Sandra Handl The conventionality of figurative language A usage-based study Language in Performance LiP The conventionality of figurative language 46 Edited by Werner Hüllen† and Rainer Schulze Advisory Board: Thomas Herbst (Erlangen), Andreas Jucker (Zürich), Manfred Krug (Bamberg), Christian Mair (Freiburg i. Br.), Ute Römer (Hannover), Andrea Sand (Trier), Hans-Jörg Schmid (München), Josef Schmied (Chemnitz) and Edgar W. Schneider (Regensburg) Sandra Handl The conventionality of figurative language A usage-based study Bibliografische Information der Deutschen Nationalbibliothek Die Deutsche Nationalbibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliografie; detaillierte bibliografische Daten sind im Internet über <http: / / dnb.d-nb.de> abrufbar. Zugl. Dissertation an der Ludwig-Maximilians-Universität München, 2008 © 2011 · Narr Francke Attempto Verlag GmbH + Co. KG Dischingerweg 5 · D-72070 Tübingen Das Werk einschließlich aller seiner Teile ist urheberrechtlich geschützt. Jede Verwertung außerhalb der engen Grenzen des Urheberrechtsgesetzes ist ohne Zustimmung des Verlages unzulässig und strafbar. Das gilt insbesondere für Vervielfältigungen, Übersetzungen, Mikroverfilmungen und die Einspeicherung und Verarbeitung in elektronischen Systemen. Gedruckt auf säurefreiem und alterungsbeständigem Werkdruckpapier. Internet: www.narr.de E-Mail: info@narr.de Printed in Germany ISSN 0939-9399 ISBN 978-3-8233-6624-9 5 Table of contents Acknowledgements 1 Introduction 2 Literality, figurativity, and conventionality 2.1 Literal vs. nonliteral: A matter of oppositeness? 2.2 The many meanings of literal 2.3 Literality on different levels 2.4 Summary 3 Metaphor, metonymy, and cognition 3.1 Metaphor and metonymy in pre-cognitive linguistics 3.2 Figurative language and figurative thought 3.2.1 The conceptual theory of metaphor........................ 27 3.2.2 Metonymy as a cognitive-linguistic notion ...........32 3.2.2.1 Basic assumptions ........................................................32 3.2.2.2 Metonymy as a reference-point construction ...............35 3.2.2.3 The ubiquity of metonymic processes.......................38 3.3 Mental spaces and conceptual integration 3.3.1 Basic assumptions ..................................................... 39 3.3.2 Metaphor, metonymy, and the theory of conceptual blending.................................................. 44 3.4 Summary 4 Motivation and conventionality in language and thought 4.1 Motivation: From structuralism to cognitive linguistics 4.1.1 Non-cognitive approaches to motivation and conventionality .......................................................... 51 4.1.2 Motivation and cognition.........................................53 4.2 Approaches to conventionalization and conventionality 4.2.1 The conventionalization of language: Institutionalization and lexicalization.................... 57 4.2.2 Conventionality and cognition: A preliminary sketch...........................................................................63 6 4.2.2.1 The notions of salience and entrenchment ...................63 4.2.2.2 Factors influencing the entrenchment and conventionalization of figurative language ..............66 4.3 Typologies of conventional and ‘dead’ metaphors and metonymies 4.3.1 Typologies of metaphor ...........................................76 4.3.1.1 Creative, conventional, dead, and literal metaphors ......76 4.3.1.2 The ‘death’ of metaphors and semantic change ......79 4.3.2 Typologies of metonymy .........................................84 4.3.2.1 Functional criteria and the synchronic-diachronic distinction ......................................................................84 4.3.2.2 Types of metonymy and the notion of domains........86 4.3.2.3 Metonymy as a prototypical notion...........................89 4.3.2.4 Criteria of conventionality and the ‘death’ of metonyms ......................................................................95 4.4 Processing nonliteral language: Linguistic theory and psycholinguistic evidence 4.4.1 The standard pragmatic model: Sequential processing ...................................................................98 4.4.2 The direct access view: The influence of context on comprehension ...................................................102 4.4.3 The entrenchment-based view: Conventionality, frequency, and language processing .................................................................109 4.4.3.1 The graded salience hypothesis .....................................109 4.4.3.2 Processing well-entrenched and less entrenched meanings ......................................................................112 4.5 Summary 5 Investigating the conventionality of figurative language 5.1 Figurative language and corpus linguistics: Methods and problems 5.1.1 General considerations ...........................................126 5.1.2 The choice of the source domain vocabulary ...... 130 5.1.2.1 The metonyms.............................................................130 5.1.2.2 The metaphors ............................................................135 5.2 The procedure: Looking for metonymies and metaphors in the BNC 5.2.1 Metonymic senses ...................................................137 7 5.2.1.1 KWIC concordances and the problem of underspecification ......................................................137 5.2.1.2 Conjoined phrases, multiple mappings, and meanings ......................................................................143 5.2.1.3 Further problems: The metaphor-metonymy interface, metonymic compounds, and metonymic chains .......................................................145 5.2.2 Metaphorical senses ................................................156 5.2.2.1 The sub-corpora ..........................................................156 5.2.2.2 Extracting and organizing metaphorical senses ....157 5.2.2.3 Problematic cases........................................................163 5.2.2.4 The presentation of the results .................................166 6 Evidence for the conventionality of metonymic meaning and thought 6.1 Frequent and less frequent metonyms: An overview 6.2 The conventionality of metonymies: A matter of linguistic idiosyncrasy or conceptual regularity? 6.2.1 Salience and the conventionality of metonymy.. 178 6.2.1.1 Wholes and parts ........................................................179 6.2.1.2 W HOLE FOR PART , PART FOR WHOLE , and PART FOR PART metonymies ........................................................181 6.2.1.3 Reversible mappings: Preferred routes and nondefault cases ................................................................186 6.2.1.4 The influence of conceptual structure and salient attributes ......................................................................200 6.2.2 Conventional and less conventional metonymies: Ontological salience and attribute salience ......................................................................206 6.2.2.1 A TTRIBUTE OF ENTITY FOR ENTITY metonymies ........206 6.2.2.2 P RINCIPAL PART FOR WHOLE metonymies ................212 6.2.2.3 C ENTRAL FACTOR FOR INSTITUTION / PROFESSION / ACTIVITY metonymies .................................................217 6.2.2.4 L OCATION FOR ENTITY metonymies ..........................220 6.2.2.5 I NSTITUTION FOR PEOPLE metonymies ......................228 6.2.2.6 W HOLE FOR PRINCIPAL PART metonymies ................229 6.2.2.7 Further metonymic mappings ..................................233 6.2.3 Types of salience and the conventionality of metonymic mappings and meanings: An intermediate summary ..................................... 234 8 6.3 Target-in-vehicle salience and the emergent structure of metonymies 6.3.1 Salient and non-salient metonymic targets ......... 238 6.3.2 Types of attribute salience and the question of emergent structure .................................................. 245 6.3.3 Summary ..................................................................249 7 Evidence for the conventionality of metaphorical meaning and thought 7.1 The results of the metaphor study: An overview 7.2 Frequent and infrequent metaphorical mappings and meanings 7.3 Preferred metaphorical construals 7.4 Metaphors and attribute salience: Emergent structure and conventional metaphorical meaning 7.4.1 The relativity of attribute salience ........................ 274 7.4.2 Lean and rich mappings and their attributes ...... 275 7.5 Summary 8 Conclusion References Dictionaries Corpora Appendix A: The results of the metonymy study Appendix B: The results of the metaphor study 9 Acknowledgements I DEAS ARE CHILDREN , as the Master Metaphor List states, and just like children cannot develop their capacities in isolation, the ideas gathered in this book owe very much to the help and guidance of those surrounding me. First of all - and most of all - I would like to thank Prof. Dr. Hans-Jörg Schmid, my dissertation advisor at the Ludwig-Maximilians-University Munich, for his constant support, invaluable advice, and patience during all stages of the project’s development. Many thanks also go to Prof. Dr. Richard W. Janney, whose critical comments often helped me see things in a different light. I am deeply indebted to Prof. Dr. Leonhard Lipka for showing me the way into the ‘wonderful world of linguistics’ and for sparking my original interest in questions of conventionality. I am very grateful to Prof. Dr. Rainer Schulze, the series editor of LiP, for his assistance in the publication process. Furthermore I would like to thank Mareike Reichelt and Karin Burger from Narr publishers for their professional expertise and their readiness to answer all my questions. I also owe many thanks to Kathleen Rabl, who proofread the entire manuscript and provided many additional examples. Special thanks go to my friend and colleague from American Studies, PD Dr. Kerstin Schmidt, for her genuine interest in my work and for her priceless help with everything. Most crucially, however, I am indebted to my parents, Brigitte Handl and Manfred Singer, and the rest of the family, Susanne Handl, Torsten Krauß, Laura Handl and Philipp Handl, for encouraging and sustaining me, and for supporting me in countless other ways. 11 1 Introduction One of the fundamental assumptions of cognitive linguistics is that the most vital mechanisms of thought are those which are entirely subconscious. This is particularly evident in cognitive approaches to figurative language. Metaphors have been demonstrated to be profuse in everyday discourse. They are so common that we no longer notice them, they are conventional. Moreover, and more importantly, conventional metaphorical expressions form coherent systems based on well-established metaphorical patterns of thought. These metaphorical ways of thinking help us to understand and cope with the world around us by helping to structure abstract concepts. In this way, metaphors determine not only what we think, but also what we do. While metaphors have been in the limelight of cognitive-linguistic research ever since the publication of Lakoff and Johnson’s (1980) classic book, metonymy has only recently attracted attention. Just like metaphor, metonymy is guided by highly systematic conceptual mappings that lead to regular linguistic expressions. And just as in the case of metaphor, it is assumed that the mappings which are cognitively the most central are those which are so deeply entrenched that the figurative nature of the linguistic expressions they have given rise to is not noticed. For both metaphor and metonymy researchers have compiled substantial lists of examples meant to prove the conventionality of the expressions and, correspondingly, also of the mappings underlying them. But are all of these examples really equally conventional ways of speaking? Is a reference to persons by way of the metonymic expression We need some new faces around here really as common as a reference to the United States by way of the metonym America? And is a metaphorical expression like I’m shopping for a better philosophy equally essential from a cognitive point of view as a construal of a verbal ARGUMENT in terms of WAR as in He attacked every weak point in my argument? Even though the concept of conventionality is the foundation of many key assumptions in the conceptual theory of figurative language and thought, the meaning of the term is frequently taken for granted. With respect to metaphors, researchers have proposed some tangible criteria, such as unidirectionality and the invariance principle, which can help to distinguish unconventional from more conventional expressions. However, the principles that are supposed to govern the conventionality of linguistic metonyms and metonymic ways of reasoning, such as the laws of ontological salience are often somewhat more vague, difficult to apply and have hardly ever been put to the test. Therefore, the primary focus will be on metonymy rather than on metaphor throughout this study, in the theoretical as well as empirical parts. 12 Given that the conventionality of metonyms and metaphors is intrinsically connected to the question of which figurative ways of thinking are at the heart of human cognition, we should attempt to understand better what it means for metonyms and metaphors to be conventional. In what follows, I will thus search for conceptual rules and regularities that govern the conventionality of figurative language: Which preconditions have to be met by figurative expressions in order to be accepted by the speech community at large? Are all types of metonyms and metaphors subject to the same constraints, or do the distinct types have to fulfil different criteria in order to gain wider currency? Are all examples that belong to one and the same type equally well-established, or are there exceptions to the more general tendencies? To answer these questions, it is necessary to introduce several basic theoretical frameworks. Since the nonliterality of conventional figurative expressions is characteristically no longer consciously apprehended, I will first address the general distinction of literal and nonliteral language and its theoretical and practical value (chapter 2). Chapter 3 discusses the key concepts of cognitively-oriented theories of figurative language. Since some of the assumptions usually considered to be unique to the cognitive framework are actually indebted or at least related to earlier, non-cognitive treatments of metaphor and metonymy, I will begin this chapter with a brief overview of some of these earlier approaches (chapter 3.1), before taking up the most central claims of the conceptual theory of figurative language and thought in chapter 3.2. Here, the first part deals with metaphor (chapter 3.2.1), the second, more extensive one, with metonymy as a cognitive-linguistic notion (chapter 3.2.2). The last section of chapter 3 then introduces the second major cognitive theory which is relevant for an examination of the conventionality of figurative language, the theory of conceptual blending (chapter 3.3). Chapter 4 concentrates on the chief concerns of my study, i.e. theories of the conventionality of figurative language and thought. The chapter begins with a consideration of the relationship between motivation and conventionality (chapter 4.1) and then turns to the process of conventionalization, its preconditions and consequences (chapter 4.2). As explicit examinations of conventionality are rare in cognitive linguistics, structuralist theories are also briefly reviewed in chapters 4.1.1 and 4.2.1. As far as the cognitive perspective is concerned, two terms, namely salience and entrenchment are of major importance for a description of the conventionalization of figurative language, and these notions are thus introduced in chapter 4.2.2.1 before an initial, rather sketchy picture of the conventionalization of figurative language is drawn in chapter 4.2.2.2. While the latter provides the basis for the discussion and critical evaluation of extant typologies of conventional and ‘dead’ metaphors and metonymies in chapter 4.3, the remainder of chapter 4 is devoted to the question of the psychological reality of the conventionality of figurative language: It examines the evidence found for 13 and against the position that the frequency and familiarity of figurative language bear directly upon its on-line processing. The factors identified as relevant for the conventionalization of linguistic metonyms and metaphors in the theoretical discussion, like frequency of occurrence and salience, are then used as a starting point for the empirical investigation of authentic linguistic data in chapters 5 to 7. This part is mainly based on the results of a large-scale corpus-linguistic study of metonymic and metaphorical expressions. Once again, as in the theoretical section, the focus will be on metonymy. In many theoretical accounts, no matter whether they belong to the realm of structuralism or cognitive linguistics, the factor of frequency is quoted as one of the main indicators of conventionality - an assumption that has been supported by many psycholinguistic experiments, and therefore the frequency of metonyms and metaphors in natural language use was analyzed. A description of the general procedure and the problems involved in such a study is given in chapter 5. In the following two chapters, the results of the corpus study are presented and the relative frequencies found for the single instances of different metonymic and metaphorical mappings are related to the criteria of conventionality established in the theoretical section. Chapter 6 focuses on metonymies, and chapter 7 on metaphors. The reasons why some figurative construals are clearly preferred to others are discussed, and it is claimed that the conventionality of figurative language cannot be adequately described if only the level of the conceptual mappings is considered. With regard to metonymies, it is shown in the first sections of chapter 6 (6.1 to 6.2.1.3) that the laws of ontological salience often proposed to account for the degree of conventionality of different mappings cannot explain all the relative frequencies of the linguistic metonyms investigated. A comprehensive account of the conventionality of different metonyms presupposes a consideration of the structures of the concepts involved in the mappings, since the degree of conventionality hinges on the salience of single attributes in speakers’ mental representations of these concepts. The notion of attribute salience is therefore introduced in chapter 6.2.1.4 and applied as an additional explanatory criterion of the corpus-linguistic results in the following sections (chapter 6.2). In contrast to ontological salience, attribute salience can explain regular cases and numerical outliers alike. This becomes evident in chapter 6.3, which is partly based on the results of an attribute-listing task (chapter 6.3.1). In addition, it is argued that metonymies can be characterized by two different types of emergent structure, salience-upholding and salience-changing emergent structure, and that the first results in high relative frequencies of the expressions, whereas the latter leads to low relative frequencies (chapter 6.3.2). Parallel to chapter 6, chapter 7 begins with a preliminary overview of the results regarding the usage of metaphors (chapter 7.1). On the whole, the scope of the metaphor study is more restricted, and the results thus 14 allow much less generalizations than for the metonymies. In chapter 7.2, it is demonstrated that many examples commonly employed to illustrate the conventionality of metaphorical speaking are actually not very frequently used in natural discourse. For example, verbs belonging to certain general frames, like COMMERCIAL EVENTS , seem to be particularly resistant to metaphorical meaning extensions. And for some conceptual metaphorical mappings, there are obvious preferences to express them with certain linguistic means. How this is connected to metaphorical highlighting is examined in chapter 7.3. The last section of chapter 7 is devoted to the role of attribute salience for the conventionality of metaphors. The application of rich mappings of the type ARGUMENT IS WAR or A THEORY IS A BUILDING does, as a rule, not disturb the saliency of the attributes of the respective source domains. Lean mappings of the type A PERSON IS A PIG , however, behave differently; they typically single out only one aspect of the source domain and this can result in changes of the salience of attributes in the source, a process which is best described as generating conventionally fixed emergent structure. Attribute salience is thus not only relevant for the conventionality of metonymies, but also for the conventionality of metaphors. The findings of the complete study are then summarized in chapter 8. 15 2 Literality, figurativity, and conventionality 2.1 Literal vs. nonliteral: A matter of oppositeness? A study of the conventionality of figurative language seems to presuppose that the notion of figurative language refers to a well-defined and unified concept. At least our folk model of language 1 is based on the assumption that there is a strict dividing line separating what is literal from what is nonliteral. Certainly, the average language user would not hesitate to categorize (1) as figurative and (2) as literal. (1) Juliet is the sun! (2) A dog is a ‘domesticated carnivorous mammal that typically has a long snout, an acute sense of smell, non-retractile claws, and a barking, howling, or whining voice’. (NODE; s.v. dog) So far, the issue seems uncontroversial. But deciding upon the literality or nonliterality of sentences or utterances becomes more difficult in cases like the following: (3) He had to digest the bad news. (4) There were many new faces around. On the one hand, (3) and (4) seem to be perfectly normal utterances that could occur in many everyday contexts. They are by no means poetic or otherwise stylistically marked. They are easy to understand, and do not, contrary to example (1), seem to leave much room for creative interpretation. On the other hand, on closer inspection it becomes apparent that (3) and (4) contain lexical items that are not used in their original and most literal senses. The verb digest literally means ‘break down (food) in the stomach and intestines into substances that can be used by the body’ and the noun face actually refers only to ‘the front part of a person’s head from forehead to the chin’ (NODE; s.v. digest, v., face, n.). These, however, are not the senses principally intended in (3) and (4), where digest describes the way a person tries to cope with some information, and faces refers to ‘people’ and not to their countenances only. Therefore, (3) and (4) are nonliteral, but much less obviously so than (1). English native speakers are quite 1 That human thinking is determined by oversimplified versions of reality is the rule rather than the exception. What is stored in speakers’ minds and forms the basis for cognitive processes are, in Lakoff’s (e.g. 1987) terminology, idealized cognitive models (ICMs), i.e. idealized versions of the world abstracted from reality in various ways. Good examples are the everyday concept of a MOTHER (cf. Lakoff 1986: 291-292, 1987: 74-76, 79-84), or the seven-day calendric cycle. Even though time comes as a continuum with no boundaries and structure, cultures have imposed an organization upon it: We have a work week of five days and a weekend that lasts two (Lakoff 1987: 68-69). 16 used to talking about ideas in terms of food, and, consequently, about understanding information as digesting it. Similarly, they are accustomed to conceptualizing persons in terms of their faces, since the face is what we look at first when we meet other people. The fact that it is so common to talk about information in terms of food and about persons in terms of their faces is responsible for the unobtrusiveness of the nonliterality of (3) and (4). If average speakers noticed it at all, they would do so only at second glace, and still might end up confused about whether it is really correct to categorize the examples as figurative, because the senses of digest and face realized in (3) and (4) are so ordinary. Merely by virtue of being so familiar, they somehow seem be or have become ‘literal’ in a way. The discussion below demonstrates that linguistic description is confronted with similar problems when it comes to distinguishing between literal and nonliteral language. For a long time linguists concentrated on the study of literal language and considered figurative uses as derived and parasitic, 2 thus implying a neat boundary between the two. More recent research in cognitive semantics, psycholinguistics and pragmatics has, however, shown that the criteria proposed for the distinction cannot account for all the cases, so that the dividing line between literality and nonliterality turns out to be much more blurred than commonly assumed. 2.2 The many meanings of literal The uncertainty about what should be regarded as literal and what belongs to the realm of figurative language is by no means restricted to laypersons’ judgements. Even though much work has been devoted to this question (e.g. Bartsch 1996; Dascal 1987, 1989; Gibbs 1984, 1989, 2002; Katz 1977; Lakoff 1986; Récanati 1989, 1995, 2004; Searle 1978), there is no single, allencompassing answer to it, the main reason being that the concept of literal meaning is not a unified one. As Lakoff (1986: 292) and Gibbs (1993, 1994: 75) point out, the term literality can be understood and has been investigated in up to five different senses: 3 1. Conventional literality (i.e., plain, direct language) 2. Subject-matter literality (i.e., expressions ordinarily used in connection with a particular subject-matter) 2 This position is not restricted to semantics. Many, more traditionally oriented pragmatic approaches also hold that any successful understanding of a nonliteral expression, e.g. an indirect speech act or a linguistic metaphor, relies on the rejection of the corresponding literal interpretation. Grice (1975; 1998) and Searle (1979) are often quoted as prime representatives of this view; cf. also chapters 3.1 and 4.4.1. 3 Other linguists also state that literal meaning has been characterized in many different ways: as direct, fixed, explicit, and necessary, as fully compositional, context-free, and truth-conditional (cf. e.g. Ariel 2002a: 362; Dascal 1987: 260-262; Katz 1977: 14; Searle 1978: 207-208). 17 3. Non-metaphorical literality (i.e., language that is not even partly understood in terms of something else) 4. Truth-conditional literality (i.e., language that fits the world) 5. Context-free literality (i.e., language as interpreted in a null context) Even if the different senses of literality are considered in isolation, they are difficult to apply. Searle (1978), for example, convincingly argues that there is no such thing as a ‘null context’. Even if we encounter language without any meaningful situational (or linguistic) context, we invoke default assumptions which are based on our experiences and determine our interpretation of the linguistic input. For Searle’s famous example The cat is on the mat, we automatically assume that the cat and the mat are within the gravitational field of the earth and not floating in outer space. But if context invariably plays a role for any interpretive process, literality cannot be defined as independent of the context and the speaker. 4 An account of literal meaning in terms of the principle of compositionality, which goes back to Frege ([1892] 1952), does not solve the problem either, as this framework presupposes that all morphemes and words have fixed literal meanings. Therefore, understanding sentences or rather their literal meanings should consist of an activation of all the meanings of the words in the sentence, supported by the hearer’s knowledge of the syntactic rules according to which the words have been combined. Apart from the fact that background assumptions will always exist, this view has another shortcoming: That word meaning is fixed cannot be maintained for all cases. Particularly good examples to illustrate this are adjectives, as their interpretation often varies with the nouns they modify. 5 Clark (1991), for example, has demonstrated that this even holds for fairly concrete adjectives like red. If red is used as a predication of hair, subjects tend to imagine an orangish colour, if it is used as a predication of wine, the meaning is shifted towards a purply red (cf. also Ariel 2002a: 371, Gibbs 1994: 39). The meaning of red is vague. 6 In such cases it seems almost impossible to speak 4 The idea that semantic description should be restricted to sentence meaning (as opposed to utterance meaning in context) is especially prominent in the generative framework (e.g. Katz and Fodor 1963). Cf. also Katz (1977) who defines the ideal speaker/ hearer’s semantic competence with reference to the so-called anonymous letter criterion. 5 For an attempt to apply the principle of compositionality to larger chunks of linguistic input, cf. Bartsch’s (1996) discussion of cut. She argues that literality should not be searched for on the word-level, but on the level of propositions. 6 The terminology used to refer to the different semantic aspects belonging to one lexeme is unfortunately not uniform. In general, one can distinguish between ambiguity and vagueness. The former describes cases where one word has various distinct senses, while the latter refers to underdetermined or underspecified meanings, as in the case of red. Another example of vagueness is cousin, which is underdetermined with respect to the sex of the person, and ambiguity can be illustrated with port ‘harbour’ vs. port ‘wine’. A useful survey of the different terminological systems in this field can be found in Schmid (1993: 95-108). Cf. also Geeraerts (1993a), Taylor (2003: 18 of one (or more) fixed literal meaning(s) of the word. There is no clear oneto-one or one-to-many correspondence between form and meaning(s) as suggested by our folk understanding of language as a conduit (cf. Reddy 1979, Ritchie 2006: 11). Truth-conditionality, too, cannot be regarded as an infallible criterion for literality. Cases in point are negated predications (cf. 5) or ambiguous sentences (cf. 6): (5) Life is not a bed of roses. (6) Anchorage is a cold city. (Leezenberg 2001: 175) Example (5) is literally true, but a literal interpretation misses the metaphorical point of the utterance. Example (6) poses a similar problem. Depending on the context, it could receive either a literal (‘low temperature’) or a nonliteral (‘unfriendly’) interpretation. If uttered with the second intention in mind, truth-conditions cannot cope with it, as - due to the existence of the literally true interpretation - the nonliteral reading does not violate them (cf. Leezenberg 2001: 173-177). Another argument against a truth-conditional approach is that not all statements that are literally false have to be figurative. There is no criterion that separates nonsensical sentences such as A dog is a plant from nonliteral ones like My husband is a pig. 7 Neither compositionality nor truth-conditionality are definitive indicators of literality. Moreover, literal meaning is not fixed and contextindependent. While these criteria fail for various reasons if considered in isolation, the weaknesses of classic accounts of literality are even more evident if the different senses of literality are applied to one and the same example, because usually, the various senses or types of literality are tacitly assumed to converge. But this is not necessarily the case. What is of particular significance for my concerns is that linguistic semantics tends to think of the first and third sense of literality, i.e. conventional and non-metaphorical literality, as being identical (Lakoff 1986: 292- 293). Metaphorical language is considered unconventional and regarded as an antonym (if not as a complemetary term) to literal language. But as the conceptual theory of metaphor 8 convincingly claims, metaphoricity and conventionality are by no means mutually exclusive. Many quite conventional and ordinary utterances such as Your claims are indefensible or This esp. chapters 6 and 8), Tuggy (1993). In cognitive linguistics, the different semantic facets of vague lexemes have also been described as active zone phenomena (cf. e.g. Langacker 1987, 1993). 7 What is more, “to specify a set of truth conditions for a sentence presupposes that some interpretation has already been given to that sentence” (Gibbs 1994: 64). Without any knowledge about Anchorage and cold and the relation between the two, a truthconditional analysis is impossible. This caused many researchers, such as Searle (1978) or Récanati (1989, 2004), to modify their concept of truth-conditionality to fit the data at hand, while others flatly reject it as a criterion for literality (e.g. Ariel 2002a; Gibbs 1994, 2002). 8 An overview of this research paradigm can be found in section 3.2.1. 19 theory needs more support have a metaphorical motivation and are thus literal in one sense, but nonliteral in another. The same holds true for metonymically motivated language (e.g. The White House isn’t saying anything, He drives a BMW), although metonymy is usually not addressed in the literature on literality. This proves that the classic approaches to literality failed mainly because they tried to capture too many, too diverse aspects with the help of one (seemingly) unified notion. A promising way out of this dilemma is Ariel’s (2002a) idea of abandoning the notion of literality and replacing it with “three concepts of minimal meaning” (2002a: 388), which will be introduced in the next section. 2.3 Literality on different levels Ariel (2002a) starts with the observation that the problems connected to the traditional notion of literality have led to so many modifications that “there are now many different concepts of ‘literal meaning’” (2002a: 362) which are only partly justified and, what is even more important, partly clash with each other. She therefore proposes giving up the search for a single, unified notion of literality and argues that many difficulties can be avoided or overcome by focusing on different types of basic, or as she calls them, ‘minimal meanings’. Minimal meanings occur on three levels of linguistic description - the linguistic, psycholinguistic, and pragmatic (or interactional) levels - and each of them reflects the primary interest of the corresponding linguistic discipline. The linguistic level is concerned with the difference between coded and inferred meanings. On this level, a meaning is minimal, if it can become active by mere decoding. Linguistically minimal meanings are not restricted to the lexeme level, but embrace larger grammaticized units such as collocations or idioms. Crucially, not only fully truth-conditional and non-figurative meanings can be linguistically minimal. In contrast to other approaches, Ariel (2002a: 392) includes figurative readings within the realm of the linguistically most central meanings. As regards the psycholinguistic level, 9 meanings are minimal if they are automatically and obligatorily accessed without causing any undue processing effort. The defining criterion on this level is “speed of accessing” (Ariel 2002a: 393). Psycholinguistically minimal meanings are different from linguistically minimal ones, since not all coded meanings are accessed 9 Originally, Ariel (2002a: 393) uses the expression “dynamic construction of meaning in context in real time”. In the strict sense, however, highly salient meanings (in Giora’s 1997, 2003 terminology) have to be stored, as salience is considered (among other things) a function of familiarity and frequency. The term meaning construction should be reserved for more dynamic processes that involve at least a full sentential context. 20 with equal speed. More frequent and more familiar meanings are processed faster than less frequent and familiar ones (cf. e.g. Giora 2003; cf. also chapter 4.4.3.1). For example, two of the encoded meanings of the form drop ‘tiny amount of liquid’ and ‘act of falling’ differ with respect to their psycholinguistic minimality, as the former is more frequent and therefore accessed faster, than the latter (cf. Ariel 2002a: 393). Only the interactional perspective takes context into account. Here, minimal meanings are less stable and therefore referred to as “possible privileged interpretations” (Ariel 2002a: 396; original emphasis). From a pragmatic point of view, meaning is always a question of negotiation, and hearers’ interpretations of utterances do not necessarily converge. Interactionally minimal meaning is characterized by its relevance for the ongoing discourse. That this level is more subjective than the other two can be seen if competing opinions occur as to which interpretational formula is the most contextually appropriate. This is demonstrated in example (7). The librarian feels it is necessary to cancel the contextually appropriate interpretation of one minute as ‘short period of time’. This illustrates the significance of interactionally appropriate meanings as those to whose sincerity the speaker is being committed. (7) Woman at the library: D’you have a minute? Librarian: Literally, one minute. (Ariel 2002b: 1009) What is gained by giving up one unified notion of literality and focusing instead on the different underlying motivations of the term? One advantage is that Ariel’s approach considers, as frequently demanded (cf. Gibbs 1984, 1989, 1994), the psychological reality of literal meaning. But even though it has been claimed that meaning is never entirely contextindependent, as seemingly context-free interpretations are invariably the result of the hearers’ assuming a default context (cf. Searle 1978), Ariel does not completely do away with the idea of literal meaning on the linguistic level either. By splitting up literality into three different minimal meanings, she takes away much of the pressure a single unified notion has to cope with. Since her model explicitly allows that a meaning is minimal on one of the levels, but non-minimal on another, it is more successful than other approaches in accounting for the contradictory findings that abound in the relevant literature. 10 The idea of minimality on different levels turns literality into a much more flexible notion. It seems reasonable to assume that a linguistically encoded but not very frequent meaning will not necessarily be accessed at all, if the context points in a different direction. According to the linguistic and psycholinguistic status of the meaning in question and the situation at hand, the different scales shift their position, to allow a unified perception of the current meaning (cf. Figure 1). 10 For an overview of these contradictory findings, cf. chapter 4.4. 21 Linguistic level: encoded inferred meaning meaning Psycholinguistic level: fast slow access access Contextual level: contextually contextually relevant irrelevant Figure 1: Three scales of minimal meaning (based on Ariel 2002a) Ariel’s approach manages more than the prototype view of literality proposed by Dascal (1987). The latter tries to retain many of the traditional ideas about literality (e.g. truth-conditionality and compositionality) no longer considered central by Ariel. Even though Dascal realizes that the conventionality of a meaning plays a significant role for its processing, his explanation is not very helpful. His argument is that the conventionalized meanings of indirect speech acts, idiomatic expressions and metaphors are “in fact the literal ones” (1987: 267). In this way, he destroys the very basis of literality as a necessary counterpart to nonliteral usage by incorporating large amounts of nonliterally motivated language into his concept of “moderate literalism” (Dascal 1987: 260; cf. also Dascal 1981, 1983, 1989). 11 The boundary between literal and figurative language is not as easy to draw as our everyday conception of language makes us think. However, giving up the notion of literality altogether, as Gibbs (e.g. 1994, 2002) proposes, does not solve the problem either. How would it be possible to identify a metaphor such as He attacked every weak point in my argument, if not by stating that we conceive of ARGUMENTS in terms of WAR , i.e. by pointing out that our conceptualization is not a literal one? Just because such expressions and the patterns of thinking they rely on are conventional, does not mean they are literal. Conventionality is not the same as literality and does not cause literality. Conventionality is a phenomenon that essentially cuts across the two fuzzy-edged categories of literality and figurativity. 11 For metaphorical language and many idioms, this corresponds to what Gibbs (1994: 10) calls “dead-metaphor view”, a view he vigorously challenges in many publications (cf. e.g. Gibbs 1993; Gibbs and O’Brien 1990). That metaphors still fulfil important conceptual functions, even if their linguistic instantiations have become conventions, is also one of the major claims of conceptual metaphor theory (cf. e.g. Barcelona 2000b: 5; Lakoff and Johnson 1980: 55); cf. also chapter 3.2.1. current meaning 22 2.4 Summary Neither literality nor figurativity can be measured in absolute terms. The traditional strict distinction of literal and figurative language has to be replaced by a psychologically realistic model. While it cannot be denied that strong contextual implications can favour even highly unfamiliar meanings, it is also safe to assume that speakers work with default assumptions, if such a strong context is lacking. Accordingly, meanings that are foremost in speakers’ minds, i.e. meanings that are highly conventional, still enjoy a somewhat privileged status. What matters is the meaning we conventionally associate with linguistic expressions. Let me illustrate this with two individual lexemes: to attack and America. The question is whether we are more likely to think first of their respective literal meanings, i.e. ‘take aggressive action against (a place or enemy forces) with weapons or armed force, typically in a battle or war’ (NODE, s.v. attack, v.) and ‘a landmass of the western hemisphere consisting of the continents of North and South America joined by the Isthmus of Panama’ (NODE, s.v. America), or whether we are so accustomed to their figurative senses, i.e. ‘criticize or oppose fiercely and publicly’ (NODE, s.v. attack, v.) and ‘name for the United States’ (NODE, s.v. America), that these are the first that come to our minds when we encounter the lexemes either in a weak context or without any context at all. Cognitive linguists have convincingly claimed that figurative thinking and speaking can be conventional. However, the extent to which individual conceptual mappings and figurative expressions are conventional is a question that has, so far, not been addressed in detail. Even though conventionality is unquestionably a matter of degrees, it seems to have been regarded as an absolute. Moreover, the question of what factors foster or inhibit the conventionality of metaphors and metonyms has not attracted much attention either. These are important issues, however, since drawing conclusions about the cognitive weight of different figurative reasoning patterns only becomes possible on the basis of an assessment of the degree of conventionality of linguistic metaphors and metonyms. To approach these questions, it is first necessary to provide an overview of the main claims of the relevant theoretical frameworks, i.e. the conceptual theory of metaphor and metonymy and the theory of conceptual blending, which will be done in the following chapter. As some of the major insights commonly ascribed to these theories are, however, not as unique to them as usually assumed, the chapter will start with a brief consideration of what other, earlier, and partly still influential theories of figurative language have to say with regard to the relationship between figurative language, thinking, and conventionality. Section 3.1 thus serves as background for the following discussion of the conceptual theories by pointing out similarities with and differences to later cognitive approaches. 23 3 Metaphor, metonymy, and cognition 3.1 Metaphor and metonymy in pre-cognitive linguistics Traditionally, metaphor was regarded as a matter of words, as a mere substitution of one word for another that is based on the criterion of similarity. However, in Aristotle’s description it is already evident that figurative language is - at least to a certain degree - a necessity, and not only an ornamental feature of language. Aristotle admits that there are cases where gaps in literal language are filled by metaphor (Leezenberg 2001: 35-39). Moreover, he does not deny that metaphor is used in everyday language and by the average language user. Still, his focus is clearly on poetic, i.e. unconventional examples, as these achieve the greatest rhetorical effects. This focus is probably responsible for the fact that Richards (1936), who developed the so-called interaction theory of metaphor, begins his account with a severe criticism of the Aristotelian tradition. He points to “the evil presence of three [...] assumptions which have ever since prevented the study of this ‘greatest thing by far’ [i.e. metaphor; SH] from taking the place it deserves” (1936: 89): 1. Good metaphors can only be made up by geniuses who have a special eye for resemblances. 2. Individuals cannot learn how to use metaphor. 3. Metaphor forms the counterpart to normal language use, is therefore exceptional. For Richards, metaphor is ubiquitous so that “[w]e cannot get through three sentences of ordinary fluid discourse without it” (1936: 92). As he (1936: 118, 130) points out, metaphorical language can become part of the lexicon, can become a conventional way of expressing ideas. This, however, does not alter the status of the expression as a figurative one. Moreover, metaphor is said to determine our thinking to a large extent (Richards 1936: 92, 94; cf. also Black 1977: 446-448): [W]hen we use a metaphor we have two thoughts of different things active together and supported by a single word, or phrase, whose meaning is a resultant of their interaction. (Richards 1936: 93) The interaction view 12 holds that much is lost if a metaphor is ‘translated’ into a literal statement (Black 1962: 46, 1977: 441), as metaphors make us see things in a new light. This is the result of the interaction between the two constituents of a metaphor, called tenor and vehicle by Richards (1925, 12 More recent versions of the interaction view can be found in Indurkhya (1992) or Kittay (1987). 24 1936). 13 Metaphor is considered a double unit in which knowledge commonly associated with the vehicle is projected onto the tenor. Which parts are selected depends on the speakers’ knowledge about the tenor. As a result, the tenor is linked to selected aspects of the vehicle and acquires novel semantic aspects. But the vehicle is also changed as the aspects used in the metaphor have to be aligned with the tenor. Consider example (8): (8) Man is a wolf. (Black 1977: 433) The speakers’ knowledge of wolves consists of a whole array of different features. They are ‘animals’, ‘have four legs’, ‘have a fur’, ‘live in groups’, are often considered ‘malicious’ and ‘aggressive’, etc. From all these different aspects, only one is chosen when wolf is connected to man. This changes the structure of the concept WOLF significantly. But the selected facet cannot be applied to the tenor in a straightforward way. The mental picture of aggressive animals has to be aligned with the knowledge about the tenor: If humans are metaphorically conceptualized as wolves, the behavioural patterns conjured up are certainly different from those of the vehicle. Interaction theory can certainly be criticized because of the unclear idea of the central term interaction and its lack of examples (cf. Jäkel 1997: 103- 105). Nevertheless, the theory anticipates some of the most vital ideas for which the conceptual metaphor theory has become famous, although proponents of the conceptual view may be reluctant to accept this (cf. Lakoff and Turner 1989: 131-133): Metaphorical language and thought are not considered dispensable. Instead it is argued that metaphors are “’cognitive instruments’” (Black 1977: 454) which make us perceive connections that are otherwise not present. The notion of metaphor is widened to include “those processes in which we perceive or think of or feel about one thing in terms of another” (Richards 1936: 116). Within pragmatically oriented approaches, exemplified by Searle’s (1979) account here, cognitive aspects, evidently, play a minor role. Searle (1979: 83) defines metaphors as follows: “a speaker utters a sentence of the form ‘S is P’ and means metaphorically that S is R”. As metaphorical utterances are characterized by the difference between what is said and what is meant, analyses of metaphors consist of two sentences: the metaphorical utterance and a literal paraphrase (Searle 1979: 81-82). (9) (MET) It’s getting hot in here. (PAR) The argument that is going on is becoming more vitupera- tive. (Searle 1979: 82) However, Searle is well aware that literal paraphrases are often “inadequate” (1979: 82) and that there are cases where no paraphrase can be 13 In Juliet is the sun! , quoted as example (1) above, Juliet is the tenor and sun the vehicle. Black, the second major early proponent of the interaction theory, uses different terms. Richards’s tenor is called principal subject (1962) or primary subject (1977) and the vehicle is called subsidiary subject (1962) or secondary subject (1977). 25 given because the language does not offer any literal way of expressing certain ideas (1979: 83). Figurativity is, thus, a necessary part of language. The relations between P and R are diverse. Searle (1979: 107-110) offers a list of eight principles, which is, as he points out (1979: 107), potentially incomplete. Interestingly, this list also embraces metonymic relations: “When one says, ‘S is P’, and means that ‘S is R’, P and R may be associated by such relations as the part-whole relation, the container-contained relation or even the clothing and wearer relation” (Searle 1979: 110). As far as metaphor proper is concerned, R can, for example, be a characteristic property of P things (e.g. Sam is a giant for ‘Sam is big’). Or P things can be stereotypically believed to be R (e.g. Richard is a gorilla for ‘Richard is mean, nasty, or prone to violence’). Some metaphors, he emphasizes, do therefore not rely on any relation between P and R that exists in the extralinguistic world. Sometimes it is just “a matter of perceptions, sensibilities, and linguistic practices” (1979: 97) that speakers see a connection between P and R. Searle quotes the example of coldness, which is conventionally associated with a lack of emotion, stresses that this metaphor is not culturespecific and that it leads to many expressions, like heated argument, warm welcome, and sexual frigidity (Searle 1979: 98), that are connected by the association of temperature with emotions (cf. also Leezenberg 2001: 122). Searle calls such instances ‘dead metaphors’, but for him they are still worth considering: Dead metaphors are especially interesting for our study, because, to speak oxymoronically, dead metaphors have lived on. They have become dead through continual use, but their continual use is a clue that they satisfy some semantic need. (Searle 1979: 83) Even though his approach is fundamentally pragmatic, it cannot be denied that metaphors and metonymies can and do play a role for the lexicon of a language. Figurative meanings can become established semantic interpretations of words, or, to put it differently, conventional. As this short overview has shown, some of the central ideas of the conceptual theory of metaphor are anticipated in earlier theories: First, metaphorical language cannot be entirely avoided, since we sometimes lack literal expressions. Thus, metaphors have, second, a firm place in everyday language, and figurative meanings can and do become conventional. Third, metaphors can go beyond language and play a role in thinking. Fourth, the relationship between the two constituents of a metaphor, whether called tenor and vehicle, or P and R, is sometimes only perceived because of the metaphor, and not otherwise given or extant. And fifth, the relationship between these constituent parts can extend to larger coherent systems of associations. The role of metonymy, however, is grossly underestimated in earlier research. It is often treated as a sub-type of metaphor (cf. Searle’s approach), and typically only mentioned in passing, even though Bredin (1984: 45), for 26 example, points out that metonymy could very well be more common than metaphor. 14 Before cognitive linguistics came into being, metonymy was usually defined as “the transfer of the name of a thing to something else that is closely associated with it” (Bredin 1984: 45). The notion of ‘close associations’ is obviously not a very precise one. Often, this close relationship has been referred to as contiguity (cf. e.g. Jakobson and Halle 1956: 76, 15 Lipka 1988: 358). As a defining criterion for metonymy the term contiguity is, however, problematic as it is always possible to see some type of contiguity between two terms (Schmid 1993: 93). The vagueness of the notion of contiguity becomes particularly evident in Ullmann’s (1962: 212, Fn. 1) definition: “’Contiguity’ should be taken [...] in a broad sense: it includes any associative relations other than those based on similarity”. Metonymy is thus assigned a secondary role. All figurative language which cannot be identified as metaphorical is metonymic. In this way, metonymy “is made the common dumping ground” (Bredin 1984: 47) within the realm of figurative language. To clarify the notion of contiguity, most authors (cf. e.g. Bredin 1984: 48; Leisi 1985: 190-191; Lipka 1988: 360-361; Ullmann 1962: 218-220) give extensive lists of possible metonymic relationships. Table 1 provides some examples. Metonymic relationship Example part for whole all hands on deck container for content a boiling kettle material for object a glass place of origin for food/ drink to drink champagne garment for wearer a blue-stocking cause for effect war is sad thing done for doer crime must be punished Table 1: Some examples of metonymic relationships It is difficult to find a single overarching principle which applies to all the examples and turns them into a coherent group, but one thing is evident: Metonymy has always been regarded as a phenomenon which leads to quite regular linguistic expressions (cf. Apresjan 1974: 16). Metonymies seem much more predictable than metaphors, because they presuppose an existing relationship between two entities, while metaphors create relationships. They function on the basis of our knowledge about the world, while 14 Cf. also Riemer (2002b: 382), who emphasizes that the role of metonymy is not unequivocal in traditional approaches. 15 In his 1971 article, Jakobson assigns a more fundamental role to contiguity. The two axes of language, the paradigmatic and syntagmatic, described by Saussure are reformulated in terms of similarity and contiguity, the two terms traditionally used to differentiate metaphor and metonymy. 27 metaphors generate new insights. Metonymy is, as Bredin (1984: 57) puts it, “irresistably [sic! ] and necessarily conventional”. It is paradigmatic of the treatment of metonymy that none of the authors provides a satisfying elaboration of the term contiguity, on which metonymy is said to be based. Metonymy was apparently not considered important enough to invest any energy in such a clarification. On the contrary, it seems that the indeterminate nature of contiguity was quite welcome, as it allowed researchers to put virtually any figurative expression which is not based on similarity into the convenient rag-bag labelled metonymy. Whether more recent approaches perform better in this respect will be discussed below, after the most fundamental tenets of the cognitive theory of metaphor and metonymy have been introduced. 3.2 Figurative language and figurative thought 3.2.1 The conceptual theory of metaphor Research into metaphor and metonymy changed significantly with the publication of Lakoff and Johnson’s book in 1980. As cognitive scientists, they see language not as an isolated phenomenon, but as a faculty that is intrinsically linked to other cognitive abilities, such as perception and reasoning. The focus of metaphor research has therefore shifted towards a search for general cognitive principles which govern the use of metaphorical language. The main argument is that metaphor is largely a matter of thought. The linguistic side is secondary, it is only a mirror image of deeply entrenched patterns of metaphorical thinking. Consider the following examples: (10) Look how far we’ve come. (11) We’re at a crossroads. (12) I don’t think this relationship is going anywhere. (13) We can’t turn back now. (14) We’ll just have to go our separate ways. (Kövecses 2002: 5) All the examples could have occurred in ordinary everyday discourse. Yet, they are metaphorical. In all of them the actual topic is a love relationship, but the language used to describe it is taken from the realm of journeys. The question is: Why don’t we talk about love in a more straightforward, ‘literal’ way? The problem we encounter when we try to talk about something as abstract as love is that we cannot help using terms from other domains of experience. Our ‘literal’ knowledge of love is, to put it mildly, limited so that it is extremely difficult to talk about it in a non-figurative way. Our whole idea of LOVE , the concept we have stored about this emotion in our minds, borrows much of its structure from other concepts. As Lakoff and Johnson (1980: 5) argue: “The essence of metaphor is understanding and experiencing one kind of thing in terms of another”. 28 Conceptual metaphors are “cross-domain mappings” 16 (Lakoff 1993: 203) which are part of the conceptual system. The general form of a metaphorical mapping is CONCEPTUAL DOMAIN A IS CONCEPTUAL DOMAIN B ; the mapping underlying examples (10) to (14) is LOVE IS A JOURNEY . The two domains involved have a different status: One does not have much structure of its own, while the other comes with a large amount of well-structured knowledge. To mark this difference, the domain which borrows its structure from another is called the target domain and the domain which provides the structure for the target is referred to as the source domain. 17 Target domain: Source domain: LOVE IS A JOURNEY the lovers ← the travellers the love relationship ← the vehicle events in the relationship ← the journey the progress made ← the distance covered the difficulties experienced ← the obstacles encountered choices about what to do ← decisions about where to go the lovers’ common goals ← the destination of the journey Table 2: L OVE IS A JOURNEY - the correspondences between target and source (after Lakoff 1993: 207, Kövecses 2002: 7) Mappings such as LOVE IS A JOURNEY rely on systematic conceptual correspondences between the target and the source (Lakoff 1993: 207), some of which are listed in Table 2. Importantly, mappings are not meant as active processing mechanisms language users induce every time they hear a metaphorical expression (Lakoff 1993: 210). They are highly conventional ways of thinking which are realized in language in many different ways. The linguistic metaphors are not random, but systematically linked to the metaphorical concepts (Lakoff and Johnson 1980: 7). L OVE is by no means the only thing we conceptualize metaphorically. Most if not all abstract concepts humans have to cope with and talk about, such as TIME , CHANGE , CAUSATION , ARGUMENTS , THEORIES , IDEAS , etc., are structured in this way and thus filled with knowledge. The main cognitive function of metaphor is therefore the explanatory one (Lakoff and Johnson 16 A domain is “any coherent organization of experience” (Kövecses 2002: 4). The term mapping is borrowed from mathematics (Dirven 1993: 207), where it refers to ‘an operation that associates each element of a given set [...] with one or more elements of a second set [...]’ (NODE, s.v. mapping). 17 The terms target and source domain largely correspond to what Weinrich (1958: 514- 515) calls bildspendendes Feld and bildempfangendes Feld respectively. The idea that metaphors link larger fields is not a novelty of conceptual metaphor theory. For a short overview of earlier theories which point to the idea that metaphor is always a relation between two domains or fields, cf. Lipka (1996). For an elaborate survey of the forerunners of conceptual metaphor theory, cf. Jäkel (1999). 29 1980: 154). 18 “Metaphor is the main mechanism through which we comprehend concepts and perform abstract reasoning” (Lakoff 1993: 244), because metaphorical mappings also allow us to employ inference schemas from one domain in another (Lakoff and Turner 1989: 64). As a consequence we can use the established conceptual mappings in a creative way to think about the target. In order to fulfil this function, mappings typically take the direction from concrete to abstract. 19 Since metaphorical thought is so common and omnipresent, metaphorical language, which is an explicit manifestation of this mode of thought, is far from exceptional. “[M]etaphor is pervasive in everyday life, not just in language but in thought and action” (Lakoff and Johnson 1980: 3). Metaphors are indispensable. We cannot do without them and metaphorical utterances cannot be paraphrased in a literal way without losing a considerable part of the information (Johnson 1987: 170). Lakoff and Johnson (1980) distinguish three major types of conceptual metaphor: structural, orientational, and ontological. L OVE IS A JOURNEY is a structural metaphor, as are THEORIES ARE BUILDINGS or ARGUMENT IS WAR , “where one concept is metaphorically structured in terms of another” (Lakoff and Johnson 1980: 14). Orientational metaphors have a more fundamental character. Here, whole conceptual systems are organized in terms of others. They “give a concept a spatial orientation” (Lakoff and Johnson 1980: 14). Examples are HAPPY IS UP (e.g. I’m feeling up), MORE IS UP (e.g. My income rose last year) or GOOD IS UP (e.g. He does high-quality work). Often, they are rooted in physical experiences: When a person is sad or depressed this shows in a drooping posture, but when someone is happy they are more likely to take on an erect posture. This is also related to GOOD IS UP . Happiness, life and well-being are all associated with being GOOD and therefore with being UP , and this can be transferred to all other things which can possibly be good for humans. Ontological metaphors are based on human experience with physical objects, first and foremost the human body. They allow us to see “events, actions, emotions, ideas, etc., as entities and substances” (Lakoff and Johnson 1980: 25). An example is THE MIND IS AN ENTITY . Such a general conceptual mapping, commonly called generic metaphor, can be elaborated in various ways, i.e. used as a basis of specific metaphors. The MIND can, for example, be conceptualized as a MACHINE (e.g. My mind just isn’t operating today, I’m a little rusty today) or as a BRITTLE OBJECT (e.g. Her ego is very fragile, I’m going to pieces). This illustrates that conceptual metaphors are organized in a hierarchical way. There are highly general mappings which form the basis for more 18 For a more detailed survey of the cognitive functions of metaphor, cf. Schmid (1998). 19 This unidirectionality of conceptual mappings has also been found to be the main principle underlying semantic change (cf. e.g. Sweetser 1990; Traugott 1895a, 1985b). Cf. also section 4.3.1.2. 30 specific ones. This is related to Grady’s (cf. e.g. Grady, Taub and Morgan 1996; Grady 1997, 1999) notion of primary metaphors. Primary metaphors are “basic […] metaphorical correspondences” (Grady 1999: 81) building on strong experiential correlations between two dimensions. 20 A good example is MORE IS UP / LESS IS DOWN : We all have experienced that when we add more of a substance to an already given quantity or amount of that substance, its level will rise. And when we take away some of it, its level will go down. Due to their firm grounding in human experience primary metaphors lead to metaphorical expressions that are usually perceived as well motivated (e.g. The prices are going up, Speak up, please), i.e. as ‘making sense’ (cf. Kövecses 2002: 70). More importantly, however, they can be combined to what Grady (e.g. 1997: 273) calls “’compound’ metaphor[s]”. This is, for example, the case in the mapping THEORIES ARE BUILDINGS , which consists of two primary metaphors: ORGANIZATION IS PHYSICAL STRUCTURE and PERSIST- ING IS REMAINING ERECT . These components allow speakers to focus on many different aspects in concrete realizations. The first is at work in all cases where the structure of a theory is manipulated in any way (constructed, rearranged, etc.) and the second highlights functional aspects. Only an erect structure can fulfil its function; if it falls down it can no longer work. As already indicated above, the same target can be conceptualized in different ways by using distinct source domains. The MIND IS A MACHINE leads to a conceptualization of the mind as having an internal mechanism that is driven by energy and can be turned on and off at will. If the MIND is considered a BRITTLE OBJECT , we highlight different aspects: Brittle objects can stand only a limited amount of pressure before they crack. While machines can usually be repaired when they are damaged, brittle objects that have been shattered will never be the same, even if one tries to repair them. They will always bear marks of the damage, whereas a machine which has been repaired works just as well as a new one would. The metaphors we choose thus determine how we think about events, actions, entities and the like. Every metaphor highlights some aspects but at the same time conceals others. When we use ARGUMENT IS WAR (e.g. He attacked every weak point in my argument) we focus on the competitive aspect. With another conceptualization, i.e. AN ARGUMENT IS A BUILDING (e.g. We’ve got a framework for a solid argument) we focus on the content and quality of an argument (Ungerer and Schmid 2006: 123). Metaphorical structuring is always “partial, not total” (Lakoff and Johnson 1980: 13). Metaphors only use parts of one domain and map them onto 20 Note that Grady (1999: 85) points out that the strong experiential correlations that motivate primary metaphors “sound strikingly like metonymic relationships between aspects of experienced scenes”, i.e. their original motivation might be a metonymic one. 31 another. The question of which parts are mapped is addressed in the invariance principle (Lakoff 1990; cf. also Brugmann 1990): Metaphorical mappings preserve the cognitive topology (that is, the imageschema structure) of the source domain, in a way consistent with the inherent structure of the target domain. (Lakoff 1993: 215) Image-schemas are schematic patterns of domains like CONTAINERS , PATHS , FORCES , etc. that are founded on basic bodily experiences (cf. e.g. Johnson 1987; Lakoff 1987; Lakoff and Turner 1989). What is to be preserved by the conceptual mappings are basic schemas like IN - OUT , UP - DOWN , FRONT - BACK , etc. (cf. Clausner and Croft 1999: 15; Croft and Cruse 2004: 45; Ungerer and Schmid 2006: 119). The invariance principle ensures that for example in the case of a CONTAINER schema, interiors of the source cannot be mapped onto exteriors of the target. An example is the CONDUIT META- PHOR (cf. Reddy 1979), where WORDS ARE CONTAINERS FOR MEANINGS . There are no instantiations of this mapping where the topology of the source is not upheld. In all cases such as His words carry little meaning or Your words are hollow the WORDS ARE THE CONTAINERS and the MEANING IS THE CONTENT . To clarify and widen the original intentions behind the invariance principle Ungerer and Schmid (2006: 119-120) have recently introduced the notion of the mapping scope, which is defined as the “set of constraints regulating which correspondences are eligible for mapping from a source concept onto a chosen target concept”. The mapping scope has three components: Besides the image-schematic structures also mentioned in the invariance principle, it contains basic correlations 21 and culture-dependent evaluations. As opposed to image-schemas, basic correlations are not experienced bodily. They are mainly needed to understand actions and events. We all know that PURPOSES correlate with GOALS and CAUSES with EFFECTS . Basic correlations between CHANGE and MOTION and PURPOSES and GOALS together with the PATH schema constrain the mapping scope of examples such as LIFE IS A JOURNEY or LOVE IS A JOURNEY . Culture-dependent evaluations can be illustrated with John is a pig. The English speaking community has agreed that what is mapped from source onto target in metaphorical construals of the type A PERSON IS A PIG are evaluative attributes like ‘dirty’ or ‘greedy’, i.e. the mapping A PERSON IS A PIG conventionally conveys that the person in question is ‘dirty’, ‘greedy’ or the like. Other speech communities may link PIGS with different associations so that metaphors which use PIG as a source domain will result in completely different meanings. This amply illustrates the role of conventionality for metaphors and their communicative success: Metaphors vary in how familiar single speakers are with their mapping scope, and only if a metaphor and its mapping scope are known to and shared by a considerable number of 21 The basic correlations bear some resemblances to Grady’s notion of primary metaphors (cf. above), but have a different function (cf. Ungerer and Schmid 2006: 120, 160, Fn. 7). 32 speakers is it possible to use it effectively in concrete communicative events (cf. Ungerer and Schmid 2006: 121). It is important to note that A PERSON IS A PIG does not have the same status as other mappings, like LIFE IS A JOURNEY or A THEORY IS A BUILDING . The latter are rich mappings (Ungerer and Schmid 2006: 125-127) since they provide an elaborate metaphorical structure for typically abstract target concepts. This is different with lean mappings like A PERSON IS A PIG . As they link a concrete source to a concrete target, the latter already comes with quite an elaborate structure of its own. The sole function of lean mappings is thus to “highlight one or a few additional attributes” (Ungerer and Schmid 2006: 126, original emphasis). Characteristically, however, metaphor is a cognitive instrument that is put to use when we conceptualize abstract targets. Parts of our knowledge about domains (Lakoff and Johnson 1980: 52-55) with which we are experientially familiar (Lakoff 1993: 239-241), such as the HUMAN BODY , PLANTS , BUILDINGS , etc., are used to understand and reason about domains which are otherwise hard to grasp, such as EMOTIONS , COMMUNICATION , TIME , LIFE , EVENTS and ACTIONS , etc. (cf. Kövecses 2002: 15-25). Because many common sources belong to the realm of universal human experiences, many metaphors occur in different languages and cultures. 22 As long as human thought and conceptualization is guided by metaphors to this extent, the conceptual mappings are very much alive (Lakoff and Johnson 1980: 55). The corresponding linguistic expressions may be “conventionally fixed” (Lakoff and Johnson 1980: 55), but this does not mean that the underlying mappings are no longer important for human thinking. The following sections provide an overview of the treatment of metonymy in cognitive linguistics. After giving a definition of metonymy, I will address its most basic cognitive and linguistic functions and then discuss the phenomenon of metonymy in the framework of reference-point constructions and active zones (cf. e.g. Langacker 1987, 1993). As in traditional linguistics, metonymy has received considerably less attention than metaphor - unduly so, because as will be shown, metonymy is at work in many different areas of language and cognition. 3.2.2 Metonymy as a cognitive-linguistic notion 3.2.2.1 Basic assumptions As pointed out above, traditional approaches did not pay much attention to metonymy. But it also took a while before cognitive linguists discovered metonymy as a phenomenon well worth studying. Lakoff and Johnson (1980), for example, only devote one of their chapters (and a very short 22 Many studies have been devoted to the question of whether metaphors are universal or culture-specific. Cf. e.g. Gibbs (1999), Kövecses (1995, 1999, 2000, 2006a), Quinn (1991). 33 one) to metonymy. In metonymies “we are using one entity to refer to another entity that is related to it” (Lakoff and Johnson 1980: 35). Even though this definition stresses the referential purpose, Lakoff and Johnson underline that metonymy has a cognitive function. It “provides understanding” (1980: 36) by allowing language users to focus on certain elements of what is actually being referred to. It is not by accident that we pick out particular aspects to refer to something or someone. When somebody is referred to as a good head, the person is conceptualized in terms of the properties associated with the linguistic expression. It is not a simple matter of PART FOR WHOLE , but a different way of conceiving of the person. The choice of the metonymic vehicle is thus always constrained by our conceptualization and functional aspects of the message: (15) The university needs more clever heads. (16) All hands on deck. (Ungerer and Schmid 2006: 128) The context determines which conceptualization is chosen. In (15) the university context makes properties associated with the head, or more accurately the brain, such as ‘intelligence’, much more relevant than properties associated with the hands. The latter are more relevant in the context given in (16), where the focus lies on ‘physical strength’ (Ungerer and Schmid 2006: 128). The communicative intentions differ remarkably in the two examples, and simple paraphrases of the kind ‘The university needs more people’ or ‘All people on deck’ lack much of the content which is conveyed by means of the metonyms (cf. also Alač and Coulson 2004; Radden 2000). Metonymies are of a conceptual nature and affect “our thought and our action” (Lakoff and Johnson 1980: 39). What metaphor and metonymy have in common is that both are systematic. Much in line with earlier approaches, Lakoff and Johnson (1980: 38-39) provide a list of metonymic relations and linguistic examples that realize them: PART FOR WHOLE (e.g. Get your butt over here! ), PRODUCER FOR PRODUCT (e.g. He bought a Ford), OB- JECT USED FOR USER (e.g. The sax has the flu today), etc. And just like metaphors, metonymies can be considered as mappings. In (15), for example, HEAD is the vehicle and the concept of the whole INTELLIGENT PERSON is the target. 23 But while metaphorical mappings are symmetrical (e.g. in CAREER IS A JOURNEY symmetric mappings hold between the beginnings and ends of journeys and careers, between the obstacles encountered in both, etc.), metonymic mappings are asymmetrical (Barcelona 2003: 233). The crucial difference between metaphors and metonymies is that the latter do not link two distinct domains of knowledge. The HEAD belongs to the PERSON , just as the HANDS do. A metonymy “involves only one concep- 23 Lakoff and Johnson (1980) analyze metonymies as involving a source and target domain. However, as Radden and Kövecses (1999) correctly point out, this is not an accurate description since metonymies function within one domain or ICM. Following their proposal, I will therefore use the term vehicle instead of source. 34 tual domain” (Lakoff and Turner 1989: 103). 24 As Barcelona (2003: 229-230) emphasizes, this standard definition is problematic since cognitive domains 25 are a private matter and vary from speaker to speaker. The following examples could thus be interpreted differently: (17) Mary is down. (18) John is a lion. (Barcelona 2003: 230, 231) At first glance, the only possible classification of both (17) and (18) seems to be as metaphors. However, one could very well hold that (17) is an example of the DOWN ( VERTICALITY ) FOR SADNESS metonymy, a more specific case of RESULT FOR CAUSE . V ERTICALITY can be regarded as a sub-domain of SAD- NESS (cf. also Barcelona 2002a: 233-236). For (18), a metonymic interpretation is less likely. But as Barcelona (2003: 231) convincingly argues, one cannot deny that the domains linked here, PEOPLE and ANIMALS , both belong to the overall domain of LIVING BEINGS . This would render (18) a metonymy, although such an analysis is clearly counter-intuitive. This illustrates that metaphor and metonymy are not sharply delineated phenomena, but form a continuum which leads from clear cases of metaphor through cases in which the distinction is hard to draw to clear instances of metonymy (cf. e.g. Dirven 1993; Radden 2000; Ruiz de Mendoza 2000). Due to being within-domain mappings, metonymic concepts are always experientially grounded (Lakoff and Johnson 1980: 39). We have experienced that a HEAD is a part of a PERSON (cf. 15) or that an INSTITUTION ultimately consists of PEOPLE (e.g. The Senate thinks abortion is immoral). Mostly, the vehicles and targets of metonymies are both concrete concepts (Ungerer and Schmid 2006: 130). The mapping is typically less complex than for metaphors. While metaphors rely on or rather establish many correspondences between source and target, the metonymic mapping scope is more skeleton-like. C ONTAINER FOR CONTENT metonymies (e.g. He drank a whole glass), for example, are based on the most basic aspects of the CONTAINER schema. And CAUSE FOR EFFECT metonymies (e.g. His native tongue is German) involve what Ungerer and Schmid (2006: 120) call “basic correlations”. In some cases, however, the mapping relies on looser associative links. In the whole table was roaring with laughter, where table stands for ‘people seated around a table’, it is less obvious where speakers see the com- 24 Lakoff and Johnson (1980: 36) and Lakoff and Turner (1989: 103) also provide other criteria. However, for Ruiz de Mendoza (2000: 113) the two-domain vs. one-domain relation is the only “true definitional” criterion. 25 While all researchers agree that metonymy is a conceptual process which involves only one mental representation, the terms to signify this mental representation vary considerably. As not all approaches and the differences between them can be discussed in detail, some examples will have to suffice: Radden and Kövecses (1999) speak of ICMs, Croft (2002) and Ruiz de Mendoza (2000; cf. also Ruiz de Mendoza and Díez Velasco 2002, 2004) of a matrix domain, Blank (1999) of a frame, Panther and Thornburg (1999) of a scenario, etc. 35 mon domain. Candidates are concepts like DINNER PARTY or SOCIAL EVENT , which are part of the speakers’ stored knowledge (Ungerer and Schmid 2006: 129). As evident from the discussion so far, the main focus of metonymy research usually is on so-called referential metonymies. 26 As Warren (1999: 123) points out, they are marked by three characteristics: (1) They have a referent, which is (2) not explicitly mentioned, but can (3) be retrieved by drawing inferences due to the conceptual connection between vehicle and target. An example is (19), where the only metonymic shift lies in the referential expression the Pentagon which stands for the ‘people responsible’. (19) The Pentagon has issued a warning. (Panther and Thornburg 2004: 97) (20) General Motors had to stop production. (Panther and Thornburg 2004: 103) This is different in (20), where the necessity to stop production stands for the occurrence of the event, or, in Panther and Thornburg’s (2004: 103) words, the OBLIGATION TO ACT stands for the ACTION . They call such cases predicational metonymies. However, (20) also involves a referential shift as General Motors actually refers to the ‘executives of GM’. Panther and Thornburg (2004: 103; cf. also Panther and Thornburg 2003b) classify utterances that contain both a referential and a predicational metonymy as propositional metonymies. Others, like Warren (1999: 122), also use this term, but in a slightly different way which is largely equivalent to Panther and Thornburg’s class of predicational metonymies. The last major class of metonymies are the speech act metonymies which emphasize the pragmatic functions of metonymy (cf. e.g. Panther and Thornburg 2003a). Panther and Thornburg (2003b: 6, 2004: 103) speak of “illocutionary metonymies” and show how metonymies can be used as inference schemas to explain conventionalized indirect speech acts like Can you close the window. According to Searle indirect requests can, for example, be made “by either asking whether or stating that a preparatory condition concerning H’s ability to do A obtains” (1979: 45; italics original). This is recast into the POTENTIALITY FOR ACTUALITY metonymy (cf. Panther and Thornburg 1998, 1999), which is said to underlie the comprehension of indirect speech acts as a general cognitive principle. 3.2.2.2 Metonymy as a reference-point construction Within the framework of Langacker’s Cognitive Grammar (1987, 1991, 2008) metonymy is considered part of a larger field of phenomena called reference-point constructions. A simple example are possessives: In an utterance 26 Referential metonymies can then be assigned to major ICMs, such as CAUSATION (e.g. RESULT FOR CAUSE as in ecstasy ‘drug’) or POSSESSION (e.g. POSSESSED FOR POSSESSOR as in He married money ‘person having money’; cf. Radden and Kövecses 1999); cf. also Warren’s (1999) classification. 36 S/ H like Sally’s dog has died the possessive construction allows the hearer to access the intended referent. The speaker refers to Sally in order to identify the DOG . In Langacker’s (1991, 1993, 2000) terminology, S ALLY is the reference point and the DOG is the target. Both belong to the same overall set of entities or domain (called dominion by Langacker). When we come across or use a reference-point construction, we establish a mental contact between the reference point and the target (cf. Figure 2). The reference point serves as an entry point into a larger domain and leads us to the right target. Once the actual target has been identified, the reference point which is foregrounded in the linguistic expression loses much of its significance: “when R is actually used as a reference point - it is the target thereby reached that now becomes prominent in the sense of being the focus of C’s [i.e. the conceptualizer’s; SH] conception” (Langacker 1993: 6; original emphasis). Figure 2: Essential aspects of the reference point ability (adapted from Langacker 1993: 6); R = reference point, T = target, S/ H = speaker/ hearer Metonymies function in much the same way. In She bought Lakoff and Johnson the metonymic expression, in this case Lakoff and Johnson, or rather its mental representation, acts as a reference point which guides the hearer to the intended target, i.e. the classic book on conceptual metaphor theory. The reference point is capable of invoking the target meaning because both belong to the same overall domain of knowledge. As Langacker puts it: the entity that is normally designated by a metonymic expression serves as a reference point affording mental access to the desired target (i.e. the entity actually being referred to). (Langacker 1993: 30) At the same time, metonymies are also instances of Langacker’s (1984, 1987: 271-274, 2000: 62-67) active zone phenomena. The following examples illustrate this: (21) We all heard the trumpet. (Langacker 1987: 271) (22) Give me a red pencil. (23) I’m in the phone book. (Langacker 1993: 32) T R Domain 37 What is meant in (21) is not the trumpet, but the sound it emits. Not all parts of the pencil referred to in (22) have to be red, it suffices when it leaves a red mark on paper. And what is in the phone book is certainly not the person, i.e. the I, mentioned in (23), but his/ her name and phone number. Those facets of an entity capable of interacting directly with a given domain or relation are referred to as the active zone of the entity with respect to the domain or relation in question. (Langacker 1987: 272-273; original emphasis) Which of the many potentially relevant facets of an entity is foregrounded, i.e. acts as the entity’s active zone, depends on the domain or relationship in question and the context. In other words, metonymies have a flexible mapping scope (cf. Ungerer and Schmid 2006: 129-131) that covers the possible relations within a single concept, but also encyclopedic cues “associated with different, though related, cognitive models” (Ungerer and Schmid 2006: 129). The metonymic expression the house of Windsor, for example, can be linked to different concepts, such as ROYAL FAMILY , MON- ARCHY , or E NGLISH HISTORY . But for the average language user this does not seem to be problematic (Ungerer and Schmid 2006: 129). Active zone phenomena are so natural and frequent that we are usually not aware of them. They allow us to communicate efficiently by striking a balance between “the need to be accurate and our inclination to focus explicit attention on those entities that most concern us” (Langacker 1993: 32). Metonymic thinking is thus a fundamental ability which is evident in many ways in language (cf. Langacker 1984). Reference-point reasoning and active zone phenomena are not restricted to relationships between wholes and parts. The active zone does not have to be a sub-part of the profiled entity (cf. example 23). It only has to be associated with the entity (Langacker 2000: 63). In this way, Langacker’s approach can be used to explain any possible metonymic relationship. Many researchers (e.g. Alač and Coulson 2004; Barcelona 2004; Radden and Kövecses 1999) have taken up Langacker’s ideas and incorporated them into their work. 27 At the same time, the range of phenomena studied 27 Langacker’s theory has, however, also given rise to criticism. Panther and Thornburg (2004: 98), for example, criticize it for being too general, as “it covers cases that in our view should not be treated as cases of metonymy”. According to Langacker the loss of my wallet would have to be interpreted metonymically in terms of the result of the loss, i.e. as ‘non-possession’. For Panther and Thornburg, such cases fall outside the realm of metonymy, because the relationship between loss and non-possession is a necessary one. In proper metonymies, they argue, the relationship is a contingent one instead. Cf. also Croft (2002), who holds that the profiling of different aspects of the domain matrix [ BOOK ] in This book is heavy and This book is a history of Iraq does not lead to true instances of metonymy. Still, he subsumes such cases under the more general heading of “domain highlighting” (2002: 179), which is also said to be at work in proper metonymies like Proust is tough to read, where, he argues, a metonymic shift 38 as metonymic processes was considerably enlarged. The following chapter gives a brief survey of the different levels on which metonymy functions. 3.2.2.3 The ubiquity of metonymic processes Langacker’s description of metonymy as a more general cognitive mechanism was not without consequences; particularly the idea of one conceptual entity granting mental access to another has been influential. This is reflected, for example, in the following definition of metonymy: Metonymy is a cognitive process in which one conceptual entity, the vehicle, provides mental access to another conceptual entity, the target, within the same idealized cognitive model. (Radden and Kövecses 1999: 21) Recent approaches to metonymy emphasize its relevance for many areas of thought and language. In some ways, metonymy is more basic than metaphor. The whole language system, for example, can be said to have a metonymic foundation as Radden and Kövecses (1999: 24) demonstrate with their notion of sign ICM. The FORM FOR CONCEPT metonymy (e.g. dollar for ‘money’; cf. also Lakoff and Turner 1989: 108) is the basis for all linguistic signs and is also applied in the reference ICM, where a FORM - CONCEPT PAIR- ING stands for a THING or EVENT (e.g. the word cow for a real cow). On the conceptual level, metonymy plays an important part in reasoning processes, as already discussed in connection with Langacker’s reference-point constructions. However, it also guides categorization to a large extent. The very existence of prototypes (cf. e.g. Rosch 1973, 1975), for example, indicates that categories are habitually construed and perceived in a metonymic way. Further support comes from Lakoff (1987: 77-90), who demonstrates that many cognitive models are based on metonymy. Often, categories are understood by way of sub-parts such as ideals, paragons, salient examples, etc. On the linguistic level, metonymy also shows up in many different guises. Simple linguistic metonyms of the type He bought a Picasso are only the tip of the iceberg. For example, metonymy is at work on the level of morphology and word-formation (cf. Ungerer and Schmid 2006: 155-157). This concerns stock examples of bahuvrihi compounds (e.g. skinhead, paperback) as well as conversions (e.g. to author or to pepper). The latter can be understood as metonymies where an AGENT or OBJECT stands for an ACTION (cf. Dirven 1999; 28 Schmid 2005: 194-196, 198-205; Ungerer and Schmid 2006: 155-157). In addition, many publications emphasize the importance of leads from one sub-domain of the domain matrix [P ROUST ] to another, i.e. from the PERSON to his WORK . 28 Dirven (1999) may have been the first to explicitly discuss conversions as metonymies (cf. Ungerer and Schmid 2006: 155), but he was certainly not the first to notice the phenomenon. The Master Metonymy List (Leite 1994: 16-18) provides many examples belonging to what is called action frame there, such as He fathered many children or Butter the toast. 39 metonymy for the explanation of pragmatic phenomena (cf. e.g. Panther and Thornburg 2003a, 2003b, 2004). Metonymy is omnipresent (cf. Barcelona 2002b): It plays a role in grammar, including word-formation (cf. e.g. Panther, Thornburg and Barcelona 2009), gesture (cf. e.g. Mittelberg and Müller in prep.), categorization, and inferencing. Moreover, metonymy often motivates or interacts with metaphor (cf. e.g. Barcelona 2000c; Geeraerts 2002, Goossens 1990; Grady 1999; Niemeier 2000; Radden 2000; Ruiz de Mendoza and Díez Velasco 2002; Ruiz de Mendoza and Pérez Hernández 2003). The extension of metonymy into the realm of thinking has been accompanied by a reconsideration of its functions. Whereas Lakoff and Johnson (1980: 36; cf. also Lakoff and Turner 1989: 103) still talk about the primacy of the referential function, the above-mentioned examples confirm that this view can no longer be upheld. Metonymy serves as a general, automatic cognitive mechanism in many central areas. This chapter has thus far summarized the main assertions of the conceptual theory of metaphor and metonymy. Most importantly, metaphor and metonymy are no longer considered purely linguistic phenomena; significant cognitive functions are attributed to them. And the more fundamental the underlying reasoning mechanism is, the more conventional are the linguistic expressions that mirror it. In the next sections, another theoretical model is introduced - the theory of conceptual blending (e.g. Fauconnier and Turner 1998, 2002). Originally developed as a means of describing the speakers’ understanding of creative language use, it can also be helpful in distinguishing conventional from less conventional figurative language. 3.3 Mental spaces and conceptual integration 3.3.1 Basic assumptions The conceptual theory of metaphor is largely based on analogical reasoning. By way of mapping a source onto a target, knowledge and inferences already available in the former also become available in the latter. The theory of conceptual blending, however, points out that the process of what is called conceptual integration leads to novel conceptualizations which cannot be explained by analogy or metaphor alone. Blending theory has its roots in mental space theory (Fauconnier 1994), an account of language understanding that underlines the constructivist nature of meaning. The major assumption underlying Fauconnier’s work is that language only provides us with “minimal, but sufficient, clues” (1994: xviii), which we use to build mental representations that serve as a starting point for reasoning processes. Meaning does not reside in language - language only 40 prompts us to construct meaning (Fauconnier 1994: xxii, 1997: 37; Turner 1991: 206). Any linguistic input triggers the formation of so-called mental spaces, i.e. theoretical “constructs distinct from linguistic structure but built up in any discourse according to guidelines provided by the linguistic expressions” (Fauconnier 1994: 16). They are partial mental representations of the input and continuously modified as a discourse unfolds. Spaces do not “have to be logically consistent”; they are not models of the world but discourse models (Fauconnier 1994: xxxvii). As they are used for on-line processing only, mental spaces are of a temporary nature. Once we have made sense of the linguistic input at hand, they are usually discarded, since they are no longer relevant. Similar to Fillmore’s (1975, 1985) frames, the elements within spaces have roles. And these roles can be mapped onto one another, i.e. the corresponding counterparts in different spaces can be identified and connected (Fauconnier 1994: chap. 2, 1997: 11-13). Let me illustrate this with a simple example: (24) Gina bought a sports car. (25) Paolo believes that Gina bought a pickup truck. (Croft and Cruse 2004: 32) Upon encountering (24) the hearer creates a mental space which contains information concerning G INA (a), the SPORTS CAR (b) and the relation between the two, which can be formalized as ‘ BUY a b’. This space acts as what Fauconnier (1997: 38) calls the base space for the understanding of ensuing utterances, which lead to the construction of further spaces. Every new space is related to an already existing space and the meaning constructed for the space which is currently in focus (the focus space) is dependent on the meaning constructed for the base space. Example (25) contains a so-called space builder, “a grammatical expression that either opens a new space or shifts focus to an existing space” (Fauconnier 1997: 40). Space builders can have many grammatical forms. Examples are expressions such as in 1929, in that story, in Susan’s opinion, etc. (Fauconnier 1997: 40). They all imply that there is or could be a difference between reality and the construal triggered by a linguistic expression. What Susan thinks or what happens in a story does not necessarily coincide with reality. And Paolo’s belief expressed in (25) does not match what is described in (24). This conflict can be illustrated as in Figure 3 (below). The structure of the two spaces is similar: a and a’ and b and b’ correspond to each other - a and a’ as the buyer and b and b’ as the object bought. The elements a and a’ also correspond to one another in terms of what Fauconnier calls role, i.e. a linguistic description of a category (Croft and Cruse 2004: 34). But the other elements differ: In the base space the role is SPORTS CAR , in the belief space we find a PICKUP TRUCK . 41 Figure 3: Mental spaces and meaning construction (based on Croft and Cruse 2004: 32-35) This basic set-up of mental space theory is the foundation for the theory of conceptual blending. Blending is a cognitive mechanism frequently used in on-line processing. Partial structures from different spaces are integrated into a single space which is marked by emergent properties (Fauconnier 1997: 22). The focus is on how information from more than one space can be used to gain new insights. To this end, a network of mental spaces is created. In its simplest version it consists of four elements: two input spaces, the generic space, and the blended space. A general model of such a network is given in Figure 4. Figure 4: The basic structure of a conceptual integration network (adapted from Fauconnier 1997: 151) a name Gina b sports car BUY a b a’ name Gina b’ pickup truck BUY a’ b’ a b a’ b’ Base Space Belief Space Generic space Input space 1 Input space 2 Blended space 42 The two input spaces hold partial mental representations of the linguistic input and are linked by cross-space mappings. The generic space is linked to each of the inputs. It “contains what the inputs have in common” (Fauconnier and Turner 1998: 137). The fourth element, the blend, contains structure it inherits from the inputs, but is also characterized by emergent structure - signified by the stars is Figure 4 - that is nowhere to be found in the inputs. A good illustration of what it means ‘to run a blend’ is Fauconnier and Turner’s (1998: 154-156) regatta example: (26) At this point, Great America II is 4.5 days ahead of Northern Light. (Fauconnier and Turner 1998: 154) Two voyages are blended here: One by the modern catamaran Great America II in the year 1993 and one by a clipper, Northern Light, which took the same route from San Francisco to Boston back in 1853. The two inputs are the temporal spaces of 1993 and 1853. Both contain parallel elements: the boats, their paths, their positions on the course, etc., which are linked by cross-space mappings. 29 But they differ with respect to two points: the overall time frame (1993 vs. 1853) and the time the boats need to cover certain distances. The projection from the inputs to the blend is partial. What is not mapped, for example, are the conditions of travel back in 1853. However, both boats are projected into the blend and as their path is basically the same, their positions on the course can be compared, i.e. one can be ahead of the other. The emergent structure leaves us with the impression of a race between the two boats. The emergent structure arises due to three major processes: composition, completion, and elaboration (Fauconnier and Turner 1998: 144, 2002: 48-49). Composition describes how relations from one space are attributed to elements in another. It provides “relations that do not exist in the separate inputs” (Fauconnier and Turner 1998: 144). Applied to (26) this means that the two boats are projected into a single space, where their positions can be compared. Completion has to do with background knowledge. If two boats travel along the same path and have different positions with one being ahead of the other, this evokes a larger knowledge structure which fits this overall layout, the RACE frame. Elaboration refers to the fact that once we have established the basic content and structure of the blend, it is possible to elaborate on them. Once the RACE frame is evoked, this will lead to further construals of the journeys as a race. If the Great America II takes less time for the whole journey than the Northern Light, it is going to win the race, etc. This process of running the blend is, in principle, open-ended (Ungerer and Schmid 2006: 260). 29 The cross-space mappings are based on the so-called vital relations such as change, identity, time, space, cause-effect, etc. (cf. Fauconnier and Turner 2002: 89-111). 43 The main point of the whole blending process is compression (Fauconnier and Turner 2002: 92-93; cf. also Fauconnier and Turner 2000). The originally complex inputs are reduced to those elements which can grant new insights. Blending is a highly efficient process since it reduces the load for our cognitive apparatus to a minimum by focusing on the relevant aspects only. At the same time, it allows a maximum of creativity, since the compressed structure in the blend can be used as a starting point for potentially open-ended reasoning processes. The main criticism blending theory has recently attracted aims at the actual computability of such integration networks (cf. Harder 2003; Ritchie 2004; Veale and O’Donoghue 2000) and the question of whether the assumptions of blending theory can be put to the test and falsified (cf. Gibbs 2000; Ritchie 2004). The processes assumed by the theory are, however, not as unconstrained as it seems at first glance. Fauconnier and Turner (2002: 325-334; cf. also Fauconnier and Turner 1998: 162-163) discuss several optimality principles which govern the blending process. According to the topology principle, for example, the blend takes over important structural aspects of the inputs. The principle of integration requires that the blended space be structured in a way that allows “its manipulation as a unit” (Fauconnier and Turner 2002: 329). These and other principles, such as the web principle, the pattern completion principle, etc., regulate the relationship between blend and inputs. Others are more fundamental (Ungerer and Schmid 2006: 266), like the principle of relevance, which states that elements in the blend should be relevant for the ongoing talk exchange by way of “indicating its connections to other spaces or indicating the lines along which the blend is to develop” (Fauconnier and Turner 2002: 333). Blending theory has sparked much research in different directions. It has been used as a framework to explain longer stretches of discourse (cf. e.g. Oakley and Hougaard 2008), constructions (cf. e.g. Mandelblit and Fauconnier 2000), word-formation (cf. e.g. Benczes 2006; Schmid forthcoming a; Ungerer and Schmid 2006: 268-275), compositionality (cf. e.g. Sweetser 1999), performativity (cf. e.g. Sweetser 2000), privative adjectives (cf. Coulson 2001: 144-157), semantic change (Grygiel 2004), etc. However, it has also been applied to metaphor - and to some extent also to metonymy (cf. e.g. Coulson and Oakley 2003; Grady, Oakley and Coulson 1999; Alač and Coulson 2004). How metaphor and metonymy research has profited from conceptual blending theory will be discussed in the next chapter. 30 30 When used as a descriptive framework for the analysis of metaphor, conceptual blending theory bears some similarities to interaction theory (cf. chapter 3.1). The proponents of the latter maintain that the two parts involved in a metaphor cooperate in meaning construction, i.e. that the resultant meaning is not only determined by the vehicle (Black 1962: 44; cf. also Jäkel 1997: 99, Fn. 22). 44 3.3.2 Metaphor, metonymy, and the theory of conceptual blending That blending theory can be fruitfully applied to metaphor was soon recognized. One of Fauconnier and Turner’s (e.g. 1998, 2002; cf. also Coulson 2001: 168-170) stock examples, digging your own grave, is a case in point. Conceptual metaphor theory treats this idiom as a straightforward mapping: Crucial elements of the source domain of GRAVE DIGGING such as the DIGGER and the GRAVE are mapped onto the target, i.e. the UNPLEASANT SITUATION of a person. The DIGGER corresponds to the AGENT in the target and the GRAVE to the UNPLEASANT SITUATION the agent is about to face. Both the elements in the source and the elements in the target stand in a causeeffect relationship: The digger’s actions cause the grave and the person’s actions cause the bad situation. There is, however, a crucial difference between them. Whereas the result of grave-digging is the digger’s intended goal, the outcome of the person’s actions in the target domain is undesired. Furthermore, and perhaps more importantly, the element DEATH plays significantly different roles in the two domains: In the source it triggers the digger’s actions (who is by the way usually not responsible for the death of the person to be buried), while in the target the undesirable situation the agent is about to face corresponds to the death of the person who is digging their own grave, i.e. death is the result of the action and not the cause. As Fauconnier and Turner (1998: 149) put it: “Foolish actions cause failure, but grave-digging does not cause death”. The causal structure as well as the event sequence in the two domains do not match. This is a serious problem for the conceptual metaphor view. Blending, however, can account for such cases. The blend recruits the elements of grave-digging from the source input, but the causal structure, the unintentionality of the action, and the event sequence come from the target. The emergent structure in the blend thus contains the otherwise fantastic assumption that “the existence of a grave causes death” (Fauconnier and Turner 1998: 150; original emphasis). 31 Another example is This surgeon is a butcher (Grady, Oakley and Coulson 1999: 103-106). Analyzed as a conceptual metaphor, the source domain is BUTCHERY , the target domain SURGERY . Various correspondences hold between them: The BUTCHER is the SURGEON , the ANIMAL the PATIENT , the CLEAVER the SCALPEL , etc. The problem with this account is that the crucial semantic element of the metaphorical utterance, i.e. the fact that the surgeon is depicted as incompetent, is nowhere to be found in the source. Once more, blending theory can do more than conceptual metaphor theory. A simplified version of Grady, Oakley and Coulson’s (1999: 105) graphic representation of the blend is given below in Figure 5. 31 This is further sanctioned by the common metonymic association between GRAVES and DEATH , and by conventional metaphors such as FAILURE IS DEATH and DOWN IS BAD (cf. Coulson 2001: 170; for the metaphors cf. e.g. Lakoff, Espenson and Schwartz 1991). 45 Input space 1 Input space 2 Blended space Figure 5: This surgeon is a butcher - a conceptual blending analysis (adapted from Grady, Oakley and Coulson 1999) The blend takes over structure from both inputs: From the source it inherits the role BUTCHER and the activities connected to it. From the target it inherits the identities 32 of the AGENT and the PATIENT and the LOCATION . The emergent structure has its roots in the incompatible means-end relationships in the two spaces. In the blend, the means of the BUTCHERY space are combined with the ends of the SURGERY space. This incongruity “leads to the central inference that the butcher is incompetent” (Grady, Oakley and Coulson 1999: 106). A blending analysis of the example is superior to an analysis as a conceptual metaphor. However, the blending account involves a vital element rooted in conceptual metaphor theory: The cross-space mappings between the AGENTS , PATIENTS , INSTRUMENTS , etc. equal the cross-domain correspondences of conceptual metaphor theory. These correspondences “shape and constrain” (Grady, Oakley and Coulson 1999: 101) the process of meaning construction. Conceptual metaphors can thus be a starting point 32 The distinction of role and identity in Grady, Oakley and Coulson’s account of this example goes back to the distinction of roles and values in Fauconnier’s (1997) mental space theory. Values are ‘role-fillers’ (Fauconnier 1997: 108), i.e. individual instances of the categories set up by the roles. Cf. also Fauconnier and Turner (2002: 98). Role: Surgeon Identity of surgeon Role: Patient (Person) Identity of patient Scalpel Operating room Goal: healing Means: surgery Role: Butcher Role: Commodity (Animal) Cleaver Abattoir Goal: severing flesh Means: butchery Identity of surgeon - Role: Butcher Identity of patient - Role: Patient (Person) Unspecified instrument Operating room Goal: healing Means butchery Incompetence 46 for more complex reasoning processes (Grady, Oakley and Coulson 1999: 110-111). The differences between blending theory and conceptual metaphor theory are evident (cf. Grady, Oakley and Coulson 1999: 102-106): Conceptual metaphor theory works with domains, i.e. relatively stable conceptual representations with an elaborate structure. Blending theory works with spaces, which are partial mental representations of ongoing discourse characteristically used for the purposes of on-line processing only. While conceptual metaphor theory holds that metaphors involve only two domains with a fixed mapping direction from source onto target, blending theory has at least four spaces. Often, it is possible to draw a parallel between the inputs of blending theory and the source and target of conceptual metaphor theory, but the process of blending is not restricted to cases with only two inputs. Most crucially, blending goes beyond cross-domain mappings and argues that the results of reasoning processes find their place in the blend, which consists of elements projected from all the inputs as well as emergent structure of its own. It is particularly this latter aspect that makes conceptual blending so attractive for metaphor research. The idea that metaphors can bring about novel insights is, however, not as unique to conceptual blending as often assumed. Tourangeau and Rips (1991), for example, observed in psycholinguistic experiments that metaphor interpretation leads to what they call “’emergent’ features” (1991: 459), 33 which can be central for the understanding of a metaphor (1991: 470). They refer to Murphy’s (1988) and Medin and Shoben’s (1988) results, who studied the comprehension of complex noun phrases such as apartment dog or empty store and found that the combination of concepts leads to a foregrounding of features which may be given somewhere in the concepts of the constituents, but which are not of particular significance if the concepts are considered in isolation. Similarly, certain metaphors tend to highlight aspects which would not be weighty in attribute lists of the target and the source concepts. 34 Probably they would not even be listed, since they do not inherently belong to the concepts, but are - at best - possible attributes. Together with their emergent structure, blends can become shared knowledge. Typically, Fauconnier and Turner concentrate on novel uses in order to reveal creative understanding mechanisms. However, they do not claim that knowledge once gained by way of conceptual blending cannot 33 In other psycholinguistic approaches, as for example in Gentner’s structure-mapping theory, it is held that metaphor interpretation relies on the shared features of tenor and vehicle (cf. e.g. Gentner and Clement 1989). In more recent versions of Gentner’s theory emergent structure is dealt with in terms of alignment (cf. e.g. Bowdle and Gentner 2005). 34 Cf. also Becker (1997) who claims that the emergent features in metaphor interpretation mostly belong to the vehicle term but are relatively unimportant in its mental representation (1997: 254). 47 be stored if it proves useful on more than one occasion. Blends can become conventional (Fauconnier and Turner 1998: 133, 136, 2002: 70; Grady, Oakley and Coulson 1999: 110): The argumentation often takes the following specific form: a particular process of meaning construction has particular input representations; during the process, inferences, emotions and event-integrations emerge which cannot reside in any of the inputs; they have been constructed dynamically in a new mental space - the blended space - linked to the inputs in systematic ways. For example, “They dug their own financial grave” draws selectively from different and incompatible input frames to construct a blended space that has its own emergent structure and that provides central inferences. In this case, the blended space has become conventional. (Fauconnier and Turner 1998: 135-136; my emphasis, SH) The relationship between conventional conceptual metaphors and blends is a two-sided one. On the one hand, a blend can become conventional and feed into a conceptual metaphor. The ship of state metaphor (cf. Grady, Oakley and Coulson 1999: 111), for example, has been originally created by a blending process, but has become a conventional metaphorical conceptualization which can be elaborated in various ways. On the other hand, blending often builds on conventional conceptual metaphors by exploiting metaphorical correspondences between domains or spaces. The two research paradigms do not exclude but complement one another. The apparent contradiction between the long-term conceptual structures emphasized by conceptual metaphor theory and the short-term mental representations proposed by conceptual blending theory is nothing but an effect of the different foci of the theories. Metonymy also plays a big role in conceptual blending. On a more abstract level, metonymy as a general reasoning process belongs to the foundations of the blending process. Fauconnier and Turner (1998: 170-171; cf. also Coulson and Oakley 2000: 186) mention metonymic tightening as one of their optimality principles: If the blend contains elements which stand in a metonymic relation to each other, the metonymic distance between them is shortened by way of compression. An example is the Grim Reaper (cf. e.g. Fauconnier and Turner 1998). In reality, the temporal distance between a person’s death and the skeleton left after decomposition is quite large. In the blend, however, the relation between cause and result is a direct one. Here, the skeleton is Death. In a similar way, metonymy plays a role in the vital relations posited by Fauconnier and Turner (2002: 92-102). What is compressed due to conceptual integration are, among other things, partwhole relations and cause-effect relations. Both are well-known types of metonymic reasoning. But metonymy does not only govern blending. The comprehension of linguistic metonyms can also be regarded as a blending process. Since metonyms are usually said to rely on knowledge structures which belong to one and the same domain or ICM (cf. chapter 3.2.2.1), there does not 48 seem to be much room for creativity in terms of emergent structure. However, metonymy is not restricted to its referential function (cf. e.g. Barcelona 2002b, Murphy 1996): Metonymic concepts allow us to conceptualize one thing by means of its relation to something else. When we think of a Picasso, we are not just thinking of a work of art alone, in and of itself. We think of it in terms of its relation to the artist, that is, his conception of art, his technique, his role in art history, etc. We act with reverence toward a Picasso, even a sketch he made as a teen-ager [sic! ], because of its relation to the artist. (Lakoff and Johnson 1980: 39) Alač and Coulson (2004: 23-24) argue convincingly for a blending account of the seemingly simple PRODUCER FOR PRODUCT metonymy a Picasso. A possible conceptualization is given in Figure 6. Input space 1 Input space 2 P ABLO P ICASSO P ICASSO ’ S ART - Picasso as artist - P’s conception of art - P’s techniques - P’s role in art history etc. the work of art referred to all of P’s works of art H knows - H’s attitudes towards P’s art the value of P’s works of art etc. Blend A P ICASSO a work of art by P important in art history of a certain style - H’s attitudes towards P’s art valuable etc. Figure 6: Emergent structure of the Picasso-blend (adapted from Alač and Coulson 2004: 24); (P = Picasso; H = hearer) The conceptual network consists of two inputs, the P ABLO P ICASSO space and a space containing ideas about P ICASSO ’ S ART . Within the P ICASSO space, the person is considered primarily as an artist. Thus, input 1 contains ideas like ‘his conception of art’, ‘his techniques’, ‘his role in art history’, etc. Input 2 is concerned with the works of art he produced. All of ‘Picasso’s works of art’ the hearer knows play a role here, as well as the hearer’s attitudes towards them and certainly the idea that - no matter if the hearer likes them or not - they are ‘valuable’. The blend, i.e. the actual interpretation of the expression, will therefore be more detailed than a simple referential account of metonymy would predict. The blend is linked to the inputs and inherits many of the aspects mentioned as their parts. The 49 style and importance of the picture are emphasized by way of the metonymic construal as well as its value, which is likely to be the most important idea, at least for the layperson. The cognitive advantage is obvious. The contents of the inputs can be easily activated and feed effortlessly into the blend. The blend, as it were, collects them and offers the language user a large amount of information for a minimum of effort - on the conceptual level due to the effortless access to a large amount of knowledge and on the linguistic level due to the shortness of the expression. 35 3.4 Summary This chapter has discussed how cognitive linguistics conceives of metaphor and metonymy. Both thought mechanisms help us to understand and cope with the world. Moreover the linguistic manifestations, i.e. figurative linguistic expressions, offer a maximally efficient way of conceptualizing and communicating rich and complex meanings. While this holds for creative examples as typically analyzed by blending theorists, it is particularly true for linguistic metaphors and metonyms that are based on deeply entrenched mappings and whose figurative meanings have become so common that their actual figurativity is no longer noticed. And it is precisely because we are not aware of the underlying mappings that they are so essential in terms of cognition. The more subconscious a reasoning procedure is, the less effort one has to invest. This renders the degree of conventionality of a mapping and its linguistic instantiations a vital factor. To put it simply, the more common and familiar conceptual mappings and the linguistic metaphors and metonyms they motivate are, the more cognitively efficient are they. In the following, I will therefore focus on how the conventionalization and conventionality of figurative language can be described and measured. 35 For a further discussion of the example a Picasso, where it is considered as a paragon, cf. chapter 4.2.2.2. Its conventionality is also addressed in chapters 6.2.1.3 and 6.2.1.4. 51 4 Motivation and conventionality in language and thought Many metaphors and metonymies belong to everyday language. Nevertheless, not all of them are equally conventional. Not all linguistic realizations that are sanctioned by the existence of conceptual mappings are also used by speakers. This chapter will look at the question of the conventionality of figurative language in more detail. Even though cognitive linguists agree that conventional figurative language is most apt to grant insights into the workings of the human mind, the meaning of the term conventional has often been more or less taken for granted. For this reason I will not only focus on the cognitive tradition, but also refer to what structuralism has to say about conventionality. The role of figurative language has been seriously underestimated in the latter framework, but many findings which were originally meant for a description of ‘literal’ language, mostly wordformations, can be usefully transferred to metaphor and metonymy. The main questions to be addressed in the following sections are: What exactly does conventionality mean and how can we describe it? In what relationship do conventionality and motivation stand to each other? What makes some linguistic expressions more conventional than others? And which characteristics of conventional figurative expressions can be operationalized for empirical research? 4.1 Motivation: From structuralism to cognitive linguistics 4.1.1 Non-cognitive approaches to motivation and conventionality The notion of conventionality is certainly associated with Ferdinand de Saussure in every linguist’s mind. When he postulates the arbitrary relationship between the two sides of the linguistic sign ([1916] 1983: 67), he points out that the fact that the one side invariably triggers the other is a matter of convention ([1916] 1983: 71). Once a speech community has incorporated an arbitrary form-meaning pairing into its collective mind, speakers are bound to it. This suggests that conventionality is of paramount importance with respect to language, since established linguistic conventions govern our linguistic behaviour to a large extent and are usually beyond our conscious control (Saussure [1916] 1983: 71-73). However, motivated signs are also subject to the established conventions of a language. Saussure ([1916] 1983: 63) details that onomatopoeia (together with exclamations, the only signs for which he speaks of a proper motivation) can lose their motivation and become arbitrary. The boundary 52 between arbitrariness and motivation is thus not fixed and stable. The effect of conventionality on (relatively) motivated signs (cf. Saussure [1916] 1983: 130) can be illustrated with the competing forms mobile phone and cell phone which both stand for the same concept (Schmid 2005: 107). 36 That the British favour mobile phone, whereas American speakers prefer cell phone 37 demonstrates the role of conventionality. As both lexemes have been institutionalized 38 they seem to be equally apt to transport the meaning in question. It is up to the speech community to decide which linguistic form is preferred. While Saussure rightly sees no contradiction between motivation and conventionality, other linguists working in his tradition, such as Ullmann (1957, 1962) treat the two terms as antonymous, if not complementary. Ullmann replaces Saussure’s notion of arbitrariness by the term conventionality and defines it as “the absence of any intrinsic motivation or justification, any ‘natural’ connection between the name and the sense” (1957: 84). The three types of motivation he distinguishes (1957: 86-89, 1962: 82-93), i.e. phonetic (e.g. cuckoo, whizz), morphological (e.g. penholder, speaker), and semantic motivation (for metaphors and metonymies, e.g. hood (of a car), the cloth ‘the clergy’), thus invariably lead to ‘unconventionality’. Ullmann’s purely language-immanent approach does not leave any room for the speech community, but it is impossible to find a satisfactory explanation for the issue of motivation and conventionality without taking both the levels of system and use into account. A tripartite terminology, such as Saussure’s, fits the problem at hand much better than Ullmann’s ideas. From the point of view of the language system, form-meaning pairings can either be arbitrary or motivated. And from the vantage point of the language community, both types can be conventional or not. 39 All form-meaning pairings - not only the arbitrary ones - have to be accepted by the speech community. This concerns not only the preference for one of several possible forms, as mentioned above for mobile phone and cell phone. For word-formation, this relates for example to the question of blocking (e.g. *stealer, *raper vs. thief, rapist). 40 But also novel complex lexemes which are not subject to any restrictions on productivity do not ne- 36 While Saussure only considers combinations of full linguistic signs, Marchand (1969: 2-3) and later Ungerer (1991b) also discuss the motivation of words which are not composed of full linguistic signs, as for example acronyms, blends or clippings. 37 Cf. LDOCE4, s.v. mobile phone and cell phone. 38 For the term institutionalization, cf. e.g. Bauer (1983); Fischer (1998); Lipka (1992, 2002); Schmid (2005); and below chapter 4.2.1. 39 A similar position is held by Ungerer (1991a; cf. also Ungerer 1991b: 155), who also points out that conventionality is a term that is intricately linked to the speech community. 40 For word-formations, linguists have described a number of criteria which block the formation and/ or incorporation of new lexemes into the lexicon and which would therefore be considered ‘unconventional’ by the average language user (cf. e.g. Bauer 1983: 84-99; Kastovsky 1982: 158-164; Plag 1999, 2003: 59-68; Schmid 2005: 115-119). 53 cessarily gain wider currency. Some words which are coined due to an immediate need do not reach the status of linguistic conventions. 41 And the same holds true for metaphors and metonyms. While some of them represent very well-established ways of talking, others seem much more creative and are immediately felt to be figurative and innovative. Conventionality is commonly associated with ideas like ‘normal’, ‘usual’, and ‘socially accepted’. A given form-meaning pairing should thus be considered conventional if it represents a generally established manner of expression that is shared by a given speech community. Speakers belonging to this community can process and understand it automatically and effortlessly, without having recourse to their knowledge about wordformation rules or metonymic or metaphorical mappings. As opposed to word-formations, the conventionality of figurative language has so far received little attention. What is at issue here is why some metonymic and metaphorical expressions stand a better chance of becoming conventional than others. Which are the factors that govern the conventionality of metonyms and metaphors? Are the speech community’s preferences in this respect idiosyncratic or is it possible to find some regularities which explain why some metonyms and metaphors are more conventional than others? One aspect that could foster the conventionality of form-meaning pairings is motivation. Well-motivated expressions are easier to understand than less motivated ones, 42 so language communities might favour them. To assess the influence of motivational factors on a possible subsequent conventionalization, I will now turn to more recent approaches to motivation. 4.1.2 Motivation and cognition Cognitive linguists hold that human behaviour is usually motivated. As a rule, there is a reason for the things we do, so that our behaviour can be characterized as actions carried out to reach desired goals. It is prompted by some conscious or unconscious stimulus, i.e. motivation, which triggers the steps taken and gives them a certain direction (Radden and Panther 2004b: 3). Language as a form of human behaviour thus should be motivated. As Lakoff (1987: 346) puts it: “In natural language, motivation seems to be more the norm than the exception”. The structuralist concept of motivation, which concentrates on language-internal factors only and draws an artificial boundary between the language and its users, has to be replaced 41 Cf. chapter 4.2.1. 42 Schmid (2008: 10-13) discusses this with regard to new words. Apart from the familiarity with the morphemic constituents and the semantic relationship between them, other factors like the amount of co(n)textual information influence how easily new words are comprehended. Cf. also his account of the relationship between semantic transparency, storage, processing, and frequency (2008: 22-26). 54 by a more comprehensive notion, which takes pragmatic and sociolinguistic as well as cognitive aspects into account (cf. Radden and Panther 2004b: 3; Ungerer 1991a: 161). Traditionally, motivation was often thought of as applicable to linguistic form only, but cognitive linguists point out that motivation is not restricted to this side of the sign. Sense extensions, for example, have therefore become prominent in cognitive-linguistic research. This can be illustrated with over. As regards the most central sense of the preposition, there are competing opinions. While Lakoff (1987: 417) argues that it consists of a combination of two semantic elements, i.e. ‘above’ and ‘across’ (e.g. The plane flew over), Tyler and Evans (2003: 64-66) emphasize that the primary meaning is the stative ‘above’-sense (e.g. The picture is over the mantel). More relevant for my present concerns is, however, that the meaning can be extended way beyond the central sense(s) to embrace metaphorical conceptualizations, as in Sam was passed over for promotion (Lakoff 1987: 436). All senses of over form meaningful chains. Even occurrences of over in larger linguistic structures such as overlook and oversee are part of these chains and motivated because “it makes sense for these words to have these meanings” (Lakoff 1987: 438; original emphasis) if one takes the underlying metaphors into account. 43 Lakoff defines motivation as “an independently existing link, L [which] makes sense” (1987: 448; original emphasis) of a relationship between two elements, A and B. Due to these links speakers feel that a given linguistic expression is a “natural” (1987: 449) way of translating ideas into words. The most comprehensive account of possible motivational links is given by Radden and Panther (2004b). They (2004b: 14-23) propose the following typology of motivational links: 1. Content motivates form 2. Form motivates content 3. Content 1 motivates content 2 4. Form 1 motivates form 2 5. Form-content pairing 1 motivates form-content pairing 2 The first type can be illustrated with onomatopoetic words, where the forms closely resemble the intended referents. The second case, where a form motivates a content, accounts, for example, for phonesthemes (Radden and Panther 2004b: 18-19), i.e. associations between forms and meanings on a nonor sub-morphemic level. Verbs beginning with / fl/ (fly, flow, flee, float, etc.), for instance, are typically associated with movement 43 For overlook Lakoff (1987: 437) mentions conceptual metaphors such as SEEING IS TOUCHING and the MIND IS A BODY (which entails LOOKING AT SOMETHING IS TAKING INTO CONSIDERATION ), and for oversee he quotes the metaphor CONTROL IS UP , which interacts with the metonymy SEEING SOMETHING DONE STANDS FOR MAKING SURE THAT IT IS DONE . 55 through air or water (cf. Marchand 1969: 412, Schmid 2005: 45). 44 The third class, the motivation of one content by another, was already illustrated above by means of the metaphorical extensions of the meaning of over. 45 Content-content relations apply when one lexeme has at least two related meanings. This relationship is of particular relevance for my concerns since it covers metonymic and metaphorical senses. In general, every figurative sense is motivated by its link to the literal sense of the same linguistic form. 46 This, however, does not entail speakers always needing to have recourse to the literal meaning in order to understand a metaphor or metonym. It only means that it makes sense to use one and the same linguistic form to denote different, but related entities. Motivational links between two linguistic forms are only found on the phonological level and concern phenomena like assimilation or vowel harmony (Radden and Panther 2004b: 20). More interesting, still, are motivational links between two form-content pairings (Radden and Panther 2004b: 22-23). This includes cases like Gosh! for ‘God! ’, where the form is slightly changed, but still quite similar to the original. The similarity of form also accounts for the semantic similarity. Cognitive linguistics stresses all possible connections between forms and meanings and yields a much more varied and complete picture of motivation than traditional approaches. Forms, contents and form-content pairings can serve as triggers for motivational processes, and the targets of motivational processes are just as diverse as their sources. The motivational resources are extended to include not only phonetic and morphological qualities of words but also graphemic characteristics, metaphorical and metonymical extensions as well as free associations connected with the meanings of words (Ungerer 1991a: 163). Motivation affects all kinds of linguistic signs, even simplex lexemes (Ungerer 1991a: 162). A large proportion of language is motivated in one way or the other. But why do speakers favour motivated links? The reason given by Lakoff is as simple as it is convincing: It is easier to learn something that is motivated than something that is arbitrary. It is also easier to remember and use motivated knowledge than arbitrary knowledge. (Lakoff 1987: 346; original emphasis) 44 Early treatments (cf. e.g. Jespersen 1925: 398) already point out that the onomatopoetic properties of such phonemes or phoneme clusters are only sensed when they happen to fit the meaning. Phonesthemes can also be associated with more than one meaning (cf. e.g. Lu 1998: 58-59). Ungerer (1991a: 163) emphasizes that phonesthemes are highly language-specific, i.e. a matter of convention. 45 Cf. also Evans and Tyler (2004), who discuss the meaning extensions of the preposition in. 46 Even though speakers are often not aware of these motivational links when they use language, they can be retrieved with relative ease in many situations. This has been observed, for example, for idioms and metaphors (cf. e.g. Gibbs and O’Brien 1990; Sweetser 1990). 56 Motivation is cognitively advantageous, because motivated language demands less effort from the language user. The single items stored in people’s minds are not isolated from each other. They form motivated and meaningful networks or chains, which are maximally informative. As Taylor (2004: 57) points out, the degree of motivation of a given linguistic structure depends on “the extent [to which] it is related to other structures in the language”. From this perspective, everything in language is motivated to a greater or lesser extent. All forms are related to some others, all meanings enter into conceptual relationships with other meanings. The first use of a new complex lexeme or of a figurative meaning of an already existing word is sanctioned by the above-mentioned motivational principles. But this is only the beginning. When a new word or a novel meaning is used or heard for the first time, its links to other items within the system are still restricted. A nonce-formation, for example, would only be linked to the forms and meanings of its constituents and a novel metonymy mainly to the literal (or other established) meaning(s) of the same form. With repeated use, the links become stronger and more and more new links or “pointers” (as Taylor 2004: 58 puts it) to other items are established so that the knowledge network becomes denser. As a consequence it becomes easier to use the new item, since it can be accessed from an increasing number of entry points. What Ungerer (1991a: 168) calls source motivation, i.e. the more linguistic aspects of motivation, like phonological, graphemic, morphological, and semantic qualities, interacts with the frequency of use (Ungerer’s frequency motivation) and can foster the processing of the linguistic items in question, which - in turn - enhances their level of motivation. 47 From a cognitive perspective, motivation and conventionality are thus intricately related. The more motivational links an item has or acquires, the more firmly it is anchored in the conceptual network and the easier its use and processing become. Therefore, chances are good that well motivated items become conventional. The motivational links, as it were, support the conventionality of linguistic items. The process of conventionalization, roughly sketched above, attracted linguists’ attention long before the advent of cognitive linguistics. The pro- 47 In his pragmatically oriented approach, Ungerer (1991a, 1991b) distinguishes three scales which influence successful communication. Referential interpretation deals with (non-)arbitrariness in the Saussurean sense, cognitive interpretation with mental processes, and sociolinguistic interpretation with the role of the speech community in terms of regional, social, etc. preferences for certain forms and meanings. For Ungerer, motivation relates to cognition only and conventionality is considered a phenomenon which depends on sociolinguistic variables. The notion of motivation used here and in other cognitive-linguistic accounts (cf. e.g. Radden and Panther 2004b), is broader and covers all of the factors Ungerer mentions. Moreover, conventionality is also regarded as a cognitive phenomenon. The boundary between speech community and cognition seems to be dispensable as sociolinguistic behaviour will undoubtedly be reflected in cognition. 57 cesses that affect linguistic units when they become firmly established parts of the language system have mainly been treated under the headings of institutionalization and lexicalization. In the following chapters, I will first provide a brief summary of structuralist linguists’ findings with respect to these processes. As they mainly focus on complex lexemes, I will assess to what extent metaphorical and metonymic meaning extensions can be described in a similar fashion. The findings of structuralism then will be reconsidered from a cognitive-linguistic angle and related to some central cognitive-linguistic notions, such as salience and entrenchment. 4.2 Approaches to conventionalization and conventionality 4.2.1 The conventionalization of language: Institutionalization and lexicalization In traditional accounts two complementary terms are used to describe the process of conventionalization: The first, institutionalization, refers to sociopragmatic aspects, while the second, lexicalization, concentrates on the word and its internal characteristics (Schmid 2005: 73, 2008: 3). In accordance with the earlier neglect of figurative language, they have been applied mostly to word-formations. 48 I will therefore briefly introduce the two notions, using classic examples from the domain of word-formation before I discuss how they can be transferred to metaphor and metonymy. Institutionalization refers to the incorporation of new lexical material into the lexicon of a language. A new word enters “the common usage of a speech community” (Fischer 1998: 15), “[i]t starts to be accepted [...] as a known lexical item” (Bauer 1983: 48). The beginning of this process is marked by the fact that any potential ambiguity of the original nonceformation 49 is typically ignored (Bauer 1983: 48). Telephone box, for example, has been institutionalized as ‘telephone kiosk’, but other meanings are theoretically possible (e.g. “box shaped like a telephone, box which is located at/ by a telephone”; Bauer 1983: 48). Institutionalized lexemes are relatively transparent, but speakers are familiar with the form and have established a connection between the form and one specific meaning - the novel item has become part of the item-familiar lexicon (cf. Meys 1975, 1985). 48 Notable exceptions are Lipka (e.g. 1998, 2002: 108-110) and Tournier (1985: 47-51), who explicitly demand the study of novel senses of existing lexical items within the field of lexicology. 49 There are many competing terms like ad-hoc formations, nonce-formations, deictic compounds (cf. esp. Downing 1977), contextuals (cf. esp. Aronoff 1980), innovations (cf. esp. Clark and Clark 1979), etc. which all refer to different types of novel lexemes or uses. Hohenhaus (1996) provides a useful overview of the terminology used within the wider field of ad-hoc formations. 58 Which factors favour the incorporation of lexemes into the common core or at least into a variety of a language? Fischer (1998) mentions several aspects. Of prime importance, however, is the frequency of use. It is relevant for the process of institutionalization in two different, but related, respects: On the one hand, it generates institutionalization, since a high frequency of use can result in institutionalization. 50 On the other, it is the effect of institutionalization, i.e. can be regarded as the “the safest measure for institutionalization” (Fischer 1998: 172). Further factors are alternative expressions which compete with the lexemes in question, the degree of productivity of word-formations which use the new word as its basis, and motivation (Fischer 1998: 178-181). Fischer (1998) has a more traditional understanding of motivation and therefore regards it as being of “limited overall importance” (1998: 179). What is not considered is that lexemes which are on the verge of becoming institutionalized are more and more linked to already stored linguistic knowledge and thus supported by increasingly dense conceptual networks. From a cognitive perspective, institutionalization is therefore heavily dependent on motivation in the sense of the place of a lexeme within the conceptual network. Not only novel lexemes can be institutionalized and thus become known to the majority of speakers of a language, but also new senses of already existing forms, i.e. metaphorical and metonymic senses (cf. e.g. Lipka 1992, 1998). As already pointed out in chapter 4.1.2, such senses are invariably motivated. The degree of motivation is probably linked to the question of how familiar speakers are with the underlying metaphorical or metonymic reasoning pattern. As opposed to word-formation patterns, which are commonly described in terms of their productivity (cf. e.g. Bauer 2001, 2005; Plag 1999, 2006), metaphor and metonymy have traditionally been said to be creative, i.e. unpredictable and not rule-governed (cf. e.g. Bauer 1983: 63; Lyons 1977: 548-549). However, in the light of recent research one can seriously doubt this conclusion. Just consider the typical mapping direction of metaphors (cf. Lakoff and Johnson 1980) or the systematicity of metonymic readings (Markert and Nissim 2003: 176). Admittedly, it is at least theoretically possible to link any two domains by way of a linguistic metaphor and metonymic meanings are often said to be openended (cf. e.g. Markert and Nissim 2002a: 204), because of the open-ended nature of contiguity (cf. e.g. Riemer 2002a: 384, 2002b: 383; Schmid 1993: 91- 93). What counts, however, is that highly creative and extraordinary metaphors, which, for example, reverse the typical direction of mapping 51 will 50 Institutionalization has to be distinguished from topicality. Topicality causes a higher frequency for a limited period of time due to a certain topic being of current interest (Fischer 1998: 174-175). An example is millennium bug, a word which was highly frequent around the millennium, but which is no longer frequent today. 51 Jäkel (1997: 65-88) tested the unidirectionality hypothesis and observed that metaphors which diverge from the usual mapping direction are much harder to understand. An example is Das Weib ist ein vernünftiges Märchen (‘The woman is a reason- 59 hardly gain currency. And the same holds for highly context-dependent metonymies, like ham sandwich ‘customer who ordered a ham sandwich’ (as in The ham sandwich is at Table 7; cf. also chapter 4.3.2.1). The latter’s conventionalization is likely to be blocked by the fact that there is no nameworthy category (cf. Downing 1977: 823) it could denote. 52 The acceptance of word-formations as well as metaphors and metonyms presupposes the existence of a relatively stable category they are meant to label. The metonymic pattern FOOD ORDERED FOR CUSTOMER is not likely to lead to institutionalized senses, since the people referred to form an unstable category at best. This poses constraints on the use of the pattern, so that it is only of a very restricted profitability (cf. Bauer 2001: 207-211) Acceptability thus determines whether a metaphor or metonym will be used frequently enough to become institutionalized. It is a graded phenomenon (Leech 1981: 212- 215). There is definitely a decrease of acceptability in the following examples: (27) Sam is an absolute: pig, rat, mouse, hawk, [...] reindeer, wombat, platypus, stegosaurus. (Leech 1981: 214) The reason why people are much more likely to be re-categorized as pigs, rats, mice or hawks than as reindeer, wombats, platypuses 53 or stegosaurs lies in our (folk) knowledge about the animals. We associate stereotypical properties with pigs, rats and the like, which can be utilized in metaphors. Pigs are supposed to be ‘dirty’ and ‘greedy’, rats are ‘sly’ and ‘cunning’, etc. But what properties would be ascribed to a person who behaves like a wombat or a platypus? At least in Central Europe speakers lack enough factual information or stereotypes to use these mappings. 54 A metaphor which employs a source that does not contain any relevant knowledge that can be applied to the target will hardly be accepted, as it cannot fulfil any explanatory function. Sam is a platypus undoubtedly raises more questions than it answers. The metaphor is uninformative because it remains unclear which category of humans is being alluded to. For metonymies, acceptability also depends on categorization and knowledge. Linguistic metonyms are potentially acceptable if they denote a more or less stable category. And the ICM within which the mapping oc- able fairy-tale’; my translation, SH). Only 46.2% of the participants of his study indicated that they had understood this metaphor, while the metaphors which obeyed the principle of unidirectionality were understood in up to 100% of the cases. 52 This is connected to the fact that the existence of a word tends to imply that there is a single, unified entity or category that can be denoted by the word. Leisi (1975: 26) refers to this phenomenon as hypostatization, Leech (1981: 32) calls it the “’conceptforming power’ of the word”. For a detailed account of the role of concepts regarding the creation and use of new words, cf. Schmid (2008). 53 The word platypus itself is actually a metonymy, as one salient aspect is picked out to refer to the animal as a whole. 54 This is related to the cognitive-linguistic notion of salience, cf. esp. chapters 4.2.2.1 and 6.2.1.4. 60 curs has to be filled with a certain amount of knowledge. Take the PRODUC- TION ICM as an example. To understand metonyms like I’m reading Shakespeare or The BMW was parked behind the house it is necessary to know that both Shakespeare and the BMW company can be regarded as producers, that both produce(d) certain kinds of products, i.e. literature and cars, and that both are well-known for exactly these products and not for others. Even though this may seem trivial, language users need a considerable amount of encyclopedic knowledge in order to find the metonymic references meaningful. Nevertheless, the factors contributing to the acceptability of a metaphor or a metonym are not more than indicators of a potential institutionalizability. Which metaphors and metonyms indeed acquire the status of linguistic conventions is not entirely predictable. When a new word-formation or a new sense of an already existing form has been institutionalized, it can be subject to the process of lexicalization. With regard to word-formation, Lipka (2002: 111) defines the term as the phenomenon that a complex lexeme once coined tends to become a single complete lexical unit, a simple lexeme. Through this process it loses the character of a syntagma to a greater or lesser degree. Lexicalization can be regarded as a gradual historical process which depends on the frequency of use. 55 It affects either the formal side of the sign, its semantic content or both. But in any case, it creates a certain amount of distance between the original morphological building blocks (and their meanings) and the complex lexeme. As a result, the lexemes lose some of their motivation (in the traditional sense) and acquire a certain degree of idiomaticity or opacity. This process is fostered by the frequency of the word in question, since a frequent usage or encounter of a word almost invariably entails it being used or heard in a variety of contexts, which leads “to a constant enrichment of the set of conceptual associations” (Schmid 2008: 25-26; cf. also Ungerer and Schmid 2006: 47-58). Formal changes can be exemplified with the lexeme forecastle. In its British spelling variant fo’c’sle, the original morphological structure is lost so that it seems to be a simple lexeme. This is mirrored on the phonological level in the British pronunciation / / . A good example to illustrate semantic changes is wheelchair (cf. e.g. Leech 1981: 226; Lipka 1992: 109, 2002: 114; Ungerer and Schmid 1998: 87-89), which cannot be interpreted accurately by drawing on the meanings of its morphological constituents only. As Ungerer and Schmid (2006: 96) point out, the cognitive categories which underlie the compound “do not only rely on the two categories sug- 55 Not all linguists agree as to whether lexicalization should be regarded as a gradual historical process that depends on the frequency of use. For different positions, cf. e.g. Bauer (1983: 48-50), Kastovsky (1982: 164), and Lipka (2002: 113). Useful overviews of the different definitions of the term have more recently been provided e.g. by Brinton and Traugott (2005) and Hohenhaus (2005). 61 gested by the linguistic form”. Knowledge from additional categories like INVALID , HOSPITAL , ENGINE and BRAKE has become an important part of the mental representation of the complex lexeme. Due to the fact that the word can no longer be understood by drawing on the mental representations of its constituent parts, it is necessary that the word is stored as an independent unit in speakers’ minds. 56 Lexicalization also applies to figurative language (cf. Lipka 1992: 109, 1998: 105). One example is dogfight 57 (Lipka 1992: 109). In its literal sense ‘a fight between dogs’ (NODE, s.v. dogfight) the word is quite transparent. Compared to this, its figurative meaning ‘a close combat between military aircraft’ is more idiomatic. This is why Lipka argues that the latter is lexicalized. However, when one looks at the figurative meaning in isolation, without considering the literal one, one could very well hold that the figurative sense is simply another institutionalized meaning, 58 given that this meaning has not developed any further since it was used for the first time. The lexeme dogfight has a second metaphorical meaning, however, i.e. ‘ferocious struggle for supremacy between interested parties’, where actual violence is no longer a necessary component (“the meeting deteriorated into a dogfight”, NODE, s.v. dogfight). This second metaphorical sense could thus be claimed to be lexicalized by virtue of being a further extension of the first figurative, military sense. Another aspect which has been cited as an indicator of the lexicalization of metaphors and metonymies is the loss of the derivational relation (cf. Grimm 1991: 92), i.e. that users are no longer aware of the semantic relationship between the literal and the figurative meanings (Hansen et al. 1990: 205). This however, presupposes the existence of such a relationship to begin with, a condition which, for example, does not hold for metaphorical word-formations which do not have a literal meaning, such as daisy wheel. This process can only be at work when lexemes with a literal meaning also have a figurative lexical unit. 59 An example could be the word culture (cf. Bowdle and Gentner 2005: 209; and chapter 4.3.1.2). Its most frequent senses, i.e. ‘the arts and other manifestations of human intellectual achievement’ or ‘the customs, arts, social institutions, and achievements of a particular nation, people, or other social group’, date back to the 56 Cf. Schmid’s (2005: 73-85, 2008) description of the process of concept-formation. 57 Cf. also Geeraerts (2002), who analyzes figuratively motivated compound expressions with the help of a prismatic model. 58 The creation of a figurative meaning is not a sufficient criterion to consider an item lexicalized, if this figurative meaning does not entirely replace the literal one. Otherwise, one would have to claim that derived figurative senses are motivated and unmotivated (or demotivated) at the same time, since structuralists usually take metaphorical and metonymic senses to be semantically motivated and hold that lexicalization causes a loss of motivation. 59 The distinction of lexemes and lexical units goes back to Cruse (1986: 49). Lexical units are connections of one form and a single sense, whereas lexemes can have more than one sense. 62 19 th century only and are actually metaphorically derived from the earlier sense ‘growing; cultivation of the soil’ (NODE, s.v. culture 1 ). As Cruse (1986) emphasizes, the process which causes the loss or reduction of the derivational link hinges upon the frequency of use of the figurative sense: If, however, a metaphor is used sufficiently frequently with a particular meaning, it loses its characteristic flavour, or piquancy, its capacity to surprise, and hearers encode the metaphorical meaning as one of the standard senses of the expression. Interpreting it then no longer requires the activation of the metaphorical strategy, working through the literal meaning, but merely requires the looking up, as it were, of a dictionary entry [...]. (Cruse 1986: 42) This quotation illustrates three points: First, it shows the deeply rooted belief that metaphorical language is, at least initially, something extraordinary and outstanding. Second, it indicates the adherence to the primacy of literal meanings, which was current at the time Cruse’s book was published. Third, it demonstrates the underestimation of the role of cognition. Metaphors which no longer require an active search for their creative meanings are considered to function on the same level as arbitrary formmeaning pairings. Merely due to the fact that they have become conventional parts of the lexicon, instances of metaphor and also of metonymy are assumed to lose a large part of the insights that can be gained by metaphorical and (to a lesser extent) by metonymic construals. But as Lakoff and Turner (1989) point out: The mistake derives from a basic confusion: it assumes that those things in our cognition that are most alive and most active are those that are conscious. On the contrary, those that are most alive and most deeply entrenched, efficient, and powerful are those that are so automatic as to be unconscious and effortless. (Lakoff and Turner 1989: 129) What Lakoff and Turner emphasize is that stored figurative senses fulfil more cognitive functions than literal senses ever can. They allow access to a larger amount of knowledge, simply by virtue of being based on mappings, and do so in a cognitively economic way. And these properties of figurative language remain intact long after the figurative senses have become institutionalized and thus become established and conventional parts of a language. This chapter has shown that many originally structuralist notions, such as productivity (referring to the existence of established mappings which can be exploited to varying degrees), acceptability (referring to the preconditions which have to be met in order for a metaphor or metonymy to become a widely accepted way of speaking), and institutionalization (referring to the actual incorporation of metaphors and metonymies into a language) can be usefully applied to approach the conventionality of figurative language in the sense of conceptual metaphor theory. However, when it comes to the functions and storage of figurative senses, one finds decisive differences between older and more recent accounts, since tradi- 63 tional approaches equate the conventionality of figurative language with a massive decrease in terms of its usefulness and informativity. In the next chapter I will re-analyze these findings from a cognitive perspective and relate them back to what has been said about motivation in chapter 4.1. The productivity and acceptability of figurative language depend heavily on the cognitive-linguistic notion of salience. And the actual institutionalization is related to the notion of entrenchment (Langacker 1987: 59). Finally, if metaphors or metonyms are subject to further semantic specializations, i.e. semantically lexicalized in structuralist terminology, they can be said to acquire what conceptual blending theory (Fauconnier and Turner 1998, 2002) calls emergent structure. This account of the basic building blocks of the conventionality of metaphors and metonymies will then lead further to a discussion of typologies of metaphors and metonymies, which take their conventionality into account. 4.2.2 Conventionality and cognition: A preliminary sketch 4.2.2.1 The notions of salience and entrenchment From a cognitive perspective, the above-mentioned stages a novel complex lexeme potentially goes through are marked by differences with regard to the fixedness of the conceptual representation (cf. Schmid 2005: 71-85, 2008). An item is institutionalized when more and more users begin to perceive the complex unit in a holistic way. This goes hand in hand with the fact that it is stored in the minds of an increasing number of speakers. Moreover, the new entry in the mental lexicon is linked to many other stored concepts. In addition to the motivational links which are responsible for the creation of a novel item in the first place and which, as has been pointed out in chapter 4.1.2, remain active long after the new unit has become an established part of the lexicon, new connections are created. What structuralists term lexicalization is characterized by the fact of the complex lexeme losing its former connections to its constituent parts. By then, the item is supported by such a large number of conceptual connections that the original building blocks are no longer of prime importance. Langacker (1987) describes this with the help of his scaffolding metaphor: [C]omponent structures are seen as scaffolding erected for the construction of a complex expression; once the complex structure is in place (established as a unit), the scaffolding is no longer essential and is eventually discarded. (Langacker 1987: 461) Similarly to novel complex lexemes, new or uncommon metaphors and metonyms can only be interpreted if speakers have recourse to their ‘scaffolding’. Here, however, the scaffolding is not found on the linguistic surface, but on the conceptual level. A creative metaphor like His theory has lots of little rooms and long, winding corridors (Lakoff and Johnson 1980: 53; cf. chapter 4.3.1.1) is supported by the entrenched mapping THEORIES ARE 64 BUILDINGS . The linguistic expression might not have been encountered before. Consequently the addressee will be forced to look for a conceptual basis which links THEORIES and BUILDINGS . Other instantiations of the same underlying mapping, such as His early results form the building blocks for his theory or They constructed this theory from the ground up (Lakoff, Espenson and Schwartz 1991: 114) are more familiar. The latter cases have, as it were, already shaken off their scaffolding and been stored. Still, this does not mean that the expressions have simply acquired the status of ‘literality’. Even though the scaffolding is no longer immediately necessary for comprehension, it remains part of speakers’ cognitive apparatus and sanctions the formation of novel linguistic expressions on the same basis. The likelihood that the ‘scaffolding’ becomes superfluous and can be dispensed with 60 is largely dependent on the frequency with which a unit is processed. An increase in frequency results in a greater degree of entrenchment of the corresponding linguistic unit in single speaker’s minds (cf. also Bybee 1985: 117): Linguistic structures are [...] conceived as falling along a continuous scale of entrenchment in cognitive organization. Every use of a structure has a positive impact on its degree of entrenchment, whereas extended periods of disuse have a negative impact. With repeated use, a novel structure becomes progressively entrenched, to the point of becoming a unit; moreover, units are variably entrenched depending on the frequency of their occurrence [...]. (Langacker 1987: 59; original emphasis) It is important to note, however, that the effect of changes in frequency is most pronounced “at the low-frequency end of the spectrum” (Forster 2007: 42). The entrenchment of a word proceeds fastest when it is heard for the first few times. Once it is entrenched to a reasonable degree, even larger changes in its frequency have less prominent effects (cf. Schmid 2008: 21) As Schmid (2007: 119) points out, it is also possible to use the notion of entrenchment to refer to the speech community taken as a whole. Single speakers’ familiarity with linguistic units influences the whole speech community’s familiarity with the items, as it determines “the frequency with which its members are exposed to them”. This is certainly true, but I prefer to keep the individual and the collective levels terminologically apart: The term entrenchment will be used to refer to the former, while conventionality will be reserved for the latter. Linguistic structures which have a higher degree of entrenchment are processed more automatically than others. The concept of entrenchment is therefore linked to the notion of salience. This term is used with two differ- 60 As Schmid (2008: 24) emphasizes, however, the access to the constituents of complex words is not entirely blocked, even if they are no longer informative about the meaning of the complex lexeme. This, obviously, only holds for cases of semantic lexicalization. In cases of highly formally lexicalized items, such as lord (from Old English hlāfweard ‘breadkeeper’; cf. Sauer 1992: 356), a morphological analysis is ruled out. 65 ent meanings in cognitive linguistics (Schmid 2007: 119-120). One of the senses, called cognitive salience by Schmid, refers to the current activation state of a cognitive unit. A mental concept can be activated either as a result of spreading activation (cf. e.g. Collins and Loftus 1975; Collins and Quillian 1969) or because the concept in question is activated in a more direct way, i.e. linguistically mentioned or situationally prominent. That concepts can enter the focus of attention due to spreading activation demonstrates the paramount importance of the position of a linguistic structure within a dense knowledge network. If a meaning, say of a metonym like seat for ‘person who occupies the seat’ as in Ask seat 19 whether he wants to swap (Nissim and Markert 2005), has never been encountered before, it cannot be arrived at by way of spreading activation. But the more links a given item has to others in speakers’ minds, the more likely it will become activated by a simple spread of activation and therefore with no processing cost worth mentioning. Good candidates for spreading activation are, for instance, concepts which are linked by way of traditional sense relations. Schmid (2007: 119) mentions the example DOG , the activation of which facilitates the activation of other concepts, such as “’bark’, ‘tail-wagging’, ‘fur’, ‘poodle’, ‘alsatian’, ‘collie’, etc.” The only way to activate the above-mentioned metonymic meaning of seat is by being exposed to an utterance like the one given above and/ or by being in a situation in which this utterance could be made, i.e. on a plane. If the utterance was heard on a plane, the situational context makes an effortless comprehension likely. Most people travelling on planes know that the seats are numbered and that passengers cannot simply choose a seat for themselves. This situationally relevant knowledge is probably activated when a person is on a plane, so that the metonymic link between the seat number and the person occupying this seat does not cause problems. Both concepts the PASSENGERS and the SEATS are part of the attentional focus, i.e. cognitively salient. In a more neutral situational context, the link is much more difficult to grasp. This is related to the second usage of the term salience in cognitive linguistics, to what can be called ontological salience (cf. Schmid 2007: 120). Ontological salience describes the attention-attracting potential of realworld entities. Some entities are more likely to attract our attention than others: Other things being equal, various principles of relative salience 61 generally hold: human > non-human; whole > part; concrete > abstract; visible > non-visible; etc. (Langacker 1993: 30) 61 Langacker (1991: 171, 1993: 30) treats these laws under the heading of cognitive salience. In general, Langacker does not distinguish between the two different facets of the term salience. This is sanctioned by the fact that the two are by no means independent of each other. 66 Seat numbers do evidently not belong to those entities which have a large attention-attracting potential. They are not ontologically salient, while humans have a high degree of ontological salience. The metonymic mapping SEAT NUMBER FOR PASSENGER therefore does not conform to the general laws of ontological salience and can only be used for successful reference if the situation or other considerations make the otherwise hardly salient vehicle highly (cognitively) salient (cf. Alač and Coulson 2004: 28; Langacker 1993: 30, 1991: 171, Fn. 14). Ontological salience is intrinsically linked to cognitive salience. Because ontologically salient entities will more frequently become a person’s focus of attention, the corresponding concepts will be activated more often, i.e. become cognitively salient. This is also where salience and entrenchment meet: The cognitive representations of ontologically salient entities are processed frequently and therefore stand a good chance of becoming deeply entrenched. This does not mean, however, that only concepts of ontologically salient entities can be entrenched. Ontologically salient categories and concepts are likely to be acquired and entrenched first (cf. Schmid 2007: 120), but once grown up, speakers will also frequently process ontologically non-salient concepts. Cognitive linguists, for example, will often process highly abstract concepts, such as SALIENCE and EN- TRENCHMENT , which, as a consequence, will be deeply entrenched despite their ontological non-salience. The next section will discuss how the two types of salience influence the entrenchment and conventionalization of figurative language. Moreover, since both entrenchment and conventionality are intricately connected with the frequency of use of a linguistic structure, it is necessary to clarify the relationship between these notions. In addition, the effect of emergent structure on the conventionality of figurative language needs to be addressed in more detail, since emergent structure can be either fixed or only of a transient nature and this has different consequences for the description of conventionality on the linguistic and conceptual levels and for the conventionality of different types of metaphor and metonymy in general (cf. also chapters 6.3.2 and 7.4.2) 4.2.2.2 Factors influencing the entrenchment and conventionalization of figurative language In what way does salience affect the conventionality of figurative language? The most obvious answer is that linguistic metaphors and metonyms which are frequently processed, i.e. often become cognitively salient, will also be more deeply entrenched than those which are rarely encountered. And the more deeply entrenched they become, the more effortlessly and automatically they will be processed, understood, and used. The relationship between frequency and entrenchment is thus seemingly not a simple one: On the one hand, a deepening of the relevant memory traces, 67 i.e. a deeper entrenchment of a linguistic structure, depends on how frequently one processes it for reasons of comprehension and production. On the other hand, since a deeper entrenchment goes hand in hand with an increasing processing ease, deeply entrenched linguistic structures are also more easily activated and therefore it is likely that they are also used frequently. Frequency can be both a cause of entrenchment and a result of it. While this may be disconcerting for the linguist interested in entrenchment, since for descriptive as well as investigative purposes one would prefer a more clear-cut cause-result relationship between frequency and entrenchment, it is nevertheless intuitively plausible. That frequency causes entrenchment and that entrenchment results in frequency, does not, however, mean that relatively frequent structures become continually more deeply entrenched and cause - in the end - an ever increasing frequency of the items in question. First, as mentioned above, changes in the frequency of already highly frequent structures do not have large effects on their degree of entrenchment. An increasing frequency of such structures does therefore not necessarily result in a further intensification of their entrenchment. Second, the frequency of a linguistic structure is not only determined by its entrenchment, but also by pragmatic factors: A linguistic structure will not be comprehended more often than one encounters it, and, similarly, it will not be activated for language production more often than one has opportunities to use it, i.e. not more often than the situational and linguistic context at hand allow its potential use. This is a simplified account, since it is not excluded that situationally non-relevant linguistic structures are activated by way of spreading activation. In addition, and especially with regard to figurative language, it is important to note that psycholinguistic research has demonstrated that at least under some conditions currently non-intended, but stored meanings of lexemes are activated during lexical access (e.g. bug ‘insect’ vs. ‘overhearing device’; cf. Swinney 1979). 62 Moreover, the speed of activation of a low-frequency form can benefit from the high frequency of a homophonic form (e.g. nun vs. none; cf. Jescheniak and Levelt 1994). As has recently been pointed out by Schmid (forthcoming b), the speed and ease of lexical access, which reflect the degree of entrenchment, are thus possibly not determined by “the word-specific frequency (e.g. of nun as opposed to none) but by the cumulative frequency of all the homophonic forms (i.e. frequency of nun plus frequency of none)”. However, cognitive linguists who apply corpus-linguistic methods to measure the degree of entrenchment of linguistic structures typically disregard this. They focus on relative frequency (Hoffmann 2004: 190) 63 only, i.e. compare the frequency of one 62 Cf. chapter 4.4 for more details. 63 Hofmann (2004) distinguishes two basic types of frequency, conceptual frequency and lexical or textual frequency. The former is extremely difficult to measure. Therefore Schmid (forthcoming b) focuses on the latter. The terms used above, i.e. relative frequency and absolute frequency, are actually sub-types of lexical or textual frequency. 68 linguistic item to the frequency of other, related linguistic phenomena. The other type of frequency, absolute frequency, is ignored. This is a serious shortcoming (Schmid forthcoming b). Ideally, both types of frequency ought to be taken into account. One problem remains, however: Thus far, we do not know how the two types of frequency are related to each other and to what extent the individual types influence the degree of entrenchment. Nevertheless, it is certainly justified to demand, as Schmid (forthcoming b) does, a complementation of corpus-linguistic frequency measures by psycholinguistic experiments. That there is no satisfying answer to the question as to how exactly frequency influences the degree of entrenchment, does not, however, mean that frequency is irrelevant with respect to entrenchment. But to achieve a better understanding of entrenchment, other factors should be considered as well. This also concerns the widespread assumption that the entrenchment of a given structure correlates with its degree of cognitive centrality (Langacker 1987: 60; Lakoff and Turner 1989: 129). The line of reasoning underlying this claim is the following: Frequency results in or reflects entrenchment. Highly frequent structures are therefore also most central in terms of cognition. Two caveats seem to be in order here: First, the relationship between frequency/ entrenchment and cognitive centrality is often taken for granted without any further consideration of the causal relationship. Are linguistic structures frequent because of their cognitive centrality or are they cognitively central because of their frequency? Second, are all cognitively central structures frequent and all the less central structures infrequent or less frequent? To address the second question first: As Gilquin (2006) has demonstrated, what is most frequent in corpora is not necessarily identical with the most salient example, i.e. the prototype of a category. She concludes her comparison of the cognitive model of prototypical causation and frequency data from the BNC by pointing out that those causation structures which are assumed to be prototypical “account for an astonishingly small proportion of the corpus data” (Gilquin 2006: 175). If one accepts that prototypes are cognitively central, this means that cognitive centrality is not always reflected in high frequency, and that not all highly frequent structures are automatically cognitively central. The relationship between frequency and cognitive centrality is not as straightforward as commonly believed. This also answers the first question raised above. If there is no reliable correlation between frequent structures and cognitively central ones, there is also no simple causal relation between frequency and cognitive centrality. Cognitive centrality is not only related to frequency, but also determined by other factors, such as how useful the concept is in terms of cognitive efficiency and pragmatic functions. For metaphors, this would for example concern their typical mapping direction from concrete to abstract and their explanatory function, and for referential metonymies it could, among other things, be related to the stability of the category referred to. 69 An investigation of the cognitive centrality of reasoning patterns therefore presupposes more than a search for frequent linguistic structures. Considering that the conventionality of metaphorical and metonymical mappings is usually regarded as a sign of their cognitive centrality, this supports my initial assumption that proving the conventionality of a metaphorical or metonymic mapping involves more than collecting some possible linguistic manifestations and claiming that all figurative expressions which rely on a conceptual mapping are more conventional than linguistic instances which do not. Even a frequency analysis of the linguistic expressions is not enough. The single linguistic instantiations have to be analyzed in more detail to find out whether their relative frequencies can be linked with generally valid cognitive principles. 64 Particularly suitable candidates for such a more differentiated approach to the conventionality of figurative language - predominantly with regard to metonymies - are the laws of ontological salience. Referring to Langacker’s ideas concerning these laws (cf. chapter 4.2.2.1), Radden and Kövecses (1999: 45-52; cf. also Kövecses and Radden 1998) point out that the choice of the metonymic vehicle is not arbitrary. It is governed by principles based on human experience and perceptual selectivity as well as cultural preferences. Only a few examples can be given here (cf. Table 3): Principle Example HUMAN OVER NON - HUMAN I have a flat tire. ‘car’ Human experience: CONCRETE OVER ABSTRACT having one’s hand on something ‘controlling something’ IMMEDIATE OVER NON - IMMEDIATE She is my joy . ‘reason for joy’ Perceptual selectivity: GOOD GESTALT OVER POOR GESTALT The car needs washing. ‘exterior’ IDEAL OVER NON - IDEAL You are a Judas . ‘betrayer par excellence’ Cultural preferences: IMPORTANT OVER LESS IMPORTANT speak a language ‘know a language’ Table 3: Some examples of cognitive principles governing the choice of metonymic vehicles (selected from Radden and Kövecses 1999: 44- 50) “[O]ur anthropocentric view of the world” (Radden and Kövecses 1999: 45) is responsible for our preference for construals where humans are metonymically foregrounded. This sanctions, for example, the use of a personal pronoun to refer to a ‘car’ in POSSESSOR FOR POSSESSED metonymies, but also 64 Furthermore, one has to take into account that the cognitive centrality of reasoning patterns is not an ever constant factor. Synchronically it is relatively stable, but as diachronic studies have shown (cf. e.g. Kövecses 2006b; Trim 2007), metaphorical conceptualizations can change or evolve though time - due to novel conceptualizations emerging and being spread within the speech community. 70 plays a role for other metonymic mappings, such as CONTROLLER FOR CON- TROLLED or PRODUCER FOR PRODUCT . Humans also display a preference for CONCRETE things as opposed to ABSTRACT concepts. This comes to the fore in the second example in Table 3: A hand is much more concrete than the concept ‘control’ it stands for in this case. Further laws mentioned by Radden and Kövecses (1999: 45-46) under the heading of human experiences are SUBJECTIVE OVER OBJECTIVE , INTERACTIONAL OVER NON - INTERACTIONAL , and FUNCTIONAL OVER NON - FUNCTIONAL . Metonymic expressions which deviate from these laws, such as car for ‘driver’, ought to be dispreferred and thus less conventional than their reversals which obey the principles. Furthermore, the human perceptual apparatus naturally focuses on what is IMMEDIATE rather than on what is NON - IMMEDIATE . This is reflected in the expression She’s my joy, an EMOTION FOR CAUSE OF EMOTION metonymy. Perceptual selectivity also accounts for the fact that GOOD GESTALTS are better equipped to function as metonymic vehicles than POOR GESTALTS . This interacts with our preference for WHOLES OVER PARTS and sanctions, for example, the use of car to refer to the ‘exterior of the car’. Other principles belonging to this class are OCCURRENT OVER NON - OCCURRENT , MORE OVER LESS , DOMINANT OVER LESS DOMINANT , BOUNDED OVER UNBOUNDED , and SPE- CIFIC OVER GENERIC (Radden and Kövecses 1999: 47-48). Cultural preferences are at issue in cases where particular members of a category are favoured. In You are a Judas, the ‘betrayer par excellence’ is chosen to refer to the general category of betrayers. Our inclination to focus on important aspects is responsible for the fact that we can talk about somebody’s ‘command of a language’ in terms of speaking this language. This is probably also related to the general human preference for CONCRETE OVER ABSTRACT . Other cultural principles include STEREOTYPICAL OVER NON - STEREOTYPICAL , TYPICAL OVER NON - TYPICAL , COMMON OVER LESS COMMON , and RARE OVER LESS RARE (Radden and Kövecses 1999: 48-50). These cognitive principles are complemented by communicative principles. The principle of clarity ( CLEAR OVER OBSCURE ) motivates expressions like The dog bit the cat. 65 “This metonymy is so natural that it cannot even be replaced by the expression of the intended target: *The dog’s teeth bit the cat ” (Radden and Kövecses 1999: 50) and is supported by many cognitive principles such as WHOLE OVER PART , VISIBLE OVER NON - VISIBLE and GOOD GES- TALT OVER POOR GESTALT . The principle of relevance 66 deals with the situational and contextual relevance of metonymic vehicles. Special situations and contexts can render otherwise less or non-salient entities more salient (cf. Langacker 1993: 30). 65 This expression would not be considered metonymic by all researchers, such as Croft (2002) and Koch (2001, 2004). 66 This is related to Sperber and Wilson’s (1995) relevance theory. The basic definition of their principle of relevance, “every act of ostensive communication communicates a presumption of its own optimal relevance” (1995: 158), also holds for metonymy (cf. Radden and Kövecses 1999: 50). 71 The cognitive and communicative principles determine the “defaultroutes” (Radden and Kövecses 1999: 52) of metonymic mappings. If a metonymy is in accordance with a large number of the cognitive principles, it has a higher 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 of metonymic mapping “help us understand [...] why certain vehicle-to-target routes have become conventionalized in the language” (Radden and Kövecses 1999: 52). Even though 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 ), Radden and Köveces (1999) regard them as guidelines as to whether a metonym has the potential for becoming conventional. If the notion of ontological salience is extended to cover general human and perceptual as well as cultural preferences, it is helpful for the description of the conventionality of metonymies. But to what extent can it account for the conventionality of metaphors? One of the principles of ontological salience seems to be of special importance here: CONCRETE OVER ABSTRACT . This principle is responsible for the overall preference for metaphors which map concrete domains onto abstract ones. That speakers favour experientially familiar concrete sources to abstract ones has been sufficiently demonstrated (cf. e.g. Lakoff and Johnson 1980, Lakoff and Turner 1989, Kövecses 2002). The HUMAN BODY , for example is a good concrete source domain, which becomes manifest in expressions such as the heart of the problem (cf. Kövecses 2002: 16), while abstract domains like COMMUNI- CATION (cf. Reddy 1979) or TIME mainly form the targets of conceptual metaphors (cf. Kövecses 2002: 15-25). The general laws of cognition and perception thus seem to govern to a large extent which metonymic and metaphorical expressions are acceptable. For metaphors speakers ought to prefer the concrete to abstract direction because it is most useful with regard to the knowledge gained by a mapping. For metonymies, speakers ought to favour mappings which grant them access to the intended targets in an effortless way. Ontological salience should therefore be able to predict which metaphors and metonymies can become entrenched in speakers’ minds and conventional parts of the language. This constitutes an empirically verifiable hypothesis. I will therefore return to the notion of ontological salience in the empirical part (chapters 6 and 7) and investigate whether the speech community indeed displays a marked preference for figurative construals that are in line with the principles introduced above. As will be demonstrated, ontological salience can only account for some more general regularities on the level of conceptual mappings. However, as not all metonymic meanings motivated by the same mapping behave in a uniform way, the laws of ontological salience have to be complemented by another important cognitive principle, i.e. target-in-vehicle salience. Only the latter is able to clarify the con- 72 ventionality of metonymy on the linguistic level and to explain why some metonymic meanings of individual lexemes are highly frequent and conventional although they are not in line with ontological salience. The cognitive notions of entrenchment and salience mirror the structuralist terms of institutionalization and acceptability. Lexicalization finds its cognitive counterpart in what conceptual blending theory calls “emergent structure” (cf. e.g. Fauconnier and Turner 1998, 2002). As discussed in chapter 3.3.2, this theory claims that the interpretation of a metaphor is not restricted to knowledge taken over from the source and aligned to the metaphorical target. Through blending spaces speakers can gain additional knowledge not contained in either of the input spaces. The question is whether metaphors whose interpretation leads to emergent structure are to be considered more or less conventional than others which do not. To answer this question one has to distinguish the conceptual and the (socio)linguistic levels. Consider once more the example My surgeon is a butcher (cf. Grady, Oakley and Coulson 1999; cf. chapter 3.3.2). The emergent structure of the blend has become an inherent part of the concept BUTCHER . This is particularly evident in the conversion to butcher that can not only describe the professional activity of a butcher, but also be used with the meaning ‘to spoil something by working carelessly’ as in That hairdresser really butchered by hair! (LDOCE4, s.v. butcher 2 ). The attribute ‘incompetence’ has become part of the meaning so that the verb can be applied in a wide set of contexts. From a strictly (socio)linguistic perspective, it is a matter of convention that the concept BUTCHER carries the attribute ‘incompetence’. It may be a somewhat idiosyncratic convention, but this is not exceptional. On the conceptual level things are different. Conventional conceptual metaphors are so efficient because they consist of systematic mappings between domains. What is mapped from the source onto the target in specific instances is largely predictable by means of the entrenched conceptual mapping and the invariance principle. If one of the linguistic expressions belonging to a conceptual mapping paradigm acquired additional semantic aspects, this would separate it from the other instances of the same mapping paradigm, 67 which lack this additional emergent structure. Metaphors enriched in this way would therefore have to be considered less conventional as they deviate from the systematicity of the entrenched mappings. Emergent structure can also be observed in metonymies (cf. chapter 3.3.2). That the emergent structure of metonymic construals is often stored, i.e. conventionally fixed, can be illustrated with the help of paragons, 67 Emergent structure thus has different consequences for the conventionality of rich and lean metaphorical mappings. The surgeon is a butcher example above relies on a leaner mapping and the emergent structure characterizing it does not prevent its conventionality. For a more detailed discussion of this difference, cf. chapter 7.4.2. 73 where a category is comprehended “in terms of individual members who present either an ideal or its opposite” (Lakoff 1987: 87): (28) This young man is a real Shakespeare. Barcelona (2004: 364) splits the development of a paragon sense into three different steps: The basis of a paragon is a metonymic, stereotypical model of a person. In (28) Shakespeare is conceived of as a writer with an outstanding literary talent. In a second step, a category of individuals is mentally created that is characterized by one (or some) properties which are vital in the conceptual representation of this stereotypical model. In my example, this would be a category of gifted literary geniuses. This category contains one or more ideal members who can function as paragons. In a third step, the conceptual process of metonymy links the stereotypical representation of S HAKESPEARE AS A LITERARY GENIUS with the category of OUTSTANDING WRITERS . This category has a prototype structure with Shakespeare forming the ‘best example’. A description of the metonym a Picasso used to refer to an adept painter would run along the same lines. But if the expression is used to refer to a picture painted by another artist, a metonymic chain is involved: PRODUCER FOR PRODUCT and PARAGON FOR ENTIRE CATEGORY . First, the PRODUCER FOR PRODUCT metonymy sanctions the use of the artist’s name to refer to his pictures. The target has attributes like ‘impressive’, ‘of a certain style’, ‘expensive’, etc. In the second step, a mental category is created which consists of pictures by other artists which have the same or similar characteristics. And the parallels between the properties of Picasso’s pictures and pictures by other painters form the basis of the actual paragon. In terms of conceptual blending theory, the network would now consist of four inputs: Picasso, his art, other artists, and their art. What matters is that they are linked by virtue of common properties, which are all likely to be connected with the respective mental representations. Picasso plays an ‘important role in art history’, his pictures are also ‘important in art history’ and therefore ‘expensive’. Other painters are also ‘important in art history’ if their pictures are similar to Picasso’s in terms of style, impressiveness, value, etc. Any work of art which can possibly compete with Picasso’s work in some way, is important in art history and expensive or impressive, can therefore be called a Picasso. Figure 7 (below) illustrates this. However, similar to example (28), this only works if the speakers agree on which ideas should be connected with Picasso and his works of art. The interpretation depends on what has been described above (cf. chapter 3.3.2) as the emergent structure of the PRODUCER FOR PRODUCT metonymy in utterances like He bought a Picasso ‘picture by Picasso’. If the speech community did not agree that a Picasso is a picture which is valuable and important in art history, i.e. if the emergent structure was not fixed and conventional, it would not be possible to use the expression as a paragon. 74 Input 1 Input 2 Input 3 Input 4 P ABLO P ICASSO P ICASSO ’ S ART OTHER PAINTERS ’ ART OTHER PAINTERS - P as artist - P’s conception of art - P’s techniques - P’s role in art history etc. all of P’s works of art H knows - H’s attitudes towards P’s art value of P’s art etc. the work of art referred to other works of art well-known painters (Braques, Dalí, etc.) unknown painters their styles, techniques etc. Blend A P ICASSO any impressive work of art of similar style expensive important in art history Figure 7: A Picasso ‘any impressive, expensive work of art’ (P = Picasso; H = hearer) Let me briefly summarize the discussion so far. There are various factors which play a role in the conventionality of metaphors and metonymies. The most obvious and maybe also most important one is their frequency of occurrence. Although the frequency of a linguistic structure is not necessarily an indicator of its cognitive centrality, it is certainly significant with regard to conventionality, because it reflects the speech community’s preferences for some linguistic structures and meanings. To what extent conceptual mappings are accepted by speakers hinges upon the direction the mappings take, which is related to the laws of ontological salience: For metaphors, mappings that obey the concrete-to-abstract direction ought to be preferred, and for metonymies those which use an ontologically more salient vehicle to refer to an ontologically less salient target. This goes hand in hand with the fact that for both metaphors and metonymies, a conventionalization ought to be more likely when the source domain or vehicle concept (for metonymies) is filled with well-structured knowledge. In addition, linguistic metonyms ought to be more ‘successful’ in the speech community as a whole, if they refer to a target which forms at least a relatively stable category. Emergent structure is also relevant with regard to the conventionality of figurative language and thought. The outcome of reasoning processes can become stored knowledge for both metaphor and me- 75 tonymy. However, while emergent structure does not necessarily create much distance between a linguistic metonym enriched in this way and the metonymic mapping it is based on, 68 this seems to be different for certain kinds of metaphor. In the following sections, I will review typologies of metaphors and metonymies that address their conventionality in order to, first, establish more support for the assumptions presented thus far, and second, to possibly identify more factors that contribute to the conventionality of metaphors and metonymies. In addition, I will deal with the relationship between conventionality and the ‘death’ of figurative language and thought. 4.3 Typologies of conventional and ‘dead’ metaphors and metonymies Since conceptual metaphors and metonymies have become prominent in research, many typologies have been proposed to explain the phenomena. Often, they focus on the levels of generality of the mappings involved, like, for example, Grady’s (1997, 1999) distinction of primary and compound metaphors or the classification as genericand specific-level metaphors (cf. e.g. Kövecses 2002: 38-39; chapter 3.2.1). Others, like Lakoff and Johnson’s (1980) typology consisting of ontological, orientational and structural metaphors, rely more on functional aspects. The same can be observed for metonymies: Sorted according to different degrees of generality, we find on the topmost level WHOLE FOR PART (e.g. America ‘United States’), PART FOR WHOLE (e.g. face ‘person’) and PART FOR PART (e.g. BMW for ‘car’, where both PRODUCER and PRODUCT are part of the larger PRODUCTION ICM) mappings. These types can be realized by more specific mappings on a lower level (e.g. WHOLE FOR PART can lead to mappings like OBJECT FOR MATERIAL as in There was cat all over the road or CATEGORY FOR ITS MEMBER as in pill for ‘contraceptive pill’; Radden and Kövecses 1999). On the other hand, as already discussed in chapter 3.2.2.1, metonymies can be differentiated according to their functions, as can be exemplified by the difference between referential and illocutionary metonymies. In what follows, I will concentrate on typologies relevant to a description of the conventionality of metaphors and metonymies. As in the previous chapters, the focus of attention will be devoted to metonymy. 68 Cf. chapter 6.3.2 for a more detailed description of the influence of emergent structure on the conventionality of metonyms. 76 4.3.1 Typologies of metaphor 4.3.1.1 Creative, conventional, dead, and literal metaphors Conceptual metaphor theory mainly discusses unobtrusive examples of the kind He’s wasting his time, He attacked my position, etc. - called conventional metaphors. These are contrasted to dead and poetic ones. According to Lakoff and Johnson (1980: 55) foot of a mountain or leg of a table are, for example, ‘dead metaphors’. They are of marginal relevance as they are idiosyncratic: There are not many systematic correspondences in a metaphor like a MOUNTAIN IS A PERSON . 69 In trying to draw a boundary between conventional and poetic or creative metaphors, Lakoff and Johnson (1980) sometimes overshoot the mark. Everyday metaphors are equated with literal language: “The language of argument is not poetic, fanciful, or rhetorical; it is literal” (Lakoff and Johnson 1980: 5). Conventional metaphors are even called “literal metaphors” (Lakoff and Johnson 1980: 53). This is certainly not only a terminological problem. Literality is not the same as conventionality. But what, then, makes a metaphor a conventional metaphor for Lakoff and Johnson (1980)? Conventional orientational, ontological and structural metaphors, we read, are experientially grounded. Structural metaphors can, in addition, rely on similarities which arise from ontological or orientational metaphors (Lakoff and Johnson 1980: 151). T IME IS A VALUABLE RESOURCE is thus conventional due to our experiences in a modern, industrialized world and IDEAS ARE FOOD is conventional since it relies on two other, more basic metaphors, i.e. IDEAS ARE OBJECTS and the MIND IS A CONTAINER . New metaphors are mostly structural and work just like conventional structural ones (Lakoff and Johnson 1980: 152): They can be based on orientational or ontological metaphors and consist of partial mappings. This suggests that the more basic and embodied a metaphorical conceptualization, the more likely is its conventionality. Orientational and ontological metaphors, like HEALTHY IS UP (e.g. He rose from the dead) and AN ACTIVITY IS A CONTAINER (e.g. I put a lot of energy into washing the windows), are therefore probably all highly conventional, while structural metaphors, which often build on the former and elaborate them in different ways (e.g. THE MIND IS A MACHINE which relies on THE MIND IS AN ENTITY ), are more likely to vary regarding their conventionality. But this does not answer the question of where exactly we can locate the difference between new and conventional metaphors. Why are some metaphors more common than others? Lakoff and Johnson (1980) do not give a satisfying answer. They merely list three types of imaginative metaphors 70 which “fall outside the domain of normal literal language” (1980: 53): 69 This is another example of a lean mapping according to Ungerer and Schmid’s (2006: 126) definition. Cf. above chapter 3.2.1. 70 Cf. also Lakoff and Turner (1989: 67-72), who discuss the ways in which poetic metaphor makes use of the ordinary metaphoric systems. 77 1. Extensions of the used part of a metaphor: e.g. These facts are the brick and mortar of my theory. 2. Instances of the unused part of the literal metaphor: e.g. His theory has thousands of little rooms and long, winding corridors. 3. Instances of novel metaphor: e.g. Classical theories are patriarchs who father many children, most of whom fight incessantly. The first type is less conventional than other instances of the THEORIES ARE BUILDINGS metaphor because it refers to the material of the outer shell of the building. While the outer shell is a commonly used part of the mapping, the material is not. The difference to the second type is that an utterance like His theory has thousands of little rooms ... metaphorically conceptualizes the aspects of a theory as the rooms of a building. Rooms, however, do not belong to those parts of the domain of BUILDINGS which are usually mapped onto THEORIES . And the third type offers an entirely novel way of thinking about THEORIES . The continuum of conventionality suggested by this is summarized in Figure 8. Novel metaphors Modified metaphors Conventional metaphors low high Degree of conventionality Figure 8: The conventionality of metaphors as a continuum (based on Lakoff and Johnson 1980: 53) The first and second type, subsumed under the heading ‘modified metaphors’ in the figure, rely on conventional mappings, but modify them to different extents, while the third type offers a novel conceptualization of THEORIES that we have probably never come across before. This continuum is rather crude and does not offer many details. The main factor that distinguishes the single types seems to be the frequency of usage, even if this is not explicitly acknowledged. Conventional metaphors differ from modified metaphors mainly in their frequency, i.e. modified metaphors are less often used and encountered than conventional ones. And novel metaphors occur even less frequently than modified ones. The reason for these differences lies in the cognitive function of the respective metaphors. A conceptualization of THEORIES as PATRIARCHS offers less insights than a conceptualization in terms of BUILDINGS . The concept PATRIARCH cannot lend as much structure to THEORIES as the BUILDING domain, which comes with a large amount of structured knowledge: Buildings have an outer shell and an interior, they need a foundation, they can be re-built, changed and renovated, etc. The relationship between conceptual mappings and linguistic expressions remains unclear, however, in Lakoff and Johnson’s (1980) typology. Is 78 the modification of THEORIES ARE BUILDINGS in These facts are the bricks and mortar of my theory only a matter of language or does it affect the conceptual mapping? Lakoff and Turner (1989: 50) rightly stress that it is a prerequisite of metaphor research to distinguish between the levels of language and thought and that the conventionality of metaphor can only be adequately described when both levels are taken into account: At the conceptual level, a metaphor is conventional to the extent that it is automatic, effortless, and generally established as a mode of thought among members of a linguistic community. For example, DEATH IS A DEPARTURE is deeply conventionalized at the conceptual level; we probably all have it. [...] Conventionalization also applies to the connection between the conceptual and linguistic levels. When [...] we speak of the degree to which a conceptual metaphor is conventionalized in the language, we mean the extent to which it underlies a range of everyday linguistic expressions. For example, DEATH IS A DEPAR- TURE is not just conventionalized as a way of conceiving of death; it is also widely conventionalized in language, underlying a wide range of expressions such as “pass away,” be “no longer with us,” “gone,” “among the dear departed,” and so on. (Lakoff and Turner 1989: 55-56) The connection of the two levels is evident in the fact that on the linguistic level conventionality correlates with how many different linguistic expressions rely on the same mapping. Separating the two levels of language and cognition for descriptive purposes, one can say that while novel metaphorical expressions are unconventional on the conceptual and the linguistic levels, both conventional and modified metaphorical linguistic expressions are largely conventional on the conceptual level, but modified metaphors are less conventional on the linguistic side. The difference between the types is, again, a matter of degree. Extending already existing correspondences, for example, leads to less ‘unconventionality’ in the conceptual representation than establishing novel correspondences. In short, idiosyncrasy of language may or may not express idiosyncrasy of thought, but idiosyncratic thought requires idiosyncratic language. (Lakoff and Turner 1989: 50) The continuum of the conventionality of metaphors mentioned above can therefore be modified to accommodate both the levels of thought and of language as in Figure 9. Linguistic level: idiosyncratic modified conventional Conventionality: low high Conceptual level: idiosyncratic conventional Figure 9: Conventionality on two levels 79 On both levels the blurred boundaries between idiosyncratic and conventional cases are largely a matter of frequency. The cognitively most entrenched metaphors thus ought to be those which have the highest possible degrees of conventionality on both levels, because they are the most automatic and unconscious ones. From a cognitive perspective, they are the ones that are “most alive” (Lakoff and Turner 1989: 127; original emphasis). More recently, Müller (2008) has proposed a typology of metaphors which also explicitly considers the two different levels. On the level of the language system metaphors can either be historical, entrenched or novel. This is distinguished from the conceptual level, which is - in her account - largely a private, subjective matter and associated with language use. It describes the degree of activation of metaphoricity when a linguistic metaphor is encountered. On this more dynamic level, single instances find their place on a gradable scale reaching from sleeping to waking metaphors (Müller 2008: chap. 6). She thus focuses very much on the potential differences between language as a collective system and its individual use by single speakers (Müller 2008: 195) - a highly interesting aspect, which is, however, not of prime concern for a study of the general conceptual preconditions and consequences of the conventionality of figurative language, which necessarily concentrates on the supraindividual level. 4.3.1.2 The ‘death’ of metaphors and semantic change In the conceptual theory, the focus of research has shifted towards what was often referred to earlier as ‘dead’ metaphors. Earlier categorizations of metaphors on scales reaching from ‘living’ and ‘active’ metaphors to ‘dead’ ones were mainly needed to separate the ones worth studying from those which were not. The typologies proposed are sometimes quite elaborate, such as Fowler’s (1983: 359), which has six different categories: stone-dead, dead, three-quarters dead, half-dead, dormant and live. Black (1977: 439- 440) proposes a categorization into “extinct, dormant, and active metaphors”, but for him only the latter cases are of interest: “a so-called ‘dead metaphor’ is not a metaphor at all” (1977: 439). 71 Searle (1979: 98) takes a slightly different position. He emphasizes that, for example, cold ‘lack of emotion’ is by now a dead metaphor, because the metaphorical sense is listed in dictionaries like the OED. But for him, dead metaphors are worth considering from a semantic perspective, as they play a role for the lexicon. 72 Even earlier, Richards (1936) pointed out that the 71 Interestingly, also some more recent, cognitive approaches, such as Croft and Cruse (2004: 206) contend that one should not so much focus on “cases such as branch, flourish or even reap” (as instances of the more general SOCIAL ORGANIZATIONS ARE PLANTS metaphor, cf. Kövecses 2002: 8). This view is most likely a consequence of their primary research interest, i.e. the interpretive process, reflected in their chapter on “Blending Theory and novel metaphors” (2004: 207-209). 72 Cf. chapter 3.1. 80 exclusion of dead metaphors results in an unnecessary and counterproductive restriction of the scope of metaphor studies. This favourite old distinction between dead and living metaphors [...] is, indeed, a device which is very often a hindrance to the play of sagacity and discernment throughout the subject. For serious purposes it needs a drastic re-examination. (Richards 1936: 102) Conceptual metaphor theory has also instigated research into semantic change. 73 The idea that metaphors follow systematic mapping patterns served as a starting point for a search for parallels between synchronic mappings and general trends in semantic change. Traugott (1985b), for example, studies SPACE as a source concept for more abstract targets. Many expressions with temporal meanings started out with spatial meanings. Before and after, for example, originally meant ‘in front of’ and ‘in back of’. In the same vein, many mental verbs have developed out of spatial verbs (e.g. the root of deduce is Lat ducere ‘lead’, and suppose comes from Lat sub + ponere ‘put under’; Traugott 1985b: 160-163). Examples like these have certainly reached a degree of conventionality that goes beyond what conceptual metaphor theorists usually have in mind when referring to a metaphor as conventional 74 and therefore lie outside the scope of my study, but they corroborate the idea that the concrete-to-abstract direction of mapping is a crucial factor for conventionality, since total semantic change could be said to form the end point of the process of conventionalization. The semantic course of words appears to be largely constrained by regularities which are based on general facts about human cognition. Spatial experiences form a significant part of human experience and are thus often used as sources of metaphorical construals. Similarly, all senses, especially sight, seem to be quite frequent source domains for diachronic semantic changes (Traugott 1985b: 169). Given that data gained by synchronic and historical studies converge (cf. Sweetser 1990: 23), cognition seems to be the place where diachronic and synchronic approaches to language meet. The criteria proposed to distinguish metaphors with regard to their degrees of conventionality are often not very clear or useful. Searle’s (1979; cf. also Jäkel 1997: 55) idea of using the existence of dictionary entries of figurative meanings as a criterion is problematic, as dictionaries will differ with regard to the meanings they record (cf. Fass 1997: 49-50). Other approaches are more promising. Traugott (1985a: 22-23), for example, treats the conventionality of metaphors as a function of three parameters: reference, 73 This is evident in volumes such as Blank and Koch (1999) or Fisiak (1985). A particular interest in diachronic issues in the study of cognition can be observed in many of Geeraerts’s publications, cf. e.g. Geeraerts (1983, 1985) or Geeraerts and Grondelaers (1995). 74 Cf. Richards (1936: 101), who maintains that metaphors are never entirely “stone dead”. They can always be ‘revived’ using quite simple methods. A little etymological knowledge suffices to see the original metaphorical motivation behind words like dis-cover or dis-close (Traugott 1985a: 21). 81 conceptualization, and distance. 75 Reference has to do with the subjectively felt figurativity of an utterance and distinguishes cases like Mary is a block of ice from less obviously metaphorical utterances like Mary left before Jane. Conceptualization measures the difference between the two domains linked by a metaphor. A re-categorization of a human being as a nonanimate object (e.g. She is a crumpled sheet of print-out) causes more conceptual changes than a metaphor like She is a worm. The parameter of distance deals with “the degree of naturalness within a domain” (Traugott 1985a: 23). A metaphor is considered more conventional if the aspects which are picked out from the source domain to be applied to the target are prominent within the source. Compare the following examples: (29) father of the eye vs. uncle of the eye (both: ‘centre of the eye’) (30) She is a battle-axe vs. She is a crumpled sheet of print-out (Traugott 1985a: 23) The reason why the first examples in (29) and (30) are less strikingly figurative than the second ones is that the sources of the former are conceptually more central and informative than those of the latter. Within the field of kinship terms, the concept FATHER takes up a more central position than UNCLE . If this wider field is chosen as a metaphorical source it is natural to pick out one of its central members. Example (30) is based on functional aspects. A battle-axe can be associated with a clear-cut function, but what about a crumpled sheet of print-out? While the former leads to an activation of a rich and neatly structured concept, the latter concept has a lean structure with no particularly important parts which could serve as a basis for the metaphor. Traugott’s last criterion is intuitively convincing and also corresponds to what has been pointed out above with regard to the acceptability and institutionalization of metaphors and metonymies. 76 The other two are more difficult to apply: The first one because it relies on subjective impressions and the second because it stands in contrast to the claims of conceptual metaphor theory. Traugott’s example implies that metaphorical mappings are more likely to be accepted if the domains involved are not too different. However, conventional conceptual metaphors usually link an abstract and a concrete domain, which implies a considerable conceptual distance. Which further criteria can be used to describe the conventionality of metaphorical expressions? One frequently mentioned point is that a metaphor is dead when its literal meaning has died out completely (cf. e.g. Bowdle and Gentner 2005: 208-210; cf. also Gentner and Bowdle 2001: 229- 230; Lakoff and Turner 1989: 129; Leech 1981: 227). But even the ‘death’ of a 75 Traugott’s parameters have also been used in psycholinguistic studies of metaphorical linguistic expressions. However, psycholinguists use different terms to refer to the same phenomena such as metaphoricity or semantic relatedness (cf. e.g. Katz et al. 1988). 76 Cf. chapter 4.2.1. 82 metaphor is not as straightforward as this suggests. Bowdle and Gentner (2005: 209) point out that there are two types of dead metaphors. The difference between what they call dead 1 and dead 2 metaphors can be illustrated with the following examples: (31) A university is a culture of knowledge. (32) The movie Star Wars was a blockbuster. (Bowdle and Gentner 2005: 209) As already mentioned in chapter 4.2.1, the most familiar everyday sense of culture is ‘the beliefs, way of life, art and customs that are shared and accepted by people in a particular society’ (LDOCE4, s.v. culture 1 ), and this is also realized in (31). From an etymological perspective, however, this sense of culture is a metaphor based on another sense of the word, which is still in use: “’a preparation for growth’” (Bowdle and Gentner 2005: 209). 77 Reconsidered in the light of this older sense, the statement in (31) suggests that a university is a place where knowledge finds ideal conditions for growth. The point is that this connection between the two senses of culture is not necessarily seen and the older sense therefore does not automatically feed into our understanding of the example. Due to its frequent use in the domain of human society and traditions, the concept CULTURE has acquired properties belonging to this latter domain (Bowdle and Gentner 2005: 209). Target domain-specific knowledge has become part of the representation of the source domain; in other words, the source has acquired an emergent structure. And as the new knowledge in the source domain is different from the original, ‘literal’ source domain knowledge, two, more distant source domain representations can result. However, one can doubt whether the two concepts are really entirely separate in speakers’ minds, since it has been pointed out that fixed meanings might not exist at all (cf. e.g. Searle 1978; Shinjo and Myers 1987: 238), which calls the whole idea of a separation of different meanings into question. 78 Dead 2 metaphors are marked by the complete absence of the literal meaning. The word blockbuster in (32) has only one meaning today, namely 77 Bowdle and Gentner’s (2005) dead 1 metaphors correspond to what Leech (1981: 227) describes as the psychological divergence of the literal and the transferred meaning and what Grimm (1991: 92) calls the loss of the derivational relation. Leech’s (1981: 227) example is crane (cf. also Leisi 1985: 184), what the Lakovians would call an image-metaphor (cf. e.g. Kövecses 2002: 50-51; Lakoff and Turner 1989: 89-96). For further interesting examples of image-metaphors, cf. also Lipka (1996). 78 It is usually assumed that the human mind makes use of maximally informative structures, Lakoff (1987), for example, calls this radial categories, but the idea goes back to Rosch’s studies of the internal structure of categories (cf. e.g. Rosch 1973; Rosch and Mervis 1975; a useful survey is given in Ungerer and Schmid 2006: chap. 1). Different meanings of one lexeme form chains or networks in the mind (cf. chapter 4.1.2) and as long as the new information in the source domain which is a consequence of the metaphorical mapping is not incompatible with the original source domain, there is no reason to believe that speakers would separate the new semantic aspect entirely from what they have already stored. This is supported by the fact that speakers see semantic relations even between real homonyms (cf. e.g. Ullmann 1962: 104). 83 ‘a book or film that is very good or successful’ (LDOCE4, s.v. blockbuster). The original, literal meaning (‘a huge aerial bomb’, NODE, s.v. blockbuster) has died out. Conventional metaphors are, according to Bowdle and Gentner (2005: 209), those where both the literal and the metaphorical meaning of the source term (what they call the base term), co-exist in the language. Both have been encountered before and are stored, but the literal meaning, they argue, is still activated during comprehension. 79 The main difference between conventional and novel metaphors lies in the fact that speakers do not possess any stored metaphorical concept or category for the latter. The two examples quoted above from Bowdle and Gentner (2005) do not rely on rich systematic conceptual mappings, but on leaner mappings in Ungerer and Schmid’s (2006: 126-127) sense. They are reminiscent of the metaphors researched within structuralist semantics under the heading of lexicalized metaphor. But if one complements Bowdle and Gentner’s (2005) approach with a consideration of systematic mappings, a useful picture of the difference between the conventionality and the ‘death’ of metaphors emerges: The meanings of conventional metaphorical expressions are stored in speakers’ minds, but still supported by the entrenched mappings motivating them and thereby linked to other linguistic metaphors relying on the same mapping. But as long as the mapping has not died out, there is no reason to believe that the metaphor has lost its cognitive function. Only when the mapping and the literal concept cease to exist, is a metaphor virtually dead. So far, I have only discussed to what degree extant typologies of metaphors can throw light on their conventionality. Mostly, the discussion has supported the criteria of conventionality already proposed in the previous chapters. The vital factors are frequency and that the source concept contains well-structured knowledge. The latter is related to experiential grounding. More or less direct experiences will more likely than not lead to well-structured stored knowledge in humans’ minds, which can then readily serve as the source of metaphorical thinking and speaking. For the most part, the mappings resulting from such experiences belong to the generic level and form the basis for many different specific metaphors. Thus they are prime candidates to fulfil the criterion of conventionality mentioned by Lakoff and Turner (1989: 55-56): They necessarily motivate many everyday expressions. In the next sections, I will summarize what typologies of metonymy contribute to our understanding of conventionality and which defining criteria are proposed by them. A short section will also be devoted to the 79 Bowdle and Gentner link the different stages of the “career of metaphor” (2005: 209) to different processing strategies. While novel ones are interpreted as comparisons, conventional metaphors can either be understood by way of categorization or comparison. Both types of dead metaphors are necessarily comprehended as categorization statements. Cf. also Shen (1992), who deals with metaphor comprehension in terms of the formation of ad-hoc categories (cf. Barsalou 1983). 84 ‘death’ of metonymies - an issue that has rarely been addressed even though metonymy is, alongside metaphor, one of the major reasons for polysemy and semantic change (cf. e.g. Bartsch 2002; Bredin 1984, Geeraerts 1997, Koch 2001, 2004, Stern 1931, Warren 1992). 80 4.3.2 Typologies of metonymy Traditionally, typologies of metonymy have consisted of extensive lists of metonymy-producing relationships (cf. chapter 3.1). More recent approaches have tried to reduce the types of metonymy by subsuming various metonymic patterns under a few more general headings (cf. Blank 1999: 178). Some of the typologies focus on functional criteria (e.g. Nunberg 1995), others emphasize the difference between synchrony and diachrony (Blank 1999). Further typologies take the notion of domains as their starting point (e.g. Ruiz de Mendoza 2000; Ruiz de Mendoza and Díez Velasco 2002, 2004) or consider metonymy as a prototypical notion (e.g. Barcelona 2003; Koch 2001; Peirsman and Geeraerts 2006; Schmid 1993). 4.3.2.1 Functional criteria and the synchronic-diachronic distinction Nunberg’s (1995) classification is based on functional criteria. He describes metonymy as a kind of meaning transfer (1995: 115). Transfers of meaning are defined as “the ensemble of productive linguistic processes that enable us to use the same expression to refer to what are intuitively distinct sorts of categories of things”. His first class is deferred indexical reference, “a process that allows a demonstrative or indexical to refer to an object that correspond [sic! ] in a certain way to the contextual element picked out by a demonstration” (1995: 111). An example is (33) in a context where a customer hands his keys to an attendant at a parking lot. This class stands in opposition to what Nunberg (1995: 111) calls predicate transfer (cf. 34). (33) This is parked out back. (34) I am parked out back. (Nunberg 1995: 110) The difference between (33) and (34) lies in the relation between the metonymic expression and the intended target. In (33), the subject of the sentence refers directly to the target, i.e. the car. In (34), the subject of the sentence, the metonymic expression I, refers to the metonymic trigger (Alač and Coulson 2004: 25). 81 80 This impression is confirmed by a simple internet search. A Google search for the term dead metaphor on English-language pages returns significantly more results than a search for the term dead metonymy. On 11 December 2009, the former term was found 17,000 times, while the latter returned only 710 hits. 81 This difference is also responsible for discrepancies in terms of conjoining predicates (Nunberg 1995: 110-111). Deferred indexical reference can only take additional predicates that also refer to the car (e.g. This is parked out back and may not start). A conjoined reference to the key is hardly possible (e.g. ? This fits only the left front door and is 85 Predicate transfer is the foundation of systematic or regular polysemy (cf. Apresjan 1974) and thus comprises all classic metonymic patterns, such as PLACE FOR INHABITANTS or PAINTERS FOR WORKS . It relies on two conditions (Nunberg 1995: 112-115): There has to be some kind of correspondence between the metonymic expression and the target (as in (34) between car and owner) and “the property contributed by the new predicate [has to be] ‘noteworthy’” (1995: 114). The noteworthiness can either be of a more general nature as in utterances like I’m in the Whitney or only be relative to the immediate purposes of the talk exchange as in (34). The latter is also true for Nunberg’s (1995: 116) third major class, the occurrent metonymies, which can be exemplified with the well-known example The ham sandwich is at Table 7. Occurrent metonymies do not lead to polysemy as the relationship between trigger and target, in this case between customer and food, is not stable enough to allow a successful reference in neutral contexts. They are always highly context-dependent. In this way, Nunberg’s (1995) approach is also helpful for a consideration of the conventionalization of metonymies: Occurrent metonymies cannot become part of the lexicon and cases of deferred indexical reference are also unlikely candidates. Only instances of predicate transfer can gain wider currency, provided that they rely on a noteworthy and generally valid property. Blank’s (1999) typology of metonymy builds on the distinction of synchrony and diachrony, but uses the terms in a novel way. His two major types are co-present and successive relations. Both are meant as superordinate domains of contiguity under which different types of metonymyproducing relationships can be subsumed. Co-present relations hold, for example, between BODY PARTS and PERSONS or INVENTORS and their PRO- DUCTS (e.g. Zeppelin). They are said to be simultaneously present, i.e. stand in a synchronic relationship. Successive relations are marked by their processual character (e.g. ACTIVITY FOR RESULT : Fr bâtiment ‘act of building’ > ‘building’, MATERIAL FOR PRODUCT : e.g. horn ‘horn of an animal’ > ‘instrument’). This classification has two weaknesses. On the one hand, it is not always clear where the boundary between synchrony and diachrony is drawn. It is possible to claim that INVENTORS stand in a synchronic relationship to their INVENTIONS , but then one could also argue that the MATERIAL is still present in the PRODUCT . While some cases such as Fr bâtiment can only be categorized in one way, others like Zeppelin could also be assigned to the opposite category. The second point is more serious: It remains unclear what is gained by the distinction between co-presence and succession, even if both are “fundamental models of human conceptualization” (Blank 1999: 179). Blank does not discuss any potential consequences or functions of this parked out back). With predicate transfer only predicates that describe the trigger can be conjoined (e.g. I am parked out back and have been waiting for 15 minutes vs. *I am parked out back and may not start). 86 distinction. And indeed, when it comes to his central interest, the question of which metonymies are most likely to be frequently used and therefore lead to semantic change, he (1999: 183) refers to the level of metonymic patterns but makes no reference to the difference between co-presence and succession. 4.3.2.2 Types of metonymy and the notion of domains As pointed out in chapter 3.2.2.1, the notion of domains is, on the one hand, a vital definitional criterion for metonymy, but on the other, hard to pin down and apply. Despite its problematic nature, Ruiz de Mendoza’s (e.g. Ruiz de Mendoza 2000; Ruiz de Mendoza and Díez Velasco 2002, 2004; Ruiz de Mendoza and Pérez Hernández 2003) typology of metonymy is based on this very notion. Most researchers (e.g. Leite 1994; Radden and Kövecses 1999) argue that metonymic mappings can be of three principal types: WHOLE FOR PART , PART FOR WHOLE , and PART FOR PART . Ruiz de Mendoza holds, however, that the third type does not exist. He only distinguishes two basic types of metonymy: source-in-target and target-in-source. In target-in-source metonymies, a whole matrix domain 82 stands for a sub-domain. Example (35) can be considered an instance of this pattern. N APOLEON ’ S ARMY is conceived of as a sub-domain of the knowledge about the overall domain N APOLEON . This type involves domain reduction by highlighting only the relevant part of a larger knowledge structure. The other type, source-in-target metonymy, 83 involves domain expansion. Here, a sub-domain stands for the matrix domain. This is the case in (36). The two types can be illustrated diagrammatically as in Figure 10: (35) Napoleon lost at Waterloo. (36) The sax has the flu today. N APOLEON N APOLEON ’ S ARMY THE SAX PLAYER THE SAX Figure 10: Target-in-source and source-in-target metonymies (based on Ruiz de Mendoza and Díez Velasco 2004: 298-299) 82 Ruiz de Mendoza’s notion of matrix domain (i.e. “the most encompassing of two domains involved in a mapping”; Ruiz de Mendoza and Díez Velasco 2004: 299-300) differs from what Langacker (1987) calls domain matrix. Cf. Ruiz de Mendoza and Díez Velasco (2002: 496, Fn. 8). 83 An interesting observation is that only source-in-target metonymies can be used predicatively (Ruiz de Mendoza and Díez Velasco 2004: 298). In John is a brilliant brain, the source is also a sub-domain of the more encompassing matrix domain J OHN . 87 One advantage of the model is that it accounts for double or chained metonymies (cf. Ruiz de Mendoza and Pérez Hernández 2001, Hilpert forthcoming) in an elegant manner. The utterance Plato is on the top shelf is not a simple instance of AUTHOR FOR WORK , but in fact involves two different mappings: AUTHOR FOR WORK and WORK FOR MEDIUM IN WHICH IT IS PRESENTED . This can be summarized as AUTHOR FOR WORK FOR MEDIUM . Figure 11: The AUTHOR FOR WORK FOR MEDIUM metonymy (adapted from Ruiz de Mendoza and Díez Velasco 2004: 311) As Figure 11 shows, P LATO ’ S WORK is a sub-domain of both the P LATO domain and the MEDIUM domain. While one can certainly regard the MEDIUM domain as larger than the domain of P LATO ’ S WORK , it is doubtful whether the actual target of the double metonymy is really the MEDIUM as such, since the speaker is not simply referring to books, but to books by a specific author. The medium seems to be of secondary importance and could also be considered part of the domain of P LATO ’ S WORK , which is necessarily presented in one medium or another. The distinction of target-in-source and source-in-target metonymies has some bearing on the use of anaphora. What is at issue is whether anaphoric reference is possible to the literal meaning of the metonymically used expression or whether potential anaphora are restricted to the metonymic meaning. Fauconnier (1994: 5-6) gives the following examples to illustrate this: (37) Plato is on the top shelf. It is bound in leather. (38) Plato is on the top shelf. You’ll find he is a very interesting author. (39) The mushroom omelet left without paying the bill. He jumped into a taxi. (40) *The mushroom omelet left without paying. It was inedible. The AUTHOR FOR WORKS metonymy in (37) and (38) allows both types of anaphora: those referring to the metonymic construal, i.e. the ‘works’, and those referring to the literal meaning, i.e. the ‘author’. For the less common metonymy FOOD ORDERED FOR CUSTOMER in (39) and (40), things are different. It is only possible to use anaphora that are conceptually linked to the target, while anaphora that refer to the source result in unacceptable utter- P LATO P LATO ’ S WORK MEDIUM OF PRESENTATION 88 ances. Once the metonymic construal has been established by the first sentence, speakers cannot simply switch back to the literal meaning. Ruiz de Mendoza and Díez Velasco (2002, 2004) explain anaphoric reference to metonymic noun phrases with their Domain Availability Principle: “whenever a metonymic noun phrase occurs in a sentence, only the domain matrix is available for anaphoric reference” (Ruiz de Mendoza and Díez Velasco 2002: 500; cf. also Ruiz de Mendoza and Pérez 2001). A UTHOR FOR WORKS is a target-in-source metonymy, 84 while FOOD FOR CUSTOMER is a source-in-target metonymy. This accounts for the difference between (39) and (40): Only the CUSTOMER is available for anaphoric reference, since he is the matrix domain. But why is it possible to use different anaphoric pronouns in the case of the AUTHOR FOR WORKS metonymy? The reason is that it actually is a double metonymy. Therefore, the language user can, in principle, choose one of the two matrix domains for anaphoric reference. This could then lead to utterances like the following: (41) *Plato is on the top shelf. He is bound in leather. (Ruiz de Mendoza and Díez Velasco 2004: 307) However, the choice of anaphoric pronouns is further constrained by the Domain Combinability Principle: [W]henever two domains are available for anaphoric reference to a metonymic noun, we tend to select the domain that is semantically more compatible with the predicate of the sentence containing the anaphoric pronoun. (Ruiz de Mendoza and Díez Velasco 2004: 311) That the property ‘bound in leather’ belongs to the realm of books rather than to the realm of authors excludes cases like (41) and is the reason why (37) is preferred. In the same vein, (38) is possible because the predication refers to the author and not to the works, while the use of anaphoric it is excluded in this case. Papafragou (1996: 193) implies a straightforward relationship between the range of possible anaphora and the conventionality of metonymy. She claims that those metonymies which exhibit restrictions as regards their possible anaphora are less conventional than those which do not. This account seems somewhat simplistic, however, since the nature of the predication has to be taken into account as well as the distinction of simple and double metonymies. However, Ruiz de Mendoza’s work once again supports the assumption that the conventionality of metonymies is linked to the direction of mapping. In Ruiz de Mendoza’s terminology, it is directly related to the 84 In Ruiz de Mendoza and Díez Velasco (2004: 310), the AUTHOR FOR WORKS mapping is described as source-in-target; in Ruiz de Mendoza and Díez Velasco (2002: 516) the same metonymy is quoted as an example of target-in-source metonymies. As the AUTHOR is the source and the WORKS form the target of this metonymy, and because the AUTHOR forms the matrix domain in this case, the description given in their 2004 article is incorrect. 89 distinction of matrix vs. sub-domains. Matrix domains are relatively welldefined, they have to be “very clear delineated concept[s]” (Ruiz de Mendoza and Díez Velasco 2004: 300). The sub-domains are less delineated. It should thus be more natural to use mappings with well-defined concepts as vehicles and less delineated sub-domains as targets. In other words, the speech community ought to prefer WHOLE FOR PART mappings to PART FOR WHOLE mappings. And indeed, the examples of the two different directions of mapping given above are intuitively felt to differ with respect to their conventionality. The source-in-target metonymy mushroom omelet for ‘customer’ is undoubtedly used less frequently than Plato meaning ‘Plato’s work’. While the assumption that WHOLE FOR PART and PART FOR WHOLE mappings differ with regard to their conventionality seems plausible and can be empirically examined, it is not as unproblematic as it seems at first glance, because the identification of matrix domains and sub-domains is not always easy: It is sometimes hard to decide whether something is an inherent part of another, larger domain or concept. Even Ruiz de Mendoza admits that different views are possible: While it is true that a ruler and the army under his command may be regarded as different elements of a common conceptual frame (a war or a battle), it is also possible to conceive of Napoleon’s army as part of our knowledge about the French general. (Ruiz de Mendoza and Díez Velasco 2004: 299) An approach that is entirely based on the distinction of domains and subdomains can therefore be no more than a tentative proposal. In the next section, I will turn to typologies which describe metonymy with the help of more than one characteristic, i.e. as a category which comprises more and less prototypical instances. 85 4.3.2.3 Metonymy as a prototypical notion As the nature of domains is vague, Peirsman and Geeraerts (2006) propose a prototype model of metonymy which is based on the notion of contiguity. This may seem daring at first glance as contiguity is not a very clear concept either (cf. chapter 3.1). Peirsman and Geeraerts (2006) argue, however, that the problems with contiguity can be avoided when it is regarded as “conceptual contiguity” (2006: 273). Contiguity is not an objectively given property, but “constituted by a conceptual act” (Dirven (1993: 14), i.e. largely a matter of construal (cf. also Schmid 1993: 91-95), and different construals vary with respect to their prototypicality. The most prototypical type is, as Peirsman and Geeraerts (2006: 278) maintain, “spatial or material contiguity”. Instances of spatial metonymy like SPATIAL PART FOR WHOLE 85 The approaches discussed in this section are based on very dissimilar ideas of prototypicality. I will, however, only discuss them with regard to their usefulness for an identification of criteria contributing to the conventionality of metonymies, so that these differences are only of minor importance. 90 (e.g. England ‘Great Britain’) or CONTAINER FOR CONTENT (e.g. He drank a whole bottle), etc. are then classified on two dimensions: strength of contact and boundedness. The first accounts for extensions of the notion of spatial contiguity to embrace containment relations, contact and mere adjacency. The second allows a conceptualization of bounded objects as part of an unbounded object (2006: 279). The resulting classification is reproduced in Figure 12 with some of Peirsman and Geeraerts’s examples. Bounded Unbounded Part - Whole Spatial part & Whole: e.g. England ‘GB’, America ‘US’ Material & Object: e.g. There was cat all over the road, Fr carton ‘cardboard box’ (chocolate ‘substance as part of the drink’) Containment Container & Contained: e.g. I drank a glass, The milk tipped over (Dutch bubbels ‘champagne’) 86 Contact Location & Located: e.g. Washington is negotiating with Moscow, Ger Universität ‘building’ Adjacency Entity & Adjacent entity: e.g. Ger Liedertafel ‘choir’ Figure 12: The prototypicality of metonymies in the spatial and material domain (adapted from Peirsman and Geeraerts 2006: 285) According to this classification, spatial PART - WHOLE configurations are the most prototypical cases as they have high values on both dimensions. They involve bounded entities and the highest possible degree of contact. With decreasing boundedness and strength of contact the instances become less prototypical. This kind of analysis is repeated by Peirsman and Geeraerts to include the domains of TIME , ACTIONS / EVENTS / PROCESSES , ASSEMBLIES and COLLECTIONS (2006: 286-311) and results in a three-dimensional model of the category METONYMY . As Croft (2006: 319) points out, Peirsman and Geeraerts (2006) do not provide any evidence that the cases they consider prototypical are really more acceptable or frequent than others. 87 This criticism is only partly valid 86 The English noun bubbly, an informal word for ‘champagne’, is similar to the Dutch example given by Peirsman and Geeraerts (2006), since it makes use of the relation between a substance and one of its properties. According to the OED (s.v. bubbly), the noun bubbly is an elliptical form of bubbly water. Its formation was therefore not primarily motivated by the metonymic containment relation. 87 In Croft’s (2006: 323-324) opinion, a further weakness of Peirsman and Geeraerts’s (2006) account is that some of their examples do not belong to the realm of metonymy proper. For Croft, metonymies are restricted to nouns. Examples like Ger Ein Mäd- 91 since, as discussed in chapter 4.2.2.2, prototypicality does not inevitably coincide with high frequency. Nevertheless Croft’s substantial list of constructed examples which show that a mere consideration of contiguity relations is not enough to sanction metonymic uses is certainly of interest for a consideration of the conventionality of metonymy. Some of them are given below. While (42) to (44) are not acceptable in general, (45) seems - contrary to Croft’s opinion - acceptable in certain contexts, e.g. if the sentence refers to a frozen milk bottle. (42) PART FOR WHOLE : *I painted the drawer. ‘desk’ (a drawer is a part of the desk) (43) WHOLE FOR PART : *The house burned down. ‘dining room’ (the house includes the dining room) (44) CONTAINER FOR CONTAINED : *The glass is red. ‘drink’ (45) CONTAINED FOR CONTAINER : *The milk cracked. ‘bottle of milk’ (Croft 2006: 318) The criterion of contiguity alone cannot explain why (42) to (44) cannot be used, but other instances of the same mappings are acceptable, such as to fill up the car or to drink a glass. 88 In Schmid’s (1993: 94-95) account of the prototypicality of metonymies, the criterion of conceptual contiguity is accompanied by three others, which are partly responsible for the creation of a conceptual contiguity in the first place. Conceptual contiguity is the result of our experiences. Entities which frequently co-occur in culturally or otherwise vital situations, events or actions and which fulfil important functions within them are particularly likely to lead to linguistic metonyms. By way of the co-occurrence, the two entities become associated in speakers’ minds and thereby acquire the status of associative conventions. Prototypical metonymies are therefore characterized by featuring: • conceptual contiguity, which is the result of • the co-occurrence of the entities involved in the metonymic relation, and leads to chen zum Altar führen (literally ‘lead a girl to the altar’, i.e. ‘marry’), he holds, are a matter of attentional selection. 88 In this case also the mapping scope proposed by Ungerer and Schmid (2006: 129-131; cf. chapter 3.2.2) is not helpful. They argue that the mapping scope restricts the interpretation of a metonymy to a “socially sanctioned encyclopaedic interpretation” (2006: 130) that typically corresponds to a cognitive model, i.e. a “stored cognitive representation[...]” (2006: 49). This, however, does not explain why the metonymic interpretations of (42) to (44) are not socially sanctioned, because, for example, our cognitive model of a HOUSE certainly includes the information that it consists of rooms, such as a dining room, and the cognitive model of a GLASS clearly contains the idea that it is used as a container for drinks. More promising is Croft’s (2006: 323) view, who proposes the laws of ontological salience as good candidates for such an explanation. This issue will be discussed in more detail in chapter 6. 92 • conventionalized associations between them, which in turn sanction • the stand-for relation. This definition allows an explanation of some of Croft’s (2006) counterexamples to Peirsman and Geeraerts’s (2006) claims. For example, it is impossible to refer to a ‘pillow’ by means of the expression Swiss Army knife even if the two are spatially adjacent, because the two items do not stand in any relation whatsoever to one another. They do not co-occur in any important situation and there is no socially sanctioned association between them. Frequently heard metonyms, however, such as the White House for ‘the US government’ are based on co-occurrence and conventionalized associations and can therefore be understood effortlessly (Schmid 1993: 94-95). But why, for example, (44) is not acceptable remains unclear, since glasses and drinks unquestionably co-occur in extralinguistic reality. Barcelona (2002a, 2003, 2004) takes the distinctness of metonymical vehicle and target as his starting point for a description of the prototypicality of metonymies. He distinguishes three major types of metonymy: schematic, typical, and prototypical. Examples of the single types are (46) to (48): (46) This book is very large. (47) This book is a history of Iraq. (48) Washington is insensitive to the needs of the people. (Barcelona 2002a: 215, 226-227) For schematic metonymies the boundaries between vehicle and target domain are not as discrete as for typical ones. That the domain of PHYSICAL OBJECTS is distinct from the domain of BOOKS as a whole (cf. example 46) is less obvious than that the SEMANTIC CONTENT is distinct from the PHYSICAL OBJECT in (47). A prototypical metonymy (cf. example 48) is defined as “a typical metonymy with individual entities as targets and as referents” (Barcelona 2002a: 228). With increasing prototypicality, metonymies also become more and more noteworthy. We are much more likely to notice the figurative motivation of prototypical cases such as (48) than that of typical or even of purely schematic instances. While this is probably true, Barcelona’s account runs into difficulties when he tries to incorporate the factor of conventionality into his classification. According to his approach both established examples like (48) as well as highly context-dependent metonymies like The ham sandwich is at Table 7 (Barcelona 2004: 222, 228; cf. also chapter 4.3.2.1) fall into the same category, i.e. prototypical metonymies. To account for this difference Barcelona (2003: 243-246, 2004: 229-232) introduces an additional class of conventional metonymies, which are marked by being socially sanctioned. At the same time, however, he draws a parallel between the increasing degree of noteworthiness (2004: 229) or “metonymicity” (2003: 246) and the likelihood of conventionalization (cf. Figure 13). 93 prototypical metonymies typical metonymies schematic metonymies Metonymicity Likelihood of conventionalization Figure 13: Barcelona’s types of metonymy and their likelihood of conventionalization (based on Barcelona 2003: 246) Schematic metonymies display the “lowest degree of likelihood” (Barcelona 2003: 246) of becoming conventional. However, considering that cases like (46) or (47) are prime examples of Langacker’s (1984, 1987, 1993) active zone phenomena or reference-point constructions and that referencepoint reasoning is quite common (cf. chapter 3.2.2.2), this claim has to be rejected. It is not unusual to refer to sub-domains which can hardly be separated from their wholes by way of mentioning the wholes. While it is certainly true that the degree of metonymicity increases over Barcelona’s different classes of metonymies, it is extremely doubtful that metonymicity and conventionality coincide. An inverse relationship between metonymicity and conventionality is more likely, 89 with the schematic cases being more conventional than the prototypical ones. The schematic ones are probably accepted by any native speaker as perfectly literal forms of speech, whereas the prototypical instances need to fulfil additional criteria to be socially sanctioned. With regard to the degree of conventionality, Koch (2001) differentiates between “metonymic ad hoc effects” (2001: 206) and conventional metonymies. The former occur in individual speech events only and are idiosyncratic. The latter lead to polysemy and therefore play a role in the lexicon (2001: 206-207). Cases falling in between the two extremes, i.e. metonyms which are not totally idiosyncratic but which are not yet entirely fixed either, are not mentioned. Whether a metonym has the potential for becoming conventional is said to depend on the type of metonymy. Let me illustrate this with some examples: (49) The buses are on strike. (50) Paul is tanned. (51) His native tongue is German. (52) The ham sandwich is getting restless at Table 20. (Koch 2001: 208, 219, 220) 89 This claim is obviously a very general, if not a too general one. It is not excluded that instances of what Barcelona calls prototypical metonymy can become as conventional as instances of his schematic metonymy. Barcelona’s assumptions, however, are not less general. 94 Example (49) differs from (50) in terms of the shift of reference involved: While the literal and the metonymic meanings of bus correspond to two distinct classes of referents, Paul ‘Paul’s skin’ in (50), does not lead to a full shift of reference. The former cases are called referent-sensitive and the latter non-referent-sensitive processes (2001: 219). Only referent-sensitive processes, Koch holds, ought to be considered proper metonymies (2001: 221). Example (52) is entirely referent-oriented. Its conventionalization potential is low because its use is restricted to a very limited number of situations. The same can be said to hold true for bus in (49). As Koch puts it: In these cases, the distinction between ad hoc metonymy and metonymic polysemy does not apply. [...] To put it more precisely, referent-oriented metonymies [...] do not pertain at all to the realm of the lexicon, but to discourse semantics [...]. (Koch 2001: 219-220) 90 Concept-oriented metonymies (Koch 2001: 219) such as (51) are “predisposed to habitualization” (Koch 2004: 28). As to where he sees the boundary between what is potentially relevant for the lexicon and what is only relevant within the context of a given discourse, Koch remains vague. One issue seems to be whether the category referred to is potentially name-worthy. L ANGUAGE is a name-worthy concept (Koch 2001: 219), while CUSTOMERS ORDERING HAM SANDWICHES are not. This explains why tongue ‘language’ is included in the concept-oriented class and why ham sandwich ‘customer’ is not. But it does not explain why bus ‘bus driver’ is also excluded from this class, since BUS DRIVERS certainly form a name-worthy category. 91 All in all, Koch (2001: 225) considers metonymy as a category with more or less prototypical members. Tongue ‘language’ is mentioned as a central member of the category as it displays all three central properties of metonymies: • referent-sensitivity, • concept-orientation, • and a figure/ ground effect. The first two criteria were already discussed. Koch illustrates the third by the difference between child ‘very young person’ (originally ‘offspring’) and book ‘content of a book’. Only the former, he argues (2001: 222), “implies a real figure/ ground effect”. The explanation offered is somewhat unclear. Book does not lead to a proper figure/ ground effect since the dif- 90 In his later article, however, Koch (2004: 27) contends that referent-oriented metonymies can become part of the lexicon. 91 One could argue that bus ‘bus driver’ is less relevant for the lexicon as there is already an established lexeme to denote the category. However, this also holds for tongue ‘language’. I do not wish to argue that the metonym bus is more conventional than the metonym tongue, but Koch’s explanation is not entirely satisfying. Cf. Appendix A, for the frequency of the metonym bus referring to ‘bus drivers’. 95 ferent semantic facets inherently belong to one and the same entity. 92 That child exhibits a ‘real’ figure/ ground effect seems to be based on the fact that it allows reference to distinct denotata in the extralinguistic world: A child does not have to be a person’s own child in order for that person to use the word child to refer to him or her. However, a child is invariably somebody’s child. This idea (i.e. the original sense of the word) might therefore very well be part of our concept VERY YOUNG PERSON . This would render the figure/ ground effect involved much less pronounced than Koch assumes and make the metonym child less prototypical. For a consideration of conventionality, however, the criteria of referentsensitivity and figure/ ground effect are of less interest. Crucial in this respect, Koch says (2004: 28), is whether a metonym is concept-oriented or not. As already pointed out above, this is inherently tied to the stability and the name-worthiness of a category - a criterion already mentioned several times in the previous sections. 4.3.2.4 Criteria of conventionality and the ‘death’ of metonyms The examination of the relevant typologies of metonymy mainly supports the assumptions about the factors contributing to the conventionality of metonymy introduced in chapter 4.2. The name-worthiness of the category referred to by way of a metonym plays a role and the relative stability of this category. Apart from this the direction of the metonymic mapping is of significance, as is evident, for example, in Ruiz de Mendoza’s work: The speech community ought to have a preference for WHOLE FOR PART mappings in contrast to PART FOR WHOLE mappings. This ties in well with what has been said about the relevance of the laws of ontological salience above (cf. chapter 4.2.2.2). In my analyses in chapter 6, I will thus take up the hypothesis that WHOLE FOR PART mappings ought to be favoured as compared to PART FOR WHOLE and PART FOR PART mappings, even though any study of this kind will be subject to the problem of having to distinguish between wholes and parts and therefore only be valid within the framework of the sometimes necessarily somewhat subjective decisions taken in analyzing single specific mappings as instances of WHOLE FOR PART or PART FOR WHOLE mappings. 93 Another factor that occurs in more than one of the typologies is that conventional metonymies are ‘socially sanctioned’. Barcelona (2003: 243- 246, 2004: 229-232) mentions this; Schmid (1993: 94-95) quotes the social sanctioning of the associations between vehicle and target as one of his criteria for the prototypicality of metonymies; Ungerer and Schmid (2006) 92 This is in line with Croft’s (2002: 179-180) argument, who also points out that cases like book have not always been regarded as metonymic, but were instead considered instances of lexical ambiguity. 93 The results of my empirical analysis concerning the relative frequency of WHOLE FOR PART , PART FOR WHOLE , and PART FOR PART mappings are discussed in chapter 6.2.1.2. 96 argue that the interpretation of metonymies is restricted to a “socially sanctioned encyclopaedic interpretation” (2006: 130) by the metonymic mapping scope (cf. above, Fn. 88). And Nunberg’s (1995: 112-115) idea that predicate transfer can only gain currency when the attributed property is generally valid can also be interpreted to mean that the relationship between vehicle and target must be socially sanctioned. That conventionality is essentially dependent on social sanctioning is an indisputable fact. However, what exactly promotes the social sanctioning in the first place is hardly ever addressed in detail. Importantly, for a comprehensive account of the conventionality of metonymies, it is not enough to look at the level of conceptual mappings (cf. Croft 2006; Schmid 1993). Single linguistic instantiations may very well differ in being more or less acceptable. And the acceptability or conventionalization potential of metonyms is largely dependent on typical associations between entities which are related to the knowledge stored within a domain or concept. But can metonyms also ‘die’? As Panther and Thornburg point out, the metonymic meaning of a linguistic expression can replace the original meaning: “the target meaning [can] become maximally prominent” (2004: 107; original emphasis). This would be an extreme case where the metonymic relation is entirely lost, and the metonym can be said to be ‘dead’. An example is Fr témoin ‘person giving testimony’ Koch (2004: 17), which can no longer be used in its former meaning ‘testimony’. On the scale of conventionality such cases only constitute one of the extremes. The other is occupied by entirely novel metonymies, and somewhere in between we find the relatively frequent, conventional cases. 94 This brings me to my last theoretical point before I turn to the empirical investigation of the conventionality of metonymy and metaphor in chapter 5, i.e. to the question of whether the distinction of conventional and more creative or uncommon figurative construals has any psychological reality. It seems intuitively plausible that the continuum of conventionality assumed here does not remain without consequences in terms of processing: Uncommon or even entirely novel figurative construals should be harder to process and comprehend than highly conventional ones. For the latter, one can reasonably assume that the figurative meanings are stored 95 and 94 Cf. also Papafragou (1996). Her continuum reaches from novel over conventionalized metonymies to fully semanticized instances. An example of the latter is Fr grève which has gone through several stages of semantic change and extension: “’bank of Seine’ → ‘spot on Seine’s bank where unemployed workers met’ → ‘act of stopping work’ - strike” (1996: 192). As Fr grève, however, can still be used in the sense ‘bank, beach’, i.e. is still polysemous, the metonym is not entirely ‘dead’. 95 As far as metonymy is concerned, this seems to be at odds with Langacker’s (e.g. 1993) reference-point approach according to which the sole function of the vehicles is to grant the language user access to the intended target. This could be taken to imply that the ‘content’ meaning of glass, for example, is not seen as a stored meaning of the lexeme, but that the interpretation of glass as a CONTAINER FOR CONTENT metonymy 97 can, accordingly, be understood rather effortlessly, while the meanings of the former have to be arrived at by more complex reasoning procedures. This leads to the realm of psycholinguistics where the question of how humans process and understand figurative language has inspired countless studies. 4.4 Processing nonliteral language: Linguistic theory and psycholinguistic evidence Much research has been devoted to studying the cognitive processes that take place when humans understand language. A first wave of psycholinguistic studies of the processing differences between strictly literal utterances and (allegedly) cognitively more demanding cases caused, for example, by indirectness in language use, was initiated as a result of the pragmatic shift in linguistics in the 1970s. Apart from indirectness, the resolution of lexical ambiguity (cf. e.g. Foss 1970; Hogaboam and Perfetti 1975; Tanenhaus, Leiman and Seidenberg 1979) and the comprehension of figurative language soon became prominent in psycholinguistic research (cf. e.g. Harris 1976; Johnson and Malgady 1979; Ortony et al. 1978). However, although - or maybe because - a wealth of studies of the processing of figurative language has been published since, no unified picture has yet emerged (cf. e.g. Gibbs 2001: 317). The evidence observed can be assigned to three different theoretical paradigms: the traditional model of sequential processing, the so-called direct access view, and a more recent view that is based on the notion of salience. In what follows, I will introduce these different positions and discuss their implications for my empirical studies of the conventionality of metaphors and metonymies in the subsequent chapters (5 to 7). Since figurative language is such a fertile field for psycholinguistic research, I can, however, not provide a detailed overview, but only briefly outline some of the most important theories and studies. The discussion will be largely limited to questions of comprehension; questions relating to the production of metaphorical and metonymic utterances can only be addressed in passing where they are of immediate relevance. As most of the research into figurative language focuses on metaphor, while the psycholinguistic evidence concerning metonymy processing is scarce (cf. Frisson and Pickering 1999: 1368), the following chapters will largely deal with metaphor comprehen- consists of a two-stage comprehension process. A direct connection between glass and the meaning ‘container’ seems to be excluded. However, similar to metaphors, where the metaphorical mappings proposed are not primarily meant to reflect the mental processing in a direct way (cf. Lakoff 1993: 210), metonymic reference-point reasoning should also be considered a more general, theoretical model and not a model of processing. 98 sion. The question of the extent to which the results for metaphor processing can be transferred to metonymy will be addressed in chapter 4.4.3.2. 4.4.1 The standard pragmatic model: Sequential processing The idea that the comprehension of figurative language invariably involves several steps has its roots in early speech act theory and Grice’s ideas on meaning in discourse. In general, speakers are bound to the Cooperative Principle, but the maxims can be exploited to communicate more than what is actually said. If a speaker deliberately flouts a maxim, this “gives rise to a conversational implicature” (Grice 1975; 1998: 152) - hence the distinction of ‘what is said’ and ‘what is implicated’ (Grice 1975; 1998; cf. also Grice 1978; 1998: 162). 96 Consider the metaphor in (53): (53) You are the cream in my coffee (Grice 1975; 1998: 156) The utterance flouts the first maxim of quality (Grice 1975; 1998: 149). But as long as the hearer has no reason to believe that the speaker is unwilling to cooperate, he or she will seek a possible relevant interpretation. It is essential that the hearer arrives at this interpretation only after going through a series of steps: 97 The first involves realizing that a maxim (or more than one) has been violated, the second consists of looking for a possible meaning - other than the surface meaning - the speaker could want to convey, and the third leads to the conclusion that the speaker actually wanted to communicate the assumed implicature because the speaker knows that the hearer can work out step two (cf. Grice 1975; 1998: 153). For his initial interpretation the hearer relies on “the conventional meaning of the words used” (Grice 1975; 1998: 153; my emphasis, SH). Only if this interpretation violates the maxims of the Cooperative Principle, does it serve as a starting point for a reasoning process that leads to a new interpretation, which mainly depends on the interlocutors’ shared background knowledge and the linguistic and situational context. It is one of the central features of conversational implicatures that they “must be capable of being worked out” (Grice 1975; 1998: 153). 98 It must, at least in principle, be possible to construct an argument consisting of the three steps indicated above. Otherwise, the implicature is not a conversational, but a conventional one. 96 Cf. also Grice’s earlier article on “Meaning” (1957), where he discusses natural vs. nonnatural meaning. 97 Grice’s steps are very similar to those Searle (1979: 105-106) proposes for metaphor interpretation. It is important to note that, although Searle explicitly mentions literal defectiveness as a triggering condition in his first step, he admits that the latter is not a necessary precondition for metaphor. 98 For other characteristics of conversational implicatures, such as cancellability and non-detachability, cf. Grice (1975; 1998: 160-161 and 1978; 1998). 99 The differentiation of conversational and conventional implicatures shows that Grice was more aware of the role of conventionality in language and understanding than is commonly realized. Even though later research tends to simplify his original intentions by equating his “conventional meaning” (1975; 1998: 153) with strictly literal language, one of Grice’s own examples proves that this is not the case: (54) He is in the grip of a vice. (Grice 1975; 1998: 147) Grice’s account of the conventional meaning of (54) does not stop at the literal sense of a person being caught in a tool. The first reading he mentions is a metaphorical interpretation, where the referent of he is “unable to rid himself of a certain kind of bad character trait” (Grice 1975; 1998: 147). Conventional figurative language is explicitly part of “what is said” (Grice 1978; 1998: 162). Only unconventional cases demand a complex strategy for a successful interpretation. In terms of on-line processing, this means that literal as well as highly conventional figurative meanings are assigned a privileged status. Only unconventional nonliteral meanings are secondary and derived. Searle (1979: 115) takes a similar position. ‘Dead metaphors’ are said to be computed directly without recourse to their ‘literal’ meanings. His examples, like heated argument or sexual frigidity (1979: 98), are highly reminiscent of what the conceptual theory considers to be conventional metaphors, in this case instances of a mapping which could be formulated as EMOTION IS TEMPERATURE , with the more specific entailments INTENSE EMOTION IS HEAT and LACK OF EMOTIONS IS COLD . Typically ignoring that according to both Grice and Searle highly conventional figurative language can be processed directly, psycholinguists took the stage-like reasoning process described by both as a point of departure for empirical research. The basic assumptions of the so-called standard pragmatic model can be summarized thus: 1. The literal meaning of an utterance is always activated 99 and invariably so before any other possible interpretation. 2. Nonliteral interpretations are only arrived at after a triggering condition. 3. Therefore, nonliteral meanings require more cognitive effort than literal ones and more contextual support. (after Gibbs 1994: 83-84; Giora 1997: 184, 2003: 105) 99 Usually, psycholinguists distinguish between access and activation. The former refers to subconscious processes which are at work until 300 ms after the stimulus has been encountered. The latter describes a later processing stage, i.e. a more conscious one. Unfortunately, however, these terms are not used consistently in the relevant literature. It is not always entirely clear whether the sources use the terms in the way just explained; they often seem to use the terms interchangeably. Since these differences are, however, not the focus of my interest, the terminology was taken over in the form in which it occurred in the sources indicated. 100 The initial processes involved in the comprehension of literal and nonliteral language are essentially the same. Both require an activation of the literal meaning. For literal utterances this is the whole story. Figurative ones, however, require additional stages. In the first test of this model, Clark and Lucy (1975) compared the times it took participants to process direct and indirect requests (e.g. Color the circle blue. vs. Can you color the circle blue? ) and observed that indirect ones took significantly longer (around one second). This seems to support the prediction that speakers do indeed construct the literal meanings of sentences during the comprehension process (1975: 68). 100 As far as figurative language in the narrower sense is concerned, psycholinguistic studies have mainly focused on metaphors and idioms. Do listeners really have to activate the literal meaning of metaphorical utterances in all situations? Blasko and Connine (1993), for example, have demonstrated that contextually inappropriate meanings are initially activated when hearers encounter novel metaphors. In cross-modal semantic priming experiments, 101 subjects were auditorily presented with metaphorical sentences and, at the same time, with visual targets either related to the literal reading of the sentence or to its metaphorical reading or entirely unconnected to the sentence meaning. An example is given in (55): (55) When she realized that HER THOUGHTS WERE A BOILING KETTLE , Marie decided she had to make the call. Metaphor target: turmoil Literal word target: whistle Control target: happening (Blasko and Connine 1993: 307) The participants had to decide whether the targets shown on a computer screen were indeed words of the English language, i.e. had to make lexical decisions. Reading hardly familiar metaphors like (55) only facilitated lexical decisions for the literal targets. The metaphorical targets took longer to verify as existing lexemes than the completely unrelated control targets (Blasko and Connine 1993: 298). For more familiar metaphors (e.g. Happiness is gold), however, the comprehension of the metaphorical target (e.g. precious for Happiness is gold) was facilitated as well. 100 That the indirect versions are longer than the direct ones might, however, also have played a role. Janus and Bever (1985: 474-475) point out another reason why Clark and Lucy’s (1975) results cannot be taken as proving the standard pragmatic model: Since the model predicts that the inference process works on the basis of context, and Clark and Lucy did not provide any context at all, Janus and Bever argue that “their findings cannot be considered definitive” (1985: 475). Cf. also Gibbs’s (1979) results on the influence of context on the comprehension of indirect requests and Gibbs (1994: 86) on the general importance of a consideration of context in psycholinguistic experiments. 101 One of the classic sources on cross-modal semantic priming is Swinney (1979). Cf. also Tabossi’s (1996) overview of this method. 101 Other studies (e.g. Gerrig 1989; Gibbs 1990; Onishi and Murphy 1993) have also shown that novel meanings can lead to initial processing difficulties. Gerrig (1989), for example, studied the comprehension of “preempting innovations” (1989: 194), i.e. words which have a conventional meaning but are used in an innovative way in a given context. Uniform, for example, can be used in the ‘clothing’ sense or metonymically, meaning ‘uniformed police officers’ or ‘soldiers’. The second reading takes more time to be arrived at because it demands more than mere sense selection - it has to be created 102 (cf. e.g. Clark 1983; Clark and Clark 1979; Clark and Gerrig 1983; Gerrig 1989). Gibbs (1990) investigated figurative referring expressions. He has observed that an expression like creampuff to refer to a cowardly ‘boxer’ is more difficult to understand than a literal reference like fighter. And metonymic references (e.g. glove to refer to a ‘baseball player’) are even more difficult to process than metaphorical references. Onishi and Murphy (1993) argue convincingly that the differences in processing times between literal and figurative expressions found in Gibbs’s (1990) study are ultimately a result of the referential form of his stimuli. Predicative metaphors taking the form of X is a Y are not necessarily processed more slowly than literal expressions (cf. e.g. Gildea and Glucksberg 1983; Keysar 1989). While in a predicative metaphor, such as My cat is a princess (Onishi and Murphy 1993: 769), both the topic and the vehicle are explicit and only the ground of the metaphor has to be inferred, referential metaphors (e.g. My princess won’t eat) leave it to the hearer to infer the topic and the ground which accounts for the extra processing time. For metonymic references things are still more difficult, as the metonymic interpretation has to compete with a contextually plausible literal alternative. In a text dealing with baseball, for example, a sentence beginning with The glove... could very well receive a literal interpretation, while a literal reference to a creampuff is much less likely. Taken together, the results of the above-mentioned studies suggest that the literal meanings of utterances which ultimately demand a figurative interpretation are indeed activated - but not invariably so. Literal meanings are only activated when the figurative expression is referential rather than predicative or when the idea conveyed is largely unfamiliar. 103 Many 102 While sense selection is driven by general world knowledge (stored as schemas, frames, scripts, etc.; cf. e.g. Fillmore 1985; Minsky 1975; Rumelhart 1980; Schank and Abelson 1977), sense creation relies on contextual knowledge shared by discourse participants (cf. Gerrig 1989: 205). However, as Gerrig (1989) has shown in his investigation of novel metonymic uses of established words (e.g. beard ‘man with a beard’), novel meanings are also created initially so that the processes of sense selection and sense creation compete with each other. 103 That novel metaphors are processed differently from literal language has also been shown using event-related brain potentials (ERPs; cf. Coulson and Van Petten 2002; Kutas and Van Petten 1994). Tartter et al. (2002), for example, observed that novel 102 other studies support the view that nonliteral meanings are not necessarily harder to find than literal ones. For idioms, which are marked by their high degree of conventionality, the literal interpretations even take longer than the nonliteral ones (cf. Gibbs 1980; Ortony et al. 1978; Swinney and Cutler 1979). The generation of nonliteral meanings can thus be obligatory (cf. Glucksberg, Gildea and Bookin 1982; Keysar 1989), and cannot be suppressed initially, even if the context is biased for a literal reading. 104 This indicates that there are mainly two factors which determine how a figurative meaning is comprehended: conventionality and context. The influence of the latter factor on language processing has shaped the second research paradigm to be discussed, the so-called direct access view. It holds that if enough contextual support is provided, nonliteral language is not more difficult to comprehend than literal language and that all meanings are arrived at directly. 4.4.2 The direct access view: The influence of context on comprehension According to the direct access view always the “contextually compatible meanings would be tapped directly” (Giora 2002: 489) - no matter whether figurative or literal. Literal language is not given priority over figurative language. Instead, context is said to exert a strong influence in the earliest stages of the comprehension process. Meanings that do not fit the context are not activated at all or are suppressed (cf. Gernsbacher 1990; Gernsbacher and Robertson 1999; Gernsbacher et al. 2001). Among the first to demonstrate that metaphor comprehension is related to context were Ortony et al. (1978). Processing ease, they maintain, is intimately connected to the length of the context given (1978: 470, 473) and depends on “the transparency of the relationship between a target and its preceding context” (1978: 473). Ortony et al. measured the time it took subjects to read target sentences that were presented subsequent to different types of contexts: literal inducing and metaphorical inducing contexts, both in either long or short versions. An example is (56): metaphors cause more processing effort, but at a very early stage of the comprehension process, which is usually not associated with semantic analysis. 104 Cf. also Tanenhaus, Leiman and Seidenberg’s (1979) study which indicates that contextually inappropriate readings of noun-verb ambiguities (e.g. rose ‘flower’ vs. past tense of rise) have to be suppressed (1979: 436). For further information on the suppression of initially activated but contextually inappropriate meanings, cf. e.g. Gernsbacher (1989) and Giora (2003: 26-27). The question of suppression and enhancement does not only pertain to the activation of whole meanings, but also to those properties made use of to create and understand metaphors (cf. e.g. Gernsbacher and Robertson 1999; Gibbs 2001: 327; Glucksberg, Newsome and Goldvarg 2001: 281-290; and chapter 7.4). 103 (56) Literal inducing context: Approaching the enemy infantry, the men were worried about touching off landmines. They were very anxious that their presence would be detected prematurely. These fears were compounded by the knowledge that they might be isolated from their reinforcements. The outlook was grim. Metaphorical inducing context: The children continued to annoy their babysitter. She told the little boys she would not tolerate any more bad behavior. Climbing all over the furniture was not allowed. She threatened to spank them if they continued to stomp, run, and scream around the room. The children knew that her spankings hurt. Target: Regardless of the danger, the troops marched on. (Ortony et al. 1978: 467) For the short context condition, the contexts were significantly reduced to Approaching the enemy infantry and The children continued to annoy their babysitter. Overall, the short contexts caused longer reading times of target sentences, with metaphorical readings taking even more time than the literal ones. The long context, however, reduced the differences in reading times between literal and metaphorical sentences to a no longer significant level (1978: 470). Reading the preceding context activates the corresponding concepts in subjects’ minds. For the short contexts, the conceptual representation created by the linguistic input is rather poor and unspecific. But in the long context condition the amount of information is large enough to construct an elaborate mental picture. 105 The interpretation of the target sentence is guided by the concepts that are activated and thus cognitively salient due to the preceding context (Ortony et al. 1978: 473-474). Integrating the metaphorical meaning of the target sentence Regardless of the danger, the troops marched on into the detailed conceptual framework evoked by the long metaphorical context is an easy task. That the relative processing ease depends not on literality or nonliterality, but rather on “relatedness to context” (Ortony et al. 1978: 475) is reflected in the equal reading times observed for the literal and nonliteral target sentences in the appropriate contexts. The contextually inappropriate literal meanings do not seem to be 105 This ‘mental picture’ largely corresponds to Van Dijk and Kintsch’s (1983: 337) situation model of discourse. They hold that to understand a text, it is necessary to be able “to imagine a situation in which certain individuals have the properties or relations indicated by the text”. We link the linguistic input to stored experiences and extant knowledge structures in order to construct a mental model of the situation described by the text, which, then, acts as a basis for our interpretation of it. For a more elaborate account of the relevance of the situation model for discourse comprehension, cf. Van Dijk and Kintsch (1983: esp. chapter 10), and Kintsch (1994, 1998). A shorter overview of the model can be found in Fletcher (1994). Cf. also Foltz (2003). 104 activated at all if the preceding context strongly supports the metaphorical interpretation. Otherwise, the consideration of the literal sense and its subsequent rejection should result in measurable processing time differences between literal and nonliteral target sentences under all conditions. Thus, Ortony et al. (1978: 475) conclude that people process literal and figurative language in exactly the same way . That literal and nonliteral language comprehension is based on the same mechanisms is also the position taken by many other psycholinguists (cf. e.g. Gibbs and Gerrig 1989; Glucksberg, Gildea and Bookin 1982; Keysar 1989, 1994). Language understanding is essentially a process of disambiguation (Gibbs and Gerrig 1989: 150-152). What counts for a successful interpretation are the common ground of speaker and hearer, world knowledge, and co(n)text. And this holds for nonliteral language as well as literal expressions, like, for example, homonyms (e.g. bank ‘side of a river’ vs. bank ‘financial institution’). 106 Metaphor or metonymy comprehension then does not demand a different amount of context than literal language comprehension. That quantity matters seems to be an undeniable fact. This has been shown in many studies, like Gibbs (1984, 1989), Gildea and Glucksberg (1983), Inhoff, Lima and Carroll (1984), Rumelhart (1993) - to list only some of the classics in this field. 107 But are there any qualitative differences between contexts that lead to differences regarding comprehension? Or, to put it differently, are there any linguistic expressions that require a special type of context for a successful interpretation? Psycholinguistic tests suggest that this is indeed the case. Some metaphorical expressions like (57) can be understood automatically and rapidly without any context (cf. Glucksberg, Gildea and Bookin 1982: 95). Others like (58) seem harder to understand (Gildea and Glucksberg 1983: 579). (57) Some surgeons are butchers. (58) All marriages are iceboxes. (Gildea and Glucksberg 1983: 578-579) The divergence results from the respective topic-vehicle combinations in the examples. Both have the general form X is a Y. Taking for granted that the statements are informative (cf. Grice 1975; 1998), the addressee will follow the assumption that the new information is to be found in the second element, 108 i.e. in the vehicles butchers and iceboxes. While interpreting 106 For more detailed information on the influence of context on the resolution of lexical ambiguities caused by homonyms, cf. e.g. Duffy, Morris and Rayner (1988); Martin et al. (1999); Sereno, Pacht and Rayner (1992); Vu, Kellas and Paul (1998); and Vu et al. (2000). Also of interest is Kintsch’s (1988) account of the activation of contextually appropriate meanings of homonyms. 107 The evidence is by no means limited to metaphor processing, but includes other types of nonliteral language, like idioms, sarcasm, proverbs, and indirect speech acts as well. 108 Conventionally, the information distribution within sentences is in accordance with what has been called the “Given-New Strategy” (Haviland and Clark 1974: 512; cf. also Clark and Haviland 1977), i.e. what is assumed to be known information syntac- 105 (57), addressees have to activate the concept BUTCHER , which not only contains neutral factual information on butchers and their profession, but is marked by stereotypical assumptions about their being brutal. With respect to the topic, they are certainly more informative than other properties of butchers, such as that they are ‘human’, ‘sell meat’, etc. That the concepts SURGEON and BUTCHER have quite a similar skeletal structure, which includes agent-patient relations, typical instruments, and locations, has already been discussed. 109 Gildea and Glucksberg (1983: 579) claim that these shared properties of topic and vehicle are “ignored” by the addressee as they are uninformative about the metaphorical statement. And this is said to be the reason why (57) draws the addressee’s attention immediately to the conflicting stereotypical assumptions about surgeons and butchers, i.e. ‘skilfulness’ vs. ‘brutality’. The latter property is applied to the topic in a successful interpretation of (57), because it is rendered most salient by the utterance. 110 However, maintaining that the other properties are totally ignored means underestimating the role of analogous structure in the right interpretation: it acts as a necessary background that enhances comprehension. But why is (58) more difficult to understand in the absence of an appropriate context than (57)? According to Glucksberg, Gildea and Bookin (1982: 96), butcher belongs to a fairly small class of stock vehicles and invariably, i.e. conventionally, transports a negative stance towards the topic and is therefore relatively easy to comprehend. (57) thus relies on a lean mapping (cf. chapter 3.2.1). Even though source and target have an analogous structure, the single correspondences are not utilized in different metaphorical expressions. Lean mappings are conventionally fixed in terms of their mapping scope as well as regarding their linguistic forms, or to put it differently, the nonliteral meanings have become part of the established meanings of the vehicle terms. Their interpretation therefore requires no, or at least no specific, context. As Gildea and Glucksberg point out, however, (58) is ambiguous: On the one hand, it could be an instantiation of the conceptual mapping LACK OF AFFECTION IS COLD . On the other, as Gildea and tically precedes the new information. This principle has its origins in theoretical linguistics and is also discussed by Halliday (1967). Importantly, it is related to the Prague School notion of functional sentence perspective (FSP) and Firbas’s concept of communicative dynamism (cf. e.g. Daneš 1974; Firbas 1964, 1971, 1992; Mathesius 1929). 109 Cf. the discussion of the example This surgeon is a butcher in chapter 3.3.2, where it is discussed from a conceptual blending perspective. 110 Gildea and Glucksberg (1983) take over the notion of salience from Ortony (1979), who defines it as “subjects’ estimates of the prominence of a particular attribute with respect to a concept to which it does or could apply” (1979: 162). He acknowledges that salience is context-dependent, but his salience imbalance model (cf. also Ortony et al. 1985) also almost exclusively focuses on the differences between the concepts involved in a metaphor and neglects the role of the shared elements to a large degree. Cf. also chapter 7.4.2, where the development of stock vehicles is discussed. 106 Glucksberg (1983: 588) suggest, it could receive an alternative interpretation like “marriages are devices that preserve perishable things”. Although this reading of (58) may seem a little far-fetched, the existence of a possible alternative interpretation 111 bears witness to the fact that conventional metaphors in the sense of Lakoff and Johnson (1980), i.e. metaphorical expressions that rely on a rich mapping, are less fixed than lean mappings: Alternative nonliteral readings are apparently not entirely ruled out if no context is provided. 112 But where does context come into play? Gildea and Glucksberg’s (1983) main argument is that potentially ambiguous vehicles, such as icebox, cannot be understood correctly without any contextual clues. Moreover, the context has to be of a certain quality: It has to activate properties of the vehicle which are relevant for the current topic. For (58), an informative context should activate the concept COLDNESS . As the results of the experiments show, a previous mention of the relevant concept facilitates the comprehension of metaphoric statements, no matter if the primes used are literal (e.g. Some winters are cold) or figurative (e.g. Some people are cold; 1983: 583). Even more general primes just referring to the relevant general semantic field (e.g. Some summers are warm) enhance comprehension (1983: 586). Reviewed in the light of conceptual metaphor theory, this would mean that effective contexts are those which guide the addressee from the linguistic input (icebox) to the relevant conceptual source ( COLD ). Considering that one conceptual metaphor often has various sub-types which rely on the same basic idea, as, for example, the two converse forms AFFECTION IS WARMTH and LACK OF AFFECTION IS COLD of the mapping between EMOTIO- NAL INVOLVEMENT and TEMPERATURE , it is by no means surprising that even linguistic input which only activates the right source dimension is helpful for a successful interpretation. However, as opposed to Gildea and Glucks- 111 That they treat the two readings of (58) as of equal status shows that Gildea and Glucksberg (1983) are not aware of the difference between conventional conceptual metaphors and other types of metaphorical expressions. There is no rich conceptual mapping underlying the ‘preservation’ meaning of (58), just as there is no rich mapping that supports (57). 112 Another difference between (57) and (58) is that the former could in principle be literally true, whereas the latter cannot, which makes it plausible that subjects had less trouble deciding about the literal falsity of (58) as compared to (57). That different quantifiers are used in the examples, i.e. some vs. all, also plays a role. Metaphorical expressions that are quantified with some lead to the so-called metaphor interference effect, i.e. a possibly true metaphorical interpretation slows down subjects’ decisions regarding the literal falsity of test sentences. When all is used instead, the effect does not occur and the metaphors are generally rated as less ‘good’ or ‘poorer’ than the versions with some (e.g. Some jobs are jails vs. All jobs are jails; cf. e.g. Glucksberg, Gildea and Bookin 1982). In the case of (58), metaphor interference did thus not occur. (57) is a ‘better’ metaphor merely due to the occurrence of some. However, if and how the potential truth of (57) influences its comprehension is, regrettably, left uncommented by Gildea and Glucksberg (1983). 107 berg (1983), Shinjo and Myers (1987: 236) come to the conclusion that general primes as well as topic primes do not facilitate but interfere with the comprehension process. 113 If Shinjo and Myers (1987) are right, metaphor processing works smoothest when the preceding context biases the specific intended nonliteral reading, i.e. when the appropriate source concept is rendered cognitively salient. So far, the findings are in line with the direct access view: Given a qualitatively and quantitatively appropriate context, the intended meaning is tapped directly and without any undue processing effort. But there are also results which are difficult to reconcile with this view. One classic example is Janus and Bever’s (1985) study. They used Ortony et al.’s (1978) material, but changed the methodology. Ortony et al. got their evidence for the direct understanding of metaphors by measuring the time it took subjects to comprehend the whole target sentence. Janus and Bever criticize this, since sentence understanding proceeds in an incremental manner and the cognitive load generally increases towards the end of clauses, as this is the place where people decide upon the meaning of the syntactic and semantic information provided by the linguistic input. This sentence wrap-up effect (cf. Just and Carpenter 1980; Levelt 1989; Hirotani, Frazier and Rayner 2006; Schlesewsky and Bornkessel 2003) 114 might have influenced Ortony et al.’s (1978) results. To get results that reflect the processes during the comprehension of the metaphor within the target sentence, it is necessary to focus on the point “where the metaphor is first encountered” (Janus and Bever 1985: 478). Thus, they examined the processing times within target sentences at the end of the phrase that constitutes the metaphor vehicle. An example is given below (the commas mark the boundaries between the phrases): (59) The fabric, had begun, to fray. (Janus and Bever 1985: 480) Janus and Bever’s (1985) results differ markedly from Ortony et al.’s (1978). Janus and Bever compared the mean reading times for the metaphorical condition (with a metaphorical inducing context) to those for the literal condition (with a literal inducing context). While in the literal condition the 113 At first glance, this seems to contradict the basic assumptions of the conceptual metaphor theory because, as pointed out above, general semantic field primes should activate the right source domain. However, the theory does not claim that metaphorical expressions are necessarily understood by way of language users actively carrying out the mapping process. On the contrary, the strength of this theory lies in the assumption that many metaphors are so conventional that we do not notice them any longer. For contradictory evidence concerning the activation of conceptual mappings during metaphor and idiom comprehension, cf. e.g. Gibbs and O’Brien (1990), Gibbs et al. (1997), Glucksberg, Brown and McGlone (1993), Glucksberg and McGlone (1999), McGlone (1996), Nayak and Gibbs (1990). 114 For the relevance of wrap-up effects for speech production, cf. e.g. Bock and Levelt (1994), Bock et al. (2003). For the locus of the effect of word frequency, cf. also Jescheniak and Levelt (1994). 108 mean was 96 centiseconds, the metaphorical condition resulted in a mean reading time of 116 centiseconds (1985: 482). This significant difference only comes to the fore when the masking effect of the extra processing necessary at sentence or clause endings is avoided by the experimental design. If one accepts the view that increased reading times reflect processing difficulty, the longer reading times suggest that the metaphors used in this study were indeed harder to understand than the literal interpretations of the same sentences. 115 Interestingly, Janus and Bever (1985: 485) explicitly mention the influence of the conventionality of metaphorical expressions on their processing. The metaphors in their study are all novel. Decoding them cannot be a matter of retrieving stored material. Highly familiar metaphors or “’[f]rozen metaphors’” (1985: 485) can, however, be understood directly without any consideration of the literal meaning. Considering that - at least theoretically - frozen and novel metaphors form the extremes of a continuum of more and less conventional metaphorical expressions, it could very well be the case that metaphors which are not exactly frozen but still have a high degree of conventionality are also processed directly. A similar view is held by Gregory and Mergler (1990: 171), who state that the dimension of familiarity “could be critical in investigating whether metaphoric processing is automatic or not”. Neither the standard pragmatic model nor the direct access view can explain all the different findings. While proponents of the direct access view, such as Gibbs, are certainly right in pointing out that figurative language comprehension does not always have to be mediated by the literal meaning, there is also reason to believe that there are cases in which the literal meaning is activated, at least initially. In realistic circumstances, different factors interact and influence the comprehension process. Context, the main issue propagated by the direct access view, undoubtedly constrains the interpretation of ambiguous utterances. But the degree of conventionality of a figurative expression also determines the speed and manner of processing. A model that openly acknowledges the difference between novel and familiar meanings is the graded salience hypothesis (e.g. Giora 1997), which serves as a basis of the discussion in the following sections of the influence of conventionality and frequency on language pro- 115 Other relevant studies concern idiom and proverb comprehension. Cacciari and Tabossi (1988), for example, observed that idioms are processed slowly: Literal meanings are activated initially, idiomatic meanings become available only after 300ms (1988: 677), if the context provides no cues towards the latter. But due to the fixed nature of idioms, even an unspecific context should result in an immediate activation of the idiomatic meaning (cf. e.g. Gibbs 1980). And Katz and Ferretti (2001: 213), who contrasted reading times for familiar and unfamiliar proverbs, conclude that unfamiliar proverbs are not read more quickly in a literally-biasing context. If context really invariably led the addressee to the meaning it favours, this result could not have been obtained (cf. also Kemper 1981). 109 cessing. It relies on the assumption that both the standard pragmatic model, with its strict separation of literal and figurative language, as well as the direct access view, with its emphasis on contextual influence, are oversimplified. 4.4.3 The entrenchment-based view: Conventionality, frequency, and language processing 4.4.3.1 The graded salience hypothesis Particularly with respect to polysemous lexical items it has been observed that not only the contextually appropriate meaning is initially activated, but also other, contextually incompatible ones. Initial processing has therefore been claimed to be exhaustive and unordered (cf. e.g. Onifer and Swinney 1981; Seidenberg et al. 1982). Others like Duffy, Morris and Rayner (1988), Rayner, and Frazier (1989), Sereno, Pacht and Rayner (1992), Simpson (1981), and Tabossi (1988) have found evidence indicating that lexical activation is generally exhaustive, but still sensitive to issues like the frequency of the respective meanings and thus propose a parallel processing model. Findings like these have led to the development of the graded salience hypothesis (cf. e.g. Giora 2003), a model of language processing which accounts for many types of linguistic ambiguities such as figurative language, idioms, jokes, irony, and innovations. It is founded on the following assumptions: 1. Salient interpretations are always activated first. 2. Arriving at less salient meanings invariably involves a sequential process, in which the more salient meaning is processed first and rejected. 3. Less salient interpretations are more difficult to derive and therefore need more and different contextual support. (after Giora 1997: 186) First, it is necessary to clarify what salience means within the framework of this model. 116 Importantly, salience is meant as a property of meanings as wholes (as opposed to the salience of single attributes, cf. Ortony et al.’s 1985 salience imbalance model). The salience of a meaning is fostered by several factors: codedness, frequency, familiarity, conventionality, and prototypicality (cf. Giora 1997: 185, 1999: 921, 2003: 15-17). In what follows, I will discuss the single aspects in turn. The first element is codedness. Salient meanings have to be stored in language users’ minds (Giora 2003: 15). 117 Retrieving a known meaning from the mental lexicon is easier 118 than inferring a new meaning from 116 For a discussion of other uses of this term, cf. chapter 4.2.2.1. 117 This ties in well with what has been pointed out regarding linguistically minimal meanings by Ariel (2002a). Cf. chapter 2.3. 118 Cf. also Gibbs and Matlock (2001: 235) who argue that even understanding stored senses requires more than just looking them up in the mental meaning store. 110 context. However, salience is not an all-or-nothing matter that only depends on the distinction between stored and unstored meanings. Not all senses collected in the mind are equally foregrounded. They differ, for example, with regard to their frequencies. If a word has more than one meaning, the different senses can have different probabilities of occurrence. On the one hand, it can be argued that frequency is largely an idiosyncratic matter. This more subjective aspect is described by the notion of familiarity. There is definitely a difference between the overall frequency of usage of a meaning by a whole speech community and the subjective, experiential familiarity of single persons with a given meaning (cf. e.g. Gernsbacher 1984). Low-frequency meanings can be highly experientially familiar and vice versa. Thus, for a shop-owner selling only Portuguese food and drink, the ‘wine’-meaning of port would certainly be more salient than the ‘harbour’-meaning. On the other hand, there are larger tendencies that concern the speech community as a whole, which can, for instance, be established by corpus-linguistic analyses. As corpora consist of large collections of texts belonging to distinct text-types, they allow us to study the frequencies of different senses of one lexeme on this more comprehensive level. The frequency of senses is crucial for their initial activation. More frequent meanings are activated sooner than less frequent ones (Hogaboam and Perfetti 1975: 271; cf. also Clarke and Nerlich 1991: 228; Récanati 1995: 212; Simpson 1981: 133; Wiley and Rayner 2001: 1016-1017; Williams 1992: 208). Moreover, frequent use has been quoted as the major factor leading to the conventionality of a given form-meaning pairing (cf. e.g. McGlone 1996: 545; Papafragou 1996: 193). Giora’s notion of conventionality is basically the same as the one advocated by Nunberg, Sag and Wasow (1994): a relation among a linguistic regularity, a situation of use, and a population that has implicitly agreed to conform to that regularity in that situation out of preference for general uniformity, rather than because there is some obvious and compelling reason to conform to that regularity instead of some other. (Nunberg, Sag and Wasow 1994: 492-493) Similar to frequency and familiarity, conventionality in Giora’s sense is not restricted to single lexical units, but can extend beyond word boundaries and include larger chunks of linguistic input which tend to be associated with a specified meaning (Giora 2003: 18-21). Gibbs (1982), for instance, mentions that the small syntactic difference between You are a fine friend and A fine friend you are leads to diverging interpretations. The latter construction is much more likely to be used to convey sarcasm than the former - a fact that is crucially related to the frequency with which it is used to convey this specific meaning (cf. e.g. Bybee 2006). 119 119 This is obviously related to construction grammar. For short introductions to the latter cf. e.g. Ungerer and Schmid (2006: 244-252) or Croft (2007). A more detailed account can be found in Croft and Cruse (2004: 225-327). The latter also discuss in more 111 The last factor contributing to the salience of a meaning discussed by Giora (2003: 17) is prototypicality (cf. e.g. Rosch 1973; Rosch 1975; Rosch and Mervis 1975). When hearing a word, people are more likely to think about prototypical examples than about marginal category members. For the stock example of the category BIRD this means that when someone hears the word bird, what comes to their mind first is the prototype of the category, in this case the robin, or at least one of the more central category members like the sparrow. 120 All these factors have to be regarded as continua. Meanings are not either frequent, familiar, conventional and prototypical or not, but more or less so. Taken together, they form the basis for the degree of salience of a meaning. As Giora (2003: 17) points out, the different aspects do not always correlate. A meaning can be infrequent, but familiar to a given person. Or a less prototypical meaning might be more frequently activated by a person than the more central members of the same category. From the subjective perspective of individual language users, this is certainly true. Personal experiences will decide upon the salience of a given meaning and thus also upon its processing. And because familiarity is the factor which reflects this subjective component best, it is said to be the most significant one (2003: 17). Considering that the number of subjects taking part in psycholinguistic experiments is usually restricted to thirty or forty, it is not astonishing that psycholinguists emphasize the individual differences between speakers, because they can heavily influence their results. However, if one approaches the question of salience with corpus-linguistic methods, as will be done in chapters 5 to 7, the subjective aspects lose much of their importance. The overall frequency of, for example, a metonymic meaning of a word counted in a large corpus is undeniably informative about the salience of this meaning in the speech community at large. Thus, frequency emerges as the decisive factor if one is interested in larger linguistic and detail the so-called usage-based model (2004: 291-327), which underlies Bybee’s work (e.g. 2006; Bybee and Eddington 2006; Bybee and Hopper 2001). It demonstrates that the frequency of use has a considerable impact on the mental representation of formmeaning pairings and is the major factor leading to grammaticalization. 120 There seems to a difference of status between frequent, familiar and conventional meanings on the one hand and prototypical meaning on the other. Prototypical meaning is more concerned with the internal structure of cognitive categories and reference (in Lyons’s 1977 sense), while the former notions deal with linguistic meaning in the narrower sense. For the prototype cases mentioned by Giora (2003), speakers can easily avoid potential ambiguities by using the subordinate category terms. This is related to onomasiological salience as described by Geeraerts, Grondelaers and Bakema 1994: esp. 134-146; cf. also Geeraerts 1993b and the discussion of salience in chapter 4.2.2.1). Avoiding the potential ambiguity is often not as easy for Giora’s examples of salience due to frequency (e.g. bank ‘side of a river’ vs. ‘financial institute’) or conventionality (e.g. direct requests, such as Pass me the salt, are considered impolite and therefore not used very often). 112 cognitive regularities. This is reminiscent of Langacker’s (1987, 2000) notion of entrenchment (cf. chapter 4.2.2.1): Entrenched linguistic structures are processed automatically and require no cognitive effort, because psychological events generally leave “some kind of trace that facilitates their re-occurrence” (Langacker 2000: 93). The more frequently we use a word with a specific meaning (as compared to the frequency of use of other possible meanings of the same word; cf. Geeraerts, Grondelaers and Bakema 1994), the easier it becomes to retrieve this meaning from semantic memory. One has to bear in mind, however, that the exact relationship between frequency and entrenchment is not yet completely understood by (psycho)linguists, as mentioned in chapter 4.2.2.2. Nevertheless, the term entrenchment with its implicit metaphor of well-trodden mental paths seems to be particularly apt to describe the issue in question. Meanings which are deeply entrenched are foremost in our minds, and dominant (cf. Brisard, van Rillaer and Sandra 2001; Duffy, Morris and Rayner 1988), and thus immediately activated upon encountering the corresponding lexical form. Sticking to the terminology introduced in chapter 4.2.2.1, I will use the term entrenchment instead of salience to refer to meanings which are activated easily and immediately due to their frequent usage. The former notion captures the essence of Giora’s ideas quite well and has the additional advantage that it helps to avoid any confusion that could otherwise arise from the various meanings of the word salience. Whereas salience either defines a current activation state or the attention-attracting potential of real-world entities (cf. Schmid 2007), entrenchment is reserved for a description of the continuum ranging from entirely novel, unfamiliar, and infrequent meanings to highly conventional, familiar and frequent ones. 4.4.3.2 Processing well-entrenched and less entrenched meanings Giora and Fein’s (1999) experiments prove that familiar, i.e. relatively entrenched metaphors, are processed differently from less entrenched ones. 121 The degree of entrenchment of the metaphors used in the study was established in pre-tests, by way of ratings on a seven-point scale. Accordingly, the metaphors were assigned to three groups: familiar, less-familiar, and unfamiliar. In the actual study, subjects read short texts consisting of a contextual passage and a target sentence. The contexts were either biased towards the literal or the metaphorical meaning of the target sentences: (60) Literally biasing context: The Saturday night party went on for hours. Drinks were poured, and we danced all night. We were probably less than inconsiderate when, the 121 Long before the development of Giora’s theory, researchers like Gregory and Mergler (1990) and Blasko and Connine (1993) pointed to the relevance of entrenchment for automatic metaphor understanding. Cf. also Marschank, Katz and Paivio (1983), and Katz et al. (1988), who mention familiarity as one of their ten psychological dimensions relevant for metaphor research. 113 next evening, we called on our friends who had been partying with us. When they opened the door we realized: Metaphorically biasing context: A bloody war has been going on in central Europe for a few years. Thousands of innocent women, men, and children got massacred, and no one budged or lifted a finger. At last, a decision was made to intervene in the fights. Target: Only now did they wake up. (after Giora and Fein 1999: 1605) Each subject read only one of the versions of the stories. After having read the targets, the subjects were presented with two word fragments, one related to the literal meaning (‘rise’) and one to the metaphorical meaning (‘act’) 122 and asked to complete them “with the first words that came to mind” (Giora and Fein 1999: 1606). Word-fragment completion tests are used to test the activation of meanings. Completing a word fragment, such as -epre-se- (depressed), is difficult without a context, but facilitated as soon as an appropriate context is given (Giora 1997: 185). Entrenched metaphorical meanings are activated not only in the metaphorically biasing, but also in the literally biasing context (Giora and Fein 1999: 1607-1610). Less entrenched metaphorical meanings, however, were hardly activated at all in the literal context condition. Moreover, even the metaphorical context did not lead to an exclusive activation of the compatible meaning. On the contrary, the contextually incompatible literal meaning was activated more than the appropriate metaphorical meaning. Context does not entirely determine which meanings are activated. Entrenched meanings are always initially activated even if the context is heavily biased towards another reading. As opposed to entrenched metaphors with at least two entrenched meanings, the literal and the metaphorical one, the figurative meanings of less entrenched metaphors are not stored. Only the literal meanings transported by the same lexical surface structures are. The latter meanings should thus be the only ones that are initially activated. Evidence from reading-time studies suggests that this is indeed the case (cf. Giora 1999: 922, 2003: 108). Whereas entrenched metaphors take about the same time to read as their literal interpretations, less entrenched and non-entrenched metaphors are read more slowly in metaphorically biasing contexts than in literally biasing contexts. The delay observed for the metaphorically biasing contexts is an effect of the initial activation of the entrenched, literal meaning. This contextually incompatible meaning has to be rejected 123 and 122 Originally, the experiment was conducted in Hebrew. This is why the glosses given here are not identical with the fragmented test words used by Giora and Fein (1999). 123 Giora and her collaborators (e.g. Giora 1997: 185; Peleg; Giora and Fein 2001: 176) propose that contextual processes run parallel to lexical ones. However, the two are said not to interact in the early processing phases. For a more detailed account of the 114 an alternative interpretation has to be searched for on the basis of the information provided by the context. Further support for the entrenchment-based view comes from studies which also investigated the processing of entrenched vs. non-entrenched metaphorical meanings. Blank (1988), for example, studied the comprehension of conventional metaphors as defined by Lakoff and Johnson (1980) and compared their processing to novel metaphors and literal and anomalous control sentences. The results of his lexical naming task 124 show that subjects did not find it more difficult to name the last word of highly conventional metaphors, such as spend your weekend (an instance of TIME IS MONEY ), than to come up with the literal counterparts (e.g. spend your income). For novel metaphors like I heard the thunder gossip, participants took more time to pronounce the targets. Blank (1988) concludes that the conventional metaphors have become part of the average language user’s mental lexicon, and are therefore understood by way of selecting the stored sense from the mind. The unfamiliar metaphors, however, require an additional stage, in which subjects look for possible sense extensions, i.e. derive the intended meanings. 125 As already mentioned, Blasko and Connine investigated how familiar and less familiar metaphors are processed. In pre-tests they had established the familiarity of their metaphors as well as their aptness. Metaphor aptness or metaphor goodness (cf. e.g. Johnson and Malgady 1979, 1980), depends on between-domain characteristics as well as within-domain features. On the one hand, good or apt metaphors rely on a connection between two distinct and semantically moderately distant domains (cf. e.g. Katz 1989: 494; Kittay 1987: 292) in order to fulfil their explanatory function. On the other hand, there must be some properties, characteristics or parts that are similar or can be regarded as similar in the two domains (cf. e.g. Johnson and Mal- influence of context on the processing of entrenched meanings, cf. Peleg, Giora and Fein (2001). As a response to Vu, Kellas and Paul’s (1998; cf. also Martin et al. 1999) findings they showed that for novel metaphors that are presented in sentence-initial position entrenchment effects cannot be outweighed even by strong contextual bias. Only in sentence-final position were contextual effects observed to be slightly faster than entrenchment effects, as the contexts were so predictive that the appropriate meanings were activated, i.e. cognitively salient, before the relevant stimulus was actually read. 124 In this task, subjects were presented with the linguistic stimuli on a computer screen. First, the sentences were shown with the actual targets replaced by a dot. After two seconds the partial sentence disappeared and the target words were presented. Subjects were asked to pronounce these words into a microphone as soon as they had recognized them (Blank 1988: 26). 125 Blasko and Connine (1993: 296) criticize Blank’s (1988) work, since his highly familiar metaphors “were stock phrases and not really metaphorical”. But while some of the metaphorical sentence completions may have been highly predictable, they were not necessarily more predictable than the literal targets (e.g. We’ve been robbed by inflation vs. We’ve been robbed by burglars). 115 gady 1980: 272). 126 Or as Tourangeau and Sternberg (1981: 27) put it: “Metaphors are [...] more apt to the extent that their terms occupy similar positions within domains that are not very similar to each other”. Blasko and Connine (1993) established that metaphor comprehension is not only influenced by the entrenchment of the metaphorical meanings, but also by the degree of aptness of a metaphor. In their second experiment, the targets were presented with a delay of 300 ms (after the metaphorical vehicle had been shown). Although the general trend was the same as in the first study, i.e. the degree of familiarity led to differences in reading times, there were additional effects that could be traced back to aptness. All the highly familiar metaphors displayed facilitation, no matter whether they were particularly apt or not. For the less familiar metaphors, however, only the highly apt ones exhibited activation, while the moderately apt ones showed inhibition (Blasko and Connine 1993: 300). The meanings of the metaphors rated as low-familiar and moderately apt were available only 750 ms after the offset of the metaphor vehicle (1993: 303). Examples are given in (61) to (63): (61) High-familiar: The belief that HARD WORK IS A LADDER is common to this generation. 127 (62) Low-familiar - High apt: Aunt Mabel showed the boys how A DUSTY AND CROWDED ATTIC IS A PARADISE that lazy Saturday afternoon. (63) Low-familiar - Moderate apt: The thought that A GOOD PROFESSOR IS AN OASIS was clung to by the entire class. (Blasko and Connine 1993: 307) Aptness thus “may play an especially important role in determining the ease of comprehension in less familiar metaphors” (Blasko and Connine 1993: 300). If a metaphorical construal is apt, even less familiar metaphorical meanings can be understood rapidly. A full account of metaphor comprehension hence has to embrace both factors, entrenchment and aptness. Unfortunately, Blasko and Connine do not indicate which and how many of their highly familiar metaphors had been rated as moderately apt. 128 But it seems reasonable to assume that more apt metaphors also stand a better 126 Moreover, the salience of properties or attributes linked by the metaphor have been said to play a role here (cf. e.g. Ortony 1979; Ortony et al. 1985; Johnson and Malgady 1979, 1980). As this issue is of importance for my later argument, it will be discussed separately; cf. esp. chapters 6.2.1.4, 6.3, 7.4. 127 Note that the familiarity of the example hard work is a ladder can also be related to participants’ knowledge of the entrenched conceptual mapping GOOD IS UP , or the more specific mappings CAREER PROGRESS IS VERTICAL MOVEMENT and STATUS IS POSI- TION (cf. Lakoff, Espenson and Schwartz 1991). 128 The metaphors were originally taken from Katz et al. (1988), but Blasko and Connine (1993) did their own aptness and familiarity ratings, because some of the metaphors had been shortened. Thus Katz et al.’s ratings cannot be used to judge the relationship of familiarity and aptness in Blasko and Connine’s study. 116 chance of being used more frequently so that the number of less apt highly familiar metaphors should be relatively low. 129 While the entrenchment-based view cannot explain why more apt metaphors are easy to understand even if they are unfamiliar (cf. Giora 1997: 191-192), it can reconcile many of the contradictory findings concerning metaphor comprehension. As entrenched meanings are always activated, literal and metaphorical meanings are activated if both are stored. If only the literal meaning is well-entrenched, the comprehension of a figurative meaning takes place in a stage-like fashion and is constrained by contextual influences in the later integration phase. Context, however, can also play a role initially. It can, in a way, predict metaphorical meanings, which are then only activated as a result of spreading activation which leads to a higher cognitive salience of the concepts and meanings in question. Only some studies explicitly address the comprehension of metonymic expressions, but their findings are largely in line with the entrenchmentbased view. 130 For example, it can explain why Gibbs (1990; cf. chapter 4.4.1) observed differences between the comprehension of literal and figurative referring expressions (cf. Giora 1997: 189-190). As his metonymic target expressions (e.g. glove), have well-entrenched literal meanings, whereas the intended figurative ones are not stored, the evocation of the figurative meanings takes longer. Also the fact that his metonymic references were more difficult to comprehend than the metaphorical ones (cf. Gibbs 1990) can be accounted for. Since the literal meaning of glove is potentially contextually relevant in the boxing context used by Gibbs, the literal meaning might have been contextually favoured and already been part of the subjects’ focus of attention before the actual metonymy occurred. When encountering the word glove, subjects thus might have found it difficult to reject this cognitively salient meaning, which is necessary to comprehend the referential meaning ‘baseball player’. For metaphorical referential expressions this is not the case, because metaphors are based on the meaningful connection of two distinct domains. Accordingly the literal 129 Although many studies focus on metaphor aptness (e.g. Katz 1989; Tourangeau and Sternberg 1981; Johnson and Malgady 1979, 1980) none of them considers the interaction of aptness and familiarity. The only source available which supports the assumption that ‘better’ metaphors are often at least felt to be more familiar than ‘poorer’ ones, is Katz et al. (1988). In their lengthy appendix they list 464 metaphors together with the results of their rating studies. As a closer look at their data confirms, the mean values of the felt familiarity ratings are often not very different from the mean values of metaphor goodness (e.g. A pond is nature’s mirror: Aptness: 5.0, Familiarity: 5.0), but there are also cases where the mean values differ markedly (e.g. Billboards are the yellow pages of the highway: Aptness: 5.03, Familiarity: 3.10). 130 My discussion is limited to the results of studies addressing the processing of referential metonyms. Further work has been done on how logical metonymy (e.g. to begin Dickens for ‘to begin reading Dickens’) is understood; cf. e.g. Lapata, Keller and Scheepers (2003), McElree, Frisson and Pickering (2006), Traxler, Pickering and McElree (2002). 117 meaning of the vehicle term can hardly become cognitively salient due to contextual influence. Metonymies are not, however, invariably more difficult to comprehend than metaphors, since the literal meaning is not rendered cognitively salient in all cases. Consider for example a text on the political relations between the United States and Great Britain published in a newspaper. The overall context of the domain of POLITICS would bias rather the metonymic reading of White House than its literal sense. Thus, the literal meaning would not be cognitively salient and not interfere with the comprehension of the metonymic meaning. Moreover, metonymic meanings can - just as other types of figurative meanings - be more entrenched than the literal meanings of the same expressions. For White House, for instance, the literal meaning occurs significantly less frequently than the metonymic BUILDING FOR INSTITUTION sense, which accounts for 72.1% of 1000 instances of White House in the BNC. 131 The metonymic meaning should be foremost in speakers’ minds and cause no processing difficulty. 132 While the BUILDING FOR INSTITUTION meaning of White House is deeply entrenched and does not need any special contextual clues to be activated, ‘baseball player’ as a meaning of glove is probably not stored in the mind of the average English speaker and needs contextual support. In Gibbs’s (1990) example We need a new glove at third base the metonymic meaning of glove can, if no context is provided, only be comprehended after the mention of the noun base in the prepositional phrase. Only then is the right domain evoked, which allows a re-reading of glove in terms of a COSTUME FOR WEARER metonymy. This aspect is also discussed by Frisson and Pickering (1999, 2001), who emphasize that many figurative expressions are not comprehended by way of activating a stored figurative meaning: “[...] a number of expressions derive their figurative interpretation not on the basis of the activation of a figurative sense of a word, but rather through the combination with other lexical elements in the same sentence” (Frisson and Pickering 2001: 153). This holds not only for attributive metaphors, like Some jobs are jails, but also for unconventional, non-entrenched metonymies, like glove. 133 Their experiments support the view that entrenchment matters with regard to 131 For a more detailed account of the frequency of different metonyms, cf. Appendix A and chapter 6. 132 Certainly, one can imagine situations in which the less entrenched literal meaning of White House could interfere with the comprehension of the metonymic sense, for example, if speaker and addressee of an utterance beginning with The White House ... are standing in front of the building at the time of speaking. But in more neutral situational contexts, which would undoubtedly surround the majority of utterances containing the expression, such an interference effect is unlikely. 133 Cf. also Pickering and Traxler’s (1998) results on metonymy processing. A biasing context can lead to a rapid metonymic interpretation of professor as ‘statue of a professor’ in a sentence like The janitor was polishing the professor as it influences the plausibility of this interpretation. But the effect did not occur when the word professor had been encountered before in a context biasing its literal reading. 118 metonymy comprehension. Frisson and Pickering (1999) investigated PLACE FOR INSTITUTION and PLACE FOR EVENT metonymies and established that unfamiliar metonymic senses require more effort to be understood than familiar ones even when the preceding context clearly requires a metonymic interpretation (1999: 1369-1377). 134 Thus, stadium in (64) is harder to understand than convent in (65). (64) That blasphemous woman had to answer to the stadium at the end of March, but did not get a lot of support. (65) That blasphemous woman had to answer to the convent at the end of March, but did not get a lot of support. (Frisson and Pickering 1999: 1369) The results of their eye-tracking study 135 are compatible with a parallel model of language processing also proposed by the entrenchment-based view. Literal and familiar metonymic senses are considered at the same time (1999: 1373). Thus entrenched metonymic meanings are “every bit as straightforward to process as [their] literal control[s]” (1999: 1377). Frisson and Pickering’s work is based on Frazier and Rayner’s (1990: 182) “partial specification hypothesis”, which holds that people do not necessarily commit to one of the possible meanings of a word if there are not enough semantic clues present. Instead they work with an underspecified semantic assumption that does not correspond to any specific sense, but is compatible with all possible senses (cf. Pickering and Frisson 2001: 158). Accordingly, words initially only receive a minimal interpretation which is in line with any of the possible stored meanings, until there is enough (contextual) evidence to guide the addressee to the right, more specific interpretation (cf. also Frisson 2009). Frisson and Pickering (1999, 2001, 2007) acknowledge that there is a difference between stored senses (what they call ‘familiar’) and novel interpretations. Stored meanings are entrenched and hence easier to deal with than the novel ones. However, Frisson and Pickering see no evidence of an ordered activation of specific senses as suggested by the entrenchmentbased view (2001: 161) and “no support for the hypothesis that the frequency of the senses influences the reading times” (1999: 1376). But as Giora (2003: 118) points out, Pickering and Frisson’s findings do not rule out 134 In their later study of PRODUCER FOR PRODUCT metonymies, however, Frisson and Pickering point out that “the presence of an appropriate context, in combination with a metonymic rule [i.e. if a conventional metonymic mapping exists that supports the linguistic metonym; SH], can facilitate the processing of an unfamiliar metonym” (2007: 605). 135 Eye-tracking studies monitor the position of subjects’ eyes during reading. The underlying assumption is that “the point at which a person is looking is extremely closely linked to the thing they are currently ‘thinking about’” (Pickering and Frisson 2001: 159). 119 the possibility of an initial activation of all salient senses. 136 What they do, however, is cast doubt on the assumption that the continuum ranging from deeply entrenched to less or not at all entrenched senses directly influences processing - with more frequent senses being more entrenched and thus easier to process than less entrenched, but stored ones. The processing difference would be reduced to the distinction of stored vs. unstored meanings. But how reliable are Frisson and Pickering’s (1999) frequency data on which the above claim is based? The frequencies of the literal and metonymic meanings of the targets used (e.g. convent, stadium) were established through an analysis of the British National Corpus (BNC). But the corpus analysis on which Frisson and Pickering’s (1999) results are based was restricted to only 55 randomly selected instances of the target words. 137 These 55 occurrences were then classified as either metonymic or literal (metonymic senses other than those intended in their examples were ignored). One can thus call into question whether results which are based on such a small number of items really reflect the overall relative frequencies of the two meanings. Regrettably, they do not mention how often the single lexemes analyzed occurred in the two different meanings. 138 Otherwise it would have been possible to verify Frisson and Pickering’s claims. Given that their samples from the BNC might have been too small to be representative of the relative frequencies of the different senses, it is also unclear whether it is justified to argue that there is no reliable interaction between relative sense frequencies and reading times. Nevertheless, Frisson and Pickering’s results largely support the entrenchment-based view. Both stadium and convent (cf. examples 64, 65) were studied as instances of the mapping LOCATION FOR INSTITUTION . Whereas convent was found to be a familiar metonymy and understood effortlessly, stadium was considered unfamiliar and caused processing difficulties. An extensive corpus analysis of the two lexemes in the BNC revealed that con- 136 Actually, Giora’s (2003: 118) argumentation is not very convincing and she readily admits that her ideas are hard to prove. She bases her account of the findings largely on verbal metaphors (e.g. to disarm a critic) and holds that the fact that the contextually incompatible meaning is retained until the later stages of processing as a part of the underspecified meaning could possibly be explained by her ‘retention hypothesis’, i.e. because the literal meaning can be useful for the metaphorical interpretation. However, Frisson and Pickering are mainly concerned with nominal metonymies, and she does not comment on how her ideas can be transferred and applied to nominal metonymic references. 137 In Frisson and Pickering (2007), sense frequencies are established on the basis of an even smaller number of instances in the BNC. Only 25 instances of each of the 24 names used in the experiment were considered (2007: 601). 138 The numbers given in Frisson and Pickering (1999: 1372) are calculations which reflect how often the different meanings of all the words analyzed occurred in their sample. It is not possible to extract the number of occurrences of a specific sense of a single word from these figures. 120 vent is indeed much more frequently used as a metonym than stadium. For convent, 12.60% of 1000 instances are metonymic and all of them instantiate the mapping LOCATION FOR INSTITUTION (cf. Appendix A). For stadium, only 5.45% of the uses rely on metonymic mappings. The LOCATION FOR INSTITU- TION pattern Pickering and Frisson (1999) did not find in their sample accounts for 2.72% of all cases, 139 whereas the remainder of the metonyms is motivated by other mappings. While the LOCATION FOR INSTITUTION sense of stadium is not as novel as Pickering and Frisson suggest, it is clearly less frequent than the ‘institution’ meaning of convent. The ‘institution’ meaning of convent can be assumed to be more entrenched than the ‘institution’ meaning of stadium. So it is not surprising that the former is easier to comprehend than the latter. On the basis of my analysis of the frequencies of the different senses of the two words in the BNC, it cannot be entirely excluded that participants had encountered the ‘institution’ sense of stadium before. However, it is certainly not the most deeply entrenched meaning and can be harder to comprehend than the literal meaning which is much more frequent. The entrenchment-based view does not argue that there is a processing difference among well-entrenched meanings. All of them are invariably activated, while the activation of less entrenched meanings requires additional cognitive effort. The ‘institution’ sense of stadium does not seem to be entrenched well enough to allow such a direct processing. Two factors are of potential relevance in this respect: On the one hand, that stadium is more frequently used to convey a literal meaning than as an instantiation of the mapping LOCATION FOR INSTITUTION , as Pickering and Frisson (1999) also acknowledge. On the other hand, and this is neglected by Pickering and Frisson, the ‘institution’ meaning targeted in the study has to compete with other metonymic senses of the same word (cf. Appendix A), which are not too different from the metonymic ‘institution’ meaning in terms of their relative frequencies, i.e. entrenched to a similar degree - a fact which might very well also play a role in terms of processing. This is, as noted above, different for convent, where all the metonymic uses I found in the BNC rely on one and the same mapping. That the ‘institution’ sense of convent is also less frequent than its literal meaning does not necessarily matter. It seems well-entrenched enough to be activated immediately. Further evidence supporting the main ideas of the entrenchment-based approach is provided by Gerrig (1989). His results concerning metonymy comprehension (cf. chapter 4.4.1) strongly favour the view that lexical ac- 139 In my more detailed analyses (cf. chapters 5 and 6, and Appendix A), the instances counted here as belonging to the pattern LOCATION FOR INSTITUTION form part of two different categories: LOCATION FOR INSTITUTION , which comprises the ‘sports club’ sense only and accounts for 0.84 % of all instances, and LOCATION FOR PEOPLE , which embraces the ‘audience’ sense and the ‘stadium management’ sense. The latter sense, which occurs in 1.88 % of the cases, could have been subsumed under the LOCATION FOR INSTITUTION sense as well and is therefore considered relevant here. 121 cess is ordered, with more deeply entrenched meanings being activated first. Of major interest with regard to his findings is that the degree of familiarity and accessibility of a stored meaning interferes with the ease of activation of unconventional metonymic senses (Gerrig 1989: 203). This indicates that there is an interrelation and interaction between meanings with different degrees of entrenchment. Although psycholinguists have not studied metonymy as extensively as metaphor, the evidence points to some significant similarities between the comprehension of both. Metonymies can and do - just like metaphors - differ with regard to their entrenchment. Novel and unfamiliar metonymic expressions are in general more difficult to process than established and stored metonymic senses. Contextual factors, however, can make them quite accessible to addressees. For familiar metonymies, which come with one (or more) deeply entrenched metonymic meaning(s), one can expect that either both the literal and the metonymic meaning are initially activated or, if the metonymic sense is more entrenched than the literal sense, that the metonymic reading is activated first (before the literal meaning is activated). The only difference between metaphor and metonymy seems to be that metonymies, which by definition consist of a within-domain relationship, are possibly more likely to lead to an initial activation of an underspecified meaning, which is specified only as soon as enough contextual clues are available. However, considering that there are linguistic expressions which are used with a metonymic rather than a literal meaning in more than half of the cases, such as BMW, Crown, or drugs (cf. Appendix A), language users thus have reason to believe that these metonymic meanings, which are so familiar to them, are the ones at issue when they encounter the corresponding linguistic expressions. It is unclear why they should prefer a weak and unspecified semantic assumption to a much more informative one, which proves useful in most of the cases. 4.5 Summary The human mind is not necessarily biased towards literal thinking, but accepts nonliteral construals as equally natural. This is, for example, reflected in the fact that ‘good’ or apt figurative expressions are often dealt with effortlessly, even if they are largely unfamiliar. The influence of context has been shown to be limited. Strong contexts can render certain concepts more readily available by increasing their cognitive salience. This can lead to a relatively smooth and quick understanding of less entrenched meanings. However, even heavily biasing contexts cannot prevent the initial activation of contextually incompatible, but deeply entrenched meanings. We cannot avoid activating those meanings which we encounter most frequently, even if they might interfere with the semantic content currently 122 at issue. Conventionality definitely constitutes a psychological reality and is more important in terms of language processing than the distinction of literal and nonliteral language. The chapter has shown that there are several factors which either contribute to the conventionality of figurative thinking and speaking or can be used to judge the conventionality of figurative language and thought. The first and most vital of these factors is frequency. This has become apparent in the theoretical considerations of the process of conventionalization in traditional and cognitive linguistics (cf. chapter 4.2), as well as in discussion of the results of psycholinguistic tests (cf. chapter 4.4). Frequency can thus be used to measure conventionality. Among the factors which foster the conventionality of metaphors and metonymies, ontological salience (cf. chapter 4.2.2) plays a significant role by predicting the preferred direction of mapping. However, since not all linguistic instantiations of a given mapping necessarily display the same degree of conventionality (cf. chapter 4.3.2.3), it is necessary to distinguish between conventionality on the conceptual and on the linguistic levels. Which linguistic exploitations of conceptual mappings stand a good chance of becoming established depends on cognitive as well as more pragmatically oriented factors. The latter have to do with the acceptability of a figurative construal, which is in turn dependent on the name-worthiness and relative stability of the category referred to by way of the corresponding linguistic expressions. The former are concerned with the structure of the source (for metaphors) or vehicle (for metonymies) concepts. Many of these assumptions are further supported by the classifications of metaphors and metonymies introduced in chapter 4.3. Ruiz de Mendoza’s (e.g. Ruiz de Mendoza and Díez Velasco 2002, 2004) typology of metonymies, for example, is in accordance with the prediction that ontological salience influences the conventionality of metonymies. It is therefore worth looking at the relationship between the relative frequencies of metonymic meanings and their either being in line or at odds with the laws of ontological salience. Nunberg’s (1995) classification of metonymies corresponds to the criteria of name-worthiness and referent-category stability: While occurrent metonymies are not likely to become conventional, predicate transfer is the main source for metonymically motivated polysemy. This can also be verified by comparing the relative frequencies of examples of occurrent metonymies to those of examples of predicate transfer. This will as well indicate if Barcelona’s (e.g. 2003, 2004) assumptions concerning conventionality are right or whether his typology of metonymies (cf. chapter 4.3.2.3) is indeed too crude to explain the issue of conventionality. The fact that not all instances of a given mapping are automatically similar in terms of their conventionality has become apparent in several typologies (e.g. Schmid 1993; Barcelona 2003, 2004; Croft 2006), mostly in terms of the rather fuzzy phrase ‘social sanctioning’. Since social sanctioning seems to play a vital role, it is worth trying to arrive at a better idea of what this 123 term actually signifies by looking at how the structure of the concepts involved in a mapping is related to the speech community’s preferences. The conventionality of metaphors seems to be largely dependent on the concrete-to-abstract direction of mapping and on the degree of specificity of the mapping (cf. chapter 4.3.1.1). That the structure of the source concept also influences the ‘social success’ of a mapping and its linguistic instantiations has become evident in Traugott’s (1985a) approach (cf. chapter 4.3.1.2). However, as has been noted in chapter 4.4.2, it is not enough to concentrate on the conceptual mapping only, since not all possible linguistic realizations of a mapping might be equally accepted by speakers. Moreover, as the discussion of Bowdle and Gentner’s approach (2005; cf. chapter 4.3.1.2) has shown, metaphorical meanings can be enriched by emergent structure and thereby acquire a certain distance to the underlying mapping. The following chapters will be devoted to an empirical investigation of many of these predictions and hypotheses. Using a corpus analysis of the relative frequencies of metaphorical and metonymic meanings in the BNC, I will address the question of which figurative construals are preferred by the speech community and which are dispreferred, both on the more comprehensive level of conceptual mappings as well as on the level of single linguistic instantiations. The empirical part of this thesis begins with a chapter on methodology, where I discuss some general problems of corpus-linguistic approaches to figurative language (cf. chapter 5.1.1), before introducing the methods and principles I used in my study and outlining how problematic cases are dealt with (cf. chapters 5.1.2 to 5.2.2.4). The results of the study are then presented in chapters 6 and 7 and related back to the criteria that emerged from the theoretical considerations, such as ontological salience and conceptual structure. 125 5 Investigating the conventionality of figurative language With the exception of the psycholinguistic results dealt with in chapter 4.4, most discussions of figurative language in general and the conventionality of figurative language in particular are entirely theoretical. Empirical studies are scarce. In what follows, I will put the predictions made in the theoretical considerations of the conventionality of figurative language to the test. The crucial factor in assessing the conventionality of figurative language and thinking is frequency: It is related to the entrenchment of linguistic structures and reasoning patterns and indicates which metaphorical and metonymic meanings can be processed effortlessly. Even if there is no reliable correlation of frequency and cognitive centrality (cf. chapter 4.2.2.2), it is nevertheless of interest why some mappings are more often applied in language than others. Moreover, extensive frequency analyses make it possible to judge single linguistic instantiations against the backdrop of the conceptual mappings. The question is whether there are exceptions to the rule, i.e. whether there are single linguistic instantiations whose relative frequencies differ markedly from other instances relying on the same mapping. A thorough consideration of such exceptions should yield good clues about the conceptual prerequisites of the conventionalization of figurative expressions. As an objective description of the conventionality of figurative language and thought thus requires a large-scale study of the relative frequencies of figurative expressions and conceptual mappings, the best way to obtain a clear and fairly objective picture of the frequency of figurative language is, obviously, a corpus study. Corpus-based accounts of metaphor are no longer exceptional and have produced a range of interesting results concerning fields like grammatical patterns or the collocational range of metaphors (cf. e.g. Deignan 2005; Stefanowitsch and Gries 2006). Metonymy, again, has, all in all, not attracted as much attention. A notable exception is Markert’s work (e.g. Markert and Nissim 2003, 2006; Nissim and Markert 2005). Even though corpus-linguistic studies seem to be predisposed for a consideration of the conventionality of metaphors and metonymies, this issue is seldom addressed explicitly. If it is addressed at all, the relevant studies are mostly restricted to one or only a few linguistic items (cf. e.g. Hilpert 2006; Hanks 2006). However, this is not enough to get an overall picture of the conventionality of different conceptual mappings and their linguistic instantiations. As reliable frequency data of metonymic and metaphorical meanings are not available, a large-scale corpus study based 126 on the British National Corpus was conducted. Before I can turn to the results of this study and to their implications for the conventionality of metonymies and metaphors, a few comments on methodological issues in corpus-linguistic studies of figurative language are in order. 5.1 Figurative language and corpus linguistics: Methods and problems 5.1.1 General considerations The growing interest in corpus-based studies of figurative language is rooted in the justified assumption that an adequate description of the role of metaphor and metonymy in language and thought has to be based on systematic analyses of representative samples of language use. Many of the central publications on metaphor and metonymy are, however, based on linguists’ intuition rather than on systematic empirical studies. A case in point is probably the largest compilation of metaphors, the Master Metaphor List (Lakoff, Espenson and Schwartz 1991). It lists mappings and examples on 212 pages and all of them are said to be conventional metaphors which structure our thought and determine our actions. When reading through this list, however, one cannot help but feel that some of the examples are more conventional than others. To be shopping for a better philosophy (Lakoff, Espenson and Schwartz 1991: 61) seems, at least at first glance, to be a more obviously ‘figurative’ expression than He’s lost his way. The question is thus whether the mappings proposed and frequently discussed in the literature are really as important as the Lakovians would have us believe. Semino (2006), for example, observed that the two communication metaphors which loom large in the literature on conceptual metaphor, i.e. the CONDUIT metaphor and ARGUMENT IS WAR , do not even account for 50% of all metaphorical construals of communication in British English. Furthermore, researchers have found mappings which had been hitherto overlooked. Markert and Nissim (e.g. 2006: 163; cf. also Nissim and Markert 2005), for example, established in their study of organization names the metonymic pattern ORGANIZATION FOR INDEX (e.g. Eurotunnel was the most active stock), where the name of an organization stands for the organization’s shares - a pattern which is not mentioned in any other source. Systematic con sider ations of real life data thus have two undeniable advantages: 127 They provide us with a more comprehensive catalogue of existing conceptual mappings and corresponding linguistic expressions and they help us find out which mappings are really deeply entrenched ways of thinking and which of them perhaps have a more marginal status. But there is still the issue of representativeness. Even the largest corpora can never be really representative of a language as a whole. While all of the major available machine-readable corpora, such as the BNC or the Bank of English, try to incorporate as many text-types as possible, some text-types are unquestionably better represented than others. The inclusion of newspaper articles or books is, from a purely practical point of view, much easier and less time-consuming than the incorporation of spoken language data. Most corpora are therefore biased towards written data and even within the written components not all text-types are equally well represented. Another shortcoming of corpora is related to the fact that most of them, such as the BNC, are of finite size. 140 They are static. Once the intended size has been reached, they are no longer updated. This means that they do not cover innovations and gradually occurring changes in the language (Deignan 1999: 196). While machine-readable corpora are limited by reflecting only a restricted and somehow arbitrary subset of the language, they are nevertheless more representative than invented data can ever be (cf. Deignan 1999: 196-199). The main problem involved in corpus-based studies on figurative language is the identification of the relevant data within the corpus (Stefanowitsch 2006a: 1). Corpora annotated for conceptual mappings hardly exist 141 so it is up to the researcher to find a method of extracting the data. Various methodologies have been proposed (cf. Deignan 2005; Stefanowitsch 2006a). Only some of them can be mentioned here: Some early studies rely on a manual search for figurative expressions by reading carefully through texts (e.g. Jäkel 1997). More recent studies either search for source domain (e.g. Deignan 1999) or target domain vocabulary (e.g. Stefanowitsch 2006b). Others (e.g. Martin 2006) combine the two strategies and look for sentences which contain vocabulary from both domains. No matter which way is chosen, the actual identification and subsequent categorization of metaphors and metonymies will always be based on the researcher’s intuition to a considerable extent (Stefanowitsch 2006a: 10; cf. also Deignan 1999). In clear-cut cases this does not pose a problem, but there are invariably instances which are difficult to categorize. 142 Dif- 140 Exceptions to this are the so-called monitor corpora, like the Bank of English. They are open-ended, because new texts are constantly added to them. 141 A notable exception is the Mascara corpus. It is based on extractions from the BNC and annotated for metonymic mappings which involve location names. Cf. the relevant articles by Markert and Nissim (e.g. 2002b, 2003). The scope of the corpus is, however, limited. 142 See chapter 5.2 for a discussion of some of the problematic cases encountered in my own study. 128 ferent methods have been proposed to avoid subjective decisions. All of them are, however, problematic if applied to large sets of authentic data. With regard to corpus-based research, Stefanowitsch (2006a: 11) argues that less clear-cut instances should be judged by more than one person and then tested for inter-rater reliability (cf. also Low 1999; Markert and Nissim 2003). Such a procedure would certainly lead to maximally objective classifications. However, due to the limited resources of most researchers this can hardly be more than wishful thinking - at least for large-scale studies. 143 However, by making the criteria for metonymy or metaphor identification explicit, the subjective element involved can at least be restricted. Recently, the Pragglejaz Group 144 has suggested a “’metaphor identification procedure’ (MIP)” (2007: 1; cf. also Steen 2007, Steen et al. in print) that aims at avoiding researchers’ subjective decisions about the metaphoricity or non-metaphoricity of words and phrases. Such an objective procedure would, if applied consistently, ascertain the comparability of different empirical analyses and, moreover, result in much more reliable insights into the actual “frequency of metaphor, its organization in discourse, and possible relations between metaphoric language and thought” (Pragglejaz Group 2007: 2). The MIP consists of a 4-step analysis: 1. Read the entire text-discourse to establish a general understanding of the meaning. 2. Determine the lexical units in the text-discourse 3. (a) For each lexical unit in the text, establish its meaning in context […] (b) For each lexical unit, determine if it has a more basic contemporary meaning in other contexts […]. [B]asic meanings tend to be [m]ore concrete […]; [r]elated to bodily action; [m]ore precise […]; [h]istorically older […]. (c) If the lexical unit has a more basic current-contemporary meaning in other contexts than the given context, decide whether the contextual meaning contrasts with the basic meaning but can be understood in comparison with it. 4. If yes, mark the lexical unit as metaphorical. (Pragglejaz Group 2007: 3) The identification of potentially more basic senses if not always an easy task. Therefore the Pragglejaz Group recommends using a dictionary in 143 If only 5% of the instances of a corpus study are unclear cases and if this study embraces not more than 20 lexical items and is restricted to 1000 concordance lines for each, this would mean that one would have to find independent raters for 1000 concordance lines, introduce them to the annotation scheme, have them analyze the data, compare the results and test for inter-rater reliability. And if the results turn out to be unreliable, one has to start again from scratch, perhaps modify the annotations scheme, discuss every single instance with the fellow-raters, etc. While such a procedure is undoubtedly a desirable standard for corpus-based research into figurative language, it conflicts with the limited resources available to most researchers. 144 The members of this group are Peter Crisp, Raymond Gibbs, Alan Cienki, Graham Low, Gerard Steen, Lynne Cameron, Elena Semino, Joseph Grady, Alice Deignan, and Zoltán Kövecses. 129 step 3. As Krennmayr (2008) points out, however, it is not always sufficient to use one dictionary only. Her research group applied the MIP to data from the BNC-Baby and found that for some words decisions with regard to the basic meanings could only be made after more than one dictionary had been consulted 145 - a fact which makes the procedure even more demanding and time-consuming. The main reason, however, why the MIP could not be used in my analysis, lies in step 1: “Read the entire text- discourse […]” (Pragglejaz Group 2007: 3). This is feasible, if one is interested in the different metaphors occurring in one or several texts, in their distribution and their functions within the text(s). But it is not, if occurrences of metaphorically used words are searched for in a corpus, and thousands of concordance lines are analyzed. Reading for each and every concordance line the whole text from which it was extracted would be extremely time-consuming. If, as was planned for my metaphor study, one wants to take 50 different lexemes into consideration and analyzes 1000 concordance lines for each of them, it is very likely that one would have to read through a fairly high amount of all the texts in the corpus. My analysis is based on the BNC and as the BNC website states: “The BNC World Edition contains 4,054 texts […]. Reading the whole corpus aloud at a fairly rapid 150 words a minute, eight hours a day, 365 days a year, would take just over four years” (http: / / www.natcorp.ox.co.uk/ corpus/ index.xml.ID =numbers; 13.12.2009). In order to assess the conventionality of metonymies and metaphors, the proportion of metonymic and metaphorical meanings of selected lexemes was investigated with material from the BNC. The idea behind this method is that the conventionality of a metonymic or metaphorical reading of a lexeme correlates with the relative frequency of this reading vis-à-vis the frequency of other readings. For the noun press, for example, the great majority (92.3%) of all records of the lexeme investigated were metonyms in the meaning ‘media’. Therefore it seems plausible to postulate a very high degree of conventionalization for this metonymic meaning. If, on the other hand, only a very small proportion of metonymic uses was found, then the metonymic meaning is arguably not highly conventionalized. The latter is, for example, the case for the noun wine, where only 0.4% of all 145 In the example analysis given by the Pragglejaz Group (2007) the researchers rely on the Macmillan English Dictionary for Advanced Learners (MED). As Krennmayr (2008) emphasizes it is important to use contemporary corpus-based dictionaries for analyzing contemporary texts. The MED fulfils this criterion and has the additional advantage that it explicitly acknowledges the existence of conventional metaphors in the sense of the conceptual theory. However, different corpus-based dictionaries are neither consistent in the number of senses they distinguish, nor with regard to the treatment of figurativity (cf. e.g. van der Meer 1999, 2002). The basic sense established in step 3 therefore is invariably dependent on the dictionary used. A possible disadvantage of using a learner’s dictionary might also be that this type of dictionary is always restricted in size. Slightly different senses are therefore often treated as one and the same lexical unit. 130 uses transport the metonymic meaning ‘bottle/ glass of wine’. While it is not quite clear how individual percentages can be interpreted, the findings do provide a good indicator for different degrees of conventionalization. In what follows, I will first present the criteria which led to the inclusion of certain types of lexical material, i.e. linguistic metonyms and metaphors, into the study (chapter 5.1.2). This is followed by an explanation of how the figurative meanings were identified in the BNC, how problems and unclear cases were dealt with in the study, and how the results gained by the corpus search are presented in the Appendix (chapter 5.2). Since the focus of my study lies on metonymy rather than on metaphor, I will, as a rule, start with a discussion of the relevant issues for the metonymy study, before turning to the metaphors. 5.1.2 The choice of the source domain vocabulary 5.1.2.1 The metonyms My analysis is based on source domain vocabulary, i.e. the approach is a semasiological one. The first step consisted of identifying potentially relevant lexical items. The obvious method was to choose words which were suspected of differing with regard to the frequencies with which they transport figurative meanings. Otherwise it would not have been possible - once their relative frequencies had been established - to further analyze and compare the frequent and less frequent (or infrequent) conceptual metonymic mappings as well as linguistic metonyms in order to find out which factors contribute to their respective conventionality or unconventionality. To find cases which potentially differ with regard to their frequencies, I consulted various sources. The most extensive source for the items was the Master Metonymy List (Leite 1994). Compared to the Master Metaphor List (Lakoff, Espenson and Schwartz 1991), this list is much less comprehensive so that other sources had to be used as well. These were: Apresjan (1974), Barnden (1997), Bredin (1984), Croft and Cruse (2004), Díez Velasco (2001), Fass (1997), Frisson and Pickering (1999), Gerrig (1989), Gibbs (1994), Knowles and Moon (2006), Markert and Hahn (2002), Markert and Nissim (2002b, 2003), Nissim and Markert (2005), Nunberg (1995), Radden and Kövecses (1999), Ruiz de Mendoza and Díez Velasco (2002), Stallard (1993), Warren (1992). Some of the sources already provide ideas regarding how often the single metonyms occur. Gerrig (1989), for example, treats the metonym uniform ‘uniformed police officers’ as a novel, unfamiliar metonym. Nissim and Markert (2005) refer to the example seat ‘person occupying the seat’ as an unconventional metonym. And Frisson and Pickering (1999) mention the LOCATION FOR INSTITUTION metonymy convent as belonging to the class of 131 familiar metonyms. Other sources which are neither corpus-linguistically oriented nor belong to the realm of psycholinguistics, seldom give information about the likely frequencies of the examples. Exceptions are Radden and Kövecses (1999), who often point out which of the two possible directions of mapping is the more common one for reversible metonymic mappings, and Leite (1994), who distinguishes “traditional metonymies” from “grammaticized metonymies” (1994: 1). The traditional examples are those she assumes are “easily recognized as ‘figures of speech’”, while the grammaticized ones have led to semantic change (1994: 1). However the difference between the two classes 146 is not always clear. It may be convincing that crown (‘royal government’) belongs to the grammaticized examples of the COSTUME FOR WEARER metonymy and suits (‘office workers, managers’) is a more obviously figurative expression based on the same mapping. But why, for example, Play ball! is categorized as a traditional metonymy and Let’s play darts as a grammaticized metonymy (both PLAY- ING PIECE FOR ENTIRE GAME ) is not evident. Nevertheless, for some cases, it was relatively certain that they would return either very low or very high frequencies. That America is frequently used as a PART FOR WHOLE metonymy in the meaning ‘United States’ is easy to predict. And that appendicitis is only comparatively rarely used as a metonym to refer to a person does not come as a surprise either. In other cases, however, the predictions were more difficult. This concerns most of the stock examples of cognitive-linguistic research into metonymies, such as bottle ( CONTAINER FOR CONTENT ), face ( BODY PART FOR PERSON ) or Picasso ( PRODUCER FOR PRODUCT ). How often they are really used in ordinary situations is unclear. That the proponents of the conceptual view usually subsume them all under the heading of conventional metonymy does not mean that they are indeed all equally conventional if analyzed from a usage-based perspective. A further point influencing the choice of lexical items was that instances of as many different mappings as possible should be included. This is of particular relevance for an extensive description of the conventionality of metonymy because the different mapping patterns vary considerably in terms of frequency and conventionality. As already mentioned, there are mappings which first and foremost lead to context-dependent linguistic manifestations, such as FOOD ORDERED FOR CUSTOMER (e.g. ham sandwich) or ILLNESS FOR PATIENT (e.g. appendicitis). On the other hand, there are mappings which can be reasonably believed to occur quite often in natural language, such as PRODUCER FOR CONSUMER PRODUCT (e.g. BMW ‘car produced by the BMW company’). 146 Actually, Leite (1994) has a third large group of metonymies. The “’cognitive’ metonymies” are generic-level mappings (1994: 2) such as PART OF ICM FOR WHOLE ICM or PRECONDITION FOR WHOLE ICM (1994: 20). These general metonymic mappings underlie the concrete cases she deals with in her other two classes. 132 Particularly interesting are, of course, reversible metonymic mappings. Cases in point are, for example, metonymic mappings between CONTAINER and CONTENT or between CONTROLLER and CONTROLLED . They can often be easily connected to the laws of ontological salience. According to these general laws, CONTAINER FOR CONTENT metonyms (e.g. kettle ‘content of kettle’) should be more frequent than CONTENT FOR CONTAINER metonyms (e.g. wine ‘wine bottle’) and CONTROLLER FOR CONTROLLED metonyms (e.g. Napoleon ‘Napoleon’s army’) should be more frequent than CONTROLLED FOR CONTROLLER metonyms (e.g. sax ‘sax player’), since CONTAINER FOR CONTENT and CONTROLLER FOR CONTROLLED obey the principles VISIBLE OVER NON - VISIBLE and HUMAN OVER NON - HUMAN respectively, while their reversals diverge from these principles. In order to investigate these predictions, which are found in many sources addressing the question of the conventionality of metonymy (e.g. Radden and Kövecses 1999; Barcelona 2002a), I tried to include instances of both mapping directions. This was, however, not always possible. Some reversible mappings are not very promising in a corpus-based study. The mapping between POSSESSOR and POSSESSED is, for example, in principle reversible. Radden and Kövecses (1999: 40) quote the following instances of the two different mapping directions: (66) POSSESSOR FOR POSSESSED : That’s me. (‘my bus’) (67) POSSESSED FOR POSSESSOR : He married money. (‘person with money’) P OSSESSOR FOR POSSESSED metonyms necessarily involve reference to persons on the linguistic surface, often in the form of pronouns (e.g. I am parked over there, You have a flat tire), but also by way of proper nouns (e.g. Bill is in the Guinness Book of Records; Radden and Kövecses 1999: 41). Taking the high frequency of pronouns in corpora into account, a search for metonymic uses would have been like looking for a needle in a haystack. But highly conventional metonymic meanings of pronouns are extremely unlikely anyway. A similar consideration lead to the exclusion of instances of Nunberg’s (1995: 110-111) deferred indexical reference (e.g. This is parked out back; cf. chapter 4.3.2.1). Demonstratives are so numerous in natural discourse that it is extremely unlikely to come across instances of metonymic use if only a limited number of occurrences can be looked at. Furthermore, the deictic nature of these metonyms would have made it hard to establish the actual intended referents, i.e. the targets of the underlying mappings - especially in the spoken part of the corpus. In general, the choice of the linguistic items was geared towards a maximization of the usefulness of the study. The consideration of as many mappings as possible means that the examples I included in my study belong to various different frames or ICMs (cf. Leite 1994; Radden and Kövecses 1999), such as: • PRODUCTION : e.g. PRODUCER FOR PRODUCT (e.g. Shakespeare ‘Shakespeare’s works’), 133 • CONSTITUTION : e.g. MATERIAL OF OBJECT FOR OBJECT (e.g. furs ‘garment made of fur’), • CONTROL : e.g. CONTROLLER FOR CONTROLLED (e.g. Napoleon ‘Napoleon’s army’), • POSSESSION : e.g. POSSESSED FOR POSSESSOR (e.g. money ‘person having money’), • CONTAINMENT : e.g. CONTAINER FOR CONTENT (e.g. packet ‘contents of packet’), etc. However, some ICMs/ frames had to be excluded. The first group which was excluded were most examples belonging to the ACTION frame or ICM. Noun-verb conversions of the type to author a book or to dust the room are metonymic, but in terms of the conventionalization potential they are not of prime interest. They are instances of a productive metonymically motivated word-formation process and certainly conventional and easy to process and understand. In their verbal forms they do not have any other competing, literal interpretations. This leads to a lack of any ground of comparison in a corpus study. It is possible to compare the frequencies with which, for example, Shakespeare is used in a literal way and in a metonymic way. But not much is gained regarding the aim of this study by establishing the frequency of the verb to author. The same is true for instances of ACTION FOR AGENT metonymies such as writer or driver and many other patterns like ACTION FOR RESULT (e.g. production) or MANNER FOR AC- TION (e.g. to tiptoe) (cf. Radden and Kövecses 1999: 37). The examples which fit my purposes best are nominal metonymies which are not products of conversion. Cases like Blondes have more fun ( AT- TRIBUTE OF ENTITY FOR ENTITY , Leite 1994: 7; DEFINING PROPERTY FOR CATE- GORY , Radden and Kövecses 1999: 35) had to be excluded as well since they also lack competing literal interpretations. 147 The study was thus restricted to non-derived common nouns, some syntactic groups (four wheel drive, set of wheels) and proper nouns. 148 The last group is particularly relevant for mappings belonging to the PRODUCTION ICM (e.g. Picasso ARTIST FOR WORK ) and to the LOCATION ICM (e.g. Vietnam LOCATION FOR EVENT ). To keep the amount of data to be inspected manageable, the number of lexical items was limited to 100. The complete list of the examples studied can be found in Appendix A. The mappings highlighted by bold print in 147 As opposed to e.g. glasses, the plural of blonde can only receive a metonymic interpretation. The potential existence of a competing literal interpretation was of prime importance as regards the choice of vocabulary. This also accounts for the inclusion of the noun sail, which might be assumed to be an instance of conversion. This is, however, not the case. As the NODE indicates, Old English had two different forms for the noun and the verb, segel and seglian. 148 What is also excluded from consideration are what Panther and Thornburg (2004) call ‘illocutionary metonymies’ (cf. chapter 3.2.2.1, cf. also Radden and Kövecses’s (1999) EVENT ICM), since the study of indirectness is certainly a field of study in its own right. 134 the appendix are those as instantiations of which the expressions were taken over from the respective sources. As a rule, the single examples were taken over in the form in which they were quoted in the sources. Mostly, the metonymically used nouns occurred in their singular forms in the sources. But I also included some examples in the plural form, for instance airlines, buses, uniforms, and suits. For some words like airlines or schools, which were both originally found in the INSTITUTION FOR PEOPLE RESPONSIBLE sense, number does not seem to matter too much. But there are other cases in which it plays a bigger role. Wings (‘plane’), for example, an ATTRIBUTE OF OBJECT FOR OBJECT metonymy, can only occur in the plural. Wings typically come in pairs and a reference to a plane by using the singular wing is highly unlikely: *The wing took off from the runway is, to say the least, odd. A similar case is heels (‘shoe’). For some examples it was thus essential to take them over in the form in which the sources quoted them. A problematic case is, in a way, glasses. The metonym was taken over from Nunberg (1995: 117) as an instance of the mapping CONTAINER FOR CONTENT (e.g. She drank two glasses). Here, the plural, however, sanctions the additional meaning ‘spectacles’ which is actually a double metonymy or, in other words, a metonymic chain consisting of MATERIAL OF OBJECT FOR OBJECT and PRINCIPAL PART FOR WHOLE . This meaning is very frequent and may have biased the results in an undue fashion towards a very high overall frequency of metonymic uses of the lexical item. In addition, the plural form also makes occurrences of the literal meaning of glass, i.e. the ‘material’ sense, somewhat more unlikely than a search for the singular form. Nevertheless, I decided to adhere to my policy of taking over the examples in the form in which I found them. Each decision has its consequences, and unwanted side-effects can never be entirely avoided. If the singular form had been used instead, the highly frequent and quite conventional metonymic sense ‘spectacles’ would not have occurred at all. At least sticking to the forms used in the sources seemed to guarantee that instances of the mappings I was looking for could be found in the first place. Before I turn to a description of the metaphor material investigated, a last word seems in order concerning the fact that it might seem odd - at least at first glance - to include metonyms such as Vietnam, Pearl Harbor/ Pearl Harbour, Pentagon or White House, which have their origin in American history and politics respectively, in a study that is based on the British National Corpus, which represents British usage. That many metonyms commonly quoted in the relevant literature are oriented towards American English and American culture rather than towards their British counterparts is understandable, taking into account that cognitive linguistics and especially the conceptual theory of figurative language and thought was initiated in the US. And the examples used to illustrate the theory are often taken over by researchers with a European background like Radden and Kövecses. Possible British equivalents such as Downing Street are hardly ever discussed. However, those examples used in my study which 135 have such an obvious American origin also all have a long acquired status as internationalisms. Due to the extreme influence of American politics for the world at large, metonyms like Pentagon or White House have spread to all other English speaking countries as well and have sometimes also been borrowed by other languages. German, for example, has a loan translation of White House, i.e. Weißes Haus, but the (originally Greek) name Pentagon has simply been borrowed for the German language. As concerns the PLACE FOR EVENT metonyms Vietnam and Pearl Harbor/ Pearl Harbour, the situation is basically the same: The incidents referred to by way of the metonyms had such severe repercussions for many countries that the metonyms have become known, or to use a technical term, institutionalized, worldwide. They are undoubtedly established parts of British English. And in order to incorporate them into British English in the first place, speakers must have found the relationship between forms and meanings established by these metonyms meaningful, which does not come as a surprise either considering that they are based on well-established and very common ways of metonymic reasoning. That their origins lie in another speech community thus does not matter for my concerns, since I’m not interested in the coining of the metonyms, but in their conventionality within a given speech community. 5.1.2.2 The metaphors The number of metaphors taken into consideration is smaller than that of metonymies. All in all, 50 different lexical items were used in the corpus search. The Master Metaphor List (Lakoff, Espenson and Schwartz 1991) is quite extensive and quotes most of the stock examples which are used recurrently in other sources on conceptual metaphor (e.g. He attacked my position on abortion, I grasped the idea, He followed his orders). It thus served as the main source of the vocabulary items included in the study. Only some were additionally collected from Lakoff and Johnson (1980). An advantage of relying mainly on the Master Metaphor List is that it presents the mappings and examples in a structured way. The list falls into four large categories: event structure, mental events, emotions, and a mixed bag called ‘other’. The single categories are then split up into several sub-sections. Once again, I tried to incorporate material of as many different mappings as possible. For metaphors, however, this proved much more difficult than for metonymies. The amount of different mappings attested in the Master Metaphor List is simply too large to cover a representative part of them. The choice of the examples is therefore largely arbitrary. An overview of the metaphorical expressions studied can be found in Appendix B. 149 Only some of the sub-categories of the four major classes could be 149 The senses in which the metaphors were taken over from the sources are indicated by bold print in Appendix B. 136 considered. Within EVENT STRUCTURE , for example, I included metaphors belonging to the following sub-classes: • PROPERTY : e.g. IMPORTANCE IS INTERIORITY (e.g. This is the core of the matter), • CAUSATION : e.g. CAUSATION IS COMMERCIAL TRANSACTION (e.g. He profited from his experience), • ( LONGTERM ) ACTION : e.g. CAREER PROGRESS IS VERTICAL MOVEMENT (e.g. He’s half-way up the corporate ladder), • EXTERNAL EVENTS : e.g. THE PAST IS A PLACE OF CONFINEMENT (e.g. I can’t escape from that memory), • HARM : e.g. HARM IS PHYSICAL INJURY DUE TO PREDATION BY VICIOUS ANI- MALS (e.g. Some lawyers are real sharks), • COMPARISON : e.g. COMPARING AND SEEKING ABSTRACT THINGS IS SHOP- PING (e.g. I’m shopping for a better philosophy), • COMPETITION : e.g. COMPETITION IS WAR (e.g. I beat him at chess), • TIME : e.g. TIME IS A VALUABLE COMMODITY (e.g. You’re wasting your time). 150 Another factor which influenced the choice of examples was, of course, that the metaphors should occur with different frequencies in natural discourse. The Master Metaphor List does not provide any hints in this respect and other sources also remain silent as regards the real frequency of conceptually based metaphorical expressions. Once again, for some cases it was not too difficult to predict that they would either occur very frequently in their metaphorical senses (e.g. wasting as in You’re wasting your time or beat as in I beat him at chess) or that they would be rather seldom used as metaphors (e.g. electricity as in I could feel the electricity between us). For the largest number, however, the predictions were difficult to make. As the above-mentioned examples indicate, the lexemes considered in the corpus study were not restricted to a single word class (as in the metonymy study). While the stock examples of metonymy are mostly nouns, metaphors are often expressed in verbs and adjectives on the linguistic surface. 151 And in all cases the metaphorical interpretation has to compete with other, more strictly literal readings. The relative frequencies of the metaphorical senses can thus be compared to the latter. 150 The examples used to illustrate the different mappings are taken from Lakoff, Espenson and Schwartz (1991) and Lakoff and Johnson (1980). Some have been adapted. 151 That metaphors often occur in the form of verbs and adjectives is also one of the reasons why no metaphors were taken over from psycholinguistic studies, although they often indicate how familiar their examples were rated by the participants of their studies. As already mentioned, psycholinguists usually have standardized examples of the form X is a Y, i.e. focus on nominal metaphors only. Due to this grammatical form the examples differ markedly from the common examples given in the literature on conceptual metaphor. 137 The grammatical form of the lexical items was again adopted from the sources. Again, it cannot be entirely excluded that this influenced the results. But a cross-check of the forms used, and other grammatical forms in which the lexical items can transport metaphorical meanings, would have increased the amount of data to be analyzed dramatically. Since figurative senses can only be found in a corpus by reading through the concordance lines, such a multiplication of the data would have gone beyond the scope of the study. One can, however, reasonably assume that if the form wasting occurs frequently as an instantiation of the conceptual metaphor TIME IS A VALUABLE COMMODITY , other grammatical forms can transport the same meaning. The question of to what extent different grammatical forms of one and the same lexical item convey a given metaphorical meaning cannot, however, be answered here. 5.2 The procedure: Looking for metonymies and metaphors in the BNC 5.2.1 Metonymic senses 5.2.1.1 KWIC concordances and the problem of underspecification After the target items had been identified, the next step was the actual corpus search. The metonyms had been chosen without controlling the overall frequencies of the words in the BNC. Some of them are highfrequency words such as car, which returns a total of 26,731 hits in the BNC or paper, which returns 16,926 hits. Others rarely occur, such as sax (87 instances), set of wheels (14 instances) or turf (13 instances). Obviously, not all instances of the high-frequency words could be analyzed. The maximum number of hits taken into account was therefore limited to 1000. To reduce the number of the concordance lines for those words which occur more often than 1000 times in the BNC the part-of-speech query function of the software (SARA) was used. In this way, one can exclude concordances which are irrelevant for the current purpose. The form skirt, for instance, occurs as a singular common noun (annotated as NN1 in the BNC) or as a verb form. For the latter, the BNC distinguishes between two potential forms: VVB, i.e. the base form of the lexical verb, except for the infinitive, and VVI, which signifies the infinitive form. Additionally, there are often cases where the annotations are ambiguous. For skirt, there are instances which are marked as NN1-VVB (101 cases) and others (14 cases) which are annotated as VVB-NN1. The instances annotated as verb forms are irrelevant for a study of metonyms, as illustrated by the following examples: (68) VVB: You skirt Godinton Park to go on to Great Chart. (CHJ-1271) (69) VVI: There’s a tunnel up ahead but you must skirt round it. (APC-563) 138 The samples of ambiguously annotated instances would have been potentially relevant. This can be shown with the help of examples (70) and (71) taken from the sample returned by a part-of-speech query for NN1-VVB forms of skirt: (70) She was wearing a blue costume with a slit skirt that opened when she crossed her legs. (ACW-592) (71) [...] we would sit in a café or at home discussing image, beauty, fashion and advertising, and skirt around the more obvious, immediate issues of how we both looked, how we were dressed. (ADG-1660) In (70), skirt is used as a noun, and would thus have been relevant for my aims, i.e. to find out about the relative frequencies of literal vs. figurative meanings. However, a consideration of these parts of the corpus would not have entirely excluded irrelevant verbal forms, as (71) demonstrates. Therefore only unambiguously annotated cases were taken into account as long as the number of such hits amounted at least to 1000. Nevertheless, this policy did not entirely avoid instances of unwanted grammatical forms in the sub-corpora. Quite frequently, the BNC is not annotated correctly. For street, for example, the part-of-speech query offers the singular common noun (NN1) or the proper noun (NP0). But even if the search is limited to NN1, some occurrences of NP0 are included in the download, such as Wall Street (e.g. AJD-1362, AJP-121), Downing Street (HHV-2523) or Regent Street (B24-191). An even better example is the query for China/ china. Actually, one might expect that instances of the metonymic mapping I was looking for, LOCATION FOR PRODUCT MADE THERE (‘porcelain’), should only turn up in the NN1 sample, but not in the NP0 sample, as china is not - or rather no longer - a proper noun in the ‘porcelain’ sense. However, a query for NN1 forms also returns instances of the NP0 form. 152 Examples are given below: (72) FROM OUR CHINA CORRESPONDENT (e.g. ABD-866; original caps) (73) [...] the American traveller, once got a lift from China to Tibet (ASJ-165) The choice of the right part(s) of speech which the queries should include was therefore nontrivial. In general, I tried to maximize the data. This is also why the search was not restricted to a particular part of speech when the overall number of hits was below 1000. Ambiguously annotated examples and examples tagged as grammatical forms which were originally not very promising in terms of finding metonyms were included in the queries in these cases, since chances were that due to the incorrect annotations of the BNC these samples embraced at least some metonyms. Which parts of speech were taken into consideration in the queries is specified in Appendix A in the second column indicating the number of instances analyzed. If no information on the part of speech is provided 152 A limitation of the search to NN1 instances of china/ China would have also excluded finding other metonymic mappings, such as COUNTRY FOR RULERS or LOCATION FOR PEOPLE . 139 there, the word-forms used for the queries were either all tagged as belonging to the same grammatical form in the BNC or no restrictions were made due to the above-mentioned considerations. A last word with respect to the lexical items and the forms used in the study concerns the proper noun Pearl Harbor. Since the query for the form Pearl Harbor only returned 60 hits, the search was widened to include the spelling variant Pearl Harbour. This raised the overall number of potential metonyms to 87. For none of the high-frequency word-forms did the limitation of the search to the relevant parts of speech result in samples of 1000 hits or less. For that reason a further step had to be taken in order to reduce the concordance lines of these cases to a manageable level. To do so, SARA offers several opportunities. In general, it is possible to determine the number of instances to be downloaded, and the user can also influence how the software chooses the examples. One can download the initial solutions, an entirely random set of the examples or one instance per text. Choosing the initial solutions option would have unduly influenced the number of different text-types taken into account. 153 Therefore random samples were chosen instead. These were downloaded and served as a basis for the search for metonymic senses. For the words with a lower overall frequency in the BNC and with only one grammatical form (e.g. ecstasy, kettle) as well as for those cases where a limitation of the search to a certain relevant part of speech resulted in less than 1000 hits (e.g. butt, pill), all instances were downloaded to a subcorpus and subsequently analyzed. The samples are presented by the program in the form of KWIC (key word in context) concordances. An example is given in Figure 14. 1 as if she could look in the soup kettle 2 Inside he filled a kettle and lit the gas-ring. 3 mble turned his attention to the kettle. 4 room with a little fridge and a kettle, but you have to share kitchen and b 5 Put the kettle on, spooned coffee, rolled a cigarett 6 An old kettle with a hole burnt in its bottom lay 7 e old house, Donald kicked the kettle in a spasm of impatience and said a 8 ine hardly had time to boil the kettle, to make a cup of coffee, to toast a sl 9 ‘Surely you won’t need the fish kettle? ’ she queried. 10 She filled the kettle and switched it on. 11 ions, I filled my saucepans and kettle with water and set them all to boil. 12 I quickly poured the kettle full of water over them, then the co 13 existence in the kitchen as the kettle boils. Figure 14: An extract of the KWIC concordance for kettle 153 Especially the spoken part of the corpus would have been neglected by taking only the initial solutions into consideration, since the spoken data form the last part of the BNC. All the written texts and text-types are to be found at the beginning. 140 The figure lists the first thirteen hits for the form kettle downloaded as a sub-corpus. The metonymic senses were then identified by considering the meaning of the key word in the given contexts. In the extract, example eight and thirteen are clear instances of the CONTAINER FOR CONTENTS mapping. The metonyms can be easily identified by the typical collocate boil. All the other concordances in Figure 14 are literal uses, with the exception of number eleven. The last instance is tricky, because the noun is linked with two predicates, each of them leading to a different interpretation of kettle, one to a literal and the second to a nonliteral, metonymic one. How such cases were dealt with will be discussed below in chapter 5.2.1.2. A perhaps prototypical instance of a literal use of kettle is example two. What is being referred to here is the container only. Example five illustrates quite a frequent situation. The kettle that is being put on is certainly filled with water. The container and its contents are treated as a conceptual unity. The container is, due to being the visible part, chosen as the active zone and used for the reference to the whole. Such cases were, in general, not counted as metonyms. It is impossible to speak of a clear CONTAINER FOR CONTENT relation. In order to treat this instance as a metonym a mapping of the kind CONTAINER FOR CONTAINER PLUS CONTENT would have to be posited. And this marks the crucial difference to other, clearer instances of the CONTAINER FOR CONTENT mapping, where the target is the CONTENT only (e.g. the kettle is boiling). 154 Instead, the meaning of the word kettle can be described as being underspecified in this case (cf. Frisson and Pickering 1999). It oscillates between the container and its content and actually comprises both. C ONTAINER FOR CONTENT mappings are, however, far from being the only ones which lead to such underspecified meanings, where it is virtually impossible to decide what is actually being referred to in a given concordance line. One word-form where this problem turned up quite frequently is buses. Here, I was first and foremost looking for the CONTROLLED FOR CONTROLLER meaning as in the well-known example The buses are on strike. To see bus drivers and their vehicles as separate entities is, as it seems, not the usual way in which the situation is construed. Rather, speakers tend to consider the drivers and the buses as one gestalt-like entity. Strictly speaking, buses do not drive themselves but are driven by persons, but they are often described as running more or less on their own or as waiting or stopping at their own discretion. Consider the following examples: 154 The only way to treat example five in Figure 14 as a metonym would be to conceive of it as a PART FOR WHOLE construction parallel to instances such as FACE FOR PERSON . However, within such an analysis the CONTAINER - CONTENT relation, which is vital for the expression as a whole, would be lost. And a combination of the PART FOR WHOLE analysis with the CONTAINER - CONTENT relation is hardly possible as the latter belongs to the larger group of PART FOR PART mappings. Such combinations of different mappings are only possible in metonymic chains, but not within one instance. 141 (74) [...] a 15-20 minute bus ride (buses run every half hour) will take you to Innsbruck (ADM-1467) (75) Pedestrians scampered blindly across the road to catch waiting buses [...] (G1W-394) (76) [...] local buses stop nearby. (ECF-434) Buses do not run without their drivers, neither can they stop without them. And buses can hardly be conceived of as being literally waiting for somebody. Yet, one can sit in one of the waiting buses (B1N-424). In fact, it is the drivers who are waiting for the passengers if they are lucky. The persons responsible for all this, however, are not the centre of the speakers’ attention. They are backgrounded for several reasons. First, even though they are human, the bus drivers could be said to be less ontologically salient than the buses: The buses are what can be seen in the street, the drivers are more or less hidden within them. In this way, one could even speak of a CONTAINER - CONTENT relationship between the vehicles and their drivers. It is the drivers’ actions which cause the buses to drive, stop, etc., but the drivers only play a secondary role. What we concentrate on is the concrete event, i.e. the result of the actions caused by the drivers: a bus that drives, stops, etc. Moreover, the drivers are not construed as individual persons. They are largely reduced to their profession, to the function of driving buses. All this leads to a gestalt-like impression, where the meaningful whole is the bus plus the driver. One without the other is literally useless for average persons. This quite common construal makes it often impossible to decide whether the concordance lines really feature metonymic uses or not. In order to analyze examples like (74) to (76) as metonymic, one would have to argue - parallel to what has been said regarding kettle - that there is a productive mapping of the kind CONTROLLED FOR CONTROLLER PLUS CON- TROLLED . For the reasons outlined above for kettle, I prefer to consider such uses of the form buses as underspecified rather then metonymic. Therefore such cases were not counted as metonyms, in contrast to other examples such as (77) and (78) which are much more obviously and unequivocally metonymic: (77) Taxis don’t turn up, buses stop for a chat [...] (G2B-1462) (78) Since deregulation, buses are only trying to get from A to B as fast as possible in order to make as much profit as possible. (G3H-1143) Buses cannot talk and cannot be said to be trying anything. The uses of buses in (77) and (78) are thus clear instances of the CONTROLLED FOR CON- TROLLER mapping with the bus drivers being the actual targets. 155 155 Regarding example (78), one has to add that the latter part of the sentence, i.e. in order to make as much profit as possible, points to another possible interpretation of buses where not the bus drivers themselves are in the centre of attention but the company, since it is usually rather a company that is interested in profit than its employees. However, because the first part of the verbal phrase clearly refers to the drivers as the 142 As already indicated, the problem of underspecification occurred with many, if not most of the linguistic items. Often, the meaning oscillates between the literal sense of a word and its possible metonymic uses. Further examples include: (79) [...] Soviet combat forces fought for a decade in Vietnam and other foreign wars [...] (A57-428) (80) [...] and the Boar combines with Adonis to produce Shakespeare’s tragic hero (AJV-980) (81) She died in 1850, having been allowed to leave the convent for the last few months of her life [...] (CA7-1552) Since the typical relationship which gives rise to metonymy is a withindomain one, such cases abound in everyday language. The Vietnam War and its location belong together. Thus it is only natural to find uses as in (79) where one cannot decide whether the speaker refers only to the location or rather to the event as a whole. The preposition in seems to indicate that Vietnam is meant to refer to a location here, but as wars and their duration are commonly construed as containers, this does not lead to a clear picture either. Likewise, strictly speaking, the tragic hero in (80) only exists in Shakespeare’s plays. So Shakespeare should be considered a PRO- DUCER FOR PRODUCT metonymy here. But is such a line of argumentation really convincing? And when the woman in (81) leaves the convent, does she leave the location only? Or does she leave the institution associated with the location? In many cases, definite answers cannot be given. The general flexibility of meaning illustrated by these examples (cf. Chandler 1991) is nothing the average speaker worries about. Rather, it is often an advantage that speakers do not always have to be very accurate in their choice of words. And hearers will also not bother about the unspecific nature of the meaning conveyed in such cases. They construct their own mental representation of the linguistic input (cf. Coulson 2001: 17, 37), and as long as they are not explicitly asked whether they think that it is one of Shakespeare’s heroes or one of the heroes in his plays, they will not even be aware of the ambiguity of the utterance. While it cannot be denied that such examples have a metonymic flavour, they cannot be analyzed in the same straightforward way as other instances. Therefore, examples featuring underspecified meanings are not part of my analyses, even though a more thorough consideration of the semantic aspects of such cases would have been quite interesting, especially as they seem to be so numerous. However, the focus of my study is figurative language proper, and underspecified meanings form the area only ones who can directly influence the speed of their buses, this instance was counted as belonging to the mapping CONTROLLED FOR CONTROLLER and not as an example of a metonymic chain which leads from the CONTROLLED (i.e. the buses) to the CONTROLLERS (i.e. the bus drivers), and then from these representatives of the company to the company itself ( PEOPLE FOR INSTITUTION ). Cf. chapter 5.2.1.2 on how conjoined phrases were tackled in the analysis. 143 where figurative language slowly blends into literality. As a rule, only instances where source and target meaning could be clearly delineated, i.e. where a clear stand-for relationship could be posited, were considered to be metonymic. 5.2.1.2 Conjoined phrases, multiple mappings, and meanings The examples in Figure 14 above illustrate one further principle adhered to in the analyses. Consider once more the example in line eleven: I filled my saucepans and kettle with water and set them all to boil. Here, the word kettle only occurs once, but due to its being related to different predications, it realizes two different meanings, which can be split up as follows: (82) I filled my kettle. (83) I set my kettle to boil. In (82), the meaning of kettle is unmistakably a literal one. In (83), however, the sense is primarily a metonymic one. What can boil is the content only. Nevertheless, such instances were not counted as metonymic. What was taken into consideration in the analyses was always only the first of two or more conjoined predications. The reason is twofold: On the one hand, this policy was followed for a very practical reason. A consideration of all of these cases would have markedly increased the reading to be done as the second parts of conjoined phrases often only occur at a considerable distance from the actual key word. The other reason is less trivial and of a more theoretical nature: In general, the relative frequency of a given metonym was measured against the background of the frequency of the literal meaning. This means that if, say, 500 of the 1000 examples looked at are metonyms, the relative frequency of the metonymic meaning amounts to 50%. But if I had analyzed a single instance of a key word as having both a literal and a metonymic meaning, I would have had to count it twice, once in a literal and once in a metonymic sense. This would have increased the total number of senses beyond the actual number of concordances analyzed. To avoid any potential confusion arising from this, I decided to stick to the first sense realized in the concordances. As mentioned in chapter 5.1.2.1, the vocabulary items used in the study were selected from various sources and with the prime intention of identifying the relative frequencies with which the mappings they served to illustrate in the sources occur in natural discourse. The analysis was, however, not restricted to these mappings. Many of the lexical items can be used in more than one metonymic sense, and the meaning I was originally looking for often turned out not to be the most frequent metonymic one realized in the BNC. An example is kitchen. The sense I was initially most interested in, i.e. ‘cuisine’ (cf. example 84), is motivated by the mapping LOCATION FOR PRODUCT MADE THERE . During the analysis, however, two 144 further metonymic mappings could be identified which can underlie metonymic uses of kitchen. Consider examples (85) to (88): (84) This favourite atrocity of the English kitchen cannot be too strongly condemned (H45-1247) (85) [...] but they have saved enough to buy a new kitchen (ED4-1592) (86) [...] you are guaranteed to get your chosen kitchen at the agreed price [...] (CG5-1321) (87) Thanks also to everyone who helped the kitchen to do so well (ANM- 999) (88) A large hotel or restaurant kitchen, preparing meals of the highest international standard [...] (EA9-120) In (85) as well as (86) kitchen is used in the sense of ‘kitchen fittings/ fitments’, i.e. as an instance of the mapping WHOLE FOR PRINCIPAL PART . What one buys and gets at the agreed price is not the whole room including the typical furniture, but only the cupboards, electrical appliances, etc. And in (87) and (88) what is meant is not the room, but the people working there. These two examples feature the sense ‘kitchen staff’ and are motivated by the more general mapping LOCATION FOR PEOPLE THERE . All of the metonymic senses found for the single words were incorporated into the study. This increased the overall number of examplemapping relations under consideration significantly. Originally, I started out with 100 lexical items and correspondingly with 100 example-mapping relations. In the end the number of mappings had more than doubled. All in all, the 100 different lexical items realize 229 instances of metonymic mappings. Evidently, this does not mean, however, that I identified 229 different mappings. Many of the common mappings recur in the analyses of different lexemes and word-forms (e.g. CONTAINER FOR CONTENT motivates uses of many different items, i.e. book, bottle, dish, glasses, kettle, label, and packet). Each of the mappings is represented in Appendix A by one line. Note that the lines do not stand for single senses, but for mappings. Quite frequently, one and the same metonymic mapping can motivate more than one sense of a given lexical item. An example is silver. There is only one mapping which leads to metonymic uses of this word - or at least this is the only one which occurred in my data: MATERIAL OF OBJECT FOR OBJECT . However, this mapping sanctions more than one concrete meaning as shown in the following examples: (89) [...] he came here and offered you money: gold and silver for your church and parishioners. (H9C-2062) (90) Kirk Sloan (Ards CC) took the silver and the new Irish under-15 champion Denis Cormican (Phoenix) the bronze. (K2D-1807) (91) Good country house furniture in mahogany and pine, plus silver, brass and pictures, are on sale. (A7D-1943) The first sense illustrated here (example 89) can be paraphrased as ‘silver coins’, the second in (90) as ‘silver medal’, and the third in (91) as ‘cut- 145 lery/ tableware’. When it was possible to clearly differentiate several metonymic senses motivated by one and the same mapping, the single senses and their frequencies are also given in Appendix A. As compared to the different mappings that can motivate metonymic uses of a lexical item or word-form, however, different meanings based on the same mapping are of secondary importance. For that reason the single senses (if any) are all indicated in the same line of the table. If no sub-senses result from one and the same metonymic mapping, meaning glosses are only provided if the resultant metonymic meaning is not obvious, as exemplified in (92) to (94): (92) [...] the why-dun-it, the book which depends for its interest on showing that [...] (FF0-391) (93) Well read the bottle and it’ll tell you? (KSR-46) (94) There were sounds of protest from the other end of the telephone. (BP7- 3077) Example (92) illustrates a metonymic meaning of book. In this case the metonymy results from a PART FOR WHOLE mapping of the more specific type MEMBER OF CATEGORY FOR CATEGORY . Book in (92) has a generic reference, it stands for a whole ‘genre’. As this sense might not be easily derivable from the mapping, it is given as a gloss. Similar cases are (93) and (94). In Read the bottle ... the WHOLE stands for a PART , i.e. the ‘label’ on the bottle. And in (94) telephone stands for the ‘connection’ established by way of the telephone. This usage is motivated by the general mapping PRINCIPAL PART FOR WHOLE . The telephone as the part we deal with is just a significant, ontologically salient part of the network of telephone connections. Many other meanings can be effortlessly derived from the metonymic mappings indicated in the Appendix. Instances of PRODUCER FOR PRODUCT mappings, for example, such as BMW or Shakespeare hardly need any explanatory glosses. It is evident what they refer to, i.e. cars and literary works respectively. America in the LOCATION FOR PEOPLE THERE sense, for example, does not require any further semantic explanation either. And the same holds true for brain when it is motivated by a BODY PART FOR PERSON mapping and many other cases, where no glosses are offered. 5.2.1.3 Further problems: The metaphor-metonymy interface, metonymic compounds, and metonymic chains An issue which has attracted considerable attention is the boundary between metaphor and metonymy. Often, figurative language is seen as a continuum, with clear-cut instances of metaphor and metonymy forming the two extremes and less clear-cut cases in the middle (cf. e.g. Dirven 1993: 15; Ruiz de Mendoza and Pérez Hernández 2003: 35; cf. also chapter 3.2.2.1). It is exactly in these fuzzy areas that the two phenomena interact in many different ways (cf. e.g. Goossens 1990; Radden 2000). As metaphormetonymy interaction is a common phenomenon, it also occurred in my samples. Only a few examples can be given here: 146 (95) When she was attacked by her venerable predecessor [...] for ‘selling off the family silver’, it added to her proud sense of detachment. (A66-590) (96) But he’s kicked football in the teeth with his tape. (CH3-6751) (97) You really shouldn’t use yourself as an aspirin for my general migraine, or whatever you called it. (H0F-2715) As the inverted commas in (95) make explicit, the whole phrase selling off the family silver is not meant literally. The main topic of the text is Margaret Thatcher and her politics. The phrase is used as a metaphor, but the metaphor ultimately relies on the metonymic mapping MATERIAL FOR OBJECT , which sanctions the ‘object’ sense of silver in the first place. In (96), football is also motivated by both metaphor and metonymy. The metaphor consists of a personification of the whole realm of football. This metaphor is, however, supported by an underlying metonymic mapping: The CENTRAL IN- STRUMENT used, i.e. the football, stands for the PEOPLE who use it - and this is not restricted to the players, but includes all others who have made football their chief occupation in one way or the other, such as managers of football clubs, coaches, referees, etc. The playing piece is what turns them into a coherent group and the idea that the INSTRUMENT can stand for the PERSONS using it (or associated with it) makes it easy to conceive of FOOT- BALL AS A PERSON . A fairly similar interactional pattern is found in (97). The lexeme aspirin is both a metonymy and a metaphor. Literally, aspirin refers to a special kind of pain relieving tablet produced by the Bayer company. But it is often used metonymically to refer to any kind of pain relieving tablet ( BRAND NAME FOR CATEGORY OF PRODUCT ). In (97), it is certainly the latter, more general meaning which sanctions the metaphorical use of aspirin to describe a person who fulfils a similar function as pain relieving tablets do. In all three cases discussed, the metonymies involved are vital for the metaphors to function. If, for example, the metonymy MATERIAL FOR OBJECT did not sanction the use of silver to refer to objects consisting of the material, the metaphor could not come into being. The metonymic meanings underlie the metaphors. Such cases were therefore counted as metonymies despite their actual metaphorical meanings. In many other cases, the lexical items considered were clearly used as metaphors without having any metonymic basis. Examples include the following: (98) You’re a headless chicken, Wallace. (FP7-2425) (99) [...] they could at least hold the lid on the Indo-Chinese kettle for the predictable if only relatively limited future. (EFA-736) (100) They say that Birmingham is a sleeping giant and it is but it needs a massive injection of money. (CEP-3725) While (98) is more likely to be a one-shot image metaphor, the other two examples are motivated by well-known metaphorical mappings. In (99), the difficult situation in Indo-China is depicted as a boiling kettle. This is reminiscent of many emotion metaphors, where ANGER IS HEAT , or more 147 specifically A HEATED FLUID IN A CONTAINER . This general idea, which relies on the basic experiences humans have with their own bodies, can be transferred to larger, more abstract scenarios. Difficulties or dangerous situations, such as an imminent war, are conceived of as heated fluids which are about to boil over or erupt from their containers. Example (100) exploits the mapping MONEY IS A FLUID for a vivid description of the current situation of a football club. The club as such is personified and the money it badly needs is the substance required to wake up the giant. The common conceptualization of MONEY AS A FLUID is a precondition for the scene evoked here, in which the MONEY serves as a healing MEDICINE which is to be injected into the sleeping giant. 156 These metaphors are certainly quite appealing in their own right, but could not be considered any further as they lack any metonymic component. The only cases where metaphorical meanings were included in the metonymy study are those where the metaphors are built on metonymies which are still active in the metaphorical construal. Another problem is what I call for a lack of a better term ‘metonymic compounds’. The term compound is used in a very broad sense here and includes not only examples of what is traditionally called compound, but also instances where a metonymically used noun forms part of a larger syntactic group or phrase. 157 Since compounds are frequently not spelt as one word in English, the concordances contained many sentences in which the lexical items to be analyzed occurred as building blocks of larger syntagmatic combinations. The question is how to deal with them in a study of metonymy. One could argue that instances like the following examples are compounds and should not be taken into consideration: (101) American Football, brain drain, chicken soup, kitchen sherry, roof tax, Vietnam veteran It is typical of the semantics of compounds that their meanings are somehow specialized. A barman is ‘a man who serves drinks in a bar’ and not any man who happens to be in a bar or who works there (Schmid 2005: 121-122; cf. also chapter 4.2.1). In the same vein, one could hold that the meanings of the compounds listed above are specialized towards metonymic senses of their constituents: The head of American Football does not refer to the playing piece, but to the entire game. In brain drain, the meaning of the modifier refers to whole persons. A chicken soup does not contain whole chickens, but only the meat, etc. But these semantic specializa- 156 For a discussion of a similar example where MONEY is conceptualized as a FLUID , cf. Brône and Feyaerts (2005: 90-92). 157 Many linguists have addressed the question as to what constitutes a compound and established criteria which are meant to distinguish compounds from syntactic groups (e.g. Bauer 1998; Marchand 1969; Plag 2003; Schmid 2005). But so far none of them has produced an inventory of criteria which allows an unambiguous allocation of single cases to either of the classes. Recent approaches (e.g. Schmid 2005) therefore favour a prototype-style approach to the phenomenon - with compounds forming a category consisting of better and more marginal examples. 148 tions are of quite a regular nature and coincide with the metonymic mappings which are also found in instances in which the lexical items do not form part of a larger syntagmatic structure: PLAYING PIECE FOR ENTIRE GAME for football, BODY PART FOR PERSON for brain, ANIMAL FOR MEAT OF ANIMAL for chicken, LOCATION FOR PRODUCT MADE THERE for kitchen, and LOCATION FOR EVENT for Vietnam. Thus, there is reason to analyze such cases as metonyms. As has been shown in chapter 4.2.1, metaphor and metonymy often interact with word-formation proper (cf. e.g. Lipka 1998). And indeed, many of the lexical items investigated are often used in a regular metonymic sense as constituent parts of word-formations. Further cases in point are (among many others): • press ( CENTRAL INSTRUMENT FOR INSTITUTION ) as in press freedom, press campaign, press coverage, press release, etc. • diesel ( ATTRIBUTE OF OBJECT FOR OBJECT ) as in diesel driver, diesel owner, diesel buyers, diesel enthusiast, etc. • drugs ( CATEGORY FOR SALIENT MEMBER ) as in drugs trade, drugs crimes, drugs war, drugs dealer, etc. If all of these instances had been discarded, the overall picture would have changed dramatically for many lexical items. Many metonymic uses would have been lost. The mere fact that a metonym is combined with other morphemes to form a larger structure does not affect the metonymic nature of the single words involved. Actually, one could even argue that the use of a metonym to form compounds or other morphologically complex words, can be taken as an indicator of the conventionality of the metonym in question. The speech community would hardly use opaque and difficult to understand building blocks to form a new word, as many of the advantages of the morphological motivation of the new word would then be lost. 158 Instead, the members of a speech community will prefer constituent parts which are easy to understand. The more frequently a metonym occurs in compounds or other complex morphological structures, the higher the probability that it is a well-known and established part of the lexicon, i.e. conventional. In addition to losing much information about the actual frequency of metonymic uses, an exclusion of compounds would have entailed the vexing problem of having to distinguish between compounds proper and syntactic groups or phrases. While the above-mentioned cases can reasonably be considered compounds, other occurrences are more problematic in this respect. Many of the expressions occur, for example, in longer phrase-like 158 This is in line with Fischer’s (1998: 180-181) claim that the use of new words as building blocks of word-formations is an indicator of the institutionalization of the neologisms. 149 structures. Good examples are instances contained in my sample of White House: (102) White House Chief of Staff, White House Office of Management and Budget, White House Council on Competitiveness In all three examples, White House does not refer to the building, but to the institution. The expressions are quite long and certainly not compounds. They have a phrase-like structure, but are largely fixed and therefore resemble word-formations to a certain extent. If compounds proper had been excluded, this would have led to the question of whether such instances should also be discarded. However, with regard to a potential figurative motivation, these cases do not differ from compounds: The simple fact that we are dealing with larger structures here does not at all affect the status of the constituents they consist of. No matter whether the lexical items to be analyzed occurred on their own or as parts of any type of larger, more or less fixed syntagmatic combinations, they were counted as metonyms as long as their meanings were motivated by a conceptual metonymic mapping. A last point which has to be discussed here concerns the general structure in which the results of the metonymy study are presented in Appendix A. In what follows, this will be illustrated with the help of some examples. A relatively simple case is the form convent (cf. Table 4). Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) convent 524 (NN1) 12.60 PART FOR WHOLE : LOCATION FOR ENTITY THERE : LOCATION FOR INSTITUTION 12.60 100.00 Table 4: The results for convent The form convent returns 527 hits in the BNC. Three of them are instances annotated as NP0. They are constituents of a street name and were therefore discarded. The actual number downloaded and analyzed is thus 524 (NN1). This is indicated in the second column labelled Number analyzed. The word occurs only in one metonymic sense in the corpus, which is motivated by the mapping LOCATION FOR INSTITUTION (e.g. It wasn’t easy to lose the mental habits of the convent, GVT-652). But the column called Mapping 1 in Table 4 contains more information than that. It assigns the mapping to the larger group of PART - FOR - WHOLE relations and to its sub-group LO- CATION FOR ENTITY THERE . This elaborate structure of the information provided on the mappings is necessary in order to compare the frequencies of metonymic uses of words on different levels of generality. One can, for example, either compare the frequencies of all words which instantiate a LOCATION FOR INSTITUTION mapping or - one level higher - compare the 150 frequencies of all meanings which are based on a LOCATION FOR ENTITY THERE mapping. By not restricting the information to the lowest possible level, one can thus get interesting results even if a lowest-level mapping should occur only with a single word in the data set. In order to maximize the results, I therefore included different levels of generality - at least two and maximally four. Since the different sources from which the metonyms were originally collected (cf. chapter 5.1.2.1) do not always phrase the mappings in the same way, the wording in which the mappings are presented had to be standardized to allow the planned comparisons. For the most part, the wording was taken over from Leite (1994) or Radden and Kövecses (1999), but in some cases it had to be adapted, for example, to accommodate cases which are not mentioned in these sources. The columns labelled MYs total (%), M1 (total; %), and M1 (MYs, %) provide the actual frequency information. In general, the abbreviation M stands for ‘mapping’ and MY for ‘metonymy’. The samples analyzed for the single lexical items and expressions were not all of the same size. The largest number of instances downloaded and analyzed was restricted to 1000, but some of the words occur markedly less frequently in the BNC. The smallest numbers of occurrences were found for four wheel drive (30), set of wheels (14) and turtleneck (13). The overall number of the occurrences of other words or rather word-forms such as airlines (916, NN2) or kettle (848) is not too distant from the maximum number analyzed, while the total numbers of occurrences of others again, such as BMW (517) or convent (524, NN1), lies somewhere in the middle. In order to render the data sets comparable, the results are given as percentages. The column labelled MYs total (%) shows how often a given word is used metonymically, irrespective of the number of different mappings found for this word. All in all, 524 instances of the form convent were considered. Of these 524 occurrences, 66 conveyed a metonymic meaning. The percentage of all metonymic uses of the word is therefore 12.60%. Since all of the metonymic instances of convent are based on the LOCATION FOR INSTI- TUTION mapping, the entry in the column labelled M1 (total, %) contains the same information. This column is only of interest when the metonymic uses of a word were found to be motivated by more than one mapping. It indicates the relationship between the specific mappings and all meanings found for a single word. Since for convent I found only the literal meaning and metonymic uses supported by the LOCATION FOR INSTITUTION mapping, the percentage is the same as for the overall number of metonyms indicated in the third column. That only one metonymic mapping plays a role in all instances of convent analyzed also becomes evident in the last column labelled M1 (MYs; %). This column indicates how often a mapping contributes to the overall number of metonymic uses observed for a word. As the LOCATION FOR INSTITUTION mapping is the only one found for convent, it accounts for 100% of the metonymic uses. 151 The latter two columns are more informative if the situation is more complex than for convent. The results for kitchen illustrate this (cf. Table 5): Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) WHOLE FOR PART : WHOLE FOR PRINCIPAL PART ‘fittings/ fitments’ 4.60 49.46 PART FOR PART : SOURCE FOR PRODUCT : LOCATION FOR PRODUCT MADE THERE ‘cuisine’ 3.90 41.94 kitchen 1000 (NN1) 9.30 PART FOR WHOLE : LO- CATION FOR PEOPLE THERE ‘staff’ 0.80 8.60 Table 5: The results for kitchen - multiple mappings A query for kitchen returns 7656 hits in the BNC. The search was restricted to occurrences annotated as NN1 and a random sample of 1000 concordance lines was downloaded and analyzed. All in all, 93 concordances contain metonymic uses of the word. The total share of metonymic meanings in relation to the overall number of instances looked at is 9.30% (as indicated in the second column). The metonyms, however, do not all rely on the same mapping, but on three different ones. The two last columns read as follows: The ‘fittings/ fitments’ meaning occurred in 46 sentences (e.g. [...] but they have saved enough to buy a new kitchen, ED4-1592), the ‘cuisine’ meaning in 39 sentences (e.g. Also on offer are 25 French Country Kitchen Cookbooks, G2D-39), and the ‘staff’ meaning in 8 (e.g. Thanks also to everyone who helped the kitchen to do so well, ANM-999; cf. chapter 5.2.1.2). The share these meanings, or rather the mappings motivating them, have of all meanings is 4.60%, 3.90% and 0.80% respectively. By establishing a relationship between the single metonymic senses only and disregarding the literal instances, one arrives at the figures given in the last column. The raw number of all metonymic uses of kitchen is 93; these 93 instances are 100% of the metonymies. The ‘fittings/ fitments’ meaning accounts for 49.46% of all metonymies found, the ‘cuisine’ meaning for 41.94%, and the ‘staff’ meaning for 8.60%. If more than one mapping was found in the analysis of one word, the single mappings were arranged according to their frequencies. The mapping which sanctions the highest proportion of metonymies is listed first, the others follow in descending order. In many cases a simple analysis in terms of different mappings is not sufficient. As already discussed, 159 metonymies often form chains where 159 Cf. chapters 4.2.2.2, 4.3.2.2, and 5.1.2.1. 152 one metonymic mapping presupposes the other. BMW can be used to demonstrate this. My results gained for this word are summarized in Table 6 below. The single senses and mappings are illustrated in examples (103) to (105): Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) PART FOR PART : PRODUCER FOR PRODUCT : PRODUCER FOR PHYSICAL PRODUCT : PRODUCER FOR CONSUMER PRODUCT 62.67 89.26 WHOLE FOR PART : INSTITU- TION FOR PEOPLE 7.54 10.74 BMW 517 70.21 PART FOR PART : PRODUCER FOR PRODUCT : PRODUCER FOR PHYSICAL PRODUCT : PRODUCER FOR CONSUMER PRODUCT PART FOR PART : CON- TROLLED FOR CONTROLLER ‘driver’ 6.61 4.64 Table 6: The results for BMW - metonymic chains (103) He revealed that BMW is prepared to withdraw from the US rather than face the economic suicide such regulations would bring. (A6X-944) (104) [...] on to the forecourt of the house immediately to the left of where the BMW was parked. (AB9-2306) (105) Police stopped the BMW travelling at speed just before midnight with only one headlight on. (CH6-7333) All in all, 517 instances of BMW were analyzed. 363 of them are metonymic, i.e. 70.21% of all uses. 7.54% of all meanings are motivated by the mapping INSTITUTION FOR PEOPLE RESPONSIBLE (cf. 103). So far, the analysis is straightforward. The other cases are partly more complex. In 300 concordance lines the intended meaning is ‘car produced by the BMW company’ (cf. 104) and in 24 cases it is ‘driver of a car produced by the BMW company’ (cf. 105). The ‘driver’ meaning, which relies on the CONTROLLED FOR CONTROLLER mapping, presupposes the ‘car’ meaning, which relies on the PRODUCER FOR PRODUCT mapping. Without the PRODUCER FOR PRODUCT mapping the use of BMW to refer to a ‘person driving a BMW’ would not 153 be possible. The conceptual distance between the company and the driver of a product by the company is too large to consider the mapping a direct one leading from the company to the user of its products. It needs to be mediated by the idea that the company produces something which can then be used by somebody. The metonymic relationship leading from the ‘company’ to the ‘driver’ meaning can be illustrated as in Figure 15. P RODUCTION ICM C ONTROL ICM PRODUCER BMW ‘company’ PRODUCT / CONTROLLED BMW ‘car’ CONTROLLER BMW ‘driver’ Figure 15: BMW as a metonymic chain involving two ICMs The two metonymic mappings belong to distinct ICMs. The first takes place in the PRODUCTION ICM, the second in the CONTROL ICM. And the target of the first is the source of the second. This means that all instances of the CONTROLLED FOR CONTROLLER mapping are also instances of the PRO- DUCER FOR PRODUCT mapping. If such cases had been counted twice, i.e. as two metonymies, the relationships between the frequencies of the single meanings would have been blurred. The total amount of metonymies (MYs total %) would have been unduly increased. In the given case, the total amount of metonymies would have reached a percentage of 74.85%, which would mean that the literal usages only amount to 25.15%, while the actual percentage is 29.79%. In order to avoid this, metonymic chains are presented in the overview (cf. Appendix A) in terms of different levels of mappings. For BMW, 62.67% of the instances analyzed are based on the PRODUCER FOR PRODUCT mapping. This is indicated in the column labelled M1 (total; %). This percentage includes all cases in which this mapping is the only one that plays a role as well as those where it forms the basis for the secondary CONTROLLED FOR CONTROLLER mapping. The instances which feature both the PRODUCER FOR PRODUCT and the CONTROLLED FOR CONTROLLER mapping are then found in the third line of the analysis of BMW. The columns dealing with Mapping 1 do not bear any frequency information here, as this is already indicated in the first line, where the information on the PRODUCER FOR PRODUCT mapping is given. The mapping is only repeated to clarify the structure of the metonymic chain. The frequency information is then found in the columns dealing with Mapping 2. The CONTROLLED FOR CONTROLLER mapping, which occurs in 24 of the 517 cases analyzed, accounts for 4.64% of all meanings (M2 - total; %). And 6.61% (M2 - MYs; %) 154 of all metonymies observed in the analysis of the lexical item were of the type CONTROLLED FOR CONTROLLER . The relationship between the columns describing metonymic chains is not one of addition, but of subtraction. The information given on the level of Mapping 1 always accounts for the total amount of metonymies found and for their relationship to the literal instances (cf. Table 7): M1 (total; %) M1 (MYs; %) PRODUCER FOR PRODUCT : 62.67% 89.26% INSTITUTION FOR PEOPLE : + 7.54% + 10.74% MYs total (%) = 70.21% = 100.00% Table 7: Information on the first level of mappings in Appendix A All the instances of metonyms which involve more than one mapping and therefore different levels of mappings are already included here, as they are all metonymic on the first level of mappings. The actual frequency of the cases which only rely on the PRODUCER FOR PRODUCT meaning can then be easily calculated. If 4.64% of the meanings are based on the metonymic chain just described and the PRODUCER FOR PRODUCT meaning is found in 62.67% of all cases, then the overall amount of those cases which only feature a PRODUCER FOR PRODUCT meaning is 58.03% (cf. Table 8). Mapping - all meanings Mapping - all metonymies PRODUCER FOR PRODUCT (all) 62.67% 89.26% PRODUCER FOR PRODUCT and CONTROLLED FOR CONTROLLER - 4.64% - 6.61% PRODUCT FOR PRODUCER only = 58.03% = 82.65% Table 8: Extracting the frequencies of single mappings from metonymic chains The results can then be linked back to the overall number of metonymic instances found and to the share the single mappings have of the overall amount of meanings and metonyms as shown in Table 9 below. The advantage of not splitting up the number of occurrences of level-one mappings into those where the mapping is the only one and those where the mapping is the basis for a second one is obvious. One can avoid indicating two different percentages for one and the same mapping. Thereby the percentages given for the single mappings and also the frequency-dependent ordering of the single mappings becomes more informative than it otherwise would. And since the single constituents of the metonymic chains are indicated, the information about the relative frequency with which a mapping is the only one motivating linguistic instantiations and the relative frequency with which it is used as a basis for another mapping is not lost. 155 Mapping - all meanings Mapping - all metonymies PRODUCER FOR PRODUCT only 58.03% 82.65% PRODUCER FOR PRODUCT and CONTROLLED FOR CONTROLLER + 4.64% + 6.61% INSTITUTION FOR PEOPLE + 7.54% + 10.74% = 70.21% (MYs total) = 100.00% (all metonymies) Table 9: Linking back single mappings to the overall amount of meanings and metonymies BMW, is quite a simple case of a metonymic chain. Others are somewhat more intricate. For the noun football, for example, all metonymic uses found in the BNC involve the mapping PLAYING PIECE FOR ENTIRE GAME . But two slightly different secondary mappings rely on this frequent ‘game’ meaning as their basis: In example (106), the game football ultimately stands for the persons who play it, since the lexical item trick implies an agent. A game cannot perform tricks, only the players can. On a more general level, the secondary mapping that applies in (106) can be formulated as ACTIVITY OF PERSON FOR PERSON . As opposed to this, example (107) illustrates one of the cases where PLAYING PIECE FOR ENTIRE GAME is combined with the secondary mapping CENTRAL INSTRUMENT FOR PROFESSION . (106) ANDREW TURNER, youngest kid in the Premier League, pulled off one of football’s oldest tricks and set White Hart Lane alight. (HAE-1794; original caps) (107) Football may be an athletic profession but it’s also a borstal for wayward genius. (B1L-72) A very complex case is glasses. The most basic mapping, MATERIAL OF OB- JECT FOR OBJECT is found in all cases of metonymic usage. But, ultimately, it underlies several different meanings. In the most straightforward examples, it is the only mapping involved and leads to sentences where glasses means ‘containers’ (cf. 108). This can be exploited to form metonyms based on the CONTAINER FOR CONTENT mapping, where glasses actually refers to the content (cf. 109). However, the mapping MATERIAL OF OBJECT FOR OBJECT can also support the second-level mapping PRINCIPAL PART FOR WHOLE and thus lead to the meaning ‘spectacles’ (cf. 110). And this meaning can form the basis for further metonymic extensions (cf. 111). (108) Her father produced a bottle of Calvados and four glasses. (A61-2303) (109) Urquhart gulped down three glasses almost without a pause. (G15-115) (110) She was blind as an owl in daylight without glasses [...] (A0l-304) (111) The dark-green glasses followed her progress indoors. (J10-4516) The last example involves three different mappings: First, MATERIAL OF OBJECT FOR OBJECT which is the basis for the second mapping, i.e. the PRINCI- PAL PART FOR WHOLE mapping that results in the meaning ‘spectacles’. The third mapping then leads from the PART FOR WHOLE mapping to the ‘person 156 wearing the glasses’. This mapping is recorded in Appendix A as ATTRIB- UTE OF PERSON FOR PERSON : COSTUME FOR WEARER . Actually, however, this is a simplification of the situation at hand. In example (111) the person who is wearing the glasses does not literally follow the woman’s progress, the person is only gazing in the direction she takes. In fact, the glasses stand for the person’s eyes and the eyes for the gaze. The COSTUME FOR WEARER metonymy is actually a COSTUME FOR PART OF WEARER metonymy which goes hand in hand with another metonymy where the responsible body part stands for the activity associated with it. This case is extremely complex and the single mappings can hardly be kept apart. Therefore I preferred to use a simplified version. 160 This chapter has provided an overview of the methodology used in the metonymy study. It was shown how the metonymies were identified and how various problematic cases were treated. In the next chapter, I will turn to the metaphor study. Parallel to what has been said regarding the metonymy study, I will introduce how the sub-corpora were created, discuss the methods used to extract the metaphorical uses from these sub-corpora, illustrate some problems encountered during the analysis, and explain how the results for the metaphor study are presented in Appendix B. 5.2.2 Metaphorical senses 5.2.2.1 The sub-corpora The study of the potentially metaphorical expressions basically followed the same procedure as described above for the metonyms. The maximum of concordance lines taken into account was again 1000. Approximately half of the word-forms I concentrated on occur fewer than 1000 times in the BNC. For example, the query for crossroads, which is annotated as NN1 in all cases, returned 482 hits, and the search for grasped, annotated as VVD, VVD-VVN, VVN-VVD, and VVN-VVD, 161 returned 616 hits. In these cases, it was possible to carry out a comprehensive analysis of all instances in the BNC. Many of the other word-forms the study embraces are highly frequent. The amount of data could thus often not be reduced to 1000 lines by limiting the search to potentially relevant grammatical forms through a part-of-speech query. In addition, it was necessary to download random sets of these samples. Once again, the part(s) of speech the search was limited to was largely determined by what I found in my sources. Escape, for instance, was taken 160 Another, more practical reason for this decision was that an indication of all the different levels involved would have been very difficult to accommodate in the table. And since this is the only case where more than three levels of mapping would have been necessary for an accurate description, it seemed preferable to give an easily comprehensible and clearly structured overview of all metonymies, even if this meant simplifying one of the cases. 161 VVD stands for the past tense of a lexical verb and VVN for the past participle. 157 over from the Master Metaphor List as a verb in the example sentence I can’t escape from that memory (Lakoff, Espenson and Schwartz 1991: 43). A partof-speech query for escape shows that the form occurs with five different annotations: NN1 (singular noun), NN1-VVB (singular common noun or base form of verb), VVB (base form of verb), VVB-NN1 (base form of verb or singular common noun), and VVI (infinitive of lexical verb). Those instances unambiguously annotated as nouns (NN1) and those which most likely function as a noun (NN1-VVB) were discarded and the search was restricted to examples of the remaining classes (VVB, VVB-NN1, VVI). A sub-corpus of 1000 hits belonging to these classes was downloaded and used as a basis for the search for metaphorical uses. As with the metonymies, the restrictions (if any) to certain parts of speech are specified in the Appendix. The table also indicates the total number of the instances analyzed for a single form. In general, the parts of speech taken into account for the metaphor study are less severely restricted than for the metonymy study. For the metaphors it would not have been useful to, say, restrict the search for escape to infinitives only and to exclude the base form. The study had to be restricted to one ‘outward appearance’ of the lexical items in any case, and any further restriction to a single type of the multiple grammatical functions of this form would have been counter-productive. 5.2.2.2 Extracting and organizing metaphorical senses Once the sub-corpora had been downloaded and were ready for analysis, the metaphorical uses of the lexical items had to be extracted. Once again, I was first and foremost looking for instances of the mappings that were realized in the examples I took over from the sources of my lexical items. For crossroads, for instance, the example sentence provided in the Master Metaphor List is She’s at a crossroads in her life (Lakoff, Espenson and Schwartz 1991: 37). According to Lakoff, Espenson and Schwartz, the mapping underlying this expression belongs to the larger group of EVENT STRUCTURE mappings and has the following complex form: LONGTERM PUR- POSEFUL ACTIVITY IS A JOURNEY : LIFE IS A JOURNEY : MAJOR CHOICES ARE CROSS- ROADS . But the lexeme crossroads also turns up in other sub-mappings belonging to the larger frame of LONGTERM PURPOSEFUL ACTIVITY IS A JOURNEY , such as LOVE IS A JOURNEY and CAREER IS A JOURNEY (1991: 37-38). This partly corresponds to what was found in the corpus. There are many instances belonging to the general mapping LONGTERM PURPOSEFUL ACTIVITY IS A JOURNEY such as (112) to (114): (112) He had reached a crossroads in his life. (A7H-1035) (113) One notable absentee from the squad is Lee Crooks, once rated invaluable by Reilly, who appears to have come to a career crossroads. (A4P- 279) 158 (114) Mr Mendham says: ‘Small firms are at a critical crossroads, with cashflow drying up and less willingness by the banks to lend. (AHT-281) Example (112) is possibly a prototypical instance of the mapping LIFE IS A JOURNEY , while (113) belongs to the group CAREER IS A JOURNEY . (114) is slightly more difficult to deal with. Here, COMPANIES are conceived of as LIVING BEINGS who have to make decisions. 162 One could hold that this is a sub-group of the mapping LIFE IS A JOURNEY , where LIFE is BUSINESS LIFE and the CHOICES made are therefore BUSINESS DECISIONS . Or it could be argued that the idea of career is taken over from the mapping CAREER IS A JOURNEY , where CAREER then refers to a superindividual level, i.e. to the success of a company. But the mapping LONGTERM PURPOSEFUL ACTIVITY IS A JOURNEY is not the only one I found in my data. Consider the following examples: (115) The most prolific refining and distribution areas [sic! ] is Ribat, which lies at the crossroads of three countries - Afghanistan, Pakistan and Iran (A89-139) (116) Bohemia, unlike Britain, was historically the crossroads of Europe. (AE8- 1068) (117) It was only a few years since Schiaparelli had been a living pan-cultural mix, a cosmopolitan crossroads of the solar system, where all the client races of Capella could co-exist in noisy harmony [...] (CJA-849) (118) [...] we are not ourselves, we are crossroads, meeting places, points on a curve, we cannot exist independently for we are nothing but signs, conjunctions, aggregations. (FB0-1628) Examples (115) to (117) share one idea: Any central or important location can be called a crossroads. This sense is quite common and also listed in dictionaries like the LDOCE4 (s.v. crossroads 3). The metaphor involved may be less interesting from a conceptual point of view, since the target domain is not an abstract one. Nevertheless, the descriptions are not literal and were therefore included in my results as instances of a mapping phrased as CENTRAL PLACES ARE CROSSROADS . The locations which are said to be crossroads in the examples of this mapping are ascribed certain properties which are taken over from the source concept. The places are certainly not deserted ones, they are full of busy people. Different people coming from different directions meet there for a short while and then leave again. And places where many people meet are potentially important: Meeting people opens up new opportunities, but also dangers. The scene inherits a certain dynamic from the CROSSROADS concept, which is otherwise hard to express in a concise way. The last example given above, (118), is different. Here, individuals themselves are said to be crossroads. The underlying idea, however, is not 162 This is apparently based on the quite common metonymic reasoning pattern that links companies or, on a more general level, institutions and the human beings working for them, i.e. INSTITUTION FOR PEOPLE RESPONSIBLE . 159 too dissimilar to (115) to (117): People are places where people meet. What is emphasized in this rather poetic example is that people are not individuals in the most literal sense - they cannot exist without other people. Certainly, this example is linked to the mapping LIFE IS A JOURNEY . People are meeting places because their lives are journeys. But the crossroads do not stand for major choices here - the established metaphor is exploited for a novel elaboration. What has been said about the analysis of the lexeme crossroads raises some important methodological questions. The first has already been mentioned above: Should instances like (115) to (117) be considered at all in an analysis of conceptual metaphors, because the target domain is not an abstract one in this case? The answer is a straightforward yes. The instances are obviously metaphorical and the mapping CENTRAL PLACES ARE CROSS- ROADS is not particularly infrequent either. 25 out of the total of 482 concordance lines containing crossroads in the BNC rely on it, i.e. 5.19% of all meanings. The mapping may not be a prototypical one, since the target concept is fairly concrete, but the Master Metaphor List includes similar mappings. An example is COMPETITORS OR ANTAGONISTS ARE PREDATORS in Lawyers are real sharks (Lakoff, Espenson and Schwartz 1991: 47), where the targets belong to the larger concept of PEOPLE , i.e. also to a concrete domain. Considering that many mappings are said to be supported by our experiences with ourselves and our bodies, human beings should be something we know much about. However, the domain of PEOPLE is not only an often exploited source domain, as in THEORIES ARE PEOPLE (e.g. The theory was eventually retired; Lakoff, Espenson and Schwartz 1991: 112) or MA- CHINES ARE PEOPLE (e.g. This battery has a life of 3 years; Lakoff, Espenson and Schwartz 1991: 181). It is also frequently the target of metaphorical construals, for example in PEOPLE ARE BATTERIES , PEOPLE ARE MACHINES , PEOPLE ARE ANIMALS , PEOPLE ARE PLANTS (cf. Lakoff, Espenson and Schwartz 1991). 163 The mappings may be leaner in these cases, since only a few aspects are highlighted. For the example crossroads, the highlighted aspects have already been mentioned and for Lawyers are sharks only some fairly stereotypical and maybe purely alleged aspects of the behaviour of predators are mapped onto the target. But the mere fact that the mappings are leaner does not preclude their entrenchment (cf. also Ungerer and Schmid 2006: 125-126; Ruiz de Mendoza 2000: 112). 163 This demonstrates that, in fact, at least some conceptual metaphors are reversible (cf. e.g. MACHINES ARE PEOPLE vs. PEOPLE ARE MACHINES ). This, however, does not necessarily stand in contrast to the general unidirectionality of metaphorical mappings themselves, since in each of the two cases the mapping proper, i.e. the ‘transfer’ of knowledge from one domain onto another has a clear direction and works only from source onto target, but not the other way round at the same time and for the identical linguistic expression used to realize the respective mapping (cf. also Kövecses 2002: 24-25). For a short discussion of the differences between MACHINES ARE PEOPLE and PEOPLE ARE MACHINES cf. also Evans and Green (2006: 297). 160 The second question concerns novel metaphors like example (118) and whether they should form part of an analysis which focuses on conventional figurative language and thinking. Again the answer is yes. When trying to establish criteria which mark conventional cases, it is certainly also worth looking at more unconventional ones by way of comparison. My aim is to give an overview of metaphor use in authentic discourse and this does not exclude more creative cases - even if they are only loosely connected to better known conceptual mappings, as is the case of (118). Excluding such cases from the outset would mean accepting preconceived ideas of the conventionality of a metaphor. Only an objective analysis can reveal whether there are perhaps any quite frequent mappings which have been overlooked so far. Therefore, novel and creative examples are also mentioned in Appendix B. The third issue that can be illustrated with the help of the abovementioned examples has to do with the usefulness of a very fine-grained analysis. As has been shown, the mapping LONGTERM PURPOSEFUL ACTIVITY IS A JOURNEY has various sub-groups, such as LIFE IS A JOURNEY and CAREER IS A JOURNEY , which are all connected with the mapping MAJOR CHOICES ARE CROSSROADS . Accordingly, the overview of the results should contain subsections, i.e. separate lines for all of them. But is this really necessary and, more importantly, is it helpful for the aims of the study? For relatively simple cases like crossroads, such a fine-grained analysis is at least possible. Even if all sub-groups had been listed separately, the lexeme crossroads would not have had more than five mappings: • LONGTERM PURPOSEFUL ACTIVITY IS A JOURNEY : COMPANIES / GOVERN - MENTS ARE LIVING BEINGS : MAJOR CHOICES ARE CROSSROADS • LONGTERM PURPOSEFUL ACTIVITY IS A JOURNEY : LIFE IS A JOURNEY : MAJOR CHOICES ARE CROSSROADS • LONGTERM PURPOSEFUL ACTIVITY IS A JOURNEY : CAREER IS A JOURNEY : MA- JOR CHOICES ARE CROSSROADS • CENTRAL PLACES ARE CROSSROADS • PEOPLE ARE CROSSROADS However, one could reasonably argue that the first three are sub-groups of the more encompassing LIFE IS A JOURNEY metaphor, since business matters and people’s individual careers form part of life as a whole. Moreover, they all share the last mapping, i.e. MAJOR CHOICES ARE CROSSROADS . The choices may be different in all three cases, but the underlying mapping is the same. As a look at Appendix B confirms, I chose the latter option of summarizing as many different cases as possible under one common mapping. While some details may be lost with this strategy, it has many important advantages. First, the overview of the results becomes more comprehensible and has a clearer structure. Second, it enhances the comparability of the different frequencies of single mappings. The figures for a more com- 161 prehensive category, such as LIFE IS A JOURNEY : MAJOR CHOICES ARE CROSS- ROADS are much more significant than the single figures for the different sub-cases. As a result, it is easier to contrast the frequency of a more general category to other mappings which are realized by the same lexeme as well as to compare it to different or similar mappings which are realized by other lexemes. In short, the decision to work with larger rather than with more fine-grained classes was made in order to avoid what has been called the ‘principle of diminishing returns’ (Lyons 1968: 152-153, 164; cf. also Lipka 2002: 166-167). The point is that by setting up ever finer subclassifications, the results are no longer generalizable and lose much of their theoretical value. As far as possible, sub-groups were thus assigned to larger classes. For some lexical items, this worked out well and only a few mappings had to be posited. But in general, the number of mappings found for one wordform exceeds those for metonymies. The mappings are much more varied and for some lexical items the meanings are so diverse that many different mappings had to be distinguished. An example is erupted. Of the 518 instances analyzed 375 were found to be used metaphorically. These had to be assigned to 15 different mappings. The most frequent ones are given in Table 10 and illustrated with examples from the BNC (119-123). Wordform Number analyzed Mapping Frequency of mapping (%) erupted 518 EVENT STRUCTURE : ACTION PHYSICAL AGGRESSION IS A HOT FLUID 28.19 MENTAL EVENTS : DEBATE DISAGREEMENTS / PROBLEMS ARE HOT FLUIDS 19.69 EMOTIONS : EMOTIONS ARE ENTITIES WITHIN A PERSON : EXTREME EMOTIONS ARE HOT FLUIDS 8.88 MENTAL EVENTS : PERCEPTION SOUNDS ARE A HOT FLUID / VOLCANO 7.53 OTHER : LIVING BEINGS LIVING BEINGS ARE HOT FLUIDS / VOLCANOES 3.47 Table 10: The most frequent mappings realized by the word-form erupted (119) Violence has erupted again on an estate [...] (K1D-2879) (120) Minutes later the inevitable argument erupted. (ANK-1227) (121) Intense feelings erupted about everything. (FU7-494) (122) [...] the hall erupted into an ovation [...] (CH2-10771) (123) Suddenly Bridget erupted into the room [...] (HTR-917) The first four of them are related in a way. Physical aggression and disagreements can be regarded as having a cause-result relationship with emotions. And human sounds like cheers, laughter or sobs are also based on or rather caused by emotions. For this reason, I chose to treat these cases as mappings with the source HOT FLUIDS to mark their emotional ground- 162 ing. While the underlying idea of HOT FLUIDS IN A CONTAINER is evident and also commonly used to describe the third group, i.e. the pure instances of emotion metaphors, the emotional grounding and therefore also the idea of HOT FLUIDS IN A CONTAINER is somewhat less obvious for the other cases, because is at work on a deeper, more ‘hidden’ level. However, the first three mappings in Table 10 differ in some important respects. P HYSICAL AGGRESSION is quite a concrete target as opposed to EMOTIONS . And on a scale reaching from concrete to abstract, DISAGREE- MENTS fall somewhere in the middle, between the abstract domain of EMO- TIONS and the concrete one of PHYSICAL VIOLENCE . They belong to different overarching frames: As a kind of action, PHYSICAL AGGRESSION belongs to what the Master Metaphor List calls EVENT STRUCTURE . D ISAGREEMENTS belong to the group of MENTAL EVENTS and EMOTIONS form a category of their own. These overarching groups, which already served as a guideline for the choice of the lexical items (cf. chapter 5.1.2.2), are simply too different to allow a coherent treatment of all cases in one category. The fourth mapping in Table 10, embraces examples where the targets are sounds. They all belong to the more general frame of PERCEPTIONS . However, this group also includes instances of non-human sounds, like the phrase sound of gunfire erupted (HJD-1531). Such cases are much less directly connected to emotions than the human sounds (cf. 122). The mapping is therefore phrased in a slightly different way here as SOUNDS ARE A HOT FLUID / VOLCANO . The main idea conveyed by all the instances of the soundrelated metaphors seems to be the suddenness, and maybe the unexpectedness, of the event. At least for the non-human sounds, the source domain VOLCANO thus might be more apt to describe the situation. Ultimately, however, the two phrasings do not differ in any important respect. No matter whether HOT FLUIDS or VOLCANOES are chosen as the source domain, the basic underlying idea is always the same: A container of some kind holds a substance of some kind, and due to pressure building up the container suddenly can no longer hold the substance. A volcano is perhaps the prototypical ‘container’ to which this applies, but to what extent this prototype or at least very good example of the category of containers which contain substances under pressure really motivates the linguistic expressions cannot be judged here. 164 The last mapping illustrated above is less frequent than the first four, but still accounts for 18 out of 518 cases, i.e. 3.47% of all meanings. The 164 Actually, the decision as to whether one uses HOT FLUIDS or VOLCANOES as a source domain in these cases depends on whether one concentrates on the container or on the content, i.e. is a question of the perspective one takes. For he erupted into loud, desperate sobs (CCA-406), one can either focus on the ‘container’, in this case a person, or on the ‘content’ erupting from it, in this case the sobs. For the first case, a description as A PERSON IS A VOLCANO would be appropriate, for the second, the mapping A SOUND IS A HOT FLUID is more convincing, since the invariance principle forbids mappings that do not uphold the topology of the source domain (cf. chapter 3.2.1). 163 mapping is also less rich than the first three: When living beings erupt from somewhere or into someone’s life, the main idea communicated is, again, the suddenness of the event. And again, it is difficult to decide whether the source concept is a HOT FLUID or rather a VOLCANO . Therefore the mapping posited here is ambiguous with regard to the source concept exploited. The same holds true for the mappings illustrated below (124-128), which occur markedly less frequent than those given in Table 10: (124) VEHICLES ARE HOT FLUIDS / VOLCANOES : The noise intensified as the locomotive erupted between two large buildings [...] (EEV-363) (125) IMAGES ARE A HOT FLUID / VOLCANOES : On every side images of love and violence erupted from glass-fronted posters advertising coming attractions. (HTT-2129) (126) NEWS IS A HOT FLUID / VOLCANO : [...] when the news of cold fusion erupted (CER-36) (127) BEAUTY / DECORATION IS A HOT FLUID / VOLCANO : There the new bride house erupted in ornate concrete splendour. (CDX-2260) (128) LIGHT IS A HOT FLUID / VOLCANO : [...] the light gathered into a curling ribbon and solidified, erupted into a rippling curtain of sparkles. (G1M- 3432) These cases are ultimately responsible for the large number of mappings which had to be posited. Mappings which occur less than ten times in the BNC account for ten of the 15 mappings found for erupted. They are much too diverse, however, to put them into one class so that a sub-classification could not be avoided here. Let me summarize briefly at this point. The mappings posited in Appendix B are guided by two principles. First, everything which could be usefully subsumed under one heading is put into one category, even if the single instances differ to a certain respect. The EMOTION metaphors for erupted, for example, could have easily been split up further into more positive feelings, such as JOY , and those with more negative connotations, such as ANGER . Second, the decision about whether it is possible to treat different groups of metaphors under one and the same heading is dependent on the larger classes of metaphorical mappings set up by the Master Metaphor List (Lakoff, Espenson and Schwartz 1991). When two classes of metaphor differ with respect to the larger frame they belong to (such as EVENT STRUC- TURE or MENTAL EVENTS ) they are treated separately. 5.2.2.3 Problematic cases The decision about whether a given concordance line is an instance of a metaphorical or literal use of the lexical item in question was based on a consideration of the item in context. In many cases, the metaphors were easy to identify and clear instances of frequently described mappings. As the context indicates, (129) is an example of IDEAS ARE FOOD : UNDERSTAND- ING IS DIGESTION . In (130) a similar idea is transferred to the realm of com- 164 puters: MACHINES ARE PEOPLE , DATA ARE IDEAS , and the INTEGRATION OF DATA in a computer system is equalled with human understanding, which is metaphorically construed as DIGESTION . This stands in marked contrast to literal uses of the verb as in (131). (129) He had been given so many things to digest, not the least that it was possible Celia might have a father still living. (CDE-1958) (130) [...] most computers [...] can digest data in only 8-bit streams. (B7F-175) (131) Diagnosed as suffering from gastroschisis, a major bowel disorder, at birth, she was unable to digest any food at all. (A0G-2197) The food-context marks the usage of digest in (131) as truly literal. However, especially imaginative prose often exploits the fact that words have both a literal and a figurative meaning. A good example is the following: (132) Really, he could use half an hour to digest the gobbets of this person. (CJJ-2210; italics original) At first glance, the sentence seems to be literal. At least this is what the word gobbets suggests. But, as a look at the larger context reveals, the meaning oscillates between the literal idea of digesting food and the metaphorical idea of understanding someone. That the intention is not an entirely literal one is already indicated by the fact that the verb is italicized in the original source. And the larger context shows that in the given sciencefiction story, a person can read another person’s mind by eating their brain. So, actually, it is the idea of understanding which is of paramount importance here. It is often not enough to look at the concordance lines in isolation. The wider context has to be considered as well in order to avoid mistakes. Further problematic cases are of a more theoretical nature. The first issue concerns the knotty problem of where metaphoricity ends and literality begins. A study of the conventionality of figurative language can certainly not exclude cases which are no longer felt to be figurative by the average language user, since exactly those instances are of vital interest. Thus, the general policy was to include anything which is based on a metaphorical mapping linking two different domains of experience. However, there are always instances where a decision as to whether they are literal or metaphorical is difficult to make. This can be amply illustrated with the lexeme warm. The sense I was originally looking for is the one supported by the mapping AFFECTION IS WARMTH . Many of the instances of warm in the BNC are based on this mapping (e.g. warm welcome, warm feelings, warm person, etc.). Other instances are less clear. Compare for example: (133) His handshake was warm and firm [...] (CCW-1840) (134) I leaned back against his warm shoulder and closed my eyes. (CEX-538) Both examples oscillate to different degrees between a literal and a nonliteral, metaphorical interpretation. Both ideas, literal warmth and metaphorical affection probably work together in these cases, so that the actual 165 meanings are a mixture of the two. The meanings realized in the given contexts are underspecified to a certain extent. However, for (133) a metaphorical understanding seems to be more likely than for (134), since a warm handshake is commonly associated with a friendly, i.e. affectionate welcome. A warm shoulder, however, as in (134), could very well also receive a literal interpretation by some speakers, even though the context makes it likely that affection also plays a role. 165 Since it is impossible to decide upon the issue and the interpretations might very well vary from speaker to speaker in such cases, underspecified meanings which are neither clearly literal nor clearly metaphorical but seem to convey both ideas at the same time were not counted as instances of metaphorical mappings. Another problem related to the distinction between literal and figurative meanings are dead metaphors (cf. chapter 4.3.1.2). While as a general rule all metaphors, no matter how conventional, were taken into account, some cases posed problems in this respect. An example is the form killed. In some texts what is being killed is not humans or other prototypical instances of living beings but body cells (cf. 135) or germs (cf. 136): (135) [...] if up to 80 per cent of the cells are killed off with a drug the embryo can regulate and quite normal mice can develop [...] (ASL-680) (136) [...], comforting herself with the thought that boiling water killed germs of all sorts. (ASE-433) At first glance, there seems to be no reason to treat such instances as figurative. Any living being can be killed as well. However, consider the obvious nonliterality of a sentence like I killed a tree. Trees are certainly living beings, but when syntagmatically combined with the verb kill other things seem to play a role as well. The most typical instances of entities which can be killed are intelligent living beings. Is it still possible to say that germs and body cells can be killed in a literal way? Or can one only do so figuratively? A clear answer cannot be given to this question. Perhaps, the original motivation is a metaphorical one, but the meaning of kill when combined with germs, body cells, and the like seems to be fairly literal - a stone dead metaphor at best. Cases like this were therefore not considered any further and not included as instances of metaphorical mappings. A lexical item where similar problems occurred is core. The most literal meaning is ‘central part of fruit’. In plainly metaphorical instances the word is not used in a spatial sense, but relates to the importance of something. In many cases, however, it is used in a spatial meaning, but not with reference to fruit (e.g. core of the city, core of the reactor). While it could be argued that the core of a city or the core of a reactor are also the most important parts and that therefore importance plays a role as well, one cannot be certain in all cases. If the context indicated that the meaning conveyed 165 The only way to find out about this would have been to ask a representative number of native speakers in all cases of doubt. But due to the amount of data analyzed this was not feasible. 166 by core is one of interiority and not one of importance, the instances were excluded and judged as literal. Subjective and perhaps partly arbitrary decisions as to which instances are metaphorical or not can unfortunately not be entirely avoided, but the boundary had to be drawn somewhere. Two last points must be raised here. The first concerns the interaction of metaphor and metonymy and the second the issue of conjoined phrases. Since metonyms are mostly found among nouns and the metaphor study chiefly contained items belonging to other word-classes, the problem of metaphor-metonymy interaction hardly ever occurred. Sometimes the lexical items were used in a metonymic rather than in a metaphorical way and were thus excluded from further consideration (e.g. thirst in Mr Oliver Reed, the thirst; BNS-160). The second issue, combinations of more than one metaphorical construal within one and the same sentence, was encountered much more often. Examples like the following illustrate the problem: (137) [...] that so much time, money and energy is poured into books that the world doesn’t need [...] (FSV-702) (138) An enormous amount of time and money has already been poured into this challenge. (H7W-810) (139) [...] and all of the love and care and time I had poured into her [...] (CCW-142) Many different things can be construed as FLUIDS , such as EMOTIONS , TIME , MONEY , etc. And as the examples show several of these metaphorical construals can be combined within one sentence or phrase. In order not to mask the real numerical relationship between the metaphorical usages and the literal ones, such cases could not be counted more than once, i.e. it was not possible to count them as instances of more than one mapping. The principle adhered to is the same as already described for the metonymy study: Only the first of such a series of objects was taken into consideration. (137) and (138) were thus counted as examples of TIME IS A FLUID , and (139) as an example of EMOTIONS ARE FLUIDS . Parallel to this I considered only the first of a series of two or more objects when a literal and a figurative object co-occurred in the same phrase. Instances like [...] you are wasting both time and money (B08-1908) were therefore treated as metaphorical. 5.2.2.4 The presentation of the results The way in which the results of the metaphor study are presented in Appendix B is basically the same as for the metonymy study. Compare the extract in Table 11 (below). The first column contains the word-form searched for in the BNC. The second indicates restrictions (if any) as to which parts of speech were analyzed and gives information about the total number of concordance lines taken into account. The third column indicates the total percentage of metaphorical uses found as compared to the literal ones. The column labelled Mapping provides the conceptual analysis of the metaphorical instances. As in the metonymy study, each line repre- 167 sents one mapping, and the abbreviation M in the last two columns again stands for ‘mapping’. These two last columns show the frequencies of the single mappings. The column labelled M - total (%) relates the frequency of a single mapping to the total amount of cases analyzed and the column labelled M - MR (%), where MR stands for ‘metaphor’, relates the frequencies of the single mappings to each other. Wordform Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) thirst 343 30.03 EMOTION : DESIRE IS HUNGER (includes special case: IDEAS ARE OBJECTS : LEARNING IS ABSORBING ) 27.41 91.26 OTHER : MACHINES MACHINES ARE HUMANS : FUEL CONSUMPTION IS THIRST 2.33 7.77 OTHER : PLANTS PLANTS ARE HUMANS 0.29 0.97 Table 11: An extract from the results of the metaphor study - the word-form thirst One difference to the table for the metonymy study in Appendix A concerns the structure of the mappings. The mappings provided for the metaphors usually consist of different levels of abstraction (e.g. TREATING AN ILLNESS IS FIGHTING A WAR : THE IMMUNE SYSTEM / A PRECAUTION IS A DEFENSE for defences or IDEAS ARE FOOD : UNDERSTANDING IS DIGESTION for digest), just like the single mappings described for the metonymies in Appendix A. However, the metonymies are often analyzed as chains, while this does not occur in Appendix B for the metaphors. The reason is that cases where entirely different mappings must have occurred in order to account for one and the same meaning are not nearly as numerous for metaphors as for metonymies. Hence it is usually enough to regard them as instances of one mapping, which can be described on different levels of generality. In the following chapters, I will present the results of my corpus study of metonymic and metaphorical uses of my target words in the BNC. Chapter 6 is concerned with the results of the metonymy study, and chapter 7 with the metaphors. For both I will first provide rather sketchy overviews of the findings, before turning to more detailed analyses in the later parts of the chapters. 169 6 Evidence for the conventionality of metonymic meaning and thought 6.1 Frequent and less frequent metonyms: An overview What do the data from the BNC reveal? How frequent are metonymies? All but one of the target items analyzed returned some metonymic readings. The only word which could not be found in a metonymic use in the BNC is ketchup, where I was originally looking for CONTAINER FOR CONTENT meanings. Seat did not occur in the metonymic sense in which it was taken over from the source (i.e. SEAT FOR PERSON OCCUPYING IT ), but in various others. Also of interest is that none of the instances is a truly ‘dead’ metonymy. For all instances literal uses were found as well. To begin this preliminary overview of the results of the metonymy study, I would like to address the question as to what extent the linguistic manifestations regularly used to illustrate the (un)conventionality of metonymy in the relevant literature are apt to do so. An oft-quoted example is Shakespeare as in I’m/ He’s reading Shakespeare. This metonym is indeed quite common: 34.90% of all uses, i.e. more than one third of all utterances which contain the proper noun, are instances of the mapping AUTHOR FOR WORKS . Other frequently mentioned examples, like BMW ‘car’ or America ‘United States’ are with 62.67% and 62.50% respectively even more conventional. Uncommon metonymic construals do, as Radden and Kövecses (1999: 58) point out, often not rely on a “clearly specifiable type of conceptual relationship”, since the relationship between vehicle and target is not a stable one. My study embraces the instance appendicitis for ‘patient suffering from appendicitis’. The query returned only 61 hits for appendicitis so that the representativeness of the results is much smaller than for many other word-forms. However, one clear example of the metonymic pattern AT- TRIBUTE OF PERSON FOR PERSON could be found: (140) Talking about patients as ‘the spleen in bed 9’, or ‘the appendicitis in the corner’, or in a maternity unit calling everyone ‘Mother’ removes individual characteristics, neutralizes them and protects the nurse from emotional demands. (G0T-690) As is evident from the context, appendicitis is not used in a concrete situation for the purpose of referring to a person, but in a metalinguistic discussion of the reasons why such metonymic descriptions are used by people working in hospitals. The metonymy thus seems to have a certain currency in the given register. In the language as a whole, however, it is not a frequent and conventional metonym. So far, the corpus study confirms the theoretical assumptions concerning the frequencies of the metonyms. But my data also demonstrate that in 170 many cases the mapping as instantiations of which the lexical items are quoted in the literature are not those which motivate the majority of the metonymic uses of these words. To put it differently, the metonymic senses of the single words that are considered conventional (no matter how true this assumption might be in the single cases) are not their most frequent metonymic meanings. Other metonymic meanings would therefore be better equipped to do the job of demonstrating the conventionality of metonymy. This can be illustrated with the lexemes sax and church. Particularly the first of them, sax, is invariably quoted as part of the sentence The sax has the flu today. But while the mapping CONTROLLED FOR CONTROLLER , which motivates the ‘sax player’ meaning, is quite rarely applied to sax (2.30%), the mapping INSTRUMENT FOR MUSIC occurs much more often (14.94%). 166 By the same token, church was taken over from the Master Metonymy List (Leite 1994) in the ‘service’ sense (motivated by the mapping LOCATION FOR AC- TIVITY OF LOCATION ), which is with only 4.00% much less frequent than the ‘institution’ sense (relying on BUILDING FOR INSTITUTION ) with 55.90%. As these examples indicate, the relative frequencies of the single metonymic meanings differ widely. Some metonymic meanings are extremely rare, others occur more frequently than the literal meanings of the lexical items. The relative frequencies established by the corpus analysis range from 0.10% to 98.40%. Table 12 (below) gives an overview of the most infrequent metonyms, more accurately of metonymic meanings which do not occur in more than 0.50% of the cases. Only first-level mappings are taken into account here, i.e. rare instances of second or third mappings of a longer metonymic chain are not included. An example of an infrequent metonymic meaning resulting from a second-level mapping is America ‘event’ (0.10% motivated by WHOLE FOR PART : WHOLE FOR PRINCIPAL PART and PART FOR WHOLE : LOCATION FOR ENTITY THERE : LOCATION FOR EVENT ). As the table shows, all three large groups of mappings generate infrequent metonymic meanings. Considering the specific-level mappings, the table contains well-known mappings for the most part. Only some of them are somewhat extraordinary. The mapping PLACE FOR SOUND EFFECT posited for stadium, for example, is rather uncommon. This usage, where the word is used to refer to a sound effect which is similar to the acoustics in a stadium, is illustrated in example (141): (141) Sony have a new digital sound processor [...], which incorporates Dolby Pro Logic and five types of hall surround effect, including Theatre, Hall, Jazz Club and Stadium (ED6-3799) 166 I am fully aware that not everybody will agree as to the metonymic status of the latter. Croft (2002), for example, would probably deny that this is a metonymy at all. It is, however, a prime example of Langacker’s (1987, 1993) reference-point constructions, where the INSTRUMENT is the active zone - it is best suited to grant access to the SOUND emitted by the instrument. 171 Wordform Mapping M - total (%) bottle WFP : WHOLE FOR PRINCIPAL PART ‘label’ 0.10 buses PFP : NON - HUMAN PRODUCER FOR PRODUCT : PRODUCER FOR PHYSICAL PRODUCT ‘exhaust’ 0.10 faces PFW : PRINCIPAL PART FOR WHOLE ‘heads’ 0.10 heads PFW : ATTRIBUTE FOR ENTITY : ATTRIBUTE FOR OBJECT ‘side of coin’ 0.10 Paris PFW : LOCATION FOR PEOPLE : LOCATION FOR SPORTS TEAM 0.10 Rome PFW : LOCATION FOR PEOPLE : LOCATION FOR SPORTS TEAM 0.10 schools WFP : INSTITUTION FOR CENTRAL FACTOR ‘curriculum’ 0.10 stadium PFP : CAUSE FOR EFFECT : PLACE FOR SOUND EFFECT 0.10 street WFP : WHOLE FOR PRINCIPAL PART ‘apartment’ 0.10 Napoleon PFP : PERSON FOR THEIR TIME / STYLE 0.13 bronze PFW : ATTRIBUTE FOR ENTITY : ATTRIBUTE OF OBJECT FOR OBJECT : ATTRIBUTE OF ANIMAL FOR ANIMAL 0.20 China/ china PFW : LOCATION FOR PEOPLE : COUNTRY FOR MILITARY FORCES 0.20 university WFP : INSTITUTION FOR PEOPLE : INSTITUTION FOR SPORTS TEAM 0.20 Napoleon PFP : CONTROLLER FOR CONTROLLED : PERSON FOR ACTIVITY OF PERSON 0.26 Picasso PFW : MEMBER OF CATEGORY FOR CATEGORY 0.25 sail PFW : CENTRAL FACTOR FOR INSTITUTION / PROFESSION : CENTRAL FACTOR FOR PROFESSION 0.28 America PFW : LOCATION FOR PEOPLE : COUNTRY FOR RULERS 0.30 China/ china PFW : LOCATION FOR PEOPLE : LOCATION FOR SPORTS TEAM 0.30 pencil PFP : CONTROLLED FOR CONTROLLER 0.30 room PFW : LOCATION FOR ENTITY : LOCATION FOR EVENT : LO- CATION FOR ACTIVITY OF LOCATION 0.30 Shakespeare PFP : PERSON FOR THEIR TIME 0.30 Shakespeare PFW : MEMBER OF CATEGORY FOR CATEGORY 0.30 airlines WFP : INSTITUTION FOR INDEX 0.33 Bordeaux PFW : LOCATION FOR PEOPLE : LOCATION FOR SPORTS TEAM 0.33 money PFW : ATTRIBUTE FOR ENTITY : ATTRIBUTE FOR PERSON : POS- SESSED FOR POSSESSOR 0.40 Moscow PFW : LOCATION FOR PEOPLE : LOCATION FOR SPORTS TEAM 0.40 schools WFP : INSTITUTION FOR PEOPLE : INSTITUTION FOR SPORTS TEAM 0.40 wine PFP : CONTENT FOR CONTAINER 0.40 Einstein PFP : PRODUCER FOR PRODUCT : PRODUCER FOR ABSTRACT PRODUCT : AUTHOR FOR WORKS 0.50 Rome PFW : LOCATION FOR ENTITY : LOCATION FOR EVENT 0.50 room PFW : LOCATION FOR ENTITY : LOCATION FOR INSTITUTION 0.50 seat PFW : LOCATION FOR ENTITY : LOCATION FOR OBJECT : LO- CATION FOR BODY PART ‘bottom’ 0.50 Vietnam PFW : LOCATION FOR PEOPLE 0.50 Washington PFW : LOCATION FOR PEOPLE : LOCATION FOR SPORTS TEAM 0.50 Table 12: The least frequent metonymic meanings ( WFP = WHOLE FOR PART ; PFW = PART FOR WHOLE ; PFP = PART FOR PART ) 172 Other less often discussed mappings embrace PERSON FOR THEIR TIME / STYLE (e.g. Napoleon in a ‘Napoleon’ armchair, CHP-27) and INSTITUTION FOR STOCK MARKET SHARES / INDEX (airlines as in American Airlines […] fell seven dollars; A4F-545). However, the majority of the mappings motivating the infrequent metonymic meanings belong to the oft-quoted stock examples in cognitive linguistics, such as PRINCIPAL PART FOR WHOLE (faces ‘heads’), ATTRIBUTE FOR ENTITY (e.g. bronze ‘bronze cow’, money ‘possessor of money’), and PRODUCER FOR PRODUCT (e.g. Einstein ‘Einstein’s works’). All in all, the mappings underlying these particularly infrequent metonymic meanings are quite varied. Only one group of mappings occurs strikingly often here: LOCATION FOR PEOPLE (e.g. China, America) and its sub-type LOCATION FOR SPORTS TEAM (e.g. Paris, Rome). The related mapping LOCATION FOR EN- TITY also appears repeatedly in Table 12. It motivates uses of Rome (as in One Scotsman looking forward to Rome is David Will, who will take over from Harry Cavan […] when the Irishman retires during the World Cup next June; A33-109) and room (as in president of the Junior Common Room; CDC-131). Apart from this, it is difficult to find any regularities among the infrequent instances. Especially interesting is that in some cases the table contains examples of both mapping directions for reversible mappings. 0.26% of all uses of Napoleon are motivated by the CONTROLLER FOR CONTROLLED mapping (with the more specific sub-mapping PERSON FOR ACTIVITY OF PER- SON as in Paris had been altered for ever and bore the mark of Napoleon III; ANR- 704), and 0.30% of all uses of pencil rely on the mapping CONTROLLED FOR CONTROLLER (e.g. so fine a pencil is worthy of encouragement; ALU-695). Bottle occurs in 0.10% of all uses as a WHOLE FOR PRINCIPAL PART metonymy (‘label’), and faces in 0.10% of all cases as a PRINCIPAL PART FOR WHOLE metonymy (‘heads’). Even though the laws of ontological salience predict a preference for CONTROLLER FOR CONTROLLED metonymies over CONTROLLED FOR CONTROLLER metonymies, because the former are in line with HUMAN OVER NON - HUMAN whereas the latter are not, both mapping directions can result in infrequent metonyms, which are, moreover, very similar in terms of their relative frequencies. Similarly, the speech community ought to prefer WHOLE FOR PRINCIPAL PART mappings over PRINCIPAL PART FOR WHOLE mappings, since wholes are better equipped to attract our attention than parts. However, both patterns can produce equally infrequent metonyms. This raises the question as to whether the laws of ontological salience are really as powerful with regard to the conventionalization potential of metonyms as often believed. Admittedly, these few cases could be nothing more than exceptions to the rule. But what they do is call for a more meticulous consideration of the predictive value of the laws of ontological salience. Even more so, as one of the most frequent metonyms found in the corpus study, i.e. football ‘game’, belongs to the group of PRINCIPAL PART FOR WHOLE mappings and is therefore motivated by a mapping that actually ought to be dispreferred by the speech community. The general preference for WHOLES over PARTS as metonymic reference points thus does not pre- 173 vent highly conventional PART FOR WHOLE meanings. This issue deserves a more detailed discussion and will be addressed in chapter 6.2.1.1. Table 13 provides an overview of the most frequent metonyms, i.e. those cases where metonymic meanings occur in more than 40% of all uses. Again, only first-level mappings are considered. Cases with highly frequent metonymic meanings resulting from a metonymic chain are rare. An instance is glasses ‘spectacles’ (motivated by PART FOR WHOLE : ATTRIBUTE FOR OBJECT : MATERIAL FOR OBJECT and PART FOR WHOLE : PRINCIPAL PART FOR WHOLE ). Wordform Mapping M - total (%) glasses PFW : ATTRIBUTE FOR ENTITY : ATTRIBUTE FOR OBJECT : MATERIAL OF OBJECT FOR OBJECT 98.40 Wall Street PFW : LOCATION FOR ENTITY : LOCATION FOR INSTITUTION 96.99 football PFW : PRINCIPAL PART FOR WHOLE : PLAYING PIECE FOR ENTIRE GAME 96.80 press PFW : CENTRAL FACTOR FOR INSTITUTION / PROFESSION : CEN - TRAL FACTOR FOR INSTITUTION 92.30 Watergate PFW : LOCATION FOR ENTITY : LOCATION FOR EVENT 84.54 Pentagon PFW : LOCATION FOR ENTITY : LOCATION FOR INSTITUTION : BUILDING FOR INSTITUTION 80.13 White House PFW : LOCATION FOR ENTITY : LOCATION FOR INSTITUTION : BUILDING FOR INSTITUTION 72.10 chicken PFP : SOURCE FOR PRODUCT : ANIMAL FOR MEAT OF ANIMAL 68.70 furs PFW : ATTRIBUTE FOR ENTITY : ATTRIBUTE FOR OBJECT : MATERIAL OF OBJECT FOR OBJECT 65.33 crown PFW : ATTRIBUTE FOR ENTITY : ATTRIBUTE FOR PERSON : COSTUME FOR WEARER 65.00 America WFP : WHOLE FOR PRINCIPAL PART 62.50 government WFP : INSTITUTION FOR PEOPLE RESPONSIBLE 56.20 church PFW : LOCATION FOR ENTITY : LOCATION FOR INSTITUTION : BUILDING FOR INSTITUTION 55.90 drugs WFP : CATEGORY FOR SALIENT MEMBER 55.50 set of wheels PFW : PRINCIPAL PART FOR WHOLE 50.00 paper PFW : ATTRIBUTE FOR ENTITY : ATTRIBUTE FOR OBJECT : MATERIAL OF OBJECT FOR OBJECT ‘piece of writing’ 49.20 turtleneck PFW : ATTRIBUTE FOR ENTITY : ATTRIBUTE FOR OBJECT : ATTRIB- UTE FOR CLOTHING 46.15 pill WFP : CATEGORY FOR SALIENT MEMBER 44.95 ecstasy PFP : EFFECT FOR CAUSE : EMOTION FOR CAUSE OF EMOTION 44.70 oak WFP : WHOLE FOR PRINCIPAL PART : MATERIAL FOR OBJECT 44.50 book PFP : CONTAINER FOR CONTENT 44.30 Renault PFP : PRODUCER FOR PRODUCT : PRODUCER FOR PHYSICAL PRODUCT : PRODUCER FOR CONSUMER PRODUCT 43.74 Table 13: The most frequent metonymic meanings ( WFP = WHOLE FOR PART ; PFW = PART FOR WHOLE ; PFP = PART FOR PART ) 174 The most frequent metonyms illustrated in Table 13 also rely on diverse mappings. Again, all three large groups of mappings occur in the table. As concerns the specific-level mappings, it is interesting to see that mappings which lead to very infrequent metonymic meanings can also result in highly frequent metonymic senses. A mapping which appears in both overviews is, for instance, CENTRAL FACTOR FOR INSTITUTION / PROFESSION . Sail is an infrequent example, and press a very frequent one. L OCATION FOR ENTITY also occurs repeatedly in Table 12 and in Table 13. Rome and room, for example, are infrequently used as instances of the mapping, while the vast majority of all uses of Watergate and Wall Street is motivated by LO- CATION FOR ENTITY . This result is not an entirely unexpected one, since it is commonly acknowledged that mappings can and do motivate novel or uncommon linguistic instances. Nevertheless, the question as to why we find these differences remains to be answered. Ontological salience alone does not seem to be enough to clarify this, as the frequent and infrequent examples rely on the same mapping, even on the more specific forms of LOCATION FOR ENTITY , i.e. LOCATION FOR EVENT and LOCATION FOR INSTITU- TION respectively. Some regularities are evident in Table 13: L OCATION FOR INSTITUTION mappings of the more specific type of BUILDING FOR INSTITUTION (e.g. church, Pentagon, White House) seem to be especially likely to produce highly frequent metonyms. The same holds true for the mapping MATERIAL FOR OBJECT (e.g. glasses, furs, paper). This demonstrates that it is often necessary to have a closer look at the data on different levels of specificity. If only the more general mappings ATTRIBUTE OF ENTITY FOR ENTITY (for the latter cases) and LOCATION FOR ENTITY THERE (for the former) are taken into account, no regularities transpire. It is necessary to consider the most specific mappings to discover tendencies. A comparison of the two tables also shows that some mappings only occur in one of them. C ONTENT FOR CONTAINER only motivates one infrequent example, i.e. wine; its reversal CONTAINER FOR CONTENT does not underlie one of the extremely infrequent metonymic uses. However, it is the basis of the relatively frequent use of book ‘content’. Another example, where the different mapping directions seem to coincide with either very low or fairly high relative frequencies of metonymic uses is the CATEGORY AND MEMBER ICM: M EMBER FOR CATEGORY leads to only 0.25% of metonymic uses of Picasso, but CATEGORY FOR MEMBER results in 55.50% of metonymic uses of drugs and 44.95% of metonymic uses of pill. For the first case, i.e. the CONTAINER ICM, the findings are thus far in line with what the laws of ontological salience predict: The route from CONTAINER to CONTENT seems to be preferred, since the container is, for example, more visible than the content. For the second case, ontological salience would, however, predict the opposite of what the tables suggest. So far, I have only dealt with the extremes of the continuum of frequency. But what about the cases which are neither particularly frequent 175 nor extremely infrequent? Table 14 provides an overview of such examples. The mean value of all metonymic uses is 26.14%. To better represent the fact that there are, in general, more metonymic meanings with lower relative frequencies, the table lists not only those first-level mapping instances with relative frequencies that come close to the mean value, but includes examples with lower relative frequencies as well. It embraces metonymic uses of relative frequencies between 15% and 30%: 167 Word-form Mapping M - total (%) label PFW : ATTRIBUTE FOR ENTITY : ATTRIBUTE OF OBJECT FOR OBJECT 29.60 Washington PFW : LOCATION FOR ENTITY : LOCATION FOR INSTITUTION : CAPITAL CITY FOR GOVERNMENT 26.80 pencil WFP : WHOLE FOR PRINCIPAL PART 26.10 China/ china PFW : LOCATION FOR PEOPLE : COUNTRY FOR RULERS 26.00 schools WFP : INSTITUTION FOR PEOPLE RESPONSIBLE 25.30 Moscow PFW : LOCATION FOR ENTITY : LOCATION FOR INSTITUTION : CAPITAL CITY FOR GOVERNMENT 24.20 newspaper PFP : PRODUCT FOR PRODUCER : PHYSICAL PRODUCT FOR PRODUCER : PHYSICAL PRODUCT FOR INSTITUTION 23.20 wood PFW : ATTRIBUTE FOR ENTITY : ATTRIBUTE FOR OBJECT : MA- TERIAL FOR OBJECT ‘forest’, ‘branch of tree’ 21.90 silver PFW : ATTRIBUTE FOR ENTITY : ATTRIBUTE FOR OBJECT : MA- TERIAL FOR OBJECT ‘cutlery’, ‘coins’, ‘medal’ 20.80 seat PFW : CENTRAL FACTOR FOR INSTITUTION / PROFESSION : CENTRAL FACTOR FOR INSTITUTION 20.70 airlines WFP : INSTITUTION FOR PEOPLE RESPONSIBLE 20.41 Vietnam PFW : LOCATION FOR ENTITY : LOCATION FOR EVENT 20.20 kettle PFP : CONTAINER FOR CONTENT 19.46 Vietnam PFW : LOCATION FOR PEOPLE : COUNTRY FOR RULERS 18.30 faces PFW : PRINCIPAL PART FOR WHOLE : BODY PART FOR PERSON 17.50 faces WFP : WHOLE FOR PRINCIPAL PART ‘facial expression’, ‘mouth’ 16.90 telephone WFP : WHOLE FOR PRINCIPAL PART ‘receiver’, ‘bell’, ‘cable’ 16.40 diesel PFW : ATTRIBUTE FOR ENTITY : ATTRIBUTE FOR OBJECT ‘car’, ‘locomotive’ 15.10 Table 14: Metonymic meanings with relative frequencies between 15% and 30% ( WFP = WHOLE FOR PART ; PFW = PART FOR WHOLE ; PFP = PART FOR PART ) Again, we find some mappings which already occurred in Tables 12 and 13, i.e. as underlying both very frequent and infrequent metonymic meanings. L OCATION FOR ENTITY THERE motivates, for example, 20.20% of all uses 167 An example where a metonymic chain of two mappings leads to a relative frequency between 15% and 30% is pencil ‘mark left by pencil’ (23.40%; motivated by WHOLE FOR PART : WHOLE FOR PRINCIPAL PART and PART FOR PART : CAUSE FOR EFFECT ). 176 of Vietnam, and ATTRIBUTE OF ENTITY FOR ENTITY underlies 15.10% of all uses of diesel. But Table 14 also includes mappings which do not appear in the overviews of the extremely infrequent and frequent instances. 23.20% of all uses of newspaper rely on the mapping PRODUCT FOR PRODUCER . Its opposite, PRODUCER FOR PRODUCT , however, results in infrequent as well as highly frequent metonyms (cf. e.g. Einstein in Table 12 and BMW in Table 13) - a fact that is hard to reconcile with the view that the speech community ought to favour one of two possible directions of mapping. Furthermore, the occurrence of the mapping MATERIAL OF OBJECT FOR OBJECT in Table 14 is of interest. This mapping also appears repeatedly in Table 13, i.e. also motivates some highly frequent metonymic uses (cf. furs, paper in Table 13). The same applies to CONTAINER FOR CONTENT , which leads to more than 40% of metonymic uses of book (cf. Table 13), but only to 19.46% of metonymic uses of kettle (cf. Table 14). To explain all these inconsistent findings, it is necessary to study the frequencies with which different mappings lead to metonymic meanings in more detail. This will be done in chapters 6.2.1.3 to 6.2.3, where the results of the corpus study are discussed in a mappingby-mapping fashion. To conclude this section, I would like to return to one of the theoretical models of the conventionality of metonymy. Barcelona’s (2002a, 2003, 2004; cf. chapter 4.3.2.3) account assumes that, in general, prototypical metonymies (where there is a clear-cut conceptual boundary between the source and target concept) ought to be more frequent than instances of schematic or typical metonymy. This is, however, not the case. Apt examples to demonstrate this are some metonymic meanings of book and paper, i.e. ‘content of the book’ and ‘piece of writing’ respectively. Both belong to Barcelona’s class of typical metonymies, since it is not easy to establish a precise boundary between a book and its content or between the material one writes on and the writing itself. They should therefore be less conventional and less frequent than prototypical cases like BMW in the meaning ‘driver’- since the car and the driver are undoubtedly two entirely different entities - or bottle in its ‘content’ meaning - since the container and its content are separate entities as well. The relative frequencies observed for these words in these specific metonymic meanings reinforce my earlier assumption that his equation of metonymicity and conventionality is untenable. The conventionality of metonymic meanings does not depend on more or less clear-cut boundaries between the source and target domains of metonymies. This becomes even more evident if various instances belonging to Barcelona’s different groups are compared (cf. Table 15, below). The randomly chosen examples in Table 15 demonstrate that there is no correlation whatsoever between the different types and their relative frequencies of occurrence. Some typical metonymies like book ‘content’ and paper ‘piece of writing’ are markedly more frequent than some prototypical instances such as BMW ‘driver’, bottle ‘content’ or buses ‘bus drivers’. For other cases, there is an inverse relationship. Some prototypical metonymies 177 like church ‘institution’ or crown ‘wearer’ are much more common than some typical ones such as car ‘motor, exterior, interior of the car’, kitchen ‘fittings/ fitments’ and plastic ‘object made of plastic’. This means that not only can Barcelona’s scale of the conventionality of metonyms be called into question, but also his scale of metonymicity. As Barcelona (2003, 2004) argues, the metonymicity rises from schematic through typical to prototypical metonymies. From a theoretical point of view, this is quite convincing. But according to the frequency data another picture emerges, at least if one understands metonymicity as felt figurativity. A metonym such as paper ‘piece of writing’ or book ‘content’ which is realized in nearly 50% of all uses of the word will hardly be felt to be figurative by native speakers. So far my results are in line with Barcelona’s predictions. However, for some prototypical metonymies such as church ‘institution’ or crown ‘wearer’, where the share of these metonymic meanings found is higher than the percentage of literal (or other figurative) uses of the same words, it is also not very likely that speakers are still conscious of their figurative nature. Typical metonymies Prototypical metonymies paper ‘piece of writing’ 49.20% crown ‘wearer’ 65.00% book ‘content’ 44.30% church ‘institution’ 55.90% kitchen ‘fittings/ fitments’ 4.60% bottle ‘content’ 9.90% car ‘motor, exterior, interior, etc.’ 4.10% BMW ‘driver’ 4.64% plastic ‘object’ 2.50% buses ‘bus drivers’ 1.00% Table 15: Some examples of Barcelona’s (2002a, 2003, 2004) typical and prototypical metonymies and their relative frequencies This section has given a first brief overview of some of the relative frequencies found for metonymic meanings and related my frequency data to some theoretical predictions and models. A realistic model of the conventionality of metonymy can hardly be based on purely theoretical grounds without any consideration of the speech community’s actual behaviour. This has been confirmed for Barcelona’s account as well as with regard to the predictive value of the laws of ontological salience. The latter will serve as a starting point for a thorough discussion of conceptual mappings and their relationship to the relative frequencies of occurrence in the next chapter, where I will address the question as to whether the reasons for the relative frequency of occurrence of a given metonymic meaning are to be sought on the conceptual or on the linguistic level. However, as already noted above, universal laws like WHOLE OVER PART or HUMAN OVER NON - HUMAN are almost certainly not sufficient to explain why some metonymic construals are evidently preferred over others. Additional criteria are needed. As will be shown, a high conventionality of metonymies characteristically correlates with the fact that the target concepts or attributes related with the target concepts are especially prominent in the vehicle concepts. 178 What is at issue now is why there are such huge differences between the frequencies of single metonymies. Why is it that some linguistic metonyms occur so often that they are probably no longer noticed as being figurative, whereas others are hardly ever used? Linguistic conventions are of a social nature. And even though one can indeed criticize the conceptual metaphor and metonymy theory for underestimating the social aspects of conventionality and the role of linguistic experience (Svanlund 2007: 50), one must not forget that the speech community is not an abstract entity. It consists of human beings whose linguistic behaviour is determined by pragmatic factors and is far from being independent of general cognitive principles. It is thus more than likely that the speech community will normally prefer a language that conforms to both its basic pragmatic and cognitive needs. In the following sections I will therefore also look for pragmatic aspects which govern the speech community’s preferences in this respect, but - due to the overall cognitive orientation of this study - mostly concentrate on conceptual regularities behind the differing relative frequencies of single metonymies. 6.2 The conventionality of metonymies: A matter of linguistic idiosyncrasy or conceptual regularity? 6.2.1 Salience and the conventionality of metonymy The most detailed account of the relationship between the conventionality of metonymy and ontological salience is given by Radden and Kövecses (1999; cf. also Kövecses and Radden 1998). They link their extensive typology of metonymies with comments on how frequent the single classes are and on their degree of felt figurativity (cf. chapter 4.2.2.2). For example, they emphasize that PART FOR WHOLE metonymies are “less ubiquitous” and “more likely to be noticed” than WHOLE FOR PART metonymies (1999: 31). P ART FOR WHOLE construals are supposed to be less conventional than WHOLE FOR PART construals since the former deviate from the general principle WHOLE OVER PART . As humans favour a holistic perception and a potential decomposition into the single parts and attributes of the whole is only an optional second step (Ungerer and Schmid 2006: 36) 168 , wholes ought to be preferred as entry points to metonymic construals. The preliminary overview of the results just given (chapter 6.1) has already demonstrated that PART FOR WHOLE metonymies are not invariably less frequent than WHOLE FOR PART metonymies. But perhaps clearer tendencies emerge when the data are analyzed in a more systematic and comprehensive manner. In what follows, I will therefore examine the assumption that, on the whole, metonyms belonging to the large group of WHOLE 168 This is related to the notion of gestalt and the gestalt principles, such as proximity, similarity, closure, and continuation (cf. e.g. Ungerer and Schmid 2006: 36). 179 FOR PART mappings ought to be more frequent than those belonging to the PART FOR WHOLE group. However, as already noted earlier (chapter 4.3.2.2), the issue of determining wholes and parts is nontrivial. There are some quite simple cases. When somebody refers to the ‘United States’ as America, there can be no doubt about the whole and the part. And when someone uses face and means ‘person’, they are definitely using a PART FOR WHOLE metonymy. But not all cases are so straightforward. 6.2.1.1 Wholes and parts It is not always easy to decide whether something is an inherent part of another, larger domain or concept. This is reflected in the fact that metonymy researchers frequently disagree as to whether a given metonymy ought to be treated as a PART FOR PART or a WHOLE FOR PART metonymy. The following examples illustrate this: (142) I smell skunk. (143) I took a walk in the wood. (144) I only eat chicken. (145) I always carry plastic. (146) Cambridge won’t publish the book. (147) Exxon has raised its prices again. While Radden and Kövecses (1999: 32) treat the CONSTITUTION ICM as a matter of wholes and parts and correspondingly argue that OBJECT FOR MATERIAL CONSTITUTING THE OBJECT in (142) is a WHOLE FOR PART metonymy and that MATERIAL CONSTITUTING AN OBJECT FOR THE OBJECT as in (143) is a PART FOR WHOLE metonymy, the Master Metonymy List (Leite 1994: 18) treats the very similar example (144) which is based on the pattern ANIMAL FOR MEAT OF ANIMAL as a PART FOR PART metonymy. The MATERIAL OF OBJECT FOR OBJECT mapping, as in (145), is, however, said to belong to the PART FOR WHOLE metonymies (Leite 1994: 8). And whereas Radden and Kövecses (1999: 41) mention example (146) under the heading of PART FOR PART , because they regard it as belonging to the LOCATION ICM, i.e. as an instance of the pattern PLACE FOR INSTITUTION , Leite (1994: 14) treats (147) as an instantiation of a more elaborate INSTITUTION FOR PEOPLE RESPONSIBLE pattern, which belongs to the INSTITUTION frame and is therefore considered a WHOLE FOR PART metonymy. In contrast to the common tripartite classification of metonymies, Ruiz de Mendoza and his collaborators deny the existence of PART FOR PART metonymies. Their binary system of source-in-target and target-in-source metonymies (cf. chapter 4.3.2.2) only allows PART FOR WHOLE and WHOLE FOR PART metonymies. Instances of what is traditionally called PART FOR PART metonymy are re-analyzed in order to be subsumed under their target-in-source metonymies (Ruiz de Mendoza and Díez Velasco 2002: 497, 2004: 298). In this way, a simple correspondence is established between traditional PART FOR PART mappings and target-in-source mappings. P RO- 180 DUCER and PRODUCT , for example, are not considered as different parts of the PRODUCTION ICM or frame, but the PRODUCT is conceived of as a part of the larger PRODUCER domain. The same re-analysis applies to mappings like CONTROLLER FOR CONTROLLED , POSSESSOR FOR POSSESSED , etc. At first glance, this view is quite attractive. Difficulties arise, however, when one considers that some of the traditional PART FOR PART mappings can be reversed (cf. e.g. Radden and Kövecses 1999). And if the CON- TROLLED stands for the CONTROLLER (e.g. The Mercedes has arrived) or the POSSESSED for the POSSESSOR (e.g. He married money), a re-analysis as targetin-source seems to be excluded. There is no way of claiming that the CON- TROLLED forms the larger, more exhaustive matrix domain, while the CON- TROLLER is just a sub-domain of the CONTROLLED . And indeed, Ruiz de Mendoza and Pérez Hernández’s (2001: 325, 345) re-analyses of the OR- DERED FOOD FOR CUSTOMER metonymy and the EFFECT FOR CAUSE metonymy are done in terms of source-in-target and not in terms of target-in-source. The relationship between the traditional notion of PART FOR PART metonymies and Ruiz de Mendoza’s typology is thus by no means simple. As these inconsistencies concerning the categorization of metonymies show, it often depends on the individual perspective whether a metonymic construal is conceived of as PART FOR PART , WHOLE FOR PART or PART FOR WHOLE , the exception being very clear instances, such as FACE FOR PERSON , where no one would deny that it is a PART FOR WHOLE metonymy. If Ruiz de Mendoza were right with his assumption that everything that is usually denoted a PART FOR PART metonymy can be usefully considered as a target-in-source metonymy, the majority of all metonymies would be of the WHOLE FOR PART type, and all WHOLE FOR PART metonymies, i.e. all target-in-source metonymies, would stand a better chance of becoming conventional than all PART FOR WHOLE metonymies. Only sometimes language users would diverge from the ‘normal’ pattern and use metonymic sources which should, according to the general laws of perception, be less prominent than the targets. A case in point is the mushroom omelet example, which is indeed not very likely to occur frequently. But also other metonymic mappings such as POSSESSED FOR POSSESSOR or CONTROLLED FOR CON- TROLLER should be markedly less frequent than those which belong to the overall frame of WHOLE FOR PART metonymies. The question is if these predictions coincide with the real relative frequencies of different metonymic patterns. For some cases, Ruiz de Mendoza’s predictions are indeed valid. Two of the target-in-source metonymies he mentions are AUTHOR FOR WORK and INSTITUTION FOR PEOPLE RESPONSIBLE . Examples of my corpus-study manifesting these patterns are, for instance, Shakespeare and Exxon. Source-intarget metonymies are, among others, appendicitis ( ATTRIBUTE FOR PERSON ) and ecstasy ( EFFECT FOR CAUSE ). A comparison of the relative frequencies of these meanings is given in Table 16. 181 Target-in-source Source-in-target Shakespeare ‘work’ 34.90% ecstasy ‘drug’ 44.70% Exxon ‘people responsible’ 35.17% appendicitis ‘patient’ 1.64% Table 16: Some examples of source-in-target and target-in-source metonymies and their relative frequencies The frequencies of the two target-in-source metonymies are in fact quite high. Both lie above the mean value for all metonymic meanings (25.85%). Up to this point, this is in line with Ruiz de Mendoza’s assumptions. However, one of the two source-in-target metonymies, i.e. ecstasy, deviates from his predictions. It is even more frequent than the target-in-source metonymies, although it allegedly uses a less well-structured sub-domain as its source. This example demonstrates that the difference between source-intarget and target-in-source metonymies does not always lead to valid predictions as to whether the speech community will accept a given metonym. As has been shown above, Ruiz de Mendoza’s two classes of metonymy are sometimes difficult to apply because it is not always clear whether what is usually considered a PART FOR PART mapping should be re-analyzed as a target-in-source or as a source-in-target metonymy. This is why I will stick to the more traditional tri-partite classification of metonymies into WHOLE FOR PART , PART FOR WHOLE , and PART FOR PART . These classes are, however, not unproblematic either as the above-mentioned different categorizations of very similar examples demonstrates. Therefore, the results discussed in the next section, which gives an overview of the frequencies of the respective patterns, are of a provisional nature, since they hinge upon the uncertain categorization of wholes and parts. 6.2.1.2 W HOLE FOR PART , PART FOR WHOLE , and PART FOR PART metonymies Considering the high degree of ontological salience of wholes as opposed to parts, WHOLE FOR PART metonymies should be abundant in language. Astonishingly, however, this type is the one for which the Master Metonymy List provides the smallest amount of examples. All in all, Leite’s (1994) 14page long list has only one page of WHOLE FOR PART examples. And even in Radden and Kövecses (1999), examples of this type are not really frequent. Since my aim was to cover as many different metonymic mappings as possible, the fact that there are so many more types of PART FOR WHOLE and PART FOR PART metonymies is reflected in my analysis in the higher overall number of PART FOR WHOLE and PART FOR PART mappings initially included in the study. The metonymies I was originally looking for were - with 50 out of 100 - primarily of the PART FOR WHOLE type. 34 belong to the PART FOR PART metonymies and 16 to the WHOLE FOR PART type. The picture changed only slightly after the study. Since many of the words analyzed can convey more than one metonymic meaning, the total number of form-meaning relationships considered increased to 229 (cf. 182 chapter 5.2.1.2), but still the majority of the mappings underlying the usage of the words are of the PART FOR WHOLE type. 122 mappings find their place here, while PART FOR PART occurs 63 times and WHOLE FOR PART only 44 times. For WHOLE FOR PART 28 new instances were found and for PART FOR PART 29. While this does not even double the amount of PART FOR PART mappings, the overall number of WHOLE FOR PART mappings is nearly three times as high as the original number. The number of PART FOR WHOLE mappings also increased dramatically - 72 additional metonymic uses were found in the corpus. This indicates that PART FOR WHOLE mappings are, on the whole, not significantly less productive than WHOLE FOR PART mappings. But what about the relative frequencies of the single mappings? Possibly WHOLE FOR PART mappings are not used with so many words, but are highly frequent for those words to which they can be applied? Table 17 gives a summary of the mean values of the relative frequencies of the single patterns and the standard deviations 169 for all three groups: Mapping Mean value (%) Standard deviation WHOLE FOR PART 14.01 18.37 PART FOR WHOLE 12.64 23.50 PART FOR PART 11.01 15.43 Table 17: Mean values and standard deviations for the three large mapping patterns The table shows that the mean value of the WHOLE FOR PART mappings is indeed slightly higher than those of the other patterns. But, as the relatively high standard deviation indicates, the relative frequencies of instances realizing the single sub-mappings of this type differ markedly. The lowest frequency found for WHOLE FOR PART mappings is 0.10% of all meanings (e.g. bottle ‘label on a bottle’ as in read the bottle; KSR-46). The lexical item which is most often used in a WHOLE FOR PART sense is America. In the majority of the cases (62.5%) the word is used to refer to the ‘United States’. This mapping is often the basis for other, secondary mappings. C OUNTRY FOR RULERS , LOCATION FOR PEOPLE , LOCATION FOR MILITARY FORCES , and LO- CATION FOR EVENT all usually presuppose an understanding of America as a WHOLE FOR PART metonym. Other very frequent WHOLE FOR PART mappings were found for government ( INSTITUTION FOR PEOPLE RESPONSIBLE : 56.20%) and drugs ( CATEGORY FOR SALIENT MEMBER : 55.50%). Instances whose relative frequencies come close the mean value are, for example, Renault ( INSTI- TUTION FOR PEOPLE RESPONSIBLE : 12.87%) or telephone ( WHOLE FOR PRINCIPAL PART , ‘receiver’, ‘bell’, ‘cable’: 16.40%). With a mean value of 12.64% of all meanings the PART FOR WHOLE mappings are a little less frequent. The standard deviation is even higher than 169 The standard deviation indicates how much members of a given mathematical set differ from the mean value (Albert and Koster 2002: 78). 183 for the WHOLE FOR PART mappings, i.e. the relative frequencies of the single mappings have a greater variance than the WHOLE FOR PART mappings. Nearly half of all cases account for 1% or less of all the meanings of a given lexical item. Many instances of LOCATION FOR PEOPLE belong to these quite infrequent PART FOR WHOLE mappings. Vietnam, for example, is used to refer to the population in only 0.50% of all cases. And Paris is used in this way in only 0.80% of the cases, but an additional percentage of 0.10% realizes the sub-mapping LOCATION FOR SPORTS TEAM . Other often infrequent mappings are, for instance, LOCATION FOR EVENT which also occurs with different proper nouns, such as Rome (0.50%), Paris (0.90%), and America (0.10%), but is not restricted to them. These low relative frequencies stand in marked contrast to cases like glasses (98.40%) Wall Street (96.99%), football (96.80%) or press (92.30%). These PART FOR WHOLE mappings are so common that the literal meanings no longer play a significant role. The deep entrenchment of the PART FOR WHOLE mappings in these cases is also evident in the fact that they are either the only ones found for these word-forms, as for press and Wall Street, or that they underlie all other mappings found - by way of forming the basis for metonymic chains as for glasses and football (cf. Appendix A). Examples whose relative frequencies as PART FOR WHOLE metonyms are not so different from the mean value are, for instance, Germany ( COUNTRY FOR RULERS : 12.00%), convent ( LOCATION FOR INSTITUTION : 12.60%), and heels ( PRINCIPAL PART FOR WHOLE ‘shoe’: 13.80%). As concerns PART FOR PART mappings, the mean frequency is slightly lower than for the other two groups. Only 11.01% of all meanings analyzed are motivated by a PART FOR PART mapping. With a standard deviation of 15.43, the variance is, however, the lowest among all three large groups. The different frequencies of the single sub-mappings do not differ so markedly as for the other two groups. The lowest frequency is, once again, 0.10%. This holds, for example, for the entirely uncommon mapping PLACE FOR SOUND EFFECT , which occurs in my sample for the form stadium (cf. example 141 above). Other lexical items which realize infrequent mappings belonging to the PART FOR PART group are, for example, pencil ( CONTROLLED FOR CONTROLLER : 0.30%), packet ( ABSTRACT PRODUCT FOR PRODUCER : 0.50%), and buses ( CONTROLLED FOR CONTROLLER : 1.00%). The highest frequencies of PART FOR PART mappings were found with chicken ( ANIMAL FOR MEAT OF ANIMAL : 68.70%) and BMW ( PRODUCER FOR CONSUMER PRODUCT : 62.67%). All the other values are already markedly lower. 44.70% of the instances of ecstasy are based on the mapping EMOTION FOR CAUSE OF EMOTION , 44.30% of the instances of book are motivated by CONTAINER FOR CONTENT , and 43.74% of all instances of Renault manifest the mapping PRODUCER FOR CONSUMER PRODUCT . Examples whose relative frequencies come close to the mean value are, for instance, Bordeaux as an example of LOCATION FOR PRODUCT MADE THERE (11.71%) or law when its use is based on WORK FOR AUTHOR (10.70%). 184 Let me summarize briefly at this point. All in all, the mean value of WHOLE FOR PART mappings is slightly higher than for the other patterns. Interestingly, however, the highest frequencies of single mappings established by the corpus study are not found within the group of the WHOLE FOR PART mappings, but in the group of PART FOR WHOLE mappings. This is shown in Table 18. WHOLE FOR PART PART FOR PART PART FOR WHOLE America 62.50% chicken 68.70% glasses 98.40% government 56.20% BMW 62.67% Wall Street 96.99% drugs 55.50% ecstasy 44.70% football 96.80% Pentagon 49.16% book 44.30% press 92.30% pill 44.95% Renault 43.74% Watergate 84.54% oak 44.50% pint 37.50% Pentagon 80.13% Senate 36.30% Shakespeare 34.90% White House 72.10% Exxon 35.17% dish 32.70% furs 65.33% heels 33.00% label 25.50% crown 65.00% pencil 26.10% pencil 23.40% church 55.90% Table 18: The ten most frequent instances of the three different groups of mappings (in descending order) 170 Within the group of the PART FOR WHOLE mappings we find the most deeply entrenched metonymic meanings with percentages that amount up to 98.40%. For PART FOR PART mappings, the most deeply entrenched metonymic meaning, i.e. chicken ( ANIMAL FOR ITS MEAT ), is realized in 68.70% of all cases. For WHOLE FOR PART mappings, the most frequent metonym is America with 62.50% of references to the ‘United States’. None of the instances of WHOLE FOR PART or PART FOR PART returns relative frequencies of 70% or higher. For the PART FOR WHOLE group, as many as seven lexical items manifest PART FOR WHOLE mappings which occur in at least 70% of the cases. Another interesting aspect is that for both WHOLE FOR PART and PART FOR PART mappings there are only a few instances which come close to the highest values. When one compares the ten most frequent instances of the single patterns, it is evident that the values decrease much faster for PART FOR PART and WHOLE FOR PART mappings than for PART FOR WHOLE mappings (cf. Table 18). 170 Two lexical items occur twice in the table. For pencil, the PART FOR PART mapping is secondary to the WHOLE FOR PART mapping. For Pentagon, the WHOLE FOR PART mapping is secondary to the PART FOR WHOLE mapping. This also explains why the two percentages given for Pentagon amount to more than 100%. Actually, 49.16% of all instances feature a double metonymy consisting of both mappings indicated. This means that the 49.16% of WHOLE FOR PART instances form part of the 80.13% of PART FOR WHOLE instances. The situation for pencil is similar. The 23.40% of PART FOR PART instances are part of the 26.10% of WHOLE FOR PART instances. 185 Evidently, it is the upper end of the continuum of frequencies which is responsible for the differences in terms of the standard deviation found for the single groups. Low percentages abound for all three groups - a situation which amply illustrates the creative potential of metonymy. In principle, almost anything goes, given the right context and background knowledge of the participants. Some uncommon examples are, for instance, chicken as in (148) or heels in (149): (148) Take the top off the chicken (KP1-2355) (149) [...] and O saw she was walking back to the child, he heard her heels [...] (AR2-973) What is heard in (149) are not the heels, but the sound they make when somebody is walking. And in (148) chicken actually refers to the frying pan where the meat of the animal is located. But these are not the uses which become entrenched and conventional. These are just possible metonymic construals sanctioned by existing mapping patterns, such as LOCATION FOR ENTITY THERE or CAUSE FOR EFFECT . What they show, however, is how familiar we are with the underlying mappings. Were it not for these wellknown patterns of thought, more uncommon cases would not have occurred with so many different lexical items. Instances which do not fit any of the mappings described in the literature were not found. Even the ‘sound effect’ meaning of stadium is not novel from a conceptual perspective, but simply a more specific case of the mapping CAUSE FOR EFFECT . Although the laws of ontological salience clearly predict a general preference for wholes as conceptual entry points to metonymic construals, the results of my study do not support this assumption. But due to the abovementioned difficulty with the distinction of wholes and parts and to the overrepresentation of mappings other than WHOLE FOR PART in my study, the general validity of this result can be questioned. A closer look at the highly frequent examples of PART FOR WHOLE metonymies reveals that many of them belong to the sub-classes of metonymic mappings whose status as PART FOR WHOLE mappings could be disputed. As mentioned above (cf. chapter 6.2.1.1), Leite (1994) and Radden and Kövecses (1999) differ in their classification of the mappings LOCATION FOR INSTITUTION and MATERIAL OF OBJECT FOR OBJECT . But six out of the ten most frequent cases are based on these mappings: church, White House, Pentagon, and Wall Street all instantiate LOCATION FOR INSTITUTION , and glasses and furs both belong to the mapping MATERIAL FOR OBJECT . Considering the abundance of highly frequent metonyms which are based on these mappings, it thus seems worthwhile to have a closer look at the relative frequencies with which lower-level mappings are used. The following chapters will therefore discuss potential regularities pertaining to the occurrence of more specific mappings. Special attention will be devoted to those mappings which are reversible. Significant preferences for one direction of mapping over the 186 other could very well be linked to laws of ontological salience that are more specific than WHOLE OVER PART . 6.2.1.3 Reversible mappings: Preferred routes and non-default cases Many metonymic mappings are reversible. The most obvious case already examined is the relationship between wholes and parts. Other instances of reversible metonymic relationships are: • CONTAINER - CONTENT : CONTAINER FOR CONTENT ; CONTENT FOR CON- TAINER • CONTROLLER - CONTROLLED : CONTROLLER FOR CONTROLLED ; CON- TROLLED FOR CONTROLLER • PRODUCER - PRODUCT : PRODUCER FOR PRODUCT ; PRODUCT FOR PRODUCER • LOCATION - EVENT : LOCATION FOR EVENT ; EVENT FOR LOCATION • MATERIAL - OBJECT : MATERIAL FOR OBJECT ; OBJECT FOR MATERIAL • CAUSE - EFFECT : CAUSE FOR EFFECT ; EFFECT FOR CAUSE . These reversible patterns are interesting for a study into the conventionality of metonymy because they are intricately connected to ontological salience. For the CONTAINMENT ICM, Radden and Kövecses (1999: 43) point out the following: “the metonymy CONTAINER FOR CONTENTS , i.e., the choice of the container as a reference point, is more natural than the metonymy CONTENTS FOR CONTAINER , i.e., the choice of the things contained”. The reason is that what we perceive is usually the container, while the content is hidden within it, often literally invisible. For the CONTROL ICM, the CON- TROLLER FOR CONTROLLED mapping should be preferred, since HUMANS are more ontologically salient than NON - HUMANS or even INANIMATE THINGS . The relationship between PRODUCER and PRODUCT is a parallel case. In their 1998 article, 171 Kövecses and Radden (1998: 57) emphasize that the mappings belonging to the PRODUCTION ICM are irreversible. According to them, we find only PRODUCER FOR PRODUCT and PLACE FOR PRODUCT MADE THERE , but not the reversed mappings PRODUCT FOR PRODUCER and PRODUCT FOR PLACE . While the latter did not occur in my study, 172 the former was found for nine of the lexical items analyzed. A prime example is perhaps newspaper. That the PRODUCT gives its name to the INSTITUTION producing it is evident here. In fact, two slightly different sub-mappings can be distinguished in this case. One where the PRODUCT stands for the more abstract 171 This is no longer claimed in the later version of the article, Radden and Kövecses (1999). 172 Pure instances of PRODUCT FOR PLACE do indeed not seem to exist. However, one could argue that the mapping occurs as the second part of a metonymic chain and motivates utterances like You’ll find what you’re looking for at BMW (‘BMW retailer’) or Go to Ford (‘Ford retailer’). In both examples PRODUCT FOR PLACE presupposes a PRO- DUCER FOR PRODUCT mapping. Another way of analyzing these examples is to argue that the retailers are part of the respective companies ( INSTITUTION FOR LOCATION ). 187 INSTITUTION producing it and one where the PRODUCT stands for concrete PEOPLE , i.e. JOURNALISTS or other REPRESENTATIVES of the newspaper company. The former sense is realized in (150), the latter in (151): (150) [...] a job as a journalist with the local newspaper [...] (B06-377) (151) Brigadier Dominic Kassiano [...] told the army newspaper [...] (A9M-340) The principle HUMAN OVER NON - HUMAN does not entirely prevent this pattern. Similarly, the preference for CONCRETE OVER ABSTRACT and GOOD GES- TALTS OVER POOR GESTALTS (cf. Radden and Kövecses 1999: 45-48) does not completely preclude the reversal of the LOCATION FOR EVENT mapping. An example is opera, where the typical EVENT gives the LOCATION its name (cf. 152). (152) The usual price for overnight accommodation at Glassdrumman for two people sharing a room, afternoon tea, transport to the opera [...] is £175 [...] (BMD-1483) For the relationship between MATERIAL and OBJECT , one would expect the mapping OBJECT FOR MATERIAL to be more productive than its reversal MA- TERIAL FOR OBJECT since OBJECTS usually have clear boundaries and therefore form GOOD GESTALTS , while a MATERIAL or a SUBSTANCE is unbounded and does not have too many gestalt-like properties (Radden and Kövecses 1999: 47-48). And for the CAUSE - EFFECT relations, the route leading from EFFECT to CAUSE should be preferred due to the more immediate nature of the effect (1999: 47). But are all these predictions confirmed by the results of the corpus study? Let me begin with quite an interesting case, the CAUSATION ICM. Table 19 summarizes the findings for this ICM. The instances are ordered according to the relative frequencies with which they transport a CAUSE FOR EF- FECT or an EFFECT FOR CAUSE meaning respectively (M - total (%), in descending order). Mapping Word-form MYs total (%) M - total (%) M - MYs (%) Ontological salience pencil 26.40 23.40 88.64 Conflict sax 17.24 14.94 86.67 Conflict pen 8.60 3.50 40.70 Conflict brain 2.50 1.00 40.00 ? Conflict heels 40.40 0.10 0.25 Conflict CAUSE FOR EFFECT : stadium 5.45 0.10 1.92 Conflict ecstasy 44.70 44.70 100.00 Typical EFFECT FOR CAUSE : joy 13.70 13.70 100.00 Typical Table 19: The relative frequencies of selected CAUSE FOR EFFECT and EFFECT FOR CAUSE metonyms 188 What is initially obvious in Table 19 is that I encountered a larger number of CAUSE FOR EFFECT metonyms than of EFFECT FOR CAUSE metonyms. Second, for some lexical items the CAUSE FOR EFFECT metonyms are quite rare. This concerns the examples heels and stadium already discussed, but also brain where the OBJECT CAUSING A MENTAL STATE stands for THE MENTAL STATE (cf. 153) and pen, whose CAUSE FOR EFFECT meaning ‘writing in pen’ is with 3.50% still quite infrequent (cf. 154). The other two lexical items, i.e. sax ( INSTRUMENT FOR MUSIC , cf. 155) and pencil, where the resultant meaning is ‘pencil marks (on paper or other surface)’ (cf. 156), instantiate this mapping more frequently. However, none of the CAUSE FOR EFFECT metonyms is nearly as frequent as the EFFECT FOR CAUSE metonym ecstasy (cf. 157). (153) They’re all full of shit, totally shallow and got no brain! (CAD-1397) (154) She had a cruel pen and a reputation for physical violence to match. (ANF-145) (155) The flute did not really go with a guitar and sax. (EDN-1457) (156) The ballerina Lydia Lopokova, pencil on card, 1919 by Pablo Picasso [...] (A47-31) (157) The death suggested a possible previously unrecognised side-effect of ecstasy. (A50-414) Even though the number of different EFFECT FOR CAUSE mappings is, with only two, not high enough to make any definite statements about the conventionality of this mapping, it seems that EFFECT FOR CAUSE leads to more entrenched metonymic usages than CAUSE FOR EFFECT does. Only two of the metonymic CAUSE FOR EFFECT meanings have relative frequencies of over 5%, i.e. those of pencil (23.40%) and sax (14.94%). Most of the relative frequencies of the CAUSE FOR EFFECT metonyms are very low (e.g. brain with 1.00%, and heels and stadium, both with only 0.10%). The relative frequencies of the two EFFECT FOR CAUSE metonyms are considerably higher than that. The slightly confusing picture in Table 19 becomes somewhat clearer if further factors are taken into account. First, the EFFECT FOR CAUSE metonymies always account for 100% of all metonyms found for the lexical items. They are, besides the literal meanings of the two lexemes, the only other possible meanings and do not have to compete with any other potential senses in speakers’ minds. And they are not combined with any other mappings to form metonymic chains. The CAUSE FOR EFFECT meanings of the six lexemes given in Table 19 all have to compete with other possible metonymic meanings and for two of them, i.e. heels and pencil, the CAUSE - EFFECT mapping is the second mapping within a metonymic chain (cf. Appendix A). This means that in these cases the CAUSE FOR EFFECT meaning can only be realized when another metonym is also at work. Second, the higher usage of EFFECT FOR CAUSE mappings is in accordance with the laws of ontological salience. Both instances of the EFFECT FOR CAUSE meaning correspond to the principle IMMEDIATE OVER NON - IMMEDIATE . Most examples of the mapping CAUSE FOR EFFECT , however, behave differently. As indicated in Table 19, five of them are not typical in this respect, but feature 189 a clear conflict of different laws of salience. Heels (‘sound caused by heels’), for example, deviates from IMMEDIATE OVER NON - IMMEDIATE , but corresponds to other laws, such as CONCRETE OVER ABSTRACT and GOOD GESTALT OVER POOR GESTALT . The same holds for sax ‘music’. Pen ‘writing in pen’ also deviates from IMMEDIATE OVER NON - IMMEDIATE , but corresponds to CONCRETE OVER ABSTRACT , just like pencil ‘marks in pencil’ and stadium ‘sound effect’. The only representative of CAUSE FOR EFFECT metonymies which can hardly be said to deviate from the principle IMMEDIATE OVER NON - IMMEDIATE is brain ( OBJECT CAUSING MENTAL STATE FOR MENTAL STATE ). One could argue that the general tendency to put BODILY OVER MENTAL neutralizes the question of immediacy. Moreover, the CAUSE FOR EFFECT meaning of brain is also supported by CONCRETE OVER ABSTRACT . Therefore one could also hold that brain is in line with the laws of ontological salience, hence the question mark in Table 19. Taken together, the different factors mentioned all point in the same direction: C AUSE FOR EFFECT seems to be the weaker partner in the CAUSATION ICM. What is interesting, however, is that the conflicts of different laws of ontological salience do not invariably lead to very low relative frequencies of the metonymic CAUSE FOR EFFECT meanings. At least the meaning ‘marks in pencil’ for pencil seems with 23.40% quite common and entrenched to a certain degree. 173 While the laws of ontological salience can serve as rough guidelines to predict whether a certain mapping will be accepted by the speech community at large, there will always be cases which deviate from these predictions. Radden and Kövecses (1999: 52-54) discuss some of these “overriding factors”. Intended rhetorical or social-communicative effects (e.g. euphemisms such as to go to the bathroom) can lead to a marked preference for metonyms which violate what they (1999: 52) call “the communicative principle CLEAR OVER OBSCURE ”. While Radden and Kövecses may be right in pointing out that examples like The pen is mightier than the sword or wash one’s hands, which also violate CLEAR OVER OBSCURE , are felt to be figurative, one can doubt whether the CAUSE FOR EFFECT meaning of pencil is considered very figurative by the average speaker. Probably it is just a very convenient means for conveying the meaning in a concise way and for avoiding any otherwise necessary circumlocutions. The demand for a linguistic form to transport this meaning seems to be quite considerable. 174 173 Based on the mean value of 7.17% and the standard deviation of 9.75 normal frequencies of CAUSE FOR EFFECT metonymies reach up to 16.92%. In general, the majority of the results (68%) fall between ‘mean value ± 1 standard deviation’ (Albert and Koster 2002: 79). The CAUSE FOR EFFECT meaning of pencil is therefore higher than expected. 174 For pen, the demand for a CAUSE FOR EFFECT meaning is markedly lower, as the lower relative frequency of the meaning shows. The reason is probably related to extralinguistic reality. The CAUSE FOR EFFECT meaning of pen is largely limited to ‘writing’, while for pencil many of the examples in the BNC imply a drawing context. 190 For the CONTAINMENT ICM the picture is much clearer. The results for CONTAINER - CONTENT relations are summarized in Table 20. The CONTAINER FOR CONTENT metonymies, which in general obey the principle VISIBLE OVER NON - VISIBLE , are obviously preferred. However, for two word-forms the relationship is not entirely typical. Glasses are by definition transparent so that the content is usually visible as well. And for dish ‘meal’ one could even hold that the food is actually more visible than the dish itself. 175 What matters here is that the CONTENTS are ‘held together’ by the respective CON- TAINERS . For glasses, the CONTENT acquires a certain degree of boundedness through the CONTAINER . And for dish, the motivating factor is that different food items belonging to a meal are considered as a coherent whole by way of being served on one dish. Book ‘content of book’, which has the highest relative frequency here, derives further support from the principle CON- CRETE OVER ABSTRACT . Mapping Wordform MYs total (%) M - total (%) M - MYs (%) Ontological salience book 49.60 44.30 89.31 Typical dish 32.70 32.70 100.00 Typical kettle 22.05 19.46 88.24 Typical bottle 10.00 9.90 99.00 Typical packet 8.10 7.40 91.36 Typical label 31.60 2.70 8.54 Typical CONTAINER FOR CONTENT : glasses 98.40 2.00 2.03 Typical wine 0.40 0.40 100.00 Deviation chicken 68.70 0.20 0.29 Deviation CONTENT FOR CONTAINER : ketchup - - - - - - - - - Deviation Table 20: The results for CONTAINER FOR CONTENT and CONTENT FOR CONTAINER The relative frequencies of the single instances of CONTAINER FOR CONTENT metonymies are not very uniform (standard deviation: 16.18). The relative frequencies of the two instances found for CONTENT FOR CONTAINER are, however, with 0.40% and 0.20% quite similar. 176 The other CONTENT FOR 175 While this is true in everyday contexts, it is different on very formal occasions where serving dishes with covers are used, which ‘hide’ the content. Another context in which food is hidden in its container is in hospitals, where the meals are also usually served to the patients in covered containers. 176 Other lexemes that can be used to convey a CONTENT FOR CONTAINER meaning, but were not part of my study, are, for instance, jam ‘jam jar’ and tomatoes ‘can of tomatoes’. However, it is doubtful that the inclusion of these and further items would have changed the overall picture significantly, because clear examples of CONTENT FOR CONTAINER are rare. This is related to the fact that containers and their contents are often treated as a conceptual unity, so that a straightforward stand-for relationship is not easy to establish (cf. my discussion of kettle and buses in chapter 5.2.1.1). In addition, corpora often do not provide enough contextual information to distinguish 191 CONTAINER metonym I was initially searching for, ketchup, could not be found in the BNC. The CONTAINER - CONTENT relations therefore behave exactly as predicted by the laws of ontological salience. Metonyms which deviate from the principle VISIBLE OVER INVISIBLE and use vehicles which are invisible, and, as in my examples, substances which do not have clear-cut boundaries, can hardly be used for a successful reference to their containers. As concerns the quite varied relative frequencies of the different instances of the mapping CONTAINER FOR CONTENT , one could maintain that the more specific the relationship between CONTENT and CONTAINER , the easier the reference to the CONTENT by way of the CONTAINER . The lowest relative frequencies were observed for bottle, packet, label, and glasses. Here, the contents are almost undetermined. For glasses and bottles, the content is typically fluid, but what kind of fluid is not specified. And packets and labels (‘containers with labels’, i.e. as second mapping in a metonymic chain that leads from an ATTRIBUTE OF THE OBJECT to the OBJECT itself and only then from the CONTAINER to the CONTENT ) are undeniably unspecific with respect to their contents. This picture changes for kettle. Kettles are almost always used as containers for water and they are conceptually linked to a specific purpose, the boiling of water. Dish is similarly specific and book even more so since books are typically containers for some kind of text (or pictures, especially when it comes to children’s books). In any case, the content of a book is more abstract that the content of any other container mentioned here. Therefore the concrete CONTAINER serves as an ideal vehicle to convey the metonymic meaning. Another ICM where my results coincide with the predictions linked to ontological salience is the CONTROL ICM. While the mapping CONTROLLED FOR CONTROLLER was found with many of the word-forms, it tends to account for a lower percentage of word meanings than its reversal CONTROL- LER FOR CONTROLLED . The results for the single instances are summarized in Table 21 (below). The CONTROLLER FOR CONTROLLED examples are all typical in terms of ontological salience since they instantiate the preference for HUMAN OVER NON - HUMAN . For airlines, this is less evident, but ultimately the institution consists of human beings, as manifest in (158) and (159). (158) is an example of the mapping INSTITUTION FOR PEOPLE , and (159) shows that the institution controls its planes, i.e. is an instance of the mapping CONTROLLER FOR CONTROLLED . For the proper nouns, Napoleon and Connery, HUMANS are used as vehicles. Connery as an instance of the mapping CONTROLLER FOR CONTROLLED is linked to the sub-mapping ACTOR FOR ROLE (cf. 160). Napoleon occurs twice, since the mapping CONTROLLER FOR CONTROLLED can be cases where, for example, tomatoes refers to the canned variety from those where it simply refers to several tomatoes, i.e. to more than one instance of the vegetable. 192 related to two different sub-mappings, i.e. RULER FOR ARMY 177 (cf. 161) and PERSON FOR ACTIVITY OF PERSON (cf. 162). Mapping Wordform MYs total (%) M - total (%) M - MYs (%) Ontological salience Napoleon 16.51 13.34 80.80 Typical airlines 30.79 10.04 32.62 Typical Connery 3.48 3.48 100.00 Typical CONTROLLER FOR CONTROLLED : Napoleon 16.51 0.26 1.60 Typical cameras 9.20 9.20 100.00 Deviation pen 8.60 5.10 59.30 Deviation crime 5.90 5.00 84.75 Deviation BMW 70.21 4.64 6.61 Conflict Renault 56.61 3.70 6.54 Conflict sax 17.24 2.30 13.33 Deviation telephone 20.50 2.30 11.22 Conflict football 96.80 1.70 1.76 Deviation scalpel 1.09 1.09 100.00 Deviation buses 9.60 1.00 10.42 Conflict CONTROLLED FOR CONTROLLER : pencil 26.40 0.30 1.14 Deviation Table 21: The results for CONTROLLER FOR CONTROLLED and CONTROLLED FOR CONTROLLER (158) The pilots resigned six weeks ago when the airlines refused a demand by the Australian Federation of Air Pilots for direct negotiations [...] (A1S- 346) (159) If airlines can fly wherever and whenever they want to [...] (ABF-356) (160) Medicine Man Sean Connery stalks the South American jungle as a crusty Scots scientist on the brink of discovering a cancer cure [...] (AYM- 53) (161) His humiliation was smaller than that to which the country had been subjected on being invaded by Napoleon in 1812 [...] (HY7-667) (162) Napoleon and Hitler gave European unity a bad name. (ABH-366) The CONTROLLED FOR CONTROLLER cases either deviate from the laws of ontological salience or at least feature a conflict of different principles. For example, scalpel meaning ‘surgeon’ is inconsistent with HUMAN OVER NON - HUMAN . Buses ‘bus drivers’ also contradicts this principle, but is in accordance with INTERACTIONAL OVER NON - INTERACTIONAL (Radden and Kövecses 1999: 51) and perhaps also with VISIBLE OVER NON - VISIBLE , if one construes the vehicle as a container. This possibility is also the reason why BMW ‘driver’ is specified as a case of conflicting laws of salience. 177 The description of the mapping as RULER FOR ARMY is not entirely accurate, since the BNC also contains concordance lines where Napoleon stands not necessarily for his army but for other people who acted on Napoleon’s orders (e.g. a road which was built by Napoleon; B0Y-3579). However, since the vast majority of contexts indicates that the target are his armed forces proper, I chose the wording RULER FOR ARMY . 193 None of the CONTROLLED FOR CONTROLLER instances reaches as high relative frequencies as the two most common CONTROLLER FOR CONTROLLED examples, i.e. airlines and Napoleon (in the RULER FOR ARMY meaning). In fact, only one of the relative frequencies of the CONTROLLED FOR CONTROL- LER items, i.e. cameras (9.20%), exceeds the median of the CONTROLLER FOR CONTROLLED ones (6.76%). The CONTROLLED FOR CONTROLLER group is fairly consistent in terms of frequency. The standard deviation of 2.58 is quite low. The two word-forms which stand out against the others are the already mentioned cameras, which is somewhat more frequent than the mean value of 3.30% and the standard deviation predict, and pencil, which is less frequent. Normal values for this pattern should lie between 0.72% and 5.88%. Again, there is a noticeable difference between the behaviours of pen and pencil. Pen has a moderately high frequency of CONTROLLED FOR CON- TROLLER occurrences if compared to the mean value of 3.30% and even more so if compared to the median of this group which lies at 2.30%. Pencil, on the other hand, has a strikingly low frequency. A possible reason is the difference between the specificity of the relations between a pen and an author 178 and a pencil and its user. The pen is maybe the prototypical writing instrument 179 and thus conceptually close to the writer. Pencils can also be used for writing, but their use is perhaps more associated with children. Moreover, they can be used for drawing. And this latter idea is transported by the CONTROLLED FOR CONTROLLER instances of pencil, as in [...] ‘so fine a pencil is worthy of encouragement’ (ALU-695). The mental path leading from a pencil to an artist is maybe not as straightforward as the one leading from a pen to a writer, considering, too, that a pencil is not so typical an instrument associated with artists in general. Remarkably, however, also instances like buses and scalpel where one could assume such a more or less straightforward mental path are relatively infrequent. Buses are usually not driven by laypersons, but by proper bus drivers, and scalpels are mainly used by surgeons or other kinds of doctors, but not very often by other persons. The weighty principle HUMAN OVER NON - HUMAN seems to prevent a frequent use of these items in a metonymic way to refer to human beings. What is also of interest with respect to Table 21 is that instances which clearly deviate from the laws of ontological salience, and instances which merely feature a conflict of different principles, do not significantly differ in terms of their relative frequencies. One might expect that cases which are entirely inconsistent with the general tendencies would be less frequent 178 Author is meant here in quite a wide sense, and covers not only people who have made writing their profession and earn their living by writing texts, such as journalists or novelists and the like, but refers to anybody who writes any kind of text. 179 How deeply entrenched the association between writers and pens is, is most obvious in more fixed expressions, such as poison pen letter. Even if written on a computer and printed out, anonymous abusive letters will still be called poison pen letters. The expression is lexicalized (cf. chapter 4.2.1). 194 than cases which are subject to competing motivations. This is, however, not the case. There is no correlation whatsoever between these two aspects, at least not for the CONTROL ICM. Other reversible mappings also exhibit a general preference for one direction of mapping. The PRODUCTION ICM is a particularly interesting case, since it allows a relatively fine-grained analysis. It is not only possible to analyze the two different mapping directions. In addition, the single mappings can be subdivided into PHYSICAL and ABSTRACT PRODUCTS . The generally preferred mapping direction leads from PRODUCER to PRODUCT . The results for this mapping are given in Table 22. Wordform MYs total (%) Mapping M - total (%) M - MYs (%) BMW 70.21 PRODUCER FOR PHYSICAL PRODUCT : PRODUCER FOR CONSUMER PRODUCT 62.67 89.26 Renault 56.61 PRODUCER FOR PHYSICAL PRODUCT : PRODUCER FOR CONSUMER PRODUCT 43.74 77.26 Shakespeare 35.50 PRODUCER FOR ABSTRACT PRODUCT : AUTHOR FOR WORKS 34.90 98.31 Picasso 8.58 PRODUCER FOR PHYSICAL PRODUCT : PRODUCER FOR ARTISTIC PRODUCT 8.32 97.06 Einstein 3.52 PRODUCER FOR ABSTRACT PRODUCT : THEORIST FOR THEORY 2.26 64.29 Einstein 3.52 PRODUCER FOR ABSTRACT PRODUCT : AUTHOR FOR WORKS 0.50 14.29 buses 9.60 NON - HUMAN PRODUCER FOR PRODUCT : PRODUCER FOR PHYSICAL PRODUCT ‘exhaust’ 0.10 1.04 Table 22: The results for PRODUCER FOR PRODUCT All instances are not typical in terms of ontological salience. The PRODUCERS may be human, but are less immediate than the PRODUCTS . And buses even clearly deviates from the general laws, since the PRODUCER of the ‘exhaust’ meant here is non-human. Leaving this totally unusual example aside and looking at the mappings which involve physical and abstract products respectively, we find that the mapping PRODUCER FOR PHYSICAL PRODUCTS , especially for CONSUMER PRO- DUCTS , sanctions many concrete uses. The mappings with an abstract product as their target are less common. The exception is Shakespeare, where the mapping PRODUCER FOR ABSTRACT PRODUCT leads to 34.90% of metonymic uses, and thus diverges strikingly from the percentages of metonymic uses relying on the same mapping found for Einstein (realizing the two slightly different sub-mappings given in Table 22). The difference becomes even more apparent when another way of sub-classifying the data is chosen. If CONSUMER PRODUCTS are considered as one group and AR- TISTIC / MENTAL PRODUCTS as another, a marked preference for PRODUCER 195 FOR CONSUMER PRODUCTS over PRODUCER FOR ARTISTIC / MENTAL PRODUCTS can be seen. Both instances realizing the former mapping (BMW and Renault) have fairly high relative frequencies, i.e. 62.67% and 43.74% respectively. As compared to this, those words instantiating the latter have, again with the exception of Shakespeare, quite low relative frequencies of 8.32%, 2.26%, and 0.50%, i.e. they typically lie beneath 10% of all uses. The metonymic construal of consumer goods, especially of cars, by way of referring to the company producing them is thus much more conventional than accessing mental products by way of referring to their respective human producers. The reason for the high relative frequencies of the metonymic construals of cars by way of the car producing companies may be that the companies are so well-known for exactly these products, since they do not produce any others. If a company produced many different kinds of consumer products, however, chances are that the metonymic construals of these products by way of the company name would be markedly less frequent. 180 In general, abstract products are not only less qualified as metonymic targets, they are also less good vehicles. Table 23 (below) provides an overview of the results for PRODUCT FOR PRODUCER . None of the instances of this mapping reaches as high relative frequencies as some of the PRODUCER FOR PRODUCT metonyms do. This is certainly interesting in its own right, but even more so if one takes into account that - similar to the CAUSATION ICM or the CONTROL ICM - this less typical mapping direction was found with more lexical items than the typical one. Like the PRODUCER FOR PRODUCT instances, all of the examples in Table 23 show conflicts concerning ontological salience. While they violate the principle HUMAN OVER NON - HUMAN , they conform to INTERACTIONAL OVER NON - INTERACTIONAL and/ or IMMEDIATE OVER NON - IMMEDIATE . Book meaning ‘author’ (e.g. The book also says that [...]; ACR-709) can be used to illustrate this: A book cannot literally say something, only its author can, but the book is what we interact with. It is much more immediate than the author. The group of abstract products embraces any kind of communicative acts, mostly in the written (e.g. book, newspaper) but also in the spoken medium (radio, opera). Nine of the 13 examples belong to this class. The mean 180 Cases in point are company names like Sony or Siemens. As these words were not part of my study, a comprehensive analysis of the two items was not carried out. A relatively simple way to verify the assumption that such company names are not frequently used as instances of the mapping PRODUCER FOR PRODUCT is, however, to search for the phrases a Sony and a Siemens, because metonymic PRODUCER FOR PRO- DUCT meanings are often marked on the linguistic surface by the occurrence of the indefinite article. A phrase query for a Siemens in the BNC, for example, only returns one hit, but as the context indicates ([…] a Siemens AG unit; CNR-32), the meaning is not a metonymic one in this case. The phrase a Sony occurs 26 times in the BNC, but typically Sony is used as a premodifier of nouns as in Sony walkman, Sony mobile phone, Sony Discman, etc., i.e. not as a PRODUCER FOR PRODUCT metonymy. 196 value of ABSTRACT PRODUCT FOR PRODUCER is 6.74%, the median value is 6.50% The remaining four examples have concrete products as their vehicles. The mean value of this group amounts to 13.20%, the median is 13.60%. The most frequent example is label, where a CONSUMER PRODUCT is the vehicle. Newspaper is also quite commonly used to refer to the institution producing it. The other two instances are less frequent. Paper for ‘newspaper company’ still occurs in 4% of the instances analyzed, but hands is a very uncommon instance. The example from the BNC is given below (cf. 163). Wordform MYs total (%) Mapping M - total (%) M - MYs (%) label 31.60 PHYSICAL PRODUCT FOR PRODUCER : CONSUMER PRODUCT FOR INSTITUTION 25.50 80.70 newspaper 35.40 PHYSICAL PRODUCT FOR PRODUCER : PHYSICAL PRODUCT FOR INSTITUTION 23.20 65.54 newspaper 35.40 ABSTRACT PRODUCT FOR PRODUCER : WORK FOR AUTHOR 12.20 34.46 law 21.00 ABSTRACT PRODUCT FOR PRODUCER : WORK FOR AUTHOR 10.70 50.95 paper 49.20 ABSTRACT PRODUCT FOR PRODUCER : WORK FOR AUTHOR 10.00 20.33 opera 16.40 ABSTRACT PRODUCT FOR PRODUCER : ABSTRACT PRODUCT FOR INSTITUTION 9.70 59.15 radio 10.30 ABSTRACT PRODUCT FOR PRODUCER : MEDIUM FOR COMMUNICATOR 6.50 63.11 book 49.60 ABSTRACT PRODUCT FOR PRODUCER : WORK FOR AUTHOR 5.30 10.69 paper 49.20 PHYSICAL PRODUCT FOR PRODUCER : PHYSICAL PRODUCT FOR INSTITUTION 4.00 8.13 radio 10.30 ABSTRACT PRODUCT FOR PRODUCER : ABSTRACT PRODUCT FOR INSTITUTION 3.80 36.89 label 31.60 ABSTRACT PRODUCT FOR PRODUCER : WORK FOR AUTHOR 2.00 6.33 packet 8.10 ABSTRACT PRODUCT FOR PRODUCER : WORK FOR AUTHOR 0.50 6.17 hands 2.40 PHYSICAL PRODUCT FOR PRODUCER : ARTISTIC PRODUCT FOR PRODUCER 0.10 4.17 Table 23: The results for PRODUCT FOR PRODUCER (163) The show emerged from the work of Arielle P. Kozloff [...], who since 1977 has been compiling a photographic census of eighteenth-century Theban wall paintings [...], with an aim towards identifying specific hands among the artisans. (EBW-926) The metonymy is actually a little more complex than the mapping given above suggests. The hands as ‘producers’ stand for the ARTISTIC PRODUCTS (with a characteristic style of painting), since what can be studied is only 197 the paintings produced by the persons’ hands, but the vehicle is additionally motivated by a BODY PART FOR PERSON metonymy. Even though the overall number of ABSTRACT PRODUCT FOR PRODUCER examples is higher, none of them occurs as often as the two comparatively frequent instances of PHYSICAL PRODUCT FOR PRODUCER . Taken together, the results for the PRODUCTION ICM indicate that the concrete-abstract continuum plays a role for the use of metonyms. For the metonymic construal of PRODUCERS in the PRODUCTION ICM, speakers prefer concrete products to abstract ones. This can be linked back to what has been said concerning the laws of ontological salience which are violated by mappings of either direction. The PRODUCER FOR PRODUCT mappings violate the principle IMMEDIATE OVER NON - IMMEDIATE . This seems to have less consequences if the products, which are actually immediate but not construed as such, are of a physical nature. The PRODUCT FOR PRODUCER mappings do not adhere to the principle HUMAN OVER NON - HUMAN . This is more likely to be socially sanctioned when the vehicle chosen belongs to the realm of concrete things. And particularly unconventional are cases where the vehicle is not typically associated with abstract communicative acts at all. The prime example to illustrate this is packet (cf. 164). (164) After sowing sieve compost over the tray to cover the seeds (unless the packet states otherwise). (ACY-1265) Packets are quite concrete and have a specific function. That most packets of consumer products bear much written information, which necessarily has an author, 181 is a much less salient aspect of packets. The conceptual path from packets to authors of written texts found on the packets is therefore difficult. If the associative link between the vehicle and abstract communicative acts is stronger, the metonym is much easier to understand - a fact which is reflected in the much more frequent occurrence of ABSTRACT PRODUCT FOR PRODUCER mappings with newspaper, paper or law. The last reversible mapping I would like to comment on is CATEGORY AND MEMBER . Only two examples of CATEGORY FOR MEMBER occur in my data, drugs and pill. The reversed mapping was found for many more lexical items. Mostly, it occurs with proper names as shown in Table 24 (below). The examples of MEMBER FOR CATEGORY are all typical with respect to ontological salience. The members are more concrete than the abstract 181 One could also argue that the relationship between the packet and the information found on it is one of CONTAINER and CONTENT . The same holds true for label, newspaper, and paper. The containment relation is, however, not very convincing in most of these cases. A sheet of paper is certainly not a typical container, just like a label. And positing a containment relation between a packet and the written information on it would have caused additional difficulties in the analysis, since the lexical item is also an example of quite a typical and concrete containment metonymy - as also evident in example (201) below, in the phrase to see what it could do. 198 category. In the case of aspirin, the concreteness of the vehicle is further supported by the cultural preference for TYPICAL OVER NON - TYPICAL (Radden and Kövecses 1999: 49). An example illustrating this usage of aspirin to refer to the whole category of painkillers is (165): (165) The problem is made worse because consumers are rarely told of the risks associated with drugs: most, not unreasonably, expect their medicine to do them good. A survey of 56 different painkillers conducted recently by Britain’s Consumers’ Association found that instructions accompanying most products were inconsistent, inadequate and unclear. Only three out of 26 aspirin bottles carried warnings against use by pregnant women. (ABH-2102) Mapping Word-form MYs total (%) M - total (%) M - MYs (%) Ontological salience aspirin 30.03 30.03 100.00 Typical Watergate 84.54 7.22 8.54 Typical Napoleon 16.51 2.77 16.80 Typical Pearl Harbo(u)r 31.03 2.30 7.41 Typical car 5.60 1.50 26.79 Typical ecstasy 44.70 0.83 1.85 Typical Vietnam 39.00 0.80 2.05 Typical Einstein 3.52 0.75 21.43 Typical Exxon 35.86 0.69 1.92 Typical Shakespeare 35.50 0.30 0.85 Typical Picasso 8.58 0.25 2.94 Typical MEMBER FOR CATEGORY : book 49.60 0.10 0.20 Typical drugs 55.50 55.50 100.00 Deviation CATEGORY FOR MEMBER : pill 44.95 44.95 100.00 Deviation Table 24: The results for MEMBER OF CATEGORY FOR CATEGORY and CATEGORY FOR MEMBER OF CATEGORY The instances of the reversed mapping do, however, deviate from what ontological salience predicts. The category is less specific and should therefore not frequently occur as a metonymic vehicle. 182 The frequencies indicated stand, however, in marked contrast to these predictions. The only frequent MEMBER FOR CATEGORY metonym is aspirin. All the other instances are much less common. The only other instance which is not entirely unusual when used as a paragon is Watergate (e.g. a French Watergate; EF6- 1208). The other paragons, such as Vietnam or Picasso, are not so frequent. The two CATEGORY FOR MEMBER metonyms, however, are deeply entrenched. A possible explanation might be that the intended targets of both 182 Alternatively, one could categorize all of the examples in Table 24 as instances where different laws of ontological salience are in conflict, if one considers the higher ontological salience of WHOLES as opposed to PARTS . But since the wholes are highly abstract in these cases, I prefer the classification given above. 199 can be regarded as potentially sensitive issues. The wish to avoid more direct and specific references to these targets may have caused the conventionalization of the CATEGORY FOR MEMBER metonyms. Another, perhaps more convincing, line of argument is that both, drugs and pill, are actually shortened versions of longer phrases (illegal drugs, contraceptive pill), which are preferred by the speech community because of their succinctness and only supported - but not really generated - by the existence of the mapping CATEGORY FOR MEMBER . The expectations raised by a consideration of the two different mapping directions and their respective relations to the laws of ontological salience are clearly not fulfilled here. The communicative need to refer to a category by way of mentioning a salient member does not arise often. Particularly utterances containing paragons are therefore probably felt to be figurative. It is often marked on the linguistic surface that the words are not used literally in those cases where the mapping MEMBER FOR CATEGORY is the second one in a metonymic chain (e.g. book, Vietnam, Watergate; cf. Appendix A) and that they are also not to be taken as realizations of other (more) common patterns. Frequently the paragons are preor postmodified, as in another Vietnam (e.g. CBC-8192), Scottish Watergate (K5D-9923), and the Shakespeare of Portugal (CA7-1833). In this chapter I have concentrated on the relationship between reversible metonymic mappings and the laws of ontological salience. In general, every analysis attempting to describe concrete instances in terms of ontological salience is bound to be problematic. One reason is that there is no fixed set of such principles. The second point is that - even if such a closed set of principles is established, as for example is done by Radden and Kövecses (1999) - it remains unclear how weighty the single principles are. It is thus hard to predict how cases which conform to one principle, but violate another will behave. And third, there is, of course, always the possibility that the speech community may throw all these principles over board and show a marked preference for metonyms, such as drugs ‘illegal substances’ and pill ‘contraceptive pill’, which clearly deviate from them. Despite the problems involved in an analysis of metonyms in terms of the laws of ontological salience, it could be demonstrated that usually one of the directions of mapping is clearly preferred. This illustrates that any consideration of the conventionality of metonymy must not be restricted to the ICM in question without taking the direction of mapping into account. For example, not all CONTAINER - CONTENT relations are equally accepted by the speakers. It is only the CONTAINER FOR CONTENT mapping which leads to linguistic metonyms that have a conventionalization potential worth mentioning. Most patterns examined in this chapter display a preference for the direction which is more in accordance with the general laws of ontological salience. Only the relationship between a CATEGORY and a MEMBER OF THE CATEGORY diverges from this tendency. But the laws of ontological salience 200 are not the only factor to play a role when it comes to the conventionality of metonyms. Otherwise the variation found within the single groups would not be so marked in many cases. For the PRODUCTION ICM it could be demonstrated that the abstract-concrete continuum is of relevance for the choice of metonymic targets and vehicles. And for the CONTAINER FOR CONTENT mapping, the specificity of the relation between a container and its content could be identified as a decisive factor. That the specificity of the relationship between vehicle and target is important has also been demonstrated for the most frequent PRODUCER FOR PRODUCT metonyms, i.e. BMW and Renault. On the more comprehensive level of mappings ontological salience is thus a useful instrument to approach the conventionality of metonymies. However, to explain the varying relative frequencies of diverse instances generated by the same metonymic mapping, further factors have to be taken into consideration. Viewed from a cognitive angle, the abovementioned specificity of the relationship between the vehicle and the target that decides upon the degrees of conventionality of different linguistic metonyms is related to the structure of the vehicle and target concepts involved: For more conventional metonyms, the associative link between the two is stronger than for less conventional metonyms. The target or at least aspects of the target are a highly relevant part in the conceptual representation of the vehicle. In other words, more conventional linguistic metonyms are characterized by the fact that attributes of the target concept are salient in the vehicle concept. How exactly salient attributes influence the conventionality of metonymies is discussed in the following chapter. 6.2.1.4 The influence of conceptual structure and salient attributes In the previous chapters (cf. esp. chapter 4.2.2.1) the notion of salience was introduced as connected to whole entities: the salience of real-world entities (ontological salience) and the salience of meanings (cognitive salience). Both have been found to play a significant role in the conventionalizing of metonymies as well as metaphors. As the examination of paragons has shown (cf. chapter 4.2.2.2), some metonymic construals are only possible if the speakers of a language agree as to which properties should be associated with a given metonymic vehicle. But not all knowledge shared by the speech community counts in such cases. Only certain aspects can be picked out for metonymic references - only those aspects which come to speakers’ minds immediately when thinking about the entity in question. In other words, it is only highly salient attributes which can form the basis of paragons. The notion of attribute salience is taken over from Ortony et al. (1985; cf. also Katz 1982; Ortony 1979: 162), who define it as “the prominence or importance of an attribute in a person’s representation of an entity or category” (1985: 570). To use their example, the category of GENERAL ANESTHET- 201 ICS AND SLEEP - INDUCING DRUGS carries a highly salient attribute ‘induces sleep’, which connects the members of the category. Cognitive salience is concerned with the temporary activation and entrenchment of concepts. But concepts themselves are complex structures made up of a number of attributes, so that not only whole concepts can vary regarding their cognitive salience, but also the attributes they consist of. The relevance of attribute salience for metonyms can be illustrated with the following example: (166) Take some aspirin, and lie down if you need to. (HA6-2613) As shown above (cf. chapter 6.2.1.3), the MEMBER OF CATEGORY FOR CATE- GORY metonym aspirin for ‘any pain relieving tablet’ is considerably more frequent than other instances of the same mapping. This raises the question as to why this specific metonymic vehicle is so appropriate to transport the intended meaning. From a pragmatic perspective, the communicative goal of the utterance in (166) is to convey the idea that any type of pain relieving tablet might help. Aspirin is undoubtedly the most typical and common member of the category of 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 communicative aim. The most prominent feature of the target concept of PAINK- ILLERS , or more accurately the attribute that turns different types of painkillers into a coherent category, is highly salient in the mental representation of the vehicle aspirin (cf. Figure 16, below). The relationship between the vehicle and the target is a very typical one. This is responsible for the fact that the metonym is highly apt and can be decoded rapidly, since with respect to metonymies, aptness depends on “the typicality of the relation between the two objects in question” (Markert and Hahn 2002: 152). 183 If the speaker, however, had said take an ibuprofen, instead of take an aspirin, the message would be a different one. Not everybody knows that ibuprofen is a name of a pain relieving substance. As a consequence the link between the category member and the category as a whole is not strong enough in order to be used for a metonymic reference. The only way to interpret the utterance would thus be as an advice to take not any pain relieving tablet, but a tablet containing the substance ibuprofen. 183 Originally, the term aptness was used to describe the potential of metaphorical expressions to convey the intended meaning successfully and without causing too much cognitive effort for the addressee and is also often referred to as metaphor goodness. Whether metaphor aptness really influences metaphor processing is a disputed issue (cf. e.g. Blasko and Connine 1993; Glucksberg and Keysar 1990; Goldvarg and Glucksberg 1998; Johnson and Malgady 1980; Shen 1992; Tourangeau and Sternberg 1981; cf. also chapter 4.3.3.2). 202 ASPIRIN PAIN RELIEVING TABLETS Aspirin Tylenol Ibuprofen etc. Figure 16: Take some aspirin - Metonymy and attribute salience Attribute salience can also clarify the difference between two examples discussed by Barcelona (2002a: 231; cf. also Radden and Kövecses 1999: 55; Taylor 2002: 324-325, 2003: 125): (167) I have just bought a Picasso. (168) *I have just bought a Jane. Both examples are PRODUCER FOR PRODUCT metonymies and both are equally well supported by the principles of ontological salience, such as HUMAN OVER NON - HUMAN or CONCRETE OVER ABSTRACT . But while (167) is not an entirely unconventional way of referring to a picture by Picasso (8.32% of all meanings; cf. chapter 6.2.1.3) and can be understood by every competent member of the speech community, (168) could hardly be used to refer to a picture painted by a woman called Jane - except perhaps in very restricted contexts. Referring to Taylor’s (1995, 2002; cf. also Taylor 2003) ideas, Barcelona (2002a: 230-231) argues that the conventionalization of the latter example is blocked by the lack of “a specific cultural [...], socialinteractional, or aesthetic principle”. Or as Taylor (2002: 325) puts it, metonymies have to be “sanctioned by a body of knowledge and beliefs encapsulated in an appropriate frame”. The point is that we have a very well-structured mental representation of P ICASSO . The concept contains knowledge in the form of attributes like ‘was an artist’, ‘produced paintings and sculptures’, ‘is important in art history’, ‘used certain techniques’, ‘was a man’, ‘was a father’, etc. For J ANE , however, things are different. The speech community does not share any specific knowledge concerning this person. The only piece of information one has or, rather, can arrive at on the basis of the name, is that the person is female. The J ANE concept is almost empty and extremely underspecified, while the P ICASSO concept is full of pieces of information of different degrees of salience. The attribute ‘was a father’, for example, will not be the aspect that comes to the language users’ mind first in a neutral context. Rather they will immediately connect the name with the attribute ‘was an artist’ (cf. Figure 17). ‘relieves pain’ 203 a Picasso a Jane P ICASSO J ANE was a man was a father was an artist used certain techniques etc. is/ was a woman ? ? ? A P ICASSO A J ANE - Guernica - The Three Dancers etc. ? ? ? Figure 17: The structure of concepts and the conventionality of metonyms Barcelona (2002a) also points out that the difference between (167) and (168) has to do with salience: Works of art form salient sub-domains 184 of stored concepts of artists, so that (167) obeys the principle SALIENT OVER NON SALIENT , whereas (168) does not. In addition, the conventionalization of (167) is favoured by the “cultural principle whereby works of art are regarded as unique products of the creative genius of artists, as an extension of their personality” (Barcelona 2002a: 231). Example (167) thus fulfils all three of the major conditions for the conventionalization of metonyms quoted by Barcelona (2002a: 230): 1. It is based on one of the entrenched conceptual mappings and is 2. socially sanctioned, because it is 2.a in accordance with general cognitive and communicative principles 2.b and with a specific cultural principle. According to Barcelona (2002a: 231), (168) also fulfils the demands in (1) and (2a), but fails as far as (2b) is concerned. While this is true with respect to the principles HUMAN OVER NON HUMAN , CONCRETE OVER ABSTRACT and GOOD GESTALT OVER POOR GESTALT mentioned above, it does not apply for the principle SALIENT OVER NON SALIENT . There is no salient sub-domain whatsoever of the concept J ANE as it is virtually empty, at least if one considers the speech community as a whole. It is therefore the aspect of salience which is responsible for the difference between the two examples and not, as Barcelona holds, the lack of a specific cultural principle. This cultural principle exists - it links painters to their pictures, authors to their 184 While Barcelona (2002a) speaks of salience with respect to domains and sub-domains, I prefer the idea of attribute salience, because the latter can also be applied for a description of metaphors, cf. chapter 7.4. 204 books, etc. - so why does it not link Jane to her works of art? The cultural principle pertains only if the concepts to which it applies (in this case the artists) contain enough and, more importantly, salient information. Shared salient knowledge is, therefore, its necessary precondition. 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. In order to apply the metonym successfully in unmarked contexts, the association between the vehicle and the target must be a distinctive one. 185 What the analysis of my data suggests is that the aptest metonymies which stand the highest chance of being accepted are those where targetrelated attributes are highly salient in the vehicle. Salient attributes are responsible for the specificity and typicality of the relationship between metonymic vehicles and targets, and result in higher-than-average relative frequencies of the corresponding metonyms. As already noted in chapter 6.2.1.3, company names are particularly suited to transport metonymic ‘product’ meanings if the companies are well-known for one type of product, i.e. if the vehicle concepts carry a specific and highly salient targetrelated attribute, in this case the typical products. This is certainly the case for the examples BMW and Renault, which returned the highest relative frequencies in the PRODUCTION ICM. As opposed to other examples like Shakespeare and Einstein, the conventionality of the former also profits from referring to physical products as well as from the fact that the violation of HUMAN OVER NON - HUMAN involved is less pertinent than for Shakespeare and Einstein, since the latter use the names of persons as vehicles. Attribute salience is thus only one of the factors enhancing the conventionality of certain metonyms, but certainly a vital one. Another example where the influence of attribute salience is obvious is the CONTAINER FOR CONTENT mapping (cf. Table 20, chapter 6.2.1.3). One can reasonably assume that the instances with the highest relative frequencies, i.e. book, dish, and kettle, are all characterized by highly salient targetrelated attributes of the form ‘contains written text’, ‘used to serve food’, and ‘used for boiling water’ respectively. This is not the case for packet, label, and glasses, i.e. for the less frequent metonyms. What is stored for packet, for example, is presumably not much more than ‘is a container’, but with regard to the content the concept is likely to be unspecified. Especially when the single instances of a mapping vary considerably in terms of their 185 Cf. also Clark and Gerrig (1983: 560). They studied the comprehension of eponymous verb phrases, such as do a Napoleon or do a Nixon and argue convincingly that this is a matter of salience. However, the degree of salience of an attribute is not entirely fixed, but can very due to contextual influence (cf. e.g. Gibbs and Gerrig 1989). This issue is particularly relevant with regard to metaphors and is discussed in chapter 7.4.1. 205 relative frequencies, as for the PRODUCER FOR PRODUCT and CONTAINER FOR CONTENT metonyms, attribute salience seems to be a useful explanatory tool. The remaining results of the metonymy study will therefore not only be discussed in the light of the laws of ontological salience, but also from the vantage point of attribute salience. As will be shown below, the latter can account for the outliers in terms of relative frequency, i.e. explain why some linguistic metonyms are much more common than other instances that rely on the same reasoning patterns. In addition, the complementation of the laws of ontological salience by attribute salience has one further advantage: Not all metonymic mappings are reversible and for some ICMs which allow two directions of mapping, my data are restricted to one of them. For example, within the POSSESSION ICM a search for POSSESSOR FOR POSSESSED was not feasible for reasons outlined above (cf. chapter 5.1.2.1). In these cases, an examination of the results with regard to ontological salience is not an easy task, since ontological salience is best equipped to describe preferences on a more comprehensive level, i.e. for one of two possible directions of mapping. Because I often lack the data for one of the directions, the frequency results cannot be related to what ontological salience predicts in the same systematic way as for ICMs where instances of both mapping directions were examined. The same holds true for cases where my data are too diverse to allow a simple comparison of the mapping directions. This concerns first and foremost the general relation between ATTRIBUTE OF ENTITY and ENTITY . Examples of ATTRIBUTE OF ENTITY FOR ENTITY abound and therefore call for a differentiation of the sub-mappings. A TTRIBUTE OF OBJECT FOR OBJECT (e.g. heads ‘side of coin’, four wheel drive ‘car’) is different from ATTRIBUTE OF PER- SON FOR PERSON (e.g. appendicitis ‘person suffering from appendicitis’). Within the two large groups, further sub-groups can be analyzed. A TTRIB- UTE OF OBJECT FOR OBJECT , for example, has the important sub-group of MA- TERIAL OF OBJECT FOR OBJECT instances (e.g. bronze ‘medal’, ‘work of art’, plastic ‘objects made from plastic’). And ATTRIBUTE OF PERSON FOR PERSON embraces the sub-group of COSTUME FOR WEARER metonymies (e.g. crown ‘wearer’, uniforms ‘persons wearing uniforms’) Theoretically, the reversal of the mapping ATTRIBUTE OF ENTITY FOR EN- TITY corresponds to the WHOLE FOR PRINCIPAL PART metonymies. These, however, are less numerous. A further complication concerning a potential comparison of the two arises from the fact that WHOLE FOR PRINCIPAL PART is not only connected to the ATTRIBUTE OF ENTITY FOR ENTITY group, but also to the PRINCIPAL PART FOR WHOLE group. The difference between the latter two is mainly one of necessity. The PRINCIPAL PART FOR WHOLE metonymies such as blood for ‘person’ have a vehicle and a target which stand in a necessary part-whole relation, i.e. a straightforward partonymy relation. The ATTRIBUTE OF ENTITY FOR ENTITY metonymies are slightly different, as the example appendicitis ‘patient suffering from appendicitis’ illustrates. This 206 sub-differentiation 186 makes a comparison of the mapping directions difficult, as three groups instead of two would have to be compared. Moreover, whereas the mappings at the more general level may suit counterparts in the different groups, the more specific levels are often tricky to connect to each other. M ATERIAL FOR OBJECT occurs quite often, but for the mapping OBJECT FOR MATERIAL I have only one example. The mapping COSTUME FOR WEARER was also found frequently, but this sub-mapping of ATTRIBUTE OF ENTITY FOR ENTITY , does not have a reversed counterpart, or at least none occurs in my data. Therefore, the results for the remainder of the data will be discussed in a mapping-by-mapping fashion and references to potential reversals of the mappings in question will be interspersed where relevant. The analyses in the following chapter will include different levels of specificity, i.e. they will proceed from more general mappings to more specific sub-groups. 6.2.2 Conventional and less conventional metonymies: Ontological salience and attribute salience 6.2.2.1 A TTRIBUTE OF ENTITY FOR ENTITY metonymies As already mentioned, ATTRIBUTE OF ENTITY FOR ENTITY has two important sub-groups, i.e. ATTRIBUTE OF OBJECT FOR OBJECT and ATTRIBUTE OF PERSON FOR PERSON . The ATTRIBUTE OF OBJECT FOR OBJECT group as a whole is not uniform. The word-form with the lowest number of occurrences for this mapping is heads ‘side of a coin’. This use was found only once in the BNC ([...] heads says you have kinds; ACR-162). The most frequent item is glasses, which occurs as an instance of this mapping in 98.40% of all cases. All in all, the group embraces 15 different metonymic uses of 14 lexical items: • bronze ‘medal/ work of art’, ‘cow with a bronze fur’ • crime ‘genre of crime novels/ films’ • diesel ‘car, locomotive’ • four wheel drive ‘car’ • furs ‘objects made from fur, e.g. coats’ • glasses ‘objects made from glass, e.g. glasses’ • heads ‘side of coin’ • label ‘objects carrying a label, e.g. clothes’ • oak ‘cask’ • paper ‘piece of writing’ • plastic ‘objects made from plastic, e.g. credit card’ • silver ‘objects made from silver, e.g. cutlery’ • turtleneck ‘piece of clothing with a turtleneck’ • wood ‘forest, (branch of) tree’ 186 The differentiation was taken over from Leite (1994). 207 The lexeme bronze occurs twice - as an instance of the sub-mapping MA- TERIAL FOR OBJECT (‘bronze medal, work of art’; e.g. [...] for him a bronze in the Seoul Olympics; A52-499) and as an example of ATTRIBUTE OF ANIMAL FOR ANIMAL (‘bronze cow’; e.g. [...] a cross between a bronze, a Norfolk Black and a Wrolstead [...]; C9F-381). If the ATTRIBUTE OF ENTITY FOR ENTITY group is considered as a whole, the standard deviation is with 29.08 quite high. This does not change when the two large sub-groups are considered. Let me begin with the MATERIAL FOR OBJECT metonymies. The results are summarized in Table 25: Word-form MYs total (%) M - total (%) M - MYs (%) glasses 98.40 98.40 100.00 furs 65.33 65.33 100.00 paper 49.20 49.20 100.00 wood 21.90 21.90 100.00 silver 20.80 20.80 100.00 bronze 9.10 8.90 97.80 plastic 2.50 2.50 100.00 oak 44.50 0.40 0.90 Table 25: The results for MATERIAL OF OBJECT FOR OBJECT What is striking is that this mapping mostly accounts for a very high percentage of the overall metonymic uses of the word-forms (cf. last column of Table 25). Mostly, it is responsible for 100% of all metonymies found for the lexical items. M ATERIALS can hardly be connected to metonymic targets other than OBJECTS MADE FROM THEM . Even the percentage for bronze is with 97.80% still fairly high. For glasses and paper, the MATERIAL OF OBJECT metonymy is of paramount importance since it forms the basis for diverse metonymic chains (cf. Appendix A). The only example which stands out is oak. The mapping is not significant here; it accounts for only 0.90% of all meanings found for the word. Considering the overall structure of the mapping for oak, however, this does not come as a surprise. Actually, oak is the only example I found of the reversed mapping OBJECT FOR MATERIAL OF OBJECT (‘wood of oak tree’). This mapping underlies all metonymic uses of oak and is combined into larger chains with various others. One of them is the mapping MATERIAL FOR OBJECT . The resultant metonymic chain is thus self-contradictory to a certain degree. It leads from the object (the tree) over the material (the wood) to another object (a cask). Even though the two objects involved differ since the first is a natural kind and the second an artefact, such self-contradictions within metonymic chains are not likely to foster their conventionality. When this quite exceptional case is disregarded, the picture is very clear. The MATERIAL FOR OBJECT mapping is very strong. If it applies to a lexeme at all, it tends to govern all or at least the majority of metonymic uses of the lexeme. 208 But what about the relation between the literal and the figurative meanings? Here, the results are not so unambiguous. The relative frequencies found for the lexical items under consideration differ enormously. Leaving oak aside once more, the lowest relative frequency is 2.50% for plastic (e.g. Fuels up at the boat garage, putting it on plastic; HH0-3979). The highest relative frequency, for glasses, amounts to as much as 98.40%. How does this difference arise? A very likely explanation is again 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 values in Table 25 are also fairly specified as to which products can be made from them: fur is used to make coats or other clothing items, and paper occurs most often in the form of paper sheets. In other words, the respective targets of these metonyms ought to be fairly salient parts of speakers’ mental representations of these materials and this fact accounts for the comparatively high relative frequencies of the respective metonyms. For plastic things are different. The noun is rarely used as a metonym and thus behaves the way ontological salience predicts. Products made from plastic are nearly omnipresent. As a matter of fact, there are not too many things which cannot be made from plastic. Specific objects consisting of the material are thus probably not salient in the concept PLASTIC . And this may very well be the reason why it is so difficult to use the material as a metonymic vehicle. On the whole, the MATERIAL FOR OBJECT mapping is not only a very strong pattern - by way of motivating the majority of all metonymic uses and thus not having to compete with other potential metonymic mappings - it also leads to metonymic expressions which have a considerable potential for becoming highly conventional, as long as our encyclopedic knowledge provides us with more or less unambiguous paths leading from the vehicles to the targets. This is astonishing, since all metonyms based on this mapping deviate from the laws of ontological salience. Materials are unbounded and should therefore not be used frequently as metonymic vehicles which stand for well-delineated objects, such as cutlery, spectacles, or coats. The structure of the vehicle concept, i.e. the salient attributes it embraces, can obviously be more important for the conventionalization of linguistic metonyms than the general laws of ontological salience. If it applies at all, the mapping IMPORTANT ATTRIBUTE FOR OBJECT also exhibits a significant tendency to govern many uses of a lexeme (cf. Table 26 below). But with 9.22% the median value for the share these mappings have of all meanings is considerably lower than for the mapping MATERIAL FOR OBJECT (21.35%). The only lexeme which is more often used as a metonym than one would expect considering the mean value of 15.88% and the standard deviation of 18.62 is turtleneck ‘clothing item with a turtleneck’. Here, however, two further factors have to be taken into consideration: On the one hand, the overall number of concordance lines analyzed was with only 13 very low, and, on the other, the metonymy turtleneck for ‘clothing 209 item with a turtleneck’ has a clear metaphorical component, i.e. ‘neck like that of a turtle’. The metonym relies on a metaphorical compound and it is doubtful that the compound is very often used in a truly literal way, i.e. to refer to the neck of a turtle. 187 It is noteworthy that the frequencies with which the lexical items occur as ATTRIBUTE FOR OBJECT metonymies can hardly be linked to the laws of ontological salience. Turtleneck, label, and bronze diverge from the principles by picking out POOR GESTALTS to refer to GOOD GESTALTS . But only the relative frequency of turtleneck is remarkably high. Label is less frequently used to refer to objects with a label, and the ATTRIBUTE FOR OBJECT use of bronze is extremely rare. Colours seem to be particularly poor metonymic vehicles, since they can hardly be connected to a specific metonymic target. Labels are more conventionally associated with concrete objects such as clothes, records or alcoholic beverages. Especially for the latter the noun label is often part of a product name (e.g. Johnnie Walker Red Label, HCK-280, Lanson Black Label, G2D-2910) - a fact which is responsible for the still considerably high relative frequency of label as an ATTRIBUTE FOR OBJECT metonym. Word-form MYs total (%) M - total (%) M - MYs (%) turtleneck 46.15 46.15 100.00 label 31.60 29.60 93.67 diesel 15.10 15.10 100.00 four wheel drive 3.33 3.33 100.00 crime 5.90 0.90 15.25 bronze 9.10 0.20 2.20 Table 26: The results for IMPORTANT ATTRIBUTE FOR OBJECT Crime obeys the principle CONCRETE OVER ABSTRACT , but is infrequent. And diesel and four wheel drive can be analyzed as cases where different laws conflict: Both are in line with FUNCTIONAL OVER NON - FUNCTIONAL , but diverge from POOR GESTALT OVER GOOD GESTALT and VISIBLE OVER NON - VISIBLE . While the former is with 15.10% not uncommon, the latter was found in only 3.33% of all instances. But the difference between the two latter cases is not entirely unpredictable if considered from a pragmatic point of view. What type of engine a car has, whether it is a diesel or not, is more important for using it on a day-to-day basis than whether it has a front, rear or four wheel drive. The former determines the type of fuel the car needs and many other properties of the car, such as its torque, which are relevant for 187 At least this is what intuition suggests and what the BNC confirms. Among the thirteen instances of turtleneck in the BNC there is none in which this meaning occurs. All concordance lines which were not judged as metonymic also refer to clothing items with a turtleneck. But cases like a turtleneck sweater (CDN-503) are not metonymic because turtleneck is used as the modifier of a compound and thus the PART FOR WHOLE relation is lost. 210 driving the car. But a four wheel drive is only important when driving on slippery or bumpy, potholed roads. The four wheel drive may be the salient feature that distinguishes this type of car from others, but this characteristic is only of secondary importance if compared to the type of engine, even if the latter is also hidden inside the vehicle and hence ‘invisible’. The second major class under the heading of ATTRIBUTE OF ENTITY FOR ENTITY is ATTRIBUTE OF PERSON FOR PERSON (cf. Table 27). Word-form MYs total (%) M - total (%) M - MYs (%) crown 69.10 65.00 94.07 uniforms 6.60 6.60 100.00 suits 3.34 3.34 100.00 power 1.90 1.90 100.00 appendicitis 1.64 1.64 100.00 skirt 0.90 0.90 100.00 heels 40.40 0.40 0.99 money 0.40 0.40 100.00 glasses 98.40 0.10 0.10 Table 27: The results for ATTRIBUTE OF PERSON FOR PERSON All examples in this class deviate from HUMAN OVER NON - HUMAN . The metonymic pattern ATTRIBUTE OF PERSON FOR PERSON also shows a marked tendency to govern the majority of the metonymic uses of the lexical items to which it can be applied. Exceptions are glasses and heels. What these two cases have in common is that the mapping ATTRIBUTE FOR PERSON is only the last one of a longer metonymic chain (cf. Appendix A). In both cases, the ATTRIBUTE FOR PERSON metonymy presupposes other, quite conventional mappings. For glasses, there is first the strong mapping MATERIAL FOR OBJECT (M1 total: 98.40%, M1 - MYs: 100%), second the mapping PRINCIPAL PART FOR WHOLE (M2 total: 55.10%; M2 - MYs: 56.00%), and only then can the mapping ATTRIBUTE FOR PERSON be applied (cf. example 111 above). For heels, the mapping ATTRIBUTE FOR PERSON presupposes the quite conventional mappings BODY PART FOR COSTUME (M1 total: 33.00%; M1 - MYs: 81.68%) and PRINCIPAL PART FOR WHOLE (M2 total: 13.80%; M2 - MYs: 34.16%). The existence of entrenched metonymic senses resulting from the application of the respective first and second mappings seems to affect the applicability of any further metonymic extensions. There are already wellknown mental paths which lead from the vehicle to two metonymic targets. Any further target which uses the same vehicle is potentially confusing. This assumption is in line with the fact that relatively few examples were found of metonymic chains which extend beyond a secondary metonymy. The cognitive advantage of metonymic construals is lost when the links between vehicle and target are remote and mediated by several mappings. 211 As regards the share of the ATTRIBUTE FOR PERSON meanings of all meanings, there is one outlier which is conspicuous due to its very high relative frequency. Crown occurs in 65.00% of all cases as a COSTUME FOR WEARER metonymy. All other instances are much less frequent. The mean value is 8.92% and the median, which is more suitable in this case as it is less sensitive to values which deviate very much from the others in a series, lies at only 1.64%. The standard deviation is with 21.13 quite high. But if crown is excluded from the calculations, the class of ATTRIBUTE FOR PERSON metonymies becomes quite uniform. The standard deviation is reduced to 2.17. With the exception of crown, none of the examples is a conventional way of referring to persons. The principle HUMAN OVER NON - HUMAN seems to be a very weighty one here, preventing humans from being conventionally reduced to one of their (often transitory) attributes. Metonyms violating this principle are possible, but largely context-dependent and - usually - lack any conventionalization potential. The ATTRIBUTE OF PERSON FOR PERSON metonymies can be split up further into COSTUME FOR WEARER and POSSESSED FOR POSSESSOR metonymies. Both sub-mappings are not very likely to lead to conventional metonymic expressions. And if the outlier crown is disregarded, both are quite homogeneous as far as the relative frequencies of single metonyms are concerned: The standard deviation for the COSTUME FOR WEARER metonyms lies at 2.74, and for the POSSESSED FOR POSSESSOR metonyms it is 1.06. In general, metonymies which are formed on the basis of the mapping ATTRIBUTE OF PERSON FOR PERSON do not stand a good chance of being adopted by the speech community at large. They do occur, but rarely ever gain wider currency. But what about the exception to this rule, crown? Why has this metonym made its way into the lexicon? The concept CROWN is undeniably conceptually linked with our idealized model of a ROYAL . In other words, the attribute ‘crown’ has a high cue validity with respect to the category ROYAL , i.e. the intended target ‘queen/ king’ is salient in the vehicle concept, since only queens and kings are entitled to wear crowns. Moreover, the metonym crown in the meaning ‘queen/ king’ basically has unique reference - it refers to a single person. Low-frequency metonyms like heels or skirt, however, are much less specified with regard to the metonymic targets. Anyone could wear them. And the fact that persons could, for example, possibly wear heels is certainly not a highly salient part of our conceptual representation of persons. In addition, metonyms such as heels or skirt refer to whole classes of referents, which are not well-defined or stable. The identification of a single person by way of these metonyms is therefore highly context-dependent. Crown as a metonym also works well in a neutral context due to the close associative link between royals and this symbol of their status. Attribute salience can also explain why uniforms, which is described as a novel metonym by Gerrig (1989), is actually not as unfamiliar as many other COSTUME FOR WEARER metonyms. The typical wearers of uniforms 212 form a highly salient attribute in the vehicle domain. The wearing of uniforms is highly indicative of the official status of the persons. The uniforms act as a symbol of this special status of police officers or soldiers and are therefore typically associated with them. They even act as a factor which makes the group of people wearing them seem coherent and, well, ‘uniform’ to us. Nevertheless, uniforms is a much less conventional COSTUME FOR WEARER metonym than crown, as it does not have unique reference. Particularly apt to transport metonymic ‘person’ meanings are vehicles which literally denote symbols standing for certain professions, such as crown or uniforms. The typical and stable relationship between the symbols and their users can override the laws of ontological salience and lead to conventional linguistic metonyms. Additionally, the conventionality of a COSTUME FOR WEARER metonym is higher if it has unique reference. Other vehicles, where the targets are not highly salient parts of the vehicle concepts and/ or which do not allow unique reference, are not conventionally used to refer to persons. They are unlikely to become everyday expressions whose figurative nature is no longer felt. Since the majority of the COSTUME FOR WEARER metonymies - actually the majority of all the ATTRIBUTE FOR PERSON metonymies - neither display a high target-in-vehicle salience nor unique reference, the mapping usually produces unconventional metonyms. Crown is thus the exception to the rule. This section has shown that ATTRIBUTE OF ENTITY FOR ENTITY mappings result in linguistic metonyms which differ remarkably in terms of their conventionality. The mapping MATERIAL FOR OBJECT has a considerable potential to produce conventional metonyms, but does so only if the materials used for metonymic reference are fairly specified with regard to products typically made from them, i.e. if these products are an important element of the information stored for the vehicles. The other sub-mappings of ATTRIBUTE OF ENTITY FOR ENTITY characteristically generate less conventional metonyms. Particularly if persons are the targets, a metonymic reference by way of picking out a single attribute is not common. A TTRIBUTE FOR PERSON metonyms can only become conventional if the speech community shares encyclopedic knowledge which forms the basis for an unambiguous relationship between the vehicle and target of a metonymy, i.e. if the target is a highly salient part of the vehicle concept. Target-in-vehicle salience can thus have a huge influence on the conventionalization potential of single metonyms. It can even lead to a disregard for otherwise strong preferences and therefore sanction metonyms whose relative frequencies cannot be explained by the laws of ontological salience and the speech community’s more general attitude towards a specific mapping. 6.2.2.2 P RINCIPAL PART FOR WHOLE metonymies The PRINCIPAL PART FOR WHOLE metonymies fall into two large groups: BODY PART FOR PERSON and IMPORTANT PART FOR WHOLE . The latter consists 213 of simple PRINCIPAL PART FOR WHOLE mappings (e.g. roof ‘house’) but also comprises more specific sub-mappings, such as GARMENT FOR OUTFIT (tails ‘evening outfit’, cf. example 173 below) or PLAYING PIECE FOR ENTIRE GAME (football ‘game’, cf. 174 below). All in all, 16 different word-forms of my sample are used as instances of one of the mappings. They are given in Tables 28 and 29 below. One of the words, i.e. faces, occurs in both large classes. As a PART FOR WHOLE metonymy it is mostly used for references to persons (e.g. [...] the future government not only needed new faces; A88-28), but it also occurs in the sense ‘head’ (e.g. Keep your faces down; FPK-1033). As a whole, the PRINCIPAL PART FOR WHOLE mapping does not form a very coherent group. This is mainly due to the variety found in one of its subgroups. The percentages of IMPORTANT PART FOR WHOLE meanings reach from 0.10% for face (‘head’) to as much as 96.80% for the lexeme football ( PLAYING PIECE FOR ENTIRE GAME ). The sub-group of BODY PART FOR PERSON metonymies is much more uniform as Table 28 shows. Word-form MYs total (%) M - total (%) M - MYs (%) faces 34.50 17.50 50.72 heads 8.50 4.40 51.76 hands 2.40 2.40 100.00 blood 1.70 1.70 100.00 brain 2.50 1.50 60.00 butt 1.43 1.43 100.00 Table 28: The results for BODY PART FOR PERSON With the notable exception of faces, the BODY PART FOR PERSON metonyms are quite rare. The median value of the BODY PART FOR PERSON meanings found for the lexical items given above is 2.05% only. This is in accordance with the fact that BODY PART FOR PERSON mappings in general violate the principle GOOD GESTALT OVER POOR GESTALT . The relatively high percentage for faces bears witness to the paramount importance humans ascribe to this body part. It is the body part we are normally most interested in when we interact with other people. One cannot even be certain as to whether this metonym really violates the principle GOOD GESTALT OVER POOR GESTALT . A face can very well be perceived as a coherent whole with relatively clearcut boundaries. That it forms only a part of the head and of the whole person seems less relevant due to its special status. In any case, the metonym is in line with the principle INTERACTIONAL OVER NON - INTERACTIONAL , which is likely to be responsible for the fact that it is - set against the backdrop of the low percentages found for other instances of the same mapping - quite conventional. But also some of the other instances derive some support from otherwise not irrelevant principles of ontological salience. For brain and hands one can very well claim a connection to FUNCTIONAL OVER NON - FUNCTIONAL . The different metonyms are not used at random. Each of them 214 points out one essential aspect of humans which is associated with particular properties (cf. also chapter 3.2.2.1). This becomes evident in the following examples taken from the BNC: (169) He noted that there were two contrasting myths attached to the symbol: the myth of Einstein as the Brain and the myth of Einstein as the Human Being. (FA9-229) (170) The stable hands eat lunch in the tack room. (HDB-494) Heads has a more general meaning when used to denote persons. The metonymic references are hardly ever connected to otherwise salient properties of the head such as ‘intelligence’ or ‘importance’. As a BODY PART FOR PERSON metonymy, it does not show a tendency to convey such a specific meaning. Actually, only one instance was found where the meaning could be said to be somewhat more specific than ‘person’ (cf. example 171). All other examples are more neutral as in (172), where the intended referents are members of a sports team: (171) Dana would not give up the chance to shine in front of all the heads of the fashion world. (H8J-213) (172) I am to concentrate on youth, with one or two old heads. (K4T-3590) B ODY PART FOR PERSON metonyms are in general not very frequently used in ordinary language. Most of the word-forms were only found a few times as instantiations of this mapping. That the fewest occurrences were found for butt is, due to the potentially offensive nature of the resultant metonym, not surprising. But also inoffensive BODY PART FOR PERSON metonyms such as hand or brain do not occur markedly more often. The relative infrequency of BODY PART FOR PERSON metonyms is in line with both ontological salience and target-in-vehicle salience. They are dispreferred because they do not obey GOOD GESTALT OVER POOR GESTALT by ‘reducing’ human beings to one of their body parts. That each and every human being has a head, hands, blood, etc. is undeniably part of our basic stored knowledge about human beings. But since everybody has these attributes, they are not very apt to function as metonymic vehicles. Most of these metonymic senses are therefore most likely not stored in speakers’ minds. The conceptual mapping itself may be quite common. At least it motivates a number of possible linguistic expressions. The linguistic metonyms are, however, not very widespread. The second group belonging to the PRINCIPAL PART FOR WHOLE mappings is less homogeneous, since it embraces diverse sub-kinds of metonymic construals, such as GARMENT FOR OUTFIT (cf. 173), PLAYING PIECE FOR ENTIRE GAME (cf. 174) or CAPITAL CITY FOR STATE (cf. 175, where the capital city stands for ‘ancient Rome’, i.e. for the ‘state’). In other cases, such as (176) to (178), the mapping PRINCIPAL PART FOR WHOLE is not linked to any further, more specific sub-mappings. In (176) glasses conveys the meaning ‘spectacles’, in (177) set of wheels stands for a ‘car’, and wings in (178) signifies ‘planes’. Basically the same IMPORTANT PART FOR WHOLE mapping is at 215 work in (179), even if this may be less apparent at first glance. Here, the telephone is one vital part of the ‘connection’ that is established to another person - or rather to their telephones - when calling them. The results for the whole group are summarized in Table 29 (below). (173) [...] the figure on the bandstand, wearing immaculate evening tails and waving a baton, was vaguely familiar. (CDS-131) (174) [...] turning up late and being more interested in fishing and shooting than in football. (A4P-324) (175) [...] with its cornucopia of cultural references to Troy, Greece, Rome [...] (A04-171) (176) He couldn’t really see without his glasses on. (AC3-759) (177) Otis Griffith, 21, has always been a car fanatic, but when it comes to his own set of wheels, he admits he’s never had much luck. (C8B-260) (178) [...] a 16 per cent reduction in active Army combat divisions, a 20 per cent cut in Air Force fighter wings, and an 11 per cent cut in naval vessels. (HKT-2532) (179) There were sounds of protest from the other end of the telephone. (BP7- 3077) Word-form and gloss MYs total (%) M - total (%) M - MYs (%) football ‘game’ 96.80 96.80 100.00 glasses ‘spectacles’ 98.40 55.10 56.00 set of wheels ‘car’ 50.00 50.00 100.00 heels ‘shoe’ 40.40 13.80 34.16 Rome ‘ancient Rome’ 21.10 6.90 32.70 roof ‘house, attic’ 6.90 6.90 100.00 tails ‘evening outfit’ 6.10 6.10 100.00 sail ‘sailboat’ 4.16 1.39 33.33 telephone ‘connection’ 20.50 0.90 4.39 wings ‘planes’ 0.80 0.80 100.00 faces ‘heads’ 34.50 0.10 0.29 Table 29: The results for IMPORTANT PART FOR WHOLE The examples of this group are not only heterogeneous with respect to the specific level mappings they are based on, but also with respect to whether they are in accordance with the laws of ontological salience. Telephone for ‘connection’ is in line with VISIBLE OVER NON - VISIBLE , IMMEDIATE OVER NON - IMMEDIATE , and CONCRETE OVER ABSTRACT . Football for the ‘game’ 188 obeys INTERACTIONAL OVER NON - INTERACTIONAL , FUNCTIONAL OVER NON - FUNC - 188 From a historical angle, one could argue that the invention of the game probably preceded the invention of the proper playing piece, i.e. of balls with properties tailored to the demands of the game. The underlying mapping would then be GAME FOR PLAYING PIECE . This would, however, also presuppose a metonymic understanding of the morphological constituent ball as an INSTRUMENT FOR ACTIVITY metonymy, as the literal sense of ball is undeniably the ‘round object’ meaning. Despite this possibility, I prefer the mapping PLAYING PIECE FOR GAME , since it is attested in the literature, simpler, and more convincing from a synchronic point of view. 216 TIONAL and CONCRETE OVER ABSTRACT . Glasses meaning ‘spectacles’ is also motivated by FUNCTIONAL OVER NON - FUNCTIONAL , but could be said to conflict with GOOD GESTALT OVER POOR GESTALT . The latter is clearly violated by roof ‘house’ and also by sail ‘sailboat’. But for sail one can safely assume that the violation of GOOD GESTALT OVER POOR GESTALT is remedied by the functionality of the object for the whole, just as in wings for ‘planes’. There is no clear correlation between examples which are typical in terms of ontological salience and high percentages of metonymic usage and examples which deviate from the general principles and low percentages. Football is a very conventional metonym and at the same time in line with ontological salience, but telephone for ‘connection’ is also typical in terms of ontological salience and occurs much less often. Tails deviates from GOOD GESTALT OVER POOR GESTALT , is not supported by any other principle, and is nearly as frequent as a PRINCIPAL PART FOR WHOLE metonymy as Rome, which is in line with CONCRETE OVER ABSTRACT . With the exception of football, glasses, and set of wheels, none of the other metonyms is really common. And for set of wheels the representativeness of the results is questionable, since the BNC returned only 14 hits for this expression. Similarly to the BODY PART FOR PERSON metonymies, the IMPORTANT PART FOR WHOLE mappings are often the ones which account for the biggest share of all metonymic uses of the expressions in question. Thus many of these metonymic meanings do not have to compete with other possible metonymic extensions. This holds true, for instance, for the high-frequency cases football and set of wheels. Both of them additionally profit from a fairly high target-in-vehicle salience: The game played with a football is naturally salient in the playing piece concept, and the maybe most prototypical means of transport with a set of wheels, i.e. the car, is certainly closely associated with the wheels. Glasses meaning ‘spectacles’, the other comparatively frequent case, is the result of a metonymic chain and has to compete with the ‘simpler’ MATERIAL FOR OBJECT sense ‘glass container’. Since we typically encounter glass in the form of these two objects in our everyday lives, both products are vital in the MATERIAL concept. The high-frequency instances are thus largely in line with target-in-vehicle salience. Two of the examples with the lowest relative frequencies (i.e. less than 1%), telephone and faces, are characterized by the fact that the lexemes are much more often used to express other metonymic meanings. Faces is more likely to stand either for ‘persons’ (50.72% of all metonymic meanings) or for persons’ ‘facial expressions’ or ‘mouths’ (48.99% of all metonymic meanings). And telephone refers in 80% of the metonymic uses to concrete parts of the object (‘receiver’, ‘bell’, ‘cable’). The more abstract idea ‘connection’ is less relevant. For wings, the third instance with less than 1% of IMPORTANT PART FOR WHOLE meanings, however, things are different. Here, the mapping is the only one that motivates metonymic uses. But in this case the metonym has a metaphorical component which might very well play a role in terms of the conventionalization potential of the expression. The same applies to 217 tails, where the non-metonymic meaning (‘part of evening outfit’), is also based on a similarity of shape. Due to this, tails as well as wings are probably more associated with the literal referents characterized by these parts, i.e. with animals, than with the wholes they can metonymically denote on a metaphorical basis. All in all, the IMPORTANT PART FOR WHOLE metonymies are for the most part not very conventional as far as the (socio)linguistic level is concerned. Regarding the low-frequency instances, the mapping either has to compete with other, better motivated and therefore stronger metonymic meanings of the lexical items or the metonymic mapping presupposes a metaphorical conceptualization. I MPORTANT PART FOR WHOLE metonyms are only frequent, if the target is especially salient in the vehicle. The only example which is presumably in line with target-in-vehicle salience, where the PART FOR WHOLE meaning does not compete with any other metonymic meanings, and which is nevertheless not frequent is roof. 6.2.2.3 C ENTRAL FACTOR FOR INSTITUTION / PROFESSION / ACTIVITY metonymies The next group to be commented on consists of the mappings CENTRAL FACTOR FOR INSTITUTION / PROFESSION / ACTIVITY . Considered as a whole, the group is not very consistent. The percentages of meanings motivated by this mapping reach from 0.10% for wings (‘airline’) to 92.30% for press. The standard deviation is consequently with 26.79 quite high. If the group is split up into its three sub-mappings, at least some regularities can be observed. The mapping found with the highest number of different lexemes is CENTRAL INSTRUMENT FOR INSTITUTION (cf. Table 30). Word-form MYs total (%) M - total (%) M - MYs (%) press 92.30 92.30 100.00 seat 22.30 20.70 92.83 screen 11.50 11.50 100.00 buses 9.60 8.50 88.54 crown 69.10 4.00 5.93 wings 0.80 0.10 12.50 Table 30: The results for CENTRAL INSTRUMENT FOR INSTITUTION Examples (180) to (185) illustrate the single CENTRAL INSTRUMENT FOR INSTI- TUTION uses. What is cheap in (180) are not the vehicles, but only the fares, i.e. the buses stand for the ‘service’ they offer. (181) and (184) are similar to one another, since in both cases a certain object stands for a political office, which can be understood as an institution. In (182) press stands for all institutionalized media, not only those which produce printed text. (183) is quite similar to (182); here screen stands for the whole institution involved 218 in the making of films. Finally, in (185) wings actually refers to an airline as the context makes clear. (180) The buses are cheap and excellent, the drivers are virtuosi [...] (A5X-172) (181) The Revolution settlement which formulated the basis for the offering of the British crown to William of Orange in 1689 [...] (AD2-143) (182) Do you know what I mean by relatively free press, Mr Wagner? (A06- 1062) (183) For the big screen, he has scripted The Postman Always rings Twice, Homicide [...] (B3K-883) (184) [...] Priyanka is only 20 (the minimum age for a seat in the lower house is 25) but is said to be sharp (ABG-1329) (185) Wings and wheels. The profit of British Airways plunged for the year ending in March [...] (ABG-3332) The examples are, with the exception of press, not very frequent, although some of them are in accordance with ontological salience. Buses for ‘bus service’, for example, obeys FUNCTIONAL OVER NON - FUNCTIONAL , INTERAC- TIONAL OVER NON - INTERACTIONAL , and CONCRETE OVER ABSTRACT . Crown, press, seat, and wings are, however, difficult to describe in terms of ontological salience. On the one hand, the institutions referred to consist of humans, so that they could be said to diverge from HUMAN OVER NON - HUMAN . On the other hand, the institutions could be seen as more abstract entities. Then, the examples conform to CONCRETE OVER ABSTRACT . That crown does not have a very high frequency here has to be seen in connection with the high relative frequency of references to the Queen (or other rulers, when it comes to texts which deal with earlier periods of British history) suggested by the lexeme. The two meanings compete, and the conceptual link between the symbol of a sovereign and the sovereign seems to be stronger than the link between the symbol and the larger, more abstract institution ‘royal government’. Metonyms based on the mapping CENTRAL INSTRUMENT FOR INSTITUTION are, on the whole, not so likely to become established parts of the lexicon. 189 The exception to this is press. The use of the word is no longer restricted to references to those media which literally need a printing press to multiply and publish texts, but embraces all media, including TV and radio. Yet, metonyms based on the mapping CENTRAL INSTRUMENT FOR INSTITUTION still stand better chances of being adopted by the speech community than those of the other two sub-groups. Both CENTRAL INSTRUMENT FOR PROFESSION metonyms and CENTRAL INSTRUMENT FOR ACTIVITY metonyms are rarer even though they are in line with ontological salience. P ROFESSIONS and ACTIVI- TIES are abstract entities. In principle, a reference to them by picking out a concrete and functional entity needed for the profession or activity does not seem implausible. However, metonymic construals of this kind are not 189 The reversed metonymy INSTITUTION FOR CENTRAL FACTOR is possible, but was only found once for the form schools. In [...] in his response to a question about the possibility of introducing deconstruction into high schools (A1A-577), schools refers to the ‘curriculum’. 219 very common. For both mappings, only a few instances were found. Only three of the lexical items analyzed realize CENTRAL INSTRUMENT FOR PROFES- SION (cf. Table 31). Word-form MYs total (%) M - total (%) M - MYs (%) law 21.00 10.30 49.05 football 96.80 1.10 1.14 sail 4.16 0.28 6.67 Table 31: The results for CENTRAL INSTRUMENT FOR PROFESSION For football, the mapping is secondary to PLAYING PIECE FOR ENTIRE GAME , which leads to the deeply entrenched ‘game’ meaning for the lexeme (cf. example 107, which is repeated here for convenience as 186). But this metonymic chain is seldom established. Sail is also rarely used to refer to a profession (cf. 187). The only lexical item for which the mapping is of any importance is law (cf. 188). (186) Football may be an athletic profession but it’s also a borstal for wayward genius (B1L-72) (187) [...] he leaned on the rail, sucked at his long clay pipe and reminisced about his days in sail [...] (H0A-1116) (188) He is now head chef at a leading law firm with 120 partners based in the City [...] (A0C-1331) Even though laws are quite abstract entities themselves, they still seem to be more accessible and concrete than the activities connected to the profession of a lawyer. In general, however, metonymic construals of PROFESSIONS by way of CENTRAL INSTRUMENTS are not very widespread. What certainly restricts the applicability of the pattern, is that not all professions can be linked to a single central instrument. 190 And according to my limited data, CENTRAL INSTRUMENT FOR ACTIVITY metonyms are even less familiar (cf. Table 32). 190 The relative infrequency of CENTRAL INSTRUMENT FOR PROFESSION is linked to the infrequency of CONTROLLED FOR CONTROLLER metonyms like buses, football, pen, pencil, and scalpel which can be connected to certain professions (bus drivers, football players, authors, artists, and surgeons; cf. chapter 6.2.1.3). The difference between the two mappings is that the former refer to a more abstract profession, while the latter pick out individual persons. The former are thus largely in line with ontological salience, while the latter are not. That, for example, pen and pencil do not occur as instances of the mapping CENTRAL INSTRUMENT FOR PROFESSION can be explained by the fact that neither of them enjoys an unambiguous relationship to one certain profession: Not only professional authors like novelists use pens (and today it is doubtful that they do so at all, since professional authors usually use computers instead), but everybody does who has a white collar job. The same holds true for pencil. Though some artists use pencils for their drawings, the use of pencils is not by and large restricted to this group of persons. As already explained in chapter 6.2.1.3, the situation is different, for example, for scalpel. However, this lexeme does not occur either in the sense ’profession of a surgeon’. 220 Word-form MYs total (%) M - total (%) M - MYs (%) sail 4.16 2.49 60.00 telephone 20.50 0.90 4.39 Table 32: The results for CENTRAL INSTRUMENT FOR ACTIVITY Note that the instances of the CENTRAL INSTRUMENT FOR ACTIVITY metonyms considered here are not verbal uses of sail and telephone. Only nominal uses were analyzed in the corpus study. Good examples to illustrate these nominal references to activities are [...] during the cheaper telephone period (CD0-219) and [...] it could turn out a wet sail (HTS-901). Since both lexical items can be used as verbs as well, a verbal construal is probably preferred and therefore metonymic references to activities by way of nouns are not abundant. All in all, the CENTRAL INSTRUMENT FOR INSTITUTION / PROFESSION / ACTIV - ITY mappings are not very productive. The mappings are ready for use, but the speech community’s demand for actual linguistic instantiations is limited. A notable exception is the INSTRUMENT FOR INSTITUTION metonym press, where the metonymic meaning is much more entrenched than the literal one. This is undeniably linked to pragmatic issues: The metonymic shorthand press is a very convenient means for transporting quite an extensive meaning, i.e. a reference to all different media. 6.2.2.4 L OCATION FOR ENTITY metonymies L OCATIONS can serve as metonymic vehicles for diverse kinds of targets. My classification distinguishes two large groups: LOCATION FOR ENTITY THERE and LOCATION FOR PEOPLE THERE . Both can be further sub-categorized as Figure 18 demonstrates. LOCATION FOR ... ENTITY THERE PEOPLE THERE EVENT INSTITUTION RULERS POPULATION MILITARY FORCES SPORTS TEAM Figure 18: Types of metonymies with LOCATIONS as vehicles The LOCATION FOR ENTITY THERE group is too heterogeneous to consider it as a coherent whole. I will therefore begin with the LOCATION FOR EVENT mapping. This mapping is often illustrated in the relevant literature with vehicles like Vietnam or Watergate. But as Table 33 shows, it is not restricted to locations which are associated with outstanding historical and/ or political events, although such important events lead to the highest relative frequencies. 221 Word-form MYs total (%) M - total (%) M - MYs (%) Watergate 84.54 84.54 100.00 Pearl Harbo(u)r 31.03 31.03 100.00 Vietnam 39.00 20.20 51.80 turf 9.98 9.98 100.00 church 59.90 4.00 6.68 stadium 5.45 1.26 23.08 Paris 4.40 0.90 20.45 Rome 21.10 0.50 2.37 room 3.00 0.30 10.00 America 63.50 0.10 0.16 Table 33: The results for LOCATION FOR EVENT All metonymic uses of Pearl Harbo(u)r and Watergate are concerned with the events that happened there. But this also holds for the common noun turf (e.g. [...] the Prince of Wales and Lord Rosebery for frequenting the turf [...]; AE6- 66). The mental connection between Vietnam and the Vietnam War is also quite robust. 51.80% of all metonymic uses of the proper noun convey the ‘war’ meaning. All the other lexemes are less exclusively associated with an event. This is also reflected in how often the single lexemes occur as instances of the LOCATION FOR EVENT mapping. The highest relative frequencies were found for those lexemes which are fairly unambiguous as to which event ought to be associated with them. A strong conceptual link between a location and a definite event clearly favours the conventionalization of LOCATION FOR EVENT metonyms. So again, the conventionality of the single metonyms hinges upon the specificity of the relation between vehicle and target. One can safely assume that the average speaker’s knowledge about the literal references of Watergate and Pearl Harbo(u)r, i.e. about the places, is extremely restricted. What we know about these places is largely limited to what happened there. The ‘event’ targets are salient in the respective vehicle concepts. This is different for other locations like Paris or America. We know much more about the locations themselves and probably also about diverse things that happened there. This explains why quite distinct relative frequencies were observed for the single lexemes, although all of the metonyms are equally ‘good’ in terms of ontological salience. All of them conform to CONCRETE OVER ABSTRACT and GOOD GESTALT OVER POOR GESTALT . A good example to illustrate the diversity of events which can - given the right amount of contextual support - be the intended targets of LO- CATION FOR EVENT metonyms is Paris. The nine examples of the mapping found in my sample from the BNC refer to three different events: one to a ballet (J1A-303), three to a sports event (e.g. K5A-2352), and five to a car fair (e.g. A6W-265). Sports events are, after historical and political events, most likely to be the targets of LOCATION FOR EVENT metonymies. They play a big role for turf, but also for stadium and proper nouns such as Paris, 222 Rome, and America. But whereas the relationship between a sports event and the lexemes turf and stadium is quite a typical one, 191 it is much less obvious for Rome, Paris or America. The only other location which is relatively unambiguously associated with one event is church. The ‘service’ meaning (e.g. Then, that Sunday, I went to church; FRD-110) is with only 4.00% of all meanings not really frequent, but compared to the ‘event’ meanings of other locations such as America, room 192 or Rome, it is more common. The prime example of a totally conventional LOCATION FOR EVENT mapping is Watergate. The literal meaning of the lexeme does not occur often. Other location names which literally refer to countries or cities are also used for metonymic references to people or institutions (cf. below). The metonymic event meaning of Watergate does not, however, have to compete with other potential metonymic meanings. The same holds for Pearl Harbo(u)r, which is the second most frequent word to convey an ‘event’ meaning in my sample. From a pragmatic angle, the only reason for metonymic uses of the nouns is to refer to these specific and salient events. The reversed mapping, EVENT FOR LOCATION , occurred with only two word-forms, i.e. opera and schools. Opera is used in 6.70% of all cases as an EVENT FOR LOCATION metonym (cf. example 152 above) and schools in 6.10% of all cases (e.g. [...] kids shooting kids in the schools [...]; AHX-809). Considering that the mapping stands in opposition to what ontological salience predicts, these frequencies are quite high. But regrettably my data are too limited to allow generalizations with respect to this pattern. The mapping LOCATION FOR INSTITUTION was found for 13 word-forms. Three of them, i.e. China, Paris, and Rome, occur as instances of two different sub-mappings, so that the overall number of examples is increased to 16. An overview is given in Table 34 (below). Again, the group as a whole is not homogeneous. The figures for the share of LOCATION FOR INSTITUTION meanings of all meanings reach from 0.50% for room to as much as 96.99% for Pentagon. This is reflected in the high standard deviation, which lies at 32.95 for the whole class. The most deeply entrenched metonymic ‘institution’ meanings are found with proper nouns. Apart from Wall Street, the most conventional LOCATION FOR INSTITUTION metonyms are examples motivated by the submapping BUILDING FOR INSTITUTION . Convent has with 12.60% the lowest share of metonymic BUILDING FOR INSTITUTION meanings (cf. 189), followed 191 Another example of an extremely specified relationship between a LOCATION and a certain SPORTS EVENT is Wimbledon. This proper noun was, however, not part of my corpus study so that nothing can be said here about the relative frequency with which it conveys such an ‘event’ meaning. 192 The LOCATION FOR EVENT mapping for room is actually motivated by the more specific LOCATION FOR ACTIVITY mapping. An example is [...] and I’ve done bar, I’ve done diner’s room, I’ve done kitchen, I’ve done (KCP-4670). It occurs only with compounds or compound-like structures, such as diner’s room, music room or search room. 223 by church with 55.90% (cf. 190). White House (72.10%; cf. 191) and Pentagon (80.13%; cf. 192) are even more frequently used as instantiations of this mapping. Word-form Mapping: LOCATION FOR ... M - total (%) M - MYs (%) Wall Street INSTITUTION 96.99 100.00 Pentagon INSTITUTION 80.13 100.00 White House INSTITUTION 72.10 100.00 church INSTITUTION 55.90 93.32 Washington GOVERNMENT 26.80 98.17 Moscow GOVERNMENT 24.20 98.37 convent INSTITUTION 12.60 100.00 Rome INSTITUTION ‘church’ 9.00 42.65 Rome GOVERNMENT 3.80 18.01 Paris GOVERNMENT 1.60 36.36 Bordeaux INSTITUTION ‘university, factory’ 1.34 7.70 Paris INSTITUTION ‘centre of fashion, stock market’ 1.00 22.73 China/ china INSTITUTION ‘economy’ 0.90 2.15 stadium INSTITUTION ‘sports club’ 0.84 15.38 China/ china STATE FORM 0.70 1.67 room INSTITUTION 0.50 16.67 Table 34: The results for LOCATION FOR INSTITUTION and its sub-mappings (189) For payments to the convent of wine for various feasts of the year. (CB6- 358) (190) Our young people deserve honesty from ourselves and have a right to it from the Church. (C8H-429) (191) [...] so not everybody at the White House knew what was going on. (ADL-705) (192) It fell to Pentagon officials to publicise this crucial information. (A8J-385) B UILDING FOR INSTITUTION belongs to those mappings which have a considerable potential to become very conventional ways of expression. Especially, if the vehicle has a unique reference, such as White House or Pentagon, they stand a very good chance of becoming established parts of the lexicon. The deep entrenchment of the relationship between the respective buildings and a single specific institution located there is also reflected in the high figures in the last column of Table 34. The sub-mapping CAPITAL CITY FOR GOVERNMENT was, evidently, found for Moscow, Paris, Rome, and Washington. The relative frequencies of the single lexemes as instances of this mapping echo the relevance of the countries with respect to world politics. While the lexemes Moscow and Washington are used in roughly a quarter of the cases as references to the respective governments, the figures are much lower for Paris and Rome. This fact illustrates once more how much the use of metonyms depends on extralinguistic reality and pragmatic considerations. Political decisions 224 made in Rome or Paris usually do not have as many consequences as decisions made in Washington or Moscow. The demand for metonymic shorthand references to the governments of Italy or France is, at least in publications in English, not so great. For Rome, the situation is somewhat more complex since the ‘government’ meaning has to compete with another LOCATION FOR INSTITUTION metonymy: the metonymic reference to the Roman Catholic Church. On the whole, the difference between the high-frequency and the lowfrequency examples in Table 34 is a matter of attribute salience. The BUILD- ING FOR INSTITUTION mappings Wall Street, Pentagon, and White House are so conventional because their unique reference fosters the high salience of the respective institutions in the different location concepts. For other location names which are comparatively frequently used as metonyms, like Moscow and Washington, one can also safely assume that their mental representations are characterized by one especially prominent piece of information, i.e. that they are the capital cities of Russia and the US respectively. Room (e.g. In the same year I was elected president of the Junior Common Room [...]; CDC-131), however, is entirely unspecific with regard to what kind of institution could possibly be located there. The same applies to Bordeaux (e.g. [...] Dr Ba Maw, graduate of Cambridge and Bordeaux [...]; FAN-22). The few metonymic references by way of the proper noun to institutions are to either a university or a factory located there, i.e. institutions which are found at many other places as well. None of them is an outstanding part of the location concept, so that the intended targets are not salient in the vehicle. The mapping as a way of thinking is a well-entrenched and also largely conventional one. This is evident in the many different examples based on it. It is, so to speak, ready for use in speakers’ minds, but how conventional the linguistic expressions are is dependent on speakers’ encyclopedic knowledge about places and what is located there. If the association is strong and specific, the metonym can be used in lean contexts. If it is weak, strong contextual support is needed. The ‘stock market’ meaning of Paris, for example, is not difficult to arrive at, if the noun occurs in an article with the heading European Business: French plan for share buying without tears and is surrounded by other meaning clues in the preceding context as in (193), since the corresponding concepts have already been activated. But presented in isolation, the lexeme would not immediately call up the meaning intended in (193). (193) European Business: French plan for share buying without tears. [...] The plans are a variation on an earlier form of PEPs which restricted investors to bonds and insurance plans. A stream of money flowing into the country’s stock market from both individual investors - the tax perks are not available to foreign buyers - and reforms also in the pipeline to the French pension industry are prompting hopes of a surge of the Paris 225 bourse. Paris is below its historic high [...] (AKD-551-557; emphasis added; SH) The second major group of mappings which use locations as vehicles, i.e. LOCATION FOR PEOPLE , was also found for many linguistic items. All in all 13 word-forms returned examples of this mapping. From a purely conceptual perspective, the mapping is - as a method of metonymic reasoning - not uncommon. The question is, however, how frequently different place names are used to refer to people in contrast to how often they transport other meanings. Most of the examples are proper names, such as America, Germany or Paris, but the class embraces common nouns as well (e.g. kitchen ‘kitchen staff’, stadium ‘audience’). The group as a whole is fairly uniform with only some outliers. The results will be presented separately for the different submappings referring to rulers, ordinary people, military forces, and sports teams. But I would like to point out in advance that, as already noted in chapter 6.1, the LOCATION FOR PEOPLE metonyms are not very frequent. If all 28 different mappings found for the 14 word-forms which were found to convey a metonymic ‘people’ meaning are taken as a basis for the calculation, the LOCATION FOR PEOPLE metonyms account for 3.14% of all meanings only. The median value calculated on the same basis is even more telling: It lies at only 0.85% and indicates distinctly that the conceptual link between LOCATIONS and diverse groups of PEOPLE is not a particularly strong one. The first sub-mapping to be discussed here is LOCATION FOR RULERS . The results are summarized in Table 35. Word-form MYs total (%) M - total (%) M - MYs (%) China 41.90 26.00 62.05 Vietnam 39.00 18.30 46.92 Germany 20.40 12.00 58.82 America 63.50 3.30 5.20 Bordeaux 17.39 1.00 5.77 America 63.50 0.30 0.47 Table 35: The results for LOCATION FOR RULERS America occurs twice here, since the LOCATION FOR RULER meaning can either be a simple metonym, if America is premodified as in North America, or it can be the second mapping of a double metonym, if America is also a WHOLE FOR PART metonymy. The sub-group LOCATION FOR RULER is quite heterogeneous: While the relative frequencies for China and Vietnam are not very low and Germany, too, is used in more than 10% of all cases to refer to its rulers, America as well as Bordeaux both occur infrequently as realizations of the mapping LOCATION FOR RULER . Considering that Bordeaux is not a country name, one could argue that metonymic references to rulers other 226 than those of whole countries are perhaps generally dispreferred. However, this does not explain why America occurs so seldom in this sense. The percentages of metonymic references to the rulers are particularly interesting when compared to the relative frequencies of LOCATION FOR ORDINARY PEOPLE metonymies, which are given in Table 36. The mean value for the LOCATION FOR ORDINARY PEOPLE meanings is only 1.79%. With a standard deviation of 1.49 the group is very homogeneous. Those lexemes which are often used to refer to rulers, such as China and Vietnam, rarely occur in the ‘ordinary people’ sense. The highest value in Table 36 is 5.50% for Germany. This, however, could very well be an effect of what Fischer (1998: 174-175) calls topicality (cf. Fn. 50 above). Many texts in the BNC in which this meaning is realized deal with the reunification of Germany, where the German population - especially the former Eastern German one - plays a large role. All in all, classic instances of location names like the names of countries or cities are not very likely to be found as metonyms referring to the population. Interestingly, common nouns are often more apt to transport this metonymic meaning. Compare examples (194) and (195) below. Word-form MYs total (%) M - total (%) M - MYs (%) Germany 20.40 5.50 26.96 stadium 5.45 3.24 59.62 Bordeaux 17.39 3.01 17.31 street 2.40 2.30 95.83 America 63.50 1.70 2.68 room 3.00 1.20 40.00 China 41.90 0.90 2.15 kitchen 9.30 0.80 8.60 Paris 4.40 0.80 18.18 Rome 21.10 0.80 3.79 America 63.50 0.70 1.10 Vietnam 39.00 0.50 1.28 Table 36: The results for LOCATION FOR ORDINARY PEOPLE (194) I could see Myrtle Augee, Mary Berkeley and Tessa Sanderson, all waving and shouting, seeming to make more noise than the rest of the stadium. (BMM-133) (195) Back in Moby Dick Terrace, the street’s only telephone began to ring. (HWN-1287) The common nouns room, stadium and street also have the highest values in the last column of Table 36, which indicates that metonymic references to ordinary people are among the most likely metonymic uses of the lexemes. For names of countries and cities, many different potential metonymic senses compete with each other. They can, for example, realize the mapping LOCATION FOR PRODUCT MADE THERE (e.g. Bordeaux, china) or can be 227 used to refer to INSTITUTIONS (e.g. Paris, Rome). Thus, not only references to different groups of people have to compete with one another in these cases, but the references to these diverse groups of people also have to compete with other metonymic patterns, which are often much stronger as they determine a larger part of the total metonymic meanings of the lexemes in question. Table 37 summarizes the results for the two sub-mappings of LOCATION FOR PEOPLE left, i.e. LOCATION FOR MILITARY FORCES and LOCATION FOR SPORTS TEAM . As the table indicates, neither are frequent. The occurrence of instances of the latter is undeniably text-type dependent. Considering, however, that the BNC records many newspaper texts and that sports events are among the stock subject matter of newspaper journalism, the very low relative frequencies are somewhat unexpected. Mapping Word-form MYs total (%) M - total (%) M - MYs (%) Germany 20.40 0.90 4.41 America 63.50 0.80 1.26 LOCATION FOR MILITARY FORCES : China 41.90 0.20 0.48 Germany 20.40 2.00 9.80 Washington 27.30 0.50 1.83 Moscow 24.60 0.40 1.63 Bordeaux 17.39 0.33 1.92 China 41.90 0.30 0.72 Paris 4.40 0.10 2.27 LOCATION FOR SPORTS TEAM : Rome 21.20 0.10 0.47 Table 37: The results for LOCATION FOR MILITARY FORCES and LOCATION FOR SPORTS TEAM The overall low frequency of LOCATION FOR PEOPLE metonyms is in general accordance with the laws of ontological salience. Humans can hardly be referred to by way of their locations. If speakers use it at all, they prefer the mapping for conveying the ‘ruler’ meaning. Rulers are at least more or less clearly associated with places. The link to ordinary people is weaker. That countries or cities have a population is undoubtedly stored in the respective concepts, but this knowledge is too trivial to sanction highly conventional metonymic ‘ordinary people’ uses of the location names. Institutions which are mentally connected with a special building are much easier to access via metonymic references. Metonyms like White House or Pentagon are cognitively profitable as the buildings are concrete entities which have a unique relationship to the institutions referred to. Other locations are often too unspecific to allow unambiguous references to institutions or people. And if such an unspecific relationship additionally violates the principle HUMAN OVER NON - HUMAN , the conventionalization potential of the resultant metonyms is low. 228 L OCATION FOR EVENT metonyms behave differently since they are in accordance with CONCRETE OVER ABSTRACT and GOOD GESTALT OVER POOR GES- TALT . Nevertheless, not all LOCATION FOR EVENT metonyms stand a good chance of becoming conventional. The conventionality of LOCATION FOR EVENT metonyms can only be explained when attribute salience is taken into account. If a location is not associated with a single event, i.e. if no specific event is salient in the vehicle concept, low relative frequencies are the consequence. 6.2.2.5 I NSTITUTION FOR PEOPLE metonymies Usually, the intended meaning of INSTITUTION FOR PEOPLE metonyms is ‘people responsible’, but institutions like schools and universities can also stand for their sports teams. The latter meaning is, however, infrequent. It occurred in 0.40% of all instances of schools (cf. 196) and in 0.20% of all instances of university (cf. 197). (196) Scotland 0, England 28 SCOTTISH Schools could not match the skill and strength of their English counterparts [...] (AK6-477) (197) Durham University are champions of divisions two and four, and Newcastle University have won division five (K52-4356) More interesting, since often much more conventional, are the INSTITUTION FOR PEOPLE RESPONSIBLE metonyms. The most common metonymic references to PEOPLE RESPONSIBLE are made via POLITICAL INSTITUTIONS as Table 38 demonstrates. Word-form MYs total (%) M total (%) M - MYs (%) government 56.20 56.20 100.00 Pentagon 80.13 49.16 61.34 Senate 36.30 36.30 100.00 Exxon 35.86 35.17 98.08 schools 31.90 25.30 79.31 airlines 30.79 20.41 66.31 Renault 56.61 12.87 22.74 church 59.90 10.30 17.20 university 8.30 7.90 95.18 BMW 70.21 7.54 10.74 opera 16.40 5.10 31.10 garage 4.60 4.60 100.00 label 31.60 4.10 12.97 seat 22.30 0.10 0.45 Table 38: The results for INSTITUTION FOR PEOPLE RESPONSIBLE The mapping does not produce a homogeneous picture in terms of the relative frequencies with which the single words transport the meaning ‘people responsible’. The standard deviation is, accordingly, with 18.05 229 quite considerable. The highest figures were found for government, Pentagon, and Senate. Political institutions are thus, not unexpectedly, the prime examples of conventional INSTITUTION FOR PEOPLE RESPONSIBLE metonyms. For other types of institutions no clear pattern emerges. Large international companies such as Exxon, Renault, and BMW differ remarkably with respect to the use of the company name for references to the people in charge. The same is true for university and schools. The lowest relative frequencies were mostly found for secondary mappings in larger metonymic chains (e.g. opera, label, seat). For opera, for example, the mapping is secondary to ABSTRACT PRODUCT FOR INSTITUTION (cf. 198) and for label it is even the third mapping in a chain leading from AT- TRIBUTE FOR OBJECT over PRODUCT FOR INSTITUTION to INSTITUTION FOR PEOPLE RESPONSIBLE (cf. 199). But that the mapping occurs as the second or third in a metonymic chain is not a safe indicator of the unconventionality of an INSTITUTION FOR PEOPLE RESPONSIBLE metonym, since for Pentagon, which is highly frequent in this meaning, the mapping is also the second one in a metonymic chain leading from BUILDING FOR INSTITUTION to INSTITUTION FOR PEOPLE RESPONSIBLE (cf. 200). (198) Opera North has just unveiled a new Britten staging [...] (A51-20) (199) A small company will probably focus its attention more on each artist’s project more than a major label. (A6A-1389) (200) Sixteen hours after the first assault, the Pentagon said the invasion force had ended ‘all organised resistance’ [...] (AAC-458) In terms of ontological salience, all examples feature a conflict of different principles. On the one hand, they all diverge from the weighty HUMAN OVER NON - HUMAN PRINCIPLE and from CONCRETE OVER ABSTRACT . On the other hand, the decision-makers often form an anonymous group of people. As a consequence the INSTITUTION as a whole forms a better gestalt than the humans behind them. Normally, the persons involved are not regarded as individuals, but mainly in their function as representatives of institutions, whose key concerns are determined by the aims of the institution they represent. In short, it is the fact that they belong to or constitute the institution that makes us consider RESPONSIBLE PEOPLE as a coherent group. Therefore, it is not necessarily uncommon to construe them metonymically by way of this unifying aspect, as the highly frequent examples show. 6.2.2.6 W HOLE FOR PRINCIPAL PART metonymies The last large group to be discussed consists of WHOLE FOR PRINCIPAL PART mappings. Even though they are typically in line with the principle GOOD GESTALT OVER POOR GESTALT , many of the lexical items in my study are seldom used as WHOLE FOR PART metonyms, like bottle for ‘label on the bottle’ (cf. example 93 above), packet also for ‘label’ (cf. 201) or kettle for ‘plug, fuse, etc.’ (cf. 202). 230 (201) They think ah look there’s a bit of drop of water in, there’s a packet of something let’s stick it in and stir it up, they never read a packet to see what it could do. (F8C-148) (202) An electric kettle, plugged into the wall [...] (FRC-345) (203) The prohibitionist thinking which affected America in the 1920s [...] (A0B-339) (204) Made and aged in Vosges oak (¼ new); vividly yellow-gold, with spicy oak, good acidity, mouthfeel and pungency [...] (AHK-1453) Other examples such as America ‘US’ (cf. 203) or the OBJECT FOR MATERIAL metonymy oak are, however, quite frequent. The results are given in Table 39. Oak occurs twice here. Once in the OBJECT FOR MATERIAL OF OBJECT meaning already mentioned (cf. chapter 6.2.2.1) and once as a secondary WHOLE FOR PART mapping, which is based on the ‘material’ meaning. This metonymic chain is quite rare and accounts for meanings such as ‘colour’ and ‘taste’ of oak (cf. 204). Word-form MYs total (%) M total (%) M - MYs (%) America 63.50 62.50 98.43 oak 44.50 44.50 100.00 pencil 26.40 26.10 98.86 faces 34.50 16.90 48.99 telephone 20.50 16.40 80.00 bathroom 7.20 7.20 100.00 kitchen 9.30 4.60 49.46 car 5.60 4.10 73.22 heads 8.50 4.00 47.06 windmill 1.63 1.63 100.00 kettle 22.05 1.42 6.42 oak 44.50 0.80 1.80 shoes 0.80 0.80 100.00 packet 8.10 0.70 8.64 bottle 10.00 0.10 1.00 street 2.40 0.10 4.17 Table 39: The results for WHOLE FOR PRINCIPAL PART Many of the words in Table 39 realize more than one concrete metonymic meaning relying on the WHOLE FOR PART mapping. This is in line with Langacker’s (e.g. 1987, 1993, 2000) idea of active zone phenomena (cf. chapter 3.2.2.2). Quite frequently, we pick out the wholes because these are the “entities that most concern us” (Langacker 1993: 32), if we want to refer to one of their parts. As long as there is a direct relationship between the wholes and these respective parts, active zone reasoning is not problematic for the language user. This accounts for the huge flexibility of the metonymic mapping scope (Ungerer and Schmid 2006: 129-131). Good examples to illustrate this are bathroom and car. The single WHOLE FOR PART meanings 231 are given in Table 40 together with their respective relative frequencies of occurrence. Word-form Meaning gloss Relative frequency of meaning (%) Relative frequency of WHOLE FOR PART mapping (%) ‘fittings/ fitments’ 3.50 ‘WC’ 3.30 ‘tub’ 0.20 ‘walls’ 0.10 bathroom ‘floor’ 0.10 7.20 ‘motor’ 1.80 ‘exterior’ 1.60 ‘interior’ 0.40 ‘tank’ 0.10 ‘number plate’ 0.10 car ‘tyres’ 0.10 4.10 Table 40: The diversity of WHOLE FOR PART meanings of bathroom and car Considering the number of different senses of bathroom and car that can be conveyed by WHOLE FOR PART mappings, makes it easier to explain why many lexemes are not often used as instantiations of this mapping. The entities as wholes form coherent gestalts and are therefore theoretically apt to function as metonymic vehicles. The problem lies in the fact that there are so many parts that could be metonymically referred to - and none of them stands out among the others as particularly important. Which part a speaker is actually referring to by way of these metonyms only becomes clear in a strong context. If a car breaks down (F9W-675) it is its engine, if it had a puncture (K1T-388) it is the tyres, if it is being filled up (KDK-278) it is its tank, etc. But all these parts are equally relevant for the car as a functional whole. This is different for pencil, which is much more often used as a WHOLE FOR PART metonym. A pencil only has one important functional part, i.e. the lead. It is thus natural to refer to this sub-part by way of the whole. The conceptual path from the whole to the part is much more straightforward here, and the metonym is not uncommon since the lead is largely hidden within the outer shell of the pencil, i.e. it is supported by the principle VISI- BLE OVER NON - VISIBLE . Other words which often instantiate a WHOLE FOR PRINCIPAL PART meaning are oak and America. In these two cases, too, the conceptual link between the wholes and one specific part is fairly clear-cut: An oak does not have too many parts that could be referred to by mentioning the whole, the material is thus an apparent candidate for such a metonymic construal. And for America, it is also our encyclopedic knowledge that tells us that the only logical interpretation is ‘US’, since the United States form the most important part of the continent. How conventionally 232 active zone reasoning is applied depends on - as Langacker (1987: 272-273) suggests - how direct the relationship between the active zone and the whole domain is. The more direct, unambiguous and straightforward the relationship, the more salient is the target in the vehicle, and the more conventional the resultant linguistic metonym. The main problem with many of the WHOLE FOR PART metonyms is that they are vague rather than ambiguous. This is also obvious (or rather nonobvious) on the linguistic surface. Many other metonyms lead to directly observable deviances in terms of truth-conditionality. When somebody reads Shakespeare or when the buses are said to be on strike, the vehicles and targets involved differ with respect to HUMAN vs. NON - HUMAN . The different parts of a car, however, just like the different parts of a pencil all belong to the realm of things. When somebody says My pencil broke, the utterance could very well be literally true. It is just a matter of experience that it is usually the lead which breaks and not the whole pencil. For typical WHOLE FOR PRINCIPAL PART metonymies the conceptual distance between vehicle and target is not a large one, since the parts referred to often form necessary and/ or functional aspects of the wholes. A car without a motor is not functional anymore, a pencil without lead is virtually useless, just like a telephone without a receiver or a bathroom without any of the typical fitments. That the parts are felt to belong together, as the wholes cannot exist independently from their parts, is the reason why the metonymies are not very obviously figurative despite their relatively low overall frequency. Compare the following examples from the BNC: (205) The car was searched and officers found drugs [...] (K35-1856) (206) I’ve just hoovered the bathroom and the bedroom (KBN-1427) In (205) it is the interior of the car which was searched and in (206) it is the floor which was hoovered. But the metonymic nature of these utterances is, as is common for active zone phenomena, very subtle. Exactly this unobtrusive character of the examples makes it likely that they are actually more frequent than my study reveals. Only cases where the linguistic context more or less plainly indicated that the concordance line in question is metonymic could be counted. Underspecified examples, which could not be assigned to a given mapping, could not be taken into consideration (cf. chapter 5.2.1.1). As a result it cannot be excluded that such vague references are more common than can be established by way of a corpus study. But what can be said on the basis of my results is that the WHOLE FOR PRINCIPAL PART mapping is not an unconventional conceptual reasoning pattern. Otherwise so many lexemes would not have been found in WHOLE FOR PART uses. And metonymic uses occur more often if the wholes have one particularly vital sub-part. This holds for America and pencil, but also for faces, which is mostly used as a WHOLE FOR PART metonymy to refer to the ‘facial expression’ (16.50% of 16.90%; e.g. with pitiful faces; ACA-1033), 233 and for telephone, which in the majority of the cases stands for the ‘receiver’ (10.50% of 16.40%; e.g. he picked up the telephone; AP0-819). 6.2.2.7 Further metonymic mappings In this chapter, I would like to comment briefly on various other mappings which also occurred in my study, but for which my data are not comprehensive enough to allow a more detailed discussion. Most of them are comparatively infrequent. Within the CONTAINMENT ICM the mapping CONTENT FOR AMOUNT OF CONTENT was found for book and kettle. For both lexemes, this mapping is the second of a metonymic chain which is based on a CONTAINER FOR CONTENT mapping. But clear instances of the mapping CONTENT FOR AMOUNT OF CONTENT are rare. For book it only accounts for 0.20% of all uses (cf. 207) and for kettle for 1.18% (cf. 208). (207) The landscape of towns, indeed, requires a whole book to itself. (FAG- 1946) (208) Odd-Knut collects a kettle of snow […]. (A6T-2230) Another rather unconventional group has TIME as its target. P ERSONS can stand for their TIME (e.g. since Shakespeare; HRL-394), but also EVENTS can stand for the TIME when the event took place (e.g. during Vietnam; CFG-15). In the latter case the mapping is necessarily the second in a metonymic chain: For Vietnam it is secondary to LOCATION FOR EVENT . For Einstein, the other example for which this mapping occurred (e.g. [...] but since Einstein, we have to accept that time goes at different rates for different observers; FYX- 1145), it is based on THEORIST FOR THEORY . Both the PERSON FOR TIME and the EVENT FOR TIME mapping are not very common. They never account for more than 0.30% of all uses of the respective lexical items. Other infrequent mappings include LOCATION FOR USE OF LOCATION , which is instantiated by the words room and seat (cf. 209), ACTIVITY FOR DURATION OF ACTIVITY , which occurred with university (cf. 210), INSTITUTION FOR STOCK MARKET SHARES / INDEX , which was found for airlines and label (cf. 211), and OBJECT FOR LOCATION OF OBJECT , where the kettle stands for the cooker (cf. 212). (209) PARTY BOOKINGS: 1 FREE seat for every twenty (and multiples thereof) booked and PAID FOR at FULL PRICE. (BPC-1968) (210) After university he became National Officer of the British Mountaineering Council [...] (EW9-1499) (211) Delta Airlines [...] also fell about a dollar. (A4F-546) (212) [...] and switched the kettle on before rummaging through the drawers. (EF1-1774) These examples bear witness to the immense flexibility of word meaning in general. However, the conceptual paths between the vehicles and targets of these mappings are far less entrenched than for mappings which are realized by a greater number of lexical items and which can lead to highly 234 frequent metonymic uses. The mapping INSTITUTION FOR INDEX may be an exception here. It is undoubtedly very common in certain registers, and often exists in newspaper articles on economy, for example, but is not very conventional in ordinary speech. The only group of metonyms not mentioned yet which is not infrequent is SOURCE FOR PRODUCT . The sub-mapping ANIMAL FOR MEAT OF ANIMAL is highly conventional for chicken. 68.70% of all instances analyzed convey this meaning. For chicken as well as for turkey, duck, goose, lamb, etc. this metonymic path works well. Nevertheless, the applicability of the mapping is restricted, since English often has special lexemes for the meat as opposed to the animal, like pork - pig, beef - cow, and veal - calf. Utterances where the ‘animal’ denotation is used to refer to the ‘meat’ would certainly be marked if not unacceptable, unless in a humorous context. The other three instances motivated by SOURCE FOR PRODUCT are Bordeaux, china, and kitchen. They can all instantiate the mapping LOCATION FOR PRODUCT MADE THERE . Although they deviate from the principle IMMEDIATE OVER NON - IMMEDIATE , they are not entirely uncommon. Bordeaux refers to ‘wine’ or ‘mustard’ in 11.71% of all cases, china to ‘porcelain’ in 12.90%, and kitchen to ‘cuisine’ in 3.90%. It seems that (originally) proper nouns are more apt to transport the PRODUCT meaning than common nouns. This suggests that it is, once again, the typicality of the conceptual link that determines how conventional a linguistic metonym is or can become. That a certain type of wine is produced in Bordeaux is certainly a salient part of our concept of the location. For the common noun kitchen the link to a special type of ‘cuisine’ is less entrenched. But is ‘porcelain’ really an extremely salient aspect of our concept of the country C HINA ? Probably not. Or rather, no longer. But at the time the first Chinese porcelain was imported to Europe, it probably was - otherwise the name of the country would not have become a common synonym for porcelain in English. China meaning ‘porcelain’ should therefore be considered a lexicalized metonym. 6.2.3 Types of salience and the conventionality of metonymic mappings and meanings: An intermediate summary The conventionality of metonymies is a complex issue. Whenever a tendency can be identified, there are also exceptions to the rule. Still, there are mappings which are indeed much less conventional and therefore result in more obviously figurative metonymic expressions than others. In chapter 6.2.1.2, it was shown that the highest level of analysis - the level of the three large groups of mappings WHOLE FOR PART , PART FOR WHOLE , and PART FOR PART - is difficult to work with in an empirical study. It is much easier to look for rules and regularities concerning reversible mappings. As Radden and Kövecses (1999) predict, there is always one direction of mapping which is preferred. On a general level, these default routes of mapping can often be related to ontological salience. E FFECTS are better vehicles 235 than CAUSES since they are more immediate. C ONTAINERS are better vehicles than CONTENTS , since they are visible and bounded. And that humans are more likely to be chosen as vehicles than as targets is evident in the discussion of the CONTROL and the PRODUCTION ICMs. But this is not the whole story. The laws of ontological salience can serve as rough guidelines, however when it comes to single cases they are not sufficient to account for obvious preferences. The principle HUMAN OVER NON - HUMAN cannot explain why we find more references to physical, especially consumer products, in the PRODUCTION ICM than to abstract products. This tendency can be better explained by having recourse to the principles VISIBLE OVER NON - VISIBLE and CONCRETE OVER ABSTRACT . In many cases it is unclear which and how many laws of ontological salience apply and which do not. And many instances that look simple at first glance turn out to be quite complex. Is it really convincing to argue for the application of the principle VISIBLE OVER NON - VISIBLE when the CONTAINER is transparent as is the case of glasses? And what if different principles conflict with one another in a given case? Is it more important for the metonym bus meaning ‘bus driver’ that the vehicle is non-human and the target human, or does it matter more that the vehicle is more interactional than the target? Does it really play a role for the conventionality of metonymies that institutions ultimately consist of human beings or is this not the way ordinary speakers perceive the world? Moreover, in some cases all laws of ontological salience are ignored and the speech community clearly prefers the mapping direction which is less in accordance with these principles, as was found for the CATEGORY and MEMBER mappings. As concerns the weightiness of the single laws of ontological salience, one can safely assume that the principle HUMAN OVER NON - HUMAN is essential. This shows up in the general preference for CONTROLLER FOR CON- TROLLED over CONTROLLED FOR CONTROLLER metonymies, in the preference for PRODUCER FOR PRODUCT over PRODUCT FOR PRODUCER metonymies, but also in the tendency of the mappings LOCATION FOR PEOPLE , BODY PART FOR PERSON as well as ATTRIBUTE OF PERSON FOR PERSON to lead to quite low relative frequencies. On the whole, humans are not very good metonymic targets. Generally speaking, it is much more likely that an object will be metonymically construed than a person. The material an object consists of is often a good metonymic reference point as well as are important attributes which can be clearly linked with a given object, such as a diesel or a four wheel drive. L OCATIONS which are famous for one single product are also good entry points to metonymic construals. And LOCATIONS are apt to conventionally function as metonymic vehicles, particularly so if they stand for one well-known event (e.g. Watergate) or a specific abstract institution (e.g. Wall Street), especially when the INSTITUTION is associated with a single BUILDING (e.g. Pentagon). 236 I NSTITUTIONS , above all political ones, such as the government, serve as useful metonymic reference points to an otherwise undefined group of persons. But, on the other hand, it is exactly because institutions are often perceived as being of quite an abstract nature that they are also frequently the targets of metonymies. C ENTRAL FACTORS are sometimes fairly conventional metonymic vehicles in this field (e.g. press, screen, seat). Or, if the institutions produce something, they are identified via their products (e.g. newspaper, label). It is thus possible to distinguish between metonymic mappings which ordinarily lead to more or less conventional metonymic expressions. Regrettably, my data are not comprehensive enough to locate all mappings analyzed on a scale of conventionality, but some trends are definitely apparent. Among the least conventional linguistic metonyms are: • CONTENT FOR CONTAINER mappings, • most of the mappings which involve PERSONS as their targets, such as CONTROLLED FOR CONTROLLER , POSSESSED FOR POSSESSOR , BODY PART FOR PERSON , and LOCATION FOR PEOPLE (especially LOCATION FOR ORDINARY PEOPLE , which was found for many different lexemes, but is not frequent for any of them), • MEMBER OF CATEGORY FOR CATEGORY , • and some of the less important mappings discussed in chapter 6.2.2.7, such as mappings with TIME as target or CONTENT FOR AMOUNT OF CON- TENT . The mappings which can produce extremely conventional linguistic metonyms are according to my data: • MATERIAL FOR OBJECT , if the material is typically used to produce only one or a very limited number of different types of objects, • CATEGORY FOR SALIENT MEMBER , • EFFECT FOR CAUSE , • LOCATION FOR INSTITUTION mappings, especially BUILDING FOR INSTITU- TION , • IMPORTANT PART FOR WHOLE , if the part used for metonymic reference is highly functional for the whole or forms an otherwise highly salient part in the vehicle concept, • and PRODUCER FOR PRODUCT mappings which refer to consumer products. Somewhere in between these two extremes we find many other mappings, which are neither particularly likely to lead to conventional metonymic expressions nor expressions which are felt to be highly figurative. Among them are: • WHOLE FOR PRINCIPAL PART , 237 • CONTAINER FOR CONTENT , • PRODUCER FOR ABSTRACT PRODUCT , • PHYSICAL PRODUCT FOR PRODUCER , • and LOCATION FOR PRODUCT MADE THERE mappings, which have a reasonably strong link between the place and one specific product. Note that these are not more than general tendencies. There are almost invariably exceptions to the rules. As has been pointed out repeatedly during the discussion of the results of the corpus study, there are often lexical items whose relative frequencies of metonymic uses are markedly higher than those instantiated by the other words which have been found to realize the same mappings. The prime example is the high relative frequency of crown as a COSTUME FOR WEARER metonymy, which is sanctioned by the extremely specific relationship between a crown and a queen or king. A highly specific relationship between a metonymic vehicle and target results in a high target-in-vehicle salience and coincides as a rule with high relative frequencies of occurrence of single linguistic metonyms: The highest relative frequencies of CONTAINER FOR CONTENT metonymies are displayed by examples with just such an unambiguous and definite vehicle-target relationship. Books and dish are thus more often used in a ‘content’ sense than packet and label are. Within the PRODUCTION ICM the high relative frequencies of instances like BMW or Renault can also be linked to the fact that these producers do not fabricate various types of products, but mainly one, i.e. cars, which renders ‘cars’ a salient piece of information in the PRO- DUCER concepts. Further support comes from the LOCATION ICM, where it could be shown that for both LOCATION FOR EVENT and LOCATION FOR INSTI- TUTION , high relative frequencies of single metonyms are dependent on a specific and clear-cut relationship between a given vehicle and a target. This is especially true for Watergate and also holds for Vietnam and Pearl Harbo(u)r. In Rome, and America, however, so many different events have taken place that none of them is especially salient, which leads to low relative frequencies in these cases. And an INSTITUTION can also only be referred to conventionally by way of its LOCATION , if the location is closely - and exclusively - associated with one specific institution. This marks the decisive difference between highly frequent metonyms like Wall Street, Pentagon and White House and distinctly less frequent cases such as Paris, China or room. Attribute salience can account for most outliers found in the corpus study. Thus far, the consideration of the structure of the concepts involved in metonymic mappings was based on intuition only. To corroborate these assumptions, the conceptual representations of a limited number of metonymic vehicles were investigated with an attribute-listing task. As the next section demonstrates (chapter 6.3.1), some metonymic targets are indeed extremely salient in the respective vehicle concepts, whereas the conceptual structure of other vehicle concepts is much less determined by their poten- 238 tial metonymic targets. Target-in-vehicle salience can thus explain extremely conventional cases and highly unconventional cases alike. A high target-in-vehicle salience can be the result of early experiences or it can be a matter of encyclopedic knowledge acquired later in life. Especially in the latter cases, the salient attributes which characterize the vehicle concepts can therefore be best described as emerging, i.e. as emergent structure in the sense of the conceptual blending theory. This aspect will be discussed in chapter 6.3.2, where it will be shown that with regard to metonymies two types of emergent structure with different consequences for the conventionality of the linguistic expressions have to be distinguished. 6.3 Target-in-vehicle salience and the emergent structure of metonymies 6.3.1 Salient and non-salient metonymic targets In the last chapter, MATERIAL OF OBJECT FOR OBJECT was described as belonging to the group of mappings which are most likely to lead to conventional linguistic metonyms. All my examples in this group diverge from the general laws of ontological salience, however, most of the metonyms are not infrequent (e.g. glasses, furs, and paper; cf. Appendix A and chapter 6.2.2.1). Plastic, however, is with 2.50% rarely used to refer to concrete objects consisting of the material. And only 0.50% of all uses instantiate the meaning I was originally looking for, i.e. ‘credit card’ (e.g. Edward has his plastic nicked; CH2-5947). That credit cards are not very prominent in speakers’ concept of PLASTIC can be shown with the help of attribute-listing tasks. A smallscale survey was conducted with a group of 20 native speakers of German to prove this point. Many metonyms work in German just as well as in English, for example ‘international’ PRODUCT FOR PRODUCER metonymies, such as Renault or BMW, but also many other metonyms, such as Vietnam, Bordeaux, etc. The survey consisted of two parts. One group of participants were presented with key words without a context. Another group was presented with the same key words in a metonymically biasing context. In both cases, the participants were asked to list any attributes they considered relevant for a description of the lexical item. For each key word they were given one minute to complete the lists. All in all, ten different metonyms working in both languages were tested: Vietnam, Flasche (‘bottle’), Plastik (‘plastic’), Busse (‘buses’), Pille (‘pill’), Renault, Bordeaux, Ketchup (‘ketchup’), Weißes Haus (‘White House’), and Picasso. 193 193 For the proper nouns investigated here, one should actually speak of associations rather than of attributes. The term attributes is commonly reserved for a consideration of common nouns or rather the categories they denote. However, I am less interested in the theoretical status of these parts of our knowledge of domains than in which of 239 Despite the restricted scope of the study, the results confirm the view that some metonymic targets or attributes that are strongly connected to the targets are highly salient in the vehicle concepts. Interestingly, those metonyms 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 words and 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. However, this does not come as a surprise, considering that German native speakers’ knowledge about Vietnam or the White House as well as their experiences with bottles and Bordeaux are probably not too different from English speakers’ knowledge of and experiences with these locations and things. The results for Plastik ‘plastic’ in the context-free condition support the assumption that the material is not typically associated with credit cards or any other specific object. All in all, 24 different attributes were mentioned. English translations or glosses of those which occurred more than once are provided in Table 41. The first of these attributes speaks for itself. The only concrete objects associated with plastic according to this limited survey are plastic bags, toys, bottles, and - somewhat astonishingly - works of art. The other attributes centre around rubbish and its recycling. Most of the 24 different attributes mentioned are linked to the material itself and its properties, such as ‘transparent’, ‘artificial’, or ‘unbreakable’. Number of occurrence Attribute 5 ‘material that can be used to produce virtually anything’ 4 ‘recycling’ 3 ‘(plastic) bag’ 3 ‘works of art’ 2 ‘rubbish’ 2 ‘toys’ 2 ‘bottle’ Table 41: Attributes of German Plastik (‘plastic’) - context-free condition In the context-condition the picture changed significantly. The metonymically biasing context was quite unambiguous (Mit Plastik zu bezahlen wird immer beliebter ‘Paying with plastic is becoming more and more popular’). This accounts for the high number of occurrences of ‘(credit) card’ in the attribute lists. All participants mentioned this. And all of the other attributes which occurred 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. This supports them are of prime importance for the average speaker. The terms attribute and association will therefore be used interchangeably in what follows. 240 Langacker’s (1987; 1993) assumption that once the target has been accessed, the reference point recedes into the background. Of the 26 different attributes given in the context-condition, only a few are clearly related to the material: ‘not tasty’, ‘can melt’, ‘industrialized’, and ‘synthetic material’. For more conventional metonyms the targets are much more salient in the vehicle domain. Examples are the PRODUCER FOR PRODUCT metonymy Renault and the CONTAINER FOR CONTENT metonymy Flasche (‘bottle’). The most frequent attribute for Renault 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 when the company name is encountered in isolation. The only attribute which is actually linked to the COMPANY is the Renault slogan ‘créateurs d’automobiles’, but the slogan makes reference to the typical product as well. Also quite salient are attributes which are connected to the domain of MOTOR RACING , but this domain is associated with cars, too. When a metonymically biasing context is given (Sie kaufte sich einen Renault ‘She bought a Renault’), the picture changes, but not as dramatically as for Plastik ‘plastic’. Only two ideas are foregrounded: the CAR and the COUNTRY it comes from. Of the eleven different attributes given, ten concern the car or its properties. ‘Car’ occurs in seven lists. The remainder of the attributes are quite varied. Only one, i.e. ‘cheap’, occurred more than once. Others include ‘driving’, ‘small’, ‘good design’, etc. The eleventh, which occurred in half of the questionnaires, is ‘France’. These results largely support the findings of the corpus study. Cars and their producers are so intricately linked to one another in speakers’ minds that metonymic references to the PRODUCTS by way of the PRODUCERS are easily understood and therefore conventionally used. Actually, the ‘car’ meaning seems to be more salient than the company itself. This is backed up by the products being more immediate and interactional than the company. The conceptual link between CONTAINERS and their CONTENT is equally strong. While the majority of the attributes (15 out of 30 different attributes) listed for the lexeme Flasche (‘bottle’) in isolation centre around the bottle, its material and properties, the content is also fairly important. Nine different attributes were given which belong to the domain of BEVERAGES . Examples are ‘beer’, ‘water’, ‘wine’, ‘one litre’, etc. Other attributes which cannot be clearly assigned to either the CONTAINER or the CONTENT are ‘baby’, which occurred three times, and some less important ideas like ‘beer mat’. In a metonymically biasing context (Sie trank eine ganze Flasche ‘She drank a whole bottle’), the CONTENT is foregrounded, but the CONTAINER does not entirely lose its prominence. Out of 26 different attributes, 14 now refer to the CONTENT (e.g. ‘alcohol’, ‘contains liquid’, ‘water’, etc.), but there are still eleven which belong to the CONTAINER concept (e.g. ‘glass’, ‘plastic’, ‘green’, etc.). The mental path from container to content is deeply entrenched. The 241 content is quite salient in the isolated CONTAINER concept, but even the metonymic reference to the content via the container does not result in a complete conceptual cancellation of the activation of the container concept. This is markedly different for CONTENT FOR CONTAINER metonyms, which belong to the least productive class of metonymies. If a CONTENT is chosen as metonymic reference point, the CONTAINER is not conceptually prominent. The CONTENT FOR CONTAINER metonymy ketchup, which could not be found in the BNC, served as an example here. In the contextcondition, 26 different attributes were mentioned. 24 of them belong to the CONTENT domain (e.g. ‘made from tomatoes’, ‘red’, ‘liquid’, ‘spicy’, etc.) and only two to the CONTAINER domain (‘in bottles’, ‘glass’). But at least the idea that ketchup usually comes ‘in bottles’ was mentioned by three of the ten participants. The results of the context-free condition corroborate that the CONTAINER is not salient in the CONTENT domain. All of the attributes given here are related to the content (e.g. ‘tomato’, ‘sweet’, ‘sugar’, etc.). Evidently, the CONTAINER - CONTENT relation has a very strong directionality (cf. Figure 19). The CONTENT is the most salient element of the CON- TAINMENT ICM. This is shown in the figure by the bold-print line surrounding the attributes the CONTENT concept consists of. The CONTAINER is, in contrast to this, much less salient. In both central concepts in the CON- TAINMENT ICM, the CONTAINER and the CONTENT , CONTENT -related attributes (signified by ‘+’ in the figure) are much more numerous and much weightier than CONTAINER -related attributes (signified by ‘o’ in the figure). The CONTENT is easily activated via the sub-domain of the CONTAINER , since the CONTAINER concept is full of CONTENT -related attributes. And because CONTAINER -related attributes play a minor role in the CONTENT concept, the CONTENT is not a suitable vehicle to refer to the CONTAINER . When the CON- TAINER is mentioned, the CONTENT is automatically activated as well, but when the CONTENT is mentioned, the CONTAINER is not salient. The associative strength between a CONTAINER and its CONTENT is thus markedly higher than the link leading from a CONTENT to its CONTAINER . In Figure 19, this is illustrated by the arrows leading into the two different directions. Figure 19: The directionality of the relationship between CONTAINER and CON- TENT and its dependency on salient attributes (+ = CONTENT -related attributes, o = CONTAINER -related attributes) CONTAINER CONTENT + + + o + + + o + + + o + + + + + o + + + + + o + + + + + + + + + + + + + + o + + + + + + + + + + + + + + o + + + + + strength of associative link 242 Even though the container is what people interact with when someone passes them the ketchup, the salt, or whatever and even though the content is not always visible while in a container, humans typically focus more on the content than on the container itself - and not astonishingly so, since the content has a much higher relevance and functionality than the container, whose sole function is that of containment. As a rule, metonymies are preferred where the target concept is salient in the vehicle concept. Note that target-in-vehicle salience is not meant as a proper part-whole relation. The target is not necessarily a sub-part of the vehicle, but is salient in the vehicle domain in terms of attributes which are related to the target. The principle that metonyms where the target stands in a highly salient relationship to the vehicle have a greater conventionalization potential than those where this is not the case can be applied to reversible and non-reversible mappings alike. For the reversible ones, there is a generally preferred direction of mapping which is usually determined by the laws of ontological salience. But single linguistic instantiations only gain currency when the target is salient in the vehicle. This marks the difference between I’m reading Shakespeare and *I’m reading Jane (cf. chapter 6.2.1.4). The dispreferred direction of mapping leads to less accepted linguistic metonyms exactly because the targets do not stand in a salient relationship to the vehicle concepts. This has been shown for the CONTENT FOR CONTAINER metonymies above, but also holds for many other cases. P RO- DUCERS are usually not very salient in PRODUCT domains, since we are much more interested in what we can do with the product than in who made it. Specific CONTROLLERS are not very salient in the domains of the CONTROLLEDS , since the functions of the objects (i.e. the CONTROLLEDS ) are more prominent than those who actually carry them out (i.e. the CONTROL- LERS ). A TTRIBUTES OF OBJECTS are not very good reference points to mentally access OBJECTS if the target is not salient in the vehicle domain. Consider the IMPORTANT ATTRIBUTE FOR OBJECT metonyms once more. Bronze for ‘cow with a bronze-coloured fur’ is so unconventional because the target is nonsalient in the vehicle; it is just one of many potential attributes. Diesel, however, is much more accepted and current, since diesel is closely associated with vehicles which are powered by this type of engine. The LOCATION FOR PRODUCT mapping has been said to occupy a middle position on the scale of conventionality given in chapter 6.2.3, leading to linguistic metonyms which display a medium degree of conventionality, but which are neither highly conventional nor extremely figurative. But this is only true for metonyms which follow the principle that the target is salient in the vehicle. For Bordeaux, for example, ‘wine’ is a very salient attribute. In my survey ten of 22 attributes given for Bordeaux in the context-free condition are related to the typical product (e.g. ‘tastes dry’, ‘red’, ‘stored in catacombs’, ‘wine-growing’). Table 42 summarizes the attributes which were mentioned more than once for Bordeaux. 243 Number of occurrence Attribute 8 ‘France’ 6 ‘(red) wine’ 2 ‘city’ 2 ‘tastes dry’ 2 ‘region’ Table 42: The attributes of Bordeaux - context-free condition However, if the target does not stand in a salient relationship with the vehicle, LOCATION FOR PRODUCT metonymies should have a low conventionalization potential. This can be demonstrated with the unacceptability of a potential metonymic use of, say, Germany. Even though Germany is well-known for its car production, cars do not form a particularly salient part of the average language user’s concept of the country. Therefore a metonymic reference to a car by saying *I just bought a Germany is impossible. L OCATION FOR EVENT and LOCATION FOR INSTITUTION metonymies are also much more common when the target is salient in the vehicle concept. The results of the context-free condition concerning Vietnam indicate this. 16 out of 34 different attributes listed are related to the Vietnam War. Attributes like ‘chemical weapons’, ‘napalm’, ‘Vietcong’, ‘war’, and ‘America/ USA’ abound here. 194 The war has become an established part of our knowledge about the country, actually the part that governs our concept of the country to an overwhelming extent. The BUILDING FOR INSTITUTION metonym White House is also in line with target-in-vehicle salience. The attributes collected for the German equivalent Weißes Haus in the context-free condition clearly show that the target is a crucial part of the vehicle. What is most prominent in the concept are aspects related to politics, such as ‘government’, ‘Pentagon’, ‘war’, ‘conservatives’, ‘Republicans’, ‘news’, ‘press conferences’, ‘power’, ‘president’, etc. Attributes which are connected with the building are rare (four out of 29; e.g. ‘garden’, ‘large’). Thus, it is not a property of the mapping itself that it necessarily leads to some quite conventional linguistic metonyms. It can do so, if and only if the target is salient in the vehicle. Target-in-vehicle salience thus has consequences for the continuum ranging from unconventional to highly conventional mappings proposed in chapter 6.2.3. The restrictions already mentioned there, i.e. that, for example, MATERIAL OF OBJECT FOR OBJECT is likely to produce highly conventional metonyms, but only if the material is typically used to produce only one or a very limited number of different types of objects, gain special importance. Most of the examples of this mapping investigated here (e.g. furs, 194 The country itself does not play too big a role in the conceptual representation of Vietnam. The only attributes that are related to the country proper are ‘country’, ‘Asia’, ‘Saigon’, ‘jungle’, and ‘Far East’. Cf. also chapter 6.3.2. 244 glasses, paper) are in line with target-in-vehicle salience, since, not surprisingly, the sources from which the metonyms were collected to begin with often pick out 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. It could also be the case that the ‘unconventional’ MATERIAL FOR OBJECT metonymy plastic (‘credit card’) is a much more characteristic example of the mapping than the conventionalized cases such as glasses or furs. To unequivocally answer this question, one would have to examine the frequencies of occurrence of as many examples of the mapping as possible and look into the conceptual structure of the single vehicle concepts. Only an exhaustive study of as many linguistic manifestations of a given mapping can reveal how conventional the mapping really is, i.e. differentiate the rule from the exceptional cases. This is especially relevant with regard to the mappings that were described as occupying a position at the upper end of the scale of conventionality, i.e. as particularly conventional. For the mappings at the other extreme of the continuum, the results are more reliable. There are almost certainly not too many exceptions to the rule that PERSONS are particularly poor metonymic targets - on the one hand, due to the weightiness of the HUMAN OVER NON - HUMAN principle, and on the other hand since PERSONS are, in general, not likely to be highly salient attributes of LOCATIONS , POS- SESSIONS , or CONTROLLEDS , simply because the fact that locations are often inhabited by human beings, and things are more often than not possessed or controlled by humans is too trivial and usually more or less taken for granted. In these cases, attribute salience supports the more general trends that can also be accounted for by the laws of ontological salience. But it is of special significance for the identification and explanation of outliers, i.e. for the description of single linguistic instances. The very existence of the mapping CATEGORY FOR SALIENT MEMBER sustains the view that the salience of the target or target-related attributes is of importance for metonymic reasoning. But target-in-vehicle salience also plays a role for the reversed pattern MEMBER OF CATEGORY FOR CATEGORY . This was already discussed for the example aspirin, but also sanctions many paragon uses. A paragon like a second Shakespeare or another Watergate only functions when the vehicles are unambiguously associated with one highly salient attribute which the target, i.e. the newly established category, inherits. When this precondition holds, the target is - automatically - highly salient in the vehicle domain, since it is inherited from the vehicle in the first place. Without target-in-vehicle salience, paragons like these could not be used at all to successfully convey the intended meanings. The conventionality of linguistic metonyms is therefore largely a matter of attribute salience. The most conventional metonymies are those which are in line with both the general laws of ontological salience and the principle of target-in-vehicle salience. But while ontological salience can only 245 lead to predictions as to general tendencies with respect to the conventionalization potential of metonyms which are based on a given conceptual mapping, attribute salience can explain exceptions to the rule. 6.3.2 Types of attribute salience and the question of emergent structure If the conventionality of metonyms is largely determined by the degree of salience of the target in the vehicle, the question arises as to where such a high salience comes from. On the one hand, it can be the result of a given, monofunctional and unambiguous relationship between a vehicle and a target. Consider the CONTAINER - CONTENT relationship once more. The key function of containers is to hold substances or items. The fairly high salience of the CONTENT in the CONTAINER concept is therefore natural. Especially if the container is not an all-purpose one, but made to hold specific substances or objects, the high salience is given in extralinguistic reality. Other metonyms are also based on monofunctionality. Football is perhaps a prime example. The FOOTBALL GAME concept is obviously highly salient for the PLAYING PIECE concept: Footballs are made to play football with. Artefacts produced to fulfil a special purpose are thus very good candidates to conventionally function as metonymic vehicles that convey meanings related to this function. In a similar vein, PRODUCERS and concrete PRODUCTS are often so inseparably connected to each other. Companies usually produce something. This is their main function, which makes the products we interact with so highly salient in the PRODUCTION ICM. In principle, almost any relationship between two contiguous concepts can be exploited to form metonymies. But only those which are (either) in accordance with the laws of ontological salience and/ or the principle of target-in-vehicle salience stand a good chance of being accepted by the speech community. As a result, target-in-vehicle salience is connected to cognitive salience. Linguistic expressions which are in line with target-invehicle salience are more likely to be used in a wider range of contexts, since they do not need much contextual support in order to be understood effortlessly. They are used more often, heard more often, and processed more often. This frequent activation is not without consequences for the ease of processing. The mental path from vehicle to target gets more deeply entrenched with every mention or processing of the linguistic metonym. Target-in-vehicle salience enhances a frequent cognitive salience, which then fosters the entrenchment of the metonymic meaning for the vehicle term. The entrenchment of the metonymic sense, i.e. the entrenchment of the target sense for the vehicle, in turn, further strengthens the salience of the target or target-related attributes within the vehicle concept. Target-invehicle salience is thus both a facilitating factor for and a result of the conventionality of a linguistic metonym. 246 But target-in-vehicle salience is not always connected to monofunctionality as indicated above for PLAYING PIECES and CONTAINERS . It can also be caused by other factors. Good examples to illustrate this point are LO- CATION FOR EVENT metonymies. The relationship between a LOCATION and an EVENT is different from the relationship between a CONTAINER and its CONTENT . The latter is one we grow up with and acquire quite early, it has an experiential basis. The former belongs to our later acquired encyclopedic knowledge. The high salience of the Vietnam War in the context-free concept V IETNAM is a matter of late learning rather than of early experience. Fostered by the frequency of the metonymic shorthand, the war has become a vital part of our concept of the country. The best way to describe this is as a conceptual blend. The country and a war which took place there have melted into a novel concept: the Vietnam War, which can stand as a paragon for destruction and cruelty. But this blend has consequences for the mental representation of the country proper. Since blends are hooked up to their inputs, the whole conceptual network can be influenced by changing mental representations and novel insights (cf. Figure 20). V IETNAM - Country and population: ‘jungle’, ‘Asia’, ‘tropical’, ‘poor’, etc. -Vietnam War: ‘war’, ‘weapons’, ‘napalm’, ‘helicopters’, ‘Vietcong’, ‘60s’, ‘hippies’, ‘Ho Chi Minh’, ‘protests’, ‘partition’, ‘USA’, etc. - Spare time: ‘food’, ‘holidays’, ‘travel’, etc. W AR IN V IETNAM ‘napalm’, ‘helicopters’, ‘USA’, ‘hippies’, ‘protests’, ‘Ho Chi Minh’, ‘partition’, etc. S PARE TIME ‘Vietnamese food’, ‘paradise’, ‘sea’, ‘coast’, ‘sun’, ‘holidays’, etc. V IETNAM W AR ‘war’, ‘napalm’, ‘weapons’, ‘fire’, ‘post-war trauma’, ‘death’, ‘no sleep’, ‘terror’, ‘terrifying’, ‘veterans’, etc. Figure 20: Emergent structure in the Vietnam-blend Figure 20 is of course simplified. But it shows very well how much the representation of the country is dependent on speakers’ encyclopedic knowledge and how the knowledge about the Vietnam War feeds back into one of the original input spaces, i.e. the COUNTRY space. 195 The COUNTRY concept and the concept WAR IN V IETNAM are blended into the V IETNAM 195 The attributes given in the figure are as far as possible based on my questionnaires. 247 W AR space. The Vietnam War is not any war located in this country, but clearly linked to specific ideas of extreme terror and its consequences. This knowledge about the long-term consequences and the brutality of this war is the emergent structure of the V IETNAM W AR space. However, the close association of the country and the war has turned the Vietnam War and its attributes into the most salient knowledge about the country itself. The emergent structure of the blend has made its way into the COUNTRY space. In addition, another space, somewhat clumsily called SPARE TIME in Figure 20, plays a role in the mental representation of the country. It deals with knowledge about going to Vietnamese restaurants or travelling to the country. This ‘holiday’ or ‘spare time’ idea has become so relevant for the country that even a context which clearly favours the metonymic ‘war’ meaning cannot suppress it. Seven out of a total of 42 different attributes mentioned for Vietnam in a metonymically biasing context belong here. It has been pointed out in chapter 3.3.2 that metonymies can regularly be described as conceptual blends. For the Picasso-blend the emergent structure has been depicted as consisting of attributes like ‘expensive’ or ‘importance in art history’. For the Vietnam-blend it also mainly contains attributes which are highly salient in input space 2, i.e. in the metonymic target, and have become salient in the vehicle. However, there are other instances of metonymies where what is foregrounded by way of the metonymic construal is not a particularly salient feature of the target. Cases in point are examples like skirt or suits. The following definitions taken from the NODE illustrate this: (213) skirt: ‘women regarded as objects of sexual desire’ (214) suit: ‘a high-ranking executive in a business or organization, typically one regarded as exercising influence in an impersonal way’ Both uses carry the label informal in the NODE and were found to be infrequent in the corpus study (0.90% and 3.34% respectively). But why are they not more common? On the one hand, they deviate from NON - HUMAN OVER HUMAN . On the other hand, one could argue that their conventionalization should be higher despite the fact that they diverge from the laws of ontological salience, since the targets are probably salient in the vehicles. After all clothes are made for wearing, and more likely than not an attribute list for skirt would feature the idea that skirts are typically worn by females and suits is likely to be connected with the idea that they are worn on more formal occasions including many white-collar jobs. The problem with this analysis is that the meanings of the two metonyms are more specific than that. They invariably carry the connotations ‘sexually attractive’ and ‘impersonal’ respectively. This certainly restricts the class of potential referents and therefore also the occasions where they can be used. More importantly, however, the additional information conveyed by the metonyms distances them from the underlying conceptual motivation. The mental path from vehicle to target is not as straightforward as for crown or uniforms, because 248 ‘sexually attractive’ and ‘impersonal’ are probably not the first things which come to speakers’ minds when the concepts WOMAN and MANAGER are activated. The situation for skirt is depicted in Figure 21. Moreover, the mental route from the vehicle to the target is obscured by the fact that not every sexually attractive woman necessarily wears skirts. SKIRT WOMAN ‘made of cloth’ ‘worn by women’ ‘long/ short’ ‘tight/ loose’ etc. ‘is female’ ‘raises children’ ‘can be housewife’ ‘can cook’ ‘sexually attractive’ ‘young/ old’ etc. A SKIRT ‘sexually attractive woman’ Figure 21: Skirt ‘sexually attractive woman’ as a conceptual blend The actual target, i.e. the full meaning of the metonym including the idea of ‘sexual attractiveness’, is not salient in the vehicle. Only the more neutral connection between skirts and females is. This makes the emergent structure of skirt different from the emergent structure of cases like Picasso or Vietnam. For the latter the emergent structure is less intricate. The key parts are highly salient in the target as well as in the vehicle. For skirt, however, the metonym accentuates an attribute which is only of average relevance for the target and certainly non-salient in the vehicle. And this otherwise not particularly salient attribute becomes the pragmatic point of using the metonym. In general, the emergent structure of metonyms is not entirely novel as it often is for metaphors, when the emergent aspects are exclusively found in the blend. However, there are two different types of emergent structure which can be part of the blend for metonyms. On the one hand, the emergent structure can consist of and correspond to highly salient attributes in the vehicle and in the target - it can uphold the saliency structure of the concepts involved. On the other hand, it can result in a marked change of attribute salience. For skirt, for example, this means that a potential referent of the metonym is seen in a different way, which is not predictable considering the overall structure of the domains involved. The distinction of these two types of emergent structure of metonyms is crucial when it comes to a description of the conventionality of single metonymic expressions. The former type, the attribute-salience upholding 249 emergent structure, which is typical of metonyms and does not disturb the structure of the domains in terms of the salience of the single attributes too much, is likely to enhance their conventionality. The latter type, which leads to a significant shift of attribute salience and thus to more creative conceptualizations, prevents rather than fosters the conventionality of the expressions in question. The second type is much less regular than the first and leads to more idiosyncratic metonymic senses. Metonyms with salience-changing emergent structure have a certain degree of arbitrariness. Admittedly, the speech community has agreed on these more specialized meanings, but as far as metonyms in general are concerned, these cases are clearly dispreferred as opposed to those with a salience-upholding emergent structure. By being so specialized, they lose some of their connection to the conceptual patterns motivating them. They are less easy to remember and to understand. The salience-changing emergent structure has to be learned, it cannot be derived from the conceptual mappings alone. Thus, the cognitive advantages usually connected to metonymies are lost to a certain degree. Together with the fact that the idiosyncrasy of the meanings severely restricts the range of potential referents, this reduces the likelihood of a deep entrenchment of the metonymic meanings. They are seldom used, 196 hence rarely cognitively salient and infrequently used mental paths are never well-entrenched. 6.3.3 Summary All in all, attribute salience is a highly relevant factor for the entrenchment and subsequent conventionality of metonymic meanings. If metonymic construals are in line with the structure of a vehicle concept in terms of its most salient attributes, they are rendered more cognitively salient. Once the mental paths connecting vehicles and targets are easy to find, the metonymic senses conveyed are likely to be stored and become more central members of the set of meanings attached to a lexical form. And eventually, they can - in the case of a fully fledged semantic change, i.e. if the original literal senses cease to exist - even become the only cognitively salient senses (cf. Geeraerts 1997, 1999). While this is not the case for any of the examples considered in the corpus study, many of the figurative senses are at least on equal footing with the literal ones, such as the PRODUCER FOR PRODUCT meaning of BMW, the MATERIAL FOR OBJECT meaning of paper, the WHOLE FOR PRINCIPAL PART meaning of America. Sometimes the figurative senses are even much more important than the literal ones, like for football, glasses, and press. That the figurative meanings will at some time entirely replace the literal ones is, however, not very likely. As long as humans play 196 Typically, salience-changing emergent structure results in metonyms with negative connotations, which severely restricts their applicability in concrete speech situations. This makes them similar to well-known stock examples of bahuvrihi compounds like egghead. 250 football using a football, use glass to fabricate containers for drinks and spectacles, and need printing presses to produce newspapers, the literal meanings will survive. As will be demonstrated below, attribute salience also plays a role for the conventionality of metaphors (cf. chapter 7.4.2). But before I turn this aspect, the results of the corpus study regarding the relative frequencies of metaphorical meanings need to be discussed. 251 7 Evidence for the conventionality of metaphorical meaning and thought 7.1 The results of the metaphor study: An overview One of the most central claims of conceptual metaphor theory is that metaphorically motivated meanings abound in everyday language. That this is not only a vague assumption has been shown, for example, by Pollio, Smith and Pollio (1990), who observed that about five words out of 100 are used in a figurative way. A brief glance at the summary of the results of my metaphor study (cf. Appendix B) confirms that all target word-forms were found in metaphorical senses. Compared to the metonymic uses found, the total amount of metaphorical uses is considerably higher. The mean value of the metaphorical uses amounts to 41.32%, whereas the mean value of the metonymical uses is only 26.14%. Whether these results can be generalized, however, is not entirely clear, since the mean values are certainly influenced by the choice of the lexical items. If other and more word-forms had been considered, the outcome might have been different. Moreover, the picture drawn here is unquestionably biased towards written language use, since the BNC consists to about 90% of written text-types. Nevertheless, there is significant variation as concerns the overall amount of metaphorical uses found for single word-forms. The largest percentage of metaphorical uses is 98.95% (for bombshell) and the lowest is 0.80% (for buy). The standard deviation for metaphors is with 24.42 almost as high as for the metonymies (25.85). Nearly half of the 50 potentially metaphorical word-forms analyzed returned values between 30% and 60%. The overall percentages established for the metaphorical uses of lexical items which often occur as standard examples of conceptual metaphor theory come close to the mean value. Examples are appetite, which is used metaphorically in 41.49% of all cases considered, attacked, where the metaphorical uses amount to an overall of 43.80%, constructed, where I found 43.80% of metaphorical uses, followed, where 39.60% of all concordance lines analyzed are metaphorical, and weigh, which features a total percentage of 44.44 of metaphorical instances. But these numbers only indicate the overall number of metaphorical meanings found. How many of them realize the meanings usually described in the literature? Let me begin with attacked. The meaning intended in the famous example sentence He attacked every weak point in my argument (Lakoff and Johnson 1980), which is based on the mapping ARGUMENT IS WAR , accounts for 35.30% of all instances analyzed and for 80.60% of all metaphorical meanings of the form attacked (cf. 215). All other metaphorical uses are of rather marginal importance, such as ACTIONS ARE WAR : OBJECTS INVOLVED / OBSTA - 252 CLES ARE ENEMIES (cf. 216), CHEMICAL / BIO - LOGICAL REACTIONS ARE WAR : SUBSTANCES / MICROBES ARE ENEMIES (cf. 217) or THE EMOTIONAL SELF IS A BRIT- TLE OBJECT : THOUGHTS / EMOTIONS ARE ENEMIES (cf. 218). None of them occurs in as many as 5% of all meanings. (215) [...] from a speech by Dr W. Starkie, the commissioner of National Education in Ireland, who attacked school managers for not being up to their job. (A07-1290) (216) Moving quickly into the lead the Mandale Harrier attacked the formidable 719 feet of climb on the Forestry Commission track [...] (K4T-10772) (217) Valerie Slimp, 39, was fighting for life as the venom from the deadly Brown Recluse Spider attacked the cells in her body. (CH6-4091) (218) But, back in my room, all sorts of hopeless thoughts again attacked me. (B0U-2332) For constructed, the overall picture is similar. The most frequent metaphorical meaning by far is motivated by the mapping THEORIES ARE BUILDINGS (23.30%; cf. 219). There is only one other meaning which is at least relatively frequent, since it accounts for 9.00% of all meanings. This meaning, an instance of the CONDUIT METAPHOR with the mapping THE STIMULUS IS A BUILDING , is illustrated by examples (220) and (221). What all instances of this mapping have in common is that a text of any kind is ‘made up’ of different ‘building blocks’. The meaning is certainly connected to the THEO- RIES ARE BUILDINGS metaphor, since texts can be considered as theoretical constructs. This underlines once more the importance of the mapping THEORIES ARE BUILDINGS . Other mappings identified as motivating uses of constructed like AN EMOTION IS A BUILDING (cf. 222) or MEALS ARE BUILDINGS (cf. 223) occur considerably less often (0.80% and 0.10% respectively): (219) This is partly the inevitable consequence of gender being socially constructed. (A6D-1388) (220) Within minutes of the prime ministers sitting down at 4.15 p.m., we had constructed the wording of what became Early Day Motion number 174, and which reads as follows (H91-1166) (221) [...] In suggesting the elements from which an answer can be constructed, we should bear in mind that there, as often in syntax, to assume a single explanatory factor risks oversimplifying matters. (HPY-427) (222) [...] resulted in a type of informal sub-structural power dynamic which in turn regenerated the culturally constructed fear of women necessary to patriarchal interests [...] (ACL-460) (223) These [i.e. the menus; SH] have fewer choices per course, but are thoughtfully constructed and generous with major-league dishes. (AHC- 1634) But similarly to what was observed for the metonymies, the meanings in which the lexical items were taken over from the sources are not always the most frequent figurative ones. Enriched, for instance, is given as an example of the mapping WELL - BEING IS WEALTH in the Master Metaphor List, as in Music has enriched my life. This sense occurs in as many as 21.43% of all 253 cases, but it is not the most common metaphorical one. Many more metaphorical instances are based on a mapping I summarized as ADDITIONAL PROPERTIES / THINGS ARE WEALTH . Examples of this mapping may be ‘less figurative’ from a conceptual point of view because at least ADDITIONAL THINGS are concrete. Some instances of this mapping therefore lack the usual concrete to abstract direction. Yet, they have a metaphorical character (cf. examples 224, 225). And where abstract properties are at issue as targets, the sentences from the BNC are clearly metaphorical (cf. examples 226, 227): (224) [...] Paris where, during the past eight or nine years, the capital has been enriched without erecting one public building in the Gothic style. (B1D- 1366) (225) [...] Brazil was now producing uranium enriched to 20 per cent [...]. (HKU-2763) (226) [...] with the concept of reform still in a state of flux it ‘must be enriched with new ideas and conclusions’. (HKW-857) (227) Everything he did was enriched by a warm concern for other people [...] (ASC-45) Since this mapping determines 50.42% of all meanings of the form enriched, this reading is undeniably deeply entrenched in speakers’ minds. With an overall of 87.39% of all uses being metaphorical, it is safe to assume that for enriched the literal rather than the figurative meanings will be extraordinary. More often than not the metaphorical use of one and the same wordform is not restricted to one mapping. The only targets where all metaphorical uses are motivated by one mapping are profited ( HAVING BENEFI- CIAL THINGS HAPPEN TO ONE IS GETTING MONEY ; 39.39%), sharks ( COMPETITORS OR ANTAGONISTS ARE PREDATORS ; 21.55%), and stoned ( INTOXICATION IS GET- TING DESTROYED ; 58.52%). For all other word-forms considered, more than one mapping resulting in metaphorical meanings could be identified. While for some words, there are only two or three different metaphorical meanings (e.g. core, pointed, thirst, smaller, warm; cf. Appendix B), the majority of the targets can realize at least four or more metaphorical mappings. For example, one of the possible meanings of nourished is motivated by the mapping HOPE IS A CHILD as in She nourished the hope that he would return (Lakoff, Espenson and Schwartz 1991: 151). But IDEAS or INFORMATION can also be metaphorically construed as NOURISHMENT , and not only HOPES can be construed as CHILDREN who have to be nourished, but also BELIEFS . Furthermore, the verb nourish occurs in one of Lakoff, Espenson and Schwartz’s examples of the mapping THEORIES ARE PLANTS (The theory was nourished by later experimental results; 1991: 112). And it is given as an illustration of the mapping PSYCHOLOGICAL BENEFIT IS PHYSICAL BENEFIT (Thoughts of his beloved nourished his soul; 1991: 131). But are all of these mappings really conventionally exploited in natural discourse? Table 43 254 gives an overview of the most frequent metaphorical uses found for nourished. As the table shows, many metaphorical mappings compete in this case. Apart from those listed in Table 43, there are five other mappings which occurred in my data (cf. Appendix B). But since they account only for a total of 7.28% of all meanings, they are not given here. Wordform MRs total (%) Mapping M - total (%) M - MRs (%) MENTAL EVENTS : PSYCHOLOGICAL FORCES PSYCHOLOGICAL FORCES ARE PHYSICAL FORCES : PSYCHOLOGICAL BENEFIT IS PHYS- ICAL BENEFIT : THE MIND / SELF / SOUL IS A CHILD 20.00 37.93 EMOTION AN EMOTION IS AN ANIMATE BEING : AN EMOTION IS A CHILD 10.91 20.69 MENTAL EVENTS : IDEAS IDEAS ARE ANIMATE BEINGS : AN IDEA IS A CHILD 10.00 18.97 nourished 52.73 EVENT STRUCTURE : ACTION AN ACTION ( COOPERATION / RITUAL ) IS A CHILD 4.55 8.62 Table 43: The most frequent metaphorical uses of nourished The most common metaphorical sense (20.00%) of those indicated by the Master Metaphor List is motivated by the mapping PSYCHOLOGICAL BENEFIT IS PHYSICAL BENEFIT , where the MIND / SELF / SOUL is conceptualized as a CHILD . An example illustrating this usage is (228). The mapping HOPE IS A CHILD (cf. 229) is subsumed in Table 43 under the more general heading AN EMO- TION IS A CHILD , which accounts for 10.91% of all meanings of nourished. The instances of HOPE IS A CHILD are the most frequent ones within the larger group of AN EMOTION IS A CHILD . The more specific mapping accounts for eight of a total of twelve instances. Other emotions which can be nourished include, for example, hatred (cf. 230) and envy (cf. 231). In all these cases, an emotion is conceived of as an animate being that is fed and taken care of. (228) [...] philosophical scepticism may have nourished and stimulated the mind of Greeks trained in civic religious observances [...] (A6B-1347) (229) And the Japanese advances in eastern Asia nourished the hopes that the United States would be completely tied down [...] (ADD-355) (230) He also told us how he had nourished the fire of hate inside him [...] (GXK-103) (231) [...] and frustrated envy is nourished. (ARC-616) Nearly the same amount of metaphorical instances (10.00%) relies on the mapping AN IDEA IS A CHILD , which could also have been called - parallel to 255 Lakoff, Espenson and Schwartz - AN IDEA IS A PLANT . An example illustrating this mapping is (232). What did not occur in my sample, which was with only 110 instances quite restricted for nourished, are examples where INFORMATION IS or IDEAS ARE NOURISHMENT (e.g. School nourishes your mind; Lakoff, Espenson and Schwartz 1991: 85). Instead, I found some instances where AN ACTION IS A CHILD , or more accurately a cooperation or ritual is construed as something that must be taken care of (cf. 233). Taken together, instances which convey this idea still account for 4.55% of all meanings investigated and for 8.62% of all metaphorical instances of nourished. (232) [...] it was they who nourished the idea of a unitary empire (AE8-233). (233) Many partnerships have grown from this single root, nourished by central government and funded substantially [...] (B2T-795) As this overview demonstrates, none of the single mappings is really highly frequently used. Even though more than 50% of all instances are metaphorical, the most frequent usage accounts only for 20% of all meanings. Other mappings mentioned in the Master Metaphor List are even less frequently exploited. And one mapping quoted by Lakoff, Espenson and Schwartz (1991) does not occur at all. Other examples, such as the already discussed enriched, should therefore have more entrenched metaphorical meanings than nourish. One of the lexical items whose meanings seem to be extremely flexible is escape. 50.70% of all instances were found to be metaphorically motivated. But none of the single meanings and mappings is particularly frequent. Miscellaneous things can be construed as limiting our freedom, as illustrated in the following examples: (234) [...] that small persistent voice will not allow us to escape our duty to ourselves (FP1-1897) (235) Most dictionaries would describe it as a ‘desire or tendency to escape from reality into fantasy’ [...] (HD8-116) (236) We probably escape those first fates rather more than the rabbit. (B04- 1039) (237) Morris Zapp [...], who elects to take a visiting professorship at Rummidge in order to escape from his marital problems (A1A-871) (238) Suppose you actually managed to escape all the harmful effects of tobacco [...] (AYK-166) (239) He accepted the pain beneath his ribs. Because he could no more escape it (H8R-4266/ 4267) (240) The one chain that may yet escape the decline is the John Lewis (‘never knowingly undersold’) Partnership [...] (ABE-1800) (241) Trying to escape him was about as feasible as trying to escape a very big shark in a very small fish tank. (H8H-935) (242) Because you see with her eyes for a moment, share her vision, escape yourself [...] (H8N-2708) (243) This is a modest and fairly technical book, but its central idea is likely to escape and assume a life of its own. (B71-1536) 256 (244) She resembled an enormous aquatic beetle, surprised by the light, overwhelmed by the sunshine, trying to escape with ineffectual effort [...] (EC8-184) (245) His penis was pressing hard against the inside of the fly, bent, desperate to escape and straighten. (HWA-694) (246) Neighbour Linda Moore told the court she often went to Coventry to escape the sound of Jim Reeves, whose hits in the 1960s included [...] (HJ4- 1414) The most common ones are any kind of CONTROL or OBLIGATION (9.40%; cf. example 234) and different aspects of LIFE and REALITY (cf. 235), such as FATE (cf. 236) or our EVERYDAY ROUTINE (7.50%). But not only entities with more negative connotations, such as PROBLEMS (cf. 237), CONSEQUENCES (cf. 238), ILLNESSES OR PHYSICAL DISCOMFORT (cf. example 239), ECONOMIC DE- CLINE (cf. example 240), etc., can be construed in this way. PEOPLE , or rather their presence, can be regarded as a PRISON (6.00%; cf. 241). The BODY can be a PRISON for the SELF (0.90%; cf. 242). And even BOOKS (cf. 243), LIGHT (cf. 244), CLOTHES (cf. 245) or MUSIC (cf. 246; each 0.20%) can be construed as limiting our freedom of action or the ‘freedom of action’ of non-human, abstract entities like ideas (cf. 243). All in all, I identified many mappings which are - with regard to the targets under consideration - only infrequently applied. Not all of them can be mentioned here, but Table 44 provides an overview of some randomly selected examples of these more uncommon mappings with relative frequencies of less than 0.50%: Word-form Mapping M - total (%) hollow A HEAD IS A CONTAINER FOR IDEAS 0.47 bright PAIN IS A LIGHT SOURCE 0.20 infected EVIL PEOPLE ARE A CONTAGIOUS DISEASE 0.20 killed PHYSICAL HARM / DISCOMFORT IS DEATH 0.20 poured THE H OLY S PIRIT IS A FLUID 0.20 shattered TIME IS A BRITTLE OBJECT 0.20 embraced SILENCE IS A LOVER : FALLING SILENT IS BEING EMBRACED BY IT 0.19 crack THE BODY IS A CONTAINER FOR THE SELF : THE BODY IS A CONTAINER FOR THE IMMORTAL SOUL 0.18 pinned PUTTING ON CLOTHES IS BEING PINNED INTO THEM 0.16 grasped AN AREA OF ART IS AN OBJECT : EXHIBITING IS GRASPING 0.16 wasting SPEED IS A VALUABLE COMMODITY 0.15 wounded A SPORT / GAME IS A LIVING BEING : NOT FOLLOWING THE RULES IS WOUNDING IT 0.11 followed EMOTIONS ARE ANIMATE BEINGS : BEING AFFECTED IS BEING FOLLOWED 0.10 bright DARKNESS IS A COVER : LIGHT IS REVEALING 0.10 Table 44: Selected examples of infrequent metaphorical meanings 257 Mappings which account only for such small percentages were found for many of the target word-forms. And many word-forms are highly flexible with respect to their metaphorical meanings, such as constructed (with 17 patterns), erupted (with 15 patterns), lost (with 18 patterns), and poured (with 18 patterns). This enormous flexibility is often connected with an overall percentage of metaphorical uses which lies beyond the mean value of 41.32%. This is especially evident for erupted, where the metaphorical uses amount to as much as 72.39%. But also escape and lost are, all in all, more likely to be encountered in one of their numerous metaphorical senses than in their respective literal meanings. For escape 50.70% of all instances analyzed were metaphors, and for lost 51.00%. While some of these words are primarily linked to one specific mapping and therefore to one meaning, others seem to be more unspecified with regard to a favoured metaphorical meaning. Table 45 summarizes this for the abovementioned examples. The word-forms which exhibit a noticeable tendency to be used as instances of one or two specific mappings are listed first. Wordform MRs total (%) Number of mappings Percentage of the single mappings related to all metaphors (in descending order) constructed 43.80 17 53.20; 20.54; 5.71; 4.79; 3.88; 3.42; 1.83; 1.14; 1.14; 0.91, 0.91; 0.68; 0.46; 0.46; 0.46; 0.23; 0.23 erupted 72.39 15 38.93; 27.20; 12.27; 10.40; 4.80; 2.13; 1.07; 1.07; 0.53; 0.27; 0.27; 0.27; 0.27; 0.27; 0.27 lost 51.00 18 32.16; 19.80; 16.86; 4.90; 3.73; 3.73; 3.53; 3.53; 2.75; 1.57; 1.37; 1.37; 1.18; 1.18; 0.98; 0.78; 0.39; 0.20 poured 29.50 18 22.37; 17.29; 14.92; 10.85; 7.12; 5.42; 4.41; 3.73; 3.39; 3.05; 1.69; 1.36; 1.36; 1.36; 0.68; 0.34; 0.34; 0.34 escape 50.70 21 18.54; 14.79; 12.82; 11.83; 10.85; 6.31; 5.33; 4.54; 3.35; 3.35; 1.78; 1.58; 0.79; 0.79; 0.79; 0.59; 0.59; 0.39; 0.39; 0.39; 0.20 Table 45: The flexibility of metaphorical meanings - some examples For all these word-forms many mappings were found which are seldom used. For example, the mapping A MEAL IS A BUILDING only occurred once for constructed. It motivates 0.23% of all metaphorical uses of the form. And low numbers like this abound in Table 45. However, some of the words seem to have a striking preference for being used as instances of one or two mappings. For example, more than half of the metaphorical meanings of constructed are motivated by THEORIES ARE BUILDINGS while all other mappings are markedly less frequent. For erupted, two mappings govern most of the uses: PHYSICAL AGGRESSION IS A HOT FLUID (e.g. After sporadic violence there from the beginning of the month, full scale rioting erupted on 8 July [...], AD2-513) and DISAGREEMENTS ARE HOT FLUIDS (e.g. The simmering dispute erupted in public [...], A95-36). If this is compared to escape, it becomes obvi- 258 ous that here none of the 21 mappings governs such an overwhelming share of the uses. The most frequent mapping, i.e. AN OBLIGA- TION / LAW / CONTROL IS A PRISON accounts for only 18.54% of all metaphorical meanings, and the second most frequent, i.e. REALITY / LIFE / PROFES - SION / FATE IS A PRISON 197 for only 14.79%. The metaphorical meaning of construct seems to be much more fixed than the metaphorical meaning of escape. For constructed, there is a close association between the word-form and one specific mapping. This is not the case for escape. The metaphorical idea linked to the word is flexible by its very nature. Admittedly, there are preferences to use escape to talk about OBLIGATIONS and different aspects of LIFE , but the metaphorical meaning seems to be much less specific than that of construct. What may be entrenched in speakers’ minds is therefore not a specific mapping such as an OBLIGATION IS A PRISON or an IDEA IS A PRISON , but something more abstract like ANYTHING FELT TO BE UNCOMFORTABLE AND / OR LIMITING YOUR FREEDOM OF ACTION IS A CAPTIVITY . Such a general mapping could account for all of the mappings found for the word-form. This nicely illustrates the ability of the human mind to work with fairly abstract mappings that cut across the different frames, such as ACTION , COMPETITION , TIME , EMOTIONS , etc., into which the conceptual theory commonly organizes different mappings. For theoretical and descriptive purposes it is useful to keep the different frames apart. That the human mind is more complex than any theoretical account could ever be does not come as a surprise. But my results suggest that what seems to be stored for escape, for example, is quite a vague metaphorical meaning along the lines of ‘to avoid or get away from a potentially bad or uncomfortable situation’. But this meaning is certainly quite conventionally associated with the word, as it is realized in more than half of the cases (50.70%). For cases like constructed, the situation is different. There is a clear preference for using the word-form to talk about abstract entities, such as THEORIES , IDEAS or LANGUAGE . Other uses are possible - for example when the form is used to talk about meals - but they do not form the centre of the metaphorical meaning. Taken together, the two most frequent metaphorical meanings account for more than 70% of all metaphorical uses. However, while the underspecified figurative meaning of escape is very frequent, the most important figurative meaning of construct is less often used: When related to all meanings realized by the word-form, it occurs in less than one third of all cases. One last aspect seems to be of interest with respect to the difference between cases like constructed, which have a more specific figurative meaning, and others like escape, whose meaning is less specified: Why is a mean- 197 As the wordings of the mappings for escape indicate, even the most frequent mappings are not really uniform. Similar aspects are subsumed under one and the same heading. 259 ing which is so schematic as the metaphorical sense of escape so frequent? It does not seem to be able to convey a very definite and detailed meaning if it can be used with nearly any kind of object. The point is that it is exactly this schematic nature which makes it applicable in so many situations. The meaning may not be specific, but it invariably expresses one important idea: What we want to escape from is construed as negative. The mapping underlying the most frequent figurative uses of constructed is much richer. T HEORIES ARE BUILDINGS relies on two of Grady’s (1997, 1999) primary metaphors, i.e. LOGICAL STRUCTURE IS PHYSICAL STRUCTURE and PERSISTING IS REMAINING ERECT , 198 and thereby draws on many experiential correlations; it is a correlation metaphor. Escape, however, can be regarded as a GENERIC - IS - SPECIFIC -metaphor (Grady 1999: 86-88), where “prototypical cases [...] stand metaphorically for generic categories” (Grady 1999: 92). The implicit source domain, i.e. PRISON or CAPTIVITY , is a prototypical instance of a larger category of unpleasant, uncomfortable situations. The mapping which takes place in all different sub-cases is quite lean. It focuses almost exclusively on the fact that somebody’s freedom of action is limited which indicates that the person wants to get away and this is what eventually sanctions its application to so many different situations. The highest percentage of metaphorical uses was found - not unexpectedly - for the noun bombshell. 98.95% of all uses are metaphors. The literal meaning is here evidently much less frequent 199 than the metaphorical ones. In the majority (82.11%) of the sentences analyzed the noun is motivated by a mapping which can be formulated as UNEXPECTED NEWS IS AN OFFENSIVE WEAPON (cf. 247). The aspects which are highlighted here centre around the suddenness of the event and its unpleasant quality. The second mapping that applies and motivates 14.74% of the concordance lines analyzed is SEXUALITY IS AN OFFENSIVE WEAPON (cf. 248). The third mapping found for bombshell belongs to the common conceptualization of COMPETI- TION AS WAR , where a POWERFUL STRIKE is construed as an OFFENSIVE WEAPON (2.11%; cf. 249). The overall number of concordance lines taken into consideration is with only 95 fairly low, but still it is obvious that the metaphorical meanings have almost entirely replaced the original literal meaning. 198 Cf. also chapter 3.2.1 on Grady’s (1997, 1999) primary metaphors. The two underlying primary metaphors given here are also the reason why only some aspects of buildings can be metaphorically utilized. Corridors or rooms - as in Lakoff and Johnson’s (1980: 53) example which illustrates the use of commonly unused parts of the source (cf. chapter 4.3.1.1) - are of secondary importance when it comes to the structure of buildings and their erectness. The primary metaphors thus influence and determine which parts of a source can be used in conventional metaphorical expressions (cf. Kövecses 2002: 83-84). 199 Only one of the cases analyzed was judged as literal and in this sentence bombshell is part of a simile ([...] dropped like a bombshell; CEK-737), i.e. also not used to refer to a weapon. 260 (247) Then he dropped his bombshell - it was only available in injectable form and I just cannot stand needles. (C8A-539) (248) [...] conceived of the male fantasy in two tired and specious extremes: the blond bombshell and the whip-bearing dominatrix. (FD7-159) (249) [..] Sheridan responded with a delightful curling drive from a 20-yard free-kick to make it 2-1, Zeyer’s bombshell came too late [...] (CEP-1523) Metaphorical uses are also highly frequent for the already discussed form enriched and for one of the key examples of conceptual metaphor theory, wasting as in You’re wasting your time. In the vast majority of all cases (76.78%), the form carries a figurative meaning and in most of the cases (61.10%) it realizes the mapping TIME IS A VALUABLE COMMODITY (cf. 250). This sense of waste is therefore undeniably deeply entrenched and its figurative nature is liable to be inconspicuous. Other mappings which can motivate uses of wasting are much less conventionally applied. The next frequent mapping, i.e. TALENT / ENERGY / EFFORT IS A VALUABLE COMMODITY (cf. 251) underlies only 7.54% of all uses, PEOPLE / BODY PARTS ARE VALUABLE COMMODITIES accounts for 4.07% of the cases (cf. 252) and the other mappings are even less frequent (cf. Appendix B). (250) However, I was wasting my time [...] (A0F-267) (251) [...] that Auden was wasting his talents. (EFX-662) (252) [...] furious with her for wasting herself on a man old enough to be her father [...] (BP1-2421) A similar situation was found for core, whose 75.60% of metaphorical uses are mainly based on IMPORTANCE IS INTERIORITY - 72.50% of all uses or 95.90% of all metaphors are based on it (e.g. [...] they should grasp the core of the political problem [...]; A64-760). Beat and pointed also belong here. In both cases the total amount of metaphorical uses is with 71.70% and 70.50% respectively fairly high. 94.56% of all metaphorical uses (or 67.80% of all meanings) of beat rely on COMPETITION IS WAR (cf. 253). The seven other mappings found for beat are only of marginal importance. None of them motivates more than 1.10% of all uses of the word. Pointed only occurred as an instantiation of two different mappings. The vast majority of the total of 70.50% of metaphorical uses is governed by the mapping THE MIND IS A BODY : REASONING IS FOLLOWING A PATH : IDEAS ARE GUIDES TO OUR REASONING . 95.04% of all metaphorical meanings (67% of all meanings) build on this idea (cf. 254). The construal IDEAS ARE LOCATIONS : REASONING IS FOLLOWING A PATH : FACTS ARE GUIDES TO OUR MOVEMENTS only accounts for 4.96% of all metaphorical uses (3.50% of all uses). Waste, core, beat, and pointed are very good, perhaps even prototypical examples of ‘metaphors we live by’. (253) ICL AND HOSKYNS BEAT COMPETITON FOR DSS CONTRACT (CNK-222; original caps) (254) Both the idea of politics as a collective activity […] and the idea of politics as rational, regulated government as opposed to the arbitrary unpredictable rule of despots pointed into the direction of democracy. (EVP-273) 261 Metaphor is undeniably a pervasive cognitive mechanism. It results in highly frequent, very conventional metaphorical meanings, as for example in cases like waste or core, but is also often applied in more creative ways. The latter becomes evident in more uncommon mappings like THE H OLY S PIRIT IS A FLUID (e.g. the Holy spirit was poured out; C8L-1315; cf. Table 44 above) - an example whose usage is probably restricted to religious texts. But also well-known mappings can only motivate a relatively small amount of the metaphorical uses of a given word-form. T IME IS A VALUABLE POSSESSION , for instance, only accounts for 3.50% of all metaphorical instances of lost. Importantly, for some of the word-forms under consideration, the typical concrete-to-abstract mappings sometimes have to compete with mappings which have a concrete target domain. Erupted, for example, is more likely to be used in a metaphorical construal of the type PHYSICAL AGGRESSION IS A HOT FLUID (38.93% of all metaphorical meanings) than as an instance of EXTREME EMOTIONS ARE HOT FLUIDS (12.27% of all metaphorical meanings). And, not astonishingly, also leaner mappings can result in high relative frequencies of metaphorical uses, i.e. are quite conventional. This holds true for the already discussed example bombshell, but to a certain extent also for sharks. The total of the metaphorical uses is much lower here (21.55%), but the metaphorical meaning resulting from the mapping COM- PETITORS OR ANTAGONISTS ARE PREDATORS does not have to compete with any other possible metaphorical construal. To conclude this section, I would like to briefly comment on some cases of metaphorical meanings with medium relative frequencies ranging from 30% to 45%. These are summarized in Table 46. Wordform Mapping M - total (%) grasped IDEAS ARE OBJECTS : UNDERSTANDING / ADOPTING IS GRASPING 42.37 embraced BELIEFS ARE LOVERS : ACCEPTING A BELIEF IS ENTERING A RELA- TIONSHIP 42.09 profited HAVING BENEFICIAL THINGS HAPPEN TO ONE IS GETTING MONEY 39.39 invites CAUSATION IS CONTROL OVER AN OBJECT : CAUSING MEN - TAL / VERBAL REACTIONS IS ALLOWING MOTION OF POSSESSIONS 36.76 attacked ARGUMENT IS WAR 35.30 crooked BEING IMMORAL IS BEING CROOKED 34.72 digest IDEAS ARE FOOD : UNDERSTANDING IS DIGESTION 32.39 crack A PROBLEM IS A LOCKED CONTAINER FOR ITS SOLUTION : SOLVING A PROBLEM INVOLVES OPENING AN OBJECT 31.82 key A PROBLEM IS A LOCKED CONTAINER FOR ITS SOLUTION : MEANS TO SOLVING IS MEANS TO OPENING 31.80 hollow CONDUIT METAPHOR : A WORD / SOUND / BOOK / ACTION / RITU - AL / VICTORY IS A CONTAINER : THE CONTENT IS CONTAINED IN THE STIMULUS 31.61 Table 46: Metaphorical meanings with relative frequencies between 30% and 45% 262 These neither infrequent nor particularly frequent instances are typically brought about by stock examples of metaphorical mappings, such as IDEAS ARE FOOD (e.g. I always advise musicians to digest at least a certain amount of knowledge about their business affairs; A6A-2308) or the CONDUIT METAPHOR (e.g. a hollow gesture; CKU-524). The metaphorical uses of invites rely on a very abstract mapping, where, for example, the CAUSATION OF MENTAL RE- ACTIONS is depicted as ALLOWING THE MOTION OF POSSESSIONS RELATIVE TO AN AFFECTED PARTY (e.g. […] constructivism invites a developmental perspective on thinking in the following sense; A0T-439). The only somewhat less rich mapping that occurs in this group is BEING IMMORAL / DISHONEST IS BEING CROOKED (e.g. Crooked PC robs corpse; CH2-3515). This section has shown that many of the examples often found in the literature on conceptual metaphor are indeed quite conventional. On the whole, metaphorical uses are more frequent than metonymic ones. From a psychological perspective, this means that a comparatively high number of metaphorical meanings are very likely to be entrenched, because speakers encounter them often in everyday discourse. Apart from this, mainly three aspects are evident: First, the enormous flexibility of metaphorical thinking and speaking, which is reflected in the high amount of metaphorical meanings with very low relative frequencies. Second, some words are preferably used to convey a specific metaphorical meaning. And third, if one and the same mapping can be instantiated by various linguistic surface expressions, the resultant relative frequencies of the metaphorical meanings are not necessarily similar. This means that the speakers sometimes favour certain verbal means to express a given metaphorical idea. With regard to the mapping ARGUMENT IS WAR , for example, it can be stated that attacked (35.30% of all meanings) is much more often used to realize this construal than, for instance, lost (0.20% of all meanings). Particularly the last aspect is related to pragmatic factors, i.e. to what parts of an event or situation speakers commonly focus on and talk about. Since the description of such regularities presupposes a much more detailed analysis of the data than can be given in a preliminary overview, this issue will be taken up again in the following section, where a more systematic of the results of the metaphor study is provided. 7.2 Frequent and infrequent metaphorical mappings and meanings All in all, the results of the metaphor study allow much less generalizations than the results for the metonymies. On the one hand, the number of wordforms studied was smaller in this case, so that the scope was more restricted right from the beginning than for the metonymy study. On the other hand, and more importantly, the metaphorical uses of lexical items are often extremely diverse. 263 Nevertheless, some tendencies could be observed, for example, regarding the infrequent metaphorical usage of some of my targets. The wordforms which returned the smallest number of metaphorical uses (with less than 10%) include those listed in Table 47. Word-form Percentage of total metaphorical uses wounded 8.18% smaller 7.50% shopping 5.77% died 5.50% sell 4.60% killed 3.90% electricity 1.50% buy 0.80% Table 47: Word-forms with less than 10% of metaphorical uses As becomes apparent in the low overall percentages of metaphorical usage for these forms, their metaphorical meanings can hardly be called deeply entrenched. For killed the 3.90% of metaphorical meanings are supported by ten different mappings. The largest group, which is based on FAILURE IS DEATH (cf. 255), only accounts for 1.10% of all meanings. And died has been found as instantiating 13 different mappings. The most frequent one is A TRADITION / TREND / HABIT IS A LIVING BEING (cf. 256) and occurs in only 1.00% of all cases. The other forms realize less different classes of mappings. Buy, for example, has only two ( BELIEFS ARE COMMODITIES : ACCEPTING A BELIEF IS BUYING IT and CAUSATION IS COMMERCIAL TRANSACTION : CAUSING BENEFICIAL EFFECTS IS GIVING MONEY ; cf. 257, 258), just like sell ( IDEAS ARE VALUABLE COMMODITIES : ARGUING FOR AN IDEA IS TRYING TO SELL IT and PEOPLE ( AND THEIR ABILITIES / CONCERNS ) ARE VALUABLE COMMODITIES ; cf. 259, 260). Compared to the cases mentioned above (chapter 7.1) where a large amount of the instances is clearly motivated by one specific mapping, i.e. which have one preferred metaphorical meaning and where this meaning is sometimes even more often realized in natural language use than the literal meaning, the potential metaphorical senses are much less entrenched and conventional. (255) Stock set a total of 166 for 8, with Taylor’s 52 the key, and visitors Danbury were killed with 73 runs to spare. (256) [...] but like many other traditions it seems to have died. (H09-1295) (257) I didn’t buy the idea of a Reagan mandate then [...] (EAY-1396) (258) Agreeing to her suggestion of a ‘meeting’ would buy them time [...] (AD9-3955) (259) But for larger events you have to ‘sell’ them - metaphorically if not literally. (ADK-728) (260) But what better, what better way to sell yourself? (KGN-438) 264 There seems to be a certain regularity underlying the relative infrequency of the metaphorical uses of some of these words. Two groups of words belonging to two separate semantic fields seem to be particularly resistant to metaphorical meaning extensions. The first field centres around COM- MERCIAL EVENTS and is represented by buy, sell, and shopping in my data, the second one covers BODILY HARM . I will discuss the two in turn. Why is vocabulary belonging to the COMMERCIAL EVENT frame (buy, sell, shopping) so rarely used in metaphorical ways? I DEAS and BELIEFS are frequently conceived of as OBJECTS (cf. Appendix B). Since ideas can be possessed it should also be possible metaphorically to sell them, buy them or shop for them. And this is indeed the case, but IDEAS and BELIEFS are not conventionally construed in this way. The conceptualization of IDEAS and BELIEFS as goods that can be sold and bought is rather infrequent. A possible explanation is that the idea that abstract mental entities can simply be bought and sold like any other commodity is not entirely in line with the invariance principle (cf. Lakoff 1990). A typical commercial event includes a buyer, a seller, and goods (cf. Fillmore 1977). In addition, it is strongly associated with money which is transferred from buyer to seller in return for the goods. And usually the act of buying something is essentially determined by the buyer’s volition. If someone has enough money, they can buy anything they like. The metaphorical construal can possibly neglect the absence of one crucial element of the COMMERCIAL EVENT frame, i.e. the money. After all, metaphorical hiding is not uncommon. However, the metaphorical sense of buy is not always a neutral one. It often conveys the connotation that the idea to be sold by one person and bought by another is not true. The LDOCE4 (s.v. buy 1 , 3) defines this meaning as ‘to believe something that someone tells you, especially when it is not likely to be true’ and marks it as informal. Both factors, the restriction to more informal situations as well as the connotation of ‘falsehood’ inherent in the metaphor, are responsible for its somewhat limited applicability. And due to the quite specific meaning the metaphor carries, it shows a marked tendency to be used in negated sentences or phrases. The LDOCE4 provides such an example (‘Let’s just say it was an accident.’ ‘He’ll never buy that’) and also the few instances of this metaphorical construal found in the BNC are typically negative (e.g. But I don’t quite buy that, C9L-755; I didn’t buy the idea of a Reagan mandate then; EAY-1396). The metaphorical meaning of sell is less specific: ‘to try to make someone accept a new idea or plan, or to become accepted’ (LDOCE4, s.v. sell 6). But it is likely that ideas which are not convincing enough to be accepted due to their credibility and therefore have to be actively ‘sold’ will often be seen as ideas that one would have to ‘buy’, i.e. as ideas that are potentially untrue. To be metaphorically shopping for an idea is slightly more common than just buying an idea (5.77% vs. 0.80%). This may be sanctioned by the fact that if one actively searches for new ideas, as the shopping metaphor implies, one accepts that one will come across better and worse goods, i.e. 265 ideas. The purpose of the shopping tour is, after all, exactly to inspect what is on offer, to compare the quality of the different items, and eventually to decide for the best of them. The second field, BODILY HARM , is represented in Table 47 on the one hand by the form died, which implies the highest possible cost for living beings, and on the other by the grammatical forms wounded and killed. For all of them, the literal meanings are clearly preferred over any metaphorical extensions. Interestingly, however, other lexical items which can also be considered to belong to the larger field of potential BODILY HARM are much more likely to be used figuratively. Cases in point are beat with a total of 71.70% of metaphorical uses and attacked with a total of 43.80%. Why are these forms so much more often used as metaphors than killed and wounded? It appears that the probability of metaphorical meaning extensions is related to the specificity and richness of the literal meanings of the verbs. Table 48 provides literal meaning glosses for the base forms of the different verbs, connects them to the effects implied by the verbs and once more provides the relative frequencies with which the forms investigated convey metaphorically extended meanings. Word-form: beat attacked wounded killed Literal meaning of base form: ‘strike repeatedly and violently’ ‘take aggressive action against s.o. or s.t. with weapons or armed force’ ‘inflict an injury on someone, typically one in which the skin is cut or broken’ ‘cause the death of a person or animal’ Effects implied: lasting effects not necessary lasting effects more likely due to use of weapons lasting effects highly probable definite lasting effect; irreversible Percentage of metaphorical uses: 71.70% 43.80% 8.18% 3.90% Table 48: Metaphorical uses of lexical items belonging to the field of POTENTIAL BODILY HARM (literal meaning glosses based on the NODE) The meaning of kill is more specified than that of wound and much more specific than the meanings of attack or beat. In other words, kill presupposes a previous attack, which leads to wounds, which in turn can cause death. Wounding someone in a literal way just entails that they have been attacked beforehand. Beat falls somewhat outside of this sequence of physically violent actions and is just an optional but not a necessary part of them. On the one hand, the literal meaning of kill is quite rich, because it can be said to elicit a whole series of actions in speakers’ minds. On the other hand, it is - in terms of the effects of these actions - extremely defi- 266 nite. In contrast to the other verbs, the action described by kill has irreversible consequences. This is less so for wounded. What wounded and killed have in common, however, is that they both express the idea of accomplishment, whereas beat and attacked lack this semantic component; they express actions with more or less vague consequences. Therefore, the less rich and less specific verbs are more likely to be used in a metaphorical way. The idea of irreversibility is certainly a very strong one for kill and is hard to cancel in a metaphorical construal, i.e. has to be mapped onto the target domain. This severely restricts the situations in which it can be used metaphorically. This assumption is strengthened by the fairly low amount of metaphorical uses of die (5.50%), which lacks the component of volitional action, but shares with kill the utter irreversibility it implies. For the much less specific and rich attack, and perhaps even more so for beat, it is much easier to map the decisive parts of the literal construal onto another domain of experience. Both only convey the idea that some harm has been inflicted or at least intended, which can easily be transferred to describe more abstract situations, like ARGUMENTS or COMPE- TITION . This general tendency is to some extent also reflected in my results concerning the mapping COMPETITION IS WAR . This mapping and some of its related sub-mappings, such as LOSING IS BEING WOUNDED , motivate uses of attacked, beat, and wounded, but also of defences, bombshell, and shattered. In accordance with what has been pointed out above, beat (cf. 261) has the highest relative frequency here. Attacked is the second most frequent verb belonging to the field of BODILY HARM motivated by this mapping, but it is used in this sense in only 2.80% of all cases. On the whole, compared to the high relative frequency of COMPETITION IS WAR meanings of beat, all the other targets considered are infrequently used to convey this idea. Especially infrequent, however, are shattered (cf. 262) and wounded (cf. 263). (261) It narrowly beat much bigger rival supermarkets group J Sainsbury to the top slot, and outshone Tesco. (CEL-24) (262) In fact Manawatu shattered the Irish with their 58-24 win [...] (CB3-1268) (263) ‘Many [i.e. textile manufacturers; S.H.] will die in the struggle, even more will be cruelly wounded.’ (J0P-994) This is summarized in Table 49 (below). The conceptual mapping is certainly quite conventional. This is evident in the fact that it motivates metaphorical uses of many different source vocabulary items which can be assigned to the wider field of physical aggression. But there are clear preferences for a construal of WINNING A COMPETITION as BEATING , as opposed to SHATTERING or WOUNDING . Example (263) demonstrates fairly well how figurative this construal is felt to be. Shatter, for example, has quite a specific and rich meaning, which leads to many metaphorical entailments: A company or a sports club that has been shattered is construed as forever bearing marks of this damage, as probably never being the same as before 267 the event. This is, however, often not really the case: Companies can ‘recover’ and sports clubs as well, the latter often quite soon, they just have to win the next match. Even though the meaning of wound is also fairly specific, it is not very informative if applied in a COMPETITION IS WAR mapping: Similar to attack (which is also relatively infrequent as an instance of COM- PETITION IS WAR ) it does not lead to unambiguous consequences in terms of who is going to win the metaphorical battle. Beat, on the other hand, comes with a leaner meaning but has the advantage of depicting the situation at hand in a very precise way: the one who is beaten is construed as the one who is inferior in the competition, and the one who beats is the one who is superior. Word-form MRs total (%) M - total (%) M - MRs (%) beat 71.70 67.80 94.56 defences 49.10 8.10 16.50 attacked 43.80 2.80 6.40 bombshell 98.95 2.11 2.13 shattered 59.00 0.90 1.53 wounded 8.18 0.44 5.33 Table 49: The results for COMPETITION IS WAR The verb attack is, however, very often used when an ARGUMENT is metaphorically construed as a WAR . In general, one can observe a tendency to metaphorically highlight only some parts of the VERBAL ARGUMENT frame (cf. Table 50). Word-form MRs total (%) M - total (%) M - MRs (%) attacked 43.80 35.30 80.60 defences 49.10 23.80 48.47 killed 3.90 0.20 5.13 lost 51.00 0.20 0.39 wounded 8.18 0.11 1.33 beat 71.70 0.10 0.14 Table 50: The results for ARGUMENT IS WAR The prototypical structure of a verbal dispute consists of a to and fro of single utterances made by two speakers. And these utterances can either be meant to offend the other party’s position or they can be intended to support one’s own point of view. This is reflected by the fact that the two words which stand for the offensive utterances, i.e. attack, and for the supportive ones, i.e. defend (here realized in its nominal form defences), respectively have the highest relative frequencies. Other aspects, such as beating (e.g. His abuse beat her in the face [...]; HTN-1850) or wounding the opponent (e.g. [...] that her words had wounded him; HNJ-948) are dispreferred. Also of interest is that the outcome of a verbal argument in terms of destroying the 268 enemy (killed) or losing the battle (lost) is hardly focused on. This is remarkable since the result should be of paramount importance for the opponents. But since my sample does not include other possible outcomes, such as winning, this assumption needs to be verified by further empirical investigations. A mapping which is quite conventional as far as the central source vocabulary is concerned is DESIRE IS HUNGER . My sample includes two of the main lexical items in this field, appetite (cf. 264) and thirst (cf. 265). (264) However, he soon discovers that she has a voracious sexual appetite, which he finds almost impossible to satisfy. (C9U-1620) (265) [...] and is aghast when there is such a thirst for education all around him. (A0U-1799) Both are frequently used as metaphorical expressions motivated by DESIRE IS HUNGER . By far the majority of the metaphorical uses of thirst belong here (91.26%). 27.41% of all meanings of thirst are motivated by this mapping. This number includes the special case IDEAS ARE OBJECTS : LEARNING IS AB- SORBING . For appetite the picture is even clearer: 26.69% of all uses instantiate DESIRE IS HUNGER : GETTING SOMETHING CONCRETE / ABSTRACT IS EATING (e.g. [...] appetite for power; A0K-540) and 4.78% are supported by the mapping DESIRE IS HUNGER in the larger domain of EMOTIONS (cf. 264). An additional 10.02% of all meanings is based on IDEAS ARE OBJECTS : IDEAS ARE FOOD : INTEREST IS APPETITE (e.g. Stanger has an appetite for apparently lost causes [...]; A9R-16). If INTEREST is construed as DESIRE , all metaphorical instances of appetite found belong to DESIRE IS HUNGER . This case also demonstrates why it is so tricky to make definite statements about the conventionality of single mappings and their instantiations. There is no exhaustive list of metaphorical mappings available and the existing categorizations all work with many, perhaps inevitable, crossclassifications. I NTEREST can be regarded as a type of DESIRE , but the INTER- EST IS APPETITE mapping is usually treated as an entailment of the more general IDEAS ARE OBJECTS and IDEAS ARE FOOD metaphors. The classical case of DESIRE IS HUNGER is, however, not concerned with IDEAS , but with EMO- TIONS . Still, INTEREST IS APPETITE is not only related to IDEAS ARE FOOD , but also to DESIRE IS HUNGER . The relationships between metaphorical mappings are often so complex that it is difficult to account for all of them. If INTEREST IS APPETITE is included in the DESIRE IS HUNGER group, other instances of IDEAS ARE FOOD , like UNDERSTANDING IS DIGESTION , should also be related to it. But here, the link to DESIRE IS HUNGER is not so evident. The chain of metaphorical entailments would be quite a long one, leading from DESIRE IS HUNGER , which is connected to IDEAS ARE OBJECTS and IDEAS ARE FOOD , over INTEREST IS APPETITE to the final UNDERSTANDING IS DIGESTION . If this view is accepted, the high relative frequency of digest as an instance of the mapping UNDERSTANDING IS DIGESTION (32.39% of all meanings; e.g. [...] more time to digest the information; CAP-811) lends further support to the 269 above-mentioned figures for appetite and thirst and therefore to the claim that DESIRE IS HUNGER is quite a conventional mapping. That a mapping with HUNGER as source domain should be conventional does not come as a surprise, considering the very firm experiential basis of this domain and the highly abstract nature of the targets to which it is connected. In accordance with the predictions of the conceptual metaphor theory, the domains which are most frequently construed by way of metaphors are highly abstract ones. A fairly wide range of my source vocabulary items are used to convey metaphorical construals of IDEAS : 200 I DEAS ARE OBJECTS , which can be constructed (2.10% of all meanings; cf. 266), grasped (42.37%; cf. 267), pinned down (4.52%; cf. 268), and lost (1.40%; cf. 269). As VALUABLE OBJECTS one can also sell (3.00%; cf. 270) them. The OBJECTS can be CONTAIN- ERS FOR MEANING and therefore be hollow (0.94%; cf. 271) and the CONTAIN- ERS can be hard to open (impenetrable 9.63%; cf. 272). I DEAS can also be HID- DEN OBJECTS one can throw up (15.85%; cf. 273). (266) Indeed, he proposed a two-stage process in which individuals’ identities are constructed. (HRM-132) (267) He has grasped the neat theological and ontological crux implicit in the captain’s statement [...] (A18-922) (268) Dostoevsky owed a lot to Molière who is to be met in the notebooks but not to be pinned down in the major fiction. (A18-1451) (269) [...] so that he lost the thread of the story Ratagan was telling him [...] (GWF-383) (270) Of course it still has to sell the idea to the management [...] (CTS-142) (271) The thought had a hollow ring to it. (FAB-2177) (272) [...] and some important issues will be clarified at a simple, accessible level which might otherwise remain impenetrable if dealt with in their modern context. (EVA-380) (273) This could be relied on to throw up ‘bad’ as well as ‘good’ factors. (EC5- 419) As already discussed, IDEAS can be FOOD (appetite 10.02%, digest 32.39%), but IDEAS can also be construed as LOCATIONS , when REASONING is conceived of as FOLLOWING A PATH (crooked 0.64%, followed 0.20%, pointed 3.50%), and they can as well be GUIDES along this path (pointed 67.00%). Or, if one cannot get rid of IDEAS anymore, they are PRISONS (escape 6.50%). Furthermore, IDEAS can be FLUIDS (poured 0.90%) and ANIMATE BEINGS with a life cycle, which can cease to live (died 0.30%) or be actively killed (0.90%). They can grasp a person (0.16%), and have to be nourished (10.00%). As this overview demonstrates, the relative frequencies of the single construals are far from uniform. I DEAS ARE LOCATIONS : REASONING IS FOL- LOWING A PATH does not belong to the favoured mappings as the low relative frequencies of crooked, followed and pointed as instantiations of this mapping show. However, if the IDEAS are conceived of as GUIDES TO OUR REASONING at the same time, as instantiated by pointed (67.00%), the map- 200 For an overview of metaphorical construals of IDEAS , cf. Schmid (1993: 170-174). 270 ping is widely applicable. Also infrequent are construals in terms of VALU- ABLE POSSESSIONS which can be lost or sold. A much better way to approach IDEAS metaphorically is in terms of OBJECTS which can be hidden and have to be found, and which can be understood by way of grasping them or digesting them. Another interesting observation is that, even though one might think that the two forms attacked and defences are prime examples to illustrate the conventionality of the mapping ( TREATING AN ) ILLNESS IS ( FIGHTING A ) WAR , they do not frequently occur as instantiations of this mapping. While defences is with a percentage of 5.00% used to refer to the immune system or precautions against a potential disease, the construal of an infection or a disease as an attack is, at least in a verbal form, quite rare: only 0.90% of all meanings of attacked are motivated by this mapping. The results would have been different, however, if the corpus study had included the noun attack as well, for example due to the existence of the institutionalized compound heart attack. A search for this lexeme in the BNC returns 644 hits, the singular noun attack occurs 7518 times in the BNC. If only heart attack is taken into consideration, the uses relying on the mapping ILLNESS IS WAR amount to 8.57%, and one can safely assume that this is not the only usage of the noun attack motivated by this mapping. Thus, the overall frequency is probably higher than that. This illustrates that different grammatical surface forms can differ in terms of how likely they are to instantiate a given mapping. In contrast to the noun, the verb attack is much more associated with a metaphorical construal of a VERBAL ARGUMENT as a WAR . Let me next turn to some examples which deviate from the usual concrete to abstract direction of mapping, i.e. which involve concrete sources as well as concrete targets. Some metaphorical uses of beat, died, erupted, hollow, and sharp, for example, describe weather conditions, or on a more general level the CLIMATE . The CLIMATE can be an ADVERSARY (beat), a LIVING BEING with a life cycle (died), a HOT FLUID / VOLCANO (erupted). More specifically, WIND can be a CONTAINER (hollow) and COLD a CUTTING INSTRUMENT (sharp). Compare Table 51, which summarizes the relative frequencies of these uses. That they are all quite low is in line with the assumption that concrete-to-abstract metaphorical construals are more useful than concreteto-concrete ones. Word-form MRs total (%) M - total (%) M - MRs (%) beat 71.70 0.90 1.26 sharp 73.30 0.60 0.82 erupted 72.39 0.39 0.53 died 5.50 0.20 3.64 hollow 45.91 0.16 0.34 Table 51: Some metaphorical conceptualizations of CLIMATE 271 But also mappings which are much more often discussed in the relevant literature on conceptual metaphor can regularly result in comparatively unimportant metaphorical meanings. M ACHINES ARE PEOPLE , for example, was found as motivating metaphorical uses of six of the 50 word-forms investigated, but none of the six forms is very likely to be encountered as an instantiation of this mapping, as shown in Table 52. Word-form MRs total (%) M - total (%) M - MRs (%) thirst 30.03 2.33 7.77 infected 10.50 2.30 21.90 digest 34.82 0.81 2.33 died 5.50 0.40 7.27 attacked 43.80 0.20 0.46 wounded 8.18 0.11 1.31 Table 52: The results for MACHINES ARE PEOPLE The only two aspects of humans which are mapped from PEOPLE onto MA- CHINES in these examples are HEALTH and FOOD and its digestion. The most common ideas conveyed are that cars have a thirst for fuel (ABS-203) and that A DESTRUCTIVE CODE IS A VIRUS (e.g. an infected MS-DOS disk; CWM-437). The latter has at least a certain overall relevance for the metaphorical meanings potentially conveyed by the form infected, since more than 20% of all metaphorical meanings of the form rely on it. For all the other word-forms given in Table 52, the mapping MACHINES ARE PEOPLE does not determine such a considerable share of the metaphorical meanings. Far from all mappings which can be observed in actual language use and could therefore be possibly posited as conceptual metaphors lead to really conventional linguistic metaphors. In fact, as already pointed out in chapter 7.1 (cf. also Appendix B), many of the examples found in this study are not very likely to be entrenched. Some metaphorical uses, like (274) to (276), are highly creative and poetic: (274) The short polar day died in bloody shadows. (FP0-593) (275) [...] you’re a bright star to march upon, and I’m setting a steady pace. (AMC-1309) (276) She had changed into a gown she seemed to have been poured into, and promised to pop out of. (FPX-366) Note that there are some important differences between these examples. While (275) is connected to the not unconventional mapping of GOODNESS IS LIGHT (bright 17.40%) and HOPE as a LIGHT SOURCE and is therefore only novel on the linguistic level, the other two lack any conventional mapping motivating them. D AYS are not commonly construed as BEINGS WITH A LIFE 272 CYCLE . 201 Rather, it is the other way round and A LIFE CYCLE is construed as a DAY . And while it is not too uncommon to regard an UNDETERMINED AMOUNT OF PEOPLE as a FLUID (6.20% of the 6.60% indicated in Appendix B for poured relying on PEOPLE / COMPANIES / A PERSON ARE / IS A FLUID ), example (276), where a SINGLE PERSON is construed as a FLUID is much more figurative (0.40% of the mapping). The unconventionality of the latter is not astonishing, considering that the mapping involves a reversal of a typical ontological metaphor by conceiving of a PERSON as a SUBSTANCE and the GARMENT as a CONTAINER . Not only metaphorical meanings that were first encountered during the analysis, but also figurative meanings taken over from the sources are not all equally conventional. Consider the form erupted once more. 202 It only occurs in 8.88% of all cases in the sense motivated by the mapping EXTREME EMOTIONS ARE HOT FLUIDS . Taking the popularity of the mapping ANGER IS A HOT FLUID in the literature on conceptual metaphor into account, this is not high, and even more so since I already phrased the mapping in a somewhat more general fashion to embrace not only anger, but any other type of extreme emotion. But the idea that extreme emotions (such as anger, excitement, fear, and anxieties as well as passion and joy) erupt is more common than that they are simply poured out (4.40% of all meanings). The latter verb is not able to transport the power that is ascribed to emotions by the former. As this indicates, the results are not necessarily valid for the mappings as wholes, their validity is restricted to the word-forms under consideration. This will be discussed in more detail in the following section. 7.3 Preferred metaphorical construals An example illustrating that some linguistic items are more strongly associated with a given metaphorical mapping than others is electricity. The word was taken over from the Master Metaphor List as an instance of the mapping DESIRES ARE FORCES BETWEEN DESIRED AND DESIRER : SEXUALITY IS AN ELECTROMAGNETIC FORCE BETWEEN DESIRED AND DESIRER , as in [...] she felt the stirrings of electricity flicker between them again (H8S-1794). Only 1.20% of all concordance lines analyzed realize this meaning. This could be taken to indicate that the source domain of ELECTROMAGNETIC FORCES is perhaps not experientially familiar enough to have much explanatory value. Other linguistic instantiations of this mapping are, however, undeniably more familiar. The words attract, attraction, and attractive demonstrate this. They were not part of my corpus study, but the LDOCE4 indicates that these 201 The metaphor occurs, but seems restricted to more poetic text-types, which intentionally exploit unconventional metaphorical conceptualizations. An instance is the wellknown hymn “Day is Dying in the West” (lyrics by Mary Artemesia Lathbury). 202 For an overview of the different senses of erupted found in the study, cf. chapter 5.2.2.2. 273 metaphorical senses are more common than their literal counterparts. The LDOCE4 lists the different senses of the lemmas included in the order of their respective frequencies and for attract as well as attraction the literal senses are the last ones mentioned. And attractive is only linked to the metaphorical idea and not at all to the literal one of electromagnetic forces. Speakers thus have obvious preferences for certain lexical items in conveying a given metaphorical mapping. That some metaphorical conceptualizations are preferably expressed with certain linguistic means is connected to how we perceive the target, i.e. to which parts of the target are supposed to be highlighted. T IME IS A VALUABLE COMMODITY , for example, is more often realized by wasting (61.10%) than by lost (1.80%). The idea that we waste time in a careless manner is more common than that we more or less simply lose it by accident. Even though the result of both wasting and losing is the same, i.e. the non-possession of the desired object, the idea of wasting fits the purposes at hand better then the idea of losing. That the SELF IS A BRITTLE OBJECT is evidently conveyed best by using a verb which clearly indicates the fragility of the target. Shattered is therefore favoured here (16.10% of all meanings), since the picture it evokes is the most detailed one. Crack is less apt and is therefore not so often used to realize the mapping (9.09%). Attacked, the third word-form which was found to instantiate the mapping, is probably too unspecific with regard to the consequences to be a good way of conveying the meaning. For other mappings the preferences for a single metaphorical construal are more likely to be a mere reflection of the noteworthiness of different aspects of the targets. Consider the CONDUIT metaphor. That LINGUISTIC STIMULI are constructed as CONTAINERS and their MEANING as the CONTENT is well-known. Remarkable features of this metaphorical relationship are, for example, that the meaning can be hard to decode, which can be conveyed by the lexeme impenetrable (19.79% of all meanings), or that the content is weighty, which can be transported by the word-form loaded (11.48% of all meanings). Due to these preferences, some lexical items have acquired strong associations with specific metaphorical target domains. The verb crack as well as the noun key are both heavily biased towards the target domain of PROB- LEMS . Crack occurs as a realization of the mapping A PROBLEM IS A LOCKED CONTAINER : SOLVING IS OPENING in 31.82% of all cases, and key transports this meaning in 31.80% of all cases. The conceptualization of problems as locked containers is quite a common one and, not astonishingly so, as lexemes which are apt to convey that problems can be solved with their metaphorically extended meanings abound in these senses. What we focus on when we are facing a problem is its solution, hence this close association. All in all, the conventionality of metaphorical meanings of single lexical items is higher when they pick out central aspects of the target domains. What is vital for us when coping with ideas, is that we understand them. 274 This is reflected in the fairly high frequency with which grasped is used as an instance of the mapping IDEAS ARE OBJECTS : UNDERSTANDING IS GRASPING (42.37%). When we construe a VERBAL DEBATE , we tend to focus on the basic aspect of exchanging arguments (e.g. attacked 35.30%). And when we compete with others, we are most interested in the outcome, i.e. in which of the two parties involved beats the other (beat 67.80%). This is in largely line with Lakoff and Johnson’s (1980: 53) original description of conventional conceptual metaphors (cf. chapter 4.3.1.1): The degree of conventionality of linguistic manifestations of metaphorical thinking depends on whether they single out parts of the source domain that are commonly used. And the commonly used parts are those which form vital aspects of our knowledge of the source and can, in addition, be usefully applied to describe the target. However, high relative frequencies of occurrence were not only found for the rich mappings which are usually in the centre of attention of cognitive linguists. Some instances are remarkably frequent, given that they do not rely on multiple correspondences between source and target, but are rather lean. This is particularly true for two of the word-forms under consideration, i.e. sharks and bombshell. Instances like these owe their success mainly to the fact that they transport a lean, but at the same time extremely unambiguous meaning. In other words, they carry one especially salient attribute and this is why they are frequent. The next sections discuss the role of attribute salience for the degree of conventionality displayed by such cases. 7.4 Metaphors and attribute salience: Emergent structure and conventional metaphorical meaning 7.4.1 The relativity of attribute salience Attribute salience is, in principle, a relative notion. The degree of salience of a given attribute is not entirely fixed, but flexible and changes with its context. It can be relative to a given speech situation or to the speakers and their common ground (cf. e.g. Gibbs and Gerrig 1989) and it can be relative to the linguistic context. The latter aspect is also emphasized by Barsalou (1982) in his distinction of context-dependent and context-independent properties (cf. also Schoen 1988) of concepts. The former are always accessed when a concept is activated, while the latter are more peripheral and become only salient in strong contexts. This distinction is of significance with regard to metonymies, as it accounts for the difference of the conventionality of metonyms with a salience-upholding emergent structure in contrast to the unconventionality of those with a salience-changing emergent structure (cf. chapter 6.3.2). However, it is even more essential for the description of attribute salience with regard to metaphors. The aspects mapped 275 from a source onto a target are determined by the target (cf. Shen 1991: 111). 203 That the perceived similarity between pairs of entities varies when they are presented in different groups was already discussed by Tversky (1977: 342) in his diagnosticity principle. The point is that the diagnostic value of features changes with the ground of the comparison. Similarly, the salience of the attributes of a given concept can change when this concept is metaphorically linked to different other concepts (Ortony et al. 1985: 570). In a neutral context, such as I met my butcher the other day, the concept BUTCHER , for example, would presumably not have a particularly negative connotation. Since most professional butchers are quite competent in what they do, the concept does not carry a salient attribute ‘incompetence’. This attribute comes to the fore only when a linguistic metaphor establishes a conceptual link between the BUTCHER concept as a source and another concept of a professional group as a target (e.g. My surgeon is a butcher, This pianist is a butcher, etc.; cf. chapter 3.3.2). The concrete form of the incompetence is then also dependent on the target concept: For surgeons it has to do with blood, injuries and potentially with death, while for pianists one could imagine a person who performs the music without any sensitivity, too loudly, and without any subtlety (Glucksberg, Gildea and Bookin 1982: 95). 204 The fact that the salience of attributes is not stable, but can vary, is thus essential for the description of the conventionality of both metonymy and metaphor. For the latter, it is of particular relevance to explain why some lean mappings, “which focus on a single property of a source concept” (Ungerer and Schmid 2006: 126), are so highly frequent and conventional, even though they cannot grant as many insights as richer mappings do. 7.4.2 Lean and rich mappings and their attributes The effortless understanding of metaphors, especially of more uncommon expressions, depends on the salient properties or attributes of the vehicle term or source domain. Compare the following examples: (277) An insult is a sheet of paper. (278) An insult is a razor. (after Ortony et al. 1985: 572) Both examples instantiate a more general mapping of the form WORDS ARE CUTTING INSTRUMENTS . But (278) is much better in transporting this idea than (277). It is a highly salient attribute of the concept RAZOR that they are cutting instruments and can cause pain. Sheets of paper can also cause 203 In a way, this is the mirror image of the invariance principle (cf. e.g. Lakoff 1990; Brugmann 1990) of conceptual metaphor theory, which states that any metaphorical mapping must uphold the image-schematic structure of the source domain. 204 This view is corroborated by the fact that one and the same source term can lead to unambiguous interpretations when linked to one target, but not when linked to another (cf. chapter 4.4.2). 276 deep cuts, but this is not what is foremost in our minds when we hear the expression sheet of paper. In Ortony et al.’s (1985: 572) terminology, the attributes the speaker wants to highlight with respect to the a-term (i.e. the target) are more salient and typical of the b-term (i.e. the source) than of the a-term itself. It is only in this way that the metaphor can be informative 205 (cf. also Chandler 1991; Glucksberg 1989). Linguistic metaphors which follow this principle by using source concepts that either carry contextindependent salient attributes or attributes which are foregrounded by way of the metaphorical expression itself should be preferred by the speech community. This also explains Leech’s (1981) observations with regard to the acceptability of metaphors (cf. example 27). In chapter 4.2.1 it was pointed out that metaphors like Sam is a wombat or Sam is a platypus are largely unacceptable because the average speaker lacks enough knowledge of the source concepts to find the metaphors meaningful. Now, this can be reformulated in terms of attribute salience. The concepts WOMBAT and PLATY- PUS are not very well-structured, they do not contain much information. For platypuses, for example, one can reasonably doubt that many Middle European speakers will have any knowledge at all, except perhaps for the fact that these animals have large bills. But even if a speaker’s mental representation contained this knowledge, it would not get them very far, as this attribute can hardly be applied to humans. For the more acceptable instances quoted by Leech (1981), it is much easier to see a connection between source and target. The average language user’s knowledge of PIGS , for example, is much larger. However, as soon as the source is applied to a human target, many otherwise important and salient aspects of the concept are disregarded (e.g. ‘are pink’, ‘have a snout’, ‘are kept for their meat’) and only stereotypical ideas, like ‘are dirty’ or ‘are greedy’, are used in the actual interpretation of the metaphor. Chiappe and Kennedy’s (2001: 260) discussion of salient attributes implies that the salient attributes which are mapped from source onto target become gradually fixed with repeated use. For the first few encounters with a specific metaphor or metonymy the user will rely on salient attrib- 205 While Ortony et al. (1985, cf. also Ortony 1980) and others (cf. e.g. Gildea and Glucksberg 1983; Glucksberg, McGlone and Manfredi 1997; Katz 1982) emphasize that the salient aspects of the source are responsible for which attributes of the target are highlighted by a metaphor, others, like Murphy (1996: 105, 1997: 196) or Grady (1999: 92) seem to believe that the aspects of the target foregrounded by a metaphor have to be highly salient within the mental representation of the target in the first place. This, of course, is related to the question of whether similarity is created by a metaphorical construal or is a precondition for it. Whereas Tourangeau and Rips (1991: 458), for example, do not believe that shared characteristics of source and target are a precondition for metaphors, Chiappe and Kennedy (2001: 259) observed that metaphors are preferred to similes when target and source are highly similar. 277 utes, 206 and when they have proved useful on these occasions, the figurative meaning becomes entrenched. As a result, the mapping is entirely limited to these salient attributes (cf. Grady 1999: 94; Lewandowska- Tomaszczyk 1985: 299) and other uses are largely excluded or at least felt to diverge from the usual meaning conveyed by the expression in question. In other words, the entrenchment of metaphorical senses which are based on lean mappings can depend on a shift in attribute salience. When such a metaphor is first encountered, the hearer has to look for correspondences which are applicable to the target. Normally not particularly salient attributes are foregrounded for the purposes at hand. With repeated use, however, the salience of these attributes is increased and they become an important part of the stored representation of the source concept. Therefore, metaphorical uses can change not only our idea of the target, but also of the source. Similar to what has been said with regard to metonymies, the attributes emphasized by the mapping can be viewed as emergent structure. That we only focus on parts of the source is in line with both the general claims of conceptual metaphor theory and those of conceptual blending theory, which holds that mental spaces set up for online-reasoning are typically partial representations. When such partial representations are stored, however, they will also influence the nature of the original inputs, in particular of the source concept. In this way, changes of attribute salience which were at first the result of on-line reasoning processes, but have become fixed, can lead to the development of stock vehicles (cf. e.g. Glucksberg, Gildea and Bookin 1982: 96). This has already been mentioned for butcher (cf. chapter 4.4.2), but can also be applied to other metaphors, provided they are based on a lean mapping. The source PIG is a case in point. Presumably, it is only the frequent use of PIG as a metaphorical source concept which has caused the high salience of attributes like ‘greedy’ and ‘dirty’ in the mental representation of the animals (cf. Figure 22, below). The original source concept has a richer and more neutral structure. The metaphorical conceptualization of a person as a pig, however, utilizes only negative attributes of the source concept and results in the blend, which can be described as ‘greedy, dirty, unpleasant person’. And due to the frequent processing of this conceptualization the general concept PIG has changed significantly. Even in everyday contexts, pigs are often associated with negative properties, which are inherited from the blend. Although these associations certainly do not have the same status as attributes proper, which are usually more ‘neutral’ in that they are connected with real-world properties of the entities in question rather than with the speakers’ attitudes or stereotypical assumptions, they are nevertheless a vital 206 The relevance of salient attributes is not restricted to the interpretation of figurative language. Cf. e.g. Bock and Clifton (2000) and Estes and Glucksberg (2000) who point out its significance for the interpretation of novel N+N compounds. 278 part of the speakers’ mental representations. Such a metaphorically influenced concept is certainly among the prime examples of ICMs. And this newly structured idealized model is now particularly apt to function as a metaphorical source. The mapping routine no longer makes use of the original source, but exploits instead the structure it has created. Target Original source concept Metaphorically influenced source concept P ERSON ‘human’, ‘two legs’, ‘two arms’, etc. ‘positive properties’, ‘negative properties’, etc. P IG ‘animal’, ‘pink’, ‘fur’, ‘snout’, ‘four legs’, etc. ‘often lives in dirty sties’, ‘kept for its meat’ ‘ ? greedy’, etc. P IG ‘animal’, ‘greedy’, ‘dirty’, ‘pink’, ‘four legs’, ‘fur’, ‘kept for its meat’, etc. ‘greedy, dirty, unpleasant person’ Blend Figure 22: The development of the stock vehicle PIG as a conceptual blend That metaphors which are based on leaner mappings are not uncommon is obvious. One of the examples from the corpus study belonging to this group is sharks. 21.55% of all uses of the word-form were found to be metaphorical. As is typical of leaner mappings, the mapping proposed by the Master Metaphor List, i.e. COMPETITORS OR ANTAGONISTS ARE PREDATORS (Lakoff, Espenson and Schwartz 1991: 47), does not have to compete with other mappings that can underlie uses of this lexeme. It is the only metaphorical sense realized by this word. As compared to examples of other, richer mappings it is neither particularly frequent nor particularly infrequent. Animal metaphors are not very likely to ever entirely lose their figurative character. The explanatory purposes they can be used for are much too restricted to allow this. They violate the general preference for abstract targets which are understood more or less entirely via concrete sources. Taking into consideration that they deviate from generally preferred metaphorical construals by having concrete targets and only one instead of many correspondences, metaphors such as sharks should, however, not be as frequent as many others which adhere to these principles. Still, they work well in everyday discourse - not so much because there are general conceptual mappings motivating them, but because the speech community 279 has tacitly agreed on which aspects of the source are foregrounded in a metaphorical construal and has sometimes even adopted these aspects as the most salient ones when the source concept is encountered in isolation. This makes metaphors like pig ‘greedy, dirty, unpleasant person’ in a way similar to metonyms like Vietnam ‘war’ (cf. chapter 6.3.2), where the most salient attributes of the vehicle are also determined by the target concept. However, they are not pre-existing in the most literal sense, but only a matter of late learning. In such cases, it is thus more appropriate to speak of a cultural rather than of a cognitive motivation behind the conventionality of the linguistic metaphors. For rich mappings, things are different. They also utilize only some aspects of the source, but the insights gained by metaphors like ARGUMENT IS WAR or A THEORY IS A BUILDING do not change the structure of the source concepts involved. More likely than not, the mappings will pick out aspects of the source which are salient in the source concept in a neutral, nonmetaphorical context. For BUILDINGS , for example, elements which enhance their structural integrity are of prime importance. And it is those aspects which are most frequently and conventionally used to shape linguistic expressions based on the mapping THEORIES ARE BUILDINGS . Rich mappings are likely to uphold the salience structure of the source concept and typically concentrate not on one, but on many salient aspects. If they don’t, low relative frequencies are the consequence. This can be illustrated with the examples buy and sell (cf. chapter 7.2): They are infrequently used because the negative aspects foregrounded by these metaphors, i.e. that the goods to be sold and bought are of a bad quality, i.e. the ideas are typically untrue, are not normally especially salient in neither the OBJECT concept nor in the COMMERCIAL EVENT concept. These rich mappings rely on - or rather result in - creative emergent structure which is based on a shift in attribute salience. However, in contrast to lean mappings, this shift is only of a temporary nature: It does not have lasting effects on the source concept and leaves its overall structure intact, and this renders the metaphors unconventional. 7.5 Summary With regard to lean metaphorical mappings, shifts in attribute salience are likely to become stored. If the speakers agree on which aspect is highlighted when the sources occur in a metaphorical context, chances are that they are accepted and re-used, i.e. become quite conventional expressions with a fairly specific figurative meaning. Importantly, the metaphorical conceptualizations have repercussions for the general structure of the source concept in these cases. This is different for rich mappings. They typically exploit and build on the speakers general and neutral knowledge about the sources. Severe and lasting repercussions of rich mappings on 280 the sources would be counter-productive, since they would render the sources less effective for other possible metaphorical construals. Consider the above-mentioned cases buy and sell once more: If the negative conceptualization of the OBJECTS (i.e. IDEAS ) bought and sold had permanent consequences for the OBJECT domain, i.e. if the OBJECT concept was changed from an attitudinally neutral into a generally negative one, this would heavily influence the usefulness of the OBJECT concept for other mappings and thus disturb the entire metaphorical system. Such shifts are typically only transitory, since the source domains of rich mappings characteristically have a very firm experiential grounding and are thus extremely stable and resistant to changes. As concerns the degree of conventionality of different linguistic expressions motivated by rich mappings, it has been demonstrated that single lexemes often tend to be strongly associated with a single metaphorical conceptualization. For example, 94.56% of all metaphorical meanings expressed by beat rely on COMPETITION IS WAR . For escape, it could be shown that even though the metaphorical uses of the form can be regarded as resulting from different mappings, one could also assume that the mappings commonly formulated by conceptual metaphor theorists are possibly somewhat too specific, and that the speakers actually work with a very general, highly schematic mapping which can account for all different meanings conveyed by the lexeme. The amount of data inspected here is not large enough to make any definite statements about the degree of conventionality of different metaphorical mappings. For example, the fact that all the relative frequencies found for those word-forms of my sample which can realize the mapping MACHINES ARE PEOPLE are very low, does not exclude that other words are more often used to instantiate the mapping. Much more extensive analyses of as many potential source domain vocabulary items as possible would be necessary to make such wide-ranging claims. What is certain, however, is that the mapping as such is not an entirely unconventional way of thinking - this is evident in the fact that it can be instantiated by various lexical means - and that those examples found in the study are unconventional as far as the linguistic level is concerned. This marks the difference to cases which are infrequent on the linguistic level, but not supported by an existing conceptual mapping either, like vast impenetrable openness (FP3-1317), where an OPEN SPACE is construed as a SOLID . 281 8 Conclusion The traditional approach which considered figurative language as a mere decorative element has long been replaced by the view that figurative language is essential for humans - not only as a way of speaking, but also as a way of thinking. The most deeply entrenched figurative ways of thinking are reflected in conventional linguistic metonymic and metaphorical expressions. At the beginning of this study, the question was raised of whether all the examples discussed by the relevant literature on the conceptual theory of metaphor and metonymy are really equally conventional. The unequivocal answer is no. The conventionality of metaphorical and metonymic mappings is usually connected to the number of different linguistic expressions to which they give rise. But it is not sufficient to propose a more or less large number of different examples which are all based on one and the same mapping and could - in principle - be used by the speakers. Of importance is whether the speakers really use them or not and do so frequently. Only when we know which of the mappings proposed by the theory are commonly used by speakers, can we identify those which are at the heart of human cognition and those which are less central from a cognitive perspective. Then we can ascertain why some mappings are preferred to others. For some time, cognitive linguistics has been a more or less theoretical discipline. Both its basic assumptions and the examples used to support them were founded mainly on introspection rather than on the analysis of authentic linguistic data. More recently, the situation has changed, and an increasing amount of usage-based accounts are the result. This is evident in monographs such as Deignan’s (2005) Metaphor and corpus linguistics and collections such as Stefanowitsch and Gries (2006). Corpus-based studies are particularly informative about the actual usage of metonymies and metaphors, since they allow researchers to study large samples of authentic language and therefore to recognize general tendencies and regularities. The corpus-based part of this study has shown that the conventionality of figurative language is not a result of arbitrary preferences of the speech community, but governed by several factors. For metaphors, the results prove that some source domains are dispreferred for metaphorical construals. The domains of SEVERE BODILY HARM and COMMERCIAL EVENTS were identified as rather unconventional sources, even though both domains pertain to areas of common human experience. The comparative infrequency of such construals was shown to be connected to either particular connotations resulting from a foregrounding of usually non-salient attributes and leading to a restricted applicability of the linguistic metaphors or to the specificity of the literal meanings of the lexical items. For lean mappings it was demonstrated that the conventionality is not lower than for 282 richer mappings, as long as the source contains highly salient attributes that can be fruitfully applied to the targets. The conventionality of lean mappings is, however, less influenced by general cognitive laws than by cultural factors. Concerning metonymies, a more detailed and extensive description of the factors which determine their conventionality could be given. Different metonymic mappings diverge immensely with respect to how often they are used in actual discourse. This is most obvious for reversible mappings, which always have one preferred direction. Often, this can be explained with the help of the general laws of ontological salience, but the application of these laws to concrete examples is not always unproblematic. In many cases, one and the same metonymic mapping is in line with one principle, but at odds with another. In addition, the analysis has shown that the examples motivated by one and the same mapping often differ markedly in terms of their relative frequencies. Therefore, it was necessary to look for criteria other than ontological salience in order to account for all the cases. Attribute salience has proved useful here. Target-in-vehicle salience - not as a part-whole relationship, but as a measure of the structure of the vehicle concept and the weight of the single attributes it contains - explains why some mappings which are normally likely to produce conventional linguistic metonyms actually generate unconventional ones and vice versa. The examples which are in line with the principle of vehicle-in-target salience are frequently so conventional that they can serve as starting points for further metonymic construals based on them, i.e. be used as the first parts of longer metonymic chains. However, he conventionality of the single items always has to be considered within the framework of a single mapping. It is always relative to the conventionality of the mapping as a whole. Since the number of the instances of the different mappings was sometimes quite limited, the results presented here are necessarily provisional. More detailed analyses of many more different instances of the mappings would be needed in order to substantiate them. The most conventional metonyms are those which fit the human cognitive apparatus best by using unambiguous and stable associative paths from vehicle to target. Since these paths are so natural, they are easier to use than others and become more and more entrenched. In this way, the nature of the cognitive construals bears a direct influence on the processing of linguistic metonyms. Metonymies are not only referential shorthands which are rarely used. Metonymic senses can, just like metaphorical ones, become quite central and even outweigh the literal senses in terms of frequency. If the psycholinguists are right, this means that they are activated and understood at least as fast and effortlessly as the literal meanings. Moreover, metonymic construals are not simply synonymous alternatives to literal references. The reference point chosen often influences the way we conceive of a target. For metonymies in general as well as for rich metaphorical mappings, speakers prefer conceptions which largely corres- 283 pond to the general structure of the concepts involved. Expressions that diverge from this tendency and highlight aspects which are usually not so important are less likely to become accepted. However, the degree of conventionality in our figurative language may differ and develop, it contributes more than was previously recognized to a generative cognitivelinguistic process. While metonymies just shift our perspectives, metaphors often enable us to see things from an entirely new and unexpected angle. 285 References Alač, Morana and Seana Coulson (2004), “The man, the key, or the car: Who or what is parked out back? ”, Cognitive Science Online 2, 21-34 <http: / / cogsci-online. ucsd.edu/ 2/ vol2_issue1.pdf> (last accessed November 26, 2006). Albert, Ruth and Cor J. Koster (2002), Empirie in Linguistik und Sprachlehrforschung. Ein methodologisches Arbeitsbuch, Tübingen: Narr. Apresjan, Jurij D. (1974), “Regular polysemy”, Linguistics 142, 5-32. Ariel, Mira (2002a), “The demise of a unique concept of literal meaning”, Journal of Pragmatics 34, 361-402. — (2002b), “Privileged interactional interpretations”, Journal of Pragmatics 34, 1003- 1044. Aronoff, Mark (1980), “Contextuals”, Language 56, 744-758. Barcelona, Antonio, ed. (2000a), Metaphor and metonymy at the crossroads: A cognitive perspective, Berlin - New York: Mouton de Gruyter. — (2000b), “Introduction. The cognitive theory of metaphor and metonymy”, in Barcelona (2000a), 1-28. — (2000c), “On the plausibility of claiming a metonymic motivation for conceptual metaphor”, in Barcelona (2000a), 31-58. — (2002a), “Clarifying and applying the notions of metaphor and metonymy within cognitive linguistics: An update”, in Dirven and Pörings (2002), 207-277. — (2002b), “On the ubiquity and multi-level operation of metonymy”, in Barbara Lewandowska-Tomaszczyk and Kamila Turewicz, eds., Cognitive linguistics today, Frankfurt/ M.: Peter Lang, 207-224. — (2003), “Metonymy in cognitive linguistics: An analysis and a few modest proposals”, in Hubert Cuyckens, Thomas Berg, René Dirven and Klaus-Uwe Panther, eds., Motivation in language. Studies in honor of Günter Radden, Amsterdam - Philadelphia: John Benjamins, 223-255. — (2004), “Metonymy behind grammar: The motivation of the seemingly ‘irregular’ grammatical behavior of English paragon names”, in Radden and Panther (2004a), 357-374. Barnden, John (1997), ATT-Meta Project Databank. Metonymy in mental state descriptions <http: / / www.cs.bham.ac.uk/ ~jab/ ATT-Meta/ Databank/ Indices/ metonymy.html> (last accessed August 20, 2007). Barsalou, Lawrence W. (1982), “Context-independent and context-dependent information in concepts”, Memory & Cognition 10, 82-93. — (1983), “Ad-hoc categories”, Memory & Cognition 11, 211-227. Bartsch, Renate (1996), “The myth of literal meaning”, in Edda Weigand and Franz Hundsnurscher, eds., Language structure and language use. Proceedings of the International Conference on Lexicology and Lexical Semantics, Münster, September 13-15, 1994, Tübingen: Niemeyer, 3-16. — (2002), “Generating polysemy: Metaphor and metonymy”, in Dirven and Pörings (2002), 49-74. Bauer, Laurie (1983), English word-formation, Cambridge: Cambridge University Press. — (1998), “When is a sequence of two nouns a compound in English? ”, English Language and Linguistics 2(1), 65-86. — (2001), Morphological productivity, Cambridge: Cambridge University Press. 286 — (2005), “Productivity: Theories”, in Štekauer and Lieber (2005), 315-334. Becker, Angela H. (1997), “Emergent and common features influence metaphor interpretation”, Metaphor and Symbol 12(4), 243-259. Benczes, Réka (2006), Creative compounding in English. The semantics of metaphorical and metonymical noun-noun combinations, Amsterdam - Philadelphia: John Benjamins. Black, Max (1962), Models and metaphors. Studies in language and philosophy, Ithaca/ NY: Cornell University Press. — (1977), “More about metaphor”, Dialectica 31(3/ 4), 431-457. Blank, Andreas (1999), “Co-presence and succession. A cognitive typology of metonymy”, in Panther and Radden (1999), 169-191. Blank, Andreas and Peter Koch, eds. (1999), Historical semantics and cognition, Berlin - New York: Mouton de Gruyter. Blank, Glenn D. (1988), “Metaphors in the lexicon”, Metaphor and Symbolic Activity 3(1), 21-36. 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. Bock, Jeannine S. and Charles Clifton, Jr. (2000), “The role of salience in conceptual combination”, Memory & Cognition 28(8), 1378-1386. Bock, Kathryn and Willem J.M. Levelt (1994), “Language production. Grammatical encoding”, in Gernsbacher (1994), 945-984. Bock, Kathryn, David E. Irwin, Douglas J. Davidson and Willem J.M. Levelt (2003), “Minding the clock”, Journal of Memory and Language 48, 653-685. Bowdle, Brian F. and Dedre Gentner (2005), “The career of metaphor”, Psychological Review 112(1), 193-216. Bredin, Hugh (1984), “Metonymy”, Poetics Today 5(1), 45-58. Brinton, Laurel J. and Elizabeth C. Traugott (2005), Lexicalization and language change, Cambridge etc.: Cambridge University Press. Brisard, Frank, Gert van Rillaer and Dominiek Sandra (2001), “Processing polysemous, homonymous, and vague adjectives”, in Cuyckens and Zawada (2001), 261- 284. Brône, Geert and Kurt Feyaerts (2005), “Headlines and cartoons in the economic press: Double grounding as a discourse supportive strategy”, in Guido Erreygers and Geert Jacobs, eds., Language, communication and the economy, Amsterdam - Philadelphia: John Benjamins, 73-99. Brugmann, Claudia (1990), “What is the invariance hypothesis”, Cognitive Linguistics 1(2), 257-266. Bybee, Joan (1985), Morphology: A study in the relation between meaning and form, Amsterdam: John Benjamins. — (2006), “From usage to grammar: The mind’s response to repetition”, Language 82(4), 711-733. Bybee, Joan and David Eddington (2006), “A usage-based approach to Spanish verbs of ‘becoming’”, Language 82(2), 323-355. Bybee, Joan and Paul Hopper (2001), “Introduction to frequency and the emergence of linguistic structure”, in Joan Bybee and Paul Hopper, eds., Frequency and the emergence of linguistic structure, Amsterdam - Philadelphia: John Benjamins, 1-24. Cacciari, Cristina and Patrizia Tabossi (1988), “The comprehension of idioms”, Journal of Memory and Language 27, 668-683. 287 Cameron, Lynne and Graham Low, eds. (1999), Researching and applying metaphor, Cambridge etc.: Cambridge University Press. Chandler, Steven R. (1991), “Metaphor comprehension: A connectionist approach to implications for the mental lexicon”, Metaphor and Symbolic Activity 6(4), 227-258. Chiappe, Dan L. and John M. Kennedy (2001), “Literal bases for metaphor and simile”, Metaphor and Symbol 16(3&4), 249-276. Clark, Eve V. and Herbert H. Clark (1979), “When nouns surface as verbs”, Language 55, 767-811. Clark, Herbert H. (1983), “Making sense of nonce sense”, in Giovanni B. Flores d’Arcais and Robert J. Jarvella, eds., The process of language understanding, New York: Wiley, 297-331. — (1991), “Words and their possibilities in the world”, in Gregory Lockhead and James R. Pomerantz, eds., The perception of structure. Essays in honor of Wendell Garner, Washington, DC: American Psychological Association, 263-277. Clark, Herbert H. and Richard J. Gerrig (1983), “Understanding old words with new meanings”, Journal of Verbal Learning and Verbal Behavior 22, 591-608. Clark, Herbert H. and Susan E. Haviland (1977), “Comprehension and the givennew contract”, in Roy O. Freedle, ed., Discourse production and comprehension, Norwood/ NJ: Ablex, 1-40. Clark, Herbert H. and Peter Lucy (1975), “Understanding what is meant form what is said: A study in conversationally conveyed requests”, Journal of Verbal Learning and Verbal Behavior 14, 56-72. Clarke, David D. and Brigitte Nerlich (1991), “Word-waves: A computational model of lexical semantic change”, Language & Communication 11(3), 227-238. Clausner, Timothy C. and William Croft (1999), “Domains and image-schemas”, Cognitive Linguistics 10(1), 1-31. Collins, Allan M. and Elizabeth F. Loftus (1975), “A spreading-activation theory of semantic processing”, Psychological Review 82(6), 407-428. Collins, Allan M. and M. Ross Quillian (1969), “Retrieval time from semantic memory”, Journal of Verbal Learning and Verbal Behaviour 8, 240-247. Coulson, Seana (2001), Semantic leaps. Frame-shifting and conceptual blending in meaning construction, Cambridge etc.: Cambridge University Press. Coulson, Seana and Todd Oakley (2000), “Blending basics”, Cognitive Linguistics 11(3/ 4), 175-196. — (2003), “Metonymy and conceptual blending”, in Panther and Thornburg (2003a), 51-79. Coulson, Seana and Cyma Van Petten (2002), “Conceptual integration and metaphor: An event-related potential study” Memory & Cognition 30(6), 958-968. Croft, William (2000), Explaining language change. An evolutionary approach, Harlow: Longman. — (2002), “The role of domains in the interpretation of metaphors and metonymies” in Dirven and Pörings (2002), 161-205. — (2006), “On explaining metonymy: Comment on Peirsman and Geeraerts, ‘Metonymy as a prototypical category’”, Cognitive Linguistics 17(3), 317-326. — (2007), “Construction grammar”, in Dirk Geeraerts and Hubert Cuyckens, eds., The Oxford handbook of cognitive linguistics, Oxford etc.: Oxford University Press, 463-508. Croft, William and D. Alan Cruse (2004), Cognitive linguistics, Cambridge etc.: Cambridge University Press. 288 Cruse, D. Alan (1986), Lexical semantics, Cambridge etc.: Cambridge University Press. Cuyckens, Hubert and Britta Zawada, eds. (2001), Polysemy in cognitive linguistics. Selected papers from the Fifth International Cognitive Linguistics Conference, Amsterdam 1997, Amsterdam - Philadelphia: John Benjamins. Daneš, František (1974), “Functional sentence perspective and the organization of the text”, in František Daneš, ed., Papers on functional sentence perspective, The Hague: Mouton, 106-128. Dascal, Marcelo (1981), “Contextualism”, in Herman Parret, Marina Sbisà and Jef Verschueren, eds., Possibilities and limitations of pragmatics. Proceedings of the Conference on Pragmatics, Urbino, July 8-14, 1979, Amsterdam: John Benjamins, 153- 177. — (1983), Pragmatics and the philosophy of mind. Vol. I: Thought in language, Amsterdam: John Benjamins. — (1987), “Defending literal meaning”, Cognitive Science 11, 259-281. — (1989), “On the roles of context and literal meaning in understanding”, Cognitive Science 13, 253-257. Deignan, Alice (1999), “Corpus-based research into metaphor”, in Cameron and Low (1999), 177-199. — (2005), Metaphor and corpus linguistics, Amsterdam - Philadephia: John Benjamins. Díez Velasco, Olga I. (2001), “The role of semantic relations in the creation of metonymic mappings”, Miscelánea 23, 11-28. Dirven, René (1993), “Metonymy and metaphor: Different mental strategies of conceptualisation”, Leuvense Bijdragen 82, 1-28. — (1999), “Conversion as a conceptual metonymy of event schemata”, in Panther and Radden (1999), 275-287. Dirven, René and Ralf Pörings, eds. (2002), Metaphor and metonymy in comparison and contrast, Berlin - New York: Mouton de Gruyter. Downing, Pamela (1977), “On the creation and use of English compound nouns”, Language 53, 810-842. Duffy, Susan A., Robin K. Morris and Keith Rayner (1988), “Lexical ambiguity and fixation times in reading”, Journal of Memory and Language 27, 429-446. Estes, Zachary and Sam Glucksberg (2000), “Interactive property attribution in concept combination”, Memory & Cognition 28(1), 28-34. Evans, Vyvyan and Melanie Green (2006), Cognitive linguistics. An introduction, Edinburgh: Edinburgh University Press. Evans, Vyvyan and Andrea Tyler (2004), “Spatial experience, lexical structure and motivation. The case of in”, in Radden and Panther (2004a), 157-192. Fass, Dan (1997), Processing metonymy and metaphor, Greenwich/ CT - London: Ablex. Fauconnier, Gilles (1994), Mental spaces: Aspects of meaning construction in natural language, Cambridge etc.: Cambridge University Press. — (1997), Mappings in thought and language, Cambridge etc.: Cambridge University Press. Fauconnier, Gilles and Mark Turner (1998), “Conceptual integration networks”, Cognitive Science 22(2), 133-187. — (2000), “Compression and global insight”, Cognitive Linguistics 11(3/ 4), 283-304. — (2002), The way we think. Conceptual blending and the mind’s hidden complexities, New York: Basic Books. 289 Fillmore, Charles J. (1975), “An alternative to checklist theories of meaning”, in Cathy Cogen, Henry Thompson, Graham Thurgood and Kenneth Whistler, eds., Proceedings of the First Annual Meeting of the Berkeley Linguistics Society, Berkeley: Berkeley Linguistics Society, 123-131. — (1977), “Topics in lexical semantics”, in Roger W. Cole, ed., Current issues in linguistic theory, Bloomington - London: Indiana University Press, 76-138. — (1985), “Frames and the semantics of understanding”, Quaderni di Semantica 6, 222-254. Firbas, Jan (1964), “On defining the theme in functional sentence perspective analysis”, Traveaux Linguistiques de Prague 1, 267-280. — (1971), “On the concept of communicative dynamism in the theory of functional sentence perspective”, Brno Studies in English 7, 12-47. — (1992), Functional sentence perspective in written and spoken communication, Cambridge etc.: Cambridge University Press. Fischer, Roswitha (1998), Lexical change in present-day English. A corpus-based study of the motivation, institutionalization, and productivity of creative neologisms, Tübingen: Narr. Fisiak, Jacek, ed. (1985), Historical semantics. Historical word-formation, Berlin etc.: Mouton. Fletcher, Charles R. (1994), “Levels of representation in memory for discourse”, in Gernsbacher (1994), 589-607. Foltz, Peter W. (2003), “Quantitative cognitive models of text and discourse processing”, in Arthur C. Graesser, Morton A. Gernsbacher and Susan R. Goldman, eds., Handbook of discourse processes, Mahwah/ NJ: Lawrence Erlbaum, 487-523. Forster, Kenneth I. (2007), “Visual word-recognition: Problems and issues”, in Gonia Jarema and Gary Libben, eds., The mental lexicon: Core perspectives, Amsterdam: Elsevier, 31-53. Foss, Donald J. (1970), “Some effects of ambiguity upon sentence comprehension”, Journal of Verbal Learning and Verbal Behavior 9, 699-706. Francis, W. Nelson and Henry Kučera (1982), Frequency analysis of English usage: Lexicon and grammar, Boston: Houghton Mifflin. Frazier, Lyn and Keith Rayner (1990), “Taking on semantic commitments: Processing multiple meanings vs. multiple senses”, Journal of Memory and Language 29, 181-200. Frege, Gottlob (1952), “On sense and reference”, in Peter T. Geach and Max Black, eds., Max Black transl., Translations from the philosophical writings of Gottlob Frege, Oxford: Basil Blackwell, 56-78. [originally published as: Frege, Gottlob (1892), “Über Sinn und Bedeutung”, Zeitschrift für Philosophie und philosophische Kritik 100, 25-50.] Frisson, Steven (2009), “Semantic underspecification in language processing”, Language and Linguistics Compass 3(1), 111-127. Frisson, Steven and Martin J. Pickering (1999), “The processing of metonymy: Evidence from eye movements”, Journal of Experimental Psychology: Learning, Memory, and Cognition 25(6), 1366-1383. — (2001), “Obtaining a figurative interpretation of a word: Support for underspecification”, Metaphor and Symbol 16(3&4), 149-171. — (2007), “The processing of familiar and novel senses of a word: Why reading Dickens is easy but reading Needham can be hard”, Language and Cognitive Processes 22(4), 595-613. 290 Geeraerts, Dirk (1983), “Prototype theory and diachronic semantics: A case study”, Indogermanische Forschungen 88, 1-32. — (1985), “Cognitive restrictions on the structure of semantic change”, in Fisiak (1985), 127-153. — (1993a), “Vagueness’s puzzles, polysemy’s vagaries”, Cognitive Linguistics 4(3), 223-272. — (1993b), “Generalised onomasiological salience”, Belgian Journal of Linguistics 8, 43-56. — (1997), Diachronic prototype semantics. A contribution to historical lexicology, Oxford: Clarendon. — (1999), “Diachronic prototype semantics. A digest”, in Blank and Koch (1999), 91-107. — (2002), “The interaction of metaphor and metonymy in composite expressions”, in Dirven and Pörings (2002), 435-465. Geeraerts, Dirk and Stefan Grondelaers (1995), “Looking back at anger: Cultural traditions and metaphorical patterns”, in Taylor and MacLaury (1995), 153-174. Geeraerts, Dirk, Stefan Grondelaers and Peter Bakema (1994), The structure of lexical variation. A descriptive framework for cognitive lexicology, Berlin - New York: Mouton de Gruyter. Gentner, Dedre and Brian F. Bowdle (2001), “Convention, form, and figurative language processing”, Metaphor and Symbol 16(3&4), 223-247. Gentner, Dedre and Catherine A. Clement (1989), “Evidence for relational selectivity in interpreting analogy and metaphor”, in Gordon H. Bower, ed., The psychology of learning and motivation, New York: Academic Press, 307-358. Gernsbacher, Morton A. (1984), “Resolving 20 years of inconsistent interactions between lexical familiarity and orthography, concreteness, and polysemy”, Journal of Experimental Psychology: General 113(2), 256-282. — (1989), “Mechanisms that improve referential access”, Cognition 32(2), 99-156. — (1990), Language comprehension as structure building, Hillsdale/ NJ: Erlbaum. — ed. (1994), Handbook of psycholinguistics, San Diego/ CA: Academic Press. Gernsbacher, Morton A., Boaz Keysar, Rachel R.W. Robertson and Necia K. Werner (2001), “The role of suppression and enhancement in understanding metaphors”, Journal of Memory and Language 45, 433-450. Gernsbacher, Morton A. and Rachel R.W. Robertson (1999), “The role of suppression in figurative language comprehension”, Journal of Pragmatics 31, 1619-1630. Gerrig, Richard J. (1989), “The time course of sense creation”, Memory & Cognition 17(2), 194-207. Gibbs, Raymond W., Jr. (1979), “Contextual effects in understanding indirect requests”, Discourse Processes 2, 1-10. — (1980), “Spilling the beans on understanding and memory for idioms in conversation”, Memory & Cognition 8, 449-456. — (1982), “A critical examination of the contribution of literal meaning to understanding nonliteral discourse”, Text 2, 9-27. — (1984), “Literal meaning and psychological theory”, Cognitive Science 8, 275-304. — (1989), “Understanding and literal meaning”, Cognitive Science 13, 243-251. — (1990), “Comprehending figurative referential descriptions”, Journal of Experimental Psychology: Learning, Memory, and Cognition 16(1), 56-66. — (1993), “Why idioms are not dead metaphors”, in Cristina Cacciari and Patrizia Tabossi, eds., Idioms: Processing, structure, and interpretation, Hillsdale/ NJ: Erlbaum, 57-78. 291 — (1994), The poetics of mind. Figurative thought, language, and understanding, Cambridge: Cambridge University Press. — (1999), “Taking metaphor out of our heads and putting it into the cultural world”, in Gibbs and Steen (1999), 146-166. — (2000), “Making good psychology out of blending theory”, Cognitive Linguistics 11(3/ 4), 347-358. — (2001), “Evaluating contemporary models of figurative language understanding”, Metaphor and Symbol 16(3/ 4), 317-333. — (2002), “A new look at literal meaning in understanding what is said and implicated”, Journal of Pragmatics 34, 457-486. Gibbs, Raymond W., Jr., Jody Bogdonovich, Jeffrey Sykes and Dale Barr (1997), “Metaphor in idiom comprehension”, Journal of Memory and Language 37, 141- 154. Gibbs, Raymond W., Jr. and Richard J. Gerrig (1989), “How context makes metaphor comprehension seem ‘special’”, Metaphor and Symbolic Activity 4(3), 145-158. Gibbs, Raymond W., Jr and Teenie Matlock (2001), “Psycholinguistic perspectives on polysemy”, in Cuyckens and Zawada (2001), 213-239. Gibbs, Raymond W., Jr. and Jennifer E. O’Brien (1990), “Idioms and mental imagery: The metaphorical motivation for idiomatic meaning”, Cognition 36, 35-68. Gibbs, Raymond W., Jr. and Gerard Steen, eds. (1999), Metaphor in cognitive linguistics. Selected papers from the Fifth International Cognitive Linguistics Conference, Amsterdam, July 1997, Amsterdam - Philadelphia: John Benjamins. Gildea, Patricia and Sam Glucksberg (1983), “On understanding metaphor: The role of context”, Journal of Verbal Learning and Verbal Behavior 22, 577-590. Giora, Rachel (1997), “Understanding figurative and literal language: The graded salience hypothesis”, Cognitive Linguistics 8(3), 183-206. — (1999), “On the priority of salient meanings: Studies of literal and figurative language”, Journal of Pragmatics 31, 919-929. — (2002), “Literal vs. figurative language: Different or equal? ”, Journal of Pragmatics 34, 487-506. — (2003), On our mind. Salience, context, and figurative language, Oxford etc.: Oxford University Press. Gilquin, Gaëtanelle (2006), “The place of prototypicality in corpus linguistics: Causation in the hot seat”, in Stefan Th. Gries and Anatol Stefanowitsch, eds., Corpora in cognitive linguistics: Corpus-based approaches to syntax and lexis, Berlin - New York: Mouton de Gruyter, 159-191. Giora, Rachel and Ofer Fein (1999), “On understanding familiar and less-familiar figurative language”, Journal of Pragmatics 31, 1601-1618. Glucksberg, Sam (1989), “Metaphors in conversation: How are they understood? Why are they used? ”, Metaphor and Symbolic Activity 4(3), 125-143. Glucksberg, Sam, Mary Brown and Matthew S. McGlone (1993), “Conceptual metaphors are not automatically accessed during idiom comprehension”, Memory & Cognition 21(5), 711-719. Glucksberg, Sam, Patricia Gildea and Howard B. Bookin (1982), “On understanding nonliteral speech: Can people ignore metaphors? ”, Journal of Verbal Learning and Verbal Behavior 21, 85-98. Glucksberg, Sam and Boaz Keysar (1990), “Understanding metaphorical comparisons: Beyond similarity”, Psychological Review 97, 3-18. Glucksberg, Sam and Matthew S. McGlone (1999), “When love is not a journey: What metaphors mean”, Journal of Pragmatics 31, 1541-1558. 292 Glucksberg, Sam, Matthew S. McGlone and Deanna Manfredi (1997), “Property attribution in metaphor comprehension”, Journal of Memory and Language 36, 50- 67. Glucksberg, Sam, Mary R. Newsome and Yevgeniya Goldvarg (2001), “Inhibition of the literal: Filtering metaphor-irrelevant information during metaphor comprehension”, Metaphor and Symbol 16(3/ 4), 277-293. Goldvarg, Yevgeniya and Sam Glucksberg (1998), “Conceptual combinations: The role of similarity”, Metaphor and Symbol 13(4), 243-255. Goossens, Louis (1990), “Metaphtonymy: The interaction of metaphor and metonymy in expressions for linguistic action”, Cognitive Linguistics 1(3), 323-340. Grady, Joseph E. (1997), “T HEORIES ARE BUILDINGS revisited”, Cognitive Linguistics 8(4), 267-290. — (1999), “A typology of motivation for conceptual metaphor. Correlation vs. resemblance”, in Gibbs and Steen (1999), 79-100. Grady, Joseph E., Todd Oakley and Seana Coulson (1999), “Blending and metaphor”, in Gibbs and Steen (1999), 101-124. Grady, Joseph E., Sarah Taub and Pamela Morgan (1996), “Primitive and compound metaphors”, in Adele Goldberg, ed., Conceptual structure, discourse, and language, Stanford/ CA: CSLI Publications, 177-187. Gregory, Monica E. and Nany L. Mergler (1990), “Metaphor comprehension: In search of literal truth, possible sense, and metaphoricity”, Metaphor and Symbolic Activity 5(3), 151-173. Grice, H. Paul (1957), “Meaning”, The Philosophical Review 66(3), 377-388. — (1975; 1998), “Logic and conversation”, in Asa Kasher, ed., Pragmatics. Critical concepts, vol. IV: Presupposition, London: Routledge, 145-161. [originally published in Peter Cole and Jerry L. Morgan, eds. (1975), Syntax and semantics, Vol. 3: Speech acts, New York: Academic Press, 41-58.] — (1978; 1998), “Further notes on logic and conversation”, in Asa Kasher, ed., Pragmatics. Critical concepts, vol. IV: Presupposition, London: Routledge, 162-176. [originally published in Peter Cole, ed. (1978), Syntax and semantics, Vol. 9: Pragmatics, New York: Academic Press, 113-127.] Grimm, Ursula (1991), Lexikalisierung im heutigen Englisch am Beispiel der -er Ableitungen, Tübingen: Narr. Grygiel, Marcin (2004), “Semantic change as a process of conceptual blending”, Annual Review of Cognitive Linguistics 2, 285-304. Halliday, Michael A.K. (1967), “Notes on transitivity and theme in English. Part II”, Journal of Linguistics 3, 199-244. Hanks, Patrick (2006), “Metaphoricity is gradable”, in Stefanowitsch and Gries (2006), 17-35. Hansen, Barbara, Klaus Hansen, Albrecht Neubert and Manfred Schentke (1990), Englische Lexikologie. Einführung in Wortbildung und lexikalische Semantik, 3 rd ed., Leipzig: VEB Verlag Enzyklopädie. Harder, Peter (2003), “Mental spaces: Exactly when do we need them? ”, Cognitive Linguistics 14(1), 91-96. Harris, Richard J. (1976), “Comprehension of metaphors: A test of the two-stage processing model”, Bulletin of the Psychonomic Society 8(4), 312-314. Haviland, Susan E. and Herbert H. Clark (1974), “What’s new? Acquiring new information as a process in comprehension”, Journal of Verbal Learning and Verbal Behavior 13(5), 512-521. 293 Hilpert, Martin (2006), “Keeping an eye on the data: Metonymies and their patterns”, in Stefanowitsch and Gries (2006), 123-151. — (to appear), “Chained metonymies”, in John Newman and Sally A. Rice, eds., Experimental and empirical methods, Stanford/ CA: CSLI Publications. Hirotani, Masako, Lyn Frazier and Keith Rayner (2006), “Punctuation and intonation effects on clause and sentence wrap-up”, Journal of Memory and Language 54, 425-443. Hoffmann, Sebastian (2004), “Are low-frequency complex prepositions grammaticized? On the limits of corpus data - and the importance of intuition”, in Hans Lindquist and Christian Mair, eds., Corpus approaches to grammaticalization in English, Amsterdam - Philadelphia: John Benjamins, 171-210. Hogaboam, Thomas W. and Charles A. Perfetti (1975), “Lexical ambiguity and sentence comprehension”, Journal of Verbal Learning and Verbal Behavior 14, 265-274. Hohenhaus, Peter (1996), Ad-hoc-Wortbildung. Terminologie, Typologie und Theorie kreativer Wortbildung im Englischen, Frankfurt/ M. etc.: Lang. — (2005), “Lexicalization and institutionalization”, in Štekauer and Lieber (2005), 353-373. Honeck, Richard P. and Robert R. Hoffman, eds. (1980), Cognition and figurative language, Hillsdale/ NJ: Lawrence Erlbaum. Indurkhya, Bipin (1992), Metaphor and cognition. An interactionist approach, Dordrecht etc.: Kluwer. Inhoff, Albrecht W., Susan D. Lima and Patrick J. Carroll (1984), “Contextual effects on metaphor comprehension”, Memory & Cognition 12, 558-567. Jäkel, Olaf (1997), Metaphern in abstrakten Diskurs-Domänen. Eine kognitiv-linguistische Untersuchung anhand der Bereiche Geistestätigkeit, Wirtschaft und Wissenschaft, Frankfurt/ M. etc.: Lang. — (1999), “Kant, Blumenberg, Weinrich. Some forgotten contributions to the cognitive theory of metaphor”, in Gibbs and Steen (1999), 9-27. Jakobson, Roman (1971), “Two aspects of language and two types of aphasic disturbances”, in Roman Jakobson, ed., Roman Jakobson. Selected writings, Vol. II: Word and Language, The Hague - Paris: Mouton, 239-259. Jakobson, Roman and Morris Halle (1956), Fundamentals of language, The Hague: Mouton. Janus, Raizi A. and Thomas G. Bever (1985), “Processing of metaphoric language: An investigation of the three-stage model of metaphor comprehension”, Journal of Psycholinguistic Research 15(5), 473-487. Jescheniak, Jörg and Willem J.M. Levelt (1994), “Word frequency effects in speech production: Retrieval of syntactic information and of phonological form”, Journal of Experimental Psychology: Learning, Memory, and Cognition 20(4), 824-843. Jespersen, Otto (1925), Die Sprache. Ihre Natur, Entwicklung und Entstehung, transl. by Rudolf Hittmair and Karl Waibel, Heidelberg: Carl Winters. Johnson, Mark (1987), The body in the mind. The bodily basis of meaning, imagination, and reason, Chicago: University of Chicago Press. Johnson, Michael G. and Robert G. Malgady (1979), “Some cognitive aspects of figurative language: Association and metaphor”, Journal of Psycholinguistic Research 8(3), 249-265. — (1980), “Toward a perceptual theory of metaphoric comprehension”, in Honeck and Hoffman (1980), 259-282. Just, Marcel A. and Patricia A. Carpenter (1980), “A theory of reading: From eye fixations to comprehension”, Psychological Review 87, 329-354. 294 Kastovsky, Dieter (1982), Wortbildung und Semantik, Düsseldorf etc.: Bagel/ Francke. Katz, Albert N. (1982), “Metaphoric relationships: The role of feature saliency”, Journal of Psycholinguistic Research 11(4), 283-296. — (1989), “On choosing the vehicles of metaphors: Referential concreteness, semantic distance, and individual differences”, Journal of Memory and Language 28, 486-499. Katz, Albert N. and Todd R. Ferretti (2001), “Moment-by moment reading of proverbs in literal and nonliteral contexts”, Metaphor and Symbol 16(3&4), 193-221. Katz, Albert N., Allan Paivio, Marc Marschank and James M. Clark (1988), “Norms for 204 literary and 260 nonliterary metaphors on 10 psychological dimensions”, Metaphor and Symbolic Activity 3(4), 191-214. Katz, Jerrold J. (1977), Propositional structure and illocutionary force. A study of the contribution of sentence meaning to speech acts, New York: Thomas Y. Crowell. Katz, Jerrold J. and Jerry A. Fodor (1963), “The structure of a semantic theory”, Language 39, 170-210. Kemper, Susan (1981), “Comprehension and the interpretation of proverbs”, Journal of Psycholinguistic Research 10(2), 179-198. Keysar, Boaz (1989), “On the functional equivalence of literal and metaphorical interpretations in discourse”, Journal of Memory and Language 28, 375-385. — (1994), “Discourse context effects: Metaphorical and literal interpretations”, Discourse Processes 18, 247-269. Kintsch, Walter (1994), “The psychology of discourse processing”, in Gernsbacher (1994), 721-739. — (1988), “The role of knowledge in discourse comprehension: A construction integration model”, Psychological Review 95(2), 163-182. Kittay, Eva Feder (1987), Metaphor: Its cognitive force and linguistic structure, Oxford: Clarendon Press. Knowles, Murray and Rosamund Moon (2006), Introducing metaphor, London - New York: Routledge. Koch, Peter (2001), “Metonymy. Unity in diversity”, Journal of Historical Pragmatics 2(2), 201-244. — (2004), “Metonymy between pragmatics, reference, and diachrony”, metaphorik.de 07/ 2004, 6-53 <http: / / www.metaphorik.de/ 07/ koch.pdf> (last accessed July 29, 2007). Kövecses, Zoltán (1995), “Anger: Its language, conceptualization, and physiology in the light of cross-cultural evidence”, in Taylor and MacLaury (1995), 181-196. — (1999), “Metaphor: Does it constitute or reflect cultural models? ”, in Gibbs and Steen (1999), 167-188. — (2000), Metaphor and emotion: Language, culture, and body in human feeling, Cambridge etc.: Cambridge University Press. — (2002), Metaphor. A practical introduction, Oxford etc.: Oxford University Press. — (2006a), Metaphor in culture: Universality and variation, Cambridge etc.: Cambridge University Press. — (2006b), “Embodiment, experiential focus, and diachronic change in metaphor”, in R.W. McConchie et al., eds., Selected proceedings of the 2005 Symposium on New Approaches in Historical Lexis (HEL-LEX), Somerville/ MA: Cascadilla Proceedings Project, 1-7 <http: / / www.lingref.com/ cpp/ hel-lex/ 2005/ paper1241.pdf> (last accessed August 20, 2009). Kövecses, Zoltán and Günter Radden (1998), “Metonymy: Developing a cognitive linguistic view”, Cognitive Linguistics 9(1), 37-77. 295 Krennmayr, Tina (2008), “Using dictionaries in linguistic metaphor identification”, in Nils-Lennart Johannesson and David C. Minugh, eds., Selected papers from the 2006 and 2007 Stockholm Metaphor Festivals, Stockholm: Department of English, Stockholm University, 97-115. Kutas, Marta and Cyma Van Petten (1994), “Psycholinguistics electrified: Eventrelated brain potential investigations”, in Gernsbacher (1994), 83-143. Lakoff, George (1986), “The meanings of literal”, Metaphor and Symbolic Activity 1(4), 291-296. — (1987), Women, fire, and dangerous things. What categories reveal about the mind, Chicago - London: University of Chicago Press. — (1990), “The invariance hypothesis: Is abstract reason based on imageschemas? ”, Cognitive Linguistics 1(1), 39-74. — (1993), “The contemporary theory of metaphor”, in Ortony (1993), 202-251. Lakoff, George, Jane Espenson and Alan Schwartz (1991), Master metaphor list, 2 nd edition, University of California at Berkeley <http: / / araw.mede.uic.edu/ ˜alansz/ metaphor/ METAPHORLIST.pdf> (last accessed August 7, 2007). 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. (1984), “Active zones”, Proceedings of the Tenth Annual Meeting of the Berkeley Linguistics Society, Berkeley: Berkeley Linguistics Society, 172-188. — (1987), Foundations of cognitive grammar. Vol. I: Theoretical prerequisites, Stanford/ CA: Stanford University Press. — (1991), Foundations of cognitive grammar. Vol. II: Descriptive application, Stanford/ CA: Stanford University Press. — (1993), “Reference-point constructions”, Cognitive Linguistics 4(1), 1-38. — (2000), Grammar and conceptualization, Berlin - New York: Mouton de Gruyter. — (2008), Cognitive grammar. A basic introduction, Oxford etc.: Oxford University Press. Lapata, Mirella, Frank Keller and Christoph Scheepers (2003), “Intra-sentential context effects on the interpretation of logical metonymy”, Cognitive Science 27(4), 649-668. Leech, Geoffrey N. (1981), Semantics. The study of meaning, 2 nd ed., London etc.: Penguin. Leezenberg, Michiel (2001), Contexts of metaphor, Amsterdam etc.: Elsevier. Leisi, Ernst (1975), Der Wortinhalt. Seine Struktur im Deutschen und Englischen, 5 th ed., Heidelberg: Winter. — (1985), Praxis der englischen Semantik, 2 nd ed., Heidelberg: Winter. Leite, Naomi, ed. (1994), Master metonymy list, Berkeley: University of California, unpublished MS. Levelt, Willem J.M. (1989), Speaking: From intention to articulation, Cambridge/ MA: MIT Press. Lewandowska-Tomaszczyk, Barbara (1985), “On semantic change in a dynamic model of language”, in Fisiak (1985), 297-323. Lipka, Leonhard (1988), “A rose is a rose is a rose: On simple and dual categorization in natural languages”, in Werner Hüllen and Rainer Schulze, eds., Understanding the lexicon. Meaning, sense and world knowledge in lexical semantics, Tübingen: Niemeyer, 355-366. 296 — (1992), “Lexicalization and institutionalization in English and German. Or: Piefke, Wendehals, smog, perestroika, AIDS, etc.”, Zeitschrift für Anglistik und Amerikanistik 40, 101-111. — (1996), “Words, metaphors and cognition: A bridge between domains”, in Jan Svartvik, ed., Words. Proceedings of an International Symposium, Lund 25.-26. August 1995, Stockholm: Almqvist & Wiksell, 49-69. — (1998), “Word-formation, metaphor and metonymy - Processes, results and their description”, Revista Canaria de Estudios Ingleses 36, 97-112. — (2002), English lexicology. Lexical structure, word semantics and word-formation, Tübingen: Narr. Low, Graham (1999), “Validating metaphor research projects”, in Cameron and Low (1999), 48-65. Lu, Angela Yi-chün (1998), Phonetic motivation. A study of the relationship between form and meaning, München: Hieronymus. Lyons, John (1968), Introduction to theoretical linguistics, Cambridge: Cambridge University Press. — (1977), Semantics, Vol. 2, Cambridge etc.: Cambridge University Press. Mandelblit, Nili and Gilles Fauconnier (2000), “How I got myself arrested. Underspecificity in grammatical blends as a source for constructional ambiguity”, in Ad Foolen and Friederike van der Leek, eds., Constructions in cognitive linguistics. Selected papers from the Fifth International Cognitive Linguistics Conference, Amsterdam, 1997, Amsterdam - Philadelphia: John Benjamins, 167-190. Marchand, Hans (1969), The categories and types of present-day English word-formation. A synchronic-diachronic approach, 2 nd ed., München: C.H. Beck. Markert, Katja and Udo Hahn (2002), “Understanding metonymies in discourse”, Artificial Intelligence 135, 145-198. Markert, Katja and Malvina Nissim (2002a), “Metonymy resolution as a classification task”, in Jan Haji and Yuji Matsumoto, eds., Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP), Philadelphia, July 2002, Association for Computational Linguistics, 204-213. — (2002b), “Towards a corpus annotated for metonymies: The case of location names”, in M.R. Rodríguez and C.P.S. Araujo, eds., Proceedings of the 3 rd International Conference on Language Resources and Evaluation, Las Palmas, Canary Islands, 2002, Gran Canaria: Imprenta Papeleria San Rafael <http: / / www.comp.leeds. ac.uk/ markert/ Papers/ lrec2002.pdf> (last accessed August 20, 2007). — (2003), “Corpus-based metonymy analysis”, Metaphor and Symbol 18(3), 175-188. — (2006), “Metonymic proper names: A corpus-based account”, in Stefanowitsch and Gries (2006), 152-174. Marschank, Marc, Albert N. Katz and Allan Paivio (1983), “Dimensions of metaphor”, Journal of Psycholinguistic Research 12(1), 17-40. Martin, Charles, Hoang Vu, George Kellas and Kimberly Metcalf (1999), “Strength of discourse context as a determinant of the subordinate bias effect”, The Quarterly Journal of Experimental Psychology 52A(4), 813-839. Martin, James H. (2006), “A corpus-based analysis of context effects on metaphor comprehension”, in Stefanowitsch and Gries (2006), 214-236. Mathesius, Vilém (1929), “Zur Satzperspektive im modernen Englisch”, Archiv für das Studium der neueren Sprachen und Literaturen 155, 202-210. McElree, Brian, Steven Frisson and Martin J. Pickering (2006), “Deferred interpretations: Why starting Dickens is taxing but reading Dickens isn’t”, Cognitive Science 30, 181-192. 297 McGlone, Matthew S. (1996), “Conceptual metaphors and figurative language interpretation: Food for thought? ”, Journal of Memory and Language 35, 544-565. Medin, Douglas L. and Edward J. Shoben (1988), “Context and structure in conceptual combination”, Cognitive Psychology 20, 158-190. Meer, Gert van der (1999), “Metaphors and dictionaries: The morass of meaning, or how to get two ideas for one”, International Journal of Lexicography 12(3), 195-208. — (2002), “Metaphors: How do dictionaries scramble out of this morass of meaning? ”, in Henrik Gottlieb, Jens E. Morgensen and Arne Zettersten, eds., Symposium on Lexicography X. Proceedings of the Tenth International Symposium on Lexicography May 4-6, 2000 at the University of Copenhagen, Tübingen: Niemeyer, 231- 242. Meys, Willem J. (1975), Compound adjectives and the ideal speaker-listener. A study of compounding in a transformational-generative framework, Amsterdam - Oxford: North Holland. — (1985), “Morphological meanings and the structure of the mental lexicon”, in Geer A.J. Hoppenbrouwers, Pieter A.M. Seuren and Anton J.M.M. Weijters, eds., Meaning and the lexicon, Dordrecht: Foris, 76-85. Minsky, Marvin (1975), “A framework for representing knowledge”, in Patrick H. Winston, ed., The psychology of computer vision, New York: McGraw Hill, 211-277. Mittelberg, Irene and Cornelia Müller, eds. (in prep.), Metonymy in gesture and sign, Special Issue of CogniTextes. Müller, Cornelia (2008), Metaphors dead and alive, sleeping and waking. A dynamic view, Chicago: University of Chicago Press. Murphy, Gregory L. (1988), “Comprehending complex concepts”, Cognitive Science 12, 529-562. — (1996), “On metaphoric representation”, Cognition 60, 173-204. — (1997), “Reasons to doubt the present evidence for metaphoric representation”, Cognition 62, 99-108. 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. Niemeier, Susanne (2000), “Straight from the heart - metonymic and metaphorical explorations”, in Barcelona (2000a), 195-214. Nissim, Malvina and Katja Markert (2005), “Learning to buy a Renault and talk to BMW: A supervised approach to conventional metonymy”, in H. Bunt, ed., Proceedings of the Sixth International Workshop on Computational Semantics, Tilburg, January 2005 <www.comp.leeds.ac.uk/ markert/ Papers/ IWCS2005.pdf> (last accessed July 4, 2005). Nunberg, Geoffrey (1995), “Transfers of meaning”, Journal of Semantics 12, 109-132. Nunberg, Geoffrey, Ivan A. Sag and Thomas Wasow (1994), “Idioms”, Language 70(3), 491-538. Oakley, Todd and Anders Hougaard, eds. (2008), Mental spaces in discourse and interaction, Amsterdam - Philadelphia: John Benjamins. Onifer, William and David A. Swinney (1981), “Accessing lexical ambiguities during sentence comprehension: Effects of frequency of meaning and contextual bias”, Memory & Cognition 9(3), 225-236. Onishi, Kristine H. and Gregory L. Murphy (1993), “Metaphoric reference: When metaphors are not understood as easily as literal expressions”, Memory & Cognition 21(6), 763-772. 298 Ortony, Andrew (1979), “Beyond literal similarity”, Psychological Review 86(3), 161- 180. — (1980), “Some psycholinguistic aspects of metaphor”, in Honeck and Hoffman (1980), 69-83. — ed. (1993), Metaphor and thought, 2 nd ed., Cambridge: Cambridge University Press. 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. 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. Panther, Klaus-Uwe and Günter Radden, eds. (1999), Metonymy in language and thought, Amsterdam - Philadelphia: John Benjamins. Panther, Klaus-Uwe and Linda L. Thornburg (1998), “A cognitive approach to inferencing in conversation”, Journal of Pragmatics 30, 755-769. — (1999), “The potentiality for actuality metonymy in English and Hungarian”, in Panther and Radden (1999), 333-357. — , eds. (2003a), Metonymy and pragmatic inferencing, Amsterdam - Philadelphia: John Benjamins. — (2003b), “Introduction. On the nature of conceptual metonymy”, in Panther and Thornburg (2003a), 1-20. — (2004), “The role of conceptual metonymy in meaning construction”, metaphorik.de 06/ 2004, 91-116 <http: / / www.metaphorik.de/ 06/ pantherthornburg. pdf> (last accessed July 30, 2007). Panther, Klaus-Uwe, Linda L. Thornburg and Antonio Barcelona, eds. (2009), Metonymy and metaphor in grammar, Amsterdam - Philadelphia: John Benjamins. Papafragou, Anna (1996), “On metonymy”, Lingua 99, 169-195. Paprotté, Wolf and René Dirven, eds. (1985), The ubiquity of metaphor. Metaphor in language and thought, Amsterdam - Philadelphia: John Benjamins. Peirsman, Yves and Dirk Geeraerts (2006), “Metonymy as a prototypical category”, Cognitive Linguistics 17(3), 269-316. Peleg, Orna, Rachel Giora and Ofer Fein (2001), “Salience and context effects: Two are better than one”, Metaphor and Symbol 16(3&4), 173-192. Pickering, Martin J. and Matthew J. Traxler (1998), “Plausibility and recovery from garden paths: An eye-tracking study”, Journal of Experimental Psychology: Learning, Memory, and Cognition 24(4), 940-961. Plag, Ingo (1999), Morphological productivity. Structural constraints in English derivation, Berlin - New York: Mouton de Gruyter. — (2003), Word-formation in English, Cambridge: Cambridge University Press. — (2006), “Productivity”, in Bas Aarts and April McMahon, eds., The handbook of English linguistics, Oxford: Blackwell, 537-556. Pollio, Howard R., Michael K. Smith and Marilyn R. Pollio (1990), “Figurative language and cognitive psychology”, Language and Cognitive Processes 5, 141-167. Pragglejaz Group (2007), “MIP: A method for identifying metaphorically used words in discourse”, Metaphor and Symbol 22(1), 1-39. Pynte, Joël, Mirielle Besson, Fabrice-Henri Robichon and Jézabel Poli (1996), “The time-course of metaphor comprehension: An event-related potential study”, Brain and Language 55, 293-316. 299 Quinn, Naomi (1991), “The cultural basis of metaphor”, in James W. Fernandez, ed., Beyond metaphor. The theory of tropes in anthropology, Stanford/ CA: Stanford University Press. Radden, Günter (2000), “How metonymic are metaphors”, in Barcelona (2000a), 93- 108. Radden, Günter and Zoltán Kövecses (1999), “Towards a theory of metonymy”, in Panther and Radden (1999), 17-59. Radden, Günter and Klaus-Uwe Panther, eds. (2004a), Studies in linguistic motivation, Berlin - New York: Mouton de Gruyter. — (2004b), “Introduction: Reflections on motivation”, in Radden and Panther (2004a), 1-46. Rayner, Keith and Lyn Frazier (1989), “Selection mechanisms in reading lexically ambiguous words”, Journal of Experimental Psychology: Learning, Memory, and Cognition 15(5), 779-790. Récanati, François (1989), “The pragmatics of what is said”, Mind and Language 4, 295-329. — (1995), “The alleged priority of literal interpretation”, Cognitive Science 19, 207- 232. — (2004), Literal meaning, Cambridge etc.: Cambridge University Press. Reddy, Michael (1979), “The conduit metaphor - A case of frame conflict in our language about language”, in Andrew Ortony, ed., Metaphor and thought, Cambridge: Cambridge University Press, 284-324. Richards, Ivor A. (1925), Principles of literary criticism, London: Kegan Paul. — (1936), The philosophy of rhetoric, London etc.: Oxford University Press. Riemer, Nick (2002a), “When is a metonymy no longer a metonymy? ”, in Dirven and Pörings (2002), 379-406. — (2002b), “Remetonymizing metaphor: Hypercategories in semantic extension”, Cognitive Linguistics 12(4), 379-401. Ritchie, L. David (2004), “Lost in ‘conceptual space’: Metaphors of conceptual integration”, Metaphor and Symbol 19(1), 31-50. — (2006), Context and connection in metaphor, Houndmills etc.: Palgrave Macmillan. Rosch, Eleanor (1973), “On the internal structure of perceptual and semantic categories”, in Timothy E. Moore, ed., Cognitive development and the acquisition of language, New York: Academic Press, 111-144. — (1975), “Cognitive reference points”, Cognitive Psychology 7, 532-547. Rosch, Eleanor and Caroline B. Mervis (1975), “Family resemblances: Studies in the internal structure of categories”, Cognitive Psychology 7, 573-605. Ruiz de Mendoza Ibáñez, Francisco J. (2000), “The role of mappings and domains in understanding metonymy”, in Barcelona (2000a), 109-132. Ruiz de Mendoza Ibáñez, Francisco J. and Olga I. Díez Velasco (2002), “Patterns of conceptual interaction”, in Dirven and Pörings (2002), 489-532. — (2004), “Metonymic motivation in anaphoric reference”, in Radden and Panther (2004a), 293-320. Ruiz de Mendoza Ibáñez, Francisco J. and Lorena Pérez Hernández (2001), “Metonymy and the grammar: Motivation, constraints and interaction”, Language & Communication 21(4), 321-357. — (2003), “Cognitive operations and pragmatic implication”, in Panther and Thornburg (2003a), 23-49. Rumelhart, David E. (1980), “Schemata: The building blocks of cognition”, in Rand J. Spiro, Bertram C. Bruce and William F. Brewer, eds., Theoretical issues in read- 300 ing comprehension: Perspectives from cognitive psychology, linguistics, artificial intelligence, and education, Hillsdale/ NJ: Erlbaum, 33-58. — (1993), “Some problems with the notion of literal meanings”, in Ortony (1993), 71-82. Sauer, Hans (1992), Nominalkomposita im Frühmittelenglischen. Mit Ausblicken auf die Geschichte der englischen Nominalkomposition, Tübingen: Niemeyer. Saussure, Ferdinand de (1916; 1983), Course in general linguistics, Charles Bally and Albert Sechehaye, eds., translated and annotated by Roy Harris, London: Duckworth. Schank, Roger C. and Robert P. Abelson (1977), Scripts, plans, goals, and understanding: An inquiry into human knowledge structures, Hillsdale/ NJ: Erlbaum. Schlesewsky, Matthias and Ina Bornkessel (2003), “On incremental interpretation: Degrees of meaning accessed during sentence comprehension”, Lingua 114 (9/ 10), 1231-1234. Schmid, Hans-Jörg (1993), Cottage und Co., idea, start vs. begin. Die Kategorisierung als Grundprinzip einer differenzierten Bedeutungsbeschreibung, Tübingen: Niemeyer. — (1998), “Functions of metaphor”, in Karin Pittner and Robert J. Pittner, eds., Beiträge zu Sprache & Sprachen 2. Vorträge der 5. Münchner Linguistik-Tage, 1995, München: Lincom Europa, 163-172. — (2005), Englische Morphologie und Wortbildung. Eine Einführung, Berlin: Erich Schmidt. — (2007), “Entrenchment, salience and basic levels”, in Dirk Geeraerts and Hubert Cuyckens, eds., Handbook of cognitive linguistics, Oxford: Oxford University Press, 117-138. — (2008), “New words in the mind: Concept-formation and entrenchment of neologisms”, Anglia 126, 1-36. — (forthcoming a), “Conceptual blending, relevance and novel N+N-compounds”, in Sandra Handl and Hans-Jörg Schmid, eds., Windows to the mind: Metaphor, metonymy, and conceptual blending, Berlin - New York: Mouton de Gruyter. — (forthcoming b), “Does frequency in text instantiate entrenchment in the cognitive system? ”, in Dylan Glynn and Kerstin Fischer, eds., Quantitative methods in Cognitive Semantics. Corpus-driven approaches, Berlin - New York: Mouton de Gruyter. Schoen, Lawrence M. (1988), “Semantic flexibility and core meaning“, Journal of Psycholinguistic Research 17(2), 113-123. Searle, John R. (1978), “Literal meaning”, Erkenntnis 13, 207-224. — (1979), Expression and meaning. Studies in the theory of speech acts, Cambridge etc.: Cambridge University Press. Seidenberg, Mark S., Michael K. Tanenhaus, James M. Leiman and Marie Bienkowski (1982), “Automatic access of the meanings of ambiguous words in context: Some limitations of knowledge-based processing”, Cognitive Psychology 14, 489- 537. Semino, Elena (2006), “A corpus-based study of metaphors for speech activity in British English”, in Stefanowitsch and Gries (2006), 36-60. Sereno, Sara C., Jeremy N. Pacht and Keith Rayner (1992), “The effect of meaning frequency on processing lexically ambiguous words: Evidence from eye fixations”, Psychological Science 3(5), 296-300. Shen, Yeshayahu (1991), “Schemata, categories, and metaphor comprehension”, Poetics Today 12(1), 111-124. — (1992), “Metaphors and categories”, Poetics Today 13(4), 771-794. 301 Shinjo, Makiko and Jerome L. Myers (1987), “The role of context in metaphor comprehension”, Journal of Memory and Language 26, 226-241. Simpson, Greg B. (1981), “Meaning dominance and semantic context in the processing of lexical ambiguity”, Journal of Verbal Learning and Verbal Behavior 20(1), 120- 136. Sperber, Dan and Deirdre Wilson (1995), Relevance. Communication and cognition, 2 nd ed., Oxford: Blackwell. Stallard, David (1993), “Two kinds of metonymy”, in no ed., Proceedings of the 31 st Annual Meeting of the Association for Computational Linguistics, Columbus, Ohio, 22- 26 June 1993, Columbus/ OH <http: / / acl.ldc.upenn.edu/ P/ P93/ P93-1012.pdf> (last accessed August 20, 2007). Steen, Gerard J. (2007), Finding metaphor in grammar and usage. A methodological analysis of theory and research, Amsterdam - Philadelphia: John Benjamins. Steen, Gerard J., Ewa A. Biernacka, Lettie Dorst, Anna A. Kaal, Irene López- Rodríguez and Tryntje Pasma (in print), “Pragglejaz in practice: Finding metaphorically used words in natural discourse”, in Graham Low, Lynne Cameron, Alice Deignan and Zazie Todd, eds., Metaphor in the real world, Amsterdam - Philadelphia: John Benjamins. Stefanowitsch, Anatol (2006a), “Corpus-based approaches to metaphor and metonymy”, in Stefanowitsch and Gries (2006), 1-16. — (2006b), “Words and their metaphors: A corpus-based approach”, in Stefanowitsch and Gries (2006), 63-105. Stefanowitsch, Anatol and Stefan Th. Gries, eds. (2006), Corpus-based approaches to metaphor and metonymy, Berlin - New York: Mouton de Gruyter. Štekauer, Pavol and Rochelle Lieber, eds. (2005), Handbook of word-formation, Dordrecht: Springer. Stern, Gustaf (1931), Meaning and change of meaning with special reference to the English language, Bloomington/ IN: Indiana University Press. Svanlund, Jan (2007), “Metaphor and convention”, Cognitive Linguistics 18(1), 47-89. Sweetser, Eve (1990), From etymology to pragmatics: Metaphorical and cultural aspects of semantic structure, Cambridge: Cambridge University Press. — (1999), “Compositionality and blending. Semantic composition in a cognitively realistic framework”, in Theo Janssen and Gisela Redeker, eds., Cognitive linguistics: Foundations, scope and methodology, Berlin - New York: Mouton de Gruyter, 129-162. — (2000), “Blended spaces and performativity”, Cognitive Linguistics 11(3/ 4), 305- 333. Swinney, David A. (1979), “Lexical access during sentence comprehension: (Re)Consideration of context effects”, Journal of Verbal Learning and Verbal Behavior 18, 645-659. Swinney, David A. and Anne Cutler (1979), “The access and processing of idiomatic expressions”, Journal of Verbal Learning and Verbal Behavior 18(5), 523-534. Tabossi, Patrizia (1988), “Accessing lexical ambiguity in different types of sentential contexts”, Journal of Memory and Language 27, 324-340. — (1996), “Cross-modal semantic priming”, Language and Cognitive Processes 11(6), 569-576. Tanenhaus, Michael K., James M. Leiman and Mark S. Seidenberg (1979), “Evidence for multiple stages in the processing of ambiguous words in syntactic contexts”, Journal of Verbal Learning and Verbal Behavior 18, 427-440. 302 Tartter, Vivien C., Hilary Gomes, Boris Dubrovsky, Sophie Molholm and Rosemarie Vala Stewart (2002), “Novel metaphors appear anomalous at least momentarily: Evidence from N400”, Brain and Language 80, 488-509. Taylor, John R. (1995), Linguistic categorization. Prototypes in linguistic theory, 2 nd ed., Oxford: Clarendon. — (2002), “Category extension by metonymy and metaphor”, in Dirven and Pörings (2002), 323-347. — (2003), Linguistic categorization, 3 rd ed., Oxford etc.: Oxford University Press. — (2004), “The ecology of constructions”, in Radden and Panther (2004a), 49-73. Taylor, John R. and Robert MacLaury, eds. (1995), Language and the cognitive construal of the world, Berlin - New York: Mouton de Gruyter. Tourangeau, Roger and Lance Rips (1991), “Interpreting and evaluating metaphors”, Journal of Memory and Language 30, 452-472. Tourangeau, Roger and Robert J. Sternberg (1981), “Aptness in metaphor”, Cognitive Psychology 13, 27-55. Tournier, Jean (1985), Introduction descriptive à la lexicogénétique de l’anglais contemporain, Paris - Genève: Champion-Slatkine. Traugott, Elizabeth C. (1985a), “’Conventional’ and ‘dead’ metaphors revisited”, in Paprotté and Dirven (1985), 17-56. — (1985b), “On regularity in semantic change”, Journal of Literary Semantics 14(2), 155-173. Traxler, Matthew J., Martin J. Pickering and Brian McElree (2002), “Coercion in sentence processing: Evidence from eye-movements and self-paced reading”, Journal of Memory and Language 47(4), 530-547. Trim, Richard (2007), Metaphor networks. The comparative evolution of figurative language, Houndmills etc.: Palgrave Macmillan. Tuggy, David (1993), “Ambiguity, polysemy, and vagueness”, Cognitive Linguistics 4(3), 273-290. Turner, Mark (1991), Reading minds. The study of English in the age of cognitive science, Princeton/ NJ: Princeton University Press. Tversky, Amos (1977), “Features of similarity”, Psychological Review 84(4), 327-352. Tyler, Andrea and Vyvyan Evans (2003), The semantics of English prepositions. Spatial scenes, embodied meaning and cognition, Cambridge etc.: Cambridge University Press. Ullmann, Stephen (1957), The principles of semantics, Oxford: Basil Blackwell. — (1962), Semantics. An introduction to the science of meaning, Oxford: Basil Blackwell. Ungerer, Friedrich (1991a), “What makes a linguistic sign successful? Towards a pragmatic interpretation of the linguistic sign”, Lingua 83, 155-181. — (1991b), “Acronyms, trade names and motivation”, Arbeiten aus Anglistik und Amerikanistik 16(2), 131-158. Ungerer, Friedrich and Hans-Jörg Schmid (1998), “Englische Komposita und Kategorisierung”, Rostocker Beiträge zur Sprachwissenschaft 5, 77-98. — (2006), An introduction to cognitive linguistics, 2 nd ed., Harlow etc.: Pearson/ Longman. Van Dijk, Teun A. and Walter Kintsch (1983), Strategies of discourse comprehension, Orlando etc.: Academic Press. Veale, Tony and Diarmuid O’Donoghue (2000), “Computation and blending”, Cognitive Linguistics 11(3/ 4), 253-281. 303 Vu, Hoang, George Kellas, Kimberly Metcalf and Ruth Herman (2000), “The influence of global discourse on lexical ambiguity resolution”, Memory & Cognition 28(2), 236-252. Vu, Hoang, George Kellas and Stephen T. Paul (1998), “Sources of sentence constraint on lexical ambiguity resolution”, Memory & Cognition 26(5), 979-1001. Warren, Beatrice (1992), Sense developments. A contrastive study of the developments of slang senses and novel standard senses in English, Stockholm: Almqvist & Wiksell. — (1999), “Aspects of referential metonymy”, in Panther and Radden (1999), 121- 135. Weinrich, Harald (1958), “Münze und Wort. Untersuchungen an einem Bildfeld”, in Heinrich Lausberg and Harald Weinrich, eds. Romanica. Festschrift für Gerhard Rohlfs, Halle: Niemeyer, 508-521. Wiley, Jennifer and Keith Rayner (2001), “Effects of titles on the processing of text and lexically ambiguous words: Evidence from eye movements”, Memory & Cognition 28(6), 1011-1021. Williams, John N. (1992), “Processing polysemous words in context: Evidence for interrelated meanings”, Journal of Psycholinguistic Research 21(3), 193-218. Dictionaries Fowler, Henry, W. (1983), A dictionary of modern English usage, 2 nd ed., Oxford: Oxford University Press. Pearsall, Judy, ed. (1998), The New Oxford Dictionary of English, Oxford: Oxford University Press. (= NODE) Rundell, Michael and Gwyneth Fox, eds. (2002), Macmillan English Dictionary for Advanced Learners, Oxford: Macmillan Education. (= MED) Simpson, John A. and Edmund S.C. Weiner, eds. (1989), The Oxford English Dictionary. Second Edition, 20 vols., Oxford: Oxford University Press. (= OED) Summers, Della, ed. (2003), Longman Dictionary of Contemporary English. 4 th ed., Harlow: Pearson Education. (= LDOCE4) Corpora The British National Corpus. World Edition, December 2000 Release, The Humanities Computing Unit of Oxford University. The Mascara corpus, <http: / / www.comp.leeds.ac.uk/ markert/ Papers/ index.html> (last accessed August 20, 2007). Appendix A: The results of the metonymy study Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) WHOLE FOR PART : INSTITU- TION FOR PEOPLE RESPON- SIBLE 20.41 66.31 PART FOR PART : CON- TROLLER FOR CON- TROLLED ‘planes’ 10.04 32.62 airlines 916 30.79 WHOLE FOR PART : INSTITU- TION FOR STOCK MARKET SHARES / INDEX 0.33 1.06 WHOLE FOR PART : WHOLE FOR PRINCIPAL PART 62.50 98.43 WHOLE FOR PART : WHOLE FOR PRINCIPAL PART PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE : COUNTRY FOR RULERS 3.30 5.20 WHOLE FOR PART : WHOLE FOR PRINCIPAL PART PART FOR WHOLE : LOCATION FOR PEOPLE THERE 1.70 2.68 WHOLE FOR PART : WHOLE FOR PRINCIPAL PART PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE : COUNTRY FOR MILITARY FORCES 0.80 1.26 PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE 0.70 1.10 America 1000 (NP0) 63.50 PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE : COUNTRY FOR RULERS 0.30 0.47 306 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) WHOLE FOR PART : WHOLE FOR PRINCIPAL PART PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR EVENT 0.10 0.16 appendicitis 61 1.64 PART FOR WHOLE : AT- TRIBUTE OF ENTITY FOR ENTITY : ATTRIBUTE OF PERSON FOR PERSON 1.64 100.00 aspirin 313 30.03 PART FOR WHOLE : MEM- BER OF CATEGORY FOR CATEGORY : BRAND NAME FOR CATEGORY OF PRO- DUCT 30.03 100.00 bathroom 1000 7.20 WHOLE FOR PART : WHOLE FOR PRINCIPAL PART ‘fitments’ ‘WC’ ‘tub’ ‘walls’ ‘floor’ 7.20 3.50 3.30 0.20 0.10 0.10 100.00 48.61 45.83 2.78 1.39 1.39 blood 1000 (NN1) 1.70 PART FOR WHOLE : PRINCI- PAL PART FOR WHOLE : BODY PART FOR PERSON 1.70 100.00 BMW 517 70.21 PART FOR PART : PRO- DUCER FOR PRODUCT : PRODUCER FOR PHYSICAL PRODUCT : PRODUCER FOR CONSUMER PRODUCT 62.67 89.26 307 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) WHOLE FOR PART : INSTITU- TION FOR PEOPLE RESPON- SIBLE 7.54 10.74 PART FOR PART : PRODUCER FOR PRODUCT : PRODUCER FOR PHYSICAL PRODUCT : PRODUCER FOR CONSUMER PRODUCT PART FOR PART : CON- TROLLED FOR CONTROL- LER ‘driver’ 4.64 6.61 PART FOR PART : CON- TAINER FOR CONTENT 44.30 89.31 PART FOR PART : PRODUCT FOR PRODUCER : ABSTRACT PRODUCT FOR PRODUCER : WORK FOR AUTHOR 5.30 10.69 PART FOR PART : CON- TAINER FOR CONTENT PART FOR WHOLE : MEMBER OF CATEGORY FOR CATEGORY ‘genre’ 0.10 0.20 book 1000 (NN1) 49.60 PART FOR PART : CON- TAINER FOR CONTENT PART FOR PART : CON- TENT FOR AMOUNT OF CONTENT 0.20 0.40 PART FOR PART : SOURCE FOR PRODUCT : LOCATION FOR PRODUCT MADE THERE 11.71 67.31 Bordeaux 299 17.39 PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE 3.01 17.31 308 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE : LOCATION FOR RULERS 1.00 5.77 PART FOR WHOLE : LOCATI- ON FOR ENTITY THERE : LO - CATION FOR INSTITUTION ‘university’ ’factory’ 1.34 0.67 0.67 7.70 3.85 3.85 PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE : LOCATION FOR SPORTS TEAM 0.33 1.92 PART FOR PART : CON- TAINER FOR CONTENT 9.90 99.00 bottle 1000 (NN1) 10.00 WHOLE FOR PART : WHOLE FOR PRINCIPAL PART ‘label’ 0.10 1.00 PART FOR WHOLE : PRINCI- PAL PART FOR WHOLE : BODY PART FOR PERSON 1.50 60.00 brain 1000 (NN1) 2.50 PART FOR PART : CAUSE FOR EFFECT : OBJECT CAUSING MENTAL STATE FOR MEN- TAL STATE ‘intelligence’ 1.00 40.00 309 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR WHOLE : ATTRI- BUTE OF ENTITY FOR ENTI - TY : ATTRIBUTE OF OBJECT FOR OBJECT : MATERIAL OF OBJECT FOR OBJECT ’medal’ ‘work of art’ 8.90 4.60 4.30 97.80 50.55 47.25 bronze 1000 (NN1) 9.10 PART FOR WHOLE : ATTRIB- UTE OF ENTITY FOR ENTITY : ATTRIBUTE OF OBJECT FOR OBJECT : ATTRIBUTE OF ANIMAL FOR ANIMAL 0.20 2.20 PART FOR WHOLE : CENT- RAL FACTOR FOR INSTITU- TION : CENTRAL INSTRU- MENT FOR INSTITUTION ‘bus service’ 8.50 88.54 PART FOR PART : CON- TROLLED FOR CONTROL- LER 1.00 10.42 buses 1000 (NN2) 9.60 PART FOR PART : NON - HUMAN PRODUCER FOR PRODUCT : PRODUCER FOR PHYSICAL PRODUCT ‘exhaust’ 0.10 1.04 310 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) butt 279 (NN1) 1.43 PART FOR WHOLE : PRINCI- PAL PART FOR WHOLE : BODY PART FOR PERSON 1.43 100.00 cameras 1000 9.20 PART FOR PART : CON- TROLLED FOR CONTROL- LER 9.20 100.00 WHOLE FOR PART : WHOLE FOR PRINCIPAL PART ‘motor’ ’exterior’ ’interior’ ’tank’ ’number plate’ ’tyres’ 4.10 1.80 1.60 0.40 0.10 0.10 0.10 73.22 32.14 28.57 7.14 1.79 1.79 1.79 car 1000 (NN1) 5.60 PART FOR WHOLE : MEMBER OF CATEGORY FOR CATE- GORY ‘model’ ‘all cars’ 1.50 1.20 0.30 26.79 21.43 5.36 PART FOR PART : SOURCE FOR PRODUCT : ANIMAL FOR MEAT OF ANIMAL 68.70 100.00 chicken 1000 (NN1) 68.70 PART FOR PART : SOURCE FOR PRODUCT : ANIMAL FOR MEAT OF ANIMAL PART FOR PART : CON- TENT FOR CONTAINER ‘oven, frying pan’ 0.20 0.29 311 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE : COUNTRY FOR RULERS 26.00 62.05 PART FOR PART : SOURCE FOR PRODUCT : LOCATION FOR PRODUCT MADE THERE 12.90 30.79 PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE 0.90 2.15 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR INSTITU- TION ‘economy’ 0.90 2.15 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR INSTITU- TION : LOCATION FOR STATE FORM 0.70 1.67 PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE : LOCATION FOR SPORTS TEAM 0.30 0.72 China/ china 1000 41.90 PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE : COUNTRY FOR MILITARY FORCES 0.20 0.48 312 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR INSTITU- TION : BUILDING FOR INSTITUTION 55.90 93.32 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR INSTITU- TION : BUILDING FOR INSTITUTION WHOLE FOR PART : INSTITUTION FOR PEOPLE RESPONSIBLE 10.30 17.20 church 1000 (NN1) 59.90 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR EVENT : LOCATION FOR ACTIVITY OF LOCATION ‘service’ 4.00 6.68 Connery 115 3.48 PART FOR PART : CON- TROLLER FOR CON- TROLLED : ACTOR FOR ROLE 3.48 100.00 convent 524 (NN1) 12.60 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR INSTITU- TION 12.60 100.00 crime 1000 5.90 PART FOR PART : CON- TROLLED FOR CONTROL- LER : ACTIVITY OF PERSON FOR PERSON 5.00 84.75 313 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR WHOLE : ATTRIB- UTE OF ENTITY FOR ENTITY : ATTRIBUTE OF OBJECT FOR OBJECT : MAIN TOPIC FOR GENRE 0.90 15.25 PART FOR WHOLE : AT- TRIBUTE OF ENTITY FOR ENTITY : ATTRIBUTE OF PERSON FOR PERSON : COSTUME FOR WEARER 65.00 94.07 crown 1000 (NN1) 69.10 PART FOR WHOLE : CENT- RAL FACTOR FOR INSTITU- TION / PROFESSION : CENT- RAL FACTOR FOR INSTITUTION 4.00 5.93 diesel 1000 15.10 PART FOR WHOLE : AT- TRIBUTE OF ENTITY FOR ENTITY : ATTRIBUTE OF OBJECT FOR OBJECT ‘car, locomotive’ 15.10 100.00 dish 1000 (NN1) 32.70 PART FOR PART : CON- TAINER FOR CONTENT ‘meal’ 32.70 100.00 drugs 1000 (NN2) 55.50 WHOLE FOR PART : CATE- GORY FOR SALIENT MEM- BER 55.50 100.00 314 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR PART : EFFECT FOR CAUSE : EMOTION FOR CAUSE OF EMOTION 44.70 100.00 ecstasy 604 44.70 PART FOR PART : EFFECT FOR CAUSE : EMOTION FOR CAUSE OF EMOTION PART FOR WHOLE : MEMBER OF CATEGORY FOR CATEGORY 0.83 1.85 PART FOR PART : PRODUC - ER FOR PRODUCT : PRO- DUCER FOR ABSTRACT PRODUCT : THEORIST FOR THEORY 2.26 64.29 PART FOR WHOLE : MEMBER OF CATEGORY FOR CATE- GORY 0.75 21.43 PART FOR PART : PRODUCER FOR PRODUCT : PRODUCER FOR ABSTRACT PRODUCT : AUTHOR FOR WORKS 0.50 14.29 Einstein 398 3.52 PART FOR PART : PRODUCER FOR PRODUCT : PRODUCER FOR ABSTRACT PRODUCT : THEORIST FOR THEORY WHOLE FOR PART : EVENT FOR POINT IN TIME 0.25 7.14 Exxon 145 35.86 WHOLE FOR PART : INSTI- TUTION FOR PEOPLE RESPONSIBLE 35.17 98.08 315 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR WHOLE : MEMBER OF CATEGORY FOR CATE- GORY 0.69 1.92 PART FOR WHOLE : PRINCI- PAL PART FOR WHOLE : BODY PART FOR PERSON 17.50 50.72 PART FOR WHOLE : PRINCI- PAL PART FOR WHOLE ’head’ 0.10 0.29 faces 1000 (NN2) 34.50 WHOLE FOR PART : WHOLE FOR PRINCIPAL PART ‘facial expression’ ’mouth’ 16.90 16.50 0.40 48.99 47.83 1.16 PART FOR WHOLE : PRINCI- PAL PART FOR WHOLE : PLAYING PIECE FOR EN- TIRE GAME 96.80 100.00 PART FOR WHOLE : PRINCI- PAL PART FOR WHOLE : PLAYING PIECE FOR ENTIRE GAME PART FOR PART : CON- TROLLED FOR CONTROL- LER : ACTIVITY OF PERSON FOR PERSON 1.70 1.76 football 1000 (NN1) 96.80 PART FOR WHOLE : PRINCI- PAL PART FOR WHOLE : PLAYING PIECE FOR ENTIRE GAME PART FOR WHOLE : CENTRAL FACTOR FOR INSTITUTION / PROFES - SION : CENTRAL INSTRU- MENT FOR PROFESSION 1.10 1.14 316 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) four wheel drive 30 3.33 PART FOR WHOLE : AT- TRIBUTE OF ENTITY FOR ENTITY : ATTRIBUTE OF OBJECT FOR OBJECT 3.33 100.00 furs 150 65.33 PART FOR WHOLE : ATTRI - BUTE OF ENTITY FOR ENTI - TY : ATTRIBUTE OF OBJECT FOR OBJECT : MATERIAL OF OBJECT FOR OBJECT 65.33 100.00 garage 1000 (NN1) 4.60 WHOLE FOR PART : INSTI- TUTION FOR PEOPLE RESPONSIBLE 4.60 100.00 PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE : COUNTRY FOR RULERS 12.00 58.82 PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE 5.50 26.96 PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE : LOCATION FOR SPORTS TEAM 2.00 9.80 Germany 1000 20.40 PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE : COUNTRY FOR MILITARY FORCES 0.90 4.41 317 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR WHOLE : ATTRIB- UTE OF ENTITY FOR ENTITY : ATTRIBUTE OF OBJECT FOR OBJECT : MATERIAL OF OBJECT FOR OBJECT 98.40 100.00 PART FOR WHOLE : ATTRIB- UTE OF OBJECT FOR OBJECT : MATERIAL OF OBJECT FOR OBJECT PART FOR WHOLE : PRINCIPAL PART FOR WHOLE ‘spectacles’ 55.10 56.00 PART FOR WHOLE : ATTRIB- UTE OF OBJECT FOR OBJECT : MATERIAL OF OBJECT FOR OBJECT PART FOR PART : CON- TAINER FOR CONTENT 2.00 2.03 glasses 1000 98.40 PART FOR WHOLE : ATTRIB- UTE OF OBJECT FOR OBJECT : MATERIAL OF OBJECT FOR OBJECT PART FOR WHOLE : PRINCIPAL PART FOR WHOLE ‘spectacles’ PART FOR WHOLE : ATTRIBUTE OF ENTITY FOR EN- TITY : ATTRIBUTE OF PERSON FOR PERSON : COSTUME FOR WEARER 0.10 0.10 government 1000 (NN1) 56.20 WHOLE FOR PART : INSTI- TUTION FOR PEOPLE RESPONSIBLE 56.20 100.00 hands 1000 (NN2) 2.40 PART FOR WHOLE : PRINCI- PAL PART FOR WHOLE : BODY PART FOR PERSON 2.40 100.00 318 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR WHOLE : PRINCI- PAL PART FOR WHOLE : BODY PART FOR PERSON PART FOR PART : PRO- DUCT FOR PRODUCER : PHYSICAL PRODUCT FOR PRODUCER : ARTISTIC PRODUCT FOR PRODUCER 0.10 4.17 PART FOR WHOLE : PRINCI- PAL PART FOR WHOLE : BODY PART FOR PERSON 4.40 51.76 PART FOR WHOLE : ATTRIB- UTE OF ENTITY FOR ENTITY : ATTRIBUTE OF OBJECT FOR OBJECT ’side of coin’ 0.10 1.18 heads 1000 (NN2) 8.50 WHOLE FOR PART : WHOLE FOR PRINCIPAL PART ‘brain’ ’face’ 4.00 3.90 0.10 47.06 45.88 1.18 PART FOR PART : BODY PART FOR CONTIGUOUS LOCA - TION 7.40 18.32 WHOLE FOR PART : OBJECT FOR ITS ATTRIBUTE : BODY PART FOR ITS COSTUME 33.00 81.68 heels 1000 40.40 WHOLE FOR PART : OBJECT FOR ITS ATTRIBUTE : BODY PART FOR ITS COSTUME PART FOR WHOLE : PRINCIPAL PART FOR WHOLE ‘shoe’ 13.80 34.16 319 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) WHOLE FOR PART : OBJECT FOR ITS ATTRIBUTE : BODY PART FOR ITS COSTUME PART FOR WHOLE : PRINCIPAL PART FOR WHOLE ‘shoe’ PART FOR WHOLE : ATTRIBUTE OF ENTITY FOR EN- TITY : ATTRIBUTE OF PERSON FOR PERSON : COSTUME FOR WEARER 0.40 0.99 WHOLE FOR PART : OBJECT FOR ITS ATTRIBUTE : BODY PART FOR ITS COSTUME PART FOR PART : CAUSE FOR EFFECT ‘sound of heels’ 0.10 0.25 joy 1000 (NN1) 13.70 PART FOR PART : EFFECT FOR CAUSE : EMOTION FOR CAUSE OF EMOTION 13.70 100.00 ketchup 84 0.00 PART FOR PART : CONTENT FOR CONTAINER 0.00 0.00 PART FOR PART : CON- TAINER FOR CONTENT 19.46 88.24 WHOLE FOR PART : WHOLE FOR PRINCIPAL PART 1.42 6.42 PART FOR PART : CON- TAINER FOR CONTENT PART FOR PART : CON- TENT FOR AMOUNT OF CONTENT 1.18 5.35 kettle 848 22.05 WHOLE FOR PART : OBJECT FOR LOCATION OF OBJECT ‘cooker’ 1.18 5.35 320 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) WHOLE FOR PART : WHOLE FOR PRINCIPAL PART ‘fittings/ fitments’ 4.60 49.46 PART FOR PART : SOURCE FOR PRODUCT : LOCATION FOR PRODUCT MADE THERE ‘cuisine’ 3.90 41.94 kitchen 1000 (NN1) 9.30 PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE ‘staff’ 0.80 8.60 PART FOR WHOLE : ATTRIB- UTE OF ENTITY FOR ENTITY : ATTRIBUTE OF OBJECT FOR OBJECT 29.60 93.67 PART FOR WHOLE : ATTRIB- UTE OF ENTITY FOR ENTITY : ATTRIBUTE OF OBJECT FOR OBJECT PART FOR PART : PRO- DUCT FOR PRODUCER : PHYSICAL PRODUCT FOR PRODUCER : CONSUMER PRODUCT FOR INSTITU- TION 25.50 80.70 label 1000 (NN1) 31.60 PART FOR WHOLE : ATTRIB- UTE OF ENTITY FOR ENTITY : ATTRIBUTE OF OBJECT FOR OBJECT PART FOR PART : PRO- DUCT FOR PRODUCER : PHYSICAL PRODUCT FOR PRODUCER : CONSUMER PRODUCT FOR INSTITU- TION WHOLE FOR PART : INSTITUTION FOR PEOPLE RESPON- SIBLE 4.10 12.97 321 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR WHOLE : ATTRIB- UTE OF ENTITY FOR ENTITY : ATTRIBUTE OF OBJECT FOR OBJECT PART FOR PART : CON- TAINER FOR CONTENT 2.70 8.54 PART FOR PART : PRODUCT FOR PRODUCER : ABSTRACT PRODUCT FOR PRODUCER : WORK FOR AUTHOR 2.00 6.33 PART FOR WHOLE : ATTRIB- UTE OF ENTITY FOR ENTITY : ATTRIBUTE OF OBJECT FOR OBJECT PART FOR PART : PRO- DUCT FOR PRODUCER : PHYSICAL PRODUCT FOR PRODUCER : CONSUMER PRODUCT FOR INSTITU- TION WHOLE FOR PART : INSTITUTION FOR INDEX 0.20 0.63 PART FOR PART : PRODUCT FOR PRODUCER : ABSTRACT PRODUCT FOR PRODUCER : WORK FOR AUTHOR 10.70 50.95 law 1000 (NN1) 21.00 PART FOR WHOLE : CENT- RAL FACTOR FOR INSTITU- TION / PROFESSION : CENT- RAL INSTRUMENT FOR PROFESSION 10.30 49.05 322 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) money 1000 0.40 PART FOR WHOLE : ATTRI - BUTE OF ENTITY FOR ENTI - TY : ATTRIBUTE OF PERSON FOR PERSON : POSSESSED FOR POSSESSOR 0.40 100.00 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR INSTITU- TION : CAPITAL CITY FOR GOVERNMENT 24.20 98.37 Moscow 1000 24.60 PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE : LOCATION FOR SPORTS TEAM 0.40 1.63 PART FOR PART : CON- TROLLER FOR CON- TROLLED : RULER FOR ARMY 13.34 80.80 PART FOR WHOLE : MEMBER OF CATEGORY FOR CATE- GORY 2.77 16.80 PART FOR PART : CONTROL- LER FOR CONTROLLED : PERSON FOR ACTIVITY OF PERSON 0.26 1.60 Napoleon 757 16.51 PART FOR PART : PERSON FOR THEIR TIME / STYLE 0.13 0.80 323 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR PART : PRODUCT FOR PRODUCER : PHYSICAL PRODUCT FOR PRODUCER : PHYSICAL PRODUCT FOR INSTITUTION 23.20 65.54 newspaper 1000 35.40 PART FOR PART : PRODUCT FOR PRODUCER : ABSTRACT PRODUCT FOR PRODUCER 12.20 34.46 WHOLE FOR PART : WHOLE FOR PRINCIPAL PART : OBJECT FOR MATERIAL OF OBJECT 44.50 100.00 WHOLE FOR PART : WHOLE FOR PRINCIPAL PART : OBJECT FOR MATERIAL OF OBJECT WHOLE FOR PART : WHOLE FOR PRINCIPAL PART ’taste’ ‘colour’ 0.80 0.50 0.30 1.80 1.13 0.67 oak 1000 (NN1) 44.50 WHOLE FOR PART : WHOLE FOR PRINCIPAL PART : OBJECT FOR MATERIAL OF OBJECT PART FOR WHOLE : ATTRIBUTE OF ENTITY FOR ENTITY : ATTRIBUTE OF OBJECT FOR OBJECT : MATERIAL OF OBJECT FOR OBJECT ‘cask’ 0.40 0.90 324 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR PART : PRODUCT FOR PRODUCER : ABSTRACT PRODUCT FOR PRODUCER : ABSTRACT PRODUCT FOR INSTITUTION 9.70 59.15 WHOLE FOR PART : EVENT FOR LOCATION 6.70 40.85 opera 1000 (NN1) 16.40 PART FOR PART : PRODUCT FOR PRODUCER : ABSTRACT PRODUCT FOR PRODUCER : ABSTRACT PRODUCT FOR INSTITUTION WHOLE FOR PART : INSTITUTION FOR PEOPLE RESPONSIBLE 5.10 31.10 PART FOR PART : CON- TAINER FOR CONTENT 7.40 91.36 WHOLE FOR PART : WHOLE FOR PRINCIPAL PART ‘label’ 0.70 8.64 packet 1000 (NN1) 8.10 WHOLE FOR PART : WHOLE FOR PRINCIPAL PART ‘label’ PART FOR PART : PRO- DUCT FOR PRODUCER : ABSTRACT PRODUCT FOR PRODUCER 0.50 6.17 paper 1000 (NN1) 49.20 PART FOR WHOLE : ATTRIB- UTE OF ENTITY FOR ENTITY : ATTRIBUTE OF OBJECT FOR OBJECT : MATERIAL OF OBJECT FOR OBJECT ‘piece of writing’ 49.20 100.00 325 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR WHOLE : ATTRIB- UTE OF ENTITY FOR ENTITY : ATTRIBUTE OF OBJECT FOR OBJECT : MATERIAL OF OBJECT FOR OBJECT PART FOR PART : PRO- DUCT FOR PRODUCER : ABSTRACT PRODUCT FOR PRODUCER : WORK FOR AUTHOR 10.00 20.33 PART FOR WHOLE : ATTRIB- UTE OF ENTITY FOR ENTITY : ATTRIBUTE OF OBJECT FOR OBJECT : MATERIAL OF OBJECT FOR OBJECT PART FOR PART : PRO- DUCT FOR PRODUCER : PHYSICAL PRODUCT FOR PRODUCER : PHYSICAL PRODUCT FOR INSTITU- TION ‘newspaper company’ 4.00 8.13 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR INSTITU- TION : CAPITAL CITY FOR GOVERNMENT 1.60 36.36 PART FOR WHOLE : LOCA- TION FOR ENTITY THERE : LOCATION FOR INSTITU- TION ‘centre of fashion’ ’stock market’ 1.00 0.80 0.20 22.73 18.18 4.55 Paris 1000 (NP0) 4.40 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR EVENT 0.90 20.45 326 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE 0.80 18.18 PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE : LOCATION FOR SPORTS TEAM 0.10 2.27 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR EVENT 31.03 100.00 Pearl Harbor/ Pearl Harbour 60/ 27 = 87 31.03 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR EVENT PART FOR WHOLE : MEMBER OF CATEGORY FOR CATEGORY 2.30 7.41 PART FOR PART : CON- TROLLED FOR CONTROL- LER ‘writer’ 5.10 59.30 pen 1000 (NN1) 8.60 PART FOR PART : CAUSE FOR EFFECT ‘writing’ 3.50 40.70 WHOLE FOR PART : WHOLE FOR PRINCIPAL PART 26.10 98.86 WHOLE FOR PART : WHOLE FOR PRINCIPAL PART PART FOR PART : CAUSE FOR EFFECT ‘mark’ 23.40 88.64 pencil 1000 (NN1) 26.40 PART FOR PART : CON- TROLLED FOR CONTROLLER 0.30 1.14 327 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR INSTITU- TION : BUILDING FOR INSTITUTION 80.13 100.00 Pentagon 297 80.13 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR INSTITU- TION : BUILDING FOR INSTITUTION WHOLE FOR PART : INSTITUTION FOR PEOPLE RESPONSIBLE 49.16 61.34 PART FOR PART : PRO- DUCER FOR PRODUCT : PRODUCER FOR PHYSICAL PRODUCT : PRODUCER FOR ARTISTIC PRODUCT 8.32 97.06 Picasso 793 8.58 PART FOR WHOLE : MEMBER OF CATEGORY FOR CATE- GORY 0.25 2.94 pill 574 (NN1) 44.95 WHOLE FOR PART : CATE- GORY FOR SALIENT MEM- BER 44.95 100.00 pint 1000 37.50 PART FOR PART : UNIT OF MEASURE FOR AMOUNT OF SUBSTANCE 37.50 100.00 328 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) plastic 1000 2.50 PART FOR WHOLE : ATTRI- BUTE OF ENTITY FOR ENTI - TY : ATTRIBUTE OF OBJECT FOR OBJECT : MATERIAL OF OBJECT FOR OBJECT ’other’ ‘credit card’ 2.50 2.00 0.50 100.00 80.00 20.00 power 1000 (NN1) 1.90 PART FOR WHOLE : ATTRI - BUTE OF ENTITY FOR ENTI - TY : ATTRIBUTE OF PERSON FOR PERSON : POSSESSED FOR POSSESSOR 1.90 100.00 press 1000 (NN1) 92.30 PART FOR WHOLE : CENT- RAL FACTOR FOR INSTITU- TION / PROFESSION : CENT- RAL INSTRUMENT FOR INSTITUTION 92.30 100.00 PART FOR PART : PRODUCT FOR PRODUCER : AB- STRACT PRODUCT FOR PRODUCER : MEDIUM FOR COMMUNICATOR 6.50 63.11 radio 1000 (NN1) 10.30 PART FOR PART : PRODUCT FOR PRODUCER : ABSTRACT PRODUCT FOR PRODUCER : ABSTRACT PRODUCT FOR INSTITUTION 3.80 36.89 329 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR PART : PRODUC - ER FOR PRODUCT : PRO- DUCER FOR PHYSICAL PRODUCT : PRODUCER FOR CONSUMER PRODUCT 43.74 77.26 WHOLE FOR PART : INSTITU- TION FOR PEOPLE RESPON- SIBLE 12.87 22.74 Renault 567 56.61 PART FOR PART : PRODUCER FOR PRODUCT : PRODUCER FOR PHYSICAL PRODUCT : PRODUCER FOR CONSUMER PRODUCT PART FOR PART : CON- TROLLED FOR CONTROL- LER ‘driver’ 3.70 6.54 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR INSTITU- TION ‘church’ 9.00 42.65 PART FOR WHOLE : PRINCI- PAL PART FOR WHOLE : CAPITAL CITY FOR STATE ‘ancient Rome’ 6.90 32.70 Rome 1000 (NP0) 21.10 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR INSTITU- TION : CAPITAL CITY FOR GOVERNMENT 3.80 18.01 330 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE 0.80 3.79 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR EVENT 0.50 2.37 PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE : LOCATION FOR SPORTS TEAM 0.10 0.47 roof 1000 (NN1) 6.90 PART FOR WHOLE : PRINCI- PAL PART FOR WHOLE ‘house’ ’attic’ 6.90 6.30 0.60 100.00 91.30 8.70 PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE 1.20 40.00 PART FOR WHOLE : LOCA - TION FOR USE OF LOCATION 1.00 33.33 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR INSTITU- TION 0.50 16.67 room 1000 (NN1) 3.00 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR EVENT : LOCATION FOR ACTIVITY OF LOCATION 0.30 10.00 331 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR WHOLE : CENT- RAL FACTOR FOR ACTIVITY : CENTRAL INSTRUMENT FOR ACTIVITY ‘voyage’ 2.49 60.00 PART FOR WHOLE : PRINCI- PAL PART FOR WHOLE ‘sailboat’ 1.39 33.33 sail 361 (NN1) 4.16 PART FOR WHOLE : CEN- TRAL FACTOR FOR INSTITU- TION / PROFESSION : CEN- TRAL INSTRUMENT FOR PROFESSION 0.28 6.67 PART FOR PART : CAUSE FOR EFFECT : INSTRUMENT FOR MUSIC 14.94 86.67 sax 87 17.24 PART FOR PART : CON- TROLLED FOR CONTROL- LER 2.30 13.33 scalpel 92 1.09 PART FOR PART : CON- TROLLED FOR CONTROL- LER 1.09 100.00 WHOLE FOR PART : INSTI- TUTION FOR PEOPLE RESPONSIBLE 25.30 79.31 schools 1000 (NN2) 31.90 WHOLE FOR PART : EVENT FOR LOCATION 6.10 19.12 332 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) WHOLE FOR PART : INSTITU- TION FOR PEOPLE : INSTITU- TION FOR SPORTS TEAM 0.40 1.25 WHOLE FOR PART : INSTITU- TION FOR CENTRAL FACTOR ‘curriculum’ 0.10 0.31 screen 1000 (NN1) 11.50 PART FOR WHOLE : CENT- RAL FACTOR FOR INSTITU- TION / PROFESSION : CENT- RAL INSTRUMENT FOR INSTITUTION 11.50 100.00 PART FOR WHOLE : CENT- RAL FACTOR FOR INSTITU- TION / PROFESSION : CENT- RAL FACTOR FOR INSTITUTION 20.70 92.83 PART FOR WHOLE : LOCA - TION FOR USE OF LOCATION 1.10 4.93 seat 1000 (NN1) 22.30 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR OBJECT : LOCATION FOR BODY PART ‘bottom’ 0.50 2.24 333 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR WHOLE : CENT- RAL FACTOR FOR INSTITU- TION / PROFESSION : CENT- RAL FACTOR FOR INSTITUTION WHOLE FOR PART : INSTITUTION FOR PEOPLE RESPONSIBLE 0.10 0.45 PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE 0.00 0.00 Senate 1000 36.30 WHOLE FOR PART : INSTI- TUTION FOR PEOPLE RESPONSIBLE 36.30 100.00 set of wheels 14 50.00 PART FOR WHOLE : PRINCI- PAL PART FOR WHOLE 50.00 100.00 PART FOR PART : PRODUC - ER FOR PRODUCT : PRO- DUCER FOR ABSTRACT PRODUCT : AUTHOR FOR WORKS 34.90 98.31 PART FOR PART : PERSON FOR THEIR TIME 0.30 0.85 Shakespeare 1000 35.50 PART FOR WHOLE : MEMBER OF CATEGORY FOR CATE- GORY 0.30 0.85 shoes 1000 0.80 WHOLE FOR PART : WHOLE FOR PRINCIPAL PART 0.80 100.00 334 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) silver 1000 (NN1) 20.80 PART FOR WHOLE : ATTRI- BUTE OF ENTITY FOR ENTI - TY : ATTRIBUTE OF OBJECT FOR OBJECT : MATERIAL OF OBJECT FOR OBJECT ‘cutlery/ tableware’ ’coins’ ’medal’ 20.80 13.60 4.20 3.00 100.00 65.38 20.19 14.42 skirt 1000 (NN1) 0.90 PART FOR WHOLE : AT- TRIBUTE OF ENTITY FOR ENTITY : ATTRIBUTE OF PERSON FOR PERSON : COSTUME FOR WEARER 0.90 100.00 PART FOR WHOLE : LOCA- TION FOR PEOPLE THERE ‘management’ ’audience’ 3.24 1.88 1.36 59.62 34.62 25.00 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR INSTITU- TION ’sports club’ 0.84 15.38 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR EVENT 1.26 23.08 stadium 955 5.45 PART FOR PART : CAUSE FOR EFFECT : PLACE FOR SOUND EFFECT 0.10 1.92 335 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE 2.30 95.83 street 1000 (NN1) 2.40 WHOLE FOR PART : WHOLE FOR PRINCIPAL PART ‘apartment’ 0.10 4.17 suits 569 (NN2) 3.34 PART FOR WHOLE : AT- TRIBUTE OF ENTITY FOR ENTITY : ATTRIBUTE OF PERSON FOR PERSON : COSTUME FOR WEARER 3.34 100.00 tails 525 (NN2) 6.10 PART FOR WHOLE : PRINCI- PAL PART FOR WHOLE : GARMENT FOR OUTFIT 6.10 100.00 WHOLE FOR PART : WHOLE FOR PRINCIPAL PART ‘receiver’ ’bell’ ’cable’ 16.40 10.50 5.70 0.20 80.00 51.22 27.80 0.98 PART FOR PART : CON- TROLLED FOR CONTROLLER 2.30 11.22 PART FOR WHOLE : CENT- RAL FACTOR FOR ACTIVITY : CENTRAL INSTRUMENT FOR ACTIVITY 0.90 4.39 telephone 1000 (NN1) 20.50 PART FOR WHOLE : PRINCI- PAL PART FOR WHOLE ‘connection’ 0.90 4.39 336 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) turf 511 (NN1) 9.98 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR EVENT 9.98 100.00 turtleneck 13 46.15 PART FOR WHOLE : AT- TRIBUTE OF ENTITY FOR ENTITY : ATTRIBUTE OF OBJECT FOR OBJECT : ATTRIBUTE OF CLOTHING FOR CLOTHING 46.15 100.00 uniforms 500 6.60 PART FOR WHOLE : AT- TRIBUTE OF ENTITY FOR ENTITY : ATTRIBUTE OF PERSON FOR PERSON : COSTUME FOR WEARER 6.60 100.00 WHOLE FOR PART : INSTI- TUTION FOR PEOPLE RESPONSIBLE 7.90 95.18 WHOLE FOR PART : INSTITU- TION FOR PEOPLE : INSTITU- TION FOR SPORTS TEAM 0.20 2.41 university 1000 (NN1) 8.30 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR EVENT : LOCATION FOR ACTIVITY OF LOCATION WHOLE FOR PART : ACTIVITY FOR DURATION OF ACTIVITY 0.20 2.41 337 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR EVENT 20.20 51.80 PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE : COUNTRY FOR RULERS 18.30 46.92 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR EVENT PART FOR WHOLE : MEMBER OF CATEGORY FOR CATEGORY 0.80 2.05 PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE 0.50 1.28 Vietnam 1000 (NP0) 39.00 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR EVENT WHOLE FOR PART : EVENT FOR DURATION / POINT IN TIME 0.30 0.77 Wall Street 498 96.99 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR INSTITU- TION 96.99 100.00 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR INSTITU- TION : CAPITAL CITY FOR GOVERNMENT 26.80 98.17 Washington 1000 (NP0) 27.30 PART FOR WHOLE : LOCA - TION FOR PEOPLE THERE : LOCATION FOR SPORTS TEAM 0.50 1.83 338 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR EVENT 84.54 100.00 Watergate 97 84.54 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR EVENT PART FOR WHOLE : MEMBER OF CATEGORY FOR CATEGORY 7.22 8.54 White House 749 72.10 PART FOR WHOLE : LOCA - TION FOR ENTITY THERE : LOCATION FOR INSTITU- TION : BUILDING FOR INSTITUTION 72.10 100.00 windmill 123 (NN1) 1.63 WHOLE FOR PART : WHOLE FOR PRINCIPAL PART 1.63 100.00 wine 1000 (NN1) 0.40 PART FOR PART : CONTENT FOR CONTAINER 0.40 100.00 PART FOR WHOLE : PRINCI- PAL PART FOR WHOLE ‘planes’ 0.80 100.00 wings 1000 (NN2) 0.80 PART FOR WHOLE : PRINCI- PAL PART FOR WHOLE ‘planes’ PART FOR WHOLE : CENTRAL FACTOR FOR INSTITUTION / PROFES- SION : CENTRAL INSTRU- MENT FOR INSTITUTION ‘airline’ 0.10 12.50 339 Wordform Number analyzed MYs total (%) Mapping 1 M1 (total; %) M1 (MYs; %) Mapping 2 M2 (total; %) M2 (MYs; %) Mapping 3 M3 (total; %) M3 (MYs; %) wood 1000 (NN1) 21.90 PART FOR WHOLE : ATTRI- BUTE OF ENTITY FOR ENTI - TY : ATTRIBUTE OF OBJECT FOR OBJECT : MATERIAL OF OBJECT FOR OBJECT ‘forest’ ’tree/ branch of tree’ 21.90 20.20 1.70 100.00 92.24 7.76 340 Appendix B: The results of the metaphor study Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) OTHER : POSSESSION : DESIRE IS HUNGER : GETTING S . T . CONCRETE / ABSTRACT IS EATING 26.69 64.33 MENTAL EVENTS : THE MIND IS A BODY : IDEAS ARE OBJECTS : IDEAS ARE FOOD : INTEREST IS APPETITE 10.02 24.16 appetite 858 41.49 EMOTION : LUST : DESIRE IS HUNGER 4.78 11.52 MENTAL EVENTS : DEBATE : ARGUMENT IS WAR 35.30 80.60 EVENT STRUCTURE : ACTION ACTIONS ARE WAR : OBJECTS INVOLVED / OBSTACLES ARE ENEMIES 3.60 8.22 EVENT STRUCTURE : COMPETITION COMPETITION IS WAR : SPORTS / ECONOMICAL COMPETITION IS WAR 2.80 6.40 OTHER : ILLNESSES TREATING AN ILLNESS IS FIGHTING A WAR : INFECTION IS AN ATTACK BY THE DISEASE 0.90 2.05 OTHER : EXTERNAL EVENTS CHEMICAL / BIOLOGICAL REACTIONS ARE WAR : SUBSTANCES / MICROBES ARE ENEMIES 0.40 0.91 EMOTION : OTHER MENTAL THE EMOTIONAL / MENTAL SELF IS A BRITTLE OBJECT : THOUGHTS / EMOTIONS ARE EN- EMIES 0.20 0.46 OTHER : MACHINES ARE PEOPLE A DESTRUCTIVE COMPUTER CODE IS A VIRUS 0.20 0.46 OTHER : PROBLEMS SOLVING A PROBLEM IS WAR : A PROBLEM IS A TARGET / ENEMY 0.10 0.23 EVENT STRUCTURE : LONGTERM ACTION LEADING A LIFE IS WAR : LIFE IS AN ENEMY 0.10 0.23 attacked 1000 (VVD) 43.80 EMOTION : LOVE LOVE IS A PLANT : DESTRUCTIVE ACTIONS ARE ENEMIES 0.10 0.23 342 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) OTHER : ANIMALS ANIMALS ARE ENEMIES 0.10 0.23 EVENT STRUCTURE : COMPETITION COMPETITION IS WAR 67.80 94.56 EVENT STRUCTURE : ACTION : MEANS OF ACHIEVING A PURPOSE IS A PATH : OBSTACLES ON THE PATH ARE ADVERSARIES 1.10 1.53 EVENT STRUCTURE : EXTERNAL EVENTS THE CLIMATE IS AN ADVERSARY 0.90 1.26 OTHER : PROBLEMS SOLVING A PROBLEM IS WAR : A PROBLEM IS A TARGET / ADVERSARY 0.90 1.26 MENTAL EVENTS : THE MIND IS A BODY MENTAL CONTROL IS PHYSICAL CONTROL : EMOTIONS / IDEAS WE CANNOT CONTROL ARE ADVERSARIES 0.60 0.84 EVENT STRUCTURE : TIME TIME IS A PHYSICAL ADVERSARY 0.20 0.28 EVENT STRUCTURE : ACTION OBSTACLES TO MOTION ARE ADVERSARIES 0.10 0.14 beat 1000 (VVD) 71.70 EVENT STRUCTURE : DEBATE VERBAL ABUSE IS WAR 0.10 0.14 MENTAL EVENTS : BELIEFS BELIEFS ARE CONSTRUCTED OBJECTS : A BELIEF NO LONGER HELD IS AN OBJECT THAT HAS BEEN DESTROYED : UNEXPECTED NEWS IS AN OFFENSIVE WEAPON 82.11 82.98 EMOTION : LUST SEXUALITY IS AN OFFENSIVE WEAPON 14.74 14.89 bombshell 95 (NN1; NN1- VVB) 98.95 EVENT STRUCTURE : COMPETITION COMPETITION IS WAR : A POWERFUL STRIKE IS AN OFFENSIVE WEAPON 2.11 2.13 bright 1000 (AJ0) 32.40 OTHER : MORALITY GOODNESS IS LIGHT 17.40 53.71 343 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) MENTAL EVENTS : OTHER MENTAL INTELLIGENCE IS A LIGHT SOUCE 13.90 42.90 MENTAL EVENTS : PERCEPTION CLEAR SOUNDS ARE A LIGHT SOURCE 0.50 1.54 MENTAL EVENTS : THE MIND IS A BODY UNDERSTANDING IS SEEING : AIDS TO SEEING ARE AIDS TO VISION 0.20 0.62 OTHER : PAIN PAIN IS A LIGHT SOURCE 0.20 0.62 EMOTION : LOVE AFFECTION IS WARMTH : DISLIKE IS COLD : COLD IS A LIGHT SOURCE 0.10 0.31 OTHER : LIGHT & DARK DARKNESS IS A COVER : LIGHT IS REVEALING 0.10 0.31 MENTAL EVENTS : BELIEFS BELIEFS ARE COMMODITIES : ACCEPTING A BELIEF IS BUYING IT 0.50 62.50 buy 1000 (VVB; VVB- NN1; VVI) 0.80 EVENT STRUCTURE : CAUSAL COMMERCE CAUSATION IS COMMERCIAL TRANSACTION : CAUSING BENEFICIAL EFFECTS IS GIVING MONEY 0.30 37.50 MENTAL EVENTS : THEORIES THEORIES ARE BUILDINGS 23.30 53.20 MENTAL EVENTS : COMMUNICATION CONDUIT METAPHOR : THE STIMULUS IS A BUILDING 9.00 20.54 OTHER : SOCIETY / GROUPS OF PEOPLE A SOCIETY / GROUP OF PEOPLE IS A BODY : A BODY IS A CONSTRUCTED OBJECT 2.50 5.71 MENTAL EVENTS THE MIND IS A BODY : IDEAS ARE OBJECTS : IDEAS ARE CONSTRUCTED OBJECTS 2.10 4.79 constructed 1000 (VVD; VVD- VVN; VVN; VVN-AJ0; VVN-VVD) 43.80 EVENT STRUCTURE : ACTION A PLANNED ACTIVITY / SITUATION IS A BUILDING 1.70 3.88 344 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) OTHER : PEOPLE BODIES / BODY PARTS ARE CONSTRUCTED OBJECTS 1.50 3.42 EMOTION AN EMOTION IS A BUILDING 0.80 1.83 OTHER : DATA A SAMPLE / DATA SET IS A CONSTRUCTED OBJECT 0.50 1.14 OTHER : RELATIONSHIPS A RELATIONSHIP IS A CONSTRUCTED OBJECT 0.50 1.14 EVENT STRUCTURE : PROPERTIES A PROPERTY IS A BUILDING 0.40 0.91 OTHER : MONEY BUDGETS / PORTFOLIOS ARE CONSTRUCTED OBJECTS 0.40 0.91 MENTAL EVENTS : PERCEPTION PERCEPTIONS ARE BUILDINGS 0.30 0.68 MENTAL EVENTS : DEBATE A CASE IS A BUILDING 0.20 0.46 EVENT STRUCTURE : LONGTERM ACTIVITY EXISTENCE / CAREER IS A BUILDING 0.20 0.46 EVENT STRUCTURE : COMPETITION THE RESULT OF A COMPETITION IS A CONSTRUCTED OBJECT 0.20 0.46 OTHER : FOOD MEALS ARE BUILDINGS 0.10 0.23 OTHER : PROBLEMS A CONFLICT IS A CONSTRUCTED OBJECT 0.10 0.23 EMOTION : ANGER ANGER IS HEAT : ABSENCE OF EMOTION IS COLD 26.20 57.46 cool 1000 (AJ0; AJ0- AV0; AJ0- NN1) 45.60 EMOTION : LOVE AFFECTION IS COLD : GOOD / FASHIONABLE IS COLD 10.00 21.93 345 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) EMOTION : LOVE AFFECTION IS WARMTH : DISLIKE IS COLD 8.50 18.64 OTHER : MONEY LARGE AMOUNT OF MONEY IS COLD 0.80 1.75 EVENT STRUCTURE : ACTION INVOLVEMENT IS COLD 0.10 0.22 EVENT STRUCTURE : PROPERTIES IMPORTANCE IS INTERIORITY 72.50 95.90 core 1000 75.60 OTHER : SELF THE BODY IS A CONTAINER FOR THE SELF : SENSITIVITY IS INTERIORITY 3.10 4.10 OTHER : PROBLEMS A PROBLEM IS A LOCKED CONTAINER FOR ITS SOLUTION : SOLVING A PROBLEM IN- VOLVES OPENING AN OBJECT 31.82 57.57 OTHER : SELF THE SELF IS A BRITTLE OBJECT 9.09 16.45 OTHER : SELF THE BODY IS A CONTAINER FOR THE SELF : A PERSON IS A LOCKED CONTAINER FOR THEIR EMOTIONS / THOUGHTS 6.73 12.17 MENTAL EVENTS : COMMUNICATION CONDUIT METAPHOR : THE CONTENT IS CONTAINED IN THE STIMULUS 3.45 6.15 OTHER : RELATIONSHIP A RELATIONSHIP IS A CONSTRUCTED OBJECT : UNITY IS A BRITTLE OBJECT 1.45 2.63 EVENT STRUCTURE PROPERTIES : A STABLE SYSTEM IS A CONSTRUCTED OBJECT : A STABLE SYSTEM IS A BRITTLE OBJECT 1.09 1.98 crack 550 (VVB; VVB- NN1; VVI) 55.27 MENTAL EVENTS : IDEAS AN IDEA / THEORY IS A CONSTRUCTED OBJECT : AN IDEA / THEORY IS A BRITTLE OBJECT 0.55 0.99 346 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) EMOTIONS A VOICE IS A LOCKED CONTAINER FOR EMOTIONS 0.55 0.99 OTHER : SELF THE BODY IS A CONTAINER FOR THE SELF : THE BODY IS A CONTAINER FOR THE IMMOR- TAL SOUL 0.18 0.33 MENTAL EVENTS : PERCEPTION SILENCE IS A BRITTLE OBJECT 0.18 0.33 EMOTIONS : LUST GAINING PHYSICAL INTIMACY IS WAR 0.18 0.33 OTHER : MORALITY BEING IMMORAL / DISHONEST IS BEING CROOKED 34.72 94.79 EVENT STRUCTURE : LONGTERM ACTION LIFE IS A JOURNEY : LIFE IS A PATH 1.27 3.48 crooked 314 (AJ0; AJ0- VVD; AJ0- VVN) 36.62 MENTAL EVENTS : IDEAS IDEAS ARE LOCATIONS : REASONING IS FOLLOWING A PATH 0.64 1.74 EVENT STRUCTURE : LONGTERM ACTION LIFE IS A JOURNEY : MAJOR CHOICES ARE CROSSROADS 19.29 78.15 OTHER : LOCATIONS CENTRAL PLACES ARE CROSSROADS 5.19 21.02 crossroads 482 24.69 OTHER : PEOPLE PEOPLE ARE CROSSROADS 0.21 0.84 MENTAL EVENTS : DEBATE ARGUMENT IS WAR 23.80 48.47 OTHER : SELF THE BODY IS A CONTAINER FOR THE SELF : GAINING EMOTIONAL / MENTAL INTIMACY ( AGAINST RESISTANCE ) IS WAR 12.10 24.64 defences 1000 (NN2) 49.10 EVENT STRUCTURE : COMPETITION SPORTS / ECONOMIC COMPETITION / PROTECTION OF VALUES / INTERESTS IS WAR 8.10 16.50 347 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) OTHER : ILLNESS TREATING AN ILLNESS / AGE IS FIGHTING A WAR : THE IMMUNE SYSTEM / A PRECAUTION IS A DEFENCE 5.00 10.18 MENTAL EVENTS : BELIEFS DANGEROUS BELIEFS ARE CONTAGIOUS DISEASES : TREATING AN ILLNESS IS FIGHTING A WAR 0.10 0.20 EVENT STRUCTURE : ACTION A TRADITION / TREND / HABIT IS A LIVING BEING W . R . T . LIFE CYCLE 1.00 18.18 EVENT STRUCTURE : ACTION : AN ACTION / AGGRESSIVE ATTACK IS A LIVING BEING W . R . T . LIFE CYCLE 0.90 16.36 MENTAL EVENTS : PERCEPTION A SOUND IS A LIVING BEING W . R . T . LIFE CYCLE 0.80 14.55 EMOTIONS AN EMOTION IS A LIVING BEING W . R . T . LIFE CYCLE 0.80 14.55 OTHER : MACHINES MACHINES ARE PEOPLE : DEGREE OF FUNCTIONING IS HEALTH 0.40 7.27 OTHER : FIRE A FIRE IS A HUMAN BEING W . R . T . LIFE CYCLE 0.40 7.27 MENTAL EVENTS : IDEAS AN IDEA IS A LIVING BEING W . R . T . LIFE CYCLE 0.30 5.45 MENTAL EVENTS : COMMUNICATION CONDUIT METAPHOR : A SIGN ( SYSTEM ) IS A LIVING BEING W . R . T . LIFE CYCLE 0.20 3.64 EVENT STRUCTURE : EXTERNAL EVENTS A COMPANY IS A LIVING BEING W . R . T . LIFE CYCLE 0.20 3.64 EVENT STRUCTURE : EXTERNAL EVENTS THE CLIMATE IS A LIVING BEING W . R . T . LIFE CYCLE 0.20 3.64 died 1000 5.50 OTHER : SELF ASPECTS OF THE SELF ARE DISTINCT INDIVIDUALS WITH A LIFE CYCLE 0.10 1.82 348 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) OTHER : HUMOUR EFFECTS OF HUMOUR ARE INJURIES 0.10 1.82 EVENT STRUCTURE : TIME A DAY IS A LIVING BEING W . R . T . LIFE CYCLE 0.10 1.82 MENTAL EVENTS : THE MIND IS A BODY IDEAS ARE FOOD : UNDERSTANDING IS DIGESTION 32.39 93.02 OTHER : PROBLEMS PROBLEMS ARE FOOD : SOLVING IS DIGESTION 0.81 2.33 OTHER : MACHINES MACHINES ARE PEOPLE 0.81 2.33 EVENT STRUCTURE : ACTION MUSIC IS FOOD : PRODUCING MUSIC IS DIGESTION 0.40 1.16 digest 247 (NN1-VVB; VVB; VVB- NN1) 34.82 EVENT STRUCTURE : ACTION A COMPANY IS FOOD : TAKE - OVER IS DIGESTION 0.40 1.16 EMOTION : LUST DESIRES ARE FORCES BETWEEN DESIRED AND DESIRER : SEXUALITY IS AN ELECTRO- MAGNETIC FORCE BETWEEEN DESIRED AND DESIRER 1.20 80.00 EMOTION PEOPLE ARE BATTERIES : A PERSON ’ S POWER ( PHYSICAL AND PSYCHOLOGICAL ) IS ELEC- TRICITY 0.20 13.33 electricity 1000 1.50 EMOTION : EMOTIONS ARE FORCES EMOTIONS ARE ELECTROMAGNETIC FORCES : EMOTIONAL INVOLVEMENT IS ELECTRICITY 0.10 6.67 MENTAL EVENTS : BELIEFS BELIEFS ARE LOVERS : ACCEPTING A BELIEF IS ENTERING A RELATIONSHIP 42.09 63.13 embraced 537 66.67 OTHER : INCLUSION INCLUSION IS AN EMBRACE 19.93 29.89 349 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) OTHER : LOCATIONS A LOCATION IS A LOVER : BEING AT A LOCATION / LIKING / ACCEPTING IT IS BEING EM- BRACED / EMBRACING IT 1.30 1.96 EMOTION : LOVE A PERSON IS A LOVER : WELCOMING / LIKING A PERSON IS EMBRACING THEM 0.93 1.40 EVENT STRUCTURE : ACTION A ROLE / AGREEMENT / OPPORTUNITY IS A LOVER : ACCEPTING A ROLE / AGREEMENT / OPPORTUNITY IS ENTERING A RELATIONSHIP 0.74 1.12 OTHER : LIGHT LIGHT IS A LOVER : BEING IN THE LIGHT IS BEING EMBRACED 0.37 0.56 OTHER : WATER WATER IS A LOVER : BEING TOUCHED IS BEING EMBRACED 0.37 0.56 EVENT STRUCTURE : ACTION POLITICAL FORCES ARE PHYSICAL FORCES : BEING AFFECTED IS BEING EMBRACED 0.19 0.28 EVENT STRUCTURE : PROPERTIES SILENCE IS A LOVER : FALLING SILENT IS BEING EMBRACED BY IT 0.19 0.28 MENTAL EVENTS : PERCEPTION A SIGHT IS A LOVER : LIKING THE SIGHT IS BEING EMBRACED 0.19 0.28 EVENT STRUCTURE : LONGTERM ACTION LIFE IS A LOVER : SEIZING IT WHOLEHEARTEDLY IS EMBRACING IT 0.19 0.28 OTHER : BODY PARTS TONGUES ARE LOVERS 0.19 0.28 EVENT STRUCTURE : PROPERTIES ADDITIONAL CONCRETE PROPERTIES / THINGS ARE WEALTH 50.42 57.69 EVENT STRUCTURE : PROPERTIES WELL - BEING IS WEALTH 21.43 24.52 enriched 238 (VVD; VVD- AJ0; VVD- VVN; VVN; VVN-AJ0; VVN-VVD) 87.39 EVENT STRUCTURE : PROPERTIES ADDITIONAL ABSTRACT PROPERTIES / THINGS ARE WEALTH 15.55 17.79 350 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) EVENT STRUCTURE : ACTION PHYSICAL AGGRESSION IS A HOT FLUID 28.19 38.93 MENTAL EVENTS : DEBATE DISAGREEMENTS / PROBLEMS ARE HOT FLUIDS 19.69 27.20 EMOTIONS EMOTIONS ARE ENTITIES WITHIN A PERSON : EXTREME EMOTIONS ARE HOT FLUIDS 8.88 12.27 MENTAL EVENTS : PERCEPTION SOUNDS ARE A HOT FLUID / VOLCANO 7.53 10.40 OTHER : LIVING BEINGS LIVING BEINGS ARE HOT FLUIDS / VOLCANOES 3.47 4.80 EVENT STRUCTURE : ACTION SUDDEN ACTIVITY IS A HOT FLUID / VOLCANO 1.54 2.13 OTHER : OBJECTS VEHICLES ARE HOT FLUIDS / VOLCANOES 0.77 1.07 OTHER : ILLNESSES AM ILLNESS IS A HOT FLUID / VOLCANO 0.77 1.07 EVENT STRUCTURE : EXTERNAL EVENTS THE CLIMATE IS A HOT FLUID / VOLCANO 0.39 0.53 MENTAL EVENTS : PSYCHOLOGICAL FORCES PSYCHOLOGICAL FORCES ARE PHYSICAL FORCES : A FORCE IS A HOT FLUID / VOLCANO 0.19 0.27 MENTAL EVENTS : PERCEPTION IMAGES ARE HOT FLUIDS / VOLCANOES 0.19 0.27 MENTAL EVENTS : COMMUNICATION NEWS IS A HOT FLUID / VOLCANO 0.19 0.27 EVENT STRUCTURE : PROPERTIES A STATE OF BEING IS A HOT FLUID / VOLCANO 0.19 0.27 erupted 518 72.39 OTHER : BEAUTY BEAUTY / DECORATION IS A HOT FLUID / VOLCANO 0.19 0.27 351 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) OTHER : LIGHT LIGHT IS A HOT FLUID / VOLCANO 0.19 0.27 OTHER : RESPONSIBILITIES AN OBLIGATION / LAW / CONTROL IS A PRISON 9.40 18.54 EVENT STRUCTURE : LONGTERM ACTION REALITY / LIFE / PROFESSION / FATE IS A PRISON 7.50 14.79 MENTAL EVENTS : IDEAS IDEAS ARE PRISONS 6.50 12.82 OTHER : PEOPLE OTHER PEOPLE ’ S PRESENCE IS A PRISON 6.00 11.83 OTHER : PROBLEMS A PROBLEM IS A PRISON 5.50 10.85 MENTAL EVENTS : PERCEPTION ATTENTION IS A PRISON 3.20 6.31 EMOTION AN EMOTION IS A PRISON 2.70 5.33 EVENT STRUCTURE : CAUSATION CONSEQUENCES ARE PRISONS 2.30 4.54 EVENT STRUCTURE : EXTERNAL EVENTS ECONOMIC DECLINE / POVERTY IS A PRISON 1.70 3.35 OTHER : LOCATION A LOCATION / HOUSE IS A PRISON 1.70 3.35 OTHER : SELF THE BODY / SELF IS A CONTAINER FOR THE SELF : THE BODY / SELF IS A PRISON 0.90 1.78 OTHER : COMPUTER PROGRAMS A COMPUTER PROGRAM IS A LOCATION : A COMPUTER PROGRAM IS A PRISON 0.80 1.58 escape 1000 (VVB; VVB- NN1; VVI) 50.70 EVENT STRUCTURE : ACTION A TRADITION / ROUTINE IS A PRISON 0.40 0.79 352 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) EVENT STRUCTURE : COMPETITION COMPETITION IS A PRISON 0.40 0.79 OTHER : ILLNESS AN ILLNESS IS A PRISON 0.40 0.79 EVENT STRUCTURE : EXTERNAL EVENTS KNOWLEDGE OF PAST EVENTS IS AN EXTERNAL EVENT EXTERING FORCE ON PRESENT EVENTS : THE PAST IS A PRISON 0.30 0.59 MENTAL EVENTS : COMMUNICATION A SIGN SYSTEM IS A PRISON 0.30 0.59 MENTAL EVENTS : COMMUNICATION A BOOK IS A PRISON 0.20 0.39 OTHER : LIGHT LIGHT IS A PRISON 0.20 0.39 OTHER : CLOTHES / CUSHIONS CLOTHES / CUSHIONS ARE A PRISON 0.20 0.39 MENTAL EVENTS : PERCEPTION MUSIC IS A PRISON 0.10 0.20 EVENT STRUCTURE : CAUSATION CAUSAL PRECEDENCE IS TEMPORAL PRECENDENCE 13.50 34.09 OTHER : RESPONSIBILITIES COMPLIANCE IS FOLLOWING 9.90 25.00 EVENT STRUCTURE : ACTION PURPOSEFUL ACTION IS DIRECTED MOTION TO A DESTINATION : GUIDED ACTION IS GUIDED MOTION : INTEREST / IMITATION IS FOLLOWING 8.80 22.23 followed 1000 39.60 OTHER : FORM FORM IS MOTION : PATHS / BODY CONTOURS ARE GUIDES : RUNNING PARALLEL IS FOL- LOWING 3.20 8.09 353 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) EVENT STRUCTURE : COMPARISON IMPORTANCE IS ANTECEDENCE 2.40 6.06 EVENT STRUCTURE : CAUSATION PEOPLE / PEOPLE ’ S ACTIONS ( GAZE ) ARE GUIDES TO ACTIONS ( GAZE ): WATCHING CARE- FULLY IS FOLLOWING 0.90 2.27 EVENT STRUCTURE : LONGTERM ACTION LIFE IS A JOURNEY : CAREER IS A GUIDE 0.20 0.51 MENTAL EVENTS : PERCEPTION PERCEPTIONS ARE ANIMATE BEINGS : HUMANS ’ PATHS ARE GUIDES TO PERCEPTIONS 0.20 0.51 MENTAL EVENTS : THE MIND IS A BODY IDEAS ARE LOCATIONS : REASONING IS FOLLOWING A PATH : FACTS ARE GUIDES OUR MOVEMENTS 0.20 0.51 MENTAL EVENTS : PERCEPTION PERCEPTIONS ARE GUIDES TO MOVEMENTS 0.10 0.25 EVENT STRUCTURE : PROPERTIES NEGATIVE EVENTS ARE ANIMATE BEINGS : BEING AFFECTED IS BEING FOLLOWED 0.10 0.25 EMOTIONS EMOTIONS ARE ANIMATE BEINGS : BEING AFFECTED IS BEING FOLLOWED 0.10 0.25 MENTAL EVENTS : IDEAS IDEAS ARE OBJECTS : UNDERSTANDING / ADOPTING IS GRASPING 42.37 80.56 EVENT STRUCTUER : OPPORTUNITIES OPPORTUNITIES ARE OBJECTS : SEIZING IS GRASPING 4.06 7.72 OTHER : PROBLEMS PROBLEMS ARE OBJECTS : SOLVING IS GRASPING 3.41 6.48 EVENT STRUCTURE : PROPERTIES A PROPERTY IS AN OBJECT : ADOPTING IS GRASPING 0.97 1.85 grasped 616 52.60 EMOTIONS EMOTIONS ARE OBJECTS : BEING AFFECTED IS GRASPING 0.49 0.93 354 355 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) MENTAL EVENTS : COMMUNICATION CONDUIT METAPHOR : WORDS ARE OBJECTS : BELIEVING IS GRASPING 0.32 0.62 EMOTIONS EMOTIONS ARE ANIMATE BEINGS : BEING AFFECTED IS BEING GRASPED 0.32 0.62 MENTAL EVENTS : IDEAS IDEAS ARE ANIMATE BEINGS : UNDERSTANDING IS BEING GRASPED 0.16 0.31 MENTAL EVENTS : PERCEPTION PERCEPTIONS ARE ANIMATE BEINGS : PERCEIVING IS BEING GRASPED 0.16 0.31 OTHER : PROBLEM A PROBLEM IS A WILD ANIMAL : SOLVING IS GRASPING 0.16 0.31 OTHER AN AREA OF ART IS AN OBJECT : EXHIBITING IS GRASPING 0.16 0.31 MENTAL EVENTS : COMMUNICATION CONDUIT METAPHOR : A WORD / SOUND / BOOK / ACTION / RITUAL / VICTORY IS A CON- TAINER : THE CONTENT IS CONTAINED IN THE STIMULUS 31.61 68.84 OTHER : SELF THE BODY IS A CONTAINER FOR THE SELF 9.91 21.58 EMOTION AN EMOTION IS A CONTAINER 1.42 3.08 MENTAL EVENTS : IDEAS AN IDEA IS AN OBJECT : AN IDEA IS A CONTAINER FOR ITS MEANING 0.94 2.05 OTHER : LOCATION A LOCATION IS AN OBJECT : A LOCATION IS A CONTAINER FOR PEOPLE 0.63 1.37 OTHER : PEOPLE A FACE IS A CONTAINER FOR EMOTIONS 0.47 1.03 hollow 636 (AJ0; AJ0- NN1) 45.91 OTHER : PEOPLE A HEAD IS A CONTAINER FOR IDEAS 0.47 1.03 356 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) EVENT STRUCTURE : TIME TIME IS AN OBJECT : A DAY IS A CONTAINER 0.16 0.34 EVENT STRUCTURE : LONGTERM ACTION LIFE IS AN OBJECT : LIFE IS A CONTAINER FOR ITS MEANING 0.16 0.34 EXTERNAL EVENTS : CLIMATE THE WIND IS A CONTAINER 0.16 0.34 MENTAL EVENTS : COMMUNICATION CONDUIT METAPHOR : THE CONTENT IS CONTAINED IN THE STIMULUS : DIFFICULTY IS HARDNESS OF OBJECT 19.79 29.37 MENTAL EVENTS : IDEAS AN IDEA IS AN OBJECT : AN IDEA IS A CONTAINER FOR ITS MEANING : DIFFICULTY IS HARDNESS OF OBJECT 9.63 14.29 OTHER : LIGHT / DARK DARKNESS IS A SOLID 9.09 13.49 OTHER : SELF THE SELF IS A SOLID : PEOPLE WHO ARE HARD TO UNDERSTAND AND REMOTE ARE SOLIDS 8.56 12.70 EMOTION THE BODY IS A CONTAINER FOR EMOTIONS : EMOTIONS ARE ENTITIES WITHIN A PERSON : EYES / FACIAL EXPRESSION IS A WINDOW TO THEIR EMOTIONS : OBSCURITY IS HARDNESS 5.35 7.94 EVENT STRUCTURE : PROPERTIES A PROPERTY IS A PHYSICAL PROPERTY : UNITY IS A SOLID : A GROUP OF PEOPLE THAT IS HARD TO GET INTO IS A SOLID 4.81 7.14 OTHER : PROBLEMS A PROBLEM IS A LOCKED CONTAINER FOR ITS SOLUTION : DIFFICULTY IS HARDNESS OF OBJECT 3.74 5.56 impenetrable 187 68.98 EMOTIONS EMOTIONS ARE OBJECTS : EMOTIONS ARE SOLIDS : EMOTIONS THAT ARE HARD TO UNDER- STAND ARE SOLIDS 2.67 3.97 357 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) EVENT STRUCTURE : ACTION AN ACTION IS AN OBJECT : AN ACTION IS A SOLID : AN ACTIVITY THAT CANNOT BE CHANGED IS A SOLID 2.14 3.17 OTHER : FORCES A FORCE THAT CANNOT BE BROKEN IS A SOLID 0.53 0.79 OTHER : MORALITY BADNESS IS DARKNESS : THE FUTURE IS A SOLID 0.53 0.79 OTHER : SPACE AN OPEN SPACE IS A SOLID 0.53 0.79 MENTAL EVENTS : IDEAS / BELIEFS DANGEROUS IDEAS / BELIEFS ARE CONTAGIOUS DISEASES 4.00 38.10 OTHER : MACHINES MACHINES ARE PEOPLE : A DESTRUCTIVE COMPUTER CODE IS A VIRUS 2.30 21.90 EMOTIONS EMOTIONS / DESIRES ARE CONTAGIOUS DISEASES 1.70 16.19 EVENT STRUCTURE : PROPERTIES DANGEROUS PROPERTIES ARE CONTAGIOUS DISEASES 1.10 10.48 EVENT STRUCTURE : EXTERNAL EVENTS POLLUTION IS A CONTAGIOUS DISEASE 0.40 3.81 OTHER : PEOPLE EVIL PEOPLE ARE CONTAGIOUS DISEASES 0.20 1.90 OTHER : PROBLEMS A PROBLEM IS A CONTAGIOUS DISEASE 0.20 1.90 EVENT STRUCTURE : ACTION AN ACTION IS A CONTAGIOUS DISEASE 0.20 1.90 infected 1000 10.50 OTHER : ANIMATE BEINGS ORGANISMS ARE CONTAGIOUS DISEASES 0.10 0.95 358 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) EVENT STRUCTURE : EXTERNAL EVENTS A SEASON IS A DISEASE : AUTUMN IS A CONTAGIOUS DISEASE 0.10 0.95 MENTAL EVENTS : COMMUNICATION A RUMOUR IS A CONTAGIOUS DISEASE 0.10 0.95 OTHER : MEDIA A MEDIUM IS A CONTAGIOUS DISEASE 0.10 0.95 EVENT STRUCTURE : CAUSATION CAUSATION IS CONTROL OVER AN OBJECT : CAUSING MENTAL / VERBAL REACTIONS IS ALLOWING MOTION OF POSSESSIONS 36.76 66.49 EVENT STRUCTURE : CAUSATION CAUSATION IS CONTROL OVER AN OBJECT : CAUSING ACTIVITIES IS ALLOWING MOTION OF POSSESSIONS 12.94 23.40 EVENT STRUCTURE : CAUSATION CAUSATION IS CONTROL OVER AN OBJECT : ALLOWING CHANGE OF PROPERTIES IS ALLOWING MOTION OF POSSESSIONS 4.41 7.98 invites 340 (NN2-VVZ; VVZ; VVZ- NN2) 55.29 EVENT STRUCTURE : CAUSATION CAUSATION IS CONTROL OVER AN OBJECT : CAUSING EMOTIONAL REACTIONS IS ALLOW- ING MOTION OF POSSESSIONS 1.18 2.13 OTHER : PROBLEM A PROBLEM IS A LOCKED CONTAINER FOR ITS SOLUTION : MEANS TO SOLVING IS MEANS TO OPENING 31.80 77.56 OTHER : ACCESS ( COMPUTERIZED ) DATA ARE A LOCKED CONTAINER : MEANS TO ACCESSING DATA IS MEANS TO OPENING 4.90 11.95 key 1000 (NN1; NN1- AJ0; NN1- VVB) 41.00 MENTAL EVENTS : COMMUNICATION CONDUIT METAPHOR : THE CONTENT IS CONTAINED IN THE STIMULUS : MEANS TO HELP UNDERSTANDING IS MEANS TO OPENING 2.60 6.34 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) EVENT STRUCTURE : PROPERTIES PROPERTIES ARE PHYSICAL PROPERTIES : IMPORTANCE IS INTERIORITY : MEANS TO ACCESS MOST IMPORTANT ASPECT IS MEANS TO ACCESS INTERIOR 1.70 4.15 EVENT STRUCTURE : COMPETITION FAILURE IS DEATH 1.10 28.21 MENTAL EVENTS : IDEAS IDEAS ARE ANIMATE BEINGS W . R . T . LIFE CYCLE : FALLING INTO DISUSE IS THEIR DEATH 0.90 23.07 EVENT STRUCTURE : OPPORTUNITIES OPPORTUNITIES ARE ANIMATE BEINGS W . R . T . LIFE CYCLE : NOT SEIZING THEM IS THEIR DEATH 0.60 15.38 EVENT STRUCTURE : HARM PSYCHOLOGICAL HARM IS PHYSICAL INJURY 0.30 7.69 EVENT STRUCTURE : HARM PHYSICAL HARM / DISCOMFORT IS DEATH 0.20 5.13 EVENT STRUCTURE : ACTION A BEHAVIOUR IS AN ANIMATE BEING W . R . T . LIFE CYCLE 0.20 5.13 EMOTIONS EMOTIONS ARE ANIMATE BEINGS W . R . T LIFE CYCLE : NO LONGER HAVING THEM IS DEATH 0.20 5.13 MENTAL EVENTS : DEBATE ARGUMENT IS WAR : CRITICISM IS DEATH 0.20 5.13 EVENT STRUCTURE : HARM PHYSICAL DISCOMFORT IS AN ANIMATE BEING W . R . T . LIFE CYCLE : FIGHTING IT IS ITS DEATH 0.10 2.56 killed 1000 (VVD; VVD- AJ0; VVD- VVN; VVN; VVN-AJ0; VVN-VVD) 3.90 OTHER : HUMOUR EFFECTS OF HUMOUR ARE INJURIES : SEVERE EFFECTS ARE DEATH 0.10 2.56 ladder 1000 (NN1; NN1- 39.10 EVENT STRUCTURE : LONGTERM ACTION CAREER PROGRESS IS VERTICAL MOVEMENT 16.80 42.97 359 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) EVENT STRUCTURE : COMPARISON STATUS IS POSITION 9.20 23.53 OTHER : OBJECTS ANYTHING THAT RESEMBLES A LADDER IS A LADDER 6.40 16.37 EVENT STRUCTURE : COMPARISON IMPORTANCE / INTENSITY IS POSITION ON VERTICAL SCALE 3.80 9.72 EVENT STRUCTURE : CHANGE PROGRESS IS VERTICAL MOVEMENT 2.20 5.63 EVENT STRUCTURE : LONGTERM ACTION LIFE / AGE IS VERTICAL MOVEMENT 0.50 1.28 EVENT STRUCTURE : PROPERTIES PROPERTIES ARE POSITIONS ON A VERTICAL SCALE 0.10 0.26 VVB; VVB- NN1) OTHER : DISTANCE DISTANCE IS A VERTICAL MOVEMENT 0.10 0.26 MENTAL EVENTS : COMMUNICATION CONDUIT METAPHOR : THE CONTENT IS CONTAINED IN THE STIMULUS : THE CONTENT IS A HEAVY LOAD 11.48 55.36 EVENT STRUCTURE : PROPERTIES WEALTH IS A HEAVY LOAD 3.70 17.86 EVENT STRUCTURE : PROPERTIES BIAS IS A HEAVY LOAD 1.85 8.93 OTHER : INTOXICATION INTOXICATION IS GETTING A BURDEN 1.11 5.36 EVENT STRUCTURE : PROPERTIES COSTS ARE A HEAVY LOAD 1.11 5.36 loaded 270 (AJ0; AJ0- VVD; AJ0- VVN) 20.74 EMOTION EMOTIONS ARE A HEAVY LOAD 0.74 3.57 360 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) OTHER : RESPONSIBILITIES OBLIGATIONS ARE POSSESSIONS : OBLIGATIONS ARE BURDENS 0.74 3.57 EVENT STRUCTURE : PROPERTIES PROPERTIES ARE OBJECTS : PROPERTIES ARE POSSESSIONS 16.40 32.16 EVENT STRUCTURE : COMPARISON STATUS IS AN OBJECT : STATUS IS A VALUABLE POSSESSION 10.10 19.80 OTHER : PEOPLE A PERSON IS AN OBJECT : A PERSON IS A VALUABLE POSSESSION 8.60 16.86 EVENT STRUCTURE : PROPERTIES WEIGHT IS AN OBJECT : WEIGHT IS A POSSESSION 2.50 4.90 EVENT STRUCTURE : ACTION DIRECTION / WAY IS A POSSESSION 1.90 3.73 EVENT STRUCTURE : CHANGE LIFE IS AN OBJECT : LIFE IS A POSSESSION : DEATH IT IS LOSING IT 1.90 3.73 EVENT STRUCTURE : OPPORTUNITIES OPPORTUNITIES ARE OBJECTS : OPPORTUNITIES ARE POSSESSIONS 1.80 3.53 EVENT STRUCTURE : TIME TIME IS AN OBJECT : TIME IS A VALUABLE POSSESSION 1.80 3.53 MENTAL EVENTS : IDEAS IDEAS ARE OBJECTS : IDEAS ARE VALUABLE POSSESSIONS 1.40 2.75 EVENT STRUCTURE : ACTION CONTACT IS AN OBJECT : CONTACT IS A VALUABLE POSSESSION 0.80 1.57 EVENT STRUCTURE : ACTION AN ACTIVITY IS AN OBJECT : AN ACTIVITY IS A POSSESSION 0.70 1.37 MENTAL EVENTS : HOPE / BELIEF A HOPE / BELIEF IS AN OBJECT : A HOPE / BELIEF IS A VALUABLE POSSESSION 0.70 1.37 lost 1000 (VVD) 51.00 MENTAL EVENTS : PERCEPTION SEEING IS TOUCHING / HAVING : EYES ARE LIMBS 0.60 1.18 361 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) EMOTION EMOTIONS ARE OBJECTS : EMOTIONS ARE VALUABLE POSSESSIONS 0.60 1.18 OTHER : SELF ASPECTS OF THE SELF ARE DISTINCT INDIVIDUALS THAT CAN BE LOST 0.50 0.98 EVENT STRUCTURE : LOCATION / SPEED A LOCATION / SPEED IS AN OBJECT : A LOCATION / SPEED IS A POSSESSION 0.40 0.78 MENTAL EVENTS : DEBATE ARGUMENT IS WAR 0.20 0.39 OTHER : ILLNESS TREATING AN ILLNESS IS FIGHTING A WAR : DEATH IS LOSING IT 0.10 0.20 MENTAL EVENTS : PSYCHOLOGICAL FORCES PSYCHOLOGICAL FORCES ARE PHYSICAL FORCES : PSYCHOLOGICAL BENEFIT IS PHYSICAL BENEFIT : THE MIND / SELF / SOUL IS A CHILD 20.00 37.93 EMOTION AN EMOTION IS AN ANIMATE BEING : AN EMOTION IS A CHILD 10.91 20.69 MENTAL EVENTS : IDEAS IDEAS ARE ANIMATE BEINGS : AN IDEA IS A CHILD 10.00 18.97 EVENT STRUCTURE : ACTION AN ACTION ( COOPERATION / RITUAL ) IS A CHILD 4.55 8.62 EVENT STRUCTURE : PROPERTIES A PROPERTY IS AN ANIMATE BEING : A PROPERTY IS A CHILD 2.73 5.17 EVENT STRUCTURE : LONGTERM ACTION A CAREER IS AN ANIMATE BEING : A CAREER IS A CHILD 1.82 3.45 OTHER : HIGHER BEINGS A HIGHER BEING IS A CHILD 0.91 1.72 nourished 110 (VVD; VVD- VVN; VVN; VVN-AJ0; VVN-VVD) 52.73 OTHER : GENRE A GENRE IS A CHILD 0.91 1.72 362 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) OTHER : OBJECTS OBJECTS NEEDED FOR A PROFESSION ARE FOOD FOR THE PEOPLE EXERCISING IT 0.91 1.72 EVENT STRUCTURE : ACTION RESTRICTIONS TO ACTION ARE OBJECTS : ATTACHING THEM TO S . B . RESULTS IS PERSON ’ S INABILITY TO ACT 8.55 32.32 EMOTIONS : HOPE A HOPE IS AN OBJECT : REASONS FOR HOPE ARE ATTACHED TO PERSONS / THINGS 6.29 23.78 MENTAL EVENTS : IDEAS AN IDEA IS AN OBJECT : GETTING AN IDEA IS PINNING IT DOWN 4.52 17.07 MENTAL EVENTS : PERCEPTION SEEING IS TOUCHING : EYES ARE LIMBS : LOOKING AT A PERSON IS ATTACHING A GLANCE 2.42 9.15 OTHER : RESPONSIBILITIES RESPONSIBILITIES ARE POSSESSIONS : RESPONSIBILITIES ARE ATTACHED TO PERSON HELD RESPONSIBLE 1.77 6.71 OTHER : FORM EXTERNAL APPEARANCE IS A COVER : ATTACHING S . T . TO THE COVER IS CHANGING IT 0.97 3.66 MENTAL EVENTS : IDEAS IDEAS ARE OBJECTS : IDEAS ARE ATTACHED TO THEIR REASONS 0.81 3.05 EVENT STRUCTURE : PROPERTIES PROPERTIES ARE OBJECTS : PROPERTIES ARE ATTACHED TO PERSON / THING DISPLAYING PROPERTY 0.65 2.44 OTHER : CLOTHES PUTTING ON CLOTHES IS BEING PINNED INTO THEM 0.16 0.61 EVENT STRUCTURE : OPPORTUNITIES SEIZING AN OPPORTUNITY IS PINNING IT DOWN 0.16 0.61 pinned 620 (VVD; VVD- VVN; VVN; VVN-AJ0; VVN-VVD) 26.45 EMOTIONS EMOTIONS ARE PHYSICAL FORCES : BEING AFFECTED BY THEM IS BEING UNABLE TO ACT 0.16 0.61 363 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) MENTAL EVENTS : THE MIND IS A BODY : REASONING IS FOLLOWING A PATH : IDEAS ARE GUIDES TO OUR REASONING 67.00 95.04 pointed 1000 (VVD; VVD- AJ0; VVD- VVN; VVN; VVN-AJ0; VVN-VVD) 70.50 MENTAL EVENTS : THE MIND IS A BODY : IDEAS ARE LOCATIONS : REASONING IS FOL- LOWING A PATH : FACTS ARE GUIDES TO OUR MOVEMENTS 3.50 4.96 OTHER : PEOPLE / COMPANIES AN UNDETERMINED LOT OF PEOPLE / COMPANIES ARE A FLUID / ONE PERSON IS A FLUID 6.60 22.37 MENTAL EVENTS : COMMUNICATION CONDUIT METAPHOR : THE STIMULUS IS A FLUID 5.10 17.29 EMOTIONS EMOTIONS ARE FLUIDS 4.40 14.92 OTHER : MONEY MONEY IS A FLUID 3.20 10.85 OTHER : LIGHT / DARK LIGHT IS A FLUID 2.10 7.12 OTHER : OBJECTS AN UNDETERMINED LOT OF OBJECTS ( GOODS / MUNITION / MATTER ETC .) IS A FLUID 1.60 5.42 EVENT STRUCTURE : ACTION EFFORT IS A FLUID 1.30 4.41 OTHER : OBJECTS GRANULAR MATERIALS ARE FLUIDS 1.10 3.73 MENTAL EVENTS : PERCEPTION MUSIC / SOUND IS A FLUID 1.00 3.39 MENTAL EVENTS : IDEAS IDEAS ARE FLUIDS 0.90 3.05 poured 1000 29.50 OTHER : FIRE / SMOKE FIRE / SMOKE IS A FLUID 0.50 1.69 364 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) EVENT STRUCTURE : ACTION ACTIONS ARE FLUIDS 0.40 1.36 EVENT STRUCTURE : TIME TIME IS A FLUID 0.40 1.36 EMOTIONS EMOTIONS ARE ENTITIES WITHIN A PERSON : THE HEART / SOUL IS A CONTAINER FOR EMOTIONS : EMOTIONS ARE FLUIDS 0.40 1.36 OTHER : HIGHER BEINGS THE H OLY S PIRIT IS A FLUID 0.20 0.68 OTHER : LIGHT / DARK SHADOW IS A FLUID 0.10 0.34 OTHER : RELATIONSHIP KINSHIP IS A FLUID 0.10 0.34 OTHER : ILLNESSES AN ILLNESS IS A FLUID 0.10 0.34 profited 99 39.39 EVENT STRUCTURE : CAUSAL COMMERCE : HAVING BENEFICIAL THINGS HAPPEN TO ONE IS GETTING MONEY 39.39 100.00 MENTAL EVENTS : IDEAS : IDEAS ARE VALUABLE COMMODITIES : ARGUING FOR AN IDEA IS TRYING TO SELL IT 3.00 65.22 sell 1000 (VVB; VVB- NN1; VVI) 4.60 OTHER : PEOPLE PEOPLE ( AND THEIR ABILITIES / CONCERNS ) ARE VALUABLE COMMODITIES 1.60 34.78 sharks 232 21.55 EVENT STRUCTURE : HARM HARM IS PHYSICAL INJURY DUE TO PREDATION BY VICIOUS ANIMALS : COMPETITORS OR ANTAGONISTS ARE PREDATORS 21.55 100.00 EVENT STRUCTURE : PERCEPTION AN OFFENSIVE SOUND / TASTE / SMELL IS A CUTTING INSTRUMENT 9.70 13.23 sharp 1000 (AJ0; AJ0- AV0; AJ0- NN1) 73.30 EVENT STRUCTURE : PROPERTIES EXACTNESS / RICHNESS OF DETAIL IS A CUTTING INSTRUMENT 9.60 13.10 365 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) MENTAL EVENTS : OTHER MENTAL INTELLIGENCE IS A CUTTING INSTRUMENT 7.60 10.37 MENTAL EVENTS : COMMUNICATION WORDS ARE WEAPONS 4.80 6.55 OTHER : FORM NON - ROUNDED / NON - CURVED SHAPE IS A CUTTING INSTRUMENT 4.50 6.14 EVENT STRUCTURE : PROPERTIES : THE HUMAN SENSES ARE SHARP INSTRUMENTS 3.90 5.32 EVENT STRUCTURE : ACTION HOSTILE BEHAVIOUR IS A CUTTING INSTRUMENT 2.40 3.27 EVENT STRUCTURE : CHANGE MARKED CHANGES ARE CUTTING INSTRUMENTS 14.30 19.51 EVENT STRUCTURE : PROPERTIES INTENSITY IS A CUTTING INSTRUMENT 11.50 15.69 EVENT STRUCTURE : CAUSATION A NEGATIVE EFFECT IS A CUTTING INSTRUMENT 1.80 2.46 EVENT STRUCTURE : PROPERTIES ATTRACTIVENESS IS A CUTTING INSTRUMENT 1.40 1.91 OTHER : PAIN PAIN IS A CUTTING INSTRUMENT 1.20 1.64 EVENT STRUCTURE : EXTERNAL EVENTS CLIMATE : COLD IS A CUTTING INSTRUMENT 0.60 0.82 EMOTION EMOTIONAL SELF IS A BRITTLE OBJECT 16.10 27.29 EMOTION : HOPE A HOPE NO LONGER HELD IS AN OBJECT THAT HAS BEEN DESTROYED 11.60 19.66 shattered 1000 59.00 EVENT STRUCTURE : PROPERTIES PHYSICAL WELL - BEING IS A BRITTLE OBJECT 6.50 11.02 366 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) EVENT STRUCTURE : PROPERTIES PEACE / QUIET / SLEEP IS A BRITTLE OBJECT 5.80 9.83 MENTAL EVENTS : IDEAS / THEORIES IDEAS / THEORIES ARE CONSTRUCTED OBJECTS : AN IDEA / THEORY NO LONGER HELD IS AN OBJECT THAT HAS BEEN DESTROYED 4.50 7.63 EVENT STRUCTURE : LONGTERM ACTION LIFE / CAREER IS A BRITTLE OBJECT 3.30 5.59 OTHER : RELATIONSHIP A ( LOVE ) RELATIONSHIP IS A CONSTRUCTED OBJECT : UNITY IS A BRITTLE OBJECT 2.60 4.41 EVENT STRUCTURE : PROPERTIES A SYSTEM / POSITION HELD IS A BRITTLE OBJECT 1.80 3.05 EMOTION : OTHER EMOTIONS EMOTIONS ARE BRITTLE OBJECTS 1.40 2.37 EVENT STRUCTURE : COMPETITION PEOPLE ARE OBJECTS : LOSING A WAR IS BEING SHATTERED 1.30 2.20 EVENT STRUCTURE : PROPERTIES ECONOMIC SUCCESS IS A BRITTLE OBJECT 1.00 1.69 EVENT STRUCTURE : COMPETITION COMPETITION IS WAR : LOSING A SPORTS COMPETITION IS BEING SHATTERED / WINNING IT IS SHATTERING THE OPPONENT 0.90 1.53 EVENT STRUCTURE : ACTION BEHAVIOUR IS A BRITTLE OBJECT 0.80 1.36 OTHER : FORM EXTERNAL APPEARANCE IS A COVER : EXTERNAL APPEARANCE IS A BRITTLE OBJECT 0.80 1.36 EVENT STRUCTURE : PERCEPTION A SOUND IS A BRITTLE OBJECT 0.40 0.68 EVENT STRUCTURE : TIME TIME IS A BRITTLE OBJECT 0.20 0.34 367 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) EVENT STRUCTURE : COMPARISON COMPARING AND SEEKING CONCRETE THINGS IS SHOPPING 3.15 54.55 EVENT STRUCTURE : COMPARISON COMPARING AND SEEKING ABSTRACT THINGS IS SHOPPING 1.99 34.55 shopping 953 (VVG; VVG- NN1) 5.77 EVENT STRUCTURE : COMPARISON COMPARING AND SEEKING CUSTOMERS IS SHOPPING 0.63 10.91 EVENT STRUCTURE : PROPERTIES PROPERTIES ARE PHYSICAL PROPERTIES : IMPORTANCE IS A PHYSICAL PROPERTY : IM- PORTANCE IS SIZE 3.60 48.00 EVENT STRUCTURE : PROPERTIES PROPERTIES ARE PHYSICAL PROPERTIES : VALUE IS SIZE 3.20 42.67 smaller 1000 7.50 EVENT STRUCTURE : PROPERTIES AGE IS SIZE 0.70 9.33 OTHER : MORALITY MORALITY IS CLEANLINESS 11.30 86.67 spotless 115 13.04 OTHER : BEAUTY BEAUTY IS CLEANLINESS 1.74 13.33 stoned 58 (AJ0; AJ0- VVD; AJ0- VVN) 58.62 OTHER : INTOXICATION INTOXICATION IS GETTING DESTROYED 58.62 100.00 EMOTION : DESIRE IS HUNGER (includes special case: IDEAS ARE OBJECTS : LEARNING IS ABSORBING ) 27.41 91.26 OTHER : MACHINES MACHINES ARE HUMANS : FUEL CONSUMPTION IS THIRST 2.33 7.77 thirst 343 30.03 OTHER : PLANTS PLANTS ARE HUMANS 0.29 0.97 368 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) MENTAL EVENTS : THE MIND IS A BODY IDEAS ARE ( HIDDEN ) OBJECTS : CREATING THEM IS MAKING THEM VISIBLE 15.85 30.59 EMOTION : MISCELLANEOUS DISGUST IS NAUSEA 12.80 24.71 OTHER : PROBLEMS A PROBLEM IS A ( HIDDEN ) OBJECT : CREATING IT IS MAKING IT VISIBLE 9.76 18.82 OTHER : ANIMATE BEINGS MAKING ANIMATE BEINGS AVAILABLE IS MAKING THEM VISIBLE 6.71 12.94 EVENT STRUCTURE : OPPORTUNITIES OPPORTUNITIES ARE OBJECTS : WASTING THEM IS THROWING THEM UP 3.05 5.88 OTHER : VALUABLE CONCRETE THINGS MAKING VALUABLE CONCRETE THINGS AVAILABLE IS MAKING THEM VISIBLE 3.05 5.88 throw up 164 51.83 EVENT STRUCTURE : OPPORTUNITIES OPPORTUNITIES ARE ( HIDDEN ) OBJECTS : CREATING THEM IS MAKING THEM VISIBLE 0.61 1.18 EMOTION INTENSE EMOTIONS ( AFFECTION / ANGER / COMFORT ) ARE WARMTH 24.90 97.26 EVENT STRUCTURE : PROPERTIES FUTURE SUCCESS IS WARMTH 0.50 1.95 warm 1000 (AJ0; AJ0- NN1) 25.60 EVENT STRUCTURE : PROPERTIES GOOD QUALITY IS WARMTH 0.20 0.78 EVENT STRUCTURE : TIME TIME IS A VALUABLE COMMODITY 61.10 79.58 EVENT STRUCTURE : PROPERTIES TALENT / ENERGY / EFFORT IS A VALUABLE COMMODITY 7.54 9.83 OTHER : PEOPLE / BODY PARTS PEOPLE / BODY PARTS ARE VALUABLE COMMODITIES 4.07 5.29 wasting 689 (VVG) 76.78 EVENT STRUCTURE : OPPORTUNITIES OPPORTUNITIES ARE VALUABLE COMMODITIES 1.89 2.46 369 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) EVENT STRUCTURE : LONGTERM ACTION LIFE IS A VALUABLE COMMODITY 1.31 1.70 EMOTION EMOTIONS ARE VALUABLE COMMODITIES 0.58 0.76 OTHER : SPACE SPACE IS A VALUABLE COMMODITY 0.15 0.19 OTHER : SPEED SPEED IS A VALUABLE COMMODITY 0.15 0.19 EVENT STRUCTURE : COMPARISON COMPARISON OF PROPERTIES IS COMPARISON OF PHYSICAL PROPERTIES : COMPARI- SON OF IMPORTANCE IS WEIGHING 24.54 55.21 EVENT STRUCTURE : PROPERTIES PROPERTIES ARE PHYSICAL PROPERTIES : IMPORTANCE IS WEIGHT 7.41 16.67 EVENT STRUCTURE : PROPERTIES PROPERTIES ARE PHYSICAL PROPERTIES : CONSIDERING A PROPERTY IS WEIGHING 5.25 11.81 EVENT STRUCTURE : COMPARISON COMPARISON OF OTHER PROPERTIES IS WEIGHING 4.63 10.42 weigh 648 (VVB; VVB- NN1; VVI) 44.44 OTHER : RESPONSIBILITIES / PROBLEMS RESPONSIBILITIES / PROBLEMS ARE BURDENS 2.62 5.90 EVENT STRUCTURE : HARM PSYCHOLOGICAL HARM IS PHYSICAL INJURY 7.20 88.00 EVENT STRUCTURE : COMPETITION COMPETITION IS WAR : LOSING IS BEING WOUNDED 0.44 5.33 MENTAL EVENTS : DEBATE THEORIES ARE LIVING BEINGS : ARGUMENT IS WAR : ATTACKING A THEORETICAL POSI- TION IS WOUNDING IT 0.11 1.33 wounded 917 (VVD; VVD- AJ0; VVD- VVN; VVN; VVN-AJ0; VVN-VVD) 8.18 MENTAL EVENTS : PERCEPTION A PIECE OF MUSIC IS A LIVING BEING : PROBLEMATIC PARTS ARE INJURIES 0.11 1.33 370 Word-form Number analyzed MRs total (%) Mapping M - total (%) M - MRs (%) OTHER : SPORT A SPORT / GAME IS A LIVING BEING : NOT FOLLOWING THE RULES IS WOUNDING IT 0.11 1.31 OTHER : MACHINES MACHINES ARE PEOPLE : BEING DAMAGED IS BEING WOUNDED 0.11 1.31 EMOTION THE BODY IS A CONTAINER FOR EMOTIONS : PENETRATION IS A WOUND 0.11 1.31 371 Narr Francke Attempto Verlag GmbH+Co. KG • Dischingerweg 5 • D-72070 Tübingen Tel. +49 (07071) 9797-0 • Fax +49 (07071) 97 97-11 • info@narr.de • www.narr.de NEUERSCHEINUNG OKTOBER 2010 JETZT BESTELLEN! Saskia Kersten The Mental Lexicon and Vocabulary Learning Implications for the foreign language classroom Language in Performance, Band 43 2010, XVI, 197 Seiten, 25 Abb., 18 Tab., €[D] 54,00/ SFr 76,90 ISBN 978-3-8233-6586-0 Lexis was, for a long time, paid scant attention to in foreign language teaching. Over the last 20 years, however, vocabulary acquisition has become a focus of academic research. In particular, the Cognitive Linguistic perspective on foreign language learning offers a rich theoretical framework for research in this area, since it encapsulates both ease of learning and a more profound knowledge of the target language. Learning vocabulary in school contexts, however, is still strongly associated with rote learning in many parts of the world, that is, the repetition of items, usually using lists with little or no contextual information. The implications of Cognitive Linguistics form the basis of an intervention study carried out in German primary schools. This study investigates whether lessons enabling learners to elaborate on words and thereby process the vocabulary more deeply lead to better long-term retention of these items. The results of this empirical study are used to evaluate the relevance and benefits of the theoretical implications of vocabulary research for primary school learners of English. 085410 Auslieferung Oktober 2010.indd 12 08.10.10 14: 44