Article body

Introduction

This article traces the genesis of the Portuguese translation of Edgar Allan Poe’s poem “Ulalume” published by Fernando Pessoa (1888-1935) in January 1925 in the Portuguese journal Athena. Using genetic analysis, it examines Pessoa’s interventions during the various stages of the translation process in relation to reflections on poetry translation found in the poet’s archive. Emphasis is placed on how Pessoa’s attention to the rhythmic aspect of the text – considered an essential element of poetry translation by the Portuguese poet – influenced his translation choices. Additionally, this study seeks to contribute to the field of Pessoan studies by filling a substantial gap in genetic-based research on Pessoa’s translations, which have received scarce attention from this perspective.

In fact, until recently, critical attention to Pessoa’s translations has been driven mainly by editorial initiatives (e.g. Pessoa, 2018; Saraiva, 1996; Baptista, 1990). Within the context of critical-genetic studies, Teresa Filipe’s article “Pessoa, tradutor sucessivo de Shakespeare [Pessoa, Successive Translator of Shakespeare]” presents a complete transcription of Pessoa’s (almost) integral translation of Shakespeare’s The Tempest, which the poet never published (Filipe, 2018). Another laudable addition is the Guimarães publishing house’s “Pessoa Editor” collection, which features a number of Pessoa’s works originally intended for his own publishing house, Olisipo, including translations of Edgar Allan Poe. Margarida Vale de Gato’s edition of Poe’s poems incorporates Pessoa’s translations, where possible. The edition also provides insight into the translations’ genetic documentation and includes facsimile reproductions and transcriptions of relevant documents in the final section (Poe, 2011).

A few works on the genesis of Fernando Pessoa’s translations that are not exclusively associated with editorial projects have been published in recent years. In his volume Do Caos Redivivo. Ensaios de Crítica Textual sobre Fernando Pessoa [Chaos Revived. Critical Essays on Fernando Pessoa] (2018), Luiz Fagundes Duarte presents an exhaustive description of the genesis of Pessoa’s translation of eight epigrams from the Greek Anthology[2] and attempted translations of three poems by Horace, as evidenced by several documents in Pessoa’s archive (see below). Also, Claudia J. Fischer’s article “Auto-tradução e experimentação interlinguística na génese de ‘O Marinheiro’ de Fernando Pessoa [Self-translation and Interlingual Experimentation in the Genesis of Fernando Pessoa’s ‘O Marinheiro’]” (2012) analyzes Pessoa’s self-translation of his play “O Marinheiro,” suggesting that what appears to be a straightforward self-translation from French to Portuguese is in fact the result of a multilingual poetic creation process. Finally, also worth mentioning is João Dionísio’s study “Latency, Inference, Interaction: Notes Towards a Blurry Picture of Translation Genetics in Portugal” (2021), included in the volume Genetic Translation Studies: Conflict and Collaboration in Liminal Spaces (Nunes, Moura and Pacheco Pinto, 2021). Dionísio argues that the genesis of Alberto Caeiro’s[3] poem “Agora que sinto amor” was likely inspired and influenced by Pessoa’s partial translation of an English poem by Alfred Austin.

Aside from these rare exceptions, the genetic study of Fernando Pessoa’s translations has received little critical attention. This situation reflects both the paucity of volumes devoted to the critical-genetic interpretation of Pessoa’s work and the overall state of research production in the field of genetic translation studies, a discipline which, situated halfway between translation studies and genetic criticism, examines the creative processes behind the genesis of a translation. There are a number of practical reasons for this. Insofar as translation often serves a mainly utilitarian function and its literary value as a creative act is sometimes still underestimated, translators rarely take the initiative of preserving their own manuscripts. Consequently, translation manuscripts currently represent “un matériau plus rare encore que les manuscrits d’écrivains” (Hersant, 2018, p. 1). According to Antonio Lavieri, this gap is directly related to the “transparency” associated with the figure of the translator, whose activity is often relegated to that of a “copyist” rather than being considered the work of an author (2015, p. 22). Most genetic documentation related to translation is typically found in the archives of canonical authors, but archival research usually focuses on the original writers’ production.

Thanks to the newfound fame of some translators, the situation has gradually changed in recent years. Libraries and institutions have begun dedicating space to translators’ archives, thus drawing attention from scholars interested in studying the translation process (Cordingley and Hersant, 2021, p. 12). As a result, the past decade has seen a moderate increase in research produced in the field of genetic translation studies. In 2014, the journal Genesis devoted its 38th issue “Translate” to genetic translation studies, and, in 2015, the journal Transalpina published a special issue “Genèse des traductions et communautés de pratique.” In the Lusophone context, the University of Lisbon held an international colloquium on the theme of “Unexpected Crossings: Translation and Genetic Criticism Studies” in November 2017. The three scholars responsible for organizing the colloquium (Ariadne Nunes, Joana Moura, and Marta Pacheco Pinto) also co-edited the above-mentioned 2021 volume Genetic Translation Studies.

Methodologically, genetic translation studies combine two complementary approaches: the first focuses on translation as a process and aims to reveal and interpret the genetic phases underlying this process; the second, closer to the field of cognitive sciences, seeks to identify the methods adopted by a translator while translating (Cordingley and Montini, 2015, p. 1). According to Jeremy Munday, the descriptive and analytical potential of translation studies methodology is intrinsically limited if not combined with the study of the translation process, which makes it possible to observe decision-making processes (2013, p. 135).

In general, and consistent with Fabienne Durand-Bogaert’s analysis, translation is a retrospective process that retraces the creation of the original text to reproduce it in a different linguistic and cultural system and is thus, in itself, the practical accomplishment of genetic procedures: “Les procédures de la génétique, la traduction les met en oeuvre à tout instant: son être même est déjà performatif” (2014, p. 5). For this reason, the critical approach adopted by translation studies can easily be applied to studying the evolution of a translation by drawing on the three methodologies proposed by Antoine Berman (1986, p. 102): analysis of stylistic characteristics, reconstruction of the context for which the translation is produced, and identification of the translator’s objectives. Durand-Bogaert maintains that, even if these three approaches were not originally intended for critical-genetic studies, they can be useful tools for interpreting the creative processes underlying translated texts.

In this context, a critical-genetic investigation focused on an original text does not significantly differ from one that aims to reconstruct the creative path of a translation. However, one of the key distinctions between these two strands of genetic study lies in the type of genetic materials available and composition of the avant-texte. Typically, the genetic materials associated with a translation are much more limited than those associated with an original production; due to the direct nature of the translatory act, there is often a lack of schemes and structuring projects (Durand-Bogaert, 2014, p. 26). However, “peri-textual” and “epi-textual” sources (Lavieri, 2015, p. 24) acquire special relevance in the genetic dossier: materials like introductions, prefaces, postfaces, and notes provide vital information on the translation process.

Although these can be a valuable resource for better understanding translation processes, Toury has foregrounded the often fictionalized nature of “extratextual sources,” particularly in the archives of professional translators (1995, p. 65). Munday emphasizes their documentary value, suggesting that while they may be intended to promote or justify the translation, they still represent a direct expression of the translator’s reflections (2013, pp. 126-127). In the case of Pessoa’s translation of “Ulalume,” such reservations seem unnecessary, as he was the literary director of the magazine in which the translation was published and therefore had no need to negotiate its publication. The extratextual sources, which consist mainly of the poet’s reflections on the theory and practice of poetic translation, can thus help us clarify his vision of translation as a practice. The statements found in these documents will be corroborated through examination of the methods adopted by Pessoa during the translation process.[4]

The following analysis will examine two autograph manuscripts which form the genetic dossier of Pessoa’s translation of “Ulalume,” along with the marginalia found in his privately owned copy of an anthology of Poe’s poetry. Both the manuscripts and the marginalia have been consulted through digital reproductions. Fernando Pessoa’s manuscripts are currently preserved in the author’s archive, located in the National Library of Portugal, in Lisbon, while Pessoa’s private library is located in the Casa Fernando Pessoa, also in Lisbon. All the volumes from the Pessoan library have been digitized and can be consulted online.[5]

Pessoa: A Translator

When the poet Fernando Pessoa published his translations of three poems by Edgar Allan Poe, “The Raven,” “Annabel Lee,” and “Ulalume,” in the Portuguese journal Athena in October 1924 and January 1925, it was not the first time he had donned the guise of translator. Returning to Lisbon after spending his teenage years (from 1896 until 1905) in Durban, South Africa, the Portuguese poet had taken advantage of the anglophone education he received at Durban High School to work as a professional translator, notably for trading companies (Zenith, 2021, p. 202).[6] Later on, Pessoa also had the opportunity to devote himself to literary translation, having been commissioned by the American editor Warren F. Kellog to translate some English and Spanish poems to be integrated into the Biblioteca internacional de obras famosas, a 24-volume anthology of the world’s greatest writers’ works translated into Portuguese.[7]

Though Pessoa’s first literary translations were published as the result of professional commitments, the documents found in the poet’s archive reveal that, besides practicing translation as a professional activity, Pessoa had already been dabbling in literary translation from a very early age, from and into many languages: Portuguese, English, French, Spanish, Latin, and even ancient Greek. His translation of eight epigrams from the Greek Anthology published in the journal Athena in 1924 were produced, as declared by the poet himself in a document found in the archive (BNP/E3, 123-102-103), using English as an intermediate language (see Lopes, 1993, p. 219). The bilingual edition, in English and Greek, of the anthology used by Pessoa for his translation is still extant in his private library (CFP 823-5).

In November 1909, Pessoa founded his own publishing house, the Empreza Ibis, Typographica e Editora, with the intention of publishing several translations, including works by Aeschylus, Byron, Shelley, and Stevenson (Pizarro, 2007, pp. 130-131). However, his plans as an editor and translator never came to fruition and the publishing house was sold in mid-1910 (Zenith, 2012, p. 155). Despite the setback, Pessoa continued to engage in literary translation projects in the years that followed. In 1921, he formed a new publishing house, the Olisipo, to publish his own Portuguese translations of works by renowned authors of anglophone literature, such as Shakespeare, Coleridge, and Poe, along with some works of classical literature, including Aeschylus, Aristotle, and Sappho (Ferreira, 2005, p. 79). Subsequently, in 1923, the Portuguese poet invited his friend and editor João de Castro to consider undertaking the translation of some of the most well-known poems of anglophone literature’s most celebrated authors, as witnessed by a letter sent by Pessoa on 20 June of that year (Pessoa, 1999, p. 13).

The plan was to translate William Shakespeare’s most important works and publish a series of anthologies compiling the most outstanding works of other renowned authors from the anglophone world, including Edgar Allan Poe. In the same letter, Pessoa stated that his translation of Poe’s “Principais Poemas [Main Poems]” was completed and only needed minor revisions. Unfortunately, Pessoa’s project fell short of his own expectations and, as mentioned in the first paragraph of this paper, only three translations of Poe’s poems were published by the Portuguese poet during his lifetime, whereas none of the other projects were realized.

The vast amount of material in Pessoa’s archive related to translation projects demonstrates how devoted the Portuguese poet was to the activity of translation, even though almost every attempt he made to effectively materialize his translation projects came to naught. The translation-related documentation is largely heterogeneous, consisting of 1) projects and lists of the works that the poet intended to translate and publish, 2) translation drafts, and 3) unpublished reflections that provide insight into the poet’s theory of and approach to translation.[8] Among the third type of document, a famous text in which Pessoa discusses his basic criterion for selecting works to translate is especially interesting, since it reveals that this selection is based on the level of difficulty:

The only interest in translations is when they are difficult, that is to say, either from a language into a widely different one, or from a very complicated poem though into a closely allied language. There is no fun in translating between, say, Spanish and Portuguese. Any one who can read one language can automatically read the other, so there seems also to be no use in translating. But to translate Shakespeare into one of the Latin languages would be an exhilarating task.[9]

BNP/E3, 141-99

As Rita Patrício has observed, Pessoa understands translation as a process that requires overcoming the obstacles created by the challenge of transposing a given textual content into a different semantic universe and cultural context (2012, p. 290). The greater the distance between the languages of the source and target texts, the greater the difficulty faced by the translator (and, according to Pessoa, the greater the enjoyment of the task).

It is therefore not surprising that when Pessoa was choosing which poems to translate and publish for the first time as a poet, and not as a professional translator working on commission, he selected the three poems by Edgar Allan Poe. The difficulty in translating Poe’s poems lay not only in the distance between the English and the Portuguese languages, but also in the complexity of the rhythmic and phonetic structure of Poe’s works, which Pessoa intended to accurately reproduce. In his view, the transposition of the rhythmic structure of the source text is a non-negligible aspect of poetry translation:

A poem is an intellectualised impression, or an idea made emotion, communicated to others by means of a rhythm. This rhythm is double in one, like the concave and convex aspects of the same arc: it is made up of a verbal or musical rhythm and of a visual or image rhythm, which concurs inwardly with it. The translation of a poem should therefore conform absolutely to the idea or emotion which constitutes the poem, to the verbal rhythm in which that idea or emotion is expressed; it should conform relatively to the inner or visual rhythm, keeping to the images themselves when it can, but keeping always to the type of image. It was on this criterion that I based my translations into Portuguese of Poe’s “Annabel Lee” and “Ulalume”, which I translated, not because of their great intrinsic worth, but because they were a standing challenge to translators.

Pessoa, 1967, p. 74

In this posthumously revealed fragment, Pessoa points out that the rhythmic aspect of a poem not only determines its musical progression, but also helps to convey the actual idea of the text. For this reason, the poet insists on emphasizing his translations’ conformity to the rhythmic aspects of the original text. When he published his translations of the three poems by Poe in the journal Athena, Pessoa included the following description below the title of each text: “Tradução de Fernando Pessoa, rhythmicamente conforme com o original [Translation by Fernando Pessoa, rhythmically in accordance with the original].”

What does this conformity to the rhythmic aspect of the source text entail? As evidenced by the above-quoted document, the Portuguese poet assigned a dual meaning to the idea of rhythm, making a distinction between the visual and the verbal rhythm of a poem. The former corresponds to the poetic imagery established by a specific text, while the latter, the verbal rhythm, seems to coincide with the semantic meaning of the word, i.e., “a regularly repeated pattern of sounds or beats.”[10] In this sense, translating “rhythmically in accordance with the original” means preserving in the target text all of the elements that contribute to shaping the rhythm of the original poem, both musically (from the external rhyme scheme, number of syllables, and position of beats in the verse, to the anaphors, alliterations, and internal rhymes) and visually (metaphors, comparisons, and other images conveyed by the text).

As I will show in the genetic analysis that follows, while Pessoa’s transposition of the visual rhythm of the poem seems almost intuitive, his adaptation of the phonetic and musical elements of the source text is often achieved only after several reformulations. An examination of the genetic process behind Pessoa’s translation of “Ulalume,” which is the focus of the remainder of this paper, is revealing of the strategies adopted by the poet—who was also a professional translator—to overcome an exceptional translation challenge: preserving the musical aspects of the source text while producing a target text that adheres to the semantic expression of the original poem.

“Ulalume”: The Genetic Dossier

The genetic dossier of “Ulalume” is particularly suited to the purpose, since it consists of a heterogeneous documentation that reveals various distinct phases of the translation process, hence providing an opportunity to observe a variety of textual developments. Two autograph manuscripts referring to the translation of “Ulalume” are currently kept in Pessoa’s archive: the first one (Figure 1) shows evidence of an initial attempt at translation, which was interrupted by the poet after the 13th line. In the upper area of the sheet, we can read the incomplete translation of the first two lines of Edgar Allan Poe’s poem “Hymn,” which Pessoa nevertheless deleted. In the lower part of the page, separated from the rest of the text by a dividing line, Pessoa wrote down some city names (“Bristol,” “Cintrey,” “Brighton”) and scrawled signatures of one of his heteronyms — Frederick Wyatt.[11]

Figure 1

Document BNP/E3 48I-15ª. First translation attempt of “Ulalume” by Fernando Pessoa.

© National Library of Portugal

-> See the list of figures

The second autograph manuscript (Figure 2)[12] is a 5-page document that bears witness to the complete translation of “Ulalume,” with numerous authorial corrections and interventions. On the last page of the manuscript, Pessoa wrote a list of 13 poems by Poe which he intended to translate and publish, including the three (“The Raven,” “Annabel Lee,” and “Ulalume”) that appeared in Athena in 1924 and 1925.[13]

Figure 2

Document BNP/E3 65-41r. Third page of the complete translation draft of “Ulalume,” by Fernando Pessoa.

© National Library of Portugal

-> See the list of figures

The genetic dossier of Pessoa’s translation of “Ulalume” is also enriched by the marginalia extant in his copy of an anthology of works by Poe: The Choice Works of Edgar Allan Poe (CFP 8-442). The volume has been part of Fernando Pessoa’s private library since 1903, when the Portuguese poet sat the entrance exam for the University of the Cape of Good Hope, in South Africa, presenting an essay that earned him, that same year, the Queen Victoria Memorial Prize. When selecting his reward, Pessoa chose, among other books, Poe’s poetry anthology (Monteiro, 2000, p. 111), indicating his admiration for Poe’s work from adolescence.

Pessoa’s writings in the margins of the text of “Ulalume” include the translation of a few scattered words (Table 1). It seems likely, as Patricio Ferrari suggests, that the marginalia represent the first jottings made by the poet at an early stage of the translation process (Ferrari, 2008, p. 88).

Table 1

Marginalia in Pessoa’s copy of “Ulalume”[14][15]

Marginalia in Pessoa’s copy of “Ulalume”1415

Table 1 (continuation)

Marginalia in Pessoa’s copy of “Ulalume”1415

-> See the list of tables

It is interesting that, in most cases, the words noted by Pessoa in the marginalia correspond to the translation of the rhyming words situated at the end of the lines.

As the analysis of the marginalia will show, Pessoa’s capacity to capture the semantic content of the English verse and reformulate it in a way that preserves the rhyme scheme of the poem in its Portuguese version is truly impressive. For instance, when scattering the translation of the first line of the sixth stanza, Pessoa takes a fundamentally literal approach (“uplifting her finger” => erguendo o seu dedo). The first verse of the stanza, according to the rhyming scheme established by Poe (ABBABABABB), rhymes with the fourth verse (“But Psyche, uplifting her finger, [...] Oh, hasten!—oh, let us not linger!”); in order to keep the poem’s original rhyme scheme intact, Pessoa identifies and isolates the semantic content of the phrase “not linger” and finds a Portuguese word that accurately conveys the meaning of the sentence, while still rhyming with the word dedo [finger]. He ends up choosing the adverb cedo [early], turning the “no wasting time” appeal of the source-text into the urgency of “come soon” (“vindes cedo,” in the translation’s published version).

The process through which Fernando Pessoa isolates the semantic elements of the text, as reflected in the way his initial translation notes convey the original text’s meaning but also establish a rhyme scheme, appears to be part of the author’s thought process, in which he formulates the translation and then pencils in the rhyming words. As Ferrari has already noted (2008, p. 88), the segments translated by Pessoa in the marginalia of Poe’s anthology (mostly rhyming words at the end of the verse, as already pointed out) passed unaltered to the published version of the translation:[16]

Mas Psyche, erguendo seu dedo,

Disse, “Nada a esta estrella me dou—

A seu pallido ser me não dou.

Não tardeis! Não tardeis! Vinde cedo

Pessoa, 1925, p. 163

[But Psyche, uplifting her finger,

Said, “I’m not giving myself to this star—

I’m not giving myself to its pale being.

Don’t be late! Don’t be late! Come soon] (my trans.)

As can be clearly observed, the idea conveyed by the sentence “let us not linger” in the source text is preserved in the Portuguese translation, as is the rhyme scheme; however, in Pessoa’s version, the syntactic structure of the fourth verse of this stanza is different from Poe’s: “let us not linger” => “Come soon!”

The analysis of the marginalia and its subsequent comparison with the further textual developments reveals the genetic processes through which Pessoa constructed his Portuguese translation of “Ulalume” by focusing on the rhyming elements of Poe’s poem. In sum, the details found in the marginalia suggest that Pessoa prioritized the transposition of the elements at the ends of the verses in order to establish the rhyme scheme of the poem. Since he never modified the translation choices made in the marginalia, one can conclude that the rhyming words were the central pivot around which Pessoa built the rest of the translation.

“Ulalume”: The Genetic Translation Process

This tendency to prioritize the translation of rhyming words also seems to have guided the translator during the composition of the first manuscript, as we can deduce by looking at the position of the blank spaces left by the poet in the first stanza below (Figure 3):

Figure 3

Document BNP/E3 65-41v. Detail of the first stanza. The four circles mark the presence of blank spaces.

© National Library of Portugal

-> See the list of figures

As we can see, the blank spaces are found primarily in the middle of the verse. This means that, during the translation process, the poet interrupted the natural process of writing by both anticipating the transposition of the rhyming words and postponing the translation of the line’s middle textual segment. In line 5, Pessoa was able to transpose only the first two words, leaving the verse incomplete, probably due to the difficulty of maintaining the rhyming scheme.

As for the textual content, this first translation draft basically consists of a literal transposition of Poe’s original text. In this regard, it is important to note the presence of the geographical proper names “Weir” and “Auber” (Table 2), which Pessoa introduces in his Portuguese version with no alteration from the source text:

Table 2

Pessoa’s first translation attempt

Pessoa’s first translation attempt

-> See the list of tables

In the subsequent stage of the translation process, as seen in the second autograph manuscript, Pessoa removes both geographical references and replaces the names “Weir” and “Auber” with periphrasis: “Era perto do pego sombrio / Na fria floresta spectral [It was near the shadowy pit / In the cold spectral forest].”

Several studies on Pessoa’s translations of Poe’s poems have emphasized the poet’s translation of proper names. As Margarida Vale de Gato has observed, in the three published translations of poems by Poe, the Portuguese poet consistently erases the name of the heroines:

In “Annabel Lee,” the heroine’s name is only kept in the title, and she becomes throughout the stanzas “the [beautiful] one I knew how to love” (aquela/a linda que eu soube amar). The same is true for “Lenora” who disappears completely from Fernando Pessoa’s “O Corvo” and whose emblematic quality is enhanced by the fact that the name’s omissions seem to prompt translation into metalanguage.

2010, p. 122

According to Portuguese essayist and translator João Barrento, Pessoa’s systematic erasure of proper names in his translations of Poe is revealing of the process through which he adapts the source text to the Portuguese poetic model, “neutralizing and deterritorializing the poem in the translation, in order to provide it with a wider universality of meaning” (2002, p. 174).

While this erasure reflects the domesticating (Venuti, 2005, p. 15) translation strategy adopted by Fernando Pessoa, the presence of “Weir” and “Auber” in the first autograph manuscript nevertheless sheds new light on his approach to the transposition of proper names. I believe that Pessoa’s decision to eliminate the explicit geographical references was due not only to his desire to adapt the text to the target culture but also to the difficulty of introducing the foreign names into the Portuguese rhyme scheme. Moreover, since this pattern is not exclusive to the translation of “Ulalume,”[17] one can assume that in the first stage of the translation’s genesis, Pessoa tried to introduce the names “Weir” and “Auber” in the target text, and then decided to substitute them with textual periphrasis later.

The only exception to the erasure of the names of Poe’s heroines is “Ulalume,” which Pessoa retains not only in the title of his Portuguese version of the poem, but also in the text itself (see Table 1). This decision was likely based on phonetics. The word “Ulalume,” a neologism created by Edgar Allan Poe, fits within the phonetics of the Portuguese language and is pronounced similarly in English and Portuguese (in both cases, the second “u” is stressed and the final “e” is silent). Poe’s neologism thus presented no difficulties for the translator, who could easily integrate it into the Portuguese rhyme scheme; in fact, the words rhyming with “Ulalume” were chosen by the poet at an early stage of the translation, having been added as apostils on the margins of The Choice Works of Edgar Allan Poe. Additionally, the word “Ulalume” is reminiscent of the Portuguese word lume, which means either “fire” or “light.” As has already been pointed out by Margarida Vale de Gato, the term “might even bear in Portuguese a greater evocative power of the lurid ‘luminescence’ the source-text ascribes to the dead beloved woman” (2010, p. 122).

In this context, Pessoa’s maintenance of the name “Ulalume” is not incompatible with his domestication strategy, which becomes more evident in the later stage of the translation process. In the autograph manuscript 65-41-43, which, as already mentioned, includes a complete translation of “Ulalume,” Pessoa radically reworks his first translation attempt. By revising the textual content of his translation, he moves away from the original poem, rejecting the mere literal transposition of the source text, as we can observe by comparing the translation of the first stanza in both manuscripts (Table 3):

Table 3

Pessoa’s second translation attempt

Pessoa’s second translation attempt

Table 3 (continuation)

Pessoa’s second translation attempt

-> See the list of tables

The lexical alterations introduced by Pessoa in the second manuscript attenuate the traces of the source text in the Portuguese version. Thus, in the first line, the adjectives sóbrio e cinzento, literal translations of “ashen and sober,” are replaced by livido e frio [livid and cold]; in the second line, the literal translation of “crisped and sere,” mirradas e seccas, is replaced by a periphrasis, de um louro mortal [deadly blondness]. In the following lines, the textual rephrasing also shows a detachment from the source text, clearly evident, as already mentioned, in the erasure of foreign names “Auber” and “Weir,” which are both replaced by vague allusions to geographical places: região spectral, pego sombrio, floresta spectral [spectral region, shadowy pit, spectral forest].

Moreover, the authorial corrections made by Pessoa in the second manuscript seem to have been realized with the intention of creating a deeper correspondence with the rhythmic pattern of the source text. In line 5, for example, the rewording that occurs during the translation, which can be verified by comparing the two autograph manuscripts, replaces D’aquelle anno □ [of that year] with No fim do meu anno fatal [At the end of my fatal year], introducing the consonantal nasal alliteration that also exists in the English version of the verse: “Of my most immemorial year.” In line 8, the substitution of lago [lake] with pego [pit] creates a phonetic redundance (perto do pego [near the pit]), which echoes the effect of the alliteration of the dental consonants existing in the source text (“down by the dank tarn”). The same happens in line 9, in which Pessoa replaces the adjective obscura [dark] with fria [cold], hence creating the repetition of the “f” sound (na fria floresta [in the cold forest]) and recreating the alliteration in the original poem (“woodland of Weir”).

In the following pages of the manuscript, we see more examples of interventions that indicate Pessoa’s efforts to maintain the same number of source-text alliterations in the target text. Note, for example, his interventions at the beginning of the third stanza (Table 4): the different textual reformulations scrawled one after the other on the page suggest that the poet was seeking a solution that would allow for the phonetic repetition of the “f” sound, thus recreating the alliteration of the original line (“our talk had been serious and sober”):

Table 4

Genetic transcription of first verse, third stanza

Genetic transcription of first verse, third stanza

-> See the list of tables

The final translation of the verse reproduces the alliteration effect of the source text (“our talk had been serious and sober” => cada um no fallar fôra frio). Furthermore, observing Pessoa’s numerous translation attempts for this line, we can deduce that some alterations have been made to maintain the stress pattern of the original verse; the verb fallaremos [we will talk] that Pessoa erased in the manuscript, despite preserving the reference to the first plural person as in Poe’s version (“our talk”), would have created an anapest foot (with two unstressed syllables followed by a stressed one), thus breaking the iambic rhythm that Poe conceived for the beginning of his line.

The example cited above shows that the establishment of an internal accentual rhythm respecting the predominantly ternary rhythm of the source text often coincides with the use of other rhythmic-phonetic aspects that are intricately connected to it. However, a number of rare, small emendations, which apparently serve to correct the stress pattern of the verse, can also be observed. This form of intervention typically affects only a minor component of the line’s structure and usually involves the use of simple expedients to modify the position of the stresses within the verse. This is the case, for instance, in line 5, in the third stanza, in which Pessoa makes use of a syncope (Nem p’la noite do anno fatal), changing the two initial words of the verse into monosyllables. This allows the first accent of the line to fall on the first syllable of noite, thus achieving correspondence with the first anapestic foot in the English verse (“And we marked not the night of the year”).[18]

In the fifth line of the seventh stanza, the lexical inversion and the addition of the plural desinence in the word esperança [hope] (<Com>/De\ <esperança> [↑belleza] and <belleza> [↑esperanças] a flux) create a subtle yet effective alteration in the text. Adding the letter “s” to esperança and changing the word’s position within the line allow the text to maintain the anapestic rhythm in the last foot of the verse, which, previously, with the synalepha of the last syllable of the word belleza and the article “a” (De esperança e belleza‿a flux), had a binary rhythm, unlike the preceding feet.

There are also other corrections made by Pessoa that seem to have been introduced with the purpose of maintaining the same position of the beats as in the source text. For example, on the fifth page of the manuscript (65-43v), Pessoa substituted the pretérito perfeito of the verbs affaguei and beijei (which corresponds to the past simple of the original line “thus I pacified Psyche and kissed her”) with the present tense of the same verbs (affago and beijo). This modification anticipates the position of the beat in the penultimate syllable (being the last word beijei, the stress would have been on the last syllable of the verse), respecting the position of the beats in the original catalectic trimeter:

Thus I pacified Psyche and kissed her

E a Psyche ‿ eu affago ‿ e a beijo[19]

The word beijei is one of the words noted by Pessoa in the apostils added to The Choice of Works of Edgar Allan Poe. It is one of the few cases in which he slightly modifies the rhyming words chosen in the first attempt at translation in the marginalia and its correction as beijo is one of the few authorial interventions found in the last two pages of the manuscript. This part of the manuscript comprises the stanzas that Pessoa had already began translating in the marginalia of the anthology. On closer inspection of the document 65-41-43, one notices a discrepancy between the portions of the text that Pessoa had already partially translated in the marginalia and the other sections of the text. The number of corrections and authorial interventions decreases over the course of the manuscript’s composition, meaning that Pessoa intervened more during the translation of the stanzas whose rhyme scheme had not been previously established. This suggests that the apostils in the margins of the anthology indeed represent an early stage in the translation process, and that the translation of the rhyming words was a decisive phase of this process.

The final version in the autograph manuscript examined in the previous paragraphs does not differ much from the print version. Through his careful translation process, Pessoa succeeded in producing a translation that respected the textual content of the source text while also “rhythmically conforming to the original.” As far as metric is concerned, he was able to reproduce the alternation between anapestic and iambic feet extant in Poe’s “Ulalume,” creating an exact correspondence of the position of the beats in 53 lines out of 94. However, both in Poe’s poem and Pessoa’s translation, most of the feet are anapestic, and iambic feet are used only at the beginning of the verse. Most often, the discrepancy between the source and target texts with respect to the position of the beats is related to Pessoa’s introduction of an iambic foot at the beginning of the line, which, in the original poem, is made up of three anapestic feet.[20] Nevertheless, a comparison between the first two stanzas of Poe’s and Pessoa’s “Ulalume” (Table 5) clearly shows that the rhythm of the two poems is very similar:

Table 5

Comparison of the rhythm in Poe’s and Pessoa’s “Ulalume”

Comparison of the rhythm in Poe’s and Pessoa’s “Ulalume”

-> See the list of tables

Fernando Pessoa was thus able to produce a translation that corresponded to the source text both in terms of textual content and rhythmic scheme, reproducing not only the number of verses and stanzas, together with the rhyme scheme, but also rhythmic elements such as metrical foot type, internal accentuation of the verse, and the presence of alliteration and assonance. Nevertheless, the target text also resonates on its own terms in the Portuguese linguistic and cultural context.

Conclusion

To summarize what Fernando Pessoa achieved with his translation of “Ulalume,” it is useful to refer to two principles of translation theory that the poet reflected on in certain documents found in his archives. The first, already mentioned, is the idea that the rhythm of a poem contributes to the definition of its content and that “the translation of a poem should therefore conform absolutely to the idea or emotion which constitutes the poem, to the verbal rhythm in which that idea or emotion is expressed” (1967, p. 74).

The second is related to the idea of the translator’s invisibility, a concept made famous by Lawrence Venuti’s essay “The Translator’s Invisibility: A History of Translation,” first published in 1995. Pessoa seems to refer to the “invisible translator” in a text in which he reflects on the interpretation of a work of art, stating that

Uma interpretação é tanto mais perfeita quanto mais consegue fazer esquecer o objecto interpretado na própria interpretação. (É assim que uma tradução é perfeita quando parece não ser uma tradução).

ibid., p.177

[An interpretation is perfect when it succeeds in making one forget the object being interpreted. (Thus, a translation is perfect when it seems not to be a translation).] (my trans.)

These two principles, although compatible, reflect different perspectives: in the first case, Pessoa highlights the importance of establishing a resonance between the “visual and verbal rhythm” of the source text and the target text; in the second case, the poet seems to defend the annihilation of the source text, whose traces must not be visible in what he considers a “perfect translation.”

In fact, these two principles seem to reflect the different strategies guiding Fernando Pessoa’s translation process: on one hand, he carefully revised his translation to reproduce the rhythmic structure of the source text as closely as possible; on the other hand, he adopted a domesticating practice that allowed him to produce a translation of “Ulalume” that could stand on its own in the Portuguese semantic universe. Pessoa’s ability to reconcile these two translation approaches is clear evidence of his skill and experience as a literary translator.

Fernando Pessoa’s view of translation, as discussed by Rita Patrício, seems to also reconcile two facets of the translator as “tanto como criadora de um texto pleno, outro, como geradora de um outro texto que ainda pretende ser o mesmo [both a creator of a full text, and a generator of another text which still strives to be the same]” (2012, p. 311). Straddling the domains of textual reproduction and textual recreation, Pessoa’s translation practice stands midway between originality and repetition. The result is the production of a target text that renders the marks of transposition invisible (ibid., p. 312).

That said, the purpose of this article has been precisely to reveal and interpret the marks of transposition sometimes hidden in the final translation. While the correlation between Pessoa’s theories on translation as reflected in the various documents preserved in the author’s estate and his translation practice was already known, closer examination of the translation manuscripts has shed new light on his process. According to Patricio, Fernando Pessoa’s role as a translator coincides with the role of “poetic theorist” (ibid., p. 287) since the act of translating represents an opportunity for him to reflect on and experiment with poetic writing. Regarding interlingual transposition as an exercise in overcoming obstacles, as suggested by Patrício, may have motivated Pessoa to take on particularly challenging translations (ibid., p. 290).

As this article has demonstrated, one of the obstacles for Pessoa in translating “Ulalume” was the transposition of the poem’s rhythm between two languages with fundamentally different prosodic systems. In a recent article dedicated to poetic translation in Fernando Pessoa’s work, Claudia J. Fischer and Patrício Ferrari observe that the large number of monosyllabic words in the English language offers a wider range of possibilities with respect to the positioning of accents within the verse, while, in Portuguese, the greater frequency of disyllabic or trisyllabic words makes it difficult to recreate rhythms that do not correspond to the natural Portuguese prosody, such as the binary rhythm (2020, p. 334). The other main obstacle encountered by Pessoa was the transposition of proper names and phonetic effects (internal rhymes, external rhymes, assonances, consonances, alliterations, etc.), which were difficult to adapt to the Portuguese linguistic context. We have seen how the poet overcame these obstacles in two different ways. On the one hand, the genetic material suggests that Pessoa had a lucky intuition at an early stage of the translation process; thus, for example, the translation of three verses from the penultimate stanza in the marginalia of Poe’s anthology remained unaltered in the published version. On the other hand, the translation manuscripts also reveal several cases in which Pessoa overcame translation obstacles through various rephrasing and authorial corrections.

New studies on the genesis of Fernando Pessoa’s translations will undoubtedly provide further insight into his usus traducendi, whose fundamental peculiarities have been partially outlined in this article. A systematic analysis of Pessoa’s translations could also help identify constants and variations to reveal how the translations’ production and publishing contexts (professional, commissioned or voluntary) may have oriented and influenced his translation practice. The growing interest in the field of genetic translation studies and in Pessoa’s work as a translator represents a propitious opportunity to further develop this line of research.