1Vladimir Nabokov’s theory of translation has long validated the scholarly image of Nabokov as modernist aesthete. That is, his belief in the “untranslatability” of literary works foregrounds a “mystical submission to aesthetic authority” (McGurl 2011: 5). This article reframes our reading of Nabokov’s prioritization of aesthetics in light of his technical, scientific, and, as I will show, even mathematical conceptualization of linguistic properties. As such, this article contributes to recent efforts to destabilize assumptions about Nabokov’s sense of artistic autonomy by reconciling this facet of his work with the broader intellectual goals of late modernism (Blackwell 2009; White 2017). In what follows, I reframe Nabokov’s approach to language in order to recognize his unique proximity to other efforts during the American midcentury to distinguish language as a scientific object of study, including William F. Friedman’s work in military cryptography, Warren Weaver’s field of machine translation, and Roman Jakobson’s structural linguistics. Together, these efforts represent the growing influence of communication theory, which popularized a mathematical theory of communication (Shannon 1948). Clarifying the historical stakes of communication theory explains why in 1957 Nabokov published a novel about a language professor whose professional difficulties derive from his opposition to a department that had come to “believe only in speech records and other mechanical devices” and who is told, in front of his colleagues, that “the world wants a machine not a Timofey” (Nabokov 1957: 143, 161). In Pnin, Nabokov’s reference to a linguistic “machine” that “the world wants” addresses the midcentury project of rendering translation a process of mathematics. As we shall see, Nabokov’s occasional resistance to and affiliation with this contemporaneous project motivate his affirmation of an inextricable link between language’s quantifiable features and its aesthetic potential.

2Even as ideas about the mathematical principles of language took hold during the 1950s, Nabokov began work on an English translation of Pushkin’s Evgenij Onegin. In this period, he produced a poem, “On Translating ‘Eugene Onegin,’” published in the January 8, 1955 issue of the New Yorker, and an essay, “Problems of Translation: Onegin in English,” published a few months later in the Partisan Review. The article in particular established the “impossibility” of literary translation, due to the inimitable power of Pushkin’s “novel in verse,” and the necessity, Nabokov argued, for “literal” translation. With “absolute exactitude,” he wrote, literal translation adheres to “the essential pattern of the text,” that which can be “scientifically studied;” the translator preserves semantic meaning and simple meter without attempting to replicate the intricate and “untranslatable” nuances of “literary” language, such as “beautifully onomatopoeic alliterations” (Nabokov 1955c: 512). Nabokov’s theory of translation depended on the distinction between a language’s determinable “patterns” and its “untranslatable” contingencies. What Nabokov identified as the two forces that codetermine literary language amounted to a division between a linguistic property that could be enumerated and quantified and that which could not. That such a distinction, between aesthetics and mathematics, was somehow related to contemporaneous efforts to automate translation by machine would become more apparent when Nabokov eventually published his translation of the Onegin text in 1964. Indeed, it was in the “Translator’s Foreword” of the Onegin translation that Nabokov declared a perfect translation that maintains rhyme and form as “mathematically impossible;” but as for the feat of rendering the basic meaning and order of words: “This a machine can do” (see Pushkin 1975 I: xxvii, xxvi). The distinction indicated Nabokov’s developing sense that his own theory of language belonged to an era in which machines did the work of translation, but also that his theory supplemented what “a machine can do” by addressing the “mathematically impossible.”

3Only a few months after Nabokov’s 1955 publications on translation, Warren Weaver, the wartime director of the Rockefeller Foundation, penned a brief foreword to a collection meant to establish the link between language and mathematics as enabling nothing less than the birth of a scientific discipline: machine translation. The foreword began: “We are told, by those who are sensitive to all the beauties of the Russian language, that it is completely futile to try to translate the poetry of Pushkin into any other language—futile not for a computer, but futile for the most able bilingual poet” (Weaver 1955a: v). All but naming outright the decade’s most able bilingual Russian poet and translator of Pushkin then living in the United States, Weaver aptly summarized Nabokov’s position in the Partisan Review essay on the futility of literary translation, particularly with regard to “the poetry of Pushkin.” Like Nabokov, Weaver asserted his own objective of “literal” translation, noting that “no reasonable person thinks that a machine translation can ever achieve elegance and style,” adding, “Pushkin need not shudder.” Just as Nabokov had deemed aesthetic elements of literature “untranslatable,” Weaver eliminated the qualities of “elegance and style” from the goals of translation. By eschewing aesthetics, Weaver clarified the work of machine translation as the practical exchange of information. Even as Nabokov took note of the translation work “a machine can do”, Weaver’s foreword indicated that Russian literature and its ostensible untranslatability was somehow related to the questions motivating the pursuit of automatic translation. It remains to be seen how Nabokov’s translation theory, influential though it was, had become embroiled in the discourse of postwar machine translation.

4One sees Nabokov’s attempt to think about acts of translation as belonging to this technological context, not just in his translation projects of the 1950s, but more so in the concurrently written campus novel Pnin (1957). In the novel, Professor Timofey Pnin struggles to master the communication channels of academia—lecturing, teaching, gossiping—without sacrificing his humanity in the eyes of his colleagues who treat Pnin’s mistranslations as fodder for their cruelty. Nabokov thus presents translation as a problem of reconciling the human body with technical processes, as expressed, for example, in the physiognomy of Pnin’s mouth—“the larynx, the velum, the lips, the tongue”—which functions, in this fictional world, according to the basic mechanics of a machine, from the “motion” of its parts to its “production” of sound (66). Pnin recalls a similar dynamic of linguistic difficulty when he remembers the wallpaper in his childhood bedroom. Once, while ill and confined to his bed, Pnin searched for the meaning of the wallpaper’s intricate design and “persevered in the struggle” to immerse himself in its “difficulties” (24). In his description of the wallpaper, Nabokov inserts those attributes of literature that allow him to argue for its inherent untranslatability. Though the precise significance of the wallpaper continues to elude him, Pnin’s efforts are rewarded when he experiences an unanticipated secondary effect:

The foliage and the flowers, with none of the intricacies of their warp disturbed, appeared to detach themselves in one undulating body from their pale-blue background which, in its turn, lost its papery flatness and dilated in depth till the spectator’s heart almost burst in response to the expansion. (Nabokov 1957: 24)

5The impenetrability of the wallpaper overwhelms Pnin, and the “difficulty” gives way to euphoria. Thus, Pnin’s feverish episode allegorizes the reading of a literary text, from the bursting of the reader’s heart to the “papery flatness” that such a vivid reality transcends.1 His state of suffering bears out the feeling evoked by the autonomy of a work of art, a feeling Nabokov famously referred to as “aesthetic bliss” (Nabokov 1955a: 315).

6In spite of this allegorized, aesthetic scene of reading, Nabokov also portrays Pnin as engaging in a form of reading that is formally mathematic. From his bed, the feverish Pnin observes that the design of flowers and branches on his wallpaper is divided into two “planes,” that each plane features the “repetition” of “elements” in a “series” to convey the illusion of a “rational pattern;” however, the failure on young Pnin’s part to identify “what system […] governed […] the recurrence of the pattern” renders the wallpaper illegible, its coherency made “meaningless” by its unreadability (23). Aesthetically pleasing, the “system” of “patterns” also constitutes a formula that refuses to be solved. Pnin’s struggle stems from his inability to reconcile the random “reappearance” of symbols, their “repetition” “here and there,” with a coherent “recurrence” that would indicate a significance that structures the design. Pnin cannot decide if the repeated symbols on his wall are the result of a random coincidence or an intentional pattern. What would be the significance of intentionally designing a beautiful yet irresolvable message? After all, the wallpaper’s mathematics is also the source of Pnin’s aesthetic revelation, his affective response to its insolubility. Focusing on this distinction between aesthetics and mathematics, coincidence and pattern, the narrator confirms Pnin’s worry, noting “if the evil designer—the destroyer of minds, the friend of fever—had concealed the key of the pattern with such monstrous care, that key must be as precious as life itself” (23).

7The “key of the pattern” mentioned by the narrator here, or Pnin’s own attempts elsewhere to read Pushkin’s poetry as a “cryptogram” (68), indicate the debt Pnin’s version of reading owes to cryptography.2 William F. Friedman, then head of the U.S. Signals and Intelligence Service and later chief cryptologist for the National Security Agency, was even then ushering cryptography into a new era through applications of mathematical pattern recognition to military communications. His methods led to the successful decipherment of the Japanese cipher preceding America’s involvement in World War II. Friedman presented his most significant contribution in a government publication titled The Index of Coincidence and Its Applications in Cryptography (1935), which historian Ronald Clark calls “the foundation stone of the new cryptography” (Clark 1977: 77). In this work, Friedman fundamentally demonstrated how to align multiple strands of information presumably containing enciphered messages and mathematically distinguish between patterns that are merely coincidences—“brought about by chance superimposition”—and those that are “causally related” recurrences, which is to say, “the resultants of the encipherments of plain-text letters” (Friedman 1935: 11). Friedman’s theories revolutionized older methods that required expertise in multiple languages and a tremendous amount of guesswork. During World War II, a single analyst, aided by Friedman’s statistical measurement of patterns and working without knowledge of the encrypted language, could crack a code mathematically. Friedman showed that languages adhere to certain laws of probability, a fact that reveals itself, in any language, through statistically significant repetitions. The trick, moreover, entailed isolating such repetitions from the insignificant coincidences that inevitably result, like the irregular repetitions on Pnin’s wallpaper, recurring “here and there” in a random and “meaningless tangle”.

8These same cryptographic terms of pattern and coincidence also proved instrumental at the time in developing a theory of reading central to efforts in machine translation. Weaver wrote a memorandum that effectively laid the groundwork for the field in 1949. Weaver wrote that he first conceived of the possibility of automatic translation when a “distinguished mathematician” developed a method of cryptography during the war that worked regardless of the language being decoded (Weaver 1955b: 15). “[O]ne naturally wonders,” Weaver wrote, “if the problem of translation could conceivably be treated as a problem in cryptography.” To illustrate, Weaver continued, “When I look at an article in Russian I say: this is really written in English, but it has been coded in some strange symbols. I will now proceed to decode.” Weaver called this conflation of different practices his “cryptographic-translation idea” (Weaver 1955b: 18). The mathematics of cryptography led Weaver to identify in language the possibility of a “code,” a series of “strange symbols” that reduced different languages to the same essential foundations. Weaver saw the patterns calculated by cryptography as “invariant properties” that must be “independent of the language used” (Weaver 1955b: 16). Once machine translators developed the underlying mathematic logic of linguistic content, or so Weaver reasoned, the process could be automated. Translation would be the work of digital machines. Weaver’s articulation of cryptographic patterns clearly denoted the influence of Freidman. Indeed, Weaver even enlisted Friedman as the government consultant for the Georgetown demonstration, a public display of an IBM machine that automatically translated phrases from Russian to English, thus giving the field its first headlines in 1954.3 In the years Nabokov was writing Pnin, then, cryptographers and translators collaborated to popularize a conception of language as the object of statistical analysis.

9To fully elaborate their mathematical theory of language, machine translators turned to structural linguistics, thus making their work the purview of the human sciences. Cryptography assimilated all languages according to statistical patterns of individual letters, such as the frequency of the letter ‘e.’ Machine translators needed to retain the cryptographic method of statistical analysis but apply it to a linguistic unit of measurement other than the individual letter. For this purpose, machine translators borrowed the concept of the phoneme from structural linguistics. The phoneme offered a unit of measurement more amenable to the idea of universal translation because it divided language into discrete parts that could be numerically coded. Roman Jakobson, the central figure of postwar structural linguistics, was pivotal in revising the work of Ferdinand de Saussure and modifying the concept of the phoneme to achieve mathematical congruity. Shortly after the war, Jakobson began collaborating with Warren Weaver and other machine translators, and by 1950, the Rockefeller Foundation had awarded Jakobson a large grant to research the logic of linguistic structures, and in particular, “to analyze the frequency and distribution of phonemes in Russian” (Geoghegan 2011: 110). Media scholar Bernard Dionysius Geoghegan argues that the Rockefeller Foundation’s collaboration with Jakobson represented the objective of a “worldwide fraternity of scientists” (Geoghegan 2011: 102).4 The description bears a remarkable similarity to Nabokov’s own characterization of “modern scientific linguistics” in Pnin, which the narrator calls “a fraternity of phonemes” committed to the invention of “certain elaborate machines” (Nabokov 1957:10). Jakobson’s work on Russian phonemes emerges in Nabokov’s novel as a subtle reference to the emergent field of machine translation.

10It was in 1952, while addressing a joint conference of anthropologists and linguists gathering to discuss information theory, that Jakobson himself first confirmed the possibility of a “translation machine” (Jakobson 1971b: 554). Moreover, he framed the problem of translation in the same manner that Weaver had, as a matter of military cryptography:

The receiver understands the message thanks to his knowledge of the code. The position of the linguist who deciphers a language he doesn’t know is different. He tries to deduce the code from the message: thus he is not a decoder; he is what is called a cryptanalyst. The decoder is a virtual addressee of the message. The American cryptanalysts who, during the war, read the Japanese secret messages were not the addressees of these messages. (Jakobson 1971b: 560)

11In confirming the influence of Weaver and machine translation on structural linguistics, Jakobson’s remarks also indicated the continued construal of Friedman’s wartime cryptographic work as synonymous with the work of translation. Jakobson validated Weaver’s “cryptographic-translation idea,” by concluding, “Obviously, the linguist must develop the technique of cryptanalysts” (Jakobson 1971b: 560). Yet even as this exchange of phonemes and cryptographic techniques contributed to machine translation’s “golden age,” Jakobson began a collaborative translation project with Nabokov that would never come to fruition.5 Nabokov and Jakobson’s projected translation, The Song of Igor’s Campaign, was meant to reflect the cooperation of two of the more formidable Russian translators of their time, but instead, politics ended the project prematurely. “Frankly,” Nabokov wrote in a letter to Jakobson, “I cannot stomach your little trips to totalitarian countries, even if these trips are prompted merely by scientific considerations” (Nabokov 1989: 216). In mentioning Jakobson’s “scientific considerations,” Nabokov indicated his disapproval of Jakobson’s travels to Moscow as part of his research on Russian phonemes. Nabokov did not clarify the source of his political denunciation. But whether or not he condemned the militarization of postwar translation, that is, its entanglement with military cryptography, he clearly rejected its attachment to the “totalitarian” regime of the USSR that émigrés such as Nabokov and Jakobson had escaped. In the months following his letter, Nabokov distanced himself from Jakobson and turned instead to the completion of his own translation of The Song of Igor’s Campaign before also returning to his ambition of translating Pushkin’s Onegin into English.

12In addition to the cryptographic metaphors in Pnin and Nabokov’s relationship to Jakobson, Nabokov’s work on the Onegin text provides a final example of his imbrication with the rise of a mathematical understanding of language. From its earliest moments to its postwar automation by machine, the field of statistical linguistics developed in the shadow of Russian literature; however, this literary legacy was only legible, in turn, through a history of Russian mathematics. According to most histories on the subject, the landmark event in the statistical conception of language is Claude E. Shannon’s publication of “A Mathematical Theory of Information” (1948). With his essay, Shannon provided the mathematical notation system that enabled a precise calculation of communicative events. For Shannon’s mathematical notation of language to travel fluidly between different disciplines, it required a foundation built on interdisciplinary ideas about language, which often meant drawing from literary works. Shannon himself used James Joyce to make assertions about the predictability of most messages, as compared to the stylistic improbabilities of Joyce’s prose (Shannon 1948). However, it was Jakobson who, in drawing connections between Shannon’s communication theory and linguistics, saw the need to recall an earlier Russian precedent for the statistical definition of language only then being realized. In “Linguistics and Communication Theory,” Jakobson reminded his readers that the work of Shannon, and indeed all work on the statistics of language, depended on the experiments of Russian mathematician Andrey Markov (Jakobson 1971a). In 1908, Markov showed that any given sequence of words conveys certain degrees of predictability and unpredictability that produce a statistically measurable balance, a concept now known in mathematics as a Markov chain. Moreover, just as Shannon had turned to Joyce, some decades earlier Markov had collected his numerical data from none other than Pushkin’s Onegin. Pushkin’s work was, in fact, the first literary text to be subjected to mathematical analysis (Liu 2010: 51). The Onegin text stood for Markov as the exemplar of the statistical qualities of language. It was later these same qualities, drawn from the same text, that Nabokov articulated as the difference between Pushkin’s metered patterns and the unparalleled complexities of meaning that remained “untranslatable,” which he later clarified as the work of “a machine” and the “mathematically impossible.” As Nabokov set to work on his translation of Pushkin, he thus took part in a broader intellectual tradition, one consummate with the beginnings of the information age.6

13From Friedman’s military work to Weaver’s field of machine translation and Jakobson’s structural linguistics, the founding of a statistical definition of language depended on concepts drawn from cryptography. Indeed, Shannon himself saw potential in the applications of statistical solutions to secrecy systems (Shannon 1949). To follow the course set by cryptography, as its influence traversed these disparate disciplines, is to clarify why, in Pnin, a Russian language professor and translator who is told that “the world wants a machine not a Timofey” continues to be plagued by the cryptographic patterns of his childhood wallpaper (Nabokov 1957: 161). We are told that Pnin cannot approach “the lofty halls of modern scientific linguistics,” but simultaneously, he finds himself no closer to those teachers whose natural “intuition” for “their difficult and beautiful tongue” allows them “to infuse a magic knowledge” of the power of language into their students (10). Pnin, that is, cannot satisfy the cultural expectations of those who maintain a purely scientific or aesthetic view of language. Stuck between these conceptions of language—“difficult and beautiful” on one hand, “scientific” and “modern” on the other—Pnin remains the awkward representative of their necessary negotiation. His position as a teacher mirrors that of his childhood difficulties with his wallpaper. The sublime, affective experience brought on by its aesthetic qualities cannot be disassociated from its mathematical patterns. Together, they achieve a particular balance. And though this makes the wallpaper a cryptographic message that cannot be deciphered, its insolubility achieves what Nabokov offers as the aesthetic potential of fiction: a kind of mathematics, or a play of patterns, that does not give up all of its secrets. Pnin offers a fitting lesson for an age newly equipped with a statistical definition of language. Nabokov’s novel shows that somewhere between “aesthetic bliss” and pure acts of communication, literature is not reducible to either.