Full download ebook at ebookname.

com

Contexts of Metaphor Current Research in the

Semantics Pragmatics Interface 1st Edition M.
Leezenberg

https://ebookname.com/product/contexts-of-metaphor-current-
research-in-the-semantics-pragmatics-interface-1st-edition-
m-leezenberg/

OR CLICK BUTTON

DOWLOAD NOW

Download more ebook from https://ebookname.com

More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Formal Pragmatics Semantics Pragmatics Presupposition

and Focus 1st Edition Nirit Kadmon

https://ebookname.com/product/formal-pragmatics-semantics-
pragmatics-presupposition-and-focus-1st-edition-nirit-kadmon/

The Interfaces of Chinese Syntax with Semantics and

Pragmatics 1st Edition Yicheng Wu

https://ebookname.com/product/the-interfaces-of-chinese-syntax-
with-semantics-and-pragmatics-1st-edition-yicheng-wu/

Meaning in Language An Introduction to Semantics and

Pragmatics 3rd Edition Cruse

https://ebookname.com/product/meaning-in-language-an-
introduction-to-semantics-and-pragmatics-3rd-edition-cruse/

The Situated Organization Case Studies in the

Pragmatics of Communication Research 1st Edition James
R. Taylor

https://ebookname.com/product/the-situated-organization-case-
studies-in-the-pragmatics-of-communication-research-1st-edition-
james-r-taylor/
Worker Participation Current Research and Future Trends
Volume 16 Research in the Sociology of Work 1st Edition
Vicki Smith

https://ebookname.com/product/worker-participation-current-
research-and-future-trends-volume-16-research-in-the-sociology-
of-work-1st-edition-vicki-smith/

Ecological Interface Design 1st Edition Catherine M.

Burns (Author)

https://ebookname.com/product/ecological-interface-design-1st-
edition-catherine-m-burns-author/

Current Topics in Elastomers Research 1st Edition Anil

K. Bhowmick

https://ebookname.com/product/current-topics-in-elastomers-
research-1st-edition-anil-k-bhowmick/

Influenza Current Research 1st Edition Qinghua Wang

https://ebookname.com/product/influenza-current-research-1st-
edition-qinghua-wang/

Current Essentials of Surgery 1st Edition Gerard M.

Doherty

https://ebookname.com/product/current-essentials-of-surgery-1st-
edition-gerard-m-doherty/
Contexts of Metaphor
Current Research in the Semantics/Pragmatics Interface
Series Editors:
K.M. Jaszczolt, University of Cambridge, UK
K. Turner, University of Brighton, UK

Related Elsevier books

In this Series:
TURNER (ed.) The Semantics/Pragmatics Interface
from Different Points of View

JASZCZOLT Discourse, Beliefs and Intentions

GEURTS Presuppositions and Pronouns

JASZCZOLT (ed.) The Pragmatics of Propositional

Attitude Reports

PEETERS (ed.) The Lexicon-Encyclopedia Interface

PAPAFRAGOU Modality: Issues in the Semantics/Pragmatics Interface

Related Elsevier journals

Journal of Pragmatics
Editor: Jacob Mey

Language and Communication

Editors: Roy Harris and Talbot J. Taylor

Language Sciences
Editor: Nigel Love

Lingua
Editors: Johan Rooryck, Neil Smith and Diane Blakemore

Free specimen copies available on request

For further information on the CRiSPI series and for

details of how to submit a proposal go to:
www. elsevier.nl/locate/series/crispi
 CONTEXTS OF
METAPHOR

Michiel Leezenberg
University of Amsterdam, The Netherlands

2001
ELSEVIER
Amsterdam - London - New York - Oxford - Paris - Shannon - Tokyo
ELSEVIER SCIENCE Ltd
The Boulevard, Langford Lane
Kidlington, Oxford OX5 1GB, UK

© 2001 Elsevier Science Ltd. All rights reserved.

This work is protected under copyright by Elsevier Science, and the following terms and conditions apply to its use:

Photocopying
Single photocopies of single chapters may be made for personal use as allowed by national copyright laws. Permission of the Publisher and payment of a fee is
required for all other photocopying, including multiple or systematic copying, copying for advertising or promotional purposes, resale, and all forms of
document delivery. Special rates are available for educational institutions that wish to make photocopies for non-profit educational classroom use.

Permissions may be sought directly from Elsevier Science Global Rights Department, PO Box 800, Oxford OX5 1DX. UK: phone: (+44) 1865 843830, fax:
(+44) 1865 853333. e-mail: [email protected]. You may also contact Global Rights directly through Elsevier's home page (http://www.elsevier.nl),
by selecting 'Obtaining Permissions'.

In the USA, users may clear permissions and make payments through the Copyright Clearance Center, Inc.. 222 Rosewood Drive, Danvers, MA 01923, USA;
phone: (+1)(978) 7508400, fax: (+1)(978) 7504744, and in the UK through the Copyright Licensing Agency Rapid Clearance Service (CLARCS), 90
Tottenham Court Road, London W1POLP, UK; phone: (+44) 171 631 5555: fax: (+44) 171 631 5500. Other countries may have a local reprographic rights
agency for payments.

Derivative Works
Tables of contents may be reproduced for internal circulation, but permission of Elsevier Science is required for external resale or distribution of such material.
Permission of the Publisher is required for all other derivative works, including compilations and translations.

Electronic Storage or Usage

Permission of the Publisher is required to store or use electronically any material contained in this work, including any chapter or part of a chapter.

Except as outlined above, no part of this work may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic,
mechanical, photocopying, recording or otherwise, without prior written permission of the Publisher.
Address permissions requests to: Elsevier Global Rights Department, at the mail, fax and e-mail addresses noted above.

Notice
No responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or
from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences,
in particular, independent verification of diagnoses and drug dosages should be made.

First edition 2001

Library of Congress Cataloging in Publication Data

A catalog record from the Library of Congress has been applied for.

British Library Cataloguing in Publication Data

A catalogue record from the British Library has been applied for.

ISBN: 008 043881 4

©The paper used in this publication meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper).
Printed in The Netherlands.
Current Research in the Semantics/Pragmatics Interface (CRiSPI)

Series Editors: K.M. Jaszczolt, University of Cambridge, UK, and

K. Turner, University of Brighton, UK

Editorial Advisory Board:

N. Asher, UK
C. Casadio, Italy
M. Dascal, Israel
B. Fraser, USA
T. Fretheim, Norway
B. Gillon, Canada
P. Gochet, Belgium
J. Groenendijk, The Netherlands
Yueguo Gu, PRC
A. Kasher, Israel
M. Krifka, Germany
Susumu Kubo, Japan
Chungmin Lee, Korea
S. Levinson, The Netherlands
T. McEnery, UK
F. Nemo, France
P. Pelyvas, Hungary
J. Peregrin, Czech Republic
A. Ramsay, UK
R. Stalnaker, USA
M. Stokhof, The Netherlands
J. van der Auwera, USA
R. van der Sandt, The Netherlands
K. von Heusinger, Germany
G. Ward, USA

The aim of this series is to focus upon the relationship between semantic and pragmatic
theories for a variety of natural language constructions. The boundary between semantics
and pragmatics can be drawn in many various ways, the relative benefits of each gave rise
to a vivid theoretical dispute in the literature in the last two decades. As a side-effect, this
variety has given rise to a certain amount of confusion and lack of purpose in the extant
publications on the topic.
This series provides a forum where the confusion within existing literature can be removed
and the issues raised by different positions can be discussed with a renewed sense of
purpose. The editors intend the contributions to this series to take further strides towards
clarity and cautious consensus.
The Way has never had borders, saying has never had norms.
It is by a 'That's it' which deems that a boundary is marked.
Chuang-tzu
(transl. A.C. Graham)
CONTENTS

Preface ix

Introduction 1

1. Chapters From the History of Metaphor 15

1.1. The Pre- and Protohistory of Metaphor 15
1.1.1. Prehistorical and Preliterate Societies 16
1.1.2. Metaphor in Mesopotamia: Sumer and Akkad 27
1.2. Aristotle on Metaphor, Comparison, and Similarity 31
1.3. 'Abd al-Qahir al-Jurjani's Mysteries of Eloquence 43
1.4. Giambattista Vico: Metaphor and the Origin of Language 56

2. Twentieth-Century Views of Metaphor 69

2.1. Semantic Approaches 69
2.1.1. Referentialist Views 69
2.1.2. Descriptivist Views 78
2.1.3. Intermezzo: Metaphor in Generative Grammar 93
2.2. Pragmatic Approaches 97
2.2.1. The Gricean Program 99
2.2.2. Searle 118
2.3. The Davidsonian Program: Metaphor Without Meaning 124
2.4. Conceptualist Approaches: Cognitive Semantics 135

3. Metaphor and Context 149

3.1. Metaphor and Context-Dependence 149
3.1.1. Kaplan's Logic of Demonstratives 151
3.1.2. Context-Dependence of Property Expressions 161
3.1.3. Metaphor and the Logic of Demonstratives 171
3.1.4. A Comparison: Stern's 'Metaphor as Demonstrative' 186
3.2. Metaphorical Assertion 195
3.2.1. Stalnakerian Views of Assertion 197
3.2.2.Pragmatic Intrusion: The Neo-Gricean Program 210
viii Contexts of Metaphor

3.2.3. Thematic Dimensions, Presupposition, and Assertion 217

3.2.4. Extensions: Scope and Shift of Dimensions 230
3.2.5. Novel Metaphor 239

4. Metaphor, Concept, and Society 251

4.1. Conceptualist Views of Metaphor: A Radical Critique 251
4.1.1. Extending Cognitive Semantics: Gibbs & Indurkhya 252
4.1.2. A Wittgensteinian Critique of Concepts 263
4.2. Metaphor and Concept Formation: A Vygotskyan Approach 270
4.2.1. Vygotsky: Concepts, Language, and Context 272
4.2.2. Similarity and the Role of Theories 280
4.2.3. Concepts and Metaphor 285
4.3. Beyond Literal Meaning 295

Bibliography 305

Index 317
PREFACE

If there is one topic that falls in the disputed border territory between semantics and
pragmatics, it is metaphor. Over the years, competing theories have been developed that take
the principles of metaphorical interpretation to be essentially semantic, essentially pragmatic,
or even as wholly distinct in nature; at present, there is nothing remotely like a consensus on
how to approach the problem. This study attempts to clear the terrain, by comparing the
different approaches in a comprehensive theoretical framework, and by focusing on the more
principled arguments that are, or can be, given for treating metaphor in either semantic or
pragmatic terms. It proceeds from the assumption that meaning is not a unitary phenomenon;
such notions as semantic entailment, presupposition, and implicature display a clearly different
behavior. One of its aims is to clarify how such distinct aspects of meaning are articulated in
metaphorical interpretation.
Although it seems difficult to give knock-down arguments for treating metaphor along either
semantic or pragmatic lines, I believe that there are rather persuasive reasons for analyzing
metaphor in terms of semantic and broadly conventional and language-specific rules rather than
intention-driven and universal pragmatic principles. Chapter 3 attempts to give such an
analysis; it emphasizes the essential role of context-dependence in metaphircal interpretation,
and argues that this context-dependence can be captured in terms of a compositional semantic
theory. Neither a semantic nor a pragmatic approach can give us a fully explanatory account of
metaphor, however; in the final chapter, I therefore present a supplementary treatment at the
conceptual level. For various reasons, the currently fashionable views of concepts or categories
are unsatisfactory; I present an alternative approach, inspired by the work of the Russian
psychologist Lev Vygotsky. Special attention will be given to the role of so-called 'ad hoc
concepts' and of social practices in concept formation, and in particular in so far as they can
help us account for the understanding of metaphor. In other words, although the main thrust of
this book is linguistic, it also incorporates ideas and insights from the social sciences.
I have made an effort to keep this book accessible to non-specialist readers, or to readers
specialized in only one of the fields discussed. I presuppose some familiarity with theoretical
terminology, but as this study covers a rather broad territory, I explain the central notions in
some detail as I go along.

In the course of writing and rewriting this book, I have profited form the comments and
criticism of various people. I owe a special debt of gratitude to Earl MacCormac, who not only
x Contexts of Metaphor

was my first mentor in philosophy, but also guided the start of my thinking about metaphor.
Although my views at present differ widely from his, I am indebted to him for rolling the ball
in the beginning. Renate Bartsch and Martin Stokhof guided the thesis upon which this study is
based. Discussions and correspondence with Jay David Atlas, Veit Bader, Bipin Indurkhya,
Rob van der Sandt and Josef Stern did much to shape and clarify the ideas developed here.
Babette Greiner, Donald Verene and Cornells Verhoeven commented on different parts of the
manuscript, and made valuable suggestions for improvements. Pieter Muysken and Gerard de
Vries gave important support at a later stage. Karel van der Leeuw helped with the Chinese
characters, and Ben Schomakers provided me with some last-minute Aristotelian scholarship. I
am also indebted to the series editors Kasia Jaszczolt and Ken Turner, and Chris Pringle and
Leighton Chipperfield of Elsevier for their enthusiasm about this project.
Sultan kept me company for much of the way, with both forbearance and the necessary
impatience; now, Shilan is here to show us wholly new ways. My mother was always there; it
was in her house that I did an important part of the thinking and writing of this study. Now
she is no longer with us. I miss her. It is to her memory that this book is dedicated.
INTRODUCTION

Before the twentieth century, metaphor has rarely commanded the attention of philosophers.
For the most part, this neglect has resulted less from any conscious choice than from an (often
unarticulated) folk theory, or 'language ideology', that each word has, or should have, a well-
determined 'proper' or 'literal' meaning of its own. An assumption like this is already present
in Plato's Cratylus, but it also appears in Hobbes (1991 [1651]: 24-31) and Locke (1975
[1689]: 508), both of whom reject metaphor as unfit for rational argument. This is not to say,
however, that metaphor has been uniformly ignored, let alone deliberately repressed, in 'the
Western tradition': authors like Aristotle and Vico devoted a fair amount of attention to it,
while for some Islamic philosophers, many of whom were strongly influenced by Greek
philosophizing, metaphor was a topic of central philosophical and theological importance with
respect to the interpretation of the Qur'an. For the most part, however, metaphor was
delegated to rhetoric and literary theory, while literal language was seen as a norm or standard
for 'serious' language usage, and as an ideal for rational argument. Ideals should not be
mistaken for realities, however: many words have nowhere near as clearly delimited a meaning
as we would like to believe, and word meanings often change over time. This insight was
anticipated by the romantic claim that language is metaphorical in origin (Rousseau (1990
[1781]); cf. Muller 1861), and gained in importance with the development of historical and
comparative linguistics from the early nineteenth century onwards. At the same time,
however, romantic literary critics tended to portray poetry, and metaphor along with it, as
essentially different from everyday language:both poetry and metaphor came to be seen as
expressing an 'emotive', rather than a 'cognitive', meaning.
These two beliefs, the folk theory that each word has its 'proper' meaning, and the
romantic idea that metaphor is qualitatively different from literal language, have survived as
guiding - if largely implicit - theoretical assumptions in much of twentieth century philosophy
of language and linguistics. Modern theories of meaning have also been greatly influenced by
the development of formal logic, where expressions are simply postulated to have a single,
well-defined meaning. Aside from the influence of logical semantics, there were two other
factors that contributed to this development: first, the shift in attention from diachronic to
synchronic linguistics, and secondly, the shift in theoretical status of language from a primarily
social and communicative to a primarily individual and cognitive phenomenon. The former
factor tended to downplay the obvious fact that linguistic meanings easily can and do change
in the course of time: with the shift of attention to synchronic meaning, such meaning changes
2 Contexts of Metaphor

became something of secondary importance, not to be accounted for in a theory of langue

(Saussure 1916) or competence (Chomsky 1965). The latter factor tended to downplay the
equally obvious fact that, even at the synchronic level, languages display great variation in
style, register, dialect and sociolect. Such variations were equally taken to fall outside the
domain of'core linguistics'.
Given these pre-theoretical assumptions and theoretical choices, metaphor has remained a
problem for semantic theory; consequently, it has often been delegated to neighbouring
disciplines like pragmatics, literary theory, or cognitive science, and assumed to have no place
in semantics proper. Partly in consequence of all this, the status of metaphor within linguistic
theory is highly controversial at present. There is a large number of rival frameworks that
fundamentally disagree not only on how to incorporate metaphor in a general theory of
language, but also on what such a theory should look like to begin with. Here, I start from the
intuition, with which many theorists would agree in one way or another, that metaphor has
something to do with change of meaning. I will defend a more specific version of this thesis,
however, namely the claim that metaphorical interpretation actually involves a change of
meaning, and this is something far fewer present-day theorists would agree with. Furthermore,
I wil try to integrate this approach to metaphor within a more general theory of language;that
is, I will try to restrict the number of notions specific to metaphor for as far as possible. This
reflects another idea I want to defend: literal and figurative language do not involve any
qualitatively different processes of interpretation.
But even the belief that metaphor can or should be treated within a linguistic theory has not
gone unchallenged. Recently, it has been argued that metaphor is a cognitive rather than a
linguistic phenomenon, and that linguistic metaphors are merely reflections of more general
cognitive processes. While such views contain more than a grain of truth, I think it is still
useful to look at metaphor within one relatively well-defined and well-studied domain of
experience, viz., language,and see what its characteristics are in that domain; afterwards, we
can try to integrate this view within a general cognitive framework. On the other hand, there
are good reasons for doubting the feasibility of a complete reduction of linguistic metaphor to
purely cognitive processes, the most important one being the fact that languages constitute
elaborate systems of conventions and intersubjective knowledge that cannot easily (or perhaps
not at all) be reduced to individual cognition. We will discuss this in more detail below.
My focus on linguistic metaphor reflects a belief that there are significant differences
between linguistic and other metaphors: for example, in the pictorial and cinematographic
metaphors discussed by Indurkhya (1992: 21-6), the tension or anomaly at the 'literal level'
that many authors see as a defining characteristic of linguistic metaphor is usually absent.
Thus, it is very well possible to appreciate Jan van Eyck's famous portrait of Giovanni
 Introduction 3

Arnolfini and Giovanna Cenami (plate 2 in Indurkhya 1992) as just representing a husband
and wife in a room, without attaching any symbolic meaning to specific items of the painting,
such as the one burning candle mounted on a seven-candle candelabrum which symbolizes
pride, one of the seven cardinal sins. Likewise, one can see Piero della Francesca's View of
Jerusalem just as part of a historic painting depicting the discovery of the True Cross without
being aware that at the same time the city of Arezzo is being portrayed. In other words, much
of what Indurkhya calls 'visual metaphor' might probably better be called 'visual symbolism'.
This does not exclude the possibility of similar processes of interpretation occurring in both,
but I think the latter term is a much broader one. As will appear from the discussion below, I
do not think that literal falsity or anomaly can be considered a defining criterion for metaphor,
but this does not preclude that many metaphors do manifest some kind of tension. Moreover,
linguistic metaphors typically involve a much more elaborate 'code' than nonlinguistic ones.
The variation and flexibility in the rules of a languageon a synchronic level is the basis for
diachronic change: if the rules of a language are not rigidly defined to begin with, changes in the
rules need no longer be seen as secondary or anomalous. The rules of a language display a
degree of looseness at any one moment; this holds for phonological and syntactic rules as well
as for semantic ones. A general line of argument in the following will be that all language is
vague and context-dependent to some extent, and that metaphorical language essentially
exploits these general traits. Seen in this light, it may perhaps become less obvious to treat
metaphor as exclusively a matter of language use rather than linguistic meaning (or, put in
Saussurean terms, as a matter of parole rather than of langue). In consequence, the strict and
apparently tidy distinction between literal and figurative language becomes blurred.
This study is, among others, an attempt to get clear about what kind of rules or principles
are involved in the metaphorical interpretation of language. A focus of attention will be the
relative weight of semantic or 'conventional' rules, as opposed to language-independent
pragmatic principles, such as Grice's maxims of conversation. Much of the discussion in
chapters 2 and 3 will therefore be concerned with what has been called the 'semantics-prag-
matics interface', that is, the disputed border territory between, and competing claims to
logical priority of, semantics and pragmatics. One of the methodological problems at stake
here is how we can distinguish semantic phenomena like entailment from pragmatic ones like
presupposition and conversational implicature. This problem is the more complex as there is a
looser connection between theory and evidence in pragmatics than in syntax or semantics:
pragmatic rules can typically be flouted or exploited, so apparent counterexamples to
pragmatic principles can often be explained away as really conforming to them (Thomason
1990: 333). Thus, it may be difficult to actually prove or disprove the claim that metaphor
involves a specific kind of pragmatic principle, and in general it is not always clear what
4 Contexts of Metaphor

constitutes evidence for or against a specific view; the most one can hope to do is to present a
persuasive picture that accounts for a large number of phenomena in a relatively straight-
forward manner. My own approach falls largely within the more familiar semantic views. The
reasons usually adduced for giving a pragmatic account, I will argue below, are at best incon-
clusive; and the move to an analysis in pragmatic terms turns out to create at least as many
problems as it is supposed to solve.
As regards the semantic theory employed, I will mostly assume an intensional, model-
theoretic semantics of the kind first formulated by Richard Montague, to be enriched with
some of the insights concerning context-dependence due to David Kaplan's work on demon-
stratives. By pragmatics, I understand theories of intention-governed principles of communi-
cation, specifically, Grice's theory of implicature and Searle's theory of speaker's meaning. Of
course, this is but one among many competing ways of characterizing the distinction between
semantics and pragmatics (cf. Turner 1999); but I think we are justified in doing so for the
moment. The central point here is that pragmatics, in the more restricted sense, essentially
involves speaker's intentions, and consequently, a crucial question to be answered below is in
how far intentions play an irreducible role in, and specific to, metaphorical interpretation. But
even within this restricted domain, semantic and pragmatic phenomena are not always
separated as neatly as one would want. The context-dependence of various linguistic express-
ions is already a bone of contention, and recent work in dynamic semantics, where the meaning
of a sentence is defined not in terms of its truth conditions but as the way in which it changes
the information of the hearer, has indicated that various phenomena that have long been
considered purely pragmatic may allow for a treatment in semantic terms after all. For this
reason, particular attention will be paid to the question of whether a principled, theoretically
based choice to treat metaphor in either semantic or in pragmatic terms can be made.

Problems of Metaphor

First, of course, we should get some idea of the phenomena to be studied and the problems
surrounding them. The existing definitions of metaphor are too theory-dependent to be dis-
cussed at this stage. For example, calling a metaphor 'an utterance in which one or more
expressions temporarily undergoes a change in meaning' presupposes a semantic treatment of
metaphor, while the claim that metaphor is a matter of 'saying one thing and meaning another'
is based on the assumption that metaphor follows pragmatic principles of interpretation that
are essentially different from semantic rules. Likewise, there is hardly anything like a well-
circumscribed set of problems related to the phenomenon of metaphor like there is for topics
 Introduction 5

such as presupposition; of the problems noted, not all are considered equally relevant by each
author; others would seem relevant, but are rarely noted, let alone discussed. Some of these
challenges, however, which any adequate theory of metaphor should come to terms with, are:

-giving a coherent account of the various constructions that can be interpreted metaphorically
(simple predications, negated and quantified sentences, antecedents of conditionals, etc.);
-the problem of recognizing metaphors: if metaphor is held to involve distinct principles of
interpretation, it should be distinguished from literal language on the one hand, and from other
figures of speech on the other;
-the question of what kind of rules or principles are involved in metaphorical interpretation;
-the question of what information, or perhaps what kind of information, is necessary for the
correct interpretation and perhaps recognition of metaphor;
-the question of whether a specifically 'metaphorical meaning' exists, and if so, where it
should be located within a linguistic theory;
-the question of what 'effect' or assertive power metaphors have, and more specifically of
whether they can be true;
-the fact that the specifically metaphorical effect of a sentence is to some extent indeterminate,
and often cannot be fully paraphrased in literal terms.

Consequently, I will not try to define metaphor even preliminarily, but rather list a number of
relatively uncontroversial examples.1 These should also help to give the reader some idea of
the wide variety of syntactic constructions that can be interpreted metaphorically. Many
discussions, in fact, limit themselves to examples of the A is B kind that are literally false, but
this is only one, relatively easily recognizable, construction among many.
Which linguistic items can be used metaphorically? It appears that in general, all express-
ions that have some descriptive content or information2 associated with them, may receive a
metaphorical interpretation under appropriate circumstances. An obvious class of expressions
in this respect is that of simple or complex common noun phrases used as predicates:

(1) Paul is a pig.

(2) Juliet is the sun.

'Actually, some of these examples may look rather worn off, but the point to be made here is merely to show the
different syntactic constructions that may be interpreted metaphorically. For a much more detailed description, see
Brooke-Rose (1958). White (1996) likewise discusses the variety of grammatical constructions involved in
metaphor, and argues that an adequate analysis should take this variety into account.
2
'Descriptive content' is the more common expression, but in order to avoid confusion with the technical notion
of content to be discussed in 3.1. below, I will speak of 'descriptive information' in the following.
6 Contexts of Metaphor

(3) John Searle is the ayatollah of speech act theory.

One might take (2) to express an identity statement rather than a predication if one takes the
sun as a proper name; for reasons of uniformity, however, I will take it to express a definite
description; thus, like a pig and the ayatollah of pragmatics, it functions as a property
expression. Note, however, that figuratively interpreted noun phrases need not be the
sentence's predicate, but may also occur in subject or object position:

(4) That donkey won't listen to me.

(5) I have reached the top of the greasy pole.

Likewise, verbal and adjectival phrases (which like common noun phrases can be seen
semantically as property expressions) are prime candidates for metaphorical interpretation:

(6) The chairman ploughed through the discussion.

(7) This is a hot moment for studying metaphor.

But also expressions that do not at first sight seem to have descriptive information, such as
proper names, can be used metaphorically. Interestingly, however, the metaphorical inter-
pretation of a name typically requires an article:

(8) John thinks he's Napoleon.

(9) John thinks he's a Napoleon.

If we hear (8), we are likely to think that John belongs in an asylum, whereas with (9), we will
at most dislike him for his tyrannical manners. Apparently, the metaphorical application of a
name does not usually lead to an identity statement, but rather to a predication. This suggests
that the addition of an article makes a name function as a property expression with descriptive
information, rather than as a purely referring term. But, parallel to (4)-(5) above, names may
also occur in subject or object position, and in that case, they do not require an article:

(10) I saw count Dracula again today.

(11) Einstein has failed his mathematics exam.

Still, both classes of examples suggest that the metaphorical interpretation of referential ex-
pressions like names has something to do with the descriptive information that is associated
 Introduction 1

with them,- even if that information need not, strictly speaking, be part of the meaning of
such expressions (cf. ch. 2 and 3 below).
Other referential terms, such as personal pronouns, likewise appear to allow for meta-
phorical interpretation primarily in so far as they carry descriptive information, e.g., indicating
gender or physical distance:

(12) There she comes, (said of an effeminate male)

(13)
'But do not say "this [woman here]"; for she is no longer.'
(Sophocles, Antigone, vs. 567)
(14) That picture is you. (cf. Cooper 1986: 57)

The Greek demonstrative ήδε, 'this', in (13) expresses physical proximity, whereas Creon,
who utters this sentence, means that Antigone has been banished to the realm of the dead by
being buried alive, and is thus figuratively 'remote'. Example (14) is rather problematic: the
personal pronoun you has little descriptive information, so it may be that in order to determine
the metaphorical interpretation of this sentence (and indeed any interpretation at all - see
chapter 3.1), we will first have to determine the contextually given referents of you and of that
picture. But I am not sure whether we should call such cases as (12) - (14), which are rare and
not very rich anyway, metaphors in the strict sense: they are largely conventionalized, and
hardly display the openness and indeterminacy characteristic of more familiar metaphors.
Prepositions in English, and corresponding elements in other languages, like postpositions
and case endings, also seem to allow for metaphorical interpretation to some extent:

(15) In 1984 he lived in poverty in a toolshed.

Intuitively, the different senses of in seem to be related; but one may object that such uses are
fully conventionalized, and hence are not properly speaking metaphors, but merely cases of
polysemy (cf. Cooper 1986: 9). The question of whether these cases count as genuine meta-
phors hinges not only on their degree of conventionalization, but also on a more precise
formulation of the relation between metaphor and polysemy, and on the question of whether a
preposition like in has any clearly delineated 'core meaning' or 'literal meaning' to begin with.
Important counterexamples to theories that define metaphor in terms of grammatical
deviance or semantic anomaly are sentences that may be literally correct, but allow for a
metaphorical interpretation in context:
8 Contexts of Metaphor

(16) Life is not a bed of roses.

(17) This is a pigsty, [said of a messy room]
(18) Anchorage is a cold city.

On a literal interpretation, (16) is trivially true, while (17) and (18) are contingently true or
false rather than structurally incorrect; but an adequate theory of metaphor will have to
account for such cases. Other sentences receive a metaphorical interpretation only because the
literal one is explicitly ruled out:

(19) My hands are tied, figuratively speaking of course.

(20) Nel mezzo del cammin di nostra vita mi ritrovai per una selva oscura.
('Halfway down the path of life, I found myself in a dark forest') (Dante,
Divina commedia)

Without the possessive phrase di nostra vita, (20) could just as well be taken as a literal
description of somebody getting lost during a walk.
It is also possible to quantify over metaphorically interpreted expressions, or to refer back
to them anaphorically:

(21) Qui n'a pas son minotaure? (Marguerite Yourcenar)

(22) [I dedicate this book] to all my Juliets. (J. Stern)
(23) So God has arrived. I met him on the 5.15 train. (Keynes on Wittgenstein)

Such examples of metaphors involving quantification are almost totally ignored in the litera-
ture, and their theoretical significance is therefore not immediately clear. One would like an
adequate theory to give at least some indication, however, of how to account for them. A
particularly difficult case in this respect is:

(24) Brulé de plus de feux que je n'en allumai. (J. Racine, Andromaque, vs. 320)
'Burned by more fires than I lighted...'

The linguistic context of (24) makes clear that the implicit second occurrence of feux is to be
taken in the literal sense, but the first in a figurative sense of being in love. I will return to the
above examples in the course of chapter 3.
Metaphors in indirect discourse contexts raise even more complex problems than those in
quantified or anaphoric ones:
 Introduction 9

(25) Peter said that John is a wolf.

Intuitively, this sentence does not seem quite as 'metaphorical' as the bare assertion that John
is a wolf: it may be seen as ambiguous between a literal and a metaphorical reading in a way
that the simple predication does not seem to be. Is, then, the speaker who utters (25) not
committed to a metaphorical interpretation of John is a wolf as Peter, who is claimed to have
uttered just that last sentence? Such cases involve the much broader problems of indirect
quotation and opaque contexts, and I will not discuss them in detail here.1
I hope the above list, although far from complete, at least gives an idea of the expressions
and constructions that may be interpreted metaphorically. Like most authors, I will limit my
discussion to metaphors in declarative sentences, while acknowledging that questions and
commands, such as (26) and (27), may equally well allow for a metaphorical interpretation:

(26) Is Sally a block of ice?

(27) Freeze!

As the semantics of questions and commands presents a whole field of questions and
problems that are not specific to metaphor, however, I will not discuss such cases here either.

Classifying Theories of Metaphor

Now how do we fit examples like those listed above into a general theory of language?Surely,
it seems a desirable goal to do so, unless one prefers to treat metaphor as a totally outlandish
phenomenon that defies all attempts at analysis, or believes that metaphor 'proves' that such
general theories as usually conceived have to be replaced by something altogether different. At
first sight, various options are open, and especially in the second half of the twentieth
century, a bewildering variety of theories and approaches has blossomed. One would like to be
able to compare the relative merits of these, and in order to do this systematically, it would be
useful to have some kind of classification of theories.
Of such classifications, there is no shortage either. The most famous distinction is perhaps
Max Black's division (1962) into 'substitution views', according to which a metaphorical
expression replaces a proper or literal term, by which it can always be paraphrased; 'compar-

1
Cf. Stern (1979: 413-39) and Lamarque (1982) for some preliminary discussion.
10 Contexts of Metaphor

ison views', which treat metaphors as abbreviated similes, and their meanings as those of the
corresponding comparisons; and 'interaction views', according to which at least one term of a
metaphor changes its meaning, often leading to a result that cannot be properly paraphrased
(cf. Beardsley 1962). Other ways of classifying theories have been proposed by, among
others, Scheffler (1979), Mooij (1976), and Ankersmit & Mooij (1993). Considerations of
space preclude an extensive discussion of these; but while each of these classifications has its
merits, none of them seems sufficiently based upon general theoretical considerations.
Mooij (1976: ch. 3) makes an important attempt at such a theory-based classification by
distinguishing between 'monistic' and 'dualistic' theories: on dualistic theories, metaphorically
applied words retains their literal reference, but 'may carry a second reference, because of their
specific (metaphorical) function', whereas on monistic theories they lose their normal refer-
ential capacity, and 'allow for at most a singular (abnormal and nonliteral) reference in the
metaphorical word' (1976: 31). But also theories that Mooij takes to be monistic will appeal
to some aspect of literal meaning; this aspect just need not lie at the level of reference. It might
be more practical, then, to focus on these different levels or aspects of meaning (such as refer-
ence, sense, and concept) than to restrict one's attention to reference, even if one construes
this notion rather broadly, as Mooij does. Moreover, his scheme cannot adequately deal with
pragmatic approaches that locate metaphorical interpretation entirely in language use rather
than in linguistic meaning; on such views, the words in a metaphor only have literal meanings.
At the risk of seeming pedantic, I would therefore like to propose yet another classification
scheme, guided by two central questions: first, at what level of a linguistic theory is meta-
phorical interpretation accounted for? Second, in virtue of what does the hearer determine a
metaphorical interpretation? Possible answers to the first question are: in syntax; in semantics;
in pragmatics; or outside linguistic theory altogether.1 Possible answers to the second question
are: in virtue of the properties that the referents have; in virtue of the descriptive information
associated with the expressions used; or in virtue of the concepts or mental representations
that are expressed by the words. Thus, one may hold that the hearer understands a metaphor
in virtue of some similarity between the objects referred to, that is, in virtue of some property
that the referents of the metaphor have in common; Black calls this a 'comparison view', and
Beardsley (1962) speaks of an 'object comparison view'. More generally, one can speak of
such views as referentialist, because they crucially involve the referents of the expressions
used. Another answer is that it is the meaning of linguistic expressions, or the descriptive
information associated with them (which may include commonplaces or stereotypes), that
determines metaphorical interpretation. This comes close to Black's 'interaction view', or to

1
In fact, the first position has only been adopted in the 1960s, when transformationalists, e.g., Chomsky (1965:
149) suggested treating metaphor as a violation of selection restrictions.
 Introduction 11

what Beardsley calls 'verbal opposition theories'. Here, I will speak of descriptivist views, as
these approaches take metaphorical interpretation to be guided by the descriptive information
associated with an expression. Note, however, that a descriptivist view does not commit one
to the position that such content is actually part of the meaning of the expression. Finally,
one might hold that metaphorical interpretation arises not from resemblance between objects
or from the descriptive information of expressions, but rather from general cognitive
mechanisms such as reasoning in analogies, the ability to see one thing as another, etc. Such
approaches I will refer to as conceptualist views: they assign a crucial role to the interpreter's
mental or conceptual capacities.
The different approaches also differ as to what kind of knowledge is involved in the
interpretation of metaphor. Very roughly, semantic approaches focus on conventionalized
knowledge of some sort (not necessarily aspects of meaning in ,the more restricted sense, but
also 'stereotypes', 'connotations' and the like); pragmatic approaches emphasize the role of
nonlinguistic or world knowledge; while conceptualistic approaches regard mental represent-
ations and the ability to think in analogies as the key to our metaphorical competence. This
brings us to the following classification scheme:
12 Contexts of Metaphor

Obviously, this scheme is no more than a rough guideline. We will discuss the different
possible positions and their relative merits in chapter 2 below; for now, it may suffice to
signal some doubts about ascribing particular positions to particular authors. For example,
should we classify Davidson's views under pragmatic theories or outside linguistic theories?
Davidson explicitly states that metaphor is a matter of how we use words rather than of what
those words mean (1979), but on the other hand, he denies metaphor anything like a cognitive
content or meaning even at such a presumably pragmatic level. Likewise, Lakoff & Johnson
may be seen as holding a semantic conceptualist view, or alternatively as maintaining that
there is no place at all for metaphor in semantic theory as usually conceived. There is ample
room for discussion here, but I hope the scheme may at least provide some clarification of the
issues at stake, and serve as a guideline through the chapters to come.
The reader will note the absence of what Black calls a substitution view in the above
scheme. There are several reasons for this: to begin with, if framed in Black's terms, a
substitution view is clearly incoherent: a metaphorical expression Mis used as a substitute for
some literal expression L, as in Richard is a lion used to communicate Richard is brave (1962:
31); but one of the reasons why people use metaphors at all is, on many such accounts,
precisely the absence of a literal equivalent (ibid., p. 33). While Black himself hesitates to call
such cases of catachresis, or the plugging of lexical gaps, as genuine cases of metaphor, there
are at present very few authors who would wish to maintain that there is a literal equivalent to
be substituted for every metaphor; but in that case, it simply makes little sense to speak of
substitution.1 Moreover, a substitution view as such does not say how the hearer arrives at the
intended meaning; to account for this, most authors claim that the substitution takes place in
virtue of some resemblance or analogy between the referents. But in that case, of course, a
more sophisticated statement of a substitution view will almost inevitably turn out to be
merely a variant of a comparison or referentialist view. In fact, the two authors Black refers to
as defending a substitution view (Richard Whately and the Oxford Dictionary) both explicitly
state that the analogy or resemblance between the literal referent and the object metaphorically
referred to serves as the basis of the transfer. What is specific to substitution views, then,
boils down to the claim that someone uttering a metaphor ' communicate[s] a meaning that
might have been expressed literally' (1962: 32, emphasis added), and this is precisely the
central trait of pragmatic views: the sentence meaning is what it is, but the speaker uses a
metaphorical expression to convey an intended speaker's meaning or implicature.2

'This claim should be distinguished from the thesis (discussed in 2.1. below) that for every metaphor we can find
a corresponding comparison. Such a view does not require a literal expression for the property which the members
of the comparison have in common.
2
In 2.2, it will appear that pragmatic approaches indeed seem bound to assume that the specifically 'metaphorical
content' can always be paraphrased in literal terms.
 Introduction 13

For various reasons, older authors do not fit in well; Aristotle, for example, seems to hover
between a semantic and a pragmatic acccount, but as he makes no clear distinction between
linguistic meaning and language use in his theory of metaphor, it would be an anachronism to
assign him to either position. This tallies well with another possible objection, namely that
some theories, especially those expressing a mentalist view of meaning, would reject the
distinction between the second and third column; on such accounts, there is little or no
difference between concept and descriptive information to begin with. In such cases, one might
collapse the second and third column into one single 'representationalist' view. However, good
arguments can be given for maintaining a distinction between the semantic notion of meaning
(or descriptive information) and the epistemological or psychological notion of a concept.

Overview

After this elementary background information, I would like to give the reader some idea of
what to expect in the following. The theory presented in chapter 3 below comes closest to a
semantic descriptivist approach, but it incorporates notions such as context-dependence and
the distinction between presupposed and asserted information, which others might think of as
pragmatic. In chapter 4,1 then give a reconstruction at the conceptual level of what happens in
metaphorical interpretation. While the semantic and conceptual aspects of metaphorical
interpretation should be kept distinct, it seems that the level of concepts ultimately has more
explanatory force. A full-fledged theory of metaphor, then, should include both a semantic and
a conceptual component. But before jumping to such conclusions, a lot of preparatory work
must be done. The first chapter presents a brief historical overview, obviously not intended to
be anywhere near complete, of several important theories of metaphor dating from a more
remote past. This overview is preceded by a, somewhat speculative, discussion of how meta-
phor came to be seen as a distinct phenomenon at all; I will argue that historically, the rise of
literacy was an essential prerequisite for the development of a concept of metaphor. The
authors discussed in chapter 1 (Aristotle, 'Abd al-Qahir al-Jurjani, Giambattista Vico) may
not all seem equally obvious choices for an historical overview, but the importance they attach
to the phenomenon of metaphor, or their influence on subsequent theorizing on the subject,
justifies to some extent the attention devoted to them.
Chapter 2 presents a systematic discussion of some of the most influential twentieth-
century theories of metaphor within the framework of the classification scheme presented
above. After the problems facing purely semantic treatments like Black's and Beardsley's
have been noted, it will be seen that pragmatic theories in the tradition of Grice and Searle
14 Contexts of Metaphor

mark no significant step forward with respect to these problems. Moreover, they do not
present any principled argumentation as to why metaphor should be treated in terms of
language use rather than linguistic meaning; an approach which, after all, involves an appeal to
quite distinct principles. It seems unlikely that any knock-down argumentation can be given;
but on the whole, metaphorical interpretation appears to depend more centrally on
'conventional' rather than on 'intentional' factors. Donald Davidson's influential views face
essentially the same difficulties as pragmatic accounts. Worse, his denying of cognitive content
to metaphors as such leads to a number of rather counterintuitive results. Another influential
approach, Lakoff & Johnson's 'cognitive semantics', wants to dispense with all talk of
reference and truth as traditionally conceived, and treats metaphor as primarily a reflection of
the way we conceptualize the world. While their approach yields some interesting insights, it
presupposes many questions regarding metaphor as solved. Moreover, its stress on cognitive
aspects downplays the crucial role of linguistic, social, and conventional factors. Both
Davidson's and Lakoff & Johnson's views rightly call attention to particular aspects of
metaphorical interpretation (respectively, its indeterminacy and its relation to concept
formation), but both overshoot the mark. It appears that a semantic treatment of metaphor
along relatively well-established lines can come a long way in accounting for the specifics of
metaphor, though it will have to be enriched with more explicit contextual parameters.
Part 3 proceeds to sketch such an alternative semantic treatment, which I call a 'direct con-
textual interpretation' approach. It stresses the context-dependence of property expressions
in both literal and metaphorical interpretation. But metaphors not only depend on the context;
they may also change it. Consequently, section 3.2 discusses the assertive power of meta-
phor, and the distinction between asserted and presupposed information in metaphorical inter-
pretation. I will argue that metaphorical interpretation is triggered by what is presupposed,
rather than by what is - or seems to be - asserted, as most earlier semantic and pragmatic
accounts hold. The semantic account cannot capture all aspects of metaphorical interpretation,
however, notably with respect to novel and 'cross-categoriaP metaphors. Part 4 therefore
attempts to supplement this semantic treatment of metaphor with an account in terms of
concepts and concept formation. Such a theory should take into account the context-
dependence of metaphor at the semantic level, as well as broader sociocultural factors like
literacy and education. Vygotsky's theory of concept formation is a plausible candidate for
this; I will modify his account on the basis of more recent insights. In conclusion, I will look at
how the notion of the literal fares in the light of this account of metaphor. If, as is suggested
here, metaphor is in no important way deviant or distinct from literal language, then the notion
of literal meaning loses much of its theoretical importance as well.
CHAPTERS FROM THE HISTORY OF METAPHOR

1.1. THE PRE- AND PROTOHISTORY OF METAPHOR

Theorizing about metaphor is usually believed to have started with Aristotle. But obviously,
metaphorical language,or at least language we would call metaphorical, has been used since
much earlier times. The oldest written texts, such as the Mesopotamian epics and the Homeric
poems, are replete with language that looks roundly figurative to us. But how was this language
experienced and interpreted by those who first heard or read these texts? How was poetic
language experienced before it was written down? To assume that the original audiences
recognized metaphor as distinct and deviant from literal language may involve an unwarranted
projection of our own theories onto other people's ways of thinking. Now some authors, like
Rousseau, Vico, and Max Miiller, have indeed argued that 'primitive' man thought in distinct
terms, and more specifically that language is metaphorical in origin. But then what gave rise to
the recognition of, and theorizing about, metaphor as somehow different from literal language?
In this section, I will look at some of the cultural prerequisites for a notion of metaphor. I
will not directly address the question of the presumed metaphorical origin of language, but
rather discuss findings from contemporary nonliterate languages and 'primitive societies'.
There is of course a danger in this: how can we know that the linguistic behavior of illiterate
members of contemporary rural, non-industrialized societies, which have a long process of
historical development behind them, is at all comparable to the way ancient peoples used and
experienced language? Nevertheless, authors like Luria( 1976) and Goody (1977) have argued
that ancient contemporary nonliterate thought and language use do display important structural
parallels, and that this parallelism is precisely located in the absence of literacy in both. I will
therefore focus on the role of literacy in linguistic development. It will appear that in preliterate
societies, it is difficult to make any strict distinction between literal and figurative language, so
that the very question of whether or not 'language is metaphorical in origin' would seem to
prejudice the problems to be addressed. These findings also have wider repercussions for the

15
16 Contexts of Metaphor

conceptual aspects of metaphorical interpretation; but except for some brief discussion in
sections 1.4 and 2.4,1 will only return to these questions in the final chapter.

1.1.1 Prehistorical and Preliterate Societies

Ethnographic descriptions often report metaphorical utterances that to us appear as bizarre as

they sound poetical; to the peoples uttering them, however, they do not seem to be very odd
or deviant. The widespread occurrence of such metaphors, and their apparent lack of perceived
oddity, present difficulties to many existing theories of metaphor, most of which view
metaphor as derivative from literal language, and as based on an awareness of the utterance
being inappropriate in its literal interpretation. It has also led authors like Rousseau and Vico to
view 'primitive man' as a poet, and to think of metaphor rather than literal languageas lying at
the origin of language in general. A modern adherent of such a view is perhaps George Lakoff,
who thinks of linguistic metaphor as a manifestation of a basic cognitive capacity to 'see one
thing in terms of another'. But a notion of metaphor as involving a mapping from one
conceptual domain to another would seem to presuppose a 'literal domain', so it is not im-
mediately clear how a romantic view of metaphor as the original form of language could be
coherently formulated. For how could metaphor as a mapping between two - logically prior -
domains be 'original' in any coherent sense? Here, I would therefore like to look at the status of
metaphor in 'primitive' societies, and at its relation to literal meaning and classification. What
are the category distinctions that members of primitive (that is, nonliterate and non-urbanized)
societies impose on the external world, and how, if at all, does metaphorical language involve a
clash with, or a transfer between, such categories?
Let me start with some examples, to begin with what is probably the most famous metaphor
in anthropology: the remark pa e-do nabure ("we are parrots") uttered by Brazilian Bororo
Indians. This sentence is uttered in rituals where Bororo males adorn themselves with feathers
of the bird in question. Even there, however, it is asserted only of males; outside of this
specific context, the Bororo seem far less inclined to assent to it, let alone utter it
spontaneously (Crocker 1985: 38). To Karl von den Steinen, the ethnologist who first noted
this usage, and to Durkheim & Mauss, who discuss it in their study of 'primitive classification'
(1963 [1903]: 6-7), it suggested that the Bororo could not distinguish between men and
animals, and more in general had difficulties in categorizing the world around them. It seems
rather implausible, however, to conclude that some people do not really categorize the objects
 Pre- and Protohistory 17

in their surroundings at all. More specifically, the utterance 'we are parrots' in context
correctly applies to men only, and it involves a tensed copula form e-do which suggests a
'customary form' rather than a permanent state of being: 'the first morpheme, e, denotes
"existence", [...] while the second morpheme, do, denotes the activity of making or doing'
(Turner 1991: 135-6). It would seem more adequate, then, to treat this utterance as a figure of
speech rather than as a false categorial statement or as an indication of the absence of
classification. Subsequent research has largely focused on the question of precisely which figure
the utterance involves,1 but the most salient fact to be noted here is that the Bororo do not
appear to consider it false, deviant or inappropriate given the linguistic and actional context of
the ritual.
As a second example, Rosaldo (1972, 1975) presents specimens of metaphor among the
Ilongots, a loosely structured and unstratified society of hunters and swidden agriculturalists in
the Philippines. She notes that the Ilongots have as many as 13 different names, each indicative
of body parts, for orchids; these names are used in magical spells for curing people. Used as
category names to indicate specific kinds of orchids, however, they are used quite
inconsistently by different speakers, and even by single speakers on different occasions;
instead, specific names 'appeared as descriptive titles, designating sets of plants as appropriate
to a certain context or kind of spell' (1972: 86). Thus, the same kind of flower may be called
either ge-lawagide, 'their fingers', or qudungde, 'their thighs', both on the basis of some
perceived similarity with the named body part; to distinguish among closely related plants,
Ilongots may also employ color terms or the adjectives 'male' or 'female'. The plants used in
spells are generally taken from places inhabited by a specific spirit, and steamed so that the
spirit's body enters that of the patient; the plant name is then used to threaten the spirit:

(1) Here are your fingers, spirit; I steam your fingers, spirit.
They will be knotted, spirit.
Make him well now (1972: 85)

The Ilongots thus appear to use body-part names metaphorically for the ordering of an
otherwise unstructured domain of orchids, on the basis not so much of specific similarities as a
'contextually relevant equivalence' (1972: 94). The use of one name rather than another is
determined by the specific purpose of the spell, rather than by the properties inherent in the

1 Crocker (1985) argues for its metaphorical character; Turner (1991) holds that it is partly metaphor, partly
metonymy, and partly synecdoche.
18 Contexts of Metaphor

plants. The Ilongots, then, classify orchids not on the basis of shared abstract features or
properties, but rather as the occasion arises, and for specific contextually determined purposes;
and the same seems to hold for the Bororo classification of men as parrots.
The apparent appropriateness of such metaphors in context, paired with an apparently
unsystematic and highly situated and context-dependent way of categorizing objects, poses a
problem for the numerous theories that treat metaphor as violating abstract linguistic rules or
category distinctions, or as exploiting universal principles of rational communication. These
theories, several of which will be discussed in chapter 2 below, assume that at the level of
literal meaning,metaphors are 'defective' in one way or another, which again presupposes a
notion of a relatively stable and context-free 'literal meaning'. More specifically, they assume
that the 'literal defectiveness' of metaphors typically consists in a categorial mistake. Thus,
the languageuser is supposed to possess a number of more or less fixed and stable semantic
and conceptual categories; but these assumptions do not square well with the way the above-
quoted metaphors function in their natural habitat.

Classification and Literacy

Assumptions like these are also present in some anthropological studies of 'primitive
classification', such as the seminal work of Durkheim & Mauss (1963 [1903]), already referred
to above. Durkheim & Mauss argue that systems of primitive classification reflect the
organization of the societies in which they occur, rather than abstract logical principles.
Members of the simplest societies, they hold, do not distinguish at all between animals, people
and inanimate things (1963: 6). It is only when a society becomes differentiated into distinct
subgroups that subdivisions come to be made among the objects in the world; but these
subdivisions reflect the social order rather than any inherently cognitive processes. Thus,
Durkheim & Mauss argue that the Australian Wakelbura Aboriginals, who are divided into two
moieties ('halves') and four marriage classes, classify humans, animals, and plants in prcisely
the same 'categories' held by links involving (for us at least) metaphor or metonymy: objects
associated with the same moiety or marriage class are also 'conceptually' placed together
(1963: 13). Durkheim & Mauss stress that these associations are not seen as figurative by the
native speakers: 'whereas for us [the expressions referring to social and other ties] are hardly
more than metaphors, originally they meant what they said... Logical relations are thus, in a
sense, domestic relations.' (1963: 84). In other words, the grouping of ontologically different
objects under the same category is not perceived as involving any kind of figurative transfer;
 Pre- and Protohistory 19

application of a term belonging to one 'category' to an object associated with another would be
considered a social rather than a cognitive transgression.
As noted, this view implies that in the least differentiated societies, few or no category
distinctions are made at all. But this does not seem plausible: if an expression can be made to
mean anything, it means nothing. It also implies the wildly implausible conclusion that mem-
bers of primitive societies would be entirely unable to tell a kangaroo from a stone, or a tortoise
from a tree. In general, Durkheim & Mauss overemphasize the phenomenon of'social thinking'
as pervasive and homogeneously spread over an entire society, and consequently do not allow
sufficiently for variation and change originating from individual contributions; their rigid social
determinism would make individual variations in language use, extremely difficult to account for
(Goody 1977: 23; cf. Needham 1963, Worsley 1956). It suggests that novel classifications
literally cannot be thought in the absence of a change in social structure.
The British anthropologist Jack Goody (1977: ch. 4) has raised the interesting objection that
the very act of theorizing about primitive classification radically transforms its character, by
describing it in abstract terms that presuppose fixed meanings and stable categories (in 4.3
below, this point will reappear in a more general form in the work of the French sociologist
Bourdieu). Durkheim and Mauss represent nonliterate classifications in the graphic (i.e., script-
based) form of a table, and thus impose an order on 'primitive thought' which it may not
possess in itself. In oral societies, Goody argues, classification is not as systematic, exhaustive,
and decontextualized as it is in literate ones, so the attempt to force a more or less coherent,
fixed system based on essentially graphic representations onto these processes automatically
involves a certain ethnocentric bias. The 'contextual flexibility' that Goody attributes to
nonliterate languageusers is also indicated by the Ilongot orchid metaphors, and by the fact
that in many languages, basic vocabulary items such as kinship terms like father or mother may
equally well be used to indicate 'real' kin relations in domestic contexts as more generally to
express relations of affection or interest in public contexts, without the 'domestic' sense
necessarily being primary (cf. Leach 1982: 138-9). In short, Durkheim & Mauss make a
number of important suggestions regardingthe social basis of classification, but their claim of
social factors as actually causing classifications, their downplaying of its potential for change,
and their assumption of classifications as systematic and coherent, seem mistaken.
Another approach to nonliterate classification, which is based on the Russian psychologist
Lev Vygotsky's (1986 [1934]) work on concept formation, seems more promising as a way of
capturing what is specific to nonliterate language behavior. Vygotsky distinguishes between
what he calls complex concepts and scientific concepts; complexes, he argues, are context-
20 Contexts of Metaphor

dependent, and based on concrete factual bounds that may vary over different applications,
whereas scientific concepts involve the conscious employment of abstracted and decon-
textualized features. Scientific concepts are only acquired at a late stage of conceptual
development, and require literacy and formal education. Complexes thus represent an earlier
stage of concept formation; but also adults who have learned scientific concepts may resort to
complex thinking for everyday problems (see also Ch. 4.2 below). Vygotsky's 'cultural-
historical approach' to concept development naturally lends itself to other terrains than the
ontogenetic development of the child, although it need not imply a view of ontogenesis as
merely recapitulating phylogenesis. Vygotsky specifically addresses the phenomenon of
'primitive thought', in particular the apparently illogicalBororo utterance quoted above. He
warns against approaching nonliterate societies in terms of our own most developed scientific
concepts: 'primitive people think in complexes, and consequently the word in their languages
does not function as a carrier of the concept, but rather as a family name for a group of
concrete objects belongingtogether, not logically, but factually' (1986: 129). Once we realize
that nonliterate peoples think in complexes rather than in systematic scientific concepts, he
argues, we no longer need to ascribe an 'illogical'or 'prelogical'way of thinking to them, as
Lévy-Bruhl (1918) and to a lesser extent Durkheim do. He takes Lévy-Bruhl to task for
analyzing the Bororo utterance "we are parrots" in terms of a literacy-based logic that involves
identity assertions and the like, whereas on his own view the Bororo term for 'parrot' express-
es a complex that includes parrots and male Bororo: 'it does not imply identity any more than
a family name shared by two related individuals implies that they are one and the same person'
(1986: 130). I agree with these remarks as far as they go, but they do not sufficiently stress the
context-dependence of the language involved in the Bororo ritual. I would add that, in the ritual
and for the purposes of the ritual only, the male Bororo become parrots by dressing up with
feathers and assuming other parrot attributes, that is, by becoming factually related to parrots
in the (to us) stricter sense. The Bororo do not seem to feel a need for a decontextualized
classification of parrots in abstraction from specific situations, ritual or other.
The Vygotskyan line of thought is taken up by Jack Goody, who concentrates on the role of
writing in processes of classification. The basis for his argument is the role that early forms of
writing seem to play. Goody notes that lists are prominent among the earliest written texts
such as Sumerian clay tablets,- indeed surprisingly prominent, as they form a kind of language
use quite remote from spoken communication, and cannot be seen as in any way continuous
with oral discourse (see 1.1.2 below). The Sumerian lists were of various kinds, such as
inventories, lists of traded goods, and household statistics. Goody holds that such lists, which
 Pre- and Protohistory 21

constitute an essentially graphic mode of representation, force a greater systematization of the

languageon the speakers, since they require unequivocal decisions as to whether or not some
item belongs to a certain class or category. Writing thus makes languagemore decontextualized
and more discontinuous, as it both enables and requires the employment of strict categorial
boundaries that are abstracted from particular contexts. In such a perspective, writing is a
precondition for the systematic, consistent codification of words and their meanings, and thus
for the recognition of 'deviant' usages as such. This suggests that literacy is also an essential
precondition for the ability to distinguish between the literal and figurative use of expressions.
In the absence of such strict and decontextualized categorial boundaries, the notion of a
'categorial mistake' that many authors see as a prerequisite for metaphorical languagedoes not
make much sense. In other words, nonliterate individuals would hardly consider a particular -
contextually appropriate - utterance like "we are parrots" as figurative or otherwise odd at all.
It would be good to see if such broad claims are corroborated by ethnographic findings.
Fortunately, there have been several empirical investigations along the lines of Vygotsky's and
Goody's work. In the early 1930s, the Russian psychologist Alexander Luria, set out to test
Vygotsky's hypotheses in extensive field investigations among illiterate peasants in Uz-
bekistan. He wanted to find out whether the profound socio-economic and cultural changes
following the Soviet Revolution had any cognitive consequences, as a materialist theory of
psychology would predict. For this purpose, he conducted various experiments among people
with different degrees of exposure to the novel and quickly changing social situation: illiterate
peasants from remote villages, collective-farm activists, and persons with a larger amount of
education (Luria 1976).1
Luria specifically tested for the effects of literacy and schooling on classification, and his
findings were spectacular indeed. In one experiment, several unschooled and illiterate peasants
were presented with pictures of a hammer, a saw, a log, and a hatchet, and asked which one did
not belong there; later, the same question was put to subjects with more schooling. Among the
peasants, the response was practically unanimous. To quote a typical case:

"They all fit there! The saw has to saw the log, the hammer has to hammer it, and the
hatchet has to chop it. ... You can't take any of these things away. There isn't any you
don't need!"

1
At the time, Vygotsky's theories were considered insufficiently Marxist in character, while Luria's field
investigations were criticized for an alleged bias against national minorities (see Kozulin's introduction to
Vygotsky 1986: xli, xliii). Consequently, Luria only published his findings in the late 1960s.
22 Contexts of Metaphor

But one fellow told me the log didn't belong here.

"Why'd he say that? If we say the log isn't like the other things and put it off to one
side, we'd be making a mistake. All these things are needed for the log."
Look, you can use one word -tools- for these but not for the log.
"What sense does it make to use one word for them all if they're not going to work
together?"
What word could you use for these things?
"The words people use: saw, hammer, hatchet. You can't use one word for them all!"
Could you call them tools?
"Yes, you could, except a log isn't a tool. Still, the way we look at it, the log has to be
there. Otherwise, what good are the others?" (Luria 1976: 58-59)

The last remark is particularly revealing: even when explicitly presented with an appropriate
abstract categorial term, illiterate peasants would typically reject it as irrelevant, and at times
even as false. They were equally unwilling, for example, to classify both fish and crows as
animals. Their classifications thus appear to have a functional rather than taxonomic character.
One might surmise that the same persons would feel equally comfortable in classifying a log of
wood together with, say, a stove, a furnace, and a fireplace,- items that belong together
functionally, though not categorially. At the same time, illiterate language users appeared to
have no qualms about the figurative application of words. Ichkari women living in remote
villages in Uzbekistan, for example, freely used object names like 'spoiled cotton' or 'decayed
teeth' to indicate color hues for which their color term vocabulary was inadequate; at the same
time, they had great difficulties in dividing different colors into groups (1976: 24-7). This
suggests that they indeed freely extended expressions to new experiences, but were unable or
unwilling to classify them according to more abstract principles in isolation from immediate
experiences and purposes.
Literate subjects displayed a totally different behavior. Even those with merely one or two
years of schooling unhesitatingly grouped objects in terms of abstract categories like 'tools'.
Luria also investigated processes of syllogistic reasoning, imagination, and perception among
subjects with varying amounts of schooling. The results were uniform throughout: by and large,
illiterate subjects were unable, or unwilling, to abstract from immediate experience, concrete
situations, and concrete goals.1 This strongly suggests that illiterate agents employ a situational

1
Luria argues that Gestalt experiences likewise depend on sociocultural conditions; experiments involving optical
illusions, for example, yield rather divergent responses among the investigated groups (Luria 1976: 31-47). This
 Pre- and Protohistory 23

kind of thinking directed towards concrete goals, rather than classifying objects by the
conscious employment of abstract, context-free principles such as similarity or shared features.
The upshot of all this is that in nonliterate societies, classification does not appear to be as
strict and systematic as in societies where writing allows the listing and codification of
linguistic expressions. Classifications seem to be related to personal experience, the actual
context of use, and the language user's more immediate situational interests, goals and needs.
When nonliterate individuals have to make a categorial distinction or sorting among objects,
they may well be able to do so, although perhaps not to state in abstract, general terms why. In
everyday communication, such abstract and decontextualized criteria of classification are
usually not of much relevance to them. As noted, this position does not commit us to Lévy-
Bruhl's claim that illiterates have a 'pre-logical' mode of thought where familiar logical
principles like the law of noncontradiction do not hold. When confronted with two
contradictory sentences in a single context, an illiterate individual will presumably try to
maintain consistency by discarding one of them, or to reconcile them, e.g., by restricting their
ranges of application, much as a literate person would. In other words, Luria's work does not
imply that illiterate individuals cannot think in terms of abstract taxonomic categories; they
simply do not classify objects on grounds they consider useless or uninteresting. In the
absence of codified norms and means of registration, utterances largely remain tied to a specific
context, and are relatively difficult to check for mutual consistency. When utterances are
written down, such comparison becomes much easier.

Does Literacy Cause Cognitive Change?

But a nagging question remains: did Luria and Goody really establish that it actually is literacy,
rather than, for example, formal schooling or urbanization, that determines the presumed radical
changes in cognitive style? Research carried out by Sylvia Scribner and Michael Cole (1981)
among the Vai in Liberia presents evidence against such a sweeping conclusion. The Vai are
rather distinct in having their own syllabary script, which is learned outside institutional
settings, alongside the Arabic and Latin alphabets which are learned in Qur'anic schools in the
villages and in the state schools where English is taught, respectively. Some members of the
community are literate in the native script; others are literate in Arabic, English, or in several
scripts, while yet others are completely illiterate. This state of affairs enabled Scribner and

contradicts the widespread view that Gestalts are psychologically basic and universal; but for as far as I am aware,
no further investigations have been made to test Luria's preliminary and somewhat impressionistic findings.
24 Contexts of Metaphor

Cole to separate the variables of literacy and formal education. They also took factors like age
and socio-economic background (such as an urban, trade-related lifestyle versus an agricultural,
'traditional' one) into account. By thus isolating literacy as a variable from other sociocultural
factors, they could test whether it had any specific influence on cognitive capacities.
Most of Scribner & Cole's findings appear to contradict the broader claim that literacy by
itself leads to a major change in general cognitive abilities like that of classification; rather,
specific kinds of literacy turn out to lead to an improvement in specific skills. Literacy in
Arabic, for example, which mostly involves the memorization of verses from the Qur'an,
yields an increase in the ability for literal verbal recall, but not in classification abilities. Other
kinds of literacy, they argue, also lead to small changes, but not to the qualitative leaps that
Luria and Goody postulate: in itself, literacy is no substitute for schooling as a way of forming
general cognitive skills (1981: 116-133). The only factors that do seem to enhance the ability to
classify objects in terms of abstract superordinate categories and to give a general verbal
explanation for doing so (e.g., the classification of eggplants and kolanuts as food, and
motivating this choice by saying something like 'both are food items') are schooling at
secondary level and urban, trade-related living. But even these increases are not dramatic, and
require prolonged education. Rather than concluding that Vai illiterates are almost equally good
in context-free classification and explanation as literates, then, one might hold that most
literates perform almost as poorly as illiterates. Theoretical (i.e., decontextualized and abstract)
concepts, in other words, are a limit or ideal case rather than the most obvious way of
classifying objects.
Other experiments, however, especially those testing the ability to reason with syllogisms
abstracted from particular contexts and personal experience, did show significant differences
between schooled and unschooled subjects (though less between literate and illiterate ones).
Scribner & Cole also found confirmation for Luria's hypothesis that the ability to define
expressions in terms of abstract class membership varies with the kind of concept expressed:
'academic concepts' such as government or name, which belong to organized bodies of
knowledge typically transmitted through schooling, were more readily defined in abstract terms
than 'mundane' concepts concerning common objects from everyday experience (1981: 150; cf.
Luria 1976: 85-6). So Scribner & Cole's findings moderate Luria's to some extent, but they are
not completely at odds with them. Specifically, they do not run counter to the suggestion made
above that mundane concepts of nonliterates, and to a lesser extent those of literates, are to be
seen as complexes rather than as 'scientific concepts' (that is, as organized in terms of abstract
features and rigid boundaries).
 Pre- and Protohistory 25

In the light of these findings, Scribner & Cole prefer to see literacy not as an ability but as a
practice, that is, a 'recurrent, goal-directed sequence of activities using a particular technology
and particular systems of knowledge' (1981: 237). Different kinds of literate practice, that is,
tend to enhance specific abilities. The difference between complex and scientific concepts, in
other words, may depend as much on the specific practice in which concepts are employed
(that is, on their particular function and on the institutional setting) as on the general
developmental status of the cognitive agent.1

Literacy and Metaphor

What does the above imply for the status of metaphor in nonliterate societies? For as far as I
am aware, this question has not been investigated empirically, but some tentative hypotheses
may be formulated. Given the findings discussed above, one may assume that - in nonliterate
societies at least - literal meanings, abstract categories and conceptual domains play a less
prominent role than they are assigned in most modern theories of metaphor. The strict
distinction between literal and metaphorical language usage presupposes an awareness of
abstract features and categorial boundaries like 'living', 'nonliving', 'human', etc., all of which
emerge at a relatively late stage in concept formation only. Moreover, the employment of such
abstract features as the main basis for classification crucially involves cultural variables like
literacy and schooling. In other words, metaphor as a deviation from literal language, and as
involving a category mistake or a mapping between conceptual domains, cannot be considered
a universal or culture-independent notion.
The findings reported by Luria and Rosaldo do indeed suggest that illiterate subjects are
unlikely to reject 'figurative' sentences as deviant, ungrammatical, or as 'category mistakes' on
the basis of which they will reconstruct the speaker's intended meaning. Rather, what counts is
whether a sentence is situationally appropriate, that is, whether it is somehow applicable in its
context of utterance. Recognition of figurative language use as such seems to be relatively
independent of, and in any case posterior to, the successful use and understanding of an
utterance, as it depends on formal education and explicit knowledge of linguistic norms. The
interpretation of the Bororo utterance "We are parrots", and even of many of the everyday
metaphors that literate individuals encounter and interpret with little difficulty or conscious
deliberation, involves, I would suggest, complexes rather than scientific concepts. Rather than

1
Street (1984) more radically views 'literacy as ideology': he takes literacy not as a cause of conceptual change or
as a neutral skill, but as embedded in a complex of cultural and ideological practices.
26 Contexts of Metaphor

starting from categorial boundaries which are perceived as given, and violated by the utterance,
the hearer relies on some contextually present or relevant feature in virtue of which, say,
Bororo males and parrots may be grouped together (the Bororo would probably look for
feathers and parrot-like behavior). This feature need not be consciously employed, as it would
be if a scientific concept were involved, as the basis for the metaphor: for complex thinking, it
suffices that there be some factual or perceptual basis for the grouping together of humans and
parrots under the same label.
How, then, do illiterate individuals handle metaphorical language? I would suggest: much like
literal language. In context, a sentence which literates would rank as deviant or figurative on the
basis of some categorial anomaly may be just as acceptable to individuals in an oral society as
ones we would consider fully literal. No real or apparent category mistake at the level of literal
meaning is involved, because there are no fixed and stable literal meanings and categories to
begin with. This is not, of course, to deny the existence of linguistic norms of correctness in
oral societies. Rather, such norms of correctness are just of a different kind than those in
literate societies: they are not explicit or codified, but flexible and highly context-bound. In the
Bororo ritual, the actants actually become parrots in a sense; for us, this contextual sense may
be at odds with the 'literal' sense indicating a biological species, but for the Bororo there is no
such decontextualized literal sense to begin with: use in a figurative sense involves no violation
of any strict rules for literal language. In other words, the very distinction between literal and
figurative would sem to be meaningless for illiterate language users. The restriction of the
Bororo utterance to a well-circumscribedritual also speaks against naively attributing to the
Bororo a 'magical'world view which invloves a 'confusion' between man, parrots, and other
categories. The language used here seems at least as much of a performative as of a descriptive
character. More in general, the language of ritual is often performative: in ritual, social facts are
created and reproduced (cf. Tambiah 1968).
In short, the romantic claim that preliterate individuals 'speak in metaphors' is accurate up
to a certain degree only, and rests on a measure of ethnocentrism in that it assumes the notion
of metaphor to be a given, culture-independent notion, which it is not. By the same token, the
notion of literal meaning is not unproblematic or culture-independent. It seems that it is
writing, which allows language users to list or spell out the current or 'literal' uses of an
expression, that constitutes the crucial cultural prerequisite for an awareness of literal and
figurative meaning, and of a distinction between the two, as will appear in the next paragraph.1

1
The prime locus where 'literal meanings' are listed and codified is, of course, the dictionary.
 Pre- and Protohistory 27

The realization that 'scientific', abstract concepts and literal meanings are culturally
determined ideal cases rather than universal or given notions should cast doubt on their
relevance for a general theory of metaphor. It may very well be that even adult literate language
users rely on complex thinking rather than on scientific concepts in interpreting everyday
occurrences of metaphor: what counts in such cases is contextual appropriateness rather than
context-free categorial or conceptual boundaries. An adequate theory of metaphor, and by
extension an adequate theory of literal meaning, would do well to take such contextual
influences on conceptual processes into account, rather than taking a literacy- and education-
based ideal of stable, abstract concepts as its starting point. In other words, the above findings
regarding language understanding and classification in nonliterate societies may have important
consequences for a theory of metaphor in literate societies as well. They suggest that it is the
context of utterance, rather than abstract categories or mappings between conceptual domains,
that plays a primary role in interpretation, both literal and metaphorical. In chapter 3, a theory
of direct contextual interpretation will be outlined that takes these ideas into account. By way
of illustration of the possible role of literacy, I will now briefly discuss the historical origins of
writing, and the results this development may have had on the literal-figurative distinction.

1.1.2 Metaphor in Mesopotamia: Sumer and Akkad

It was in Southern Mesopotamia, towards the end of the fourth millennium BCE, that one of
the most momentous revolutions in the development of human societies took place: the
invention of writing. Around 3300 BCE, clay tablets with pictographic symbols appear in the
archives of the E-Anna temple in Uruk. The symbols which were used quickly became more
abstract, and through the introduction of grammatical particles and the like, a full-fledged
written languagegradually emerged. From the start, however, the same sign could be used for
different but related concepts (e.g. a sign of a foot could also mean 'to go' or 'to bring') or for
unrelated concepts pronounced with the same sound, e.g. a sign of a bow could (meto-
nymically) stand for 'arrow' (ti), or for 'to live' (ti or til). The precise interpretation of a sign
was determined by its linguistic context or by grammatical particles; for this reason, it is not
possible today to interpret with precision the earliest texts, which lack such particles (Roux
1992: 75).
For the present discussion, the relevance of this development is that it made possible the
codification of words and the determination or fixation of their meanings, and thereby the
28 Contexts of Metaphor

recognition of 'deviant' usages. In other words, writing is an essential precondition for the
ability to distinguish between the literal and the figurative uses of expressions. It should not be
thought, however, that the introduction of writing implies a radical dichotomy between 'oral'
and 'literal' or 'primitive' and 'advanced' societies. Goody (1977: ch. 1) has argued the need
for a developmental perspective, as literacy remained the privilege of a small social class for a
longtime, and did not become widespread until the nineteenth century. And in effect, we can
still see traces of oral traditions in the earliest written versions of Mesopotamian epics: apart
from a high degree of repetition, these show a 'magical' attitude to spoken words held to be
characteristic of oral societies, although this attitude may more adequately be said to reflect the
performative character of such ritualized language. For example, a name was believed to express
the nature of the named object; consequently, the same individual could be given different
names in different contexts (cf. Bottéro 1987: ch. 3). Thus, in tablet VII of the Enuma Elish,
the Babylonian Epic of Creation, the patron god Marduk is given no less than fifty names, each
expressing a different power. One might be tempted to conclude from this that 'primitive'
proper names function as property expressions that ascribe specific attributes to individuals in
virtue of their descriptive content, but perhaps such a statement already prejudices the
question of how these expressions signify, as it takes the notion of abstract attributes for
granted. Authors like Vico and Vygotsky raise doubts about such an assumption (see also 1.4
and 4.2 below).
As noted, Goody (1977, 1987) has stressed the significance of the fact that there are
disproportionately many lists among the earliest written texts: lists, he argues, are an important
aid in decontextualizing and systematizing the employment of linguistic items. Of particular
interest to us are the lexical lists that appear around 3000 BCE, but first occur in appreciable
quantities in Shuruppak around 500 years later (cf. Goody 1977: ch. 5). These lists seem to
have been primarily used for educational reasons: a good many of them contain bilingual
Sumerian-Akkadian glossaries 'which could be organized according to words with similar
meaning or even root structure, or as homonyms' (Geller 1987: x). To items grouped in a
particular lexical category, a determinative is often added, e.g., the sign mushen, 'bird', which
does not feature in the same way in the spoken language. In other words, writing is not just the
phonetic reproduction of spoken sounds, but actually imposes new structures and constraints
on the use of language.
 Pre- and Protohistory 29

Lists and Metaphors

It need not surprise us that figurative language occurs quite frequently in early Sumerian and
Akkadian writings, whether literary or administrative in kind. We already saw that a single
written sign could stand for various semantically related concepts; but a similar variability
occcurs in all known spoken languages. It is often difficult, however, to assess whether we are
dealing with live metaphors, dead ones, or simply with polysemic expressions that do not have
any clear-cut 'basic' meaning. Thus, Assyrian traders in Anatolia introduced terms like 'dying
tablets' or 'killing tablets' to fill in gaps in their commercial vocabulary; in this case, a technical
expression for the cancelling of records was introduced that did not imply the physical
destruction of the tablets on which they were written, as the near-synonymous terms 'erase' or
'break' would (Veenhof 1987). Moreover, according to Veenhof, this usage expresses the
Assyrian belief of the power inherent in the tablets, and draws the economic notions of guilt
and debt into the religious sphere.1 He does not explicate, however, in how far tablets were
actually believed to be living objects, as the vocabulary suggests; but then it may be not at all
easy to say anything substantial in these matters. As much of the basic lexicographicalwork in
Assyriology is concerned precisely with establishing the 'exact' meanings of expressions in
Sumerian and Akkadian texts, these problems are far from trivial: they involve extensive
investigations of the linguistic and cultural context of the written documents, of the available
synonymous expressions, of the basic world views of the language users, etc. The question of
whether a specific metaphorical usage is alive or dead at the moment it is written down is
certainly not the easiest question to be answered in this connection. It may not be the most
fruitful question either, as it assumes that a clear-cut distinction between metaphorical and
literal, and between novel and conventionalized metaphor can be made; and this may not be the
case.
Plainly, both 'spoken' and 'scriptural' kinds of metaphor arise from the need to express a
large number of concepts with a small number of signs, but there are some interesting differ-
ences between the two: spoken language already has a rich vocabulary, but faces difficulties for
the expression of highly specialized, culture- and domain-specific concepts. Written signs, by
contrast, are initially pictograms representing concrete objects, and subsequently come to
represent abstract concepts without their number becoming unmanageable.

1
Wilcke (1987) presents other examples of the catachretic use (i.e., use to fill lexical gaps) of metaphors in the
technical and scholarly vocabulary of Mesopotamian languages.
30 Contexts of Metaphor

A different question altogether is whether the Akkadian scribes were at all aware of the
figurative processes underlying the different applications of lexical items. There is some
evidence for thinking that they were. Geller (1987: x-xii) quotes from a lexical commentary
explaining the rare word li-id (Sumerian reading NI), meaning li-ti-ik-tu ('true measure') in
Akkadian:

gis
(The sign) NI (to be vocalized as) 1 i d (means) litiktu (which means mas ('twin-
rod'). (Quote): 'Father Enlil carried the lidda(-rod)..., the lord of the land carried the
lidda(-rody. (Quote); 'In order to .[..] the "twins" of lentil and sesame.' (Materials for
the Sumerian Lexicon 14 268)

gis
He argues that lid, meaning'litiktu', is explained by mas ('twin-rod') as the first quote is
explained by the second; at the same time, / i d relates to the first quotation as gismas relates to
the second. Thus, a relatively unfamiliar word A is explained by the analogies A : B = C: D and
A : C = B : D. As Geller notes, this explanation reminds one of Aristotle's definition of
analogon-metaphors in the Poetics (see 1.2 below).
According to Geller (ibid.), this example 'represents textual criticism based upon the
conscious use of metaphor and analogy'; there is no evidence, however, of an awareness of
deviance of any kind. To this it may be replied that metaphor involved here is probably
conventionalized in any case; but if what was said above is correct, this may not even be the
right kind of distinction to make. At this stage, words were just beginning to lose their 'event-
like' character, and still in the process of being turned into isolated and relatively autonomous
units of meaning. In other words, strict and decontextualized norms of 'literal meaning' were
only just beginning to emerge from the varying contextual uses of words. The lists may thus be
said to be as much descriptive as they are normative: the lexical lists are an essential aid in
distinguishing the different meanings of a word to begin with. There was not yet any strict
distinction between 'live' and 'dead' metaphors, as each use of a word in spoken languagewas
to some extent a 'unique event' (cf. Malinowski 1930).
In other words, the Akkadian scribes appear to have organized lexical items at least in part
employing ideas of figurative transfer; however, they do not seem to have had a single general
term to indicate this transfer, nor a general distinction between the literal and the figurative
application of a word. These 'theoretical' notions require a further elaboration of, and
abstraction from, the items of the various lists. At this point in time, it seems, such reflection
upon what was being done did not yet occur: the different uses were still in the process of
 Pre- and Protohistory 31

being codified, which is a necessary, but not yet a sufficient, condition for reflective thinking on
one's own language use. Consequently, we may call this phase the 'protohistory' of metaphor,
as the scribes do seem aware of the figurative processes involved in the different uses of a
single lexical item, but do not yet have a general concept of 'figurative' or 'metaphorical'
language use. The distinction between literal and figurative language, in other words, is not the
starting point of linguistic development, but the end product of a process of codification and
systematization. The concepts of metaphor and literal meaning essentially depend on script-
based and decontextualized categories.

1.2 ARISTOTLE ON METAPHOR, COMPARISON AND SIMILARITY

Many authors credit Aristotle with being the first to outline a theory of metaphor, and note its
influence on subsequent writings on the topic; others criticize him for precisely the same
reason. Thus, Johnson (1981: 5-8) sees Aristotle as responsible for initiating a pernicious and
all-encompassing Western tradition that treats metaphor not only as based on similarity, but
also as deviant from literal language, and as a matter of language rather than thought. Such
opinions, however, are not fully justified, because Aristotle's remarks are insufficiently
detailed and precise to allow the ascription of any such doctrine to him with certainty: his brief
discussions hardly add up to a full-fledged theory of metaphorical language. Nonetheless, his
views merit discussion, if only for the tremendous prestige they enjoy: hardly a single
twentieth-century study of metaphor passes over Aristotle in silence.1
Although the word μεταφορά in a linguistic or rhetorical sense occurs in a few presocratic
fragments, Aristotle was probably the first Greek philosopher to discuss the notion of
metaphor at some length. The main source for his views is the famous discussion in Chapter 21
of the Poetics. This treatment is limited to metaphor in poetical language, especially the
languageof tragedy. But even for the restricted purposes of poetics, the brief passage in the
Poetics (1457bl-33) remains rather incomplete. Aristotle merely lists four types of metaphor,
and discusses the ways in which these work, without addressing questions such as what makes

1
For the text of the Poetics, I have followed D. W. Lucas's edition (Oxford 1968). For the Rhetoric, I have used
the Loeb translation by J.H. Freese (1926); Further, I have used J.L. Ackrill's translation and commentary of
Categories and De Interpretatione, and W.D. Ross's English edition of Organon (1928) and Rhetoric (1924).
Although I have profited from various commentaries and interpretations (e.g., Stutterheim, Vahlen, Lucas,
Ricoeur, and Cooper), I will only refer to them incidentally, to avoid an inordinate lengthening of this section.
32 Contexts of Metaphor

a good metaphor, what happens to the meanings of the words involved in figurative speech,
and whether metaphors can be true. Some of these questions are discussed in Book III of the
Rhetoric (1404b25ff), which deals with prose language used for convincing an audience in
debate. Moreover, stray remarks on metaphor and related notions like ambiguity and similarity
appear in other works of the Organon.

Poetics

It is difficult to avoid anachronism in discussing Aristotle's views, and consequently, one can
hardly assign him an unambiguous place in the classification scheme presented in the
introduction; for example, he does not make a distinction between what we would call
'semantic' and pragmatic' aspects of meaning, and at times it is unclear whether he talks about
the words themselves, their meanings, the concepts associated with them, or their referents.
With these caveats in mind, let us turn to the Poetics. There, Aristotle treats metaphor at the
level of words (όυόματα);1 later authors, by contrast, see the distinguishing traits of metaphor
at the level of sentences or utterances. Another, and rather surprising, feature is the absence of
a general distinction between literal and figurative language. All words, he says (1457bl-3), are
either current (κνριον), strange (γλώττα), metaphorical (μεταφορά), ornamental (Koauoq),
coined (πεποιημενον), lengthened (έπεκτεταμενον), contracted (ύφηρημενον), or altered
(έξηλλαγμενον). Two things stand out here. First, it is less a classification based on any single
or unified criterion than a descriptive enumeration. As will become clearer below, it appeals to
rather different levels of linguistic analysis: phonological, semantic, and 'sociolinguistic'.
Although it seems a classification of word types, at times it implicitly appeals to utterances or
tokens of words; for example, a word can hardly be metaphorical in itself, apart from its
occurring in a specific sentence. Secondly, in this passage hurion, 'proper' as it is often
translated, is not contrasted with metaphora as much as with glôtta, 'strange' or 'uncommon,'
so in this context 'current' is a more appropriate rendering. The label kurion applies to a
word's status within a language (sub-)community: for example, the word σιγυνον is the current
word for 'spear' in the Cypriot dialect of Greek, but strange for the Athenian dialect (1457b4-
7). Stutterheim (1941: 68) argues that Aristotle, in Chapter 22, makes a general stylistic

1
stutterheim (1941: 70) argues that όνομα should be taken to mean 'name' rather than 'word' here, but Aristotle
himself gives examples of metaphorically used verbs, such as 'the sun sowing its rays' (1457b30). It seems wise
to follow the generic definition όνομα as 'spoken sounds significant by convention' (De Int. 16bl9), which
includes both nouns and (uninflected) verbs.
 Aristotle 33

distinction between kurion and xenikon uses ('foreign' or 'uncommon', defined as 'everything
that goes against current usage',-a blanket term covering all the other kinds listed); but even
such a broad distinction would not yet establish the former as signifying 'literal' in a purely
linguistic rather than a stylistic or sociolinguistic sense. Aristotle postulates that metaphors
employed in the distinguished language of tragedy and other forms of poetry should be
'uncommon' (1458a21ff), not that metaphors in general are in any way out of the ordinary. I
will return to this in the discussion of the Rhetoric below. In other words, 'current' as
characterized in Chapter 21 does not seem to be opposed to 'metaphorical', let alone to
'figurative'.
Next, Aristotle generically defines metaphora as a 'transfer of a name belongingelsewhere'
(1457b8). By this he means, according to Lucas (1968: 204), both the process of transfer of a
word and the word thus transferred. This ambiguity shows that we cannot with certainty
describe Aristotle's theory as either semantic (i.e., involving words and their meanings) or
pragmatic (involving the use of language). Significantly, this definition does not involve
'referents' (things) or 'meanings' (concepts). On his view, metaphors just involve a relocation
of words, and his definition does not yet yield any precise doctrine as to how the
interpretation of metaphor works. It does not give rise, either, to seeing metaphor as 'deviant'
or improper': the phrase 'elsewhere' merely implies that a metaphorical word (or word use)
involves placing a word in a new (verbal) context, without in any way suggesting that it is out
of place there.
Only when more concrete examples are discussed does anything like a theory of figurative
interpretation emerge. Aristotle distinguishes four kinds of 'metaphorical' transfer, from which
it appears that his notion of metaphora is much broader than the present-day 'metaphor':

1. from genus to species, i.e., using a more general term instead of the available more specific
one. For example, the use of the general term έσταναι ('to stand') applied to a ship, instead of
the more specific ('to lie at anchor');

2. from species to genus, e.g., a specific number ('a thousand') for the more general 'many';

3. from species to species, i.e., 'the use of a term of a different class' (Lucas ad loc.), as in the
swapped usage of begging and praying in 'begging to the gods' and 'praying for an aim';
34 Contexts of Metaphor

4. 'according to the analogous' (kata to analogori) or 'proportionally': for example, Ares and
the shield are related in the same manner as Dionysus and the phial, so one can call the shield
'the phial of Ares,' and the phial 'the shield of Dionysus'; in other words, where there is an
equation A : B = C : D, the term A (e.g., 'phial') may be used instead of C ('shield') in the
verbal context of D, yielding 'the phial of Ares'. One may also deny a proper (oikeion)
attribute of the transferred word, and call Ares's shield 'the wineless phial.'

Obviously, not all of these transfers belong to what we would nowadays call metaphor.
Benson and Prosser (1972: 245n2), following Cope (1867: 375), prefer to see the first two
kinds as synecdoche, and the third as metonymy. However, the third case is not necessarily
restricted to metonymy, i.e., transfer on the basis of factual contiguity; it may also apply to
cases of transfer which we would consider genuinely metaphorical. In fact, in Aristotle's own
example, there is no factual relation between cutting and drawing off, but merely a resemblance
between the two. Indeed, species-to-species transfer seems a prime example of metaphor on a
referentialist view, as it is based on a similarity. If two terms are used for species of the same
genus, they must be similar in some respect. In fact, it is the third rather than the fourth type
of μεταφορά which is based on similarity in the stricter sense of sharing a property. Both
notions involved in a species-species metaphor are similar precisely in falling under the same
genus; in the Rhetoric (1405a23), actions like praying and beggingare called similar in so far as
they belong to the same genus, that of asking.1 This kind of transfer, however, also seems to be
constrained by other factors, which Aristotle does not mention: calling a dog a cat, which
would be allowed in the above scheme because both are species of the genus animal, would
seem to be not a metaphor but a literal falsehood. It remains unclear how he would distinguish
the two. Neither does he discuss cases where two objects only figuratively share a property, as
in sweet words: whatever property words may have that makes them 'sweet', it is not the
same kind of sweetness as that of sugar or other foodstuffs. Presumably, this is where type-4
metaphors come in.
The fourth type discussed by Aristotle is the one of most interest to us. In fact, most
present-day authors consider this type the only 'genuine' case of metaphor. However, it is not
immediately clear how this type is actually distinct from the third; perhaps the species-to-
species metaphors are best seen as a subtype of the analogon-metstphors, as they are restricted
to terms falling under the same genus; analogon-metaphots do not have this restriction, as they

'A contemporary follower of Aristotle in this respect is John Searle (1979: 96), who argues that literal language
involves similarity just as much as metaphor does, so similarity is not a distinguishing feature of the latter.
 Aristotle 35

seem to involve an identity of relations rather than properties. Still, it is rather easy to
reconstruct many type-3 metaphors as type-4 ones: one might argue that praying stands to
gods as begging stands to passers-by, and because of this analogy the two verbs may be used
interchangeably.1
It should be clear that the term analogon is not identical with 'similarity' or 'comparison';
Aristotle uses the term homoiotès for the former, and eikôn for the latter. He defines the notion
as follows (1457bl7): 'I speak of analogy, when the second is related to the first in a
similar/identical way as the fourth is to the third' [emph. added]. The central word, homoiôs,
may mean either 'similar' or 'identical'. It seems that Aristotle himself tends towards the latter
meaning, witness, for example, the otherwise identical definition of proportion in the Nico-
machaean Ethics V.3.8 (1131a32) as an 'equality of ratios, implying four terms at the least'.
Consequently, in the Poetics he does not claim that a shield is like a phial, but rather that both
are related to Ares and Dionysus in the same way. In other (perhaps overly anachronistic)
words, analogy is not strictly speaking a relation of resemblance, i.e., an identity of (one-place
or 'internal') properties, but an identity of (two-place) relations (cf. 4.2 below; Miller (1979)).
This still leaves us with the question of how we find the appropriate analogy, i.e., how we
supply any possibly missing parts of the equation. If all four terms are known or explicitly
given, the transfer is obvious; but in other cases, where not all the related items are present, it is
less clear precisely which relation is involved. This is a familiar objection against a comparison
view (cf. Black 1962, Levinson 1983), but it would equally well apply to a conceptualist
version: just as the referents of words may stand in an indefinite number of relations to other
objects, so concepts may have an indefinite array of associations. To pick out the right one
from all of these possibilities, the hearer will need additional linguistic or contextual clues.
Aristotle seems to presume that typically, all four elements of this equation are given, unless
no word is yet available for a specific object or activity: for example, there is no word for what
the sun does with its light, so by analogy with the sowing of seed, it may be said to 'sow light'
(1457b30). Thus, the analogy would be an equation of four members with at most one
unknown (or at least not linguistically realized) variable.
The example of the sun makes clear that Aristotle does not deny the name of 'metaphor' to
cases where there is no conventional (κειμενον) expression to be replaced by the transferred

1
Levin's (1982) attempt to derive all four types distinguished by Aristotle from the fourth kind would give the
theory a unity that it does not possess at first sight. It leads to some rather far-fetchedreconstructions, however,
e.g. of species-genus metaphors to 'proportions' like deed : ten thousand : : deed: many. Levin himself already
notes that this does not really involve an analogy, as 'deed' figures twice in the relation (1982: 28). It remains
unclear, therefore, what this reanalysis adds to Aristotle's account.
36 Contexts of Metaphor

one (1457b25); moreover, there is no reason why such a metaphorical term cannot become
kurion, that is, the current way of expressing, if it is taken over by the speech community at
large. In this he also differs from many recent authors who restrict the proper use, so to speak,
of the term 'metaphor' to cases in which a literal expression is available, and are reluctant to
consider catachresis, i.e. the use of a word to fill a lexical gap, as genuine instances of metaphor
(Black 1962: 33, Davidson 1979: 32, and Searle 1979: 97-8, 100). In so far as a metaphorical
transfer fulfills a semantic need by filling a lexical gap, they argue, it will lose its metaphorical
status, and become the 'literal' way of expressing a fact. Aristotle is not committed to any such
sharp distinction between 'live' and 'dead' or 'genuine' and catachretic metaphors.1
Superficially, the definitions of species-species and analogon transfer may seem to justify
the attribution of a referentialist view to Aristotle: they strongly suggest that metaphors work
in virtue of some similarity between the literal referents of the expressions used. Many twen-
tieth-century authors have indeed claimed that Aristotle was the first adherent of a 'com-
parison' view of metaphor (e.g., Black 1962: 36n., Beardsley 1967: 285, Mooij 1975: 64f.,
Johnson 1981: 6). Now the ascription of a referentialist view to Aristotle requires that it is
indeed the referents that determine the metaphorical interpretation, rather than some other
aspect of meaning, or the concept associated with the word; but Aristotle nowhere unam-
biguously commits himself to such a view. A related, though distinct, question is whether in
metaphor (some aspect of) word meaning changes, and if so, which aspect: the sense, the
reference, or the concept. As Aristotle defines metaphor as just the transfer of a word, it is
difficult to credit him with any specific doctrine regarding these questions. In general, the
problem with ascribing a referentialist view to Aristotle is that it presumes an unequivocal
answer to the question of exactly what is transferred in a metaphor, and on the basis of what,
or in virtue of what, this transfer occurs.2 To determine whether or not Aristotle actually held a
referentialist view, we should be careful to distinguish two questions: first, whether metaphors
are interpreted on the basis of a similarity between objects; and second, whether they are a kind
of abbreviated comparison. For now, I will concentrate on the former question; the latter will
be addressed in the context of the Rhetoric.

1 Levin (1982: 27), too, notes that novelty 'is not an overriding consideration' for Aristotle.
2
One source of this general attribution of a referentialist view to Aristotle may be the fact that many authors use
By water's (1909) translation of the Poetics, which erroneously translates onomatos allotriou epifora as 'giving
the thing a name that belongs to something else'. The phrases in italics do not appear in the Greek original.
Bickerton (1969: 50n) already noted that Bywater 'smuggles into [the translation] several illegitimate
suppositions about the nature of language... with no shadow of foundation in the original text'.
 Aristotle 37

One remark in Chapter 22 of the Poetics has widely been taken as evidence that Aristotle
holds a referentialist view: he says that the gift for appropriate (πρεπη) metaphor cannot be
learned from others, because 'to metaphorize well is to observe the similar' (1459a8). It seems
to have been almost universally overlooked, however, that this remark is merely a claim about
what makes a good metaphor, not one about the workings of metaphor in general.1
The central question is precisely which are the elements of the analogon-relation A : B = C :
D. Are they the words themselves, the referents of the terms used, or the concepts expressed
by the words? Aristotle nowhere explictly says which one of these is involved, nor does he
distinguish them by typographical or other means, so the matter is not easy to decide.
Apparently, at least three members of the analogon-relation have to be either linguistically
realized or otherwise retrievable from the context of utterance, otherwise the equation A : B =
C : D cannot be solved. Moreover, in several of Aristotle's own examples, it can hardly be the
referents of the terms that are involved, as there simply are no actual individuals called
'Dionysus' or 'Ares'; the same holds for the names of fictional heroes in some other examples,
such as 'Achilles' and 'Odysseus'. This suggests that the interpretation of a metaphor like
'The phial is the shield of Dionysus' cannot proceed just in virtue of the referents and their
properties, but must involve representations or concepts. Aristotle himself does not appear to
notice this problem; but it seems to constitute a more general difficulty for his theory of
meaning, which does not contain an account of fictional objects.
Indeed, several of the main difficulties with Aristotle's views on metaphor derive in part
from his more general ideas on the relation between language, thought and reality. In chapter 1
of De Interpretations, he states that words in different languages are merely different symbols
of the same 'affections of the soul', which in turn are likenesses of things; he considers both the
affections and the soul and the things to be the same for all people.2 Aristotle thus assumes an
isomorphism between linguistic terms, mental entities, and things. Because of this isomorph-
ism, referentialist and conceptualist views of metaphor would seem to collapse on his treat-
ment; his remarks in different places are equally germane to both views. It would be somewhat

1
in fact, the only place where Aristotle does seem to express a clearly referentialist view is Topica (140a9):
'metaphor somehow makes the signified object known because of similarity; for all who metaphorize do so
according to some similarity' (emph. added). The explicit use of 'the signified' here suggests that it is indeed the
referents that are similar; but as said, the Rhetoric and Poetics are less explicit on this point.
2
Cf. De Int. 16a6ff.: 'but what these [spoken sounds] are in the first place signs of- affections of the soul - are the
same for all; and what these affections are likenesses of - actual things - are also the same' (transl. Ackrill). Ackrill
(ad loc.) notes how problematic this claim is; on Aristotle's view that language and thought are isomorphic, cf.
Benveniste (1966).
38 Contexts of Metaphor

more difficult to attribute a descriptivist view to him, as he does not talk of reified meanings as
distinct from concepts; usually he employs the transitive verb 'to signify' (σημαινειν) rather
than phrases like 'to have a meaning'.
It is difficult to decide whether Aristotle takes the 'things' or the 'affections of the soul' as
the basis for metaphorical interpretation. It is unfortunate that he does not discuss the special
status of non-denoting terms in this connection; for, as will be seen in 2.1 below, these provide
an important argument against strictly referentialist views. Obviously, the word Achilles may
very well stand for a meaning, concept or 'affection in the soul' without there being any object
of which the concept is a likeness. Perhaps this would justify us in ascribing to Aristotle a
more strictly conceptualist theory of meaning, which would allow for expressions to have a
meaning (i.e., to symbolize a concept) even in the absence of a referent. Analogon-metaphors
involving non-denoting terms could then be thought of as expressing analogies between
concepts rather than objects; but as said, Aristotle makes no explicit claims about this.
In short, the brief discussion of different kinds of metaphor in the Poetics leaves many
questions unanswered. Some of Aristotle's remarks point to a referentialist view, but in other
respects his views look more conceptualistic. Likewise, he makes no single, unequivocal
distinction between literal and metaphorical language. It is thus questionable to describe him as
a hard-nosed referentialist, for whom metaphor depends on similarity, is a matter of language
rather than thought, and is deviant by definition, as some authors (in particular Johnson 1981)
have done. Because Aristotle's views cannot be called strictly 'semantic' or 'pragmatic' either,
one objection that has been raised against them can quickly be dispensed with. Harris and
Taylor (1989: 32ff.) believe that his treatment of metaphor clashes with his conventionalist
theory of meaning: if words, being spoken sounds significant by convention, are symbols of
affections in the soul (De Int., Ch. 1), what kind of affections do words symbolize in
metaphorical use, which is not in general entirely based on convention? This objection can be
raised against any semantic approach to metaphor, and we will encounter it again below: how
can a theory that explains metaphor in terms of (conventionalized) meanings deal with novel
metaphors and the 'creation of new senses'? Harris & Taylor's argument, however, in part
rests on the unresolved problem of whether metaphor in Aristotle's theory involves a transfer
of meanings, as they believe (ibid.), or merely a transfer of words. More importantly, De Int.
16a26 merely requires that a sound functions as a symbol in order to be conventional, not that
it has a wholly fixed conventional signification. Further, conventions may obviously change
over time, and there is no reason to think that Aristotle would deny that metaphorical use may
 Aristotle 39

involve precisely such a change of convention: especially metaphors used to name the nameless
may easily become the conventional or current way of expressing something.

Rhetoric

Aristotle returns to the notion of metaphor in Book III of the Rhetoric, where he mostly
discusses the 'pragmatics', so to speak, of metaphor, i.e., its successful employment in
discursive or argumentative prose. It is in this context that the notion of the 'proper' (πρέπον)
use of metaphor appears. For Aristotle, the main virtue of oratory prose is clarity or
perspicuity, because its purpose is convincing the audience; for this reason, he considers it best
to speak with a somewhat exotic or 'foreign' air, while at the same time hiding the artifice
(ποιησις). This effect can be created by the use of words that are either current, familiar or
metaphorical. Strange, compound or coined words are to be avoided; these are best employed in
poetical language,which is not used to persuade but to give pleasure by showing its artifice,
i.e., by its loftiness and deliberate 'artificiality'. Thus, the Rhetoric shows even more clearly
than the Poetics that Aristotle does not think of metaphor as a deviation from regular usage: he
explicitly says that 'everybody uses metaphors, current and familiar terms in conversation'
(1404b33ff.); as in the Poetics, a general distinction between literal and figurative is absent.1
From this discussion it also appears that prepon, 'proper', is a property of style, not of lan-
guage, and thus cannot serve as an equivalent of 'literal'. A proper or appropriate style should,
among others, be in accordance with the subject matter, so it should vary with the purpose of
discourse. Aristotle was well aware that a style involving no figures of speech at all was
definitely not 'neutral:' he calls the styles of Cleophon and Sthenelos, which apparently lacked
all figuration, 'mean' (1458a20). This is another indication that he does not consider metaphor
deviant from any supposedly fixed, unequivocal and context-free norm of strictly 'literal'
language: such a norm would itself yield a very marked style.2
Aristotle further requires that a metaphor be appropriate (harmottousas), otherwise the
utterance will be 'improper' (aprepes),- improper, that is, to the purpose of the message
(1405al3ff). This appropriateness arises from the proportional: a metaphor puts things in a

1
This belies the interpretations of more recent authors like Johnson (1981: 6), who mistakenly attribute to
Aristotle a view of metaphor as being deviant by definition.
2
Likewise, Ricoeur (1975) repeatedly stresses that there is no such thing as a strictly 'neutral' and literal style.
The rather restricted stylistic or 'pragmatic' meaning of prepon also casts doubt on Derrida's attempt (1974) to
attribute an entire 'metaphysics oftheprepon' to Aristotle.
Another random document with
no related content on Scribd:
 Illustrationsverzeichnis.
Seite
Ta f e l n a u f b e s o n d e r e n B l ä t t e r n :
1. Der Verfasser mit Ashluslayfreunden. Neben dem Titel.
2. Der Calileguaberg 13
3. Dorf des Chorotihäuptlings „Waldhuhn“ 30
4. Fische essender Ashluslay 48
5. Mit Mais vom Acker kommende Ashluslaykinder 52
6. Chorotimama mit ihrem kleinen Jungen und dessen
Spielkameraden 62
7. Ashluslay-Tänzer 78
8. Tanzende Ashluslaymänner 84
9. Chorotifrau auf dem Heimweg mit eingesammelten wilden
Früchten und Holz 94
10. Matacoindianer rösten „Palometas“ und andere Fische 112
11. Chorotifrauen tragen wilde Früchte in ihren
Caraguatátaschen nach Hause 124
12. Ashluslaykrieger 136
13. Ashluslaymann im Magenpanzer 138
14. Ashluslayfischer gehen über den Rio Pilcomayo 142
15. Ashluslayfrau auf der Wanderung 146
16. Caraguatá 176
17. Palmenwald, unweit des Rio Pilcomayo 192
18. Chanéfrau mit Kind 220
19. Die Frau des Chanéhäuptlings Vocapoy malt ein Tongefäß 242
20. Sagenerzähler. Chané 258

A b b i l d u n g e n i m Te x t e :
1. Matacomädchen 8
2. Hütte der Mataco-Guisnay 9
3. Ashluslayfischer 19
4. Eigentumsmarken auf Mänteln, Ashluslay 36
5. Ashluslaypapa mit seinem kleinen Jungen 37
6. Kochhütte der Chorotis 40
7. Algarrobofrüchte kauende Ashluslayfrauen 44
8. Ashluslayindianer mit Sperrnetzen 45
9. Nadeln zum Aufreihen der Fische 46
10. Spaten 47
11. Sperrung des Rio Pilcomayo mit fischendem Choroti 49
12. Chorotikinder spielen, daß sie den Fluß sperren 50
13. Eine Chorotifrau trägt Wasser nach Hause 55
14. Ashluslayfrau seiht Algarrobomehl 57
15. Eßbürste 59
16. In die Erde gegrabener Ofen 60
17. Hölzernes Messer 60
18. Hölzerne Messer zum Essen von Wassermelonen 61
19. „Reibeisen“ aus Holz 61
20. Chorotiknabe mit Boleadora 64
21. Das Kleine führt seinen blinden Großvater „abseits“ 65
22. Die Mama geht mit den Kindern zum Fluß 67
23. Spielzeugflinte von den Chiriguanos 68
24. Boleadora 69
25. Fadenfiguren knüpfende Chorotiknaben 71
26. Ballspielende Matacoindianer 72
27. Spielmarken 73
28. Tätowierung und Gesichtsbemalung 75
29. Tätowierung und Gesichtsbemalung 76
30. Alte Chorotifrau, die den Verf. tätowiert hat 77
31. Ashluslaymann 79
32. Stempel zur Gesichtsbemalung 80
33. Chorotielegant 81
34. Ashluslay mit einer mit Schneckenschalenperlen besetzten
Mütze 83
35. Junger Chorotimann am Alltag 85
36. Pfeife 87
37. Boxhandschuh 89
38. Tongefäß 95
39. Ashluslay mit einer Kalebasse Algarrobobier 97
40. „Bowle“ 99
41–43. Pfeifenköpfe 101
44. Krankenstuhl 105
45. Geist 109
46. Tongefäß 117
47. Puppen 118
48. Strumpf, Taschen 119
49. Grabestock 120
50. Säge aus hartem Holz 120
51. Scharre aus Muschelschalen 120
52. Webstuhl 121
53. Von Ashluslays gewebter kleiner Mantel 122
54. Chorotifrau, ein Tongefäß bauend 123
55. Töpferin 124
56. Hölzernes Gerät 125
57. Trommel aus einem Tongefäß 126
58. Bierkrug 126
59. Wasserkrug 126
60. Kalebasse 127
61. Kalebasse 127
62. Kalebaßschale 129
63. Mais sammelnde Frau (Zeichnung) 130
64. Zeichnungen des Chorotimädchens Ashlisi 130
65. Zeichnungen eines Ashluslayknaben 131
66. Zeichnungen eines 20jährigen Ashluslaymannes 132
67. Ashluslaykrieger 133
68. Federschmuck 134
69. Federschmuck 135
70. Skalp eines Tobapilaga 136
71. Streitkolben 136
72. Ashluslaytänzer zum Besuch bei den Chorotis 139
73. Ashluslayfischer 145
74. Vocapoys Dorf am Rio Itiyuro 149
75. Chanéindianer 150
76. Chanéindianer 151
77. Tongefäß 158
78. Tongefäß 159
79. Alter Chiriguano mit großem Lippenpflock 165
80. Pfeife 169
81. Festtracht für Männer 169
82. Tongefäß 172
83. Feuerstätte zum Maisbierkochen 174
84. Maisscheune 175
85. Sitzbank 177
86. Haken zum Aufhängen der Sachen 177
87. Chanéfrau 178
88. Tabakspfeife 179
89. Spatenstiel 179
 90. Fischfang mit Kalebasse 185
91. Netz 187
92. Dämpfapparat 188
93. Spielregel für „Daro“ 191
94. Chunquanti spielende Chanéknaben 192
95. Stäbchen zum Tshúcaretaspiel 193
96. Máma wird ausgelegt 194
97. Die Stäbchen werden geworfen 195
98. Die Spielenden sehen nach, wie die Stäbchen gefallen sind 196
99. Geteilter Ball zum Tocorórespiel 197
100. Spielstock 197
101. Rakett 197
102. Maiskolbenpfeil 198
103. Stäbchen zum Huirahuahuaspiel 198
104. Spielzeug 199
105. Puppen aus Wachs 200
106. Chanéfrau 201
107. Gesichtsbemalung 202
108. Tätowierter Frauenarm 203
109. Kämme 204
110. Chanéknabe 209
111. Der Chiriguanohäuptling Mandepora 213
112. Alte Frau 215
113. Chiriguanograb 219
114. Junge Chanéfrau entblößt den Oberkörper, um sich
photographieren zu lassen 226
115. Kalebasse 231
116. Chanémädchen stoßen Mais in einem Mörser 235
117. Kochen des Maisbieres 237
118. Suppenspatel 239
119. Tanzmaske 241
120. Serérepfeife 243
121. Tongefäß 244
122. Brennen von Tongefäßen 245
123. Webstuhl 247
124. Sieb 248
125. Korb 248
126. Verzierte Kalebaßschale 249
127. Tongefäß 264
128. Kalebaßschale 279
129. Tongefäß 281
130. Silberne Nadel 287
131. Brustschmuck aus Silber 290
132. Chanékinder 299
133. Tongefäß 301
134. Tsirakuafrau 307
135. Taubstummenzeichen 316
136. Taubstummenzeichen 318
137. Taubstummenzeichen 319
138. Sandale aus Tapirhaut 321
139. Grabekeule 323
140. Wurfkeulen 325
141. Stück Zeug 326
142. Pfeife aus Holz 327
 Alphabetisches Register.
(Die Ziffern bedeuten die Seitenzahl; f nach der Zahl = u. folgende.)

Abstandsangabe 166.
Abtritte 205.
Aderlassen 54.
Aguararenta, s. Totenreich.
Aguaratunpa 222, 258, 274, 283, 285, 291.
Algarrobillo 121.
Algarrobo (Bier) 16, 17, 18, 22, 45, 49, 58, 96, 98, 111, 236, 260,
296.
Altersklassen zwischen Kindern 67.
Amulette 53.
Aña. Añatunpa 166, 251, 255 f., 286 f.
Angeschossene Tiere, Leiden 54.
Aphrodisiakum 225.
Araukanier 80.
Arbeit 4 f., 180 f., 190;
für die Weißen 4 f., 180 f., 300 f.
Arbeitsteilung zwischen den Geschlechtern 92 f.
Aristokratie 230.
Arowaken 22, 156, 169.
Aseptik 217.
Ashluslay 5, 16–147, 164, 167, 248.
Astronomie 294 f.
Atsahuaca 182.
Aussichtsposten 131.
Ava = Chiriguano.
Aymara 13, 182, 207.

Bäder 84, 203.

Ballspiele 71, 193 f., 264.
Bastian 114.
Baumwolle 51, 182.
Begräbnis 14, 218, 309;
mit dem Kopf nach unten 219.
Beigaben 108, 310.
Besitzrecht 35, 50, 232.
Bestialität 225.
Betten 41, 176.
Bewertung von Gegenständen 126, 140.
Blattschneideameise 242.
Blinde 216.
Blitz 11, 269.
Boas 114.
Bogen 52, 133, 183.
Boggiani 26, 48, 51, 138.
Bogoras 114.
Bohnen 52, 182.
Boleadora 70, 112, 183.
Boman 243.
Bordelle (Indianer, welche die Bordelle der Weißen besuchen) 9.
Boro 182.
Boten (Eil-) 131.
Boxhandschuhe 87.
Brandpfeile 134.
Branntwein 8, 11, 100, 240, 298, 302.
Brunnen 57, 328.

Caduvei 138.
Calileguaberg 11 f.
Campana, Domenico del 138, 161, 254, 259.
Campos 137, 142.
Cangui = Maisbier.
Caraguatá 23, 119 f., 244, 269, 320, 324, 326.
Caraguatátaschen 42, 49, 61, 116, 140, 326.
Cardus 324.
Chacobo 174, 182, 207.
Chamacoco 138.
Chañar (Bier) 20, 45, 49, 59, 62, 96, 98.
Chanés 5, 7, 14, 113, 134, 148 bis 303, 310 f., 322.
Chicundapa 258.
Chiquéri = Chiquéritunpa 258, 264, 277, 296.
Chiquitos 194.
Chiriguano 5, 7, 13, 48, 86, 91, 94, 134, 138, 139, 148 bis 303,
310 f.
Chomé, P. 210, 211, 212.
Chorotis 5, 6, 7, 16 bis 147, 164, 167, 221, 248, 254, 310, 320.
Chuchio 166, 280.
Chuña 70, 112.
Churápa 224.
Coca 12, 182.
Corrado 38, 208, 228.
Couvade, s. Wochenbett 92, 206.
Crevaux 26.
Crysocol 179, 201, 211, 240, 289.

Dampfkochen 189.
Daro 190.
Diät s. Speiseverbot.
Diebstahl 34, 231.
Diener, s. Klassenunterschied.
Dolmetscher 34.
Donner 252, 258, 271, 296.
Donnergott = Chiquéritunpa.
Dörfer 32, 33, 173.
Duelle 232.
Dyori 271, 273 f.

Ehe 89 f., 211 f.

Ehrenreich 22, 114, 255, 293.
Ehrgefühl 43.
Ehrlichkeit 144 f.
Eidechsen 22 — s. auch Iguana und Téyuhuasu.
Eifersucht 38, 87.
Eigentumsmarken 35.
Eisen, Handel mit 138, 311, 326.
Eitelkeit 78.
Elternmord 38, 231.
Empörung gegen die Weißen 5, 134, 168, 233.
Erbschaften 7, 232.
Erdbeben 12.
Erdhöhlen 21, 327.
Erdratten 22, 58, 222.
Erinnerungen, für die Indianer kostbare Gegenstände als 171.
Erzeugung (eigentümliche) 112, 114, 271, 276.
Erziehung 10, 63, 209.
Esperanza, Zuckerfabrik 3, 27.
Eßbürsten für Honig 59.

Fadenspiel 70.
Fahrzeuge 48, 186.
Familienhaus 174.
Farbensinn 81.
Federarbeiten 126.
Feldbau 51, 131, 324.
Feldmanöver 133.
Feuerraub 13, 22, 110, 252 f., 312.
Feuerstätte 41, 176.
Feuerzeug 41, 252.
Fische, Handel 139.
Fischerei, Fischgerätschaften 46, 98, 184.
Flamingo (Sagen) 112, 284.
Fledermaus (Sage) 264.
Fliege (Sagen) 263, 288.
Floh (Sage) 262.
Frau 74;
Stellung im Gemeinwesen 90;
Häuptling 229;
unverheiratete 95.
Friede 129 f.
Friederici 18, 24.
Frosch (Sage) 254, 314.
Fruchtbestände, wilde, Besitzer der 50.
Früchte (Einsammeln wilder) 49.
Fuchs 219;
Sagen 113, 222, 256, 269, 289 f.
Fuchsgott = Aguaratunpa.

Garcilasso de la Vega 167.

Gären von berauschenden Getränken 96, 237.
Gastfreiheit 144, 190.
Geduld bei Schmerzen 107.
Gefangene, Auslösung 24, 135, 306.
Geier (Sagen) 22, 112, 263, 314.
Geister, s. auch Aña, 108, 257.
Gemeinwesen 29, 232.
Geograph, der Indianer als 163 f.
Gerben 123, 247.
Geruch 85.
Gesang 6, 18, 85, 97, 141, 157 f., 179, 223, 276, 325.
Geschlechtskrankheiten 9.
Geschwisterehe 90, 111, 212, 254.
Gesichtsbemalung 78, 203.
Giftige Pflanzen als Nahrung 51.
Grabekeulen 322.
Gräber 14, 108, 218, 310.
Grabestock 49, 120.
Grabhaus 14.
Graburnen 14, 218.
Guanako 155.
Gúarayús 52, 165, 207.
Gumilla 195.
Gummibälle 193, 264.
Gummigegenden 298, 305.
Gürteltiere 22.
Gürteltiergott = Tatutunpa.

Haare, Ausreißen 78, 84;

 Seil aus 121;
weißes 216;
Abschneiden bei der ersten Menstruation 210;
Kurzschneiden bei Trauer 108, 219, 253, 310.
Hagel 12.
Handel 137 f., 248.
Hängematten 177.
Hartt 293.
Hasardspiele, s. Spiele.
Häuptlinge 33, 170, 212, 228 f.
Hausbau 40.
Haustiere 55, 187.
Hawtrey 57, 328.
Heilmittel 107, 217.
Heirat zwischen Cousin und Cousine 90.
Herrmann 27, 51, 142, 300.
Heuschrecke (Sage) 269.
Historiker, der Indianer als 167 f.
Höchstes Wesen 259.
Holzalter 93.
Hölzerne Pfrieme 54.
Hölzerne Sägen 50, 120.
Holzschnitzerei 125.
Holzstäbchen, Spiel mit 72, 191.
Homosexualität 221.
Honig 54;
 Einsammeln 54.
Humor 239.
Hunde 53, 111, 184, 187, 314;
Begräbnis 56;
Verhexung 106;
windhundähnliche, Canis jubatus 22.
Hütten 39, 173.

Ibareta 26.
Iguana, s. auch Téyuhuasu.
Im Thurn 204, 255.
Industrie 111 f., 242.
Inómu 271, 283.

Jagd 52, 170 f.

Jagdhütte 54, 251.
Jagdtrophäen 53.
Jaguar 12, 20, 23, 111, 218, 272, 289, 291.
Jahreszeiten, Bestimmung der 52, 183.

Käferlarve (Nahrung) 54.

Kalebassen 52, 125, 247, 268.
Kämme 205.
Kara-Kara, Vogel (Sagen) 112, 263.
Karneval 240.
Kartenskizze 149.
Kartoffeln (süße, Bier) 52, 236.
Kegelspiel 192.
Keramik 60, 122, 159, 178, 234, 243.
Keulen 53, 183, 324.
Kinder 63, 92, 193 f., 207 f.
Kindermord 38, 88, 208, 231.
Kindermörderinnen 38.
Klassenunterschied 14, 33, 231;
zwischen den Stämmen 14, 94, 311.
Kleidertracht 80, 83, 171, 200 f., 240, 325.
Klistier 224.
Knochenschmerzen, Mittel gegen 12.
Knöcherne Pfrieme 54, 87.
Koch-Grünberg 90, 127, 221.
Kochhütte 40.
Kolibri (Sage) 262, 292.
Kommunismus 34.
Kondor = Ururuti.
Konservatismus 125, 147, 211, 220, 235.
Konserven 42, 59.
Kopfkissen 41.
Kopftrophäen 172.
Korbarbeiten 61, 125, 248.
Krankenstuhl 105.
Kreuz auf dem First 11;
auf Gräbern 328.
Krieg 19, 39, 68, 129 f., 171, 306.
Kugelgürteltier (Sage) 112, 291.
Kultureinflüsse von Peru 12, 242.
Kulturgrenze 249.
Kulturmythen s. Sagen.
Kulturpflanzen 51, 182, 324.
Kunst 116, 234, 242.
Kürbis 51, 252, 267, 271, 314;
Raub 13, 111.
Küsse 88.

Laufspiele 70.
Läuseessen 85.
Lausen (Sage) 286.
Leben im Jenseits (Glaube an) 108, 255 f.
Lehmann-Nitsche 27, 86.
Lenguas 48, 51, 57, 137, 224, 249, 328.
Liebe (freie) 78, 86, 87, 94.
Lippenpflock 211, 312.
Lozano 28.
Lügenhaftigkeit 36 f.

Mahlzeiten 59, 110.

Mais (Bier) 12, 51, 96, 98, 161, 181, 188, 237, 244, 258, 275,
278.
Mais, wie die Tapiete Mais bekamen 312.
Maiskolbenpfeile 196.
Maisraub 12, 112, 270.
Mandioca 51, 182.
Männer, unverheiratete 95.
Männerhäuser 33, 97.
Manslé 137.
Mariutunpa 258.
Marschfähigkeit 166.
Maße 139, 192, 196, 315.
Matacos 4, 5, 7, 15, 18, 22, 26, 28, 48, 70, 72, 105 f., 107, 108,
110 f., 127, 130, 132, 135, 225, 248, 320.
Mataco-Guisnays 17, 26, 89, 101, 135.
Mataco-Notén 89.
Mataco-Vejos 13, 125, 231.
Medizinmänner 12, 34, 103 f., 159, 216, 230, 252, 258, 296.
Meerschweinchen (Sage) 13, 111.
Menstruation 74, 107, 210.
Meteor 110, 295.
Militärgesetz 302.
Minderjährig, der Indianer wird betrachtet als 34.
Missionen, Missionare 109, 148, 151 f., 161 f., 168, 214, 216, 220,
246, 259, 297 f.
Mistol 45, 204, 278.
Mode in Schmucksachen 81.
Mogeln beim Spiel 73.
Mond 52, 110, 273, 295.
Mondfinsternis 110, 294.
Mord 38, 231;
s. Kindermord, Elternmord.
Mörser 60, 237.
Musikbogen 86, 199.
Musikinstrumente 86.
Mutterbaum 260.
Mützen 80.
Mystische Wesen 109.

Namen (Spott-) 208.

Namengeben 92, 208.

Öfen 60, 299.

Ohren, Durchbohren der 74.
Ohrenklötze 18, 74.
Onanie 112, 221, 225.
d’Orbigny 194, 277.
Orientierungsgabe 165.
Orion 294.
Ornamentik 78 f., 116 f., 125 f., 234, 244, 246 f.
Ortsnamen 32, 166.

Pachamama 11.
Paddeln 48.
Palometa 23, 108, 186.
Papagei (Sage) 266, 270, 276, 312.
Parfüm 85.
Paterson, Dr. 107.
Pelleschi 115.
Pelzmäntel 84, 124.
Pfeife 86, 171, 242 f., 249, 265, 285, 325.
Pfeifen, Tabak 102, 200.
Pfeile 52, 133, 183;
mit Stumpfspitzen 53, 209, 272.
Pfeilwerfen 196.
Pferde, Begräbnis 55;
gestohlene 140.
Phantasie 117.
Plejaden 52, 183, 294.
Pocken 108, 158.
Prostitution 95.
Puppen 69, 117, 248.

Quebracho 111, 269, 272.

Quichuas 11, 12, 13, 153, 182, 207, 219, 242, 247, 249, 321.

Raketts 193.
Regen 296.
Regenbogen 296.
Rehböcke 22.
Reifen (schlagen) 70, 199.
Reinlichkeit 84, 190, 203.
Reisezehrung 59.
Religiöse Vorstellungen 109 f., 114, 250 f.
Rosen, Eric von 23, 27, 41, 115, 138.

Säen 52, 183.

Sagen 4, 13, 21, 109 f., 222 f., 250 f., 312 f.
Sägen aus Holz 50.
Sagenwanderung 114, 293.
Salz 58, 189.
Sandalen 80, 203, 306, 326.
Sandfloh 63.
Schafe, wie die Tapiete sie bekamen 313.
Schamgefühl 88, 225.
Scheibenschießen (Spiel) 10, 69.
Scheunen 174 f., 183.
Schießwaffen 10.
Schildkröte (Sage) 53, 291.
Schimpfworte 157.
Schlaf 43.
Schlägereien zwischen den Frauen 38, 87, 141.
Schlange (Riesen-) 110, 281, 288.
Schlangen 23.
Schlangenbiß (Heilung mit Menstruation) 107.
Schleudern 12, 53.
Schlingen 54, 183.
Schmucksachen 80 f., 82 f., 240.
Schneckenschalen, Handel 137.
Schöpfungssage 260.
Schwalben verkünden Regen 296.
Schwangerschaft 206.
Schwirrholz 10.
Seelenwanderung 256.
Selbstmord 224.
Siebe 61.
Silberne Schmucksachen 240, 263, 288.
Siluroid 23.
„Sintflut“ 253.
Sipoy 22, 311.
Sirionos 155, 322.
Sitzbank 177.
Skalpe, Skalpieren 17, 24, 68, 90, 132.
Sonne 252, 282, 295.
Sonnendach 33.
Sonnenfinsternis 294.
Sonnenzelt 33.
Späher 131.
Spaten 52, 183, 266, 269.
Speiseverbot 190, 206, 210.
Sperrung des Flusses 48, 129.
Spiele 65, 190 f., 224, 281 f.
Spielsachen 68, 193 f., 291.
Sprache 28 f., 157, 310, 324.
Stammbaum 229.
Stammnamen 28.
Steine, Armut an 92.
Steinen, K. v. d. 119, 127, 207, 221.
Stelzen 70.
Stempeln der Ornamente 79 f.
Sternbilder 294.
Sternhaufen 295.
Sternschnuppen 295.
Stier (Sage) 104.
Stillen 63;
junger Hunde 56.
Strauße 20, 54;
(Sage) 264, 292 f.
Strümpfe, zum Schutz gegen Fische 23.
Sturm (Sage) 252, 296.

Taba 190.
Tabakraucher (Anbau des Tabaks) 51, 101, 182.
Tacumbocumba = Viscacha.
Tänze 17, 78, 85, 106, 141, 240, 325;
bei der Menstruation 174.
Tanzmasken 239.
Tanzspiele 71.
Tapietes 5, 28, 137, 139, 151, 153, 154, 164, 248, 304 f.
Tapietes = Yanaygua.
Tapuy = Chané.
Tasi 22, 45.
Tatutunpa 258, 264 f., 269, 271 f., 285 f.
Tätowierung 74f. , 89, 203, 312, 323.
Taubstumme 207, 315.
Tauschhandel, s. Handel.
Téyuhuasu 222, 276, 283.
Theokratie 230.
Tiere, wilde, werden zahm gehalten 41, 56, 187.
Tipaytunpa 258.
Tiri 277.
Tobas 5, 6, 7, 10, 15, 19, 24, 28, 39, 48, 50, 129, 132, 134, 135,
139, 161, 248.
Tobas-Pilagas 18, 132.
Tod 218, 309.
Tongefäße, s. Keramik.
Tonkugelbogen 53, 184.
Tonscherben als Spielsteine 197.
Tote 255 f.
Totenreich 157, 255 f.
Totora 125.
Trachten, s. Kleidertrachten.
Tränengruß 24.
Trauer 24, 108, 179, 219, 253, 309.
Träume 257.
Trigo, L. 134, 142, 146, 308.
Trinkgelage 17, 91, 96, 132, 212, 234, 252, 257, 274, 283, 291.
Trommeln 86, 97, 115.
Trophäen, s. Kopftrophäen und Skalpe.
Tsirakuas 60, 80, 137, 153, 155, 164, 171, 305 f., 322 f.
Tsirióno = Sirióno.
Tukan (Sage) 287.
Tunpa 251, 257, 314.
Türkisperlen 179, 201, 211, 240, 289.
Tusca 22, 45, 49, 96, 121.

Ungeziefer 85, 176.

Urapua (der schwarze Geier, Sage) 274.
Uruku, Handel mit 138.
Ururuti, der weiße Kondor (Sage) 256, 262 f.
Uturunco 12, 219.

Vejos, s. Matacos.
Verbrechen, s. auch Vertreibung der Leibesfrucht, Kindermord
usw. 231.
Verführung der Frau eines anderen 232.
Verhexen 103, 106, 159, 216, 232, 239, 258.
Verjüngung 268.
Versprechen 17.
Vertreibung der Leibesfrucht 38, 88, 208.
Viedma 168, 174, 200, 203.
Viehzucht 55, 187.
Vielweiberei 90, 212.
Viscacha (Sage) 260.
Vogelnetze 183.

Wachs 54.
Waldfeuer 22.
Waldhuhn (Sage) 222.
Wanderungen 5 f., 32, 204, 311.
Wasserfall im Rio Pilcomayo 20.
Wassermangel 17, 25, 52, 311, 329.
Wassermelone 51, 96.
Waten 186.
Weben 121, 246, 326.
Wege 24, 33.
Wegweiser 164.
Wegzeichen 131.
Weiße, Verhältnis zu, s. auch Arbeit, 142 f., 300 f.
Weltbrand 13, 21, 111.
Weltuntergang, s. auch Weltbrand, 251.
Werbung (Braut-) 212.
Westermarck 224.
Wettlaufsage 292.
Wildkatze (Sagen) 13 = Embarakaya 273.
Worte (häßliche) 157, 221.
Würfelspiel 190 f.

Yahuanau, schwarze Wesen 166.

Yahuapinta (Puma) 273.
Yahuéte = zweiköpfiger Jaguar = Yahuaróhui.
Yamandutunpa 257.
Yanaygua = Tapiete.
Yuracáre 22, 52, 79, 224, 277, 315.
Zahlwörter 30, 73, 315.
Zähne, Schwärzen 203.
Zahnschmerzen, Entstehen 314.
Zamuco 324, 328.
Zapallo, s. Kürbis.
Zauberer, s. Medizinmänner.
Zauberspaten 266, 269.
Zecken (Sage) 262, 292.
Zeichen der Würde 228.
Zeichnungen, ausgeführt von Indianern 127 f., 248.
Zeiten der Not 50 f., 60.
Zeremoniell bei den Festen 238.
Zubereitung der Speisen 58, 188.
Zweiköpfiger Jaguar 272, 294.
Zwerge 327.
 Berichtigung.

Seite 22 Zeile 12 v.o. „Tasifrucht“ anstatt „Tuscafrucht“.

116 8 v.u. „Caraguatábast“ anstatt „Caraguatárinde“.
125 14 v.o. „Korbarbeiten“ anstatt „Kostbarkeiten“.
138 3 v.o. „Schneckenschalen“ anstatt „Muschelschalen“.
211 14 v.o. „Pflock“ anstatt „Knopf“.

Nördlicher Teil der Karte