Stephanie Strickland's and Nick Montfort's Sea and Spar Between is in many respects a translational challenge that in some languages might seem an impossible task. Polish, our target language, imposes some serious constraints: one-syllable words become disyllabic or multisyllabic; kennings have different morphological, lexical, and grammatical arrangement, and most of the generative rhetoric of the original (like anaphors) must take into consideration the grammatical gender of Polish words. As a result, the javascript code, instructions that accompany the javascript file, and arrays of words that this poetry generator draws from, needed to be expanded and rewritten. Moreover, in several crucial points of this rule-driven work, natural language forced us to modify the code.
In translating Sea and Spar Between, the process of negotiation between the source language and the target language involves more factors than in the case of traditional translation. Strickland and Montfort read Dickinson1 and Melville and parse their readings into a computer program (in itself a translation, or port, from Python to javascript) which combines them in almost countless ways. This collision of cultures, languages, and tools becomes amplified if one wants to transpose it into a different language. This transposition involves the original authors of Sea and Spar Between, the four original translators of Dickinson and Melville into Polish, and we, turning into a multilayered translational challenge, something we propose to call a distributed translation. While testing the language and the potential of poetry translation in the digital age, the experiment – we hope – has produced some fascinating and thought-provoking poetry.

KEYWORDS: electronic literature translation, poetry generator, Sea and Spar Between, open-source, adaptation, generative literature

Translational Game / Distributed Translation

If “a struggle" or “an uphill battle” were not far from how we were describing our task when we decided to attempt to translate Sea and Spar Between – and to report on the project’s expected failure – a second look at the intricacy of the work revealed many rewards that were coming along with the struggle. Quite soon it became evident that the translation of Sea and Spar Between in Polish is not a mere “translational puzzle” but an elaborate “translational game” with many agents involved.
These agents/actors include:

1. Emily Dickinson and Herman Melville, with all the differences in their language and imaginary.
2. Stephanie Strickland and Nick Montfort, and their vision of how to combine Dickinson and Melville into a coherent and meaningful framework.
3. The javascript code responsible for the generation of poetic stanzas, and the requirements of HTML5 canvas element which is responsible for drawing the text within the browser’s window.
4. Existing translations of Melville and Dickinson into Polish (one and only translation of Moby Dick) and three competitive translatological approaches to Dickinson offered by Kazimiera Iłłakowiczówna, Ludmiła Marjańska, and Stanisław Barańczak.
5. Our own, at times differing, understandings of the task and the scope of translation.
6. Polish version of the javascript code, adapted to target language grammar by Jan Argasiński, Nick Montfort, and Mariusz Pisarski.

If traditional translation may be seen as an exchange between three different levels of agency, the first marked by the original author and their language, the second by various considerations related to the published artifact (different editions, editorial paratexts, critical annotations), and the third by the target language (Fig. 1), then in the realm of digital translation, and specifically within such a poly-vocal work as Sea and Spar Between, all of the above factors are literally multiplied! To emphasize the differences between this project and some more univocal translational projects, whether in analog or digital medium, we propose to call the translational intricacy we have encountered a distributed translation (Fig. 2).

Choosing between Existing Polish Translations: Contextual vs. Modular Translation

Since there exists only one translation of Melville’s Moby Dick into Polish, by Bogusław Zieliński, and because Melville’s word pool in Sea and Spar Between is slightly smaller than Dickinson’s (for example, one of the seven generative algorithms that can form a stanza is called “DickinsonLess”, but it has no Melville equivalent), our decisions regarding translation of Melville’s database of words were quite straightforward. Many of these words were nouns belonging to the nautical vocabulary used by sailors and whale hunters (“buck,” “jack,” “dock,” “hook,” “pike,” “sack,” “rail”). The very nature of Moby Dick, as a work of prose and not poetry, further simplified our task, sometimes reducing it to simple word search within the original English version and to finding equivalents in the Polish translation. In the case of Dickinson, who is famous for her dense, idiosyncratic vocabulary and imagery constrained by rhythm, rhyme and amplified by poetic tensions around enjambments, our task was far from easy. Emily Dickinson’s poems have been translated into Polish by numerous authors, including Kazimiera Iłłakowiczówna, Ludmiła Marjańska, Stanisław Barańczak and many others. While building the Polish glossary for Sea and Spar Between generator, we sometimes had to include several proposals for a given word. Most often, we opted for Polish words with the lowest syllabic count. For example, we chose monosyllabic “dłoń” rather than disyllabic “ręka” (the equivalents of the English word “hand,” which can both be identified in Barańczak’s translations), or monosyllabic “kunszt” instead of disyllabic “sztuka” (“art”), striving to preserve the syllabic structure and the graphic layout of the stanzas generated by the Sea and Spar Between script (Fig.3).

However, it soon turned out that taking words from poetry translations, or even finding Dickinson words used in Sea and Spar Between in Polish translations of her poems, was at least problematic, if not impossible. The constraints of poetic algorithms of Sea and Spar Between, for example, generating words with the suffix “less” out of a selection of common Dickinson words, or forming kennings derived from two (Melville and Dickinson) pools of nouns (for example, “chopbliss”/“blisschop”/“blissliss”/“chopchop”), more often than not required us to translate a given word from scratch, in order for it to fit the generative process. One good example of this is Dickinson’s “buzz,” a noun which in the Polish translation by Barańczak expands into a phrase of a fly repeatedly hitting the windows2 (the adjective “footless,” rendered in his translation by two words, “lacking foot” / “stóp pozbawiony,” can serve as another example). All the considerations that were behind Barańczak’s decision to replace a word with an image within the context of a single poem have to be put aside, if one wants to “translate” the workings of the generative rules of Sea and Spar Between. That’s why, as in the above case and in many others, we had to follow a modular and lexical line of translation, dependent on the constraints of the computer code behind Sea and Spar Between, rather than the contextual and figurative translations of Barańczak and Marjańska which relied on poetic constraints of a given work. In other words, our poetic factor was situated somewhere else – in the outcome of the generation process – and this is where we had to aim, for example, while searching for the Polish equivalent of a Dickinson word that might match a Melville word in a way that most closely resembled some of the poetic effects one encounters in the original Sea and Spar Between.
Nevertheless, in our own task we were guided by the translational methods of Stanisław Barańczak, due to his unquestionable mastery and – several translatologists agree on that point3 – a supremely faithful rendition of Dickinson’s oeuvre. Barańczak wrote: “Iview Dickinson . . . as a fundamentally modern poet, one who demands that her idiosyncratic stylistic traits be not smoothed over or made to sound conventional4”. In her book Stanisław Barańczak – poeta i tłumacz, Ewa Rajewska points out that the unparalleled value of Barańczak's translation lies in the variety and richness of his vocabulary.

It is this variety and richness that constitute the phenomenon of his translation. For many common, quite hackneyed, neutral English words, the translator finds nontransparent, sometimes archaic, but always vivid equivalents.5

We took Rajewska’s remark as a principle while translating Dickinson’s words. If we wanted to give one example, it would be “doll,” which – according to Emily Dickinson Lexicon – is a “little one,” a “delicate creature,” a “small human-shaped form that children play6”. Even though the closest Polish equivalent for “doll” is “lalka,” we decided to use the by far more expressive “kukła,” which means “puppet,” assuming that the image of someone who is easily controllable, whose behavior is determined by the will of others (or by the forces of nature) is in perfect accord with the vision of someone “barely afloat” clinging to a spar – a vision that forms the poetic image that is crucial for Sea and Spar Between7. At the same time, due to the semantic space it opens up, the Polish word “kukła” forms intriguing kennings when it meets words from the pool of Melville’s nouns (e.g. “dollchap”, “bulldoll”, “dollbag”).
As Stephanie Strickland and Nick Montfort emphasized in our correspondence, this approach was in line with their vision behind Sea and Spar Between, where the computer code’s unexpected grouping of words conforms in its own way to Dickinson’s poetics. The choice of Barańczak as our guide to Dickinson’s poetry translations was further motivated by his dialogic perspective on translation, where every translation is a dialogue with the original text, in which the translator’s creative individuality is taken into account.

From Grammar to Code and Back

The javascript code in which Sea and Spar Between poetry generation rules were written has greatly influenced the Polish translation. The modular approach, with words extracted from the original context in poems regardless of their translated form in Polish, had to be adopted in order to approximate the most important trait of Sea and Spar Between – the generative concept underlying the formation of kennings and random selection of lines with Dickinson’s “less” words. But this movement of special protocols sent by the demands of code to the target language was reciprocal. We would frequently encounter an inverse phenomenon, where the target language demanded changes in the code; as a result, some fragments of the code had to be rewritten.
A major problem with the Polish grammar and English-oriented generative algorithms stemmed from the inherent differences between the two languages: different syntax, gendered verbs and longer words in the target language (the syllable count of a single word is often two syllables higher in Polish8) required some important additions to the javascript code. The easiest one involved the necessary splitting up of “dickinsonLess”, “dickinsonSyllable” and “melvilleSyllable” groups into three separate sub-groups, in order to differentiate between feminine, masculine and neuter grammatical genders and the addition of new, gender-sensitive variables to the code. Instead of a single variable “var dickinsonFlatButLessLess,” the Polish version has three variables: “var dickinsonFlatButFLessLess,” “var dickinsonFlatButNLessLess” and “var dickinsonFlatButMLessLess” (F, N, and M for grammatical genders). Grammatical considerations caused even more complications in butLine() (its instances might be “then worthless is the sky,” “but guiltless is the earth”). The last of the building blocks of the line, butEnding(), may have only four values: “sky”, “sea”, “earth”, or “sun”. The preceding phrase “worthless is the” can be drawn from the same pool of “dickinsonLess” words as in the original in three cases (“sky,” “sea,” and “sun,” which are neuter in Polish). But it was enough for just one word (“earth”) to have a different gender (feminine in Polish), with the target language unable to offer any sensible equivalent of neuter gender, that the code had to be rewritten to accommodate this difference. As a result, our programmer had to write a new variable and a new function:

function butFLine(n)
{
var a, b, c = n % butFEnding.length //n % dickinsonButFLessLess.length;
n = Math.floor(n / dickinsonButFLessLess.length);
b = n % dickinsonFlatButFLessLess.length;
n = Math.floor(n / butBeginning.length);
a = n % butBeginning.length;
return butBeginning[a] + ' ' + dickinsonFlatButFLessLess[b] + ' jest ' + butFEnding[c];
}

Kennings

One of the biggest challenges on the level of both grammar and code was the task of translating Sea and Spar Between’s compound words (kennings) and maintaining them within the limited space of a single line and its pre-made structure9. The idea behind the use of kennings, a poetic structure of Viking nautical origins, here inspired by Melville, was to juxtapose words from two authors using contrasting thematic vocabularies, and have the generator randomly reshuffle their order in the compound. As such, kennings, generated by two arrays of one-syllable nouns (“melvilleSyllable” and “dickinsonSyllable”) are the heart and soul of Sea and Spar Between. Unfortunately, in Polish the number of one-syllable nouns is quite limited, with two- and three-syllable verbs forming the majority of the language reservoir. This would mean that the line containing a kenning might extend up to four syllables and effectively break the process of drawing the stanzas on HTML5 (digital) canvas. Additionally, the traditional structure of the kenning in Polish poetry is much different from the English one, taking the shape of two separate nouns remaining in a fixed genitive relationship, with grammar-enforced reshaping of the first possessive element. For example, “kot” and “kołyska” (“cat” and “cradle”) can form a kenning “kocia kołyska” (not “kot kołyska”), which is the equivalent of “cat’s cradle”; yet with the first element altered by the suffix “cia,” this involves much more modification than using the simple possessive “s” in English10. Two main problems that emerged – how to avoid exceeding the length of the original by too many syllables, and how to employ a completely different kenning structure in the code and the composition of the respective line – were solved by interventions in the syntax of the line and by additions to the code. The necessary change in the composition of the compoundCourseLine() was the most drastic one, but owing to a fortunate grammatical coincidence, it also turned out nonintrusive and natural. The three-component structure “set upon the” + kenning + “course” is generally not allowed in Polish unless the last element, “course,” stands at the beginning. Therefore we have decided to (syntactically) merge the first and the third component into an expanded entity that would (semantically) retain the meaning and the role of the two. The compoundCourseLine() took the shape of “set the course to” + kenning; this also let us save some syllables, as illustrated in figures 3 (Fig. 4) and 4 (Fig. 5):

The necessary change in the javascript code was similar to the gendered multiplication of “dickinsonSyllable” group but consisted of splitting up the rules for compoundCourseLine() into two: “syllableNominative” and “syllableGenitive” arrays. Overall, the danger of overpopulating one line and making it disproportionately longer than other instances of Sea and Spar Between verses was eliminated rather smoothly, although reducing the modular aspect of stanza generation from four elements (“courseStart” + syllable + syllable + “course”) to just three (“courseStart” + “syllableNominative” + “syllableGenitive”) might seem controversial. The polished outcome of the Polish version was also due to a bit of luck: “course” in Polish (“ster”) is also a one-syllable word, which results in an identical syllable count for the Polish variant of “course start.” In other words, the doubled semantics of the Polish “courseStart,” which had to combine “courseStart” expression and “course” string, are of the same length as the English original.

Dickinson “Less” Words and Neologisms

As mentioned earlier, more often than not in we were “forced” to build the vocabulary used in the Polish version from scratch. The task was performed almost literally, with little or no help from the existing translations, while working on adjectival compounds ending with the suffix “less11”. These word structures, so distinctive of Dickinsonian style, are used in two of the seven rules for stanza generation. Finding the Polish equivalents for words ending with “less” was an intriguing and engaging process. The Sea and Spar Between code generates them in a modular manner, adding the suffix “less” to a one-, two- or three-syllable stem. Implementation of an identical structure in Polish is virtually impossible for a number of reasons, including the three gendered suffixes for adjectives and frequent stem alternations depending on the etymological traits of a given word. On top of that, “less” structures in Polish use two alternatives for the very “less” in question: “nie” or “bez.” Thus we decided to simulate (rather than recreate) the generative process by populating the arrays of “less” words with pre-made conjoined structures. Decisions whether to use one of the Polish equivalents of “less” – “nie”, or “bez” – before the stem, as well as other decisions, had to be made according to the rules of grammar, word formation and – last but not least – to the semantic scope of the generator. A long array of words with more than 60 adjectives was created and divided into three sub-arrays of feminine, masculine, and neuter forms. There were 23 neologisms among them. Although it would be relatively easy to find in a Polish lexicon words that would fit the meaning of a given Dickinsonian neologism, we wanted to grasp one of the most crucial qualities of Dickinson’s style – its novelty and idiosyncracy. Thus instead of translating the word “droughtless” simply as “niewysychający,” which means “perennial,” the neologism “niedoschły” was created, which apart from containing the “perennial” also implies “might have been” / “would be.” The aim of this word play was to achieve a multiplication of meaning and semantic condensation within a formalized language structure, something much in line with the methodology of Stanisław Barańczak, although he had created fewer than 10 Dickinsonian neologisms (most of the potential neologisms were periphrased, a tool which in our case could not be used because the generator demanded single words12).

Summary

The case of “less” formations, in which the target language reinforces the structure of code and philosophy of the original poetry generator, was just one of several issues we faced while translating Sea and Spar Between. Many of them, if resolved inappropriately, might have started a process resulting in a certain imbalance of relations between the most characteristic traits of the original. For example, a traditionally oriented translator might focus mostly on the on-screen output, disregarding the code or at the most treating it as a means to an end, a tool for generating stylistically smooth and grammatically correct poetry. Saving the fact that neither Melville's fiction nor Dickinson's poetry were smooth or transparent linguistically and stylistically, such an approach would result in overpopulating the code with numerous exceptions, additional functions and variables tailored to accommodate the hardly modular grammar of our Slavic language. The end result of such a treatment would certainly lack the computational, generative feel of the resultant sea of stanzas. Overpopulation of code in this case would mean “decomputation” of the on-screen outcome, while failing to apprehend Sea and Spar Between as a framework for unexpected, even gritty and rough encounters of words. One can also imagine a code-sensitive translator whose ambition is to match precisely or as closely as possible the two javascript files, English and Polish. This would result in a version disregarding gender differences and, consequently, with fewer stanzas on canvas. Subtracting just one word, “earth,” from the basic pool of four nouns in the butLine() (in order to preserve the original number of variables and to avoid additional lines of code for feminine gender “ziemia”) would decrease the number of generated stanzas by at least a million if not more in a work with the output counted in trillions.

These two examples mark the extremities of a wide spectrum of approaches toward multi-layered digital works with many actors involved both in the process of creation and in the process of translation. Finding the balance between modular and non-modular (manual, or arbitrary) methods and minding the risks of overpopulation or depopulation (of code or canvas) was something we were aware of. To be honest, we found ourselves gravitating toward the traditionalist end of the scope, sometimes wanting to smooth things out too much, but fortunately, Stephanie Strickland and Nick Montfort pointed out on several occasions that this would not be the best of options.

The conclusions that come from our conversations with the authors of Sea and Spar Between and from our own readings of the translated generator, encourage us to let the generative nature of the work reveal itself rather than remain hidden. This is true both in the case of kennings (similar generative roughness occurs in the original and in the translation), and in the case of gendered structures (occurring only in Polish). The lesson we have learned over time, after several attempts to smooth things out, makes us increasingly inclined to avoid polishing up the work excessively while translating it into Polish, as this will help us remain faithful to the original spirit of Sea and Spar Between.

Notes

1. 1. As also discussed in Drucker’s later publications.back

2. 2. I refer here not only to interactive literary installations (instrumental texts) such as Text Rain, but also to the much more popular mobile literary applications which allow the reader to interact with illustrations (like Pinocchio by Elastico) and printed books which speak to the reader by their texture (like Świątynia kamienia by the Polish author Andrzej Bednarczyk, which contains a real stone inside). All such texts involve what Bouchardon calls “figures of manipulation” (Bouchardon 2014).back

3. 3. As I am comparing new media literary practices to print ones, it might be better in this context to use, instead of the category multimedial, the term (proposed by Gunther Kress and Theo van Leeuwen in Multimodal Discourse: The Modes and Media of Contemporary Communication, 2001) multimodal, as understood by Grzegorz Maziarczyk in his The Novel as Book. Textual Materiality in Contemporary Fiction in English (2013). Multimodality seems more useful than transmedial descriptions because it permits describing usage of different levels of text materiality (ex. words, images, photos, page typographic arrangement or book construction) in one medium (as print) whereas they could also been used (even in the same configuration) in the other medium. back

4. 4. And, as the back cover advertisement explains, you always can close the paper covers if you are too horrified by the dinosaurs. back

5. 5. Actually the most popular form of interaction with AR-projections – used also in newest Carlton Books publications – are AR-cards, keyboard use, which is a little inconvenient, is not necessary. [6] The full documentation of the project (sufficient to observe some puzzling aspects) is available on Youtube (http://www.youtube.com/watch?v=M7bZxZXy2o8). back

6. 6. back

7. 7. Exemplified by, but not limited to, incoherent illustrations (in “Dinosaurs Alive”, for instance, there is a dinosaur which does not open its jaws when roaring). back

8. 8. It is sometimes made easier on mobile devices. back

9. 9. This fact seems important as Billinghurst (having been working with AR technology for almost 20 years) underlines that good AR technology implementation offer the user an experience which is intuitive and in which there is no training necessary (http://www.youtube.com/watch?v=5-1CdrFxT8k). back

10. 10. It is worth to underline that in the quoted excerpt Billinghurst uses the phrase “visually augmented book”. back

11. 11. So, in the case of “Oka-leczenie” (regarded as a key liberary work), they choose an invisible text form to tell about that which cannot be seen. They also choose to compose the book of three volumes combined in a way that the one which describes what is between death (the topic of the first volume) and birth (the topic of the third volume) is situated exactly between the other ones. But, as this part of our life (the time between death and birth) is considered a “normal” one and we is often unnoticed, the second volume cannot be seen at first glance due to the fact that it is supposed to be opened from the opposite side. For more on this book, see Przybyszewska 2014. back

12. 12. Also quoted by Borsuk as an example of an inspiring work (Poole 2012). back

13. 13. Page also wants to define the relation between the characters in this way. back

14. 14. It could be added here that the Italian Scaramuccia (Eng. Scaramouch) means “skirmish”, and this is exactly the word used to describe Page’s and Screen’s correspondence in the book. The name, of course, has the same PIE root as the hero’s name. back

15. 15. This is a rule in the whole book: the animations develop the letters presented in a classic way. back

16. 16. Pressman talks here about bookishness. back

17. 17. Drucker points out: “The peculiar construction of the materially insignificant but materially based nature of signifier is essential to the paradoxical structure of Saussure’s sign” (p. 22). And she adds: “De Saussure created a difficult-to-resolve paradox between a sign independent of all materiality and also dependent upon it” (p. 27). back

18. 18. From the Polish theory of liberature one could quote here Radosław Nowakowski, who posits that we have to remember about the corporeality of letters, which are not speechless (Nowakowski 2002). Similarly, it is worth to underline that the liberary sign was described by Katarzyna Bazarnik as one which joins and uses, in the process of communication, all levels of the signifiant and signifié (Fajfer 2010: 162-163). back

19. 19. Shklovsky’s essay Art as Device speaks about the automatization of our lives in greater detail. back

20. 20. The intuitive reception about which Bouse was dreaming is certainly something difrent that automatic reception. back

21. 21. E.g. the book form in the case of print literature. back

22. 22. “Liberature” (and “liberariness” or “liberary”) openly plays with “literature” (and “literariness” or “literary”), underlining that every liberary work is designed to be read because it is an instance of a literary work. The Latin roots of word “Liberature”, however, should be noted as well: 1. lĭběr, lĭbri: a book (being the central element of liberary theory) 2. līběr, līběră, līběrŭm: free (because the liberary author should be free from all conventions, including the convention of a book form), and 3. lībră, lībrae: balance (because every element of a liberary text, even of its book form, is scrupulously balanced). back

23. 23. It could be illustrated by Fajfer’s words from 2002: “The physical object ceases to be a mere carrier of text;/the book does not contain a literary work, but/ it is itself the literary work” (Fajfer 2010: 44). back