[Cover Feature] Disintegration of Language: A Translator’s Self-defense in the Era of AI
The task of Hermes—that is what Olga Tokarczuk calls the work of a translator. Hermes is the Greek god who flies through the sky in winged sandals, wearing a winged helmet and carrying a staff entwined by two serpents. He is distinguished by his ability to cross over from heaven to the underworld, between the world of gods and the world of humans, delivering the gods’ will to men. He is one who crosses over, the one who delivers. A translator is like Hermes in that respect, save for the lack of winged sandals, helmet, and staff. What she has instead is a stooped back, stiff shoulders, and flattened buttocks. The body of a translator, who sits hunched in front of a manuscript for extended periods of time, is her tool. When I picture the translators I admire, the image that comes to my mind is people with long-enduring buttocks, not wings; people who carry words with caution, as though they were bricks, instead of moving them all at once. If asked the secret to their crossing over, these translators would doubtless say time. Time expended to translate a manuscript. Time expended—what an uncompetitive weapon in the era of AI.
They say that in the near future, translation will be the first profession to disappear. A translator’s time seems fragile in a world that demands maximum results with minimum cost and time. Many predict that translators working for subsistence will be replaced by AI, and that only a handful will participate as supervisors. The outlook need not be so gloomy, of course, but there will come a moment when translators must prove their competitiveness against AI.
When everyone was talking about ChatGPT, the first question that came to my mind was: why should humans translate books when AI is faster, cheaper, and more accurate? From time to time, I like to compare a sentence I translated with a Google translation. Google Translate is inadequate for translating long sentences and literary expressions, but on occasion, it generates surprisingly accurate results, filling me with an odd sense of shame as a professional translator. I am led to wonder if my translation isn’t swayed by my own personal interpretation, and if machine translation, which interprets a text in an objective, statistical manner, may not expand the possibility of literary translation.
I once imagined: I am at a publisher’s office, sitting side by side with an AI translator. We are translator candidates, competing for the same piece of work (in reality, of course, a translator is not selected in this way). To be chosen, I must make my case as to why I am better suited for the job than the AI translator. In what ways, as a human, am I more capable of successfully performing the task?
To answer the question, one must first study the principle of AI translation. Recent AI translation tools use Neural Machine Translation (NMT) technology to translate languages sentence by sentence. Neural network systems consist of an input layer and an output layer. When a sentence is input into the system, it outputs the coordinate values of words, syntax, word order, and so on by understanding the context through deep learning. The key is to acquire as much data as possible and input high-quality corpora.1 Simply put, AI translation involves statistics, probability, and calculation. The pros of AI translation include quick handling of large amounts of texts. This method involves a spontaneous and direct movement from one language to another, and is a way of expressing a one-to-one relationship that binds two languages into one. In other words, it is a return to simple language.
Is simple language, then, appropriate for literary translation? In this regard, a small seed of hope begins to grow in the mind of the human translator who feels small and insignificant in the face of efficiency. Literature is a fluid and complex system of words that commands specific, emotional, and connotative language, is open to various interpretations depending on the reader, and can change with the times. The first thing a human translator must do to translate intricate language is to read. Reading a work of literature is vastly more than obtaining information. The context must be identified, and further, reconstituted, during which process the reader’s imagination and subjective senses are mobilized. Thus, reading is not a passive act in which one accepts written language as it is, but an active response, a creative act, even. A translator—the first reader—seeks to translate in a creative manner through this process of reading. She ponders the meaning between the lines, and studies the context as much as she studies the text itself. This reading is a task in which translators invest as much time as they do in translating.
I studied drama in school, not literary translation. I began to translate because translating theatrical texts appealed to me. Before I start on a translation, I review all the TV shows, radio programs, and newspaper articles in which the author has appeared. I do this to find a voice: the author’s voice. I create a narrator necessary for the task of translation, constantly replaying in my mind the voice of the author, like that of an actor playing a certain role, and read the text in the way the author would breathe and see things. The narrator, of course, is not the author, having been created from my imagination and subjective senses. To be precise, the narrator is someone created within myself whose origin lies with the author.
Several years ago, I translated a couple of short texts by Marguerite Duras compiled in “Summer 80,” a collection of ten short pieces she had written for Libération between June and August of 1980. Working on the translation, I looked out every day at the Trouville Beach, where Duras had stayed while writing the book, and imagined her reading her own words to me. I was the stenographer setting down her words and breath on paper. There was a unique rhythm and intonation to her words, and I hoped that her voice had infused itself into the translated text, like a song from a foreign land unfamiliar to one’s ears. A certain hesitance around the border that keeps one from smoothly crossing over, a sense of displacement, suits the language of Duras, I thought. This, of course, is subjective interpretation and feeling—which is how I know what dangers lie in this method of interpretation. Creative translation entails the possibility of mistranslation. With each translated sentence comes continuing conflict. What seems a faithful translation to some could, to me, seem an awkward literal translation—a failure; to another, creativity could seem to be a betrayal of translation. Then there are limits imposed on my time, space, and experience. Translation is an attempt to simultaneously reach beyond a linguistic border and a translator’s limit, and something always goes missing or lost in the process. I once dreamed that the words and sentences I had missed transformed themselves into the author and tormented me. Each time I translate, I feel myself a failure and resolve to do better next time, but I’m flooded by feelings of stagnation—because I don’t know the right answer. What is a good translation? I have yet to find a sure answer to that question. I only know that the narrator I create should not be a reenactor for Duras, or an imitator of a certain language; and that the translation must be done in a language that is whole and intact. Will it be possible if I read and polish the manuscript again and again? As I lose myself in these thoughts, the clock ticks toward the deadline. I wonder if I have ever submitted a translation manuscript that is perfect. All of them are full of holes. Will the day ever come when I’ll be able to say, this time it’s perfect? What is a perfect translation, anyway?
Translator Jung Young-Mok said, “The task of the translator is not to achieve a perfect translation, but to perfect the language.”2 I think the statement is based on a beautiful and fascinating perspective that focuses not only on translation but also on the art of language as a whole. A language is not fixed to a certain text, so something entirely new can emerge as a text is translated from A language to B language; and a perspective that acknowledges this new creation, so that language may be perfected, sets the translator free. The holes and dents that occur as A and B come into conflict may become a sort of literary valley, which does not need to be laboriously filled. The valley itself can be magnificent. And if each translator creates a different valley—as each has her own language—how rich and colorful the view would be. If translation is a creative art, then its wholeness and beauty, I believe, have their source in this rich variety. Ten translators working on one text results in ten different texts, as the language we read is not simply source language, but a personal language with different histories and narratives. This language is handed down from parents to children and develops according to education, environment, encounters, circumstances, and personalities; then it undergoes progress, degeneration, and changes with time, becoming one’s unique trait. Thus, the coming together of an author’s language and a translator’s language is a conversation of sorts that can go in any number of unexpected directions, not one in which the answers can be predicted as in a conversation with AI. It is through this openness that we become aware that this world does not move in a single direction, nor is there only one aspect to it; there are so many different facets to this world, which can proceed in multiple directions.
Sometimes I try to think from an author’s stance instead of a translator’s. If my work is translated into another language, I would certainly welcome the possibility of varied interpretations. I would also be happy if the work transformed so much that it surprised me, as that would evidence the aliveness of my language. There is a life force that comes into existence only when a living entity changes and disintegrates. Yoko Tawada said, “Art must disintegrate in an artistic manner.”3 Translation, perhaps, is a process of disintegrating in an artistic way and gaining new vitality from the debris. If my words crumble away in a beautiful way, giving birth to something new, then that, too, would bring me joy.
Back in the shoes of a translator, I consider the act of crumbling in a beautiful way. If we believe that language affects thought, and that the words we write and translate ought to be imbued with the morals, ethics, and values of the day, the first thing that must be eliminated from the language of translation is discriminatory language, and particularly language that lacks gender awareness. The expression I’ve been wrestling with lately is “그녀,” the Korean word for “she.” In a Korean dictionary, “그녀” is a third-person pronoun referring to a woman previously talked about. The gender-equal dictionary compiled by the Seoul Foundation of Women & Family, however, points out that the word is used from the standpoint of a man to indicate a woman. Considering that a corresponding expression—“그남”—does not exist, we note that “그녀” does put an emphasis on a woman’s gender. Replacing every instance of “그녀” in a text with “±×,” however, may lead to confusion when a foreign language with separate male and female pronouns, such as “he/she” or “il/elle” are translated into Korean. Further, in translating sentences written by an author who lays emphasis on the narrator’s identity as a woman (for instance, in author Clarice Lispector’s works, it is important to reveal that the narrator is a woman), the term “그녀” cannot be excluded. And let’s say that there’s a story about someone whose gender is irrelevant, or someone with both masculine and feminine characteristics. In such a case, do we use “그” or “그녀”? Shouldn’t a fresh neuter pronoun be invented? None of these questions can arise in a simple language system in which two languages are placed in a one-to-one relationship. The questions become possible when language is not merely seen as a means of communication uttered through the vocal organs, but perceived as a complex and multilayered system fraught with social and cultural significance. Without such questions, the originality and creativity found in a translation language will seem nothing but mistranslation.
So, this is all there is to my self-defense as a human translator. The AI translator might already have finished and submitted the translation while I’ve been talking away about these hypotheses and theories. In the end, the winner of this competition will be determined by what the reader wants, which is the most reliable criterion. What does someone reading literature want? What does it mean to read literature?
Finally, I’ll borrow once again from Olga Tokarczuk in an effort to persuade readers of literature: “Literature is thus that particular moment when the most individual language meets the language of others.”4
With what kind of language do you want to greet this particular moment? My answer, as a reader, is clear: a language with a voice, a thinking language, a creative language, a contemplating language. And this answer is the hope, as well as the urging whip, that I hand myself—reader to translator.
Translated by Yewon Jung
Did you enjoy this article? Please rate your experience