Mixing metaphor: a descriptive and prescriptive analysis 9789027202109, 9789027267504, 9027202109, 9027267502

Citation preview

Metaphor in Language, Cognition & Communication

Mixing Metaphor

6

John Benjamins Publishing Company

 

Raymond W. Gibbs, Jr.

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 communication. 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

Alan Cienki

Tony Berber Sardinha

Joep Cornelissen

Victoria University of Wellington Pontifical University of São Paulo

Lynne Cameron

The Open University

Daniel Casasanto

University of Chicago

VU University Amsterdam & Moscow State Linguistic University VU University Amsterdam

Alice Deignan

University of Leeds

Thomas Fuyin Li

Beihang University (BUAA)

Cornelia Müller

Europa Universität Viadrina, Frankfurt/Oder

Francisco José Ruiz de Mendoza Ibáñez

Raymond W. Gibbs, Jr.

University of La Rioja, Logroño

Joe Grady

Lancaster University

University of California, Santa Cruz Cultural Logic, Providence

Volume 6 Mixing Metaphor Edited by Raymond W. Gibbs, Jr.

Elena Semino

Mixing Metaphor A descriptive and prescriptive analysis

Edited by

Raymond W. Gibbs, Jr. University of California, Santa Cruz

John Benjamins Publishing Company Amsterdam / Philadelphia

8

TM

The paper used in this publication meets the minimum requirements of 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 Raymond W. Gibbs, Jr.

vii

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? Lynne Cameron

17

chapter 3 Why mixed metaphors make sense Cornelia Müller

31

chapter 4 Tackling mixed metaphors in discourse: New corpus and psychological evidence Julia E. Lonergan & Raymond W. Gibbs, Jr.

57

Part II.  Reasons for Mixing Metaphor chapter 5 Mixed metaphor: Its depth, its breadth, and a pretence-based approach John Barnden

75

chapter 6 Mixed metaphor is a question of deliberateness Gerard Steen

113

chapter 7 When languages and cultures meet: Mixed metaphors in the discourse of ­Spanish speakers of English Fiona MacArthur

133

 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 Jonathan Charteris-Black

155

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

179

chapter 10 A corpus-based study of ‘mixed metaphor’ as a metalinguistic comment Elena Semino

203

chapter 11 Mixing in pictorial and multimodal metaphors? Charles Forceville

223

chapter 12 Extended metaphor in the web of discourse Anita Naciscione Index

241 267

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 question. 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 history, 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. Nonetheless, 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 language that may reflect his varied metaphorical understandings of the city tax proposal 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 metaphors 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 elaborative reasoning about a single source domain to reveal some of its less known or hidden meanings. Mixing metaphor in discourse does not necessarily cause listeners 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 metaphorical 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 conceptual 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 conceptual 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 discourse 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 perspective: 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 messages 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. Readers 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 abilities to respond to moment-by-moment affordances within different communicative 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 overproduction of butter in the EU has led to huge amounts of butter awaiting distribution 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 distribution). Bringing forth background knowledge when speaking and gesturing activates 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 activation for source domains as discourse unfolds in interpersonal interactions. Julia Lonergan and Raymond Gibbs’s chapter is titled “Tackling mixed metaphors 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 different 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, ATTMeta, 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 adaptation 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 primarily 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 deliberateness.” Although many observers claim that mixed metaphors are the products 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, diplomatic 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 metaphorical messages precisely when people recognize their deliberate composition and use. Steen explores different cases where metaphors are deliberately and nondeliberately 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 nonnative 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 languages, but non-native speakers do this quite often, as when one person, talking 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 listeners 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 experiences 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 concludes 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 merging 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 important 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” meaning 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 understanding 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 metaphorical experiences in language, thought, and communication. In this manner, the study of mixing metaphor may offer significant insights into contrasting theories 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 comprehensive 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, conceptually 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 producing 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 mappings. 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 metaphorically 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 fighting 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 metaphors 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 expressions (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 metaphorical expression, red tape, which renders this meaning appropriately in a conventionalized 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 producing 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 (constituted 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 mentioned; for example, the issue of what kind of social unity is provided by the family 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 metaphorically 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 coherence 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 commonly 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 consistent 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 certain 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 connection 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 something 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 discussed 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 immigrant 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 metaphors 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 somehow 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 processing” (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 incompatibility 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 activation 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 mappings 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 propositions. 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 conventionalization 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 comprehend 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 activation level goes beyond the threshold? We know it because although we understand 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 metaphors, 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 discourses 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 throughout 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 complexity 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 dominant 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 homogeneous 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 conceptual 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 metaphors 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 metaphors. 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 interfere 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 produce 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 applying metaphor (pp. 29–47). Cambridge: Cambridge University Press. Kimmel, M. (2010). Why we mix metaphors (and mix them well): Discourse coherence, conceptual 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 perspective, 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 dissonance, or producing any paralinguistic feature, such as pausing, hesitation, or laughter, to suggest that they have even noticed the juxtaposition. Mixed metaphors 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 mixing 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’ metaphor 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 juxtaposed 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 making use of metaphors to explain ideas from my research, and how multiple metaphors 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 spoken, 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 processes 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 identifies metaphor vehicles as words and phrases that contrast with the on-going discourse 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 participants 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 conversations 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 metaphorically 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 conflict. 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 metaphor 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 metaphors’ 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 308 309 310

.. I think it’s so easy to – er, … perpetuate.. bitterness, 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 carries reminders of what has already been said. Metaphors are mixed in the chain bitterness – breaking that cycle – of bitterness, but any dissonance is only superficial; 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 sufficiently 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 contextual 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 conversations, 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: responsibility, 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 discourse topics together, with a layering effect in which healing seems to be discursively 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 conversation. 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 suggest, 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, anaphoric reference and lexico-conceptual pacts, coherence via metaphorical scenarios, and layering of conventionalised and systematic metaphors. Analysis of metaphor as a discourse dynamic phenomenon shows the ‘mixed metaphor problem’ 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 science 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 journeys 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, motivated 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 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656

there’s an , to that er, …(1.0) you know, er, …(2.0) figure of speech you know, bridges. … bridges can be built. … and that is if you, .. actively – er, .. create, er, .. distances. … barriers. … or what are they? they are exclusions …(1.0) and er, .. a thing I believe absolutely fundamentally, is that er, …(1.0) if you exclude anybody’s voice, …(1.0) you know, … you’re se- – 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 closeness. 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 provide 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 feelings, 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 adding 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 accurately 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 metaphors 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 avoiding 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 connecting 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. Having 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 construct 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 logical relations of semantic fields. The discourse dynamics theory of metaphor holds that large-scale systematic 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 principle’ (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 understanding 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 constructions. 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 conceptual 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 describing and explaining what people do with metaphor and thus what metaphor ‘is’. Furthermore, the scientific method of reduction(ism) that seems to underpin 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 generalising 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 illusion 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 & Symbol, 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? Cognitive 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 discourse. 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 concepts: 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 conventionalized 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 communicative encounter they are immersed in. Mixing metaphors is a consequence of shifting one’s attention to uncommon aspects of metaphoric meaning. By foregrounding 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, crem 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 conventional metaphoric expressions: the compound “butter mountain” and the idiomatic 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 semantic 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 metaphors 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 consequences 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 activating metaphoricity.

3.2  Why should mixing metaphors be problematic? Linguistic Metaphor Theories assume that metaphors require a dual semantic structure and they locate metaphoric meaning on the level of the semantics of language only. For them metaphor is matter of language, not of thought ­(Müller 2008a: 40–61). Linguistic Metaphor Theories (LMT) have often been characterized 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 meaning) 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 semantics (Cohen 1993; Abraham 1998; Davidson 1978; van Dijk 1975;), in pragmatics (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 Metaphor Theories (CogMT), which consider metaphor either as a specific form of thought, or more generally as a cognitive phenomenon. When it comes to asserting 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 generally 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 Theory tend to depart from the assumptions that metaphor is a matter of poetic language 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 differently 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 tradition 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 tradition 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 Cognitive 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 fundamental principle of human cognition (Lakoff and Johnson 1980, 1999). In a nutshell CMT transforms metaphor theory into an experientially grounded theory 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 fundamental form of thought and understanding and accordingly a pervasive and fundamental 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 understanding one thing in terms of another is held to be the primary nature of metaphor. 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 meanings 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 considering 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 considered 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 conventionalized 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 metaphors 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 speakers 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 conscious 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 Conceptual 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 Linguistic Metaphor Theories is what is regarded as most ‘alive’ in Conceptual Metaphor 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 metaphorically 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 grammatical 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 combinations would involve at least one shared entailment, e.g., at least one epistemic correspondence 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 coherent 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 metaphors, but it does not explain, why people nevertheless do it. If, indeed, the conceptual 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 meaning. 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 consciousness and attention play out in the process of metaphor use. This is why they both have difficulties seeing the rationale guiding this widespread linguistic phenomenon. 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 meaning 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 conversations, and which is the line of argument we will pursue. For this particular research Cienki and Swan coded three conversations between students for all syntactic 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 coherence and consistency of conceptual metaphors (including the sharing of entailments), 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 straightforward (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 commonly elaborated on with an adjective like high, so, upholding high standards” (Cienki and Swan 2001: 18, underlining for metaphoric expressions in the original). 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 disposition 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 random errors? For Linguistic Metaphor Theories the answer is relatively simple: mixed metaphors make sense on the level of figurative meaning, the semantic violation 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 mountain, 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 substances. 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 prepositional 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 mixing of metaphors creates a semantic inconsistency on the level of the literal meaning. 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 competence, 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 perspective 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 operating 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 particular 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 of something (awaits distribution)

(a huge amount of something) awaits distribution

sthg = solid substance (butter)

sthg = liquid substance (oil, water, gaz)

The buttermountain has been in the pipeline for some time

temporal and aspectual framing

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 followed. The salience structures of those elements of meaning that go into metaphoric 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 metaphors 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 meaning (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, becoming 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 semantic 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 linguistic 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 unconventional reading concerns not only and maybe even not primarily two incompatible 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 semantically 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. Typically, ‘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. Presumably her attentional focus was on core aspects of the figurative meaning, disregarding 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 formulated in the apodosis ‘… no further activities will be realized’ (input space 2). What the speaker/writer presumably conceptually integrated, is the bare conditional 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 meaning 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 2

INPUT SPACE 1 Protasis

Apodosis

Event structure (P) “If all activity options have been tried without success…“

Event structure (A) “…no further activities will be realized“

Implications for Apodosis: “…a positive solution will be found“

Presupposition for protasis: “repeated unsuccessfull performances of the 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 encounter a particularly interesting case, which is rather complex and stands in opposition 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 contamination 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 different 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 motivated 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 qualities. 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 presumably 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 connotation of the borrowed plumes, a stolen outfit) and create new metaphoric meaning. The example reveals that source domains of conventionalized metaphoric expressions 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 metaphors 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 metaphors 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 

Isar is … itself molting

INPUT SPACE 

sthing is … itself embellishing

sthing is … itself embellishing

– positive connotations – change (child – adult grey – color)

– negative connotations – change at some other person’s account

Isar is … itself adorned with borrowed plumes

“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 metaphoricity 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 foregrounding and backgrounding of aspects of metaphoric meaning. We believe that these observations call for a dynamic view on metaphor, a view that regards conventionalized 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 evidence 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 activated 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 meaning. 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 conventionalized 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 phenomenon ­(Kappelhoff & ­Müller 2011). We argue that in the process of perceiving 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 spectator 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 composition of cinematic movement. Therefore, Kappelhoff and Müller (2011) suggest that, in metaphoric meaning 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 metaphor is created “like a spark that flashes when two semantic fields that were hitherto 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 metaphoricity 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 strategies have shown that this process can be empirically reconstructed by attending to multimodal activation cues, such as a gesture, a picture or a verbal elaboration 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 metaphoric 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 paradox (Steen 2008)” (Steen 2013: 193). Suggesting that only deliberately used metaphors 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 metaphoricity view proposed here. To conclude, Linguistic as well as Conceptual Metaphor Theory cannot convincingly 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 metaphor, 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 processing as comparison, which they consider characteristic for novel metaphors. Obviously 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 multidirectional, 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 conceived 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 process, 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 perspective, 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 postconflict 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 hidden 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 similarity. 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.), Metaphor 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.), Idioms, 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 semantics 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 Empfindsamkeit. 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 movement. (pp. 201–226). Amsterdam: John Benjamins.  doi: 10.1075/aicr.84.16kol Köller, W. (1975). Semiotik und Metapher. Untersuchungen zur grammatischen Struktur und kommunikativen 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 Literaturwissenschaft. München: Max Hueber. McArthur, T. (Ed.). (1992). The Oxford companion to the English language. Oxford: Oxford University 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 linguistischen 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 metaphor (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 memory. In A. Ortony (Ed.), Metaphor and thought (2nd ed., pp. 307–328). Cambridge: Cambridge 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 cognitive 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 published 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 metaphor. 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: Processing, structure, and interpretation. Hillsdale, NJ: Lawrence Erlbaum. Sweetser, E. (1996). Mental spaces and the grammar of conditional constructions. In G. Fauconnier & 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 comprehension. 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 highschool 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 creating an overall meaning of the different metaphoric phrases together. Of course, these two challenges are relevant to understanding any instance of connected discourse, metaphorical or not. Still, the mixing of metaphors, with their very different 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 sometimes perceived as funny because we mock speakers for thinking and speaking in this jumbled manner. But under this hypothesis, people should find mixed metaphorical narratives difficult to interpret. Hypothesis 2, adopted from “conceptual metaphor theory,” states that mixed metaphors reflect the relatively coherent combining of two active conceptual metaphors (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 underu 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 different 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 common 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 matter 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, conceptual metaphor theory has provided abundant data showing how individual target 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 conceive 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 people’s reasoning about source and target domain mappings extends beyond the single conventional parallel mappings typically seen in conceptual metaphors theory. 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 reasoning (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 relationships 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 journey, 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, ancillary 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 submetaphor 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 economic 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 different 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 figurative 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 presents 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 Procedure” (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 contextual 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 followup 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 underestimates 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 metaphorically 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 language 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 previously 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 meaningful given that several BTM excerpts referred to several different target domains (e.g., examining an abstract situation, judging that a problem existed and was getting 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 different 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 members 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 metaphoric 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 referring 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 metaphorical 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 consistent 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 participated 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 meaning?” 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 metaphor (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 ‘generally’) also can give indications of their metaphorical understandings of the different phrases. We reviewed each response for whether it indicated a metaphorical or nonmetaphorical 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 conveying 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 inferring, in this case, that ideas are objects given that the object to be comprehended in this context was abstract, namely the economic crisis. Our examination of the participants’ responses showed that 88% of these alluded to different conceptual metaphors as partly motivating the meanings of the various metaphorical 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 underlying 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 conceptual 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 understanding 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 perceptually 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 speaker’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 people”), 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 participants’ 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 different metaphorical phrases were not isolated, but that they were integrating these interpretations across the various verbal metaphors within each vignette. Consider 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 economic 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 understanding 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 available 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 different 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 understanding 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 metaphorical 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 individual 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 referring to, or alluding to, through the use of different mixed metaphors. In this manner, the present findings support proposals from Barden (this volume, and also see Lee & Barnden, 2001), also given as Hypothesis 3, on the importance of nonparallel 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 metaphorical language, exactly as has been shown in extensive linguistic and psychological 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 understanding of metaphorical language. As the present research shows, people understand 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 verbal 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 conceptual 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 explicitly 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 postconflict 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 University 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 system. Metaphor and Symbol, 16, 29–42.  doi: 10.1207/S15327868MS1601&2_3 Pragglejaz Group. (2007). MIP: A method for identifying metaphorically-used words in discourse. 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 individual 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 continue 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 metaphor 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 ordinary 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-terrain. 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 particular 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 general 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 metonymy. 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 fundamental dimensions, and allowing compromises and overlaps between metaphor and metonymy. That work extends the idea of some other authors that there is a spectrum 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 metaphor 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 handling 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 implemented in computer software – many other implementations of the broad principles 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 mappings 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 ATTMeta’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 ATTMeta 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 people (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 utterance 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 (metaphorically corresponds to, is metaphorical mapped to) at least one sub-person having 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 different 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 metaphorical 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, pragmatically, 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 something they believe it, this sub-person presumably believes that Mick is not adorable. 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 economical, accessible and vivid manner the presence of a conflicted state of mind. This also illustrates one principle of ATT-Meta. The approach, while mappingbased, tries to avoid as far as possible the on-line creation, during the understanding 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 mentioned “part” nor the insisting is mapped to anything. ATT-Meta is focussed on sentences like (7) that, arguably, contain metaphorical elements that are not lexicalized for the hearer (the hearer has no stored ­target-scenario meaning for them) and that, moreover, go beyond the metaphorical 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 following 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 metaphor. 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 continuing 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. However, 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 punctuation 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 requirements 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 serving to flesh out, the source scenarios implied by the sentences. As so far explained, ATT-Meta’s approach resonates with the notion of “metaphorical entailments” in much work on conceptual metaphor theory, based initially 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 similarities 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 linguistics, 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 thinking through what would have happened had Obama lost the 2012 US presidential 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 scenario 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 adorable, 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 subperson 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 reality, including within other pretences (see Section 5.3.1), mappings serve to bridge between a pretence space and the surrounding reasoning space – the space immediately 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 narrative. The ATT-Meta approach is therefore akin to fictionalist approaches to metaphor 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 Relevance 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 central 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 metaphors, 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 linking, 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 ATTMeta 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 metaphorical 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 openendedness 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 correspondence 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 themselves. It is heuristically and presentationally useful to regard ATT-Meta’s mappings 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 subject 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 Mick is adorable”

says(p, adorable(mick)) 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, associated 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 utterance 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 ATTMeta 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 subperson 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 process 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 somewhat 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 conscious 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 assumption 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 consciousself-person, which is in the main part of Anne’s mind-region. This lack of physical 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. Copiability 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 possible 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 unimportant, through being relatively quiet. So, it can be inferred a fortiori that the subperson’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 metaphor 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 conscious 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” Marigold, 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 scenario, 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 pretence into reality in order to get the two connotations listed above. This mapping is done in ATT-Meta by two further VNMAs. One maps qualitative 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) corresponds 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 temporal 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 distress 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), emotional/mental states, ease, and degrees. In all this the only view-specific mapping 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 premises 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) handling affective states, this can be converted into the goal of investigating her affective 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 pretence, 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 blendingtheory 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 metaphorical 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 unconsciously 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 commonly 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) hanging–over(TSA, mind(M))

is–person(CS) in–main–part(CS, mind(M))

REALITY SPACE

is–terrain(mind–of(M)) is–near(TSA, CS) PERSISTENCE INFERENCE is–angry(TSA) is–person(TSA)

persistently (strongly (feels–threatnd(CS,TSA))

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

by VNMAs

persistently (strongly (feels–threatnd (M,TSA)) OUTER 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 inferencing that ­derives is-near(TSA, CS) is not shown

Marigold’s mind and the thought are in all three spaces: reality, the outer pretence 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 corresponds 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 experience 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 conscious 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 threatened 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 Marigold feels threatened by that thought. Further, much as for (19), the VNMA that deals with the temporal characteristics 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 commonsense 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 natural 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 mixing with ideas as internal utterances. The following are some further realdiscourse 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 plausible 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 (weakness) as a heavy physical object.] (28) “[The Dean’s] gaze went slowly up to the ceiling [where there were depictions 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. Provisional contributions to the surrounding space are then drawn by mappings from the two pretences, and these contributions can combine in whatever way is possible 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 realdiscourse 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 suggested 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 various 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 understanding, 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 (accessed 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 illustration, there could be three outer pretences, two of which contain an inner pretence. 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 practice. 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 nevertheless be reported as if it were so. Notice that even if the person has the sensation 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 something 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 animate 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 metaphorically or hyperbolically senseless, though the metaphors here may be lexicalized 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 complex 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 discourse. 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 parallelism. 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 sourcewise parallelism are a matter of viewpoint, and are not objective clear-cut matters. 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 existence 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 viewspecific 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, correspondence rules opportunistically decide from themselves whether they are relevant or not (to put it metaphorically!). This gives great operational flexibility. No topdown 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 fragmentary. 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 metaphorical 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 parallel 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 intuitively 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 assumptions 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 mappings, 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 conclusion 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 reality. 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 parallel 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 parallel 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 metaphorical view of soldiers in the [British] army as chocolate soldiers and (ii) a metaphorical 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. Similarly, 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 separately 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 mixing of (i) the metaphorical view of the soldiers as chocolate ones, and (ii) a metaphorical view of chocolate’s melting as fading. There may not be a single correct analysis. It may be more a matter of personal (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 fading, 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 mappings as opposed to whole metaphorical views, allows a finer grain of analysis and computational processing for mixing purposes. The use of guards in mapping 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 metaphor 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 system 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 Science, 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 linguistics: 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 Psicolingüísticos: 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. Cognitive 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 comprehension. 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: Psycholinguistic 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 educational 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.), Communication 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 including 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 structures 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 metaphorical 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 attention 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 unconscious, how can so many language and communication advisers spend so much time telling writers and speakers to avoid mixing metaphors? Such advice presupposes 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 recognized 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 idiomatic 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, conflicting 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 generates an incorrect idiomatic construction that is then correctly recognized as mixed metaphor. The other two examples both involve the use of one figurative idiom followed by another that is so different that it breaks the expected sense of cohesion between two consecutive parts of discourse, two figurative constructions combining 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 concerns 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 nonmetaphorical 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 nonmetaphorical, 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 distinction 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 metaphors 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 intentionally 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 directedness 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 referent 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 metaphorically 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 metaphor 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 deliberate 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 differentiated 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 distinction 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 thinking 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. Perhaps 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 attention 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 disambiguation, 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 metaphorical. I would contend that this is precisely because it has been used deliberately 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 metaphorical 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 conventional whereas none of them is deliberate, somehow drawing attention to itself as being metaphorical, but ‘setting up new governments’ involves a conventionally 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 metaphor 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 ‘communicative’ dimension of metaphor: it deals with the communicative status of metaphor 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 dism course, i.e. the one that distinguishes between different linguistic forms of metaphor, 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 phenomenon of mixed metaphor as well. As will be appreciated, deliberate metaphor use can depend on very local stylistic 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 nonmetaphorical 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 deliberate metaphor, here are some more conspicuous examples of deliberate metaphor from the same book. The most convincing and extreme case of deliberate metaphor 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, offering 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 functional 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 fundamentally 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 flowering 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 novelty. 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 covered 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 utterance is therefore no accidental use of metaphor. This illustrates how deliberate metaphor use comprises both conventional as well as novel metaphorical structures 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 therefore 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 intention 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 adverbial 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 preceding 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 tearing the veil from the eyes of committed communists. This is a heavy metaphorical 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 mapping 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 personification, 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 dictionary, 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 metaphorical 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 ‘struggle’ 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 deviant 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 iceberg 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 cannibalism 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 winter 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 archipelago, 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 construction: 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 idiomatic construction may have increased this visibility.



Chapter 6.  Mixed metaphor is a question of deliberateness 

Example (18) has also textually reduced two coordinated deliberate metaphors: 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 cannibalistic 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 irreconcilable 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 referents 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 metaphorical 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 combination of two incompatible deliberate metaphors. It is somewhat attenuated, however, if the colon is read as introducing a reformulation of the previous metaphor, 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 deliberate 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 spontaneously 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 talking 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 represented 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 communicative force of one of the two metaphors involved in the conceptual clash is nondeliberate, 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 easily 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 construction 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 problematic. 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 collocation 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 maturity’ 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 metaphorical 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 ‘edifice’ 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 destruction’, 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 situation 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 nondeliberate, 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 looking 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 metaphorical images of peoples thrust back into a grimy, forgotten corner of their own continent 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 nineteenthcentury 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 transcended 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 metaphorical 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 consciousness 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, conceptual 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 discourse. 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 metaphor. 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. metaphorik.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 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 characteristic 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 foreign 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 expressive 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 language (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 language 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 incoherent 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 (OncinsMartínez, 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 “nativespeaker 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 conversational 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 target 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 nonuniversal 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, possibly novel metaphor, in lieu of a more widely known or used one. From the perspective 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 conventionally 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 interplay of metaphors produced by non-native speakers of English, illustrating the type of hybrid metaphors that such speakers produce, and how communicatively successful 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 communicative interaction. I hope to show that efforts to guide teacher intervention or proposed remedies for infelicitous metaphor use in the interlanguage of learners 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 metaphorical 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 learner’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 lexical 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 conceptualizations 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 SubSaharan 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 metaphorical 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 speakers 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 language 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 language. For example, a recent study carried out by Littlemore et al. (2014) 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 qualitative 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 metaphorical 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. (2014, p. 141), 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 Fernsehreporter 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 culture 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 conventionally 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 conceptualizations 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 compared 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 proficiency, the influence of culture-specific conceptualizations and wordings may often be discerned when we examine the way that such speakers express themselves metaphorically, often giving rise to what I have termed hybrid metaphors. Philip (2010) has argued that it is not always a question of conceptual interference 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 conventionally 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 “conceptual 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 ability to express themselves in ways that are coherent with their own social and cultural 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 particular 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 purposes, 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 imperialism implied by the research that seeks to foster adoption of native speaker metaphors 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 communication. 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 varieties 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 metaphorical 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 something 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 completely opaque to ordinary speakers of English. Interestingly, however, for Spanish learners of English (as well as speakers of other Romance languages) it is not. Typically, 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 simply 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 processing (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 particularly 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 metaphorical 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 conventional collocates of ‘hold back’ or ‘nerves’, or that we have here an incomprehensible 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 conventional 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 different factors (cf. Kövecses’ [2011] discussion of context-induced variation for example). It seems likely that the hybrid metaphors of L2 speakers arise from the “pressure of coherence” (Kövecses, 2005, 2009, 2010), in this case the pressure to be coherent with one’s own native metaphorical language and conceptualizations, 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 consider 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 better 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 English 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 metaphor, while the first writer has relied on superficial markers of textual cohesion 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 characteristic 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 English (MacArthur & Littlemore, 2011). In the following extract from a conversation 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 creating 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 reconciliation dialogues studied by Cameron (2007). In their study, Gibbs and Cameron relate the emergence and stabilisation of metaphor in discourse or the ‘metaphoreme’ (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 referring 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 Carmen, 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 ‘outline’ 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. However, 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 expression 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 understand 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 understanding 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 foregoing 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 create 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 implying 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 continua 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 similar 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. Debbie 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. Differences such as these are likely to prove important in conversations between native

 Fiona MacArthur

and non-native speakers and contribute to different understandings and evaluations of non-native speakers’ use of hybrid metaphors. According to Brennan and Clark, conceptual pacts are specific to the interactions 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 (undesirable) 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 comparison. 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: so maybe she likes pu:ppies = 217 S2 =oh man no but this you’re a woman 218 S3 is that what’s bothering you? your shaving @ @@@ 219 S1 @@@@ 220 S2 i don’t MIND a puppy look from time to time? 221 S1 from time to time 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 no it’s my fa:ce we’re talking about (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 conventionally 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 particular 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 idiomatic coinages devised to meet an immediate communicative need” (­ Seidlhofer, 2009, pp. 197–8). When considered in the light of Brennan and Clark’s findings 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” comparison (“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 mixing 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, 2008) 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 conversation 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 maximal 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. Nevertheless, it still seems possible to single out some features of the use of hybrid metaphors 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 metaphors (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 monologic 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 conceptualizations 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 metaphorical 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 different 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 “nativespeaker norms” (whatever they are taken to be) has largely to be abandoned as inappropriate in a world where English is a vehicle of international communication, 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 Linguistics, 27(4), 671–690.  doi: 10.1093/applin/aml032 Chater, N., & Manning, C. D. (2006). Probabilistic models of language processing and acquisition. 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 possibilities. Australian Review of Applied Linguistics, 31(3), 28.1–28.16. Danesi, M. (1995). Learning and teaching languages: The role of “conceptual fluency”. International 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 consciousness 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 meanings 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 prepositions. 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 differences 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. (2014). An investigation into metaphor use at different levels of second language writing. Applied Linguistics, 35(2), 117–144. 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 understood 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 English 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 metaphoric 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. OSUWPL, 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 University 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, deviation 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 interviews 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 communicative 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 conveying 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 purposeful 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 experience, 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 control. 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 medication 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 metaphor 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 metaphor

Semantically Convergent

Extended metaphor

Elaborated metaphor

METAPHOR VEHICLES

Pain as under control

Figure 8.1.  Complex uses of metaphor

Mixed metaphor

Semantically Divergent Pain as beyond control

 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, drawing 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 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 taking 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’ emotions 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 conceptual 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 discourse – 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 listener; 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 particular 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 metaphor 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 qualitative and treats mixed metaphors as normal rather than exceptional. Blending theory (Fauconnier and Turner 2002) and the discourse dynamics approach (Cameron 2010) provide further evidence that metaphors are typically 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) metaphor 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