Beruflich Dokumente
Kultur Dokumente
Dennis Sobolev
904
carried out over the last decades have shown the extreme complexity
of this seemingly simple operation of synthesis; metaphor has revealed
itself to be an embodiment of analytical difficulties, as well as a source
of irresolvable contradictions between different approaches.
Although most scholars who have written about metaphor agree that it
is central to the understanding of poetic language, this seems to be the
only subject upon which a relative consensus has been achieved. None
of the theories propounded in the field has been able to convince the
majority of the academic community; and for every theory that has been
proposed, numerous counterexamples have been found. In light of this
situation, two mutually exclusive hypotheses become possible. One can
either state that the failure to find a sufficiently cogent solution to this
problem is just a temporary difficulty, and thus the philosophers of language and literary critics should continue looking for it. Or, conversely,
one can suggest that the whole discussion has been misguided and that
there is no such thing as the structure of metaphor, or, similarly, that
there is no such thing as metaphor in the traditional sense of the term.
I believe, however, that both radical positions are misguided. This essay
aims to demonstrate that metaphor, even though it is formally identifiable,
is not a single unified structure, but rather a field of heterogeneous possibilities, which is organized along several independent axes and is limited
by border parameters. Correspondingly, it is the culturally constructed
possibility of identifying metaphor that is responsible for the illusion
of the existence of its singular essence, while it is the heterogeneity and
multiplicity of metaphorical functions that is responsible for both its
actual applications and the proliferation of the theories of metaphor.
The firstpreliminaryproblem of metaphorical discourse is terminological. In order to account for the structure of metaphor, different sets of concepts have been proposed; the most known of them are
I. A. Richardss distinction between tenor and vehicle, and Max Blacks
double dichotomy of focus and frame,3 principal and subsidiary
subjects.4 In Blacks later essay, the latter dichotomy was renamed as
that between primary and subsidiary subjects.5 Significantly, as Black
stresses, the difference between his system and that of Richards is not
only terminological; and their dichotomies do not overlap.6 Therefore,
the choice of one of these conceptual sets has far reaching implications.
It should be stressed, however, that the analysis of metaphors in terms
of a dichotomy (of any dichotomy) implies that the general structure
of metaphor is bipartite; and this has yet to be proven. Moreover, to my
mind, this is a misconception. This structure includes a frame, which
should be interpreted literally, a word (or several words) that are used
figuratively (in other words, whose literal interpretation seems to be
impossible or insufficient in a given case), and a theme (an object,
metaphor revisited
905
906
metaphor revisited
907
908
tion, in turn, can clarify the meaning of the bipartite structure of the
identification of metaphor as a whole. To be precise, the above attributes
are not exclusive features of metaphor; they rather indicate the presence
of a rhetorical operation. Some of them can also indicate a metonymy,
while others may be the intimations of irony. Thus, in order to conclude
that a given figure is a metaphor, an additional condition (or conditions)
must be fulfilled. Once again, following Kittays suggestion, I will call it
the sufficient condition. At the same time, I can hardly agree with her
description of this condition.
Defining these sufficient conditions, Kittay writes that it is sufficient
for an utterance to be a metaphorical utterance either if the utterance
was intended to be understood metaphorically or if it is possible to attribute a metaphorical interpretation . . . [it] must be either intended
as such or must be taken as such.12 To my mind, this definition has
three serious shortcomings. First, in principle, almost any utterance
can be taken metaphorically, and thus she endlessly widens the subject.
Second, she defines metaphor in terms of pragmatics, and moreover,
introduces intentional and affective components into the very definition. For literary criticism, this would mean that the language of Homer,
whose intentions we cannot know, is metaphoric only if taken as such;
and this is definitely counterintuitive. Finally, Kittays functional definition in the style of Ernst Cassirer seems to be based on a logical circle.
When she writes that metaphor must be intended or taken as having
a metaphorical function, she actually defines the structure by means of
which metaphor can be identified (the structure of identification in
terms of this essay) through a reference to its function. Yet, as has already
been stressed, the question of the functioning of metaphor is the most
complicated and controversial question in the whole field of rhetoric,
and it is closely associated with the same problem of identification. Consequently, to say that metaphor can be called metaphor only if it was
intended or interpreted as metaphor merely redirects the discussion of
the procedure of identification to a much more complicated question:
what is metaphor? or how does it function? In reality, however, such a
circle is not indispensable. The sufficient condition of the identification
of metaphor has been widely known since Aristotle: this is similarity or
resemblance.13 However, although it was the basis for the understanding of metaphor in traditional rhetoric, in the fifties it fell into disfavor:
Black questioned its centrality,14 while Nelson Goodman dismissed it completely.15 Yet, subsequently, Paul Ricoeurs compelling analysis of Blacks
and Goodmans case against resemblance restored its importance for
the understanding of the functioning of metaphor.16 Donald Davidson
underscored it;17 George Miller proposed a theory of metaphor that focuses on relational (rather than substantial) similarities and analogies;18
metaphor revisited
909
910
a lion in a physical fight. Thus, this rhetorical figure fulfills both one of
the necessary conditions (Samson was not a lion; it is false on empirical grounds) and the sufficient condition (it emphasizes a similarity);
and hence, it is a metaphor. By contrast, calling the epoch of cars the
epoch of Ford does not foreground any similarity whatsoever between
Ford the man and the car (even though, after a while, one can easily
find many); the relation upon which this figure focuses is the relation
of production, which can be roughly classified as a causal one.
At first sight, this bipartite structure of identification might seem
commonsensical and even trivial, were it not able to clarify one of the
most controversial issueswithout any reference to other problematic
subjectslike the problem of the functioning of metaphor. At this point
of analysis, the reason for separating the question of identification from
the question of its functioning becomes clear; it has made it possible
to single out the object of analysis before the beginning of the analysis
itself and thus to avoid a vicious circle. The level of cultural competence
this analysis presupposes does not transcend that of an average native
speaker: the person must be able to identify explicit logical contradictions,
categorical incongruities, and clear contextual irrelevance, as well as to
possess basic competence in figurative language, which is necessary to
determine whether a given construction foregrounds similarity or not.
It also becomes clear why most metaphors involve a transition from one
semantic field to another. Within one semantic field, there are numerous
and diverse relations between different entities (part and whole, cause
and effect, they may belong to one class, and the like), and hence a
relation of similarity is often obscured by other relations. In contrast, if
the terms of metaphor belong to different semantic fields, a relation of
similarity becomes clear and visible.
To summarize: there is a peculiar type of discourse that we identify
as metaphoric; and there is a formal procedure through which one can
pinpoint this type of discourse. Now it is high time we addressed the main
issue of this essay. Having analyzed the identification of metaphorical
discourse, one has to analyze its functioning and its impact. As has already
been mentioned above, the range of answers to this question is impressive
and even somewhat overwhelming. Moreover, long and spirited debates
between different schools have complicated the situation even further.
It is not my intention, however, to analyze here the existing theories of
metaphor, their refutations, and the refutations of refutations. Instead,
I would like to propound a different framework for the understanding
of metaphoric utterances, within which many of the ideas and findings
of previous research cease to be mutually contradictory, while othersas
well as some objectionsreveal themselves to be misguided.
metaphor revisited
911
912
between the primary and the secondary terms of all metaphors. Even
Black has admitted that there are other types of metaphors, in addition
to the type he described (47), but, according to him, it is only his type
that is of importance in philosophy (45). Leaving aside Blacks evaluation of philosophical significance, I would like to say a few words about
different types of metaphor.
On the one hand, there are indeed such metaphors as man is wolf,
where the interaction between the terms is quite limited: the primary
term (man) limits to a certain degree the type of attributes that can
be transferred to it (those that are applicable to a human being), and
then these associated commonplaces are transferred quite mechanically,
without any juxtapositions, between the concepts of man and wolf. Yet
there are still other metaphors as well. For example, in order to make
sense of such a metaphor as Bill is a barn door,24 it is necessary to make
a close examination and comparison between John and the idea of the
barn door, not only because there is no standard system of commonplaces
associated with the barn door, but also because it is far from being evident that there are any points of similarity and interaction between the
two. Nevertheless, there are contexts in which such a metaphor can be
quite telling and meaningful. In these contexts, the interaction between
its terms can bring to the fore certain hidden similarities between them.
In such cases, the reader feels puzzled for a moment and then activates
the process of hermeneutic search and detailed comparison frequently
assisted by visualizationand often resorts to the hermeneutic strategies
described by Searle with reference to Gricean pragmatics.25 In these cases,
the resultant meaning of a metaphor is produced through an intellectual search rather than straightforward linguistic predication. One can
define this search process as the gradual foregrounding of a common
attribute. It is to this type of metaphor that Donald Davidsons remark
applies, when he writes: It may be remarked with justice that the claim
that a metaphor provokes or invites a certain view of its subject rather
than saying it straight out is a commonplace; so it is This view is neatly
summed up by what Heraclites said of the Delphic oracle: It does not say
and it does not hide, it intimates.26 However, in contrast to the popular
interpretation of the theory of interaction,27 which states the complete
symmetry between the terms of metaphor, this process of metaphorical
interaction is still asymmetrical; eventually, it results in the foregrounding
of certain attributes of the primary term in the first turn.
It is noteworthy, however, that in most empirical metaphors the
principles of transference and foregrounding are combined. Such
metaphors imply both the transference of associated attributes and
the necessity of a comparison between their termseven though these
components are mixed in different proportions. When, for example,
metaphor revisited
913
914
metaphor revisited
915
ately notices that almost all the cognitive content of this metaphor has
vanished. Actually, the relation between this parameter of metaphoric
utterance and the previous one is so close that one may even suggest
that it is the same parameter, although since it is viewed from a different
perspective, it entails essentially different types of theoretical problems. At
the present stage of research, however, I do not see definitive arguments
for its contingency, as well as its independence. Therefore, heuristically,
it is defined as an independent parameter, at least for the time being.
3. The problems of the type of relation between the terms of metaphor,
discussed above, are closely associated with another parameter: that of
the type of similarity upon which a metaphor is based. As has already been
stressed, metaphor always foregrounds similarity, although this similarity
can be of different types. Sometimes this similarity seems to be evident
and objective, and the metaphor is used only to emphasize it, as in case
of Achilles is a lion. This similarity can be labeled as a given. In
other cases, metaphors yoke togetherto use Dr. Johnsons unforgettable expressionthe most dissimilar things: twin compasses and lovers
in John Donne or evening . . . patient etherized in T. S. Eliot. Black
writes: it would be more illuminating in some of these cases to say that
the metaphor creates the similarity than to say that it formulates some
similarity antecedently existing.37 Kittay quotes Wallace Stevenss line,
The garden was a slum of bloom, and asks: What objectively given
similarity is there between a garden and a slum prior to the formation of
the metaphor itself?38 Trying to come to grips with the same problem,
Andrew Ortony introduces a distinction between predicate promotion
metaphors and predicate introduction metaphors,39 by which he means
a distinction between metaphors that just highlight the knowledge the
reader already has and those that state something new. Once again,
in most cases, these are only ideal poles. On the one hand, only a few
metaphors refer to subjects about which we know absolutely nothing; as
a result, any introduction of a new attribute reorganizes the structure
and the hierarchy of the existing ones. On the other hand, there are
very many metaphors that do not make it possible to deduce from their
secondary terms new cognitive information in relation to the primary
onesin addition to foregrounding and strengthening the importance
of some of its attributes. Moreover, in most cases, metaphors both use
evident, given similarities and produce new ones; such as Hopkinss
mornings minion, kingdom of daylights dauphin, dapple-dawn-drawn
Falcon.40 Clearly, there are numerous given similarities upon which
these metaphors are based, but they also draw the attention of the reader
to other similarities, which are not evident at all. In other words, again,
one does not deal here with the two opposed values that a metaphor
916
may take, but rather, with another axis, along which all metaphors are
situated, at different distances from its poles.
There is another question, however, that should be immediately asked.
Perhaps this is not an independent axis but only another perspective
from which we can analyze the already familiar parameters. Indeed, at
first sight, it seems that the transference of an attribute must be based
upon a given similarity, whereas metaphorical interaction creates this
similarity. In some instances, this is indeed the case; in others, however,
this intuition is misleading, and these axes are independent. John is a
pig functions differently in two different cases: in the case where the
listener knows much about John and in the case where he knows little
or nothing. In both cases, a standard attribute of the secondary term is
transferredand hence the value on the first axis is close to the ideal
pole transference. Yet, in the first example, this statement is based
upon a given similarity (and if there is contrast rather than similarity,
this statement is metaphorically false), while in the latter this similarity
is produced. In this latter case, if the axis of the type of relation has the
value transference and the axis of the type of similaritythe value of
production, the principal term of metaphorical statement (John in
the above example) acquires new attributes, structures, and relations. In
most cases, however, metaphor both uses existing knowledge and adds
something new. The same holds true for the opposite value on the axis
of the type of similarity: interaction. John is a kettle, which implies
interaction between the terms rather than transference, can be based
on a given similarity (at the present moment, John boils), as well as
create a new, perhaps far-fetched, resemblance. Taking a richer example,
one can see that the traditional metaphorical representation of death
as a departure to a final destinationa metaphor that may also imply
interaction, as it does in Emily Dickinsontransfers to death the structures that death, as a physical event, does not have.41 To put it briefly,
the third axis of the production of metaphorical meaning is that of the
type of similarity, which is independent from the previous axes and is
limited by the ideal pole values given and produced.
4. Another distinction, whose understanding is crucial, is that between
the metaphors of creation and those of elucidationwhich mark the
ideal poles of the axis of the object of metaphorical synthesis. Near the creation pole of this axis, one should place all the catachreses that actually create their objects, for which no other word in the world exists.
In this sense, it is important to remember that the significance of the
notion of catachresis is not restricted to such evidentand essentially
nonmetaphoriccases as the leg of the table or electromagnetic wave;
the elements of catachresis are present in all the metaphors that cannot
metaphor revisited
917
918
metaphor revisited
919
the perceptible pole, one will find, alongside others, those metaphors
that focus on visual similarities. Again, the necessity of stressing these
two possibilities results from continuous polemics upon the question of
whether the procession of metaphor must imply visualization or not.50
In reality, metaphors that are close to the former pole require little or
no visualization, whereas those belonging to the perceptible part of
the spectrum imply it as an integral part of the process of interpretation.
Thus, on the one hand, when in The Wreck of the Deutschland Hopkins
writes, I am soft sift / In a hourglass, he stresses his transparence before
God, who sees into the heart, as well as the slow, gradual, and inescapable dynamics of his spiritual progress.51 As a result, no visualization
is required for the understanding of this metaphor, even though the
visualization of an hourglass can become a significant assent for some
readers. Near the other pole of this axis, one will find Andr Bretons
line, My wife . . . whose waist is an hourglass, which underscores the
visual similarity between the waist of a woman and the shape of an
hourglass. Its correct interpretation does require a mental image of an
hourglass. In both examples, metaphors use the secondary terms in
order to make a statement about the primary term and to underscore
some of its features, but the mechanism of focus and projection works
in two essentially different domains. This, in turn, means thatas in all
the previous casesthe polemics about whether the interpretation of
metaphors uses mental images, whether it requires an effort of imagination and an act of visualization, is somewhat misguided: it may require
them, but this requirement is not universal.
8. Another major typological dichotomy of the ideal poles of metaphor
is between the metaphors of identification and the metaphors of juxtaposition
along the axis of the configuration of the terms. In the former case, the
identity between the primary and the secondary terms of metaphor is
explicitly stated (John is a pig); in the latterthe existence of a relation, if there is one, must be deduced by the recipient (for example, by
the reader); the former states a similarity clearly, while the latter only
implies one. It seems that it is this, or a comparable, distinction that
Philip Wheelwright had in mind when he coined his slightly enigmatic
dichotomy of two major types of metaphor. He wrote that epiphora is
standing for the outreach and extension of meaning through comparison, whereas diaphora causes the creation of new meaning by juxtaposition and synthesis.52 Epiphora often implies explicit predication (A is
B), while diaphora is based on simple juxtaposition; as an example of
the latter category, he quotes Pounds In a Station of the Metro: The
apparition of these faces in the crowd; / Petals on a wet, black bough.
If in epiphora semantic exchange results from a statement of identity or
920
metaphor revisited
921
922
one, and the change in the attitude towards it coincided with that in
evaluation of the frequency of use of metaphor in everyday speech and
communication.
Since the mid-1970s, numerous studies have demonstrated the ubiquitous presence of metaphors in human speech and thought. In all probability, the most influential of them was by Lakoff and Mark Johnson,
who cogently showed that different metaphors, far from being singular
and arbitrary instances of transference of an isolated sign, are often
based on more general conceptual transferences, which organize conceptual systems and structure human understanding of different aspects
of existential experience. This approach was developed in the works of
Lakoff, Gibbs, Turner, and many others. Finally, Lakoff suggested that all
the studies of metaphors should be reduced to the study of conceptual
transferences (or mappings in his terms), which he labels as conceptual metaphors.59 Correspondingly, metaphor in the traditional sense
began to be seen as a surface manifestation of conceptual metaphor,
and referred to as linguistic metaphor.60 It is this suggestion of the
cognitive school that must be corrected: not every metaphor is based on
conventional conceptual transference. Furthermore, to the best of my
knowledge, at least half of the most memorable literary metaphors are
not. It is impossible to say, for example, what the standard conceptual
transference behind Hopkinss unforgettable fell of dark is: DARKNESS
HAS BODY or NIGHT IS A CROCODILE.61 The same holds true for his
famous blue-bleak embers that fall, gall themselves, and gash goldvermilion.62 At the same time, the case of Hopkinss metaphor the mind
has mountains, cliffs of fall is more complicated. The representation of
the mind as a landscape, which is behind it, is indeed conventional. Yet,
it is not a part of the language in its common everyday form, but rather,
that of a more specific conceptual system of European Romanticism.
In addition, there are whole groups of metaphors, which are not
based on conceptual transferences; the so-called scientific metaphors
are among them. Thus, for example, from different points of view, light
can be described as either a wave (which has a front and a length)
or corpuscular (which has a charge). These metaphors have been
initially taken from essentially different conceptual domains; and their
relations have been completely restructured according to scientific
needs and concerns. Another group of metaphors, which exist outside
of the transferences of conceptual schemes, includes those idiosyncratic
metaphors, which make sense only in certain contexts but are completely
meaningless in others, like Searles Sally is a block of ice.63 Certainly,
one can say that this statement is an example of the conceptual metaphor MAN IS AN OBJECT, but this answer only foregrounds the general
problem of this theory as a whole. The problem is that the following
metaphor revisited
923
924
metaphor revisited
925
926
927
metaphor revisited
not all the collocations of values of these parameters are possible; and
this significantly limits the number of possible structures in comparison
to pure combinatory estimation. Thus, for example, the radical version
of creation along the axis of the basis of metaphorical synthesis is
incompatible with the value juxtaposition on the axis of configuration:
in simpler words, the same metaphor cannot be both a diaphora and a
catachresis, for diaphora requires both its terms to be explicitly present.
This does not mean that these axes are not independent, but rather,
that in the space of metaphor there exist combinations of values that
are prohibited for theoretical reasonsexactly as certain combinations
of parameters might be prohibited in physics or in chemistry.
To put it briefly, metaphor as such is only a field of different possibilities,
which is unified owing to the existence of the structure of identification
(and, consequently, the external form of metaphor), rather than a clear
and unambiguous essence. As a result, any attempt to define a single
essence of metaphor may be compared to the old Buddhist parable
of the blind men who try to describe an elephant: one, who holds its
tail, believes that it is a rope, while the others hold its trunk, leg, belly,
and ear. Like an elephant, metaphor is neither a rope, nor a trumpet
or a pillar, nor a winnowing basket or a storage bin; but in a sense, it
can become any of them. The pure eidetic concept of metaphor, like
pure existence, is not an essence but only a field of possibilities. And
this, in turn, is the answer to the question why most metaphors are not
paraphrasablebecause of the multiplicity and heterogeneity of possible
metaphorical functions. Indeed, in principle, one can find a functional
analogue for almost any single parameter, when it is isolated from the
whole structure, but to find another linguistic entity that will perform
a dozen different functions exactly as a given entity does is, in most
cases, impossiblesomething that is quite predictable from both logical
and theoretical points of view. To return to one of the examples above,
metaphor is a metaphor is a metaphor is a metaphor.
University of Haifa
Notes
1 Jonathan Culler, Part One: Structuralism and Linguistic Models, in Structuralist Poetics:
Structuralism, Linguistics and the Study of Literature (Ithaca, NY: Cornell Univ. Press, 1975),
3109.
2 Roman Jakobson, Language in Literature, ed. Krystyna Pomarska and Stephen Rudy
(Cambridge, MA: Belknap Press, 1987), 95117.
3 Max Black, Metaphor, in Models and Metaphors: Studies in Language and Philosophy
(Ithaca, NY: Cornell Univ. Press, 1962), 28.
4 Black, Metaphor, 39.
5 Black, More About Metaphor, in Metaphor and Thought, ed. Andrew Ortony (Cambridge: Cambridge Univ. Press, 1993), 27.
928
metaphor revisited
929