No translation is an exact copy of its original, no matter how faithfully the translator attempts the translation (Mikhailov & Villikka 2001:378). Mikhailov and Villikka further assert that even if two translators translate the same text ‘as faithfully to the original as possible’, the result would be ‘two clearly different translations’. This is owing to the fact that each person employs his or her own idiolect (Landers 2001:90). It has been widely argued in translation studies that literary translators should adhere to the style of the original author (Chesterman 1997; Landers 2001; Nida & Taber 1974). Landers (2001:49–52) argues against this notion because in his view, products of such a practice are awkward and incomprehensible. He is of the opinion that translators should decide what kind of text they want to create and whether they want to be visible or not. To him, producing a text that ‘leaves as little evidence as possible’ of the translation process is what translators strive for (Landers, 2001:49). He views the translator as the person in control of the process, the decision maker. Jones (2011:143–144) draws attention to the fact that there is tension among translation scholars regarding this issue. Some advocate for replicating the style of the author, while others argue that translators are writers in their own right and inevitably leave their mark. He further asserts that literary translation is a manipulative exercise because it involves issues of identity and ideology, which influence the translator’s decision-making and style of writing (Jones 2011:156–157).

It has also been argued that autobiographies are not only sensitive or intimate texts (Brierley 2000) but also historical artefacts that should be translated as faithfully to the original as possible (Aurell 2006; Honey 2006). However, this is very difficult to achieve since translators are individuals with unique thinking patterns (Landers 2001:90). When considering the fact that translation is a cognitive process which involves decoding, encoding and decision-making (Albir and Alves 2009; Hewson & Martin 1991; Lewandowska-Tomaszczyk 2010), the final product cannot be the same as the original. Every translator aims to achieve fluency; as a result, they make decisions that will ensure that they achieve this goal (Landers 2001). These decisions manifest themselves through the mode of expression they select, which sometimes reveals itself through certain patterns. It is against this background that the researchers decided to explore the style of the two translators who translated Mandela’s autobiography. They wanted to discover how the translators dealt with the book as a literary text and autobiography. The question this study therefore seeks to answer is: What kind of decisions did the latter make and what kind of recurring patterns that distinguish the two translators can be identified from their translations?

The aim of this paper is to present a comparative study of these translators’ styles. Style in translation is revealed by the choices translators make during the translation process (Baker 2000). The focus of the study is therefore placed on the identification of recurring features of translator style exhibited by choices made by Peter Mtuze and Bheki Ntuli in their renderings of Mandela’s Long Walk to Freedom into isiXhosa and isiZulu as Indlela ende eya enkululekweni and Uhambo olude oluya enkululekweni, respectively. Long Walk to Freedom was translated into four South African languages, that is, Afrikaans, isiXhosa, isiZulu and Sesotho sa Leboa. IsiXhosa, isiZulu and Sesotho sa Leboa are three of the previously marginalised languages of South Africa, which together with eight others such as the formerly dominant Afrikaans and English now constitute the 11 official languages of the democratic South Africa.

The authors first present a brief biographical sketch scrutinising the translators. The second section presents the definition of style followed by a review of existing literature on translator style and the theoretical framework underpinning the study. A concise description of the methodology followed in the analysis is outlined; thereafter, the interpretation of the findings is discussed. The conclusion is then presented by way of a summary.

Nida and Taber (1974:207) define style as ‘the patterning of choices made by a particular author within the resources and limitations of language and literary genre in which he is working’. They also maintain that style ‘gives a text its uniqueness’. Landers (2001:90) explains it as ‘a characteristic mode of expression’ displayed by the translator, whether consciously or unconsciously. The emphasis is on patterning of choices rather than one-off occurrences (Baker 2000:245; Munday 2008). Bakhtin (1981:276 in Munday 2008) makes a very important point when he points out that ‘any utterance – oral or written … can reflect the individuality of the speaker (or writer); that is, it possesses individual style’. These definitions also highlight the visibility of the translator that has been advocated by Venuti (1986). Venuti argues that translators should make themselves visible by applying strategies that would make their intervention in the text recognisable. By so doing, they display their individual styles.

The definition that has been adopted in this paper is that of Saldanha (2011) which she adapted from Short’s (1996 in Saldanha 2011:28) definition of authorial writing and thereafter applied to translation. Saldanha argues that identifying translator style is more complex than identifying authorial style because there is always a link with or reference to the original author or text. Saldanha (2011:31) proposes the following definition:

A ‘way of translating’ which:

The last point of her definition sets it apart from other definitions like that of Boase-Beier (2006), for example. Boase-Beier (2006:5) proposes four viewpoints from which style in translation can be considered:

Boase-Beier’s approach places emphasis on the text instead of the translator while Saldanha’s definition emphasises the fact that the style of the author is not used as a yardstick but as a reference to confirm the translator’s shifts. Its use is optional.

This paper represents an attempt to reveal stylistic patterns in translated texts. By focusing on and comparing the isiXhosa and isiZulu translations of Long Walk to Freedom, the study adds an African perspective to the existing literature on corpus-based studies of translators’ styles.

Before analysing the styles of the two translators, it is necessary to briefly mention the theoretical framework underpinning this study.

This study is guided by the Descriptive Translation Studies (DTS) and functionalist theories. These theories were propounded at more or less the same time as a result of dissatisfaction with the prescriptive equivalence-based theories of translation. Both these theories advocate for target-oriented translation. The aim of DTS theorists was to describe translation phenomena and to encourage the inclusion of the target culture and socio-political characteristics of the target audience (Toury 1980, 1995). Proponents of this approach examine a corpus of authentic texts in an attempt to understand the nature of the texts and the constraints operating on them. In this article, the researchers aim to determine the styles of the translators by describing the techniques they adopted, in line with Toury’s (1982:24) assertion that ‘descriptive studies are actually the best means of testing, refuting, and especially amending and modifying the underlying theory’.

Being a form of communication, translations are regarded as having a ‘skopos’ (Vermeer 2000), a term coined by Hans Vermeer referring to the aim or purpose of a translation. The functionalists view translation as a human action with a purpose. It is therefore this function of the text in the target system that determines the translation process, that is, which translation strategies can be adopted (Nord 1997). The translator is regarded as ‘the’ expert in the translational action to carry out these functions (Vermeer 2000:228). There are four text functions that were identified by functionalist theorists, that is, the referential function, which has the informative function as a sub-function, the expressive, the phatic and the appellative functions. As experts, the initiators of the translational act entrust translators with the responsibility of carrying out the translation functions.

The biographical background of the translators that is provided in the Prof. PT Mtuze and Prof. DBZ Ntuli sections above confirms that the translators under investigation are experts. According to Mtuze (2003:141), the translation of Long Walk to Freedom was initiated by Vivlia Publishers towards the end of 1999. Although the purpose of the translation is not clearly stipulated in Mtuze’s report, it is assumed that it was meant to inform all South Africans about the struggles against the apartheid government as reflected in Mandela’s life story. In his article, Mtuze (2003:141) mentions that all four translators, that is, Krog (Afrikaans), Mtuze (isiXhosa), Ntuli (isiZulu) and Serudu (Sesotho sa Leboa), held a meeting with Mandela to discuss certain cultural aspects they would come across in the translation process. This shows that they were concerned about their responsibility for translating the text as well as wanting to avoid misinterpretation of anything. They did not want to transgress the loyalty principle (Nord 1997) and misrepresent the author. They as ‘the experts’ had to decide how they would translate and fulfil the informative function of the text.

The following section outlines the studies that have been conducted on translator style in order to place this study in context.

The question of translator style was introduced by Baker (2000). After noticing the general preoccupation with the style of a literary writer and acknowledging that translators are writers in their own right, she became interested in finding a way to reveal the distinctive style of a literary translator. She was aware that translators are individuals with unique personalities which may manifest themselves in their translations. She argues that it is impossible to produce a stretch of language without leaving one’s ‘fingerprints on it’ (Baker 2000:244). This concurs with Herman’s notion (1996) that there is a translator’s voice in every translation (in Boase-Beier 2006). Baker (2000:245) defined style as ‘a kind of thumb-print that is expressed in a range of linguistic – as well as non-linguistic – features’. She further explains that style may encompass, among other issues, consistent use of specific translation strategies, glossing and manner of expression. The aim of her paper was to propose a methodology for identifying consistent stylistic characteristics that could be attributed to the translator. She proposed a corpus-based investigation where the researcher would study and compare certain patterns displayed by translators and then establish whether the patterns identified could be found in other works translated by the same translator. This was achieved by exploring the use of the English reporting verb SAY and its variations by Peter Clark and Peter Bush in their translations. In her findings, she noted that Clark used SAY more frequently, especially its past tense form, than Bush. This, she argues, could be attributed to the influence of the source language which is Arabic. She used the British National Corpus and Translational English Corpus as her control corpora. Since Baker’s (2000) publication, numerous studies investigating elements of individual translators’ styles have been published.

Mikhailov and Villikka (2001) conducted research to determine whether translators do have ‘stylistic fingerprints’. They compared fiction texts originally written in Russian with their Finnish translations. They collected a corpus of texts originally written by the same author and other authors in Russian and compared these with Finnish translations of different texts by the same translator as well as translations of the same text by different translators. They used richness of vocabulary, frequent words, word, sentence and paragraph counts and favourite words as their basis of comparison. With regard to vocabulary and frequent words, they discovered that these were not reliable indicators of individual style as the translators tended to stick to the source text. However, they found that ‘the use of modal words, particles, conjunctions, grammar forms, etc., as well as splitting or joining sentences and paragraphs and expanding or shortening the text’ can be taken as indicators of personal features (2001:383).

In 2004 and 2007, Winters conducted two separate studies to determine features of translator style in the translation of Fitzgerald’s The Beautiful and Damned as Die Schönen und Verdammten. The novel was translated into German by two translators, Renate Orth-Guttmann and Hans-Christian Oeser in 1998. In her first study (Winters 2004), she explored the use of loan words and code switches. In her second study (Winters 2007), she investigated the use of speech–act report verbs as features of translator style. Both studies were corpus based. In her investigation of the use of loan words and code switches, Winters used WordSmith Tools and Multiconc concordancers to interrogate her data. WordSmith Tools was used to obtain key words, frequency lists and concordance lists while Multiconc was useful for producing parallel concordances of the translations. In her findings, Winters revealed that Orth-Guttmann preferred to germanise loan words whilst Oeser tended to code switch. In other words, Orth-Guttmann tends to domesticate whilst Oeser is inclined to keep the foreign words.

As in her analysis of loan words and code switching, Winters used WordSmith Tools and Multiconc to facilitate the scrutiny of speech–act report verbs. She discovered that Oeser preferred lexical repetition whilst Orth-Guttmann avoided it. Orth-Guttmann, she noticed, exercised her creativity by employing a greater variety of verbs whilst Oeser was inclined to repeat verbs that appear in the source text. Both studies undertaken by Winters confirm that no two translators can produce identical translations because each has a different personality which influence their translation style.

Translator style manifests itself in various ways. A pilot study conducted by Kamenická in 2007 revealed that an ‘explicitation profile’ can be used to distinguish translators and thus contribute to identifying translators’ individual styles. Kamenická utilised a parallel corpus of translations based on Halliday’s language metafunctions: experiential, logical, interpersonal and textual. The focus of comparison was on measuring the frequency, distribution and variability of explicitation phenomena in the translations of two novels: David Lodge’s Small World by A. Přidal and John Cheever’s Falconer by R. Nenadál. Her definition of an explicitation profile includes implicitation since this is usually studied alongside explicitation (Kamenická 2007:118).

Adding to the existing literature, Munday (2008) explored Harriet de Onis’s translation, amongst others, focusing on the use of condensed pre-modifiers, a variety of verbs or process forms and prosodic elements. He also studied ideological intervention when examining lexical choices, demonstrating that stylistic choices and ideological orientation do contribute to a translator’s individual style.

In 2010, Pekkanen conducted a very thought-provoking research in which she confirmed the position adopted by earlier studies that argued it is possible to talk about translators’ style. She studied translations undertaken by Saarikoski, Matson, Mäkinen and Linturi of different novels from English into Finnish. Her findings demonstrated how translation shifts reveal the individual style of a translator. She also proposed a framework for identifying translator style where the researcher investigates the micro-level data and the effects thereof at the macro-level. Pekkanen drew attention to the fact that translator style cannot be interrogated by computer software alone, owing to the different structures of languages. The software may not be able to recognise inflected lexical items found in certain languages.

The studies cited above underscore the fact that translators do tend to leave a personal imprint on the texts they manipulate and that to a large extent it is possible to identify such characteristics. Observable in these studies is the fact that they are corpus based. Most of them used two types of corpora: the core corpus which consisted of translations under investigation and a control corpus, which served as a reference to confirm the personal imprint. The present study differs from these studies in that it does not employ a control corpus because the focus of style is limited to the translation of Nelson Mandela’s autobiography by the two translators. The researchers believe that if ‘style is inextricably intertwined with one’s idiolect’ (Landers 2001:90), this research is legitimate and a control corpus is not necessary; hence, the first point in Saldanha’s definition of style (par 3) has not been considered. The study will provide fresh evidence of translator style to the existing corpus-based studies.

The section that follows presents the method of analysis that was adopted.

For the present study, the authors compiled a corpus, which consisted of the aforementioned isiXhosa and isiZulu translations of Nelson Mandela’s Long Walk to Freedom, entitled Indlela ende eya enkululekweni and Uhambo olude oluya enkululekweni. The original text was used for reference.

The authors proposed a data-driven approach to the study, aiming to uncover patterns that distinguish between the two translations. Barlow’s (2001) ParaConc Concordancer which produces parallel concordances was used for ease of analysis. The three texts, that is, the English original and the isiXhosa and isiZulu translations, were aligned sentence by sentence. For the purposes of this study, just part eight and nine were uploaded on ParaConc for analysis. Although all three texts could be uploaded on ParaConc software, the researchers were able to view only two texts at a time, the original and one translation. The two translations were also aligned separately. The graphic in Figure 1 is a window with the English and isiXhosa texts. On the bottom left corner of the window, it is indicated that two parallel texts were uploaded. On the bottom right corner of the window is recorded the number of words for each text, starting with the source text, English text and then the target texts, isiXhosa and isiZulu. Figure 2 illustrates the alignment of the two translations. ParaConc allows the researcher to see instances of search words in the source text on one screen and the parallel sentences in the two translations, as in Figure 3. For the texts to be uploaded on the software programme, they have to be scanned, converted to text format and thoroughly cleaned to remove foreign characters that are created during the conversion process. This takes a great deal of time; as a result, only selected sections of the book were prepared for the software. A manual inspection of the layout of the translations was also carried out and this yielded useful data for analysis too. Both methods of analysis have advantages in that concordancers allow quicker and easier analysis, while a manual analysis is not preoccupied with the results. The isiXhosa and isiZulu texts were given Afrikaans and Swedish labels since ParaConc does not have these languages in its built-in list.

Figure 1 and Figure 2 did not just depict the alignment of the uploaded texts but furnished word count information as well. Looking at the bottom right corner of Figure 1, one notices the word count for each text. The word count for parts eight and nine is 51 149 for the English text; 38 077 for isiXhosa and 37 180 words for isiZulu. Figure 2 portrays the word count of part eight only. From these figures, one notices the difference in word count between the two translations: isiXhosa uses 20 522 and isiZulu 19 775 words. Figure 3 displays the results of the word search for ‘High Organ’. The concordance lines show that this word was rendered as ‘High Organ’ (iSigqeba esiPhezulu) in isiXhosa, a loan word plus the isiXhosa equivalent in brackets; and as ‘High Organ’ in isiZulu. What cannot be seen from the window are the italicised words. These can only be observed in the hard copy.

When the researchers were looking at the hard copies of the translations, they could note immediately that the book covers have the same design, but there is a slight difference in the title and contents. Mtuze rendered the title Long Walk to Freedom as Indlela ende eya enkululekweni (Long road to freedom) while Ntuli used Uhambo olude oluya enkululekweni (Long walk to freedom), which means the same as the source text. Looking at the page numbers, it was noticed that there is a difference of approximately 20 pages in length between the two translations. This corresponds with the difference in word count that has been noted above. Another difference that was noted is that the isiZulu translation has a translator’s preface, while the isiXhosa one does not. Instead, it contains a brief note on the list of technical terms that appear in the book. This is found at the end of the book. In his preface, Ntuli reports on the challenges he encountered whilst he was translating the autobiography. One of those he mentions, for instance, is that of indigenising loan words but at the same time not adhering strictly to isiZulu orthographic rules. Instead of writing ihelikhophutha, he chose to write ihelikhoptha because the former does not sound right: ‘kuzwakala kungemnandi’ (Mandela 2001b). Mtuze however reported on the challenges he faced when he was translating the text in the form of a journal article (Mtuze 2003). These differences prompted a more detailed analysis of the texts. The following characteristics were noted and examined: use of loan words for place names and months, loan words in italics plus target equivalents, loan words with paraphrase, use of indigenised words with paraphrase and expansion and deletion.

This section presents the interpretation of findings.

As may be observed from the discussion above, translator style involves decision-making during the translation process. The translator displays her or his personal touch through the words he selects from his or her lexicon or ‘idiolect’. The most important aspect in our adapted definition of style is the recurrence or patterning of certain features. It has also been shown that style can be investigated from different angles, such as use of grammatical aspects, explicitation or, as in the case of this study, translation strategies. The literature furthermore revealed that style can be identified by comparing translated texts with other translations or with non-translated texts. In this study, we compared isiXhosa and isiZulu translations of Mandela’s autobiography Long Walk to Freedom. The investigation of the use of loan words in general suggests that Mtuze is closer to the source text than Ntuli. The frequent use of target equivalents alongside the English expressions not only facilitates understanding but also teaches translation students about translation strategies. Although the use of italics cannot be said to be a unique feature attributable to Mtuze, when compared with Ntuli’s work, it cannot be ignored. In the same vein, by not italicising foreign words in the isiZulu text, Ntuli made evident his thought processes, which differentiates him from Mtuze. By avoiding foreign words as much as possible and adopting the paraphrasing technique, Ntuli was moving the text closer to the reader.

As proposed by Mikhailov and Villikka (2001), expansion and contraction are other features which clearly distinguish the two translators. Mtuze displayed tendencies towards expansion while Ntuli was inclined to contraction. Although contraction and deletion could be disadvantageous to the target reader, in that crucial information could be lost, such a strategy simplifies the text and eases the tension whilst reading; sometimes too-long sentences hamper the reading process. Although sentence construction was not the focus of this study, sentence by sentence alignment suggests that Ntuli prefers to break up longer sentences. This finding is not conclusive but is an area that could be explored further. The analysis presented here illustrates that the function of the translation and the target reader were always at the back of the translators’ minds. It also suggests that Mtuze and Ntuli have different translating styles and have subsequently left indelible fingerprints on their respective translations of Long Walk to Freedom. Whether consciously or unconsciously, the two translators exemplified Venuti’s (1986) advocacy for visibility. This is evident in their translation of this autobiography. Doing so did not deter them from displaying their personal imprints as creative writers. As Pekkanen (2010) has indicated that translator style can be investigated in various ways, more features could be examined in future to reveal the style of the two translators or of others. It is hoped that this article will pave the way for further research into comparative translation studies as regards the indigenous languages of South Africa.