Poema de Sete+ Traduções: A Study of Carlos Drummond de Andrade’s “Poema de Sete Faces” Anne Connor São relações perigosas mas somente no papel. Que o tradutor inocente não vá preso pro quartel. They are dangerous relationships yet only on paper. For the innocent translator is not sent prisoner to the barracks.1 Although the above excerpt from Carlos Drummond de Andrade’s poem “Versinhos de Tradutor” may seem a wry commentary on the unaccountability of literary translators, it also underlines the difficulties translators face in their work. The act of poetic translation is indeed precarious, for it constitutes the most extreme form of interpretation. This assertion stems from the fact that, unlike poetic analysis, the translator’s work in many respects limits the multiple possibilities of meaning suggested by each signifier in the poem. Forced to come up with a single equivalent in another language, the translator must ultimately make a decision as to the poem’s meaning and steadfastly adhere to it. Of course this process is not only radical, but also extremely difficult, especially when one confronts the complexities of the poetic genre, for, beyond meaning, rhyme, rhythm, technique and poetic space must be considered. Moreover, the role of the historical, political and cultural context of not only the original but also the translation itself comes into play. As Bassnett and Trivedi point out, “Translation is not an innocent, transparent activity but is highly charged with significance at every stage; it rarely, if ever, involves a relationship of equality between texts, authors or systems.”2 An excellent example of the multiple difficulties involved in the translation process can be found in the various translations of Carlos Drummond de Andrade’s landmark poem “Poema de Sete Faces” (Alguma Poesia 1930). For many years, the number 1 Carlos Drummond de Andrade, “Versinhos de Tradutor,” Poesía completa e prosa (Rio de Janeiro: Companhia José Aguilar Editora, 1973), p. 630. Translations are mine unless otherwise noted. 2 Susan Bassnett and Harish Trivedi, Post-Colonial Translation: Theory and Practice (London: Routledge, 1998), p. 2. Anne Connor 47 of translations (spanning from 1965 to 1997) fittingly mirrored the number of stanzas mentioned in the poem’s title. With the recent addition of Richard Zenith’s translation in 2015, there are now eight published translations of the poem.3 By analyzing these distinct translations and at the same time offering my own English approximation, this study reflects on the limits inherent in translating poetry, while also demonstrating how the art of translation can actually further our comprehension of “Poema de Sete Faces.” Frequently anthologized, “Poema de Sete Faces” emblematizes the Modernist’s search for a new means of expression. The fragmented self-portrait playfully examines conflicting aspects of the poet’s identity and his perceptions of the world. Much like a cubist painting, each strophe stands on its own, providing a different perspective from the others. The resulting disjointedness reflects a simultaneity of perception, which in turn opens the possibilities of interpretation. This innovative structure has led some critics to not only use “Poema de Sete Faces” as a symbol of Brazilian Modernism, but to also employ its concept of multidimensionality as an appropriate metaphor for Brazilian poetry of the entire twentieth century. Charles A. Perrone, in his study of contemporary Brazilian poetry, whose title—Seven Faces: Brazilian Poetry Since Modernism—is directly inspired from “Poema de Sete Faces,” describes the poem as emblematic of the search for a new artistic expression: In a rhetorical sense, the poetic fragmentation is a move away from linearity and dominant discursiveness. Thus the internal conflicts of the poem are not defeating, but rather constitutive of a searching self and of an artistic newness.4 Indeed, there can be little doubt as to the importance of the translation of a poem that has been used to represent the entire Brazilian contemporary poetic project. In the following analysis of the poem’s eight translations, while recognizing that many theorists consider faithful translation of poetry to be “by definition impossible” and prefer instead “creative transposition,”5 I seek to 3 See the accompanying article in this issue of Delos for a transcription of the eight translations plus my own version, as well as Drummond de Andrade’s original poem, from Poesía completa e prosa, p. 53. 4 Charles A. Perrone, Seven Faces: Brazilian Poetry since Modernism (Durham: Duke UP, 1996), p. 16. 5 Roman Jakobson, “On linguistic aspects of translation” (1956) in Theories of 48 Poema de Sete+ Traduções privilege the original’s multiplicity of meanings while at the same time keeping in mind its important positioning as a landmark Modernist poem in the greater Brazilian literary context. To begin with, a simple inspection of the different titles given to the poem in the English translations reflects the disparity among them. While both Elizabeth Bishop (1972), Mark Strand (1976) and Zenith (2015) entitled their respective translations “Seven(-) Sided Poem,” Virginia de Araújo chose a more literal route with the title “Poem with Seven Faces” (1980). In Portuguese face can mean “face,” “countenance,” “visage,” “look,” “expression,” “appearance,” “surface,” or “side.” As the English equivalent “face” may signify both “side” and “appearance,” it at first seems a better choice. However, the preposition “with” does not correspond to the Portuguese preposition “de” which in the title signifies both “made, or constituted of” and “from” (ie, poem related from seven points of view). John A. Nist (1965) and Frederick G. Williams (1996) include the element of constitution and origin in their respective literal translations: “Poem of Seven Faces.” Nevertheless, Nist commits the grave error of separating the poem into eight stanzas, neglecting in this way to connect the title to the poem’s structure. The title “Seven-Sided Poem,” which not only refers to different “sides” or faces, but also indicates the different “perspectives” or viewpoints in the poem, seems to be just as fruitful. Another option, though, would be “Seven-Faced Poem,” which includes the concept of appearance. However, there is awkwardness to the phrase “Seven-Faced Poem” which is absent in the title “Seven-Sided Poem.” Duane Ackerson and Ricardo Sternberg attempted to incorporate the signifier “face” in their title through a clever adaptation: “Poem of Seven Facets” (1972). This selection privileges the definition of face which means “side” or “plane surfaces” of a geometrical shape, while at the same time underlining the other definition of “facet” which implicates the sides or aspects of someone’s personality. The missing connotation in the word “facet,” however, is “expression” or “countenance”; nevertheless, it could be argued that the signifier “facet” does contain within it the word “face” and thus these missing meanings are implicitly suggested. Jack Tomlins’ solution, on the other hand, was to give the title its geometrical equivalent: “Heptagonal Poem” (1978). Unfortunately, although Tomlins’ title stands out for its originality, his attempt severely Translation: An Anthology of Essays from Dryden to Derrida, ed. R. Schulte and J. Biguenet (Chicago: University of Chicago, 1992), p. 151. Anne Connor 49 limits the interpretation of the poem by losing the meanings “perspective” or “view” which the word “side” generates. Eight translations have five distinct titles; this fact is really a tribute to Drummond’s artistry as a poet. Each title offers a slightly different interpretation of the poem and, read together, they serve to open the possibilities of poetic analysis. Of course, one title has to be chosen over the others in order for a translation to exist, and in this case I prefer the literal translation, as it preserves the word “face” with all of its possible connotations. In this way, Nist’s and Williams’ titles “Poem of Seven Faces” preserve the multiplicity of meanings to be found in the original. The translations of the first stanza also offer a number of marked differences. Whereas almost all of the translators opted to render the enigmatic image in the first verse, “um anjo torto,” as “a crooked angel,” Tomlins deviates from this pattern in his English rendition: “an unjust angel.” Nist and Zenith take the opposite tack by choosing the adjective “twisted.” In Portuguese, “torto” means “twisted,” “crooked,” “bent,” “distorted,” and “deformed” as well as “wrong,” “unfair,” “unjust,” or “askew.” The choice of the other five translators, “a crooked angel,” makes an effort to incorporate the idea of “twisted” with that of “dishonest” or “not to be trusted.” Again, I argue that the more successful translations are those that capture, at least to some degree, the multiple meanings which were hinted at in the original poem. In the next two lines, the tone of the poem becomes an issue for the translators. Modernism, it will be recalled, was a revolt against the staid Parnassianism, with its reverence for form and prestigious language, so the maintenance of Drummond’s conversational tone is essential in capturing the Modernist spirit. The verb “vivem” is translated as “live” in all of the English versions except those of Ackerson-Sternberg and Tomlins. The signifier “inhabit,” used by Ackerson and Sternberg, has a more scientific connotation than “vivem,” setting a tone far from the colloquialism of the original. Likewise, Tomlins’ “dwell,” which poetically may constitute an improvement on the original, is not as faithful a rendering as the verb used on an everyday basis: “live.” The third line plays quite an important role in any interpretation of “Poema de Sete Faces,” not only in the establishment of the poem’s tone, but also in the construction of the Modernist poetic project. Here, the poetic voice offers direct advice on the way one should approach life, and by extension on the manner in which one should approach literature. Perrone defines the French word 50 Poema de Sete+ Traduções gauche as “maladjusted” or “marginalized” and sees this concept as crucial to Drummond’s poetics: “the oddly formed ‘Poema de Sete Faces’ incarnates the general break from strictures of the past.”6 It seems quite clear that Drummond’s tone is highly informal at this point of the poem: “Vai, Carlos, ser gauche na vida.” The playful and rather mischievous tone of the fallen angel’s declaration is best captured in Mark Strand’s translation: “Go on, Carlos, be gauche in life.” The other translations are either too formal (Tomlins’ “Go forth, Carlos!”) or overbearingly authoritative (“Go Carlos!” from Nist, or “Carlos, go on!” from Bishop). Departing from the other translations, Zenith decided against leaving the French term in his text: “Carlos, get ready to be a misfit in life!” While the term “misfit” certainly maintains a colloquial tone, the phrase “get ready to become” implies that Carlos’ lot in life has been predestined by the angel. Yet in the original, the angel is encouraging him to go on and be gauche, suggesting such an attitude belied an inclination already within him. An interesting interpretation of the poem, which helps to support my belief that the tone should be mischievously playful, has been presented by Thomas Colchie, who identifies a connection between “Poema de Sete Faces” and another well- known poem by Drummond: “Canto ao homem do povo Charlie Chaplin” [“Song for that Man of the People Charlie Chaplin”7]: The figure of Carlitos, as Chaplin was known in Brazil, offers perhaps the greatest single key to Drummond’s poetics: the consummate artist who appears not to be an artist at all; the down-and-out clown who manages to stumble along life’s tightrope, forever nearly, yet not ever quite falling off: “Carlos, go on! Be gauche in life!”8 The connection between Carlos Drummond de Andrade and Charlie Chaplin adds yet another dimension to the autobiographical component of the poem and certainly favors not only the more playful translation established by Strand, but also the suggestion that Carlos’ gauche tendencies come from within, and are only encouraged by the angel. 6 Perrone, Seven Faces, p. 17. Italics mine. 7 Translated by Thomas Colchie. Review: Literature and Arts of the Americas 10.17 (1976), p. 71. 8 Thomas Colchie and Mark Strand, eds., Traveling in the Family, Selected Poems by Carlos Drummond de Andrade (New York: Random House, 1986), p. X. Anne Connor 51 The second stanza serves to highlight some more intriguing differences between the translations. The first two lines seem simple enough in Portuguese: “As casas espiam os homens / que correm atrás de mulheres.” However, they have been translated in seven different ways. For example, although Araújo’s translation appears much more poetically formal (“The houses mount surveillance / on the woman-chasers”), it betrays the simplicity of the original. Ackerson and Sternberg, on the other hand, at first avoid the literal translation, “the houses look at men” (not “spy” on men), but then choose literalness in “who run after women,” when the correct idiomatic expression in English would be “who chase after women.” Williams and Tomlins reverse these decisions with much more success, as the literal version followed by the idiomatic expression maintains the simplicity of the verses as well as their original meaning. The third and fourth lines in the second stanza also present interesting difficulties in the translation process. The original verse “A tarde talvez fosse azul/ não houvesse tantos desejos” is translated by Ackerson and Sternberg as “The afternoon might be clear/ If it weren’t for so many desires.” To begin with, their decision to translate “azul” as “clear” seemingly implies a narrowing of the text’s meaning. Logically, a blue sky is a clear one; however, in Portuguese “blue” also connotes happiness, as in “tudo azul,” [everything is fine] and therefore maintaining the color in the translation accentuates both the clarity and element of happiness. However, by choosing the adjective “clear,” they avoid the color blue’s negative connotation in English. The fact that in English “blue” can mean sad, as in “the blues,” creates an alternate reading in English, absent in the Portuguese. More important, however, is the fact that the ellipsis of the conjunction “se” [if] in the conditional clause of the third and fourth verses has led to two principle groups of translations. In the original verse “A tarde talvez fosse azul / não houvesse tantos desejos” Andrade’s omission of “se” has led some translators to insert it at the beginning of the lines, “If the evening perhaps were blue/ there wouldn’t be so many desires” (Williams) and “If the afternoon had been blue / there might have been less desire” (Bishop), whereas others place it in the middle: “The afternoon might be clear / If it weren’t for so many desires” (Ackerson and Sternberg) and “If desire weren’t so rampant, / the afternoon might be blue” (Zenith). These two interpretations bring up the question, is it the desires which determine the clarity/happiness of the afternoon? Or rather the lack of blueness/clarity which affects the quantity of 52 Poema de Sete+ Traduções desires? In both instances, the causal relationship between the poet’s interior feelings and the exterior natural world imply both the strength of his emotions and the frustration of not being able to control them. In my reading of the poem, the “houses” that are spying on men who chase after women refer to the societal strictures surrounding sexual desire, which has manifested itself in clearly binary, passive/aggressive gender roles. The suggestion that “perhaps if the evening were blue / there wouldn’t be so much desire,” implies that pent-up yearnings (desire) might grow stronger because of the grey skies (social restrictions). Alternatively, if “blue” refers to “clarity” and “happiness,” and the causality is reversed, “The afternoon might be clear / If it weren’t for so many desires,” then those inner desires are what is causing the outside world to be dark. The poetic voice in this case would clearly link desire with shame brought on by Catholic guilt. Either way, there is an affirmation of sensual desire and frustration as an essential part of one of the poet’s “faces.” Another seemingly small point of difference in the translation of these lines can be found in Williams’ version. Here “tarde” is translated as “evening” instead of “afternoon,” as the other five translators had done (1996, p. 223). Although “afternoon” may be accurate literally, “evening” does constitute one of the lexical exceptions for “tarde,” and also adds to the sexual tension that continues into the poem’s third stanza, as the poet is preoccupied with bare legs. O bonde passa cheia de pernas: pernas brancas pretas amarelas Para que tanta perna, meu Deus, pergunta meu coração Porém meus olhos não perguntam nada. (1973 [1930], p. 53) The third stanza loses its musicality in English no matter how it is translated. In Portuguese, the agreement of the color adjectives with the feminine plural noun produces a strong phonetic effect: “pernas / pernas brancas pretas amarelas.” Both the sameness in sound and the omission of the comma between each color work to group the modifiers together, contributing to the suggestion that all of the legs, no matter the color, are equally attractive to the poetic voice. It is impossible to achieve the same rhythm with the English word “legs.” For example, Bishop’s repetition of the word “legs” is overdone, causing the verse to sound almost like a poem by Dr. Seuss (“white legs, black legs, yellow legs”). Likewise, Williams’ and Araújo’s decision to separate the legs Anne Connor 53 into two groups, “white legs, black and yellow ones,” creates a dichotomy which did not exist in the original, privileging “white” over “black and yellow.” However, Zenith’s solution to repeat the conjunction “and” in “white and black and yellow legs” helps to emphasize the sameness in the poetic voice’s physical attraction. This stanza serves to confirm the inescapability of desire for the lyric “I”, while at the same time introducing the guilt arising from religious constraints. Even though the poetic voice’s heart might ask God why there should be so much temptation, his eyes do not question the multitude of attractive legs at all, suggesting he will go on being naturally gauche despite the Catholic social norms that try to impede him. The fourth stanza offers one noteworthy discrepancy between the translations with the adjective “raros” in the fourth line: “Tem poucos, raros amigos.” The dictionary defines “raros” as “rare,” “seldom,” “unusual,” or “scarce,” which seems to be Tomlin’s interpretation in his verse “His friends are few and far between,” as well as Williams’ rather unimaginative attempt: “He has few, very few friends.” However, other translators interpret “raro” as “rare,” in the sense of unusual and therefore prized: “He has few and precious friends” (Strand). In this instance I believe both interpretations are valid; however, in order to open the English version to both possibilities, it seems fitting to simply use the cognate “rare,” as Nist did in his original translation: “He has a few, rare friends.” The translators can be divided into two groups regarding their interpretations of the first line in the fifth stanza: “Meu Deus, por que me abandonaste.” Whereas Araújo, Bishop, Strand, and Tomlins adopt the King James Bible version of Jesus’ lament: “My God, why hast Thou forsaken me” (Matthew 27:46), which though formal sounding, when considering the influence of Shakespeare’s era on modern-day English, is arguably the most well-known translation of the Biblical verses in English. On the other hand, Williams and Ackerson-Sternberg opt to follow Andrade’s translation of Matthew 27:46 to Portuguese, without adding the embellishments from Old English: “My God why have you abandoned me” and “My God, why did you abandon me.” This latter selection by Ackerson-Sternberg seems to be the more appropriate of the two, since the original text is written in the preterit, and it again keeps in line with Andrade’s colloquial tone. 54 Poema de Sete+ Traduções The sixth stanza causes the most difficulties in the translation due to the rhyme both in the first and second verses, as well as the third and fifth verses: Mundo mundo vasto mundo se eu me chamasse Raimundo seria una rima, não seria uma solução. Mundo mundo vasto mundo, mais vasto é meu coração. Strand and Tomlins tried to recreate the first rhyme with “world” and “Harold” in their versions. The problem with this approach, however, lies in the weakness of this rhyme in English. In the context of this stanza, the poetic voice disparages rhyme for its empty simplicity. “World” and “Harold,” however, constitute a difficult or contrived rhyme which in turn goes against the poet’s argument of rhyme being simple or “easy.” The problems with Bishop’s translation are numerous. First of all, “universe” goes beyond the idea of “mundo” or “world” implied in Drummond’s poem. Secondly, the rhyme created with Eugene and “mean” ignores the structure as well as the content of the original poem. The search for a solution to a literary dilemma is completely absent in Bishop’s version. Araújo’s attempt is admirable because it manages to incorporate the poet’s central concern about the art of writing into the very same rhyme scheme in Drummond’s original: Sphere, sphere vast sphere If I were a sonneteer It would be a rhyme, not a solution. Sphere sphere vast sphere More vast is my emotion. The rhyme scheme AABAB is maintained just as it is in the original. In addition, the noun “sonneteer” recalls the Parnassians, who were of course disparaged by the Modernists. In this case, the translation helps the reader to better understand the original stanza and its implied criticism of the old poetic order. However, the noun “sphere” does not encompass the same meaning as “world” and the declaration that the poetic voice’s emotion is more vast than the vast sphere does not quite capture the original, in which the poetic voice declared his heart to be vaster than the world. Both Williams and Ackerson-Sternberg also provide interesting solutions to this dilemma. However, these two translations add a religious element that, while missing in these specific lines of the Anne Connor 55 original, certainly contribute to the overall theme of Catholic guilt within the poem. Williams uses the word “christened” instead of “named” which establishes the poetic voice’s Christianity, but the name “McEarld,” which has a very Irish ring to it, places him within an entirely different cultural context, possibly problematic for a poem that specifically represents Brazilian Modernism.9 Ackerson and Sternberg’s version also presents complications. The selection of the name Job, of course, sends us to the biblical character who lost his family, wealth, and health, yet remained faithful to God and was eventually restored to good fortune in his old age. This is an interesting choice in name, especially in light of the previous stanza in which God has apparently abandoned the poetic voice. However, the added dimension has little to do with the original poem, for Job is a very uncommon name, unlike Raimundo. Zenith takes the opposite stance by choosing the rather common name, Clyde. Unfortunately, “Clyde” is not free from cultural connotations either, and is actually used in slang to refer to an “unsophisticated person” or “yokel.”10 An important consideration missing from all of the translations is the meaning within the name Raimundo itself. Defined as “sábio protetor,” [wise protector] “aquele que protege com seus conselhos,” [he who protects with his counsel]11 the name not only remits us to Portuguese royalty, Raimundo de Borgonha (1070–1107), but also contains within it the semantemes, “rei” (king) and “mundo” (world).12 In this way, by writing “se eu me chamasse Raimundo” Drummond is pondering not just what would happen if his name were different, but if he were king of the world. González and Treece, who translate just the sixth stanza of “Poema de sete faces” in their analysis of Brazilian Modernist poetry, make an impressive attempt to include Drummond’s reference to royalty in their version: World world world so vast if my name were Henry the Last 9 It should be added, however, that a translator’s motivations are not limited to solely capturing the original context of the poem. If the intent is to recreate the poem in a new cultural context, then Williams’ Irish version offers not only a name that rhymes perfectly, but also an invitation to compare Brazilian and Irish cultural constructs. 10 “Clyde,” Def. 1. Green’s Dictionary of Slang, 2011. Oxfordreference.com. 22 Nov. 2016. 11 “Raimundo,” Def. 1. Dicionario de Nomes Proprios. Dicionariodenomes proprios.com.br. 22 Nov. 2016. 12 I am grateful to Charles Perrone for pointing out this reading to me. http://oxfordreference.com https://www.dicionariodenomesproprios.com.br/ https://www.dicionariodenomesproprios.com.br/ 56 Poema de Sete+ Traduções it would be a rhyme, not a solution. World world world so vast much vaster is my heart.13 However, their attempt fails to recreate the rhyme in the third and fifth verses, and while “the Last” refers to royal lineage, it also implies an end to it. González and Treece contend that while the “Modernists of the 1920s were the first to engage imaginatively, critically, and consciously with the processes of economic and social change which brought the initial phase of capitalist modernization in Brazil” (p. 66), Drummond preferred to adopt an alternative, ironic stance regarding how his identity tied in with economic and social change. Still, I read the problematic sixth stanza of “Poema de sete faces” as deftly speaking to the emptiness of his poetic predecessors, the Parnassians, and their facile rhymes, as well as voicing his clear allegiance to the heart over wealth and fame in the economic power struggles of the time. Given the difficulty of the sixth stanza, no existing translation of which fully expresses the original meaning, in my own translation I propose the following compromise: World world vast world If my name were Earl It would be a rhyme, but not a solution World world vast world More vast is my heart’s resolution. The assonant rhyme in “world” and “Earl” seem preferable to me over the imperfect consonance of “world” and “Harold.” While “Earl” does not imply “king of the world” it functions as a semanteme of earl, literally meaning “nobleman of a rank between viscount and marquess,” so in that way it mimics Drummond’s use of Raimundo.However, it still does not form a perfect consonant rhyme. Another problem with this translation is the change in meaning of the fifth verse. In order to achieve the rhyme with “solution” in the third verse, it is no longer the poetic voice’s “heart” which is vaster than the world, but rather his “heart’s resolution.” This, to me, however, is an acceptable modification. To end with the monosyllabic signifier “heart” for “coração” leaves the reader feeling empty-handed (see, for example, Tomlins’ version), as the stanza lacks the tight closure found in the original. The “heart’s 13 Mike González and David Treece, The Gathering of Voices: Twentieth- Century Poetry of Latin America (New York: Verso, 1992), p. 72. Anne Connor 57 resolution,” on the other hand, more successfully emulates the original rhyme scheme in the Portuguese; plus, the word “resolution” implies the adoption of a new way of thinking—or in this case of feeling, since it is the heart’s resolution—which goes along with the poetic voice’s affirmation of the heart’s eminence over wealth and power. For Drummond, poetry should no longer be limited to inflexible rhyme schemes that stultify the poet’s authentic expression. On the contrary, poetry should take into account the vastness of the poet’s heart, allowing in this way full freedom of expression. Still, even after having studied nine distinct translations, I can make no claims of superiority in my approach to the ever-elusive sixth stanza. Without a doubt, the complexity of the sixth stanza, with its interweaving of rhyme, poetic space, and meaning—not only within the poem, but on a metapoetic level—in a sense dooms “Poema de Sete Faces” to be forever lost in translation. One of the best solutions to this difficulty, it seems to me, is to footnote any translation of the poem with a brief summary of the difficulties inherent in the translation of the sixth stanza, or, at the very least, to footnote the translation with a copy of the original stanza in Portuguese. In this way, the reader will at least be aware of this loss. Finally, the last stanza deserves some attention: Mas essa lua mas esse conhaque botam a gente comovido como o diabo. The ironic note evident in the last line of the original poem has been translated into various idiomatic phrases referencing hell or the devil. Of the three translations utilizing hell—“shake a person up like hell” (Nist), “are hell on a person’s feelings” (Strand), and “make me sentimental as hell” (Zenith) — Zenith’s stands out as capturing the poet’s “comovido” or “sentimental” state. However, the decision to employ the image of hell, rather than the devil, causes the reader to miss the connection between the devil and the anjo torto at the beginning of the poem. Worse yet, Williams’ version, “put one in a crazy mood” completely eliminates the infernal overtones in the poetic voice’s predicament. Of the attempts to incorporate the reference to the devil, Araújo’s is unfortunately quite awkward (“work on a man’s feelings like a devil”) due to its lack of idiomatic accuracy. Ackerson and Sternberg come closer to the original with “moves me like the devil,” whereas Tomlins’ verse, “turn me on like the very devil” 58 Poema de Sete+ Traduções places an overtly positive emphasis on the theme of sexual desire in the poem, although such an upbeat connotation is missing from the original. In comparison, Elizabeth Bishop’s translation of this last stanza is by far the superior interpretation. Her incorporation of an English idiomatic expression, “play the devil with one’s emotions,” not only reflects and helps to explain the ambiguity of the different perspectives throughout the poem, but it also adds a light, comic note, and thus underlines the difference between the Parnassian approach to poetry, which took itself very seriously and, in general, did not create a space for humor, and its Modernist counterpart, which embraced not only humor but all topics previously deemed “unworthy” of poetry. As Carlos Drummond de Andrade indicated in his poem “Versinhos de Tradutor,” the art of translation constitutes a dangerous enterprise. Not only does it require the most radical act of interpretation, but it also brings with it the heavy burden involved in adequately representing the work of a writer from a different culture and time period, while at the same time attempting to create a consistent literary aesthetic for the poem in the target language.14 Through the eight published translations of Drummond’s landmark piece “Poema de Sete Faces,” and my own attempt, we have seen how easy it is for the original poem to slide through the many cracks that exist between languages. Although one could conclude that “Poema de Sete Faces” is doomed, in part, to be lost in translation; I believe, on the contrary, that the critical dialogue which results from a contrast and comparison of distinct translations of a single poem can actually help us better understand the original poem’s meaning. In addition, this process can only improve the art of translation, as the translators can thus learn from the shortcomings and achievements of others. 14 Of course, there can be many different motivations in translating poetry, including cultural transplantation, or the desire to “domesticize” culture-specific elements in order to make the original more accessible for readers in the target language; see Louise M. Haywood, Thinking Spanish Translation (London: Routledge, 1995), p. 80. In addition, poetry translations can be motivated primarily by creativity, leading to a much more “free” definition of translation, an idea taken to its extreme by Yves Bonnefoy: “You can translate by simply declaring one poem the translation of another”; see Yves Bonnefoy, “Translating Poetry” in Theories of Translation, ed. Schulte and Biguenet, pp. 186-192, at 186.