Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
EDITORS
Marcelo Dascal (Tel Aviv University)
Raymond Gibbs (University of California at Santa Cruz)
Jan Nuyts (University of Antwerp)
Editorial address: Jan Nuyts, University of Antwerp, Dept. of Linguistics (GER),
Universiteitsplein 1, B 2610 Wilrijk, Belgium, e-mail: nuyts@uia.ua.ac.be
Volume 4
Edited by
KLAUS-UWE PANTHER
GÜNTER RADDEN
University of Hamburg
Introduction 1
Klaus-Uwe Panther and Günter Radden
1. Background
Eighteen years after Lakoff and Johnson’s (1980) seminal work on the role of
metaphor in conceptualization, which sparked a vast amount of research in
cognitive linguistics, it has become increasingly apparent that metonymy is a
cognitive phenomenon that may be even more fundamental than metaphor.
We believe that the contributions give a fair view of the state of the art in
metonymic research, although we are also aware of the fact that a great many
questions about metonymy still remain unanswered, some of which will be
addressed below.
The cognitive understanding of metaphor and metonymy is certainly at
variance with both naive and traditional scholarly views, which have strongly
been influenced by centuries of rhetorical and literary studies. The cleavage
between literal and figurative language, which was taken for granted by
traditional rhetoric and linguistics, has recently been challenged by Gibbs
(1994: 24–79; and this volume). Still, we owe the first basic insights into the
nature of tropes to Greek, Roman and medieval scholars, modern literary critics
and linguists. Many different classifications of tropes have been proposed,
starting with Aristotle, who subsumed metonymy and synecdoche under
metaphor, and more recently by the Groupe de Liège or Groupe µ, which
subsumed metaphor and metonymy under synecdoche (see Schofer and Rice
1977). Some of these ideas on metonymy definitely have a modern, cognitive
tinge. Various contributors to this volume (Koch; Blank; and Nerlich, Todd and
Clarke) link their cognitive approach to metonymy to this rhetorical tradition.
2 Klaus-Uwe Panther and Günter Radden
Three papers address the role of metonymy in language and thought from a
broader theoretical perspective. The issue of the conceptual nature of me-
tonymy is investigated by Günter Radden and Zoltán Kövecses in their
paper “Towards a theory of metonymy.” Their approach is based on the notion
of idealized cognitive model (ICM) as proposed by Lakoff (1987). Metonymy
is understood as a conceptual process in which one conceptual entity, the
‘target,’ is made mentally accessible by means of another conceptual entity,
the ‘vehicle,’ within the same ICM. In principle, either of the two conceptual
entities related may stand for the other, i.e., metonymy is basically a reversible
process. There are, however, a number of cognitive principles which govern
the selection of a preferred vehicle. These principles lead to natural, or
‘default’ cases of metonymy and often escape our awareness. These principles
Introduction 3
upon, but the whole event is conceptually involved. For example, in the action
schema the participants patient (fish), instrument (hook), and manner (pearl
fishing) are most frequently converted into new verbs (to fish, to hook, to fish
pearls, respectively). Five participant types are regularly exploited to yield new
verbs. In conclusion, Dirven raises the question if this selectivity is the result of
sociocultural saliency or rather a matter of linguistic preference.
Another semantic study, Christian Voßhagen’s paper “Opposition as a
metonymic principle,” focuses on antonymy as a metonymic relation. This
metonymy shows up in irony, where usually a positive concept metonymically
stands for a negative concept, and in some conventionalized lexical items such
as terribly in It was terribly amusing. As a rule, the metonymy applies to
evaluative concepts, which are semantically scalar but are reinterpreted as
complementary.
In his paper “Metonymic hierarchies: The conceptualization of stupidity
in German idiomatic expressions,” Kurt Feyaerts studies the metonymic
structure of everyday expressions of stupidity. For example, an expression
such as Du bist wohl nicht von hier? ‘You are not from here, are you?’
exemplifies the metonymic folk model OUTGROUP ORIGIN FOR STUPIDITY .
Feyaerts shows that metaphorically and metonymically organized hierarchies
have major structural characteristics in common. Higher-level metonymies
tend to be cross-culturally valid, while lower-level metonymies are more
culture-specific.
Klaus-Uwe Panther and Linda Thornburg emphasize the importance
of a cross-linguistic comparison of conceptual metonymies. In their paper “The
POTENTIALITY FOR ACTUALITY metonymy in English and Hungarian,” they
analyze the extent to which this metonymy is exploited across two genetically
unrelated languages, English and Hungarian. They explore its operation in
seven conceptual domains: sense perceptions, mental states and processes,
hedged performatives, indirect speech acts, (extralinguistic) actions, character
dispositions, and acquired skills. In some of these domains, the POTENTIALITY
FOR ACTUALITY metonymy is much more productive in English than in
Hungarian. The most striking contrast between the two languages emerges in
the domain of sense perceptions: whereas English systematically exploits the
metonymy in sentences such as I can taste the vanilla (for I taste the vanilla),
Hungarian systematically excludes the metonymy and resorts to a non-modal
construction in the indicative mood. The authors also discuss the relationship
between Gricean maxims, conversational implicatures and metonymy.
8 Klaus-Uwe Panther and Günter Radden
We assume that metonymy is not, as has often been taken for granted, merely
a matter of the substitution of linguistic expressions but a cognitive process
that evokes a conceptual frame. The notion of ‘conceptual frame’ is meant
here as a cover term for what is variously called ‘domain,’ ‘idealized cognitive
model’ (ICM), ‘schema,’ ‘scenario,’ ‘script,’ etc. in the cognitive-linguistic
literature (cf. also Blank, this volume; and Koch, this volume). The ‘substitu-
tion view’ of metonymy claims that the name of one thing is used in place of
that of another thing to which it is related. As will be shown below, this view
has serious draw-backs. Following Langacker (1993: 30), we assume that
“metonymy is basically a reference-point phenomenon [...] affording mental
access to the desired target.” Let us consider the conceptual frame of a
straightforward case of metonymy as exemplified in:
(1) The first violin has the flu.
The concept ‘the first violin’ is part of a knowledge structure that it evokes. As
a musical instrument, a violin is immediately associated with a violinist as the
player of that instrument. Moreover, the first violinist is defined as a member
of a larger group of musicians, the symphony orchestra. Among the musicians
of the orchestra, the first violinist is the most outstanding member. Finally, our
knowledge of orchestras includes, among other things, the notion of music and
its representation in scores. The predication has the flu as well as the attribute
first trigger a non-literal interpretation of the noun phrase the first violin. Thus,
the metonymic reading in (1) involves a shift from the instrument to the
musician as the most readily available element in the frame. Through this
metonymic shift, the reference point (‘the first violin’) is backgrounded and
the desired target (‘the first violinist’) is foregrounded. This conceptual shift is
reflected in grammatical form: thus the second sentence of (2a), in which she
anaphorically refers back to the target, is a felicitous continuation of (1),
whereas the second sentence of (2b), in which the pronoun is coreferential
with the reference point, is not:
10 Klaus-Uwe Panther and Günter Radden
(2) a. The first violin has the flu. She cannot practice today.
b. # The first violin has the flu. It is a Stradivarius.
But now consider a situation described by the following sentence:
(3) My ex-husband is parked on the upper deck.
The expression my ex-husband evokes a rich mental script involving mar-
riage, divorce, etc., all of which, however, do not seem to play a role in the
metonymic interpretation of this utterance. In contrast to (1), it is not the
conceptual frame of the noun phrase that is exploited for the metonymic
interpretation, but the predicate is parked on the upper deck. The predicate
helps identify the target of the metonymic shift, i.e., ‘my ex-husband’s ve-
hicle.’ Concomitantly, the metonymic reference point (‘my ex-husband’) is
foregrounded while the conceptual target (‘my ex-husband’s vehicle’) is
backgrounded. This analysis is corroborated by the linguistic fact that an
anaphoric pronoun cannot refer back to the target expression as in (4a), but
only to the reference-point expression as in (4b):
(4) a. # My ex-husband is parked on the upper deck. It has a California
license plate.
b. My ex-husband is parked on the upper deck. He is taking the
bus today.
On the basis of pronominal facts as in (4), Nunberg (1995: 111) claims that in
sentences such as (3) it is not the subject that is used metonymically but the
predicate, which “contributes a property of persons, the property they possess
in virtue of the locations of their cars.” This type of analysis thus postulates
that the metonymic shift is not achieved through the noun phrase but involves
a “predicate transfer” (for a critique of Nunberg’s theory cf. Kleiber 1995). It
is possibly more plausible and intuitively more satisfying, however, to view
the metonymy in (3) as an instance of referential shift, i.e., to understand my
ex-husband in the sense of ‘my ex-husband’s vehicle.’ We suggest that the
choice of the pronoun might be governed by a general cognitive principle
according to which humans take precedence over non-humans (see also
Radden and Kövecses, this volume). This principle would account for the fact
that the human entity in the frame seems to be foregrounded irrespective of
whether it is the reference point or the target.
A further point in need of clarification relates to the relationship among
the elements in the frame. In the case of an artifact as in example (1), the user
Introduction 11
of the artifact is so tightly integrated into the frame that the metonymic reading
has become lexicalized and is listed as a separate sense in dictionaries. In
contrast, with the exception of well-known individuals such as Shakespeare,
Mozart and Einstein, who are closely associated with their artistic or scientific
products, humans do not seem to be consistently tied to a frame which leads to
lexicalized metonymic senses. Thus it is highly unlikely that ex-husband
would have ‘car’ as one of its conventional senses. This will even hold for
human nouns such as car-dealer, which explicitly contains the concept of
‘car’ as an integral part of its frame.
We believe that both reference point and target are always present as
elements of the conceptual frame, but are highlighted to different degrees.
This can be shown by the following minimal pair which exemplifies two ways
of highlighting frame elements:
(5) a. The harpsichord has the flu. His part has been taken over by the
grand piano.
b. The harpsichord has the flu. Its part has been taken over by the
grand piano.
In contrast to (2a), in which only the human target can be foregrounded, the
sentences under (5) seem to allow the foregrounding of either the human
performer or the instrument. The possessive pronoun his in (5a) anaphorically
refers to the musician who is metonymically targeted by the harpsichord,
whereas the pronoun its in (5b) is grammatically congruent with the reference-
point expression, but conceptually relates to the part assigned to the harpsi-
chord in the score.
The car-parking situation described in (3), however, does not lend itself
to similar highlighting of either the reference point or the target as in (5). It is
much more difficult to foreground the target when the reference point is
human and the target is non-human. It seems, however, possible to say (6), in
which the anaphoric pronoun they highlights the cars and not their owners:
(6) ? Myex-husband and his girlfriend are parked next to each other.
They are both Fords.
In this sentence, the noun phrases my ex-husband and his girlfriend are
metonymically interpreted as ‘my ex-husband’s car’ and ‘his girlfriend’s car,’
respectively, i.e., there is a referential shift from HUMAN to NON - HUMAN .
The discussion thus far has looked at one area in which metonymic
12 Klaus-Uwe Panther and Günter Radden
Literally, sentence (8a) states the ability of the well-known basketball team to
win the championship, but metonymically this utterance implicates their
actually winning the game. The metonymy involved may be described as a
metonymic shift from POTENTIAL TO ACTUAL (see Panther and Thornburg,
this volume). Why should a sports commentator choose the metonymic predi-
cation were able to nail down rather than nailed down, which, after all, seems
to be the more economical wording of the two? Again, an explanation may be
provided within the framework of relevance theory. Sentence (8a) triggers
more contextual effects, i.e., pragmatic implications, than sentence (8b). Both
utterances convey the actuality of winning the championship, which is expli-
citly stated in (8b), but only conversationally implicated in (8a). Yet, the
metonymic wording in (8a) has the advantage of communicating additional
information: in stating ability, the predicate were able to strongly implicates
the notions of ‘effort,’ ‘difficulty,’ and ‘positive achievement,’ none of which
is present in (8b). The greater length of (8a) is thus more than compensated for
by the number of desirable contextual effects that it triggers.
As examples (7) and (8) demonstrate, a metonymic expression is hardly
ever completely equivalent in its pragmatic force to its ‘literal’ counterpart.
Thus, these data provide more evidence against the traditional ‘substitution
view’ of metonymy. In conclusion, this view of metonymy as a means of
providing maximal contextual effects with a minimum of processual effort
certainly opens new avenues of future research on the role of this as well as
other figurative modes of thought.
References
Nunberg, Geoffrey
1995 Transfers of meaning. Journal of Semantics 17: 109–132.
Schofer, Peter, Donald Rice
1977 Metaphor, metonymy, and synecdoche revis(it)ed. Semiotica 21: 121–147.
Sperber, Dan, Deirdre Wilson
1995 Relevance: Communication and Cognition. Second edition. Oxford and
Cambridge, MA: Blackwell.
Sweetser, Eve
1990 From Etymology to Pragmatics: Metaphorical and Cultural Aspects of
Semantic Structure. Cambridge: Cambridge University Press.
Part I
Günter Radden
University of Hamburg
Zoltán Kövecses
Eötvös Loránd University, Budapest
expectation: ?She is a pretty person but does not have a pretty face. The two
metonymies, THE FACE FOR THE PERSON and THE PERSON FOR THE FACE, thus
complement each other: A person’s face evokes the person and a person
evokes the person’s face. Metonymy does not simply substitute one entity for
another entity, but interrelates them to form a new, complex meaning.2 To use
Warren’s (this volume) example: “We do not refer to music in I like Mozart,
but to music composed by Mozart; we do not refer to water in The bathtub is
running over, but to the water in the bathtub.” Metonymic relationships should
therefore more adequately be represented by using an additive notation such
as X PLUS Y, as suggested by Radden (in print). For the sake of simplicity, we
will keep the traditional formula X FOR Y with the proviso, however, that the
metonymic process is not understood to be one of substitution.
The metonymic process consists in mentally accessing one conceptual
entity via another entity. This is the cognitive explanation Langacker (1993:
30) offers for metonymy. He conceives of metonymy as a reference-point
phenomenon in which one conceptual entity, the reference point, affords
mental access to another conceptual entity, the desired target.3 We will refer
to the reference-point entity as the ‘vehicle’ and the desired target simply as
the ‘target.’ In the example of She’s a pretty face, the ‘pretty face’ serves as
the vehicle for accessing the ‘person’ as the target; in the reverse description,
She’s a pretty person, the ‘person’ serves as the vehicle for accessing the
person’s ‘pretty face’ as the target. In either construal, both the vehicle and the
target are conceptually present. However, one of them is seen as being more
salient than the other and is therefore selected as the vehicle.
question concerns the nature of the relationship between the vehicle and one
or more targets. Metonymy tends to make use of stereotypical, or idealized,
relationships within an ICM. Thus, certain places tend to be associated with
events which typically occur at the place. For example, the expression to go to
bed may, depending on the situation, evoke the metonymic targets ‘to go to
sleep,’ ‘to have sex’ or ‘to be sick.’ All these events are stereotypically
associated with beds, irrespective of the particular context that triggers the
situationally relevant target. More generally, we may describe the conceptual
relationship between space and event as one that is entrenched and may be
exploited by metonymy. The question that needs to be answered here is what
types of conceptual relationships in an ICM may give rise to metonymy.
A third question pertains to the choice of vehicle and target. Thus far, we
have only considered conceptual relationships between two entities either of
which may become the vehicle or the target as shown in the examples of She’s
a pretty face and She’s a pretty person. Unlike metaphorical mappings, which
tend to be unidirectional, metonymic mappings are in principle reversible.
This was already implicitly noticed in traditional approaches by listing both
directions of a metonymic relationship such as CAUSE FOR EFFECT and EFFECT
FOR CAUSE, GENUS FOR SPECIES and SPECIES FOR GENUS, etc. Such theoretically
possible alternatives have to be distinguished from the speaker’s choice of a
particular vehicle as the ‘entry point’ into the ICM. We therefore need to ask if
there are any preferred metonymic construals and, if this is the case, what
‘cognitive principles’ govern the selection of one type of vehicle entity over
another. Such precedence principles were already hinted at by Langacker
(1993: 30). To the extent that there are such preferred routes, these will define
the unmarked, or ‘default,’ cases of metonymy.
Given that there are such default routes, we need to ask, as a fourth
question, if there are any principles that determine the choice of a vehicle other
than by default construal. If this is the case, these metonymic construals yield
marked, or ‘non-default,’ instances of metonymy.
The following four sections of this paper will be devoted to finding answers
to these central questions which, for convenience, are summarized below:
ii(i) What are the ontological realms in which ICMs and metonymic relation-
ships may occur? (Section 2);
i(ii) What are the types of conceptual relationships that may give rise to
metonymy? (Section 3);
(iii) Are there any conceptual entities that can better direct attention to an
intended target than others? If this is the case, are there any cognitive
Towards a Theory of Metonymy 23
The following three ontological realms are distinguished for the present
purpose: the world of ‘concepts,’ the world of ‘forms,’ in particular, forms of
language, and the world of ‘things’ and ‘events.’ These realms roughly
correspond to the three entities that comprise the well-known semiotic triangle
as developed by Ogden and Richards (1923: 11): thought, symbol and refer-
ent. The interrelations between entities of the same or from different ontologi-
cal realms lead to various ICMs and possibilities for metonymy.
An important distinction has to be made between ICMs which interrelate
entities of different ontological realms within the same semiotic unit and ICMs
which interrelate entities of different semiotic units within the same ontological
realm or realms. The former situation of interrelated ontological realms gives
rise to two ICMs: the pairing of a concept and a form establishes a sign and may
be described as ‘Sign ICM’; the pairing of a thing or event and a sign, form or
concept establishes a referential situation and may be described as ‘Reference
ICM.’ In as far as these ICMs lead to metonymy, the metonymies will be
described as ‘sign metonymy’ and ‘reference metonymy,’ respectively. The
latter situation of interrelated semiotic units involves concepts, typically in
conjunction with forms. These ICMs will be referred to as ‘Concept ICMs,’ and
a metonymy based on a Concept ICM will be described as ‘concept metonymy.’
Figure 1 illustrates the semiotic relationships which lead to the sign metonymy
(1) and three types of reference metonymies (2)–(4) on the one hand and one
type of concept metonymy (5) on the other hand. The arrows indicate the
direction of the metonymic mapping which will be discussed below.
The Sign ICM unites a form and one or more concepts. Thus, the word form
dollar or the dollar sign $ are intimately linked with the ‘currency denomina-
tion of dollar,’ ‘currency,’ or ‘money’ in general. As a rule, the form met-
onymically stands for the concept it denotes.
(1) FORM FOR CONCEPT: dollar for ‘money’
The very nature of language is based on this metonymic principle, which
Lakoff and Turner (1989: 108) describe as WORDS STAND FOR THE CONCEPTS
THEY EXPRESS. Since we have no other means of expressing and communicat-
ing our concepts than by using forms, language as well as other communica-
tion systems are of necessity metonymic. It is also for that reason that we fail
to notice the metonymic character of language. This general FORM FOR CON-
CEPT metonymy has to be distinguished from a FORM FOR CONCEPT metonymy
involving specific signs (see (47)).
The Sign ICM only seems to lead to the FORM FOR CONCEPT metonymy,
but not to the reverse metonymy of CONCEPT FOR FORM. A metonymic situation
of this kind is difficult to imagine. It may be said to occur in the ‘tip of the
tongue’ experience, in which we have a certain concept but cannot think of the
corresponding word-form, or in the context of foreign-language learning,
where learners must find a form in the foreign language for a concept with
which they are familiar.
tween these two worlds by comparing the extensional and intensional mean-
ings of the verb to punch. In the extensional world of reality, an event of
punching involves a series of subevents: folding one’s fist, moving one’s arm,
bringing it into contact with an object, and recoiling it. A punching event thus
has duration. Intensionally, however, to punch is a punctual verb and, as such,
cannot be used to describe a durational event as in ??It took five minutes to
punch him. The projected world we refer to by means of a sign thus only partly
matches reality.8 We do, however, firmly believe that words refer to the
extensional world so that metonymy (2) has psychological validity.
Signs can only be said to stand for things or events they refer to — the
reverse metonymic situation, THING/EVENT FOR FORM-CONCEPT, is hard to
imagine. A thing or event may evoke the word denoting it, but it certainly
cannot be argued that when we see a cow and the word cow comes to our
mind, the thing ‘cow’ metonymically stands for the sign cow.
The above folk theory of reference has two variants in which either the
concept or the form of a sign is focused upon. Lakoff (1987: 168f) describes
the former situation as “reference via meaning,” and the latter as “doctrine of
direct reference.”
According to the Reference-via-Meaning ICM, “words have inherent
meanings (called intensions) and designate objects by virtue of those mean-
ings” (Lakoff 1987: 168f). In this view, the meaning associated with the word
cow is assumed to stand for any cow in the world of reality — in contrast to the
set-theoretic account, in which ‘cow’ denotes the set/class of cows.
(3) CONCEPT FOR THING/EVENT: concept ‘cow’ for a real cow
The reversal of this metonymic relationship, THING/EVENT FOR CONCEPT, may
occur in special situations. One such situation would apply to people who are
associated with certain outstanding properties. For example, the former En-
glish soccer player Bobby Charlton may stand for the concept of sportsman-
ship. Another situation might be that of a game of chess in which a missing
piece is replaced by an object such as an eraser or a button. At first sight, this
situation might be seen as involving things in reality only, i.e., THINGA FOR
THINGB. This is, however, not the case. In assigning an eraser the function of,
say, a white rook, we have the thing stand for the concept of a rook.
The Direct-Reference ICM and the type of referential metonymy in-
volved most clearly apply to the use of proper names for persons of that name.
The name John Smith directly refers to the bearer of this name. In our folk
26 Günter Radden and Zoltán Kövecses
theory of language, the Direct-Reference ICM also has a much wider met-
onymic application. Stephen Tylor (1978:168) points out that in our common-
sense view of language words are names of things, not names of classes:
“There are words and the things we speak of. The word ‘cow’ is the name of
the object ‘cow.’ The word ‘cow’ stands for the object ‘cow’ as a substitute for
it, and we know the meaning of the word ‘cow’ because we have a prior
knowledge of the object it stands for.”
(4) FORM FOR THING/EVENT : word-form cow for a real cow
Also Ogden and Richards (1923) draw attention to people’s tendency to see an
inherent connection between words and referents: Words are seen as part of
the thing and may even provide power over reality in superstition, magic and
spells.
The reversal of this metonymic relationship is hardly possible. A situa-
tion that comes closest to the THING/EVENT FOR FORM metonymy might be that
of language learning, where a child or student points to an object and wants to
be given its name.
The metonymic relationship may also appear in its reverse form. For
example, the Control ICM may also give rise to the metonymy CONTROLLER
FOR CONTROLLED, as in I am parked over there.
used to stand for the latter. The metonymy PLACE FOR INSTITUTION thus
accounts for our understanding of The White House did not intervene in the
sense of ‘the American government did not intervene.’
Metonymic polysemies seem to take a preferred direction, but they are in
principle reversible. Thus, an institution may also be used to stand for a place.
For example, Let’s have an oyster dish at Central Station illustrates the
metonymy INSTITUTION FOR PLACE.
(8) FORMA-CONCEPTA FOR FORMB-CONCEPTA: UN for United Nations
In the metonymic situation in (8) the form of an expression changes while the
concept roughly remains the same, which might also be indicated by a
notation such as CONCEPTA/B. This metonymy applies to reductions of form as
in the abbreviation UN, acronyms such as NATO, clippings such as exam for
examination, modifications of form as in the euphemism What the heck are
you doing? for What the hell are you doing? and substitutions by pro-forms
and equivalence in translation, in which the language changes but the content
is intended to be preserved. In all these cases, the form of language changes
but, ideally, not its conceptual content.
The reversal of this metonymic relationship does not occur freely. In the
case of abbreviations, this would imply that we understand a full expression
such as United Nations as standing for its abbreviated form, UN. In the case of
translation equivalence, however, either of the languages involved may of
course serve as the vehicle or target.
2.4. Summary
one sign stands for another sign, is the only type of metonymy that is in
principle reversible; it is also the only type of metonymy that is characterized
by a variety of conceptual relationships between vehicle and target — the
other types of metonymic relationships carry very little additional content
apart from interrelating vehicle and target. Both these factors have certainly
contributed to focusing on this type of metonymy.
Third, the direction of metonymization allows us to recognize a hierarchy
for the preferred choice of an ontological vehicle. With sign metonymies (1),
the form serves to access a concept; with reference metonymies, the form-
concept unit (2), the concept (3) or the form (4) serves to access a thing or
event in reality; and with concept metonymy (6), the form-concept unit
accesses a concept. Three of the ontological metonymies, (5), (7) and (8), do
not display a preferred directionality because they operate at the same onto-
logical level. We therefore find the following hierarchy for the choice of a
metonymic vehicle:
form > form-concept > concept > reality
The cognitive motivation for these directionalities will be investigated in
Section 4.
Since this paper aims at providing an overall cognitive framework of
metonymy, which finds its richest variability in metonymies involving lan-
guage, its focus will mainly be on metonymies which most visibly, but not
exclusively, manifest themselves in the shape of linguistic expressions.
Conceptual relationships within an ICM which may give rise to metonymy will
be called ‘metonymy-producing relationships.’ Thus, the conceptual relation-
ship that holds between a container and the thing(s) contained may produce the
metonymies CONTAINER FOR CONTENTS and CONTENTS FOR CONTAINER. How-
ever, not all relationships within ICMs can produce metonymies. One example
of this is the ICM of the human face. A face has several parts that are closely
related in space, yet do not lead to metonymy. For example, the nose cannot
stand for the mouth metonymically and vice versa, i.e., I hit him in the nose will
not be understood to mean ‘I hit him in the mouth.’ On the other hand, we
understand the utterance Can you answer the door? in a metonymic sense, in in
30 Günter Radden and Zoltán Kövecses
which the door is meant to refer to a person standing behind the door. Both
examples involve a relationship of spatial proximity, but only the conceptual
distinctness between objects and people in the latter example leads to me-
tonymy. Metonymy may only arise when the intended target is uniquely
accessible and “the addressee’s attention is directed to the intended target”
(Langacker 1993: 30). The greater the conceptual contrast between vehicle and
target, the better is a relationship suited to be exploited metonymically. Thus,
a whole ICM is generally conceptually distinct enough from its parts to allow
metonymic processes to operate freely from whole to part or part to whole.
The distinction between whole and part is in fact of paramount impor-
tance for metonymic processes. Given that our knowledge about the world is
organized by structured ICMs which we perceive as wholes with parts, we
suggest that the types of metonymy-producing relationships may be subsumed
under two general conceptual configurations:
i(i) Whole ICM and its part(s)
(ii) Parts of an ICM
Configuration (i) may lead to metonymies in which we access a part of an ICM
via its whole or a whole ICM via one of its parts; configuration (ii) may lead to
metonymies in which we access a part via another part of an ICM. This, of
course, implies that the whole ICM is still present in the background.
The following typology of metonymy-producing relationships and me-
tonymies is not meant to be exhaustive. It includes those types that are most
frequently listed in classifications of metonymies and seem to reflect the most
entrenched metonymic routes.
The relationship between a whole and a part typically applies to things and
their parts, where the notion of ‘thing’ is to be understood here in the
schematic sense of Langacker (1991). The prominent status which whole-part
relationships have in our cognition has been demonstrated by Tversky and
Hemenway (1984), who found that, in tasks involving attributes of basic-level
items, subjects predominantly listed parts as attributes. This may have to do
with the conceptually autonomous status which we attach to things. Things, in
particular physical objects, are typically conceived of as forming a gestalt with
well-delineated boundaries and as internally composed of various parts.
Towards a Theory of Metonymy 31
‘Whole-part configurations’ are also assumed to underlie the Scale ICM, the
Constitution ICM, the Event ICM, the Category-and-Member ICM, the Cat-
egory-and-Property ICM, and the Reduction ICM.
(i) Thing-and-Part ICM. This ICM may lead to the two metonymic
variants:
(9) a. WHOLE THING FOR A PART OF THE THING:
America for ‘United States’
b. PART OF A THING FOR THE WHOLE THING:
England for ‘Great Britain’
People often speak of America but mean one of its geographical parts, the
United States; conversely, people, especially foreigners, often speak of
England but mean Great Britain, including Wales and Scotland.
The WHOLE FOR PART metonymy is widely found in situations which
Langacker (1993: 31) describes as “active-zone/profile discrepancies,” where
an entity’s active zone is defined as comprising “those portions of the entity
that participate most directly and crucially in that relationship.” For example,
in He hit me or The car needs washing, the profiled whole things he and the
car may be said to stand for the active-zone parts ‘his fist’ and ‘the car’s
body,’ respectively. Even clearly separate things may be conceptualized as
active-zone parts of an over-arching whole ICM. Thus, it does not strike us as
unnatural to speak of lighting the Christmas tree for ‘lighting the candles on
the Christmas tree.’ The use of a whole entity for an active-zone part comes so
natural that it usually does not appear to us, including most scholars of
metonymy, as metonymic.9
The PART FOR WHOLE metonymy has traditionally been given special
attention and classified as a type of its own under the name of ‘synecdoche.’10
Synecdoches are less ubiquitous than WHOLE FOR PART metonymies and,
hence, more likely to be noticed. This applies to deliberate metonymic usages
such as Those are cool wheels you have there as well as to the widespread use
of body parts such as hand, face, head or leg for a person.11 In such situations,
the entity that is understood to be most crucially involved in the ICM is
metonymically highlighted. Especially more abstract things such as ‘hearing,’
‘intellect’ or ‘control’ tend to be metonymically expressed by one of their
concrete parts: Thus, ear may stand for ‘hearing,’ brain for ‘intellect’ and
hand for ‘control’ as in Things got out of hand for ‘things got out of control.’
(ii) Scale ICM. Scales are a special class of things and the scalar units are
32 Günter Radden and Zoltán Kövecses
parts of them. Typically, a scale as a whole is used to stand for its upper end
and the upper end of a scale is used to stand for the scale as a whole:
(10) a. WHOLE SCALE FOR UPPER END OF THE SCALE:
Henry is speeding again for ‘Henry is going too fast’
b. UPPER END OF A SCALE FOR WHOLE SCALE:
How old are you? for ‘what is your age?’
The expression speed defines the whole scale of velocity but we locate the
velocity in (10a) at, or even beyond, the upper end of the scale. Conversely,
mention of the positive end of the scale in (10b) evokes the whole scale, and it
is only for the purpose of achieving special effects that the negative end of a
scale may be used as in How young are you?
(iii) Constitution ICM. This ICM involves matter, material or substances
which are seen as constituting a thing. Properly speaking, such ‘substance-
things’ do not have parts but are constituted by their very substance. Sub-
stances are, amongst other things, characterized as being unbounded and
uncountable.12 A substance or material may be metonymically conceived of
as an object and is then construed as bounded and coded as a count noun as in
(11a). Conversely, an object may be conceived of as substance-like and is then
construed as unbounded and coded as a mass noun as in (11b).
(11) a. OBJECT FOR MATERIAL CONSTITUTING THE OBJECT:
I smell skunk.
b. MATERIAL CONSTITUTING AN OBJECT FOR THE OBJECT:
wood for ‘forest’
(iv) Event ICM. Events may be metaphorically viewed as things which may
have parts. Thus, one can speak of the theoretical and the practical ‘parts’ of a
driving test. As with things, an event as a whole may stand for one of its
subevents and a subevent may stand for the whole event.
(12) a. WHOLE EVENT FOR SUBEVENT: Bill smoked marijuana.
b. SUBEVENT FOR WHOLE EVENT: Mary speaks Spanish.
instances of the metaphor GENERIC IS SPECIFIC. Since both the specific and the
generic levels belong to the same ICM, however, we prefer to analyze them as
instances of the metonymy SPECIFIC FOR GENERIC.
Subtypes of this metonymy are AN INDIVIDUAL (AS A TYPICAL MEMBER OF
A CATEGORY) FOR A CATEGORY as in every Tom, Dick and Harry and SPECIFIC
CASE FOR GENERAL RULE, which “holds between laws and their concrete
instantiations generally” (Norrick 1981: 37). This metonymic relationship is
not, as claimed by Norrick, irreversible. One can easily think of situations in
which a general rule is invoked in describing a specific situation as in I always
leave my umbrella at home when it rains.
(vi) Category-and-Property ICM. Properties may either be seen meta-
phorically as possessed objects (PROPERTIES ARE POSSESSIONS) or metonymi-
cally as parts of an object. If categories are intensionally defined by a set of
properties, these properties are necessarily part of the category. Categories
typically evoke, and metonymically stand for, one of their defining or other-
wise essential properties and, conversely, a defining or essential property may
evoke, or stand for, the category it defines (cf. Norrick 1981).
(18) a. CATEGORY FOR DEFINING PROPERTY: jerk for ‘stupidity’
b. DEFINING PROPERTY FOR CATEGORY: blacks for ‘black people’
Most of the members of a category which Lakoff (1987: Ch. 5) discusses in
connection with metonymic models are much rather characterizable as salient
properties. For example, an ideal husband is characterized by the property of
faithfulness as in Wilbert doesn’t go to topless bars; he is my husband, while a
stereotypical husband is, amongst other things, characterized by the property
of dullness as in Wilbert will stay at home; he is my husband. Some categories
conventionally stand for specific properties such as Judas for ‘treacherous’ or
Cadillac for ‘the best of.’ Also, in referring to an upcoming star in linguistics
as a second Chomsky, we have in mind his or her intellectual brilliance.
Stereotypical properties are also evoked in our interpretation of ‘collo-
quial tautologies’ such as Boys will be boys. Since a tautology is literally
uninformative, it can only be meaningfully interpreted in the sense of a
stereotypical property associated with the category. The tautology in Boys will
be boys may, depending on the context, mean ‘boys are unruly’ or ‘boys are
cute and adorable’ (Gibbs 1994: 345–351). All these examples are instances
of a WHOLE FOR PART metonymy, which may be characterized as in (19a);
conversely, salient properties as parts of a category may stand for the category
as a whole as in (19b):
36 Günter Radden and Zoltán Kövecses
ICM, the Control ICM, the Possession ICM, the Containment ICM, the
Location ICM, the Sign and Reference ICMs and the Modification ICM.
(i) Action ICM. Action ICMs involve a variety of Participants which may
be related to the predicate expressing the action or to each other. There are,
thus, specific relationships such as those between an INSTRUMENT and the
ACTION, the RESULT of an action and the ACTION , etc., all of which are parts of
the Action ICM. These relationships may in turn be instantiated as specific
types of metonymy. The Action ICM includes the following types of met-
onymic relationships, the first four of which are reversible:
(21) a. AGENT FOR ACTION: to author a new book; to butcher
the cow
b. ACTION FOR AGENT: writer; driver
(22) a. INSTRUMENT FOR ACTION: to ski; to hammer
b. ACTION FOR INSTRUMENT: pencil sharpener; screwdriver
(23) a. OBJECT FOR ACTION: to blanket the bed; to dust the room
b. ACTION FOR OBJECT: the best bites; the flight is waiting
to depart
(24) a. RESULT FOR ACTION: to landscape the garden
b. ACTION FOR RESULT: the production; the product
(25) MANNER FOR ACTION : to tiptoe into the room
(26) MEANS FOR ACTION: He sneezed the tissue off the
table.15
(27) TIME FOR ACTION: to summer in Paris
(28) DESTINATION FOR MOTION: to porch the newspaper
(29) INSTRUMENT FOR AGENT: the pen for ‘writer’
With the exception of (29), INSTRUMENT FOR AGENT, all the Action metonymies
listed above involve predicates either as the vehicle or the target and typically
also involve changes of their word class: Participants are converted into verbs
and predicates are nominalized. Noun-verb conversion and nominalization
can therefore be seen as two complementary morphological processes leading
to the two types of reversible metonymies. What makes these morphological
derivations special types of metonymy, however, is their conflation of vehicle
and target: the original word class describes the metonymic vehicle and the
morphologically recategorized form expresses the target. The conflation pro-
cess echoes the incorporation of elements in the verb described by Talmy
(1985). The situations analyzed by him, however, are not metonymic. The
38 Günter Radden and Zoltán Kövecses
as I am parked over there (for ‘my car’); You have a flat tire (for ‘your car’) or
Bill is in the Guinness Book of Records (for ‘his name’). This is also reflected
in the use of anaphoric pronouns, which, if they can be used at all, refer to the
human vehicle as in Bill is in the Guinness Book of Records; he is on page 7
and not to the target as in #Bill is in the Guinness Book of Records; it is on page
7. Conversely, anaphoric pronouns in POSSESSED FOR POSSESSOR metonymies
refer to the human target as in Many big names have turned up and he was one
of them and not to the vehicle as in #Many big names have turned up and it was
one of them. POSSESSED FOR POSSESSOR metonymies apply to both alienable
possession (The ham sandwich had a side dish of salad) and inalienable
possession (appendicitis for ‘patient’), and in both cases they are somehow
felt to be marked.
(vii) Containment ICM. The image-schematic situation of containment is
so basic and well-entrenched that it deserves to be treated as an ICM of its own
among locational relations. As a rule, we are more interested in the contents of
a container than in the mere container so that we commonly find metonymies
which target the contents via the container as in (43a) rather than the reverse
metonymic relationship as in (43b).
(43) a. CONTAINER FOR CONTENTS: The bottle is sour for ‘milk’
b. CONTENTS FOR CONTAINER:
The milk tipped over for ‘the milk container tipped over’
(Norrick 1981: 58)
(viii) Location ICMs. Places are often associated with people living there,16
well-known institutions located there, events which occur or occurred there
and goods produced or shipped from there (see (40)). Hence, we find the
following metonymies:
(44) a. PLACE FOR INHABITANTS:
The whole town showed up for ‘the people’
b. INHABITANTS FOR PLACE:
The French hosted the World Cup Soccer Games for ‘France’
(45) a. PLACE FOR INSTITUTION:
Cambridge won’t publish the book for ‘Cambridge Univer-
sity Press’
b. INSTITUTION FOR PLACE:
I live close to the University.
42 Günter Radden and Zoltán Kövecses
EGORY, also applies here. These members are, however, less central to the
category of swear words than genuine ‘four-letter words’ such as shit and fuck.
(x) Modification ICM. This ICM mainly applies to variant forms of a sign
apart from reduction. Reduced forms were accounted for by the PART FOR
WHOLE metonymy (20); variant forms stand in a Part-and-Part relationship of
an ICM. More specifically, we may distinguish between genuine cases of
modification as in (48) and substitution as in (49), both of which seem to be
unique to language and only operate in one direction:
(48) MODIFIED FORM FOR ORIGINAL FORM: effing for fucking
(49) SUBSTITUTE FORM FOR ORIGINAL FORM:
Do you still love me? — Yes, I do.
3.3. Summary
CONTENTS FOR CONTAINER, i.e., the choice of the things contained. Similarly,
metonymy (41a), CONTROLLER FOR CONTROLLED, is more natural than me-
tonymy (41b), CONTROLLED FOR CONTROLLER. Thus, Schwartzkopf defeated
Iraq evokes the metonymic reading in which Schwartzkopf stands for the US
Army that did the fighting. The US Army defeated Iraq, however, does not
evoke the controller reading. It is these metonymic preferences which will be
explored in the following section.
The former factor relates to communicative aspects and will be described here
in terms of communicative principles, the latter factor pertains to cognitive
aspects and will be described in terms of cognitive principles.
Section 4.1. will be concerned with cognitive principles of relative sa-
lience. Since metonymic reference points are cognitive in nature, it is only
natural to expect that these principles have a cognitive basis. Since metony-
mies also play a major role in communication, we may also expect communi-
cative principles to contribute in determining the selection of a vehicle
expression. Section 4.2. will briefly examine the issue of communicative
principles of relative salience. The principles themselves are assumed to have
the status of preferential tendencies and will be stated in the form of X OVER Y.
Some initial ideas concerning cognitive principles which determine the rela-
tive salience of an entity were developed by Langacker (1993: 30): “Other
things being equal, various principles of relative salience generally hold:
human > non-human; whole > part; concrete > abstract; visible > non-visible;
Towards a Theory of Metonymy 45
etc.” Some of the semantic constraints which Cooper and Ross (1975) found
to determine the fixed ordering of bipartite expressions are also relevant for
default metonymies.17 The cognitive principles which have been identified
here relate to three general determinants of conceptual organization: human
experience, perceptual selectivity, and cultural preference, which, however,
tend to interact and overlap in several ways.
BODILY OVER MENTAL (brain for ‘intellect’) and BODILY OVER PERCEPTUAL
(good ear for ‘good hearing’). Since concrete objects are visible and abstract
things invisible, the principle also entails VISIBLE OVER INVISIBLE, which is
reflected in metonymies such as to save one’s skin for ‘to save one’s life.’
Visibility, in its turn, accounts for the default metonymy (43a) CONTAINER FOR
CONTENTS, since things in a container are typically hidden inside the container
and, thus, invisible. At a more general level, the CONCRETE OVER ABSTRACT
principle also accounts for metonymies (1) FORM FOR CONCEPT, and (47)
WORDS FOR THE CONCEPTS THEY EXPRESS, in which the concrete visual or
acoustic shape of a sign stands for the abstract concept denoted by the
(word-)form.
(iv) INTERACTIONAL OVER NON-INTERACTIONAL. Our experience of the world
mainly derives from our interacting in it. Entities we interact with form better
reference points than entities with which we do not interact. We often interact
with parts of a whole so that this principle provides a default motivation for
PART FOR WHOLE metonymies. For example, the part we interact with in
driving is the steering wheel so that we speak of sitting behind the wheel. We
mainly use our hands in interacting with the world and hence speak of hand-
on demonstration, we use our fingers in typing on the keyboard and thus speak
of having the world at our fingertips when we log into the Internet. Our
interaction with things is closely related to their function.
(v) FUNCTIONAL OVER NON-FUNCTIONAL. As shown by Tversky and
Hemenway (1984), we attach particular salience to functional parts such as the
engine of a car. There is a vested human interest in seeing things, in particular
artifacts, from a functional point of view. Thus, the part of a TV-set that appears
to be the most functional is its tube so that we speak of spending the whole day
in front of the tube; the parts of a car that are functional to its driving are the
wheels, the motor and the steering wheel so that we speak of a 24-wheeler, a
motorway and sitting behind the wheel. Parts which are not important in driving
such as the doors, the windshield wipers, or the fenders are, of course, highly
unlikely to be selected as metonymic reference points for the car.
have clear boundaries and are specific instances. These foci of perceptual
selectivity can be stated in the following principles of cognitive preference.
(i) IMMEDIATE OVER NON-IMMEDIATE. This cognitive principle accounts for
selecting stimuli in our spatial, temporal and causal immediacy. The me-
tonymy in I’ll answer the phone for ‘I’ll answer the person speaking at the
other end of the line’ is motivated by spatial immediacy. Metonymies (13)
PRESENT FOR HABITUAL as in I always take the 9 o’clock train and (14) PRESENT
FOR FUTURE as in I am off for ‘I will be off’ are motivated by temporal
immediacy. Metonymy (33) EMOTION FOR CAUSE OF EMOTION, as in She is my
joy for ‘she makes me be happy’ is motivated by causal immediacy. Here, the
emotion as an effect of some stimulus serves as the metonymic vehicle. In
general, effects are more perceptible and affect us more than causes. The
immediacy principle also accounts for many emotion metonymies in which
physiological and behavioral responses produced by emotions are used to
stand for the emotions themselves as in He got cold feet for ‘he became
frightened’ (see Kövecses 1990).
(ii) OCCURRENT OVER NON-OCCURRENT. This principle reflects our preferen-
tial concern with real, factual, and occurrent experiences. It accounts for
metonymy (15a) ACTUAL FOR POTENTIAL in expressions such as He is an angry
person or This is a fast car. Here, the occurrent senses of the words angry and
fast as found in their predicative usages stand for their non-occurrent, poten-
tial senses.
(iii) MORE OVER LESS. More of something is usually more salient perceptu-
ally than less of something. This principle accounts for the naturalness of
using expressions denoting the upper, but not the lower, end of a scale for the
whole scale as in How tall are you?, where tall refers to any size. In the social
and political domains, size is related to power and dominance, which may be
seen as metaphorical sizes.
(iv) DOMINANT OVER LESS DOMINANT. This principle explains the metonymic
use of the biggest and most powerful country or part of a country for a larger
geographical unit as in (9b) England for ‘Great Britain,’ Holland for ‘the
Netherlands’ and Russia for the former ‘Soviet Union.’ This principle may
also account for the traditional acceptance of masculine forms in a generic
sense as in mankind, postman or you guys.
(v) GOOD GESTALT OVER POOR GESTALT. A powerful organizing principle for
our perception is the tendency to see whole gestalts rather than the sum of their
parts. The gestalt-perceptual principle also applies to the selection of a pre-
48 Günter Radden and Zoltán Kövecses
ferred vehicle in metonymy and accounts for the wide-spread use of humans
and whole objects for their active-zone parts. Metonymy (9a) WHOLE THING
FOR A PART OF THE THING as in The car needs washing is thus well-motivated.
Unlike the perceptual gestalt of things, the gestalt of events is exclusively
conceptual in nature. The borderlines of events tend to be much fuzzier than
those of things. A subevent of a larger event may no longer be understood as a
part but may gain the status of an independent event gestalt of its own. As a
result, the metonymy SUBEVENT FOR THE WHOLE EVENT as in I am going to the
doctor for ‘I am seeing a doctor for medical treatment’ is widespread and well-
motivated.
An essential requirement of any gestalt is that it has clearly delineated
boundaries. Consequently, the gestalt-perceptual principle entails as its
subprinciple:
(vi) BOUNDED OVER UNBOUNDED. The productive metonymy (11a) OBJECT
FOR MATERIAL CONSTITUTING THE OBJECT allows us to construe bounded things
as substances as in We had chicken today. Its reverse metonymy (11b)
MATERIAL CONSTITUTING AN OBJECT FOR THE OBJECT as in I sent you an e-mail is
much less productive.
(vii) SPECIFIC OVER GENERIC. This cognitive principle relates to our prefer-
ence for good gestalts as well as for concreteness, immediacy and occurrence.
Specific and definite instances form better gestalts than general and unspecific
entities. The cognitive principle underlies metonymy (17b) SPECIFIC FOR GE-
NERIC and its subtypes. At a purely conceptual level, this principle accounts for
people’s tendency to generalize. For example, O.J. Simpson’s verdict of ‘not
guilty’ was taken by many Americans as a verdict for all black people.
in a culture. Ideal cases are social constructs and defined with respect to
desirability. Some concepts are inherently defined as desirable by a culture
such as ‘ideal love’ (see Kövecses 1988), others are represented by a paragon
like Babe Ruth for ‘ideal baseball players’ (Lakoff 1987). Also, negative
categories may have ideal examples that can stand for the whole category.
When we say You are a Judas, we use an individual who is a betrayer par
excellence in our culture to stand for betrayers and betrayal in general.
(iii) TYPICAL OVER NON-TYPICAL. Typical members of a category are often
picked out when a category as a whole is described. For example, one may talk
about the symptoms of sneezing and coughing in referring to a cold as in
You’ve got a bad cough.
(iv) CENTRAL OVER PERIPHERAL. The cultural impact of centrality is nicely
illustrated in Feyaerts’ (this volume) study of the conceptualization of stupid-
ity in German. Expressions such as You are not from here, are you? demon-
strate that people who are characterized as stupid live on the periphery of a
linguistic culture. Here, our folk understanding of inferior mental abilities is
based on the metonymy OUTGROUP ORIGIN FOR STUPIDITY.
(v) INITIAL OR FINAL OVER MIDDLE. In our conception of events, an initial or
final phase may be seen as being more important than the central phase. To
pull the trigger for ‘to shoot’ focuses on an event’s initial phase, to sign a
contract for ‘to make a contract’ focuses on an event’s final phase. The
etymologies of creed and mass provide nice historical illustrations of the two
aspects of this principle: creed derives from the first word of the Apostles’
Creed, Credo in unum Deum ‘I believe in one God,’ while mass for ‘service’
goes back to a formula said at the end of medieval church services, Ite, missa
est (contio) ‘go now, the meeting is dismissed’ (Ullmann 1972: 219).
(vi) BASIC OVER NON-BASIC. This principle applies to simple and well-known
‘ground’ routines or affairs as in Lakoff’s (1987: 88f) generators and submodels
and in our preference for basic level categories. The use of expressions such as
I’ve told you a hundred times for ‘several times’ may be said to exemplify a
metonymy BASIC FOR NON-BASIC because a hundred is a basic number.
(vii) IMPORTANT OVER LESS IMPORTANT. This principle accounts for the use
of stage for ‘theater’ as the most important part of the Theater ICM, or the
expression speaking a language for ‘knowing a language’ or the identification
of a capital city with a country.
(viii) COMMON OVER LESS COMMON and
(ix) RARE OVER LESS RARE. Common members of a category are culturally
50 Günter Radden and Zoltán Kövecses
given reference points and may be used metonymically such as aspirin for any
pain-relieving tablet, while rare members stand out because of their unique-
ness as in Lakoff’s (1987) example of a DC-10 crash, which people general-
ized to the extent that they refused to fly any DC-10.
It is, without doubt, possible to identify more such cognitive principles
which, however, partly overlap with the ones discussed above. Among these
we would probably have to list UNEXPECTED OVER EXPECTED, NEW OVER OLD,
and TRADITIONAL OVER NON-TRADITIONAL.
In light of the previous section, it can be suggested that the more cognitive
principles apply, the greater the cognitive motivation of a metonymy. As a rule,
more than one cognitive principle does in fact apply to a particular case of
metonymy. For example, the metonymy ARTIST FOR HIS WORK as in We are
reading Shakespeare (for ‘Shakespeare’s plays’) is motivated by a bundle of
cognitive principles: HUMAN OVER NON-HUMAN, CONCRETE OVER ABSTRACT, and
GOOD GESTALT OVER POOR GESTALT. Most instances of metonymy, however, are
not ‘fully’ motivated in the sense that not all cognitive principles converge.
Rather, we have a continuum of motivation ranging from fully motivated default
metonymies to weakly or unmotivated non-default metonymies.
Consider again Lakoff and Johnson’s example The buses are on strike for
‘the bus drivers are on strike.’ Since passengers ‘interact’ with the buses and
buses are more relevant to them than their drivers, the metonymy is motivated
by the cognitive principle INTERACTIONAL OVER NON-INTERACTIONAL and the
communicative principle RELEVANT OVER IRRELEVANT, but it is inconsistent
with the cognitive principle HUMAN OVER NON-HUMAN. The metonymy in I’ll
answer the phone is consistent with the principle IMMEDIATE OVER NON-
IMMEDIATE, but is in conflict with the principle HUMAN OVER NON-HUMAN. The
metonymic expression paper for ‘essay on a subject’ is motivated by the
principle CONCRETE OVER ABSTRACT, in particular, VISIBLE OVER INVISIBLE, but,
since paper is prototypically a mass noun, the principle BOUNDED OVER
UNBOUNDED is reversed. In all these cases, conflicting motivations decrease
the naturalness of the overall motivation of the metonymy.
52 Günter Radden and Zoltán Kövecses
4.4. Summary
5. Overriding factors
which was replaced by bathroom and restroom, which in their turn were
supplanted by expressions such as facilities and comfort station. Violations of
the clarity principle also abound in officialese as in equal opportunity em-
ployer and other areas that are prone to using jargon.
These types of ‘vivid’ metonymy have traditionally been studied in
rhetoric and literary criticism. In the cognitivist view of metonymy presented
here they now appear as non-default cases, in which cognitive principles and/
or communicative principles are deliberately overridden. Since the primary
goal of this paper is to isolate the principles which determine default cases, the
issue of non-default metonymies shall not be explored any further.
6. Conclusion
Notes
1. This paper elaborates upon an earlier version which appeared in Cognitive Linguistics 9–
1 (1998). We are grateful to our anonymous reviewers for their insightful and helpful
comments. We also wish to express our thanks to Elizabeth Mathis and Gary Palmer for
their valuable linguistic contributions to this paper.
2. This view corresponds to Dirven’s (1993: 14) characterization of metonymy as opposed
to metaphor: “[...] in metonymy the two domains both remain intact, but they are seen to
be in line, whereas in metaphor only one domain, viz. the target domain is kept, and the
other domain, viz. the source domain disappears, so to speak.”
3. The attraction of Langacker’s cognitive explanation of metonymy lies in the pervasive-
ness of reference-point phenomena in language structure, in particular possessive expres-
sions. The view of metonymy as a reference-point phenomenon is, however, not
unproblematic. The process of first making mental contact to a reference point before
accessing the target should take longer than that of accessing a conceptual entity directly,
which, however, has not been confirmed experimentally in terms of processing time
(Gibbs 1993).
4. See the discussion of the notion of ‘contiguity’ in Koch (this volume). The notion of
contiguity is also present in cognitive definitions as in Croft’s (1993: 347) definition of
metonymy as “a shift of a word meaning from the entity it stands for to a ‘contiguous’
entity.”
5. For similar views see Lakoff and Johnson (1980), Lakoff (1987), Croft (1993) and Blank
(this volume).
6. The word hearse derives from Latin (h)irpicem ‘large rake used as a harrow’ and Samnite
(h)irpus ‘wolf,’ i.e., the pins of the harrow were metaphorically related to the wolf’s teeth.
7. In distinguishing metaphor from metonymy, Lakoff and Johnson (1980: 36) and Lakoff
and Turner (1989: 103) view metonymy as primarily having a referential function, as
opposed to the function of understanding associated with metaphor. But this view of
metonymy is at odds with the many other areas in which metonymy is found.
8. Frawley (1992: 19–20) illustrates the difference between the extensional world and the
mentally projected world by the well-known case of tautologies as in George Bush is
George Bush. This sentence is meaningfully interpretable in an attitudinal sense, which
may be made explicit by attitude predicates such as wonder, doubt, be uncertain, etc. as in
I wonder if George Bush is George Bush. Gibbs (1994: 345–351) has shown that such
tautologies are metonymic in that a category is used to refer to specific salient parts or
attributes. This analysis is also adopted here (cf. 3.1.(vi)).
9. Cf. Langacker (1993: 31): “it [the active-zone/profile discrepancy] has to be regarded as
natural and expected rather than pathological.” Similarly, Herskovits (1985: 363) points
out in her discussion of the salience principle that the metonymic shift to a salient part
“has come to be so natural that it is hard to think that it involves any special process such
as metonymy.” Langacker (1991: 191) even argues that “It is in fact quite difficult to find
convincing examples [...] where all aspects of the designated entity participate equally in
a relationship.”
Towards a Theory of Metonymy 57
10. Cf. the careful distinction between metonymy and synecdoche made by Seto (this
volume).
11. On metonymies relating to the hand, see Kövecses and Szabó (1996).
12. See Langacker (1991: Ch. 3). Prototypical substances are characterized by internal
homogeneity and divisibility, prototypical things are characterized by internal heteroge-
neity and integration of its parts into a whole.
13. For a more general distinction between metonymies based on co-presence and succes-
sion, see Blank (this volume).
14. The metonymy ACTUAL FOR POTENTIAL also accounts for the development of the possibil-
ity sense of the modals can and may. The epistemic sense of can may have developed
from the original sense of ‘know’ via ‘knowledge how to do X’ and ‘being able to act’ to
‘increased possibility of acting.’ In a similar chain of sense development, the original
sense of may ‘might, power’ may have led to the modal sense of possibility. Studies on
grammaticalization have convincingly explained the steps involved in these semantic
changes by implicature and pragmatic strengthening.
15. Cf. Lakoff in the discussion on “Semantic Accommodation” in Cogling, May 7, 1994.
16. Objects and animals may, of course, also be associated with a place. A nice example of
metonymic association is the proper name Canary Islands, which goes back to the name
Canaria given to it by the Romans on account of the many dogs seen there and which
later on provided the name for the bird canary, which the Spanish found on the islands.
17. The following semantic constraints identified by Cooper and Ross (1975) correspond to
the cognitive principles as used here: Here and Now correspond to IMMEDIATE OVER NON-
IMMEDIATE, Singular corresponds to SPECIFIC OVER GENERIC, Animate and Agentive corre-
spond to HUMAN OVER NON-HUMAN, and Count corresponds to BOUNDED OVER UNBOUNDED.
Possibly also the remaining semantic constraints are relevant for metonymy.
18. Violation of the communicative principle CLEAR OVER OBSCURE amounts to the same thing
as flouting Grice’s maxim of manner.
19. Cf. Morgan (1978: 263), who analyzes this example, which goes back to Robin Lakoff, as
conventionalized conversational implicature.
References
Dirven, René
1993 Metonymy and metaphor: different mental strategies of conceptualization.
Leuvense Bijdragen 82: 1–28.
Frawley, William
1992 Linguistic Semantics. Hillsdale, NJ: Lawrence Erlbaum.
Gibbs, Raymond W., Jr.
1993 Processes and products in making sense of tropes. In A. Ortony (ed.),
Metaphor and Thought. Cambridge: Cambridge University Press. Second
edition. 252–276.
1994 The Poetics of Mind: Figurative Thought, Language and Understanding.
Cambridge: Cambridge University Press.
Grice, H. Paul
1975 Logic and conversation. In P. Cole, J.L. Morgan (eds.), Syntax and Seman-
tics. Vol. 3: Speech Acts. New York: Academic Press, 41–58.
Herskovits, Annette
1985 Semantics and pragmatics of locative expressions. Cognitive Science 9:
341–378.
Jongen, René
1985 Polysemy, tropes and cognition or the non-Magrittian art of closing curtains
whilst opening them. In W. Paprotté, R. Dirven (eds.), The Ubiquity of
Metaphor (Current Issues in Linguistic Theory 29). Amsterdam and Phila-
delphia: Benjamins, 121–139.
Kövecses, Zoltán
1988 The Language of Love. Lewisburgh: Associated University Press.
1990 Emotion Concepts. New York: Springer-Verlag.
Kövecses, Zoltán, Péter Szabó
1996 Idioms: A view from cognitive semantics. Applied Linguistics 17: 326–355.
Kövecses, Zoltán, Günter Radden
1998 Metonymy: Developing a cognitive linguistic view. Cognitive Linguistics
9: 37–77.
Lakoff, George
1987 Women, Fire and Dangerous Things: What Categories Reveal about the
Mind. Chicago: The University of Chicago Press.
Lakoff, George, Mark Johnson
1980 Metaphors We Live By. Chicago: The University of Chicago Press.
Lakoff, George, Mark Turner
1989 More than Cool Reason: A Field Guide to Poetic Metaphor. Chicago: The
University of Chicago Press.
Langacker, Ronald W.
1991 Foundations of Cognitive Grammar. Vol. II: Descriptive Application.
Stanford, California: Stanford University Press.
1993 Reference-point constructions. Cognitive Linguistics 4: 1–38.
Morgan, Jerry L.
1978 Two types of convention in indirect speech acts. In P. Cole (ed.), Syntax
and Semantics. Vol. 9: Pragmatics. San Diego: Academic Press, 261–280.
Towards a Theory of Metonymy 59
Norrick, Neal R.
1981 Semiotic Principles in Semantic Theory. Amsterdam and Philadelphia:
Benjamins.
Norvig, Peter, George Lakoff
1987 Taking: a study in lexical network theory. Proceedings of the Thirteenth
Annual Meeting of the Berkeley Linguistics Society. Berkeley, California:
Berkeley Linguistics Society, 195–206.
Ogden, Charles A., Ivor A. Richards
1923 The Meaning of Meaning: A Study on the Influence of Language upon
Thought and of the Science of Symbolism. London: Routledge and Kegan
Paul.
Radden, Günter
in print How metonymic are metaphors? In A. Barcelona (ed.), Metaphor and
Metonymy at the Crossroads. Berlin, New York: Mouton de Gruyter.
Talmy, Leonard
1985 Lexicalization patterns: semantic structure in lexical forms. In T. Shopen
(ed.), Language Typology and Syntactic Description. Vol 3. Cambridge:
Cambridge University Press, 36–149.
Taylor, John R.
1995 Linguistic Categorization: Prototypes in Linguistic Theory. Oxford:
Clarendon Press. Second edition.
Thornburg, Linda, Klaus-Uwe Panther
1997 Speech act metonymies. In W.-A. Liebert et al. (eds.), Discourse and
Perspective in Cognitive Linguistics. Amsterdam and Philadelphia:
Benjamins, 205–219.
Tversky, Barbara, Kathleen Hemenway
1984 Objects, parts and categories. Journal of Experimental Psychology 113:
169–193.
Tylor, Stephen A.
1978 The Said and the Unsaid: Mind, Meaning and Culture. New York: Aca-
demic Press.
Ullmann, Stephen
1972 Semantics: An Introduction to the Science of Meaning. Oxford: Blackwell.
Webster’s Third New International Dictionary of the English Language
1971 Edited by Philip B. Gove. Chicago et al.: Merriam Co.
Speaking and Thinking with Metonymy
1. Introduction
The impulse to speak and think with metonymy is a significant part of our
everyday experience. Traditionally viewed as just one of many tropes, and
clearly subservient in most scholars’ minds to the master trope of metaphor,
metonymy shapes the way we think and speak of ordinary events and is the
basis for many symbolic comparisons in art and literature. Consider, as one
example, the French novelist Honoré de Balzac. Like many 19th century
fiction writers, Balzac provides wonderful examples of metonymic descrip-
tions in which the concrete depiction of some object or person stands for or
represents larger objects or domains of experience. For instance, in his novel
Père Goriot, Balzac opens with a long description of a boarding-house owned
by Madame Vauquer (Balzac 1951: 33):
Madame Vauquer is at home in its stuffy air, she can breathe without being
sickened by it. Her face, fresh with the chill freshness of the first frosty
autumn day, her wrinkled eyes, her expression, varying from the conven-
tional set smile of the ballet-dancer to the sour frown of the discounter of bills,
her whole person, in short, provides a clue to the boarding-house, just as the
boarding house implies the existence of such a person as she is.
metaphorical, while The third baseman is like a glove does not, and thus is
metonymic.
Metonymy is related to synecdoche in that both tropes exploit the rela-
tionship of larger entities and lesser ones. Synecdoche substitutes the part for
the whole as in
(3) They’re taking on new hands down at the factory.
where the term hands stands for men. Metonymy also substitutes the token for
the type or a specific instance, property, or characteristic for the general
principle or function. For instance, in
(4) They prefer the bullet to the ballot box.
the term bullet represents armed conflict, while ballot box refers to peaceful
democratic processes.
Some linguistic expressions involve both synecdoche and metonymy.
For instance, consider
(5) General Schwartzkopf had 400,000 fatigues at his command.
Here fatigues, the specific type of uniform worn by common soldiers, refers to
soldiers, reflecting a part-whole relationship, or to warlike power, reflecting a
token-for-type relationship. Moreover, many linguistic statements reflect both
metaphor and metonymy (cf. Goossens 1990, this volume). A significant
challenge for scholars is to discover some of the complex ways that metonymy
interacts with other tropes, both within individual statements and across
longer stretches of discourse.
Wit, hands, and heart are metonyms that stand for the familiar tripartite
division of a human into mind, body, and soul. Like realistic novelists or
biographers, poets, such as Shakespeare, rely heavily on synecdochic detail to
evoke scene, characters, and cultural experience. The poet Philip Larkin, to
take another example, evokes the past glories of race horses in the following
stanza from “At Grass” (Larkin, 1988).
Silks at the start: against the sky
Numbers and parasols: outside,
Squadrons of empty cars, and heat,
And littered grass: then the long cry
Hanging unhushed till it subside
To stop-press columns on the street.
principle by which a place may stand for an institution located at that place.
Thus, a place like Wall Street stands for the particularly salient institutions
located at that place, namely the stock exchange and major banks. Various
metonymic models in our conceptual system underlie the use of many kinds of
figurative and conventional expressions (e.g., OBJECT USED FOR USER, CON-
TROLLER FOR CONTROLLED, THE PLACE FOR THE EVENT, and THE PLACE FOR AN
INSTITUTION LOCATED AT THAT PLACE).
Finally, there are thousands of ordinary verbs that are based on me-
tonymy. Consider the main verbs in the following sentences (cf. Dirven, this
volume):
(10) The librarian shelved the books.
(11) The scientists eyeballed the data.
(12) The maid dusted the table.
(13) The pilot dusted the crops.
(14) He boycotted the story.
(15) We can’t stomach any more articles on metonymy.
(16) He toed the line.
People experience little problem interpreting these expressions, despite the
fact that, for example, one could, in principle shelve books by hitting books
with shelves. But our deep background knowledge of the typical relationship
that shelves have with books allows us to figure out exactly which salient act
one does when shelving books, eyeballing data, and so on.
In other cases, we can create novel verbs based on our knowledge of
particular actions associated with specific people, called ‘eponymous verbs.’
Consider the case of a friend of mine who noted about another person we both
knew
(17) He’s going to OJ his way out of his marriage.
I immediately understood the rather complex meaning of this expression by
drawing a metonymic inference about the salient acts associated with OJ (for
the famous American football player and murder suspect O.J. Simpson),
namely that one can murder one’s wife to get out of the marriage (a comment
that my friend intended sarcastically about the other person we knew). Again,
our ability to think metonymically ‘on the fly’ enables us to make immediate
sense of these novel phrases.
66 Raymond W. Gibbs, Jr.
Traveling from one place to another involves a series of actions where people
find some vehicle to take them to the desired location, get into the vehicle, ride
in it to the destination, arrive and get out. An idealized cognitive model (ICM)
of this series of events includes the following (Lakoff 1987):
Preconditions: You have access to the vehicle.
Embarkation: You get into the vehicle and start it up.
Center: You drive (row, fly, etc.) to your destination.
Finish: You park and get out.
End point: You are at your destination.
It is conventional to use one part of this idealized model to evoke the entire
model. Thus, people can simply mention either the Precondition, Embarka-
tion, or Center to stand for the entire series of events that make up the travel
scenario. In the above brief exchange, speaker B mentions a Precondition (i.e.,
getting access to a taxi by hailing one) to represent the entire travel scenario.
Other possible responses that might work equally well specify other parts of
the idealized model, such as:
(19) I drove my car. (Center)
(20) I called my friend. (Precondition)
(21) I hopped on a bus. (Embarkation)
(22) I stuck out my thumb. (Embarkation)
By metonymically mentioning a subpart of the travel scenario to stand for the
whole scenario, speakers get listeners to draw the right inference about what is
meant. It is interesting, and significant, to note that many cases of conversa-
tional implicature are understood via metonymic reasoning. Consider the
following brief exchange from Grice (1975: 51):
(23) A: Smith doesn’t seem to have a girlfriend these days.
B: He has been paying a lot of visits to New York lately.
Grice argued with this example that what B said only expresses part of what he
meant by his utterance. Successful interpretation of B’s remark demands that
the listener make this inference about what the speaker meant. Grice called
this kind of inference a conversational implicature. Thus, although B simply
stated a fact about Smith’s recent visits to New York, B likely intended for A
to understand that Smith has, or may have, a girlfriend in New York. Follow-
ing Grice, we can say that B implicates the proposition just mentioned by
68 Raymond W. Gibbs, Jr.
virtue of what is said along with various background knowledge and beliefs
shared with A, including maxims of conversation that participants in talk-
exchange are mutually expected to observe (Grice 1975).
In many instances, our ability to draw implicatures requires us to see how
a speaker’s utterance metonymically refers to a whole organized sequence of
activities. Thus, suggesting one part of a likely scenario (e.g., men often go to
other places to be with their girlfriends) activates a whole scenario and implies
other unstated parts (e.g., that Smith actually has a girlfriend these days).
Different psychological research, unknowingly, provides good evidence
that people reason metonymically when understanding language. Consider the
following pair of utterances (Gernsbacher 1991):
(24) I need to call the garage (where my car was being serviced).
They said they’d have it ready by five o’clock.
Note that there is a plural pronoun in the second sentence, but not in the first.
But antecedents of pronouns must agree in person, number, and case. None-
theless, Gernsbacher (1991) has shown that people rate as more natural and
are faster to understand the above pair of sentences with ‘conceptual
anaphors’ than they do pairs of sentences with appropriate singular pronouns.
This is so because the singular entity mentioned (garage) metonymically
stands for some conceptual set (the people working at the garage). Plural
pronouns are natural and easy to understand precisely because of our ability to
think metonymically about people, places, events, and objects. We see that the
mention of the subpart metonymically stands for the whole event frame.
Finally, our ability to conceptualize of people, objects, and events in
metonymic terms provides the basis for much of the way we reason and make
inferences during text processing. Many studies show that people metonymi-
cally infer entire sequences of actions having only read some salient subpart in
a story. Consider the following simple tale.
(25) John was hungry and went into a restaurant.
He ordered lobster from the waiter.
It took a long time to prepare.
Because of this he only put down a small tip when he left.
When people hear this brief episode, they presumably activate their knowl-
edge of the activities normally associated with eating in a restaurant and use
this information to fill in the gaps to make the story coherent. This type of
Speaking and Thinking with Metonymy 69
I have described some of the forms that metonymy takes in language and
suggested that people understand many kinds of language, and not just me-
tonymy per se, because of their ability to think metonymically about people,
70 Raymond W. Gibbs, Jr.
objects, and events. Let us now consider in a bit more detail how it is that
people understand metonymic language.
Consider the following metonymy (Nunberg 1979):
(26) The ham sandwich is getting impatient for his check.
Understanding this expression requires that a process of ‘sense creation’ must
operate to supplement ordinary ‘sense selection.’ One model, following the
Gricean view, assumes utterances like (26) violate maxims of truthfulness.
This ‘error recovery model’ assumes that sense creation is initiated only after
conventional meaning has been found to be in error. On the other hand, a
‘concurrent processing model’ claims that sense creation and sense selection
operate simultaneously, perhaps in competition with each other.
To test these models, Gerrig (1989) had participants read stories ending
with statements like The horse race was the most popular event. In conven-
tional context, this phrase referred to a standard race between horses. In
innovative context, this phrase referred to a unique situation where snails
competed in a race that was the length of King Louis’s horse. Readers took
roughly the same time to read this statement in both kinds of context. Thus, the
error recovery model seems incorrect. Instead, readers seem to be creating and
selecting metonymic meanings at the same time (i.e., the concurrent process-
ing model). So, understanding metonymy does not require that listeners
realize that this expression violates specific maxims of conversation.
The ability to quickly employ both sense selection and sense creation
processes is nicely illustrated in the following passage. One of the best
examples of metonymy I know of comes from the American satirist Erma
Bombeck, who once wrote in one of her newspaper columns about her
daughter’s difficulties finding a suitable roommate (Clark 1983). Consider
what Bombeck says as she quotes her daughter:
We thought we were onto a steam iron yesterday, but we were too late. Steam
irons never have any trouble finding roommates. She could pick her own pad
and not even have to share a bathroom. Stereos are a dime a dozen.
Everyone’s got their own systems. We’ve just had a streak of bad luck. First,
our Mr. Coffee flunked out of school and went back home. When we replaced
her, our electric typewriter got married and split, and we got stuck with a girl
who said she was getting a leather coat, but she just said that to get the room.
It seems odd, literally speaking, to talk about steam irons having trouble
finding roommates or electric typewriters getting married. Traditional theo-
Speaking and Thinking with Metonymy 71
ries of parsing will fail to handle many of these phrases even though we, for
the most part, can easily understand what Bombeck’s daughter is saying
(Clark 1983). Consider the sentence Steam irons never have any trouble
finding roommates. Most parsers will search their lexicons for the sense of
steam irons that is intended, namely ‘a person who owns a steam iron,’ and
will fail to find anything like this meaning. But the fact that we think about and
talk about people in terms of steam irons, stereos, Mr. Coffee and so on,
reflects the common metonymic mapping whereby we use a salient aspect of
an object or event to stand for the thing or event as a whole.
Clark and Gerrig (1983) have proposed that many ‘contextual expres-
sions’ such as Our electric typewriter got married and split are understood via
the following goal hierarchy:
1. Bombeck wants readers to recognize that she is using electric typewriter
to denote an object used for typing.
2. Bombeck wants readers to recognize that her assertion about getting
married is the kind of thing that she has good reason to believe that on this
occasion we can readily compute uniquely on the basis of our common
ground such that this kind of thing has something to do with electric
typewriters.
3. Bombeck wants readers to recognize that she is using electric typewriter
to denote the individual (one of her daughter’s roommates) who owns an
electric typewriter.
By inferring the lowest subgoal, (1), from the fact that the speaker is using the
noun electric typewriter, we are to infer the next subgoal up, (2), from the fact
that electric typewriter is being used in connection with the verb married.
Finally, we can infer the highest subgoal, (3), from our understanding of (2).
Key among this list of goals is the assumption that the listener/reader will
interpret the speaker’s current utterance given what the speaker and listener at
that moment mutually know (called their ‘common ground’).
Only by considering our common ground with Bombeck can we recognize
when electric typewriter is to be construed as having a ‘stands-for’ relationship
in which certain people, places, events, and things may stand for other people,
places, events, and things. Many of the inferences required to understand what
is meant in this passage are fundamentally metonymic in that each of these
contextual expressions requires readers to understand that the object mentioned
(e.g., steam irons, stereos, Mr. Coffee, and electric typewriter) stands for the
72 Raymond W. Gibbs, Jr.
people who own these items (i.e., PEOPLE FOR THEIR POSSESSIONS). Many
contextual expressions, but not all, will be readily understood when interpreted
in light of conventional metonymic mappings such as OBJECTS USED FOR THEIR
USERS, PEOPLE FOR THEIR POSSESSIONS, THE PLACE FOR AN INSTITUTION LOCATED
AT THAT PLACE, or PRODUCER FOR THE PRODUCT. These ‘stands-for’ relationships
reflect pre-existing patterns of metonymic thought that substantially constrain,
in many cases, the kinds of inferences listeners are likely to draw to make sense
of what speakers say. Some psycholinguistic research has shown that people can
determine without great difficulty the appropriate referents for metonymic
expressions in discourse (Gibbs 1990).
Metonymic reasoning, where people infer wholes from parts and parts from
wholes, is also important in acts of reference where speakers make requests of
listeners (cf. Panther and Thornburg, this volume). Consider the following
types of requests:
(27) Can you shut the door?
(28) I’d like the door shut.
(29) Would you mind shutting the door?
(30) How about shutting the door?
(31) It’s getting cold in here.
Indirect speech acts arise when speakers plan a social transaction in which the
speaker exchanges something with the appointed addressee for the desired
information or action. To do this, the speaker must first assess what reasons there
may be for the addressee not to give the desired information. The speaker will
then formulate an utterance to deal with the greatest potential obstacles. By
picking out salient obstacles, even ones that are more apparent than real,
speakers assume that listeners can metonymically infer the entire sequence of
actions that must occur for the transaction of goods to be completed.
Research from both naturalistic and laboratory studies has shown that
people formulate their requests to best specify the main potential obstacles for
addresses (Francik and Clark 1985; Gibbs 1986). Thus, in a situation in which
the ability of the addressee to lend the speaker a blue sweater is salient,
speakers prefer the expression Can you possibly lend me your blue sweater?
Speaking and Thinking with Metonymy 73
over the request Would you mind lending me your blue sweater? Reading-time
experiments show that readers are faster to comprehend indirect requests that
specify addressees’ projected obstacles (Gibbs 1986). Thus, what makes some
indirect requests ‘conventional’ is the appropriateness of the sentence form in
matching the obstacles present for addressees in a social context. These
studies emphasize the importance of metonymic reasoning in people’s use and
understanding of indirect speech acts.
Another version of metonymy that has become quite colloquial for speak-
ers is to refer to aspects of people, objects, and events through tautological
statements. Consider the following brief exchange between two parents
(Gibbs and McCarrell 1990). A mother asks her husband Did the children ever
clean up their rooms? The father shakes his head and responds Well, boys will
be boys. At first glance, the father’s response to his wife’s question seems
nonsensical. The phrase Boys will be boys is true by virtue of its logical form
alone (as a nominal tautology) and, superficially, contributes no new informa-
tion to the conversation. But the utterance Boys will be boys is readily
interpretable and most listeners would agree that the father intended to convey
a particular meaning, something like ‘boys will be unruly and it is often
difficult to get them to do what you want.’ Nominal tautologies are found with
surprising frequency in everyday speech, literature (e.g., Gertrude Stein’s
famous line A rose is a rose is a rose), and advertising (e.g., Motor oil is motor
oil). These expressions are metonymic because the speaker refers to a general
category (e.g., boys) to refer to specific salient parts or attributes of that
category (e.g., unruly behavior) (Gibbs and McCarrell 1990). Interpreting
colloquial tautologies requires metonymic reasoning. This ability to infer parts
from whole also underlies the interpretation of expressions like
(32) The New York Times is late for the President’s press conference.
where the speaker means that a reporter representing the New York Times is
late for the news conference.
7. Conclusion
My main argument in this paper has been that metonymy is a significant part
of how people ordinarily think and speak. Although certain individuals, such
as great poets and writers, often create spectacular poetic examples of me-
74 Raymond W. Gibbs, Jr.
tonymy, virtually all people are quite familiar with many conventional forms
of metonymy in both language and thought. I have suggested that several
sources of evidence, ranging from work in linguistics to psycholinguistics,
highlight the idea that people experience little difficulty thinking of, speaking
of, and understanding metonymic language. At the same time, evidence from
cognitive psychology and psycholinguistics raises the possibility that people
also reason metonymically (i.e., inferring wholes from parts and parts from
wholes) to solve problems, and draw conversational inferences about lan-
guage that, strictly speaking, is not metonymic. The proper study of me-
tonymy surely extends beyond looking at metonymic language alone. In this
sense, the approach I advocate in this chapter extends beyond looking at
metonymy as a lexical phenomenon, and seeks to discover the ways that
patterns of metonymy in language reflect patterns of metonymic thought.
More generally, my review of research from cognitive psychology suggests
that different forms of metonymic thinking can be found in experimental
situations which did not set out to study metonymy at all. I believe that
psychologists, and others, would benefit greatly from examining a wide
variety of research findings in light of possible links to metonymic thought.
The enthusiasm exhibited for metonymy in this article, and in the contri-
butions of this volume, is encouraging and should greatly boast metonymy’s
reputation from its present status as a secondary trope below metaphor. I must,
nonetheless, raise a cautionary note. Linguists, philosophers, literary theorists,
and psychologists, like myself, must be careful not to assume that a direct link
exists between metonymy in language and metonymy in thought. People may,
for instance, comprehend conventional metonymic language without neces-
sarily drawing metonymic mappings given our experience with many familiar
forms of metonymy. Additional psycholinguistic studies must be conducted to
determine how and when conceptual metonymies are inferred when under-
standing both metonymic and non-metonymic language. Furthermore, me-
tonymy scholars must be careful not to assume that particular mental
processes must operate for metonymy to be successfully understood. Thus,
many scholars have incorrectly conjectured that metonymies can be under-
stood only after some violation of a Gricean maxim has been recognized, an
idea that is not consistent with the available psycholinguistic data. Most
generally, as cognitive scientists from a variety of disciplines move forward to
further pursue the topic of metonymy, we must acknowledge the limitations of
our respective methodologies in drawing conclusions about the role that
Speaking and Thinking with Metonymy 75
References
1990 Why boys will be boys and girls will be girls. Journal of Psycholinguistic
Research 19: 125–145.
Gibbs, Raymond W., Jr., Y. Tenney
1980 The concept of scripts in understanding stories. Journal of Psycholinguistic
Research 9: 275–284.
Goossens, Louis
1990 Metaphtonymy: the interaction of metaphor and metonymy in expressions
for linguistic action. Cognitive Linguistics 1: 323–340.
Graesser, Arthur C., S. Woll, D. Kowalski, D. Smith
1980 Memory for typical and atypical actions in scripted activities. Journal of
Experimental Psychology: Human Learning and Memory 6: 503–515.
Grice, H. Paul
1975 Logic and conversation. In P. Cole, J. Morgan (eds.), Syntax and Semantics,
Vol. 3: Speech Acts. New York: Academic Press, 41–58.
Jakobson, Roman
1971 Two types of language and two types of aphasic disturbances. In R.
Jakobson (ed.), Selected Writings, Vol. 2. Cambridge, Massachusetts: The
MIT Press, 239–259.
Lakoff, George
1987 Women, Fire, and Dangerous Things: What Categories Reveal about the
Mind. Chicago: The University of Chicago Press.
Lakoff, George, Mark Johnson
1980 Metaphors We Live By. Chicago: The University of Chicago Press.
Larkin, Philip
1988 Philip Larkin: Collected Poems. Edited with an introduction by A. Thwaite.
London: Faber.
Levin, Samuel R.
1977 The Semantics of Metaphor. Baltimore: John Hopkins University Press.
Lodge, David
1977 The Modes of Modern Writing. London: Arnold.
1981 Modernism, antimodernism, postmodernism. In D. Lodge, Working with
Structuralism. London: Routledge and Kegan Paul, 3–16.
Nunberg, Geoffrey D.
1979 The non-uniqueness of semantic solutions: polysemy. Linguistics and Phi-
losophy 3: 143–184.
Rosch, Eleanor
1978 Principles of categorization. In E. Rosch, B. Lloyd (eds.), Cognition and
Categorization. Hillsdale, NJ: Erlbaum, 28–49.
Schank, Roger C., Robert P. Abelson
1977 Scripts, Plans, Goals, and Understanding: An Inquiry into Human Knowl-
edge Structures. Hillsdale, NJ: Erlbaum.
Searle, John
1979 Metaphor. In A. Ortony (ed.), Metaphor and Thought. New York: Cam-
bridge University Press, 92–123.
Metonymy and Conceptual Integration
Gilles Fauconnier
University of California, San Diego
Mark Turner
University of Maryland
1. Conceptual integration
Table 1. Conceptual correspondences in the ANGER IS HEAT metaphor and in the metonymy
linking emotions to physiological effects
SOURCE TARGET
‘physical events’ ‘emotions’ ‘physiology’
container person person
heat anger body heat
steam sign of anger perspiration, redness
explode show extreme anger acute shaking, loss of
physiological control
boiling point highest degree of emotion
If the blend stood by itself, it could not be interpreted in the real world
because anger does not produce smoke or explosion. But in the network
model, the blend remains linked to the inputs. A sentence like He was so mad
I could see the smoke coming out of his ears directly identifies the blend, not
the inputs, but inferences in the blend can be projected back to the target input
spaces. For example, in the blend, we infer from the smoke’s coming out of his
ears that he is extremely angry and showing physiological signs of it, because
in the blend, smoke is a physiological sign of great anger. We project that
inference back to the target input space: he is extremely angry and showing
physiological signs of it. (We do not project back to the target input space the
specific nature of the physiological signs in the blend, where, e.g., physiology
includes emitting smoke.) Of course, the structure of the blend itself is highly
dependent on the conventional metaphorical mapping of heat to anger.
In addition, we find an explanation for the actual grammatical structure of
Metonymy and Conceptual Integration 83
an expression like He exploded; I could see the smoke coming out of his ears.
Specifically, we see how vocabulary that picks out apparently incompatible
conceptual elements from quite different domains (smoke should not go with
coming out of his ears since smoke comes from fire but people cannot have
fire inside them) can be used in a clause to evoke a ‘single integrated scene.’
That single integrated scene is available in the blend, even though it is
unavailable from the source or the target. The blend provides a frame (seeing
somebody in an abnormal and dangerous state, with corresponding emotions,
etc.) not available in the source or target.
Next, the blend can have a life of its own, not fully determined by the
inputs. So, we can say, with some hyperbole:
(2) God, was he ever mad. I could see the smoke coming out of his ears
— I thought his hat would catch fire!
It is easy to see how this works: in the blend, the hat on fire is a sign of
even greater heat, hence even greater anger, emotions, etc. But there is no
counterpart for the hat in the source: the elaboration is in the blend, where the
frame of somebody on fire or at least very hot is used (not the boiling kettle
anymore), and the existing mapping operates toward the source (greater heat)
and toward the target (greater anger, but also greater loss of control, greater
social danger, etc.).
The analysis by Kövecses (1986) and Lakoff (1987) underscores the
essential role of physiological reaction metonymies in the formation of the
metaphorical system for emotions. The metonymic correspondences are in the
target — body heat, redness, etc. These map directly onto the blend, in the
sense that in the blend (but not in the target), the physiological reactions are
smoke, explosion, etc. This is done by mapping hot (in the target, for people,
with a certain physiology) to hot in the source (for containers with quite
different physical properties), and then from source to blend, where the new
set of physiological reactions is constructed.
Web: Manipulating the blend as a unit must maintain the web of appropri-
ate connections to the input spaces easily and without additional surveil-
lance or computation.
Unpacking: It is optimal for the blend alone to allow reconstruction of the
inputs, the cross-space mapping, the generic space, and the network of
connections between all these spaces.
Topology: For any input space and any element in that space projected
into the blend, it is optimal for the relations of the element in the blend to
match the relations of its counterpart.
Good reason: All things being equal, if an element appears in the blend,
there will be pressure to find significance for this element. Significance
will include relevant links to other spaces and relevant functions in
running the blend.
There is a last optimality constraint that is the crucial one in a discussion
of metonymy. It concerns the projection of metonymic links from inputs to the
blend:
Metonymy projection constraint: When an element is projected from an
input to the blend and a second element from that input is projected
because of its metonymic link to the first, shorten the metonymic distance
between them in the blend.
We saw above that blending can combine non-counterpart elements from
a single input, such as death, the cowl of the priest, and the skeleton of the
person who has died. The metonymic distance is large between abstract death
as the general cause of all deaths and the cowl worn by a certain kind of
participant in a ritual associated with particular deaths. But in the blend, the
metonymic connection is direct: the cowl is the attire of death. Similarly, the
skeleton that remains after the corpse has decomposed is a distant product of
death. But in the blend, the skeleton is actually a body part of death, indeed the
body part that provides the form of death. The fact that metonymy is preserved
at all in such blends can be viewed as a consequence of ‘topology’: a
metonymic relation in the input corresponds to a metonymic relation in the
blend. The ‘metonymy projection constraint’ additionally specifies that me-
tonymies get tighter under projection.
Satisfying the metonymy projection constraint is not a matter of blindly
projecting metonymic links. The internal integration of the blend provides
opportunities for some acceptable metonymies but not for others. Since death
86 Gilles Fauconnier and Mark Turner
is an active person in the blend, and active persons are known to have
skeletons (although they are not normally visible), the PART FOR WHOLE
metonymy connecting a body to its skeleton becomes available as the counter-
part of the distant metonymy in the input. Tightening metonymies as they are
projected from input to blend can help optimize integration in the blend by
creating a tighter, more easily manipulated unit.
Now consider some additional cases that show how metonymy projection
operates. Consider a cartoon illustration of a report that a powerful newspaper
company has succeeded in its hostile takeover of a weaker automobile com-
pany and will now sell off the automobile company’s assets, thereby eliminat-
ing the company. The cartoon shows a giant screw-type printing press
smashing a car. Obviously, the printing press represents the newspaper com-
pany, the car represents the automobile company, and the smashing of the car
by the printing press represents the elimination of the automobile company by
the newspaper company. This cartoon is a metaphorical blend: input one has a
stronger object and a weaker object; input two has the contest between the two
companies. The cross-space mapping is the basic metaphor that maps stronger
objects destroying weaker objects to winning and losing. The strong heavy
object is mapped onto the powerful newspaper company; the weaker object is
mapped onto the weaker automobile company. But in the blend, we find the
printing press as the strong heavy object and the car as the weak object. This is
an efficient exploitation of metonymic connections: the printing press is a
salient instrument of producing newspapers and cars are the salient products
manufactured by automobile companies. In the input with the companies, the
printing press is not an instrument of destruction, but it does have a force-
dynamic function that can be associated with a car-smashing machine of the
sort used in recycling automobiles. In the blend, the printing press is fused
with both the company and the car-smashing machine.
What is going on here? The blend must achieve three goals. First, given
that the cartoon is a visual representation, the blend must be concrete and
specific. Second, it must fit the frame that has the stronger object and the
weaker object. Third, these objects in the blend must be properly connected to
the companies in input two. The companies in input two, being abstract,
cannot in themselves provide the corresponding concrete elements in the
blend. The weaker and stronger objects in input one are concrete but not
specific, and so cannot in themselves provide the corresponding specific
elements in the blend. But we can exploit internal connections in the inputs to
Metonymy and Conceptual Integration 87
make the elements in the blend adequate. The printing press and the car are
concrete, specific objects associated metonymically with the companies, and
they can be fit into the frame of the stronger object destroying the weaker
object. They fit this frame in part because a printing press is already framed as
having force-dynamic structure that would make it capable of destruction
even though it is not intended for destruction, and in part because we are
familiar with car-smashing machines. In the blend, the two elements are
simultaneously (1) the two concrete, specific objects; (2) a stronger object
destroying a weaker object; and (3) two companies.
Clearly, such a blend is creative. Not just any connections will do. There
has to be a search for elements that simultaneously satisfy a number of
constraints. The printing press and car have topology in the blend (namely, the
printing press crushes objects and the car is the patient of an event of crushing)
that their counterparts in input two do not have (the press in input two is an
instrument of making newspapers, not of crushing, and the car is a salient
product of the automobile company, not a patient of an event of crushing).
Additionally, the printing press and car in input two have no counterparts in
input one. Interestingly, the printing press and the car, which are the elements
that did not project their input-topology to the blend, end up being the only
objects in the blend. (That is, in the input with the companies, the printing
press prints and the car is a working, undamaged product; but this topology
does not project to the blend; in the blend the printing press does not print, but
rather crushes, and the car is not a working, undamaged product, but a crushed
object.) The cartoon is remarkable because it is a case where integration and
topology are maximized by recruiting special metonymic connections in input
two. Because the topologies of strong and weak object on the one hand and
competing companies on the other will match only at a very abstract level, we
find that in addition to the companies, objects metonymically connected to
them are projected to the blend in a way that closely matches and elaborates
the topology in input one of strong and weak objects.
This example emphasizes that conceptual projection is a dynamic process
that cannot be adequately represented by a static drawing. Once the concep-
tual projection is achieved, it may look as if the printing press has always
corresponded to the stronger object and the car to the weaker. But in the cross-
space mapping, the printing press and the car play no role; they have no
counterparts in input one. Rather, the cross-space counterparts are stronger
object and newspaper company, weaker object and automobile company.
88 Gilles Fauconnier and Mark Turner
Under metonymy projection from input two, the printing press in the blend
becomes the counterpart of the stronger object in input one, and the car in the
blend becomes the counterpart of the weaker object in input one.
This example also shows that identity is metonymy of zero distance. The
metonymic relation in input one between company and commercial product is
transformed into identity in the blend, where the printing press is identically
both a printing press and the newspaper company to which it is, in the input,
metonymically related as an instrument.
Suppose the cartoon now contained the newspaper magnate operating the
printing press to smash the car, which is being driven by the car magnate. Here
the blend structure becomes elaborate through the recruitment to the blend of
an additional adversaries-with-instruments frame in which adversaries fight
with opposing instruments, and in which the winning adversary has the
superior instrument. Now the printing press and car in input two have counter-
parts in the adversaries-with-instruments frame: in input two, the printing
press is a symbol of a capacity for productivity that is an instrument of
corporate competition, and the car is a product that is an instrument of
corporate competition; these instruments in input two are the counterparts of
the instruments in the adversaries-with-instruments frame. Now, the topology
of opposing instruments in the blend matches the topology of opposing
instruments in the adversaries-with-instruments frame. This frame has the
useful property of aligning superiority of instrument with superiority of adver-
sary. In this case, we see that exploiting special internal connections in input
two makes it possible to recruit a frame that makes topology much stronger in
the blend structure.
Given the structural and dynamic mechanisms of blending and the opti-
mality constraints that guide it, metonymy plays an important role in con-
structing conceptual integration networks. Therefore, the various basic
cognitive phenomena that are varieties of conceptual integration — e.g.,
framing, provisional category extension, analogy, metaphor, the construction
of counterfactual spaces, aspects of grammar — will interact systematically
with metonymy according to uniform principles.
Note
References
Coulson, Seana
1996 The Menendez Brothers virus: analogical mapping in blended spaces. In A.
Goldberg (ed.), Conceptual Structure, Discourse, and Language. Stanford:
Center for the Study of Language and Information, 67–81.
1997 Semantic Leaps: The Role of Frame-shifting and Conceptual Blending in
Meaning Construction. Ph.D. dissertation, UC San Diego.
Fauconnier, Gilles
1997 Mappings in Thought and Language. Cambridge: Cambridge University
Press.
Fauconnier, Gilles, Mark Turner
1994 Conceptual projection and middle spaces. UCSD Cognitive Science Techni-
cal Report 9401. San Diego. [Available from http://cogsci.ucsd.edu and
from http://www.wam.umd.edu/ ~mturn]
1996 Blending as a central process of grammar. In A. Goldberg (ed.), Discourse
and cognition: Bridging the Gap. Stanford: Center for the Study of Lan-
guage and Information, 113–130.
1998a Conceptual integration networks. Cognitive Science 22/2: 133–187.
1998b Principles of conceptual integration. In J.-P. Koenig (ed.), Discourse and
Cognition: Bridging the Gap, Vol. II. Stanford: Center for the Study of
Language and Information, 269–283
Kövecses, Zoltán
1986 Metaphors of Anger, Pride and Love: A Lexical Approach to the Structure
of Concepts. Amsterdam and Philadelphia: Benjamins.
Lakoff, George
1987 Women, Fire, and Dangerous Things: What Categories Reveal about the
Mind. Chicago: The University of Chicago Press.
90 Gilles Fauconnier and Mark Turner
Turner, Mark
1996 The Literary Mind. New York: Oxford University Press.
Turner, Mark, Gilles Fauconnier
1995 Conceptual integration and formal expression. Metaphor and Symbolic
Activity 10/3: 183–203.
1998 Conceptual integration in counterfactuals. In J.-P. Koenig (ed.), Discourse
and Cognition: Bridging the Gap, Vol. II. Stanford: Center for the Study of
Language and Information, 285–296.
in print A mechanism of creativity. Poetics Today.
Distinguishing Metonymy from Synecdoche
Ken-ichi Seto
Osaka City University
1. Introduction
Several terms in (1) and (2) need explanation: among others, ‘E-related’ or ‘E-
relation,’ ‘contiguity,’ and ‘entity.’ In addition to these, there are two more
terms that are often used in defining synecdoche and metonymy: ‘whole’ and
‘part.’ Both these terms, which will inevitably come into play with the above
notions, also need to be precisely defined because their ambiguity has often
caused confusion and misunderstanding in the discussion of metonymy and
synecdoche. Not only this, but the ambiguity has contributed to the general
supposition that synecdoche is a (mere) subtype of metonymy. It will be
argued in this paper that synecdoche should be independent of metonymy.
Traditionally, synecdoche is defined as a relation in which a part stands for a
whole or a whole stands for a part and a genus for a species or a species for a
genus. I will reserve the notion of ‘synecdoche’ to a C(ategory)–related
transfer, while the notion ‘metonymy’ will be applied to an E(ntity)–related
transfer. Synecdoche, as used in this paper, will be defined as:
(3) Synecdoche is a conceptual transfer phenomenon based on the
semantic inclusion between a more comprehensive and a less com-
prehensive category.
This definition of synecdoche is abbreviated as:
(4) Synecdoche is a C(ategory)–related transfer.
Section 2 argues that the distinction between partonomy and taxonomy
provides a firm basis for the disambiguation of ‘whole’ and ‘part,’ and,
ultimately, for the all-important distinction between E-relation and C-relation.
Section 3 concentrates on the notion of ‘contiguity,’ the most important, but
ambiguous, term used to characterize metonymy. Section 4 defines the notion
of ‘entity,’ and identifies three different types: spatial, temporal, and abstract.
Section 5 describes various types and subtypes of metonymy according to the
three kinds of entities and other norms. Finally, Section 6, after discussing
synecdoche briefly, concludes that synecdoche, which is C-related, should be
independent of metonymy, which is E-related.
tree body
fir arm
The next step is to extend the notion of metonymy, since the referential
transfer based on partonomy is just one kind of metonymy. Therefore it is
necessary to clarify the sense of ‘contiguity’ introduced in (1) because par-
tonomy is just one of several contiguous relations. Again, the term ‘contigu-
ity’ has been used in a number of different ways. The best way to proceed is to
delimit the notion of contiguity to ‘spatio-temporal contiguity between two
entities in the world.’3 Contiguity in this sense covers a wider range of
relations than partonomy. For instance, what is the relation between a kettle
and the water in it? The water is not part of the kettle, but is just in contact with
it. Yet, the kettle can refer to ‘the water’ in it in sentence (6):
(6) The kettle is boiling.
This example shows that contiguity is a notion wider in range than the whole-
part relation.
96 Ken-ichi Seto
4. What is an entity?
5. Kinds of metonymy
The above discussion has shown that there are different kinds of metonymy,
depending not only on the kinds of entities (spatial, temporal, and abstract),
but also on the types of reference, i.e., the way one entity refers to another
(whole-part, container-contents, process-result, etc.). Figure 3 is a simplified
diagram of major types of metonymy classified in terms of E-relations.
metonymy
(E-relation)
In what follows, we shall see how each major type organizes its subtypes and
how they are related with one another.
The spatial metonymy provides a model for other types of metonymy. It has
two major subtypes: the whole-part and the container-contents type. The
whole-part subtype is represented by the relation between a physical object
and its components such as the body and its parts. The typical example of the
container-contents subtype is seen in the relation between a kettle and the
water in it.
bers are minimum units of the structure. By a minimum unit it is meant that a
constituent member belongs to an organization, not as a person per se, but,
rather, as a role unit or a substitutable part. This reasoning may have support
from the sense of substitute as in:
(15) Coming on as a substitute, he scored four crucial goals for
Cameroon.
A substitute has the same function as spare parts of, say, a motor because both
are potential parts of a whole, one for a team, the other for a machine.
Although a substitute is a metaphorical part and a motor part is non-meta-
phorical, the question is not whether a given part is metaphorical or real, but, if
metaphorical, what it is compared to and how it is related with the thing to
which it is compared. The group-member relation is a metaphorical whole-
part, hence, an E-relation. Therefore, the entity-transfer based on the group-
member relation or, more generally, the organization-member relation, is
metonymy.
Still, there may be a certain complication, which brings us to the second
of the Cruse’s three categories: classes. Cruse (1986: 179) refers to cases
where, he supposes, “[a]ny taxonomy can be thought of in part-whole terms
[...] a class can be looked on as a whole whose parts are its sub-classes.” This
is a dangerous step that might jeopardize the very concepts of, and the basic
distinction between, taxonomy and partonomy. It has already been seen that
the congruency of Figure 1 and Figure 2 is an illusion. But it may still be worth
mentioning the difference between organizations and classes, building on
insight gained from the two paragraphs above, so as not to fall victim to
superficial similarities in the figures. Figure 4 shows an organization (a
European-style university) with subclasses superimposed on its members at
the same time.
The column represents an organization: a (specific) linguistics depart-
ment. The department, which itself is usually a part of a larger organization
such as a (specific) faculty, may consist of one professor, three lecturers, and
two assistants. They are all parts of the linguistics department, but at the same
time each of them belongs to one of the three academic ranks: professor,
lecturer, and assistant. These layers represent categories. The professor of the
linguistics department is an element of the category ‘professor.’ The category
‘professor,’ on the other hand, may belong to a supercategory ‘university
teacher.’ It follows, then, that the particular professor has a double function:
102 Ken-ichi Seto
professor
lecturer
assistant
university teacher
strictly speaking, not part of the girl who wears it (see example (7a)), it must
be classified otherwise, together with the following examples:
(24) a. The red cap who was on duty last night got drunk.
b. He looked at his wrist. “I’d better get back to work.”
The relative pronoun who clearly indicates that the red cap refers to a person
in (24a); in (24b) his wrist ought to refer to ‘his wristwatch.’ These examples
may suggest the possibility of setting up a new category of metonymy,
something like the possessor-possessed or, perhaps more generally, the CON-
TROLLER-CONTROLLED type, as some scholars in fact do (Lakoff and Johnson
1980; Bredin 1984; Radden and Kövecses, this volume). I shall not pursue this
issue here.
Instead, let us briefly consider the implications that the following sen-
tence, perhaps the most famous example of metonymy (Nunberg 1978: 22),
may have in our context:
(25) The ham sandwich is getting restless at table 20.
In a restaurant situation, one waiter might say (25) to another, intending to
refer to a customer who ordered a ham sandwich. What is, then, the relation
between the customer and the ham sandwich? Is this spatial? In a sense it is; in
another it is not. A ham sandwich means primarily something to eat, and its
relation with the customer may be spatial even before it is brought to the table.
However, a ham sandwich is also a name (form) that is conventionally
associated with the referent so named (this association may be, again, met-
onymic as claimed by Radden and Kövecses, this volume), and the name may
be functioning in (25) as a kind of label stuck on the customer who ordered the
dish. In the latter interpretation, the ham sandwich in (25) can be a property, in
fact, the most salient property in the restaurant situation, that characterizes the
customer in question most effectively to a server. Perhaps this is one of the
reasons that prompt some scholars to adopt broader definitions of metonymy.
Thus Taylor (1995: 123f) suggests that
the essence of metonymy resides in the possibility of establishing connections
between entities which co-occur within a given conceptual structure [and
notes that the] entities need not be contiguous, in any spatial sense. Neither is
metonymy restricted to the act of reference. On this broader view, metonymy
turns out to be one of the most fundamental processes of meaning extension,
more basic, perhaps, even than metaphor.
106 Ken-ichi Seto
Temporal metonymy divides into two categories: one is based on the relation
between a whole event and a subevent, and the other on the relation between a
preceding and an ensuing situation.
(v)
(vi)
As the arrows in Figure 5 show, there are six different routes available for a
metonymic transfer: (i) PRECEDING SITUATION TO PROCESS, (ii) PROCESS TO
PRECEDING SITUATION, (iii) PROCESS TO ENSUING SITUATION, (iv) ENSUING SITUA-
TION TO PROCESS, (v) PRECEDING SITUATION TO ENSUING SITUATION, and (vi)
ENSUING SITUATION TO PRECEDING SITUATION. Among them, the transfers from
process to the preceding situation and from process to the ensuing situation are
particularly frequent as is shown in what follows. The preceding situation
means whatever precedes the process, and the ensuing situation means what-
ever comes out through the process, especially products and (new) states. As
to the preceding situation, four major factors need to be recognized: (a)
material, (b) agent, (c) instrument, and (d) place. Thus the prototypical preced-
ing situation involves an agent who (intentionally) does something some-
where, using some material and some instrument. Keeping all this in mind, let
us now turn to specific transfer patterns of temporal metonymy.
(i) PRECEDING SITUATION TO PROCESS. This transfer pattern is illustrated by
the sentences in (32):
(32) a. He took off the uniform at last.
b. Mrs. Djiak spooned coffee into the pot.
Taking off the uniform in (32a) could metonymically mean ‘retiring from an
organization’: it describes an event preceding that of retiring. Likewise, to
open one’s purse could mean ‘to spend money.’ Spoon in (32b) is a denominal
verb. Clark and Clark (1979) examine denominal verbs and note that the
instrumental verb, of which spoon is an example, is the most frequent and
productive type, followed by locatum, location, goal, and agent verbs. These
verbs can all be (re)interpreted as metonymic because, for example, spoon,
Distinguishing Metonymy from Synecdoche 109
which is an instrument as a noun, means as a verb a process (‘to put coffee into
the pot’) in which the instrument is involved (cf. Dirven, this volume).
(ii) PROCESS TO PRECEDING SITUATION. This transfer pattern typically in-
volves deverbal nouns such as supplies, allurements, toothpick and walk in the
following sentences:10
(33) a. The only grocery stores are holes in the wall that also sell beer,
liquor, and school supplies. (material)
b. There are a lot of allurements in big cities. (agent)
c. The bartender chewed on a toothpick. (instrument)
d. The Milford Track is the finest walk in the world. (place)
Also derived nouns such as invitation belong to this group:
(34) Have you sent out the invitations yet?
Invitations refers to ‘invitation cards’ (instrument) that are used in inviting
people (process). Another subtype of (ii) is the case of washing (process)
referring to ‘clothes that need to be washed’ as in (10a). The basic semantic
function of -ing is assumed to indicate processes taken holistically (cf.
Langacker 1991: 26).
(iii) PROCESS TO ENSUING SITUATION. This type is also productive:
(35) a. I carried the sheaf of printouts to a crowded table in the
periodicals room.
b. Anyone who came to work with a three-day growth, shoulder-
length hair, and a dirty sport jacket hanging over jeans would
be instantly slung into a black hole.
Printouts is a conversion from the corresponding verb, and growth is a derived
noun; both mean the outcome of a particular process. Another subtype of the
PROCESS TO RESULT type is characterized by the -ing-form; unlike the -ing
subtype in (ii), it works the opposite way, from process to result:
(36) My furnishings are Spartan gleanings from police auctions and
resale shops.
If we add more common examples like painting, lettering, clipping, etc. to this
class, it will expand quickly.
(iv) ENSUING SITUATION TO PROCESS. Regular examples of this class are
found in denominal verbs such as to cash, to cripple, to group, etc. They refer
110 Ken-ichi Seto
metonymy are also easy to find: the pen as in live by the pen or The pen is
mightier than the sword, and the brush as in the brush of Turner. As for
examples of the PLACE TO RESULT (product) metonymy, there are Bordeaux
(‘wine’), Scotch (‘whisky’), and so on, where place names seem to function
partly as agents because they are closely connected with the local people who
have inherited a special knowledge of the products there. This may be related
to the fact that there are as many examples of the AGENT TO RESULT metonymy,
which is exemplified by:
(40) a. That was a truly beautiful Picasso.
b. I found myself whistling Mozart under my breath.
Not only can persons be the cause (agent = producer) of valuable products, but
companies can also be agents:
(41) a. I pulled the cover from my mother’s old Olivetti.
b. I eat anything. Cold pizza, McDonald’s, you name it.
(vi) ENSUING SITUATION TO PRECEDING SITUATION. Representative examples
of this transfer pattern are:
(42) a. She is my pride and joy.
b. The news was a great satisfaction to all of us.
c. Self-complacency is the death of the artist.
Pride and joy in (42a) mean what causes them. Likewise, if the basic sense of
satisfaction is assumed to be ‘the resultant state of being satisfied,’ a (great)
satisfaction in (42b) is a metonymic transfer from it, meaning ‘what satisfies
one.’ Finally, death in (42c) means ‘cause of death,’ just as death (‘bullets’) in
Death fell in showers.
It has already been seen in (11) how a property can behave as an entity. Some
properties may be essential in the sense that they are inherent attributes that
characterize the whole entities in question; others may be accidental in the
sense that their saliency is highly dependent on context. Clearly, there is a
gradience of properties from the most essential to the most accidental. The
essential end is usually conventionalized and wholly or partly lexicalized,
which is exemplified by:
112 Ken-ichi Seto
Traditional rhetoric has assumed that synecdoche describes the WHOLE FOR
PART (or, more specifically, the PART FOR WHOLE) kind of metonymy, without
distinguishing between the WHOLE-PART and the GENUS-SPECIES relations. Con-
sequently, no meaningful distinction has been drawn between metonymy and
synecdoche. More fundamentally, this is because the C-relation has never
been clearly distinguished from the E-relation. Synecdoche is now defined as
114 Ken-ichi Seto
the C-related transfer, i.e., the categorical transfer based on the semantic
inclusion between a more comprehensive and a less comprehensive category.
Thus defined, synecdoche becomes an internally consistent and externally
independent category. However, questions remain about the status of synec-
doche. Is it a category that should be comparable to metonymy, or is it a far
smaller category of minor importance? Moreover, whereas metonymy is rich
in variety as has been demonstrated in Section 5, synecdoche in the sense
defined above may look simple and poor in variety, because taxonomy ex-
hausts the C-relation. It is true, indeed, that synecdoche has no more than two
subtypes, but this fact should not in the least reduce the significance of
synecdoche. The categorical transfer based on taxonomy is, in fact, so preva-
lent in our daily language that its importance in the cognitive system could not
be overemphasized. The ubiquity of synecdoche in this sense is such that
actual examples are easy to overlook, with the consequence that synecdoche
has scarcely received the attention it deserves.12
Logically, there are two types of synecdoche: one is the transfer from a
less comprehensive category (species) to a more comprehensive (genus); the
other is the transfer from a more comprehensive (genus) to a less comprehen-
sive (species). The SPECIES TO GENUS transfer is exemplified by walkman: it
was and is the trademark for Sony’s personal stereo, but the once proper name
Walkman (species) has become a common noun, walkman (genus), for any
personal stereo. Likewise, the proper name Spa became a common noun, spa.
Another kind of example is pencil case where pencil (species) means ‘writing
instruments’ (genus). Bread as in to earn one’s (daily) bread is a similar
example. The transfer pattern from SPECIES TO GENUS is also observable in
verbs:
(48) If you wanted to bury, burn, or ship refuse in the Chicago area, you
had to cut him in on the action.
Ship is an instrumental denominal verb, but the referential range of the
instrument has been extended from marine vessels to any means of transporta-
tion.
The second subtype of synecdoche is the transfer from GENUS TO SPECIES:
(49) a. I have a temperature. (‘fever’)
b. I got a ticket again. (‘traffic ticket’)
c. My, my, you’re certainly a sight. (‘terrible sight’)
Distinguishing Metonymy from Synecdoche 115
These are all general nouns used to convey more specific senses than they
literally mean. We do not always have to specify the details of things; when
the situation permits, we tend to use words with more general meanings that
require less effort. This phenomenon may have something to do with the
maxim of informativeness, the second of the Gricean four maxims (cf. Grice
1975). An extreme case of this sort is observed in:
(50) Now that he’s been promoted, he thinks he’s really somebody. (‘an
important person’)
Some -able adjectives also show the GENUS TO SPECIES pattern of synecdoche:
(51) Naperville is ringed by genteel tract houses on sizable lots.
The adjective sizable means ‘rather large,’ though whatever has a size, regard-
less of whether it is large or small, can be, literally, size-able. Readable,
considerable, etc. are of this type, too.
Perhaps somewhat harder to notice are verbal examples of the GENUS TO
SPECIES synecdoche. For instance, suppose one fashion model criticizes an-
other model’s manner of walking on the stage by uttering:
(52) She can’t walk.
Example (52) means ‘she can’t walk professionally (on the stage).’ Once it is
understood that walk in (51) is an example of the GENUS TO SPECIES synecdo-
che, the number of the examples of this class become virtually infinite; any
word can be used in a more specific sense than it normally means. Of course,
the opposite also holds true. The English language (and probably any other
languages) provides some metalinguistic devices to deal with the elasticity of
meaning of this kind: strictly speaking, loosely speaking, real (e.g., Mary’s
husband is a real bachelor), fake, etc. (Lakoff 1972; Taylor 1995) as well as in
a narrow sense and in a broad sense. What should be stressed here is that
practically all words (perhaps with a few exceptions) are capable of changing
their own categorical range (within limits) with no assistance from any of the
above devices. To limit ourselves to the GENUS TO SPECIES synecdoche, we
could say, for example, I can’t sleep or You can’t hit, meaning, respectively, ‘I
can’t sleep well’ and ‘you can’t hit (a ball) well.’ Or suppose a mother asks her
son to go and buy a dozen eggs, then she would expect him to come back with
a dozen chicken eggs, not a dozen reptile eggs. Sometimes professional people
may take advantage of this type of synecdoche. For example, stone for ‘jewel,’
116 Ken-ichi Seto
7. Conclusion
I have discussed why the essential nature of the term ‘metonymy’ has not yet
been agreed upon and given some suggestions for improving upon the present
situation. One obstacle which has prevented a proper understanding of me-
tonymy is the failure to distinguish the C-relation from the E-relation. The E-
relation, which is the contiguous relation between one entity and another in the
world, is quite different from the C-relation, which is the conceptual relation
between a more comprehensive (more inclusive) and a less comprehensive
(less inclusive) category. It must be stressed here again that a category and an
entity are cognitively two different things. It is important to note this because
confusing them is so frequent in the literature of metonymy (and synecdoche).
Another obstacle, closely connected with the first, is the failure to notice
the seeming ambiguity of the terms ‘whole’ and ‘part,’ which led to the
traditional view to treat synecdoche as a mere subtype of metonymy. In
people’s folk view, taxonomies (‘kind of’ relations) tend to be equated with
partonomies (‘part of’ relations). It has been argued in this paper that synecdo-
che, if it is to be a consistent category, should take only taxonomy, leaving
partonomy to metonymy because taxonomy is equivalent to the C-relation
while partonomy is one type of the E-relation. Therefore, it is concluded that
synecdoche, which is C-related, should be independent from metonymy,
which is E-related.
Notes
1. Part of this paper was first read at the 11th New Zealand Linguistic Society Conference
(Victoria University, 19 May, 1995). I would like to thank Claudia Brugman, Elizabeth
Mathis, Günter Radden, Jae Jung Song, Tomoko Tsujimoto, Kimihiro Yoshimura and
Beatrice Warren for their valuable comments and suggestions on earlier versions of this
paper. I am especially indebted to Brigitte Nerlich for a number of critical comments on
the manuscript. My special thanks are due to John Taylor for reading the entire draft and
making helpful suggestions and constructive criticisms.
Distinguishing Metonymy from Synecdoche 117
confirmed by the fact that all the examples in this section, like most of the metonymic
examples in the other sections, translate acceptably into Japanese word for word.
10. One may find it difficult to accept some of the following examples as instances of
metonymy because the nominalization process itself causes the shift in meaning. It may
be argued, however, that the meaning shift in the nominalization process itself is
metonymic. Thus supplies seems doubly metonymic: one from supply (v) to supplies (n);
the other from supply (n) as in a regular supply of fresh vegetables to supplies (n).
11. Most metonymic examples in this paper have come from Sara Paretsky’s V. I.
Warshawski series: Guardian Angel, Toxic Shock, Burn Marks, Tunnel Vision, and
Indemnity Only. A few examples are from dictionaries and other sources.
12. In diachronic semantics synecdoche in our sense has been known as ‘generalization’ and
‘specialization.’ Thus Geeraerts (1994) refers to specialization, generalization, me-
tonymy, and metaphor as “the classical quartet,” which I would like to call ‘the classical
trio’ in terms of synecdoche, metonymy, and metaphor. Nerlich (forthcoming) is clear in
appreciating the cognitive difference between metonymy and synecdoche along the line
of the present paper: “Metonymy is based on our world-knowledge about space and time,
cause and effect, part and whole, whereas synecdoche is based on our taxonomic or
categorical knowledge. Metonymy exploits our knowledge of how the world is, synecdo-
che of how it is ordered in our mind.”
References
Bredin, Hugh
1984 Metonymy. Poetics Today 5: 45–58.
Clark, Eve V., Herbert Clark
1979 When nouns surface as verbs. Language 55: 767–811.
Cooper, David E.
1986 Metaphor. Oxford: Basil Blackwell.
Cruse, D. A.
1979 On the transitivity of the part-whole relation. Journal of Linguistics 15: 29–
38.
1986 Lexical Semantics. London: Cambridge University Press.
DCE
1995 Longman Dictionary of Contemporary English. Essex: Longman.
Dirven, René
1993 Metonymy and metaphor: different mental strategies of conceptualisation.
Leuvense Bijdragen 82: 1–28.
Geeraerts, Dirk
1994 Historical semantics. In R.E. Asher (ed.), The Encyclopedia of Language
and Linguistics. Oxford: Pergamon Press, 1567–1570.
Goossens, Louis
1990 Metaphtonymy: the interaction of metaphor and metonymy in expressions
for linguistic action. Cognitive Linguistics 1: 323–340.
Distinguishing Metonymy from Synecdoche 119
Grice, H. Paul
1975 Logic and conversation. In P. Cole, J.L. Morgan (eds.), Syntax and Seman-
tics. Vol. 3: Speech Acts. New York: Academic Press, 41–58.
Herskovits, Annette
1986 Language and Spatial Cognition: An Interdisciplinary Study of the Prepo-
sitions in English. Cambridge: Cambridge University Press.
Jakobson, Roman
1971 [1956] Two aspects of language and two types of aphasic disturbances. In
R. Jakobson (ed.), Roman Jakobson: Selected Writings. Vol. 2: Word and
Language. The Hague, etc.: Mouton, 239–259.
Lakoff, George
1972 Hedges: a study in meaning criteria and the logic of fuzzy concepts. CLS 8:
183–228.
1987 Women, Fire, and Dangerous Things: What Categories Reveal about the
Mind. Chicago: The University of Chicago Press.
Lakoff, George, Mark Johnson
1980 Metaphors We Live By. Chicago: The University of Chicago Press.
Lakoff, George, Mark Turner
1989 More than Cool Reason. Chicago: The University of Chicago Press.
Langacker, Ronald W.
1987 Foundations of Cognitive Grammar. Volume I: Theoretical Prerequisites.
Stanford, California: Stanford University Press.
1991 Foundations of Cognitive Grammar. Volume II: Descriptive Application.
Stanford, California: Stanford University Press.
Lindner, Susan
1981 A Lexico-Semantic Analysis of English Verb Particle Constructions with
OUT and UP. Bloomington: The Indiana University Linguistics Club.
Lyons, John
1995 Linguistic Semantics: An Introduction. Cambridge: Cambridge University
Press.
Nerlich, Brigitte
forthcoming Synecdoche: a trope, a whole trope, and nothing but a trope. In N.
Norrick, A. Burkhardt (eds.), Tropic Truth. Amsterdam and Philadelphia:
Benjamins.
Nunberg, Geoffrey D.
1978 The Pragmatics of Reference. Bloomington: The Indiana University Lin-
guistics Club.
Ogden, Charles K., Ivor A. Richards
1956 [1923] The Meaning of Meaning. London: Routledge and Kegan Paul.
Pustejovsky, James
1995 The Generative Lexicon. Cambridge, Massachusetts: The MIT Press.
Pustejovsky, James, Pierrette Bouillon
1995 Aspectual coercion and logical polysemy. Journal of Semantics 12: 133–
162.
120 Ken-ichi Seto
Reddy, Michael J.
1979 The conduit metaphor: a case of frame conflict in our language about
language. In A. Ortony (ed.), 1993. Metaphor and Thought. Second edition.
Cambridge: Cambridge University Press, 164–201.
Sorensen, Roy A.
1990 Process vagueness. Linguistics and Philosophy 13: 589–618.
Talmy, Leonard
1983 How language structures space. In H.L. Pick, Jr., L.P. Acredolo (eds.),
Spatial Orientation: Theory, Research, and Application. New York, etc.:
Plenum Press, 225–282.
Taylor, John R.
1995 Linguistic Categorization: Prototypes in Linguistic Theory. Second edi-
tion. New York: Oxford University Press.
Tversky, Barbara
1990 Where partonomies and taxonomies meet. In S.L. Tsohatzidis (ed.), Mean-
ings and Prototypes: Studies in Linguistic Categorization. London:
Routledge, 334–344.
Tversky, Barbara, Kathleen Hemenway
1984 Objects, parts, and categories. Journal of Experimental Psychology: Gen-
eral 113: 169–191.
Warren, Beatrice
1995 Distinguishing between metaphor and metonymy. In G. Melchers, B. War-
ren (eds.), Studies in Anglistics. Stockholm: Almquist and Wiksell Interna-
tional, 137–151.
Weinrich, Harald
1964 Typen der Gedächtnismetaphorik. Archiv für Begriffsgeschichte 9: 96–104.
Wells, Rulon S.
1977 Metonymy and misunderstanding: an aspect of language change. In R. W.
Cole (ed.), Current Issues in Linguistic Theory. Bloomington: Indiana
University Press, 195–214.
Wierzbicka, Anna
1984 Apples are not a kind of fruit: the semantics of human categorization.
American Ethnologist 11: 313–328.
Winston, Morton E., Roger Chaffin, Douglas Herrmann
1987 A taxonomy of part-whole relations. Cognitive Science 11: 417–444.
Aspects of Referential Metonymy
Beatrice Warren
Lund University
1. Introduction
One of the most significant effects of Grice’s seminal paper in 1975 was that it
highlighted the importance of implicitness in human communication. There is
now general acceptance that the principle of compositionality is not sufficient
to account for the interpretation of utterances. To specify the conditions which
allow the unsaid to be communicated has therefore become an essential task of
linguistic theory of today. The current interest in metonymy can be seen as
concerned with a particular aspect of this very task.
Some examples of metonymy discussed in more recent literature are
given in (1)–(6) below. (The metonymic expression is in italics; the interpreta-
tion taken to be intended is within square brackets.)
(1) It won’t happen while I still breathe. [live]
(Halliday 1994: 340)
(2) A: How did you get to the airport?
B: I waved down a taxi. [A taxi took me there]
(Gibbs 1994: 327)
(3) She turned pale. [was frightened]
(Kövecses, quoted from Ungerer and Schmid 1996: 132)
122 Beatrice Warren
(4) “Oh dear,” she giggled , “I’d quite forgotten!” [she said while
giggling]
(Goossens 1990: 328)
(5) A customer at a parking lot handing a car key to the attendant:
This is parked out there. [the car to which the key belongs]
(Nunberg 1996: 110)
(6) She has her father’s eyes. [eyes like those of her father]
(Warren, in print)
In all these examples the speaker feels confident that the hearer will elaborate
on what has been said or on a certain part of it in a certain direction and that the
hearer will take these elaborations to be part of the conveyed message. The
examples also have in common that that which triggers the inference is a
salient subpart of a state, situation or entity: in order to live one must breathe;
getting from A to B in a taxi involves first of all getting it to stop; paleness
tends to accompany fright; keys will go to some object with a lock, etc. Our
ability to access an entire state, situation or object from the mention of some
part is seen by some as the hallmark of metonymic thinking (cf. in particular
Gibbs 1994: 319). It is this type of thinking that the speaker intuitively knows
is common to all of us and which makes it possible for her/him to trust that her/
his message will be understood in the intended manner.
However, a second look at the examples in (1)–(6) reveals that they are
not quite equivalent from a conceptual point of view. For instance, breathing
is a condition for living, whereas paling is not a condition for, but a common
effect of, fear and keys are neither results of nor conditions for cars. Also from
a linguistic point of view, there are differences. Some are clearly propositional
(notably (2)); others are clearly referential ((5) and (6)). Above all, some
violate truth conditions ((4)–(6)); others do not ((1)–(3)). Therefore, although
I am prepared to accept the possibility that there is a common cognitive basis
for the italicized expressions in (1)–(6), I suggest that there are different types
of metonymy which have different constraints and which behave linguistically
in different ways. Making subdivisions of metonymy should have the advan-
tage of providing homogeneous sets of examples, which obviously makes it
easier to work out the processes according to which the examples in question
are formed and deciphered. In this spirit, the focus of the present contribution
is the particular type of metonymy which I will refer to as referential me-
tonymy.
Aspects of Referential Metonymy 123
Finally, as just pointed out, the relations described are not restricted to
compounds, denominal verbs and referential metonymy, but occur also in
other constructions. An important point to make in this connection is that not
all relations occur in all constructions. There appear to be constraints, some of
which are natural, whereas others are more intriguing. This, however, will
have to be the topic of another paper.
These omissions might be excused since the main purposes of Table 1 are
to reveal the basic semantic structure of referential metonyms and its striking
similarity to some other linguistic constructions. It is also meant to bring out
the similarity between the relations described and case endings, such as
essives, ergatives, instrumentals, possessive genetives, locatives, ablatives,
and consequently, deep case roles. Consider AGENT, INSTRUMENT, RESULT,
SOURCE, LOCATIVE and EXPERIENCER (a type of container or possessor). If the
interpretation of sentences involves assigning roles, i.e., working out unex-
pressed relations — as most linguists assume — then it is perhaps not very
surprising that we should find these same types of relations also in phrases and
other constructions involving implicit relations.
The problem facing linguists working with case roles is that it appears
impossible to give an exhaustive list of these. The same problem faces us here.
Although the 120 examples in my corpus can be taken to be based on one or
the other of the specified relations, arguably there are counterexamples.
Consider Nixon bombed Hanoi. The most natural papraphrase for the
metonym Nixon here appears to be ‘those who Nixon controlled or could
activate,’ rather than ‘those who Nixon represents.’ Consider also the much
quoted example The ham sandwich is waiting for his check and examples such
as The appendicitis is in ward 8 (cf. Dirven, this volume), which appear to be
counterexamples, although with some stretching it could also be maintained
that these are based on types of possession relations.4
Nevertheless, the regularity and ease with which referential metonyms
involving composition, causation, possession, etc. are formed and interpreted
and the difficulty of finding convincing counterexamples warrant the sugges-
tion that the relations in Table 1 are default relations. (This suggestion is also
in line with Langacker’s solution to the problem of case roles. That is, there
are certain role archetypes (cf. Langacker 1991: 236).) This would also
explain why they tend to be grammaticalized. In fact, we may reverse the
argument: if there are certain relations which are so frequently expressed that
they tend to be grammaticalized, they ought to be predominant. The reason
Aspects of Referential Metonymy 127
As has just been pointed out, I see strong parallels between noun-noun
compounds and referential metonyms. Both constructions involve at least two
referents,5 and both involve an implicit link. The difference is that the refer-
ring item is explicit in compounds, whereas it is implied in metonyms.
Consider a few examples:
(7) I am bugged [the place I am at]
(8) We are just across the river
[the place we live/work in]
(9) Are you in the printer?6 [that which you have produced, i.e.,
your paper]
(10) The bathtub is running over [that which is in the tub, i.e., the wa-
ter]
(11) Answer the door! [the person at the door]
(12) I like Mozart. [that which Mozart has produced,
i.e., his music]
128 Beatrice Warren
trigger and target, the hearer is expected to retrieve the feature or features that
form the basis of the similarity. Consider the interpretation of (17):
(17) Mary is the Cinderella in the family.
This involves working out some such features as ‘unjustifiably neglected’ and
possibly also ‘a person with qualities as yet not appreciated but which are
superior to her siblings (or the equivalent) and which will eventually raise her
above them in status.’ Compare this interpretation to that of Cinderella in (18),
which simply involves working out that Mary is the person representing
Cinderella in the play.
(18) Mary is Cinderella in the play. [the person representing Cinderella]
My suggestion concerning the crucial difference between referential me-
tonymy and metaphor is, therefore, that in the case of referential metonymy
the link between trigger and target is a relation (and one relation only),
whereas in the case of metaphor, it involves one or more attributes.
Working out several links (i.e., similar attributes) between trigger and
target may cause the interpretation to be much richer, but also less straightfor-
ward than is the case of metonymy. Indeed interpretations of metaphors often
vary with the interpreter, probably to a much greater extent than in the case of
metonymy. This indefiniteness may in turn make metaphors enigmatic and
interesting in a way that is not so common in the case of metonymy.
Perhaps the most important difference between metaphor and metonymy
is that metaphors often involve hypothetical thinking. When we interpret the
metaphor in (19), for instance, we see information as if it were a fluid seeping
through a container supposed to hold it (the White House).
(19) Information about the matter leaked from the White House.
Metonymy does not seem to involve hypotheses. Perhaps we can explain this
difference by pointing out that metonymy is based on relations which presup-
pose actual coincidence, whereas metaphor, which involves finding a match
for an attribute among all the mentally stored attributes, is freed from con-
straints of what could actually occur or coincide.
132 Beatrice Warren
6. Conclusion
The speaker expects the hearer to be able to fill in that which is left implicit,
which will result in a description of the intended referent.
I have also suggested that metonymic links are not quite as ad hoc as the
much quoted example The ham sandwich is waiting for his check may
suggest, but that there is a limited set of default relations which are of great
linguistic relevance (cf. Radden and Kövecses, this volume). This is evi-
denced by the fact that they are not restricted to metonymy, but are part of a set
of unexpressed relations that are important in the semantics of phrases and
sentences.
The illustration above also reveals that I look upon referential metonymy
as a kind of abbreviation having potentials as a naming and/or rhetorical
device.
I further argued that we should make a distinction between concomitance
relations between propositions (i.e., ANTECEDENT-CONSEQUENT relations) and
concomitance relations between entities (i.e., relations such possession, loca-
tion, causation, etc.). The former type of relation may be the basis of what is
sometimes referred to as pragmatic meanings of words which do not cause
violations of truth conditions and which may be cancellable. The latter type of
Aspects of Referential Metonymy 133
Notes
1. By “not explicitly mentioned” I mean that there is no morpheme which could be taken to
represent it. Printer, for instance, would not be considered a metonym, since the suffix -
er can be taken to formally — although opaquely — indicate the nominal status of the
expression, reflected in the interpretation ‘that which prints.’
2. The terms ‘trigger’ and ‘target’ are adopted from Fauconnier (1985).
3. These are the links involved in turning count nouns into non-count nouns and vice versa,
as in there is egg on the knife and two coffees, please. The former process is sometimes
referred to as ‘grinding’ and the latter as ‘chunking.’ Cf. Copestake and Briscoe (1996).
4. Nerlich et al. (this volume) see an unconventional metonymic relation in the compound
Treacle people juice, formed by her son Matthew. The connection between Treacle
people and juice is that Matthew was served a particular juice after having watched a TV
play called “The Treacle People.” In my view, the connection fits quite well with the
OBJECT -TIME pattern: ‘that which (practically) co-occurred with The Treacle People.’ The
expression is strange, however, in that the play will hardly affect the character of the
drink, i.e., be a salient subpart of it.
5. The basic structure of both compounds and metonyms can be expanded. Consider
compound within compounds (cowboy hat) and metaphors within metonyms or
metonyms within metaphors as in yellow belly in the sense ‘a native of the fens’: ‘sb who
is like sb who has a yellow belly (i.e., a frog).’ Cf. Warren (1992: 94–97).
6. This is a translation from Swedish, i.e., from Är det du som ligger i skrivaren?
7. There are, of course, other possible interpretations of (7), viz. ‘I am irritated’ or ‘I have
bugs on me.’
8. An increasing number of linguists do, however, approach metonymy in this spirit. Cf. in
particular Jackendoff (1997), Nunberg (1996), Fass (1997) and Copestake and Briscoe
(1996).
9. A terminological distinction of this kind is not infrequently made. Consider, for instance,
Stern (1965), who refers to sense shifts developed from the former type of relation as
permutations.
134 Beatrice Warren
References
Bergsten, Nils
1911 A Study of Compound Substantives in English. Uppsala: Almquist and
Wiksell.
Clark, Eve V., Herbert H. Clark
1979 When nouns surface as verbs. Language 55: 767–811.
Copestake, Ann, Ted Briscoe
1996 Semi-productive polysemy and sense extension. In J. Pustejovsky, B.
Boguraev (eds.), Lexical Semantics: The Problem of Polysemy. Oxford:
Clarendon Press, 15–67.
Fass, Dan
1997 Processing Metonymy and Metaphor. Greenwich, Connecticut, etc.: Ablex
Publishing Corporation.
Fauconnier, Gilles
1985 Mental Spaces. Cambridge, Massachusetts: The MIT Press.
Gibbs, Raymond W.
1994 The Poetics of Mind: Figurative Thought, Language and Understanding.
Cambridge: Cambridge University Press.
Goossens, Louis
1990 Metaphtonymy: the interaction of metaphor and metonymy in expressions
for linguistic action. Cognitive Linguistics 1: 323–340.
Grice, H. Paul
1975 Logic and conversation. In P. Cole, J. Morgan (eds.), Syntax and Semantics.
Volume 3: Speech Acts. New York: Academic Press, 41–58.
Halliday, Michael A.K.
1994 An Introduction to Functional Grammar. London, Melbourne, Auckland:
Edward Arnold.
Jackendoff, Ray
1997 The Architecture of the Language Faculty. Cambridge, Massachusetts: The
MIT Press.
Langacker, Ronald W.
1991 Concept, Image and Symbol. Berlin, New York: Mouton de Gruyter.
Leisi, Ernst
1985 Praxis der Englischen Semantik. Heidelberg: Winter.
Leitner, Gerhard
1974 Denominale Verbalisierung im Englischen. Tübingen: Niemeyer.
Levi, Judith
1978 The Syntax and Semantics of Complex Nominals. New York, San Fran-
cisco, London: Academic Press.
Lipka, Leonhard
1988 A rose is a rose is a rose: on simple and dual categorization in natural
languages. In W. Hüllen, R. Schulze (eds.), Understanding the Lexicon.
Tübingen: Niemeyer, 355–366.
Aspects of Referential Metonymy 135
Mellenius, Ingmarie
1997 The Acquisition of Nominal Compounding in Swedish. Lund: Lund Univer-
sity Press
Nunberg, Geoffrey
1996 Transfer of meaning. In J. Pustejovsky, B. Boguraev (eds.), Lexical Seman-
tics: The Problem of Polysemy. Oxford: Clarendon Press, 107–132.
OED
1989 The Oxford English Dictionary. Second edition. Oxford: Clarendon Press
Stern, Gustaf
1965 [1931] Meaning and Change of Meaning. Bloomington: Indiana Univer-
sity Press.
Ungerer, Friedrich, Hans-Jörg Schmid
1996 An Introduction to Cognitive Linguistics. London, New York: Longman.
Warren, Beatrice
1978 Semantic Patterns of Noun-Noun Compounds. Gothenburg Studies in En-
glish 41.
1992 Sense Developments. Stockholm: Almquist and Wiksell International.
1995 Distinguishing between metaphor and metonymy. In G. Melchers, B. War-
ren (eds.), Studies in Anglistics. Stockholm: Almquist and Wiksell Interna-
tional, 137–151.
1998 What is metonymy? In M. Hogg, L. van Bergen (eds.), Historical Linguis-
tics 1995. Amsterdam and Philadelphia: Benjamins, 301-310.
Part II
Peter Koch
University of Tübingen
1. Introduction1
To my knowledge, it is here for the first time that the cognitive component of
metonymy is explicitly stated in terms of associationist psychology (and in
accordance with the classical rhetorical tradition):
(5) Changements résultant d’une association par contiguïté entre les
idées. Tous les changements que l’on a appelés changements par
connexité ou par métonymie (devenue inconsciente) appartiennent
à cette catégorie. (Roudet 1921: 690)
Take our example (2). In a public house we normally find a counter across
which liquor and food are served. Thus, there is a spatial contiguity between
the ideas COUNTER and PUBLIC HOUSE. Their contiguity must have triggered the
metonymic change of meaning (from ‘counter’ to ‘public house’) that under-
lies the metonymic polysemy of Eng. bar ‘counter; public house.’
Now let us consider another example (cf. DHLF, s.v. prison).
Frame and Contiguity 143
Let us begin with the semiotic problem. As we have already seen (cf. quota-
tion (5)), according to Roudet, metonymy is based on a contiguity between
ideas, that is — in more up-to-date terms — ‘conceptual’ contiguity. In
contrast to this, Table 2 reveals that Jakobson’s authoritative two-axes theory
does not deal with concepts, but with linguistic ‘signs.’ So in Jakobson’s
approach, metonymy seems to presuppose a (semantic) contiguity between
linguistic signs. Similarly, in the period following, many scholars accounted
for metonymy and metonymic change in terms of contiguity between the
‘senses’ of two ‘words.’
Consider for instance Ullmann’s systematics of the ‘essence’ of semantic
change, which only superficially seems identical to Roudet’s system (Table
1), by which it has indeed been inspired (cf. Ullmann 1962: 211–227; Blank
1997a: 19f, 35–38).
PUBLIC HOUSE
COUNTER
With regard to example (7), there is a FIREPLACE frame (Figure 2), one of
whose elements is obviously FIRE.
FIREPLACE
FIRE
DEPRIVATION OF LIBERTY
PRISONER
ACT OF
CAPTIVITY PRISON
SEIZING
PENALTY OF
IMPRISONMENT
MARRIAGE
TRUST,
ENGAGEMENT
MARRIAGE
CONTRACT WEDDING
PARENTS
MARRY1 PRAYER
= UNITE IN MARRY2 VOW,
BETROTHAL MATRIMONY = TAKE AS OATH
WIFE/HUSBAND
WIFE/ UNION OF
BRIDE
FIANCÉ(E) HUSBAND WIFE AND
(GROOM)
HUSBAND
SET UP MOTHERHOOD
HOUSE
But there still remain some problems with contiguity and even with frames.
The original paradigm of associationist philosophy and psychology (see Sec-
tion 3), which leans heavily on contiguity, takes a rather mechanistic ap-
proach.9 Whenever there is contiguity, there has to be an associative link. But
is not everything contiguous to everything else? So why do we not associate
everything with something or something with anything or everything? Frame
models seem to be more subtle because they represent non-accidental net-
works of contiguities. But even a frame model conceived as categorical
(something is or is not in a frame: this is inevitable, I think, in Artificial
Intelligence) would not serve our purpose. Quite on the contrary:
– We have to acknowledge (see (2)) that there are perhaps public houses
without a counter, which we would nevertheless call bars.
– We have to acknowledge (see (7)) that there are fireplaces without a
(burning) fire and that fire is not confined to fireplaces; and, nevertheless,
speakers of Vulgar Latin would have used the word focus to denote FIRE.
150 Peter Koch
PUBLIC contiguity
COUNTER
HOUSE
When we introduce notions like ‘prototype’ and ‘salience’ into frame theory,
we recognize the ‘gestalt’ character of frames and contiguities.12 It is only by
leaning on salience effects within frames that we can avoid the aporias of a
mechanistic associationist approach or of categorical frame models.
More precisely, from a gestalt perspective, metonymy turns out to be a
‘figure/ground’ effect. Consider Figure 6 as a traditional example of a percep-
tual figure/ground constellation. In this figure, we can perceive a white cross
on black ground, but alternatively, it seems to be a black cross on white
ground (cf. Wittgenstein 1958: 207).
Coming now to the conceptual level, we can claim that every concept desig-
nated by a given lexical item appears as a figure in relation to (at least) another
contiguous concept that — for the time being — remains the ground within the
same frame. But at some moment, while we are using the same lexical item,
certain pragmatic, conceptual or emotional factors may highlight the ground
concept so that figure and ground become inverted. That is what we call
metonymy.
Thus, in one of the senses of OFr. prison (‘captivity’; see above (6)),
CAPTIVITY was the figure and PRISON one possible ground (Figure 7a). By
highlighting PRISON into the figure and backgrounding CAPTIVITY, prison was
acquiring the new metonymic sense ‘prison’ (Figure 7b).
According to Croft (1993: 348), we can interpret metonymy as a concep-
tual effect of domain highlighting within one domain matrix (opposing it to
metaphor as a conceptual effect of domain mapping across different domain
matrices).13 In the case of OFr. prison ‘captivity; prison,’ we shift, say, from
the domain (HUMAN) CONDITION to the domain LOCATION within the domain
matrix for prison. Note, however, two significant differences between the
domain-matrix approach and the frame approach proposed above:
1. Frame-internal relations, in as far as they represent contiguity relations,
exclude any similarity or taxonomic relations (relevant to metaphor and
extension/restriction of meaning only). For domain (matrices) this is not
so clear: the domain structure underlying the concept of the letter ‘T,’ for
DEPRIVATION OF LIBERTY
CAPTIVITY PRISON
DEPRIVATION OF LIBERTY
CAPTIVITY PRISON
tradition, we can distinguish metonymy from synecdoche, but the latter often
appears to be only a special case of the former (Lausberg 1973: §§572–577;
Bredin 1984: 45f). However, as Le Guern (1973: 36ff) has pointed out,
synecdoche is not, in reality, a unitary trope (see also Section 7.1.): traditional
synecdoche comprises (or rather: confuses) cases of taxonomic extension/
generalization (e.g., bread ‘foodstuff’) or restriction/specification (e.g., mor-
tal ‘man’) on the one hand, which have nothing to do with metonymy, and
cases of pars pro toto/totum pro parte on the other hand (e.g., roof ‘house’;
America ‘USA’).
In my opinion, we should integrate pars pro toto and totum pro parte into
metonymy (cf. also Ullmann 1962: 212; Schifko 1979: 247; Lakoff and
Johnson 1980: 36; Croft 1993: 350; Warren 1992: 64ff, 1995; Blank, this
volume; Seto, this volume). Separating metonymy and pars/totum ‘synecdo-
che’ would be artificial because the difference between pars/totum relations
and (other) contiguities is often not so easy to pin down. For instance, are the
relations COUNTER – PUBLIC HOUSE or FIRE – FIREPLACE to be considered pars/
totum relations or relations of location?14 Behind these possible differences
and uncertainties, we nevertheless perceive a fundamental constant: the pars/
totum tropes, like any metonymic trope, involve a figure/ground effect: in pars
pro toto, the totum — as a ground that becomes the figure — is a whole frame,
and the pars — as a figure that becomes the ground — is one of the concepts
of this frame (and vice versa for totum pro parte).
So we can retain our definition given in Section 5: contiguity is a salient
relation that exists between the elements (or sub-frames) of a conceptual
frame or between the frame as a whole and its elements. Consequently,
metonymy implies a contiguity-based figure/ground effect between elements
of a conceptual frame or between the frame as a whole and one of its elements
(or vice versa).
A new differentiation between two contiguity-based changes of meaning
has been suggested by Warren, who distinguishes ‘metonymy’ from ‘implica-
tion’ (cf. Warren 1992: 51–72, 101, 1995). Metonymy (in her narrow sense of
the term) is a non-literal use of a word that causes an abrupt shift of meaning,
restricted to a few, well-defined contiguity relations and — normally — to
nouns. Implication, on the other hand, is the gradual development of novel,
coexisting senses, based on if-then contiguities of any kind and occurring in
adjectives, verbs, and nouns. I reject this differentiation on the following
grounds.
Frame and Contiguity 155
First of all, it can easily be shown that metonymies (in this narrow sense)
are possible even with verbs and adjectives:
(9) Fr. descendre ‘to go down,’ hence: ‘to take down’ (with many
parallels in other languages; cf. Koch 1991: 294; Haspelmath
1993: 92–94, 101, 104, 112–120; see also our example Eng.
marry in (8f))
(10) Eng. sad ‘distressed,’ hence: ‘distressing’ (with many parallels in
other languages; cf. BDE, s.v. sad)
Secondly, we have to acknowledge that a frame-based figure/ground
effect is present in ‘implications’ as well as in ‘metonymies’ (in the narrow
sense). I would claim that the difference resides rather in the pragmatic,
referential and expressive conditions in which these metonymies (in the broad
sense) emerge in discourse, as we will see in the next section.
lexical item. Traugott and König call this process ‘pragmatic strengthening’ or
‘conventionalization of conversational implicatures/inferences.’ In this case,
it is the hearer (as a virtual speaker) who triggers the metonymic innovation
which induces a metonymic change later on (cf. König and Traugott 1988;
Traugott and König 1991: 193ff).
A similar analysis probably applies to one of the steps of semantic change
in our example (6), which may be represented as in Table 5.
Table 5. Metonymic innovation through inference: PRISON
I suppose that this type of figure/ground effect is the one Warren calls
‘implication.’
A second type of ad hoc figure/ground effect is triggered by the speaker,
as illustrated in Table 6. A Latin speaker wants to refer to the thigh of a person,
but instead of using the ‘exact’ lexical item (Lat. femur), he imprecisely recurs
to another item (coxa) which designates a contiguous concept, namely HIP (cf.
Blank 1997a: 388f).
Table 6. Metonymic innovation through imprecision: THIGH
11. Conclusion
Notes
1. I express my gratitude to Mary Copple (Berlin) for the stylistic revision of this paper.
2. Concerning the intimate link between semantic change and polysemy, cf. Bréal (1921:
143ff, 284–287), Werth (1974: 377f), Koch (1991: 283), Blank (1993: 31), for me-
160 Peter Koch
tonymy, see especially Schifko (1979: 248–251). The lexicalization steps for metaphor
are described in Koch (1994: 203–209); a general framework for lexicalization (and
delexicalization) processes within semantic change has been established in Blank
(1997a: 116–130). Note the important difference between ‘innovation’ and ‘change’
proposed by Coseriu (1958: 44–46).
3. Cf. Amin (1973: 19–81, especially 38). The first to explicitly take up Aristotle seems to
be the Scotsman Thomas Brown, who occupies a somewhat particular position with
respect to associationism (cf. ibid.: 72).
4. Cf. the critical outlines of Wundt’s approach in Roudet (1921: 681–686); Nerlich (1992:
77–80); Blank (1997a: 18f).
5. From Hjelmslev onwards, Saussure’s rather open concept of ‘associative’ relation is
replaced by the technical, more specific concept of ‘paradigmatic’ relation (EITHER/
OR-relation of substitution in a given syntagmatic environment); cf. Hjelmslev (1963:
33–40); Happ (1985: 52–59).
6. Cf. Jakobson (1956: 90f); Holenstein (1974: 81–91); Happ (1985: 75–79, 130–139).
According to Raible (1981), the distinction between semantic(-paradigmatic) contiguity
(hut – thatch) and (semantic-)syntagmatic contiguity (hut – burnt out) is essential for
preferences in human verbal associations.
7. Rastier (1997: 140–145) underscores the undeniable tendency of cognitive linguistics to
spatialize its objects of description.
8. Cf. BDE, s.v. marry; DCECH, s.vv. boda, casar, esposo; DE, s.v. Ehe; DELL, s.v. mater;
DHLF, s.vv. époux, marier; DSSPIL, s.vv. 2.33 marry, 2.34 marriage; REW, s.vv. casa,
sponsus, votum.
9. Cf. the critical observations in Amin (1973, especially 39–42, 111–115).
10. For prototype theory, cf. Rosch (1973); Fillmore (1975); Taylor (1989); Kleiber (1990);
Cordier (1993).
11. See also Geeraerts (1997: 74f). Concerning the difference between salience effects within
a frame and prototypicality effects within a category, cf. Cordier (1993: 122–124, 135–
149).
12. For gestalt theory in general, cf. for example Wertheimer (1922/23); Köhler (1947);
Metzger (1986). Amin (1973: 97–155, 201f) stresses the holistic character of gestalt
psychology as opposed to the mechanistic associationist approach, but he nevertheless
does not definitively exclude a synthesis of association and ‘gestalt’ (cf. also Raible
1981: 5f).
13. Cf. also Taylor (1989: 90). For the notions ‘domain’ and ‘domain matrix,’ cf. Langacker
(1987: 147ff). — As to the totally different character of metaphor, cf., for instance,
Bühler (1965: 342–350); Black (1954: 1977); Lakoff and Johnson (1980); Bredin (1984:
57); Liebert (1992: 28–82); Koch (1994: 209–214).
14. Cf. also Cordier (1993: 123). Within the type of contiguity called ‘co-presence’ by Blank
(this volume), there are several relations that can hardly be distinguished from pars/
totum: TYPICAL ASPECT — ACTIVITY, TYPICAL ASPECT — FRAME, TYPICAL ASPECT — OBJECT,
FUNCTION — OBJECT, OBJECT — PLACE. On the other hand, Blank’s distinction between
Frame and Contiguity 161
References
Amin, Ismail
1973 Assoziationspsychologie und Gestaltpsychologie: Eine problemgeschicht-
liche Studie mit besonderer Berücksichtigung der Berliner Schule. Bern,
Frankfurt/M.: Lang.
Anonymous
1894 Incerti auctoris De ratione dicendi ad C. Herennium libri IV. Leipzig:
Teubner (reprint Hildesheim: Olms, 1966)
Aristotle
1975 Aristotle in Twenty-Three Volumes. VIII: On the Soul. Parva Naturalia. On
Breath. With an English translation by W.S. Hett. Cambridge, MA, Lon-
don: Harvard University Press.
Barsalou, Lawrence W.
1992 Frames, concepts, and conceptual fields. In A. Lehrer, E.F. Kittay (eds.),
162 Peter Koch
Frames, Fields, and Contrasts: New Essays in Semantic and Lexical Orga-
nization. Hillsdale, NJ, London: Erlbaum, 21–74.
Bartlett, Frederick Ch.
1932 Remembering: A Study in Experimental, Social Psychology. Cambridge:
Cambridge University Press.
BDE
1988 The Barnhart Dictionary of Etymology. Robert K. Barnhart, New York:
Wilson.
Black, Max
1954 Metaphor. Proceedings of the Aristotelian Society 55: 273–294.
1977 More about metaphor. Dialectica 31: 431–457.
Blank, Andreas
1993 Polysemie und semantische Relationen im Lexikon. In W. Börner, K.
Vogel (eds.), Wortschatz und Fremdsprachenerwerb. Bochum: AKS, 22–
56.
1997a Prinzipien des lexikalischen Bedeutungswandels am Beispiel der romani-
schen Sprachen. Tübingen: Niemeyer.
1997b Il senso di una semantica dei prototipi e dei ‘frames’: osservazioni
decostruttive e ricostruttive. In M. Carapezza et al. (eds.), Linguaggio e
cognizione. Atti del XXVIII Congresso della Società di Linguistica Italiana,
Palermo, 27–29 ottobre 1994, Roma: Bulzoni, 89–103.
Bonhomme, Marc
1987 Linguistique de la métonymie. Bern: Lang.
Bréal, Michel
1921 Essai de sémantique (science des significations). Fifth edition. Paris:
Hachette.
Bredin, Hugh
1984 Metonymy. Poetics Today 5: 45–58.
Bühler, Karl
1965 Sprachtheorie: Die Darstellungsfunktion der Sprache. Second edition.
Stuttgart: Fischer.
Cordier, Françoise
1993 Les représentations cognitives privilégiées: Typicalité et niveau de base.
Lille: Presses Universitaires de Lille.
Coseriu, Eugenio
1958 Sincronía, diacronía e historia: El problema del cambio lingüístico. Uni-
versidad de Montevideo.
Croft, William
1993 The role of domains in the interpretation of metaphors and metonymies.
Cognitive Linguistics 4: 335–370.
DCECH
1980–83 Diccionario crítico etimológico castellano e hispánico. 5 vol. Second
edition. J. Corominas, J.A. Pascual (eds.), Madrid: Gredos.
DE
1989 Duden. Etymologie. Herkunftswörterbuch der deutschen Sprache. Second
edition. G. Drosdowski, Mannheim: Dudenverlag.
Frame and Contiguity 163
DELL
1951 Dictionnaire étymologique de la langue latine: Histoire des mots. A.
Ernout, A. Meillet (eds.), Paris: Klincksieck.
DHLF
1992 Dictionnaire Historique de la Langue Française. 2 vol. Alain Rey, Paris:
Dictionnaires le Robert.
Dik, Simon C.
1977 Inductive generalisation in semantic change. In P.J. Hopper (ed.), Studies in
Descriptive and Historical Linguistics. Festschrift for W.P. Lehmann.
Amsterdam and Philadelphia: Benjamins, 283–300.
DSSPIL
1971 A Dictionary of Selected Synonyms in the Principal Indo-European Lan-
guages: A Contribution to the History of Ideas. Third edition. C.D. Buck,
Chicago: The University of Chigaco Press.
Dubois, Jacques, Francis Edeline, Jean-Marie Klinkenberg, Philippe Minguet, François
Pire, Hadelin Trinon
1970 Rhétorique générale. Paris: Larousse.
Eco, Umberto
1984 Semiotica e filosofia del linguaggio. Torino: Einaudi.
Fillmore, Charles J.
1975 An alternative to checklist theories of meaning. Proceedings of the Annual
Meeting of the Berkeley Linguistics Society 1: 123–131.
1982 Towards a descriptive framework for spatial deixis. In R.J. Jarvella, W.
Klein (eds.), Speech, Place, Action: Studies in Deixis, Related Topics.
Chichester: Wiley, 31–59.
1985 Frames and the semantics of understanding. Quaderni di Semantica 4: 217–
240.
Gauger, Hans-Martin
1971 Durchsichtige Wörter: Zur Theorie der Wortbildung. Heidelberg: Winter.
Geeraerts, Dirk
1988 Prototypicality as a prototypical notion. Communication and Cognition 21:
343–355.
1997 Diachronic Prototype Semantics: A Contribution to Historical Lexicology.
Oxford: Clarendon.
Grice, H. Paul
1975 Logic and conversation. In P. Cole, J.L. Morgan (eds.), Syntax and Seman-
tics. Vol. 3: Speech Acts. New York: Academic Press, 41–58.
Happ, Heinz
1985 ‘Paradigmatisch’ – ‘syntagmatisch’: Zur Bestimmung und Klärung zweier
Grundbegriffe der Sprachwissenschaft. Heidelberg: Winter.
Haspelmath, Martin
1993 More on the typology of inchoative/causative verb alternations. In B.
Comrie, M. Polinsky (eds.), Causatives and Transitivity. Amsterdam and
Philadelphia: Benjamins, 87–120.
164 Peter Koch
Hjelmslev, Louis
1963 Prolegomena to a Theory of Language. Second edition. Madison: Univer-
sity of Wisconsin Press.
Holenstein, Elmar
1974 Die zwei Achsen der Sprache und ihre Grundlagen. In E. Holenstein (ed.),
1976. Linguistik – Semiotik – Hermeneutik: Plädoyers für eine strukturale
Phänomenologie. Frankfurt/M.: Suhrkamp, 76–113.
1975 Roman Jakobsons phänomenologischer Strukturalismus. Frankfurt/M.:
Suhrkamp.
Jakobson, Roman
1956 Two aspects of language and two types of aphasic disturbances. In R.
Jakobson, M. Halle, 1971. Fundamentals of Language. Den Haag, Paris:
Mouton, 67–96
Jongen, René
1985 Polysemy, tropes, cognition or the non-Magrittian art of closing curtains
whilst opening them. In W. Paprotté, R. Dirven (eds.), The Ubiquity of
Metaphor. Amsterdam and Philadelphia: Benjamins, 121–139.
Kleiber, Georges
1990 La sémantique du prototype. Catégories et sens lexical. Paris: Presses
Universitaires de France.
Koch, Peter
1991 Semantische Valenz, Polysemie und Bedeutungswandel bei romanischen
Verben. In P. Koch, T. Krefeld (eds.), Connexiones Romanicae. Dependenz
und Valenz in romanischen Sprachen. Tübingen: Niemeyer, 279–306.
1993 Kyenbé – tyonbo. Wurzeln kreolischer Lexik. In C. Foltys, T. Kotschi
(eds.), Berliner Romanistische Studien: Für Horst Ochse. Berlin: Institut
für Romanische Philologie der Freien Universität, 259–287.
1994 Gedanken zur Metapher – und zu ihrer Alltäglichkeit. In A. Sabban, C.
Schmitt (eds.), Sprachlicher Alltag: Linguistik – Rhetorik – Literaturwis-
senschaft. Festschrift für Wolf-Dieter Stempel. Tübingen: Niemeyer, 201–
225.
1995 Der Beitrag der Prototypentheorie zur Historischen Semantik: eine kritische
Bestandsaufnahme. Romanistisches Jahrbuch 46: 27–46.
1996a Le prototype entre signifié, désigné et référent. In H. Dupuy-Engelhardt
(ed.), Questions de méthode et de délimitation en sémantique lexicale:
Actes d’EUROSEM 1994. Reims: Presses Universitaires de Reims, 113–
135.
1996b La sémantique du prototype: sémasiologie ou onomasiologie? Zeitschrift
für französische Sprache und Literatur 106: 223–240.
1997 La diacronia quale campo empirico della semantica cognitiva. In M.
Carapezza et al. (eds.), Linguaggio e cognizione. Atti del XXVIII Congresso
della Società di Linguistica Italiana, Palermo, 27–29 ottobre 1994, Roma:
Bulzoni, 225–246.
in print TREE and FRUIT: A cognitive-onomasiological approach. Studi Italiani di
Linguistica Teorica ed Applicata.
Frame and Contiguity 165
Rastier, François
1997 De l’antinomie spatio-temporelle aux zones anthropiques. In H. Dupuy-
Engelhardt, M.-J. Montibus (eds.), L’organisation lexicale et cognitive des
dimensions spatiale et temporelle. Actes d’EUROSEM 1996. Reims:
Presses Universitaires de Reims, 139–153.
REW
1992 Romanisches etymologisches Wörterbuch. Second edition. W. Meyer-
Lübke. Heidelberg: Winter.
Rosch, Eleanor
1973 On the internal structure of perceptual and semantic categories. In Th.E.
Moore (ed.), Cognitive Development and the Acquisition of Language.
New York: Academic Press, 111–144.
Roudet, Léonce
1921 Sur la classification psychologique des changements sémantiques. Journal
de Psychologie normale et pathologique 18: 676–692.
Saussure, Ferdinand de
1916 Cours de linguistique générale. Paris, Lausanne: Payot.
Schifko, Peter
1979 Die Metonymie als universales sprachliches Strukturprinzip. Grazer Lin-
guistische Studien 10: 240–264.
Schwarze, C.
1995 Grammatik der italienischen Sprache. Second edition. Tübingen:
Niemeyer.
Stempel, Wolf-Dieter
1983 ‘Ich vergesse alles’. Bemerkungen zur Hyperbolik in der Alltagsrhetorik. In
M. Faust et al. (eds.), Allgemeine Sprachwissenschaft, Sprachtypologie und
Textlinguistik. Festschrift für Peter Hartmann. Tübingen: Narr, 87–98.
Tannen, Deborah
1979 What’s in a frame? In R.O. Freedle (ed.), New Directions in Discourse
Processing. Norwood, NJ: Ablex, 137–181.
Taylor, John R.
1989 Linguistic Categorization. Prototypes in Linguistic Theory. Oxford:
Clarendon.
Traugott, Elizabeth C., Ekkehard König
1991 The semantics-pragmatics of grammaticalization revisited. In E. Traugott,
B. Heine (eds.), Approaches to Grammaticalization. 2 vol. Amsterdam and
Philadelphia: Benjamins, 189–218.
Ullmann, Stephen
1962 Semantics: An Introduction to the Science of Meaning. Oxford: Blackwell.
Warren, Beatrice
1992 Sense Developments: A Contrastive Study of the Development of Slang
Senses and Novel Standard Senses in English. Stockholm: Almquist and
Wiksel.
1995 What is metonymy? Paper read at the 12th ICHL, Manchester, August
1995.
Frame and Contiguity 167
Weinrich, Harald
1987 Zur Definition der Metonymie und zu ihrer Stellung in der rhetorischen
Kunst. In A. Arens (ed.), Text-Etymologie: Untersuchungen zu Textkörper
und Textinhalt. Festschrift für Heinrich Lausberg zum 75. Geburtstag.
Stuttgart: Steiner, 105–110.
Werth, Paul
1974 Accounting for semantic change in current linguistic theory. In J.M. Ander-
son, Ch. Jones (eds.), Historical Linguistics I: Syntax and Morphology,
Internal and Comparative Reconstruction. Amsterdam: North-Holland,
377–415.
Wertheimer, Max
1922/23 Untersuchungen zur Lehre von der Gestalt. Psychologische Forschun-
gen 1: 47–58.
Wittgenstein, Ludwig
1958 Philosophical Investigations. Second edition. Oxford, Cambridge:
Blackwell.
Wundt, Wilhelm
1912 [1900] Völkerpsychologie: Eine Untersuchung der Entwicklungsgesetze
von Sprache, Mythus und Sitte. Bd. II: Die Sprache. Third edition. Leipzig:
Engelmann.
Co-presence and Succession
A Cognitive Typology of Metonymy
Andreas Blank
University of Marburg
1. Introduction
Over a decade ago, Hugh Bredin (1984) complained about the marginal status
of metonymy studies up to that time.1 And in fact, typologies of metonymy
from Fontanier (1968 [1830]) to Schifko (1979) or Lakoff and Johnson (1980)
lack consistency in that there are usually metonymies remaining that do not fit
into any category. Especially, the widely accepted tripartition into ‘spatial,’
‘temporal’ and ‘causal’ metonymies (cf. Ullmann 1962; Duchácek 1967;
Schifko 1979) is insufficient, as it does not provide an explanation for metony-
mies such as:
(1) OE sælig ‘blessed’ > ModE silly ‘stupid’
(2) OGr aggelos ‘messenger’ > ‘angel’2
(3) Arab al wazir ‘minister,’ ‘mayor’ > Sp alguacil ‘lower law-court
official’
The main concern of the following reflections is how metonymies can best be
classified. In Section 4 I will introduce a couple of older and more recent
170 Andreas Blank
another facet of the old meaning, as in (1) OE sælig ‘blessed’ > ModE silly
‘stupid,’ which of course, initially, was intended to be a euphemistic synonym
of OE dumb or contemporary synonyms. In all cases, the metonymy relies on
extralinguistic world knowledge (cf. Blank 1993, 1997a; Warren 1995; Koch,
this volume): cattle served as money in former times, angels are God’s
messengers, and the most positive thing to say about a stupid person is that he
or she is blessed by Jesus.7
Thus, the associative relation is rather between the two concepts BLESSED
– STUPID, MESSENGER – ANGEL or CATTLE – MONEY. It is important to point out
that this conceptual contiguity exists before the metonymic transfer is done
(cf. Section 3.1.). As Hugh Bredin (1984: 57) says, “[...] metaphor creates the
relation between its objects, while metonymy presupposes that relation.”
Ullmann’s interpretation goes back to a scarcely known article by Léonce
Roudet (1921), who appears to be the first to have brought the terms ‘contigu-
ity’ and ‘similarity’ (Roudet says ‘resemblance’) into historical linguistics.
Roudet (1921: 688f) distinguishes between semantic change due to an asso-
ciation “par contiguïté entre les idées,” “par ressemblance entre les idées” as
well as “changements résultant des rapports syntagmatiques entre les mots.”8
The last type of change is generally called ‘ellipsis,’ the second ‘metaphor’
and the first ‘metonymy’ (cf. Blank 1997b: 157–190, 230–269, 281–302).
Roudet’s view of an association between ‘ideas’ is more plausible than
Ullmann’s ‘contiguity of senses’ according to our conception of the psycho-
logical grounding of metonymy.9
More popular than Roudet’s lucid classification became Jakobson’s (1971)
application of the terms ‘contiguity’ and ‘similarity’ to linguistics (cf. Koch, this
volume). The major point of Jakobson’s paper is the construction of a parallel
between communicative strategies of aphasic speakers and both metaphor and
metonymy. He begins by stating that any linguistic sign can either be ‘com-
bined’ with other signs or be ‘substituted’ by others. The first “mode of
arrangement” is called “combination” and is based on the in praesentia
contiguity of signs in an utterance, the second is named “selection” and is based
on an in absentia similarity of concepts in a paradigm (1971: 74f). He further
states that one type of aphasia (Wernicke and amnesic aphasia) is characterized
by a loss of semantic knowledge, the mode of selection being disturbed. These
speakers often have recourse to syntagmatic strategies, e.g., they give a
paraphrase of the word they cannot find — or create a metonymy. The other
major type of aphasia (Broca) is characterized by a loss of syntactic knowledge.
172 Andreas Blank
In the preceding section, I claimed that the relation upon which metonymies
are based is a conceptual one, and that these concepts are contiguous to each
Co-presence and Succession 173
MORNING NOON
SUGAR PUB FAST
TOAST
DRINK
BUTTER BREAKFAST LUNCH
MILK PEAS
SALT
MARMELADE CHIPS
EGG DISH
KETCHUP
HAM STEAK
Furthermore, this example shows that frames and their contents are
entirely culture-dependent: they not only differ from one linguistic commu-
nity to another but they can also vary even within the same community. An
English breakfast, for example, typically includes bacon and eggs, buttered
toast, etc., whereas in Scotland porridge made of oats is an important part of
breakfast.17
Finally, these two frames show that even strong contiguity between
concepts in a frame does not automatically lead to metonymy.18 Frame-
relations and frame-networks only build the cognitive foundation necessary
for a metonymic transfer. In this case, a salient conceptual relation is high-
lighted (Croft 1993: 348) and the target concept is verbalized by the word
usually related to the concept that serves as donator. As Langacker (1993: 30f)
puts it, the donator concept serves as a ‘reference point’ for accessing the
target. The metonymic innovation may remain a hapax legomenon or lead to
‘lexicalized metonymy’ by semantic change when being adopted by a group
of speakers or the whole speech community.19
Thus, we can say that metonymy is a linguistic device based on salient
conceptual relations within a frame-network. In this frame-network, three
forms of conceptual contiguity are relevant for metonymy:20
1. Relations between concepts within a frame, as in (5):
(5) L praeco ‘messenger’ > OSp pregón, OPt pregão ‘message’
2. Relations between concepts and the superordinate frame, as in (6):
(6) OF travail ‘pain’ > MF ‘work’; ME travail ‘pain’ > ModE travel
‘journey’
3. Or even relations between related frames, as in (7):
(7) OF disner ‘to have the first meal of the day’ > ModF dîner ‘to have
lunch’ > ‘to have dinner.’
life is typically organized and how the ‘things of life’ are interrelated. This
makes metonymies very efficient tools for resolving different tasks in commu-
nication.21
As we have already seen, metonymies can express new concepts which
have not been verbalized before but which are closely related to other con-
cepts in a frame; this was the case in examples (2) and (4). We have also
learned that they can be used for euphemistic, attenuating speech. This caused
the metonymic transfers in (1), (6) and also in (3), where a more prestigious
expression was used in addressing an inferior person in order to please him. In
time, it became impossible for Spanish speakers to address the ‘lower law-
court official’ in any other way than by calling him alguacil. A third possibil-
ity offered by metonymy is to reflect changes in reality and, consequently, in
the conceptual representation of the real world. This was the case in (7).
When, in the 16th century, the upper classes of Paris became used to getting
up late, they shifted the first (and most important) meal of the day from the
middle of the morning to noon. The term dîner, however, was kept. The same
thing happened a century ago, as people increasingly had to stay away from
home all day in order to work and, consequently, moved the most important
meal to the evening.
Metonymy is further used to turn deverbal nouns into more prototypical
nouns, as in:
(8) L prehensio ‘act of seizing someone’ > F prison ‘captivity’ >
‘prison’
(9) F bâtiment ‘act of building’ > ‘building’
(10) L civitas ‘citizenship’ > ‘community of citizens’ > OF cité, I città,
Sp ciudad, Pt cidade ‘town,’ ‘city’
As in the above examples, we often find that names for localities are
created by metonymic processes (cf. Koch 1997).
Finally, we find that quite often two concepts are strongly related in the
speech context so that a single word can accidentally refer to either of them.
The shared speech context enables us to understand utterances like The ham
sandwich is waiting for his check (Lakoff and Johnson 1980: 35), where a
waiter refers to a customer by the dish the customer has ordered. When this
conceptual constellation satisfies the communicative needs of a defined group
of speakers or the whole speech community, the metonymy or the type of
176 Andreas Blank
4. Typologies of metonymy
In this section, I shall try to integrate the theoretical issues on contiguity and
frames, as introduced in Section 3, into a typological approach to metonymy
different from the taxonomies presented above. Instead of listing types of
conceptual relation, the focus will be set on two rather abstract conceptual
frameworks that derive directly from the two fundamental ways of conceptu-
alizing real-life situations, viz. as static frames or as dynamic scenarios.
TIME FRAME
PERIOD FUNCTION
ACTIVITY
STATE INVENTOR
COLLECTIVE ACTOR PRODUCT
BODY AUTHOR
PRODUCER
PLACE
Figure 2. Co-presence
Examples:
(a) ACTOR – OBJECT: L praeco ‘messenger’ > OSp pregón, OPt pregão
‘message’; F courrier ‘postman’ > ‘mail’
(b) INVENTOR – PRODUCT: G Zeppelin ‘proper name’ > ‘long, cylindrical,
dirigible balloon filled with gas’28
(c) ACTOR – TYPICAL ASPECT: OE sælig ‘blessed’ > ModE silly ‘stupid’; OGr
aggelos ‘messenger’ > ‘angel’
(d) OBJECT – TYPICAL ASPECT: L pecunia ‘cattle (as a kind of currency)’ >
‘money’
(e) OBJECT – FUNCTION: E target ‘little round shield’ > ‘object to be aimed at
in shooting’
(f) ACTIVITY – OBJECT: L vestis ‘act of dressing’ > ‘clothes’
(g) ACTIVITY – TIME: It vendemmia ‘gathering of grapes’ > ‘vintage (season)’
(h) ACTIVITY – PLACE: F comédie ‘comedy’ > ‘theater (building)’
(i) STATE – PLACE: OF prison ‘captivity’ > ‘prison’
(j) TYPICAL ASPECT – ACTIVITY: F baiser ‘to kiss’ > ‘to make love’
(k) TYPICAL ASPECT – FRAME: ME travail ‘pain’ > ModE travel ‘journey’
Co-presence and Succession 181
CAUSE RESULT
PURPOSE INSTRUMENT
AIM TOOL
(ACTOR/AUTHOR /
PRODUCER/INVENTOR)
Figure 3. Succession
182 Andreas Blank
Examples:
(a) ACTIVITY – CONSECUTIVE ACTIVITY: L plicare ‘to fold’ > Rum a pleca ‘to
leave,’ Sp llegar ‘to arrive’
(b) ACTIVITY – RESULT: F desfier ‘to break the vassal’s oath’ > ‘to defy’; L
prehensio ‘act of seizing someone’ > F prison ‘captivity’ > ‘prison’; F
bâtiment ‘act of building’ > ‘building’
(c) ACTIVITY – AIM: F pour ‘for (exchange)’ > ‘in order to’
(d) CAUSE – AIM: It perché, Sp porque ‘because’ > ‘in order that’
(e) RESULT – CAUSE: L reus ‘accused’ > Rum rau ‘bad’
(f) PRECONDITION – ACTIVITY: L parare ‘to prepare,’ ‘to arm’ > It parare, Pt
parar ‘to ward off, to repulse’; E to realize ‘to make real’ > ‘to recognize’
(g) MATERIAL – PRODUCT: E horn ‘horn of an animal’ > ‘wind-instrument’
(h) INSTRUMENT – PRODUCT: F plume ‘pen’ > ‘style (of an author)’
(i) PERIOD – PERIOD: Sp. mañana, ME morwe, morrow ‘morning’ > ‘the
following day’
(j) PLACE – PLACE: OL *loukos ‘clearing, open space’ > L lucus ‘grove,’
‘wood’
(k) FRAME – FRAME: OF disner ‘to have the first meal of the day’ > ModF
dîner ‘to have lunch’ > ‘to have dinner’
The two lists of contiguity types are certainly incomplete, as one might always
find metonymies grounded on conceptual contiguities other than those men-
tioned here. Up to now, however, every type of metonymy I have found
belongs to one of the two superordinate domains of conceptual contiguity, as
the underlying concepts are either co-present or successive to each other.29
The typology of metonymy proposed here involves three levels of ab-
straction (cf. Figure 4): two superordinate ‘domains of contiguity’ (co-pres-
ence and succession), a principally open list of ‘types of contiguity’ or
‘contiguity schemas’ (or even ‘conceptual metonymies’; cf. Radden and
Kövecses, this volume), and finally — on the level of the language — concrete
metonymies.
In our everyday life we observe, experience and learn concrete contigui-
ties and we are subsequently learning and using metonymies that result from
this knowledge. For speakers it is important to know which types of contiguity
are salient and which are usually realized by metonymy. This knowledge is
Co-presence and Succession 183
not only necessary to understand existing metonymies but it is very useful for
producing new ones. The rather abstract ‘domains of contiguity’ are less
relevant for speakers than for the purpose of linguistic description. They are
kinds of ‘meta-frames’ which show the same level of abstraction as the
conceptual meta-frames behind some types of word formation.30 Meta-frames
generally highlight a more general aspect underlying and structuring different
sorts of frames, as shown in Figures 2 and 3.31
DOMAINS OF CONTIGUITY
CO-PRESENCE SUCCESSION
CONCRETE METONYMIES
mon to use the contiguity schema AUTHOR – PRODUCT for referring to a cake by
the name of the baker (in a sentence like Mary was delicious).32
The three traditional types of metonymy as mentioned in Section 4
(spatial, temporal and causal relations) are so to speak ‘transverse’ to the two
superordinate types of conceptual contiguity. In both domains we can find
spatial relations (e.g., PART – WHOLE vs. CONTIGUOUS PLACES), temporal rela-
tions (e.g., ACTIVITY – TIME WHEN THE ACTIVITY OCCURS vs. PREVIOUS STATE –
CONSECUTIVE STATE) and causal relations (e.g., ACTIVITY – AFFECTED OBJECT
vs. PROCESS OF PRODUCTION – PRODUCT).
6. Conclusion
In this paper I have tried to show that a theory and typology of metonymy first
of all has to define the cognitive background within which it works. Further-
more it has to be made clear what is understood by ‘contiguity.’ Considering
this, I come to the following conclusions:
1. Metonymy as a linguistic device is the transfer of a word to another
concept on the basis of conceptual contiguity between a donator and a
target concept. Any spontaneous metonymy can be adopted by the speech
community and thus become lexicalized.
2. Conceptual contiguity results from a relation between these concepts
within cognitive frames, between a concept and the frame itself or be-
tween two related frames.
3. The conceptual relations in these frames and between frames are repre-
sented as types of contiguity. Any concrete metonymy is rooted in such a
conventionalized contiguity schema.
4. All conceptual relations relevant to metonymy are either co-present or
successive in time. These two very fundamental aspects in human con-
ceptualization constitute meta-frames which contain typical convention-
alized contiguity schemas. The co-presence and the succession domains
are mentally stored abstractions from the schemas behind actual metony-
mies.
Frame theory and the three-level conception of metonymy make metonymy a
very powerful and efficient tool in communication and semantic change. To
highlight a given concept we only have to apply the appropriate contiguity
schema to the frame in question. In contrast to metaphor, no new relation
Co-presence and Succession 185
needs to be established (by the speaker) and interpreted (by the hearer), as
both frames and contiguity schemas already belong to our knowledge.
Notes
1. I express my gratitude to Mary Copple for the stylistic revision of this paper.
2. A semantic loan of a metonymy in Hebrew (m’l’k).
3. Indeed, most of the newer typologies of metonymy mentioned in this paper integrate
PART-WHOLE, WHOLE-PART and even PART-PART relations, which in rhetoric usually fall
within the bounds of synecdoche (cf. Dumarsais 1818: 115; Lausberg 1960: § 572; Le
Guern 1973; Garcia Arance 1979; Ruwet 1983). In fact synecdoche is a highly problem-
atic category as it comprises contiguity-based types like those already mentioned (e.g., E
soul ‘spiritual part of man’ > ‘human being’), and similarity-based transfers like E bread
‘food made out of flour or meal’ > ‘food,’ where the relation is between the more specific
and the general (cf. Nerlich, forthcoming: Section 5; Seto, this volume).
4. A cognitive foundation of metonymy and other contiguity-based phenomena in the
lexicon akin to mine is discussed in detail in Koch, this volume. Focusing in my
contribution on the cognitive roots of metonymy is nevertheless necessary for a better
comprehension of the typology suggested in Section 5.
5. “denominatio est quae ab rebus propinquis et finitimis trahit orationem, qua possit
intellegi res quae non suo vocabulo sit appellata.” (Rhetorica ad Herennium IV: 32, 43;
my italics).
6. “[...] désignation d’un objet par le nom d’un autre objet qui fait comme lui un tout
absolument à part, mais qui lui doit ou à qui il doit lui-même plus ou moins, ou pour son
existence, ou pour sa manière d’être” (Fontanier 1968: 79). — “Die Metonymie
verwendet [...] ein Wort in der Bedeutung eines anderen Wortes, das semantisch mit dem
verwendeten Wort in einer realen Beziehung steht” (Lausberg 1960: §565; my italics). —
“[...] entità qualsiasi mediante il nome di un’altra entità che stia alla prima come la causa
sta all’effetto e viceversa, oppure che le corrisponda per legami di reciproca dipendenza”
(Mortara Garavelli 1988: 149).
7. Allusion to The Sermon on the Mount (Matthew 5, 3).
8. A fourth category, changes resulting from “rapports associatifs entre les mots,” concerns
semantic change in word formation.
9. The same direction had already been taken some years earlier by Kristoffer Nyrop (1913:
80): “Une chose ne se présente jamais à l’état isolé, elle est toujours accompagnée de
différentes circonstances qui la complètent ou l’expliquent; elle provoque des associa-
tions d’idées, dues à des rapports de ressemblance, de contiguïté, de cause à effet etc.”
10. For semantic disorders in aphasia, cf. Gurd and Marshall (1993); for syntactic disorders,
cf. De Bleser and Bayer (1993); for a contemporary definition of the different types of
aphasia, cf. Huber et al. (1989: 107–132).
186 Andreas Blank
for OSp pregón and OPt pregão. In this case, one of the two meanings tends to be
expressed by other means, as for example by derivation: NSp pregonero, NPt pregoeiro
‘messenger.’
24. Cf. Bredin’s (1984: 45) critique of this kind of typology: “For what is presented [...] is not
a true definition, but an enumeration of instances.”
25. E.g., Nixon bombed Vietnam is, according to Fass, an AGENT FOR INSTRUMENT substitution,
which highlights the agent responsible, while Exxon has raised its prices again is an
INSTRUMENT FOR AGENT substitution.
26. Indeed, the idea of case-role substitution in sentences could suggest that metonymy relies
on syntagmatic contiguity. This is clearly not the case: in sentences such as The buses are
on strike the syntactic position of the subject remains the same. Only the semantic role of
this subject has been replaced by another role taken out of the paradigm.
27. Warren’s typology is a good example of how underestimating the importance of the
underlying associations can lead to confusion in distinguishing different types of me-
tonymy, but also metonymy and metaphor: cf. Warren (1992: 64f): “[...] it is difficult to
see exactly what the difference is between metonyms and metaphors, or indeed between
metonyms and implications.”
28. As proper names do not have meanings, transfers based on this conceptual relation are
simple ‘namings’ and not semantic changes.
29. Take, for example, the ACTOR (INVENTOR, etc.) – PRODUCT relation: at the moment of initial
production it belongs of course to the co-presence domain, but later moves to the
succession domain. As is the case with much in language, it all depends on perspective.
30. Cf. the ‘conceptual models’ of Italian word formation as described in Schwarze (1988);
cf. also Blank (1997a: 96f, and 1998).
31. An example from Schwarze (1988: 438–443) is the ‘Tätigkeitsmodell’ [“model of
activity”], which introduces different types of affixation as being conceptually centered
(if not to say “derived”) around the concept ACTIVITY.
32. Gilles Fauconnier, however, has drawn my attention to the fact that in very specific
contexts — for example in a cookery contest — this metonymy would be easily under-
stood.
References
Anonymous
1954 Ad C. Herennium, de ratione dicendi (Rhetorica ad Herennium). London:
Heinemann.
Andor, József
1985 On the psychological relevance of frames. Quaderni di Semantica 6: 212–
221.
188 Andreas Blank
Aristotle
1958 Kategorien. Lehre vom Satz (Organon I/II). Translated and edited by Eugen
Rolfes. Hamburg: Meiner.
Blank, Andreas
1993 Polysemie und semantische Relationen im Lexikon. In W. Börner, K.
Vogel (eds.), Wortschatz und Fremdsprachenerwerb. Bochum: AKS-
Verlag, 22–56.
1996 Tyson est aux anges – Zur Semantik französischer Funktionsverbgefüge.
Zeitschrift für französische Sprache und Literatur 106: 113–130.
1997a Il senso di una semantica dei prototipi e dei frames: osservazioni deco-
struttive e ricostruttive. In M. Carapezza, D. Gambarara, F. Lo Piparo
(eds.), Linguaggio e cognizione. Rom: Bulzoni, 89–103.
1997b Prinzipien des lexikalischen Bedeutungswandels am Beispiel der roma-
nischen Sprachen. Tübingen: Niemeyer.
1998 Kognitive italienische Wortbildungslehre. Italienische Studien 19: 5-27.
forthcoming Pathways of lexicalization. In M. Haspelmath et al. (eds.), Language
Typology and Universals: An International Handbook. Berlin, New York:
Mouton de Gruyter.
Bonhomme, Marc
1987 Linguistique de la métonymie. Bern, Frankfurt: Lang.
Bredin, Hugh
1984 Metonymy. Poetics Today 5: 45–58.
Coseriu, Eugenio
1958 Sincronía, diacronía e historia: El problema del cambio lingüístico.
Montevideo: Universidad de la República.
Croft, William
1993 The role of domains in the interpretation of metaphors and metonymies.
Cognitive Linguistics 4: 335–370.
De Bleser, Ria, Joseph Bayer
1993 Syntactic disorders in Aphasia. In G. Blanken et al. (eds.), Linguistic
Disorders and Pathologies: An International Handbook. Berlin, New
York: Mouton de Gruyter, 160–169.
Dik, Simon C.
1977 Inductive generalisation in semantic change. In P.J. Hopper (ed.), Studies in
Descriptive and Historical Linguistics. Festschrift for Winfred P. Lehmann.
Amsterdam and Philadelphia: Benjamins, 283–300.
Duchácek, Otto
1967 Précis de sémantique française. Brünn: Univ. J.E. Purkyne.
Dumarsais, César Chesneau
1818 Les Tropes. Edited and commented by Pierre Fontanier. Paris: Belin-le-
Prieur (Reprint Geneva: Slatkine 1967).
Fass, Dan
1991 Metonymy, case role substitution and sense ambiguity. In: D. Fass et al.
(eds.), Proceedings of the IJCAI-91 Workshop on Computational Ap-
proaches to Non-Literal Language (Sydney, Australia). Boulder, CO: Uni-
versity of Boulder, 42–51.
Co-presence and Succession 189
Fauconnier, Gilles
1984 Espaces mentaux: Aspects de la construction du sens dans les langues
naturelles. Paris: Éditions de Minuit.
Fillmore, Charles
1968 The case for case. In E. Bach, R.T. Harms (eds.), Universals in Linguistic
Theory. New York: Holt, Rinehart & Winston, 1–90.
1975 An alternative to checklist theories of meaning. Proceedings of the 1st
Annual Meeting of the Berkeley Linguistics Society. Berkeley: Berkeley
Linguistics Society, 123–131.
1977 Scenes-and-frames semantics. In A. Zampolli (ed.), Linguistic Structures
Processing. Amsterdam and Philadelphia: Benjamins, 55–81.
1985 Frames and the semantics of understanding. Quaderni di Semantica 6: 222–
254.
Fontanier, Pierre
1968 [1830] Les figures du discours. Paris: Flammarion.
Garcia Arance, Maria del Rosario
1979 Semantica de la metonimia y de la sinecdoque. Valladolid: Universidad de
Valladolid.
Gurd, Jennifer M., John C. Marshall
1993 Semantic disorders in aphasia. In G. Blanken et al. (eds.), Linguistic Disor-
ders and Pathologies: An International Handbook. Berlin, New York:
Mouton de Gruyter, 153–160.
Happ, Heinz
1985 ‘Paradigmatisch’ – ‘syntagmatisch’: Zur Bestimmung und Klärung zweier
Grundbegriffe der Sprachwissenschaft. Heidelberg: Winter.
Huber, Walter, Klaus Poeck, Dorothea Weniger
1989 Aphasie. In K. Poeck (ed.), Klinische Neuropsychologie. Second edition.
Stuttgart, New York: Thieme, 89–137.
Jakobson, Roman
1971 [1956] Two aspects of language and two types of aphasic disturbances. In
R. Jakobson, M. Halle, Fundamentals of Language. Second edition. The
Hague, Paris: Mouton, 67–96.
Johnson, Mark
1987 The Body in the Mind: The Bodily Basis of Reason and Imagination.
Chicago: The University of Chicago Press.
Johnson, Samuel
1785 A Dictionary of the English Language: In Which the Words Are Deduced
from Their Originals, and Illustrated in Their Different Significations. 2
vols. Seventh edition. London.
Keller, Rudi
1994 Sprachwandel: Von der unsichtbaren Hand in der Sprache. Tübingen:
Francke.
Koch, Peter
1993 Kyenbé – Tyonbo: Wurzeln kreolischer Lexik. In Ch. Foltys, Th. Kotschi
(eds.), Berliner Romanistische Arbeiten. Für Horst Ochse. Berlin: Freie
Universität, 259–287.
190 Andreas Blank
Louis Goossens
University of Antwerp
1. Introduction
This paper will be concerned with the development of the English modal must
from deontic modality to epistemic modality, more particularly with the
way(s) in which this development came about. This shift is not restricted to
English must, of course, or to English, for that matter. As Bybee et al. (1994:
Ch. 6) have pointed out, the more general shift is one from agent-oriented to
epistemic modality, and it is attested in modals in a large number of languages
across the world. The development can be interpreted as involving:
(a) a metaphorical shift (cf., for example, Sweetser 1990 for a mapping of the
socio-physical domain onto the epistemic domain),
(b) a context-induced interpretation (Heine 1995) or the conventionalization
of an implicature (Traugott 1989).
As Heine has observed, the two interpretations differ in level of abstraction:
(a) is a macro-level view of the change, (b) is to be situated at the micro-level.
In Goossens (1990) and also in Goossens et al. (1995) I have argued that there
is considerable interaction between metonymy and metaphor. The most fre-
quent pattern is what I have called ‘metaphor from metonymy,’ which is
exemplified in (1).
194 Louis Goossens
A B A B
D E D E
2. Research question
With respect to English must, the question arises whether there are sufficient
grounds to consider the development of its epistemic sense as a metaphor from
metonymy. In the following quotation, Traugott and König clearly argue that
the change was induced from contexts which are open to both a deontic and an
epistemic interpretation and which would therefore give it a metonymic basis:
[...] must in the epistemic sense of ‘I conclude that’ derived from the obliga-
tive sense of ‘ought to’ by strengthening of conversational inferences and
subjectification. If I say She must be married in the obligation sense, I invite
the inference that she will indeed get married. This inference is of course
epistemic, pertaining to a state of affairs that is anticipated to be true at some
later time. (1991: 209)
Bybee et al. (1994: 196) generally favor the thesis that the overall change from
agent-oriented to epistemic “involves the conventionalization of an
implicature, by which the inferences that can be made from the meaning of a
particular modal become part of the meaning of that modal.” Also here, in the
terminology adopted, we find acceptance of the metonymic origin. In the
specific case of English must, however, they are opposed to such an origin:
Since the epistemic use of must arises in contexts with aspectual interpreta-
tions distinct from the obligation uses, it appears that metaphor may be at
work in this change. Metaphorical change involves a shift to a different
domain — in this case from the domain of social obligations and physical
necessities applied to an agent, to the epistemic domain that speaks of the
necessary conditions under which a proposition can be true. (Bybee et al.
1994: 201)
The argument focuses on the essential difference between a state of affairs that
combines with deontic must, which is typically dynamic or at least controlled,
whereas for epistemic must we need a ‘State,’ that is, a state of affairs
conceived of as non-dynamic and non-controlled.1 As I have pointed out in
Goossens (1987), the qualification of a controlled/dynamic nuclear predica-
tion by such aspectual notions as Imperfective/Progressive or Perfect gives it a
State-character.
Traugott and König’s example would seem to indicate that there are
contexts which are compatible with both the deontic and the epistemic inter-
pretation. But then their example is clearly a constructed one. It remains to be
demonstrated whether language use actually provides contexts in which the
196 Louis Goossens
two interpretations are simultaneously relevant, and more generally, what the
precise basis for the development may have been. In what follows, I will study
contexts which are relevant to a better understanding of this modal shift in
must, first in contemporary English (Section 3), then in older language stages
(Section 4).
Before we tackle the instances that are directly relevant to our purposes, allow
me to make a number of preliminary observations. For obvious reasons the
undecidable (‘?’) cases do not receive any further discussion. As far as the
deontic/necessity examples are concerned, they outnumber the epistemic ones
in all the samples (but more clearly in the written ones). Among them there were
23 instances of must not in the spoken data, but only two in the written samples
(more specifically in LOB). Of special interest are the four cases of must not
which I have assigned to ‘subjective epistemic,’ because we do not expect
epistemic readings of must to occur with not (cf. Coates 1983: Section 4.1.3).
They are all in the interrogative-negative (three out of four come in tags) and
have the effect of trying to elicit agreement of the addressee with an inference
of the speaker; (2) is a case in point. Neither this instance, however, nor the other
examples of this type, are relevant to the issue that concerns us here.
(2) I like to think of those days # and how tough it was # for the average
Englishman #. what a hard life they must have had # – and mustn’t
there be #. endless stories about this mansion # –3
It is the ‘inferable necessity’ and the ‘transitional’ uses that are our focus.
They will be discussed in the next subsection.
3.2. Bridges
(6) (BROWN)
Rossi and Ferguson have been across the street, talking to the kid.
They’ve found some sort of new evidence, a bundle of clothes or
something, and it must link the kid even stronger to the crime. Why
won’t you accept facts? The two kids were together a lot, they [...]
Both (5) and (6) present an inference for which the evidence is available
contextually, so that the hearer can arrive at the same conclusion as the
speaker — although it remains the case that it is the speaker who draws the
conclusion (since he/she does the talking). Some five percent of the data have
to be accounted for in this way. Although these instances are not metonymic
between deontic and epistemic as the ‘transitional’ cases are, they may never-
theless be considered as stepping stones to the full-scale subjective epistemic
use which is typical of present-day English. The diachronic data to be dis-
cussed in Section 4 will confirm the transitional character of this type of usage
toward a full-scale (subjective) epistemic use of English must.
The following conclusions can be drawn from this brief discussion:
(i) Although the normal understanding of ‘epistemic’ implies that it is a
subjective inference of the speaker (this is, for instance, what is indicated by
‘subjectification’ in the above quotation from Traugott and König), we also
have to consider that inferences may be ‘objective,’ i.e., the inferential ground
is understood to be available to both speaker and hearer, as in instances (5) and
(6). Without claiming that these ‘objective epistemic’ instances provide us
with a metonymic bridge between the deontic and the epistemic usage, I
would like to hypothesize that somehow they must have played a part in the
shift to the full epistemic sense typical of contemporary epistemic must.
(ii) Occasionally an utterance/sentence with must is compatible with both
a deontic and an objective, or subjective, inferential reading. A condition for
such a ‘dual’ interpretation is that the state of affairs represented can be taken
to be both non-controlled or controlled (or at least controllable). This is the
case for remember in instance (4) (something may spontaneously come to
mind, or may be the result of a conscious effort to recall it); the normal
interpretation of be hard in (3) is that it is a State, but in the deontic reading
assigned to it above this is viewed as controllable (the subject will (have to)
treat them in such a way that they become/are hard). Uses like these could be
regarded as the metonymic bridge we have been looking for.
Armed with these synchronic observations let us now move on to some
diachronic data to see whether and how the rise of epistemic must confirms the
200 Louis Goossens
As it happened, the first clear uses of subjective epistemic must were found in
sample Early Modern English 3 (1640–1710). The four instances found are
listed here as (7)–(10).
(7) (CEDIAR3B4 , 457–461) (EMOE3)
That was opposed & spoke against with such vehemency by my L.
Clarendon (her owne Unkle) as putt himm by all preferments,
which must doubtlesse, [{have{] been greate, as could have ben
given him5
(8) (CEOFFIC3, 180–184) (EMOE3)
The manner of doing it, so unsought for and unexpected, must take,
with any good mind, more then the thing itself, and I am sure the
sense must always last with mee.
(9) (CEPRIV3, 1200–1204) (EMOE3)
[...] I did thinke never to imploy him, but he being all your tailors I
have altered my resolution; beleev I must not have a gown again
this seaven yeare this cost soe much; four pound four shillings the
outside cost me ready mony [...]
Metonymic Bridges in Modal Shifts 201
Table 2 gives the distribution of the different uses of must, classifying them into
the categories used for the analysis of the present-day items (deontic/necessity,
inferable necessity, transitional, and (subjective) epistemic). For the Middle
English data we need an additional category ‘wish.’ This use builds on the older
permission sense; it is clearly recessive in Middle English and disappears
completely in the EMOE samples. As such, it will not concern us any further.
A first observation is the absolute predominance of the category Deontic/
Necessity, which is much more pronounced than in present-day English. As
pointed out above, it also includes instances which as such do not express an
obligation. (11) and (12) illustrate this more general necessity use.
(11) (CMGAYTRY, 360–364) (ME4)
[...] for-thynkynge þat we hafe of oure syn, with-owtten will of
thoghte to turne agayne to it. And þis sacrament must haue thre
thynges: — Ane es, sorowe in oure herte þat we hafe synnede.[...]
(12) (CEBOETH2, 304–308) (EMOE2)
[...] All that is so, long must last & holde togither, as it is one, but
must needes perish & decay, whan so it leaves to be; [...]
In addition, note that in ME3, ME4 and EMOE1 instances, in which the
speaker can be assumed to be in authority, or at least strongly associates
himself/herself with some other authority/necessity source, are markedly
lower in frequency than those where this is not the case. Among those that do
not involve speaker authority, must is still regularly used in instances where it
is a past tense (is used in past time main clause contexts), a usage which
decreases as we come closer to present-day English.
The most important observations to be made, however, relate to the other
three categories (inferable necessity, transitional and epistemic). As was
pointed out in the preceding section, the fully (subjective) epistemic usage
does not come in until EMOE3 (four instances). The ‘transitional’ usage is
found from EMOE1 onwards, but is lacking in the ME samples. Inferable
necessity, however, is well established from our earliest sample (ME3) on-
wards, and remains so in all the subsequent ones. It is there that inferential
must finds its starting-point, as I will try to demonstrate in the next section.
Let us first offer a few instances of ‘inferable necessity,’ one from each period.
(13) (CMBOETH, 375–379) (ME3)
But I have wel concluded that blisfulnesse and God ben the
sovereyn good; for which it mote nede be that sovereyne blisfulnes
is sovereyn devynite
Metonymic Bridges in Modal Shifts 203
subject’s, knowledge of the world. In (19) the necessity for them to flee can be
inferred from the absence of fresh water (mentioned in the preceding clause).
In (20) the necessity for the subject to be a beggar follows ‘by good reason’
from the fact that he belongs to successive generations of beggars, as is made
explicit in what precedes. Also (21) and (22) come in a context in which a
reasoning is presented.
General necessity instances like the ones discussed here can therefore be
taken to offer a natural transition between necessity and objective inference/
inferable necessity.
As was pointed out in Section 4.3., instances which are transitional towards
(subjective) epistemic usage do not occur until the EMOE samples, and only
sparingly in EMOE1 and EMOE2. The following instances are representative
of the kind of transition we came across; the first (EMOE1) instance is from an
educational treatise, the second (EMOE2) from The Trial of Sir Walter Raleigh,
and the other two (EMOE3) from Pepys’s Diary and The Earl of Essex’s
Correspondence, respectively. I add a brief discussion to each instance.
(23) (CEEDUC1B, 71–75) (EMOE1)
A childe shall learne of the better of them, that, which an other daie,
if he be wise, and cum to iudgement, he must be faine to vnlearne
againe
Contrary to what the reader may have been led to expect on the basis of Section
4.4., (23) does not build on an inferable necessity, but rather on a general
necessity with a deontic tinge, since the speaker can be assumed to give his
backing to this necessity. Note in this respect that be faine to vnlearne againe
is a process that may come about in an uncontrolled as well as in a controlled
way. At the same time, the example can be interpreted epistemically, but the best
paraphrase in that interpretation is will rather than must. In contemporary
English, the instance with will would be predictive rather than inferential.
(24) (CETRI2B, 73–77) (EMOE2)
Nay, I will prove all: thou art a Monster; thou hast an (^English^)
Face, but a (^Spanish^) Heart. Now you must have Money:
(^Aremberg^) was no sooner in (^England^) (I charge thee (^Ra-
leigh^)) but thou incitedst (^Cobham^) to go unto him, [...]7
206 Louis Goossens
In (24), now you must have Money is part of an argumentation in a trial; now
indicates the next step in this argumentation. Must is used to express the
speaker’s inference (as such it is subjective), but the speaker later goes out of
his way to give the grounds for his inference, which gives the inference an
‘objective’ backing. In other words, ‘inferable necessity’ supports the
speaker’s inference.
(25) (CEDIAR3A, 312–315)(EMOE3)
We were full in discourse of the sad state of our times. And the
horrid shame brought on the King’s service by the just clamours of
the poor seamen. And that we must be undone in a little time.
This is a nice mixture of inferable necessity and epistemic use (again predic-
tive; here the paraphrase is would rather than will because this is reported
speech or, more specifically, erlebte Rede ‘semi-indirect speech’). A deontic/
general necessity reading is out of the question here.
(26) (CEOFFIC3, 735–739) (EMOE3)
[...] although in a matter so nice as this yr Excellency must needs
bee ye best Judge. I believe ye Parliament is like to sitt longer [...]
On the one hand, this may express the speaker’s (subjective) judgment, but
can also be interpreted as expressing a more objective necessity (in line with
the presence of the adverb needs). On the other hand we can take it as
expressing the speaker’s advice, which gives it a (speaker-backed) deontic
overtone.
Summing up, these transitional uses confirm that the epistemic usage of
must typically builds on inferable necessity must: this is illustrated by (24),
(25) and, be it less obviously, in (26). However, there are transitional cases
which exhibit other features. Besides the subjective epistemic ingredient, a
deontic interpretation proved possible in (23) and perhaps also in (26). Note
also that in three out of four instances, must is rather prospective and para-
phrasable in contemporary English by will/would.
The conclusion at this point seems to be that the rise of epistemic must is
rooted in the ‘inferable necessity’ uses, but that complexes of deontic and
epistemic uses may have played a supportive role. In the next section we add
another factor that appears to have been operative to complete this picture.
Metonymic Bridges in Modal Shifts 207
4.7. Subjectification
A factor signaled by Traugott and König (cf. the quotation in Section 2.) is
subjectification. For Langacker this is a crucial ingredient in the semantic
make-up of the present-day modal auxiliaries of English; it is this
subjectification that turns them into ‘grounding predications’ (Langacker
1990 and 1991: 6.3).8 Although a detailed account of how must develops into
a subjectified item is outside the scope of this paper, I would like to confirm
the basic correctness of the claim that subjectification played a part in the rise
of epistemic must. More particularly, it can be argued that it is subjectification
in the deontic area which paved the way for epistemic must, which, as we have
defined the category, is necessarily subjectified (it reflects the speaker’s
subjective inference).
An initial analysis of the deontic/necessity uses in our data revealed that,
although speaker-backed deontic uses are already firmly established in the
ME samples, it is not until we reach EMOE1 that those subjectified deontic
instances begin to outnumber the non-subjectified ones in this category, and
that in EMOE3, where the first clear instances of epistemic must occur, the
ratio is roughly 75% (for the speaker-backed deontic uses) as opposed to 25%
(for the ‘not-speaker-backed’ deontic/necessity instances). There is no doubt,
it seems to me, that the increase and, finally, the predominance of subjectified
uses in the deontic area, made the rise of an essentially subjectified epistemic/
inferential must possible.
5. Conclusion
(i) The rise of epistemic must took place against the background of a gradual
meaning development of must from a general necessity sense to the expression
of inferable necessity, which amounts to an objective, non-defeasible infer-
ence. Early instances which are transitional to epistemic must still show
elements of an objective inference.
On the other hand, another type of transition is found in instances where a
deontic and an inferential interpretation are possible simultaneously. In those
cases, however, must is prospective and parapharasable by present-day En-
glish will, rather than must. Although they must have contributed to the rise of
epistemic must and indicate that we cannot explain everything in terms of a
208 Louis Goossens
single pattern of contextual shift, they are less essential for our understanding
of how present-day inferential must developed. What is crucial, however, is
the marked increase in subjectification in the deontic senses. This was an
essential factor in the development of subjectified inferential/epistemic must.
(ii) With respect to the research question formulated in Section 2, it is
clear that our empirical findings speak strongly against a purely metaphorical
shift hypothesis. The shifts that we have observed are minimal and very
gradual. There is no evidence that there was at any given point in the develop-
ment of English must a single conceptual shift whereby an element from the
sociophysical domain was mapped onto the epistemic domain.
(iii) More importantly, at least from the point of view of a collective
volume whose focus is on the workings of metonymy, the view that must
extended its meaning by way of contextualizations in which the deontic and
the epistemic sense are intertwined (in other words, the view that the change
from deontic to epistemic must is a case of conceptual metonymy) cannot be
upheld either. The concatenation that we have witnessed is one that involves
shifts that do not reach a cognitive salience which can be interpreted as the
conceptualization of one ‘element’ in terms of another within the same (com-
plex) domain. Moreover, more than one type of minimal shift appears to have
contributed simultaneously.
(iv) Finally, we have emphasized the important contribution of a more
global shift, namely that of subjectification in the deontic area. This is a
conceptualization and grammaticalization operation which eventually af-
fected the inferential domain that had become part of the semantic make-up of
must. This process of subjectification is in need of further investigation, but
beyond the scope of what I set out to clarify in this paper.9
Notes
1. My terminology for the differentiation into states of affairs is in line with that of
Functional Grammar (cf. Dik 1989).
2. The samples were collected randomly by Ludo Lejeune. For details, cf. Lejeune (1995
and 1996).
3. The quotations from LOLU are presented in a somewhat simplified form. The diacritic
signs retained are #, which marks intonation units, * *, which signals overlap, and those
that indicate pauses of varying lengths.
Metonymic Bridges in Modal Shifts 209
4. The abbreviations for the different texts are taken over from the Helsinki Corpus. The
sources, in the order of their first occurrence in the examples quoted as (7)–(26), are as
follows: CEDIAR3B = Evelyn, The Diary of John Evelyn; CEOFFIC3 = Letters, Non-
private; CEPRIV = Letters, Private; CETRAV3B = Fryer, A New Account of East India;
CMGAYTRY = Gaytridge, Dan Jon Gaytridge’s Sermon; CEBOETH2 = Elizabeth,
Boethius; CMBOETH = Chaucer, Boethius; CMCAXPRO = Caxton, The Prologues and
Epilogues; CESCIE1B = Record, the Path-way … of Geometrie; CEEDUC2A = Brinsley,
Ludus Literarius or the Grammar Schoole; CEHAND3B = Langford, Plain and Full
Instructions to Raise All Sorts of Fruit-trees; CMCTPROS = Chaucer, The Parson’s Tale;
CMSIEGE = The Siege of Jerusalem in Prose; CEFICT1B = Harman, A Caveat … for
Commen Cursetors; CESERM3A = Tillotson, Sermons; CEEDUC1B = Ascham, the
Scholemaster; CETRI2B = The Trial of the Earl of Essex; CEDIAR3A = The Diary of
Samuel Pepys. For further details, see Kytö (1993).
5. The notation [{.....}] indicates an italicized emendation in the source text. Note that the
text makes sense with and without the emendation; without the emendation we have to
read been as the equivalent of Modern English be.
6. Again in the terminology of Functional Grammar, i.e., dynamic, but not controlled. Note
that in (16) the Process character is the result of passivization.
7. The notation (^.....^) in the Helsinki Corpus indicates a typographical shift from the basic
font in the source text.
8. For a critical discussion with respect to contemporary must, cf. Goossens (1996: 2.3.3)
9. This process of subjectification is in focus in Goossens, forthcoming.
References
Olaf Jäkel
University of Halle
1. Introduction
How are surnames generated in German and in other languages? Are there
principles or patterns after all in the realm of anthroponymy? Or can we at best
reconstruct the etymology of each single name, finally to establish overall
arbitrariness? Like all proper names, surnames belong to a special class of
linguistic expressions that is in marked contrast to appellatives. Unlike these,
their semantic function lies not in the categorization of objects or in descrip-
tive conceptualization, but in the naming and identifying of individuals.
With several people bearing the same first name in one community,
unambiguous reference to a certain person became a problem. The solution
was found in giving by-names, e.g., ‘James son of Zebedee,’ ‘John the
Baptist,’ ‘Richard Lionheart,’ ‘Erik the Red,’ or ‘Thomas Aquinas.’1 In by-
names like these lies the origin of surnames as we know them, as Adolf Bach
confirms: “Our surnames are by-names inherited from ancestors” (Bach 1952:
231; cf. Reaney 1967: 19f).2 The custom of hereditary surnames did not
spread over Germany until the twelfth century, taking almost five hundred
years until complete proliferation: “The custom of bearing a surname was not
generally established in Germany until the year 1600” (Gottschald 1982: 47).
In the field of onomastics, the notion of semantic motivation has always
been more than a mere working hypothesis. It is simply regarded as a “fact that
proper names, at least at the time of origin, are more or less motivated”
(Naumann 1987: 16). Any original motivation, though, may later vanish from
the conscious knowledge of the linguistic community.3 This is why, from
212 Olaf Jäkel
today’s point of view, most onomasticians see proper names as having etymo-
logical meaning rather than lexical meaning.4 It is this etymological meaning
of surnames that we will be dealing with in the following account of various
naming principles.
All surnames used as examples in this paper have been taken from the
Hamburg Telephone Directory of 1993/94 (Deutsche Bundespost Telekom
1992), which served as corpus material in this investigation. After a short
glance at surnames motivated by ‘genealogy’ or ‘profession’ (Section 2) we
will turn to metonymic strategies of surnaming after certain ‘utensils’ or
‘qualities’ (Section 3). Section 4 focuses on surnaming motivated by ‘location
metonymy,’ which shows a rich structure of subtypes that can be analyzed
profitably from a cognitive linguistic perspective. Section 5 presents a sum-
mary of the suggested taxonomy of motivated surnames. As a final point, the
evidence of this onomastic investigation is brought to bear on the controver-
sial discussion of the role of metonymy currently taking place among cogni-
tive linguists.
2. Motivated surnames
3. Metonymic strategies
reasons, and with more or less friendly intentions. But what if the bearer
acquired his surname because he worked as a fishmonger, or because he was a
famous fisherman, or because he was a notorious lover of smoked herring, or
because he once had an accident or a unique encounter in which a herring
played a special part? Then, though in most cases the precise motivation might
be difficult or impossible to confirm (cf. Naumann 1987: 28; Bach 1952: 284;
and Dorward 1995: x), we would be fully justified in calling this metonymy
proper, namely of the type IMPORTANT UTENSIL FOR PERSON.
But the precise motivation of naming can also be more transparent. In the
following examples, adjectives have been converted into surnames:
(9) Klein ‘short,’ Groß ‘tall,’ Langer ‘long’ + deriv. morph. ‘person,’
Starke ‘strong,’ Schön ‘pretty,’ Treu ‘faithful,’ Alt ‘old,’ Jung
‘young,’ Schwarz ‘black,’ Braun ‘brown,’ Roth ‘red,’ Lustig
‘merry,’ Wunderlich ‘strange’
Most of these cases exemplify the metonymic mapping SALIENT QUALITY FOR
PERSON. The same sort of adjectives can also enter into compounds with the
morpheme -mann ‘man’:
(10) Dickmann ‘fat man,’ Hartmann ‘hard man,’ Klugmann ‘wise man,’
Bangemann ‘scared man’
Furthermore, salient qualities of the original name-bearer can even be ex-
pressed by regular synecdoche (11), or hyperbolically (12):
(11) Kahlkopf ‘bald head,’ Weißhaar ‘white hair,’ Langnese ‘long nose’
(Low German), Krummbein ‘crooked leg’
(12) Riese ‘giant,’ Wicht ‘midget’
Examples such as those in (11) name a salient part of the body (like ‘bald
head,’ ‘white hair,’ ‘long nose,’ ‘crooked leg’) that stands for a person as pars
pro toto. The hyperbolic terms (‘giant,’ ‘midget’) in examples (12) obviously
refer to the height of the original name-bearer.
4. Location metonymy
With utensil metonymy and sometimes also with quality metonymy, we are on
fairly slippery ground as far as the particular details of motivational diagnosis
216 Olaf Jäkel
Compounds
Attribute/ Head
Modification
(Geographical) Landmarks
natural man-made
– Landscapes/areas – Farmland
– Lakes and rivers – Fords and dikes
– Woods and copses – Buildings
– Settlements
Directional prepositions are also represented in the corpus, with zu ‘to, to-
wards’ (22) probably defining a place of residence ‘on the way to’ the
landmark in question, and von ‘from’ (23) in contrast referring to the place of
origin of the name-bearer (cf. the discussion in Section 4.5.):
(22) a. Zumsande ‘towards the sand’
b. Zumwalde ‘towards the wood,’ Zumholz ‘towards the copse,’
Zumbusch ‘towards the bush’
c. Zum Felde ‘towards the field’
d. Zurmöhle ‘towards the mill’ (dial.)
(23) a. von der Heide ‘from the heath’
b. von der Lippe ‘from the Lippe’ (river name)
Um ‘around’ is rare (24), probably because this preposition defines only a
very imprecise location. By contrast, außen ‘outside’ is quite frequent as a
modifier in compound surnames (25).
(24) Umland ‘around land/surrounding land’
(25) a. Außendorf ‘outside/outer village,’ Utendörfer ‘outer villages’
(dial.)
b. Utermöhle ‘outer mill’ (dial.)
c. Butenhoff ‘outside farm/yard’ (dial.)
d. Butendeich ‘outside the dyke’ (dial.)
The last example (25d) in particular goes to show how important the in/out-
orientation in relation to certain landmarks was at the time of surnaming. In
settlements owing their existence to huge dikes around low-lying land prone
to flooding, the unprotected area ‘outside the dike’ was the place of residence
of the underprivileged. Thus the surname Butendeich ‘outside the dike’ is
quite frequent in the southern parts of Hamburg by the river Elbe and espe-
cially on the island of Finkenwerder.
As mentioned above, the distinction between rather natural and rather man-
made landmarks should not be overrated. The only relevant criterion for
location-metonymic surnaming is that the locations should be proper land-
marks, i.e., distinctive, fairly unmistakable points prominent in the landscape.
As subgroups of man-made landmarks we find farmland, certain paths, fords
and dikes, buildings, and larger settlements (cf. Figure 2).
Farmland, which is clearly demarcated, can serve better than the open
plain (see above, examples 30d and 30e) to pinpoint residents. Quite frequent
as heads of surnames are fields (33a, b), pastures (33c), and gardens (33d):
(33) a. Grothefeld ‘large field’ (dial.), Grünfeld ‘green field,’ Sommer-
feld ‘summer field,’ Wüstefeld ‘desert field,’ Strohfeld ‘straw
field’
b. Langacker ‘long field,’ Übelacker ‘bad field’
222 Olaf Jäkel
of naming inhabitants who had moved there after their origin” (Gottschald
1982: 48; cf. Smith 1950: 47; and Reaney 1967: 36ff). Thus, incomers could
be named after their place of birth, but also after their native region or county,
and even country or nationality.
From a distance, all these supraregional landmarks are distinctive enough
to serve as surnames without further specification. This is why there are hardly
any compounds in this group. However, the development of town names to
proper and lasting surnames usually included a phase in which they were
preceded by the preposition von ‘of, from’ (37a) (see also example (23)),
which in most cases was dropped later (37b) (cf. Reaney 1958: xiv f).
Furthermore, the derivational morpheme -er ‘person‘ could be suffixed to a
town name to create a surname for an incomer (37c).
(37) a. von Coelln
b. Berlin, Utrecht, Bern, Buxtehude
c. Bremer, Achner
While in German surnames comprising regional associations or counties are
quite common (38), reference to countries or nationalities (39) is rarer. Pre-
sumably this is because during the centuries of active surnaming, international
influx into the German provinces was low.
(38) Friese ‘Frisian,’ Bayer ‘Bavarian,’ Preuß ‘Prussian,’ Böhme ‘Bo-
hemian,’ Hesse ‘Hessian’
(39) Deutscher ‘German,’ Holländer ‘Dutch,’ Norweger ‘Norwegian’
(40) Ausländer ‘foreigner,’ lit.: ‘out-land-er,’ Undeutsch ‘un-German’
In this connection, the examples under (40) certainly are a reflection of the
mentality of the name-givers. If somebody is named Ausländer ‘foreigner’ or
even Undeutsch ‘un-German,’ the intention — then as now — is clearly not so
much to characterize the name-bearer precisely, but to brand him as excluded
from the community.10
The final example (41) provides confirmation ex negativo of the impor-
tance of location metonymy for surnaming:
(41) Wanderer ‘wanderer’
Here it is exactly the ‘non-locatability’ of the person named which to the
name-givers seemed significant enough to identify and distinguish him from
more settled first-namesakes.
224 Olaf Jäkel
Surnaming
unmotivated motivated
(motivation not
recovered)
genealogical professional metonymic
To a great extent, our results are in line with the vast body of expert
knowledge represented in the standard onomastic literature. A slight diver-
gence in classification results from the fact that nowhere in the standard
literature is metonymy recognized as an important motivating principle in
surnaming.
The classification that has become the standard for German onomastics
comes from Adolf Bach (1952), who distinguishes surnames after ‘first
names’ (Rufnamen), ‘place of origin’ (Herkunft), ‘place of residence’ (Wohn-
stätte), ‘occupation and status’ (Beruf und Stand), and ‘nicknames’
(Übernamen). Among English onomasticians, the subdivision of surnames
into the four classes of ‘local,’ ‘occupational,’ ‘patronymical’ (or ‘relational’)
and ‘descriptive’ (or ‘nicknames’) is consensual since the work of Elsdon C.
Smith (Smith 1950: 44; cf. Reaney 1958: xii and 1967: 20; Cottle 1978: 9; and
Dorward 1995: vi).11 What is correct in these traditional classifications is
preserved in our suggested taxonomy of surnaming strategies, which also has
some additional advantages.
The incorporation of metonymy as a motivating principle in surnaming will
certainly not revolutionize onomastics. Yet it seems desirable to differentiate
between utensil metonymy, whose analysis remains speculative, and the much
clearer quality metonymy. In traditional onomastics, surnames of both patterns
are jumbled up in that huge and ill-defined residual category of ‘nicknames,’
which is usually left completely unanalyzed. Another nontrivial result is the
finding that surnaming after place of residence and after place of origin is based
on the same location-metonymic principle. Our discussion has shown the
immense productivity of this principle in the generation of surnames.
The surnaming patterns investigated here on the basis of German names
also account for English surnames (this was shown in Jäkel 1996). In fact,
English and German surnames do not only follow the same motivational
patterns, but they even share most of the morphological details discussed
above. Our taxonomy of naming strategies was even found to apply to Non-
Indo-European languages like Hungarian and Japanese.12
As regards the relevance of the results of our investigation for the
controversial discussion of the general role of metonymy currently taking
place among cognitive linguists, our claims are modest. With our results we
have supplied firm evidence to corroborate the view that metonymy — in
contrast to metaphor — is basically a ‘naming strategy.’ To distinguish
metonymy from metaphor, which has primarily an ‘explanatory function’ (see
226 Olaf Jäkel
Jäkel 1997), Lakoff and Johnson (1980: 36) state: “Metonymy, on the other
hand, has primarily a referential function, that is, it allows us to use one entity
to stand for another.”13 And that this referential function really is about the
‘naming of individuals’ can be confirmed by two quotes from Lakoff (1987).
There, metonymy is defined as follows: “With respect to naming, A stands for
B” (Lakoff 1987: 19). And in the context of his theory of ‘idealized cognitive
models’ Lakoff (1987: 85) explains: “Most metonymic models are, in fact, not
models of categories; they are models of individuals.”
Yet again, this should not be misunderstood. Metonymic structures can
definitely have cognitive status (this point is also made by Lakoff 1987: 90).
They just do not have it as strategies that enhance conceptual understanding or
offer conceptual explanations, but as principles for the naming of individuals.
Humans obviously have a strong cognitive need for distinctive yet economical
naming. Nowhere can this be seen more clearly than in the supposedly
antiquated domain of onomastic research. And that metonymic patterns of
various kinds are at work in exactly this of all domains is no coincidence: just
because of its general naming function there is a vast field of operation for
metonymy in onomastics.
Notes
1. It will be seen later that these by-names anticipate all known strategies of surnaming.
2. This and all remaining quotations from German sources are given in translation by the
author.
3. Cf. Fleischer (1962: 30): “At the moment of a name’s origin the relation is present, later
to congeal.”
4. This is stated explicitly by Debus (1966: 14) and at least implicitly held by many
onomasticians; for example, see Bach (1952: 206ff), Fleischer (1962: 30ff), Witkowski
(1964: 51), Cottle (1978: 10ff), and Dorward (1995: xvi). Cf. also Lyons (1977: 223).
5. See Matthews (1966: 71): “But there always remain at least three or four per cent about
which the honest investigator can only state, ‘I don’t know’.” Another problem which
cannot be considered here is the possibility of alternative reconstructions for one and the
same name. For a treatment of this issue see e.g., Bach (1952: 235) and Matthews (1966:
71f).
6. Note that in the Scandinavian tradition the surname does not carry the same weight as in
German. Its character is still more like that of a by-name. Concerning Icelandic in
particular, cf. Wiktorin (1991: 57): “Unlike our common practice, Icelandic gives more
prominence to the first-name. [...] It is only logical that in the Icelandic telephone
directory entries are arranged alphabetically according to first-names.”
Metonymy in Onomastics 227
7. Of course the spelling of these surnames may vary. Thus, Schmidt is more frequent than
Schmied, Böttger is found as well as Böttcher, and Köster alongside Küster, to name but
a few examples.
8. See Wimmer (1973: 27f), who analyses this development as the transition from an
appellative use of Müller to a proper name Müller.
9. That professional surnaming should not be counted among metonymic strategies is
confirmed by Norrick (1981). Although he proposes no less than 17 different metonymic
principles, he deems no special metonymic principle necessary if occupation stands for
person: “[...] the theory of reference [...] predicts that one can refer to a person by
identifying his role and conversely” (Norrick 1981: 100). This distinguishes professional
naming from those strategies discussed below as truly metonymic. However, other
authors who subscribe to a more extensive notion of metonymy (e.g., some of the
contributions to this volume) would notch up even professional and genealogical naming
patterns under metonymic strategies.
10. For a synchronic analysis of the concept ‘Ausländer’ and its current motivation see Jäkel
(1993).
11. Matthews (1966: 69) confirms this consensus: “[...] nearly all writers on the subject of
surnames have classified them into the four types of Locality, Relationship, Occupation
and Nicknames.” And Smith (1950: 45) even pronounces the universal status of this four-
part classification: “In general, surnames in all countries originate in one of the above
four ways if they are not consciously adopted.”
12. I thank Zoltán Kövecses for providing information on Hungarian surnames, and Ken-Ichi
Seto for his detailed explanation of Japanese surnames. According to his estimate, no less
than 90 % of Japanese surnames are of the location-metonymic type, which is much more
than in German or English.
13. But cf. Taylor (1989: 122ff), who seems to be taking the other side, though without
voicing any convincing examples or arguments to support his view. For more controver-
sial positions on the status of metonymy, see the other contributions in this volume.
References
Bach, Adolf
1952 Deutsche Namenkunde. Bd. I: Die deutschen Personennamen. Second
edition. Heidelberg: Winter.
Cottle, Basil
1978 The Penguin Dictionary of Surnames. Second edition. Harmondsworth:
Penguin.
Debus, Friedhelm
1966 Aspekte zum Verhältnis Name – Wort. Groningen: Wolters.
Deutsche Bundespost Telekom (eds.)
1992 Amtliches Telefonbuch der Deutschen Bundespost Telekom, Ortsnetzbereich
Hamburg, 1993/94 edition. Frankfurt a.M.
228 Olaf Jäkel
Dorward, David
1995 Scottish Surnames. Glasgow: Harper Collins.
Fleischer, Wolfgang
1962 Zur Frage der Namenfelder. In W. Fleischer (ed.), Name und Text.
Tübingen: Niemeyer, 25–42.
Gottschald, Max
1982 Deutsche Namenkunde: Unsere Familiennamen nach ihrer Entstehung und
Bedeutung. Fifth edition. Berlin, New York: Mouton de Gruyter.
Jäkel, Olaf
1993 Wer oder was ist eigentlich ein Ausländer? Anmerkungen zu einer
seltsamen Kategorie. In Pressestelle der Universität Hamburg (eds.), Uni hh
2: 44.
1996 Metonymy as a cognitive principle in onomastics. Paper presented at the
XIXth International Congress of Onomastic Sciences, University of Aber-
deen, Scotland, August 4–11, 1996.
1997 Metaphern in abstrakten Diskurs-Domänen: Eine kognitiv-linguistische
Untersuchung anhand der Bereiche Geistestätigkeit, Wirtschaft und
Wissenschaft. Frankfurt a.M.: Peter Lang.
Lakoff, George
1987 Women, Fire, and Dangerous Things: What Categories Reveal about the
Mind. Chicago: The University of Chicago Press.
Lakoff, George, Mark Johnson
1980 Metaphors We Live By. Chicago: The University of Chicago Press.
Lyons, John
1977 Semantics. Volume I. Cambridge: Cambridge University Press.
Matthews, Constance M.
1966 English Surnames. London: Weidenfeld and Nicolson.
Naumann, Horst (ed.)
1987 Familiennamenbuch. Leipzig: Bibliographisches Institut.
Norrick, Neal R.
1981 Semiotic Principles in Semantic Theory. Amsterdam and Philadelphia:
Benjamins.
Reaney, Percy H.
1958 A Dictionary of British Surnames. London: Routledge and Kegan Paul.
1967 The Origin of English Surnames. London: Routledge and Kegan Paul.
Smith, Elsdon C.
1950 The Story of Our Names. New York: Harper.
Taylor, John R.
1989 Linguistic Categorization: Prototypes in Linguistic Theory. Oxford:
Clarendon.
Wiktorin, Karl
1991 Island erfahren: Reiseinformationen. Eichstätt: Lundipress.
Wimmer, Rainer
1973 Der Eigenname im Deutschen: Ein Beitrag zu seiner linguistischen
Beschreibung. Tübingen: Niemeyer.
Metonymy in Onomastics 229
Witkowski, Teodolius
1964 Grundbegriffe der Namenkunde. Berlin: Akademie-Verlag.
Part III
Richard Waltereit
University of Tübingen
1. Introduction
cally. Grammatically, the only possible reading is that the real Ringo hits the
statue. One has to consider the relation between the person (Ringo) and his
physical representation (the statue) a metonymic one. Now, the possibility of
metonymic reference relation is clearly dependent on the grammatical context,
as (1) shows. The nature and the form of this dependency are far from obvious.
In this paper, I hope to shed some light on these questions, but it will of course
be impossible to study the issue exhaustively.
As for the theory of metonymy, I assume that metonymy is a device for
reference that exploits a contiguity relation between two entities (cf. Koch,
this volume; as for non-referential approaches to metonymy, cf., e.g., Radden
and Kövecses, this volume; Feyaerts, this volume; Panther and Thornburg,
this volume; and Pankhurst, this volume). Contiguity can be a relation of
spatial vicinity, a PART-WHOLE relation, a CONTAINER-CONTENTS relation, an
AGENT-ACTION relation, etc. It seems less important to try to enumerate the
possible types of contiguity than to acknowledge that contiguity is a relation
of experiential ‘togetherness,’ where experience is to be understood in the
broadest sense. Given this assumption, contiguity can take virtually any form,
provided speakers construe a relation between the entities involved and take
the relation as communicatively relevant (Nunberg 1995). To the extent that
the respective contiguity belongs to a relatively stable relation, it can be
addressed as part of a frame (taking this notion as a cover term for what has
been called ‘domain,’ ‘script,’ ‘scene,’ etc.) (cf. Koch, this volume).2
In order to clarify the notions that will be used throughout this paper, a basic
distinction with respect to metonymic relations in verbal semantics has to be
introduced. Consider (2) and (3):
(2) a. La soupe aux poissons n’a pas donné de pourboire.
‘The fish soup didn’t leave a tip.’
b. Michel n’a pas donné de pourboire.
‘Michel didn’t leave a tip.’
(3) a. Ils ont enfin servi la soupe aux poissons.
‘They finally served the fish soup.’
b. Ils ont enfin servi Michel.
‘They finally served Michel.’
Grammatical Constraints on Metonymic Reference 235
In (2a) we deal with a metonymy of the classical type: the subject noun phrase
does not actually refer to the fish soup, but of course to the guest who ordered
it. We grasp this immediately because the predicate n’a pas donné de
pourboire normally requires a human subject (as is the case in (2b)). The
selection restrictions of the predicate are therefore violated in (2a). In order to
get an appropriate referent for the actual subject the hearer has to search for
some entity that meets the ‘human’ requirement and that is at the same time
semantically contiguous with the actual subject noun phrase. There is clearly a
contiguity of the ‘guest’ and the ‘ordered dish’ in restaurant communication
that makes it easy to choose the guest as the appropriate referent. That is the
way metonymies usually work: the given noun phrase violates the verb’s
selection restrictions, and this violation has to be accommodated by contigu-
ity-based reference. The selection restrictions of the predicate tell the hearer
what kind of linguistic expressions would ordinarily be expected in the given
syntactic slot; the contiguity relation ensures that the expression will neverthe-
less be correctly understood. I will refer to this as a metonymy on the
‘insertional level’ (because here we are dealing with lexical insertions into
subcategorization frames).
Now, things are quite different in (3). The verb servir is polysemous, as is
its English counterpart to serve. It allows two different thematic roles in the
direct object position, or rather two different classes of objects: the ‘served
dish’ and the ‘served person.’ These are two semantically contiguous roles,
and their relation is the same as in (2). But this time, there is no metonymic
reference. It is really the dish and really the person, respectively, that are
served. In order to understand (3a) and (3b) it is not necessary to activate a
metonymic reading. Rather, the contiguity relation of the served dish and the
served person is lexically encoded. The two different thematic roles are part of
the verb’s lexical content. I will refer to this phenomenon as contiguity on the
‘role level.’ The verb permits two different thematic roles in the direct object
position without formally indicating this variation. This state of affairs entitles
us to regard the verb as polysemous. Polysemy means that a lexical item has
two separate but related meanings compatible with the same surface manifes-
tation. Furthermore, a polysemy is always a form of lexical ambiguity that is
not predictable from grammatical or other rules. A polysemy of a word is
always an individual, ‘accidental’ phenomenon; ultimately, it is the result of a
diachronic semantic change (this was already noticed by Bréal 1897).3
This leads us to the next crucial assumption: role-level contiguities arise
236 Richard Waltereit
Regularities of a slightly different kind are found with the locative and
‘swarm’-alternations that have received a lot of attention especially in the
generative literature (cf. Fillmore 1968; Anderson 1971; Baker 1988; Olsen
1994). These can always be reduced to some kind of ‘local’ relation. Locative
alternations are of the type illustrated in (11) and (12):
(11) a. Ils ont chargé du charbon sur le bateau.
‘They have loaded coal onto the ship.’
b. Ils ont chargé le bateau de charbon.
‘They have loaded the ship with coal.’
(12) a. Il faut substituer ce chiffre par un autre.
‘You have to replace this number with a different one.’
b. Il faut substituer ce chiffre à un autre.
‘You have to put this number into another number’s place.’
What is alternating in these alternations? The object position sustains two
different semantic roles. These two roles are clearly contiguous to each other,
they establish a (be it metaphorical) CONTAINER-CONTENTS relation. This is
obvious in (11), where the ship and the coal are actually regarded as a
container and its contents. It is less obvious but nevertheless plausible in (12).
The verb substituer ‘to substitute’ evokes the idea of a base from which the
substituted thing is detached as a profile (in the sense of Langacker 1987: 183–
189). The highlighted profile is part of the base: one can only substitute parts
in wholes. The PART-WHOLE relation is in turn a relation of ‘topological
inclusion,’ a CONTAINER-CONTENTS relation (Winston et al. 1987: 431–435).
Note that in these cases it is only possible to speak of polysemy with
respect to the direct object position. Only in the direct object position are there
alternating contiguous roles without any surface indication of the alternation.
The oblique objects (sur le bateau ‘onto the ship,’ de charbon ‘with coal,’
etc.) present alternating roles that are in a relation of contiguity, too, but
formally indicate the semantic variation by a varying preposition. The formal
marking of the oblique object prevents these from being semantically ambigu-
240 Richard Waltereit
The so-called ‘swarm’-alternation owes its name to the famous example Bees
are swarming in the garden/The garden is swarming with bees. The point of
this alternation is that two thematic roles can be coded either in the subject
position or in a prepositional phrase. Interestingly, the contiguity relation
between the two roles is usually a CONTAINER-CONTENTS relation:
(13) a. L’eau déborde de la baignoire.
‘The water is flowing over the bathtub.’
b. La baignoire déborde d’eau.
‘The bathtub is overflowing with water.’
In (13) there is clearly a contiguity between the contents, the water, and the
container, the bathtub.
(14) a. La pluie ruisselait sur les murs.
‘The rain was streaming down the walls.’
b. Il ruisselait de cold-cream, de sueur et de vin. (Flaubert)
‘He was streaming with cold cream, sweat and wine.’
The CONTAINER-CONTENTS relation is less obvious here, but at any rate we are
dealing with a relation of local contact with some surface. The ‘container’ is,
so to speak, ‘flat.’
With both verbs, the subject can express both the contents and the
container. This flexibility has to be explained by polysemy. For a monose-
mous interpretation to be possible, we would have to find one label that would
appropriately characterize the selection restrictions of the subject. But this
Grammatical Constraints on Metonymic Reference 241
One might think that there is no use looking for syntactic restrictions in the
area of insertional metonymies, because any constituent can in principle be
affected by a metonymic shift. This is true, of course. One would be surprised
to find, for example, that a dative phrase cannot be metonymic. And, of
course, I will not put forward such a claim. There are, however, plausible
arguments in support of the hypothesis that also in this domain the direct
object has certain privileges. Certain ‘marked’ syntactic constructions that
inherently involve contiguity relations allow metonymic reference for the
direct object but less naturally or not at all for other arguments. A construc-
tion, in this sense, is a specific arrangement of syntactic ‘slots’ which can be
filled from an open set of lexical items. It may be restricted to a certain class of
lexical items but its properties are not entirely deducible from that class. The
construction types involved here include reflexive cliticization and some types
of inalienable possession constructions.
Reflexive cliticization, with the reflexive clitic pronoun being a direct object,
is a construction that is particularly suitable for metonymies. In (17) and (18) I
present some examples where the reflexive clitic allows metonymic reference
and a corresponding lexical direct object which does not:
(17) a. Pierre s’économise pour être en forme le jour du match. (Zribi-
Hertz 1978)
‘Pierre saves his energy in order to be in form on the day of the
match.’
b. *L’entraîneur l’économise pour qu’il soit en forme le jour du
match.
‘The coachi saves hisj energy for himj so that hej will be in form
on the day of the match.’
In (17a) the reflexive is clearly metonymic, because Pierre will surely not
save himself, but rather something about himself, i.e., his energy. This met-
onymic relation only works with the reflexive; we cannot have a non-reflexive
pronoun in the same slot. The asymmetry is the same in (18).
Grammatical Constraints on Metonymic Reference 243
A
B A
The syntax is the same in the (a) sentences and (b) sentences, respectively.
Both possessive relations (human being-feet vs. car-roof) may be coded in the
‘ordinary’ type of construction with the adnominal phrase, as (21a) and (22a).
But only with the BODY-PART relation in (21b) may the possessor (the son) be
coded in a dative phrase. Only the possessive relations that are intuitively
conceived of as ‘highly dense,’ ‘inalienable’ permit this marked construction
type. (The precise semantic restrictions of the construction in French cannot
be investigated here; cf. Spanoghe 1995; König and Haspelmath 1998 and
references cited therein for analyses.) Now, the relation between a possessor
and an (inalienable) possession is clearly a contiguity relation, because they
belong to the same domain (the human body): a foot, for example, is semanti-
cally contiguous to its bearer, because people know that people usually have
feet. This contiguity has, in this case, even been solidified as a grammatical
rule of French insofar as the possessor may be coded as a dative phrase. The
traditional term ‘inalienable possession’ is misleading, because it tells us little
about the nature of the relation in question. What is traditionally thought of as
inalienability boils down to a special instance of contiguity. Therefore, in-
alienable possession is another marked construction that inherently involves
contiguity.
The contiguity relation of the kind in question may give rise to a genuine
metonymy. There is a construction where the whole slides metonymically into
the syntactic slot initially reserved for a part. In French it is possible to say
(23a) as an alternative to (23b):
(23) a. Sylvie est jolie des yeux. (Frei 1972 [1939])
b. Les yeux de Sylvie sont jolis.
‘Sylvie’s eyes are beautiful.’
(23a) really says something about Sylvie’s eyes, but it is the entire person that
appears in the subject position. We are dealing with a kind of PART-WHOLE
metonymy in which a part (the eyes) is replaced by a whole (the entire person).
In Relational Grammar this is referred to as ‘possessor ascension’ (cf. Blake
1984).
In French and other contemporary European languages this construction
is only available for intransitive predicates. Only the subject can be affected
by such a metonymy, as König and Haspelmath 1998 have noted. So once
again, the hierarchy of arguments is confirmed: the subject is affected only if
there is no direct object. Now it would be nice, of course, to have examples
246 Richard Waltereit
where possessor ascension applies to the direct object in order to have even
better evidence for the hierarchy. Here I draw again on the study carried out by
König and Haspelmath. There are languages where other arguments are
implied, and in most cases it is the direct object:
(24) a. Homeric Greek (König and Haspelmath 1998: 564)
tón rh’ ébalen kephalen hupèr oúatos
this.one PAST-hit head above ear
oxéi khalkoi
javelin:DAT of.bronze
‘He hit his head above the ear with the bronze javelin.’
b. Sierra Popoluca (Mixe-Zoque, Mexico) (Blake 1984: 437)
šéwan a-kuʔd-aʔy an-sék
John me-eat-indirect my-beans
‘John ate my beans.’
c. Korean (König and Haspelmath 1998: 562)
acessi-ka sikey-lul cwul-ul kochi-ess-ta
uncle-NOM watch-ACC chain-ACC repair-PAST-IND
‘The uncle repaired the watch-chain.’
(24b) literally says: ‘John ate me with respect to my beans.’ This is the same
kind of synecdochic relation as illustrated in the French example (23a), but
this time the direct object is concerned.
As König and Haspelmath report, this kind of construction involving
possessor ascension most often occurs with the subject or the direct object. All
their examples with possessor ascension affecting the subject contain intransi-
tive predicates. With the transitive predicates they report it is always an object,
never the subject, that is affected by the metonymy. This seems to provide
fairly good evidence for the claim that possessor ascension preferably applies
to direct objects.
There is an alternation that seemingly provides evidence against the claim that
metonymies privilege the direct object. As is well-known, action verbs allow
the agent role to be replaced by the instrument role:
(25) a. Marie ouvre la porte avec la clé.
‘Marie opens the door with the key.’
Grammatical Constraints on Metonymic Reference 247
The obvious question now is: why should the direct object — in an accusative
language like French — be so vulnerable to metonymic interpretation? I am
unable to offer a completely satisfying explanation, but here are some possible
suggestions.
Firstly, the direct object belongs — with the subject — to those argu-
ments that are semantically opaque. Oblique functions (dative, locative and
temporal adjuncts, etc.) are most often semantically transparent, i.e., their
preposition- or case-marking permits only one thematic role (or few of them).
The dative phrase usually has to be interpreted as an experiencer or a benefi-
ciary argument; locative prepositions clearly mark locative arguments, etc.
Subjects and direct objects, on the other hand, are wide open as to their
thematic interpretation, the exception being that the direct object cannot mark
an agent (cf. Givón 1984: 135–185). Clearly, ambiguities in verb arguments
are only possible within the range permitted by the overt marking of each
particular argument. It is therefore unexpected that verbs display a polysemy
that relies on an ambiguity of an oblique object. But subjects and direct objects
should easily facilitate contiguity-based polysemies, because these arguments
allow for a wide range of thematic interpretations.9 (The same can also be
seen in the examples of metonymy given in Fass 1991.) This point clearly
argues for the supremacy of the subject in relation to verbal polysemy, even
over the object. But perhaps the following argument makes us understand why
the subject is outranked by the object in the end.
The ‘outranking of the subject by the object’ only concerns transitive
sentences, because only in transitive sentences is there both a subject and an
object. The object in transitive sentences and the subject in intransitive
(unaccusative) sentences are most often ‘themes’ (according to common
inventories of thematic roles). It seems natural to assume that their sharing this
property should be relevant for an explanation of the hierarchy (4). Now, the
roles of individual verbs labeled ‘themes’ generally seem, in a very particular
sense, semantically more ‘specific’ than roles which are eligible for subjects in
transitives (agents, instruments, experiencers). Whereas it is possible to have
Grammatical Constraints on Metonymic Reference 249
Notes
1. I am indebted to Andreas Blank, Mary Copple, Ulrich Detges, Paul Gévaudan, Martin
Haspelmath, Hendrikje Scholl, and two anonymous reviewers for useful comments on
earlier versions of this paper.
2. The relevant literature on frames includes Fillmore 1975, 1976, 1977, 1985; Tannen
1979, Schank and Abelson 1977; Barsalou 1992.
3. As pointed out by an anonymous reviewer, it is sometimes difficult to determine whether
we are dealing with two alternating roles of one predicate or whether the different
interpretations merely reflect a polysemy of the entire predicate. In fact, both analyses are
normally possible. Therefore the decision is less an empirical than a theoretical one. It
corresponds to what is known as the ‘predicate-independent’ vs. the ‘predicate-depen-
dent’ approach to thematic roles. Given that this paper focuses on arguments, not on
predicates, it seems reasonable to opt for the predicate-independent approach, i.e., to
assume that different interpretations of a verb as in (3a) and (b) actually point to
alternations of thematic roles and not to differences of predicates, of which role alterna-
tions would only be ‘reflexes.’ But note that the polysemy approach adopted here forces
us to consider roles not in isolation, but as parts of the verb they are employed with. Only
250 Richard Waltereit
verbs, by virtue of being lexical items, can be polysemous. Roles alone cannot, because
they are not lexical items. In short, the alternation in (3a) and (b) is a verbal polysemy
with respect to the direct object’s interpretation.
4. The historical data used in this paper are drawn from Le Robert Historique.
5. Cf., e.g., hériter de quelque chose/de quelqu’un ‘inherit sth./to be s.o.’s heir.’ In this case,
the contiguity of the inheritance and the heir applies to a prepositional object.
6. An anonymous reviewer objected that these alternations might simply boil down to
active-zone phenomena (Langacker 1984). Standard examples of active zones are We all
heard the trumpet or Don’t ever believe Gerald. The semantic content of their direct
objects does not precisely coincide with the entity designated by these objects, which
participates most directly in the process. A more ‘literal’ version would be We all heard
the sound of the trumpet or Don’t ever believe what Gerald says. Consequently, active-
zone phenomena overlap with metonymies (cf. Langacker 1993: 31). But there is a
crucial difference between active-zone phenomena and the alternations (5) to (10):
whereas active-zone examples are felt as non-literal speech and can be paraphrased by a
synonymous ‘literal’ sentence (as in the cited examples), frame-based alternations can-
not. Take examples (5a) and (b): the variants of balayer present slight but important
differences in meaning. Balayer ‘to sweep’ in (5a) has an incremental theme (the cleaned
room, place, etc. gets clean ‘step by step’) and this necessarily implies duration. The
variant (5b), however, has a holistic theme (things may be swept up ‘all in one go’) and it
does not necessarily imply duration. Accordingly, it is not always possible to replace the
variants one for each other (which again indicates that they actually differ in semantic
content). On the one hand, someone who has ‘swept up something’ has not always
thereby ‘swept a place.’ On the other hand, someone who has cleaned an already clean
place (i.e., without sweeping up anything), can nevertheless claim to ‘have swept it.’
7. This (ungrammatical) construction must, of course, not be confused with the impeccable
one se préoccuper de quelque chose ‘to worry about something.’
8. The PART-WHOLE relation is even iconically reflected in the reflexive clitic’s slight
morphological and phonetic weight, as was suggested to me by Peter Koch. Even on the
level of linguistic form, the reflexive’s antecedent is bigger than the reflexive itself, just as
the whole is bigger than one of its parts. Interestingly, the asymmetry between the reflexive
and its antecedent fades away as the reflexive marker grows in phonetic shape (which again
confirms the iconic nature of the part-whole relation in question). Préoccuper-type
constructions are sometimes better with a strong reflexive than with a reflexive clitic: je me
choque moi-même is better than je me choque ‘I shock myself’ (cf. Belletti and Rizzi 1988:
297f for analogous examples in Italian).
9. Languages may differ considerably with respect to the semantic range permitted by the
subject and the direct object. Clearly, arguments such as subject and direct object are
language-specific categories. The semantic spectrum of a given argument is therefore
highly dependent on particularities of the language’s grammar which could not be studied
here, e.g., the case system, word order rules, etc. The flexibility for metonymic reference
is also part of these language-specific properties. Cf. Hawkins (1986: 53–73) for differ-
ences of this kind between English and German, and Müller-Gotama 1994 for a cross-
linguistic study on semantic transparency in argument linking.
10. Koch 1981 offers an articulated theory on abstractness vs. concreteness of description of
thematic roles.
Grammatical Constraints on Metonymic Reference 251
References
Anderson, Stephen
1971 On the role of deep structure in semantic interpretation. Foundations of
Language 7: 387–396.
Baker, Mark
1988 Incorporation: A Theory of Grammatical Function Changing. Chicago:
The University of Chicago Press.
Barsalou, Lawrence W.
1992 Frames, concepts, and conceptual fields. In A. Lehrer, E.F. Kittay (eds.),
Frames, Fields, and Concepts: New Essays in Semantic and Lexical Orga-
nization. Hillsdale: Erlbaum, 21–74.
Belletti, Adriana, Luigi Rizzi
1988 Psych-verbs and -theory. Natural Language and Linguistic Theory 6: 291–
352.
Blake, Barry J.
1984 Problems for possessor ascension: some Australian examples. Linguistics
22: 437–453.
Blank, Andreas
1997 Prinzipien des lexikalischen Bedeutungswandels am Beispiel der roma-
nischen Sprachen. Tübingen: Niemeyer.
Busse, Winfried, Jean-Pierre Dubost
1983 Französisches Verblexikon: Die Konstruktion der Verben im Franzö-
sischen. Second edition. Stuttgart: Klett.
Bréal, Michel
1897 Essai de sémantique. Paris.
Croft, William
1993 The role of domains in the interpretation of metaphors and metonymies.
Cognitive Linguistics 4: 335–370.
Fass, Dan
1991 Metonymy, case role substitution and sense ambiguity. In D. Fass et al.
(eds.), Proceedings of the IJCAI-91 Workshop on Computational Ap-
proaches to Non-literal Language. Boulder, Colorado (Technical Report
CU-CS-SS0–91), 42–51.
Fillmore, Charles
1968 The case for case. In E. Bach, R.T. Harms (eds.), Universals in Linguistic
Theory. New York: Holt, Rinehart and Winston, 1–88.
1975 An alternative to checklist theories of meaning. Proceedings of the First
Conference of the Berkeley Linguistics Society. Berkeley, California: BLS,
123–131.
1976 Frame semantics and the nature of language. Annals of the New York
Academy of Sciences 280: 20–32.
1977 Scenes-and-frames semantics. In A. Zampolli (ed.), Linguistic Structures
Processing. Amsterdam: North-Holland, 55–81.
252 Richard Waltereit
Nunberg, Geoffrey D.
1995 Transfers of meaning. Journal of Semantics 12: 109–132.
Olsen, Susan
1994 Lokativalternationen im Deutschen und Englischen. Zeitschrift für
Sprachwissenschaft 13: 201–235.
Schank, Roger C., Robert P. Abelson
1977 Scripts, Plans, Goals, and Understanding: An Inquiry into Human Knowl-
edge Structures. Hillsdale, NJ: Erlbaum.
Spanoghe, Anne-Marie
1995 La syntaxe de l’appartenance inaliénable en français, en espagnol et en
portugais. Frankfurt a.M.: Lang.
Tannen, Deborah
1979 What’s in a frame? Surface evidence for underlying expectations. In R.O.
Freedle (ed.), New Directions in Discourse Processing. Norwood, NJ:
Ablex, 137–181.
Winston, Morton E., Roger Chaffin, Douglas Herrmann
1987 A taxonomy of part-whole relations. Cognitive Science 11: 417–444.
Zribi-Hertz, Anne
1978 Economisons-nous: A propos d’une classe de formes réflexives
métonymiques en français. Langue Française 39: 104–128.
Putting Metonymy in its Place
Paul Pauwels
KVH Antwerpen
1. Characterizing Metonymy1
When Lakoff and Johnson (1980) put figurative language at the top of the
agenda, claiming that it involved basic cognitive processes rather than deviant
usage, their work gave an enormous boost to metaphor research, but not to the
study of metonymy. After all, they did call their work Metaphors We Live By,
and in devoting only a single chapter to metonymy, they seemed to imply it
was a minor process in comparison. Moreover, they mainly defined me-
tonymy in comparison with metaphor.
Metaphor is principally a way of conceiving one thing in terms of another,
and its primary function is understanding. Metonymy, on the other hand, has
primarily a referential function, that is, it allows us to use one entity to stand
for another. But metonymy is not merely a referential device. It also serves the
function of enhancing understanding. [...] determines which aspect we are
focusing on. (Lakoff and Johnson 1980: 36)
It is this focusing behavior which gives metonymy its special cognitive status.
Examples (1) and (2) illustrate what is meant by ‘focusing.’
(1) The Times hasn’t arrived at the press conference yet.
(2) We need some good heads on the project.
In (1), which follows the WHOLE FOR PART pattern (institution for member of
the institution), the metonym is said to focus on the status of the reporter thus
referred to. In (2), the (body)part which is selected to refer to the whole
person, focuses the intelligence of that person as a particularly relevant
256 Paul Pauwels
At the same time, he notices a similarity between metonymy and other, non-
figurative, kinds of lexical ambiguity. Unlike Taylor (1989) Croft does want
258 Paul Pauwels
to differentiate between the two. The shift in (7a) and (b), for example, does
not involve figurativeness, since there is no ‘shift of reference,’ while there is
such a shift in (8a) and (b).
(7) a. This book is heavy. (in the literal sense, i.e., weighs a lot)
b. This book is a history of Iraq.
(8) a. The Times is owned by R. Murdoch.
b. The Times hasn’t arrived yet.
(7a) and (7b) highlight different primary domains of the concept book, viz. the
object-domain in (7a) and the contents-domain in (7b); the difference between
(8a) and (8b), on the other hand, lies in the fact that only (8a) highlights a
primary domain — focusing the newspaper as a company — while (8b)
highlights a secondary domain — focusing the newspaper as an entity which
can be represented by a correspondent. So far, the characterization is relatively
straightforward.
Croft then considers further examples. In a footnote he describes
Goossens’ example (4) of metaphor from metonymy as a case of domain
mapping, which is complicated by the fact that the source domain (sound) is
one of the domains “which happens to be in the matrix of the target” (speak-
ing) — hence the appearance of being metonymy (1993: 367). Here, Croft’s
argument is not as convincing. Within the domain matrix of linguistic action,
there obviously is a subdomain relating to the manner of speaking. In some
cases, this manner can indeed involve giggling (in a way which would not be
possible for say ‘barking,’ or indeed, ‘thundering’ which do not describe
human sound in their prototypical meaning — which is Goossens’ point). In
those cases, the use of giggle would simply be the highlighting of a secondary
domain, i.e., metonymy. In situations where the manner of speaking is con-
ceived in terms of giggling, i.e., speaking in a particular way as if it were
giggling, though actually no giggling is involved, there is domain mapping
and metaphor. The difference between such cases of metonymy and the more
central ones like (8b) above is twofold. First of all, these cases are situationally
invoked; there is no permanent link between the concepts involved. Secondly
— and this may relate to the first difference — the metonym deals with a
conceptualization expressed by a verbal expression which has a different
(probably more complex) kind of domain matrix than a conceptualization
involving a nominal expression. As things stand, Croft does not seem to want
Putting Metonymy in its Place 259
to make room for overlap between metonymy and metaphor in his model.
In what seems to be a further move away from recognizing distinctness of
domains in metonymy, he even questions the relation between metonymy and
highlighting.
It may not be the case that domain highlighting within the domain matrix of a
word is involved in all cases of metonymy. In some cases the shift in domain
prominence in the matrix is quite subtle, and sensitive to the semantics of the
associated words. [...] [I]t appears that no domain selection is involved. (Croft
1993: 350)
In example (9), which contains a metonym of the type PART FOR WHOLE, there
is no novel domain selection:
(9) We need a couple of strong bodies for our team.
The domain of the body is subsumed under the domain of the human being, and
if one subscribes to the encyclopaedic view on meaning, this means there is no
domain selection, but a “subtle shift in domain prominence” (Croft 1993: 350).
Summarizing, Croft distinguishes five possible cases: distinct domains
and mapping between them (metaphor); distinct domains of which one hap-
pens to be in the matrix of the other, and mapping (metaphor); single domain
matrix and highlighting of a secondary domain (metonymy); single domain
matrix and a difference in domain prominence (metonymy); and single do-
main matrix and highlighting of a different primary domain (literal lexical
ambiguity).
Finally, using Langacker’s (1987: 300) distinction between ‘dependent
predications’ (which depend on other predications for their elaboration) and
‘autonomous predications’ (which elaborate the dependent ones), Croft offers
an explanation for the high incidence of nominal examples in most discussions
of metonymy. He suggests highlighting occurs on autonomous predications,
while mapping occurs on dependent ones. He takes care, however, to empha-
size that metonymy is not restricted to nouns — the autonomous predications
par excellence — but that also verbs can behave autonomously with respect to
adverbs, and that nouns can behave as dependent predications with respect to
other nouns.
In most of the above, the issue seems to be the way to distinguish between
the working of metaphor and metonymy. Lakoff and Johnson (1980), Lakoff
(1987) and also Croft (1993) draw the line by referring to the distinctness of
domains, while Goossens (1990, 1995) wants to accommodate overlap. The
260 Paul Pauwels
notion of domain matrix and the composition of domain matrices seem to play
a crucial role in all analyses.
In a recent paper, Langacker (1994) comments on the importance which
distinctness, as opposed to continuity, plays in conceptualization. In his view,
most phenomena which have been described as continuous will yield to an
alternative analysis in terms of discrete categories upon further research.
There may be an exception, though, which is highly relevant to the present
discussion:
An important aspect of linguistic semantics is our ability to construe one
structure against the background provided by another. [...] I wish to empha-
size a parameter that may well be continuous: the salience of the background
structure, i.e. its level of activation in the construal of the target. (1994: 17)
the corpus which allow us to see what is more central and what is more
marginal to metonymy.
In our discussion we will refer to the notions of ‘cognitive domain,’
‘source’ and ‘target domain,’ ‘domain matrix,’ ‘mapping’ and ‘highlighting,’
and ‘dependent’ and ‘autonomous predications,’ as characterized above. We
will also make use of Langacker’s (1987: 231ff) distinction between ‘trajec-
tor’ (TR) and ‘landmark’ (LM) to describe the profiling of the state of affairs
in the sentence. In our corpus of manipulation verbs, the prototypical pattern is
a three-place predicate profiling two TR-LM relationships, as in (10).
(10) He put the book on the table.
(11) The constable put the spade in once.
(12) Adams was lighting and placing candles.
(11) and (12) are common alternatives. In (11) the secondary LM is not
expressed, or hidden, while in (12) the verb is used as a two-place predicate
profiling only one TR-LM relationship.
the concept of bet and the actual object which is being placed is of a more
complex nature. The bet is created by the act of placing money in a certain space
on the table. The relationship can be described as that between the result of an
action and an object which is instrumental in achieving that result, with the
RESULT standing for the INSTRUMENT. By referring to the result, the metonym
focuses the function of the action. Similar examples were found in put on a
disguise, lay bait and place an order. In the latter, however, the physical act of
placing is already much less central in the process, so much so, even, that the
example would preferably be analyzed as a metaphor from metonymy, where
there is a mapping from the domain of manipulation to the domain of ordering
goods.
(16) Father? Put your heart into it. Love me, come on. Have a shot at
filial devotion.
(17) I’m putting it aside for my old age.
(18) a. I do put my foot down about discussing them.
b. Just the once. I had to put my foot down there.
Examples (16)–(18) provide further illustration of the interpenetration of
metonymy and metaphor. (16) is a case of metonymy within metaphor:
making an effort to feel is metaphorically described as putting in, i.e., an act of
manipulation, while heart refers metonymically to the effort in which it is
crucially involved. Examples (17) and (18), although metaphorical to all
intents and purposes, have a metonymic basis: in both cases, situations where
the describing and the described act co-occur are easily conceivable —
indeed, it is probable that such situations are at the basis of the expressions —
which makes them typical cases of metaphor from metonymy. Examples (17)
and (18) are different from the foregoing in another respect: they could also be
taken literally. In (17) and (18a), however, there is still an element inside the
sentence pointing to the need for a reinterpretation in figurative terms. In (17)
there is the prepositional phrase introducing the element of long-term plan-
ning, and in (18a) the activity is linked to linguistic action by a prepositional
phrase. In (18b), however, the figurative interpretation depends on the wider
context.
(19) The power of life and death — no need of other vices. If you’ve
once put on the black cap, everything else tastes like wax fruit.
Putting Metonymy in its Place 263
(20) The board is laying a cable along a seven mile route in Surrey.
(21) The church has no views on drains, gasworks or bricklaying.
Examples (19)–(21) clearly demonstrate that also pure metonymy can be
context induced. In all three cases, the act of manipulation is a necessary part
of the action complex for which it stands; an ‘as if’ interpretation is ruled out.
The metonyms are at the same time referential and clarifying, since they focus
on the most visually salient aspect of the action complexes. In (20) there is
even a further metonym which, to put it in Croft’s terms, is induced by the
dependent predication (laying a cable) on the autonomous one (the board):
the board is the institution responsible for the action but is not the actual agent.
Most of the above examples (13)–(21) are characterized by a tension
between the concrete/specific and the abstract/general. In (13) and (15) there
are two factors contributing to this tension. First, there is the fact that the verbs
profile only one TR-LM relationship, which means there is an inherent
vagueness in comparison to the fully descriptive pattern (exemplified by (10)).
The less tangible character of the LM (bets, tea) outside the domain of
manipulatable objects is the second element. In (14), which is closer to the
fully descriptive pattern, the secondary LM is hidden, which again creates
vagueness, and the primary LM (tea) is again less tangible. In (19), a hidden
secondary LM combines with a primary LM which is marked as generic by the
definite article (the black cap), and in (20) the indefinite article introduces
indeterminacy into the primary LM (a cable), while a second metonym makes
the TR more abstract. (21) is probably the most striking example of the shift
towards abstraction, since the specific description has been turned into a
(necessarily) more general concept. The ‘exceptions’ are (16)–(18), where
metonymy and metaphor interact.
A first salient source domain involves the manipulation of body parts. In all,
some 60 expressions containing references to hand, finger, fist, foot, knee,
eyes, face and heart were found.
(22) She has glasses of whisky standing around at vantage points, to
which she would put her lips when so disposed.
(23) He was unlikely ever to set foot in his kingdom again.
(24) a. I’d strangle any man as much as laid a hand on her.
b. The first person who sets a finger on anyone of his family...
c. Trouble is I’m practically certain I’ve never laid eyes on you,
mate. (set)
d. I’ll tie a lovers’ knot with his arms and throw him in the Tiber if
I ever lay my tentacles on him.
In (22) drinking is described in terms of an action preparatory to the drinking.
The effect is to draw attention to the behavior of the person, rather than to the
actual drinking. Whether she is just wetting her lips or whether she is drinking
hard is left to be inferred. The effect is euphemistic, but there is of course the
irony added by the first part of the sentence.
It is probably best to consider cases like (23) as metaphorical in present-
day usage; the means of entering a country are so diverse and the prototypical
case of bodily movement is now only used in a context of overstatement.
There is a similar connotation of overstatement about the examples in
(24). In (24a) and (b) assaulting someone is referred to in terms of physical
contact — even the slightest touch is prohibited. In (24a) the overstatement is
as much the effect of the intensifier as much as, as of the description lay a
hand (not even two hands); (24b) minimizes the extent of contact (not even a
complete hand) to achieve a similar effect. Seen in the context of (24a) and (b),
(24c) describes an even more tenuous kind of contact. The expression derives
its value from the tension between the laying of a body part which would seem
to indicate physical contact, and eyes which, in this case, denotes a greater
distance. In (24d) the overstatement runs in the other direction: rather than
minimize the extent of physical contact as in the threats (24a) and (b) or the
disclaimer (24c), the extent of physical contact is maximized for effect. Again,
there is a shift away from the metonymic origins — these are still clearly
Putting Metonymy in its Place 265
present in an example like (24a), in so far as the hands are the prototypical
means of assault. (24c) and (24d), however, should be considered as pure
metaphor. In (24c) ‘seeing’ is grasped in terms of physical contact, with eyes
as a metonymic bridge (a case of metonymy within metaphor), while in (24d)
getting hold of someone with the aim to inflict physical harm is described in
terms of the octopus’s, usually unwelcome, attentions.
In sum, roughly 50 of the examples on this domain carry a connotation of
euphemism or, conversely, of overstatement. This connotation is often con-
textually supported, and the sentences often contain conditionals or markers of
exceptionalness (if ever, the first who, the first time, as soon as, (neg+) before,
once, everything, ...). The recipient domains in most cases refer to socially
unacceptable behavior such as excessive drinking, suicide/killing, fighting,
harassing, anger, or shame.
Again we note the tension between specific and general in most ex-
amples. In (23) there is a difference in ‘scale’ between the primary LM (foot)
and the secondary LM describing its target-location (kingdom). It could be
argued that this case is an exception; in other examples like in this house, in
this room the tension is less obvious. Still, the fact remains that literal
descriptions usually have as secondary LM either a physical object, a sub-
stance, or a surface. Similarly, in (24), there is a tension between the primary
LM (body part) and the secondary LM (person as a whole), while in literal
descriptions the focus is more narrowly on the contact between body parts as
in lay a hand on someone’s arm. It is this tension which is resolved in the
metonymic interpretation. In (22) the tension is between the specificness of
the description of the contact between the lips and a glass, and the frequency
suggested by the plural glasses in the antecedent.
body, spanking, toilet-training) are (or used to be) taboo topics. The subdo-
main of manipulation is not only selected because it avoids the taboo; there is
also the concreteness of manipulation which provides the necessary salience,
and the focus on the control exerted by one participant over another, which is
a central element in the event.
(35) a. If that’ll be all sir, I’ll let her in and put a kettle on.
b. I’ll put on an egg.
A second set of examples, where put is the central verb, all involve the starting
up of a process. In (33) the act of writing is referred to by means of the
manipulation of the prototypical instrument involved, or rather the first move-
ment of that instrument. In so far as this expression is used to refer to ‘manual’
writing, it can be considered as metonymy. If it were used, consciously, to
refer to writing by means of other instruments, this would be a case of
metaphor from metonymy (cf. also Pauwels 1995: 150). Similarly, in (34), a
fairly simple act of manipulation which occurs at the beginning of a sequence
of activating a recordplayer and playing a record, is used to refer to the
complete sequence. In this case, the metaphorical potential is possibly greater,
for two reasons. For one, the classic turntable is on its way out and being
replaced by the CD-player, where the record is not put on in the literal sense.2
On top of that, this might activate another path via which the usage can be
sanctioned, i.e., the metaphorical put on as in put on the radio. Example (35a)
falls into the same pattern as the above, with both the potential of a metonymic
reading (‘on the stove’) and a metaphorical reading (‘electric kettle’). An
example like (35b) clearly demonstrates a further shift away from the literal,
descriptive usage, since the instrument involved in the boiling of the egg
(indeed, boiling, rather than any other way of cooking an egg, is referred to
thus) is not mentioned.
In cases like the above, the metonyms appear to be mainly referring
devices. Referring to a complex activity by means of its fairly concrete first or
last stage is basically a time-saving strategy. Rather than focus on a specific
aspect, or euphemistically avoid another (the main strategies encountered so
far), this type of metonymy tends toward imprecise language use, exploiting
Grice’s maxims, one of which states one need not be more specific/precise
than necessary. The fact that quite a few of these examples involve closely
scripted events undoubtedly plays a role here. The selection of lay to concep-
tualize ‘stopping’ and put to conceptualize ‘starting’ relates to the prototypical
meanings of these verbs (see further Section 6) — a full discussion is beyond
the scope of this paper.
Not all the examples of manipulating instruments are of the same type,
however. Example (36) is clearly much more closely linked with the examples
in the previous sections as far as the motivating factor behind the metonym is
concerned.
Putting Metonymy in its Place 269
(36) “Put a pillow on his face and get him out of it,” said Paul to himself.
Although the actual act of putting a pillow on someone’s face falls into the
pattern of referring to the start of a scripted sequence of actions, there is, in this
case, an additional aspect of euphemism which provides further motivation for
the partial selection.
This is also true of (37), which is however even more complex, since
there is a complete interpenetration of metaphor and metonymy on different
levels.
(37) Wipe that smile off your face and get up, before I put you on the
governor’s report.
At the highest level, this is again a case of the first stage in a sequence standing
for the complete sequence: writing someone’s name down in an official
register is used to refer to the consequent punishment. However, the writing
down itself is metaphorized in terms of manipulation, where that which is
written down is referred to metonymically. The result of the complete se-
quence is a ‘demetonymization’ (cf. also Goossens 1990: 335) since in the end
the person is literally the affected entity.
In this section I will take a closer look at the nature of the link between source
and target domain from the perspective of the target domain. Of the examples
found in the corpus, quite a few, with put as the central element, have already
been discussed in some detail in Pauwels (1995). Also, examples from previ-
ous sections fit in here.
(18) a. I do put my foot down about discussing them.
(30) It is not difficult to lay down The Conquest of Peru or Volume 2 of
The Cambridge Medieval History, once it has been taken up; but it
is not so easy to feel altogether happy about never taking it up again.
(33) Plant plastic gnomes where you walk and even, perhaps, put pen to
paper and relate your experiences in the force.
(37) Wipe that smile off your face and get up, before I put you on the
governor’s report.
270 Paul Pauwels
have traditionally been described as PART FOR WHOLE or WHOLE FOR PART.
Looking at, for example, Lakoff and Johnson’s list (1980: 38f) it is clear that
most of their cases can be construed as more specific kinds of these two basic
types. PRODUCER FOR PRODUCT, OBJECT USED FOR USER, CONTROLLER FOR
CONTROLLED, INSTITUTION FOR PEOPLE RESPONSIBLE, PLACE FOR INSTITUTION,
PLACE FOR EVENT can all be characterized by invoking a ‘container’ relation-
ship between two domains: a superordinate one and a subsumed one. In
analyzing these examples as metonymy, the stand-for relationship is taken to
resolve the semantic tension.
However, the tension can be, and, judging from the evidence, often is,
resolved in another way. Take example (40), which can be described as a
straightforward case of metonymy, where table stands for the objects which
are being laid on it.
(40) I’ll just lay the table. (LET/Ayckbourn)
The autonomous predication (table) is the locus of the effect induced by the
dependent one (lay). This is however not the full story: the preferred status of
goal-objects appears to have resulted in an accommodation of this fact in the
semantics of the verb, which has resulted in a generalization. Rather than
‘manipulate,’ lay can be taken to mean ‘arrange, fix, prepare or make ready.’ In
other words, the meaning of the dependent predication appears to have shifted
as well. Also, the syntax of the verb has been affected by this: rather than profile
the primary LM of the lay-prototype (the goal-object which is being moved, as
in example (12)), this two-place predicate only profiles the secondary LM of
that prototype (the location involved in the movement). That this pattern has
proved productive, should be clear from the list of examples under (41).
(41) set the table, set the alarm, set a fuse, set the thermostat, set a watch,
set the oven to the required temperature, ...
Frequencies in the corpus seem to indicate that set has evolved furthest in this
direction. There are some examples with lay in this pattern, but no examples at
all for put and place. For put, this could be explained by the fact that the
original prototype concerns bodily movement (pushing, shoving) rather than
manipulation. In this prototype, the secondary LM with respect to which the
movement can be characterized is the agent her/himself. The use of put to
profile manipulation of an object with respect to a location is a later, extended
use, and in this literal meaning put needs the presence of at least a hidden
272 Paul Pauwels
secondary LM, which can then not move up in the hierarchy, so to speak, to
take the place of the primary one. Place, finally, is a much more recent
addition to the language: set and lay are attested in OE, put from late OE
onwards, while place is a 15th century loan from French.
7. Conclusion
Notes
References
Croft, William
1993 The role of domains in the interpretation of metaphors and metonymies.
Cognitive Linguistics 4: 335–370.
Geens, D., L.K. Engels, W. Martin
1975 Leuven Drama Corpus and Frequency List. PAL. KU Leuven.
Goossens, Louis
1990 Metaphtonymy: the interaction of metaphor and metonymy in figurative
expressions for linguistic action. Cognitive Linguistics 1: 323–340.
1995 From three respectable horses’ mouths: metonymy and conventionalization
in a diachronically differentiated data base. In L. Goossens et al. (eds.), By
Word of Mouth: Metaphor, Metonymy and Linguistic Action in a Cognitive
Perspective. Amsterdam and Philadelphia: Benjamins, 175–204.
Johansson, Stig, Geoffrey N. Leech, Hellen Goodluck
1978 Lancaster-Oslo/Bergen Corpus of British English, for use with digital
computers.
Lakoff, George
1987 Women, Fire and Dangerous Things: What Categories Reveal about the
Mind. Chicago: The University of Chicago Press.
Lakoff, George, Mark Johnson
1980 Metaphors We Live By. Chicago: The University of Chicago Press.
Langacker, Ronald W.
1987 Foundations of Cognitive Grammar. Volume I: Theoretical Prerequisites.
Stanford, California: Stanford University Press.
1994 The limits of continuity: discreteness in cognitive semantics. In C. Fuchs,
B. Victorri (eds.), Continuity in Linguistic Semantics. Amsterdam and
Philadelphia: Benjamins, 9–20.
Pauwels, Paul
1995 Levels of metaphorization: the case of put. In L. Goossens et al. (eds.), By
Word of Mouth: Metaphor, Metonymy and Linguistic Action in a Cognitive
Perspective. Amsterdam and Philadelphia: Benjamins, 125–158.
Pauwels, Paul, Anne-Marie Simon-Vandenbergen
1995 Body parts in linguistic action: underlying schemata and value judgements.
In L. Goossens et al. (eds.), By Word of Mouth: Metaphor, Metonymy and
Linguistic Action in a Cognitive Perspective. Amsterdam and Philadelphia:
Benjamins, 35–70.
Taylor, John R.
1989 Linguistic Categorization: Prototypes in Linguistic Theory. Oxford:
Clarendon Press.
Conversion as a Conceptual Metonymy
of Event Schemata
René Dirven
Gerhard Mercator University of Duisburg
1. Definition
2. Application to conversion
feature for the agent, which can only be realized at the clause or sentence
level. Statistically, this referencing function seems to be intimately related to
the agent function, or at least to the subject function. Indeed, the HAMBURGER
metonymy is far less likely in direct object or indirect object position.
(5) a. ? I asked the hamburger to pay at the counter.
b. ? I showed the hamburger the way to the station.
This suggests that even the referencing type of metonymy cannot occur in
every possible context, but that it is highly constrained to one prototypical
configuration, i.e., one in which the agent/subject function, at least for hu-
mans, is salient.
Conversion differs from this agent/subject or referencing type of me-
tonymy in that it does not apply at the sentence or clause level, but rather at the
predicate-argument or nucleus level. Any of the participants in an action
schema or in any other event schema, except the agent/subject, can become
the bearer of the saliency feature in the appropriate configuration and then
serve as input for the conversion process. Thus in reporting a soccer game, we
can describe certain ways of handling the ball with the head as He headed the
ball into the goal, whereby the exact timing of the ball-to-head contact and the
exact force and direction given to the ball by the head are of paramount
importance. In the total action schema The player sends the ball into the goal
with his head the instrument participant with the head is so salient that it can
stand for the action itself. This does not mean, however, that the instrument
slot in the action schema disappears. Although instrument verbs like to kick, to
head, to hammer, etc. do not normally require an instrument, they can take one
— if it is sufficiently specified:
(6) a. He kicked the ball with his left foot.
b. He headed the ball with the back of his head.
c. He hammered the nail into the wall with his shoe.
This means that the conversion process does not necessarily delete the partici-
pant role which is denoted by the converted lexical item in the total action
schema. Since the type of metonymy that we find in the conversion process
takes place at the nucleus level, it cannot become a means of referencing as
metonymy at the clause level does; indeed, a nucleus is by definition not
grounded in communicative space.
Whatever differences there may be between these two processes of
Conversion as a Conceptual Metonymy of Event Schemata 279
ner, goal or source and essive roles (e.g., Kastovsky 1974, 1977; Leitner 1977;
Clark and Clark 1979). Elaborating on this previous research, I proposed five
classes of converted verbs based on underlying semantic roles holding be-
tween a putative predicate and a semantic role (Dirven 1986: 321ff):
(9) a. object verbs (to fish, to crew, to anger)
b. instrument verbs (to harpoon, to head, to veto)
c. manner verbs (to queue, to balloon, to spoon)
d. locative verbs (to bottle, to shelve, to record)
e. essive verbs (to author, to nurse, to knight)
As was already suggested in Section 2, these five categories can be seen as
labels which summarize a full-fledged event schema in which several partici-
pants operate on each other.
There are three canonical event schemata underlying conversions:
(i) the action schema, in which an agent acts upon a patient, often using an
instrument and involving a certain way or manner of doing this as in to
fish;
(ii) the location or motion schema, in which the agent may perform an action
aiming at a localized effect as in to bottle;
(iii) the essive schema, or the schema for ‘beingness,’ in which the status of
class membership or an attribute is assigned to an entity.
These three event schemata will now be analyzed in more detail, whereby the
semantic role selected is seen as the element responsible for the metonymy
underlying the event schema as a whole.
The action schema conceptually synthesizes the flow of energy from an agent
to a patient via an instrument in a certain manner. In the instantiation of
someone trying very patiently to catch fish with a fishhook, each of these
semantic roles, except for the agent itself, can be focused upon, assume a
predicate’s relational function of linking the semantic roles, and in the process
highlight one aspect of the action schema, but always imply the whole of it.
Thus the patient itself, the instrument, and in some contexts, the manner, can
come to stand for the whole action schema as in (10):
Conversion as a Conceptual Metonymy of Event Schemata 281
In its simplest form, the motion schema comprises a moving patient and one or
more elements of the motion’s trajectory, i.e., source, path and goal. Given the
fact that in the human sphere of interest, goals are more salient than sources
(Ikegami’s 1987 goal-over-source principle), it need not surprise us that
locative verbs resulting from conversion processes are predominantly goal
verbs.
It is, moreover, not astonishing that the natural topography of the goal
also determines the type of arrival. The type of physical end-point of the
motion determines the specific meanings of each converted verb as
contextualized in the following sentences:
(12) a. The tide had gone out, leaving the boat stranded on the rocks.
b. Before going home, the fisherman beached his boat.
c. On the cruise we’ll first land in Casa Blanca.
d. The plane was forced to land in Cairo.
e. The submarine surfaced again.
f. The plane was grounded there for 24 hours by the hijackers.
The metaphoric expression leave somebody or something stranded, as in
(12a), historically relates to the now obsolete noun strand, meaning ‘the land
bordering a sea, lake, or river’ — i.e., the coastline or margin, be it rocks, sand,
etc. (still found in the name of a street in London, i.e., The Strand). Thus, for
example, a boat which has run aground during a storm has been left helpless
and unable to move. Since no successful arrival was achieved, the boat got
‘stranded.’ The type of arrival denoted by to beach is different, as (12b)
shows. The Dictionary of Contemporary English defines the meaning of the
Conversion as a Conceptual Metonymy of Event Schemata 283
verb as ‘to pull a boat onto the shore away from the water.’ Thus, if we leave a
boat on a beach, by running or hauling it there, we have ‘beached’ the boat. It
is only when we reach the land, as opposed to the sea or the air, that we have
‘landed’ as in (12c, d). In each of these instances, the physical strip or area of
land which constitutes the goal stands for the motion as a whole. Likewise to
surface in (12e) picks out the top of the water as its goal and to ground in (12f)
takes the earth seen in opposition to the air as its goal.
These instances of conversion may also show that in the conversion process
there is never a purely physical goal, but that the goal also stands for the
circumstances surrounding the scene. This also applies to their metaphorical
senses. Thus to strand evokes and reflects the shipwreck scene or a situation in
which a child is left unattended in a strange place. To beach brings to mind a
spontaneous situation in which, for example, the people jump out of the boat as
it reaches the beach, haul it up onto the sand anywhere away from the water and
just leave it there. To land suggests a purposeful arrival, either from the sea or
the air, as well as in the phrase ‘landing a job.’ To surface calls up an image of
a swimming or underwater-motion scene, as well as, metaphorically, the
coming to light of scandals or lies. To ground evokes a scene of forced contact,
which can also be noted in the metaphorical usage of ‘grounding a naughty
child.’ In other words, the use of metonymy is not restricted to the aspect of
motion as such, but it also encompasses the wider experiential scene of which
it is one specific facet. This wider experiential scene is in fact the real motivation
for locative conversion verbs — and probably for all conversions.
For locative verbs such as to bottle, to box, to can, etc., it is the scene of
food preservation that provides the context. Other more ‘abstract’ goals may
be shelter as in to house, to jail, or to harbor; display as in to bench or to field;
or else giving things a specific shape or form as in to bundle, to pile, or to slice.
Even purely abstract mental motion leading to a new artifact involves the
motion schema as in to book, to map or to register.
woman, who tends the sick, injured, or infirm’ or the more specialized
meaning of ‘a woman employed to breast-feed another woman’s child’
(CED). The six meanings of the converted verb to nurse can be neatly
associated with either of these two meanings:
(13) a. Mary nursed the sick soldiers (= to tend an injured person)
b. Mary nursed her father’s ailment (= to tend the wounds)
c. Cathy nursed the crying child in her arms (= to clasp)
d. Cathy nursed the baby five mornings per week (= to care for)
e. Gilly nursed the baby since the baby’s mother could not breast-
feed her child (= to breast-feed)
f. The baby nursed at the woman’s breast (= to suckle)
To put it in a somewhat simplistic way, we could say that one need not be a
nurse in order to do all the ‘nursing acts’ described in (13). Indeed, the verb’s
meaning is far more general than that of the corresponding noun. The two
domains covered by to nurse are ‘tending sick or injured people’ and ‘taking
care of young children, either as a care-giver or else in the breast-feeding role.’
The far richer meaning of the verb to nurse as compared with the noun nurse
also reveals another aspect of the metonymic nature of the conversion process.
Whereas the noun can only denote more or less fixed categories or classes
and their members, the verb is far more flexible and somehow breaks the
constraints imposed by the boundaries of the noun’s class membership mean-
ings. Thus the conceptual entity ‘a nurse’ can only, in her profession, tend sick
or injured people or breast-feed another woman’s child. But the looser con-
ceptual entity of ‘a person who nurses’ can do both: tend sick or injured people
or breast-feed young children. When nursing young children, she can be
imagined in a specific act such as holding a child in her arms, in her overall
assignment, or in her ‘technical’ role of breast-feeding a child. In fact, it is
again the various overall scenes that may be associated with her work as a
nurse that are conceptualized in the conversion process.
We can also combine the essive schema with the action schema: for
example, an agent may turn a patient into a member of a class. Thus, in to
knight, the action schema and the essive schema combine to express the
institutionalized act of elevating a person to knighthood. This is, however, a
very unproductive pattern.
The third type of essive relationship is that of assigning an attribute to a
patient. This relationship is only exceptionally expressed by nouns, the typical
Conversion as a Conceptual Metonymy of Event Schemata 285
word class for this purpose being adjectives. Here again, the essive schema
combines with the action schema as in to clean the table, which is a highly
productive pattern in English.
district can never mean that people are turned into ‘police officers of a district’
(essive relation), but rather it can only involve the patient role, i.e., that you put
a sufficient number of policemen in a district so that it may become safe again.
Here the police are not understood as agents, but as patients, or instruments that
in an atmosphere of street violence may restore a neighborhood’s feeling of
safety.
The deeper explanation for the fact that typically human roles such as
agents or datives do not enter into conversion processes may have to do with the
clash between two cognitive principles, i.e., the principle of anthropocentrism
and the principle of metonymic focusing. Conceptually, it is indeed very
difficult to see how a human being could be part of some larger configuration
to which it could stand in a metonymic relationship. Whereas the opposite
relationships abound, i.e., a book being referred to by means of its author or a
customer being referred to by the food he ate, a metonymic relationship of
humans to some higher conceptual category is hardly imaginable and, hence,
agents and dative roles do not easily enter into conversion processes.
References
Leitner, Gerhard
1977 Zur Vorhersagbarkeit von Derivation: Teil von Nomina als Basen. In H.E.
Brekle, D. Kastovsky (eds.), Perspektiven der Wortbildungsforschung.
Bonn: Bouvien-Verlag Herbert Grundmann, 140–154.
Marchand, Hans
1969 The Categories and Types of Present-Day English Word Formation: A
Synchronic-Diachronic Approach. München: Beck.
OED
1994 The Oxford English Dictionary. Second Edition on Compact Disc for the
IBM PC. Oxford: Oxford University Press.
Zandvoort, Rainard W.
1961 A Handbook of English Grammar. Groningen: Wolters.
Opposition as a Metonymic Principle
Christian Voßhagen
University of Hamburg
1. Introduction
This paper discusses the notion that forms of language use in which something
is uttered to convey its opposite are metonymic. In this view, a conceptual
entity can be used to provide mental access to its opposite, which is closely
associated with it within a conceptual structure.
The basis of this discussion is the observation that opposites generally
form part of one conceptual domain, and that mappings which occur within
such domains involving opposites are influenced by factors that also influence
other metonymic mappings. Such factors include conceptual contiguity, sa-
lience, social factors, idealized structuring of conceptual domains, and the
highlighting function of metonymies for expressive purposes. These factors
will be discussed as motivations of metonymic mappings of one concept onto
its opposite in figurative language including euphemism, socially motivated
reversals, irony, and expressive use of negatives to convey positive evalua-
tions.
Speakers frequently use, and make sense of, expressions which involve saying
one thing and meaning the opposite. These expressions often contain adjec-
tives and seem to be used in two general ways. First, something positive can
stand for something negative, as in the following example:
290 Christian Voßhagen
(1) Y has been cheated by her friend X and says: “X is a fine friend.”
Here, the adjective fine conveys a negative, ironic meaning. Second, some-
thing negative can stand for something positive. A well-known example of
this is the adjective bad in its use in slang:
(2) bad
a. ‘eminently suitable or appropriate; excellent; wonderful [...]’
(Wentworth and Flexner 1967);
b. ‘a simple reversal of the white standard, the very best’ (Major
1971);
c. ‘good. Originally from the terminology of the poorest black
Americans, either as simple irony or based on the assumption
that what is bad in the eyes of the white establishment is good
for them [...]’ (Thorne 1990).
These processes lead to the following, related phenomena. First, expressions
can be ambiguous between a literal and its opposite meaning, as illustrated in
the following description given by an African American speaker (Folb 1980:
205):
(3) “Like ‘funky’ — it kin be really good or really bad. All depends.
You kin tell by d’ way dude say it — watch his face, move his body
certain way. Like, ‘He’s funky man!’ Move way from d’ dude —
he stink! ‘Hey, dude’s funky!’ Nigger be smiling, make d’ fist —
dude’s okay!”
Second, expressions can acquire a meaning that is the opposite of their
original meaning. The adjective terrific with its now commonly accepted
positive meaning is an example of this. This paper regards instances of
figurative speech of this kind as manifestations of a widespread metonymy: A
CONCEPT STANDS FOR ITS OPPOSITE.
In this view, using an expression in order to convey its opposite falls under
definitions of metonymy such as the one formulated by Lakoff (1987: 84):
– There is a “target” concept A to be understood for some purpose in some
context.
– There is a conceptual structure containing both A and another concept B.
– B is either part of A or closely associated with it in that conceptual
structure. Typically, a choice of B will uniquely determine A, within that
conceptual structure.
Opposition as a Metonymic Principle 291
Opposition is without doubt one of the most basic semantic relations (cf. Cruse
1986: 198; Deese 1970: 102). A vast number of lexical items are related by
opposition. In contrast to other relations distinguished in semantics, opposition
is readily comprehended by speakers and is among the first semantic relations
children learn to distinguish accurately (Landis, Herrmann and Chaffin 1987).
292 Christian Voßhagen
0
‘bad’ ‘neutral’ ‘good’
Figure 1. Concepts of evaluative antonyms
The labels ‘neutral’ and ‘0’ imply that there is a portion of the scale where
neither of the concepts applies; thus, it is possible to say something is neither
good nor bad. For the same reason, negation of one pole does not necessarily
imply assertion of the other — it is possible to say something is not good, but
not bad either. The arrows at the ends of the scale indicate that both concepts
are graded and can be intensified; thus, it is possible to say something is very
good or very bad.
In the analysis that follows, I would like to suggest a differentiation
between two groups of antonyms: ‘evaluative antonyms,’ exemplified by
good-bad, and antonyms belonging to the domain of ‘physical measurement,’
exemplified by big-small. The difference between these groups can be seen in
comparing Figures 1 and 2:
‘dead’ ‘alive’
From a logical perspective, there are only two values possible in the concep-
tual domain of, say, ‘organisms’ — not alive implies ‘dead,’ not dead implies
‘alive.’ This may be called a ‘two-valued organization’ of the domain.
closely associated within a conceptual domain, and that one of them can be
used as a reference point affording mental access to a target (Langacker 1993).
The target is, in this case, the conceptual domain that includes both concepts.
The systematic occurrence of this phenomenon in language supports the claim
made in the previous section that contiguity between opposites is not only
conditioned by lexical association. It can be argued that just as the adjective
denoting the larger amount of the quality is linguistically unmarked, the
corresponding salient concept is perceptually unmarked with regard to its
opposite, making it both “simpler” (Cruse 1986: 248) and “more basic”
(Lakoff 1987: 60).
Evaluative concepts show a different kind of asymmetry. Whereas the
markedness of physical measurement antonyms is perceptually motivated,
evaluative antonyms involve what may be said to be a socially motivated
markedness. The adjectives that express a socially desirable, positive concept
may be called unmarked, whereas those that express a negative, socially
undesirable concept may be regarded as marked. As an illustration of social
markedness, consider the following euphemism:
(5) pretty ear
‘an ear deformed from being hit repeatedly; a cauliflower ear’
(Wentworth and Flexner 1967)
The adjective pretty is used to denote something that may well be described in
terms of its antonym, ugly. Mention of the negative, socially marked term is
avoided, and its opposite is used instead to stand for the undesirable quality.
The motivating factor is not perceptual salience, but is closely connected with
social-pragmatic factors such as politeness.
In the following section, other social factors will be shown to motivate
metonymic mappings onto opposite concepts.
(6) insane
‘positive, healthy state of mind’ (Major 1971)
Why do speakers avoid a positive, unmarked expression like sane and use a
negative, marked one to access the positive concept? A plausible explanation
results if conflicting social or cultural attitudes are considered as a motivating
factor. For example, the slang use of bad as illustrated in example (1) gains its
metonymic sense from what might be called a negative attitude. Dictionary
definitions speak of a ‘reversal’ and, more specifically, of the assumption that
what is bad for one social or ethnic group must be good for the other.
When social attitudes differ, linguistic expressions may take on different,
but related meanings that correspond to these attitudes. When the attitudes are
opposed, linguistic expressions may take on opposite meanings. Thus, the
adjective bad had a conventionally accepted negative, marked meaning for
white mainstream culture, and a positive one for many African American
speakers who opposed the attitudes of the dominant group. In this sense, the
metonymic use of expressions like bad, wicked, etc., is motivated by social
opposition.
A similar motivation seems to underlie other cases where the opposition
is not necessarily between ethnic groups, but between subculture and main-
stream culture, as in the following examples (all from Wentworth and Flexner
1967):
(7) nasty
‘excellent; “wicked” [...]’
(8) mean
‘[...] psychologically exciting, satisfying, and exhaustive; Mean —
the best; the greatest’
(9) evil
‘wonderful; specifically, thrilling, very satisfying [...] Implies that,
like sex, that which is really thrilling and satisfying is considered
sinful by puritanical people.’
As will be shown below, there are many expressions, especially in American
slang, that reverse attitudes of mainstream culture. Along with other factors,
this aspect of social evaluation seems to be a basic motivating factor for this
kind of metonymy.
Opposition as a Metonymic Principle 297
6. Two-valued orientation
The examples given so far show that both negative and positive concepts can
stand for their opposites. In euphemism and irony, positive concepts tend to
stand for their negative opposites; in expressions motivated by opposition of
social or cultural attitudes, negative concepts stand for their positive oppo-
sites. It is remarkable that such non-literal uses of language always seem to
involve evaluative oppositions such as between good and bad rather than
oppositions such as between big and small, which may be called denotative in
contrast. Opposition metonymies only seem to apply to evaluative (connota-
tive), not denotative oppositions.3 Thus, a more specific formulation of the
metonymic principle suggested here could be AN EVALUATIVE CONCEPT MAY
STAND FOR ITS OPPOSITE.
This further specification is necessary because evaluative opposition
often seems to be conceptualized in ‘non-semantic’ terms by speakers, which
sometimes results in an instance of opposition being seen as absolute, not
scalar. Hayakawa (1978: 211) described this phenomenon with the following
words:
In the expression “We must listen to both sides of every question,” there is an
assumption, frequently unexamined, that every question has two sides — and
only two sides. We tend to think in opposites, to feel that what is not good
must be bad and what is not bad must be good. [...] This penchant to divide the
world into two opposing forces — “right” versus “wrong,” “good” versus
“evil” — and to ignore or deny the existence of any middle ground, may be
termed the two-valued orientation.
Hayakawa was mainly thinking of the world of politics, but as the following
examples and observations illustrate, two-valued orientation is a much more
general phenomenon:
300 Christian Voßhagen
(i) Speakers often interpret the negation of one evaluative term as an asser-
tion of its opposite although the intermediary term is equally possible. In
answering the question Is it a good movie?, the reply with No tends to be
understood in the sense of ‘It is a bad movie,’ although it may be neither
good nor bad (cf. Lyons 1977: 278).
(ii) In irony, speakers often make use of the opposite, positive pole in order to
state something negative rather than using intermediary terms.
(iii) In euphemism, the positive opposite of an avoided expression is often
used instead of a neutral term. Consider the following examples from
Italian, French, and English, in which the use of euphemisms is motivated
by religious taboo:
(16) benedetto
‘maledetto’ (Kainz 1965 I: 255)
(17) sacred animal
‘cursed beast’ (ibid.)4
(18) a holy terror
‘(sl.) [...] mischievous, embarrassing child’ (Hornby 1974)
(iv) An expression that literally denotes something salient can be metaphori-
cally used to convey both something very negative and something very
positive, but nothing in between, which leads to an abrupt transition from
positive to negative in the dictionary entries:
(19) smash
‘A total failure [...] a popular success; [...] a hit [...]’ (Wentworth
and Flexner 1967)
(20) tough
‘great, wonderful, difficult, terrible’ (Major 1971)
(v) Two-valuedness can be seen in a very interesting way in the following
slang expression:
(21) laugh at the other side of one’s face
‘to cry; to change one’s mood from happy to sad [...]’ (Wentworth
and Flexner 1967)
Laugh stands for ‘cry,’ positive and negative emotions are treated as two sides
Opposition as a Metonymic Principle 301
of one thing, i.e., the negative side is the other side of the positive rather than
detached or separated from it.
All these examples point to a general human tendency to think of evalua-
tive concepts in terms of opposites. This view is also reflected in the expres-
sions cited by Hayakawa and in folk notions such as the two sides of a story,
there are two sides to everything, there is a thin line between love and hate,5
or in the common metaphor to think in black and white rather than in shades
of gray.
Two-valued orientation in domains of evaluation seems to be a basic
conceptual phenomenon, an idealized kind of opposition. There is thus a
discrepancy between semantic and conceptual opposition. Semantically, an
opposition between gradable terms involves a whole scale, including a neu-
tral, middle ground; on the conceptual level, however, it may only involve the
two opposite terms. This state of affairs is represented in Figure 4:
‘good’ ‘bad’
This figure illustrates that, under certain conditions, the relation between
evaluative scalar concepts may be similar to that between complementary
ones (cf. Figure 3). Just as either of the complementary concepts ‘dead’ or
‘alive’ applies in the domain of organisms, either of the concepts ‘good’ or
‘bad’ applies in the evaluative domain. Therefore, the two-valued orientation
reduces the number of potential meanings of evaluative expressions: they are
either meant literally or are understood metonymically in the complementary
sense of their opposites. Such mappings seem to occur much more frequently
in domains of evaluation than in others. That is, whereas it would be very
strange to say big and mean ‘small’ in a purely descriptive, non-ironic context,
it is both possible to utter a positive term in order to convey a negative concept
(as in irony and euphemism), or to utter a negative term in order to convey a
positive concept, as in the socially motivated reversals discussed in Section 5.
A further form of metonymic use of negative concepts that seems to be
motivated by factors other than those mentioned so far will be discussed in the
following section.
302 Christian Voßhagen
George [a trifle abstracted]: Oh, I’ve had it awhile. Did you like
that?
Martha [giggling]: You bastard.
[...]
George [leaning over Martha]: You liked that, did you?
Martha: Yeah... That was pretty good. [softer] C’mon... Give me a
kiss.
The expression bastard is clearly not intended in the sense of a swear word in
this dialogue: Martha giggles when she utters it and then expresses the wish to
be kissed. In this case, Martha uses this term to express affection. For Wundt,
the main motivation for using language in this way is expressiveness: “It is
only the wish to emphasize the emotion in the strongest possible way that
leads to this result” (Wundt 1912: 576; my translation).
In a very similar context, many native speakers of English find it per-
fectly natural that a mother says You brat! to her child in a tender tone of
voice. There does not really seem to be a conflict between the intended
positive message and the negative semantic content of the utterance. It may be
that a negative emotion concept such as anger or fear can afford mental access
to, and vivid expression of, a positive emotion concept exactly because it is
perceptually or conceptually more prominent and because it therefore allows
the speaker to focus on the intensity of the target concept.
Modern inquiry into the nature of emotions has found that every emotion,
positive or negative, has its own specific effects and intensity, and most
importantly, that certain negative emotions are experienced more intensely
than positive ones; in particular, physiological effects such as heart rate seem
to increase more in the negative emotions fear and anger than in the positive
emotion happiness (cf. Ekman et al. 1983): Following Wundt’s suggestion
noted above, it might be said that negative emotion concepts are, paradoxi-
cally, very suitable for the expression of highly positive concepts. There
seems to be a general difference between cases like these and cases in which
something positive stands for something negative. POSITIVE FOR NEGATIVE, as
in an utterance like Oh, excellent, you idiot!, always seems to involve an ironic
intention of the speaker. When something negative is used to stand for
something positive, this is not the case.
Among the numerous negative slang expressions that are used to stand
for something positive, there is a very interesting group in which expressions
Opposition as a Metonymic Principle 305
originally connected with ‘fear’ are used to stand for something ‘pleasant.’
Consider the following entries from slang dictionaries:
(28) scare
‘[...] a pleasant surprise’ (Major 1971); “Wow, what a scare!”
(29) panicky
‘extreme pleasure or excitement’ (Major 1971); ‘excellent, very
satisfying or exciting’ (Wentworth and Flexner 1967)
(30) terrible
‘wonderful; great; the best; the most [...] “Terrible — the best; the
greatest.”’ (Wentworth and Flexner 1967)
In these instances the negative concept is used to afford mental access to its
positive opposite because it is more expressive in that it highlights the aspect
of excitement. I would like to argue that these are expressive metonymies,
which accounts for the motivation of the slang expressions mentioned earlier
as well as for the more conventional instances of opposition such as the
adjective terrific mentioned at the beginning.
8. Conclusion
Notes
1. Antonymy and complementarity are not the only kinds of lexical opposition that are
distinguished in lexical semantics, but they constitute two very basic groups and the
difference between them corresponds largely to the distinction in logic bewteen contra-
diction and contrariness (cf. Lyons 1977: 271).
2. The positive senses of these expressions are also metaphorically motivated: salient,
positive physical measurement antonyms have obviously acquired a positive sense
through metaphorical use.
3. This is not meant to deny that evaluative contrasts are also denotative; rather, evaluative
or connotative opposition exists in many conceptual domains in addition to denotative
opposition.
4. In the same context, Kainz (1965) points out that the word taboo itself could mean both
‘holy’ and ‘cursed’ in its original use by native Australians; the same holds for Latin
sacer.
5. This expression was used in a preview of a thriller on American TV.
Opposition as a Metonymic Principle 307
References
Aitchison, Jean
1987 Words in the Mind. Oxford: Basil Blackwell.
Albee, Edward
1962 Who’s Afraid of Virginia Woolf. Harmondsworth: Penguin, 1965.
Clark, Herbert C., Richard Gerrig
1984 On the pretense theory of irony. Journal of Experimental Psychology:
General 113: 121–126.
Croft, William
1993 The role of domains in the interpretation of metaphors and metonymies.
Cognitive Linguistics 4: 335–370.
Cruse, D.A.
1986 Lexical Semantics. Cambridge: Cambridge University Press.
Deese, James
1962 On the structure of associative meaning. Psychological Review 69: 161–
175.
1970 Psycholinguistics. Boston: Allyn and Baron.
Ekman, P., R.W. Levenson, W.V. Friesen
1983 Autonomic nervous system activity distinguishes among emotions. Science
221: 1218-1220.
Esper, Erwin A.
1973 Analogy and Association in Linguistics and Psychology. Athens: Univer-
sity of Georgia Press.
Folb, Edith A.
1980 Runnin’ Down Some Lines: The Language and Culture of Black Teenagers.
Cambridge, MA: Harvard University Press.
Fromkin, Victoria
1973 Speech Errors as Linguistic Evidence. The Hague: Mouton.
Garman, Michael
1990 Psycholinguistics. Cambridge: Cambridge University Press.
Givón, Talmy
1978 Negation in language: pragmatics, function, ontology. In P. Cole (ed.),
Syntax and Semantics. Vol. 9: Pragmatics. London: Academic Press, 69–
112.
Hayakawa, Samuel I.
1939 [1978] Language in Thought and Action. Fourth edition. New York:
Harcourt Brace Jovanovich.
Hornby, Albert S.
1974 Oxford Advanced Learner’s Dictionary of Contemporary English. Oxford:
Oxford University Press.
Kainz, Friedrich
1965 Psychologie der Sprache. 4 vols. Stuttgart: Ehmke.
308 Christian Voßhagen
Lakoff, George
1987 Women, Fire, and Dangerous Things: What Categories Reveal about the
Mind. Chicago: The University of Chicago Press.
Lakoff, George, Mark Johnson
1980 Metaphors We Live By. Chicago: The University of Chicago Press.
Landis, T.Y., D.J. Herrmann, R. Chaffin
1987 Developmental differences in the comprehension of semantic relations.
Zeitschrift für Psychologie 195: 129–139.
Langacker, Ronald W.
1993 Reference-point constructions. Cognitive Linguistics 4: 1–38.
Lehrer, Adrienne
1990 Polysemy, conventionality, and the structure of the lexicon. Cognitive
Linguistics 1: 207–246.
Lyons, John
1977 Semantics. 2 vols. Cambridge: Cambridge University Press.
Major, Clarence
1971 Black Slang. London: Routledge.
Murphy, George L., Jane M. Andrew
1993 The conceptual basis of antonymy and synonymy. Journal of Memory and
Language 32: 301–319.
Sperber, Dan, Deirdre Wilson
1995 Relevance. Second edition. London: Basil Blackwell.
Thorne, Tony
1990 Dictionary of Contemporary Slang. London: Bloomsbury.
Wentworth, Harold, Stuard Flexner
1967 Dictionary of American Slang. New York: Crowell.
Wundt, Wilhelm
1912 Völkerpsychologie. Band 2. Leipzig: Engelmann.
Metonymic Hierarchies
Kurt Feyaerts
University of Leuven (KUL - KULAK)
1. Introduction
One of the tenets of cognitive semantic is the focus on figurative language use
as a central topic of cognitive semantic research.1 As opposed to structuralist
linguistic approaches, where phenomena like metaphor and metonymy are
considered mere rhetorical or semantically peripheral devices, cognitive se-
mantics demonstrates that metaphor and metonymy are essentially ‘concep-
tual phenomena’ which show up only secondarily in linguistic expressions.
Up until now, the field of cognitive semantics has witnessed an overwhelming
interest in the role of metaphor as a conceptual mechanism structuring large
parts of our knowledge. Although the notion of metonymy was never entirely
absent, it was mostly relegated to the secondary status of being introduced or
mentioned in an essentially metaphorical context. However, recent years have
seen an increasing interest in the role metonymy plays in the structuring of our
conceptual system. Several studies seem to indicate that metonymy is catching
up to metaphor as a relevant area of study in the field of conceptual structure.
This article forms part of this renewed interest in metonymy. Its purpose
is to extrapolate the notion of ‘metaphoric hierarchies’ to metonymy, thus
showing that there are also large-scale metonymic structures to be found in
which several individual metonymies take part. I illustrate this with German
idiomatic expressions which all designate some kind of stupidity. It will
310 Kurt Feyaerts
become clear that most of the expressions in this domain involve a metonymic
relationship between source and target structure.
This paper is structured as follows. First, in Section 2, I will situate this
topic within the context of my larger research project. In Section 3, I will
briefly bring to mind some major characteristics of metaphoric hierarchies. In
Section 4, I will discuss some theoretical aspects of metonymy which are
related to the kind of material examined in this study. Finally, Section 5 will
concentrate on some examples of metonymic hierarchic structures as they
appear in this material.
2. Preliminary remarks
most northern part of the French-speaking culture and speak French with a
heavy regional accent. Belgians are consequently called stupid by the Dutch
as well as by the French. What makes the double use of this metonymic model
with respect to Belgians particularly interesting is the fact that it hides another
metonymy according to which, in both cases, the whole stands for one of its
parts. Accordingly, the Dutch as well as the French envisage in their stupidity
jokes the Belgians and not the Flemish or the Walloons, respectively.
Of particular interest for the present purpose is the observation that in this
cognitive-cultural model for stupidity, a functional relationship is established
between people’s origin on the one hand and their mental abilities on the
other. To put it in linguistic terms, in (1)–(3) both properties are ‘conceptual-
ized’ as being contiguous. In this case contiguity is realized as a kind of causal
relationship according to which stupidity is determined (or caused) by a
specific origin. In this construed relationship, stupidity is linked to another
property as its effect or manifestation.
Before I continue the analysis one point needs to be made very clear. What I
am aiming at here is the description of a ‘folk model’ for stupidity as it appears
in linguistic expressions rather than the verification of these models from the
point of view of observing experts. Logically, one knows that this kind of
functional link between geographic origin and mental ability is not generally
‘true’ or justifiable. Instead, I aim to find the conceptual structures which are
activated (un)consciously in the spontaneous categorization of a person as
STUPID. In this description I thus abstract from people’s reflection over this
value judgment. Lakoff (1987: 85f) describes these culturally determined
conceptual patterns as ‘social stereotypes,’ which he classifies as a subcat-
egory of metonymically structured cognitive models:
[...] they define cultural expectations, they are used in reasoning and espe-
cially in what is called ‘jumping to conclusions.’ However, they are usually
recognized as not being accurate, and their use in reasoning may be overtly
challenged.
These social stereotypes function as cognitive models and they reflect culturally
established preferences and norm concepts which play a decisive role in
people’s judgments about typical properties of all kinds of personality types. As
an example of a contemporary American stereotype, Lakoff describes the
Metonymic Hierarchies 313
In sum, one can say that cognitively oriented person perception theory
does not provide any grounding for the specific linking of conceptually
elaborated properties with each other. Instead, its value for the present analy-
sis must be situated on a more schematic and structural level where it provides
experiential grounding for the conceptualized contingency between two or
more seemingly unrelated properties. Its specific importance for the present
purpose is its function as a grounding principle for the metonymic structure of
the target concept STUPID.
3. Metaphoric hierarchies
T IM E S A R E T H IN G S E V E N T S A R E A C T IO N S S T A T E S / A T T R IBU T E S
A R E T H IN G S
T IM E S A R E T IM E S A R E S TAT ES AR E A T T R IBU T E S
E N T IT IE S L O C A T IO N S L O C A T IO N S AR E
P O S S E S S IBL E
O BJ E C T S
P A S S IN G O F P A S S IN G O F L IF E IS A C H A N G IN G C H A N G E IS
T IM E IS T IM E IS P E R S O N JOUR NEY S T A T E IS M O VEM E NT O F
M O V IN G E N T IT Y M O V IN G W IT H C HANGE OF O BJ E C T
R ES P EC T TO L O C A T IO N
L O C A T IO N
F U T U R E IS S e v e ra l e l a b o r a t i o n s : C H A N G E IS
E N T IT Y M O V IN G M OVEM ENT OF
TO WAR DS L IN E O F C O N D U C T I S A P A T H O BJ E C T
P R ES E NT (H e g o t o f f t h e ri g h t t rack )
T he tim e of (M IS )F O R T U N E IS A P A T H H e h as l o s t h i s
l eav i n g w i l l s o o n (S h e ran i n t o b ad l u ck ) l u ck
co m e
D Y IN G IS G O IN G T O A F IN A L
D E S T IN A T IO N
(H e m o v ed t o t h e o t h er w o rl d;
H e p as s ed aw ay )
4. Defining metonymy
INSTITUTION FOR PEOPLE RESPONSIBLE Exxon has raised its prices again.
THE PLACE FOR THE INSTITUTION The White House isn’t saying
anything.
THE PLACE FOR THE EVENT Remember the Alamo.
Expressions pointing to an imaginative, ‘non-realistic’ contiguity relationship
are only scarcely mentioned, although precisely in this kind of relationship the
conceptual status of a metonymic extension becomes very clear. The conclu-
sion that we are dealing with two different types of metonymy, i.e., ‘literal/
realistic’ vs. ‘figurative/imaginative’ metonymy, cannot be maintained. In
both cases, the conceptual-semantic relationship involves two interactional
(attributed) properties. Accordingly, even a contiguity relation which is expe-
rienced as being ‘realistic,’ inevitably pertains to a conceptually mediated
reality. This understanding is not unimportant for an adequate analysis of
metonymy as a conceptual mechanism. It shows that in both cases a contiguity
relationship, which is to be considered the conceptual basis of metonymy, is
involved.
Of central importance for an adequate analysis of examples such as in
(5)–(7) is the insight that phenomena like hyperbole and metonymy are
situated on two different semasiological dimensions. Metonymy, on the one
hand, concerns the contiguity relationship between two semantic structures
and thus basically pertains to the ‘conceptual-semantic’ level. Compare, for
example, the relationship between the semantic source structure ‘unable to
count to three’ and the target structure ‘stupid’ in (5). The observation that a
property like ‘unable to count to three’ is ‘not realistic,’ is of no relevance for
the purpose of this study, since the relationship I am interested in involves
nothing but attributed semantic structures. Hyperbole, on the other hand,
arises from the ‘inappropriate’ use of a linguistic expression in a particular
situation and thus primarily concerns a specific ‘referential’ aspect of a
semantic structure. To be more precise, the experience of an utterance being
hyperbolic has to be understood against the background of a narrow interpre-
tation of the concept of reference as ‘correct reference’: “if the meaning of the
expression is a true description of the thing it indicates, then the expression
refers correctly” (Geiger 1993: 278). In normal everyday communication, we
expect (most of the) utterances to be referentially true, but in using hyperboles,
we deliberately violate this communication principle for stylistic,
argumentational or other reasons. This narrow concept of reference is of no
Metonymic Hierarchies 323
interest to a semantic analysis, where the notion of truth does not play a
prominent role. Accordingly, in describing metonymy, my attention is drawn
by the relationship between two semantic structures, regardless of whether
these semantic values are ‘true’ or ‘real’ for a referent or not. Lyons (1977:
184) points out that the search for truth should be left to philosophers:
philosophers are professionally concerned with the explication of the notions
of truth, knowledge, belief and existence. The fundamental problem for the
linguist, as far as reference is concerned, is to elucidate and to describe the
way in which we use language to draw attention to what we are talking about.
5. Metonymic hierarchies
On the basis of this characterization of metonymy, I can now turn to some large-
scale metonymic structures that generalize over several linguistic instantiations.
I will show that metonymic hierarchical systems display similar structural
characteristics to the ones I described for metaphoric hierarchies (see above).
First, I will discuss two specific examples and then I will present a schematic
picture of the conceptual structure of STUPID which will demonstrate the
structural role metonymy can play as a pervasive conceptual mechanism.
(8) Er steht da wie der Ochse vor dem Scheunentor.
‘He stands there like the ox in front of the barn-door.’
This example concerns the conceptual background of a non-conventionalized
expression which profiles stupidity in terms of a particular behavior: a person
who is doing nothing else than just standing there like an immobilized object
(NOT MOVING). More specifically, an expression such as (8) incorporates an
overlap between the two concepts DEVIANT BEHAVIOR and DEVIANT APPEAR-
ANCE. A deviant behavior in its turn elaborates both concepts DEVIANT SOCIAL
PROPERTY and DEVIANT PHYSICAL PROPERTY. On a higher schematic level, these
latter concepts are schematized as SALIENT DEVIANT HUMAN PROPERTY. This
324 Kurt Feyaerts
D E V IA N T H U M A N P R O P E R T Y
D E V IA N T S O C I A L P R O P E R T Y D E V IA N T P H Y S IC A L P R O P E R T Y
D E V IA N T BE H A V IO R D E V IA N T A P P E A R A N C E
N O T M O V IN G
E r s t eh t da w i e der O ch s e v o r dem S ch eu n en t o r
S TU P ID
M ANTAF AHR ER
S TU P ID
lower-level schemas which further elaborate parts of this structure. Notice that
I consider the target structure STUPID as an elaboration of DEFICIENT MENTAL
ABILITIES and that, normally, one would expect the abstract and undifferenti-
ated concept STUPID to appear right below this schema. In order to keep this
figure as clear as possible, however, I moved it to the bottom. The numbers
between brackets refer to instantiating expressions which correspond to ex-
amples mentioned below or elsewhere in the text.
An interesting observation concerns the complex source structure which
can be observed in (5). On the one hand, not being able to count to three
elaborates both mental as well as practical abilities. On the other hand,
however, this deficiency also relates metonymically to little children who are
not able to count properly so that this notion — albeit not very prominently —
also takes part in the conceptual structure of the source domain. It thus appears
that — in applying the theory of complex conceptual networks — a complex
source domain structure may consist of schematically-related, as well as
(metonymically) extended concepts. A similar conceptual pattern underlies
the expression in (10), which instantiates a deviant appearance (the most
prominent source structure) and at the same time metonymically refers to a
small (newborn) child (DEVIANT AGE) which typically has this kind of appear-
ance. Both conceptual relationships combine into the complex source struc-
ture which metonymically stands for the target STUPID.
Metonymic Hierarchies 327
de v i a n t h um a n p ro p e rt y
s ev i a n t s o c i a l p r o p er t y de v i a n t p h y s i c a l p ro p e rt y
de fi c i e n t m ar k e d de v i a n t de v i a n t de v i a n t de fi c i e n t
a b i l i t i es ag e b eh av i o r s t a t us a p p e a ra n c e b o dy p a rt
(1 3 )
def i c i e n t de fi ci en t de fi c i en t low o ut g ro up de fi ci en t de fi c i e n t
s o ci al p ra c t i c a l m e n t a l p ro fe s s i o n o ri g i n h e a d/ b r a i n s e n s o ry
ab i l i t i es ab i l i t i es ab i l i t i es (1 4 ) (1 - 3 ) (1 2 ) a p p a ra t us
(1 5 ) (7 ) (1 1 )
(8 )
(5 )
(9 )
(1 0 )
S T U P ID
(14) Das ist noch unter dem Nachtwächter./Er ist ein geistiger
Untergefreiter.
‘That is even below the night watchman.’/‘He is an intellectual
“under-corporal”.’
(15) Er ist zu dumm, um einen Eimer Wasser umzustoßen.
‘He is even too stupid to knock over a bucket of water.’
Most of these examples speak for themselves, except perhaps for (13), which is
not an expression that profiles stupidity in a direct way. On the contrary, it
profiles a positive mental property. However, the image which is used confirms
the social stereotype according to which mental ability can be inferred from
somebody’s appearance (‘appearance as a manifestation of mental abilities’).
From a more profound analysis of the data, it follows that the prototypical
schematic source structure for STUPID is DEFICIENT HEAD/BRAIN (compare (12)).
More than the other source structures this schematic concept hides a complex
network of multiple interacting conceptual relationships. Correspondingly, this
conceptual source domain displays the highest instantiation rate.
With regard to a structural comparison between metaphorically and
metonymically organized conceptual hierarchies, it seems that both systems
have major structural characteristics in common. Although I am aware of the
different conceptual strategies which are involved in both mechanisms (cf.
Dirven 1993), Figure 4 clearly shows that generalized metonymy, just like its
metaphoric counterpart, operates on different levels of schematicity (‘first
similarity feature’). At the same time this broader focus allows one to identify
more easily the patterns of interplay and overlap between different metony-
mies in the hierarchy (‘second similarity feature’). In this respect I have briefly
discussed some interesting examples (compare (5), (8), (9), (10)) that profile
stupidity on the basis of a complex source domain. Such patterns of overlap
can be described as a complex structure situated somewhere in conceptual
space, defined by different relationships with other conceptual structures and
activated as (part of) the source domain matrix of the profiled concept.
The examples cited above display two possible patterns of overlap be-
tween concepts/metonymies. The first one resides in the elaboration of two
schematic structures, whereas the second one consists of a metonymic asso-
ciation with another concept. With this description, I apply Langacker’s
(1987) view on schematic networks as the organizing principle of our knowl-
edge structures. Accordingly, the conceptual background (domain matrix) of a
profile requires “specifications in numerous domains” (ibid.: 163). All these
Metonymic Hierarchies 329
6. Conclusion
Notes
1. I wish to thank Dirk Geeraerts for valuable help with the preparation of the talk on which
this article is based, Günter Radden for helpful comments on an earlier version of this
paper and Jessica Dobratz for proofreading major parts of the text. This article presup-
poses a certain acquaintance with basic theses and terminology of cognitive semantics
and of the Lakovian theory of metaphor in particular.
2. Compare Kövecses (1989: 44): “each and every expression related to a concept has to be
examined if we wish to uncover the minute details of the concept,” but see also Geeraerts
and Grondelaers (1995: 174ff).
3. In this respect I refer to Langacker’s (1993: 30) characterization of metonymy in terms of
reference points: “[...] a well-chosen metonymic expression lets us mention one entity
that is salient and easily coded, and thereby evoke — essentially automatically — a target
that is either of lesser interest or harder to name. [...] Other things being equal, various
principles of relative salience generally hold: human > non-human; whole > part;
concrete > abstract; visible > non-visible; etc.”
4. The cognitive basis of person perception makes it a discipline of great interest for
cognitive semanticists: “In dealing with person perception we are concerned with it as it
occurs in the layman, as a process affecting ordinary human interaction, and not with the
person as an object to be diagnosed by the expert” (Tagiuri 1958: xii).
5. Cf. Heider (1988: 341): “Our main concern is not the exact measure of one’s own ability,
but right away the relation to others. If somebody would tell us exactly what our ability
measures, we still would ask: but how do others do?”
6. Cf. Lakoff (1993) and Kövecses (1995a) for recent studies in this respect.
7. Several examples illustrate that LIFE IS A JOURNEY also elaborates the object dual of both
schematic metaphors which are involved here. Compare, for instance, expressions like
He lost his life (CHANGE OF STATE IS MOVEMENT OF POSSESSED OBJECT) or His youth has
abandoned him (TIME PASSING IS MOVING ENTITY).
8. See also Kövecses (1995b: 193), who distinguishes between a generic, mainly physiologi-
cally motivated conceptual level and a specific, mainly culturally determined conceptual
level.
Metonymic Hierarchies 331
9. Argumentation theory provides a good basis for an adequate description of the concep-
tual relationships underlying these expressions. Crucially, the kind of argumentation (or
conceptual linking) involved here is situated on a rhetorical rather than a strictly logical
level.
10. Obviously, as this theoretical perspective is inherent in any linguistic research, no
linguistic analysis can ever claim to reflect the mental richness of a conceptual structure
in full detail.
11. I refer to Sandra and Rice (1995), where the issue of the psychological reality of
theoretical notions such as ‘domain’ is discussed in greater detail.
12. According to Dirven (1993: 18), figurative meaning “only arises — or at least can arise
— if the conceptual distance between the two (sub)domains or things referred to is large
enough.”
13. In order not to complicate the picture too much, not every relationship between source
and target or between one source concept and another can be explicated in one single
figure. For example, the contiguity relationship and potential metonymy (CAUSE/EFFECT)
among the source concepts DEFICIENT SENSORY APPARATUS, DEFICIENT HEAD/BRAIN and
DEFICIENT ( PRACTICAL) ABILITIES is not represented in this schema.
References
Klaus-Uwe Panther
University of Hamburg
Linda Thornburg
Eötvös Loránd University, Budapest
1. Introduction
1.1. Metonymy
From the scenario given in (7) for the specific illocutionary type ‘directives’
we can abstract a more general State-of-Affairs Scenario. Speech acts are just
special instances of what we call ‘states of affairs,’ e.g., states, events, pro-
cesses, and actions, which, of course, include linguistic actions. We assume
that the distinction between BEFORE conditions, the CORE, the immediate
RESULT or EFFECT of the core, and the AFTER, which we have already posited for
the Directive Speech Act Scenario in (7), also holds for states of affairs. Thus,
we propose the scenario given in (9) for propositional contents that describe
existing (actual) states of affairs.
(9) The State-of-Affairs (SoA) Scenario
(i) the BEFORE: necessary preconditions: motivations, potentiali-
ties, capabilities, abilities, dispositions, etc., which
can bring about the SoA
(ii) the CORE: the existing/true SoA
the EFFECTS: necessary consequences immediately following
from the SoA
(iii) the AFTER: non-necessary consequences of the SoA
As an example of a state of affairs, consider a description of the act of opening
a window, such as in (10):
(10) Mary opened the window.
A necessary pre-condition (a BEFORE condition) for the existence of the state of
affairs described in sentence (10) is Mary’s ability to bring it about. An even
338 Klaus-Uwe Panther and Linda Thornburg
more basic BEFORE condition would be that it is at all possible (for anyone) to
open the window. Other BEFORE conditions which might be regarded as
specific subcases of the ability condition would be, e.g., that Mary has the
strength, the time, the patience, etc. to open the window (see Panther 1981). If
Mary is not able to open the window or is not strong enough or does not have
the time to open the window, then the window-opening simply does not come
about. Certain mental states also count as BEFORE conditions in our account,
e.g., the desire or intention to open the window, which can be regarded as
motivations for the action. The CORE of the state of affairs is the actual
performance of the action. The immediate EFFECT of a successful act of
opening the window is another state of affairs, that is, that the window is open.
An intended but not necessarily following consequence of opening the win-
dow (the AFTER) might be that fresh air flows through the window. An
unintended consequence might be that flies come through the open window.
In general, the AFTER is only loosely associated with the CORE, whereas the
EFFECT is a necessary part of it.4
The SoA Scenario can be metonymically exploited just like the more
specific Scenario for Directive Speech Acts. For example, in answering the
question (11a) with (11b) the answerer B may exploit the metonymy EFFECT
FOR CORE :
The first conceptual domain that we analyze concerns perceptual events like
seeing, hearing, feeling, tasting, and smelling. All the English sentences from
(12)–(20), apart from examples (15) and (16), are about actually occurring
sense perceptions although, literally, they either question or assert what we
call the BEFORE condition of the Sense Perception Scenario, i.e., they question
or assert the ability for the particular sense perception. For example, in (12)
and (13) the speaker wants to know whether the hearer actually sees someone
or something. In (17) the speaker actually heard the sneering laughter. In (18)
the gambler actually has a sense of being lucky. In (19) and (20) the taste and
smell are actually experienced. In each of these English examples the ability
for the sense perception ‘stands for’ the actual perceiving. Thus the POTENTI-
ALITY FOR ACTUALITY metonymy is fully exploited for the expression of sense
perceptions in English.6
(a) see
(12) Can you see him?
Látod? (Keresztes 1992: 34)
see-2.SG.PRES.IND.DEF.
(13) Can you see well?
Jól látsz? (Keresztes 1992: 34)
well see-2.SG.PRES.IND.INDEF.
(14) I can’t see the movie screen while you have that hat on.
(Searle 1975)
Nem látom a mozivásznat a kalapodtól.
not see-1.SG.PRES.IND.DEF. the moviescreen-ACC. the hat-
2.SG.POSSESS.-FROM/WITH
340 Klaus-Uwe Panther and Linda Thornburg
In the conceptual domain of mental processes and states, English again fully
exploits the POTENTIALITY FOR ACTUALITY metonymy, as can be seen in sen-
tences (21)–(24). The second sentence in (21) strongly implies that the
speaker is actually in a state of understanding. In (22) the speaker actually
remembers the past event. In (23) the speaker conveys his actual belief that
Steve is guilty and in (24) he actually imagines how it happened.
(a) see (= ‘understand’)
(21) I can’t let her down just like that, yet one day it will have to come.
I can see that now. (LOB)
…Most már látom.
…now already see-1.SG.PRES.IND.DEF.
342 Klaus-Uwe Panther and Linda Thornburg
(b) remember
(22) I can remember when we got our first TV.
Emlékszem, mikor…
remember-1.SG.PRES.IND.DEF. when…
(c) believe
(23) Mary can’t believe that Steve is guilty, but I can.
Mari nem hiszi, hogy Pista b²
unös, de én igen.
Mary not believe-3.SG.PRES.IND.DEF. that…
(d) imagine
(24) I can imagine how it happened.
El tudom képzelni, hogyan…
PRT able-1.SG.PRES .IND.DEF. imagine-INF . how…
For the limited data we have for Hungarian the situation is less clear. Although
the Hungarian sentences in (21) and (22) have no modality markers, (24) does
contain the modal verb tud ‘be able’ in the Present Indicative. Sentence (24)
thus instantiates the POTENTIALITY FOR ACTUALITY metonymy. Example (23) is
unclear because in the English sentence can is negated. Our Hungarian
informants opted for a translation without a modality marker but they also told
us that reversing the sequence of the clauses to I can believe that Steve is
guilty, but Mary can’t would allow the modality affix -het- ‘possible’ in the
relevant metonymic sense.
To conclude this section of our analysis regarding the mental process/state
verbs, the English examples strongly suggest that the POTENTIALITY FOR ACTU-
ALITY metonymy is again fully exploited to render the pragmatic effect of
actuality. However, for Hungarian, despite the fact that the modality markers
co-occur with some verbs of mental states/processes and apparently have the
relevant metonymic function, the POTENTIALITY FOR ACTUALITY metonymy
seems to operate in this conceptual domain only to a limited extent. We now turn
to what we call speech act metonymies; the first type we consider is ‘hedged
performatives,’ i.e., explicit performative utterances that are hedged by modal
or modal-like expression (cf. Fraser 1975; Bach and Harnish 1979).
The Potentiality for Actuality Metonymy in English and Hungarian 343
(34) Mary, you can peel the potatoes; John, you can set the table.
Mari, te megpucolhatnád a krumplit, Jani, te megteríthetnéd az
asztalt.
Mary you PRT.-peel-POSS.-2.SG.PRES.COND.DEF. the potato-ACC.;
John, you PRT.-set-POSS.-2.SG.PRES.COND.DEF. the table-ACC.
(35) You can forget it.
Elfelejtheted.
PRT.-forget-POSS.-2.SG.PRES.IND.DEF.
The same analysis holds true for Hungarian but with one distinction: the
grammaticalization of the modality marker -hat-/-het- is much stronger in
Hungarian than that of can in English. This means that no literal ‘ability’
interpretation is available in Hungarian; in other words, the request interpreta-
The Potentiality for Actuality Metonymy in English and Hungarian 347
in (38) where ability occurs in a past time frame, actuality is strongly impli-
cated, though in some contexts it is cancelable.
For the English and Hungarian sentences in (39), where ability occurs in
a future time frame, the metonymic link between potentiality and actuality is
less strong. In Hungarian the verb tud in (39a) focuses on the ability reading in
contrast to (39b), which according to one informant, focuses on the commit-
ment of the speaker to a future action. In other words, the actuality effect in
(39a) is very weak and easily cancelable, whereas in (39b) it is fairly strong
and therefore difficult to cancel (cf. ‘commissives’ in Section 2.3.2.). In fact
the Hungarian sentences in (39a, b) can be conjoined without an effect of
redundancy. The pragmatic sense of the resulting conjunct is paraphraseable
as I can come to your party and I will.
(38) John was able to finish his paper before the deadline.
John a határid²
o el²
ott be tudta fejezni a dolgozatát.
John the deadline before PRT. able-3.SG.PAST.IND.DEF. finish-INF.
the work-ACC.
(39) I can come to your party on Friday.
a. El tudok jönni a pénteki bulidra.
PRT. able-1.SG.PRES .IND.INDEF. come-INF. the Friday-ADJ. party-
YOUR-TO
b. Eljöhetek…
PRT.-come-POSS.-1.SG.PRES.IND.INDEF.…
tud, the focus particle is, and the infinitival form of the verb lenni ‘to be.’ The
metonymic effect of occasional actuality is not cancelable, e.g., it is impos-
sible to assert at the same time that Dogs can be dangerous but have never
been observed to be dangerous.
(40) Dogs can be dangerous.
a. A kutyák veszélyesek is tudnak lenni.
the dog-PL. dangerous-PL. PRT. able-3.PL.PRES.COND.INDEF. be-
INF.
b. A kutyák veszélyesek lehetnek.
the dog-PL. dangerous-PL. will-be-POSS.-3.PL.PRES.COND.INDEF.
(41) He can be very unfriendly.
Néha nagyon barátságtalan is tud lenni.
sometimes very unfriendly PRT. able-3.SG.PRES.IND.INDEF. be-INF.
(42) She can be very generous.
Nagylelk²u is tud lenni.
very-generous PRT. able-3.SG.PRES.IND.INDEF. be-INF.
The last metonymy that we briefly explore in this paper and illustrate in (43)–
(45) concerns skills or acquired abilities. This metonymy is similar to the
character disposition metonymy in that the link between potentiality and
display of behavior is restricted to certain occasions. Since skills are acquired
and tend to be exercised, we expect to find in language use a strong met-
onymic link between the ability to exercise a skill and its actual use. For
example, if you can speak five languages then, by implicature, you will
probably actually speak them. In English this metonymy is systematically
exploited but seems to be only partially exploited in Hungarian.
(43) Can the baby walk now?
Jár már a baba?
walk-3.SG.PRES.IND.INDEF. the baby
(44) Mary can speak five languages.
Mari öt nyelven beszél.
Mary five language-ON speak-3.SG.PRES.IND.INDEF.
350 Klaus-Uwe Panther and Linda Thornburg
3. Results
The results of our analysis are summarized in Tables 1 and 2. The tables
include the modality forms, information for each conceptual domain regard-
ing whether or not and to what extent the POTENTIALITY FOR ACTUALITY
metonymy is exploited, and how strong the metonymic link from potentiality
to actuality is. We describe the degree of exploitation as ranging from ‘sys-
tematic,’ ‘weakly exploited,’ ‘not exploited,’ to ‘not exploitable.’ The strength
of the metonymic link is also indicated in scalar terms: from ‘enforced,’ i.e.,
non-cancelable and grammaticalized; ‘strong,’ i.e., cancelable in some con-
texts; ‘fairly strong’; ‘weak’; to ‘zero.’ With regard to indirect speech acts, the
tables also present the degree to which the modal has been grammaticalized as
a marker of illocutionary force. Interesting contrasts in exploitation and de-
gree of metonymic link are highlighted in bold face.
Two major contrasts between English and Hungarian emerge from our
study. The most important difference can be seen in the conceptual domains of
sense perceptions and mental states/processes. First, the degree of exploitation
of the POTENTIALITY FOR ACTUALITY metonymy in these two domains is nearly
opposite: English exploits the metonymy systematically, whereas in Hungar-
ian it is either systematically blocked (not exploitable) or only weakly ex-
ploited. Second, the metonymic link between potentiality and actuality is very
strong in English, i.e., almost not cancelable, whereas in Hungarian, it is either
non-existent (zero) or only weakly present.
Another contrast, more subtle yet highly interesting, occurs in the domain
of indirect directives. In both languages statements or questions containing
modality markers for ability/potentiality are systematically used as requests.
However, an important difference between the two languages is that in Hun-
garian this practice has become grammaticalized, whereas in English there is
still the possibility of canceling the request reading (cf. Searle 1975). As we
have shown in Section 1.2. (also see Thornburg and Panther 1997), the
hearer’s ability to perform the requested action is a BEFORE condition in the
Request Scenario and can stand for the whole scenario or other parts of the
The Potentiality for Actuality Metonymy in English and Hungarian 351
scenario such as the AFTER condition that the hearer will actually perform the
action. Notice, however, that the metonymic link between potentiality and
actuality for indirect directives is relatively weak. Unlike commissives, whose
satisfaction is guaranteed by the speaker, requested actions are hearer-depen-
dent, i.e., not under the speaker’s control, and therefore not guaranteed.
In this section we will briefly touch upon the problem of how the POTENTIAL-
ITY FOR ACTUALITY metonymy relates to two well-known pragmatic principles,
namely Grice’s (1975) First and Second Maxims of Quantity: Make your
contribution as informative as is required (for the current purpose of the
exchange) and Do not make your contribution more informative than is
required. The POTENTIALITY FOR ACTUALITY metonymy seems to be in conflict
with the first maxim. The first maxim triggers the implicature that if a speaker
says that something is potential or possible then it is not actual. A speaker
saying for example
(46) Professor Smith may be in her office.
conveys that she cannot make the stronger statement
(47) Professor Smith is in her office.
In such cases the First Maxim of Quantity seems to take precedence over the
POTENTIALITY FOR ACTUALITY metonymy. In other words, the route from
potentiality to actuality is barred: may be definitely cannot stand for is because
the first quantity maxim triggers the implicature that the speaker is not certain
about the actuality of the proposition.
However, in a sentence like
(48) I could hear him rummaging around in the basement.
the POTENTIALITY FOR ACTUALITY metonymy licenses the interpretation ‘I heard
him rummaging around in the basement.’ This interpretation can also be
derived in a Gricean framework, namely, via the Second Maxim of Quantity
(cf. what Horn 1989: 194ff calls an R-based implicature). The problem of
which maxim takes precedence in cases of conflict is a complex issue that is
beyond the scope of this paper (cf. Horn 1989: Chs. 3ff for detailed discussion).
The Potentiality for Actuality Metonymy in English and Hungarian 353
There are however cases where the First Maxim of Quantity and the
POTENTIALITY FOR ACTUALITY metonymy (or Grice’s Second Maxim of Quan-
tity) are not in conflict but seem to interact. Consider (49) (from Sweetser
1990: 70):
(49) He may be a university professor, but he sure is dumb.
Sweetser remarks that, despite the modality, the first clause states, or rather
concedes, that the subject is a professor.12 In our terminology, the POTENTIAL-
ITY FOR ACTUALITY metonymy is operative in example (49). Yet the first clause
of the utterance additionally conveys a reservation, which the non-modal
predicate is a university professor would not imply. That is, to make the
actuality statement
(50) He is a university professor,…
is to evoke stereotypical professorial attributes in the mind of the hearer, e.g.,
that professors are highly educated, intelligent, well-read, etc., which can then
be canceled by a subsequent but clause. However, in using may be, as in (49),
the speaker signals from the very start, within the initial clause, that some
stereotypical implicature(s) such as ‘he is intelligent’ or ‘he is well-read’ cannot
be anticipated. Thus, in contrast to the actuality predicate is a professor, the use
of may be a professor contravenes a possible conclusion that would be
warranted on the basis of the stereotypical assumptions held about professors.
In the first clause of (49) the POTENTIALITY FOR ACTUALITY metonymy sanctions
the conclusion that the subject actually is a professor, whereas the First Maxim
of Quantity licenses the inference that he is a professor to less than the fullest
extent, an interpretation which is confirmed by the but clause.
Likewise, in the conventionalized metaphorical use of see ‘understand’
the potentiality marker can in sentences like
(51) I can see your point.
seems to have the same dual function. On the one hand, the POTENTIALITY FOR
ACTUALITY metonymy yields the interpretation ‘I see your point,’ on the other
hand, since the stronger statement I see your point has not been made, the First
Maxim of Quantity licenses the implicature ‘…but I have reservations.’ In
uttering I can see your point the speaker gives to understand that she cannot
make the stronger statement I see your point because understanding to the
fullest possible extent has not occurred.
354 Klaus-Uwe Panther and Linda Thornburg
In examples like (49) and (51) the conflict between the First Maxim of
Quantity and the POTENTIALITY FOR ACTUALITY metonymy is resolved in a
compromise: actuality is conceded but not to the fullest extent. In general then,
utterances with a potentiality marker like can — even when they are used
metonymically to evoke actuality — seem to be weaker than their correspond-
ing non-modal counterparts.
To summarize, the relationship between conceptual metonymies and
Gricean implicatures is far from being clear at the present stage of research.
In recent work Kövecses and Radden (1998) and Radden and Kövecses (this
volume) have proposed that the choice of metonymic vehicles for accessing
various targets is not random. They postulate default metonymies that repre-
sent the normal, i.e., preferred, way of indicating an intended target. In their
framework the cognitive principle ACTUALITY OVER POTENTIALITY is a default
case whereby ‘actual’ vehicles are selected to access ‘potential’ targets rather
than the other way around. They interpret our data from indirect speech acts as
exceptional to their default principle since these data instantiate a metonymy
that links a ‘potential’ vehicle to an ‘actual’ target. They argue that these
‘exceptions’ result from a conventionalized flouting of their principle CLEAR
OVER LESS CLEAR, a flouting that they claim is motivated by social norms such
as politeness. In this way they preserve the primacy of their cognitive default
principle ACTUALITY OVER POTENTIALITY.
However, our research calls into question the validity of the proposed
default principle ACTUALITY OVER POTENTIALITY. As we have amply demon-
strated in this paper, in addition to indirect speech acts, at least six other
cognitive domains exist in which the POTENTIALITY FOR ACTUALITY metonymy
is systematically exploited in English and to varying degrees in Hungarian.
Unlike indirect speech acts, the tokens of POTENTIALITY FOR ACTUALITY in
these other domains cannot be accounted for in terms of social conventions, in
particular, politeness principles. The use of an utterance such as I can smell the
garlic over I smell the garlic is not motivated by rules of politeness. The
pervasiveness of the POTENTIALITY FOR ACTUALITY metonymy in both English
and Hungarian, its systematic exploitation, and its degree of conventionaliza-
tion below the level of conscious awareness pose a serious challenge to a
The Potentiality for Actuality Metonymy in English and Hungarian 355
6. Conclusion
Our pilot study was motivated by a curiosity to see how conceptual metony-
mies are realized in two very different languages. By using a scenario ap-
proach to describe states of affairs and choosing one basic metonymic
principle, the POTENTIALITY FOR ACTUALITY metonymy, we were able to deter-
mine (i) that this particular conceptual principle is more systematically ex-
ploited in English than in Hungarian in various cognitive domains and (ii) that
it is pervasive in disparate cognitive domains, a finding that suggests it is a
default rather than a non-default cognitive principle.
We believe that future research should cover a larger sample of languages
in which additional conceptual metonymies and cognitive domains are ex-
plored. Moreover, such an extended research program should elucidate the
relationship between Gricean implicatures and the kind of metonymic infer-
ences discussed in this paper.
Notes
which instantiate the POTENTIALITY FOR ACTUALITY metonymy and asked them to give a
natural translation in Hungarian. With one exception — the sentence in (38) with be able
— the English data occur with the modal auxiliary can; the Hungarian translations
contain, where possible, either the affix -hat-/-het-, which denotes possibility in a general
sense, or the verb tud, which is roughly equivalent to ‘be able, know how.’ In particular,
we wanted to know whether the ‘possibility’ and ‘ability’ modality markers -hat-/-het-
and tud conveyed the same metonymic effect (i.e., implicating actuality) as in the English
examples. We would like to thank various friends, colleagues and students for their time
and assistance in providing most of the Hungarian data: Andrea Szirmai, Éva Szabó,
Balazs Lövenberg, Zoltán Kövecses, and especially Rita Brdar-Szabó, who supplied
many helpful grammatical points.
6. For the glosses of the Hungarian examples the following conventions will be used: words
are separated by a space; morpheme boundaries are indicated by a hyphen; grammatical
functions are indicated in small capitals. The abbreviations used are: ACC = accusative,
ADJ = adjectival suffix, COND = conditional, DEF = definite conjugation, IMP = imperative,
IND = indicative, INDEF = indefinite conjugation, PART = participle, PL = plural, POSS =
possibitity, POSSESS = possessive, PRES = present tense, PRT = particle, SG = singular.
7. The abbreviation LOB stands for Lancaster-Oslo-Bergen Corpus.
8. In examples (15) and (16) the -ható complex may be regarded as an adjective-forming
suffix deriving ‘possibility-passive’ adjectives from verbal bases. We are grateful to Rita
Brdar-Szabó for providing the arguments in support of this claim and regard this analysis
as additional support for ours.
9. The factors that determine whether a hedged performative can actually be used to
perform the illocutionary act denoted by the performative verb are complex and will not
be dealt with in any detail. Suffice it to say that I can ask you to leave does not constitute
an act of asking nor does I must promise to be there constitute a promise.
10. With indirect commissives (offers) like (30), our Hungarian language consultants often
preferred the more indirect conditional over the plainer indicative mood for reasons of
politeness. This preference also holds for indirect directives.
11. We are grateful to Rita Brdar-Szabó for making this point clear to us.
12. This is an example of what Sweetser calls ‘speech act modality.’ Unlike (46), utterance
(49) is not available as a topic-introducing or discourse-initiating device and seems to be
restricted to a reactive slot in a discourse.
References
Fraser, Bruce
1975 Hedged performatives. In P. Cole, J. Morgan (eds.), Syntax and Semantics.
Vol. 3: Speech Acts. New York: Academic Press, 187–210.
Grice, H. Paul
1975 Logic and conversation. In P. Cole, J. Morgan (eds.), Syntax and Semantics.
Vol. 3: Speech Acts. New York: Academic Press, 41–58.
Horn, Laurence R.
1989 A Natural History of Negation. Chicago: The University of Chicago Press.
Keresztes, László
1992 A Practical Hungarian Grammar. Debrecen: Debreceni Nyári Egyetem.
Kövecses, Zoltán, Günter Radden
1998 Metonymy: developing a cognitive linguistic view. Cognitive Linguistics 9:
37–77.
Lakoff, George
1987 Women, Fire and Dangerous Things: What Categories Reveal about the
Mind. Chicago: The University of Chicago Press.
Panther, Klaus-Uwe
1981 Indirect speech act markers or why some linguistic signs are non-arbitrary.
In R.A. Hendricks et al. (eds.), Papers from the Seventeenth Regional
Meeting. Chicago Linguistic Society, April 30 - May 1, 1981. Chicago:
Chicago Linguistic Society, 295–302.
Panther, Klaus-Uwe, Linda Thornburg
1998 A cognitive approach to inferencing in conversation. Journal of Pragmatics
30: 755–769.
Searle, John
1969 Speech Acts: An Essay in the Philosophy of Language. Cambridge: Cam-
bridge University Press.
1975 Indirect speech acts. In P. Cole, J. Morgan (eds.), Syntax and Semantics.
Vol. 3: Speech Acts . New York: Academic Press, 59–82.
Sweetser, Eve
1990 From Etymology to Pragmatics: Metaphorical and Cultural Aspects of
Semantic Structure. Cambridge: Cambridge University Press.
Thornburg, Linda, Klaus-Uwe Panther
1997 Speech act metonymies. In W.A. Liebert et al. (eds.), Discourse and
Perspectives in Cognitive Linguistics. Amsterdam and Philadelphia:
Benjamins, 205–219.
Part IV
Applications of Metonymy
“Mummy, I like being a sandwich”
Zazie Todd
University of Leicester
1. Introduction
The first use of the term ‘metonymy’ can be found in ancient Greek philoso-
phy, and more specifically in the famous debate about the arbitrariness or
naturalness of signs. In his account of Plato’s contribution to linguistics, Fred
Householder (1995: 93) points out that
Democritus (as quoted in Proclus’ commentary on the Cratylus 16) offered
four arguments (with four specially coined names) in favour of arbitrariness:
(a) ‘homonymy’ or ‘polysemy,’ i.e., the same sequence of phonemes may be
associated with two or more unrelated meanings; (b) ‘polyonymy’ or ‘isorro-
phy,’ i.e., the existence of synonyms; (c) ‘metonymy,’ i.e., the fact that words
and meanings change; (d) ‘nonymy,’ i.e., the non-existence of single words
for simple or familiar ideas.
362 Brigitte Nerlich, David D. Clarke and Zazie Todd
shrinking’ despite the fact that the children could express the same meaning
with the words they know. Later still, in adults, we can observe the use of
metonymies in, literally speaking, cost-effective communication, as for ex-
ample in headlines like “Brains of British Museum crack Crystal Skull
Riddle” (The Observer 26/6/96). After England’s victory over Holland in the
Euro 96 football championship one could read: “Our Boys done Gouda” (The
Sun, June 1996). We understand these headlines instantly, and yet they require
a lot of linguistic, contextual and conceptual ‘unpacking.’
2. Typologies of metonymies
Studies of how children produce and, especially, how they understand meta-
phor, proliferate in the literature on language acquisition in general and
semantic development in particular. However, no work has been done up to
now on metonymy in language acquisition, and the possibility of having
metonymic overextensions in child language has hardly ever been noted
before (cf., however, Melissa Bowerman’s work, e.g., Bowerman 1978;
Gibbs 1994).
Our first hypothesis was that metonymical relations might be exploited in
overextensions produced by children up to age 2;5. We called these ‘com-
pelled metonymical overextensions’ because they are based on the fact that at
this age a child’s vocabulary, category and conceptual systems are still rela-
tively small and unstructured. This scarcity compels them to extend already
known words to cope with increasing communicative needs, to comment on
what they see and to request what they want.
A good overview of the research done on over- and underextensions is
provided by Eve Clark in her book The Lexicon in Acquisition. She writes:
‘Mummy, I Like Being a Sandwich’ 365
Children do not start out already knowing the meanings of the words they are
attempting to pronounce and use. They must first assign some meaning to
them, and that takes time. In fact, children’s earliest mappings of meanings
onto forms diverge from adult usage in a variety of ways. (Clark 1993: 32)
One of the most important ways of mapping the meaning a child wants to
express onto a form which he or she has already acquired is overextension,
that is, by applying a word to members of the adult category and to members
of other categories that are perceptually similar, conceptually contiguous or
spatio-temporally related. Consider these examples:
(1) a. ball for balls of all kinds, and also for round hanging lampshades,
doorknobs, and the moon [based on perceptual similarity]
b. door for corks, jar-lids, box-lids, and gates when wishing to
have the relevant object opened or closed [based on functional
similarity]
Eve Clark distinguishes between two types of overextensions: (i) ‘overin-
clusions,’ where children extend a term to other entities from the same
taxonomy, e.g.:
(2) a. dada used for both father and mother
b. baby used for self-reference and all children
c. apple for apples and for oranges
and (ii) ‘analogical extensions,’ where children use a term for objects from
other taxonomies on the basis of perceptual similarity, e.g.:
(3) a. cotty-bars for abacus on the wall and picture of a building with
a columned façade
b. comb for centipede
Whereas analogical overextensions are based on recognizing and construing
similarities (and can therefore be regarded as the training-ground for use of
metaphors), overinclusions are based on both perceptual similarity and con-
ceptual contiguity (we shall later call them synecdochical overextensions, and
they are the basis for the construction of taxonomies).
That is, the words being extended are applied to instances of other categories
within the same or an adjacent conceptual domain. For example, a child may
over-extend a term within the domain of animals, as when horse is applied to
goats, cows, and sheep or within the domain of vehicles, as when truck is
applied to buses, tractors, and vans; or within the domain of clothing, as when
hat is applied to a crown. (Clark 1993: 34)
366 Brigitte Nerlich, David D. Clarke and Zazie Todd
However, it seems to us that Eve Clark should have discussed in more depth a
third type of overextension based not so much on conceptual contiguity, but
on spatio-temporal or functional contiguity. As we shall see, such contiguities
are based on visual perception and are therefore the precursors to actual
metonymies.
This type of overextension has been studied from Piaget and Vygotsky
onwards under the heading of ‘complexive associations.’ Those studying
complexive associations have claimed that object names initially encode
complexive groupings that incorporate actions or locations associated with
objects (cf. Huttenlocher and Smiley 1993: 240).
Even the words themselves are said to be initially just auditory features of
situations where they habitually occur [...]. In short, it has been suggested that
young children group their experiences in a fundamentally different way than
older children or adults — in terms of co-occurring aspects of situations —
and that object names, rather than standing for particular types of objects, are
just another type of associate. (ibid.: 222)
Huttenlocher and Smiley contest this claim and argue that the extension of
children’s early object words match those of adults to a large extent. However,
after distinguishing more clearly between a word’s extension or meaning and
its use, they found that in certain cases other than naming contexts, such as
comments or requests, such ‘overextensions’ did occur; as when a child said
toy when seeing a certain bag which habitually contained toys, or when a child
said shoe when she saw a foot (ibid.: 226f). This means that children use
certain ‘metonymically’ based pragmatic strategies, just as adults do, so as to
achieve certain communicative effects.
When a child comprehends a word correctly but overextends it in production,
Clark [1978] concludes the overextension is not the result of an immature or
nonadult word concept but is rather a deliberate attempt to communicate
concepts for which the appropriate word cannot be remembered. (Pease et al.
1989: 113)
b. Matthew (5;0)
Matthew was wearing a pullover that went down to his knees — we
laughed and said it was a mini-dress, whereupon he pretended to be a
girl and started to curtsey. As he was not very good at it, he practiced
all afternoon. At bed-time he pulled off the pullover and said “I have
enough of it, I have been wearing this curtsey all day.”
c. Matthew (5;1)
Matthew was playing with his Playmobil operating theater. He was
‘Mummy, I Like Being a Sandwich’ 371
d. Matthew (5;1)
While eating cheese on biscuits, he said: “I am eating a bone-sand-
wich.” — “Why?” — “Because cheese is good for your bones.”
e. Matthew (5;3)
We were in Germany, Easter-Sunday, Matthew was trying to find his
chocolate eggs, rabbits, toys, etc. Having filled up his basket to
overflowing, he looked round and said: “Are there any more Easters
to find?” Later he repeats the same sentence in German: “Sind da noch
ein paar Ostern?”
f. Matthew (5;3)
We were at home, Matthew tried to push me out of the house and said
“Uff,” then “I uffed you really hard,” and a bit later “I am pushing
you.” [Uff is a German interjection expressing effort.]
g. Matthew (5;4)
We were playing postman. Matthew told me to go into the living room
and close the door so that he could post a letter to me underneath it. I
forgot to close the door. Matthew came in with his letter and shouted:
“Mummy, where’s your door?”
h. Matthew (5;4)
Despite all our efforts to prevent such ‘utterances’ — here it is:
Holding out his finger as a gun, Matthew said “Mummy, I want to
have a toy gun, I don’t want the baddies just to be fingered.” And, on
a similarly macabre line, playing with a police car and a fire engine, he
said: “I’ll put the hose there so that I can spray people with water to
see if they wake up when they are not dead” and “I’ll put this [a
different hose attachment] on the police car to gun people.”
i. Matthew (5;4)
Matthew watched Daddy build a climbing frame. First came the
normal nuts and bolts, then the dome nuts, whereupon Matthew said
“Oh dear, now you have to do another lot of nutting.”
372 Brigitte Nerlich, David D. Clarke and Zazie Todd
j. Matthew (5;4)
After watching a program called “The Treacle People,” we concocted
a drink for him, consisting of orange, apple and passion-fruit juice,
plus orange squash, plus water which Matthew then baptized
“Treacle-people juice.” And he now refers to this kind of juice quite
commonly in sentences like “Mummy, not treacle-people juice
again!”…
k. Matthew (5;4)
Matthew was making ‘potions,’ that is, mixing various juices and
water and filling them into pots. The first was a potion against
earache, the second, he decided, was “another earacher.”
l. Matthew (5;4)
This spring was the coldest one on record, so when I dared to wear a
T-shirt one Saturday instead of a woolly jumper, Matthew looked at
me with big eyes and said: “Mummy why are you all summered?”
Later in the day, he put on a T-shirt himself and said: “Look, I am all
summered too now!”
m. Matthew (5;5)
Playing with Lego, building a film studio, using a fire engine exten-
sion ladder as a camera stand, he said: “They want to camera them.”
n. Matthew (5;5)
Matthew was listening to pop music on the radio, started jumping
around, waving hands and feet, and said: “Mummy, I am drumming
and guitaring at the same time.”
5. Understanding metonymies
Although we would in no way dispute the fact that children have “some ability
to think metonymically,” there seems to be a gap between the production and
understanding of metonymies (a well-known asymmetry in language acquisi-
tion). As far as the observation of Matthew can be regarded as evidence, we
found that he was quite happy to produce metonymies, first in overextensions
(although that was before his utterances were recorded), then in a more
creative way, but that he did not readily understand conventional or dead
metonymies or accept novel metonymies uttered by adults. At an older age,
the following examples of his non-comprehension were noted:
(9) a. Matthew (4;5)
When Matthew was 4;5 the trains were on strike. On our way back
from the childminder we used to walk over a railway bridge and watch
the trains rush through the tunnel. However, in the autumn of 1994 we
were standing there and nothing came. When Brigitte realized what
was going on she said: “The trains are on strike.” Matthew then
asked “What are they doing, can you explain?”
b. Matthew (4;1)
At around the same time Brigitte read a story to him entitled “Jake and
the jumble panic.” This time Brigitte asked him: “Did you understand
this?” and he said that he did not, so she explained to him what “I am
Electrical” meant (i.e., I am the man that sells the electrical goods) —
and now this is one of his favorite stories, he really savors it.
c. Matthew (4;1)
Matthew came home from school and announced that there would be
a table-top sale on Saturday morning — and he would like to go there
because he really could do with another table top (the week before we
‘Mummy, I Like Being a Sandwich’ 375
had just bought him a ‘table-top,’ that is, a very wide shelf that can
also serve as a drawing-table). In this case he had not learnt the
meaning of a dead or conventional metonymy yet. In the next case he
understood, but did not accept a novel metonymy.
d. Matthew (4;4)
Matthew came home from school. I said: “Wow, you have eaten
your whole lunch box!” Matthew burst out laughing and said: “Oh
Mummy, you got that wrong — it’s: You have eaten everything out of
your lunch box!”
The examples given above tell us a lot about one child’s development of
metonymy comprehension, but it is critical also to collect experimental data
from a larger group of children. An embedded story framework was used to
test the comprehension of metonymy by children in two age groups, and to see
whether the presence of clues in the text would help to improve understanding.
5.1.1. Method
The participants in this experiment were sixteen children at the local univer-
sity play group. Ten children were in the two and three-year-old group, and six
in the four and five-year-old group. There were approximately equal numbers
of girls and boys, and only children whose first language was English took
part in the study.
Two short stories were used as stimuli. One was “Jake and the Jumble
Panic,” taken from a ‘Ladybird’ book by Joan Stimson (Stimson 1992). This
story was selected because it contained four metonymies. The second story
was about some pirates and a treasure hunt, and was written especially for the
experiment. Although this story contained a total of sixteen metonymies, only
six were used as test stimuli in the experiment. Both stories were deemed
suitable for the age group, and none of the children had heard them before.
The metonymies in the stories were independent, such that failure to under-
stand one would not in itself cause problems in understanding the others. A
copy of the Pirate Story can be found in the Appendix.
Children were tested in small groups of two or three. The experimenter
read the two stories, pausing each time one of the metonymies was reached.
Children were then asked to point to which of two pictures showed what was
happening in the story. One of the pictures was always a literal (incorrect)
interpretation of the metonymy, while the other picture correctly depicted the
story. The pictures were presented concurrently, with each taking up half of a
sheet of writing paper. The ordering of the pictures was random so that
children could not consistently point in the same direction to give the correct
answer. The pictures were presented in such a way that no child could see
which picture the other child had chosen, i.e., they had to make the choice for
themselves. After a picture had been chosen, the experimenter continued with
the story without informing the children whether they had been correct.
‘Mummy, I Like Being a Sandwich’ 377
4 Comprehension
0
2-3 years 4-5 years
old old
children were therefore analyzed according to whether or not there were clues
present in the text. There were four metonymies with clues, and six without.
The results showed that on average 72% of the metonymies were under-
stood when clues were present, compared to 40.5% when there were no clues.
A Wilcoxon test (Wilcoxon T = 9, p < 0.01) showed that there was a
significant improvement in comprehension for the metonymies for which
clues were available in the text compared to ‘stand-alone’ metonymies. A
graph showing these results can be found in Figure 2.
80
70
60
50
40 Comprehension
30
20
10
0
No clues Clues
This experiment shows that the four and five-year-olds have a signifi-
cantly better understanding of metonymy than the two and three-year-olds,
which is what might be expected from what has been said elsewhere in this
paper and from experimental studies of metaphor comprehension (e.g.,
Vosniadou 1987). In addition this experiment shows that children are able to
use clues in the text of a story to help them to understand metonymies. More
evidence would be needed to establish whether clues such as this enable
children to develop some understanding of metonymy in the first place, or if
they are just used in some instances once children already have some basic
knowledge of how metonymy works. One observation that could be worth
following up is that for the metonymy Twenty sailor hats were marching down
the gangplank, most of the children who correctly selected the picture of
‘Mummy, I Like Being a Sandwich’ 379
sailors pointed to the sailors hats rather than another part of the picture. The
children may have realized that hats cannot march on their own and used that
as the basis for selecting a picture (although clearly there are many children’s
cartoons and books in which it is possible for inanimate objects to march, walk
and even talk). It may be that children’s general cognitive development or
understanding of the real-pretend distinction is as important as any under-
standing of metonymy itself. Further experiments could also be carried out to
investigate other aspects of metonymy comprehension and in particular to see
if there are links between development in comprehension of metaphor and
metonymy.
6. Conclusion
Note
1. An example of a (rather rare) SPECIES FOR GENUS synecdoche can be found on a toy
recently bought at a “Sealife Centre.” It consists of a plastic dolphin, whale, shark or
alligator which you put into water where it expands 200 times in size. On the package it
says: “Put sealife into a pan, sink or pool…” “Sealife can grow to 22 inches, wet” (on
synecdoche in general, cf. Nerlich in print and Nerlich and Clarke in print).
References
Anonymous [Cicero]
1954 Ad C. Herennium, de ratione dicendi (Rhetorica ad Herennium). London:
Heinemann.
380 Brigitte Nerlich, David D. Clarke and Zazie Todd
Barrett, Martyn D.
1982 Distinguishing between prototypes: the early acquisition of the meaning of
object names. In S.A. Kuczaj (ed.), Language Development. Vol. I: Syntax
and Semantics. Hillsdale, NJ: Lawrence Erlbaum, 313–334.
Bowerman, Melissa
1978 Systematizing semantic knowledge: changes over time in the child’s orga-
nization of meaning. Child Development 49: 977–987.
Braine, Martin D.S.
1976 Children’s First Word Combinations. Monographs of the Society of Re-
search in Child Development 41 (Serial No. 164).
Bréal, Michel
1991 [1884] How words are organized in the mind. In G. Wolf (ed.), The
Beginnings of Semantics: Essays, Lectures and Reviews. London:
Duckworth, 145–151.
Clark, Eve V.
1978 Strategies for communication. Child Development 49: 953–959.
1982 The young word maker: a case study of innovation in the child’s lexicon. In
E. Wanner, L.R. Gleitman (eds.), Language Acquisition: The State of the
Art. Cambridge, Massachusetts: The MIT Press, 390–428.
1993 The Lexicon in Acquisition. Cambridge: Cambridge University Press.
Esnault, Gaston
1925 L’imagination populaire: Métaphores occidentales. Paris: Presses Univer-
sitaires de France.
Fass, Dan
1991 Metonymy, case role substitution and sense ambiguity. In D. Fass et al.
(eds.), Proceedings of the IJCAI Workshop on Computational Approaches
to Non-Literal Language: Metaphor, Metonymy, Idiom, Speech Acts and
Implicature, held at the Twelfth International Joint Conference on Artificial
Intelligence (24 August 1991, Sydney, Australia), 42–51.
Gibbs, Raymond W.
1994 The Poetics of Mind: Figurative Thought, Language, and Understanding.
Cambridge: Cambridge University Press.
Householder, Fred W.
1995 Plato and his predecessors. In E.F.K. Koerner, R.E. Asher (eds.), Concise
History of the Language Sciences. Oxford: Pergamon, 9–93.
Huttenlocher, Janellen, Patricia Smiley
1993 Early word meanings: the case of object names. In P. Bloom (ed.), Lan-
guage Acquisition. New York: Harvester and Wheatsheaf, 222–247. (First
published 1987)
Lakoff, George, Mark Johnson
1980 Metaphors We Live By. Chicago: The University of Chicago Press.
Leopold, Werner F.
1939 Speech Development of a Bilingual Child: A Linguist’s Record. Vol. 1:
Vocabulary Growth in the First Two Years. Evanston, IL: Northwestern
University Press.
‘Mummy, I Like Being a Sandwich’ 381
Appendix
The Pirate Story used in the comprehension experiment: ‘The metonymic treasure hunt.’
Tom and Mary had been promised a special adventure for their birthday. For one day they
would be able to live and work on a pirate ship. When they came to the harbor, dressed up
as pirates, there she was, the Black Princess swaying gently on the waves. “What a
beauty,” said Tom. “But where are the pirates?” asked Mary. And there they came —
twenty sailor hats were marching down the gang plank, stood in a row before them and
saluted them. Mary and Tom wanted to greet them, but then they realized that they didn’t
know their names. “What are your names?” asked Tom. “Har-har, my dears, this is the first
mystery you have to solve,” said the captain of the ship, who had just stepped in front of the
pirates as if from nowhere. “My name is Captain Hatty, by the way,” announced the
Captain. Look at the sailors closely and try to find out what their names are.
Peggy (-leg)
(Eye) Patchy
Stripy
Dotty
Tatty
and
Fatty
Now the children knew the names of the pirates and everybody went aboard. “All hands on
deck,” cried the captain, “set sails!” And off they went. The Black Princess was heading for
adventure. When it was lunch time, the ship’s cook shouted “Twelve bells, Twelve bells,”
come and eat your grub. Everybody sat down in the tiny galley. The cook, Fatty, brought
the meal, and looking round, he said jokingly: “Oh dear, there are two more mouths to feed.
Let’s hope we have enough fish fingers!” To celebrate the arrival of Tom and Mary, the
captain got out a bottle of lemonade and everybody got a glass. But just as they were about
to speak a toast, the ship wobbled a bit and the lemonade tipped over, but what was worse,
the treasure-map, that the captain was just about to show to the children, slid on the floor
and got a bit dirty. Quickly, Tom jumped up and saved it before it got all dusty. The captain
took it and nailed it down on the table, to be on the safe side. After a few days they arrived
on a desert island. Everybody got off the ship and explored the beach. Tom had brought a
camera and said to Mary. “Stand there between the rocks and do a Captain Hook for me!”
Then they all went into the deep dark jungle. The sun shone down and everybody got
thirsty. Mary said: “I wish I could have a Ribena.” “No Ribena here my deary,” said the
captain, “you’ll have to do with this nice fresh spring water.” They all had a drink and went
on. Finally they arrived at the spot where there was a cross on the treasure map and started
digging. They dug and they dug. But they couldn’t find anything. Then Tom had a look at
the map that Dotty had pinned to a tree and said: “But it’s upside down!” The captain took
Dotty by his collar and shook him, saying “Don’t you have any brain! We have wasted half
a day now. Let’s have a rest, before we start the search again.” They all put their heads
down for an hour and then started looking for the treasure again — this time in the right
spot. And after a little digging, the whole crew shouted: “There it is!” They could all see a
‘Mummy, I Like Being a Sandwich’ 383
huge treasure chest. They opened it and found gold coins, jewels and, what a surprise, toys
for Tom and Mary. Everybody was very pleased. They went back to the ship and sailed
home.
Recontextualization of Metonymy in
Narrative and the Case of Morrison’s
Song of Solomon
Anne Pankhurst
University of Edinburgh
1. Introduction
the same time evokes the real world of the reader’s experience. It is more than
a description. The earring has numerous features which refer the reader to
more than one concept. This metonymy also has a number of meta-narrative
functions serving to structure the overall development of thematic material in
the text. Its recurrence in new contexts is an important element of the narra-
tive. It reactivates the reader’s memory and highlights important episodes.
vast scope and liberty to the fancy: it both adorns and invigorates our style.”
But metonymy retains its rhetorical function of persuasion. In this case, it
persuades the reader of who Pilate is, and what the earring is, in the domain of
physical appearance. It also plays an important part in marking and extending
associated conceptual domains within her person.
This well-known literary device can be related to the reader’s understand-
ing of what is intended through an interface that has to be structured if new
elements are to be satisfactorily communicated. Many repetitions may be
needed before metonymies “develop into subtexts that mirror the whole of the
text in which they are embedded and facilitate the text’s interpretation”
(Riffaterre 1990: 21). There must be a prolonged sequence, dispersed through-
out the narrative and weaving in and out of it, forming part of the referential
frame of the text. In this case, the development of the earring as a subtext is a
partial explanation of its effect. The interpretation of what has been under-
stood also depends on the reader recognizing that the conceptualization moves
through different contexts of experience while preserving its unity of domain.
How is this achieved? The metonymic reference is extended from its immedi-
ate context to the whole textual structure, by means of recontextualizations.
For Riffaterre, repetition of a linguistic device is an indicator of its importance
in the narrative, but it is also more than that. It enables positive or negative
marking of the episode in which it occurs. The earring, for example, is
described as ‘foolish’ in Macon’s negative view; but also as ‘wonderful,’
when Pilate’s dominance is unquestioned. There are affective as well as
cognitive effects.
The attempt to create reality or at least a possible world in fiction requires
a more complex function for metonymic references, which are known to be
‘two-faceted.’ The inference that the true referent of an earring is not the
object but its wearer (or indeed further implied characteristics of either the
ornament or the person) is relevant to this discussion. The fictional world must
have its own consistency or truth, understood by the assumed reader in terms
of a real experienced world and a rich personal encyclopedia of knowledge
and beliefs (cf., e.g., Langacker 1987). The single word ‘earring’ contains a
potentially rich chain of associations. Through a series of links, initially
following the metonymic pathway which allows substitution of POSSESSOR
FOR POSSESSION, the earring represents the person wearing it. The earring
attracts attention by its unusual form (a brass box) and therefore its function as
a container, standing for the personal treasure contained, is made salient.
Morrison’s Song of Solomon 389
The first point examined here is the actual description of the earring and the
question of its complexity. Metonymy requires that an object stand for some-
thing with which it has close associations, but this earring is more than a
simple decorative object. It is known to contain a special piece of paper on
which the name ‘Pilate’ is written. It was the only word her illiterate father
ever wrote and is the only family document known to exist. It is therefore a
container for an important legacy and may be read as a means of accessing the
past.
The earring is first mentioned in an account of how members of the
family acquired their names. Macon Dead II recalls the birth and naming of his
sister, Pilate, and his father’s insistence that the name randomly chosen and
copied from the family Bible should be kept.
And it did stay there, until the baby girl turned twelve and took it out, folded
it up into a tiny knot and put it in a little brass box, and strung the entire
contraption through her left earlobe. (Song of Solomon: 21)
The earring is the vehicle of Macon II’s ambivalent feelings about his sister,
Pilate, once “the dearest thing in the world to him” and now “odd, murky, a
regular source of embarrassment.” Her “foolish earring” (Song of Solomon:
22) derives from her ragged clothes and disreputable lifestyle, transferring the
foolishness of the woman to the object by the epithet. Her poverty contrasts
with her brother’s material success and aspirations. Macon tells his son,
Milkman, the secret of the earring’s contents in warning him to stay away
from Pilate who “can’t teach you a thing you can use in this world” (Song of
Solomon: 53). Since Pilate is Macon’s sister and Milkman’s aunt, our curiosity
focuses on trying to understand why both object and wearer are ‘foolish,’ and
we therefore infer that there is a special motivation for this attitude yet to be
discovered.
Wearing the earring is a metonymy for Pilate’s physical appearance, but
it also stands for her family’s origins. It encapsulates her respect for her father,
Morrison’s Song of Solomon 391
her power over her brother and nephew, and her status as matriarch. To
account for it as synecdochic detail, a part for the whole, is insufficient. The
known implications expand the temporal and conceptual domains to include
the past, and other people’s memories besides those of Pilate herself. Within
this frame, the earring acquires the properties of a special message-bearing
object, symbolically representing the history of the wearer’s family.
Riffaterre invokes psychological plausibility in claiming that the impor-
tance of an object as a poetic device is heightened by repetitions, which confer
authority on it. As it is recontextualized, the cumulative effect becomes
stronger. In this case, Macon II narrated briefly the making of an earring out of
a brass box. A second mention of the earring, in another episode, permits the
reader to access another character’s experience and point of view. Pilate’s
single earring enhances her role and status, in the eyes of outsiders as well as
to her own family, but it is of particular importance to her nephew. When the
boy is brought to see her by his friend, Guitar, she has an overwhelming effect
on him. She will exercise great power and become the instrument for
Milkman’s rebellion against social common sense and his father’s wishes. The
earring, although only one of the striking elements of Pilate’s appearance, is
the salient feature of an extraordinary woman.
As they came closer and saw the brass box dangling from her ear, Milkman
knew that what with the earring, the orange and the angled black cloth,
nothing — not the wisdom of his father or the caution of the world — could
keep him from her. (Song of Solomon: 36)
While it could be argued that the earring is only one of the ways in which
Pilate influences Milkman in that episode (others involve food, story-telling,
assumption of seniority and the right to direct him), it is the highlighted
element. The consequences of this visit motivate much of the subsequent
narrative, within the theme of family origins.
At a later stage of the narrative the account of the making of the earring is
expanded with a larger number of details of past history, after the potentially
dangerous relations between Pilate and Milkman have been established. The
story-teller is again Pilate’s estranged brother, Macon II, who relates family
history to Milkman (Macon Dead III). The brother and sister had been made to
abandon their father’s farm when it was forcibly acquired by white farmers
392 Anne Pankhurst
who killed their father, Macon Dead I (Jake). Pilate created the earring herself
during a period of isolation when she and Macon II were hidden for their own
safety by a woman called Circe. The stay in the house, with its intertextual
allusion to the stay of Ulysses and his sailors with the enchantress Circe, is
exile and imprisonment for the two children, deprived of the Edenic environ-
ment of their home. The repetition recreates and expands Macon II’s memory
of the past, but the negative marking of the earring as foolish is absent.
Before they left the farm she’d taken the scrap of brown paper with her name
on it from the Bible, and after a long time trying to make up her mind between
a snuffbox and a sunbonnet with blue ribbons on it, she took the little brass
box that had belonged to her mother. Her miserable days in the mansion were
spent planning how to make an earring out of the box which would house her
name. She found a piece of wire but couldn’t get it through. Finally, after
much begging and whining, Circe got a Negro blacksmith to solder a bit of
gold wire to the box. Pilate rubbed her ear until it was numb, burned the end
of the wire, and punched it through her earlobe. (Song of Solomon:167)
The choice of this special container for the word ensures that by inferen-
tial association the reader will understand that it has several levels of impor-
tance both for the characters in the story, and for the reader. Because the name
Pilate was given by her father, unlike their surname ‘Dead,’ which was
allocated officially and meaninglessly, the name will remain in the box as a
record of her true family history. Like a phylactery or amulet, it is bound to its
wearer and contains a special reminder that justification for power and identity
can be contained in one single word. Pilate’s action of creating the earring
preserves her identity and is her self-affirmation against powerlessness. Para-
doxically, lexical values in the phrases — the scrap of brown paper and the
little brass box — indicate that before the flight, the paper and the box were of
no special significance. They became valuable in the state of exile to which
Pilate and her brother were condemned, and which is structured by using the
earring as a mnemonic device.
Up to this point in the narrative, the earring has been no more than an
object by which Pilate can be identified. As the relationship between the boy
and his aunt grows, the earring acquires still greater significance. The different
implications combine the person with her symbolic role. Pilate is identifiably
different from all other people by the fact that she wears this single earring. As
metonymy, in the relationship of physical and mental contiguity within the
domain of the person, it stands for her sense of family and ancestry, her love
for her dead father, her strangeness in society’s eyes and her personal identity
Morrison’s Song of Solomon 393
Through Milkman’s experience, which verifies the truth of the family stories,
the reader understands that the earring and its contents exist in the past, present
and future of the narrative time. Given the complex spatio-temporal digres-
sions in this story, it functions as a necessary unifying or connective device for
the various concepts it contains (cf. Fauconnier and Turner, this volume).
The power of the earring as a referential frame to the text develops more with
each episode, notably at critical moments in Pilate’s family history. The most
394 Anne Pankhurst
The theories of Jakobson and Riffaterre, which have so far provided a frame to
the consideration of metonymy in fictional narrative, do not attribute a special
role to closure. Yet it is plausible to assume that communication with the
reader must be influenced by the nature of the ending, whether it is open or
Morrison’s Song of Solomon 395
This event returns the earring to the realm of the literal, but does not
negate its metonymic value. Although depersonalized and demythologized, it
still has a function in the structure of the narrative, insofar as it now could
motivate either of two potential outcomes. Because the metonymy has carried
an important amount of reference in the narrative, the first inference drawn is
that the closure is clear and dramatic. The earring is no longer defined as a
treasure; it is a shiny, ownerless thing. The move to anonymity is meaningful
in the thematic context, demonstrating to the reader that the writer’s intention
has been to focus only on Pilate’s life by means of the metonymic referential-
ity of the earring. In this sense, the metonymy has a meta-narrative function as
a mark of fictionality. The other potential outcome is open-ended. We do not
know what has become of the earring, but we can speculate as to its new
functions in some other narrative yet to be constructed.
396 Anne Pankhurst
4. Discussion
In Song of Solomon the earring is part of the physical description of its wearer,
always present and salient in her appearance, therefore she can be identified
by mentioning it. As a part for the whole, within the domain of the person, it is
a synecdochic detail of self-presentation. Its function is not, however, limited
to local use. In the macro-structure of the narrative, the earring is highlighted
by different focalizing voices at dramatic moments. It has been created during
the children’s flight from persecution in their home, it creates the bond
between Milkman and Pilate, it is the focus of Pilate’s matriarchal power
during the funeral of Hagar. It confirms Milkman’s discovery of his family’s
origins, and disappears at the death of Pilate. It becomes a short cut to our
understanding of Pilate’s importance for her family and reactivates memory of
previous events both for characters within the narrative and for the assumed
reader. Thus, it is evidence of a rich conceptual domain extending beyond the
person to family, race and social disempowerment.
The complex use of metonymy in this fictional narrative cannot be
accounted for by a single theory. Jakobson proposed that metonymy, retaining
its character as a figure defined by contiguity, facilitates forward movement of
the narrative when seen in a macro-context. This is not enough to explain how
the reader comes to the understanding that the single word ‘earring’ has the
property of multiple referentiality. The other part of Jakobson’s theory, the
provision of synecdochic detail, is also oriented towards exegesis of the text
rather than the reader’s process of understanding and interpretation. A more
satisfactory analysis is provided if we follow Riffaterre’s view that metonymy
is instrumental in creating subtexts which by force of repetition are crucial to
the reader’s grasp of the fictional world and may provide the basis for an
allegorical reading of the narrative.
Riffaterre defines subtext as a means of creating an actual or potential
relationship between the fictional world and the real world, achieved through
multiple references. When recontextualized, subtexts have a meta-narrational
function. They hold together disparate elements within the macrotext by a
network of inference because the patterns set up match existing patterns of
cognition (cf. Black 1993 for a discussion of similar effects generated by
metaphor). The availability of complex yet coherent multiple reference is of
great importance for the systematically cohesive development of longer narra-
tive (Miller 1985; Schulz 1992). If the fictional world is to be perceived as
Morrison’s Song of Solomon 397
true, albeit possessing fictional truth, the reader requires a number of entry
points into this world, created by means of reactivating what is already known
or believed in the real world. Understanding is, after all, an interactive process
involving the reader. “Narrative truth is a linguistic phenomenon experienced
through reading, a performative event with participation on the reader’s part”
(Riffaterre 1990: xiv).
The cognitive theory of metonymy (Gibbs 1994) addresses the impor-
tance of metonymic models of thought. Metonymy, as well as having linguis-
tic realizations, is a universal cognitive process which enables understanding
to take place. So in a very general way, it can be used in any kind of
communication for specific purposes, to express meaning, and will be under-
stood automatically.
People’s knowledge in long-term memory of coherent, mundane series of
events can be metonymically referred to by the mere mention of one salient
subpart of these events. We see that the mention of the subpart metonymically
stands for the whole event. (Gibbs 1994: 330f)
Notes
1. Jakobson’s theory of metonymy has been developed by other literary critics, notably
Lodge (1977) who argued strongly for the co-presence of metonymy and metaphor in
post-modernist fiction.
2. This function is not exclusive to metonymy. In a longer narrative, both metaphor and
metonymy can function as underlying structural devices in a text (Lodge 1977; Werth
1994).
References
Austen, Jane
1990 [1814] Mansfield Park. London: Penguin.
Black, Elizabeth
1993 Metaphor, simile and cognition in Golding’s The Inheritors. Language and
Literature 2: 37–48.
Bunyan, John
1974 [1678] A Pilgrim’s Progress. London: Collins.
Dirven, René
1993 Metonymy and metaphor: different mental strategies of conceptualization.
Leuvense Bijdragen 82: 1–28.
Gibbons, Thomas
1969 [1767] Rhetoric. Menston: The Scolar Press.
Gibbs, Raymond W., Jr.
1994 The Poetics of Mind: Figurative Thought, Language and Understanding.
Cambridge: Cambridge University Press.
Hoebel, Edgar
1972 Anthropology: The Study of Man. New York: McGraw Hill.
Jakobson, Roman
1956 Two aspects of language and two types of aphasic disturbances. In R.
Jakobson, M. Halle (eds.), Fundamentals of Language. The Hague: Mou-
ton, 55–82.
Labov, William
1972 Language in the Inner City. Philadelphia: University of Pennsylvania Press.
Lakoff, George, Mark Turner
1989 More than Cool Reason. Chicago: The University of Chicago Press.
Langacker, Ronald W.
1987 Foundations of Cognitive Grammar. Volume I: Theoretical Prerequisites.
Stanford, California: Stanford University Press.
Leech, Geoffrey N., Michael Short
1981 Style in Fiction. London: Longman.
Lodge, David
1977 The Modes of Modern Writing: Metaphor, Metonymy and the Typology of
Modern Literature. London: Edward Arnold.
Morrison’s Song of Solomon 399
Miller, J. Hillis
1985 Introduction to Bleak House. London: Penguin.
Morrison, Toni
1977 [1989] Song of Solomon. London: Picador.
Norrick, Neil
1981 Semiotic Principles in Semantic Theory. Amsterdam and Philadelphia:
Benjamins.
Ortony, Andrew
1993 Metaphor and Thought. Second edition. Cambridge: Cambridge University
Press.
Propp, Vladimir
1928 Morphology of the Folk Tale. Austin: University of Texas Press.
Riffaterre, Michael
1982 Trollope’s metonymies. Nineteenth Century Fiction 37/3: 272–292.
1990 Fictional Truth. Baltimore: Johns Hopkins University Press.
Schulz, Victor
1992 ‘Deep structure signals’ in fiction. In D. Stein (ed.), Co-operating with
Written Texts. Berlin: Mouton de Gruyter, 335–360.
Steen, Gerard
1994 Understanding Metaphor in Literature. London: Longman.
Werth, Paul
1994 Extended metaphor, a text-world account. Language and Literature 3/2:
79–103.
List of Contributors
(Present addresses)
Andreas Blank
Philipps-Universität Marburg, Institut für Romanische Philologie
Wilhelm-Roepke-Str. 6D, D-35032 Marburg, Germany
E-mail: andreas.blank@mailer.uni-marburg.de
David D. Clarke
University of Nottingham, Dept. of Psychology
University Park, Nottingham NG7 2RD, UK
E-mail: David.Clarke@nottingham.ac.uk
René Dirven
Gerhard-Mercator-Universität Duisburg, Fachbereich 3, Sprach- und Li-
teraturwissenschaften
Postfach 10 15 03, D-47057 Duisburg, Germany
E-mail: Rene.Dirven@orchis.be
Gilles Fauconnier
University of California, San Diego, Cognitive Science Center
La Jolla, CA 92093, USA
E-mail: faucon@cogsci.ucsd.edu
Kurt Feyaerts
KU Leuven, Dept. Linguistiek
Blijde-Inkomststraat 21, B-3000 Leuven, Belgium
E-mail: kurt.feyaerts@arts.kuleuven.ac.be
Raymond W. Gibbs, Jr.
University of California, Santa Cruz, Dept. of Psychology
Santa Cruz, CA 95064, USA
E-mail: gibbs@cats.ucsc.edu
402 List of Contributors
Louis Goossens
Universitaire Instelling Antwerpen, Dept. Taal- en Letterkunde: Ger-
maanse talen
Universiteitsplein 1, B-2610 Antwerpen (Wilrijk), Belgium
E-mail: louis.goossens@uia.ua.ac.be
Olaf Jäkel
Matin-Luther-Universität Halle-Wittenberg, Fachbereich Sprach- und
Literaturwissenschaften, Institut für Anglistik und Amerikanistik
D-06099 Halle (Saale), Germany
E-mail: jaekel@anglistik.uni-halle.de
Peter Koch
Eberhard-Karls-Universität Tübingen, Romanisches Seminar
Wilhelmstraße 50, D-72074 Tübingen, Germany
E-mail: peter.koch@uni-tuebingen.de
Zoltán Kövecses
Eötvös Loránd University, Dept. of American Studies
Ajtósi Dürer sor 19-21, H-1146 Budapest, Hungary
E-mail: kovecses@isis.elte.hu
Brigitte Nerlich
University of Nottingham, Dept. of Psychology
University Park, Nottingham NG7 2RD, UK
E-mail: bn@psychology.nottingham.ac.uk
Anne Pankhurst
University of Edinburgh, Dept. of Applied Linguistics
14 Buccleuch Place, Edinburgh EH8 9LN, UK
E-mail: afp@holyrood.ed.ac.uk
Klaus-Uwe Panther
Universität Hamburg, Seminar für Englische Sprache und Kultur
Von-Melle-Park 6, D-20146 Hamburg, Germany
E-mail: panther@rrz.uni-hamburg.de
Paul Pauwels
Katholieke Vlaamse Hogeschool
St. Andriesstraat 2, B-2000 Antwerpen, Belgium
E-mail: jverscha@pophost.innet.be
List of Contributors 403
Günter Radden
Universität Hamburg, Seminar für Englische Sprache und Kultur
Von-Melle-Park 6, D-20146 Hamburg, Germany
E-mail: radden@rrz.uni-hamburg.de
Ken-ichi Seto
Osaka City University, Faculty of Literature
3-3-138 Sugimotocho, Sumiyoshi-ku, Osaka, Japan 558
E-mail: seto@lit.osaka-cu.ac.jp
Linda Thornburg
Eötvös Loránd University, Dept. of American Studies
Ajtósi Dürer sor 19-21, H-1146 Budapest, Hungary
E-mail: lthornburg@isis.elte.hu
Zazie Todd
University of Leicester, Dept. of Psychology
University Road, Leicester LE1 7RH, UK
E-mail: ztl@le.ac.uk
Mark Turner
University of Maryland, Dept. of English Language and Literature
1102 Francis Scott Key Hall, College Park, MD 20742
E-mail: mturn@wam.umd.edu
Christian Voßhagen
Universität Hamburg, Seminar für Englische Sprache und Kultur
Von-Melle-Park 6, D-20146 Hamburg, Germany
E-mail: fs6a023@rrz.uni-hamburg.de
Richard Waltereit
Eberhard-Karls-Universität Tübingen, Romanisches Seminar
Wilhelmstraße 50, D-72074 Tübingen, Germany
E-mail: richard.waltereit@uni-tuebingen.de
Beatrice Warren
Lund University, Dept. of English
Helgonabacken 14, S-22362 Lund, Sweden
E-mail: beatrice.warren@englund.lu.se
Subject index
essive H
role 280 hapax legomenon 174
schema 280, 283-285 head 220-221
euphemism 264-266, 269, 272, 289, hedged performative 342-344, 351, 356
295, 299-301, 306 hierarchy
metonymic 53, 373 conceptual 328
evaluation/evaluative 292, 294-296, metaphoric 309-310, 314, 324
301-302, 306 metonymic 309, 321, 323
Event ICM 32-33 of grammatical relations 236, 248
event schema 275, 278-280, 285 of metonymic vehicles 29, 52
experiencer 248-249, 285 highlighting 11, 153, 174, 184, 258-259,
experiencing schema 285 270, 280, 302, 305
expressive 140, 155-156 hyperbole 215, 321-323
extension(al) 25, 153-154, 270, 317, hyponymy 94
320-321, 325, 365, 367
analogical 365, 367 I
metaphorical 281 ICM (Idealized Cognitive Model) 9, 17,
metonymic 150, 153, 325 19-23, 26-29, 67, 186, 256, 334
iconicity 250, 277, 282, 373
F idiomatic expression 309-310, 334
face-threatening act 343 illocution 176
figurative language 309 illocutionary
figure/ground 151-152, 154-157, 159 act 335, 343-344, 356
focusing, metonymic 286 force 350, 368
folk model 212 image schema 319
force-dynamic structure 87 implication 154-156, 178, 187
Form ICM 36 pragmatic 13
frame 9, 139, 145-155, 157-160, 172- implicature 193, 195, 349, 353-355
175, 178-180, 185, 234, 238, 276, 369 conversational 62, 67, 156-157, 334
generic 79 imprecation 344
frame model/theory 149, 151, 159 inalienable possession 242, 244-245
function of metonymy, pragmatic 12 indexicality 373
indirect object 243
G indirect speech 72
gender change 157-158 indirect speech act 62, 72, 335-336, 344,
generalization 118, 270 347, 351
generic space 77 inference 155, 198-199, 203, 205-206,
genitive 275 345
genus 114 metonymic 62, 355
gestalt psychology 151, 159-160 inheritance hierarchy 315
goal 280, 282, 285 innovation, metonymic 156-157, 174,
goal-over-source principle 282 375
grammaticalization 126, 346-347, 350 input space 77
Grice’s maxims 268 input-topology 87
instrument 246-249, 267-268, 277-281,
408 Subject index
typology of metonymy 173, 176, 178, vehicle 19, 21-22, 29, 334, 354
182, 185, 363 vehicle-to-target route 52
violation of cognitive and communica-
U tive principles 52-54
understatement 303
unpacking 85 W
web 85
V Whole ICM and its part(s) 30
vagueness 266 Whole-Part ICM 36
valence semantics 177 word formation 139, 153, 157-159, 187
value judgment 313 world knowledge 173-174 (see also
variation 261 encyclopaedic knowledge)
Author index
A Broca 171
Abbott, Black, and Smith 69 Brown 160
Aitchison 292 Bruner 313
Albee 298, 303 Bühler 160
Amin 160 Bunyan, John 389
Anderson 239 Busse 236
Andor 173 Bybee 195
Arance 185 Bybee et al. 193
Aristotle 1, 141, 160, 179, 362
Auden, W.H. 63 C
Clark 70, 71, 364, 365, 372, 373, 375
B Clark and Clark 108, 125, 280
Bach 211, 215, 226 Clark and Gerrig 71, 298
Bach, Adolf 211, 225 Clark, Eve 366
Bach and Harnish 342 Coates 197
Bain 141 Cooper 117
Baker 239 Cooper and Ross 45, 57
Balzac, Honoré de 61, 63 Copestake and Briscoe 133
Barrett 367 Cordier 146, 160
Barsalou 146, 249 Coseriu 160, 170
Bartlett 146 Cottle 216, 225, 226
Belletti and Rizzi 243, 250 Coulson 77, 78
Bergsten 125 Croft 19, 56, 62, 66, 153, 154, 173, 174,
Black 160, 396 186, 233, 238, 257, 258, 259, 263, 302,
Blake 245, 246 317, 319, 329, 334
Blank 141, 144, 145, 150, 159, 160, 170, Cruse 93, 100, 101, 117, 291, 292, 294,
171, 172, 186, 187, 233, 236 295
Bombeck 70
Bonhomme 145, 178, 179, 186 D
Bower, Black, and Turner 69 Darmesteter 141
Bowerman 364 Davies 311
Braine 367, 368 De Bleser and Bayer 185
Bréal 141, 159, 235, 361 Debus 226
Bredin 62, 105, 139, 145, 146, 154, 160, Deese 291, 292
169, 171, 177, 187 Dik 150, 173, 208
414 Author index
Dirven 56, 117, 280, 317, 320, 328, 331, Gibbs and McCarrell 73
389 Gibbs and Tenney 69
Dorward 215, 216, 225, 226 Givón 247, 248, 294
Dubois 144 Goossens 63, 103, 122, 193, 195, 209,
Dubost 236 256, 257, 258, 259, 269, 272, 317, 319,
Duchácek 169, 177 320
Dumarsais 177, 185 Goossens et al. 193
Gottschald 211, 223
E Graesser, Woll, Kowalski, and Smith 69
Eco 145 Grice 50, 57, 67, 68, 115, 121, 157, 268,
Eemeren et al. 318 335, 352
Ekman et al. 304 Grimshaw 243
Eliot, T.S. 63 Groupe de Liège 1
Esnault 362, 363 Gurd and Marshall 185
Esper 292
H
F Halliday 121, 129
Fass 133, 177, 248, 363 Happ 142, 143, 145, 160, 172
Fauconnier 77, 133, 186 Haspelmath 155
Fauconnier and Turner 77, 84 Hawkins 250
Fauconnier, Gilles 187 Hayakawa 299
Fillmore 146, 150, 160, 173, 177, 186, Heider 313, 330
239, 249, 266 Heine 193
Fleischer 226 Hemenway 30
Fontanier 169, 170, 177, 185 Herennium, auctor ad 140-141, 170, 185,
Francik and Clark 72 362
Fraser 342 Herskovits 56, 104
Frawley 24, 56 Hjelmslev 160
Frei 245 Hobbes 141
Fromkin 292 Hoebel 387
Holenstein 142, 143, 160
G Horn 352
Garavelli 170, 185 Hornby 300
Garman 292 Householder 361
Gauger 158 Huber et al. 185
Geens 260 Hume 141
Geeraerts 118, 150, 160, 321 Huttenlocher and Smiley 366
Geeraerts and Grondelaers 330
Geiger 322 I
Genette 62 Ikegami 282
Gernsbacher 68
Gerrig 70 J
Gibbons 387 Jackendoff 133, 233
Gibbs 1, 8, 18, 33, 35, 56, 62, 72, 73, 121, Jäkel 225, 226, 227
122, 364, 372, 374, 379, 385, 389, 397 Jakobson 62, 117, 143, 144, 160, 171,
Author index 415
172, 186, 386, 387, 396, 398 Leech and Short 389
Johansson 260 Lehrer 292
Johnson 183, 186, 255, 256 Leisi 125
Jongen 150 Leitner 125, 280
Joyce, James 63 Lejeune 208
Leopold 367
K Levi 125
Kainz 300, 306 Levin 62
Kastovsky 280 Liebert 160
Keller 176 Lindner 104
Kleiber 10, 12, 150, 160 Lipka 125
Koch 141, 145, 150, 155, 158, 159, 160, Locke 141
161, 173, 175, 186, 238, 250 Lodge 62, 63, 398
Koch and Oesterreicher 161 Lyons 117, 226, 292, 300, 306, 323
Köhler 160
König 195 M
König and Haspelmath 245, 246 Major 290, 296, 297, 300, 305
König and Traugott 156, 159 Marchand 279
Kövecses 3, 48, 49, 80, 81, 83, 121, 330 Martin 144
Kövecses and Radden 334, 354 Matthews 213, 226, 227
Kövecses and Szabó 57 Mellenius 125
Kruszewski 142, 143 Metzger 160
Kytö 209 Mill, James and John St. 141
Miller 396
L Minsky 146, 173
Labov 386 Morgan 57
Lakoff 2, 3, 19, 20, 25, 27, 33, 35, 38, 39, Morrison, Toni 385, 397
49, 50, 56, 57, 66, 67, 80, 81, 83, 94, Müller-Gotama 250
95, 115, 150, 186, 226, 233, 255, 256, Murphy and Andrews 292
257, 259, 290, 291, 294, 295, 310, 312,
313, 314, 315, 316, 317, 330, 334 N
Lakoff and Johnson 1, 2, 18, 19, 26, 40, Naumann 211, 215
51, 56, 64, 94, 95, 105, 154, 160, 169, Nerlich 117, 118, 141, 160, 185, 379
175, 177, 183, 226, 255, 256, 259, 271, Nerlich and Clarke 373, 379
321, 363 Norrick 34, 35, 41, 227, 363, 387
Lakoff and Turner 24, 34, 42, 56, 95, 385, Norvig and Lakoff 39
389 Nunberg 10, 12, 70, 122, 128, 133, 233,
Landis, Herrmann and Chaffin 291 234
Langacker 9, 19, 30, 31, 44, 56, 57, 96, Nyrop 141, 185, 363
109, 126, 127, 160, 174, 176, 186, 207,
233, 238, 239, 250, 259, 260, 261, 275, O
295, 310, 317, 328, 330, 388 Ogden and Richards 23, 26, 110
Larkin, Philip 64 Olsen 239
Lausberg 141, 154, 170, 177, 185 Ortony 389
Le Guern 145, 154, 185, 186, 233 Orwell, George 63
416 Author index
P Steen 389
Panther and Thornburg 335 Stein, Gertrude 73
Paretsky 113, 118 Stempel 161
Paul 141 Stern 133, 177, 363
Pauwels 268, 269 Stimson 376
Pauwels and Simon-Vandenbergen 260 Sweetser 5, 193, 353, 356
Pease et al. 366
Piaget 366 T
Propp 386 Tagiuri 330
Pustejovsky 104 Talmy 37, 104
Pustejovsky and Bouillon 104 Tannen 146, 249
Taylor 27, 55, 104, 105, 115, 150, 160,
Q 183, 227, 256, 257, 317
Quintilian 177 Thornburg and Panther 33, 333, 336,
344, 350, 355
R Thorne 290
Raible 144, 160, 186 Traugott 193, 195
Rastier 160 Traugott and König 156, 195, 199, 207
Reaney 211, 213, 216, 223, 225 Turner 77
Reddy 104 Turner and Fauconnier 77, 83
Reisig 141 Tversky 30, 93
Rice 331 Tversky and Hemenway 46, 93
Riffaterre 8, 386, 387, 388, 389, 391, Tylor 26
396, 397
Rosch 66, 160 U
Roudet 142, 143, 144, 160, 171, 172 Ullmann 49, 144, 154, 169, 170, 171,
Ruwet 185 177, 363
Ungerer and Schmid 121
S
Sandra 331 V
Sappan 363 Vosniadou 378
Saussure 142, 143, 160 Vygotsky 366
Schank and Abelson 69, 249
Schifko 154, 160, 161, 169, 177 W
Schmid 317 Warren 110, 117, 122, 123, 125, 133,
Schofer and Rice 1 150, 154, 171, 177, 178, 187
Schulz 396 Weinrich 104, 146
Schwarze 161, 187 Wells 110
Searle 62, 335, 336, 344, 345, 350 Wentworth and Flexner 290, 292, 295,
Seto 367 296, 297, 300, 302, 305
Shakespeare 64 Wernicke 171
Smith 213, 223, 225, 227 Werth 159, 398
Sorensen 110 Wertheimer 160
Spanoghe 245 Wierzbicka 93
Sperber and Wilson 12, 50, 176, 297, 298 Wiktorin 226
Author index 417
Wimmer 227 Y
Winston 239 Yamanashi 363
Winston et al. 100
Witkowski 226 Z
Wittgenstein 151 Zandvoort 279
Wundt 142, 303-304
Metonymy and metaphor index
D H
defining property for category 35 hamburger for customer 277
destination for motion 37 human over non-human 45, 51, 53, 57
deviant behavior for stupidity 324 human to non-human 11
disposition for occasional behavior 348
dominant over less dominant 47 I
ideal over non-ideal 48
E immediate over non-immediate 47, 51,
effect for cause 22, 38, 39, 43, 321, 333 57
effect for core 338 important over less important 49
emotion for cause of emotion 39, 47 important utensil for person 215, 224
ensuing situation to preceding situation inhabitants for place 41
108, 111 initial or final over middle 49, 53
ensuing situation to process 108, 109 institution for people responsible 271,
evaluative concept for its opposite 299 322
event for place 42 institution for place 28, 41
event – subevent 106 instrument – actor 181
events are actions 314 instrument – product 182
instrument for action 37
F instrument for agent 37, 187
face for the person 18, 19 instrument for product 40
fire – fireplace 154 instrument for result 125
form for concept 24, 42, 46 instrument to result 110
form for content 333 instrument/organ of perception for the
form for thing/event 26 perception 38
form-concept for thing/event 24, 46 interactional over non-interactional 46, 51
form-conceptA for conceptB 27 inventor – product 180
formA-conceptA for formA-conceptB 27 inventor for the thing invented 39, 176
formA-conceptA for formB-conceptA
28, 36 K
formA-conceptA for formB-conceptB 26 knowing is seeing 319
Metonymy and metaphor index 421
L organization – member 99
life is a journey 314, 316, 330 origin – object 125
living things – human beings 153 outgroup origin for stupidity 49, 311
M P
manner for action 37, 43 pars pro toto 154
manner of perception for the perception part – part 181
38 part – whole 181, 184
material – product 182 part for part 36
material constituting an object for the part for whole 31, 36, 43, 46, 84, 86, 106,
object 32, 48 113, 125, 259, 261, 271, 321, 389
material to result 110 part of a form for the whole form 36
matter for artifact 125 part of a thing for the whole thing 31
means for action 37 part – part 185
member of a category for the category part – whole 124, 179, 185, 234, 239,
27, 34, 42 243, 244, 245, 250, 344, 362
mental/physical state for object/person passing of time is moving entity 315
causing it 39 people for their possessions 72
messenger – angel 171 perception for result of perception 319
modified form for original form 43 perception for thing perceived 38, 45
more over less 47 period – period 182
person for the face 19
N physical/behavioral effect for emotion
necessity for motivation 333 causing it 39
new over old 50 place – object 181
place – place 182
O place (of origin or residence) for person
object – collective body 181 216, 224
object – function 180 place for activity performed at that place
object – typical aspect 180 42
object – place 160 place for event 42, 65, 271, 322
object for action 37 place for inhabitants 41
object for material constituting the object place for institution 28, 41, 65, 72 125,
32, 48 271, 322
object for owner 387 place for an institution located at that
object for user of the object 40 place 65, 72
object used for user 65, 271, 321 place for product made there 40
object – component 99 place for the event 65, 322
object – material 99 place for the institution 322
object – origin 125 place to result 111
object – place 124 place – object 124
object – property 317 positive end of a scale stands for the
objects used for their users 72 whole scale 294
occurrent over non-occurrent 47 positive for negative 304
ordered dish – customer 183 possessed for possessor 40, 41
422 Metonymy and metaphor index
V
virtuality for actuality 343 whole scale for upper end of the scale 32
visible over invisible 46, 51 whole thing for a part of the thing 31, 48
whole – part 99, 103, 104, 113, 124,
W 179, 185
whole event for subevent 32 words for the concepts they express 24,
whole event – subevent 106, 107 42, 46
whole for part 31, 35, 50, 113, 125, 255, 271
In the series HUMAN COGNITIVE PROCESSING (HCP) the following titles have been
published thus far or are scheduled for publication:
1. NING YU: The Contemporary Theory of Metaphor. A perspective from Chinese. 1998.
2. COOPER, David L.: Linguistic Attractors. The cognitive dynamics of language acquisition
and change. 1999.
3. FUCHS, Catherine and Stéphane ROBERT (eds.): Language Diversity and Cognitive
Representations. 1999.
4. PANTHER, Klaus-Uwe and Günter RADDEN (eds.); Metonymy in Language and Thought.
1999.
5. NUYTS, Jan: Epistemic Modality, Language, and Conceptualization. A cognitive-
pragmatic perspective. 2001.
6. FORTESCUE, Michael: Pattern and Process. A Whiteheadian perspective on linguistics.
2001.
7. SCHLESINGER, Izchak, Tamar KEREN-PORTNOY and Tamar PARUSH: The
Structure of Arguments. n.y.p.
8. SANDERS, Ted, Joost SCHILPEROORD and Wilbert SPOOREN (eds.): Text
Representation. Linguistic and psycholinguistic aspects. n.y.p.