Article body

1. Introduction

To what extent can we meaningfully apply translation theory outside of the study of translation? Can we, in other words, speak of translational elements beyond translation, of properties that can be studied with the tools of translation theory but are not limited to a translational context alone? This article examines this question from the vantage point of collaborative self-translation. I focus especially on a largely overlooked form of consecutive self-translation (Grutman 2016: 121) whereby a bilingual author outsources their work’s translation with the understanding that the hired translator’s version is to follow the original closely and an assumption of its provisional nature; the translator may or may not be contractually bound to this arrangement (the closest term proposed for this practice to date is Dasilva’s [2016] notion of semi-self-translation).

The goal of the article is twofold. First, it aims to draw attention to the use of intermediary translations as a form of collaborative self-translation and its potential as a perspective on a range of issues in Translation Studies. Secondly, to highlight one such issue, it applies a genetic editing approach (see Grutman 2016; Cordingley and Montini 2015; Munday 2013) to the changes made by the author-translator to intermediary translations in order to stretch our notion of translation to texts produced by a process other than translation. To quote van Doorslaer, “the blurring of borders is […] an ongoing process between TS and adjacent research fields that deal with other text-modifying practices […]” that “[…] will inevitably affect the seemingly never-ending debate on the name of the discipline” (2020: 10). I refer to textual elements of such “other text-modifying practices” that yield to the tools of translation theory collectively as translational beyond translation.[1] This establishes the interdisciplinary relevance of translation in an inductive manner in an attempt to regain control over the extension of the field’s boundaries. Shifting the emphasis beyond translation acknowledges the fact that certain elements thus analysed may be fruitfully ascribed to cognitive constraints that translation happens to share with other forms of writing (Kotze 2020).

Self-translation is a contested ground for Translation Studies, a phenomenon where translation spills over into other literary processes, not to say its “other.” There is little argument about what “faithful” self-translation is about—it is when a self-translator behaves like a third-party translator. Self-translators can, indeed, surprise researchers by how closely they, at times, follow the original (Meng and Cai 2020; Anselmi 2018: 2-3; Edel-Roy 2010; Molnar 1995).

Despite these documented exceptions, many self-translators do not. Enter rewriting, an umbrella term for many of the strategies the self-translator uses to “endow her work with an aura of authenticity that is rarely, if ever, granted to ‘standard’ translations” (Grutman and Van Bolderen 2014: 324). The by now sizable body of self-translation scholarship reflects an interdisciplinary tension coming from the need to take a stand on whether self-translation is ultimately a form of translation (Canli 2018; Grutman 2009: 257; Tanqueiro 2009; Santoyo 2004; Popovič 1976: 19), or a manifestation of literary bilingualism resulting in two equi-valent, rather than equivalent, original texts (Sindicic-Sabljo 2011; Oustinoff 2001). Susan Bassnett perhaps goes the furthest to argue that the very concept of self-translation “is misleading and unnecessary. All these writers negotiate between languages in different ways, and for different reasons” (2013: 13).

Although I find the concept of self-translation useful, I follow Bassnett’s stance in viewing it as a ground of negotiation among several literary processes, one of which is translation. I thus consider self-translation to be a continuum of choices applied, in turn, to each translation unit of the original. At one end of this continuum is literal translation, defined here as translation that aims to be as close to word-for-word as fluent literary expression allows (since the “left-side player in the long and frustrating game of squash between ‘literal’ and ‘free’ doesn’t really exist,”; Bellos 2011: 116). At the other end is original writing, that is not translation. Self-translators’ choices are not binary but can be defined, rather, as points on that continuum. This is not to imply that allograph translations, in contrast, must be faithful or literal. Any translator is technically operating within the same continuum; yet the generally accepted legitimacy of choices on its “free” end in self-translation makes for a particularly rich context for an exploration of the grey area between free translation and non-translation (Halverson 2010: 378-384; Chesterman and Arrojo 2000: 153).

Following along our continuum, we first pass by several acceptable options of free translation. Yet, eventually we must enter the zone of what most self-translation scholars would see as rewriting, something other than translation. This is a point usually defined by the given choice’s acceptability were it to be made by a third-party translator. Quod licet Iovi, non licet bovi. Or, to adapt the adage, what’s permitted to the author is not permitted to the translator. Past this point, we can potentially speak of rewriting. Many choices considered off bounds to a translator, however, remain, as I argue, fundamentally translational in nature. Considering such choices systematically allow for a novel point of view on the tensions in self-translation and creates a precedent where translation theory clarifies, inductively, something that is not translation. It thus contributes to interdisciplinary applications of translation theory while calling for a clearer distinction between such applications as a genuine extension of translation theory and translation used metaphorically.

Figure 1

Freedom in translation continuum in translation studies

Freedom in translation continuum in translation studies

-> See the list of figures

2. Allograph Intermediary Translations as an Object of Study

The notion of collaboration in translation, and especially in self-translation, has only recently started to gain critical attention.[2] Where it comes to collaborative processes whereby the authors work, often consecutively, on a pre-existing allograph translation, revising it extensively, the field is still close to uncharted. Hersant refers to this collaborative self-translational strategy as “authorial appropriation” (2017: 101-102), Dasilva as “semi-self-translation” (2016: 29-32) and Peñalver as “autotraducción parcialmente autorial revisora” (2011: 206). However, existing case studies tend to treat each case primarily as a stand-alone exception, while attempts to identify this kind of collaborative self-translation as a distinct phenomenon have not yet focused on the theoretical potential of a genetic editing approach.[3] Nabokov is perhaps the best-known example. His post-1940s working model involved “malleable” or “docile” translators who were then expected to “provide him with a literal translation, which the writer abundantly amended” (Anokhina 2017: 114; Grayson 2000: 989).[4] The writer kept his right to the last word, quite literally.[5] Joseph Brodsky tried similar approaches involving what one embittered translator whom Brodsky failed to warn that his work would not be the final translation, calls “полуфабрикаты,” that is “semifinished products” (quoted in Ishov 2020: 74; see Rulyova 2020), and so did Paul Celan (Grutman 2016; Weissmann 2003; Walcott 1998).[6] In all these cases, we deal with a setting where a bilingual author-translator commissions an accurate translation they never intend to be publishable, produced without their involvement and viewed by them as a largely discardable first draft.

In some cases, we have direct evidence that the intermediary translator was explicitly instructed to follow what Nabokov’s translator calls “the servile path” (Scammel 2001: n.p.). Nabokov himself once tried to enlist a translator with a request for a “точный и грамотный перевод, который вероятно потом раздраконю” (“an exact and competent translation that I will then probably dragonize,” quoted in Anokhina 2015: 204). Romain Gary, who rewrote Camilla Sykes’ “plain but by no means incompetent version” of La Danse de Gengis Cohn “top to bottom, adding chapters, Yiddish jokes and topical references” (Bellos 2010: 335), is likely to have shared this expectation. Indeed, Dasilva (2016) goes as far as to consider working with a malleable translator, such as a friend or a relative, competent but too dependent to venture away from the original or to mind extensive changes to their work, an integral part of his account of semi-self-translation. Hence the peculiar role played by the intermediary translation and the unusual power relations between the parties.[7] We have no precise term for the intermediary translation in such cases, but the interlinear is perhaps the closest translation practice to what is at stake (Tyšš and Gromová 2020; Witt 2017); no alternative term comes any closer to capturing the role of the intermediary translation or offering an applicable theoretical framework.

The question of theoretical context is especially delicate. On the one hand, the assumption of subsequent editing makes it justifiable to regard such arrangements as self-translation that involves an allograph translation rather than as translation edited by the author (Dasilva 2017). On the other hand, the proportion of the translator’s work in the published text can be considerable (although in some documented cases it may be as little as 20%; see Anokhina 2017: 114). In practice, the resulting work may be published as a translation revised by the author, a new original or a self-translation with a translator’s assistance.

The study of authors’ use of provisional intermediary translations offers an exceptionally advantageous setting for the kind of thought experiment proposed in this article. These self-translators vote with their feet to confirm that relying on, and paying for, the additional labour of an allograph translation is worthwhile. This is emphatically unrelated to any need of assistance in the target language; these writers are fully competent bilinguals with experience in solo self-translation and could have self-translated their work themselves. We must therefore conclude that they regard the translator’s involvement as either outsourcing an aspect of their would-be self-translation process or adjusting it.

The self-translator’s process, indeed, changes radically: to translate like a third-party translator, the self-translator now needs to do nothing at all, that is let the actual third-party translators’ version stand for the given unit of translation. Translation thus remains the default choice for any given unit of translation, material and accessible both to the self-translator and (conveniently for us) the researcher. Deviations from it must now be framed as a series of deliberate and tangible decisions to reject each suggestion made by the hired translator. In this sense, the self-translator’s process is brought artificially closer to Toury’s monitor model (1995/2012: 225) whereby the translator first translates according to formal correspondences by default and only then, in their capacity as reader of that default version, introduces changes. Their range and the motivations behind them are naturally different when that reader is a self-translator. Unlike a third-party translator, a self-translator does not necessarily translate according to formal correspondences by default (Grutman and Van Bolderen 2014); when they bring in a hired hand instructed to stick to the original, they do.

Short of a controlled experiment with think-aloud protocols, this is perhaps the closest we can get to a simplified model of the self-translator’s differential choices in terms of our “word-for-word translation-original work” continuum. The writer works with an embodied (rather than mental) representation of what their original work may look like in the target language were it to be translated by a third-party translator. They can now either give flesh to this translational vision by simply leaving it as is (except, perhaps, minor improvements), or else opt for something else, local choice by local choice.

This working model is rigged in favour of translation, making a self-translator hypothetically less likely to introduce drastic changes to the text.[8] However, whenever they reject the translational vision, this triggers a self-translational, editorial, or even writing process. It also allows the self-translator to work, when they choose to, within the target language, which would have been impossible in solo self-translation. In other words, this kind of self-translation relies on several text-modifying processes from all over the continuum.

3. Romain Gary’s Collaborative Self-Translation Practice

Romain Gary made literary history when he hoodwinked the French literary establishment into giving the coveted Goncourt prize to him twice under different names. This unprecedented and illegal distinction attested, as it were, to the validity of the writer’s multiple lives. Born in Vilnius, Gary spoke Russian at home, Polish at school, and was a member of a largely Yiddish-speaking community. Curiously, though, while all these languages feature in his novels, he chose to write, instead, in French, which he learned at the age of fourteen, and in English, which he acquired as a war pilot stationed in London (later the native language of both his wives). That is also where he wrote, between raids, his first French novel, Education européenne. As Gary penned most of his English novels, he was a diplomat representing France in the US, and a commander of the French Legion of Honour (see Bellos 2010: 428-434).

Sustaining a triple existence as one English and two French novelists—not to mention a diplomat, war hero, celebrity spouse and filmmaker—contributed to Gary’s habit of reworking and recycling his work. Most of his novels were translated from French into English and vice versa, many with no mention of a foreign-language original and under a new title. The success of Gary’s strategy of “opaque self-translation” (Dasilva 2016) was so complete that until Anissimov (2006) and Bellos (2010), biographers assumed that he self-translated all his work. This seems to reflect Gary’s sense of ownership over his writing process: whichever text came first, and whatever its genesis, it is the “original” work of Romain Gary.

A closer look, however, reveals a much more complex reality, with the author’s involvement varying from solo self-translation to working with a ghost translator. Translators worked on at least three novels: La Danse de Gengis Cohn as it was translated into English (Bellos 2010: 472), Lady L as it was translated into French (Anissimov 2006: 303) and The Gasp (Grutman 2016: 125). Manuscripts of another novel, White Dog/Chien blanc also reveal the presence of a ghost translator (Kelbert 2015: 154-159; see Bellos 2010: 430). Keen to conceal the presence of a translator in most cases, Gary invented one in others. Notably, he attributed his English self-translation of La Promesse de l’aube to John Markham Beach, one of his many pseudonyms, who also “translated” The Talent Scout (in fact, the novel’s English original; Bellos 2010: 172); Joseph Barnes’ English translation of L’homme à la colombe was unpublished, but Gary planned to present it as the original and mark the French original as a translation from “l’américain” (Bellos 2010: 251).

Much is still unknown about the nature of these collaborations, real or invented. Did the initiative to involve translators come from Gary or his publisher, and what was the relationship between Gary and his translators? While the publisher signed the contracts, Gary, scrupulous in financial terms if not in giving credit, doubled the translator’s wage on at least one occasion (Anissimov 2006: 303). This suggests a professional arrangement both the publisher and the writer must have been comfortable with. Manuscript revisions on the transcripts indicate that Gary reworked a finished manuscript post-factum, presumably alone given his travelling schedule, although there were exceptions when he worked with the translator in person, notably on the translation of Lady L (Anissimov 2006: 303; see Anokhina 2015: 204 on Nabokov).

Gary’s translators were not normally acknowledged and the work was presented as an original. The one exception is the acknowledgement of the “assistance” of Camilla Sykes in translating La Danse de Gengis Cohn into English; in the cases of The Gasp, Lady L and Chien blanc, we are dealing with ghost translators. While we do not know whether Gary informed his translators that their version would be revised extensively without their control, the confidentiality clause in the translators’ contracts with Gallimard seems to suggest that this was the implicit, if not explicit, understanding. Ironically, Gary found himself at the other end of the stick years later when, having published under the fake identity of Emile Ajar, he found himself invisible and powerless vis-à-vis his English translator whose work he tried and mostly failed to amend (Bellos 2010: 394).

The analysis that follows is based on the surviving manuscripts (where they could be consulted) and the final published versions of three novels: Chien blanc, La Danse de Gengis Cohn and Lady L.[9] The manuscripts of Chien blanc offer the most detail on the collaboration process. They include the intermediary transcript by an unidentified ghost translator blotted with the author’s extensive corrections, as well as (for many passages) handwritten pages Gary rewrote entirely. With La Danse de Gengis Cohn, only parts of the transcript survive with relatively few corrections, while much of the material is rewritten in Gary’s hand, representing, it seems, a later stage in the revision process.[10] This material was compared to the published French originals and the published English translations. For Lady L, only the two published texts survive, perhaps because Gary worked with the translator, Jean Rosenthal, closely in person in this instance.

4. Translational Processes in and Beyond Translation in Chien blanc/White Dog

Consider how a collaborative self-translation arrangement plays out in Chien blanc (White Dog in English). This novel came out in 1970 in both languages, both as “originals,” although the manuscripts make it possible to identify French as the novel’s original language (Kelbert 2015: 153-168). The eponymous dog is a friendly German shepherd Gary allegedly took in only to discover, when it nearly mauled a delivery man, that this was a retired police dog from Alabama trained specifically to attack Black people. Mortified, Gary tries to have the foundling retrained. As the dog becomes a metaphor for all the ingrained prejudices of American society, a parallel narrative unfolds of Gary’s relationship with his wife, Jean Seberg. This characteristic passage paraphrases Jean’s dialogue with the dog’s coach, Keys:

There, you’ve got me. Là, vous me tenez. J’ai toujours aimé les bêtes, depuis que j’étais môme. C’est pour ça que j’ai choisi ce métier. Bientôt, j’aurai mon chenil à moi. Je vais m’établir à mon compte. Je suis un professionnel, un vrai. Si je réussis avec ce toutou, ça voudra dire que je suis le meilleur. Yes, ma’am. Le meilleur…

Tout cela devait se passer dans une odeur de roses. Je dois laisser derrière moi un vide extraordinaire, parce que dès que je pars, je suis immédiatement remplacé par des dizaines de bouquets de roses. Il en vient de tous les côtés. Avec des cartes de visite. C’est très flatteur de savoir que, dès que vous quittez votre ravissante épouse, un nombre impressionnant d’individus se précipitent chez les fleuristes pour tenter de remplacer le parfum envolé.

Gary 1970a: 66

First, the passage was translated into English by the invisible translator. The parts Gary kept for the published translation are underlined.

There you’ve got me. I’ve always liked animals, since I was a kid. That’s why I chose this job. Soon, I’ll have a kennel of my own. I’m going to set up on my own. I’m a real pro. If I make it with this pooch, that’ll mean I’m the best. Yes, Ma’am, the best…”

The smell of roses must have been in the air while all this happened. I must be terribly missed, because as soon as I leave I’m immediately replaced by dozens of bouquets of roses? They come from all directions. With visiting cards. It’s very flattering to know that as soon as you leave your gorgeous wife an impressive number of people rush to the florist’s and try to replace your sweet-smelling self.

Both the length and the syntax are translated competently. The subject changes in the sentence that first introduces the roses, and “le vide extraordinaire” is translated freely by “terribly missed.” We also observe a certain liberty in the translation of “parfum envolé” as “sweet-smelling self.” Overall, though, the translator clearly aims to render the source text as the sum total of literal translations of its constituent parts. The overlap between the published version of this passage and the interlinear (underlined above) reflects the default setting of translating faithfully that an intermediary translation provides. Three shorter sentences remain the unaltered contribution of the ghost translator, as well as the core of multiple other phrases. Overall, while most pages of the typescript have few, if any, sentences not written over or improved, a reasonable proportion of the published novel does belong to the hired translator’s pen.

However, Gary takes and transforms this version. Gary’s most significant changes to the intermediary translation are underlined:

There you’ve got me. I’ve always liked animals, since I was a kid. That’s why I chose this kind of work. I wanted to be a vet, but I dropped out of school in the eighth grade. Had to work. Well, I’ll soon be in business for myself. There’s not one real good kennel in the whole black community here. I feel they could do with more dogs, a lot more.And I’m a real pro with dogs. The best. Yes, Ma’am, the best. If I can make it with your friend here, that’ll prove it. Those things get around.

All this must have been happening in a wonderful smell of roses. Whenever I leave Jean alone, I am immediately replaced by bouquets of roses. Dozens of them come to fill the void, all with visiting cards, and I have estimated at various times that my flower value is about a dozen roses per pound. It is flattering and very satisfying to know that as soon as you leave your gorgeous wife alone, an impressive number of people rush to the florist’s in the admirable hope of replacing with roses your sweet-smelling self.

Gary 1970b: 68-69

The manuscripts of this passage include the translator’s transcript crossed out with a single line and Gary’s own handwritten text with multiple phrases started and rewritten as he went (Figure 1). This draft was made with close reference to the intermediary version (for example, Gary keeps, even after multiple attempts to rewrite the sentence, the translator’s elegant solution of “sweet-smelling self”). The rewritten page is similarly crossed out, presumably to indicate the existence of further revisions.

Most of the changes are additions; yet the writer does more than merely add. Closer analysis reveals different kinds of changes. Some can be interpreted as editing a translation, for example joining some of the phrases—such as the mention of visiting cards—that would otherwise sound abrupt in English. The choice of “be in business for myself” as opposed to “set up on my own” is a matter of taste; both are perfectly valid translations of “j’aurai mon chenil à moi.” In some cases, we can guess the reason for an edit. For example, the translator suggests “this job” for “ce métier,” and Gary adds a lost nuance present in the original with “this kind of work”: ‘métier’ really is not just about a job but a vocation. Similarly, syntax is restored in the phrase “All this must have been happening in a wonderful smell of roses”; here, the translator’s solution probably came across as too free to the writer. The changes in these cases are aimed at making the result a better translation.

Other changes to the intermediary translation—and consequently to the original—are ones no third-party translator, however invested in a creative vision of the profession (Perteghella and Loffredo 2006), would be comfortable to make; these are a matter of a different kind of editing, similar to revising an original text. The information contained in a unit of translation is changed or added to, such as removing the phrase “They come from all directions” or adding detail about Key’s life or an estimate of Gary’s own “flower value.” This last addition exemplifies an original writing process replacing translation—the opposite end of our continuum of choices.

Figure 2

Manuscript page of White Dog; Gary’s initial rewriting of the transcript

Manuscript page of White Dog; Gary’s initial rewriting of the transcript

myself. We There is not one ken kennel that you can real good kennel in all the black community here. I Very few dogs I feel they could do with more dogs, a lot more. If I unclear And I’m a real pro with dogs. The best. Yes, Ma’am, the best. If I can pull the make it with your unclear here, that will unclear prove it. It’s o you I hope you’ll recommend me to your friends… So I’ll

 All this was ha must have been happening, I am sure, in a wonderful smell of roses. I unclear guess must be seem to leave behind be (sic) an extraordinary void, whenever I leave Jean alone, for I am immediately replaced

-> See the list of figures

Several further changes, however, belong to neither translation nor original writing. These are generally seen as appropriate in self-translation under authorial licence (Grutman and Van Bolderen 2014: 324). Yet they remain deeply translational in terms of their likely motivation and the decision-making mechanism they expose. One characteristic example is the compensation strategy in adding the phrase “to fill the void.” As translations go, this is decidedly free and borders on rewriting; there is no direct analogue in the original text. Yet we can see what happened here: one key element that the hired translator happened to omit is present in both the French and in Gary’s translation: the word void. Gary obviously cares about this word and restores it. The translator, not knowing Gary’s priorities, must have considered a literal translation such as “I must be leaving behind an extraordinary void” (which Gary first attempts, see Figure 1) unidiomatic and found a smoother alternative. Gary agrees that literal translation will not do but refashions the entire passage to keep the word he wants. We can hypothesise that this led him to the added joke about his flower value, triggering an instance of original writing. The motivation behind the original change, however, is clearly translational: the author-translator is technically rewriting the phrase, but to find an equivalent for the one element that matters.

For a further example, consider the phrase “Tout cela devait se passer dans une odeur de roses” from the extract quoted above (literally “all this must have been happening in a smell of roses”). The translator tries to be literal but, presumably recognising that the phrase “in a smell of roses” sounds awkward in English, compromises by changing the subject: “The smell of roses must have been in the air while all this happened.” Gary disagrees and replaces the original word order. To make this palatable, however, he introduces an extra epithet: “All this must have been happening in a wonderful smell of roses.” Here, too, an element of apparent rewriting serves a translational purpose, enabling a more accurate translation of the remainder of the sentence.

Elsewhere in Chien blanc, the narrator locks the dog up “avec une pâtée royale” (sic; 37). There is no easily available English equivalent (the brand Royal Canin, founded a couple of years before the translation was published, was presumably still unfamiliar to Gary), but the published translation opts for “a royal helping of dog food,” clearly inspired by the French expression (30). We know nothing of the size of the dog’s meal from the original text, but this solution reveals something about Gary’s intention, namely that the treat itself was unimportant to him compared to the connotation of royal treatment. In this instance, the hired translator makes a guess that proves to correspond to the writer’s vision—their version, too, is “a royal helping of dog food,” which Gary then changes to “a royal helping of doggie food,” and then back to the translator’s original suggestion in the published version. As with the term void discussed above, Gary rewrites an element of the text to translate what mattered: the effect rather than the original’s means to achieving it (see Bassnett 2013; Grayson 2000: 990).

This last example, although it changes the meaning from a description of the food to a description of the helping, is an instance of rewriting that a translator might plausibly choose as a compensation strategy, and indeed, in this instance, does. There is little difference, in principle, between compensation strategies in our first passage and the “pâtée royale” example. Translators, too, are often compelled to sacrifice formal or semantic correspondence to translate an effect (Pym 2010: 66). The French théorie du sens or théorie interpretative de la traduction (TIT), alongside Skopos theory, have contributed significantly to the understanding that the most accurate equivalent in translation is often one that translates the function rather than the literal meaning of the words in the original (Lederer 2010; Israël 2006). However, some compensatory changes observed in the first example—adding adjectives, rehauling the structure of two sentences and removing elements that got in the way—belong to what self-translation scholarship tends to place under the umbrella concept of authorial licence (see Montini 2010: 306).

These examples push the line of what is considered appropriate for an allograph translation, and somewhere between them this line has arguably been crossed. In the same passage, Gary’s manuscripts show that he was unhappy with pooch and looking for a better way to translate toutou. He then ends the paragraph with “I hope you’ll recommend me to your friends… So I’ll.” In the published version, however, he decides on “your friend here” for “toutou” and thus replaces the other mention of friends with “Those things get around” in a further example of how the wish to translate a key element or its function can inform a clear deviation from the original, here, an added sentence. A third-party translator could not have made such an addition, nor would they have known to prioritise the word void, but the author can, and does. Authorial licence, then, can bear traces of problems of translation successfully (or not so successfully) resolved. Deviations from translation can be translational.

5. One Step Further Beyond Translation: Explicitation in Gary’s Self-Translations

Compensational rewriting may border on translation, but explicitational rewriting takes us further still towards the right side of our continuum. It has been claimed that translators tend to explicitate—to verbalise aspects of the message that, while present or potentially present, were not verbalised in the original (Klaudy 2009). There is no clear agreement on what exactly constitutes explicitation on the level of grammar or content, although the general consensus revolves around variants of “syntactic expansion (for example “the girl I saw” vs. “the girl that I saw”) and the provision of lexical information that is considered common knowledge to ST users but not to TT users (for example “‘Huesca’ might become ‘the city of Huesca in northern Spain,’” Pym 2011: 81). Blum-Kulka’s explicitation hypothesis “postulates an observed cohesive explicitness from SL to TL texts regardless of the increase traceable to differences between the two linguistic and textual systems involved” (1986: 19). Accordingly, most research that seeks to prove or disprove the explicitation hypothesis has been limited by the very task at hand: to quantify and to empirically prove an effect. Existing studies range from the insertion of a connective (Olohan and Baker 2000; Øverås 1998) to verbalising the subject of a phrase implicit in the original (Konsalova 2007). While Pápai identifies sixteen strategies in total (2004), this approach still implies demonstrating the validity of the hypothesis for the given patterns alone.

The supposed reason for explicitation, however, lies in interpretation and a desire to be understood. As Blum-Kulka explains, “[t]he process of interpretation performed by the translator on the source text might lead to a [target language] text which is more redundant than the [source language] text” (1986: 19). In other words, translators explicitate because they question whether something implicit in the source language will be as understandable in the target language. Hence the addition of verbalised equivalents of the non-verbalised aspects of original utterances, and perhaps to a tendency to overcompensate, leading to redundancy.

What, then, of self-translation? If translators are guided by “a process of interpretation,” self-translators are carriers of interpretation. It is therefore logical that self-translators would have a similar but stronger motivation to explicitate. Their access to a larger arsenal of culturally acceptable choices makes them likely to overcompensate in more visible ways, including semantic rather than merely grammatical. One way to test this hypothesis is to consider Gary’s changes to the original novel, and where available to the intermediary translation. Explicitation, in this enlarged self-translation context, includes semantic additions that have no direct analogue in the original but reinforce or spell out tacit information present in it.

An interlinear produces a setting where every change is deliberate insofar as each choice responds to a pre-existing target language version. We could therefore assume that, if explicitation is largely an unconscious process (as data from stylistic patterns over large corpora seems to imply, for example Pápai 2004), we would not see further additions. Yet Gary’s corrections still reveal a tendency to verbalise otherwise implicit information, often leading to overcompensation or redundancy, despite the writer’s working, though with reference to the original, within the target language.

In our passage above, two or three instances of apparent rewriting could be seen as explicitational and, consequently, translational. The added epithet wonderful is one example. Similarly, “C’est très flatteur de savoir que” translated word by word by the hired translator becomes in Gary’s revision “It is flattering and very satisfying.” Elsewhere, the reinforcing repetition of the contextual information “with roses” in “replacing with roses your sweet-smelling self” remained implicit both in the original and in the intermediary translation.

Let us see how this kind of explicitation functions more generally in Gary’s self-translated novels. La Danse de Gengis Cohn, self-translated into English with Camilla Sykes (see Huston 2001), is a tragicomedy; the protagonist, a Jewish comedian, is a dybbuk, a Holocaust victim turned spirit who now haunts his murderer’s consciousness. The unwilling host of Cohn’s spirit, Schatz, is now a denazified police officer and not at all thrilled to be woken up to recite the Kaddish or forced to cook gefilte fish—these being harmless jokes on Cohn’s part intended, he tells us, to foster a tender friendship between the pair. In the meantime, Schatz is engaged in solving a series of mysterious murders.

Compared with the French original, the English version is strewn with examples that are surprisingly close to Blum-Kulka’s (1986) definition of explicitation in spirit, even though they may technically appear as deviations from translation. At one point, Cohn claims that one “ne se serait jamais doutée du trésor que Schatz cache en lui” (the treasure being himself, in a pun on the officer’s last name; Gary 1967/1995: 19). In English, this reads “aware of the hidden treasure that Schatz carries within him” (Gary 1969: 16). The treasure within Schatz is now qualified as “hidden,” a redundancy; Stevenson did not see the need to call his novel “The Hidden Treasure Island,” and this has never prevented readers from racking their brains over where the treasure was buried. A few lines down, Schatz seeks to justify the Holocaust: “There was an ideology, strong political and philosophical motivations…” (1969: 16). The French original only has “une idéologie,” but what is an ideology if not strong political and philosophical motivations? In the self-translation, the writer expands on the word and, essentially, repeats himself.

For this translation, just a few pages of the intermediary translation survive with Gary’s corrections and these can be compared to the published novel. At one point, for example, Cohn and his host realise that the author, in whose subconscious they now both dwell, is trying to expunge them, together. But why me along with the Nazi? wonders Cohn. Schatz responds “Nous sommes associés dans son esprit” (Gary 1967/1995: 148). The translator renders the phrase as “In his mind, we are associated.” Gary’s hand appends to this a reinforcing “forever” (this is also the published version; Gary 196SIC9: 127). A few lines down, he concludes “J’y suis, j’y reste” (Gary 1967/1995: 149). The intermediary translation renders this as “Here I am, and here I’m staying.” In the published translation, Gary changes this to “I am here to stay. He is not going to get rid of me” (Gary 1969: 128). The first phrase alone constitutes a reasonably accurate translation of the original phrase, the second reinforces the message but is otherwise redundant. Elsewhere, Cohn pins up an article to make Schatz feel “moins seul” in the bathroom (Gary 1967/1995: 20). In the translation, the irony is replaced by a clear statement of its real meaning: “so that he wouldn’t feel safe from me, even in there” (Gary 1969: 18). Although some redundancy is also removed, occasional implicitation is vastly outnumbered by instances of explicitation.

We can thus speak of the additions, in such cases, as instances of explicitation beyond translation. In most cases, semantic redundancy is expressed through additions to the text. Sometimes, however, the effect extends to new information as well. Perhaps the most elegant example of this strategy is the “petit bouquet de fleurs” (1967/1995: 21) that Genghis Cohn presents to his host: in the English, the flowers are named and the name is “forget-me-nots” (1969: 19). Here, the name is technically new information, but poetically, it amplifies and reinforces the spirit of the gesture.

Humour is often at stake in cases of redundancy and Gary’s explicitation strategy may be explained away by a desire to make his humour more accessible for an American audience. However, explicitation—taken a notch further, and a notch further from translation—is also observed in Lady L, a novel Gary self-translated from English into French. This novel, which Gary claimed was his favourite “au point de vue orfèvrerie artistique” (Huston 1995: 56), was the first Gary wrote directly in English. The French version was commissioned to “a distinguished translator working directly for Gallimard” (Bellos 2010: 297), Jean Rosenthal, and reworked by the writer. De Gaulle sent Gary a congratulatory note on reading the translation, although it never became a bestseller the way it did in the US.

Lady L is a former Parisian prostitute who fell in love with an anarchist and, through an elaborate scheme devised by her lover, became an English lady. The novel opens as she turns eighty and decides to tell her long-standing admirer Percy, the British Poet Laureate, her story. The English original develops the suspense; the reader can infer that she has a secret and a past, but Gary gives no leads as explicit as her thought, at the outset of the French translation: “Je suis un peu anarchiste” (Gary 1963: 12). Here, the original English text has “I am afraid I am still quite, quite romantic, she thought” (Gary 1958: 3). Is anarchiste a possible translation for the English romantic? Hardly! Yet, at the deeper level of the novel’s structure, the heroine’s anarchist past is the implicit information connoted by the English term and made explicit in French. Compare this to another passage, where Lady L’s little grandson runs up to her. As we later learn, her son was really her anarchist lover’s. Her partiality to the little grandson is due to his having inherited some of her lover’s features and the English text tells us this indirectly, mostly by drawing our attention to the boy’s eyes:

And so she rose to join her grandchildren and her great-grandchildren, who were waiting for her. She didn’t like them particularly, except the very young one, the little boy—he had such lovely dark eyes and fine features and she loved him. The others merely smelled of milk.

Gary 1958: 7

Compare this to the same passage in French:

Elle se leva donc pour rejoindre ses petits et arrière-petits-enfants. Elle n’en aimait qu’un, le benjamin, qui avait de beaux yeux sombres et impudents, des boucles aux reflets fauves et une impétuosité, une virilité naissante qui l’enchantaient: la ressemblance était vraiment extraordinaire. Il paraît que l’hérédité se manifeste souvent ainsi, en sautant une ou deux générations. Elle était sûre qu’il allait faire des choses terribles lorsqu’il serait grand: il était de la graine d’extrémiste, cela se sentait immédiatement. Peut-être avait-elle donné à l’Angleterre un futur Hitler ou un Lénine, qui allait tout casser. Elle mettait tous ses espoirs en lui. Avec des yeux pareils, il allait certainement faire parler de lui. Quant aux autres moutards dont elle confondait toujours les noms, ils sentaient le lait et il n’y avait rien d’autre à en dire.

Gary 1963: 16

Not only are the eyes of the little boy mentioned but the fact that he is someone’s likeness, and that this someone must be the boy’s grandfather (since a generation is skipped). We even learn that the resemblance lies in an extremist streak. In other words, only a very thick reader could now miss the key to the story. These changes to the original include an abundance of new detail that only the author could have introduced. Not implicit in the words of the English original, they are very much implicit in its spirit.

The essence of what Gary is doing here and elsewhere corresponds to Blum-Kulka’s definition of explicitation both in the reader’s being “able to infer from the context, her world knowledge or from other inferential sources if it were not verbalized” (quoted in Becher 2010: 2) and in its reference to redundancy. This is yet another step away from translation compared to our first few examples. Such profound and extensive changes to the original appear to be the very opposite of translation; what is more, explicitation effects have been observed in settings other than translation or even self-translation, notably in “various forms of bilingualism-influenced communication” and are linked to “contexts of high cognitive demand” (Kotze 2020: 120). Yet the nature of such changes, in self-translation and potentially further afield, can be fruitfully accounted for in translational terms. In self-translation, the redundancy is a matter of semantic and even thematic structure rather than, say, reintroduction of an implicit conjunction as is usually the case in ‘standard’ translation (see Edel-Roy 2010). Such self-translational explicitation constitutes flagrant rewriting that professional translators would not consider themselves empowered to practice. It is also, however, another example where the term translational has a precise application and can elucidate phenomena beyond translation.

6. Discussion

The study of intermediary allograph translations as used in self-translation offers a fruitful ground for applying translational approaches inductively to textual material that pushes, and even extends, the borders of translation in concrete and tangible ways. Most of the examples considered in this study would be classed as instances of authorial licence to rewrite instead of translating. This dichotomy behind even the subtler accounts of negotiations between accuracy and freedom in self-translation leads to important nuances being overlooked. Umberto Eco speaks for both translators and self-translators when he points to “rewriting as a free form of translation,” and cites examples from his own work, translated by others, where “to obtain the same effect that the text was designed to provide, we have to rewrite” (2001: 57). Indeed, Eco encouraged his translators to behave as self-translators in cases that, he believed, revealed “that we are on the borders of the notion of translation” (2001: 61). Not all rewriting is translational, nor are the examples analysed here translational in every respect. Yet Gary’s practice makes it clear that a descriptive technical extension of what we consider translational beyond translation seems to be called for.

Based on the present analysis, we can revisit the continuum proposed earlier in this article to include, to the right side of free translation, a swath of choices made by the self-translator that either fulfil a translational purpose by means that are not usually available to a third-party translator or, without being translation, can nevertheless be understood in translational terms (Figure 3).

Figure 3

Freedom in translation continuum revisited

Freedom in translation continuum revisited

-> See the list of figures

As the example of explicitation suggests, there is no reason to doubt that the continuum can be extended even further and enriched with additional examples and nuances to fill in the blank space on the continuum as depicted above. This potential for future research makes this approach an important precedent for current debates on interdisciplinary applications of translation theory. Translation has long been recognised as a key conceptual metaphor in adjacent fields, notably in literary studies (see Guldin 2016; Walkowitz 2015; Simon 2013; Apter 2006). While this development contributes to the growing perceived relevance of Translation Studies, many theorists of translation find it concerning. Bassnett (2011: 73-74) and Bellos (2011: 312) justly comment that translation is becoming devalued or marginalised as the term is increasingly used metaphorically in criticism (see; Saussy 2017: 1-2; Bachmann-Medick 2013: 188). As a result, the boundaries between translation as metaphor and translation as the attempt to convey in one language something originally expressed in another have grown increasingly blurred. In this context, an inductive approach allows us to gradually extend our notion of translation and offers a path towards an alternative to more metaphorical extensions of the term in interdisciplinary contexts.

Whether this means that we ought to extend our definition of translation or admit that translational mechanisms can be present beyond translational contexts per se is a discussion best left for another venue. Both conclusions are possible, but the first would have challenging theoretical implications. Extending the concept of translation may imply that, without access to authorial intention or consciousness, faithful translation may often prove deficient. In other words, such an extended definition would go against the idea, near-axiomatic in Translation Studies, that however we define a text’s meaning, the text within its context is, on the whole, self-sufficient where translation is concerned (Chesterman and Arrojo 2000). It would also potentially imply a liberation of the third-party translator along the lines of the self-translator’s access to translational processes in rewriting. This argument, already made by some (for example Bantinaki 2020), is fraught with complication.

The second interpretation, that is that translation theory applies beyond translation itself, is easier to accept. It is also more promising in terms of making sense of interdisciplinary discussions of translation. Insofar as translational analysis may be extended in a meaningful yet specific way, this gives reassurance that uses of the term in areas such as literary analysis may be similarly justifiable. At the same time, the question arises—to what extent? At some point, the specific connection to translation theory may cease to be traceable. In this sense, the proposed inductive approach can potentially be built on as a model for making clearer distinctions between the literal and the metaphorical uses of the term in interdisciplinary contexts.

7. Conclusion

To ensure continuity of the processes analysed, this article has deliberately limited its argument to novels translated with an intermediary allograph translation. This elicits more deliberate choice-making between different writing processes than in standard self-translation where the writer presumably also works with the mental possibility of literal translation, yet may lose track of where they deviate from the original. At each textual juncture, the allograph intermediary translation defines the default choice not only as a product of translation but also as a text bound to the original’s division into semantic units. At each juncture, the writer must make a deliberate decision to either leave the semantic unit in question, improve on it or put it aside to replace it with something new: between, in other words, translation, editing (which, as demonstrated above, is also very often translational) and original writing. This does away with the controversial category of “rewriting” in self-translation and replaces it by more specific categories. Approaching self-translation from this perspective may also help account for reasons why self-translators surprise researchers, often in the very same text, now by the degree of freedom they are willing to exercise, now by a contrasting degree of self-erasing faithfulness (see Oustinoff 2011: 121).

This study’s contribution consists in highlighting the use of provisional allograph translations in self-translation as a phenomenon with significant theoretical potential for the study of both self-translation and translation. The study of intermediary translations can inform, notably, our understanding of self-translation, collaborative translation and the interlinear. Its particular value lies, however, in the way it helps model and clarify how different drives are negotiated within self-translation. Changes made to the intermediary translation can be classified, helping us observe how these processes coexist and in which contexts one becomes preferable to another. For the study of translation more generally, this analysis thus exemplifies how our notions of what is translational may meaningfully extend beyond translation.

This study aims, further, to begin a new strand in the discussion of what is translational by trying to trace it gradually from free translation to and then beyond the very border of the traditional span of the term in Translation Studies. It thus shows inductively that there is, indeed, room to expand our notion of a translational process and, by pursuing a study of such processes in borderline settings such as self-translation, to take its application beyond such settings. The study of texts that are translational beyond translation may be one avenue towards fostering translation studies scholarship that would account for the increasing relevance of translation while building on existing achievements of translation studies in a concrete way. Without seeking to draw hasty conclusions, it aims to open a new critical perspective on the potential, as well as the potential limits, of interdisciplinary applications of Translation Studies.