By Word of Mouth

BY WORD OF MOUTH
Pragmatics & Beyond

New Series
Editors:
Jacob L. Mey
(Odense University)
Herman Parret
(Belgian National Science Foundation, Universities of Louvain and Antwerp)
Jef Verschueren
(Belgian National Science Foundation, University of Antwerp)
Editorial Address:
Linguistics (GER)
University of Antwerp (UIA)
Universiteitsplein l
B-2610 Wilrijk
Belgium
Editorial Board:
Norbert Dittmar {Free University of Berlin)
Bruce Fraser (Boston University)
John Heritage {University of California at Los Angeles)
David Holdcroft {University of Leeds)
Catherine Kerbrat-Orecchioni {University of Lyon 2)
Beatriz Lavandera {University of Buenos Aires)
Marina Sbisà {University of Trieste)
Emanuel A. Schegloff (University of California at Los Angeles)
Paul O. Takahara (Kobe City University of Foreign Studies)
Sandra Thompson {University of California at Santa Barbara)
Daniel Vanderveken {University of Quebec at Trois-Rivières)
Teun A. van Dijk (University of Amsterdam)
33
Louis Goossens, Paul Pauwels, Brygida Rudzka-Ostyn,

Anne-Marie Simon-Vandenbergen and Johan Vanparys
By Word of Mouth
BY WORD OF
MOUTH
METAPHOR, METONYMY AND LINGUISTIC
ACTION IN A COGNITIVE PERSPECTIVE
LOUIS GOOSSENS
PAUL PAUWELS
BRYGIDA RUDZKA-OSTYN
ANNE-MARIE SIMON-VANDENBERGEN
JOHAN VANPARYS
JOHN BENJAMINS PUBLISHING COMPANY

AMSTERDAM/PHILADELPHIA
The paper used in this publication meets the minimum requirements of
American National Standard for Information Sciences — Permanence of
Paper for Printed Library Materials, ANSI Z39.48-1984.
Library of Congress Cataloging-in-Publication Data

By word of mouth : metaphor, metonymy, and linguistic action in a cognitive perspective /
Louis Goossens ... [et al.].
p. cm. -- (Pragmatics & beyond, ISSN 0922-842X ; new ser. 33)
Includes bibliographical references and indexes.
1. Metaphor. 2. Metonyms. 3. Pragmatics. 4. Cognitive grammar. 5. Values. I. Goossens,
Louis. II. Series.
P99.4.M48B96 1995
306.4'4--dc20 95-34671
ISBN 90 272 5045 6 (Eur.) / 1-55619-326-2 (US) (alk. paper) CIP
© Copyright 1995 - John Benjamins B.V.
No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any
other means, without written permission from the publisher.
John Benjamins Publishing Co. • P.O.Box 75577 • 1070 AN Amsterdam • The Netherlands
John Benjamins North America • P.O.Box 27519 • Philadelphia PA 19118-0519 • USA
Table of Contents
Introduction vii
A Survey of Metalinguistic Metaphors 1

Johan Vanparys
Body Parts in Linguistic Action: Underlying Schemata and Value Judge- 35

ments
Paul Pauwels & Anne-Marie Simon-Vandenberg en
Assessing Linguistic Behaviour: A Study of Value Judgements 71

Anne-Marie Simon-V andenbergen
Levels of Metaphorization: The Case of Put 125

Paul Pauwels
Metaphtonymy: The Interaction of Metaphor and Metonymy in Figura- 159

tive Expressions for Linguistic Action
Louis Goossens
From Three Respectable Horses' Mouths: Metonymy and Convention- 175

alization in a Diachronically Differentiated Data Base
Louis Goossens
Metaphor, Schema, Invariance: The Case of Verbs of Answering 205

Brygida Rudzka-Ostyn
References 245
Subject Index 251

Introduction
This volume contains the results of a series of empirical investigations into the
expression and conceptualization of the domain of linguistic action in English,
with a focus on figurative extensions into (and in one case also from) that do-
main. More specifically, all these studies are concerned with metaphorical and
metonymic extension as generally understood in cognitive linguistics and are
thus explorations into the nature of these conceptual processes.
We would like to characterize our book as empirical, because the research it

presents is corpus-based, in the sense that it relies on language data which have
been gleaned systematically from a variety of sources, supplemented in one in-
stance by data elicited from native speakers by means of questionnaires.
Since we want to contribute to the understanding of the ways in which
linguistic action is expressed and conceptualized by speakers of English, our
approach can be taken to be conceptual-empirical in the sense of Verschueren
(1985a and 1987a). The best way to situate our contributions, however, is to in-
scribe them into the growing body of cognitive linguistic work. Our main points
of reference therefore include Lakoff and Johnson (1980), Johnson (1987),
Lakoff (1987), Langacker (1987 and 1991), a variety of articles in Rudzka-
Ostyn, ed. (1988), Geiger and Rudzka-Ostyn, eds. (1993) and the journal
Cognitive Linguistics.
Our concentration on metaphorical and metonymic meaning extensions is moti-
vated by the view that metaphor and metonymy are central aspects of our con-
ceptual apparatus. Although there is a vast literature on these extension proc-
esses, it is our belief that the cognitive linguistic approach applied to a vast em-
pirical body of linguistic action data has yielded new insights into these crucial
mechanisms of meaning extension and hence into the manner in which human
conceptualization works. For one thing because the domain of linguistic
(inter)action is highly complex and plays an irreplaceable part in our everyday
lives, for another because our data bases are sufficiently large and varied to re-
flect how this complexity is given expression in very diversified and often subtle
ways.
v111 INTRODUCTION
There are seven contributions. The first five (Vanparys, Pauwels/Simon-

Vandenbergen, Simon-Vandenbergen, Pauwels and Goossens-1) are synchronic,
in the sense that they explore data from contemporary English. The main data-
base here was the Longman Dictionary of Contemporary English (LDOCE),
which we scanned systematically for figurative expressions relating to linguistic
action. For details about this, see Vanparys's article. It may suffice at this point
to say that our systematic search yielded 1916 entries. These data were supple-
mented from other sources, mainly lexicographic ones, whereas Anne-Marie
Simon-Vandenbergen relies on elicitation data. The last two chapters have a
diachronic orientation, and obviously use data from older language stages: Æ1-
fric, Chaucer and Shakespeare in Goossens-2, the Oxford English Dictionary
for Rudzka-Ostyn (though for her paper an extensive contemporary data base
was used as well).
An important feature of the synchronic data base, which correlates di-
rectly with the fact that it largely comes from general usage dictionaries, is that
it is made up of conventionalized, rather than creative metaphors. In other
words, the metaphors (and metonymies) under scrutiny are those that are part
of most native speakers' everyday talking and thinking; or, to put it differently,
those that have stood the test of time, and which possibly reveal something
about the factors that make a metaphor or metonym successful. This adds to the
representativeness of our exploration from the point of view of con-
ventionalization (though, of course, it characterizes our material as predomi-
nantly non-literary). This feature allows Vanparys, for example, to survey the
overall distribution of source domains operative in the conceptualization of lin-
guistic action in Present-day English. For the other papers, which highlight more
specific aspects of metaphorization and metonymization, it means that the in-
vestigations are rooted in a general usage database, and that it becomes possible
to include statements about the relative frequency of the types of metaphor that
are focused upon.
It may have become clear by now that, although there is considerable common
ground underlying them, the following studies have their own individual vantage
points, and are only partially intended to form one coherent whole. To bring the
reader's expectations in line with what follows, let us survey the conliibutions
one by one, in the order in which they succeed each other in the volume.
In À survey of metalinguistic metaphors, Johan Vanparys first explains
in some detail how our main database (from LDOCE?) was established, and
INTRODUCTION iX
evaluates its merits and its biases. Taking for granted that by and large it reflects
general usage, he then proceeds to a systematic survey of the source domains
showing up in it. He is able to give an idea of how the diversity of source do-
mains ties up with the diverse complexity of the linguistic action domain. One of
his conclusions is that Reddy's (1979) claim that our language about linguistic
communication is almost exclusively dominated by the conduit metaphor is a
gross overstatement.
In Body parts in linguistic action: underlying schemata and value
judgements, Paul Pauwels and Anne-Marie Simon-Vandenbergen focus on
metaphors of linguistic action whose source domain is the (human) body and its
functioning. As the title of their paper makes clear, they investigate a specific as-
pect of metaphorical extension, namely the expression of value judgements.
They examine various concrete subdomains which are frequently exploited, and
for which, because of this concreteness, the experiential grounding is fairly ob-
vious. Such concrete experiential grounding is not always available, however.
For a considerable number of instances they have to fall back on more abstract
image schemata (in the sense of Johnson (1987)). The way in which the con-
crete subdomains interact with the more abstract image schemata is also ex-
plored. This yields interesting conclusions about scales that are relevant to value
judgements, as well as new insights into image schemata: they emphasize the
importance of a control schema (which is not provided for by Johnson), the pos-
sibility of coming up with a hierarchization among image schemata, and the
overall importance of image schemata in the metaphorical expression of value
judgements.
In more than one respect Anne-Marie Simon-Vandenbergen's paper As-
sessing linguistic behaviour: a study of value judgements is complementary
to the preceding one. Making use of British native speakers' intuitions (which
were obtained via questionnaires), she examines to what extent conventional-
ized metaphors reflect positive or negative evaluations of certain aspects of lin-
guistic behaviour. She finds out that the expression of value judgements is the
rule, and that most of the time they are negative. Her study also sheds light on
the contribution of specific donor domains to the expression of value judge-
ments, where it is revealed, amongst other things, that the value judgement may
be transferred from the source domain directly, or may come in only after the
extension to a specific aspect of the target domain. Also the role of the image
schema of scale in speakers' assessments of linguistic (inter)action is clarified.
The whole is put into further perspective by emphasizing that value judgements
X INTRODUCTION
associated with the metaphorical expressions investigated exhibit both a con-

text-independent and a context-dependent dimension, and that in the assessment
of the experiments which form the basis of her contribution a certain bias must
be taken into account which comes from the specific subculture to which the
(comparatively young) subjects belong.
Paul Pauwels (Levels of metaphorization. The case of put) next con-
cerns himself with the observation that not all metaphors can be assigned the
same degree of cognitive salience. His database consists of 42 expressions with
the verb put, whose common denominator is the domain of object manipulation,
but which link up with a host of more specific (source) domains at the same
time. It appears that metaphoricity, as he calls it, comes in different degrees, and
that these degrees correlate with similar degrees of recoverability. Generally his
data confront us with the question at what level of specificity transfers from one
domain to another are made, and how these different levels affect the recover-
ability of the donor domain. Pauwels also emphasizes that in a number of in-
stances the motivation for the metaphor is to be situated in metonymy, a point
which paves the way for the next contribution.
In Metaphtonymy: the interaction of metaphor and metonymy in ex-
pressions for linguistic action, Louis Goossens explores the transitional zone
between metonymy and metaphor. In collecting our data, as well as in our in-
vestigation of them, we were repeatedly confronted with the fact that in some
instances a clear-cut separation of metaphor from metonymy was not possible.
This problem is investigated on the basis of data from three donor domains
(violent action, sound, body parts). It appears that two types of interaction be-
tween metonymy and metaphor (metaphtonymy, in terms of the neologism in
the title) predominate, viz. metaphor from metonymy and metonymy within
metaphor. Metonymy from metaphor, on the other hand, as well as metaphor
within metonymy are extremely rare. The paper gives a number of suggestions
as to why this asymmetry should exist.
Goossens' second article, From three respectable horses' mouths: me-
tonymy and conventionalization in a diachronically differentiated data
base, is centrally concerned with metonymy in a diachronic perspective. The
data come from (expressions with) the symbolic item mup, mouth(e), mouth, as
it appears in Ælfric, Chaucer and Shakespeare respectively. Throughout the
history of English, mouth has exhibited frequent extension to linguistic action.
Initially the mapping appears to be almost exclusively metonymic, with a sudden
growth of metaphorical items in Shakespeare. An analysis of these data permits
INTRODUCTION XI
Goossens to shed light on two issues. The first is the literal-metonymy-

metaphor continuum, where the end point are metaphors from metonymy; but,
as can be expected, there are also other patterns of interpenetration between
metonymy and metaphor. The second point concerns the problem of con-
ventionalization. A number of factors are uncovered that appear to have pro-
moted, or, as the case may be, have worked against long-term conventionaliza-
tion.
Brygida Rudzka-Ostyn, finally, confronts verbs of answering (answer, re-
join, reply, respond and retort), with a focus on the relation between metaphor
and schematization (Metaphor, schema, invariance: the case of verbs of an-
swering). She presents a schematic network analysis based on an extensive con-
temporary corpus for each of these verbs, as well as a diachronic network based
on the OED that shows the successive meaning extensions over time. She un-
covers the image-schemata which are/were operative in these extensions, both
from a synchronic and from a diachronic point of view (the prior-subsequent
schema, the direction reversal schema, the correspondence schema, etc.). This
allows her, for example, to confront the Invariance Hypothesis as formulated by
Lakoff and Turner. Rather than saying that the image-schematic structure of the
target must be preserved in metaphorical extension, and that as much of the
image-schematic structure of the source as is consistent with that preservation
must be imported into the target, she emphasizes the inescapable abstraction (or
schematization) that is involved. Metaphor, she concludes on this point, pre-
serves a perception of similarity as expressed by a schema applicable to both the
source and the target domains.
Summarizing, the following areas of investigation have received attention.

• The differentiation in source domains for the metaphorical expression of lin-
guistic communication in English, and their relation to the complexities of
linguistic action as a target domain.
• The role of the axiological parameter in the motivation of metaphors for lin-
guistic action, as well as the experiential grounding of metaphors that express
value judgements. More specifically, the part played in this by image-
schemata of various sorts, the way in which positive and (especially) negative
value judgements come about, and the importance of the socio-cultural em-
bedding of value judgements.
XII INTRODUCTION
• The graded character of metaphoricity and its correlation with degrees of

cognitive recoverability/salience. This vantage point again ties up with the
general question of what motivates a given metaphorical expression.
• The interaction and interpenetration of metonymy and metaphor. One aspect
of this study is the question what factors are involved in the conventionaliza-
tion of metonymic expressions, which in turn includes the perspective that
conventionalized metaphors frequently have metonymies as their origin.
• The part played by image-schemata in the organization and the development
of a given lexical subfield, which raises new questions on the nature of meta-
phorical extensions, the identification of source and target domains, as well
as on the Invariance Hypothesis.
The research underlying this book has been carried out under the auspices of the
National Science Fund of the Flemish Community, and for two of the authors
(Louis Goossens and Brygida Rudzka-Ostyn) in the context of a research pro-
gram supported by a Belgian government grant (IUAP-II, contract number 27).
It constitutes the continuation of work by the research group which issued The
scene of linguistic action and its perspectivization by SPEAK, TALK, SAY and
TELL (by René Dirven, Louis Goossens, Yvan Putseys and Emma Vorlat; 1982,
also in the series Pragmatics and Beyond, vol. 111:6).
Other work that is the direct or indirect outcome of our collaboration is
reported on in Goossens (1986, 1987, 1989, 1990a, 1990b, 1993a, 1993b and
1994), Rudzka-Ostyn (1988) and (1989), and Simon-Vandenbergen (1991).
Three contributions included here (with slight modifications) have been pub-
lished elsewhere, as Goossens (1990), Pauwels and Simon-Vandenbergen
(1993) and Rudzka-Ostyn (1994). We are grateful to Mouton De Gruyter
(Berlin) and to Peeters (Leuven) for permission to reprint this material.
Each of the authors takes responsibility for her or his contribution(s),
though it must be emphasized that we have greatly profited from each other's
comments. This introduction is the result of combined efforts as well. Finally,
we wish to thank Hans Paulussen (Notre-Dame de la Paix University, Namur)
for sharing his computer expertise with us.
A Survey of Metalinguistic Metaphors
Johan Vanparys
Notre-Dame de la Paix University, Namur
0. Introduction
The goal of cognitive linguistics has been defined as an attempt "to characterize
those psychological structures that constitute a speaker's linguistic ability, i.e.
his grasp of established linguistic convention" (Langacker, 1988: 130). Our
knowledge of language is largely implicit, so that it needs to be explicated by
expert theories. However, there are also intuitive conceptions of language, as
reflected in the metalinguistic repertoire of the lexicon. For instance, the
existence of verbs such as promise, request, assert, apologize, etc., points to
pre-theoretical conceptions about speech acts (cf. Searle 1979, Searle &
Vanderveken 1985). Similarly, hedges like par excellence, loosely/strictly
speaking, technically, etc. indicate our intuitive understanding of linguistic
categorization (cf. Lakoff 1973). Such conceptions are of special relevance to
linguistics, since what we do with language is at least in part determined by
what we think we do (Verschueren 1985a).
It is therefore not surprising that metalinguistic terms have received due
attention in linguistics (eg Ballmer & Brennenstuhl (1981), Dirven et al. (1982),
Leech (1983: ch. 9), Rudzka-Ostyn (1989), Searle & Vanderveken (1985: ch.
9), Vanparys (1993), Verschueren (1985a, (ed.) 1987), Wierzbicka (1985a,
1985b, 1987)). Most studies deal with items that literally designate aspects of
verbal communication. However, "a great many concepts (perhaps most
abstract concepts) are structured metaphorically" (Lakoff 1989: 103). As a
result, metaphorical conceptualizations need to be investigated, as well.
2 JOHAN VANPARYS
1. Previous research
The most influential account of metalinguistic metaphors to date has been

Reddy's (1979) identification of the so-called 'conduit metaphor', according to
which ideas (or meanings) are objects and words (or other linguistic expres-
sions) are containers; speakers put ideas into words and transfer them through a
kind of conduit to the hearer, who extracts the ideas from the words. This
metaphor is well-documented with examples, such as
(1) Don't force your meanings into the wrong words. (id.: 287)
(2) Apparently, your reasons came through to John quite clearly. (id.:
311)
(3) I have to struggle to get any meaning at all out of the sentence.
(id.: 314)
Reddy claims that this is the "preferred framework for conceptualizing
communication" (id.: 286) in English and concludes that "as a metalanguage
[...] English is its own worst enemy" (ibid.) since the conduit metaphor "can
bias our thought" (id.: 285), for metaphorical expressions tend to highlight
certain aspects of their designatum and hide others. He also points to a "minor
framework", which "overlooks words as containers and allows ideas and
feelings to flow [...] into a kind of ambient space between human heads" (1979:
291): ideas are ejected by speakers into external space, where they exist in
reified form, independent of any human beings; they may or may not find their
way back into human heads. Examples include:
(4) I feel some responsibility to get these ideas out where they can do
some good. (id.: 316)
(5) That concept has been floating around for centuries. (id.: 318)
(6) Harry just won't let certain kinds of thoughts sink in. (id.: 319)
Given the abundance of examples that can be adduced to support the conduit
metaphor, in both versions, there is no reason to doubt its existence nor even its
pervasiveness. It is my contention, however, that Reddy's Whorfian pessimism
is unfounded. About his methodology, he says: "for the past several years, I
have been collecting some new facts and talking about them with many different
people" (id.: 285), revealing that the data were not collected in a systematic
way. Expressions illustrating the conduit metaphor are put into focus, alterna-
tives are overlooked. In this light, it is interesting to compare Reddy's results
with an example from Lakoff & Johnson (1980), who have looked in detail at
the ways speakers of English conceptualize arguments. Numerous examples
A SURVEY OF METALINGUISTIC METAPHORS 3
reveal the pervasiveness of THE ARGUMENT IS WAR metaphor, for example (id.:
4):
(7) Your claims are indefensible.
(8) He attacked every weak point in my argument.
(9) His criticisms were right on the target.
This is not the only metaphor in terms of which we frame arguments; alterna-
tives include AN ARGUMENT IS A JOURNEY (id.: 89 ff.) and AN ARGUMENT IS A
BUILDING (id.: 92 ff.). These metaphors bear little if any affinity at all to the
conduit metaphor. They show that, in general, we employ various, sometimes
mutually inconsistent, metaphors to conceptualize the same phenomenon and,
regarding linguistic communication, there exist alternatives to the conduit
metaphor, which focuses exclusively on the encoding-transfer-decoding process.
To conclude, partial or skewed sampling of data may distort the relative
importance attributed to certain metaphors.
More recently, metaphors of verbal communication have been studied by
Rudzka-Ostyn (1988). She has compiled a corpus of 700 examples from various
sources and narrowed it down to the 367 items that involve movement in space.
Within this subset, the primary focus is put on phrasal verbs. A detailed analysis
of these data reveals a large variety of metaphorical extensions from the domain
of spatial motion into that of verbal communication. This study differs in several
important respects from Reddy's. First, the data - although restricted from a
semantic point of view, too - have not been selected on an a priori basis.
Second, the analysis is much more fine-grained. The conduit metaphor operates
with notions such as 'object', 'container' and 'transfer', all of them situated at
the level of abstraction of Johnson's (1987) image schemata. Rudzka-Ostyn
goes into greater detail, adding richer images that have more content. Both
abstract image schemata and rich images play a role in figurative language (cf.
Lakoff 1987: 444 ff.). Finally, Rudzka-Ostyn is aware of her study's limits
("Many [...] areas remain unexplored." (id.: 551)) and counters Reddy's claim
that English is its own worst enemy as a metalanguage by stressing that the
metaphors she has uncovered offer "an extremely powerful means of expres-
sion" (ibid.), which leads to the conclusion that English is "its own best friend"
(id.: 552). (See also Goossens (1994) on the conduit metaphor.)
The study undertaken by Verschueren (1985b), based on a series of news-
paper articles covering the 1960 U2-incident, is restricted not only in scope but
also regarding its objectives (id.: 38). The limitations inherent in such a case
study allow a very detailed analysis of the role played by metalinguistic
4 JOHAN VANPARYS
metaphors in one genre. The conclusions drawn from it, however, constitute but
one piece of a huge jigsaw puzzle that still remains to be assembled.
Other clues can be found in research that is not specifically geared to-
wards verbal communication. Talmy's (1985) work on force dynamics,
elaborated by Johnson (1987: 57-61), shows how the force schema pervades
our thinking about interpersonal relations and communication. Folk beliefs
about emotions and their expression have been studied in detail by Kövecses
(1986) and Lakoff (1987: 380-416).
To conclude, existing research results on metalinguistic metaphors offer
only a partial, somewhat distorted, survey of the topic.
2. Objectives
The aim of this paper is to remedy this situation. An attempt will be made to
construct a representative list of conventional metalinguistic metaphors, which
will then be narrowed down to the set of verbials, that is verbs and longer units
built around a verb (eg phrasal verbs, idiomatic clauses, etc.; cf. Verschueren
1985a).
The general background is the seminal study of Lakoff & Johnson (1980),
which entails a number of commitments. The "essence of metaphor is under-
standing and experiencing one kind of thing in terms of another" (Lakoff &
Johnson 1980: 5). Thus, a speaker who says (1) understands, and consequently
codes, words in terms of containers. In a metaphorical expression, two
conceptual domains are involved: a donor domain and a recipient domain (or
source and target domain). In example (1), the referent of words - an entity in
the domain of verbal communication - is understood in terms of physical
objects, more specifically containers. Very often, the donor domain is more
concrete than the recipient domains, since we tend to understand abstract
phenomena in terms of more concrete things (Johnson 1987). The main
objective of this paper is to identify the range of donor domains serving as the
source for metalinguistic metaphors.
Not anything can be conceptualized in terms of anything else. In order for
a metaphor to be successful, there must be some perceived structural resem-
blance between the two domains. For example, the ARGUMENT IS WAR metaphor
relies on an identification of the speaker (S) with the aggressor, the hearer (H)
with the defendant, the words with weapons, the different stages in an argument
with the stages in a fight, etc. (Lakoff & Johnson 1980). For an adequate
understanding of a metaphorical utterance or a set of related metaphors, one
needs therefore a structural description of boh domains involved. The domain of
verbal communication has been described in great detail by Dirven et al. (1982),
Rudzka-Ostyn (1988) and Verschueren (1987b), among others. In this paper,
the ways in which elements from this recipient domain are receive mappings
from various donor domains will be explored.
Typically, a metaphorical mapping between two domains is instantiated by
a whole cluster of examples; the number of instantiations is to some extent
indicative of a metaphor's cognitive salience. Thus, the evidence adduced by
Lakoff & Johnson for the ARGUMENT IS WAR metaphor and the numerous
illustrations provided by Reddy for the conduit metaphor point to a high degree
of salience. Contrary to Reddy and Lakoff & Johnson, the material on which
this paper relies is corpus-based. Since the examples are taken from a corpus-
based general-usage dictionary, the analyst's creativity in making up metaphori-
cal expressions plays no role, which lends more plausibility to the pervasiveness
of the metaphors under investigation. '
3. The data
In principle, there are several ways of constructing a corpus. One might, for
instance, make up one's own examples. This is a useful procedure if the aim is
to verify/falsify a given theory, but the potential pitfalls are obvious: one risks
coming up with artificial examples, self-concocted items reveal nothing about
their relative salience, one cannot be sure that important metaphors have not
been overlooked, etc. Another possibility is to scan texts for metalinguistic
metaphors. This will yield a database with rich contextual information. In order
to arrive at a representative sample, however, one would have to go through a
huge amount of text representing different genres, since metalinguistic
metaphors do not have a high frequency. A text-based corpus, therefore, is only
efficient for case studies with a restricted scope (eg Verschueren (1985b)).
Authenticity and efficiency can be combined in a third option: a diction-
ary-based corpus. Indeed, every dictionary contains a wealth of material that is
relevant to the issue at hand. Moreover, the items found in a general-usage
dictionary constitute a representative sample. They cover all types of language
use and they are not restricted to one particular topic. In addition, conventional-
ized metaphors, when compared with novel constructions, are of special rele-
6 JOHAN VANPARYS
vance (cf. Traugott 1985). When looked at from a diachronic viewpoint, they
may be said to have been more successful. In order for a new lexical item or a
novel sense of an existing item to penetrate the lexicon, it must be used to a
considerable extent by a sufficient number of speakers. In order to stay there, it
must moreover stand the test of time. From a synchronic point of view,
conventionalized metaphors are part of a language's lexicon to the extent that
they have acquired unit status. They are used automatically and spontaneously,
without any constructive effort on the part of S. In sum, it is the dead meta-
phors that we really live by (cf. Mackenzie (1985: 69)), and many of these can
be found in a dictionary.
The corpus serving as the basis of this paper and several other contribu-
tions to this volume is taken from the Longman Dictionary of Contemporary
English (LDOCE). It was compiled in two stages. First, the contents of the
dictionary were narrowed down to entries that involve aspects of verbal
communication. Part of this work was done by computer. This database was
then split up into literal and figurative (metaphorical and metonymic) expres-
sions.
LDOCE was chosen for several reasons. It is a corpus-based general-
usage dictionary representing ordinary, present-day English. Though British
English is the main locus of attention, other varieties are taken into account as
well. The feature that renders LDOCE the ideal dictionary for our purposes is
its restricted defining vocabulary: lexical items are defined by means of a set of
about 2,000 defining words. We hypothesized that definitions of metalinguistic
items (literal and figurative) would include one or more metalinguistic defining
words themselves. Consider the following examples (my italics, J.V.):
accuse to charge (someone) with doing wrong or breaking the law
beg 1 to ask humbly for (food, mone, or other necessary things) 2 to ask
humbly (something not material)
exclaim (mostly used with the actual words of the speaker) to say sud-
denly, because of strong feeling
prophesy 1 to give (a warning, statement about some future event, etc.) as
a result of a religious experience directed by God or a god 2 to say in ad-
vance
shout to give a loud cry (of); speak or say very loudly
LDOCE defines accuse in terms of charge, beg is explained by means of ask,
the definition of exclaim includes terms such as words, speaker and say, the
paraphrase of prophesy includes warning and statement and shout, finally, is
defined on the basis of cry, speak and say. The hypothesis was tested on the
basis of the 240 speech act verbs listed in Wierzbicka (1987: 395-397) and
largely confirmed. Most of these verbs are defined in LDOCE by means of one
or more metalinguistic terms. Those that are not turn out to be only marginal
speech act verbs.
(i) Some do not necessarily involve verbal communication:
• calculate, conjecture, compare, conclude, credit, deduce, estimate,
guess, presume, prove, reckon, speculate, suppose, and suspect belong
more to the epistemic than the metalinguistic domain;
• agree, approve, blame, deplore, disapprove, object and resolve refer to
mental states and actions that can but need not be expressed overtly;
• blackmail, dissuade, ridicule and reassure denote actions that can be
brought about by non-linguistic means;
• counter profiles retaliative acts in general, not only those that are lin-
guistic.
(ii) Institutional declarations (acquit, appoint, book, convict, excommunicate,
reserve, resign, sentence) represent "a very special category of speech acts"
(Searle 1979: 18-19). They require a set of extra-linguistic, institutional
rules for their successful performance. Thus, they are situated on the area
of overlap between linguistic and institutional action. Moreover, they are
not essentially communicative, which sets them apart from central speech
acts (cf. Bach & Harnish 1979: 117). Farewell features only as an interjec-
tion in LDOCE, not as a verb.
(iii) The rest of the verbs listed by Wierzbicka but not defined in LDOCE by
means of metalinguistic items are:
bemoan to be very sorry because of
bet to risk (money) on the result of a future event
criticize 1 to find fault with (someone or something); judge severely
2 to make judgments about the good and bad points of (someone or
something)
enthuse to show enthusiasm
exult to rejoice; to show delight (in victory, success, etc.)
justify to give a good reason for
note to call attention to; make known; show
point out to draw attention to (something or someone)
The definitions given for bemoan and criticize are somewhat inaccurate.
They make reference only to one's having a certain feeling/opinion, whereas
other dictionaries indicate that these psychological states are also ex-
8 JOHAN VANPARYS
pressed. The paraphrases for bet, note and point out make no explicit refer-
ence to verbal communication. The element show in the definitions of en-
thuse and exult refers to the exteriorization of emotions, without however
putting the use of language into focus. Justify, finally, is defined in a meta-
phorical way by means of give.
These verbs reveal a disadvantage of LDOCE: the commitment to a
restricted defining vocabulary sometimes results in poor definitions. The
Collins COBUILD English Language Dictionary, for instance, makes ex-
plicit reference to verbal communication for seven out of the eight re-
maining verbs. The exception is bet, a notorious troublemaker in speech act
theory (cf. Fotion (1981)). All in all, however, it seems justified to rely on
the LDOCE definitions for drawing up a list of metalinguistic lexicaliza-
tions. In quantitative terms, the loss - 8 items out of 240 - is negligible.
Moreover, the loss is random, the absent items do not go together semanti-
cally, so that the representativeness remains unaffected. In our analyses, the
poverty of the LDOCE definitions has been compensated for on the basis of
other dictionaries.
An additional source of information is the example sentences given in LDOCE.
Consider the following items (abridged from LDOCE):
airy (derog) having little substance; empty Nothing results from his airy
plans and promises
hover (of people) to wait around one place (fig) A question hovered on his
lips
mingle to mix (different things) together a speech that contained praise
mingled with blame
plaster to put wet plaster on (fig) They plastered over the difficulties with
fine words
spice interest or excitement, esp. as added to something else a few good
stories to add spice to the speech
Though the definitions make no explicit reference to verbal communication, the
examples clearly show the item's relevance. The import of such evidence should
not be underestimated. The point of example sentences in dictionaries,
especially those aiming at a non-native public, is to illustrate typical usages. The
presence of metalinguistic metaphors in the example sentences, especially those
labelled 'fig' indicates therefore that they are to some extent conventionalized;
they are on their way to acquiring unit status. Moreover, LDOCE is to a great
extent corpus-based.
Working from the observation that most metalinguistic terms, whether

literal or metaphorical, are defined in terms of other metalinguistic terms and
that additional material can be found in the example sentences, we drew up a list
of relevant words from the defining vocabulary. This list was checked against
samples of the dictionary and revised where necessary. The resulting list consists
of the following items:
account, address, adjective, admission, admit, advice, advise, agree,

agreement, answer, argue, argument, ask, beg, blame, call, charge, claim,
command, complain, complaint, conversation, conversational, cry, curse,
declaration, declare, demand, describe, description, descriptive, enquire,
enquiry, explain, explanation, express, expression, greet, greeting, in-
form, information, informative, inquire, inquiry, instruct, instruction, in-
structive, interrupt, interruption, introduce, introduction, invitation, in-
vite, joke, language, lie, mention, message, name, news, note, noun, or-
der, persuade, persuasion, phrase, praise, pray, prayer, promise, pro-
nounce, pronunciation, quarrel, question, remark, remind, reply, report,
request, say, scold, sentence, shout, speak, speaker, speech, state, state-
ment, story, suggest, suggestion, swear, talk, tell, thank, urge, verb, voice,
warn, warning, word, write, writer
10 JOHAN VANPARYS
The content of LDOCE with regard to our purpose is illustrated diagrammati-

cally in Figure 1, together with the decision algorithm that was applied. A
computer program was run on LDOCE to retrieve all items that included one of
the defining words in their definition or the example sentences. On this basis, the
dictionary's content could be reduced to a database of manageable size on
which further work could proceed manually. Manual filtering was required for
several reasons. Several defining words are polysemous or homonymous, with
senses unrelated to verbal communication (eg charge, order, lie, state), thus
yielding redundant items. Moreover, not every dictionary entry containing a
metalinguistic term is itself metalinguistic. Though tell and state figure in the
information of clobber and tenable, the latter items are clearly not metalinguis-
tic:
clobber to strike severely and repeatedly I'll clobber you if you don't do
what you're told
tenable (of an office, position, etc.) that can be held by somebody (for a sta-
ted period of time)
So, entries unrelated to verbal communication had to be filtered out. Then the
final split was made between literal and figurative entries. Obviously, the
difference between metalinguistic and other terms on the one hand and between
literal and figurative items on the other is by no means clear-cut. Borderline
cases were discussed in group, in order to work out selection criteria. Given the
size of the material, we set ourselves the task of eliminating as much as seemed
justified. We preferred to have a manageable database with clear cases rather
than an exhaustive collection replete with dubious instances. With regard to the
first step in the decision algorithm (± verbal communication), we stipulated that
items must necessarily or typically involve verbal communication, thus
eliminating those that do so only occasionally. The following examples may
serve to illustrate this principle:
abrupt (of behaviour, speech, character, etc.) rough and impolite
amiss out of place; (a) bad idea A few pleasant remarks may not be amiss
asinine foolish; stupid what an asinine remark
blare out (esp. of a machine) to produce (sounds or words) loudly and un-
pleasantly The radio blared out the news
floor to confuse The news floored me
watch to take care of, be careful with, or pay attention to [.. J Watch your
health as you grow older. You have to watch your words when you talk to
the general
The adjective abrupt was eliminated because of its applicability to human

behaviour in general, of which verbal communication is only one aspect. The
definition of blare out includes the metalinguistic item words, but it also reveals
that the verb pertains principally to the production of sounds by a non-human
source. Amiss, asinine, floor and watch owe their presence in the computer
output to the presence of a metalinguistic marker in the example sentences
rather than the definition itself. In each case it is clear that verbal communica-
tion is not the preferred domain evoked by the items: not only remarks or other
speech acts can be amiss or asinine, one can be floored by other things than
news or other information and one's words are but one of the numerous things
one can watch.
Banner illustrates another elimination principle. Verbs of speaking (here,
said) often take a non-human information conveyor for subject. Though this is a
major metaphor (personification) in the field of verbal communication, the entry
banner was not included in the database because the metaphor in question bears
no specific relevance to the noun banner. In other words, banner as such does
not specifically evoke a metaphor of verbal communication; here it does so only
by coincidence, through the example sentence LDOCE happens to provide:
banner a long piece of cloth on which a sign is painted, usu. carried be-
tween 2 poles The marchers' banners all said: "We want work"
Regarding the second step in the decision algorithm (± figurative), the require-
ment that the item must still be felt to be a figurative expression on a synchronic
basis served as a major elimination criterion. This principle applies primarily to
words of Latinate origin, such as accede, acquiesce, asseverate, effusion,
express, or prevaricate, whose non-communicative orgin is no longer transpar-
ent for ordinary speakers of present-day English.
As in the first step of the decision process, metaphors that surface only
'by coincidence' were excluded. The metaphoricity in the examples below is
triggered by full and make, it has no specific bearing on the dictionary entries
ambiguity and arch:
ambiguity something that is ambiguous His reply was full of ambiguities
arch supercilious When he made that silly remark, she gave him an arch
look
The resulting corpus contains 1916 entries, consisting of the LDOCE headword,
its definition and in most cases one or more example sentences. It includes not
only one-word items (mostly verbs, nouns and adjectives) but also longer
expressions, such as phrasal verbs and idioms:
12 JOHAN VANPARYS
barrage (of speech or writing) a large number of things put forward at al-
most the same time or very quickly aftert he other a barrage of questions
blurt out to say (something which should not be said) suddenly and with-
out thinking Peter blurted out the news before he considered its effect
breath: waste one's - to talk uselessly, without effect
The corpus takes the shape of an alphabetical list of dictionary entries. In order
to improve semantic accessibility, it was indexed on key-words (metalinguistic
and other) occurring in the definitions. In this way, lists were generated by
computer of items defined in terms of the same defining word. Thus the defining
word secret yielded semantically related lists such as the following, which reveal
the metaphor TO MAKE KNOWN IS TO MAKE VISIBLE:
leak to make known (news, facts, etc. that ought to be secret)
leak out (of news, facts etc. that ought to be secret) to become known
let the cat out of the bag to tell a secret (often unintentionally)
spill to let out or tell (secret information)
spill the beans to tell a secret too soon or to the wrong person
Describe yields a cluster illustrating the same kind of extension from the visual
to the communicative realm, containing chalk out, delineate, depict, picture,
portray, and sketch. Under the defining word interrupt, several verb+in(to)
combinations are listed (barge into, break in (on), break into, burst in on, butt
in, chime in, cut in, get in, horn in, push in, put in) revealing the metaphor A
CONVERSATION IS A CONTAINER. These lists proved to be useful instruments for
bringing together semantically related items that were scattered across the
alphabetically arranged corpus.
4. The conduit metaphor revisited
Reification is no doubt one of the major metaphorical patterns, as it relates

abstract conceptions to concrete experiences. Language is largely an abstract
phenomenon; the only concrete aspect to it being the soundwaves produced in
speaking (or the characters in writing). Their transient, non-tangible nature,
moreover, renders them atypical with respect to concreteness. The conduit
metaphor is therefore a natural one (cf. Goossens 1994). It relates to the
process of information transfer, in which a speaker who entertains a conceptu-
alization tries to find or constructs an appropriate linguistic expression
(encoding) and transfers this coded conceptual content towards H by means of
soundwaves. H's task, then, consists in decoding, that is reconstructing S's
conceptualization on the basis of the acoustic input. The conduit metaphor

consists in a reification of this process. Words and ideas are equated with
objects: the former with containers and the latter with objects of undefined
nature that fit into the container-words. Encoding consists in putting ideas into
words, which are then passed towards H, whose task it is to take the words out
of the containers. In sum, the conduit metaphor can be resolved into several,
interrelated metaphorical equations:
(a) IDEAS ARE OBJECTS.
(b) LINGUISTIC EXPRESSIONS (WORDS, SENTENCES, TEXTS, ...) ARE CON-
TAINERS.
(c) ENCODING CONSISTS IN PUTTING IDEA-OBJECTS INTO WORD-CONTAI-
NERS.
(d) SPEAKING CONSISTS IN TRANSFERRING WORD-CONTAINERS FILLED WITH
IDEA-OBJECTS TO H.
(e) HEARING INFORMATION CONSISTS IN RECEIVING WORD-CONTAINERS
STUFFED WITH IDEA-OBJECTS.
(f) DECODING CONSISTS IN TAKING THE IDEA-OBJECTS OUT OF THE WORD-
CONTAINERS.
Since the data of this study consists of verbials, the focus will be on (c)-(f), for
verbs typically denote actions. However, (a) and (b) are presupposed by (c)-(f).
4.1 Linguistic expressions as containers
The corpus contains several items supporting (a)-(c):

(10) ... an important letter containing further details of the contract ...
(COBUILD)
(11) They cannot put their feelings into words. (COBUILD)
(12) His anger at last found expression in loud cursing. (LDOCE)
(13) Words embody thought. (LDOCE)
(14) This book embraces many different subjects. (LDOCE)
(15) He had learnt how to wrap up bad news. (COBUILD)
(16) The offer was simply an old one dressed up in new terms.
(COBUILD)
(17) The government's refusal was couched in unfriendly language.
(LDOCE)
(18) Ker ideas were caparisoned in beautiful poetry. (LDOCE)
(19) In his book he tried to capture the beauty of Venice. (LDOCE)
14 JOHAN VANPARYS
The container metaphor is evoked in a rather neutral fashion by contain (10)

and verb + in combinations (11, 12), while other verbs add richer information.
The human body - a container par excellence (Johnson 1987) - is evoked by
embody (13) and embrace (14). Wrap up (15), dress up (16) and couch (17)
highlight the disguising or ornamental function of the container, thus creating a
tension between words and content, resulting in a pejorative connotation. No
such tension is present in (18), where the container matches the nature of the
content rather than concealing it. Capture (19) requires control, giving rise to a
positive connotation: the author really masters his subject. The conduit
metaphor alone does not suffice to understand these instances: other notions
apart from containment have to be invoked for a full appreciation of their
complexity.
4.2 Linguistic expressions as objects
Very often, linguistic expressions are reified, not as containers for ideas, but in
some different fashion. As objects, linguistic expressions can undergo all sorts
of manipulations:
4.2.1 Verbs of making

A noun denoting a linguistic expression frequently functions as the object of
make. Thus one can make a promise, an answer, a point, a suggestion, etc.
Similarly, a sentence can be constructed. Contrary to the container metaphor,
where S's only task is to put idea-objects into pre-existing containers, make,
construct and the like highlight the constructive effort required of the encoder.
There is often an element of insincerity, as in cook up (20), fabricate, manufac-
ture, spin and weave (eg a story or evidence) or concoct (eg an excuse, an
explanation). In these cases, the presupposition is that true information exists
first, while lies need to be produced before being expressed. With grind out and
thrash out (eg stories), finally, S is dehumanized, acting like a machine
continually at work.
(20) Jean cooked up a story to explain why she was late for work but
her employer did not believe her. (LDOCE)
4.2.2 Verbs of adorning

As objects, linguistic expressions or their content can be manipulated in various
ways. For instance, they can be embellished or serve to adorn their content (cf.
4.1):
(21) I asked him to tell the simple truth and not to embellish it with
ideas of his own. (LDOCE)
(22) He adorned his story with all sorts of adventures that never hap-
pened. (LDOCE)
One way of changing an object's outer appearance is to change its colour. Thus,
a claim can be rendered more credible by adding colour to it:
(23) give a false colour to (= 'give a false account of) (LDOCE)
Similarly, a story can be caparisoned or embroidered to make it more attractive
and one's language needs polishing if it is not mastered properly:
(24) Her ideas were caparisoned in beautiful poetry. (LDOCE)
(25) This mystery of the sea has been retold and embroidered.
(COBUILD)
(26) My German was not very polished. (COBUILD)
If the object is food, its attractiveness can be enhanced by larding or spicing it:
(27) Why do you lard your conversation with bits of slang?
(COBUILD)
(28) His speech was spiced with anti-imperialist sentiments.
(COBUILD)
4.2.3 Verbs of expanding and compressing

Another way of changing objects is to make them larger or smaller by expand-
ing or compressing, or by adding or removing parts. Thus ideas can be
amplified, a report can be condensed, compressed, pruned or spun out, it can be
confined to a certain topic and something can be tacked or tagged on to it or
clipped from it. A subject can be dilated, enlarged or expanded on. Each
metaphor profiles the change in size in a different manner, determined by its
meaning in the donor domain. For instance, if a long report is con-
densed/compressed or spun out, the substance-content does not alter; it is only
presented in a shorter, respectively longer form. Pruning and clipping on the
other hand, involve the removal of superfluous parts, while tacking or tagging
something on consists in the addition of something inessential. Here, the content
does undergo a change.
16 JOHAN VANPARYS
4.2.4 Other verbs of changing

In addition to being adorned or undergoing changes in size, linguistic expres-
sions as objects can be modified in a variety of ways: a report can be angled, it
can be distorted, mutilated, perverted or twisted. Its effect can be diluted or
watered down. In these cases, the change in form entails a change in meaning,
evaluated negatively as in the donor domain. If someone is said to turn a
phrase, on the other hand, he is praised for his original formulation. Here the
change in shape is a change for the better.
4.2.5 Visual perception

When an abstract concept is reified, one of the effects is that it may be visible or
be concealed. With regard to the message of a speech event, this gives rise to
the metaphorical concepts TO MAKE KNOWN IS TO MAKE VISIBLE and TO KEEP
UNKNOWN Is TO MAKE/KEEP INVISIBLE. For instance, a secret can be disclosed
or covered up.
The verbials bear out, disclose, expose, highlight, reflect and confront s.o.
with (eg the facts) represent different variations on the theme TO MAKE KNOWN
IS TO MAKE VISIBLE. So do verbs of drawing or painting: chalk out, chart,
delineate, depict, illustrate, paint a picture, picture, portray, sketch (out),
outline.
Though richer in content, the idioms spill the beans, let the cat out of the
bag and lay/put one's cards on the table hinge on the same equation (cf Lakoff
1987). Invisibility is conveyed by sweep s.th. under the carpet and draw a
curtain over s.th.
Leak (out) equates the message (usually a secret) with a liquid, passed on
through some kind of pipeline to those who are entitled to being informed. A
leak makes the information accessible to unauthorized recipients. The expres-
sion broach a/the subject, too, represents the content as a liquid. In this case it
is contained in a bottle or a barrel. When the container is opened, the informa-
tion becomes accessible.
In break the news, on the other hand, the message is portrayed as a solid
object. When it is broken, its interior becomes visible. The concept 'access to
the inside' also motivates probe and pry into. The interior is rendered less
accessible by cloak ('to hide'), plaster (with euphemisms) and wrap up ('to say
indirectly').
Impaired visibility, as instantiated by fog and throw dust in s.o. 's eyes, re-
sults in a lack of understanding or misunderstanding. When visibility is restored,
things becomes 'clear': clear up (eg a mystery), clear the air.
In sum, a whole range of verbials present linguistic expressions as objects
that can be produced (4.2.1), adorned (4.2.2), expanded or compressed (4.2.3),
altered in some other way (4.2.4) or perceived (4.2.5). Unlike the conduit
metaphor, these reifications do not specifically involve the container schema
posited by Reddy. These cases indicate that the conduit model is not the only
framework in terms of which we conceptualize (or even reify) linguistic
expressions.
4.3 Transfer of information
As specified by the conduit metaphor, speakers communicate by physically

transferring word-objects to their addressees. This metaphor takes on many
different shapes:
4.3.1 Verbs of giving

The corpus includes numerous verbs of giving, such as deliver (eg a lecture, a
speech, etc.), furnish (eg an answer), give, give away (eg a secret), give the
game/show away, give an order/permission etc., give s.o. to believe/understand
etc., give/send s.o. one's love/respect, grant, have to hand it to someone, offer
(eg a prayer), pass (eg a comment, a remark), peddle (eg gossip), ply s.o. with
(eg questions), present, return:
(29) Rudolph gave her Julie's number. (COBUILD)
(30) It was her policy with her never to pass any comment when she
was in their company. (LDOCE)
(31 ) Today I want to present the student view. (LDOCE)
A prototypical scenario of giving might be characterized as follows:
Two persons are involved, a donor and a recipient. The donor owns an
object, which he holds in his hands. He wants the other to have the object
and assumes a similar desire on the other's part. He passes the object into
the recipient's hands, who accepts it and thereby becomes the new owner.
Note that this scenario, like any other prototype-based characterization, is quite
flexible. More than two persons may be involved, there need not be exactly one
donor and one recipient. The donor need not possess the object and giving need
18 JOHAN VANPARYS
not involve transferral of ownership. Moreover, the object need not pass from
hands to hands. However, the elements of the prototype and their interrelations
correspond quite closely with those involved in a typical speech-act scenario:
the two participants (speaker/donor, hearer/recipient, S entertains a conceptu-
alization (owns an object) and assumes a desire on the part of H to become
informed of it.
An example like (29) is fairly neutral with respect to these parameters of
variation. Here, give simply profiles a donor (subject), an object (direct object)
and a recipient (indirect object). Ownership does not seem to be involved, the
exact location of the object (in the hands or elsewhere) is irrelevant. However,
the principal participants are there: speaker/donor, the hearer/recipient, and the
idea/object. In addition there is willingness on the part of S to give the
information (compare he gave her the number with he let slip the number) and
a desire on the part of H to receive it.
(32) No one in the class could furnish the right answer to the question.
(LDOCE)
(33) ... the unreliable gossip that had been peddled for so long ...
(COBUILD)
Other verbials highlight specific elements not predicted by the conduit meta-
phor. Deliver (a lecture/speech) and furnish (eg an answer) place the act of
giving in a commercial environment, where the donor satisfies a demand. In
(32), this element is transferred into the domain of verbal communication in a
classroom setting, where the teacher indicates a desire to be told something.
Peddle (33), on the other hand, derives its pejorative connotation from the
negative value judgments our culture stereotypically places on peddlers.
(34) I didn't feel like giving away more information than I had to.
(COBUILD)
(35) The examiners have given away the answer. (LDOCE)
(36) to give the game/show away (= 'to tell a secret5)
Verbs of giving combined with away focus on a different aspect. Away profiles
movement of an entity from its initial location into another, unspecified region.
In literal giving away, this is carried over into a loss of control over the object in
question. When used as a metaphor of verbal communication (34-36), lack of
intentionality is highlighted.
Grant (37) and offer (38), on the other hand, are motivated by a desire on
the part of the recipient/hearer to receive/be told something:
(37) I had to grant him the reasonableness of his argument. (LDOCE)
(38) May I offer my congratulations? (COBUILD)

Return profiles movement back to the mover's initial location. It is therefore a
suitable verb for talking metaphorically about replies:
(39) "Yes, if you like!" she returned smiling. (= 'answer') (LDOCE)
(40) He returned his earnings as £3,000 on the tax declaration.
{LDOCE)
(41) return a/the compliment
4.3.2 Verbs of bringing

Both prototypical giving and speaking require spatial proximity of the two
participants. This condition motivates the use of verbs of bringing as metalin-
guistic metaphors; bring (forward), bring a charge against, convey, drag in.
Bringing differs from giving in the fact that it involves a relocation of the donor
towards the recipient. Thus we get:
(42) The neighbours brought a complaint against the fast driver.
{LDOCE)
(43) Can the prisoner bring forward any proof of his story? {LDOCE)
(44) A plan was brought forward to allow workers to share in the prof-
its. {LDOCE)
(45) They disapproved of dragging in his wealth. {COBUILD)
Here bring (42) is the most neutral item, whereas the other verbs focus on
specific aspects of the act of bringing. Bring forward (43, 44) is typically used
when there is an audience rather than a single addressee, a situation requiring S
to step forward in order to become visible and/or audible. Drag in, on the other
hand, makes reference to a heavy object that resists transportation. In (45), this
resistance is motivated by the audience's disapproval of the topic.
4.3.3 Lay, put, drop and related verbs

Prototypical giving involves an object placed in the hands of the recipient.
Several verbials, however, profile a different kind of movement of the object:
drop (eg a hint), drop a brick/clanger, hand down, lay (down), lay claim to, lay
down the law, let drop/fall (eg a suggestion), lodge (eg a complaint), name-
drop, place an order, put (eg a question), put in, put it (eg mildly), put it to s.o.
that, not to put too fine a point, table.
Lay, as well as lodge, typically appears in official contexts (lay claim to,
lay a charge against), or in situations where S occupies an authoritative
20 JOHAN VANPARYS
position {lay down conditions, lay down the law). These examples are subject to
two different, though not mutually exclusive, explanations. For one, they may
involve the placement of documents on a table, for instance by a person filing an
official complaint. This is the preferred interpretation for table (eg a report). On
this account, the metaphors have a metonymic origin. In a more abstract
interpretation, the object is of unspecified nature, undergoing downward move-
ment instigated by S, ending within the immediate proximity of H and resting
there. The downward movement is then explicable in terms of the superordinate
position of S vis-à-vis H (equally in hand down). A difference between laying
down something and giving it to a person resides in the terminal stage of the
process. In a prototypical act of giving, the recipient accepts the object. That is,
the object ends up in the recipient's hands. When I lay it down (in front of him),
however, it may take some time before he takes it up - or he may refuse to
accept it at all. Indeed, official claims, charges etc. are not dealt with on the spot
and may be subject to an acceptance procedure.
(46) put a question
(47) Let me put to you a practical problem. {LDOCE)
(48) The organization put forward eight candidates for the NUS execu-
tive. {COBUILD)
(49) They rejected every proposal put forward. {COBUILD)
(50) I put in a request for an interview {COBUILD)
(51) not to put too fine a point
(52) put it briefly/simply/mildly ...
Put is frequently used in connection with verbal communication (cf. Pau-
wels, this volume). It shares with lay the fact that in the terminal stage the
object rests at a certain location. Examples (46)-(50) all involve speech acts that
require further action on the part of H, in the form of providing an answer (46),
working out a solution (47), considering a proposal (49) or a request (50). The
verb put profiles these speech acts as 'waiting' to be acted on. Regarding put
forward, a similar interpretation as for bring forward can be given: in bringing
the object/message to the fore, S makes it readily accessible.
4.3.4 Across, home, through

Verbs of movement frequently combine with one of these three particles. If
one's point comes or gets home, or if one manages to drive or press it home,
then one has succeeded in communicating effectively. The particle home
indicates that the object/message has reached its destination/hearer. If one's
ideas come or get across, or if one manages to get or put them across or get
them through, or if one's speech goes over well, the object/message equally
reaches its destination, but this time after overcoming some obstacle (across,
over) or following some predefined trajectory (through). In other words, across,
over and through indicate successful communication involving some effort on
the part of S, owing to the presence of resistance.
4.3.5 Violent transfer

Several verbials combine transfer with aggression or violence on the part of S:
aim/fling/hurl/throw (a remark at s.o.), barrage/besiege/bombard/pelt (s.o. with
questions), level (a charge at), throw out. Similarly, evade and fend off (eg a
question), indicate that the hearer/victim tries to avoid the objects/words while
parry denotes a defensive act:
(53) Instead of answering he parries with another question.
(COBUILD)
4.3.6 Verbs of retracting

If to speak is to give words, then it is possible to make a speech act undone by
taking back the objects/words: draw back/retract/take back/withdraw (a
promise, a remark, etc.).
4.3.7 Verbs of keeping

Numerous verbials related to keep appear as metaphorical verba tacendi,
involving unwillingness on the part of S to release information: hold s.th. back,
hold out on s.o., keep a secret, keep s.th. back, keep s.th. to oneself, withhold
(information). Similarly, in (54) a promise is withheld, thus serving as bait:
(54) He dangled a trip to Paris in front of her to get her to work for
him. (LDOCE)
4.3.8 Verbs of taking

While the verbials above indicate movement of objects/words from speaker to H
and take S as subject, verbials of taking focus on H's side. The most neutral
verbs here are get and take, indicating comprehension and belief respectively:
(55) I didn't really get the point of the story. (LDOCE)
(56) Don't get me wrong - I'm not complaining. (LDOCE)
(57) take it from me / take my word
22 JOHAN VANPARYS
Similarly, to take s. th. for granted denotes uncritical acceptance/belief. Various

other verbs imply some effort or difficulty on the part of H:
(58) I don't think I caught the name. (COBUILD)
(59) Do you mind repeating what you said, I didn't quite catch or?
(LDOCE)
(60) I didn't gather much from the confused story he told me.
(LDOCE)
(61) I grasped the main points of the speech. (LDOCE)
Certain items denote a mere attempt on the part of H to receive the ob-
jects/words: angle/fish for (compliments, information), hold out for (=
demand).
4.3.9 Verbs of exchanging

In conversations, the two participants assume the roles of S and H in turns. As a
result, we find not only verbials of giving and other verbials of transfer where
the subject denotes S, but also verbials of exchanging, such as bandy/exchange
words with, where the subject instantiates both S and H roles (cf. Pauwels, this
volume).
4.3.10 Verbs of disseminating

Not every speech act is directed at a particular H; people often speak to a large
group or 'to whom it may concern'. This situation is implicitly present in the
minor framework identified by Reddy, in which ideas are ejected by speakers
into external space, where they exist in reified form:
(62) Startling rumours began to get about/round in next to no time.
(COBUILD)
(63) A rumour was put about/around to the effect that he had been
drunk. (COBUILD)
The particles about and (a)round specify an entity's motion as not being goal-
oriented. The absence of a clear destination stands for the absence of a specific
addressee: reified ideas float around in the air, from person to person. The same
idea is present in broadcast and disseminate:
(64) Don't broadcast it, but I'm looking for another job. (COBUILD)
(65) The agency would collect and disseminate information.
(COBUILD)
4.4 Self-propelled motion
According to the conduit metaphor, in both versions, it is S who transfers word-

objects to H or ejects them into the air. A great many verbials, however, make
no reference to S's initiative, as if words lived a life of their own. Examples
include various combinations with come (come before / on / up / home), get (get
about / across / home / round), go (go, go the rounds, go down / over) and run.
Self-evident ideas even go without saying! More specific verbials add richer
informative content. For instance, several metaphors - typically verbials
indicating movement of large quantities of liquids - highlight the overwhelming
effect of large quantities of speech acts:
(66) They were deluged with requests to play the song. (COBUILD)
(67) Hospitals were being inundated with requests for help.
(COBUILD)
(68) Requests for information flooded in after the advertisement.
(LDOCE)
(69) snowed under with invitations to parties (LDOCE)
Other verbials highlight specific ways in which words-liquids move. For
instance, flow (70-71) indicates smoothness, perhaps even elegance, while filter
(72) involves reduction in contents slowness:
(70) As the ladies drank their tea, conversation began to flow.
(LDOCE)
(71) Her words flowed on, in perfect sentences. (COBUILD)
(72) Disturbing rumours filtered back from the East. (COBUILD)
In other expressions, words fly through the air (73) - or they fall down (74). In
the latter case, they fail to reach their destination; that is, they fail to produce
the desired effect.
(73) Rumours were flying in official circles. (COBUILD)
(74) His jokes fell flat and amused nobody. (LDOCE)
4.5 Target-directed transfer
In the conduit metaphor, as formulated by Reddy, words are transferred by S

towards H. However, several items reveal that linguistic expressions can take a
different route, namely towards a target: lay/put the blame on ..., lay s.th. at
s.o. 's door, fix the blame/crime on s.o., heap praise/criticism (up)on ..., lay (eg
a charge or responsibility) (up)on s.o., give the lie to s.o., pour scorn on ... In
24 JOHAN VANPARYS
these cases, the destination of the word-objects does not correspond with H, but
with the person positively or negatively affected by the speech act. Several items
evoke more than mere patient-directed transfer. For instance, when one lays or
puts the blame on someone, the speech act places a burden on him, something
that is fixed may be hard to detach, while heap and pour imply abundance.
In sum (4.3-4.5), verbials indicating information transfer suggest various
qualifications of the conduit metaphor. First, the notion 'conduit' itself is
virtually absent from the database: despite the large number of expressions
denoting transfer of information, this process is rarely presented as taking place
through the kind of pipeline suggested by Reddy. (See also Goossens 1994).)
Second, though the verbials in question do introduce ideas as objects
(reification), most of them can be understood without making reference to
words-as-containers. Third, in most cases the image evoked by an expression is
much richer in content than suggested by Reddy's account, so that other, more
specific conceptual elements need to be drawn into the analysis.
4.6 S and H as containers
Not only linguistic expressions, but also their source (S) and target (H) are
often presented as containers. Thus, the corpus includes numerous combinations
with out: bleat out, blubber out, blurt out, bluster out, boom out, dish out, give
out, grind out, hand out, ladle out, let out, pour out, put out, roll out, spill out,
spout out, thrash out, throw out. This particle designates movement of an entity
away from its initial location, where it was enclosed in a container. In the
examples, S is therefore conceptualized as a container where information is
stored in reified form. Speaking, then, consists in releasing the idea-objects.
(75) The date of the election will be given out soon. {LDOCE)
(76) He put out a statement denouncing the commission's conclusions.
(COBUILD)
(77) throw out (= 'to say s.th. carelessly and without considering the re-
sult')
(78) He likes dishing out good advice. {LDOCE)
(79) the knowledge that is ladled out daily in high schools {COBUILD)
(80) I don't need you handing out that sort of advice. {LDOCE)
(81) You were always spouting some theory to us at the table.
{LDOCE)
Give out (75) and put out (76) refer to this process in a neutral fashion, while
the other verbials add extra dimensions. The majority focuses on the acoustic
aspect of speech: bleat out, blubber out, blurt out, bluster out, boom out, roll
out. Let out (eg a cry of pain) and spill out (eg a secret) highlight the uninten-
tional nature of the speech act. Throw out (77) evokes the negative result of
speaking carelessly, motivated by the destructive action of projectiles. Dish out
(78), ladle out (79) and hand out (80), on the other hand, focus on the authori-
tative position assumed by the speaker-distributor and the abundance of his
output. As already explained above, grind out and thrash out evoke monoto-
nous abundance, by evoking S as a machine. The same effect is produced by the
fountain-metaphor spout (81).
The speaker-as-container metaphor is instantiated not only by verb + out
combinations, but also by other verbs requiring a container-like subject (82) or
object (83):
(82) John emitted a few curses. (LDOCE)
(83) A whistle of surprise escaped him. (LDOCE)
A different realization of the same metaphor can be found in verbials profiling a
person trying to extract an object from a speaker-container. Most of them
indicate unwillingness on the part of S to release information and a concomitant
effort on the part of his interlocutor:
(84) The truth had to be dragged out of him. (COBUILD)
(85) It was impossible to draw the truth out of him. (COBUILD)
(86) Sir James had extracted from Francis a fairly detailed account.
(COBUILD)
(87) She pretended that she couldn't hear, so that they would not force
an answer from her. (LDOCE)
(88) I managed to pry out of the adults the reason for her disappear-
ance. (COBUILD)
(89) I pumped him discreetly about his past. (COBUILD)
(90) The truth had been wormed out of him by his lawyers.
(COBUILD)
A speaker-container is also evoked by verbials of opening and closing,
designating unimpeded speech and termination respectively:
(91) She found it difficult to open out to people. (COBUILD)
(92) He was impressive on paper, but he completely clammed up in the
interview. (COBUILD)
(93) Shut up.
H is metaphorized as a container by verb + in(to) combinations and other
verbials requiring a container-like object:
26 JOHAN VANPARYS
(94) I'll fill you in / up on the details now. (COBUILD)

(95) I had it dinned into me at school. {COBUILD)
(96) Somehow we have to drill into them the importance of saving en-
ergy.
(97) drive into (= 'to teach (s.th.) with effort to (s.o.)')
(98) These facts had been drummed into him.
(99) 'How old are you, Mr Sharpe?' - 'All right, no need to rub it in.'
{COBUILD)
(100) The information is fed back to the government department con-
cerned. {LDOCE)
Fill in/up (94) instantiates the hearer-as-container metaphor in a neutral way.
Din (95), drill (96), drive (97), drum (98) and rub (99) point to some effort on
the part of S, presupposing resistance by H. Din, drill, drive, drum and feed
(100) indicate, in addition, repetitiousness.
4.7 Motion on the part of S
Lakoff & Johnson (1980) have identified the CONVERSATION-AS-A-JOURNEY

METAPHOR, which is instantiated by numerous verbials in the corpus:
4.7.1 Motion towards H

One precondition for successful oral communication is that S and H should be
within hearing distance of each other. A change of location may be required to
fulfill this condition. In other words, a communicative act may involve motion
of S towards H. For instance, S may approach H in order to be able to establish
communication. This element provides the metonymic basis for those uses of
approach that denote a communicative act:
(101) Did he approach you about lending him some money? {LDOCE)
In (101), approach is used as a verb of speaking, which is also indicated by the
presence of about to introduce the topic. The metonym may give rise to a
metaphor when no motion is involved at all. In this case, approach simply
means speak to.
Other verbials involving motion of S towards H include make advances
to, certain combinations with come (come again, come back, come forward,
come round, come out against/for/with, come up with) and go (go to arbitra-
tion, go to the people), as well as some of the verbials of bringing discussed in
4.3.2, such as bring, bring forward and bring a charge against. Despite their
semantic divergence in the recipient domain: these expressions have a common

element in the donor domain, motion of S towards H. With to, a specific
destination/hearer is indicated. In other cases, the intended audience may be
vaguer, as in the combinations with forward and out, which express a direction
and an origin, respectively.
The objective of goal-oriented motion is to reach a destination. In the case
of motion towards H, this element translates into physical contact: to contact
someone, to make contact with someone, to get in touch with someone.
Although these expressions do not necessarily indicate motion, the underlying
metaphorical basis is the same: establishing contact qualifies as the intended
endpoint of S's motion towards H.
4.7.2 Motion towards the topic

In the conversation-as-a-journey metaphor, not only H but also the topic may
feature as S's destination or the entity to establish contact with. Indeed, several
verbials indicate motion of S towards the topic: come back, come/get to the
point, get down to brass tacks, get onto/on to, go into detail, go straight to the
point, jump, meander, move on, return to, revert to. Some expressions bring
additional elements to the topic-as-destination concept. Verbials of returning
(come back, return to, revert to) presuppose that a topic has been left in favour
of another and indicates that it becomes the current topic again. Move on and
meander evoke more than one topic, as well. In the donor domain of motion in
space, move on refers to a change of location, which translates into a change of
topic in the recipient domain of verbal communication. Meander evokes a richer
image, that of a river flowing slowly from one point to another as if it had to
clear destination. When applied to verbal communication, the verb indicates an
excessive rate of topic switches, that is speech that does not move on fast
enough to its aim. According to the same metaphor, a digression is conceptual-
ized as movement away from the topic, as with depart from. (See also Simon-
Vandenbergen, this volume.)
The purpose is to establish contact with the topic, as revealed by
be/go/keep on about, be/go/keep/get on at, come/get to grips with, dwell on, get
onto/on to, touch (on). When S A fails to establish contact, on the other hand,
the topic becomes an obstacle and S is said to pass/slide/slur over it.
28 JOHAN VANPARYS
4.7.3 Towards the goal of the conversation

If the conversation is goal-oriented, its purpose may serve as the destination: to
arrive at (eg a decision), to come to terms. A lack of goal-orientedness is
implied by ramble.
4.7.4 Manner of movement

The motion metaphors discussed above concentrate on the destination (H, the
topic or the goal). Other verbials focus on the manner of movement: speed
(change/shift gear, dash off (eg an essay), maunder, rattle through (eg a
speech)), clumsiness or lack of control {falter, flounder, put a foot wrong, put
one's foot in it, stumble, trip over), skill {take a hurdle), secretivenss {sneak
on), monotony {trout out) or pride {swagger).
4.8 The conversation as a container
Since virtually any activity can be conceptualized as a container (Johnson 1987),

the CONVERSATION IS A CONTAINER metaphor is a natural one. Here we find
mainly combinations of verbs with in(to), such as fall into (a conversation), get
in (a conversation), horn in, plunge into, step in, weigh in. Depending on the
verb, additional aspects of the speech activity are highlighted. Someone who
intervenes in an ongoing conversation enters the conversation-container. This is
often experienced as rude (barge into, break in on/into, burst in on, butt in,
chip in, cut in, horn in, push in). Alternatively, joining in a conversation may be
conceptualized as bringing word-objects into it (not get a word in edgeways, put
in). The introduction of a new topic may be conceptualized as bringing it into
the conversation/container (drag in, introduce into).
4.9 Bodyparts used in speaking and listening
4.9.1 Respiration
To speak is to modify one's breath in such a way that one produces conven-
tionalized sounds. In other words, respiration is part of the speaking process.
This gives rise to a number of metonymic expressions involving respiration or
bodyparts used in speech production (the mouth, lips and tongue). Given the
metonymic relationship between breathing and producing speech sounds,
verbials indicating manner of respiration can be used to indicate manner of
speaking. The most neutral verb here is breathe, in the sense of 'to whisper'.
Other verbs add extra dimensions, following the pattern 'to V = to say by/as if
V-ing' (to gasp (out), to pant, to puff, to snivel, to snort, to wheeze). The notion
of breath as a limited resource is highlighted by save/waste one's breath. The
expression take one's breath awayrefers to one's inability to speak.
4.9.2 The mouth

The mouth (with the lips and the tongue) is not only a crucial but also a highly
visible speech organ. Figurative expressions involving the mouth are therefore
quite abundant (cf. Goossens, this volume). In open/shut one's mouth and stop
one's mouth/lips, the metonym TO SPEAK IS TO USE ONE'S MOUTH (TONGUE,
LIPS) singles out the mouth as the active speech organs. The verbs add
additional semantic content: the mouth as a container that is open when active
and closed when passive (open/shut) or the mouth and lips as moving bodyparts
(stop). Similar variations on the same theme can be found in give tongue, set
tongues wagging and give s.o. the rough side of one's tongue.
The speech organs are controlled by the mind. When the body takes over,
things go wrong. This is the background assumption behind bridle/guard one's
tongue: when S loses control over himself, he is conceptualized as losing
control over his body. The same assumption motivates the expression find one's
tongue, in which S has lost his ability to locate his tongue. When S has regained
control, he may wish to punish the rebellious body part: bite one's tongue off.
The distinction between the mind and the speech organs also motivates
mouth ('to speak without understanding or sincerity') and pay lip service to ('to
support insincerely'): when S says something he does not mean or understand,
the situation is conceptualized as one where the body acts out of line with the
mind.
The expression put words into s.o. 's mouth has two meanings: 'to tell s.o.
what to say' and 'to claim falsely that s.o. said something'. Both are motivated
by a situation where someone else (instead of S's body) has taken over control.
Take the words out of someone's mouth and hover ('a question hovered on
his lips'), just like put words into s.o.'s mouth, profile the mouth as a region
where words-objects are situated before they are actually spoken. In both cases,
they remain unspoken: someone else has taken over control or S decides to
wait. In the case get one's tongue around (a word that is hard to pronounce),
the word-object is also located in the mouth, where it functions as an obstacle
for the tongue.
30 JOHAN VANPARYS
Of course, the primary function of the mouth is to take in food. Its func-
tion as the interface between the digestive system and the outer world gives rise
to a cluster of metaphorical expressions where speaking is conceptualized as the
exteriorization of words, either in the form of food or of saliva.
In the case of regurgitate (as in the case of a student regurgitating what
the teacher said), S is represented as exteriorizing undigested food, which
stands for words he does not understand. The underlying metaphor equates
understanding with digestion. In the case of drool and slobber, on the other
hand, S is portrayed as exteriorizing saliva, which we usually associate with
foolish behaviour.
Pour ('Curses poured from his lips' LDOCE) represents the words as an
excessive amount of some liquid substance. Whether it is food, saliva or some
other specific substance remains indeterminate. Similarly, spit (it) out and cough
(up) characterize the words as an indeterminate substance that clogs the
digestive or respiratory channel. (Bodyparts as a donor domain are discussed in
greater detail by Pauwels & Simon-Vandenbergen, this volume.)
4.10 Cognitive and emotional effects as physical effects
Speech acts produce certain (intended or non-intended) effects in H. These, too,

are subject to reification.
Directive speech acts (attempts to get H to do something) may require a
certain degree of (verbal) pressure, which is often rendered as PHYSICAL
PRESSURE. For instance S may be said to bulldoze, charge, press, pressure,
pressurize, prod, push (about/around) or spur (on) H to get him to act or press
for action.
Some speech acts, especially persuasives and directives, have the effect of
MOVING H, that is changing his location. Persuasion can be expressed by bring
s.o. (a)round ('to persuade into a change of opinion'), bring s.o. to terms ('to
force to agree') or talk s.o. round ('to persuade'). Directives include bring s.o.
out, draw s.o. on/out ('to encourage s.o. to talk'), goad s.o. into doing s.th.,
shove s.o. around ('to annoy s.o. with many orders'). Natural candidates for
metaphorization by a motion verbial are requests or orders for H to come or go
somewhere: bring s.o. in ('to ask to come to one's help'), haul s.o. ('to
summon s.o.'), move s.o. on ('to order s.o. to go away'), turn away ('reject, eg
prospective students').
The obligation resulting from commissives, directives and official decla-

rations can be phrased in terms of a BINDING impact: bind s.o./oneself to ...,
bind over (in court), bind s.o. over to keep the peace, fasten s.o. down to (eg a
date), nail s.o. down to. Fascination, too, may have a binding effect: grip ('to
fascinate'), hold (s.o.'s attention, s.o. silent), magnetize. Surprise, on the other
hand, brings H out of BALANCE: bowl s.o. over, jolt, knock s.o. (back), knock
s.o. cold/out/sideways/for a loop/for six, sweep s.o. off his feet.
Additional examples of cognitive or emotional effects phrased in terms of
physical effects include:
(101) His funny story really broke me up. (LDOCE)
(102) I will not burden you with a lengthy account. (LDOCE)
(103) News that the king had imprisoned many parliamentary officals
threatened to convulse the country. (LDOCE)
(104) That joke really creased me. (LDOCE)
(105) a thrilling story (LDOCE)
4.11 Verbal aggression as physical aggression
One of the better known metaphorical clusters introduced by Lakoff and

Johnson (1980) is ARGUMENT IS WAR (see also Lakoff 1992). The corpus
includes numerous instanstiations, though they suggest a higher level of
generality: VERBAL AGGRESSION is PHYSICAL AGGRESSION. Obviously, this
equation is compatible with the one introduced in the previous section:
COGNITIVE AND EMOTIONAL EFFECTS ARE PHYSICAL EFFECTS. It also Subsumes
the verbials of violent transfer mentioned in 4.3.5. Extensions from this category
include fling/throw dirt at s.o., fire away, let fly at s.o., people who live in glass
houses shouldn't throw stones and sling/throw mud at s.o. The difference with
the verbials in 4.3.5 is that these extensions no longer make literal reference to
S's words, giving rise to a higher degree of metaphorization. Compare fling a
remark at s.o. (4.3.5) and fling dirt at s.o.
In addition, the corpus includes a diverse array of verbials related to
fighting or the infliction of pain. Note that most do not necessarily make
reference to a situation of war, which motivates the more general phrasing
suggested above. Examples include annihilate s.o., cut up s.o., fight, fight (it)
out, make the dust/feathers/fur/sparks fly, get (at) s.o., go for s.o., hit out
at/against s.o., rule (s.o.) with a rod of iron, jump on s.o./s.th., jump down s.o. 's
throat, kick s.o. around, knock s.o., lace into s.o., lambaste s.o., lash (out at)
s.o., lay about/into s.o., let out at s.o., massacre (eg one's opponents), maul
32 JOHAN VANPARYS
s.o., make mincemeat of s.o., needle s.o., rub s.o.'s nose in the dirt, pull/tear
s.o. apart/to pieces, pulverize s.o., cut s.o. to the quick, round on s.o., run s.o.
down, slam s.o., slash s.o., tackle s.o., take/tear s.o. apart, tear into s.o., tear
s.o. off a strip, tear s.o. to shreds, cast/throw in s.o. 's teeth, tilt at s.o. and wade
into s.o.
Other verbials do situate the aggressive speech event in a context of war-
fare, eg campaign/crusade (eg for a cause), cross swords with s.o., give/lose
ground, hold/shift one's ground, bring up one's big guns and stick to one's
guns.
For a related category of aggression metaphors, the donor domain is tor-
ture or punishment: crucify, drag/haul s.o. over the coals, excoriate, flay, grill,
keelhaul, pillory, rack, give s.o. a (good/real) roasting.
4.12 Sounds
Since speech involves an acoustic component, the domain of verbal communi-

cation comprises the domain of sound production as one of its subdomains.
Various linguistic action verbials (eg scream, shout, whisper, yell, ...) pick out
the acoustic quality of the reported speech event as their major meaning
component. Metaphor comes into play when the verbial is borrowed from a
domain that involves the production of sounds in another context than verbal
communication (eg thunder). Two donor domains stand out for their promi-
nence: animal sounds and music. Indeed, a whole range of verbials expressing
the production of sounds by animals can be applied to human communication.
Here is just a small sample: bark, bellow, bill and coo, bleat (out), cackle,
caterwaul, chirp, crow, drone away/on, purr, snarl, squawk, squeak, twitter.
Extensions from the domain of music include drum up, pipe, and trumpet.
4.13 Extensions from non-verbal to verbal communication and within the

domain of verbal communication
Communication may also take place by non-verbal means, for instance by

shaking hands to greet someone, patting a person on the back to praise him or
pointing at a person to criticize him. As a result, verbials such as give s.b. the
glad hand, pat s.o. on the back, put the finger on s.o. and point the finger (of
scorn) at s.o. qualify as (literal) verbs of communication. When applied to
communicative events where these gestures are accompanied by words, they
acquire metonymic status (pars pro toto). Metaphorization takes place when the
same expressions are used to describe speech events where the designated
gesture is not made. (See also Goossens, this volume.)
In quite a number of cases, the non-verbal act of communication qualifies
as a more or less ritualized event in an institutionalized setting, such as duels
(throw down/take/pick up the gauntlet) performances (applaud s.th., blow one's
own trumpet/horn), hierarchies (bow to s.o., bow down the neck to s.o.,
bow/bend the knee (to) s.o., fall/go on one's knees), religion and supersition
(beat one's breast, cross one's heart and hope to die), the navy and merchant
navy (lower/haul down one's colours/flag, nail one's colours to the mast, sail
under false colours).
Even within the domain of verbal communication, extensions are possible
between various subdomains. For instance, the following verbials require a
religious setting in their literal senses but they can also be applied to describe
other speech events: say amen to, cast aspersions on, damn, pronounce s.o.'s
doom, speak in parables, pontificate, pray, preach, prophesy, swear.
4.14 Complex categories
In the first subsections of section 4, metaphorical items were discussed in terms

of fairly schematic terms, such as 'object', 'container', 'expansion and com-
pression', 'motion', 'transfer' etc. Gradually, richer images have come into play,
such as the human body and its functions, warfare, religion, etc. In order to
round off this (potentially endless) survey of communicative metaphors, I would
like to indicate a few of these more complex domains that frequently serve as a
source of metaphorization, such as cooking and serving meals: cook up (eg a
story), dish/ladle out (eg advice), dish up (eg arguments), feed (eg s.o. with
information), lard ('ornament'), mince one's words, regurgitate, sauce or
spatchcok.
Another productive donor domain is sports: the ball is in your court,
bandy words with s.o., floor s.o., overshoot the mark, carry/gain one's point,
score (a point (off/over s.o.)), beat s.o. to the punch, call one's shot, spar,
stonewall, argue the toss, throw in the towel/sponge, ...
Animals and their behaviour, finally, give rise to such expressions as ape,
badger, bark up the wrong tree, bay at the moon, bitch, have a bone to pick
with s.o., clam up, crab, fawn, have/put out feelers, make the dust/feathers/fur
fly, fox, hawk, pick holes in s.th., you can lead a horse to water but you can't
34 JOHAN VANPARYS
make it drink, parrot, pick over, pick people's brains, rabbit, trot out, talk
turkey, ...
5. Conclusion
Based on a large corpus, several clusters of related metaphors have been

identified. Most are reifications. A linguistic expression (a word, a sentence, a
story ...) can be conceptualized as a container for its content (4.1) or as an
object that can be produced (4.2.1), adorned (4.2.1), expanded or compressed
(4.2.3), or modified in some other way (4.2.4). Reification also results in
visualisation (4.2.5). Information transfer can be conceptualized in terms of
motion of ideas towards H (4.3) or a towards a patient (4.5). In other meta-
phorical clusters, S and H are viewed as containers (4.6). S can moreover be
conceived of as an entity that moves towards H (4.7.1), the topic (4.7.2) or the
goal (4.7.3) of the conversation, which may take the shape of a container (4.8).
The effects of communication can also be reified in a variety of ways (4.10).
More complex metaphorical clusters involve the equation of verbal aggression
with physical aggression (4.11) and extensions from non-verbal to verbal
communication (4.13) or conceptualizations based on cultural notions (4.14).
What about the conduit metaphor? Although certain aspects of it are con-
firmed (cf. 4.1, 4.3), Reddy's claims need to be qualified. For one, plenty if not
most of the metaphorical clusters revealed by the data do not invoke the conduit
model at all. The wide range of conceptualizations conventionalized in the
English lexicon certainly undermines Reddy's pessimistic conclusion: rather than
imposing one single model, the English language as its own metalanguage gives
its users the opportunity to select from a rich gamut of expressions. For another,
even in those cases that can be related to the conduit metaphor, this model
usually fails to offer sufficient clues to appreciate their complexity. In other
words, Reddy oversimplifies matters and additional elements have to be drawn
into the analysis if one wants to achieve a full understanding of a given instance.
Acknowledgement
The LDOCE computer-tape was kindly put at our disposal by Prof. Em. Dr.
L.K. Engels.
Body Parts in Linguistic Action
Underlying Schemata and Value Judgements
Paul Pauwels
Catholic Flemish Institute of Higher Education
Anne-Marie Simon-Vandenbergen
University of Ghent
0. Introduction
In this paper we will take a closer look at metaphors which originate in the
domain of 'body parts' 1 . These are metaphors which either contain explicit
names of parts of the body or refer to processes in which body parts are implicit
but play a central role. Our material consists of 175 items, selected from the
corpus of linguistic action metaphors described by Vanparys in this volume.
The view of metaphor we subscribe to has been outlined in Lakoff and
Johnson (1980) and Johnson (1987). In this view, metaphor is not merely an-
other, non-congruent way of referring to phenomena (processes and partici-
pants), but rather an important mode of understanding and a way of structuring
experience. The concepts used in metaphorisation are "concepts for natural
kinds of experience and objects [which] are structured clearly enough and with
enough of the right kind of internal structure to do the job of defining other
concepts. [These other concepts] are less concrete or less clearly delineated in
their own terms" (Lakoff and Johnson 1980: 118). One important function of
metaphor is then to structure abstract domains by means of projections from
more concrete domains. The domain of body parts and bodily functions is one
such structured concrete domain. The study of metaphors of linguistic action
which refer to parts of the body and their functioning may thus contribute to a
clearer understanding of how physical experience is projected onto linguistic
action. More specifically, the aims of this study are the following.
36 PAUL PAUWELS & ANNE-MARIE SIMON-VANDENBERGEN
• First, we wish to examine which aspects of linguistic action are grasped

through metaphorical projection from the physical domain. This involves
looking at ways in which the linguistic action domain is structured by means
of various concrete domains all involving the use of body parts. We shall also
look at the way in which a number of basic schemata (in the sense of Johnson
1987) form the experiential grounding of the body part metaphors.
• In the second part of the paper we shall look at the way in which the meta-
phors of the corpus express value judgements in the domain of linguistic ac-
tion. The hypothesis is that value judgements are an important motivating
factor in the creation of metaphors (see also Rudzka-Ostyn 1988a: 524). In
reporting someone else's speech act, a reporter may express value judge-
ments about any aspect of the primary speech act situation.2 This paper aims
to discover which aspects of the components of the speech situation (the re-
cipient domain) are valued positively and which negatively, and how these
judgements correlate with donor domains.
1. Structuring the linguistic action domain: use of the body
1.1 Donor domains
Through linguistic action (henceforth LA) human beings interact both with one
another and with their environment. However, LA is only one type of action and
interaction, and other types are more concrete; they involve the body more
'tangibly'. It is therefore logical according to our view of metaphor that the
latter types of interaction should be used as donor domains for the metaphori-
sation of LA.
In what follows we will present a survey of the major types of donor do-
mains involving body parts and outline how they give structure to the LA do-
main.
1.1.1 Donor domains involving articulators in non-speech functions3

A first set of metaphors involves body parts which function in speaking but
which are put to a different use. Such donor domains are, for instance, eating
and breathing. We want to emphasize from the start that not all domains are so
specific in nature, and that vaguer domains like use of the mouth, tongue and
throat should also be recognized. Some examples may clarify the picture.
BODY PARTS IN LINGUISTIC ACTION 37
Eating and feeding

The donor domain of eating is used to structure LA in different ways. First,
metaphors like feed and force/ram/thrust something down someone's throat
describe the interaction between two persons; the speaker transfers something
to the hearer, and eating stands for receiving information. The ground for the
metaphor is not the use of the mouth, common to both the activities of eating
and of speaking, but the aspect of interiorization found in both eating and lis-
tening.
A second subset describes speaking in terms of eating (or part of the eat-
ing process). In chew the rag and chew the f at (meaning either 'complain' or
'chat' (LDOCE)4 the focus is on the duration of the action - both rag and fat
are substances one can chew on a long time - and the fact that not much is go-
ing to come out of it. The LA is described as of long duration and pointless. The
object of the chewing can also be seen as describing the subject/content of the
LA as old, leftover or of poor quality (cf. the definition 'discuss, esp. old griev-
ances' (OALDCE)). The metaphor eat one's words ('admit one has said some-
thing wrong' (LDOCE)) provides another kind of structure for the act of
speaking. It uses the directionality of eating (ingesting) as contrasted with that
of speaking (exteriorizing). The act is an undoing of a previous LA; the offen-
sive utterance is destroyed by making it go back to where it came from: once
inside the speaker it becomes invisible and cannot have an effect on the audience
anymore. In contrast, regurgitate does have the same directionality as speaking.
It describes the utterance as something which has been ingested, but not di-
gested. The substance (food / information) has not been inside the speaker long
enough to be processed. Also relevant is the meaning element 'repetition'.
Breathing
The structure of the donor domain of breathing differs from that of eating in one
important respect, viz. directionality. In this case there is no inherent opposition
to the directionality of speaking, since breathing involves both interiorizing and
exteriorizing. Moreover, there is a much closer, metonymic link with LA, since
one uses one's breath in speech. For instance, in metaphors such as breathe (as
in the example from LDOCE: "He breathed words of love into her ear"), puff
(out), waste one's breath, the salience of air as an ingredient of speech is very
high. Further, the domain of breathing frequently interacts with another domain.
In waste one's breath, it is the domain of valuable objects which comes into
play: the metaphor stresses the importance of breathing for the functioning of
the body, and advocates an economical use of one's resources. The LA is de-
scribed as useless because it does not have an effect; it does not pay off. In
cough up, the focus is on the removal from the body of something (blood,
phlegm) which causes discomfort. While breathing involves both in- and exha-
lation, the directionality of coughing is obviously an outward one. The domain
interacting with that of breathing is clearly that of illness, and relief can be at-
tained by pushing air out from the throat. The parallel with speech is to be
sought then not only in the use of air, but in the removal of some disturbing
substance from the body (donor domain) and from the mind (LA domain). A
relevant example is: "Even after the police threatened him, Smith refused to
cough" {LDOCE). In choke back, the speaker prevents his feelings from escap-
ing his body by holding them in the throat. The metaphor describes controlled
silence.
Other metaphors focus on silence that is beyond the control of the
speaker. For instance, take someone's breath away ('make unable to speak'
{LDOCE)) attributes the inability to speak to an external agent.
Use of articulators
The metaphor drool ('talk foolishly' (LDOCE)), in which the mouth is neces-
sarily involved, describes a natural reflex which human beings should have learnt
to control. The metaphor brings in three elements: the speaker's behaviour is
described as uncontrolled, the substance brought out is characterized as val-
ueless, and the action as a whole as distasteful and socially unacceptable. In the
metaphor spit out 'say or express with effort, force or anger' {LDOCE), the
value of the substance (whether it be food or phlegm) and the unacceptability of
the action are likewise meaning elements.
A number of expressions focus on the movement of the visible speech or-
gans, and clearly have a metonymic basis:5 keep one's mouth shut, open one's
lips, and close lipped describe presence or absence of speech; lip, chinwag and
jaw describe a type of LA, and are more clearly metaphorical. Other examples
are tongue in cheek and lie in one's teeth/throat. The listener's role is focused
on in turn a deaf ear to, go in one ear and out the other and have a word in
someone's ear. (Although ears are not articulators, they are salient body parts
involved in oral communication). All of these metaphors seem to rely exclu-
sively on the body part domain, in contrast to the majority of instances in our
corpus, where the body part domain plays only a secondary role in the meta-
phorisation, by being the chief instrument in other kinds of (inter)action.
1.1.2 Domains involving other body parts

Some metaphors are based on the use of the body in other types of (inter)action.
Closest to the domain of LA, one finds the domains of nonverbal communi-
cation and sensory perception (seeing, touching, smelling). Apart from these,
fighting, physical punishment, restricted movement, the manipulation of
objects and walking are also used as donors.
Nonverbal communication
In metaphors based on nonverbal communication, there is usually a metonymic
link with the LA domain, since the verbal and nonverbal actions frequently
cooccur. This is the case in expressions like pat on the back, bow/bend the
knee/neck to, tip someone the wink, put the finger on. However, since the ac-
tions do not need to cooccur, these expressions are also metaphorical. For in-
stance, tip someone the wink 'let someone know quietly or secretly, so that they
can take immediate action' (COBUILD) is a kind of communication which does
not necessarily involve the use of nonverbal means. The relevant feature which
gives rise to the metaphors, however, is the discreetness of the signal and the
secrecy of the act.
Some of the metaphors stand out because they involve a kind of nonver-
bal communication which is somewhat out of the ordinary. In pat oneself on the
back the focus is on the ridiculousness, as much as on the far more abstract
unacceptability of self-praise. In blow one's own trumpet - which has more or
less the same meaning - the LA is described as a way of attracting attention to
oneself. Here, too, the deviation from normal behaviour (normally important
persons are heralded by someone else) is significant since it emphasizes the
inappropriateness.
Some metaphors are based on animal behaviour to which a communica-
tive function is attached. For instance, in snarl ('speak, say in an angry,
bad-tempered way' (LDOCE)) the transfer to the LA domain is quite straight-
forward. The fact that we are dealing with a particular kind of animal behaviour
as contrasted with human behaviour often accounts for the negative value
judgement expressed by metaphors. (See further section 2).
Sensory perception
Metaphors based on sensory perception structure the LA domain in an unex-
pected way: instead of describing the act of hearing, they are indeed often used
to focus on a relevant aspect of the act of speaking. This curious reversal is
accounted for in different ways.
The metaphor sniff ('say something in a proud, complaining way'

(LDOCE)) is based on the fact that this act of perception is combined with a
production of noise. The noise of sniffing, moreover, has a communicative
value, since it signals that the agent smells something. This value is strengthened
because the action is usually performed when the smell is objectionable in one
way or another. The metaphor could also be classified in the domain of nonver-
bal communication.
Poke one's nose into something describes an action preparatory to sen-
sory perception. The aim is to obtain information. There are two possible con-
crete donor contexts: the human intruder in the kitchen who goes around lifting
the lids of pots and sniffing at the contents, or the animal (dog or cat) investi-
gating dustbins. In both cases, what is investigated belongs to others, and hence
the act is slightly objectionable, which is another salient aspect in the metapho-
risation.
Other metaphors clearly rely on the animal domain, since they involve
body parts which humans do not have or refer to actions which humans do not
perform. Prick up one's ears describes the act of listening, while put out feelers
again focuses on the use of language to elicit information.
Violent actions
Within the domain of violent actions, one subdomain is that of fighting. The
activity of boxing gives rise to expressions like spar, pull one's punches, beat
someone to the punch, punchline. We are here dealing with an organized, rule-
governed activity in which the violence is acceptable. Hence, the focus of the
metaphor is primarily on another aspect of the interaction. The grounding obvi-
ously lies in the description of the LA as involving two opponents. In spar,
('argue, but not in an unpleasant or serious way'(COBUILD)) the focus is on
the preparatory character and the playful nature of the argument. In pull one's
punches the speaker is described as being soft on the hearer; he does not
'attack' him as strongly as he could or should. Throw dust in someone's eyes
describes unfair tactics used by a speaker who hides his actions from the hearer.
Perhaps metaphors like tongue twister and jawbreaker also belong here, because
in these cases a word or phrase is given the status of a fighter, who succeeds in
incapacitating his opponent - the speaker. The focus is on the autonomy of
words and phrases, and their control over (tongue twister) or destruction of
(jawbreaker) the body parts involved in the LA. Hyperbole is clearly present
here.
A second subset of expressions has punishment as donor domain. There

is punishment as of schoolboys (rap someone over the knuckles, burn someone's
ears), in a military context (tear someone off a strip), as torture (excoriate) or
as atonement (bite one's tongue off). The focus is on the experiencer of the
painful sensation. The first two metaphors describe the act of scolding as caus-
ing someone pain. There is probably a metonymic basis since the physical ac-
tions are often accompanied by scolding or reprimanding. Excoriate ('express a
very bad opinion of (a book, play, performance etc.)' (LDOCE)) adds an ele-
ment of destruction which is also common with other metaphorisations of the
act of criticizing (see further). Bite one's tongue off describes silence as causing
grave discomfort to the agent: the non-expression of certain feelings is seen as
having a devastating effect. The same applies to eat one's heart out, which
moreover relies on the domain of eating. The domain of punishment extends
into the domain of human-animal interaction with metaphors like muzzle and
rub someone's nose in the dirt (treatment of puppies).
By far the largest subgroup involves the description of violent actions
pure and simple. In metaphors like boot out, kick someone around, tear apart,
choke off, the relationship between agent and experiencer is transferred onto the
LA domain. The relationship is asymmetrical, in that the experiencer is clearly in
no position to fight back. In the first two cases, the asymmetry is expressed
through the use of the foot; the experiencer is literally in a lower position. Both
describe a LA which is an assertion of authority. The metaphors tear apart and
choke off focus mainly on the effect, which is destruction: the extreme nature of
the violence transfers onto the LA domain, where the actions of scolding or
making someone shut up subject the experiencer to the authority of the speaker.
In actions like kick, kick up a row/rumpus/stink/shindy, the violence is not
directed at any experiencer. This aimlessness plays an important role in the
metaphors: the violence is useless, the LA is described in the same way. Again
this set of metaphors involves use of the foot; indeed, people often start kicking
at things to express their anger. In kick against the pricks ('complain uselessly
about something that cannot be changed' (LDOCE)), the donor domain is the
use of the legs by cattle to kick against the goad. Here also the meaning of
useless linguistic action derives from that of physical action by means of which
the animals only hurt themselves.
Most of these metaphors describe violent actions performed by animals:
make the fur fly, backbite, bite (N), snap, snap at, snappish, snap someone's
head off, bite someone's head off (and maybe also talk/shout someone's head
off and tear apart) focus on the violent interaction between animals. The vio-
lence is often extreme, and in the metaphorisation the focus is, in most cases, on
the unmotivated nature of the action. The LA is described as disproportionately
hostile under the circumstances. The agent is overreacting, not controlling him-
self. Backbite describes a LA in which the victim is defenceless because of his
absence. Bite merely describes the effect of the LA on the hearer in terms of a
painful sensation. (The example from LDOCE is: "His words had a cruel bite to
them which we all felt keenly.")
Restricted movement
The domain of restricted movement gives rise to metaphors such as: tonguetied,
muzzle, bridle one's tongue, guard one's tongue and (less unambiguously)
watch one's words6.. The first metaphor describes the inability to speak, the
second one the imposing of silence, while the other three describe self-control.
In muzzle and bridle one's tongue the donor domain is clearly that of animal
control, while in the others the donor is less specific. All metaphors describe
silence or carefully controlled speech. The movement of the tongue as an articu-
lator is a prominent element in three of the metaphors.
Manipulating objects
Another important donor domain is that of manipulating objects. The aspects in
focus are the (lack of) skill of the manipulator (fumble, shoot one's mouth off),
his physical capacities (heavyhanded, a lefthanded compliment), or linguistic
action as exchange of information (hand it to someone).
The metaphor fumble has three slightly different uses, which are illus-
trated by the following examples:
(1) He often has to fumble for the right word. (LDOCE)
(2) [We] are made to feel inferior if we fumble an unusual word.
(Webster's)
(3) What I'm fumbling to say is that I felt different about you.
(COBUILD)
(4) He fumbled in answering and made them suspicious. (Webster's)
In (1) the act is clearly that of looking for an object. The metaphor says that the
speaker is unable to produce a word, and this inability is put down to word-
finding difficulty. Words are conceived as objects, which are stored away and
have to be produced quickly if needed. In (2) the situation is different. The
focus is in this case on the maltreatment of the object. The word is produced,
but as a result of clumsiness, it comes out wrong. The descriptions in (3) and
(4) pay attention to the clumsiness of the speaker, rather than to the manipula-
tion of the objects. This clumsiness is of course the common ground in all these
examples, but the first two stand out in that they allow a more specific transfer
from the donor domain to the LA domain.
The metaphor shoot one's mouth off is still more specific. The mouth is
described as a gun, which in this case causes harm because it is handled care-
lessly or unskilfully. In shooting off a gun, one wastes a bullet and one attracts
attention; one might also hurt someone. Similarly, when one shoots one's mouth
off, one talks 'foolishly about what one does not know about or should not talk
about' (LDOCE).
Pick a quarrel and split hairs are instances of skill used wrongly, in the
first case because of the domain of application, in the second because of the
excessive and useless nature of the action.
Walking
A final domain is that of walking, used to describe the LA as a movement of the
speaker. Again, different metaphors focus on different aspects. Still, the ele-
ments of skill and efficiency seem to play a role in all of them. In backtrack the
aspects in focus are the direction of the movement, and the fact that this is the
consequence of a mistake. The LA is shown as a correction of an error, which
involves a movement opposite to the normal direction of speech. In put one's
foot in it, and never put a foot wrong, the focus is mainly on the (un)skilful
handling of the body. In the first metaphor there is also the aspect of stepping
into some substance which is obviously to be avoided.
1.2. The basis for the metaphors: an image-schematic approach
1.2.0 Introduction
Following Johnson's (1987) approach, we will now describe the metaphors
from another angle. According to Johnson, metaphors are based on our abstract
bodily experience of the world, which we translate into basic schemata. He
defines a schema as "a recurrent pattern, shape, and regularity in, or of, [our]
ongoing ordering activities. These patterns emerge as meaningful structures for
us chiefly at the level of our bodily movements through space, our manipulation
of objects and our perceptual interactions." (Johnson 1987: 29). Johnson dis-
tinguishes several such basic schemata, and he provides a list (1987: 126) which
is said to be highly selective, but contains those schemata which Johnson thinks
most pervasive. We have included the list here, and will refer to it in the discus-
sion:
CONTAINER BALANCE COMPULSION BLOCKAGE

COUNTERFORCE RESTRAINT REMOVAL ENABLEMENT ATTRACTION
MASS/COUNT PATH LINK CENTRE/PERIPHERY
CYCLE NEAR/FAR SCALE PART/WHOLE
MERGING SPLITTING FULL/EMPTY MATCHING
SUPERIMPOSITION ITERATION CONTACT PROCESS
SURFACE OBJECT COLLECTION
Some of these schemata are further specifications of a more general experience:

COMPULSION, BLOCKAGE, COUNTERFORCE, RESTRAINT REMOVAL, ENABLEMENT
and ATTRACTION all specify the FORCE schema. Schemata also interact, as in the
case of the CONTAINER schema and the FULL-EMPTY schema, or the COM-
PULSION and the PATH schema. The schemata can be further analysed in several
component elements. For instance, for the PATH schema these would be the
source, the goal, and a sequence of locations in between; the schema is usually
given a directionality which relates to the prototypical movement along a path.
In what follows, an attempt will be made to generalize over the interac-
tion between the different donor domains and the LA domain; the focus will be
on the experiential grounding of the metaphor. To exemplify the different types
of interaction, we will mainly rely on those metaphors which we did not classify
in specific donor domains. It can indeed be argued that metaphors in which the
similarity between donor domain and LA is less obvious than is the case in for
instance metaphors from the domain of eating (use of articulators), nonverbal
communication (communication) or physical violence (interaction between hu-
man beings), rely most heavily on the image-schematic relationship. At the same
time, however, we will also pay attention to those metaphors partly explainable
in terms of the donor domains identified in 1.1 above. We will try to show that
the image-schematic basis can make a specific contribution in these cases, too.
First, a number of metaphors will be described which rely on the
CONTAINER schema. It appears that this schema interacts with a number of oth-
ers, which we will have to include in the discussion. In the remainder of our
account we will exemplify some basic uses of other schemata - notably FORCE,
BALANCE, PATH, CONTACT and CONTROL - but we will not treat them exhaus-
tively.
1.2.1 The container schema

The CONTAINER schema is used to describe different aspects of the linguistic
action. Among the metaphors referring to the speaker's actions, there are a
number which describe his body as a container for language. The lips and mouth
are then the place where the container can be opened or closed, as in open one's
lips, shut one's mouth, close-lipped. When the container is opened, there is LA,
when it is closed, there is silence. The BLOCKAGE schema (Johnson 1987:45) is
also relevant here, since the lips form a blockage to prevent LA from taking
place. Open one's lips is based on the experience of the REMOVAL OF RESTRAINT
(Johnson 1987:45). In bite one's lips, the closing itself is combined with strong
force.
In unbosom ('tell the secret feelings, esp. troubles and worries of oneself'
(LDOCE)), the CONTAINER schema is combined with a PATH schema. The meta-
phor describes LA as the removing of feelings inside the bosom (which is a
container within a container), and bringing them into the open. The schema
which contrasts CENTRE and PERIPHERY is used implicitly, as the bosom is seen
as centrally located, far from the periphery of mouth and lips, used in speech.
The container is regarded as a hiding place: things inside are invisible and secret
and become visible, perceptible in the exteriorization. This metaphor is one
which relies on the combination of image-schemata, rather than on any distinct
donor domain; there is no act of unbosoming outside the LA domain. Another
aspect which plays a role here is the folk-theory which conceives of the bosom
as the seat of emotions. However, the role of folk-theory with respect to meta-
phorisation is beyond the scope of this paper.
In lie in one's teeth the container is also seen as a hiding-place, but in this
case a defective one, because of its being on the periphery, and because it is far
from prototypical. Hence, the metaphor expresses shameless lying, almost with-
out an attempt to conceal one's mendacity.
For the metaphor shit ('worthless talk' (LDOCE)) the CONTAINER image
also provides the grounding - it is something which has been removed from the
body. This basis for the metaphor however is not at all in focus in this case; the
quality and the social judgement of the substance are more important.
None of the metaphors we have considered so far pay attention to the ex-
periencer in their description of the LA.7 In the following, the CONTAINER
schema is used to describe the hearer, while the speaker's action is described as
the exertion of some kind of FORCE on the container. There are two basic situa-
tion types: the first describes the speaker as forcing entry; the second describes
him as using force to keep the container closed. The emphasis differs from
metaphor to metaphor. In cast/throw something in someone's face/teeth, the
intensity of the force is salient; in have a word in someone's ear, the actual entry
is in focus; get an earful pays attention to the lack of control on the part of the
experiencer, while rub in, force/thrust/ram something down someone's throat,
shut someone's mouth combine all three aspects.
It may be noted that the metaphors describing entry are not just restricted
to the normal way of entry in LA (the ears), but also use the mouth as an en-
trance, which links up with the donor domain of eating, or in this case
force-feeding. Tear into and rub in stand apart in this respect. Tear into
('criticize very strongly' (COBUILD)) does not describe the locus of entry; it
simply focuses on the destructive violence involved in effecting entrance. In rub
in, the nature of the force - not extreme, but repeated - is at stake. The defini-
tions in different dictionaries stress different aspects of this metaphor: 'say re-
peatedly' (LDOCE), 'force a lesson into someone's mind' (OALDCE), 'remind
someone of something he does not want to be reminded of, usu. because it is
embarrassing' (COBUILD). The LDOCE definition focuses on the repetitive-
ness of the rubbing, the OALDCE definition combines the force used in the
process with the description of the intended effect (lesson), while the COBUILD
definition refers to the side effect on the hearer (embarrassment).
The extent of entry into the container may vary, as can be seen from put
words into someone's mouth and force/ram/thrust something down someone's
throat. The extent of entry here correlates inversely with the subtlety of the
agent's action. Furthermore there is another difference: whereas the second
metaphor is prototypical of forced entry in that it describes a LA in which the
speaker controls the effect on the hearer, the first metaphor draws attention to
the hearer as speaker in another LA which either follows ('to tell someone what
to say' (LDOCE)) or precedes ('to claim, falsely, that someone has said a par-
ticular thing' (LDOCE)). In either case, the focus is on the control of someone
else's speech.
Yet another use of the CONTAINER image describes the LA as the removal
of something from a container: pick someone's brains, brainwash, take the
words out of someone's mouth, take one's breath away. The actions violate the
self control of the affected person. His words, ideas or capacity to speak/act are
influenced by an outsider. In pick someone's brains the interaction described is
that of eliciting information. The LA is conceived as an act of stealing, but be-
cause of the skilfulness the experiencer does not notice this (cf. picking some-
one's pocket). The information, ideas, etc. stored in the brain are looked upon
as valuable possessions, which should not be interfered with8. The same applies
to brainwash.
A final use of the CONTAINER schema is found in poke one's nose into
something and horn in; here an ongoing process is described as a container, and
the LA as an unacceptable attempt at entering, because it is construed as a
violation of private territory. In horn in, this violation is combined with the use
of force.
To conclude this section, let us survey the role of the body parts in these
metaphors. They function as parts of the container (especially exit and entry:
lips, mouth, teeth, skin), as contents (heart, breath), as containers (brain, mouth,
bosom, cheek, throat, teeth) and as instruments used in the exertion of force
(hands, implicitly in most cases, and horns).
7.2.2 Force
According to Johnson (1987: 42), the FORCE schema is necessarily involved in
our interaction with the environment. Our body exerts and/or undergoes force in
any kind of (inter)action. In normal circumstances, however, our awareness of
these forces is very low, because we take them for granted, and we only notice
them "when they are extraordinarily strong, or when they are not balanced off
by other forces." (Johnson 1987: 42).
Such experiences of force should have the required salience to be used in
metaphorisation. It is therefore not accidental that a large group of instances in
the corpus actually have violent actions or fighting as a donor domain. In
these cases force is clearly in focus, either because of its extreme nature, or
because of the opposition of forces from different sources. The elements of the
FORCE schema which are most salient in that context are the degree of force
and/or the interactional character. In many cases, however, the FORCE schema is
further specified in different ways through the combination with other schemata
such as PATH, CONTAINER, and BALANCE.
Compare the following:
(5) He had studied very little and has no reason to kick about low
grades. (Webster's)
(6) He is always kicking against the system. (COBUILD)
(7) He kicks his children around a good deal. (Webster's)
In the first two examples, the metaphor kick is shown to rely exclusively on the
FORCE schema, though the emphasis differs: in (5) the intensity of the force is at
stake, in (6) there is specification of a target as well. Example (7) combines the
FORCE schema with the PATH schema. The movement of the experiencer ("his
children") becomes a relevant element.
An example of the combination of FORCE with the CONTAINER and
BALANCE schemata can be found in:
(8) The widow sat fuming and blowing off steam. (OED)
Here, the FORCE schema plays a secondary role, as the focus is clearly on the act
of restoring balance by relieving pressure in the container of the body.
The different types of FORCE schemata distinguished by Johnson were
found in our corpus. Example (7) above illustrates the COMPULSION schema, as
is clear from the combination with the PATH schema. The experiencer becomes
subject to the force and loses control of his movement. The same explanation
would apply to a metaphor like boot out. Examples of BLOCKAGE can be found
most clearly in combination with the CONTAINER schema. Here, the force is used
to effect either exit from the container (spit out) or entry into it
(force/ram/thrust something down someone's throat). Clear examples of
COUNTERFORCE were not found in this corpus, although muzzle could be ex-
plained in those terms, and perhaps also bite one's lips. An example of
REMOVAL OF RESTRAINT is found in bang out ('to write in haste' (LDOCE)).
The image of FORCE is again combined with the CONTAINER schema.
1.2.3 Path
A number of metaphors from various concrete donor domains can be explained
in terms of the PATH schema. From the domain of walking we have the metaphor
backtrack, which focuses on the directionality of the movement along the path.
In some metaphors from the domain of eating, the path becomes relevant be-
cause of the opposition between the directionality inherent in eating and that
inherent in speaking. Examples are found in metaphors like eat one's words and
eat crow, which are instantiations of a more general image of 'taking back one's
words'.
Earlier on we discussed some other combinations of the CONTAINER and
PATH schemata, where the path referred to the movement in the container (bite
back, force/ram/thrust something down someone's throat). In these, as in digest
and regurgitate, the image schema of CENTRE-PERIPHERY also plays a role. In
bite back, the path is cut off at the periphery, in force/ram/thrust something
down someone's throat an attempt is made to reach beyond the periphery (oral
cavity); regurgitate describes how the substance did not succeed in reaching the
centre or in staying there long enough; and digest says that it did.
Finally, the concept of removal, which operated in some of the cases dis-
cussed under the CONTAINER schema, also involves a PATH schema (pick some-
one's brains, take the words out of someone's mouth). In a number of other
metaphors, the schema plays a similar role, but the CONTAINER schema is absent.
Tear a strip off someone combines PATH and FORCE, as does get something off
one's chest. These exemplify several types of removal, however. In the former
case it violates the control of someone over his property or domain. In the latter
instance, removal involves a reassertion by the agent of his self-control. Johnson
(1987) would describe the second metaphor as an example of BALANCE, more
specifically of freedom from external pressure.
1.2.4 Balance, or control?

Johnson (1987: 85) describes the BALANCE schema as "consisting of force vec-
tors (...) and some point or axis or plane in relation to which those forces are
distributed. In every case, BALANCE involves a symmetrical (or proportional)
arrangement of forces around a point or axis." The schema is closely linked to
the exertion of force. In get something off one's chest, force is used to remove a
blockage, which caused an imbalance. The agent was subject to a certain op-
pressive force, which he now succeeds in removing. In doing so, he restores his
sense of well-being, of being in control again. Other examples are blow off
steam, cough and cough up.
It seems, then, that the BALANCE schema would correlate with the ab-
sence of FORCE. Johnson indeed implicitly suggests this when he defines FORCE
(1987: 42), adding that we only notice forces "when they are not balanced off
by other forces" (our emphasis). This, according to us, raises some questions as
to the status of the BALANCE schema. We do not wish to suggest that the expe-
rience of balance would not exist; Johnson has clearly demonstrated that it does.
What we do claim is that it could be reinterpreted in terms of another bodily
experience which - at least in some cases - perhaps has greater explanatory
value. The schema we are suggesting could be called CONTROL.
In our view this schema is basic, in the sense Johnson gives to the word.
CONTROL is a schema which operates both when it is present and when it is
absent. It basically functions on the level of human interaction with the envi-
ronment or with other human beings. A metaphor like slip of the tongue may be
explained as loss of control, drool as lack of control, while muzzle involves the
control of one actor over another. A human being tries to control himself and
his environment in several ways, and this is a precondition for being able to
make use of that environment. In fumble, for instance, there is a clear absence
of control over the object which the agent tries to manipulate.
The CONTROL schema of course very often interacts with the FORCE
schema where the control of outside agents is involved. Going through the
corpus again, we found that a lot of metaphors can be explained from this angle:
pick someone's brain, brainwash, take the words out of someone's mouth, put
words into someone's mouth all violate the self-control of the experiencer, talk
over someone's head can be construed as language passing the hearer by beyond
his controlled environment. In one of its interpretations, the metaphor mouth-
piece relies on the lack of control of the speaker over the content of his utter-
ances: he is only the channel through which the information passes. (See also
2.2.2. below).
The prototypical instantiations of the BALANCE schema, too, can be rein-
terpreted in terms of SELF-CONTROL. Metaphors like get something off one 's
chest and blow off steam describe the reassertion of self-control after an experi-
ence of outside or inside pressure. A metaphor like cough (up) describes an
uncontrollable action, a temporary loss of self-control. Since coughing can also
be interpreted as having a liberating effect, as taking away internal pressure, it
can also be a restoration of self-control. The context will decide which interpre-
tation is called for.
1.2.5 Contact
The final schema we would like to discuss is the CONTACT schema, which is also
at work in a wide range of metaphorical expressions. In a number of metaphors
it is, however, of secondary importance; it may be implicit, as in muzzle or kick,
or explicit, as in kick against the pricks, which primarily focuses on the useless
exertion of force.
In those metaphors where the schema is basic, it is used to describe two
different aspects of the LA. In in/out of touch, catch someone's ear, have some-
one's ear, the schema is used to describe the relationship between interactants,
while in touch ('to deal with or treat in speech or writing' (LDOCE), touch on
and fumble it describes the relationship between speaker and utterance.
The type of contact, its intensity, duration and the effort involved also play
a role in the metaphors. Touch and in/out of touch (with) are fairly neutral in this
respect; they simply use the presence/absence of contact to describe the pres-
ence/absence of a relationship between speakers or between the speaker and his
subject. With touch on and fumble the contact with the object is neither exten-
sive, nor firm - the metaphors rely on the contrast between mere contact and
control. Absence of firm, extensive contact leads to absence of control. In have
someone's ear and catch someone's ear the contact is with a body part obvi-
ously involved in listening. Have someone's ear involves both contact and con-
trol, or control in terms of contact, and stresses the permanent nature of the
communicative channel. Catch someone's ear focuses on the effort involved in
establishing such a channel, which results in achieving contact and control.
1.2.6 Image schemata and linguistic action

In retrospect, it is obvious that several image schemata should be at work to-
gether in any metaphorisation of LA. LA indeed has a complex structure involv-
ing several entities in different relationships. Individual schemata can hope to
account for only part of this, since it is their nature to represent basic and salient
elements of bodily experience, which are automatically less complex.
The explanatory value of schemata for the metaphorisation of LA depends
on their complexity. The CONTAINER schema, with the implications of contents,
inside-outside orientation, central and peripheral elements, the containing
structure, the possibility of things moving in and out, the possibility of quanti-
fying the contents (FULL-EMPTY), is obviously better placed in this respect than
the CONTACT schema, which is much simpler in structure. Still, we hope to have
demonstrated that this schema also goes a long way towards explaining some of
our metaphors.
We also hope to have shown the necessity of complementing the do-
nor-domain approach with the image-schematic approach, since the former does
not succeed in (fully) explaining all of our metaphors. Even with those it can
explain, image-schematic notions may indeed clarify the case. On the whole, the
image-schematic approach makes clear why a similarity can be perceived be-
tween the donor domain and the LA domain.
If we look at metaphor as a case of 'conceiving-as', we should be aware
that this process may take place at different levels of abstraction. For instance,
when we compare a metaphor like spar with bare one's heart/soul, it is clear
that the former relies on a more concrete experience of similarity than the latter;
spar can indeed be explained on the basis of a donor domain situation, whereas
for bare one's heart/soul this causes problems, since a situation of this kind is
just not available. It is there that the image-schematic approach is most reward-
ing, or, to put it differently, has a greater explanatory value.
2. Value judgements
2.0 Value judgement in the linguistic action scene
When describing linguistic action, a speaker may at the same time express a
judgement or give an evaluation of that LA or of any component of that LA.
When we look at the LA scene, it is clear that the value judgement should be
located at the level of the secondary speech act, the describing act, and that it
refers to aspects of the primary speech act. The primary speaker's intentions and
behaviour, the linguistic form, manner of presentation, hearer's attitude etc. and
the relations between these components all come in for evaluation (see Ver-
schueren 1984 and Rudzka-Ostyn 1988).
2.1 The basis of value judgements: donor concepts and scales
In the metaphors examined, two types of criteria are used for evaluating LA:
more concrete ones and more abstract ones. The more concrete criteria are
based on donor concepts; within the donor domains, certain types of behaviour
are approved or disapproved of on various grounds, and these evaluations are
then transferred to the recipient domain.
Apart from such criteria provided by concepts in the donor domains, there
are also more abstract criteria based on a number of VALUE SCALES.
2.1.1 Donor concepts

When we say that a particular metaphor contains a value judgement, this means
that the reporter (or speaker in the secondary speech situation, henceforth S2),
in using the metaphor, expresses a judgement of some aspect of the primary
speech situation. In other words, he evaluates some aspect of the linguistic
action referred to as positive or negative. The criteria on which such evaluations
of LA are based may be of a rather concrete type, such as social unacceptability
or aggression on the negative side, and skilfulness or sociability on the positive
side. These criteria are used to judge behaviour in both the donor and the re-
cipient domains. For instance, there are certain social rules which dictate what is
acceptable or unacceptable in the action of eating. Spitting out food (as spitting
in general) is socially disapproved of in our society, so that a negative value
judgement accompanies the LA verb spit out. In the domain of fighting, unfair
behaviour and tactics are objectionable, as in throw dust in someone's eyes.
Lack of skilfulness is valued negatively in the domain of manipulating objects,
so that fumble as a LA verb expresses disapproval, too.
The adjective silver-tongued expresses a positive value judgement of the
manner of speaking. The donor domain of manipulating objects here structures
the linguistic action metonymically as the manipulation of an instrument, more
specifically of the articulator 'tongue'. The positive evaluation is carried over
from the donor domain, in the sense that silver, as a precious metal, has positive
connotations in our society. (Cutlery is such an example where the silver in-
strument gets a positive evaluation as a marker of 'class'.)
The verb backbite, on the other hand, carries a negative evaluation of the
linguistic action. Coming from the domain of violent actions performed by ani-
mals, it expresses a strong disapproval by S2. The negative features which are
carried over from the donor domain are 'aggression', 'pain inflicted on the
victim', and the cowardly aspect of 'attacking someone from the back', i.e.
when he is defenceless. The link between donor domain and recipient domain is
that the mental harm and pain are viewed as similar to physical harm and injury.
2.7.2 Scales
The more concrete criteria illustrated above, however, do not always suffice as
an explanation for the evaluative meaning of metaphors. In addition, more ab-
stract criteria are relied on, based on VALUE SCALES. Johnson includes the image
schema of SCALE in his list (cf. 1.2.0. above), and points out that this structure
is "experientially basic" and "one of the most pervasive image-schematic struc-
tures in our understanding". (1987:123). Further, what is typical of scales is
their normative character: "Having more or less of something may be either
good or bad, desirable or undesirable" (1987:123). In what follows we use the
SCALE schema to account for value judgements expressed by the metaphors in
our corpus. A SCALE schema is based on our experience of 'more' or 'less', and
this 'more' of 'less' takes various more specific forms, depending on what is
measured. The SCALES which we found to be relevant are those of what we will
call 'INTENSITY', 'QUANTITY', 'FREQUENCY', 'SPEED', and 'DURATION'. These
scales interact whith the image schemata of FORCE, CONTROL, CONTAINER, PATH
and CONTACT We shall now take a look at how they operate in metaphorisation.
Intensity
As pointed out in 1.2.2. above, the image schema of FORCE is particularly salient
in the metaphors with violent actions or fighting as donor domains. Force can be
measured, so that a LA can be evaluated in terms of having stronger or weaker
INTENSITY. One may compare kick someone when he's down and rap someone
over the knuckles in this respect. The former metaphor involves more force and
refers to more violent action than the latter. The value judgement is also more
negative in the former, because of the use of the feet to kick someone who is
already defenceless. The feature 'aggression' combines with the feature 'unfair',
and with the high degree of force being exerted. The INTENSITY SCALE is also
salient in a metaphor such as pull one's punches, where the speaker is restrain-
ing his force. Here a weak degree of force expresses less violent linguistic action
than might be possible in the circumstances.
The INTENSITY SCALE is however not only relevant in the violent action
metaphors. It is also present in metaphors such as shut one's mouth. Here the
mouth as a container is shut, which implies a sudden forceful closing action. The
FORCE structure operating here is that of installing a BLOCKAGE: the closing of
the mouth raises a barrier to LA. The degree of force exerted is clearly stronger
than in the non-metaphorical expression speak no longer. The same applies to
keep one's mouth shut, which involves more deliberate force than e.g. not
speak, or remain silent. However, whether force in such examples is evaluated
positively or negatively is not so clear, and will depend on the context in which
they are used.
Similarly there may be different degrees of intensity in the making of a
speech act. As Johnson says (1987: 59): "The difference lies in the force with
which the sentence-container is thrust upon the hearer". For instance, a warning
may be mild or strong. A strong warning is made more forcefully. This, as John-
son says, is mainly a question of emphasis. In a number of metaphors in the
corpus, such degrees of force of the speech act are particularly salient. Good
examples are applaud and raise one's eyebrows, which both have nonverbal
communication as a donor domain. The meaning of applaud can be paraphrased
as 'express strong agreement with' (LDOCE). Whether the nonverbal act ac-
companies the linguistic action or not, it is clear that the latter is judged by S2
to be a forceful form of approval, agreement, support etc. In contrast, the meta-
phor raise one's eyebrows means 'express surprise, doubt, displeasure by or as if
moving the eyebrows upwards' (LDOCE). The speech act must here be a 'mild'
one, as can be judged from the fact that it would be inappropriate in a context
where S1 starts shouting, crying etc. to express feelings of displeasure.
In conclusion, strong and weak FORCE, measured on the INTENSITY
SCALE, are not positive or negative on their own. Degrees of force may however
combine with other concepts and thus acquire a positive or a negative value.
When force combines with violent behaviour, for instance, the negative value
judgement will be strengthened.
The bodily experience of CONTROL was suggested in 1.2.4. above as a
way of correlating the schemata of FORCE and BALANCE. In general, there are
three types of control:
• control over oneself, which is typically evaluated positively,
• control over others, which is typically evaluated negatively, and
• control over the environment, which is also typically valued positively.
Hence, what may be focused on in metaphorisation is the simple presence or
absence of control. A metaphor in which control is salient is flounder. The
donor domain of this metaphor is walking, and its literal meaning is 'move about
with great difficulty'. Non-fluent speech is referred to in terms of lack of control
of one's movements. The value judgement is hence clearly negative. In muzzle,
the presence of control is valued negatively, because it is exerted over others.
The donor domain of controlling animals adds a strong negative value judge-
ment to the expression, since a strong form of control of humans over other
humans in the way one normally deals with animals is disapproved of. Finally,
instances such as shoot one's mouth off illustrate salience of the third type of
control. The donor domain is the manipulation of objects, here a gun. One
should be in control of objects, and not shoot off a gun when it is not ap-
propriate to do so. In the metaphor, which refers to 'talking foolishly about
something one should not talk about' (LDOCE), lack of control is valued nega-
tively.
Where CONTROL involves FORCE it may also be graded in terms of FORCE
and hence entail an evaluation on the INTENSITY SCALE. For instance, rub
something in involves less force , and hence a weaker form of control than force
something down someone's throat
Quantity
The relevance of QUANTITY may be exemplified by full-blooded (argument,
style, etc.). In this metaphor, the donor domain is 'blood', as the source of life
and energy, and the positive value judgement of a forceful manner of speak-
ing/writing is related to the QUANTITATIVE SCALE
The SCALE OF QUANTITY is salient in a number of metaphors (all based on
the CONTAINER schema), though there is once again not a one-to-one rela-
tionship between much/little on the one hand and positive/negative on the other.
Clearly, when 'too much' is the meaning, the value judgement is negative, since
too means lack of balance. An instance of this is loudmouth. This metonymy
expresses a negative evaluation of someone who talks too much. High degree of
loudness (phonetic aspect) here expresses a form of 'obtrusive', hence disturb-
ing linguistic action.
In other cases, such as pour from/out, the value judgement is not clear. In
such utterances as (9), (10) and (11) below, the focus is on amount and speed,
combined with the physical schema of a CONTAINER. Whether S2 evaluates this
as positive or negative depends on other factors:
(9) Curses poured from his lips.
(10) Words poured out of her.
(11) She poured out her thoughts/feelings.
Frequency
As with the other scales, when extremes are focused on, this is valued nega-
tively. Examples are rattle-brained, rub something in and brainwash. In all
three, repeated talking is disapproved of. In rattle-brained, the brain is com-
pared to an instrument making a lot of quick little noises which however do not
mean anything. The adjective is used metonymically of a speaker who talks all
the time. The verbal rub something in refers, in its donor domain, to a repetitive
movement of the hands applying an ointment to the skin, until it is absorbed. It
is based on the CONTAINER, FORCE and CONTACT schemata, and the FREQUENCY
of the movement is required for effect. In the same way, repeated LA leads to
'absorption' of the message. It is valued negatively because the experi-
encer/listener loses control. Brainwash is very similar: again repeated LA is
valued negatively because it forces the listener to absorb the speaker's ideas.
Speed
LA may be perceived by S2 as being fast or slow. In such cases as pour out,
quantity and speed go together, but are not particularly approved or disap-
proved of. SPEED may however combine with a value judgement. Snap provides
a clear example. In the sense of 'say quickly, in an angry way' (LDOCE) it ex-
presses a negative evaluation of the LA, both through the feature 'violence'
(from the animal donor domain) and through the feature 'quickness': the unex-
pectedness and quickness make the victim defenceless. SPEED adds to the ag-
gression in this case.
In other cases, speed may imply fluency and hence be valued positively,
while slow speech is often associated with non-fluency and awkwardness. The
positive value judgement is apparent in have a ready tongue (ready here mean-
ing 'quick'), while the negative evaluation is clear in instances such as falter,
flounder, fumble.
Duration
DURATION interacts with FORCE, to the extent that the exertion of force may be
of long or short duration. An instance is chew the fat, which has a high value on
the duration scale. It may also relate to CONTACT, as in touch on, where the
contact is brief.
Further, the schema of CONTROL may interact with the schema of
CONTACT. In such metaphors, control may be graded in terms of DURATION. For
instance, have someone's ear involves permanent control and contact (see
1.2.5.). In contrast, go in at one ear and out the other implies a very brief con-
tact and hence almost total absence of control.
In c o n c l u s i o n , the SCALES of INTENSITY, QUANTITY, FREQUENCY, SPEED,
and DURATION may interact both with one another, and with a positive-negative
value judgement. A value judgement in terms of positive/negative is, however,
not a necessary corollary, and there is certainly no one-to-one relationship be-
tween say a high or low value on one of the scales and a positive or negative
value judgement.
2.2 Role of context
Metaphors expressing value judgements can be grouped into two main classes,
viz. those cases where the value judgement is context-independent, i.e. inherent
in the metaphor, and those where it is context-dependent.
2.2.1 Context-independent value judgements

Context-independent value judgements are prototypically either positive or
negative, no matter in what type of context the metaphor is used. (But see our
reservations expressed below). Fumble may serve as an example. In all con-
texts, this metaphor expresses a negative value judgement of the manner of
speaking. In the donor domain the verb refers to awkward movements with the
hands, hence inefficiency, and this evaluation is carried over into the LA domain
to refer to clumsiness in speech.
Another instance of context-free value judgement is found in drool. The
meaning of the verb in the donor domain, viz. 'letting liquid flow from the
mouth' (COBUILD) gives the metaphor its negative value judgement of 'talking
nonsense, talking foolishly' (LDOCE). The action in the donor domain is indeed
associated with imbecility, i.e. lack of CONTROL over one's own body, and
therefore meets with social disapproval. In the LA domain this negative evalua-
tion is always present.
2.2.2. Context-dependent value judgements

There are different types of context dependency to be distinguished.
In one type, the metaphor displays ambiguity, in the sense that it has two
clearly distinct meanings, a positive and a negative one. The context in which
the metaphor is used will make clear which interpretation is the appropriate one.
This is for instance the case with the noun jaw. This metaphor has two mean-
ings, viz. a positive one ('chat', as in (12)), and a negative one ('impudent talk',
as in (13)):
(12) I had a good jaw with Sally yesterday afternoon. (COBUILD)
(13) ...don't have to take any of his jaw. (Webster's)
In the donor domain, the jaw as an articulator has no value judgement attached
to it. In the metaphor, the focus is on one aspect of LA, viz. the peripheral as-
pect of the visible movement of the jaws in speech. By reducing LA to visible
articulation only, the reporter implicitly expresses that the content of the mes-
sage is not important. Absence of importance of subject-matter may be valued
positively, because it is part of informal conversation, the contact function of
language. On the other hand, this metonymicai expression may imply disap-
proval by focusing on manner (viz. strong movement of the jaws, perhaps due
to the loudness of speech) rather than on content.
It is interesting that both meanings are separately present in two other
metonymic expressions, viz. lip and chinwag. The noun lip means 'impudent
talk' as in (14), while chinwag means 'an enjoyable conversation' as in (15):
(14) I'll have none of your lip. (Webster's)
(15) I had a good chinwag with my sister yesterday. (COBUILD)
A second type of metaphor in this category is neutral in some contexts and
value-laden in other contexts. An example is mouthpiece. This metaphor may be
neutral, simply denoting a person who publicly states the opinions, policies etc.
of another person or organization. In sentences like (16), there is no positive or
negative evaluation:
(16) He became the official mouthpiece of the moderate leadership.
(COBUILD)
However, the metaphor is frequently derogatory. The negative value judgement
comes from the fact that the speaker is compared to an instrument used by
someone else. Thus, whereas the metaphor may be used in a neutral way, the
negative evaluation is clear in other contexts.
A third type of metaphor combines a positive with a negative value
judgement. The type may be illustrated by means of blow off steam. As ex-
plained in 1.2.4. above, the image schemata of FORCE and BALANCE (or
CONTROL) are particularly salient in this metaphor. The restoration of balance is
valued positively: by expressing one's feelings one restores mental balance. On
the other hand the comparison of a human being with a steam engine may ex-
press a certain amount of scorn, contempt, etc. Compare example (17):
(17) They can let off steam in pubs where nobody knows them.
(COBUILD)
This type is different from the type illustrated by jaw: whereas jaw is a case of
ambiguity, blow off steam is a case of merger. The two value judgements are
present at the same time, and the context may strengthen the positive or the
negative side of the coin.
Another example of the merger type is find one's tongue. Finding an ob-
ject which one needs for a certain purpose is a positive experience. Hence, the
ability to speak again (after the removal of psychological barriers) is valued
positively. However, finding implies previous loss, which is valued negatively.
Example (18) implies annoyance of the speaker, which may also be present in
(19):
(18) Lost your tongue ?
(19) Oh, you've found your tongue, have you. (COBUILD)
In (19), contextual factors reinforce the negative value judgement; the expres-
sion of scorn is salient.
It needs to be emphasized that the role of context - both situational and
linguistic - cannot be overestimated. First, even though a particular item proto-
typically expresses a positive or negative value judgement, pragmatic factors
may always reverse it. In utterances such as (20), the metaphor which negatively
refers to excessive control over someone, has a positive meaning for the
speaker, because of his ideology:
(20) Fortunately, we managed to brainwash him.
However, such factors do not detract from the inherently negative value judge-
ment present in the metaphor.
Secondly, the semantic-syntactic contexts in which the metaphors fre-
quently or typically occur may contribute to the value judgement. For instance,
pull punches and put a foot wrong nearly always occur in the negative, as in
(21) and (22):
(21) I didn't pull any punches.
(22) He never puts a foot wrong.
Although the former is inherently positive, and the latter inherently negative,
their most usual contexts, i.e. negatives, reverse the value judgement of the LA.
Another example is the metaphor spit out. The imperative (23) gives a positive
value judgement to the LA referred to (though the expected message still quali-
fies as 'difficult', 'possibly unwelcome'), as contrasted with an utterance like
(24), in which the value judgement is negative. Mood thus also contributes to
value judgement:
(23) Spit it out. (LDOCE)
(24) She spat out the hateful word. (Webster's)
2.3 Types of transfe r
There are two types of interaction between value judgements in donor domains
and the recipient domain of LA: the value judgement may be directly trans-
ferred, or it may be new.
2.3.1 Direct transfer

This means that the value judgements in donor domain and recipient domain are
both positive or negative. In other words, the value judgement expressed by the
metaphor can be explained from the value judgement in the donor domain. In
make a clean breast of something, for example, clean has a positive meaning in
the donor domain, and is associated with openness and virtue. The folk vision of
sin, lies etc. as weighing on someone's breast, and thus disturbing one's balance
is also salient here. By telling the truth, the dirt and weight are removed, and
one's balance is restored.
In throw dust in someone's eyes, the value judgement of aggression, in

this case a form of fighting in which one prevents one's opponent from seeing
and thus from defending himself, is negative. This evaluation is carried over to
the recipient domain, where verbal deception is seen as preventing someone
from understanding (or seeing).
Direct transfer is the most frequent type.
2.3.2 New value judgement

In the case of new value judgements the actions are neutral in the donor domain
but acquire positive or negative connotations in the recipient domain. Some
examples may illustrate this.
The donor domain of chew someone out is eating. In the donor domain,
chew as an action performed on food is neutral with respect to value judgement.
When applied to LA, however, the destruction of a person by verbal means
acquires a negative value judgement.
The donor domain of beat one's breast is nonverbal communication. The
gesture expresses sorrow. In the LA domain, the focus on the visual aspect
(nonverbal communication is visible) may suggest dishonesty: what is visible
may be in contradiction with one's intentions. Hence, exaggerated display of
emotions - the metaphor has a high value on the INTENSITY SCALE - is suspect.
(Cf. the definition in LDOCE: 'to make a noisy open show of sorrow that may
be partly pretence').
A final example is hand out (advice). In the donor domain, the verb is
neutral in such contexts as (25):
(25) He handed out the books to the pupils.
In the LA domain, however, handing out advice implies a superior attitude on
the part of the speaker, because the advice, being given freely, may not have
been asked for by the recipient, as in (26):
(26) I don't need you handing me out that sort of advice. (LDOCE)
It may be noted that no instances of reversed value judgement were found in the
corpus. A metaphor like brainwash, which at first sight looked like an example
(because wash in its sense of 'removing dirt', and thus 'making something
clean' is evaluated positively), was on second thoughts not classified as a case
of reversed value judgement. The negative value judgement must indeed be
ascribed to the incompatibility of 'brain' and 'wash'.
2.4 Links between the most important donor domains, image schemata, and
value judgements
2.4.1 Eating and feeding

As pointed out above, the donor domain of eating and feeding may structure the
LA in different ways.
One aspect focused on is eating as 'taking in', or 'incorporating' food.
The relevant body parts are then the mouth, the oesophagus, the stomach. In the
same way, LA may be seen as taking in information. Shared schemata are: the
body as CONTAINER, and the directionality of movement along a PATH. In both
domains, the movement is one from outside the container into the container.
Two metaphors which focus on LA as transfer of information are based on these
schemata, viz. feed information to someone and force/thrust/ram something
down someone's throat. In both, the speaker has control over the listener, but in
the former metaphor the degree of force is obviously much weaker than in the
latter. Because of the high degree of force used in the latter, the value judge-
ment is negative. A different type of metaphor which is based on the same
structures is regurgitate. The movement from within the container to outside it,
presupposes a previous taking in of food, information or ideas. Here, the re-
verse directionality is focused on, entailing the negative evaluation of a distur-
bance of the normal process: undigested food is compared to ideas which have
not been understood.
Another aspect of eating is taste. The experience of good and bad food is
valued positively and negatively, respectively. An example of a neutral metaphor
is flavour. A story may have a romantic, macabre, etc. flavour. The value
judgement will thus depend on the context, but the basic motivating factor is the
amount. (The focus is on little.) The adjective mealy-mouthed, on the other
hand, is clearly negative. This is an interesting example in that the negative
experience overrides the structure in the following way. Mealy is a negative
property of certain kinds of fruit: peaches, pears, etc. are mealy when the flesh
is powdery. Hence, the formal nature of the substance is the primary donor
domain. The negative experience of eating such kinds of fruit is the second
donor domain, and the link between eating and LA is explicit in the second part
of the word, viz. mouthed. However, the directionality is reversed in the meta-
phor, since 'unclear, ambiguous language' comes out of the mouth instead of
going into it, as food does. The aspects focused on are the lack of clear
boundaries (hence ambiguity, etc.) and the negative experience of tasting bad
food.
Thirdly, eating involves the destruction of food. In the same way, LA may
lead to the destruction of someone. In such metaphors, strong force exerted on
a person is valued negatively. An example is chew someone out.
Finally, eating may be focused on as an activity with a certain duration.
This can be valued positively (focus on the social aspect) or negatively (focus
on DURATION without much result). Both evaluations may be expressed in chew
the rag/fat, meaning 'have a conversation together' (positive) or 'complain
together' (negative).
2.4.2 Breathing
The link between breathing and LA is that air is an essential component of

speech. In waste one's breath the value judgement is negative, through the
negative evaluation of wasting an essential object. Conversely, save one's breath
implies an economical use, i.e. control, of one's resources, as in "I don't know
why I speak to my daughter. I might as well save my breath" (COBUILD). If
something takes your breath away, it renders you unable to speak, hence forces
you into silence. Again, then, the schema of CONTROL may play an important
role. In choke back, for instance, the speaker is seen as being in control. Hence,
it may carry a positive value judgement, as control over one's emotions is felt to
be necessary in certain contexts. (Cf. the example in LDOCE: "She tried to
choke back her anger, but failed").
2A3 Nonverbal communication

The most important SCALE in this donor domain is that of INTENSITY. Depending
on how much physical force is involved in the gesture, and on the extent to
which the gesture is salient, the speech act will be denoted as forceful or weak.
Weak intensity is salient in raise one's eyebrows, while strong intensity is inher-
ent in such metaphors as fall on one's knees, applaud, bow the neck, put one's
foot down, etc. In some, though not in all cases, strong intensity may imply
exaggeration when applied to LA. This seems to be the case in beat one's breast
and fall on one's knees. The explanation for the negative value judgement in
such cases is then that focus on visibility may imply dishonesty. (Compare also:
give someone the glad hand). Another relevant schema here, apart from FORCE,
is CENTRE-PERIPHERY: what is visible is 'outer', and may be different from what
is 'inner'. (cf. Johnson 1987: 124ff.). It may be noted in passing that the
CENTRE-PERIPHERY schema is also particularly salient in metaphors referring to
the expression of feelings, in which the heart is seen as the centre of our inner
feelings: bare one's heart, mbosom, hearty, heartwarming.
2.4.4 Violent actions

LA seen as aggressive behaviour may be referred to by metaphors which either
have animal behaviour or human fighting as donor domains.
In either case, the schema of CONTROL exerted over someone else is sali-
ent. The degree of force, inversely correlating with the degree of self-control of
the victim, is responsible for the extent to which the activity is experienced as
aggressive, and hence valued negatively. For example, rap someone over the
knuckles is less violent than kick someone when he's down. Similarly, restraining
one's force is valued positively, as in pull one's punches and spar. Most fre-
quently referred to in these metaphors are the speech acts of criticizing, scolding
and blaming someone.
Another type of metaphor is that where human treatment of animals is
donor domain, as in muzzle. Here again, excessive control over another person
is valued negatively.
2.4.5 Manipulating objects

One aspect focused on in these metaphors is awkwardness, inefficient handling
of objects (the CONTROL schema). The negative evaluation may relate to manner
of speaking, i.e. non-fluency, as in fumble (for the right word) (see above
2.2.1.), or to LA as obeying social norms. Violation of social norms is salient in
e.g. heavy-handed.
Secondly, hands are used for giving something, i.e. transferring an object
from one person to another one (the PATH schema), as in hand it to someone,
hand out (advice), hand down (information). These metaphors are neutral,
except for hand out advice, where control is again exerted: unrequested advice
is forced upon one.
2.4.6 Walking
LA as compared with movement along a path, is valued negatively in the fol-
lowing two cases.
First, awkward movement is related to the schemata of CONTROL and

BALANCE. Lack of self-control is negative. Again, as with the metaphors based
on the manipulation of objects, the evaluation may relate to the manner of
speaking (non-fluency) or to social norms. Examples are: slip of the tongue,
falter, flounder, never put afoot wrong, put one's foot in it
Second, the movement back along a path one has just covered, is seen as
purposeless: the previous activity has been useless. This schema is present in
backtrack and backpedal.
3. Summary, discussion and further perspectives
In this final section we shall summarize the main findings of our research and
point out which aspects need further exploration.
Although the main principle underlying the collection of our corpus was
the explicit or implicit reliance of the metaphors on the functioning of the body,
it is clear that there are also important differences cutting across the expressions
composing the material.
• One distinction, mentioned in 1.1. above, is that between metaphors based on
(i) those body parts used in both speaking and some other activity, and
(ii) those body parts not present in LA, such as feet, hands etc.
Within each of these classes, further distinctions can be made. In class (i), a
distinction can be drawn between metaphors where another domain is clearly
relevant (e.g. eating, breathing) and those where no such other domain seems
to be involved. An instance of the latter type is open one's lips. However,
this type is, in our view, not merely metonymicai, because at a higher level
transfer does indeed take place. In the case of open one's lips, for instance,
the body is seen as a container for language. The focus on language as com-
ing out of a container (the speaker's body) and going into one (the listener's
body) links a large number of at first sight unrelated expressions. At this
higher level, such expressions as open one's lips, shut up, pour out, rub it in,
have a word in someone's ear, etc. can all be shown to be related conceptu-
ally. In class (ii) above, a further distinction can be made between those ex-
pressions where body parts are used in nonverbal communication and those
where body parts are used in some type of non-linguistic (inter)action (e.g.
fighting, walking, etc.).
• A second major distinction can be drawn between expressions also used to

refer to non-linguistic action (e.g. muzzle a dog vs muzzle a person) and ex-
pressions which can only refer to LA (e.g. backbite, catch someone's ear).
The fact that the latter type of items can only refer to LA does again not
mean that they are not metaphorical: they can only be understood with refer-
ence to a more general metaphor, such as 'LA is violent behaviour' (for
backbite), or 'LA is physical CONTACT between speaker and listener' /
'catching someone is CONTACT and CONTROL' (for catch someone's ear).
• A third major distinction which has appeared from the corpus is that between
those metaphors which can be explained from very specific donor domains
and those having less specific donor domains. However, this is not a
clear-cut distinction but rather a question of variation on a scale: metaphors
can be situated on a scale from more general to more specific donor domains.
For the most specific ones, encyclopaedic knowledge is essential. Examples
of such expressions are: get it straight from the horse's mouth
(encyclopaedic knowledge of customs in the horse trading business is re-
quired), kick against the pricks (cattle breeding context), tear someone off a
strip (military context). In such metaphors the body part domain is only
marginally important. What is interesting about them is that, although in
terms of specific donor domains they may at first sight look like isolated ex-
pressions, they actually fit into more general ways of conceptualizing LA,
viz. 'exchange of information understood in terms of PATH' (get it straight
from the horse's mouth), 'LA as FORCE' (kick against the pricks), and 'LA as
FORCE and CONTROL' (tear someone off a strip).
• In comparison, expressions such as eat one's words or kick someone around
have donor domains which are less idiosyncratic: both eating and fighting
are, as shown, frequently drawn upon as sources of metaphorisation of LA.
These expressions can be understood, without encyclopaedic knowledge,
with reference to the general conceptual metaphors 'LA is eating' and 'LA is
violence'.
• Still further away from the specific context extreme are expressions such as
get something off one's chest, unbosom, bare one's heart. These cannot be
related to a specific donor domain. They are understood with reference to
general folk models of the location of feelings and their effect on one's body
(see e.g. Kövecses 1986 and Lakoff 1987). Because there is no specific do-
nor domain, the image schemata become more directly relevant: BALANCE,
PATH, CONTROL, CONTAINER, CENTRE/PERIPHERY.
• Finally, expressions such as put words into someone's mouth, take the words
out of someone's mouth have no donor domains except the most general
ones of image schemata: 'the body as a CONTAINER' and 'CONTROL' (see
Pauwels, this volume, for a detailed discussion of such metaphors).
The conclusion we draw from these types of expressions is that the image
schematic approach to metaphorisation is essential, both for establishing links
between metaphorical expressions originating within a particular domain and for
generalizing over domains. However, some metaphors seem to work only on
this most general level, while others, though ultimately also based on image
schemata, depend more directly on very specific domains of experience. It fol-
lows that the relevance of these different levels of abstraction may vary from
expression to expression, depending on its position on a scale. The exact rela-
tionship between domains and image schemata needs further investigation.
Value judgements are among the most important factors that facilitate
metaphorical transfer. As pointed out, they may either simply be explained from
more specific donor domains or they may, in addition or exclusively, be based
on a number of SCALES. In other words, donor-based judgements also involve
scales but there are judgements which seem to be based on scales only. Since
'too' is negative, talking too much, too slowly, too long, too frequently etc. is
valued negatively: e.g. rub something in derives its negative value judgement
from 'too frequently'. It was further emphasized that scales may be relevant
without involving a value judgement: in touch on {a subject) the brevity of the
contact is relevant in the metaphorisation, but the expression is not inherently
positive or negative.
Only when we show how all the factors mentioned (i.e. the relevant image
schemata, domains, scales, value judgements) interact, can we capture the intri-
cacies of metaphorisation.
It is suggested in this paper that the schematic framework proposed by
Johnson (1987) may have to be modified and/or refined. One such refinement
involves the introduction of the CONTROL schema. Presence or absence of con-
trol and the type of control indeed appear to be important variables determining
the value judgement. Its exact status in terms of links with the other schemata
still needs to be examined, but so does, in our view, the status of Johnson's
schemata. It seems possible, for instance, that some schemata are more basic
than others, or, in other words, that there may be a hierarchy of schemata. Fur-
ther, we have found that the image schema of SCALE is indeed pervasive in that
it interacts with the other relevant schemata to make metaphorical expressions

value-laden.
Finally, the notion of value judgement itself needs closer examination. In
the first place, stricter criteria will have to be set up for distinguishing between
value judgement introduced in the metaphorisation process on the one hand and
value judgement pronounced on the primary speech act, independent of the
metaphor, on the other hand. This involves, for instance, distinguishing between
the negative connotation of scold someone (non-metaphorical) and that of rap
someone over the knuckles (metaphorical). Secondly, it would be interesting to
integrate value judgements in a larger pragmatic framework, relating them to
factors such as politeness and cultural or sub-cultural values.
Acknowledgements
We are grateful to George Lakoff for his suggestions at the Symposium on

Cognitive Linguistics, Duisburg 28 March - 1 April 1989, where we gave an
oral presentation on this topic. We remain, clearly, solely responsible for the
contents of this article.
Notes
1. The notion of domain is used here as in Rudzka-Ostyn (1988a: 509).

2. Rudzka-Ostyn (1988: 512) uses this terminology to distinguish between the describing
act (secondary speech act) and the described one (primary speech act).
3. The term "articulator" is used in the widest sense here, to include all the body parts
involved in speaking.
4. Apart from the Longman Dictionary of Contemporary English (1982) (LDOCE), a
number of other dictionaries were used for the clarification of definitions and for exam-
ples, where Longman did not provide them. They are: The Oxford English Dictionary
(1933, 1972-1986) (OED), Hornby's Oxford Advanced Learner's Dictionary of Current
English (19743) (OALDCE), Collins COBUILD English Language Dictionary (1987)
(COBUILD) and Webster's Third New International Dictionary of the English Language
(1971) (Webster's).
5. Goossens (1990b) provides a more thorough explanation of the interaction between the
metonymicai and the metaphorical in these expressions, on the basis of a similar corpus
of examples.
6. Guard one's tongue, tonguetied and bridle one's tongue have been classified here, rather
than under 1.1, since the tongue only provides the link with the LA domain, while the
main metaphorical element is clearly the absence of movement/speech.
7. The term "experiencer" is used here as a cover term for listener, addressee and patient.
8. Another possible interpretation for this metaphor was suggested by L.Goossens. Birds of
prey also pick the brains of their prey. The container schema is still salient in this inter-
pretation.
Assessing Linguistic Behaviour
A Study of Value Judgements
Anne-Marie Simon-Vandenbergen
University of Ghent
0. Introduction
In Pauwels & Simon-Vandenbergen (this volume) linguistic action (LA) meta-

phors in which body parts play a central role are examined for the kinds of value
judgements they express. The conclusions we arrived at are the following:
• Value judgements may be of two types, viz. those based on specific and
concrete donor domains, and those based on the scales of quantity, fre-
quency, speed, duration and intensity, which may be regarded as very ab-
stract domains. These scales are relevant in different concrete domains.
Evaluations which refer to positions on one of the scales may be positive or
negative but - it is argued - they are not necessarily so. For instance, LA may
be judged to be slow, which means it has a low value on the scale of speed.
Whether this is positive or negative depends on the individual metaphor.
• Value judgements may be context-dependent or context-independent. Value
judgements which are context-independent are inherent in the meaning of the
item. An example is to backbite, which prototypically expresses a negative
value judgement, in the sense that it will be easily and fairly consistently
categorised as a negative word (see e.g. Jaeger 1986 for this use of the term
'prototypical'). With other expressions the context will favour a positive,
negative or neutral interpretation of the item. An example given by Pauwels
& Simon-Vandenbergen is mouthpiece. However, pragmatic factors - it is ar-
gued - may always override a prototypical value judgement, as in Fortu-
nately, we managed to brainwash him. In this example, brainwash, which
prototypically expresses a negative value judgement, is used in a context
which is positive from the speaker's point of view.
72 ANNE-MARIE SIMON-VANDENBERGEN
• Value judgements may be directly transferred from donor domains, or they

may be 'new' in the metaphor.
Pauwels & Simon-Vandenbergen conclude, firstly that it is necessary to distin-
guish more clearly between value judgements of LA which are independent of
metaphorisation and those introduced by the metaphor; secondly, that it would
be interesting to relate such value judgements to cultural and sub-cultural val-
ues.
This paper picks up some of the points touched upon in Pauwels & Si-
mon-Vandenbergen. In particular, it attempts to give an answer to the following
questions:
• To what extent do lexicalised metaphors of LA reflect value judgements of
speakers of English?
• Which donor domains are typically used for the expression of value judge-
ments?
• How relevant/useful are the scales referred to in Pauwels & Simon-Vanden-
bergen?
• How realistic is the dichotomy context-dependent v. context-independent
value judgements?
1. Approach
Whereas Pauwels & Simon-Vandenbergen examine a corpus of LA metaphors

whose donor domain is that of 'body parts', the corpus used for this study is not
restricted with regard to donor domain but with regard to recipient domain.
Only those metaphors have been selected which refer to specific aspects of lin-
guistic behaviour. The following components of the LA scene are looked at:
• 'linguistic interaction' (including turn-taking and topic management), and
• 'manner of speaking' (including speed, loudness, quantity, repeated linguistic
action and directness).
The data comprise the following items:
• metaphors from the Longman Dictionary of Contemporary English corpus
(cf. Vanparys, this volume) referring to the specific aspects mentioned above;
• metaphorical and non-metaphorical expressions in the same semantic fields
taken from Roget's Thesaurus.
ASSESSING LINGUISTIC BEHAVIOUR 73
The non-metaphorical expressions were added in order to compare value

judgements attaching to them with those attaching to metaphorical items with
similar meanings.
Value judgements are decided on in two ways, viz. on the basis of dic-
tionaries and on the basis of questionnaires.
Dictionaries have been used in the following way. Definitions and exam-
ples have been looked up in the Longman Dictionary of Contemporary English
1978 {LDOCE), Collins COBUILD English Language Dictionary 1987
(COBUILD), the Oxford Advanced Learner's Dictionary of Current English
1914 (OALDCE), and The Concise Oxford Dictionary of Current English 1976
(CODCE). The definitions and illustrative examples are taken as evidence of
value judgements attaching to those items. In some cases, dictionaries use ex-
plicit labels such as derogatory, appreciative, humorous. In other cases the
definitions themselves contain items with a positive or negative value, such as
foolish (talk), friendly (talk), empty talk), unpleasant (sound), etc. Thirdly,
examples may at least point to possible value judgements in some contexts,
and/or to possible collocations. The treatment of the non-metaphorical item
chatty in the LDOCE and COBUILD may illustrate this. Compare:
chatty LDOCE fond of talking I wish she weren't so chatty; she's always
annoying me when I'm trying to read. COBUILD fond of talking in a
friendly informal way' She wrote a nice, chatty letter, full of news and gos-
sip.
Whereas 'fond of talking' is neutral with respect to evaluation of linguistic
action, it is clear that the word can have both positive and negative connota-
tions.
Intuitive reactions of native speakers of British English to a number of
items were tested by means of questionnaires.
Three types of questionnaires were used, which I shall refer to as QI, QII
and QUI. All three types were presented to first-year undergraduate students at
the School of English of the University of Newcastle-upon-Tyne.
• Questionnaire I (presented in the academic year 1989-1990) consists of two
sets, each containing both metaphorical and non-metaphorical items. One set
(A) was presented to twenty students, the other one (B) to another group of
twenty students. The complete questionnaire I is added in the Appendix as
Tables A1-A7 and B1-B7. As can be seen, the items are decontextualised,
and no justification of answers is asked for.
• Questionnaire II was presented to only seven students in the academic year

1990-1991. It consisted of a list of sixty items which informants were again
asked to evaluate as positive, negative or neutral. In addition, however, in
each case they were requested to motivate their answers (Why?), and to use
the item in a sentence. The results of QII are - given the small number of in-
formants - of no statistical value, but they do provide some clarifications of
value judgements given in QI. I shall occasionally refer to these data.
• Finally, a third questionnaire, which contained ten items, was presented in
1991-1992. It was completed by 10 students. All the items were metaphori-
cal expressions, which informants were asked to evaluate. In order to find
out more about underlying motives and associations I again (as in QII) asked
informants why they evaluated the item in a particular way, and to give a
sentence illustrating the use of the expression. QUI can be found in the Ap-
pendix, too.
2. The value of the questionnaires
QI is valuable in that it gives an idea of intuitive reactions of informants to the

items presented. If there is a high degree of agreement among informants, we
can conclude that the item in question evokes primarily positive or negative
connotations. On the other hand I am fully aware of the limitations of this way
of eliciting information. Before looking at the data (section 3 below) it is there-
fore useful to discuss the way in which results of the questionnaires should be
interpreted.
First of all, it is obvious that twenty informants make up a fairly small
group. Moreover, in a number of cases some informants left all slots blank, for
reasons unknown to me. Perhaps they did not know the item or they were not
sure what to answer, or they simply overlooked the item. In any case, the result
is that in those cases we have only 19 to 16 answers. (Tables A6 and A7 were,
incidentally, completely skipped by one student). However, percentages have
always been calculated out of 20, so that if for instance an item is said to have
been evaluated as positive by 80% of informants, this actually means that 16
students ticked it as positive (while, say, 2 may have ticked the neutral box, and
2 may have left all slots blank). As can be seen in the Appendix, percentages are
added for clarity's sake where ten or more informants filled in a particular slot.
It seems to me that percentages between 100 and 70 are significant, i.e.

indicative of general trends. However, significance is clearly scalar: items which
receive 100% agreement are more clearly associated with particular value
judgements than those which get, say, 70% agreement.
Also interesting are cases where an item gets answers divided over two
slots only (e.g. positive and neutral, but not negative). The total number of
items evaluated as positive, negative or neutral in QI is 125. The results - in
terms of agreement among 20 informants on their 'value' - are given in Table 1.
Table 1: % of agreement among informants in QI
%of %of %of %of

agreement the items agreement the items
100 1.6 70 10.4
95 2.4 65 10.4
90 8 60 9.6
85 8 55 7.2
80 8 50 10.4
75 9.6
The figures indicate that between 100% and 70% agreement was reached in
48% of the cases; and that at least 50% agreement was reached in 85.6% of the
cases. The two items on which all informants agreed were fluent and expand on
a subject. Both were judged to express a positive evaluation, the former expect-
edly so, the latter slightly more surprisingly so. We shall come back to fluent
below. At this stage, the unanimous agreement on expand on a subject needs to
be examined because it points to a second 'limitation' of the questionnaire.
It is possible that in some cases the order in which items were presented
may have influenced the answers. The fact that all informants judged expand on
a subject as positive, may have been caused by the fact that it occurred in a list
of definitely (more) negative items such as keep on about a subject, be on about
a subject, carry on about a subject, etc. (See Table B3). This may explain why
it is considered a positive rather than a neutral item, which it appears to be,
judging from dictionary definitions. (See e.g. COBUILD 'give more information
or details about it when you write or talk about it'). This explanation finds some
support in the results for go into a subject, which, whereas I also expected it to
be judged as neutral, only received 3 ticks in the 'neutral' box, while 85% of
informants saw this expression also as a positive one.
On the other hand, the explanation does not account for the fact that go
into a subject does not reach a 100% agreement. Perhaps such small
'deviations' must be allowed for in this kind of questionnaire and interpreted as
a matter of coincidence, perhaps there is an explanation in terms of the kinds of
contexts with which informants associate items. In the case of expand on a
subject, it seems to me possible that its unanimously positive evaluation may
have something to do with the academic context in which students meet the
word, especially in comments on essays, in interviews, etc. in the exhortative
form: 'Expand on this!' or 'Could you expand on that?'. In other words, ex-
panding on a subject is positive because it suggests that expectations are met.
Further, the person who can 'expand' is knowledgeable and able to present
information in a structured way.
That the order in which items are presented may influence answers, how-
ever, is further apparent in the reactions to an item which was presented twice,
i.e. once in set A and once in set B. The expression stick to a subject was pre-
sented as no. 10 in Table A5 and as no. 10 in Table B5. As the figures show, an-
swers were very much divided over positive and neutral in A5 (9 positive - 8
neutral), while 90% of informants judged it to be a positive item in B5 (i.e. 18
positive v. only 2 neutral). The explanation may be that the preceding item in B5
was meander, which is almost like an antonym of stick to a subject and which
was judged rather negatively. The preceding item in A5, viz. tackle a subject, is
semantically less close to stick to a subject and would therefore not cause this
kind of reaction.
On the other hand, it seems to me important that neither in A5 nor in B5
was the item judged to be negative (only 2 ticks in A5, and none in B5). I there-
fore believe that the general trends are significant - on condition the results are
interpreted with caution.
Another item which occurred twice was chatty, this time in the same set
(A4), once as no. 2, once as the very last item, no. 20. This means that in this
case the same informants gave their opinion twice on the same item. The figures
show that the answers are largely, though not completely, the same: 85% posi-
tive v. 75% positive; 3 neutral answers in both cases; 1 negative and 1 blank in
no. 20. In other words, my informants on the whole primarily thought of chatty
as a positive word, but it is obvious that one may also think of contexts in which
it acquires a negative meaning. (See the example in LDOCE, quoted in section 1
above, and see further below).
A third shortcoming of QI is that we have no information on the infor-

mants' motivation behind their answers, nor on the way they interpreted the
choices. What, for instance, did they mean by 'neutral'? What kind of contexts
were they thinking of when they answered in one way or another? In order to
find out more about such things, I set up questionnaires II and III, which will be
referred to in the discussion of the data.
Before looking at the data, I shall now examine in some detail a number of
items presented in QI, especially with a view to clarifying the notions of con-
text-dependent and context-independent value judgements introduced in Pau-
wels & Simon-Vandenbergen.
Pauwels & Simon-Vandenbergen (this volume, 2.2.1) use the term
'context-independent' value judgements to refer to value judgements which 'are
always either positive or negative, no matter in what type of context the meta-
phor is used'. We there give examples such as fumble (negative), drool
(negative). Context-dependent value judgements, on the other hand, are said to
be those which vary with context. There are three types of context-dependency,
viz. ambiguity (e.g. jaw), merger (e.g. blow off steam) and metaphors which are
neutral in some contexts and value-laden in others (e.g. mouthpiece).
Further, Pauwels & Simon-Vandenbergen point out that prototypical value
judgements may always be overridden by pragmatic factors. We shall now exam-
ine the above notions by looking at some examples.
The word fluent (Table A4) was judged by all informants to express a
positive value judgement. This means that a sentence such as 'He's a fluent
speaker' expresses a positive quality of the subject rather than a criticism. Dic-
tionaries define the word as follows:
fluent COBUILD ... speaks smoothly and clearly with no hesitations or
mistakes LDOCE speaking ... in an easy smooth manner
Smoothness, ease and clarity are characteristics of speech which are evaluated
positively in our society because they show skill on the speaker's part and make
speech pleasant to listen to from the hearer's point of view. The following two
utterances are therefore pragmatically odd:
(1) Unfortunately he's fluent.
(2) His main shortcoming is his fluency.
However, as pointed out by Pauwels & Simon-Vandenbergen, the prototypical
value can always be overriden through marked contextualisation, as in:
(3) I hate his fluency (because I'm jealous of it).
One may compare:

(4) Unfortunately, she's beautiful and clever. (Because she possesses
these two characteristics she's dangerous).
In other words, items such as fluent seem to be prototypically positive, although
the overall judgement may be negative, due to contextual factors. Still, the term
'context-independent' value judgement applies to such words precisely because
of their inherently positive or negative connotations. The unanimous agreement
of informants is a signal of this prototypical value judgement.
Let us compare this with an item such as chatty. As pointed out above,
the item was judged by the majority of informants to be positive, though a few
considered it neutral, and one person valued it as negative (Table A4, nos.
2,20). This means that - unlike fluent - this word has different connotations with
different speakers. As indicated above (section 1), both COBUILD and LDOCE
use the terms 'fond of talking' in their definitions, though COBUILD adds 'in a
friendly informal way'. It is likely that COBUILD's addition of the element
'friendly' accounts for the predominantly positive evaluation.
How do we explain the remaining answers? LDOCE's example gives us a
clue: T wish she weren't so chatty; she's always annoying me when I'm trying
to read'. In other words, there are situations where one does not like to be dis-
turbed even by friendly talk. In addition, it very much depends on one's per-
sonality whether 'chattiness' is appreciated most of the time or not. This ac-
counts for the negative answer, and also for the neutral ones, which probably
mean that the value judgement can go in either direction.
Hence, sentences such as the following are not at all odd:
(5) She's nice; very chatty, though.
(6) He's a bit too chatty, but O.K.
(7) She made a good impression, though she's very chatty.
In contrast with example (7), the following is pragmatically doubtful:
(8) She made a good impression, though she's very fluent.
In conclusion, the value judgement of chatty seems to be mostly positive, al-
though a negative interpretation is not excluded. To some speakers the term
may even primarily be negative (depending on their personality), to others con-
textual factors may favour a positive or negative interpretation. The positive
meaning element is 'friendly and sociable', the negative one is 'time wasting'.
To the extent that results of QI give quick intuitive reactions, the basically posi-
tive value judgement attached to chatty is significant.
In comparison with chatty, expressions such as speak plainly have less

obvious connotations. This expression was judged to be positive by 65% of
informants and neutral by 35% (Table B6). Here we are dealing with a much
lower degree of agreement and the question arises how we should interpret such
results.
Dictionaries define the item as follows:
(speak/say) plainly LDOCE in a plain manner (plain is defined as 'clear,
easy to understand') COBUILD say something in a way that is easy to un-
derstand and cannot be mistaken
These definitions point to a positive rather than a negative value judgement,
although COBUILD's definition of plain statements hints at the possibility of a
negative meaning: 'direct and easy to understand and perhaps not very polite'.
Whether subjects who rated speak plainly as neutral were thinking of 'neither
positive nor negative' or of 'positive or negative, depending on the context' is
impossible to say. On the other hand it seems to me important that the negative
slots were empty for use plain words, speak plainly, tell someone in plain
terms, tell someone straight and give a straight-out answer (Tables A6 and B6).
This points to a basically positive evaluation of directness of speech by my
informants. Directness is thus primarily associated with the positive features of
clarity and honesty, and only secondarily with lack of politeness. This means
perhaps, that young British people (aged 18-19) value frankness more than
tactfulness (see further section 3.2.5).
A different case again is blow off steam, which was judged by 60% as
positive, by 3 people as negative and by 5 as neutral (Table B7). This metaphor
is said to be a case of 'merger' in Pauwels & Simon-Vandenbergen. This means
that 'two value judgements are present at the same time and the context may
strengthen the positive or the negative side of the coin' (2.2.2). The positive
element is 'restoration of balance', the negative one is 'lack of control'. These
two meanings are clearly present in answers given by informants in QIL On the
positive side they gave explanations such as 'release of pent-up energy', on the
negative side they wrote 'suggests impatience' and 'out of control, but fairly
harmless', while those who ticked neutral motivated their answer as 'can be
necessary at times' and 'could mean to rid oneself of negative feelings on a
subject or to have an argument'. It is clear that one's evaluation depends on the
extent to which one thinks the restoration of balance is more or less important
than self-control.
In conclusion, the distinction between context-independent and context-

dependent value judgements which Pauwels & Simon-Vandenbergen make is
useful to the extent that some items are prototypically positive or negative,
while others have connotations which either depend on their use in a particular
context or on the speaker's individual value system and personality. On the
other hand a scalar view is more realistic than a hard-and-fast distinction be-
tween two types: some items need less context than others in order to evoke
positive or negative associations. The percentages in the tables of QI are a clear
indication of the position of an item on that scale from 'context-independent' to
'context-dependent' (see also section 6 below).
3. The data
3.1 Linguistic action as interaction
In this section I shall look at expressions referring to the social function of talk,
to various aspects of turn-taking, and to various ways of dealing with a topic.
3.1.1 The social function of talk

Establishing and/or maintaining contact is the primary function usually attrib-
uted to 'casual conversation'. Cloran (1987: 90) formulates it as follows:
The general purpose of participants in casual conversations with friends or
indeed with strangers is sociability (...). It is possible, then, to view the
context of casual conversation as being 'about' social relationships - their
establishment (...) or their maintenance (...).
Conventions to be observed by participants in order to preserve 'an easy bal-
ance of power' (cf. Good 1979) are, according to Good:
(i) equal access to speaker and hearer roles;
(ii) equal distribution of such speech roles as that of questioner;
(iii) the introduction of topics to which the conversational others can address
themselves;
(iv) the avoidance of unfamiliar or specialised vocabulary.
Conventions (i) and (ii) are further looked at in section 3.1.2. below, conven-
tions (iii) and (iv) have one implication which seems to be relevant in the present
context, viz. that the subject-matter of pleasant conversation is 'light', 'non-
specialised'. The questions which arise here are: Firstly, do people evaluate this
type of discourse (i.e. talking for the sake of contact) as positive, negative, or
neutral? Secondly, what types of metaphors refer to conversation?
A list of twenty nouns all referring to verbal events was presented to the
twenty informants, who were asked to evaluate the verbal events as positive,
negative or neutral. The answers show a number of interesting tendencies (cf.
Table Al):
Table 2
positive neutral
conversation: 50% 50%
talk: 50% 50%
dialogue: 35% 65%
(1) The nouns which one would intuitively classify as 'neutral', viz. conversa-
tion, talk , and dialogue also get a fairly high score for 'positive' (cf. Table 2). It
appears that the more formal word dialogue, which also refers to a more formal
occasion, gets a higher percentage of 'neutral' evaluations than the more every-
day words. It is further noteworthy that the word colloquy (60% neutral) had 2
ticks under 'negative' as well. This may be due to the formality of the event
referred to. On the other hand it is possible that a number of the students had no
sharp intuitions about the word. In any case, there is nothing in the definition
given in dictionaries to predict a negative value judgement (cf. LDOCE: 'a
formal conversation'; COBUILD: 'conversation or meeting; a formal word'),
unless formality has negative connotations to some subjects. It is important,
though, that the other three words had empty slots for 'negative'. (Dialogue:
35% positive; talk/conversation: 50% positive). This shows that, if a value
judgement is attached to these words - out of context - it is a positive one.
Hence, the mere act of 'talking, having a conversation' is felt to be a positive
one, possibly because it is a signal of non-inimical relations. (Compare: not to be
on speaking terms).
(2) Words which scored higher for 'positive' are listed in Table 3. What they
share, as opposed to the four discussed above, is reference to 'informality', and
'lightness of topic' (talking about unimportant things), which are felt to be
pleasant.
Table 3 Table 4
positive light shallow

chat: 80% chit-chat: 100% 85%
chit-chat: 60% natter: 95% 85%
chinwag: 55% chat: 95% 60%
natter: 55% chinwag: 90% 80%
In order to test this hypothesis, the same twenty words were presented to an-
other group of twenty informants, who were asked to evaluate the verbal events
referred to as: a) light, heavy, neither; and b) as: deep, shallow, neither.
(Appendix, Table Bl). The results were the following for the above words:
(i) Colloquy and dialogue are the only two words which have more than 50%
answers for 'heavy' and 'deep' (Colloquy: 70% heavy; 65% deep; dialogue:
90% heavy; 90% deep). This seems to indicate that it is the 'seriousness' which
gives these words a less positive connotation than talk and conversation (cf.
talk: 95% neither heavy nor light; 75% neither deep nor shallow; empty boxes
for light and shallow; conversation: 55% neither heavy nor light; 55% neither
deep nor shallow; empty boxes for light and shallow.) These figures show that
'light' and 'shallow' are not associated with these four words, and that talk is
the most neutral one in these respects.
(ii) In contrast, the more positive words chat, chit-chat, chinwag and natter are
associated with 'light' and 'shallow' (cf. Table 4). These figures show two
things: first, that lightness of subject-matter may be valued positively; secondly,
that chat, which scores lower for 'shallow' than the other three positive words,
is the most positive word of all (80%). If one compares chat and chinwag, there
is nothing in the dictionary definitions which would explain a less positive value
judgement in chinwag (55% positive). It is defined in LDOCE as 'informal
conversation; chat'; and in COBUILD as 'long enjoyable conversation between
friends'. Hence, the slightly more negative connotation must come from the
feature 'shallowness', which is less pervasive in chat (cf. further on the meta-
phor chinwag). This hypothesis was confirmed by QII and QUI. Students who
evaluated chinwag, chit-chat, and natter as negative words all referred to the
trivial, inconsequential nature of the activity. It further appears from QII and
QUI that the absence of a high degree of agreement shown up by QI must be
explained from both the context and collocations that come to informants'
minds first, as well as from their sub-culture and personalities. Examples given
by those evaluating the above words as positive were such as:
(9) Let's have a good chinwag.
(10) We must meet and have a chinwag.
(11) They sat down for a chit-chat.
(12) We had a good old natter this morning.
Typical collocations of chinwag seem to be, judging from the examples: a good
-, a nice -, a jolly old -.
Examples given to illustrate the negative connotation of these same words
were such as:
(13) I want positive ideas, not chit-chat.
(14) Don't waste your time on chit-chat.
(15) This parish magazine is full of nothing but chit-chat.
Strikingly, an illustration of the negative connotation of chinwag was the fol-
lowing:
(16) Ethel came over for coffee and we had a bit of a chinwag.
Although the example does not make clear the negative connotation, the stu-
dent's comment does: 'It's a bit old-ladyish'. Such evaluations of words and of
the contexts they are associated with point to the relevance of the sub-culture.
In this case, age seems to be an important factor. Other instances will confirm
the role played by the sub-culture.
(3) Words which combined the features 'negative' - 'light' - 'shallow' were, in
descending order of 'negativeness': prattle (80%), babble (75%), yap (75%),
idle talk (70%), trivial talk (70%), gossip (65%), tattle (65%), small talk
(60%), jaw (50%). It is clear, however, that the relevance of the features 'light'
and 'shallow' varies: in prattle, for instance, they are very salient and directly
relevant to the negative value judgement (See Appendix, Table B1: 90% light;
85% shallow); in contrast, jaw scores only 55% on both 'light' and 'shallow',
which means that the negative judgement may be due to other factors. (On the
metaphor jaw, see further below).
In any case, it appears that 'light' and 'shallow' can go together with
positive or negative evaluations. 'Shallowness' in a more extreme form may
indeed become 'foolishness', 'emptiness'.
Let us now have a closer look at the metaphors in this set, viz. chinwag,
jaw, small talk, cackle, yap (Tables Al and Bl in the Appendix).
Chinwag and jaw are comparable in that they share the same donor do-
main, viz. body parts. They both refer to parts of the face which are seen mov-
ing when one talks. Hence, in both cases LA is referred to in terms of something
which is visible, an outward sign of talk, articulation rather than meaning or
content. This focus on the superficial level of language conveys the meaning
aspect 'talk for talk's sake', i.e. the primary function of casual conversation. In
both cases, also, no value judgement is attached to the words in the donor do-
main. What about the recipient domain?
When compared with chat, a non-metaphorical expression which has
more or less the same denotative meaning, chinwag is seen to contribute a
negative element to the concept of 'conversation', while the shift towards the
negative pole is even more obvious with jaw. Compare: chat: 80% positive (no
negatives); chinwag: 55% positive (other answers divided over negative and
neutral); jaw: 50% negative (other answers divided over positive and neutral).
Hence, the metaphors seem to have the function of expressing a negative value
judgement. Chinwag, as pointed out above, emphasizes the lightness or shal-
lowness of the subject-matter, expressed metaphorically by reference to the
surface level of articulation. In contrast, jaw seems to go in the other direction:
it scores considerably lower on 'light' (compare: chat: 95%; chinwag: 90%;
jaw: 55%). The definitions do point to a difference in value judgement through
the absence of the element 'enjoyable', 'friendly' in jaw. Compare:
chat COBUILD informal, friendly conversation, usu. about things which
are not serious or important LDOCE friendly informal conversation
chinwag COBUILD long enjoyable conversation between friends LDOCE
informal conversation; chat
jaw (vb) COBUILD talk to each other, often for a long time, without saying
anything important LDOCE (informal, sometimes derog.) talk/to talk
Furthermore, jaw can also mean 'sermonizing talk, lecture' (CODCE), which
may have caused the negative responses.
The less positive meaning of jaw as compared with chinwag can in fact be
explained from the donor domain: on the one hand the movement of the jaws is
a slower, heavier one than that of the chin, so that speed and frequency of
movement seem to be associated with 'lightness'; on the other hand, the word
wag may have associations with 'tail wagging', an outward sign of wellbeing in
animals.
The metaphors cackle and yap are based on the animal domain. Both ex-
press a negative value judgement, according to dictionary definitions as well as
according to questionnaire responses (Compare: yap: 75% negative; cackle:
85% negative). These expressions are discussed further in section 3.2.2, because
they focus on manner of speaking (loudness) rather than on the function of
'maintaining contact'. Their relevance here lies in the feature 'empty, unim-
portant subject-matter'. Compare:
yap LDOCE noisy empty talk; vb: talk noisily about unimportant things
cackle LDOCE foolish useless talk
It appears, then, that 'unimportant subject-matter' may have positive as well as
negative connotations, which lie on a scale (cf. Figure 1). Different metaphors
focus on different 'degrees' on that scale. 'Talk for talk's sake' becomes nega-
tive when seen as 'useless noise'. The animal domain seems to emphasize this
particular meaning aspect: the barking of dogs in yap; the noise made by hens in
cackle.
Figure 1
The metaphor small talk was valued by 60% of informants as having a negative
connotation. Yet, there is nothing in dictionary definitions to point to a negative
meaning element. Dictionaries refer to the lightness of subject-matter. Compare:
small talk COBUILD light conversation that people make at social occa-
sions about unimportant things LDOCE light conversation on unimportant
or nonserious subjects CODCE unimportant social conversations
QII yielded the following negative comments on small talk: 'very trivial',
'banal', 'a waste of time', 'semi-ritualistic practice', 'inconsequential'.
One explanation for its more negative connotation than chat or chinwag
may be the social context associated with it, i.e. talk for talk's sake between
strangers, not friends. Talking about unimportant subjects is on such occasions a
signal of the absence of real contact or of the superficiality of the relationships.
The donor domain is space: the word small denotes 'unimportant'. Although it
is neutral in the donor domain, it frequently expresses a negative value judge-
ment when it acquires a metaphorical sense in various collocations. Compare:
feel small ('feel ashamed or humble'); small fry ('young or unimportant per-
son'); small beer (s1. 'unimportant': He thinks he's wonderful, but he's really
rather small beer). (Definitions from LDOCE). Such examples support and are
coherent with Lakoff & Johnson's UP-DOWN spatialisation metaphors: '"Bigger
is better" is coherent with MORE IS UP and GOOD IS UP "Smaller is better" is not

coherent with them.' (Lakoff & Johnson 1980: 22).
3.1.2 Turn-taking
As pointed out by Good (1979), quoted above, equal access to speaker and
hearer roles is a convention to be observed by participants who want to pre-
serve a balance of power. It appears from value judgements that speakers who
break the unwritten rules of conversational behaviour in this respect are evalu-
ated negatively.
3.1.2.1 Duration of a turn

Dragging one's turn is generally valued negatively. For instance, the non-meta-
phorical expression bore someone is explained in LDOCE as 'make someone
tired, or uninterested, esp. by continual dull talk'. 'Continual' and 'dull' are
indeed frequently seen as companions which, in combination, put a burden on
the addressee. (cf. burden s.o. with a lengthy account, with a long story, etc.).
Metaphors for dragging one's turn are e.g. go on, run on, keep on, drone
on (Table A2). In go on and run on, the donor domain is forward physical
movement. The particle on also has a basically spatial meaning, viz. 'supported
by, attached to, covering, enclosing, or touching, something' (CODCE). In
combination with such dynamic processes as go and run, denoting motion, it
acquires the meaning of 'in some direction, ... farther forward' (CODCE), and
is used metaphorically for continuity in time. (Cf. Johnson 1987). In the ques-
tionnaire, run on was judged negatively by 75% of informants. Go on is nega-
tive in such contexts as: 'Don't go on about your problems so much' (QII). It is
the prolonged dwelling on the same surface which causes the negative evalua-
tions.
Drone on was found by 90% of informants to be a negative term. The
shift towards the negative extreme of the value judgement scale is due to the
reference to the animal domain, in which drone denotes the sound made by
bees. In the metaphor it is the monotony, the lack of variation in level or speed
which is focused on.
Another particle which focuses on duration is away, as in jabber away.
Here again, the primary meaning is spatial (cf. CODCE 'to or at a distance'),
and the negative connotation in the recipient domain comes from the view of the
speaker as moving forward on his own, leaving listeners (and their concerns)
behind.
A second consequence of dragging one's turn, apart from boring one's

audience, is that the other party has difficulties getting a turn. This negative
consequence is focused on in the expression not to get a word in edgeways:
conversation is seen as a container into which speakers put their contributions
(cf. further on). The negative evaluation of dragging one's turn is further appar-
ent from the results for a number of expressions in the Appendix, Table B3:
harp on about a subject (85% negative), carry on about a subject (75% nega-
tive), keep on about a subject (75% negative). Note again the use of the particle
on to signal duration.
3.1.2.2 Interruption o f a turn and violent turn-taking

Stainton (1987:89) claims that
... in every case an interruption functions as a means to take a turn at a
non-specified point in the current speaker's utterance. This fact causes in-
terruption to be viewed as an infelicitous feature of discourse.
Metaphors for (violent) turn-taking and/or interruption are: barge into a con-
versation, break in ('He broke in with some ideas of his own'), break into a
conversation, butt in, chip in, cut in, cut someone short, get in ('May I get a
word in?'), horn in, push in, put in, step in.
The most salient feature in these metaphors is the preposition into or the
particle in. As Lakoff and Johnson point out (1980:30-32), every activity is a
space which one can enter, and Rudzka-Ostyn (1988) shows that verbal com-
munication can be viewed as an entity whose boundaries may be transgressed.
In several expressions conversation is indeed seen as such a bounded space
which one enters with varying degrees of force. The more force is used, the
more negative the value judgement attached to the LA becomes. Thus, while
step in is neutral, break in, push in and barge into are definitely negative. Butt
in and horn in have the animal domain as their donor, and derive a strongly
negative value judgement from that.
Chip in and cut in refer to conversation as an object whose intact form is
destroyed. They focus on breaking up the other party's 'flow of talk': 'Don't
cut in while I'm talking'. The same view of LA is present in cut someone short,
break off someone's speech.
Finally, get (a word) in and put (a word) in also have the container view of
conversation as a basis, but here the speaker does not enter that container him-
self but moves objects (words) into it. These expressions are neutral with regard
to an evaluation of the LA, because get and put are themselves neutral items.
In conclusion, the degree of force which is used in entering conversation
directly correlates with the extent to which the LA is viewed negatively. De-
grees of force are expressed by means of types of physical movement and ways
of damaging or destroying objects.
3.1.3 Topic management
3.1.3.1 Bringing up a topic

Starting a topic is typically expressed in a metaphorical way as bringing up a
topic. The metaphor is a spatial one, in which LA is viewed as an upward
movement. As Rudzka-Ostyn (1988:539) convincingly argues, the 'landmark'
which is reached in such expressions with up is the hearer, 'his awareness and
receptivity'. Bring up (a topic) does not carry a positive or negative value
judgement. Similar is raise (a topic), in which the same spatial elements are
present. It was slightly surprising therefore that both expressions were valued
positively in the questionnaire: 65% for bring up, 75% for raise (Table A5).
This could mean either that actively contributing to conversation is in itself
valued positively, or that the expressions were valued positively in contrast with
other items in the list presented to informants. The following received negative
scores: drag in a subject (90% negative), drag up a subject (90% negative),
rake up a subject (80% negative) (Tables A5, B5).
In drag in/up, the focus is on the force which is needed to 'mention' a
subject: if you drag something, a certain amount of resistance is implied. The
metaphor refers to the raising of an undesirable topic that had better be left
alone: 'Paul dragged up the subject of the £5 I owe him'. (Example sentence in
QII).
In rake up, the donor domain is a more specific one, viz. that of garden-
ing: dead leaves are raked up. The negative value judgement may come from the
knowledge that dead leaves are ill-smelling and should be left covered. Hence,
raking up (a subject) means 'talking about something that should be forgotten'.
The same vertical movement underlies several other metaphorical ex-
pressions for 'raising' a topic: come up ('Your question came up at the meet-
ing'), pick up ('I'd like to pick up the point you made a moment ago'), take up
('I'll take up the story where I finished yesterday'). These are, however, neutral
with regard to value judgement.
Another view which underlies expressions is that of the topic as a con-

tainer, as in broach (a subject). (Cf. CODCE: broach = 'pierce (cask) to draw
liquor; open and start using contents of (bale, box, cargo, etc.)').
Starting a topic is also expressed in terms of physical motion towards it,
as in get onto (a subject), move onto (a new subject). Such physical movement
can take various more specific forms to refer to ways of discussing topics, as
shown in the next section.
3.1.3.2 Discussing or re-introducing a topic
S = speaker, T = topic, p = point
Figure 2
When a subject has been brought up, one starts talking about it, which is typi-
cally expressed as physical movement. An example is come to (the point).
Again, the way of moving is responsible for differences in value judge-
ment: plunge into (a subject) adds the meaning of 'suddenly or hastily': 'She
plunged at once into a description of her latest illness'. (Example from
LDOCE). Here, the scales of speed and force are relevant. As Table A5 in the
Appendix shows, the expression was valued positively by 70% of informants,
while 2 informants ticked the negative slot and 4 the neutral one. QII clarifies
these evaluations. On the positive side the comments were: 'suggests enthusi-
asm', 'suggests dedication'; on the negative side the comments referred to lack
of thought. One informant, who ticked the neutral slot, explained his answer as
'it shows initiative, but may reflect stupidity'.
LA as physical movement underlies a great many metaphorical expres-

sions referring to various aspects of dealing with topics: dealing seriously with a
topic is go into it or cover the ground', talking about a topic again is go back to,
return to, revert to, run back over a topic; one can range over many subjects,
and pass over, slide over a subject; one can come to the point, go off the point',
something may be beside the point (Tables A5, B5). In all of these, conversation
is a space in which speakers move around; topics are specific places or bounded
spaces within that larger space, and 'points' are specific and central places
within topics.
Another view of a topic is that of an object which can be handled in vari-
ous ways. Broaching (a subject) was mentioned above. Belabour (a subject) has
a strongly negative connotation (85% negative responses in questionnaire B3).
It is defined in LDOCE as 'to work on or talk about to silly lengths'. Its literal
meaning, according to LDOCE is 'to beat heavily with repeated blows, as with a
stick'. Inefficiency and clumsiness are part of the meaning. In the recipient do-
main the negative value judgement may be linked with the negative evaluation
of 'dragging a turn' in general.
If one tackles a subject, or comes to grips with it, one is definitely in con-
trol. The donor domain is that of 'manipulating', and the actions are seen as
successful. Hence, the positive value judgements (Table A5).
3.1.3.3 Deviating from a topic

The same metaphor of speakers 'moving around' accounts for a number of
expressions referring to whether one keeps dealing with one topic or not. Com-
pare: stay with (a subject), stick to (a subject), go off the point, jump (from one
subject to another). Here again, the types of movements in the donor domain
account for different types of LA denoted by the metaphorical expressions.
Jumping from one subject to another is a different type of linguistic behaviour
from meandering, which has the slow, turning flow of rivers as its donor do-
main, and ramble, which denotes a disordered and wandering way of talking.
With regard to value judgements the questionnaire results (Table B5)
point to a positive evaluation of stay with a subject (80%) and stick to a subject
(90%), while meander was felt by 55% to be negative and by 45% to be neutral
and go off the point was valued negatively by 65%. It looks as if deviating from
a theme is evaluated as undesirable linguistic behaviour. The comments in QII
confirm this. For instance, explanations for the negative evaluation of go off the
point include: 'suggests indecision', 'suggests a wandering mind', 'means losing
track of what you are saying'. Hence, lack of control is involved, which is dis-
approved of.
3.2 Manner of speaking
In this section I will discuss expressions referring to various aspects of a per-

son's speech style, i.e. the speed, the loudness, the quantity, the repetition and
the degree of directness.
3.2.1 Speed
When a person is said to speak fast or slowly, this is an evaluation on the scale
of speed, which does not necessarily imply a positive or negative judgement.
The question arises, however, to what extent a positive or negative evaluation
correlates with either fast or slow speech, and how relevant the scale of speed is
in the LA lexicon.
3.2.1.1 Fast speech

From a list of twenty verbs denoting LA, twelve were rated by informants as
referring to fast speech. Percentages ranged from 100% to 65%. These were:
chatter (100%), rattle (100%), babble (100%), jabber (95%), twitter (95%),
gabble (95%), prattle (90%), talk nineteen to the dozen (85%), natter (75%),
patter (75%), gas (70%), run on (65%) (Table B2).
Another group of twenty informants were given the same words, but were
asked to evaluate the type of verbal event denoted by these verbs as positive,
negative or neutral. With the exception of chatter (65% positive), natter
(divided) and patter (divided), all verbs were rated negative by 65% to 90% of
informants. In other words, 'fast' and 'negative value judgement' correlate in 9
out of 12 cases. (See Appendix, Table A2.)
If we look at the above list of words denoting fast speech, to which we
can add burble (60% negative) and splutter (75% negative), which were not
given a 'fast' rating by the majority of informants (probably because speed is not
the most salient feature of these verbs) but which both have the feature 'quickly'
in the LDOCE definitions, we see that fast speech is typically referred to by
means of sound-imitative words in -er and -le, suffixes with a frequentative
meaning.
Focusing on these sound-imitative verbs, we see then that 8 out of 10 ex-

press a negative value judgement, viz: prattle, jabber, rattle, gabble, twitter,
splutter, burble, babble. The verbs chatter and natter are, as said, not negative.
As Cassirer (1953) points out, the essence of onomatopoeic expressions is
that they are not merely mimetic:
In truth, ..., the meaning of each form cannot be sought in what it ex-
presses, but only in the manner and modality, the inner law of the ex-
pression itself. In this law of formation, and consequently not in proximity
to the immediately given but in progressive removal from it, lie the value
and the specific character of linguistic as of artistic formation (1953: 189).
Cassirer argues that onomatopoeic words have developed from mere mimicry
and imitation through the process of analogy to the purely symbolic expression
which "precisely in and by virtue of its otherness, becomes the vehicle of a new
and deeper spiritual content." (1953: 197).
Thus, while the formal markers -er and -le have come to denote repeated
sound, the resulting words, when referring to a form of linguistic action, are not
merely imitative but express the 'reporter's' perception of and attitude towards
that type of linguistic action. It appears that such attitudes tend to be negative
rather than positive. More specifically, the linguistic action denoted by these
words, is perceived as lacking deeper content. Let us have a closer look at
connotations of fast speech.
One consequence of fast speech is that it is often unclear, difficult to hear
or understand, or confused. These meanings are given for: jabber, babble, bur-
ble, gabble and splutter. Whereas 'unclear' and 'difficult to hear' still refer to
the sound level, 'confused' refers to the organisation of content. Hence, a link is
established between form and content.
This link becomes more obvious when fast speech is associated with lack
of content. Reference to unimportant subject-matter is explicit in dictionary
definitions of:
rattle on/away COBUILD talk quickly and for a long time, usu. about
something which is not very interesting or important
twitter COBUILD speak very fast in a high-pitched voice about things that
are not important, usu. because they are nervous
The definitions of the following words include reference to emptiness, foolish-
ness:
babble LDOCE talk quickly and foolishly in a way that is hard to under-
stand
burble LDOCE talk quickly but foolishly or in a way that is hard to hear
clearly
jabber CODCE 1 speak volubly and with little sense 2 utter (words) rap-
idly and indistinctly
prattle LDOCE talk meaninglessly or lightly and continually in a simple
way, or like a child
In this respect, then, the sound-imitative words are metaphorical in a broad
sense of the term, in that reference to the sound-level implies absence of the
deeper level of meaning. The explanation for the frequent association of fast
speech with lack of deeper meaning may be that thought requires time, so that if
one talks too quickly, one just produces sounds without meaning. Further, the
features 'fast, quickly' frequently co-occur with 'continually'. Examples are
rattle on, twitter, burble, prattle, but also talk nineteen to the dozen, pour
('flow steadily and rapidly'), etc. The same explanation applies to continuous
speech, viz. that if one speaks continually, LA becomes merely the 'mechanical'
production of sound. That such an association of 'fast' with 'continual' is a real
one is also apparent from the fact that, although to gas and to run on have no
feature 'fast' in the dictionary definitions, which only refer to continuous speech
and empty talk, 70% of informants attributed the feature 'fast' to gas, and 65%
did so for run on. (In the latter case, the donor domain is clearly salient). (See
further: 3.2.3)
The same link between high speed and lack of deeper meaning or value is
present in the metaphors knock off ('write quickly') and throw off ('say/write
something easily'). The focus here is on the quick separation of words from the
speaker (the particle off): the donor domain is that of manipulating objects in a
brusque manner, or more generally, it is spatial.
Another connotation of fast speech may be that it is aggressive. A suc-
cession of rapid questions is felt to be threatening because the addressee is
pushed into a position where he has to produce answers quickly, which is face-
threatening. It can be seen that a series of questions put rapidly one after the
other is typically expressed by means of metaphors from the donor domain of
war and fighting: a barrage of questions, besiege/bombard/pelt s.o. with ques-
tions. The metaphor a volley (as in 'He directed a volley of curses at the driver'
LDOCE) fits into the same pattern.
The explanation for these metaphors is that an object moving at high
speed will hit a target harder. Hence, if that object is directed at someone with
the intention to attack, speed becomes threatening. As there is in these meta-
phors a link between the scales of speed and frequency, they are discussed fur-
ther in section 3.2.4.
Summarising, we can say that fast speech frequently has the negative con-
notations of lack of clarity and lack of content. A typical donor domain is that
of sound, in that a high number of items referring to fast speech are sound-
imitative. Some of those have links with the animal domain, such as twitter
(birds); others are also used for sounds made by objects, such as rattle, clack
(cf. CODCE: 'sharp sound as of clogs on stone or of boards struck together'),
rap (LDOCE: 'sound of a quick light blow: a blow on the door'); still other
sound-imitative words are also used for the sound made by water flowing, such
as burble (cf LDOCE: 1 'to make a sound like a stream (flowing over stones)').
The same link between liquid and fast speech is present in pour (out), which has
no positive or negative value judgement but merely refers to a smooth move-
ment. Other donor domains are the brusque manipulation of objects and fight-
ing. The latter gives rise to the connotation of aggression, sometimes a con-
comitant of rapid talk directed at someone.
3.2.1.2 Slow or impaired speech

The question arises what sort of value judgement attaches to slow speech.
In the questionnaire, the words listed in Table 6 were judged by infor-
mants to have the feature 'slow' (cf. Appendix, Tables A2 and B2).
A possible explanation for converse may be that it is judged to be slow because
of the seriousness of the subject-matter and the formality of the type of occasion
which the word is associated with. The other three items are, as can be seen,
definitely judged to express negative value judgements.
Table 5
converse: 85% slow; 100% neutral or positive

drone on: 95% slow; 90% negative
falter: 75% slow; 65% negative
flounder: 70% slow; 70% negative
Two associations seem to emerge as typical of words denoting slow verbal

events: slow speech is non-fluent and boring. Lack of fluency is for instance
expressed by: flounder, falter, stumble, fumble (for words). Donor domains are
physical movement and the manipulation of objects. In flounder, falter and
stumble, non-fluency is compared with difficult progress: walking is hampered.
(Compare with their opposite run on !). To fumble focuses on awkwardness in

handling objects or words. What all of these items share is that they associate
non-fluent speech with lack of control over one's own body. Non-fluency is
unintentional, due to negative factors hindering normal, efficient movement, i.e.
speech. Comments on falter in QII all mentioned such features as 'suggests
failure / indecisiveness / uncertainty / hesitation'. In all of these metaphors,
however, the feature 'slow' seems to be the result of the more salient feature
'non-fluent'.
The verb drone on (valued as slow and negative in the questionnaires:
Appendix, Tables A2 and B2) is defined in COBUILD as 'talk boringly' and in
LDOCE as 'in a low dull voice'. The donor domain is that of animal sound, viz.
the sound made by bees. Although the item was almost unanimously judged to
denote slow speech, the slowness again seems to be less relevant than the mo-
notony. As in the non-fluency metaphors, slowness is concomitant rather than
central. The metaphor drone on is also discussed in 3.1.2.1. above, because of
the particle on denoting continual talk.
3.2.2 Loudness
When one examines the metaphors in the LDOCE corpus which in their defini-
tions have the terms 'loudly, noisily, etc', the following connotations emerge.
Loud speech is uninhibited, free, forceful. The assumption is that, if
someone speaks loudly, he wants to be heard by everyone, and hence is not
afraid of reactions. These features are present in the non-metaphorical items
vocal and vociferous, on which informants were divided with regard to value
judgement in terms of positive, negative or neutral (Table A4). The metaphor to
sound off on the other hand, was found by 50% of informants to be negative
(Table A6). The explanation is that the positive characteristic 'uninhibited'
becomes negative when speech is 'imposed' on an audience: loud speech cannot
be ignored. The definition in COBUILD makes this clear: 'express your opin-
ions strongly or loudly to everyone without being asked'. This also explains the
divided answers for vociferous (8 positive, 8 negative, 2 neutral). Comments in
QII indeed show that the positive evaluation applies to speakers who 'state their
opinion clearly', 'are well-spoken and coherent'. Comments on the negative
evaluation were: 'implies over-loud arguing', 'suggests someone who is inclined
to be overbearing'.
A similar metaphor is to air, which literally means 'expose to air', and re-
fers to 'make known to others (one's opinions, ideas, complaints, etc.), esp. in a
noisy manner' (LDOCE). The example given in LDOCE is: 'They were tired of
the doctor airing his knowledge'. However, to air one's ideas was given in QUI
and valued by 7 out of 10 students as positive, and by the 3 others as neutral.
The 'positive' comments referred to: 'sharing one's views with others',
'circulating ideas, so that others can think about them', 'suggests free discus-
sion, a forum for sharing thoughts', 'implies healthy discussion and exchange of
thought'. It seems, then, that the feature which was valued positively by my
informants is 'openness'. The example given by LDOCE to illustrate the nega-
tive connotations collocates to air with knowledge, which conveys a different
attitude as compared with ideas, since the former suggests authority and control
over others. This shows that the linguistic context in which such metaphors
occur may determine which meaning aspect is more salient. Further, I would
hypothesise that the sub-culture plays a role, as well: young people seem to
value openness.
One step further, loudness becomes offensive. The metonym loudmouth
indeed refers to someone who talks 'in an unpleasant, offensive or stupid way'
(COBUILD). The feature of aggression is present in animal-sound based meta-
phors for loud speech, such as bark ('say something in a sharp loud voice'
LDOCE), and bellow ('shout something in a deep voice' LDOCE).
The donor domain of nature gives rise to the metaphors bluster and thun-
der. In both cases, loud speech is compared to threatening weather conditions,
i.e. wind blowing roughly and thunder respectively. Finally, the aggressive con-
notation of loud speech is expressed by reference to the manipulation of objects
in hurl, whose literal meaning is 'throw with force, throw violently from some
position' and whose metaphorical meaning is 'shout out violently' (LDOCE).
The general pattern in these metaphors is that loud speech is forceful
speech, which becomes negative when force is directed against an addressee, in
which case it is threatening.
A different value judgement attaching to loudness is stupidity. This feature
is present in loudmouth (cf. above), as well as in a number of sound-imitative
words which combine 'fast and loud', such as jabber, rattle on, gabble, at least
according to questionnaire responses. All of these were valued negatively. Jab-
ber also refers to the noise made by monkeys, and a number of other metaphors
show that animal sounds are a typical donor domain for loud, stupid talk: cackle
(hens), yap (dogs), gaggle (geese). The latter, according to LDOCE, refers to:
'1. number of geese together; 2. group of noisy people, esp. women, who talk a
lot.'
The opposite of loud speech, viz. soft, quiet speech, appears to be less sa-
lient, since I did not find metaphors for it in the corpus. The word softspoken
seems to indicate, though, that it is valued positively. This may help to explain
why loud speech should be associated with stupidity. On the one hand, loud
speech reminds one of sounds made by animals (cf. the animal-based meta-
phors), on the other hand it is a signal of anger, aggression, and hence lack of
self-control. The positive appreciation of soft speech may reflect the high value
put on 'education', 'fine breeding', etc. (Cf. 'Keep your voice down'). It is
possible that such a negative appreciation of loud speech should be culturally
determined, and possibly absent in other cultures.
3.2.3 Quantity
3.2.3.1 Talking a lot

The word chatty, which is defined in LDOCE as 'fond of talking' is rated as a
positive term by 85% of the informants (cf. Appendix, Table A4). The explana-
tion is that a chat is a friendly, informal conversation with a fundamentally social
function, and hence that someone who likes to engage in that sort of activity is
seen as a sociable person. Most words, on the other hand, which refer to
'talking a lot' have a negative connotation.
First of all, it is striking that most verbs which denote rapid speech also
denote continuous speech, viz. clack, chatter, natter, babble, burble, etc., i.e.
the sound-imitative ones. The endings -er and -le in these words are linguistic
signs for repeated sound (cf.3.2.1.1. above). Rapid repetition is associated with
triviality, and hence the connotations of 'unimportant subject-matter', 'lack of
content', 'emptiness', 'foolishness', which were found to be attached to quick
speech are also attached to continuous speech.
For instance, chatter, still a positive word according to questionnaire re-
sults (see Appendix, Table A2: 65% positive) is yet less positive than chat. The
element -er accounts for that. In chatterbox the mechanical, continual produc-
tion of talk is focused on even more clearly.
Words such as blabber, prattle, palaver refer to talking a long time with-
out saying anything important, or even to talking too much and foolishly.
This connotation of 'lack of serious content' is not only present in the
sound-imitative words but also in metaphors like gas and gasbag. Compare:
g a s LDOCE talk a long time about unimportant things
gasbag COBUILD person who talks a lot, esp. about things that are not
important (informal and offensive word) CODCE bag for holding gas; bal-
loon's or airship's gascontainer; (derog.) empty talker
In these metaphors the link between donor and recipient domains is emptiness.
It is significant, though, that both words have the meaning element 'large
quantity of speech'.
The explanation for the frequent association of continual talk with empti-
ness or foolishness must be the same as that suggested above in connection with
quick speech: serious thinking requires time; if one talks continually, there can-
not be much thinking behind it. Emptiness of content is expressed by focusing
on the sound level in the sound-imitative words, or by reference to something as
rare or light as 'gas'.
Talking (too) much is further associated with lack of control, because
there is the risk of saying something one did not really want to say or something
one should not have said. The metaphor let something slip is based on the do-
main of movement, and has a negative meaning due to the loss of balance and
control, i.e. to the element 'involuntary'. In the questionnaire, 80% of in-
formants valued the term as a negative one (Table A7). The importance of self-
control is further corroborated if one looks at evaluations of the opposites, viz.
of terms denoting carefulness in speech, such as guard one's tongue (75% posi-
tive).
However, the issue is slightly more complicated. The expression watch
one's words, which seems to be near-synonymous with guard one's tongue, was
rated as positive by only 65% of informants (Table A7). This may be an indica-
tion of a gradient: self-control is positive, but over-prudence is negative. The re-
sults of QUI for watch one's words further reveal the importance of collocation
and situation in value judgement. The expression was indeed valued negatively
by 8 out of 10 informants. (One ticked 'neutral', one left it blank). The explana-
tions all refer to over-carefulness imposed on one. Examples given by subjects
to illustrate the expression were:
(17) You have to watch your workaround him; he's easily offended.
(18) You'd better watch your words. She's his girl-friend, you know.
(19) If you don't watch your words, you'll find yourself in a difficult
situation.
(20) Make sure you watch your words when she arrives.
(21 ) The teacher told the boy to watch his words.
In all cases, the metaphor occurs in a warning or an obligation imposed on

someone. Hence, the negative evaluation results from a rejection of outside
control, and subjects identifying with the controlled party. On the other hand it
is likely that guard one's tongue also predominantly occurs in such contexts (cf.
LDOCE: 'You must guard your tongue carefully'), and hence that the difference
between 75% positive answers and 65% is not significant. The important point,
however, is that 'carefulness' as a result of self-control is valued positively,
while imposed carefulness (hence outside control) is valued negatively. Infor-
mants' answers may have depended on whether they evaluated the former (self-
control) or the latter (outside control).
Connected with quantity of speech is the meaning 'telling secrets'. Hence,
a babbler, which means 'someone who talks too much', can also mean 'a teller
of secrets'. Similarly, blabbermouth means 'person who tells secrets by talking
too much'. The metaphor have a loose tongue is defined in COBUILD as 'talk
about things that should really be kept secret', and in LDOCE as 'not well con-
trolled'. In the questionnaire, it was valued as negative by 90% of informants
(Table B7).
Also somewhat related are metaphors referring to social blunders: drop a
brick, drop a clanger, heavy-handed, put a foot wrong. Expressions with this
meaning typically convey lack of control and lack of skill. The scale of quantity
is only marginally relevant here, however, since the salient aspect is control.
A number of metaphors refer to the production of a great quantity of
words, without apparently expressing a positive or negative value judgement.
These are such nouns as a flood of..., a flow of ..., a gush of ..., outpourings of
..., a torrent of ...; such verbs as pour (from one's lips), pour out, spout; such
adjectives as brimful of etc. It can be seen that the donor domain is that of
liquid: words follow one another so fast and continuously that they are com-
pared to flowing, streaming water. If one looks at examples given in the dic-
tionaries, it appears that all of these metaphors have their own specific conno-
tations, related to the specific type of movement of water in the donor domain,
and that positive and negative value judgements in these cases are context-
dependent.
A flood 'of complaints' (LDOCE), 'of grateful telegrams and letters'
(COBUILD), 'of words' (OALDCE) focuses on the quantity reaching someone.
Donor domain is 'water in a place which is normally dry' (LDOCE). The scale
of quantity is definitely relevant here, but the value judgement may change de-
100 A N N E - M A R I E SIMON-VANDENBERGEN
pending on whether the message is a positive or negative one: flood is simply a

hyperbole.
A flow is also context-dependent with regard to value judgement. Com-
pare: a flow 'of meaningless words' {LDOCE), 'of angry words' {OALDCE),
'of apologies' {COBUILD). The focus here is on the smoothness of the move-
ment, so that flow can easily acquire a positive meaning of fluency. Compare:
flow COBUILD If somebody's words flow, they are spoken smoothly and
continuously without hesitation the words flowed on, in perfect sentences
Somebody who is in full flow is talking fluently and easily and seems likely
to go on talking for some time He stopped me when I was in full flow.
The metaphor gush focuses on the sudden and rapid flow of a great quantity of
liquid, i.e. speech. Compare the examples: a gush 'of expressions of sorrow'
{LDOCE), 'of anger/enthusiasm' {OALDCE), 'of pleasure, enthusiasm, con-
fessions, etc.' {COBUILD). The sudden, rapid movement is linked with strong
emotions, which can again be positive or negative ones.
The same applies to outpourings, defined in LDOCE as 'continuous
strong expression of feelings'. However, the second meaning given in CO-
BUILD points to the fact that a great flow of words to express one's feelings
may be a signal of lack of control, and may as such acquire a negative value
judgement. Compare:
o u t p o u r i n g s COBUILD things that you say or write which seem uncon-
trolled and irrational and indicate that you are very angry and upset about
something the hysterical outpourings of fanatics
The metaphor torrent has the features 'great quantity', 'speed' and 'force'.
LDOCE gives as its literal meaning: 'violently rushing stream, esp. of water
(fig.) a torrent of tears/bad language'. Compare:
torrent OALDCE violent outpourings a - of words I abuse I insults CO-
BUILD a torrent is a lot of water falling or flowing rapidly or violently - a
torrent of questions, abuse, etc. is a lot of questions, directed continuously
at someone
It is clear that the feature 'violent' in this metaphor tends to make it a more
negative word than the others discussed so far.
Just like the noun outpourings, the verbs pour (intransitive) and pour out
(transitive) may be associated with positive or negative messages. The verb
spout, however, has a negative meaning:
s p o u t (out) COBUILD If someone spouts or spouts out words, they say sth
in a way which is often boring and which gives people the impression that
they are not thinking about what they are saying.
It appears, then, that the same connotation of 'lack of thought' already

found to be typical of continuous as well as fast speech recurs in this metaphor.
3.2.3.2 Saying little

One connotation of using few words is carefulness in speech, expressed in terms
of control: cf. such expressions as bridle one's tongue, guard one's tongue,
watch one's words, weigh one's words. Control is metaphorically expressed as
control over animals (bridle), over people (watch, guard), and over objects
(weigh). In the recipient domain, the 'controlled' are one's own words or one's
own articulator (tongue), and hence control as self-control is valued positively.
However, as pointed out above (section 3.2.3.1.), the 'self-control' may be
imposed and hence become external control, which is valued negatively when
one identifies with the controlled person.
Outside control is also valued negatively in a metaphor like tongue-tied,
which indicates that one is controlled by such psychological states as nervous
ness, embarrassment, etc.
t o n g u e - t i e d COBUILD unable to say anything because they feel shy or
nervous LDOCE unable to speak freely, as through awkwardness i n the
presence of others
The donor domain here is a physical impediment to speech (cf CODCE: tongue-
tie = 'impediment in speech due to shortness of fraenum of tongue'.)
The same explanation applies to the expression Lost your tongue?, which
is negative and refers to the inability to speak due to some factor beyond one's
own will.
More difficult to evaluate is the metaphor keep a stiff upper lip, defined in
COBUILD as '... they are behaving in a firm way and are not showing any
emotion even though it is difficult for them not to'. This expression very
strongly focuses on self-control, and yet opinions in the questionnaire were
divided: 6 positive, 9 negative, 5 neutral (Table B7). This again points to per
sonal values playing a role in assessing expressions such as these. The results of
QII and QUI (which both included the item) clarify the picture. Comments on
the positive evaluation pointed to: 'firm resolve', 'strength and bravery',
'courage', 'control', 'someone who is calm, cool and collected'. Hence, they
support the cultural value put on self-control. On the other hand, comments on
the negative evaluation were such as: 'suggests class bias', 'old British army',
'very antiquated', 'keeping up appearances', 'suppression of emotions'. Hence,
these informants reject the upper-class values of reserve and restraint.
Being silent is also valued negatively when it shows un-cooperativeness,

as in close-lipped (90% negative) and tight-lipped (85% negative) (see Appen
dix, Table A4). The same focus on control over one's articulators is present in
my lips are sealed, referring to keeping secrets. The value judgement here is
positive rather than negative. (Cf. 3.2.3.1: telling secrets is a negative conse
quence of talking a lot).
In conclusion, there seem to be fewer negative connotations of saying lit
tle than of saying a lot. This also shows up in the greater number of metaphori
cal expressions for talking a lot than for using few words. Metaphors seem to be
primarily coined for expressing negative value judgements.
3.2.4 Repeated linguistic action
Repeated L A is especially focused on in two types of speech acts, both of which

demand something from the addressee, viz. questioning someone and urging
someone to act. Both of these, when repeated, put high pressure on the ad
dressee and hence are felt to be threatening.
Repeated questions are seen in terms of physical attack: be
siege/bombard/pelt s.o. with questions, grill s.o. with questions, and even
badger s.o. with questions. In this latter metaphor the donor domain is hunting.
(Cf. CODCE: 'pester as dogs worry a badger; torment, tease'). In a barrage of
questions we notice a heaping up of metaphors: the first meaning of a 'dam' led
to the metaphorical meaning in the domain of fighting, which in its turn led to
the metaphorical meaning in the domain of linguistic action.
The metaphor push somebody ('force s.o. to do something by continual
urging' LDOCE) has physical force as its donor domain: the exertion of force
on someone leads to the experiencer's loss of control. Rub something in shows
a parallel in that here also the speaker says something repeatedly, and this is
metaphorically expressed in taking control over the addressee's body.
Repeated L A of the types described above is valued negatively because of
the dominant position assumed by one party, the mental control of one party
over the other one.
3.2.5 Directness
I shall now address the question to what extent being direct or indirect is
evaluated positively or negatively by speakers of English.
ASSESSING LINGUISTIC B E H A V I O U R 103
Results from the questionnaire show that being direct tends to be evalu
ated positively, though not unreservedly so. Compare: give a straight-out an
swer (95%), tell s.o. in plain terms (90%), tell s.o. straight (90%), speak one's
mind (80%), tell s.o. flat out (70%), use plain words (70%), speak plainly
(65%), call a spade a spade (60%), not to mince one's words (60%), tell s.o.
point-blank (55%) (Tables A6, B6).
There is clearly a gradient here, ranging from near-unanimous agreement
on the positive connotations to a nearly equal distribution of positive and nega
tive evaluations. The explanation is that 'directness' implies honesty and clarity,
which are evaluated positively, but may lead to rudeness, and thus hurt people.
The comments given in QII and QUI support this. Again, as with other items
discussed above, however, it is not the case that each individual expression has
either the positive or the negative connotation. Rather, informants agree on the
connotations to a greater or lesser extent, so that here again individual appre
ciations of types of behaviour come into the picture. A comment on not to
mince one's words given in QIII may clarify this:
It implies someone who is not afraid to say what they think and although I
think people should consider other people's feelings, we spend too much
time skirting around a subject for the sake of unnecessary taboos and con
ventions in our society.
Although expressions on which opinion is particularly divided would get more

agreement when used in contexts, such comments do show the role played by
individual and sub-cultural value systems. A comment on call a spade a spade,
given by a Japanese student (whose answers, since they did not come from a
native speaker of English, were not included in the counts), shows that cultural
values definitely determine evaluations:
In my opinion, this expression is a negative one because we usually avoid

saying something clearly or directly: we've got the custom to respect other
people's opinions and feelings.
There is apparently a clear link between the negative evaluations and flouting
politeness principles which require indirect speech acts.
The following items received a negative evaluation: plaster over an issue
(95%), hedge one's answer (90%), wrap up one's meaning (85%), beat about
the bush (85%), give an evasive answer (85%), talk round the issue (70%)
(Tables A6, B6). Two factors in particular are involved here, viz. deceit and lack
of clarity. The reason why plaster over an issue gets near-unanimous agree-
ment, as compared with say talk round the issue is that in the former deceit is
salient, in the latter mere absence of clarity.
Opinion was very much divided on speak baldly (Table 6). The expla
nation must be that judgements depended on which features of the expression
informants had in mind. (cf. the LDOCE definition: 'spoken plainly, even cru
elly'). Further, as suggested before, the informant's personality and his whole
set of values play a role.
When we look at the donor domains which provide metaphors for direct
ness and indirectness, it appears that there are two major ones.
The first important domain is that of movement in space. Directness in
speech is expressed in terms of movement along a straight path, without turn
ings. Indirectness is compared with the opposite. Contrast: a straight-out an
swer, flat out v. talk round an issue, an evasive answer.
The second domain is that of covering and concealing. Underlying these
expressions is the metaphor 'Seeing is understanding'. The more one covers up
the meaning, the greater the indirectness. Compare: plain words, speak plainly
v. wrap up one 's meaning, plaster over an issue, hedge one 's answer. In (not) to
mince one's words, the object in question is food, which is (not) made more
palatable by mincing it.
This second donor domain leads us to another group of metaphors which
express some form of 'indirectness', viz. by referring to various ways of
'dressing up' one's language (to use another metaphor). Examples are: adorn,
embellish, polish՛, florid, flowery, frilly, fustian, refined. These words also
focus on speech as an object which is treated, decorated and/or covered up in
various ways. Value judgements here seem to depend on whether the decora
tions are exaggerated or not. The expression polish one's speech was included
in Q I , and opinion was divided: 9 positive, 5 negative, 5 neutral (Table 6).
This was surprising in view of the rather positive definitions in LDOCE and
COBUILD. The latter even adds 'used showing approval'. The expression was
therefore included in QII and QUI, and the following 'negative' comments were
given: 'if one has to polish one's speech it suggests one feels inferior',
'snobbery', 'false, rather pretentious speech'. These comments clearly testify to
an ideology which favours 'the natural' rather than adaptation to certain norms.
The evaluations of and comments on a refined way of speaking go in the same
direction. In QUI, 4 out of 10 informants thought it was a negative term, and
gave the comments 'put on', 'false', 'snobbish' 'posh'.
Finally, there are the definitely negative terms to refer to particular styles,
such as inflated, swollen, turgid, padding. The motivating factor in these is the
contrast between a big shape and inner emptiness.
4. Donor domains for the evaluation of linguistic behaviour
I shall now examine more closely the donor domains most frequently drawn
upon for the expression of the various aspects of conversational behaviour
discussed above. These domains vary in degrees of abstraction and may inter
sect. For instance, the domains of 'movement in space' and 'liquids' intersect,
and so do the domains of 'liquids' and 'sounds'. Hence, when a metaphorical
expression is given as an example of a particular donor domain, this does not
imply exclusive membership of that domain. Rather, the implication is that the
feature which is most relevant to the meaning of the metaphor comes from that
domain.
4.1 Body parts
Chinwag and jaw refer in the donor domain to parts of the face which are seen
moving when one talks. When these body parts are used to refer to informal
conversation the focus is on the superficial level of language, viz. articulation of
sounds. This emphasis conveys the meaning of 'absence of a deeper content'.
The noun loudmouth used metonymically to refer to a person who talks
loudly and in an offensive or stupid way focuses on the most important body
part used in speech. Again, the articulatory or phonetic level is used to convey
something about the content of the message. The same negative value judge
ment is present in blabbermouth to refer to a person who talks too much and
hence cannot keep a secret.
Expressions with tongue such as bridle one's tongue, guard one 's tongue,
tongue-tied, have a loose tongue, lost your tongue? and with lips such as close-
lipped, tight-lipped, keep a stiff upper lip, my lips are sealed, focus on the most
movable articulators and convey a positive value judgement or a negative one
depending on
• the extent to which the speaker is able to control his articulators;
• the extent to which the speaker is seen as behaving in a 'sociable' way. (Self-
control is positively valued only insofar as it does not hinder social contact
and co-operation.);
• the extent to which the speaker is (not) controlled by an outside force.
4.2 Movement in space
The donor domain of movement is a pervasive one in expressions for LA.

Speakers are seen as moving forward in various ways; e.g. falter, stumble,
flounder, go on, run on, step in, push in, barge into (a conversation), get onto,
move into, plunge into, go back to, return to, pass over, jump etc. Further,
speech acts such as answers may be straight-out, or evasive.
Value judgements here depend on various factors, the most important
ones being:
• degree of force or violence used;
• degree of control over one's movements;
• degree of 'efficiency' in getting to a place directly (cf. the negative evaluation
of meander, ramble, evasive).
4.3 Liquid
L A is frequently compared to flowing water, mainly in three ways:

(i) L A is compared to rivers, streams, etc. for the sound produced in this way;
(ii) L A is compared to water for the effect of a large quantity of something
coming from some source;
(iii) L A is compared to rivers, streams, etc. for the way in which they move.
Examples of (i) are babble, burble՛, examples of (ii) are outpourings, spout,
flow, gush՛, examples of (iii) are meander, gush, torrent. These metaphors are
particularly used to refer to speed and quantity of speech produced as well as to
the manner in which speakers deal with conversational topics. Meander is an
example of the latter function: speakers are seen as taking many 'turnings', i.e.
as not 'coming straight to the point'. (cf. 4.2).
4.4 Other natural elements
Apart from rivers and streams, other elements in nature are used as donor do
mains in my corpus: bluster and thunder are both used for the element 'force',
combined with loud noise. Hence, they are used to link loudness with aggres
sion.
4.5 Animals
The donor domain of animals gives rise to expressions conveying a negative

evaluation of L A mainly for two reasons:
• the focus on the similarity in sound between animal noises and human speech
conveys the meaning of stupidity;
• animal behaviour is used as a donor domain because of the feature
'aggression' (e.g. horn in, bark).
4.6 Air, gas
Expressions based on air or gas typically convey a negative value judgement

because of the empty nature of the LA referred to: e.g. to gas, windbag, gasbag.
Emptiness is also focused on in turgid, swollen, inflated, padding.
4.7 Manipulating objects
Speakers are seen as dealing with words, messages, topics in the way one ma
nipulates objects. Value judgements depend on such factors as:
• the speaker's skill or absence of it;
• the extent to which the speaker-manipulator wants to 'disguise' the real
object by treating it in various ways;
• the relevance and usefulness of his actions (e.g. belabour is negative);
• the extent to which the speaker's action is destructive (e.g. chip in, cut in are
negative).
4.8 War and fighting
Expressions based on the domain of war and fighting all express a negative
value judgement of the type of L A referred to. They are used to convey the
meaning of aggressive behaviour. In this corpus they were found to be typically
used for the expression of repeated questions (cf. 3.2.4. above), which are
'threatening' to the experiencer.
4.9 Sound
Apart from those expressions where sounds specifically come from water (4.3),
natural elements (4.4) or animals (4.5), there are a great number of sound-imi-
tative expressions which refer to L A only. They typically end in the suffixes -le
and -er denoting the frequentative repetitive nature of the sound. They are
metaphorical in the broad sense of the term only. Though primarily referring to
fast, continuous and/or loud speech, they frequently express a negative value
judgement, in that focus on the sound level tends to imply absence of content,
hence foolishness, stupidity, etc.
5. Scales and value judgements
It is clear from the present corpus that in the evaluation of L A the scales of
quantity, frequency, speed, duration and intensity, which are suggested as rele
vant in Pauwels & Simon-Vandenbergen, play an important role.
Firstly, there is a direct relationship between the scale of speed and the
evaluation of fast/slow talk. It appears that metaphorical expressions for fast
talk nearly always express a negative value judgement. This judgement is typi
cally expressed in words focusing either on the sound level (the sound-imitative
ones) or on the mechanical production of speech (by means of donor domains in
which the 'sources' are machines or guns). This means that when fast speech is
expressed metaphorically it is because it is judged to be 'too fast', i.e. an ex
treme position on the scale is focused on.
The other extreme, i.e. speaking too slowly, is also valued negatively, ei
ther because of the lack of variation (leading to boredom) or because of the lack
of smoothness. The latter aspect is typically expressed by drawing on the donor
domain of bodily movement and skill, especially through focusing on lack of
control.
It appears, however, that 'fast speech' is more frequently metaphorised
than 'slow speech', and that in those items which were judged to denote slow
speech, other features are more salient.
Secondly, quantity and duration are focused on in metaphors for talking
too much or too little. Again, as with speed, 'too much', 'too long' are more
frequently metaphorised than 'too little'.
It is striking that in a great number of cases the sound-imitative words de
note 'fast and continuous' speech, which means that the same value judgements
attach to both types of LA. Mere quantity is focused on in a number of expres
sions based on the donor domain of 'liquid', and value judgements depend on
the context. Again, quantity and speed may be combined (as in gush, torrent).
Quantity of speech is typically associated with lack of thinking or of rationality,

either in the case of 'foolishness, stupidity' or in the case of the expression of
strong emotions. In the latter case lack of control may be involved, leading to a
negative evaluation. Lack of control is also responsible for 'awkwardness',
which is donor domain for expressions for social blunders (e.g. drop a brick, put
a foot wrong).
On the other hand, saying little is frequently associated with self-control
(control over one's articulators or words/ message), which is valued positively,
except when 'saying too little' is felt to be a form of uncooperativeness, asocial
behaviour.
Thirdly, frequency is an important scale as well. First of all, the suffixes -
er and -le, typically used to form words for 'fast and continuous' speech have a
'frequentative' meaning: they focus on the repetitive nature of the articulated
sounds. Further, the scale is focused on in the terms for 'repeated questioning'
which are typically based on the donor domain of war and fighting (cf. barrage,
bombard, pelt, besiege) as well as in an expression such a rub it in, which is
neutral in the donor domain but becomes negative in the recipient domain by
focusing on 'repetition of movement' in order to achieve one's aim, which is to
force the hearer to absorb one's message.
Finally, the 'vaguest' or most abstract scale is that of intensity. It is,
however, an important one when it is taken in the general sense of 'strength (in
quality or feeling)' (LDOCE). In this sense it is frequently linked to the image
schema of force.
Force is focused on in metaphors for L A that is seen as aggressive, threat
ening, or violent. This is the case in some metaphors for loud speech (such as
hurl), metaphors for the expression of strong negative feelings (such as a torrent
of abuse), metaphors for violent turn-taking (such as break in, butt in, horn in,
push in). The degree of force used is highly relevant for the difference in value
between different metaphorical expressions in the same semantic field. Compare
step in, break in, push in, horn in to refer to various ways of interrupting an on
going conversation, and a flow of, outpourings of, a gush of, a torrent of to
refer to various ways of expressing emotions.
Force is also relevant for the difference in value between such items as
bring up a subject (low value on the intensity scale) and drag up a subject (high
value on the intensity scale). It also explains the difference between get onto a
subject and plunge into a subject: the latter has a higher value on the intensity
scale.
In conclusion, the scales are relevant in the evaluation of linguistic action

in that metaphorisation is often grounded on the perception of L A as being
'extreme' in some sense: too slow, too loud, etc. The five scales of speed, fre
quency, duration, quantity and intensity are frequently focused on in metaphori
cal expressions drawing on various more concrete donor domains.
It was pointed out in the introduction (Section 0.) that the scales may be
regarded as (very) abstract domains. The corpus shows that there are two types
of domains which are relevant: the more concrete ones and the image schamata.
Although abstraction is necessarily involved in all types of metaphorisation,
some domains are indeed less abstract than others. For instance the donor do
main of animals gives rise to horn in, to refer to a particular type of linguistic
action. In the metaphorisation process, various features in the donor domain are
abstracted from. When we compare, however, horn in, with hurl, there is no
concrete donor domain which they share, and yet they are linked in the abstract
domains of 'force' and 'intensity'. The most abstract donor domains (the scales)
are arrived at via the establishment of links between what is coded in the lexi
con. At the same time, they are a reflection of the pervasiveness of the image
schema of SCALE in our experience (Johnson 1987: 121-124).
6. Context-dependent and context-independent value judgements
A final question asked at the outset was how realistic the dichotomy context-
dependent v. context-independent value judgements would be. The findings of
this study suggest the following answers.
• Firstly, apart from just a few exceptions, the words presented in the ques
tionnaires did not receive unanimous judgements. On the other hand it is
clear that some expressions invite a greater amount of agreement than others.
Therefore a gradient-view seems realistic. Words which are judged to be
positive by say 90% of informants must be different from those which are
judged as positive by say 45%, as negative by 25% and as neutral by 30%.
The difference between the types must lie in the fact that some words are
prototypically positive or negative, whereas others are not.
• Secondly, we must account for the lack of agreement. It appears from the
results of the questionnaires that informants in some cases had different col
locations and contexts in mind. The former are particularly interesting and
worth further examination. In any case it is obvious that the same expression
may, depending on contextual factors, be positive or negative. Compare:
(22) We had a good old natter this morning and it cleared the air.
(Positive).
(23) These women nattering all the time got on his nerves. (Negative).
Of course, pragmatic factors may always override value judgements (even
prototypical ones), as in:
(24) I hate nice and cosy chats. (Negative).
The word chat, whose positive meaning is reinforced by nice and cosy, here
occurs in an utterance which, through the superimposition of the speaker's
individual judgement, expresses negative feelings about what are typically
positive events. However, it is useful to distinguish between such clear cases
of 'reversal' of judgements and others where it is not so obvious whether the
positive or the negative meaning aspect is predominant.
• Thirdly, it appears that sub-cultural as well as individual values and attitudes
play a role in judgements. The same expression may evoke contexts which by
one group/individual are evaluated in a positive way and by another one in a
negative way. For instance, self-control and refinement are obviously not al
ways valued positively by everybody. The association of types of behaviour
with particular evaluations depends on one's broader value system. (See e.g.
the discussion of keep a stijf upper lip in section 3.2.3.2.).
7. General conclusions
This paper has concentrated on value judgements of various aspects of linguistic

behaviour. It has focused on how speakers of English evaluate types of conver
sations, manner of speech (in terms of speed, loudness, quantity, repetition and
directness), rules governing turn-taking, and ways of introducing and dealing
with topics.
Value judgements have been studied out of context. In other words, con
textual factors which may override the 'usual value' of a term have been left out
of consideration. On the basis of dictionary definitions and comments as well as
of questionnaire responses it has been possible to tease out general tendencies,
which can be summarised as follows:
In nearly all cases metaphors are coined for expressing value judgements.
These can be judgements in terms of relative positions on scales (i.e. the scales
of speed, loudness, quantity, duration, and intensity) and/or in terms of the

positive - negative binary opposition.
Donor domains may simply transfer value judgements (type: put a foot
wrong) or they may furnish terms which acquire a specific value in the recipient
domain. In the latter case one specific aspect of the meaning of the donor term
is focused on and 'transferred'. For instance 'monotony' is transferred in drone
on, 'violence' in horn in, 'volatility' in gasbag, 'force needed' in drag in, etc.
Some donor domains are more pervasive than others for the expression of
LA. Within those, individual expressions may denote specific value judgements.
One such donor domain is that of 'movement', in the sense that speakers are
'moving about' within the 'conversational area' in various ways. Hence, within
such a general domain, some expressions will be 'neutral' while others will be
value-laden. Compare e.g. return to a subject (neutral) with meander
(negative). Another example is that of 'manipulating objects': treat a subject is
neutral, while belabour a subject is negative and come to grips with a subject is
positive.
In most cases value judgements expressed by metaphors are negative.
This has to do with the fact that metaphorical expressions will be coined par
ticularly when L A is perceived as being 'out of the ordinary', 'extreme' in one
way or another, i.e. 'too much or too little of something'.
Some of the expressions examined have suggested the importance of col
location. Further research into probabilities with regard to the syntactic and
semantic context of L A metaphors seems promising.
The questionnaires were presented to young students (aged 18-19). In
some cases it was evident that ideology played a major role in the evaluation. It
would be interesting to compare reactions of different age groups.
Value judgements as they are expressed by means of lexicalised metaphors
reflect attitudes of the speakers of the language towards certain types of dis
course. A cross-linguistic study would be interesting, in that it would reveal
whether the same types of discourse are metaphorised in the same ways, with
the same value judgements, in different languages and cultures.
Acknowledgements
I wish to thank Joan Beal (School of English at the University of Newcastle-

upon-Tyne) for her generous cooperation in handing out the questionnaires to
her students and for devoting some of her lecturing time to their completion. I
am very grateful to the Newcastle students themselves for their invaluable help.
Appendix: the questionnaires
Questionnaire I: set A
I Evaluate the type of verbal event denoted by the following nouns as basically
positive/pleasant, negative/unpleasant or neutral.
Put a tick in the appropriate column.
Table Al
Noun Positive Negative Neutral Blank
1 chat 16 = 80% 4
2 babble 3 15 = 75% 2
3 dialogue 7 13 = 65%
4 chinwag 11 =55% 6 2 1
5 conversation 10 = 5 0 % 10 = 50%
6 jaw 4 10 = 50% 5 1
7 natter 11=55% 4 4 1
8 Smalltalk 2 12 = 60% 6
9 gossip 2 13 = 65% 4 1
10 cackle 1 17 = 85% 1 1
11 pow-wow 7 1 9 3
12 prattle 2 16 = 80% 1 1
13 colloquy 4 2 12 = 60% 2
14 chit-chat 12 = 60% 4 2 2
15 tattle 1 13 = 65% 4 2
16 yap 2 15 = 75% 2 1
17 talk 10 = 50% 10 = 50%
18 idle talk 2 14 = 70% 4
19 trivial talk 2 14 = 70% 4
20 light talk 9 11=55%
II Evaluate the type of verbal event denoted by the following verbs as positive,
negative or neutral.
Table A2
Verb Positive Negative Neutral Blank

1 chatter 13 = 65% 3 4
2 natter 9 9 1 1
3 converse 10 = 5 0 % 10 = 5 0 %
4 palaver 1 13 = 65% 4 2
5 drone on 18 = 90% 1
6 gas 2 16 = 80% 1
7 rattle 1 14 = 70% 5
8 babble 2 14 = 70% 3
9 clack 1 14 = 70% 3 2
10 gabble 1 14 = 70% 4
11 run on 1 15 = 75% 3
12 prattle 1 17 = 85% 1
13 jabber 1 16 = 80% 2
14 patter 4 6 9
15 splutter 1 15 = 75% 3
16 twitter 1 18 = 90% 1
17 talk nineteen to the dozen 3 12 = 60% 4
18 falter 13 = 65% 6
19 burble 4 12 = 60% 3
20 flounder 1 14 = 70% 4
III According to you, do the following descriptions typically apply to a man, or

woman, or both?
Table A3
Description Male Female Both Blank

1 a chatterer 12 = 6 0 % 8
2 a windbag 5 7 7 1
3 a loudmouth 12 = 60% 1 7
4 a rattler 4 6 8 2
5 a talker 3 1 16 = 80%
6 a magpie 1 6 9 4
7 a chatterbox 11 = 55% 9
8 a babbler 5 14 = 7 0 % 1
9 a driveller 3 1 15 = 75% 1
10 a gasbag 3 9 7 1
IV Evaluate the following descriptions as positive, negative or neutral.

Table A4
Description Positive Negative Neutral Blank

1 loquacious 9 2 5 4
2 chatty 17 = 85% 3
3 reticent 1 9 9 1
4 taciturn 2 7 8 3
5 flippant 1 15 = 75% 3 1
6 sparing of words 6 3 11 = 55%
7 vocal 10 = 50% 3 7
8 curt 1 17 = 85% 1 1
9 close-lipped 18 = 9 0 % 2
10 slick 5 10 = 5 0 % 4 1
11 vociferous 8 8 2 2
12 reserved 5 2 12 = 6 0 % 1
13 glib 1 15 = 75% 3 1
14 fluent 2 0 : = 100%
15 voluble 7 5 5 3
16 prosy 1 10 = 50% 7 2
17 tight-lipped 1 17 = 85% 1 1
18 silent 1 5 14 = 7 0 %
19 tongue-tied 1 15 = 75% 4
20 chatty 15 = 75% 1 3 1
V Evaluate the following expressions as positive, negative or neutral.

Table A5
Expression Positive Negative Neutral Blank

1 bring up a subject 13 = 65% 7
2 drag i n a subject 18 = 9 0 % 2
3 plunge into a subject 14 = 7 0 % 2 4
4 raise a subject 15 = 75% 1 4
5 table a subject 6 2 11 = 5 5 % 1
6 rake up a subject 16 = 80% 3 1
7 get onto a subject 4 2 14 = 7 0 %
8 pass over a subject 1 10 = 50% 9
9 tackle a subject 16 = 80% 2 1 1
10 stick to a subject 9 2 8 1
VI According to you, do the following sentences express a positive, negative or

neutral evaluation of X's behaviour?
Table A6
Sentences Positive Negative Neutral Blank
1 X told me flat out 14 = 70% 2 2 2
2 X hedged his answer 18 = 90% 2
3 X used plain words 14 = 70% 4 2
4 X talked round the issue 2 14 = 70% 3 1
5 X spoke his mind 16 = 80% 1 1 2
6 X called a spade a spade 12 = 60% 1 5 2
7 X gave an evasive answer 17 = 85% 1 2
8 X sounded off his opinion 4 10 = 50% 5 1
9 X told me straight 18 = 90% 1 1
10 X spoke baldly 2 13 = 65% 4 1
VII Evaluate the following expressions as positive, negative or neutral.

Table A7
1 guard one's tongue 15 = 75% 1 3
2 let something slip 16 = 80% 3
3 spit out one's feelings 9 7 3
4 watch one's words 13 = 65% 6
5 bite one's lip 7 6 6
Questionnaire I: set 
The nouns below denote a verbal event. Which adjectives from the following sets
do you primarily associate with them: (A) light or heavy or neither; (B) deep or
shallow or neither. Tick twice: choose one item from set A and one item from set
B. Put ticks in the appropriate columns.
Table BI
A 
Noun Light Heavy Neither Deep Shallow Neither Blank
1 chat 19 1 1 12 7
= 95% = 60%
2 babble 17 3 16 4
= 85% = 80%
3 dialogue 18 2 18 2
= 90% = 90%
4 chinwag 18 1 1 3 16 1
= 90% = 80%
5 conversation 9 11 9 11
= 55% = 55%
6 jaw 11 5 4 4 11 5
= 55% = 55%
7 natter 19 1 2 17 1
= 95% = 85%
8 Smalltalk 17 3 1 15 4
= 85% = 75%
9 gossip 17 3 3 15 2
= 85% = 75%
10 cackle 13 1 6 2 8 10
= 65% = 50%
11 pow-wow 3 10 7 9 4 7
= 50%
12 prattle 18 1 1 2 17 1
= 90% = 85%
13 colloquy 1 14 5 13 1 6
= 70% = 65%
14 chit-chat 20 1 17 2
= 100% = 85%
15 tattle 18 1 1 1 17 2
= 90% = 85%
16 yap 17 3 2 15 3
= 85% = 75%
17 talk 1 19 5 15
= 95% = 75%
18 idle talk 17 1 2 2 16 2
= 85% = 80%
19 trivial talk 13 1 6 2 14 4
= 65% = 70%
20 light talk 16 4 2 11 7
= 80% = 55%
II The verbs below denote a verbal event. Which adjectives from the following sets
do you primarily associate with them: (A) loud or soft or neither; (B) fast or slow
or neither. Tick twice: choose one item from set A and one item from set B. Put
ticks in the appropriate columns.
Table B2
A 
Verb Loud Soft Neither Fast Slow Neither Blank
1 chatter 10 4 6 20
= 50% = 100%
2 natter 5 8 7 15 5
= 75%
3 converse 1 14 5 17 3
= 70% = 85%
4 palaver 16 1 2 10 5 4 1
= 80% = 50%
5 drone on 7 6 7 19 1
= 95%
6 gas 7 2 11 14 1 5
= 55% = 70%
7 rattle 11 2 7 20
= 55% = 100%
8 babble 9 6 5 20
= 100%
9 clack 14 2 4 8 2 11
= 70% = 55%
10 gabble 12 1 7 19 1
= 60% = 95%
11 run on 4 4 12 13 3 4
= 60% = 65%
12 prattle 4 4 12 18 2
= 60% = 90%
13 jabber 14 6 19 1
= 70% = 95%
14 patter 1 12 7 15 2 3
= 60% = 75%
15 splutter 7 4 9 9 1 10
= 50%
16 twitter 1 15 4 19 1
= 75% = 95%
17 talk nineteen 11 1 8 17 2 1
to the dozen = 55% = 85%
18 falter 12 8 1 15 4
= 60% = 75%
19 burble 3 11 6 9 6 5
= 55%
20 flounder 1 8 11 14 6
= 55% = 70%
III Evaluate the following expressions as positive, negative or neutral.

Table B3

1 belabour a subject 1 17 = 85% 2
2 exhaust a subject 3 9 8
3 hold forth on a subject 11 = 5 5 % 3 6
4 pick over a subject 4 10 = 5 0 % 6
5 dilate on a subject 4 5 9 1
6 go into a subject 17 = 85% 3
7 harp on about a subject 2 17 = 85% 1
8 talk out a subject 13 = 65% 7
9 carry on about a subject 3 15 = 7 5 % 2
10 thrash out a subject 3 12 = 6 0 % 4 1
11 dwell on a subject 7 7 6
12 keep on about a subject 3 15 = 7 5 % 2
13 descant on a subject 4 3 12 = 6 0 % 1
14 expand on a subject 20 = 100%
15 be on about a subject 2 7 11 = 5 5 %
IV Indicate whether the following expressions are typically associated with men,
women or both.
Table B4
Verb Male Female Both Blank

1 chatter 16 = 80% 4
2 palaver 3 2 14 = 7 0 % 1
3 rattle 1 8 11=55%
4 clack 2 14 = 7 0 % 4
5 natter 18 = 9 0 % 2
6 converse 5 15 = 75%
7 drone on 9 2 9
8 gas 12 = 6 0 % 7 1
9 babble 9 11 = 5 5 %
10 twitter 16 = 80% 3 1
V Evaluate the following expressions as positive, negative or neutral.

Table B5

1 drag up a subject 1 18 = 9 0 % 1
2 slide over a subject 1 10 = 50% 9
3 go o f f the point 2 13 = 65% 5
4 touch on a subject 9 1 10 = 5 0 %
5 stay with a subject 16 = 80% 4
6 slur over a subject 19 = 95% 1
7 cover the ground 16 = 80% 4
8 steer the conversation into a subject 12 = 60% 8
9 meander 11 = 5 5 % 9
10 stick to a subject 18 = 90% 2
VI According to you, do the following sentences express a positive, negative or

neutral evaluation of X ' s behaviour?
Put a tick in the aprropriate column.
Table B6
Sentences Positive Negative Neutral Blank

1 X beat about the bush 1 17 = 85% 2
2 X told me in plain terms 18 = 90% 2
3 X did not mince his words 12 = 60% 3 5
4 X gave a straight-out answer 19 = 95% 1
5 X wrapped up his meaning 1 17 = 85% 2
6 X told me point-blank 11 =55% 2 7
7 X polished his speech 9 5 5 1
8 X spoke badly 7 9 4
9 X plastered over the issue 1 19 = 95%
10 X spoke plainly 13 = 65% 7
VII Evaluate the following expressions as positive, negative or neutral.

Table B7
1 blow off steam 12 = 6 0 % 3 5
2 have a loose tongue 1 18 = 90% 1
3 throw something out 5 2 13 = 65%
4 bridle one's tongue 7 3 10 = 50%
5 keep a stiff upper lip 6 9 5
Questionnaire III
The expressions below all denote verbal events, types of verbal behaviour, ways of
talking, etc.
a) According to you, do they express a positive, a negative or a neutral value judge
ment? Put a tick in the appropriate slot.
b) Briefly explain why you judge the expression to be positive, negative or neutral.
c) Use the expression in a sentence to show the kind of context which you associate
with it.
1. a chinwag
a) Positive □ Negative □ Neutral □
b) Why?
c) Example sentence
2. to watch one's words

b) Why?
c) Example sentence
3. to polish one's speech

b) Why?
c) Example sentence
4. to r u n on
b) Why?
c) Example sentence
5. to keep a stiff upper lip

b) Why?
c) Example sentence
6. to air (one's ideas)

b) Why?
c) Example sentence
7. to twitter
b) Why?
c) Example sentence
8. a refined way of speaking

b) Why?
c) Example sentence
9. a chatterbox
b) Why?
c) Example sentence
10. not to mince one's words

b) Why?
c) Example sentence
11. a) do you know the word colloquy? yes □ no □

b) I f yes,agreeable
do you associate with colloquy a verbal event which is
□
i) disagreeable
□
neither
□
ii) formal
□
informal
□
Levels of Metaphorization
The Case of Put
Paul Pauwels
Catholic Flemish Institute of Higher Education
1. Aims and means
Within the framework of a cognitive approach to metaphor (Lakoff and Johnson

1980, Johnson 1987, Lakoff 1987, Rudzka-Ostyn 1988b), metaphor has been
looked at from two basic angles. It has been described in terms of transfer from
one cognitive (donor) domain to a another (recipient) domain, while especially
Johnson (1987) focuses on the grounding of such mappings in terms of image
schemata, which he defines as "a recurrent pattern, shape and regularity in, or of
[our] ongoing ordering activities" (1987: 126), and which constitute the basis
for any act of conceptualization, of which metaphor is but one example. In
Pauwels & Simon-Vandenbergen (this volume), we demonstrate that metaphors
can be situated on a scale from more general to more specific donor domains.
For Kövecses (1986) there are only two classes of metaphor. "Basic level meta
phors" use "basic level concepts that are linked more directly to experience,
concepts that are information-rich and rich in conventional mental imagery"
(1986: 117), while "constitutive metaphors", are based on "superordinate con
cepts" (1986: 116-117). Brugman (1990) also addresses the issue and points out
that the notion of image schema itself is not without problems.1
Against this background this paper will focus on degrees of metaphoricity:
to what extent are expressions conceived as metaphorical by language users,
and on which basis. To this end, a first part will introduce the notion of
'recoverability' and discuss how it relates to metaphoricity and specificity. The
second part of the paper is an analysis of a limited corpus of metaphorical ex
pressions (42 items). All the expressions have two things in common: the re
cipient domain of 'linguistic action' (henceforth L A ) , and the expression put,
which situates the metaphors in related donor domains. The analysis provides
126 P A U L PAUWELS
further insight into the interaction of different donor domains, the existence of
different levels of metaphorization, and the role of image-schematic structures.
2. Recoverability, specificity, metaphoricity
Because so many of the concepts that are important to us are either ab
stract or not clearly delineated in our experience [...] , we need to get a
grasp on them by means of other concepts we understand in clearer terms.
(Lakoff and Johnson, 1980: 115)
From the above, it is clear that an essential part of metaphor is the presence of a
salient concept, which is used to structure a non-salient experience. Salience can
be a consequence of concreteness, visual impact, specificity, but also of perva
siveness and familiarity. It is important to note that salience is not an absolute
property, inherent in a concept, but that it is relative, in so far as it depends on
the experience of the language users who constitute the linguistic community. A
linguistic community, moreover, is never homogeneous, and the experiences of
its members may differ substantially, which is why salience, too, may differ for
individual language users. Furthermore, linguistic communities and their lan
guage evolve, and concepts may gain or lose salience. Because salience is rela
tive, metaphor, which relies on salient concepts, is relative too, and it becomes
important that one should think in terms of metaphoricity, i.e. metaphor as a
matter of degree.
In my account, I propose to use the notion of 'recoverability of donor do
main/situation' to describe the extent to which a specific donor concept, and the
way in which the metaphorical expression relates to it, are salient for a linguistic
community.
One problem which any investigation of a concept relying on the knowl
edge of the 'average language user' is confronted with, is verifiability. For the
purposes of this paper, I will take it that the expressions and definitions con
tained in a collection of explanatory dictionaries such as COBUILD, OALDCE,
LDOCE and Webster's provide a picture which certainly does not fall short of
the average language user's lexical competence. More specifically, I take it to be
the case that concepts and meanings not represented there, constitute specialist
knowledge, available to some language users, but certainly not to the majority
of the present-day linguistic community of native speakers of English. This leads
me to posit two basic kinds of metaphorical expression: those in which the
LEVELS OF METAPHORIZATION 127
metaphor relies on specialist knowledge, versus those where the concepts are
fully salient. When faced with metaphors relying on non-salient concepts, the
average language user will either (i) perceive/learn them as nonfigurative, or (ii)
perceive them as figurative and use them, or try to interpret them, on the basis
of knowledge available to him. In the second case he may well recover different
concepts and donor domains to give meaning to the metaphor than were at
work in the original - from a diachronic point of view - metaphorizationprocess.
But even when the concept used in a metaphor can safely be con
sidered as salient and transparent, the problem of verifiability remains: the
knowledge contained in general explanatory dictionaries can only provide an
insight in the possibilities which are open to most of the language community,
but does not show which of that knowledge will, or indeed needs to, be acti
vated in actual situations. It is likely that the use of a metaphorical expression in
a familiar pattern often precludes the need to recover its non-figurative basis,
whereas for example an ironic or punning use reactivates the metaphor and its
non-figurative basis, and heightens its metaphoricity.
In the remainder of this section, I will try to demonstrate how different
degrees of metaphoricity correspond to the degree to which, and the way in
which, a donor domain/situation can be recovered. Some of the examples used
here were taken from Pauwels & Simon-Vandenbergen (this volume).
A metaphor like bandy words with 'argue, quarrel with' has, for present-
day users of English, no recoverable donor domain. Bandy used to be a ball-
game like tennis, and the meaning the metaphor relies on, i.e. 'throw or strike a
ball to and fro, as in the games of tennis and bandy' (OED) is, like other literal
meanings of bandy, indicated as archaic; in our set of explanatory dictionaries
bandy is only mentioned in its LA-meaning. This metaphor would, in conven
tional terminology, be called 'dead' - the primary meaning of the expression
now lies in the L A domain, and the expression does not point the language user
to another domain. 2
The picture is different for expressions like tear off a strip and kick
against the pricks, which may not be completely recoverable as military degra
dation resp. the futile protest of oxen against the goad, but which are at least
partly recoverable. Both clearly fit into the more general donor domain of vio
lent actions, which is a frequent donor for LA.
An expression of the type get it straight from the horse 's mouth is in all
probability recoverable to a greater degree: it benefits from the presence of an
other expression requiring the same encyclopaedic knowledge (looking a gift
128 P A U L PAUWELS
horse in the mouth) and maybe also from the fact that the donor domain has a
greater degree of salience than in the cases described above.
Some expressions stand to be misinterpreted because the lack of specialist
knowledge may lead to wrong inferences. One such example is burden (of an
argument, story, message...), meaning the 'main subject or point' (LDOCE).
(1) The burden of his message did not strike me as being very original.
(COBUILD)
Here the meaning which should be recovered is not 'something that is carried, a
load' (OALDCE). This appears on closer scrutiny, since this meaning would
involve a reversal in value judgement from donor domain to recipient domain -
something which is unlikely, as is shown by the data examined in Pauwels &
Simon-Vandenbergen (this volume) - as well as a reversal of roles in the situa
tion: the burden in the metaphor has a neutral to positive connotation, and it is
described as being at the basis of the argument. Rather, the meaning which
should be recovered in that of bourdon 'bass, undersong, or accompaniment
'(OED), from which it was derived via 'the refrain or chorus of a song'(OED)
via figurative uses like (2) to refer to 'the chief theme, leading idea, prevailing
sentiment' (OED) of an LA.
(2) This was the burthen of all his song - 'Everything which we could
reasonably hope from war, would be obtained from treaty.' (OED)
It is not unlikely that there are other metaphors for which this kind of mistaken
identity is possible. The question which needs to be investigated is in how far
this influences metaphoricity.3
The viability of these metaphors probably depends to a high degree on the
way in which they fit into more general ways of conceptualizing, and in how far
this is recognizable. Another example may illustrate the point.
In an expression like lace into someone 'attack someone with words' the
use of lace to mean 'whip' may not be immediately recoverable. The presence
of other, similar, collocations in expressions like tear into someone and even
lam into someone - where lam, too, is intransparent - may help the language
user recover the relevant donor domain of violent action. The actual donor
situation might remain beyond his grasp, but the collocation with into someone
will probably serve to situate the expression with other, similar ones.
With blow the gaff 'let some secret be known' (LDOCE) the situation is
different. For the average language user, the original donor domain of sailing is
far from recoverable,4 and although a link to the domain of breathing may pro-
vide some ground for understanding (blowing as contrasted with keeping one's
mouth shut), it does not go a long way. Individuals with a specialist knowledge
of sailing will find the element gaff less of a problem. For them, the metaphoricity
of the expression is higher. A similar case, also from the domain of
ships, is bilge 'foolish talk' (LDOCE). In this case, though, the item does not
form part of a larger pattern, nor is there a fixed collocation, so the lay person
has no basis for a possible (re)interpretation.
Of course, quite a few metaphors have easily recoverable donor domains.
Metaphors like eat one 's words and kick someone around are clearly and com
pletely recoverable from their donor domains of resp. eating and fighting.
Moreover, it is unlikely that these metaphors will ever lose their recoverability,
since they refer to basic bodily experiences, and as such have a near-ontological
basis.
The picture which emerges from the above can be outlined as follows.
Metaphors originate in situations/domains which are recoverable for the lan
guage users. Historical change may cause a change in salience of those concep
tual situations/domains, so that the metaphor becomes irrecoverable (dead, in
traditional terminology). In cases where the original literal meaning of a given
item falls out of use, the process of 'dying' can only be slowed down through
the presence of a (large) number of current metaphors drawn from the same
domain, which provide clues for each other's interpretation.
However, historical change can also have the opposite effect. Domains
which used to be 'specialist' can acquire the necessary overall salience to war
rant recoverability. The growing importance of computer and computer science
has given rise to a number of metaphors with this domain for a donor, and most
of these should be classified as recoverable. Feedback, for example, was origi
nally a metaphorization of the use of energy in an (electric) circuit: 'the return
of a fraction of the output signal from one stage of a circuit, amplifier etc. to the
input of the same or preceding stage ... to increase, decrease the amplification
etc.; also a signal so returned'(OED). From there, it was extended to the com
puter domain, where it later came to metaphorize the flow of (electronically
coded) information. In a final stage, it has become a metaphor for L A , describ
ing 'a response' (OED) as in (3).
(3) We began to get a fairly good feedback from most people who
know about it, and it looked as though the concerts would be
good scenes. (OED)
130 PAUL PAUWELS
Notice how the purpose described in the first of the above OED definitions is
paralleled in the aim of the L A of giving feedback, viz. the modification of the
behaviour of the system/person addressed. Further, it is interesting to see how
the metaphorization in this case also relies on an existing and at the time already
well-established pattern of meaning extension for the verb feed, which can mean
'gratify, minister to the demands of (a person's vanity, desire of vengeance, or
other passion); to sustain or comfort a person with (usu. fallacious) hopes'
{OED). In this usage, exemplified by (4), feeding also describes an action which
affects the mind of the addressee.
(4) He feeds himself and his friends with hopes of a speedy peace.
{OED)
Other expressions from the domain of computer technology, like input, output,
interface, etc. have been extended in similar directions. Together with the in
creasing popularity of the domain (or because of it), there occurs an increase in
the number of metaphors which relate to it as a donor domain.
The metaphors we have been discussing up to this moment all rely on a
specific donor situation, which may or may not be recoverable. In Kövecses'
(1986) terms, these would all be "basic level metaphors" which are "linked more
directly to experience". Kövecses' other class of metaphors, "constitutive meta
phors", are based on "superordinate concepts". Lie in one's throat/teeth 'lie
shamelessly' {LDOCE), as in (5), would be a case in point.
(5) Whoever asserts that lies in his throat. {OED)
The relevant domain for this metaphor is that of movement in space. More spe
cifically, any explanation would probably refer to image-schematic notions like
container, path, centre-periphery, as distinguished by Johnson (1989: 126). The
key factor in the metaphor is that the origin of the L A is situated along the path
of speech, but nearer the periphery of the body-container than one would expect
of something which, according to our cultural conventions, should preferably be
kept hidden. At this stage, again, the question of metaphoricity arises. Indeed,
the metaphor can be explained, but in how far the image it relies on is recover
able for the average language user seems a moot point.
From the foregoing, it should be clear that Kövecses' (1986) binary clas
sification is insufficient to fully grasp the complexity of different types of meta
phor. Although it is probably an accurate picture of the two basic ways in which
original metaphors can arise, it fails as a model to explain how conventionalized
metaphors are used and experienced, and how, and to what extent, the average
language user recovers which basis for a metaphorical interpretation. In what
follows, I have attempted a more systematic approach to recoverability. To this

aim, I have selected a restricted number of metaphors which in their origin can
be traced to one specific action on the domain of manipulating objects (putting)
and which have a fairly specific recipient domain (LA) in common. In this way I
hope to be able to demonstrate how, and at which level of specificity, transfers
are made, and how this affects the recoverability of the donor domain and the
metaphoricity of the expressions.
3. һ put-metaphors
3.1 The corpus
For this investigation use was made of the LDOCE-bastd corpus of L A meta
phors as described in Vanparys (this volume), which was then further extended
on the basis of other explanatory dictionaries: Oxford Advanced Learner's Dic
tionary of Current English (1989 4 ) (OALDCE), Collins COBUILD English
Language Dictionary (1987) (COBUILD) and the Longman Dictionary of
Contemporary English (1987 2 ) (LDOCE).
The set of expressions under investigation is restricted in two ways. First,
there is the syntactic restriction to verbal expressions. Metaphorization patterns
for verbials, nominals and adjectives will probably exhibit differences, but these
are not our concern here. A second restriction imposed for this investigation of
a semantic nature. As in the main corpus, we chose to only investigate LA-
metaphors. Rather than focus on the variety of donor-domains used in meta-
phorizing LA, 5 we chose to focus on the extensions from one central scene
which is lexicalized by the verb put.
In all, the corpus consists of 87 sentences exemplifying 42 expressions.
The corpus is limited, and as such we cannot hope to provide a comprehensive
picture of metaphorization. Still, it is well suited to demonstrate the complexity
of metaphorical mappings, even in this limited set, and further highlight the
problems of recoverability and metaphoricity.
3.2 Main senses of'put
This section will provide a survey of the main senses of put in the spatial do
main, which is shown by a frequency count6 to be the primary domain for put.
132 PAUL PAUWELS
We will first describe the relations between those meanings, before attempting
to describe the extensions into the L A domain.
Basically, put can be described as a three place predicate which profiles
two trajector-landmark relationships (Rudzka 1988b: 508, henceforth TR and
L M ) . The basic patterns are exemplified by (6-13).
(6) She put the book on the table. (OALDCE)
(7) Did you put sugar in my tea? (OALDCE)
(8) She put the baby in its cot.
(9) It's time to put the baby to bed. (OALDCE)
(10) She put her arm around his shoulders. (OALDCE)
(11) Put down that knife before you hurt someone. (OALDCE)
(12) She put her hat and coat on. (OALDCE)
(13) She put some makeup on.
The prototype7 for put is exemplified by (6). We can try and describe this pat
tern in terms of basic image schemata. First of all, it singles out an entity as a
(manipulated) object. This object is presupposed to be something the agent is
holding in his hands - presupposed, because this is not something which is fo
cused by the verb. Put describes a movement away from the agent, which entails
a path, a direction, and two types of location i.e. an origin (which is not pro
filed) and a destination. The agent is in control of the object until it has reached
its destination. This destination is focused separately in the syntactic structure,
and is expressed by means of a prepositional phrase describing a location in the
spatial domain. The first TR-LM relationship holds between the agent and the
object, the second between the object and the destination. Two elements from
this prototype, i.e. the presence of a human agent as primary TR and the move
ment in space of a secondary TR to a secondary L M , constitute the common
core of the basic meanings.
The pattern exemplified by (7) is characterized by a difference in the na
ture of the secondary TR and L M . The sugar referred to can be (but need not
be) a substance rather than a single solid object, and the manipulation would
then be indirect, by means of an instrument such as a spoon. Also L M 2 is non-
prototypical, and the final relationship between TR2 and L M 2 is not one of
support, but rather one of 'merging'.
In (8) the secondary TR/primary L M is also different in nature, but this
does not affect the relationship with the primary TR. A baby can be conceptual
ized as being sufficiently small and manageable to be lifted, carried and left
somewhere in the same way as an object. This pattern with a human being as
LEVELS OF ΜETAPHORIZATION 133
primary L M is further extended via human beings not in physical control of their
bodies (asleep, drunk, ill or unconscious) and hence manipulatable, to human
beings not in mental control of themselves (e.g. prisoners). Example (9) shows
another extension of the pattern, which is metonymic in nature. The ritual of
putting a baby to sleep is referred to by means of a description of the central
element in the process.
Sentence (10) exemplifies another highly frequent pattern in the use of put
in the spatial domain. In this case, the primary L M is not an object, but a body-
part. The extension is quite straightforward, since hand and arm are both central
elements of the prototypical act of manipulating as conceptualized by put. Still,
the movement of other bodyparts such as head and foot is also conceptualized
by means of put.
Although the meanings illustrated by (7,8,10) are central, I would argue
they are also extensions from the prototype in that they all conceive the primary
L M as an object, which involves crossing a (sub)domain boundary. The question
is now in how far these domain boundaries are salient to the average language
user in this context. In view of the frequency and the well established nature of
these patterns, it is likely that metaphoricity is low. Still, the use of put strength
ens the conceptualization of a variety of primary LMs as manipulatable objects.
In cases like (10), the inherent duality of the human being between the mind as
the origin of agency and the body as the primary instrument of that agency is
exploited and emphasized.
In examples (11-13), LM2 is not expressed. According to Rudzka-Ostyn
(1988b) we are dealing with a 'covert' or 'hidden' landmark, which "does not
mean that the landmark is absent from our conceptualization of the scene; it
simply means that it is either too obvious or not worth bothering about" (p520).
Still, it should be noted that in all of these cases there is a particle/adverb further
specifying the verb. There are two main subtypes in this set. In (11), the hidden
L M can be inferred as being the agent (the primary TR) who functions as source
of the movement. In (12) and (13), it is the agent functioning as destination of
the movement. Example (13) can be seen as a further extension of the pattern in
(12), in that the putting involves a substance rather than an object. Still, (12) too
is already further away from the prototype, since it involves a more complex act
of manipulation, a nonprototypical secondary L M and a more complex final
relationship between the secondary TR and L M .
In sum, it should be clear that put has several (non)figurative extensions in
the spatial domain, and even in the more specific subdomain of manipulating
134 P A U L PAUWELS
objects, several subtypes can be distinguished. The variation can be accounted

for on the basis of the variations in the secondary TR and LM, and the relation
ship between them.
3.3 Metaphorical extensions into the LA domain
For each of the expressions under investigation, I will try to determine the ex
tent to which a donor situation can be recovered from the spatial domain, and
the relevance of the prototypical meaning of put. I hope to demonstrate that
transfers occur at different levels, and that metaphor is a relative notion.
Figure la
Put one's cards on the table 'clearly state one's intentions' (LDOCE). A unifying scene
in the donor domain at the basis of the metaphorical transfer is unequivocally recov
erable. (Full lines: in focus; dotted lines: out of focus; T: transfer; R: recovery.)
The main distinction to be made is that between expressions which rely on one
clearly recoverable donor scene (Fig. la), and those that can only be recovered
from different (superordinate) donor domains (Fig. lb). In the second group,
present-day average language users have no access to a nonfigurative scene
where the elements present in the metaphor are unified. In the first group, we
find expressions which are recoverable from the domain of the manipulation of
objects (3.3.1), specific subdomains of this more general domain (3.3.2.), other
domains which have been metaphorized through the domain of manipulating
objects (3.3.3), especially the domain of bodily movement (3.3.4).
Figure lb
Put in for something 'apply for ' (LDOCE) There is no single unifying scene on which
the metaphor relies, for present day users. Recovery in various (superordinate) do
mains.
3.3.1 Donor domain: manipulating objects
Four expressions rely exclusively on the domain of manipulation of objects: put

a question (to so) 'ask (officially) (for something) to be considered' (LDOCE),
put the case that 'suggest that' (LDOCE), put it (to someone) that 'suggest,
invite someone to consider that' (LDOCE) and put something (to someone)
'express, communicate or submit' (OALDCE).
These expressions only use part of the prototype established in section
3.2. They do not focus on the destination involved in the act of putting, which
shows either in the absence of a prepositional phrase expressing the secondary
L M , or in the combination with a preposition which is not normally associated
with the verb, and falls outside the metaphorization: the preposition to is the
prototypical preposition to indicate the addressee on the LA-scene. Although it
is metaphorical, and although it metaphorizes the addressee as a destination, it is
not part of the metaphorization achieved by the use of put. The difference with
136 P A U L PAUWELS
other expressions, where the preposition does belong to the metaphorization,

will be clear from what follows.
The expression put something to someone differs from the other three, as
is more clearly illustrated by example (14).
( 14) Your proposal will be put to the board of directors.
Here, we are dealing with an act of reporting, which involves the 'expressing,
communicating or submitting' of a (possibly written) L A by an intermediary to a
third party. The scene here is different from the standard L A scene: whether the
'proposal' is delivered orally or in a written way, it is clearly something which
has been prepared prior to the 'putting'. It is much more straightforward to
conceive of proposals (and the like) as finished products, on which other opera
tions can be performed. The distance between the described L A and the describ
ing act is less great than in the other three cases. Especially in the case of writ
ten communication the expression benefits from a metonymic link: the report
will in all probability be put before the secondary addressee who can then read
it.8
In the other three cases, completion is a more central element of the meta
phorization. The L A is described as completed, transferred away from the
speaker and put before the addressee. Although the speaker relinquishes physi
cal control over the 'object' in doing so, he at the same time challenges the ad
dressee to take it up, by moving it into the open. In conceptualizing the secon
dary L M by means of to, the metaphors introduce a contrast with the prototype,
where L M 2 is locational, rather than directional. Examples (15) and (16) illus
trate the point.
(15) The lawyer put several questions to the witness. (LDOCE)
(16) I put it to you that you haven't told the whole truth. (LDOCE)
The key point is, that in order to take up the question/suggestion that is being
put to him/her, the addressee will have to make an effort and move out of
his/her selfcontained domain of silence. The addressee will have to meet the
speaker, as it were, halfway. This becomes even clearer if one compares the
value of put to with give: the relevant difference is that in the case of put to the
object is not moved into the addressee's domain of control.
Finally, put is also used in combination with manner adverbials:
(17) She is, how shall I put it, not exactly fat, but rather well built for
her age. (LDOCE)
(18) At 6'4" she's tall, to put it mildly. (OALDCE)
(19) To put it bluntly, you're fired. (OALDCE)

(20) His ideas were cleverly put. {LDOCE)
Except for the vague how in (17), none of the manner adverbials is combinable
with put in its literal use. They are not part of the metaphorization achieved by
put, but make separate contribution - metaphorical as in bluntly, or literal as in
cleverly. The function of, or the motivation for the use of put in these cases is
not immediately obvious, since there are variants with LA-verbs for most of
these (e.g. (19b)).
(19b) He told them bluntly what was acceptable and what not.
{COBUILD)
The only possibly relevant difference which exists with the put-version - apart
from the difference in syntactic structure - is that the object of putting remains,
and can be scrutinized, whereas L A is volatile. It remains questionable whether
this element of the metaphorization is really salient.
3.3.2 Subdomains as donor
Some expressions can be recovered from a more specific subdomain of manipu

lation of objects. In put one's cards on the table, the prototypical act of putting
is specified by means of a specific type of secondary TR, as in (21).
(21) We can only reach an agreement if we both put our cards on the
table. {OALDCE)
The metaphor clearly relies on the possibility in some card games to play an
open hand. This is often accompanied by the player's statement of his intention.
In some card games this action is usually performed by one player only, who
then lets the others try to trap him. The metaphor focuses on the openness
achieved by the action of 'being completely honest, saying openly what one
intends to do' {LDOCE).
Another set of expressions are less clearly specified. They are: put on the
agony, put the blame on someone, put the mouth on someone, put emphasis on
something, put someone right/straight, put the record straight, put something
up, put someone up (for something), and put across/over oneself/one's mean
ing. The secondary TR is situated outside the physical domain, and refers to
L A , to a person (speaker or subject of L A ) , or to an abstract notion like blame,
or agony. The secondary L M , too, is unhelpful in the recovery of the expres
sion, since it either also refers to LA, and participants in the L A scene, or is
simply not mentioned. The expressions are less directly recoverable, since the
138 PAUL PAUWELS
information available in the verb + preposition/ particle/ adverb combination

often leads to more than one possible donor domain.
In the set with put on, reference is made to four different donor domains
at least. In put the blame on someone 'blame' {LDOCE) the L A is conceived in
terms of putting a burden on someone, which is a more general way of con
ceiving responsibility, guilt etc. (cf. metaphorical uses of weight, burden). The
mention of an affected person as secondary L M in this case also promotes the
recoverability of the donor situation.
The expression put the mouth on someone 'make someone's actions or
attempts unsuccessful by saying he's doing very well' (LDOCE), there is at first
sight a metonymic basis, with mouth referring to LA. In this reading the L A
would also be conceived as a burden. Here, though, salience is probably less
strong, because there is no general conception of language as a burden, espe
cially not if it is appreciation. The praise referred to is probably ironic, and irony
does have a restrictive influence on one's actions. In a diachronic survey of L A
metaphors including the concept mouth, Goossens (1993b) demonstrates that
most of these metaphors have a biblical origin, and that the earliest uses were
mainly metonymic. Thinking along these lines, we could hypothesize that the
action referred to in the metaphor is actually the Judas kiss; on the surface a
mark of respect for his master, it actually was a betrayal which led to the arrest
of Jesus. This situation is clearly marked by the same kind of contradiction of
surface value (positive) and underlying intent (negative) as the metaphor. In this
reading, we would be dealing with a transfer from a domain which involves a
different use of an articulator (kissing); it is a subdomain of the movement of
bodyparts which is conceived in terms of manipulating objects (see also 3.3.4.).
The question arises in how far these interpretations are recoverable for the aver
age language user. There is no nonfigurative interpretation for putting the
mouth on in present-day English, and the hypothetical scene of the Judas kiss
does not immediately come to mind. The alternative, viz. conceiving praise as a
burden might be more readily recoverable, but it is only weakly signalled in the
expression.
Put emphasis on something 'emphasize' {LDOCE) relies on a situation of
marking the surface of an object with some substance or sign. The metaphor
dissociates emphasis and emphasized, and in such way allows for e.g. a quanti
fication of emphasis {a lot of emphasis, how much emphasis ...). The meaning
which should be recovered (marking) is already an extension from the proto
type: the secondary TR is not a prototypical object, and the relation between
secondary TR and L M is more complex. The path of extension probably runs

via the domain of nonverbal communication, in which case the process should
be described as metaphor from metonymy (see Goossens (this volume), for an
in-depth discussion of the contribution of metonymy to metaphor).
Put on the agony poses a problem for unequivocal recovery: there are
several equally plausible donor domains with respect to the definition 'say that
one's sufferings, feelings etc. are stronger, greater or worse than they really are,
esp. in order to influence others' (LDOCE). Putting on could be taken to refer
to the putting on of a mask. This action hides the face, where feelings usually
show. Similarly, one could construe the meaning on the basis of the situation of
putting on make up, which can be used to hide or enhance. Both methods are
used in theatrical contexts to help actors convey certain emotions. The same
context provides another explanation. A theatre company can be said to put on
a play - itself a metonymy based on the location of the action on the stage. From
this derives a metaphor put on an act 'pretend' (OALDCE), of which put on the
agony can be considered as a specification. Conversely, the metaphor could be
taken to rely on a completely different kind of putting on, namely that of setting
in motion some kind of machine. The emphasis would then be on the control the
persons exert on their so-called emotions. There is evidence for the different
readings in other, parallel, expressions: put on a brave face seems to support the
first interpretation, whereas turn on the agony (LDOCE), turn on one's charms
(COBUILD) and turn on the waterworks 'cry in order to get attention'
(COBUILD) support the third interpretation. Recoverability will undoubtedly be
influenced by the (lack of) awareness of these related expressions. Even the
donor situation of putting on a burden cannot be ruled out, for it is supported by
another variant of the expression, namely pile on the agony. One reason for the
variety of interpretations is the absence of a clear landmark. Contrary to what
was said earlier (section 3.2.), the hidden landmark seems neither obvious nor
dispensable in this case. Of these five readings, only the final one directly relates
to the prototype. The first two relate to well established extensions into the do
main of manipulating objects/substances, the third is a case of metaphor from
metonymy, the fourth one relies on an intermediary metaphorization of setting in
motion of some kind of machine (see also section 3.3.3.). Fig. 2 provides a sur
vey of the different possible paths of recovery.
PAUL PAUWELS
Figure 2
Put on the agony 'to say that one's sufferings, feelings etc. are stronger, greater, or
worse than they really are, esp. in order to influence others ' (LDOCE)
There are four possible paths of recovery, viz.
(1) putting on a mask, hence hiding;
(2) putting on a play, hence acting;
(3) putting on a machine; hence activating;
(4) putting on an extra measure; hence exaggerating.
These rely on two different strategies of recovery, viz.
(a) that which is being 'put on': a mask (1), or a machine (3);
(b) that which it is being 'put on ': a stage (2) or a pile (4).
Put the record straight 'give a correct account of facts or events' (COBUILD)
and put someone right/straight 'give correct information to someone who has a
wrong opinion, often rather sharply'(LDOCE) both rely on the donor domain of
arranging objects. As in arranging, what is correct or right depends on the
judgement of the agent. This is especially relevant in the second metaphor,
where control over the hearer-object is in focus. On top of that, this expression
has another possible donor situation where persons are actually manipulated;
soldiers on parade (or schoolchildren for that matter) are put straight in the
literal sense.9 This may be done by hand, or by means of language, which again
provides a metonymic basis for the metaphor. In the latter interpretation, the
metaphor is once again based on an extension from the prototype in which the
secondary TR is a person not in control of his own body.
Put someone up (for something) 'suggest (someone) for a job, position
etc.' (LDOCE) relies on the situation of putting up notices or pictures. In cer
tain contexts it has an element of metonymy as well, as in (22)
(22) They put up several candidates in Cornwall... (COBUILD)
The names and pictures of candidates in elections are literally put up on walls,
hoardings etc. This metonymic link was probably the basis for the metaphor,
which later got a more general meaning. Put something up 'present (an idea
etc.) for discussion or consideration'(OALDCE) seems to have a similar basis.
Both expressions ultimately rely on a more general mapping of spatial concepts
where up means visible, as contrasted with down.
Another metaphor which is probably recoverable from a more or less
clearly recognizable donor situation in the field of the manipulation of objects is
put across/over (OALDCE) oneself/one's meaning 'cause to be understood'
(LDOCE), which is clearly regarded as a skill, judging from examples (23-28).
(23) That speaker doesn't know how to put himself across. (LDOCE)
(24) I'm not putting my meaning across very well. (LDOCE)
(25) It's very hard to put across the facts. (COBUILD)
(26) He doesn't know how to put himself across at interviews.
(OALDCE)
(27) You need the skill to put your ideas across. (COBUILD)
(28) She's very good at putting her ideas across. (OALDCE)
The donor situation is that of moving something across a barrier or a gap, to
another person. Again, one can trace a metonymie element, as people involved
in L A are prototypically seated across each other, and in quite a few situations
with an object/obstacle in between (table). In the case of speaking before an
audience, there is even a real gap, which consists of the physical distance and
very often also the difference in physical level between speaker and audience. In
putting oneself across, as in (23) and (26) there is a metonymy embedded within
the metaphor, of the type 'speaker for utterance' (see Goossens (this volume).
The collocation with the negative not seems to focus the lack of (self)-control
or skill exhibited by the speakers.
142 PAUL PAUWELS
3.3.3 Metaphorical donor domain

A third set of metaphors are recoverable from domains outside the domain of
movement. In some cases we discussed earlier, like put on the agony, it con
cerned an alternative reading of the metaphor.
With not to put too fine a point on it 'speak plainly and directly, in a way
that some people may find rude' (COBUILD), as in (29), the donor domain can
be described as that of sharpening.
(29) I don't much like modern music. In fact, not to put too fine a point
on it, I hate it. (OALDCE)
The fairly complex act of sharpening has been conceptualized by means of put
which makes it possible to specify the quality of the point. However, the donor
domain does not provide an answer of why one should refrain from putting on
too fine a point; there is no sufficiently salient situation available in the donor
domain for a full interpretation of the metaphor (especially since the secondary
L M is vague and we do not know what is being sharpened). The metaphor does
appear to fit into a larger pattern of conceptualization; this is evident from re
lated expressions like be blunt 'speaking directly and simply, without making
any effort to be polite or to avoid upsetting people' (COBUILD) as in (29b).
(29b) To be blunt, I am afraid that sooner or later you will no longer be
needed here.
In both cases the utterance is conceived as an instrument which can be used to
perform the linguistic action of informing someone of unpleasant news. In such
a situation, delicacy is the social norm. Instead, however, of taking care and
using a precision-instrument in performing the act, the speaker in this case
chooses for a more forceful approach, and does not care for niceties. It could be
argued that cases like (29) should be considered as an (originally jocular) un
derstatement for cases like (29b). The concept of 'refined LA' is also present in
other contexts, as exemplified by (29c).
(29c) I missed some of the finer points of the argument.
The status of put/set someone's mind at rest/ease does not seem to be quite
clear. One definition, i.e. 'tell them something that stops them worrying'
(COBUILD) marks it as LA, while another 'cause or enable to stop worrying'
(OALDCE) does not. The available examples (30) and (31) show the same du
ality.
(30) Let me set your mind at rest. There is nothing wrong with your
heart. (COBUILD)
(31) The letter from her daughter set her mind at rest. (L2)
We are dealing with descriptions of an intended perlocutionary act, and in (31),
for example, it is not even clear whether it is the contents of the letter or the fact
of getting the letter which does the trick. It could be claimed in view of other
evidence (expressions like peace of mind, turn something over in one 's mind)
that we are actually dealing with a metaphorization of a state of mind, on which
another metaphor has been superimposed. The mind is conceived as moving,
worry as excessive movement, and rest as lack of movement. Expressions like
turn/set one's mind to something, and a mind working at full speed demonstrate
that the mind is conceived as a piece of machinery. On this piece of machinery,
various operations can be performed, among other things via LA: in (30-31)
affecting the mind by L A is described as a manipulation of the machinery - some
kind of adjustment, or tuning. The metaphorization of the L A is in this case
shown to be closely linked with, or even dependent upon the metaphorization of
the mind. Given the generally accepted relation between language and thought,
this should not come as a surprise. The core of the LA-metaphor is the control
which is achieved by means of language.
3.3.4. Donor domain: manipulating bodyparts
A major subset of the metaphors is recoverable from the domain of manipulat

ing bodyparts. Apart from put the mouth on someone (kissing), which we dealt
with earlier, we can fit in expressions like put out feelers (animal perception),
put afoot wrong (walking/dancing), put one's foot in it (walking), put the finger
on and put one's foot down (nonverbal communication). One should consider,
however, whether the domain of manipulating objects does not contribute to
these metaphors in some way.
For put a foot wrong and put one's foot in it the contribution of put seems
significant. Here, the aspect which is stressed by the paraphrase is that of con
trol over one's body. As in all acts of manipulation, a certain skill is required and
expected, and it is this which is emphasized. One could refer to the use of
(nonfigurative) expressions like watch where you put your feet, which are either
used as a warning (you could step into something) or a reproach (you're step
ping on my toes). In both cases, skilful movement is in focus.
Put the finger on 'tell the police about a criminal' (LDOCE) and put one's
foot down both have a metonymic basis: for the first this is the way witnesses
identify criminals by looking at photographs, for the second the expression of
anger by stamping one's foot. Both these actions occur in the company of an
144 PAUL PAUWELS
L A of the stated type. Both expressions have shifted away from metonymy to
cover less specific contexts as well, and this is made possible by the meta
phorical pattern they fit into. As in the other metaphors, control over one's body
is stressed in the paraphrase. Put one 's foot down may also fit into the more
general pattern of making a stand, and could hence be looked at from a com
pletely different angle: the situation of horsemen dismounting to stand and fight.
We would then be dealing with an example of the 'argument is war' system.10
Here, however, the recoverability is questionable.
3.4. Interacting donor domains
As was already clear from the foregoing, metaphors very often rely on more
than one cognitive donor domain. The previous set of expressions could still be
situated in one rather specific domain, with in some cases a secondary aspect
contributed by another domain. In the set which now follows, various domains
make equally important contributions.
At the outset (3.2.) it was stated that the put-prototype could be specified
through the type of secondary TR and L M , and the relation between them as
expressed in the preposition/particle. In the previous paragraphs we have seen
how such specifications allowed us to recover donor domains in different ways:
through the secondary TR (put one's foot in it), through both secondary TR and
L M (put the blame on someone), through the secondary L M (put the mouth on
someone), and, finally, via relations which are typical of certain situations (put
someone up for something). In the latter case, the metonymic basis for the
metaphor also played an important role.
In what follows, the specifications do not support the recovery of one do
nor domain: the secondary TR and L M , if specified, are usually specified within
the L A domain, and the prepositions which specify relations between them are
either atypical in combination with put (put over on, put off) or too vague (put
in). There is one exception to this pattern. In put words into someone's mouth,
the secondary L M is specified outside the L A domain - but mouth is of course
still related to LA. Still, the specification does point to a clearly identifiable do
main, so we should consider whether this expression is maybe a borderline case.
3.4.1 Put words into someone 's mouth
Put words into someone's mouth is also treated in Pauwels & SimonVandenbergen
(this volume). It is classified there as a case of 'abstract' meta-
phor, to be understood in terms of the container֊ and control-schemata. The

metaphor is described as having a dual meaning by LDOCE: 'tell someone
what to say' or 'suggest or claim, falsely, that someone has said a particular
thing'. Examples (32) and (33) from COBUILD and OALDCE and also their
definitions only refer to the second meaning, which is probably the main mean
ing.
(32) She accused the journalist of putting words into her mouth.
(OALDCE)
(33) "Now don't go putting words into my mouth please" Jean retorted
getting cross. (COBUILD)
The most obvious 'thing' to be put in the mouth is food, and the donor domain
for the metaphor would seem to be feeding. Still, it is not the prototypical
reading of that situation which is relevant here: it is not helping someone who is
unable to perform the action of feeding himself by putting food into his mouth.
Rather, the focus is on the fact that someone is being manipulated. The meta
phor can only be fully grasped if we reinterpret the situation of feeding in terms
of the prototypical relationships on the recipient domain. It is clear then that the
action described by the metaphor violates (a) the presupposed self-control of a
speaker, and (b) the directionality of speech, i.e. out of the mouth. In terms of
recoverability, however, one could wonder whether the domain of feeding is not
more readily recoverable for the average language user than such abstract image
schemata like container, path, and control. Johnson (1987) argues that such is
not necessarily the case, given the overall salience of image schemata, but meta
phors like this one do not seem to confirm this.
The other meaning of the metaphor 'tell someone what to say' can more
easily be grasped in terms of feeding: here, the central element seems to be the
control of the speaker over someone else. The speaker feeds someone informa
tion which that person will be expected to 'regurgitate'.
The question that remains concerns the contribution of put in these in
stances. It is a fact that put something in someone's mouth is not a standard
description of the scene of feeding. In using this description, the process is ana
lysed and the most tangible/visible aspects are focused. It is this reduction which
is typical of the conceptualization by means of put. At the same time, it is this
analytic description which allows for the (partial) mapping onto the L A domain.
In this way, the metaphor relies on the prototypical sense of put.
146 P A U L PAUWELS
3.4.2 Atypical combinations

Put someone/oneself forward (for something) 'offer for a position' (LDOCE)
and put forward an idea, suggestion, plan, argument etc. 'to offer for consid
eration' (OALDCE) are almost identical in meaning to put someone up for
something and put something up (cp 3.3.2). The first item once again has in all
probability a metonymic basis, in a situation where the act of volunteering is
accompanied by stepping forward (military, school, etc.). In the domain of ma
nipulating objects, however, there is no clearly recoverable donor situation. In a
selling scene, objects are of course moved forward to show them to the cus
tomer, and to offer them for inspection, but this action is not lexicalized by
means of put forward, which must result in a lesser degree of salience. Another
possible context is that of a game of chess, where a forward move gives rise to
consideration on the part of the opponent; it is also part of a plan which unfolds.
But again, putting forward is not normally used with reference to the domain of
chess,11
It seems then, that we have to look elsewhere for a plausible motivation of
the metaphor. The metaphor fits into a more general way of conceiving L A as a
forward movement, also present in metaphors like backtrack, retract, make
progress etc. In that case, we are not dealing with a manipulation of objects any
more, but with the domain of movement in space.
In the same way as put up (cp 3.3.2.) relies on the prototypical upright
orientation of the human body, put forward relies on its prototypical forward
movement. Up and forward are the most prominent positions in space - objects
which are so located will attract attention. It is this general pattern of conceiv
ing which lies at the basis of these four metaphors. For put forward this is
probably more relevant, since a sufficiently salient donor situation seems absent.
In put someone forward, the combination of put with a human L M em
phasizes control: the affected person is manipulated by the speaker.
Next, there is a set of atypical combinations which centers around the
contact-schema. Put someone off, put someone on to someone/something and
put someone up to something, describe LAs which establish or avoid contact.
For none of these metaphors is there a clearly definable donor situation; they
derive their value from their component parts. The put-element is in all three
cases combined with a human L M , which highlights the aspect of control by the
speaker. In put someone off, the preposition describes a lack of contact; some
one is being kept at a distance.
Put someone on to someone/something clearly refers to a surface-schema,

which is open to two interpretations. On can be taken to refer to the sur
face/path along which the addressee will move in order to reach his destination,
which is then indicated by to. In that case L A is seen as a means to steer some
one in the right direction. Conversely, onto as unit can be taken to refer to the
final point in the destination, which would then as a whole lie within the control
of the speaker. Examples (34-35) can be taken to illustrate this difference.
(34) Detectives hunting the gang were put on to them by an anonymous
telephone call. (LDOCE)
(35) I can't advise you in this matter, but I can put you onto a good
lawyer. (OALDCE)
In put someone up to something, up and to need to be considered separately. To
refers to the contact with the secondary L M , in this case a course of action,
while up refers to the direction of movement. Upward movement normally en
tails effort and intent (read force and control) on the part of the mover, and it is
exactly these two elements which are as it were taken over by the speaker. The
L A is again responsible for the contact established.
Put it/one/that old story/an old story across someone also relies on con
tact and surface schemata. Here, however, it is the addressee who is described
as the surface over which the L A moves; he is clearly the underdog of the story.
The L A is conceived in terms of a movement which completely crosses a sur
face. In other words the movement is in no way hampered by the surface. The
concept of control provides the link with the L A meaning 'deceive someone
into believing or doing something' (LDOCE). The agent/speaker is in complete
control of this smooth movement, the patient/addressee has no influence what
soever.
Put something over on someone 'persuade someone to accept a claim,
story etc. that is untrue or worthless' (OALDCE) works along similar lines. Put
over is again a case of 'hidden landmark'. In this case the landmark can be con
strued as a barrier, which necessitates force/effort of the part of the speaker.
This accounts for the aspect of persuasion. The preposition on, moreover,
clearly singles out the addressee as dupe: a burden is placed on his shoulders.
Moreover, on also signals the presence of a greater force exerted on the ad
dressee, which strengthens the element of persuasion.
A final metaphor in this category is put about lies/bad or false news
'spread' (LDOCE). This metaphor exploits another aspect of the put-prototype.
Although putting, as an act of manipulation, implies control, it also implies let-
148 P A U L PAUWELS
ting go of something. In this metaphor, the second element, about, brings this is
focus: when things are about they are per definition scattered at vague loca
tions. This metaphor uses the contradiction between the control which is present
when one gives information to an individual, and the lack of control once that
information reaches more than one person and starts to spread.
3.4.3 Vagueness
The final set of expressions do rely on prototypical uses of the scene of putting,
but it is in no way clear on exactly which use each relies. Put in(to) and put out
occur in so many different situations, that it would be unacceptable to pin them
down on just one donor domain. The notion which is central here is the con
tainer schema. It is used to describe different aspects of the L A situation: the
addressee (put in a request/claim, put in for), the ongoing L A or conversation
(put in, put in a good word for), language (put feelings into words), the speaker
(put out a statement).
Put feelings into words, as in (36) and (37), conceives of words as con
tainers of meaning (a frequent metaphor), and feelings as manipulated objects.
This metaphor fits the conduit metaphor, as described by M.Reddy (1979: 286
ff).
(36) They cannot put their feelings into words. (COBUILD)
(37) ... trying to put her feelings into words. (LDOCE)
The examples seem to emphasize the difficulty and the skill required - feelings
are difficult to 'grasp' and to control. A variant on this metaphor allows for
quantification, which is implicit in (36-37), but comes out more clearly in (38).
(38) He put a lot of feeling into it. (my example)
In put in and put in a good word for, we are once again dealing with a hidden
landmark, which can be construed as ongoing LA, conceived as a container (see
also Rudzka-Ostyn 1988b: 529). The speaker is perceived as putting a contri
bution into this container. One of the effects is to reify that contribution, which
is then open to quantification. The container itself is in both metaphors con
ceived as closed, and rather difficult to get into, which results into connotations
of interrupting as in (39) and (40) or conceives entry as a privilege as in (40)
and (41). The ongoing interaction is conceived as the privileged territory of the
participants, which remains closed to 'outsiders'.
(39) "But what about us," he put in. (LDOCE)
(40) Could I put in a word at this point ? (OALDCE)
(41) I'll put in a good word at the meeting if I get a chance ...
(COBUILD)
Figure 3
Recovery on superordinate domain; donor situation not fullly salient, or insufficient
for a full grasp.
1. put words into someone's mouth 'tell so what to say/suggest or claim falsely that so
has said a particular thing' (LDOCE);
2. put forward 'offer (a suggestion) for consideration' (LDOCE);
3. put in for 'apply for' (LDOCE);
Similarly, put in a request/claim, put in for and put out a statement, where ad
dressee, resp. speaker are conceived as containers, seem to rely on the more
general idea of territories closed to outsiders. The addressee/resp. speaker in
these metaphors is an organisation, which is often metonymically identified with
its office, which is of course literally inside a building. The relevant aspect in the
metaphors is again lack of access, which leads to lack of information about what
goes on inside. It is possible to find a parallel here with the metaphors based on
put up (3.3.2) and put forward (3.4.1). In all cases there is a probable me
tonymic basis in the manipulation of written documents. On the basis of the
parallel, we should construe the hidden L M 1 of put in for as oneself. Self-
determination/control would again be a relevant feature in the metaphorization,
which is logical in context like those provided by (42) and (43).
150 P A U L PAUWELS
(42) Are you going to put in for that job? (OALDCE)

(43) They've put in for membership of the club. (LDOCE)
Figure 3 presents a survey of the types of metaphors which rely on more than
one clearly delineable donor domain. In all these cases, the metaphorizations
rely on various superordinate domains or image schemata.
3.5 Metonymy
Finally, the corpus contains five expressions which are basically metonymic. In
put, put (the) pen to paper, put something down, put someone down for some
thing, and put a sock in it the action described is usually contiguous with L A ,
i.e. "the mapping occurs within a single domain." (Goossens 1990b: 325)
Put in isolation can be used to refer to the process of writing, as exempli
fied in (44-46).
(44) I don't know what to put. {LDOCE)
(44b) Put a question mark. (LDOCE)
(45) Put your name here. (OALDCE)
(46) Put all the details on the card. (COBUILD)
(46b) He couldn't read what Ken had put for his address. (COBUILD)
From these examples, and the definitions (44'-46') which accompany them, it is
clear that the process of writing here is only considered in its 'physical' aspect,
and not as a full LA.
(44') write down, make a written mark (LDOCE)
(45') write or mark something on something (OALDCE)
(46') write or type (a particular word, sentence etc. somewhere)
(COBUILD)
In other words, we are not really dealing with a metonymy for L A , but with a
metonymy based on an act of manipulation which accompanies the L A of writ
ing. The mapping occurs within the L A domain, so we are clearly dealing with
metonymy.
Put (the) pen to paper 'start to write' (LDOCE) is also a clearcut case.
The only divergence from a purely metonymic interpretation here lies in the
possible use of another instrument of writing; in that case the pen would not be
part of the scene which it is used to describe. The requirement of contiguity
would be broken. Still, one would hesitate to call this metaphor, since we are
still moving within the same cognitive domain of writing. In cases where there is
an obvious conflict - e.g. someone sitting down at a PC and saying (47) - the
divergence would in all probability be understood as ironic and metaphorical.
(47) Let's put pen to paper. (my example)
Also put something down relies on the prototypical scene of writing, where the
movement of the pen (or other writing utensil) towards the paper follows a
downward trajectory. Again contiguity is not assured in all cases - writing on a
wall would be an obvious case, but then again it would not normally be de
scribed by means of put down. Here, though, there is a metaphorical potential in
the combination which can be interpreted as an act of controlling - for instance
in order to remember as in (48).
(48) Let me put down your telephone number before I forget.
(LDOCE)
Writing moreover is a less volatile means of communication than speech, which
strengthens this possibility of interpretation.
Put someone down for something is basically the same. Here, it is the
presence of a human object for put brings out the dimension of control. Again,
this corresponds to the generally perceived value of a name on a list: if your
name is put down, this is usually binding. Moreover, the expression contains a
metonymy of the type person-for-name. This second metonymy further en
hances the importance of the control element, since it is not 'just' the name
which is on the list.
Put a sock in it, finally, stands apart in that it is only used performatively,
as in (49).
(49) "Why don't you put a sock in it. I'm trying to work." (OALDCE)
Here, the act of gagging, which entails silence, is called up indirectly, since we
are dealing with an action the addressee is supposed to perform on himself. The
expression again illustrates the way in which metaphor and metonymy interact.
There is an element of contiguity, since gagging entails silence, but the relation
ship is not symmetrical, since silence does not entail gagging. It is not clear
whether in this case one could say that the expression is used to 'refer to' the
act of shutting up. This would lead us to posit a metaphorical interpretation, in
which shutting up (in a certain context) is explained as an act of gagging one
self. The expression draws attention to an apparent unwillingness of the ad
dressee to shut up, the fact that this is considered to be a nuisance, and the pos
sible solution to this problem.
152 P A U L PAUWELS
The metonymic examples are summarized in Figure 4. In this schemati-

zation, it becomes clear that the distinction between metaphor and metonymy is,
in one respect, scalar. One could argue a metaphorical interpretation is called for
from the moment another domain - which may overlap with the target domain -
is sufficiently salient to function as a donor in its own right. As long as the rele
vant aspects of this source domain are considered to be subsumed under the
target domain, a situation of metonymy is invoked.
Figure 4
Metonymy and the extension into metaphorization (dotted lines)
put a pen to paper 'start to write', put sth down 'write sth down'
4. Conclusions
4.1 Recoverability, specificity, metaphoricity
This investigation has shown that different types of metaphors can be distin
guished along different dimensions. First of all, there is the dimension of meta
phoricity - i.e. the degree to which an expression is perceived as metaphorical
by the present-day average language users - which depends on the recoverability
of salient donor situations or donor concepts. Second, there is the dimension of
specificity. Some metaphors rely on clearly recoverable, highly specific, situa
tions, which are salient because of their visual, sensory etc. imminence. Other
metaphors rely on the recoverability of image schematic structures, which are
salient because of their pervasiveness.
The picture is complicated by the fact that these dimensions interact in

such a way that there is no direct correlation between high metaphoricity and
high specificity. Indeed, metaphors like lace into someone or to take another
example kick against the pricks (see also Pauwels & Simon-Vandenbergen (this
volume)) in origin relied on highly specific, clearly recoverable donor situations
- resp. 'whip someone' and 'oxen kicking in protest when they are pricked with
the goad'. Both situations are irrecoverable for the present-day language user;
in the first case because lace has lost its meaning of 'whip' and, in the second
case, because the scene as a whole has lost its salience. A full metaphorical
reading is therefore unlikely for such expressions. The presence of other, similar,
expressions may still account for some metaphoricity. Awareness of lam into
someone and tear into someone may enable a language user to classify lace into
someone as 'aggressive verbal behaviour' with the same donor domain of vio
lent action. Similarly, awareness of kick or kick against the system may help the
language user interpret kick against the pricks as 'a kind of protest'. In extreme
cases, like burden, where the nature of the expression - noun, no colloca-
tional/colligational information - precludes fitting it into a larger pattern of ex
tensions, this can lead to a reinterpretation of the metaphor, as was outlined in
section 1.
For put, such cases of complete irrecoverability are unlikely, because of
the frequency of the verb and the action it refers to in everyday interaction. On
the other hand, it is precisely the frequency and the omniprensence of actions of
putting which makes it difficult to point to a single recoverable scene as the do
nor for a particular metaphorization. On the basis of our investigation, we can
distinguish the following:
• A first set of expressions can be recovered unequivocally because of a full
specification of TR2 and LM2 outside the domain of LA, on a specific sub-
domain of the manipulation of objects. An example in point would be put
one 's cards on the table
• Next there are metaphors where only TR2 is clearly recoverable outside the
LA-domain. Examples are put one's foot in it, and put on the agony. In put
one's foot in it, the donor scene is clearly recoverable outside the domain of
manipulating objects. It is walking, reinterpreted in terms of manipulating the
body, which draws the attention to the lack of skill and self-control on the
part of the agent. In put on the agony there are several possible paths of recoverability,
depending on the framework in which the expression is em
bedded. If it is conceived as a parallel to pile on the agony, the language user
154 P A U L PAUWELS
will recover a context in which adding is the central element; if it is con

ceived as a parallel to turn on the agony, the central image is one of ac
tivating; finally, it can be conceived in terms of putting on a mask/make up,
in which case the element of hiding gets focused. In these cases, recoverabil-
ity is problematic because TR2, although clearly non-LA, is not a manipulatable
object either, so that the verb+particle combination is the main basis
for the recovery of a donor scene. Put on, however, is ambiguous, and can be
recovered from different scenes.
• With some expressions, the overt signalling of metaphoricity has been re
duced even more. Neither TR2 nor L M 2 are specified outside the L A do
main, and the only signalling in the expression is done by put+particle. In
some cases the combination is still recoverable from a limited set of donor
scenes, as with put something up. In other cases, the combination is either
vague (put in) or not used outside the domain of L A (put over on). In those
cases, the combination of the donor domains of the manipulation of objects
and its image schematic values, with the spatial domain and its image sche
matic values, provides the explanation.
• In a final set of metaphorical expressions, put itself is the only element re
coverable outside the L A domain. It is embedded in a non-prototypical
structure: there is no secondary L M , or it collocates with the preposition to
(+addressee) or with manner adverbials which also occur with L A verbs.
There is no signalling which facilitates recovery of a specific scene from the
domain of manipulating objects. In some cases (put something to someone)
the metaphor strengthens the conception of communication as transfer, in
other cases (put the case that, put it bluntly) the producing of L A is disso
ciated from the product of LA, which is reminiscent of Reddy's conduit-
metaphor. In all of the above there is probably a low degree of metaphoricity.
• Finally, there are 5 examples of metonymy. For three of them, metaphorical
readings seem possible. The transfer is again situated at the image schematic
level.
4.2 The contribution of put
The contribution of put varies according to the specificity of the metaphor. With
the most specific cases, put at most only adds a secondary aspect. In ex
pressions like put one 's cards on the table the contribution is minor: it is the
scene as a whole which is recovered, and the focus is not on the mapping of put
onto an element of the L A domain. The same goes for those

put+preposition/particle combinations which are clearly recoverable (put the
blame on, put emphasis on). For less clearly recoverable metaphors, put plays
already a more important role; the contrast with other, similar, verbs is often
relevant, as was clearly demonstrated in the discussion of put on the agony,
which stands in contrast with turn on/pile on the agony. Also in put someone up
for something, put has a clearly distinguishable contribution: the manipulation of
a human L M focuses the issue of control.
For the metaphors which were found to have a donor domain outside the
domain of manipulating objects, put definitely contributes to the metaphoriza
tion. In put one's foot in it and never put a foot wrong, put describes skill in
terms of body-control. In put the finger on and put one's foot down, deliberation
is described in terms of body-control.
The role of put is central in most of the metaphors which rely on more
than one domain. In expressions where the secondary TR or the secondary L M
are persons, put focuses the control-schema. In put about there is the interac
tion between put (control) and about (lack of control), in put feelings into
words it stresses the skill/difficulty involved.
4.3 The contribution of the container schema
Next, I would like to come back to a number of metaphors based on the con
tainer schema. Put in, put in a good word for, put in for, put in a request/claim
and put out a statement do not seem to rely on a specific donor domain. Hence,
one would assume that they follow the pattern of metaphors based on different
donor domains, but this does not seem to be the case. Conversely, the meta
phors seem to rely on the conception of the act of putting in/out, without fur
ther specification.
One reason for this could be the fact that the container schema, in contrast
with other image schemata, is still a relatively rich concept, with a large number
of entailments, such as contents (fluid or object), central and peripheral ele
ments, an inside-outside orientation, a containing structure, and a potential for
movement in and out and for quantification (full-empty). Container is a superordinate
concept, as is clearly demonstrated by basic level realizations like
cup, box, glass etc. Still, one could hypothesize that this is only true in one per
spective, while from another angle (the domain of objects), it is more specific.
156 P A U L PAUWELS
Another factor which plays a role in this set of expressions, is the strong
interaction between put in and the inferred destination (i.e. addressee, or ongo
ing L A ) . To begin with, it is striking that we are dealing with particle verbs,
where the destination of the putting is not specified, in cases where there is a
hidden landmark. The interaction, then, concerns the conflict between an act of
putting in and the perceived nature of the destination, on the image schematic
level. In putting in, one clearly enters a domain, and if that domain is another
person (addressee) or something which is perceived as belonging to someone
(ongoing L A ) , this entails a clear violation of territory, and of control. The
conflict can in some cases be resolved by relinquishing control over that which
is put in. In put in a good word for, for example, the limited nature of the con
tribution seems to indicate a minimal violation of territory. The good word is put
in, and then left to undergo the forces of ongoing interaction. With put in, on
the other hand, the direct object allows for more variation. Examples (39) and
(40) show that the situation is more confrontational; in being a request, (40)
clearly shows who is perceived to be in control of the situation.
(39) "But what about us," he put in. (LDOCE)
(40) Could I put in a word at this point ? (OALDCE)
Put out a statement is different, since there is no conflict situation. The speaker
is in full control, but chooses to move something out of his domain so that oth
ers can act upon it. That which is put out gives the outsiders an idea of what is
inside, or going on inside. In this way they do acquire some kind of control.
Again it is the perceived nature of the object which plays a central role: in this
case the statement is perceived as still linked to its originator.
4.4 Hidden landmarks
In a large number of instances, the expressions are perspectivized with reference

to a hidden landmark (as in put something over on someone) or a vague land
mark (as in put one's foot in IT). These landmarks seem to be recoverable in
three distinct ways. In a first set of expressions, the landmark can be recovered
in the donor domain. These include put across ideas, put something over on
someone, put one's foot in it, not to put too fine a point to it. A second set rely
on a landmark that can be recovered from the recipient (LA) domain. Here we
could classify put in, put in a good word for, put in for something (addressee or
ongoing L A ) and put something/someone/oneself down (for something) (paper
as an instrument in writing), put a sock in it (mouth as an instrument of speak-
ing). The largest set of expressions refer to the agent as landmark, or to his en
vironment with respect to his position in that environment. Here we include put
about lies, put someone off, put something/someone/oneselfforward, put out a
statement. In two expressions the movement of a bodypart is profiled with re
spect to the agent (put one's foot down, put out feelers), and one expression
uses a bodypart (face) as landmark (put on the agony). In this respect, the AmE
expression put one's foot in one 's mouth offers an interesting insight in the in
terpretation of metaphors. It makes explicit the construal of the landmarks in
terms of the recipient domain, whereas the BrE expression put one's foot in it
relies on the donor domain for the recovery of its vague landmark.
4.5 Metaphorical interaction
An example like the above seems to point to an interaction between metaphori

cal expressions which are perceived as similar. Support for the existence of such
an interaction can also be found in the ability to interpret expressions like lace
into someone on the basis of similar metaphors which have a negative effect on
the affected entity or person, like tear into someone. The reliance on such inter
action seems to be especially relevant for metaphors with a hidden landmark, as
is further illustrated by put on the agony, which can be interpreted in terms of
'masking' in relation to, for example, put on a brave face, or in terms of 'setting
in motion' in relation to turn on the agony, turn on the waterworks, or, finally,
in terms of 'burdening' in relation to pile on the agony.
4.6 A note on recoverability
In this paper I have introduced the notion of recoverability to describe the ex
tent to which, and the way in which, the average language user is able to link a
metaphorical expression to a (usually non-figurative) donor scene. I have pre
ferred recoverability to salience, because, to my mind, it is not the overall sali
ence of the donor scene or concept which is at stake, but the salience of that
scene together with the availability of the scene in using the metaphor. In other
words, a scene or situation may be salient, but still not be recoverable to the
language user in the context of a certain metaphor. What makes the donor scene
recoverable seems to be
(i) the overt signalling in the metaphorical expression,
(ii) the awareness of other, similar, metaphorical patterns or mappings,
(iii) the overall salience of the donor scene or concepts, and
158 P A U L PAUWELS
(iv) possibly also the wider context in which the metaphor is embedded.
The contributions of (i-iii) have been discussed and analyzed in this paper. The
role of (iv) remains beyond the scope of this investigation, since we worked
from a lexicographic database, and we have no access ot the context beyond the
example sentences.
Finally, the judgements on recoverability in this paper are subject to two
limitations. To begin with, they rely on a notion of average language user as
defined in relation to a number of explanatory dictionaries; furthermore, the
interpretation of the data is my own, and although it has been verified with a
number of competent language users, there has been no empirical testing of
either the specific interpretations or indeed the model of interpretation.
Notes
1. The role of image schema is further elaborated on by Rudzka-Ostyn (this volume).

2. Note, however, that the presence of the prepositional phrase ' w i t h ' accompanied by ref
erence to a person does provide a clue as to the possible interpretation. The LA is perspectivized
as an interaction between a speaker and a hearer.
3. It might very well be that such situations lead to a reinterpretation of the metaphor. This
w i l l only surface in a full-scale diachronic investigation, which is beyond the scope of
this paper.
4. 'Gaff' here is short for 'gaff-topsail', " A light triangular or quadrilateral sail, the head
being extended on a small gaff which hoists on the topmast, and the foot on the lower
gaff' (voile à corne) (OED) ' B l o w ' is used in the meaning of f i l l i n g with air, making
big. I n the combination, then, the relevant elements are the highly visible position of the
gaff-topsail, which is enhanced by the act of blowing it.
5. The variety of donor domains is more fully described in Rudzka-Ostyn (1988a) and Van-
parys (this volume).
6. The frequency count was based on a corpus of 1,000 examples, which was compiled by
means of a KWIC-index on the L.O.B. and Leuven Drama corpora.
7. On the problems with the notion 'prototype' see Lakoff (1987: 39ff) and Geeraerts
(1989).
8. This type of link is investigated in more detail in Goossens (1990), where it is described
as metaphor from metonymy.
9. Also this metaphor fits into a more general pattern of metaphorizations, of which out of
line is another example.
10. See L a k o f f and Johnson (1980: 79ff).
11. I n the corpus which is also referred to in note 6, containing non-figurative meanings of
put, there is not a single example of put forward; the same is true of the set of explana
tory dictionaries used in compiling the corpus for this investigation.
Metaphtonymy
The Interaction of Metaphor and Metonymy in Figurative
Expressions for Linguistic Action
L o u i s Goossens
University of Antwerp
1. Introduction
The purpose of this paper is to deepen our insight into the ways in which
metonymy interacts with metaphor in figurative language. Although in principle
metaphor and metonymy are distinct cognitive processes, it appears to be the
case that the two are not mutually exclusive. They may be found in combination
in actual natural language expressions. In that sense there might be room for the
neologism in the title of this paper, for which I suggest the phonological
realization [metæf'tɔnimi] to help the reader along if (s)he wishes to know
whether the word is also pronounceable. It will be shown, however, that the
interaction can take several forms, for which a single term may be misleading
rather than helpful. In other words, I would like to assign metaphtonymy the
status of a mere cover term which should help to increase our awareness of the
fact that metaphor and metonymy can be intertwined.
To explore the interaction I have used a data base made up of figurative
expressions where the target domain is that of linguistic action. This database
is restricted and can therefore not be expected to provide an exhaustive account
of the possible interaction patterns. On the other hand linguistic action is
sufficiently complex and the data base exhibits enough diversity to allow us to
come up with the main patterns, which in the final section of this paper will be
put into a somewhat broader perspective.
In what follows I first remind the reader of a couple of basic insights into
metaphor and metonymy (section 2) as well as into the target domain (section
3). Next, I provide a brief account of the data base and of the donor domains
160 Louis GOOSSENS
figuring in it (section 4). Sections 5, 6 and especially 7 constitute the bulk of the
paper: in them I explore the different ways in which metaphor and metonymy go
together for the three donor domains in the data base in succession. Section 8
surveys these findings and tries to come up with a few generalizations about the
interplay between metaphor and metonymy.
2. Metaphor and metonymy
As a representative of the traditional approach to metaphor and metonymy, the

definitions in Halliday (1985: 319-320) can be quoted. Note that in this view a
distinction is made between synecdoche and metonymy, though, obviously,
synecdoche is a subtype of metonymy.
• Metaphor. "A word is used for something resembling that which it usually
refers to; for example, flood ... poured in, [...] in A flood of protests
poured in following the announcement (a large q u a n t i t y ... came
i n ) . [...] If the fact of resemblance is explicitly signalled by a word such as
like, as in protests came in like a flood, this is considered to be not
metaphor, but simile".
• Metonymy. " A word is used for some thing related to that which it usually
refers to; for example eye [...] in keep your eye on the ball (gaze)".
• Synecdoche. "A word is used for some larger whole of which that which it
refers to is a part; for example strings [...] i n At this point the strings
take over (stringed instruments)".
For an instance where synecdoche is viewed as part of metonymy, we can refer
to Ullmann (1962: 212), who differentiates metonymy from metaphor as
involving contiguity as opposed to similarity, where contiguity "includes any
associative relations other than those based on similarity".
In cognitive treatments metaphor and metonymy are viewed as concep
tual processes in which the notion of domains plays a crucial role. Lakoff
(1987), for example, offers the following definitions:
• "[...] metaphoric mapping involves a source domain and a target domain.
[...] The mapping is typically partial. It maps the structure in the source
domain onto a corresponding structure in the target domain"
• "[...] a metonymic mapping occurs within a single conceptual domain which
is structured by an I C M [= an Idealized Cognitive Model]"
METAPHTONYMY 161
In other words, the crucial difference between metonymy (as well as synecdo
che) and metaphor is that in a metaphoric mapping two discrete domains are
involved, whereas in a metonymy the mapping occurs within a single domain.
Obviously the hierarchy among cognitive domains, as well as their de
limitation, which are important areas for exploration within cognitive linguistics
anyway (see e.g. Langacker (1987: chapter 4)), are important issues in this
context. For the purposes of what follows we simply posit the existence of
complex domains built up by the combination of other domains which them
selves may either be complex or basic in the sense of Langacker (1987). It
should also be expected in this view that the boundary lines between domains
are often fuzzy, which is one of the reasons why metaphor and metonymy may
interpenetrate.
3. The target domain: linguistic (inter)action
For discussions of the complexity involved in the domain of linguistic

(inter)action, we refer the reader to Dirven et al. (1982), Verschueren (1984)
and (1985) and Rudzka-Ostyn (1988). Without going into details, I would like
to emphasize two aspects of this complexity here.
• In linguistic (inter) action a speaker produces utterances by means of natural
language to make known his ideas, beliefs, wishes to one or more hearers
who process those utterances and, in turn, may become speakers to do the
same. Talking about linguistic interaction involves secondary speakers who
produce utterances in which they report to secondary hearers what was
said by primary speakers to primary hearers. Both primary and secondary
speakers are equipped with their own beliefs and emotions and make use of
specific linguistic forms as well as specific communication channels. All this
produces a complex network of relationships and stuctures, which is never
theless conceived as hanging together, in other words, as one complex
conceptual domain.
• As a complex domain, the domain of linguistic (inter)action intersects with
(or: partially incorporates) several basic and non-basic domains, such as
sound, language, human actions, emotions, human cognition, percep
tion, etc. For a (tentative) schema we refer to Rudzka-Ostyn (1988: 510).
162 Louis GOOSSENS
4. Database and donor domains
4.1 Donor domains
We have studied the interaction of metaphor and metonymy in a database of

figurative expressions for linguistic (inter)action from three different donor
domains which are fairly discrete (though there is a slight degree of overlap
between them). They are body parts, sound and violent action. Let me give a
brief characterization of each of these in turn.
4.1 1 Body parts
There appear to be a considerable number of figurative expressions for linguistic

(inter)action which contain lexical items denoting parts of the body, more
specifically of the human body. To the extent that certain human body parts are
instrumental in linguistic (inter)action, we can expect there to be an intersection
with the target domain, but evidently this does not hold for all of them. It should
also be emphasized from the start that as a rule the body parts are not donor for
linguistic action on their own. In the majority of cases the body part fits into a
more complex domain or scene which has to be processed with reference to
linguistic (inter)action in its own right.
4.1.2 Sound
An obvious restriction here is that sound is to be understood as sound that can

be perceived by the human ear. Another one is that literal references to linguistic
sounds (as in shout or whisper) are excluded. Otherwise it is also the case here
that a given figurative expression usually relates to a donor scene for which a
more specific characterization than sound is required.
4.1.3 Violent action
More specifically, our third donor domain is that of physical violent action,
which itself is a subdomain of the vast domain of human action. Again this is an
important donor domain for linguistic (inter)action, not unexpectedly, given the
connection with the Argument-is-war-metaphor identified by Lakoff and
Johnson (1980).
METAPHTONYMY 163
4.2 The database
The database consists of 309 items, distributed as follows:

• Body parts, 109 items; 86 verbials (i.e. verbs or expressions with verbs
which, in turn, may contain nominals), 12 adjectivals (which also include
some participial items) and 11 nominals
• Sound, 100 items, all verbials
• Violent action, 100 items, also all verbials.
The main source for this data base is Longman's Dictionary of Contemporary
English; the material is predominantly, but not exclusively, British English. For
the way the data were collected we refer to Vanparys (this volume). The sound
and violent action expressions also include a few items from other contempo
rary lexicographic sources, in actual fact I have used the databases established
by Van Deun (1988) (for the sound corpus) and by Govaers (1988) (for the
items where violent action is donor. Note also that the body part data differ
somewhat from the corpus used by Pauwels and Simon-Vandenbergen (this
volume) in spite of the common core.
Given the fact that these data originate from a contemporary diction
ary/dictionaries which itself/themselves is/are based on an extensive database,
we can safely assume them to be representative of everyday metaphor and
metonymy, and in that sense of the figurative language that the speakers of
(mainly British) English 'live by'.
5. Analysis of the sound data
Van Deun (1988: 68-79) distinguishes the following subcategories (according to

the type of sound involved):
• human sound (27 items; applaud, giggle, snort, wheeze and the like)
• animal sound (43 items; bark, cackle, purr, squeal, etc.)
• non-human, non-animal, natural sound (8 items; blast, thunder, etc.)
• artificial sound produced by musical instruments (9 items; blow one's
own trumpet, harp on, pipe down, etc.)
• artificial sound not made by musical instruments (the remaining 13
items).
164 Louis GOOSSENS
As a rule the donor domain is clearly distinct from the target domain, there is a
mapping from one domain into another, hence we get pure metaphors. Typical
examples are:
• bark 'say something in a sharp loud voice': the loud, penetrating sound of
barking dogs is mapped onto linguistic action where the sound is perceived
as loud, harsh or sharp;
• blow one's own trumpet 'say good things about oneself, perhaps immodestly,
so that others will know them' : the public and festive character of trumpet
blowing, in combination with the added reflexive dimension, is mapped onto
self-praise. To the extent that we may conceive of a scene in which the
trumpet blowing is followed by a public statement in which the announcer
'says good things about himself', we might accept a metonymic basis for the
expression. Since such an interpretation is far removed from the prototypical
scene of trumpet blowing, however, such a metonymic basis is very weak, to
say the least.
The items in the first group, on the other hand, but only those, usually have a
metonymie ingredient. Let us have a closer look at giggle 'express by or utter
with a giggle' as a paradigm case. A typical example would be (1).
(1) 'Oh dear', she giggled, ' I ' d quite forgotten'
One interpretation is that she said this while giggling: in that case there is a
synecdochic relationship; we express part for the whole, we have a pure
metonymy. Another way to interpret it is that she said this as if giggling; hence
there is a crossing of domain boundaries, we have a metaphor. Very often,
however, it is not clear whether the domain of giggling and that of
(light-hearted) linguistic action are conceived separately. We are still aware of
the metonymie basis in the metaphorical interpretation. This is what I would like
to call metaphor from metonymy. Figure 1 tries to visualize this. On the left
hand side of the figure two potentially discrete domains, A and B, intersect; they
are fused in a single scene (the surrounding circle). On the right hand side, A
and  are separated, but, as the broken arrow indicates, there remains a
conceptual link with the scene in which the two are together.
The double possibility (metaphor from metonymy or metonymy only)
holds for most items in group (i) (19 out of 27). Obviously for some of them the
metaphorical reading is the natural one, for example, for applaud 'express
strong agreement with (a person, idea, etc.)', as in (2).
(2) These changes will be applauded
METAPHTONYMY 165
Another example of this sort is snort 'express anger in a snorting way', where it
is difficult to conceive of a scene where the non-linguistic and the linguistic
activity go together and where we get an as-if-reading, hence a (pure) meta
phor.
Figure l Metaphor from metonymy
An (exceptional) example of a metaphor from metonymy from the other

groups is snap at 'say or answer in an angry or rude way'. The literal meaning
denotes the quick closing of jaws, especially of an animal, for example of a dog.
Since, however, human beings also have the capacity to close their jaws quickly
and forcefully, and since this may occasionally accompany angry speech, the
expression can also be said to have a (weak) metonymic basis. Note, for that
matter, that the donor domain for snap at can also be taken to be violent
(animal) action, as well as animal sound; in actual fact snap at was also
included in the violent action corpus.
In conclusion, we find that the donor domain sound gives rise to several
metonymies, or metaphors from metonymy, precisely in those cases where
sound hangs together with a human activity that can naturally co-occur with
linguistic action. Typically, these items have a hybrid character, in that they are
metonyms in some contexts and metaphors from metonymy in others.
6. The violent action data
Physical violent action is sufficiently distinct from linguistic (inter)action not to

overlap or coexist with it in the very great majority of cases. As a rule therefore
the figurative expressions in our data base involve a mapping from one domain
onto another, in other words, they are metaphors.
166 Louis GOOSSENS
In the whole subcorpus of 100 verbials there were only 6 or 7 items for
which a metonymic ingredient can be suggested. All of these are of a type where
the violent action could be accompanied by verbal action, for example, throw
mud at 'speak badly of, especially so as to spoil someone's good name unnec
essarily'. It is conceivable that people may combine the violent action with
shouting names, which is linguistic action: this would be an instance of
metonymy. The metaphorical interpretation can easily be established, however,
without this metonymic backing. What this adds to our insight, is that metaphor
from metonymy occurs with varying degrees of cognitive saliency; instances like
the one discussed here provide us with the limiting case. Another instance of
this sort is give a rap on/over the knuckles 'attack with sharp words', an
item which also occurs in the body part corpus.
7. The body part corpus
7.1 Some further characterization of the data
Before embarking on a discussion of the interplay of metaphor and metonymy,

we first provide some further characterization of the 109 items making up our
data base. We do this from two points of view.
To begin with, the corpus can be subclassified according to different
groups of body parts:
• 49 items contain a body part which could be instrumental in the speech act:
15 with mouth, 15 with tongue, 7 with lip(s), 3 with breath (not really a
body part, but so closely associated with the body that we decided to include
it), 2 with jaw, 2 with throat, one with chin and one with voice (an item
for which the same remark holds as for breath);
• 26 items contain a body part which is connected with the head but not
potentially functional in the act of speaking (this includes the item head
itself): 6 with head (I have listed here tête-à-tête, which is actually from
French), another 6 with ear(s) (which, of course, may be instrumental in the
perception of speech), 3 with neck, 2 with nose, 2 with eye(s), 2 with
brain, one with brow, eyebrow, cheek, hairs and profile;
• The trunk of the body is involved in items with heart (6), breast (2), chest
(1), bosom (1), belly (2), back (3) (15 in all).
METAPHTONYMY 167
• 10 items are connected with the hand, including hand itself (6 times),
palm (1), finger (1), knuckles (2).
• The leg or part of it are represented 7 times: legs (1), foot/feet (4), knee(s)
(2).
• Finally, there are two items with blood, again a 'body part' only in the loose
sense of the word.
Secondly, it must be pointed out that the role played by the body part plays
varies according to whether we have a verbial, an adjectival, or a nominal. In
the case of verbials and adjectivals the body part is necessarily integrated into
some broader scene. Nominals, on the other hand, may be directly related to an
aspect of linguistic action, though also here there may be a combination with
another item, so that the body part is instrumental in a broader scene as well. As
we shall find, this considerably increases the complexity with which metaphor
and metonymy may interact.
7.2 Metaphor and metonymy in the body part corpus
7.2.1 General survey

Table 1. Distribution in the body part data
Database Verbials Adjectivals Nominals

109 88 11 10
Pure metaphors 42 36 5 1
Pure metonyms 8 5 0 3
Mixed cases 59 47 6 6
- Metonymy in metaphor (27) (19) (4) (4)
֊ Metaphor frommetonymy (27) (24) (2) (1)
- Special cases (5) (4) (0) (1)
Obviously, the data show up a considerable portion of what we may refer to as

pure metaphors and metonyms. This is only to be expected, since in collecting
the items the criterion was that they should be 'figurative'. As was indicated for
the other two subcorpora, this results in a set of data which are predominantly
metaphorical. The striking fact about the body part data therefore is rather that
there are so many instances with a metonymic ingredient, i.e. either pure
metonyms or mixed cases (where mixed implies that there is some interplay of
metaphor and metonymy). Table 1 surveys the proportion of pure metaphors,
168 Louis GOOSSENS
pure metonyms, and mixed cases in the corpus. I have added the distribution
over the verbials, adjectivals and nominals, because it is not insignificant.
In the context of this paper it is, of course, the mixed cases that are of interest;
they will be explored under the following subheadings. Before proceeding with
that discussion, let me draw the reader's attention to the high proportion of pure
metonyms for the nominals as opposed to their complete absence for adjectivals.
This hangs together with the fact that it is easier to select entities which are part
of, or otherwise associated with, other entities as representatives for those other
entities than it is to represent properties by partial or associated properties
(where I take for granted that the categorial meaning for nominals is the
denotation of entities and for adjectivals the denotation of properties). As will
appear from the instantiations for the mixed cases, it is usually (but not always)
the integration of a nominal element into the verbial or the adjectival that is
responsible for the metonymic ingredient in an otherwise metaphorical context.
7.2.2 Metaphor from metonymy
This pattern, which frequently occurs when the donor domain is (non-linguistic)
human sound (see section 5), is also well represented in the body part data. In
my analysis there are 27 items (24 verbials, 2 adjectivals, 1 nominal), i.e.
practically one fourth of the data, that belong here. Again the boundary lines
with pure metaphors and pure metonyms are sometimes a little hazy, but there is
no doubt that the great majority of those 27 can safely be assigned to this type.
For all of them it is possible to use them metonymically, that is with reference to
a scene where both the non-linguistic and the linguistic action reading are
relevant, and it is that metonymic reading which is the basis for the metaphorical
use. As a rule, however, there is an idea of transfer from a distinct scene; in
other words, we get metaphors for which there is a link with their metonymie
origin. In the following exemplification it will also appear that the relevant
scenes have to be characterized in their own right; the body part is just an
ingredient in a broader scene. This accounts for the fact that there is no
significant correlelation with any of the subgroups distinguished in 7.1. Let me
provide a few instances now with a word of explanation.
• Say something /speak / talk with one's tongue in one's cheek 'say
something and mean the opposite, esp. in an insincere or ironic way'. The
metonymie basis is a scene in which someone literally (and visibly) pushes his
tongue into his cheek while saying something that he does not really mean; in
this metonymie reading - unlike in the (admittedly improbable) literal inter-
METAPHTONYMY 169
pretation - the tongue in the cheek is taken to be intentionally linked up with

the ironic impact of what the speaker says. As a rule, however, we use the
expression to express that the primary speaker says something as if he had
his tongue in his cheek; there is a mapping from a donor scene onto the
target scene. When we use it of insincere, rather than ironic speech, the
expression is even necessarily metaphorical, at least to the extent that an
insincere speaker does not want to give away that what is said is not really
meant.
• Beat one's breast 'make a noisy open show of sorrow that may be partly
pretence'. Here the metonymic basis is the religious practice of beating one's
breast while one confesses one's sins publicly.
• Close-lipped 'silent or saying little'. Let me emphasize here that the
metonymic reading and the interpretation as metaphor from metonymy can
be expected to be equally frequent. Close-lipped can be paraphrased as
literally meaning 'having the lips close together' or as 'having the lips
closed'; when close-lipped is used to indicate that a person is literally silent,
we therefore need the metonymie reading. If, on the other hand, we describe
as close-lipped someone who is actually talking a lot, but does not give
away what one would really want to hear from him, we have a metaphor
(and given the saliency of the metonymie basis, a metaphor from metonymy).
A general point which should have emerged from the discussion is that the
chances that metaphors from metonymy are used purely metonymically are
variable. They are probably better for items like have a word in someone's
ear 'speak secretly' or raise one's eyebrows at 'express surprise, doubt,
displeasure or disapproval (at)' than for beat one's breast or put one's foot
down 'speak or act firmly on a particular matter'.
7.2.3 Metonymy within metaphor
In this pattern, which appears in the body part data only, we get metaphors
(involving therefore a mapping from a donor domain A onto the discrete target
domain B, which in our data evidently is linguistic (inter)action), but with a
built-in metonymy. This metonymy involves the body-part which is a shared
element in both domains (A and B). This situation can be pictured as in Figure
2, where the shared element, the body-part, is represented as x. Because of its
different function in the two domains, it is differentiated as  and x' in the donor
and recipient domain.
170 Louis GOOSSENS
Figure 2. Metonymy within metaphor
Note that this representation does not yet give us the whole story about the
shared item x/x'. A couple of examples will show that as a rule it functions
metonymically in the target domain only, whereas it is interpreted literally or
(more often) (re)interpreted metaphorically in the donor domain.
• Bite one's tongue off <informal> 'be sorry for what one has just said',
typically in contexts like (3).
(3) I should/could bite my tongue off
Here tongue can be processed literally in the donor scene. Because of the
counterfactual contextualization this donor scene can only be one that does
not directly tie up with everyday experience. Perhaps the best way to charac
terize it is in terms of self-punishment, where the punishment hyperbolically
involves a rather unlikely kind of self-mutilation. Mapping this onto linguistic
action we get something like 'depriving oneself of one's ability to speak',
where the metonymy is from tongue to the speech faculty as a whole. The
hyperbolic nature then generates an implicature in the Gricean sense along
the lines of 'I'm terribly sorry for having said something so foolish, rude, or
the like'.
• Shoot one's mouth off 'talk foolishly about what one does not know about
or should not talk about'.
The donor domain is the foolish or uncontrolled use of firearms: the foolish
(and therefore potentially, though not intentionally, dangerous) use of a gun
is mapped onto unthoughtful linguistic action. By integrating mouth into a
scene relating to the use of firearms it is reinterpreted as having properties of
a gun in the donor domain; this is the metaphorization in the donor domain.
In the recipient domain, however, there is a first level of interpretation which
amounts to something like 'using one's mouth foolishly', in which mouth is
a metonymy for speech faculty. Hence an utterance like (4) comes to mean
'Don't say anything rash'.
METAPHTONYMY 171
(4) Don't shoot your mouth off.

• Catch someone's ear 'catch someone's sympathetic attention or notice' as
in (5).
(5) She caught the minister's ear and persuaded him to accept her
plan.
This invokes a scene of an entity (animal, bird, insect, or even human being)
running or flying around which one tries to get hold of. From the point of
view of the donor domain ear has to be reinterpreted (metaphorically) as an
entity on the move; in the target domain it is used metonymically for the
hearer's attention. Note, by the way, that the paraphrase, which was taken
over from Longman's Dictionary, makes use itself of the 'frozen metaphor'
catch someone's attention.
Again, the pattern is comparatively frequent in this subcorpus: nearly one
fourth of the items are of this type. With two or three exceptions they are all
verbials where the body part is involved in a broader scene. Note also that
here there is a very strong correlation with the body parts that can be func
tional in linguistic action: all instances come from groups (1) and (2).
7.2.4 Demetonymization inside a metaphor
There is at least one instance of what can be described as a demetonymization

inside a metaphor: pay lip service to 'support in words, but not in fact; give
loyalty, interest etc. in speech, while thinking the opposite'.
At first sight this may seem to be another example of metonymy within
metaphor. Paying suggests a scene of financial transaction; that scene is the
'embedding metaphor'. Lip service is 'service with the lip(s)', where lip(s)
stands for speaking, which is a metonymy (one with a biblical origin, see
Goossens (1993a), but no doubt 'secularized' for most speakers of English
today). However, to make the figurative expression work, we have to expand
our paraphrase for lip service into 'service as if with the lips only'; the part is
dissociated from the whole for which it was made to stand in the earlier
processing stage, it is 'demetonymized'.
7.2.5 Metaphor within metonymy
Also this type is represented by one instance only: be/get up on one's hind legs
'stand up in order to say or argue something, esp. in public'.
172 Louis GOOSSENS
The peculiarity about this item is perhaps best revealed if we leave out
hind: being I getting up on one's legs with reference to 'standing up in
order to say something in public' is metonymic, there is an overall scene of
somebody standing up and saying something publicly. The addition of hind
forces us to reinterpret in terms of an animal standing up. This suggests a
greater effort, an event which attracts more attention. At the same time there is
a bathetic effect, because a human being is interpreted as being involved in the
pseudo-achievement of standing on two legs. One may, of course, also argue
that the addition of hind makes the expression as a whole metaphorical; it is
only to the extent that we process it with an awareness of the metonymy, that it
is more adequate to view this as a metaphor embedded into a metonymy.
8. Some further perspective
The foregoing analyses have given us an initial and probably representative

picture of the ways in which metaphor and metonymy can interact. Let us first
review the patterns that were observed. I list them with an indication of the
frequency with which they occurred.
(i) Metaphor from metonymy. This was a frequent type in the figurative
expressions where the donor domain is human (non-linguistic) sound, and
well represented in the body part data. The main point here is that underly
ing the metaphor there is an awareness that the donor domain and the tar
get domain can be joined together naturally in one complex scene, in
which case they produce a metonymy, of course. The actual contexts into
which these items fit will be decisive for the interpretation as either a me
tonymy or a metaphor from metonymy, with, of course, a fuzzy area where
it is difficult to decide which of the two is the more relevant interpretation.
(ii) Metonymy within metaphor. Although less frequent than (i) in our data
base, this pattern was also quite current, be it only in the body part corpus.
The typical case for (ii) is that a metonymically used entity is embedded in
a (complex) metaphorical expression. The metonymy functions within the
target domain. As we found out in the instances we analysed, this often,
but not necessarily, goes together with a metaphorical reinterpretation of
the relevant entity in the donor domain.
(iii) Metaphor within metonymy. This type is extremely rare in our data and I
assume that it is rare in general. Probably this hangs together with the fact
METAPHTONYMY 173
that if we embed a metaphor into a metonymy, it tends to 'metaphorize'

the whole expression. It is only in instances where the metonymic reading
remains relevant (as in be/get up on one's hind legs, which was dis
cussed above) that this pattern occurs.
(iv) Demetonymization in a metaphorical context. This is also an excep
tional type. In the example we found (pay lip service to) it turned out
that the metonymic reading (lip for saying something) was relevant, but
that at the same time the overall metaphorical context favours an interpre
tation in which the metonymie extension is abandoned ('service by means
of the lips only').
These findings raise a couple of questions which I will briefly go into next.
First, we may wonder whether the mirror image for type (i) is possible,
i.e. metonymy from metaphor. I would like to suggest that it is, though it is
rather difficult to conceive and therefore very rare. Let me try to construct an
example with an item which occurred in my discussion of the sound data in
section 5. I pointed out there that blow one's own trumpet is an instance of
metaphor because it is difficult to conceive of the scene of trumpet blowing and
that of self-praise as being combined. Suppose, however, that the two do occur
together and that we use an utterance like (6) to describe this (admittedly un
likely) scene.
(6) Remarkable, the chap is blowing his own trumpet!
In such a case we would be forced to become aware of the metonymie interpre
tation, but to the extent that we are also conscious of the fact that we know that
the expression is basically metaphorical, we will process it as a metonymy from
metaphor.
This leads to a second question. Why is it that metaphor from metonymy
is quite current, whereas it is difficult to come up with good instances of
metonymy from metaphor?
Let me repeat in this context that for metaphor we map an element from a
donor domain onto an element of a discrete recipient domain. For a metonymy
the mapping is from an element A to an element  within the same (structured)
conceptual domain. Metaphor from metonymy implies that a given figurative
expression functions as a mapping between elements in two discrete domains,
but.that the perception of 'similarity' is established on the basis of our aware
ness that A and  are often 'contiguous' within the same domain. This frequent
174 Louis GOOSSENS
contiguity provides us with a 'natural', experiential, grounding for our mapping

between two discrete domains.
Going from metaphor to metonymy is conceptually more difficult, because
here it is implied that the two domains are in principle discrete. The case where
the mapped elements in a basically metaphorical expression can be interpreted
as belonging to the same (complex) domain is rare as it were by definition,
because, if it were frequent, we would automatically get a metaphor from
metonymy.
This does not yet explain why metomymy within metaphor occurs fre
quently, but not metaphor within metonymy. In both cases we get a complex
mapping, where for metonymy within metaphor a metonymic mapping is
inserted into a metaphoric one and for a metaphor within metonymy a metaphor
becomes an ingredient in a metonymic expression. As will have become clear
from the discussion of the examples in section 7.2.3, metonymy within metaphor
is possible only if in the donor domain the element which becomes metonymie in
the recipient scene can either be processed literally or be reinterpreted meta
phorically. In other words, the metonymy is integrated into the metaphor, but
the metaphor maintains itself, it is not 'destroyed' by this metonymie ingredient.
In the case of a metaphor within metonymy, on the other hand, at least in the
single example we have found in our data {be I get up on one's hind legs) the
addition of an element from a discrete domain {hind in our instance) tends to
metaphorize the whole expression; it is only by virtue of the strong cognitive
salience of the metonymie alternative {be/get up on one 's legs) that the complex
interpretation as metaphor within metonymy becomes relevant. A metaphor
inserted into a metonym would seem to metaphorize the whole, whereas a
metonym integrated into a metaphor does not appear to have the power to
metonymize the metaphor.
Rounding off, we found that there are two current patterns for the inter
play between metaphor and metonymy. One in which the experiential basis for
the metaphor is a metonym, yielding what we called metaphor from metonymy.
The other in which a metonym functioning in the target domain is embedded
within a metaphor, i.e. metonymy within metaphor. The other types that we
found are a lot more difficult to conceptualize and therefore exceptional.
Obviously, the generalizability of these findings will have to be verified
with figurative data for other domains, more specifically for other recipient
domains than linguistic interaction.
From Three Respectable Horses' Mouths
Metonymy and Conventionalization in a Diachronically
Differentiated Data Base
L o u i s Goossens
University of Antwerp
1. Introduction
1.1 Aims and scope
In this paper I will try to add to our insight into metonymy, its relationship to
'literal' usage and to metaphor, and its conventionalization. Making use of data
from Ælfric, Chaucer and Shakespeare (the three 'respectable horses' of my
title), as well as of the categorization of metonymic expressions with mouth
proposed in Goossens (1993b), I will clarify some of the ways in which
metonymy works as a tool for meaning extension in a diachronically diversified
data base. As regards my understanding of metonymy within the framework of
Cognitive Grammar I have profited from Croft (1993).
It will be shown, amongst other things, that metonymy shades off into
'literal' meaning extensions, but also that it interacts with metaphor along the
lines indicated in my Metaphtonymy paper (this volume), and that there is
something of a metonymy-metaphor continuum. The diachronic spread of our
data (which range from Old English over Middle English to Early Modern
English, and occasionally include Present-day items) will allow us to have an
eye for at least some aspects of conventionalization with respect to metonymy.
As such this paper brings together a number of insights about metonymy
formulated in three earlier papers of mine (Goossens 1990b, 1993a and 1993b),
and supplements them in various ways.
176 Louis GOOSSENS
1.2 Structure
This contribution is structured as follows. Section 2 provides an account of

metonymy for which I am especially indebted to Croft (1993). In section 3 I
briefly list the relevant interaction types between metonymy and metaphor.
Section 4 characterizes my database, which will turn out to be somewhat
restricted, but, as pointed out already, has the advantage of providing us with
some interesting diachronic diversification. A preliminary exploration of the
domains that are relevant to our understanding of the symbolic unit mouth (or
mouth(e) orÞfor that matter), around which my data are centred, is offered
in section 5. An exploration of the metonymies at hand from several angles
comes in the four subsections of section 6, followed in section 7 by a series of
remarks on conventionalization. The whole is put into perspective in the final
section.
2. Metonymy
Whereas a discussion of metonymy is virtually absent in Langacker (1987), it is

accorded an important position in Lakoff (1987). In his concern with categori
zation he points out that metonymy plays a major role in the structuring of
radial categories. The metonymic model is one of the 'links' that may exist
between central and less central members of a category. A metonymy involves a
function (or mapping) from one element onto another "within a single domain
structured by an I C M [an Idealized Cognitive Model]" (Lakoff 1987:288); this
single domain obviously has some degree of complexity (what we have is, in
Langacker's terms, a domain matrix). Metaphor, on the other hand, typically
involves two distinct domains, a source domain and a target domain. Typical
instances of metonymy are synecdoches, in which a part stands for the whole,
as in We need a couple of strong bodies for our team {= strong people or
players) (example from Lakoff and Johnson 1980:36), or vice versa, but the
relation between the two elements concerned is manifold, as other examples
from Lakoff and Johnson (1980: 38-39) illustrate.
(1) He bought a Ford producer for PRODUCT)
The gun he hired wanted fifty grand {object for USER)
A Mercedes rear-ended me {controller for CONTROLLED)
I don't approve of the government's actions {institution for PEOPLE
RESPONSIBLE)
E R O M THREE RESPECTABLE HORSES' M O U T H S 177
Watergate changed our politics (place for EVENT)

As Lakoff and Johnson point out, metonymic concepts are systematic. They
"allow us to conceptualize one thing by means of its relation to something else";
and they "are grounded in our experience" (p. 39).
Working along the lines of Cognitive Grammar as understood by Langacker,
Croft (1993) proposes an analysis of metonymy as "the highlighting of
an aspect of a concept's profile in a domain somewhere in the entire domain
matrix or domain structure underlying the profiled concept" (p. 354). In the
encyclopedic view of semantics held by CG, he points out, such a view of
metonymy would seem to include meanings which are normally considered as
literal. An instance is (2a-b) (Croft's (21) and (22)).
(2) a The book is heavy
b The book is a history of Iraq
Whereas "[t]he concept [BOOK] is profiled in (at least) two primary domains,
the domain of physical objects and the domain of meaning or semantic content",
we find that in (2a), in its non-metaphorical reading, "the physical object domain
of book is highlighted by virtue of the requirements of the predicate heavy". In
(2b) on the other hand, "the semantic content of book is highlighted, again due
to the requirements of the predicate be a history of Iraq" (Croft 1993: 349).
These are not to be regarded as examples of metonymy, however,
"because the elements profiled in each domain are highly intrinsic; no reference
is made to external entities. For both of these reasons, the word book is not
always treated as metonymic, or even ambiguous, in these sentences" (p. 349).
When these conditions are absent, however, we have classic cases of metonymy.
Metonymy in this sense is then illustrated by (3) and (4), Croft's (18) and (20).
(3) Proust is tough to read
(4) Time took over Sunset magazine, and it's gone downhill ever
since.
In spite of the fact that in the encyclopedic view of semantics the works of
Proust and the company that produces Time magazine are part of the concepts
[PROUST] and [TIME-MAGAZINE] respectively, they are less central and - more
importantly - less intrinsic than the semantic content and physical domain
aspects of the profile of [BOOK]. On the other hand, the domain matrix of an
encyclopedic characterization of [PROUST] will include the domain of human
activity, or more specifically human creative activity (the problem of the proper
individuation of domains arises here). This domain is probably not as central as
the domains of physical objecthood and living things to the meaning of Proust,
178 Louis GOOSSENS
but "[s]ince Proust's claim to fame is that he is a writer, and the work produced
is a salient element in the domain of creative activity, the metonymic shift is
quite natural" (p. 348).
Still following Croft, we accept that the difference between intrinsic and
extrinsic is a continuum, where (2) exemplifies the highlighting of very intrinsic
facets of a concept, and (3) and (4) the highlighting of clearly extrinsic ones. In
between are instances like (5) and (6).
(5) She came in through the bathroom window (Croft's (24))
(6) We need a couple of strong bodies for our team (Croft's (25),
taken over, as the reader may have found out, from Lakoff and
Johnson)
In (5) "[t]he interpretation of [WINDOW] as an opening in the shape domain is
somewhat extrinsic because it makes crucial reference to what is around it -
contrast the use of window to describe a physical object in a hardware store
showroom - though it appears to be less extrinsic to the concept [WINDOW] than
the publishing company and writings in examples (18) and (20) above [our (3)
and (4)]" (p. 349-350). Whereas (5) is still intrinsic enough not to be taken as
an instance of metonymy, (6), which exemplifies synecdoche, is generally
treated as a subtype of metonymy: the selection of bodies "is sanctioned by the
need to highlight the physical strength/ability domain underlying the domain of
human beings" (p. 350).
Summarizing this position, we accept with Croft that metonymy is part of
a more general process, domain highlighting, where metonymy involves the
highlighting of an extrinsic entity/facet in a possibly nonprimary domain of the
concept. I take instances like (5) to be out, but synecdoches like (6) to be in; but
it follows from this conception of metonymy that the boundary with 'literal
uses' will not always be easy to draw.
3. Metaphtonymy
After having been concerned with the definition of metonymy in relation to

literal meaning, we must now consider its interaction with metaphor. Since that
was the subject of the preceding paper in this volume, I can restrict myself here
to listing the types discussed there that are relevant to our purposes.
It will not come as a surprise to hear that the two main types, viz. meta
phor from metonymy and metonymy within metaphor are also the ones that
F R O M THREE RESPECTABLE HORSES' M O U T H S 179
turned up in our diachronic data: see sections 6.2 and 6.3 respectively. In
addition, also demetonymization, one of the less current types, showed up.
Again we do no more here than refer the reader to the metaphtonymy paper and
to the discussion in 6.4 in the present one.
The only point that we would like to emphasize here, is that in the same
way as we can discern a continuum from 'literal' to 'metonymic', there is one
from metonymy to metaphor. The difference between metonymy and metaphor
is that for the former the mapping occurs within the domain matrix of a given
concept, whereas for the latter the mapping is between two discrete domains.
To the extent that two domains can on some occasions be conceptualized as
joined together in one domain matrix, and on others as being dissociated, which
is exactly what we claim to be characteristic of metaphor from metonymy, an
area can be postulated where the distinction between metonymy and metaphor
gets blurred. The rule is, however, that we have a clear notion in specific
contexts which of the two is at stake. In our corpus a few instances of metaphor
from metonymy turned up in context, though it also appeared here that its main
importance is as a diachronic process.
4. The data
The data on which I rely to explore the role of metonymy in a diachronic

perspective are those studied in Goossens (1993b). They are all instances with
the item mouth (mouth(e), Þ), coming from Ælfric (late Old English),
Chaucer (Middle English) and Shakespeare (early Modern English). All of them
are connected with linguistic action one way or another. For the purposes of
this paper I restrict myself to those items that are relevant to metonymy, which
amounts to 55 cases from Ælfric, 26 from Chaucer and 36 from Shakespeare
(on reconsidering the data from Shakespeare, I have removed one instance from
the group of metonymies). In actual fact, these samples include all the uses
relevant to metonymy from Chaucer, and a clear majority of those in the works
of Ælfric and Shakespeare. For details on how they were arrived at, see
Goossens 1993b, sections 3 to 6).
The reason for concentrating on the metonymic mappings involving
mouth in this paper, as well as in the earlier ones, is that, as far as the
'figurative' use of body parts in relation to linguistic action is concerned, mouth
comes out as a clear first, not only in entries in the LDOCE, which are
representative of Present-day English, but also in the Toronto Concordance to
180 Louis GOOSSENS
Old English (as I found out in the preparation of Goossens (1993a; for some
evidence, see there). Given its frequency in Chaucer, and especially in Shake
speare, moreover, there are good grounds to believe that it is an extremely
suitable item for the kind of diachronic investigation that I wish to undertake
here.
Granting then that this is a somewhat restricted database in more than one
respect (only 108 items in all; domain of linguistic action as the only common
denominator), it will appear that a considerable number of complexities
concerning metonymy can be illustrated from it, and that at least some
diachronic perspective on metonymy can be gleaned from it.
5. A domain matrix for Þ, mouth(e), mouth
An important point with respect to the concept symbolized by -mouth(e) -

mouth is that it is remarkably stable over the span of time covered by our data
(nearly 1,000 years if we include Present-day English mouth into the picture).
The meaning definition in the Longman Dictionary of Contemporary
English under (7) is applicable to the Early Modern English, the Middle
English, and the Old English item as well.
(7) the opening on the face through which a person or animal can take
food into the body, and speak or make sounds (LDOCE, mouth-1,
la)
The only noteworthy difference is one between Old English Þ and its
counterparts in the later stages of the language: the Old English item does not
share the meaning quoted under (8), because in that sense Old English uses a
derivative from Þ, viz. Þ.
(8) an opening, entrance or way out (as in the mouth of a river)
(LDOCE, mouth-1, 2)
Whereas from the point of view of the later stages of English, the meaning
under (8) can be regarded as a metaphorical extension of the meaning under (7),
it is from a diachronic viewpoint a syncretism of two (related) items into one.
Since this syncretism does not concern linguistic action (and does not involve
metonymy, for that matter) it need not occupy us here any further.
Limiting ourselves to the conceptualization formulated as (7) then, let us
see what the domain matrix of mouth amounts to.
F R O M T H R E E RESPECTABLE HORSES' M O U T H S 181
Figure 1: The domain matrix of mouth
(i) Its primary domain (as understood by Langacker (1987:165)) is the

human face, with ultimately the human body as its basic domain
(ibid.: 148); for animals that can be taken to have mouths, the primary do
main would be the head (to the extent that animals are conceived as hav
ing no faces).
Among the secondary domains two stand out which both relate to basic
functions of mouths.
(ii) That of eating and drinking (the intake of food and drink into the body).
(iii) That of speech production, or linguistic action (in humans) or more
generally sound production (for humans and animals).
Other functions, which can be taken to yield tertiary domains in that they are
less central in our conceptualization of what mouths do (and which are not
included in the definition from LDOCE quoted above) are:
(iv) vomiting
(v) spitting
(vi) sucking (although that may be regarded as a partial specification of the
intake of food)
(vii) kissing, and no doubt others.
182 Louis GOOSSENS
Certain features are shared by some of these domains: (3) and (7) are
'communicative', for example, the others are not; (2), (6) and - at least partially
- (7) are 'ingressive', whereas (3), (4) and (5) are 'egressive'.
The diagram above (Fig.1) is an informal way to represent this. The bold
circle indicates the 'thing' [MOUTH], the relevant domains are indicated as
ellipses, with a full line for the primary domain, a broken line for the secondary
domains and dotted lines for the tertiary domains.
The prototype for mouth is no doubt the human mouth and can be charac
terized, following the paraphrase in (7), as "the opening on the face through
which a person can take food/drink into the body, and speak or make sounds",
but this characterization is, of course, not absolute, since there may be variation
in the amount of detail that is included.
Note, furthermore, that what we are going to be concerned with, is the
mapping of mouth onto an element or structure in its secondary domain (iii)
only, viz. linguistic action. This implies that in our corpus data speakers choose
to access elements or structures in the domain of linguistic action through a
symbolic unit that denotes a 'thing' belonging to a secondary domain from its
viewpoint.
6. Metonymy in perspective
6.1 Preliminaries
In our exploration of metonymy on the basis of the diachronically differentiated

data in whichÞ-mouth(e) - mouth (henceforth mouth as a cover term) is
used in relation to linguistic action, the following steps will be taken.
First I review the different types of metonymy that were adopted in
Goossens (1993b); it will appear that, in spite of their diversity, they are
motivated by a shared conceptual feature (see 6.2.). Next, I consider the
continuum literal ... metonymic to which attention was drawn in section 2
(6.3). Finally, we consider the interpenetration of metaphor and metonymy, the
'metaphtonymic' side of our metonyms (6.4).
F R O M THREE RESPECTABLE HORSES ' M O U T H S 183
6.2 Metonymic patterns
Table 1. Five types of metonymy
Ælfric Chaucer Shakespeare

(55/59) (26?/27) (36/61)
0) = what is said/ words/speech 11 14 14
(ii) = X's words = X as speaker 24 8 8
(iii) = speaker 3 2 5
(iv) = speech faculty 6 1 3
(v) = metonymic expressions with verb 11 1? 6
In my earlier analysis five types of metonymy appeared to be relevant for each

of the three samples under scrutiny here. The table above lists them, together
with a synopsis of how they were represented in the Ælfric, Chaucer and
Shakespeare data respectively. Between brackets the number of metonymies on
the total of linguistic action cases for each given sample. As was pointed out in
section 4, I have reduced the group of metonymies from Shakespeare by one,
more specifically in group (v).
Even if not all of these types are equally productive, and the decision where a
particular instance should be assigned is not always a straightforward one, it is
of interest to see how they pattern together. Before we go into this question, we
adduce two instances for each type.
(i) mouth — > [WHAT IS SAID/WORDS/SPEECH]
(9) (Chaucer, C.T., Parson, 955-60)
Synne [sinning] is every word and dede, and al that men coveyten
[desire], agayn [against] the lawe of Jhesu Crist; and this for to
synne in herte, in mouth and in dede [in mouth = 'in speaking']
(10) (Shakespeare, Measure for Measure, v 1 1309)
Is 't not enough thou hast suborn'd these women
To accuse this worthy man, but in foul mouth
And in the witness of his proper ear
To call him villain?
(ii) X's mouth/mouth ofX - - > [X'S WORDS] - - >[X AS SPEAKER]
(11) (Ælfric Hom. 11,9 78.192)
and he ðurh ðæra wealhstoda muð
and he through of-the-interpreters mouth
184 LOUIS GOOSSENS
Þam cyninge and his leode

to-the-king and his people
godes word bodade
God's words preached
(12) (Chaucer, C T . , Knight 1744)
This is a short conclusioun
Your owene mouth, by youre confessioun
Hath dampned yow
(III) mouth - - > [SPEAKER]
(13) (ÆHom 1,10 160.8)

Gewite seo sawul ut: ne mæig se muÞ
go the soul out not can the mouth
clypian
speak/cry out
(14) (Sh., Henry V I I I i 2 60)
this makes bold mouths՛,
Tongues spit their duties out and cold hearts freeze
Allegiance in them
(iv) mouth - - > [SPEECH FACULTY]
(15) (ÆHom I I 42 310, 14)

Ic soðlice sylle eow muð and wisdom
I truly shall-give you mouth and wisdom
Þam ne magon wiðstandan ne wiðcweðan
to-which not can resist nor contradict
ealle eower wiðerwinnan
all of-you adversaries
(16) (Sh., As you like it iii 2 239)
Rosalind. [...] Answer me in one word.
Celia. You must borrow me Gargantua'smouth first; 't is a
word too great for any mouth of this age's size
(v) metonymic expressions with verbs
(17) (ÆHom I, 36 548.13)
and he undyde his Þ and hi lærde
and he opened his mouth and them taught
Þus cweðende
thus saying
(18) (Sh., Much ado ii 1 322)

Stop his mouth with a kiss, and let him not speak neither
[= 'prevent him from speaking']
As I have pointed out in section 5, what happens in these data, is that speakers
choose to access an element or structure in the L A domain through the symbolic
unit mouth, which denotes a 'thing' belonging to its domain matrix. Obviously,
type (v) is somewhat special, in that there it is a larger expression (containing a
nominal whose head is the item mouth) which is mapped onto some linguistic
action as a whole (including the decision not to open one's mouth, i.e. not to
say anything), whereas in the other types the nominal with mouth is directly
mapped onto an ingredient of LA, viz. [SPEECH/WHAT IS SATO] (type (i)), the
[SPEAKER] (type (iii)), the speaker's [SPEECH FACULTY] (type (iv)); and type (iii)
involves both the mapping of mouth onto [WHAT IS SATO] and the mapping of
X'smouth onto [XAS SPEAKER].
What primarily motivates all of these metonymic types is the functionality
of the mouth in linguistic action:
• no [SPEECH] (of the prototypical sort) can be effected without the use of our
mouths ('instrument' for EFFECTED)
• [SPEAKERS] need their mouths to produce speech ('functional part' for
WHOLE)
• our somewhat abstract [SPEECH FACULTY] instrumentally presupposes the use
of our mouths to produce speech ('instrument' for POTENTIALITY  USE IT)
• [SPEAKING] (as an action; or, other events related to the speech act) involves
using (or not using) the mouth ('partial event in which the mouth is instru
mental ' for MORE COMPLEX LINGUISTIC EVENT).
In other words, whether synecdochic (types (iii) and (v), and to some extent
type (ii)), or not (types (i) and (iv)), it is the experientially salient and 'concrete'
element which is mapped onto the more complex and more abstract (aspect of)
linguistic action. The diversity of the metonymic uses then derives from the
complexity of the linguistic action domain.
6.3 Blurred, weak, and strong metonymy
The foregoing discussion is not complete, though. For one thing, because the
salience with which the elements/aspects of linguistic action are made accessible
by the item mouth is variable, and should in each instance be evaluated
186 Louis GOOSSENS
individually; for another, because the distinction between the different types may
be blurred.
Let's tackle the blurring among types first. An obvious case is type (ii), in
which the phrase X's mouth is mapped onto both [x's WORDS/WHAT X SAYS OR
SATO] and on [X AS SPEAKER], according to whether we process mouth indi
vidually or carry out the mapping for the complex noun phrase with the genitive
(or of-phrase). Clearly, we can execute the conceptual mapping in one or both
of these ways. Note that in (12), for example, the selection restrictions for
dampne(n) 'damn' invite a personal interpretation of mouth (as [SPEAKER]), but
the third person of the finite verb militates against a simple paraphrase of your
owene mouth as 'you'.
Another instance quoted above which is open to more than one me
tonymic interpretation is (15).  can be mapped onto something like 'the
capacity to say what has to be said' (which for me would come under [SPEECH
FACULTY]), as well as onto 'the right words to be spoken' (which would rather
belong to type (i)). Note that in this instance the absence of the article increases
the necessity of a metonymic reading.
This lack of determinacy is not exceptional in my data. Ultimately, it is a
principle like the Gricean maxim of Relevance that will help us to decide on the
appropriate interpretation. In some instances, for that matter, a contextual
implicature will have to be inferred on top of the metonymie interpretation, as in
the next two examples which can be adduced to illustrate this point.
(19) (Chaucer, Boece V, prosa 3, 124)
But not only to trowe [believe] that God is disseyved [mistaken],
but for to speke it with mouthe, it is a felonous [wicked] synne
This instance belongs to type (i), in that mouthe is mapped onto 'words',
[SPEECH] in the typology adopted; however, the contextual interpretation has to
be pushed beyond this: with mouthe implicates [ALOUD], or even [PUBLICLY].
The fact that we end up in contextually, or, if you wish, pragmatically deter
mined meaning is, of course, no difficulty for CG, which rejects the difference
semantics/pragmatics anyway.
Something similar holds for the following instance from Shakespeare.
(20) (As  like it, iii 1 11)
Thy lands, and all things that thou dost call thine
Worth seizure, do we seize into our hands,
Till thou canst quit thee by thy brother's mouth
Of what we think against thee
Thy brother's mouth was assigned to category (i), mouth can be said to stand
for [WORDS]. At the same time, thy brother's mouth can be interpreted as [THY
BROTHER (AS SPEAKER)], type (ii). But ultimately the most adequate meaning
specification for this combination with quit (= 'acquit') would be [THY
BROTHER'S TESTIMONY], which builds on the interpretation as type (i).
We are still left with what we can refer to as the saliency question. What I
mean by this, is that in some instances in our data the metonymic mapping is
indispensable in order to make sense contextually, whereas in others it would
seem that a literal interpretation might do as well.
None of the instances in my corpus has the salience of such show exam
ples of metonymy as the ham sandwich in Lakoff and Johnson's The ham
sandwich is waiting for his check. Still, instances like (9), (10), (14) and (15)
will not work in a literal interpretation; we might say that the metonymy has
medium saliency there. Cases like (11), (13), (16) and (17), on the other hand,
could also be interpreted literally, though ultimately a more satisfactory
interpretation is obtained if we read them metonymically. Again this is no real
difficulty with a CG view of meaning. Indeed, it only proves that a particular
item which gives access to a meaning network can be made compatible with a
given context in more than one way. Our point about additional implicatures
that can be built into contextual interpretations is in the same line of thinking.
6.4 Interpenetration of metonymy and metaphor
In the same way as 'literal' and 'metonymic' form a continuum, or, more
generally, in the same way as it is difficult to keep literal meaning and metonymy
apart in certain instances, there is also an interpenetration of metonymy and
metaphor.
Following up on my contribution on metaphtonymy and section 3 in this
paper, I will now look at this interpenetration, as it shows up in my data, from
four angles:
• the blurring of the distinction between metonymy and metaphor (6.4.1.)
• instantiations of metaphor from metonymy (6.4.2.)
• metonymy within metaphor (6.4.3.)
• demetonymization in a metaphorical context (6.4.4.).
Note beforehand that these rubrics are not necessarily a hundred percent
discrete.
188 Louis GOOSSENS
6.4.1 Blurred distinction

To set the tone for what follows, I offer three instances where we could
simultaneously decide on a metonymic and on a metaphorical interpretation.
The first is from Ælfric.
(21) (ÆHom l l 34 259.127)
ða halgan bec ða ðurh godes 
the holy books which through God's mouth
oððe Þuһ godes gast gedihte wæron
or through God's spirit composed were
Taking for granted that 'the composition of holy books' provides us with a LA-
context (in the broad sense), there is a first (metonymic) level of interpretation
in which godes  can be mapped onto [GOD AS SPEAKER], our second type.
Such an interpretation is compatible with a scene in which God is conceived,
rather anthropomorphically, but not so much out of the way for a medieval
frame of mind, as having dictated ða halgan bec. On the other hand, the
alternative offered by Ælfric himself in the phrase oððe purh godes gast points
to another reading, where we have to map godes  onto a spiritual type of
revelation, which would involve a mapping onto a discrete domain. To the
extent that the two interpretations are relevant together we can describe this as
a case of blurring between metonymy and metaphor.
Mutatis mutandis the same can be said about (22) and (23).
(22) (Chaucer, House of Fame I I , 758 (250))
Lo, this sentence [pronouncement] ys knowen kouth [known]
Of every philosophes mouth
As Aristotle and daun [don, sir] Plato
And other clerkys many oon [one]
(23) (Henry V iv 7 45)
It is not well done, mark you now, to take the tales out of
my mouth ere it is made and finished
In (22) every philosophes mouth can be mapped onto [PHILOSOPHER (AS
SPEAKER)], if we think of an oral origin for their pronouncements, or onto
[PHILOSOPHER (AS WRITER)], in which case a discrete target domain is involved.
(23) invites an interpretation of my mouth as [ME (AS SPEAKER)], as well as an
overall processing of take the tales out of my mouth as a mapping from a source
domain relating to food (more particularly as a scene, where one person takes
food out of another one's mouth). Perhaps the latter interpretation is not
directly shared by everybody; still, the continuation ere it is made and finished
minimally suggests an interpretation in terms of an object to be fashioned and

prepared properly, before 'it is taken out' (note, moreover, the shift from the
plural tales to the singular it in the ere-clause, as if Shakespeare had something
like food in mind when formulating this); note as well that he quite frequently
conceptualizes linguistic action in terms of eating and drinking (for a couple of
instances see section 6 in Goossens (1993b)).
Obviously, these examples can also be taken to be instances of vagueness.
The important point for our purposes is the alternation, or the simultaneous
presence, of a metonymic and a metaphorical interpretation. In this respect they
must be distinguished clearly from the instances in the next section, which
illustrate the possible continuum that may exist between metonymy and
metaphor.
6.4.2 Metaphor from metonymy
We get 'metaphor from metonymy', if we highlight an element in a domain

which can both be taken to belong to the domain matrix of some concept in one
interpretation and to a discrete source domain with partial mapping onto an
aspect of a discrete target domain in another, such that the two interpretations
from as it were a continuum.
An instance that nicely fits the above characterization is (24).
(24) (Two gentlemen of Verona ii 3.51)
Launce [...] and in losing thy voyage, lose thy master and in
losing thy master, lose thy service; and in losing thy
service, - Why dost thou stop my mouth?
Panthino For fear thou shouldst lose thy tongue.
Although there is no stage direction to that effect, it is clear that Panthino must
have made a gesture with his fingers in the direction of Launce's mouth. In that
sense he does something that stops his mouth (from moving, and consequently
from talking on). But the same gesture can be interpreted 'as if', i.e. as if it
stopped Launce's mouth. At the same time therefore we can say that (literally)
stopping Launce's mouth is in the domain matrix of this specific scene (the
mapping onto 'causing Launce to stop talking' is therefore a metonymy), as
well as claim that an element from a discrete source domain is mapped onto the
target domain (and that consequently this is a case of metaphor).
Something similar is illustrated by (25).
(25) (C.T., Sir Thopas 890 (2081))
Now holde your mouth, par charitee,
190 Louis GOOSSENS
Bothe knyght and lady free,

And[...]
It can be interpreted both as an instruction to hold one's mouth literally, and
hence to stop talking (metonymy), or to behave 'as if' holding one's mouth and
therefore stop talking (metaphor). Note in passing that instances that fit
metaphor from metonymy as a rule denote event structures where the two
relevant sub-events can be interpreted as occurring together or not.
Most of the time, however, potential metaphors from metonymy are either
one or the other in context: (26), for example is metonymic, (27) is a metaphor.
(26) (Richard II V 1 95)
One kiss shall stop our mouths, and numbly part
Thus give I mine, and thus take I thy heart
(27) (Henry V v 2 298)
We are the makers of manners, Kate; and the liberty that
follows our places stops the mouth of all find-faults
Still, I would at least call (26) a potential metaphor from metonymy, in that it
can easily give rise to an as if reading (one kiss will stop his mouth would then
mean something like ' I will prevent him from saying something embarrassing or
critical by being very sweet or flattering to him'); (27) can actually be taken to
build on such metonymic instances as (26).
The observation in the metaphtonymy paper that metaphor from meton
ymy is a frequent pattern follows from the data studied there. I worked with
dictionary instances which I judged to be metaphors from metonymy if they
could be used both ways, i.e. either as metonymies or as metaphors (and
occasionally in a dual interpretation). For the data investigated here, however,
the main point to be made is that metaphors from metonymy illustrate the
continuum from metonymy to metaphor, both synchronically and as a diachronic
process.
6.4.3 Metonymy within metaphor
This kind of combination, where metonymy and metaphor 'hold their own', is
well represented in our data, especially in the Shakespeare sample, though it is
present from Ælfric onwards. I adduce a couple of examples with a minimum of
discussion.
(28) (ÆLS (Auguries) 34)
[...] wyrgendras Þæra.  bið symle
[...] evil-doers whose mouth is always
mid geættrode wyrigunge afylled

with poisoned abuse filled
 in this instance invites mapping onto [SPEECH] (type (i)), but in the
(relative) clause as a whole it requires a metaphorical reinterpretation as
denoting a container with poisonous substance, or perhaps as a snake's mouth.
The clause then will be interpreted as something like [WHO ARE ALWAYS
CURSING] in the target domain.
The Shakespeare sample has at least 11 instances of this sort, where in the
metaphorical reading the mouth is to viewed as some sort of container, with
properties that one way or another are different from ordinary human mouths,
but where at the same time it can be mapped individually onto some aspect of
linguistic action. Without further comment I list a few instances with a
specification of the metonymy type.
(29) (King John, iii 1 306)
O hushand hear me! ֊ay alack, how new
Is husband in my mouth
([SPEECH] - type (i))
(30) (King John, iv 2 161)
- Hear' st thou the news abroad, who are arrived?
- The French my lord; men's mouths are full of it
(again type (i))
(31) (Richard II i 3 94)
Take from my mouth the wish of happy years
([ME AS SPEAKER] as w e l l as [WHAT I'M SAYING] -type (ii))
6.4.4 Demetonymization
Demetonymization, as understood in my metaphtonymy contribution, is also

rare in the present corpus. Still, the relevant instances may shed additional light
on this phenomenon. (32) and (33) are cases in point.
(32) (Measure for measure ii 4 4)
[...] heaven hath my empty words
Whilst my invention [imagination] hearing not my tongue
Anchors on Isabel; Heaven in my mouth,
As if I did but only chew his name,
And in my heart, the strong and swelling evil
Of my conception [thought]
192 Louis GOOSSENS
(33) (King John ii 1 457)

[...] Here's a large mouth, indeed,
That spits forth death and mountains, rocks and seas,
Talks as familiarly of roaring lions,
As maids of thirteen do of puppy dogs!
In both cases we get a similar pattern. There is a succession of two contexts (or
even three in (33)) in which mouth is active. With respect to the first of these
contexts mouth requires a metonymic interpretation: mouth for [WHAT I'M
SAYING], strictly speaking a metonymy within a metaphor, in (32), since it fits a
metaphorical context; mouth for [SPEAKER] in (33), where large mouth amounts
to [EXCESSIVE TALKER]. The second context requires a literal reading of mouth in
the overall metaphorical context. In (32) as if I did but only chew his name
builds on the literal sense of mouth ('opening on the face which is functional in
eating, more particularly in chewing') to arrive at an overall metaphorical
interpretation. For (33) the same holds for that spits forth, etc. Finally, there is a
third context to be taken into account for (33), where we have to resort to the
initial metonymic interpretation again {mouth for [SPEAKER] in Talks a
familiarly, etc.).
What these instances (and those analysed in the preceding sections) amply
demonstrate, is the great conceptual flexibility that may be required, and that we
are capable of, in the interpretation of natural language: the same symbolic item
can occasionally be used, as in the instances discussed just now, with a shift
from one sense to another, as we proceed in a given context. In addition it is
clear that the CG view of meaning, which views symbolic units as paths of
access to networks of meaning on which we may proceed to varying extents in
accordance with specific contexts, is better equipped than other theories of
meaning to capture what is going on in decoding (and encoding) natural
language utterances.
7. Conventionalization
7.1 Preliminaries
It is obvious that our three samples can do no more than give us a glimpse of
the developments connected with the symbolic unit mouth and its metonymie
mapping onto various aspects of linguistic action. What we find in this material
is the reflexes of the metonymie uses of mouth in three bodies of writing which
differ not only in time, but also in register and in dialect. It should be remem
bered, for instance, that there is no direct continuation between the late West-
Saxon standard used by the homilist Ælfric and the early modern English
standard of Shakespeare, whereas Chaucer's language is no more than an
important ingredient in the standard that is emerging at the end of the Middle
English period.
On the other hand, we have found that there is a certain stability in the
metonymic subtypes with which we have worked. A question that can therefore
be asked is to what extent the different metonymic uses are conventionalized,
where we understand conventionalization as degree of conceptual en
trenchment, which is directly connected with ease of activation.
Whether and to what extent a particular usage is conventionalized is diffi
cult to make out for older language stages; sometimes it is not easy for Present-
day English either. A prototypical usage can, of course, safely be taken to be
fully conventionalized. Metonymie uses, which we defined as involving
relatively extrinsic meaning elements (section 2) should be expected to have a
low degree of conventionalization, unless they occur with considerable
frequency.
It is with this in mind that we will confront the different metonymie types
in the three samples with Present-day English. As a criterion for conventionali
zation in today's English we consider whether a particular pattern has filtered
through into general usage dictionaries like LDOCE, COBUILD, OALDCE,
Heritage, and Webster's Third.
72 Observations
In order to detect some of the factors that are operative in the conventionaliza
tion of metonymies, let us review the subtypes we distinguished earlier from that
point of view. The (restricted) generalizations that follow from these observa
tions come in 7.3.
(i) mouth - - > [WHAT IS SAID]
Judging from the constant frequency with which this pattern occurs (Ælfric 11
times; Chaucer 14; Shakespeare 14), it would seem that it is fairly well
conventionalized.
Note, however, that the Ælfric instances are still largely an echo of their
biblical origin (with the Psalms as the chief donor source). Though Ælfric shows
some degree of independent conceptualization of linguistic action in terms of
194 Louis GOOSSENS
the symbolic unit  (as I hope to have demonstrated in Goossens (1993a)),
his works are still characterizable as belonging to the religious register for which
the Hebraistic tradition of the Bible remains an obvious ingredient. Instance (28)
quoted above is representative of this, as are (34)-(36:.
(34) (ÆCHom I I , 39 1 294.231)
and cwæð to ðam leasan mid
and said to the wicked-one with
gelærnedum ð
learned mouth
(35) (ÆCHom I, 39 604.21)
hæbbe eower gehwylc halwende lare on 
have of-you each holy lore in mouth
and sealmboc on handa
and psalm book in hand
(36) (ÆLS (Basil) 116)
he cwæð to Basilie, beo ðin muð afylled mid
he said to Basil be your mouth filled with
haligre herunge æfter Þinre bena
holy praise after your prayer/service
All of these retain their biblical ring when transposed into Present-day English;
conventionalization must be taken to be restricted to the religious register. Note,
moreover, that in (35), the addition of a determiner would be required ('in your
mouth').
Chaucer is less dependent on religious contexts, witness (37).
(37) (Boece I, Prosa 5, 57-60)
al [though] myghten tho [those] same thynges betere and plentevously
[abundantly] ben couth [be known] in the mouth of the
peple that knoweth al this
Still, the vast majority of instances occur in the Parson's tale, which is
somewhere in between a homily and a treatise of moral theology. A clearly
conventionalized pattern for Chaucer is in mouth, of mouth, with mouth(e) in
the meaning 'oral(ly)', occasionally even 'public(ly)'. It is illustrated in (9)
above and in (38).
(38) (.., Parson, 167)
Contriccioun [remorse] of herte, and shrift [confession] of mouth
As the comparison with Present-day English reveals, such (temporary)
conventionalization is no guarantee that it will be retained in later language
F R O M THREE RESPECTABLE HORSES ' M O U T H S 195
stages. Again the treatment of mouth as an uncountable, as in (35), may have

been a barrier against further conventionalization (though contemporary English
does retain a few phrases where mouth occurs without a determiner, see below).
In the Shakespeare sample restriction to the religious register is fully
abandoned. The treatment of mouth as an uncountable is less current than in
Chaucer, but it is still around; (11) is an instance in point (but, again, it could
not be transposed as such into today's English). Another example is (39).
(39) (Twelfth Night, ii 3 141)
Do't knight. I'll write thee a challenge; or I'll deliver
thy indignation by word of mouth
Here we get the phrase by word of mouth, which as such, i.e. in its phrasal form,
has made it into contemporary usage. What we witness is not the generalized
survival of uncountable mouth for [WHAT is SAID/WORDS/SPEECH], but a
conventionalization of a particular phrase with this metonym.
Otherwise, we get several instances where the metonymy is embedded
into an overall metaphorical context, which gives the whole an idiosyncratic
character; (40) and (41) exemplify this.
(40) (K.John iv 2 187)
Old men and beldams [grandmothers] in the streets
Do prophesy upon it dangerously
Young Arthur's death is common in their mouths
(41) (Richard I I V 6 37)
From your own mouth, my lord, did I this deed
Although these are processable from the point of view of the Present-day
language, they cannot be taken to be fully conventionalized. Note that both of
them illustrate the conceptualization of the mouth as a container for speech, a
conceptualization which is still operative today. But it is the broader meta
phorical contexts which make these instances idiosyncratic.
( i i ) X's mouth/mouth of X- -> [X'S WORDS] - -> [X AS SPEAKER]
In spite of the important decrease after Ælfric (24 occurrences, as op
posed to 8 for both Chaucer and Shakespeare), this metonymic pattern is fairly
stable. Looking at the way in which the relevant instances translate into Present-
day English (see instances (11) and (12), as well as (42)), we can consider this
type as being largely conventionalized.
(42) (Winter's Tale i 2 445)
[...]; nor shall you be safer
196 Louis GOOSSENS
Than one condemned by the King's own mouth

The reason for this, it seems to me, is that the metonymic mapping is double
here: from mouth onto [WORDS], and from the whole phrase X's mouth/mouth
of X onto [(X AS) SPEAKER]. This at least adds to the accessibility in a
conceptual network.
That not all instances are easily transposable into Present-day English (see
for example (21), (22) and (23)) hangs together again with the idiosyncratic
metaphorical context, witness also (43).
(43) (2 Henry V I iv 7 17)
My mouth shall be the parliament of England
(  ) mouth --> [SPEAKER] ( b u t not - - > [WHAT IS SAID, ETC.])
As can be inferred from Table 1, this pattern is comparatively rare (3 occur
rences in Ælfric, 2 in Chaucer, 5 in Shakespeare). Indeed, the vast majority of
cases where mouth maps onto [SPEAKER] are to be found under pattern (ii).
Proportionally, however, we find an increase of type (iii) in the later samples.
This indicates an increasing degree of conventionalization. Not only for
phrases like that in (44) (cf. Webster's from mouth to mouth 'from person to
person by word of mouth'),
(44) (Chaucer, House of Fame I I I 2076 (986))
Thus north and south
Wente every tydyng fro mouth to mouth
And that encresing ever moo [more]
but also for the pattern where mouth is modified by an adjective or some other
attributive word, especially one with prejorative connotations. Instance (14) can
be adduced to illustrate the latter. The conventionalized items that we find in
Present-day English are variations on this theme, see (45).
(45) (LDOCE)
loudmouth infml. derog. 'a person who talks too much and in an
offensive way'
It is cases like this that come closest to filling lexical gaps. The combination of
the adjective with mouth provides us with a compact way to denote a specific
type of speaker. Note, moreover, that these instances are largely synecdochic.
( i v ) mouth --> [SPEECH FACULTY]
This is also a rare type, which, unlike the preceding one, shows a decrease after
the Old English period (Ælfric 6, Chaucer 1?, Shakespeare 3). The instances in
the Ælfric sample clearly show dependence on their biblical origin (see, for
example, (15)). The three Shakesperean instances come with markedly

metaphorical contexts, see (16) and (46).
(46) (Love's Labour Lost ii 1 251)
But to speak that in words which his eye hath disclosed:
I only have made a mouth of his eye,
By adding a tongue which I know will not lie
Generally, this use shows a very low degree of conventionalization, witness also
its absence in dictionaries of contemporary English.
(v) Verbal expressions with mouth i n the D.O.
Given the diversity among the metonymic mappings of this type, it is worth
while to have a look at individual subtypes.
In the eleven Ælfric instances, four are of the kind illustrated in (17): he
undyde/geopenode hisÞfor 'began to speak' (or, with the negation, 'did not
speak', 'kept silent'). This type can be taken to be conventionalized in Present-
day English. I would like to suggest that conventionalization is facilitated by the
fact that the link between '(not) opening one's mouth' and 'beginning to speak'
(or, alternatively, 'remaining silent') is synecdochic; the event of opening one's
mouth is part of the speaking event. In (47) the added to-phrase could not be
parallelled in today's English, but the verbal is expression is otherwise transposable.
(47) (ÆHom 16 184)

Þæt he fordyttan mihte Þa idelan spræca, and
that he stop might the worthless speeches and
undon his Þ to wisdomes spræcum
open his mouth to of-wisdom speeches
The other instances can still be parallelled in Present-day English, but are
virtually restricted to the biblical register in which these items have originated.
They are the causative uses exemplified in (48) ('take away the capacity to
speak' and 'give the capacity to speak') and the type in (49).
(48) (ÆLS (Apollinaris) 83)
and gebæd for hine ðus, Drihten Hælend Crist իս
and prayed for him thus Lord Saviour Christ you
ðe beclysedest ðyses mannes , [...] geopena
who shut this man's mouth open
nu his muÞ Þæt he mærsige Þinne naman
now his mouth that he may-praise your name
198 Louis GOOSSENS
(49) (ÆCHom I, 26 366.24)

hi habbað dumbne Þ, and blinde eagan,
they have dumb mouth and blind eyes
deafe earan
deaf ears
Turning to the single occurrence of this type in Chaucer, quoted above as (25),
we have an instance which is perhaps metaphorical rather than metonymic
(hence the question mark in Table 1). As such it is still operative in Present-day
English, hence conventionalized (in my terminology) as a metaphor from
metonymy.
The six relevant examples in Shakespeare all contain the expression stop
X's mouth (his, my, our mouth or mouth of X), which metonymically amounts to
(cause to) stop speaking. Instances (18), (24), (26) and (27) amply illustrate it.
As the discussion of those examples has shown, they are not all on a par, but the
frequency of the combination points to conventionalization for Shakespeare.
In Present-day English the expression survives, with an extra metaphorical
interpretation, see (50).
(50) (Webster's1 stop)
stop one's mouth: to make silent; specific: K I L L <threatened to
stop his mouth>
Note also that the expanded expression stop somebody's mouth with a kiss (as
in (18)) must be the indirect origin for the related expression in (51).
(51 ) (LDOCE, first edition, mouth1)
put the mouth on someone Br.Ε & Austr sl. to (seem to) make
someone's actions or attempts unsuccessful by saying that he is
doing very well
Though the expression in (51) cannot be taken to be a conventionalization of
the expressions in our corpus, I assume that these expressions must have helped
to build this metaphor from metonymy.
Minimally, this discussion of verbal expressions with mouth as D.O. has
shown that if a synecdochic relationship is at stake it may be a factor that
promotes conventionalization. Secondly, that these metonymic expressions may
be the basis of conventionalized metaphors from metonymy.
7.3 Tentative conclusions
Perhaps the most obvious, but not necessarily unimportant, conclusion is that
the metonymies we have investigated here do not follow rectilinear paths to
conventionalization.
Type (ii) comes closest to straightforward conventionalization (in the
sense that it can be transposed into Present-day English without difficulty), as
do some (but not all) of the verbal expressions with mouth as (part of) the direct
object considered under type (v). For type (i) the conventionalization appears to
be more restricted than its sheer frequency would make us believe at first sight.
Types (iii) and (iv) have the lowest frequency in our database; but, whereas type
(iii), in which mouth is mapped onto [SPEAKER], appears to be on the path of
increasing conventionalization, type (iv), where mouth is used for [SPEECH
FACULTY], would seem to go the opposite way.
As to the factors which turned out to be relevant in the conven
tionalization process (or its opposite), the following are worth pointing out.
• The (relative) continuity of the metonymic patterning hangs together with the
stability of the conceptual network that mouth gives access to. This includes
the stability of its prototype as well as of the conceptual ingredients involved
in linguistic action with which this prototype is connected. In other words,
there do not appear to be noteworthy differences between the basic concep
tualization of our mouths and its functionality in speaking by the speakers of
Old English and by those of Modern English.
• Conventionalization may be restricted to a specific register. As we have
noted, most uses in Ælfric still betray their biblical origin. It is in this bibli
cal/religious register that they often continue to date, whereas they have not
percolated into general usage.
• Among the factors that appear to work against conventionalization, the
following may be singled out.
- A first barrier, though not an absolute one, is the shift in subcategorization
in the instances where mouth came to be used as an uncountable (for ex
ample Chaucer's in mouth, of mouth 'oral(ly)'; in a way this shift points to
increased conventionalization in Chaucer's idiolect (and beyond Chaucer,
given the currency of these phrases in other Middle English texts), but
apparently the conventionalization was not strong enough in most of these
instances for the increased conceptual distance that this status as uncount-
200 Louis GOOSSENS
able entails not to militate against a smooth transition of the pattern to the
following generations.
- Secondly, the embedding of the metonym in an idiosyncratic metaphorical
context (which we often find in Shakespeare) usually appears to be a rea
son for the conventionalization not to go through.
- Thirdly, the fact that a given metonym does not fill a lexical gap (in other
words, the fact that there is another conventionalized way of saying what
the metonym expresses) may be another barrier against smooth conven
tionalization of the metonym.
• Among the factors that promote conventionalization, there is first the
salience of the mapping that is involved. This may account, for example, for
the fact that of the two least frequent patterns, type three {mouth for
[SPEAKER]) and type four {mouth for [SPEECH FACULTY]), it is the mapping
onto [SPEAKER] that appears to be most conventionalized in Present-day
English. Experientially the connection between mouth and speaker is more
obvious ('salient') than that between the mouth and the associated speaker's
speech capacity.
With respect to type (ii), which is the one that is most easily transposable to
later language stages, there is the fact that this metonymy can be accessed
doubly. First, through a mapping of mouth onto [WHAT is SAID/SPEECH],
secondly, by way of mapping the phrase as a whole onto [X AS SPEAKER]:
double access promotes ease of activation, and apparently conventionaliza
tion as well.
Then we found that in some instances phrases tend to be conventionalized as
phrases. This applies to by word of mouth and from mouth to mouth. Why
exactly these are picked out and not others is difficult to explain. Perhaps the
availability of the frames by {way) of and from {time) to (time)/from (case) to
(case) may have helped. In the second example it may also have to do with
the fact that it provides speakers with a compact way to express a relatively
complex conceptualization.
Finally, the reason why some of the verbal expressions in group (v) may be
so accessible and open to conventionalization would seem to be that they are
close to being synecdochic, in the sense that they express a more complex
(linguistic action) event by way of a simpler, partial event (e.g. open one's
mouth). Note that also type (iii), and to some extent type (ii), are synecdo
chic.
FROM THREE RESPECTABLE HORSES' MOUTHS 201
• It is also in the group of expressions that we find conventionalized metaphors

from metonymy. Typically it is when the synecdochic relationship is no
longer relevant that the metaphorical 'leap' is/has to be taken; it is the me
tonymy that provides the experiential basis for the metaphor.
8. Some final perspective
8.1 The unobtrusive power of metonymy
As I have argued in Goossens (1993b), metonomy is an important strategy to

extend the meaning of symbolic units, probably even more pervasive than
metaphor. Although the latter point does not directly appear from this paper, it
may at least have clarified to some extent why metonymy often goes unnoticed.
For one thing, we have emphasized the thin line that separates the
'intrinsic' (or literal) from the 'extrinsic' uses, and the fact that in some contexts
both the literal and the metonymic reading may yield acceptable interpretations.
For another, the diversity of the metonymic mappings of mouth onto various
aspects of linguistic action showed up the shared feature of the functionality of
the mouth in linguistic action; this functionality can be taken to be in the
boundary zone between the intrinsic and extrinsic properties of mouth in that it
is an essential ingredient in a secondary domain of its domain matrix. This
obviously contributes to the inconspicuous character of the metonymies
concerned.
This unobtrusiveness, however, does not reduce the general import of
metonymy.
8.2 Context and relevance
Not only the difference between literal and metonymie uses may be blurred in
context, we also observed that it is sometimes difficult to distinguish among the
different metonymie subtypes. Some instances appear to be compatible with
more than one interpretation, according to the stretch of context considered. A
notable example of this sort was our second type, where mouth can be mapped
onto [WORDS, WHAT IS SAID SPEECH], whereas the nominal phrase as a whole
(X's mouth/mouth of X) stands for [X AS SPEAKER]. Ultimately, relevance with
respect to overall context is decisive as to which interpretation fits best (where
the possibility of a double fit must be one of the options).
202 Louis GOOSSENS
Reliance on the decoder's capacity to assess the relevance of possible in

terpretations with respect to the actual context was found to be exploited in
rather extreme ways in the cases discussed under demetonymization (not
surprisingly, instances from Shakespeare). Here considerable conceptual
flexibility is required to assign multiple interpretations as the context is
developing: a clear invitation, if that might still be required, to conceive of the
meanings expressed by natural language units as processes intertwined with
unfolding conceptualizations.
8.3 Interaction of metonymy with metaphor
In the same way as there is a continuum between the literal and the metonymic
senses of the item mouth, we find indications of a continuum metonymy-
metaphor, be it only for the verbal expressions where mouth is part of the object
nominal. As we have pointed out, these metonymic expressions may give rise to
metaphorical ones (metaphors from metonymy). In our diachronic data base it
was revealed that some contexts are compatible with both the metonymie and
the metaphorical interpretation, but that more often context will disambiguate
the expression as one or the other. Metaphor from metonymy therefore as a rule
denotes the quite current type of metaphor for which a metonymie origin can be
postulated; the metonymy, then, gives the metaphor its experiential underpin
ning.
Otherwise metaphtonymy manifests itself as the insertion of a metonym
into a metaphorical context. Again this requires considerable conceptual
flexibility from the decoder, and tends to give the combination an idiosyncratic
effect.
8.4 Conventionalization
Conventionalization is an evasive issue, for a number of reasons. It is, to begin

with, impossible to come up with data that reflect the continuity of successive
generations learning the language from preceding ones; our corpus is no more
than a very remote approximation of the ideal. Even though an item like mouth
remains quite stable in its semantic make-up over the centuries, and although
the conceptualization of what is going on in linguistic action has not changed in
any drastic way, the very nature of metonymy as involving extrinsic meaning
facets (or, ingredients belonging to a secondary domain in its domain matrix)
makes it likely that the meaning extensions that we are concerned with should
FROM THREE RESPECTABLE HORSES' MOUTHS 203
have a peripheral status. In other words, we expect metonymies to be marginal

in the network of conventional usages that relate to the symbolic item under
scrutiny, and therefore no more than occasional candidates for conventionaliza
tion. Especially, as was the case here, if the metonymic extensions do not relate
to specific lexicalization needs (i.e. do not fill lexical gaps).
Still, we have found a few factors that appear to be operative in either
working against or in promoting conventionalization of metonymies. A shift in
subcategorization (in casu from countable to uncountable), although not out of
tune with processes of this type elsewhere in the language, may be taken to
increase the conceptual distance from the prototypical sense(s); it appears to
have been a barrier against conventionalization (obviously not an absolute one).
Similarly, metonyms embedded in metaphorical contexts require some extra
conceptualization effort, and are therefore more difficult to assimilate.
Positive factors would seem to be experiential obviousness as regards the
connection between the prototypical sense and the metonymic extensions
(where it is especially synecdoches that fulfil this condition), lexical convenience
(as was already pointed out above), multiple access as in our second type, and
use of well entrenched lexico-semantic frames to fit in the metonymy. An
external factor would be the stability within a particular register, as was pointed
out for the biblical metonyms first registered in our Ælfric sample.
With respect to metaphors from metonymy, we found considerable proof
for the thesis that metaphors quite often have their experiential basis in some
metonymie origin.
All in all, of course, the picture is fragmentary, and the conventionaliza
tion issue requires much more extensive study.
8.5 Rounding off
Metonymies, at least of the current types studied here, form a continuum with
the so-called literal senses with which they are connected on the one hand, and,
at least in the instance of metaphor from metonymy, with metaphor on the other.
This makes it difficult sometimes to notice them, especially because there may
be some blurring among different metonymie extensions, as in the data studied
here. In addition, they often require close scrutiny of the contexts in which they
come in order to be interpreted properly: it is important to give them their me
tonymie reading also if they occur in metaphorical contexts, or to be prepared
for multiple interpretation as the context unfolds. All this make them a
204 Louis GOOSSENS
sometimes elusive, but probably even more important feature of meaning

extension than is mostly believed.
It is difficult to say anything definitive about the way in which they are
conventionalized, but this study has at least revealed some of the factors
involved, and invites further exploration of the problem.
Metaphor, Schema, Invariance
The Case of Verbs of Answering
Brygida Rudzka-Ostyn
University of Leuven
1. Introduction
Metaphor keeps eluding us notwithstanding the ever intensifying research into

its nature and functioning. We are told that it is at work when one domain is
structured in terms of another, when elements of some, usually more concrete
and tangible, domain are mapped onto a more abstract one. Immediately the
problem arises as to what is or is not preserved in the course of the mapping.
What constrains a cross-domain transfer? A number of specific factors facilitat
ing and blocking metaphorization have been suggested, but what is the su-
perordinate, most general constraint? Despite the attention it has received from
philosophers, linguists and psychologists,1 this question remains open. Recently
an answer has been formulated as the Invariance Hypothesis in Lakoff (1989,
1990).
This hypothesis claims that "metaphorical mappings preserve the cognitive
topology [i.e. the image-schematic structure] (...) of the source domain"
(Lakoff 1990: 54). Reexamining the claim, Turner (1993) argues that it is the
image-schematic structure of the target domain that matters. This structure
must be preserved, and the image-schematic structure of the source domain
which participates in the mapping must be "consistent with that preservation".
We will return to Turner's position at a later stage. What concerns us here is the
entailment of the Invariance Hypothesis, namely that abstract reasoning as such
may rely on metaphorical mappings, that most if not all our abstractions may
eventually turn out to be grounded in material reality (cf. Lakoff 1990: 39, 73).
When developing the Invariance Hypothesis, Lakoff points to the meta
phorical dimension of the most basic concepts used in talking about language.
The one that is of special interest from our vantage point is the concept of cate-
206 BRYGIDA RUDZKA-OSTYN
gory. This concept, too, is metaphorically understood. Its classical variant is

usually viewed as a bounded region or container. As Lakoff observes, the prop
erties ascribed to classical categories reflect the "topological properties of con
tainers" (1990: 52). To a certain extent, Lakoff's observation is no less appli
cable to the concepts of word, semantic field, or meaning. These, too, are
viewed as bounded regions or containers, and then not only by proponents of
classical semantics. When Langacker says that nouns designate things, and a
thing is "a region in some domain" (1988a: 95) - a bounded region in the case
of a count noun — he appeals to our metaphoric understanding of region and
boundary. When we talk of meanings as conceptual clusters, we again resort to
metaphorically charged descriptive tools. Incidentally, these clusters often come
to function as containers which the constitutive concepts are meant to fill.
Likewise, semantic fields are conceptualized as containers for semantically re
lated words. And since the container-contained opposition can be viewed in
terms of the whole-part opposition, a link between these two oppositions is eas
ily established, which in turn opens up interesting possibilities for metonymic
extensions. Leaving specifics aside, it should be clear that the CATEGORY-IS-
CONTAINER metaphor is applicable to categories of different kind and size.
Ubiquitous and valuable as a knowledge-organizing device, this metaphor
has its limitations, and so linguists have promoted a new one over the last two
decades. It is the network conception of categorial structure. 'The logic of clas
sical categories is the logic of containers", states Lakoff in the (1990) article on
the Invariance Hypothesis (p. 52). Following Lakoff, we could say that the logic
of today's categories is the logic of networks. Abstract categorial structure is
now understood in terms of nodes linked with one another in various ways.
Obviously there are no links or nodes in any physical sense. What we are deal
ing with is again a metaphoric construct. Unlike its predecessor, this construct
allows us better to capture the dynamic nature of categories, the way entities
can fade into one another, and as such is more in keeping with our relativistic
view of the universe and its components, a view supported by modern physics
and other disciplines.
In what follows we will show the relativity of metaphoric mechanisms
themselves. It will be argued that the distinction between a meaning extension
and its source actually hinges on the perspective from which the process of se
mantic change is examined. Panta rei ('everything flows'), and where in the
continuous flow the conceptualizer or language analyst places his vantage point
will have an influence on his conclusions. Lakoff, as already mentioned, claims
METAPHOR, SCHEMA, INVARIANCE 207
that a metaphoric mapping must preserve the image-schematic structure of the

source domain; for Turner, on the contrary, the structure of the target domain is
of primary importance. We will see in the course of the discussion that the
identity of the source and target domains is not always self-evident. Problems
also arise with respect to the very notion of image-schematic structure. How
much structural ground does or should it cover? Regarding highly abstract do
mains like time, one could ask "whether source-domain properties which are
maintained in a metaphorical mapping create the image-schematic structure in
the target domain, rather than merely being consistent with preexisting image-
schematic structure" (Brugman 1990: 259). And, as Brugman goes on to ask,
what would constitute such preexisting structure in the case of the temporal
domain? Finally, questions arise with regard to the interaction between meta
phor and metonymy as well as between the process of metaphorization and gen
eralization.
To probe some of these questions, we will investigate certain metaphoric
extensions operating in the field of verbs of answering. We will examine not
only the internal semantics of each verb but also the semantic organization of
the whole field. At the same time, we will test the explanatory potential of the
network conception of categorial structure, specifically the schematic network
model as proposed by Langacker (1987, 1988b). While other contributions to
this volume concentrate mainly on extensions into the domain of communica
tion, we will also look at extensions from communication into other domains.
Speech-act verbs constitute an interesting source domain and, as we shall see,
their study can be revealing with respect to general metaphoric issues.
2. Verbs of Answering
The verbs to be discussed are answer, rejoin, reply, respond, and retort. The
discussion will be based on examples drawn from several sources, including the
Leuven University Theatre Corpus.2 The focus here being on mechanisms of
meaning extension, we will present the verbs' semantics without an analysis of
the lexico-syntactic frames in which they occur. Needless to say, these frames
will play a major role in our attempts to identify the different meanings of each
verb.
2.1 Answer
The most common of the five verbs is undoubtedly answer. It has already re
ceived a great deal of attention, though primarily in its role as a verb of speak
ing. Of special interest are Wierzbicka's (1987: 37If.) observations on this vari
ant of answer. In her paraphrases intended to capture its meaning, she mentions
the speaker's awareness of being expected to react verbally to another's utter
ance as well as his/her awareness of having to react appropriately. The speaker
knows that what (s)he is going to say must somehow fit that utterance, but not
necessarily in terms of providing missing information. Wierzbicka (op. cit.: 372)
correctly observes that the meaning of answer cannot be restricted to question-
answer contexts, as is often the case in the literature. The verb is quite common
in responses to non-interrogative utterances, as in:
(1) "My God, it's true!" Yossarian shrieked and collapsed against
Nately in terror.
"There is no God," answered Dunbar calmly, coming up with a
slight stagger. (CAT: 129)
(2) "Goodbye, Tony. God bless you." Tony doesn't answer. They go
out. (TC)
However, the fact that analysts tend to link answer with questions may indicate
that this usage is typical for the verb. But then why is it typical? The explanation
is straightforward: since an answer is normally given in a context where a verbal
response is expected, and since questions, at least prototypical ones, epitomize
such expectations, the verb answer becomes most natural, and indeed proto
typical, in designating responses to questions rather than other types of utter
ance (Wierzbicka idem; Rudzka-Ostyn 1989). Moreover, as has often been
pointed out, 3 a true question is incomplete; it contains an empty slot, which is
precisely what a response needs to count for an answer. Thus, although other
speech acts may be followed by answers, questions seem to provide an optimal
context for this type of response. Hence the frequent occurrence of examples
like the following:
(3) "Aren"t you Chaplain Tappman?" demanded the obese colonel.
"He's the one," sergeant Whitcomb answered. (CAT: 387)
(4) I'm willing to answer any questions you like to put. (TC)
While genuine questions are inherently incomplete, other types of utterance of
ten acquire this property via their interaction with various pragmatic factors. To
return to the situations coded by (1) and (2) above, the utterance which pre-
cedes the response is not a question, but because it (a) allows "the assumption
of a gap (...) to be filled" (Wierzbicka op. cit.: 372) and (b) the response given
fits that gap, one can use answer to report the event. And as the gap created is
not prompted by a desire for information, the response does not have to be in
formative. Informativeness is required, however, in the case of responses typi
cally associated with answer.4 Wierzbicka (op. cit.: 373) identifies them as
"knowledge-providing devices" and rightly points out that the knowledge sup
plied is supposed to be sufficient and satisfactory. To these concepts we could
add that of control: under normal circumstances the speaker is in control of
what he says and may therefore be held responsible for his utterance, be it a true
answer or not.
To sum up, a speech act reported by answer is typically subsequent vis-à-
vis another speech act and is expected to be informative, appropriate and suffi
cient. In less typical codings, some of these notions may be suspended or elabo
rated further, and thus restricted to specific situations. In (1) and (2) above, the
requirement of informativeness is dropped, whereas in (5) and (6) answering is
confined to a very restricted context of accusations or criticism. It amounts to
acts of verbal self-defence. We will return to this particular sense of answer at a
later stage because this was the original meaning of the verb.
(5) ... was held in bail to answer a charge of petty larceny. (Webster's)
(6) How would you answer the criticism that your government has in
creased the level of taxation? (LDOCE)
Given the fact that the concepts invoked in the description of answer-related
speech events also characterize non-linguistic events, it should be possible to
use answer in situations that have nothing to do with verbal exchange. This is
indeed what happens, but the extensions are not unconstrained. Answer cannot
be used with reference to just any event that follows another; the use must con
form to some schema, and to discover this schema it is necessary to go into the
place of answers and speech acts in general within a broader categorial context.
As a category comparable with other categories, speech events constitute
a subclass of what we can call "acts" or "events". The latter subdivide along
various, often intersecting, parameters. One of them pertains to the acciden
tal/non-accidental opposition. Non-accidental acts include acts performed in
reaction to or caused by other acts, and it is, among others, to this broad sub
category that answers belong. As a verbal response to another's utterance, an
answer is a subcategory of acts triggered by other acts. It is an elaboration, a
special case of something done in reaction to something else. How such acts
link up with speech acts in general and answers in particular can be represented
as in Figure 1, where full-line arrows indicate a relationship of instantia
tion/elaboration and its opposite, schematicity, and interrupted-line arrows in
dicate extension. Sequences of three dots indicate elements that could be added
to complete the legend or the network of (sub)categories.
Simplistic as it is, Figure 1 depicts several paths by which answer could gain
access to non-linguistic acts. To see how exactly answer arrived at these ex
tended senses, it may be instructive to go back to its prototypical version. As
already mentioned, this version is appropriate when a speaker reacts verbally to
another's question in a satisfactory manner, i.e. provides the information sought.
Since questions are a subcategory of utterances, the meaning of answer can
easily be generalized to include verbal responses to any triggering speech act
(cf. (1) and (2) above). And since speech acts are a subtype of acts, the process
of generalization can be carried beyond the domain of linguistic action. At this
point, answer acquires a very general, metaphoric meaning 'to react to an
other's action'.
(7) Jesus told us how to answer a blow on the cheek but he never told
us how to answer a kiss. (TC)
(8) He answered his country's call when he wasn't much older than
you are ... (TC)
As in the case of questions and answers, one event comes as a reaction to an
other, and it is this prior-subsequent schema5 that licenses the extension. An
other motivating factor is the nature of the prior event; like a question it is as
sumed to require, or at least be able to trigger, a response. It is not surprising
then that, in addition to this general meaning of answer as illustrated above,
speakers of English have come up with more specific variants focusing on
counteracts and responses to acts meant as signals. Examples (9)-(14) may
serve as illustrations:
(9) ... a major incursion which would be answered by massive retalia
tion. (COBUILD)
(10) ... answered the enemy's fire shell for shell (Webster's)
(11) Sam answered her look with a grin. (COBUILD)
(12) A spotlight comes up between the towers. Two answering owl
hoots are heard. (TC)
(13) ... the second voice answers in the dominant... (Webster's)
(14) ... underground treasures ... seldom answer clearly when they are
queried by the geologist's instruments. (Webster's)
In the situations coded by (9) and (10) the reaction is meant as an act of retalia
tion. What comes to the fore is the concept of opposition, a concept that is not
unexpected if we bear in mind the etymology of answer and its present-day use
in the context of accusations and criticism. This concept is, however, absent
from the variants illustrated in (11)-(14), where making a responsive sound or a
signal matters and the responding entity does not have to be human or even
animate. English has further extended this use of answer to such events as
picking up the telephone when it rings or going to the door when one hears a
knock or the bell (see examples (15)-(16)). Again the extension is in keeping
with the prior-subsequent schema and the expectancy pattern associated with
prototypical answer. We have here a combination of metaphor and metonymy --
a metaphtonymy, as Goossens (1990) aptly identifies such combinations -- be
cause the front door and the telephone actually refer to sounds signalling a de
sire to be let in or to communicate.
(15) The telephone rings. Ben answers it. (TC)
(16) Andrew answers the front door. (TC)
The prior-subsequent schema itself incorporates information captured by a
higher-order correspondence schema. At a certain level of abstraction, trigger
ing acts and responses to these acts can be seen as having some common
ground by virtue of either belonging to the same category of socio-perceptual
signals or complementing one another in terms of informativeness or reflecting
the same kind of intention, good or bad, on the part of the agent. Unlike the
prior-subsequent schema, which dominates sequences of events, the correspon
dence schema generalizes over dynamic and static relations. This being the case,
it is easy to understand why answer has come to mean simply 'correspond,
match, be similar (in number, size, shape, position, character, etc.)', as in:
(17) A man answering his description has been seen in the Bedford
area. (COBUILD)
(18) parts that answer each other on a blueprint (Webster's)
Whereas in the examples discussed so far the notion of semi-similarity between
prior and subsequent events determines the path for metaphoric extension, in
other cases the concepts of fulfilment and appropriateness become a critical
factor. As already pointed out, a prototypical answer must satisfy a desire for
information and it must be appropriate. It must fit the gap created by the given
question. Since the notions of appropriateness, fulfilment or conformity are also
present in the semantics of other domains, a new possibility of metaphoric ex
tension arises. This possibility has been exploited in English, and so answer can
be used in contexts in which verbs such as fit, suit, satisfy, fulfil, serve, or meet
normally occur. Here are a few examples, one of which (cf.(20)) lends itself to a
'metaphtonymic' interpretation:
(19) These discussion groups are obviously answering a need.
(COBUILD)
(20) The government just didn't answer our hopes. (LDOCE)

(21) a first-aid kit designed to answer all common emergencies
(Webster's)
(22) The tool will answer (for) (our needs) very well. (LDOCE)
(23) a sofa that was quite often made to answer for a bed (Webster's)
What these extensions make clear is that schematic relations hold not only be
tween whole conceptual clusters but may involve particular elements of these
clusters as well. In other words, components of a conceptual multiplexity6 may
instantiate more general concepts which are not necessarily schematic for the
whole multiplexity. In our case the concept of suitability is schematic vis-à-vis
the conformity required between a prototypical answer and the question that
triggers it, but it is not schematic with respect to some of the other notions as
sociated with the central variants of answer.
A few of these notions relate to what is expected of someone confronted
with questions. Under normal circumstances, questions are meant to be an
swered and the person who complies is held responsibe for what he says. From
his vantage point, in turn, answering a question means accepting the conse
quences of being addressed. The original sense of answer, 'defend oneself
against a charge', has no doubt contributed to associating the concepts of re
sponsibility or accountability with this verb.7
Again, these concepts are associated with non-communicative acts, so it is
possible to use answer in reference to such acts. Answer is, of course, also pos
sible with combinations of verbal and non-verbal elements as non-linguistic acts
often follow or accompany verbally expressed commitments (cf. (26)-(27) be
low). In examples (24) and (25), the meaning of answer reduces to 'pay for',
'suffer the consequences of', and in (26) and (27) to 'assume responsibility',
'vouch for'. The concept of responsibility combines here with that of guarantee
ing the right outcome, which provides an interesting link with the variants of
answer which cluster around the notions of suitability and appropriateness. As
elsewhere, the semantic shift affects the collocational properties of answer,
when answer encroaches upon the semantic space dominated by verbs of paying
or vouching, it takes for, a preposition very common with these verbs.
(24) a crime for which he answered with his life ... (Webster's)
(25) You will answer to me for any damage to the car. (OALDCE)
(26) I can answer for his loyalty. (COBUILD)
(27) I will answer to you for his safety. (LDOCE)
BRYGIDA RUDZKA-OSTYN
Figure 2
a) = exs. 5, 6; b) = exs. , 4; ) = exs. 1, 2; d) = exs. 7, 8; e), f) = exs.
17, 18; g) = exs. 11, 12; h) = exs. 9, 10; i) = exs. 13, 14; j) = exs. 15, 16;
k) = exs. 24, 25; l) = exs. 26, 27; m) = exs. 19, 20; n) = exs. 21, 22, 23.
Figure 2 brings together all the variants of answer discussed in this section and
shows the main interconnections between them. The descriptions enclosed in
the boxes are not meant as exhaustive definitions. Rather, they function as
shorthand intended to label the most important concepts associated with a par-
ticular variant. To express the relationships between these concepts, we repre

sent them by means of phrases strung together and not by isolated words. Be
hind each of the representations is in fact a complex mental model of the event
which a particular submeaning evokes, and to describe it fully we would need to
bring in numerous parameters and values. Regarding the marking devices intro
duced in Figure 2, the prototypical variant is marked by bold face. A squiggly-
line arrow indicates a relationship mediated by metonymy, and a dotted-line box
stands for the domain in which the given variants operate.
The emergent network of senses demonstrates the intimate link between
metaphoricity and schematicity along the lines suggested in Rudzka-Ostyn
(1989). A higher-order concept is schematic vis-à-vis the more specific concepts
that elaborate it. From the perspective of the elaborating elements, however, the
higher-order concept is an extension; to arrive at it a great deal of specific in
formation must be suspended. Since schemata are more abstract than their in
stantiations, they often belong to primary domains other than the one in which
the instantiations are located. And if indeed this is the case, the path from in
stantiation to schema crosses domain boundaries and leads to metaphor. Such
metaphoric extension is at work when, for instance, answer is used in the sense
of 'correspond', 'fit', or 'satisfy'. If, however, a schema and its instantiation
belong to the same domain, the relationship between them is non-metaphoric.8
Vis-à-vis prototypical answer (box b), the variant which denotes a verbal reac
tion to any type of utterance is more abstract and thus extended. But because
the relationship of abstraction holds within the same domain, no metaphor re
sults. Although metaphorization commonly involves generalization, these two
processes are not identical.
It also becomes evident that in the case of present-day answer its ex
tended uses are grounded in speech acts. The domain of linguistic action is the
source domain for all the metaphoric mappings noted. Has it always been so?
To answer this question, a brief look at the verb's historical development will be
necessary.
Answer comes from and/ondswarian or and/ondswerian which meant
'swear/affirm against' (and-/ond- 'against'; swarian/swerian 'swear, affirm'). In
Old English it had already acquired the meaning 'react verbally to a charge' (L.
Goossens p.c.). Although etymologically confined to the context of rebutting
charges, andswarian also covered verbal responses to questions and other types
of utterance. These uses were recorded at the same time, i.e. in the 9th-10th
century. Soon various submeanings developed, many of which have survived to
our day. The similarity between their organizational pattern and what we have
identified above as a correspondence schema gave rise to numerous extensions
of andswarian into the domain of imitative acts as well as that of different cor
respondence relations. In turn, the similarity between a speaker's readiness to
meet the expectations of his interlocutor and acts of recompense, assumption of
responsibility or atonement led to extensions into the semantic territory nor
mally controlled by verbs such as pay, repay, be responsible, and atone. By
around 1200, andswarian was already common as a non-communicative verb
meaning 'correspond, be similar', 'assume responsibility for' and 'atone for'.
Then came variants whose meanings approached those of recompense, return,
and retaliate. And finally a number of extensions developed around the concepts
of satisfaction and fulfilment, allowing answer to become a synonym of satisfy,
fulfil and accomplish. Several of the metaphoric and non-metaphoric variants of
andswarian/ nswer acquired more specific senses like 'reply favourably to a
petition', 'sing antiphonally', 'react to a sign/signal', etc.
Figure 3 presents the main stages in the evolution of andswarian/answer
together with a few illustrations of the more interesting developments. All the
'historical' examples cited below as well as in the following sections come from
the Oxford English Dictionary or its Supplement. The dates provided inside the
network indicate the first and last occurrence of the given variant as recorded in
this dictionary. If only one date is given, the variant is still in use. In any event,
the dates are mere approximations as the different variants of answer must have
been well entrenched in the language before they were recorded. Note, moreo
ver, that the meanings isolated here do not always conform to the groupings
distinguished in the Oxford Dictionary due to the considerable semantic overlap
among these groupings.
As is evident, not all variants of andswarian have remained productive. In
some cases, moreover, their range of applications has changed. At the beginning
of the 16th century, answer acquired the meaning 'react verbally to another's
utterance in an impertinent way'. This meaning is still alive, but it now requires
the presence of back and an intransitive frame. Restrictions of a different kind
affected sense (e), which now occurs primarily in the context coded by Your
prayers have been answered. To draw attention to one more example, the vari
ant of answer which lexicalized the correspondence schema could in the 18th
century be used in causative constructions (see (o)). This is no longer possible.
METAPHOR, S C H E M A , INVARIANCE 217
Figure 3
d) How they will answer it ... at the last day I know not (1680); e) With
reluctancy to answer my Request (1689); f) Hush, Frank, never answer
your father (1853); g) The Glebe will answer to the Sylvan Reign, Great
Heats will follow, and large Crops of Grain (1697); The girl instantly an
swered to the action in her sculling (1865); i) Able to answere feast with
feast (1601);Answering love for love (1827); j) Knock but at the gate,
and he himself will answer (1597); k) The holwe rokkis answerden hire a-
gayn (1385); m) Or hit ountrid h to aunsware Ector agayne (1400); n)
Whiche bordure is answering to the degrees of the equinoxial (1391); The

Terms of our Law ... will hardly find Words that answer them in the
Spanish, or Italian (1690); The right arm answers accurately to the lefi
both in size and shape (1794); o) He could not answer to his fame The
triumphs of that stubborn dame (1713); p) If it were so, it were a griev
ous fault And grievouslie hath Ceasar answered for it (1601); r) The
lordes of euery toun wher suche thyng shold be taxed ... shold ansuere to
the tyng thereof (1480); Yet did they not answer either the threshers la
bour, or the owners measure (1596); s) Here godfaderes sullen for hem
answerie bifore pe prest ate fanstone (1200); How euery good woman
ought to ansuere for her lord in al thinge (1483); t) This he wel foresaw
and the event truly answered it (1673); He offered him a beast he thought
wad answer him wel eneugh (1816); u) In both Cases the Ends of Self
Love are equally answered (1714); v) Their labour was almost in vain
before, but now it answers (1783).
Our short diachronic survey also demonstrates that the metaphoric senses of
answer are extensions from the domain of communication. This domain is their
source domain, at least as far as the Old English data go. The factors that de
termined the growth of the andswarian category are the same as the ones in
voked to account for the uses of Modern English answer. The reason why this
should be so is clear: a word that has taken root in a language cannot acquire
just any new sense. There must be a link between what the word already means
and the sense it is to adopt. And since the present-day network of senses is the
result of earlier processes, whatever has motivated them is necessarily reflected
in this network.
2.2 Reply
Another very common verb of answering is reply. Like answer, it reports speech
acts triggered by other speech acts. But while typical answers fill some informa
tion gap, typical replies seem to focus more on "covering the same ground,
giving an appropriate return" (Webster's). This return may, but does not have to
be, triggered by a question, and, as Wierzbicka (1987: 373-75) rightly notes, its
relation to the triggering act can be quite loose. My data base provides numer
ous examples of reply (both as a verb and a noun) referring to the second
speaker's views on or reactions to earlier utterances. Here is just a small sample:
(1) "... Your job is to bomb the ammunition dumps at Bologna."
"But Fm going to be killed at Bologna," Yossarian pleaded. "We

are all going to be killed."
"Then you'll just have to be killed," replied ex-P.F.C. Wintergreen.
(CAT: 126)
(2) "Don't worry, sir," Colonel Cathcart promised General Dreedle
with a threatening look at Yossarian. "You have my personal word
for it that this man will be severely punished."
"What the hell do I care if he is punished or not?" General Dreedle
replied with surprise and irritation. (CAT: 224)
(3) "Darling, we're going to have a baby again," she would say to
Yossarian every month.
"You're out of your goddam head," he would reply. (CAT: 72)
(4) I once said to him, "It's all a game," and his reply was, "but some
times it gets serious." (TC)
When discussing the distinction between answers and replies, Wierzbicka (1987:
373f.) points also to the different motivation behind them. In her opinion, "the
person who answers feels that he has to say something in response to what has
been said; the person who replies feels that he wants to say something in re
ponse" (op. cit.: 374). The fact that answer, but not reply, extended into con
texts in which "external compulsion" is stressed (idem.; cf also examples (5)-(6)
in section 2.1), confirms her opinion, despite the considerable overlap between
the two verbs. An examination of our data base reveals that whereas reply is
quite common with questions, answer is not uncommon with triggering utter
ances which are not questions. Both, moreover, combine with lexically identical
objects, though in different syntactic frames (cf. answer letters and reply to let
ters).
Wierzbicka further observes that, unlike answers, in which the emphasis is
more on a relationship between two utterances and which can thus be given
independently of the answering person (think of answers to exercises/problems
in textbooks), "replying implies a relationship between two people", and as such
"involves the human activity of turn-taking in speech" (op. cit.: 374). This being
the case, reply, but not answer, can be used to mark turn-taking in dialogues
which involve a whole sequence of 'free' verbal responses to other responses.
The comparatively loose links between triggering utterances and replies is found
to be reflected in the syntactic frames that the two verbs take: "one can answer
a question but one can't *reply a question"; likewise, it is possible to say "This
was left unanswered", but not *"This was left unreplied (to)" (idem). There is
also a stylistic difference between the two verbs, noted by Wierzbicka as well.
Reply is certainly more formal than answer, which may have something to do
with the Latin origin of reply.
The origin is worth examining in connection with the other meanings that
reply developed over the centuries. Dictionaries link reply with Old French
replier and Latin replicare, both of which meant 'fold back, fold again, turn
back' {Webster's, OED). This meaning was still recorded between 1450 and
1574, but already in a 1385 work by Chaucer reply occurred as a verb of com
munication. In addition to the general meanings it exhibits now, it also had more
specific senses such as 'say that one did not mean what one said, retract' and
'say in response to the defendant's plea', as in these illustrations:
(5) Whiche thing is wonder, that they knowing me saiying but soth
arne nowe tempted to reply her olde praysinges. (1388)
(6) To that that he hath aunsuerd y have replyed yn such wyse ...
(1453)
(7) The plaintiff may plead again, and reply to the defendant's plea.
(1768)
From the vantage point of their Latin and Old French ancestors, the communi
cative versions of reply functioning in Middle English appear to be metaphoric
extensions -- extensions facilitated by the structural similarity exibited by acts of
folding/turning an object back and utterances made in response to other ut
terances. In both cases we witness a repetition of an earlier act and a reversal of
directionality. An object can be folded back only once it has been unfolded;
likewise, a speaker can reply only after he has been spoken to. Moreover, like
the object that returns, by being folded back, to its earlier position or form, a
verbal reply goes back to the speaker who initiated the exchange. We will have
more to say about the similarity between responsive speech acts and this type of
physical manipulation at a later stage. What needs stressing now is that since the
communicative and non-communicative variants were operating in English at
the same time, they can all be regarded as instantiations of one and the same
schema. Once the schema became available as a sufficiently salient frame, it
could be applied to any new domain organized around a similar schema, which
explains the use of replien in reference to acts of repudiation, as in:
(8) The quene Gwendolyne ... Whome Kyng Locryne forsoke and re
ply ed, And Estrylde weddid againe. (1470)
In (8), the extended sense 'send away, reject' was no doubt motivated by the
prior-subsequent schema and the concept of reversing the effect of the prior act.
Figure 4
a) = exs. 1, 2, 3, 4; b) = exs. 6, 7; c) = ex. 5; d) = exs. 9, 10, 11; e) = 12,
13; f) = exs. 14, 15; g) = ex. 8.
Within the domain of speech acts, new concepts came to be associated with this
schema, concepts that the schema did not have in the domain of physical move
ment. One of them was the notion of producing an utterance in reaction to the
triggering act. Via generalization this notion gave rise to an extended meaning
of reply: 'do something in direct/immediate reaction to'. In the situation coded
by (8), the subsequent act reverses the effect of some prior act which could have
been performed many years earlier. The subsequent act cannot be said to consti-
tute a direct and immediate reaction to whatever preceded it. The concept of
such a reaction is associated with the variant exemplified in (9)-(11):
(9) .... see ... thine eyes replying To the hues of yon fair heaven.
(1818)
(10) He sang his song, and I replied with mine. (1848)
(11) He replied with a nod. (OALDCE)
Related to the above variant is a more specific sense which can be paraphrased
as 'return an attack'. This retaliatory sense may have been reinforced by the
legal use of communicative reply as well as the concept of counterforce incor
porated in the original meaning of the verb. Folding or turning back an object is
not necessarily done in reaction to some prior act but it may involve overcoming
the object's resistance. Like the general sense, the more specific one (see (12)-
(13) below) is still alive.
( 12) The besieged replied sharply. (1829)
(13) ... poured broadside after broadside into the forts, which replied
continuously. (Webster's)
The general sense can be linked to yet another very specific meaning, para-
phrasable roughly as 'return a sound'. As the following examples indicate, this
meaning has also stood the test of time.
(14) What man that in the wodes crieth, Withoute faile Eccho replieth.
(1390)
(15) We were calling for help, but only the rocks replied.
The most important senses of reply are presented in Figure 4, which combines
synchronic and diachronic information. From the vantage point of Modern Eng
lish, sense (a) appears to be prototypical for the reply category. Note again that
if only one date is provided, the variant is still in use. Note also that a two-
headed arrow (like the one linking the meaning of Latin replicare with Middle
English replien) symbolizes a relationship of identity.
While in the case of answer all the extended uses could be related to the
domain of speech acts as their source domain, the metaphoric extensions of
reply pose a few problems. At first glance directed physical motion appears to
be the source domain for the extensions, though not for all. Variant (d), for in
stance, dominates much more directly the communicative and acoustic variants
of reply than the variant denoting physical motion. It should therefore be treated
as an extension from these variants. The fact that they were in use much earlier
would support such a conclusion (but see section 3). On the other hand, variant
(d) is not unrelated to the original meaning of replye. What is then its source
domain? If its direct linkage with communicative and acoustic reply is taken to
be the decisive factor, we are still left with two source domains: communication
and acoustics. Which one should we choose? Or should we argue that all three
domains are in fact source domains for the most general variant of reply? We
will return to these questions in the final section.
2.3 Respond
A somewhat different story is told by respond. The earliest texts give it as a

verb of communication. Its thirteenth-century predecessor respoun(d) expressed
the meaning 'react verbally to another's utterance', a meaning also associated
with the verb's most obvious sources: Old French respondre and Latin respon
dere. Webster's and Longman include Modern English respond among verbs of
answering, whereas Wierzbicka (1987) does not mention it at all. An oversight?
Certainly not. Rather, the reason for excluding the verb from consideration
seems to have been its growing non-communicative uses. Respond can now
designate many kinds of nonverbal reaction, and its role as a verb of speaking
has somewhat lost in significance. But by no means is it non-existent. On the
contrary, it is still quite common9, although, as COBUILD observes, it is limited
to rather formal contexts.
As a verb of speaking, respond has a fairly general meaning 'react verbally
to a triggering utterance' and more restricted meanings 'react verbally to a
question' and 'react verbally to an utterance according to some conventional
pattern (as e.g. in liturgy)'. What all three uses share is again the concepts of
causation, reversal in directionality and the second speaker's readiness to react.
When compared with answer and reply, respond seems to be closer to reply in
that its most typical uses involve the second speaker's willingness to react ver
bally and not so much the information gap created by the triggering utterance.
In many cases there is no gap whatsoever. Like reply, respond is rather loosely
connected with the contents of the triggering utterance, which is reflected in its
co-occurrence with the to-adjunct as well as its non-occurrence with nominal
direct and indirect objects. One can neither *respond someone nor *respond a
question. The following may serve as illustrations of the major communicative
senses:
(1) "Appleby, you've got flies in your eyes," he whispered helpfully as
they passed by each other . . . .
"What?" Appleby responded sharply, thrown into confusion by the
fact that Yossarian had spoken to him at all. (CAT:. 48)
(2) ... I've heard you say you were a gnu and Ann respond with the
unlikely information that she's a hippopotamus ... (TC)
(3) "Peckem?" repeated General Dreedle, still squinting with bewil
derment. "Just what the hell does Peckem have to do with it?" ...
"Absolutely nothing, sir!" Colonel Cathcart responded sprucely ...
(CAT: 224)
(4) ... responded negatively to the question. (Webster's)
(5) The priest says "The Lord be with you," and the people respond,
"And also with you." (LDOCE)
Whereas answers and replies are prototypically verbal reactions to utterances,
responses are often reactions to the attending circumstances, including the first
speaker's behaviour or intentions. In other words, a verbal response may be as
much a reaction to verbal as to non-verbal stimuli. Moreover, the response to a
verbal stimulus may be non-verbal. Consider in this respect example (6):
(6) He responded to my suggestion with a laugh/by laughing.
(LDOCE)
One may, of course, argue that the laughter is a symbolic representation of a
scornful utterance, but the fact that answer and reply would not fit as well in the
context of (6) shows that respond has a yet broader range of applications. The
differences become even more conspicuous in the frame suggested by L.
Goossens and A.-M. Vandenbergen (p.c.): "Do you want it?" He didn't an֊
swer/reply/*respond; he just nodded. What makes respond odd here is precisely
its stronger association with non-verbal acts. Its linkage with verbal stimuli may
be suspended altogether, in which case the meaning of respond comes very
close to that of react and can be paraphrased as 'do in reaction to', 'react to a
stimulus'. Such extended meanings are given below:
(7) ... responding to the threat of death with behavior that is a degra
dation of the human spirit. (Webster's)
(8) An electric circuit will respond most readily to impulses which
come timed to its own natural rate of vibration. (OED-S)
(9) The pupil of the eye responds to change of light intensity.
(Webster's)
As in the case of verbal responses, we have here a succession of events, one
being triggered by another. Like a speaker that reacts to another's utterance, an
inanimate entity reacts in accordance with the stimulus. This reaction is concep
tualized as originating with this entity, its directionality is thus opposite to the
stimulating force or influence. The motivating force behind such extensions is
again the similarity perceived in the structure of verbal responses and other
types of responsive acts.
Frequently respond means not only 'react' but 'react favorably', as the
examples below testify.
(10) This case proves the importance of adults responding to children's
needs as and when they arise. {COBUILD)
(11) The disease failed to respond to drugs. (LDOCE)
We could consider such uses as a case of extension via specialization and say
that speakers of English decided at one point to associate the concept of benefit
with the verb respond, and with time this association became conventionalized.
The question remains, however, what prompted them to do so in the case of re
spond and not, for instance, reply, which takes us back to Latin respondere.
This ancestor of respond already carried positive overtones and meant, among
other things, 'to pledge, promise (in return), warrant'. The concept of a positive
reaction has been carried into English and, as etymological evidence indicates,
has always been linked with respond. For instance, in the 17th century it
emerged in the variant of respond which came very close to the verb recipro
cate. Instructive here is the following example:
(12) The King should not be denied the means, by which he may re
spond the great confidence placed on him. (1642)
Against this variant, the uses illustrated in (10)-(11) can be seen as cases of gen
eralization rather than specialization since the specific constraints operating on
acts of reciprocation have been suspended. What this example makes clear is
that the notions of specialization and generalization are relative, not absolute,
and their application cannot be divorced from one's vantage point and standard
of comparison. With respect to the very general meaning 'react to some stimu
lus' as expressed by certain variants of respond, the meaning 'react favorably'
(cf. exs. 10-11) is a specialization; but with respect to the 'reciprocate' variant
of respond (ex.12), it is a case of generalization. There is, however, a usage (see
ex. 13) which appears to be a specialization vis-à-vis the 'reciprocate' variant:
(13) If one diamond were opened and partner responded two clubs,
there would not be a sound rebid. (OED-S)
Here the favorable reaction is confined to a very restrictive context of bridge
playing, specifically to acts of making a bid in reply to a partner's bid. These
acts may combine verbal and non-verbal elements.
At a certain level of abstraction, all the uses of respond discussed so far

conform to the same pattern: one act comes as a reaction to another act. Al
though the two acts in themselves may be very different, it is possible to per
ceive them as similar, and disregard the differences, because of the causal link
between them. It is possible to conceptualize them as corresponding to each
other. This possibility was pursued both in Latin and English, where the verb
came to mean 'correspond to/with', as in:
(14) The principall pointes ... do diuide the Zodiacke into foure partes
or quarters, responding to the foure parts and seasons of the
yeare. (1591)
(15) His great deedes respond his speeches great. (1600)
Figure 5
a) = exs. 1, 2; b) = exs. 3, 4; c) = ex. 5; d) = ex. 13; e) = ex. 6; f) = ex.
7; g) = ex.14; h) = exs. 7, 8, 9; i) = ex. 12; j) = exs. 10, 11.
This variant of respond gradually fell out of use, but as long as it was employed
it was fully motivated by the structural similarities holding between sequences of
triggering acts plus responses and entities brought into a relationship of corre
spondence. An additional motivating factor was no doubt the morpho-phonetic
affinity between respond and correspond.
With respect to the original meaning of respond, the meanings illustrated
in examples (7)-(12) and (14)-(15) are metaphoric extensions. The source do
main is that of verbal communication and the recipient domains include a host of
acts performed in reaction to other acts or stimuli. But from the present-day
perspective, that is, from the perspective of the very general meaning 'do
something in reaction to', the communicative uses of respond may be seen as
cases of specialization. If indeed this general meaning has become for speakers
of English the most representative meaning of the whole respond category, what
is the exact status of the more restrictive senses vis-à-vis the general sense? Can
some of them be treated as metaphoric extensions, or are they all de facto cases
of synecdoche? This is another problem which we will take up in the final sec
tion.
Figure 5 is an attempt to bring together the different variants of respond
discussed and to make explicit the links between them. As elsewhere, the de
scriptions in the boxes are intended to signal the most important concepts ex
pressed by the given variants. A question mark accompanying a box means that
it is unclear when the particular sense was recorded. Although some speakers of
English see sense (f) as prototypical for the whole respond category, others are
inclined to accord similar status to sense (a). In view of this discrepancy, we
have not used any prototypicality markings in Figure 5.
2.4 Rejoin
Unlike the verbs discussed so far, rejoin refers to speech acts which express
disagreement and/or annoyance on the part of the second speaker. The utterance
of the first speaker peeves him, which explains why his response is often quick
and sharp. The triggering utterance may, but need not, be a question; what is
important is that the response be related to it and that, as in the case of answers
and replies, some symmetry obtain between the two speech acts. A rejoinder,
however, not only takes up a point made by the first speaker but it also counters
that point, thus introducing an element of opposition. The example below illus
trates a typical context for the use of rejoin:
(1) "He can't sleep comfortably on that ship", she said. "In his present
state of mind", rejoined Andrew, "he might not sleep comfortably
anywhere". (Webster's)
The notion of opposition has been carried over from the original Middle English
use of rejoin. The verb was first used, and in fact still is, to denote a defendant's
answer to a plaintiff's replication (Webster's, OED). Webster also defines Mid
dle English rejoinen as 'join one's own plea to that of the plaintiff'. Given the
fact that the verb's French ancestor meant 'join two things together', the com
municative use of rejoin could have resulted from a metonymic extension: what
at one stage designated a legal document joined to another legal document, both
bearing on the same case, came to be linked with the content of one of those
documents and, by further extension, with a reply to a legal charge or to just
any utterance likely to arouse opposition. Indeed, in example (1) above any ref
erence to a legal context is absent; the only concept that has been preserved is
that of symmetry and comparatively mild opposition.
A similar symmetry underlies the non-communicative uses of rejoin. As
examples (2)-(5) below illustrate, in each case a subsequent event is a repetition
of a prior event, but with the opposite effect. Specifically, two or more entities
come or are brought together after they had (been) separated. And it does not
matter whether the entities are parts of a concrete object, a spatial configu
ration, or members of a group.
(2) Rejoin the two wires. (LDOCE)
(3) ... the road rejoins the highway two miles east (Webster's)
(4) He was determined to rejoin the RAF. (COBUILD)
(5) You are entitled to rejoin your escort if you wish. (TC)
To a certain extent, Modern English keeps the two versions of rejoin apart by
assigning to them different stress patterns and phonetic values. According to
dictionaries, communicative rejoin is pronounced as [ri'dჳƆin], and the non-
communicative variant as[ 1 ri'dჳƆin]. 10However, this difference is not observed
in all contexts. Non-communicative rejoin often assumes the phonetic value of
its communicative counterpart (cf. examples (3), (4), and (5)), which does not
mean that speakers of English are necessarily aware of the semantic links be
tween them.
In view of all this ambivalence, what is then the categorial status of the
uses illustrated in (2)-(5) above? Are they to be regarded as metaphoric exten
sions of communicative rejoin? Although the earliest (1456, 1530) records lo
cate rejoin in a communicative, specifically legal context, the uses encountered
since the verb became well entrenched in English indicate that the com
municative and non-communicative meanings developed simultaneously. As
already mentioned, communicative rejoin spread into the domain of more gen
eral replies, and non-communicative rejoin extended the concepts of bringing
and coming together to people and abstractions. The following examples illus
trate this evolution:
communicative rejoin
(6) This is the reioynyng of Nicholl Marshall vnto the replicacion of
Robert Bale. (1456)
(7) To this aunswere the Duke of Orliaunce replyed, and King Henry
reioyned. (1568)
(8) Knox rejoyndeth, it is not enough. (1637)
(9) Several passages both of the Preface and Body of the Discourse I
am rejoyning to. (1665)
non-communicative rejoin
(10) Whan they be drye sewe them subtylly and the lyppes wyl reioyne
togydre. (1541)
(11) As tin-soder doth knit and rejoyne a crackt peece of brasse. (1603)
(12) Her great spirite, rejoyned to the spirite Of this great masse, is in
the same enwombed. (1591)
(13) Receive the one, and soon the other Will follow to rejoin his
brother. (1769)
Taking into account the simultaneity noted and the very plausible metonymic
origin of communicative rejoin, the non-communicative uses cannot be viewed
as its extensions. If my interpretation of the historical data is correct, non-com
municative rejoin simply continues to express the meanings it has always had,
and it is communicative rejoin that is de facto an extension. It was possible to
use a verb which denotes acts of physical reunion in the context of this particu
lar type of speech act because of the underlying similarity. Whether one rebuts a
legal charge or rejoins broken pieces, one tries to undo the effects of the first
event, i.e. of the accusation and the breaking respectively. This concept of undo
ing, of reversing, some prior effect has allowed rejoin to extend into the domain
of communication. Interestingly, rejoin has not encroached upon the territory of
such otherwise symmetrically oriented speech acts as greetings because a greet

ing is never meant to undo the effect of another greeting.
As a verb of linguistic action, rejoin once had a more general sense 'say in
response to' (cf. ex. (9) above). At one time it also meant 'say in response to a
reply' (see ex. (7)), which was a case of generalization vis-à-vis the original
legal variant of the verb. A typical use of communicative rejoin preserves, how
ever, the concept of opposing the first speaker's utterance, and it is also this
concept, it seems, that has prevented the verb from uncontrolled extension into
the territory of other symmetrical sequences of speech events.
Figure 6 summarizes the development of the rejoin category.
Figure 6
a) = ex. 6; b) = ex. 1; c) = exs. 8, 9; d) = ex. 7; e) = exs. 2, 11; f) = exs.
3, 4, 5, 10, 12, 13.
2.5 Retort
The last verb to be discussed is retort. While Webster and Longman include it
among verbs of answering, Wierzbicka (1987: 136f.) groups it together with
disagree, refute, dispute, deny, contradict and the like under the rubric of
'argue verbs'. This fact alone points to the fluctuation of field boundaries. The
network approach accommodates such fluctuation in a straightforward manner
and, what is more, it provides a rationale for the divergent groupings. Once
meaning representations are viewed as conceptual clusters, and these clusters
are found to include one or more identical concepts, it is easy to show what
licenses the proposed division.
In the case of retort, there are several concepts that link it with both verbs
of answering and arguing. In addition to being subsequent vis-à-vis another act,
a retort is triggered off by that act. It comes as a reaction to blame or criticism
which, as Wierzbicka (1987: 136) rightly observes, may be implied rather than
explicit. There is again a certain symmetry, but not necessarily at the level of
expected information. Rather a retort is meant to counter an assumed attack,
and this countering may involve repeating the words used by the attacking party
(see exs. (1) and (2) below). It is often an act of retaliation, quick11 and rather
aggressive, one "that turns the first speaker's statement or argument against
him" (Webster's). A typical retort exhibits considerable skill and wit on the part
of the speaker, though it is primarily the concepts of countering, implied criti
cism, and disagreement that connect the verb retort with the field of arguing.
Another common element is the second speaker's annoyance at the hostile im
plications. The double affinity — with verbs of answering and verbs of arguing ~
is reflected in the syntactic frames in which retort can occur. It can take the
same prepositional adjuncts and nominal complements as certain verbs of ar
guing. In addition, it can combine with sentential complements and take the
noun question as its direct object, which brings it close to verbs of answering.
The following may serve as illustrations:
(1) "You've got peanut brittle crumbs on your face," Appleby re
marked to him.
"I'd rather have peanut brittle crumbs on my face than flies in my
eyes," Havermeyer retorted. (CAT: 48)
(2) "Are you ready?" Why should I be ready when you're not?" she
retorted... (LDOCE)
(3) ... will retort the question ... by another question ... (Webster 's)
(4) "We'd probably shoot you," ex-P.F.C. Wintergreen replied.
"We?" Yossarian cried in surprise. "What do you mean, wel Since

when are you on their side?"
"If you're going to be shot, whose side do you expect me to be
on?" ex-P.F.C. Wintergreenretorted. (CAT: 60)
(5) ... the plundering soldier in Georgia who retorted to Sherman's
sermon with "you can't expect all the cardinal virtues for thirteen
dollars a month". (Webster's)
To sum up, as a verb of linguistic action retort typically evokes a speech act
which is meant to counter a real or assumed verbal attack. Whatever criticism
the first speaker may have implied is rendered ineffective by the second speech
act, and in this sense is turned against him. The verb has, of course, many other
meanings, and again we are faced with the problem of their relation to its uses
as a speech-act verb. Can we still uphold the view that form-meaning pairings
are non-arbitrary? Furthermore, how does retort compare with other verbs of
answering?
Interesting in this respect is the historical development of retort. Related
to Latin retorquere 'to turn/bend back', the verb originally (i.e. from the 1550s
on) meant 'to return', 'to pay back', 'to cause to return upon or against the of
fender', 'to throw/hurl back (a weapon)', 'to turn back (a blow) upon a striker'
{OED). Soon, however, it came to be used as a verb of speaking, meaning 'to
reply in kind to', 'to reply in a sharp or aggressive manner', 'to allege in return',
'to turn an opponent's statement against him' {OED).
At first glance, there seems to be a long way from turning an object back
to saying something in reaction to something else. However, a closer look at the
two situations unveils a common ground. Turning something back presupposes
a prior act of turning; it may be viewed as returning to the initial position, as a
repetition of the earlier act, but in the opposite direction. The same pattern un
derlies an act of 'returning' a charge, an epithet, or an accusation. The object
'returned' goes back where it came from, with the different locations translating
metaphorically into the first and the second speaker. Since turning or bending
back involves one and the same object, it is not surprising that the extended uses
of retorquere should exploit the similarity between the subsequent and prior
acts, and thus enter the territory of verbs pertaining to retaliation and reci
procity. Compare the following examples:
(6) And then both the Devil and Man ... shall have their Malices re
torted upon them. (c. 1559)
(7) Thus I retort the dart thou threwst at me. (1592)
(8) We shall retort these kind favours with all alacrity of spirit. (1598-
9)
Once retort becomes entrenched in the context of returning weapons, blows, or
injuries, an analogy with negatively charged utterances is easily established.
Verbal retaliation becomes a special case of returning ill-treatment in general.
Hence by the end of the 16th century retort had already occurred as a verb of
communication:
(9) I shall straight retort all the blame ... upon yourself. (1596)
(10) I scorne to retort the obtuse jeast of a foole. (1602)
(11) Hipponax retorted their pleasantry with such keen strokes of satire
that they hanged themselves. (1734)
As retaliation involves acting against somebody, it should be possible to use
retort where intellectual opposition rather than sheer verbal abuse is highlighted.
This possibility was indeed pursued, and so retort came to designate acts of
responding to propositions by means of similar propositions to the contrary.
Here are two illustrations:
(12) Our Experiments may ... enable us to retort their Arguments
against themselves. (1660)
(13) So apt is the Comparison in most Respects ... that my Adversary
in vain Labours to retort it. (1713)
From acts of casting back weapons, returning blows, wrongs and incivilities, we
have come to acts of countering arguments. The objects acted upon have be
come increasingly abstract, but the organizational pattern has remained the
same. In each case, the second act is an imitation, or, better, a mirror image, of
the first one, which leads to a reversal of the attacker-attacked relationship.
When used as a verb of speaking, retort does not necessarily mean that
the first speaker's utterance is an explict attack on the one who responds. As
already mentioned, bad intentions may be only implied, which brings us back to
the central variants of retort as it functions in Modern English. From the dia-
chronic perspective sketched above, these variants may be viewed as meta-
phoric extensions of a schema originally grounded in physical manipulation. And
when the present-day perspective is adopted? What status should we accord to
communicative retort and its non-communicative counterparts? Note that even
now retort is still used in reference to acts which have nothing to do with verbal
communication. It can mean 'return', 'reflect' or 'retaliate', as in:
(14) retort heat (Webster's)
(15) retort a wrong (Webster's)
(16) There exists in the animals the impulse to retort upon offenders.
(Webster's)
We have already identified a few factors motivating the use of retort in the do
main of retaliatory acts. Very similar factors account for its reference to reflec
tion of heat or light. In order to be reflected, heat and light must first reach the
given surface, at least according to naïve physics. When retorted, they go back
to their initial locus, in which they resemble questions and incivilities 'returned'
to the first speaker. But, to go back to our question, should these uses be
treated as metaphoric extensions of communicative retort? We usually think of
metaphor when a structure from a more concrete domain is mapped onto a
more abstract one. The domain of speech acts is experientially more salient than
that of retaliatory acts in general, and probably more concrete than the domain
of light and heat transmission. From a synchronic perspective, the domain of
communication may then be treated as the source domain for the variants illus
trated in (14)-(16). Yet etymological data like (17) and (18) below point to a
more direct connection with retorquere. Such non-communicative uses were
recorded at the same time as, or even before, certain communicative variants of
retort developed, so diachronically, they are all extensions from the domain of
physical motion. Once again, the concept of source domain turns out to be inti
mately related to the perspective adopted.
(17) As when his vertues shining vpon others, heate them, and they re
tort that heate againe To the first giuer. (1606)
(18) ... a precious stone which, as a burning glasse, receiueth, and re
tortes the Sunne-beames (1611)
An alternative interpretation of the data is possible, however. When we look at
16th-18th century English, we notice that retort had a much broader range of
meanings than today. And since several of them arose almost simultaneously, it
could well be that the metaphoric uses resulted not so much from transfers from
more concrete to less concrete domains but from the application of the schema
underlying retorquere and its early English translations to an increasingly wider
range of domains. What I am suggesting here is that once a schema has been
abstracted and recognized as an organizational pattern for a given linguistic
category, be it a lexical item, a morpheme or a syntactic construction, it can
serve as a link with categories exhibiting similar patterns, and thus mediate ex
tensions across category boundaries. Already in the 17th century retort en
croached upon the territories of retaliate, reflect, return, respond, reciprocate,
and the like simply because all these verbs exhibited similar schematic structure.
Figure 7 below is an attempt to represent these similarities.
Figure 7
a) = exs. 1, 2, 3, 4, 5; b) = exs. 12, 13; ) =exs. 9, 10, 11; d) = ex. 6, 15,
16; e) = ex. 7; f) = exs. 14, 17, 18; g) = ex. 8.
Although several of the old uses of retort have been lost, to account for the pre
sent-day situation we have to postulate a network that is almost identical to the
one reflecting the historical development. Why? Because it charts conceptually
plausible links. In fact, on the basis of this network one could guess what paths
the item had followed in the past. Again, it does not mean that the evolution
was totally determined from the start since the actual exploitation of the given
possibilities depends on a number of factors, including pressure from neighbour
ing categories. In the case of retort, the choice went in the direction of negative
connotations, so the sense of 'reciprocate' gradually fell out of use.
3. Discussion
Having looked into the semantics of individual verbs of answering, we are now
in a position to identify the concepts that hold the whole field together. One of
them is undoubtedly the prior-subsequent schema. On several occasions we
have seen that all the central variants of the verbs in question and most of their
extended uses denote acts that follow other acts. The prior acts are the cause ~
direct or indirect — of the subsequent acts, and here we come to the next organ
izing schema which indicates that the subsequent acts are performed in reaction
to what precedes them. As a result, the effect of the prior act is, depending on
the type of activity, reversed or undone, and the primary agent becomes the goal
of the subsequent act. It is at him that the activity is now directed. This notion
of effect/direction reversal is another element that binds the verbs to one an
other. And since not only linguistic responses but also acts of retaliation in gen
eral and certain acoustic and thermodynamic phenomena are organized around
the same element, it is no wonder that these verbs expanded their semantic terri
tory the way they did.
In the case of retort, rejoin and, to a certain extent, reply, another concept
emerged as a unifying element, namely that of verbal opposition. The opposition
schema is an elaboration of the effect-reversal schema, and so an additional link
with the other verbs of answering is established.
While the semantics of retort and rejoin revolves around verbal opposition
and physical movement, answer, respond and reply allow more abstract senses.
The highest degree of abstraction has been reached by answer (and, in the past,
by respond՛, see Figure 5), which brings us to the correspondence schema. This
schema evokes situations in which two entities — things or relations — are
brought together, usually for the sake of comparison. The standard of compari-
son may, but does not have to, involve intrinsic properties. Totally different en
tities may be brought into a correspondence relation simply because they are
functionally equivalent or because one is the cause of the other. Since such
causal links exist between responses, verbal or non-verbal, and the
acts/situations that trigger them, we can regard the sequences as elaborations of
the correspondence schema. Another type of elaboration is effected by a schema
which says that the given entities are similar because e.g. they are of the same
kind. This schema appears to operate more on intrinsic elements, although the
distinction between correspondence and similarity may well involve other fac
tors. The two concepts certainly deserve further investigation, also because
categorization in general involves similarity judgments.
Of more immediate concern here is the role of the similarity schema in the
semantics of verbs of answering. As Figures 2, 4, 5, 6, and 7 demonstrate, it
provides a bridge between these verbs, specifically the variants which denote
acts of retaliation, reciprocal bids (as in bridge), echoes, or antiphonal singing.
Another unifying element is its subschema which relates to value judgments. In
extensive studies on the role of values in semantic extensions, Simon-Vanden-
bergen (this volume), Pauwels and Simon-Vandenbergen (this volume), and
Krzeszowski (1993) all point to the importance of the axiological parameter to
categorial growth (cf. also Rudzka-Ostyn 1988). This finding is confirmed by
the present investigation as responses to prior acts often subdivide along ax
iological lines.
Figure 8 presents the upper part of the schematic hierarchy that structures
the semantic space covered by verbs of answering and shows how this space is
parcelled out among them. Parentheses indicate that the verb in question plays a
minor role in instantiating the given schema,12 whereas square brackets mean
that the verb no longer instantiates it. Rejoin and retort belong to the latter
category for they have lost their general communicative sense of 'react verbally
to another's utterance'.13 They are mentioned simply to highlight the common
semantic core that the field had at one time. In order not to overload the figure,
we have introduced schemata that contain disjunctive information. The left-most
schema, for instance, comprises at least six different schemata, and to see how
exactly it relates to the individual verbs, the preceding figures must be con
sulted.
Figure 8
When we follow the evolution of the semantics of these verbs, we notice that
sense extensions do not necessarily proceed step by step. Often intermediate
stages that might be postulated on the basis of all the meanings expressed by a
given verb are skipped. Highly abstract variants come into use before less ab
stract ones. To take an example, answer had lexicalized the correspondence
schema 300-400 years before certain of its more concrete submeanings devel
oped (see Figure 3: (d), (e), (1), or (m)). Similarly, respond had expressed cor
respondence relations before it came to denote acts performed in reaction to
other acts (see Figure 5). One could, of course, say that this development was
strongly influenced by translations from Latin. But such influence cannot be
invoked in the case of answer, and so a few words of explanation are due. Be
fore, however, attempts are made in this direction, it may be useful to return to
the Invariance Hypothesis.
Its latest version, which both Lakoff and Turner accept, reads as follows:
"In metaphoric mapping, for those components of the source and target do
mains determined to be involved in the mapping, preserve the image-schematic
structure of the target, and import as much image schematic structure of the
source as is consistent with that preservation" (Turner 1990: 254). 14 We want
to know, of course, what would constitute the relevant image-schematic struc
ture in the case of our source and target domains. More concretely, when an
swer extends into e.g. the domain of satisfaction and fulfilment, what precisely is
preserved? On the side of the source we have verbal responses which, proto-
typically, are meant to meet the primary speaker's desire for information; and on
the target side we have a multitude of acts intended to satisfy someone's mate
rial needs, feelings, hopes, or expectations. We would argue that answer is used
in this particular target domain because the schema that can be abstracted from
the multitude of acts is the same as the one that underlies typical verbal answers.
In each case, the subsequent act or situation corresponds to what is wanted.
The concept of meeting a desire constitutes the common schema that sanctions
the use of answer in place of satisfy or fulfil. It is this schema, it seems, that is
preserved by the metaphoric mapping. Most, if not all, of the other concepts
associated with answer are suspended, and what is mapped is consistent with
the structure of the target. So Lakoff and Turner are right? Yes and no. As al
ways, there is a snag.
We have already mentioned problems posed by targets which do not have
any preexisting image-schematic structure (Brugman 1990). The target domains
encountered in this study are not unproblematic either. To return to the domain
of satisfaction, it subsumes many specific acts which have their individual

structural properties; but what is the structure of the whole domain? Is it not the
very abstract and vague schema which says that some state of affairs meets a
desire or an expectation? If this indeed is the case, and if it turns out that target
properties commonly overlap with the higher-order schema that can be ab
stracted from both target and source, then we may have to reformulate the In
variance Hypothesis. Since targets are in general much more abstract than
source domains, there is reason to suspect that the examples of overlap noted in
this study are not isolated cases. Tentatively, we could thus say that metaphoric
mapping preserves primarily a perception of similarity as expressed by a
schema applicable to the source and target domains.
With this reformulation we are back at the schematic network model and
the historical evolution of answer. In the context of the preceding discussion,
the evolution is no longer erratic; answer could extend into the domain of cor
respondence relations before it developed many of its more concrete senses be
cause the schema that dominates such relations (e.g. the concept of symmetry)
is also applicable to the communicative senses of the verb, and speakers rec
ognize the applicability. Flexible as it is, the human mind is able to make com
parisons at different levels of abstraction and does not have to follow a step-by-
step procedure.
Lakoff suggests at one point that perhaps "all abstract human reasoning is
a metaphorical version of imagistic reasoning" (1990: 39). But isn't the reverse
also true? While many "abstract concepts arise from metaphorical mappings of
spatial concepts" (op. cit.: 73), our encoding and decoding of such mappings
crucially involves the ability to abstract and generalize. The situation we are
facing here resembles the eternal dilemma: what comes first ~ the idea, the
frame, the schema or its instantiation? The form-giving concept or its imple
mentation? The present study supports the conclusion that metaphorization and
abstraction are in fact interdependent. The schematic-network model (itself a
metaphoric construct) has allowed us to highlight at least certain aspects of this
interdependence.
At the same time difficulties with such basic concepts as source and target
domain became apparent. Vanparys (this volume) has explored many of the
complexities inherent in the concept of source domain while our research has
brought its relativity to the fore. As the discussion of retort and respond makes
clear, it is not always obvious which of the domains that participate in semantic
extension is to be regarded as source and which as target. They depend vitally,
though not exclusively, on the perspective adopted. If you view the communi
cative meanings of retort from the perspective of its earlier uses as a verb of
physical retaliation, the communicative uses are an extension via mapping. If,
however, you take the present-day situation as your vantage point and, follow
ing some dictionaries, treat the communicative uses as primary, then all the
other uses may be regarded as extensions from the communicative meanings. In
the case of respond matters become further complicated by the shift of its centre
of semantic gravity towards non-communicative domains. For some speakers
the very general meaning 'do something in reaction to prior act' has become the
most representative for the whole category. What is then the status of the other
meanings? The communicative uses of respond are instantiations of the general
meaning, but instantiations of what kind?
This is where the concept of metonymy may be brought in. In his work on
metaphtonymy, Goossens (1990) makes us aware of the close symbiosis be
tween metonymy and metaphor and explains its mechanism. Once we adopt the
schematic-network approach, a yet broader dimension of this symbiosis reveals
itself. To appreciate its value, let us pause for a moment and consider the notion
of metonymy. Traditionally, the term has been used to cover a variety of conti
guity relations. It has applied primarily to referents of concepts. Alongside this
prototypical conception of metonymy, an extended one has arisen. Lakoff's
(1987) research into categorial structure has shown that a very similar type of
relation is attributed to concepts themselves. Underlying this attribution is again
a metaphor, this time a metaphor according to which concepts that are associ
ated with one another are viewed in terms of physically contiguous objects.
Since both language users and language analysts seem to rely on such an ex
tended conception of metonymy (cf. Rudzka-Ostyn 1989), we may want to
probe it further. One result of this probing is that any extension effected via ab
straction, metaphoric or not, can be seen as involving a metonymic disso-
ciation. Answer, which typically denotes the whole cluster of concepts isolated
in section 2.1, comes to be metonymically used to denote only part of the clus
ter. Where meaning extension involves generalization, i.e. concept suspension,
metonymic mechanisms are at work. Questions immediately arise with respect
to the reverse process. What happens when a meaning is extended via speciali
zation, via acquisition of new concepts? And is this process still metaphoric?
These questions bring us back to the semantic structure of respond. When
queried about the relationship between the most general meaning of the verb
and its communicative uses, native speakers of English refused to identify it as
metaphoric, which confirms the general tendency to confine metaphor to exten

sions from more concrete to less concrete domains. The only notion that covers
extensions in the opposite direction is that of synecdoche. But to apply it, we
again have to resort to a metaphor which equates conceptual clusters with
physical wholes and individual concepts with parts of those wholes. Thus, as
Lakoff predicted, there is no escape from metaphoric mappings. On the other
hand, it is abstract reason that makes such mappings possible.
This point finds further support in the widespread use of the prior-subse
quent schema. The schema manifests itself in the linguistic representation of
spatial and temporal phenomena as well as causality and event-participant roles.
In Mandarin, for instance, 'source' locatives are placed to the left of 'goals', re-
sultative adverbial adjuncts follow the main verb, and the ordering of clauses
reflects the order of the events to which they refer (Newman 1993). As New-
man amply demonstrates, "spatial source vs. goal, earlier vs. later in a temporal
sense, and cause vs. effect can all be seen as instantiations of an ordering be
tween entities involving some kind of starting or 'prior' point and some kind of
end or 'subsequent' point" (op. cit.: 464f.).
A very similar instantiation pattern operates in other languages, including
English. As we have seen, English, too, expresses causes and effects in terms of
spatio-temporal relations. To account for both the Mandarin and the English
data, we could invoke a conceptual metaphor that equates causes with prior
entities and effects with subsequent entities. But once more we would have to
ask what allows speakers of English, Mandarin and, in fact, many other unre
lated languages to make such equations. And the answer would again point to
the structure of the human mind, its ability to abstract and thus perceive the
relations between causes and effects as similar to spatio-temporal priority and
subsequence. If there is no escape from metaphoric reasoning, there is also no
escape from schematization.
Notes
* I am grateful to Peter Kelly for his valuable suggestions and to Hans Paulussen for com-
puterdrawing my figures.
1. Cf. Lehrer (1978, 1990), Ortony (1979), Honeck & Hoffman (1980), Lakoff & Johnson
(1980), Kittay & Lehrer (1981), Johnson (1981, 1987), Miall (1982), Paprotté & Dirven
(1985), and Kittay (1987).
2. The following abbreviations will be used to refer to the sources:
CAT = Catch-22 by Joseph Heller, New York: Dell, 1968.
COBUILD = Collins COBUILD English Language Dictionary

LDOCE = Longman Dictionary of Contemporary English
OED = The Oxford English Dictionary
OALDCE = Oxford Advanced Learner's Dictionary. New Edition.
OED-S = Supplement to OED
TC = Leuven University Theatre Corpus
Webster's = Webster's Third New International Dictionary of the English Lan
guage.
For further details, see references.
3. See for instance Lyons (1977), Miller & Johnson-Laird (1976), or Hiž (1978).
4. That this is, indeed, so is evidenced by the annoyance of speakers when the information
sought is not provided. Characteristic in this respect is the following example from my
data base: If you answer a question with a question again, I'll hit you. (TC)
5. Newman (1993) evokes this schema in his investigation into the semantics of Mandarin
gěi. My use of the same label ('prior-subsequent') is meant to underscore the broad ap
plicability of the schema. See the final section for further comments.
6. The term is taken from Talmy (1988).
7. It is thus no coincidence that English answerable has become a synonym of respon
sible/accountable. Interestingly, in many languages the equivalents of responsibility are
derived from verbs of answering.
8. Obviously, there are borderline cases, and the situation may not be so clear-cut; specific
examples are discussed in Rudzka-Ostyn (1985, 1989).
9. It is the third most frequent verb of answering in the Leuven University Theatre Corpus.
10. I thank Louis Goossens and Anne-Marie Vandenbergen for drawing my attention to this
distinction.
11. One would not normally say ?It takes him hours to retort as one could in the case of an
swer, reply or respond.
12. Reply, for instance, instantiates the verbal-opposition schema only in its peripheral
meaning 'react verbally to defendant's plea'.
13. Some dictionaries of Modern English still list this meaning, but I have not come across
any convincing examples.
14. Note that Turner wrote his (1990) paper after the (1993) one.
References
Bach, Kent & Robert M. Harnish

1979 Linguistic Communication and Speech Acts. Cambridge, Mass: MIT Press.
Ballmer, Thomas & Waltraud Brennenstuhl
1981 Speech Act Classification. A Study in the Lexical Analysis of English Speech Ac
tivity Verbs. Berlin, Heidelberg/New York: Springer Verlag.
Brugman, Claudia
1990 "What is the Invariance Hypothesis?" Cognitive Linguistics 1.257-266.
Cassirer, Ernst
1953 The Philosophy of Symbolic Forms. Vol. I: Language. New Haven: Yale Univer
sity Press.
Cloran, Carmel
1987 "Negotiating New Contexts in Conversation." Occasional Papers in Systemic
Linguistics 1.85-110.
Collins Cobuild English Language Dictionary. 1987. London/Glasgow: Collins. [COBUILD]
Concise Oxford Dictionary of Current English. 1976. London: Oxford University Press.
[CODCE]
Croft, William
1993. "The Role of Domains in the Interpretation of Metaphors and Metonymies."
Cognitive Linguistics 4.335-370.
Dirven, René, Louis Goossens, Yvan Putseys & Emma Vorlat
1982 Speak, Talk, Say and Tell Amsterdam/Philadelphia: Benjamins.
Fotion, Nicholas
1981 "I'll Bet You $10 That Betting is Not a Speech Act." In H. Parret, M. Sbisà & J.
Verschueren (eds.) Possibilities and Limitations in Pragmatics. Amsterdam: John
Benjamins, 211-224
Geeraerts, Dirk
1989 "Introduction: Prospects and Problems of Prototype Theory." Linguistics 27. (ed.
by Dirk Geeraerts) 587-612.
Geiger, R. & . Rudzka-Ostyn (eds.)
1993 Conceptualization and Mental Processing in Language. Berlin/New York: Mou
ton de Gruyter.
Good, Colin
1979 "Language as Social Activity: Negotiating Conversation." Journal of Pragmatics
3.151-167.
246 REFERENCES
Goossens, Louis
1986 "Framing the Linguistic Action Scene in Old and Present-day English: Old Eng
lish cwepan, secgan, sprecan and Present-day English speak, talk, say and tell
compared." In J. Fisiak (ed.) Papers from the Sixth International Conference on
Historical Linguistics. Amsterdam: John Benjamins, 149-170.
1987 "Dealing with Linguistic Action Verbs in Depth." In J. Verschueren (ed.) Lin
guistic Action: Some Empirical-Conceptual Studies. Norwood, N.J.: Ablex Pub
lishing Corporation, 83-106.
1989 "Prototypical and Non-prototypical Meaning: Say Revisited." In F.J. Heyvaert &
F Steurs (eds.) Worlds Behind Words. Essays in Honour of F.G. Droste on the
Occasion of His Sixtieth Birthday Leuven: Leuven University Press, 155-161.
1990a "Framing the Linguistic Communication Scene: ask vs. acsian and biddan." In H.
Andersen & K. Koerner (eds.) Historical Linguistics 1987, Amster
dam/Philadelphia: John Benjamins, 191-209.
1990b "Metaphtonymy: The Interaction of Metaphor and Metonomy in Expressions for
Linguistic Action." Cognitive Linguistics 1.323-340.
1993a "The Rise of a New Conceptualization Pattern: Old English lippe, weler, tunge
and  with Reference to Linguistic Action." In H. Aertsen & R. Jeffers (eds.)
Historical Linguistics 1989. Amsterdam/Philadelphia: John Benjamins, 141-151.
1993b ", mouth(e), mouth, Denoting Linguistic Action: Aspects of the Development
of a Radial Category." Acta Linguistica Hungarica 38.61-81.
1994 "Metonymy in the Pipeline: Another Way of Looking at the Container Metaphor".
In K. Carlon, . Davidse & . Rudzka-Ostyn (eds.) Perspectives on English.
Studies in Honour of Professor Emma Vorlat. Leuven: Peeters, 386-394.
Hiź, Henryk, ed.
1978 Questions. Dordrecht: Reidel.
Honeck, Richard P. & Robert R. Hoffman, (eds.)
1980 Cognition and Figurative Language. Hillsdale, N.J.: Erlbaum.
Jaeger, Jeri J.
1986 "Concept Formation as a Tool for Linguistic Research." In Jeri J. Jaeger & John
Ohala (eds.) Experimental Phonology. Orlando: Academic Press, 211-237.
Johnson, Mark
1987 The Body in the Mind. The Bodily Basis of Meaning, Imagination and Reason.
Chicago/London: University of Chicago Press.
----- (ed.)
1981 Philosophical Perspectives on Metaphor. Minneapolis: University of Minnesota
Press.
Kay, Paul
1987 "Linguistic Competence and Folk Theories of Language. Two English Hedges."
In D. Holland & N. Quinn (eds.) Cultural Models in Language and Thought.
Cambridge: Cambridge University Press, 66-77.
Kittay, Eva F.
1987 Metaphor: Its Cognitive Force and Linguistic Structure. Oxford: Oxford Uni
versity Press.
REFERENCES 247
----- & Adrienne Lehrer

1981 "Semantic Fields and the Structure of Metaphor." Studies in Language 5.31-63.
Kövecses, Zoltan
1986 Metaphors of Anger, Pride and Love. Amsterdam/Philadelphia: Benjamins.
Krzeszowski, Tomasz P.
1993 "The Axiological Parameter in Preconceptual Image Schemata." In R. Geiger &
. Rudzka-Ostyn (eds.) Conceptualization and Mental Processing in Language.
Berlin/New York: Mouton de Gruyter, 307-329.
Lakoff, George
1973 "Hedges. A Study in Meaning Criteria and the Logic of Fuzzy Concepts." Journal
of Philosophical Logic 2.458-508.
1987 Women, Fire and Dangerous Things: What Categories Reveal about the Mind.
Chicago/London: University of Chicago Press.
1989 "The Invariance Hypothesis: Do Metaphors Preserve Cognitive Topology?" Duis
burg: LAUD.
1990 "The Invariance Hypothesis: Is Abstract Reason Based on Image-Schemas?"
Cognitive Linguistics 1.39-74.
1992 "Metaphors and War: The Metaphor System Used to Justify War in the Gulf." In
M. Pütz (ed.) Thirty Years of Linguistic Evolution. Studies in Honour of René
Dirven on the Occasion of His Sixtieth Birthday. Philadelphia/Amsterdam: John
Benjamins, 463-482.
& Mark Johnson
1980 Metaphors We Live By. Chicago/London: University of Chicago Press.
& Mark Turner
1989 More than Cool Reason: A Field Guide to Poetic Metaphor. Chicago/London:
University of Chicago Press.
Langacker, Ronald W.
1987 Foundations of Cognitive Grammar. Vol. 1: Theoretical Prerequisites. Stanford:
Stanford University Press.
1988a "The Nature of Grammatical Valence." In B. Rudzka-Ostyn (ed.) Topics in Cog
nitive Linguistics. Amsterdam/Philadelphia: John Benjamins, 91-125.
1988b "A Usage-based Model." In B. Rudzka-Ostyn (ed.) Topics in Cognitive Linguis
tics. Amsterdam/Philadelphia: John enjamins, 127-161.
1991 Foundations of Cognitive Grammar. Vol. 2: Descriptive Application. Stanford:
Stanford University Press.
Leech, Geoffrey
1983 Principles of Pragmatics. London/New York: Longman.
Lehrer, Adrienne
1978 "Structure of the Lexicon and Transfer of Meaning." Lingua 45.95-123.
1990 "Polysemy, Conventionality and the Structure of the Lexicon." Cognitive Lin
guistics 1.207-246.
Longman Dictionary of Contemporary English. 1978, 1987. Harlow: Longman. [LDOCE]
Lyons, John
1977 Semantics. Vol 2. Cambridge: Cambridge University Press.
248 REFERENCES
Mackenzie, J. Lachlan
1985 "Metaphor in Contemporary Semantics." Dutch Quarterly Review of Anglo-
American Lettters 15.52-72.
Miall, David S. (ed.)
1982 Metaphor: Problems and Perspectives. Hassocks, Sussex: Harvester Press.
Miller, George A. & Philip N. Johnson-Laird
1976 Language and Perception. Cambridge, Mass.: Harvard University Press.
Newman, John
1993 "The Semantics of Giving in Mandarin." In R. Geiger & . Rudzka-Ostyn (eds.)
Conceptualization and Mental Processing in Language. Berlin/New York: Mou
ton de Gruyter, 433-485.
Oxford Advanced Learner's Dictionary of Current English. 1980, 1989 Oxford: Oxford Uni
versity Press. [OALDCE]
The Oxford English Dictionary. 1933. Oxford. Oxford University Press. [OED]
Papproté, Wolf & René Dirven (eds.)
1985 The Ubiquity of Metaphor. Amsterdam/Philadelphia: John Benjamins.
Pauwels, Paul & Anne-Marie Simon-Vandenbergen
1993 "Value Judgement in the Metaphorization of Linguistic Action." In R. Geiger &
. Rudzka-Ostyn (eds.) Conceptualization and Mental Processing in Language.
Berlin/New York: Mouton de Gruyter, 331-367.
The Penguin Roget's Thesaurus. 1966. Harmondsworth: Penguin.
Reddy, Michael
1979 "The Conduit Metaphor. A Case of Frame Conflict in Our Language about Lan
guage." In A. Ortony (ed.) Metaphor and Thought, Cambridge: Cambridge Uni
versity Press, 284-324.
Rudzka-Ostyn, Brygida
1985 "Metaphoric Processes in Word Formation." In W Papproté & R. Dirven (eds.)
The Ubiquity of Metaphor. Amsterdam/Philadelphia: John Benjamins, 209-241.
1988 "Semantic Extension into the Domain of Verbal Communication." In B. Rudzka-
Ostyn (ed.) Topics in Cognitive Linguistics. Amsterdam/Philadelphia: John Ben
jamins, 507-553.
1989 "Prototypes, Schemas and Cross-Category Correspondences: The Case of Ask."
1994 "Metaphor, Schema, Invariance: The Case of Verbs of Answering." In K. Carlon,
. Davidse and . Rudzka-Ostyn (eds.) Perspectives on English. Studies in Hon
our of Professor Emma Vorlat Leuven/Paris: Peeters.
----- (ed.)
1988 Topics in Cognitive Linguistics. Amsterdam/Philadelphia: John Benjamins.
Searle, John R.
1979 Expression and Meaning. Studies in the Theory of Speech Acts. Cambridge:
Cambridge University Press.
& Daniel Vanderveken
1985 Foundations of Illocutionary Logic. Cambridge: Cambridge University Press.
REFERENCES 249
Simon-Vandenbergen, Anne-Marie
1991 "The Body in Talking about Language. A Study of Metaphors." In S. Granger
(ed.) Perspectives on the English Lexicon. A Tribute to Jacques Van Roey. Ca
hiers de l'Institution de Linguistique de Louvain 17 (1-3), Louvain-la-Neuve, 51-
63.
1993 "Speech, Music and Dehumanization in George Orwell's Nineteen Eighty-Four:
A Linguistic Study of Metaphors." Language and Literature 2.157-182.
1994 "Metaphorical Verbal Clauses in George Orwell's Nineteen Eighty-Four." In K.
Carlon, K. Davidse & . Rudzka-Ostyn (eds.) Perspectives on English. Studies in
Honour of Professor Emma Vorlat. Leuven: Peeters, 395-407.
Stainton, Caroline
1987 "Interruptions: A Marker of Social Distance?" Occasional Papers in Systemic
A Supplement to the Oxford English Dictionary. Vols 1-4. 1972-1986. Oxford: Oxford Uni
versity Press. [OED-S]
Talmy, Leonard
1985 "Force Dynamics in Language and Thought." In W.H. Eilfort, P.D. Kroeber &
K.L. Peterson (eds.) Papers from the Parasession on Causatives and Agentivity at
the Twenty-First Regional Meeting of the Chicago Linguistics Society. Chicago:
Chicago Linguistics Society, 293-337.
1988 "The Relation of Grammar to Cognition". In B. Rudzka-Ostyn (ed.) Topics in
Cognitive Linguistics. Amsterdam/Philadelphia: John Benjamins, 165-205.
Traugott, Elizabeth Closs
1985 "'Conventional' and 'Dead' Metaphors Revisited." In W. Papproté & R. Dirven
(eds.) The Ubiquity of Metaphor. Amsterdam/Philadelphia: John Benjamins, 17-
56.
Turner, Mark
1990 "Aspects of the Invariance Hypothesis." Cognitive Linguistics 1.247-255.
1993 "An Image-Schematic Constraint on Metaphor". In R. Geiger & . Rudzka-Ostyn
(eds.) Conceptualization and Mental Processing in Language. Berlin/New York:
Mouton de Gruyter, 291-306.
Vanparys, Johan
1993 The Categorization of Illocutionary Verbs. Doctoral Dissertation, K.U.Leuven.
Verschueren, Jef
1981 "Problems of Lexical Semantics." Lingua 53.317-351.
1984 "Basic Linguistic Action Verbs. A Questionnaire." Antwerp Papers in Linguistics
37.
1985a What People Say They Do With Words: Prolegomena to an Empirical-Conceptual
Approach to Linguistic Action. Norwood, N.J.: Ablex Publishing Corporation.
1985b International News Reporting: Metapragmatic Metaphors and the U2 Incident.
Amsterdam/Philadelphia: John Benjamins.
1987 "Metapragmatics and Universals of Linguistic Action." In J. Verschueren (ed.)
Linguistic Action: Some Empirical-Conceptual Studies. Norwood, N.J.: Ablex
Publishing Corporation, 125-140.
250 REFERENCES
Webster's Third New International Dictionary of the English Language. 1976. Springfield,
Mass.: Merriam. [Webster's]
Wierzbicka, Anna
1985a "A Semantic Metalanguage for a Crosscultural Comparison of Speech Acts and
Speech Genres." Language in Society 14.491-514.
1985b "Different Cultures, Different Languages, Different Speech Acts. Polish vs Eng
lish." Journal of Pragmatics 9.145-178.
1987 English Speech Act Verbs. Sydney: Academic Press.
Subject Index
Categories 206, 209f in relation to image schemata and

as containers 206 value judgements 35f, 108f
network conception of 206f, 209- problems with identification 126f,
211, 214-217, 221f, 226f, 229f, 139, 207, 222, 227, 229f, 233f,
236-242 240f
Conduit metaphor 2-3 Image schemata
specific variants of and alternatives and linguistic action 43f, 51
to 4-5, 12-34 in relation to donor domains and
Container schema 45f, 148-149, 155- value judgements 52f, 108f
156 types of 44
Continuum See also Container schema; Control
from literal to metonymic 176-178, schema; Correspondence
182-187 schema; Prior-subsequent
from metonymy to metaphor 150- schema; Invariance Hypothesis
152, 164-165, 168-169, 174, Invariance Hypothesis
187-189, 202 Landmark
Control schema 50, 67, 79-80, 90-91, hidden 133, 139, 147, 148, 156-157
146-148 Linguistic action 159, 161, 180-182
Conventionalization, See Metaphor; vs. other types of action 36f, 209-
Metonymy 214
Correspondence schema 212, 214, 226, as donor domain 211-218, 220-222,
235-238 224-227, 236-238
Demetonymization 171, 173, 179, 191- Meaning 206
192 extension of 202, 203
Domains, See Recipient domain; Donor metaphoric vs. non-metaphoric 133,
domain 215
Donor domain 2f, 160, 162, 164, 205- See also Metaphor; Metonymy;
207, 215, 239-241 Metaphtonymy
complex 33f, 51, 144-150, 209f prototypical 71, 111,215
for evaluating linguistic behaviour instability of 227, 241
62f, 105f Metaphor 4-5
involving body parts 28f, 35f, 105, and image schemata 43f, 146-150,
143-144, 162, 166-172 240-243
and value judgements 7 If, 232, 238
252 SUBJECT INDEX
conventionalized 5-6 Target domain, See Recipient domain

purpose of 111f Value judgements
related to linguistic action as based on specific and concrete
interaction 8Of donor domains 52f, 6If, 66, 71,
related to manner of speaking 91f 110
See also Conduit metaphor; based on scales 53f, 108f
Metaphtonymy; Metonymy context-(in)dependent 57f, 71, 78,
Metaphorization 205-207, 239-242 80, 110f
facilitating factors 126f, 236-239 in relation to donor domains and
levels of 125f, 211-217, 236-239 image schemata 53f, 62f, 66,
vs. schematization/generalization 105f
215, 240-242 introduced by metaphor 52f, 71f
Metaphtonymy 159-174, 178-179,212 merged 59, 77, 79
types of 150-152, 164-165, 168- prototypical 71, 78, 111
173, 178-179, 169-192
See also Metaphor; Metonymy
Metonymy 212, 228, 241
and metaphor, See Metaphtonymy
and literal usage 175-178
as means of categorial extension
176-178, 241f
conventionalization of 192-198,
199-201
types of 183-187
Onomatopoeic expressions 92, 108
Prior-subsequent schema 211f, 236,
242
in Mandarin 242
Recipient domain 14f, 145, 160-161,
205,207,209-211
Recoverability 125f, 157-158
and metaphoricity 125f, 154
and specificity 125f, 152-154
Salience 125-126, 157,200
Scales
and value judgements 53f, 62f, 108f
Source domain, See Donor domain
Synecdoche 176-177, 197, 200, 227,
241f
See also Metonymy

By Word of Mouth

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

By Word of Mouth

Uploaded by

Copyright:

Available Formats

BY WORD OF MOUTH

Pragmatics & Beyond

Louis Goossens, Paul Pauwels, Brygida Rudzka-Ostyn,

JOHN BENJAMINS PUBLISHING COMPANY

Library of Congress Cataloging-in-Publication Data

A Survey of Metalinguistic Metaphors 1

Body Parts in Linguistic Action: Underlying Schemata and Value Judge- 35

Assessing Linguistic Behaviour: A Study of Value Judgements 71

Levels of Metaphorization: The Case of Put 125

Metaphtonymy: The Interaction of Metaphor and Metonymy in Figura- 159

From Three Respectable Horses' Mouths: Metonymy and Convention- 175

Metaphor, Schema, Invariance: The Case of Verbs of Answering 205

Subject Index 251

We would like to characterize our book as empirical, because the research it

There are seven contributions. The first five (Vanparys, Pauwels/Simon-

associated with the metaphorical expressions investigated exhibit both a con-

Goossens to shed light on two issues. The first is the literal-metonymy-

Summarizing, the following areas of investigation have received attention.

• The graded character of metaphoricity and its correlation with degrees of

The most influential account of metalinguistic metaphors to date has been

Working from the observation that most metalinguistic terms, whether

account, address, adjective, admission, admit, advice, advise, agree,

The content of LDOCE with regard to our purpose is illustrated diagrammati-

The adjective abrupt was eliminated because of its applicability to human

4. The conduit metaphor revisited

Reification is no doubt one of the major metaphorical patterns, as it relates

conceptualization on the basis of the acoustic input. The conduit metaphor

4.1 Linguistic expressions as containers

The corpus contains several items supporting (a)-(c):

The container metaphor is evoked in a rather neutral fashion by contain (10)

4.2 Linguistic expressions as objects

4.2.1 Verbs of making

4.2.2 Verbs of adorning

4.2.3 Verbs of expanding and compressing

4.2.4 Other verbs of changing

4.2.5 Visual perception

4.3 Transfer of information

As specified by the conduit metaphor, speakers communicate by physically

4.3.1 Verbs of giving

(38) May I offer my congratulations? (COBUILD)

4.3.2 Verbs of bringing

4.3.3 Lay, put, drop and related verbs

4.3.4 Across, home, through

4.3.5 Violent transfer

4.3.6 Verbs of retracting

4.3.7 Verbs of keeping

4.3.8 Verbs of taking

Similarly, to take s. th. for granted denotes uncritical acceptance/belief. Various

4.3.9 Verbs of exchanging

4.3.10 Verbs of disseminating

4.4 Self-propelled motion

According to the conduit metaphor, in both versions, it is S who transfers word-

4.5 Target-directed transfer

In the conduit metaphor, as formulated by Reddy, words are transferred by S

4.6 S and H as containers

(94) I'll fill you in / up on the details now. (COBUILD)

4.7 Motion on the part of S

Lakoff & Johnson (1980) have identified the CONVERSATION-AS-A-JOURNEY

4.7.1 Motion towards H

semantic divergence in the recipient domain: these expressions have a common

4.7.2 Motion towards the topic

4.7.3 Towards the goal of the conversation

4.7.4 Manner of movement