Windows to the Mind: Metaphor, Metonymy and Conceptual Blending 9783110238198, 9783110238181

Citation preview

Windows to the Mind

Cognitive Linguistics Research 48

Editors Dirk Geeraerts John R. Taylor Honorary editors Rene´ Dirven Ronald W. Langacker

De Gruyter Mouton

Windows to the Mind Metaphor, Metonymy and Conceptual Blending Edited by Sandra Handl Hans-Jörg Schmid

De Gruyter Mouton

ISBN 978-3-11-023818-1 e-ISBN 978-3-11-023819-8 ISSN 1861-4132 Library of Congress Cataloging-in-Publication Data Windows to the mind : metaphor, metonymy and conceptual blending / edited by Sandra Handl, Hans-Jörg Schmid. p. cm. ⫺ (Cognitive linguistics research ; 48) Includes bibliographical references and index. ISBN 978-3-11-023818-1 (alk. paper) 1. Creativity (Linguistics) 2. Metaphor. 3. Metonyms. I. Handl, Sandra. II. Schmid, Hans-Jörg. P37.5.C74W56 2011 415⫺dc22 2010039878

Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.d-nb.de. 쑔 2011 Walter de Gruyter GmbH & Co. KG, Berlin/New York Printing: Hubert & Co. GmbH & Co. KG, Göttingen ⬁ Printed on acid-free paper Printed in Germany www.degruyter.com

Table of contents List of contributors ......................................................................................vii

Introduction................................................................................................... 1 Sandra Handl and Hans-Jörg Schmid

Part I: Metaphor and metonymy: Fundamental issues Methodological issues in conceptual metaphor theory............................... 23 Zoltán Kövecses The structure of metaphor and idiom semantics (a cognitive approach).....41 Dmitrij Dobrovol'skij Why focus on target domains? The importance of domain knowledge in children’s understanding of metaphors.................................63 Aivars Glaznieks Salience and the conventionality of metonymies........................................85 Sandra Handl

Part II: Metaphor and metonymy: Usage-based investigations The role of metaphor scenarios in disease management discourses: Foot and mouth disease and avian influenza.............................................115 Brigitte Nerlich “Overt” vs. “covert” cultural variance in metaphor usage: ‘Europe’ vs. Malta and the EU-membership debate..................................143 Monica Petrica Examining conceptual metaphor models through lexical frequency patterns: A case study of U.S. presidential speeches.................................167 Kathleen Ahrens

vi

Table of contents

Metaphor, constructional ambiguity and the causative resultatives...........185 Beate Hampe

Part III: Conceptual blending Conceptual blending, relevance and novel N+N-compounds....................219 Hans-Jörg Schmid Blending and creativity in metaphorical compounds: A diachronic investigation.........................................................................247 Réka Benczes Reference points in adjective-noun conceptual integration networks.......269 Elena Tribushinina Conceptual blending, evaluation and common ground: George W. Bush and Saddam as friend or foe?.........................................291 Siaohui Kok and Wolfram Bublitz Index..........................................................................................................311

List of contributors

Kathleen Ahrens Professor and Head Language Centre Hong Kong Baptist University 224 Waterloo Road Kowloon Tong Kowloon Hong Kong E-mail: [email protected] Réka Benczes Assistant Professor Department of American Studies School of English and American Studies Eötvös Loránd University Rákóczi út 5. 1088 Budapest Hungary E-mail: [email protected] Wolfram Bublitz Professor of English Linguistics University of Augsburg 86135 Augsburg Germany E-mail: [email protected] Dmitrij Dobrovol’skij Professor of Linguistics Russian Language Institute Russian Academy of Sciences Volkhonka 18/2 119019 Moscow Russia E-mail: [email protected]

viii

List of contributors

Aivars Glaznieks Senior Researcher Institute for Specialised Communication and Multilingualism EURAC research Viale Druso 1 39100 Bolzano Italy E-mail: [email protected] Beate Hampe Professor of English Linguistics Institute of Linguistics University of Erfurt Nordhäuser Str. 63 99089 Erfurt Germany E-mail: [email protected] Sandra Handl Assistant Professor Ludwig Maximilians University Munich Department of English and American Studies Schellingstr. 3 80799 Munich Germany E-mail: [email protected] Siaohui Kok Lecturer of English Linguistics University of Augsburg Universitätsstraße 10 86135 Augsburg Germany E-mail: [email protected] Zoltán Kövecses Professor of Linguistics Department of American Studies Eötvös Loránd University

List of contributors

Rákóczi út 5. 1088 Budapest Hungary E-mail: [email protected] Brigitte Nerlich Professor of Science, Language and Society Institute for Science and Society University of Nottingham Law and Social Sciences Building, West Wing University Park Nottingham NG7 2RD United Kingdom E-mail: [email protected] Monica Petrica Georgenstr. 80 80799 Munich Germany E-mail: [email protected] Hans-Jörg Schmid Chair of Modern English Linguistics Ludwig Maximilians University Munich Department of English and American Studies Schellingstr. 3 80799 Munich Germany E-mail: [email protected] Elena Tribushinina Senior Researcher Computational Linguistics and Psycholinguistics Research Centre University of Antwerp Prinsstraat 13 2000 Antwerp Belgium E-mail: [email protected]

ix

Introduction Sandra Handl and Hans-Jörg Schmid

1. Windows to the mind: Metaphor, metonymy, and conceptual blending The cognitive turn in linguistics, triggered to a large extent by key publications such as Lakoff and Johnson (1980), Lakoff (1987), and Langacker (1987), has led to the now widely shared view that our linguistic behaviour is constrained by the way we experience and perceive the world and by how we conceptualize and construe these experiences and perceptions in our minds. This suggests that the study of language allows us to catch a glimpse of otherwise hidden mechanisms of human thinking. In addition to opening up windows to the mind, the structure and use of language arguably also has an influence on the way our minds work (cf. Pederson 2007). Right from the beginning of cognitive linguistics, the realm of figurative language proved to be an especially fruitful area for studying this reciprocal relation between language and other cognitive abilities. Mostly concentrating on metaphor, research has shown that figuratively motivated expressions abound in everyday language. These conventional figurative expressions can be traced back to deeply entrenched mappings, i.e. wellestablished mental connections between different domains of experience, characteristically between a more concrete source domain and an abstract target domain (cf. e.g. Lakoff and Johnson 1980; Lakoff and Turner 1989). Starting out from typically inconspicuous linguistic examples, such as (1) or (2), conceptual metaphor theorists identify underlying patterns of thinking: (1)

He has strong beliefs.1

(2)

That belief died out years ago.

In both examples, mental issues are assigned the ontological status of concrete entities. (1) bears witness to the fact that essentially abstract concepts such as BELIEFS, IDEAS, and the like can be conceptualized as concrete entities one can possess. In (2), BELIEFS are construed differently, i.e. as

2 Sandra Handl and Hans-Jörg Schmid BEINGS WITH A LIFE CYCLE.

Both types of conceptualization lead to further, related metaphorical ideas: Possessions, for example, can be acquired, bought, and sold, therefore it is possible to do the same with beliefs (cf. 3– 5). When BELIEFS are conceptualized as LIVING BEINGS, they can be regarded as PLANTS, whose growth stands for the development of the beliefs (cf. 6), whose roots signify the basis for the beliefs (cf. 7), and whose cultivation entails encouragement of the beliefs (cf. 8). Beliefs can, however, just as easily be construed as BELOVED (HUMAN) BEINGS (cf. 9), especially as CHILDREN or PETS one has to take care of (cf. 10–11). (3)

He acquired his beliefs during childhood.

(4)

I really buy what he’s saying.

(5)

He tried to sell me a load of hooey.

(6)

This is a flourishing belief in his culture.

(7)

This is a deeply rooted belief.

(8)

I cultivated a belief in my infallibility among my subordinates.

(9)

He espoused that belief publicly.

(10)

He nourished his belief with weekly church visits.

(11)

He fostered the belief within himself.

Illustrative as these examples may be, they also raise some methodological questions, mainly regarding the identification of metaphorical expressions and conceptual mappings. Frequently, studies of conceptual metaphor (e.g. Lakoff and Johnson 1980, 1999; Kövecses 2000, 2002) use invented examples to prove the existence of conceptual mappings. One can even suspect that, at least in some cases, what researchers have in mind first is the mapping rather than the examples, i.e. that they construct examples to fit the mappings proposed. This is certainly a problem, as what is at issue are not the possible conceptualizations language users have at their disposal, but those which are frequently used and shared by the majority of the members of a given speech community, i.e. the conventional metaphors. It cannot be

Introduction

3

denied that in strong contexts speakers are able to use and understand almost any metaphorical conceptualization.2 This, however, only reveals something about speakers’ competence with regard to conceptualizing and decoding, but not about how the mind is structured, about how humans commonly perceive and understand the world. And while examples like (1) – (11) sound natural enough, this does not tell us anything about their authentic use in everyday language. For this reason, the focus of more recent metaphor research has shifted towards usage-based studies (cf. e.g. Cameron 2003, 2007; Deignan 1999, 2005; Nerlich 2004; Nerlich and Halliday 2007; Stefanowitsch and Gries 2007; Steen 2007). They concentrate mostly on finding out how frequently different metaphorical mappings are actually used either in a language as a whole, by relying on large corpora like the British National Corpus, or in various more specific types of discourse, such as political discourse or journalistic discourse. These data-driven approaches go hand in hand with a shift towards more functional considerations. Since metaphors are first of all ways of thinking about topics, they are not only informative about how speakers or writers conceive of a given issue. Especially in text-types such as newspaper articles and political speeches, they can be and certainly sometimes are used consciously to influence the hearers’ or readers’ perception of certain issues. Just as it matters whether a BELIEF is construed as a POSSESSION one can acquire, buy, and sell more or less at one’s one discretion, or whether it is construed as a PET or CHILD one has the moral obligation to take care of and cater to, metaphorical conceptualizations of current events or problems proposed and publicized by politicians or journalists are apt to affect our views of these issues. The language chosen to talk about something thus also has effects on the addressees’ minds, whose current metaphorical structures are therefore continuously updated by linguistic input. It can be argued that the figurative structures entrenched in a person’s mind arise from, and are sustained by, linguistic as well as non-linguistic sources, which constantly influence each other reciprocally (cf. Figure 1). One the one hand, taking a ‘Whorfian’ perspective, figurative thought is influenced by the conventionalized figurative expressions which are part of and current in the surrounding language(s). For instance, if a speaker’s native language ‘teaches’ her or him to talk about TIME in terms of MONEY, it may not seem far-fetched to argue that they will eventually come to conceptualize TIME that way. On the other hand, an individual’s system of figurative thought is shaped by (non-linguistic) perceptions and experiences. These can rely on individual and personal memories, opinions or attitudes,

4 Sandra Handl and Hans-Jörg Schmid which, however, do not tend to develop in isolation, but rather under the influence of socio-cultural models and values shared by larger groups of people (e.g. the culture-specific Japanese conceptualization of ANGER as being located in the hara, literally ‘belly’; cf. e.g. Matsuki 1995). In addition to social factors, universal, as it were pan-human, ones such as bodily experience play a role, manifested for example in the metaphorical conceptualization of GOOD as UP. Closing the feedback loop and again taking a Whorfian stance, the way these essentially non-linguistic memories and experiences are processed and structured by individuals may be influenced by linguistic structures and patterns. The conceptualization of GOOD as UP just mentioned, in addition to being based on the fact that people usually adopt an erect posture when they are happy, may to some degree also be an effect of linguistic conventions. In short, the figurative expressions conventionalized in a given language function both as a central determinant and a mirror image of how the minds of the speakers of the language are structured and work. It is from this perspective that figurative language can be seen as opening up a (methodological) window to the notorious black box. language, communication

cognition

individual experiences/opinions

linguistic structures and use

individual metaphorical system

socio-cultural models and values

embodied experience

Figure 1. Factors determining an individual’s mental metaphorical system

As suggested by the crucial role attributed to shared experiences, cultural models, and, last but not least, shared knowledge about a language, patterns

Introduction

5

of figurative thought entrenched in one individual’s mind can be assumed to be similar to patterns in the minds of speakers with a comparable linguistic and cultural background. This is essentially what conventionality is all about (cf. Langacker 2008: 21). However, it is far from exceptional that we come across novel or previously unfamiliar ways of conceptualizing entities or events. And this concerns not only novel or unfamiliar figurative cognitive construals, but also any other kind of conceptually multi-layered expression. One theory which has considerable potential to explain how we deal with such new or unusual cognitive construals is conceptual blending (also called conceptual integration theory), introduced by Turner and Fauconnier (1995) and further developed in multiple publications, notably Fauconnier and Turner (1998) and (2002). As opposed to conceptual metaphor theory, conceptual blending emphasizes the on-line processes which lead to our understanding of linguistic expressions. Blending theory developed out of Fauconnier’s (1994) mental space theory, an account which underlines that language only prompts us to construct meaning, since it does no more than provide us with “minimal, but sufficient, clues” (1994: xviii). Accordingly, any linguistic input leads to the formation of temporary mental representations, called mental spaces, i.e. “constructs distinct from linguistic structure but built up in any discourse according to guidelines provided by the linguistic expressions” (Fauconnier 1994: 16). A good example to illustrate this are simple metaphorical utterances like (5), repeated here for convenience as (12): (12)

He tried to sell me a load of hooey.

Leaving aside the effect of the verb tried for the time being, conceptual blending would begin by arguing that the two key words sell and hooey will call up two related mental spaces in the hearer’s mind, dubbed ‘commercial transaction’ and ‘communication’ in Figure 2. As the internal structures of these spaces are based on corresponding frames stored in long-term memory and their components (indicated in the figure), the activation of these mental spaces is presumably automatic and effortless. The next assumption of conceptual blending theory is that hearers construct a blended space by projecting selected information from the two input spaces and integrating it. The details of what is projected and how it is integrated depend on a number of so-called vital relations such as identity, similarity, and cause-effect and are restricted by a set of governing principles, among

6 Sandra Handl and Hans-Jörg Schmid them compression, integration, and relevance (see the papers in Part III for more details). This is in fact where the verb tried comes in, since the collocation tried to sell conjures up a scene where it is the seller rather than the buyer who profits from the commercial transaction. This idea is integrated Input 1: Commercial transaction AGENT: seller ACTION: exchange

Input 2: Communication

identity

goods for money THEME: goods BENEFACTIVE: buyer PURPOSE: profit for seller (cf. tried to sell)

AGENT: he ACTION: talk THEME: hooey BENEFACTIVE: me

(=

speaker) PURPOSE:

get ideas

across

AGENT: deceiver ACTION: talk THEME: nonsense BENEFACTIVE:

speaker PURPOSE: make believe, deceive

Blended space Figure 2. Conceptual network of He tried to sell me a lot of hooey

with information projected from Input Space 2, especially the strongly evaluative expression a load of hooey, in such a way that the hearer arrives at the interpretation, represented in the blended space, that the referent of he is trying to deceive the speaker or at least to make him or her believe things that may not be true. While conceptual metaphor theory would presumably try to trace this example to conventionalized metaphors such as IDEAS ARE OBJECTS and the well-known CONDUIT-metaphor of communication (cf. Reddy 1993), it would leave unexplained central components of the interpretation emerging from the juxtaposition of the two domains. These, on the other hand, play an important role in conceptual blending theory and are

Introduction

7

accounted for in terms of notions like compression, integration, and emergent structure. 4 Like conceptual metaphor theory, the theory of conceptual blending has attracted much criticism, since – at least in its early versions, before the optimality principles controlling the most effective generation of blends had been introduced – it seemed much too unconstrained (cf. e.g. Gibbs 2000). However, it is possibly also the open-ended and all-encompassing nature of the cognitive process of conceptual integration proposed by this theory that has made it so attractive to researchers interested in quite diverse types of linguistic structures of different sizes: Blending has proven a powerful tool in explaining long stretches of discourse (cf. e.g. Oakley and Hougaard 2008), advertising texts (cf. e.g. Herrero Ruiz 2006; Joy, Sherry, and Deschenes 2009), riddles and jokes (cf. e.g. Coulson 2001: 179–185; Fauconnier and Turner 1998: 136–142), metaphorical and non-metaphorical phrases and sentences (cf. e.g. Coulson 2001: 125–161; Grady, Oakley, and Coulson 1999), counterfactuals (cf. e.g. Coulson 2001: 203–212; Pérez Hernández 2002), constructions (cf. e.g. Broccias 2006; Mandelblit and Fauconnier 2000), as well as word-formation processes (cf. e.g. Benczes 2006, Ungerer 2007). Similarly to conceptual metaphor theory,5 blending theory thus elucidates structural and regular principles of human cognition as well as pragmatic phenomena. However, there are also some noteworthy differences between the two theories. While blending theory has always been more oriented towards real-life examples, conceptual metaphor theory had to come of age before it was put to the test with data-driven approaches. A further difference between the two theories already alluded to is that blending theory focuses more on the decoding of creative examples, whereas conceptual metaphor theory is well-known for its interest in conventional examples and mappings, i.e. in what is stored in people’s minds. But again, the difference is one of degree and not an absolute one. Blending processes can be routinized and stored if their outcome proves to be useful on more than one occasion. And conceptual metaphor theory is able to explain and accommodate novel figurative linguistic expressions as long as they are compatible with the more general metaphorical makeup of the human mind. Another, perhaps somewhat less important difference lies in the fact that while from the start conceptual blending has pointed to the importance of metonymic construals and thinking for cognitive processes (cf. e.g. Fauconnier and Turner 1998: 158–162), the conceptual metaphor paradigm has long underestimated the role of metonymy, a fact already evident in the

8 Sandra Handl and Hans-Jörg Schmid name commonly used to refer to the theoretical framework. Even though metonymy is already mentioned in Lakoff and Johnson (1980), the book which largely triggered the by now uncountable publications in this area, and even though it has been repeatedly underlined that metonymy might very well be the cognitively more fundamental process (cf. e.g. Lakoff and Turner 1989: 108; Radden and Kövecses 1999: 24; cf. also Lakoff 1987: 77–90), the lion’s share of attention is still devoted to metaphor. This is probably also the reason why equally appropriate names for the more general area of research, like conceptual theory of metaphor and metonymy or conceptual theory of figurative language, still sound somewhat strange and unfamiliar. Although both conceptual metaphor (and metonymy) theory and conceptual blending theory are no longer new and have undergone considerable scrutiny, both theoretical and empirical, there are still fundamental questions to be answered. For the conceptual metaphor paradigm, this relates to questions such as how the conventionality of linguistic expressions and conceptual mappings can be established or the extent to which conceptual mappings as such are cognitively real, i.e. the role adults’ and children’s knowledge of source domains plays in the understanding of a metaphor. For blending theory, this pertains, among other issues, to the cognitive status and relative weighting of the above-mentioned optimality principles, i.e. to the question as to how exactly the generation of a blend is governed by aspects such as integration, unpacking, or relevance. In addition, and despite several attempts to redress this shortcoming (cf. e.g. Gibbs 2000, Stefanowitsch 2007), both theories still suffer from a certain lack of methodological rigour which (indeed) invites justified criticism. The articles in this volume are intended as a contribution to a better understanding of the explanatory potential as well as possible limitations of the two frameworks by taking up basic methodological questions and providing empirical foundations for contested theoretical assumptions.

2. The articles in this volume The present collection originated mainly from the Second International Conference of the German Cognitive Linguistics Association, held in Munich on 5 – 7 October 2006, with some additional, solicited papers which fit the overall focus of the volume. The articles assembled here all share the central idea that cognitive approaches to the study of language open a win-

Introduction

9

dow to how the human mind works and is possibly influenced by available linguistic structures and choices. The volume is divided into three parts. The first and second build in various ways on the conceptual theory of metaphor and metonymy, while the third is devoted to studies set in the framework of conceptual blending theory. The first part addresses fundamental issues in the study of metaphor and metonymy. It begins with a strong, albeit controversial, methodological statement by Zoltán Kövecses. His article is a contribution to the ongoing discussion on the extent to which analyses of conceptual metaphor which are not data-driven can be informative about the role metaphors play in language users’ minds. Kövecses tackles criticism recently levelled at more ‘traditional’ studies of conceptual metaphor by proponents of usage-based, bottom-up approaches, such as Dobrovol’skij and Piirainen (2005) and Stefanowitsch (2007), concerning three different but related points: Firstly, regarding what has been called intuitive metaphor analysis, i.e. the fact that many researchers in the field base their arguments on introspection. Such an approach entails that, secondly, traditional studies potentially miss out on the irregular character of metaphorical language found when looking into natural data. And thirdly, that owing to their intuitive methods, they are hardly able to draw a complete picture of all the possible metaphorical conceptualizations of different target domains. While Kövecses admits that all these criticisms are justified to a certain extent, he builds a strong case for the theoretical and practical value of intuitive studies, mainly by claiming that the results of data-driven research have so far confirmed rather than refuted the assumptions based on intuitive analyses. Dmitrij Dobrovol’skij’s paper focuses on the relationship between the semantics of idioms and their conceptual grounding, and argues that the linguistic description of the semantics and syntax of idioms can profit very much from insights gained by cognitive research. The fact that many idioms like to spill the beans or to let the cat out of the bag are motivated by underlying metaphors has been amply illustrated within cognitive-linguistic research (cf. e.g. Gibbs and O’Brien 1990; Nayak and Gibbs 1990). Dobrovol’skij addresses the problem of the semantic analyzability or decomposability of idioms, a phenomenon which has been the subject of many, also non-cognitively-oriented, publications (cf. e.g. Abel 2003; Geeraerts 1995; Gibbs, Nayak, and Cutting 1989; Nunberg, Sag, and Wasow 1994). Analyzability is related to the more or less autonomous semantic status of some of the constituents of the idiom within the actual, non-literal meaning conveyed by the idiom as a whole. Dobrovol’skij holds that whether or not the

10 Sandra Handl and Hans-Jörg Schmid status of the constituents can be seen as autonomous depends on the mental metaphors underlying the idiomatic expressions. If the structure of the metaphorical mental image and that of the idiom’s lexicalized meaning correlate, the idiom is analyzable. The fact that this also has considerable effects on the discursive behaviour, i.e. the syntactic flexibility, of idioms, is illustrated with natural data taken from the internet. Fundamental questions related to conceptual metaphor theory are also addressed by Aivars Glaznieks. Like Dmitrij Dobrovol’skij, he investigates metaphorically-based idiomatic expressions, but Glaznieks focuses on how children’s understanding of such expressions develops. At the age of four, children have acquired the general ability to comprehend metaphors, i.e. metaphorical competence. Still, not each and every metaphorical expression is understood at this age. It has been found that the further development of children’s metaphorical competence is dependent on their knowledge of the domains involved in the metaphorical mappings (cf. Keil 1986). It could be assumed that it is their knowledge of the source domains rather than that of the target domains that is vital in this respect, since the source domains act as explanatory devices for the targets. Glaznieks, however, provides experimental evidence from children aged five, eight and ten, suggesting that knowledge about the source domains of metaphors may in fact be less important for their acquisition and understanding than was previously believed. Shifting the focus to metonymy, Sandra Handl’s contribution proposes an empirical framework for investigating the hitherto much neglected issue of the conventionality and salience of metonymic meanings. Handl discusses the results of a usage-based study which show that metonymic construals vary a great deal in terms of their conventionality, operationalized as being mirrored in the relative frequency of metonymic meanings of lexemes and expressions in natural discourse. She demonstrates that the conventionality of metonymy can be approached, especially as far as reversible mappings are concerned (e.g. PRODUCER FOR PRODUCT vs. PRODUCT FOR PRODUCER), by applying the laws of ontological salience, as proposed for example by Kövecses and Radden (1998). However, it is argued that a full account of the phenomenon, which explains conceptual regularities and linguistic irregularities alike, can only be given if these more general preferences are supplemented by a consideration of what Handl calls target-invehicle salience, a term which captures the degree to which target-related attributes are salient in the vehicle concept that is used to convey a metonymic meaning.

Introduction

11

The second part of the volume collects papers which share a strong empirical grounding in authentic data and the goal of applying the cognitivelinguistic theory of metaphor in the service of superordinate aims. Both Brigitte Nerlich and Monica Petrica study strategies, exploitations, and effects of the use of metaphor in public discourse. Nerlich examines the role of metaphor in disease management discourses relating to two recent types of disease which received considerable media coverage in the last years, foot and mouth disease and avian influenza. Using UK print media as the source for her empirical investigations, she shows how different metaphor scenarios are created and employed in the media, which then heavily influence public opinion about such socio-political issues (cf. also Musolff 2006). Nerlich suggests that the metaphorical conceptualization of diseases and its change over time can, in general, be explained by a source-path-goal schema, which entails the extensive use of journey metaphors. Accordingly, a virus which has not yet ‘arrived’ in a given country, is construed as travelling. However, once it has reached its goal, i.e. the country, the conceptualization changes and war metaphors prevail. The variance of metaphor usage is also the topic of Monica Petrica’s contribution. She looks into the Maltese journalistic discourse covering the EU-membership of the country. Based on a corpus of English-language newspapers, she identifies metaphor variance of two types: overt and covert. Overt variation describes the more obvious differences between metaphors commonly used in countries like Great Britain or Germany, i.e. the more powerful member states, and Maltese metaphors, i.e. the metaphors of one of the weaker members. These intercultural differences between European and nation-specific metaphors manifest themselves in the use of different source domains. While the former are dominated by sources like FAMILY, GAMES, or BUILDING, the latter depict the EU as a body exercising pressure upon Malta or even as abusing it. Covert variation designates two different forms of variation: Firstly, the use of identical source domains across countries which are, however, linked to different targets in the different states. Secondly, cases in which it seems at first glance as if the sources and targets employed were the same as in other countries, whereas a closer analysis reveals that the sources are actually conceptually different. Petrica shows that the intra-cultural, covert variation in particular can only be noticed and analyzed if the cultural context is taken into account to a sufficient degree. Kathleen Ahrens’ paper is also concerned with political discourse. Her aim lies in uncovering the underlying cognitive models in the speeches of

12 Sandra Handl and Hans-Jörg Schmid US presidents Ronald Reagan, George H.W. Bush Sr., Bill Clinton, and George W. Bush Jr. Ahrens takes the criticism of Lakoff’s (1996, 2002) ideas concerning the two dominant cognitive models related to the two political parties in the US – i.e. the strict father model (MORALITY IS STRENGTH) for the Republicans and the nurturant parent model (MORALITY IS NURTURANCE) for the Democrats – as her starting-point, and proposes a methodology for the identification of metaphorical models through the examination of lexical frequency and co-occurrence patterns in small computerized corpora. An analysis of the frequencies of keywords associated with the two different models proposed by Lakoff as well as a subsequent examination of collocational patterns is revealing in two respects, as Ahrens demonstrates: Firstly, with regard to the more general political convictions of the different presidents, and secondly, concerning how they adjust their metaphors to different types of audiences. Like Ahrens’ paper, Beate Hampe’s contribution relies on corpus data and has a strong methodological focus. Hampe investigates the semantics of grammar and combines metaphor theory and construction grammar in her study of the so-called causative resultatives, which include the CausedMotion Construction (e.g. The warm air pushes other air out of the way), and the Resultative Construction (e.g. If you have fresh maggots, riddle them clean of the sawdust; both examples taken from the International Corpus of English – GB). By way of collostructional analysis (cf. Stefanowitsch and Gries 2003, Gries and Stefanowitsch 2004), it is demonstrated that the postulation of the Resultative Construction and its extensions does not exhaustively account for the semantic potential of the complex-transitive argument structure with adjectival predicative, as there are multiple form-function mappings. In particular, there is a strong, nonresultative verb class, which is referred to by Hampe as the attributive class. This class covers cognition verbs, and the constructional meaning underlying these expressions can be described as (X THINK [Y BE Z]). While metaphorical polysemy links can account for a wide variety of uses of the two types of causative resultatives, it is shown that is is not likely that attributive uses of this argument structure are derived via a metaphorical inheritance link from resultatives ones. Based on this main finding, Hampe differentiates metaphorical links between constructions on different levels of generality, i.e. the schematic and the local level. The third and last part of the volume reflects the growing interest in conceptual blending theory, and is structured along the size of the linguistic units investigated. The section starts with Hans-Jörg Schmid’s study of the

Introduction

13

understanding of novel N+N-compounds. Based on data on the comprehension of invented compounds such as bean-garden or hamburger-shrub investigated by Ryder (1994), Schmid tests the predictions made by conceptual blending theory as to how humans are likely to cope with situations in which they are forced to make sense of novel combinations of existing lexical material. The theory predicts that the process of ‘running the blend’ is constrained by the governing or optimality principles (cf. Fauconnier and Turner 1998, 2002). It turns out that the principles of relevance as well as the maximization of vital relations like CHANGE, SPACE, IDENTITY, and CAUSE-EFFECT can explain large parts of the data analyzed. However, some of the vital relations, i.e. ROLE, REPRESENTATION, ANALOGY, and DISANALOGY, are not confirmed by the data. Due to the restricted data set, this, however, does not falsify Fauconnier and Turner’s assumptions. More importantly, the data suggest further conceptual links not yet explicitly covered by blending theory, such as CONTAINER- or MADE-OF-relations, which are all motivated by the relevance principle hitherto quite unspecified within the framework of blending. Schmid therefore concludes that this principle should be strengthened and amended by adopting a simplified notion of optimal relevance in line with Sperber and Wilson’s (e.g. 1995) relevance theory. The paper by Réka Benczes also applies blending theory to compounds. Benczes tests the potential of the theory to explain creative ad-hoc metaphorical and metonymic N+N-compounds, which have been largely neglected by traditional approaches due to their semantic non-transparency. After an introduction to the general explanatory potential of blending with respect to creative compounding, Benczes’s contribution provides detailed accounts of the meanings of two such compounds, sandwich generation and flame sandwich. It is argued that their actual meanings have developed out of a sequence of different blending operations, all initiated by a first, physical-material blending process which has led to the original meaning of the word sandwich. The paper ends with some theoretical remarks on the justification of using of blending theory to explain N+N-compounds. Elena Tribushinina’s contribution takes the section on blending from word-formation to the semantic structure of premodified noun phrases. In her analyses, which concentrate on ‘simple’ noun modifications via predicating colour adjectives (e.g. red house as opposed to more exotic cases like dolphin-safe or fool-proof), she combines blending theory and ideas from Langacker’s Cognitive Grammar, especially his notions of active zones (e.g. 1984, 1987) and reference points (1993). It is shown that, con-

14 Sandra Handl and Hans-Jörg Schmid trary to what has been pointed out by Murphy (1990), for example, even the understanding of ‘simple’ predicating adjectives like red is contextdependent. It varies with the active zone of the ENTITY SPACE, i.e. the space containing information about the modified noun, which is determined by factors such as e.g. perceptual salience, and discourse relevance. The active zone of the PROPERTY SPACE, i.e. the space containing information about the colour, is accessed, however, via a number of reference points within the spectrum of a given colour. What is more, it is argued that the emergent structure, typically described as being a characteristic of the blended space only, is not restricted to this space. Emergent structure is said to pertain to the whole conceptual integration network, since no one fixed and predetermined reference point exists in the PROPERTY SPACE in the case of premodified noun phrases, but rather different ones among which the decoder has to choose in order to establish mental contact with the relevant active zone. The section closes with Siaohui Kok and Wolfram Bublitz’s contribution, which takes up the register of political discourse also investigated by Nerlich, Petrica as well as Ahrens, but exploits the potential of blending theory to explain the fundamental pragmatic phenomena of common ground and stance/evaluation. They provide detailed analyses of two texts, one political joke and one short extract from a political speech, where the evaluative meaning is not encoded in the lexical or structural surface, but has to be arrived at by way of more complex cognitive processes. Pragmatic theory alone, it is argued, is not sufficient to account for how what is actually meant is inferred from what is said in such cases. In line with blending theory, it is proposed that the addressees’ construal of evaluative meaning depends on setting up and mapping mental spaces which allow them to align their ‘inside-world’ to the speaker’s/writer’s. By doing so, common ground is created, which is accordingly characterized as an emergent configuration composed of semantic as well as attitudinal aspects. Only when this empathetic process of creating common ground is successfully accomplished can the intended evaluative meaning be derived or inferred – either by relying on stored cognitive domains or frames or by constructing short-lived mental spaces.

Introduction

15

Notes 1.

2.

3.

4.

5.

The examples in this section are all taken from the Master Metaphor List (Lakoff, Espenson, and Schwartz 1991). Some of them have been slightly modified. The influence of context on the comprehension of metaphors has been tested in many psycholinguistic experiments (cf. e.g. Ortony et al. 1978; Gibbs and Gerrig 1989; Giora and Fein 1999; Gong and Ahrens 2007). Even though the results are by no means homogeneous, and it has been pointed out that other factors such as familiarity also play a role, most researchers agree that strong contexts facilitate comprehension. Fauconnier and Turner (e.g. 1998, 2002) typically use integration networks consisting of a minimum of four spaces. The so-called generic space which “contains what the inputs have in common” and is linked to each of the input spaces (Fauconnier and Turner 1998: 137) is neglected here. A further aspect which would be highlighted more by blending theory than by conceptual metaphor theory is the following difference between selling goods and convincing somebody of an idea: Once sold, objects belong exclusively to the buyer, but ‘sold’ ideas are not ‘possessed’ solely by the person recently convinced of them. They are usually still shared by the person convincing the other as well. The invariance principle proposed by conceptual metaphor theory (cf. e.g. Lakoff 1990) to solve such problems is not too successful in explaining this inconsistency, since both the source and the target involve events which have largely the same schematic structure. Blending is much more flexible and explicitly allows inconsistencies between mental representations which are related by a conceptual integration network. A concise and useful overview of further similarities and differences between the conceptual theory of metaphor and conceptual blending theory is provided, for example, by Grady, Oakley, and Coulson (1999).

References Abel, Beate 2003 Sprecherurteile zur Dekomponierbarkeit englischer Idiome: Entwicklung eines Modells der lexikalischen und konzeptuellen Repräsentation von Idiomen bei Muttersprachlern und Nichtmuttersprachlern. Tübingen: Niemeyer. Benczes, Réka 2006 Creative Compounding in English. The Semantics of Metaphorical and Metonymical Noun-noun Combinations. Amsterdam/Philadelphia: Benjamins.

16 Sandra Handl and Hans-Jörg Schmid Broccias, Cristiano 2006 Cognitive approaches to grammar. In Cognitive Linguistics. Current Applications and Future Perspectives, Gitte Kristiansen, Michel Achard, René Dirven, and Francisco J. Ruiz de Mendoza Ibáñez (eds.), 81–115. Berlin/New York: Mouton de Gruyter. Cameron, Lynne 2003 Metaphor in Educational Discourse. London: Continuum. Cameron, Lynne 2007 Confrontation or complementarity?: Metaphor in language use and cognitive metaphor theory. Annual Review of Cognitive Linguistics 5 (1): 107–135. Coulson, Seana 2001 Semantic Leaps. Frame-shifting and Conceptual Blending in Meaning Construction. Cambridge etc.: Cambridge University Press. Deignan, Alice 1999 Corpus-based research into metaphor. In Researching and Applying Metaphor, Lynne Cameron, and Graham Low (eds.), 177–199. Cambridge etc.: Cambridge University Press. Deignan, Alice 2005 Metaphor and Corpus Linguistics. Amsterdam/Philadelphia: Benjamins. Dobrovol’skij, Dmitrij, and Elisabeth Piirainen 2005 Figurative Language: Cross-cultural and Cross-linguistic Perspectives. Amsterdam: Elsevier. Fauconnier, Gilles 1994 Mental Spaces. Aspects of Meaning Construction in Natural Language. Cambridge etc.. Cambridge University Press. Fauconnier, Gilles, and Mark Turner 1998 Conceptual integration networks. Cognitive Science 22 (2): 133–187. Fauconnier, Gilles, and Mark Turner 2002 The Way We Think. Conceptual Blending and the Mind’s Hidden Complexities. New York: Basic Books. Geeraerts, Dirk 1995 Specialisation and reinterpretation in idioms. In Idioms: Structural and Psychological Perspectives, Martin Everaert, Erik-Jan van der Linden, André Schenk, and Robert Schreuder (eds.), 57–73. Hillsdale, New Jersey: Lawrence Erlbaum. Gibbs, Raymond W. Jr. 2000 Making good psychology out of blending theory. Cognitive Linguistics 11 (3/4): 347–358. Gibbs, Raymond W. Jr., and Richard J. Gerrig 1989 How context makes metaphor comprehension seem ‘special’. Metaphor and Symbolic Activity 4 (3): 145–158.

Introduction

17

Gibbs, Raymond W. Jr., Nandini P. Nayak, and Copper Cutting 1989 How to kick the bucket and not decompose: Analyzability and idiom processing. Journal of Memory and Language 28: 576–593. Gibbs, Raymond W. Jr., and Jennifer E. O’Brien 1990 Idioms and mental imagery: The metaphorical motivation for idiomatic meaning. Cognition 36: 35–68. Giora, Rachel, and Ofer Fein 1999 On understanding familiar and less-familiar figurative language. Journal of Pragmatics 31: 1601–1618. Gong, Shu-Ping, and Kathleen Ahrens 2007 Processing conceptual metaphors in on-going discourse. Metaphor and Symbol 22 (4): 313–330. Grady, Joseph E., Todd Oakley, and Seana Coulson 1999 Blending and metaphor. In Metaphor in Cognitive Linguistics. Selected Papers from the Fifth International Cognitive Linguistics Conference, Amsterdam, July 1997, Raymond W. Gibbs Jr., and Gerard J. Steen (eds.), 101–124. Amsterdam/Philadelphia: Benjamins. Gries, Stefan Th., and Anatol Stefanowitsch 2004 Extending collostructional analysis: a corpus-based perspective on ‘alternations’. International Journal of Corpus Linguistics 9: 97– 129. Herrero Ruiz, Javier 2006 The role of metaphor, metonymy, and conceptual blending in understanding advertisements: The case of drug prevention ads. Revista Alicantina de Estudios Ingleses 19: 169–190. Joy, Annamma, John F. Sherry Jr., and Jonathan Deschenes 2009 Conceptual blending in advertising. Journal of Business Research 62 (1): 39–49. Keil, Frank C. 1986 Conceptual domains and the acquisition of metaphor. Cognitive Development 1 (1): 73–96. Kövecses, Zoltán 2000 Metaphor and Emotion: Language, Culture, and Body in Human Feeling. Cambridge etc.: Cambridge University Press. Kövecses, Zoltán 2002 Metaphor. A Practical Introduction. Oxford etc.: Oxford University Press Kövecses, Zoltán, and Günter Radden 1998 Metonymy: Developing a cognitive linguistic view. Cognitive Linguistics 9 (1): 37–77. Lakoff, George 1987 Women, Fire, and Dangerous Things. What Categories Reveal about the Mind. Chicago/London: University of Chicago Press.

18 Sandra Handl and Hans-Jörg Schmid Lakoff, George 1990 The invariance hypothesis. Is abstract reason based on imageschemas? Cognitive Linguistics 1: 39–74. Lakoff, George 1996 Moral Politics: What Conservatives Know that Liberals Don’t. Chicago: University of Chicago Press. Lakoff, George 2002 Moral Politics: How Liberals and Conservatives Think. 2nd ed. Chicago: University of Chicago Press. Lakoff, George, Jane Espenson, and Alan Schwartz 1991 Master Metaphor List. 2nd ed. Berkeley: University of California. Lakoff, George, and Mark Johnson 1980 Metaphors We Live By. Chicago/London: University of Chicago Press. Lakoff, George, and Mark Johnson 1999 Philosophy in the Flesh. The Embodied Mind and its Challenge to Western Thought. New York: Basic Books. Lakoff, George, and Mark Turner 1989 More Than Cool Reason. A Field Guide to Poetic Metaphor. Chicago/London: University of Chicago Press. Langacker, Ronald W. 1984 Active zones. Proceedings of the Annual Meeting of the Berkeley Linguistics Society 10: 172–188. Langacker, Ronald W. 1987 Foundations of Cognitive Grammar. Vol. I: Theoretical Prerequisites. Stanford/CA: Stanford University Press. Langacker, Ronald W. 1993 Reference-point constructions. Cognitive Linguistics 4 (1): 1–38. Langacker, Ronald W. 2008 Cognitive Grammar. A Basic Introduction, Oxford: Oxford University Press. Mandelblit, Nili, and Gilles Fauconnier 2000 How I got myself arrested. Underspecificity in grammatical blends as a source for constructional ambiguity. In Constructions in Cognitive Linguistics. Selected Papers from the Fifth International Cognitive Linguistics Conference, Amsterdam, 1997, Ad Foolen, and Friederike van der Leek (eds.), 167–190. Amsterdam/Philadelphia: Benjamins. Matsuki, Keiko 1995 Metaphors of anger in Japanese. In Language and the Cognitive Construal of the World, John R. Taylor, and Robert E. MacLaury (eds.), 137–151. Berlin/New York: Mouton de Gruyter.

Introduction

19

Murphy, Gregory L. 1990 Noun phrase interpretation and conceptual combination. Journal of Memory and Language 29: 259–288. Musolff, Andreas 2006 Metaphor scenarios in public discourse. Metaphor and Symbol 2 (1): 23–38. Nayak, Nandini P., and Raymond W. Gibbs Jr. 1990 Conceptual knowledge in the interpretation of idioms. Journal of Experimental Psychology: General 119 (3): 315–330. Nerlich, Brigitte 2004 Towards a cultural understanding of agriculture: The case of the “war” on foot and mouth disease. Agriculture and Human Values 21 (1):15–25. Nerlich, Brigitte, and Christopher Halliday 2007 Avian flu: The creation of expectations in the interplay between science and the media. Sociology of Health and Illness 29 (1): 46– 69. Nunberg, Geoffrey, Ivan A. Sag, and Thomas Wasow 1994 Idioms. Language 70: 491–538. Oakley, Todd, and Anders Hougaard (eds.) 2008 Mental Spaces in Discourse and Interaction. Amsterdam/Philadelphia: Benjamins. Ortony, Andrew, Diane L. Schallert, Ralph E. Reynolds, and Stephen J. Antos 1978 Interpreting metaphors and idioms: Some effects of context on comprehension. Journal of Verbal Learning and Verbal Behavior 17: 465–477. Pederson, Eric 2007 Cognitive Linguistics and Linguistic Relativity. In Oxford Handbook of Cognitive Linguistics, Dirk Geeraerts, and Hubert Cuyckens (eds.), 1012–1044, Oxford etc.: Oxford University Press. Pérez Hernández, Lorena 2002 Blending vs. conceptual integration in the construction of illocutionary meaning: Counterfactual pieces of advice. Revista Alicantina de Estudios Ingleses 15: 181–194. Radden, Günter, and Zoltán Kövecses 1999 Towards a theory of metonymy. In Metonymy in Language and Thought, Klaus-Uwe Panther, and Günter Radden (eds.), 17–59. Amsterdam/Philadelphia: Benjamins. Reddy, Michael J. 1993 The conduit metaphor – a case of frame conflict in our language about language. In Metaphor and thought, 2nd ed., Andrew Ortony (ed.), 164–201. Cambridge: Cambridge University Press.

20 Sandra Handl and Hans-Jörg Schmid Ryder, Mary Ellen 1994 Ordered Chaos. The Interpretation of English Noun-noun Compounds. Berkeley etc.: University of California Press. Sperber, Dan, and Deirdre Wilson 1995 Relevance. Communication and Cognition. 2nd ed. Oxford: Blackwell. Steen, Gerard J. 2007 Finding Metaphor in Grammar and Usage. A Methodological Analysis of Theory and Research. Amsterdam/Philadelphia: Benjamins. Stefanowitsch, Anatol 2007 Words and their metaphors. In Corpus-based Approaches to Metaphor and Metonymy, Anatol Stefanowitsch and Stefan Th. Gries (eds.), 64–105. Berlin: Mouton de Gruyter. Stefanowitsch, Anatol, and Stefan Th. Gries 2003 Collostructions: Investigating the interaction of words and constructions. International Journal of Corpus Linguistics 8 (2): 209–243. Stefanowitsch, Anatol, and Stefan Th. Gries (eds.) 2007 Corpus-based Approaches to Metaphor and Metonymy. Berlin/New York: Mouton de Gruyter. Turner, Mark, and Gilles Fauconnier 1995 Conceptual integration and formal expression. Journal of Metaphor and Symbolic Activity 10 (3): 183–204. Ungerer, Friedrich 2007 Word-formation. In The Oxford Handbook of Cognitive Linguistics, Dirk Geeraerts, and Hubert Cuyckens (eds.), 650–675. Oxford etc.: Oxford University Press. Ungerer, Friedrich, and Hans-Jörg Schmid 2006 An Introduction to Cognitive Linguistics. 2nd ed. Harlow: Pearson Education.

Part I: Metaphor and metonymy: Fundamental issues

Methodological issues in conceptual metaphor theory Zoltán Kövecses

1. Introduction1 In the past 25 years the metaphor theory worked out originally by Lakoff and Johnson (1980) has become one of the most influential theories of metaphor. At the same time, however, the theory, most commonly known as conceptual metaphor theory, has been strongly criticized over the years for a variety of reasons. Researchers from several disciplines using different perspectives have raised objections to conceptual metaphor theory. In this paper I will examine some of the criticism that has been leveled at conceptual metaphor theory from a methodological perspective. In particular, I will take up the challenge offered by various metaphor researchers concerning three distinct (though related) issues: (1) the idea that practitioners of “traditional” conceptual metaphor theory use intuitive metaphor analyses; (2) the notion that such researchers do not pay sufficient attention to the highly irregular character of linguistic metaphors; and (3) the claim that these researchers work with highly impoverished collections of metaphor. I want to make clear at the outset that, in the main, I agree with these criticisms. However, I also want the critics to recognize that in all three cases we need to keep in mind that our research goals and methods may be justifiably different. Such justifiable differences should not make us assume that the goals and methods of other metaphor researchers are incompatible with our own or that they are inferior to ours. That is to say, I wish to suggest that we should see our different goals and methods in metaphor research as complementary and compatible with each other and not as being superior or inferior to one another.

24 Zoltán Kövecses 2. Intuitive metaphor analysis Several researchers suggest that traditional conceptual metaphor analysis is intuitive and as such lacks scientific rigor and objectivity. Most representative of this kind of criticism is the Pragglejaz Group (2007), of which I am a member (the “z” at the very end; “Pragglejaz” consists of the initial letters of the first names of the participating scholars). Members of the group make the charge that traditional metaphor researchers examine their own mental lexicons or the data found in dictionaries and thesauri, and on the basis of some intuitively found linguistic examples they arrive at conceptual metaphors. The Pragglejaz Group criticizes the approach for two reasons. On the one hand, they claim that traditional metaphor researchers take for granted which expressions are metaphorical, and, on the other, they also suggest that the approach does not pay attention to which actual metaphorical expressions are used of particular target domains by real speakers in natural discourse. They propose, in light of these objections, that we need to construct a reliable methodology to identify metaphorical expressions and we need to use real corpora in the course of identifying such expressions (Pragglejaz Group 2007). I, for one, in addition to Lakoff and Johnson and many others, have been “guilty” of this practice for a long time. But, in my view, the main problem has not been with the intuitive manner in which we arrived at what are called conceptual metaphors but with calling what we arrived at conceptual metaphors, thereby suggesting that we found something actually existing in the human mind. Whether the structures, or systems, we found do or do not exist in the mind cannot be decided on the basis of the intuitive linguistic method that was employed. At the same time, however, several cognitive psychologists, such as Gibbs (1994), Boroditsky (2000), and others, have (to my mind at least) convincingly shown that there is a mental reality behind the structures that were discovered. For this reason, although I admit that we as linguists can only make suggestions for the hypothetical existence of conceptual metaphors, intuitive metaphor analysis has been and still is very useful and necessary for several reasons. Perhaps most importantly, it is useful because its suggestions turned out to be correct in many cases, as psychological experiments indicate. Furthermore, intuitive metaphor analysis saves time in research. In the past 25 years, hundreds of conceptual metaphors have been identified, and, as a result, we are beginning to see at least the vague outlines of the large and intricate system of metaphors that characterizes human conceptual

Issues in conceptual metaphor theory

25

systems. Although imprecise, incomplete, and vague, a large amount of knowledge has been accumulated concerning the nature, content, structure, and universality (and non-universality) of this system. Had we waited for a ‘rigorous’, ‘objective’, and ‘scientific’ methodology to reveal all this, we would probably still not have the idea of conceptual metaphor in its entire complexity, pervasiveness, and social-cognitive power. But we do, and it is time to bring together what we know about what it means to do good science with what we know about conceptual metaphor. First of all, it seems to me that we should be able to recognize the distinct levels at which metaphors function: The supra-individual, the individual, and the sub-individual levels (Kövecses 2002: 239–245, 2010: 305– 311). Conceptual metaphor theory has been “traditionally” formulated at the supra-individual level (the level of de-contextualized metaphorical expressions and the assumed metaphorical conceptual structures based on them), whereas the criticism is formulated at the individual level (the level of fully contextualized metaphorical expressions). Clearly, criticism formulated at one level does not apply to another level. At the supra-individual level, we are typically dealing with largely decontextualized conventionalized metaphorical expressions as native, or what’s worse, non-native, speakers of a language. We establish what are called mappings between the source and target domains. The conceptual metaphors that we find in a language constitute large systems. Two large metaphor systems have been identified: The great chain metaphor that characterizes “things” (Lakoff and Turner 1989: 160–213; Kövecses 2002: 124–134, 2010: 152–162) and the event structure metaphor that characterizes “relations” (Lakoff 1990, 1993). The source domains may consist of a great deal of knowledge that can be carried over to the target in the form of metaphorical entailments, or inferences. Many of the conceptual metaphors that have been identified on the basis of language can also be found in all kinds of cultural institutions (as these are broadly conceived), such as art, science, politics, sports, and so forth (for an overview, see, e.g., Kövecses 2005, 2006). Do people actually have the metaphors in their conceptual system that cognitive linguists discover on the basis of their ‘intuitive’ linguistic analyses? Research shows that conceptual metaphors actually exist in the heads of individual speakers (see, e.g., Gibbs 1994; Boroditsky 2000). This is what I call the individual level. But the same research also showed that the match between the supra-individual and the individual levels is not perfect or complete. The incompleteness of the fit may result from a variety of

26 Zoltán Kövecses factors. The entire range of metaphors at the supra-individual level is not utilized by every single speaker of a language. The individual level is the level at which individual speakers of a given language use the metaphors that are available to them at the supra-individual level in actual communicative situations, but this is also the level at which they create new metaphors. This is the level that is characterized by such issues as the selection of metaphors for particular communicative purposes; how people think online when using metaphors; how the context of communication constrains the use of metaphors; and how metaphors can organize or otherwise structure actual texts or discourses (Kövecses 2002: 239–245). In addition, not all the metaphors that have been, or could be, identified at the supraindividual level are available to all speakers of a language. Both individuals and social groups vary in the kinds of metaphors they use and they also often invent new conceptual metaphors. This is what I called “withinculture” variation in metaphor (Kövecses 2005: 88–113). When people engage in on-line thinking in the course of actual communication, they commonly create blends – both in language and thought. They blend elements of the source with elements of the target (Fauconnier and Turner 2002). Blending is a broader phenomenon than metaphor. We do not need metaphorical source and target domains to get blends; people often use blends on-line or in real time in the course of working conceptually with input domains of any kind. The use of metaphors also depends on the context of communication as broadly conceived. Differential experiences, including the kinds of concerns speakers have, their life histories, the nature and properties of discourse, and even the physical context (such as the particular season in which they communicate), can significantly contribute to the production of metaphors (see, e.g., Kövecses forthcoming). Individuals may also differ in whether or not they make use of all the mappings of a metaphor that are associated with it supra-individually when they use a particular metaphor in particular communicative situations. More often than not, only a selection of a set of conventional mappings is utilized in actual speech situations, depending on one’s communicative needs. Thus it is not the case that all of the mappings arrived at by cognitive linguists at the supra-individual level are activated by individual speakers in the course of on-line thinking and communication in the real world. What I call the sub-individual level of metaphor is the level at which the conceptualization of a conceptual domain (the target) by means of another conceptual domain (the source) is made natural and motivated for speakers.

Issues in conceptual metaphor theory

27

Since the bringing together of the two domains into a conceptual metaphor often involves sensorimotor experiences of various kinds, and since human beings share these experiences (see, e.g., Grady 1997; Lakoff and Johnson 1999; Gibbs 2006), this is the level where universal metaphors are likely to be found. But this is not to claim that each and every conceptual metaphor is based on such correlations in experience. Many are not, and these may obtain their motivation from what we called “perceived structural similarity”, or even real, objective, and preexisting similarity. The two types of motivation (correlations in experience and resemblance or similarity) should be seen as complementary, rather than mutually exclusive. People in different cultures may take the same thing to be similar to different things, and different cultures can have unique concepts that may function as either source or target domains. Because of these possibilities, the sub-individual level of metaphor is only partially universal – to the degree to which motivation is based on correlations in experience. All in all, then, the systematic identification of linguistic metaphors in natural discourse is a goal that is connected with what I call the individual level. For this reason, the objections that are based on this level do not invalidate the goals of the supra-individual level: namely, to propose conceptual metaphors on the basis of linguistic expressions that researchers intuitively take to be metaphorical. At the same time, however, the goals of the two levels complement each other, in that the metaphors suggested on an intuitive basis may prove to be useful in organizing the systematically identified linguistic metaphors into ‘larger’ conceptual metaphors at the individual level.

3. Irregularity of linguistic metaphors Closely related to the previous criticism, several critics raise the issue of the direction of analysis, that is, the issue whether the analysis of metaphors should proceed “top-down” or “bottom-up” (Dobrovol’skij and Piirainen 2005; Stefanowitsch 2007). To make the criticism more specific, let us consider figurative idioms, as discussed by Dobrovol’skij and Piirainen (the discussion that follows in this section borrows from Kövecses 2006). Topdown approaches to figurative idioms find certain data, make particular generalizations given that data, and suggest global cognitive structures (such as conceptual metaphors) that ‘underlie’ and explain the data. This is what many cognitive linguists in the tradition of Lakoff and Johnson have

28 Zoltán Kövecses customarily done in their work on figurative language, including figurative idioms. By contrast, researchers such as Dobrovol’skij and Piirainen begin with an extensive (not selective) set of data, make minimal generalizations about the data, and are much less in the business of suggesting global cognitive structures that account for the data. As a result, cognitive linguists in the tradition of Lakoff and Johnson emphasize the importance of global cognitive structures, while researchers like Dobrovol’skij and Piirainen lay stress on finding a much less general and global conceptual machinery that is needed to account for the total meaning of each and every individual figurative expression whose meaning could not be explained with the help of global cognitive structures alone. Bottom-up research is primarily concerned with the systematic study of metaphorical expressions in real discourse or in large corpora, while top-down research is much less concerned with metaphor in natural discourse or language in general than with metaphorical structures in thought. More and more researchers are studying metaphor in a bottom-up fashion (e.g., Dobrovol’skij and Piirainen 2005). These researchers emphasize the many irregularities in the data they are dealing with, and they suggest that such irregularities cannot be adequately handled by an approach to metaphor that works top-down. In the words of Dobrovol’skij and Piirainen: Central to our approach is the idea that even if the production of CFUs [conventional figurative units; ZK] is governed by some general principles of human cognition, they remain, above all, irregular units of the lexicon. Thus, the most salient features of their semantic structure and discursive behaviour cannot be captured by metalinguistic tools aimed at exclusively discovering regular characteristics. Large portions of CFUs came into being under the influence of certain culture-specific phenomena (Dobrovol’skij and Piirainen 2005: 355–356).

That is to say, in bottom-up approaches a large number of expressions are studied (e.g., an entire corpus), the metaphorical expression are identified on the basis of a well established protocol (Pragglejaz Group 2007), the metaphorical expressions are checked for their detailed behavior (semantic, structural, pragmatic, esthetic, etc.) in concrete contexts of use, and finally conceptual metaphors are established as a result of a multi-stage procedure (see, e.g., Steen 1999). In this kind of approach, what is in the center of attention is language and linguistic metaphors, as well as their behavior in specific contexts. Clearly, those researchers who follow the ‘traditional‘ practice of conceptual-metaphor-theory type of analysis follow the topdown direction, since on the basis of a small number of examples they pos-

Issues in conceptual metaphor theory

29

tulate conceptual metaphors and then they examine the internal structure of these metaphors (mappings, entailments, etc.). In such an approach, what is in the center of attention is (what they conceive of as) the conceptual metaphor itself as a higher-level cognitive structure. If we are primarily concerned with linguistic structures and processes (as opposed to cognitive structures and processes), then we are likely to find irregularities, rather than regularities, in the corpora we are working with. This means that the individual metaphorical expressions will be found predominantly irregular as regards their semantic behavior despite the fact that, in the main, they come into existence as a result of regular cognitive processes, such as conceptual metaphors. That is to say, the main objection is that the top-down approach that emphasizes regularities (like conceptual metaphors) cannot account for the unique and irregular semantic behavior of many metaphorical expressions. According to Dobrovol’skij and Piirainen, the reason for this is that these irregular metaphorical expressions (such as metaphor-based figurative idioms) have often emerged from and have been shaped by certain culture-specific factors and influences, and therefore, they cannot be explained by larger, often universal conceptual metaphors. In other words, a major problem that representatives of the bottom-up approach point out in connection with the top-down approach is that conceptual metaphors do not provide an account for the meaning and, more generally, the irregular linguistic-semantic behavior of many metaphorical expressions. These critics of traditional conceptual metaphor theory believe that the linguistic behavior of metaphors is more irregular than regular, and that this dominant feature of linguistic metaphors is hidden by an approach that emphasizes global cognitive structures, such as conceptual metaphors. In my view, the criticism is partly valid, partly not. Let us take as an example the metaphorical expression split hairs (taken from Dobrovol’skij and Piirainen 2005). How can we explain the meaning of the expression (i.e., its semantic behavior) in light of traditional conceptual metaphor theory? In order to be able to do that, we would have to know which global conceptual metaphor underlies the expression, that is, the metaphor that motivates the meaning ‘to pay too much attention to small and unimportant differences in an argument’. It is not easy to find a global conceptual metaphor that can readily and naturally account for this meaning. This situation is unlike cases where we have well known metaphors, such as ANGER IS A HOT FLUID IN A CONTAINER (to account for the word boil), THEORIES ARE BUILDINGS (to account for foundation), or LIFE IS A JOURNEY (to account

30 Zoltán Kövecses for bumpy road), that easily come to mind and readily motivate the behavior of many metaphorical expressions. If there is no easily retrievable conceptual metaphor that accounts for the meaning of the expression split hairs in a natural way, then the expression will remain hopelessly unmotivated, that is, irregular, as regards its meaning. But even if there is such a global metaphor in this case (though it’s not easy to find for the example), it is likely that there are many other cases where no global conceptual metaphors can be discovered and used to explain the semantic behavior of particular metaphorical words and expressions (see also Deignan 1999). Given this, we cannot, at least in my judgment, claim that there is a global conceptual metaphor behind, or underlying, each and every metaphorical expression. Thus, in this respect, I find the criticism valid. But, in another respect, it is not. In Dobrovol’skij and Piirainen’s (2005) view, conceptual metaphor theory cannot in many cases account for the dissimilar behavior of metaphorical linguistic expressions that are instances of the same conceptual metaphor. According to Dobrovol’skij and Piirainen, the fact that two or more linguistic expressions belong to the same conceptual metaphor does not explain the different meanings these expressions have. Let us take the following expressions as examples: Add fuel to the fire and flare up, as in His stupid comment just added fuel to the fire and The argument flared up between them. Although we can be sure that what motivates the expressions is the ARGUMENT IS FIRE conceptual metaphor, Dobrovol’skij and Piirainen would maintain that the difference in their meaning is not explained by the conceptual metaphor alone. The meaning ‘to increase the intensity of the argument’ (in the case of add fuel to the fire) and the meaning ‘the intensity of the argument increases suddenly’ (in the case of flare up) are significantly different, but the difference is not captured by the theory of conceptual metaphor. What Dobrovol’skij and Piirainen leave out of their account is that conceptual metaphor theory is not exhausted by setting up global conceptual metaphors, like ARGUMENT IS FIRE, or ANGER IS A HOT FLUID IN A CONTAINER, or THEORIES ARE BUILDINGS. As a matter of fact, the setting up of such metaphors is merely a first step in the course of analysis. What we need to do in addition is to see which elements of the source correspond to which elements of the target domain. These are the correspondences (or mappings) that crucially constitute conceptual metaphors. In the case of the ARGUMENT IS FIRE metaphor, the mappings display a complex fine-grained structure. Since the ARGUMENT IS FIRE metaphor is part of a more general metaphor (INTENSITY IS HEAT), it is appropriate to deal with the mappings

Issues in conceptual metaphor theory

31

of the ARGUMENT IS FIRE metaphor within the framework of this more general, global metaphor (see Kövecses 2002: 112–116, 2010: 140–144). The INTENSITY IS HEAT metaphor consists of the following mappings (examples taken from Deignan 1995): INTENSITY IS HEAT THE HIGHEST DEGREE OF INTENSITY IS THE HIGHEST DEGREE OF HEAT

His eyes blazed intently into mine. The president launched his anti-drug campaign in a blaze of publicity. The career that began in a blaze of glory has ended in his forced retirement. CHANGE OF INTENSITY IS CHANGE IN HEAT

Then, in the last couple of years, the movement for democracy began to heat up. The battle for the Formula One Championship hotted up. In a clear bid to take the heat out of the rebellion, he authorized an interest rate cut. I think that the Scottish problem might cool off. CAUSING INTENSITY TO INCREASE IS STARTING HEAT (LIGHTING)

She has failed to ignite what could have been a lively debate. The strike was sparked by a demand for higher pay. MOTIVATION TO DO SOMETHING INTENSELY IS AN INTERNAL CAUSE OF HEAT

He said they were looking for someone with a bit of spark as the new technical director. CONTROLLING THE SITUATION IS CONTROLLING THE HEAT

This proved insufficient to dampen the fires of controversy. MAINTAINING INTENSITY IS MAINTAINING HEAT

The fact is that the very lack of evidence seems to fan the flames of suspicion. The president warned that this will fuel the fires of nationalism.

32 Zoltán Kövecses A SUDDEN INCREASE IN INTENSITY IS A SUDDEN INCREASE IN HEAT

Dozens of people were injured as the fighting flared up. Dale stayed clear of the disease for six years until it flared up last summer. LATENT INTENSITY IS POTENTIAL HEAT

The government was foundering on an issue that had smoldered for years. INTENSITY CEASING IS THE HEAT GOING OUT

Some were simply burnt out, exhausted. … a burnt-out business executive. This analysis of the linguistic metaphors reveals a great deal of patterning. It strongly suggests (in the form of a hypothesis) that, at least on an unconscious level, the linguistic metaphors manifest an elaborate conceptual structure as well. Of these mappings, different ones will apply to His stupid comment just added fuel to the fire and The argument flared up between them. The meaning of the expression add fuel to the fire is based on and motivated by the mapping ‘maintaining heat Æ maintaining intensity’, whereas the meaning of the expression flare up is based on and motivated by the mapping ‘a sudden increase in the degree of heat Æ a sudden increase in the degree of intensity’. Although this does not fully account for the meaning of the expressions, we can see how the detailed and specific mappings can systematically motivate subtle differences between the semantics of linguistic expressions that belong to the same conceptual metaphor. In sum, it can be suggested the bottom-up approach places too much emphasis on irregularities in the same way as the top-down approach places too much emphasis on what is regular in our metaphorical conceptual system. Ideally, both directions of analysis are necessary for a full and mutually satisfying understanding of how metaphors work in language, culture and thought. To my mind, the interesting question is what percentage of the metaphors found in natural discourse is regular or irregular in the sense above.

Issues in conceptual metaphor theory

33

4. Partial inventory of metaphors The third objection that is commonly raised against the traditional methodology of conceptual metaphor theory is that researchers use not only intuitively arrived-at metaphors but also impoverished collections of them.2 The suggestion is that our overall goal should be to find each and every linguistic and conceptual metaphor relating to a target in a given corpus. The claim is especially characteristic of practitioners of corpus linguistics (see, e.g., Stefanowitsch 2007). Again, while I agree with the general claim that it is desirable to identify all linguistic and conceptual metaphors in a corpus, I feel we need to raise a deeper issue in this connection. The question we need to ask is why, indeed, it is so important for us to find metaphors for particular target domains. Is it because it is inherently good to find the complete repertoire of metaphors and to produce a complete list in connection with a target domain or is it because a complete list could be used for some useful purpose? No doubt, completeness (i.e., to have exhaustive sets of metaphors for particular concepts) is a good thing in itself, but I believe this is not the most important reason for the pursuit of metaphor in cognitive linguistics. The most important reason seems to me to be able to see to what extent the metaphors contribute to the conceptualization of abstract concepts, as well as their cognitive representation, and what content they contribute. After all, our ultimate goal is to understand, as best as we can, the mind that consists of a large number of concepts that function as target domains associated with an equally large number of source domains. As an example of such cognitive-cultural models, consider the hypothetical model below I proposed elsewhere (Kövecses 1991, 2002: 88–90, 2010: 113–114) for what I called “happiness as an immediate response”. I also suggested that this form of happiness is most commonly accessed in everyday communication by means of the word joy. Cause of joy: You want to achieve something. You achieve it. There is an immediate emotional response to this on your part. Existence of joy: You are satisfied.

34 Zoltán Kövecses You display a variety of expressive and behavioural responses including brightness of the eyes, smiling, laughing, jumping up and down. You feel energized. You also experience physiological responses, including body warmth and agitation/excitement. The context for the state is commonly a social one involving celebrations. You have a positive outlook on the world. You feel a need to communicate your feelings to others. The feeling you have may “spread” to others. You experience your state as a pleasurable one. You feel that you are in harmony with the world. You can’t help what you feel; you are passive in relation to your feelings. The intensity of your feelings and experiences is high. Beyond a certain limit, an increase in intensity implies a social danger for you to become dysfunctional, that is, to lose control. It is not entirely acceptable for you to communicate and/or give free expression to what you feel (i.e., to lose control). Attempt at control: Because it is not entirely acceptable to communicate and/or give free expression of what you feel, you try to keep the emotion under control: You attempt not to engage in the behavioral responses and/or not to display the expressive responses and/or not communicate what you feel. Loss of control: You nevertheless lose control. Action: You engage in the behavioral responses and/or display expressive responses and/or communicate what you feel. You may, in addition, exhibit wild, uncontrolled behavior (often in the form of dancing, singing, and energetic behavior with a lot of movement). We can think of the emergence of this model from the metaphors, metonymies, and related concepts in the following way: Take, for instance, the idea that when we are very happy, there is some loss of control involved. An indication of this idea is given in a number of conceptual metaphors,

Issues in conceptual metaphor theory

35

such as HAPPINESS AS A NATURAL FORCE, HAPPINESS AS AN OPPONENT, HAPPINESS AS A CAPTIVE ANIMAL, and HAPPINESS IS INSANITY. The typical linguistic examples of these metaphors suggest that the person who is intensely happy is likely to undergo some loss of control (we are overwhelmed, we are seized, we go crazy, etc.). Thus, the language we use about happiness reveals the way we (at least assume to) think about happiness, and the way we think about it is given in a prototypical cognitive model. I arrived at this model by means of studying the linguistic and conceptual metaphors I found at the supra-individual level. The corpus-linguistic approach can identify a great deal more metaphorical expressions for particular target domains than the intuitive, ‘traditional’, approach can (e.g., Stefanowitsch 2007). At the same time, however, this quantitative advantage does not necessarily lead to a qualitative advantage. Stefanowitsch examined five words (anger, fear, joy, sadness, and disgust) related to emotions in the British National Corpus. His results, in the main, support the results of top-down researchers: In English, the most characteristic metaphor for ANGER is HEAT, the one for HAPPINESS is an UPWARD ORIENTATION, and so forth. The corpus-linguistic approach did not find most of the specific-level emotion metaphors as being significantly unique to particular emotion concepts. At the same time, these are the metaphors that serve as the basis of our knowledge about emotion concepts. Metaphors such as EMOTIONS ARE OPPONENTS, EMOTIONS ARE NATURAL FORCES, EMOTION IS INSANITY, and so forth, define key aspects of emotion concepts (such as control, passivity, lack of control). This discrepancy between the results makes it clear that our intuitions (in this case, about emotion concepts) are inevitable for the interpretation of the results produced by corpus linguistics. Generic-level metaphors make up the majority of emotion metaphors in Stefanowitsch’s study. These are such event structure metaphors as (EMOTIONAL) STATES ARE CONTAINERS, (EMOTIONAL) STATES ARE OBJECTS, CAUSES (OF EMOTION) ARE FORCES, and (EMOTIONAL) CHANGE IS MOTION. These generic-level metaphors apply to any state, cause, or change, and for this reason they have limited usefulness in characterizing the specific content and structure of prototypical models of emotion concepts. This is so even though we find several interesting differences between emotion concepts at the level of generic metaphors. Therefore, the identification of such generic-level metaphors in corpus studies leads to a somewhat disappointing result; namely, that the most

36 Zoltán Kövecses common metaphors in the data are the least interesting ones, given the goal of characterizing the cultural-cognitive models associated with target concepts. The fact that approximately 70% of, say, the happiness metaphors consist of the object version of the event structure metaphor (as in giving someone joy) is perhaps not the most revealing feature of the culturalcognitive model of happiness. I believe that corpus studies of particular target concepts should pay more attention to the analysis of those conceptual metaphors that can be considered “central” with regard to target concepts. These are the metaphors that contribute most to the structure and content of abstract concepts. I feel that corpus linguistics cannot really meet our expectations without taking into account the suggestions for hypothetically proposed culturalcognitive models that are based on central metaphors (and metonymies, etc.) that researchers working in the traditional mode have to offer. As a matter of fact, it seems reasonable to expect corpus linguists to design research programs that could help us discover how our target concepts are constituted and represented in the corpora they study, since the structure and content of concepts must be, at least to some degree, retrievable from the large corpora corpus linguists typically work with. It seems to me that a large part of this could be done without necessarily looking for all the linguistic and conceptual metaphors in corpora. A final aspect of corpus studies I would like to mention bears on the issue of who is in the best position to study target concepts in a language in a useful way. I think corpus linguistics is an excellent way of studying target concepts for native and non-native speakers alike. In the same way as corpus studies can correct mistaken beliefs by both native and non-native speakers about the behavior of all sorts of syntactic and morphological phenomena, they can also lead to the clarification of misconceptions by both native and non-native speakers concerning the structure and content of concepts. Such clarifications can be done by non-native speakers just as much as by native speakers using the techniques of corpus linguistics. This could considerably open up research possibilities on concepts to non-native speakers. In sum, as regards the third objection concerning the direction of analysis, in my view the identification of linguistic and conceptual metaphors should be as complete as possible, but, as we have seen above, quantitative metaphor analysis must be supplemented by intuitive qualitative analysis.

Issues in conceptual metaphor theory

37

5. Conclusions I have examined three recent methodological challenges to the practice of conceptual metaphor theory: (1) the objection that traditional conceptual metaphor theory uses intuitive metaphor analyses; (2) the objection that there is much more irregularity in linguistic metaphors than traditional cognitive linguists care to admit; and (3) the objection that traditional conceptual metaphor theory works with impoverished sets of data. As for the first criticism, I have argued that the goals of metaphor analysis on one level cannot be used to invalidate the goals of metaphor analysis on another level. As for the second, I suggested the bottom-up approach places too much emphasis on irregularities (and thus ignores regularity) just as the top-down approach places too much emphasis on what is regular in our metaphorical conceptual system (and thus ignores irregularity). As for the third, quantitative metaphor analysis must be supplemented by intuitive qualitative analysis. In all three cases, my suggestion was that the traditional and the more recent approaches must complement each other in specific ways. It follows from this that neither approach is in any way superior to the other. Metaphor analysis is only complete (if this is the right word) if it takes advantage of both the intuitive and the more objective methods, if it pays equal attention to regularity and irregularity in the use of metaphors, and if the qualitative and the quantitative approaches are duly balanced in our analyses.

Notes 1. 2.

I am grateful to Linda Thornburg, Réka Benczes, and Péter Csatár for reading and commenting on the paper. This section is, in part, a reply to Anatol Stefanowitsch’s (2007) critique of Kövecses (1998).

References Boroditsky, Lera 2000 Metaphoric Structuring: Understanding time through spatial metaphors. Cognition 75 (1): 1–28.

38 Zoltán Kövecses Deignan, Alice 1995 Collins Cobuild English Guides 7: Metaphor. London: HarperCollins Publishers. Deignan, Alice 1999 Corpus-based research into metaphor. In Researching and Applying Metaphor, Lynne Cameron and Graham Low (eds.), 177–199. Cambridge: Cambridge University Press. Dobrovol’skij, Dmitrij and Elisabeth Piirainen 2005 Figurative Language: Cross-cultural and Cross-linguistic Perspectives. Amsterdam: Elsevier. Fauconnier, Gilles and Mark Turner 2002 The Way We Think: Conceptual Blending and the Mind’s Hidden Complexities.New York: Basic Books. Gibbs, Raymond W. 1994 The Poetics of Mind: Figurative Thought, Language, and Understanding. Cambridge: Cambridge University Press. Gibbs, Raymond W. 2006 Embodiment and Cognition. Cambridge: Cambridge University Press. Grady, Joseph 1997 Foundations of Meaning: Primary Metaphors and Primary Scenes. Ph. D. diss., Department of Linguistics, University of California at Berkeley. Kövecses, Zoltán 1991 Happiness: A definitional effort. Metaphor and Symbolic Activity 6: 29–46. Kövecses, Zoltán 1998 Are there any emotion metaphors? In Speaking of Emotions, Angeliki Athanasiadou and Elzbieta Tabakowska (eds.), 125–151. Berlin: Mouton de Gruyter. Kövecses, Zoltán 2002 Metaphor: A Practical Introduction. New York: Oxford University Press. Kövecses, Zoltán 2005 Metaphor in Culture: Universality and Variation. New York: Cambridge University Press. Kövecses, Zoltán 2006 Language, Mind, and Culture: A Practical Introduction. New York: Oxford University Press. Kövecses, Zoltán 2009 Metaphor, culture, and discourse: The pressure of coherence. In Metaphor and Discourses, Andreas Musolff and Hans-Joerg Zinken, (eds.), 11–24. Houndsmills/New York: Palgrave Macmillan.

Issues in conceptual metaphor theory

39

Kövecses, Zoltán 2010 Metaphor: A Practical Introduction. 2nd ed. New York: Oxford University Press. Lakoff, George 1990 The invariance hypothesis: Is abstract reason based on image schemas? Cognitive Linguistics 1 (1): 39–74. Lakoff, George 1993 The contemporary theory of metaphor. In Metaphor and Thought, Andrew Ortony (ed.), 202–251. New York: Cambridge University Press. Lakoff, George and Mark Johnson 1980 Metaphors We Live By. Chicago: University of Chicago Press. Lakoff, George and Mark Johnson 1999 Philosophy in the Flesh: The Embodied Mind and its Challenge to Western Thought. New York: Basic Books. Lakoff, George and Mark Turner 1989 More Than Cool Reason: A Field Guide to Poetic Metaphor. Chicago: University of Chicago Press. Pragglejaz Group 2007 MIP: A method for identifying metaphorically used words in discourse. Metaphor and Symbol 22 (1): 1–39. Stefanowitsch, Anatol 2007 Words and their metaphors. In Corpus-based Approaches to Metaphor and Metonymy, Anatol Stefanowitsch, and Stefan Th. Gries (eds.), 64–105. Berlin: Mouton de Gruyter. Steen, Gerard J. 1999 From linguistic to conceptual metaphor in five steps. In Metaphor in Cognitive Linguistics: Selected Papers from the Fifth International Cognitive Linguistics Conference. Amsterdam, July 1997, Raymond W. Gibbs, and Gerard J. Steen (eds.), 57–77. Amsterdam/Philadelphia: John Benjamins.

The structure of metaphor and idiom semantics (a cognitive approach) Dmitrij Dobrovol’skij

1. Theoretical prerequisites The main postulate of cognitive linguistics, from which all its methods and heuristics can be derived, is the idea that there are conceptual structures standing behind the linguistic structures and that, therefore, investigation into linguistic structures has to involve addressing relevant conceptual structures (cf., for example, Talmy 2000; Croft and Cruse 2004). This basic assumption is what distinguishes the cognitive approach from other directions of linguistic research. From a theoretical point of view some questions arise in connection with the idea of the relevance of knowledge structures corresponding to linguistic facts under consideration. 1. 2. 3. 4. 5.

How is it possible to analyze knowledge structures independently from language? How can the explanations involving knowledge structures be verified? What metalanguages have to be developed in order to be able to take into account relevant conceptual structures? How can cognitively oriented results be properly embedded in linguistic descriptions? In which domains of linguistic research is the use of cognitive methods especially promising?

As for the first two questions, they cannot be answered in the framework of linguistics. To be able to investigate conceptual structures and to verify assumptions about their features relevant to language functioning one has to address disciplines other than linguistics and to use methods of psychological empirical research. Even in this case, questions of this nature can be answered only in part because the concepts are not directly accessible. However, the last three questions can reasonably be answered within the scope of linguistics. This paper is to be regarded as a contribution to an-

42 Dmitrij Dobrovol’skij swering the question as to how to include results of cognitively oriented idiom research in their purely linguistic descriptions, or how linguistic semantics and syntax can profit from cognitive findings. The paper discusses the phenomenon of semantic analyzability of idioms – with its cognitive foundations – as the conceptual basis for some specific features of their discursive behaviour. This phenomenon is one of the central issues in cognitively oriented idiom research because it is related to the structure of the underlying metaphor and therefore to the conceptual basis of idiom semantics.

2. Semantic analyzability of idioms: General issues The notions of semantic analyzability or decomposability have been frequently debated, both within and outside of the field of cognitive linguistics (see, e.g., Rajxštejn 1980; Dobrovol’skij 1982, 1997, 2000, 2004; Wasow, Sag, and Nunberg 1983; Gibbs, Nayak, and Cutting 1989; Geeraerts and Bakema 1993; Geeraerts 1995; Nunberg, Sag, and Wasow 1994; Dobrovol’skij and Piirainen 1997, 2005; Keil 1997; Titone and Connine 1999; Abel 2003; Langlotz 2003, 2006). Analyzable idioms are expressions in which certain constituents have more or less autonomous meanings in the scope of the lexicalized figurative meaning,1 e.g. to throw the baby out with the bath (water) ‘to destroy what is good or important about a situation by mistake while trying to change and improve it’. This idiom can be considered to be analyzable or decomposable: autonomous meanings can be ascribed to baby (meaning roughly ‘the good’), throw out (‘get rid of’) and bath water (‘the entity not wanted anymore’). In this case, we are dealing with a kind of homomorphism between the structure of the lexicalized meaning and the structure of the underlying metaphor. Homomorphism of this nature is cognitively real; it means that the speakers who know this idiom are able to associate its constituent parts with the corresponding parts of its lexicalized meaning (compare, e.g., Gibbs, Nayak, and Cutting 1989; Gibbs and Colston 2007: 826–827). This leads to the analyzability or decomposability of such expressions including the impact on their discursive behaviour; e.g. possible syntactic transformations such as that was the baby that was thrown out with the bath water. A certain freedom in the discursive behaviour of idioms, including alterations of their lexical and syntactic structure, is typically considered to be the main argument in favour of their

Structure of metaphor and idiom semantics

43

semantic analyzability, i.e. semantic autonomy of their constituent parts (compare, e.g., Wasow, Sag, and Nunberg 1983: 108). Discussing notions like semantic analyzability, decomposition, idiomatic combinations, etc. we have to bear in mind that they are, so to speak, of secondary nature, i.e. all these terms refer to semantic autonomy ex post factum. To be able to handle an idiom as a “combining expression” or an “idiomatic combination” (cf. Nunberg, Sag, and Wasow 1994: 497) the speakers have to know the lexicalized meaning of a given idiom, i.e. the knowledge of the overall meaning of the idiom is the mandatory prerequisite for ascribing quasi-autonomous meanings to its constituents. These quasi-autonomous meanings of the idiom constituents are obviously not identical with their literal meanings. Further English idioms displaying semantic analyzability are to grasp the nettle, to pull strings, to spill the beans, to let the cat out of the bag, to keep/juggle the balls in the air, to cross one’s bridges before one has come to them. Some linguists believe that most idioms behave in this way; cf., for instance, Wasow, Sag, and Nunberg (1983: 109): “we claim that pieces of an idiom typically have identifiable meanings which combine to produce the meaning of the whole”. The question remains open whether the idioms are typically analyzable or rather typically holistic, i.e. semantically nondecomposable. In order to answer this question, additional empirical data from a wide range of languages is needed. One point is clear: just as with many other linguistic phenomena, the analyzability of idiom semantics is a matter of degree. In many cases, the judgement of idiom analyzability is individually or contextually based, i.e. differs from person to person2 and from context to context. Thus, it often makes no sense to argue over the issue whether a certain individual idiom is decomposable or not. What is important is the fact that if I perceive an idiom as being analyzable, I will modify it in my discourse in a way similar to to throw the baby out with the bath (water). If another speaker disagrees with this view, this speaker would not modify the given idiom in such a way. Anyway, in many cases most speakers would agree in their judgements about an idiom’s analyzability. So, e.g., they would agree that it is not possible to ascribe more or less autonomous meanings to the constituents of English idioms such as to go west, to bite the dust, to have something at one’s fingertips, to stick one’s neck out, to rattle someone’s cage, to split hairs. These idioms are non-analyzable, at least in a linguistically relevant sense. This means that they do not allow alterations which are acceptable with idioms like to throw the baby out with the bath (water). This does not mean that the constituents

44 Dmitrij Dobrovol’skij of such idioms have no autonomous status whatsoever from a purely psychological point of view. Nor does this mean that non-analyzable idioms do not undergo playful variation in puns and jokes. What is crucial for our further discussion is that these idioms are not subject to more or less regular syntactic transformations and adaptations in conventional speech, i.e. outside playing on words, and that their constituent parts do not show any clear correlation with the parts of their lexicalized meaning, hence are not autonomous with regard to lexical semantics. Different views on the notion of semantic autonomy will be discussed in more detail in Section 3. There are several reasons for the growing interest in the notion of semantic analyzability. First of all, this phenomenon is theoretically challenging. As is well known, the traditional “school-grammar” view of idioms, as well as their generative treatment is based on the idea of their total noncompositionality (cf. the widespread metaphor “idioms as long words” or the generativist definition of idioms as “anomalies of interpretation”). The fact that some idioms behave as lexical units with structural parts which are more or less autonomous semantically demands a rethinking of many categories established in idiom research. Second, there is some empirical evidence that the constraints on syntactic alterations of a given idiom partly depend on the degree of its semantic analyzability. The notion of semantic analyzability therefore gets a certain operational value or, at least, an explanatory potential for research on idiom syntax. As for cognitive linguistics, it regards this phenomenon as one of the most crucial aspects of idiom semantics. There are psycholinguistic findings which show that speakers process analyzable idioms in a different way than the non-analyzable ones, including their syntactic behaviour (cf., e.g., Gibbs and Nayak 1989). The question which idioms are to be considered analyzable and why cannot be answered at the level of linguistic structures. Phenomena of this kind are obviously conceptual by nature. The decisive criterion here is the structure of the underlying metaphor (cf. Dobrovol’skij 2000). In this paper, I briefly discuss some specific features of the discursive behaviour of idioms with regard to their analyzability, the possibility of applying some operational criteria in this field, essential cognitive foundations of this phenomenon, above all, the internal structure of the underlying metaphor, and the relationships between semantic analyzability and motivation.

Structure of metaphor and idiom semantics

45

3. Discursive behaviour of analyzable idioms The crucial question for the theory of phraseology is whether or not the internal semantic structure of an idiom influences its discursive behaviour, including its ability to participate in syntactic transformations. If so, then a powerful explanatory tool can be developed, namely that the degree of idiom semantic analyzability (on condition that it can be measured more or less objectively) would be able to predict or, at least, explain what kinds of discursive alterations are applicable to a given idiom. If not, then the notion of semantic analyzability itself appears to be of little value for linguistic research. Why should we develop a new category with its relevant criteria if it cannot be used for predicting or explaining empirical data? The answer to the question of the discursive relevance of analyzability depends on how we define this notion. In the literature on idiom research, the following three views of this concept can be found. First, it has been repeatedly stressed that idioms have an internal syntactic structure. In other words, idioms are not “long words” which have to be embedded into sentences as a whole but undergo the mandatory morphosyntactic adaptations, which make them fit in the sentence structure, i.e. “they never function as a lexical unit, i.e. they are not inserted as a unit at any one point of derivation” (Newmeyer 1974: 339). This is self-evident and does not help to correlate the degree of discursive freedom with the degree of analyzability. In this case, not the semantic, but the purely syntactic analyzability is at issue. Just like semantically decomposable idioms such as to throw the baby out with the bath (water), to keep tabs (on somebody) or to pull strings,3 most nondecomposable expressions undergo systematic (i.e. standard) morphosyntactic operations or inflectional variations, which are normal for their wordclass and necessary to meet the embedding conditions. Verbs, for example, can be marked by a particular person, tense, and the like. Compare: X will bite the dust / would bite the dust / has bitten the dust; he bites / they bite the dust, etc. Second, the analyzability of an idiom, and therefore the “semantic autonomy” of its constituents can be understood in the sense that they contribute semantically to its overall meaning (cf. Cacciari and Glucksberg 1991). In most cases, this is an incontrovertible fact, but has no relevance for the discursive behaviour of a given idiom, at least with regard to issues under discussion. The contribution of the particular constituents to the actual meaning manifests itself, among other things, in the fact that general semantico-syntactic properties of the verb-constituents (such as aspect or

46 Dmitrij Dobrovol’skij Aktionsart) are inherited by the overall figurative meaning of the idiom. So, kicking the bucket does not denote the situation of a slow death (Hamblin and Gibbs 1999: 25–26; Gibbs and Colston 2007: 826; cf. also Tronenko 2003 based on Russian data). Furthermore, the semantic role of particular constituents can be shown while comparing idioms from various languages that are semantically similar, but not identical. The German idiom Eulen nach Athen tragen lit. ‘to carry owls to Athens’ and the Russian idiom exat’ v Tulu so svoim samovarom lit. ‘to go to Tula with their own samovar’, meaning nearly the same as the English idiom to carry coals to Newcastle, display significant differences in their combinatorial properties. The Russian idiom can combine with words denoting human beings in the position of the direct object. That is, if one can take someone to a place where there are already many people of this kind, this idiom can be used as an ironic comment on this action. The German idiom cannot be used in this combinatorial surrounding. The reason for it seems to be that the verb tragen ‘to carry’, taken in its literal meaning, prevents this kind of direct object: normally one does not carry other people from place to place. The verb exat’ s ‘to go with’, allows for the interpretation in the sense ‘to take somebody somewhere with you’. This example demonstrates that semantic properties of particular constituents can (but, of course, need not) have impact on the combinatorial profile of the idiom in question. Another fact demonstrating the relevance of the content plane of single constituents for the actual meaning is the stylistic sensitivity of idioms towards their constituents. Idioms with vulgar words in their lexical structure are always perceived as vulgar, even if there is nothing vulgar in their lexicalized meaning. Idioms with old-fashioned, formal or poetic constituents are normally perceived as belonging to the corresponding stylistic levels. Such a view on analyzability is often referred to as the compositional approach, meaning that it emphasizes the fundamental ability of idiom constituents to contribute to the overall meaning. One version of this view is presented by Titone and Connine (1999) who propose a model of idiom representation and processing that ascribes non-compositional and compositional characteristics to idioms. In this view, idiomatic expressions function simultaneously as semantically arbitrary word sequences and compositional phrases. Similar to the aforementioned syntactic view on analyzability, the “compositional approach” does not help to answer the question whether or not the internal semantic structure of an idiom influences its discursive behaviour.

Structure of metaphor and idiom semantics

47

What would be much more helpful from the perspective of idiom syntax is a view of analyzability that would allow us to explain why certain idioms resist standard transformations and adaptations while others do not. Why can we say the beans have been spilt, but not *the bucket has been kicked? Intuitively it is obvious that speakers address semantic properties of idioms to understand that *the bucket has been kicked 4 and the like is senseless whereas utterances such as the beans have been spilt make sense, rather than learning lists of possible alterations for every idiom. The theory of idioms has to develop means for describing this kind of semantic intuition. In this case, the reason is that the word bucket from the idiom to kick the bucket does not allow for a concrete-referential use while the word beans from the idiom to spill the beans does. Thus, bucket is a meaningless constituent of the sentence structure, and beans is a meaningful one. The passivization in English in standard usage is communicatively reasonable only in those cases where the noun phrase that is moved into the subject position, and by so doing topicalized, is a meaningful constituent of the sentence structure.5 Therefore, the idiom parts that are moved from object to subject, or undergo topicalization, clefting, adnominal modification and the like must be meaningful parts of the idiom structure, at least to a certain extent. Speakers seem to have rather clear intuitions about this issue, that is to say, they feel in which cases idiom constituents are meaningful, i.e. semantically autonomous, and in which cases they are meaningless (not in the sense that they do not contribute to the overall meaning or that they are “non-words” morphosyntactically, but in the sense that they are not relatively autonomous semantic entities). To put it another way, idioms such as to kick the bucket, on the one hand, and to spill the beans, on the other, display significant differences not only in their discursive behaviour, but also in their internal semantic structures. It is reasonable to assume that the semantics is the trigger for their syntactic behaviour, and not vice versa. Thus, the notion of semantic analyzability that is based on the idea of semantic quasi-autonomy of particular idiom constituents seems to be a useful theoretical instrument for explaining not only the discursive behaviour of idioms, but also speakers’ background intuition. That is why such a view on analyzability is favoured by cognitive semantics. This view of analyzability is preferred by many linguists working in the field of idiom research regardless of their involvement in cognitively oriented research communities. Compare the typical argumentation in Nunberg, Sag, and Wasow (1994: 504):

48 Dmitrij Dobrovol’skij To say that an idiom is an idiomatically combining expression is to say that the conventional mapping from literal to idiomatic interpretation is homomorphic with respect to certain properties of the interpretations of the idiom’s components. In the case of an idiom like pull strings, this is quite easy to see: the literal situation-type involves a pulling activity and an affected object that is a set of strings. The idiomatic situation-type that this is mapped to involves a different activity, but one that preserves certain properties of pulling, and an affected object that participates in the idiomatic activity in a way that is similar in certain key respects to the way strings are pulled.

What remains unclear in this description is the idea of the similarity of situation-types. What is the difference between semantically motivated but non-analyzable idioms, such as to stick one’s neck out, and semantically analyzable idioms like to pull strings? Both display a certain degree of similarity between the situation-type described by the idiom’s lexical structure taken literally and the situation-type denoted by the actual meaning. For more detail about this issue see Section 4. The idioms that are considered to be analyzable in the sense of homomorphism between their lexical and semantic structure regularly participate in syntactic transformations such as passivization; compare the following examples found on the Internet. (1)

“So I hear you are moving your factory to China,” I said. Ted’s eyes bulged out in a way I had never seen before as he asked, “who told you that?” Laughing pretty hard, I choked out, “your boss.” Ted just shook his head, speechless. The big secret was out and the beans had been spilt by the main man himself. (http://www. telemarktips.com/TeleNews3.html)

(2)

I continued to write about my life ... Married and the father of two small children, I was never home, drunk a good deal of the time, and apparently felt it necessary to sleep with every waitress in North America and The British Isles. But guess what? All these beans have also been spilt in song. (http://www.paramuspicture show.com/Loudon.html)

(3)

Fonteyn’s private life, however, remained private, even when she published her autobiography. She belonged to an earlier age of discretion, as did her great friend and admirer, Leslie Edwards. No

Structure of metaphor and idiom semantics

49

beans spilt in his book. (http://www.guardian.co.uk/books/2003/ dec/21/bestbooksof2003) It is obvious that in all these examples the constituents spill and beans possess a certain semantic autonomy. In (1) the context explicitly relates the beans to the big secret, and had been spilt to was out. In (2) the constituent beans is modified by all these; the effect is that beans gets a concretereferential status. In (3) the word beans appears in the scope of negation, which is only possible with meaningful constituents. The semantic autonomy of idiom parts such as beans becomes especially evident in contexts with focussing, profiling function, i.e. in contexts where they are modified by relative clauses, possessive or demonstrative pronouns, and the like; cf. (4). (4)

They spilled the beans and gave up the gory details... As a listener, you felt you were getting something at its source, something simple, direct, and easy to identify with because, it turns out, their beans were not unlike your own. Everybody has pretty much the same gory details… (http://www.westport-news.com/news/article/ Wainwright-Live-in-Westport-252898.php)

It seems that for a constituent to be systematically modified in the discourse it is necessary that this constituent not only has a certain potential semantic autonomy, but also a clear referential status. Compare sentences (5) and (6) taken from Fellbaum (1993: 285). (5)

John spilled the beans (about his girlfriend).

(6)

John spilled the beans that his girlfriend once worked for the CIA.

Fellbaum points out that (5) is much more acceptable than (6). The reason is that the beans must refer to a secret whose existence and contents are known to the discourse participants. In (6) the contents of the secret are made known to the listeners after the idiom is already used, which makes this sentence sound odd. To put it differently, not only the knowledge about the potential semantic analyzability of this idiom is a part of its content plane, but also the knowledge that its noun-constituent is preferably used in concrete-referential function. The approach to the description of the systematic discursive behaviour of idioms as an epiphenomenon of their internal semantic structure is con-

50 Dmitrij Dobrovol’skij vincingly advocated in Langlotz (2003). “The systematic lexicogrammatical behaviour of idioms can be explained in terms of the speaker’s ability to manipulate an idiomatic construction to make it fulfil its cognitive modeling function in an efficient way” (Langlotz 2003: 463). Investigating the discursive behaviour of English idioms in relation to their analyzability, he points to some important restrictions. Above all, only systematic syntactical operations are subject to these generalizations. That is, all possible ad-hoc manipulations with idioms do not have to depend on their internal semantic structures (cf. in this regard also note 4 in the present paper). This is the first restriction that we have to bear in mind when looking for regularities in the field of idiom variability. The second restriction concerns the communicative structure of the context in question (cf. examples (5) and (6) discussed above). If there is no communicative need for a given transformation or adaptation it cannot be implemented, even if the idiom is potentially able to undergo this alteration. For example, in order to passivize an idiom we need not only semantic and structural prerequisites within the idiom itself, but also contextual conditions which make it communicatively necessary to remove the focus of attention from the Agent, and maybe to profile the Patient or Theme (see for more detail Dobrovol’skij 2001). The third restriction is usage-based by nature. Some idioms resist certain syntactic operations even if they are in principle possible. For example, although contexts such as (1) to (3) can easily be found, it is much more typical of the idiom to spill the beans to be used in active voice sentences. Compare Moon’s (1998: 109) comment on this issue: “There may indeed be semantic motivations here, but phraseological patterning also plays a part. Cases like spill the beans show a strong fossilization in an active structure, irrespective of potential passives and deep semantics”. Thus, not only the internal semantic structure of idioms with their underlying conceptual backgrounds is decisive for their discursive behaviour, but also usage routines and discourse preferences. From what has been said so far, the semantic analyzability of idioms is not a sufficient condition for their systematic discursive variation. Moreover, it is also not a necessary condition. There are syntactic operations which are not bound to the semantic analyzability. Compare the distinction between meaningful operations, i.e. operations on meaningful expressions (control and topicalization, clefting, adnominal modification and the like) and meaningless operations (such as raising or verb-second) made by Schenk (1992: 104). Schenk advocates the view that idioms, in principle,

Structure of metaphor and idiom semantics

51

are unable to undergo meaningful operations. This appears to be an overgeneralization that contradicts the empirical data, but still the distinction between these two kinds of operations remains relevant to idioms. Whereas for meaningful operations it is necessary that idioms be analyzable, meaningless operations can also be implemented with non-analyzable idioms. Especially if the idiom as a whole is involved in the syntactic operation, there are no constraints based on its internal semantic structure. This issue becomes even more complicated when other languages are involved in the analysis. The same syntactic operation can, in one language, be stronger semantically grounded and less in another. For instance, object fronting or verb movement have different semantic values in English and German. Compare the profound semantic differences between English topicalization and German scrambling. Whereas topicalization is defined as the emphasis placed on the topic of a sentence by preposing it to the beginning of the sentence (cf. Cigarettes, you couldn't pay me to smoke them), the term scrambling refers to word order variation of argument NPs with respect to each other and/or with respect to adverbial phrases, i.e. scrambling is a phenomenon typical of languages with free word order. To sum up, knowing that a given idiom is semantically analyzable or non-analyzable is not sufficient to predict its discursive behaviour. It can only be assumed that the internal semantic structure allows for a separation of idiom parts as more or less autonomous pieces and for an implementation of the syntactic operations which should be applicable to autonomous constituents of a given syntactic structure. It is impossible to make concrete predictions about individual operations on individual idioms because, on the one hand, there are other factors also influencing their discursive behaviour, and on the other, idiom variability in discourse is heterogeneous by nature. An additional problem can be attributed to differences in idiom processing by individual speakers. Since idiom variability depends on idiom representation, a given idiom cannot be unequivocally regarded as being equally variable for all speakers. This makes the implementation of whatever operational criteria in this field extremely difficult. What cognitive semantics can offer in this field is only a more or less reasonable heuristics for predicting the scope of variation for large groups of idioms and for ruling out certain alteration types. However, the most efficient contribution offered by cognitive semantics to phraseological theory in the field of idiom discursive behaviour lies in its explanatory rather than its predictive potential.

52 Dmitrij Dobrovol’skij This explanatory potential can be realized only if the very notion of semantic analyzability is understood as a conceptual property of a given idiom based on homomorphism between its lexical and semantic structure (and not in the sense of its syntactic transparency nor in the sense that single constituents usually contribute semantically to its overall meaning). The interpretation of the semantic analyzability of idioms which goes back to the idea of relevant homomorphism is favoured here. The implementation of this view on analyzability presupposes addressing the metaphoric foundation of the idiom’s figurative meaning because the structure of the underlying metaphor is responsible for the conceptual relationship between the lexically fixed word-chain and its actual meaning, i.e. between what is said and what is meant by a given idiom.

4. Semantic analyzability, metaphor structure and motivation To be able to efficiently implement the notion of semantic analyzability we need to find criteria to discriminate between analyzable and non-analyzable idioms, at least approximately, bearing in mind that operational criteria stricto sensu are difficult to establish. It is obvious that certain lexical and syntactic alterations, such as the modification of an idiom constituent by a possessive or demonstrative pronoun or its topicalization, clearly indicate its semantic autonomy. Such instruments can be used for quasi-operational purposes, but implementing such criteria would be essentially circular. If we derive the systematic discursive behaviour of idioms from their internal semantic structures, then the relevant variations cannot be seen to be the source of semantic analyzability (Dobrovol’skij 2004). Hence, semantic analyzability as a conceptually grounded entity must be separated from syntactic properties. In the early literature on semantic analyzability attempts can be found to bind the internal semantic structure to the structure of the definition of the idiom meaning. The problem is that the same idiom can be defined in very different ways. This means that a homomorphous definition, such as to spill the beans § ‘to divulge the information’ where to spill clearly correlates with ‘to divulge’, and the beans with ‘the information’, has to be regarded as an artefact of linguistic description having no ontological status. The correct way to deal with this phenomenon is to address the underlying conceptual structure, i.e. the mental image standing behind the lexicalized meaning. If the structure of the underlying metaphor homomorphically

Structure of metaphor and idiom semantics

53

correlates with the structure of the situation-type fixed in the lexicalized meaning, the idiom in question is semantically analyzable. This theory on the nature of semantic analyzability was proposed in Dobrovol’skij (2000, 2004). Compare a similar view in Geeraerts (1995) and Geeraerts and Bakema (1993), where semantic analyzability is described as the result of a conceptual projection, or Langlotz (2003, 2006). Consider the idiom to let the cat out of the bag. The underlying metaphor seems to be transparent to most speakers. The motivation is not only based on the overall mapping of the source frame on the target frame, reflected by the actual meaning, which can roughly be defined as ‘to disclose a secret’. There are also obvious associations between the part of the metaphor expressed by the noun cat and the part of the actual meaning ‘secret’. Until the moment highlighted in the lexical structure the cat was kept in a bag. This knowledge structure from the source frame corresponds to the knowledge structure of the target frame:6 ‘the secret was kept’. Then the cat (usually unintentionally) was let out of the bag, i.e. ‘the secret (usually unintentionally) was disclosed’. Thus, the structure of the underlying metaphor is homomorphic (partly isomorphic) to the structure of the actual meaning. The correspondences between these two conceptual structures explain why the noun cat is semantically autonomous while the noun bag is not. The quasi-autonomous status of the cat ‘a piece of information hidden from a hitherto unwitting third party’ can be demonstrated by the following contextual examples found on the Internet. (7)

This French cat’s out of the bag. There’s something irresistible about being in on a secret ... The same holds true for film music. It’s immensely satisfying to savor a great score before most people know about its talented composer. But such possessiveness is sure to be short-lived, for nothing can keep a lid on real talent for long. Certainly that’s the case with Alexandre Desplat, whose sublime scores for “Girl With a Pearl Earring” (2003) and “Birth” (2004) attracted widespread attention among aficionados. (http://www. variety.com/article/VR1117933727.html?categoryid=1985&cs=1)

(8)

The Israeli cat is out of the bag. One of Ariel Sharon’s closest advisers, in fact, his former chief of staff, Dov Weisglass, has revealed to the Israeli newspaper Haaretz the ulterior motives behind Sharon’s unilateral decision to withdraw from the Gaza Strip. (http://www.commondreams.org/views04/1008-28.htm)

54 Dmitrij Dobrovol’skij (9)

Letting the Radical Islamist Cat Out of the Bag. May 6, 2005 – A statement posted by the purported deputy of al Qaeda in Iraq, Abu Abdel-Rahman al-Iraqi, very possibly may have completely redefined the War on Terror. It was such a slip of the tongue that one has to wonder whether or not Abu will be allowed to keep his after Abu Musab al Zarqawi gets his hands on him. (http://www.gopusa. com/commentary/fsalvato/2005/fs_05061.shtml)

In contexts (7) to (9) the constituent cat takes modifiers which show its concrete-referential status. Moreover, these modifiers interact semantically not only with the part of the lexicalized meaning expressed by the word cat, but also with the corresponding part of the underlying metaphor. The role of the metaphor structure becomes especially obvious in contexts such as (10) where the writer elaborates on the image component of this idiom. (10)

Nikon Coolpix SQ: The cat takes another step out of the bag? Thus far, though, we’ve seen only speculation as to specifications from readers and in forums and newsgroups – nothing official has been released by Nikon, and none of the websites we’ve seen have posted anything beyond what was in the leaked Shockwave teaser files. If you’ll pardon the corruption of the cliche, the cat has poked its head out of the bag, but it hasn’t really come out yet. (http://www. imaging-resource.com/EVENTS/PMAS03/1044924660.html)

Although creative variations of this kind are not systematic by nature and cannot be used as evidence in favour of the semantic analyzability of idioms or in favour of the semantic autonomy of their parts (see Section 3), this example shows how the underlying metaphor works. The situation described by the metaphor (source frame) includes the action of ‘letting out’ with its three participants: (a) (b) (c)

Agent – the person who ‘lets the cat out of the bag’, Patient – ‘the cat’, and Container – ‘the bag’.

This structure allows for operations such as removing the Agent and promoting the Patient into the subject position. Then the cat, like in (10), starts working actively. The understanding of such contexts is guaranteed by people’s ability to relate the participants of the source-situation with the

Structure of metaphor and idiom semantics

55

participants of the situation denoted by the lexicalized meaning (target frame). It has to be stressed that the domain of relevant correspondences is conceptual by nature, i.e. what correlates is, strictly speaking, not the lexical structure (word chain) and the idiom’s actual meaning, but two conceptual structures: source frame and target frame. The participants of the target frame are: (i) (ii) (iii)

Agent – the person who ‘discloses the secret’, Theme – ‘a piece of information hidden from a hitherto unwitting third party’, and Experiencer – ‘the third party’.

While being mapped, the source and the target display partial correspondences. The Agent of the source corresponds to the Agent of the target, the Patient to the Theme, but participant-slot (c) of the source frame (Container) has no corresponding participant-slot in the target frame. Likewise participant-slot (iii) of the target frame (Experiencer) does not have a corresponding slot in the source frame. (a) ĺ (i) (b) ĺ (ii) (c) does not map onto (iii) This is the reason why the idiom to let the cat out of the bag has no Experiencer-valency in its argument structure, and why the bag is not semantically autonomous whereas the cat is. Strictly speaking, the idea of a secret corresponds to the concept ‘the cat in the bag’ rather than to the constituent the cat considered in isolation. Therefore, it would be wrong to claim that the word cat is fully autonomous. Here we are rather dealing with a partial correspondence of a secondary nature, and it is more correct to speak in this regard about homomorphism and not about isomorphism. To say that the structure of the underlying metaphor homomorphically correlates with the structure of the lexicalized meaning is to say that the participants of the source frame (partly) correspond to the participants of the target frame. In this case we are dealing with analyzable idioms. If the source frame is mapped on the target frame as a whole, and their lexically expressed participant-slots cannot be put into reasonable correspondences to each other, the idiom is perceived as being non-analyzable. However, the

56 Dmitrij Dobrovol’skij idiom may still be well motivated in this case. This leads us to the next issue, namely the relationship between analyzability and motivation. Some cognitively oriented studies equate the motivation of an idiom with the analyzability of its semantic structure (compare, e.g., Gibbs 1990). However, the motivation of an idiom does not necessarily result from its analyzability. Motivation presupposes conceptual links between source and target whereas analyzability addresses their structural properties. There are many idioms which are not semantically analyzable, and yet they are motivated, e.g. to rattle someone’s cage ‘to do something that annoys or frightens someone; to make someone angry, usually deliberately’. This idiom is not analyzable since its constituents rattle and cage are neither associated with meaningful parts of the underlying metaphor nor with the parts of the target frame. Nevertheless this idiom is motivated because the sense encoded in its lexical structure provides a clear basis for its actual meaning, ‘as if one were sitting in a cage like a bird and somebody else rattled the bird’s cage’. One of the consequences would be that the bird would become annoyed or frightened. Thus, non-analyzable idioms can be motivated. But can unmotivated, i.e. semantically opaque idioms also be analyzable? Different answers to this question can be found in linguistic studies (cf., among others, Geeraerts and Bakema 1993; Geeraerts 1995; Nunberg, Sag, and Wasow 1994; Keil 1997). The general impression is that the differences in this regard are due to the differences in the interpretation of either motivation or analyzability. Compare the view advocated in Nunberg, Sag, and Wasow (1994: 497), that analyzable idioms do not need to be motivated: “When we hear spill the beans used to mean ‘divulge the information’, for example, we can assume that spill denotes the relation of divulging and beans the information that is divulged, even if we cannot say why beans should have been used in this expression rather than succotash”. From this passage it is obvious that Nunberg, Sag, and Wasow (1994) regard motivation as a predicting property. However, if we define motivation as transparency of conceptual links between source and target, the idiom to spill the beans must be characterized as being motivated.7 Firstly, the motivating link is provided by the quasi-universal conceptual metaphors standing behind this idiom: IDEAS ARE PHYSICAL ENTITIES and THE MIND IS A CONTAINER. Secondly, the literal meaning of the verb to spill provides the idea of an unintentional action, the result of which is that entities lying inside a container are suddenly in plain view.

Structure of metaphor and idiom semantics

57

If the notions of motivation and semantic analyzability are defined along the lines advocated in this paper the most convincing way to deal with the relationships between these phenomena would be to consider analyzability to be a particular case of motivation. This results from the idea that the existence of relevant conceptual links (motivation) is a necessary prerequisite for comparing the structuring of both conceptual levels of the content plane of idioms. Much more important than to clarify the relations between analyzability and motivation is to investigate how they interact in idiom processing. Compare Langlotz’s (2006: 105–106) concept of “idiomatic activation set” including both motivation and homomorphism (“isomorphism” in his terminology). For instance, the conceptual structures relevant for the idiom to miss the boat/bus ‘to fail to take advantage of an opportunity’ and operations on them are represented as follows in Langlotz (2003: 41): 1. 2. 3.

MOVEMENT → ACTIVITY JOURNEY → PURPOSEFUL ACTIVITY VEHICLE → SUPPORTING MEANS FOR PROGRESS/DEVELOPMENT BUS/BOAT → SUPPORTING MEANS FOR PROGRESS/DEVELOPMENT MISS (THE BOAT/BUS) → NOT MAKE USE OF A MEANS FOR PROGRESS/DEVELOPMENT = MISS A CHANCE

Thus, knowledge structures relevant for understanding this idiom and activated for its processing concern both its holistic semantic motivation (cf., for example, the mapping from MOVEMENT to ACTIVITY) and its semantic analyzability (cf. the words bus and boat in their quasi-autonomous reading ‘supporting means for progress/development’). To sum up, semantic analyzability as well as motivation as a whole are cognitively-based phenomena, which, however, have no real predicting power, especially on the level of actual linguistic expressions. The speakers have to know the overall figurative meaning to be able to (re)construct the links between the conceptual structures involved in the content plane of a given idiom (cf. Keysar and Bly 1999). The way in which they do this varies from speaker to speaker. Also, the degree of semantic analyzability is, to a large extent, an individual phenomenon, i.e. the same as motivation, it is based on interpretive strategies. The consequence is that here we are dealing with tendencies rather than with rules. Maybe the only case where we are dealing with semantic autonomy stricto sensu is with idioms containing words denoting particular cultural

58 Dmitrij Dobrovol’skij symbols, as they are interpreted in Dobrovol’skij and Piirainen (1997, 2005) within Conventional Figurative Language Theory. Such words are semantically autonomous not only because of the structuring of relevant conceptual domains, but also because of their semantics outside a given idiom. Compare gold in an idiom such as to have a heart of gold ‘to be a friendly, generous, forgiving person whose qualities are much appreciated’. The constituent gold (i.e. the concept GOLD) can be singled out, and an autonomous meaning – ‘something extremely good and valuable’ – can be attributed to it. This idiom cannot be understood as a metaphor because it is not possible to imagine that someone’s heart is really made of gold. What is meant is that the speaker or hearer is able to activate a special component of his or her cultural knowledge, namely the knowledge of certain nonliteral functions of the concept GOLD. The symbolic meaning ‘something very positive, valuable’ is fixed in both language and culture. In this case we can speak of semantic autonomy in the strict sense of the term. In most cases discussed above we are instead dealing with an interpretive strategy which maps the structure of the underlying metaphor onto the structure of the idiom’s lexicalized meaning. Notions such as semantic analyzability belong to the metalevel of linguistic description, and cannot be used to predict, for instance, the discursive behaviour of every particular idiom or to “technically” describe its variation, but the notion does explain how the underlying cognitive mechanisms work. Though not predictable in detail, the behaviour of idioms in discourse is not absolutely arbitrary. On the contrary, it reveals a certain cognitively-based dependence on relevant conceptual and semantic properties, and above all, on the structure of the underlying metaphor.

Notes 1. 2.

This semantic structure can also be called actual meaning. In what follows, I will use these terms as synonyms. This obvious restriction concerns not only the phenomenon of semantic analyzability of idioms, but also their semantic motivation and variation, thus such limitations must always be taken into account while investigating idioms by cognitive methods. Compare in this regard (Langlotz 2003: 464): “Since idiom transparency has been described as a cognitive phenomenon, speakers must be expected to vary in terms of how they understand, store, and process different idioms in their minds. A given idiom cannot be expected to be equally represented and motivated for all speakers… Thus, since idiom varia-

Structure of metaphor and idiom semantics

3.

4.

5.

6.

7.

59

tion seems to be dependent on idiom representation, different idioms cannot be unequivocally regarded as being equally variable for all speakers. To explore this problem more extensively, more sophisticated psycholinguistic experiments on idiom variation and representation are needed”. Nunberg, Sag, and Wasow (1994: 525) also point out: “Individuals will naturally differ regarding how readily they perceive the metaphorical basis of an idiom, resulting in variability of judgments”. Cf. examples from Nunberg, Sag, and Wasow (1994: 502) which convincingly demonstrate the analyzability of the idioms to keep tabs (on somebody) or to pull strings (for somebody): Although the FBI kept tabs on Jane Fonda, the CIA kept them on Vanessa Redgrave; Those strings, he wouldn’t pull for you. I would like to stress again that this restriction concerns only the conventional use of language. In certain situations, in which the standard usage norms are violated in order to achieve specific communicative goals, the passivization of the idiom to kick the bucket seems to be possible. Compare the following situation described in Gibbs and Colston (2007: 825): “Two friends, Maria and Sven who have not spoken in a few weeks are having a conversation. The last time they had spoken, Maria learned that Sven’s very old pet dog was in poor health. During the present conversation, Maria asks Sven about his dog and Sven replies, ‘The bucket was kicked’”. This rule has certain exceptions. One of them is the so-called pied-piping, i.e. a syntactic operation which moves a larger constituent if, semantically, only a subpart of this constituent is affected. For example, sentences such as I think my leg is being pulled are allowed because of this syntactic phenomenon. It is necessary to move the constituent my leg into the subject position of the clause if the speaker wants to topicalize the possessor (in this case, I standing behind my). In the case of metaphorical motivation, two frames (source and target) are involved. These concepts go back to the ideas of the Cognitive Theory of Metaphor, developed by Lakoff and his colleagues (cf., for example, Lakoff and Johnson 1980; Lakoff 1987, 1993); compare the metalinguistic apparatus of conceptual metaphor providing the mapping from source domains onto target domains. Compare Langlotz (2003: 463): “idiom transparency results from a speaker’s ability to (re)motivate the semantic structure of an idiom relative to wellestablished cognitive models and entrenched patterns of semantic extension”.

60 Dmitrij Dobrovol’skij References Abel, Beate 2003 Sprecherurteile zur Dekomponierbarkeit englischer Idiome: Entwicklung eines Modells der lexikalischen und konzeptuellen Repräsenation von Idiomen bei Muttersprachlern und Nichtmuttersprachlern. Tübingen: Niemeyer. Cacciari, Cristina, and Sam Glucksberg 1991 Understanding idiomatic expressions: The contribution of word meanings. In Understanding Word and Sentence, Greg B. Simpson (ed.), 217–240. Amsterdam/New York/Oxford/Tokyo: Elsevier. Croft, William, and D. Alan Cruse 2004 Cognitive Linguistics. Cambridge: Cambridge University Press. Dobrovol’skij, Dmitrij 1982 Zum Problem der phraseologisch gebundenen Bedeutung. Beiträge zur Erforschung der deutschen Sprache 2: 52–67. Dobrovol’skij, Dmitrij 1997 Idiome im mentalen Lexikon: Ziele und Methoden der kognitivbasierten Phraseologieforschung. Trier: Wissenschaftlicher Verlag Trier. Dobrovol’skij, Dmitrij 2000 Ist die Semantik von Idiomen nichtkompositionell? In Sprachspiel und Bedeutung: Festschrift für Franz Hundsnurscher zum 65. Geburtstag, Susanne Beckmann, Peter-Paul König, and Georg Wolf (eds.), 113– 124. Tübingen: Niemeyer. Dobrovol’skij, Dmitrij 2001 Pragmatische Faktoren bei der syntaktischen Modifizierbarkeit von Idiomen. In: Pragmatische Syntax, Frank Liedtke, and Franz Hundsnurscher (eds.), 271–308. Tübingen: Niemeyer. Dobrovol’skij, Dmitrij 2004 Semantische Teilbarkeit der Idiomstruktur: zu operationalen Kriterien. In Europhras 2000: Internationale Tagung zur Phraseologie vom 15.–18. Juni 2000 in Aske/Schweden, Christine Palm-Meister (ed.), 61–68. Tübingen: Stauffenburg. Dobrovol’skij, Dmitrij, and Elisabeth Piirainen 1997 Symbole in Sprache und Kultur: Studien zur Phraseologie aus kultursemiotischer Perspektive. Bochum: Brockmeyer. 2005 Figurative Language: Cross-Cultural and Cross-Linguistic Perspectives. Amsterdam/Oxford: Elsevier. Fellbaum, Christiane 1993 The determiner in English idioms. In Idioms: Processing, Structure, and Interpretation, Cristina Cacciari, and Patrizia Tabossi (eds.), 271–295. Hillsdale, New Jersey: Lawrence Erlbaum.

Structure of metaphor and idiom semantics

61

Geeraerts, Dirk 1995 Specialisation and reinterpretation in idioms. In Idioms: Structural and Psychological Perspectives, Martin Everaert, Erik-Jan van der Linden, André Schenk, and Robert Schreuder (eds.), 57–73. Hillsdale, New Jersey: Lawrence Erlbaum. Geeraerts, Dirk, and Peter Bakema 1993 De prismatische semantiek van idiomen en composita [The prismatic semantics of idioms and compounds]. Leuvense Bijdragen 82: 185– 226. Gibbs, Raymond W. 1990 Psycholinguistic studies on the conceptual basis of idiomaticity. Cognitive Linguistics 1 (4): 417–451. Gibbs, Raymond W., and Herbert L. Colston 2007 Psycholinguistic aspects of phraseology: American tradition. In Phraseologie/Phraseology: An International Handbook of Contemporary Research. Vol. 2, Harald Burger, Dmitrij Dobrovol’skij, Peter Kühn, and Neal R. Norrick (eds.), 819–836. Berlin/New York: Walter de Gruyter. Gibbs, Raymond W., and Nandini P. Nayak 1989 Psycholinguistic studies on the syntactic behavior of idioms. Cognitive Psychology 21: 100–138. Gibbs, Raymond W., Nandini P. Nayak, and Copper Cutting 1989 How to kick the bucket and not decompose: Analyzability and idiom processing. Journal of Memory and Language 28: 576–593. Hamblin, Jennifer L., and Raymond W. Gibbs 1999 Why you can’t kick the bucket as you slowly die: Verbs in idiom comprehension. Journal of Psycholinguistic Research 28 (1): 25–39. Keysar, Boaz, and Bridget M. Bly 1999 Swimming against the current: Do idioms reflect conceptual structure? Journal of Pragmatics 31 (12): 1559–1578. Keil, Martina 1997 Wort für Wort: Repräsentation und Verarbeitung verbaler Phraseologismen (Phraseo-Lex). Tübingen: Niemeyer. Lakoff, George 1987 Women, Fire and Dangerous Things: What Categories Reveal About the Mind. Chicago/London: The University of Chicago Press. Lakoff, George 1993 The contemporary theory of metaphor. In Metaphor and Thought, 2nd ed. Andrew Ortony (ed.), 202–251. Cambridge: Cambridge University Press. Lakoff, George, and Mark Johnson 1980 Metaphors We Live By. Chicago/London: The University of Chicago Press.

62 Dmitrij Dobrovol’skij Langlotz, Andreas 2003 Idiomatic creativity: A cognitive linguistic model of idiom representation and variation for English idioms. Ph.D. diss., Department of English, University of Basel. Langlotz, Andreas 2006 Idiomatic Creativity. A Cognitive-linguistic Model of Idiomrepresentation and Idiom-variation in English. Amsterdam/Philadelphia: John Benjamins. Moon, Rosamund 1998 Fixed Expressions and Idioms in English: A Corpus-based Approach. Oxford: Oxford University Press. Newmeyer, Frederick J. 1974 The regularity of idiom behavior. Lingua 34: 327–342. Nunberg, Geoffrey, Ivan A. Sag, and Thomas Wasow 1994 Idioms. Language 70: 491–538. Rajxštejn, Aleksandr D. 1980 Sopostavitel’nyj analiz nemeckoj i russkoj frazeologii [Contrastive analysis of German and Russian phraseology]. Moscow: Vysšaja škola. Schenk, André 1992 The syntactic behaviour of idioms. In Proceedings of IDIOMS, Martin Everaert, Erik-Jan van der Linden, André Schenk, and Robert Schreuder (eds.), 97–110. Tilburg: Tilburg University. Talmy, Leonard 2000 Toward a Cognitive Semantics. 2 vols. Cambridge, Massachusetts: MIT Press. Titone, Debra A., and Cynthia M. Connine 1999 On the compositional and noncompositional nature of idiomatic expressions. Journal of Pragmatics 31: 1655–1674. Tronenko, Natalia 2003 Regularities in the Behaviour of Russian Phrasal Idioms. Bern et al.: Peter Lang. Wasow, Thomas, Ivan A. Sag, and Geoffrey Nunberg 1983 Idioms: An interim report. In Proceedings of the XIIIth International Congress of Linguists, Shirǀ Hattori, and Kazuko Inoue (eds.), 102– 115. Tokyo: CIPL.

Why focus on target domains? The importance of domain knowledge in children’s understanding of metaphors Aivars Glaznieks

1. Introduction There is a wealth of psychological literature dealing with the question of how children understand metaphors, or rather, metaphorical expressions.1 Some researchers seek to trace the moment when children begin to understand their first metaphorical expression (e.g. Augst 1978; Billow 1975); some the first production of a metaphorical expression (e.g. Gardner et al. 1975; Winner 1979; Winner, McCarthy, and Gardner 1980). Others focus on the structure of metaphors and on how differences in structure influence the comprehension of metaphorical expressions (e.g. Dent 1987; Gentner 1988); and a few on various other factors that affect the development of a so-called metaphorical competence, such as the syntactic construction of metaphorical expressions, the linguistic and non-linguistic contexts in which metaphorical expressions appear, or the general cognitive development that is necessary for the understanding of figurative speech (e.g. Cometa and Eson 1978; Johnson and Pascual-Leone 1989; Gernsbacher et al. 2001; Vosniadu 1987). One precondition for understanding metaphorical expressions was pointed out by Winner in 1988: to understand a metaphorical expression, she claims, a child must have the ability to perceive similarities between two ‘things’ in the world. For example, two objects can be similar in aspects like colour, shape, and function. These kinds of similarities can serve as the ground of a metaphorical mapping. Yet this competence is not specific to the understanding of metaphorical expressions. It is crucial for every kind of learning, and children are equipped with this ability from birth on (Winner 1988: 68–70). A metaphorical competence in this sense thus becomes apparent in children’s symbolic play. For example, when a child at the age of two puts her/his foot into a wastebasket and says “boot”, the child displays her/his competence in perceiving analogies. Winner calls

64 Aivars Glaznieks this kind of renaming of familiar physical objects symbolic play metaphor (Winner 1988: 94–97). Although children at the age of two or three may show some metaphorical competence in their own utterances,2 they are far from understanding metaphorical expressions in spoken language. From a psychological point of view, children under the age of four lack a crucial cognitive prerequisite: their ability to read the other’s mind is not yet fully developed, i.e. their ability to make assumptions about the assumptions of their partners in verbal and nonverbal interactions is restricted by their cognitive development. The so-called theory of mind is a competence that is fundamental to understanding utterances that are not ‘literally’ comprehensible, including all sorts of figurative expressions (e.g. metaphor, irony, hyperbole). Thus, children are unable to understand figurative speech unless they have developed a theory of mind. After children are able to make elaborate assumptions about the other’s intentions around the age of four, they should also begin to understand metaphorical expressions (cf. Happé 1995: 283).3 Once children have acquired the cognitive prerequisites for understanding metaphorical expressions at about the age of four, they still do not understand every metaphorical expression they hear. A further factor in understanding metaphorical expressions is the structure of knowledge children have acquired in the domains that are connected by a certain metaphorical expression. Keil (1986) shows that the comprehension of metaphorical expressions develops in a domain-specific way (“domain by domain”) in correlation with the child’s knowledge. Domains that are proximate to the child’s experience, like human nutrition, can easily serve as a source domain that can be mapped on a familiar target domain. Thus, very young children are able to understand metaphorical expressions such as the car was thirsty. On the other hand, unfamiliar domains block this competence and impede the understanding of metaphorical expressions. Thus, expressions like he was a smooth person are less likely to be understood by young children because their knowledge of domains like surface texture or personality traits may not be developed well enough to recognize the relation between a trait and surface texture. Additionally, Keil finds that children’s comprehension of metaphorical expressions “tends to emerge in a unified, all-or-none, manner” (Keil 1986: 91), i.e. children who understand one metaphorical expression (e.g. the car was thirsty) which connects certain domains (here an animate property is mapped on CAR) will also understand other metaphorical expressions (e.g. the car was sick, tired,

Why focus on target domains?

65

greedy, etc.) with the same domain features (again animate properties are mapped on CAR). Keil’s (1986) findings are not specific to children’s development of a metaphorical competence. Under certain circumstances, adults show similar behaviour in understanding metaphorical expressions. For example, a physicist might tend to create metaphorical expressions that use physical processes as a source domain. Another adult person may not be too familiar with physics, and so she/he does not understand what the physicist is trying to say employing his novel figurative expressions. In contrast, the physicist’s colleagues are very likely to understand these novel metaphorical expressions because they have explicit knowledge of the field of physics. Obviously, it is not undeveloped and rudimentary cognitive skills that prevent the adult from understanding these novel metaphorical expressions: she/he simply lacks the necessary specific knowledge about physics. According to these approaches in developmental psychology, understanding of metaphorical expressions in children older than four depends mainly on individual knowledge of the domains involved. Accepting this assumption, another important question remains: on knowledge of which domain does the understanding of metaphorical expressions depend? Does it depend on a child’s individual knowledge of the source domain, of the target domain, or of both the source and target domains? In light of a widespread theoretical assumption about the function of metaphors, there seems to be a clear answer to this question: since source domains serve to explain target domains, knowledge of the source domain will be crucial. From the perspective of Cognitive Linguistics, too, this answer might seem to be correct. Lakoff and Johnson state that “the essence of metaphor is understanding and experiencing one kind of thing in terms of another” (Lakoff and Johnson 1980: 5), and some studies have shown the influence of listeners’ tacit knowledge of metaphors in interpreting metaphorical expressions (e.g. Gibbs and O’Brien 1990).4 While these studies have not taken into account the development of a metaphorical competence during language acquisition,5 the main question of this paper is: how much impact, if at all, do source domains have on children’s understanding of metaphorical expressions? In the following, I will present some preliminary data from a study with five-year-old German-speaking preschoolers and pupils aged eight and ten. In two comprehension tasks subjects were tested with several metaphorical expressions denoting the emotions fear and anger in German. In Test 1, participants had to sort several metaphorical and non-metaphorical expres-

66 Aivars Glaznieks sions for fear and anger under the aspect of similarity of meaning. In Test 2, participants were confronted with several short stories in a multiplechoice task. All stories concluded with a question about the protagonist’s emotion. Participants had to choose between several possible metaphorical answers. The data show that comprehension of metaphorical expressions develops in a domain-specific way only with respect to target domains. By contrast, effects of domain specificity with respect to source domains cannot be confirmed.

2. How do children understand metaphorical expressions in the domains of FEAR and ANGER? 2.1. Source domains of FEAR and ANGER in German From a cognitive linguistic view, emotions like fear or anger are conceptualized metaphorically and metonymically.6 There are several metaphors that guide our perception of FEAR and ANGER. For example, FEAR in German is conceptualized via three7 metaphors (cf. Dobrovol’skij 1997: 185– 187): ʊ

FEAR IS COLD: eine Gänsehaut bekommen (to get goose bumps, ‘to get the jitters’), es läuft einem eiskalt den Rücken herunter (something icecold runs down one’s back, ‘something gives one the creeps’).

ʊ

FEAR IS DEFECATION: sich in die Hose machen (to make in one’s pants, ‘to be scared shitless’), die Hosen voll haben (to have one’s pants full (of excrements), ‘to be in a blue funk’).

ʊ

FEAR IS WEAKNESS: weiche Knie bekommen (to get soft knees, ‘to have one’s legs turn to jelly’), Blut und Wasser schwitzen (to sweat blood and water, ‘to be in a cold sweat’).

In the domain of ANGER, numerous metaphors can be observed. Lakoff and Kövecses (1983) first described a variety of them for American English. The following examples of source domains for the conceptualization of ANGER in German are adopted from common source domains from Lakoff and Kövecses’ investigation.8 Thus, ANGER in German can be conceptualized in the following ways (cf. Lakoff and Kövecses 1983: 9–19):

Why focus on target domains?

67

ʊ

ANGER IS HEAT OF A FLUID IN A CONTAINER:

jemandem kocht das Blut in den Adern (someone’s blood is boiling in her/his veins, ‘to boil with rage’), an die Decke gehen (to go up to the roof, ‘to hit the roof’).

ʊ

ANGER IS SICKNESS:

Schaum vor dem Mund bekommen (to get foam at the mouth, ‘to fret and fume’), Gift und Galle speien (to spit poison and gall, ‘to breathe fire and brimstone’).

ʊ

ANGER IS INSANITY:

wild werden (to become wild, ‘to be enraged’), auf 180 sein (to be on 180, ‘raging’).

The basic assumption of the domain-by-domain-account illustrated earlier in this paper is that domain knowledge has a crucial impact on the understanding of metaphorical expressions. According to this assumption, metaphor comprehension develops “domain by domain” during language acquisition. Understanding of metaphorical expressions develops earlier in domains closer to the child’s experience than in unfamiliar domains. Since no answer has been given to the question as to which domain has a stronger influence on the development of understanding metaphorical expressions, I will frame two hypotheses that follow from the basic assumption: ʊ

(H01): The development of the understanding of metaphorical expressions proceeds separately in different target domains.

ʊ

(H02): The development of the understanding of metaphorical expressions within one target domain proceeds along source domains.

Adapting the two hypotheses to the domains of FEAR and ANGER, they can be phrased in a more precise way: ʊ

(H1): The development of the understanding of metaphorical expressions in the target domains of FEAR and ANGER proceeds separately.

According to H1, the understanding of metaphorical expressions develops in different ways in the two target domains of FEAR and ANGER. Different results in understanding metaphorical expressions denoting FEAR and ANGER are expected.

68 Aivars Glaznieks ʊ

(H2): The development of understanding metaphorical expressions within the target domains of FEAR and ANGER proceeds along source domains.

H2 is a reformulation of the “all-or-none-effect” stated by Keil (1986: 91): if participants understand one metaphorical expression, they will also understand other metaphorical expressions with both the same target domain and the same source domain (i.e. they belong to the same metaphor).

2.2. Method 84 German-speaking children (N = 84) were tested with a set of metaphorical expressions that designate the emotions fear and anger in German. Participants consisted of three groups: 24 preschoolers around age five (M = 5;5, range 5;1–6;0), 30 second-graders around age eight (M = 8;1, range 7;7–8;11), and 30 fourth-graders around age ten (M = 10;7, range 9;10–11;9). All participants were monolingual native speakers of German from comparable socio-economic backgrounds. Groups of participants included equal numbers of boys and girls. They were tested during their attendance at kindergarten and elementary school, respectively. In addition, all participants were tested for non-verbal IQ (Raven-Matrices), general vocabulary (HAWIK-III/HAWIVA), and advanced theory of mind (in a German version adopted from Happé 1994).

2.3. Design The understanding of metaphorical expressions was examined using two different comprehension tasks. In Test 1, participants had to sort 50 metaphorical and non-metaphorical items from the domains of FEAR, ANGER, and JOY (serving as distractor). Participants were asked to group together all items that are similar in meaning. Unknown items were to be put aside. Afterwards, participants had to name the piles they had formed. Test 2 was a multiple-choice test. 14 short stories were presented. At the end of each story, the protagonist – a child named Petra – found herself in a situation that is evocative of a certain emotion (again: FEAR, ANGER, and JOY). One of the stories was as follows:

Why focus on target domains?

69

Petra soll für ihre Mutter etwas aus dem Keller holen. Sie geht allein die Treppe hinunter, als sie plötzlich ein unbekanntes Geräusch hört und das Licht ausgeht. ‘Petra was asked to get something from the basement. While she was going downstairs, she suddenly heard a strange noise and the lights went out.’

The stories concluded with the question: Wie geht es Petra? ‘How is Petra?’ Four possible answers were presented for each story: (a) a metaphorical answer denoting the intended emotion (e.g. Ihr läuft es eiskalt den Rücken herunter ‘something gives her the creeps’), (b) a metaphorical answer denoting a different emotion (distractor, e.g. Sie möchte ihr Herz ausschütten ‘she wants to bare her soul’), (c) a metaphorical answer denoting something other than an emotion (distractor, e.g. Sie hat viele Jahre auf dem Buckel ‘she has many years under her belt’), and (d) the option of rejecting all of the presented answers. The order of the answers (a) – (c) was randomized for each story. The metaphorical items used in Test 1 were the figurative idioms mentioned as examples in Section 2.1. The metaphorical answers in Test 2 contained these figurative idioms as well.9 The figurative idioms used in Test 1 and Test 2 are very conventional metaphorical expressions, however, they are figurative and decomposable instances of the before stated metaphors in the domains of FEAR and ANGER. Test 1 and Test 2 feature fundamental differences. A comparison of the results of the two tests has to take into account their basic diversity with respect to what is demanded from the participants. Test 1 requires sorting of more or less related lexical items that were presented without linguistic context; a predetermined thematic context was not given to the participants either. By contrast, to pass Test 2 the child had to choose from specified answers the one that fits best to a given narrative. Besides the different cognitive demands of sorting-tasks and choosing-tasks, the extent to which linguistic and thematic context was provided in each task might also be a considerable difference of the two tests.10

2.4. Interpretation The data were submitted to quantitative analysis. Participants’ performance was analyzed by a binary-coded evaluation process. In Test 1, every item was evaluated as correct when it was sorted onto the appropriate pile, representing, respectively, the domains of FEAR and ANGER; if not, the item

70 Aivars Glaznieks was rated as incorrect. Unknown items were also rated as incorrect. In Test 2, only metaphorical expressions denoting the intended emotion of fear or anger were evaluated as correct answers. All other options were rated as incorrect. A sum score was formed for each participant with regard to each test and each domain. This score was used in subsequent calculations.

2.5. Results Figure 1 shows the means for overall metaphorical understanding in both Test 1 and Test 2 for all three age groups. In Figure 2 means are split into the domains FEAR and ANGER. Figure 1 shows only small differences between the findings in Tests 1 and 2. This suggests that the differences between Test 1 and Test 2 such as the potentially different effect of linguistic context did neither impede nor facilitate the understanding of metaphorical expressions.

percentage of participants

1 0,9 0,8 0,7 0,6

Test 1

0,5

Test 2

0,4 0,3 0,2 0,1 0

5-year-olds

8-year-olds

10-year-olds

Figure 1. Correct understanding of metaphorical expressions: Test 1 vs. Test 2

A first result is that the understanding of metaphorical expressions in the domains of FEAR and ANGER improves with age. In Test 1, preschoolers understood 19% of the items tested, second-graders understood nearly 50%, and fourth-graders had an average rating of 78%. In Test 2, there was a similar finding: preschoolers understood 16% of the items presented in a linguistic context, second-graders understood 47%, and fourth-graders,

Why focus on target domains?

71

71%. The developmental pattern was analyzed using one-way ANOVAs. A main effect for age was found in Test 1 and Test 2 (cf. Table 1) in the domains of FEAR and ANGER, Test 1 – FEAR: F(2, 82) = 42.70, p < .01; Test 1 – ANGER: F(2, 82) = 38.73, p < .01; Test 2 – FEAR: F(2,77) = 46.67, p < .01; Test 2 – ANGER: F(2,77) = 31.79, p < .01. Table 1. Statistics of the one-way ANOVAs analyzing age effects, split up by test and domain Test 1

5-year-olds 8-year-olds 10-year-olds

Test 2

FEAR

ANGER

FEAR

ANGER

M 1.32 4.17 4.83

M 0.96 2.40 4.37

M 0.92 3.38 4.57

M 1.04 2.26 3.97

F(2, 82) = 42.70 F(2, 82) = 38.73 F(2,77) = 46.67 F(2,77) = 31.79 p < .01 p < .01 p < .01 p < .01

Post-hoc Scheffé contrasts revealed that mean differences between all groups were significant except the one between 8-year-olds and 10-yearolds in the domain of FEAR in Test 1 (cf. Table 2). Table 2. Post-hoc Scheffé contrasts analyzing age effects Test 1

Test 2

FEAR

ANGER

FEAR

ANGER

mean difference

mean difference

mean difference

mean difference

5-year-olds v 8-year-olds

-2,85*

-1,44*

-2,47*

-1,22*

8-year-olds v 10-year-olds

-,67

-1,97*

-1,18*

-1,71*

5-year-olds v 10-year-olds * p < .01

-3,51*

-3,41*

-3,65*

-2,92*

Figure 2 displays the participants’ performance in the domains of FEAR and ANGER separately. There are only small differences in preschoolers’ performance in both domains FEAR and ANGER. Paired-samples t-tests revealed no significant differences between FEAR and ANGER in either Test 1

72 Aivars Glaznieks

percentage of participants

(t(24) = 1.027, p = .315) or Test 2 (t(23) = -0.592, p = .56) for the fiveyear-olds. Second-graders, in contrast, showed significant differences in understanding metaphorical expressions depending on which domain they belong to. In Test 1 (t(29) = 5.707, p < .01) and Test 2 (t(24) = 5.041, p < .01), eight-year-old participants understood metaphorical expressions from the domain of FEAR significantly better than expressions from the domain of ANGER. This mean difference decreases in the ten-year-olds’ group. Although the trend stays the same, the differences are only marginally significant in Test 1 (t(29) = 1.882, p = .070) and Test 2 (t(28) = 2.01, p = .054). 1 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0

c a a

b

c

a d

c d b

b

d

5-year-olds

8-year-olds

10-year-olds

Figure 2. Correct understanding of metaphorical expressions separated between target domains in Test 1 and Test 2. The bars display the participants’ performance in the domain of a) ANGER (Test 1), b) ANGER (Test 2), c) FEAR (Test 1), and d) FEAR (Test 2).

So far, it has been shown that the development of the understanding of metaphorical expressions is domain-specific in the following respect: although the general ability to understand metaphorical expressions increases with age, eight-year-old participants understand significantly more metaphorical expressions from the domain of FEAR than from the domain of ANGER. In the results of the ten-year-olds this difference decreases slightly and is only approaching significance. Thus, development seems to be domain-specific with respect to target domains. H1 can be considered as veri-

Why focus on target domains?

73

fied: the development of the understanding of metaphorical expressions in the target domains of FEAR and ANGER proceeds separately. I will now analyze the data with specific consideration of source domains and their influence on the understanding of metaphorical expressions. Recall the hypothesis stated earlier in this paper: according to the domain-by-domain-account, the development of the understanding of metaphorical expressions within the target domains of FEAR and ANGER proceeds along source domains (H2). Figures 3 through 6 illustrate the understanding of metaphorical expressions. Expressions that belong to the same source domain are indicated by the same initial letter.

percentage of participants

1,00

0,80

0,60 ca

ca

0,40

aa

ab

cb ba

ca

0,20

ab cb

0,00

aa ab ba bb

5-year-olds

ba

bb

cb aa

bb

8-year-olds

10-year-olds

Figure 3. Correct understanding of metaphorical expressions within the target domain of ANGER (Test 1) based on the source domains of aa) HEAT (an die Decke gehen ‘to go up to the roof’), ab) HEAT (jemandem kocht das Blut in den Andern ‘one’s blood is boiling in her/his veins’), ba) SICKNESS (Gift und Galle speien ‘to spit poison and gall’), bb) SICKNESS (Schaum vor dem Mund bekommen ‘to get foam at the mouth’), ca) INSANITY (wild werden ‘to become wild’), cb) INSANITY (auf 180 sein ‘to be on 180’)

74 Aivars Glaznieks

percentage of participants

1,00

c

a

a

c d

e

d e

0,80 b

b f

0,60 0,40

c

aa d

a b 0,20 aaab

ba

ba bb

bb ca

ab

e

ba

aa

bb ca

ab

cb

f

f ca cb

cb

0,00 5-year-olds

8-year-olds

10-year-olds

Figure 4. Correct understanding of metaphorical expressions within the target domain of FEAR (Test 1) based on the source domains of aa) DEFECATION (sich in die Hose machen ‘to make in one’s pants’), ab) DEFECATION (die Hose voll haben ‘to have one’s pants full (of excrements)’), ba) COLD (eine Gänsehaut bekommen ‘to get goose bumps’), bb) COLD (einem läuft es einkalt den Rücken herunter ‘something ice-cold runs down one’s back’), ca) WEAKNESS (weiche Knie bekommen ‘to get soft knees’), and cb) WEAKNESS (Blut und Wasser schwitzen ‘to sweat blood and water’)

percentage of participants

1 0,8 0,6 ca ca

0,4

ba

aa ab

0,2

aa

ab ba bb ca cb

0 5-year-olds

aa

ab

cb

ba

cb bb

bb

8-year-olds

10-year-olds

Figure 5. Correct understanding of metaphorical expressions within the target domain of ANGER (Test 2) based on the source domains of aa) HEAT (an die Decke gehen ‘to go up to the roof’), ab) HEAT (jemandem kocht das

Why focus on target domains?

75

Blut in den Andern ‘one’s blood is boiling in her/his veins’), ba) SICK(Gift und Galle speien ‘to spit poison and gall’), bb) SICKNESS (Schaum vor dem Mund bekommen ‘to get foam at the mouth’), ca) INSANITY (wild werden ‘to become wild’), cb) INSANITY (auf 180 sein ‘to be on 180’) NESS

percentage of participants

1 0,8 0,6 ba

0,4

ba

aa

bb

bb

ca

ca 0,2 aa ab ba

bb

ca

cb

cb

cb

aa ab

ab

0 5-year-olds

8-year-olds

10-year-olds

Figure 6. Correct understanding metaphorical expressions within the target domain of FEAR (Test 2) based on the source domains of aa) DEFECATION (sich in die Hose machen ‘to make in one’s pants’), ab) DEFECATION (die Hose voll haben ‘to have one’s pants full (of excrements)’), ba) COLD (eine Gänsehaut bekommen ‘to get goose bumps’), bb) COLD (einem läuft es einkalt den Rücken herunter ‘something ice-cold runs down one’s back’), ca) WEAKNESS (weiche Knie bekommen ‘to get soft knees’), and cb) WEAKNESS (kaum noch atmen können ‘hardly be able to breath’)

According to H2, one would have expected these diagrams to paint a different picture. Differences in understanding between the tested metaphorical expressions should have followed the divisions between source domains. Instead, it seems that participants’ understanding is independent of source domains. H2 must therefore be rejected.

76 Aivars Glaznieks 2.6. Summary of results As a first result of this study, data show that the percentages of participants understanding metaphorical expressions in the domains of FEAR and ANGER increase with age. This finding is not surprising; nonetheless, it is worth noting that five-year-old participants understand about 20% of the metaphorical expressions, whereas eight-year-olds understand about 50%, and the oldest participants about 75% of the metaphorical expressions tested in this study. Secondly, there are no significant differences between the participants’ results in Tests 1 and 2. This suggests that the differences between the tasks such as the diverse cognitive demands and the extent to which linguistic context was provided do not impede or facilitate the understanding of metaphorical expressions. This finding may at first seem surprising. I will discuss this point later (see Section 2.7). Finally, the main result of this paper regards the hypotheses stated earlier: the development of the understanding of metaphorical expressions proceeds separately in the domains of FEAR and ANGER. H1 is verified by the data presented. While five-year-olds start at a very low level of understanding in both target domains FEAR and ANGER, eight-year-olds score differently in different target domains: their understanding in the domain of FEAR is significantly better than in the domain of ANGER. This difference persists in the ten-year-olds, although it decreases slightly and is not significant. In contrast, the development of the understanding of metaphorical expressions does not proceed separately with respect to several source domains within the domains of FEAR and ANGER, respectively. The data do not confirm H2. Development does not proceed domain-by-domain either in the group of preschoolers nor in the groups of second-graders and fourthgraders; instead, it proceeds independently of source domain boundaries.

2.7. Discussion From a cognitive linguistic point of view, all metaphorical expressions tested in the present study are instances of common metaphors in the domains of FEAR and ANGER. They thus refer to the way people perceive the domains of the emotions. Speakers of German comprehend and conceptualize FEAR as COLDNESS, WEAKNESS, and DEFECATION. In contrast, ANGER is conceptualized as HEAT OF A FLUID IN A CONTAINER, SICKNESS, and INSAN-

Why focus on target domains?

77

(in both cases, among other conceptualizations). In accordance with this assumption, knowledge about source domains was expected to influence the ability to understand metaphorical expressions because “the essence of metaphor is understanding and experiencing one kind of thing in terms of another” (Lakoff and Johnson 1980: 5). The results suggest that participants’ understanding of metaphorical expressions is not necessarily constrained by their knowledge of source domains, since their understanding is not bound to them. This means that children do not understand metaphorical expressions by projecting from the source domain, on the basis of pre-existing metaphorical conceptual structures. The findings suggest a different explanation: all metaphorical expressions used in this study were figurative idioms. Many studies investigated the way idioms are understood. Gibbs and O’Brien (1990) for example found that adults draw on their tacit knowledge about metaphors when they are asked to explain and interpret idioms. In addition, Gibbs and his colleagues have shown that “conceptual metaphors can under some circumstances be quickly accessed during immediate idiom comprehension” (Gibbs et al. 1997: 149). Even though metaphors might not always be activated and not automatically accessed whenever people are confronted with idioms in discourse, metaphors “clearly help people make sense of why idioms mean what they do” (Gibbs et al. 1997: 149). In contrast, children normally do not analyze idioms in this way; nonetheless, they usually do not have problems with many figurative idioms in natural verbal interaction, because “they treat idioms as if they were simply instances of ordinary language” (Glucksberg 2001: 86). According to Häcki Buhofer (1997: 219), idioms in natural verbal interactions are not understood “bottom-up”, i.e., the form is not processed by disassembling the utterance into its parts, and meaning is not constructed by taking into consideration all elements of the idiomatic expression. Instead, idioms are understood “top-down”, i.e., as one part of the whole linguistic and non-linguistic context in which they appear. In this respect, idioms are comparable to words; they are learned as fixed phrases during early childhood. Moreover, young children are not aware of the double semantic meaning of an idiom. When the latter is presented in context, children do not rely on the meaning of the idiom’s elements; however, the literal meaning of the idiom’s elements can influence understanding (Häcki Buhofer 1997: 223–226). The picture presented by the data as displayed in Figures 3 through 6 could not have appeared under the assumption that specific cognitive capa-

ITY

78 Aivars Glaznieks bilities evolve during certain phases of children’s development; the understanding of different metaphorical expressions, at least those from identical source domains, should under this assumption evolve in lock-step. For example, only 35% of the ten-year-olds understand the item die Hosen voll haben, whereas nearly 90% understand the item sich in die Hose machen (cf. Figure 5). High variation in understanding different metaphorical expressions within one age group (cf. Figure 3–6) supports Häcki Buhofer’s position. This variation in understanding can be interpreted as a result of the typical way idioms are acquired (cf. Burger, Buhofer, and Sialm 1982). The irregular distribution of children’s understanding of metaphorical expressions within both target domains and source domains, then, may be better explained by variations in the frequency with which certain metaphorical expressions appear in the language surrounding children.11 Metaphorical expressions that are frequently used by parents, siblings, teachers, educators, and peers are understood, whereas unfamiliar expressions are not understood. Due to the design of this study, participants were not able to learn unfamiliar metaphorical expressions during the comprehension tasks. Even Test 2 may not give enough contextual information for an understanding of the tested expressions unless participants have learned them before, although metaphorical expressions were presented in a thematic and linguistic context. This might explain why almost no differences are observed between Tests 1 (no linguistic context) and 2 (linguistic context; cf. Figures 1 and 2).

3. Why focus on target domains? The following conclusions must be drawn from the data of this study. Although figurative, metaphorical expressions in the domains of FEAR and ANGER are not understood by children in the way theoretical assumptions from developmental psychology and Cognitive Linguistics predict. The main reason behind this finding is probably that the metaphorical expressions used in this study were figurative idioms. Studies in developmental psychology predominantly use novel metaphors in their research of metaphor understanding. It is nonetheless surprising that source domains do not seem to have an impact on children’s understanding of metaphorical expressions in this case even though the mappings are very common in the German language, and even the youngest children in this study have obvi-

Why focus on target domains?

79

ously already conceptualized fear and anger. That leads to the question of how children at preschool and elementary-school ages conceptualize emotions. Do children conceptualize emotions metaphorically at all, or is a metaphorically structured concept the result of a constant use of metaphorical speech and thus to be expected later in the child’s development? Janke (2002) conducted several studies with German-speaking preschool and elementary-school children testing their knowledge about emotions. Although children at this age show astonishing knowledge, Janke shows that seven-year-olds, for example, do not display adult-like knowledge about body sensations accompanying the emotions fear, anger, joy, and sadness. In contrast, ten-year-olds do not differ from adults in respect to their knowledge. This finding suggests that children before the age of ten lack relevant knowledge about emotions they may need to understand metaphorical expressions in a proper way. The construction of metaphorical conceptual structures around emotion concepts may be an ongoing process at elementary-school age. Once this structure is set up, children may be able to understand unfamiliar metaphorical expressions with regard to the underlying metaphor. Unless these structures are established, reasoning does not play a role in children’s understanding of metaphorical expressions. Nonetheless, it is important to investigate what kinds of metaphorical expressions are part of the child’s vocabulary at this stage of development. In a usage-based model, metaphorical expressions frequently heard and used lead to metaphorical thinking. Tomasello describes the underlying general process as follows: Adults tell children with regularity, for example, to ‘toe the line’ or to ‘put that out of your mind’ or not to ‘lose patience’. Comprehending these figurative ways of talking takes children down the path of drawing analogies between the concrete domains they know from their sensory-motor experiences and the more abstract domains of adult interaction and social and mental life that they are in the process of learning about. After enough of certain kinds of metaphorical expressions, children should presumably be able to construct the kinds of broad and pervasive metaphorical understandings that lead to productivity – as in the famous ‘love is a journey’ metaphor of Lakoff and Johnson (1980). (Tomasello 1999: 168)

Once a “metaphorical understanding” is established, the child may use this understanding to grasp unfamiliar and novel metaphors, as pointed out e.g. by Gibbs and O’Brien (1990). This process calls for empirical investigation.

80 Aivars Glaznieks The data of this study show that there are differences in children’s familiarity with conventionalized metaphorical expressions in both domains FEAR and ANGER. This is particularly apparent in the eight-year-old participants. Thus, further investigations will have to examine how this difference has come into being. To begin with, if this difference is a direct result of a discrepancy between frequencies in the input language, then metaphorical expressions denoting fear should be more frequent in the adults’ language. Subsequent studies will therefore have to investigate the frequency of metaphorical expressions in the target domains. Secondly, if the difference appears due to participants’ unequal vocabulary development and varying knowledge about the domains of FEAR and ANGER, which leads in turn to a different understanding of metaphorical expressions, then investigations of semantic and conceptual knowledge in children will be fruitful. According to Janke (2002), preschool and school age children display a great deal of knowledge about emotions; nevertheless, knowledge structures are still developing, and it is worth examining this development with respect to children’s capacity to understand metaphorical expressions. The focus of further studies, then, must be placed on children’s knowledge of the target domains.

Notes 1.

2.

3.

Although this paper starts with theoretical assumptions from developmental psychology, I will adopt the terminological distinction between metaphor and metaphorical expression from the conceptual metaphor theory. To avoid any terminological confusion, metaphor will be used with the meaning “a crossdomain mapping in the conceptual system”, while metaphorical expression “refers to a linguistic expression … that is the surface realization of such a cross-domain mapping” (Lakoff [1993] 2003: 186). There is a crucial difference between symbolic play metaphors and metaphorical expressions usually produced by adults: they differ in the directionality of the mapping. In traditional theory, a metaphor maps features of a vehicle (source) onto a topic (target) to make a proposition regarding the topic (target). In contrast, a symbolic play metaphor focuses on the vehicle (source) (cf. Winner 1988: 98). Some researchers question whether symbolic play metaphors are metaphors at all (e.g. Burger 2006: 341–342; Gibbs 1994: 404). In several studies, Happé finds a strong correlation between the absence of a theory of mind and the inability to understand metaphorical expressions in autistic participants. Autistic subjects that passed the theory of mind test (false believe task) were able to understand metaphorical expressions but they were

Why focus on target domains?

81

not able to understand irony (Happé 1993, 1994, 1995). Irony, thus, requires a more elaborate comprehension of the speaker’s intention – a so-called “second order theory of mind” (Sodian and Thoermer 2005: 554); see Norbury (2005: 394–396) for a critique of Happé. 4. There was an extended debate over this issue during the 1990s and the earliest years of this century; see McGlone (2001) for a review of this debate and Glucksberg (2001: 81–89) for an opposite opinion; see also Croft and Cruse (2004: 195) for a critique of Glucksberg. 5. See Gibbs (1994: 399–433) for an explanation based on the assumption that the acquisition of metaphors depends on the development of early imageschematic structures. 6. In the following I will speak only of metaphors. I will not make a distinction between metaphors and metonymies because metonymies that are relevant in this paper are body-based metonymies like eine Gänsehaut bekommen (to get goose bumps, ‘something gives someone the jitters’). Dobrovol’skij (1997: 183) argues that although symptoms like getting goose bumps are clearly a part of the emotion of fear, there is no strong connection between the emotion fear and the experience of getting goose bumps; i.e., one can be afraid of something without necessarily getting goose bumps. Nevertheless, utterances like ich bekomme eine Gänsehaut are still possible. Hence, the figurative expression in such a situation is used metaphorically. 7. Dobrovol’skij (1997: 181–187) states four main models of figurativeness in the domain of FEAR in German. For methodological reasons, I have left aside the model FEAR IS A PREDATOR. 8. Unfortunately, there is no systematic investigation of anger metaphors in German. See Herrera Burstein (1997) and Weber (1993) for studies of Spanish and French anger metaphors. 9. The item kaum noch atment können (hardly be able to breathe ‘one’s breath caught’) replaced the item Blut und Wasser schwitzen (to sweat blood and water ‘to be in a cold sweat’) that was used in Test 1. Since both metaphorical expressions belong to the same metaphor (FEAR IS WEAKNESS), the difference should not have any impact on the results. 10. Because of the configuration of Test 1, in which similar metaphorical and non-metaphorical expressions were presented, a restricted thematic context also occurred. However, this configuration is dissimilar from the linguistic context given in Test 2 where utterance turns were imitated. 11. Unfortunately, no study has been made of the frequency of metaphorical expressions in either adult verbal interaction, child-directed speech or family contexts. Recently, some attempts have been made by Stefanowitsch within a usage-based model (Stefanowitsch 2006).

82 Aivars Glaznieks

References Augst, Gerhard 1978 Zur Ontogenese des Metaphernerwerbs – eine empirische Pilotstudie. In Spracherwerb von 6 bis 16: Linguistische, psychologische, soziologische Grundlagen, Gerhard Augst (ed.), 220–232. Düsseldorf: Pädagogischer Verlag Schwann. Billow, Richard M. 1975 A cognitive developmental study of metaphor comprehension. Developmental Psychology 11 (4): 415–423. Burger, Harald 2006 Phraseologie im Spracherwerb – Evidenz aus der Spontansprache. In Phraseology in Motion 1: Methode und Kritik. Akten der internationalen Tagung zur Phraseologie (Basel, 2004), Annelies Häcki Buhofer, and Harald Burger (eds.), 339–356. Bultmannsweiler: Schneider Verlag Hohengehren. Burger, Harald, Annelies Buhofer, and Ambros Sialm 1982 Handbuch der Phraseologie. Berlin/New York: Walter de Gruyter. Cometa, Michael S., and Morris E. Eson 1978 Logical operations and metaphor interpretation: A Piagetian model. Child Development 49 (3): 649–659. Croft, William, and D. Alan Cruse 2004 Cognitive Linguistics. Cambridge: Cambridge University Press. Dent, Cathy H. 1987 Developmental studies of perception and metaphor: The twain shall meet. Metaphor and Symbolic Activity 2 (1): 53–71. Dobrovols’kij, Dimitrij 1997 Idiome im mentalen Lexikon: Ziele und Methoden der kognitivbasierten Phraseologieforschung. Trier: Wissenschaftlicher Verlag. Gardner, Howard, Mary Kirchner, Ellen Winner, and David Perkins 1975 Children’s metaphoric productions and preferences. Journal of Child Language 2: 125–141. Gentner, Dedre 1988 Metaphor as structure mapping: The relational shift. Child Development 59 (1): 47–59. Gernsbacher, Morton Ann, Boaz Keysar, Rachel R. W. Robertson, Necia K. Werner 2001 The role of suppression and enhancement in understanding metaphors. Journal of Memory and Language 45 (3): 433–450. Gibbs, Raymond W. 1994 The Poetics of Mind: Figurative Thought, Language, and Understanding. Cambridge: Cambridge University Press.

Why focus on target domains?

83

Gibbs, Raymond W., Josephine M. Bogdanovich, Jeffrey R. Sykes, and Daniel J. Barr 1997 Metaphor in idiom comprehension. Journal of Memory and Language 37: 141–154. Gibbs, Raymond W., and Jennifer E. O’Brien 1990 Idioms and mental imagery: The metaphorical motivation for idiomatic meaning. Cognition 36: 35–68. Glucksberg, Samuel 2001 Understanding Figurative Language: From Metaphors to Idioms. Oxford: Oxford Unversity Press. Häcki Buhofer, Annelies 1997 Phraseologismen im Spracherwerb. In Wortbildung und Phraseologie, Rainer Wimmer, and Franz-Josef Berens (eds.), 209–232. Tübingen: Gunter Narr Verlag. Happé, Francesca G. E. 1993 Communicative competence and theory of mind in autism: A test of relevance theory. Cognition 48 (2): 101–119. Happé, Francesca G. E. 1994 An advanced test of theory of mind: Understanding of story characters’ thoughts and feelings by able autistic, mentally handicapped, and normal children and adults. Journal of Autism and Developmental Disorders 24 (2): 129–154. Happé, Francesca G. E. 1995 Understanding minds and metaphors: Insights from the study of figurative language in autism. Metaphor and Symbolic Activity 10 (4): 275–295. Herrera Burstein, Marcos 1997 Sprachliches Erfassen von Emotionen im Spanischen. Dissertation, University of Freiburg. Janke, Bettina 2002 Entwicklung des Emotionswissens bei Kindern. Göttingen: Hogrefe. Johnson, Janice, and Juan Pascual-Leone 1989 Developmental levels of processing in metaphor interpretation. Journal of Experimental Child Psychology 48 (1): 1–31. Keil, Frank C. 1986 Conceptual domains and the acquisition of metaphor. Cognitive Development 1 (1): 73–96. Lakoff, George 2003 Reprint. The contemporary theory of metaphor. In Cognitive Linguistics: Basic Readings, Dirk Geeraerts (ed.), 185–238. Berlin/New York: Mouton de Gruyter. Originally published in Metaphor and Thought, 2nd ed., Andrew Ortony (ed.), 202–251. Cambridge: Cambridge University Press, 1993.

84 Aivars Glaznieks Lakoff, George, and Mark Johnson 1980 Metaphors We Live By. Chicago: University of Chicago Press. Lakoff, George, and Zoltán Kövecses 1983 The Cognitive Model of Anger Inherent in American English. Trier: L.A.U.T. (Linguistic Agency University of Trier). McGlone, Matthew S. 2001 Concepts as metaphors. In Understanding Figurative Language: From Metaphors to Idioms, Sam Glucksberg, 90–107. Oxford: Oxford University Press. Norbury, Courtenay Frazier 2005 The relationship between theory of mind and metaphor. Evidence from children with language impairment and autistic spectrum disorder. British Journal of Developmental Psychology 23: 383–399. Sodian, Beate, and Claudia Thoermer 2005 Theory of mind. In Enzyklopädie für Psychologie, Serie V: Entwicklung. Vol. 2, Wolfgang Schneider, and Beate Sodian (eds.), 495–608. Göttingen: Hogrefe. Stefanowitsch, Anatol 2006 Words and their metaphors: A corpus-based approach. In Corpusbased Approaches to Metaphor and Metonymy, Anatol Sefanowitsch, and Stefan Th. Gries (eds.), 61–105. Berlin/New York: Mouton de Gruyter. Tomasello, Michael 1999 The Cultural Origins of Human Cognition. Cambridge/London: Harvard University Press. Vosniadu, Stella 1987 Contextual and linguistic factors in children’s comprehension of nonliteral language. Metaphor and Symbolic Activity 2 (1): 1–11. Weber, Frauke 1993 Denken in Metaphern. Kognitive Semantik und französische Gefühlsmetaphorik. Frankfurt am Main: Peter Lang Verlag. Winner, Ellen 1979 New names for old things: The emergence of metaphoric language. Journal of Child Language 6: 469–491. Winner, Ellen 1988 The Points of Words: Children’s Understanding of Metaphor and Irony. Cambridge: Harvard University Press. Winner, Ellen, Margaret McCarthy, and Howard Gardner 1980 The ontogenesis of metaphor. In Cognition and Figurative Language, Richard P. Honeck, and Robert R. Hoffman (eds.), 341–361. Hillsdale: Erlbaum.

Salience and the conventionality of metonymies Sandra Handl

1. Introduction For good reasons, cognitive linguists’ descriptions of metaphor and metonymy typically focus on examples which are inconspicuous with regard to their figurative nature. Linguistic instantiations such as Your theory needs more support or The kettle is boiling are well equipped to illustrate the fundamental role of metaphor and metonymy for human cognition because they demonstrate that figurative thinking can be so entrenched that speakers are no longer aware of it. Thus, it is the so-called conventional linguistic metaphors and metonymies which grant us access to those patterns of figurative thinking which are automatic and therefore govern our cognition to a considerable extent. While it is largely undisputed1 that conventional metaphors and metonymies have this special status, the question as to what the label conventional actually refers to has, astonishingly, not attracted too much attention. This is especially true for metonymy, where the relevant literature often remains silent with respect to any criteria which could be used to determine the degree of conventionality of the examples discussed. The situation is slightly different for metaphor. Here, even some of the classic sources such as Lakoff and Johnson (1980) or Lakoff and Turner (1989) offer valuable ideas on conventionality. In what follows, I will first suggest a working definition of conventionality and briefly summarize the criteria of conventionality which have been proposed for metaphors. In a second step, I will show that only one of them can be applied to metonymy, whereas the others work with metaphor only. It will be argued that this lack of defining criteria can be compensated by taking the notion of salience into account. This may not sound new to those familiar with the field, since most of the few researchers interested in the subject have stressed the interrelation between conventionality and ontological salience. However, as will be demonstrated with the help of the results of a corpus-based study, a consideration of ontological salience is not enough for a full account of the conventionality of metonymy. It has to be complemented by an analysis of the second facet of salience: cognitive

86 Sandra Handl salience or – more accurately – the cognitive salience of the attributes involved in the metonymical construal at hand.

2. Conventionality and figurative language 2.1. The conventionality of metaphors Basically, a given form-meaning pairing can be considered conventional if it represents a generally established way of expression that is shared by the members of a speech community. As an intersubjective phenomenon, conventionality has to be distinguished from the related, but not identical notion of entrenchment (cf. Langacker 1987: 59). The latter term describes how firmly linguistic structures are rooted in the minds of individual speakers, so that the degree of entrenchment determines how effortlessly these structures can be processed and understood. Whereas entrenchment is a precondition for conventionality, i.e. conventional linguistic structures can be expected to be entrenched in the minds of most, if not all, speakers of a given speech community, entrenchment does not necessarily cause conventionality. Conventionality is inseparably linked to the speech community, a fact already evident in Saussure’s famous description of the linguistic sign, where he states that form-meaning pairings which have been incorporated into the speech community’s collective mind can hardly be changed voluntarily ([1916] 1983: 71). Once they are firmly established, speakers are bound to them; they govern their linguistic behaviour (1983: 71–73). Somewhat surprisingly, Saussure’s claims tie in well with later, cognitive accounts of the role of conventionality. Here, too, it is said to determine human behaviour, but the levels on which it plays a role are widened to include cognition and action as well as language. It is thus one of the main arguments of the conceptual theory of metaphor and metonymy that conventional patterns of figurative thinking shape not only the way we speak, but also how we conceptualize and experience the world and act. From a cognitive angle, one therefore has to distinguish between conventionality on the linguistic and on the conceptual level. For metaphor, Lakoff and Turner (1989: 55) propose that conventionalization on the conceptual level can be measured by looking at how “automatic, effortless, and generally established” a conceptual mapping is. Three factors contribute to conventionality on this level: First, the mapping has to

Salience and the conventionality of metonymies

87

involve a set of systematic correspondences. This criterion separates isolated and unsystematic metaphors like A MOUNTAIN IS A PERSON from conventional ones like A THEORY IS A BUILDING (Lakoff and Turner 1989: 55). Second, conventional conceptual metaphors follow the principle of unidirectionality and map more concrete domains onto more abstract ones. Reversing the source and the target domain is usually excluded (Kövecses 2002: 16). The third criterion links the conceptual to the linguistic level and deals with the number of linguistic instantiations motivated by a given conceptual metaphor. For instance, Lakoff and Johnson provide many examples of conventional conceptual metaphors such as A THEORY IS A BUILDING (1980: 46), but do not supply a single one illustrating the only unconventional mapping they discuss in detail, LOVE IS A COLLABORATIVE WORK OF ART (1980: 139–143). In accordance with the theory's focus on the conceptual nature of figurative language, conventionality on the linguistic level cannot be assessed without taking cognition into account: metaphorical expressions are conventional if they rely on one of the conventional conceptual mappings, i.e. conventionality on the conceptual level is, as it were, a prerequisite for conventionality on the linguistic level. However, not all expressions which are motivated by a conventional mapping are inevitably equally conventional. Some, for example, exploit the mappings in more creative ways and refer to parts that are not conventionally used (e.g. His theory has thousands of little rooms and long, winding corridors; Lakoff and Johnson 1980: 53). The difference between conventional and unconventional cases is thus determined by the creative use of existing patterns2 and the frequency of occurrence of the single linguistic instantiations. The factor of frequency is often quoted in psycholinguistics as well, a discipline which has also contributed a considerable share to the understanding of conventionality. Various studies have shown that subjects’ familiarity with a form-meaning pairing influences its ease of processing (e.g. Giora 1997, 2003; Blasko and Connine 1993). The second major factor which plays a role in psycholinguistic studies is aptness. The term aptness describes how good a metaphor is in getting across the intended figurative meaning and tries to relate this to the conceptual distance between the domains linked by a metaphorical mapping. Even though the term is by no means an undisputed one, most researchers such as Katz (1989), Blasko and Connine (1993), Chiappe, Kennedy, and Smykowski (2003), and many others hold that metaphors are ‘better’ or ‘apter’ if the domains involved are distant in semantic space and if the metaphors make use of features that

88 Sandra Handl are highly similar or shared by the two domains. Blasko and Connine (1993), for example, found that metaphorical expressions like a good professor is an oasis or a ritual is a prison are not only less familiar but also less apt than examples like hard work is a ladder or books are treasure chests, because it seems easier to find similar features of BOOKS and VALUABLE COMMODITIES than of a PROFESSOR and an OASIS. As this section has shown, the relevant cognitive-linguistic and psycholinguistic literature provides useful hints as to the criteria which render a metaphor conventional. The question if they can be applied to metonymy as well will be addressed next.

2.2. Metaphor vs. metonymy On the whole, one would expect the conventionality of metonymies to be governed by the same principles as the conventionality of metaphors: A conceptual metonymy should be considered conventional if it is shared by the members of a speech community. And a linguistic metonym is conventional if it is based on one of these well-established ways of thinking and does not apply the mapping in an innovative way. So far the general theory. But is it also possible to transfer any of the more specific criteria to metonymy? Kövecses and Radden (1998: 39) define metonymy as follows: a cognitive process in which one conceptual entity, the vehicle, provides mental access to another conceptual entity, the target, within the same domain, or ICM.

This entails some significant differences to metaphor: First, metonymy is a within-domain process, and second, it is not used to understand one domain in terms of another, but as a means of activation and reference. Conceptual metonymies do not involve whole sets of systematic correspondences between domains. While there is an interrelation between the number of correspondences and the linguistic productivity of a metaphorical mapping, this does not hold for metonymy. The only way in which metonymies can be said to be systematic is that for metonymy we find, just like for metaphor, a limited number of conceptual mappings (such as CONTAINER FOR CONTENT in The kettle is boiling or PRODUCER FOR PRODUCT in He’s reading Shakespeare, etc.) which are shared by the members of a speech community or a given culture – a fact which becomes manifest in their multiple linguistic instantiations.

Salience and the conventionality of metonymies

89

The criterion of unidirectionality is also difficult to apply to metonymy. When a metonymy such as The White House isn’t saying anything is used to refer to the INSTITUTION by way of mentioning the PLACE, the metonymical vehicle loses its conceptual prominence as soon as the addressee has accessed the target (cf. Langacker 1993: 6). In this sense, metonymies are unidirectional. On the other hand, as the work of Kövecses and Radden (1998) and others indicates metonymic mappings are, in contrast to metaphorical ones, often reversible. There are, for example, WHOLE FOR PART metonymies (e.g. America for ‘United States’) as well as PART FOR WHOLE metonymies (e.g. England for ‘Great Britain’), CONTAINER FOR CONTENT metonymies (e.g. kettle for ‘contents’) as well as CONTENT FOR CONTAINER metonymies (e.g. to uncork the wine for ‘bottle of wine’). So, unidirectionality cannot be used as a hard and fast criterion to differentiate between conventional and unconventional patterns, because one cannot establish a single overarching principle that covers all conventional cases and excludes all unconventional ones. On the linguistic level, one is dealing with a scale of conventionality that has unconventional examples at one of its poles and highly conventional ones at the other. The unconventional ones need more contextual support for their interpretation and can be assigned to two groups: those whose unconventionality is a direct result of the limited applicability of a conceptual mapping (e.g. ORDER FOR CUSTOMER in The ham sandwich wants his check), and those which are formed on the basis of a quite common pattern of metonymic reasoning, but whose conventionalization potential is severely limited, since they build on concepts that are only shared by a very restricted number of speakers (e.g. PRODUCER FOR PRODUCT in *I’ve just bought a Jane ‘a picture painted by a person called Jane’). They cannot gain wider currency, as the communicative advantage that is usually inherent in using metonymic shortcuts when a metonym links concepts that (1) contain well-structured knowledge that is (2) shared by the majority of a speech community, is not given (cf. Section 4.5. below for a more detailed discussion of the role of the structure of concepts for the conventionality of metonyms). The examples at the other extreme of the scale, like the White House standing for ‘the President of the US and the people who advise him’ (LDOCE4, s.v. White House) have already found their way into most dictionaries as fully established senses of the words in question. Whereas linguistic metaphors can exploit the conventional mappings in novel or uncommon ways, metonymies lack this creative potential. For metonymies it

90 Sandra Handl is thus merely the frequency of occurrence which differentiates more conventional from less conventional cases. Aptness, the last criterion mentioned to account for the conventionality of metaphors, also cannot be transferred to metonymy without problems. Neither can we differentiate metonymic mappings with regard to the conceptual distance between target and vehicle, as both belong by definition to the same domain or ICM, nor can we speak of a ‘similarity of features’. To apply aptness to metonymic construals, the term has to be defined in a different way. As Markert and Hahn (2002: 152) put it, the degree of aptness of a given metonymy depends on “the typicality of the relation between the two objects [i.e., vehicle and target; SH] in question”. To sum up, the only criterion that can be transferred from metaphors to metonymies without any difficulties is that conventional metonymies (just like conventional metaphors) are those which are shared by a speech community and which therefore lead to a whole range of linguistic expressions motivated by them. The question is thus: Which other factors can be used to describe and measure the conventionality of metonymies? What exactly is it that makes some metonyms such as bottle (‘contents’) in example (1) intuitively sound much more common than others like suits (‘businessmen’) in (2), even though both have been said to be supported by conventional conceptual metonymic mappings (CONTAINER FOR CONTENT and COSTUME FOR WEARER respectively)? (1)

He drank another bottle. (BNC-1010)3

(2)

Seemingly, the only criterion adhered to by these ‘starmakers’ in signing new talent is not to upset the suits upstairs. (C9N-2157)

As will be shown in the following sections, the difference can be explained with the help of the notion of salience. Whereas earlier research has mostly concentrated on only one of the facets of salience, namely on its ontological aspects, it will be suggested that an accurate account of the conventionality of metonymy can only be given if attribute salience, i.e. the salience of the attributes which are associated with the concepts involved in the metonymic mapping, is considered as well.

Salience and the conventionality of metonymies

91

3. Salience as a measure of the conventionality of metonymies 3.1. The notion of salience As Schmid (2007: 119–120) states, the term salience can be used in two different but related senses. On the one hand, salience – or more accurately ontological salience – refers to the fact that some entities are more likely to attract our attention than others. On the other hand, the term can be used to describe the temporary activation of a concept, i.e. when a concept is currently activated – no matter why – it is cognitively salient and can be processed with a minimum of cognitive effort. The two aspects are obviously linked: since ontologically salient entities will often be the focus of somebody’s attention, the corresponding concepts become more frequently activated, i.e. cognitively salient, than those subsuming ontologically nonsalient entities. The frequent activation of these concepts fosters their entrenchment so that, according to Langacker (1987: 59), their activation becomes even more automatic.

3.2. Ontological salience and preferred routes of metonymic mapping The idea of ontological salience is especially prominent in Langacker’s approach to metonymy, who conceives of metonymies as reference-point constructions. In a metonymic expression such as I filled up the car the metonymic vehicle car is used to establish mental contact with an associated concept, the metonymic target, in this case the petrol tank. The aptness of concepts to function as cognitive entry points depends on their “relative salience”, which Langacker (1993: 30) describes as follows: Other things being equal, various principles of relative salience generally hold: human > non-human; whole > part; concrete > abstract; visible > nonvisible; etc.

The most detailed account of the relationship between these general principles and the preferred direction of mapping in metonymies is given by Kövecses and Radden (1998) and Radden and Kövecses (1999). They propose several principles that are grounded in human experience, perceptual selectivity, and cultural preferences. These principles determine the “default-routes” (Radden and Kövecses 1999: 52) of metonymic mappings. Only a few examples can be given here to illustrate their point (cf. Table 1, below).

92 Sandra Handl Table 1. Some examples of cognitive principles governing the choice of metonymic vehicles (selected from Radden and Kövecses 1999: 44–50) Principle Human experience:

HUMAN OVER NON-HUMAN CONCRETE OVER ABSTRACT

Example I have a flat tire. ‘car’ having one’s hand on s.th. ‘controlling something’

Perceptual selectivity:

IMMEDIATE OVER NONIMMEDIATE GOOD GESTALT OVER POOR GESTALT

She is my joy. ‘reason for joy’ The car needs washing. ‘exterior’

Cultural preferences:

IDEAL OVER NON-IDEAL

You are a Judas. ‘betrayer par excellence’ speak a language ‘know a language’

IMPORTANT OVER LESS IMPORTANT

If a metonymy is in accordance with a large number of the principles, it has a high degree of cognitive motivation. And the larger the motivation of a metonymy, the more likely it will be adopted by the speech community: The default-routes “help us understand ... why certain vehicle-to-target routes have become conventionalized in the language” (Radden and Kövecses 1999: 52).4 As a consequence, those linguistic instantiations of metonymic mappings which follow the preferred routes are felt to be less figurative, i.e. are more conventional, than other instances which don’t. According to this classification the above-mentioned case car for ‘petrol tank’, a WHOLE FOR PART metonymy, is hardly ever noticed as having a figurative motivation, whereas set of wheels for ‘car’, a PART FOR WHOLE metonymy, is more obviously non-literal because speakers prefer wholes as reference points. This suggests that the (un)conventionality of metonymies should be relatively easy to determine. Those which obey the principles of ontological salience are conventional, whereas those which do not are unconventional. Unfortunately, however, things are a bit more difficult. When one tries to apply the laws of ontological salience, one is faced with two problems: First, there is no established hierarchy of these principles. So it remains unclear when and why, for example, a principle like HUMAN OVER NONHUMAN will be overridden by other considerations. And second, it is not necessarily the case that all linguistic examples which are motivated by the same metonymic mapping behave alike with respect to either obeying or

Salience and the conventionality of metonymies

93

disobeying the principles of ontological salience. Therefore it is not always possible to describe the conventionality of metonymies on the more comprehensive level of conceptual mappings. Sometimes one has to have recourse to the level of single linguistic instantiations, which of course allows less generalizations. In what follows, I will illustrate these two problems with the help of the results of a corpus-based study.

4. A corpus study of the conventionality of selected metonymies 4.1. The method In order to assess the conventionality of metonymies, the proportion of metonymic meanings of selected lexemes was investigated with material from the British National Corpus (BNC). The idea behind this method is that the conventionality of a metonymic reading of a lexeme correlates with the relative frequency of this reading vis-à-vis the frequency of other readings. If, as was for example the case for the noun press, the great majority of all records of the lexemes investigated were metonyms (for press ‘media’, the metonymic uses amount to 92.3%), then it seems plausible to postulate a very high degree of conventionalization for these metonymic meanings. If, on the other hand, only a very small proportion of metonymic uses were found (e.g. 0.4% for the noun wine ‘bottle/glass of wine’), then the metonymic meaning is arguably not highly conventionalized. While it is not quite clear how individual percentages can be interpreted, the findings do provide a good indicator for different degrees of conventionalization. I extracted 100 potentially metonymic expressions from the relevant literature and used them as a basis for examining how often the metonymies discussed in cognitive linguistics are actually used in ordinary language. If the potentially metonymic word-form occurred less than 1000 times in the BNC, all instances were taken into consideration. In cases where a search for the word-form returned more than 1000 hits, a sub-corpus of 1000 uses of the form was created by downloading a random sample. In a next step, all concordance lines were analyzed in terms of whether they realized a metonymic reading or not. Since many of the words that formed part of the study conveyed more than one metonymic meaning, i.e. realized more than one metonymic mapping, not only 100 but as many as 229 instances of metonymic mappings could be identified. The form airlines, for example, instantiates three different mappings: INSTITUTION FOR PEOPLE RESPONSI-

94 Sandra Handl BLE, CONTROLLER FOR CONTROLLED (‘planes’), and INSTITUTION FOR INDEX (‘shares’; cf. e.g. Markert and Nissim 2006: 163). Each metonymic

usage of a given word was assigned to a conceptual mapping in order to be able to compare different instantiations of one metonymic mapping in terms of their frequency. After all instances of the 100 word-forms had been analyzed and all metonymic meanings had been established, all 229 examples were evaluated in terms of the ontological salience of the concepts involved in the single mappings. The results are discussed in the following sections, beginning with an overview of the findings.

4.2. Overview of findings concerning laws of ontological salience Interestingly, only roughly one third of the 229 examples investigated are entirely in line with the laws of ontological salience. Typical instances include, for example: (3)

... do you realise you’ve eaten the whole packet! (KST-4483)

(4)

... ahead of Nancy Reagan, America’s First Lady ... (A0B-339)

The CONTAINER FOR CONTENT metonymy packet (‘content of packet’) in (3) is in line with VISIBLE OVER NON-VISIBLE, and the WHOLE FOR PRINCIPAL PART metonymy America (‘US’) in (4) is supported by DOMINANT OVER LESS DOMINANT. And both metonyms do not violate any of the other principles. In roughly another third of the cases, however, I found that different principles of ontological salience clash with each other: (5)

The Pentagon is advising the president about troop reductions. (AAC-633)

(6)

A blue BMW, neat but not gaudy, not souped-up, but a fast, useful car (AB9-2282)

(7)

... the people on the estate need buses which drive carefully. (K541459)

Salience and the conventionality of metonymies

95

The BUILDING FOR INSTITUTION metonymy Pentagon (‘US Department of Defense’) in (5) follows the laws of ontological salience by using a concrete entity to refer to an abstract target concept. However, the institution referred to ultimately consists of human beings so that the metonym can be said to diverge from HUMAN OVER NON-HUMAN. Similarly, the PRODUCER FOR PRODUCT metonymy BMW (‘car’) in (6) can be analyzed as conforming to HUMAN OVER NON-HUMAN, since the company again consists of human beings. In this sense, the example is in line with ontological salience. At the same time, however, the example violates the principle INTERACTIONAL OVER NON-INTERACTIONAL because what people interact with is usually the car rather than its producers. And in the CONTROLLED FOR CONTROLLER metonymy buses for ‘bus drivers’ in (7) INTERACTIONAL OVER NON-INTERACTIONAL applies, but HUMAN OVER NON-HUMAN is disregarded. The last third of the metonymies in my corpus do not at all follow the laws of ontological salience. Good examples are ATTRIBUTE FOR PERSON metonymies like appendicitis (‘person suffering from appendicitis’) or crown (‘queen/king’) which violate HUMAN OVER NON-HUMAN and are not in line with any of the other laws either. Other cases in point are CONTENT FOR CONTAINER metonymies such as wine (‘bottle/glass of wine’), which deviates from BOUNDED OVER UNBOUNDED and is not entirely in line with VISIBLE OVER NON-VISIBLE, or MATERIAL FOR OBJECT metonymies, like furs (‘coat’) or paper (‘piece of writing’), which disobey BOUNDED OVER UNBOUNDED. The first, quite general conclusion which can be drawn from these examples is that the laws of ontological salience seem to be less essential for the use of metonyms than expected. Otherwise one would not find so many metonyms which feature a conflict of principles or which entirely deviate from the predictions. More interesting, however, is that ontological salience does not seem to be a predictor of the relative frequencies of occurrence of metonymic meanings. This is most evident when one considers reversible mappings.

4.3. Evidence from reversible mappings For reversible mappings it was mostly possible to identify a marked preference for one of the directions of mapping. In some cases, this preference is evidently motivated by the laws of ontological salience. As regards the

96 Sandra Handl ICM, for example, the CONTENT FOR CONTAINER mapping was hardly ever found. One of the examples I was originally looking for in the ‘container’ meaning, ketchup ‘bottle of ketchup’ (taken over from Díez Velasco 2001: 21), could not be found at all in the BNC. The only two words of my sample which occurred as instantiations of this mapping are chicken (Take the top off the chicken; KP1-2355) and wine (... Benstede ran his finger round the rim of his wine up; BMN-487). However, both words realize this mapping infrequently: Only 0.2% of 1000 instances of chicken and 0.4% of 1000 instances of wine were found in the ‘container’ meaning. This stands in marked contrast to the results gained for the CONTAINER FOR CONTENT mapping, which was found with seven different word-forms and tends to motivate a higher percentage of uses. The relative frequencies of the examined forms that occurred as instances of the CONTAINER FOR CONTENT mapping are given in Table 2. The second column dubbed ‘Mapping – total meanings’ indicates how many instances of all records investigated instantiated a ‘content’ meaning. The third one (‘Mapping – all metonymic meanings’) specifies the relationship between the CONTAINER FOR CONTENT mapping and other metonymic meanings for the given word-forms, i.e. it illustrates how many of the metonymic readings of a given word-form were motivated by the CONTAINER FOR CONTENT mapping. CONTAINMENT

Table 2. Relative frequencies of selected CONTAINER FOR CONTENT mappings in the BNC Word-form book dish kettle bottle packet label glasses

Mapping – total meanings (%)

Mapping – all metonymic meanings (%)

44.30 32.70 19.46 9.90 7.40 2.70 2.00

89.31 100 88.24 99.00 91.36 8.54 2.03

Admittedly, the percentages of meanings motivated by the CONTAINER FOR CONTENT mapping vary considerably: while the ‘content’ meaning with 44.3% seems to be quite common for book, it is only of marginal relevance for label (2.7%) and glasses (2%). The low percentages found for glasses and label have to be seen in connection with the fact that the two forms are frequently used as instances of other metonymic mappings; only 8.54% of all metonymic uses of label and 2.03% of all metonymic uses of glasses are

Salience and the conventionality of metonymies

97

motivated by CONTAINER FOR CONTENT. For the other words, the CONTAINER FOR CONTENT mapping motivates the majority of all metonymic uses which results in higher frequencies of the ‘content’ meaning, reaching up to more than 44%. On the whole, then, the CONTAINER FOR CONTENT mapping is obviously used more often than its reversal. It was also possible to establish a preferred route of mapping for the PRODUCTION ICM, even though both mappings, i.e. PRODUCER FOR PRODUCT and PRODUCT FOR PRODUCER, invariably lead to conflicts in terms of ontological salience. Typically, examples of the PRODUCER FOR PRODUCT mapping (e.g. Picasso ‘work of art’) are in line with HUMAN OVER NONHUMAN, but violate IMMEDIATE OVER NON-IMMEDIATE, whereas PRODUCT FOR PRODUCER examples (e.g. radio ‘communicator’) conform to IMMEDIATE OVER NON-IMMEDIATE or INTERACTIONAL OVER NON-INTERACTIONAL, but defy HUMAN OVER NON-HUMAN. Here, the laws of ontological salience do not lead to unambiguous predictions as to which of the two should be favoured. The relative frequencies of the examples investigated of the two different directions of mapping are given in Table 3. Table 3. The relative frequencies of selected metonyms in the PRODUCTION ICM Mapping

Word-form

Metonymic sense

PRODUCER FOR PRODUCT

BMW Renault Shakespeare Picasso Einstein Einstein buses label newspaper newspaper law paper opera radio book paper radio label packet hands

‘car’ ‘car’ ‘writing’ ‘work of art’ ‘theory’ ‘writing’ ‘exhaust’ ‘institution’ ‘institution’ ‘author’ ‘author’ ‘author’ ‘institution’ ‘communicator’ ‘author’ ‘institution’ ‘institution’ ‘author’ ‘author’ ‘artist’

PRODUCT FOR PRODUCER

Mapping – total meanings (%) 62.67 43.74 34.90 8.32 2.26 0.50 0.10 25.50 23.20 12.20 10.70 10.00 9.70 6.50 5.30 4.00 3.80 2.00 0.50 0.10

98 Sandra Handl The results are slightly confusing at first glance, since some of the wordforms instantiate different sub-mappings within the two larger groups of PRODUCER FOR PRODUCT and PRODUCT FOR PRODUCER, and therefore lead to different metonymic meanings. In addition, the single relative frequencies of the metonymic ‘producer’ and ‘product’ uses are quite heterogeneous. What is remarkable, however, is that although the group of the PRODUCT FOR PRODUCER examples investigated was larger, the highest relative frequencies could be established for the PRODUCER FOR PRODUCT group. None of the 13 examples that were found in a ‘producer’ meaning is as frequently used metonymically as the most common instances of the reversed mapping. Even though the mapping PRODUCT FOR PRODUCER is, all in all, a productive one, motivating many different metonyms, it does not seem to lead to highly conventional metonymic readings. In terms of the conventionalization potential of metonymic senses, the PRODUCER FOR PRODUCT mapping thus seems to be preferred. Particularly sub-mappings which have physical consumer products as targets can cause highly conventional metonymic senses (e.g. BMW, Renault). This suggests that HUMAN OVER NON-HUMAN is more important than questions of immediacy or interactionality so that mappings which violate the former are dispreferred. And indeed, a closer look at metonymies which generally defy HUMAN OVER NON-HUMAN such as ATTRIBUTE FOR PERSON reveals that they are infrequently employed in natural discourse. For example, only 0.4% of all uses of heels (‘person wearing heels’) are motivated by this mapping, 0.4% of the uses of money (‘rich people’), 1.9% of all uses of power (‘powerful people’), 0.9% of all uses of skirt (‘sexually attractive woman’), and 3.34% of all uses of suits (‘businessmen’). Yet, there is also one example whose metonymic meaning is highly frequent: crown (‘queen/king’) is used in 65% of all cases as an instance of the ATTRIBUTE FOR PERSON mapping. This raises the question if the HUMAN OVER NONHUMAN principle is after all not as vital as the other examples seem to suggest or if the example crown is simply the exception which proves the rule.5 Before this question can be discussed in more detail (cf. Section 5), I would like to illustrate some further cases which demonstrate that ontological salience alone cannot explain why some metonymies have become conventional while others have not.

Salience and the conventionality of metonymies

99

4.4. Further evidence against the laws of ontological salience As already mentioned, sometimes not all examples motivated by the same conceptual mapping behave alike with regard to either obeying or disobeying the laws of ontological salience. My examples of the CONTROLLER FOR CONTROLLED mapping are all typical in terms of ontological salience, since they all instantiate the preference for metonymic construals which use humans as reference points (e.g. Napoleon ‘Napoleon’s army, Napoleon’s activities’). This is probably also the reason why this direction of mapping is preferred according to my data. More interesting for the current purposes is, however, that some instances of CONTROLLED FOR CONTROLLER clearly deviate from the laws of ontological salience, while others only feature a conflict of different principles. BMW, for example, when used to refer to the ‘driver’ of the car, leads to a conflict, since the metonym violates HUMAN OVER NON-HUMAN, but adheres to VISIBLE OVER NON-VISIBLE as the driver is more or less ‘hidden’ within the car. Cameras, when used to refer to ‘cameramen’ or ‘photographers’, deviates from HUMAN OVER NONHUMAN, but is not in line with any other principle either. The photographers are, for instance, just as visible and immediate as their instruments, they form a gestalt which is as good as that of their instruments, etc. Astonishingly, however, the difference between the cases where a conflict of principles can be observed and cases which deviate entirely from the laws of ontological salience is not reflected in any way in the relative frequencies of the targetted items, i.e. entirely deviant items are not necessarily less frequently used than those where a conflict occurs (cf. Table 4). Table 4. Relative frequencies of selected CONTROLLED FOR CONTROLLER mappings and deviations and conflicts in terms of ontological salience Word-form cameras pen crime BMW Renault sax telephone football scalpel buses pencil

Mapping – total meanings (%)

Ontological salience

9.20 5.10 5.00 4.64 3.70 2.30 2.30 1.70 1.09 1.00 0.30

Deviation Deviation Deviation Conflict Conflict Deviation Conflict Deviation Deviation Conflict Deviation

100 Sandra Handl What this boils down to is that the laws of ontological salience can at best serve as general guidelines when it comes to the description of the conventionality of metonymies. They can lead to valid predictions regarding the preferred direction of mapping, but they are difficult to apply when it comes to single metonymic expressions. But in order to give a comprehensive account of conventionality, one needs a tool which is able to cope with the level of single linguistic instances as well, since – as has been shown above – the frequencies of linguistic metonyms which rely on the same metonymic mapping are often not very homogeneous. In addition, there are also metonymic mappings which are not reversible, such as LOCATION FOR PRODUCT MADE THERE (e.g. Bordeaux ‘wine, mustard’), where ontological salience does not help much in order to describe or predict their conventionality. To cover conventionality on the level of single linguistic instantiations and to explain non-reversible metonymic mappings, one has to turn to other factors. And this is where the second facet of salience, i.e. cognitive salience, comes into play.

4.5. Cognitive salience and the structure of concepts Cognitive salience is concerned with the temporary activation and entrenchment of concepts. But concepts themselves are complex structures marked by a number of attributes. Thus, not only concepts as wholes can vary regarding their cognitive salience, but also the attributes they are associated with. The idea of attribute salience is taken over from Ortony et al. (1985: 570), who define it as “the prominence or importance of an attribute in a person’s representation of an entity or category”. To take their example, for a category like GENERAL ANESTHETICS one would find the attribute ‘induces sleep’ which is highly salient for all the members of the category and holds the members of the category together. The relevance of attribute salience for metonyms can be illustrated with the help of the following example: (8)

Take some aspirin, and lie down if you need to. (HA6-2613)

Radden and Kövecses (1999: 34) describe the metonym aspirin for ‘any pain-relieving tablet’ as an instance of the mapping MEMBER OF CATEGORY FOR THE CATEGORY. But why is this specific metonymic vehicle chosen? The communicative goal is to convey the idea that any type of pain reliev-

Salience and the conventionality of metonymies

101

ing tablet might help. Aspirin is undoubtedly the most typical member of the category PAIN RELIEVING TABLETS and as most members of the speech community link the attribute ‘pain relieving’ most effortlessly with aspirin, the metonym is the most efficient way to achieve this aim. In other words, it is highly apt and can thus be decoded rapidly. The idea of attribute salience can also explain the difference between two PRODUCER FOR PRODUCT metonyms discussed by Barcelona (2002: 231) and Taylor (2003: 125): (9)

I’ve just bought a Picasso.

(10)

*I’ve just bought a Jane.

Both examples are equally supported by the laws of ontological salience, such as HUMAN OVER NON-HUMAN or CONCRETE OVER ABSTRACT. But while (9) is a conventional way of referring to a picture by Picasso, (10) could hardly be used to refer to a picture painted by a woman called Jane – except perhaps in a very restricted range of contexts. Referring to Taylor’s (2002, 2003) ideas, Barcelona (2002: 230–231) argues that the conventionalization of the latter example is blocked by the lack of “a specific cultural ..., social-interactional, or aesthetic principle”. The ultimate reason why (9) is a quite conventional way of referring to Picasso’s art, but (10) would not be accepted by the speech community is that most speakers have a clearly structured mental representation of PI6 CASSO, but lack one for JANE. The concept PICASSO includes knowledge or, in my terminology, contains attributes like ‘was a man’, ‘was a father’, ‘was an artist’, ‘produced paintings/sculptures’, ‘used certain techniques’, ‘is important in art history’, etc. For JANE, things are different. The only piece of information one has or can gather from the name is that the person is female. The JANE concept is almost empty, while the PICASSO concept is full of information of different degrees of salience. That he ‘was a father’, for example, will not be the aspect that comes to speakers’ minds first in a neutral context. Rather, the mention of the name will immediately evoke the attribute ‘was an artist’. It is thus not the lack of a ‘specific cultural principle’ which accounts for the unconventionality of (10), as Barcelona (2002: 230–231) holds. The cultural principle exists – it links painters to their pictures, authors to their books, etc. – so why does it not link Jane to her works of art? The cultural principle pertains only if the concepts to

102 Sandra Handl which it applies (in this case the artists) hold enough and salient information. Shared salient knowledge is, as it were, its necessary precondition. That the PICASSO concept includes attributes which differ with regard to salience is also the reason for the aptness of this PRODUCER FOR PRODUCT metonymy. If Picasso were equally well-known for his works of art and, say, his cooking, the relation would no longer be unambiguous and the metonymy a Picasso therefore less likely to gain wider currency. The association between vehicle and target must be a distinctive one, at least in unmarked contexts (cf. Clark and Gerrig 1983: 560).7 The conventionality of metonymies therefore depends on several factors: First, a linguistic metonym should rely on one of the entrenched conceptual mappings. Second, it is certainly helpful if the direction of mapping is not entirely at odds with the laws of ontological salience. And third, metonyms stand a better chance of being accepted as conventional ways of expression if they draw on attributes which are highly salient in the vehicle. As the following section will show, the latter factor can override the second, i.e. attribute salience can be more important than ontological salience, and cause a high conventionality of metonyms which should actually be ‘unconventional’ if one considers ontological salience only.

5. Target-in-vehicle salience as a measure of the conventionality of metonymies The communicative success of a metonymic construal is related to the structure of the concepts involved in the mapping. They have to be filled with knowledge shared by the members of the speech community. And there has to be one attribute that is especially salient. However, not any highly salient attribute will do. In (8), the attribute ‘relieves pain’, which is the common denominator of the target category of PAIN RELIEVING TABLETS, is probably the most salient attribute of the vehicle concept ASPIRIN. Similarly, there is a close associative link between the target domain A PICASSO and the highly salient information in the PICASSO AS PERSON concept that he ‘was an artist’ or ‘created works of art’ in (9). In both cases, the acceptability of the metonyms depends on the fact that the target concept is a highly salient part of the vehicle concept. In general, the conventionality of a given linguistic metonym can only be judged in relation to other instantiations of the same conceptual mapping. Consider the COSTUME FOR WEARER mapping. All of the examples I

Salience and the conventionality of metonymies

103

investigated deviate from HUMAN OVER NON-HUMAN, but while instances like heels or skirt, for example, are rarely used as metonyms (0.4% and 0.9% respectively), the metonymic meaning of crown is more likely to be encountered than the literal one (cf. Table 5). Table 5. Relative frequencies of selected COSTUME FOR WEARER mappings Word-form crown uniforms suits skirt heels glasses

Mapping – total meanings (%) 65.00 6.60 3.34 0.90 0.40 0.10

How can this be explained with attribute salience? The concept CROWN is undeniably a salient part of the idealized model of a ROYAL and it stands in an unambiguous relationship to its wearer. In neutral contexts, there are only a few people who are entitled to wear crowns, i.e. kings or queens. The attribute ‘crown’ has a high cue validity with respect to the category ROYAL. In other words, the target concept is salient in the vehicle concept. This is different for heels or skirt. That heels, for example, are usually worn by persons as part of their shoes is probably self-evident, but presumably not the most central idea that is connected with the concept. Moreover, while the metonym crown for ‘queen/king’ basically has unique reference, heels and skirt can refer to whole classes of entities. The identification of a single person by way of a metonym like heels or skirt is therefore contextdependent, whereas crown works well in neutral contexts due to the close link between royals and this symbol of their status. Another mapping where the relevance of attribute salience becomes manifest is MATERIAL FOR OBJECT. Even though this mapping invariably violates the principle BOUNDED OVER UNBOUNDED, it seems to lead to linguistic metonyms which have a considerable potential to become conventional. This is at least suggested by the number of examples in my data that have high relative frequencies of ‘object’ meanings (cf. Table 6, below). The frequencies found for the single items differ enormously. A likely explanation for these differences is the specificity of the relation between the materials and the objects made from them. Glass is mainly used for producing windows, spectacles, and various types of containers. Other materials which have high percentages in Table 6 are also fairly specified as to which products are typically made of them: fur is used to make coats or

104 Sandra Handl other clothing items, and paper occurs most often in the form of sheets of paper. The range of products which can be made of plastic is, however, nearly unrestricted. This may be the reason why it is so difficult to use the material in order to identify and refer to a specific object.8 Table 6. Relative frequencies of selected MATERIAL OF OBJECT FOR OBJECT mappings Word-form glasses furs paper wood silver bronze plastic oak

Mapping – total meanings (%) 98.40 65.33 49.20 21.90 20.80 8.90 2.50 0.40

To verify this assumption, a small-scale study was carried out with 20 native speakers of German. For the study, metonyms were selected from my sample which work in both English and German (Vietnam, Flasche ‘bottle’, Plastik ‘plastic’, Busse ‘buses’, Pille ‘pill’, Renault, Bordeaux, Ketchup ‘ketchup’, Weißes Haus ‘White House’, and Picasso). The study consisted of two parts, a context condition and a context-free condition. To half of the participants the potentially metonymic expressions, i.e. the target words, were presented in isolation and to the other half embedded in a context which was biased towards a metonymic reading (e.g. Mit Plastik zu bezahlen wird immer beliebter ‘Paying with plastic is becoming more and more popular’). In both cases, the participants were asked to list any attributes of the targets that immediately came to their mind, either upon encountering the targets in isolation or after they had read the sentences in which the targets were embedded. The aim of the study was to elicit salient attributes of the concepts investigated in the two conditions. Despite the restricted scope of the study, the results corroborate the view that some metonymic targets are highly salient in the vehicle concepts. Interestingly, those for which a high target-in-vehicle salience could be observed in the survey with the German native speakers are relatively frequent in English, while those expressions which are not often used metonymically in English correspond to those where the target is not salient in the vehicle concept of German speakers. This does, however, not come as a surprise, considering that the experiences of Germans with bottles, plastic,

Salience and the conventionality of metonymies

105

and the like are probably not too different from those of English native speakers. The context-free part showed that Plastik ‘plastic’ is associated with many different objects, but that none of them is especially salient. The most frequently mentioned attribute (listed by half of the participants) was ‘material that can be used to produce virtually anything’, which is certainly a telling result. Other, less important attributes which are related to objects made of plastic include ‘bags’, ‘works of art’ (each mentioned three times), ‘toys’, and ‘bottle’ (each mentioned twice). Apart from purely objectrelated attributes, another relevant group of attributes centres around the idea of waste (mentioned twice) and its recycling (mentioned four times). The remaining attributes of the 24 that could be collected for Plastik were each mentioned by one participant only and typically referred to the material itself and its properties (e.g. ‘transparent’, ‘artificial’, ‘unbreakable’). In the context-condition the picture changed significantly. The metonymically biasing context was quite unambiguous, which accounts for the high number of occurrences of ‘(credit) card’ in the attribute lists. All participants mentioned this. All the other attributes which appeared more than once are related to the intended metonymic meaning: ‘cashless’, ‘money’, and ‘handy’. The material – even though it is metonymically foregrounded – is not of particular relevance.9 Of the 26 different attributes collected in the context-condition, only a few are related to the material: ‘not tasty’, ‘can melt’, ‘industrialized’, and ‘synthetic material’. If the results of the corpus study are considered in isolation, it seems as if the mapping MATERIAL FOR OBJECT were a metonymic pattern which regularly leads to quite conventional linguistic metonyms, despite its being at odds with the laws of ontological salience. However, if attribute salience is taken into account as well, the findings can be interpreted differently. It is not necessarily the case that the mapping is, in general, particularly likely to lead to conventional linguistic metonyms. It only happened to be the case for my sample of metonyms motivated by this mapping, since most of them are in line with the preference for metonyms whose target is salient in the vehicle: ‘sheets of paper’ is salient in the PAPER concept, ‘coats’ in the FUR concept, etc. Not surprisingly, the sources from which the metonymies were collected for the study in the first place often use examples which can serve the purpose of demonstrating that metonymies can indeed be quite common. But this does not mean that the examples are necessarily typical of the mapping itself. This indicates that a high target-in-vehicle salience can possibly override the laws of ontological salience and lead to highly

106 Sandra Handl conventional linguistic metonyms which are based on a mapping which is, on the whole, not too conventional. The findings of the attribute listing task together with the results of the corpus study support the view that linguistic metonyms are more likely to become conventional if the target is a salient part of the vehicle concept. To take a further example, the survey showed that the PRODUCER FOR PRODUCT metonym Renault, which is instantiated by as much as 43.74% of all occurrences of the word, is in line with target-in-vehicle salience. The most frequent attribute of Renault in the context-free condition is ‘car’, which occurred in eight of the ten lists. Second are references to ‘France’ (seven times). Further attributes which occurred more than once are ‘ugly’, ‘world champion’, ‘Formula One’, ‘Twingo’, and ‘créateurs d’automobiles’. The conceptual link between the COMPANY and its PRODUCTS is so strong that the products are most salient even when the company name is encountered in isolation. Attributes belonging to the domain of MOTOR RACING are also quite salient for Renault, whereas the only attribute which can be said to be connected to the COMPANY is the slogan créateurs d’automobiles. Other items tested also confirm the relevance of target-in-vehicle salience for the conventionality of metonyms. For Flasche ‘bottle’, for example, the most essential attributes centre around the bottle itself, its material and properties. Content-related attributes (e.g. ‘beer’, ‘water’, ‘wine’, ‘one litre’) are already important in the context-free condition, but their significance increases as soon as a metonymically biasing context is provided. This supports the results of the corpus study, according to which the ‘content’ meaning is with 9.9% not unconventional, but not highly conventional either. In the conceptual representation of Ketchup ‘ketchup’, an example of the reversed mapping, i.e. CONTENT FOR CONTAINER, however, container-related attributes are not salient at all. This is probably the reason why no instance of this metonym could be found in the BNC. The LOCATION FOR PRODUCT metonym Bordeaux, which was found with 11.71% of the occurrences of the word, again features the metonymic target as a salient subpart of the vehicle. In my survey, ten out of 22 attributes given in the context-free condition for Bordeaux are connected to the typical product (e.g. ‘tastes dry’, ‘red’, ‘stored in catacombs’, ‘wine-growing’). All in all, there seems to be a strong relation between linguistic metonyms which are relatively frequent in natural discourse – always considered within the framework of the general productivity of the mapping motivating them – and a high degree of target-in-vehicle salience and vice versa. Target-in-vehicle salience thus can be helpful in more than one way: it can

Salience and the conventionality of metonymies

107

lend further support to assumptions concerning the preferred direction of mapping for reversible metonymic patterns, but its applicability is not restricted to such cases. It can also be used as an indicator of ‘exceptions to the rule’. If various examples of the same mapping are compared in terms of their frequencies, it can explain why some linguistic metonyms are much more or much less frequent and conventional than the overall productivity of the mapping implies. In this way, taget-in-vehicle salience can be used to explain the rule and any exceptions alike.

6. Conclusions While the criteria of the conventionality of metonymies which have been proposed by researchers such as Radden and Kövecses (1999), Barcelona (2002, 2003), and Taylor (2002, 2003) offer useful starting points, they are often not specific enough to allow a description of all cases. The application of the laws of ontological salience, which are mentioned by most sources, is bound to be problematic. On the one hand, the lack of a clear hierarchy of the different principles renders predictions with regard to the conventionality of many mappings difficult. To complicate matters further, not all instances of the same mapping inevitably behave in the same way with respect to either obeying or violating them. This means that, even though ontological salience is primarily intended as a criterion for the description of whole conceptual mappings, it cannot entirely fulfil this function. It cannot cope with cases which feature conflicts of different principles, it cannot explain why there is no correlation between the frequency of usage of different examples of the same mapping and their either obeying or disobeying the principles, and it cannot offer any explanation why there are some examples which are either markedly more or less frequent than other instances of the same mapping. In short, it cannot explain any of the exceptions. The principles are only useful as guidelines, for example, when it comes to predictions as to which of two possible mapping directions is normally preferred. But to validate these predictions, corpus studies have to be carried out to judge the actual conventionality of different mappings and single linguistic examples. However, if considered in isolation, even corpus-linguistic data are not necessarily a safe indicator of the conventionality of single linguistic instances. Given that any corpus-linguistic investigation of the frequency of metonymic uses of words involves the researcher manually analyzing every

108 Sandra Handl concordance line, corpus studies will always be limited in scope. Thus, the choice of the test items can, as has been shown above, unduly influence the results and lead to erroneous assumptions concerning the overall conventionality of a given mapping. An adequate description of the conventionality of metonymies can only be given if attribute salience is taken into account, since a high degree of target-in-vehicle salience can override the more general preferences described by ontological salience. Accordingly, attribute salience can be useful to distinguish those cases which behave regularly, i.e. which behave as examples of a given mapping commonly tend to do, from exceptions to these general tendencies and it also serves as a means of analyzing these exceptions. Further investigations dealing with the conventionality of metonymies should therefore consider both levels, the more comprehensive conceptual level of metonymic mappings and the level of single linguistic examples and draw a clear dividing line between them. Not every conventional metonym is representative of the conventionality of a mapping as a whole and hence does not necessarily allow one to argue that the underlying conceptual mapping is a deeply entrenched way of thinking.

Notes 1. 2.

3. 4.

5.

For a slightly different position, cf. Croft and Cruse (2004: 206). Evidently, the boundary between highly conventional examples and those which exploit existing patterns in a more imaginative way is much fuzzier than the dividing line that separates entirely novel metaphorical conceptualizations from conventional ones. This is also pointed out by Clausner and Croft (1997), who propose to reformulate the underlying conceptual mappings in more specific ways in order to clarify the conceptual difference between examples like We need to buttress the theory with solid arguments and *That line of reasoning has no plumbing. The examples are taken from the British National Corpus. Obviously, these principles can be overridden by other considerations such as the avoidance of taboo (e.g. Where can I wash my hands? which conflicts with CENTRAL OVER PERIPHERAL and RELEVANT OVER IRRELEVANT; cf. Radden and Kövecses 1999: 53). To assess this, it would be useful to relate the frequency of the metonym crown (‘queen/king’) to the frequencies with which other (literal) linguistic expressions are used to refer to queens and kings. Since, however, my study focused on metonymies only, this issue cannot be addressed here.

Salience and the conventionality of metonymies 6.

7.

8.

9.

109

According to the standard view of the philosophy of language, names do not evoke concepts, but are only used for reference. This is, for example, evident in Frege’s article “On sense and reference”: “a proper name ... has as its reference a definite object ..., but not a concept or a relation” (Frege [1952] 1993: 23). However, it can hardly be denied that the mention of names of celebrities like Picasso will invariably lead to an activation of stored knowledge concerning the person. That this is indeed the case can be shown with the help of a simple attribute listing task. The subjects who partook in the small-scale study described in Section 5 found the task of listing attributes for Picasso a meaningful one and experienced no difficulties in coming up with a considerable number of ideas connected to the PICASSO AS PERSON concept. Given the right circumstances and a particular common ground of the speakers, literally any metonymy can be used and understood as long as it follows one of the entrenched conceptual patterns, but a conventionalization in the whole speech community depends on shared knowledge on a larger scale. Instances of the ‘object’ meaning of oak (‘cask’) presuppose another mapping, i.e. OBJECT FOR MATERIAL OF OBJECT, which motivates the meaning ‘wood of oak tree’. This latter mapping underlies all metonymic uses of the word oak and can be combined with various others. The cases in which it is combined with the MATERIAL FOR OBJECT mapping are probably rare, since the metonymic chain leads from an object (the tree) over the material (the wood) to another object (the cask). The chain is thus quasi-self-contradictory as it involves a combination of a mapping and its reversal. This amply supports Langacker’s (1987, 1993) assumption that once the target has been accessed, the reference point recedes into the background.

References Barcelona, Antonio 2002 Clarifying and applying the notions of metaphor and metonymy within cognitive linguistics: An update. In Metaphor and Metonymy in Comparison and Contrast, René Dirven, and Ralf Pörings (eds.), 207–277. Berlin/New York: Mouton de Gruyter. Barcelona, Antonio 2003 Metonymy in cognitive linguistics. An analysis and a few modest proposals. In Motivation in Language. Studies in Honor of Günter Radden, Hubert Cuyckens, Thomas Berg, René Dirven, and KlausUwe Panther (eds.), 223–255. Amsterdam/Philadelphia: John Benjamins.

110 Sandra Handl Blasko, Dawn G., and Cynthia M. Connine 1993 Effects of familiarity and aptness on metaphor processing. Journal of Experimental Psychology: Learning, Memory, and Cognition 19 (2): 295–308. Chiappe, Dan L., John M. Kennedy, and Tim Smykowski 2003 Reversibility, aptness, and the conventionality of metaphors and similes. Metaphor and Symbol 18 (2): 85–105. Clark, Herbert H., and Richard J. Gerrig 1983 Understanding old words with new meanings. Journal of Verbal Learning and Verbal Behavior 22: 591–608. Clausner, Timothy C., and William Croft 1997 Productivity and schematicity in metaphors. Cognitive Science 21 (3): 247–282. Croft, William, and D. Alan Cruse 2004 Cognitive Linguistics. Cambridge etc.: Cambridge University Press. Díez Velasco, Olga I. 2001 The role of semantic relations in the creation of metonymic mappings. Miscelánea 23: 11–28. Frege, Gottlob 1993 Reprint. “On sense and reference”, In Meaning and Reference, Adrian W. Moore (ed.), 23–42. Oxford etc.: Oxford University Press. Originally published in Translations from the Philosophical Writings of Gottlob Frege, Peter Geach, and Max Black (eds.), 56– 78, Oxford: Blackwell, 1952. Giora, Rachel 1997 Understanding figurative and literal language: The graded salience hypothesis. Cognitive Linguistics 8 (3): 183–206. Giora, Rachel 2003 On Our Mind. Salience, Context, and Figurative Language. Oxford etc.: Oxford University Press. Katz, Albert N. 1989 On choosing the vehicles of metaphors: Referential concreteness, semantic distance, and individual differences. Journal of Memory and Language 28: 486–499. Kövecses, Zoltán 2002 Metaphor. A Practical Introduction. Oxford etc.: Oxford University Press. Kövecses, Zoltán, and Günter Radden 1998 Metonymy: Developing a cognitive linguistic view. Cognitive Linguistics 9 (1): 37–77.

Salience and the conventionality of metonymies

111

Lakoff, George, and Mark Johnson 1980 Metaphors We Live By. Chicago/London: University of Chicago Press. Lakoff, George, and Mark Turner 1989 More Than Cool Reason. A Field Guide to Poetic Metaphor. Chicago/London: University of Chicago Press. Langacker, Ronald W. 1987 Foundations of Cognitive Grammar. Vol. I: Theoretical Prerequisites. Stanford/CA: Stanford University Press. Langacker, Ronald W. 1993 Reference-point constructions. Cognitive Linguistics 4 (1): 1–38. Markert, Katja, and Udo Hahn 2002 Understanding metonymies in discourse. Artificial Intelligence 135: 145–198. Markert, Katja, and Malvina Nissim 2006 Metonymic proper names: A corpus-based account. In Corpus-based Approaches to Metaphor and Metonymy, Anatol Stefanowitsch, and Stefan Th. Gries (eds.), 152–174. Berlin/New York: Mouton de Gruyter. Ortony, Andrew, Richard J. Vondruska, Mark A. Foss, and Lawrence E. Jones 1985 Salience, similes, and the asymmetry of similarity. Journal of Memory and Language 24: 569–594. Radden, Günter, and Zoltán Kövecses 1999 Towards a theory of metonymy. In Metonymy in Language and Thought, Klaus-Uwe Panther, and Günter Radden (eds.), 17–59. Amsterdam/Philadelphia: John Benjamins. Saussure, Ferdinand de 1983 Course in General Linguistics, Charles Bally, and Albert Sechehaye (eds.). Translated and annotated by Roy Harris. London: Duckworth. Original edition 1916. Schmid, Hans-Jörg 2007 Entrenchment, salience, and basic levels. In The Oxford Handbook of Cognitive Linguistics, Dirk Geeraerts, and Hubert Cuyckens (eds.), 117–138. Oxford etc.: Oxford University Press. Summers, Della (ed.) 2003 Longman Dictionary of Contemporary English. 4th ed. (LDOCE4). Harlow: Pearson Education. Taylor, John R. 2002 Category extension by metonymy and metaphor. In Metaphor and Metonymy in Comparison and Contrast, René Dirven, and Ralf Pörings (eds.), 323–347. Berlin/New York: Mouton de Gruyter.

112 Sandra Handl Taylor, John R. 2003 Linguistic Categorization. 3rd ed. Oxford etc.: Oxford University Press.

Part II: Metaphor and metonymy: Usage-based investigations

The role of metaphor scenarios in disease management discourses: Foot and mouth disease and avian influenza Brigitte Nerlich

1. Introduction This chapter compares the role and function of various metaphor scenarios in two contested areas of science-society discourse, especially media discourses dealing with disease management relating to foot and mouth disease (FMD) and avian influenza respectively. It focuses on media coverage in the autumn of 2005 with some references to developments in early 2006 and early 2007.1 FMD is an epizootic, that is, an outbreak or epidemic of disease in animal populations. In 2001, it affected sheep, cattle and pigs in the UK with devastating effects for the rural economy. Avian flu or bird flu, especially the highly pathogenic H5N1 strain, is a zoonotic, an emergent disease that can be passed from animals, whether wild (in this case wild birds, especially geese and ducks) or domesticated (in this case poultry), to humans. So far it has only caused a limited number of human deaths caused by close contact with poultry, but it is feared that the virus might mutate and trigger a type of pandemic flu transmissible between humans.2 This chapter examines systematic similarities and differences in the uses of metaphor scenarios in discourses relating to these two types of diseases and explores the emergence of metaphor use during noteworthy political and social events. UK media coverage3 has been chosen as the best source through which a comparative analysis can be achieved, on the grounds that the print media provide a major location where the strategic and implicit metaphors employed by different actors are filtered into the public domain.

2. Conceptual framework The work presented here is based on two traditions. On the one hand, it has emerged from a critique of traditional metaphor analysis in cognitive lin-

116 Brigitte Nerlich guistics; on the other it is linked to work on metaphor in the social studies of science and the cultural study of agriculture, especially relating to infectious diseases in humans and animals (Joffe and Haarhoff 2002; Joffe and Lee 2004; Washer 2004, 2006; Nerlich 2004; Wallis and Nerlich 2005; Larson, Nerlich, and Wallis 2005; Nerlich and Halliday 2007; Chiang 2007; Koteyko, Brown, and Crawford 2008). Since around 1980 cognitive linguists have studied conceptual metaphors which pervade our thinking and talking and which may guide actions and interactions between actors. According to conceptual metaphor theory a conceptual metaphor is defined as understanding one conceptual domain in terms of another conceptual domain, e.g. LOVE in terms of a JOURNEY. In the conceptual metaphor LOVE IS A JOURNEY the concept of JOURNEY, the source concept, is mapped onto the concept of LOVE, the target. Such conceptual metaphors can find linguistic expression in utterances, such as Our relationship has reached a cross-road or We have come to the end of the road (see Lakoff and Johnson 1980). In recent years a new type of study has emerged which deals not only with such conceptual metaphors and their cognitive potential, but also with discourse metaphors and their social and political potential (for an overview see Frank 2008). In contrast to conceptual metaphors for which cognitive linguists claim a quasi-universal status, discourse metaphors (see Zinken, Hellsten, and Nerlich 2008; Zinken 2007) are regarded as relatively stable metaphorical mappings that function as key framing devices within a particular discourse over a certain period of time. They are conceptually grounded but their meaning is also shaped by their use at a given time and in the context of a debate about a certain topic. The source concepts of discourse metaphors refer to phenomenologically salient real or fictitious objects that are part of interactional space (i.e., can be pointed at, like machines or houses) and/or occupy an important place in cultural imagination (e.g. war, monsters, plague, apocalypse, etc.); and, conversely, discourse metaphors themselves highlight salient aspects of a socially, culturally or politically relevant topic. Marianne van den Boomen (2005) has recently summarized the main reasons why the notion of discourse metaphor has been introduced by some critics of traditional conceptual metaphor theory. The most compelling reason is that Lakoff and Johnson’s tradition of “cognitive” metaphor analysis cannot readily account for cultural differences in concept interpretation and mapping and tends to overlook social transformations in metaphor use over time. By contrast, distinguishing between discourse meta-

Metaphor scenarios in disease management

117

phors – which frame social assemblages of thought, expression and action – and conceptual metaphors – which frame cognitive assemblages of thought, expression and action – the discursive politics of metaphors can be brought into focus. Discourse metaphors frame and organize shared narratives (be it in the form of public opinion, political agendas, research programs, or world views); they are embedded in discursive formations and networks of power; and they are constitutive of certain views of the world, of society, and of how things work. In a study of metaphors used during the outbreak of FMD in Britain in 2001, Nerlich, Hamilton, and Rowe (2002) examined conceptual metaphors as part of stereotypical narratives of war, contest, journey or plague and as used in conjunction with potent photographic images, e.g. of burning pyres. We showed how traditional metaphors of war and battle assumed a distinctive social relevance as part of political narratives. In a recent article, the linguist Musolff has explored the use of metaphors in a different type of public discourse (Musolff 2006). Like the proponents of discourse metaphors, he has shifted his focus from studying conceptual metaphors to examining what he calls metaphor scenarios that organize source concepts into mini-narratives. Musolff characterizes a “scenario” as a set of assumptions made by competent members of a discourse community about ‘typical’ aspects of a source-situation [e.g. marriage], for example, its participants and their roles, the ‘dramatic’ storylines and outcomes, and conventional evaluations of whether they count as successful or unsuccessful, normal or abnormal, permissible or illegitimate, etc. (Musolff 2006: 28)4

Metaphor scenarios are cognitively, emotionally and politically important, as they enable the speakers to not only apply source to target concepts but to draw on them to build narrative frames for the conceptualisation and assessment of sociopolitical issues and to ‘spin out’ these narratives into emergent discourse traditions that are characteristic of their respective community. (Musolff 2006: 36)

Musolff studied scenarios, such as MARRIAGE (discursive source), that structured journalistic discourses about EUROPE (discursive target) in Germany and the UK. This chapter uses the concepts of discourse metaphor and metaphor scenario to analyze media discourses about FMD and avian flu in the UK.

118 Brigitte Nerlich The media coverage of FMD and avian flu is partially concerned with debating the risk of disease outbreaks to animals, humans and the economy. As Adams has pointed out in his seminal book about Risk: Risk is all in the mind. It is a word that refers to a future that exists only in the imagination. … We imagine the future, arriving at the subjective probabilities of both positive and negative outcomes, by using instinct, intuition and our experience of circumstances that appear similar to those that we have encountered before. (Adams 2005: 36)

This chapter investigates how the risks carried by two types of diseases are framed by metaphor scenarios that are rooted in instinct, intuition and our experience of circumstances that appear similar to those that we have encountered before: war, journey and house. The war scenario is in some sense a super-scenario that subsumes other scenarios or mini-narratives, which themselves can be connected to other scenarios and form discursive metaphor clusters. Koller (2003a, 2003b) has explored the war metaphor scenario extensively. I only want to list some of its sub-components and briefly show how they can be linked to other scenarios and conceptual metaphors relevant to the subsequent analysis. The concept of WAR is associated with concepts of BATTLE, FIGHT, ATTACK, DEFENCE, KILLER, ENEMY, FRONTLINE and FORTRESS amongst many others. Each of the sub-scenarios involved can provide food for metaphors that exploit aspects of such scenarios or narratives, as will be shown in case of the use of the frontline scenario. In case of disease the most stable metaphors based on war seem to be DISEASE IS A WAR/INVASION and TRYING TO CONTROL DISEASE IS WAR (on war metaphors see e.g. Hodgkin 1985; Warren 1991; Kövecses 2000; Sarasin 2003, 2004). These metaphors construct disease as an attack by foreign bodies that have to be destroyed. Apart from war two other scenarios were salient in the press coverage of FMD and avian flu: The journey scenario and the house scenario. In the context of viral spread and viral infection this generated discourse metaphors such as THE SPREAD OF A VIRUS IS A JOURNEY (see “Bird flu journey” http://news.bbc.co.uk/1/shared/spl/hi/world/05/bird_flu_map/html/1.stm = website about the spread of the virus) and THE APPEARANCE OF A VIRUS IS AN INVASION. The house scenario as a source domain also framed various other discourse metaphors and entailments, depending on the target domain onto which it was mapped. The house scenario can, for example, be mapped onto an abstract entity: THE NATION IS A HOUSE (and has to be protected against invaders, real or metaphorically framed ones); or onto a concrete entity: THE BODY IS A HOUSE (and has to be protected against invaders

Metaphor scenarios in disease management

119

[in this case invaders can only be of the metaphorically framed sort, e.g. viruses]); or onto a different type of concrete entity: THE FARM PREMISES ARE A HOUSE (and have to be protected against invaders, real or metaphorical) (in this case the metaphoricity is reduced as the virus is seen as entering not only the premises but also real houses or buildings, be they hen houses, cattle sheds or the farm house itself). Both the house scenario and the journey scenario are based on more basic conceptual metaphors or image schemata: The container metaphor and the source-path-goal schema respectively. Both can become negatively connotated in the context of war when an “enemy” (the virus) is seen as “invading” the container (nation, body, farm) which again triggers various other war scenarios of attack and defence. Under these circumstances the house can also become conceptualized as a fortress. The house scenario or frame itself has sub-components, such as the gate leading to the house, the door that leads into the house, which can become metaphorically active (letting the virus in, etc.) and can attract negative connotations in the context of the war scenario (entering a house becomes an invasion), or even the doorstep (which can be crossed). And finally, in the context of disease, especially human disease, the war or battle scenario is well-entrenched (see Montgomery 1991; Gwyn 2002) and can be superimposed onto the fight that a nation or farm fights against a virus. The intersection of these various scenarios, image schemas and conceptual metaphors can be represented as follows: remoteness

proximity

source

path

goal

Virus = agent

Container/ House/fortress/ nation/body Journey/invasion

containment

war

Figure 1. Virus on the march – source-path-goal schema

Let us now examine which parts of this metaphor field were activated in what way in the UK press coverage of FMD and avian flu respectively and what disease management strategies or policy options they reflect, suggest or render plausible.

120 Brigitte Nerlich

3. Foot and mouth disease FMD is a highly infectious disease affecting cloven-hoofed animals, in particular cattle, sheep, pigs, goats, and deer. While FMD is not normally fatal to adult animals, it is debilitating and damages productivity, reducing milk yields and causing lameness. In the UK, 2,026 farms were infected in Britain and four in Northern Ireland between February and September 2001, with infections peaking at the end of March (see Royal Society report 2002). This was the worst outbreak of FMD in UK history. It is thought that the virus was introduced in early February 2001 when waste meat products (probably illegally imported airline food) were mixed into pigswill. The rapid spread of the virus was attributed to the intensification of international and domestic animal movements and the closure of local abattoirs. FMD was thus seen, almost immediately, as a product of the global economy and the global movement of infectious organisms. FMD is a virus that infects animals world-wide and that can travel on global trade routes of meat, legally and illegally. It emerged in the UK unexpectedly after a more than 30-year absence. It was unclear from where it had travelled, that is, there was uncertainty about its source, but once it had arrived/invaded, it travelled quickly inside the country from farm to farm. Again it was uncertain, however, how the virus travelled, on the tyres of vehicles, on the wind, carried by wild birds, etc. Farmers blamed the government for “letting the virus” into the UK (THE NATION IS A HOUSE) in the first place, whereas the government blamed farmers for not doing enough to keep the virus out of their farms (PREMISES ARE A HOUSE; ANIMAL BODIES ARE A HOUSE) (see Donaldson and Wood 2004). Using the source-path-goal image schema, one can say that both the global source and global path of the FMD virus were uncertain, and once it had reached its goal, the UK, the internal paths or transmission routes it used were relatively uncertain as well. Given the certainty that the virus was here, i.e. had reached its target from whatever source, and given the uncertainties about its spread, government fell back on what seemed to be a relatively certain, i.e. historically entrenched policy option (slaughter, culling or killing the virus hosts and suspected virus hosts), which came with a well-established metaphor scenario: war. To deal with FMD, the UK government implemented a slaughter policy, a war-like response well-entrenched in British and European legislation (Woods 2004), which involved not only rural movement restrictions, airport food checks, and stringent biosecurity measures, including cleansing,

Metaphor scenarios in disease management

121

hygiene and disinfectant mats, but also the mobilization of the armed forces in support of civil power. As the FMD virus “invaded” farms and spread through the national herd, biosecurity became an all-important watchword, more commonly referred to as trying to keep the virus out of the farm (Nerlich and Wright 2006). Farmers spent immense amounts of time and money placing disinfectant mats across access roads, and cleaning possible virus hosts. The effect of these measures is debatable, with even the then Agriculture Minister describing the disinfectant matting as “more symbolic than real” (Hansard 2001). Nonetheless, they allowed farmers at least some sense of tangible control over the virus, alleviating feelings of helplessness and insecurity. Biosecurity became part of “fortress farming” adopted to keep the FMD virus away from farms. The phrase “fortress farming” and its cognates (fortress mentality, fortress attitude, fortress policy, fortress measure),5 together with others such as “siege mentality”, allowed farmers and politicians to conceptualize the fight against the FMD virus as one against a foreign invader or, as John Law has pointed out, an incoming flood against which barriers had to be erected (Law 2004). In a study devoted to cultural aspects of biosecurity Samantha Twigg Johnson has argued that the purgative zeal of biosecurity was energized by broader cultural anxieties surrounding the integrity of the British nation, and the capacity of the state to protect it in an age of globalization (Twigg Johnson 2009). Success in resisting foreign invasion has long been a prominent aspect in British national identity, one still constantly reinforced by cultural references to war stories and the ritual retelling and commemoration of war narratives. As William Shakespeare wrote, evoking various parts of the metaphor scenario or mini-narrative explored in this paper: This fortress built by Nature for herself Against infection and the hand of war, This happy breed of men, this little world, This precious stone set in the silver sea, Which serves it in the office of a wall, Or as a moat defensive to a house, Against the envy of less happier lands, This blessed plot, this earth, this realm, this England, This nurse, this teeming womb of royal kings. (Richard II, 2.1.32–48)

122 Brigitte Nerlich Regional heritages sometimes sharpen this view of “England” through a focus on the castles that dot the countryside. For example, The Sun described how, during the FMD outbreak, the “gentrified locals” of “Blair Castle” were “bolting down the hatches again” (i.e. closing the door to the virus) at the “scene of the last siege in British history – back in 1745” (The Sun, 05/04/01). Framing the control of FMD in this culturally salient way was important, as, throughout the FMD epidemic, scientific and political uncertainty reigned. Scientists were unsure about the origins and mode of transmission of the virus, and debated endlessly whether vaccination would work better than culling. Policy makers had no real contingency plans and often pursued confused and confusing measures. In such a situation, “the military metaphor is one way to galvanize the society and to marshal its resources for an effective counterattack” (Childress 1991: 71; cf. Childress 2001: 191). By comparison, there seems to be a scientific and media consensus that the origins of the avian flu virus are relatively well-known as well its routes of transmission (however, in principle, a mutation of the bird flu virus could originate in a different and unpredictable “place”). What is uncertain in this case is the risk to humans, as it is impossible to predict when the bird flu virus will mutate and turn bird flu into pandemic flu and what response to adopt in that event, as vaccines will not be readily available. Compared to FMD a policy response to both a spread of bird flu to Europe and the US as well as the mutation of the bird flu virus into a human flu virus, cannot be “taken off the shelf”, and a response needs to make sense globally and not only locally or nationally. This might be the reason why military metaphors are as yet relatively few and far between, indicative of the uncertainty about what to do about a human pandemic “in waiting”. As will be shown below, when military metaphors were used in reports on avian flu, they were mainly used with reference to Far Eastern countries where bird flu is endemic in the domestic bird population,6 in this similar to the UK reporting on SARS (see Larson, Nerlich and, Wallis 2005; Wallis and Nerlich 2005). Different uncertainties seem to call for different metaphorical framings, a topic that might need further exploration. FMD was, by contrast, an animal epidemic that was there, that had been “fought” before and that could be fought again. It had invaded the UK without anybody knowing, had spread silently, before eventually spreading with quite an explosive force. In this context it was much easier to conceptualize the handling of the disease first and foremost as a war, fight, or

Metaphor scenarios in disease management

123

combat with an invisible and treacherous enemy, the virus.7 A few examples will suffice to illustrate the rhetoric based on various war scenarios and mini-narratives based on a virus/enemy that travels along a route: (1)

a. b.

c.

d.

The army has compared the disease to the enemy in a war. There is a battle to stay ahead. (The Guardian, 30/03/01) But is a prime minister who presents himself as the commander-in-chief of a country at war with an enemy as threatening as foot-and-mouth disease, able to devote himself to an election campaign? (The Independent, 30/03/01) We can follow the route of foot and mouth on a map, just like we follow the enemy in a war. (The Daily Telegraph, 29/03/01) We need to find its route and follow it like we are seeking out opposing forces. (The Daily Telegraph, 29/03/01).

In this war, as well as a combatant, the virus was variously personified as a criminal, a supernatural enemy, such as the Grim Reaper, or one of the horsemen of the Apocalypse. Conversely, veterinarians and farmers became soldiers. However, in framing the virus as an enemy in a war, the animals, the actual victims of the disease, came to be recast as enemies and villains alongside the virus. Even as they were killed in their millions, it was possible not to see the animals “as” the real victims (Stibbe 2002; see also Tovey 2003). This is even worse for poultry in the case of avian flu, as they are almost invisible until they land on your plate. Linked to the more visible metaphors of FMD as an enemy in a war were other types of more hidden but very pervasive fields of metaphoricity that ran through the FMD discourse: FMD was engaged in a hand-to-hand combat with farmers and officials (FMD hits, strikes, deals a blow, a double whammy); FMD was a killer and a criminal – two ways to conceptualize a disease or invader that were also prominent in some discourses about SARS and are now emerging in the avian flu media coverage; and FMD was seen as being on a long journey (FMD goes, marches, travels, reappears). This last metaphor scenario of a journey also structures the way avian flu is reported as marching across the globe. In the UK avian flu discourse it is at present still more dominant than the war scenario, for reasons to be explored later. Although FMD is a global problem, the response to FMD in the UK in 2001 was a national, and to some extent nationalistic, one. FMD does not

124 Brigitte Nerlich pose a serious risk to human health and is rarely fatal for animals. Its threat is largely economic: That of loosing “FMD-free status”. Against this background the role of war discourse in this epidemic was highly traditional and appeared to be the natural and most plausible policy response. The immense “collateral” damage this entailed for the farming and non-farming economy, rural communities, and animal and human welfare, has arguably changed this situation. Avian flu, by contrast, is seen predominantly as a global problem, just like SARS had been in 2003, and poses direct risks to human health. Managing avian flu might therefore provoke different metaphorical responses or different uses of familiar ones, as well as different policy responses.

4. Avian flu Lately, fear of disease has been fuelled yet again by the emergence of a new highly pathogenic virus strain of avian influenza that could jump the species barrier between birds and humans, and, in a worst-case scenario, cause the next global pandemic of influenza. Although there had been an influenza pandemic before the 20th century, such as the “Asiatic flu” of 1889/1890, the popular story of global influenza pandemics, based on the first well-documented event, began in 1918 when 50 million people were infected world-wide with an influenza strain subsequently identified as avian in origin and recently reconstructed in an American laboratory (Tumpey et al. 2005). The 1918 pandemic (“Spanish flu”) was followed by two other 20th-century pandemics in 1957 (“Asian flu”) and 1968 (“Hong Kong flu”), but 1918 seems to serve as an “index” date in news coverage of existing and emerging epidemics and pandemics. The story of a new 21st-century virus that could lead to a new pandemic began as early as 1997 when bird flu broke out in Hong Kong and infected humans. This was a new strain of avian flu, the highly pathogenic H5N1 strain. 18 people were infected and six people died; all had had close contact with chickens. Chicken exports were banned, about 1.6 million chickens were slaughtered and the virus seemed to have been eradicated. The UK government responded with a contingency plan, but nobody took any real notice. This changed after 2003 (which was also the year of the SARS outbreak) when another outbreak of avian flu, this time the H7N7 strain, occurred in the Netherlands, which killed a veterinarian who had visited a poultry farm. At the same time the H5N1 strain re-emerged and devastated

Metaphor scenarios in disease management

125

a commercial poultry farm in South Korea. In 2004, outbreaks of H5N1 occurred in Thailand, Vietnam, Cambodia, Indonesia and China, where pigs and wild geese were also infected. So far about 170 humans have been infected and about half of them have died. In the autumn of 2005, the virus spread from East Asia to Russia and to Eastern Europe, probably carried by migrating birds. A consignment of finches infected with H5N1 arrived in Britain and a parrot died in a quarantine centre. In 2006, it spread to Italy, Greece and Bulgaria, but also to Nigeria in Africa and also to Germany and France. In February 2007, it arrived in Hungary and then on a poultry farm in the UK but was quickly brought under control (see Nerlich, Wright, and Brown forthcoming). Although known to affect humans since 1997, the danger arising from the H5N1 virus was only picked up by the media at the end of 2004 when the World Health Organization (WHO) engaged in a concerted campaign of “early warnings” that fuelled expectations of an impending pandemic. The warnings began to be heard in early 2005 when governments around the world started to prepare for a possible bird flu pandemic. The UK government published a first version of a pandemic preparedness plan in March 2005, which triggered a first wave of media interest, then a full version in October 2005, a time when the media went into something like a bird flu panic. By coincidence, this was the month when a parrot in a UK quarantine centre died of H5N1, a quarantine centre that was located quite close to the abattoir where the first case of FMD had been detected – or so some newspapers speculated (see Rose and Elliot 2005). In a recent article (Nerlich and Halliday 2007) we tried to uncover certain patterns of scientific and media discourses about avian flu in early 2005, a time when bird flu had still been seen as a remote illness. These patterns clustered around scientific and expert discourses of “early warnings”, which, linguistically amplified in certain ways, contributed to a “rhetoric of fear” in UK newspapers and a governmental discourse of “wait and see”. This in turn sparked a “rhetoric of blame” when the government was accused of not doing enough to prevent the virus from reaching the UK or not doing enough in the way of stockpiling antiviral drugs. The two discourses, by scientists and the government, raised quite different expectations, one of urgent action and one of caution, a clash that might have contributed to a general “discourse of uncertainty”, quite similar to the situation during the FMD outbreak. In the case of FMD, some semblance of certainty had been imposed on the situation by the ritualistic adherence to biosecurity measures, such as the use of disinfectant mats, a situation that

126 Brigitte Nerlich seems to be partly mirrored by the stockpiling of antiviral drugs in the case of avian flu. In late 2005, early warnings about bird flu were replaced by more urgent warnings that the virus was now closing in on Europe. What had, until then, been a far-flung illness became a more national and individual concern. The “discourse of fear” was complemented by a growing “discourse of war and containment” which had been relatively absent in early 2005. As shown in Nerlich and Halliday (2007), the initial media coverage of a possible pandemic was not couched in terms of war and plague, but was metaphorically framed in terms of natural disaster metaphors that reflected a view of an illness that was still remote but could come closer any time soon. The virus was conceptualized as an impending flood (similar to FMD, see Law 2004), a storm or a volcano – and this by scientists, not the media. However, as the virus spread/travelled from East Asia to the outskirts of Europe, and the threat level increased, war metaphors began to creep into the language used by experts and the media. In various interviews the WHO Regional Director for the Western Pacific Dr. Shigeru Omi has, for example, exploited the war metaphor. On 4 July 2005,8 he pointed out that “We must have an all-out war against this virus”. On 26 October 2005, at the height of the media coverage about avian flu in the UK, he proclaimed that “Asia remains ‘ground zero’ in the war on avian influenza” (http://www.euro.who.int/mediacentre/PR/2005/20051027_1). To investigate this shift in the perception of the threat from a remote natural disaster to an impending war “on our doorstep” (which increased its newsworthiness quite substantially), a number of searches on the Lexis Nexis data base of articles published in UK national newspapers between September and December 2005 were conducted and compared with coverage earlier in the year. This was accompanied by a more in-depth qualitative analysis of a smaller sample of avian flu coverage in October 2005. As about 350 articles were published that month on avian flu in the UK national newspapers alone, only articles published by one tabloid and one broadsheet with the highest output over a ten-day period (17 to 26 October) were sampled. Uses of words such as war, fight, battle did not increase as dramatically as expected between early 2005 and late 2005.9 However, there was a slight surge in battle metaphors in late 2005 especially at the end of October and the beginning of November when there was increased activity amongst governments and global organizations, such as the WHO, the United Nations, the Food and Agriculture Organization of the United Nations and the

Metaphor scenarios in disease management

127

World Organization for Animal Health to draw up what one might call bird flu battle plans. There were however two significant shifts in the reporting on avian flu between early 2005 and late 2005, which coincide with the virus travelling from East to West and closing in on Europe. From September 2005 onwards, the tabloids began to use the term killer flu or killer virus, which had also been their favourite during the SARS outbreak (see Wallis and Nerlich 2005)10, whereas the broadsheets began to use the word frontline with increased frequency.11 A closer inspection reveals that the metaphor scenario of war, of which frontline forms a sub-part, can shape various representations depending on what part of the frontline scenario or frontline mini-narrative is highlighted: Its geographical position, the people that fight on it or the weapons used. The frontline can either be visualized geographically and in relatively abstract terms as being drawn in the Far East, the source of the virus, or in the UK, the target of the virus; or else the focus can be on the people working on the frontline (the “soldiers”), either individually or metonymically as parts or larger organizations; or else a certain drug can be metaphorically evoked as a frontline “weapon” – in this case tapping into the metaphor scenarios of illness and invasion, where the body is conceptualized as being invaded by pathogens and medicines as defence mechanisms to ward off invaders. It is then only one more step to conceptualizing drug or vaccine manufacturers as constituting the frontline in the war against the virus. Similar uses of the frontline scenario were ubiquitous during the FMD crisis. Examples of the various perspectivizations of the frontline scenario in the broadsheet coverage of avian flu in 2005 are: Geography: (2) a.

f.

Asian countries on the front lines of the battle against bird flu Asian countries on the front lines of the battle to prevent … Asian countries on the front lines of the disease Remote front line in the war on bird flu Pang Tru has become the front line of the battle against the disease … on a tour of the bird flu front lines

a.

Back on the front line in Gloucestershire

b. c. d. e.

(3)

128 Brigitte Nerlich b. c. d. People: (4) a. b. c. d. e. f.

[farmers] are expected to be on the front line because of the wild geese and ducks … … poultry farms would be in the front line of an outbreak Birdwatchers join the front line

… jabs, which were restricted to front-line emergency staff Front-line NHS workers … GPs would be on the frontline … waiting game with its own front-line health staff … the trust is on the front line of the battle against the avian … United Nations organisations in the frontline of action against avian …

Weapon: (5) Stockpile Tamiflu as a front line of defence against … Weapon manufacturer: (6) Remote front line in the war on bird flu: In a small laboratory in a Budapest suburb, scientists are developing a vaccine which could prevent a global pandemic ... unlikely front line This exploitation of frontline scenarios by the broadsheets provides a more nuanced understanding of the dangers posed by avian flu than the rather clichéd use of the killer metaphor by the tabloids. After this overview, a small sample of UK national press coverage was examined in more detail, that is, articles published by the Daily Mail and The Times between 17 and 26 October 2005, a time when the virus was reaching Europe and then the UK and when the final version of the UK contingency plan was published. Using the search term bird flu 43 articles published by the Daily Mail and 40 articles published by The Times were analyzed. Comparing the metaphors used in the two newspapers revealed a striking difference, which is not entirely unexpected though. Whereas the Daily Mail exploited a network of metaphor scenarios which will be studied below, The Times coverage was almost devoid of metaphors – with one major exception. Professor John Oxford, a famous virologist and what one could call “pandemic pundit” who was seldom out of the news at that time, published an article on the comments pages of The Times which exploited

Metaphor scenarios in disease management

129

the war metaphor scenario, as well as the journey and house ones, quite skillfully. He writes that we are accelerating “towards the opening shots of the first global influenza outbreak of the 21st century”, that “a new bird virus, H5N1, is the enemy at the gate”, and that: At the moment, we have plans galore but not much in the way of ammunition. Yet any investment now in influenza antivirals and vaccines will not be wasted. We deserve a properly planned war against our old enemy, the flu. (20/10/05)

In the following section the use of metaphor scenarios by the tabloid the Daily Mail will be examined, which overlaps with those tapped into by John Oxford. Before doing so, it might however be useful to reflect briefly on the status of Tamiflu in the “fight” against avian flu. Tamiflu is an anti-viral drug that can reduce the severity of flu symptoms. It is believed to be the only available treatment for avian flu for which no vaccine exists as yet. Some experts surmise that the use of Tamiflu could buy time so that vaccines targeted against any particular strain of flu, mutated from avian viruses, could be developed. As shown elsewhere (Nerlich and Wright 2006), during the FMD outbreak visible biosecurity measures, such as putting down disinfectant mats and putting up “Keep out” signs, were not only used rationally to prevent the spread of the virus, but they were also used as physical evidence of efforts to deflect blame. Furthermore, the demarcation of boundaries was offered as evidence of “doing one’s bit”, but it might also have been conceived as a magical “cordon sanitaire”. Visible, physical biosecurity actions became ways to deflect blame and were self-evaluated by farmers as reducing guilt in terms of “how could we live with ourselves if we had done nothing and caught it”. Stockpiling Tamiflu by individuals as well as governments may have similar functions, some of which might also be, as in the case of disinfectant mats, “more symbolic than real”. Mental images of piles of Tamiflu might evoke an almost magical and “visible” line of defence against an invisible enemy. Demonstrating that Tamiflu is being stockpiled might also serve governments as “physical” evidence that they are “doing” their bit in a situation of uncertainty and reduce blame by self or others. Paradoxically, during the FMD crisis biosecurity control measures (including visible symbols of these measures, such as “Keep out” signs and disinfectant mats) became, in many ways, the FMD crisis. Similarly, visualizing the various heights of Tamiflu stockpiles on pharmacy shelves or in government stock rooms has become part of the pandemic panic. Let us now look at how other “lines of defence” against an

130 Brigitte Nerlich invisible enemy were conceptualized in the tabloids, in this instance, the Daily Mail. The first article in our sample starts with the sentence “Bird flu’s deadly knock is at our very door” (Daily Mail, 17/10/05). This echoes John Oxford’s “enemy at the gate” metaphor and also corresponds to the journey scenario used in a statement in The Times a day later that “death comes from the East in a flapping of wings” (The Times, 18/10/05). However, unlike in the tabloid, the broadsheet articles, sampled for ten days, never seriously exploited the networks of metaphor scenarios that are indexed by these linguistic expressions, namely the war, journey and house scenarios. The metaphor scenario evoked by a virus on a journey is linked to the image schema designated as source-path-goal (see Lakoff and Johnson 1980), onto which the personification of the virus as a moving agent is superimposed, and it is linked to the discourse metaphor THE NATION (THE UK) IS A HOUSE. The source-path-goal schema, which was also exploited in conceptualizations of the spread of FMD (see Figure 1), has the following sub-components that map neatly onto the spread of the bird flu virus as the trajector: – – – – – – – – – –

A trajector that moves A source location (the starting point) A target or goal, an intended destination of the trajector A route from the source to the target The actual trajectory of motion The position of the trajector at a given time The direction of the trajector at that time The position of the trajector at a given time The actual final location of the trajector, which may or may not be the intended destination

The source in this instance is the Far East where the virus, H5N1, originates. The path is the spread of the virus from East to West, carried by real agents, such as people trading/moving animals (poultry or wild birds) or migratory/moving birds (which follow well-known routes). This path leads the virus ever closer to the goal that is the UK. As the UK as a nation is conceptualized as a house, the virus is seen as approaching the house and knocking on the door or as approaching the gate. Defence against the virus is conceptualized as stopping it from travelling, i.e. escaping from the source (containment) or, once it has started to travel, stopping it from com-

Metaphor scenarios in disease management

131

ing through the door or gate, keeping it out (which reminds one vividly of the “Keep out” signs displayed at farm gates during the FMD crisis – and which reappeared quickly when a swan infected with H5N1 was discovered in Scotland in the spring of 2006 and when poultry were infected in Suffolk in 2007, i.e. preventing it from reaching its goal by not keeping the door open or closing holes in it). This type of defence against a virus can then be couched in terms of war, fight or battle, in which case the house becomes, as in the case of FMD, a fortress. Some of the linguistic expressions that instantiate this network of metaphor scenarios are: (7)

a. b.

c. d. e. f. g.

Virus reached the edge of Europe. (Daily Mail, 17/10/05) The Ministers stark warning [Sir Liam Donaldson saying that 50,000 people if not more will die; BN] came as scientists revealed the disease continued its march across the globe. (Daily Mail, 18/10/05) Coordinating Britain’s battle to stop the march of the virus. (Daily Mail, 19/10/05) Is this [pet markets] how bird flu will get through? (Daily Mail, 22/10/05) – it turned out that it did! Killer strain has reached Britain. (Daily Mail, 24/10/05) Wide open to deadly threat. (Daily Mail, 25/10/05) Holes in our defence. (Daily Mail, 25/10/05)

The frontline metaphor from the metaphorical war against a disease scenario is in this context superimposed upon the threshold/door metaphor: (8)

a.

b.

Frontline of the birdflu war … a rickety row of sheds next to the abattoir where foot-and-mouth began. (Daily Mail, 25/10/05) Sniffer dogs have gone on duty at Britain’s airports as the first line of defence against bird flu. (Daily Mail, 19/10/05)

Such lines of defence against a possible risk (which are all imaginary in a way) can be multiplied; there may be a second line of defence (for example Tamiflu), a third line of defence (for example shutting schools) and so on. The institutions defending these lines or barriers can be conceptualized as a “prime bulwark”, albeit ironically. Here the experience of the past, be it the distant past when real bulwarks still worked or the more recent past, the handling of FMD, when metaphorical ones didn’t, colours how we perceive

132 Brigitte Nerlich the future, that is the handling of a possible flu pandemic. This is especially important since an outbreak of bird flu in the UK would involve applying a similar stamping out policy as the one adopted during the FMD outbreak which created so much collateral damage. Incredible as it may seem, the prime bulwark to prevent the virus taking hold are from the selfsame cackhanded crew the Chief Vet’s outfit at the Department of the Environment, Food and Rural Affairs that brought us the tragic fiascos of BSE and foot and mouth disease. (Daily Mail, 17/10/05)

The house/door metaphor used in the Daily Mail found a more sinister expression in a book published in November 2005, entitled The Monster at our Door (Davis 2005). As a science journalist wrote in his review of Mike Davis’s book: “Is the threat of avian flu ‘the monster at our door’ or, to use an alternative and zoologically mixed metaphor, a frightened chicken crying wolf?” (Watts 2005; see also Lean 2006; Ferguson 2006). The latter view still seems to be prevailing even after H5N1 arrived in Britain.

5. Conclusion and comparison Three metaphor scenarios seem to structure the media coverage of FMD and avian flu in various ways between 2001 and early 2007: The journey/invasion scenario, the war scenario and the house scenario. In the case of avian flu the journey scenario still dominates press coverage, as, at the time of drafting this article, the virus had not yet reached/invaded either the UK poultry population or the UK human population and neither had it mutated into a form in which it can easily be transmitted between humans. Bird flu is still a pandemic in waiting. This is comparable to the situation of individual farms during the FMD crisis when farmers were watching, waiting, mapping and monitoring the virus to see when it would finally reach their farm. In the case of the FMD crisis the virus had travelled undetected to the UK and got in “through the back door”, so to speak, before anybody noticed it. Once inside the UK it then travelled in various ways from county to county and farm to farm. So the travel scenario was still important, but the war scenario had more relevance, as the UK had to fight an invader on its territory. To protect against invasion by a travelling virus, be it of the nation or the farm, barriers and fortifications have to be erected to make the nation, the farm, the animals and the humans safe. In the case of FMD such barriers were disinfectant mats, etc.; in the case of avian flu such barriers

Metaphor scenarios in disease management

133

might be the stockpiling of anti-viral drugs. However, in the case of pandemic flu various other policy options are still open, from monitoring to eliminating or containing the virus at source, whereas in the case of FMD where the virus had already reached its goal, these options were no longer available. War seemed the natural option to choose over vaccination for example, especially as one was dealing with “dispensable” animals and not humans and the aim was to safeguard an industry not individuals. Using the source-path-goal schema (see Figure 1), our analysis seems to suggest that the further along the path a virus travels and the closer it gets to its goal, the more war metaphors one might find in the media coverage. It also seems to be the case that there is a difference in metaphorical framing and the framing of disease management options that depends on whether the crisis is seen as a national or global event. Once the virus has reached its goal, the nationalistic invasion scenario seems to be triggered and war metaphors become an even more natural or instinctive way of talking about disease management. Once the virus is in a country the travel scenario too changes, as the virus can now be conceptualized as adopting a wider variety of modes of travel, as illustrated by the FMD virus that not only marches, advances and reaches, but also speeds up, slows down, jumps, appears, disappears and so on. War metaphors also seem to be more prevalent in a situation, like the FMD crisis, where the source of the virus is unknown, were its trajectory before arriving in the UK is unknown and where its modes of travelling are unknown. By contrast, in the case of avian flu we know its source, we know its trajectory and we know who can carry it, i.e. migratory birds, traded poultry and traded wild birds, how it is transmitted between birds and humans (through close contact) – until the time when it mutates and can be transmitted between people and across the globe. Once this happens metaphor scenarios around infection as invasion will probably become more prevalent in media language and the war against the virus will happen on a more individual and corporeal level. Just after this paragraph was written cases of bird flu in humans exploded in Turkey (beginning of January 2006). As in East Asia, people were infected via direct contact with infected birds. 18 people fell ill and three children died (so far, 14/01/06). This brought bird flu very close to other countries in Western Europe and alarmed Turkish politicians, health officials and citizens, as this might not only have implications for individual health but also for the tourist industry. A quick sample of Turkish newspaper articles around 13 January 2006 revealed that war metaphors

134 Brigitte Nerlich were almost everywhere. One headline from 13 January (accompanied by a cannon used to scare migrating birds away) proclaimed that war had been declared against bird flu with weapons (“Kuú gribiyle toplu-tüfekli savaú!” see http://www.radikal.com.tr/haber.php?haberno=166792). The article went on to say for example: (9)

a. b. c. d. e.

f.

This battle cannot be won with this kind of organization. The soldiers of the organization that will conduct the war are the veterinarians. Is it possible to win a war with such a small number of soldiers? How will we win this war in this downbeat mood? Senior veterinarians are the commanders of this war whereas institutes are the ‘high command’ [base from which you conduct the war]. This war requires a budget.

In a letter from the president of the veterinary association of Turkey from 8 January 2006 one can read that “struggling with (or challenging) this disease means declaring a war”; “the high command and the organization of the war is the Ministry of Farming and Villages” (see http://www.zaman.com.tr/?hn=244945&bl=yorumlar&trh=20060108). One could even find a headline on the Turkish version of Sky News, declaring on 12 January 2006 that: “Bad weather conditions make it harder to wage war against bird flu” (http://www.skyturk.tv/h_49952_1.htlm).12 This seems to confirm the hypothesis that war metaphors predominate in a situation in which a threat from disease is perceived as a direct threat to national space, individual health and the economy at large.13 However, when H5N1 reached the UK not in the form of one diseased parrot (October 2005) or one diseased swan (April 2006), but killed 2000 chicken at a Bernard Matthew’s “factory” in February 2007 something quite remarkable happened: The press did not go overboard with headlines of doom and apocalypse. Why this was not the case is still a small mystery. The immediate “reassurance” offensive by the government and the poultry industry might have something to do with this non-panic, which included a discourse that can be summarized as: “We here in Britain are made of sterner stuff than those panicking people abroad – look at what happened in Italy in 2005 when everybody stopped eating chicken – how silly.” Others

Metaphor scenarios in disease management

135

have speculated that once the unknown became a known, panic was reduced: That fear of bird flu was greater when it was a distant threat, compared with now when the disease is right here in Britain, is revealing. It suggests the politics of fear becomes more animated by the spectre of far-away threats, by distant doom lurking beyond our borders; indeed, it seems it was the very unpredictable, unknowable nature of bird flu in 2005, the fact that its movement could not be closely monitored or its potential impact on our societies measured, that made it a natural focus for the free-floating culture of fear. (O’Neill 2007)

However, this does not quite match up with what happened during the FMD crisis. It might be that flames of fear and panic burn brighter when somebody other than oneself is to blame – and that is something that still happened during the FMD crisis: In 2005, the bird flu panic was also fanned by prejudices about teeming cities in the Third World, which are apparently hotbeds of disease and a threat to stability in the West. One commentator said the potential for a human pandemic was heightened by the fact that ‘throughout the Third World, impoverished human beings have been gathering in far greater urban concentrations than anything imaginable a century ago, and any of these are potential hatcheries for a pandemic’. Another argued that these ‘large concentrations of humans ... increase the speed of evolution of viruses’. Mirroring the debate about terrorism, the 2005 great bird flu panic was driven in part by fear and suspicion of Johnny Foreigner – and by barely concealed prejudice against industrialisation in the Third World. This wasn’t only an immeasurable disease; it was also from ‘over there’, yet another toxic threat to Western civilisation from that ‘hatchery’ of disease that is the developing world. (O’Neill 2007)

As Musolff (2006: 36) has pointed out, metaphor scenarios “enable the speakers to not only apply source to target concepts but to draw on them to build narrative frames for the conceptualization and assessment of sociopolitical issues”. Such scenarios provide important narrative and discursive framing devices for journalists, but, as demonstrated in this chapter, these discursive frames are also used by scientists and policy makers – and sometimes the same scenarios are exploited by experts and (tabloid) journalists alike, sometimes the same scenario is used in different ways. As Donald Schön (1979) has pointed out, metaphors construct or frame views of “reality”, which can be used in policy-making and planning. Although these views of reality may be framed by similar metaphor scenarios (war, jour-

136 Brigitte Nerlich ney, house), the policy scenarios that result from these framings will differ depending, in our two cases, on the perceived proximity of the disease threat as well as the perception of the threat as global or local/national. In both cases, however, fundamental metaphor scenarios, based on culturally or phenomenologically salient objects or experiences, such as houses and journeys, allow the media, scientists and policy-makers to reduce the complexity of the threat posed by a disease and to engage in disease management strategies or risk management strategies that appear to make instinctive or intuitive sense to experts and the public.

Acknowledgement Work for this article was supported by a grant from the Economic and Social Research Council entitled “Talking Cleanliness in Health and Agriculture” (grant number: RES000231306). I would like to thank Robert Dingwall for his, as ever, insightful comments on a first draft of this paper and Ilke Ozdemir for her help with translating examples from the Turkish press coverage of bird flu.

Notes 1.

2.

3.

4.

The paper will not take into account developments in August 2007 when foot and mouth disease re-emerged in England. This outbreak, caused by a biosecurity breach and leak of the virus at a government laboratory, was however very different from the 2001 outbreak (see Nerlich, Brown, and Wright 2009). It is important to distinguish between three different phenomena to which the label “bird flu” can be applied: (1) avian influenza in birds, (2) avian influenza in people, and (3) pandemic influenza, that is, a mutated form of an avian flu virus that has acquired the ability to spread easily between humans. To study the media coverage of FMD and avian flu in the UK we used Lexis Nexis Professional, which provides full text access to all British newspapers. It will be made clear in each instance which type of sampling provided the basis for analysis. Musolff refers to Fillmore’s type of frame analysis as an inspiration for his definition of metaphor scenarios (see Fillmore 1985). However, Goffman’s concept of frame also seems to fit very well: “I assume that definitions of a situation are built up in accordance with principles of organization which govern events … and our subjective involvement in them; frame is the word I use to refer to such of these basic elements as I am able to identify.” (Goffman 1974: 10–11)

Metaphor scenarios in disease management 5.

6. 7.

8. 9.

10.

11.

12. 13.

137

Such phrases were found 67 times in the press coverage of the FMD crisis by UK national newspapers between 22 February 2001 and 30 June 2001, most often used in Northern Ireland. A disease is endemic when it is constantly present to greater or lesser extent in a particular locality. We examined 614 articles published in The Guardian from February to July 2001. 170 contained historical allusions, metaphors, and symbolic references that were extracted “manually” (see Nerlich, Hamilton, and Rowe 2002). We also looked more closely at the use of war metaphors at the height of the crisis. We searched all UK national newspapers in Lexis Nexis database from 20 March to 30 March 30 2001, using the key word foot and mouth disease in conjunction with the key words war, battle and fight. We found 159 occurrences of war, 125 of battle and 93 of fight in this ten day sample (see Larson, Nerlich, and Wallis. 2005). See: http://society.guardian.co.uk/publichealth/story/0,11098,1521621,00. html A search for war and bird flu in UK national newspapers between February and April 2005 produced three hits, for September to November 2004 four hits; a search for fight produced 13 hits for February to April; 14 for September to November. A search for battle by contrast produced four hits between February and April and 20 hits between September and November. A search for killer flu produced no hits for February to April 2005, but 17 hits between September and November 2005, out of which 13 uses were found in the tabloids. A search for killer virus produced two hits for the period between February and April; 25 for the period between September and November, out of which 20 uses could be found in the tabloids. A search for frontline produced no hits for a search between February and April, but 22 hits for September to November, predominantly in the broadsheets. I would like to thank Ilke Ozdemir for finding these examples of war metaphors for me. See also the headline: “France on a war footing” when first cases of bird flu were found in France in February 2006 (http://news.bbc.co.uk/1/hi/world/ europe/4730156.stm). This was also the case during the SARS outbreak when China “declared war on SARS”, but war metaphors were very infrequent in UK media reporting (see Wallis and Nerlich 2005).

138 Brigitte Nerlich

References Adams, John 2005 Risk Management: Its not rocket science: Its more complicated: . Boomen, Marianne van den 2005 Networking by metaphors: . Chiang, Wen Yu 2007 Conceptual metaphors for SARS: “War” between whom? Discourse and Society 18 (5): 579–602. Childress, John F. 1991 Mandatory HIV screening and testing. In AIDS and Ethics, Frederic G. Reamer (ed.), 50–76. New York: Columbia University Press. Reprint: . Childress, John F. 2001 The war metaphor in public policy. Some moral reflections. In The Leader's Imperative: Ethics, Integrity, and Responsibility, J. Carl Ficarrotta (ed.), 181–197. West Lafayette, Indiana: Purdue University Press: . Davis, Mike 2005 The Monster at Our Door: The Global Threat of Avian Flu. New York: The New Press. Donaldson, Andrew, and David Wood 2004 Surveilling strange materialities: Categorisation in the evolving geographies of FMD biosecurity. Environment and Planning D: Society and Space 22 (3): 373–391. Ferguson, Neil 2006 Well-founded fears of some fowl times yet to come: Review of Davis, 2005. Times Higher Education Supplement, 13/01/2006: 22– 23. Fillmore, Charles 1985 Frames and the semantics of understanding. Quaderni de Semantica 6 (2): 222–254. Frank, Roslyn 2008 The language-organism-species analogy: A complex adaptive systems approach to shifting perspectives on ‘language’. In Body, Language and Mind. Vol. 2. Sociocultural Situatedness, Roslyn M. Frank, René Dirven, Tom Ziemke, and Enrique Bernárdez (eds.), 215–262. Berlin: Mouton de Gruyter.

Metaphor scenarios in disease management

139

Goffman, Erving 1974 Frame Analysis: An Essay on the Organization of Experience. New York/New York et al.: Harper & Row. Gwyn, Richard 2002 Communicating Health and Illness. London: Sage. Hansard 2001 House of Commons. Vol. 366: Column 713 (9 April), 2001. Hodgkin, Paul 1985 Medicine is war: And other medical metaphors. British Medical Journal 291: 21–28. Joffe, Hélène, and Georgina Haarhoff 2002 Representations of far-flung illnesses: The case of Ebola in Britain. Social Science and Medicine 54 (6): 955–969. Joffe, Hélène, and N. Y. Louis Lee 2004 Social representation of a food risk: The Hong Kong avian bird flu epidemic. Journal of Health Psychology 9 (4): 517–533. Koller, Veronika 2003a Metaphor Clusters in Business Media Discourse: A Social Cognition Approach. Diss. Vienna University: . Koller, Veronika 2003b Metaphor clusters, metaphor chains: Analyzing the multifunctionality of metaphors in text. metaphorik.de 5: 115–134: . Koteyko, Nelya, Brian Brown, and Paul Crawford 2008 The dead parrot and the dying swan: The role of metaphor scenarios in UK press coverage of avian flu in the UK in 2005-2006. Metaphor and Symbol 23 (4): 242–261. Kövecses, Zoltán 2000 Metaphor and Emotion: Language, Culture, and Body in Human Feeling. Cambridge: Cambridge University Press. Lakoff, George, and Mark Johnson 1980 Metaphors We Live By. Chicago: Chicago University Press. Larson, Brendan M. H., Brigitte Nerlich, and Patrick Wallis 2005 Metaphors and biorisks: The war on infectious diseases and invasive species. Science Communication 26: 243–268. Law, John 2004 Disaster in agriculture: Or foot and mouth mobilities. Published by the Centre for Science Studies, Lancaster University, Lancaster LA1 4YL, UK at . Lean, George 2006 An epidemic of complacency. Daily Mail, 14/01/2006: 26–27.

140 Brigitte Nerlich Montgomery, Scott L. 1991 Codes and combat in biomedical discourse. Science as Culture 2 (3): 341–390. Musolff, Andreas 2006 Metaphor scenarios in public discourse. Metaphor and Symbol 2 (1): 23–38. Nerlich, Brigitte 2004 Towards a cultural understanding of agriculture: The case of the “war” on foot and mouth disease. Agriculture and Human Values 21 (1): 15–25. Nerlich, Brigitte, Brian Brown, and Nick Wright 2009 The ‘ins and outs’ of biosecurity: Bird flu in East Anglia and the spatialisation of risk. Sociologica Ruralis 49 (4): 344–359. Nerlich, Brigitte, and Christopher Halliday 2007 Avian flu: The creation of expectations in the interplay between science and the media. Sociology of Health and Illness 29 (1): 46– 69. Nerlich, Brigitte, Craig Hamilton, and Victoria Rowe 2002 Conceptualising foot and mouth disease: The socio-cultural role of metaphors, frames and narratives. metaphorik.de 2: . Nerlich, Brigitte, and Nick Wright 2006 Biosecurity and insecurity: The interaction between policy and ritual during the foot and mouth crisis. Environmental Values 15 (4): 441– 462. O’Neill, Brendon 2007 A tale of two panics. Spiked online: . Rose, David, and Valerie Elliot 2005 Village at heart of bird flu scare was where foot-and-mouth began. The Times 25/10/2005: . Royal Society report 2002 Infectious diseases in livestock. . Sarasin, Philipp 2003 Infizierte Körper, kontaminierte Sprachen: Metaphern als Gegenstand der Wissenschaftsgeschichte. In Geschichtswissenschaft und Diskursanalyse, Philipp Sarasin, 191–230. Frankfurt/M.: Suhrkamp. Sarasin, Philipp 2004 Die Visualisierung des Feindes: Über metaphorische Technologien der frühen Bakteriologie. Geschichte und Gesellschaft 30 (2): 250–276.

Metaphor scenarios in disease management

141

Schön, Donald 1979 Generative metaphor: A perspective on problem-setting in social policy. In Metaphor and Thought, Anthony Ortony (ed.), 254–284. Cambridge: Cambridge University Press. Stibbe, Arran 2002 From flu-like virus to deadly disease: Ideology and the media. Journal of Media Psychology: . Tovey, Hilary 2003 Theorising nature and society in sociology: the invisibility of animals. Sociologia Ruralis 43 (3): 196–215. Tumpey, Terrence M., Christopher F. Basler, Patricia V. Aguilar, Hui Zeng, Alicia Solórzano, David E. Swayne, Nancy J. Cox, Jacqueline M. Katz, Jeffery K. Taubenberger, Peter Palese, and Adolpho García-Sastre 2005 Characterization of the reconstructed 1918 Spanish Influenza pandemic virus. Science 310 (5745): 77–80. Twigg Johnson, Samantha 2009 Biosecurity: idyllic England in millennial Britain. In From Mayhem to Meaning: Towards a Cultural Understanding of the 2001 Outbreak of Foot and Mouth Disease in Britain, Martin Doering, and Brigitte Nerlich (eds.), 133–149. Manchester: Manchester University Press. Wallis, Patrick, and Brigitte Nerlich 2005 Disease metaphors in new epidemics: The UK media framing of the 2003 SARS epidemic. Social Science and Medicine 60: 2629–2639. Warren, Virginia L. 1991 The “medicine is war” metaphor. H. E. C. Forum 3: 39–50. Washer, Peter 2004 Representations of SARS in the British newspapers. Social Science and Medicine 59: 2561–2571. Washer, Peter 2006 Representations of mad cow disease. Social Science and Medicine 62 (2): 457–466. Watts, Geoff 2005 Review of Davis, 2005. British Medical Journal 331: 1275, 26/11/2005: Woods, Abigail 2004 A Manufactured Plague: The History of Foot-and-Mouth Disease in Britain. London: Earthscan. Zinken, Jörg 2007 Discourse metaphors: The link between figurative language and habitual analogies. Cognitive Linguistics 18 (3): 445–466.

142 Brigitte Nerlich Zinken, Jörg, Iina Hellsten, and Brigitte Nerlich 2008 Discourse metaphors. In: Body, Language and Mind. Volume 2: Interrelations between Biology, Linguistics and Culture, Roslyn M. Frank, René. Dirven, Tom Ziemke, and Enrique Bernárdez (eds.), 363–386. Amsterdam: John Benjamins.

“Overt” vs. “covert” cultural variance in metaphor usage: ‘Europe’ vs. Malta and the EU-membership debate Monica Petrica

1. Introduction In linguistics, studies on the usage of metaphors in political discourse currently abound. The analysis of metaphors depicting the forging of the European Union and other EU-related issues has also appealed to cognitive linguists. Initially, these studies mainly focused on metaphorical political discourse in well-developed countries, like Germany and Great Britain. Recently, however, the analysis of this linguistic phenomenon in the discourse of less developed countries has come under scrutiny.1 The present paper primarily investigates the metaphors common in political and journalistic discourse in Malta with regard to EU-membership and also aims to disclose potential culture-specific conceptual metaphors. For various reasons, Malta is a very interesting example in the research on EU discourse. Politically speaking, Malta constitutes a rare phenomenon as the country is characterized by a two-party system.2 What is more, nowadays the two parties, the Nationalist Party and Malta Labour Party, tend to be defined in terms of their approval or disapproval of the EU-membership. It is often asserted in the press that the National Party came to power due to the discourse favouring the EU. My analysis is based on a small corpus which includes metaphors that occurred in the English-language press of Malta between 2000 and 2008.The data suggest that while many of the conceptual metaphors overlap with those found in political discourse all over Europe, several specific metaphors can be detected as well. For example, the metaphorical conceptualization of the European Union as a spanker or sodomizer who is going to spank/sodomize Malta seems to be specific to Maltese EU discourse. Chances are that such metaphors are scarce in the political discourse of countries like Germany, which is one of the founding members of the European Union and therefore ‘in control’.

144 Monica Petrica The comparison between the metaphors recurrent in the Maltese discourse and the ones illustrative for the German or British discourse is made possible by the data in the Eurometa-Corpus of Metaphors used in Eurodebates in Britain and Germany (available online) and by the interpretation of these data published in Andreas Musolff’s books (2000; 2004). EU metaphors specific to the Maltese discourse of the spanker/sodomizer-type are highly conspicuous, because their source domains are very special. I will call this type of international variation overt variation. However, in addition, there are Maltese metaphors whose source domains are identical to metaphors known to be used in many other EU countries, which are still special in that the source domains have specific significance in the socio-cultural and physical environment of Malta. For example, the common metaphor of the EU as a family of nations conjures up specifically Maltese associations in a Maltese context, which are determined by the cognitive model of the FAMILY prevalent in Maltese society. Metaphors of this type will be seen as involving covert variation. Following a clarification of central terms, Sections 3 and 4 of this paper will focus on intercultural and intracultural overt variation respectively, and Section 5 on covert variation.

2. Overt and covert variation – European vs. nation-specific metaphors The scale from overt to covert variation manifests itself in the degree of obviousness. Thus, overt variation refers to clear, incontestable cases of source domain variation, which are recognized as such at first sight. In such cases, it is not only selective mappings that diverge, but the whole source domain. Overt variation remains obvious when the metaphors are looked at in separation from their original context. In contrast, covert variation is only salient if the metaphors are analyzed within their original linguistic and cultural context, which implies that prototypical cultural models play an important role. This paper starts out from the pivotal finding in cognitive linguistics that suggests the existence of universal metaphors, i.e. widespread metaphors that are bodily and experientially grounded and thus shared by human beings irrespective of their geographic origin. In contrast to universal metaphors, culture-specific metaphors are – as the term suggests – culturally sensitive and rooted in the interaction of language, mind, and culture.

Cultural variance in metaphor usage

145

As the object of analysis is restricted to the discourse on the European Union, the term universal metaphor would not suit my purpose. Therefore, I suggest replacing the well-known dichotomy of universal vs. culturespecific metaphors (cf. e.g. Kövecses 2005: 35-36) by the more restricted opposition European vs. nation-specific metaphors. European metaphors are metaphors shared by the EU-related political discourse(s) in Germany, France, England, as well as Malta and other smaller European countries, whether EU-members or aspiring EU-members. Nation-specific metaphors, on the other hand, are those unique, or at least specific, to the discourse of individual countries, here specifically Malta. However, since metaphor variation within European discourse is particularly determined by economic differences, it should be emphasized that nation-specific Maltese metaphors may to some extent be shared by the public discourse of other countries characterized by a weak economy (e.g. countries of the former Eastern Bloc). Moreover, it cannot be excluded that these conceptual metaphors, which are prevalent in the public discourse in Malta (and Eastern European countries), will also occur in the European discourse of economically strong countries like Germany or France. In fact, the main difference often lies in the frequency of occurrence: while nation-specific metaphorical expressions are frequent in the discourse of some, e.g. the smaller and weaker, members, the odd occurrence in the discourse of well-developed countries can still occur. Nevertheless it is this difference in usage that reflects the political makeup of a country, and more often than not, the situation is of course not uniform even within one nation due to different political opinions. For example, as one of the two major political parties in Malta is said to be a Europhobe party, it goes without saying that specific metaphors deviating from the middle-of-the-road European discourse will abound in Maltese public discourse coming from sources associated with this party. The fact that many conceptual metaphors are shared by the European discourse(s) of the member states becomes obvious if one looks at the findings in the literature on EU metaphors. A particularly useful basis for the investigation of the public debate on the EU in Britain and Germany is the Eurometa-Corpus compiled by Andreas Musolff and his research team. This database is structured around 20 source domains shared by British and German discourses, such as LOVE AND FAMILY, GROUP/CLASS/CLUB, GAMES/SPORTS, TRAIN, LIFE/BODY/HEALTH, DISCIPLINE/AUTHORITY/ SCHOOL, HOUSE AND BUILDING, etc. Research into the EU-related discourse in Eastern Europe has shown that many of these source domains are impor-

146 Monica Petrica tant there as well. For example, Šariü (2005) analyzed metaphorical models in the Croatian Media and detected similar source domains, including JOURNEY, HOUSE/BUILDING, TRAIN, PATH/MOVEMENT, HEALTH/DISEASE, SPORT/RACE/GAME. However, as mentioned above, despite striking similarities it can be misleading to offer such lists of shared source domains, since closer investigation may reveal covert variation due to different conceptualizations of these domains in different cultures and societies. What follows is an analysis of the main metaphors that distinguish the Maltese discourse from the dominant European discourse represented mainly by German/British sources. The next section focuses on metaphors which – due to their universal raw material – are likely to occur in the discourse of other low-developed countries as well.

3. Intercultural overt variation As already mentioned, it cannot be excluded that metaphors recurrent in Maltese discourse also occur in the discourse of other member states characterized by a low-developed economy and, marginally, even in economically well-developed countries, in Europhobic speeches. To begin with, a discourse of PRESSURE can be recognized in the debate on the EU membership in smaller member-states. Not only in negative contexts, but also in positive ones, an inventory of the lexical field of pressure can be compiled, including items such as press, coerce, force, etc. These lexical items are present in contexts referring to all areas of life, from divorce laws to hunting and bird shooting, which would have to be adjusted to the European norms. Metaphors of pressure also prevail in Maltese discourse on the EU institutions: (1)

From the great defender of makku’s sovereignty, Labour now rushes to report the government to the ‘big brother’ in Brussels whenever it drags its feet on any of those – hitherto – costly, bureaucratic, burdensome and useless maze of straightjacket EU laws (http://www.timesofmalta.com/core/article.php?id=259431, emphases mine, MP).

Given their frequency in all kinds of discourse it is not surprising that anthropomorphic metaphors are also quite common in the discourse on Europe. But whereas in the German and British discourse Europe itself is

Cultural variance in metaphor usage

147

personified, in the Maltese discourse Malta as an individual state is regarded as a human being. In one of the metaphorical instances, Malta as a member is conceptualized as a ‘baby’: (2)

“We need the money to be spent now – it’s like a baby that needs a full bottle of milk but is only given half now and the other half kept in the fridge. Why, if the baby needs it all now? And our economy needs these EU funds to be spent now.” (http://www.maltatoday. com.mt/2009/03/15/t3.html, quotation marks in the original)

In this example, the conceptualization of the country as a baby, conflated with the conceptual metaphor MONEY IS A NUTRITIOUS FLUID, gives rise to the framing of the EU as a nurturant parent – a metaphor which will be explored in greater detail in Section 5.2. Not surprisingly, negative images are not uncommon – especially in the discourse of Eurosceptic parties (see Section 4 below) – in economically less well developed countries like Malta. The EU is the embodiment of negative forces, of monsters and other disastrous phenomena. An example of how such tendencies are embedded in the specific socio-historical and linguistic context of Malta is Joe Brincat’s conception of Europe as a ‘spanker’ or a ‘sodomizer’. (3)

To spank or to sodomise – Brincat and Bondi cross swords On Sunday 9 June the MLP deputy leader, Dr Joe Brincat addressed the public in Gudja and during his speech, he voiced his opinion on the state of affairs concerning the EU – in plain Maltese language. ‘Nispiccaw biz-zokk f’idejna … pero anke nispiccaw bizzokk fuq il-warrani … iz-zokk nuzawh fuq il-warrani.’3 These phrases formed part of his speech and immediately elicited giggles from members of the listening crowd, who immediately tuned to one track of the comments … He claimed that what he said had been in plain Maltese language, referring to a spanking with a stick and in no way had he meant to refer to anything remotely vulgar. On Friday 14 June in his regular column to The Times, Lou Bondi stated that Joe Brincat had spoken in an extremely vulgar manner to say the least, and claimed that Brincat had said that ‘the European Union was trying to sodomise us’ and that ‘According to the MLP deputy leader we should, as it were, turn around and sodom-

148 Monica Petrica

ise Europe back.’ (http://www.maltatoday.com.mt/2002/0616/ l6.html) Due to a recurrent pun in the Maltese language, in which the Labour Party representative’s speech was delivered, an ambiguity arises. Thus, zokk, which means ‘stick or branch of a tree’, is similar to the word zobb (‘penis’) and is often used euphemistically in common parlance to refer to ‘penis’. Furthermore, placed in the context with warrani (‘backside, bum’), the lexeme zokk suggests two alternative meanings or, more precisely, leaves it open to the hearer to decode the utterance in one way or the other. If one analyzes such metaphors against the Maltese cultural background, the assumption that they arise from the Maltese colonization heritage is not far-fetched. Therefore, it can be concluded that such metaphorical expressions distinguish the Maltese public discourse from those of other countries for which colonization is not an historical constituent. The long centuries of colonization underlie the conceptual system of the Maltese people and therefore emerge in metaphoric usage. Another example of overt variation is the conceptual metaphor THE EU IS A FORTRESS. As suggested by the example below, Malta would be reduced to the status of a remote outpost, which both recalls the colonization period and stresses the lack of importance and influence of small states within the EU. (4)

Malta At The Crossroads Again (Speech by Dr Harry Vassallo, Chairperson AD – The Green Party To The Commonwealth Foundation, Visit of Commonwealth Fellows 15th March, 2002) The opposition expresses the fear of some of the partners in our small business: that we are not prepared for such a leap, that the human and economic cost of adjustment to the new reality will be catastrophic, that we have done well so far by being different, by exploiting our uniqueness just beyond the borders of Europe, membership will reduce us to the status of remote outpost. (http://www.alternattiva.org.mt/speeches.html#SELFACtstud)

Interestingly, politicians also make direct reference to colonial times and compare EU-membership to the status of being a colony. Consider, for example, the following fragment from the article “Alfred Sant (on Wednesday) Profession”:

Cultural variance in metaphor usage

(5)

149

[T]he argument has been that, in the EU context, sovereignty can and should be shared. So what is wrong with Malta giving up part of what it had at last gained in 1964 and 1979? After all, by doing so it gains a voice in the way by which continental policies will be shaped. For those who genuinely believe this fantasy, it at least offers an escape route by which to fudge the return to the politics of colonialism in Malta. I cannot believe – or pretend to believe – in such an escape route. (The Times of Malta, 16/04/2003)

Another rich source of figurative language is Malta’s geographical position as an island. It is presumably not too far-fetched to argue that this special experience of space becomes embedded in people’s mental representations; and since the cognitive cannot be separated from the affective, it can be assumes that the physical setting also influences the psychological make-up of the Maltese. Consider the following quotation: (6)

And here’s Mrs Mizzi’s ‘vote for moi’ advertisement in today’s newspapers: “I would like to see the insular mentality, so characteristic of islanders, to be diluted into a healthy blend of ‘Europeanism’. I would like to give children at an early age, a craving for learning and a yearning for knowledge.” Yes, ma’am, we believe you. That’s why you voted in the 2003 general election for a man determined to keep us out of the European Union, because you wanted to dilute our insularity and give our children chances. (http://daphnecaruanagalizia.com/2009/05/21/hallina-marlene/, emphases mine, MP)

This example illustrates how the Maltese project the bounded landmark of the island as a real ecological space upon the mind space so that the ‘insularity’ feature becomes a characteristic of the mind. Further, insularity loses concreteness and can be blended upon the space of the “European mind” in an attempt to create a new conceptual organization. From a negative point of view, an island is characterized by isolation and, directly related to it, ignorance and limited relations to the surrounding territories as well as scarce resources. These are exactly the arguments used by people favoring EU-membership. A dominant conceptual metaphor is EU MEMBERSHIP IS A GEOGRAPHY CHANGER. In the following excerpt from the article “Staying out means remaining a backwater”, a series of complements (a backwater, ignored, bypassed, and insular) convey the idea that

150 Monica Petrica not joining the EU would preserve and/or contribute to Malta’s isolation. If this proves true, it goes without saying that the opposite is also true, i.e. that EU membership is apt to amend the geographical (insular) position of the island. (7)

The decision to join the European Union is like the decision to marry. The heart and mind both play a part. You have to use both when coming to your decision … By staying out of the European Union, I fear that we will remain a backwater, ignored, bypassed, and insular. I find this far more frightening than any one of the scare stories being promoted by the ‘no’ lobby. (The Times of Malta, 04/03/2003)

The discourse of insularity/marginality is based on a centre-periphery schema with the EU constituting the center and Malta standing for the periphery. As the insular geography cannot be contested, it is obvious that this center-periphery schema underlies both positive and negative contexts. However, the negative contexts are characterized by a centrifugal propensity, whereas in the more positively intended texts a centripetal tendency prevails. As already mentioned, the issue of size is also present in articles and speeches conveying a positive attitude towards the European Union: (8)

Malta, the size of a small town in Europe, will be standing next to giants, sharing the same experiences and making its own contribution, whatever the cynics, in Malta and abroad, may think of the island. We will be leaving behind those still caught in a time warp. (The Times of Malta, 28/04/2003).

This quotation is ripe with metaphors: the metaphor POWERFUL COUNTRIES ARE LARGE CREATURES, implying the metaphor SMALL COUNTRIES ARE PYGMIES dominates the passage. However, the collocation sharing the same experiences places different sizes on the same level; indicating that size is a relational value. Size is a category that may undergo changes by losing some of its most intrinsic features. A closer inspection reveals that in positive contexts (as in the above paragraph), the centre-periphery schema sometimes disappears. Size is still a fact, but it does not indicate importance any more since Malta, as a member, will stand in the immediate proximity of other prominent countries.

Cultural variance in metaphor usage

151

Therefore, the metonymy SIZE FOR IMPORTANCE will become irrelevant after joining the EU (or in the visionary texts concerning the time after having become a member of the EU).

4. Intra-cultural overt variation As a result of the tension characterizing the political field in Malta, a strong intra-cultural variation becomes manifest within the Maltese public discourse. Malta’s two dominant political parties, the Nationalist Party and the Malta Labour Party can be defined in terms of their approval or disapproval of Malta’s EU-membership. If one compares the source domains employed by the Nationalist Party and the available evidence on British/German discourse mentioned in Section 1 above, it goes without saying that the discourse of the former, which has been in favour of the European Union from the very beginning, resembles the ‘sanctioned’ European discourse in Britain/Germany (and presumably that of other well-developed EU member countries) whereas the discourse of Malta’s Labour Party, a ferocious EU opponent, deploys a whole range of novel metaphors. It should be also pointed out that the victory of the Nationalist Party as well as the defeat of the Labour Party at the General Election in April 2003 are explained by their approving/disapproving position regarding the European Union. Malta is a very interesting case inasmuch as the outcome of the EU Referendum in March 2003 closely mirrors the results of the General Election in April 2003.4 Thus, apart from the cultural differences, which account for variance in metaphoric usage, another explanation for the existence of shared and variant metaphors within the context of EU-enlargement is the division of the population into Europhiles and Europhobes. Quite plausibly, the Europhiles will tend to resort to metaphors common in the ‘European’ discourse, while the Europhobes are supposedly more likely to create their own, novel metaphors. I do not claim that the Europhobes do not resort to conventional metaphors present in the EU discourse around Europe, i.e. European metaphors, as, logically, they would use in their discourse some of the metaphors present in the discourse of other Europhobes in Europe. Nevertheless, I assume that the Europhobes are more likely to create original, more convincing metaphors, as the general tendency is to enter the EU, so that the Europhobes’ lack of control is apt to trigger a larger usage of novel, attention-seeking metaphors. And again, because the Europhobes are more ori-

152 Monica Petrica ented towards national values, they would undoubtedly tend to resort to the country’s national heritage and local culture when creating new, original metaphors. This is supported by the following set of examples: (9)

Two years ago people overwhelmingly endorsed the European project. Euphoria was running high. Many people believed in Europe as the panacea for the country’s economic ills. Some remained highly suspicious others reserved judgement. With the passage of time people started to draw a more realistic picture. (http://www.maltatoday.com.mt/2006/04/30/editorial.html)

This laudatory metaphorization of the EU as a panacea becomes an ominous figure when the Nationalist Party position is ridiculed by their political opponents: for the latter, Europe turns out to be a quack: (10)

They [The Nationalist Party; MP] projected EU membership as a cure for all ills” rather than the better way forward, on balance. The imagery was of a quack selling mysterious bottles [emphasis mine]which would guarantee health, virility, hair growth, and sweet-smelling perspiration. That raised and fattened expectations. (http://www.timesofmalta.com/core/article.php?id=222530).

For those opposed to the membership, e.g. the Labour Party and its supporters, Europe is conceptualized as an enemy (embodying zoomorphic or apocalyptic features): (11)

In other words, the MLP is convinced that they – the EU – are all out get at us and poor little Malta desperately needs someone to defend her: the MEPs elected from the MLP list of candidates. Is the rest of the EU, therefore, the enemy threatening to swallow us up? (www.maltatoday.com.mt/2004/05/09/opinion3.html)

5. Covert variation As mentioned above, a great range of metaphors seem to be shared by all European countries. Nevertheless, closer analysis reveals that even among the European metaphors what could be referred to as false friends can be identified. Two types of such covert variation can be distinguished. In the

Cultural variance in metaphor usage

153

first, source domains are identical across countries and languages but are associated with different target domains (cf. Section 5.1). Secondly, there are cases of covert variance where metaphors only seem to share the same source and target domains while, on closer observation, one notices that the source domain is actually different. As suggested by Kövecses (2005: 118), such culture-specific construals of a fairly general source domain may lead to multiple variants of a conceptual metaphor that seem identical at a superficial glance. The two source domains FAMILY and HOUSE, which have a strong cultural basis and illustrate this type of variation particularly well, will be discussed in Section 5.2

5.1. Identical source domains – different targets Consider the following examples of the JOURNEY metaphor from the British (12) and the Maltese (13) press respectively: (12)

Europe stands at a crossroads (cited after Eurometa-Corpus)

(13)

the country [Malta] is being driven into a dead end alley” (www.maltatoday.com.mt/2004/05/09/opinion3.html)

In example (12), the target domain is EUROPE, whereas in (13) it is MALTA. This difference reveals two alternative attitude patterns underlying the discourses on Europe: the first conceives of the (aspiring) member as part of the whole, i.e. Europe whereas the second pattern conveys an exterior and dynamic stance. The JOURNEY, or more specifically, ROAD metaphor is among the best examples of source domains shared by the European discourse as a whole, but applied to different targets. As in the rest of Europe, the ‘road’ metaphor with its sub-metaphor ‘means of transport’ is also largely used in the discourse about Europe in Malta. However, depending on whether the membership is seen from a positive or a negative viewpoint, the ROAD metaphor is realized as a promising or a futile journey. Thus, it can be predicted that in speeches made by affiliate members of the National Party the road or journey into a promising future will recur, whereas the members of the MLP are likely to employ the opposite elaboration of this conceptual metaphor. Consider, for example, the following realization of the JOURNEY METAPHOR as leading into a dead end alley:

154 Monica Petrica (14)

Speaking to MLP supporters following the traditional May 1 demonstration, Alfred Sant said that as a result of Malta joining the EU, the PN government had led the country into a dead end resulting in an economic and social crisis. He then went on to add that the MLP is committed to get Malta out of this cul-de-sac … However, this time the notion that the people made a bad decision as they were tricked by the Government into joining the EU was even more emphasised than was normal in the last twelve months – to the extent of Malta’s EU accession being referred to as a serious mistake, akin to the country being driven into a dead end alley. (www.maltatoday.com.mt/2004/05/09/opinion3.html; emphases mine, MP)

Prima facie, the metaphors seem identical with the movement metaphors, for example, in British discourse. It should nevertheless be pointed out that the perspective encoded in these metaphors is radically different. Whereas the British metaphors focus on Europe ‘moving’ (e.g. ‘The European Community at its most saintly fudges its way’(The Guardian, October 29, 1991, cited after Eurometa) or being at a crossroads, one notices that in the Maltese texts, it is Malta that is at a crossroads (rather than Europe). And again, it is Malta on the bumpy road to Europe and not the community as a whole. Undoubtedly, this is due to a different perspective, which in the first case determines the visualization of Europe as an Actor, while in the Maltese discourse ‘Malta’ becomes active in its movement towards the destination, i.e. Europe as a Goal. Another example of shared metaphors that are instantiated in specific ways is the CONTAINER metaphor, which also pervades the European discourse. Thus, in the discourse of powerful members, this metaphor will primarily be used with reference to new members that should be taken in or let in, whereas in the discourse of weaker members the metaphor conceptualizes the dilemma of staying out’ or going in – cf. “the crossroads in this nation’s history over whether it should join the EU or stay out” (The Times of Malta, March 6, 2003) – or even bringing (the country) out: (15)

Once the party had taken its stand against European Union membership, Dr Mifsud Bonnici argues, it had to respect the views of those it persuaded to vote for it and continue to sustain this view. Moreover, it must promise to bring Malta out of the EU as soon as it regains the support of the majority of the people and finds itself

Cultural variance in metaphor usage

155

in power … If anything, as has been stressed time and again by the Nationalist Party, one would have expected the Labour Party to be more eager than the Nationalist’s to join Europe, considering the prevalence of social democratic traditions and rights of workers inside the EU. (http://www.doi.gov.mt/EN/commentaries/2003/ 07/ind30.asp; emphases mine, MP) Furthermore, the orientational metaphors directly related to the container metaphors gain persuasive importance: OUT IS COLD and IN IS WARM. (16)

He [Fenech Adami, MP] said the EU had been born out of a vision of three great Christians who led Germany, Italy and France after the war. They wanted peace and strove to eliminate cruelty. “People have a choice whether they want to form part of the new EU with 25 member states or to stay out in the cold”, he said. (The Times of Malta, 05/03/2003)

Two particularly fruitful source domains which show how metaphors that are used in a seemingly identical way in the discourses around Europe acquire different meanings in different socio-cultural contexts are FAMILY and HOUSE. These will be looked at in greater detail in the next section.

5.2. Identical source domains – different cultural models The FAMILY metaphor plays an important role in the present analysis of metaphors. It is an illustrative example of how cultural variation works, even when prima facie the metaphor seems to have a universal status (e.g. SOCIETY IS A FAMILY). Moreover, the FAMILY metaphor is not only an example of intercultural variation, but also of intracultural variation since within one society there is normally more than one co-existing family model, as will be shown in the following section. The definition and analysis of a ‘universal’ family as the nucleus of every society imply a high degree of abstraction and a radical reduction of the particular features of family models around the world. For the present case, I will compare the ‘Western family’ ideal with the idealized Maltese model of ‘family’.5 While it is difficult to gather objective non-linguistic evidence on cultural models, informal discussions with Maltese people about family values

156 Monica Petrica and typical behavior strongly suggest that a striking difference between the Western model of the family and the Maltese familial model appears to manifest itself in the moving-out patterns of young people. This aspect is – in my opinion – likely to affect, and reflect, the understanding of the ‘family’ metaphor. In Western Europe teenagers more easily and much earlier achieve their independence than the youth in Malta, where young people generally only leave home after marriage. This is first and foremost due to the country’s religiosity. In addition to the religious as well as economic reasons, the explanation for the young people’s living with their parents until getting married has to do with the small size of the island and the types of social networks resulting from this. Thus, no matter where young people would choose to move, it will never be too far from the parental home, so that parents would still easily find out everything about their offspring and could keep on meddling in their affairs. If we transfer this feature to the metaphor EUROPE IS A FAMILY, we can assume that the Maltese people would expect the European Union to be a ‘family’ in which members would know everything about each other and could easily get involved in each other’s affairs. Given the dominance of the European metaphor EUROPE IS A FAMILY, I would argue that counter to the expectation in other European countries, this metaphor has contributed to the fear that the Labourites showed towards the option of Malta’s entering the EU family in the form of ‘full membership’. An overarching metaphor dominating Maltese journalist discourse is the representation of Europe as a NURTURANT PARENT (Lakoff 2002: 108). Leaving aside the metaphorical expressions in which Europe is directly portrayed as a family, all the arguments circulating before Malta’s Referendum on Europe were based on the conceptualization of the European Union as a ‘giver’, ‘a provider’ whereas Malta would be the ‘recipient’, ‘the beneficiary’: (17)

Many people look at the EU as if the queen was coming back to Malta; instead of milking funds from the queen we will be milking funds from the EU. This is not the case and this is a mentality we have to change. (Malta Today, 01/02/2004)

However, even within Malta, cognitive models of FAMILY are not uniform, but subject to variation and historical change. Carmel Tabone posits the coexistence of at least five types of families in contemporary Malta: the traditional family, the conventional family, the modern family, the deprived

Cultural variance in metaphor usage

157

family and the progressive family (Tabone 1994: 247-249).6 Given the hypothesis that stored mental representations guide our meaning construction, I can assume that this intracultural variation also has an effect on the EU membership debate and on people’s envisionment of the future reality (i.e. after EU accession).

5.2.1. Different family models within changing Malta On the Maltese political scene prior to the Referendum in March 2003, there were two alternatives open for Malta: firstly, full membership, supported by the Nationalist Party and secondly, partnership supported by the Labour Party. After a more thorough analysis, it becomes obvious that the two alternatives actually stand for two version of ‘familial unions’: (18)

EU membership is like being married. Before the marriage you have the engagement, which is when you lay down the rules and regulations . If you agree, you get married. That’s it. You’re in. But you must bear in mind that the dominant partner may change the rules and regulations after the marriage. No divorce is possible. Now with partnership, it’s like two people moving in and living together: Initially, no rules and regulations, these are made up as you go along, to the common good of both partners. If you don’t agree to the rules etc. you can walk out and start again. Nobody gets hurt. (Malta Independent, 05/03/2003)

In other words, the ‘full membership’ version stands for the accepted type of familial union based on a marriage licence or other legal document. The ‘partnership’ would merely be a form of domestic partnership agreement, in which the two partners are not joined in a traditional marriage or a civil union. The advantage of the latter would be that the union could be more easily dissolved in case of a bad match. This is exactly why Alfred Sant, the leader of the Labour Party, rejected the idea of ‘full membership’ from the very beginning. However, it should be pointed out that even in the case of a partnership, the roles of Europe as a ‘provider’ and of Malta as a ‘receiver’ remain the same. Only the prerogatives of Europe would be modified as Europe would have fewer rights to interfere in Malta’s affairs. As the family and family life are central issues in the Maltese people’s lives, contributing a great deal to moulding personalities and shaping val-

158 Monica Petrica ues, one can assume that the family model to which an individual is attached might have determined the vote for one or the other form of ‘joining’ the EU. In his account of American morality and politics, Lakoff argues that individuals brought up with a ‘strict-father’ family model are likely to adopt conservative ideas whereas people brought up in a ‘nurturing’ milieu would be attracted to liberal ideals instead (Lakoff 2002: 12). Following a similar line of argument, it might be assumed that in the accession referendum the Maltese utterly devoted to the idea of a family grounded in the Catholic tradition may have voted for the ‘full membership’, while the ones with more liberal ideas and who accept a domestic partnership as an alternative to the traditional family might have supported the political partnership between Malta and the European Union. However, considering the extensive religiosity of the Maltese population, it is surprising that the percentage of the electorate that voted for ‘full membership’ is only slightly larger than the percentage of those who voted for political partnership. If one takes these results to be indicative of the frames dominating the Maltese society, it can be concluded that the election outcome shows that a reframing must have taken place (Lakoff 2006: 12–13).

5.2.2. EUROPE IS A FAMILY vs. MALTA IS A FAMILY As has already been stated, the importance of the family is paramount in Malta. Due to the small size of the island, the whole nation is regarded as a ‘family’ based on the frequent argument that in Malta ‘everybody knows everybody’. In other words, the ‘NATION IS A FAMILY’ metaphor could be almost taken literally in Malta as a consequence of the closely-knit social network characterizing social life on the island.7 Taking into account that in Maltese literature, the nation is conceptualized as a family and Malta (as a country) is referred to as the ‘mother’, along with the argument that Malta is not really interested in the EUmembership on a basis of shared interests, but for financial reasons, it follows that the fusion of the ‘Maltese family’ with the large family, i.e. Europe, is unlikely to be effected. As a direct consequence, Malta will only be a family within the ‘larger family’, without necessarily being part of an extended family in which ‘kinship ties’ are perfectly maintained. The fact that the some Maltese are obsessed with the idea of the family as a nuclear unit often leads to familism, which, according to Lakoff’s

Cultural variance in metaphor usage

159

(2002: 312–315) account of moral deviation, would be a form of moral pathology. To put it simply, individuals are likely to put familial interests above everything else and thus jeopardize the relation with the larger social group. A reflection of this pathological attitude is the striking cleanliness of the Maltese houses as compared to the dirt on the path in front of the house: The fact that the Maltese are meticulous in cleaning their house but quite reckless in matters concerning public cleanliness is symptomatic of such an attitude. And if they do clean in front of their doorways it is only because they do not want the dirt to enter inside. This does not occur in cleanliness only but also in various other aspects of social life. (Tabone 1994: 237)

If one extends this attitude to the larger scale of the ‘national family’, it follows that what happens outside the space of ‘one’s own’ family (i.e. Malta) is not of great interest for the Maltese. The same could be said of the metaphor EUROPE IS A HOUSE. For the Maltese, Malta is the ‘House’, so that the island would just be a ‘house’ within a larger ‘house’, where the walls of the Maltese house should not necessarily come into contact with the walls of the larger building, so that the former would not thereby lose its own characteristics or boundaries.

5.2.3. EUROPE IS A HOUSE The metaphor just mentioned, EUROPE IS A HOUSE, is another obvious case of covert variation. As in the case of family, this metaphor differs not only cross-culturally, but also within one culture. Schäffner (1996) offers a diachronic overview of this controversial metaphor. The metaphor of the ‘common European house’, introduced into political discourse in the mid-1980s by the leader of the Soviet Union at the time, Mikhail Gorbachev, has been constant in the European debate and has undergone an essential change, i.e. the modifier ‘common’ has been discarded. According to Schäffner, this omission is due to the decoding of this metaphor via the French prototypical house. Whereas Gorbachev imagined the ‘European house’ as a multi-storey apartment block with several entrances, shared by several families, each dwelling in their own flat, the French interpretation of “L’Europe notre maison commune” was based on the one-family house, which would allow the dwellers free movement within the house (Schäffner 1996: 33–34). As expected, this metaphor, which seems to be one European metaphor par excellence, differs from one culture to another. In Malta, this concep-

160 Monica Petrica tual metaphor occurs both in the Europhile discourse and in the Europhobic one. If these two types of discourse are compared, it becomes obvious that, even intraculturally, there are two different instantiations of the HOUSE metaphor: EUROPE IS A HOUSE 1 vs. EUROPE IS A HOUSE 2. If the frame for an understanding of the EU is as a family (of nations), the source domain in the metaphor EUROPE IS A HOUSE will most probably be understood as a house shared by the members of the family (of nations), i.e. the metaphor has connotations which entail togetherness and fellowship. Consider, for example, the following quotation from the article “A new beginning for Malta” that appeared in The Times of Malta (17 April 2003): (19)

The European leaders, including Dr. Fenech Adami and President de Marco, were also part of the largest-ever European family photo. … ‘We consider the EU to be our home,’ he [Fenech Adami; MP] said.

Another example is taken from the article “The ‘Yes’ vote of a European Maltese” (Mario de Marco) which appeared in The Times of Malta, 4 March 2003: (20)

On March 8 we will be voting so that a European Malta will take its rightful place in our maison commune, in our common European home. (emphasis mine, MP)

As the above examples show, the members of the Nationalist Party (the Europhiles) understand the ‘European house’ as a shared house where all the nation members live together like in a family. Interestingly, in the Nationalists’ vision, the European home is not an artifact, but a natural thing: (21)

The choice before us is clear: do we want to be part of the European Union or do we want to detached from our natural home, a home of common values and aspirations? The EU is a success story and has brought tremendous benefits to its citizens. ... Malta cannot get a better deal from the EU by remaining out of union, which is what those who oppose EU membership are saying. ... Can Malta afford to stay out? Absolutely not. European Union enlargement is an opportunity not to be missed. By joining the EU, Malta would be returning to its natural home, a home of common values and aspi-

Cultural variance in metaphor usage

161

rations. (Malta Independent, 06/03/2003, Referendum campaign, Eddie Fenech Adami) In the above quotation home is not only a “dwelling-place” but also an identity symbol. New elements are added to the HOME metaphor, so that a new metaphor emerges: THE EU ACCESSION IS A HOMEWARD JOURNEY metaphor. EU is a natural home to which Malta is returning after long wanderings, i.e. the movement through its history of colonization. Effecting the European homecoming has great implications for the Maltese. The Semitic origin of the Maltese language together with the dark complexion of the Maltese people led to the Maltese being regarded as ‘Arabs’, i.e., as nonwhites (Pirotta 1994: 103-104). Thus, the EU membership stands for a proof of their identity as Europeans and coming homeward would mean reasserting their basic values (such as Christian religion) after a long history of political and cultural insecurity. It should also be mentioned that the connotations of the lexeme home, in opposition to house, i.e. the neutral term as the basic term for the category BUILDING, must not be overlooked. Moreover, I am convinced that the noun home, modified by the possessive adjective our is connotatively loaded. It should be emphasized as well that in the Maltese language there is only one lexeme, dar, used both for ‘house’ and ‘home’, which would also explain the constant presence of ‘home’ in contexts in which other languages prefer ‘house’. As the interpretation attributed to Gorbachev illustrates, living in the European House can be conceptualized as having your own house within a larger house. In Malta, it is this realization of the metaphor that is present in the discourse of the Eurosceptics. Consider, for example, the following excerpt from the Malta Labour Party Manifesto 1998: (22)

On the other hand, the New Labour Government appreciates and supports the process of economic, social, and political integration spearheaded by the European Union. It is also aware that this country can only enjoy the concrete benefits of full membership in this Union once we have adequately consolidated the economic foundations of our own home – our Maltese Home in Europe. (Malta Labour Party Manifesto 1998; emphasis mine, MP)

Although the collocation our Maltese Home in Europe does not refer to a block of flats, but to a house inside a larger house, the model put forward

162 Monica Petrica by Gorbachev seems to resemble the Labour Party’s vision of living together, but without interference from the other dwellers. The two models activated by the ‘house’ metaphor therefore do not entirely overlap: in the house pictured by the Nationalist Party, the family members live together in harmony (French model: ‘maison commune’) whereas in the house portrayed by the Labour Party, the Maltese are isolated inside their own house (‘house-in-house’ converges with the Russian model: ‘block of flats’). They are independent and no one else can interfere in the household’s affairs.

6. Conclusions The close analysis of EU metaphors recurrent in the Maltese discourse as well as their comparison to prevalent metaphors in the German discourse leads to the conclusion that cases of both “overt” and “covert” variation can be detected in the discourse on the European Union throughout Europe. The results of the analysis suggest that many metaphors prevalent in the Maltese public discourse overlap conceptually with metaphors occurring in the European public discourse in general. One explanation, which has been suggested to account for the existence of what I defined as “European” metaphors (EU-specific metaphors that might be identified in the EU-discourse of various countries) is the transfer of metaphors from one language to another, e.g. via translations (Šaric 2005: 3). However, even in such cases of similarity, a thorough analysis indicates that these shared metaphors are characterized by different metaphorical conceptualizations and entailments. These covert differences can be explained on the basis of socio-cultural and even environmental differences: as suggested by Kövecses, generic level schemas are filled out with socio-cultural substance and thus multiple instantiations are achieved at the specific level (2005: 68).8 The source domains presented to support this argument are HOUSE and FAMILY, two source domains that seem very prolific in the discourse on the EU. It has been shown that the FAMILY frame can lead to the emergence of competing metaphors even within the same culture due to co-existing worldviews. Worldviews may determine the preference for a certain metaphor and thus determine a certain course of action. In its turn, the preference for one metaphor and the rejection of another can presumably be demonstrated on the basis of people’s concrete actions (e.g. voting for a

Cultural variance in metaphor usage

163

political party/political vision), can indicate a reframing or value reappraisal. Close attention has also been given to the metaphors that are apt to distinguish the Maltese discourse from the discourse of other countries. In this respect, it has been demonstrated that not only the long history of colonization has formed the source domains in Malta, but also the country’s geographical setup surfaces in the discourse devices in use. Insularity and smallness accounts for a number of metaphors and other tropes, so that one can speak of a discourse of smallness and insularity in the EU-membership debate. At the level of metaphors, the insularity and small size are reflected in a range of expressions conveying the lack of national importance sensed by the inhabitants. It should, however, be mentioned that the discourse of insularity is not only an attribute of the discourse on the EU, but is also a feature of the discourse regarding Maltese identity. Nevertheless, I would like to emphasize that the metaphors presented as culture-specific and thus as distinguishing the Maltese discourse from the discourse of other countries (e.g. Germany) are likely to occur in the discourse of other nations that tend to share some of Malta’s distinctive features, such as geographical characteristics and socio-economic or historical features.

Notes 1.

2.

3.

The most important source for my comparative purposes is the ARC project carried out at the University of Durham in collaboration with the Institut für deutsche Sprache, Mannheim under the Anglo-German Research Collaboration Programme (ARC) and funded by the British Council and the German Academic Exchange Service. The result of this project is the book Attitudes towards ‘Europe’ (Musolff et al. 2001). Another example is the larger project funded by the Asko Europa-Foundation in Saarbrücken, called “European discourses in Germany, France and other EU member states” (my translation). Its results have been compiled in the book Denkart Europa. Schriften zur Europäischen Politik, Wirtschaft und Kultur (2006). The number of projects concentrating on the EU discourse is vast. This is only a very limited selection. “Since the early 1970s, this polarization has intensified to such a degree that 98 per cent or more of the electorate now vote for one of the two main parties” (Cini 2002: 6–7). “We will end up with the member in our hand, also in our butt... we use it on our butt (my translation, MP).” This sentence was translated on the basis of

164 Monica Petrica

4.

5. 6.

7. 8.

three Maltese native speakers’ comments. The vulgar words in the original were euphemized in the translation. The outcome of the EU Referendum on March 9, 2003 was a narrow pro-EU vote (53.6 per cent). This made Alfred Sant question the result of the Referendum and request the holding of general elections as soon as possible with the hope that the Labour Party would win and thus the invalidation of the pro-EU vote would become legally binding. The results of the election mirrored closely the result of the referendum as the Nationalist Party won with a slightly higher number of votes and thus the pro-EU vote was regarded as valid. Therefore, it is often argued that Malta voted twice for Europe: once in the Referendum and once in the General Elections as the vote cast for the Nationalist Party is a covert vote for Europe (Henderson, 2004: 155). For my analysis, I primarily use the description of the Maltese family as presented in Carmel Tabone’s (1994, 1995) studies. According to Carmel Tabone, the traditional family cherish fundamental values and resist change; the conventional family accepts traditional values, which are part of its’ members conscious worldview, but in practice adopts a way of life that is incompatible with these values; the modern family opposes the traditional family model and adjust to progress and to the needs of the contemporary society; the deprived family is characterized by lack of satisfaction in life, which makes its members adopt a different value system from that of the traditional family; the progressive family tries to follow the trends of development but preserves at the same time the basic traditional values in their behaviour patterns (Tabone 1994: 247–249). Term used by Milroy (2002: 414), borrowed from Blom and Gumperz (1972). The generic-level metaphor was introduced by Lakoff and Turner (1989).

References ASKO Europa-Stiftung 2006 Denkart Europa. Schriften zur Europäischen Politik, Wirtschaft und Kultur. Baden-Baden: Nomos-Verlag. Blom, Jan-Petter, and John J. Gumperz 1972 Social meaning in linguistic structure: Code-switching in Norway. In Directions in Sociolinguistics: The Ethnography of Communication, John J. Gumperz, and Dell Hymes (eds.), 407–434. New York/London: Holt, Rinehart and Winston. Chambers, Jack K., Peter Trudgill, and Natalie Shilling-Estes (eds.) 2002 The Handbook of Language Variation and Change. Malden, Massachusetts/Oxford/Garitón, Australia: Blackwell.

Cultural variance in metaphor usage

165

Cini, Michelle 2002 A divided nation: Polarization and the two-party system in Malta. South European Society and Politics 1: 6–23. Henderson, Karen 2004 Developments in the applicant states. In The European Union. The Annual Review 2003/2004, Special issue of Journal of Common Market Studies, Lee Miles (ed.), 153–168. Oxford: Wiley-Blackwell. Kövecses, Zoltán 2005 Metaphor in Culture: Universality and Variation. Cambridge: Cambridge University Press. Lakoff, George 2002 Moral Politics: How Liberals and Conservatives Think. Chicago: University of Chicago Press. Lakoff, George 2006 Whose Freedom? The Battle over America’s Most Important Idea. New York: Farrar, Straus and Giroux. Lakoff, George, and Mark Turner 1989 More Than Cool Reason: A Field Guide to Poetic Metaphor. Chicago: University of Chicago Press. Miles, Lee (ed.) 2004 The European Union. The Annual Review 2003/2004, Special issue of Journal of Common Market Studies. Oxford: Wiley-Blackwell. Milroy, Lesley 2002 Social networks. In The Handbook of Language Variation and Change, Jack Chambers, Peter Trudgill, and Natalie Schilling-Estes (eds.), 549–568. Malden, Massachusetts/Oxford/Garitón, Australia: Blackwell Publishing. Musolff, Andreas 2000 Mirror Images of Europe: Metaphors in the Public Debate about European Britain and Germany. München: Iudicium. Musolff, Andreas 2004 Metaphor and Political Discourse: Analogical Reasoning in Debates about Europe. Houndmills: Palgrave Macmillan. Musolff, Andreas, Colin Good, Petra Points, and Ruth Wittlinger (eds.) 2001 Attitudes Towards Europe: Language in the Unification Process. Aldershot et al.: Ashgate. Musolff, Andreas, Christina Schäffner, and Michael Townson (eds.) 1996 Conceiving of Europe: Diversity in Unity. Aldershot et al.: Dartmouth. Pirotta, Godfrey A. 1994 Maltese political parties and political modernization. In Maltese Society: A Sociocultural Inquiry, Ronald Sultana, and Godfrey Baldacchino (eds.), 95–112. Msida: Minerva Publications.

166 Monica Petrica Šariü, Ljiljana 2005 Metaphorical models in EU discourse in the Croatian media. Linguistics 6 (1-2): 145–170 (). Schalley, Andrea, and Dietmar Zaefferer (eds.) 2007 Ontolinguistics. How Ontological Status Shapes the Linguistic Coding of Concepts. Berlin/New York: Mouton de Gruyter. Schäffner, Christina 1996 Building a European house? Or at two speeds into a dead end? Metaphors in the debate on the united Europe. In Conceiving of Europe: Diversity in Unity, Andreas Musolff, Christina Schäffner, and Michael Townson, 31–59. Aldershot et al.: Dartmouth. Tabone, Carmel 1994 The Maltese family in the context of social change. In Maltese Society: A Sociocultural Inquiry, Ronald Sultana, and Godfrey Baldacchino (eds.), 229–253. Msida: Minerva Publications. Tabone, Carmel 1995 Maltese Families in Transition: A Sociological Investigation. Malta: Ministry for Social Development.

Sources Eurometa-Corpus: The Guardian 29/10/1991. Malta Independent 05/03/2003, 06/03/2003. Malta Labour Party Manifesto, 1998, Malta Today 01/02/2004. The Times of Malta 04/03/2003, 05/03/2003, 06/03/2003, 16/04/2003, 17/04/2003, 28/04/2003.







Examining conceptual metaphor models through lexical frequency patterns: A case study of U.S. presidential speeches Kathleen Ahrens

1. Introduction One issue of concern in conceptual metaphor theory has to do with establishing the basis for a particular cognitive model. For example, it is essential to avoid postulating a cognitive model based on conceptual metaphors and then saying that the conceptual metaphors are evidence for that particular cognitive model. This circulus in probando can be avoided if it is first shown that a particular cognitive model exists based on evidence other than conceptual metaphors themselves or if, once conceptual metaphors have been postulated, other linguistic evidence is brought to bear on the issue. Lakoff’s two proposed cognitive models of political/moral systems (1996, 2002) have run into just such a quandary. Lakoff postulated that two specific models of the family organize conceptual metaphors into coherent systems that give rise to unambiguous moral rules. These moral rules relating to the concept of family underlie the values of the two predominant political parties in America, the Republicans and the Democrats, especially since the election of Ronald Reagan in 1980. The strict father (SF) model is based on a traditional nuclear family where the father has the primary responsibility to support and protect the family. The primary metaphor for this model is that MORALITY IS (UNDERSTOOD AS) STRENGTH. Lakoff postulates that the Republican Party in America bases its value paradigm on this model, and, thus, the Republicans will primarily use metaphors (such as MORALITY AS STRENGTH) that support this model. The alternate model, the nurturant parent (NP) model, is postulated to be used primarily by the Democrats. The NP model is based on a family where responsibility is shared among family members, and there is mutual caring and support given to all family members. The primary metaphor for this model is MORALITY AS NURTURANCE. These cognitive models, while seemingly intuitively correct in their assessment of the reasons for the fundamentally different views held by De-

168 Kathleen Ahrens mocrats and Republicans, have not received support in the linguistic literature. Charteris-Black (2004, 2005), for example, extensively analyzed metaphors in U.S. presidential speeches, yet did not present evidence supporting these two models (although he did present evidence for a related metaphor, the moral accounting metaphor). Cienki (2005) specifically searched for instances of these metaphors in transcripts of three televised debates between the 2004 presidential candidates, George Bush Jr. (a Republican) and Al Gore (a Democrat), and found only forty-eight instances of conceptual metaphors (in a 41,000-word corpus) that support these two cognitive models. The scarcity of these instances led Cienki to expand his search to include entailments, which could be either metaphorical expressions or nonmetaphorical expressions, such as “I think it’s important for NATO to be strong and confident” (Bush Jr., cited in Cienki, 2005: 290). Unfortunately, the steps for determining what constitute a metaphorical or nonmetaphorical entailment are often open to interpretation. However, the idea of looking at non-metaphorical entailments to support the models is similar to the hypothesis I am proposing here: Patterns of lexical usage found in well-defined corpora can reflect underlying cognitive models. This idea of looking at lexical frequency patterns has recently been utilized in analyses of lexical change (Lim 2002, 2008). Lim, for example, showed that over a 200-year period, U.S. presidents used an increasingly democratic rhetoric. He also demonstrated that words having to do with families (i.e., kinship terms) have increased substantially since Franklin Roosevelt’s presidency, and the words children and youth have jumped considerably since Jimmy Carter’s term. Lim’s studies have demonstrated that socio-cultural shifts can be detected through keyword analyses. In another diachronic corpus-analysis, Ahrens (2006) examined the frequency of use of male-biased words, such as the generic use of man and mankind, between 1945 and 2005 in State of the Union Addresses. She found that there was a marked drop in these usages after Title IX of the Civil Rights Act was encoded into law (during Gerald Ford’s term). Title IX specifically outlawed discrimination on the basis of gender. Although the generic uses of mankind are kept alive by Republican presidents, the use of man to stand for “people” has now dropped out of the presidential lexicon. Ahrens uses this data to argue that analyses of lexical use in small, but meaningfully designed, corpora can shed light on subtle bias with regard to language use. In addition to the above studies regarding the examination of lexical frequency patterns diachronically, a useful starting point to study and con-

Examining lexical frequency patterns 169

trast the political ideology of a particular speaker with others has been to determine prominent keywords in a text and then evaluate the reason for their salience (i.e., Hart et al. 2005; Koller and Semino 2009; Semino and Koller 2009). This work has been built on the insight gained from Scott (2001), who hypothesizes that a corpus reflects the accumulated exposure a speaker has to a language, and suggests that if a word occurs more often in a specific text or specific corpus, compared to a general corpus, then this word is a “keyword” for that text and reflects what the text is about. A similar assumption is made in this study with a slight variation: The assumption is that a corpus of an individual speaker reflects the speaker’s viewpoint within the constraints of that particular corpus. A further independent assumption is made that conceptual metaphor models can be independently and objectively associated with a set of keywords for that model. (The methodology of implementation of this assumption is given in section 2.) Finally, a hypothesis is then postulated wherein a comparison of the frequency of different sets of keywords in a corpus of speeches will reflect either the different ideological leanings of that speaker or the different audience demands of a particular speech corpus. Furthermore, speakers with different ideological leanings may have different lexical frequency patterns in their respective corpora, again within the constraints demanded by a particular audience. Recent work along these lines (Ahrens and Lee 2009) did not find evidence supporting these two conceptual models when the Senate Floor Speeches of U.S. Democratic and Republican Senators were contrasted as two large groups. However, it was noted that there were strong individual differences among senators, although these individual differences were outside the scope of that particular study. Thus, it will be useful to contrast the speeches of four different U.S. presidents to determine whether their individual differences in lexical choice reflect their respective political ideologies, and, furthermore, to determine whether they modulate their lexical usage based on the audiences that they are addressing. To test this hypothesis, corpora were downloaded from on-line sources, including the State of the Union Addresses (SOUAs) for U.S. presidents from 1981 to 2006 (http://c-span.org) and Radio Addresses (RAs) for the same period (http://presidency.ucsb.edu), in order to demonstrate that there are different patterns of frequency in presidential usage for lexemes relating to STRENGTH/AUTHORITY (SF model) and NURTURANCE/EMPATHY (NP model). I predict, following Lakoff (1996, 2002), that Republican presi-

170 Kathleen Ahrens dents will use more lexemes related to the SF model, while Democratic presidents will use more lexemes related to the NP model.

2. Methodology The choice of relevant lexemes was determined by using WordNet (http://wordnet.princeton.edu), an on-line lexical reference system in which nouns, verbs, adjectives, and adverbs are organized into synonym sets, each representing one underlying concept. Using WordNet 3.0, the appropriate sense was selected for each of the following four words: strength, authority, nurturance, and empathy. These four words were chosen because they are the top two metaphors for each model.1 The appropriate senses are listed in (1) through (4) below.2 The direct hypernyms given in WordNet are also listed. All concrete nouns and verbs in the definition and hypernym were then identified and are underlined as follows: (1)

Selected sense and its direct hypernym for strength Selected sense: (n) force, forcefulness, strength (physical energy or intensity) he hit with all the force he could muster; it was destroyed by the strength of the gale; a government has not the vitality and forcefulness of a living man Direct hypernym: (n) intensity, intensiveness (high level or degree; the property of being intense)

(2)

Selected sense and its direct hypernym for authority Selected sense: (n) authority, authorization, authorisation, potency, dominance, say-so (the power or right to give orders or make decisions) he has the authority to issue warrants; deputies are given authorization to make arrests; a place of potency in the state Direct hypernym: (n) control (power to direct or determine) under control

(3)

Selected sense and its direct hypernym for nurturance Selected sense: (n) nurturance (physical and emotional care and nourishment) Direct hypernym: (n) care, attention, aid, tending (the work of providing treatment for or attending to someone or something) no medical care was required; the old car needs constant attention

Examining lexical frequency patterns 171

(4)

Selected sense and its direct hypernym for empathy Selected sense: (n) empathy (understanding and entering into another’s feelings) Direct hypernym: (n) sympathy, fellow feeling (sharing the feelings of others [especially feelings of sorrow or anguish])

After these lexemes were selected, they were listed in a table with their frequency from the British National Corpus (BNC) from Sketch Engine (http://www.sketchengine.co.uk/). I attempted to equate the lists for frequency, but as the lexemes from the SF model were much higher in frequency than the lexemes from the NP model, I decided to randomly select nine lexemes from each group. The list of lexemes associated with the NP model is: sympathy, nourishment, sorrow, feeling, care, aid, attend, nurture, and nourish; while the list for the SF model is: dominance, authorization, potency, intensity, force, control, dominate, strengthen, and authorize. A word list was created to search for all inflectional forms of these lexemes. These eighteen words, and their associated inflections, were then searched for in the corpora.

3. Corpus creation The SOUA corpus was downloaded one speech at a time from the C-Span website (http://c-span.org), starting with Reagan in 1981.3 All SOUAs from 1981 to 2006 were directly downloaded to text files. Each file was then imported into Microsoft Word, and the president’s words in the speech were highlighted and counted with the word-count feature. In most cases, this meant that the heading was omitted (i.e., “President Ronald Reagan’s Address before a Joint Session of Congress”). In other cases, information about where and when the speech was given, or who introduced the president, also had to be omitted from the word count. In the case of Bush Jr., indications of “(Applause.)” had to be deleted as well.4 In short, every effort was made to include the words used by the president himself in the word count.5 In addition, it is important to note that these speeches were given orally from a prepared text. The version that is being examined here is the version that was provided for the written, historical record, and the content may therefore vary slightly from the actual words that the president spoke. The total and average word counts for the SOUAs for Democratic and Republican presidents are given in Table 1. Since some presidents gave

172 Kathleen Ahrens more than one SOUA in a given year (e.g., Bush Jr. gave two addresses in 2001, one on 27 February and one on 20 September), there are a total of twenty-eight speeches in this corpus. Table 1. SOUAs included in current corpora6 Word count Reagan 1981–1988 8 36,652 Bush Sr. 1989–1992 5* 20,461 Clinton 1993–2000 8 60,541 Bush Jr. 2001–2006 7* 32,336 Total 28 149,990 *Presidents gave more than one SOUA in a given year Name

Year

Total

Political party Republican Republican Democratic Republican

Average no. of words/speech 4581.50 4092.20 7567.63 4619.43 5356.79

Bill Clinton is, as has been noted by many pundits, the most prolix speaker in terms of total number of words. All RAs were taken from the website http://www.presidency. ucsb.edu/satradio.php. All transcripts were saved as text files and all headers were removed. Headers comprised titles such as “The President’s Radio Address” and the date “April 7, 2001.” All word counts shown below in Table 2 are word counts without headers. After saving all 1,066 speeches as individual text files, a meta-file was created for each president (Reagan, Bush Sr., Clinton, and Bush Jr.), and word searches were run using Wordsmith, Version 3 (http://www.lexically.net/wordsmith/). Table 2. RAs included in current corpora Name

Year

Total

Reagan Bush Sr. Clinton Bush Jr. Total

1981–1988 1989–1992 1993–2000 2001–2006

334 18 407 307 1,066

Political party Republican Republican Democratic Republican

Word count 280,075 11,724 365,616 184,802 842,217

Average no. of words/speech 838.55 651.33 898.32 601.96 790.07

Wordsmith creates a concordance for all instances of the lexical item chosen, with links to the full-text. The concordances for each term were then printed out and read through. Acronyms were excluded from the analysis.7 Data on the number of instances found were then saved into Excel spreadsheets for further analysis.

Examining lexical frequency patterns 173

4. Data analyses For the analysis, I calculated the total token count and normalized that to the number of total tokens per 10,000 words in order to determine the overall pattern of usage. I then compared the frequency of use of NP lexemes and SF lexemes in each corpus for each president with the overall frequency of use in the BNC using z-statistics. All presidents use the NP lexemes and the SF lexemes significantly more frequently than is found in the BNC (p < .05). The fact that both SF lexemes and NP lexemes occur more frequently in RAs and SOUAs than in the general corpus indicates that these lexemes represent a distinctive part of political language. Next, I compared whether proportions of lexemes used by the presidents significantly differ.8 It can be seen in Table 3 below that Clinton differs significantly from the Republicans overall in terms of NP lexeme usage, both in the two individual corpora and in the combined corpus. In addition, in terms of individual comparisons, Clinton differs in NP lexical usage from Reagan in both the two single corpora and in the combined corpus. He also differs from Bush Jr. in the combined corpus and in the RAs but not in the SOUAs. Lastly, he does not differ significantly from Bush Sr. either in the individual corpora or in the combined corpus. The first finding, that Clinton uses NP lexemes more often than the three Republican presidents, supports the hypothesis put forward here – that Clinton’s viewpoint is fundamentally different from that of the Republicans. As a Democrat, he values the concepts of nurturance and empathy more highly, as evidenced by the fact that he uses words related to these concepts more often in both genres, the SOUAs, which speak to the general American public, and the RAs, which speak to his core constituents. While Republicans as a group also use these lexemes, they do so significantly less frequently than Clinton. However, individual comparisons between the presidents also paint an interesting picture. While Clinton uses NP lexemes more often than Bush Sr., this difference does not reach significance in either corpus, indicating that perhaps political pundits were correct in claiming that Bush Sr. moderated his language to a certain extent to appeal to both Democrats and Republicans. Bush Jr., on the other hand, uses significantly fewer NP lexemes than Clinton in RAs but not in SOUAs. This seems to indicate his ability to control his language so that he can appeal to both the general audience of Americans and the more narrow audience of his core constituency, who is more likely to listen to his RAs.

174 Kathleen Ahrens This core constituency of Republicans is hypothesized to be cognizant and accepting of the viewpoint found in the SF model, while the general American audience can be considered to hold a wider range of views. In Table 4 below, the pattern is slightly different. As predicted, Republicans do use lexemes from the SF model more often than Clinton does in the SOUAs, RAs, and the combined corpus. In addition, with the exception of the comparison between Bush Jr. and Clinton in the SOUAs and Bush Sr. in the RAs, Republican presidents use significantly more SF lexemes in the SOUAs, the RAs, and the combined corpus. The results from Tables 3 and 4 demonstrate that Reagan follows the conceptual viewpoint prototypically postulated for a Republican president – he uses more SF lexemes and fewer NP lexemes than Clinton. Bush Sr., however, shows no difference from Clinton in terms of the frequency of use of NP-related lexemes, but he does use more SF-related lexemes in his SOUAs, and, surprisingly, shows no difference from Clinton with respect to the use of SF-related lexemes in the RAs.9 This contrasts with what is found for Bush Jr. When his lexeme usage is compared with Clinton’s, Bush Jr. doesn’t differ significantly for either SF lexemes or NP lexemes in the SOUAs. However, when speaking to his core constituency in his RAs, he does differ significantly from Clinton by using more SF lexemes and fewer NP lexemes. Bush Jr. is able to modulate his language to appeal to the general American population by using more compassionate terminology (hence, the reason for his title as a “compassionate conservative”) in the general SOUAs, while using more authoritative and strength-related lexemes when talking to his base in the RAs. In Table 5 below, a direct comparison is made between the proportion of SF lexemes and NP lexemes used by each president.10 When the proportion of the NP lexemes versus SF lexemes in the SOUA and RA corpora is compared for each president, Clinton and Reagan both clearly follow the predictions made. Clinton uses significantly more NP lexemes than SF lexemes. Reagan, on the other hand, uses significantly more SF lexemes than NP lexemes. Bush Sr. does not show any difference in usage between SF lexemes and NP lexemes in either corpus. Bush Jr. uses roughly the same number of NP lexemes and SF lexemes in his SOUAs, but, as noted in the above discussion, significantly more SF lexemes in his RAs. Thus, unlike his father, Bush Jr. adjusts his language for the general American public in his SOUAs, but speaks the language of the SF model when he is talking to

+ RA

SOUA

RA

SOUA

Corpora

Democrat

Clinton Clinton Clinton Clinton Clinton Clinton Clinton Clinton Clinton Clinton Clinton Clinton

Republicans

Reagan Bush Sr. Bush Jr. Republicans Reagan Bush Sr. Bush Jr. Republicans Reagan Bush Sr. Bush Jr. Republicans

Rep. Pres. 31 36 65 132 207 19 278 504 238 55 343 636

Total lexemes (raw)

36652 20461 32336 89449 280075 11724 184802 476601 316727 32185 217138 566050

Rep. Pres.

Corpora size Clinton 151 151 151 151 816 816 816 816 967 967 967 967

Total lexemes (raw) Clinton 60541 60541 60541 60541 365616 365616 365616 365616 426157 426157 426157 426157

Corpora size

-5.76 -1.89 -1.46 -4.46 -14.95 -1.39 -5.72 -13.5 -16.08 -2.0547 -5.8007 -14.06

z

0.00** 0.03 0.07 0.00** 0.00** 0.08 0.00** 0.00** 0.00** 0.02 0.00** 0.00**

p

Table 3. Z-statistic for Republicans versus Clinton (Democrat) in SOUAs, RAs, and SOUAs+RAs, comparing lexemes in the NP model

Examining lexical frequency patterns 175

+ RA

SOUA

RA

SOUA

Corpora

Democrat

Clinton Clinton Clinton Clinton Clinton Clinton Clinton Clinton Clinton Clinton Clinton Clinton

Republicans

Reagan Bush Sr. Bush Jr. Republicans Reagan Bush Sr. Bush Jr. Republicans Reagan Bush Sr. Bush Jr. Republicans

Rep. Pres. 157 50 67 274 546 17 524 1087 703 67 591 1361

Total lexemes (raw)

36652 20461 32336 89449 280075 11724 184802 476601 316727 32185 217138 566050

Rep. Pres.

Corpora size Clinton 92 92 92 92 570 570 570 570 662 662 662 662

Total lexemes (raw) Clinton 60541 60541 60541 60541 365616 365616 365616 365616 426157 426157 426157 426157

Corpora size

8.26 2.73 1.94 5.94 3.74 -0.29 10.04 7.41 6.63 2.29 10.05 9.30

z

0.00** 0.00** 0.03 0.00** 0.00** 0.38 0.00** 0.00** 0.00** 0.01* 0.00** 0.00**

p

Table 4. Z-statistic for Republicans versus Clinton (Democrat) in SOUAs, RAs, and SOUAs+RAs, comparing lexemes in the SF model

176 Kathleen Ahrens

Examining lexical frequency patterns 177

a narrower audience, one that agrees with his own worldview. However, Bush Sr.’s small corpus size, especially with respect to his RAs, leaves open the possibility that the findings presented for his corpus could potentially change if the word count increased. Table 5. Comparisons of NP lexemes versus SF lexemes (raw frequency) in SOUAs, RAs, and SOUAs+RAs corpora Corpora

SOUA

RA

SOUA + RA

Presidents Reagan Bush Sr. Bush Jr. Clinton Reagan Bush Sr. Bush Jr. Clinton Reagan Bush Sr. Bush Jr. Clinton

Total NP lexemes 31 36 65 151 207 19 278 816 238 55 343 967

Total SF lexemes 157 50 67 92 546 17 524 570 703 67 591 662

Corpora size

z

36,652 20,461 32,336 60,541 280,075 11,724 184,802 365,616 316,727 32,185 217,138 426,157

-9.19 -1.51 -0.17 3.78 -12.35 0.33 -8.69 6.61 -15.16 -1.09 -8.11 7.56

p 0.00** 0.07 0.43 0.00** 0.00** 0.37 0.00** 0.00** 0.00** 0.14 0.00** 0.00**

In sum, the findings from this study on lexical frequency patterns suggest that it is possible to search for patterns of lexical usage to examine whether speakers are biased toward a particular conceptual model, and, moreover, whether they adjust that model depending on the audience they are addressing.

5. Collocational patterns The above analyses looked solely at the numbers of instances in which particular lexemes were used and have drawn conclusions based on statistical analyses of these lexical usages. However, collocational patterns can also elucidate the differences between the Republican presidents and Clinton. For example, a comparison between Tables 6 and 7 below demonstrates that Clinton is more concerned with health care and child care than Republicans are (especially since the Republican’s corpus is over 100,000 words larger). Clinton also uses the form care more often in active predicates such as to care and take care. In addition, Clinton’s discussion of health care includes quality care, preventive care, managed care, home

178 Kathleen Ahrens

care, term care, and medical care. The Republicans only refer to managed care, term care, and medical care and do so much less frequently than Clinton. Table 6. Lexemes found immediately to the left of care in Clinton’s RAs Lexeme health child to medical managed

Raw frequency count 406 65 28 24 21

Lexeme quality home preventive take term

Raw frequency count 16 12 11 11 10

Table 7. Lexemes found immediately to the left of care in Republican’s RAs Lexeme health child and medical to

Raw frequency count 137 11 7 7 7

Lexeme managed of take term the

Raw frequency count 4 4 4 4 4

This suggests that Clinton’s use of care follows the hypotheses put forward in the NP model, where the nation is a family, and the morality of a family is judged upon its ability to nurture and take care of its members. In addition, when we compare the lexemes found immediately to the left of force in Clinton’s RAs with the Republican RAs in Tables 8 and 9 below, it is obvious that Clinton’s focus is on a work force or task force while Republicans are split between the concepts of a work/task force and the air/multinational/military force. Table 8. Lexemes found immediately to the left of force in Clinton’s RAs Lexeme work task to air the

Raw Frequency count 24 12 7 6 6

Lexeme strongest a and fighting implementation

Raw Frequency count 5 4 4 4 4

Examining lexical frequency patterns 179 Table 9. Lexemes found immediately to the left of force in the Republican’s RAs Lexeme air task work a to

Raw Frequency count 23 22 20 10 10

Lexeme multinational of by military and

Raw Frequency count 9 8 7 6 5

In addition, if lexical meanings are taken into account, such that the use of force in work force is not counted as an instance that has to do with STRENGTH, the data presented above would change. However, although a detailed analysis of the use of each lexeme is precluded by the length of this paper, a preliminary analysis indicates that counting only the related senses would continue to support the hypothesis that Clinton (a Democrat) uses more word senses with the NP model, and the Republicans use more word senses associated with the SF model. In line with the proposal that word senses should be taken into account, it should also be noted that it is, in fact, difficult in many cases to assign word senses in context. On the one hand, done on an individual basis, it is time consuming. On the other hand, words may have more than one sense within a particular context, or it may be difficult to decide which sense is intended by the speaker in that context (Ahrens, Chang, Chen, and Huang, 1998). One way to deal with this issue would be to look at the collocations, as in the above tables, and identify which collocational patterns are related to the meaning of strength (e.g., one possibility is that a force, to force, military force, multinational force, and air force count as being related to the sense of strength). Of course, different researchers may have different ideas about whether or not these meanings (or others) should be classified as having to do with strength. This is why, for this first study in using lexical frequency patterns, the lexeme was considered the appropriate level on which to run the analyses. However, as long as the sense identification method is principled (something that was not explicated in the above example) and replicable, it should be considered as an additional level of analysis in future lexical frequency pattern studies.

180 Kathleen Ahrens

6. Conclusion The proposal put forth in this paper is that lexical frequency patterns may shed light on the underlying conceptual model utilized by a speaker for a particular audience. This proposal assumes that once a conceptual model is proposed, key lexemes related to the model can be identified, and associated lexical items can be found through WordNet’s hypernyms. Next, these items are searched for in the corpora, and proportional differences are examined between groups. And last, collocational patterns (i.e., one to the left and one to the right) of key words should be analyzed to determine whether the detailed linguistic data support the coarser frequency pattern account. In the particular case analyzed in this paper, the conceptual model was identified by Lakoff (1996, 2002) on the basis of conceptual metaphor analysis. The top two source domains for each conceptual paradigm (i.e., STRENGTH and AUTHORITY from the strict father paradigm, and NURTURANCE and EMPATHY from the nurturing parent paradigm) were identified, and associated words were found by taking the nouns and verbs in the WordNet hypernym associated with each concept. A subset of these associated words was searched for in well-defined corpora for two particular political speech genres – the State of the Union Addresses and Radio Addresses. When the proportional data is examined, Lakoff’s (1996, 2002) hypothesis that Democrats and Republicans view the world differently receives support from the Reagan and Clinton data. It also suggests, as political pundits have noted, that Bush Jr. is adept at directing his messages to a particular audience, since he is using his RAs to renew his ties with his core party members, while moderating his message with lexical items from both conceptual paradigms when he speaks to mainstream Americans in the SOUAs. Preliminary analyses of the collocational patterns of two key words indicate that analyses based on the appropriate senses of the lexemes would also support the above findings, although further study is needed for a more in-depth understanding of the collocational data. In sum, this paper suggests that small, narrowly focused corpora are suitable for identifying different viewpoints through an examination of lexical frequency patterns. This method of testing cognitive models avoids the circularity apparent in many attempts to establish such models based solely on conceptual metaphors and paves the way for a greater understanding of the way humans organize and use language to conceptualize and to persuade.

Examining lexical frequency patterns 181

Acknowledgements An earlier co-authored version of this paper was presented at the Second International Conference of the German Cognitive Linguistics Association in Munich from October 5 to 7, 2006, under the title “Establishing cognitive models with small corpora: A case study of U.S. political speeches” with Siaw-Fong Chung. I would like to thank Siaw-Fong for her help in presenting the paper and for her comments on later versions. Thanks also go to the participants of the conference for their comments on the earlier version. In addition, I would also like to thank Sherry Wu for running the corpora searches and Professor Chung-Ping Cheng for running the statistical analyses, as well as John Edwards and an anonymous reviewer for commenting on this paper. Lastly, I would like to acknowledge NSC research funding for this work (Grant # 94-2411-H-002-038). All remaining errors are my own. Correspondence may be directed to: Kathleen Ahrens, Hong Kong Baptist University, 224 Waterloo Road, Kowloon Tong, Kowloon, Hong Kong, or to: [email protected].

Notes 1.

2.

3.

4. 5.

See Lakoff (2002: 99) for the list of metaphors in the SF model, and Lakoff (2002: 135) for metaphors in the NP model. Lakoff discusses how these metaphors are hierarchically ordered. That is, while the SF model also has NURTURANCE as a metaphor, it is lowest on the hierarchy, and, furthermore, exists to promote MORALITY AS STRENGTH. Note that we selected one sense for strength and authority from the list of senses in WordNet 3.0 (sense three for strength and sense one for authority). Only one sense is listed in WordNet for nurturance and empathy. C-Span incorrectly lists Carter as giving a SOUA in 1981. The actual file under Carter’s name is, in fact, Reagan’s first SOUA (as accessed on February 6, 2006, at http://www.c-span.org/executive/transcript.asp?cat=current event& code=bush admin&year=1981). This has not been included in the Carter corpus compiled for this paper, but it is, of course, included in the Reagan corpus. “Bush Sr.” refers to George H. W. Bush, and “Bush Jr.” refers to George W. Bush, following the usage in Charteris-Black (2005). Microsoft Word counts the dash punctuation mark as one word when it is written as two short hyphens close together with a space on either side or as one short hyphen with a space on either side (i.e., as “ -- ” or as “ - ”). When it

182 Kathleen Ahrens is written as a long, unbroken line without spaces on either side, as in “—”, the program does not count it as a word. In addition, in some speeches, only a short hyphen is used without a space on either side, which the word-count program then interprets as a hyphenated word. Ideally, for the most precise word count possible, each speech should be re-edited for uniformity among the various types of dash marks used. However, such editing carries the risk of altering the intent of the original and was not carried out for this study. 6. Please note that the word count differs slightly from Ahrens (2006), as the headers for the speeches were removed for this study. 7. AIDS was the only acronym found that needed to be excluded. 8. Since we are comparing proportions, we calculate the relevant z-statistic and set the alpha-level for significance at .05 for comparison between Democrats (i.e., Clinton) and Republicans (one asterisk indicates p < .05, and two asterisks indicate p < .01). However, in order to correct for multiple comparisons, we use the Bonferroni correction when comparing directly between individual presidents. In these cases, the alpha-level is .01 (one asterisk indicates p < .0167, and two asterisks indicate p < .0033). The z-statistic is used to compare proportions. For example, in this paper, two kinds of lexemes (A1 and A2) are being compared in a particular president’s corpus of speeches. The proportion of A1 (p1) and the proportion of A2 (p2) cannot be compared directly because frequencies of lexemes differ overall in a large-scale corpus. However, this can be dealt with by testing the ratios of frequencies based on the result of Scott and Seber (1983). For example, if the ratio of frequencies of two kinds of lexemes in a large-scale corpus is a constant, say a (in other words, p1 / p2 = a), then what is at issue is whether the ratio in a politician’s speech is higher than the ratio in a large-scale corpus. So the null hypothesis can be set as p1 = a * p2, or, equivalently, as p1 − a * p2 = 0. Thus, what is being compared is the difference between p1 and a*p2 rather than the difference between p1 and p2. If the null hypothesis is rejected, and p1 − a * p2 < 0, then it means that this politician says more A2 lexemes than A1 lexemes relative to a large-scale corpus. If the null hypothesis is rejected, and p1 − a * p2 > 0, then this politician says more A1 lexemes than A2 lexemes relative to a large-scale corpus (Professor Cheng Chung-Ping, personal communication, 12 January 2009). 9. It must be noted in this regard that the corpus size for Bush Sr.’s RAs is very small compared to those of other presidents. 10. Again, in order to correct for multiple comparisons, we use the Bonferroni correction when comparing the proportion of SF lexemes and NP lexemes used. Thus, in Table 5, the alpha-level is .01 (one asterisk indicates p < .0167, and two asterisks indicate p < .0033).

Examining lexical frequency patterns 183

References Ahrens, Kathleen 2006 Using a small corpus to test linguistic hypotheses: Evaluating “people” in the State of the Union Addresses. International Journal of Computational Linguistics and Chinese Language Processing 11 (4): 377–392. Ahrens, Kathleen, Li-Li Chang, Kei-Jiann Chen, and Chu-Ren Huang 1998 Meaning representation and meaning instantiation for Chinese nominals. International Journal of Computational Linguistics and Chinese Language Processing 3 (1): 45–60. Ahrens, Kathleen, and Sophia Lee 2009 Gender versus Politics: When conceptual models collide in the US Senate. In Politics, Gender and Conceptual Metaphors, Kathleen Ahrens (ed.), 62–82. London: Palgrave Macmillan. Charteris-Black, Jonathan 2004 Corpus Approaches to Critical Metaphor Analysis. London: Palgrave Macmillan. Charteris-Black, Jonathan 2005 Politicians and Rhetoric: The Persuasive Power of Metaphor. London: Palgrave Macmillan. Cienki, Alan 2005 Metaphor in the “Strict Father” and “Nurturant Parent” cognitive models: Theoretical issues raised in an empirical study. Cognitive Linguistics 16 (2): 279–312. Hart, Roderick, Sharon Jarvis, William Jennings, and Deborah Smith-Howell 2005 Political Keywords: Using Language that Uses Us. Oxford: Oxford University Press. Koller, Veronica, and Elena Semino 2009 Metaphor, politics and gender: A case study from Germany. In Politics, Gender and Conceptual Metaphors, Kathleen Ahrens (ed.), 9– 35. London: Palgrave Macmillan. Lakoff, George 1996 Moral Politics: What Conservatives Know that Liberals Don’t. Chicago: University of Chicago Press. Lakoff, George 2002 Moral Politics: How Liberals and Conservatives Think. 2d ed. Chicago: University of Chicago Press. Lim, Elvin 2002 Five trends in presidential rhetoric: Analysis of rhetoric from George Washington to Bill Clinton. Presidential Studies Quarterly 32: 328– 366.

184 Kathleen Ahrens Lim, Elvin 2008 The Anti-Intellectual Presidency. Oxford: Oxford University Press. Scott, Alastair J., and George A. F. Seber 1983 Difference of proportions from the same survey. The American Statistician 37: 319–320. Scott, Mike 2001 Mapping key words to problem and solution. In Patterns of Text: In Honour of Michael Hoey, Mike Scott, and Geoff Thompson (eds.), 109–127. Amsterdam/Philadelphia: Benjamins. Semino, Elena, and Veronica Koller 2009 Metaphor, politics and gender: A case study from Italy. In Politics, Gender and Conceptual Metaphors, Kathleen Ahrens (ed.), 36–61. London: Palgrave Macmillan.

Metaphor, constructional ambiguity and the causative resultatives Beate Hampe

1. Introduction: Metaphor in the causative resultatives Of all argument structures in the “resultative family” (Goldberg and Jackendoff 2004), the so-called causative resultatives, comprising the Caused-Motion Construction (CMC, cf. 1)1 and the Resultative Construction (RC, cf. 2), have received by far the largest amount of attention from construction grammarians (Goldberg 1995; Goldberg and Jackendoff 2004: 535–538). This is partly due to early postulations in Construction Grammar, which claimed the latter to be a metaphorical version of the former (Goldberg 1995: 81–94). (1)

The warm air pushes other air [PP out of the way]. (W2B-025)

(2)

If you have fresh maggots, riddle them [AdjP clean] of the sawdust. (W2D-017)

Traditionally, the causative resultatives are also known as “complextransitive” (henceforth cxtr.) clause patterns (e.g. Quirk et al. 1985), because in these patterns the object-NP is followed by an object-related adverbial (cf. 1) or complement/predicative (cf. 2), formally most typically realized as a prepositional phrase or an adjectival phrase, respectively. In construction grammar, object-related adverbials and predicatives have been subsumed under the notion of the resultative phrase (RP), because both designate the endpoint of a change – either of location (cf. 3a) or of state (cf. 3b) – which is intentionally caused by the referent of the subject-NP and undergone by the object-NP referent, the so-called object host (Goldberg and Jackendoff 2004). (3)

a. b.

Semantics CMC: Semantics RC:

[X CAUSE (Y MOVE Z)] [X CAUSE (Y BECOME Z)]

186 Beate Hampe It is well-known that RPs in the CMC can either specify a literal goal, as in (1) above, or a metaphorical one, as in (4) below (Goldberg 1995: 81–94). Literal caused-motion examples relate to a scenario of direct causation, in which some sort of force is transferred from an agent to a patient bringing about the patient’s change of location. Metaphorical caused-motion examples denote the causation of a change of state, in accordance with a range of primary metaphors from the location branch of the EVENT-STRUCTURE 2 METAPHOR complex (Lakoff 1993; Grady 1997). Consequently, causedmotion verbs take on resultative meanings in metaphorical instances of the CMC (cf. 4a–c) and will more often than not also appear in the RC itself (cf. 5a–c).3 (4)

a. b. c.

It’s just a question of putting them [RP in order]. (S1A-040) A fateful process had been set [RP in motion]. (S2B-029) At times it drove his audience [RP to anger and boredom]. (BNC-AA9)

(5)

a. b. c.

So he puts this [RP right] and I sort of stand by. (S1A-082) Oil wells have been wantonly set [RP alight]. (S2B-014) There 's an echo in here that 's going to drive me [RP mad]. (S2A-052)

In the standard construction-based account (cf. Goldberg 1995: 81–94), these observations motivated the postulation of a metaphorical “inheritance link” between the two argument-structure constructions, i.e. the claim that the RC itself “crucially involves a metaphorical interpretation of the result phrase as metaphorical type of goal” (cf. Goldberg 1995: 93–94). From this perspective, (rare) occurrences of resultative verbs like make in causedmotion syntax look like re-motivations making the original metaphor visible: (6)

The technical difficulty at the moment with these new superconductor materials is to turn them [PP into wires] because the they’re ceramic and so uh as you can imagine it’s very hard to make them [PP into wires]. (S1A-088)

The standard account has been criticized mainly because individual verbs usually participate in this web of metaphorical extensions in a highly idiosyncratic fashion only, thus restricting the productivity of the metaphorical

Metaphor, constructional ambiguity and the causative resultatives 187

inheritance link (cf., e.g., Taub 1996; Boas 2003; Goldberg and Jackendoff 2004: 558–563). Existing combinations of individual verbs with non-spatial RPs behave like idiomatic expressions, such that the lexicalization of a specific pairing of a verb with a metaphorical adverbial or an adjectival predicative as RP usually pre-empts alternative realizations. Examples (7a, b) show the metaphorical instantiations of the CMC given in (4a, b) to be relatively resistant to even moderate lexical or syntactic variation. (7)

a.

b.

It’s just a question of putting them [RP in order]/ ~ ?[RP in chaos]/ ~ [RP right]/ ~ ??[RP correct]/ ~ *[RP wrong]/ ~ *[RP in the right]/ ~ *[RP in the wrong]. A fateful process had been set [RP in motion]/ ~ ?[RP in progress]/ ~ *[RP moving].

In view of the considerable semantic overlap between the CMC and the RC and the apparent limitations of the productivity of the postulated metaphorical inheritance link, the more recent literature has emphasized the similarities between the various kinds of causative resultatives and tended to reject the hypothesis of a metaphorical inheritance link between the CMC and the RC (cf. Boas 2003: 116). ‘Change-of-location’ is frequently viewed as a special case of the more generic category of ‘change-of-state’ (cf. e.g. Bencini and Goldberg 2000: 646; Boas 2003: 4–11; Broccias 2003: 2; Goldberg and Jackendoff 2004). This is also reflected by the inclination to subsume both object-related adverbials and object-complements under the formally and functionally underspecified category of the resultative phrase (RP). Although well-motivated, this move downplays the differences between the phrases serving as RPs: the presence of an object-related directional/ locative adverbial in the CMC always profiles a spatial event-construal, which is retained in all metaphorical uses of caused-motion verbs (cf. 4b, 5b) and even in cases where a resultative verb appears in the CMC (cf. 6). Though Goldberg’s original distinction between the CMC and the RC is transformed, for instance, into Goldberg and Jackendoff’s (2004) semantic distinction between ‘path’ and ‘property’ resultatives, this does not suffice, as analytical distinctions without a clear separation of formal and functional criteria cannot handle metaphorical instances of the CMC, which are formally ‘path’ resultatives, but semantically ‘property’ resultatives.

188 Beate Hampe

2. Goal and methods of this study The present study is usage-based, both in the well-known sense of a theoretical position which assumes that usage both shapes and reflects the language users knowledge of the language (cf. Barlow and Kemmer 2000), and in a more narrowly defined, corpus-linguistic sense. It relies on wellestablished methods from quantitative corpus linguistics (Stefanowitsch and Gries 2003, 2005; Gries and Stefanowitsch 2004a, 2004b) in order to adequately characterize usage in the first place. The paper is concerned with two related issues, the first of which concerns the empirical validation of the original distinction between the CMC and the RC and of the respective semantic characterizations of these argument structures as symbolic units. Secondly, the results of this analysis will provide the basis of a re-assessment of the role of conceptual metaphor in the family of cxtr. argument-structure constructions. Specifically, it will be shown that, for adult speakers, the semantic potential of the syntactic pattern known as the RC is not exhaustively captured by existing postulations about the constructional semantics and its extensions by regular polysemy links (Goldberg 1995). The presence of a second central use of this syntactic pattern, which has recently been described as belonging to a separate family of constructions (cf. GonzálvezGarcía 2008) and whose relation to the causative resultatives has not been investigated at all, threatens the semantic unity of the RC as a symbolic unit. This study thus revisits mechanisms so far postulated for extending constructional networks both semantically, i.e. within constructions (via polysemy links), and syntactically, i.e. across constructions (via inheritance links). Viewing metaphorical extensions as a strictly local, lexically determined phenomenon, and emphasizing the role of verb-class based constructions (vis-à-vis totally schematic ASCs), this study works towards an alternative account of the growth of a constructional network. Concerning the corpora employed, the main part of this study investigates the full set of the relevant constructions in the adult data provided by the syntactically parsed ICE-GB (1 million words). Additionally, child language data are made use of, but in a less systematic way. A provisional patch-work corpus compiled from various files of the CHILDES database provides the material for two age groups. The Manchester subcorpus, containing data from the regularly recorded play sessions of 12 children, was used for the age group from 24 to 36 months. Due to extreme data sparsity in the age group from 6 to 8 years, however, the data used come from all

Metaphor, constructional ambiguity and the causative resultatives 189

British/American subcorpora of CHILDES, which provide any substantial amount of relevant material at all (see Appendix, Table 4, Note 1). All verbs parsed as in the ICE-GB were manually coded with respect to the syntactic realization of their RPs: Going beyond the standard characterization of the CMC, I coded all verbs with object-related adverbials in their complementation as instances of the CMC, no matter whether these adverbials were directional or locative ones, and no matter whether these were formally realized as prepositional phrases (cf. 1, 3), adverb phrases (cf. 8a), spatial noun phrases (cf. 8b), or even spatial adverbial particles (cf. 8c). The latter were treated as intransitive counterparts of prepositions, where the landmark of the spatial relation (i.e. the object of the preposition) is left implicit. (8)

a. b. c.

(9)

And Martin Brundle will bring Jaguar number three [AdvP home] in third place. (S2A-012) If we take the water away, we can push the reaction [NP the other way]. (S2A-034) It is certainly the case that there has been a tendency uh to bring [PP into those schemes] more uh part-time workers … with the result that many of those groups uh have been increasingly brought [Adv Part in] (S1B-058) [ + passive construction]

This makes Bonfire Night [NP the enemy of advertising]. (W2E-003)

In contrast, cxtr. argument structures with predicatives in the form of NPs (cf. 9) were separated from those with adjectival predicatives and excluded from further consideration for two reasons. Firstly, the standard account (Goldberg 1995) characterizes the RC as exhibiting an adjectival RP. Secondly, the conceptualizations referred to by nominal RPs differ in kind from those denoted by adjectival ones. While the former denote relatively autonomous, non-relational conceptualizations, i.e. “regions” in some cognitive domain (Langacker 1991: 63–74; Taylor 2002: 343–365), the latter modify the conceptual content associated with nouns, i.e. denote relations that take nouns as their trajectors. Such a formal variation in an argument structure must be assumed to incur considerable functional differences. Hampe (submitted) shows that the cxtr. pattern with nominal objectcomplement is indeed an ASC of its own (i.e., the English Denominative Construction, DC) with a central meaning different from that of the RC.

190 Beate Hampe Lexical items such as call, date, name, term, entitle, and label are representative of its most closely associated collexeme class. Issues of further constructional distinctions based on differences in the formal realization of the predicative, however, are beyond the scope of the present paper. The frequency lists obtained by these retrieval and coding procedures were analyzed by means of the various methods from collostruction analysis (Stefanowitsch and Gries 2003, 2005; Gries and Stefanowitsch 2004a, 2004b), i.e. simple, distinctive, and co-varying collexeme analyses.4 A simple collexeme analysis determines the degree of association or attraction between a constructional pattern and the lexemes instantiating it in a corpus. Crucially, the top-rank collexemes give access to an adequate semantic characterization of the constructional pattern itself. A distinctive collexeme analysis makes apparent semantic differences between closely related constructions by directly comparing the respective collexeme sets with one another. A co-varying collexeme analysis of all realizations of one construction in a corpus, finally, can determine patterns of lexical co-variation between two of its syntactic slots. Concerning this study, the two most relevant syntactic slots are represented by the main verb and the head of the resultative phrase. Overall, the quantitative analysis of the adult data is employed to determine the major uses of the cxtr. argument structures with object-related adverbials and adjectival predicatives, respectively, – and thus the range of constructional distinctions to be made for these argument structures. The developmental data gathered from the CHILDES database are used to check the general plausibility of potential motivational links between the various central uses of the cxtr. argument structures. A collostructional analysis of the child language data was not carried out because of the provisional nature of the corpus itself and the kind of data retrieval performed. For reasons of feasibility, only a subset of all cxtr. verb uses in the child data was retrieved by means of a lexical search for all occurrences of the most closely associated and most highly distinctive adult collexemes of the argument structures with object-related adverbials and adjectival predicatives (cf. Appendix, Table 5, Note 2).

3. Results and discussion of the collexeme analyses The ICE-GB query for all VPs parsed as returned 4,019 sentences, which were found to contain 3,514 cxtr. argument structures containing a

Metaphor, constructional ambiguity and the causative resultatives 191

total of 3,707 RPs (due to occurrences of multiple RPs in cases of verbal ellipsis). Of these, 1,937 verb tokens occur with one or more object-related adverbials, and 908 verbs occur with one or more adjectival predicatives. If corrected for verbal ellipsis, the total number of true hits increases to 2,083 and 951, respectively.

3.1. Lexical overlap and the distinction between the CMC and the RC The above-mentioned semantic overlap between the CMC and the RC is reflected by the large lexical overlap in the instantiations of the respective argument structures, consisting of 37 shared lexical types amounting to 10.8% of all types instantiating these argument structures in the corpus. Among these are also the two highly frequent generic verbs put and make (cf. 10, 11), which are usually characterized as “path-breaking” in the acquisition of the CMC and RC, respectively (e.g., Goldberg 2006: 77–79). (10)

a. b. c.

(11)

a. b.

CMC, spatial: And we put cucumber and lemon and orange [PP in the Pimms]. (S1A-005) CMC, metaphorical: I thought I’d be able to put his mind [PP at rest] quite easily. (W2F-011) RC: So he puts this [AdjP right] and I sort of stand by. (S1A082) RC: But I think making people [AdjP aware that anybody can do it], uhm, is is quite important. (S1A-002) CMC: Tables 3, 4, 5, for example, could usefully be made [PP into one table]. (W1B-025)

Observations of this sort cannot be used to argue against the distinction between the CMC and the RC, however, because the collexeme analysis shows that these verbs are not “attracted to” the argument structures illustrated in (10c) and (11b), respectively, and occur in it at no more than chance level: 8 out of 774 of all tokens of put in the ICE-GB occur in the RC (coll. str.: 0.850, p > 0.05). Similarly, only 24 out of the 1,951 tokens of make in the ICE-GB occur in the CMC (coll. str.: -0.521, p > 0.05). Secondly, the lexical restrictions concerning the realizations of their RP heads are severe. At a level of significance smaller than 0.001, make only correlates with spatial RPs with into (coll. str.: 4.75, rank 12 in the co-

192 Beate Hampe varying collexeme analysis), mimicking the second-strongest co-varying collexeme pair in the ICE-GB data (turn into, coll. str.: 19.91) and redundantly importing its metaphor of change. The caused-motion verbs put and leave only correlate with the adjectival RP heads right and alone, respectively (coll. str.: 8.209 and 8.030, p < 0.001). What is noteworthy here is that in all these cases the verbs are not by themselves attracted to the alternative syntactic pattern, while their combinations with very few RP heads (put X right, leave X alone, make X into Y, etc.) are among the two patterns’ strongest co-varying pairs (see Appendix, Table 4). Though the limited corpus size of the ICE-GB prevents the full identification of the lower-level schemas involved, the results suggest that the respective expressions should be treated as symbolic units at a much lower level of generality than that of the ASC they instantiate. They should best be regarded as phraseological units with lexically schematic NP slots, akin to other multi-word verbs. It will be shown next that the results of the collexeme analysis clearly support the constructional distinction between the CMC and the RC. A closer look at the top ranks of the collexeme lists also makes clear, however, that – at least for adult speakers – the semantic potential of the argument structure with adjectival predicatives is wider than previously claimed.

3.2. Central collexeme classes in the AS with object-related adverbial The collexemes most strongly attracted to and most distinctive for the AS with object-related spatial adverbials are verbs that express the subject-NP referent’s manipulation of the locations of object-NP referents (see Appendix, Table 1). The results suggest the centrality of two basic uses, corresponding to the difference between directional and locative adverbials. These are represented by two high-ranked collexeme classes, the strongest representatives of which are the lexical causatives put and keep, respectively. In accordance with existing claims about the CMC, the first, most central group realizes the caused-motion use of this AS (cf. 12a): the verbs take directional adverbials and, when used literally, the resulting expressions refer to the causation of a change of the object-referent’s location. The verbs in the second group (cf. 12b) take locative adverbials and the resulting expressions refer to the prevention rather than causation of motion, i.e. to the forced maintenance of an object-referent’s location. I will

Metaphor, constructional ambiguity and the causative resultatives 193

subsequently call these the forced-location uses of the argument structure with object-related adverbials. (12)

Top collexemes in the AS with object-related adverbial (rank): put (1), place (3), bring (4), get (5), set (6), take (8), turn a. (9), send (12), push (13), shove (15), force (16), lay (18) keep (2), leave (7), bear (11), hold (14), base (17) b.

As the difference between directional and locative adverbials is not reliably morpho-syntactically marked in English, with many prepositions exhibiting both a locative and a directional meaning, the difference between causedmotion and forced-location uses must be regarded as a lexically determined one, achieved only in conjunction with the verb. As forced-location uses are moreover force-dynamically related to caused-motion uses (cf. also Goldberg 2006: 108), a moderately extended description of the semantics of the CMC (cf. 3a) which encompasses both caused-motion and forcedlocation uses is well motivated: [X CAUSE (Y MOVE TO/REMAIN AT Z)]. To these two central verb groups a third and again closely related one must be added, referred to here as the perceived-location use of this argument structure. It is represented by the strong, but isolated collexeme no 10 (find), which takes locative adverbials and denotes the subject-referent’s identification or perception of the object-referent’s location.5 Its RP is most typically headed by the preposition in, its strongest co-varying collexeme (coll. str.: 20.002, p < 0.001). Note also that the co-varying collexeme pair find X in is the strongest one of all the lexical realizations of this AS in the ICE-GB (see Appendix, Table 4). Analogous to what is known about the various metaphorical extensions of single caused-motion verbs, the location identified can be either a literal or a metaphorical one (cf. 13a, b). In the latter case, the meaning expressed concerns the identification of a state or condition, rather than location, though the construal is still spatial. The metaphorical use is motivated by a set of primary metaphors, which partially overlaps with those responsible for the resultative uses of causedmotion verbs. These include STATES ARE LOCATIONS/BOUNDED REGIONS and CIRCUMSTANCES/CONDITIONS ARE SURROUNDINGS in conjunction with KNOWING IS PERCEIVING (cf. Grady 1997, Appendix). (13)

a.

literal (perceived location): Their car was found [PP in the desert]. (S2B-004) [+ passive construction]

194 Beate Hampe b.

metaphorical: I must accept the conditions I find myself in. (W1B-008) [+ relative clause construction]

In line with what has been claimed about the CMC, the statistical analysis of the corpus data shows caused-motion lexemes which are lexical causatives and represented best by put to be absolutely central: Firstly, a very large class of caused-motion verbs occupies nearly all of the top ranks of the collexeme list (cf. 12a). Secondly, the distinctive collexeme analysis reveals that exactly this group of caused-motion verbs (put, bring, place, take, turn, send) is also distinctive for the pattern in direct comparison with the pattern with adjectival object-complements, so far described as the RC (see Appendix, Table 3). And thirdly, caused-motion verbs from this group are the first ones to occur in cxtr. syntax in child language. They present the overwhelming majority of all verb uses within the entire network of transitive causatives during childhood (both in terms of token and type frequencies), and still dominate the complex transitive network in adulthood (see Appendix, Table 5). The collexeme class representing the forced-location use is secondary in that it contains much fewer closely associated lexical types (keep, leave, have, bear). Complex-transitive find is an isolated collexeme even.6 Moreover, neither the two strongest forced-location collexemes, keep and leave,7 nor cxtr. find are distinctive for the AS with adverbial in the comparison to the AS with adjectival object-complement, as they are also very strongly attracted to the latter. In sum, the traditional characterization of the AS with adverbial as the CMC characterizes its prototypical use and thus presents an abbreviation for a range of closely related meanings to do with the manipulation and perception of object locations.

3.3. Central collexeme classes in the AS with adjectival predicative The results of the simple and distinctive collexeme analyses for the AS with adjectival object-complement are mixed in a way that clearly goes beyond the standard characterization of this AS as the RC (cf. Appendix, Table 2). On the one hand, existing claims about the RC are corroborated insofar as lexical causatives such as make, render, and declare are representative of its most strongly attracted and most distinctive collexeme

Metaphor, constructional ambiguity and the causative resultatives 195

group, with generic make being by far the most strongly attracted and most distinctive item in that AS (cf. 14a, see also Appendix, Table 3). (14)

Top collexemes in the cxtr. AS with adjectival predicative (rank): a. make (1), render (5), get (7), declare (9), set (10), adjudge (16) b. keep (3), leave (4), have (7), bear (8) find (cogn.) (2), consider (6), deem (8) c.

Exactly paralleling the results for the cxtr. AS with adverbial, there are secondly the force-dynamically related verbs of the keep class (cf. 14b), which denote the prevention of change or the forced maintenance of a state, henceforth called forced state. The cxtr. AS with adjectival predicative thus centrally expresses ‘manipulation of state’ scenes. To capture this result, the semantics of the RC (cf. 3b) needs to be broadened to include forcedstate uses: [X CAUSE (X BECOME/REMAIN Z)]. On the other hand, the exclusively resultative character of this AS is questionable, because in the adult data an additional collexeme class with a relatively high type variation is strongly attracted to and highly distinctive for this AS (cf. 14c, see also Appendix, Table 3). Its semantics differs from that of the RC and involves the subject-referent’s ascription of a property/attribute to the object referent: [X THINK (X BE Z)]. Accordingly, I will call this additional central collexeme class attributive (cf. Hampe and Schönefeld 2006). The verbs involved in such attributive uses are not lexical causatives, but mental verbs (cf. 15), the strongest of which is find: (15)

a. b.

The Franks found it [AdjP hard] to rule Brittany. (W1A-003) Certain political and religious beliefs are considered [AdjP normal] elsewhere. (W1A-007) [+ passive construction]

The strongest co-varying collexemes of find in that AS in the ICE-GB (difficult, hard, useful, guilty, see Appendix, Table 4) show that find is used here in an extended, truly cognitive sense, which is more abstract than its perceived-location use (both literal and metaphorical). Crucially, the leading attributive collexeme, find(cogn.), immediately follows make in the collexeme ranking. Moreover, the attributive collexemes find, think, and consider even surpass the resultative collexemes render and declare in terms of their distinctiveness values for this AS in comparison with the AS with object-related adverbials.8

196 Beate Hampe Typically, such attributive meanings are expressed by another cxtr. ASC, the so-called as-predicative, which exhibits formally highly variable RPs (cf. 16a–d), but is semantically more restricted in that it strictly excludes resultative uses (cf. Gries, Hampe, and Schönefeld 2005, to appear). (16)

a. b. c. d.

Writing has to be considered as [NP a turning point in human development]. (W1A-012) Say uhm our language tends to be known as [AdjP rich and specific]. (S1B-003) We see the hard ecu as [NFC -ing being useful in the fight against inflation]. (S2B-007) Prince Charles regards what has been projected as [PP at odds with the surroundings]. (W2A-005)

To conclude, the AS with adjectival complements clearly expresses a wider range of meanings than acknowledged in its standard characterization as RC. In addition to the subject-NP referent’s manipulation of the state of the object-NP referent, for which the term resultative shall be kept, the structure also centrally denotes the ascription of a property to the object-NP referent. The latter use is referred to here as the attributive use of the cxtr. argument structure with adjectival object-complements. The identification of two major uses of the adult AS with adjectival object-complements, raises the question, however, whether these two uses present a case of constructional polysemy or homonymy. What is at stake here in case of the latter is the status of this AS as a construction (i.e., the RC), as the existence of two unrelated meanings threatens the semantic unity of the symbolic unit.

4. Metaphorical motivations in the cxtr. argument structures 4.1. Emergence and motivation of attributive meanings A polysemy account would require attributive meanings to be connected to resultative ones by motivating links. The question is therefore whether the attributive use can be shown to fit the standard polysemy account of the RC (Goldberg 1995). This account focuses firstly on metonymic aspects of the causation-of-change scenario, thus listing such polysemy links as

Metaphor, constructional ambiguity and the causative resultatives 197

‘means/manner’, or ‘instrument’ of causation (cf. 17). Relevant verbs are found relatively high up on the collexeme list (see Appendix, Table 2): (17)

a. b.

Every Egyptian scribe must have possessed a knife in order to cut letters free from the whole papyrus roll (S2A-048) She has quickly combed her hair tidy. (W2F-019)

Another well-established type of link is provided by metaphorical extension. It is thus very tempting to hypothesize a metaphorical polysemy link within the RC connecting the attributive collexeme class back to the resultative one. A widely attested metaphor motivating this link would be provided by IDEAS ARE OBJECTS. In this way, the central ‘manipulation-ofobjects’ meaning of the ASC could be extended to a ‘manipulation-ofcategories’ use, in line with (i) well-known metaphorical extensions frequently observed at the lexical level, and (ii) previously established mechanisms of extension in polysemy networks. Despite its apparent elegance, however, the child-language data available to me suggest this approach to be relatively implausible. The attributive use of the syntactic pattern with adjectival RPs equals its close relative, the as-predicative, in that it occurs so late in acquisition that it must in fact be regarded as an adult extension of a pre-established constructional network.9 While non-metaphorical attributive collexemes like consider or deem do not occur at all in the entire British and American files of the CHILDES database, (18) presents the only attributive example in CHILDES that comes from an adolescent, rather than a parent or teacher. (18)

I found that [AdjP more interesting] than the psychology courses that I’ve taken so far. (age: 16, Carterette: adult.cha, line 4240)

Inspired by the theory of primary metaphor (Grady 1997; Grady and Johnson 2003) as well as by usage-based models of language acquisition emphasizing the item-based, piece-meal development of grammatical constructions (Dąbrowska 2004; Diessel 2004; Tomasello 2004), I hypothesize that, within constructional networks, metaphorical extensions are strictly local, i.e. lexically determined and operate on the low-level generalizations provided by (the strongest representatives of) the collexeme classes identified, rather than on entirely schematic ASCs. I assume the leading representatives of the central collexeme classes to be “path-breaking”, not only with respect to the acquisition of syntactic form, but also concerning the creation

198 Beate Hampe of metaphorical extensions. My hypothesis is that attributive uses of find(cogn.) are motivated by occurrences of the perceived-location verb find in the AS with object-related adverbials. Even in adult speech, cxtr. find predominantly expresses the perception/identification of location. 82% of all adult uses of the construction in the ICE-GB have spatial, though often specialized RPs, while the remaining ones are licensed by the metaphors expounded above (cf. 13). In these metaphorical examples, the complements of the RP head, in, come from a narrow range of abstract nouns: state, condition, situation, conditions, etc., so that the metaphorical construal is only marked by the presence of the locative adverbial (cf. 13b, repeated here as 19a). Moreover, 6 of the 10 uses of find with metaphorical RPs in the ICE-GB exhibit a reflexive fake object. Their overtones of the unanticipated are carried over from specific perceived-location uses of the frame [find NP (fake reflexive) in NP], which are literally and metaphorically true, with unexpected spatial surroundings being judged to present highly unusual circumstances (cf. 19b): (19)

a. b.

I must accept I find myself [PP in ].. (W1B-008) In his fourth adventure …, the November Man finds himself [PP in frozen Helsinki]. (W2B-005)

Literal/spatial uses of find in the syntax of perceived location are relatively frequent in the adult child-directed speech recorded in the British and American files of CHILDES, while truly attributive uses of find are exceedingly rare in the entire database. In the British files of CHILDES, the seven clearly attributive uses of find come from teachers speaking to school children (RP heads: difficult, easy). In the American files, there are eight adolescent/adult instances of attributive find (RP heads: okay, interesting, boring, wrong). None of those instances appears in speech directed at children less than 5 years old, and there are no instances of this use in the speech of young children. The remaining 169 cxtr. instances of find in the entire CHILDES database, some of which come from adult care-takers or from children older than five, present perceived-location uses. Not only are their locative RPs purely spatial in nature, the children’s locatives moreover usually refer to aspects of their immediate environment. This can be directly read off from the highly frequent occurrence of the deictics here and there as RPs or as prepositional complements within the RP (in here, over there), which con-

Metaphor, constructional ambiguity and the causative resultatives 199

trasts sharply with the realization of the locatives in the adult uses of find in the ICE-GB. In the British files of CHILDES, for example, 32 out of 67 instances (= 47%) are deictic, while only 1 out of 56 (= 1.8%) of the adult use of cxtr. find in the ICE-GB is deictic. The difference is highly significant (p (Fisher-Yates) = 1.165E-09; odds ratio = 49.01). As regards the lexical realizations of the RP heads, the child data prefigure the adult preferences. It was mentioned above that the strongest covarying collexeme of find in the adult data is in (cf. 14).10 This preposition is also the single most frequent RP head to co-occur with find in the child data: In all British and American CHILDES files, roughly one third of all examples with object-related adverbial (60 of 169 occurrences = 35.5%) contain the preposition in (with only 16.9 expected under the assumption of a random distribution of all 10 spatial prepositions observed, p(chi-square) < 2.2e-16), suggesting that the identification of object locations frequently relates to the discovery of objects in containers. The element of surprise on the child’s side is carried over to subsequent, extended uses of find. The only metaphorical extension of the literal identification-of-location use of find in the children’s utterances is provided in the file of a 10-yearold child by an expression with a fake reflexive object (20a) equalling the adult expression from (19b). As in the adult example, the subject referent is an experiencer, who suddenly becomes aware of his/her own unanticipated location/surroundings in the scene denoted. It was indicated above that the identification of the specific surroundings goes together with a subjective judgement about the very unusual situation. (20)

a.

b.

and then when I got back I found myself [PP down at the bottom of the pool]. (age: 10; Carterette: fifth.cha, line 4959) I found this fork [PP on the way back]. (age: 2;6.17; Manchester: aran22a.cha, line 523)

Such situations have the properties of a primary scene (Grady and Johnson 2003). If repeatedly experienced in (early) infancy, a primary scene gives rise to a primary metaphor, because a perceptually determined sourcedomain concept (here: location/surroundings) and a subjective response constituting the non-perceptual target-domain concept (here: state/conditions/situation) are simultaneously active. While the two basic concepts are conflated in early childhood (as probably in 20b), they present separate concepts connected by a stable mapping in adult life. Even granting that

200 Beate Hampe cognition metaphors are among the later ones to be acquired, the relevant primary metaphors (STATES ARE LOCATIONS, CONDITIONS/ CIRCUMSTANCES ARE SURROUNDINGS, KNOWING IS PERCEIVING) should be established before the age of ten. It should thus be a very small step from an example like (20a) to the fully metaphorical adult example of similar shape provided in (19a). Nevertheless, this step does not seem to be completed until well into late childhood – there are no examples, at least, in the entire files of CHILDES that would document a non-adult, fully metaphorical, perceived-location use of find in the AS with locative adverbial. Truly attributive uses of find in the AS with adjectival predicative (cf. 18) are entirely missing from the children’s utterances in all of the British and American CHILDES files. The child data available on the use of find (however sparse they may be) suggest a series of small changes in its cxtr. use, which achieve both the semantic shift from the perceived-location verb to the attributive one and the syntactic shift from the object-related adverbial to the adjectival predicative, making the hypothesis of a metaphorical inheritance link from resultative to attributive uses within the AS with adjectival object-complements rather unlikely. First of all, it should be stressed that the syntactic change in the complementation of find from the locative adverbial to the adjectival objectcomplement does (at least initially) not require any conceptual metaphor. An adjectival phrase can appear next to a locative adverbial, or even without an overt locative adverbial, if reference is made to an unexpected (and thus salient) perceivable property of the object referent. For the purpose of this discussion, I shall call such uses the perceived-state uses of find. (21a) presents one of many relevant adult examples from the ICE-GB.11 (21)

a. b.

c.

A young married couple have been found [AdjP dead] [PP in a caravan]. (S2B-016) A young married couple was found (in a caravan): dead. (= ‘A young married couple, who were dead, was found (in a caravan)’) entailment: A young married couple was found.

The AdjP in (21a) allows for co-occurrence with a spatial adverbial, and accepts various forms of re-ordering and/or deletion (see 21b). One entailment of (21a) shows that the meaning of the verb is still very much identical with that in its perceived-location uses (cf. 21c). This is not the case

Metaphor, constructional ambiguity and the causative resultatives 201

with the metaphorical uses in (19a) and (20a) or the fully attributive use of find given in (18). None of these entails anything like ‘I found NP’, nor does the RP in (18) allow for the co-occurrence of a spatial adverbial. Adjectival phrases of the type illustrated in (21a) are thus not fully integrated in a truly cxtr. ASC, but rather present depictive adjuncts expressing an unexpected (and thus highly salient) perceivable property of the object-NP referent. The data from CHILDES suggest that such perceived-state uses with depictive adjectival predicates constitute children’s first precursors to adult attributive uses of cxtr. find (cf. 22–24).12 In contrast to the latter, the children’s depictives can also be realized by participles and non-finite -ing clauses.13 In all of the early examples, find still literally denotes the identification of the object-referent’s location, which can be seen from the entailments given in (22b)–(24b), while the depictive phrases additionally specify one or more salient properties. (22)

a.

b. (23)

a. b.

(24)

a.

b.

… the two cowboys found a man [AdjP dead] … and they told Abbott that he killed that man. (age: 5,3; McWhinney: boys59.cha, line 1723) entailment: The cowboys found a man. … you found me [part drowned]. (age: playmate of the fiveyear old target child; Hall3: maa.cha, line 4183) entailment: You found me. when we woke up we found the bear [NFC -ing right laying on our floor], [AdjP ready to kill us]. (age: 8; Carterette: third.cha, line 2971) entailment: We found the bear.

Only one step removed from these uses is a second precursor of the attributive adult use of find given in (25a, b): Here, the verbal meaning exclusively denotes the identification of an unexpected concrete property of the object referent, but not simultaneously the identification of an unexpected location, the entailments thus differ from those given for (22)–(24). (25a) does not entail ‘I found a foot’; neither does (25b) entail ‘I found Marky’. These examples are thus not depictive, but already form a part of a cxtr. predicate. They nevertheless diverge from adult attributive find both formally and semantically. Formally, they still allow -ing participles or even

202 Beate Hampe full -ing clauses as object-complements. Semantically, they still denote nothing more than a perceived state and could be paraphrased as ‘I noted that my foot was bleeding’ and ‘I noted that Mark is speaking some real Pittsburghese’, respectively. (25)

a. b.

I have a hurt foot because I suddenly found it [part blee:ding]. (age: about 5; Hall 3: gas.cha, line 5698) ya [: you] know I found Marky [NFC -ing speaking some real Pittsburghese lately]. (age: 5;3; McWhinney: boys76.cha, line 5180))

What is missing is thus only the very last step from a perceptual to a nonperceptual/metaphorical, i.e. truly cognitive, use of find as for instance provided by the adult example from CHILDES given in (26a). The primary metaphor licensing this last extension is KNOWING IS PERCEIVING. Judging from the data, this step is completed very late indeed. (26)

a. b.

If you find that [AdjP difficult], there are spoons here. (teacher; Gathburn: 33.cha, line 746) Diagnosis is not considered [AdjP important] because such conflicts are not thought of as illnesses. (W1A-007) [+ passive construction]

The adult collexeme ranking suggests that, once metaphorical find(cogn.) is established as a cxtr. verb, it can be “path-breaking”. Its syntactic frame provides a well-motivated and coherent way to express the cognitive scenario we called attributive. Other, non-metaphorical cognition verbs, such as deem, consider, think, etc., can be analogically modelled after find in that AS (cf. 26b). These cognition verbs do no longer betray the motivational chain from the perceived-location uses to the perceived-state uses of find and their final metaphorical extension. They thus present a kind of nonmetaphorical paraphrase of attributive find, and together create the attributive verb class in the AS with adjectival object-complements. In the childdirected adult speech recorded in the CHILDES files, however, none of these truly adult, relatively formal ‘paraphrases’ of cognitive find occurs. Cognition verbs like consider expand the strictly verb-based schema to a verb-class based one (Croft 2003). As this generalization is fully schematic and clearly paired with a meaning of its own [X THINK (Y BE Z)], it will be regarded as an ASC and henceforth be referred to as the Attributive Con-

Metaphor, constructional ambiguity and the causative resultatives 203

struction (AC). The AC is productive in the adult ICE-GB data, in that some of its instantiations exhibit verbs of cognition and perception, and less typically even verbs of liking and desire (cf. Gonzálvez-García 2008), that may not be attracted to the constructional pattern (at least in the ICE-GB), but are modelled in close analogy to the leading attributive collexemes (cf. 27): (27)

a. b.

… which is why I haven't uh seen it necessary to encourage you. (S1A-075) In all walks of life there exists exist people who feel it necessary to take on the role of the schoolbully. (W1A-010)

In sum, the AC is a very late, adult addition to the cxtr. AS with adjectival object-complement. Motivated by earlier uses of find, it does not present metaphorical extension from the make-class in the RC. It can best be described as a verb-class based construction that is built onto the previously established network of the CMC and the RC and their various related meanings to do with the subject referent’s perception and manipulation of the locations and states of object referents.

4.2. On the metaphorical inheritance link between the CMC and RC In the light of the argument presented above, the developmental data presently available can also shed some new light on the issue of the motivating relation between the CMC and the RC. As mentioned before, the CMC is the first of the cxtr. argument structures to be established and continues to dominate the cxtr. network even in adulthood. In childhood, this dominance is overwhelming, which can be gleaned from both the token and type frequencies of the CMC in the age groups of 2- to 3-year-olds and 6- to 8year-olds, respectively (see Appendix, Table 5). From the beginning of the period documented in the Manchester corpus, i.e. from 24 months of age, all twelve children use put and various other manipulation-of-location collexemes that are most central to the adult CMC. At the end of the period recorded (i.e., from 33 months), all children have used the most central adult manipulation-of-location verbs literally hundreds of times (total token frequency of put: 2,472; total token frequency of the CMC: 4,918), even though the sparse corpus documents only a tiny fraction of their overall language production. Despite this, the assumption of a metaphorical inheri-

204 Beate Hampe tance link between the CMC and the RC at the level of the ASC appears to be implausible from a developmental perspective, for two main reasons: Firstly, it can be assumed that scenarios where the states of concrete objects are manipulated are directly accessible to children, as long as the results are immediately achieved and directly perceivable.14 All of the first instantiations of the RC in the Manchester corpus relate to such scenarios of perceivable change, with the comparative form of the adjectival RP head often additionally expressing the immediate change of state achieved (and perceived): make, for instance, co-occurs with bright, loud(er), higher, big(ger), taller, small, strong, flat, warm, hot, cold, cool, nice and warm, better/worse (of health), dry, dirty, nice and clean, better and washed, awake, tired, and dead. As this trend holds across all twelve children, with big(ger) being used by eight of the twelve children, the probability that this is an artefact of the very sparse corpus is low. Despite the corpus sparsity, the data suggest that children extend RPs to expressions with more subjective/evaluative meanings in a second step. Examples like those given in (28) consistently occur only after a range of concrete instances denoting perceived change have been produced by the child. (28)

make Mummy happy (age 2;9.02, ruth28a.cha, line 641), make it super for Andy (age 2;9.02, aran29a.cha, line 665), making mine’s wrong (age 2;6.29, anne25a.cha, line 741), making the back safer (age 2;10.28, aran34b.cha, line 363), make it the the pan lovely (age 2;8.06, gail25b.cha, line 697), this toe is making me crazy (age 2;8.28, liz29a.cha, line 238)

Secondly, resultative make occurs very early in the AS with adjectival object-complements, at a time when metaphorical instances of the CMC are still extremely rare, though the CMC itself is well established. As children clearly manage to manipulate object locations earlier than object states, it comes as no surprise that make with an adjectival RP starts to emerge somewhat later than the manipulation-of-location collexemes of the put and keep classes in the CMC: half of the 12 children use make with an adjectival resultative phrase at the beginning of the period documented in the corpus (24–27 months), and 11 of the 12 children have used it by the end, though the overall token frequency of cxtr. make in the corpus is dramatically lower than that of put (total token frequency: 258). Roughly at the same time, children also start to use call or name with a nominal object-

Metaphor, constructional ambiguity and the causative resultatives 205

complement (total token frequency: 153), thus initiating the acquisition of the Denominative Construction (DC), too (cf. Hampe, submitted). Most central to the argument suggested here is the fact that the first metaphorical instances of caused-motion verbs occur in the corpus only after the first resultative verbs with an adjectival complement are used. This is true for both the occurrence of metaphorical adverbials as RPs (with either caused-motion or resultative verbs) and for the occurrence of causedmotion verbs with adjectival object-complements. There are no counterexamples to this tendency in the data recorded of the 12 children in the Manchester corpus, so that it is again rather unlikely that these observations are merely due to data sparsity. As regards the occurrence of metaphorical RPs in the CMC, only 2 children out of 12 do use non-spatial adverbial RPs at all: In the Liz file, 22 uses of the RC with make and get are recorded from the age of 2;2.02, before the following metaphorical RP appears at the age of 2;8.28: we are just putting it into lumps (liz29a.cha, line 260). Similarly, in the Warren file, 10 uses of make with adjectival RPs (better, warm, taller, dry) are recorded from the age of 2;3.02, before he uses the CMC metaphorically to denote change (albeit with make instead of the causedmotion verb turn) at the age of 2;7.08 and 2;9.16, respectively: make it in a big ball (warr27b.cha, line 245), make it into car shape (warr33a.cha, line 409). Concerning the second possibility, children make occasional use of caused-motion verbs in the RC forming such combinations as put asleep, put straight, put nice and warm, put right, leave open. There are two comments to make, both of which support my line of argument: Firstly, some of these combinations are in fact also highly frequent combinations in adult speech and thus probably rote-learned rather than formed. And secondly, some of the children’s combinations, such as those with put (~ asleep, ~ nice and warm, ~ straight) resemble the perceived-state uses of find in that they refer to scenarios which have the properties of primary scenes: The change of state denoted by the adjective is literally achieved by an act of putting, and the intended outcome is simultaneously a location and a state associated with that location. The fact that some of the expressions used by the children would not exactly count as idiomatic from an adult perspective suggests that children may manipulate the RP according to what they focus on more, the resulting location or the resulting state. Leaving aside the issue of whether such (semi-)metaphorical uses facilitate the acquisition of more adult uses of put in the RC (put right, etc.), the point relevant here is that none of these early uses of caused-motion verbs with either metaphori-

206 Beate Hampe cal or adjectival RPs are needed as motivating links from the CMC to the RC, because the latter’s central uses occur earlier. Ruth’s expression putting that nice and warm (ruth25a.cha, line 382), which she immediately corrects to make the baby nice and warm, only follows other recorded resultatives with make (~ better, ~ clean). Similarly, Anne uses put baby asleep (anne24a.cha, line 251) after 13 recorded occurrences of make with adjectival complements. To summarize, a metaphorical inheritance link between the CMC and the RC is implausible from a developmental perspective (i) because causation is directly accessible if an immediate change of state can be perceived, and (ii) because cxtr. expressions with resultative verbs, notably but not exclusively make, occur before metaphorical instances of the CMC and before instances of caused-motion verbs in the RC. This is not to say, however, that the two ASCs are unrelated. Even though they may not be linked via a metaphorical inheritance link, it is certainly the case that the CMC makes available to the child the coherent expression of a specialized (and developmentally primary) causation scenario which provides a syntactic slot – the object-related adverbial as RP – for the expression of the resulting object-location achieved. Very soon after the first caused-motion examples – and before the first metaphorical instances of the CMC – this slot is used to express a perceivable resultant state in other scenarios of direct causation.

5. Conclusions This paper has presented a collostruction analysis of data from the ICE-GB on the complex-transitive argument structures with object-related adverbials and adjectival object-complements, better known as the CMC and RC, respectively. The standard distinction between the CMC and RC was shown to be valid in as far as lexemes denoting the causation of motion and the causation of change, respectively, are most strongly associated with and most highly distinctive for the argument structures with object-related adverbials and adjectival object-complements. At the same time, both of these argument structures were shown to centrally code for a slightly wider range of meanings, including the force-dynamically related forced-location and forced-state uses. This extension does not undermine the semantic unity of these argument structures as symbolic units or challenge existing descriptions of these argument structures as the CMC and RC, though the child-

Metaphor, constructional ambiguity and the causative resultatives 207

language data consulted make a metaphorical inheritance link from the CMC to the RC at the level of the fully schematic ASC itself seem highly unlikely. Less conveniently, the collostruction analyses also brought to light another central verb class of the AS so far described as the RC, here referred to as the attributive class, because the meaning expressed can roughly be determined as (X THINK [Y BE Z]) and the verbs occurring in this class are cognition verbs. As the leading collexemes of the attributive class were both very strongly attracted to and highly distinctive for the AS with adjectival object-complement, this class poses a real threat to the semantic unity required by its description as the RC. Although independently existing primary metaphors would in principle allow to derive attributive meanings via a metaphorical polysemy link (IDEAS ARE OBJECTS) from the core constructional uses and thus to regard the existence of resultative and attributive meanings in the RC as a regular case of constructional polysemy in line with the standard model (Goldberg 1995, 1997), developmental data on the emergence of attributive meanings suggest another, strictly local mechanism of network extension to be more plausible. In the account suggested here, the attributive verb class in the pattern with adjectival predicatives is linked back to early perceivedlocation uses of its leading collexeme find in the cxtr. pattern with adverbials. It was argued that series of small shifts in the use of this item-based syntactic frame extend the semantic scope of this verb from the earliest spatio-perceptual uses in the structure with adverbials via perceived-state uses with depictive predicates denoting a salient/unexpected state of the respective object referent, and syntactically integrated expressions of perceived-state to fully attributive uses with adjectival complements motivated by the primary metaphor PERCEIVING IS KNOWING. Assuming this still hypothetical model to be largely correct, the presence of the attributive verb class in the pattern known as the RC, though no case of regular constructional polysemy, is no case of constructional homonymy either, as the attributive verb class is well motivated within the network of causative resultatives as a whole. The line of motivations was shown to bridge across constructional patterns like an inheritance link. Very unlike an inheritance link, however, it is found at a much lower level of generality. Rather than being productive at the entirely schematic level of the ASC itself, primary metaphors were argued to operate in a strictly local, item-based fashion allowing metaphorical extensions to be applied to concrete (frame-based) conceptualizations only.

208 Beate Hampe In view of these considerations, this paper postulated the Attributive Construction (AC) as an argument-structure construction, best characterized as a verb-class based generalization of its own, rather than a regular subsense of the RC arrived at by metaphorical extension from its most central resultative uses – although it is noteworthy that the resulting two main uses of the AS with adjectival object-complements, i.e. the RC and the AC, are fully consistent due to independent cognition metaphors such as IDEAS ARE OBJECTS. Finally, a word of caution needs to be repeated for emphasis: Because of obvious data sparsity in general and the provisional corpus base for the age group of the 6- to 8-year-olds, the account of the metaphorical mechanism of network extension presented here is hypothetical and suggestive rather than fully conclusive. To test the hypotheses presented, much denser longitudinal child data are required documenting the individual development of children after 6 years of age and preferably up to 12 years, which are presently not available from the CHILDES database. It is thus left open to further research to confirm the model of the local/item-based metaphorical extension of constructional networks sketched out here.

Appendix Table 1. Top 20 collexemes in the cxtr. argument structure with object-related adverbial of location/direction [Subj Pred ObjDir AdvObj] Collexeme Analysis I Rank

Collexeme

Observed freq /expected freq

Collostruction Strength*

1 2 3 4 5 6 7 8 9 10 11 12 13

put (dir.) keep (loc.) place (dir.) bring (dir.) get (dir.) set (dir.) leave (loc.) take (dir.) turn (dir.) find (loc) bear (loc.) send (dir.) push (dir.)

369 / 10.84 115 / 5.76 66 / 1.66 78 / 6.45 175 / 45.81 35 / 2.80 50 / 8.16 81 / 23.12 37 / 4.73 54 / 13.16 20 / 2.13 26 / 4.84 15 / 1.38

111.823 88.789 57.814 50.256 26.846 23.240 21.007 20.970 17.095 13.204 11.146 10.981

INFINITE

Metaphor, constructional ambiguity and the causative resultatives 209 Table 1. cont. 14 15 16 17 18 19 20

hold (loc.) shove (dir.) force (dir.) base (loc.) lay (loc.) tuck (loc.) locate (loc.)

23 / 4.32 6 / 0.11 13 / 1.30 15 / 1.99 13 / 1.54 6 / 0.18 8 / 0.48

9.864 9.692 9.158 8.722 8.244 7.931 7.721

(*Collostruction strength: -log (Fisher exact, 10); coll. str. >3: p < 0.001)

Table 2. Top 20 collexemes in the cxtr. argument structure with adjectival objectcomplement/predicative [Subj Pred ObjDir ComplObj ]. Collexeme Analysis II Rank

Collexeme

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

make 395 / 12.79 find 135 / 6.17 keep 87 / 2.70 leave 43 / 3.82 render 13 / 0.12 consider 17 / 1.74 get 56 / 21.47 deem 5 / 0.06 declare 7 / 0.35 set 11 / 2.12 prise 2 / 0.01 strip 3 / 0.10 drive 5 / 0.85 shoot 3 / 0.38 call 10 / 4.26 adjudge 1 / 0.01 comb 1 / 0.01 jerk, riddle, 1 / 0.02 tickle, wrench clamp, stuff 1 / 0.03 cut 4 / 1.14

19 20

Observed freq /expected freq

Collostruction Strength* INFINITE

133.138 101.695 30.302 24.500 11.420 9.760 8.830 7.161 4.919 4.367 3.919 2.757 2.178 1.931 1.884 1.884 1.709 1.585 1.548

(* Collostruction strength: -log (Fisher exact, 10); coll. str. >3: p < 0.001; coll. str. >2: p < 0.01; coll. str. > 1.301: p < 0.05)

210 Beate Hampe Table 3. Comparison of the cxtr. argument structures Distinctive Collexeme Analysis AS with object-related adverbial No Collexeme (obs./exp.) 1 put (384/269) 2 bring (87/60) 3 place (68/47) 4 take (95/69) 5 turn (38/27) 6 send (28/19) 7 have (107/89) 8 bear (20/14) 9 get (196/174) 10 play (27/20)

Collostr. Str.* (p-value) 55.474 (p