mc NO 17 2021 manuscript cultures Hamburg | Centre for the Study of Manuscript Cultures ISSN 1867–9617 PUBLISHING INFORMATION | MANUSRIPT CULTURES Publishing Information One Text, Many Forms – A Comparative View of the Variability of Swahili Manuscripts Edited by Ridder H. Samsom and Clarissa Vierke Editors Editorial Office Prof. Dr Michael Friedrich Dr Irina Wandrey Universität Hamburg Universität Hamburg Asien-Afrika-Institut Centre for the Study of Manuscript Cultures Edmund-Siemers-Allee 1 / Flügel Ost Warburgstraße 26 D-20146 Hamburg D-20354 Hamburg Tel. no. +49 (0)40 42838 7127 Tel. No.: +49 40 42838 - 9420 Fax no. +49 (0)40 42838 4899 Fax No.: +49 40 42838 - 4899 michael.friedrich@uni-hamburg.de irina.wandrey@uni-hamburg.de Prof. Dr Jörg H. Quenzer Layout Universität Hamburg Nora Harms Asien-Afrika-Institut Edmund-Siemers-Allee 1 / Flügel Ost Cover D-20146 Hamburg First page of an Arabic Qas īda Hamziyya manuscript, copied by Abī Bakr Tel. no. +49 (0)40 42838 7203 bin Sulṭān Aḥmad in 1311 AH/1894 CE, with annotations in Swahili and Fax no. +49 (0)40 42838 6200 Arabic. Private collection of Sayyid Ahmad Badawy al-Hussainy (1932- joerg.quenzer@uni-hamburg.de 2012) and Bi Tume Shee, Mombasa. Translations and Copy-editing Carl Carter, Amper Translation Service Print optimal media GmbH, Röbel/Müritz Printed in Germany www.csmc.uni-hamburg.de ISSN 1867-9617 © 2021 Centre for the Study of Manuscript Cultures Universität Hamburg Warburgstr. 26 20354 Hamburg manuscript cultures mc NO 17 1 CONTENTS | ONE TEXT, MANY FORMS CONTENTS EDITORIAL 2| One Text, Many Forms: A Comparative View of the Variability of Swahili Manuscripts Ridder H. Samsom and Clarissa Vierke ARTICLES 13 | Writing in Swahili on Stone and on Paper Ann Biersteker 29 | Swahili Manuscripts from Northern Mozambique: Some Notes on Ajami Correspondence Letters Chapane Mutiua 53 | Arabic-Swahili Hamziyya Manuscripts: Observations on Two Testimonies of the Text Ahmed Parkar 65 | A Network of Copies: Transmission and Textual Variants of Manuscript Traditions from the J. W. T. Allen Collection (Dar es Salaam) Annachiara Raia 87 | Writing Songs, Singing Texts: Orality and Literacy in Swahili Manuscripts Clarissa Vierke 105 | Contributors manuscript cultures mc NO 17 2 SAMSOM AND VIERKE | ONE TEXT, MANY FORMS Fig. 1: Maalim Moliti preparing for a madrasa class by writing down translations of Arabic formulas in Swahili and, partly, Emakhuwa; Mozambique Island, Mozambique. manuscript cultures mc NO 17 3 SAMSOM AND VIERKE | ONE TEXT, MANY FORMS Editorial One Text, Many Forms: A Comparative View of the Variability of Swahili Manuscripts Ridder H. Samsom and Clarissa Vierke | Hamburg and Bayreuth, Germany The papers in this issue of manuscript cultures explore the variability of Swahili manuscripts in terms of their graphical representation and material design and address the question of how manuscripts differ in form across time, regions and text traditions. The articles also discuss the reasons for variability and the interdependence of manuscripts and practices: to what extent can variations in manuscripts’ representation be attributed to differences in the practice and purpose for which they were used? How have writing practices changed over time? These papers are the result of a workshop with the title ‘One Text, Many Forms: A Comparative View of the Variability of Swahili Manuscripts’ that was held at the Sonderforschungsbereich 950 ‘Manuskriptkulturen in Asien, Afrika und Europa’ and within the scope of the Centre for the Study of Manuscript Cultures at the University of Hamburg on 21 and 22 April 2017. Regarding manuscript cultures found in Africa, the coast of the western Indian Ocean, which is often referred to as ‘the Swahili coast’, is commonly mentioned as an area with a long-standing tradition of writing linked to its Muslim tradition of learning. With the arrival of Islam, literacy in Arabic and the Arabic script reached the East African coast before the end of the first millennium. The oldest preserved writings can be found on coins from the coastal sultanates and in stone inscriptions dating back to the twelfth and thirteenth centuries.1 The art of writing had reached a high standard by the seventeenth century, as shown by artwork and inscriptions in stone, ivory and metal2 – an aspect further explored by Ann Biersteker in her contribution to this volume. However, it is mostly in the eighteenth century that we can find substantial proof of Swahili texts being written in Arabic script, namely on paper.3 1 Hichens 1939, 119. 2 Hichens 1939, 120. 3 Whiteley 1969, 38; Miehe 1990, 202. mc NO 17 Apart from letters (see the paper by Chapane Mutiua) and contracts, poetry is the genre that accounts for the largest body of ancient texts in Swahili written in Arabic script, and it is these pieces of writing that have also attracted considerable interest among scholars since the end of the nineteenth century. Approximately eighty text editions based on Swahili manuscripts in Arabic script have been published by around thirty academics and experts in the Swahili language.4 These publications of Swahili manuscripts are mainly in the form of critical text editions which emphasise the content of a particular text and analyse it historically, linguistically or as literature. Despite the sizeable corpus and these critical text editions of various literary achievements such as the laudatory poem Hamziyya, Chuo cha Herkal, an epic poem about battles in the early history of Islam, or Al-Inkishafi, a philosophical poem on the brevity of life, as well as the poems of Fumo Liyongo, the earliest East African hero and Mwana Kupona, a guideline for marriage life, or polemical, satirical and dialogue verses of Muyaka in the tradition of the mashairi genre, there are no established conventions of codicological research yet from which any clear generalisations about all genres of Swahili manuscript culture can be made. To date, only a small number of scholars have asked questions about the manuscripts themselves such as who wrote them, when and what for, and how and why they were written; William Hichens, John W. T. Allen and Mohammed H. Abdulaziz are among the few who have attempted to describe the material characteristics of the manuscripts they have been dealing with.5 Apart from the short articles by 4 Steere 1876, Taylor 1891, Stigand 1915, Büttner 1892, Neuhaus 1896, Velten 1901, Werner 1917, Meinhof 1925, Hichens 1939, Sacleux 1939, Dammann 1940, Allen 1945, Hinawy 1950, Harries 1950, Lambert 1952, Whiteley 1957, Knappert 1958, 1979, Harries 1962, Alpers 1967, Allen 1971, Abdulaziz 1979, Mgeni bin Faqihi 1979, Biersteker and Shariff 1995, Mulokozi and Sengo 1995, Liyongo Working Group 2004, Mutiso 2005, Saavedra 2007, Miehe and Vierke 2010, Vierke 2011, Raia 2017. 5 Hichens 1939; Allen 1971; Abdulaziz 1979. manuscript cultures 4 SAMSOM AND VIERKE | ONE TEXT, MANY FORMS Fig. 2: First opening of Arabic Qa īda Hamziyya manuscript of 66 pages, copied by Abī Bakr bin Sulṭān Aḥmad in 1311 AH/1894 CE, with annotations in Swahili manuscript cultures mc NO 17 SAMSOM AND VIERKE | ONE TEXT, MANY FORMS 5 and Arabic. Private collection of Sayyid Ahmad Badawy al-Hussainy (1932-2012) and Bi Tume Shee, Mombasa. mc NO 17 manuscript cultures 6 Fig. 3: Qa īda Hamziyya, detail showing pencil annotations, Mombasa. Private collection Sayyid Ahmad Badawy al-Hussainy and Bi Tuma Shee. Simon Digby and James de Vere Allen, we hardly have any information at our disposal about the production of Swahili manuscripts before the printing era started in the second half of the nineteenth century.6 The contributions to this issue written by Ahmed Parkar, Annachiara Raia and Clarissa Vierke seek to dig deeper into history before practices of printing and reading fostered by Western school education introduced in colonial times left their imprint on Swahili manuscript culture and on other co-existing textual practices. However, as the paper that Chapane Mutiua has authored concerning letter-writing and correspondence during the era of Portuguese colonialism underlines, the ‘modern’ colonial state and its practices of reading and writing did not just make the local manuscript culture vanish, but it actually fostered the continued usage of Swahili in Arabic script. So far, the interface between Swahili manuscript culture and oral poetic traditions and performance practices has hardly been considered, even though they co-existed in the past and still do to this day. This is surprising since in a society such as the Swahili-speaking one where literacy was and still is restricted to a certain extent and orality is emphasised, Swahili manuscripts cannot be 6 SAMSOM AND VIERKE | ONE TEXT, MANY FORMS considered as being independent from oral performance and composition. Thus, one of the main concerns that some of the contributions deal with, like the articles by Annachiara Raia and Clarissa Vierke, is the relationship between manuscripts and oral practices, which is the root cause of much of the variability that exists. The earliest Swahili manuscripts known so far emerged from a practice of translation involving Arabic texts that are generally regarded as canonical in Swahili Islamic culture. In the predominantly oral culture of the Swahili coast, where knowledge was transmitted by word of mouth (and still is to a large extent), translation, rendition and adaptation from Arabic texts can be considered the main incentive for Swahili literacy and Swahili manuscript culture. Consequently, early translations from Arabic into Swahili, which were not commissioned by Europeans, cover a particularly interesting period and text corpus. These texts include corpuses of narrative poetry like the Chuo cha Herkal or the Qissat alYusuf (dealt with in Annachiara Raia’s contribution) and the hymnal Tabaraka, the Hamziyya (which Ahmed Parkar discusses in his paper) and the Maulidi Barzanji, the poem narrating the Prophet’s birth. These are just a few examples of texts written again and again by scribes who rendered them in different ways and styles in manuscripts. As Ahmed Parkar has shown, for instance, apart from idiosyncrasies in the usage of Arabic script, much of the variability one can find is due to the differing arrangement of lines, variable marking of pauses and the insertion of glosses and interlinear translations. Although these manuscripts and their variations provide unique access to an epoch when manuscripts started to be used systematically to transmit knowledge in Swahili and when forms of written representation of oral narrations were only just starting to emerge, they have hardly been studied yet. The manuscripts may also help us to reconstruct a history of Swahili manuscript production and reception. One pertinent question here is how Swahili scribes adapted Arabic manuscript conventions. In the case of bilingual manuscripts, how are the two languages arranged or referred to? How is Swahili oral practice ‘translated’ into a manuscript page? How have Quranic recitations and commentaries been adapted in various contexts of ‘vernacularisation’, for instance? And how do translation and commentary practices in either written or oral form mediate between the text, the performer and the audience? Simon Digby 1975; Allen 1981. manuscript cultures mc NO 17 SAMSOM AND VIERKE | ONE TEXT, MANY FORMS Following the comparative perspective suggested, the presentations and articles chosen for this issue cover a large geographical and historical spectrum of Swahili manuscript culture. They include evidence of early Swahili literacy in stone and wood, three different bilingual testimonies in Swahili and Arabic of the canonical Arabic qaṣīda by al-Būṣīrī, the Hamziyya, which speak of early translation practices and oral performances, nineteenth-century letter-writing in northern Mozambique and its hinterland during Portuguese colonialism, and the transmission, collecting, copying, transliterating and translating of textual practices concerning the popular Utendi wa Yusuf, the story of the Prophet Yusuf, and, finally, the essential question about the meaning of the concept of the written text in relation to its performance or vice versa.7 In her article, Ann Biersteker (emerita Michigan State University) considers whether ancient carvings and inscriptions might contribute to our understanding of ‘the variability in design, layout and phrasing’ found in later Swahili manuscripts. Epitaphs on tombs seem to provide evidence of early and continuous writing in code-mixed Arabic and Swahili and in Swahili along the East African coast. The evidence they provide is limited, though, as it primarily consists of Swahili honorific titles and proper names that are also Swahili words. Chapane Mutiua (University of Hamburg, Germany and Eduardo Mondlane University, Maputo, Mozambique) shows in his contribution that the process of diffusion of Ajami literacy through Islamic education and the use of Arabic script in Kiswahili link this literary tradition to the wider Swahili region. Its integration as part of the colonial system of communication and the use of elements of Portuguese and northern Mozambican languages as loanwords make it peculiar to northern Mozambique. Written correspondence also shows that although people in the region spoke different languages such as Kimwani, Ekoti, Enattembo, Emmakhuwa and Portuguese, Kiswahili was the dominant language for literary production and letter-writing. 7 Ahmed Parkar (Pwani University, Kilifi, Kenya) compares three manuscripts relating to the Hamziyya and explains that one major feature in them is their divergent textual layout. He also shows that a single text in Swahili society can be written in various ways in a given time and place. Two of the manuscripts under discussion show they were probably used as an aid for chanting and for private reading as well. This is because their textual layout contains the double column system and is narrowly spaced, which indicates that they were mainly used for preparing oral performances. One of the manuscripts was probably used for teaching and learning purposes. Fig. 4: Mr. Sh Zuheri Ali Hemed (1944–2015) showing a manuscript of 400 pages with tafsīr in Swahili, composed and written by Sh Aliy Hemed Abdallah alBuhriy (c.1870–1957). Private collection of the family of Sh Zuheri Ali Hemed. 7 Variability in the way the Arabic script has been used for the Swahili language has been dealt with in two presentations (by Prof Mohammed Abdulaziz and Sh Mahmoud Ahmad Abdulkadir ‘Mau’) which are not contained in this volume, nor is the presentation by Eugeniusz Rzewuski on the recurrence and variability in ‘Shirazian’ sagas from northern Mozambique. mc NO 17 manuscript cultures 8 SAMSOM AND VIERKE | ONE TEXT, MANY FORMS Fig. 5: tafsīr manuscript in Swahili and the Qur’anic Arabic in red ink, composed and written by Sh Aliy Hemed Abdallah al-Buhriy (c.1870–1957), p. 1. Private collection of the family of Sh Zuheri Ali Hemed (1944–2015). manuscript cultures mc NO 17 9 SAMSOM AND VIERKE | ONE TEXT, MANY FORMS Fig. 6: tafsīr manuscript in Swahili and the Qur’anic Arabic in red ink, composed and written by Sh Aliy Hemed Abdallah al-Buhriy (c.1870–1957), p. 28. Private collection of the family of Sh Zuheri Ali Hemed (1944–2015). mc NO 17 manuscript cultures 10 Annachiara Raia (University of Leiden, Netherlands) concludes that what has emerged from the overview of the different Utendi wa Yusuf manuscript groups is a mosaiclike scenario showing how much the texts have diverged, as they were handwritten and re-written over the centuries. Side by side with the figure of the poet/adapter committed to producing a copy of the poem, she has also introduced the figure of the singer/adapter committed to giving voice to a written text and editing it at his or her own discretion. Such a ‘re-performance’ must also be seen as another form through which a story is re-formulated and transmitted. Clarissa Vierke (University of Bayreuth, Germany) underlines in her paper that writing is not a straightforward process, but can serve different roles and functions. On the one hand, it can be strongly linked to the very idea of a poem such that prosodic features become visual features of the script. On the other hand, however, she uses examples from the genres of the ancient utumbuizo and competitive poetry in shairi form to show that writing can be so entangled with orality that the question of what a text is, where it starts and where it ends becomes difficult to answer. In the latter case, each version of the text is but a temporary constellation of its parts, which recur in a different arrangement in the next manuscript and performance. The utenzi genre with its strong emphasis on writing and fixity despite its own variability is the more peculiar case. It mimics the use, reading and writing of Arabic texts in the Swahili context more closely, probably also because it is clearly derived from a practice of adaptation or translation that emphasises the Arabic original. The editors of this volume are of the opinion that each of the five papers published here contributes to the development of a full set of instruments for dealing with past literacy in Swahili on the one hand and Swahili’s prominence in orality on the other. The papers demonstrate that codicological and manuscriptological methods of analysing Swahili manuscripts can be on a par with the arts of literary criticism, linguistics, history and other social sciences that are needed to answer the question of what role Swahili manuscripts in Arabic script once played – and still play today – in the transfer of knowledge in Swahili society. SAMSOM AND VIERKE | ONE TEXT, MANY FORMS REFERENCES Abdulaziz, Mohamed H. (1979), Muyaka: 19th Century Swahili Popular Poetry, Nairobi: Kenya Literature Bureau. Allen, James de Vere (1981), ‘Swahili Book Production’, Kenya Past and Present, 13: 17–23. Allen, John W. T. (1945), ‘Arabic Script for Students of Swahili Tanganyika Notes and Records’, Tanganyika Notes and Records, Supplement November 1945: 7–78. —— (1970), The Swahili and Arabic Manuscripts and Tapes in the Library of the University College Dar-es-Salaam, Leiden: Brill. Allen, John W. T. (ed.) (1971), Tendi: Six Examples of a Swahili Classical Verse Form with Translations & Notes, New York: Africana Publishing Corporation. Alpers, Edward A. (1967), ‘A Revised Chronology of the Sultans of Kilwa in the Eighteenth and Nineteenth Centuries’, Azania, 2: 145–163. Biersteker, Ann and Ibrahim Noor Shariff (1995), Mashairi ya Vita vya Kuduhu: War Poetry in Kiswahili Exchanged at the Time of the Battle of Kuduhu, East Lansing: Michigan State University Press. Boxer, Charles R. (1957), ‘A Glimpse of the Goa Archives’, Bulletin of the School of Oriental and African Studies, 14/2: 299–325. Büttner, Carl Gotthilf (1892), Suaheli-Schriftstücke in arabischer Schrift, mit lateinischer Schrift umschrieben, übersetzt und erklärt von Dr. C. G. Büttner (Lehrbücher des Seminars für Orientalische Sprachen zu Berlin, 10), Stuttgart/Berlin: W. Spemann. Dammann, Ernst (1940), Dichtungen in der Lamu-Mundart des Suaheli, Hamburg: Friederichsen. Digby, Simon (1975), ‘A Qur’an from the East African Coast’, Art and Archeology Research Papers, 7: 49–55. manuscript cultures mc NO 17 SAMSOM AND VIERKE | ONE TEXT, MANY FORMS 11 Harries, Lyndon (1950), ‘Swahili Epic Literature’, Africa, 20: 55–59. Mgeni bin Faqihi (1979), Utenzi wa Rasi ‘lGhuli, ed. by Leo van Kessel, Dar es Salaam: Tanzania Publishing House. —— (1962), Swahili Poetry, Oxford: Clarendon. Mulokozi, M. M. and Sengo, T. S. (1995), History of Kiswahili Poetry: A.D. 1000–2000, a Report, Dar es Salaam: Institute of Kiswahili Research University. Hichens, William (1938), ‘The Lamu Chronicle’, Bantu Studies, 12: 2–33. —— (1939), Al-Inkishafi: The Soul’s Awakening, London: Sheldon Press. Hinawy, Mbarak Ali (1950), Al-Akida and Fort Jesus, Mombasa, London: Macmillan and Co. Knappert, Jan (1958), Het epos van Heraklios. Een proeve van Swahili poëzie. Tekst en vertaling, voorzien van inleiding, kritisch commentaar en aantekeningen, PhD thesis, Leiden University, Alkmaar: Hofman. —— (1979), Four Centuries of Swahili Verse: A Literary History and Anthology, London/Nairobi: Heinemann. Lambert, H. E. (1952), ‘A Specimen of Lamu Prose’, Bulletin of the East African Inter-Territorial Language (Swahili) Committee, 22: 14–27. Liyongo Working Group (2004), Liyongo Songs: Poems attributed to Fumo Liyongo, collected and edited by the Liyongo Working Group, Cologne: Köppe. Meinhof, Carl (1924/25), ‘Das Lied des Liongo, kritisch bearbeitet und übersetzt’, Zeitschrift für EingeborenenSprachen, 15: 241–265 (English translation in Liyongo Working Group 2004). Miehe, Gudrun (1990), ‘Die Perioden der Swahililiteratur und ihre sprachliche Form’, Paideuma, 36: 201–215. —— and Clarissa Vierke (2010), Muhamadi Kijuma: Texts from the Dammann Collections and other Papers, in collaboration with Sauda Barwani and Ahmed S. Nabahany, Cologne: Köppe. mc NO 17 Mutiso, Kineene wa (1997), Utenzi wa Hamziyyah, Dar es Salaam: Taasisi ya Uchunguzi wa Kiswahili. Neuhaus, Gustav (1896), Suaheli-Manuskripte in photolithographirten Originalen, Berlin: Reichsdruckerei. Raia, Annachiara (2017), The Utendi wa Yusufu: A Critical Edition of the Swahili Poem of Yusuf and a Study of its Adaptation at the Swahili Coast, unpublished PhD thesis, University of Bayreuth and University of Naples L’Orientale. Saavedra, José Casco (2007), Utenzi, War Poems, and the German Conquest of East Africa: Swahili Poetry as Historical Source, Trenton, NJ/Asmara: Africa World Press. Sacleux, Charles (1939), Dictionnaire Swahili–Français (Travaux et Mémoires de l’Institut d’Ethnologie de l’Université de Paris, vols 36/37), Paris: Institut d’Ethnologie. Samsom, Ridder H. (2015), ‘Swahili Manuscripts: Looking in East African Collections for Swahili Manuscripts in Arabic Script’, in: Hannelore Vögele, Uta Reuster-Jahn, Raimund Kastenholz and Lutz Diegner (eds), From the Tana River to Lake Chad: In memoriam Thomas Geider, Köln: Rüdiger Köppe Verlag, 243–284. Steere, Edward (1876), The Arabic Alphabet as Used in Writing Swahili, Zanzibar: Central African Mission. Stigand, C. H. (1915), A Grammar of Dialect Changes in the Kiswahili Language, Cambridge: Cambridge University Press. manuscript cultures 12 SAMSOM AND VIERKE | ONE TEXT, MANY FORMS Taylor, William E. (1891), African Aphorisms: Or, Saws from Swahili-Land, London: Society for Promoting Christian Knowledge. Velten, Carl (1901), Praktische Anleitung zur Erlernung der Schrift der Suaheli, Stuttgart/Berlin: Spemann. Vierke, Clarissa (2011), The Poem of the Palanquin: A Text Edition of the ‘Utendi wa Haudaji’ together with a Textual Analysis Approaching the Style of the Nineteenth-Century Swahili Utendi, Berlin/Münster: Lit. Werner, Alice (1917), ‘The Utendi wa Mwana Kupona’, in O. Bates (ed.), Harvard Studies, vol. 1: Varia Africana I, Cambridge, MA: Peabody Museum, 147–181. Whiteley, Wilfred (1969), Swahili: The Rise of a National Language, London: Methuen. Whitely (sic), W. H. (1974), Maisha ya Hamed bin Muhammed el Murjabi yaani Tippu Tip kwa maneno yake mwenyewe (Kampala/Nairobi/Dar es Salaam: East African Literature Bureau; reprint of 1958/1959 Supplement to The East African Swahili Committee Journals No. 28/2, July 1958 and No. 29/1 January 1959). PICTURE CREDITS Fig. 1: © Photography by Clarissa Vierke, Mozambique Island, 5 August 2015. Figs 2–3: © Photography by Ridder H. Samsom, Mombasa, Kenya, 11 October 2011. Figs 4–6: © Photography by Ridder H. Samsom, Tanga (Tanzania) 27 August 2012. manuscript cultures mc NO 17 13 BIERSTEKER | WRITING IN SWAHILI Article Writing in Swahili on Stone and on Paper Ann Biersteker | Milwaukee, WI, USA We know that manuscripts and likely collections of manuscripts existed along the East African coast during the late fifteenth and early sixteenth century. The leaders of East African coastal city-states appear to have seen their cities as learned communities and to have respected other such communities. In March 1497 when Nicolau Coelho entered what later became Cabacceira in Mozambique, the leader of that community asked to see Coelho’s ‘books of the law’.1 Luís Vaz de Camões’ Os Lusíadas also indicates that coastal peoples asked the Portuguese for their ‘books of our laws, commandments, or faith’.2 As Adrien Delmas has noted,3 Mutual recognition by respectively presenting sacred texts should not come as a surprise. For the Portuguese, the idea was to find Christian allies in their endeavour to circumvent Islam, and this ceremonial exchange was repeated along the East African coast, in Mombasa and Malindi, as well as on the way back in Mogadiscio and Kilwa, but also in other places in Africa and the Indian Ocean. For the Africans, the need to identify newcomers was obvious, and while they were hoping to size up their strength by the sound of their cannons, it was through their writings that they intended to find out their intentions. Subsequently, during the first attack the Portuguese made along the East African coast, the items confiscated included ‘books of the law’.4 The other items seized were distributed to the sailors involved in the looting, but the ‘books of the law’ were ‘kept back to be shown to the king’.5 We do not know why these books were held separately for the King of Portugal; perhaps because of their aesthetic qualities or because they provided evidence that these were Islamic societies. Nor do we know what happened to these ‘books’ – 1 João de Sá or Alvaro Velho 1898. 2 Camões 1997, 15. 3 Delmas 2017, 182. 4 João de Sá or Alvaro Velho 1898. 5 João de Sá or Alvaro Velho 1898. mc NO 17 if they were eventually given to the King of Portugal and are now in some Portuguese archive or if they sank in one of the Portuguese ships lost during the journey. Presumably, these works were written in Arabic and they were books that guided people in these societies in some sense. They may have been copies of the Qur'ān or other religious or legal texts. It is almost certain that they were manuscripts rather than printed books as the first printed Qur'ān was not produced until 1538 and it was printed in Venice for Christian missionaries (albeit with many errors);6 the first Qur'ān printed by Muslims for Muslims was not published until 1787 CE.7 There was much earlier block printing in Arabic for decorative purposes, but according to Jonathan Bloom, ‘The earliest Arabic printed text that has survived is the Kitāb salāt al-sawā'ī, a book of hours produced in Fano by the master Venetian printer Gregorio de’ Gregori in 1514’.8 With reference to the late nineteenth century, Anne K. Bang speaks of ‘a sphere of Islamic learning that encompassed the Indian Ocean shores, from the East African coast to the archipelago of South-East Asia’.9 To what extent did such a sphere exist in the fifteenth, sixteenth, seventeenth, and eighteenth centuries? We will clearly need to study a wide range of texts written on numerous types of surfaces to have any sense of this.10 In addition to knowing a little about other types of manuscripts produced before the nineteenth century,11 6 Bloom 2001, 10. 7 Bloom 2001, 10. 8 Bloom 2001, 219. 9 Bang 2011, 90. 10 In an earlier paper, ‘The Inscribed Object: The Textures and Textuality of Writing in Eastern Africa’ (Biersteker 2018), I considered the history of writing in coastal East Africa. 11 The oldest confirmed Swahili writing on paper is correspondence written in the early eighteenth century (1711–1724): letters that were confiscated by the Portuguese and taken to the Portuguese archives in Goa; see Alpers, 1975; Sultan Fatima binti Muhammad Mkubwa 2007, Omar and Frankl 1994, and Ashley Lynn 2005. The dating of earlier manuscripts including ‘Hamziyya,’ ‘Chuo cha Herkal’/ ‘Utendi wa Tambuka,’ and ‘Swifa ya Mwana Manga’ is disputed. manuscript cultures 14 we know that local rulers in East Africa composed correspondence during the period of Portuguese control, presumably mostly in Arabic, but also possibly in Portuguese, given that there were Swahili in the employ of the Portuguese who translated from Swahili to Portuguese.12 According to Randall Pouwels citing Olfert Dapper, the King of Malindi employed scribes who managed his correspondence in Arabic.13 After the destruction of Mombasa in 1505 CE, the King of Mombasa wrote a letter to the King of Malindi describing in vivid detail what had happened in Mombasa. 14 We do not know whether the King wrote the letter in Arabic or in Swahili, but the letter is evidence that the rulers of these city-states communicated with each other in writing in the early sixteenth century. Jeremy Prestholdt also cites letters written by the kings of Malindi to the King of Portugal,15and Liazzat Bonate mentions these as well as ‘a letter from the ruler of Mozambique [Mozambique Island], Sharif Muhammad al-Alawi, to the king Manuel dated May 27, 1517’.16 Correspondence continued throughout the sixteenth century. According to Thomas Vernet, In 1596 or 1597 Sultan Muhammad, king of Malindi and Mombasa, wrote to Philip II of Spain to ask him, among other things, for his ships to sail freely and tax free to the fortresses of the Portuguese Empire. He also asked [for] the right to send a ship of pilgrims to Mecca (i.e. Jeddah), presumably each year, on the basis that this right had already been granted by the Portuguese Crown to the Muslim rulers of India.17 12 See Alpers 1975, 98–99. 13 Pouwels 2002, 397. 14 The letter read: ‘May God protect you Sayyid Ali. I have to inform you that we have been visited by a mighty ruler who has brought fire and destruction amongst us. He raged in our town with such might and terror that no one, neither man nor woman, neither the old nor the young, nor even the children, however small was spared to live. His wrath was to be escaped only by flight. Not only people, but even the birds in the heavens were killed and burnt. The stench from the corpses is so overpowering that I dare not enter the town, and I cannot begin to give you an idea of the immense amount of booty which they took from the town. Pray hearken to the news of these sad events, that you yourself may be preserved’, Strandes 1968, 64. Strandes cites Mayr fol. 10a ff.; Mayr was a German representative who traveled with Francisco de Almeida. A Portuguese translation of his account was included in a manuscript Strandes found in Munich (Strandes 1968, 56, n. 7). Digby 1975, 54; see de Sousa 1788 as well. 19 Jones 1993, 477. 20 Burke III 2009, 176–177. 21 See Saad 1979, 177–207 and Delmas 2017, 181–206. 22 Perkins, 2015. 23 Delmas, 2017, 194. Presumably he is referring to manuscripts. 24 Sheriff 2009, 181. Bonate 2016, 65. 25 Sheriff 2009, 181. Vernet 2015, 167. 26 Personal communication. Prestholdt 2001, 397. Presthold cites Sousa (1790, 123–125) and Silva da Rego (1962–1989 [6], 44–46). 17 Since there is no evidence to the contrary, it seems reasonable to assume that this correspondence was on paper, although there is no documentation of East African sources of paper until the eighteenth century.18 We do know that paper from Europe was available in Malaysia by the early part of the sixteenth century19 and that paper-making technology and paper itself seem to have diffused with the spread of Islam.20 Early evidence of letter exchanges makes it seem likely that paper was in use in East Africa when the Portuguese arrived and may even have been in use there much earlier. Unfortunately, to the best of our knowledge, aside from these letters and the Portuguese and Arabic versions of the Kilwa Chronicle, there are no surviving manuscripts from this period or until nearly two centuries later.21 Our evidence of writing in East Africa during this two-century gap is found in the extensive stone and wood carvings still extant along the East African coast. The question in this essay is whether consideration of these carvings might contribute to our understanding of later Swahili manuscripts, and how this might be done. Arabic is used in most inscriptions on tombs and in mosques in Eastern Africa and is the only language employed in quotations from the Qur'ān on these structures. Arabic was also used on coins minted in East Africa even earlier.22 Delmas claims that ‘[t]he earliest Swahili ‘ajamī dates to the seventeenth century’,23 yet Swahili occurs quite often in epitaphs that were written well before that. The first Swahili inscription on stone may actually be on the mihrab of the mosque in Kizimkazi on Zanzibar. This 500 AH / 1107 CE inscription refers to Sayyid Abi Imran Mfahamu al-Hasan b. Muhammad as the builder of the mosque.24 Abdul Sheriff comments that ‘the middle title Mfahamu’ is probably ‘among the earliest evidence of a Swahili term for ruler, mfaume/mfalme’.25 Professor Sheriff also observes that this ‘may be the first inscription of a Swahili word on stone’.26 Aldrick has observed that the inscription on this 18 15 16 BIERSTEKER | WRITING IN SWAHILI manuscript cultures mc NO 17 15 BIERSTEKER | WRITING IN SWAHILI tomb ‘has foliate scroll work within the calligraphy’ and that this ‘is an exception and is quite different from the later tomb inscriptions found in East Africa.’27 According to G. S. P. Freeman-Grenville and B. G. Martin, At Kunduchi, a little north of Dar es Salaam, the tomb of Sultan Shaf la-Hajji, dated a.h. 1081/a.d. 1670–1, in giving the name of his father Mwinyi Mtumaini, is the first wholly certain record of Swahili words that we possess [...].28 They recorded many earlier Swahili examples themselves, though, such as ‘Sayyida 'A'isha bint Mawlana’ on a 1360 or 1550 tomb in Kilwa, ‘Mwana wa Bwana’ on a tomb dated to 1462 near Kilindini, and ‘Mwinyi Shummua [sic] b. Mwinyi Shomari’ on a 1664 tomb in Kwale. Numerous additional epitaphs on later tombs found in East Africa include Swahili honorific titles and other Swahili terms. Those recorded by Freeman-Grenville and Martin (F&M) are as shown in Table 1. It is significant that many of these honorific titles are Swahili phrases (Bwana Mkuu, Bwana She, Nana She, Mwinyi Mui, Mwinyi Mkuu, Mwinyi Mtumaini, Mtu Mkuu, Mkuu Mtumaini). The epitaphs also include names that are Swahili words, such as Simba [‘Lion’], Mauwa/Mawa [‘Flowers’?], and Mwinyi Sawa29 [‘The Fair Ruler’]. The use of ‘la’ as a possessive in descriptions of lineages is of linguistic interest; there is one example of a ‘wa’ possessive (Mwana wa Bwana binti Mwidani), but ‘la’ is used in the following cases: Sometimes the ‘la’ precedes a name (la-Muhammad) and at other times a Swahili phrase (la Mtu Mkuu) or a Swahilicized version of an Arabic term, e.g. ‘Haji’. This use of ‘la’ probably has its source in augmentative usages of Class 5 and refers to the status of the person to whom reference is made. It is noteworthy that these ‘la’ examples are all found on tombs located in Kunduchi or elsewhere on the Mrima Coast or on Zanzibar or Pemba, while the only ‘wa’ example is from Kilindini on Mombasa Island. Many of the usages are associated with the clan or family name ‘Shonvi’ and/or with ‘al-Hatimi’ or ‘al-Barawi’, all of which are associated with people who came to the Mrima Coast from Yemen via Brava.30 The widespread and long-standing use of both ‘mwinyi’ and ‘mwana’ is also of significance, as is the use of ‘maulana’. Some of the epitaphs are almost entirely in Swahili, such as Sultan Diwan La Simba b. Mwinyi 'Amiri31 and Mwana Mauwa bint Mwinyi Sawa Kal(?) b. Mwinyi Hajj. Most are in a mixture of Swahili and Arabic, but Swahili readers still understand them. Examples of Swahili/Arabic code-mixing and -shifting clearly have long and extensive histories in East Africa. The epitaphs also make it clear that many of those memorialized, including a number of women, had made the hajj (Table 2). 1. Malwana bint Sultan Shaf la-Mtu Mkuu b. Sultan Shaf Muhammad al Barawi 2. Sultan Shaf la-Hajji b. Sultan Shaf la-Muhammad b. Sultan Mwinyi Mtumaini 3. Mwinyi Mkuu b. Mwinyi Kab b. Mwinyi Khamis Kab b. al-Sultan Shonvi la-Muhammad 4. Sultan Diwan La Simba b. Mwinyi 'Amiri 5. Mas'ud b. Sultan Shonvi 'Ali b. Sultan Shonvi la-Muhammad al-Barawi 6. Mwinyi Mkuu Mwinyi Abu Bakr b. Lamta ... Mkuu ... Sultan Shonvi la-Mtu Mkuu al-Hatimi al-Barawi 27 Aldrick 1988, 38. 28 Freeman-Grenville and Martin 1973, 99. 30 Pouwels 2002, 419. This might also be ‘Sawakal’, the meaning of which is uncertain. 31 Freeman-Grenville and Martin 1973. 29 mc NO 17 manuscript cultures 16 BIERSTEKER | WRITING IN SWAHILI Table 1: Swahili terminology in tomb inscriptions. Epitaph Translation Location Date Source ‘Hajj 'Umar b. Bwana Mkuu b. Bwana b. Bwana 'Umar b. Bwana Mkuu b. Bwana Daya (locally pronounced Dayo) b. Bwana She [sic] Hajji’ Chovai, Somalia n.d. F&M,10832 ‘Sultan Bwana’ (on a mosque) Pate n.d. F&M, 109 Sultan Abu Bakr b. Sultan Bwana Mkuu (?) (on a mosque) Pate n.d. F&M, 109 Nana She of Jum'a (?) b. Shaykh 'Idi b. Hamad b. Shaykh Ahmad b. Shaykh Husayn b. Shaykh Abu Bakr b. Salim Pate n.d. F&M, 109 Malwana bint Sultan Shaf la-Mtu Mkuu b. Sultan Shaf Muhammad al Barawi Kunduchi n.d. F&M, 118 Sultan Mwinyi Mtumaini b. Mwinyi Jum'a b. Ka … b. Sultan al-Shonvi laMuhammad al-Barawi Kunduchi date erased F&M, 118 Sayyid Abi Imran Mfahamu al-Hasan b. Muhammad Kizimkazi 1107 Sheriff 33 Sayyida 'A'isha bint Mawlana Amir 'Ali b. Mawlana Sultan Sulayman Kilwa Kisiwani 1359–60 or 1553–54 F&M, 120 Mwana wa Bwana binti Mwidani Tuaka near Kilindini 1462 F&M, 114 Mwinyi Shummua [sic] b. Mwinyi Shomari Kwale 1663–64 or 1664–65 F&M, 120 Sultan Shaf la-Hajji b. Sultan Shaf la-Muhammad b. Sultan Mwinyi Mtumaini Kunduchi 1670–71 F&M, 118 Mwana (Miyan) b. Mwana Madi b. Mwana Sa'id al-Malindani Kilwa Kisiwani 1712 F&M, 120 Mwinyi Mkuu b. Mwinyi Kab b. Mwinyi Khamis Kab b. al-Sultan Shonvi la-Muhammad Ukutani, Pemba 1712–13, 1825–26, or 1845–46 F&M, 118 Hajjiya Bibi bint Muqaddam Hajj Muhammad Mogadishu 1726 F&M, 104 Sultan Mwinyi Mtumaini b. Sultan Shaf la-Muhammad al-Barawi Kunduchi 1748–49 F&M, 118 Mwana Siti bint Diwan Hasan (on a mosque) Wasin 1749 F&M, 115 32 References are to Freeman-Grenville and Martin 1973, 98–122 33 Reference is to Sheriff 2009, 181. manuscript cultures mc NO 17 17 BIERSTEKER | WRITING IN SWAHILI Shaykh b. Mwinyi Mui al-Khazraji al-Ba Urii [Bauri], name of the carver of the epitaph above Wasin 1749 F&M, 115 Sultan Shonvi la-Hajji Abu Bakr Kunduchi 1754 F&M, 118 Mwinyi Saqafu b. Diwan Ruga Vumba 1786–87 F&M, 115 Sultan Diwan Nungu (?) b. al-Walid Mwinyi 'Amiri b. Sultan Diwan La Simba (on a mosque) Kipumbwe Mji Mkuu 1789 F&M, 116 Sultan Diwan La Simba b. Mwinyi 'Amiri Kipumbwe Mji Mkuu 1789–90 or 1823–24 F&M, 116 Mas'ud b. Sultan Shonvi 'Ali b. Sultan Shonvi la-Muhammad al-Barawi Mbweni, Zanzibar 1791–92 F&M, 117 Mahalimu [Mwalimu] Abu Bakr bin Mkandi Pate 1792 O&W Mwana Majid bint Ma'adi b. Wabi al-Kindiya Mombasa? 1793 F&M, 114 Mwana Jub bint Mabwana Hajj [sic] 'Umar b. Shaykh b. Da'ud al-Mustafi Provenance not known 1799 F&M, 119 Tammam (? Tamim) bint Mwinyi Jum'a Mbweni 1802–03 F&M, 117 Sultan Muhammad b. Abu Bakr b. Sultan Bwana Mkuu al-Nabhani al Barawi Pate 1809 or 1616 or 1624 O&W F&M, 108 Mwana Mauwa bint Mwinyi Sawa Kal(?) b. Mwinyi Hajj Mbweni 1813 or 1832–33 F&M, 117 Bwana Musa Kidonge, Pemba 1817–18 F&M, 121 Mwana Furno (?), the daughter of Sultan Muhammad bin Sultan Pate 1824 O&W34 Mawa35 Ghamr bint al-Sultan Mwinyi Mtumaini al-Barawi Kunduchi 1832–33 F&M, 118 Mwinyi Mkuu . . . Mkuu Mtumaini Kab b. Sultan Shonvi la-Muhammad al-Hatimi al-Barawi Kunduchi or Mrima Coast 1838–39 F&M, 119 Mwinyi Mkuu Mwinyi Abu Bakr b. Lamta ... Mkuu . . . Sultan Shonvi la-Mtu Mkuu al-Hatimi al-Barawi Perhaps Kunduchi or Mrima Coast 1843 F&M, 119 34 Reference is to Wilson and Omar 1997. 35 It is uncertain if this is a proper name or a shortening of ‘mwana wa’. mc NO 17 manuscript cultures 18 BIERSTEKER | WRITING IN SWAHILI Table 2: Swahili terminology in tomb inscriptions containing the honorific title Hajj. Epitaph Translation Location Date Source 1. This . . . 2. for the imam of God 3. God ….Hajji4….to all… 5. made it for him….6. [May] God most High love him! Mogadishu n.d. F&M, 102 Hajji Mahmud (plate, not tomb) Mogadishu n.d. F&M, 105 Hajj 'Umar b. Bwana Mkuu b. Bwana b. Bwana 'Umar b. Bwana Mkuu b. Bwana Daya […] b. Bwana She Chovai, Somalia n.d. F&M, 108 Hajj Shanid b. Abi Bakr 'Umar b. 'Uthman b. Husayn (on a mosque) Barawa (Brava) 1105 F&M, 108 'Umar b. al-Hajj Sharif b. Abi Bakr b. Hajj Da'ud Mogadishu 1269? F&M, 103 al-Hajj Sa'id al-Khazraji Mogadishu 13th c. F&M, 103 Abu Bakr b. al-Hajj Yaqut al-Hadrami Mogadishu 1358 F&M, 104 Abu Bakr b. Muham mad b. Hajj Ahmad al-Madani Mogadishu 1364 F&M, 104 Hajj Yusuf b. Abu Bakr b. Hajj Da'ud Mogadishu 1365 F&M, 104 Shaykh al-Hajj . . . . b. al-Shaykh 'Uthman b. Isma'il Mogadishu 1368–9 F&M, 106 Abu al-DIn [sic] b. al-Faqih al Hajj al-Qahtani Mogadishu 1607 F&M, 104 Sultan Shaf la-Hajji b. Sultan Shaf la-Muhammad b. Sultan Mwinyi Mtumaini Kunduchi 1670–71 F&M, 118 Hajjiya Bibi bint Muqaddam Hajj Muhammad Mogadishu 1726 F&M, 104 al-Hajj 'Ali b. Ahmad al-Sadiq al-Bakri (on a door, not a tomb) Mogadishu 1738 F&M, 107 Sultan Shonvi la-Hajji Abu Bakr Kunduchi 1754 F&M, 118 Hajj 'Umar al-Thani Ibrahim (plate, not a tomb) Mogadishu 1778 F&M, 105 Dedicatory inscription in Arabic in the Shikely Mosque, stating that it was built at the time of Ahmad b. Muhammad al-Mazru'i's pilgrimage to Mecca Mombasa 1793–4 F&M, 118 Inscription in black ink on the wall of the Mazru'i Audience Chamber in Fort Jesus, recording the pilgrimage of Ahmad b. Muhammad al-Mazru'i to Mecca in 1208 AH/793–4 CE. Mombasa 1793–4 F&M, 113 Mwana Jub bint Mabwana Hajj [sic] 'Umar b. Shaykh Unknown, but in the National Museum, Dar es Salaam 1799 F&M, 119 Mwana Mauwa bint Mwinyi Sawa Kal(?) b. Mwinyi Hajj, Dar es Salaam 1813 or 1832 F&M, 117 Hajja bint Hajj Shah Kunduchi 1847–8 F&M, 118 Hajj 'Abdallah al-Balushi Kaole 1848 F&M, 117 manuscript cultures mc NO 17 19 BIERSTEKER | WRITING IN SWAHILI It seems likely from the epitaphs that many people made the hajj in family groups that probably included servants or slaves. We can safely assume that those memorialized for having made the journey were only a small percentage of those who actually undertook the arduous trip. Because their journey by ship was dependent on the monsoon winds, this would have meant them having to spend months on the Arabian Peninsula where they no doubt met up with other pilgrims from elsewhere in the Islamic world. Citing Jomier, Russell King states that 15,000 people from Africa south of the Sahara made the hajj in 1324 CE. King also indicates that a major route to Mecca from East Africa was by dhow to the coast of Yemen and then via Sanaa through the highlands to Mecca.36 Kenneth McPherson has suggested that most of those who made the hajj from Indian Ocean regions ‘carried trade goods to help defray the heavy expenses of the pilgrimage’.37 Some of these people may have engaged in study and have copied or purchased manuscripts during their journey. In his thesis, Usam Isa Ghaidan mentions a qadhi from Lamu, for example, who is commented upon in the work of the Arab historian Al-Maqrizi:38 the Lamu qadhi who visited Mecca in 1441 CE ‘impressed the Arab historian by his scholarship.’39 There is only one case we know of in which the name of the person who carved an inscription has been preserved: the author of an inscription made on a mosque in Wasin in 1749 CE was ‘Shaykh b. Mwinyi Mui al-Khazraji al-Ba Urii (??)’.40 We can assume at least in this case that the carver was someone with religious knowledge and the son of a ruler, but we cannot generalize beyond this except to say that his background seems quite similar to that of another much later carver about whom we know a great deal: Muhamadi Kijuma. Kijuma, who was a poet, musician, calligrapher, scribe, and carver, was from an elite ruling family, studied Islam with leading scholars, and made the hajj at least four times.41 As Clarissa Vierke has observed, ‘[t]he Swahili term nakshi refers to all kinds of ornaments, no matter whether inscribed in wood, modelled in plaster or drawn on paper’42 (or carved in stone, as in this case). She also notes the resemblance between Kijuma’s design work on different surfaces. Presumably, this was also the case for earlier artists. This little piece of information raises some intriguing questions about who the carvers were and how they were trained, which materials were used, and what kind of relations existed between carvers, scribes, and authors. Freeman-Grenville and Martin have observed that ‘[i]t is noteworthy that no dated inscriptions are known in Mogadishu between 1365 and 1607.’ There are also significant gaps in other places, and the majority of the inscriptions in many sites have become unreadable. Freeman-Grenville and Martin saw their ‘handlist’ ‘only as a beginning’ and encouraged readers to contribute to future publications so as to create a corpus of ‘Islamic epigraphy of the coast’.43 This is a project that has yet to be undertaken, but it continues to be a worthy goal. They recorded 249 inscriptions in all. One hundred and two of those are illegible, partially illegible, or not yet deciphered. Thirty-three of those that are readable contain Swahili terms. Most of the inscriptions on tombs and mosques were on coral rag, but at least one inscription was on sandstone. Marble was also imported from India to Mogadishu and Kilwa. Elizabeth Lambourn suggests that inscriptions on marble found in East Africa were commissioned and argues that it is likely artisans traveled from Cambay in India to Mogadishu and Kilwa to carry out this work because of the differences between carving on coral rag and carving on marble.44 Of course, it is also possible that artisans traveled from Mogadishu and Kilwa and learned to carve on marble in Cambay, or that texts were sent from East Africa to Cambay to be carved on marble slabs. Two woodworked minbars [steps used as a platform in a mosque] with inscriptions from the early sixteenth century have also survived: one in Siu dated 1523–1524 and one in Lamu dated 1511–1512.45 Both include decorative carving in addition to inscriptions in Arabic. Aldrick describes the carving on these doors as being in an ‘interlaced strapwork pattern called guilloche, which is seen particularly in Ethiopian and Somali art’.46 In addition, Freeman-Grenville and Martin refer to an ‘unread inscription on an ivory horn 36 King 1972, 65. 37 McPherson 1988, 50. 38 Al-Mas'udi 1964. 43 Freeman-Grenville and Martin 1973, 98. Lambourn 1999, 61–86. 39 Ghaidan 1974, 46. 44 40 Possibly ‘al-Bauri’. 45 41 Vierke 2010, 41–42 42 Vierke 2010, 44. mc NO 17 Freeman-Grenville and Martin 1973, 105, 108, 109; photograph of the Siu minbar inscription. Also see Aldrick 1988. 46 Aldrick 1988, 37. manuscript cultures 20 BIERSTEKER | WRITING IN SWAHILI Fig. 1: ‘Heading Sura 36’ taken from Digby 1975, eighteenth-century Qur'ān from Witu. or siwa, made in Pate in A.D. 1696’.47 Presumably, the differences were not as great in carving on coral rag and carving on wood or on ivory as the differences between carving on these surfaces and marble, but were there specialists in particular materials? And did many of those who carved on these surfaces also copy or write manuscript texts? In his description of an eighteenth-century Qur'ān from Witu, Simon Digby observed the following (Fig. 1): On the central field of each rectangular panel, reserved against a black background with rounded corners, is set the name of the Sura in a compressed thuluth-like hand with some cursive contractions, a hand which recalls that of stone epigraphs. the guard-strip is decorated in reserve against black with lacing rope ornament, in the second, also in reserve against black with a chain of lozenges.48 The similarities between tomb inscription designs and manuscript design are also evident in the design of the fourteenth-century tomb near the Mosque by the Sea in the Jumba la Mtwana ruins (Fig. 2),49 in the design of the sixteenthcentury Siu minbar (Fig. 3),50 and in the design of the epitaph of Mwana wa Bwana binti Mwidani, who died in 1462 CE, on her tomb in Tuaka near Kilindini (Fig. 4).51 The inscription design on each of these tombs consists of a border design surrounding a text, which is remarkably similar to that described by Digby. At the rounded corners red-coloured spandrels complete the rectangle. The rectangular field is enclosed within a guard strip with red ruled lines on either side. In the first heading 47 Freeman-Grenville and Martin 1973, 110. manuscript cultures 48 Digby 1975, 52. 49 Jumba la Mtwana Guide n.d., 9. 50 Freeman-Grenville and Martin 1973, plate 1. 51 Freeman-Grenville and Martin 1973, plate 5. mc NO 17 21 BIERSTEKER | WRITING IN SWAHILI Fig: 2: Jumba la Mtwana mosque inscription. The Jumba la Mtwana Guide (n.d., 6) gives the following description: ‘The inscription itself consists of a finely carved slab of coral. The surrounding text is from the Koran, Chapter 3, verse 185 and may be translated: “Every soul shall taste death. You will simply be paid your wages in full on the Day of Resurrection. He who is removed from the fire and made to enter heaven, he it is who has won the victory. The earthly life is only delusion.”’ mc NO 17 manuscript cultures 22 BIERSTEKER | WRITING IN SWAHILI Fig. 3: The Siu Minbar; Freeman-Grenville and Martin 1973, plate 1. ‘Inscription on the wooden minbar in the Friday Mosque A.H. 930/15231524 CE. This is the earliest known inscription on wood in Eastern Africa’. manuscript cultures mc NO 17 23 BIERSTEKER | WRITING IN SWAHILI Fig. 4: Epitaph of Mwana wa Bwana binti Mwidani, Freeman-Grenville and Martin 1973, plate 5. ‘d. 29 Rajab, 866 A.H. / 29 April 1462 CE, from Tuaka, near Kilindini, Mombasa Island’. mc NO 17 manuscript cultures 24 BIERSTEKER | WRITING IN SWAHILI Fig. 5: Diwān al-'Adanī [The Collected Poetry of Abū Bakr al-'Adanī], copied by Sālim b. Yusallim b. 'Awaḍ Bā Safar for Habib Saleh in 1927 CE (EAP466/1/19, image 2). These tomb designs look remarkably like pages, although I have not seen any similar border designs on Swahili manuscripts. The closest design of which I am aware is the twentieth-century border of a signboard that Muhamadi Kijuma made for Ernst Dammann.52 Another similar border design is found in an early poetry manuscript in the Riyadha Mosque collection (Fig. 5). There is certainly a need to study the styles of carved writing in a similar way to carving styles.53 Are there any regional differences? Did expert carvers travel as we know later scribes and translators did? To what extent were the styles used in epitaphs and inscriptions consistent across Swahili-speaking areas? Comparisons of the epitaphs found on tombs from the eleventh to mid-nineteenth century with nineteenth- and twentieth-century Swahili names recorded as authors’ names in manuscripts and in legal and waqf [inheritance dedication] documents suggest that the use of Swahili in such formal naming contexts may have declined rather than expanded during this more recent period. For example, Anne K. Bang records only one author, scribe, or waqf name containing a Swahili term (‘Qadi Hisham b. Abi Bakr b. Bwana Kwab [or Kub] al-Lami’) in her provisional catalogue entries for the 43 nineteenth- and twentieth-century manuscripts discovered in the Riyadha Mosque to which she refers in her article.54 The waqf documents examined by John O. Hunwick and Rex Sean O’Fahey55 (which are also from Lamu) contain more Swahili terms, but none of the ‘wa-’ or ‘la-’ constructions found on epitaphs on tombs further south. The names that include Swahili terms are the following: 52 54 Bang 2015, 164-172. 55 Hunwick and O’Fahey. 2002, 1–19. 53 Miehe and Vierke 2010, 532. Athman 1996, 11–29. Also see Aldrick 1988. manuscript cultures mc NO 17 25 BIERSTEKER | WRITING IN SWAHILI Amina bint bani/a [= bwana?] (p. 10) Bwana Mkubwa b. al-Shaykh b. Abi Bakr al-Khatimi (pp. 11, 13) Mwana Amina bint bwana Sacd (p. 15, 17) Bwana Sacd b. bwana cAliyu (p. 15) Bwana Sacd b. bwana Ubu (p. 17) Yayi bint bwana cAliyu (p. 17). In his recent study of nineteenth-century contracts made in Zanzibar, Thomas F. McDow also records a number of names that include Swahili terms that were used in contracts or other official contexts: Diwan Makambi (or Mwekambi) Juma bin Ahmad (p. 52) Binti Bana (Bwana) Waziri al-Mufazii (p. 52) Bwana Heri bin Juma al-Mafazii (p. 52) Mwinyi Kidogo (p. 99) Mwinyi Kheri (p. 141). Again, these are honorific titles and there is no other evidence of Swahili usage. Writing on other surfaces provides even less evidence. While ninth- to fourteenth-century Swahili coins were unique in that writing on the coins ran from right to left around the coins and from front to back, the writing was always in Arabic; there is no evidence of Swahili names, titles, or grammatical features on Swahili coins56 or on other types of artifacts until the late nineteenth and early twentieth century.57 Conclusion Epitaphs on tombs seem to provide evidence of early and continuous writing in code-mixed Arabic and Swahili and solely in Swahili along the East African coast. The evidence they provide is limited in that it consists primarily of Swahili honorific titles and proper names that are also Swahili words. The strongest evidence of early Swahili writing is the use of the possessive forms ‘la-’ and ‘wa-’ as this is evidence of use of Swahili grammatical structures, but these seem to be associated with honorific titles. There are clearly some regional differences in the extent to which Swahili forms are used in epitaphs since the possessive forms are only used in some areas. Further analysis may also reveal regional differences in styles and decorative features as well as differences across time periods. The epitaphs additionally indicate that at least some elite East Africans, including scholars, traveled extensively. Epitaphs of this kind challenge the assumption that texts were exclusively or primarily brought to East Africa by outsiders, because they make the extensive travel and scholarship of East Africans from 1200 onwards clear as well as their extensive contact with other Arabic speakers. Epitaphs of this type also raise questions concerning the intellectual and artistic work of carvers and the extent to which they were composers and/or commissioned to create works using specific materials. Design parallels in different materials are evident, but require additional study. 56 See Perkins 2015; Perkins, Fleisher, and Wynne-Jones 2014, 102–116; Wynne-Jones and Fleisher 2012. 57 See Biersteker 2018. mc NO 17 manuscript cultures 26 BIERSTEKER | WRITING IN SWAHILI REFERENCES Al-Mas'udi (1938), Muruj-ul-Dhabab wa Ma'adin-ulJawhar (Cairo: Mahmud Hilmi). Aldrick, Judith Sophia (1988), The Nineteenth Century Carved doors of Mombasa and the East African Coast in two volumes, Master’s thesis, Durham University: Durham E-Theses vol. 1, 39 <http://etheses.dur.ac.uk/6619/>. Alpers, Edward A. (1975), Ivory & Slaves in East Central Africa, Berkeley: University of California Press. Athman, Athman H. (1996), ‘Styles of Swahili Carving’, Afrikanistische Arbeitspapiere (AAP), 47: 11–29. Bang, Anne K. (2011), ‘Authority and Piety, Writing and Print: A Preliminary Study of the Circulation of Islamic Texts in Late Nineteenth and Early Twentieth Century Zanzibar’, Africa, 81/I: 89–107. —— (2015), ‘Localising Islamic Knowledge: Acquisition and Copying of the Riyadha Mosque Collection in Lamu, Kenya’, in Maja Kominko (ed.), From Dust to Digital: Ten Years of the Endangered Archives Programme, Cambridge: Open Book Publishers, 135–172. Biersteker, Ann (2018), ‘The Inscribed Object: The Textures and Textuality of Writing in Eastern Africa’, in Prita S. Meier and Allyson Purpura (eds), World on the Horizon: Swahili Arts Across the Indian Ocean, Champaign, IL: University of Illinois Press, 310–336. Bloom, Jonathan M (2001), Paper Before Print: The History and Impact of Paper in the Islamic World, New Haven, CT: Yale University Press. Bonate, Liazzat J. K. (2016), ‘Islam and Literacy in Northern Mozambique: Historical Records on the Secular Uses of the Arabic Script’, Islamic Africa, 7: 60–80. Burke III, Edmund (2009), ‘Islam at the Center: Technological Complexes and the Roots of Modernity’, Journal of World History, 20/2: 165–186. manuscript cultures Camões, Luis Vaz de (1997), Os Lusíads/The Lusiads, translated by Landeg White, Oxford: Oxford University Press. De Sá João or Alvaro Velho (1898), A Journal of the First Voyage of Vasco da Gama 1497–1499, translated by E. G. Ravenstein, London: The Hakluyt Society <https:// archive.org/stream/worksissuedbyha00unkngoog/ worksissuedbyha00unkngoog_djvu.txt>. De Sousa, João (1790), Documentos Arabicos para Historia Portugueza copiados dos originaes da Torre do Tombo, Lisbon: Officina Da Academia Real Das Sciencias. Delmas, Adrien (2017), ‘Writing in Africa: The Kilwa Chronicle and other Sixteenth-Century Portuguese Testimonies’, in Andrea Brigaglia and Mauro Nobili (eds), The Arts and Crafts of Literacy: Islamic Manuscript Cultures in Sub-Saharan Africa, Berlin: Walter de Gruyter, 181–220. Digby, Simon (1975), ‘A Qur’an from the East African Coast’, Art and Archeology Research Papers (AARP), 7: 49–55. Freeman-Grenville, G. S. P. and B. G. Martin (1973), ‘A Preliminary Handlist of the Arabic Inscriptions of the Eastern African Coast’, The Journal of the Royal Asiatic Society of Great Britain and Ireland, 2: 98–122. Ghaidan, Usam Isa (1974), Lamu: Case Study of the Swahili Town, Master’s thesis, Architecture, University of Nairobi. Hunwick, John O. and R. S. O’Fahey (2002), ‘Some Waqf Documents from Lamu’, Sudanic Africa, 13: 1–19. Jones, Russell (1993), ‘European and Asian Papers in Malay Manuscripts; A Provisional Assessment’, Bijdragen tot de Taal-, Land- en Volkenkunde. Manuscripts of Indonesia, 149/3: 474–502. Jumba la Mtwana Guide (1981), Nairobi: Kenya Museum Society. mc NO 17 BIERSTEKER | WRITING IN SWAHILI King, Russell (1972), ‘The Pilgrimage to Mecca: Some Geographical and Historical Aspects’, Erdkunde, 26/1: 61–73. Lambourn, E. (1999), ‘The Decoration of the Fakhr alDīn Mosque in Mogadishu and Other Pieces of Gujarati Marble Carving on the East African Coast’, Azania: Archaeological Research in Africa, 34/1: 61–86. Lynn, Ashley Ridgway (2005), Early Written Kiswahili: History and Translation of Kiswahili in Arabic Script, Master’s thesis, African Studies, Yale University. McPherson, Kenneth (1988), ‘Maritime Passenger Traffic in the Indian Ocean Region before the Nineteenth Century’, The Great Circle, 10/1: 49–61. Miehe, Gudrun and Clarissa Vierke (eds) (2010), Muhamadi Kijuma: Texts from the Dammann Papers and Other Collections, Cologne: Rüdiger Köppe. Omar, Yahya Ali and P. J. L. Frank (1994), ‘A 12th/18th Century Swahili Letter from Kilwa Kisiwani (being a study of one folio from the Goa Archive)’, Afrika und Übersee, 77: 263–272. Perkins, John, Fleisher, Jeffrey, and Stephanie Wynne-Jones (2014), ‘A Deposit of Kilwa-type Coins from Songo Mnara, Tanzania’, Azania: Archaeological Research in Africa, 49/1: 102–116. —— (2015) ‘The Indian Ocean and Swahili Coast Coins, International Networks and Local Developments’, Afriques, 6 <http://journals.openedition.org/afriques/1769>. Pouwels, Randall L. (2002), ‘Eastern Africa and the Indian Ocean to 1800: Reviewing Relations in Historical Perspective’, The International Journal of African Historical Studies, 35/2 –3: 385–425. Prestholdt, Jeremy (2001), ‘Portuguese Conceptual Categories and the “Other”: Encounter on the Swahili Coast’, Journal of Asian and African Studies, 36: 383–406. mc NO 17 27 Saad, Elias (1979), ‘Kilwa Dynastic Historiography: A Critical Study’, History in Africa, 6: 177–207. Sheriff, Abdul (2009), ‘The Persian Gulf and the Swahili Coast: A History of Acculturation over the Longue Durée’, in Lawrence G. Potter (ed.), The Persian Gulf in History, New York: Palgrave Macmillan, 173–188. Silva da Rego, R. (1962–89) (ed.), Documentos sobre os Portugueses em Moçambique e na África Central, vols 1–9, Lisbon: Centro de Estudos Historicos Ultramarinos. Strandes, Justus (1968), The Portuguese Period in East Africa, translated by Jean F. Wallwork, ed. J. S. Kirkman, Nairobi: East African Literature Bureau. Sultan Fatima binti Muhammad Mkubwa (2007), ‘Peace and Security’ [Letter]. Trans. Ann Biersteker, in Amandina Lihamba, Fulata L. Moyo, M. M. Mulokozi, Naomi L. Shitemi, and Saida Yahya-Othman (eds), Women Writing Africa: The Eastern Region, New York: The Feminist Press at City University of New York, 71–72. Vernet, Thomas (2015), ‘East African Travelers and Traders in the Indian Ocean: Swahili Ships, Swahili Mobilities ca. 1500–1800’, in M. Pearson (ed.), Trade, Circulation, and Flow in the Indian Ocean World, New York: Palgrave Macmillan, 167–202. Vierke, Clarissa (2010), ‘Between the Lines: Life and Work of the Lamuan Artist and Cultural Broker Muhamadi Kijuma’, in Gudrun Miehe and Clarissa Vierke (eds), Muhamadi Kijuma: Texts from the Dammann Papers and Other Collections, Cologne: Rüdiger Köppe, 41–62. Wilson, Thomas H. and Athman Lali Omar (1997), ‘Archaeological Investigations at Pate’, Azania: Archaeological Research in Africa 32/1: 31–76. Wynne-Jones, Stephanie and Jeffrey Fleisher (2012), ‘Coins in Context: Local Economy, Value and Practice on the East African Swahili Coast’, Cambridge Archaeological Journal, 22: 19–36. manuscript cultures 28 BIERSTEKER | WRITING IN SWAHILI PICTURE CREDITS Fig. 1: © Digby 1975, 52. Fig. 2: © Photography by Sonia Marie Gloss. Fig. 3: © Freeman-Grenville and Martin 1973, 108. Fig. 4: © Freeman-Grenville and Martin 1973, 114. Fig. 5: CC BY-ND, taken from Bang 2015, 157. manuscript cultures mc NO 17 29 MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE Article Swahili Manuscripts from Northern Mozambique: Some Notes on Ajami Correspondence Letters Chapane Mutiua | Maputo, Mozambique and Hamburg, Germany 1. Introduction This study provides a historical and descriptive account of northern Mozambique ajami manuscripts and focuses on the ajami correspondence kept at the Arquivo Histórico de Moҫambique (Mozambican Historical Archive, AHM). All the manuscripts discussed in this paper are in ajami, i.e. Swahili written in Arabic script with some Portuguese, Koti and Mwani loanwords. The spread of Islam along the western coast of the Indian Ocean increased geographically and socially during the nineteenth century. Consequently, Islamic education expanded quickly to all the coastal regions of northern Mozambique and its immediate hinterland. The expansion of Qur’anic schools popularised the use of Arabic script for religious and secular affairs. The ajami correspondence of northern Mozambique in the AHM’s collection is evidence of this process. This paper aims to take a brief look at letter correspondence as part of ajami writing practice and at the demands for systematic research aiming at its preservation and at the diffusion of knowledge thereof. My paper is based on archival and fieldwork research and will help readers to understand the socio-historical value of northern Mozambican Swahili manuscripts as a significant component of the Swahili-Islamic heritage. 2. Swahili and ajami in northern Mozambique: some background information Ajami literacy in northern Mozambique must be considered in a local context as well as from a broader perspective. It is viewed as a result of a long-lasting process involving cultural, political and economic dynamics that integrated the region into a much larger cultural entity along the east coast of Africa demarcated by the Indian Ocean. The spread of ajami literacy followed the paths of the expansion of Islam and the Indian Ocean trade networks that had been in place ever since the first millennium CE. From the very beginning, city-states – not only in what is now Kenya and Tanzania, but along the Mozambican coast as well – were involved in the Indian Ocean network and its exchange of goods, mc NO 17 knowledge and texts as well as in networks along the East African coast. Apart from maritime trade, the Mozambican city-states shared a number of cultural features with the rest of the East African coast like Islam. They also shared coral stone architecture, which archaeological sites still speak of, as well as language: languages of the city-states like Mwani and Koti, which are mentioned in this article, have such a high number of Swahili features that Swahili was obviously not just used as a vernacular language all along the coast, but can even be considered an ancestor of these languages.1 Later they became highly influenced by other Bantu languages from the hinterland such as Makonde and Makhuwa, so the Swahili linguistic features lost their prominence; the affinity to hinterland Bantu languages like Makhuwa, Makonde and Yao, which were feasible in terms of their linguistic structure and lexicon, became stronger. For Schadeberg and Mucanheia, although these languages may sometimes be considered Maka (from Manga, i.e. related to the Arabian Peninsula) due to the huge number of Swahili and Arabic words and some phonetic and grammatical features of Swahili, they have more ties to Makhuwa in the case of Koti, and Makonde and Makhuwa in the case of Mwani and Makwe.2 The linguistic change mirrors the disruption that took place from the fifteenth and sixteenth century onwards when the Portuguese usurped the gold trade and its entrepot Sofala and destroyed Kilwa, the major cultural and economic centre for the whole of what is now the Tanzanian and Mozambican coast. For some time, Mozambican city-states became rather isolated from the rest of the East African coast and Swahili networks continued to play a lesser role in comparison to relations with the hinterland. However, thanks to the socio-political and cultural revolution brought on by the establishment of Zanzibar as the seat of Busaid Omani rule in the 1830s, the usage and 1 Schadeberg and Mucanheia 2000. 2 Schadeberg and Mucanheia 2000, 1–7; Prata 1983. manuscript cultures 30 diffusion of Kiswahili increased again in nineteenth-century northern Mozambique, as did the use of Arabic script. Besides extending their political influence to most of the western Indian Ocean, the Busaid rulers of Zanzibar succeeded in attracting intellectuals from everywhere in the Muslim world, which transformed the island into the main intellectual and cultural centre in the region. The social impact of this cultural revolution was the introduction of the term ustaarabu in the Swahili societies, meaning ‘assimilation to the Arab way of life’. Among other things, this included the popularisation of Arabic literacy, a phenomenon that had started several centuries before that.3 The northern Mozambique Swahili states which also came under the political, economic and cultural influence of Zanzibar benefited from the acceleration of Swahili ajami literacy through a scholarly network linking Zanzibar, Comoro and (north-)western Madagascar.4 At that time, Kiswahili was spoken widely as a lingua franca by the political and trade elite of the northern Mozambican coast. According to Eugenius Rzewuski, the process of migration combined with a practice of intermarriage caused a linguistic contact zone to be created that was marked by bilingualism and multilingualism.5 Thus, in some areas, like Palma and Quionga, women spoke Makwe in their household, while men spoke Swahili in the business forum and public sphere.6 While this situation may not be extrapolated to southern regions of the country, the influence of Kiswahili in the existing written evidence suggests that it was used widely in areas where it had never been the first language of household communication. 3. The AHM’s collection of ajami correspondence As a further result of Zanzibari influence, Swahili literature and the use of Swahili ajami correspondence became widely disseminated in the coastal societies of northern Mozambique. In fact, they became so widespread and such an efficient means of communication that even the Portuguese colonial government decided to adopt them: rather than Portuguese or Arabic, Swahili became the official language of written correspondence in northern Mozambique. While Swahili was also partly employed by the German and British colonial governments in Kenya and Tanganyika, 3 Pouwels 1987, 3; Eastman 1994; Declich 2001, 47; Khamis 2001, 18; Mutiua 2015, 206. 4 Vilhena 1905, 19, Hafkin 1973. 5 Rzewuski 1991, 270–271. 6 Rzewuski 1991, 271–272. manuscript cultures MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE the Mozambican case is particular since even at that time, Swahili was not a first language to the people living along the coast, but rather a second or third language used in restricted contexts involving transregional communication. It was not only in trade relations, but in transregional networks of Islamic learning that Swahili played an important role as a language of instruction – on Zanzibar as well as the Comoros, to name just two prominent centres. The colonial government did not initiate the practice of letter-writing in Swahili, but rather made used of an established written register for its own purposes. The origin of correspondence in Arabic script in the public sphere in East Africa and northern Mozambique must have started before the Portuguese arrived in the region. However, the earliest Arabic correspondence that Mozambican rulers sent to the Portuguese is a letter that Sharif Muhammad el-Alawi of Mozambique wrote to King Dom Manuel of Portugal at the beginning of the sixteenth century.7 Although Kiswahili was well established as a lingua franca in the area by this time (mainly among the political and commercial elite),8 it seems that Arabic was used for correspondence by letter. A sample of this earlier Arabic correspondence between the African rulers and the Portuguese was published by Father Frei João de Sousa in 1788 CE.9 Upon Vasco da Gama’s arrival in East Africa at the end of the fifteenth century, Arabic was the main language and the script used for international correspondence in the region.10 Although there is not much evidence of the ‘moment’ when use of the Arabic language shifted to Kiswahili, later correspondence 11 reveals the important role of Kiswahili, which had become a lingua franca in the political and trading arena all over East Africa, and in northern Mozambique where it had been widely spoken since the eighteenth century,12 it became the language of official communication between all the trading and political actors, including the Portuguese. The lack of people who could speak Portuguese and read and write using the Latin alphabet transformed Kiswahili and the Arabic alphabet into the most important tools for written communication between Portuguese colonial officials and 7 Prestholdt 1998. 8 Ferreira 1976; Mutiua 2015, 208. 9 Sousa [1788], 85–86; Prestholdt 1998. 10 See João de Sousa 1789; Alpers 2000, 304. 11 See the ajami manuscript collection at the Arquivo Histórico de Moçambique. 12 See Bonate 2016, 66–67; Hafkin 1973. mc NO 17 31 MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE local rulers, at least until the end of the Campanhas de Ocupação efectiva (1895–1918), which established de facto colonial rule in Mozambique. It was during this period that the Portuguese abolished the use of Kiswahili and the Arabic alphabet in official communication. In northern Mozambique, as evidenced in the National Archives collection, the second half of the nineteenth century was the prime period for the use of ajami correspondence (in Swahili with Mwani and Portuguese loanwords) in the official realm between the Portuguese colonial officers and the African rulers. This was also an important period for the expansion of Islam and Islamic education when ‘court Islam’ was replaced by ‘popular Islam’.13 Consequently, a significant proliferation of Islamic and Swahili literature can be found in this period. The Swahili manuscript collection kept by the Mozambican Historical Archive (AHM) is part of the ‘Nineteenth-century Collection’ (Fundo do Século XIX) and is a corpus of official Swahili letter correspondence dating back to the second half of the nineteenth century (Fig. 1)14. Some of the letters are accompanied by tentative Portuguese translations, related reports and Portuguese correspondence. The Portuguese reports attached to these manuscripts are fundamental in helping us understand the meaning and historical context of the correspondence. 3.1. The correspondence: languages and scripts and the role of the lingoa do estado The nineteenth-century administrative manuscripts kept at the Mozambican Historical Archive reveal the use of numerous scripts and languages. Portuguese was the language of administration, but Kiswahili dominated official correspondence with local rulers. However, the aforementioned Mwani, the major language in coastal Cabo Delgado, was also used in the correspondence, specifically in the areas of Quissanga, Olumboa, Quirimizi, Ibo and Mocimboa. In this region, people spoke local languages such as Mwani, Koti, Sankaci and even the linguistically more distant Bantu language Makhuwa, while Kiswahili was the literary language used for correspondence, written poetry and other literary productions. 13 On the spread of Islam in East Africa, see Robinson 2004; Trimingham 1964. Regarding the spread of Islam in Mozambique, see Vilhena 1905; Hafkin 1973; Bonate 2007; Alpers 1999. 14 Source: AHM, Fundo do Século XIX, caixa (“box”), maço (“volume”) 2, Letter from Boana Shaki b. Abdulatifo al-Mafazi, Sheikh of Quissanga, 1892. mc NO 17 Sometimes, the Gujarati script and language were also used, not just the Latin and Arabic script. This indicates a specific situation in which languages and scripts were employed in a colonial system. It gave more prominence to the role of translators, or lingoas do estado, a particular position in the colonial system sometimes known as intérprete. The decision to rely on other scripts as well as Swahili was a strategic one: the Portuguese, who apparently controlled the region in this period in order to maintain constant access to the hinterland, where most of the foodstuffs and commodities came from, depended on the local rulers – people who could not read and write in Latin script or the Portuguese language. Accordingly, the easiest way of communicating by letter was by using Kiswahili written in Arabic script. In addition, hinterland trade was controlled by Indian traders, some of whom wrote their correspondence in Gujarati (both the language and script). The diversity of scripts and languages can sometimes be seen in a single manuscript. A letter written in Arabic script may be signed in Latin script, for instance, or vice versa, and letters written in Gujarati can be signed in Latin script or vice versa. The use of Kiswahili and Mwani was a natural choice for the local rulers as the latter is the language of coastal Cabo Delgado societies, and they had become familiar with the use of the former in the literary, political and trade spheres through their long-lasting integration into the western Indian Ocean networks. For the Portuguese, the use of the two languages in their offices was a pure necessity, which they were forced to adopt by the practical situation they were in, as evidenced by some of the correspondence. Momba ibn Ishaka, who was Sheikh of Quirimizi (Cabo Delgado), for instance in one of his undated letters, asks the governor of the district of Cabo Delgado not to send him letters in Portuguese, but in Kimoro (literally meaning the ‘language of the Moors’, i.e. Mwani or Kiswahili),15 as we can see from the following words found in one letter (Fig. 2): ‘aidha ukitaka kupereka ofiso [ofício] kwangu usipereke letera [letra] ya kizungu, hakuna muno ajui kufoma kizungu. Pereka hati ya kimoro’. Literally, the Sheikh informs his addressee in Kiswahili with a heavy Mwani influence: ‘if you want to send official correspondence (ofiso) to me, do not send it in the Latin alphabet because there is no-one [here] to read it. Send it in the Arabic alphabet (kimoro)’.16 15 See the letter by Momba ibn Ishaka, Sheikh of Quirimizi, in AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 10, maço 2. 16 Letter by Momba ibn Ishaka, Sheikh of Quirimizi, in AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 10, maço 2, lines 9–11. manuscript cultures 32 MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE Fig. 1a: Sample of northern Mozambican Swahili correspondence held at the Mozambican Historical Archive (AHM). Letter from Boana Shaki b. Abdulatifo al-Mafazi, Sheikh of Quissanga, 1892. AHM, Fundo do Século XIX, caixa, maço 2. manuscript cultures mc NO 17 33 MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE Fig. 1b: Sample of northern Mozambican Swahili correspondence held at the Mozambican Historical Archive (AHM). Letter from Boana Shaki b. Abdulatifo al-Mafazi, Sheikh of Quissanga, 1892. AHM, Fundo do Século XIX, caixa, maço 2. mc NO 17 manuscript cultures 34 MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE Fig. 2: Letter written by Momba ibn Ishaka, the Sheikh of Quirimizi (1880). AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 10, maço 2. manuscript cultures mc NO 17 35 MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE Fig. 3: Letter by Said ibn Musa, a lingoa do estado in Tungi. AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 14, maço 1. mc NO 17 manuscript cultures 36 MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE Table 1: Tansliteration and transcription of the letter by Said ibn Musa, a lingoa do estado in Tungi (Fig. 3). Line A1: Hua Allahu taala Him, the Great Allah17 Line A2: Al-dula al-purtugisi In the Portuguese land Line A3: Bismillah rahman Rahim In the name of Allah, the Merciful Line B1: ila janabi al-akhi siyuru kufurunadulu I wibu Mola [probably Moura] Gulushão This letter to my brother, Mr Governor of Ibo, Mola [or Moura] Gulushão [Garção]. Line B2: nakuarifu falume wangu nimetoka katika inti ya wibu I inform you, my King, I come from the land of Ibo. Line B3: anakasidi kuja kukaa barua kwa kazi kama alio nipa falume wangu I came to stay as my King ordered me in the letter of employment. Line B4: nami nikawasili barua hata kwa kumidanti wa barua akanitwaa kali [kazi] And I took the letter to the commander, who gave me the work. Line B5: sani [sana] nimekaa mwezi mmoja basi hata siku yakupokea amenambia I stayed for a long time, worked for one month until the day of payment. He told me Line B6: toka sikutaki na sababu yake sijui mwanzo hata mwishowe wake ‘Go away. I don’t need you’. I don’t know what the reason was [for the dismissal]. Line B7: nami kazi yakusoma barua nalikuwa tayari kula wakati And as for the work of reading the letters, I was ready to do it any time. Line B8: atakao nimeona kuniongoza basi nawe mfalume wetu ujue I think you must direct me – you, my King, should know this, Line B9: ya kuwa mtu uliomweka katika inti si mwema wa tamma si mwema that the person you placed in the land is no good [wa tamma], he is no good, Line B10: Ushahidi na kama huo kunifukuza mimi língua wake basi mujue the evidence is this, [is] that he wants to fire me, his [own] translator, get yourself informed Line B11: yakuwa si mutu mwema na kama mukubali ni hayo na kama hakukubali na zari that he is not a good person and if you agree and if they do not agree then Line B12: yaku mfalme wangu ojue inti etaharibika ina kula siku you my ruler know that the land is being destroyed every day. Line B13: hamushika katika majuba ya masuludadu basi ka wanawake basi The soldiers entre into our houses for the women. Line B14: Ndio kazi iliopo wasalam wakatabahu al-waraka Sa’idi ibn Musa That is the work they do. Greetings. Yours Sa’idi ibn Musa, Line B15: língua al-portugesi mimi língua nimifunguza kwa mambo basi katika I am the official interpreter of Portuguese and I inform you about the news in this land. Line B16: inti hiyi tutakaji nasi ni wana wa inti wenyewe We are the inhabitants of this land, we are the real sons of this land. 17 The text in Arabic script does not have punctuation, but I added some in the English text to make it easier to understand. manuscript cultures mc NO 17 37 MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE Another example can be found in a letter by Bwana Shaki ibn Daly, the ‘Governor of Moors’ of Quissanga, who informs the Governor of Cabo Delgado, João Lobo Teixeira de Barros, in an undated letter that he could not reply to a letter because he was unable to find anyone who could read Portuguese and translate it for him.18 Due to this situation, the Portuguese used lingoas do estado (official translators), who were local intellectuals19 who translated and wrote ajami letters for the Portuguese officers. The position of the lingoas do estado was problematic sometimes, however, and was influenced by the political context of the epoch and region, as we can see in a report by Said ibn Musa, the official translator of Tungi. In his letter dated 20 February 1888 CE (Fig. 3, Table 1), he reports that the military commander of Tungi was not paying him his wage for the position of the official translator, although Said had been recommended for the post by the governor of the district. He also informs us that this military commander, whose name is not actually mentioned in the letter, was not behaving properly towards the people of Tungi either.20 This may not be a problem that the translator only encountered, but rather part of a more complex political problem in this region, as witnessed by several correspondences from the period. The transliteration and translation of several lines of the letter shown in Figure 3 attests the complex role of the lingoas do estado (see Table 1). The letters are thus linked to the local context of Islamic literacy and spread through Islamic education, as mentioned above. Although there is no clear evidence of the presence of Qur’anic schools in the region in the letters, the historiography of the area suggests that their authors, and in some cases their scribes, attended a Qur’anic school in Mozambique or elsewhere in East Africa where Islamic education was propagated by trading ulama who came from different regions of East Africa. This point of view links the process of ajami literacy in northern Mozambique to the wider area of East Africa through Islamic scholarship, trade and political connections.21 18 See the letter by Bwana Shaki ibn Daly in AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 10, maço 2. 19 See Bonate 2016, 67 for more details about the role of the lingoa do estado. 20 See Said ibn Musa’s letter to the Governor of the District of Cabo Delgado, 20 February 1888, in AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 14, maço 1. 21 See Hafkin 1973. mc NO 17 In contrast, ajami literacy came to play an important role in the secular context of the colonial administration as part of official communication between the Portuguese officers and the local rulers. Thus, although the process which gave birth to these manuscripts is part of the dynamics of Islam’s diffusion in the area, the documents cannot be classified as religious texts. Although the letters reveal so many Islamic words and sentences, for instance in addressing people, what needs to be understood is that their purpose or use was not a religious one. The secular and administrative use of these letters marks the interface of two distinct cultural entities: the Portuguese on the one hand and the northern Mozambique Swahili rulers on the other. This twofold representation makes this correspondence different from other Swahili ajami correspondence in the region. Its peculiarity can be evidenced by several Portuguese loanwords that made their way into the introductory formulas, in the dates of the letters or even in its corpus. Thus, while the introductory formulas are common to East and Central African ajami correspondence,22 the integration of Portuguese words such as senhor (Sir or Mister), governador (governor), mando recado (I send my salutations), ordem (order), capitãomor (captain-major), comandante (commander), surdado/ soldado (soldier), receber (to receive), licença (licence), excelência (excellence) and perdoar (to forgive) reflect the specific northern Mozambican context. 3.2. Size of the collection and date range The Fundo do Século XIX of the Mozambican Historical Archive is divided into a host of regional and administrative collections and includes a total of 665 ajami manuscripts. The collection of ajami manuscripts comes from two main regions in northern Mozambique: the District of Mozambique, now Nampula Province, where the capital city (in the Ilha de Moçambique) was located during most of the period, and the District of Cabo Delgado comprising the current territories of Niassa and Cabo Delgado Province. The former District of Mozambique has 109 letters (representing 16.3% of all the works in the collection), all located in the ‘Fundo do Governo Geral de Moçambique’, ‘Fundo do Governo do Distrito de Moçambique’ and the ‘Fundo do Governo do Distrito de Angoche’. The Cabo Delgado collection has 556 letters (representing 83.6% of the collection) located in the ‘Fundo do Governo do Distrito de Cabo Delgado’, most of 22 See Luffin 2014. manuscript cultures 38 MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE Letters by various authors from the district of Mozambique Ni ra In M m tu pa anu din -m e l S un Pi a m o, res sh Ab de ad Sh a Al in eik k h M H us ar ib m e id n a se a Xa hm in b Sua Sh oo un di H ei co a d M k h M jj Se bu Bur a wa n l i m uh ma Bw him Su uhe a mm n b ana lta da va ad un n Ib bun bu Ali S u ra hi Mu n Bw ira z m y i b hug ana n o d S ( a Sa W e lim Fara u le a h z ir an llah ib Sh ibn i) nH ar ajj iff Raj ib ah M Ali n uh ib Al M n am a Ib o m ra r la w i ad hi -m bu m a un n l-M u Al Pe i an oro op d A ni le of li bu n Xa Mus … S i Ha m h Sh b j eik eik Na j b u i h hA A n m bd hm bu Ali i-m ul a d lah bu un Sh eik Abd i ibn n N u h B ull M as ah uh im o w an a i am bn I mm ad m a a d ib m F n Bw a qh an ad au 20 18 16 14 12 10 8 6 4 2 0 1850/1859 1860/1869 1870/1879 1880/1889 1890/1899 1900/1909 No date Total by author Fig. 4: Diagram showing the number of letters per author from the district of Mozambique. them addressed to the island of Ibo, where the capital of the district was established as the main Portuguese settlement in the area. As regards the period covered, the collection ranges from 1858 to 1898 CE: the earliest letters come from Cabo Delgado, from Fumo Matika of Shanga (1858) and from Sultan Said Ali ibn Sultan Abdallah of Messanja in Pemba Bay (1858), while the most recent ones come from Naguema and Muhaburika (1898) in Mozambique district and Mussaka bun23 Mweka (1898) in Cabo Delgado. Figure 4 shows the number of letters by author in the district of Mozambique. 3.3. The authors The two regions (Mozambique district and Cabo Delgado district) have a total of 116 authors between them: 39 came from the former (Fig. 4) and 77 from the latter (Fig. 5). The most important authors of these letters according to local oral history24 and Mozambique historiography include Farallahi, Hussein Ibrahim, Mussa Phiri, Saleh ibn Hajj Ali, Maulid 23 Bun is a local variant of the more common ibn/bin ‘son’. 24 Collected during fieldwork research for my MA thesis in Sancul, Moginqual, Sangage, Angoche, Quissanga and Pangane in 2013 and 2014. manuscript cultures Volay, Bwanamad ibn Banadau, Xa Hajj Ali, Nunu Fatima binti Zakariya, who were all from Mozambique, and Mwaliya Mwidala, Bwana Shaki ibn Abdulatifo, Abdulgafur ibn Abdulatifo, Said Ali ibn Sultan Abdallah, Aburary ibn Sultan Abdurabi, Yussuf ibn Abubakr and Mussaka ibn Mweka from the Cabo Delgado region.25 The Portuguese authors in this collection include the governors of Cabo Delgado, Valentim Hermenegildo de Campos and Duarte (or Eduardo) Humberto de Oliveira, and the local-born mestiços of European descendants, who even had Portuguese names like João Carrilho, Francisco Valente and Dona Maria Lopes (the Queen of Arimba), who were normally considered Wana wa Wibo26 and were able to speak Portuguese and Kiswahili and other local languages and wrote their correspondence in Swahili ajami. In Mozambique, Portuguese authors such as Manoel Pires de Almeida and Agostinho Teixeira de Almeida can be found (both held the post of Capitão-mor das Terras da Coroa, that is, the person responsible for the administration of the Mussoril mainland). 25 Mutiua 2014; Conceição 2006; Pelissièr 1987; Medeiros 1999; Hafkin 1973; Amorim 1910; Albuquerque 1897; Coutinho 1935. 26 ‘Offspring of Ibo’, indicating that they were indigenous people! mc NO 17 1 Muhaburika 1 Sheikh Ayuba ibn Yussuf 1 2 Sheikh Yussuf ibn Abdallah Letters by various authors from Cabo Delgado Sheikh Muhammad inb Ali MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE 350 300 250 200 150 100 50 0 18 Saleh ibn Ali Ibrahim (Marave) Fig. 5: The number of letters per author in the district of Cabo Delgado. 12 Maulid Volay Fig. 6: The number of letters per author in Sancul. mc NO 17 1 Ali ibn Harunan and Hussein ibn Muhammed 39 manuscript cultures Sultuane Metarika ibn Kadimu Sai d ibn Mussa Bakar ibn Is’haca Boana Makasar i bn Abdulatifo Muhammad ibn Sheikh M. Jamal Francisco Rodrigues Sal im ibn Nasry João “Namunho” Francisco Dias Ali ibn Munzi Francisco José Valente Imam ibn Nasry Sal im ibn Ibrahim Mas’ud ibn Muhammad Bwana Mushamo ibn Abubakr Duarte Humberto de Oliveira Namariya ibn Najri Regulo/Sul tuane Mazeze Faque Caetano Ummar ibn Mutubula Bwana Hajj ibn Sumaila et. al l Bwana Hajj ibn Sumaila Sultane Mugabo ibn Reman… Bakar ibn Muhammad et all Nunu Sultane Muriba al-Sahriri Seliman ibn Atumane et all Sheikh Funzi ibn Isma’il al-… Sheik h Yussuf ibn Abubakr Fumo Muguiya Sheik h Abubakar ibn Mussa al-… Abdul Gafur ibn Abdulati fo al-… Abdulatifo ibn Daly al-Mafazy Abubakar ibn Ahmad Abubakar João Calheiro Badri ibn Yahay a Hajj Musa ibn Yussuf No sender's ID 40 Collective authors can also be found, mentioned in letters written on behalf of many people or signed by more than one sender, like the soldiers of Matibane, the people of Musimbi (Quitangonha Sheikhdom) and certain noblemen or political and community leaders who wrote on behalf of their people, like Ali ibn Harun, Hussein bun Ummar, Muhammad bun Ali and Ali bun Abdullah, all of whom came from Sancul. Cabo Delgado not only leads in terms of the number of letters and individual authors, but also in the number of collective authors, some examples being the people of Pangane and noblemen from Tungi and Arimba, who all wrote on behalf of their people – complaining about the appointment of colonial officers who were not of their approval and requesting to postpone the payment of taxes and even open shops in their villages. The correspondence mostly consists of official and administrative reports. Consequently, the collection is not a good source for examining the biographies of the authors, although it does provide some notes, which can help in understanding the family trees of some of the writers and their succession in the political structure. The figures above, which show the dynamics of correspondence per author, are not conclusive, but they do indicate trends for certain authors and regions such as Sancul (Fig. 6) and Quissanga (Fig. 7). Figures 3 and 4 can help to clarify the tendencies found in Figures 1 and 2 as the former show the distribution of the letters in terms of their authors and the areas where they were written. In Cabo Delgado, 235 letters corresponding to 42.26% of the collection from the region (556 letters) come from Quissanga, and 220 of these letters, or 93.6% of them, come from the al-Mafazi family, whose most important representative was Bwana Shaki ibn Abdulatifo al-Mafazi, an author who wrote 147 letters ranging from 1872 to 1893. The letters from Quissanga correspond to 12 authors among the 77 from Cabo Delgado. This distribution illustrates the important role Quissanga and its rulers played during a specific historical period in the Cabo Delgado region due to its key geo-strategic location, lying on the mainland opposite the most important island in the Quirimba Archipelago. The first capital of the district of Cabo Delgado was established on the Island of Ibo in the sixteenth century and became the most important harbour and seat of representative companies later on. It was moved to Porto Amélia (Pemba) during the rule of the Nyassa Company (1893–1929). The geo-strategic role of Quissanga as a source of power that the Bwana Shaki manuscript cultures MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE family wielded in this period is illustrated by this collection, which tells us about the integration of local rulers into the Portuguese colonial administration (vassalage treaties), the ivory trade, slavery, the impact of mfeqani27 in the region, crime and the administration of justice, demographic movements and payment of taxes, among other things. Quissanga and Sancul have certain similarities which deserve our attention when it comes to studying northern Mozambican ajami manuscripts. The sheikhdom of Sancul was located in front of Mozambique Island to the south. It was through the territories of Sancul (to the south) and the sheikhdoms of Matibane and Quitangonha that Mozambique Island accessed the mainland of Mussoril and the rest of Makhuwana in the hinterland. Access to the hinterland was of vital importance for the development of trade that connected Mozambique Island to the wider Indian Ocean, but it was also key to the supply of foodstuffs, manpower and later on for effective occupation of the region as well. Sancul additionally gave access to Infussi and Moginqual to the south across the coast on a route which took the caravans to Angoche through Sangage, while Matibane linked Mozambique Island to the north through a coastal route going to Quitangonha, Lúrio and the southern areas of Cabo Delgado. Out of a total of 109 letters from the district, 39 come from Sancul, representing 35.7% of the total number of letters in Mozambique. The most important authors in this collection are Saleh ibn Hajj Ali Ibrahim al-Moroni with 12 letters (1886– 1896) and Maulid Volay with 18 letters (1887–1896). Maulid Volay and Marave were members of the same family, but fought for the leadership of the sheikhdom during most of the late nineteenth century. Both of them claimed or were appointed to the chair of Sheikh of Sancul and were appointed one after the other as Capitão-mor28 of Sancul by the Portuguese, which created even more rivalry between them. 27 The mafiti or maviti and maguanguara raids that reached Quissanga and other coastal villages between 1874 and 1876 and interrupted the caravan routes that linked the Zambezi valley and the margins of Lake Nyasa to Ibo. For more on the impact of mfeqani in northern Mozambique, see Medeiros 1999; Palmer and Newitt 2016; Hafkin 1973. 28 Captain-major, as the title is known in some English bibliographies, was a designation bestowed on certain members of the local political elite who demonstrated – or were expected to demonstrate – some loyalty to the Portuguese. According to Newitt (1995), the rank was equivalent to honorary consul, representing Portuguese sovereignty. Although a capitão-mor did not automatically receive any administrative or military powers, those who were appointed to such a position did gain some authority. When bestowed upon a Portuguese citizen (like the title of Capitão-mor das Terras Firmes do Mossuril, for example), the position always went hand in hand with exercising military and administrative authority. mc NO 17 41 MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE Fig. 7: Map of the coastal area in northern Mozambique.29 Sancul was the closest sheikhdom30 to Mozambique, where the Portuguese had established their government, and controlled the main trade routes which linked Mozambique Island to the mainland, while Quissanga played the same role in Cabo Delgado, where Ibo Island was the main centre. These strategic roles provided more political power to their rulers in negotiating with the Portuguese settlers on Ibo and Mozambique Island. 29 3.4. The 1880s and the growth of ajami correspondence The figures below31 suggest that the 1880s CE were the most active decade in the district of Cabo Delgado in terms of ajami correspondence and, as indicated, the same tendency occurs in Mozambique as well (Fig. 8). As a consequence of the Portuguese military occupation and the vassalage treaties with local rulers, there was a sudden growth in correspondence from the 1850s, reaching its zenith in the 1880s. No letters from the first decade of the twentieth century have been found in the district of Cabo Delgado. Taken from Medeiros 1988, 9. 30 Ruled by the local Swahili elite, who had moved from Mozambique Island after the Portuguese occupation. See Hafkin 1973. mc NO 17 31 These figures are not definitive as there are almost 150 letters which still need to be dated (see the column entitled ‘No date’). manuscript cultures 42 MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE Letters from Cabo Delgado in chronological order 250 200 150 Number of letters 100 50 0 1850/1859 1860/1869 1870/1879 1880/1889 1890/1899 1900/1909 No date Fig. 8: Number of letters from the district of Cabo Delgado in chronological order. Letters from the district of Mozambique in chronological order 60 50 40 30 20 10 0 1850/1859 1860/1869 1870/1879 1880/1889 1890/1899 1900/1909 No date Fig. 9: Number of letters from the district of Mozambique in chronological order. manuscript cultures mc NO 17 43 MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE In Mozambique district, although one letter exists from the 1860s and another from the 1870s, the figure suggests that regular correspondence in Swahili between the Portuguese and local rulers must have peaked in the 1880s and began to decline in the following decade (1890s) until the first decade of the twentieth century when it was suppressed from the official realm (Fig. 9). The concentration of letters in the 1880s raises the question of why this happened in this period – a question which could be answered in several ways. From the 1880s onwards, the European colonial powers were engaged in the partition of Africa with the establishment of certain rules set out at the Berlin Conference (1884–1885).32 This meeting was preceded by exploration campaigns (in the 1870s to 1880s) such as those conducted by Serpa Pinto, Cameron, Livingstone and others.33 Serpa Pinto, who was in Cabo Delgado and Niassa from 1884 to 1886, is mentioned in some of the letters by Bwana Shaki ibn Abulatifo al-Mafazi, Abdulgafur ibn Abdulatifo al-Mafazi, Muhammad ibn Sheikh and Mwaliya34 as an ally whom they worked with or helped during his trips from coastal Cabo Delgado to Nyassa. This international conference affected Portuguese–African relations afterwards. Many treaties were signed during the period to ‘legitimise’ Portugal’s colonial power over the territory. As Portugal did not have sufficient military power to ensure control over the vast territory of Mozambique, the Portuguese had to rely on local African rulers through vassalage treaties (with certain sheikhs and sultans) or even bestowing the titles of Capitão-mor and Sargentomor in a system that Hafkin has called ‘survival strategy’.35 32 Newitt 1995, 352. 33 Ki-Zerbo 1992, 67–87. Fig. 10. Letter from Mussaka ibn Mafiga Maniga, 16 August 1888. AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 8, maço 4. In this system, local rulers agreed to hoist the Portuguese flag in their own territories and received a monthly allowance in return. In practice, however, they often broke the agreement and rebelled against the Portuguese.36 Correspondence from this period provides evidence of these dynamics. In two letters from Mussaka ibn Mafiga, we can see just how fragile the relationship between the local rulers and the Portuguese was (see Figs 10 and 11). Mussaka ibn Mafiga was the most powerful Swahili leader of all in the region of Palma, Cabo Delgado, and he often helped the Portuguese. He also rebelled against the colonial power on occasion, though. His letters display both friendship and enmity (see Table 2). 34 See letters from Bwana Shaki ibn Abdulatifo al-Mafazi to the Governor of the District of Cabo Delgado, 26 October 1885 and 11 August 1885 in AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 10, maço 2; letters from Abdulgafur ibn Abdulatifo al-Mafazi to Secretary Júlio, 21 December 1884 and 28 December 1884 and to the Governor of Cabo Delgado, Francisco de Ornela Pery da Câmara, 13 February 1885 and 29 April 1885; letters from Muhammad ibn Sheikh to the Secretary of the Governor of Ibo, 14 July 1885 and 31 August 1885, all in AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 10, maço 2; and letter from Mwaliya Mwidala to the Governor of Cabo Delgado, (26) December 1884 in AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 8, maço 3. 35 Hafkin 1973. mc NO 17 36 Pelissièr 1987; Medeiros 1999; Hafkin 1973. manuscript cultures 44 MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE Fig. 11: Letter from Musaka ibn Mafiga Maniga, 23 October 1888. AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 8, maço 4. Table 2: Transliteration and translation of the letter from Musaka ibn Mafiga Maniga, 16 August 1888 (Fig. 10). 1st line: Ila janabi shehe al-aziz indana sinyuru shirijeto [sargento] salamuhullahu taʿala This letter [is] to the leader, the great [one], to Senhor Sargento, [may the] peace of Allah be upon you. 2nd line: wama ba aduhu nakuarifu yakama hali yangu jema wa zaidi I inform you that I am fine. Then 3rd line: ya habari nimesikia Adurabi umufungu basi nataka I heard that you arrested Adurabi (Abdurabi). I want 4th line: wufungue allah allah tazama allah allah wufungue kwa amuri yangu tafadhali you to release him! By God, consider it, by God, please, release him as if by my order! 5th line: usifanye mengine usifanye matata na usifanye matata nadari yako Don’t create any further problems, don’t cause any further trouble and don’t cause trouble to you. 6th line: basi nami nafanya matata ila ufungue basi ukifungua basi Because I will cause trouble unless you release him. If you release him, [everything] will be fine. 7th line: hapana matata mimi nawe walakini iwapo aukufungua fahamu There will be no conflict between me and you, but if you do not release him, remember, 8th line: mimi nawe hatupatani maisha yangu kana wataka wema mimi nawe you and I will not make up as long as I live. If you want peace between us 9th line: ili fungue mwinyi Adurabi basi ndi habari yangu release Mwinyi Adurabi. This is my concern. 10th line: wasalam bwana Musaka ibn Mafiga Maniga Greetings, Bwana Musaka ibn Mafiga Maniga manuscript cultures mc NO 17 45 MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE Table 3: Transliteration and translation of the letter from Musaka ibn Mafiga Maniga, 23 October 1888 (Fig. 11). 1st line: Ila janabi al-aziz al-akram al-mukarram al-akhi sinyuru Kapitamoro Mutapa To the great, respectful, respected, brother Senhor Capitão-mor Mutapa 2nd line: Salamuhullahu taala insha allahu taala baada ladhi nakuarifu barua yako imewasili [May the] Peace of Allah, the Highest, be upon you. After that I inform you that your letter has arrived. 3rd line: nimesoma nimefahamu yakunita nije ndiyo wajibu wako wewe na mimi nakuliko habari ya zita I have read it, I have understood that you summoned me to come [to Ibo] – this is your responsibility. And I will inform you about the war. 4th line: Ukitaka menyewe ni Mikumba menyewe Mikumba kama wataka menyewe Mikumba ni menyewe zita Mikumba na huyo If you want to know who caused the war, it was Mikumba. If you want to know the instigator, Mikumba is the instigator of the war, [but], 5th line: ndugu yango si menyewe zita mufungueni [nabuda kuniwa] basi walakini zita si menyewe ya Mikumba my brother, is not the one who started the war. Release him […]; it is not his war, it is Mikumba’s. 6th line: Na mimi nimepereka barua na kazi makusudi ushahidi marufu yakua zita ya Mikumba na huyo niwapo And I sent a letter with the intention to provide an important testimony that the war is due to Mikumba and that 7th line: Mrugu munafiki mkubwa na wewe sinyuru kapitamoro mufungue ndugu yangu mana si menyewe zita menyewe Mikumba Mrugu is a great destabiliser and you, Senhor Capitão-mor, release my brother because it is not his war. It is Mikumba’s war. 8th line: Nahie(?) zita si yako wewe wala matajiri wala mutu mwingine ila Mikumba nimenyewe zita And this war is not yours, nor the traders’, nor any other person’s except Mikumba’s; it is his war. 9th line: Ni kama yalikua ya (A)durabi ni wazungu mugoti ningalikata ninkalikenda Murue And if it was [a war] caused by Adurabi and the Portuguese, I would kneel down to apologise and I would go to Murue 10th line/right margin: Au kisiwani ningalikuja walakini makusudi yangu Mikumba na wewe or I would come to the island [of Ibo]. But my purpose [is to tell you] that it is Mikumba’s fault. And you, 11th line/right margin: Kama wataka imani ila unipatiye Mikumba unipereke- if you want peace, you have to get Mikumba for me, bring him to me or arrest re wama mufunge him. 12th line/right margin: Na sababu mimi na Mikumba hatupatani na rafiki hungalikufa ila Mikumba rauki yaru (?) Because Mikumba and I do not get along. And it was Mikumba who killed my friend(?).37 13th line/ top margin: Ukinipatie Mikumba wama ukinifungie itie amani Get Mikumba for me or arrest him for me, and make peace, 14th line/top margin: Amani iwapo hukunipatie Mikumba ao kufunga hakuna so there is peace. If you don’t get Mikumba for me or don’t arrest him, there imani allah allah mufungue ndugu yangu will not be any peace. By God, release my brother! 15th line/top margin: Wala usisikie habari ya Mikumba mufungue wakatabahu Sultane Musaka ibn Mafiga And don’t listen to what Mikumba says, [just] release him! Written by Sultan Mussaka ibn Mafiga 37 The two lines 12 and 13 can be interpreted in different ways and there are two words I cannot discern (‘rauki yaru’) at the end of line 12. But based on the Portuguese interpretation available in the same archival file, it seems that Mikumba was responsible of causing war and of killing João Carrilho who was a friend of Mussaka. And this was the topic in these lines. However, the Portuguese document is an interpretation and I tried to make a translation of the letter. mc NO 17 manuscript cultures 46 MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE Letters per area in the district of Cabo Delgado 250 200 150 100 Letters 0 Metarica Quiterajo Tungi Mocimba Quirimizi Quissanga Arimba Medo No place Memba Palma Ibo Puto Shanga Quionga Olumboa Pangane Pemba Bay Mucojo Muluse Ancuabe Messanja Ubiri Mushishila 50 Fig. 12: Letters per area from the district of Cabo Delgado. The vassalage treaties that could grant Portugal the right to act as the colonising power could only be signed if the African local rulers agreed to them. How could this be achieved? The African rulers of northern Mozambique had been involved in the slave trade, which accorded them respect among their partners on the western Indian Ocean coast and among some French and Portuguese slave dealers and made them wealthy until the late nineteenth century. How and why, then, would they change their opinion and accept Portuguese authority, as demonstrated in their correspondence? The answer is a complex one because there are so many factors involved. First of all, Zanzibar had been under British influence in the 1880s, which led to the Protectorate being created in 189038 and the power of the Sultan, who was a ‘guiding force’ for many East African rulers, decreasing. Second, the British were engaged in the promotion of legitimate commerce in the region and the abolition of the slave trade, which was also the aim of Brito Capello, Serpa Pinto and Roberto Ivens, who all started their African campaigns in 1877.39 Thus, some of the local rulers accepted the treaties as a way of guaranteeing 38 the survival of their political and social status and having Portuguese protection against attacks by their enemies, who were mostly neighbouring local rulers. In Cabo Delgado, for instance, mavite or mafiti warriors attacked several Makhuwa and Mwani communities in Cabo Delgado, while in the district of Mozambique, hinterland Makhuwa kingdoms and those of the coast shifted their relations between enmity and friendship according to their immediate interests. Others accepted the treaties to distract the Portuguese and continue trafficking slaves out of Portuguese and British inspections. 3.5. The content of the letters The content of the letters includes various topics relating to the social, political and economic history of the northern Mozambique region with a focus on the establishment of Portuguese colonial administration, which meant the integration of the local political elite (sheikhs, sultans and mwenes)40 into the administration structure under the rule of Portuguese colonial officers as Capitão-mor, Sargentomor and Cabo, all of whom performed administrative functions and were under the subordination of the Sheikh Bang 2003, 7. 39 Pelissiér 1987, 138. These Portuguese explorers were forced to travel by the pressure of other European colonial powers who demanded the effective occupation of the territories to confirm them as colonial possessions, as agreed at the Berlin Conference. Another factor was the Portuguese anti-slavery decrees of 1832 and 1854/69 and the Anglo-Portuguese agreement on the abolition of slavery, which was signed in 1842. See Medeiros 1988, 28–31. manuscript cultures 40 The title of ‘Sheikh’ is used according the designation given to the heads of northern Mozambican Swahili states such as Quissanga, Sangage, Sancul or Quitangonha. In the same way, ‘Sultan’ is a title used to refer to the heads of the Sultanates of Tungi and Angoche. The term mwene used in this article is the Makhuwa word for ‘King’, although its etymology can also be interpreted in Sacleux’s way (cf. Sacleux 1939). mc NO 17 47 MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE Boana Shaki ibn Daly al-Mafazy Abdulatifo ibn Daly al-Mafazy Boana Makasar ibn Abdulatifo Ali ibn Faque Muhammad ibn Sheikh Muhammad J Musa Pira Salim Ibrahim Fig. 13: The authors of the letters from Quissanga. (in the Sheikhdom) and the Sultan (in the Sultanate). This also meant the legitimation of the sheikhs and sultans by the Portuguese Military Authority through their official appointment and inauguration of their seats of power. This resulted in many conflicts among the African ruling elites in the region, who wanted to gain advantages from the new ‘client’. The word ‘client’ in this context is suitable when trying to understand the way this process developed in this period as the ‘collaboration system’ between the Portuguese colonial officers and the local rulers. In practice, it only worked on the basis of specific interests. The Portuguese were interested in securing trade routes and extending their authority to the hinterland through the local rulers, who used the opportunity to maintain their independence and earn a monthly ‘allowance’ while continuing to practise the slave trade behind their backs.41 See the letter from Capitão-mor Saleh bun Ali Ibrahim from 1894, to quote just one example among many; in this letter, Saleh bun Ali Ibrahim (Marave) reports on the attack perpetrated by Maulid Volay (his uncle, 41 To quote just one example, see the letter from Capitão-mor Saleh bun Ali Ibrahim in AHM, Governo Geral de Moçambique, Fundo do Século XIX, caixa 8, maço 1, 1894; in this letter, Saleh bun Ali Ibrahim (Marave) reports on the attack perpetrated by Maulid Volay (his uncle, and Sheikh of Sancul) on his land. Volay was himself capitão-mor before Marave was appointed one. And Marave was latter reported to be a rebel and support the War of Namarral Makhuwas against the Portuguese. For more details on the ‘collaboration system’ that defined the integration of local rulers into the Portuguese administration before the end of Campanhas de Ocupação efectiva, see Hafkin 1973; Pelissièr 1987; Medeiros 1999. mc NO 17 and Sheikh of Sancul) on his land. Volay was Capitão-mor himself prior to Marave, and Marave was later reported to be a rebel and support the War of Namarral Makhuwas against the Portuguese.42 The slave trade had developed a social and economic network through which East African rulers, including those of northern Mozambique, could access commodities from Europe and Asia. The declining benefits of this network as the slave trade was abolished provoked a scarcity of these commodities among northern Mozambique rulers. The ‘alliance or integration’ into the Portuguese Authority was an alternative method for the local political elite to gain access to prestigious goods through the Portuguese-controlled networks. Thus, we can find requests for cloth, umbrellas, paper, guns and gunpowder in some of the correspondence.43 It was also a strategic political option to maintain their political and social authority among their people in the context of growing Portuguese military power aiming to achieve an ‘effective occupation’ (Ocupação efectiva), which was the campaign of military submission of all of Mozambique’s territories, precipitated by the agreements made at the Berlin Conference (1884–1885), which stipulated 42 For more details on the ‘collaboration system’ that defined the integration of local rulers into the Portuguese administration before the end of the Campanhas de Ocupação efectiva, see Hafkin 1973, Pelissièr 1987 and Medeiros 1999. 43 Mutiua 2014 and 2015. manuscript cultures 48 that all the colonial powers should exert effective authority in their colonial territories. In Mozambique, its severest period lasted for 20-30 years from 1880.44 The political, social and economic integration of local societies into the Portuguese colonial order by such means as the demographic census, registration of slaves and freed slaves,45 reports on raping of local noblemen’s wives by Portuguese soldiers and the end of slave trafficking,46 the influence of Indian merchants on the development of trade 47 and the payment of décimas industriais (‘industrial tax’)48 all serve to exemplify the variety of topics broached in these manuscripts. Other topics are also mentioned such as the continuation of the slave trade (in many other ways), the ivory trade from the hinterland (Nyassa, Zambezia and Medo – in the south of Cabo Delgado, integrating the areas of Montepuez, Balama, Namuno, Ancuabe and Chiúre), crime, trade and links to the Swahili coast. MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE The links to the Swahili coast are not a specific issue in the manuscripts, although some letters refer to the appearance of ‘Wajojo’ in the region, who were associated with the slave trade,49 like those reported in Francisco Valente’s undated letter,50 or the presence of ‘Waswahili’, such as an employee of Sultan Said Barghash of Zanzibar51 who travelled along the coast of Cabo Delgado from Tungi to Ibo looking for a Zanzibari mlunguana52 who was supposed to be in that area. The presence of European foreign citizens (nonPortuguese citizens) is also reported in the correspondence. In his letter of 26 April 1886, for instance, Abdulgafur ibn Abdulatifo al-Mafazi reports that soldiers were looking for two Welsh citizens who had apparently got lost in the Quissanga hinterland, while on 6 and 7 July 1884 the same author reported the presence of an English citizen who was looking for a house to rent or a place to build his own house. In the district of Mozambique, Sheikh Yussuf ibn Abdallah of Sancul reports that an English consul was crossing the River Mutiquiti on his travels.53 49 44 Kizerbo 1973; Pelissièr 1987; Newitt 1995. 45 See the letters that Boana Shaki ibn Daly sent the Governor of Cabo Delgado, João Lobo Teixeira Barros, on 15 November 1861: AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 10, maço 2. 46 See the letter by Ali ibn Munzi (of Palma), 2 November 1893, AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 28, maço 2. Also see the letters from Francisco Valente in AHM, Governo do Distrito de Cabo Delgado, caixa 8, maço 2; Francisco Valente also reports about the state of trade, which was mainly under the control of Indian and Makhuwa merchants. 47 See the letter by Capitão-mor Boana Shaki, for instance (undated), asking the Governor of Cabo Delgado to send Indian traders to his land (Quissanga), in AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 10, maço 2, and the letter by Maulid Volay (25 February 1893) reporting about Indian shops in his land (Sancul-Quivolane), in AHM, Fundo do Século XIX, Governo Geral de Moçambique, caixa 150, maço 1. These are just a few of the numerous letters reporting on these issues (see Mutiua 2014 for more examples and details). 48 Letter by Boana Shaki ibn Abdulatifo al-mafazi, 3 October 1877, in AHM, Governo do Distrito de Cabo Delgado, caixa 10, maço 2. manuscript cultures The letter by Maulid Volay (2 January 1886) in AHM, Fundo do Século XIX, Governo Geral de Moçambique, caixa 152, maço 2 refers to Wajojo (Comorians) who were trading in slaves in Infusi with the local sheikh and Sultan of Angoche. Sheikh Ayuba ibn Yussuf (the interim sheikh of Sancul in 1874) also reported on the presence of Wajojo in the same area of Infusi for the purpose of trading slaves. In his letter from February 1886 (in AHM, Fundo do Século XIX, Governo Geral de Moçambique, caixa 149, maço 1), Sheikh Saleh ibn Xa Ibrahim of Moginqual also reported on the presence of Wajojo slave-dealers in his land. The arrival of eight Wajojo dhows was also reported in Arimba (Quissanga region, in Cabo Delgado) in an undated letter written by sargento-mor Mzungu Xico (Francisco) Valente, in AHM, Governo do Distrito de Cabo Delgado, caixa 8, maço 2. Juma ibn Hamisi and Omar ibn Maulid reported in their letter dated 11 March 1885 that a dhow of Wajojo was embarking slaves in Tandanhangue (Quissanga), in AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 9, maço 3. 50 See the letter by Francisco Valente (signed mzungo Xico Valente) in AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 8, maço 2. 51 See the letter from Tahiri ibn Mussa al-shirazi, Capitão-mor of Quiterajo (Cabo Delgado), 13 August 1885, in AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 10, maço 3. 52 ‘Noble man’ in Kiswahili. 53 See letter from Sheikh Yussuf ibn Abdallah, 27 August 1881, in AHM, Fundo do Século XIX, Governo Geral de Moçambique, caixa 147, 1; letter from Francisco Valente, undated, in AHM, Fundo do Seculo XIX, Governo do Distrito de Cabo Delgado, caixa 8, maço 4; letters from Abdulgafur ibn Abdulatifo al-Mafazi, 26 April 1886, 6 and 7 July 1884, in AHM, Fundo do Seculo XIX, Governo do Distrito de Cabo Delgado, caixa 10, maço 2. mc NO 17 49 MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE The relationship between the local African rulers in northern Mozambique can also be seen in these manuscripts. As mentioned earlier, the opportunity for forging an alliance with the Portuguese as an alternative method of accessing material wealth exacerbated the competition for power and prestige among the African rulers. The case of Saleh bun Ali Ibrahim (Marave) and Maulid Volay, both of whom came from Sancul, is a vivid example of this rivalry. More examples can be found between the Guernea and Morla, the two most important rulers of the Imbamela Makhuwa. However, the manuscripts also provide evidence of friendly relations such as those between Bwana Shaki and Mwaliya of Medo. Mwaliya was a Makhuwa king (mwene) of the southern hinterland of Cabo Delgado who studied the Qurʾān in Quissanga and became a Muslim, adopting Swahili culture. This is also attested in a letter written by the newly appointed mwene Mwaliya, who wrote to the Portuguese governor in Ibo to inform him about the death of his uncle, Mwaliya Mwidala, and requested cloth – ‘not trousers, but cotton fabric as I want to dress like Wajojo do’. 4. Conclusion The ajami manuscripts of northern Mozambique held by the Mozambican Historical Archive (or Mozambique National Archives – AHM) are a testimony to the role that local rulers played in the process of establishing the Portuguese precolonial administration. In this sense, ajami literacy emerged from the religious sphere and became an important medium of communication in the secular, political and administrative realm. The process of diffusing literacy through Islamic education and the use of Arabic script in Kiswahili links this literary tradition to the wider Swahili region. Its integration as part of the colonial system of communication and the use of loanwords from Portuguese and northern Mozambican languages make ajami manuscripts unique to northern Mozambique. REFERENCES Albuquerque, Joaquim Mouzinho de (1897), A Campanha contra os Namarrais: Relatórios, Lisboa: Imprensa Nacional. Alpers, Edward (1999), ‘Islam in the service of colonialism? Portuguese strategy during the armed struggle in Mozambique’, Lusotopie, 1999: 165–184. Alpers, Edward (2001), ‘A complex relationship: Mozambique and the Comoro Islands in the nineteenth and twentieth centuries’, Cahiers d’Etudes Africaines, 41, 161: 73–95. Amorim, Pedro Massano de (1910), Informações Relativa á Região de Angoche: Noticia Historica sobre o Distrito de Moçambique, Lourenço Marques: Imprensa Nacional. Bang, Anne (2003), Sufis and Scholars of the Sea: Family Networks in East Africa, 1860–1925, Oxford: Routledge Curzon. Bonate, Liazzat J. K. (2007), Tradition and Transition: Islam and Chiefship in Northern Mozambique, ca. 1850–1975, Cape Town: UCT, PhD thesis. Bonate, Liazzat (2016), ‘Islam and literacy in northern Mozambique: historical records on the secular uses of Arabic script’, Islamic Africa, 7: 60–80. Conceição, Rafael (2006), Entre o Mar e a terra: situações identitárias do Norte de Moçambique, Maputo: Promédia, Identidades. Coutinho, João de Azevedo (1935), As duas conquistas de Angoche, Lisboa: Pelo Império, no 11. Declich, L. (2001), ‘The Arabic manuscripts of Zanzibar National Archives: source for the study of popular Islam during the 19th century’, in Biancamaria Scarcia Amoretti (ed.), Islam in East Africa: New Sources (Archives, Manuscripts and Written Historical Sources. Oral History, Archaeology), Rome: Herder, 47–57. mc NO 17 manuscript cultures 50 MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE Ferreira, Eduardo de Sousa (1977), O fim de uma Era: o colonialismo português em África, Lisboa: Sá da Costa Editora. Palmer, Hilary C. and Malyn D. D. Newitt (2016), Northern Mozambique in the Nineteenth Century: The Travels and Explorations of H. E. O’Neill (Leiden/Boston: Brill). Hafkin, Nancy J. (1973), Trade, Society, and Politics in Northern Mozambique, ca. 1753–1913, Boston: Boston University, PhD thesis. Pelissiér, René (1987), História de Moçambique: formação e oposição, 1854–1918, vol. 2, Lisboa: Editoria Estampa. Khamis, Said A. M. (2002), ‘Wondering about change: The taarab lyrics and global openness’, Nordic Journal of African Studies, 11/2: 198–205. Pouwels, Randall (1987), Horn and Crescent: Cultural Change and Traditional Islam on the East African Coast, 800–1900, Cambridge: Cambridge University Press. Ki-Zerbo, Joseph (1992), História da África Negra, 3 ed., vol. 1: Europa-América. Prata, A. Pires (1983), A influência da língua portuguesa sobre o suahíli e quatro línguas de Moçambique, Lisboa: Instituto de Investigação Científica Tropical. Luffin, Xavier (2014), ‘Swahili documents from Congo (19th century): variation in orthography’, in Meikel Mumin and Kees Versteegh, The Arabic Script in Africa: Studies in the Use of a Writing System, Leiden/Boston: Brill, 312–317. Prestholdt, Jeremy G. (1998), ‘As artistry permits and custom may ordain: the social fabric of material consumption in the Swahili World, circa 1450 to 1600’, PAS Working Papers 3, Evanston: Northwestern University. Medeiros, Eduardo (1988), As etapas da escravatura no norte de Moçambique (Maputo: Arquivo Histórico de Moçambique, Estudos, 4. Robinson, David (2004), Muslim Societies in African History (New Approaches to African History) Cambridge: Cambridge University Press. Medeiros, Eduardo (1997), História de Cabo Delgado e Niassa (c. 1836–1929), Maputo. Schadeberg, Thilo C. and Mucanheia, Francisco Ussene (2000), Ekoti: the Maka or Swahili Language of Angoche, Cologne: Rüdiger Köppe Verlag. rd Mutiua, Chapane (2014), Ajami Literacy, Class and Portuguese Colonial Administration in Northern Mozambique, Cape Town: UCT, Faculty of Humanities, MA thesis. Mutiua, Chapane (2015), O Islao e o Processo de Literacia no norte de Mocambique entre os finais do seculo XIX e principios do seculo XX, in Teresa Cruz e Silva and Isabel Casimiro (eds), A Ciencia ao Servico do Desenvolvimento?, Dakar: CODESRIA, 205–220. Sousa, João de [1788], Documentos arábicos da história portugueza copiados dos originais da Torre do Tombo, Lisboa: Comissão da Academia Real das Sciencias. Trimingham, J. S. (1971), Islam in East Africa, Oxford: Clarendon Press. Vilhena, Ernesto J. (1905), A Companhia do Nyassa: Relatórios e Memórias sobre os Territórios, Lisboa: A Editora. Newitt, Malyn (1995), A History of Mozambique, London: Hurst and Co. Ngunga, Armindo (2004), Introduçao à Linguística Bantu, Maputo: Imprensa Universitária. manuscript cultures mc NO 17 MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE 51 INTERVIEWS Interview with Mussa Abudo ‘Na Tambuka’, Tibane, Sancul, 2013. AHM, Governo do Distrito de Cabo Delgado, caixa 8, maço 2: letters from Francisco Valente (undated). LIST OF LETTERS FROM THE MOZAMBIQUE HISTORICAL ARCHIVE AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 10, maço 2: letter from Capitão-mor Boana Shaki (undated). AHM, F. S. XIX, GDCD, caixa 8, Mç. 3: letter from Mwaliyya, 5 June 1888. AHM, Fundo do Século XIX, caixa 10, maço 2: letter from Boana Shaki b. Abdulatifo al-Mafazi, Sheikh of Quissanga, 1892. AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 10, maço 2: letters from Bwana Shaki ibn Abdulatifo al-Mafazi to the Governor of the District of Cabo Delgado, 26 October 1885 and 11 August 1885. AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 10, maço 2: letters from Abdulgafur ibn Abdulatifo al-Mafazi to Secretary Júlio, 21 December 1884 and 28 December 1884 and to the Governor of Cabo Delgado, Francisco de Ornela Pery da Câmara, 13 February 1885 and 29 April 1885; letters from Muhammad ibn Sheikh to the Secretary of the Governor of Ibo,14 July 1885 and 31 August 1885. AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 8, maço 3: letter from Mwaliya Mwidala to the Governor of Cabo Delgado, 26 December 1884. AHM, Governo Geral de Moçambique, Fundo do Século XIX, caixa 8, maço 1: letter from Capitão-mor Saleh bun Ali Ibrahim, 1894. AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 10, maço 2: letters from Boana Shaki ibn Daly to the Governor of Cabo Delgado, João Lobo Teixeira Barros, 15 November 1861. AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 28, maço 2: letter from Ali ibn Munzi (of Palma), 2 November 1893. mc NO 17 AHM, Fundo do Século XIX, Governo Geral de Moçambique, caixa 150, maço 1: letter from Maulid Volay, 25 February 1893. AHM, Governo do Distrito de Cabo Delgado, caixa 10, maço 2: letter from Boana Shaki ibn Abdulatifo al-mafazi, 3 October 1877. AHM, Fundo do Século XIX, Governo Geral de Moçambique, caixa 152, maço 2: letter from Maulid Volay, 2 January 1886. AHM, Fundo do Século XIX, Governo Geral de Moçambique, caixa 149, maço 1: letter from Sheikh Saleh ibn Xa Ibrahim of Moginqual, February 1886. AHM, Governo do Distrito de Cabo Delgado, caixa 8, maço 2: letter from Sargento-mor Mzungu Xico (Francisco) Valente (undated). AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 9, maço 3: letter from Juma ibn Hamisi and Omar ibn Maulid, 11 March 1885. AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 10, maço 3: letter from Capitão-mor of Quiterajo (Cabo Delgado) Tahiri ibn Mussa al-shirazi, 13 August 1885. AHM, Fundo do Século XIX, Governo Geral de Moçambique, caixa 147, maço 1: letter from Sheikh Yussuf ibn Abdallah, 27 August 1881. AHM, Fundo do Seculo XIX, Governo do Distrito de Cabo Delgado, caixa 10, maço2: letters from Abdulgafur ibn Abdulatifo al-Mafazi, 26 April 1886 and 6 and 7 July 1884. manuscript cultures 52 MUTIUA | SWAHILI MANUSCRIPTS FROM NORTHERN MOZAMBIQUE AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 8, maço 3: letter from Mwaliyya, 5 June 1888. AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 10, maço 2: letter from Momba ibn Is´haka, sheikh of Quirimizi, [1880]. AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 10, maço 2: letter from Bwana Shaki ibn Daly (undated). AHM, Fundo do Século XIX, Governo do Distrito de Cabo Delgado, caixa 14, maço 1: letter from Said ibn Musa to Governor of the District of Cabo Delgado, 20 February 1888. PICTURE CREDITS Figs 1–3: © Mozambican Historical Archive (AHM), IT Department. Figs 4–6: The author. Fig. 7: © Medeiros 1988, 9. Figs 8–9: The author. Fig. 10–11: © Mozambican Historical Archive (AHM), IT Department. Figs 12–13: The author. manuscript cultures mc NO 17 53 PARKAR | ARABIC-SWAHILI HAMZIYYA MANUSCRIPTS Article Arabic-Swahili Hamziyya Manuscripts: Observations on Two Testimonies of the Text Ahmed Parkar | Kilifi, Kenya and Hamburg, Germany 1. Introduction The general aim of this paper is to show how a single text, namely the Qaṣīda Hamziyya (‘QH’), an ode to the Prophet Muḥammad originally produced in Arabic, had dynamic codicological ‘lives’ in Arabic and Swahili over a period of a century on the East African coast. The main focus of the science of codicology is on the handwritten book or a manuscript originally produced as a craft.1 This field of study is quite recent and tries to answer such questions as how, when, where and for what purpose a given manuscript was made. Generally, the field explores all the techniques involved in the making of a manuscript.2 In this paper, I intend to focus on particular codicological elements, such as the mise-en-page (page layout) of two QH manuscripts. One, shelf-marked as Ms. 541, which I shall simply call ‘manuscript D’ here, is currently preserved in the East Africana Collection in Dr Wilbert Chagula Library at the University of Dar es Salaam (UDSM),3 Tanzania, and the second one, which I shall call ‘manuscript M’, is part of a private collection in Mombasa, Kenya that I was kindly able to access.4 In the context of codicology, the term mise-en-page means the arrangement of the various elements appearing on a page, not only with respect to the main text, but to the margins and decorations as well, along with the relationship between these different elements.5 While looking at the textual elements of the two manuscripts (D and M), I shall also try to compare them with a third one, ‘manuscript L’, which is currently shelf-marked as Ms. 53823 in the library of the School of Oriental and African Studies (SOAS), University of London. This manuscript is the oldest known Swahili-Islamic manuscript that contains the Hamziyya ode. There are various approaches to analysing the textual layout arrangements in manuscripts and their means of production in general.6 Prior to embarking on a discussion of the layout of the QH manuscripts, however, I wish to outline my own approach, which is by gleaning information and terminology from several disciplines: Swahili, the field of manuscriptology, Arabic, and Islamic studies. This multidisciplinary view is necessary because it can be very difficult to find appropriate terms in a single field of study. 1 Gacek 2009, 64 2 2. The state of the art Déroche 2006, Gacek 2009. Scholars in Swahili studies have already examined many aspects of the Swahili Hamziyya. Their publications include critical analysis of the poem,7 overviews of the text and its vocabulary,8 commentaries9 and works on 3 Many thanks to the Director of Kiswahili Studies, UDSM, Dr E. S. Mosha, the deputy, Dr M. M. Hans, the librarian Ms Lavena, my host Sayyid Ahmad Mwinyibaba, ʿAydarūs Bahasani and all those whom I have not mentioned by name, but who assisted me in getting access to Ms. 541 (D) at the UDSM library during my fieldwork in Dar es Salaam, Tanzania in 2015. 4 Refer to Samsom 2015, 263–265 for more background information on this manuscript. The same manuscript is also mentioned in Samsom 2016, 49–51. Drs Ridder Samsom took the pictures of the manuscript in 2011, which was made accessible to him by courtesy of Sayyid Aḥmad Badawy b. Sayyid Muḥammad al-Ḥussainy (d.2012) and his wife Bi Tuma Shee, Mombasa, Kenya. I am very grateful to the al-Hussainy family and Ridder Samsom for their assistance and for photographing the manuscript. I would also like to express my sincere gratitude to the DFG, SFB 950 and the C07 project for their generosity in funding my research work and staging the workshop that made this article possible. Many thanks to the University of Hamburg, CSMC, Hamburg, Germany, Director Prof. Michael Friedrich, the dean of the Graduate school, Prof. Oliver Huck, the organisers of the workshop, Prof. Roland Kießling (my supervisor), Prof. Alessandro Gori (my co-supervisor/co-presenter), Prof. Clarissa Vierke and drs Ridder Samsom for their continuous support of my work in mc NO 17 Swahili manuscript studies. I am also very grateful to Pwani University (PU), Kilifi for its support and granting me leave for my PhD research; many thanks to Prof. Mohamed Rajab, the Vice Chancellor of PU, Kilifi, and the head of the Philosophy and Religious Studies Department, Dr Ali Hemed and all the Pwani University fraternity members who assisted me in so many ways. 5 Déroche 2006,167–184. 6 Déroche 2006, 159–184. 7 Knappert 1968 (partially), Mutiso 2005. 8 Mkelle 1976. 9 Hichens 1936 (partially), Mutiso 2005. manuscript cultures 54 oral performances.10 Other scholars, such as Hichens and Knappert, have attempted to produce an English translation of the text.11 Codicological studies on the QH manuscripts such as articles examining their divergent textual forms and layout are still lacking, however. So far, the textual layout of the Burda, Dalāʾil Ḫayrat and Šifāʾ manuscripts from North Africa has been examined. Frederike Daub’s work12 covers the mise-en-page elements of more than 200 manuscripts and shows how these features reflect the relationship between textual forms and functions of the manuscripts. In her view, poetic canonical manuscripts in the Islamic world such as those containing the Burda may be designed in many different forms, depending on the reason for creating them.13 Small manuscripts of approximately 10 cm by 7 cm are used as talismanic items, for example, since they can easily be put into one’s pocket and carried around wherever one wishes. Larger manuscripts approximately 20 cm by 16 cm in size are designed for ordinary reading, both private and public.14 There are a number of elements that need to be considered when analysing the layout of poetic texts in a manuscript. For instance, the poetic stanzas may be narrowly spaced between each line, arranged in two columns and decorated with symbols such as the intahā (‫)ه‬, a figure which marks the caesura points at the end of each hemistich. Graphical elements of this kind are not only meant to beautify the text, but to facilitate its reading and memorisation. Similar textual layout arrangements in Swahili manuscripts are yet to be examined, especially those in bilingual (Arabic and Swahili) texts that are written in a variety of Arabic scripts. 2.1. Qaṣīda Hamziyya: origin, contents and forms This section examines the origin, contents and forms of the Qaṣīda Hamziyya. According to Abdulaziz, the term qaṣīda (Ar. pl. qaṣāʾid; Sw. kasida, meaning ‘hymn’) specifically came to mean ‘panegyrics eulogising the Prophet’, and also strictly religious poetry which is sung or chanted [on] religious occasions.15 In Swahili, the word adopted for poetry in general and secular poetry in particular is mashairi PARKAR | ARABIC-SWAHILI HAMZIYYA MANUSCRIPTS (derived from Arabic Shiʿr)’. In this paper, the word qaṣīda will be used to mean a panegyric poem (or ode) in praise of the Prophet. Al-Qaṣīda al-Hamziyya fī al-Madāʾīḥī al-Nabawiyya, (‘The Hamza-rhymed Panegyric Poem in Praise of the Prophet’), also known as Umm al-Qurā, or ‘Mother of Villages’, is a panegyric poem in praise of the Prophet Muḥammad. It was originally composed in Arabic by the Egyptian Ṣūfī cleric Šaraf al-Dīn Abu ʿAbdallah Muḥammad bin Saʿīd bin Ḥammād bin Muḥsin bin ʿAbdallah bin alṢanhajī al-Būṣīrī (1212–1294 CE). It has been rendered into Swahili by various scholars, including Šayḫ ʿAydarūs bin ʿUthmān bin ʿAlī bin Šayḫ Abūbakar bin Sālim of Lamu in 1749 CE.16 It is said that at a certain stage of his life, al-Būṣīrī became seriously ill and his doctor was of the opinion that he might not get cured. Thus, al-Būṣīrī composed the Qaṣīda alBurda,17 which has been mentioned before (Daub 2016) and contains 163 verses in all, which are concerned with madḥ, i.e. ‘the praising of the Prophet’, and seeking God’s pardon.18 It is believed that he was mysteriously healed after reciting the Burda and saying prayers to God.19 Due to its madīḥ contents, many Ṣūfī followers associate the Burda with Baraka, ‘blessings’ and healing powers. Hence, the Burda is the most recited and copied ode in the Muslim world.20 Later on, al-Būṣīrī composed the Hamziyya, which seems to be his second most important (and popular) ode among Ṣūfī followers. Al-Būṣīrī’s Hamziyya, which I shall call matn, consists of 456 verses. The term matn in this paper is used to mean the main text, which may appear together with other subsidiary texts such as a translation, commentary, poetic rendition, and/or glosses and colophons. Swahili has a very similar word, matini, which only means ‘text’.21 However, in this article, I prefer to use the term matn to refer specifically to al-Būṣīrī’s Hamziyya (which is an Arabic text). The themes of Al-Būṣīrī’s Hamziyya are the Prophet’s praises, his biography from birth to his migration to Medina 16 Hichens 1936, 2. 17 A poem by Kaʾb is generally believed to be the original Burda poem, while the one by al-Būṣīrī is regarded as the second. See Stetkevych 2010 for more information on the two Burda poems. 10 Olali 2012. 11 Hichens 1936 (partially) and Knappert 1968 (partially). 12 Daub 2016. 18 Aslan 2008, 77. 13 Daub 2016, 42. 19 Aslan 2008, Schimmel 1985, Stetkevych 2010, Daub 2016. Daub 2016, 73. 20 Sperl and Shackle 1996, 85. Also see Daub 2016. Abdulaziz 1995, 152. 21 Mohamed 2011, 438. 14 15 manuscript cultures mc NO 17 55 PARKAR | ARABIC-SWAHILI HAMZIYYA MANUSCRIPTS (verses 1–100), his habits, miracles and virtues of the Qurʾān (101–198), resistance of the Jews and Christians and their hostility towards Islam (199–252), criticism about the hypocrites of Medina and the unbelievers of Mecca, the poet’s desire to visit Medina (253–280), his desert journey, praise for the Prophet and his family household (281–325), and, finally, the poet’s lamentations and seeking of the Prophet’s intercession for his (the poet’s) shortcomings (326–456). The Swahili Hamziyya, which I shall refer to as tarjama, is the poetic rendition of the Arabic Hamziyya. Tarjama is used to mean a translation.22 The Swahili also have the words tarjumi or tarjuma23 and tafsiri, which may be used interchangeably to mean either a translation or an explanation of a given text.24 In Arabic, the word tafsīr means ‘explanation’, ‘interpretation’, ‘commentary, especially of the Qurʾān’.25 To avoid any confusion of the two Swahili words tarjumi and tafsiri, I prefer to employ the Arabic term tarjama here, which is simply a Swahili poetic rendition of ʿAydarūs on al-Būṣīrī’s Hamziyya in this paper. In Swahililand, the QH manuscripts are written in divergent forms. This is evident from the data identified in the fieldwork in East Africa conducted as part of SFB 950’s C07 project. The identified corpus, which consists of 16 Qaṣīda Hamziyya manuscripts, shows that the manuscripts vary in many ways in terms of their language representations and textual units; they may contain texts that are written in monolingual (Swahili) or bilingual (Arabic and Swahili) form, for example. The narration style of the texts can be found in prose as well as in poetry and is a mixture of poetic and prose elucidation. A detailed explanation of the forms of each of the 16 QH manuscripts discussed here is beyond the scope of this paper. However, I shall mention a few cases in order to give the reader a rough idea about these divergent features.26 Two models may be observed among the monolingual ones, for instance: i) the manuscript QA, QAYYIM, NABAHANY: MSA, 00127 only contains a 22 Cowan 1976, 93. 23 Sheikh Ali Muhsin al-Barwani (1995) uses the word tarjuma in Swahili to mean ‘a translation’, i.e. the translation of a tafsīr (commentary) of the Qurʾān in Swahili. 24 Mohamed 2011, 721 and 735. 25 Cowan 1976, 713. 26 The images of the 16 QH manuscripts are preserved at CSMC’s digital repository in Hamburg, Germany. 27 I have coded the manuscript QA, QAYYIM, NABAHANY: MSA, 001, based on the name of the copyist, who was Qayyim bin Amfar Bani Shahardin. The manuscript contains the Swahili Hamziyya, written in Arabic mc NO 17 tarjama (a Swahili translated text), and its poetic lines are narrowly spaced and arranged in two columns; and ii) the manuscript NC, NABAHANY: MSA 00228 contains two texts: a tarjama plus a commentary about it (which is in prose). Both texts are arranged in a single column, and their lines are narrowly spaced.29 Bilingual manuscripts may contain complex combinations of Arabic and Swahili texts. To give the reader a rough idea about them here, the manuscripts and their contents may be summarised as follows: i) Manuscript L30 contains the matn in Arabic, which is interlinearised with the tarjama in Swahili. Both texts are arranged in two columns, and their poetic lines are narrowly spaced. ii) Manuscript D is a five-stanza (taḫmīs)31 poem in Arabic which is interlinearised with the tarjama in Swahili. In this case, the additional texts are centralised and placed before the matn’s, which are in double columns. The poetic lines are also narrowly spaced. iii) Manuscript M only contains the matn in Arabic, the poetic lines of which are arranged in double columns, but are widely spaced to allow annotations to be made (those added are in Arabic and Swahili). iv) Manuscript BE, BEYDH: MAMBRUI 00132 contains a list of words in Arabic and Swahili33 based on the Hamziyya vocabulary. script. Ridder Samsom photographed it in Ahmed Sheikh Nabahany’s private collection in Mombasa in 2013. 28 I have also coded the manuscript NC, NABAHANY: MSA 002, where ‘NC’ stands for ‘Nabahany’s commentary’. The manuscript was in Nabahany’s private collection and so I added the number 002. Ridder Samsom photographed it in Mombasa in 2013. 29 The two monolingual texts in (ii) are only in Swahili and, unlike the aforementioned ones, are written in Latin script. The rest of the texts are written in Arabic script. 30 Approximately half of the 16 QH manuscripts accessed in Swahililand are designed like manuscript L. The older manuscripts such as manuscript KA, KAME: NDAU, 001 are among them. This manuscript’s colophon date is 1863 CE. I coded the manuscript ‘KA’ based on the name of the custodian, Ustadh ʿAli Lali Kame, a madrassa teacher from Ndau Island, Lamu, Kenya. The manuscript contains the matn interlinearised with ʿAydarūs’ tarjama. I photographed it on Lamu Island in 2017. 31 See section 3.1 for further elaboration of the term taḫmīs. 32 I coded the manuscript BE, BEYDH: MAMBRUI 001 based on its author’s name, Sayyid Muhammad bin al-Sharif Salim Al-Beydh (also spelt al-Baydh). Ahmad Badawy, who was working on phase I of the SFB 950 C07 project at the time, photographed the manuscript in Mambrui, Kenya in 2011. 33 The folios of the manuscripts containing the word-list are divided into two columns. One column contains Arabic words obtained from the Hamziyya vocabulary and the other one contains their Swahili equivalents. The matn and tarjama verses do not appear in the manuscript. manuscript cultures 56 All of the above-mentioned bilingual manuscripts (i–iv) are written in Arabic script. In order to provide a broader picture and examples of such textual and language variations, I have chosen three manuscripts – L, D and M – which I shall elaborate on in this paper. Meanwhile, to understand how and why the Swahili wrote their texts in Arabic and Swahili, we need to briefly examine the emergence of Ṣūfī traditions on the East African coast and try to show why such texts were highly regarded in Swahililand. 2.2. The emergence of Ṣūfī traditions on the East African coast The mutual contact between Arabs, Persians, Asians and Swahili via the Indian Ocean trade routes from Somalia to Madagascar lasted over a millennium and has influenced the lives of the Swahili people in many ways, especially in terms of their religious culture.34 Ṣūfī scholars from Hadhramawt established a cultural and religious network between the Arabian Peninsula, Asia and the East African coast as early as the fifteenth century. A number of Arab traders, and specifically the Banī ʿAlawy who adhere to the Al-Alawiyya order, came to settle in East Africa. They were largely involved in the spread of Islam, establishing mosques and Islamic schools wherever they settled in the region.35 Reading, writing and learning the basics of the Arabic language took place for centuries in these institutions.36 In the eighteenth and nineteenth centuries, the religious elites led by sāda (sing. sayyid), the descendants of the Prophet, and šuyūḫ, ‘elders’, had a great impact on the development of religious qaṣīda texts. They highly revered some canonical Islamic poetical texts in Swahili, including the Burda, Hamziyya, Banāt Suʾād and Tabāraka, among others.37 The poetic texts are concerned with seeking baraka (‘blessings’), shufaa (‘healing powers’) and msamaha (‘God’s pardon’) by eulogising the Prophet of Islam.38 Chanting kasida plays a central role in the lives of certain Swahili Sufi tarīqa (‘order’) adherents in East Africa, such as the Banī ʿAlawiyya. The Hamziyya, for instance, is chanted in its Arabic form as well as in its Swahili version by the PARKAR | ARABIC-SWAHILI HAMZIYYA MANUSCRIPTS Mabingwa wa Pate (‘the masters of Pate’)39 on a number of occasions, such as on visiting the grave of Ḥabīb Ṣāleḥ (1844– 1935 CE),40 the mawlid celebration in the Lamu archipelago marking the birth of the Prophet, and sometimes during the delivery of a baby41 and marriage ceremonies.42 The Burda and Hamziyya odes are also recited (in Arabic) prior to the darsa (pl. durūs, ‘lecture’) on Qurʾān tafsīr ‘exegesis’ during the month of Ramadan, in the Riyadha Mosque in Lamu and in the Masjid Anisa in Mombasa, Kenya. Apart from its chanting in a ceremonial context, the Hamziyya is also used in academic circles: it is recited and taught at the Halqa (‘educational forums’) in mosques and during lessons at Islamic higher institutions of learning, where the subject of Arabic literature is offered in the curriculum.43 The teachers present the Hamziyya text in Arabic and briefly explain its Arabic words in modern Swahili. Ṣūfī scholars from Hadhramawt contributed to the development of Islamic scholarship, oral tradition and the production of manuscripts in East Africa. Some of them wrote Swahili in Arabic script and translated a number of Islamic poetic canonical texts into Swahili. These scholars wanted the texts to be understood by those who were unfamiliar with the Arabic language, but could read the Arabic script. Kasida are still sung and chanted at many Swahili gatherings today. 3. The manuscripts 3.1. The description of the Hamziyya manuscripts This section briefly outlines the physical descriptions and the texts of the three QH manuscripts in their respective categories: type A: manuscript L; type B: manuscript D; and type C: manuscript M. Type A: manuscript L Manuscript L contains 82 folios, is 230 mm by 170 mm in size and is bound in a red cardboard jacket. Its colophon date is 14 Ḏū-al-ḥijja 1207 AH / 23 July 1793 CE44 and the copyist was ʿUthmān bin al-Qāḍī bin Mbwarahaji bin 39 Ḥabīb Ṣāleḥ Jamal al-Layl was the founder of the Riyadha Mosque College, Lamu, Kenya. 41 Personal communication with Mwalimu Dini, Pate, 2015. Trimingham 1964, Olali 2012. 34 Knappert 1979, Abdulaziz 1995 and 1996, Bang 2013. 42 35 Trimingham 1964. 43 36 Bakari and Yahya 1995. 37 Knappert 1968 and 1971. 38 Trimingham 1964. manuscript cultures Olali 2012. 40 Personal communication with Sheikh Bahsan (Mombasa, 2014) and Sheikh Hadi (Lamu, 2015). 44 1793 CE, the date of manuscript L, is disputed by some scholars of Swahili; Hichens (1936, 5) confirmed the date, but Knappert (1968, 55) is of the opinion that the manuscript is fifty years younger. mc NO 17 57 PARKAR | ARABIC-SWAHILI HAMZIYYA MANUSCRIPTS Matn line Rhyme of hamza Tarjama line Rhyme of mīm Fig. 1: London, SOAS Library, shelf mark 53823 (manuscript L ), fol. 25. al-Marḥūm al-Fazī.45 The manuscript contains Hichens’ and Hinawy’s notes (in English), which introduce the ode (fols 1–14), the matn (in Arabic, fols 15–80) and ʿAydarūs’s tarjama (in Swahili, fols 15–80) plus paratexts; the colophon (in Arabic) (fol. 80) and an endowment statement (in Arabic, fol. 81). The term ‘paratext’ refers to the threshold or liminal devices and conventions employed in a written work, such as the titles, forewords, notes and epilogues that are used to mediate a book to its readers.46 Recently, several researchers have redefined the term and widened its scope to include a variety of components, such as orality, films, digital media, glosses, glossaries and even subscriptions.47 In this paper, the term ‘paratexts’ will be used for titles, opening statements, waqfiyya (‘endowment’) and tamalluk (‘ownership’) statements, glosses and colophons. The arrangement of the poetic lines is in a specific form. As shown in figure 1, the matn text is interlinearised with the tarjama, whereby both texts run throughout the poem in an alternating stanza-wise fashion. The poetic lines are narrowly spaced and each matn stanza is strategically placed before the Swahili one. Moreover, the characters of the matn 45 The words in italics are titles and are given here as they are scribed in the manuscript’s colophon itself. Al-Qāḍī means ‘the Muslim judge’ and al-Marḥūm means ‘the deceased one (and may he be blessed)’ 46 Genette 1987, xviii. 47 Watson 2010, Ciotti and Lin 2016, vii. mc NO 17 are larger and bolder compared to the tarjama’s counterparts. The matn contains two hemistichs per stanza and so does the tarjama. Each hemistich has been put in its ‘own’ column in the two poems. These columns are regarded as ‘pseudocolumns’, however (Daub 2016), because the reader is not required to read each column independently, unlike modern books or newspapers that contain texts in double columns. In other words, one should read the first hemistich in the right-hand column and then read the second hemistich in the next one (i.e. the left-hand column) (Fig. 1). This procedure should then be repeated for each subsequent stanza. The QH’s bilingual interlinear layout seems to be the original form as scribed by ʿAydarūs himself. The evidence we have so far to back up this argument is based on the layout arrangement of manuscript L, which, as mentioned earlier, is the earliest known preserved bilingual Hamziyya manuscript in Swahili manuscript culture. The Swahili poetic rendition itself was probably done considerably earlier, however, namely in 1749 CE.48 Apart from the layout system of the QH’s poetic lines, there are a number of components that ought to be considered if we want to get a clear picture of how a QH bilingual manuscript is organised. The order of these components is as follows: 48 Hichens 1939, 19. Knappert, 1968, 55 has even argued that the date of translation was 1652. manuscript cultures 58 1) the title, Hamziyya, 2) basmallah, the Islamic opening formula,49 3) eight prologue verses composed by ʿAydarūs, 5) a bilingual interlinear section, namely the matn interlinearised with the tarjama in a stanza-wise fashion, 6) an epilogue of seventeen verses composed by ʿAydarūs’ and a copyist’s colophon. Looking at the organisation of the components of the old QH manuscripts, such as L, it is very likely that, at its initial stage, the layout of matn and tarjama verses was sandwiched between the prologue and epilogue verses. Otherwise, what could have been the reason for ʿAydarūs to invent his own new prologue and epilogue verses? The rhyming schemes of the two poems in manuscript L are ab for the matn and xy for the tarjama, whereas b and y are constant throughout the poems (Fig. 1). As the figure shows, b ends with hamza, a ‘glottal stop’ symbol, and y ends with the rhyming syllable ma. In Arabic prosodic terms, the matn is usually known by its rhyme scheme. In this case, it has been given the title of Hamziyya due to the fact that it is the character for hamza. We may also wish to call the Swahili version ‘mimmiya’ 50 after the final rhyming character mīm. However, in Swahili prosodic terms, the usage of a mid-rhyme or a master rhyme (kina cha bahari) throughout the poem may not determine the name of a poem; a utenzi (‘epic poem’) may have a master rhyme of ma, for instance, but it will always be called a utenzi and not a mimmiya. In the two testimonies in question (i.e. manuscripts D and M; see section 3.2. and 3.3), new forms of texts emerge. Various new textual layout arrangements exist, such as creating more space between the poetic lines to allow annotations to be seen when compared to the layout form of manuscript L. Type B: manuscript D Manuscript D is made of 148 folios and measures 220 mm by 170 mm. The folios are unbound and the manuscript does not have a cover. Currently, its loose leaves are kept in a brown paper envelope at the University of Dar es Salaam. As the colophon page is missing, we have no information on the copyist or scribe or on the date of its production. PARKAR | ARABIC-SWAHILI HAMZIYYA MANUSCRIPTS As a result, I can only guess that it is a twentieth-century work (probably from the 1930s) as the manuscript’s leaves do not seem to be very old. The manuscript was originally obtained from ʿAfuā binti Ḥassanī, Siyu, in 1966 CE.51 The contents of the codex include the Taḫmīs al-Hamziyya by al-Būṣīrī and the tarjama verses by ʿAydarūs in Swahili. The term taḫmīs in Arabic prosody means ‘to make five’, i.e. expand a well-known qaṣīda by adding three taḫmīs lines/ hemistichs prior to the original two hemistichs (of a matn) to make a five-line poem.52 In Arabic prosody, a different author usually composes the taḫmīs lines.53 In this article, the new additional stanzas shall be referred to as the three taḫmīs lines and the term ‘taḫmīs poem’ will generally be used to refer to the Taḫmīs al-Hamziyya that al-Būṣīrī authored. Section 3.2 contains further details of the taḫmīs layout system and illustrates it. I have not used the Swahili term takhmisa or Utano here because, firstly, the taḫmīs alHamziyya poem I am referring to (in manuscript D) is in Arabic and it is technically known by this name in Arabic prosody. Secondly, the Swahili word takhmisa means any poem that contains five lines, which can sometimes adopt middle and end rhymes. Thirdly, a Swahili takhmisa poem is usually composed by one person, unlike an Arabic one.54 Manuscript D contains three types of texts, which were composed by different authors at different times: i) the three taḫmīs (Arabic) lines, which were composed by ʿAbd al-Bāqī bin Sulaymān al-Fārūqī (d. 1861), ii) the matn (Arabic) lines by al-Būṣīrī (d. 1294), and iii) the tarjama (Swahili) lines added by ʿAydarūs (in 1749 CE). In the codex, the taḫmīs text appears up to verse 49. From verse 50 up to the end of it (verse 456), only the matn and its tarjama have been copied. The copyist did not complete the taḫmīs lines for some reason. Type C: manuscript M Manuscript M contains 79 folios and is 270 mm by 175 mm in size. It is bound in a cardboard material, which is wrapped in a brown leather jacket. The colophon on fol. 79 dates the 51 Allen 1970, 32. 52 Abdulaziz 1979, 56–57. 53 The Islamic opening formula states: ‘In the name of God, the Merciful, the Compassionate’. Time of composition and place of the two combined poems may differ. The composer of taḫmīs lines may be a person of a current generation or may be located in another geographical region or nation. 50 54 49 Hichens 1936, Olali 2012. manuscript cultures King’ei and Kemoli 2001, 20. mc NO 17 59 PARKAR | ARABIC-SWAHILI HAMZIYYA MANUSCRIPTS Taḫmīs lines Taḫmīs lines Matn line Tarjama (AV) line Matn line Tarjama (AV) line Taḫmīs lines Intahā symbols Matn line Tarjama (AV) line Fig. 2: University of Dar es Salaam, Dr. Chagula Library (manuscript D ), fol. 3. Fig. 3: An illustration of a taḫmīs model showing the combination of texts and rhyme schemes in manuscript D. manuscript to 7 Šaḥr al-Qaʿad, 1311 AH / 14 May 1894 CE and the name of the copyist is Abī Bakr bin Marhūm Sulṭān Aḥmad. The custodian was Sayyid Ahmad Badawy al-Hussainy, Mombasa. After Sayyid Ahmad’s death, the manuscript was removed from his residence. I tried to locate it during my fieldwork in 2015, but I was unable to trace it. The contents are the matn (in Arabic) plus ample annotations (roughly 30% in Arabic and 70% in Swahili). Manuscript M contains one folio that is not annotated (fol. 7), of which an annotated one also exists, and one folio (fol. 6) that has only been lightly annotated, which also exists as a more heavily annotated version. These doublets show that there were at least two separate manuscripts, some folios from one of them being inserted into the codex. is positioned prior to the tarjama line56 and (iii) the tarjama line follows (Fig. 2). The total number of lines for the three combined texts is therefore seven (three for the taḫmīs, two for the matn and two for the tarjama). When the taḫmīs lines are united with the matn lines, a new rhyming scheme is created: the final syllables of the three lines (a or b) rhyme with the final syllable of the matn’s first hemistich (a or b). As Fig. 3 shows, the rhyming scheme of the five lines is aaaaxfy, bbbbxgy, ccccxhy and so on, whereas the rhyming scheme of the three additional taḫmīs lines is aaa, the matn’s is ax and that of the tarjama is fy. Fig. 3 also shows that the alternating rhymes (x and y) remain constant throughout the poem, while the rhyme of the taḫmīs lines keeps on changing due to the last syllables of the first hemistich of the matn. Apart from the new rhyming scheme, the layout of the poetic lines has also been redesigned. The first stanza of the taḫmīs poem can be illustrated as follows: 3.2. Analysis of manuscript D: the taḫmīs and tarjama lines In Arabic prosody, the addition of extra lines in a given poem is known as tasmīt or musammat, literally defined as ‘the missing link’. Technically speaking, the term tasmīt implies an insertion of new hemistichs: two or more into the already well-known qaṣīda, which rhyme with it.55 Manuscript D contains the taḫmīs poem (with a five-hemistich structure), the last line of which is interlinearised with ʿAydarūs’s tarjama (Fig. 2). A close examination of manuscript D reveals a complex combination of the textual layout system, i.e. (i) the taḫmīs lines are positioned prior to the matn lines, (ii) the matn line Li-ʿulā al-rusuli ʿan ʿulāka anṭiwāu (i) Wa-ūlū al-ʿazmi taḥta šāʾwika jāʾu (ii) Wa-li-murqāka dānati al-aṣfiyāʾu (iii) Kaifa tarqā ruqiyyaka al-anbiyāu (iv) yasamāʾ mā ṭāwalatʾhā samāu (v) Hali wakwelaye kukwelako mitume yonte (vi) Uwingu usiyo kulotewa ni moja sama (vii) 56 55 Vierke 2009, 52. mc NO 17 The term ‘double columns’ can also be used instead of bi-colons, see Daub 2016, 46. manuscript cultures 60 PARKAR | ARABIC-SWAHILI HAMZIYYA MANUSCRIPTS Caesura marks i) The high-ranked angels have congregated for you in the high heavens [, O Muḥammad!] ii) And the arch-prophets came along [too and are] just beneath them iii) And your utmost position [above them all] has been reserved iv) How can [the other] prophets reach your highness, [O Muḥammad!], v) [by your ascension] to heaven, with which no one vies?57 Moreover, the taḫmīs lines are positioned in a single column in the middle of the folio, while the matn and tarjama lines maintain their two-columnar system (see Fig. 3). All the poetic lines are narrowly spaced. An intahā symbol ( ‫ ) ه‬is positioned in the middle of the matn and tarjama lines (Figs 2 and 5). It can be argued here that the layout arrangement of a manuscript may reflect its functions. Manuscript D was probably used as a chanting and reading aid on account of its unique textual organisation. The layout of its single and bi-colon system helps the reader or memoriser of the text to follow every line’s starting and end points due to the wellorganised rhyme schemes and arrangements of the columns. In other words, the trained reader can easily identify the taḫmīs, matn and tarjama line at a glance. 3.3. Analysis of manuscript M: the matn and its paratexts Manuscript M is a good example of a bilingual Hamziyya manuscript in Arabic and Swahili produced in a different form in Swahililand. The manuscript contains the matn as well as paratexts, but does not have a tarjama and thus differs from manuscript L’s layout in some other bilingual interlinear manuscripts. Looking closely at manuscript M, it becomes apparent that several different types of paratexts exist: the title, AlHamziyya (in Arabic), the tamalluk (‘ownership’ statement, in Arabic), the waqfiyya (‘endowment’ statement, in Arabic), annotations (in Arabic and Swahili) and the colophon (in Arabic). The tamalluk statement simply mentions the name of the owner of the manuscript along with the date and place of its acquisition, as in Hāḏā al-kitāb milk li-faqīr Allāhi taʿālā, ‘This book is owned by a poor servant of God, ‘The Most High’, Muḥammad bin sayyid Abdallāh bin sayyid Fig. 4: Mombasa, private collection of Sayyid Ahmad Badawy b. Muhammad al-Hussainy’s family and Bi Tuma Shee (manuscript M), fol. 22. Fig. 5: University of Dar es Salaam, Dr. Chagula Library (manuscript D ), fol. 14. Ḥasan al-Bāʿlawī al-Masalī.58 I will not provide any more details of the paratexts in manuscript M here, but shall restrict myself mainly to explaining the forms of the textual layout features of the matn and its glosses. Manuscript M contains matn stanzas, which are widely spaced (approximately 3.5 cm apart) and contain ample annotations in Arabic and Swahili. The matn is written in nasḫ ‘vocalised’ style and in bold black ink. Its characters are bigger, while the characters of the annotations are smaller and thinner. A small, flowerlike figure with an inverted comma in red ink on top of it has been inserted after each hemistich to mark the caesura of each stanza (Fig. 4). Manuscript M’s Swahili annotations are more numerous than the Arabic ones (especially from fols 8 to 65) and their proportion is roughly 25 to 75 per cent of the Arabic notes. The layout of the annotations is a haphazard one: they are scribed horizontally, marginally and diagonally and in the blank spaces beside the matn lines in particular folios. This type of glossing is found in Arabic-Islamic and other manuscript cultures, such as the Hausa and the Malay ones.59 I cannot tell whether the same person who wrote the matn also wrote the annotations; this is subject to further scientific verification and may require ink analysis and computer software to study the types of handwriting involved, which is beyond the scope of this paper. 58 Wādī Masīla is probably a village in Yemen. The tamalluk is in Arabic; the date is not provided, but it may have been written around 1894 as the handwriting of the tamalluk resembles the one used for the annotations. 57 59 Unless otherwise stated, the translation is provided by the researcher. manuscript cultures Gacek 2009. mc NO 17 61 PARKAR | ARABIC-SWAHILI HAMZIYYA MANUSCRIPTS Matn line Annotations in Arabic Annotations in Swahili Fig. 6: Mombasa, private collection of Sayyid Ahmad Badawy b. Muhammad al-Hussainy’s family and Bi Tuma Shee (manuscript M), fol. 19. 4. Conclusion The annotations in Arabic and Swahili give the meaning of selected words from the matn verbatim. For instance, the Arabic words of the matn that are annotated in Swahili are Ar. lam yusāūka = Sw. Hawalingani nawe ‘they are not equal to you’ (fol. 1), ḥamamah = njiwa ‘dove’ (fol. 12), and wa dumuʿī = matozi ‘tears’; standard Sw. = machozi (fol. 42). The Swahili words are usually in nasḫ vocalised script. The Arabic words of the matn that are annotated in Arabic can be exemplified as follows: lam yusāūka = ay fī sharafuk ‘that means in your noble status’ (fol. 1), fa atat-hu = ay jāt ilayhi ‘that means she came to him’ (fol. 6) and warqāu = ay abyaḍ launūhu wa aswad ‘that means it [the dove] had black and white spots’ (fol. 12). The Arabic annotations are in riqaʿ script. The main text is usually written first. The scribe purposely leaves wide spaces between the lines for the annotations to be filled at a later stage by either a student or a teacher. The riqaʿ script gives us a clue that the scribe is either an advanced student in Islamic studies or a teacher. In East Africa, even today, this type of handwriting is usually taught to advanced learners. Manuscript M was probably used for teaching and learning as it contains widely spaced stanzas and a large number of annotations that explain the difficult Arabic words in simple Swahili (see Fig. 4). It is hard to say whether the manuscript was used by a scholar to decipher the Arabic text as preparation for giving a lecture (darsa) in a mosque or whether a student used it to insert his lecturer’s remarks during lessons. All in all, the manuscript indicates that it was used for academic activity in Swahililand. mc NO 17 This article provides some background information about the emergence of Islamic scholarship and Swahili manuscript tradition with specific reference to Qaṣīda Hamziyya manuscripts. One important feature of these manuscripts is their divergent visual organisation. My paper also shows that a single text in Swahili society can be written in various ways at a given time and place. For example, manuscript M (from Mombasa) and manuscript D (from Siyu, now preserved at UDSM, Dar es Salaam), which were written in the nineteenth and twentieth century respectively, have divergent textual features. Manuscript L (from Pate, now preserved at SOAS, London) was written in the eighteenth century and contains the matn (in Arabic) and tarjama (in Swahili), while manuscript D contains the taḫmīs (in Arabic) and the tarjama (in Swahili). Both texts are narrowly spaced. Manuscript M contains the matn (in Arabic) and annotations (in Arabic and Swahili). Manuscript L (which is the oldest one of them all) has its own kind of textual layout: it contains a prologue (in Swahili), the matn (in Arabic), the tarjama (in Swahili) and an epilogue (in Swahili). Manuscripts L and D indicate that they were probably used as an aid for chanting and private reading. This is because their textual layout employs the double-column system and is narrowly spaced, features that show the manuscripts were mainly used for the preparation of oral performances. Manuscript M was probably used for teaching and learning purposes because it contains the matn, which is widely spaced to allow plenty of annotations to be made, which are arranged haphazardly between the poetical stanzas and along the margins of the main text. manuscript cultures 62 PARKAR | ARABIC-SWAHILI HAMZIYYA MANUSCRIPTS 5. APPENDIX: ARABIC-LATIN TRANSLITERATION TABLE REFERENCES Arabic symbol Transliteration Arabic symbol Transliteration ‫ا‬ a ‫ض‬ ḍ (dhw) ‫ب‬ b ‫ط‬ ṭ ‫ت‬ t ‫ظ‬ ẓ (dh) ‫ث‬ th ‫ع‬ ʿ ‫ج‬ j ‫غ‬ gh ‫ح‬ ḥ ‫ف‬ f ‫خ‬ ḫ (kh) ‫ك‬ k ‫د‬ d ‫ق‬ q ‫ذ‬ dh ‫ل‬ l ‫ر‬ r ‫م‬ m ‫ز‬ z ‫ن‬ n ‫س‬ s ‫و‬ w ‫ش‬ š (sh) ‫ه‬ h ‫ص‬ ṣ (sw) ‫ي‬ y Abdulaziz, Mohamed H. (1979), Muyaka: 19th Century Swahili Popular Poetry, Nairobi: Kenya Literature Bureau. —— (1995), ‘Impact of Islam on the Development of Swahili Culture’, in Mohammed Bakari and Saad Yahya (1995), Islam in Kenya, Mombasa: MEWA, 142–157. —— (1996), ‘The Influence of the Qasida on the Development of Swahili Rhymed and Metered Verse’, in Stefan Sperl and Christopher Shackle (eds) (1996), Qasida Poetry in Islamic Asia and Africa, Leiden: Brill, 411–428. Allen, J. W. T. (1970), The Swahili and Arabic Manuscripts and Tapes in the Library of the University College of Dares-Salaam – A Catalogue, Leiden: Brill. Aslan, Rose (2008), Understanding the Poem of Burdah in Sufi Commentaries, MA thesis, The American University in Cairo < http://dar.aucegypt.edu/handle/10526/1334 > (accessed on 17 March 2018). Bakari, Mohammed and Saad Yahya (eds) (1995), Islam in Kenya, Mombasa: MEWA. Bang, Anne (2013), Sufi and Scholars of the Sea: Family Networks in East Africa, 1860–1925, Oxon: Routledge Curzon Group. Biersteker, Ann and Mark Plane (1989), ‘Swahili Manuscripts and the Study of Swahili Literature’, Research in African Literatures, 20: 449–472. Ciotti, Giovanni and Hang Lin (eds) (2016), Tracing Manuscripts in Time and Space through Paratexts, Berlin: De Gruyter. Cowan, J. Milton (ed.) (1976), The Hans Wehr Dictionary of Modern Written Arabic, New York: Spoken Language Services, Inc. manuscript cultures mc NO 17 PARKAR | ARABIC-SWAHILI HAMZIYYA MANUSCRIPTS 63 Daub, Frederike-Wiebke (2016), Formen und Funktionen des Layouts in arabischen Manuskripten anhand von Abschriften religiöser Texte: al-Būṣīrīs Burda, alǦazūlīs Dalāʾl und die Šifāʾ von Qāḍī ʿIyāḍ, Wiesbaden: Harrassowitz Verlag. Mutiso, Kineene wa (2005), Utenzi wa Hamziyyah, Dar es Salaam: Institute of Swahili Research. Déroche, François et al. (2006), Islamic Codicology: An Introduction to the Study of Manuscripts in Arabic Script, London: Al-Furqān Islamic Heritage Foundation. Samsom, Ridder H. (2015), ‘Swahili Manuscripts: Looking in East African Collections for Swahili Manuscripts in Arabic Script’, in Hannelore Vögele, Uta Reuster-Jahn, Raimund Kastenholz and Lutz Diegner (eds), From the Tana River to Lake Chad: Research in African Oratures and Literatures, in memoriam Thomas Geider, Cologne: Rüdiger Köppe Verlag, 243–284. Gacek, Adam (2009), Arabic Manuscripts – A Vademecum for Readers, Leiden: Brill. Gennete, Gerard (1997), Paratexts: Thresholds of Interpretation, tr. Jane E. Lewin, Cambridge: University Press. Hichens, William (1936), ‘Notes on Hamziyya’, notes attached to Ms. 53823, SOAS, London, UK, published online in SOAS Special Collections <https://digital.soas. ac.uk/LOAA000083/00001/1x>. Olali, Tom (2012), Performing Arts in Lamu: Maulidi Festival, Saarbrücken: LAP Lambert. —— (2016), ‘A Case Study of Transfer of Islamic Knowledge in Swahili Manuscripts in Arabic Script: A Tafsir by Ali Hemed Abdallah Said Abdallah Masudi alBuhry (1889–1957 CE)’, in Wafik Kelliny (ed.), Research and Surveys in National and International Archives, vol. 6, Muscat: National Records & Archives Authority, 31–56. —— (1939), Al-Inkishafi: The Soul’s Awakening, Dar es Salaam: Oxford University Press [repr. 1972]. Schimmel, Annemarie (1985), And Muhammad Is His Messenger: The Veneration of the Prophet in Islamic Piety, Chapel Hill / London: University of North Carolina. King’ei, Kitula and Amata Kemoli (2001), Taaluma ya Ushairi, Nairobi: Acacia Stantex Publishers. Sperl, Stefan and Christopher Shackle (eds) (1996), Qasida Poetry in Islamic Asia and Africa, Leiden: Brill. Knappert, Jan (1968), ‘The Hamziyyah Deciphered’, African Language Studies, 9: 52–81. Stetkevych, Suzanne (2010), The Mantle Odes, Bloomington: Indiana University Press. —— (1971), Swahili Islamic Poetry, 3 vols, Leiden: Brill. Trimingham, J. Spencer (1964), Islam in East Africa, Oxford: Clarendon Press. —— (1979), Four Centuries of Swahili Verse: A Literary History and Anthology, Nairobi: Heinemann. Mkelle, M. Burhan (1976), ‘Hamziyyah – The Oldest Swahili Translation’, Kiswahili, 46/1: 71–75. Mohamed, A. Mohamed (2011), Comprehensive Swahili– English Dictionary, Nairobi: East African Educational Publishers Ltd. mc NO 17 Vierke, Clarissa (2009), On the Poetics of the Utendi, Berlin: LIT Verlag. Watson, Alex (2010), ‘Thirteen Ways of Glossing “To a Haggis”: Disputing the Borders of Robert Burns’ Paratexts’, International Journal of Scottish Literature, 6: 1–24. manuscript cultures 64 PARKAR | ARABIC-SWAHILI HAMZIYYA MANUSCRIPTS PICTURE CREDITS Fig. 1: Image courtesy of SOAS Library, London. Figs 2–3: Photography by Ahmad Parkar, courtesy of University of Dar es Salaam, Dr Wilbert Chagula Library, Dar es Salaam, Tanzania, 2015. Fig. 4: Photography by drs Ridder Samsom, Mombasa, 2011. Fig. 5: Photography by Ahmad Parkar, courtesy of University of Dar es Salaam, Dr Wilbert Chagula Library, Special Collections, Dar es Salaam. Fig. 6: Photography by drs Ridder Samsom, Mombasa, 2011. MANUSCRIPTS Ms. 541, courtesy of the University of Dar es Salaam, Dr Wilbert Chagula Library, Dar es Salaam, Tanzania (manuscript D). Ms. 53823, courtesy of SOAS Library, London, UK (manuscript L). Ms. al-Hussainy, Mombasa, private collection, courtesy of Sayyid Ahmad Badawy b. Muhammad al-Hussainy’s family, Mombasa, Kenya (manuscript M). manuscript cultures mc NO 17 65 RAIA | TRANSMISSION AND TEXTUAL VARIANTS Article A Network of Copies: Transmission and Textual Variants of Manuscript Traditions from the J. W. T. Allen Collection (Dar es Salaam) Annachiara Raia | Leiden, The Netherlands Introduction The aim of this paper is twofold: firstly, to examine what exactly ‘textual practices’ – such as transmission, collecting, copying, transliterating and translating a handwritten text – tell us about variability and adaptation, and secondly, to question the idea of ‘one text – one original archetype’.1 The text I shall refer to is the poem of Yusuf, son of Yaqub, known in Swahili as Utendi wa Yusuf (but also Hadithi ya Yusufu, ‘The story of Yusuf’), Kisa cha Yusufu, ‘The account of Yusuf’ or Utenzi wa kisa cha Nabii Yusuf, ‘The poem of the story of the Prophet Yusuf’). The story was inspired by earlier Muslim texts such as Sura 12 from the Qur’ān and Tha‘labī’s prose text Qiṣaṣ al-Anbiyā’ (‘The Prophets’ stories’), which was originally adapted in the utendi form, an important Swahili poetic genre, and Arabic script in the eighteenth and nineteenth century. While geographically speaking, its manuscripts are all from the northern Swahili coast, the story has also travelled and been re-adapted across the centuries, so DVD copies are not only sold in shops in Lamu and Mombasa now, but in Mozambique as well. Furthermore, Mwalimu Evaristo M. Mahimbi edited the first utenzi adaptation of the story of Yusuf based merely on the biblical account in Tanzania in 1975 CE.2 The beginning of a comparative study of tales about Yusuf from the Swahili region can be found in Ridder Samsom’s dense article from 1 The idea for this paper was conceived on the occasion of the CSMC Workshop ‘One text – Many forms: A comparative view of the variability of Swahili manuscripts’, held in Hamburg in April 2017. On that occasion, I presented and started posing some of the new and necessary questions that had emerged on the verge of my PhD submission and that required further consideration. The presentation, entitled ‘Visually pleasant texts or imperfect copies? The many ways to copy the “Story of Yusuf”’, intended to address the question of the relationships between manuscript copies. My heartfelt thanks to the organisers for having invited me to participate, and to Clarissa Vierke, Ridder Samsom, Abdilatif Abdalla and Antonella Brita for their useful remarks during my presentation and afterwards. 2 Mahimbi 1975. mc NO 17 1997 CE, in which he underlines the widespread nature of the story and points out the far-reaching network of texts and scholarship related to ‘Yusufism’ (Yusufukia) that is so promising: kwa hakika elimu ya Fasihi Linganishi (Comparative Literature) ingeweza kuanzisha elimu peke yake ya ‘Yusuf’, tuseme ‘Yusufukia’. Lakini maudhui yenyewe yanaanzia wapi na kuishia wapi? Maudhui yenyewe ni ya kilimwengu [...] (‘Indeed, comparative literature can initiate an exclusive study about Yusuf, let’s say a Yusuf-ism … However, where does this topic begin and end? This is a worldwide topic …’).3 My focus in this paper will be on the several manuscripts in Arabic script listed under the label of Utendi wa Yusuf in J. W. T. Allen’s catalogue and in the East Africana Section of the Library of the University of Dar es Salaam, which I consulted in September 2016. At the core of the paper, I shall consider variants and versions of the texts, including formal and textual variants as well as incomplete copies of the Utendi wa Yusuf manuscript from Allen’s catalogue from 1970. My interest in variants and versions is imbued and inspired by the concept(s) of textual instability. As pointed out by Judy Quinn and Emily Lethbridge, Zumthor’s concept of mouvance as the ‘mobilité essentielle du texte médiéval’4 has been fundamental to Cerquiglini’s Praise of the Variant.5 Rather than trying to build stemmata, the ‘new’ philologists have begun treating each manuscript as a cultural artefact (and not a deficient aberration of an original version), including the most recent reworkings of earlier material, also placing emphasis on the editing of the whole manuscript. Before I consider the variants here, I shall introduce the reader to Allen’s Swahili 3 Samsom 1997, 91–92. My own translation (AR). 4 Zumthor 1972, 171 quoted in Quinn and Lethbridge 2010, 92. 5 Cerquiglini 1989, engl. transl. 1999. manuscript cultures 66 manuscripts collection in order to delve into Lamuan textual practices, which show that the act of penning a Swahili poem should not be reduced to the hand-made product of a single person, but rather a social practice rooted primarily in people’s own memory and transferred to paper by a network of craftsmen. I will proceed by shedding some light on the entangled Utendi wa Yusuf manuscript traditions listed in J. W. T. Allen’s catalogue and further investigated by myself as a way of re-problematising the transmission and stemmatics of Utendi wa Yusuf manuscripts. 1. The transmission and collection of Swahili manuscripts on the Swahili coast before the -Second World War If texts were written down, they were not intended for publication and reading, but rather for conservation and performance.6 In the nineteenth century, Europeans ‘discovered’ the Swahili manuscript tradition and started to study it. Wellknown authorities on Swahili language and poetry in Arabic script facilitated the awareness, copying and collection of such manuscripts in what were mostly British and German academic circles.7 In the course of collecting material, the British scholar and Anglican priest William Taylor, who spent ten years on the Swahili coast (mostly in Mombasa), came into close contact with reputed Islamic scholars and poets like Mwalimu Sikujua bin Abdallah and Sheikh Muhammed bin Ahmad al-Mambassy.8 The scribe and poet Mwalimu Sikujua (died in 1890) contributed greatly to collecting the poetry of the famous poet Muyaka bin Haji, which the scribe had mostly copied in Arabic script and annotated in the late 1880s for a volume that Taylor planned to publish along with his own transliteration in Latin script and translation. A record of these activities is preserved in a shairi verse by Sikujua himself:9 Kingozi cha K’ongowea chote tulikikusanya; Hawa kukipangalia, na fasiri kuifanya; Maana hamtajia pamoja na kumwonya, Na yeye akanandanya maanaye akapata. RAIA | TRANSMISSION AND TEXTUAL VARIANTS We collected all the Kingozi of Kongowea (Mombasa); Then I began arranging it and interpreting its meaning. I told him (Taylor) what it meant and explained to it to him carefully, And he followed and understood it well. The activity of ‘collecting’ (-kusanya) and interpreting the meaning (-fanya fasiri) of the poetry written in the dialect of Mombasa, formerly known as Kongowea nda mvumo (‘Kongowea the famous’)10 reflects common practices of copying Swahili compositions and translating them for those early scholars who arrived and stayed on the coast.11 Another well-known figure was Muhammad bin Abubakar Kijuma, who also copied manuscripts of the Utendi wa Yusuf several times. Kijuma played a key role in introducing Swahili poetry to European scholars. More frequently than any other scribe at that time, he was commissioned by European missionaries and scholars such as Alice Werner, Ernst Dammann and William Hichens to write and copy poems, which ended up in important European collections in Berlin, Hamburg and London.12 He had a whole network of clients whom he communicated with both in person, like Alice Werner and Ernst Dammann, and through written correspondence, like Carl Meinhof and William Hichens. Alice Werner had visited the coast before World War I broke out and, thanks to the assistance of Muhammad Kijuma in Lamu (and Abu Bakar bin ‘Umar es-Sawiyy in Siu), she came to possess several manuscript copies, on which she based some of the earliest publications of important Swahili poems such as the Utendi wa Mwana Kupona (‘The Poem of Mwana Kupona’), Ayubu (‘Job’) and Hadithi ya Mikidadi na Mayasa (‘The Story of Mikidadi na Mayasa’).13 Like Alice Werner, Ernst Dammann also particularly benefited from Muhammad Kijuma’s consultancy. After his arrival on Lamu in 1936, Dammann undertook research not only on the Utendi wa Tambuka (‘The Poem of Tambuka’), but also on numerous other poems preserved in manuscripts that 10 The W. E. Taylor Collection, which includes Muyaka’s poetry as well as prose and letters, is now stored in the library of the School of Oriental and African Studies in London. 12 6 De Kreij 2015, 18. 7 Vierke 2010, 51–60. 8 Abdulaziz 1979, 4. 9 Abdulaziz 1979, 68. manuscript cultures Sacleux 1939, 437. 11 For an overview, see Miehe and Vierke 2010. 13 Regarding the Ayubu, the poem about Job, Alice Werner was able to buy a manuscript that was ‘an imperfect copy of the Ayubu’– although more complete than Steere’s in terms of the number of stanzas it contained (197 instead of 46). Later on, she also obtained a complete copy of the poem – from Muhammad Kijuma – on which she largely based her own edition (Werner 1921–1923, 85. Also see Miehe and Vierke 2010, 26). mc NO 17 67 RAIA | TRANSMISSION AND TEXTUAL VARIANTS Kijuma had supplied him with or that the German scholar commissioned from him even after he had left the island, such as the Utendi wa Yusuf (‘The Poem of Yusuf’), Utendi wa Mwana Esha (‘The Poem of Lady Aisha’), Kufa kwa Muhammadi (‘Muhamad’s Death’) and the acrostic poem A I U.14 In the case of poems in manuscripts that were in a particularly poor condition, like the Utendi wa Isbani (‘Isbani’s Poem’) (also known as the Utendi wa Qatirifu), personal letters attest to the scribe Kijuma’s efforts to handcopy poems that another scribe had already been paid to copy. 15 his shows some of the intricacies involved in the process of copying, which subsumes the scribe’s wish to emulate a text while imitating it, that is, to render a manuscript a better copy than the previous one. Kijuma hand-copied a manuscript that had already been copied in the past; what the recipient would have received was therefore a ‘copy of the copy’, albeit a more refined copy of a poorer version.16 The sharing of manuscripts and the collaboration between master poets and scribes from the coast point to the existence of a Swahili coastal network of copying and commissioning, comprised of local poets and copyists in collaboration with missionaries and scholars. Copying for different scholars contributed to the wide dissemination of Swahili manuscripts in Europe. Furthermore, at the time, Kijuma’s manuscripts – adapted to European preferences, tastes and reading habits – started to reflect the commercialisation of manuscript production: to a large extent, he earned a living by copying manuscripts.17 In most of the cases where the author of a manuscript was unknown, copies of the anonymous manuscript continued to be copied either by well-known scribes or sometimes even very young assistants. In 1936, for instance, Ernst Dammann’s wife Ruth copied several manuscripts by adopting the same method that primary-school children often use when replicating script: although she was not even acquainted with the Arabic script, she put a piece of transparent paper over the ‘original’ manuscript and copied it by tracing the shape of the Arabic letters with a pencil.18 Some of the copyists were very young, actually: the scribe who copied the Utendi wa Nabii Yusuf (‘The Poem of the Prophet Yusuf’) in 1964,19 for instance, was said to be only 15 years old at that time.20 Before starting to compare variant readings, I would like to highlight John W. T. Allen’s collection and his specific network of people who contributed to the collection of Swahili manuscripts after the Second World War. 1.1 Allen’s catalogue and the Lamuan network: poets, agents, scribes and assistants at work Literary works ‘are fundamentally social rather than personal or psychological products’.21 John W. T. Allen studied in Oxford, joined the British Colonial Service and served in Tanganyika and the Western Aden Protectorate. It was with the East Africana Section of the Library of the University College of Dar es Salaam that he started to build up a huge collection of Swahili manuscripts. Thanks to his wife Winifred Ethel Emma Brooke, he collected and purchased pre-European Swahili manuscripts all along the East African coast and on the Comoro Islands. Apparently, they were a ‘unique and most effective team: they both carried further their study of Swahili, he especially as a trained classical philologist, she as someone who could enter deeply into the lives of the womenfolk who are the guardians of some of the greatest achievements of Swahili civilization and refinement of culture’.22 18 14 The Arab Sayyid Ahmed, known as ‘She’, was another local expert that Kijuma introduced to Ernst Dammann and from whom he obtained manuscripts in exchange for recipes – for cakes! (For more on the acrostic poem, see Miehe and Vierke 2010, 65–68, 82, 94.) 15 Kijuma’s letter 7 (Miehe and Vierke 2010, 89–93). 16 Na thendi inshallah utapata wa Isbani nimepata kwa mthu umerarukararuka; Nimempa mapesa awandike kwa khati yake; kisa mimi ninakili kwa khati; lakini imewekwa sharti name nimwandikie mngine nimekubali – ‘As far as the tendi are concerned – if God wishes – you will get Isbani’s. I got it from someone and it is completely torn. I gave him money so he would write in his own handwriting so I can finally copy it in [my own] handwriting. A condition was set that I should write another one for him; I agreed.’ (Kijuma’s letter 7, in Miehe and Vierke 2010, 89–93). 17 For further criticism on Kijuma’s figure, see Vierke 2010, 41–60 and Abou Egl 1983. mc NO 17 This is the case for the Berlin manuscripts Hs. Or. 9954 and 9955, for instance, which contain Bwana Zahidi Mnugumi’s poems (Miehe 2010). 19 Ms. 603, Allen 1971, 35, 114. 20 Personal communication by Ustadh Mau (February 2018, Lamu) 21 Mcgann 1983, 43–44, quoted in Driscoll 2010, 42–43. 22 As Wilkening has attested, ‘manuscripts were not necessarily bought but mostly borrowed and microfilmed for a honorarium so that families did not have to part with their treasure’ (Wilkening 2000, 240–243). Being surrounded by experts who share a manuscript with ‘other’ scholars, thanks first and foremost to the appreciation of their own literary treasures, is not something that happens very often, especially nowadays when manuscripts’ owners or poets may not want to display or be able to share what is kept in their house by inheritance. This is also the reason why I wanted to shed light on this network of ‘beautiful minds’ and hardworking people without whom the textual criticism of classical tendi would be much more difficult, if not impossible. manuscript cultures 68 The East Africana collection of the University Library of Dar es Salaam contains Kiswahili and Arabic manuscripts. Over 1,180 titles of manuscripts have been microfilmed at the time of writing this article and around 890 Kiswahili and Arabic manuscripts are yet to be labelled, typed and microfiched.23 John W. T. Allen24 started his work as Rockefeller Research Fellow at the University College of Dar es Salaam, Tanzania in 1965 under the project title ‘The Collection of Manuscripts of Literary, Linguistic and Historic Interest’. The result of his three years of cataloguing, editing, translating and publishing some of the huge trove of Swahili manuscripts he was able to access is recorded in his catalogue, which was published in 1970.25 During my short stay in Dar es Salaam in September 2016, a plethora of manuscripts, typescripts and microfilm copies of the Utendi wa Yusufu that emerged from the dusty but tidily arranged and numbered brown paper envelopes were kindly made accessible to me by the librarians. My consultation of Allen’s catalogue in the East Africana section of the University of Dar es Salaam Library proved that there are many versions of the poem which used to be part of private collections.26 Their importance is also reflected in the post-World War II activity of the Swahili manuscript collection.27 Allen’s research started with a first check of those manuscripts that were already part of the collection of the University of Dar es Salaam Library, as Allen himself reports: ‘We spent a month in Dar es Salaam making a check of the manuscripts already in the collection. It was necessary to examine every manuscript to ascertain whether it was a good copy, so as to avoid duplication, or a corrupt copy of which a better one would be valuable’.28 23 University of Dar es Salaam Library: https://library.udsm.ac.tz/index. php/east-africana, accessed on 1 September 2018. 24 John W. T. Allen (1904–1979) mainly gathered Swahili manuscripts on the East African coast in the 1960s. Alice Werner (1859–1935) started collecting even earlier and was followed by Ernst Dammann (1904–2003) and William Hichens (?–1944). For a more detailed discussion of the early British scholars and their successors, see Miehe and Vierke 2010, 18–33. 25 Allen, J. W. T. and Dar-es-Salaam. University College (1970), The Swahili and Arabic Manuscripts and Tapes in the Library of the University College of Dar es Salaam: A Catalogue (Leiden: Brill). 26 On the importance of private collections, also see Varvaro 2012, 49. 27 As attested in Miehe and Vierke‚ the post-war period is characterised by the Swahili Committee’s increasing efforts to secure their poetic heritage. ‘In particular, the Committee’s Journal played a more and more important role in this respect. Here, as well as in the supplements to the Journal and in the later series ‘Johari za Kiswahili’, scholars like Harold E. Lambert (1893–1967), John W. T. Allen (1904–1979), and later on Jan Knappert (1927–2005) made the manuscripts accessible to the public, which were collected and kept in the Committee’s library’ (Miehe and Vierke 2010, 30). 28 Allen 1968, 113. manuscript cultures RAIA | TRANSMISSION AND TEXTUAL VARIANTS However, it was especially on the island of Lamu, and with the co-operation of an exceptional, erudite local network, that he was able to access so much additional manuscript material, amounting to 7,000 microfiche frames in all.29 The initial frustration that Allen and his wife felt after arriving on Lamu in December 1965 disappeared when Ahmed Sheikh Nabahany, the clerical officer of the district of Lamu at that time, introduced them to his grandmother, Amina Mohamed Sheikh.30 She had an unparalleled talent: even when Allen was unable to obtain a reliable copy of a particular manuscript, she was able to retrieve the text from memory. One such example is the ‘copying’ of the Utendi wa Ngamia na Paa (‘The Poem of the Camel and the Gazelle’), which was first published by Dammann in Dichtungen in der Lamu-Mundart des Suaheli (‘Poems in the Swahili Lamu Dialect’) (1940). In Allen’s introduction to his second edition, the author relates the story as follows: ‘We were unable to obtain a reliable manuscript of the poem; but with Dammann’s edition to refresh her memory, Bi. Amina binti Sheikh Nabhany of Lamu was able to reconstruct the complete poem with confidence, remembering it from early childhood, and it is interesting to note how often Dammann notes a probable lacuna which she has been able to fill’.31 This hints at the important role of women in preserving Swahili poetic traditions. The main contacts through whom Allen was able to gather and collect manuscripts were, indeed, elderly women, the custodians of the knowledge and culture of Swahili poetic manuscripts. The master poetess Zaharia binti Maimun, for instance, and a woman called Asiya with whom Winifred (Allen’s wife) was in close contact, used to visit them and provide them with the manuscripts they had ordered along with others that they had not expected to get. In fact, Zaharia binti Maimun became Allen’s agent and made an effort to provide him with manuscripts not only from Lamu, but from Pate as well.32 29 Only the Utendi wa Mikidadi na Mayasa was not obtainable on Lamu. According to Allen, the poem was ‘totally unknown’ there. He was unable to find anyone who had ever heard about it or possessed a copy of it (Allen 1970, 269). The discovery of the Utendi wa Qiyama (‘The Poem on the Judgment’) was another story: its sources came not only from Lamu, but from different areas, from versions ascribed to Saada binti Maawia el Maawy on Lamu, to another witness in the Pemba dialect, ascribed to Hemedi Abdalla (Allen 1970, 429–432). 30 Allen 1968, 113–114. The poem edited by herself and Ahmad Sheikh Nabhany was published in Tendi (Allen 1971, 55–71). 31 Allen 1971, 77. 32 Zaharia binti Maimun, daughter of Bibi Khadija Muhammad al-Rudeyn, is also among the people to whom Ibrahim Noor Shariff was particularly grateful for having collected compositions and sung them for him (Shariff 1988, ii). The talent Swahili women have when it comes to recalling poems mc NO 17 69 RAIA | TRANSMISSION AND TEXTUAL VARIANTS Fig. 1: Colophon of manuscript 352 from Allen 1971, 21 (Utenzi wa Yaaqubu). Somehow the network of people he was introduced to over the years formed a local team that helped him regarding the circulation of Swahili manuscripts. The owner of a manuscript would make his or her manuscript copies (nakala) available to those who asked about them. The lender, in turn, would either copy the manuscript him- or herself (kunakili kwa khati) or share the copying task with a professional scribe (mwandishi) or a young person in order to produce a copy of it. The manuscript would only have been sold to Allen once the cycle was completed.33 In the lucky event that a manuscript is preserved in its entirety and includes a closing colophon, it is often possible to retrieve some names and spot traces of the manuscriptcopying chain. The sample excerpt is from ‘Utenzi wa Yaaqubu, manuscript 352’. The colophon includes the date of composition, 1329 AH (1911 CE), along with the names of two people and a reference to the provenance of the manuscript itself (Fig. 1): by heart also extends to the Utendi wa Yusuf ascribed to Kijuma. As Abou Egl states, for instance, ‘the mother of Bi. Maryamu M. al-Bakariy used to recite the Utendi wa Yusuf by heart’ (Abou Egl 1983, 223). Bi. Maryamu M. al-Bakariy was Mama Rukiya, the mother of Zena Mahmoud, Asiya Mahmoud and Maryam Mahmoud, the following generation of poetesses and custodians of Swahili tenzi on Lamu. Abubakr Mukhsin Seyyid Ali, a contemporary mwimbaji (‘singer’) and madrassa teacher on Lamu, affirmed in an interview I conducted with her that the most talented poets and musicians in Lamu are women. Despite their prowess, they do not want to appear in public, however. I visited Bi Khadija Ahmed, who was living in the Riyadah area of Lamu at the time, and Bi Ridhai Sufiyan, who was living in Pate. They both composed poems on commission. 33 According to Ustadh Ahmed Abdulkadir, a local poet and scholar, the payment for writing an original manuscript was 30 shillings and it was ten shillings for making a copy (malipo yaalan kwa msuwada asili ni shilingi 30 na kwa kuandika upya ni shilingi 10). Ten shillings was a huge amount in the 1960s. As a rough comparison, in 1975 when Ustadh Mau was already married, three daily meals would have cost twenty shillings. Personal communication, March and September 2018. mc NO 17 Transcription: Man kataba hadha alkharufu Ahmad bin ‘Abdallah bin Muhamad An‘adi tarehe yaum ٢١ Rabi’a awali sanat ١٣2٩34 wa salamu / Mali Sharifu Shehe Hamadi wa shehe Pate / wasalla llahu ’ala Sayyidna Muhamadi wa alihi wa sahbi wa salam Translation: The one who wrote this manuscript is Ahmad bin ‘Abdallah Muhamad An‘adi. Date: the 21st of the third month in the year 1329 (AH); greetings. The owner of this manuscript is Sharif Shehe Hamadi, the son of the Sheikh of Pate. Peace be upon our lord Mohammed and his family and companions. The two names mentioned in this colophon have two different roles ascribed to them: while the first one, Ahmad bin ‘Abdallah Muhamad An‘adi, was the person in charge of copying the manuscript, as the Arabic words kataba hadha alkharufu (‘he wrote this manuscript’) clearly indicate, Shehe Hamadi wa Shehe Pate was the manuscript’s owner, which the noun mali (‘property , possession, goods’) tells us. Besides these two figures, it is also relevant to mention the role played by Zaharia binti Maimun, who brought the manuscript to Lamu for Allen from Pate.35 As is evident here, no reference to the poem’s author can be discerned in the colophon – it is almost as if he or she turned into an anonymous and invisible figure once the poem had been composed. The words Man kataba hadha al-kharufu refer to Ahmad bin ‘Abdallah Muhamad An‘adi as being the 34 The decimal number of the date in the colophon looks like the number ‘2’ rather than its Arabic form. 35 However, it is also possible that, despite the Pate origins of She Hamadi, he had already settled on Lamu by 1911, when the manuscript was copied. She Ahmadi wa Pate lived on Lamu; his house was visible from those of nearby Zaharia binti Maimun and Asia; they were all neighbours. manuscript cultures 70 scribe who hand-copied the poem, but it does not actually refer to its author, who likewise cannot be identified with the manuscript’s owner. Thus, the author is not mentioned, whereas the scribe and the owner of the copy are. The author’s name – which would make us think about the ‘original/archetype’ issue if it were provided – is not an aspect which seems to have mattered in this handwritten copy. This exemplar is a witness to the fact that the text has been transmitted and handed over to other people – such as the owner and the copyist. As emphasis was put on owning or copying the text, the role of the author and the issue of the original became irrelevant. To sum up, then, I have highlighted the network of people involved in copying texts, pointing out that people often had various roles, being scribes and poets at the same time. Even those who are widely known as just the ‘owners’ of tendi often have multiple roles, such as Ahmed bin Abdalla ‘Boke’, who is said to be not just the owner of a copy, for instance, but also the author of several manuscripts, like Loho ya Kihindi and Shairi la Shilingi.36 2. The Utendi wa Yusuf manuscripts’ list in Allen’s catalogue: entangled traditions The Yusuf manuscripts listed in J. W. T. Allen’s catalogue which I have been able to go through are the following, listed here according to the original labels and numbering (provided in brackets).37 Allen provided the poem’s first line in italics; he also included the prosodic pattern for some of the manuscripts listed, e.g. ‘8.4.656 –a’ where the first number (‘8’) refers to the number of syllables (mizani). In an utendi composition, every manuscript line (mstari) contains four verses (vipande) of eight syllables each. The second number, ‘4’, refers to the four verses making up a manuscript line, whereas the last number, ‘656’, refers to the number of stanzas (sing. ubeti, pl. beti) occurring in the utendi poem. The letter ‘-a’ denotes the vowel of the final rhyme (bahari or kina cha utendi ‘rhyme of the utendi’ or kina cha kikomo ‘end-rhyme’) occurring at the end of each ubeti’s manuscript line. What is missing in this prosodic label, but is worth 36 These poems were originally ascribed to Kijuma in Allen’s catalogue (1970, 87, 106), but according to the singer who performed these poems, Zainubu l-Abideen of Mombasa, this authorship still needs to be confirmed (Abou Egl 1983, 245–246). As for Abdalla Boke, he was a famously rich man ‘throughout the period of economic decline in Lamu’ (Romero 1997, 143). The nickname given to him by the British administrators was a pejorative one, meaning ‘someone touched in the head’. 37 Allen 1970, 114. manuscript cultures RAIA | TRANSMISSION AND TEXTUAL VARIANTS pointing out here are two further important components featuring the utendi metre: the caesura between the verses, namely kituo (pl. vituo) – which is sometimes represented in manuscripts by a symbol like a small reversed heart or is graphically absent – and the division of the manuscript line in two 16-syllable half-lines, namely mishororo. Below in Table 1, I have listed all the versions of the Utendi wa Yusuf found in Allen’s catalogue. Besides outlining prosodic information, the catalogue also provides the first line of the poem and refers to other publications of it. Table 1: J. W. T. Allen’s list of the Utendi wa Yusuf (from Allen 1970, 114). Yusufu and / or Yaaqubu ‘The manuscripts given this title are extremely complicated’. 38 Yusufu, Ut wa (Yaaqubu na) (1) Mwando wangu kukutubu 8.4.656 –a Pub. Knappert, Four Swahili Epics Ms. 118; [182] 183, p. 104; 351; 352 pp. 1–32; 438 Yusufu (2) Kwa kuli kuitukuza So described at the end of Ms.; perhaps meaning part II Yusufu (3) Taani niwakhubiri 8.4.1700 –a Ms. 333, pp. 1–12; 353; 708 Yusufu (4) Bismillahi qahari 8.4.621 –a Ms. 603, pp. 1–87 Yusufu (5) Siyo yako Ms. 333, pp. 117–243 Perhaps a continuation of ‘Yusufu 2’ Yusufu (6) No beginning or end, but in the same hand as 708 Ms. 354 Yusufu (7) Namshukuru Muweza 8.4.620 –a By Said Karama Pub. Coronation Printers, Mombasa, 1964 No Ms. 38 Allen 1970, 114. mc NO 17 71 RAIA | TRANSMISSION AND TEXTUAL VARIANTS Some of the manuscripts are ascribed to different copyists and scribes, while in other cases, the scribes and copyists are unknown. Before delving into a comparative analysis of textual variants of excerpts from these documents, it is worth offering a short overview of the mechanisms involved in the production of these copies, namely their textual transmission (Überlieferungsgeschichte), specifically focusing on the period of the 1960s, when J. W. T. Allen started collecting Swahili manuscripts.39 Taking this transmission into account allows us to better understand the striking presence of at least seven witnesses (complete and incomplete) of the Yusuf story from the East Africana Section in Dar es Salaam, to which we should add the further witnesses attested in libraries in Europe, chiefly Hamburg, Berlin and London.40 This certainly makes the story one of the most widely copied versions of the classical utendi manuscript compositions. The witnesses are so different that it is obvious they do not go back to a single original source. Rather, as Driscoll has pointed out, ‘[…] we are obliged to view them as representing separate versions or redactions’.41 Taking Allen’s list in account, I felt the need to consider some aspects which I concluded from it. It seems that Allen subsumed the manuscripts into different groups using different numbers and titles (in bold) to label them (e.g. ‘Utendi wa Yaaqubu na Yusufu (1)’, ‘Yusufu (3)’, ‘Yusufu (4)’, etc.). Thus, I started to have a closer look at the groups of texts which may have belonged to the same text tradition – because of the same incipit – and could be derived from the same hypothetical archetype. I chiefly focused on the three following groups: ‘Utendi wa Yaaqubu na Yusufu (1)’, ‘Yusufu (3)’ and ‘Yusufu (4)’. As table 2 shows, I have grouped them under Roman numerals (I, II and III) and indicated the first line/half-line (kipande) of the poem. The three groups show three different versions of the story of Yusuf. I have grouped the manuscripts according to their 39 It was in the 1930s that Allen started pursuing his interest in the topic of Swahili literature in Arabic script. However, at that time and until after the war, he ‘did little more than collect one manuscript of an unedited poem at a time and [did] not look for another until [he] had finished with it. Not until about 1953 did [he] begin to realize that there was serious danger of this culture being largely lost to posterity if urgent steps were not taken to preserve it’, Allen 1970. 40 See Raia 2017, 210–211 for a complete overview and list of all the manuscripts on the Utendi wa Yusuf. It is beyond the scope of this paper to present the manuscripts listed in Dammann’s catalogue and the SOAS catalogue. 41 Quoted from Driscoll 2010, 87. For further discussion, see the chapter on classical philology entitled ‘Ci fu sempre un archetipo?’ (‘Was there always an archetype?’) in Pasquali 1934-XII: 14–21. mc NO 17 ‘closeness’ in terms of story-line. The richest tradition (I) houses three manuscripts and two typescripts, the second tradition (II) houses three manuscripts, and the third tradition is represented by one manuscript. Table 2: Diagram of three selected manuscript groups from Allen’s catalogue. Groups – created by Group I the author Group II Group III Yusufu (3) Yusufu (4) Mwando wangu Tanena ni Poem’s first half-line nakutubu/ wakhubiri kukutubu Bismillahi Qahari Allen’s label Texts Utendi wa (Yaaqubu na Yusufu) (1) Ms. 351, Ms. 352, Ms. 182, Typescript 183, Typescript 118 Ms. 333, Ms. 353, Ms. 708 Ms. 603 2.1 The manuscripts’ recension Group I: Mwando wangu nakutubu (Mss. 351, 352, 182, 183, 118) Ms. 351 has 63 pages and is 760 stanzas long, which makes it one of the longest available manuscripts ascribed to Muhamadi Kijuma to date. The manuscript, which does not have a title, opens with the basmala at the very beginning (Fig. 2). It includes a colophon containing Kijuma’s signature and the date of its composition, which is written as follows: tammati 12 jumad at-thani, Sanaa 1309 [‫ ]ھ‬bi yad Muhammad bin Abubakar Kijuma, i.e. ‘[The poem] finished on 12 Jumad at-Thani 1309 AH [Wednesday, 13 January 1892 CE], by the hand of Muhamad b. A. Kijuma’. Unlike two further manuscripts written by Kijuma which were composed deliberately for Western scholars,42 there are 42 Qissati Yusufu Hs. Or. 9893 no. 375; Staatsbibliothek zu Berlin (Dammann 1993,166–167) and Hadithi ya Yaaqubu na Yusufu Seminar 1465 H73, no. 3, Library of the Institute for Asia and Africa, University of Hamburg (Dammann 1993, 33–35). manuscript cultures 72 RAIA | TRANSMISSION AND TEXTUAL VARIANTS Fig. 2: The first two pages of Ms. 351, Utendi wa Yaaqubu. Fig. 4: The opening page of Ms. 182, Utenzi wa Kisa cha Nabii Yusufu, found together with typescript 183 in Roman script (shown in the background). Fig. 3: The opening page of Ms. 352, Utendi wa Yaaqubu. no references to any commission. The date of composition, which it is possible to obtain from the colophon, qualifies Manuscript 351 as Kijuma’s earliest copy. Furthermore, the stanzas’ arrangement is typical for Kijuma: each stanza occupies one line, and the vipande are divided from each other by the common symbol of a reversed heart marking the caesurae (vituo). The monosyllabic end-rhyme also stays the manuscript cultures same and is rendered by the Arabic letter alif. The stanzas’ numeration in Latin numbers roughly every ten stanzas seems to have been added later in blue ink, most probably by someone else. Ms. 352 is 36 pages long (Fig. 3). It spans 590 stanzas and includes 20 Qur’ānic quotations. Although there is no title piece and the opening only consists of the basmala, Manuscript 352 includes a closing colophon in which the date of its composition, 1911 CE (1329 AH), is mentioned along with the names of two people, Abdalla bin Muhamad Anhadi and Sharifu Shehe Hamadi from Pate. The poem’s layout features one stanza per line, although there is no particular ‘stop’ (kituo) sign between the vipande, hence it gives the general impression of each stanza being one full line. The monosyllabic rhyme -iya is also rendered as alif+ya here, and the handwriting appears quite clear and well vocalised despite the fact that no variant signs were adopted to distinguish the vowels /e/, /i/ and /o/, /u/. mc NO 17 73 RAIA | TRANSMISSION AND TEXTUAL VARIANTS Ms. 182 was written down in two exercise books and takes up 209 pages (20 × 16 cm). There is also a typescript copy (manuscript 183) in the same paper folder as the two notebooks (manuscripts 182.1 and 182.2). Allen states43 that this was obtained from Yahya Ali in 1963. The manuscript’s title (Fig. 4), written in Swahili in Arabic script in blue ink, reads as follows: ‫مالسلا هيلع فسوي يبَن َچ هصِق َو ِزْنٖتُﺃ‬ Utenzi wa Kisa cha Nabii Yusuf, peace upon him. The following page, before the poem starts, is occupied by the quotation of the well-known sura 12:3, written by the same hand and in blue ink. The typed transcription in Roman script that corresponds to it (in manuscript 183) was made by Sharifu al-Badawy from Mombasa and contains fewer stanzas; it does not start with the quotation from sura 12, nor the basmala. Manuscript 182 is the same text used by Knappert for his edition.44 He had already noted the similarity to the typescript prepared by Badawy for the East African Swahili Committee the same year. The layout of this manuscript splits each verse over two lines: the first 16-syllable half-line (mshororo) of the stanza is followed by the second one below it. Generally speaking, the stanzas seem well arranged, are aligned centrally on each page on which they appear and are each numbered with Latin numerals. Sheikh Yahya’s handwriting is clear, although it seems very elementary: the letters look bigger, squarer and less stylised than Kijuma’s, and the Swahili vowels /o/ and /e/ are differentiated in Arabic script by using the inverted ḍamma sign and the vertical kasra. Typescript 118 is a further transcription in Roman script obtained from Sharifu Abdirahman al-Badawy in 1963. 65 pages in length, it has 648 stanzas without any Qur’ānic quotations. Unlike typescript 183 above, the stanzas in this typescript (118) are arranged according to the common pattern based on one manuscript line (mstari) per line, which could lead to a misconception of the utendi stanza as a quatrain. Group II: Tanena niwakhubiri (Mss. 333, 353, 708) The authors of the manuscripts belonging to this group are all unknown. However, this manuscript group, tanena niwakhubiri, can be considered to transmit another branch of the story’s tradition. The first witness I will briefly introduce is manuscript 333 (with its constituent parts). Ms. 333 is a 243-page poem that Allen obtained from a woman called Asia on Lamu in 1965 and then microfilmed (Fig. 5).45 The microfilm is divided into three sections, and manuscript 333 (called ‘Yusufu 3’ in Allen’s catalogue) corresponds to the first 12 pages of it, although Allen’s question mark shows that he did not actually know what the manuscript’s contents were for sure.46 The handwriting, which is not very legible on the first few pages, becomes clearer and slightly different on the following pages, giving room to the assumption that the manuscript may have been written by more than one hand.47 There are no references to either the scribe(s) or the date of composition at the end of the poem, however. In Allen’s catalogue (where it is named ‘Yusufu 2’), a 30-page manuscript can be found on pages 86 to 116 of the microfilm. A second part of manuscript 333’s first 12 pages can be considered what Allen lists as ‘Yusufu 2’, which belongs to the same film. This assumption is supported by the fact that, despite the Swahili manuscript not bearing a title, this second part of manuscript 333 displays its own title on the top of the first page, where the number ‘2’ occurs (supposedly meaning ‘second part’). This part contains around 340 stanzas (351 if we include the last nine pages), and some stanzas seem to have been erased. The end-rhyme representation varies between -ya, -yā and -wa. The last three pages display different handwriting, thicker and less clear. There is also a third section of manuscript 333 that has been microfilmed. Listed in Allen’s catalogue as ‘Yusuf 5’, this part is likely to be a continuation of the second part already described above. This third section is 126 pages long (117 to 243), and the first line provided in Allen’s catalogue is siyo yako. For the following analysis, I will focus particularly on the first part of this tripartite microfilm, which is shown below: 45 Allen 1970, 19–20, 114. 46 Allen 1970, 20. 47 43 Allen 1970, 12. 44 Knappert 1964, 6. mc NO 17 Nonetheless, looking at the sample page of Utendi wa Haudaji (‘The Poem of the Palanquin’), manuscript A1 (Vierke 2011, 456) leads us to believe that Yusuf manuscript 333 and the Haudaji manuscript were written by one and the same scribe. manuscript cultures 74 RAIA | TRANSMISSION AND TEXTUAL VARIANTS Fig. 5: The first page of Ms. 333, simply entitled ‘Yusuf’ in Allen’s catalogue. Fig. 6: The first page of Ms. 708, Kaani niwakhubiri. Fig. 7: The first page of Ms. 353, Utendi wa Yusufu (incomplete). Fig. 8: The first page of Ms. 603, Bismillahi Qahari. manuscript cultures mc NO 17 75 RAIA | TRANSMISSION AND TEXTUAL VARIANTS Ms. 708 is a further manuscript (Fig. 6), the pages of which were found together with manuscript 333. This manuscript was obtained from Asia in 1965, so she provided Allen with two different manuscripts that same year (the entire manuscript 333 and manuscript 708).48 The poem contains 448 stanzas, although the reading of some of them is not very clear, which makes its numbering a difficult task. The third manuscript belonging to this group is the longest and most complete copy: manuscript 353 (Fig. 7).49 This document was obtained by Allen from Zaharia binti Maimun in 1965. Although it is also incomplete, it is 32 pages long and contains 484 stanzas, which makes it longer than the other witnesses. Group III: Bismillahi Qahari (Ms. 603) Obtained from Zaharia binti Maimun on Lamu in 1966, manuscript 603 in Arabic script occupies the first 87 pages of a 103-page notebook.50 The structure of the introduction of this manuscript suggests ascribing it to Ustadh Said Karama, whose edited transliteration was published in Mombasa51 and whose introduction follows the same structure as the Arabic script version: the 620 stanzas that comprise the manuscript are prefaced by a dibaji (‘prologue’) in prose form, which occupies the first page and is indented and set in quotation marks – features very unusual in manuscripts in Arabic script. The dibaji is followed by the maoni ya mtungaji (‘the author’s opinion’), composed in utendi metre, which opens with Bismillahi qahari (Fig. 8). Compared to the published version, manuscript K also shares similar closing remarks, the maoni ya msahihishi (‘the editor’s opinion’), written in verse. In a similar way to manuscript 182, the stanzas in K are arranged in two 16-syllable half-lines, one halfline (mshororo) per line. The caesura between vipande is graphically represented by the commonly reversed heart, and each ubeiti is numbered by an Arabic numeral between brackets. The blue-ink handwriting sometimes alternates with black ink on sporadic pages at the beginning of the poem. 3. Variants and versions Zumthor, Cerquiglini and the ‘new’ philologists have all argued that textual instability (variance, mouvance, ‘unfixedness’) is so fundamental a feature of chirographically transmitted texts that rather than trying to bring order to this chaos we should celebrate it.52 Variants and versions are two recurrent key terms which I am going to use in the following part of this paper, where I will venture to describe the complicated relationships between the manuscripts known as, or circulated under the name of, the Utenzi wa Yusuf (or also Hadithi ya Yusufu, Kisa cha Yusufu, Qissat il-Yusuf or Utenzi wa kisa cha Nabii Yusuf).53 I intend to describe the variability in terms of orthography and dialectal features as well as the adaptation of the story in other rewritings and media, e.g. from paper to recorded versions, and in more or less complete copies. The section is divided into three roughly equal parts. Firstly, in section 3.1, I am going to discuss dialectal and scribal variants and verse layout, looking not only at handcopied manuscripts in Arabic script, but also at typed transliterations in Roman script. Secondly, I will focus on textual variants (section 3.2) where I will chiefly enquire about the narrative frame, e.g. the incipit, through which the poet-adapter sets his or her story tradition.54 For this analysis, I will take the tanena niwakhubiri manuscripts group into account (Group II) along with a recorded version – less well known and never analysed before – which will allow me to compare a specific text with its new medium of adaptation. To conclude, this chapter will also entail a note on some apparently ‘incomplete copies’ (section 3.3) which, in line with Cerquiglini’s Praising the Variant, are worth being analysed and fit the context of this paper well. 52 48 Allen 1971, 41, 114. 49 Allen 1970, 21, 114. 50 Allen 1970, 35, 114. Somewhat like an Italian zibaldone (lit. ‘scartafaccio’, ‘scribbling pad’, ‘heap of things’), the notebook includes a miscellaneous section: not only the Utendi wa Yusuf, but a further poem in Arabic script, namely the Utendi wa Asha (pp. 88–103); see Allen 1970, 35. 51 Karama, 1968. mc NO 17 Quinn and Lethbridge 2010, 141. 53 The section where all the extant sources on the Utendi wa Yusuf are listed in Allen’s catalogue indeed opens with the author himself claiming the following: ‘The manuscripts given under this title are extremely complicated’, Allen 1970, 114. 54 It is beyond the scope of this article to delve into a comparative analysis of the narrative variations. To find out more about adaptation as a form of amplification, see the comparative textual analysis undertaken for three manuscripts of the Utendi wa Yusuf edited in Raia 2017, 103. manuscript cultures 76 3.1 Formal variants in Arabic script The first and most common type of variant in philology is what is known as a ‘formal variant’, which may be a graphic, phonetic or morphological variant of the shape of a single word within the text.55 The following variant readings belonging to this category are from three different exemplars which are from Group I (manuscript 351, manuscript 352 and manuscript 182): while manuscript 351 and manuscript 352 are two microfilmed manuscripts (Figs 9 and 10), manuscript 182 is a 209-page notebook (Fig. 11). The excerpts below contain the first stanza of the Utendi wa Yusuf, and as I will show, while the content and meaning are almost the same, the three excerpts have different handwriting and the scribes copied and adapted Arabic letters to Swahili based on their own choices and skill. The images below show the first manuscript line (mstari) of the Utendi wa Yusufu in the three different manuscripts. The first two scribes structured the poem by kipande (8-syllable lines): 1. Mwando wangu nakutubu; 2. Jina la Mola Wahabu; 3. Hadithi ya Yaaqubu; 4. Nimependa kuwambiya. In manuscript 182, however, Yahya Ali Omar arranged each stanza into two half-lines (mishororo) of 16 syllables each (two 8-syllable lines): 1. Mwando wangu nakutubu/Jina la Mola Wahabu, 2. Hadithi ya Yaaqubu/Nimependa kuwambiya. RAIA | TRANSMISSION AND TEXTUAL VARIANTS Fig. 9: Ms. 351, (‘Yaaqubu, Ut. wa’)56, Muhammad Kijuma, 1309 AH/1892 CE; microfilm, Arabic. Fig. 10: Ms. 352, (‘Yaaqubu, Ut. wa’)57, Abdalla bin Muhamad Anhadi, Sharifu She Hamadi (of Pate) and Anhadi Hamadi (of Pate), 1329 AH /1911 CE; microfilm, Arabic. Fig. 11: Ms. 182, (Ut. wa kisa cha nabii Yusufu)58, Yahya A. Omar; Arabic. Obtained from Yahya Ali Omar, January 1963. Kiswahili written in Arabic script has no standard orthography and leaves room for variation. The different conventions in writing Swahili, which Knappert59 has estimated at half a dozen, have allowed for the possibility of variant readings in the philological analysis of every Swahili composition in Arabic script;60 nasals and glides are written sometimes, but not always, for instance. Vocalic differences between /u/ and /o/ and /i/ and /e/ are not indicated consistently. Arabic words are sometimes written according to their Arabic orthography or sometimes according to their Swahili pronunciation. As the samples show, each scribe (Kijuma, Anhadi and Yahya Ali Omar) wrote down the same stanza, adopting different conventions – some more or less helpful to readers, some more accurate than others – in which the idiosyncrasies of each scribe can be detected. In Table 3, the formal variant readings are listed accordingly by 8-syllable verse line (kipande). 56 Allen 1971, 21, 114. 57 Allen 1971 58 Allen 1971, 12. 59 Knappert 1989, 81. 60 55 Stussi 2011, 132–36. manuscript cultures This is not only true of classical utendi compositions, which exist in many copies, or Muyaka’s nineteenth-century shairi verses (Abdulaziz 1979), but also of the several tumbuizo attributed to Liyongo Fumo (Miehe 2004) and the eighteenth-century gungu songs attributed to Bwana Zahidi Ngumi. mc NO 17 77 RAIA | TRANSMISSION AND TEXTUAL VARIANTS Table 3: Formal variant readings in manuscripts from Group I. Kipande 1 Mwando wangu kukutubu ‘[this is] the beginning of my writing’ Variant reading: Mwando wangu nakutubu ‘I write the beginning [of my writing]’ Ms. 351 Ms. 352 – The glide /w/ in mwando is represented as ‫ َوْم‬, comprised of ‫ و‬preceded by ‫ م‬+ sukūn, exclusively in Ms. 182; the nasal compound /nd/ in the same word (mwando) consists of just ‫د‬ in Ms. 352, whereas in Ms. 351 and 182, a ‫ ن‬appears before the ‫د‬. – The nasal cluster /ng/ in the possessive adjective wangu is rendered only as ‫ غ‬in mss. 351 and 352. In Ms. 182, the compound is rendered as ‫ ن‬+ ‫ڠ‬. In Ms. 352, the scribe has placed ‫ ا‬between the‫ و‬and ‫ غ‬in wangu. – The infinitive prefix ku- in the verb kukutubu in mss. 351 and 352 corresponds to a different prefix in Ms. 182: na-, which yields a first person singular in the present-A-tense form. Morphologically speaking, both readings (kukutubu / nakutubu) are feasible. Ms. 182 Kipande 2 Jina la Mola Wahabu ‘In the name of God, the Giver’ Ms. 351 Ms. 352 Ms. 182 mc NO 17 – In all the three texts, the Kiamu form ‘ina’ is used instead of the standard ‘jina’. Whereas in Ms. 351 /i/ in ina is spelt ‫ ا‬+ kasra, where ‫ ا‬is followed by the long vowel ‫ي‬, in Ms. 352 it is not. Ms. 182 reads alif hamza ‫ إ‬+ kasra rather than a simple ‫ا‬. – he syntagma la Mola is only written as two morphologically separate words in Ms. 182, where an isolated ‫ ل‬precedes the noun Mola. In Ms. 352 it is written as one word, with the consonant ‫ ل‬attached to Mola. In Ms. 351, in place of ‫ل‬, the reader is tempted to read‫ ن‬, which would yield a slightly different translation: ina na Mola‚ ‘the name and God’. – Only in Ms. 182 is the vowel /o/ in the word Mola precisely rendered by a dhuma ya kupinduwa, or ‘overturned ḍamma’, while Ms. 351 and Ms. 352 have a simple classical ḍamma. – The Arabic word wahhābu accurately bears a shadda on the consonant ‫ ه‬of wahhābu in Ms. 182, whereas it is lacking in Ms. 351 and Ms. 352. The latter two also omit the long vowel fatḥa + ‫ا‬. manuscript cultures 78 RAIA | TRANSMISSION AND TEXTUAL VARIANTS Mshororo 2 (‘second half-column’); kipande 3 Hadithi ya Ya‘aqubu ‘The story of Ya‘aqūbu’ Ms. 351 Ms. 352 Ms. 182 – For the word of Arabic origin ḥadīth, in Ms. 352, the copyist uses ‫ ه‬instead of ‫ح‬, while the copyist of Ms. 351 omits the long vowel ‫ي‬. Still, all three variants have hadithi with a kasra under the ‫ ث‬which makes it a Swahili word and not an Arabic one. – The Arabic name Ya‘aqūbu is only spelt correctly in Ms. 182; the possessive concord ya (‘of’) is rendered by the isolated single consonant ‫ ي‬in Ms. 182, whereas in Ms. 351 and Ms. 352 it is rendered by ‫ ي‬+ ‫) اَي ( ا‬. The ‘incorrect’ spelling in Ms. 351 and Ms. 352 may have been influenced by the need for eight syllables. Interestingly enough, the ‘correct’ spelling in Ms. 182 only has seven syllables. – In Ms. 352, the qāf and wāw, which are both found in the name Ya‘aqūbu, overlap in the writing: ‫ و‬has two diacritics on top, which could make the letter look identical to ‫ ; ق‬thus the character seems to represent two Arabic consonants in one. Kipande 4 Nimependa kuwambiya ‘I want to tell you’ Ms. 351 Ms. 352 Ms. 182 manuscript cultures – The voiceless occlusive /p/ in nimependa is rendered by the Persian pe ‫ پ‬, whereas Ms. 352 merely opts for the Arabic ‫ ب‬. – The nasal cluster /nd/ is clearly spelt as ‫ ن‬+ ‫ د‬only in Ms. 182, whereas the nasal is omitted in Ms. 351 and Ms. 352 (phonotactically speaking, this is better). – The vowel /e/ is rendered by kasra in Ms. 351 and Ms. 352 and as such is indistinguishable from /i/; in Ms. 182, on the other hand, the copyist usually wrote the vowel as a vertical kasra, namely kasiri ya kusimama ‘an upright kasra’, to distinguish it from the vowel /i/, although he did not do it in this line. – The bahari (last monosyllabic end rhyme) -iya is rendered by ‫ ي‬+‫ ا‬in all cases in Ms. 351 and Ms. 352 (‫)اَي‬, but only by ‫ ي‬in Ms. 182 (‫)َي‬. mc NO 17 79 RAIA | TRANSMISSION AND TEXTUAL VARIANTS Manuscript 182 is one of the most recent exemplars and the one that best respects a set of rules for writing Swahili in Arabic script, which became established in the second half of the twentieth century on the basis of the Swahili used in Mombasa.61 Indeed, its accuracy is particularly apparent in the rendering of nasal clusters and glides, in the separation of the possessive connector from the noun (e.g. la written in isolation in jina la Mola v.2, or ya in hadithi ya Yaaqubu v.3), which might reflect the influence of Swahili orthography on Latin script, and in the correct transcription of Arabic proper names such as Ya‘aqūbu. After all, it was in the course of the twentieth century that an effort was made to establish some conventions for writing Swahili in Arabic script: the rendering of the vowels /e/ and /o/, the representation of homorganic nasals in front of plosives, the velars g and ng, the voiced fricative v, aspiration and the differentiation between dental and alveolar plosives.62 While the relationship between the tradition of a text and the evolution of its orthography is not necessarily straightforward and does not need to be a conscious decision,63 the introduction of specific new ‘Swahili’ definitions (such as damma kupinduwa, ‘an overturned damma’ and kasra ya kusimama, ‘an upright kasra’, shows the scribes’ conscious effort to deal with the Arabic alphabet, and the meta-vocabulary speaks of their familiarity with it. 3.2 Variants in framing the story In the following, I will consider other witnesses under the title of Utendi wa Yusuf, namely those that differ from the manuscripts analysed above more substantially, going beyond variants at the level of the single line. The further witnesses, which I am going to analyse here, come from Group II and can be grouped together because of the incipit in their first 8-syllable line (kipande): tanena niwakhubiri. The story is still about Yusuf and in utendi metre, but in totally different words. 61 62 Frankl and Omar 1997. Knappert 1989, 81, Frankl and Omar 1997, 61. 63 Pasquali 1934, 17. My own translation from the original Italian version: ‘Che peculiarità ortografiche non provano nulla è risaputo. Specie da filologi greci e romanisti, che hanno osservato come un testo si modernizzi nell’ortografia senza che per questo bisogni supporre l’opera conscia di un amanuense; l’ortografia dipende almeno altrettanto dal tempo della copia che da quello degli originali da cui furono copiati: essa non appartiene alla tradizione se non là dove contrasta con le abitudini delle scuole scrittorie, dunque arcaizza’. mc NO 17 Rather than exploring the formal variants in this group of witnesses, an interesting element I would like to underscore in this paragraph instead is ‘the narrative frame in which the story’s transmission is set’, looking particularly at the preface – the dibaji – of three manuscripts. From the very beginning (stanzas 4–5), the anonymous composers inform the audience of the story that they are about to tell in all three witnesses in this group (Mss. 333, 708 and 353) (Table 4). The two manuscript lines (mistari) convey the desire to understand what has been read or heard from the Qur’ān. There is a clear reference to the Qur’ān as the inspiring source that has prompted the composer to explain/translate (fasiri) the story in order to make people aware of it (kuarifiya). As the transliteration shows, what stanza five says in manuscript 708 and manuscript 353 is different to what can be found in manuscript 333. Two moments of ‘interpretation/translation’ are highlighted in manuscript 708 and 353: fasiri ilo twayibu and ziyada fasiri piya. The usage of the verb fasiri twice invites an in-depth understanding of the meaning of fasiri in Swahili, which does not actually mean ‘translate’, but rather ‘explain, interpret, comment on’. As Talento has suggested, ‘the verb kufasiri/kutafsiri (‘to translate’) also referred to a variegated series of re-writing processes’ and while it was used interchangeably with kutarjumi until the nineteenth century, in recent times the latter mostly hints at ‘simultaneous interpretation’.64 Beyond the variant reading offered in stanza five by the manuscripts belonging to the same tradition, it is generally possible to see from this very early incipit how the manuscripts starting with tanena niwakhubiri put the story into a different context of reception (textus receptus). While the poet refers to an ‘original’ Arabic text and the Qur’ān, in the manuscripts from the mwando wangu nakutubu group, the poet expresses his pleasure (kupenda) in telling the story as an author or adapter here: Hadithi ya Ya’aqubu / Nimependa kuwambiya, ‘The story of Ya’aqub, I want to tell you’. Furthermore, while in the manuscripts from tanena niwakhubiri, there is a focus on the Sifa za thumwa Yusufu (‘The praises of the prophet Yusuf’– st. 5) right from the beginning, in the mwando wangu nakutubu group the focus is on Ya’aqub, Yusuf’s father, it being entitled Hadithi ya Ya’aqubu (‘The story of Ya’aqub’ – st. 1), which the poet wishes to narrate. 64 Talento 2013, 86. manuscript cultures 80 RAIA | TRANSMISSION AND TEXTUAL VARIANTS Table 4: Comparison of prefaces (dijabi). St. 4 St. 4 Ms. 333 St. 4 St. 5 Ms. 708 St. 4 St. 5 Ms. 353 Transliteration, stanzas 4–5 Naliona kitabuni / fasiri ya Qurani / moyo ukatamani / ili kuarifiya St. 4 – Mss. 333, 708, 353 Kitaja kisifu / sifa za thumwa Yusuf / kapenda kuarifu / kama niliyosikia St. 5 – Ms. 333 Variant reading of st. 5 Naliona Khatwibu / ya swahihi mujarabu / fasiri ilo twayibu / ziyada fasiri piya St. 5 – Mss. 708, 353 Translation, stanzas 4–5 I saw it in the book and the Qur’ān’s translation; my heart desired to let you know. St. 4 – Mss. 333, 708, 353 I praise the prophet Yusuf’s qualities; I love to tell you what I heard. St. 5 – Ms. 333 I saw Khatwibu. He tried to interpret properly and translate it all St. 5 – Mss. 708, 353 manuscript cultures mc NO 17 81 RAIA | TRANSMISSION AND TEXTUAL VARIANTS On the one hand, there are variant readings in the manuscripts belonging to the same tanena niwakhubiri manuscript tradition (e.g. st. 5 in Ms. 333, which differs from manuscripts 708 and 353) which show altogether different attitudes adopted by the poets/adapters to the poem. While setting the story within an Islamic frame of reference hinting at the Qurʾān is a narrative device that is very common in most of the classical tendi, not citing the Islamic pedigree at the beginning is rather unusual. This reveals a certain degree of freedom that the poet avails himself of in reshaping the story right from its incipit. The attention in the mwando wangu nakutubu manuscripts is more focused on the poet’s appropriation of the story and his own pleasure in telling it to his audience. It seems that whereas the Qur’ān is without any doubt the source which inspired the Swahili poet/interpreter to write it down in Swahili verse form in the tanena niwakhubiri manuscripts, in the mwando wangu nakutubu group the poet puts some distance between himself and the prior sources – as if making references to specific texts was not an important concern for him. Within the mwando wangu nakutubu group the poet/adapter does indeed seem far less concerned with historical reliability: his major intention is to depict the story in vivid terms. 3.2.1 From a written to a sung version: the Bismillahi Qahari (Ms. 603) A process of re-enacting the Yusuf text can also be found in chanted versions. To what extent is a text effected by transformation and variability if it is performed orally? Does a process of tafsiri take place for a singer as well (a mwimbaji)? Furthermore, is the work of a singer comparable to the interpretation which a composer goes through from the original to his/her own Swahili text? In the following, I will briefly focus on the Bismillahi Qahari (Ms. 603), which belongs to the third manuscript group I have selected from Allen’s list. In addition to the original manuscript in Arabic script and its version in Roman script, there is another late version of this poem composed by Karama and sung by Muhammad Abdalla Kadara, which has been recorded on a compact disc.65 Born in Lamu but now living in Mombasa, Kadara’s original name is Muhamad Abdalla Bakathir. His father was a poet and businessman on Lamu. Kadara himself is mainly a singer, 65 A video recording of the Qissati Yusuf by Malim Yahya also exists. The reading was organised by Ridder Samsom at SOAS in 1997. In 2016, Samson kindly gave me a copy of this recording for my private use. mc NO 17 but he also composes poems on commission. He regularly visits and spends his afternoons at the Bwana Radio Service in Kibokoni (Mombasa) where I had the chance to meet him in February 2018. Like most tendi recorded on CDs, the recitation opens with the singer stating both his own name and that of the poem’s author, along with information on where to buy the CD version that is about to start. At this point, the performer already defines implicitly what he is in relation to this text: neither the author nor a scribe who copied the text. Rather, he re-enacts it using his voice – albeit with a certain amount of liberty towards the text. The singer – although allegedly relying on a copy of the utendi while recording his own version – does not chant the same amount of stanzas found in the written copy of the poem. While scanning the poem and setting it to a melody (kutia sauti/maadhi), he skips one or more stanzas even though the performer knows the poem and has read it at least once before recording it.66 Like a scribe who dictates the poem to himself before writing it down,67 Kadara may have tested and selected the right melody for the poem and then decided on certain editorial measures such as dropping particular stanzas, changing a word to modify the rhyme, and the nolens volens repetition of a line. In the case of this sung version, on the one hand, these editorial interventions might be due to some constraints of the recording, such as the necessity to fit the sung version of the poem within the time constraints of the cassette or CD.68 On the other hand, further variations such as the singer changing the words and (in so doing) the meaning of a line could be seen as an independent decision made by the singer himself, 66 Kadara in my interview with him in Mombasa, February 2018. 67 In philology and to explain the presence of mistakes within texts that have been transmitted, self-dictation or inner pronunciation of a text is considered one of the four stages that take place while copying: visual perception of the letters, memorisation, self-dictation and reproduction, see Isella 2009 and Stussi 2001, 86. 68 The master copy from which I obtained the sung version of the Utendi wa Yusuf by Kadara on Lamu consists of two CDs lasting 45 minutes each, making a total duration of 90 minutes. Before the advent of CDs, recording on cassettes created some constraints for singers. The young mwimbaji Abubakr Mukhsin Sayyid Ali of Lamu told me how he had had to struggle to find the proper tempo for recording the Ramani ya Maisha ya Ndowa or Haki za Watoto written by Ustadh Mau on 60- or 90-minute cassettes. To avoid wasting money on more than one cassette, the mwimbaji had to adapt the length of the poem to his rhythm, which in turn had to fit the available tape length. manuscript cultures 82 who somehow felt like the new author of that sung version.69 In Kadara’s recorded version, in the first 30 stanzas alone, I noticed that he has skipped the following verses: st. 3, 6–12, 18–21, 24, 26 and 28–33. In addition to this, the attentive listener who has Karama’s handwritten version at his or her disposal will notice that in stanza 13, while the ‘original’ version reads kisa hicho maarufu ‘this is a famous story’, the recorded version recites nina kisa maarufu ‘I have a famous story’.70 Moreover, in stanza 27, the second mshororo of the poet’s version reads na mwilini hana kovu, uzuri umezidia ‘On his body he (Yusuf) has no scar; his beauty has grown’, whereas the singer’s version recites na mwilini hana kovu, uzuri amezidia ‘On his body he has no scar; he has increased in beauty’. As in the bismillahi qahari manuscript, where the composer (Karama) refers to writing down the meaning of the story while changing its language (Maana ni kikutubu lugha kuwafasiria, st. 12), in the sung version the singer, Kadara, engages himself in transforming the written version. Tafsiri and kutia sauti are both practices through which reappropriation and transformation are enacted and a text takes on a new form and life. Conclusion The aim of my paper was to reflect upon the variability of one text. Beginning with considerations on manuscript collection and arrangement, I first sought to present some implications of transmission and copying practices. I put my focus on the network of people involved in preserving a text, which can be put in several forms (from a manuscript to a memorised version and then to a microfilm) by different agents in the course of its transmission, like the manuscript’s owner, the poet, the scribe and the collector. Texts are always acted out, and not only writing but memorisation is an important way in which they are transmitted from one generation to the next. In a further section of the article, I focused on the internal variability of texts. In transmitting and copying the Utendi wa Yusuf, we saw how the various copies allegedly belonging to the same text tradition, such as the mwando wangu nakutubu manuscript group, are shaped differently in terms of scribal 69 This free new adaptation of the lines was not well received by the ‘original’ author of the poem. Ustadh Mahmoud Ahmed Abdulkadir (Mau) told me he was not entirely pleased with the fact that the mwimbaji Muhamad Abdalla Kadara had changed some of his poems here and there (Mau, March 2018). It goes without saying that the maoni ya mtungaji (‘the poet’s opinion’) is completely omitted in the sung version. 70 RAIA | TRANSMISSION AND TEXTUAL VARIANTS conventions, although they relate the same story in the same metre. The handwritten form of each manuscript represents the craftsmanship of a known or unknown young apprentice or an experienced scribe shaping letters, glides and vowels and formatting the poem in different ways. Regardless of whether we are confronted with several copies ascribed to the same author or we are dealing with copies ascribed to different copyists, the poet-scribe always re-shapes the poem. Formal orthographic variants mainly concerning the adoption and adaptation of Arabic script to render Swahili sounds, as well as the utendi stanza layout vis-à-vis typed versions, have clearly shown how difficult it is to trace a fixed set of conventions among scribes. Variability even increases if a manuscript has been copied by more than one hand, as in the case of manuscript 333. The plethora of Utendi wa Yusuf exemplars does not merely reflect the different aesthetic conventions adopted by the scribes, but also differences in the way of framing the story and showing its transformation. This latter issue has allowed me to talk about other versions, redactions and traditions of the story of the same prophet, Yusuf. Examining the second manuscripts group – tanena niwakhubiri – has illustrated how much each manuscript enacts a ‘re-performance’ in which the alleged author – known or unknown – commits himself or herself to understand the meaning of the story (kufanya fasiri ‘to translate’) and conveying it to his or her own audience in order to make it known. A re-formulation or re-writing of something already said or written is involved. What is important to stress here is that the Utendi wa Yusuf text has always undergone a process of form-giving and changing which has concerned and affected the text’s language and metre, its interpretation (tafsiri) and writing (kutubu) as well as its voicing. Side by side with the figure of the poet/adapter committed to producing a copy of the poem, I have also introduced the figure of the singer/adapter committed to giving voice to a written text and editing it at his own liberty. Such a reperformance must also be seen as another form in which the story is re-formulated and transmitted. Accordingly, the task of deriving all versions from one archetype is a utopian task. What has emerged from the overview of the different Utendi wa Yusuf manuscript groups is a mosaic-like scenario, showing how much the texts have diverged as they were written down and re-written over the Karama 1994, 3. manuscript cultures mc NO 17 83 RAIA | TRANSMISSION AND TEXTUAL VARIANTS centuries.71 It is interesting to consider that, up to now, it is often only in well-known and well-copied manuscripts commissioned by Europeans in the first half of the twentieth century, where we find a particular care of the scribe or composer. The latter put much effort into embellishing the colophon of the manuscripts, which is missing in the other manuscripts which were not written on commission and which this paper has been focused on. For instance, the colophon of the Qissati Yusufu (Fig. 12) by Muhamadi Kijuma written for Ernst Dammann reflects how much European scholars have imposed their expectations and needs on Swahili manuscripts, whereas for a local audience, the meaning and the importance is not on paper, but in the (memorised) living text. This also explains the abundant presence of incomplete copies (e.g. Mss. 333, 353): these manuscripts did not only waste away over time, but show that it was not important to be complete on paper or to be faithful to a paper copy, as Kadara’s practices of singing and the abundance of scribal variants have indeed shown. The ‘carelessness’ evident in written copies is a sign revealing that manuscripts did not circulate very far. As Samsom has attested, ‘[t]he manuscripts, books, papers and other items that make up a specific collection are in the Swahili context normally perceived as belonging to the family’.72 And the facts contributing to the dissolution and disappearance of many collections of manuscripts in East Africa may also hold true in the case of the Utendi wa Yusuf manuscripts. These include ‘… the falling apart of extended family structures safeguarding the common heritage, a very weak tradition of paper conservation in a devastating climate, [and] a general absence of preservation practices’.73 The hope is that other philologists – while immersed in catalogues, archives and household collections –will feel inspired not only to look for the most reliable copy of a classic text, but also for its many other fascinating exemplars. Fig. 12: Staatsbibliothek zu Berlin, Hs. Or. 9893 Qissati Yusufu , 375; p. 1 (detail from the colophon).74 71 Ms. 351 is dated 1309 AH/1892 CE, which makes it the first existing manuscript to be attested, whereas the latest manuscript version is Ms. 603, which dates back to 1384 AH /1964 CE. 72 Samsom 2015, 208. mc NO 17 73 Samsom 2015, 208. 74 Dammann 1993, 166–167. manuscript cultures 84 RAIA | TRANSMISSION AND TEXTUAL VARIANTS REFERENCES Abdulaziz, Mohammed (1979), Muyaka: 19th Century Swahili Popular Poetry, Nairobi: Kenya Literature Bureau. Allen, John, W. T. (1968), ‘The Collection and Preservation of Manuscripts of the Swahili World’, Swahili Journal of the Institute of Swahili Research, 38/2: 109–117, Dar es Salaam: University College. —— (1970), The Swahili and Arabic Manuscripts and Tapes in the Library of the University College Dar-es-Salaam, Leiden: Brill. —— (ed.) (1971), Tendi: Six Examples of a Swahili Classical Verse Form with Translations & Notes, New York: Africana Pub. Corp. Bausi, Alessandro and Alessandro Gori (2006), Tradizioni orientali del ‘Martirio di Areta’: La prima recensione araba e la versione etiopica; edizione critica e traduzione, (Quaderni di semitistica, 27), Florence: Ist. di Linguistica e di Lingue Orientali, University of Florence. Cerquiglini, Bernard (1989), Éloge de la variante: Histoire critique de la philologie (Paris: Éditions du Seuil). Dammann, Ernst (1940), Dichtungen in der Lamu-Mundart des Suaheli: Gesammelt, herausgegeben und übersetzt von Ernst Dammann (Schriften des Kolonialinstituts der Hansischen Universität, 3), Hamburg: Friederichsen, De Gruyter & Co. —— (1993), Afrikanische Handschriften (Verzeichnis der orientalischen Handschriften in Deutschland, 24), Stuttgart: Steiner. De Kreij, Mark (2015), ‘Transmission and Textual Variants: Divergent Fragments of Sappho’s Songs Examined’, in André Lardinois, Sophie Levie, Hans Hoeken and Christoph Lüthy (eds), Texts, Transmissions, Receptions, Leiden: Brill, 17–34. Driscoll, Matthew James (2010), ‘The Words on the Page: Thoughts on Philology, Old and New’, in Judy Quinn and Emily Lethbridge (eds), Creating the Medieval Saga: Versions, Variability and Editorial Interpretations of Old Norse Saga Literature, Denmark: University Press of Southern Denmark, 87–104. Egl, Abou (1983), The Life and Works of Muhamadi Kijuma, unpublished PhD thesis, School of Oriental and African Studies, London, microfilm. Frankl, Peter and Yahya Ali Omar (1997), ‘An Historical Review of the Arabic Rendering of Swahili: Together with Proposals for the Development of a Swahili Writing System in Arabic Script (Based on the Swahili of Mombasa)’, Journal of the Royal Asiatic Society, 3rd ser., 71/1: 55–71. Genette, Gerard (1982), Palimpsestes: La littérature au second degré, Paris: Éditions du Seuil. Harries, Lyndon (1962), Swahili Poetry, Oxford: Clarendon Press. Isella, Dante (2009), Le carte mescolate vecchie e nuove, Torino: Piccola Biblioteca Einaudi. Karama, S. (1964) (ed.), Utenzi wa Nabii Yusu, (Mombasa: Haji Mohamed and Sons [repr. 1994]. Knappert, Jan (1964), Four Swahili Epics, Leiden: Drukkerij Luctor et Emergo. Macé, Caroline (2015), ‘Textual Criticism and Text Editing’, in Alessandro Bausi, Pier Giorgio Borbone, Françoise Briquel-Chatonnet, Paola Buzi, Jost Gippert, Caroline Macé, Marilena Maniaci, Zisis Melissakis, Laura Parodi and Witold Witakowski (eds), Comparative Oriental Manuscript Studies: An Introduction, Hamburg: COMSt, 321–465. Mahimbi, Evaristo (1975), Utenzi wa Yusufu, Tanzania: Ndanda Mission Press. manuscript cultures mc NO 17 RAIA | TRANSMISSION AND TEXTUAL VARIANTS McGahn, Jerome (1983), A Critique of Modern Textual Criticism, Charlottesville: University of Virginia Press. —— (2013), ‘Philology in a New Key’, Critical Inquiry, 39/2: 327–346. Miehe, Gudrun (1979), Die Sprache der älteren SwahiliDichtung: Phonologie und Morphologie (Marburger Studien zur Afrika- und Asienkunde, 18), Berlin: Reimer. Miehe, Gudrun (2002), Kala Shairi: German East Africa in Swahili Poems, Cologne: Rüdiger Köppe Verlag. —— (2008), ‘Kwamba uko Pate-Unga, nami niko Kiwandeo: The Paths of Zahidi Mngumi’s Powerful Poetic Language from Indian Ocean Shores to the Western World’ (unpublished paper). —— and Clarissa Vierke (2010), Muhamadi Kijuma: Texts from the Dammann Papers and other Collections, Cologne: Rüdiger Köppe Verlag. Pasquali, Giorgio (1934), Storia della tradizione e critica del testo (Florence: Felice Le Monnier). Raia, Annachiara (2017), The Utendi wa Yusufu: A Critical Edition of the Swahili Poem of Yusuf and a Study of its Adaptation at the Swahili Coast, unpublished PhD dissertation, University of Bayreuth and University of Naples L’Orientale. Romero, Patricia (1997), Lamu: History, Society and Family in an East African Port City, Princeton, NJ: Markus Wiener. Sacleux, Charles (1939), Dictionnaire swahili-français (Travaux et mémoires de l’Institut d’ethnologie, 36), Paris: Université de Paris, Inst. d’Ethnologie. 85 —— (2015), ‘The Dissolved Collection of Sheikh Aliy Hemed Abdallah al-Buhriy (1889–1957)’, Islamic Africa, 6: 201–208. Shariff, Ibrahim Noor (1988), Tungo zetu: Msingi wa Mashairi na Tungo nyinginezo, Trenton, NJ: The Red Sea Press. Stussi, Alfredo (2011), Introduzione agli studi di filologia italiana, Bologna: Il Mulino. Talento, Serena (2013), ‘Of Presences/Absences, Identity and Power: The Ideological Role Of Translation Into Swahili During Late Pre-Colonial And Early Colonial Times’, Swahili Forum, 20: 85–101. Varvaro, Antonio (2012), Prima lezione di filologia, Bari: Laterza. Vierke, Clarissa (2011), On the Poetics of the Utendi: A Critical Edition of the Nineteenth-Century Swahili Poem ‘Utendi wa Haudaji’ together with a Stylistic Analysis (Beiträge zur Afrikaforschung, 50), Zurich: LIT. Werner, Alice (1921–23), ‘Utendi wa Ayubu’, Bulletin of the School of Oriental Studies, 2: 85–115; 297–320; 347–416. Wilkening, Friederike (2000), ‘Who is J.W.T. Allen?’, Swahili Forum, 7: 237–258. Zhukov, Andrey (2004), ‘Old Swahili-Arabic Script and the Development of Swahili Literary Language’, Sudanic Africa, 15: 1–15. Zumthor, Paul (1972), Essai de poétique mediévale, Paris: Editions du Seuil. Samsom, Ridder (1997), ‘“Hammatbihi w ahammabiha”: Fasihi ya kiswahlili na kisa cha Yusuf’, Swahili Forum, AAP 51: 83–94. mc NO 17 manuscript cultures 86 PICTURE CREDITS RAIA | TRANSMISSION AND TEXTUAL VARIANTS LIST OF MANUSCRIPTS FROM DAMMANN’S CATALOGUE (1993) CITED IN THIS ARTICLE Fig. 1: © Allen 1971. Figs 2–11: © University Library of Dar es Salaam, East Africana Section. Hs. Or. 9893 no. 375 Qissati Yusufu – Staatsbibliothek zu Berlin – Preußischer Kulturbesitz (Berlin State Library – Prussian Cultural Heritage) Fig. 12: © Staatsbibliothek zu Berlin – Preußischer Kulturbesitz (Berlin State Library – Prussian Cultural Heritage). Seminar 1465 H73, no. 3 Hadithi ya Yaaqubu na Yusufu – Universität Hamburg, Asien-Afrika-Institut (Institute of Asian and African Studies) Tables 3 and 4: Images used © University Library of Dar es Salaam, East Africana Section. LIST OF MANUSCRIPTS FROM ALLEN’S CATALOGUE (1970) (DAR ES SALAAM LIBRARY, EAST AFRICANA SECTION) CITED IN THIS ARTICLE Ms. 118 Ut. wa Yusufu Ms. 182 Utenzi wa Kisa cha Nabii Yusufu Ms. 183 Ut. wa Yusufu Ms. 333 Yusuf? Ms. 351 Ut. wa Yaaqubu Ms. 352 Ut. wa Yaaqubu Ms. 353 Ut. wa Yusufu (incomplete) Ms. 354a Ut. wa Yusufu (incomplete) Ms. 438 Ut. wa Yusufu (incomplete) Ms. 603 Ut. wa Nabii Yusuf Ms. 708 Kaani niwakhubiri manuscript cultures mc NO 17 87 VIERKE | WRITING SONGS, SINGING TEXTS Article Writing Songs, Singing Texts: Orality and Literacy in Swahili Manuscripts Clarissa Vierke | Bayreuth, Germany 1. Introduction1 The aim of this article is to consider how much a Swahili manuscript can tell us about the oral or written nature of a text. How much is a text a product of writing? How much is it rather a documentation of an oral performance? How much does a poem ‘live’ in the manuscript? And what can the text and manuscript tell us about the oral nature of a composition? Are there some written clues – some textual evidence – about the performance of a text? I will specifically take variability – the many forms of a text – as a hint at the latter. I base my argument on different poetic genres from the eighteenth and nineteenth centuries when Swahili manuscript production flourished. It seems reasonable to assume that Swahili poetry fundamentally relied on oral performance, as it also does nowadays to a large extent: most of the poems were not read silently by individuals, but rather performed out loud; some of them are never actually committed to writing (or only fortuitously). In fact, we do not have any secondary sources that could tell us how poetry was once performed and written on the East African coast in the eighteenth and nineteenth centuries, nor do we have any paper evidence from the seventeenth century or earlier. Thus, I will attempt a kind of ‘textual archaeology’ here and try to interpret the evidence from the manuscripts and texts in a bid to understand the extent to which compositions are shaped by their oral presentation as well as by writing. By looking at the manuscripts in this way, I hope to find some written clues about the repertoire and resources poets and copyists had at their disposal and how they used to compose, 1 Clarissa Vierke was a Petra Kappert Fellow at the Centre for the Study of Manuscript Cultures from 1 April 2017 to 30 June 2017. This article is based on a previous version presented at the workshop entitled ‘One Text – Many Forms’, held at the Centre on 21 and 22 April 2017 and co-organised with the researchers of subgroup C07 of the Sonderforschungsbereich 950, ‘Manuskriptkulturen in Asien, Afrika und Europa’, Universität Hamburg. I am very grateful for all the generous support I have received from CSMC. I thank Annachiara Raia and Ridder Samsom for comments on the text and Kristen de Joseph and Carl Carter for proofreading it. mc NO 17 copy and preserve texts. I will first argue that Swahili manuscripts show traces of oral performance. Secondly, the dependence on oral performance differs greatly from genre to genre. The extent to which a text is tied to paper can vary a great deal, not only from culture to culture, but also within one culture or language. Thirdly, as I will show in the last part of the paper, writing does not simply replace memory and sound as a form of preservation and communication. As I will show, reading, copying and even composing drew very much from oral reception and oral techniques. This paper has been inspired by our recent work on Swahili dance poetry from the late eighteenth and early nineteenth centuries, a genre that was committed to manuscripts chiefly towards the end of the nineteenth century. Together with a group of scholars,2 I have been working on a text edition of poems in this genre as exchanged between the charismatic political leaders of the city-states of Northern Kenya, chiefly Pate, Lamu and Mombasa. As opposed to the narrative tenzi (sing. utenzi)3, which is another mainly narrative genre I had previously worked on, these poems seemed so fragmentary and polyphonic as to constantly evade interpretation. Not only do the voices change from one stanza to the other, but the poems also make broad reference to musical instruments and their sounds. The poems are far more auditory than visual. The dynamic presence evoked in these texts seems to find an echo in their mercurial nature on the manuscript page, on which I have chiefly concentrated in this article. 2 The group includes Abdilalif Aballa, Mahmoud A. Abdulkadir, Ann Biersteker, Annmarie Drury, Jasmin Mahazi, Gudrun Miehe, Ahmed Parkar, Annachiara Raia, Ridder Samsom and Farouk Topan. Stefanie Kolbusa and Natalie Kontny participated in one of the three workshops held in 2015, 2016 and 2017. The last one took place at the Centre for the Study of Manuscript Cultures. All the translations of the Mngumi poetry below are a joint product of the working group. 3 Singular utendi and plural tendi are used in the Northern Swahili dialects of Lamu and Pate. manuscript cultures 88 In the following, I will try to explore these aspects in more detail. To make the contrast clearer and to show the variability of manuscript usage, I will start by describing aspects of the utenzi genre, which is far more biased towards the written page, before turning to poetry of a more oral nature. 2. Written poetry The utenzi is a poetic genre that was predominantly used in the eighteenth and nineteenth century to narrate the battles that took place during the Prophet Muhammad’s lifetime or recount the stories of the prophets, such as Yusuf and Musa.4 These creative and poetic adaptations in Swahili, which mainly drew from Arabic prose source texts, became particularly popular in the course of the nineteenth century.5 The production of manuscripts containing tenzi poetry reached a (final) peak in the first half of the twentieth century, fostered by the demand of philologically inclined European scholars working on Swahili poetry. The writing of Swahili manuscripts did not start in the nineteenth century. By the beginning of the eighteenth century, writing Swahili manuscripts in Arabic script was a well-established practice.6 A local, intellectual elite of Islamic scholars, mostly ulama of Hadramy origin who were well trained in Arabic, started translating important Arabic poetry into Swahili, like the Qaṣīdat al-Hamziyya, the Qaṣīdat al-Burda and the Mawlid Barzanjī.7 They intended to foster Swahili as a written language of learning because they sought to pass on Islamic ideals, doctrines and new forms of piety to a broader public. A vernacular literacy started to flourish, and Swahili manuscript production in Arabic script, mostly containing poetry, became a marker of coastal Swahili identity. 4 See Raia 2017 for an in-depth comparison of the qiṣaṣ al-anbiyā ‘the Prophets’ stories’ and the Utenzi wa Yusuf depicting the life of Prophet Yusuf. For an overview of the relationship of the so-called maghāzi tradition – focusing on the Prophet’s battles and those of his companions – to Swahili tenzi, see Vierke 2011, 420–424 and Abel 1938 for a detailed study. 5 According to Jan Knappert in 1958, the oldest manuscript still existing (that of the Utendi wa Tambuka) dates back to 1728, although this has been called into doubt (e.g. by Zhukow in 1992). However, since 2011 the manuscript has been reported missing at the holdings of the library at the Asia-Africa Institute at the University of Hamburg. Most remaining utenzi manuscripts date back to the nineteenth and twentieth century. VIERKE | WRITING SONGS, SINGING TEXTS Some poetic genres, like the utenzi, were and still are more strongly tied to the written form and co-existed with other genres, like nyimbo (‘songs’), which have hardly ever been committed to paper. Although utenzi was also meant to be recited, chiefly at wake services, it is strongly linked to writing and its representation on the manuscript page. I have shown in the past that its textual nature is shaped by the written medium, but in what follows, I will summarise a few points to underline this idea in contrast to the more oralitybased poems.8 2.1 The text evokes writing Tenzi often highlight their own written character. In the first stanza of the Utendi wa Katirifu (‘Utendi of Katirifu’), 9 which is printed below, writing utensils are evoked such as a stylus (kalamu), good ‘Syrian paper’ (karatasi ya Shamu) and a writing tablet (kibao) on which ‘cords of woven silk’ (upote wa hariri) stretched parallel to each other provided the writer with a grid to follow the lines: Akhi, patia kalamu, ‘O friend, obtain for me a stylus, Na karatasi ya Shamu And some Syrian paper; Na kibao muhakamu A writing tablet well chosen, Na upote wa hariri And some cord of woven silk’ The utensils, on the one hand, echo the Swahili material culture of manuscript writing and seem to reflect a process of composition by writing. However, on the other hand, given that this kind of evocation recurs in many utenzi manuscripts and is also recited in performances, it has become a formulaic way of marking the beginning of the text rather than reflecting the actual situation of writing.10 2.2 The written text as a unified text Unlike many other genres where meta-information about the text like the author and the date of composition is not included, the utenzi presents itself as a self-contained text that does not depend on a knowing audience. Often the author’s name, date of composition and number of verses are printed at the end of the manuscript, as shown in the following sample stanzas: 6 The oldest Swahili writings that have come to light are a number of letters found in Goa dating back to 1724 (Omar and Frankl 1994). Arabic writing in Arabic on the Swahili coast dates back to an even earlier period. For instance, the Lamu chronicle reports that early ‘Arab’ settlers on Lamu exchanged letters in the seventh century (Hichens 1938). 8 Vierke 2014; Vierke 2011, 212–227. 9 Hichens 1939, 121. 7 10 See Vierke 2014; Samsom 2015, 202; Vierke 2016, 228. manuscript cultures Vierke 2011, 217. mc NO 17 89 VIERKE | WRITING SONGS, SINGING TEXTS Name of the composer: Mezoandika, yuwani / mwane Saidi Amini / wa Saidi Uthmani / kabila Mahadalia ‘The one who has written it, know it, is the daughter of Said Amin, the son of Said Uthman of the tribe Mahadali’.11 Date of the composition: Na hijira ya rasuli / alifu na mia mbili / na ishirini na mbili / idadi imetimia ‘And it is 1,222 years after the Prophet’s hijra’.12 Number of stanzas: Na baitize yuani / zilizo karatasani / ni alfu kwa yakini / bila moya kuzidia ‘Know that the verses which are on the paper are exactly one thousand without one more’.13 The meta-information provides the utenzi with categories that make it identifiable and single out the manuscript from numerous instances of performance. The counting of the stanzas, for instance, hints at the written nature of the text; counting is hardly possible while reciting. There is a fixed association between the author, date of composition and the text, which makes it different from other versions of the text and instances of performance. A sense of textual unity also attends the utenzi’s narrative nature: the utenzi typically presents a coherent, closed and plot-driven narrative with an inaugurating formula, an introduction, a main part and an end. Narrative closure is a constant of this form of writing, which contributes to its self-contained nature. By using terms like fixity or unity, I do not mean to suggest that when an utenzi is copied, it is reproduced in exactly the same way on another piece of paper. Stanzas are often omitted or added, and there is a substantial amount of variation at the stanza and word level. The text is variable and is constantly re-enacted in processes of performing and copying it (see below). Yet adding meta-information to each text categorises the individual manuscript; it singles it out from the chain of transmission, reinforcing the idea of an original text by referring to the author and date of composition (and sometimes the copyist, too) as well as by providing the original number of stanzas or the number of stanzas in the copy. 11 Utendi wa Fatuma (‘Utendi of Fatuma’), Dammann 1940, 140; my translation. 12 Utendi wa Fatuma (‘Utendi of Fatuma’), Dammann 1940, 140; my translation. 13 Utenzi wa Abdurahmani na Sufiyani (‘Utendi of Abdurahman and Sufiyan’), El-Buhriy 1961, 118–119. mc NO 17 Fig. 1: The first page of the Utendi wa Haudaji (MS 279888-7, SOAS), showing its method of representing rhythm visually. 2.3 The utenzi’s notions of prosody as related to writing In writing, sounds become visual signs. The temporality of speech, the rhythm, is translated into a spatial arrangement: prosodic units like verses become lines in a manuscript, and pauses are marked by space or, for instance, punctuation or indention. The utenzi shows a familiarity with translating speech time into a spatial form. The symmetrical form of the first manuscript page of the Utendi wa Haudaji (‘Utendi of the Palanquin’) printed below already suggests a regular rhythm.14 In the extract shown above, each utenzi stanza (ubeti; pl. beti) corresponds to one manuscript line (mstari) (see below as well). Space has been used systematically to mark further prosodic breaks: each utenzi stanza (ubeti) consists of four vipande (sg. kipande), i.e. verses of equal syllable length, that yield four strings of words separated by spaces. Accordingly, the manuscript is divided into four columns visually. The Arabic script ensures the equal length of the vipande. The 32 syllables required for each ubeti ideally correspond to 32 consonants per written line; the eight consonants per kipande (verse) are represented by eight syllables.15 14 For a critical edition of the Utendi wa Haudaji, see Vierke 2011. 15 This makes the stanza in Arabic script differ from its representation in Roman script. Apart from the different shapes of the letters, the main reason is that certain phonemic differences are not marked in Arabic script – prenasalised consonants like mb and nd are not differentiated from b and d in Arabic script, for instance. This is in line with verse prosody: both count the same in terms of metrics. manuscript cultures 90 VIERKE | WRITING SONGS, SINGING TEXTS Fig. 2: A manuscript line from the Utendi wa Haudaji (MS 279888-7, SOAS corresponding to a stanza (ubeti) comprised of four parts (vipande). 16 16 Uwambie nimekuya / Maka kwethu nimengiya / kwa utume wa nabiya / Muhamadi mfadhaa. ‘Tell them that I have come, I have entered our home, Mecca, on an errand for the Prophet, the excellent Muhammad.’ Furthermore, as Steere has pointed out, the rhyme at the end of each kipande is ‘to the eye more than to the ear’.17 The rhyme – which consists in the sameness of the last syllable, such as the ya at the end of the first three vipande in the example above (i.e. nimekuya, nimengiya, nabiya) – is more of a visual than an audible hallmark. Since it is not emphasised in Swahili, it is the sameness of the written letter that marks it.18 The rhyme of the ubeti is the same for the first three vipande, but varies from stanza to stanza, while the rhyme at the end of the ubeti runs throughout the whole poem (in the example above, it is aa, which can also change to wa, ya or ʿa). In contrast to older genres where the last letter is not emphasised, the utenzi enforces a certain strictness on the end rhyme, which seems to have been conditioned by its representation in written form. On a manuscript page, the uniformity of the end rhyme enhances the sense of visual symmetry. In short, the concept of prosody is strongly linked to writing. Even units like rhyme – typically a sound feature – correspond to letters, and the syllabic metre (mizani) is a measure of letters. This also corresponds to its form of recitation: the utenzi is typically performed on the basis of a written text, which the performer chants half by reading the text and half by remembering it – an aspect that I will consider more specifically in the last section. 3. ‘Oral poems’ written on paper As I have tried to show, the utenzi very much lives on paper and in written form, a fact that becomes particularly obvious if one compares it to other genres like the utumbuizo and the shairi, which I will look at below. As far as we know, the 16 See Vierke 2011, 500, stanza 102; also see Vierke 2014, 329–334 on the representation of prosodic units on a page. 17 Steere 1870, xii. 18 Also see Abdulaziz 1996, 416. manuscript cultures utumbuizo is one of the oldest Swahili genres, which, to my knowledge, is no longer used to compose poetry nowadays. Most of the works of this genre to have survived are archaic poetic texts in an old language. Although dating of the texts is difficult, there is a considerable gap between the composition of tumbuizo and their commitment to paper: the tumbuizo manuscripts that have survived mostly date back to the nineteenth century, while some utumbuizo traditions, like those associated with the ancient hero Fumo Liyongo, are considerably older.19 Thus, as opposed to the utenzi genre, which seems to have been written from its onset and has adapted themes from written Arabic texts, for the greater part of its history, the utumbuizo was a genre of oral poetry. Only at the end of the nineteenth century did European interest in the local traditions stimulate the locally felt need to preserve what by then already appeared to be an endangered tradition. Still, the manuscripts often bear traces of the rather oral nature of the poem, which I will consider below. 3.1 The incongruence between the audible and visual text First of all, no writing conventions have been established to set prosodic units apart in utumbuizo. Unlike the utenzi, there is no neat analogy between the written and prosodic line, as we can see in the manuscript of the Utumbuizo wa Kikowa (‘The Kikowa Meal’),20 a poem associated with Liyongo. Spacing is not used, but the scribe does use three small circles arranged in the form of a triangle to mark prosodic boundaries. These are used inconsistently, however: while they mark the end of the first line, for instance, the scribe does not use them at the end of the second line. Below, I will use the first two lines to illustrate this inconsistency as well as the fact that the line breaks are arbitrary. Accordingly, there is also no neat end rhyme, which serves as a visual hallmark to denote the end of every poetic line in the utenzi. 19 His life remains a topic of controversy; he has been variously suggested to have lived before the thirteenth century (Shariff 1988, 53), the sixteenth century (Knappert 1979, 68) and the seventeenth century (Freeman Grenville 1973), for instance. 20 Miehe et al. 2004, 36, 37, 89–93. Manuscript MS 47754, Taylor Collection, SOAS, University of London. mc NO 17 91 VIERKE | WRITING SONGS, SINGING TEXTS Manuscript lines Transcription of the units according to how they are depicted in the first two lines of the manuscript21 1. T'eze na Mbwasho na K'undazi pijani phembe / vikomamule na towazi pija / muwiwa / kumbuka / 2. Mwana wa shangazi / yu wapi simba / ezi li kana / mtembezi Fumo wa Sha(u)nga / sikia T'eze na Mbwasho na K'undazi pijiani phembe vikomamule na towazi pija muwiwa kumbuka Mwana wa shangazi yu wapi simba ezi li kana mtembezi Fumo wa Sha(u)nga / sikia Prosodic units Let me dance with Mbwasho and K’undazi Strike for me the horns, the long drums and the cymbals, Strike, you who owe a debt, so that I may remember (my) cousin. Where is the mighty lion? He is like an inveterate wanderer! Fumo of Shanga, reckon well… Fig. 3: First page of the Utumbuizo wa Kikowa (MS 47754, SOAS) and a closer analysis of the first two lines. The coloured text shows the incongruence between prosodic units and the parts of the manuscript visually demarcated by three circles. 21 The slash (/) corresponds to the three circles in the manuscript. mc NO 17 manuscript cultures 92 I have marked the units that conclude with the visual caesura – the three circles – in different colours; one can see at a glance that they are of very different lengths. T’eze na Mbwasho na K’undazi pijani phembe (marked in red), for instance, is the first unit of the manuscript and concludes with three circles. In the manuscript above, I have used the respective colour to underline the part that corresponds to the transcription. If one compares it with the actual rhythm, which I print under ‘prosodic units’ above (where each colour corresponds to the visually marked units of the manuscript), one can see that the first prosodic unit is not delimited in the manuscript: t’eze na Mbwasho na K’undazi is not separated from pijiani phembe by three circles, although it comprises a prosodic (and syntactic) unit that is even marked by rhyme (-zi). The same holds true for the part underlined in blue: there is no marking of the metrical and syntactic break, although there is a clear break after towazi and before pija, as indicated. The green and purple part, by contrast, consists of just one word: here the scribe used more visual breaks (three circles) than the metre requires. Moreover, the line break severs a metrically and syntactically meaningful unit, namely kumbuka Mwana wa shangazi ‘so that I may remember (my) cousin’; mwana wa shangazi ‘my cousin’ (literally ‘the child of my paternal aunt’) appears at the beginning of the second line. I take this as evidence that the visually accurate representation of the text in the manuscript, including the systematic use of visual caesurae (the three circles) and line breaks, did not matter very much, as the use of these features appears to be rather random. The poem’s prosody is hardly captured on the page, even though the reciter would have needed to know how to sing the piece. Unlike the utenzi, where the manuscript seems like a guide on how to read the text rhythmically and where to pause, a great deal of knowledge about prosody and performance is required for the utumbuizo that cannot be inferred from the paper. 3.2 The indeterminacy of text and authorship The impression that the oral dimension of the text is hardly represented by the manuscript is also echoed in the indeterminate nature of the poem. The songs ascribed to Fumo Liyongo are different to utenzi, which present a selfcontained, consistent and coherent narrative. The authorship has not been marked for many of the tumbuizo: in the manuscript above, qala shair (‘the poet spoke’) suffices as a title to set the poem apart from others, and the poems often shade into each other. Not only the authorship, but also data manuscript cultures VIERKE | WRITING SONGS, SINGING TEXTS such as the number of verses and the date of composition has been omitted, just like text-structuring devices such as opening formulas, introductions and conclusions. The single text found in the manuscript seems far less defined and more like a momentary constellation that could easily morph into a different one in another context or on another page. Furthermore, the Liyongo songs hardly display a coherent narrative: the audience is thrown into a scene with a multitude of voices, as even the brief example above shows. One is left wondering who the person is who is talking. Who wants to dance with Mbwasho and Kundazi (in the first line above)? What kind of debt is the poet talking about? Whom does the ‘mighty lion’ refer to? In contrast to the utenzi, where the audience can gradually get all the information they need to understand the narrative, the tumbuizo hardly relate a plot; instead, voices and themes seem to be spliced into each other. The polycentric tumbuizo loosely add up to a cycle of songs: vaguely interlocking and often held together by similar characters or tone, or by occasions of recitation now unknown to us, rather than by a narrative driven by cause and consequence – a stark contrast to the utenzi, which gradually guides the audience into the story and explicitly introduces the characters that will accompany their journey to its successful end.22 The utumbuizo conveys the voice of a narrator whose words can only be understood in light of a larger narrative embedded in a specific situation (of performance) that is not implicit in the words themselves. Although nowadays the tumbuizo are mostly obscure, even on the northern Swahili coast, people on Lamu can still give prose accounts of different episodes of Fumo Liyongo’s life – the gigantic hero who defied all his enemies. While connecting the songs with the episodes is difficult now, previously the audience must have been able to make sense of the songs in light of the overall story of Liyongo, as told, for instance, in the prose version recorded by Steere on Zanzibar in the second half of the nineteenth century.23 Furthermore, it is also possible that the tumbuizo were once interspersed within a larger prose narrative. However, I tend to think that one does not necessarily need to assume a previously unified and plot-driven text, 22 Vierke 2011, 284–288. 23 Steere 1870; see Miehe et al. 2004, 5 as well. There is also an Utenzi wa Fumo Liyongo, which is a more recent poem, however, and also makes an effort to create a consistent narrative from the anecdotes and fragments (Abdalla 1973). mc NO 17 93 VIERKE | WRITING SONGS, SINGING TEXTS which now only seems to exist in a few fragmented songs. Rather, considering the context of an oral culture, where culturally essential texts are common knowledge, I think that a narrative account was neither deemed necessary nor valued. The audience would have taken great pleasure in listening to and decoding new, creative allusions to and evocations of episodes, enjoyed the dynamics of the many voices and relied on their cultural knowledge to interpret the texts. Furthermore, the utumbuizo has certainly been shaped by the context and practice of combined dance and poetry performance, to which the poems also make reference. As opposed to tenzi, which insist on writing materials, tumbuizo typically start by calling for dance, music and instruments, as we see for instance in the Utumbuizo wa Kikowa above. The narrating voice demands to dance – ‘Let me dance with Mbwasho and Kundazi’ – and asks the musicians to ‘strike … the horns, the long drums and the cymbals’. Kucheza ‘to dance’ and ngoma ‘dance’, which are reinforced by reference to the instruments used, like the drums, horns and cymbals, imply more than the involvement and rhythmic movement of the body; they also set the ground for poetic competitions in which only men of high social standing could participate. Poetry provided a platform where questions of debt, marriage alliances and political leadership were addressed, but only in a highly allusive and metaphorical form that is hard to understand nowadays. 3.3 The fluidity of poetry in performance: nineteenth-century mashairi poems Similarly, this also holds true for shairi (pl. mashairi), a genre that is younger than the utumbuizo. This was adapted from Arabic and flourished in Mombasa and the northern parts of the Swahili coast at the end of the eighteenth and nineteenth century. It seems to have shared features with the utumbuizo, like its metaphorical and enigmatic language, for instance, but it also had a link to a tradition of dialogue verse or verbal duelling. Dialogic mashairi offered an arena to playfully act out rivalries between neighbourhoods and dance associations by solving riddles based on word play, but they also served to negotiate burning social issues. The most prominent example is the mashairi exchanged between Muyaka bin Haji al-Ghassany, the court poet of the ruling dynasty of the Mazrui in Mombasa, and Bwana Zahidi Mngumi, the political leader of Lamu at the beginning of the nineteenth century. They hurled provocative lines at each mc NO 17 other, which anticipated a violent confrontation, the Battle of Shela, where Lamu surprisingly defeated the strong coalition of Mombasa and Pate.24 Zahidi Mngumi earned an enduring reputation, and the poems he exchanged with his opponents became popular throughout the region. They also made their way into chronicles25 recorded in the nineteenth century as well as poetic anthologies preserved in manuscripts, like the ones presented below showing the first six stanzas. In this selection, Bwana Zahidi Mngumi is calling on his fellow Lamu men to gather for a dance, where they can reach an agreement to band together and fight the common enemy from Pate. Bwana Zahidi Mngumi underlines his certainty: he sees through the plans of the enemies (stanza 4). He knows that war is going to come, but Lamu is prepared for the fight and to protect its citizens: ‘we have closed the doors’ (stanza 1). In the first manuscript (HH), which the renowned scribe Muhamadi Kijuma (c.1885–1945) copied for Ernst Dammann, the scribe makes an effort to write in a way that is easily decipherable for his European client. He separates the stanzas, each composed of four half-lines (vipande), from each other with a line.26 The inverted heart divides each line into two vipande. The stanza is presented as a quatrain in the transcription below. In the second manuscript (SBB), written by an unknown scribe, indentations are used to mark off stanzas, and circles (or hastily drawn inverted hearts?) have been added to separate vipande. The transcription of each manuscript is included to highlight their variations. They start with the same stanzas. SBB has an additional stanza not found in HH. In stanza four, HH and SBB start to differ in terms of how the stanzas are ordered. 24 Biersteker and Shariff 1995. 25 See the Lamu chronicle, for instance (Hichens 1938). 26 The scribe Muhamadi Kijuma, who copied an enormous number of Swahili manuscripts for chiefly British and German scholars, adapted his style of writing and also introduced some calligraphic elements to please his European clients. His style is markedly different from that of other scribes. manuscript cultures 94 VIERKE | WRITING SONGS, SINGING TEXTS Fig. 4: Example of a ‘war poem’ by Bwana Zahidi Mngumi: the first page of the poem Nenda wa Asha Hamadi, as written by Muhamadi Kijuma on commission for a European scholar (MS 1219 H56, Universität Hamburg, Seminar für Afrikanische Sprachen und Kulturen; labelled ‘HH’ below). The second manuscript (Hs. or. 9954, State Library of Berlin, Oriental Department; labelled ‘SBB’) was written by an unknown scribe. manuscript cultures mc NO 17 95 VIERKE | WRITING SONGS, SINGING TEXTS HH SBB 1. Nendra na Asha Hamadi – wakhubirini malenga Ayao napije hodi – milango tumeifunga tumeikomeza midi – na magogo ya kupinga kupa hirimu kutanga – kivuza kilicho ndrani 1. Zizi na Asha Hamadi – wakhubirini malenga Ayao napije hodi – milango tumeifunga tumeikomea midi – na magogo ya kupinga kupa hirimu kutanga – kivuza kilicho ndrani Go with Asha Hamadi/Zizi and Asha Hamadi, proclaim to the master poets Anyone who comes should say ‘hodi’, we have closed the doors We have bolted the doors and [put] logs [there] to block the way So those outside will wonder and ask [themselves] what is inside 2. Ya mshabaha vurungu – avumizapo muyumbe khusi zenu matawangu – nazandamane na phembe tukeshe yothe makungu – kula mwenye lake ambe maneno tusiyafumbe – wala tusiyakhizini 2. Ya mshabaha vurungu – avumizapo muyumbe khusi zenu matawangu – nazandamane na phembe tukeshe yothe makungu – kula mwenye lake ambe maneno tusiyafumbe – walau tusiyakhizini It is like a wind instrument, when the messenger blows it forcefully Your clapping, beautiful women, should accompany the horns Let us stay awake the whole night, so everyone may speak out frankly Let us not speak in riddles nor hold back our thoughts Thathasa patu mpambe – patanisha zibod Avumizapo muyumbe – siwa yetu ya Mreo Na yand5amane na phembe – mishindro iwe thangao Tukumbuke yapitao – mapotofu ya zamani No corresponding stanza in HH Strike the cymbal, beautiful lady, beat with the braided palm leaves When the messenger blows our siwa from Proud Lamu It should accompany the horn, the sound should announce [the dance] Let us remember the past, the past stupidity/evil 3. Tukumbuke ulimwengu – umezozinga duniya kizazi na moya fungu – kupindruliyana niya kuzengeleyana phingu – kutiwa wenye kutiya kula mthu mwenye haya – achambiwa mtiyeni Tukumbuke ulimwengu – umezozinga duniya kizazi na moya fungu – kupindruliyana niya kuzengeleyana phingu – kutiwa wenye kutiya kula mthu mwenye haya – achambiwa mtiyeni Let us remember how the ways of the world have changed The descendants of the same clan betray each other Putting each other into fetters, the shacklers are being shackled Everyone involved is told: ‘Imprison him!’ After the third stanza, which is found in both manuscripts, the arrangement of stanzas varies, as shown below. mc NO 17 manuscript cultures 96 VIERKE | WRITING SONGS, SINGING TEXTS HH SBB 4. Kimbelembele waume – cha wakokota zisiki Taka nasi kitwandame – kwa ukhadimu wa nyuki pindrani kisitwegeme – kwa muomo wa bunduki tufiliye yetu haki – tusiwekwe tuli duni Kula mwenye mbovu niya – mwenye dhamiri nasi Rabbi tatuhukumiya – twokowe zethu nafusi kwa auni ya Jaliya – aliyo yuu Mkwasi kufuata mfuasi – hilo halipatikani Rush on, men, the way people drag tree stumps They want mishap to befall us, let us move forward like the diligent bees Let us strive so that misfortune will not befall us through the barrel of the gun Indeed, let us fight to the death for our rights so we won’t be regarded as worthless All those with evil intentions and ill will towards us, God will judge them for us and may He save our souls, By the help of God, the Most High, the Self-Sufficient, To follow the follower – that will never happen! 5. Kula mwenye niya mbovu – ngwampindrukiza yeye mwoka kiwana cha ivu – nale na aliyo naye aifanyao mwelevu –penye uwinga akae asikome mayutoye – afiliye mayutoni Kula mwenye niya mbovu – ngwampindrukiza yeye mwoka kiwana cha ivu – nale na aliyo naye aifanyao welevu –penye uwinga akae asikome mayutoye – afiliye mayutoni Anyone with an evil intention, may God turn it on him The baker who ruined the bread, let him share it with his companion The one who pretends to be clever, let him remain in ignorance May his regrets be endless, may he die with them Anyone with an evil intention, may God turn it on him The baker who ruined the bread, let him share it with his companion The one who pretends to be clever, let him remain in ignorance May his regrets be endless, may he die with them 6. Kula mwenye mbovu niya – mwenye dhamiri na sisi Rabbi tatuhukumiya – twokowe zethu nafusi kwa auni ya Jaliya – aliyo yuu Mkwasi kufuata mfuasi – hilo halipatikani Kimbelembele waume – cha wakokota zisiki Takanani ktwandame – kwa ukhadimu wa nyuki pindrani kisitwegeme – kwa muomo wa bunduki tufiliye yetu haki – tusiwekwe tuli duni All those with evil intentions and ill will towards us God will judge them for us and may He save our souls, By the help of God, the Most High, the Self-Sufficient, To follow the follower – that will never happen! Rush on, men, the way people drag tree stumps They want mishap to befall us; let us move forward like the diligent bees Let us strive so that misfortune will not befall us through the barrel of the gun Indeed, let us fight to the death for our rights so we won’t be regarded as worthless. Like many other poems ascribed to Bwana Zahidi Mngumia and his opponents, the poem was copied many times and was also memorised until quite recently. The different sources vary considerably, not only in terms of the length of the poems and the sequence of the lines, but also in terms of the ascribed authorship found in the sources.27 The following table is meant to present these variations in a simplified schematic form, which allows us to take more sources into consideration. The column at the far left gives the name of the poem. Apart from Nenda wa Asha Hamadi, I also refer to another poem, La Kutenda Situuze (‘Don’t Ask Us What We Should Do’), since in one source, DA, the 27 Also see Miehe 1976. manuscript cultures stanzas of the latter are part of Nenda wa Asha Hamadi. The ‘Verse’ column lists the beginning of the stanzas. In the top row, abbreviations like HH (also found in Fig. 4) and FA stand for the various sources of the poem.28 28 Apart from three manuscripts in Arabic script (HH, FA and SBB), I have also taken published and unpublished editions and notes based on manuscripts into consideration: HH = manuscript, Universität Hamburg, Seminar für Afrikanische Sprachen und Kulturen, MS 1219 H56; FA = manuscript written by Faraj Bwana Mkuu, Nabahany Collection, Mombasa; SBB = manuscript, State Library of Berlin, Oriental Department, MS Hs. or. 9954; DA = transcript (of SBB) by Ernst Dammann with notes and stanzas added by Ahmed Nabahany, DEVA Bayreuth. MA = unpublished booklet: Maawy, Ali A. A. (n. d.), The Lamu Hero: The Story of Bwana Zahidi Mngumi (1760–1832). mc NO 17 97 VIERKE | WRITING SONGS, SINGING TEXTS Table 1: An overview of stanza variation in the sources of Nenda na Asha Hamadi. Poem Verse HH FA SBB DA NA MA SHA BSH Nenda na Asha Hamadi 1 = ZM 1 = ZM 1= ? 1= SAA 1 = ZM 1 = ZM 1 = ZM 1 = ZM Ya mshabaka vurungu 2 2 2 2 3 2 2 2 3 3 3 2 3 3 3 Tatasa patu mpambe Nenda na Asha Hamadi Tukumbuke ulimwengu 3 4 4 4 4 4 4 4 Kimbelembele waume 4 7 7 7 5 7 7 7 Kulla mwenye nia mbovu 5 6 6 6 6 6 6 Kulla mwenye mbovu nia 6 5 5 5 6 5 5 5 7 1 = ZM 1 = ZM 1 = ZM 8 2 2 2 3 3 3 Tuna konde za asili Mla shokowa la Pemba 7 Phani shauri Suudi La Kutenda Situuze 8 1 8 9 2 9 La kutenda situuze 1 1 3 = ZM 10 1 = AJ 1 = SAA 1 = SAA 1 = AJ Hela tunda kwa dalili 2 2 4 = AJ 11 2 1 = AJ 1 = AJ 2 As Table 1 shows, these differ first in terms of their length: Nenda wa Asha Hamadi has seven stanzas in HH, but nine in FA, for instance. Secondly, they also differ with respect to the order of stanzas. Thirdly, the boundary between texts, i.e. the question of where one poem ends and the next one starts, is not clear: in DA, for instance, Nenda na Asha Hamadi is one poem of eleven stanzas. However, in many other sources (HH, FA, NA and BSH), the last two stanzas are part of a separate poem, La Kutenda Situuze. The colours are additionally meant to mark the different textual boundaries. In MA and SHA, the two stanzas of La Kutenda Situuze do NA = unpublished paper by Nabahany, Ahmed Sheikh (n. d.), Mapisi ya Bwana Zahidi Mngumi aliyekuwa mwenye mui yaana raisi wa Kiwandeo (Lamu). CHI = Chiraghdin 1987, 13. SHA = Shariff 1988, 119. BSH = Biersteker and Shariff 1995, 21, 22. mc NO 17 not form one poem, but should be attributed to two different authors. The attributions of authorship vary considerably as well: what is clearly attributed to Zahidi Mngumi (ZM) in some sources is attributed to Sheikh Ali bin Ahmed (SAA) or Abubakari bin Jabir (AJ) in others. In order to understand the text’s life beyond the manuscript, parallel or prior to it, one must consider the degree of variation it exhibits. First of all, the poems were very popular in Lamu as well as in Mombasa; in fact, they may have changed by being repeated aloud and copied so often.29 As I will try to argue below, copying, like recitation in performance, does not preclude variation, but on the contrary, it fosters it. In the context of their oral performance, which has to be kept fluid and where the text takes shape on the spot, the performer 29 See Miehe 1976 for a detailed comparison of poetic lines attributed to Muyaka bin Haji or Zahidi Mngumi. manuscript cultures 98 rearranges lines. The compact nature of the poetic line, which is a clearly defined syntactic, semantic and metrical unit, makes it more easily detachable and reusable in another context: the lines stick in the mind of the poet, who can make use of them again. This is even more so for the stanza: according to local poetic conventions, the ubeti needs to adhere to the rule of kutosheleza (‘to be self-contained’).30 It is a compact unit that is apt to be rearranged within new texts and contexts. Furthermore, given the polyphonic nature of the poetry, there is no narrative sequence that could be destroyed by rearranging stanzas or leaving a stanza out. This is in line with what Barber highlights as preconditions for detaching the text from a specific context and making it repeatable so that it can be performed in a new context; Barber considers this a form of ‘entextualisation’.31 In her view, the lack of deictic definiteness in time, space and, as one might add, the sequence of events driven by cause and consequence contributes to the lack of definite reference, i.e. the text’s ‘deliberate indeterminacy’ along with the strongly allusive or metaphorical nature of the text.32 They foster the extraction and rearrangement of the text in a new context where stanzas can acquire new meaning.33 Accordingly, lines and stanzas also ‘slip’ easily from one poem to another, making the boundaries between the texts unclear. The poet (and the performer and copyist as well) relies on a repertoire of poetic lines and formulas that ideally fit the metrical environment. Therefore, the question of authorship becomes elusive because poets rely on their stock repertoires and take lines from one context to use in others, so for them and for the audience, the idea of original authorship does not exist as it does for the utenzi. The line between the performer, scribe and composer becomes blurred in such a context. In this case, poetic speech, even if it is written down, is full of preconceived lines that are taken over wholesale. The mashairi, which originate in a context of oral performance, namely kujibizana, or dialogic poetry, have to be composed on the spot in the context of performance, and the poet has to give a swift reply to his or her opponent. He or she therefore makes use of quotes and recycled formulas and builds on the words of his or her predecessor and other poets. It is also possible that Zahidi 30 Abdulaziz 1996, 416. 31 Barber 2007, 22. 32 Barber 2007, 91. 33 Barber 2007, 72–74. manuscript cultures VIERKE | WRITING SONGS, SINGING TEXTS Mngumi took over stanzas and verses from his opponents and vice versa. Moreover, after his death, in writing down, copying or re-performing his poetry, performers and scribes may have added lines that other sources attribute to other poets. We also find many of Muyaka’s lines in the aphorisms collected by William Taylor, where it is often not even clear whether the aphorism is a line quoted from Muyaka bin Haji’s poem or, conversely, if Muyaka bin Haji quoted a line that was already in circulation.34 Thus, to conclude, the largely oral performance of the text – where the text only takes concrete shape in performance, and where the poet relies on a repertoire that he or she can ideally rearrange – corresponds to the undetermined nature of the text, which appears obscure as its meaning lies somewhere other than in the words as such. On the one hand, it largely depends on interpretation, and on the other, its indeterminacy also allows for textual variability, i.e. the rearrangement of the text. This fluid but formulaic nature of the text is echoed by its presentation in writing as well as by textual variability from one manuscript to the other. Most of the time, unless written for a European audience, a manuscript is of a secondary nature, documenting some lines whose collocation appears random: they may easily end up in a new constellation in the next document. Authorship is also variable. In Kenya, I have occasionally heard people quarrelling over authorship and the fact that ‘poets steal from other poets’, attributing this to the present-day lack of care and concern for old poetry. Yet I rather think that the fluidity of lines can be attributed to the performative aspect of the poetry and the obscure language full of mafumbo ‘riddles’ that emerges from and contributes to it. The poem is quite deliberately formed from a reservoir of existing lines and imagery. On the manuscript page, the poem becomes a written poem, but in a much more restricted sense compared to the utenzi. Memory and knowledge of the text play an important role, such that accurate marking of prosodic units, punctuation, line breaks and spacing, not to mention a full account of the story, are not deemed necessary. In a context where the rearrangement of lines to build a powerful poetic statement is more important than questions of authorship, not only textual boundaries but also questions of textual length become relative. Each text exists only as a momentary constellation of poetic lines. 34 Taylor 1891. mc NO 17 99 VIERKE | WRITING SONGS, SINGING TEXTS Lastly, the minor importance attributed to the manuscripts is also echoed by the hasty writing style in manuscripts that were not commissioned by European scholars: the orthography in them is inconsistent. Nor do calligraphy or ornamentation play a role, as we notice in the following lines attributed to Fumo Liyongo, which Mwalimu Sikujua wrote towards the end of the nineteenth century, where words have been crossed out and letters written down in a hurry. Furthermore, the weak manuscript conservation practices on the Swahili coast35 could also be considered indicative of the minor importance attributed to Swahili manuscripts. In this sense, Swahili manuscripts are also treated differently compared to the carefully adorned and valuable Arabic books and manuscripts from the coast, which are often better preserved. Thus, at the same time and in the same place, there were fundamentally different conventions for writing Arabic and for writing Swahili. Arabic and Swahili texts depend on different ideas of literacy: while in a Swahili literary environment each Arabic text is essentially a written text, and thus the written word is cherished, only some aspects of Swahili texts, which are part of a bigger oral tradition, are presented in the manuscripts.36 A reader needs a great deal of predictive knowledge about the text to be able to access it. 4. Copying as a ‘scribal performance’ I would like to argue for Swahili manuscripts’ inclination towards orality by considering one more aspect that has not been explored with respect to Swahili poetry: oral techniques as used in reading and copying manuscripts. The basic idea is that copying a manuscript in this predominantly oral context is different from copying in the sense of deciphering manuscripts word by word, letter by letter, and reproducing the same signs on paper again. Rather, in copying, the scribe or copyist voices the text, performs it and thereby introduces variation. I am drawing on Katherine O’Brien O’Keeffe’s Visible Song, a study of Old English manuscripts, in which she considers the many variants that occur in the texts as traces of the residue of oral practice. In copying texts, scribes refashion them, because the copyist recasts the text in his or her own voice before writing it down. Accordingly, a text 35 Fig. 6: Example of a hastily written, unadorned Swahili manuscript page of the Utumbuizo wa Dhiki (‘Song of Agony’), most probably written by Mwalimu Sikujua (MS Nabahany Collection, Mombasa). changes from manuscript to manuscript similarly to how the text changes from oral performance to oral performance. The copyist relies on techniques known from the context of oral performances. In a similar vein, Doane describes a ‘scribal performance’ in the following way: Whenever scribes who are part of the oral traditional culture write or copy traditional oral works, they do not merely mechanically hand them down; they rehear them, ‘mouth’ them, ‘re-perform’ them in the act of writing in such a way that the text may change but remain authentic, just as a completely oral poet’s text changes from performance to performance without losing authenticity.37 In such contexts, the idea of unfaithfulness to the text or of textual corruption does not make any sense. The scribe does not reproduce a visual sign on paper, but is foremost a reader who reads aloud, makes sense of the text and later ‘translates’ his voice into visual signs again. There are several ‘traces’ of this technique of revoicing found in the manuscripts, which I will consider below. Samsom 2015, 207 36 O’Brien O’Keeffe 1990, 19 highlights a similar contrast between Old English vernacular literature and Latin texts; while Latin was predominantly a written language, never spoken, Old English was a spoken language as well as the language of commentary – and this dichotomy is also reflected in the writing style. mc NO 17 37 Doane 1991, 80, 91. manuscript cultures 100 4.1 Dialectal variation A scribe may adapt the text to his own dialect, for instance. In the Swahili context, there a host of examples of dialectal variation in different versions of the same text. On the one hand, Kiamu – the dialect of the city-state of Lamu in Northern Kenya, which turned into a cultural and literary centre in the course of the nineteenth century – became a prestigious literary register that was also used to write poetry well away from Lamu. In Northern Mozambique, for instance, poems in Kiamu were not only circulated, but local poets also made a deliberate effort to use Kiamu features in their poetic compositions. On the other hand, particularly in the rival city-state of Mombasa, texts and stanzas composed in Kiamu were adapted to the local Kimvita. Even in the venerated poem of Al-Inkishafi – composed on Pate by Said Abdallah Said Ali bin Nasir bin Sheikh Abubakr bin Salim – the famous scribe and copyist Mwalimu Sikujua from Mombasa did not stick to the mostly Kiamu forms of the text, but brought in his own dialect, as echoed for instance in sound features (alveolar instead of rhotacised nd and nt) as well as in forms like ikomile ‘it has finished’ (stanza 23) or the class prefix vi-, for instance in vitwa instead of zitwa ‘heads’ (stanza 35), the form normally used in Kiamu.38 Sometimes, scholars referred to his manuscript version as a ‘southern recension’ (Taylor 1915) whose divergence from the northern version stirred some discussion.39 Even so, it was hardly a deliberate effort to produce a Mombasan version of the text, as Kimvita, the Mombasan dialect, is not used exclusively or consistently. While in his edition, Hichens dismisses the many variants as false ‘corrections’ or ‘copyist errors’ or gap-fillers, I rather consider them to reflect the scribe’s practice of recasting the text in his own words.40 Therefore, not only variations from different dialects are introduced but also transpositions or interpolations of lines, which Hichens notes as mistakes. 4.2 Quoting sound Furthermore, the way of writing Arabic words and lines embedded in Swahili poetry, i.e. not according to Arabic orthography, but according to local pronunciation, can be considered further evidence of the practice of voicing text. For instance, the lines quoted from the Quran in the Swahili VIERKE | WRITING SONGS, SINGING TEXTS version of the ‘Story of Yusuf’ (Utenzi wa Yusufu) appear to violate Arabic orthography.41 Muhamadi Kijuma, the scribe, seems to have quoted the Quran by heart, whereby he ‘corrupted’ the Arabic lines. This seems to suggest that he did not refer to the Quran in order to copy from it and adhere to its conventional orthography, but quoted the lines from memory, i.e. how he remembered the sound.42 4.3 Half-reading the text There are further hints of practices linked to an oral culture. O’Brien O’Keeffe suggests that rather than copying lines word by word, scribes of Old English verse half-predicted the text while writing it down.43 Thus, the copyist only half-read the text and ‘guessed’ the rest of the text, which is possible because the text is so dependent on formulas. Thus, the reader and copyist used his or her ‘knowledge of the conventions of the verse to “predict” what is on the page’. 44While working on old manuscripts with the poet and scholar Ahmed Sheikh Nabahany, I often saw him copying an utenzi in a similar way: he started reading a stanza, but often would not finish reading it – as I saw his eyes moving away from the paper – instead adding the rest of the line or stanza from memory. This technique is possible because many lines and stanzas make use of formulas that ideally fit the metrical pattern of the utenzi, so if one knows the beginning, one can fill in the rest of the line. In the line kwa dunia na akhera ‘in this world and the hereafter’, for instance, he would only have to read kwa dunia to be able to add na akhera from memory, just as he could add whole lines from the many stock phrases he had memorised. He also offered to ‘fix’ lines that had become unreadable in such a way. The variants in Nenda Asha wa Hamadi listed above provide another example. On the one hand, one finds variants that simply come from reading the manuscript in different ways. In stanza four (see above), for instance, Nabahany, who transliterated the manuscript in the 1980s, read kimbelembele as kipilipili, which can be attributed to a different reading of the Arabic script. (In Arabic script, there is often no difference marked between be and pe, nor are nasals marked.) Nyuki, which Nabahany transcribed as nyoki 41 Raia 2017, 146 and Raia forthcoming. 42 In a similar way, many Arabic loanwords figuring in the text do not appear in their Arabic orthography, their standardised visual shape, but how the poet hears and writes them. 38 See, for, instance, the editions by Hichens 1939 and Taylor 1915. 39 See for instance, Hichens 1939, 108. 43 O’Brien O’Keeffe 1990, 40. Hichens 1939, 109. 44 O’Brien O’Keeffe 1990, 40. 40 manuscript cultures mc NO 17 101 VIERKE | WRITING SONGS, SINGING TEXTS and Maawy as wanyoki in the second line of the same stanza, is a similar example, as the reading of vowels as well as that of word boundaries is subject to interpretation. On the other hand, however, there is another kind of variability found in the first line of the same poem: other sources do not say Nenda Asha wa Hamadi, as we find the manuscript (HH) written by Kijuma, but Zizi na Asha Hamadi (CHI and BSH) or Zizi Asha na Hamadi (MA). These are all variants that could hardly come from copyists reading the letters differently; rather, they hint at processes of half-reading, where the obscurity of the reference – who is or who are Zizi and Asha? – fosters variability in reading, and hence reformulation. These variants are also possible within the metrical framework of the verse. A copyist not only copied a manuscript, but also made considerable use of his reservoir of formulas. The poet halfreads the text and then fills in the gaps, thereby producing variants in many cases. Lexical variation, where we find certain words replaced by others, can be considered as evidence of this. Consider stanza three of Nenda Asha wa Hamadi: the second half-line of Tukumbuke ulimwengu – umezozinga duniya ‘Let us remember how the ways of the world have changed’ appears as ugeukao duniya in the manuscript from Berlin (SBB): here the copyist used a synonym of -zinga, namely the verb -geuka ‘to change, to turn’. In stanza four (see above), tusiwekwe tuli duni ‘so that we won’t be regarded as worthless’ appears as tusiweke cheo duni ‘so that we won’t have a low rank’ in many sources (SBB, MA, NA, SHA, BSH, FA); in DA, it even takes the form of tusiwe kishindoni ‘so that we may not quarrel’. When considering Swahili poetry, I have often been struck by the enormous amount of textual variation that exists. Multiple versions of one text are never the same, even if they were copied again and again by the same copyist.45 The copyist can add or omit lines or stanzas, but he or she also introduces variation at the level of the line or stanza. Still, generally speaking, variants do little to alter the basic narrative or theme of the text and, as I hoped to show here, are often based on formulas. Variants emerge, conditioned and restricted by the metrical, syntactic and semantic framework of the stanza, and often consist of preexisting strings of words that are part of the poet’s mental 45 This became most clear to me while I was working on the manuscripts copied by Muhamadi Kijuma, who copied many poems several times to sell them to European scholars, in particular William Hichens in London and Ernst Dammann in Hamburg. mc NO 17 vocabulary. Therefore, the practice of copying is similar to the oral performance of a text, in which the performer essentially resorts to stock phrases that he can adapt to the context of performance. In both cases, literal faithfulness to a previous text is not the primary concern: neither the written document nor the previous oral version is more authentic than the current oral version or document. No effort is made to reproduce a text verbatim, either in a performance of it or in copying. On the one hand, one can consider formulaic techniques as being conditioned by the necessity to keep the oral presentation fluid and, accordingly, also as a way to cope with the idiosyncrasy and variability of handwriting and the lack of a standard orthography, which also turns reading and copying Swahili manuscripts into a difficult task. The scribe necessarily resorts to the practice of improvisation, which he knows from contexts of oral performance. This view would be in line with the functional perspective that dominates the debate on oral formulaic theory. However, it fails to recognise the active role of the scribe and his or her engagement with the text. In this context, copying does not mean reproducing the poem mechanically, but ‘re-voicing’ the text. The reader also contemplates the text. Copying is thus not a dull technique with which to reproduce a text, but rather an inspiring and stimulating exercise in the same vein as memorisation and recitation, which are both appreciated.46 Learning the Quran by heart, for instance, is not only a way of preserving the text, but a form of contemplation, which implies that understanding comes gradually, through repetition, which continuous engagement with the text renders possible. Similarly, for Swahili poetry, the copyist appropriates the text, and the variability echoes this active engagement with it. Apart from variability at the level of the stanza, copyists also often add stanzas in an effort to reformulate the scene. Tenzi in particular tend to grow over time, as the scribe, in voicing the text, contemplates scenes and ideas; while pondering them, he amplifies them by adding more details or stressing certain sentiments, for instance.47 The scribe recites and rethinks the pre-existing text in writing and reshapes it, which is easily conflated with composing.48 46 Furthermore, as Samsom 2013 argues, writing as such might also be considered an adequate way of not only preserving, but also venerating the divine text. 47 See Raia 2017, 103–117 and Vierke 2016. 48 O’Brien O’Keeffe 1990, 41 also underlines the similarity between reading, copying and composing because formulaic language plays an important role in all of these activities. manuscript cultures 102 5. Conclusion A main idea of this paper is to underline the fact that writing is not a straightforward process, but can take on different roles and functions. On the one hand, it can be strongly linked with the very idea of the poem, such that prosodic features become visual features of the script. On the other hand, however, writing can be so entangled with orality that the question of what a text is, where it starts and where it ends becomes difficult to answer – and also reading is linked with memorisation. In the latter context, each version of the text is but a temporary constellation of parts, which recur in a different arrangement in the next manuscript and performance. Questions about the original text are elusive in this context since they imply the idea of erosion, namely of a unified text that gradually diverges over time. The samples of tumbuizo and mashairi taken into consideration here, however, suggest that the constant flux of preconceived strings of words and newly created lines, which then become part of a repertoire, is rather a rule of composition. In the Swahili context where not only the spoken word but the repertoire of poems is key, composing tends to be conflated with copying and reading anyway. In this sense, it is rather the utenzi with its strong emphasis on writing and fixity despite its own variability that is the more peculiar case. Probably also because it is clearly derived from a practice of adaptation or translation that emphasises the Arabic original, it more closely mimics the use, reading and writing of Arabic texts in the Swahili context VIERKE | WRITING SONGS, SINGING TEXTS REFERENCES Abdalla, Abdilatif (ed.) (1973), Utenzi wa Fumo Liyongo, Dar es Salaam: Chuo cha Uchunguzi wa Lugha ya Kiswahili, Chuo Kikuu cha Dar es Salaam. Abdulaziz, Mohamad (1979), Muyaka: 19th Century Swahili Popular Poetry, Nairobi: Kenya Literature Bureau. —— (1996), ‘The Influence of the Qasida on the Development of Swahili Rhymed and Metred Verse’, in Stefan Sperl and Christopher Shackle (eds), Qasida Poetry in Islamic Asia and Africa, Leiden: Brill, 411–428. Abel, Martin (1938), Die arabische Vorlage des SuaheliEpos Chuo Cha Herkal: Textkrit. Ed. u. Übers. Ein Beitr. zur Kenntnis d. legendären Maġāzi-Literatur (Zeitschrift für Eingeborenensprachen, Beiheft 18), Berlin: Reimer. Barber, Karen (2007), The Anthropology of Texts, Persons and Publics: Oral and Written Culture in Africa and Beyond, Cambridge: Cambridge University Press. Biersteker, Ann, and Ibrahim Noor Shariff (1995), Mashairi ya Vita vya Kuduhu. War Poetry in Kiswahili Exchanged at the Time of the Battle of Kuduhu, East Lansing: Michigan State University Press. Chiraghdin, Shihabuddin (1987), Malenga wa Karne Moja: Utangulizi na Ahmed Sheikh Nabahany, Nairobi: Longman. Dammann, Ernst (1940), Dichtungen in der Lamu-Mundart des Suaheli: gesammelt, herausgegeben und übersetzt von Ernst Dammann, Hamburg: Friederichsen, De Gruyter & Co. Doane, Alger Nick (1991), ‘Oral Texts, Intertexts, and Intratexts: Editing Old English’, in Jay Clayton and Eric Rothstein (eds), Influence and Intertextuality in Literary History, Madison: University of Wisconsin Press, 75–113. El-Buhriy, Hemed Abdallah Saidi Abdallah Masudi (1961), Utenzi wa Abdirrahmani na Sufiyani: The History of Abdurrahman and Sufian, edited by John Allen, Dar es Salaam: East African Literature Bureau. manuscript cultures mc NO 17 VIERKE | WRITING SONGS, SINGING TEXTS Grenville, Freeman (1973), Chronology of African History, Oxford: Oxford University Press. Hichens, William (1938), ‘The Lamu Chronicle’, Bantu Studies, 12: 2–33. —— (ed.) (1939), Al-Inkishafi: The Soul’s Awakening, London: Sheldon Press. Knappert, Jan (1958), Het epos van Heraklios: een proeve van Swahili poëzie, tekst en vertaling, voorzien van inleiding, kritisch commentaar en aantekeningen, PhD thesis, Leiden University, Alkmaar: Hofman. —— (1979), Four Centuries of Swahili Verse: A Literary History and Anthology, London: Heinemann. Miehe, Gudrun, Abdilatif Abdalla, Nassir Juma Ahmad Bhalo, Ahmed Nabahany, Angelica Baschiera, Clarissa Dittemer, Farouk Topan, Mohamed Abdulaziz, Said Ahmed Mohamed Khamis, Yahya Ali Omar, and Zeina Mahmoud Fadhil Al-Bakary (2004), Liyongo Songs: Poems Attributed to Fumo Liyongo, Cologne: Köppe. —— (1976), ‘Bemerkungen zur Muyaka-Rezeption in Lamu’, Africana Marburgensia, 9/1: 33–58. ——, and Vierke, Clarissa (eds, 2010), Muhamadi Kijuma: Texts from the Dammann Papers and Other Collections, edited in collaboration with Sauda Barwani and Ahmed Sheikh Nabahany, Cologne: Köppe. O’Brien O’Keeffe, Katherine (1990), Visible Song: Transitional Literacy in Old English Verse (Cambridge: Cambridge University Press). Omar, Yahya Ali, and Peter J. Frankl (1994), ‘A 12/18th Century Swahili Letter from Kilwa Kisiwani (Being a Study of One Folio from the Goa Archives)’, Afrika und Übersee, 77: 263–272. mc NO 17 103 Raia, Annachiara (2017), The Utendi wa Yusufu: A Critical Edition of the Swahili Poem of Yusuf and a Study of its Adaptation at the Swahili Coast, unpublished PhD thesis, University of Bayreuth and University of Naples L’Orientale. —— (forthcoming), ‘Āya na Ubeti: Qur'ānic Quotations in Muhamadi Kijuma’s Swahili Tendi Manuscripts: Citation Aspects and the Authorial Intervention’, in Antonella Brita, Giovanni Ciotti, Florinda De Simini and Amneris Roselli (eds), Textual and Material Craftsmanship: What Does Copying a Manuscript Mean, Naples: Dipartimento Asia, Africa e Mediterraneo. Samsom, Ridder (2013), ‘Why write in the abandoned Arabo-Swahili script?’, manuscript of the month 11/2013, Hamburg: CSMC <https://www.manuscript-cultures.unihamburg.de/mom/2013_11_mom_e.html>. —— (2015), ‘The Dissolved Collection of Sheikh Aliy Ahmed Abdallah al-Buhriy (1889–1957)’, Islamic Africa, 6: 201–208. Shariff, Ibrahim Noor (1988), Tungo Zetu: Msingi wa Mashairi na Tungo Nyinginezo, Trenton, NJ: The Red Sea Press. Steere, Edward (1870), Swahili Tales, as Told by Natives of Zanzibar, London: Bell & Daldy. Taylor, William (1891), African Aphorisms: Or, Saws from Swahili-Land, London: Society for Promoting Christian Knowledge. —— (1915), ‘Inkishafi, A Swahili Speculum Mundi’, in Chauceny Stigand (ed.), A Grammar of the Dialectic Changes in the Kiswahili Language: With an Introduction and a Recension and Poetical Translation of the Poem INKISHAFI, A Swahili Speculum Mundi by the Rev. W. E. Taylor, M.A., Cambridge: Cambridge University Press, 73–105. manuscript cultures 104 Vierke, Clarissa (2011), On the Poetics of the Utendi: A Critical Edition of the Nineteenth-Century Swahili Poem ‘Utendi wa Haudaji’ Together with a Stylistic Analysis, Berlin: LIT. —— (2014), ‘Akhi patia kalamu: Writing Swahili Poetry in Arabic Script’, in Meikal Mumin and Kees Versteegh (eds), Arabic Script in Africa: Studies on the Usage of a Writing System, Leiden: Brill, 311–331. —— (2016), ‘From Across the Ocean: Considering Travelling Literary Figurations as Part of Swahili Intellectual History’, Journal of African Cultural Studies, 28/2: 225–240. VIERKE | WRITING SONGS, SINGING TEXTS Utumbuizo wa Dhiki, (Nabahany Collection, Mombasa, now possibly kept at the Research Institute of Swahili Studies of Eastern Africa in Mombasa). Utumbuizo wa Kikowa, MS 47754 (Taylor Collection, SOAS, University of London). Ziza na Asha Hamadi, Hs. or. 9954 (State Library of Berlin, Oriental Department). Transcript (of Ziza na Asha Hamadi) by Ernst Dammann with notes and stanzas added by Ahmed Nabahany (DEVA Bayreuth). Zhukov, Andrey (1992), ‘The Dating of Old Swahili Manuscripts: Towards Swahili Paleography’, Swahili Language and Society, Notes and News, 9: 60–68. UNPUBLISHED SOURCES PICTURE CREDITS El-Maawy, Ali A. (n.d.), The Reconstruction of the Early Struggles for Independence (1807–1873). The Lamu Hero: The Story of Bwana Zahidi Mngumi (unpublished booklet). Figs 1–2: © London, SOAS. Nabahany, Ahmed Sheikh (n.d.), Mapisi ya Bwana Zahidi Mngumi Aliyekuwa Mwenye Mui Yaani Raisi wa Kiwandeo (Lamu) (unpublished paper). Fig. 4: © Universität Hamburg, Asien-Afrika-Institut (Institute of Asian and African Studies); Staatsbibliothek zu Berlin – Preußischer Kulturbesitz (Berlin State Library – Prussian Cultural Heritage). Nenda wa Asha Hamadi, MS 1219 H56 (University of Hamburg, Seminar für Afrikanische Sprachen und Kulturen). Fig. 3: © The author. Fig. 5: © The author. Fig. 6: © Nabahany Collection, Mombasa. Utendi wa Haudaji, MS 279888-7 (Allen Collection, SOAS, University of London). manuscript cultures mc NO 17 105 CONTRIBUTORS | MANUSCRIPT CULTURES Contributors Ann Biersteker Annachiara Raia Associate Director Afrika-Studiecentrum African Studies Center at Michigan State University, USA Pieter de la Court 427 N. Shaw Lane Wassenaarseweg 52 East Lansing MI 48824-1035, USA NL-2333 AK Leiden, The Netherlands email: abierst@yahoo.com email: a.raia@hum.leidenuniv.nl Chapane Mutiua Ridder Samsom Universidade Eduardo Mondlane Universität Hamburg Centro de Estudos Africanos Centre for the Study of Manuscript Cultures Av. Julius Nyerere, Rua da Reserva da Universidade, 1993 Warburgstr. 26 Maputo 1993, Mosambik D-20354 Hamburg, Germany Universität Hamburg Centre for the Study of Manuscript Cultures Warburgstr. 26 D-20354 Hamburg, Germany email: mutirua@yahoo.com.br email: ridder.samsom@uni-hamburg.de Clarissa Vierke Universität Bayreuth Faculty of Languages & Literatures Literatures in African Languages Universitätsstr. 30 Ahmed Parkar Pwani University Department of Philosophy and Religious Studies D-95440 Bayreuth, Germany email: Clarissa.Vierke@uni-bayreuth.de PO BOX 195 80108, Kilifi, Kenya email: a.ahmed@pu.ac.ke mc NO 17 manuscript cultures mc NO 17 2021 ISSN 1867–9617 © 2021 Centre for the Study of Manuscript Cultures (CSMC) Universität Hamburg Warburgstraße 26 D-20354 Hamburg www.csmc.uni-hamburg.de