Why think about what translation means today?
There are at least three versions of translation that might be relevant. First,
more or less straightforward or utilitarian translation from language to
language, for instance literary translation from French to English to publish a
new novel. Most of the traditional issues surrounding the difficulty of
translation—denotative versus connotative meaning, style, and so on—might be encountered
in this category. This type of translation has of course been going on for
thousands of years. Second, an issue particularly relevant to the digital
humanities, is translation into computer code of literary corpi as this material
is digitized and made searchable. Organizations such as the Text Encoding
Initiative maintain standards for such translation, which is in many respects a
type of conversion between formats. Lastly, and the type of translation I would
like to discuss here, is a form that is much closer to the contemporary essence
of the digital itself than coding: artificial neural networks. Here we have not
only the transference of meaning and information from one language or format to
the next, but a machine translation with a high degree of accuracy and the potential
to apply the translation process across different contexts. This latter capacity
is known as "artificial general intelligence." By considering this
mode of translation through the lens of what Walter Benjamin characterized as a
posthuman translatability, the supplementarity of pure language, and the
symbolizing agency of language, we can reference an expanded dimensionality of
the digital.
Artificial neural networks have been
through different phases of development but effectively started in the 1940s as
an attempt to mimic the actual biological neural networks being discovered in
the human brain. Computer scientists had conceptualized how a digital version
of the biological networks might function at that point but technological advances,
primarily computer memory capacity, have only now begun to catch up with the core
of the earlier insights. Essentially, neural networks are grounded in the idea
of, rather than setting up logical rules by which a computer operates, importing
large amounts of data from which the computer then recognizes patterns. The
recognition takes place through a trial and error process where connections
between data are tested until a particular output is reached. This process is
often called "machine learning" when a single neural network layer or
pattern is involved, and then "deep learning" when
additional layers are added in, the computer working to recognize patterns
within patterns. The complexity that emerges from the operation of artificial
neural network systems is profoundly different from what we have traditionally
known as the potential of the digital computer.
A December 2016 article in the New York Times Magazine, "The Great A.I.
Awakening," quotes the CEO of Google, Sundar Pichai, as saying
that the future of Google is "AI first," meaning that the company
will be prioritizing machine or deep learning artificial intelligence (AI). The
unit within Google that is charged with this initiative is Google Brain, a group of leaders in
the artificial intelligence community that has been developing neural network technology.
As the article relates, the first test case for its progress with AI (Facebook
and Amazon, among others, are also developing it) was Google Translate. The
article describes how the Google Brain team implemented AI software within
Google Translate to produce translations of complex literary texts that had
effectively no identifying mark that might indicate that the translations were
computer generated. The example provided is the opening paragraph of the Ernest
Hemingway story "The Snows of Kilimanjaro." Even though I'm simply
paraphrasing the New York Times
article, I want to reproduce this example here, since it provides a very clear
impression of the look and feel of neural network outputs. What follows is (1)
Hemingway's original, (2) the translation by the pre-neural network Google
Translate, and (3) the updated, neural-network version from Google Translate
(both of these latter are translations back into English from a Japanese
translation of the original):
1.
Kilimanjaro is a snow-covered mountain
19,710 feet high, and is said to be the highest mountain in Africa. Its western
summit is called the Masai “Ngaje Ngai,” the House of God. Close to the western
summit there is the dried and frozen carcass of a leopard. No one has explained
what the leopard was seeking at that altitude.
2.
Kilimanjaro is 19,710 feet of the
mountain covered with snow, and it is said that the highest mountain in Africa.
Top of the west, “Ngaje Ngai” in the Maasai language, has been referred to as
the house of God. The top close to the west, there is a dry, frozen carcass of
a leopard. Whether the leopard had what the demand at that altitude, there is
no that nobody explained.
3.
Kilimanjaro is a mountain of 19,710 feet
covered with snow and is said to be the highest mountain in Africa. The summit
of the west is called “Ngaje Ngai” in Masai, the house of God. Near the top of
the west there is a dry and frozen dead body of leopard. No one has ever
explained what leopard wanted at that altitude.
The point here is, first of all, that
Hemingway's original—what we have come to know as crisp and clean English diction
and syntax, his trademark, an American trademark—is not being altered beyond
recognition in #3, as it is in many ways in #2,. More importantly, however, the
quality of the neural-network translation possesses a strong readability, being absent of errors that
would inhibit our reading process. It brings another order of experience of
language with it. There is a quality of the language in this example that
accesses another register of our very experience with language, an unexpected
register of the human. Just to highlight the multiple conclusions that might be
drawn here: the quality of translations has increased; difficult translations
are handled effectively; the computer is able to duplicate the symbolic or
connotative implications of literary language; the computer is able to effect
translations that are not noticeably computer generated. Finally, in a
significant way, we see here Hemingway's—an excellent example in many ways—ability
to be a stand-in for American identity, and literature itself. The conclusions
we should entertain are quite clearly that not only has human translation
become a redundancy, but in some sense literature itself may be next, if a
computer can recognize and reproduce connotative language. The original,
Hemingway himself and whatever he might represent, is situated in a context of
the nonhuman, of the breakdown of distinctions between original and
translation, one that arises from the transformation of symbolic language.
I want to combine this look at artificial
neural networks with three little-discussed aspects of Walter Benjamin's 1921
essay "The Task of the Translator." Benjamin there raises the idea of
a translatability that exists in the world but that is beyond human perception.
How do we account for such potential if translatability is accessed digitally?
Also referenced in Benjamin's essay is a complexity that emerges as a product
of the supplementarity of all languages. Should we consider artificial neural
networks as that complexity itself? Finally, there is a turn referenced in the
essay, a turn that happens within language itself, from passive to active, from
being that which is symbolized to the symbolizing, a turn that parallels the
potentials being developed with multilayered neural networks. Investigating
Benjamin's text in these three areas can perhaps uncover aspects of the
contemporary digital context that repositions and expand our notion of
translation, its dimensionality and greater significance as a form (as Benjamin
calls it)[1]
in a world constituted by a feedback loop between human and artificial neural
networks.
Translatability
Benjamin constructs an idea of the
nonhuman through a consideration of what aspects of an original can possibly be
translated. He attributes then to each original a part of its essence that
lends itself to being translated, or an aspect of "translatability"
that is part of every original. Every origin, in order to be an origin, must
have within itself its own particular non-originality, which is that aspect of
an original that allows it, that calls for it, to be brought into the realm of
the communicable, of what is translated. Once Benjamin settles on this idea he
then posits a realm of originals whose translatability will never be accessed.
He sees no reason why something might never be translated but still have an
existent aspect of translatability, just as an original can exist without our
being aware of it. An original might not find anyone who recognizes its
translatability, who is able or desires to translate it. It may also be the
case that an original's translatability does not "call for" being
translated. Benjamin compares this situation to the sense in which something
may be said to be "unforgettable" but then is entirely forgotten by any
living person. It can't be said that designating something as unforgettable is
an error, just that the claim is "unfulfilled by men." His comment
then is: "Analogously, the translatability of linguistic creations ought
to be considered even if men should prove unable to translate them. Given a
strict concept of translation, would they not really [still] be translatable to
some degree?" He continues:
Even in times of narrowly prejudiced
thought, there was an inkling that life was not limited to organic
corporeality. . . . The concept of life is given its due only if everything
that has a history of its own, and is not merely the setting for history, is
credited with life. (255)
On this model, translatability[2]
exists beyond the realm of human cognition. We only experience those things
that are actually brought into translation. When we speak of artificial neural
network translation then, we are better able to access its potential by
considering this idea, since with AI translation we would have every
expectation of not only more translation taking place than would have before
(and Google Translate had some 500 million monthly users even before the
introduction of neural network technology), but translation of what would not
have otherwise been translated. The idea is that neural network technology is
an example of exceeding the universe as we know it by accessing this "unforgettable
forgotten" or translatability, bringing into communicative exchange
aspects of the world that might exist but that would not otherwise be
perceived.
Fair enough. This does not however deny
the existence of a translatability even neural networks cannot address; this
does not mean that originals are fully accessed by technology, or even accessed
at all. We are still dealing only with what is translated. "It is evident
that no translation, however good it may be, can have any significance as
regards the original." Benjamin here compares the original to life itself,
where any given manifestation of life, an individual, a species, and so on,
does not effect the overall quality or existence of life in general. And it is
with this qualification in mind that it is useful to do a second pass through
some of the language Benjamin uses to frame his argument for nonhuman translatability,
for he is constantly qualifying his descriptions, putting them in place as a
play of mental and material constructs, "linguistic" but only in an
expanded sense of that term. We need only return to the example of the
"unforgettable forgotten" to uncover this quality to Benjamin's
entire argument in "The Task of the Translator." To quote the passage
at more length:
One might, for example, speak of an
unforgettable life or moment even if all men had forgotten it. If the nature of
such a life or moment required that it be unforgotten, that predicate would
imply not a falsehood but merely a claim unfulfilled by men, and probably also
a reference to a realm in which it is
fulfilled: God's remembrance. Analogously, the translatability of linguistic
creations ought to be considered even if men should prove unable to translate
them. (254)[3]
What we can note here is the emphasis on
language, the way the nonhuman realm is posited as an effect of language. The nonhuman realm is built in to words like
"unforgettable," called out as a "predicate," since clearly
anything could be forgotten by all living persons. Language then seeks its own fulfillment,
which may in fact be the truly nonhuman, contained within language itself. And
that realm is here specifically associate with the Judeo-Christian
"God", a set of values and metaphysical assumptions that finally, as
here, does not include the human. As earlier in this same paragraph of
Benjamin's essay, logic itself is also called into question and associated with
the linguistic, our certainty that there is nonhuman translatability being
"apodictic," and there here "analogously" as an adverbial access
to the character of translatability. Benjamin's point here seems quite clearly
to be that translatability, and by extension any concept of an
"original," are ideas that come to us as part of the
"technological unconscious"[4]
of language itself. This concept of language contains the nonlinguistic.
The implication for artificial neural
networks, particularly for translation, and the way in which the digital is evolving
overall, is that what is technologically enabled, perceived as the introduction
of nonhuman capacity into the known world, is in fact already contained in a material/immaterial
binary carried within the structure of language itself. Translatability, and
original, are situated concepts that operate objectively. Any perceived
transcendence in the world, and the world may in fact entirely consist of the
perception of such transcendence, is reducible to what may almost be termed the
agency of language itself, the blueprints, the laws it hands down to humanity.
"The laws governing the translation lie within the original, contained in
the issue of its translatability." In a sense, language itself is our
original, containing the laws of how neural networks will recognize and
reproduce the world, its present, and hence its past, and hence its future. On
this logic, we re-evaluate technicity as not only a human construct, rather
than in any way nonhuman, but one that remains in every respect embedded in the
deepest humanistic traditions at its very root. We have to ask, is there a
recursivity or repetition in how we imagine any given technological solution or
utopia?
Supplementarity
Benjamin confronts the idea of what the
common translatable element between languages is by introducing the idea of
"kinship." In this sense, he might as well be attempting to define
the digital itself, as a global phenomena that by definition crosses all
languages and cultures. Is this kinship then what is most biologically human,
for instance the basic structure of the brain, the neural structure? We might
say that Google seems to think so and that the slogan "AI first" has
this tenet behind it. It seems that corporations are able to push the digital
to its furthest reaches most successfully through a mimesis of what is most
universally human, even as in this case that universal human element is only
able to be recognized, detailed, and discussed through advanced science. Technology
talks to itself,[5]
as it were, using the idea of the "human" as a mere reference point.
And Benjamin's text both reflects and
extends these tendencies in important ways. As his text works its way toward an
understanding of what this kinship might be, he rejects the idea that kinship
is found in "vague resemblance," and writes that "it cannot be
defined adequately by an identity of origin," though the "concept of
'origin' remains indispensable." Thus kinship, commonality, does not
substantively hold up as a result of either similarity or shared origin. It may
be broken in both these cases, mostly since they are both historically
oriented. Lasting kinship is rather found in the "suprahistorical"
aspect of languages where "in every one of them as a whole, one and the
same thing is meant." What is at stake is that the thing that is in fact
signified by different words is the same across languages and cultures. There
is a way in which different words for "bread" mean the same thing,
bread, across cultures. In this sense, languages intend the same thing, and for
Benjamin, globally, it is by conceptualizing the "totality of their intentions"
as in fact supplementing each other that the different languages combine this
element of their commonality to produce actual kinship, or what Benjamin calls
"pure language." "Whereas all individual elements of foreign
languages—words, sentences, associations—are mutually exclusive, these languages
supplement each other in their intentions."
The description of this supplementarity tracks
the description of pattern recognition in artificial neural networks. In the
trial and error at the base of AI, connections between data points are tested
and given different "weights" based on the number of accumulated relevant
responses.[6]
Over time patterns emerge and contribute to outputs with a high degree of
precision. What we would be theorizing here, through "The Task of the
Translator," is that pure language results from a supplementary
aggregation of intentions, an aggregation that parallels the accumulation of
"weights" and that enables accurate translation. In this way, pure
language serves as a cross-cultural semantic architecture that parallels in its
function and makeup artificial neural networks.
But so what? There are perhaps plenty of
network structures that might reflect Benjamin's pure language. But what is
important here about the "pure language" of Benjamin's essay is its
further development into a situatedness reminiscent of the originality/translatability
problematic discussed above. What Benjamin theorizes through the kinship of
languages that produces translation is the very presence of and relation to
difference itself. He accesses this register of critique through the seemingly
innocuous example of variant words for "bread," the French pain and German Brot:
. . . we must draw a distinction, in the
concept of "intention," between what is meant and the way of meaning
it. In the words Brot and pain, what is meant is the same, but the
way of meaning it is not. This difference in the way of meaning permits the
word Brot to mean something other to
a German[7]
than what the word pain means to a
Frenchman, so that these words are not interchangeable for them; in fact, they
strive to exclude each other. As to what is meant, however, the two words
signify the very same thing. Even though the way of meaning in these two words
is in such conflict, it supplements itself in each of the two languages from
which the words are derived; to be more specific, the way of meaning in them is
supplemented in its relation to what is meant.
We can note here the peculiar
"strive to exclude each other" and "such conflict" as
references to the World War I context out of which Europe had emerged only
three years before this essay was published, and that had defined much of
Benjamin's adult life to this point. It seems clear that the German/French
conflict is not coincidental in this passage, and that "kinship"
likely refers to the relation between these two nations. In this sense, the
astonishing destruction of the recent war is at stake in locating a stable idea
of kinship. The fact that the "ways of meaning" rule each other out,
that "these words are not interchangeable" both defines local
identity and serves as the loci of human destruction. But the hints of nonhuman
agency persist here as well, since the syntax itself announces that the "difference
in the way of meaning" is an aspect of language beyond its specific use,
and this difference itself "permits" the German/French divide. And
who or what is it that "strives to exclude"? The English syntax is indeterminate:
the referent in fact is "these words," even as it is also in some
sense people themselves. Again, it is the "way of meaning in these two words" that is in
conflict, though they can supplement themselves in their relation to what is
meant to access a unified pure language.[8]
Kinship as a historical force is
intermingled with the function of nonhuman agency contained within language
itself. As we refine neural network translation capacity to the point of
singularity, the non-recognizability of computer generated presence, the
decommissioning of the Turing Test, it makes sense to factor in the persistence
and gravity of local forces that not only approach the cyborg as
"users" or opportunities for interaction, but that are extensions of
the very same linguistic structures that constitute those very same neural
networks. What Benjamin accesses in his essay is these specific historic links
that compose the technological, the digital, the copy, the translation that
also retains aspect of a type of "purity." What Benjamin also points
to is the compromised nature of symbolic meaning itself.
Symbolization[9]
In this last section I want to hold side
by side two key moments, one in the progress of "The Task of the
Translator" and one in the development of neural networks, enabling them
to perhaps read each other more deeply, enabling us as readers perhaps to
position ourselves within the form of translation, as I have partially
approached it in one of the footnotes to this paper.
In the penultimate paragraph of
Benjamin's essay he contrasts the two overriding concerns of any translation,
fidelity and freedom, and writes of how a strong commitment to the exact
informational sense of an original leads to the most substantive experience of
the freedom of translation. This is a complicated notion, within which we have
to allow for the fact that translational freedom is precisely the thing that
disallows for or dismantles an overreliance on the basic sense of an original. Benjamin
anchors his discussion in the area of this extra-literal content:
Only if the sense of a linguistic
creation may be equated with that of the information it conveys does some
ultimate, decisive element remain beyond all communication. . . . In all
language and linguistic creations, there remains in addition to what can be
conveyed something that cannot be communicated; depending on the context in
which it appears, it is something that symbolizes or something symbolized.
Thus this particular
"noncommunicated" element exists in relation to the symbolic function
of language. Here what one might expect is that this mysterious element might
well be something that is symbolized, what language or translation attempts to
bring into communicative discourse. We think here of any original that is fed
into an automated translation system, and given the capacity of neural networks
this would include an original at all levels, from the informational through
every level of the symbolic (as indicated in the Hemingway excerpt). What might
be unexpected however is that this extra-linguistic element itself is doing the
symbolizing: "it is something that symbolizes." What Benjamin posits
in this noncommunicable but core area of translation is nonhuman agency that is
operating at what we perceive as the most important level of symbolic meaning.
He continues:
And that which seeks to represent, indeed
to produce, itself in the evolving of languages is that very nucleus of the
pure language; yet though this nucleus remains present in life as that which is
symbolized itself, albeit hidden and fragmentary, it persists in linguistic
creations only in its symbolizing capacity. (261)
The language of "The Task of the
Translator" is laden with words and sentences that say one thing and mean
another, and it's important to stay aware of these exchanges, divides, or
alternations in any reading. Here the oscillation appears in our understanding
of language itself, which occupies multiple positions in this particular quote.
Language is something evolving, transforming over time according to natural
force. It is pure language, or the more static effect of the supplementarity of
intentions of all languages. Language is "present in life" as what is
symbolized (brought about indirectly), hence lacking agency, while at the same
time having independent agency, "capacity," as something that
symbolizes. And there is a way in which Benjamin makes the tautological
statement here that language is present in language. One of the points of the
essay, however, is that we need to hold each of these elements in suspension in
order to grasp not only translation but language itself, as well as both human
and nonhuman agency and the manner in which they are superposed with each
other. Language is about and conveys what we perceive as its own agency as well
as whatever agency we perceive as originating from the human.
Interestingly, the nonhuman aspect of
pure language operates according to a biological metaphor, the nucleus, in the
same was as artificial neural networks circulate around the biological metaphor
of the neurons of the human brain. It's as if the advanced state of AI, where
deep learning takes place independently of human input, is constantly faced
with its own human reference point, that the "intelligence" of "artificial
intelligence" will always structurally be compromised by the fact that it
is human intelligence that is at stake. However, we would always also have in
view the fact that the nonhuman or extralinguistic is a constitutive element of
language itself, just as we saw in the "unforgettable forgotten"
example. In any case, this moment of symbolization, where what is symbolized
trades places with that thing that is doing the symbolizing, occupies the
absolute forefront of digital science. It is situated as the very subject of the
many discussions about technological singularity, or the point at which artificial
intelligence takes on more capacity than the intelligence of human beings. The
Lewis-Kraus article in the New York Times
describes this moment, in terms of multilayer neural networks, as follows. The
members of Google Brain and others in the artificial intelligence community realized
that:
. . . neural networks with more than a
billion "synaptic" connections . . . could observe raw, unlabeled
data and pick out for [themselves] a high-order human concept. . . . Out of the
welter of the pure sensorium the network had isolated a stable pattern. . . .
The machine reached directly into the world and seized the idea for itself. . .
. Machines could . . . deal with raw unlabeled data, perhaps even data of which
humans had no established foreknowledge.
The terms here are strikingly similar to
those used by Benjamin if we think of the substantive content or overall import
of technological advance being the displacement of agency. It just so happens
that through the lens of neural networks and the specific type of capacity they
introduce into computing, particularly through Google Translate, this shift is
understood as a version of "symbolization," or the ability of the
computer to identify what have hitherto been perceived as noninformational
elements of language and the world in general. Benjamin continues to elaborate:
Whereas in the various tongues that
ultimate essence, the pure language, is tied only to linguistic elements and their
changes, in linguistic creations it is weighted with a heavy, alien meaning. To
relieve it of this, to turn the symbolizing into the symbolized itself, to
regain pure language fully formed from the linguistic flux, is the tremendous
and only capacity of translation. (261)
As we have seen, the supplementariy of
intentions produces out of the "various tongues" pure language, but
the nonhuman or "alien" aspect of language carries through as another
aspect of linguistic being as well. And here again we are faced with the
transition between symbolizing and symbolized, this time in the reverse
direction, from what has agency, the symbolizing, to what does not, the
symbolized. But we must note that removing the aspect in which language appears
to have alien or nonhuman agency or import is in fact bringing it back to the
realm of the merely symbolized. Translation here finds its true significance,
as that which displaces nonhuman agency and restores the world to its
human-centeredness.
On this logic the extra-human aspects of
neural network technology need always to restore the world as we have come to
know it, even as that world, operating as an extension of the symbolized,
rather than the symbolizing, reaches beyond itself. We arrive here at a concept
of translation that works at the border between the human and nonhuman both at
the root of language and of technology, such that neither of these two
nominally distinct entities has priority. Benjamin finally outlines a realm in
which contrary tendencies cannel each other out:
In this pure language—which no longer
means or expresses anything but is, as expressionless and creative Word, that
which is meant in all languages—all information, all sense, and all intention
finally encounter a stratum in which they are destined to be extinguished.
(261)
The messianic implications are clear but
equally as clear is the foreboding of how language and technology are grounded
in the dialectical decommissioning of what we know as human, and as with the
reference to World War I this "extinguishing" has every indication
that it may well play out physically and not only according to what we
understand as the immaterial. There is a kind of destined material and
theological stasis and self-cancellation at the farthest reaches of translation
and the human. How this plays out in day to day reality is perhaps up for
grabs.
Conclusion
Translation can be understood to be
multidimensional. But when we begin to apply the concept to the digital it
takes on not only the more straightforward and mundane versions of a transfer
of meaning between languages or formats, but a sense in which, as we can see Benjamin
developing it "The Task of the Translator," it opens onto a
conceptual territory that informs some of our most advanced notions of the
digital. Through an analytic that invokes the key Benjaminian concepts of
translatability, supplementarity, and symbolization, it's possible to construct
a figuration of deep learning and neural networks that understands them from a
humanistic perspective, in the course of which analyses we reclaim central
notions of what we mean by the digital as a constitutive element of both
language and technology.
[1]
Translation is a form. Benjamin makes the statement suddenly and
matter-of-factly. He is concerned with "comprehending" translation as
a form, as if his essay translates translation itself, inhabits that task, his
task, which is not to convey the original directly (an impossible task) but to
do so in a way that is "derivative, ultimate, ideational" (259). In
some sense we dispense with our own being to become the translation.
[2]
Translation is a form, like other forms. In this case, like art itself is a form.
In Wolfgang Ernst's sense, form is the "real technological condition of
expression." Art is the form of the "original," and "The
Task of the Translator" wants to point out, to say, that art works are
not, in their essential or most basic qualities, concerned to communicate or
transmit information. This lack is what defines a work of art, and that is
exactly what translation wants to translate, the non-informative, the
non-denotative status of a work. Since that is what the work is. Anything else
is "bad translation." Translation takes on its status as a form when
its intention is to get at, to translate, this essential quality of an original
that ostensibly has nothing to do with the transmission of information, that is
not concerned with the "receiver," that is not concerned with
audience "attentiveness." The task of the translator must be to
convey the original as an original, which is the original as something that is
not translatable, since that quality makes up its originality. In this way
translation as a form is like other forms in their relation to content. They
can only be provisional or temporary, based in the moment of human perception
since any given content is not summarizable by our attempts to summarize it.
Here the idea that all form is a form of translation makes sense. Form is
translation. Real expression is a technological condition.
[3]
As you may have noticed, there is a contradiction (at least one) running
through Benjamin's essay, a parallel between how we come to understand what an
original is and what a translation, or copy, is. The point almost seems to be
to convey their interchangeability, how the one operates in terms of the other,
so that finally this quality or this dynamic is what Benjamin's essay itself is
translating to us, is in some sense transmitting to us, communicating, and at
the same time resisting, embodying. And it is my work here in discussing the
essay that must then parallel that task. It is my work here in reading the
essay.
[4]
For the formulation of the technological unconscious, see Nigel Thrift's Knowing Capitalism, Sage, 2005.
[5]
In any case, when we consider an original, it is important to note that
"the laws governing the translation lie within the original, contained in
the issue of its translatability." Thus translation in fact finds its way
into the heart of any original, which to be an original must have in some sense
already planned to be translated or to be taken away from its own nature as an
untranslatable original. "Translatability is an essential quality of
certain works." It is in this sense that translation becomes (or is
revealed to be) even more solidified as a form itself, one that appears in the
world and through which the world takes shape. An original must in some sense
correspond to the dictates of translation—only "certain works" do
this—to achieve its continued existence as something that is original. Here we
can see that translation or the cult of the copy has already in this early
essay taken a significant place in Benjamin's work, anticipating later essays
like "The Work of Art in the Age of Its Mechanical Reproduction."
Here there are two distinct dynamics that Benjamin posits as part of the
translation process.
[6]
See Bishop, Christopher M. Neural
Networks for Pattern Recognition. Clarendon Press, 1995.
[7]
First is that there is a realm of non-human translatability that may well never
be accessed. "The translatability of linguistic creations ought to be
considered even if men should prove unable to translate them." Thus there
is a quality of originals that is translatable but that may never actually be
translated, or may only ever exist in pure unrealized potentiality. This could
be due to either never finding a translator, the right translator, or to not
having any part of its nature that lends itself to translation as we know it,
having a nature that does not call for translation. If we take this latter case
as true, we might then turn back and ask whether or not the original then had
any translatability within it at all. And it may not. Translatability relies on
human perception. It's here that we need to note the linguistic and conceptual
nature of translatability. There is an aspect of Benjamin's argument here that
stays entirely within language itself. His single example of an unrealized translatability
is an analogy to the linguistic conundrum of an "unforgettable life or
moment" that is in fact forgotten by all. Certainly this might be possible
but Benjamin draws two implications said to be contained in the language
itself, in the "predicate": that the claim of being unforgotten is
"merely a claim unfulfilled by men" or that it "references a
realm in which it is fulfilled: God's remembrance." Thus translatability
is a presupposition contained within language itself but that carries with it the
idea of the existence of non-human potentiality and of a kind of theology.
Translation as form in this way (among others) takes priority over originals
existing purely in their own right, without any regard to how they line up with
what might conceivably be translated.
[8]
The second major aspect of the translation process is the "afterlife"
within which everything that is translated operates or takes shape. Even though
an essential part of an original is its translatability, that does not mean
that its being translated will "have any significance as regards the
original," or, that is, that the original will be affected by whether or
not it is translated. That said, the translation will stand "in closest
relationship" to the original, since it and nothing else will partake in
that essential translatability of the original. The analogy here is to the
existence of life itself—the "original" is placed in this same
realm—which will not depend for its existence on any given manifestation it
might take in the world, such as a person, a horse, a species, and so on. No
matter what happens to those manifestations life will still exist, beyond human
cognizance. "Even in times of narrowly prejudiced thought, there was an
inkling that life was not limited to organic corporeality." These manifestations
of life, these translations of an original entity, are then known as an
"afterlife," mostly since they must come after, follow, life or the
original (very much echoing the phenomena of "afterbirth"). An
extraordinary statement Benjamin makes here is that "The idea of life and
afterlife in works of art should be regarded with an entirely unmetaphorical
objectivity."
[9]
Here translation as form takes on a basic relevance to the phenomena of life
itself, broadly defined. Translation of originals and manifestations of life do
not have to do with what might have a soul or experience animal sensation and
the like, but with whatever might have a history of any kind. This relates
specifically to the non-human as a key component of language itself, since
translatability and hence translation introduce what has not yet been credited
with life or originality into the readability of the present, into our
understanding of what does in fact have life. In this sense translation can be
equated with the life force. Exchanging the role of the philosopher with that
of the translator, Benjamin writes that "The philosopher's task consists
in comprehending all of natural life through the more encompassing life of
history." What we perceive then is a realm of translatability only, and
even though "no translation, however good it may be, can have any
significance as regards the original," in its manifestations and
afterlives, "the life of the originals attains its latest, continually
renewed, and most complete unfolding."
That any given manifestation or
translation has to do with this aspect of a single unified life force or
original is what forms the unified underlying structure of language.
"Translation thus ultimately serves the purpose of expressing the
innermost relationship of languages to one another." A Benjamin writes,
this purposiveness "is almost beyond the grasp of the intellect," but
what it has to do with
If we attempt to define free translation
we get the benefit of the idea appearing near the end of Benjamin's essay
"The Task of the Translator," which means that it must already
contain within it the progress of the rest of the essays. The paragraphs are
recursive and contradictory in nature, meaning they circle back on themselves,
sometimes containing blatant repetition, and that they often end up in a place
that seems to contradict where they started. For us to pick up the concluding
concept of free translation means that we should be able to touch on most of
the major points of the essay, means that the network of conclusions and
starting places the essay embodies should begin to surface. This a method of
digging back into the content of the essay, one that translates the essay by
imitating the method of the original.
In this paper it's also possible to look
at the article and the quite central role of the "linguistic turn,"
which is phrased here as a parallel to that move in philosophy, which declared
the centrality of language to all existence. There is no outside the text. Our
perception of the highest state of innovation makes its way to us as the
journalistic climax and mimetic of the step difference between what had been
functional as AI and this new edition, which is the entire point of the article
and the thing that purports to take high literature into the cybernetic fold.
This is the place at which translation truly operates, even though translation
is happening on multiple levels, from the journalist translation of complex
science, to the translations the scientists and CEOs are doing of the human into
the technological, when then translates the cyborg back into profits that then
further mechanize the cyborg. Agency here makes its way straight through
symbolic thinking, so that the nonwhite author of the article and the nonwhite
CEO of Google effect a transfer point for symbolization to take effect at the
deepest level possible. If we consider the way web 2.0 manifested itself as key
factor in the Arab Spring, which then led to the reaction of repressive
regimes, that then led to the migration crisis, it is remarkable that Google
CEO T then pushes forward to offer web technology as a solution. But this makes
sense exactly in the realm of a step change for symbolization.
No comments:
Post a Comment