Article body

1. Introduction

The concepts of source text and target text are central in translation studies, yet they seem to have remained undertheorized – just like the adjacent concept of the original (Baer 2017). If translation is a text for which “there is another text, in another language/culture, which has both chronological and logical priority over it” (Toury 1995/2012: 29; see also Apter 2005), it may become necessary to locate this other text – the source text (ST). If STs are not correctly identified, the comparison of translations with their assumed STs is on shaky ground (see, for example, Toury 1995/2012: 100-101; Shengyu 2018: 38) and, by consequence, so are the theories derived from such comparisons. One problem is that the ST does not necessarily equal what is commonly understood as “the original text.”

Our initial idea was to compare Otto Joutsen’s 1916 indirect Finnish translation of Jules Verne’s Vingt mille lieues sous les mers (1869/1870)[1] with four (direct) Finnish (re)translations of the novel to see if translating indirectly really results in “a lesser degree of precision and an increasing number of deviations” (Edström 1991: 12; see also Dollerup 2000: 23) as compared with translating directly. However, the five translations are so different from each other that comparison was difficult. This observation led us to question whether these translations really were based on the same ST – even if they are all translations of the same novel. As Paloposki and Koskinen (2010: 41) put it, “categorization and labeling may be misleading” and in fact our initial categorization of the four translations as direct translations proved questionable.

As Toury (1995/2012: 94) points out, “there will always be the possibility that the assumed translation under study will be found not to have been derived from a particular assumed ST after all, or not from it alone.” In fact, although “[t]he standard Western model of translation posits a kind of exclusive, binary and unidirectional relationship between source text and target text” (Delabastita 2008: 239), the reality is often more complicated (see also Meylaerts 2006). Unfortunately, the information regarding the source text(s)/language(s) of translations on title pages and bibliographies may be inaccurate, incomplete or even lacking (Toury 1995; Poupaud, Pym, et al. 2009; Paloposki and Koskinen 2010; among others), and therefore it can be easy to arrive at wrong conclusions. The careful assessment of the ST(s) of a translation, however, could lead to the discovery that the translation has no ST and that it is actually a pseudotranslation, “a text that is presented as a translation while it is in fact not a translation” (Du Pont 2005: 328; see also Toury 1995; Popovič 1976). Or, it could turn out that the ST is not in the language in which the text was first written, as is the case of indirect translation, which is understood as “a translation based on a text (or texts) other than (only) the ultimate source text” (Ivaska and Paloposki 2018: 43, note 1), or that it is based on several STs (Rizzi 2008; Ivaska and Paloposki 2018), in which case it could be labeled a compilative translation. Similarly, retranslations, defined as “second or later translation[s] of a single source text into the same target language” (Koskinen and Paloposki 2010: 294; emphasis added), could be established to actually have different STs (see, for instance, Paloposki and Koskinen 2010), leading to the conclusions that they are, strictly speaking, not retranslations.

The importance of identifying the ST(s) of translations, however, does not lie only in being able to correctly categorize them. Rather, this is also important because, just as “any account of an instance of translation that is wrongly located in space and time […] is bound to be misleading and result in shaky or wrong accounts” (Toury 1995/2012: 19, note 2), identifying the wrong ST(s) can also lead into invalid conclusions. For example, thinking that Janina (1847/1882), Emilia Dobrzańska’s Polish translation of Charlotte Brontë’s Jane Eyre (1847), was based (solely) on an English ST, Hadyna (2016: 79) first attributed “one of the most notable errors […] to the translator’s misunderstanding of the original.” However, she later found out that the passage in question had actually been translated from a French mediating text and that the Polish translation renders the meaning of the French – the de facto ST of the passage – correctly. In other words, there is no translation error, but Hadyna could reach this conclusion only after carefully reassessing what is – or, in this case, are – the translation’s STs.

In this article, we discuss the difficulty of identifying and defining source texts and present a case study of five assumed (re)translations of one novel, of which the (re)translations nevertheless have different source texts. We adopt the notions of work and text to be able to discuss such complex textual relationships and to show how the source text does not (always) correspond to the “original text.” In addition, we show how the categorizing of translations and the kind of questions one can ask change depending on how source text is defined.

2. Theoretical background

2.1. Definitions of source text

How does translation studies define source text? Rather vaguely, if at all, it seems. For example, the Routledge Encyclopedia of Translation Studies (Baker and Saldanha 1998/2009), the Handbook of Translation Studies (Gambier and van Doorslaer 2010–2014), The Oxford Handbook of Translation Studies (Malmkjaer and Windle 2011) and The Routledge Handbook of Translation Studies (Millán and Bartina 2013) all lack an article on source text, and even on text in general. The focus on the target culture/text in recent years may explain this (see Baer 2017), but it is nevertheless surprising considering how central a concept source text is for translation studies.

Palumbo (2009: 108) defines source text succinctly, yet not very successfully, as “[t]he text to be translated, sometimes also called ‘foreign text’.” The introduction of the term foreign text does not lead to a better understanding of what a source text is. Also, we could debate whether Palumbo’s choice of tense is correct: is a text already a sourcetext before any translating takes place – even if the text is prepared with translation in mind (see Dollerup 2000) – or does it become one only after translating takes place (see Emmerich 2011; Littau 1997)?

A more extended definition of source text can be found in Shuttleworth and Cowie’s Dictionary of Translation Studies (1997):

The text (written or spoken) which provides the point of departure for translation. Except in the case of INTERSEMIOTIC and INTRALINGUAL translation, the source text will be in a different language (SOURCE LANGUAGE) from the translation (or TARGET TEXT) which the translator produces from it. The source text will typically be an original text written in SL; however, in the case of INDIRECT TRANSLATION […], it may itself be a translation of another text in another language.

Shuttleworth and Cowie 1997: 157-158

This definition views a ST as a singular, unified entity (“the text”; “the source text”; “a translation of another text”). Although translating from a translation is acknowledged, the possibility for a compilative translation, defined as “a translation which makes use of a number of source texts” (Toury 1995/2012: 100, note 4; see also Assis Rosa, Pięta, et al. 2017; Crisafulli 1999; Ivaska 2021), is not considered, and the same applies to support translation, which entails translators checking other translations of the text “in order to see whether colleagues have found satisfactory solutions to certain problems” (Dollerup 2000: 23).[2] According to the above definition, the ST and the translation are in the same language only in cases of intersemiotic and intralingual translation. Thus, a retranslation using a previous translation into the same language as a (supporting) source would not count as interlingual (re)translation.[3]

Another problem with this definition is that ST and original text are considered equal (“The source text will typically be an original text”). This relationship is problematic, as Pym discusses in his Translation Research Terms: A Tentative Glossary for Moments of Perplexity and Dispute, in which he defines source text as the “[s]tandard term for the text that you translate from” (Pym 2011: 92). He also notes that “[t]he theoretical problem is that all texts incorporate elements from previous texts, so in principle no text can be a primal ‘source.’ Common parlance refers more readily to ‘the original,’ which promotes the same illusion of primacy” (Pym 2011: 92). Pym seems to be suggesting that no text is ever truly original within the universe of all the texts in the world because some level of (unintended) intertextuality is unavoidable (see also Littau 1997; Bassnett 1998; Paloposki 2001; Scott 2006). This illusion of originality exists also on the level of an individual work. Novels, for example, can exist simultaneously in several forms, such as editions and (re)translations in various languages. Sometimes there might even be small differences between the hardback and the paperback – because someone has spotted typos or because the translator has changed their mind – like Emily Wilson, who in 2017 translated Homer’s Odyssey into English, recently disclosed on Twitter.[4] In such situations, which version is the “original”? Complexities of this and similar kinds make equating (source) text with original an oversimplification: sometimes it can be difficult – if not impossible – to pinpoint one version that is more “original” than the others.

2.2. The multiple, unstable (source) text

There are often several candidates for a ST from which the translator – or whoever is responsible for the choice of the ST(s) (see for instance Emmerich 2017) – needs to choose (Crisafulli 1999; see also Frei 2012; Kothari 2005; Wu and Fernández Díaz 2017). This can make it difficult to identify the correct ST(s) of translations, which in turn complicates the categorization of the translations. For example, Virginia Woolf’s novels were “typeset and printed in both the U.K. and the U.S. from two separate sets of proofs” (Emmerich 2017: 2), and naturally either version can serve as a ST. What if one translation is done from the U.K. and another from the U.S. version – are they translations of the same novel? Or if the first translation into language A is from the U.K. and the second from the U.S. version – is the latter a retranslation?[5] What if one part of a translation is based on the U.K. and another on the U.S. version (see Wu and Férnandez Díaz 2017)? Or what if the author revises a published text (see Baer 2017; Toury 1995/2012; van Hulle 2015) or self-translates[6] their works – which version should be used as a ST, and what if different translators use different versions (see Woods 2006)?

Because novels sometimes exist simultaneously in several forms, establishing the exact ST(s) may require extensive detective work (see, for instance, Fernández Muñiz 2016). However, as “the published text is but one phase in the text’s evolution” (Cordingley and Montini 2015: 2), an evolution that begins before the text’s publication, we need to ask: does a text become a text only once it is published, or already in manuscript form – can an unpublished text serve as a ST? And what should we do about the fact that “behind every manuscript there lurks other, lost manuscripts and any number of unrecorded hours of mental activity” (Scott 2006: 108) – can a ST exist in oral or other non-physical form (see Apter 2005; Hung 2005)?

When the possibility of a non-physical ST is taken into account, establishing what served as the ST(s) of a translation might become outright impossible. For example, the ST may have existed only in the mind of the translator, without ever taking any tangible form, when they were producing a pseudotranslation (see Apter 2005) or a compilative translation (see Shengyu 2018; Ivaska 2021).[7] Locating the ST may also be challenging if the ST is “one of a kind, […] utterly unique” (Emmerich 2017: 6), such as Karen Emmerich’s ST for her English translation of Vassilis Vassilikos’ novel Γλαύκος Θρασάκης (1974), translated as The Few Things I Know About Glafkos Thrassakis (2002). Since the publisher of the translation demanded that the book be shortened, the author himself tore out 150 pages of a Greek version, leaving the translator with a “physically altered copy of an out-of-print edition of a novel that had already appeared in numerous other versions in Greek” (Emmerich 2017: 6) as her ST. In fact, sometimes the translator may be the only one with access to and/or knowledge on what the ST(s) of a particular translation were, and the researcher may, unfortunately, have “no way, let alone a foolproof one, to distinguish between texts whose sources have simply vanished and texts which never had a single-text source” (Toury 1995/2012: 48).

In other words, identifying the exact ST(s) may be difficult because there can be several candidates, because the translator may have used a unique copy, or because the text never had a physical form. However, picking up a version and comparing it with the translation might lead to incorrect conclusions. Dedner, for example, has observed one such instance:

Earlier scholars compared Büchner’s translation with Hugo’s original [of the plays Lucrèce Borgia and Marie Tudor] and were amazed at the liberties the translator Büchner had allowed himself. Their amazement was ill-founded since it was they who had taken the liberty of comparing Büchner’s translation from 1835 with a much later and much changed version of the French text.

Dedner 2012: 125-126

To avoid this kind of missteps, the ST(s) should be carefully established before making comparative analyses. This seems especially important when dealing with phenomena like retranslation, pseudotranslation, indirect translation, revision, back-translation, adaptation and self-translation, as these categories are defined and distinguished from each other at least partly on the basis of their relationship with their ST(s) (see Gambier 1994, among others).

2.3. The work, the (source) text, the translation

If, for research purposes, it is necessary to identify the ST(s) of a translation, how to confront the fact that a text may exist in multiple forms, and how to conceptualize the connections between the various translations of a novel that were done from different STs? In other words, how to describe the relationship between different texts that stem from the same root? Or what to do if one is unable to locate the exact ST(s) of a translation – does it mean that research becomes impossible? One approach to dealing with such situations can be found in textual scholarship,[8] in which a literary product can be described with two distinct terms, text and work: “From the receiver’s perspective, a work is the imagined whole implied by all differing forms of a text that we conceive as representing a single literary creation” (Shillingsburg 1984/1996: 43; emphasis added). In other words, a text is one form of a work – or, the other way round, texts are different representations of a work.

The evolution of a work may begin as a non-physical text (an idea) and may then transform into various physical texts – in one or several languages. Then, the “process of textual transformation continues […] after the work’s publication through its re-editions, its retranslations and its different reception by heterogeneous communities of readers” (Cordingley and Montini 2015: 2; emphasis in the original). In brief, works, whether translated or not, often exist in several forms – texts – none of which is necessarily the last or the definitive version. Similarly, trying to locate the first or the original version is difficult. Even authors themselves can create multiple texts of their works. For example, as van Hulle (2015) explains, Beckett self-translated his L’innommable (1953) from French into English (The Unnameable, 1958) adding a phrase to the ending – thus making “the original less complete” (van Hulle 2015: 46) – and later, when the French version was republished, this additional phrase was also included in that language-version. In other words, the relationship between a ST and a target text is not always as straightforward as it may seem: translations can cause their STs to evolve (see also Emmerich 2017) or they can even function as STs for further (indirect) translations, implying that the relationship between STs and translations is not unidirectional and binary (see Dollerup 2004; Delabastita 2008). However, both the French and the English versions of Beckett’s novel are texts of the same work. Similarly, in the case of the Virginia Woolf novel whose U.K. typesetting and printing differ from those of the U.S. version, the story Woolf imagined is the work, whereas both the U.K. and the U.S. print versions are its texts. Regardless of which of these versions is used as a ST, all subsequent translations are representations of the same work.[9]

3. The five Finnish translations of Vingt mille lieues sous les mers

Jules Verne’s Vingt mille lieues sous les mers was originally published between 20 March 1869 and 20 June 1870 in Magasin d’Éducation et de Récréation, a periodical aimed at young readers. It has since been published in several versions (texts), including new editions and translations.[10] For this case study, we have identified five book-length translations into Finnish (see Table 1). We will study these translations to see if they have the same source text. The aim is to understand whether they really are retranslations of the same source text, keeping in mind the definition of retranslation as a “second or later translation of a single source text into the same target language” (Koskinen and Paloposki 2010: 294; emphasis added).

The first Finnish translation, Otto Aleksanteri Joutsen’s Sukelluslaivalla maapallon ympäri [Around the globe in a submarine], came out in 1916, forty-seven years after the story was first published in French. The title page does not mention Joutsen’s source language/text. The translation, published by WSOY, is abridged, at 400 pages and approximately 79,000 words. It was reprinted in 1918 and 1934.

In 1926, Verne’s novel was published in a (re)translation by another publisher, Karisto. This version was initially published in two parts: Merten alitse [Crossing under the seas], translated by Väinö Hämeen-Anttila, and Kapteeni Nemo [Captain Nemo], translated by Urho Kivimäki. The title page explicitly states that the translations were done from French. There were no reprints in this two-volume format, but in 1968 the translations were published in a second revised edition. The revisions were done by an anonymous reviser – both Hämeen-Anttila and Kivimäki had passed away – and the novel was turned into a trilogy, the parts being titled Merten alitse, Kapteeni Nemo, and Nautilus. The trilogy was reprinted in 1969 and 1977, whereas the fifth edition in 1991 was published in one volume under the name Kapteeni Nemo ja Nautilus [Captain Nemo and the Nautilus]. The sixth and latest reprint appeared in 1998, again in one volume. In any format, the translation by this duo, with approximately 530 pages and about 105,000 words, is the longest of the Finnish translations.

A Finnish translation of the novel also appeared in the USSR in 1934-1935, published by Valtion Kustannusliike Kirja. The publisher was based in Petrozavodsk, the current capital of the Republic of Karelia in the modern-day Russian Federation. At the time of the translation’s publication, Finnish was one of Soviet Karelia’s official languages, which explains why a Finnish translation was published there. The translation consisted of two parts, Sukelluslaivalla maapallon ympäri I [Around the globe in a submarine I] (1934) and Sukelluslaivalla maapallon ympäri II [Around the globe in a submarine II] (1935). No translator is mentioned on the title page, but the names of the editors, H. Mäkelin (Part I) and Aune Rautio (Part II), are given. Length-wise, this version, having 406 pages and about 79,000 words, seems slightly abridged.

The publishing house WSOY, which had published Joutsen’s translation in 1916, released the novel in a new translation in 1955. This translation, by Martta Tynni, carries the same title as Joutsen’s translation, Sukelluslaivalla maapallon ympäri [Around the Globe in a Submarine]. Tynni’s translation is 317 pages and about 67,000 words long (whereas Joutsen’s is 400 pages and around 79,000 words) and, according to the title page, it has been slightly abridged. Tynni’s translation has been reprinted six times (1957, 1964, 1970, 1976, 1995, and 2000).

Lastly, Vingt mille lieues sous les mers was translated into Finnish by Kristina Haataja and published by Minerva in 2008. Haataja’s translation, titled Kapteeni Nemo: Merten syvyyksissä [Captain Nemo: in the depths of the seas], is the shortest of all with only 287 pages and roughly 57,000 words, and it is also the only one with no illustrations. According to the title page, the book was translated from French. At the time of writing, the translation has been reprinted once, in 2011.

Table 1

Bibliographic information of the Finnish translations of the novel[11]

Bibliographic information of the Finnish translations of the novel11

Table 1 (continuation)

Bibliographic information of the Finnish translations of the novel11

-> See the list of tables

4. Methods

To identify the source text(s)/language(s) of each Finnish translation, we ask: Is the ST in French? If not in French, then what language(s)? We first looked for clues on the source texts/languages in materials, such as correspondence, found in the translators’ archives. Joutsen discusses the translating of Vingt mille lieues sous les mers in letters to the publisher, archived at the National Archives of Finland (NAFI). Hämeen-Anttila’s archives at NAFI and the Finnish Literature Society Archives, like Kivimäki’s archives at NAFI, do not contain any mentions of translating Verne. As for Tynni, we have been unable to locate her archives, and the same applies to Haataja (who is still alive). We looked for additional information on the translators’ lives in various sources to find further clues as to their possible source texts/languages and to contextualize their translations.

Textual comparison of the translations with their possible STs is another way to identify the source text(s)/language(s), and it forms the core of our analysis. To enable a rich analysis, we analyze four chapters from different parts of the novel that deal with different themes: one is the first chapter (“Un écueil fuyant” in the original), two are from the middle (“Une forêt sous-marine” and “La foudre du capitaine Nemo”) and one from the end (“Les poulpes”). The comparison builds on Huuhtanen (2016). When the analysis requires comparison with a French version, we refer to a version that was easily accessible: the 40th, illustrated edition published in 2010 by Librairie Générale Française.

5. Uncovering the source text(s)/language(s) of the five Finnish translations

Otto Joutsen’s correspondence with his publisher reveals that he translated the novel indirectly from English and Swedish. It was common to translate French literature indirectly in that era (Suomela-Härmä 2007), and according to online database Fennica,[12] the National Bibliography of Finland, at least two other novels by Verne from the late 1800s are indirect translations.[13] In a letter, Joutsen explains how he had – without success – tried to get a copy of Vingt mille lieues sous les mers in French because he was not happy with using the English version he had at his disposal:[14] he feared that converting measurements back from the English system into the metric system would not result in the same numbers that Verne had used, especially since the numbers were often preceded by the word “about” in the English translation.[15] Therefore, the publisher sent Joutsen a Swedish translation, but because Joutsen suspected that it had also been translated from English, he concluded that it was best to use both translations side by side.[16] Thus, Joutsen’s translation is not only indirect, but also compilative. However, Joutsen refers to the English as “the original”[17], which suggests that he might have used it as his primary ST – or perhaps the English version was “original” for him because it was chronologically his first ST.

Väinö Hämeen-Anttila worked at the Karisto publishing house as an editor and CEO (Simojoki 1950), wrote and edited books, and translated English and French authors such as Defoe, Kipling, and Dumas. Urho Kivimäki also worked at the Karisto publishing house (Simojoki 1950) and translated literature, mostly from German and French. Comparison of Kivimäki’s and Hämeen-Anttila’s translations with a French version of the novel has produced no evidence suggesting that they did not translate Vingt mille lieues sous les mers from French. Of course, it is difficult to ascertain that they did not use any other language-version(s) as their ST(s). In any case, the most interesting aspect of their translation, from the point of view of this study, is that it was successively recycled, as will be discussed next.

The anonymous 1934-1935 USSR translation appropriates Joutsen’s and Kivimäki’s translations. Comparison makes evident that the first part of Sukelluslaivalla maapallon ympäri is an almost exact copy of the first 192 pages of Joutsen’s translation, except that it has been shortened from about 38,000 to approximately 30,000 words and a few words have been changed, apparently to modernize the style (see Examples 1a, Joutsen’s translation, and 1b, the USSR translation; texts in square brackets are our glosses of the examples that are being analyzed). The second part of the translation is not a copy of Joutsen’s translation, but a slightly revised and shortened (from approximately 57,000 to around 49,000 words) version of Kivimäki’s Kapteeni Nemo with the wording being overall very similar (see Examples 2a, Kivimäki’ version, and 2b, the anonymous Soviet translation). In other words, this version is some kind of intralingual compilative plagiarism or revision.

According to Fennica, Martta Tynni’s translation career was short (1955-1961) with three works by Verne, a novel by another French author, Paul Berna, and six children’s books by the French writer Madeleine Grize, which Tynni translated together with L. Aro. Tynni was also a lecturer of French at the University of Helsinki from 1971 to 1973;[18] she must have had an excellent command of French at least by the 1970s. Considering that Tynni’s translation has the same publisher and title as Joutsen’s, one might expect for it to be a revision, but it is not. Joutsen used English and Swedish STs, but the textual analysis of Tynni’s translation of Vingt mille lieues sous les mers suggests that she used a French ST, as Examples 3a (Joutsen) and 3b (Tynni) demonstrate. It also shows that Joutsen was rightly worried about converting measurements from the English system back into the metric system:

The title page of Tynni’s translation states that the translation is abridged. Milton (2001) observes that there are several ways to abridge a translation, the easiest being to translate an existing abridged version. Tynni’s ST may have been an abridged French version or a full-length version. In the latter case, the ST could be the same as Hämeen-Anttila and Kivimäki’s, but considering how heavily Tynni’s translation is abridged (it has about 67,000 words, as compared to the approximately Hämeen-Anttila and Kivimäki’s 105,000), it becomes difficult – if not impossible – to compare her version to Hämeen-Anttila and Kivimäki’s to verify similarities and differences, let alone to compare Tynni’s translation to French versions in order to figure out the exact ST. In any case, we can say with certainty that her ST is different at least from Joutsen’s and the USSR version.

Kristina Haataja is a translator, language teacher, and journalist who has spent most of her life in Paris. Prior to translating Vingt mille lieues sous les mers in 2008 – as well as Verne’s Le tour du monde en 80 jours (1872) under the title Maailman ympäri 80 päivässä (2008) and Cinq Semaines en ballon (1863), titled Kuumailmapallolla Afrikan halki (2009) in Finnish – she had already translated other French authors, such as Marguerite Duras, Virginie Despentes, and Emmanuel Carrère. She has also written fiction both in French and in Finnish, and hence we have no reason to suspect that she could not have translated Verne directly from French. However, comparison with the other Finnish translations reveals that Haataja’s translation draws on Hämeen-Anttila and Kivimäki’s translation. Even if Haataja’s abridged translation is much shorter than Hämeen-Anttila and Kivimäki’s (57,000 as compared to 105,000 words), her translation is divided into paragraphs following Hämeen-Anttila and Kivimäki’s example, the syntax is very similar, and some word choices are duplicated so systematically that the source of influence is indisputable, such as in the case of the not-so-common and poetic word lakkapää (see Example 4):

However, some passages suggest that Haataja used also a French ST:

It remains unclear why, instead of translating simply from a French ST, Haataja ended up making what Alvstad and Assis Rosa (2015: 17) would call a compilative inter- and intralingual retranslation, which they define as the “use of the source text and of one or several previous translations into the target language,” or what Washbourne (2013: 619) calls a comparison-revision/retranslation, in which “the most successful parts are retained, the weaker parts shored up.”[19] Perhaps Haataja was familiar with Hämeen-Anttila and Kivimäki’s translation as a reader, and their version thus represents for her the canonical translation from which she did not want to deviate too much;[20] or, it might have been simply the publisher’s decision or demand that Haataja should stay close to Hämeen-Anttila and Kivimäki’s translation. Milton (2001) suggests that it can be cheaper to recycle an existing translation rather than translate from scratch, but as Şahin, Duman, et al. (2015: 197) note, retranslations that recycle previous translators’ work have not been discussed in detail “partly due to the complexity of the problem and partly because of the vulnerability of the situation.” Şahin, Duman, et al. presumably refer to the fact that phenomena such as editing, revising, and recycling raise the question of “how much change can there be in the revision process for the translation still to be the same, i.e. under the name of the previous translator, and where is the line to be drawn to a new translation?” (Paloposki and Koskinen 2010: 44). Whatever the motivations behind the choice to use a French version and a previous Finnish translation as the STs, Haataja’s translation presents an interesting case of yet another type of a (combination of) ST(s) that can be found behind a translation. What makes it particularly interesting for this study is that following Shuttleworth and Cowie’s (1997) definition of source text, this translation would need to be categorized as an intralingual translation because one of the STs is in the same language as the translation itself.

To sum up the findings (also presented in Table 2), we can conclude that Joutsen’s translation is an indirect, compilative translation: his correspondence reveals that the STs are an English and a Swedish translation, and that he had no access to a French version. As for Hämeen-Anttila and Kivimäki’s translation, we have found no evidence suggesting that it is anything but a direct translation, whereas the anonymous USSR translation is some sort of an intralingual compilative plagiarism/revision that appropriates Joutsen’s and Kivimäki’s translations. Two interpretations are possible when it comes to the fourth translation, by Tynni, which is done directly from French like Hämeen-Anttila and Kivimäki’s translation, but is abridged: either Tynni used an abridged French version as her ST, in which case her ST is different from Hämeen-Anttila and Kivimäki’s, or she had a full-length French version as her ST which she translated only partly in her Finnish translation, in which case the ST Tynni had at her disposal might be the same version of the novel as Hämeen-Anttila and Kivimäki’s. Finally, the fifth translation, by Haataja, is perhaps the most interesting as it shows use of both a French version and a previous Finnish translation as its sources; it evades clear-cut categorizations and could be described as a compilative inter- and intralingual retranslation, or as a comparison-revision/retranslation.

This careful assessment of each translation’s source texts has made new qualities of the source and target texts become evident, which translates into new research opportunities: we could compare Joutsen’s, the USSR and Haataja’s versions as they are all compilative in nature, or we could study Tynni’s and Haataja’s translations, which are both abridgements, or examine closer the USSR and Haataja’s versions, which both make explicit reference to earlier Finnish translations. Similarly, delving deeper into the forms of source texts and the categorization of translations can help notice new research avenues to further our understanding of what source text is. In this case study, some of the topics that could be further explored include: How to hypothesize compilative retranslations which draw from previous translations into the same language, that is, what hypotheses should one apply to Haataja’s translation, for example: the ones regarding retranslation, those formulated on indirect translation or something else, such as hypotheses on intralingual translation? How much can a translator rely on previous translations before their translation becomes compilative? Where is the line between retranslation and revision? What to do with the cases where one part of a retranslation/plagiarism is based on one and the other part on another text, like the anonymous USSR translation?

Table 2

The STs and translation types of the Finnish translations

The STs and translation types of the Finnish translations

-> See the list of tables

6. Conclusions

The concept of source text remains undertheorized. As this study shows, it is not sufficient to conceptualize source text as equal to an “original” – in fact, “the concept of the original […] is a modern invention, belonging to a materialist life, and carries with it all kinds of commercial implications about translation, originality and textual ownership” (Bassnett 1998: 38; see also Hung and Wakabayashi 2005; Baer 2017). Adopting the notions of work and text helps in establishing the relationships between the different versions – editions, (re)translations; in the first language or in translation – of a novel, for example: a work is represented by the different texts, that is, its different versions. If source text is considered equal to “original,” the existence of different texts is neglected. Such a discourse seems to liken source text with work, although the work, as Shillingburg’s (1996: 43) definition states, exists only through the texts that represent it.

Although identifying the source texts of translations is not always easy, it can be crucial in order to reach meaningful conclusions when dealing with topics where the source text-target text relationship is central, such as retranslation. The analysis of the five Finnish (re)translations of Jules Verne’s Vingt mille lieues sous les mers makes evident that the five translations have different source texts (see Table 2). This raises the question whether it is meaningful to compare them to learn something about retranslation, at least if we adopt Koskinen and Paloposki’s (2010: 294) definition of retranslation as a “second or later translation of a single source text into the same target language.” Another way to look at the situation is to acknowledge that the five Finnish translations are all texts of the same work – Jules Verne’s Vingt mille lieues sous les mers. This means that these translations could be compared among them to see what kind of elements the different texts of a work share or how they differ from each other. In this case, the comparison could also include versions in languages other than Finnish, as they are also texts of the same work by Verne. In other words, the points of view from which the five translations can be studied depends on whether source text is defined as a text or as a work.