1Postcolonial approaches to translation studies (Bandia 2021; Tymoczko 1999) have often stressed that the postcolonial text, which “involves multiple linguistic and cultural systems” (Bandia 2021, 129), is already a translation itself and that “there are […] parallels between a translator’s choice in transmitting a source text and the choices made by a minority-culture writer in representing his/her indigenous culture in an alien dominant language” (137).

2The works of Abdulrazak Gurnah, particularly those devoted to the history of East Africa – such as Paradise (1994), Desertion (2005) or Afterlives (2020) – fall under this category as Gurnah mediates his texts through English in order to convey significant aspects of East African culture in its context to Western readers. While numerous articles have studied how translators deal with such linguistic and cultural elements when translating into imperial languages (Bandia 2021; Mayanja 2019; Fabusuyi and Olagunju 2019; Brahima 2012, to name but a few), it is less common to focus on an almost reverse trajectory. It will be the case here, since the Swahili translation of Paradise is a way for the original text to travel back to its source – Uswahilini, where the novel’s plot is set, and through its major language, Kiswahili.

Reasons and ways of coming back home

The English novel is written in the first place by English people for English people. That is what makes it the English novel. The Russian novel is written by Russians for Russians. But the African novel is not written by Africans for Africans. African novelists may write about Africa, about African experiences, but they seem to me to be glancing over their shoulder all the time they write, at the foreigners who will read them. Whether they like it or not, they have accepted the role of interpreter, interpreting Africa to their readers. Yet how can you explore a World in all its depth if at the same time you have to explain it to outsiders? (2003, 51).

6How can this be done? Two major conditions are required in the African context:
a) the book must be written or translated in a language accessible to readers and
b) it must be published locally, at an affordable price.

7It is worth noting that digital technology now improves the circulation of texts in East Africa, whether written in African languages or not. In Francophone Africa, for quite a long time – and today still – authors had their books published in France, especially Paris; this made it difficult, and most of all expensive, to access them in Africa. In 2006, Francophone author Véronique Tadjo expressed concern that her work L’Ombre d’Imana, which addresses the 1994 Rwandan genocide, was largely inaccessible to readers in Africa. This awareness, coupled with the support of the Alliance Internationale de l’Édition Indépendante, led to the creation of the “Terres solidaires” collection. Through this initiative, the Alliance acquires rights to books authored by African writers but initially published in France, with the objective of co-publishing them in Africa. This approach significantly improves accessibility by offering these works at prices adapted to local markets – approximately two to three euros – thus enabling, for instance, students to purchase them. L’Ombre d’Imana was the first book to be disseminated through this model, paving the way for numerous others. However, the linguistic medium remained French.

8The issue of the language gap, a prevalent one in Africa, has been discussed by some African writers, from Ngũgĩ wa Thiong’o to Boubacar Boris Diop, who both decided to stop using imperial languages like English or French in their fictions, aiming to develop their writing through their mother tongues – be they Gikuyu or Wolof. In doing so, they sought to engage their African readers – the audience they value most – more effectively and directly. In an article dealing with the issue of readership, based on the case-study of Ngũgĩ’s Petals of blood, Kenyan writer Abdilatif Abdalla highlights the importance of choosing an African language to reach one’s audience and even goes further, presenting the written form as an obstacle to the dissemination of a text:

• 4 “Lakini riwaya ya mwandishi huyo huyo, Petals of blood, haikuwatia shughuli wala wasiwasi kwa sabab (...)

But Ngũgĩ wa Thiong’o’s novel Petals of Blood didn’t worry them much, because they (those in government) knew that its message – even though it criticized the government – would reach only a few Kenyans, for two reasons: first, it’s a written text, and few Kenyans can read. Second, the book is written in English, which makes its audience even smaller. (2016, 95; my translation)4

9This reception contrasts sharply with that of Matigari wa Njinũũji (wa Thiong’o’ 1986), by the same author, which was written and published in Gikuyu and provoked such alarm among the authorities that they sought to suppress not only all the copies of the book but even its main character, Matigari, whom they had mistaken for a “flesh and blood” political agitator. While it is quite unlikely that Gurnah seeks the kind of reception envisioned by Marxist writers such as Ngũgĩ or Diop, reaching an East African audience nevertheless remains significant, given that several of his narratives are deeply rooted in East African history and culture and that no one ought to be denied the possibility of knowing more about their own history. As Ida Hadjivayanis notes in the preface to her Swahili translation of Paradise, this concern was one of her main motivations when she took up the challenge:

• 5 “Nilichagua kutafsiri riwaya ya Paradise kwa Kiswahili kwa sababu nilitaka kuipeleka kazi hii nyumb (...)

I decided to translate Paradise into Kiswahili because I wanted to bring this novel back home. I say this because Abdulrazak Gurnah portrays undeniable realities of life in East Africa before and during colonization, as well as after independence and the revolution. His writings resonate strongly with theories of memory studies and address themes of identity, exile, and the lives of Zanzibaris and, more globally, Swahili people. He has both the knowledge and the ability to unfold our world before our eyes with truth and realism, while giving us the weapons to better understand ourselves and the world we live in. (2022, iii; my translation)5

10The Swahili translation of Paradise has enhanced its circulation within East Africa through a local edition published by Mkuki na Nyota, thereby enabling readership in Kiswahili. However, to the best of my knowledge, the text has yet to be incorporated into the school curriculum despite the government purchasing one thousand copies and distributing them to colleges across the country (Hadjivayanis 2025b).

11What is at stake between original and translation when, so to speak, a text is repatriated into its cultural context? What does it mean to translate an original version that has already “encoded” Swahili culture in English? If the author already assumes the role of interpreter, what, then, constitutes “the Task of the Translator,” to invoke Walter Benjamin’s words (1923)?

In the house of the interpreter

12When I argue that the English text already “encodes” Swahili culture, I mean that since Paradise was first published in the UK for a mostly European audience, Gurnah – as described by J.M. Coetzee – had to assume the role of the “interpreter.” He already had to think about techniques that would allow the British reader to discover and understand Swahili culture in a specific historical context, that of the caravans. This interconnects with Venuti’s theory, which identifies two strategies in translation: that of “domestication” and that of “foreignization” (1995). Hadjivayanis refers to these strategies in her foreword using the Swahili verbs of “kuzowesha” (to domesticate) and “kugenisha” (to foreignize). In a way, Gurnah had already “foreignized” his original text. One must ask what the translator does with that.

13One of the most classical strategies when a writer wants a realia or expression to be understood without having to use a footnote is to explain its meaning directly in the text. Gurnah often uses this technique, as we can see in the few examples below:
- “but he had made a good friend who was a boatmaster of a dhow, a nahodha” (Gurnah 2021, 14; my emphasis);
- “The dhow they travelled in was called Jicho, the Eye” (2021, 15; my emphasis);
- “The mnyampara wa safari, the foreman of the journey, would be waiting for the expedition somewhere in the interior.” (2021, 34; my emphasis);

14Quite logically, these explanations disappear in the Swahili translation. As Hadjivayanis underlines in her foreword:

• 6 “Matukio huingizwa Uswahilini, yaani katika dunia ya Kiswahili. Hii inamwezesha msomaji kuelewa sim (...)

the events take place in the Swahili world, that is, in the world of the Swahili language. This allows the readers to understand the story without the need for explicit explanations, because what they read mirrors the very environment they live in and therefore understand.Words such as barza or kibarazani, for example, do not need to be defined.6 (2022, iv; my translation)

15Elements that require explanation, interpretation, and mediation in the English version are largely unnecessary in the Swahili translation, although, sometimes, explanatory passages from the English text are retained in Kiswahili.

16Occasionally, the exact term that had been replaced by an equivalent in English reappears (for example “adhuhuri” stands for “midday prayers”) as if the “undertext” of the original were finally resurfacing, like a photo negative being revealed through translation: a point that had until now remained invisible (or rather invisibilized, overlaid by the Western language) emerges with greater precision and clarity. Hadjivayanis, who states that she translates into Kiswahili for a readership of the coast (mwambao), explains that greeting formulas – highly specific in Kiswahili – assert themselves with particular force in contexts where English offers fewer nuances. She gives the example of “hello mother,” rendered as cheichei, a greeting specific to Zanzibar and used by children when addressing an elder (Hadjivayanis 2025b). This is the expression that imposes itself upon her, insofar as it is expected by a readership thoroughly familiar with the distinctive social practices of everyday life along the East African coast.

17This “moment of revelation” can be considered as the reemergence of the “Third text” of Paradise into the open. The concept of “Dritter Text” (“Third text” in English) was introduced into the field of African literary studies by Shaban Mayanja (1999). It refers to “the unwritten text, which serves as a primary source for the author and from which he or she draws inspiration” (1999; qtd. in Fabusuyi and Olagunju 1999, 55). Mayanja argues that when an African writer composes a text in a European language, its mental genesis originates from the language, culture, orality of their vernacular language. One of the major challenges in translation is to identify and convey this subtext into other languages. Here, because the translation is going back to the Swahili subtext (and also because Ida Hadjivayanis is a native speaker), we feel that the translation and the Third text seem to coincide.

18Here are two other examples of the slightly different – and more precise – contextualization, given by the use of the Swahili language. The first refers to the word “uncle,” which is translated as “ami” in Peponi. There are different ways of referring to an “uncle” in Kiswahili (Hadjivayanis 2026), depending on whether the man is from the mother’s side (mjomba) or whether he is younger or older than the father (baba mdogo/mkubwa). By using “ami,” a word of Arabic origin, the translation may focus on the relation between Aziz and Yusuf’s father but, more probably, on the fact that the action takes place in a setting where Arab people and Islam, the culture of Uswahilini is important, as “ami” is also the way people living in the Interior address a man from the Coast even if, as stated in the text this man “ain’t your uncle” (Gurnah 2021, 25).

Euphemism, indirect language and veiled communication

20Indirect communication is widely recognized as a characteristic of traditional African cultures. A very clear message is often perceived as superficial, suited only to those unable to interpret or decipher, such as children. By contrast, conveying meaning through allusion, the use of well-chosen proverbs or images, is highly valued. Although Gurnah’s prose is not among the most allusive within African literatures, indirect language holds a high place in Swahili culture. This is why I was particularly interested in examining how the translator approached the sexual jokes and innuendoes exchanged at length by certain characters in Paradise, given that Kiswahili is not a language characterized by lexical vulgarity.

“In between he sang snatches of love songs and puffed at a biri. ‘Please be kind to me for once, you goatfuckers. I’d love nothing better than to sit here and and stroke your anuses all day, but I’ve got other loads to pick up, bob. So can you put a bit of shoulder into it and stop sniffing each other’s shit. […] There’s a funny smell in this place. I am sure I can pick up the perfume of rotting cocks, but maybe that’s the food they give you. Hey, baba! What do they give you here? There’s a lot of grease in that sweat running down your backs. They love fatty meat up there where you’re going, bob, so watch where you put your fannies. Come on, my brother, stop scratching yourself there.’ […] The porters roared and cackled with laughter at the driver’s foul mouth. Using such a language in front of the merchant! When he faltered, they taunted him with insults about his mother and father, and crude insinuations about his own children. ‘Come and suck my cock,’ he would say, clutching his crotch. Then he would start off again.” (2021, 54)

“Kati ya yote aliimba nyimbo za mapenzi huku akivuta sigara ya biri. ‘Tafadhali kuweni na huruma na mie angalau mara hii, nyinyi waparamia mbuzi. Nisingependa kitu zaidi ya kuwashika shika tu hapo kutwa nzima, lakini nina mizigo mingine ya kuchukua, bob. Kwa hivyo mnaweza kujishughulisha kidogo na muwache kunusananusana hapo. […] Kuna harufu ya ajabu mahali hapa. Nina hakika nasikia harufu ya vijogoo vilivyooza, lakini labda ndicho chakula wanachowapa. Wee, baba! Wanakupa nini hapa? Mbona kuna mafuta mengi hivyo ndani ya jasho linalowatiririka miongoni mwenu. Wanapenda nyama za mafuta huko mnakoenda, bob, kwa hivyo angalieni mtakapowatumbukiza hao wajomba zenu. Haya wee ndugu yangu acha kujikuna hapo.’ […] Wapagazi waliangua vicheko vya kelele kwa mdomo mchafu wa derewa. Yaani huyu alithubutu kutumia lugha kama hiyo mbele ya Mfanyabiashara! Alipoyumba, walimdhihaki kwa matusi kuhusu mama na baba yake, na maneno machafu juu ya wanawe yeye mwenyewe. ‘Njoo njoo umnyonye jogoo wangu’ angesema, akishika sehemu zake za siri. Kisha angeanza tena.(2022, 55)

22Most of the time, as in Swahili life, the words will not be crude, but innuendoes will be understood without difficulties. Here, it is interesting to note that Gurnah, writing “Stop scratching yourself there” (2021, 54; emphasis mine), seems to be hinting at what can be done in Kiswahili. As we can see, Ida Hadjivayanis uses localization on two other occasions thanks to “hapo” (meaning “here”/ “there”) which stands for “anuses” and “shit” in her translation.

23The rude word “fannies” also disappears but the crudeness of the allusion can all the same be felt through the use of the verb “kutambukiza” (meaning to “introduce,” “put into,” “dip”) and a word-to-word translation of the image would be: “So that you better watch where you are going to introduce/dip your uncles.” “Crotch” or other words to refer to genitalia or loins will be paraphrased by “sehemu za siri” (secret parts) or “sehemu za chini” (the parts below) as it is nearly always the case in the everyday life of the Swahili people. As for “cock,” it is either “jogoo” (rooster) or its diminutive form “kijogoo” that is used but it is widely known that this “jogoo” also stands for the other “cock” in English. In other passages, the nickname “abdalla” or “little abdallas” (viabdalla) appears, as in the original by Gurnah.

24We could therefore argue that if not directly expressed in the same manner, erasing or toning down the crude words (for example “goatfuckers” is translated as “people riding/grasping goats”), clarifies the intention all the same and, besides, conforms to the Swahili way of expressing things, by images, innuendoes that convey the same salacious notes. The use of reduplication “kuwashikashika tu” or “kunusananusana” is also very evocative and conveys what is implied in the English text. All this tends to show that, even if a lot was already in the original English novel in terms of local reception, some characteristics are to be credited to the translation, which manages to tell the story “in a Swahili way” and by using the literary means of the Swahili language.

The Swahili way of telling a story

26Here are a few examples:

“She was not to give herself airs just because she has grown up on the coast among civilized people” (2021, 13).	“Mwanamke asijidai kuringa ati kwa sababu amekulia pwani kati ya wastaarabu” (2022, 14).
“Would you like to go on a little trip, little octopus?” (16)	“Eti wee pweza mdogo, ungependa kusafiri kidogo?” (16)
“The porter shouted for people to make way – ‘Let the seyyid pass, waungwana’” (20).	“Mbeba mizigo alipiga kelele kwa watu kumpisha, ‘Hebu, mwacheni Seyyid apite, waungwana’” (21).
“He told the story of the beautiful woman who, after complicated stratagems on his part, agrees to take him home, then turns out to be a man. Or the old lady he negotiates with, taking her to be a pimp, but who turns to be the prostitute he was paying for” (86).	“Alielezea mkasa wa mwanamke mmoja mzuri, ambaye baada ya yeye kutumia mbinu zote za kumtongoza alikubali kumchukua nyumbani kwake huyo mwanamke, lakini kumbe alikuwa ni mwanamme. Au pia bi kizee ambaye yeye akipatana naye kama kuwadi, kumbe yeye mwenyewe huyo bi kizee ndo alikuwa malaya ambaye akimlipa” (88).

27It must be noted that playing with interjections or onomatopoeia can be done the other way round, as the translation pays close attention to rhythms and sonorities. Hadjivayanis thus transposed the evocative “chomp chomp” into a sequence of verbs with rhyming endings, using such a succession – quite common in Kiswahili – to express that nothing is left.

28Another resource of Swahili is “reduplication,” which conveys intensity while also being very close to what is felt in the English text, as can be seen here with the play on sonorities:

“Grumble, mumble, what’s the good of that?” (87)

“Kazi kunung’unikanung’unika tu, inasaidia nini?” (89)

29Conversely, although Ida Hadjivayanis in some ways recovers a manner of storytelling “in a Swahili way,” she also preserves some aspects of Gurnah’s original, particularly his use of punctuation. While Swahili texts are typically less punctuated than English or French ones, a comparison of the original and its Swahili translation shows how close she sticks to Gurnah’s text – “down to the comma,” as one might say. In this respect, Hadjivayanis departs from a long-dominant tradition of domestication in Swahili translation (Talento, forthcoming). Her retranslation of Alice in Wonderland (2015) can illustrate this shift: she restores Alice to her English setting, whereas the earlier translation had recast her as an African child in an East African context. She likewise aims to reproduce Carroll’s rhythm and syntax, as she notes in her introduction:

• 11 “Tofauti kuu kati ya tafsiri hii na ile ya awali ni kwamba tafsiri hii imejitahidi kujengeka katika (...)

The main difference between this translation and the previous one is that I built mine on what Alice really is. She is a young English girl who lives in England. I also try to respect closely the style of Carroll and used the same long sentences, that develop themselves and connect with different punctuation such as commas and dashes. (2015, vii; my translation)11

30We thus see the translator’s delicate negotiation between close adherence to the original English text and adaptation to the target language and culture.

Another “Third text”: when spoken and written words in Kiswahili travel back through fiction

31While the “Third text” is by nature elusive, what is very challenging in the case of Paradise, is that the novel draws on real testimonies, told and written in Kiswahili. The actual existence of Swahili subtext allows a comparison of the texts across centuries. For this reason, I now turn my attention to two specific excerpts that warrant closer examination. Both are taken from “The Gates of Flames,” a section of Paradise that heavily draws on written sources in Kiswahili, namely the Safari yangu ya barra Afrika, told by Sleman bin Mwenye Chande at the end of the nineteenth century. This oral travelogue (or khabar to use the Swahili genre) was transcribed from dictation and published by Carl Velten at the beginning of the twentieth century (Velten 1901; Carré 2016; Hadjivayanis 2023). Gurnah, who may have read either the Swahili or English version, has used the material to build on his story (Carré 2014; 2021).

32The first passage under consideration occurs at the moment when the caravan enters Tayari (Gurnah 2021, 130–31). A line-by-line comparison would be too long, but the initial sections of the three texts already illustrate key features.

Auwali Tabora Kanyenye, mji wa Mw’arabu, jina lake Muhina bin Slemani Le’urubi pamoja na Sa’idi bin ‘Othmani. Na mji wa pili unakuitwa Bahareni, wa Mw’arabu, na mwenyewe jina lake Sa’id bin Ali. Mji wa tatu unakwitwa Lufita, mwenyewe Sa’idi bin Salum pamoja na Mwenyi Mlenda, mtu wa Mrima. (Mwneye Chande 1901, 8)

Kwanza kulikuwa na Kanyenye, ambayo ilikuwa mali ya Mwarabu kwa jina la Muhina bin Seleman El-Urubi. Na sehemu ya pili ya mji iliyoitwa Bahareni ilikuwa ya Mwarabu aliyeitwa Said bin Ali. Na ya tatu ikiitwa Lufita, na hiyo ilikuwa mali ya Mwenye Mlenda, aliyetoka Mrima huko mwambao. (Hadjivayanis 2022, 133)

First, there was Kanyenye, which belonged to an Arab whose name was Muhina bin Seleman El-Urubi. And the second part of the town was called Bahareni, and the Arab it belonged to was Said bin Ali. And the third was called Lufita, which belonged to Mwenye Mlenda, a man from Mrima on the coast. (Gurnah 2021, 130–131)

34Another subtle change is that the word “awali” disappears in Peponi. In the nineteenth century, and before, many Swahili chronicles used to begin in such a way, as for example the Akhbar Pate (the Pate Chronicle) or others. Using this word in telling his story, Mwenye Chande is signalling that he is including a very short – but trustworthy – history of Tabora. This “mini-chronicle” is to be found, almost unchanged, in the novel where a fragment of “real history” is embedded in fiction (though this may be impossible to identify for the reader, especially in Europe, even if the reference to Tippu Tip can attract attention). The second extract takes place when Yusuf and the other members of the caravan are crossing Lake Tanganyika to go to Sultan Chatu’s territory. A storm rises, and in the face of danger, offerings are made to the spirit Pembe. Here again, Gurnah uses Mwenye Chande’s testimony: a single paragraph by Mwenye Chande (1901, 23) becomes more than two pages in Gurnah (Gurnah 2021, 146–48). I have previously examined the relationship between the two texts in an earlier article (2021) and will not revisit that discussion here. However, a close reading of the texts reveals distinct constellations of words and meanings.

36We thus observe a disjunction between the terminology of the palimpsest’s initial layer and that of Paradise. By affirming this, I certainly do not want to suggest that this “first” text has been “betrayed.” Any author is free to use his or her sources as (s)he chooses (and, in fact, the narratives are largely similar here). I would rather suggest that the choice of particular words and the way they echo one another, create networks of meaning that are worth analyzing, because they sustain imaginaries, even if they operate beneath the surface. In this episode, the wind (upepo) subsides because the spirit (pepo) has received offerings. The link between pepo (spirit) and upepo (wind) is not truly perceptible in the English version. By contrast, this connection is clearly present in both Swahili versions and, quite surprisingly, even more strongly in Peponi, where the same word (pepo) is used to refer both to the violent wind and to the spirit, as we can see below:

The spirit of the shrine is Pembe (Gurnah, 2021, 146)	Pepo wa tambiko hili ni Pembe (Hadjivayanis, 2022, 149)
By the time they had returned to the boats, the wind had dropped. (Gurnah, 2021, 147)	Ilipofikia wakati wa kurudi kwenye mitumbwi, pepo zilikuwa zimetulia ghafla. Hadjivayanis, 2022, 150)

38The loop created by the translation of Paradise into Kiswahili reveals a remarkable plurality of influences, all woven together. Gurnah – an “interpreter,” in Coetzee’s term – finds his novel reshaped in another language and by another mediator, the translator. The English original, infused with a “Third text,” informed by oral traditions as well as Swahili texts, gives birth to new constellations of meanings that shine through translation. If such circulations characterize all literary works, it is only fair that the stories should be carried into African languages for the benefit of readers who have been marginalized for a long time.