True to form, J.M. Coetzee plays games with the reader in his latest novel, El Polaco. At the time of writing still only available in Spanish translation, this is the story of a Polish pianist told by the barcelonesa Beatriz, the object of the pianist’s desire. The name Beatriz recalls, very deliberately, Dante’s Beatrice, and one could do worse than to read the story as a modern-day travesty of Dante’s high-strung Vita nuova, leading from pathos to bathos. If the ageing pianist persists in acting like a Dante, Beatriz seems both amused and abhorred by his advances. What drives the story forward, not least, is her ambivalence: rather than shutting him off, she continues to interact with Witold – to begin with passive-aggressively, but gradually more sincerely.

In minimalist fashion, Coetzee manages at almost every turn of the page to complicate an apparently simple story. He re-uses the technique of numbered passages that he previously employed in In the Heart of the Country and Diary of a Bad Year, as though to underscore that we as readers are participating in a textual game rather than a fleshed-out fiction. His writing, as Ben Etherington has discussed, has always had a »shallow amplitude,« and perhaps nowhere more so than here.Footnote 1 By this phrase, Etherington is referring to the sparseness of »social and physical backgrounds« of Coetzee’s novels, a point that speaks to the idea I wish to present here: that the future of literature might lie in the vanishing of the text, or rather, its transformation into an absent core that generates a potentially endless proliferation of other texts.Footnote 2 Less is here quite literally more.

»Shallow amplitude,« of course, is a polite way of saying that Coetzee provides little of what many readers want from literature: the thick description of lived life with its web of relations and sensuous intensities. Something more than shallow amplitude is however afoot in El Polaco. This is a novel where so much of the action – let’s call it that – has to do with the mentioning of or relating to texts that are not quite »there,« that is, legible in the novel itself. This may seem trivial. Post-Barthes, post-Kristeva, post-Genette, we are well versed in the workings of intertextuality. No text can work without this seed-bed of previous texts, and Vita nuova, as mentioned, haunts El Polaco throughout. But my point differs from the retrospective gaze of intertextual analysis. Rather, I am interested in how contemporary fictions seem to anticipate their own malleability and translingual manipulability in an age of boundless loquacity. In clearer terms, it is the productivity of commentary, translation and indeed chatter in all their verbal-visual, streamed, social-media and AI manifestations that I see as a backdrop to what El Polaco but also other contemporary literary works are doing. I will speak of the literary response to that predicament as a strategy of textual evasion.

What looms behind is of course the threat to literature’s relevance. If the established, codex format of the printed book remains with us even today, in the digitised 2020s, and if new novels are still published widely and in great numbers (490 in France between August and October 2022, for example),Footnote 3 the symbolic value of what is conventionally meant by »literature« is in decline in many societies. Connected to such a decline is our current state of semiotic overload and the scarcity of attention it entails. If the individual subject’s experience of semiosis in much of the world today is one of constant overflow, literary culture’s inherent claim that only some few texts are worthy of attention in a public sphere still often assumed to be »national« seems less and less viable. And if we widen the scope to cover »world literature« in the open-ended, Damroschian sense of works in transnational circulation, the restricted conception of literature becomes even harder to sustain. There are good reasons for this: the Eurocentrism and »Eurochronology« of literature as it was conceived in the nineteenth and twentieth centuries has reached a dead end – hence the importance of world literature as a field of enquiry.Footnote 4 But this does not spell the end of canonicity as such. It’s just that the wider the scope, the more diversified and thinner the canon becomes. Or rather, there is simultaneously a diversification of discrete canonical constellations and a narrowing of what Damrosch has called the »hypercanon.«Footnote 5 Today, if we exclude contemporary writers, Shakespeare, Austen, Proust, Borges, Achebe, Morrison, possibly Dante, possibly Ibsen can claim global recognition. Not many more. This is a heavily Western list (minus Achebe): old-world literature rather than world literature. And even those names might be stretching it. If name-recognition among the greatest number of people in the greatest number of countries is what counts, then only Shakespeare remains. In the meantime, the tidal wave of semiosis presses on, swamping literature as a publicly recognised cultural value, for good or for bad.

In scholarship, this situation was prefigured by Franco Moretti in his seminal article »Conjectures on World Literature,« where he famously coined »distant reading« as the solution to managing the mass of printed matter confronting the researcher wanting to move beyond »the canonical fraction« that traditionally staked out the boundaries of literary studies.Footnote 6 Tongue in cheek, he suggested that »we know how to read texts, now let’s learn how not to read them.«Footnote 7 He was talking, we should note, mainly about novels published in the nineteenth century. In that respect, semiotic overload is nothing new, it has just accelerated.

Moretti’s trajectory since then has been remarkably consistent with that initial formulation. In a fully logical development, he turned to computational literary studies (CLS) at the Stanford Literary Lab, where distant reading at a completely new and different level was made possible. As he writes in pamphlet 15, this activity should not even be confused with reading:

What is at stake is not reading, it’s the continuity between reading and (a certain kind of) knowledge. I read books; but when I work in the Literary Lab they’re not the basis of my work. Corpora are; ideally, those 200,000 novels. Here, size is again crucial, because a corpus is not »just like a text, only more of it.« A text is a »communicative event«: written by someone, in specific circumstances, to convey a specific meaning. But no one writes corpora; they are not »communicative events« – they are not events at all, they are artificial objects created by a researcher. A text is meant to address us; to »speak« to us. Corpora don’t speak to us; which is to say, they have no meaning in the usual sense of the word.Footnote 8

This is plain speaking: what interests Moretti as a researcher are the objects of knowledge that can be constructed through corpora. Reading, and the meaning it can provide, seems to be relegated here to the domain of private pleasures.

With the advent of ChatGPT, however, the game might be changing: not only is AI already able to produce marketable verbal narratives, but we can also outsource the labour of reading to AI.Footnote 9 Indeed, by one description, AI is little more than a machine for reading, working at a pace and scale inconceivable to humans. Trawling enormous corpora, it provides not just quantitative information but indeed »meaning« of some kind, however off-kilter it may prove to be. Out of the corpus, new texts emerge: intertextuality 2.0. One future of literature might therefore be as background noise – a textual engine humming in the background, unencumbered by the intervention of a human readerly subjectivity. The anti-subjectivist formulations of intertextuality by Kristeva and Barthes in the 1960s would finally have reached full realisation.Footnote 10

As we have learnt since the launch of ChatGPT, this is not merely a bizarre joke. AI’s increasingly powerful technology of data management will eventually be self-sustaining and perhaps also threatening to humanity – although hardly in the linear fashion imagined by some of the doom-sayers. Regardless, we seem to have entered a stage where the »textual condition« of which Jerome McGann once wisely wrote, is no longer just the transformations that happen to the text in its sequence of editorial mediations, but the condition for which texts are designed and to which they are destined.Footnote 11

But where AI will lead us is not predetermined. In a dialectical reversal, the evacuation of human subjectivity in the digital era could also result in its revalorisation. This is where literature, as a qualified human textual practice, can strike back. Old forms of writing and reading could be revived precisely because they are the mark of a non-digital human domain of meaning-making. There is something exquisitely frugal about Gutenberg-style literature – black letters on white paper – that can count as a value compared to the bloated multimedial excess of digital technologies. Taking a slightly different tack, as I will discuss later, we might also be witnessing an oralisation of textuality that could work in a humanising fashion to democratic and decolonising ends. A line of defence might therefore be to say that computers have their games, humans have theirs, and the two need not always meet.

The risk with such a view is that it assumes that human subjectivity could remain unaffected by the onslaught of digital technologies. However, again in a dialectical spirit, I would think of such a subjectivity not as pristine and separate from the digital, but as a synthesis of sorts, enabled by our emergent new condition. Meaning, and also readerly pleasure, deserve however to be claimed as human. AI, by contrast, mimics meaning and has no experience of either pleasure or pain.

If literature, then, has a future as a voluntaristic claiming of textuality – and specifically the game of textuality we call literature – it is all the more striking to note how contemporary writers use this resource to the ends of textual evasion. Could it be that such hide-and-seek is what lies beyond the reach of the humourless (but sometimes comical) positivism of digital technology? A text that is not actually there can’t be processed by AI. Take, for example, the final section of El Polaco, where a collection of poems composed by the pianist plays a major role. As can be expected, these poems are mostly spoken about; only two of them appear in the novel. What is more, we only encounter these two poems in translation – in a novel that is itself published as a translation. »Originally« written in Polish, Beatriz pays dearly for a translator to transpose the poems into Spanish, and the very labour of physically retrieving them from Poland constitute the main focus of this section of the novel. As poems, they are therefore an absent centre around which the events making up the novel proper revolve. This is how it can sound when Beatriz, stumped by the fact that the poems are in Polish, first attempts to make sense of them:

Ella abre un programa de traducción, polaco a español, e ingresa el primer poema de los ochenta y cuatro, teniendo cuidado de incluir cada punto y trazo y cada curva. Lo que aparece después de haber apretado el botón no tiene mucho sentido. Hay tres hombres en el poema: Homero, Dante Alighieri y un vagabundo sin nombre que, con un animal a su lado – es de suponer un perro – les sigue los pasos, rondando por ciudades atestadas y pidiendo a la gente dinero.Footnote 12

In English, this could read as follows:

She opens a translation program, Polish to Spanish, and enters the first poem of the eighty-four, taking care to include every dot and stroke and every curve. What appears after you have pressed the button does not make much sense. There are three men in the poem: Homer, Dante Alighieri, and a nameless tramp who, with an animal at his side – a dog, presumably – follows in their footsteps, wandering through crowded cities and asking people for money.Footnote 13

Note that these last words are not Coetzee’s, but a version produced by Google translate – AI in action. Stylistically, it is not obviously bad, except for the »you have,« which should rather have been »she has.« By cheekily substituting this translation for the English version of The Pole, which will only be made available in September 2023, I am underscoring both my own point about textual evasion and the rationale of Coetzee’s publication strategy, which issues a challenge to Anglophone hegemony. Most importantly: where do we locate, and to whom do we attribute, the text of El Polaco? The productivity of evasion can be noted here at two distinct ontological levels. The Spanish words above come not from Coetzee but from the real-life translator Mariana Dimópulos, whereas the passage itself thematises the translation of poems that only exist as a fictional conceit. Having »pressed the button,« what materialises makes little sense to Beatriz. It is the disappointing result of the machine translation that leads her to approach an actual translator. Towards the end, poems numbers 2 and 20 in the sequence also appear on the page, with the remark that they are in Spanish, without the rhyme.

As we can see, translation looms large in the novel, as though to underscore that there never is just one text locatable in El Polaco. Notably, all of its many instances of what Robert Stockhammer calls »lingualism« – an intensification or marking of linguistic difference – will be rendered differently not just in Coetzee’s English-language »original,« but in subsequent translations of the novel.Footnote 14 What we find in El Polaco, then, is a novel that meta-textually thematises the productivity of literature, understood both as works of writing and as a system of publication, translation and reception, yet it does so around an absent core. If I revisit a term I’ve used elsewhere, Coetzee has deftly designed and curated El Polaco as an expandable »textual zone,« in keeping with his practice in earlier works such as Summertime.Footnote 15

But it’s not just Coetzee. Another celebrated contemporary novel, Mohamed Mbougar Sarr’s Goncourt-winning La plus secrète mémoire des hommes, raises the political stakes of textual evasion. Here, the entire architecture of the story is built around an impossibly magnificent, forgotten masterpiece from the 1930s by an elusive Senegalese writer, T.C. Elimane. Called Le labyrinthe de l’inhumain, any encounter with this clandestinely circulating text – only a few copies remain – changes the lives of its readers, as in the case of novel’s narrator, Diégane. We readers of Sarr’s novel never get to read Le labyrinthe, of course, but we learn that it was published in 1938 to great acclaim, only to provoke tremendous controversy immediately afterwards. Critics doubted whether it »really« could have been written by an African, and, even more damningly, Elimane was accused of plagiarism, leading to his mysterious disappearance.

The story alludes to the fate of the Malian writer Yambo Ouologuem and his novel Le devoir de violence. Published in 1968, this novel was indeed accused of plagiarism – the bad form of intertextuality – at the very moment when »intertextuality« was being coined as a concept.Footnote 16 In the twists and turns of Sarr’s novel, Elimane’s fate becomes an allegory of the missing African text in the modern conception of literature and, indeed, its intertextuality. By building a story where characters to a large extent talk about and around the fictional masterpiece, Sarr never lets his readers know if this cult of literature – inspired by Roberto Bolaño, among others – is sincere or ironic. The reason for this, in my reading, has precisely to do with the still equivocal positioning of African and black writers in the »world republic of letters.«Footnote 17 Sarr’s own performance of textual evasion, however, turns this equivocation to its own advantage by distributing the narrative across three continents and ending up in Senegal, thereby ultimately provincialising the assumed centrality of France and Europe in the world republic. Even more to the point, one could offer a historicising, bifurcated reading of the story: Elimane’s Le labyrinthe stands in for old-world literature, an institution to which African writers weren’t admitted, even as it nurtured »l’illusion empoisonnée de notre élévation symbolique.«Footnote 18La plus secrète mémoire, by contrast, is literature reinvented for our transcultural, postcolonial and digitised age.

Such a displacement of old-world literature is evident also in my final example, the South African-Scottish writer Zoë Wicomb’s novel Still Life. In what supposedly is a historical fiction about the Scottish poet and abolitionist Thomas Pringle (1789–1834), Wicomb enacts, in an intensively intertextual manner, a form of evasion that privileges instead those on the margins of the story of Pringle’s life. The story begins in an indeterminate space, a »pondok« or shack of some kind where ghostly figures are huddling. What we hear is the voice of the narrator mulling over what she should do with them. But gradually, a dialogue develops. The figures speak back to her, especially the former slave Mary Prince (whose autobiography Pringle edited), and expect to be returned to memory: »So very little we ask of you, nothing more than allowing us to be, setting us free. […] But look, you’re free to improvise, to use your own fancy words.«Footnote 19 If Pringle, ostensibly, occupies centre-stage in the novel, it is in other words those he wrote about or collaborated with who come into focus here. Even as Wicomb intertextually absorbs Pringle’s work (and also that of other writers, such as Virginia Woolf), the novel displaces the authority of old-world literature on behalf of the once enslaved and colonised, the truly absent texts in literary history.

All three of these novels can be described as garrulous, but in a controlled way: the talking about, as well as the translations, stand in for the absent texts. Do we encounter here a form of oralisation that opens literature to a dialogical, decolonial mode of writing? One that employs the excessive textuality of the digital era to its own ends? If so, does this mean that a certain tyranny of textuality, which logocentrically fetishised the singular text as immutable and thereby maintained a given Eurocentric order of knowledge and value, has reached its logical end? Or is the real story of the AI era the exact opposite: that it maintains the order of things? There is after all a »coloniality of data« that is triply damaging to the extent that it excludes, appropriates and exploits all at once.Footnote 20 The operative logic of AI is profoundly extractivist, which means that it mines what is already available, thereby reproducing and aggravating imbalances of knowledge and cultural value. In that respect a negative dialectic of literature in our rapidly evolving digital age might prove to be crucial. Working as it does around the empty spaces, textual evasion could be a tool for planetary democratic survival.