The strain of constraint

Loyalty, elasticity, and feasibility in translation

Francosphères (2021), 10, (2), 265–282.


This article seeks to define and illustrate the notion of ‘constraint’ as it applies to literary translation. After a brief discussion of various ways in which the concept of constraint intersects with literary translation, the focus turns to the notion of ‘elasticity’, which describes tensions exerted upon the translator by factors particular to a given translation project, whether these are stylistic, formal, lexical, or intentional literary constraints. This tension forces the translator to work ‘otherwise’ and dictates to a certain extent where the translator must situate him or herself along a continuum of ‘faithfulness’ that ranges from material form to semantic meaning. Four examples are taken from the author’s own work as a literary translator, drawing on translations of ‘constrained’ texts of progressive difficulty by Marcel Schwob, Raymond Queneau, Olivier Salon, and J.-A. Soubira. Finally, this illustration of varying textual elasticity and constraint is examined from a sociological angle, which seeks to explore practical constraints of literary translation in today’s American literary marketplace.

Cet article vise à définir et à illustrer la notion de ‘contrainte’ dans le contexte de la traduction littéraire. Après une brève discussion des différentes façons dont le concept de contrainte s’entrecroise avec la traduction littéraire, l’accent est mis sur la notion d’‘élasticité’, qui décrit les tensions exercées sur le traducteur par divers facteurs propres à un projet de traduction donné, qu’elles soient d’ordre stylistique, formel, lexical ou relevant de contraintes littéraires intentionnelles. Cette tension oblige le traducteur à travailler « autrement » et lui dicte dans une certaine mesure où il doit se situer le long d’un continuum de ‘fidélité’ allant de la forme matérielle à la signification sémantique. Quatre exemples sont tirés des travaux de l’auteur en tant que traducteur littéraire, à partir de traductions de textes ‘à contraintes’ et de difficulté croissante, de Marcel Schwob, Raymond Queneau, Olivier Salon et J.-A. Soubira. Enfin, cette illustration de différentes contraintes et élasticités textuelles est abordée d’un point de vue sociologique, qui tente d’explorer les contraintes pratiques pesant sur la traduction littéraire dans le marché américain contemporain.

The strain of constraint

Loyalty, elasticity, and feasibility in translation


This article seeks to define and illustrate the notion of ‘constraint’ as it applies to literary translation. After a brief discussion of various ways in which the concept of constraint intersects with literary translation, the focus turns to the notion of ‘elasticity’, which describes tensions exerted upon the translator by factors particular to a given translation project, whether these are stylistic, formal, lexical, or intentional literary constraints. This tension forces the translator to work ‘otherwise’ and dictates to a certain extent where the translator must situate him or herself along a continuum of ‘faithfulness’ that ranges from material form to semantic meaning. Four examples are taken from the author’s own work as a literary translator, drawing on translations of ‘constrained’ texts of progressive difficulty by Marcel Schwob, Raymond Queneau, Olivier Salon, and J.-A. Soubira. Finally, this illustration of varying textual elasticity and constraint is examined from a sociological angle, which seeks to explore practical constraints of literary translation in today’s American literary marketplace.

Cet article vise à définir et à illustrer la notion de ‘contrainte’ dans le contexte de la traduction littéraire. Après une brève discussion des différentes façons dont le concept de contrainte s’entrecroise avec la traduction littéraire, l’accent est mis sur la notion d’‘élasticité’, qui décrit les tensions exercées sur le traducteur par divers facteurs propres à un projet de traduction donné, qu’elles soient d’ordre stylistique, formel, lexical ou relevant de contraintes littéraires intentionnelles. Cette tension oblige le traducteur à travailler « autrement » et lui dicte dans une certaine mesure où il doit se situer le long d’un continuum de ‘fidélité’ allant de la forme matérielle à la signification sémantique. Quatre exemples sont tirés des travaux de l’auteur en tant que traducteur littéraire, à partir de traductions de textes ‘à contraintes’ et de difficulté croissante, de Marcel Schwob, Raymond Queneau, Olivier Salon et J.-A. Soubira. Enfin, cette illustration de différentes contraintes et élasticités textuelles est abordée d’un point de vue sociologique, qui tente d’explorer les contraintes pratiques pesant sur la traduction littéraire dans le marché américain contemporain.


In my experience as a literary translator and translation studies scholar, there are a variety of meanings of the word ‘constraint’, which relate to translation in different ways. It can refer to the constraints of translation in general, to translating constrained literature, or to ‘constrained translation’, creative translation that makes use of intentional constraints to produce new texts. It can also be used to describe difficulties inherent to the practice of translation as an artform and a trade skill. For this article, I have been asked to draw from my own practice as a literary translator with a predilection for difficult projects, to investigate the notion of constraint and how it relates to literary translation. I will provide some specific examples of how I see the notion of constraint operating within literary translation, after which I will explain certain problems I see in how such difficult translations fit (or do not fit) within the current publishing framework in the United States.

Defining constraints

Before I offer some examples of constraint in translation drawn from practice, I think it worthwhile to consider the word itself: what we mean by the word ‘constraint’, and more specifically, how we can relate this to literature and to translation. Somewhat amusingly, from the point of view of a translator who supposedly specializes in the translation of such complicated texts, an etymological examination of the word’s history shows that in the fourteenth century, the word constreinte was used to mean ‘distress and oppression’,1 which is indeed what is sometimes felt by the translator as he or she attempts to wrap his or her head around the gauntlet to be run. The term was borrowed from the Old French, which, earlier still, meant ‘binding, constraint, or compulsion’. This, in turn, was spawned by the Vulgar Latin constrinctus, from the Latin constrictus, itself a derivation of the verb constringere, which meant ‘to bind together, tie tightly, fetter, shackle, chain’.

Why would one intentionally undertake a translation that caused feelings of distress or oppression? ‘Because it’s a challenge’ is the simplest answer, a puzzle or a challenge to overcome. The mid-sixteenth-century meaning of ‘constraint’ fits here as well, as by the 1530s it meant ‘coercion, compulsion, irresistible force or its effect to restrict or compel’. This is accurate; some of us are by our nature drawn to such puzzles. However, much as when it comes to writing with constraints, another reason is to force us to work ‘otherwise’. A contemporary French definition of ‘contrainte’, used more broadly, defines it as ‘[v]iolence physique ou morale exercée contre une personne afin de l’obliger à agir contre sa volonté’.2 Physical or moral violence exercised against a person in order to oblige them to act against their own will. I might translate this to our purposes, as a ‘linguistic, grammatical, or literary roadblock erected before an artist that obliges them to work in ways that deviate from the habits to which they have become accustomed’. Earlier French etymologies also convey this: circa 1174, contraindre had the sense ‘empêcher quelqu’un de suivre son penchant naturel, l’obliger à se gêner’.3 To force someone out of their comfort zone, to prevent them from doing what they always do the way they have always done it before.

Finally, if we reach back to Godefroy’s Dictionnaire de l’ancien et moyen français (1881), which covers the ninth to fifteenth centuries, we find under numerous variants a set of synonyms: resserrer, tenir serré, presser, faire pesant. To squeeze or to grip, to compress, to hold tight, to press against or apply pressure to, to make heavy or cumbersome. This series, interestingly enough, overlaps with a separate concept I am drawn to when considering the translation of constrained texts. Within this last lexical field, I would like to focus on these tactile notions, these pressures and weights. To keep the metaphor in the realm of the visual and the tactile, when discussing the translation of constrained literature, I tend toward the term ‘elasticity’.

In Friedrich Schleiermacher’s On the Different Methods of Translating, from a lecture he delivered in 1813, the key takeaway for the discipline of translation studies has long been his theory of two poles, his belief that there are two methods of translation available to a translator:

What paths are open to the translator […]? In my opinion, there are only two. Either the translator leaves the writer alone as much as possible and moves the reader toward the writer, or he leaves the reader alone as much as possible and moves the writer toward the reader. Both paths are so completely different from one another that one of them must definitely be adhered to as strictly as possible, since a highly unreliable result would emerge from mixing them, and it is likely that author and reader would not come together at all.4

Schleiermacher believes that we have two options as translators. Some translators think they have a tendency to stick quite closely to the syntax, form, and lexical materiality of the source text, while others choose to take a freer approach, reworking the language and syntax to a greater degree. While Schleiermacher’s theory puts it in terms of choosing a side, between the text or the reader, we may rephrase this in contemporary language as being a question of any of the buzz words you might see in typical reviews of translations: ‘fidelity’, ‘faithfulness’, or ‘loyalty’. Loyalty to the materiality of the source text, or a focus on our target language. The French have their terminology for this as well, often employing the terms sourcier and cibliste, coined in 1983 by Jean-René Ladmiral. The sourcier is tied to the source text, the cibliste is bound to the target. While there is a degree of sense to this, in retrospect it seems to be a rather simplistic view, despite Schleiermacher’s protestation that anything in between would lead to ‘a highly unreliable result’. Instead, in the modern industry of literary translation, a translator will situate him or herself somewhere along a continuum between these two poles, as if balancing upon a tightrope or a string, or perhaps multiple modes can be used in the same text, depending on what it demands. Whatever the translator’s decision, we can think of these two poles as being extremes denoting, on one side, the semantic value of a text, and on the other, its physical (material) properties. Is one to be more faithful to the text ‘as it is written’, or is one to lean more towards the conveyance of meaning, the transmission of information, with a preference for communicating that semantic content in the best possible way afforded by the target language system? Really, we could use these poles to describe any incompatibilities within a translation project, any two or more systems that interfere with each other.

Overall, I do not believe that a translator truly decides to work one way or the other, but instead that it is often the text - and by this I mean the translator’s ‘reading’ of a text - that dictates where, on the continuum connecting the poles, the translator must be situated at a given moment. Picture this string as two elastics, two bungee cords, each one fastened to a pole at one end and to the translator in the middle. These two elements to be preserved, in this case meaning and form, exert more or less pressure on the translator depending on their role in the text and how they function among other systems, and this pressure will dictate where along the continuum the translator must be stationed.

Adding intentional literary constraints to the process tends to multiply the tension exerted by these different poles in drastic ways. The more restrictive the constraint, or the more disruptive it is within the system(s), the more tension the translator is subjected to. This tension, in extreme cases, must inevitably pull him or her away from the semantic information he or she is trying so desperately to convey. When external constraints are introduced into a text, especially if they are a key feature of the text - which, in the case of constrained writing practiced by groups such as the Oulipo in France, they often are - they must be prioritized. And when a constraint forces a translator to lean too far in one direction, in this case ‘toward’ the preservation of the literary constraint, the elastic tension this produces can make it very difficult to maintain one’s grasp on the other elements in the source text that the translator wishes to preserve or recreate.

To demonstrate the varying elasticity involved in the translation of constrained texts, I will offer examples drawn from several different projects I have worked on. It would be possible for us to consider the question of whether or not all acts of translation can be thought of as a form of constrained writing. However, in keeping with the scope of this article (constrained as it is), I will leave that overarching generalization aside and focus on illustrating a few specific points in hopes that they might add to this overall conception of constraint in translation and the practical difficulties that arise from it.

Marcel Schwob’s Imaginary Lives

This first example occupies a middle-ground between the natural constraints of more traditional literary translation and the intentionally constrained problems that occur during the translation of texts by the Oulipians and others like them. Here, I will describe the chief difficulty I encountered in translating Marcel Schwob’s Imaginary Lives (1896), which is primarily lexical in nature.5 In this collection, Schwob wrote short biographical vignettes drawing on available historical sources, describing the lives of figures spanning from Empedocles in the fifth century BCE to Burke and Hare, grave robbers in early nineteenth-century Scotland. The problem with translating this text to its fullest literary potential is one of commitment, research, and time; this difficulty lies in the fact that the historical sources Schwob drew from were numerous, they were in multiple languages, they spanned thousands of years, and fragments of these sources were drawn into his own writing. The easiest approach would have been to translate the text as Schwob published it, taking for granted that the historical information it contained had been fully assimilated into his composition. However, a straight translation of this collage-based text poses another a problem: by incorporating literary citations, unmarked and buried within the prose itself, Schwob has produced an additional level of reading and association and made it available to the reader. If I had ignored these buried citations, the text would have undergone a process of ‘smoothing’ during the translation, as not only do these citations point outward, referring a reader back to what Italo Calvino once termed his ‘hypothetical bookshelf’, but they also tend to produce a shift in tone, register, rhythm, vocabulary, and/or syntax.6 These fragments stick out because they are the words of another writer, composed in his own personal style, and employing his idiolectal vocabulary. Translating the entire text as a homogenous unit would thus have the effect of lessening this shift at occurrence, thereby erasing some of the richness of the text, and preventing readers from following the threads to their sources.

Schwob was not only a writer, but a voracious reader as well as a literary translator, and accordingly, it was important for me to determine what exactly it was that he had inserted into his text, whether it was from Italian, French, English, Latin, or German, all of which Schwob was capable of working with. Fortunately, history of the sort Schwob was drawing from is only to be found in so many places. When he spoke of Paolo Uccello, it was evident that the fragments he was working with were drawn from Vasari’s Lives of the Painters. When he wrote of pirates, his primary source was The General History of Pyrates (1724), often attributed to Daniel Defoe; on Pocahontas, The General History of Virginia, New England, and the Summer Isles by Captain John Smith (1624), and Edward Neill’s History of the Virginia Company of London (1869), and so on. Schwob pulled fragments of these works directly into his own, translating them himself if they were written in languages other than French. And, if Defoe describes Blackbeard’s head as having been ‘hung up at the boltsprit end’, drawing himself on transcribed court-room proceedings, then these details are as historically accurate as they can possibly be. No amount of imagination on Schwob’s part could make these details any truer, and any truth claim lies with the original author(s). His imagination and skill were instead focused on the selection of the details of which he made use, and in extrapolating other information that was not included in the histories.

Once I had located as many of these fragments as possible among dozens of books, libraries, and databases, and translated them in isolation if the case required, I inserted them back into my translation. I preserved two final versions of the text: an unmarked copy to submit to my publisher, and another with each fragment marked, to make certain that during the editing process, none were lost or altered. This way, when my editor suggested that a phrase read awkwardly, such as ‘[i]f a man remained inclined for drinking after that hour, he was to drink upon the deck, under the open sky’,7 it was straightforward for me to argue that the phrase (in this example) came directly from ‘Black Bart’s Pirate Code’ of 1724. A more direct translation of this text would have made for a much quicker turnaround, but it would not have preserved the many intertextual traces of Schwob’s source texts, and the resulting translation would be much poorer for it.

Raymond Queneau’s Exercises in Style

Shifting to more formally constrained texts, my next example is Raymond Queneau’s Exercises in Style.8 Queneau began writing his Exercises in the early 1940s, over fifteen years before he co-founded the Oulipo with François Le Lionnais in 1960. For my contribution to the expanded anniversary edition of this book, the translation of many of the new exercises involved working through an expansive list of different stylistic, syntactic, or lexical problems, each one presenting the translator with its own difficulties. However, certain of these texts do pre-figure what would later be known as ‘hard’ constraints, and especially some of those Queneau wrote later, as he added to the French edition years after the fact, removing some exercises and adding others.

Writing in the style of Edgar Allen Poe or adopting the form of a radio advertisement can be described as constrained writing in their own right, but other examples in this collection can better help to showcase some of the extra work that a translator is subjected to when working with hard constraints. For example, among these new exercises are a lipogram in ‘E’, a pair of S+7 texts, and some light mathematics-based variations which he referred to as ‘permutations’. The lipogram is a problem I will return to further on, but for the moment, let us examine the additional steps involved in properly translating these other relatively simple examples of hard-constraint texts.9

The permutational texts Queneau composed are all manipulations of what I call a floating base text. Once the translator has determined the nature of the manipulations, those manipulations can typically be reversed, leaving a clean, unadorned version of the story. In the case of Queneau’s Exercises, the base texts that result from this reverse-engineering are quite similar to the two most straightforward versions of the story to be found in the collection: ‘Notation’ and ‘Narrative’, although they are not a precise match with either one. The first step, then, is to reverse the constraint in the original French and determine the French base text. In each instance, I then had to compare what I ended up with to the other base texts, as Queneau did slightly vary his base from one exercise to another. Then, I translated the ‘new’ base text into English,10 after which the constraint was re-applied to the resulting English text. For example, in the case of the Lescurian S+7s, which involve looking up a set of lexical items (typically nouns) in a dictionary and replacing them with the word (or noun) found seven entries later, after reverting to the base text and translating it, I reactivated the constraint using a dictionary that was as close as possible to that which Queneau had employed. For instance, he produced one of the examples using the basic lexicon in Le Français Élémentaire (1954), to which my closest English parallel was Charles Ogden’s 1930 lexicon from Basic English: A General Introduction with Rules and Grammar. A similar base text reversal was used for the translation of his ‘Permutations by Groups of Letters’ or ‘by Groups of Words’, and several other examples.

The result of these necessities is arrived at after I have fought off the elastic pressure from two or more poles: first, that of representing the semantic content of the text; then, that of the stylistic systems, the materiality of the text; and finally, the tension exerted by the constraint itself. This is crucial, of course, in the translation of an Oulipian text that features hard constraints. There is little point of translating a lipogram in ‘E’ if I do not abide by the same constraining stipulation myself. With the S+7s, a large part of their novelty (aside from the potential absurdity of their new content) resides in the fact that the Oulipian reader can reverse engineer the text in his or her head, especially if the base text is familiar, thus enjoying the surprise potential of reading two texts in one. And, of course, it would have been futile to simply translate the results of Queneau’s dictionary manipulation as they were written on the source-text page, for the resulting English text would bear little resemblance to the base text, and thus erase any trace of the possible second text.

Olivier Salon’s ‘The Stations of the Cry’

In translating Olivier Salon’s ‘The Stations of the Cry’, the elasticity provided much greater tension, in spots interfering with the semantic ‘faithfulness’ of the translation on a relatively extreme level.11 This short text is, as Salon describes it, a ‘retrograde progressive lipogram’. It is twenty-six paragraphs long, the first paragraph of which contains all of the letters of the alphabet; then, after each successive paragraph, its author was barred the use of a letter of the alphabet, beginning with ‘Z’. To highlight this (and, for the translator, to make matters worse), the author has alliteratively focused each paragraph on the next letter the text is about to lose, and, as is his fashion, Salon capped each paragraph off with a pun that in some way described the loss of the given letter. For example, in the culminating line of the seventh paragraph, it ‘is no longer time for tea’, then, from the next paragraph on, ‘T’ does not appear again. The final paragraph of this masterful piece of constrained writing, of course, reads: ‘Aaaaaaaaaaaaaaaa!’, which accurately described my state of mind by the end of my translation.

As to the elasticity of this constraint, its pull was considerable. And, interestingly, the differences between the language systems of French and English, on a purely material level, made visible certain disparities that I had not previously considered. I can now argue, for instance, that Salon’s text was in certain ways more difficult to write in English than it was in French, thanks to one particular quirk of the English language: many of our most common words, words we rarely consider, such as simple pronouns, articles, and prepositions, tend to include letters that come later in the alphabet than do their semantic equivalents in French. This means that in English, such simple-but-necessary words became unavailable to the writer earlier in the text. In fact, of the one hundred most common words in English, according to Oxford University Press, sixty-four of them were unavailable by the end of the seventh paragraph (‘T’), including what, where, when, why, who, with, the, to, it, she, you, at, and etc. By the time I reached the thirteenth paragraph, having lost my ‘N’, all subsequent nouns had to begin with a consonant, as only the indefinite article ‘a’ remained.

While this perhaps gives an idea of the difficulty of translating this text, it also offers an example of the manner in which constraint elasticity exerts itself on the translator, forcing him further away from the semantic, and thereby rendering content loyalty increasingly difficult. In these extreme cases, I cling to the meaning as best I can, struggling against the tension generated by the constraint. At times, this results in a form of rewriting that stems more from overall interpretation, from my own reading of the text and my instinctual feel for the author’s intention, or from reflection on how the given fragment relates to the text as a whole. This was the case with the paragraph featuring the soon-to-be-defunct letter ‘Q’, as I was no longer permitted to use the letter ‘U’, lost several paragraphs earlier, and there are not many words in English that contain the former without the latter. Salon, in composing his French text, was able to play with the loaded French national symbol of the ‘coq’. However, my re-using this French loan-word in English felt like it would fall outside the rules of the game, and I had set a ground-rule for myself at the outset: all words I employed must be attested in my English dictionary. This left me with only a few attested loan-words for ‘Q’ without ‘U’ that did not come from French; since most of these came to English from Arabic, a precise recreation of the semantic content in this paragraph became impossible and creative substitution was necessary. In that paragraph, my only path was to reflect on my reading of Salon’s overall message, and to attempt to build a bridge out of the constraint-permitted lexicon I had at my disposal. For comparison, here is Salon’s ‘Q’ paragraph followed by that from my English translation:

On a manié l’idée, l’idée de ce coq. Là, ce coq coi. On place le coq en plein champ, champ de blé, on a l’idée de ce champ de coq. L’idée… le coq piaille comme la limande, comme l’animal de l’océan.12

Each idea half-baked in flame, an idea held in check like a niqab cloaking a face. And a niqab can hide a look, a glance. A niqab made of qiana, concealing a dimpled chin, a long black mane, enfolding a bold confidence. An idea… a niqab can conceal a face like an idiom can hide an idea, like an idea can cloak a being.13

It is evident that these two paragraphs differ a great deal. In this case, due to the limitations placed on me by the constraint, instead of translating what Salon has written in his paragraph, I have had to settle for substituting a paragraph that I felt remained within the spirit of the text, and which provided a suitable bridge from the ‘R’ paragraph to the ‘P’ paragraph, while maintaining the strict limitations of the constraint. In this case and others like it, the elastic tension exerted on the translator is so strong in one direction that all that remains is for him to gesture towards the original semantic content. And once again, to ignore the constraint at this stage would be to concede that the text was untranslatable, to throw in the towel and give up.

Soubira’s 666

For a final, extreme example of elasticity in the translation of hard constraints, I will turn to a work in progress, a pet project which is sure to take me many more years to complete. As far as the translation of hard constraints goes, I am yet to encounter a text that exerts as much elastic tension as this one does. The text is a poem by Jacob-Abraham Soubira, a self-styled prophet who lived in the southwest of France in the early nineteenth century. A prolific writer, Soubira was a doomsday poet, and penned two different versions of his masterpiece, in 1824 and 1828. The first is comprised of nine quatrains, the second an expanded eighteen cinquains. I have started work on the shorter of the two, hoping that as I work out the kinks and become more adept at managing this constraint, I will, like Soubira, be able to work more quickly and cleverly on the second piece. Soubira makes use of a hard constraint that defies the imagination, one which must have been very difficult to work with, and is even more difficult to translate, considering that one cannot write oneself out of trouble but must cling to the semantic content. Soubira also had the advantage of being a ‘delegate of the Messiah’, which surely helped him with this process, but I have no such inside information to guide me. The poem recounts the downfall of France through history and toward the coming apocalypse, as perpetrated by heathens and government. Here is the first quatrain:

Le 19e siècle hissera de l’orage! Son mondain zéphir En altérera le paysage Et déracinera le visir!14

At a glance, it does not seem so overwhelming, but there is a catch. Soubira devised a numerological system, attributing a value to each letter of the alphabet, where A=1, B=2, C=3, and so on to 10, 20, 30, through 100, 110, 120, and up to 150, with two exceptions influencing the order: there is no ‘J’, that being Judas’s letter, for which he instead employed G+E=12, and then ‘W’ carries a value of 120, it being in French double V, or twice the value of ‘V’. With this system established in the background, the constraint is defined by the poem’s title: ‘666’, the Number of the Beast.15 Each line of the poem, when you add up the value of each of its letters, has an aggregate value of 666.

How does one go about such a challenge? If I were more of a mathematician, perhaps I would be able to develop a shortcut, but that not being the case, I have borrowed a page from Georges Perec. I have been fashioning myself a dictionary of values, much in the way he constructed a lipogrammatical lexicon for writing his novel La Disparition.16 The problems are numerous. First, much as with ‘Stations of the Cry’, I have noticed that English, as compared to French, is at somewhat of a disadvantage with this constraint, with more of English’s basic words being spelled with letters that come very late in the alphabet. While I am not prevented from using them outright, as I was in the lipogram, it is instead a case of not being able to fit them within the maximum value for the line. My method so far is an extension of the way I work with rhyming metered verse, which involves a much simpler counting of syllables: I have been giving preference to the load-bearing substantives and verbs that make up the semantic backbone of each line, and then construct the rest of the line with filler words that fit the remaining value allotment. To help expedite my determination of suitable syntactic padding, I have prepared a set of tables that includes the most common prepositions, determiners, and qualifiers, conjunctions, and verbs in each tense. Each of these tables I have reorganized in three different fashions, once by order of frequency in English, once in alphabetical order so I can find them quickly, once in numerical value to help with sums.

For each line, I set about translating the key semantic words first, being pulled by the constraint further and further away from the semantic content I am trying to preserve, until finally settling on a possibility. If I can make that option fit with the others needed for the line, great. If not, I step further away. For instance, take the second line, which simply reads, ‘Son mondain zéphir’, which I might typically translate as, ‘Its worldly Zephyr’. Unfortunately, ‘Zephyr’ in English takes a ‘y’, and so the word carries a high value of 443, already two thirds of my allotment for the line, forcing me to follow the semantic thread further. What is a ‘Zephyr’? In Greek mythology, it is the personification of the west wind, which does not help me, ‘west’ alone having a value of 435. And on I continue: ‘Zephyrus’ was equated by the Romans to ‘Favonius’, one of their wind gods, known to both the Ancient Greeks and Romans as the ‘anemoi’, which has also been transliterated in mythology as the ‘anemi’, which has a low value of eighty-five. Some semantic distance has been introduced between the source term and the English term I have opted for, but it is a reasonably close starting point that provides me with enough room for the line’s other terms. Now that I have completed this line and a few more, I feel that the resulting text is (thus far) still semantically similar enough to the source, a paraphrase if not an equivalent, and the constraint has been respected in full.

A translator’s grasp on the semantic is tenuous in an exercise like this, and very strictly dictated by the stipulations of the hard constraint. The pressure exerted by the physical rules of the system are dominant, and any semantically linked substitution that is made has been stretched thin by the time the constraint has been met. I have managed a little over a quatrain’s worth so far, and only work on the project on rare occasions, leaving it a source of occasional amusement and frustration: a vanity project in the truest sense of the words.

The constraints of the American literary translation industry

As a postface to this discussion of translating constrained texts, I will briefly address a very different form of constraint in translation, one to which I feel the above project descriptions are complementary. Alongside the notion of textual constraint and the difficulty that texts of this sort can pose to their translators, there is an everyday light under which we must also consider the relationship between translation and constraint: the often-unspoken constraints of being a translator in general. Since I produce my translations for the American literary market, I will address this from an American standpoint. As a second proviso, I must also note that I typically translate from French to English, and focus here on ‘literary translation’, as opposed to ‘technical’ translation or simultaneous interpreting. In literary translation, beyond the many constraints of the texts themselves, there remains the economic constraint of trying to earn a living as a translator. This differs from one country to the next, but at this point in time, it seems to be particularly difficult in the United States.

Sources of accurate information on the American translation industry are few, far between, and often out of date, but a number of recent surveys allow me to paint a picture by way of statistics. These figures are drawn from two surveys: the first by the Conseil Européen des Associations de Traducteurs Littéraires (CEATL), from 2007,17 and the second, a more recent American survey conducted by Alex Zucker and Jessica Cohen with the participation of the Author’s Guild.18 Of course, one of the most evident problems translators face in the United States is in finding a venue to publish their work; there are relatively few, and the limited number of publications-in-translation make it very competitive. This is where the term coined by Chad Post, ‘The Three Percent’, comes into play. Originally, Post used this to describe the percentage of the American domestic book market that was translated material.19 This figure included all varieties of books, including non-fiction, technical materials such as government publications and scientific reports, operating and instruction manuals, etc. To put this in comparison, according to the CEATL study, in 2005-2006, over 40% of new literature published in France was literature in translation. While different quantities are involved due to population base, this is a very considerable difference.

When it comes to fees and royalties, of the eighteen European countries surveyed, five countries had set agreements with publishers at the time of the survey with regards to basic fees and royalty percentages; Germany and Spain were at that time in the process of negotiating this, and over a dozen others had associations who had at least published recommendations with regards to fees and royalties. This is not the case in the United States, as attempts to set a reasonable rate by associations has been repeatedly met with government pushback. As Alex Zucker shared with me:

The American Translators Association published recommended rates between ‘the latter part of 1987 and the end of 1989. […] In December 1990, the ATA20 received word from the Federal Trade Commission that it was the subject of a “non-public investigation” regarding its now-defunct rate guidelines. That investigation was not completed until May 1994’.21

Responding to the investigation cost [the] ATA a ton of money, enough that it scared off not only the ATA itself but also ALTA from ever going near discussion of rates again. It even got to the point where some translators thought (or were told) that they couldn’t *discuss* rates with their peers on the site of an ATA conference while the conference was going on.22

Whether the translator’s fee is based on a page of thirty lines such as in Germany, the number of keystrokes, or the number of words in the source or target language, it appears that in Europe at the time of the survey, the average fees were highest in Belgium, Norway, and France, followed by Ireland and the UK.23

In the more recent American survey, of the members of the multiple associations who were surveyed, only 7% of respondents indicated that their entire income was translation-related. Instead, 79% of translators queried stated that they earned less than 50% of their income from translation. When queried on the gross income earned by full-time translators, which the survey defined as those who spend more than 50% of their working time translating, 32% of translators said they earned less than $10,000, 33% between $10,000 and $20,000, 18% between $20,000 and $30,000, and considerably less in any higher bracket. When we consider this next to the statistics about education, which found that 43% of respondents hold a PhD, and then consider that the poverty line in the US was about $12,000 in 2016, it is easy to see that there is a problem. As far as the rates themselves, Zucker and Cohen had this to say in their conclusions:

When asked to specify the rate for their most recent book translation, responses for prose translators varied from $0 to more than $0.20 per word, with the two largest segments charging $0.00-$0.07 per word and $0.14 or $0.15 per word, closely followed by $0.10 per word. Considering the rate ‘that UK publishers are prepared to pay’, as reported by the Society of Authors in the UK (‘in the region of £95 per 1,000 words’, or about $0.13 per word at the time of this writing), and the rate prescribed by the Canada Council for the Arts ($0.18 Canadian per word, or $0.14 U.S., it is clear that a large number of U.S. translators are being paid rates that make it difficult, if not impossible, to earn a living.24

With this obvious problem in mind, allow me to paint an even darker picture than that suggested by the Three Percent or the averages found in these surveys. I do not dispute these figures in any way; on the contrary, I applaud the work of these individuals and wish there were more like them around the industry. But one thing these numbers do not illustrate, as far as literary translation in the United States is concerned, is the question of ‘what it is that is being published in translation’. I will not provide statistics for what follows; this topic has been addressed in part by other scholars, including Gisèle Sapiro for the French market.25 Instead, these are simply the impressions of a reader, an academic, and a translator who has lived in both the United States and France.

If 40% or more of French literature published each year is literature in translation, this figure suggests above all that it is made up of a different body of literature than that published in translation in the United States. When you wander through the paperback literature section in a major French bookstore, you will see an abundance of writing in translation. And yet, for every experimental or literary work you see, you will also see shelf upon shelf of what we often refer to in the United States as middle-brow or even low-brow writing. Airport fiction. Full runs of romance novels by Nora Roberts and Fern Michaels; mainstream thrillers, from David Baldacci to Scott Turow; young adult books; all manner of Shopaholics or Mormon vampires, cat detectives, and as many shades of gray as you can imagine. This is decidedly not the case in the United States. The vast majority of literature in translation in the United States is made up of the hyper-literary, of avant-garde and experimental literature, poetry, high literary classics, and lost literary oddities; and, more recently, a push towards literature from non-traditional markets: in French, from the Maghreb, sub-Saharan Africa, and the Caribbean, for example, which itself increases the need for linguistic specialization, cultural and historical research, and sometimes even travel.26

As my examples of constrained literature have shown, there is a wide range of difficulty associated with literary translation. I readily admit that the examples I have presented above are extremes, and I am in no way suggesting that there is a publishing logic to translators being paid a high hourly wage for translating a text such as Soubira’s ‘666’. A project of that nature is clearly a work of passion, an experiment and exploration. But what I do hope to make clear is that in comparing base pay rates between countries, in this case, the United States and France, we must further compound the discrepancy in pay by the fact that what literary translators are being paid for at a by-the-word or by-the-page rate cannot always be evaluated by the same metric. Even if pay were equal, fourteen cents a word for a romance novel or a straightforward thriller - projects for which stylistics, research, and specialty skills are often quite unnecessary to rendering the content and to the telling of a good story - cannot be seen in the same light as pay of fourteen cents a word for high literature, where a text must be researched, contextualized, and reworked repeatedly until the desired literary effects are reached. The training required and the time involved are very different. The case for experimental and avant-garde literature is even stronger. I do not intend, by this, to denigrate the work of translators of low- and middle-brow literature, as I am aware that many of them are very capable translators and feel they deserve fair pay for their work as well. Instead, I am suggesting that since very little of this sort of literature is published in translation in the United States, for the industry to be a viable career option, either there needs to be a shift in what we are translating or in what publishers are willing to publish, or, conversely, a shift in pay commensurate to the difficulty of the task at hand. Perhaps we might think of it like competitive figure skating, where fewer points are scored for a triple axel than for a quad. If the complexity of a work dictates that to effectuate the translation to its highest potential, it will take two, three, or even ten times longer than a simpler project of the same length, it makes little sense that both projects are paid out at the same rate, whether that is eight cents per word or fifteen. If these are truly the translations that American readers are wanting to purchase, and if it is true that the market has no need for the French Twilight, the Spanish Da Vinci Code, or the Croatian Hunger Games - seeing as the American literary market produces more than enough low- and middle-brow literature itself without looking beyond its borders - then we must reconsider our priorities. The downside is that if things remain the same, if translators hope to make ends meet, they will need to limit the time they invest in a project, which will make it considerably more unlikely that they are able to produce the quality translations that important works deserve. This is currently a constraint with which the literary translation industry has been forced to live, but if we are to consider translating constrained literature, we also have to reconsider the constraints of literary translation.

Etymologies are drawn from unless otherwise noted.

As per the Trésor de la Langue Française Informatisée >>.

As per the CNRTL’s (Centre Nationale de Ressources Textuelles et Lexicales) ‘Ortolang’ (Outils et ressources pour un Traitement Optimisé de la LANGue), this etymology filed under ‘contraindre’ >>.

Friedrich Schleiermacher, ‘From “On the Different Methods of Translating”’, trans. by Waltraud Bartsch, in Theories of Translation: An Anthology of Essays from Dryden to Derrida, ed. by Rainer Schulte and John Biguenet. (Chicago: University of Chicago Press, 1992), pp. 36-54 (pp. 41-42).

Marcel Schwob, Imaginary Lives, trans. by Chris Clarke (Cambridge, MA: Wakefield Press, 2018).

See Italo Calvino, ‘Whom Do We Write For? or The Hypothetical Bookshelf’, in The Uses of Literature, trans. by Patrick Creagh (San Diego: Harcourt Brace Jovanovich, 1986), pp. 81-88.

Schwob, p. 129.

Raymond Queneau, Exercises in Style, trans. by Barbara Wright and Chris Clarke, rev. edn (New York: New Directions, 2012).

There was an additional constraint that I set for myself when I began working on these additional exercises. As these new translations were meant to complement the existing ninety-nine exercises translated by Barbara Wright in 1947, it was necessary for me to re-employ the vocabulary she had used when it fit, as I did not want these additional texts to seem drastically out of place in the collection, and so my translations had to refer back constantly to the previously translated exercises.

Again, keeping in mind Wright’s lexical predilections from throughout the collection (see previous note).

Olivier Salon, ‘The Stations of the Cry’, trans. by Chris Clarke, Words without Borders, December 2013 >> [accessed 23 August 2021].

Olivier Salon, ‘Les Stations du cri’, Words without Borders, December 2013 >> [accessed 24 August 2021].

Salon, ‘The Stations of the Cry’, trans. by Clarke.

Jacob-Abraham Soubira, ‘666’, cited in Raymond Queneau, Aux Confins des ténèbres: Les Fous littéraires, ed. by Madeleine Velguth (Paris: Gallimard, 2002), p. 305.

This is Soubira’s personal take on the age-old mystic divination method known as ‘arithmancy’ or ‘arithmomancy’, which the Encyclopedia Britanica describes as a precursor to numerology that was ‘practiced by the ancient Greeks, Chaldeans, and Hebrews’; in this form of mysticism, ‘the letters of an alphabet are assigned numbers by some rule, typically A = 1, B = 2,…., Z=26 or its equivalent’. This entry goes on to describe 666, ‘the Number of the Beast’, as being the ‘best-known instance’, found in Revelation (13:18). Apparently, ‘Revelation is the 66th book in the Bible, and the Number of the Beast occurs in verse 18, which is 6 + 6 + 6’ >>.

Oulipo member Georges Perec’s 1969 novel La Disparition, published in an English translation by Gilbert Adair, A Void, in 1995. The most evident constraint (of many) at play in this novel is that it is a lipogram in ‘E’, meaning that the letter ‘E’ does not appear once in its 300+ pages. Further information on Perec’s preparatory word lists can be found in his papers, as well as in Yû Maeyama’s article, ‘Les notes préparatoires à “La Disparition” de Georges Perec: genèse de la saga d’une famille brisée’, La Licorne, 122 (2016), 233-58.

CEATL: Conseil Européen des Associations de Traducteurs Littéraires, ‘Comparative Income of Literary Translators in Europe’, conducted by Holger Fock, Martin de Haan, and Alena Lhotová, CEATL 2007/08 >> [accessed 23 August 2021].

Alex Zucker, Jessica Cohen, and Author’s Guild, ‘2017 Authors Guild Survey of Literary Translators’ Working Conditions: A Summary’ >> [accessed 23 August 2021].

See, for instance, the Three Percent homepage on the University of Rochester website: >> [accessed 23 August 2021].

The American Translator’s Association (ATA), not to be confused with ALTA, the American Literary Translators Association, which is now a separate entity.

The quoted passage is taken from an online post by Rdovan Pletka, ‘ATA and FTC Money Mess: How Much It Cost, Why It Happened’ >> [accessed 22 September 2021].

Alex Zucker, via email with the author, 10 November 2018.

The UK, France, and Germany were lower than most of the higher-paid countries represented in the survey, showing a wider gap in possible salary.

Zucker, Cohen, and Author’s Guild, p. 3.

See, for example, Gisèle Sapiro, ‘Les échanges littéraires entre Paris et New York à l’ère de la globalisation’ [Rapport de recherche] (Paris: Centre européen de sociologie et de science politique/Le MOtif (Observatoire du livre d’Ile de France), 2010) >hal-01621786> [accessed 10 August 2021].

Sapiro elaborates on the differences in translated literature in France and the USA in the report cited in the note above.


Author details

Clarke, Chris