7. THE SPLITTING OF THE VOICE

To grasp improbability, the notion of style strikes me as quite imperfect. It is often used to point toward a layer of language detached from the real or from political friction. We speak of a “stylist,” or of an author who cultivates fine sentences, as of an aesthete embroidering motifs on the surface of language.

The notion takes on a variety of meanings in literary criticism. In using the expression “spontaneous prose” (and, in my tests, “minimalist style” and “hallucinatory prose”), I adopted a definition of style as the expression of a somewhat “general” way of writing that brings together distinct authors (without there being, fortunately, any school or manual of spontaneous prose). By contrast, the word style is often used to designate the particular “signature” of a given author. At that point, the concept begins to flirt — and sometimes to merge — with that of “voice.”

The notion of voice has the advantage of not suggesting pure formalism or a separation between writing and life. It carries a kind of aura of the real that the notion of style does not. It points to an origin of speech, yet it summons something that does not have its source in the Self. It sketches a provenance if not exterior to the subject, then at least offset from it.

The notion unfolds at least three facets:

• Living speech;
• The gap between the subject and “what speaks”;
• The stylistic signature.

Living speech, because at root, voice stands opposed to writing; it plunges its roots into orality and into the primacy of dialogue over writing in Plato. Indeed, the Phaedrus is unconsciously invoked — in the media, in popular thought, and in scientific thought — each time technological anxiety resurfaces. In that dialogue, Plato was saying in substance: writing kills living memory. Since the 2000s, we have heard the same discourse about the Internet, an anxiety that today extends to artificial intelligence as well: these “prostheses” would atrophy not only our memory but also our thought, our reasoning, our ability to focus, and so on, devitalizing the biological faculties of the brain. It is fair to say that artificial intelligence is dead speech, in the same way that writing is dead speech. It is the same paradox that runs through both human writing and artificial writing, viewed through the lens of voice. At first glance, AI seems to strip us of our voice — a living organ — replacing it with flat, inert, grainless prose.

At the same time, the notion opens a gap between the one who speaks and what speaks. Nowhere is this better expressed than in Samuel Beckett’s The Unnamable:

Where now? Who now? When now? Unquestioning. I, say I. Unbelieving. Questions, hypotheses, call them that. Keep going, going on, call that going, call that on. Can it be that one day, off it goes on, that one day I simply stayed in, in where, instead of going out, in the old way, out to spend day and night as far away as possible, it wasn’t far. Perhaps that is how it began. I can’t be bothered for anything anymore. You think you are simply resting, the better to act when the time comes, or for no reason, and you soon find yourself powerless ever to do anything again. No matter how it happened. It, say it, not knowing what. Perhaps I simply assented to an old thing. But I did nothing. I seem to speak, it is not I, about me, it is not about me.¹

Beckett drives a wedge between the subject and what speaks. Voice is no longer something one possesses by essence; it is an outward movement that flees, and that one chases to the point of dizziness. The Unnamable drags the shadow of modern literature into harsh daylight: on the reverse side of the internalization of the origin of speech exists this force of externality that refuses to let itself be forgotten. I may delude myself into believing that “I write”; in fact, “It writes.” This is what the split prose of Beckett — and of other word-artists of the twentieth century — obstinately reminds us. Artificial intelligence confirms this radical push toward the outside. AI now makes it visible: the subject is no longer the seat of the voice. It speaks before oneself, in spite of oneself. It speaks in the machine.

This is where the work begins: how can we make these scribes called Claude, Gemini, ChatGPT, or Kimi speak? How can we become the ventriloquists of AI, animating the mouth of the silicon puppet in a way that reflects the vivacity of voices? Here, I am not thinking only of general styles, which have something reductive about them (Kerouac is not merely the “spontaneous prose” label slapped onto him; Duras is not merely a “minimalist”; and so on). I am thinking of the extreme and magnificent diversity of ways of expressing oneself, the multiplicity of syntaxes, of lexical choices, of actualizations of improbability.

I cannot help thinking of what Henry David Thoreau said in Walden about the foliate forms traced by ice crystals as they form: “Even ice begins with delicate crystal leaves, as if it had flowed into moulds which the fronds of water plants have impressed on the watery mirror.”² These precious “crystal leaves” that are “impressed” on the ice: it is the perfect metaphor for what I am trying to express. Voices are imprints, varied and delicate impressions, that could take the form of a leaf to be deciphered.

The particularity of a voice does not come, in my view, from the uniqueness of the subject-self in essence, but rather from the diversity of experiences that gives each language its vocal imprint. It is the outside that gives shape to the voice; it is the real that impresses itself upon it, fashioning idiosyncratic forms within it, defining a grain.

Voice deposits a kind of mystery, a secret DNA. One spends a whole lifetime searching for it without ever putting one’s finger on it. One goes on writing, and text after text, one hopes to finally draw closer to it. And it is because one never coincides with it that one goes on writing. One speaks, one misspeaks, one fails, one fails better, and what remains is sometimes called literature.

With AI and the powerful movement of externalization it sets in motion, the relationship to voice will no longer be the same. We will henceforth have at our disposal an extremely powerful machine for — this is at least a possibility — finding or recovering the voice.

---

1. Samuel Beckett, The Unnamable, New York, Grove Press, [1953] 1958, p. 1. Translated from the French by the author. In the original: « Où maintenant? Quand maintenant? Qui maintenant? Sans me le demander. Dire je. Sans le penser. Appeler ça des questions, des hypothèses. Aller de l’avant, appeler ça aller, appeler ça de l’avant. […] J’ai l’air de parler, ce n’est pas moi, de moi, ce n’est pas de moi. » (L’innommable, Paris, Minuit, [1953] 2004, p. 7.) ↩

2. Henry David Thoreau, Walden; or, Life in the Woods, Boston, Ticknor and Fields, 1854, “Spring.” In the original French translation by L. Fabulet: « Il n’est pas jusqu’à la glace qui ne débute par de délicates feuilles de cristal, comme si elle avait coulé dans les moules que les frondes des plantes d’eau ont imprimés sur l’aquatique miroir. » (Walden, ou, la Vie dans les bois, Paris, Gallimard, coll. « L’imaginaire », p. 250.) ↩