Both propositions, I would argue, are
true: science-fiction writers have been writing about machine translators
for decades now, but only, it seems to me, because they (at least those
working in English) are not really interested in how languages work, how
human translators develop translation skills, how a working machine translator
might be developed. Science-fiction writers are interested in writing
speculative adventure stories about other times and other places than
our own. Those stories almost invariably bring characters like the writer
human English-speakers into close cultural and linguistic
contact with members of other species; but the writers dont want
to get bogged down dealing with the intricacies of interspecies/intercultural/interlinguistic
communication (they want to get on with the story), so they invent a gizmo
to handle all translation problems for them. Some sort of box that intrepid
space explorers lug around with them. A hearing-aid-like thing that they
put in their ears, which somehow magically makes it possible for them
to understand everything the bug-eyed monsters say and to speak BEMese.
Lets look at some of the forms of MT as imagined by science fiction writers over the past few decades, relying on the thumbnail sketches compiled by Walter Meyers in Aliens and Linguists.
And so on. There are more; let these
stand in for the larger group covered by Meyer. Overall, Meyer notes,
the passion for machine translation among sf writers is somewhat remarkable,
given their general lack of sentimental regard for their heroes
For the most part, the writers Meyer
summarizes mention the translation machine once and then forget about
it; once the obvious language problem is dispensed with, translation drops
out of the story altogether, no longer of any interest. A few sf writers,
however, do recur to the problems of translation typically, though,
without much sense of how languages work or what translators do. The writers
of the Darmok episode on Star Trek: The Next Generation
invented a people who speak an apparently untranslatable language
that has the Enterprises universal translator close
to system overload. Finally the linguistically sophisticated members of
the crew have to put their heads together and solve the problem,
which in fact is a laughably bogus problem in the first place. For the
Darmok language is not really a language at all; it is a collection
of about six highly metaphorical and allusive English sentences, which
the members of this language community keep repeating doggedly,
over and over, throughout the episode. Nor does the fact that the sentences
are in English seem to be merely a representational device, shorthand
for theyre really speaking in some totally weird language,
and weve managed to translate it this far into English, but it still
doesnt make sense for everyone on the Enterprise immediately
understands the words in the six-fingered handful of bombastic pronouncements,
without the aid of the ships universal translator. They
just cant figure out what the sentences taken as a whole mean. One
way of putting it: they really arent trying to translate a foreign
language; theyre trying to interpret a very brief English-language
poem (say a six-line national epic, with important historical allusions).
They finally succeed by running computer searches on all the puzzling
lexical items in the mysterious language (all three of them),
and here these crew-members astonishing linguistic genius
really shines through getting the computer to cross-reference the
search in a whole network of cultural systems. They get several hits.
They figure out what the words mean. Now all of a sudden they understand
Another typical attempt to focus more
serious scrutiny on machine translation in science fiction is Alan Dean
Fosters in The False Mirror. Foster manages to raise some
interesting issues regarding machine translation in this novel about a
war between humans and several other species (the Svan, the Hivistahm,
and the Wais, who seem to be born interpreters); but it seems he is not
himself quite aware that they are interesting issues, since he passes
over them quickly, even unconsciously. Let me just give a series of quotes
with pointed questions:
On the scope of translatability:
Does the translation machine imitate
voices? Machine translators are not normally thought of as impressionists;
but perhaps we should begin thinking along those lines? Our own machine-translation
researchers have typically tended to reduce translation to the replication
of meaning; is that model inadequate?
Does this imply that all grunts are untranslatable,
or just that this particular one was? Can the translators distinguish
between language and nonlinguistic noise? Or is nonlinguistic noise simply
anything that it cannot translate?
Do the machine translators do body language?
Or would this fall into the same category as untranslatable grunts?
Everyone in the novel seems to have been
fitted with a hearing-aid-style translator, and theyre always fiddling
with it, making adjustments to it, trying to make it perform better:
Adjusted his translator to do what? Rearrange
his syntax? But in this book all Hivistahm use inverted syntax, whether
they are speaking to each other in their own language[?] or to other species
Does this imply that there are settings
on the translator that allow the user to specify the target language?
Elsewhere it is called a universal translator (89)
why then these adjustments?
Foster here seems to be operating on
the analogy of a radio tuner: you fiddle with the dial to make a station
come in more clearly. But even on a radio, is it possible to fiddle
with the transmitter to improve some unknown listeners reception?
How does the sender even know when the receiver is comprehending his or
her transmission clearly? And even if it were possible to know such a
thing, how could it be remote-controlled? And what would fiddling
do to a machine translator? In our current state of machine-translation
incompetence, we would dearly like to know how to fine-tune a system so
as to optimize target-language (re)production and so maximize receptor
comprehension. But so far all our attempts remain woefully inadequate.
What must this sf technology be like that a user, untrained in computational
linguistics, can improve the results of translation by fiddling
with the instrument?
How does he know its finely tuned?
Or is this the narrators omniscience at work?
And finally on the socioeconomics of
Does the existence of a species that
has specialized in translating imply that machine translation technology
is not universally available? Perhaps it is too expensive for some people,
some groups, some planets to purchase, and they must rely on Wais? And
are we to infer from this that nonmachine translation is actually cheaper
than machine translation?
Does that disdaining imply
that proficiency in foreign languages is socially valued more highly than
Not all science-fiction treatments of
translation are this cursory, however. Susanne Elgin Hayden puts translation
at the very center of her novel Native Tongue and its sequels:
the ruling family in the books society handles all interspecies
translation, giving them inordinate power over trade and knowledge, indeed
to the point of rendering them the societys aristocrats. This power
is fiercely resented by the ordinary citizens (including the politicians
that ostensibly run the country) who do not translate, and they are constantly
trying to seize it from the family, by kidnapping babies from the familys
nurseries (the skill must be biological!) and a variety of other stratagems.
Nothing works. Whenever the ordinary citizens try to train their own children,
or even the kidnapped children of the translator family, in alien languages,
the children go crazy and die, proving Wittgensteins point that
Wenn ein Löwe sprechen könnte, wir könnten ihn nicht
verstehen (if a lion could talk, we couldnt understand him).
Indeed Hayden takes Wittgenstein one step further: not only could we not
understand him; attempting to do so would prove fatal to us. Interesting
Three other observations.
Another example of a more complex treatment
of translation in science fiction, this one specifically a machine translator:
C3PO in the Star Wars movies. He is a prissy, fastidious robot
who speaks and can translate to and from 1.2 million languages. Just how
he has been programmed to do this, we are never told, of course; in this
sense George Lucass imagination of machine translation is just as
simple as any other recent science fiction writers. But C3PO also
has features that raise him above the herd of his sf colleagues.
Symbolically he is even more interesting.
He looks like a human stylized in gold (far more humanoid than his droid
pal R2D2) and sounds and acts like a comic British aristocrat, ruler of
the world in the 19th century, reduced to a golden stage-and-screen buffoon
in the 20th. C3PO is specifically a caricature of the effete Old World
intellectual, literally stuffed full of book learning, a physical coward,
a moral prig and a translator, along for the ride because they
need a translator.
What a contrast between this Old World
buffoon and the dashing New World heroes Luke Skywalker and Han Solo!
Of course, Han can get by in a foreign language or two he understands
them (Chewbaccas, Jabba the Huts), but he never actually speaks
them. He just speaks English, and they understand him: a rough-and-ready
mode of communication that is in fact quite common in the real world,
each speaker speaking his or her own native language and understanding
enough of the other to keep the conversation going. Luke is the Percival
figure, the rube from the backwoods, so of course he speaks no foreign
languages; interesting that his Jedi training under Yoda involves no work
in languages either, and that Yoda has to speak to Luke in English, which
he obviously doesnt feel entirely at home in.
Also, of course, Star Wars is
a movie trilogy about empire and rebellion in which the main rebels are
white North Americans and the main imperials are vaguely European (icky
British and German accents) the movies symbolically refight the
U.S. Revolutionary War, or (more problematically) World War II (the US
fought WWII from a position of strength, not the weakness of a tiny band
of rebels). This makes it a useful tool in the resymbolization of American
imperialism, in Vietnam, in Latin America, in Africa; which also gives
it an enormous cultural (imperial) power worldwide. U.S. presidents (especially
Reagan and Bush) invoke the movies to garner additional ideological support
for their policies on the crumbling Soviet Union (which Reagan refers
to as the Evil Empire), Grenada, Libya, and Iraq. In the 1980s the Strategic
Defense Initiative, designed to shoot the other guys incoming missiles
out of the sky, was commonly called Star Wars technology, again associating
the other guy with the Evil Empire and the US with the plucky band of
right-minded rebels. The resymbolization of the U.S. empire as small (small
is beautiful), vastly outnumbered by evil forces, and pure of heart
akin to the resymbolization of the powerful conservative backlash
as a small but valiant band of rebels against liberal (PC)
The camera follows a female tourist of
middle years in green clothes that shimmer in the sun, antigrav jewelry,
and virtual makeup. She carries a holo map, and checks it frequently against
street signs. She's flashy but elegant, an upper-middle-class matron from
one of the hightech planets. Front view: furrowed brows, intent looks.
Rear view: heels clicking, envious glances. Classical music plays over
the crowd noise: zydeco. She stops in front of a large public building,
checks the address against her map, and pushes through the door. Well-dressed
functionaries stand behind a counter. She looks them over, then walks
confidently up to a nice-looking man a few years her junior. He smiles
MAN (with a friendly smile)
The woman, puzzled at not understanding,
taps at her ear with a long shiny retractable fingernail, enameled the
same shimmery green as her dress.
MAN (same friendly smile)
The woman slides her nail carefully into
her right ear and pulls out the translator, taps it again, turns it over
between thumb and forefinger. The green power light is on, however, and
the battery level is high. She shrugs, inserts the translator back in
her ear, and walks out with a mute apology.
Zoom on functionary's head and upper
torso. He smiles conspiratorially and taps his breast pocket. As the camera
zooms closer, he pulls out what appears to be a pack of cigarettes. He
holds it out to the camera's view. It says "KOOL Lites" on the
side, but it looks strangely heavy.
Cut to product still of the ZD2000 Translation
Jammer in three different disguises: the cigarette pack that we've just
seen in action, a transistor radio, and a pocket diet transmitter.
It has become customary to distinguish
among several degrees of human-machine interface in translation: machine
translation (MT) proper, computer-aided translation (CAT), and human translation
(HT). This seems a bit silly, since the extremes on that cline, MT and
HT, are so rare as to be virtually nonexistent, and thus serve mostly
as markers on a purely hypothetical scale. Who translates without computer
aid these days? I know one literary translator (mostly of poetry) who
does all his work with a pencil and paper, then has a secretary type it
up on a computer in the department office; such people do still exist.
But as well see in a moment, it is a narrow conception of translation
indeed that would exclude the secretarys work from the total overall
translation act; and even if my friend did not have her key his translations
into digital form, in order to publish them he would need to send his
handwritten scribbles to a press, where someone would key them into publishing
software (and probably copy-edit them on-screen). Again, some people define
translation narrowly enough that the process of editing and publishing
a translation is not part of the total act of translation you have
the translator who translates, the typist who types, the copy-editor who
copy-edits, etc. but this seems excessively atomistic in an age
of systems theory, better suited perhaps to the eighteenth century. I
will be returning to a systems approach to translation in a later section.
It is also somewhat misleading to speak
of MT proper and this is the main burden of my remarks here. Where
in the world do we find strong MT, MT that is not HAMT (human-aided
machine translation), MT without human pre-editing of the source text
(input) and post-editing of the target text (output)? Strong MT remains
an ideal, a goal to shoot for; it is not a reality. Indeed some MT researchers
are now saying we will never make it a reality, and we should stop shooting
for it. Its impossible. That may just be pessimism born of frustration.
Still, as strong HT disappears, it is not currently being replaced by
A few years ago the machine translation
laboratory at Carnegie-Mellon University produced an elaborate system
for Caterpillar Corporation that they touted in their patent application
as strong MT, because the input needed no pre-editing and the output needed
no post-editing. The way they managed that, however, was to add a step
to the creation of input texts, involving a process by which the computer
guided a human tech writer in the writing/editing of natural source texts
into what they called constrained source language (CSL), or
what machine translation researchers now call domain-specific controlled
language: High-quality machine translation is currently feasible
only when the text to be translated is highly restricted. It must be restricted
to the vocabulary of some narrow domain of knowledge, such as a particular
brand of photocopiers, and it must be straightforward in style and grammar.
This type of text is often called a controlled-language text (Melby
1). The tech writer would either originally write a source text specifically
for the MT system or edit an existing one to meet its disambiguation needs,
attempting to avoid or eliminate all pre-identified types of ambiguity.
Every time s/he would fail every time s/he would introduce some
sort of ambiguity into an original text or leave one in a pre-existing
text that the MT system could not disambiguate the computer would
beep and not let him or her proceed until the ambiguity was cleared up
to the computers satisfaction.
According to all reports (I used to live
in the town where the CMU/Caterpillar system has been primarily implemented,
and know several of the implementers personally), the system is not working
as expected. The labor of loading the 70,000 technical terms Cat uses
in each language into the system is itself herculean, but that is the
easy part. Someone next has to identify (in general, in the abstract)
all potential ambiguities between English and every other language Cat
works with; if they miss even one, the controlled-language input texts
are insufficiently disambiguated for the system and its output will require
human post-editing. (And in fact human post-editing has almost without
exception been required.) Finally, the human tech writers hate writing
controlled-language texts, and especially hate being beeped at by the
computer every time they fail to write it to system specs.
Clearly, the CMU/Caterpillar system is
not an example of strong MT.
What I want to suggest in this essay,
in fact, is that all translation today falls in the middle category, computer-aided
translation (CAT). More radically still: I would argue further that for
the foreseeable future say for the next several hundred years,
or as long as computers are around to help us do our work say until
the apocalypse, or the collapse of our current information society
say until we no longer have the electricity to run computers or the knowhow
to build and maintain them all translation will continue to be
Or, more neutrally: all translation will
continue to require a human-machine interface. Indeed even if strong MT
becomes a reality, it will continue to need humans to program it: to feed
the vast quantities of terminology and parsing rules into it, to identify
possibly damaging ambiguities between source and target languages, and
to develop disambiguation procedures. And in a broader focus the systemic
network that takes a source text and produces for it an acceptable target
text will continue to involve a significant human contribution, in the
form not only of programmers but of project managers and assistants that
keep the process flowing smoothly. Even the strongest MT system will never
be fully automated.
We have a term for the human-machine
interface, derived from science fiction (of course!) and adapted persuasively
for social theory by Donna Haraway: the cyborg (cybernetic organism).
In the traditions of Western science and politics,
Haraway writes in her now-legendary Cyborg Manifesto (1986),
the tradition of racist, male-dominant capitalism; the tradition
of progress; the tradition of the appropriation of nature as resource
from the reflections of the other the relation between organism
and machine has been a border war (150). MT, CAT, HT: border war.
Where shall we draw the boundaries between the human and the machine?
Who or what translates? Nature and culture are reworked, she
adds; the one can no longer be the resource for appropriation or
incorporation by the other (151). Or, well, more specifically they
are both resources for the appropriation and incorporation by the other,
which means that neither can ever entirely appropriate or incorporate
the other, for it too is forever (to use the term of the Star Trek Borg)
being assimilated. Human-becoming-machine, machine-becoming
human. With the result that, By the late twentieth century, our
time, a mythic time, we are all chimeras, theorized and fabricated hybrids
of machine and organism; in short, we are cyborgs (150).
I suggest, then, that it may be more
fruitful to shift the discussion of human-machine interfaces in the translation
field from the MT/CAT/HT triad into the realm of cyborg translation. The
motto of this new approach would be: all translators are cyborgs.
The family at dinner: Mom, Dad, teenaged
daughter. Dad is serving the rest of the family. Meatloaf, potatoes, and
boiled carrots. As he holds the tray for each in turn, they talk.
Mom brushes her hair back with her right
hand, casually touches her ear on the way past. She looks troubled.
Dad finishes serving his womenfolk and
takes the tray into the kitchen. The camera follows; halfway down the
hall, he turns his head slightly and winks. In the kitchen he sets the
tray down and turns to face the camera, pulling what seems to be a rubber
coin purse from his pants pocket.
Cut to product still of the ZD3000 Genderlect
Translation Jammer in three different disguises, the rubber coin purse
that we've just seen in action, a compact for the ladies, and a Sony Walkman
for the youngsters.
Do you instinctively resist the notion
that when you translate you too are a cyborg? You sit at the computer,
yes. You value the many support functions the computer offers you, not
just in typing (no need to retype when you make a mistake!) but in spell-checking,
global search-and-replace functions, term-management functions, and formatting,
and, once the computer is hooked up to the internet through a modem, access
to web resources (on-line terminological databases, contextualized term
searches through specialized websites) and, via e-mail, connection to
translators and other experts worldwide, both individually and on listservs
like LANTRA-L. All this is important, certainly. In fact it is difficult
to imagine how you got along without it, just a few years ago, typing
your translations on a typewriter, relying much more heavily on print
dictionaries, making phone calls to experts.
But still: you do the translating.
The computer helps; but you are the translator. You are a human being.
The computer cannot translate; its a translation tool. Only you
A classical definition of the cyborg,
offered back in 1960 by Manfred E. Clynes and Nathan S. Kline in an Astronautics
article entitled Cyborgs and Space: any exogenously
extended organizational complex functioning as an integrated homeostatic
system unconsciously (Clynes and Kline 30-31). Five keywords: exogenously
extended, organizational complex, integrated, homeostatic, system, and
Exogenously extended because the
cyborg extends human agency outward, into the world, into environments
(such as deep space or the bottom of the ocean) where humans cannot survive
without such extensions, into performances (such as fantastically rapid
recall and analysis, or manipulation requiring superhuman strength) of
which humans alone would not be capable. The cyborg translator translates
faster and more accurately than the human translator. It edits translated
text far more reliably, both bringing together humans and machines in
widely separated geographical places through sophisticated communication
technology and enabling rigorous checking and testing processes.
Organizational complex and system
(to consolidate my definitional task somewhat) because in cyborg functioning
it is impossible to reduce agency to the will of a single human being,
or any other part or unit in a larger system.
The translator sitting at the computer, internet connection open, connected
by incoming e-mail and ICQ to translators around the world, ready to check
problematic terms on the web, seeing words in his or her translation that
are not found in the word-processors lexicon highlighted with little
red dotted lines on the screen ... this translator is shaped by the computer
interface. S/he would translate differently without it. S/he would be
a different kind of translator without it. S/he would live a different
kind of life without it. (For example, a good friend of mine lives in
rural Washington State and translates from Swedish, almost exclusively
for Sweden-based clients. Because of the instantaneity of e-mail connections,
she finds it professionally essential to work during Swedish business
hours, which in U.S. terms basically means the graveyard shift. So she
wakes up shortly after midnight and works till morning, then naps and
gets her kids off to school and does various household chores and chats
on ICQ with translator friends and edits her nocturnal production throughout
the day. Grabs a few more hours of sleep in the evening, and is ready
to start over again.) Together the translator and the computer form a
complexly organized translator-system a fortiori when the cyborg
translator involves several humans and several machines in a single complex
Homeostatic, or self-regulating, because only homeostatic systems can continue to function stably in their particular environment. This is in fact the problem with machine translation, as it has been idealized in the AI community: stand-alone machines cannot translate homeostatically. They are in constant need of adjustment. Alan Melby calls this process tuning. In 1954, for example (Melby tells us), the Georgetown/IBM MT project gave an influential demonstration
This sort of tuning is not so much avoided
in cyborg translation as it is internalized. It becomes an integrated
part of the systems functioning. The system becomes self-regulating,
or homeostatic. When problems arise, the system adjusts. This is, after
all, what stand-alone human translators used to do; cyborg translators
now do it more effectively. It is an historical irony that the key frustration
for MT researchers was not that their machines seemed unable to keep adjusting
to the translation problems encountered; they were not thinking cyborgically.
Their frustration rather was that natural (non-domain-specific) language
seemed to resist their every attempt to reduce it to a single set of universalized
rules that could be learned once and for all by a digital computer. They
thought adjustment was a necessary evil, a somewhat embarrassing transitional
stage preceding the advent of universalized linguistic logarithms. They
were wrong. Adjustment is the name of the game.
Bring all the separate parts together into a single organizational complex, integrate them into a system. Make the system function homeostatically, constantly adjusting to changes in its environment. Make those adjustments autonomic, automatic, never (or rarely) requiring the system to stop and pay attention to them. Let the result extend natural human function and agency out into the world in facilitative ways, enabling both enhanced efficiency in the performance of existing tasks and breakthrough creativity in the invention or discovery of new tasks.
Music: Mozart, "Eine kleine Nachtmusik."
A patient in his mid-50s is walking around
the room wearing a virtual reality headset, muttering to himself. The
camera follows him from over the psychiatrist's left shoulder. She watches
his progress on a 21" console monitor with three windows open: one
a graphic representation of his mental images, the second a running transcription
of his subvocalizations, the third a translation of the unconscious material
repressed in these subvocalizations. The psychiatrist watches for a few
minutes, increasingly bothered by something. Finally she reaches a decision.
The psychiatrist taps the monitor screen,
disturbed. The running translation has followed the patient's vocalizations
almost verbatim. As she ends the session and the patient removes the headgear
and prepares to go, she runs her finger down a list of phone numbers till
she stops on "Translator Repair."
The patient has his coat on and is ready
to leave. The camera follows as he shakes hands with the doctor and walks
down the hall to the receptionist's desk to pay, then out the front door.
Only when he is safely out in the street does he turn to the camera and
smile, beckoning us closer with a crooked forefinger.
He reaches in his pocket and pulls out
what appears to be a pet rock.
I came close to losing my job based on
a negative psychiatric evaluation. Then I found one of these ZD4000 Psychiatric
Translation Jammers at my local Radio Shack, and boy has that made a difference!
Even my boss's boss is convinced that the real problem is the dickhead's
poor communication skills!
Cut to product still of the ZD4000 Psychiatric
Translation Jammer in three different disguises, the pet rock that we've
just seen in action, a rabbit's foot, and a television remote.
In The Cyborg Handbook David Hess muses on what he calls low-tech cyborgs:
A low-tech cyborg for Hess is simply
any ad hoc or temporary human-machine interface, like driving a car or
watching television. Or sitting down to the computer to translate. (A
high-tech cyborg translator would be C3PO, say something we do
not currently have the technology to create, a fully functional translator
android.) The machine exists. It is in many ways a technological marvel
in its own right. But there are many things it simply cannot do without
a human partner. The car cannot pull out of the driveway, stop at the
four-way stop sign and then proceed when its turn comes, pause before
taking a free right, merge on the freeway, turn off its cruise control
when both lanes ahead are blocked by slower-moving vehicles, or decide
when it is safe to switch it back on again; the car cant even stay
between the white lines, though we do currently have the technology to
build cars that could. And no, the computer cant translate. But
the car with a human driver (or the human driver with a car) can do all
those miraculous things. And the human sitting at the computer (or the
computer run by a human) can translate. Together they form a temporary
More importantly, the kind of systemic
thinking explored by cyborg theory is increasingly endemic to our society.
We still do engage in petty border wars with machines; but even in our
own strategic thinking about those wars, we are losing. Losing not in
the sense of losing ground, feeling the boundaries being moved uncomfortably
close, but in the sense rather of blurring the boundaries in our own minds.
Sandy Stone, in her article in The Cyborg Handbook, tells the story
of discussing virtual communities with some of her colleagues. They scoff
at the notion of virtual community, jerking a thumb at the computer screen
and saying Not much serious work gets done in there
which always gives me a chuckle, Stone writes, since
with that airy gesture they simultaneously accept the interface metaphor
and dismiss its implications (393). They both recognize and refuse
to recognize the blurring of boundaries in their own thought. They recognize
that blurring with their thumb and deny it with their words.
What cyborg theorists are calling for
in all of the various fields where its paradigms are proving revolutionary
is a deliberate blurring of those boundaries. Gary Lee Downey, for example,
in anthropology: A crucial first step in blurring the human-centred
boundaries of anthropological discourse is to grant membership to the
cyborg image, i.e. to recognize in our writing that human actors routinely
produce themselves and their machines as part human and part machine,
and that machines have positioning strategies too (369).
According to the editors of The Cyborg
Handbook, cyborg technologies take four different forms: restorative,
normalizing, reconfiguring, and enhancing (Gray/Mentor/Figueroa-Sarriera
3). Cyborg translators are currently thought of almost exclusively as
enhancing: improving existing translation processes by speeding them up,
making them more reliable and cost-effective. And there is no reason why
cyborg translation should be anything more than enhancing.
Still, trying to imagine how cyborg translation
might fit the other categories as well may prove heuristically productive.
A restorative cyborg translator technology, for example, would seek to
replace translation function in a system that had lost it, like fitting
an amputee with a prosthetic limb; and in some sense that is in fact what
happens when a noncyborg translator becomes computer-literate, enters
the world of cyborg translation. The computer becomes a prosthetic device
to replace (or, perhaps, better, to normalize) translation
function that the professional translation market regards as missing
in the noncyborg translator: the ability, for example, to receive and
deliver text by e-mail, or to format it in Microsoft Word. Electronic
formatting and delivery are basic professional requirements in the translation
marketplace today; a translator who lacks those capabilities doesnt
have a leg to stand on, needs a prosthetic leg, a computer needs,
in other words, to become a cyborg translator. This restores
or normalizes translation function.
Other specialized cyborg types:
The ontology of cyborgology,
the Handbook editors write aphoristically, is embodiment
(Gray/Mentor/Figueroa-Sarriera 12). What is embodied in the cyborg is
typically an idea or a function, an abstraction of some sort in
this sense cyborg engineering is, in Coleridges terms, a dim analogue
of Creation. This abstraction is commonly referred to as agency:
I use the term agency, Gary Lee Downey writes,
to refer to the act of positioning itself. CAD/CAM developers and
users transcribe human agency into their technologies by abstracting informational
content from engineering activities, translating it into binary code,
and then reinserting the empowered technological agents back into those
activities as active participants (366). Thus Automating the
drafting process involves transcribing drawing practices into computer
graphics programs and then inserting those programs back into the drawing
activities. 2D CAD/CAM is constructed on the image of a drafter at a drawing
board (Downey 366-67).
But this is only the lowest level of
cyborg automation. The individual human translator does certain things,
performs certain activities, utilizes a certain kind of translating agency;
all this is replicated electronically and fed back into the human translators
own functioning, enhancing it in a series of feedback loops.
And the same kind of thinking could be
applied to cyborg translators, specifically hypercyborg translators operating
at a higher level of generality than a human translator sitting at a computer.
What is the infrastructure in which disaggregated translation agencies
function? A publications manager at a large manufacturing firm needs a
document translated: what steps are taken, what individual semicyborgs
enlisted and linked, what technological systems created, to bring that
translation about? The entire disaggregated translation agency would include,
just in terms of human bodies, at least one project manager, one editorial
assistant, one research resource person, and one freelance translator;
larger jobs might involve two or more of each, plus new bodies like financial
managers or communication managers. These human bodies would be variously
hooked up to computers and telephones and, through them, to internet resources
like on-line databases and e-mail listservs. Some of the computers would
be personal units, outfitted with a free-standing operating system and
various software packages, especially for word-processing, term-management,
cross-platform conversions, web-browsing, and e-mail; others would be
LANned, hooked into local-area networks, with mainframe support. Special
hypercyborg connections may be set up: conference calls, webboards, listservs.
So: if one were to automate more of this disaggregated agency and inject its empowered technological forms back into the hypercyborg, what infrastructural shifts would one want to undertake? Term-management software has come out of the perception that translators need to remember term sets and use them consistently, and that practicing translators often reopen old files to see how they translated a term in a previous job; what sort of software might come out of the perception that translators, editors, and project managers all collaborate on big jobs, send materials back and forth, do research for each other, check each others work? How might listservs and webboards be adapted to the communicational needs of hypercyborg translators? How might on-screen copy-editing software be modified for the creation and editing of collective texts, with emendations flagged and coded for historical sequence? One of the most cumbersome aspects of collective or hypercyborg translation today is that, while the target text is being created, it exists on different computers and in different heads in various versions and fragments that are exceedingly difficult to collate and compare: what pieces go where? what replaces what? which parts have been edited, by whom, in what order, and what other parts still need to go to which people? Hypercyborg translation might generate a hypertextual environment that is densely linked and layered for lexicographical and editorial history and made instantaneously available to all participants in the hypercyborg agency at every stage of the textualization process.
Music: a selection of light ganthan dining
The clientele is mostly human, the servers
five-foot-tall silvery lizard-like bipeds with a vermillion stripe down
Pan across the dining room, then zoom
on two human women at a table.
A server lumbers up and speaks in heavily
WOMAN (with a wink at the camera)
Exit server, bowing. The first woman
has been looking at the second woman in unconcealed amazement.
WOMAN 1 (with a sigh)
Turns to face camera. Zoom on her face
as she taps her head.
The image of the cyborg constitutes quite
a serious ideological assault on individualism, and as such makes many
people profoundly uneasy. If I am only one part of the actual agency doing
the translation, it becomes increasingly silly to worry about my boundaries.
What were my words, what became edited versions of my words, what words
were provided by somebody else? This kind of individualistic anality quickly
proves irritating to more adept hypercyborg translators. Who cares whose
words they are? Who cares who does the work? We all do. Agency becomes
loosely collectivized. Subjectivity becomes deterritorialized. In Star
Trek terms, rugged individuals like Captain Picard are assimilated into
the Borg (if only temporarily).
In her Cyborg Handbook piece Sandy
Stone discusses the unsettling implications of this critical shift:
The critical location technologies for
hypercyborg translators: time zones and billing units. What time is it
right now for x, y, and z? The key question is: are they awake and at
their computers? Freelance translators today are semicyborgs: they only
become integral parts of cyborg translation systems when actually sitting
at their computers, awake and alert enough to read words off the screen
and type new ones there. They typically complain about spending too much
time there, but in fact most of them feel more comfortable at their computers
than anywhere else. They also complain about not having a life
their so-called have-a-life coefficient, an only half-humorous
measure of their time spent outside cyborg translation systems, is typically
low, which individualist society has conditioned them to think of as a
bad thing, something to complain about. But their complaints typically
also have an air of boasting about them: I just spent twenty hours
straight, translating a big job that had to be delivered by nine a.m.
today. Im exhausted. I should really go to bed. Instead here I sit,
writing about my accomplishment to all of you. Congratulate me!
And the congratulations pour in. The complaining/boasting translator stays
on-line a little longer to read them, swilling coffee, to share in the
euphoria of cyborg translation. An old-time individualist would be asleep
long ago. But while sleeping, the semicyborg is cut off from the integrated
systemic cyborg body, like a prosthetic limb removed for the night.
Because of the human bodys need
for sleep and the human workers need for payment, of course, location
technologies remain important in hypercyborg translation systems. We need
to know when all the scattered semicyborg translators are in place (when
the human bodies are awake and at their computers) so we can contact them
immediately during the job; and we need to know what contribution each
individual semicyborg translator made to the overall hypercyborg project
after the job, so the funds available for the project can be properly
allocated. Stone calls this process warranting the
production and maintenance of a link between a bodys physical and
discursive space, or between the physical body (which needs sleep and
needs to be fed with the money obtained from work) and what she calls
the legible body, or textually mediated physicality
The legible body is probably most clearly
visible in hypercyborg translation in dedicated translator chat channels
and ICQ connections; less obviously on translator listservs. In the immediacy
of chat, the semicyborg translator writes the reality of his or her human
body: just woke up, drinking my first coffee of the day, got a big job
but dont want to start on it just yet, etc. To his or her correspondents,
s/he (a very cyborgic construction!) is just words on the
screen; but again, in the immediacy of e-mail chat, where every keystroke
is visible, it seems essential to flesh out a larger body, a legible body,
a textually mediated physicality. Humor makes that textually-mediated
physicality emit a very big grin (vbg) or laugh out
loud (lol) or roll on the floor laughing (rotfl), and
even more poignantly, it seems to me, makes it spray the computer screen
with coffee an event for which there is yet no accepted internet
abbreviation. The power in that last image from a cyborg (or what Stone
calls a prosthetic sociality ) point of view is that
the physical effect of humor is not just laughter in the legible body
but a legible connection across the human-machine interface. In some sense,
legible coffee drops on the computer screen signal the creation of a textually-mediated
physicality that contains the entire semicyborg unit, human and machine.
The computer screen is not just the place where the human reads the words
of his or her virtual friends or hypercyborg colleagues in chat; it is
also the place where (dis- or re-)embodied humor is inscribed. And, of
course, as s/he rewrites his/her physicality as text, it reappears on
that screen, gleaming through the gelid drops of coffee running down it,
first as an editable unsent chat message, then as a real utterance
in the ongoing flow of chat, where the other textually mediated bodies
can read it, indeed can scroll up to reread it, or save it to disk, even
to print it.
Translators working in hypercyborg translation
systems make friends by something like this process. You subscribe to
a translator listserv in order to get terminological help with translation
jobs youre doing. You get that help, often more help than you need.
Sometimes generous listserv members will do web searches for you and send
you (and the rest of the list) the urls. Sometimes, given a sentence or
paragraph as context for a problematic term, they will translate the whole
passage, thus becoming ever more integral (if unpaid and unrecognized)
parts of the hypercyborg translator doing the job. Terminological help
often veers into on-list discussions of issues like translator pay and
ethics, or political and cultural questions regarding political correctness,
or current events in this or that country. You disagree violently with
some posts, agree strongly with others. Gradually the list begins to take
on a shape for you the shape of virtual community. You begin to
recognize patterns in the textually-mediated bodies or personalities of
the most vocal list members. Particularly impressed with a specific term
explication or cultural statement, or disturbed by a flame, you contact
the poster by private e-mail; the two of you exchange messages for a while,
begin to ask about each others lives, where you live, who you live
with, how you feel about this and that. You exchange ICQ UINs (universal
identification numbers), and meet there regularly to chat; you visit the
lists IRC chatroom and make more friends. With each interchange
the legible bodies of your friends become richer, more complex, more fully
fleshed out; you are doing the same with your own, for them. You exchange
.gifs, graphic image files; your legible bodies take on two-dimensional
visual form. Finally, through the travel of one or more friends, you meet
in real life (irl). But in what sense is it real?
(What is reality?) Your conversation when you meet is all about the list,
which drives non-list-subscriber spouses in attendance to distraction;
your perception of your friends physical body is powerfully shaped
by the legible body you have been interacting with for months, perhaps
years. Briefly, tentatively, you attempt to fill the legible body in with
perceived details from the physical body; you may even have a brief fling,
get naked together, come home with really intimate details of the physical
body. Still, the directionality is not toward the physical body so much
as through the physical body to an enhanced conception of the legible
body: for after your visit you return home and go back to interacting
with your virtual friend through the social prosthetic of the computer
and the modem.
Sandy Stone writes about traditional
friendships and acquaintances, social relationships mediated by shared
physical spaces (workplaces, neighborhoods) rather than communication
prosthetics, and notes that It is this constellation of fixed relationships
that virtual communication prosthetics, instantiated as cyberspace, disrupts
so thoroughly (400). But I wonder: is the disruption really so thorough?
And is it even so new? Stone herself notes that traditional social and
professional networks have quite comfortably assimilated the telephone
the first electronic network prosthesis (402)
into their physical mediations. Many of us have friends with
whom we talk on the phone almost every day, and whom we also see at work
or at the store or at the kids school events with traditional regularity.
Those phone conversations are not perceived as a threat to traditional
embodied friendship, because, perhaps, a friends voice over the
phone is still his or her voice, part of his or her physical body. Somehow
a textual trace of that same body, sent over the phone lines to someone
you have never met, is a totally different thing? Maybe it is and maybe
In her attempts to get at this difference
Stone is particularly fascinated by phone sex, since phone sex bridges
the conceptual gaps between chatting vocally with a friend on the phone,
which we think of as traditional premodern embodied friendship, and chatting
textually over an internet chat line, which we think of as radical postmodern
disembodied friendship. Phone sex, Stone suggests, is a form of data compression:
Usually sex involves as many of the senses as possible: taste, touch,
smell, sight, hearing and, for all I know, short-range psychic
interactions all work together to heighten the erotic sense. Consciously
or unconsciously, phone sex workers translate all the modalities of experience
into audible form (396). Using only sound, they try to evoke all
the senses that are not currently in use. They describe, as sensually
as they can, what they are wearing or not wearing, how they look, feel,
smell. Their callers then use their imaginations to create a mental image
of their physical bodies that is complete enough to be erotically stimulating.
Physically separated lovers have of course been doing the same with telephones
for decades; the only innovation in phone sex is that now it is being
done with strangers whose physical bodies are not, and almost certainly
never will be, available to the caller. Phone sex workers bodies
are almost purely audible the telephone version of the legible
body. Their bodies too, like the bodies of e-mail friends, are textually
mediated. The only difference is that the text is oral rather than written.
And yes, the audible body is more physical
than the written body constructed over chat lines and other internet connections
even more physical than live sex chat, where internet subscribers
watch a live streaming video feed of a naked woman or man or couple on
a bed splayed out around a computer keyboard, which she or he or they
use to communicate with the legible bodies watching her/him/them. They
say do this or that, and the person or people on the screen do(es) this
or that, meanwhile writing this or that in response. No sound so
yes, phone sex is probably more physical. But what is being compressed
in all of these technologies, the telephone and the purely textual chat
line and the live video chat line, is still bodies. What is being constructed
is still a virtual image of the body.
This may seem to have taken us rather far afield from translation, but the scattered, disembodied subjectivities of hypercyborg translators are finally not so different from the extended audible cyborg body of Stephen Hawking. Who translates when the human sits at the computer, and fingers move on keyboards, and words appear on the screen? Sometimes the words appearing on the screen were keyed by the person sitting there; other times they were keyed by someone halfway around the world, on a translator list or chat line. Where does the body of the human translator end and the body of the machine, or of other translators, begin? As long as we think of physical bodies, the human translator and the computer are separate, distinct, radically different entities, the one flesh and blood, able to talk and learn and love, the other metal and plastic. But our cyborg bodies flow easily across such artificial boundaries, fusing and splitting with other cyborg bodies, exchanging and compressing (dis)embodied data. In this perspective, computers arent tools; they are an integral part of our cyborg selves.
Street scene. An inner-city sidewalk.
Trash, leaves, debris.
Zoom on a building recess in deep shadow
with three or four homeless people huddled in blankets and torn sleeping
bags. Slow pan.
Take Julian. Julian was a marketing director
at a large multiplanetary manufacturing company. His company provided
him with the latest in machine translation technology, cutting-edge software
for the mainframe as well as various language-specific portable devices
for ear or pocket use. But these were cumbersome in Julian's work, which
required that he be able to operate simultaneously in three or four languages.
Some of the languages he dealt with, especially on mining planets, were
simply unavailable in machine form, and he was forced to rely on living
interpreters. This was not only awkward and unreliable, it was extremely
costly. Dealers filing customers' warranty claims would blame machine
breakdowns on machine translator glitches; Julian knew for a fact that
many of them regularly jammed translations, precisely in order to avoid
Neurotranslator implant technology seemed
like the perfect solution to the problem. Julian and the other marketers
would in effect become translator cyborgs, mastering dozens of languages
almost instantaneously, never having to rely on machine translators or
living interpreters again, never having to worry about translation jammers.
The company funded implant surgery for 1225 men and women in various marketing
divisions on thirteen planets. The results exceeded all expectations.
Dozens of new markets were discovered and successfully tapped. Hundreds
of delivery glitches were ironed out. Warranty claims dropped to an all-time
Within the first year after implant,
two of the marketers were telekinetic (and claimed to be gods); fourteen
were schizophrenic; three, including Julian, were catatonic. The class
action suit filed against the employer and the implant manufacturer by
all nineteen former employees' dependents is still pending; the defense
lawyers have argued that there is no provable connection between the implant
surgery and the subsequent etiology of their mental disorders. They claim,
in effect, that all nineteen highly respected employees would have developed
mental problems regardless of the surgical procedures which they themselves
elected to undergo.
Plushly appointed doctor's office. Leather
armchairs, massive carved mahogany desk, cherry bookcases lined with medical
books, framed diplomas on one wall.
A handsome middle-aged male doctor with
gray at his temples sits behind the desk.DOCTOR
Hello. My name is Dr. Truman Killebrew,
and I'm chief medical officer at Planetary General Hospital in Kyoto.
I'm here tonight to talk to you about yes, brain implants. There
has been some unfortunate hysteria in the media recently about the dangers
of brain implants: they will drive you crazy! They will give you supernatural
powers! I'm afraid there is no substance to these reports whatsoever.
They are the work of irresponsible rumor-mongers, not the established
interplanetary medical community. There is no danger, I repeat, there
is no danger to implant wearers, provided those implants were installed
by a reputable physician licensed for brain implant procedures by the
Interplanetary Medical Association. If you have any qualms at all about
your implant, don't hesitate to check with your family doctor or local
hospital emergency room. A simple risk-free brain scan fully covered by
most health insurance plans will quickly remove any uncertainty you may
have. So please, don't listen to idle gossip. Get the facts.
Slide: THE FOLLOWING IS A PAID EDITORIAL.
Cut to Chol Altrine, sitting behind a
bridge table on a folding chair in front of a bright aquamarine backing,
which bleeds onto and across her tweed suit during her editorial.
18 living interpreters file in and stand
in a row behind her: human, theon, gantha, others, representing the 18
most politically powerful language communities in the station's broadcast
area. Slow pan, left to right. As each interpreter's head fills the screen,
the image becomes a window that tumbles across the screen and locks into
place around the outside rim. Within 20 seconds the 18 separate images
have formed a rectangular ring around the editorialist's head in the center.
The second image from the left on the bottom row has a flashing border,
signaling the interpreter whose words (in English) will be heard in this
All 18 heads nod yes to the question
they have all just interpreted.
(Simultaneous interpretation: Very well,
let us begin. There has been much discussion of late about brain implants
that adapt themselves to your brain and help you learn foreign languages
quickly and easily. It has been said, quite rightly, that this is unethical,
indeed inhuman. Never mind whether these translator implants will drive
you insane; only an insane society would invent such a device. And a human
being who allows a doctor to install one of these implants in his brain
is every bit as insane as his society whether the implant "works"
as it is supposed to, or fails.
But does this judgment go far enough?
No it does not.
The time has come to condemn the insanity
of all human-machine interfaces, everything that turns human beings into
cyborgs: not just neurotranslator implants, but contraceptive and diet
and tranquilizer implants as well; not just implanted translation devices
but all machine translation; not just computer chips but the use of any
technological device to enhance our senses, our memories, our reproductive
systems, our immune systems, our communication skills. For once let us
take the bold step; let us have the courage of our convictions! Let us
rid ourselves of heating and cooling; of lighting and plumbing; of writing
and counting. In fact, let us at last be bold enough to rid ourselves
of television itself. Let
A commotion interrupts the speaker. One
of her interpreters has raised his voice and deliberately drowned her
out. Burly station security moves to expel the disruptor, but Chol Altrine
motions them away and signals for the interpreter to continue. She sits
quiet mid-screen while the other interpreters improvise, each picking
another interpreter whose language he or she knows and interpreting it
into a second, which is then picked up and interpreted into a third, and
so on. Soon all 18 interpreters are speaking, but it is impossible to
tell which is the "speaker." Chol Altrine surveys the scene
with equanimity, even perhaps a slight smile of satisfaction.
Gradually the interpreters give up and
walk off, vacating their little on-screen windows. Finally only the faintly
smiling face of Chol Altrine is left mid-screen, surrounded by 18 empty
blue windows, one still with its flashing border.
Fade to black.
The author's study.
The author at his desk, back to the camera,
typing at his computer. Zoom on computer screen a laptop nestled
in amongst the keyboard and diskettes and cables for his desktop, an older
computer apparently no longer in use.
AUTHOR (the words appearing on-screen
as he types)