You are on page 1of 14

History and Theory, Theme Issue 50 (December 2011), 136-149

Wesleyan University 2011 ISSN: 0018-2656

THE TRANSFIGURATION OF DISTANCE INTO FUNCTION


Frank Ankersmit
Abstract

The point of departure of this essay is the intuition that the relationship between the past
and the present (or between the past as the object of historical investigation and what is
said about it by the historian) should be conceived of in terms of temporal distance. The
spatial metaphor of distance at work in this intuition is thought to provide the basis for the
epistemological model appropriate for understanding the nature of historical knowledge.
This results in two claims: 1) epistemology is the philosophical instrument we must rely
upon for understanding historical writing, and 2) the metaphor of distance iswhether one
is aware of it or notthe model for most, if not all, epistemological thought. This essay
discusses the pros and cons of these two claims. It argues that the two claims are indeed the
best way to begin our analysis of the relationship between the past and the historical text
or representation. However, we cannot afford to stop there; indeed, we must ask ourselves
where the associations we have with the metaphor of temporal distance may, in the end, be
misleading. This will enable us to recognize that the notion of distance will, finally, have to
yield its prerogatives to that of the notion of function. Historical writing is functionalist in
the sense that the historical text is a substitute for the past discussed in it. That is its function. Hence the essays title.
Keywords: distance, metaphor, epistemology, function, representation, Donald Davidson
I

Distance is a spatial concept: its proper and literal use refers us to objects in different places in (three-dimensional) space. Self-evidently, this should make us wary
of using the notion outside contexts where spatial distance is at stake. For in such
contexts the notion can make sense only in a metaphorical way. Such metaphorical uses of the notion of distance may yield illuminating insights, but the validity
of such insights can be ascertained only after having translated the language of
spatial metaphor into that of non-metaphorical literal prose.
But it is often far from easy to satisfy the demand for the literal translation of
spatial metaphors. This is the danger of spatial metaphor, in the sciences no less
than in philosophy.
Think of history. History investigates the past, hence what is temporally remote
from us. But can we conceive of what is temporally remote from us without an
appeal to the notion of temporal distanceand, hence, of the spatial metaphor of
distance? Anyone thinking of the past has to rely on the axis of time on which the
so-called arrow of time moves from left to right, where the past is to the left of

THE TRANSFIGURATION OF DISTANCE INTO FUNCTION

137

us, the future is on the right, and the present is in the middle. Suppose you were
required to imagine the past without any reference, either explicit or implicit, to
this openly spatial notion of distance. I must confess I have no idea how to do this.
In matters of time, and of the past, there is indeed something ineluctable about the
spatial metaphor of distanceand this undoubtedly is what made the organizers of this conference put the notion of distance on the agenda. It is as if you can
speak about time and the past only metaphorically. Or, to put it differently, you
can clarify the notion of time only in terms of that of spaceand vice versa.
Kants first Critique seems to confirm this intuition. In his transcendental aesthetics Kant deals with space and time as forms of perception (Anschauungsformen), and hence as a priori schemata that are conditional of all experience and
knowledge. It must strike us to what extent Kants accounts of space and time are
similareven in the way both are phrased: sometimes he simply copies verbatim
what was said on space to what he tells us about time.1 Especially suggestive is
Kants chiastic comparison of space and time: different times do not coincide
with each other, but follow each other successively, just as different spaces do
not follow each other successively, but do coincide with each other.2 Again, the
implication is that what is typical of space is best expressed in terms of its contrast
with timeand vice versa.
Kants transcendental aesthetics is instructive for one more reason. Kants account of the nature of space and time comes close to that of Leibniz, in fact so
close that Kant feels compelled to invest a lot of time and energy in demarcating
his own position from that of his illustrious predecessor. As Kant himself insists,
the big difference between his own account and Leibnizs is that, whereas Leibniz
completely disconnects the ideality of space and time from sensory perception, he
presents both as the schemata conditional of all sensory experience and knowledge.3 Obviously, Kants argument here for the ideality of space and time might
invite us to conclude that the past is, in the end, a mental construction. For does
the ideality of time not compel us to embrace the ideality of the past as well? Of
course, not in the sense that the past is a reality existing only in our imagination
(as Leibniz might perhaps choose to see itthough he might wish to qualify his
claim by saying that the past certainly belongs to the category of the phenomena
bene fundata), but rather in the transcendental sense of being a category that we
project onto the world such that it is not part of the world an sich.
But putting things this way confronts us with an unpleasant and worrying dilemma from the perspective of historical writing. For does the past, as investigated by the historian, then belong to phenomenological or to noumenal reality?
If the former, the historian investigates basically the transcendental schemata we
project onto the world. But it is not easy to tally this with what we know historical writing to be like. If the latter, all historical writing could be accused of being
guilty of the transcendental abuse of reason. The sciences can avoid this dilemma
1. Especially in the two sections on the metaphysical clarification of space and time. See I. Kant,
Kritik der reinen Vernunft (Hamburg: Meiner Verlag, 1956), B 38, 39 and B 46, 47.
2. Ibid., B 47 (my translation).
3. F. C. Beiser, German Idealism: The Struggle against Subjectivism (Cambridge, MA: Harvard
University Press, 2002), 80.

138

frank ankersmit

since the transcendental conditions of their possibility do not invade their subject
matter, as is the case with history.4 For in history no content or meaning that one
might wish to give to the notion of the noumenal past could possibly prevent it
from being a potential object of historical research. To put it succinctly, the notion
of the history of transcendentalism or of epistemology is not a contradictio in
terminisas Gadamer was at pains to point out. History invades and digests epistemological barriers as easily as anything else. Epistemology can be historicized
(think of Foucault, Latour, Rorty, or social constructivism), but history cannot be
epistemologized. In sum, transcendentalism, whatever one may think of its merits, surely is an option for the sciences, but not for history.
II

But the immense metaphorical power of the notion of distancequite rightly recognized as such by the organizers of this conferencearguably has had more
unwitting victims than our concept of the past. For epistemology itself, as it has
developed since Descartes, is also quite hard to conceptualize without the notion
of distance. Epistemologists discuss the relationship between a knowing subject
on the one hand and a known object on the otherand then the perennial epistemological problem is how to bridge in some way or other, yes, the distance
between the two of them. Of course, one might now protest that Im projecting
here a primitive metaphor onto a philosophical subdiscipline of unparalleled technical refinement where one never, or only in some particularly sad cases, allows
oneself to be guided by the kind of associations we happen to have with the notion
of distance.
I wonder, however, whether this protest does sufficient justice to the facts. Simply recall what goes on in your mind when reading and commenting on the work
of the great epistemologists of the past and present. Is the situation not that, when
you ponder what they say, there is always some spatial model in your mind and
onto which you project their argument? Just try to think away this spatial model,
with its quasi-spatial distance between subject and object, between language
and the world, and observe what is left. Your mind will then irrevocably turn into
a tangle of incoherent ideas in which all sense of directionone more spatial
metaphor, by the wayis lost. Just as epistemology presents us with a scheme for
4. Just think of the impossibility of differentiating, la Kant, between a phenomenological and
noumenal conception of the past. What is one here, and what the other? We lack any believable criterion that might help us to distinguish between the two. It is as if history inexorably places us in the
realm of the Spinozist One Substance comprising both subject and objectand it might be added that
there is no better way of expressing this than Vicos verum et factum convertuntur. It might be argued
that hermeneutics has always been an attempt to find the juste milieu between these two equally unattractive Kantian alternatives. A still better way of dealing with the problem would be to simply drop
time (and space) as potential candidates for a transcendentalist account of historical writing, and to
replace them by the notion of narrative. That notion undoubtedly is far more appropriate for an understanding of the nature of historical writing than those of space and time. For an example of what such
an approach in fact might look like, see H. M. Baumgartner, Kontinuitt und Geschichte: Zur Kritik
and Metakritik der historischen Vernunft (Frankfurt am Main: Suhrkamp, 1972), chapter 6. But the
very best solution would be to retain the gist of Baumgartners argument while refraining from formulating it in transcendentalist language. This is how one might describe Louis Minks position; see
L. O. Mink, Historical Understanding (Ithaca, NY, and London: Cornell University Press, 1987).

THE TRANSFIGURATION OF DISTANCE INTO FUNCTION

139

conceiving of the relationship between word and world, so only spatial metaphor
presents us with the scheme in terms of which epistemological thought can adequately articulate itself. It is the precondition for epistemology.
Or think of the situation in which you have to compare different epistemologies
with each other. For example, you have to explain to your students Foucaults The
Order of Things. Spatial metaphors are then hard to avoid. You will then say that
in the Renaissance episteme the two planes of language and the world are still
entangled with each other, so that words may give us access to a things nature no
less than the thing itselfand you will tell your students that this is why in the
sixteenth century, etymology and text-commentaries could so strangely fascinate
the scholars of that time. For if language is seen as just one more thing in the
world, it may be just as informative about what the world is like as anything else
that is part of the world. Obviously, were not talking here about what is stated
expressis verbis about the world in language, but about language as being no less
a thesaurus of fascinating and unexpected discoveries about the world than nature
as investigated by the physicist.5 You would then presumably go on to say that in
the classical episteme the plane of language emancipated itself from reality, with
the result that the plane of language and that of the world are then presented as
two parallel planes, allowing you to move from the one to the other, just as you
may argue from some place on a map to a certain place on the globes surface
and vice versa. Whereas, finally, in the modern episteme beginning in the early
1800s both parallel planes begin to shift with regard to each other, so that any
correspondence between the two of them will have to be dated historically. This
inaugurated the age of the human sciences, and of man, as Foucault notoriously
claimed, where asking epistemological questions inexorably reduces us to history.
When expounding this shift from epistemology to history Foucault quite openly
embraces spatial metaphors himself:
[T]he domain of the modern episteme should be represented rather as a volume of space
open in three dimensions. In one of these dimensions we would situate the mathematical
and physical sciences. . . . In the second dimension there would be the sciences (such as
those of language, life and the production and distribution of wealth) that proceed by relating discontinuous but analogous elements in such a way that they are then able to establish
causal and structural constants between them. The third dimension would be that of philosophical reflection, which develops as a thought of the Same.6

So the idea is that we should situate the human sciences in what Foucault characterizes, with one more spatial metaphor, as an epistemological trihedron, and
that all the uncertainties peculiar to the human sciences arise from the impossibility of fixing once and for all the epistemological relationships between language
and the world in this trihedron. As Foucault puts it:

5. What Foucault describes as the Renaissance episteme has an eerie resemblance to how Rorty
conceived of the relationship between language and the world as defined by his so-called pure philosophy of language, in distinction to the classical episteme, which introduced impure philosophy
of language in Western philosophy. See also note 10.
6. M. Foucault, The Order of Things: An Archaeology of the Human Sciences (New York:
Pantheon Books, 1970), 347.

140

frank ankersmit

What explains the difficulty of the human sciences, their precariousness, their uncertainty as sciences, their dangerous familiarity with philosophy, their ill-defined reliance upon
other domains of knowledge, their perpetually secondary and derived character, and also
their claim to universality, is not, as is often stated, the extreme density of their object; it is
not the metaphysical status or the inerasable transcendence of this man they speak of, but
rather the complexity of the epistemological configuration in which they find themselves
placed, their constant relation to the three dimensions that give them their space.7

Hence, it is spatial metaphors we shall have to rely upon when accounting, with
Foucault, for the great epistemological sea-changes of the last five hundred years,
culminating in the emergence of the human sciences and how these, as exemplified by psychoanalysis, ethnology, and history, result in the demise of the regime
of epistemology. So both the triumphs of epistemology and its nemesis in the
historicism of the human sciences can only understandably be accounted for if we
tell our story in terms of spatial metaphors.
This is worrying for epistemology. Admittedly, metaphors may deepen our insight by explaining one thing in terms of another, for example by saying that
the heart is a pump or that our present economic system is no less threatened by
bubbles than it was before the credit crisis. In these examples what we ordinarily
associated with pumps or bubbles is projected onto hearts and economic systems
in order to achieve a better understanding of them. And as many theorists since
Max Black and Mary Hesse have insisted, this may be very illuminating, no less
in the sciences than in ordinary life.8 But the omnipresence of spatial metaphors
in epistemological thought is not to be considered with such equanimity, especially since epistemologists themselves never openly and candidly explain what
their argument owes to spatial metaphors. Indeed, spatial metaphors can safely
be assumed to function like the double bottom of the boxes that conjurers use for
their tricks, thereby rendering the cogency of the arguments of epistemologists to
considerations tacitly smuggled into them, considerations that have not been subjected to the most rigorous and painstaking philosophical scrutiny. Debate among
epistemologists then feeds on assumptions that were never explicitly stated. The
fact that spatial metaphor seems to be ubiquitous in epistemological argument,
though its presence is never actually signaled, is sufficient reason for alarm about
epistemological thought as it has developed in Western philosophy since the days
of Descartes, Locke, and Kant down to the present.
More specifically, dependence on spatial metaphor may explain why epistemology even in the heyday of logical positivism resisted all attempts at formalization:
metaphors mix two different vocabularies, and no formalization can cope with
that. Their incommensurability will effectively ruin any effort at formalization.9
One could read this as one more argument in favor of Rortys condemnation of
7. Ibid., 348.
8. See, for example, Knowledge and Metaphor. Volume III: Metaphor and Knowledge, ed. F. R.
Ankersmit and J. J. A. Mooij (Dordrecht and Boston: Kluwer Academic Publishers, 1993).
9. And insofar as historical writing can be described as using one vocabulary (that of the present)
to account for another (that of the past), it follows that all attempts to formalize historical knowledge
in one way or another are doomed to fail. It does not follow, obviously, that there should be no room
for formalization in theoretical reflection on the nature of historical knowledge.

THE TRANSFIGURATION OF DISTANCE INTO FUNCTION

141

epistemology as impure philosophy of language.10 Epistemology has a chance


of success only on the condition that it is radically cleaned of all spatial models.
I shall return to the issue of formalization and its relation to epistemology at the
end of this essay.
III

Now, truth has traditionally been the main topic of research in epistemology. Indeed, the spatial metaphor of the distance between subject and object, between
language and the world, has always invited the question of how this distance
could be bridged by truth. Just think of the correspondence theory of truth and
the more than two thousand years of philosophical debate occasioned by it. Or of
epistemological research on how the notion of reference ties together world and
language in much the same way as you might nail together two pieces of wood.
This should provoke our interest in Donald Davidsons philosophy of language.
For Davidson presents us with an account of truth that apparently leaves no room
for spatial metaphors such as that of distance between subject and object, or
between language and the world. At the same time, Davidsons thought developed
more or less naturally from the still predominantly epistemological concerns of
his former teacher W. V. O. Quine. This may give rise to the question of whether
we may credit Davidson with having developed a philosophy of language that is
happily free of spatial metaphors, and whether it may guide us toward a pure
epistemology la Rorty ready for its desired formalist take-off.
When accounting for the features of Davidsons philosophy of language relevant in the present context, I had best begin with the question of meaninghence
with the question of what it means to say s means that p (a phrase, by the way,
that is typically inhospitable to spatial metaphor: it is simply impossible to project
any spatial metaphor onto s means that p).11 The phrase s means that p is suggestive of substituting s with pand what spatial metaphor could adequately express what we associate with substitution? Substitution is basically anti-, or rather,
a-spatial. So that looks promising. Now, the main step in Davidsons philosophy
of language is to conceive of meaning in terms of truth, as defined in Tarskis semantic theory of truth. For Tarski the truth-predicate typically appears in contexts
in which we use a sentence in order to express the truth of some other sentence.
The sentence predicating truth is then in a meta-language that has an object-language as its object, an object-language that includes the sentence to which truth is
being predicated. In agreement with this picture, Tarski argues that the predicate
is true can be defined for each individual sentence s in an object-language by
10. For Rorty, impure philosophy of language originates from our tendency to juxtapose language
and the world as the battle-lines of two hostile armies before actual fighting begins. R. Rorty, Philosophy
and the Mirror of Nature (Oxford: Basil Blackwell, 1980), 257ff. In F. R. Ankersmit, Rorty and
History, New Literary History 39 (Winter 2008), 79-101, I made a first attempt to relate philosophy of
history to the pure philosophy of language that Rorty opposes to the impure variant.
11. Meaning can be described as legitimating semantic substitution: s means that p can be read as
s can be substituted with p without loss of meaning. Meaning can therefore be said to be the domain
of application of the substitution theory of representation that will be discussed in the next section. In
both cases there is no room for spatial metaphor.

142

frank ankersmit

providing a matching sentence p in the meta-language. For example, the sentence


in the German object-language Schnee ist weiss is true if and only if snow is
white (where snow is white is in the English meta-language). Such sentences,
hence sentences of the form s is true if and only if p are called truth theories,
since they formulate empirically verifiable theories about the correct use of the
truth-predicate in the object-language. Obviously, one would need, in principle,
such a truth theory for each sentence in the object-language.
We will all agree that Schnee ist weiss means the same as snow is white.
It follows that in Tarskis theory of truth, questions of truth and meaning begin to
shade off into each othersuch that one can understand truth in terms of meaning
(as Tarski preferred to do) or, inversely, explain meaning in terms of truth (which
is Davidsons preferred strategy). Davidson explains meaning in terms of truth
by replacing in the statement s means that p the component means that by
Tarskis is true if and only if. The result is that s means that p becomes s is
true if and only if p, so that we now have a satisfactory theory of truth for the
sentences in an object-language, which, moreover, will enable us to explain meaning in terms of truth. Finally, all this became possible without ever referring to
or relying on ontological or epistemological argument and the spatial metaphors
always haunting it. So here we have a theory of truth where truth bypasses the
traditional confrontation model of language and the world, and that seems to
avoidI explicitly say here seems to avoid, for we shall see in a moment that,
in fact, things are differentthe perennial temptation to measure the distance between the two of them in terms of reference or of epistemological categories such
as subject and object. The demons of spatial metaphor would then successfully
have been exorcized.
However, this claim may turn out to have been premature. When commenting on Davidsons appeal to Tarskis theory of truth, Bjorn Ramberg writes: [I]
ts promise lies in the fact that whatever the logical and ontological resources the
theory uses, it produces theorems which state the truth-conditions for the sentences of the object-language by giving the translations of these sentences in the
meta-language. This means that no resources available in the object-language are
needed to see the theorems are true, as long as we take for granted the correctness
of the translation.12
It must strike us in this passage that, whereas we had believed we were talking about meaning and truth, it now becomes clear that we have been talking all
along about translation. Indeed, Davidson uses Tarskis truth theory to give us
a translation of sentences in an object-language, such as the (German) sentence
Schnee ist weiss, into a sentence in our own (English) meta-language, namely
snow is white. And this is no mere accidental feature of Davidsons application
of Tarski, but truly belongs to the essence of his argument, as will be clear if we
turn to what Davidson called radical interpretation. For in his essay entitled
Radical Interpretation Davidson explicitly argues that the operation of translation is structurally identical with his Tarskian definition of the truth-conditions of
12. B. R. Ramberg, Donald Davidsons Philosophy of Language (Oxford: Wiley-Blackwell,
1989), 53-54.

THE TRANSFIGURATION OF DISTANCE INTO FUNCTION

143

sentences in an object language that we discussed a moment ago.13 This need not
surprise us, if we recall that I began my account of Davidson a moment ago with
the statement s means that p before getting to Davidsons Tarskian truth-theory
for s. For what else is translation from one language into another but the generation of statements of the form s means that p?
So the fates of Davidsons theory of truth and that of his theory of radical interpretation are identical: the final truth about the one is also the final truth about the
other, and vice versa. This provides all the more reason to closely scrutinize his
theory of radical interpretation. Decisive in this context is Davidsons linguistic
holism, which he took over from Quine, and which several of his commentators
believe to be the Achilles heel of Davidsons philosophy of language. Davidson
is compelled to this linguistic holism since his theory of truth pretends to be a
theory for all of a language, that is, for a language as a whole. Now, as Quine had
already insisted in Word and Object, each body of evidence always allows for several competing theories, with the result that if all translation must always remain
merely tentative, it will always be based on a certain body of evidence and therefore be subject to correction by new evidence. So all we can strive for is an ever
better match of the meanings of sentences in the object-language and those in the
meta-languagewith the ever unattainable ideal of achieving a complete fusion
of horizons of the two languages, to use the most appropriate terminology here.
Now, as we all know, this notion of the fusion of horizons (Horizontverschmelzung) is a pivotal concept in Gadamers Heideggerian hermeneutics. We
will recall that it is a spatial metaphor that Gadamer relied upon to explain what
historical understanding (Verstehen) is all about, and that he discussed in a section of Truth and Method characteristically entitled the hermeneutic meaning of
temporal distance.14 The notion of the fusion of horizons, in terms of which
Gadamer wishes to capture what is at stake in translating past into present meanings, is, in its turn, a spatial metaphor, if ever there was one. It need not surprise
us, then, that it has become a thriving intellectual industry to compare Davidson
and Gadamer, and, next, that the spatial metaphors abounding in hermeneutic theory from Schleiermacher to Gadamer can now also be projected onto Davidsons
radical interpretation. In this way Davidsons theory of truth can, contrary to our
initial intuitions, be said to tacitly depend on and employ all the spatial metaphors
functioning as the skeleton of hermeneutics and (impure) epistemology.
This claim finds extra support in Davidsons notion of the so-called Charity
Principle. The principle has its origins in Quines radical translation, which, as
the name suggests, is the ancestor of Davidsons own radical interpretation. In
Quines radical translation, words are translated from one language into another
in a context in which we have nothing else to go on than how the world presents
itself to us. For example, if the speaker of a foreign language always uses the
word Gavagai when a rabbit is around, we may infer from this that Gavagai
means rabbitbut on the condition that the speaker of the foreign language
13. D. Davidson, Radical Translation, in idem, Inquiries into Truth and Interpretation (Oxford:
Clarendon Press, 1985), 130, 131.
14. My italics. H. G. Gadamer, Wahrheit und Methode: Grundzge einer philosophischen
Hermeneutik (Tbingen: Mohr, 1972), 275-290. See especially 289, 290.

144

frank ankersmit

is not trying to systematically fool us and will generally speak the truth in his
language. The assumption that this condition is met is what Davidson calls the
Charity Principle, which he defines as the principle assigning truth conditions
to alien sentences that make native speakers right when plausibly possible, according, of course, to our own view of what is right.15 However, speaking the
truth as understood in the Charity Principle is not Tarskian truth, but truth as
defined by the correspondence theoryhence the matching of word and fact (a
rabbit manifests itself under that tree over there, and we utter the word rabbit or
Gavagai). Self-evidently, this gets us back to the epistemologists confrontation
model of language and the world, where truth is expected to bridge the distance
between the two. Indeed, the ground has then been prepared for a return of all
spatial metaphors in the philosophy of language.
Speaking more generally, in their accounts of radical translation and radical
interpretation Quine and Davidson seem to make use of two conceptions of truth
instead of just one. On the one hand, there is the conception of truth they explicitly
discuss when dealing with the question of how their holism affects the relationship between word and world, but on the other their argument also presupposes
the embrace of a fairly trivial and commonsensical variant of the correspondence
theory of truth.16 One may well wonder what will be left of their theory of truth in
the former sense if it is radically purged of its less illustrious rival.
IV

This sends us back to our initial question of whether we can conceive of a philosophy of language that avoids an implicit or explicit reliance on spatial metaphors. I
shall now argue that a philosophy of language that sees it as giving us representations of the world satisfies this condition. Representation is an imprecise term, so
I should clarify how I shall understand it here. The best way to do so is to distinguish
between two theories of representation: the resemblance theory and the substitution
theory. According to the resemblance theory, a representation should resemble what
it representsmeaning that there should be certain rules or algorithms allowing us
to move safely from represented reality to its representation and vice versa. The resemblance theory has an obvious elective affinity with epistemological thoughtit
is based on the assumption of a clear distinction between language and its objects,
and proposes a way to bring them back togetherwhich is why we may expect it to
be just as hospitable to spatial metaphor as epistemology itself.
In spite of its immediate appeal, the resemblance theory is riddled with thorny
problems, and this is why most (though not all) people nowadays prefer the alternative, the substitution theory of representation. The theory was first expressed
by Edmund Burke in his essay on the sublime and the beautiful, where he pointed
out that words may represent things not because words resemble thingsfor they
do notbut because they are linguistic substitutes for things (hence, the theorys
15. Davidson, Radical Translation, 137.
16. It is of interest, in this context, that in the definition of the Charity Principle mentioned a
moment ago, Davidson uses the notions of truth and of what is rightwhere the latter stands for
this trivial variant of the correspondence theory of truth that remains unanalyzed in his argument.

THE TRANSFIGURATION OF DISTANCE INTO FUNCTION

145

name).17 But it was Ernst Gombrich who gave us the canonical definition of the
substitution theory.18 Going back to the images we associate with the birth of art,
Gombrich writes:
In many case these images represent in the sense of substitution. The clay horse or
servant buried in the tomb of the mighty takes the place of the living. The idol takes the
place of the God. The question whether it represents the external form of the particular
divinity . . . does not come in at all. The idol serves as the substitute of the God in worship
and ritualit is a man-made God in precisely the sense that the hobby horse is a man-made
horse. . . . Can our substitute take us further? Perhaps if we consider how it could become
a substitute. The first hobby horse . . . was probably no image at all. Just a stick which
qualified as a horse because one could ride on it. The tertium comparationis, the common
factor, was function rather than form.19

This gives us all we need. To begin with, as Gombrich insists, the relationship
between the represented and its representation cannot be defined or explained in
terms of schemes or frameworks encompassing both the represented world and
its representations and enabling us to define their relationship. For that would
reduce us again to the resemblance theory, which he explicitly rejects. There is
no planeanother spatial metaphor!embracing the represented reality and its
representation as their tertium comparationis that both have in common so that
we can map their relations in terms of that planes coordinates. However, there is
a tertium comparationisbut, as Gombrich insists, this tertium comparationis is
not a matter of form but of function. So representation is basically functionalist,
and decisive is whether a representation can function as what it represents. This,
then, gets us out of the magical circle of spatial metaphors, as I shall argue.
We need only think, in this context, of the prototypical function, that in mathematics. Take the equation: f(x) = x2. Admittedly, we can draw on paper a diagram of that functionand then we do have a figure in space. We might then
feel tempted to say that the diagram is a representation of the functionwhich
would reduce representationalist functionalism to spatial metaphors again. But
we should resist this temptation. The function defines a rule for coupling a certain
numerical value to a certain value of x. This is what it does, and this is all it does.
It does not actually state what number corresponds to a certain value of x, as is the
case in the diagram. The function gives the rule, and the diagram the application
of the rule. Observe, furthermore, that there is no room for truth in the equation
itself. It is a rule, and the truth-predicate does not apply to rules. But it does to the
17. But descriptive poetry operates chiefly by substitution; by the means of sounds, which by
custom have the effect of realities. Nothing is an imitation further than as it resembles some other
thing; and words have no sort of resemblance to the ideas for which they stand; see E. Burke, A
Philosophical Enquiry into the Origin of our Ideas of the Sublime and the Beautiful [1757] (Oxford:
Oxford University Press, 1990), 157. But perhaps one could even go back as far Herodotus and
Thucydides. They distinguished between autopsythe perception of something with ones own
eyesand historia, which gives us an account of what we have seen. They then argue that the latter
should have just as much presence, evidence, or energeia as the formerwhich clearly anticipates
Burkes formulation. See F. Hartog, vidence de lhistoire (Paris: Editions de lEcole des hautes
tudes en sciences sociales, 2005), 11-21, 169-191.
18. It is worthwhile to remind the reader here that we found in section III that substitution is
essentially a-spatial.
19. E. H. Gombrich, Meditation on a Hobby Horse, in Aesthetics Today. rev. ed., ed. E.
Philipson and P. J. Gudel (New York: Meridian, 1980), 175 (my italics).

146

frank ankersmit

diagram. For the diagram truthfully states that the values of 3, 4, and 5 for x will
give us the numbers 9, 16, and 25 (or falsely, if other numbers are given). It is
much the same with (historical) representation. For one might say that a historical representation proposes20 a rule for how to conceive of part of (past) reality.
That is its (functionalist) meaning. And, indeed, applying the rule may then give
us a mental image or picture of the past, just like the diagram of a function. But
the representation itself does not do this; again, it only proposes the rule. Nor are
representations true or false,21 as I have argued on many occasions, though their
application may give rise to questions of truth and falsitythough not necessarily
so.22 Finally, the functionalism of historical representation enables us to decide the
old question of whether historians describe or explain the past: they do neither.
The foregoing also sheds some more light on the shortcomings of the resemblance theory of representation. Correct is the insight that there should be a rule
for connecting the represented world and its representation. But wrong in the resemblance theory is conceiving these rules as telling representers in general, and
historians in particular, how to move from represented reality to its representation.
Instead, the rules require us to move in the opposite direction, and hence from
representation to how we are required to imagine part of the past. They tell us what
to project onto the past. And since the past can only passively undergo such projections without protest, these rules can never be taken in an epistemological sense.23
Since metaphor is an operator knowing of few restrictionsalmost anything
can be seen in terms of anything elsewe could say that, though the substitutional account gets us out of the magical circle of spatial metaphors, the functions
diagram, and the pictures or images of the past that history provides, are spatial
metaphors of the function or, as the case may be, of a historical representation.
This is a useful observation, for it correctly reflects the hierarchical relationship
between representationalist functionalism and spatial metaphors by making admirably clear that the latter can never express relevant information about the former
that is not yet present in the former. This, then, is why representationalist functionalism can never fall into the epistemologists trap of inferring claims from the
spatial metaphors we associate with epistemological argument, claims that have
no support in that argument itself. Trying to do so would be repeating the error of
the resemblance theory.

20. I deliberately use here the word propose and avoid saying that representations are rules;
representations do not say what the world is like; instead they invite us to conceive of it in a certain
way. This is one more reason why epistemology is unable to account for representation.
21. Even though they consist for the better part of statements about the past that are either true
or false.
22. In the last section of this essay I shall argue that truth arises from representationalist meaning.
But this transition from meaning to truth is possible only if the web of our representations of the world
is so dense and so closely knitted together that it will permit formalization. This is typically true of our
representations of daily realityand, indeed, then there is room for truth and falsitybut it will rarely
be the case with historical representations where overlap will at best be only very partial. Moreover,
historical representations are ordinarily written with the explicit intention of avoiding overlap. For
only this will attract the interest of other historians; overlap does not.
23. I emphasize that this should certainly not be taken to imply that the rules in question are wholly
arbitrary: though many hats will fit your head, some of them will fit better than others.

THE TRANSFIGURATION OF DISTANCE INTO FUNCTION

147

Finally, I wish to address the admittedly somewhat egotistical question of how the
foregoing argument fits into my own intellectual development. But I may perhaps
be forgiven for addressing it since it may clarify some of my previous arguments.
In my 1983 book on narrative logic I developed a theory of the historical text or
representation (the terminology I adopted in the 1990s) from which I have never
deviated. That theory is, basically, the same as the one that I presented here as
the substitution theory of representation, according to which there exist no fixed
rules, algorithms, or epistemological schemata tying a representation to what it
represents. This is why I warmly welcomed Rortys notorious attack on epistemology in his Philosophy and the Mirror of Naturewhich I read shortly after
the publication of Narrative Logic.
I had my doubts, nevertheless. In fact, I was, and still am, a little less radical
than Rorty. Rorty never discusses texts, whereas I have always felt the need to
strictly distinguish between how singular true statements and texts relate to the
world (when speaking of texts, I primarily have in mind historical texts or representations). It is an illusion to believe that the singular true statement gives all one
needs for explaining how texts relate to the world. For even though the historical
text consists of singular true statements, its logical form cannot be reduced to, or
modelled on, the singular true statement. So when discussing epistemology we
must make sure to be aware of whether we are talking about statements or about
texts as exemplified by the historical text or representation.
Precisely this is why philosophy of history is of so much interest for philosophy
of language and where it may substantially add to contemporary philosophy of
language. The significance of philosophy of history exceeds by far that of being a
theoretical account of the discipline of historiography: it fills an intolerable lacuna
in philosophy of language insofar as the latter has somehow succeeded in ignoring, from the days of Frege up till the present day, the problem of how we may account for how complex texts represent a complex reality. This problem is, indeed,
best exemplified by historical representation, but it is really a substantial, if not
the major, part of all of our use of language. Philosophy of language will remain
a mere torso as long as it goes on stubbornly ignoring the problem of the text. So
whereas the sciences have been for one and half centuries the main compass for
philosophy of language, the latter should now be ready to learn from the writing
of history as well. This is the idea that has inspired all of my writings and to which
I shall adhere until the moment that the pen drops from my fingers.
Or, to rephrase my doubts about Rorty in the terminology adopted in this paper,
whereas I agree with Rortys rejection of epistemology from the perspective of
(historical) representation, I am less skeptical of epistemology when we are dealing with singular true statements. For example, though I am fascinated by Davidsons account of the truth of singular statements, I believe that its anti-epistemological implications derive from its weaknesses rather than from its strengths.

148

frank ankersmit

Let me put it this way. Davidsons holism is an important ingredient of what one
might see, with Rorty, as the anti-epistemological implications of his philosophy
of language. He took holism over from Quine, who had argued that scientific
statements face the tribunal of sense experience not individually but collectively.
But if one decides to agree here with Quine, it does not automatically follow that
such holism also holds for natural languages. Rather the reverse, for the statements made in natural languages can only rarely be pulled together within the
kind of holist webs of scientific statements that Quine had in mind. Moreover, if
one removes this dimension of holism from the statements made about the world
in natural languages, they seem to face the tribunal of sense experience individually rather than collectively. Historical writing is illustrative here: there is no
web tying together a statement about, say, the reign of Septimius Severus to one
on Schillers aestheticsunless one embraces some speculative philosophy of
history24 that provides you with a common ground for these two statements about
the past.25 In the context of singular statements the questions that were typically
asked by the epistemologist make sense again.26 As a result, the less coherence
you have on the level of language (and history certainly has less coherence on that
level than science) the stronger the case for epistemology will be.
But as far as (historical) representation is concerned, I am an anti-epistemologist. I am so for roughly for the same reasons that Nelson Goodman rejected
the resemblance theory of representation (though I hasten to add that there is a
good deal of ambivalence in Goodmans criticism of it in his Languages of Art
and in his later writings, owing to his unfortunate metaphor of the language of
art. For that metaphor invites us to see language as the most appropriate model
for an understanding of artand of representation and the text, and then all the
epistemological specters driven out the front door reappear at the back door.27).
This raises the question of whether my position reflects an inconsistent attitude
toward epistemology. How can one denywith meto epistemology its raison
dtre in one case, while acknowledging it in another?
The main idea in replying to this question is the singular obtrusiveness of the
notion of the I, whether as a transcendental, empirical, or simply as the knowing
subject. Fichte was right: it all begins with the I positing a world. But why is
24. The historians so characteristically unreliable counterpart to the scientists scientific theories.
25. This may also explain why historians feel a natural affinity with an empiricist account of their
discipline, and why this account is, in fact, less naive than is often argued to be the case.
26. It is, therefore, most paradoxical that the linguistic turn has been welcomed so warmly in
the reflection on the humanities because of its holism. For the holism that Quine had in mind has no
counterpart in historical writing. It might be objected to this that 1) historical texts or representations
mutually define each other, as is suggested by so-called inter-textualism, going back to Saussures
theory of the sign, and 2) that this implies holism as well. The objection is to the point. However,
holism in history and the humanities is a holism of meaning and not of truth (as is the case with
Quines and Davidsons holism). So, indeed, there undoubtedly is some truth in the current popular
argument that Quines and Davidsons holism builds a bridge between the sciences and the humanities. But this is a most treacherous bridge, for in spite of its air of solidity, one will fall through it as
soon as one steps upon it.
27. Illustrative is Goodmans effort to break down narrative in artfor example, the story of
Psyche as told in the painting by Jacopo del Sellaio in the Museum of Fine Arts in Bostoninto a
chronologically ordered set of statements about some (imaginary) history. The place where narrative
and representation do not accord with the model of the statement is then eliminated. See N. Goodman,
Of Mind and Other Matters (Cambridge, MA: Harvard University Press, 1984), 11ff.

THE TRANSFIGURATION OF DISTANCE INTO FUNCTION

149

this so: why does every philosophy begin by embracing this insolent and imperialist I? Certainly, the notion of the I has its genealogy as well; it must have
emerged somehow from some primeval situation in which there was not yet an
I. It was not there at the creation of the world. But this is where representation
comes in, for representation does not need an I. For example, a painting is a representationbut where is the I of the painting? Representations can exist without Is, just as things can. There may still be representations after humanity has
ceased to exist. The case is different with knowledge: you cannot have knowledge
without someone having it. Knowledge dies with the last man, whereas representationsfor example, paintings and history booksmay well survive him.
Representation precedes the I, as is clear from the quasi-Heideggerian observation that one should not say I have my representations but rather I am
my representations, where only the latter can do justice to how our conception
of selfhood emerges from our representations. Self-evidently, this emergence of
the notion of selfhood always goes together with the production of a not-self. So
in this way the unity of representation falls apart in the duality of the I on the
one hand and the world on the other. Representation, not knowledge, gives the
deathblow to solipsism. Recurring patterns in representations guide this process
of pulling apart representations into a knowing self and a known objectthus
paving the way for epistemological reflection. Representationalist meaning then
bifurcates into truth (the subject) and what truth is true of (the world)and only
then does epistemology become possible.
So the main error of epistemology has always been to start with truth instead of
with representationalist meaning. When this happens, spatial metaphors are then
inevitable. If, however, epistemology derives truth from representationalist meaning, spatial metaphors can be avoided. Of course, then epistemology will have
to pay the price of recognizing its dependence on representation: no representation, then no epistemology either. This is, again, why history is so important to
philosophy (of language). History is the discipline that still reminds us of the fact
that representation and meaning, not truth, refer us to the ultimate origins of our
cognitive conquest of the world; truth merely registers or codifies recurrent patterns in representationalist meaning. There is no knowledge and no truth without
first having representation and meaning. The more technical details of this evolution from representation to truth I presented already in my book on narrative logic,
now some thirty years ago.28 As may be clear from this, I have never changed my
mind about this issue since then. Whether this is proof of an admirable constancy
or of an intolerable self-conceit is for others to decide.
University of Groningen

28. F. R. Ankersmit, Narrative Logic (Dordrecht and Boston: Martinus Nijhoff, 1983), 155-169. I
shall add a new dimension to the story told there in my forthcoming book The Semantics of Historical
Representation (Ithaca, NY, and London: Cornell University Press, 2012). The books main thesis
is, indeed, that whereas in most of contemporary philosophy of language one moves from Truth to
Meaning, in (historical) representation Truth can be derived from Meaning.

You might also like