Commentary

Author(s): Jonathan Culler

Reviewed work(s):
Source: New Literary History, Vol. 6, No. 1, On Metaphor (Autumn, 1974), pp. 219-229
Published by: The Johns Hopkins University Press
Stable URL: http://www.jstor.org/stable/468350 .
Accessed: 27/03/2012 10:27
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
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 [email protected].

The Johns Hopkins University Press is collaborating with JSTOR to digitize, preserve and extend access to
New Literary History.

http://www.jstor.org

Commentary
Jonathan Culler

HERE SEEM ESSENTIALLY

two ways of thinking about metaphor,

which we might christen the via philosophica and the via

rhetorica. The former locates metaphor in the gap between
sense and reference, in the process of thinking of an object as something. The latter situates metaphor in the space between one meaning
and another, between the literal or "proper" verbal expression and
its periphrastic substitute. And so whereas the former makes metaphor
a necessary and pervasive feature of all language, which with its verbal
detours gestures obliquely towards a world of objects, the latter makes
it a special use of language which can be isolated and studied against
the background of a nonmetaphoric use of language.
Stated thus, these positions appear alternatives: two definitions, each
of which delimits a different domain and calls it metaphor. If this
were indeed so, all would be well: one would choose a definition, give
principled reasons for one's choice, and proceed happily with one's
discussion. But in fact, as soon as one tries to justify one's choice one
finds the object slipping away towards the other pole of definition. One
continually catches oneself presupposing the position one had rejected;
and one can only conclude from the unhappy throes of theory that the
domain of metaphor is rent and distended by the paradoxical relationship between these two approaches. They must be kept distinct if
one is to define a coherent object, but can they be kept distinct if one's
own discourse is to remain coherent?
Both the importance and the resiliency of this problem are amply
illustrated by the treatment it receives in Jacques Derrida's and F. E.
Sparshott's essays. Sparshott maintains in a footnote (n. 25) that
the distinction between thinking of something as what it is and
thinking of it as what it is not does not coincide with the distinction
between the literal and the metaphorical, "which is a distinction between ways of using words." The via philosophica and the via rhetorica
are different roads. It is striking that this claim should be relegated
to a footnote, though it is, as he admits, the topic of his essay. It

220

NEW LITERARY HISTORY

cannot be made to emerge as the conclusion of a coherent argument,

and indeed, his discussion elegantly explores the paradoxes that arise
from the inevitable interplay of the two approaches. Jacques Derrida,
on the other hand, argues that the two enterprises cannot be carried
out separately. One cannot study metaphor as a use of language in the
philosophic text because metaphor implies a whole philosophic framework (of essence and accident, identity and difference) which prevents
it from being a use of language in the text and makes it a condition
of the language of the text. Yet when he tries to imagine a metaphilosophic study that would subsume both approaches to metaphor,
he offers as specimen Bachelard's work on the constitutive functions
of metaphor in poetic and scientific discourse: a project which relies
on the possibility of identifying, against a background of nonmetaphoric language, the metaphoric uses of language in a text. The preliminary arguments which led Derrida to Bachelard's method seem
contradicted by the presuppositions of that method.
It may be, of course, that these two approaches engage us in a dialectic whose synthesis we are as vet unable to imagine, but it might also
be that the most salutary lesson we can learn from metaphor is how
to live a conceptual contradiction, how happily to situate ourselves at
the metaphoric nexus where incompatibles are said to be fused and yet
not fused. Fortunately, the function of a commentator is not to produce a brilliant Hegelian sublation of the problems which exercise contributors but to comment. Let us briefly explore some of the implications and difficulties of these two approaches.

I
The staple argument of the via philosophica is that abstract terms
are metaphorical: "grasping" an argument involves metaphorical
grasping just as a stormy countenance involves metaphorical storms.
Indeed, this Lockean approach which locates "reality" in objects and
sensations makes all conceptualization a metaphorical process: to
think of a discrete particular under a heading is to think of it as something, and such thinking is of interest and value only because the
"something" in question is other than the particular. The only nonmetaphoric language, it seems, would be logically proper names, and
paradoxically, as Sparshott and others have pointed out, logically
proper names are the one thing natural languages do not have. To
call anything by a name in a natural language is to ascribe to it some
properties, however valgue, and thus to bring it under some loose head-

COMMENTARY

22 I

ing. As soon as we begin to speak of a thing we engage in metaphor.1

Of course such an approach collapses a distinction which we
initially wished to make and obliges us to invent a new term if we would
distinguish calling my love a rose from calling that flower a rose. Still,
this collapse might be justified by the conceptual gain, were it not for
the fact that our conclusion contradicts our initial premise. We began
by distinguishing between real grasping and metaphorical grasping,
between a sign given its proper meaning and a sign given an extended
or distended meaning, yet this distinction between two types of linguistic usage was the instrument which led us to exclude from language altogether the "proper," nonmetaphoric use of language. Our
whole reflection was based on the notion of a "proper" use of signs,
yet our conclusion, as Derrida notes, destroys any such notion.
What tempers this contradiction, makes it livable, is a temporal
perspective: in any given state of a language we can distinguish between metaphorical and nonmetaphorical usage, but what is now seen
as nonmetaphorical is metaphoric with respect to an earlier state of
the language, and the proper usage of that state is in turn metaphorical with respect to a still earlier state. Derrida says that by reading
"in a concept the hidden history of a metaphor we are giving a privileged position to diachrony at the expense of system," locating the
truth of a sign in its origins rather than in its present function; but
in fact this move is designed to tame time and its effects and by so
doing to reduce the arbitrariness of signs. The metaphorical detour
is an historical space, an ambiguous space of usure ("wear and tear"
but also "usury" which adds value); but its ambiguities are motivated,
deposits of a fundamental rationality which the synchronic state of a
language may conceal. To call abstract terms metaphors is to say
that they are not arbitrary names but motivated extensions or developments. And indeed, eighteenth-century etymological reasoning, which
is only the most extreme version of this approach, shows that time is
invoked not in the interests of historical accuracy but as a fiction which
enables one to motivate signs. Consider an example from Horne
Tooke's Diversions of Purley:
A bar in all its uses is a defence: that by which any thing is fortified,
strengthened, or defended. A barn is a covered enclosure, in which the
grain, etc. is protected or defended from the weather, from depredations,
etc. A baron is an armed, defenceful, or powerful man. A barge is a strong
boat. A bargain is a confirmed, strengthened agreement. . . . A bark is a
I In addition to Derrida's essay, see Paul de Man, "Theory of Metaphor in
Rousseau's Second Discourse," Romanticism, ed. D. Thorburn and G. Hartman
(Ithaca, 1973).

222

NEW LITERARY HISTORY

stout vessel. The bark of a tree is its defence: that by which the tree is
defended from the weather, etc.
The bark of a dog is that by which we are defended by that animal.2
The nouns of our language are not arbitrary signs but metaphors, motivated by a resemblance with a primitive arbitrary sign (bar). So once
again we are brought up short by a fundamental paradox: an approach which makes the proper use of language an arbitrary naming
deploys all its energy in showing that language as we know it is improper, metaphorical, and, for that very reason, eminently natural
and rational. Along this road paradoxes and contradictions continue
to accumulate; if one cannot fight one's way through to a coherent
discourse the best strategy is to retreat and try another road.

II
The via rhetorica seems at first to offer firm ground: we are enjoined to locate ourselves firmly within language and to study one
specific type of usage. In metaphor one name is replaced by another,
and the task of the student of metaphor, on this model, is to investigate
the types of replacement, the various relations between intended meaning or tenor and expression or vehicle. The via rhetorica, which
Aristotle was only the first to explore, leads to a typology such as
Tzvetan Todorov offers. Among the relations between one verbal
form and another-between what is meant and what is said-metaphor comprises one group, as opposed to the relations of other figures,
such as synecdoche and metonymy. This approach, as Paul Ricoeur
notes, works best for the least interesting metaphors, those "raids
on the bestiary" where the sign replaced by the animal name is
ready to hand. Problems arise in the case of creative metaphors
where we find translation difficult, either because we know what object
the metaphor refers to but are uncertain about the precise grounds
of comparison (e.g., Shakespeare's "bare ruined choirs where late the
sweet birds sang"), or because we grasp the general tenor of the metaphor but are uncertain what it is replacing (e.g., Eliot's "I have heard
the mermaids singing, each to each"). In these cases it is not easy
to define the relevant relations between the initial object, the metaphorical expression, and the qualities ascribed to the object by that
expression; but this does not make them less metaphorical than the
2 Diversions of Purley (London, 1805), II, 182-83. Quoted by Hans Aarsleff,
The Study of Language in England, 1780-i860
(Princeton, 1967), pp. 63-64.

COMMENTARY

223

definable cases. On the contrary, precisely because of their typological

indeterminacy they are rather better examples of the metaphorical
function as practiced and valued in literature than are the pallid replacement figures which theorists discuss.
The difficulties of the via rhetorica are nicely illustrated by Pierre
Fontanier's discussion of catachresis, whose salient moments Derrida
cites. Espousing the traditional view of metaphor as replacement,
Fontanier is led to exclude from the domain of figures proper those
tropes where he finds it difficult to speak of replacement. Catachresis
occurs where "a sign already assigned to a first idea is also assigned
to a new idea which had no expression."3 Thus catachresis includes
many expressions which to speakers of a language seem "le mot
propre," the only expression furnished by the language for the idea in
question. In this sense they are irreducibly different from figures of
replacement, but Fontanier argues that they are "determined by the
same relations" as are rhetorical figures proper: "correspondence,
connection, or resemblance between ideas; and they operate in the
same way: as metonymy, synecdoche, or metaphor."4 Thus the notion
of replacement, on which the whole typology of figures depends, seems
to fall away as irrelevant when tropes not based on replacement are
placed in the same taxonomic categories. Metonymy, synecdoche, and
metaphor are defined by the rhetorical approach in terms of replacement, but then tropes not involving replacement are brought under
these three headings.
That is only the first contradiction. The definition of catachresis
brings together the two phenomena which are excluded from figures
of replacement: the "dead" metaphor or metonymy which in the
language functions as a nonmetaphorical expression, though we can
perceive its metaphoric origin, and the truly creative live metaphor
which is not simply a substitute for another expression. In both cases
a sign is used to express an idea for which no other expression exists,
yet from the point of view of the function of tropes these two cases are
irreducibly different, and a rhetorical approach to metaphor, supposedly predicated upon the suasive and expressive functions of language, ought to be able to define this difference, which after all is much
more fundamental than the technical distinction between synecdoche
and metonymy.
Third, the notion of catachresis brings together in a curious amalgam
both the natural and the unnatural. The word "catachresis," Fontanier
writes, "expresses very well their nature and use, since it signifies abuse,
3
4

Les Figures du discours (1821; rpt. Paris, 1968), p. 213.

Ibid., p. 214.

NEW

224

LITERARY HISTORY

and the extension of meaning is a kind of abuse."5 One forces a

word out of its proper sense into another sense, and this is not merely
a poetic license which expires at the end of the poem but a true abuse
of the word since, in the absence of another term which might succeed
its temporary replacement, the word thus forced becomes the proper
expression of the new idea. Yet precisely because there is no other
expression, the abuse becomes "natural." The wings of an army and
of a building, the arms of the sea or a chair, the head of a nail or a
head of lettuce, are "so many forced metaphors, although doubtless
just and natural."' And after another list he asks, "without these
forced metaphors, these catachreses, how could we succeed in capturing and representing these ideas?" 7 The distinctions between the proper
and improper, between tropes of replacement and of nonreplacement,
and between the abusive and the natural, on which the rhetorical
analysis of metaphor had seemed to depend, collapse in the attempt to
deal with the varieties of metaphor. Moreover, the via rhetorica seems
to have led us back to the imbroglio of word and object, of proper
meanings with rhetorical origins, which led us to take it when retreating from the via philosophica.

III
These problems may be insoluble, may be constitutive of the
domain of metaphor, but it is at least worth inquiring whether we
might succeed in avoiding them by ignoring the concept of metaphor
and approaching obliquely, from a different perspective, the phenomena usually discussed as metaphor. The typologists of the via
rhetorica adopt the perspective of the writer or creator and thus beg
the most interesting questions, for if one assumes first the presence of
an idea to be expressed and then the selection of a metaphor to express
it, one is dealing with cases where the expression is already defined as
,a metaphor and as the sign of a particular idea. The question of what
makes something a metaphor and how it might signify is set aside
as already resolved. If, on the other hand, we reverse the perspective
and take, as Paul Ricoeur recommends, the point of view of the reader
confronted with a written text (language cut off from an originating
5

Ibid., p. 77-

Ibid., p. 2 16.
7 Ibid., p. 217. The French reads "comment . . . eit-on pu parvenir a retracer
ces idees?" which does not mean, pace Derrida's translator, to "trace these ideas
The ideas exist virtually in the mind and are traced,
back to their origin."
recaptured, represented by the new sign.
6

COMMENTARY

225

intention), we may then ask what makes something a metaphor for

the reader, and a rather more interesting series of problems arises.
Metaphor becomes a derived rather than primary phenomenon, the
detritus of a particular interpretive operation, whose conditions and
determinations are more interesting than the phenomenon itself.
The question of metaphor arises only when there is a problem in
the text, a perception of incongruity. And therefore in the first instance the notion of metaphor depends on a series of models of
vraisemblance and coherence which allow us to perceive a statement as incongruous or incoherent. Our models of appropriate discourse tell us that there is something wrong with Dylan Thomas'
line, "And there this night I walk in the white giant's thigh," and
that some kind of interpretive operation must be performed on it. But
this conclusion depends in turn on conventions about the unity and
referentiality of literary works. If we were to take Thomas' poem as
referring to a world of utter fantasy, then we could read his assertion
quite literally; it would require no interpretation (though the fact
that he is writing about this kind of fantasy world would need to be
interpreted). Or if our conventions of literature did not enjoin us
to integrate this assertion with the rest of the poem, we might allow
it to stand forth as an example of literal unintegrated fantasy.
No statement, in short, is metaphorical in itself. We make it metaphorical when we yield to the reality principle and strive to produce
an accustomed intelligibility; when we assume the unity, the vraisemblance and the referentiality of the text. "Look, the morn in russet
mantle clad!/Walks o'er the dew of yon high eastward hill" becomes
metaphorical only because we take Hamlet as generally referring to a
familiar world and because we assume a unity which would proscribe
a moment of pure hallucinatory fantasy (Horatio is allowed to see
ghosts but not to see the morning walk or wear mantles; such are our
purely conventional interpretive constraints). Committing ourselves to
taming the strange, to bringing it wholly within our ken, we extract
from the sequence what might "really" belong to dawn, label the
phrases "metaphor," and read the constructed detour as an index
of wonder and poetic intensity.
That metaphor is not a primary element of literature but only the
detritus of conventional interpretive operations is amply shown by our
difficulties in identifying metaphor in the texts of surrealism or dadaism,
where accustomed standards of unity and referentiality are questioned.
Consider the beginning of John Ashbery's poem "They dream only of
America":

NEW LITERARY HISTORY

226

They dream only of America

To be lost among the thirteen million pillarsof grass:
"This honey is delicious
Though it burns the throat."
And hiding from darknessin barns
They can be grownups now
And the murderer'sashtrayis more easilyThe lake a lilac cube.
We cannot identify some phrases as metaphors because we have difficulty constructing an imagined world and deriving from it a level
of vraisemblance which would indicate those incongruities to be reduced. Nor can we be sure which items should be allowed to stand free
as unconnected fragments of another discourse and which should be
reduced in the interests of unity. Any particular reading we offer would
throw up some metaphors, depending on what we take literally and
make a constituent of this world.8 These metaphors are the jetsam
of interpretation: what cannot be integrated as such and is not allowed
to stand apart but is raided for its content before being sloughed off
as a poetic indicator.
In the perspective of a poetics, which focuses on the conventions and
operations of reading, the conditions under which the reader produces
metaphor are more vital components of literature than the operation
of metaphorical interpretation, which is but a special and extremely
limiting instance of interpretation.9 When we call something a metaphor,
we invoke a traditional model of replacement which says that the reduction of the strange is easily accomplished, that we can replace what
is said by what is meant, because we have discovered what propertNy
the former and the latter share. We pass through the class of beautiful
things from "red, red rose" to "my love," or through the class of sly
creatures from the "fox" to "that man." When one comes upon this
basis of metaphor in the context of a theory of reading one sees what
a paltry thing it is and how minor a role it plays in our actual experience of literature. The incongruous phrases in a work which interest
and exercise us are never naturalized by a simple metaphorical operation. If one seeks examples of metaphor in literature, clear cases prove
extremely elusive (which is why theorists have frequent recourse to
the bestiary), for most figures bring into play a variety of strategies
of literary interpretation.
If we call Eliot's "I have heard the mermaid's singing, each to each"
8

For discussion

see my Structuralist

Poetics

(Ithaca,

1974),

pp. 169-70,

metaphor generally, pp. 78-79, 85-90, 94, 178-83.

9

For this general

perspective,

see Structuralist

Poetics,

Chs. vi-viii.

and for

COMMENTARY

227

a metaphor, we imply that we know what it stands for. But if we wish

to linger in these chambers of the sea, we must discard the metaphoric
model so as to perform other interpretive operations: balancing
Donne's invocation of an impossibility ("Teach me to hear mermaids
singing, or to keep of envy's stinging") against the assertion of an
experience, weaving the delicate nostalgia of exclusion into the structure of values which must include the marine, the exotic, the different
feminine milieux. The point is not, let us hasten to add, that the
phrase synthesizes all this as the acolytes of metaphoric fusion would
have us believe. Rather, its incongruity, its initial invraisemblance,
enjoins us to set to work, to explore, to split up the phrase which is incongruent per se into components which we can trace through the
poem, discovering how they are associated, transformed, opposed, and
valued. The phrase is not a synthesis, a distillation of a meaning, but
a stimulus to deconstruction, a nexus of semantic trails which we
can follow in various directions.
Every sign, says C. S. Peirce, requires an interpretant, some explanation or argument which enables it to be used; and since that interpretant
is itself another sign, it will require further interpretants in the form
of yet other signs.'0 Interpretation is woven in a potentially infinite
process and it is only the thoroughly conventional metaphor which
precludes the developments natural to interpretation. If that man is
a fox, he is sly and nothing more, but if he is a hippopotamus, or a
tattered coat upon a stick, or a pair of ragged claws scuttling across
the floors of silent seas, the phrase gestures towards an open meaning,
functions as a signifier which designates a space of signification. It
provides material which we can trace through the text, subjecting it to
various developments, so as to fill up the space of meaning which it
designates, piling sign upon sign. Any interesting incongruity can call
forth minor metaphoric, synecdochic, metonymic, and symbolic operations as we strive to release what Derrida calls its secret story: the
developments to which it gives rise and which we organize, calling
upon our models of unity and vraisemblance, into an extended naturalization of the phrase.
For a theory of reading, the interesting cases of so-called "metaphor"
are, as Paul Ricoeur implies, occasions for the exercise of interpretation in its most general and complex forms. Nothing is gained by
calling these moments of the text "metaphor." Indeed, much may be
lost, since by bringing them under this heading we imply that there
is substitution at work and thus commit ourselves to the notion of an
io

Collected Papers (Cambridge, Mass., 1931-58),

II, 136-37-

228

NEW

LITERARY

HISTORY

easily definable reduction of the strange. By identifying "metaphors"

we trivialize the processes of interpretation.
Indeed, it will no doubt prove true that those who resist abandoning
the term do so not from any love of identifying metaphors, nor even
from a desire to discuss metaphors. Image-criticism has been frequently attacked and its day seems to have passed. If it attempts a
revival by studying, as Robert Weimann suggests, the tenor or ultimate
meaning of metaphors rather than concentrating on vehicles, there
would no longer be any reason for it to limit its objects to supposed metaphors." The resiliency of the term "metaphor," its continued presence
in literary criticism, may derive rather, as Cyrus Hamlin's essay indicates, from the love critics bear not for metaphor but for the metaphoric, not for metaphors themselves but for the concept of the metaphor. If poetry attempts to fuse particulars without forgetting their
particularity, if it seeks to lodge the eternal in the temporal or the
transcendent in the empirical while suggesting the noble futility of
this task, then what better name is there for the poetic than the metaphoric. Rejecting the notion of symbol as standing for an ultimate
fusion which we believe impossible, Hamlin writes about the importance
in Romantic aesthetics of the metaphoric, conceived as an ironic or
self-conscious tension between unity and disparity, atemporality and
temporality. He cites no actual metaphors in his discussion, and indeed
there is no reason why he should, because this aspect or this function
of poetry does not especially depend on poetic metaphors.'2 The metaphoric here is simply a name for a theory of literature-a theory which
is, of course, that late blossom of Romantic aesthetics, the theory of the
New Criticism. Scrapping the term "metaphor" would not harm
such theories and would, on the contrary, force them to look more closely
at the interpretive processes which produce a tension between unity
and disparity and at the various textual indeterminacies which provoke
such reading.
It is, however, important to point out, pace Weimann, that the error of
I
image-criticism was not excessive formalism and that indeed there is nothing
formalistic about it. To relate "the slings and arrows of outrageous fortune" to
methods of primitive warfare is immediately to reduce form to content without
even allowing the content of the speech to take shape. To prefer the content
implied by isolated words (the verbal material of the play) to the content of
words as organized by and in the dramatic structure is the most radical of
content-centered approaches and the furthest remove from a formalist approach
to metaphor. For an example of the latter see G'eoffrey Hartman, "The Voice of
the Shuttle," Beyond Formalism (New Haven, 1970).
12 Canonical metaphors, such as "George is a fox," do not have these exalted
poetic functions. Indeterminate metaphors, such as "I have heard the mermaids
singing each to each" may do so (only because they are not simple figures of replacement), but so may metonymies, synecdoches, and other stimuli to interpretive
displacement, if in their context they have the right kind of ambiguity.

COMMENTARY

229

The change in perspective achieved by abandoning metaphor would

also permit an elegant resolution of awkward problems raised by the
rhetorical concept of metaphor. Speculating on "impossible metaphors," Todorov asks why it is that we cannot say "fork" for "knife,"
"door" for "window," "wolf" for "fox." To the rhetorician this is a
difficult problem, since the first term is related to the second by resemblance, just as in the case of possible metaphors. But if we scrap this
model and turn to the perspective of reading, the reasons become
plain. "Fork" is never a metaphor for "knife" because we never
encounter incongruous sentences containing the word "fork" where
substituting the word "knife" would be a likely and satisfactory solution. "I sharpened the blade of my fork" suggests simply an odd
fork. "I stood at the bedroom door looking out through its upper pane
into the street" implies an oddly designed house; and "the thin reddishbrown snout of the wolf" gives us an unusual wolf. We never make the
first terms into metaphors for the second because the interpretive operations which would vield this result are never necessary. The rhetorical
approach to metaphor simply throws up false problems which reading
ignores.
If it is difficult to think coherently about metaphor or in terms of
metaphor, it is perhaps because we have long tried to hold together
through this notion two things which are fundamentally incompatible,
and on whose very incompatibility the power of literature depends: the
notion of a definable rhetorical operation in which I say X and mean
Y and the open-ended violations of vraisemblance through which we
are invited to explore, develop, and fill in a space of signification.
Literature is of interest only because its signs do not simply replace
other "proper" signs, or rather because its signs do, precisely, stand for
something else, for the "other" in its possible developments, and thus
are different from a meaning which they defer. Literature's power has
been thought to lie in metaphor, but in fact it is precisely literature's
resistance to metaphor, resistance to replacement operations, which is
the source of this power.
To say that the notion of metaphor should be scrapped, that it is a
positive hindrance to our understanding of reading because it conceals
the complexities of interpretation, is an ungenerous conclusion to an
issue on metaphor; but the best essays here presented point this way.
They explore the paradoxes and impossibilities of the notion (sometimes by illustrating its triviality), or else escape from it to the general
problems of interpretation.
BRASENOSE
OXFORD

COLLEGE,