making. In M. Goded Rambaud & A. Poves Luelmo (Eds.), Proceedings for the 1st Con-
gress on linguistic approaches to food and wine descriptions (pp. 57–72). Madrid: UNED
­University Press.
Paradis, C. (2015a). Meanings of words: Theory and application. In U. Hass & P. Storjohann (Eds.),
Handbuch Wort und Wortschatz (Handbücher Sprachwissen-HSW Band 3) (pp. 274–294).
Berlin: Mouton de Gruyter.  doi: 10.1515/9783110296013-012
Paradis, C. (2015b). Conceptual spaces at work in sensuous cognition: Domains, dimensions
and distances. In F. Zenker & P. Gärdenfors (Eds.), Applications of Conceptual spaces:
Geometric knowledge representation (pp. 33–55). Dordrecht: Springer Verlag.
doi: 10.1007/978-3-319-15021-5
 Chapter 9.  We drink with our eyes first 

Paradis, C., & Eeg-Olofsson, M. (2013). Describing sensory experience. The genre of wine
reviews. Metaphor & Symbol, 28(1), 22–40.  doi: 10.1080/10926488.2013.742838
Parker, R. Available at https://www.erobertparker.com/entrance.aspx
Parr, W., Heatherbell, D., & White, G. (2002). Demystifying wine expertise: Olfactory threshold,
percepetual skill, and semantic memory in expert and novice wine judges. Chemical senses,
27(8), 747–755.  doi: 10.1093/chemse/27.8.747
Plümacher, M., & Holz, P. (Eds.). (2007). Speaking of colors and odors. Amsterdam: John
Benjamins.  doi: 10.1075/celcr.8
Popova, Y. (2003). ‘The fool sees with his nose’: Metaphorical mappings in the sense of smell in
Patrick Süskind’s Perfume. Language and Literature, 12, 135–151.
doi: 10.1177/0963947003012002296
Popova, Y. (2005). Image schemas and verbal synaesthesia. In B. Hampe (Ed.), From percep-
tion to meaning: Image schemas in cognitive linguistics (pp. 1–26). Berlin & New York: de
Gruyter.
Robinson, J. Available at http://www.jancisrobinson.com/
Shen, Y. (1997). Cognitive constraints on poetic figures. Cognitive Linguistics, 8(1), 33–71.
doi: 10.1515/cogl.1997.8.1.33
Shesgreen, S. (2003). Wet dogs and gushing oranges: Winespeak for a new millennium. Chroni-
cle Review, (online), 49. Available at http://chronicle.com/free/v49/i26/26b01501.htm
Silverstein, M. (2003). Indexical order and the dialectics of sociolinguistic life. Langauge and
Communication, 23, 193–229.  doi: 10.1016/S0271-5309(03)00013–2
Sweetser, E. (1990). From etymology to pragmatics: Metaphorical and cultural aspects of semantic
structure. Cambridge: Cambridge University Press.  doi: 10.1017/CBO9780511620904
Suárez Toste, E. (2007). Metaphor inside the wine cellar: On the ubiquity of personification
schemas in winespeak. Metaphoric.de, 12, 53–64.
Vanhove, M. (2008). Semantic associations between sensory modalities, prehension and mental
perceptions: a crosslinguistic perspective. In M. Vanhove (Ed.), From polysemy to semantic
change: Towards a typology of lexical semantic associations (pp. 343–370). Amsterdam: John
Benjamins.  doi: 10.1075/slcs.106
Viberg, Å. (1984). The verbs of perception: A typological study. Linguistics, 21(1), 123–162.
Waugh, E. (2003). Brideshead revisited. London: Penguin.
Zucco, G. M. (2007). The unique nature of a memory system. In M. Plümacher & P. Holz (Eds.),
Speaking of colors and odors (pp. 155–165). Amsterdam: John Benjamins.
doi: 
10.1075/celcr.8.09zuc
 chapter 10

A corpus-based study of ‘mixed metaphor’

as a metalinguistic comment

Elena Semino
Lancaster University, UK

This chapter investigates the use of the expression ‘mixed metaphor’ as a

metalinguistic comment in the two-billion-word Oxford English Corpus. I
consider the co-text of 141 occurrences of ‘mixed metaphor’ in the corpus, in
order to shed light on the kinds of uses of metaphors that writers opt to explicitly
draw attention to as involving ‘mixing’. I show how folk understandings of
‘mixed metaphor’ include phenomena that do not correspond to the technical
use of the term in the specialist literature, and reflect on the implications of these
findings for metaphor theory. Some attention is given to the use of the phrase
‘mixed metaphor’ in different genres, the relevance of grammatical boundaries to
perception of ‘mixing’ between metaphors, and the possible pragmatic motivation
for using ‘mixed metaphor’ as a metalinguistic label. The study broadly confirms
the prevailing view that the notion of ‘mixed metaphor’ often involves a negative
evaluation of a particular stretch of language and of the speaker/writer who
produced it. However, in a substantial minority of cases, the phrase is used
humorously to point out what are in fact creative, witty and highly effective uses
of metaphor.

10.1  I ntroduction: Mixed metaphor as a ‘folk’ concept

and a technical term

In this chapter I investigate the use of the phrase ‘mixed metaphor’ as a meta-
linguistic comment in a large corpus of English. I show that what is described as
‘mixed metaphor’ in authentic discourse does not always correspond with the defi-
nitions that are proposed in the specialist literature or in guides on good writing.
More specifically, my data provides some insights into the ways in which speakers
of English attempt to achieve coherence between neighbouring uses of different
metaphors, and into the circumstances in which they perceive enough of a clash

doi 10.1075/milcc.6.10sem
© 2016 John Benjamins Publishing Company
 Elena Semino

or incongruity in their own or others’ language choices to use the term ‘mixed
metaphor’. My study broadly confirms the prevailing view that the label ‘mixed
metaphor’ often involves a negative evaluation of a particular stretch of language
and of the speaker/writer who produced it. However, in a substantial minority of
cases, the phrase is used humorously to point out what are in fact creative, witty
and highly effective uses of metaphor.
Why is it relevant to study how the phrase ‘mixed metaphor’ is applied in
general language use? I would argue that ‘mixed metaphor’ is, first and foremost,
a ‘folk’ concept. The label is normally used to refer to an incongruity between two
or more metaphors that are used in close proximity to one another, and to suggest
a negative evaluation of such uses of metaphor and of those who produce them.
This negative evaluation is usually based on the idea that clashes between contigu-
ous metaphors may result in confusion and/or unintended humour. This explains
a general prescriptive injunction to avoid mixed metaphors. The definition below
is taken from the fourth edition of a practical guide to public speaking aimed at
students, which was published in the US in 2012:
A mixed metaphor makes illogical comparisons between two or more things.
When speakers mix their metaphors, they begin with one metaphor and then
switch to another midstream. The confusion, if not humour, that results from
mixed metaphors is apparent in the following examples […]. […] In short, mixed
metaphors bring together too many or contradictory associations and are difficult
to visualize. To avoid mixed metaphors, take a close look at the metaphors you
want to use to be sure the words in the phrase refer to the same category, event or
thing.(Griffin 2012: 213–14)

The following extract from an online review included in my corpus data is repre-
sentative of the uses of metaphor that are usually described as ‘mixed’ in manuals
such as the above and in the academic literature on this topic:
(1) But there’s no getting around the fact that Chappelle’s sudden disappearance
was pretty much bat-@#$% crazy, and though Murphy and Rawlings
do their best to defuse the big elephant in the room (there’s a mixed
metaphor for you), I never quite escaped the feeling that this was just
weird, and perhaps unnecessary. (http://www.dvdverdict.com/reviews/
chappellesshowlost.php)1

Here I assume that the writer labels as mixed metaphor the expression: ‘defuse
the big elephant in the room’. In this expression, ‘defuse’ metaphorically describes
the attempt to deal with an awkward issue in terms of bomb disposal, while ‘big

.  All webpages mentioned in this chapter were accessed in July 2012.
 Chapter 10.  A corpus-based study of ‘mixed metaphor’ as a metalinguistic comment 

elephant in the room’ metaphorically describes the issue itself in terms of a large
animal whose presence in a confined space is impossible to ignore. The whole
description arguably poses no comprehension problems, partly because the
metaphorical uses of both ‘defuse’ and ‘elephant in the room’ are conventional in
English. However, if one attempts to make sense of the whole expression at the
figurative level, one ends up with an incongruous and potentially humorous sce-
nario where the actions that are appropriate to the disposal of bombs are applied to
a big elephant in a room. Readers may of course not imagine such a scenario at all,
but the writer’s parenthetical comment ‘there’s a mixed metaphor for you’ suggests
that he or she is aware of this possibility. Indeed, the use of the label ‘mixed meta-
phor’ itself pre-empts the possible criticism that the two incongruous metaphors
are used unwittingly, thus safeguarding the writer’s own credentials as a competent
user of the English language.
The example above is adequately accounted for by different approaches
to ‘mixed metaphors’ in the scholarly literature. In Leech’s (1969) fairly stan-
dard approach to the topic in relation to poetry, for example, mixed metaphors
occur when:
dead metaphors, which have lost their imaginative force, are brought incongru-
ously  together so that a conflict in their literal meanings, which normally go
­unnoticed, is forced upon our attention. (Leech 1969: 161)

From Leech’s point of view, the use of both ‘defuse’ and ‘big elephant in the room’
in my example would be described as ‘dead metaphors’. The use of the two expres-
sions in close proximity to each other would therefore bring their literal meanings
to the writer’s attention, so that a conflict between them is perceived.
In Lakoff and Johnson’s (1980: 91–6) Conceptual Metaphor Theory, ‘mixed
metaphor’ potentially arises when, within a narrow stretch of text, a particular
target domain is talked about in terms of metaphorical expressions that draw
from different source domains. Within this theory, different source domains
may cater for different aspects of the same target domain, and hence may be
applied to that target domain for different purposes. For example, Lakoff and
Johnson suggest that the journey source domain is conventionally mapped onto
the ­argument target domain in order to capture the notion of the ‘progress’ of
arguments, as in the expression ‘at this point in our argument’. In contrast, the
container source domain is conventionally mapped onto the argument target
domain in order to capture the notion of the ‘content’ of arguments, as in the
expression ‘I’m tired of your empty arguments’. According to Lakoff and J­ ohnson,
it is possible to mention both aspects of arguments within a single stretch of
text, as in the expression ‘At this point our argument does not have much con-
tent’. This kind of expression is described as a ‘permissible’ mixed metaphor, for
 Elena Semino

two ­reasons:  (a)  each of the two metaphorical expressions is used to express a
different aspect of the ­argument, and (b) the conceptual metaphors involving,
respectively, the journey source domain and the container source domain
have a shared entailment: as an a­ rgument is developed, more ‘surface’ is created
within both source domains, i.e., a longer path in the journey source domain
and a larger surface in the c ­ ontainer source domain. This shared entailment
makes the two metaphors mutually coherent, even though they are not mutu-
ally consistent, as there is ‘no “single image” that completely fits both metaphors’
(Lakoff and ­Johnson 1980: 95). The attempt to combine the main purposes of the
two different domains into a single metaphor thus results into what Lakoff and
Johnson call ‘impermissible’ mixed metaphors, such as ‘The content of the argu-
ment proceeds as follows.’ Regardless of exactly what Lakoff and Johnson mean by
‘permissible’ and ‘impermissible’, it is clear that Example 1 above would count as
an ‘impermissible’ mixed metaphor. There are no metaphorical entailments, in
Lakoff and Johnson’s sense, that are ­straightforwardly shared between what we
may call the bomb disposal source domain and the large entity in confined
space source domain when they are mapped onto the general target domain that
may be labelled difficult issues.
Similarly, my example counts as a clear case of a ‘mixed’ metaphor cluster
in terms of Kimmel’s (2010) study, which provides an exceptionally systematic
and methodologically sophisticated account of metaphor clusters generally, and
mixed metaphor in particular, in authentic data. In Kimmel’s terms, the expres-
sions ‘defuse’ and ‘the big elephant in the room’ form a mixed metaphor because
they ‘occur in textual adjacency’, and share ‘no source-domain similarity’ and
no ‘direct inferential entailments’ (Kimmel 2010: 98–101). Kimmel addition-
ally argues that different metaphors occurring in separate but textually adjacent
clauses are not normally perceived as mixed or clashing, since, in his view, lan-
guage users attempt to integrate the literal meanings of metaphorical expres-
sions only within as opposed to across clause boundaries. As I have mentioned,
the two different metaphorical expressions in my example occur within a single
clause, and are therefore candidates for the perception of a clash within Kimmel’s
approach.
While all three scholarly approaches to mixed metaphor apply to my proto-
typical example, the concept of ‘mixed metaphor’ poses some difficulties for tech-
nical definitions. Describing metaphors as being ‘mixed’ suggests, implicitly or
explicitly, that language users activate the literal meanings of metaphorical expres-
sions, and perceive a clash between them that results in potential confusion and/or
negative evaluation. This is problematic for two main reasons. On the one hand, it
is difficult to predict when, how and for whom the literal meanings of metaphori-
cal expressions are involved in language production and reception (Steen 2008 and
 Chapter 10.  A corpus-based study of ‘mixed metaphor’ as a metalinguistic comment 

Gibbs 2011 are an example of recent debates on this issue). On the other hand,
there is increasing evidence that different metaphorical expressions often occur
in close proximity to one another without causing any comprehension problems,
as shown, among others, in a number of recent studies on metaphor ‘clustering’
(Corts and Pollio 1999, Cameron and Stelma 2004, Kimmel 2010; the findings of
Shen and Balaban’s 1999 study are also relevant). This raises the issue of which
cases of clustering may be aptly described as ‘mixed’ (Kimmel 2010). Moreover,
the combination of different (conventional) metaphors in poetry and other text
types is often regarded as a manifestation of creativity and exceptional mastery
of language, as in Leech’s (1969) ‘compound metaphors’ and Lakoff and T ­ urner’s
(1989) ‘composite’ metaphors. However, the distinction between a clumsily
‘mixed’ metaphor and a skilfully ‘compound/composite’ metaphor is not always
clear-cut, and can be, once again, partly a matter of individual subjective percep-
tion (Müller 2008).
These considerations lie behind this study of ‘mixed metaphor’ as a metalin-
guistic comment in a corpus of English. My goal is to consider the kinds of expres-
sions that are described as ‘mixed’ in authentic language use, in order to gain some
understanding of the semantic, grammatical and pragmatic circumstances in
which ‘ordinary’ language users feel that the explicit use of the label is appropriate.

10.2  Data

The data for this study was collected from the Oxford English Corpus (OEC),
which contains over 2 billion words of 21st century written English (http://oxford-
dictionaries.com/words/the-oxford-english-corpus). The materials included in
the corpus are mostly drawn from the World Wide Web and are divided into broad
subject areas, such as news, leisure, business, and so on.
I used the concordance facility in the online tool Sketchengine in order to
extract from the OEC all occurrences of the phrase ‘mixed metaphor’ (http://www.
sketchengine.co.uk/; Kilgarriff et al. 2004). More specifically, I conducted a lemma
search of ‘mix’ and ‘metaphor’ immediately following each other: this provides
any instance of the word form ‘mix’ when it is followed by any instance of the
word form ‘metaphor’. As a result, the concordance I obtained includes the phrase
‘mixed metaphor’ and a number of variants, such as ‘mixed metaphors’, ‘to mix
metaphors’, ‘mixing metaphors’. For the sake of simplicity, in the rest of this paper
I will still use ‘mixed metaphor’ to refer generally to the small range of variants
included in my concordance.
The search I have just described produced 322 hits, which were ­distributed
across 16 different sections of the corpus. For the purposes of my analysis,
 Elena Semino

I  focused on the two sections of the corpus that have the highest numbers of
­occurrences of the search string, namely: ‘weblog’ (88 occurrences) and ‘arts’ (53
occurrences). The former section contains extracts from a wide variety of online
blogs, which vary considerably in terms of topics and level of formality. The lat-
ter section includes criticism and reviews, in the areas of literature, film, music,
art, and so on. The total number of instances of ‘mixed metaphor’ I analysed is
therefore 141.
Given my choice of data, I cannot claim that my study reflects the views on
‘mixed metaphor’ that are held by the majority of speakers of English, or by an
idealised ‘average’ user of the language. The OEC only includes writing that is avail-
able online, and is not therefore representative of general language use, in spite of
the increasing opportunities afforded by Web 2.0. In any case, ‘mixed metaphor’ is
not likely to be part of the productive verbal repertoire of most speakers of ­English.
Hence my findings can only be generalised to those speakers of English who use
the term and produce the kinds of texts included in the sections of the corpus I ana-
lyzed. On the other hand, those sections include a considerable variety of texts and
writers: the blogs vary from personal, autobiographical online diaries to more for-
mal political commentaries; similarly, the ‘arts’ section ranges from extracts from
printed books to brief online film reviews. There is also considerable variation in
terms of the kinds of ‘English’ represented in the data, given that the World Wide
Web was exploited in the composition of the corpus. The writers are geographi-
cally spread around the world, but with a prevalence of countries where English is
mostly spoken as a first language, such as the USA, the UK and ­Australia. While,
from my observations, most writers represented in my examples seem to be native
speakers of English, I have no way of establishing that all of them are.

10.3  Patterns in the use of ‘mixed metaphor’ in the data

In this section I begin by discussing some general patterns in the data before focus-
ing in more detail on the semantic and grammatical characteristics of the uses of
language that tend to be described as ‘mixed metaphor’.
I used the collocation tool in Sketchengine to extract a list of the most fre-
quent collocates of ‘mixed metaphor’ in the OEC, i.e., words that occur unusually
frequently in close proximity to the search string.2 The top ten collocates confirm

.  Collocates were calculated by means of the default settings in Sketchengine, namely:
within a window of 5 words to the left and 5 words to the right of the search string; with a
minimum frequency in the corpus of 5 occurrences; with a minimum frequency in the given
 Chapter 10.  A corpus-based study of ‘mixed metaphor’ as a metalinguistic comment 

the general assumption that the use of mixed metaphors tends to be perceived
negatively. They include:

–– ‘Cliché’ and ‘clichés’ – namely, references to other uses of language that are
negatively evaluated;
–– ‘horribly’ and ‘unfortunate’ – namely, an adverb and an adjective that indicate
a negative evaluation of entities, actions or events;
–– ‘pardon’, ‘apologies’, ‘excuse’, ‘alert’ – namely, expressions that convey the writ-
er’s awareness that the co-text includes a form of words that readers may find
objectionable.

I then proceeded to classify each of the 141 selected occurrences of ‘mixed meta-
phor’ in terms of two main aspects of variation:

i. Whether the writer applies the label ‘mixed metaphor’ to their own language
use or to somebody elses’s language use;
ii. Whether the writer applies the label ‘mixed metaphor’ to an instance of lan-
guage use that is provided in the co-text or not.

In Example 1 above, for instance, ‘mixed metaphor’ is applied to an immediately

preceding use of language that has been produced by the writer him- or herself. In
contrast, in Example 2 below from the ‘arts’ section of the OEC, the writer refers
generally to another person’s language use without providing any actual examples
of what is regarded as ‘mixed metaphors’:
Trixie is a film about its namesake, a woman with an education that didn’t
(2) 
go past the fifth grade, a medical history of blunt head trauma from a
bowling ball, and a propensity for mixed metaphors and crucified cliches.
(http://www.cinescene.com/reviews/trixie.html; ‘arts’ section of OEC;
doc.id: 02.03.0016.005)

Table 10.1 below shows that, overall, my data includes slightly more instances
of references to one’s own use of ‘mixed metaphor’ rather than other people’s.
However, the two sections of the corpus differ in this respect. The majority of
instances of ‘mixed metaphor’ in the ‘arts’ section (72%) are applied to someone
else’s language use, while in the ‘weblog’ section the majority of instances of ‘mixed
metaphor’ (75%) are used as references to expressions used by the writer him- or

range of 3 occurrences; and on the basis of the logDice measure of statistical significance
(Rychly 2008), which, in the results I obtained, produces the same set of top collocates as
Mutual Information.
 Elena Semino

herself.3 This is not surprising, as the texts collected under ‘arts’ involve critiques
and evaluation of other people’s artistic productions, while blogs can be more
autobiographical and self-conscious in orientation.

Table 10.1.  Distribution of ‘mixed metaphor’ as a metalinguistic

­comment on one’s own vs. others’ uses of language
Text type Self (% of total in row) Other (% of total in row) Total

Arts 15 (28%) 38 (72%) 53

Blogs 66 (75%) 22 (25%) 88
Total 81 (57%) 60 (43%) 141

Table 10.2 shows a further contrast between the two genres in terms of the pres-
ence in the co-text of (an instance of) the expression that the writer describes as
‘mixed metaphor’. In the ‘arts’ section, just below half of the instances of ‘mixed
metaphor’ are preceded or followed by the relevant expression(s); in contrast,
in the ‘weblog’ section the vast majority of instances of ‘mixed metaphor’ (87%)
describe choices of words that are present in the co-text.4 This difference is partly
linked with the differences illustrated in Table 10.1: the authors of the texts in the
‘arts’ section often make general claims about the language (or images) used in
the works they are discussing, while the authors of blogs tend to comment on the
linguistic choices that they themselves make as they write.

Table 10.2.  Distribution of ‘mixed metaphor’ as a label with or without an example

Text type Label only (% of total in row) Label and example (% of total in row) Total

Arts 30 (57%) 23 (43%) 53

Blogs 11 (13%) 77 (87%) 88
Total 41 (29%) 100 (71%) 141

The rest of this section is devoted to an analysis of those instances of ‘mixed meta-
phor’ that are in close textual proximity to the relevant uses of language, which,
as shown in Table 10.2, amount to exactly 100 examples. In each case, I attempted
to identify the precise expressions that the writer describes as ‘mixed metaphor’,
and classified each instance in terms of the kind of semantic relationship or clash

.  This difference between the two genres is statistically significant at p<0.0001 (Fisher exact
test, df1).
.  This difference between the two genres is statistically significant at p<0.0001 (Fisher exact
test, df1).
 Chapter 10.  A corpus-based study of ‘mixed metaphor’ as a metalinguistic comment 

that appears to have led the writer to use the metalinguistic label. As shown in
Table 10.3, I identified three broad but distinct patterns:

–– Pattern (a) – Clashes between the literal meanings of different uses of meta-
phor that, in terms of Conceptual Metaphor Theory, can be subsumed under
different source domains;
–– Pattern (b) – Clashes between the literal meanings of different uses of meta-
phor that, in terms of Conceptual Metaphor Theory, can be subsumed under
the same source domain;
–– Pattern (c) – Similarities-cum-clashes between the literal meaning of a meta-
phorical expression and some aspect of the current topic.

Table 10.3.  Types of semantic relations that are labelled ‘mixed

­metaphor’ where an ­example is provided
Semantic relations Occurrences

a.  Clashes across source domains 32

b.  Clashes within source domains 40
c.  Relationship between source domain and topic 28
Total 100

Pattern (a) includes prototypical instances of mixed metaphor, such as the one I
discussed earlier as Example (1). The other two patterns were more surprising,
however. The uses of metaphor that are included under pattern (b) do not consti-
tute prototypical instances of ‘mixed metaphor’: they would not, for example, be
classified as ‘mixed’ within Kimmel’s (2010) typology, nor in terms of Lakoff and
Johnson’s definition of both ‘permissible’ and ‘impermissible’ mixed metaphors.
The choices of language that are included under pattern (c) would not count as
‘mixed metaphor’ according to any existing definition. Table 10.3 shows that the
three patterns are roughly similar in frequency in my data, with a slight prevalence
of pattern (b).
In the rest of this section I first consider pattern (b) in more detail; secondly,
I discuss the respective grammatical positions of the uses of metaphor that are
described as ‘mixed’ within both pattern (a) and pattern (b); and finally I discuss
pattern (c).

10.3.1  C
 lashes between different uses of metaphor involving the same
source domain
As shown in Table 10.3, 40 out of the 100 concrete uses of language that are
described as ‘mixed metaphor’ in my data involve metaphorical expressions that,
 Elena Semino

in the terms used within Conceptual Metaphor Theory, can be explained in terms
of a single source domain (Cameron et al.’s 2010 notion of ‘vehicle grouping’ could
also be applied in a similar way). Example 3 below is taken from the ‘weblog’
­section of the corpus:

(3) A half full glass in a toenail of a day*

[…]
*I am aware that this is a mixed metaphor, but it is a much more original
one than something the Prime Minister is reported to have said today:
‘There are big bumps in the road ahead; it’s not all going to be smooth
sailing.’ (http://stilllifewithcat.blogspot.co.uk/search?q=mixed+metaphor;
‘weblog’ section of OEC; doc.id: not provided)

Here the term ‘mixed metaphor’ is used in what looks like a note or afterthought
to the main blog entry. The writer makes clear that she is aware that she has used
‘a mixed metaphor’ in the title of her post, and somewhat defensively describes it
as ‘much more original’ than an utterance that has recently been attributed to the
Australian Prime Minister.
The title of the entry (‘A half full glass in a toenail of a day’) involves two uses
of metaphor that draw from different source domains, and was therefore classi-
fied under pattern (a) for the purposes of my analysis. The second use of meta-
phor that the blogger describes as ‘mixed’ is different, however. In the quotation
attributed to the Prime Minister, future difficulties are described metaphorically as
‘big bumps in the road ahead’ and in terms of ‘it’s not all going to be smooth sail-
ing.’ Here there is clearly a contrast between two different metaphorical scenarios
(‘road travel’ and ‘sea faring’; Musolff 2006, Semino 2008). However, both uses of
metaphor arguably involve what, within Conceptual Metaphor Theory, would be
called the journey source domain. Indeed, in both cases the notion of problems
on a journey is mapped onto the notion of abstract difficulties; and, in both cases,
the metaphorical scenario involves an entailment whereby the greater the impedi-
ment to travel, the greater the difficulties experienced by the people or institutions
the Prime Minister is alluding to.
In other words, in the latter case what the blogger describes as ‘mixed’ are two
uses of metaphor that are in fact rather similar to and broadly compatible with
each other. In Lakoff and Johnson’s (1980) terms, the metaphorical e­ xpressions
­attributed to the Prime Minister would not in fact even constitute a case of
­‘permissible’ mixed metaphor (as a result of shared entailments), but rather as
instantiations of the same conceptual metaphor: life is a journey. In Kimmel’s
terms, this particular case would be classified as a conceptually complementary
metaphorical cluster in view of the sharing of source domain ontology, and hence
not as a ‘mixed cluster’ at all. As Kimmel (2010: 101) puts it, ‘To count mixed
 Chapter 10.  A corpus-based study of ‘mixed metaphor’ as a metalinguistic comment 

clusters, the most important thing to exclude is that no source-domain similarity

obtains between metaphors.’5
Similar considerations apply to the following three examples, all from the
‘weblog’ section of the corpus:
(4) Anyhow, I think I’m going to be keeping a lot of this stuff in my ­paper
­journals, so to avoid boring y’all, and to keep things on a different
note, instead of the same record getting played over and over again
­(mixing metaphors is a hobby of mine). (http://eiram.blogspot.co.uk/
archives/2002_09_01_eiram_archive.html; ‘weblog’ section of OEC;
doc.id: 20.03.0015.006)
(5) This is frustrating for me, because I rely on the words always being at
my fingertips and on the tip of my tongue (with apologies for the mixed
metaphor ); but, currently, they’re not. (http://www.whereveryouare.org.
uk/weblog/archives/week_2002_07_28.html; ‘weblog’ section of OEC;
doc.id: 20.03.0007.009)
(6) Most entertaining of all has been the mimicking by the DUP of the
canary-in-the-mine routine. For years, Mitchel McLoughlin, Sinn Fein
Chairman, has been sent out to test out the atmosphere north and south
by saying the unexpected. If the kite flies (if you’ll excuse the mixed
metaphor), Gerry Adams rushes forth and brandishes it. Lo! No sooner
had Jeffrey Donaldson changed all his stationery to reflect his new
political allegiance, but he became the DUP’s canary, (http://www.
sluggerotoole.com/archives/2004/09/index.php; ‘weblog’ section of OEC;
doc.id: 20.02.0008.007)

In Example 4, the perception of ‘mixing’ involves two musical metaphors: ‘keep

things on a different note’ and ‘the same record getting played over and over again’.
There is of course a difference between keeping the same note and playing the
same record. However, in both cases, the idea of lack of variation is metaphori-
cally described in terms of repetition of the same music. In Example 5, the writer
apologises for using two different metaphorical idioms for the ability to remember
and say things: ‘at my fingertips’ and ‘on the tip of my tongue’. Here there is a con-
trast in body parts, but, in both cases, the things that the writer is able to say and/
or remember are located at the tip of entities at the front of the body, and hence
physically accessible.

.  Kimmel (2010) tagged his examples both in terms of relevant source domains and in
terms of relevant image schemas. When a similarity was found between two nearby uses of
metaphor in terms of at least one of these levels, the cluster was not regarded as mixed.
 Elena Semino

Example 6, which is concerned with politics in Northern Ireland, involves

a longer stretch of text and more creative metaphorical expressions. The writer
uses two different metaphors in order to describe a strategy attributed to the Irish
Republican Party Sinn Fein, which, it is argued, is being copied by the DUP (Dem-
ocratic Unionist Party). The strategy involves dispatching a particular representa-
tive to hold exploratory talks before a more senior politician becomes involved:
‘the canary-in-the-mine routine’ (and later ‘the DUP’s canary’) and ‘if the kite flies’.
The two metaphorical scenarios are quite different, of course. The first has its basis
in the practice of miners taking a caged canary down the tunnels of a mine, as the
bird would die before humans if dangerous fumes escaped: here the politician who
is sent forth to say something ‘unexpected’ is humorously described as the canary.
The second scenario involves the successful flying of a kite, which presumably
stands for political success in negotiations. The two uses of metaphor share the
same source domain in a much looser sense than in previous cases. Nonetheless,
the two scenarios can be seen as variants of the same generic schema: both involve
entities that can fly, in the sense of being suspended in the sky; and, in both cases,
flying, or the continued potential ability to fly, corresponds to success in the politi-
cal domain.
It could be tempting to see these examples as evidence that ‘ordinary’ language
users often apply the label ‘mixed metaphor’ inappropriately. Rather, I suggest that
these examples give us an insight into the ways in which writers may attempt to
establish coherence across different uses of metaphors, and into the circumstances
in which they consciously perceive some kind of clash or incoherence among dif-
ferent metaphors that occur in close proximity to one another. This is particu-
larly important given that, as I mentioned earlier, the tendency for metaphors to
occur in clusters poses a challenge to any clear-cut technical definition of ‘mixed
metaphor’.
I have shown that, in a substantial proportion of cases, a fundamental simi-
larity between different metaphors used in textual proximity seems to lead to an
increased awareness of the contrasts between those metaphors. For example, the
fact that both of the metaphors used in Example 4 involve music may have made
the writer more consciously aware of the differences between the literal mean-
ings of the two metaphors, i.e. between continuing to play the same note vs.
playing the same record over and over again. This is in fact consistent with the
notion that relatively conventionalized expressions may be ‘revitalized’ in particu-
lar textual circumstances, including when metaphors that share the same source
domain occur in close proximity to one another (Müller 2008: 6, Semino 2008: 19)
What my data additionally suggests is that, in such circumstances, the similari-
ties between different uses of metaphors (e.g., in terms of shared source domains
and entailments) may possibly lead speakers/writers to expect and attempt to
 Chapter 10.  A corpus-based study of ‘mixed metaphor’ as a metalinguistic comment 

achieve a greater degree of coherence between those metaphors than when no

such ­similarity applies. Against the background of those expectations of coher-
ence, the contrasts between the different metaphors may become salient enough
to lead to metalinguistic references to ‘mixing’. This does not necessarily mean that
those uses of metaphor may be perceived as ambiguous or difficult to understand,
but simply as involving some kind of clash or incongruity that the writer feels the
need to point out.

10.3.2  Grammatical boundaries and ‘mixed metaphor’

The notion of ‘mixed metaphor’ raises a further methodological issue, namely:
how close do two different uses of metaphor have to be in order to be potentially
described or perceived as ‘mixed’? This issue is seldom addressed explicitly or
systematically, but the examples given in the literature tend to involve meta-
phors that occur in the same clause or in the same sentence. Kimmel’s (2010)
study is, however, an exception in terms of methodological explicitness, includ-
ing with respect to the role of grammatical boundaries in the identification of
potential mixed metaphors. In coding his corpus for the presence of mixed
clusters, ­Kimmel makes a three-way distinction between (i) metaphors that
occur in adjacent sentences, (ii) metaphors that occur in adjacent clauses within
the same sentence, and (iii) metaphors that occur within the same clause. He
argues that the tightness of the grammatical connection between the units that
contain different metaphors ‘affects our tendency to process the metaphors
integratively or not’ (Kimmel 2010: 110; cf.  Goatly 2011: 288), and proposes
that clusters should only be described as ‘mixed’ when they involve metaphors
that occur within the same clause (and that cannot be explained in terms of the
same source domain or image schema, as I mentioned earlier). Kimmel puts it
as follows:
Only the close syntactic integration of two metaphors within a clause can enforce
or foster a close integration of their semantic content qua imagery. Where mixed
metaphors occur across clauses no ontological clashes will be felt to begin with
and secondary mechanisms to keep the clash at bay are dispensable. If this is
correct, making sense of mixed metaphors is a natural by-product of default
clause processing. (Kimmel 2010: 110)

It should be noted that the goal of Kimmel’s (2010) study was to explain why,
by and large, the use of different metaphors in close proximity to one another
does not appear to pose comprehension problems, or lead to the kind of percep-
tions of incongruity and verbal clumsiness that are associated with the notion
of ‘mixed metaphor’. In contrast, my study attempts to explain how and why the
label ‘mixed metaphor’ is actually used in discourse. My data suggests that neither
 Elena Semino

clause boundaries nor sentence boundaries can, in principle, block the perception
of clashes between metaphors that are described as ‘mixed’.
More specifically, I classified in terms of relevant grammatical boundaries all
expressions described as ‘mixed metaphor’ within semantic patterns (a) and (b) in
Table 10.3 above. This amounts to 72 examples, in total, namely:

–– 32 instances of textually proximal metaphors involving different source

domains, such as Example 1 above (pattern a); and
–– 40 instances of textually proximal metaphors involving shared source domains,
as in the examples discussed in the previous sub-section (pattern b).

The results of the grammatical analysis are presented in Table 10.4. The table
shows that in only 40% of cases the label ‘mixed metaphor’ describes different uses
of metaphor that occur within the same clause, as in Examples 1 and 5 above. In a
similar proportion of cases, a clause boundary separates the two uses of metaphor,
as in the second instance of mixed metaphor in Example 3. Perhaps surprisingly,
in the remaining 14% of the cases, the two relevant uses of metaphor occur in dif-
ferent sentences, as in Example 6.

Table 10.4.  The label ‘mixed metaphor’ and grammatical boundaries

Grammar Occurrences (and % proportion)

Across sentences 10 (14%)

Across clauses 33 (46%)
Within clause 29 (40%)
Total 72 (100%)

On the one hand, these findings suggest that neither clause nor sentence
boundaries necessarily block writers’ attempts to achieve coherence among the
literal meanings of different metaphorical expressions. On the other hand, in all
cases the relevant uses of metaphor apply to the same topic, and that seems to
account for any search for or expectation of coherence, regardless of the presence
of clause and sentence boundaries. In a number of cases, moreover, there is some
kind of grammatical parallelism between the units within which the different met-
aphors occur. This applies, for example, to the prepositional phrases in Example 5
(‘at my fingertips and on the tip of my tongue’). This grammatical parallelism may
contribute to expectations about semantic relationships of similarity or contrast,
which contribute to explain the perception of clashes between metaphors (Short
1996: 14–15).
 Chapter 10.  A corpus-based study of ‘mixed metaphor’ as a metalinguistic comment 

10.3.3  ‘Mixed metaphor’ and relationships between source

domain and topic
In this section I turn to those cases where the label ‘mixed metaphor’ is applied to
describe stretches of text that do not in fact involve two or more uses of metaphor.
Rather, the label seems to indicate the perception of a relationship of overlap-cum-
contrast between the source domain of a particular use of metaphor and the cur-
rent topic. As shown in Table 10.3 above, this applies in 28 of the 100 cases in my
data where ‘mixed metaphor’ describes a use of language that is available in the
co-text.
In Example 7 below a blogger is talking about his musical activities:
(7) Rehearsal on Monday went well, but I have come down with a cold and
am not feeling 100% at the moment. Finally got my hair cut this morning
before heading into town and playing for AC at the VCA. She’s working on
a large song-cycle by Ned Rorem and will be performing it in November,
then again at the Woodend Festival (?) in February. It certainly keeps me on
my sight-reading toes (now, there’s an interesting mixed metaphor). (http://
fcprain.blogspot.co.uk/2004_09_01_fcprain_archive.html; ‘weblog’ section
of OEC; doc.id: 20.03.0057.011)

It appears that what is described as ‘mixed metaphor’ in this extract is the

expression ‘keeps me on my sight-reading toes’. The metaphorical idiom ‘to
be kept of one’s toes’ has a metonymic basis: by standing on our toes we gain
a larger field of vision, so that this particular physical posture is metonymi-
cally associated with being alert and ready to take action. Here, however, the
idiom applies metaphorically to the blogger’s preparedness to perform music at
a high level, particularly by being able to sight-read music effectively. Hence,
the reference to sight-reading is, according to my analysis, literal or, at any rate,
non-metaphorical. In the noun phrase ‘my sight-reading toes’, the literal ref-
erence to sight-reading functions as pre-modifier to the metaphorically used
head noun ‘toes’. This creates a contrast between the metaphorical toes and
another part of the body, the eyes, which are literally involved in sight-reading.
If one attempted to integrate the literal meaning of ‘sight-reading’ and the lit-
eral meaning of ‘toes’, one would end up with a fantastic image of toes that are
endowed with eyes and can therefore sight-read. In any case, what seems to be
described as ‘mixed metaphor’ is a relationship between a metaphorical refer-
ence to a part of the body and a literal reference to an activity that is performed
with a different part of the body. In other words, there is both similarity and
contrast between an aspect of the current topic and the literal meaning of a
metaphorical expression.
 Elena Semino

Example 8 relies on a more sophisticated combination of similarity and con-

trast between a metaphorical scenario and the current topic:
(8) So Soviet Planners, always quite thorough, figured out beforehand each
street Reagan would see – and then re-paved and re-painted all along each
route. Wherever Reagan traveled, Moscow looked good. It’s no coincidence
that “Potemkin Village” is a Russian expression.
 Now the Chinese are heirs to the Communist throne, to mix
­metaphors. (http://vodkapundit.com/archives/2002_02.php; ‘weblog’
­section of OEC; doc.id: 20.99.0037.005)

The general topic of the extract is practices in Communist regimes. The blogger
starts by talking about the Soviet Union and then mentions ‘the Chinese’, who are
described metaphorically as ‘heirs to the Communist throne’. The expressions ‘heir’
and ‘throne’ are conventionally used, both separately and together, to describe
metaphorically a situation where someone takes over a particular role from some-
body else. In the extract, however, a metaphorical reference to monarchic practices
is used to describe the relationships between two Communist regimes. On the
one hand, the topic and the source domain are not as separate as is normally the
case with metaphor, as they both involve types of political regimes: in that sense,
they arguably belong to the same broad conceptual domain. On the other hand,
there is a fundamental contrast between the two types of regimes, and particularly
between the egalitarian aspirations of Communism and the monarchic practice of
handing over power to blood relatives. The blogger’s metalinguistic comment, ‘to
mix metaphors’, seems to be a reaction to the perception of this partial and awk-
ward overlap between the topic and the source domain of the metaphor.
These examples are in fact reminiscent of ‘topic-triggered metaphors’, where the
choice of source domain or scenario seems to be inspired by the topic under discus-
sion (Koller 2004, Semino 2008). This is one of the situations in which conventional-
ized metaphorical expressions may be ‘revitalized’, resulting in a kind of metaphorical
punning (Goatly 2011: 195–6). These puns are often used for humorous purposes
in newspaper headlines, as, for example in ‘Plastic surgery laws go under the knife’
(The Guardian, 28th January 2005). In the examples I discussed in this section, in
contrast, the writers seem to become aware of some unintended ambiguity and/or
incongruity in their choice of metaphor, and use the term ‘mixed metaphor’ in spite
of the absence of two separate uses of metaphor that can potentially be ‘mixed’.

10.4  Concluding remarks

Because of the way in which my data was collected, I cannot claim that the exam-
ples I have discussed in this chapter are representative of what counts as ‘mixed
 Chapter 10.  A corpus-based study of ‘mixed metaphor’ as a metalinguistic comment 

metaphor’ according to scholarly definitions, nor to the unspoken assumptions

of English speakers. Nonetheless, a number of conclusions can be drawn from
my analysis of 141 occurrences of the expression ‘mixed metaphor’ in the Oxford
English Corpus.
The use of the metalinguistic label varies, in part, according to genre. In the
‘arts’ section of the corpus, the majority of instances apply to somebody else’s
language use, and there is an even split between cases where the relevant use of
language is available in the co-text and cases where this is not. This reflects the
fact that the ‘arts’ section contains criticism and reviews, where the topic is the
artistic production of other people, and where writers often make general state-
ments about somebody else’s style without necessarily providing relevant quotes.
In contrast, in the ‘weblog’ section of the corpus, ‘mixed metaphor’ is used in ref-
erence to the writer’s own writing in three quarters of the cases, and most uses of
the label apply to a stretch of language that is provided in the co-text. This reflects
the tendency for blogs to have an autobiographical focus, and for bloggers to write
rather self-consciously.
My analysis of the 100 instances where ‘mixed metaphor’ is applied to a co-
textually available stretch of text has revealed that prototypical cases of mixed met-
aphor occur in just under a third of instances. By ‘prototypical’ instances I mean
cases such as ‘defuse the big elephant in the room’, where the literal meanings of
different metaphorical expressions differ in terms of the source domains they
potentially evoke, and are too incompatible to be semantically integrated in any
way. In all other cases, it could be argued that the expressions that are described
as ‘mixing’ actually involve the same broad conceptual domain. I have discussed
two different patterns where this applies. In pattern (b) above, the literal meanings
of two different metaphorical expressions involve different scenarios that can be
subsumed under a single source domain (as well as often sharing entailments).
This kind of case may be regarded as ‘mixed metaphor’ according to some general
definitions, such as Leech’s (1969), but not within approaches to the topic that are
based on Conceptual Metaphor Theory, as these require the presence of contrast-
ing source domains for ‘mixing’ to occur. In pattern (c) above, the literal meaning
of a single use of metaphor involves a source domain that overlaps in part with the
current topic. This kind of case does not qualify as ‘mixed metaphor’ according
to any technical definition, as it does not involve more than one use of metaphor.
In all instances of both patterns (b) and (c), it is possible to identify a seman-
tic contrast between entities or actions that seems to lead to the perception of
‘mixing’, such as between two different parts of the body, two different modes of
travel, or two different political regimes. In fact, I have argued that this contrast
may be more keenly perceived precisely because the relevant expressions are quite
similar in the first place. Earlier studies would account for the fact that the lit-
eral meanings of metaphorical expressions are more likely to be activated when
 Elena Semino

they co-occur with other expressions that are semantically related. The examples
I have discussed also suggest that the expressions that are described as ‘mixed’
often involve some degree of creativity and/or the presence of semi-fixed multi-
word metaphorical phrases (Müller’s 2008 examples show the same phenomena).
Both aspects may be relevant to explaining why, in some cases, writers become
so consciously aware of their own or others’ uses of metaphor that they explicitly
describe them as ‘mixed’. However, the potential role of similarity as a possible
trigger of perceptions of contrasts described as ‘mixed metaphor’ had not, to my
knowledge, been discussed before.
My findings have some relevance for Conceptual Metaphor Theory. On the
one hand, both of the patterns I have discussed in detail provide examples where
the same conceptual domain arguably accounts for several adjacent linguistic
choices, whether literal or metaphorical. Pattern (b), in particular, may be seen
as evidence that, once a source domain is active, it may be exploited more than
once as a source of metaphorical expressions within relatively short stretches
of text (Shen and Balaban 1999, however, suggest that this may only apply with
relatively novel and explicit uses of metaphor; cf. Steen’s 2008 notion of ‘deliber-
ate metaphor’). On the other hand, pattern (b) in particular suggests that shared
source domains across different metaphorical expressions are not sufficient for
perceptions of coherence. The writers in my data seem to be sensitive to coher-
ence or lack of coherence at the level of specific scenarios rather than broad source
domains, so that even scenarios with shared entailments and source domains are
described as mixed. Indeed, as I have repeatedly pointed out, my data suggests that
expectations of coherence and perceptions of contrasts may increase when two
metaphorical scenarios are broadly similar and can be subsumed under the same
source domain, such as for example, travelling by means of different vehicles, or
making music in different ways. In this sense, Lakoff and Johnson (1980: 94) were
right to stress the importance of different metaphors sharing a ‘single image’ for
the purposes of what they call mutual ‘consistency’. What they did not consider
was the possibility that similar scenarios from the same broad source domain may
not in fact satisfy this criterion from the perspective of language users.
My analysis does not of course undermine the claim made by Kimmel (2010)
and others that, by and large, the use of different metaphors in close proximity goes
unnoticed and poses no comprehension problems. What my study suggests is that
we may be more acutely aware of contrasts between textually proximal metaphors
when those metaphors are in some respect similar to each other. This should, to
my mind, be taken into account when attempting to provide technical definitions
of ‘mixed metaphor’. The examples I included under pattern (c) also involve per-
ceptions of contrasts within aspects of the same broad conceptual domain, but do
not actually involve relationships between different uses of metaphor.
 Chapter 10.  A corpus-based study of ‘mixed metaphor’ as a metalinguistic comment 

Finally, my analysis allows me to draw some conclusions concerning the

personal, interpersonal and generally pragmatic circumstances in which the
metalinguistic label ‘mixed metaphor’ is used. The use of the label in my data
appears to be often motivated by considerations of ‘face’, and specifically one’s
own credentials as a critic, writer, or language user more generally (Goffman
1967; Brown and Levinson 1987). By describing one’s own or others’ language
use as ‘mixed metaphor’, the writers in my data display their own sensitivity to
language use, and sometimes anticipate and prevent potential negative judge-
ments on their own linguistic competence: they cannot be accused of poor criti-
cal discernment, or poor command of their own language use, if they show that
they are consciously aware of potential incongruities in their own or others’ lan-
guage choices. More specifically, the use of ‘mixed metaphor’ when talking about
others’ language use is for the most part critical and judgemental. The situation
is more nuanced when ‘mixed metaphor’ is applied to one’s own language use.
In many cases, the writer’s strategy seems to be primarily defensive, as when a
blogger says ‘apologies for the horribly mixed metaphors’ (Example  3 is also
relevant). A substantial minority of writers, however, seem to confidently claim
the right to use expressions that others may describe as ‘mixed’ because of a
narrow-minded, prescriptive approach to language. These are the cases where
the label is applied to particularly creative and witty uses of metaphor, that in
fact display the writer’s verbal flair rather than being evidence of clumsiness, as
in Examples 3 and 6 above. On a few occasions, writers humorously play with
the (metaphorical) notion of ‘mixing’ itself, as when a blogger comments: ‘well,
if you’re going to mix metaphors, use a blender’.
Overall, my study seems to question whether ‘mixed metaphor’ can be made
into a viable and operationalizable technical term. More positively, however,
there is evidence of considerable consensus among language users as to what
they describe as ‘mixed metaphor’. This consensus can be summarized as follows:
contrasts between at least two textually proximal uses of metaphor; similarity-
cum-contrast between at least two textually proximal uses of metaphor; and
­similarity-cum-contrast between the source domain of a metaphorical expres-
sion and the current topic. Although technical definitions should not of course be
overly swayed by ‘folk’ concepts, this three-way typology can, in my view, provide
a useful foundation for future studies.

References

Brown, P., & Levinson, S. C. (1987). Politeness: Some Universals in language usage. Cambridge:
Cambridge University Press.
 Elena Semino

Cameron, L., & Stelma, J. H. (2004). Metaphor clusters in discourse. Journal of Applied Linguis-
tics, 1(2), 107–136.  doi: 10.1558/japl.2004.1.2.107
Cameron, L., Low, G., & Maslen, R. (2010). Finding systematicity in metaphor use. In
L. ­Cameron & R. Maslen (Eds.), Metaphor analysis: Research practice in applied linguistics,
social sciences and the humanities (pp. 116–146). London: Equinox.
Corts, D. P., & Pollio, H. R. (1999). Spontaneous production of figurative language and gesture
in college lectures, Metaphor and Symbol, 14(2), 81–100.  doi: 10.1207/s15327868ms1402_1
Gibbs, R. W. Jr. (2011). Are “deliberate” metaphors really deliberate? A question of human con-
sciousness and action. Metaphor and the Social World, 1(1), 26–52.
doi: 10.1075/msw.1.1.03gib
Goatly, A. (2011). The language of metaphors. 2nd ed. London: Routledge.
Goffman, E. (1967). Interaction ritual: Essays in face-to-face behavior. Garden City, NY: Anchor
Books.
Griffin, C. L. (2012). Invitation to public speaking. 4th ed. Boston, MA: Wadsworth Publishing.
Kilgarriff, A., Rychly, P., Smrz, P., & Tugwell, D. (2004). The Sketch Engine, Proc. Euralex. Lori-
ent: France, 105–116.
Kimmel, M. (2010). Why we mix metaphors (and mix them well): Discourse coherence, concep-
tual metaphor, and beyond. Journal of Pragmatics, 42, 97–115.
doi: 10.1016/j.pragma.2009.05.017
Koller, V. (2004). Businesswomen and war metaphors: ‘Possessive, jealous and pugnacious?’
Journal of Sociolinguistics, 8(1), 3–22.  doi: 10.1111/j.1467–9841.2004.00249.x
Lakoff, G., & Johnson, M. (1980b). Metaphors we live by. Chicago: University of Chicago Press.
Leech, G. N. (1969). A linguistic guide to English poetry. London: Longman.
Lakoff, G., & Turner, M. (1989). More than cool reason: A field guide to poetic metaphor. Chicago:
University of Chicago Press.  doi: 10.7208/chicago/9780226470986.001.0001
Müller, C. (2008). Metaphors dead and alive, sleeping and waking: A dynamic view. Chicago:
University of Chicago Press.  doi: 10.7208/chicago/9780226548265.001.0001
Musolff, A. (2006). Metaphor scenarios in public discourse. Metaphor and Symbol, 21(1),
23–38.  doi: 10.1207/s15327868ms2101_2
Rychly, P. (2008). A lexicographer-friendly association score. In P. Sojka & A. Horák (Eds.), Pro-
ceedings of the 2nd Workshop on Recent Advances in Slavonic Natural Languages Processing,
RASLAN 2008. Brno: Masaryk University.
Semino, E. (2008). Metaphor in discourse. Cambridge: Cambridge University Press.
Shen, Y., & Balaban, N. (1999). Metaphorical (in)coherence in discourse. Discourse Processes,
28(2), 139–153.  doi: 10.1080/01638539909545077
Short, M. (1996). Exploring the language of poems, plays and prose. London: Longman.
Steen, G. (2008). The paradox of metaphor: Why we need a three-dimensional model of meta-
phor. Metaphor and Symbol, 23(4), 213–241.  doi: 10.1080/10926480802426753
 chapter 11

Mixing in pictorial and multimodal

metaphors?

Charles Forceville
University of Amsterdam

“Mixed metaphors” in language use two or more different source domains to

predicate something about the same target domain in a short stretch of discourse.
This often leads to unintendedly humorous results and is usually considered
bad style. Given that metaphors may be expressed pictorially or multimodally as
well as verbally, one may ask whether non-verbal modalities can also give rise to
metaphors of the “mixed” kind. If so, would such instances be considered odd,
humorous, or stylistically awkward? And what, if anything, would make such
“mixed metaphors” different from metaphoric blends with three input spaces (one
target and two sources)? The provisional conclusion is: we should, for the time
being, not adopt “mixed pictorial/multimodal metaphor” as a technical term;
but the discussion provides leads for further research from which both metaphor
theory and multimodal discourse analysis will benefit.

11.1  Introduction

Cognitive Metaphor Theory (CMT) claims that metaphors play a central role in
the way human beings conceptualize the world. Metaphorizing therefore is pri-
marily a mental activity and only derivatively a verbal one. Robust evaluation
of this trail-blazing idea requires taking into account work on non-verbal and
multimodal manifestations of metaphor. After all, an exclusive focus on verbal
metaphors may incur the objection that the supposed conceptual metaphors and
their verbal surface manifestations differ from one another only in that the former
are conventionally written in small capitals of the a is b type (see Gibbs and
Colston 1995, Cienki 1998, Forceville 2006). Even if this criticism should not be
considered valid, it remains necessary to study metaphors in other modalities than
language. Metaphors wholly or partly recruiting pictorial, gestural, sonic, olfactory
and/or tactile modalities after all in some respects behave differently than verbal

doi 10.1075/milcc.6.11for
© 2016 John Benjamins Publishing Company
 Charles Forceville

ones (e.g., Carroll 1996, Forceville 1996, Forceville and Renckens 2013, Kromhout
and Forceville 2013, Koetsier and Forceville 2014, Coëgnarts and Kravanja 2012,
Kappelhoff and Müller 2012). There are various reasons for this deviant behaviour.
The most important is that, while multimodal discourses have structure, they do
not have grammar. Therefore, while any aspect of verbal metaphor deserves to be
considered with its potential pertinence to its non-verbal and multimodal sisters
in mind, we should always be alert to where the latter may be systematically differ-
ent on the basis of mode-specific affordances and constraints.
In this chapter I will explore whether it makes sense to postulate that picto-
rial and multimodal metaphors can be “mixed.” If so, what form would or could
such metaphors take? Are they, like many verbal mixed metaphors, unintention-
ally humorous? Are they stylistically awkward, creatively playful, or unobtrusively
normal? Or is it better to reject the notion of pictorial/multimodal mixed meta-
phors altogether, and find alternative ways to account for metaphors that have,
what I will call, “multiple source domains”?

11.2  Characteristics of mixed verbal metaphors

Before addressing the question what mixed metaphors in non-verbal modalities

could look like, it is necessary to try and formulate more precisely what counts as
a mixed metaphor in language. A quick scan of the “mixed metaphors” found on
the site at http://therussler.tripod.com/dtps/mixed_metaphors.html (last accessed
22 May 2014; all verbal specimens discussed come from this source) reveals first of
all that by no means all examples given here qualify as mixed metaphors. The very
first one, “a car comes up behind you, flashing his horn,” surely only features a fun-
nily incorrect collocation: the user conflates “flashing a car’s lights” and “sounding
a car’s horn.” But in the absence of further context, this is to be taken as a mistake
in a literal, not a metaphorical sentence. The same holds for “a leopard can’t change
his stripes” (attributed to Al Gore). Others, such as “a rolling stone is worth two
in the bush” and “a stitch in time is worth a pound of cure” conflate two proverbs,
and while proverbs are themselves to be understood metaphorically (Lakoff and
Turner 1989: 160 et passim), these are not what we prototypically think of as mixed
metaphors either. By contrast, “a carpenter was the low rung on a totem pole”
(attributed to Charles Hodge) is a bona fide mixed metaphor, conflating status is
spatial position on a ladder and status is spatial depiction on a totem
pole. What the two source domains share is that they recruit the verticality
schema in the service of the structural metaphor high status is up/low status
is down. That the sentence nonetheless sounds odd is due both to the confusion
between two vertical (often: wooden) entities, and to the fact that the activity of a
 Chapter 11.  Mixing in pictorial and multimodal metaphors? 

person climbing is conflated with that of the static position of a carved face on a
totem pole. Here is another mixed metaphor: “a heart as big as gold” (attributed
to Kathy Scott), combining degree of empathy/generosity (metonymically
located in the heart) is size and degree of empathy/generosity is quality of
metal. Other ways of formulating the underlying conceptual metaphors are good
is big and good is gold. Both are conventional metaphors, the latter presum-
ably tying in with the “Great Chain of Being” metaphor (Lakoff and Turner 1989:
Chapter 3; see also Tillyard 1976).
What turns these into mixed metaphors is that two metaphors are squeezed
into a single grammatical expression. It is generally accepted within CMT that
there are many different source domains that can be used to help characterize
a single, complex target domain (Kövecses discusses this phenomenon as the
“range” of metaphor, 2010: 215). It is to be noted that using metaphors with dif-
ferent source domains one after the other is no problem. For instance, if instead
of saying, as above, “my sister has a heart as big as gold,” somebody would say,
“my sister has a heart of gold; really, she is such a generous person; her heart is
the size of a mountain,” this would already sound far less odd. And if there were
many more sentences in between the two metaphorical ones, the oddness would
further decrease. The Biblical Solomon’s Song is a true festival of metaphors all
celebrating the body parts and attributes of the beloved, and sometimes the same
body part is praised using two different source domains: “Thy two breasts are like
two young roes that are twins” (4.5 and 7.3) and a few verses later “This thy stature
is like to a palm tree, and thy breasts to clusters of grapes” (7.7). Surely we would
not complain here that this a case of mixing metaphors – the ecstatic lover seeks
different source domains to do full justice to his enthusiasm about the beloved’s
breasts. The metaphorizer thus presents distinct source domains after each other
rather than mixing them. In the same Bible book we come across an example
where the source domains appear even closer together: “A garden inclosed is my
sister, my spouse; a spring shut up, a fountain sealed” (4.12). Arguably the beloved
is described using three source domains: an enclosed garden, a blocked spring,
and a sealed fountain. But even here, we would not say that this is a mixed meta-
phor. It would have been if the sentence had contained for instance the phrase “a
spring enclosed.” What distinguishes a series of metaphors with shifting source
domains from mixed metaphors, I submit, is that the former are part of different
metaphor scenarios (Musolff 2006), and are deliberately envisaged as such by the
metaphorizer, while the latter are mistakenly presented as belonging to the same
metaphor scenario. Usually, these conflated scenarios appear within phrasal units
such as sentences or clauses, but this is not necessarily true, as a mixed metaphor
can straddle two sentences. Consider “The future of the church depends on pass-
ing the torch to the next generation. Tonight’s speaker is one who has taken hold
 Charles Forceville

of the baton” (ascribed to Robert Taylor). This praise confuses ensuring future
success is passing on a torch to a younger person and ­ensuring future
success is passing on a baton to the next runner in a relay race. Even
without further context we infer that the speaker thinks he remains within the
same metaphor scenario, the shared feature being the passing on of an important
object to another person. Nonetheless, in many (most?) cases, mixed metaphors’
conflation of source domain scenarios takes place within a clause or sentence.
Here a first problem arises for the question whether it makes sense to pos-
tulate the possibility of pictorial or multimodal mixed metaphors: there is no
“natural” equivalent in pictorial or multimodal discourse for what counts as a
clause or sentence in verbal discourse. Related to this is the issue of semantic
incompatibility. In the examples of mixed metaphor discussed above, we have
seen that there is a semantic incompatibility between the two metaphor sce-
narios. But it is doubtful that there is such a thing in pictures. Pictures, at least
realistic pictures, have structure in that for instance the location, size, orienta-
tion, and/or colour of their salient elements can usually not be randomly varied
without affecting their meaning, or simply making the whole incomprehensible.
But that is not the same as claiming that there is a set of rules determining how
visual elements must be strung together to make an acceptable whole in the way
a verbal grammar does this. For this reason the word “grammar” in Kress and
Van Leeuwen’s (2006) title Reading Images: The Grammar of Visual Design is only
acceptable when taken non-literally (see Forceville 1999). (Incidentally, there
is not only no visual syntax, there is also no equivalent in pictures of a verbal
vocabulary. We have dictionaries that, at a given moment, more or less exhaus-
tively describe all the words in a language, while we at best can have infinitely
incomplete thesauruses of pictorial elements.)

11.3  Mixed metaphor in pictures and multimodal discourse: Candidates

Before deciding on the impossibility of mixed pictorial and multimodal metaphor

straightaway on the ground that only language has grammar and can thus mix
metaphors, let me push matters by considering some cases that might nonetheless
qualify as fitting this label. The guiding idea will be that the criterion for consider-
ing a metaphor “mixed” is that it conflates two or more source domains predicat-
ing something about a single target domain within a stretch of discourse that is
experienced as, somehow, an undividable unit.
Case study 1 Consider Figure 11.1, a car ad for the VW Golf (I owe the exam-
ple and some of the discussion to Robin van Parera, a former student of mine). The
pictorial part of this advertisement features three semantic domains: (1) a motor;
 Chapter 11.  Mixing in pictorial and multimodal metaphors? 

(2) a compressed orchestra; (3) a work of art. Recognition of the domains of course
requires background knowledge: not everybody may recognize a motor in this
hybrid (although the text “perfectly tuned” may help identification), and one
has to have some experience visiting art museums to recognize the set-up of this
“thing,” suspended from the ceiling by thin cables and displayed against a white
wall, as an art object. And there may even be people who lack any familiarity with
the visual appearance of musical instruments, which then prevents them from rec-
ognizing this domain. But let us assume that all three scenarios are available to, as
well as activated by, the viewer.
Without the pragmatic knowledge that this is an advertisement for VW cars,
we would be at a loss whether any of the three domains qualifies for “target
domain” status. This knowledge, I submit is cued by the logo, the text, and the
place where this picture would typically be accessed: in a marked place in a
journal or magazine or as a billboard in an outdoor space designated for adver-
tising. My guess is that in the unlikely case we were to see this picture without
the logo and text, and completely decontextualized, say as a result in a google
image search, it would be up to us to privilege any of the three domains – a
choice (perhaps not even consciously made) that is triggered by whether we are
primarily art, music, or car lovers. If we were to privilege the art domain, we
might reason, “This is a piece of art resembling a car’s motor and consisting of
musical instruments.” We might then construe, and ponder an interpretation

Figure 11.1.  Advertisement for VW. Advertising Agency: DDB Stockholm, Sweden
 Charles Forceville

of, for instance, the metaphor motor is orchestra or of its reverse, orches-
tra is motor, without having to choose one over the other. This situation is
similar to the one in René Magritte’s series of “Le viol” paintings, in which we
need not, indeed cannot, choose between face is torso or torso is face (see
Forceville  1988). But the fact is that we do not see this picture without logo
and text, and that these latter “anchor” (Barthes 1986) the picture: they help us
zoom in on its salient elements. Consequently we realize that this is an ad for a
VW car, and that since motors have a more direct metonymical relation to cars
than either orchestras or artistic installations we are to construe a metaphor
here: vw’s motor is orchestra/artistic installation. The music domain is
further anchored by the pay-off: “Golf R32. Perfectly tuned,” which allows for a
further mapping onto the target domain.
This ad, then, appears to fit the criterion that (1) a metaphor needs to be
construed (2) which draws on two source domains at the same time, and (3) is
part of a single “gestalt.” It would be quite natural to say that the VW motor
is simultaneously an orchestra and an art installation, both of which provide
mappings to the motor domain, and by extension to the VW Golf R32. I pro-
pose, then, that for the time being we consider this a multimodal equivalent of
a mixed verbal metaphor (the logo and anchoring text make this into a multi-
modal rather than a purely pictorial metaphor, see Forceville 2006, Eggertsson
and Forceville 2009, Bounegru and Forceville 2011 for discussion). If this is
accepted, we have to conclude that there is nothing unintendedly funny, awk-
ward, or clumsy about the mixed metaphor – on the contrary, I submit that it
is a very creative one.
Case study 2 Figure 11.2 shows another hybrid whose potential qualifi-
cation as a visual “mixed metaphor” is food for thought. As in case study 1,
the picture triggers three semantic domains: bear, shark, and octopus, but
none of these is “privileged” as target domain. It is only the verbally cued con-
text, namely a site of one Matt Ufford called “Nine badass hybrid monsters that
deserve their own syfy [sic; presumably “scifi” is meant, ChF] original movie,”
which suggests that we should construe a metaphor with “monster” as target
domain, and “bear,” “shark,” and “octopus” as its various source domains. It is
relevant to remember here that a metaphor cannot only be used to provide a
new perspective on a pre-structured target domain; it can also impose that very
structure itself upon a target domain that has little or no structure of its own
(Indurkhya 1991). Many monsters acquire their identity precisely by mappings
from one or more other source domains that trigger negative, awe-inspiring,
fearsome etc. connotations. The three source domains chosen suggest that the
monster is strong, predatory, with sharp teeth, and can perhaps strangle its
victims using its tentacles.
 Chapter 11.  Mixing in pictorial and multimodal metaphors? 

Figure 11.2.  “bearsharktopus” http://www.uproxx.com/feature/2010/02/nine-badass-

hybrid-monsters-that-deserve-their-own-syfy-original-movie/ (last accessed 22 May 2014)

Of course in the science fiction and fantasy genres, there is far more freedom
to create metaphorical hybrids than in other discourses, but this example, how-
ever odd, conforms to the criteria of a metaphor with an attestable target domain
(“monster”) that is simultaneously to be understood in terms of three different
fear-inspiring animals, each of them triggering its own species-specific dangerous
mappings, culminating in a single visual “gestalt.” Since the target domain is ren-
dered verbally in the context, namely in the URL’s title, this is a multimodal rather
than a purely pictorial mixed metaphor. It is worth commenting upon that none
of the three source domains in this examples appear to be privileged; all of them
are of equal importance. (This example reminded an anonymous reviewer of an
­earlier draft of this chapter of the “ManBearPig” instalment of South Park, which
satirizes Al Gore’s documentary An Inconvenient Truth about the dangers of global
warming. The Gore character keeps warning against a mythical threat to human-
ity called “­ ManBearPig,” and he chides someone for referring to this monster as
­“PigBearMan.” By insisting on the “right” order of the elements in the mythical
creature’s name, Gore seems to suggest a degree of structure in the monster that
nobody else in this South Park instalment perceives, and this joke thus further helps
poke fun at Gore.)
Case study 3 For case study 3 I re-visit a film scene I was alerted to in a paper
by Anton Kanis, a former student of mine. It occurs in the film American ­Psycho
(Mary Harron, USA 2000). In this justly popular scene Patrick Bateman, the
eponymous hero, and his colleagues outbid each other with the stylishness of their
business cards in the context of a ruthless alpha-male business ­environment. It is
completely possible to enjoy the scene at face value as a ridiculous fight between
 Charles Forceville

over-paid, empty-headed businessmen, where the metaphor businesscard

exchange is battle is triggered. However, one could on the basis of certain ges-
tures and sounds understand the battle in more precise metaphoric terms. One
man brandishes his card as if attacking his interlocutor with a knife, an action
accompanied by a “swoosh”-sound, while a close up shows a silver cardholder
flicking open as one would open a stiletto. Patrick himself, overpowered by the
astonishing elegance of “Paul Allen’s card” that he has been handed, drops it as
if wounded. So a viewer that picks up these cues may construe the metaphor
­business card is knife or comparing business cards is knifing duel.
But it is also possible to understand the scene slightly differently, building
on the activation of a different battle domain: that of a card game (Antonio
­Barcelona, Rosario Caballero, personal communication at the Researching and
Applying Metaphor conference in Leeds, 2006). This interpretation is enabled by
the fact that business cards have a rectangular form, just as playing cards; that they
are called “cards” too; that the cards are thrown on the table in a manner remi-
niscent of card-playing, and that card-playing cues the idea of “trumping.” Con-
sciously or subconsciously activating the card-playing (or even more specifically:
poker-playing) source domain activates different aspects of the target domain than
the knifing duel scenario: psychological rather than physical warfare. As a mat-
ter of fact, the possibilities do not stop there. We can also see the men as young
boys getting their thingies out of their flies to see “who’s got the biggest” ….
Whether three, two, one, or none of these source domains are activated
depends on viewers’ familiarity with the source domains as well as on their ability,
at the moment of watching the scene, to activate them. Let us assume that there
are viewers who cue all three source domains more or less simultaneously: com-
paring business cards is (1) knifing duel; (2) playing poker; (3) comparing
dicks. We could then say that the situation is not really different from the one in
the previous case studies: we here have a metaphor that is to be verbalized as a is b,
c and d. Activation of all three source domains enhances the richness of the meta-
phor rather than make it seem stylistically awkward or inept. What aids this posi-
tive assessment is that there is nothing “homospatially non-compossible” (Carroll
1996) about the conflation of the source domains (unlike in Case study no. 2):
the audio-visual cues potentially trigger three different mental scenarios that can
serve as source domains of a metaphor, but even activating all three of them does
not make anything in the scene impossible or even improbable.
Case study 4 The examples discussed in the first three case studies are spec-
imens  of creative metaphor in the sense of Black (1979) and Forceville (1996).
If we provisionally accept them as examples of “mixed metaphors,” the question
now presents itself whether structural metaphors, too, can be of a mixed nature.
­Kövecses (1986, 2000, 2008) analyses and charts the source domains that are used
 Chapter 11.  Mixing in pictorial and multimodal metaphors? 

to metaphorically structure emotions. The overarching metaphor he identifies is

emotions are physical forces (2008: 385). But physical forces come in many
forms. In his analyses of anger, Kövecses (1986) mentions anger is hot fluid in
a pressurized container as one of the most salient varieties. In one paper I inves-
tigated visual manifestations of anger in a comics album, Asterix and the Roman
Agent, and conclude that my findings are all “at least commensurate” (Forceville
2005: 80) with those of Kövecses. This is particularly true for his anger is a hot
fluid in a pressurized container metaphor (see also Eerden 2009, F ­ orceville
2011a). But in fact, one could also distinguish in some examples I discussed
manifestations of other metaphors Kövecses identified: anger is a ­dangerous
­animal, with as one of the correspondences “the aggressive behavior of the
­dangerous animal is angry behavior” (Kövecses 1986: 25). Indeed Kövecses him-
self already indicated that a phrase such as “he breathed fire” appears to draw on
both the pressurized container metaphor (although it is a gaseous rather than a
fluid entity that gets out) and the dangerous animal metaphor – specifically per-
taining to the behavior of a dragon (Ibidem: 23). Yet another metaphor Kövecses
identifies is anger is a burden (Ibidem: 27). Now consider Figures 11.3 and 11.4.

Figure 11.3.  Alcazar is angry (panel in Hergé, Tintin et les Picaros, Casterman 1974)

Figure 11.4.  Kwabbernoot is angry (panel in Franquin & Jidéhem, Flaterfestijn, Dupuis 1977)
 Charles Forceville

Clearly in both cases it makes sense to analyze the metaphor in terms of the anger
is hot fluid in a pressurized container: the anger that is bottled up inside the
container of the body tries to get out. In Alcazar’s case it does come out in the form
of a loud “enough!”; Kwabbernoot succeeds, at least for the time being, in keeping
the anger inside. The visual cues for this assessment include facial expressions (and
a lobster-red face in Figure 11.4) and various pictorial runes (spirals and droplets),
the “smoke” pictogram (Figure 11.4), and the jagged text balloon (­ Figure 11.3; for
more discussion of these issues, see Forceville 2005, 2011a, F ­ orceville et al. 2010).
But is it not also possible to understand Alcazar’s expression, with open mouth
and bared teeth (Figure 11.3) as cueing angry human behaviour is aggressive
animal behaviour? And does Kwabbernoot’s huddled position (Figure 11.4) not
also suggest that anger is a burden? And if these metaphorical interpretations
are both acceptable, why should we have to choose? I think that a viewer can well
simultaneously (albeit not necessarily consciously) activate, in Figure  11.3, the
hot fluid metaphor and the angry animal metaphor, and, in Figure 11.4, the
hot fluid metaphor and the burden metaphor. If we are ready to accept that in
both cases not one, but two domains are activated, here, too, we could speak of
mixed metaphors very much as in the previous three case studies.
Case studies 5 and 6 Let me end by considering two animation films I have
discussed before (in Forceville 2011b), though not from the perspective of possi-
bly mixed metaphors. The first is the wordless The Life (Jun-ki Kim, South Korea
2003, 9′45″ last accessed at http://www.youtube.com/watch?v=faArmZxwtT0
on 22 May 2014). We see a middle-aged man with his son, in a basket on his
back, climbing up what turns out to be a huge totem pole. This is a very difficult,
exhausting activity, which moreover is dangerous: the father is sometimes ham-
pered by unfavourable weather conditions, and he always needs to be careful to
maintain his balance. In one shot we see other totem poles in the background,
which presumably are also being climbed. As the story develops, the man and
his son grow older, and at one stage the son takes the lead. When the father is so
old he cannot go on climbing, he remains behind, probably to die. At the end the
son, now a grown-up man himself, arrives at the top. There he unpacks a stone
that he and his father had carried with them all the time, and fits it in an empty
slot at the top of the totem pole. We now understand that the son and his father
have been preceded by numerous ancestors that each helped build the family
totem pole.
The “climbing” is an instance of purposive activity – in fact it is what the
men spend their lives doing. So one metaphor that we are invited to activate is
­purposive activity is self-propelled movement toward a destination,
which is a specific formulation of the more general life is a journey. But there
is a second source domain that helps structure the inferences viewers are to make
 Chapter 11.  Mixing in pictorial and multimodal metaphors? 

to interpret this existential film: purposive activity is making an artifact. It

is ­noteworthy that the two source domains interact. Viewers undoubtedly quickly
realize they have to activate the journey metaphor in their sense-making activ-
ities; but it is not until the very end that they realize the precise nature of the
purpose for which the journey has been undertaken. The metaphor purposive
activity is making an artifact could well have been used independently, but
as it is recruited here, the two source domains seamlessly merge, and their combi-
nation makes for an impressive, moving result.
The second example is the (again: wordless) Polish animation film The Road/
Droga (Mirosław Kijowicz, Poland 1971, 4′23″). We see a man, from behind, walk-
ing on a road. When he comes at a Y-crossing, he hesitates whether to go left or
right. Eventually he splits himself into two halves; the left half of his body takes
the left turn whereas the right half takes the right turn. When after some time
the roads converge again, and the man wants to unite his two halves, he succeeds
indeed in doing this, but he finds that there is no longer a perfect match, since his
left half has grown while his right half has retained its original size. He walks on as
best as he can in his new, ill-matched body.
As in The Life, the dominant metaphor is purposive activity is self-­
propelled movement toward a destination. Even though we never get
to know what exactly the destination of the man in The Road is, I propose that
activating this metaphor is a precondition for finding any interest in this short
film. Often, the journey domain is exploited to structure metaphorically a target
domain that can variously be described as searching for meaning in life, or
looking for one’s role in life, or searching for one’s identity in life (see
Forceville 2006, 2011b, 2013; Forceville and Jeulink 2011). As in any story, we
expect some crisis to befall the protagonist. In this case the protagonist is faced
with an either/or choice. This is a problem given the structural metaphor iden-
tity is wholeness. By splitting himself into two, the fragmentation in his body
maps metaphorically onto a fragmentation in his identity. When the two halves
meet, they cannot merge into the whole that existed before.
My argument here hinges on the assumption that appreciation of the film
depends both on the journey and on the wholeness metaphor. I submit that
it is not mere word play to formulate what is going on as a combination of find-
ing one’s identity in life is a journey and finding one’s identity in life is
(maintaining/seeking) wholeness (see also Koetsier and Forceville 2014). If
this is accepted, we have the situation in which a given target domain is combined
with two source domains. Given the brief of this chapter, the crucial question
would be: but are the two source domains simultaneously activated? Is the entirety
of the film similar to a stretch of verbal text in which two metaphors are conflated
in such a way as to call them “mixed?”
 Charles Forceville

This is a difficult question. In the beginning of the film, only the journey
metaphor is active – just as in The Life. However, it is the fragmented iden-
tity metaphor that provides the twist that we expect in a story and that makes it
creative and memorable. For a satisfactory interpretation we need to recruit both
metaphors. So on these grounds, one might say that at least from the moment of
the “bodily split” onwards, it would be defensible to label this an instance of mixed
metaphor. If we do so, it is crystal clear once more that the result is anything but
awkward or odd or the pictorial equivalent of “ungrammatical”; on the contrary,
despite (or thanks to) the simplicity of pictorial execution of The Road, we here
have a very profound, thought-provoking mixed metaphor.

11.4  Mixed metaphors and Blending Theory

In a mixed metaphor what happens is that a certain target domain is understood

not in terms of a single source domain, but in terms of two or more (conflated)
source domains. It is worth noticing that this peculiarity (more than one source
domain is activated) means that mixed metaphors can be nicely modeled in terms
of metaphoric blends (Fauconnier and Turner 2002). As I have argued elsewhere
(Forceville 2004), Blending Theory (or “conceptual integration theory”) has hith-
erto insufficiently exploited one very important difference with Conceptual Meta-
phor Theory: the ability of a blend to accommodate more than two input spaces.
While a conventional metaphor has no more and no less than two “input spaces,”
namely a target domain and a source domain, the blending model allows for mod-
eling metaphors with three or more input spaces: a target domain/input space and
at least two source domains/input spaces.
This holds both for the verbal mixed metaphors discussed in Section 2 and for
the creative visual case studies discussed in Section 3 (VW-ad, “Bearsharktopus,”
and American Psycho scene): all of them require or invite the viewer to construe
metaphors with two source domains, each of which imparts unique features to
the “blended space” – which in metaphorical blends doubles as the target domain.
What turns the verbal mixed metaphors into laughable blunders is that their mak-
ers were supposedly not aware that they conflated two source domains; they cre-
ated, one could say, “metaphorical contaminations.” By contrast, the creative visual
examples reflect hybrids that were intentionally created. But it is worth pointing
out that both the unintentionally humorous verbal examples and the emphatically
created visual ones result in a richer target domain than if the metaphor would
have accommodated mappings from just one source domain.
Please note that all examples count as metaphorical blends, as this highlights
the usefulness of bringing metaphor theory and blending theory together in
 Chapter 11.  Mixing in pictorial and multimodal metaphors? 

­ iscussing these hybrids. What metaphor theory brings to them is that we can
d
clearly distinguish and label a target domain; what blending theory brings to them
is that it allows for two (or more) source-domain-input spaces that jointly impose
(new) structure upon this target domain. I dwell on this issue because we should
never forget that the category of blends is the overarching category, within which
metaphors are a subcategory (counterfactuals such as exemplified in “if I were
you …” being another). This point is made more explicitly by Grady et al. (1999)
than by Fauconnier and Turner (2002). For instance, Fauconnier and Turner in
several publications discuss “landyacht” as a blend between a car and a ship. But
while these latter are the input spaces, we can still distinguish a target and a source
in this blend, as it presents a car in terms of a yacht, not a yacht in terms of a car.
Indeed, in Forceville (2012: 115) I claim that as far as metaphors are concerned,
­Fauconnier and Turner’s (2002) blending model is a sophisticated formalization of
Black’s (1979) “interaction theory” of metaphor.
In short, the creative visual or multimodal examples of mixed metaphor dis-
cussed in Section  4 can all be considered as metaphorical blends with three or
more input spaces: one target and two or more source domains. There is a differ-
ence, though, between the first two and the third examples in that in the first two
the various source domains are more emphatically cued than in the third.
Can we say about the structural metaphors in case studies 4–6, too, that they
could be modeled in a Blending Theory format? In Figures 11.3 and 11.4 we can
indeed postulate the pressurized container, angry animal, and burden
source domains as input spaces that, in different combinations, help structure
the target domain anger. As in their structural verbal counterparts, however, the
mixing of source domains is probably not so much an intentional act as a “natural”
way of representing this emotion (for which the artists cannot help draw sponta-
neously on deep-rooted embodied metaphor; cf. Lakoff 1986 for a discussion of
everyday language usage that draws on conceptual metaphors).
For the two animation films, too, both source domains in each case can be mod-
eled as input spaces structuring the journey target domains/blends. Nonetheless,
the mixing of structural metaphors in the two seems to be of a different nature in
that here the two the source domains do not provide mere enrichments of the tar-
get domain; they complement each other. We do not miss any essential information
whether we activate anger is hot fluid in a pressurized container or angry
behavior is a wild animal’s aggressiveness or anger is a burden in case
study 4. But a satisfactory interpretation of the two animation films in case studies
5 and 6 requires the activation of both source domains – j­ ourney and building,
and journey and wholeness, respectively – as well as the Vital ­Relation of “time
compression” presented by Fauconnier and Turner (2002:  Chapter 5), since both
films present substantial stretches of the protagonists’ lives in mere minutes.
 Charles Forceville

11.5  Concluding remarks and further research

I have argued in this chapter that mixed metaphors in the verbal modality exem-
plify targets that are combined with two or more different source domains in the
same short stretch of discourse. What makes the few verbal examples discussed
unintentionally funny is that their speakers seem to be unaware of drawing on
two source domain scenarios, rather than on one. The pictorial examples I have
discussed fit the first part of this condition, but not the second part. They are not
“unintentionally funny”; they are intended and (more or less) clever or even admi-
rably creative.
So let me come back to the central question of this chapter: does it make
sense to postulate the existence of “mixed metaphors” in pictures? For the time
being, my answer is “not really.” Language has grammar and conventional ways of
combining elements within a single semantic unit – be it phrase, clause, sentence,
paragraph, or text – that are too radically different from the “structure” that is its
pictorial equivalent to warrant the use of the label “mixed pictorial/multimodal
metaphor.” Rather than here introduce such a half-baked concept, I would pro-
pose that the examples analysed in Section 4 can well be described using existing
terminology. Thus, we could either say that these are metaphorical blends drawing
on two or more source-domain input spaces, or alternatively that we sometimes
encounter discourse situations in which a given target domain is coupled simulta-
neously with two or more source domains; for the latter formulation I propose the
term “multiple source domain metaphors.”
Even if my suggestion to abandon the notion of “mixed visual/multimodal
metaphors” in favour of “multiple source domain metaphors” is accepted, this
chapter has hopefully yielded some insights that can spawn further research. In the
first place, it is worth considering whether multiple source domain metaphors are
somehow typical of certain genres or developments. The metaphor in the VW ad
(Figure 11.1) is unusual; I have hitherto rarely seen specimens of the a is b and c.
But this may well become a new strand in advertising metaphors. Despite the rela-
tively realistic nature of the American Psycho example, I would guess that multiple
source domain metaphors are more likely to arise in films that verge toward fan-
tasy than in more realistic genres. Science fiction is full of metaphorical hybrids
that can draw on more than one source domain, while the medium of animation
easily allows transformations of a given phenomenon in a series of other phenom-
ena in such a way that we could see this is a chain of metaphors with the same tar-
get (note that we would verbalize this as a is b and then a is c – or perhaps b is c).
A second pertinent parameter is monomodality versus multimodality. The
examples discussed under case studies 4, 5, and 6 are monomodal in the sense that
both target and sources are rendered predominantly in the pictorial modality. But
 Chapter 11.  Mixing in pictorial and multimodal metaphors? 

the metaphors in The Life and The Road both draw on the sonic modality to cue
the strong wind (The Life) and the protagonist’s footsteps (The Road) – which are
metonymically related to the journey domain. By contrast, the first example is
truly multimodal, at least if we consider the VW logo as verbal rather than picto-
rial: VW motor is orchestra/art installation. But the verbalization motor
is orchestra/art installation would be monomodal. The bearsharktopus
in case study 2 hovers between being monomodal and multimodal – depending
on whether we need the accompanying text to supply the target domain monster.
An overall question here could be: can any patterns be detected in when (if at all)
a multiple source domain metaphor has source domains that are rendered in two
or more different modalities?
A third issue that invites further research first requires the compilation of a
corpus of multiple source metaphors of the subtle creative kind of case studies 1
and 3 and experimentally test which of the source domains is cued (first). One
could imagine that (sub)cultural background plays a role in the preferences here.
Monitoring which source domain is cued (first) allows insight into the specific
background information certain communities of people bring to their interpreta-
tion of metaphors.
In the structural type of metaphors with multiple source domains, finally, we
might want to examine more systematically which source domains can actually be
“mixed,” and which cannot. This may allow us to say something about the conditions
under which (or the genres in which) specific source domains are typically clus-
tered. This in turn feeds into theorizing about image schemata (see Hampe 2005).

Acknowledgments

I am indebted to Ray Gibbs for challenging me to reflect on the issue of “mixed metaphors” in
pictorial and multimodal discourse, and to two anonymous reviewers for their suggestions and
comments. Much of this paper was written on the outdoor terrace of the Carmen de la Victoria,
with a view of the Alhambra, during my stay as visiting scholar at the University of Granada,
Spain, in March 2012. I warmly thank Belén Soria for hosting this visit.
*Every effort has been made to trace copyright holders of Figure 11.1 and Figure 11.4 to
obtain their permission for the use of copyright material. The publisher regrets any unwitting
infringements.

References

Barthes, R. (1986 [1964]). Rhetoric of the image. In The Responsibility of forms (pp. 21–40).
Trans. by R. Howard. Oxford: Blackwell.
 Charles Forceville

Black, M. (1979). More about metaphor. In A. Ortony (Ed.), Metaphor and thought (pp. 19–43).
Cambridge: Cambridge University Press.
Bounegru, L., & Forceville, C. (2011). Metaphors in editorial cartoons representing the global
financial crisis. Visual Communication, 10, 209–229.  doi: 10.1177/1470357211398446
Carroll, N. (1996). Theorizing the moving image. Cambridge: Cambridge ­University Press.
Cienki, A. (1998). Metaphoric gestures and some of their relations to verbal metaphoric expres-
sions. In J.-P. Koenig (Ed.), Discourse and cognition: Bridging the gap (pp. 189–204).
­Stanford, CA: Center for the Study of Language and Information.
Coëgnarts, M., & Kravanja, P. (2012). From thought to modality: A theoretical framework
for analysing structural-conceptual metaphor and image metaphor in film. Image [&]
­Narrative, 13(1), 96–113.
Eerden, B. (2009). Anger in Asterix: The metaphorical representation of anger in comics
and animated films. In C. Forceville & E. Urios-Aparisi (Eds.), Multimodal metaphor
(pp. ­243–264). Berlin: Mouton de Gruyter.
Eggertsson, G. T., & Forceville, C. (2009). The human victim is animal metaphor in hor-
ror films. In C. Forceville & E. Urios-Aparisi (Eds.), Multimodal metaphor (pp. 429–449).
­Berlin: Mouton de Gruyter.
Fauconnier, G., & Turner, M. (2002). The way we think: Conceptual blending and the mind’s hid-
den complexities. New York: Basic Books.
Forceville, C. (1988). The case for pictorial metaphor: René Magritte and other Surrealists.
In A. Erjavec (Ed.), Vestnik 9 (pp. 150–60). Ljubljana, YU: Inštitut za Marksistične
Študije.
Forceville, C. (1996). Pictorial metaphor in advertising. London: Routledge.
doi: 10.4324/9780203272305
Forceville, C. (1999). Educating the eye? Kress and Van Leeuwen’s Reading images: The grammar
of visual design (1996). Language and Literature, 8, 163–178.
doi: 10.1177/096394709900800204
Forceville, C. (2004). Review of Fauconnier and Turner (2002). Metaphor and Symbol, 19,
83–89.  doi: 10.1207/S15327868MS1901_5
Forceville, C. (2005). Visual representations of the idealized cognitive model of anger in the
Asterix album La Zizanie. Journal of Pragmatics, 37, 69–88.
doi: 10.1016/j.pragma.2003.10.002
Forceville, C. (2006). Non-verbal and multimodal metaphor in a cognitivist framework: ­Agendas
for research. In G. Kristiansen, M. Achard, R. Dirven, & F. Ruiz de Mendoza Ibáñez (Eds.),
Cognitive linguistics: Current applications and future perspectives (pp. 379–402). ­Berlin:
Mouton de Gruyter.
Forceville, C. (2009). Pictorial and multimodal metaphor in fiction film, Lecture 5/8 of Course
in pictorial and multimodal metaphor. http://semioticon.com/sio/courses/pictorial-multi
modal-metaphor/
Forceville, C. (2011a). Pictorial runes in Tintin and the Picaros. Journal of Pragmatics, 43,
­875–890.  doi: 10.1016/j.pragma.2010.07.014
Forceville, C. (2011b). Structural pictorial and multimodal metaphor. ­Lecture 7/8 of Course
in pictorial and m­ ultimodal metaphor. http://semioticon.com/sio/courses/pictorial-multi
modal-metaphor/
Forceville, C. (2012). Creativity in pictorial and multimodal advertising metaphors. In R. Jones
(Ed.), Discourse and creativity (pp. 113–132). Harlow: Pearson/Longman.
 Chapter 11.  Mixing in pictorial and multimodal metaphors? 

Forceville, C. (2013). Metaphor and symbol: searching for one’s identity is looking for a
home in animation film. Review of Cognitive Linguistics, 11, 250–268. (Special issue on mul-
timodality and linguistics issue, guest editor: Maria Jesús Pinár). doi: 10.1075/rcl.11.2.03for
Forceville, C., Veale, T., & Feyaerts, K. (2010). Balloonics: The visuals of balloons in comics. In
J. Goggin & D. Hassler-Forest (Eds.), The rise and reason of comics and graphic literature:
Critical essays on the form (pp. 56–73). Jefferson NC: McFarland.
Forceville, C., & Jeulink, M. (2011). The flesh and blood of embodied understanding: The
source-path-goal schema in animation film. Pragmatics & Cognition, 19, 37–59.
doi: 10.1075/pc.19.1.02for
Forceville, C., & Renckens, T. (2013). The good is light and bad is dark metaphors in feature
films. Metaphor and the Social Word, 3, 160–179.  doi: 10.1075/msw.3.2.03for
Gibbs, R. W., Jr., & Colston, H. L. (1995). The cognitive psychological reality of image schemas
and their transformations. Cognitive Linguistics, 6, 347–378.  doi: 10.1515/cogl.1995.6.4.347
Grady, J., Oakley, T., & Coulson, S. (1999). Blending and metaphor. In R. W. Gibbs, Jr., &
G.  J.  Steen (Eds.), Metaphor in cognitive linguistics (pp. 101–124). Amsterdam: John
­Benjamins.  doi: 10.1075/cilt.175.07gra
Hampe, B. (Ed.). (2005). From perception to meaning: Image schemas in cognitive linguistics.
Berlin: Mouton de Gruyter.  doi: 10.1515/9783110197532
Indurkhya, B. (1991). Modes of metaphor. Metaphor and Symbolic Activity, 6, 1–27.
doi: 10.1207/s15327868ms0601_1
Kappelhoff, H., & Müller, C. (2012). Embodied meaning construction: Multimodal metaphor
and expressive movement in speech, gesture, and feature film. Metaphor and the Social
World, 1, 121–153.  doi: 10.1075/msw.1.2.02kap
Koetsier, J., & Forceville, C. (2014). Embodied identity in werewolf films of the 1980s. Image [&]
Narrative, 15(1), 44–55.
Kövecses, Z. (1986). Metaphors of anger, pride and love. Amsterdam: John Benjamins.
doi: 10.1075/pb.vii.8
Kövecses, Z. (2000). Metaphor and emotion: Language, culture, and body in human feeling. Cam-
bridge: Cambridge University Press.
Kövecses, Z. (2008). Metaphor and emotion. In R. W. Gibbs, Jr. (Ed.), The Cambridge handbook
of metaphor and thought (pp. 380–396). Cambridge: Cambridge University Press.
doi: 10.1017/CBO9780511816802.023
Kövecses, Z. (2010). Metaphor: A practical introduction. 2nd ed. Oxford: Oxford University
Press.
Kress, G., & van Leeuwen, T. (2006 [1996]). Reading images: The grammar of visual design. 2nd
ed. London: Routledge.
Kromhout, R., & Forceville, C. (2013). life is a journey: the source-path-goal schema in the
videogames Half-Life, Heavy Rain, and Grim Fandango. Metaphor and the Social World, 3,
100–116.  doi: 10.1075/msw.3.1.05for
Lakoff, G. (1986). The meanings of literal. Metaphor and Symbolic Activity, 1, 291–296.
doi: 10.1207/s15327868ms0104_3
Lakoff, G., & Turner, M. (1989). More than cool reason: A field guide to poetic metaphor. Chicago:
University of Chicago Press.  doi: 10.7208/chicago/9780226470986.001.0001
Musolff, A. (2006). Metaphor scenarios in public discourse. Metaphor and Symbol, 21,
23–38.  doi: 10.1207/s15327868ms2101_2
Tillyard, E. M. W. (1976 [1943]). The Elizabethan world picture. Harmondsworth: Penguin.
 chapter 12

Extended metaphor in the web of discourse

Anita Naciscione
Latvian Academy of Culture

This chapter explores extended metaphor in the cognitive stylistic framework.

Extended metaphor defines as an entrenched stylistic pattern of both thought
and language, reflecting extended figurative thought. It is a cognitive inference
tool, applicable in new figurative thought instantiations. A metaphor can be
extended only by extension of its metaphorical image: by creating a metaphorical
sub‑image or a string of sub-images, which relate metonymically by associations
of contiguity. Metonymy is invariably present in each instantiation of
metaphorical extension. Thus, extended metaphor is “mixed” by definition.
Metaphor is not alone in figurative meaning construction. Apart from
metonymy, extended metaphor may incorporate other figurative modes (pun,
allusion, personification, euphemism, hyperbole, irony), forming figurative
networks and representing a process and a result of human thought and a
conceptualisation of experience. Only a detailed semantic and stylistic analysis
will reveal the interaction and interrelationships of direct and figurative meanings
in the web of discourse, which is not a “mix” but a natural flow of figurative
thought in natural discourse.

12.1  Introduction: A cognitive perspective

Cognitive research has established that figurative language forms part of human
cognitive processes. People think and conceptualise their experience and the
external world in figurative terms. In the cognitive linguistic view, one of the most
important tenets is that metaphor constitutes an inherent part of thought, lan-
guage and culture.
Stylistic patterns appear to be linked to the immemorial past of human-
ity of which we have no written records. Metaphor and extended metaphor are
­centuries-old techniques of both thought and language. My aim is to explore
extended metaphor as a stylistic pattern from the cognitive stylistic perspective.
As to the English language, extended metaphor has existed throughout the course

doi 10.1075/milcc.6.12nac
© 2016 John Benjamins Publishing Company
 Anita Naciscione

of its ­history as a way of thinking. Old English texts and dictionary attestations
reveal that written records of extended metaphor go back to the earliest stages
of existence of written text in English, that is, to the eighth century. Significantly,
while the verbal manifestation of an extended metaphor does not repeat itself, pat-
tern does. What patterns have in common is that the development of metaphorical
thought forms new extensions in real contexts, using the same language means.
This chapter attempts to provide insights into the common properties of
extended metaphor that emerge in actual language use. Observation and analysis
are based on a large corpus of stylistic use of lexical metaphors and metaphori-
cal phraseological units (PUs)1 in discourse. Exploration of extended metaphor
across sentence boundaries is supported by ample textual illustrations of stylis-
tic use ranging from Old English (OE) to Modern English (MoE). The chapter is
structured as follows. Section 2 argues for pattern of stylistic use as a structure of
thought and a cognitive inference mechanism that is reproduced in new diverse
stylistic instantiations. Extended metaphor is one of the most widespread stylistic
patterns in actual use. Section 3 brings out the variety of discoursal manifesta-
tions of extended metaphor, presents its common types and uncovers the role of
metonymic links in the extension of the metaphorical image. A study of metaphor
in discourse naturally leads to the concept of extended metaphor. In Section 4, a
diachronic approach reveals the stability of the pattern of extended metaphor. The
pattern is viewed as part of the mental lexicon, stored in the long-term memory of
the language user. Section 5 takes a closer look at sustainability of extended meta-
phor that is regarded as a natural phenomenon. As thought is sustainable, so is
figurative language. A sustained metaphorical image acquires a discourse dimen-
sion and promotes semantic and stylistic cohesion of the text. This Section exam-
ines the involvement of other tropes in an extended metaphorical instantiation:
metonymy, pun and allusion. Section 6 presents a case study of multimodal repre-
sentation of figurative thought, involving stylistic techniques from more than one
semiotic mode of expression. Extended metaphor exists not only in thought and
language but also in visual representation and its perception. Finally, in Section 7,
I argue against the terms “mixed metaphor”, “impermissible mixed metaphor” and
“contaminated metaphor”. In this chapter, I do not deal with the beliefs of prescrip-
tive linguistics. I am interested in the use of metaphor in real discourse environ-
ments. The illustrations analysed reveal that the flow of figurative thought does not
obey the rules of formal logic. The use of metaphor along with other metaphors

.  I hold that the phraseological unit is a stable, cohesive combination of words with a fully
or partially figurative meaning.
 Chapter 12.  Extended metaphor in the web of discourse 

and stylistic patterns is a natural discourse phenomenon. That is why I view the
traditional notion of “mixed metaphor” (and hence the term) as inappropriate.2
The analysis of my empirical material brings me to the conclusion that
extended metaphor is a mode of reflecting extension of figurative thought; it is
one of the patterns by which people conceptualise their thoughts, feelings, and
experiences. Thus it is an integral part of our conceptual system.

12.2  Stylistic pattern of extended metaphor as a structure of thought

Stylistic pattern is an essential element in figurative meaning construction. It is

a cognitive inference tool and a mental stylistic technique which is applicable
in new figurative thought representations. Pattern is an archetypal conception
and an integral part of mental cognitive structures. As “an abstract framework”
(Gibbs 2003), pattern helps to form new creative instantiations in use. Patterns are
used to construct meaning.
Stylistic pattern forms a set of common rules of use in discourse. Each pat-
tern is characterised by a number of formal and semantic features which are com-
pulsory for new instantiations designed on the basis of the pattern. As typified
recurring techniques, patterns are elements of the language system which can be
reproduced. Thus, patterns of stylistic use are reproducible elements, generating
innumerable particular manifestations in discourse (Naciscione 1982, 2010).
Extended metaphor is one of the most widespread stylistic patterns in instan-
tial stylistic use3 in most discourses. It is a characteristic of the figurative mind:
one of the figures of thought that is used to conceptualise experience. For instance,
John Fowles’ The French Lieutenant’s Woman is a love story focusing on the com-
pelling mystery and passionate relationship between Charles and Sarah:
fire4
He had thought by his brief gesture and assurance to take the first step towards
p u t t i n g o u t the fire the doctor had told him he h a d l i t ; but when one is

.  In the cognitive linguistic view metaphorical terms are theory constitutive metaphors. See
Gibbs [1994] 1999: 169–172.
.  Instantial stylistic use is a particular instance of a unique stylistic application of a word
or a phraseological unit in discourse resulting in significant changes in its form and meaning
determined by the thought and the context.
.  In this chapter, stylistic instantiation has been highlighted for emphasis: base forms are
marked bold and underlined; i n s t a n t i a l e l e m e n t s are spaced and underlined; ­r eplaced
­elements are underlined double and spaced; cues are marked with an ­interrupted line.
 Anita Naciscione

oneself t h e f u e l , f i r e f i g h t i n g is a hopeless task. Sarah was a l l f l a m e .

Her eyes were a l l f l a m e and she threw a passionate look back at Charles.
 J. Fowles, The French Lieutenant’s Woman

The whole instantiation of fire is a metaphorical network, consisting of the base

metaphor5 and five metaphorical sub-images,6 all relating to it by associations
of contiguity; this means that they are linked metonymically.7 The presence of
metonymy is an intrinsic element in every metaphorical extension. This instantia-
tion is a linguistic manifestation of the conceptual metaphor love is fire. I would
argue against the use of decontextualised phrases and sentences to illustrate either
conceptual metaphors or their linguistic manifestations. I believe in a discourse-
based approach to stylistic phenomena.
It is not only lexical metaphors but also phraseological metaphors that may be
extended across sentence boundaries in any type of discourse (e.g., prose, poetry,
drama, print media texts, advertisements). The stability and cohesion of the base
form of the PU provides ample opportunities for extension of figurative thought
even in a short stretch of text, for instance:
to burn daylight
Mercutio: Come, we burn daylight, ho!
Romeo: Nay, that’s not so.
Mercutio: I mean, sir, in delay
We w a s t e o u r l i g h t s in vain, l i k e l a m p s b y d a y .
 W. Shakespeare, Romeo and Juliet, Act 1, Sc. 4

Out of all the structural types of PUs, it is proverbs8 that lend themselves most
frequently to stylistic use. Mieder rightly concludes that “it is exactly the metaphor

.  By a base metaphor I understand a lexical or a phraseological metaphor that serves as a

basis for metaphorical extension.
.  A sub-image is an extension of the metaphorical image in a direct way or through other
sub-images. The sub-images become part of the associative figurative network created and
sustained on the basis of the metaphorical image.
.  Gibbs urges metaphor researchers to study “the interaction of metaphor and metonymy
in expressions for linguistic actions” (Gibbs 1999: 36–37). It is essential to understand the
interrelationship between metaphor and metonymy not only at the level of the system of lan-
guage but also in the web of discourse.
.  I believe that proverbs are part of phraseology as language units, e.g., the road to hell
is paved with good intentions. Proverbs are stable, cohesive combinations of words, with a
fully or partially figurative meaning in their base form, the same as other PUs. Proverbs
form one structural type of PUs. In the cognitive stylistic view, the same cognitive processes
 Chapter 12.  Extended metaphor in the web of discourse 

of the proverb which enables us to employ proverbs in so many different contexts”

(Mieder 1989: 21). E.g.:
the road to hell is paved with good intentions
Hell isn’t merely paved with good intentions; i t i s w a l l e d a n d r o o f e d
w i t h t h e m . Ye s , a n d f u r n i s h e d t o o .
 A. Huxley, Time Must Have a Stop

Shakespeare’s plays offer numerous sophisticated illustrations of extended figurative

thought that cover longer stretches of text, including monologues, e.g., the Ghost
who is the spirit of Hamlet’s late father delivers the following message to Hamlet:
to make one’s hair stand on end
I could a tale unfold whose lightest word
Would harrow up thy soul, freeze thy young blood,
Make thy two eyes, like stars, start from their spheres,
Thy knotted and combined locks to part,
A n d e a c h p a r t i c u l a r hair to stand on end,
Like quills upon the fretful porpentine.
 W. Shakespeare, Hamlet, Act 1, Sc. 5

The extended image conjures up the horror of the whole scene and brings Hamlet
to the revelation that his father had been most foully murdered. The development
of this horrifying thought forms an extended metaphor with metonymic links in
between the sub-images, creating a cohesive network of associations.
As thought develops, metaphor develops too. Extended metaphor reflects
extended figurative thought. The pattern of extended metaphor is a way to instan-
tiate a new meaning in discourse. Strings of sub-images bring out the cohesive
strength of the pattern: metaphorical ties and metonymic associations of conti-
guity with the base image provide semantic and stylistic cohesion in discourse.
Creation of an extended metaphor is a cognitive skill, part of the human ability
for continued abstraction. Interpretation of the figurative network calls for a cog-
nitive approach to use of both metaphor and metonymy. My observations fall in
with recent advances in cognitive linguistics and “the novel claim that metonymy
is a form of thought” (Gibbs 2007: 31). In extended metaphor, metonymy does
not function on its own but in close interaction with metaphor and other stylistic
techniques. Thus, extended metaphor is an instantial pattern involving a string
of metaphorical sub-images sustained and linked together by the base metaphor,
creating a cohesive network of associative metaphorical and metonymic ties.

determine stylistic changes of proverbs in actual use and the same stylistic patterns are
employed in discourse.
 Anita Naciscione

12.3  Types of extended metaphor

Several types of extended metaphor exist; these are subsets of the instantia-
tion of extended figurative thought. The simplest type of extended metaphor
consists in the use of one sub-image, metonymically going back to the base
metaphor. It is widespread in all types of genre and all periods of the develop-
ment of English. The following is a commercial for Nissan cars. The metaphor
is extended, resorting to use of Eastern mysticism. Interestingly, the commer-
cial is an instantiation of the conceptual metaphor life is a journey that also
appears in the text:
a journey
Remember, young man, l i f e i s a journey. E n j o y t h e r i d e .
 Nissan TV commercial, 1996

base metaphor

sub–image

Figure 12.1.  Extended metaphor Type 1

The most common type of extended metaphor presents successive use of meta-
phorical sub-images. It is a linear development of figurative thought. A sequence
of sub-images emerges, strung out in a line, securing a sustained mental picture of
the image of the base metaphor. Metaphorical development proceeds like a chain
reaction aroused by the thought of the previous sub-image, with each instantial
metaphorical item bringing about the subsequent sub-image:
a new broom/a new broom sweeps clean
A new broom d o e s n ’ t a l w a y s s w e e p c l e a n , it just b r u s h e s s o m e
o f t h e w o r s t d i r t u n d e r t h e c a r p e t for a while.
 Collins Cobuild Dictionary of Idioms, p.51

base metaphor

sub–image 1

sub–image 2

sub–image 3 etc.

Figure 12.2.  Extended metaphor Type 2

 Chapter 12.  Extended metaphor in the web of discourse 

This figure reflects figurative extension, constituting a process and resulting

in a string, a whole semantic network of novel associative sub-images, linked
­metonymically, that are sustained on the basis of the image of the PU, covering an
entire area of experience.
Another type of extended metaphor is the extension of two or several notional
base constituents. The development of several constituents provides parallel lines
of metaphorical thinking, creating a ramified semantic structure, which at the
same time remains part of the image of the PU and its instantiation in the given
context. In the following example from Chaucer, both asse and harpe are extended,
securing a cohesive network of associative ties over five lines of verse:
lyk an asse to the harpe 9
Or artow lyk an asse to the harpe,
That hereth soun, whan men the strenges plye,
But in his minde of that no melodye
May sinken, him to glade, for that he
So dul is of his bestialitee?
 1380 G. Chaucer, Troilus and Criseyde

base metaphor

sub–image 1 sub–image 2

sub–image 3

sub–image 5
sub–image 4

Figure 12.3.  Extended metaphor Type 3

In the course of time, stylistic patterns acquire new features; this is a logical dia-
chronic development of stylistic patterns. Usually the metaphorical extension fol-
lows the base metaphor. In EMoE10 a new type appears where a sub-image or

.  The PU lyk an asse to the harpe is obsolete MoE.

.  EMoE – Early Modern English is a period from the late 15th century to the transition to
Modern English in the late 17th century, also called Early New English (ENE).
 Anita Naciscione

several sub-images may also precede the base metaphor as we see it in Shake-
speare’s plays. For instance:
to hoist with one’s own petard
Let it work;
For ‘tis the sport t o h a v e t h e e n g i n e r
Hoist with his own petard: and ’t shall go hard
But I w i l l d e l v e o n e y a r d b e l o w t h e i r m i n e s ,
And blow them at the moon.
 W. Shakespeare, Hamlet, Act III, Sc. 4

sub–image 1

base metaphor

sub–image 2

sub–image 3
etc.

Figure 12.4.  Extended metaphor Type 4

Sustainable development of figurative thought is secured by metonymically extend-

ing the base image, e.g. petard → enginer… delve one yard below the mines…blow
them at the moon. However, the whole instantiation is perceived holistically: the
result of image extension is the metaphorisation of the whole context. It is a repre-
sentation of a metaphorical flow of thought in turning the plot against the plotter.
Use of a military image11 reveals the true nature of Hamlet’s enemies. He feels
their hostility and is encircled by them, but he is not going to give in without a
struggle. The metaphorical image is extended to reveal Hamlet’s state of mind and
his feelings.
We know that Shakespeare’s plays are built on figurative language. Extended
metaphor stands out as a very frequent phenomenon in his plays, reaching

.  In Shakespeare’s day it was the practice of engineers (constructors of military engines)
in a military siege operation to dig a mine or an underground gallery under the defences and
blow them up with a charge. The defenders would try to dig a counter-mine under it and set
off a charge that would blow up the men and explosives in the original mine.
 Chapter 12.  Extended metaphor in the web of discourse 

s­ ophisticated heights. For a typical instantiation of this type of extended metaphor,

see the famous monologue “All the world’s a stage” in Shakespeare’s play As You
Like It, Act II, Sc. 7. The whole extension of figurative thought covers 27 lines, cre-
ating a closely-knit network of metaphorical sub-images that not only follow the
base metaphor stage → players … exits … entrances … play many parts … acts …
plays his part … scene, but also precede the base metaphor: wide and universal
theatre … pageants … scene … to play in. The whole monologue is an instantiation
of the conceptual metaphor the world is theatre.

12.4  A diachronic approach to the stylistic use of extended metaphor

Patterns are part of the mental lexicon, stored in the long-term memory of the
language user. They are characterised by stability across centuries. New inimitable
instantial forms are constantly generated in accordance with existing language
patterns.
As a way of thinking, extended metaphor has existed since Old English
throughout the course of the history of the English language, judging by the texts
that have come down to us (Naciscione 1982, 2010). The pattern is characterised
by diachronic stability across centuries.
se æppel on his eagan12
Geheald me … and beorh me, swa swa man byrhð þam æplum on his eagum
mid his bræwum.
c900 Paris Psalter 31 (16.8)

Extended metaphor is more frequently used and more extended in MiE than in
OE as far as we can judge from recorded texts. Perhaps this could be explained
by a difference in the prevalence of stylistic techniques in the two periods. OE
verse mostly relies on alliteration and inner rhyme. In OE stylistic use, the
central role is played by kennings: a special type of compounding based on
metaphorical periphrasis, e.g., OE texts feature numerous kennings that are
poetic synonyms for the sea: hron-rād (whale-road), hwæl-weġ (whale’s way),
seġl-rād (sail road), swan-rād (swan-road). In MiE the stylistic use of metaphor
becomes prominent:

.  OE se æppel on his eagan – MoE the apple of one’s eye. The first references to this PU are
known from the Bible (five cases in core use). The PU first appeared in OE in King Alfred’s
translation Gregory’s Pastoral Care; it is also used in his work Boethius.
 Anita Naciscione

to catche a thorn13
And how-so she hath hard ben her-biforn,
To god hope I she hath now caught a thorn,
S h e s h a l n o t p u l l e i t o u t this nexte wyke;
God sende m o s w i c h t h o r n e s o n t o p y k e !
 c1385 G. Chaucer, Troilus and Criseyde, II, 1271–1274
an adder/serpent in the bosom14
O servant traitour, false hoomly hewe,
L y k t o the naddre in bosom s l y u n t r e w e .
 c1395 G. Chaucer, The Canterbury Tales, E, 1785–1786

Earlier periods of development of English testify to the historical evolution of con-

ceptual mapping (Trim 2007, 2011).
However, it is not only conceptual metaphors that are linked to pre-existing
ones and that are traceable back in time: the same applies to stylistic patterns too,
extended metaphor included. As cohesive figurative combinations of words, PUs
are also characterised by cross-century stability, e.g., the PU an adder/serpent in
the bosom is used by W. Caxton a century later according to recorded texts:
a serpent in one’s bosom
But they, lyke as the serpent t h a t p r y c k e t h o r s t y n g e t h h y m t h a t
k e p e t h h y m w a r m e in his bosomme.
 1481 W. Caxton Godeffroy, 66.34–35

Empirical observations allow me to draw generalisations about patterns across cen-

turies and various types of discourse, seeing sameness in difference. The stylistic
pattern is a reproducible and dynamic element of the system of language. As a dia-
chronically recurring element, pattern is inherently stable (Naciscione 1976, 1982).
Diachronic evidence has it that PUs also evolve through time, e.g., the PU a
snake/serpent in/under the grass/flowers has been recorded in English texts since
the end of the MiE period. Chaucer’s works contain two instances of its use, both
striking extended metaphors (see Naciscione 2010: 154; 241), which means that
the PU was already in use before Chaucer, borrowed into MiE as a loan translation
of the Latin proverb latet anguis in herba (Apperson 1969: 583). The PU has been

.  The PU to catche a thorn is obsolete in MoE.

.  The variant a serpent in one’s bosom is commonly used in MiE and EMoE. In MoE, the
most common form is a snake in one’s bosom; the full form is to nourish/cherish a snake/a
viper in one’s bosom. The PU was borrowed into MiE from the Latin proverb: in sinu viperam
habere. It goes back to Aesop’s fable The Farmer and the Viper. Thus, several variants have
existed in the history of English; however, the metaphorical meaning has been sustainable.
 Chapter 12.  Extended metaphor in the web of discourse 

extended in countless instances since Chaucer, creating instantial networks and

permeating the stretch of text with metaphorical meaning:
His felle malys he gan to close and hide,
L y c h e a snake t h a t i s w o n t t o g l y d e
W i t h h i s v e n y m under f r e s c h e floures.
1420 Troy I 18.209–211

Extended metaphor remains the most common stylistic technique used in EMoE
and subsequent periods, as we can see from the following instantiation of another
variant of the same PU: a serpent under grass:15
But the serpent l u r k e d vnder the grasse, and v n d e r s u g e r e d s p e a c h e
was hide pestiferous poyson.
 1548 E. Hall, Chronicle

In the course of development, the base form of the PU has had a number of lexi-
cal variants; it evolved into a snake in the grass in MoE. Recorded examples since
Chaucer’s day reveal cross-century stability of the PU in use, not fixedness, fro-
zenness, or non-compositionality. The PU is intrinsically stable in the face of dia-
chronic changes – the existence of lexical variants in its base form. Semantic and
stylistic stability is due to the metaphorical image that reflects human experience;
it is a figurative thought preserved through the centuries. As the cited examples
show, the PU a snake in the grass is a reproducible language unit that possesses
both stability and a great deal of flexibility in use depending on the context and
thought expressed, which determine the given stylistic instantiation.
MoE can boast innumerable cases of novel applications of the pattern of
extended metaphor:
a silver lining16
She always discovered silver linings t o t h e b l a c k e s t o f c l o u d s , but
now, scrutinize them as she might, she c o u l d d e t e c t i n t h e m n o n e
b u t t h e m o s t s o m b r e h u e s . Her imagination had worked out a
dazzling future for this portrait.
 E. F. Benson, Trouble for Lucia

.  Shakespeare used the PU a serpent under flowers twice in his plays (see Naciscione
2010: 109–110); these cases are sophisticated instances of allusion, which is a powerful stylistic
technique, an implicit verbal reference to the image of the PU, evoking associations with the
whole unit with the help of separate constituents.
.  A silver lining – a metaphorical PU derived from the proverb every cloud has a silver
lining, see Kunin (1967: 560); Collins Cobuild Dictionary of Idioms (1995: 352). The same
meaning has been preserved: a ray of hope, a metaphor for hope and optimism.
 Anita Naciscione

Development of stylistic patterns over the course of time presents a vast field of
research. Patterns can only be discerned by detailed and systematic analogies,
which help to ascertain similarities and establish their basic range. However,
even the few examples cited here provide a diachronic dimension: an ongoing
diachronic process of pattern evolution as a framework for metaphor develop-
ment and networking. My aim is to show that these patterns are not an inven-
tion of recent centuries, but a logical diachronic development of thought and
language.
Thus, the stylistic pattern of extended metaphor is a mental pattern that is
characterised by cross-century stability. It is viewed as a structure of thought that
is reproduced in new creative instantiations. A fundamental tenet proposed here is
sustainability of the stylistic pattern of extended metaphor throughout the history
of the English language.

12.5  Sustainability of figurative thought: A discourse dimension

The stylistic pattern is inherently dynamic in discourse. It provides for the devel-
opment and sustainability of metaphorical thought and language in discourse; it
goes across sentence boundaries, contributing to the semantic and stylistic cohe-
sion of text:
like looking for a needle in a haystack
Then of course we’ve got to find the antibody that we want, and that’s the
problem with the new technology. It’s really like t h e needle in t h e haystack.
We’ve got t o g o t h r o u g h t h e haystack, s t r a w b y s t r a w ,
p u l l i n g   o u t t h e needle.
 Collins Cobuild Dictionary of Idioms 1995: 272

It is the cohesive sustainable development provided by gradual extension of con-

tiguous links between sub-images that lends a discourse dimension to extended
metaphor. The extension of the figurative mode of thought is clearly seen in turn-
taking in dialogical discourse. The following stretch of dialogue is built on one
proverb; related elements form chains, in which the metaphorical network is inter-
twined with associations of contiguity, bringing out the cohesive force of figurative
thought:
rats desert a sinking ship
“You seem to have drawn yourselves into the limelight, which is a thing I prefer
to shun.”
“Sounds like a case of rats and a sinking ship,” put in Voles, upon whom Menton
immediately turned.
 Chapter 12.  Extended metaphor in the web of discourse 

“Don’t be a fool,” he snapped. “There is no question of a sinking ship if we s t e e r

a s e n s i b l e c o u r s e .”
“And there’s no question of me following a r a t ’ s tendencies, either,” said
Burnham.
 W. le Queux, The Mystery of Mademoiselle

It may be surprising to discover the strength of extended metaphorical image in

texts of great English poets, such as Chaucer, Shakespeare, Milton, Byron and
many others. Let me turn to Chaucer’s use of the tappe of lyf17 in The Reves18 Pro-
logue to his tale in The Canterbury Tales:

the tappe19 of lyf

And yet ik have alwey a coltes tooth,
As many a yeer as it is passed henne
S i n t h a t my tappe of lyf b i g a n t o r e n n e .
For sikerly, whan I was bore, anon
D e e t h d r o g h the tappe of lyf a n d l e e t i t g o n ;
A n d e v e r s i t h h a t h s o the tappe y - r o n n e ,
Til that almost al empty is the tonne.
The streem of lyf now droppeth on the chimbe.
 G. Chaucer, The Canterbury Tales, A, 3888–3895

Chaucer’s greatness has been noted by many scholars. He has been called the
father of English poetry, the father of English iambus, the maker of the national
language; his narrative techniques have been highly praised. All this is very true.
Even so, Chaucer’s greatness in the creative use of language certainly remains
underrated.
In the above quotation, the metaphor the tappe of lyf has been extended
across six lines of poetic text. This extended metaphor is one instance of the
greatness of Chaucer’s creative mind in language use. These lines delineate the
predestined course of a person’s life: death draws the tap of life when a person is
born and lets it run till almost empty is the tun, that is, until the barrel is almost
empty. Interestingly, in this instantiation Chaucer uses a synonymous metaphor

.  The tappe of lyf is obsolete MoE.

.  MiE reve – MoE reeve, bailiff.
.  Some of the more difficult MiE words in this instantiation: MiE tappe – MoE tap; MiE
lyf – MoE life; MiE drogh – MoE draw; MiE sith – MoE since; MiE tonne – MoE tun, barrel;
MiE chimbe – MoE rim of the barrel.
 Anita Naciscione

the streem of lyf that is frequently used later in English literature along with the
sands of life.20
In discourse, extended metaphor frequently turns into a narrative technique.
It sustains a narrative and creates a metaphorical continuum, a network of associa-
tive strings, constituting part of the same metaphorical application. Sustainability
is secured by recourse to the base metaphor over successive phrases or sentences,
or over longer stretches of text. For instance, the PU an odd/strange fish has been
extended over a large stretch of 25 pages by D.H. Lawrence in his novel The Lost
Girl to reveal the thoughts, emotions and experience of the lost girl who cannot
bring herself to marry a man whom she does not love.21
However, extended metaphor may be instantiated not only by gradual exten-
sion of contiguous links between sub-images, as is so in all the above cases. Exten-
sion of a figurative image over considerable stretches of text may also be achieved
by means of other stylistic patterns, e.g., pun, allusion and others.
Let me have a closer look at the use of the proverb all that glisters is not gold,22
a metaphorical image which has been extended over a whole scene by William
Shakespeare in his play The Merchant of Venice (Act II, Sc. 7). From the first lines of
the scene we find out that a rich heiress, Portia, is facing the difficult task of choos-
ing from among her suitors the right man whom she would marry. They have a
choice of three caskets: one of gold, one of silver and one of lead. The right casket
contains her picture. The first to choose is the Prince of Morocco. He chooses the
golden casket with the following saying, engraved in gold: “Who chooseth me
shall gain what many men desire”. The whole scene is centred around the image of
gold: the word “gold” or its derivative “golden” appears nine times within a short
scene of 80 lines, used in both direct and metaphorical meanings, e.g., “A golden
mind stoops not to shows of dross”. At the end of the scene the Prince opens the
golden casket to find no picture of Portia inside but a written scroll instead:
all that glisters is not gold
All that glisters is not gold;
Often have you heard that told:
Many a man his life hath sold

.  See the use of this metaphor in R. Burns’ famous poem A Red, Red Rose:

I will luve thee still, my dear,

While the sands o’ life s h a l l r u n .
.  For a full analysis of this instantiation, see Naciscione 2010: 230–233.
.  The form all that glisters is not gold was used in EMoE, it is obsolete in MoE. The MoE
form is all that glitters is not gold.
 Chapter 12.  Extended metaphor in the web of discourse 

But my outside to behold:

Gilded tombs do worms infold.
Had you been as wise as bold,
Young in limbs, in judgment old,
Your answer had not been inscroll’d:
Fare you well; your suit is cold.
 W. Shakespeare, The Merchant of Venice, Act II, Sc. 7

This monologue starts with the proverb all that glisters is not gold. The image of
gold is extended further into gilded tombs do worms infold which is literal and
metaphorical at the same time in this instantiation. This case study illustrates the
role of metaphor in the web of discourse, acquiring a discourse dimension. This
is not a “pure” metaphor; it is “mixed”, instantiated in a pun that encompasses a
whole scene. This is another type of figurative thought extension: extended pun.
The metaphorical thought all that glisters is not gold covers a long stretch of text:
a whole scene, providing semantic and stylistic cohesion and sustaining figurative
thought. Only identification of the whole figurative network brings out the flow of
figurative thought in a longer stretch of text.
Print media texts present an enormous variety of use of extended metaphor:
an area of research in its own right. I would just like to offer a brief insight into
the enormous potential of this stylistic technique in figurative conceptualisa-
tions. Metaphorical PUs are often employed in newspaper headlines, acquir-
ing significance for the whole discourse.23 The metaphor used in the headline
conveys an overtone and takes on a prominence in relation to the entire text.
For instance, the Financial Times frequently publishes serious articles in which
headlines contain a metaphor that is extended over the whole text, creating a
figurative network of semantic and stylistic relationships. In headlines, many
authors choose PUs as the best way to represent the quintessence of the article,
e.g., “What do we do now the c l i m a t e wolf is at the door?”24 (Financial
Times, 12 July, 2006, p. 2).
The first paragraph, which usually serves to set out the gist of the text in
greater detail, is a tight network of metaphorical ties. We find out that the little
boy was gobbled up as punishment for his earlier pranks. We read a reference to

.  I call this technique umbrella use: it refers to and covers the whole of the text. More on
umbrella use see: Naciscione 2010: 163–170.
.  The wolf is at the door means “the threat of poverty is upon us” (Spears [2002] 2006). This
proverb has developed from the PU to keep the wolf from the door (“to avert poverty or starva-
tion”) that has existed in the English language across centuries: MiE to kepe the wolf frome the
gate/dore; EMoE to kepe the wolfe from the dur.
 Anita Naciscione

Malthusians who “have been crying wolf25 for a couple of centuries, but in global
warming they may well have seen a real one”. The first paragraph ends with a rhe-
torical question, “Is global warming a wolf at our door?”26 Thus, the stretch of
text is permeated with figurative meaning created by use of three PUs united by a
common extended metaphorical image.
The end of the article is a warning that “global warming is going, like the wolf,
to gobble us up”. Reiteration of the PU or its sub-images both in the initial and final
paragraphs of newspaper articles performs a sustainable cohesive text-embracing
function. It is a frame construction, a common technique in print media texts.

12.6  Extended metaphor in multimodal discourse: A case study

It is revealing to explore the representation of figurative thought not only in verbal,

but also in visual discourse, and the intricate relationship between the verbal and
the visual, often involving other modes of expression, which is generally known as
multimodality.27 A fruitful approach has been suggested by Forceville, who holds
that in multimodal metaphor target, source, and/or mappable features are repre-
sented or suggested by at least two different sign systems or modes of perception
(2008: 463). I have tried to sum up the key traits of multimodal discourse from
the cognitive perspective: multimodal discourse applies stylistic techniques from
more than one semiotic mode of expression; the verbal works together with the
non-verbal in construction of new meaning in figurative conceptualisations which
reveal patterns of thought that are manifest in verbal and visual representations
(Naciscione 2010: 253).
Multimodal discourse calls for new ways of both creation and interpretation.
“Creative multimodality reveals how language functions” (Goodman 2006: 244).
Importantly, I would argue that multimodal discourse also reveals how thought
functions: it features the development and sustainability of figurative thought
in text.
Let me turn to an illustration of sustained figurative networks in print media.
The strange death of modern advertising is the title of a lengthy analytical ­article

.  The PU to cry wolf means “to give false alarm”. It goes back to Aesop’s fable The Boy Who
Cried Wolf.
.  The meaning of the PU a wolf at the door is a debt collector or a creditor. It is a later
development of the wolf is at the door (see The American Heritage Dictionary of the English
Language [2000] 2006).
.  For more on extended metaphor in multimodal discourse, see Naciscione 2010: 184–200.
 Chapter 12.  Extended metaphor in the web of discourse 

from the Financial Times, published at a time when advertising underwent a

severe crisis in 2006, threatened by the growth of digital advertising. It is written
as an ironic obituary to modern advertising with a striking visual representation
in the centre of the article (Figure 12.5).

Figure 12.5.  Financial Times, 22 June, 2006, p. 13

The text of the article is a tightly-knit web of figurative strands. The leading sty-
listic pattern is extended metaphor that covers the whole text and incorporates a
number of other stylistic patterns, all proceeding from the key metaphor death28
in the headline. What we first see visually is a newly-dug grave and a headstone
with a short epitaph:
ADVERTISING
R.I.P.
The picture and the inscription represent death, which is not only a metaphor but
also personification at the same time: funerals are organised for people, not for
abstract concepts. It is frequent that metaphor works together with personification
in figurative conceptualisations. R.I.P.29 is typically used to wish eternal rest and

.  Most dictionaries give death as the second meaning of the grave (literary).
.  The abbreviation R.I.P. stands for Rest in Peace (Latin: Requiescat in pace). It is a short
epitaph, commonly appearing on headstones.
 Anita Naciscione

peace to someone who has passed away. The grave is a visual metaphor for death.30
The design of the grave spotlights a man who is, metaphorically speaking, with
one foot in the grave.31 Hyperbolically, he is already half in the grave, desperately
trying to find the magic word which will be the Saviour – the right brand name
before his advertising company disappears altogether. This is the only way to avoid
bankruptcy which is as good as death for a company. The whole pictorial represen-
tation is a visual pun on the death of advertising.
The text of the article presents networks of metaphorical sub-images, chains
or strands of figurative thought, linked metonymically to the base metaphor by
associations of contiguity, all constituting an extended metaphor:

1. death → illness:
a symptom, a diagnosis, an affliction;
2. death → a funeral:
the deceased, to be buried, gravediggers;
3. death → a funeral service:
funeral rites, death knell, mourners, standing at the graveside, praying, the Bible
placed in the box, salvation, eternal life, the word is the Saviour.

Metonymy is invariably part of each instantiation of metaphorical extension. Thus,

extended metaphor is “mixed” by definition in every instance. Metaphor is not
alone in figurative meaning construction. In this text, extended metaphor incor-
porates other figurative modes: metonymy, personification, euphemism, allusion,
visual pun, hyperbole, irony. This is the reason why I would argue for the term
figurative networks,32 not merely metaphorical networks. It takes all sorts of tropes
to make a discourse. The great diversity of the reflection of figurative thought in
actual use is lost if the term metaphor is used as an umbrella term for all types of
figurative patterns, as was fairly common in the first decades of the development
of cognitive linguistics. Metaphor certainly remains the leading stylistic pattern;
however, we cannot deny the existence and role of other patterns of figurative use.
In this case study, the unpleasantness of an imminent funeral is mitigated by
euphemistic turns of phrase, e.g., death is referred to as the last stage of affliction
and the deceased stands for the dead body. The word is the Saviour is clearly an

.  Visual metaphor in advertising has been widely discussed. See Forceville 1991;
Mieder 1993; Fiedler 2012 and many others.
.  To have one foot in the grave means to be close to death or in a terrible condition. It is a
figurative hyperbolic phrase, first recorded in 1566 (see Ammer 1997).
.  For more on the functioning of figurative networks in discourse, see Naciscione 2010: 51,
77, 112, 116, 178, 207.
 Chapter 12.  Extended metaphor in the web of discourse 

allusion, an implicit reference to the Bible; it is the very beginning of all things.33
Thus the right word (viz., the right brand name) is the beginning of all things and
the maker of all things. The whole discourse is a hyperbole with a tone of subtle
euphemistic irony. All these stylistic patterns form figurative networks, represent-
ing a process and a result of human thought, and a conceptualisation of experi-
ence. The article is a verbal and visual instantiation of the conceptual metaphor
death is departure. Importantly, this conceptual mapping is not reflected in a
single sentence but in a longer stretch of multimodal discourse.
In stylistic use, visual representation of figurative thought performs an essen-
tial semantic and stylistic function: it enhances and interprets the image, bring-
ing the literal meaning to the fore; it enables us to make human thought visible,
creating a visual effect. The visual mode of perception provides food for thought
or, as Arnheim puts it, it forms visual thinking ([1969] 1997). Comprehension
and interpretation of discourse relies on the tie between the visual and the verbal.
The visual expression goes together with the verbal, creating cohesion of figurative
meaning and providing sustainability of figurative thought.
Thus, multimodal use is naturally “mixed”. It is a perfect “mix”. It is impossible
to find metaphor in a “pure” shape in multimodal use. It is only natural as we do
not sort out stylistic patterns in our mind; they interact and intermingle in the
process of meaning construction, reflecting the flow of figurative thought whose
development in discourse does not obey the rules of formal logic: to keep every-
thing separate, clear and precise.
The case study reveals that multimodal discourse is a verbal and visual repre-
sentation of figurative meaning with the aim of creating a coherent and cohesive
narrative; it is a “mini-narrative” (Forceville and Urios-Aparisi 2009: 11) in which
the image is evoked pictorially, and cohesion of figurative meaning is retained in the
web of discourse. Visual representation is a means of sustaining a figurative thought.

12.7  “Mixed” metaphors

It is believed that the issue of “mixed” metaphors has emerged over recent decades.
However, it has existed in traditional stylistics for a long time.34 In ­cognitive

.  It is an allusion to the New Testament. St John’s Gospel reads, “In the beginning was the
Word, and the Word was with God, and the Word was God”.
.  For instance, Nesfield writes about “confusion of metaphors”. Metaphors borrowed from
more than one source must not be combined in the same phrase or in the same sentence.
(1898: 416).
 Anita Naciscione

l­inguistics it came to the fore with Lakoff and Johnson’s division of metaphors
into “permissible mixed metaphors” and “impermissible mixed metaphors”
(Lakoff and Johnson [1980] 2003: 95). The immediate question that arises is,
“What is ­permissible and what is impermissible in natural language use?” Lakoff
and ­Johnson’s theoretical framework is as follows. Two metaphors that are used
in close proximity (in one sentence) are permissible if they come from the same
conceptual domain. However, if the two metaphors belong to different conceptual
domains, they are impermissible, e.g., “The content of the argument proceeds as
follows” (ibid.).
This approach has been explained in great detail in Müller’s book Metaphors
Dead and Alive, Sleeping and Waking: A Dynamic View (2008). Müller believes
that mixing of metaphors happens because speakers or writers combine meta-
phors from different and often contradictory domains. She also speaks about
mixing incompatible metaphors (2008: 136–137) and advises to avoid mixing or
contaminating metaphors (2008: 158). Müller quotes the unfortunate example:
“The butter mountain has been in the pipeline for some time” I call it unfortunate
because it is difficult to judge about its actual use as in both the books it is given
out of discourse even though it has grammatical context: it is used in a full simple
extended sentence. This is not sufficient for cognitive stylistic analysis. It is no use
considering that “a butter mountain just does not fit into a pipeline of a normal
size” as Müller suggests (2008: 134). It is perfectly clear that the meaning is figura-
tive. We understand it online35 if we read it in an appropriate discourse context.
The notorious EU butter mountain appeared in 1987, following the crisis in the
dairy industry as a result of the Common Agricultural Policy. A butter mountain
is a metaphorical term that in EU language usually denotes the problem that the
phenomenon causes. As Müller herself deciphers it on p. 143, what is meant is the
problem of Europe’s butter mountain. If that is the case, the metaphorical term
the butter mountain is a metonymy in actual use, as it stands for the problem of
overproduction. New metaphorical terms keep springing up in EU vocabulary
and are regularly used in debates in EU institutions, as my own experience as a
freelance interpreter has it; and an endless line of metaphorical terms is always in
the pipeline.
In the quoted example, in the pipeline does not mean a type of transporta-
tion but is used in its figurative meaning, which is part of its semantic structure
recorded in most dictionaries. Large institutions often use this metaphorical term
for plans or projects that are in process at the discussion phase.

.  For more on online processing of figurative thought, see Gibbs [1994] 1999: 286.
 Chapter 12.  Extended metaphor in the web of discourse 

It is important to accept the interplay of two or more metaphors “in authentic

use”36 as a natural discourse phenomenon. Let me examine the use by famous
writers of two metaphorical PUs in one context:
to smell a rat
without rhyme or reason
The P.P.R.S. was so imposing a concern, and he had been connected with it so
short a time, that it seemed presumptuous to smell a rat; especially as he would
have to leave the Board and the thousand a year he earned on it if he ra i s e d
s m e l l o f r a t without r a t or reason. B u t w h a t i f t h e r e w e r e a
r a t ? That was the trouble!
 J. Galsworthy, The White Monkey

Is this use impermissible? Do these metaphors contaminate the text? Shall we take
our red pen and mark it as an error or a logical fallacy? The question why two
metaphors cannot be used in one context if they come from different conceptual
domains has not been answered by the mixed metaphor theorists yet.
In the above example, two PUs are intertwined in one context. The instan-
tiation creates a network of figurative ties; it becomes the centre of interest: it is
outside the reader’s experience, as it has not been encountered before. It is a focal
point, attracting attention and increasing emotional suspense. The reiteration dis-
closes emotional unrest, apprehension, and anxiety. A rat appears again as part
of an indirect interior monologue of the character, bringing out inner hesitation
and wavering before making a decision. The interfusive effect is enhanced by the
instantial replacement of rhyme by rat and set off by the onomatopoeic effect of
alliteration: [r] is common to the base forms of both PUs.
In discourse, it is fairly common to have two metaphorical PUs running close
to each other and complementing the metaphorical image. Actual use shows that
two PUs may also come from two different conceptual domains to emphasise the
thought:
don’t change horses in mid-stream
to look before you leap
He glanced at Blanche. “Changing horses, love? I should look before you leap.”
 D. May, Revenger’s Comedy

The phenomenon of use of two or more metaphors, coming from different con-
ceptual domains and following one another in close proximity, is nothing new in

.  Charteris-Black 2012: 5.

 Anita Naciscione

the history of English literature. Chaucer’s poetical works contain a number of

instantiations of this type, e.g.:
to hold by the brydel37
at the staves ende38
as an arowe
His newe lady holdeth him t o n a r o w e
U p by the brydel, at the staves ende,
That every word, he dradde hit as an arowe.
 c1379 G. Chaucer, Anelida and Arcite, VII, 183–185

There is a good reason for use of several metaphors running; they intensify the
conceptualised experience: Arcite’s new mistress is so cruel that she holds him
tightly by the bridle and keeps him at the staff ’s end. These two PUs are followed
by another metaphor: he dreads every word she says like an arrow.
The aim of having “unmixed” metaphors may be hard to attain. In figurative
language, the presence of another mode of expression is practically inevitable. The
simplest way to explain it is to turn to the base forms of PUs, e.g.,: many proverbs,
such as two heads are better than one, a bird in hand is worth two in the bush, there’s
more than meets the eye etc., are metaphorical, at the same time they include a
striking metonymy, based on the metonymic mapping part for whole; out of
sight, out of mind features metaphor, metonymy, parallelism, assonance and ana-
paest; when Adam delved and Eve span who was then the gentleman? is a metaphor-
ical PU; however, its semantic structure also contains a number of other stylistic
means: allusion, inner rhyme, rhetorical question. They all have a role to play in
shaping the figurative meaning of the PU. Importantly, in a creative instantiation
in discourse, the base form of the PU retains its stylistic potential and acquires a
novel shape, creating significant changes in form and meaning determined by the
development of thought in the process of meaning construction.
In phraseology, allusion is a complex stylistic pattern: an implicit mental ref-
erence to the image of the PU. In the case of allusion, the PU is represented by one
or more explicit image-bearing constituents in discourse, e.g.:
the devil can quote Scripture for his own ends
His use of a biblical phrase gave her a touch of shivers, of d i a b l e r i e 39 – the
devil a t h i s o l d g a m e o f quot i n g scripture.
 G. Greene, Loser Takes All

.  MiE to hold by the brydel – MoE to hold a bridle; synonymous PUs: to hold a double
bridle, to hold the reins, to hold the double reins.
.  MiE at the staves ende – at the staff ’s end; obsolete in MoE.
.  Diablerie – italicised by G. Greene.
 Chapter 12.  Extended metaphor in the web of discourse 

The explicit constituents the devil and quoting scripture remain metaphorical but
at the same time they perform a metonymic function, acting like a recall cue, and
evoking associations with and alluding to the metaphorical proverb.
Phraseological allusion may introduce elements in the contextual structure of
the PU that do not come from the same conceptual domain:
a snake/serpent in one’s bosom
What is most interesting about the author’s research is that it reveals in its
subtext  an underlying pattern at the core of all organized white resistance in
America to black advancement, a pattern that still exists today even as we prepare
to elect our first Black President. This pattern is t h e p o i s o n o u s snake
c o i l e d in the bosom o f A m e r i c a n d e m o c r a c y .
 H. L. Calhoun, Who won the Civil War?

As an abstract notion, democracy means “power of the people”, which metonymi-

cally stands for “government by the people” that is recorded in all dictionaries
as part of the semantic structure of the word. American democracy would be an
“alien” element in the framework of the base form of the PU: it is in no way part of
the image of a snake in the bosom. However, it is perfectly natural in instantiation
of figurative language. Stylistically, American democracy is personification in this
context (in the bosom of American democracy), which is part and parcel of the use
of the metaphorical PU.
Seeming incompatibility appears in the use of a number of stylistic patterns.40
One of them is zeugma, which often contains an element of the non-rational and
does not follow the rules of logic, e.g.:
This Troilus, with h e r t e a n d eres spradde,
H e r d e al this thing devysen to and fro.
 c1385 G. Chaucer, Troilus and Criseyde, IV, 1422–1423

The instantiation is based on a common expression with spread ears that usually
refers to elephants. In this poem Chaucer uses it metaphorically: Troilus herde
(listened to) all that is told to him with herte and eres spradde (with heart and ears
spread). By definition, zeugma joins two different senses that may have a contra-
dictory relation to each other, in this case herte and eres.
Thus, the development of figurative thought plays an important role in intro-
ducing new figurative elements that may be seemingly incompatible, but which

.  The very system of stylistic patterns includes devices which do not respect the logical
principles of non-contradiction and are intrinsically based on the absurdity of seem-
ingly irreconcilable constituent elements, thus displaying logical incompatibilities, e.g.,
oxymora, paradoxes, malapropisms, which are commonly used in English (Naciscione
2008: 199–201).
 Anita Naciscione

make sense in the given stylistic instantiation. “Mixing” or rather the use of several
figurative elements in close proximity is a natural process of figurative meaning
construction.

12.8  Conclusion

Use of extended metaphor is one of the resources to convey sustained human expe-
rience. It gives freedom and space for creativity. Extended metaphor is an instan-
tial stylistic pattern, involving a string of sub-images sustained and tied together
by the base metaphor, creating a cohesive network of associative metaphorical
and metonymic bonds. The metaphorical sub-images are linked metonymically
by associations of contiguity. Thus by definition metonymy is an integral part of
extended metaphor.
In discourse, extended metaphor incorporates other metaphors and figurative
modes: metonymy, allusion, pun, personification, euphemism, hyperbole, irony;
they all become part and parcel of the extension of the base metaphor, creating fig-
urative networks. In figurative use, formation of figurative networks and “mixing”
extended metaphor with other stylistic patterns is a natural discourse phenom-
enon. Metaphor works together with other patterns in figurative conceptualisa-
tions, representing people’s experiences and the external world.
Diachronic studies reveal that extended metaphor defines as an entrenched
figurative pattern and a way of thinking. Pattern is characterised by cross-­century
stability. New stylistic instantiations emerge as a manifestation of the human
imaginative process on the basis of existing stylistic patterns. In the cognitive sty-
listic view, extended metaphor forms a pattern of both thought and language that
is stored in the long-term memory of the language user. Instantiation of extended
metaphor is a cognitive process that reflects extended metaphorical thought. It is
an extension of the metaphorical image.
Stylistic pattern is an essential element in figurative meaning construction: it
provides a thinking framework. Use of stylistic patterns in the web of both verbal
and multimodal discourse is a fruitful pathway for further exploration in cognitive
stylistics.

References

Ammer, C. (1997). The American heritage dictionary of idioms. San Diego, CA: Houghton
Mifflin.
Apperson, G. L. (1969). Proverbs and proverbial phrases: A historical dictionary. Detroit: Gale
Research Co.
 Chapter 12.  Extended metaphor in the web of discourse 

Arnheim, R. [1969] (1997). Visual thinking. Berkeley, CA: University of California Press.
Charteris-Black, J. (2012). Forensic deliberations on ‘purposeful metaphor’. In L. Cameron  &
G.  Low (Eds.), Metaphor and the social world (Vol. 2(1), pp. 1–21). Amsterdam: John
Benjamins.
Collins Cobuild Dictionary of Idioms. (1995). London: Harper Collins Publishers.
Fiedler, S. (2012). Englische Redewendungen und Sprichwörter in der Praxis. Leipzig: Engelsdor-
fer Verlag.
Forceville, C. (1991). Verbo-pictorial metaphor in advertisement. Parlance, 3, 7–20.
Forceville, C. (2008). Metaphor in pictures and multimodal representations. In R. W. Gibbs
(Ed.), The Cambridge handbook of metaphor and thought (pp. 462–482). Cambridge:
­Cambridge University Press.  doi: 10.1017/CBO9780511816802.028
Forceville, C., & Urios-Aparasi, E. (2009). Introduction. In C. J. Forceville & E. Urios-Aparasi
(Eds.), Multimodal Metaphor (pp. 3–17). Berlin: Mouton de Gruyter.
doi: 10.1515/9783110215366
Gibbs, R. W. (1994). The poetics of mind: Figurative thought, language and understanding.
­Cambridge: Cambridge University Press.
Gibbs, R. W. Jr. (1999). Researching Metaphor. In L. Cameron & G. Low (Eds.), Researching and
applying metaphor (pp. 29–47). Cambridge: Cambridge University Press.
Gibbs, R. W. (2003). Prototypes in dynamic meaning construal. In J. Gavins & G. Steen (Eds.),
Cognitive poetics in practice (pp. 27–40). London: Routledge.
Gibbs, R. W. (2007). Experimental tests of figurative meaning construction. In G. Radden, K.-M.
Köpke, T. Berg, & P. Siemund (Eds.), Aspects of meaning construction (pp. 19–32). Amster-
dam: John Benjamins.  doi: 10.1075/z.136.04gib
Goodman, S. (2006). Word and image. In S. Goodman & K. O’Haloran (Eds.), The art of English:
Literary creativity (pp. 244–277). New York: Palgrave/Macmillan.
Kunin, A. V. (1967). Anglo-russkiy frazeologicheskiy slovar’ [English-Russian Dictionary of
Phraseology]. Moskva: Sovetskaya entsiklopediya.
Lakoff, G., & Johnson, M. (2003). Metaphors we live by. Chicago: University of Chicago
Press.  doi: 10.7208/chicago/9780226470993.001.0001
Mieder, W. (1989). American proverbs: A study of texts and contexts. Bern, Frankfurt/M., New
York & Paris: Peter Lang.
Mieder, W. (1993). Proverbs are never out of season: Popular wisdom in the modern age. Oxford:
Oxford University Press.
Müller, C. (2008). Metaphors dead and alive, sleeping and waking: A dynamic view. Chicago &
London: University of Chicago Press.  doi: 10.7208/chicago/9780226548265.001.0001
Naciscione, A. (1976). Okkazional’noye stilisticheskoye ispol’zovaniye frazeologitcheskikh edinits
(na materiale proizvedeniy Chosera) [Occasional Stylistic Changes of Phraseological Units:
Based on the Complete Works of Chaucer]. Ph.D. dissertation. Moscow: Moskovskiy peda-
gogicheskiy institut inostrannykh yazykov imeni M. Toreza.
Naciscione, A. (1982). Reproducibility of patterns of occasional changes of phraseological units
in speech, 〈http://www.naciscione.com/resources/reproducibility1982.pdf 〉 (19 November,
2012).
Naciscione, A. (2008). Uncommon Nonsense: The Reflection Of The Non-Rational In Stylistic Use,
〈http://www.naciscione.com/resources/uncommon_nonsense.pdf 〉 (19 November, 2012).
Naciscione, A. (2010). Stylistic use of phraseological units in discourse. Amsterdam: John
­Benjamins.  doi: 10.1075/z.159
 Anita Naciscione

Nesfield, J. C. (1898). Aids to the study and composition of English. London: Macmillan and Co.,
Limited.
Spears R. A. ( 2006). McGraw-Hill’s dictionary of American idioms and phrasal verbs. New York:
McGraw-Hill Companies.
The American heritage dictionary of the English language. (2006). Boston & New York: Houghton
Mifflin Company.
Trim, R. (2007). Metaphor networks: The comparative evolution of figurative language.
­Basingstoke: Palgrave Macmillan.
Trim, R. (2011). Metaphor and the historical evolution of conceptual mapping. Basingstoke:
­Palgrave Macmilla.  doi: 10.1057/9780230337053
 Index

A conditional  43–44 D
A is B metaphor structure  cognitive flexibility  viii, ix, x, deviations
128, 130 66–67, 191 from dominant
advertisement  226–228, Cognitive Metaphor Theories semantics  123
246, 257 (CogMT),  34–35 from phraseological
allusion  262 comprehension norm  141
American Psycho (Mary of mixed metaphor  11–13, from register  125
Harron)  229–230 64–68, 206–207 dictionary  61, 82, 96, 114,
anaphoric reference  22–23, 24 conceptual blending  41–42, 122, 124, 129, 140, 146,
anger  231–232 44, 47–48, 159, 160–161, 147, 263
animation  232–234 162, 165, 170–173, 234–235 discourse
Aristotle  33 and ATT-Meta  95 metaphorically
attention  20, 41–43, 44, 47 conceptual metaphor homogenous  7–9, 13–14
ATT-Meta  xi, 75, 78–109 theory  ix, 3, 11, 28–29, 35, multimodal  226, 256–259
advantages of  104–108 40, 58, 205, 212, 220, 223 natural  76–77, 117, 159,
and mixed metaphor  95–109 shortcomings of  37–39, 59, 242–243
pretense/fictional 69, 86 discourse dynamics  19–29,
spaces  83–89, 91–94, 96, conceptual packs  xii, 145–146, 150, 159, 165, 171–172, 174
97–100, 102–103 148–149 discourse strategies  xiv, 151
view-neutral mapping connotation  46–47, 92–93, 228
adjuncts (VNMAs)  context  135. See also contextual E
60–61, 68–69, 90–94, meaning embodied experience  140,
96–100, 107–108 cultural  10. See also culture 156–157, 168, 174
local and global  9–10 embodied simulation  157, 161,
B conventionalization  12, 36, 163–164
Barden, John  xi 119. See also metaphor(s), emotion  82, 94, 99, 142,
Bible  8–9, 225, 259 conventional 197, 231
Black, Max  34–35 corpus empathic response  161, 164
blending theory. See British National Corpus empathy  19, 25–28, 164
conceptual blending (BNC)  61–62, 82, 138, 142
“Block that Metaphor!”  collocates  139, 142, 208–209 F
x, 57–59, 61 concordance  166, 207 figurative language
Brideshead Revisited (Evelyn data/evidence  62, 189–190, in wine discourse  188–189
Waugh)  192–193 204, 207–208 processing speed of  84
interview  156, 158, figurative thought,
C 165–166, 174 sustainability of  252–259
Cameron, Lynne  ix Oxford English Corpus Forceville, Charles  xiii
chaining. See mixing, serial (OEC)  207–208 foregrouding  32, 41, 44, 47,
Charteris-Black, Charles  xii VOICE  141, 148 50–52
Chaucer  247, 250–251, 253, creativity  141–142, 151–152, 207,
262, 263 220, 228 G
cinematic expressive culture genre  219, 229
movement  50 cultural conceptualizations  Gibbs, Raymond  x
clauses  4 144 goal-directed reasoning 
and mixed metaphor  206, socio-cultural knowledge  69, 94–95
215–216 193–196, 237 group identity  191–195
 Mixing Metaphor: The State of the Art

H Metaphor Identification purposeful  156–158, 163, 175

Hommerberg, Charlotte  xii Procedure (MIP)  repeated  143–149, 158, 167,
humor  12, 64, 119, 205, 218, 221 61–62, 166 168–170
metaphor(s). See also stylistic use of  13–14,
I metaphoric language/ 218, 243–245, 252–256,
iconic description  183 expressions 259–264
idiomatic expression  37, alive/vital vs. dead  35–38, systematic/structural  19, 23,
46–47, 82, 115–116, 126, 217 143, 205 69, 165
idioms. See idiomatic and influence of L1  137–139 three-dimensional
expression and metonymy  77, 241, 244, apporoach  119
image schemas  140, 237 245. See also metonymy topic-triggered  218
imagery, discursive role asocial view of  29 uncounscious use of  xiv,
of  191–197 cinematic  50 38, 114, 157, 159
internet  vii, viii, 63, 114, 208 clustering of  4, 19–20, 25, metaphoreme  145, 149
148, 207 metaphoric language/
K conceptual  vii, 3, 38, 39, expressions
Kimmel, M.  3–4, 6–7, 118, 121, 58–59, 60, 66–67, 69, communicative purpose
206, 211, 212–213, 215 84–85, 114, 171, 205–206, of  13–14, 119–123, 139,
Kövecses, Zoltán  ix 212, 224–226, 244, 246, 259 150–151, 164–165, 167,
consciously created  xiv, 170–174, 191–197
L 218, 220–221 conceptual clashes in  4,
Language conventional  10, 36–38, 97, 107, 115, 116–117, 126, 131,
English as lingua franca  135 118–119, 123, 214. See also 149, 151, 161, 205–206,
ESL  134 conventionalization 211, 215
interlanguage  133, 136, 139 deliberate use of  xi, xiv, didactic purposes of  8–9
Latin  139–140 13–14, 25, 51, 114–117, literal meaning of  32, 41, 47,
modern English  249, 251 120–132 206–207, 211, 214, 217
old English  242, 249 deliberate vs. metaphorical interpretation
Romance languages  139–142 nondeliberate  117–120 of  66
speaker, non-native  xi, 133 dynamic view of  x, 28–29, noncoherence of  39–40
lexico-conceptual pact  31–32, 49–52 rhetorical purpose/function
22–23, 24 elaborated  162, 170–174 of  xi, xii, xiii, 119, 122,
Linguistic Metaphor Theories extended  120, 162, 165, 158, 167, 175, 218, 220
(LMT)  33–34, 36 170–174, 241–264 semantic dissonance/
local idiomaticity  150 generic  5 incompatibility in 
Lonergan, Julia  x homogenous. See discourse, 18–19, 226
metaphorically semantic opposition in  158
M homogenous variations in  135, 142
MacArthur, Fiona  xi hybrid  xi, 134, 136–139, metaphorical (figurative)
mapping  78–83, 84 140–152 network  244–245,
(non)parallel  59–60, 69 layered/embedded  255–256, 258
correspondence rule  85–89, 24–25, 104 sub-images  244–249, 258
91–94, 105–107 map-transcending  81–83, 91 metaphoricity  32, 33, 43,
source-to-target  4, 7, 38, mixed. See mixed metaphor 49–52
41, 212 multimodal  49, 223–224, awareness of  36–38, 151
meaning 236–237, 256–259 method
contextual  18–23, 60–62, multiple source error analysis  134–135
135–136 domain  236–237 talking-and-thinking  19
semantic inconsistency  41, multiple  19–28 write out after
43–44, 46, 115–116, 205, 226 novel  10, 60, 68, 122, interpreting  64
source-domain  84 141–142, 171 metonymy. See also
The Merchant of Venice primary  84 metaphor(s), and
(Shakespeare)  254–255 product vs. processing  39 metonymy  244, 245
 Index 
mixed metaphor
adjacent (textually
proximal)  118, 128, 206,
214, 215
felicitous vs. infelicitous  23,
75, 101–102, 107, 135–136,
143, 190
grammatical boundaries
of  215–216
in pictures, and multimodal
discourse  226–237
metalinguistic
comment  203, 207–221
negative comment
about  vii–viii, 18, 32, 58,
114, 135–136, 159, 204, 209
permissible vs.
impermissible  xiv, 38,
205–206, 211–212, 242,
260–261, 211, 260
mixing
combination of  103–104
parallel  76–77, 97, 100–103,
106–107, 108
serial  76–77, 97, 104
Müller, Cornelia  x
multimodality  164, 256
N cross-modal  185–186,
Naciscione, Anita  xiii
The New Yorker  x, 57–58, 61
O 189–190
Obama  83, 115
P Shakespeare  244, 245,
pain, chronic  156–157, 161–164,
166–175
parable  8
Paradis, Carita  xii
parallelism
grammatical  216
source/taget-wise  105, 135
pattern, stylistic  243–245,
249–252
perception, sensory  180–184
description of  184–187, 189
imagery  187–198
personification  96, 123–124,
127, 171
phraseological units
(PUs)  242, 244, 247,
250–251, 261
policy  12
Postwar: A history of Europe
since 1945
 (Tony
Judt)  115, 118, 120–122, 127
proverb  37, 224, 244–245,
250–251, 252–255, 258, 262
pun  218, 255
R source domain reasoning
reference book  43, 204
relevance theory  84
Richards, Ivor  34–35
S 120, 243
Schwarzenegger, Arnold  10
science fiction  228–229
Semino, Elena  xiii
sense(s)  180
190–191
smell (olfactory
sensations)  184, 187,
vision (visual
inspection)  184, 189
248–249, 254
shallow processing  11–12
simile  77, 187–188, 192–193,
196–197
source domain
information  ix, 5, 68,
195–197
activation of  11–13, 45, 47,
212, 230, 232, 233–234
ancillary assumptions  x,
xi, 60, 64, 67, 83, 89–91,
98, 107
complexity of  13
divergent  9–10, 156–157, 205
incompatibility of  6, 19, 44
juxtaposition of  7, 11, 18,
21–22
metaphorised  24
overlap-cum-contrast
between  217–218
semantic convergence  157
similarity between  212–215
source-scenario
inferences  80, 83, 90, 91
suppressed  41
 
59–61, 64
empirical evidence of  61–68
Steen, Gerald  xi
stylistics  xiv, 18, 41, 44,
submetaphor  60, 64, 67
sub-person  79–80, 87–89, 91
superamodal
representation  190–191
syncretism, lexical  190–191
synesthesia  185, 189–190
V
visual cues  232
visual representation  259
W
Weinrich, Harald  34–35
wine reviews  180
terminologies  185–187
textual structure of  182
Z
zeugma  119, 263
The ability to recognise, discuss and evaluate one’s educational
beliefs and working practices in metaphoric terms has for
several years been seen as a highly valuable tool for increasing
self-awareness, facilitating learning (or teaching), and/or
predicting behaviour. This is the first edited book solely devoted
to the topic of researching elicited metaphor in education,
and brings together key researchers from China, Poland,
Puerto Rico, South America, UK and USA. The 12 chapters
involve overviews and state-of-the art articles, articles focussing
on methodology and validation, as well as reflections on the
effectiveness of techniques and research reports of recent
empirical studies. The bulk of the articles relate to literacy
(L1 and L2) and teacher education, but science education is
also addressed. The book offers useful models for academics,
professionals and PhD students in these areas, and provides
solutions for improving the validity of elicited metaphor
techniques in educational research.

isbn 978 90 272 0207 9

John Benjamins Publishing Company