Metonymy

Author(s): Hugh Bredin

Source: Poetics Today, Vol. 5, No. 1 (1984), pp. 45-58
Published by: Duke University Press
Stable URL: http://www.jstor.org/stable/1772425
Accessed: 04/02/2010 14:04

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=duke.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Duke University Press is collaborating with JSTOR to digitize, preserve and extend access to Poetics Today.

http://www.jstor.org
 METONYMY

HUGH BREDIN
Scholastic Philosophy, Queen's

One of the distinguishing marks of modem literary culture is the

dominance given to metaphor over all the other figures and tropes.
It is not immediately clear why this should be so: a history of
metaphor's rise to power would be most interesting, and would
reveal much about how we currently conceive of, and respond to,
literary and other verbal discourse. Metonymy may in fact be more
common than metaphor, and indeed is held by some to be the
distinguishing mark of realist prose,1 yet it is seldom subjected to the
detailed and lengthy investigations that metaphor undergoes. And
not only is it not widely studied, but most accounts of it are un-
satisfactory. In most contemporary reference works, as well as
traditional treatises on rhetoric, the typical entry on metonymy runs
something as follows: metonymy is the transfer of the name of a
thing to something else that is closely associated with it - such as
cause and effect, container and contained, possessor and possessed,
and so on; for example, "crown" or "throne" for monarchy.
The giveaway here is the expression "and so on." For what is
presented in this kind of formulation is not a true definition, but an
enumeration of instances. Further research quickly reveals that the
various authorities cannot even agree on what these instances are:
a list of metonymy types compiled from even a small selection of
rhetoricians and reference books stretches out in a most alarming
way, and offers no obvious principle of order.
Confusion piles upon obscurity when we consider the treatment
given to synecdoche. There is fairly general agreement, not on how
synecdoche is defined, but on its various types; but there is general
disagreement on its relation to metonymy. In traditional Rhetoric
1. Th. locus classicus of this theory is in Jakobson (1971), but it has been investigated
in greatest depth by Lodge (1977).

Poetics Today, Vol. 5:1 (1984) 45-58

46 HUGH BREDIN

it was considered to be quite distinct from metonymy, but in more

recent times it has been said to be a type of metonymy (Whately,
Preminger, Jakobson, Henry). Others again claim that synecdoche
is the fundamental trope, of which metaphor and metonymy are
derivations (Group j). And at least one work makes the claim that
synecdoche and metonymy are varieties of metaphor (Abrams) - a
claim which, interestingly, takes us back to Aristotle's use of the
word "metaphor."
However, the standard account given of metonymy a moment
ago contains at least the beginnings of a definition, in the words
"the transfer of the name of a thing to something else." For these
words are nothing less than a definition of trope itself. Sometimes
trope is thought to be, or to involve, a transfer or alteration of
meaning, but this is clearly wrong. For what occurs in a trope is not
a change in the meaning of a word, but a change in the object to
which it normally refers. To use Frege's distinction, it is an alteration
in reference, not in sense. Therefore the phrase "the transfer of the
name of a thing to something else" will be part of the definition
of any trope, a statement of the genus to which it belongs. The
definition must then be completed by a statement of the specific
differentia, which distinguishes it from other tropes. It is here that
difficulty arises.
The type of differentia required is easy enough to characterize
in general terms. The difference among the various tropes is a
difference in the relation that holds between the two objects be-
tween which the transfer of names takes place. Thus, in the case of
metaphor the two objects are related by similarity. In irony2 the
objects are related by the fact that the names normally given to
them contrast in meaning. In the subspecies3 of trope known as
personification, one of the objects is a person and the other is
inanimate. In the subspecies antonomasia, one of the objects has a
proper name and the other is an attribute of the thing whose name
it is.
In the case of metonymy, however, the relation between the
objects concerned in the transfer is precisely the area of obscurity
and dissension. If we look at what generally pass for definitions of
metonymy, we find that some writers simply describe it as the
substitution of one name for another, and fail thus to distinguish
it from trope in general (Quintilian, Bede, Doyle); while others tell
us merely that the two objects are "closely associated" or "closely

2. "Irony" is being used here as the name of a trope.

3. "Subspecies" because personification can be metonymic as well as metaphoric. Antono-
masia is referred to as a subspecies for the same reason.
 METONYMY 47

related" ("Cicero," Abrams, Preminger). With this latter vague

affiliation we may lump "agreement" (Fraunce) and "contiguity"
(Jakobson).
It might be tempting to leave matters like this - in effect to
exploit metonymy as a raggle-taggle collection of those tropes for
which we can find no other name. In such a purview, metonymy
would have no distinctive character, but would include all kinds of
relations other than, say, similarity or the relation of part to whole;
it could then be defined as any trope that is not metaphor or
synecdoche. This, indeed, is not unlike what Jakobson has done to
metonymy. He describes metaphor as a trope based upon similarity,
a description at once traditional and clear. But to say, as he does,
that metonymy is based upon "contiguity" is to cloud it in a con-
venient but unrevealing metaphor: contiguity seems in fact to refer
to any type of relation whatsoever, other than similarity. In the
realm of figurative language, it is made the common dumping
ground.
But metonymy cannot be banished from the city quite as easily
as that. It is one of the great survivors among tropes. When Peter
Ramus reduced Quintilian's twelve tropes4 to four, metonymy was
one of the four, along with metaphor, synecdoche, and irony.5
When, in our own time, Jakobson reduced the four to two, me-
tonymy again remained, if not alongside metaphor at least in its
wake. In all the main treatises on rhetoric, metonymy is considered
to be a trope of some importance, second only to metaphor. Clearly,
there is some deeply rooted intuition about figurative language which
continues to generate the notion of metonymy with stubborn
persistence, and it is our duty to try to sharpen that intuition into a
clear conception, one capable, we must hope, of unambiguous
definition.
One way of attempting this task, and one which I here essay, is to
construct a list of metonymies - or, to be precise, of types of
metonymical relation - in order to see if they can be forced to yield
some principle of unity (Table 1). I have compiled the list from a
selection of rhetoricians, and contemporary reference books, which
if not exhaustive is at least representative. In Table 1, the pairs
of words enumerated on the left hand side indicate the relations
between the objects between which the transfer of names takes
place. These relations are two-way, and this is why there are two
illustrations of each. For example, in l(a) the name of a cause

4. Thirteen if simile is counted.

5. These are the "four master tropes" discussed by Kenneth Burke in his A Grammar of
Motives. It is not clear whether he is aware of the Ramist origins of this list.
48 HUGH BREDIN

Table 1

Metonymical Relations Examples

1. Cause Effect (a) War is sad.

(b) She is my pride and joy.
2. Inventor Invented (a) She was reading Virgil.
(b) An enterprising scheme.
3. User Instrument (a) A walking stick.
(b) A hired gun.
4. Doer Thing done (a) That car is a lovely driver.
(b) Crime must be punished.
5. Passion Object of passion (a) She is my true love.
(b) She is a slave to wealth.
6. Container Contained (a) A boiling kettle.
(b) Milks, sugars (for milk-jugs and
sugar-bowls).
7. Place Object in place (a) Wall Street panicked.
(b) I take my holidays at the seaside.
8. Time Object in time (a) A warlike century.
(b) The age of science.
9. Possessor Possessed (a) The violinist broke a string.
(b) The smart money is in computer
software.
10. Sign Signified (a) Crown (for monarchy).
(b) Give me a Scotland (to a seller of
rosettes at a football stadium).
11. Concrete Abstract (a) The soldier is the enemy of
mankind.
(b) Youth is giddy and irresponsible.

("war") is transferred to its effect, while in l(b) the name of an

effect ("pride and joy") is transferred to its cause. The same applies,
mutatis mutandis, to the rest.
It can be seen at once that any attempt to unify this unruly bunch
will require unusual finesse. In fact there is a strong temptation
to come up with an imposed solution, in other words with some
definition which will fit as many as possible of the types listed
and will simply exclude the rest. It could be argued, after all, that
traditional and current usage is conceptually chaotic, that there
simply cannot be a single trope which includes all the types that
people have taken it upon themselves to call metonymies. It could
be pointed out that traditional Rhetoric was never systematic in the
first place, that it was seldom more than a report on usage, and that
all or most of its technical terms are as loose and arbitrary as every-
day language, with the added disadvantage of accretions and erosions
they have suffered in over two thousand years of use.
 METONYMY 49

However, this would be to give up too soon. I will look upon

stipulative definition as a last resort, and proceed on the hypothesis
that there is indeed some principle of unity, that the impulse to
classify all the tropes on the list as metonymies has in fact a justifi-
cation. But before making the attempt, I wish to look at two recent
accounts of metonymy, in the hope that one of them may have hit
upon the answer and so saved us from unnecessary work.

First, I will examine the account of metonymy given in A General

Rhetoric, by Group ,. This brilliant work, which has helped to
revolutionize the study of rhetoric, provides a complete system of all
the figures and tropes. One of its most distinctive contributions to
the theory of the three major tropes is its argument that synecdoche
is the most fundamental trope, metaphor and metonymy being,
each in its own way, a combination of two synecdoches.6 However,
we need not concern ourselves here with this general system. Me-
tonymy, according to the Group p rhetoricians, is the substitution of
one verbal expression for another, whenever the expressions are
related to one another within a web of connotative associations.
They give no formal definition of connotation, but it is clear that
they are using the word, not in its logical sense, but in the sense of
a penumbra of related meanings, a cluster of semantic associations.
There are, they continue, two distinct modes of connotative
association underlying metonymy. In one mode, the referents of
the two expressions are members of the same "material whole," or
"spatio-temporal totality." For example, in the metonymy "reading
Caesar," for reading Caesar's De Bello Gallico, the expressions
"Caesar" and "De Bello Gallico" refer to objects belonging to a
totality which includes "the life of the famous consul, his loves,
his literary works, his wars, his times, his city" (p. 121). All of
these things constitute a web of connotations in virtue of which an
expression referring to one of them may metonymically replace an
expression referring to another.
The other mode of connotative connection is conceptual rather
than referential. That is, the two expressions are associated, not
primarily through their referents, but through their meaning; they
belong to a semantic rather than a material whole. The authors
unfortunately provide us with no illustration of this type, and indeed

6. It is not clear in what sense they are using the word "fundamental" (see p. 218). It seems
just as arguable that metaphor (or metonymy) is a fundamental trope, with synecdoche the
product of its dissolution.
50 HUGH BREDIN

it is remarkably difficult to think of one. But perhaps the following

will do: when we speak of "a hired gun," for a hired gunman, the
concept of a gunman contains the concept of a gun as one of its
semantic constituents. Thus the connotative link which sustains the
metonymical relation of "gun" to "gunman" is a conceptual rather
than a referential one.
It might reasonably be objected here that conceptual relations
among words are ultimately the product of "real" connections
among their referents; that language is not a self-sustaining system
but is controlled and determined by its use in the service of knowl-
edge and truth. However, to say this would provoke a sharp disagree-
ment with Group ,L, which has been deeply infected by the anti-
realist metaphysics of Structuralism (see p. 221). I will therefore not
pursue this line of argument, but look instead at whether the notion
of connotation casts any light upon metonymy, irrespective of
whether it has one mode or two.
In fact it casts very little light. It is no news to anyone that the
two items involved in a metonymy are related to one another: if they
weren't, it wouldn't be a metonymy but a mistake. However, to call
metonymical relations connotative is to give us very little in the way
of specific information about them, other than to tell us that they
are relations between verbal expressions and not relations between
objects. The problem is that webs of connotative links among words
are both extremely large and extremely various, but that not every
such link generates a metonymy. Thus "inflation" is connected with
"economy," but the replacing of one expression by the other would
be neither figurative nor intelligible. The same can be said about
"penal system" and "code of law," "Herodotus" and "Thucydides,"
"three little pigs" and "big bad wolf," and so on indefinitely.
Furthermore, in Britain the words "penny" and "copper" are con-
nected, but the trope in which pennies are referred to as "coppers"
is a synecdoche, not a metonymy. Thus, even if we were to concede
that metonymical relations were connotative relations, it would
remain the case that some connotations were metonymical and
others were not.
The Group pI theory is therefore not specific enough. It is not
enough simply to classify connotations as semantic or referential.
For what we want to know is which connotative relations are, in
addition to being connotative, metonymical as well. It may be this
lack of specificity, of a positive characterization of metonymy,
which compels the group to suggest that metonymy is the trope
that is left over - that metonymical relations are those which are
neither metaphoric, synecdochic, nor antiphrastic:
 METONYMY 51

[The reader] will first see whether the figure is synecdoche, metaphor, or
antiphrasis. It is only when . . . [this] . . . has failed that he will seek connota-
tive extrapolations that will allow him to identify metonymy (p. 122).

It is clear that the essence of metonymy still eludes us.

Another recent theory of metonymy is found in Albert Henry's
Aletonymie et Metaphore. It starts interestingly, by referring us to
Gaston Esnault's distinction between metonymy and synecdoche.7
Esnault had said that metonymy has to do with the intension
[comprehension] of words, while synecdoche has to do with their
extension. However, Henry becomes alarmed at the realism under-
lying this distinction - the fact that intension and extension are
determined by such things as properties and objects - and argues
that it is not logical but "psychological" intension and extension
which are manipulated in the two tropes.
By psychological intension Henry means something rather like
Group ,u's connotations, though it is at first sight a less indefinite
and loose conception. The psychological intension of a word is the
set of constituent concepts which, taken together, constitute the
"concept entity" designated by the word. This set of constituent
concepts is also ieferred to as the word's "semic field." For example,
the semic field associated with a certain French coin is composed
of the concepts flat, round, gold, to pay, effigy of King Louis: these
are thus the constituent concepts of the concept entity which
French speakers have in mind when they think of or speak of the
coin. It is Henry's contention that when the verbal sign for one of
these constituent concepts, "louis," is used to designate the whole
concept entity of which it is a part, we get a metonymy.
One difficulty with this theory is mentioned by Henry himself:
that it is difficult to say just what are the constituent concepts of
a given concept entity, or semic field. For example, does the semic
field of the word "poison" include the concept elimination of a
political rival? No firm answer can be given. It is clear that, once we
begin to loosen the notion of logical intension, we embark on a
slippery slope which has no end. Between Henry's semic field and
Group J's web of connotations there is only a difference of degree.
Both of them alike stretch out inexorably and indefinitely.
Henry's account of metonymy is significantly different, however,
from that of Group p. If it were the same, he would have had to say
that, in a metonymy, the word normally designating one of the
constituent concepts in a semic field was used to designate another
of the constituent concepts - and not, as he claims, to designate

7. The reference given in Henry's bibliography is Gaston Esnault, L'Imagination populaire,

Metaphores occidentales, Paris, 1925.
52 HUGH BREDIN

the whole concept entity of which it is a part. None the less, the
criticisms which I have made of Group p can easily be adapted to fit
Henry as well. For we can construct semic fields where metonymies
are less easy to come by than is the case with the French louis. Take
the expression "publishing company": its semic field includes such
concepts as book, distribution, and profit; yet it is impossible to
employ the name of any of these as a metonymical replacement for
"publishing company." Furthermore, if Henry's theory were correct,
the use of the word "nickel" for a five-cent coin should be a me-
tonymy; whereas in fact it is a synecdoche.
Both Henry and Group ,i are unusual in trying to analyze and
describe metonymical relations at all. As we have seen, most writers
on the subject are content to say that the relationship is a "close
association," or "close relationship," or "contiguity" - expressions,
these, which are not in fact descriptions but tautologies. In the first
century B.C., the influential Rhetorica ad Herennium stated:
Metonymy is the figure which draws from an object closely akin or associated
an expression suggesting the object meant, but not called by its own name
(Bk.IV, 32).
Variations on this formula have been repeated down to and including
our own day. What is missing from it is what Lakoff and Johnson
call the "systematic" character of metonymy. By this they mean that
metonymical connections are not random, but are specific types of
connection. It is this specificity which is lacking in Group p and in
Henry, despite their best efforts.

An assumption shared by Henry and the Group p rhetoricians is

that the relations exploited in metonymy, and indeed in all kinds
of tropes, are linguistic relations. It seems a plausible enough assump-
tion. Yet it has proved to be a singularly unfruitful one: for neither
the relations subsisting in webs of connotations, nor those in fields
of psychological intension, have enough in the way of positive
features to give them a clear identity, nor enough in the way of inner
diversity to sustain a distinction between metonymical and non-
metonymical relations.
I propose, therefore, to assume instead that metonymical relations
are relations between things, not between words. This is not at all
unreasonable, since the ways in which we normally describe the
various types of metonymy - container-contained, possessor-
possessed, and so on - quite clearly refer to things, not words.
However, I make no claims about the ontology of relations, about
 METONYMY 53

whether they are secundum rem or are categories of knowledge,

or whatever. My assumption here is a methodological one, designed
to make available a language of description and analysis more diverse
and resourceful than the metalanguages of Rhetoric and Linguistics.
The problem, then, is to find some way of characterizing all and
only metonymical relations between objects. In such an enterprise,
clearly, the concept of relation has a central and pervasive part to
play. Relation is one of those fundamental concepts, like existence
or inference, which are incapable of explication in any simpler terms.
There are, however, two ways in which relations can usefully be
discussed. One of these is to look upon them as predicates, and
investigate their logical properties: most recent discussion has been
of this kind, and various properties of two-place relations, such as
transitivity and connectedness, have been isolated and examined by
a number of philosophers from at least Russell onwards. The other
way of studying relations, though, is in a sense more fundamental:
it builds upon the presumption that there may be fundamentally
different kinds of relational predicates - different, that is, not just
in their logical properties but in their truth conditions. Or to put it in
a different language, relations are looked upon as a kind of onto-
logical cement holding the world together, and the challenge is to
distinguish their various types.
It is this second kind of task which I take to have relevance to the
enterprise in hand, that of finding what it is that distinguishes
metonymical relations from others. What we want, therefore, is a
typology of relations. There are some famous typologies, notably
Aristotle's, Hume's, and Kant's, which might be explored at this
point. However, I will simply state that none of these three gives a
satisfactory account of what it is that is common to all the metony-
mies listed above. It is therefore necessary to construct a typology
which will, and this is what I attempt below. As it turns out, it
explains not only metonymy, but synecdoche and metaphor as well.
First, then, we may distinguish between structural relations and
extrinsic relations. Structural relations are relations within things,
extrinsic relations are relations among things. Or to be precise, they
are relations within or among what are taken, in the context, to be
particulars. Thus, the relations of an object to the parts of which
it is composed, or the material out of which it is made, are structural
relations. So, too, is the relation between two concepts whenever
the extension of one includes the extension of the other - the
relation, for instance, between man and mortal. It can be seen at
once that we can identify structural relations with those involved
in synecdoche, the traditional account of which is that it is a transfer
of names between part and whole, species and genus, and object
54 HUGH BREDIN

and material. These are not a heterogeneous bunch of relations,

but variants of a single type.
The theory I want to propound is that synecdochic relations are
structural, and metonymical relations are extrinsic - relations, in
the one case, between particulars and their parts, and in the other
case between particulars and other particulars. Metaphorical rela-
tions, as we shall see, belong to the latter category as well. It is of
great importance, though, to note that what is taken to be the
particular in any given case is dependent upon the context, and not
necessarily upon some inherent nature in things. In other words, the
concept particular, as it is being used here, has a strongly epistemic
character - as perhaps it always has. For example, "wheels" is a
synecdoche for an automobile; but if a racing driver is given the nick-
name "Wheels," this is a metonymy. In one case the particular is an
automobile, and wheels are part of it, structurally related to the
automobile as part to whole. In the other case, wheels are a parti-
cular, and are extrinsically related to the driver, who is another
particular.
We must also note that, in the synecdochic relation of genus to
species, the concept of the genus is taken as a particular, even though
it is a concept. One reason is that, when we begin to analyze a con-
cept into its constituents, its having those constituents is the having
of certain properties: thus, mortal has the property of including the
extension of the concept man; it may be said to instantiate the
having of that property as a particular instantiates a universal.
Turning now to extrinsic relations, we may distinguish between
two kinds, which I will call simple relations and dependent relations.
The difference between them can best be approached by considering
dependent relations first. In a dependent extrinsic relation, the
objects are related in respect of some property possessed by those
objects. Thus, in "A is older than B," A is related to B in respect of
their ages. In "A is longer than B," A is related to B in respect of
their lengths. In "A is less than B," A is related to B in respect of
their quantity. And so on and so forth. In dependent relations, the
identity of the relation depends upon some property possessed by
both of the relata. This is why we cannot say, for instance, that a
meter is longer than a drink of tea: only one of these relata has the
property of length, and so the relation "is longer than" is, in this
case, without identity, and thus absurd.
Similarity is a dependent relation. The proposition "A is like B"
is in a sense incomplete, for one thing is like another always in some
- as it is
respect or other. Even if the respect is not made explicit
in "A is as long as B," for instance - the very notion of likeness
presupposes none the less that there must be some property or other
 METONYMY 55

in respect of which any pair of objects is asserted to be alike.

Similarity is the basis of metaphor, which may therefore be classed
with assertions of dependent relations. Metonymy, on the other
hand, exploits simple relations between objects. A simple relation
is any extrinsic relation that is not dependent. Cause and effect, for
example, are simply related, since there is no property, possessed by
both a cause and its effect, with respect to which the one is the cause
of the other; "is the cause of" is not with respect to anything. The
same can be said of all of the relations listed in the table of metony-
mies - is the inventor of, is situated in, is the sign of, and the rest. In
no case does the identity of the relation depend upon a property
shared by the relata.
The theory which I have just outlined can now be represented in
the following tree:

Relations

Structural Extrinsic
(Synecdoche)

Simple Dependent
(Metonymy)

~I !1~III
Similarity Others
(Metaphor)

It can be objected that there is a certain arbitrariness in this

schema - as there is, perhaps, in any attempt to delineate the order
of things, even if the order fits the facts. With sufficient ingenuity,
a dozen different orders could be constructed, all fitting the facts.
Order by itself compels no assent, unless we can see the ground from
which it springs.
However, it is not as arbitrary as it may seem at first sight. For in
fact it corresponds to radical distinctions among various modalities
of consciousness. The foundation of the schema is the well-known
distinction of analysis and synthesis. Cognition is a connected series
of encounters between subject and object in which we analyze things
into their constituents, or synthetically combine objects in
larger
or more abstract wholes, such as concepts, classes, and kinds. The
life of the mind, in its cognitive dimension, may be conceived of as
a continual and intricate dialectic of analytic and synthetic opera-
56 HUGH BREDIN

tions. It is easy to see that what I have called structural relations,

relations within things, can be regarded as objects of the analytic
mode of consciousness; while extrinsic relations, relations among
things, are objects of the synthetic mode.
The synthetic mode can then be further subdivided, for there are
two ways in which things may be brought together in consciousness.
In one, the bringing of them together in an explicit relation pre-
supposes that they have already been brought under a certain
concept. For example, the relation "is longer than" presupposes
that the objects which it relates are already apprehended as objects
which have length. There is thus a twofold synthesis involved in such
relations: a conceptual synthesis based upon a shared property of
the relata (A and B have length), and then the synthesis of the
relation itself (A is longer than B). In the other manner of synthesis,
there is a simple, singlefold relating of objects, which are not brought
under any concept, apart from whatever may be consequent upon
the relation itself.
The distinction of these two modes of synthesis corresponds, of
course, to the distinction of simple and dependent relations. It turns
out, therefore, that the typology of relations given here is not just
an ingenious but arbitrary ordering of things, designed to satisfy
the desire to construct a theory of tropes. Instead, it is part of a
more comprehensive theory in which language, thought, and things
are found to enter into profound correspondences with one another.
One of the loose ends in this theory, which should perhaps be
dealt with here, is the nature of the difference between the relations
asserted in metaphor, and other dependent relations. For of course,
while "A is as long as B" asserts a similarity, it is not metaphorical.
The difference is that, although metaphoric relations are dependent,
there is no explicit property shared by the related objects, no
definite concept that they are brought under. In "A is as long as B,"
the respect in which A is like B is explicit. But in metaphor, the
respect or respects in which the two objects are alike do not possess
the sharp and narrow clarity of concepts. The metaphor "He is a
long string of misery" cannot be elucidated conceptually. We might
try such statements as "He is tall," "He is as thin as string," "He is
like misery," but still we fall short of the meaning of the metaphor.
Indeed, it could be argued that the last of these statements cannot be
conceptually elucidated even in principle: for, in what respect is he
like misery? Under what concept could the person and the quality
fall? Or to take a simpler, though hackneyed, example, "Man is a
wolf" - we cannot say precisely what the properties are which men
are being said to share with wolves, even if we should agree that the
statement is true. In metaphor, thought has not hardened into
 METONYMY 57

concepts; in metaphor, some would wish to add, our thoughts often

move in realms that lie outside the range of concepts.
Returning now to metonymy, I will conclude that it is a transfer
of names between objects which are related to one another extrinsi-
cally and simply. Its role in language and in thought is that it
articulates the enterprise of combining our objects of thought into
larger wholes. However, since metonymical relations are simple,
and not dependent, they lack both the explicit conceptual connec-
tions presupposed in most dependent relations, and the spontane-
ously intuited connections found in metaphor. A metonymy neither
states nor implies the connection between the objects involved in
it. For this reason, it relies wholly upon those relations between
objects that are habitually and conventionally known and accepted.
We must already know that the objects are related, if the metonymy
is to be devised or understood. Thus, metaphor creates the relation
between its objects, while metonymy presupposes that relation.
This is why metonymy can never articulate a newly discovered
insight, why it lacks the creative depth of metaphor. Metonymy is
irresistably and necessarily conventional.
Because it is conventional, it is subject to limitations and changes
imposed upon it by inherited knowledge and culture. There are
many pairs of related objects which in theory could be the basis of
metonymies, but in practice are not. For example, there is no reason
why metonymies could not express the relations of husband and
wife, parent and child, or employer and employee; but in fact any
such metonymies would be found confusing and culturally dis-
tasteful. Again, the metonymies which once generated occupational
surnames such as Smith or Miller, or clan surnames such as McGregor
or O'Brien, are no longer possible. Or again, it is conceivable that
there could be a society so ordered that possessor-possessed metony-
mies would not occur or be understood. Metonymies, furthermore,
are subject to linguistic restrictions. If there were a language, for
instance, in which there was no clear distinction between doer and
thing done - so that the same word was used for "crime" and
"criminal" - then that type of metonymy would be improbable or
impossible.
These types of restrictions on metonymy have brought it about
that any list of the metonymy types actually used by, say, a twentieth-
century English speaker has a certain arbitrariness. And it is no doubt
this arbitrariness which has caused the remarkable and general failure
to discern its true nature. But it is "arbitrary" whether a house is
built of brick or wood, yet we can still define a house. Metonymy
does not refer to a congeries of vague and heterogeneous associa-
tions, but has a definite essence and structure. It may lack the
58 HUGH BREDIN

brilliance and glamor of metaphor, but it is more closely knit into

the fabric of language. In its very conventionality it retains and
expresses many of our everyday values and prejudices, and our
inherited knowledge of the world. This is perhaps the only reason
for saying that it is a mark of realist prose. I suspect, though, that
it is a mark of all language everywhere, and that its presence or
absence, or dominance over other tropes, cannot usefully be em-
ployed in the classification of discourse.

REFERENCES
M. H. Abrams, 1971. A Glossary of Literary Terms (New York: Holt, Rinehart & Winston).
Venerable Bede, 1973. De schematibus et tropis, in: J. M. Miller, M. H. Prosser, T. W.
Benson, eds. Readings in Medieval Rhetoric (Indiana UP).
George Campbell, 1776. The Philosophy of Rhetoric (London).
"Cicero," 1954. Rhetorica ad Herennium, trans. Harry Caplan, Loeb Classical Library. (An
anonymous work for long attributed to Cicero, and still indexed under his name in
most libraries.)
J. A. Cuddon, 1982. A Dictionary of Literary Terms (Harmondsworth: Penguin).
Francis Cuthbert Doyle, O.S.B., 1893. An Introduction to the Study of Rhetoric (New
York).
Abraham Fraunce, 1588. The Arcadian Rhetorike (London).
Geoffrey of Vinsauf, 1968. Documentum de modo et arte dictandi et versificandi, trans.
Roger P. Parr (Marquette UP).
Group M(J. Dubois et al.), 1981. A General Rhetoric, trans. Paul B. Burrell and Edgar M.
Slotkin (ohns Hopkins UP).
Albert Henry, 1971. Metonymie et Metaphore (Paris: Klincksieck).
Roman Jakobson, 1971. "Two Aspects of Language and Two Types of Aphasic Disturb-
ances," in his Selected Writings,Vol. II (The Hague: Mouton), pp. 239-259.
George Lakoff and MarkJohnson, 1980. Metaphors We Live By (Chicago UP).
Richard A. Lanham, 1968. A Handlist of Rhetorical Terms (California UP).
David Lodge, 1977. The Modes of Modern Writing (London: Arnold).
Matthew of Vend6me, 1968. Ars versificatoria, trans. Aubrey E. Galyon (Iowa State UP).
Alex Preminger, ed., 1974. Princeton Encyclopedia of Poetry and Poetics (Princeton UP).
Quintilian, 1920. Institutio Oratoria, trans. H. E. Butler (Loeb Classical Library).
Peter Ramus, 1578. Scholarum Rhetoricarum in Ciceronis Oratorem (Basle).
Richard Whately, 1828. Elements of Rhetoric (Oxford).