African Arabic: Approaches to Dialectology 9783110292343, 9783110292329

Citation preview

African Arabic: Approaches to Dialectology

Trends in Linguistics Studies and Monographs 258

Editor

Volker Gast Founding Editor

Werner Winter Editorial Board

Walter Bisang Hans Henrich Hock Heiko Narrog Matthias Schlesewsky Niina Ning Zhang Editors responsible for this volume

Walter Bisang Volker Gast

De Gruyter Mouton

African Arabic: Approaches to Dialectology

edited by

Mena Lafkioui

De Gruyter Mouton

ISBN 978-3-11-029232-9 e-ISBN 978-3-11-029234-3 ISSN 1861-4302 Library of Congress Cataloging-in-Publication Data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.dnb.de. ” 2013 Walter de Gruyter GmbH, Berlin/Boston Printing: Hubert & Co. GmbH & Co. KG, Göttingen 앝 Printed on acid-free paper 앪 Printed in Germany www.degruyter.com

Contents Addresses of contributors ................................................................ ....................................................................... ....................................... vii Introduction Mena Lafkioui........................................................................................... 1 Grammar studies in African Arabic dialectology Chapter 1 Stefano Manfredi Native and non-native varieties of Arabic in an emerging urban centre of western Sudan. Evidence from Kadugli .................. 13 Chapter 2 Mena Lafkioui Reinventing negation patterns in Moroccan Arabic ......................... 51 Chapter 3 Shuichiro Nakao The prosody of Juba Arabic: split prosody, morphophonology, and slang ........................................................................................... 95 Chapter 4 Catherine Taine-Cheikh Grammaticalized uses of the verb a(a) in Arabic: a Maghrebian specificity?............................................................... 121 Chapter 5 Xavier Luffin Some new information about Bongor Arabic ................................. 161

vi

Contents

Lexicological Lexicological studies studies in African Arabic dialectology Chapter 6 Giorgio Banti Strata on loanwords from Arabic and other Semitic languages in Northern Somali ......................................................................... 185 Chapter 7 Lameen Souag Sub-Saharan lexical influence in North African Arabic and Berber ...................................................................................... 211 Chapter 8 Peter Behnstedt Lexical aspects of Maghrebi Arabic ............................................... 237 Chapter 9 Vermondo Brugnatelli Arab-Berber contacts in the Middle Ages and ancient Arabic dialects: new evidence from an old Ibā7ite religious text .............. 271 Index ................................................................ ................................................................................................ .................................................................... .................................... 293

Addresses of contributors Giorgio Banti Dipartimento Asia Africa e Mediterraneo Università degli Studi di Napoli “L’Orientale” Palazzo Corigliano, P.zza S. Domenico Maggiore 12 I-80134 Napoli, Italy Email: [email protected] Peter Behnstedt Avenida de Sevilla 84 E-11550 Chipiona (Cádiz), Spain Orientalisches Seminar, Universität Hamburg (Emeritus), Germany Email: [email protected] Vermondo Brugnatelli Dipartimento di Scienze Umane per la Formazione Università di Milano-Bicocca Piazza dell’Ateneo Nuovo, 1 20126, Milano, Italy Email: [email protected] Mena Lafkioui Dipartimento di Scienze Umane per la Formazione Università di Milano-Bicocca Piazza dell’Ateneo Nuovo, 1 20126, Milano, Italy Email: [email protected] Ghent University Department of Languages and Cultures Rozier 44 9000 Gent, Belgium Email: [email protected]

viii

Contributors

Xavier Luffin Université Libre de Bruxelles Faculté de Philosophie et Lettres ULB CP175, avenue F.D. Roosevelt 50 1050 Bruxelles, Belgium Email: [email protected] Stefano Manfredi CNRS – LLACAN (UMR 8135) 7, rue Guy Môquet, Bât. D 94801 Villejuif Cédex, France Email: [email protected] Shuichiro Nakao Department of Linguistics Graduate Kyoto University, Kyoto Universtiy 606-8501, Yoshida-Honmachi, Sakyo-ku Kyoto, Japan Email: [email protected] Lameen Souag CNRS – LACITO (UMR 7107) 7, rue Guy Môquet, Bât. D 94801 Villejuif Cédex, France Email: [email protected] Catherine TaineTaine-Cheikh CNRS – LACITO (UMR 7107) 7, rue Guy Môquet. Bât. D 94801 Villejuif Cédex, France Email: [email protected]

Introduction Mena Lafkioui Dialectology considers variation in language to be most commonly based on geographical distribution and its correlated phonological, grammatical and lexical features. There are two approaches to dialectology: synchronic and diachronic ones. Synchronic dialectology studies the synchronic linguistic proximity of geographically contiguous language varieties. Diachronic, or historical, dialectology describes and explains the language evolution and diffusion processes which lead to certain variations in form, meaning and structure by means of sociohistorical reconstructions. As these two approaches are interrelated, no clear-cut distinction can be made between them. Indeed, in variational linguistic research (viz. dialectology), any geographical diversity necessarily refers to temporal diversity (de Saussure 1973 [1916]; Lafkioui 2007). In addition to the horizontal (geographical, synchronic) and vertical (historical, diachronic) dimensions, linguistic variation also involves social diversity and individual diversity. The former – social diversity – includes parameters such as habitat (rural/urban), country of origin or country of immigration, gender, ethnicity, age, social class, educational level and educational type. The latter – individual diversity – partly overlaps with social variation, but also incorporates the ideological and psychohistorical aspects of the individual speakers (Lafkioui 2007, 2011a, Fc). In the case of African Arabic, it is historical rather than synchronic dialectology which is in need of additional and advanced study. A notable exception is the research conducted in creolistics, which is not only beneficial to dialectological but also to linguistic geographical studies (e.g. Kaye 1985; Miller 2002, 2004; Owens 2001; Versteegh 1993). The present volume aims at filling this gap, at least to a certain extent, by presenting the state-of-the-art in research on African varieties of Arabic. It covers all the major areas of linguistic analysis (i.e. phonetics, phonology, morphology, syntax and lexicology), and presents in a clear and concise way data and issues from a wide range of less-commonly-quoted languages, with special emphasis on language contact phenomena. It comprises nine closely linked but self-contained chapters. Chapters one to five are mainly devoted to grammar studies (including phonetics and phonology) and examine

2

Mena Lafkioui

theoretical and methodological questions related to African Arabic dialectology. They list some specific case studies of little studied or unstudied languages and language varieties marked by or derived from contact situations. Chapters six to nine address lexicological subjects, providing several unpublished datasets and casting fresh light onto the interface between lexical forms and semantic contents embedded in various socio-historical contexts. The different glossonyms used in this book are left unchanged to the authors’ preference (e.g. Nubi = Ki-Nubi; Maghreb(i) Arabic = North African Arabic; Bongor Arabic = túrku = túrgu), so as to better represent both the diversity and the convergence in viewpoints. The same goes for the selected transcription systems, which reflect different theoretical accounts. In the first chapter, Stefano Manfredi presents an in-depth sociolinguistic study of the different Arabic varieties attested in Kadugli, the capital city of the Southern Kordofan state in western Sudan. In this emergent multilingual “peripheral” urban centre, in which Arabic-speaking and nonArabic-speaking ethnic groups from the Nuba Mountains come in contact with each other, the Arabic language has developed linguistic varieties that indicate the different socio-economic statuses of the respective migrant communities. Dialect levelling towards Sudanese Standard Arabic is one of the most observed phenomena in Kadugli: regardless of the retention trend of the phonological reflexes (e.g. *ġ = [q]) produced by the urbanized Baggara Arabic-speaking Bedouins, these locutors are little by little losing the most socially marked morphological and lexical features of their dialect (e.g. feminine as morphological category, genitive exponent hān). The city of Kadugli is also interesting for the study of phenomena created by the coexistence of vernacular (L1) and vehicular (L2) varieties of Arabic, such as a high degree of phonological variation due to second language acquisition (e.g. * = [x, h, k]) among early-urbanized non-Arabic speakers. Compared to second-generation migrants, who are more affected by linguistic normalization, the language practices of the early-urbanized speakers seem to obstruct the surfacing of a local urban koiné. Drawing on a quantitative investigation on language uses and language attitudes in Kadugli, the author presents a theoretical framework for the analysis of dialect variation in urban settings, discovering general trends as well as more local issues in Arabic urban sociolinguistics. Chapter two is dedicated to the investigation of how new negation patterns are engendered in Moroccan Arabic (Oujda region, North East Morocco) by contact with Berber (Tarifit, North Morocco). The several

Introduction

3

discussed cases are related to contact-induced innovation processes in Moroccan Arabic negation, through which the morphological data as well as the syntactic structuring and semantic functioning have been transformed by analogy with Berber negation. North African Arabic negation is essentially based on the morphosyntactic opposition between “verbal negation”, having a verbal constituent as the incidence point, and “non-verbal negation”, which concerns different nominal constituents that usually function as a predicate. While the discontinuous marker ma ___ š (ši/šay) and its optional or conditioned (modality and expressive conditions) variants are used for verbal negation in existential configurations, non-verbal predication is negated by means of the same discontinuous marker in existential contexts as well as by the continuous morpheme maši ___ or muš ___ , so as to convey semantic values that point to an identification of the referential object, its localization or its attribution to a specific semantic category. Based on her concepts of basic negation versus marked negation, the author classifies the North African Arabic negation system (especially the Moroccan Arabic one) as asymmetrical on the basis of detailed analyses and accounts of the various obligatory and optional marking procedures. With respect to this negation system, the Moroccan Arabic variety of Oujda has invented a new discontinuous marker, ma ___ bu, of which not only the post-head element is borrowed from Tarifit (morphological analogy) – which is the only Berber language where this morpheme occurs (Lafkioui, 1996, 1999, 2007, 2011a) – but also its combinatorial (structural analogy) and functional properties (functional analogy). The creation of this new hybrid negator ma ___ bu has triggered a restructuring of the negation system of Moroccan Arabic of Oujda on both the morphosyntactic and the semantic level. Additionally, this contribution confirms that the variational development of contact-induced innovations and their diffusion can also be motivated by system-based factors. In other words, it shows how extra-linguistic (contact) and linguistic (formal and functional) parameters interact precisely and, hence, how they can both be responsible for language change and diffusion (Lafkioui, 2009, 2011b). The third chapter examines the prosodic systems of Juba Arabic, a partially creolized Arabic-based Pidgin mainly spoken in urban areas of Equatoria in Southern Sudan. A contrastive dialectological study between (1) the prosody structure of Juba Arabic based on Shuichiro Nakao’s own fieldwork research with Päri and Tenet youngsters in Juba – confronted with established Juba Arabic accent descriptions (e.g. Nhial 1975; Nyombe 1986) – and (2) the Nubi prosody structure worked out by Gussenhoven

4

Mena Lafkioui

(2006), suggests that Juba Arabic mainly distinguishes between three independent prosodic systems as a result of profound language contact with Arabic and local languages of Southern Sudan. Furthermore, this contribution proposes to consider the mixed prosodic systems of Juba Arabic as a kind of “split prosody”, defined as one in which “the majority of its words are marked for pitch accent but an important minority are marked for tone” (Good 2004). Chapter four is devoted to a thorough study of verb grammaticalization in Arabic of North Africa (Maghreb), and more particularly the case of the verb raɉā, which originally signifies ‘to see’ and which is currently grammaticalized into an enunciative particle that, together with a pronominal clitic, introduces utterances for temporal, aspectual and modal purposes. Catherine Taine-Cheikh confronts the remarkable variety of patterns and values observed for this morphosyntactic phenomenon throughout the whole dialectical area. Her comparative and explanatory findings are of great interest for studying the history of the Arabic dialects of Africa, as well as for understanding the universal process of grammaticalization in general. In the fifth and last chapter of the grammar part of this volume, Xavier Luffin gives a detailed description of Bongor Arabic, an Arabic Pidgin/ Creole spoken today in Chad, in the city of Bongor (Mayo-Kebi). Even if this variety proves to have various linguistic features in common with Ki-Nubi and Juba Arabic, it also exhibits important differences in phonology, morphology and lexicon. Based on new data recorded in Chad as well as on the former studies about Turku (e.g. Muraz 1930; Tosco and Owens 1993), this chapter aims at determining which features are attributed to the first lexifier(s), Sudanese and Egyptian Colloquial Arabic, which were brought to Chad in the 19th century, and whose features relate to the most recent lexifier(s) of which mainly Chadian Colloquial Arabic is relevant. In chapter six – which is also the first chapter devoted to lexicological topics – Giorgio Banti discusses in some detail the main layers of Semitic loanwords in Northern Somali. The focus is here on identifying and analyzing a number of loanwords and other contact phenomena in Northern Somali, which can be ascribed to three major Semitic language groups against the background of the described historical sceneries: (a) Ancient South Arabian, (b) Arabic, and (c) Southern Ethiosemitic. Banti’s analysis is based on Stein’s (2008, 2012) suggestion that Himyaritic was a

Introduction

5

development of Sabaic.1 Different phonological treatments of a number of Arabic sounds are examined as possible cues of an indirect transmission of some of these sounds through Southern Ethiosemitic, rather than a direct transmission from different Arabic dialects. Besides the more obvious loanwords from Harari, Northern Somali also provides evidence for the existence of several Southern Ethiosemitic loanwords in its core lexicon, such as maalín ‘day’ (and máanta ‘today’) and badán ‘be much, be many’. These data and analyses are preceded by a description of the East Cushitic dialects and languages Northern Somali is part of, i.e. the Somali cluster. Phonological distinctive isoglosses for Northern Somali and its closest relatives are presented with comparative illustrations, as well as the major isogloss that detaches eastern Northern Somali from its other varieties. Additionally, the languages that Northern Somali is now in contact with are described and a reconstruction is provided of the contact situations experienced in the past by the communities living in the areas where Northern Somali is spoken today. The Periplus of the Erythraean Sea points to close contacts of the northern coast of present-day Somaliland with the port of Muza in south-eastern Yemen during the 1st century CE, in the area where the Himyarites would emerge as the dominant group in the following centuries. Moreover, this contribution examines the two major avenues through which knowledge of Arabic and of Islam spread to the eastern Horn: (a) the sea route along the northern and eastern coasts, and (b) the inland route of the caravans and medieval towns that connected the ports of Zeyla and Berbera to the Chercher Mountains and the lakes of the Rift Valley. Along this second route, speakers of Northern Somali also had early contacts with mainly Moslem speakers of different varieties of Southern Ethiosemitic languages. Even later, Harar was an important religious, political and commercial centre for many communities of the eastern Horn, among them also those who spoke Northern Somali. In chapter seven, Lameen Souag investigates the impact of transSaharan trade – in particular the importation of slaves from the Sahel (especially across the oases of the Sahara and in some regions even further north) – on the linguistic landscape and the history of the region, a field that remains virtually undocumented despite the existence of significant study material. He provides an overview of sub-Saharan loans across the Maghreb and in doing so seeks to reveal which sub-Saharan African languages have influenced North Africa, and to what extent. His examination indi1.

See also Contini (1994) for more information about the origin of Himyaritic and its connection with Ancient South Arabic.

6

Mena Lafkioui

cates that such loans appear to be particularly conspicuous in the area of crop names. Unsurprisingly, these are most numerous in Hassaniya, e.g. mâ+u ‘rice’ (< Wolof) or the many Songhay loans documented in Heath (2004), but are not absent further north: probable cases include the widespread Maghreb Arabic term kabuya/bkewa/etc. (and Berber equivalents, e.g. Figuig tkabiwt) for ‘pumpkin’ (probably < Hausa); the Libyan Arabic term gafuli for ‘sorghum’ (< Kanuri); the Timimoun Berber term dəmši for ‘peas’ (< Songhay); the Ghadames Berber term bƮrkano for ‘chilli pepper’ (< Hausa, cf. Lanfry 1973); Tidikelt Arabic aga/u ‘gourd’ (< Songhay, via Tuareg); and the widespread dəmman ‘sheep breed’, popularly believed to have come from the Sahel (< Kanuri?). In some cases, the sense of the word has changed slightly, illustrating environmental differences between the source and the destination. A more socially restricted environment for sub-Saharan loans, preserving a richer selection of them, is the specific context of Gnaoua brotherhoods, their songs and their Arabic-based argot. This context has received some prior attention, as in Aguadé (2002) or Berjaoui (1997). However, the origin of most of these loans has not yet been pinpointed. The argot material provides a particularly rich source, making it possible to locate riverine Songhay and Bambara sources as well as Hausa; not all the terms have a clear origin, suggesting a greater diversity of sources. The borrowed names of certain musical instruments used by the Gnaoua, in particular the ganga drum (< Hausa or Songhay), also fit within this context, yet are different as they are widely known. Another “argotic” usage has so far only been attested in the Timimoun region: in curses and insults, words for body parts of subSaharan origin are substituted. Based on the analyzed evidence, the author confirms that the principal sub-Saharan influence on Maghreb Arabic comes from languages of the Sahel. For the most part, each region seems to be primarily influenced by the principal lingua franca of the part of the Sahel directly to its south, as might be predicted from the mostly northsouth orientation of medieval trans-Saharan trade routes. However, the influence of Hausa at least is more widespread than this generalization would imply, and the distribution of argot terms indicates that, at least in some areas, speakers of many different languages must have mingled – a conclusion confirmed for some parts of the Sahara by first-hand testimonies. The differences between western (Maghreb) and eastern Arabic dialects are considerable and the problems of mutual understanding are explained mainly by lexical variation, which is the topic under investigation in

Introduction

7

chapter eight, illustrated by nineteen maps from the Wortatlas der arabischen Dialekte (Behnstedt and Woidich 2011; vols. I and II). Problems of understanding may result from “false friends”.2 Some of the lexical differences are explained by semantic change such as extension and particularization, e.g. Classical Arabic ūt ‘fish, big fish’, but means ‘fish generically’ in Maghreb Arabic, while it stands for ‘whale’ in eastern Arabic. In this lexicological research, Peter Behnstedt postulates that many synonyms of Classical Arabic at the time of their compilation by the lexicographers were regional variants, a suggestion which is backed up by evidence presented in the form of maps exposing the dialectological variation for terms such as ‘threshing place’, ‘yoke’, ‘sesame’ and ‘beehive’. Furthermore, it is shown that some Maghreb Arabic forms refer back to Yemeni dialectal forms. Other lexical particularities of Maghreb Arabic are explained by loans from the Berber substratum, especially strong in Moroccan and Mauritanian Arabic, as exemplified by the designations for ‘key’ and ‘elbow’. However, the origin of some typical Maghreb Arabic forms like ‘atrūs for ‘billy-goat’, sardūk for ‘cock’ is still obscure. Different Mediterranean languages also play an important role in the formation of the Maghreb Arabic lexicon. Examples are given for older loans from Greek and Latin (e.g. ‘chick’, ‘rabbit’, ‘stork’). In modern times, the lexical impact of the colonial languages such as French, Spanish and Italian is quite high, which is confirmed by the lexical variation map for ‘kitchen’, displaying forms from Italian and French. Other loans are also looked at on the various presented maps. As for loans from Turkish, they are present but less important in Maghreb Arabic than in eastern dialects. This is corroborated by two lexicological maps (‘spoon’ and ‘socks’), on which also loans from Spanish, French, Italian, Berber and Wolof can be found. The last chapter of this volume deals with an original and lengthy manuscript containing an Arabic-Berber text of the Middle Ages, the Kitāb al-barbariyya, also known under the name Mudawwana of Ibn R’anem or of Abū Ġānim al-:urāsānī, which provides the first evidence of different phenomena related to language contact between Berber and Arabic in the Middle Ages. The exact date of composition of the text is not known, but the data (including the spelling of [g], transcribed with instead of ) point to an early period (between the 10th and the 15th century), at least 2.

Some examples of these “false friends” are xallas? ‘to pay’ in Maghreb Arabic, meaning ‘to finish’ in eastern Arabic dialects; bisbās ‘fennel’ in Maghreb Arabic, signifying ‘macis’ in eastern dialects and even ‘hot peppers’ in Yemeni Arabic.

8

Mena Lafkioui

prior to the Hilali invasion or very close to it. In this respect, some remarkable linguistic phenomena can be found in the domain of borrowings. Even at an early stage, the text displays a number of Arabic loanwords in Berber, mainly related to the field of religion, while the language of this document preserves much of the native lexicon. An example showing this contact phenomenon is the following: f. 2a.21-2b.2: elferay@ fell-aneA yefre@ Baba-nneA, amm_uDum, am teDalliE, am az-zakawat, am el eLJLJ i w’as-yezemren, am elLJehaG eLJLJ_ebriG en Yuc, am temmert (?) n elmeJruf, am ennehi af elmenker, am ele quq en yedaddayen, am ele quq en elLJiran, am twalaE en yemeslem am tebratt en yir eLJLJiG ‘the obligations God imposed upon us, like fast, prayer, pilgrimage for him who can afford it, the effort on God’s path, the promotion of virtue and the prevention of vice, the respect of parents, the respect of neighbours, friendship of Muslims and hostility to evildoers’

In different parts of the manuscript, native and borrowed lexicon is used differently. Vermondo Brugnatelli attributes this to the fact that they were composed at a different time. This is for example the case with the Berber noun asersur, corresponding to Arabic uLJLJa ‘(divine) proof, argument’ (f.96a, l.12). In the opening part, it is substituted by the loanword l uLJLJeE (f.1a l.1 and passim), as this part was most likely composed after the rest of the work. Moreover, in some cases, the phonetic shape of the loanwords refers to particular phonetic and morphological processes in spoken Maghreb Arabic – which was already in the distant past distinct from “Classical Arabic”. The best way to prove this is by looking at the numerals of the second decade: Enāš ‘twelve’ (f. 124a, l. 8), tmenMac ‘eighteen’ (f. 62b, l. 15), where it is plain that the the loss of the last part of the numerals and the pharyngalization of the internal t (> M) is a trace of a dropped ‘ayin. The presence of these forms as loanwords in this ancient Berber text demonstrates that these phenomena, considered by Ferguson as belonging to an old “Arabic koiné”, had already taken place in early times (Ferguson 1959: 626, Brugnatelli 1982: 43). Borrowed numerals also provide proof for the early appearance of another feature typical of Maghreb Arabic, namely constructions with a particle n: xamsMac n yum ‘fifteen days’ (f. 42b, l. 3), Enāš en wuqiyya ‘twelve ounces’, Enāš en dirham ‘twelve dirhams’ (f. 57b, l. 16). Sometimes, a borrowing even helps to find the original meaning of Arabic words which were apparently already forgotten or disputed in ancient time. For example, the text contains instances of the word taklalt that is borrowed from Arabic kalāla, a term that several scholars have attempted to explain (Cilardo 2005). Many of

Introduction

9

them point to ‘weakness’ or ‘fatigue’ and to ‘distant relatives’ or ‘collateral relatives’ in the context of heritages. On the basis of cognates in other Semitic languages, it was also proposed to be a kinship term. Brugnatelli’s study confirms some of these analyses as taklalt/kalāla is repeatedly explained by the gloss ɇaO min al-ɇumma, that is to say, ‘maternal halfbrother’ (e.g. f. 304b, l. 7, 15) in the presented manuscript. This book brings together leading scholars from around the world who address topics that deal with language contact, the development of Arabicbased Pidgins and Creoles, synchronic language variation and diachronic language reconstruction. Furthermore, it includes the reports on their fieldwork and points out new interesting theoretical and methodological viewpoints on African Arabic dialectology and general dialectology as well as on contact linguistics. Its outcomes offer important insights for all linguistic studies and approaches, and directly connect with other research fields such as sociolinguistics, ethnolinguistics and language acquisition.

References Aguadé, Jordi 2002 La langue des gnâwa. In Aspects of the Dialects of Arabic Today: proceedings of the 4th Conference of the International Arabic Dialectology Association (AIDA, Marrakesh, Apr. 1–4. 2000), Abderrahim Youssi, Fouzi Benjelloun, Mohamed Dahbi, and Zakia Iraqui-Sinaceur (eds.), 405–411. Rabat: Amapatril. Behnstedt, Peter, and Manfred Woidich 2011 Wortatlas der arabischen Dialekte. Band I: Mensch, Natur, Fauna, Flora. Leiden/Boston: Brill. 2012 Wortatlas der arabischen Dialekte. Band II: Materielle Kultur. Leiden/Boston: Brill. Berjaoui, Nasser 1997 Parlers secrets d’el-Jadida: Notes préliminaires. Estudios de Dialectología Norteafricana y Andalusí 2: 147–158. Brugnatelli, Vermondo 1982 Questioni di morfologia e sintassi dei numerali cardinali semitici. Firenze: La Nuova Italia. Cilardo, Agostino 2005 The Qurɇanic Term Kalāla: Studies in Arabic Language and Poetry, XadīZ, Tafsīr, and Fiqh: Notes on the Origins of Islamic Law. Edinburgh: Edinburgh University Press. Contini, Riccardo. 1994 I suffissi in velare del perfetto in arabo yemenita. In Sem Cam Iafet. Atti della 7a Giornata di studi camito-semitici e indoeuropei (Milano, 10 giugno 1993), Vermondo Brugnatelli (ed.), 77–93. Milano: Centro Studi Camito-Semitici. Ferguson, Charles A. 1959 The Arabic Koine, Language 35(4): 616–630. Good, Jeff 2004 Split Prosody and Creole Simplicity: The Case of Saramaccan. Journal of Portuguese Linguistics 3: 11–30. Gussenhoven, Carlos 2006 Between Stress and Tone in Nubi Word Prosody. Phonology 23: 193–223. Heath, Jeffrey 2004 Hassaniya Arabic (Mali)-English-French Dictionary. Wiesbaden: Harrassowitz. Kaye, Alan. S. 1985 On the Importance of Pidgins and Creoles for Historical Linguistics. Diachronica: International Journal for Historical Linguistics 2(2): 201–230.

Introduction

11

Lafkioui, Mena 1996 La négation en tarifit. In La négation en berbère et en arabe maghrébin, Salem Chaker, and Dominique Caubet (eds.), 49–77. Paris: L’Harmattan. 1999 Syntaxe intégrée de l’énoncé non-verbal berbère. PhD thesis, INALCO (Paris). 2007 Atlas linguistique des variétés berbères du Rif. Köln: Köppe. 2009 Synchronic and diachronic linguistic variation as an indicator of language change and diffusion in Tarifit. Studi Maġrebini (Nuova Serie) 6: 65–84. 2011a Etudes de la variation et de la structuration linguistiques et sociolinguistiques en berbère. Köln: Köppe. 2011b How system-internal linguistic factors indicate language change and diffusion. A geolinguistic analysis of Berber data. Dialectologia et Geolinguistica 19: 62–80. Fc Dialectology and Language Geography. In The Oxford Handbook of African Languages, Rainer Vossen (ed.), in press. Oxford: Oxford University Press. Lanfry, Jacques 1973 Ghadamès II: Glossaire (parler des Ayt Waziten). Fort-National (Algérie): Le Fichier Périodique. Miller, Catherine 2002 The relevance of Arabic-based pidgins-creoles for Arabic linguistics. In Al-lugha, Gerda Mansur, and Madiha Doss (eds.), 7–46. Cairo: Arab Development Center. 2004 Variation and change in Arabic urban vernaculars. In Approaches to Arabic Dialects: Collection of Articles presented to Manfred Woidich on the Occasion of his Sixtieth Birthday. Martine Haak, Kees Versteegh, and Rudolf De Jong (eds.), 177–206. Amsterdam: Brill. Muraz, Gaston 1930 Vocabulaire du patois arabe tchadien ou ‘tourkou’ et des dialectes Sara-Madjinngaye et Sara Mbaye. Nancy: Charles-Lavauzelle. Nhial, Abdon Agaw Jok 1975 Ki-Nubi and Juba Arabic: A Comparative Study. In Directions in Sudanese Linguistics and Folklore, Herman Bell, and Sayed Hamid Hurreiz (eds.), 81–93. Khartoum: Khartoum University Press Nyombe, Bureng George Vincent 1986 Juba Arabic from a Bari Perspective. In Current Approaches to African Linguistics, Gerrit J. Dimmendaal (ed.), 71–78. Dordrecht: Foris.

12

Mena Lafkioui

Owens, Jonathan 2001 Creole Arabic: The Orphan of All Orphans, Anthropological Linguistics 43(3): 348–378 Saussure, Ferdinand de 1973 Reprint. Cours de linguistique générale. Tulio de Mauro (ed.), Paris: Payothèque. Original edition, 1916. Stein, Peter 2008 The “Ximyaritic” language in pre-Islamic Yemen: a critical re-evaluation. Semitica et Classica 1: 203–212. 2012 Ancient South Arabian. In The Semitic languages: an international handbook, Stefan Weninger (ed.), 1042–1073. Berlin/Boston: Mouton De Gruyter. Tosco, Mauro, and Jonathan Owens 1993 Turku: a descriptive and comparative study. Sprache und Geschichte in Afrika 14: 177–267. Versteegh, Kees. Leveling in the Sudan: From Arabic Creole to Arabic Dialect. In1993 ternational Journal of the Sociology of Language 99: 65–79.

Native and nonnon-native varieties of Arabic in an emerging urban centre of western western Sudan. Evidence from Kadugli Stefano Manfredi 1. Introduction The linguistic impact of urbanization in Arab countries has been relatively well studied during the last decades. However, specialists of Arabic dialectology have often limited their attention to well-established urban centres in core Arab countries. In this overall situation, we still lack of in-depth linguistic information from emergent arabophone cities in peripheral regions. This study tries to partially fill this gap through the description of the different Arabic varieties spoken in Kadugli, the chief town of the Southern Kordofan state in the western Republic of Sudan. Kadugli can be broadly described as an emerging urban centre in a complex multilingual setting. More in particular, Kadugli is characterized by the coexistence of both native and non-native varieties of Arabic. On the one hand, the sociolinguistic variation affecting native varieties of Arabic reflects the different socio-economic statuses of arabophone communities in Kadugli. On the other hand, non-native varieties of Arabic are subject to a strong individual variation that gives evidence of different modalities of language contact and language acquisition. After providing some linguistic and historical background (Section 2), this chapter describes the present-day ethnolinguistic setting of Kadugli (Section 3). In Section 4, I analyze the main phonological and morphosyntactic features of both native and non-native Arabic varieties spoken in Kadugli. The study finally states that, despite its relatively high degree of Arabization, Kadugli still does not possess a shared dialectal norm and that the ongoing linguistic convergence towards the prestigious national koiné impedes the emergence of an innovative urban dialect.1 1.

The following glosses and symbols are used: ACT ‘active’, ANT ‘anterior’, CS ‘construct state’, DEF ‘definite article’, DIST ‘distal’, EXS ‘existential copula’, F ‘feminine’, FOC ‘focus’, IMP ‘imperative’, IPFV ‘imperfective’, FUT ‘futur’, IND ‘indicative’, INDF ‘indefinite’, IDPH ‘ideophone’, M ‘mas-

14

Stefano Manfredi

2. Linguistic and historical background Kadugli is located in the south-western part of the Nuba Mountains (20°43ɑ00ɑɑE, 11°1ɑ0ɑɑN), one of the most linguistically heterogeneous areas in Africa both in terms of the number of languages spoken and the concentration of different language groups (Thelwall and Schadeberg 1983; Bell 1995). Even if the indigenous languages of the Nuba Mountains are still largely under-described, they can be broadly classified into two main phyla: − Kordofanian languages - These languages are spoken in the eastern and southern part of the Nuba Mountains. They can be divided into four sub-groups: Rashad (Tagoi, Tegali), Heiban (e.g. Koalib, Tira, Moro), Talodi (e.g. Ngile, Acheron, Lafofa) and (probably) KatlaTima. Kordofanian languages are believed to constitute one of the numerous branches of the Niger-Congo linguistic family (Schadeberg 1989; Quint 2006). − Nilo-Saharan languages - Nilo-Saharan languages are spoken in the northern and western parts of the Nuba Mountains. These can also be divided into five sub-groups: Hill-Nubian (e.g. Dilling, Debri, Dair, Kadaro, Karko), Nyamang-Afitti, Temein, Daju (e.g. Daju, Logorí, Soborí, Shatt) and Kadu (e.g. Kadugli, Miri, Katcha, Tulishi, Keiga, Korongo). The Kadu group was earlier classified as part of Kordofanian, but since Schadeberg (1981) it is considered as an isolated subgroup of Nilo-Saharan. At the end of the 18th century, mainly in consequence of the migration of Baggara cattle-herders from Dar Fur, Arabic started to spread over the Nuba Mountains (Manfredi Fc-a). Baggara Arabs used the Nuba Mountains area for pasture, which brought them into direct contact with local nonArab groups. Furthermore, Baggara Arabs have been systematically engaged in expeditions of slave raiding among local populations during the Turko-Egyptian rule (1821–1884). The absence of a local lingua franca in the Nuba Mountains favored the gradual adoption of Arabic as an interethnic means of communication. Later on, the process of acquisition of Arabic as both a native and a non-native language was sensibly incremented by the process of urbanization experienced by indigenous communities of the region. culine’, PROX ‘proximal’, PTCP ‘participle’, REL ‘relative pronoun’, SG ‘singular’, SING ‘singulative’, 1 ‘1st person’, 2 ‘2nd person’, 3 ‘3rd person’, ‘affix boundary’, = ‘clitic boundary’, \ ‘ablaut’, ~ ‘reduplication’.

Native and non-native varieties of Arabic in Kadugli 15

Map 1.

Sudan and neighbouring countries

With regard to Kadugli,2 it should be remarked that before the end of the Mahdiya period (1899) there were no urban settlements in the area, but just an assembly of self-reliant Kadugli-speaking communities located in the nowadays quarters of Lajar al-Makk, Kulbá and Murtá (see Section 3, map 2). The urban centre of Kadugli started to develop as a garrison station during the first phase of the Anglo-Egyptian condominium in Sudan. After that, due to the introduction of cotton as a cash crop in 1923, Kadugli rapidly became one of the largest markets of the Nuba Mountains and throughout the first half of the 20th century attracted many local non-Arab groups as well as numerous Arabic-speaking traders from northern Sudan. 2.

According to Schadeberg (1994: 12), the toponym ‘Kadugli’ finds its origins in the Kadugli compound noun kad.ii ‘people’ -guli ‘spring (of water)’.

16

Stefano Manfredi

Following the Sudanese independence in 1956, Kadugli was included into the Kordofan region whose administrative centre was established in El-Obeyd. It was not until the Addis-Ababa peace agreements in 1972 that the Nuba Mountains area was elevated to be an administratively autonomous state (Southern Kordofan) with its chief town in Kadugli. At that time, a massive process of urbanization started, which led Kadugli to acquire the appear-ance of a booming urban centre with probably 40.000 inhabitants in 1979. This process of socio-economic integration also interested Baggara cattle-herders that gradually started to settle in the eastern fringes of the urban circle. The urbanization of Kadugli went on along all the eighties, and it was further boosted by the outbreak of the civil war in the Nuba Mountains in 1989. The last national census of Sudan carried out in 1993 reported an urban population of almost 120.000 in Kadugli.3 At the time of my fieldwork (2008–2010), Kadugli had a population of almost 200.000 inhabitants coming from every region of the previous unified Sudan. However, it should be stressed that this figure is currently susceptible to change because of the ongoing urbanization of rural communities due to the eruption of a new armed conflict in 2011.

3. The ethnolinguistic setting of Kadugli This section aims at illustrating the ethnolinguistic setting of Kadugli on the basis of a quantitative survey conducted on the issues of ethnic identity and bilingualism.4 The sample inquired through questionnaire consists of 1020 individuals located in 17 different quarters of Kadugli (60 respondents for each quarter). The sample as a whole has been equally divided according to age (five age classes: 5–15, 16–30, 31–45, 45–60, > 60; i.e. 12 respondents for each class) and gender (i.e. 6 men/women for each age class). The survey included questions about the ethnic identity of the respondents, their period of urbanization, their degree of literacy in Arabic, and finally about their native and non-native languages. 3.

4.

In truth, the latest population census of Sudan was carried out in 2009, but the SPLM (Sudanese People Liberation Movement) refused to accept the results concerning Southern Kordofan because of alleged statistical manipulation by the central government. The following survey was carried out between January and May 2008. The whole work would have been impossible without the crucial collaboration of Ammar N. M. and Munawwar J. N. to whom I am deeply grateful.

Native and non-native varieties of Arabic in Kadugli 17

Map 2.

Kadugli

Table 1. Kadugli quarters

1) Layy al-Sūk 2) Darja al-Tālta 3) Layy al-MaUāni‘ 4) Zendiyya 5) Redīf 6) Malikiyya 7) Ga‘r al-Lajar 8) Banjedīd 9) Tāfāri

10) Lajar al-Makk 11) Layy al-Kōz 12) Layy al-Fagara 13) Mukhayyamāt 14) Kelimó 15) al-Samma 16) Umm BattāY 17) Kulbá

18

Stefano Manfredi

Before analyzing the social variables linked to bilingualism, some preliminary remarks about ethnic self-identification in Kadugli will be presented. From an inter-actionist perspective, ethnicity depends on two mutually interdependent processes, namely the internal definition and external categorization of ethnic boundaries (Barth 1969: 12). Since the beginning of the British colonial rule, the main ethnic definition/ categorization in act in the Nuba Mountains is that opposing Nuba and Arab identities. Nevertheless, there are also ethnic groups that fall outside this dichotomy. The main ethnic groups in Kadugli are:5 − Nuba: the term ‘Nuba’ is a cover ethnonym used for referring to all indigenous non-Arab communities of Southern Kordofan (Stevenson 1984). Given the cultural and linguistic heterogeneity of Nuba groups, their collective identity is not defined on the basis of what the different communities may partake in but rather on the basis of what can distinguish them from Arabs. The largest Nuba communities found in Kadugli are Kadugli, Miri, Logorí and Moro. It should be stressed that the majority of the respondents who declared to be ethnically defined as Nuba is indeed represented by Arabic monolinguals. Thus, the Nuba identity in Kadugli does not seem to be necessarily related to language. Nuba groups are virtually scattered in the whole urban area of Kadugli. − Baggara: ‘Baggara’ means no more than ‘cattlemen’ (from Old Arabic *baqar “cow”). That being so, the term Baggara has neither ethnic nor genealogical pertinence, but it rather stresses the specificity of a production system (Manfredi Fc-a). The majority of urbanized Baggara in Kadugli are wage laborers with some degree of education who retain close-knit relationships with their nomadic relatives living in mobile camps around Kadugli. Even if Baggara Arabs are largely Arabic monolinguals, it is still not rare to find elderly people who acquired a local language during the diffusion of large-scale cotton crop in the Nuba Mountains. Baggara Arabs in Kadugli are mainly found in the eastern quarters of Layy al-Kōz and Layy al-Fagara. − Fellata: the cover term ‘Fellata’ is historically applied to people of west-African origins that have settled in Sudan on their way back from pilgrimage to Mecca. These are mainly Hausa and Fulani 5.

Because of the political sensitivity of the argument, I decided not to display any statistical figure about ethnicity. Thus, I only give some indispensable information about the major ethnic groups living in Kadugli.

Native and non-native varieties of Arabic in Kadugli 19

speakers coming from Nigeria and western Chad. Many Fellata migrated to the Nuba Mountains in search of work in the cotton fields during the 1920s. At the present time, the majority of Fellata living in Kadugli are second/third generation migrants; nonetheless Fellata newcomers constantly arrive from abroad. In these conditions, Fellata people tend to retain their ancestral languages more than Nuba groups do. Fellata are mainly located in the western quarters of Banjedīd, Malikiyya and Tāfāri. − Jallaba: the term ‘Jallaba’ is applied by both Nuba and Baggara to northern Sudanese traders and administrators who settled in the area over the last two centuries. Differently from Baggara Arabs, the Jallāba have never been affected by the use of local languages and they are generally characterized by a high degree of education when compared to local groups. Jallaba people are mainly located in the central quarters of Layy al-Sūk and Layy al-MaUāni‘. − Shawābna: the Shawabna came from Jebel Shaibūn in the northwestern Nuba Mountains and they have a contested position between the lines of being Arab or being Nuba (Ille 2012). All the Shawābna are Arabic monolinguals and they mainly reside in the southern quarters of Mukhayyamāt and al-Samma. − Other outsider groups: this category includes different outsider nonArab groups who recently arrived in Kadugli from western and southern Sudan. These are respectively referred to as gharrāba ‘westerners’ and dʒanūbīn ‘southerners’ and they speak very different languages side by side with different regional varieties of Arabic. They are mainly located in the south-western quarters of Kadugli. Concerning bilingualism in Kadugli, it should be stressed that statistical data drawn from questionnaires could not realistically reflect the sociolinguistic situation of a given community as they depend on the respondent’s subjectivity in determining an arbitrary cutoff point between native and non-native language (Manfredi Fc-b). Nonetheless, some important trends can be observed. As a matter of fact, the process of urbanization in northern Sudan is directly linked with the acquisition of Arabic as both native and non-native language (Bell 1995; Miller 2006: 69). The effects of such linguistic normalization are also evident in a relatively peripheral city like Kadugli that can well be considered as a largely arabophone centre. Actually, all the 1020 respondents claim to speak some form of Arabic. More in detail, 676

20

Stefano Manfredi

informants (66,3%) consider themselves as Arabic monolinguals, against 314 individuals (30,8%) who affirmed to be bilingual speakers of Arabic. Only 29 informants (2,9%) affirmed to be able to speak more than two languages. Apart from bilingual Fellata born in Kadugli, all other bilingual respondents consider Arabic as their second language. Table 2. Minority languages in Kadugli

On a total of 343 bilingual/trilingual speakers Languages Speakers Kadugli (Nilo-Sarahan, Central Kadu) 59 Miri (Nilo-Saharan, Central Kadu) 51 Hausa (Afro-Asiatic, Chadic) 46 Logorí (Nilo-Saharan, Daju) 30 Moro (Kordofanian, Western Heiban) 24 Korongo (Nilo-Saharan, Eastern Kadu) 23 Katcha (Nilo-Saharan, Central Kadu) 20 Fulfude (Niger-Congo, Fula) 19 Keiga (Nilo-Saharan, Western Kadu) 15 Tira (Kordofanian, Talodi) 15 Katla (Kordofanian, Katla) 11 Other 30 TOT 343

% 17,2% 14,8% 13,4% 8,8% 6,9% 6,7% 5,8% 5,5% 4,3% 4,3% 3,6% 8,7% 100%

Minority languages (ru-āna, PL ra-īn, ru-ān-āt in Sudanese Arabic) have a functional context restricted to domestic communication, and they are highly territorialized. Obviously, there are several social variables that affect the use of minority languages in Kadugli.6

6.

The category ‘other’ embraces languages spoken by less than 3% of the sample. These include Ngile (Kordofanian, Talodi), Koalib (Kordofanian, Central Heiban), Otoro (Kordofanian, Heiban), Lafofa (Kordofanian, Talodi), Daju (Nilo-Saharan, Daju), Temein (Nilo-Saharan), Dilling (Nilo-Saharan, HillNubian), Kadaro (Nilo-Saharan, Hill-Nubian), Fur (Nilo-Saharan, Fur), Masalit (Nilo-Saharan, Maba), Beria (Nilo-Saharan, Zaghawa), Kanuri (NiloSaharan, Western Saharan), Dinka (Nilo-Saharan, Western Nilotic) and Nuer (Nilo-Saharan, Western Nilotic).

Native and non-native varieties of Arabic in Kadugli 21 Table 3. Detailed information about bilingual/trilingual speakers in Kadugli

On a total of 343 bilingual/trilingual speakers Individuals Gender men 158 women 185 Age .> 60 93 46–60 142 31–45 73 15–30 23 5–15 12 Literacy literate 198 illiterate 145 Urbanization less than 15 years 285 more than 15 years 43 born in Kadugli 15

% 46% 54% 27,2% 41,4% 21,3% 6,7% 3,4% 58% 42% 83,2% 12,5% 4,3%

First of all, the huge majority of bilingual/trilingual speakers (83,2%) arrived in Kadugli less than fifteen years ago. Thus, early urbanization is the main factor inducing the abandon of minority languages in favor of Arabic. Furthermore, if we look at the gender variable, we can observe that the regression of minority languages is less evident among women, since they represent 54% of bilingual/trilingual speakers in Kadugli. The age variable also affects bilingualism given that 68,6 % of bilingual/trilingual are more than 45 years old. On the contrary, it seems that the progression of formal education in the Nuba Mountains is not directly related to language abandon since 58% of bilingual/trilingual speakers affirmed to have some degree of literacy in Arabic. As I will show (Section 4.3.), the structures of non-native Arabic varieties spoken by non-Arab bilingual speakers in Kadugli vary a great deal according to social and historical dynamics of language contact.

22

Stefano Manfredi

Map 3.

Geographical distribution of minority languages in Kadugli 7

4. Arabic varieties in Kadugli At least four different varieties of Arabic can be distinguished in the urban centre of Kadugli: − Sudanese koiné: this is the prestigious dialectal norm spoken in the major cities of northern and central Sudan. Its centre of diffusion is Khartoum, and it is subjected to very little diatopic variation. The Sudanese urban koiné is an Eastern-type Arabic dialect that seems to 7.

Note that the present map shows the geographical distribution of minority languages spoken by more than 5% of bilingual/trilingual respondents.

Native and non-native varieties of Arabic in Kadugli 23

be closely related to the dialects of Upper Egypt, although with some influences from Peninsular Arabic. As far as Kadugli is concerned, the Sudanese koiné represents a minority native variety of Arabic spoken by Jallāba people. − Emerging Kadugli Arabic: this is the native variety of Arabic spoken by monolingual non-Arab groups born in Kadugli. In truth, the emerging urban dialect of Kadugli is characterized by few phonological traits when compared to the Sudanese koiné. However, these phonological characteristics are more and more subjected to levelling towards the prestigious norm of Sudan. This native variety of Arabic is not geographically restricted, being virtually spoken in the whole urban area of Kadugli. − Baggara Arabic: applying the designation ‘Baggara’ to this variety, I point to its inclusion into the dialect type spoken by Arab cattleherders scattered in a large area running from Lake Chad to the Blue Nile (the so called Baggara Belt). Being a Bedouin dialect, Baggara Arabic is characterized by several conservative traits. Nonetheless, Baggara Arabic is also strongly affected by dialect levelling towards the Sudanese urban koiné. Geographically speaking, Baggara Arabic is spoken in the eastern quarters of Layy al-Kōz, Layy al-Fagara and Kulbá as well as in the nomadic camps inhabited by Baggara nomads during the dry season (approximately from November to May). − Non-native Arabic: this label refers to the different non-native varieties of Arabic spoken by bilingual/trilingual non-Arab individuals living in Kadugli. With regard to this, it should be stated that acquisition of a given language as a non-native means of communication often results in a number of linguistic processes associated with language contact and simplification, but it does not necessarily lead to a drastic linguistic restructuring as in the case of pidgins and creoles (Miller 2002: 35, Tosco and Manfredi Fc). Nonnative Arabic is mainly spoken on the western fringes of Kadugli, as well as in highly mixed quarters such as Darja al-Tālta and Lajar alMakk.

24

Stefano Manfredi

Map 4. Arabic varieties in Kadugli

The different Arabic varieties present in Kadugli can be defined on the basis of several phonological and morphosyntactic traits, albeit they also share a number of features that characterize them as Sudanic dialects. Among the most important pan-Sudanic features found in Kadugli we can enumerate the diachronic development *q > g, the presence of c [tʃ] as a phonemic consonant, the absence of CCC sequences, the absence of ha- in demonstrative pronouns, the generalized use of the pre-formative a- in the formation of imperatives, and the presence of the auxiliary gā(ȥ)id for expressing the progressive aspect of transitive verbs (Owens 1993: 171– 172). In the following paragraphs I will take a closer look at the emerging urban dialect of Kadugli as well as at the Baggara dialect as spoken by urbanized cattlemen and the non-native varieties of Arabic used by local

Native and non-native varieties of Arabic in Kadugli 25

and outsider non-Arab groups. The analysis is based on a selected corpus of almost three hours of recordings (21000 words) embodying spontaneous and semi-spontaneous texts. The duration of the recordings varies from 5 to 20 minutes. This material is representative of fifteen speakers (five speakers for each variety) differentiated according to six social parameters: age, gender, ethnicity, urbanization, kind/degree of education, and pattern of bilingualism/trilingualism.

4.1. Emerging Kadugli Arabic This native variety of Arabic is predominantly spoken by second/third generation in-migrants of non-Arab origins. On that account, all the selected speakers of this dialect were born in Kadugli. If the different ethnic backgrounds of the speakers do not influence the phonology of the emerging urban dialect of Kadugli, some variation can be due to the diverse exposure to formal education in Arabic. This is particularly true for speakers who attended university in Khartoum or in other northern Sudanese urban centres. As regards the gender variable, it seems to not affect the grammatical features of Kadugli Arabic. Apart from inf.5a, who speaks Hausa as second language, all the informants are Arabic monolinguals. Table (4) summarizes the metadata for non-Arab native speakers of Arabic in Kadugli. Table 4. Kadugli Arabic – speakers profile

Age Gender Ethnicity Urbanization inf.1a inf.2a inf.3a inf.4a inf.5a

27 33 33 22 34

F F M M M

Katcha Miri Miri Kadu Fellata

born in Kadugli born in Kadugli born in Kadugli born in Kadugli born in Kadugli

Education Pattern of bilingualism primary Arabic university Arabic primary Arabic secondary Arabic secondary Arabic/ Hausa

The main phonological feature distinguishing the emergent dialect of Kadugli in comparison with the Sudanese urban koiné is the absence of pharyngeal sounds. In line with the sedentary Arabic dialects of Chad (Owens 1993: 161; Roth-Laly 1994: 77; Abu Absi 1995: 12; Jullien de Pommerol 2006: 365), the voiceless pharyngeal fricative ? is generally realized as a laryngeal [h] (ex. 1 ?ādʒa > hādʒa ‘thing’; ex. 3 ?abōba >

26

Stefano Manfredi

habōba ‘grandmother’) and the pharyngeal fricative ȥ converges in a glottal stop [ʔ] when occurring in initial position (ex. 2 ȥomda > ʔomda ‘tribal leader’) or it is elided in final and middle position (ex.2 bitāȥ > bitā ‘of’). (1)

ma=indi=na

hādʒa fi kādūgli thing in Kadugli ‘We don’t have anything here in Kadugli.’ (inf.1a)

NEG=at=1PL

(2)

abū=y ʔomda zātu bitā miri father=1SG tribal_leader FOC POSS Miri_tribe ‘My father is the very tribal leader of the Miri.’ (inf.3a)

(3)

habōba gāl-at lē=na al=kalām=da grandmother say-3SG.F to=1PL DEF=speech=PROX.SG.M ‘(Our) grandmother told us this story.’ (inf.5a)

Highly educated speakers sometimes realize etymological pharyngeal consonants in restricted lexical contexts (ex. 4 ?adāra ‘civilization’): (4)

al=?adāra at=turāsiy-ya fi=ha mawdʒūd-a DEF=civilization DEF=traditional-F EXS=3SG.F present-SG.F ‘The traditional culture is also present (in Kadugli).’ (inf.2a)

All speakers of Kadugli Arabic retain pharyngealized consonants (ex. 5 naEīf ‘clean’; ex. 6 Fabā? > Fabāh). The reflex of the pharyngealized voiced interdental E is a pharyngealized voiced alveolar fricative G (ex. 7 Gull ‘shadow’). In contrast with the Sudanese urban koiné (Dickins 2006: 560), the pharyngealized alveolar trill H and alveolar lateral approximant I are missing. (5)

kalām=u naEīf ʃadīd speech=3SG.M clean very ‘His language is very clear.’ (inf.4a)

(6)

namman biga Fabāh bi=na-miʃ when become\3SG.M morning IPFV=1PL-go ‘At the daybreak we go to the market.’ (inf.1a)

as=sūk DEF=market

Native and non-native varieties of Arabic in Kadugli 27

(7)

o-góod fi G=Gull IMP-sit in DEF=shadow ‘Sit down in the shadow.’ (inf.4a)

As for the etymological velar fricatives x and M, they are not subjected to change (ex. 8 MaraE ‘scope’; ex. 9 xallē-t=u ‘I left him’): (8)

ma=ind=i

MaraE m=o scope with=3SG.M ‘I have nothing to share with him.’ (inf.1a)

NEG=at=1SG

(9)

ali xallē-t=u gāid Ali leave-1SG=3SG.M sit\ACT.PTCP.SG.M ‘As for Ali, I left him at home.’ (inf.5a)

fi in

l=bēt DEF=house

Similarly to non-native speakers of Arabic (Section 4.3.), we can note that native speakers of Kadugli Arabic tend to realize the voiced palatal stop dʒ as a palatal glide y [j] when occurring close to central vowels (ex. 10 dʒāy=k > yāy=k ‘coming to you’; ex. 11. ȥawādʒa > ʔawāya ‘trouble’): (10) ana yāy=k 1SG come\ACT.PTCP.SG.M=2SG.M ‘I am coming to you.’ (inf.3a) (11) māfi ʔawāya NEG.EXS trouble ‘There’s no problem.’ (inf.2a) Looking at the morphology, Kadugli Arabic displays a typical urban pronominal/verbal paradigm that does not distinguish feminine plural as a morphological category. With regards to independent pronouns, we can note that, in common with sedentary Chadian dialects (Owens 1993: 108, 161), the 1PL has the form anīna corresponding to anī?na in Baggara Arabic.

28

Stefano Manfredi

Table 5. Kadugli Arabic – Personal pronouns and verb conjugation

1SG 2SG.M 2SG.F 3SG.M 3SG.F 1SG 2SG.M 2SG.F 3SG.M 3SG.F

Independent

Bound

ana inta inti huwa hiya Suffixed CvCvC -ta CvCvC -ta CvCvC -ti CvCvC CvCvC -at

=í, =ní =ak =ik =(h)u =(h)a Prefixed a-CCvC ta-CCvC ta-CCvC-i ya-CCvC ta-CCvC

Independent

Bound

1PL 2PL

anīna intum

=na =kum

3PL

hum

=hum

1PL 2PL

Suffixed CvCvC -na CvCvC -tu

Prefixed na- CCvC ta- CCvC -u

3PL

CvCvC -o

ya- CCvC -u

As regards the prefixed conjugation of two-consonantal verbs, we can note that the majority of the speakers have the tendency to elide the final vowel of verbal roots with C3 = y. The elision of final -i induces the insertion of an epenthetic vowel between C1 and C2 (ex. 12 a-mʃi > a-miʃ ‘go’). In these conditions, a contrast exists between 2SG.M and 3SG.F: ta-miʃ vs. tamʃi. According to Owens (1993: 124) this is a common characteristic of western Sudanic dialects spoken in Chad and Nigeria (see also Section 4.2. for Baggara Arabic). (12) a-miʃ as=sūk wa dʒīb nēt IMP-go DEF=market and brig\IMP joss stick ‘Go to the market and bring some joss stick.’ (inf.3a) Also in this case, educated speakers tend to align to the Sudanese koiné where final -i is never elided. (13) ma=ta-mʃi henāk NEG=2SG.M-go there ‘Don’t go there!’ (inf.4a) Another phono-morphological feature that Kadugli Arabic shares with western Sudanic dialects and that distinguishes it from Sudanese Arabic is the front stem vowel in the verb ʃāf ‘see’ (ex.14 b=a-ʃīf ‘I see’). This feature seems not to be particularly affected by dialect levelling.

Native and non-native varieties of Arabic in Kadugli 29

(14) ma=b=a-ʃīf kweyyis NEG=IPFV-1SG-see well ‘I can’t see well enough.’ (inf.5a) In Kadugli Arabic the preverbal marker b(i)= generally expresses an imperfective aspect associated with habitual, generic present and future meanings. (15) yōmi ana b=a-ʃrab daily 1SG IPFV=1SG-drink ‘I drink coffee every day.’ (inf.1a) (16) kandʒo b=i-gra fi Kanjo IPFV=3SG-study in ‘Kanjo studies at school.’ (inf.3a)

dʒabana coffee l=madrasa DEF=school

(17) inʃalla b=a-dʒi=kum bukra God_willing IPFV=1SG-come=2PL tomorrow ’God willing, I will come to visit you tomorrow.’ (inf.5a) In addition, youngest speakers are gradually integrating the Egyptian preverbal marker ?a= > ha= for expressing a future tense. According to Dickins (2006: 569) this tense marker is a 20th century borrowing that is now well established in Khartoum Arabic. (18) bukra h=a-kūn fi l=bēt tomorrow FUT=1SG-be in DEF=house ‘Tomorrow I will be at home.’ (inf.4a) As a final remark, we can note that analytic possession in Kadugli Arabic is expressed by two different particles: ?agg > hagg that is the default possessive marker in urban Sudanese Arabic and bitāȥ > bitā, which has been recently borrowed from Egyptian Arabic. In this regard, it should be remarked that, differently from the urban dialect of Khartoum (Dickins 2006: 570), in Kadugli possessive particles do not agree in gender and number with the possessed noun phrase.

30

Stefano Manfredi

(19) al=láab-āt hagg al=arab barā=hum DEF=dance-PL.F POSS DEF=arabs alone=3PL ‘Arabs’ dances stand alone.’ (inf.5a) (20) al=adʒala=dīk bitā=i DEF=bicycle=DIST.SG.F POSS=1SG ‘That bicycle is mine.’ (inf.3a)

4.2. Baggara Arabic Baggara dialects are generally classified within western Sudanic varieties of Arabic (Owens 1993: 145). Given the marginal socio-economic position occupied by Baggara Arabs in Kadugli, their Bedouin dialect is strongly affected by dialect levelling towards the Sudanese urban koiné. In view of that, Baggara informants have been chosen according to different periods of urbanization as well as to different degrees of formal education. In this regard, it should also be remarked that female speakers are generally less influenced by dialect levelling since they generally have a lower degree of formal education. Apart from inf.1b, who claims to speak Logorí as a second language, all the other Baggara informants are Arabic monolinguals. Table (6) summarizes the metadata for Baggara informants in Kadugli. Table 6. Baggara Arabic – Speakers profile

Age Gender Ethnicity inf.1b inf.2b inf.3b inf.4b inf.5b

54 61 25 29 35

F M F M M

Baggara Baggara Baggara Baggara Baggara

Urbanization Education Pattern of bilingualism 28 years none Arabic / Logorí 40 years religious Arabic 4 years none Arabic born in Kadugli university Arabic 3 years primary Arabic

In contrast with both the emerging urban dialect of Kadugli (Section 4.1.) and non-native varieties of Arabic (Section 4.3.), Baggara Arabic always retains the etymological pharyngeal consonants ? and ȥ (ex. 21 ?aya-t ‘life’, taȥb ‘tiredness’; ex. 22 ?ille ‘village’, gāȥid ‘sitting’). This phonological feature is not subject to levelling and, in the urban context of

Native and non-native varieties of Arabic in Kadugli 31

Kadugli, it can be considered as a communal isogloss restricted to speakers of Arab origins. (21) ?aya-t al=masār fōg=ha life-CS DEF=transhumance on=3SG.F ‘Nomadic life is wearying.’ (inf.1b) (22) al REL

fi in

l=?ille DEF=village

taȥb tiredness

gāȥid sit\ACT.PTCP.SG.M

wa dāk sāyir and DIST.SG.M move\ ACT.PTCP.SG.M ‘The one who lives in the village is sedentary while the other one moves for the transhumance’ (inf.2b) At the same time, Baggara Arabic retains all the etymological pharyngealized consonants (ex. 23 muFāt ‘wooden spoon’; ex. 24 an-i=ní ‘give me’) and, in line with other Arabic Bedouin dialects, it also presents non-etymological pharyngealized sounds H, Q and I (ex. 25 Hās=í ‘my head): (23) al=ȥasīde b=u-Fūt-u=ha be=muFāt DEF=porridge IND=3-stirr-PL.M=3SG.F by=wooden_spoon ‘The porridge is stirred with a wooden spoon.’ (inf.3b) (24) an-i=ní lē=ha give\IMP=1SG to=3SG.F ‘Give it to me.’ (inf.5b) (25) Hās=í gāȥd-e head=1SG sit\ACT.PTCP-SG.F ‘I have a headache.’ (inf.5b)

ti-wadʒdʒaȥ=ní 3SG.F-ache=1SG

In this context, dialect levelling affects only non-etymological pharyngealized consonants that, among early urbanized Baggara, tend to be realized as their non-pharyngealized counterparts (ex. 26 umm=i ‘my mother’ vs. aQQ=í among rural Baggara):

32

Stefano Manfredi

(26) umm=í sākn-a mother=1SG inhabit\ACT.PTCP-SG.F ‘My mother lives in Korom.’ (inf.4b)

fi in

korom Korom

It should also be stressed that, contrary to the urban dialect of Kadugli (Section 4.1.), in Baggara Arabic the pharyngealized voiced interdental E converges in the pharyngealized voiced dental stop E (ex. 27 Eull ‘shadow’), though educated speakers tend to realize it as a pharyn-gealized voiced alveolar fricative G in restricted lexical contexts (ex. b=i-?fiG ‘he preserves’). (27) anī?na bi=n-gūl aQQ=sanagīr Eull=ak ti?it 1PL IND=1PL-say high_noon shadow=2SG.M below ‘We call it aQQ sanagīr, when your shadow is just below you.’ (inf.1b) (28) aIIa b=i-?fiG nās God IND=3SG.M-preserve people ‘God protects straight people.’ (inf.4b)

al=xēr DEF=good

Like in other Bedouin Arabic dialects, in Baggara the voiced velar fricative M is realized as a voiceless uvular stop q (ex. 29 qanam ‘goats’). Also in this case, dialect levelling can induce more standardized realization among educated speakers (ex. 30 Māli ‘expensive’). (29) ȥindi=na bagar qanam wa Eān at=1PL cows goats and sheep ‘We have cows, goats and sheep.’ (inf.3b) (30) al=ȥēʃ biga Māli DEF=wheat become\3SG.M expensive ‘The mil became very expensive.’ (inf.4b)

ʃadīd very

In Baggara dialects backness harmony is the main phonological factor determining vowel quality (Owens 1993: 101). In compliance with backness harmony, three classes of vowels are distinguished: front vowels i, ī, e, ē; back vowels u, ū, o, ō; and neutral vowels a, ā. A given vowel can only occur along with vowels of the same class. Neutral vowels a, ā indifferently appear together with front and back vowels. Vowel harmony is either

Native and non-native varieties of Arabic in Kadugli 33

phonologically or morphologically determined. In the latter case, vowel harmony mainly intervenes in consequence of suffixation: back suffix vowels cause the vowels of a given stem to be realized as back vowels (ex. 31 farīg ‘camp’, furūk=hum ‘their camp’). This feature is also subjected to levelling among early urbanized Baggara speakers (ex. 32 lē=ku ‘to you’ vs. lō=ku). (31) nās hāʃi gāȥd-īn fu furuk=hum people Hashi sit\ACT.PTCP-PL.M in camp=3PL.M ‘Hashi’s relatives are in their camp.’ (inf. 1b) (32) sārih lē=ku atú graze\ACT.PTCP.SG.M to=2PL.M who ‘Who’s grazing for you today?’ (inf.2b)

al=yōm DEF=day

Looking at syllable structure, Baggara Arabic aligns with other Sudanic Bedouin dialects (Reichmuth 1983: 23) with regards to the presence of syllable internal cluster in the Old Arabic scheme C1vC2C3 (ex. 33 kalb ‘dog’). These syllable internal clusters tend to disappear among Baggara speakers born in Kadugli (ex. 34 Marib ‘west’ vs. qarb). (33) al=kalb ?iss=a DEF=dog voice=3SG.M ‘The dog barks.’ (inf. 5b)

b=i-nba? IND=3SG.M-bark

(34) al=fellāta dʒ-o min al=Marib DEF=Fellata come-3PL.M from DEF=west ‘Fellata people came from the west.’ (inf. 4b) In the same way as Baggara Arabic dialects of Chad and Nigeria (Owens 1993: 93), the vowel suffix marking feminine singular has the two forms -a and -e depending on the presence of front vowels in the nominal stem (ex. 35 baggāriy-e ‘Baggara-SG.F’). However, as a consequence of dialect levelling towards the Sudanese urban koiné, urbanized Baggara tend to generalize the use of the feminine marker -a (ex. 36 kabīr-a ‘big-SG.F). (35) al=maHa=di baggāriy-e tax DEF=woman=PROX.SG.F baggara-SG.F IDPH ‘This women is a truthfully Baggara’ (inf.5b)

34

Stefano Manfredi

(36) aF=Furȥa=di kabīr-a DEF=wrestling=PROX.SG.F big-SG.F ‘This wresting match is huge.’ (inf.2b)

marra wā?id time one

Being a Bedouin dialect, Baggara Arabic presents a maximally contrastive pronominal/verbal paradigm distinguishing feminine plural as an independent morphological category (Manfredi Fc-a). Table 7. Baggara Arabic – Personal pronouns and verb conjugation

Independent ana itte itti hu hi Suffixed 1SG CvCvTC(-t) 2SG.M CvCvTC(-t) 2SG.F CvCvC -ti 3SG.M CvTCvC 3SG.F CvCvC -at 1SG 2SG.M 2SG.F 3SG.M 3SG.F

Bound =í, =ní =ak =ki =a =ha Prefixed a-/nv- CCvC tv- CCvC tv- CCvC -i i- CCvC tv- CCvC

Independent

Bound

anī?na =na uttu =ku ittan =kin hum =hum hin =hin Suffixed Prefixed 1PL CvCvC -na nv- CCvC(-u) 2PL.M CvCvC -tu tv- CCvC -u 2PL.F CvCvC -tan tv- CCvC -an 3PL.M CvCvC -o i- CCvC -u/-o 3PL.F CvCvC -an i- CCvC -an 1PL 2PL.M 2PL.F 3PL.M 3PL.F

Even if the majority of urbanized Baggara Arabs retain feminine plural as a morphological category, some speakers tend to generalize the use of masculine plural indexes (ex. 37 b=i-gr-u ‘they M/F study’). (37) al=banāt b=i-gr-u fi l=madrasa DEF=girl\PL IND=3-study-PL in DEF=school ‘The girls are attending school.’ (inf.4b) Looking at the suffixed conjugation of three consonantal verbs, the most striking feature of Baggara Arabic is the elision of the morpheme -t marking 1SG and 2SG.M pronominal subjects: katáb ‘I, you M wrote’ vs. kátab ‘he wrote’. In truth, the morpheme -t regularly appears after twoconsonantal and doubled verbs and before bound pronouns and lexical objects: ma∫ē-t ‘I, you M went’, Earab-t=a ‘I beat him’. Besides, rural Baggara Arabic also presents the allomorph -ta, which occurs before consonant-initial objects: wiFil-ta kudugli ‘I, you M arrived in Kadugli’. Ur-

Native and non-native varieties of Arabic in Kadugli 35

banized Baggara, for their part, increasingly adopt the personal index -ta as found in the Sudanese prestigious norm (ex. 38 akal-ta ‘I ate’). (38) ya towwa ma=akal-ta until now NEG=eat-1SG ‘I still did not ate.’ (inf.2b) As shown in table (7), the Baggara dialect of Kordofan presents a mixed prefixed paradigm in which a- and n- co-occur in marking the 1SG person and n- marks plural in optional combination with the plural suffix -u. However, at the present time, the majority of urbanized Baggara speakers adopted the urban opposition a- 1SG (ex. 39 b=a-dirr ‘I pour’) vs. n- 1PL (ex. 40 na-?dir ‘we attend): (39) b=a-dirr al=mulā? fōg al=ȥēʃ IND=1SG-pour DEF=sauce on DEF=wheat ‘I pour the sauce on the wheat.’ (inf.1b) (40) lāzim na-?dir al=ȥirse=di must 1PL-attend DEF=marriage=PROX.SG.F ‘We have to attend this marriage.’ (inf.4b) b(i)= is the only preverbal marker found in Baggara Arabic. b(i)= correlates with an indicative mood related to ordinary objective statements in the present and in the future. Differently from the speakers of the emerging urban dialect of Kadugli, Baggara Arabs did not integrate the Egyptian future marker ?a= (Section 4.1.). (41) ana yōt b=a-gabl=a 1SG daily IND=1SG-meet=3SG.M ‘I meet him daily.’ (inf.2b) (42) ma=b=i-lgá be=suhūla NEG=IND=3SG.M-find\3SG.M with=simplicity ‘He won’t find him effortless.’ (inf.3b) In Baggara Arabic verbal roots with C3=y elide the final vowel in the prefixed conjugation (ex. 43 a-miʃ ‘go’) but, like in Kadugli Arabic, this west-

36

Stefano Manfredi

ern Sudanic morphological trait is affected by dialect levelling towards the Sudanese urban koiné (ex. 44 ta-mʃi ‘you go’). (43) a-miʃ ke wa ȥaddi bedʒāy IMP-go like_this and head.IMP by_this_side ‘Go this way and then head for this side’ (inf.5b) (44) ta-mʃi wala ta-dʒi ? 2SG.M-go or 2SG.M-come ‘Are you leaving or you are coming?’ (inf.4b) As a final remark, the Baggara dialect of Kordofan is characterized by the co-occurrence of two etymologically unrelated possessive markers. These are hān (SG.F hint, PL hinē), which is characteristic of western Sudanic dialects (Roth-Laly 1979: 139; Zeltner and Tourneux 1986: 45; Owens 1993: 64–66, Abu-Absi 1995: 45), and hūl (SG.F hīl, PL hilēl), which is instead found in Bedouin dialects of eastern Sudan (Reichmuth 1983: 111– 112). In this overall situation, early urbanized Baggara use more and more the urban possessive marker ?agg (ex. 47 ?aggi=na ‘our’) but they have not yet integrated the Egyptian bitāȥ. (45) bi=n-daffig ille al=elmi hān aH=Huwāba IND=1PL-spill except DEF=water POSS.SG.M DEF= curdled_milk ‘We throw away only the water of the curdled milk.’ (inf.3b) (46) al=bagar-a=di DEF=cows-SING=PROX.SG.F ‘This cow is mine.’ (inf.2b)

POSS.SG.F=1SG

hīl=í

(47) aʃ=ʃoMol=da DEF=stuff=PROX.SG.M ‘This stuff is ours.’ (inf.4b)

POSS=1PL

ana 1SG

?aggi=na

4.3. Non-native Arabic in Kadugli Non-native varieties of Arabic used as a vehicular means of communication have been attested in several African countries. Still, information about those varieties is rather scarce, mainly as a consequence of their high

Native and non-native varieties of Arabic in Kadugli 37

degree of individual variation. Ferguson (1970) analyzed the use of Arabic as a market lingua franca on the Ethiopian-Sudanese border. Jullien de Pommerol (1997), on his part, described the vehicular variety of Arabic spoken by sedentary non-Arab groups of western Chad. More recently, Simeone-Senelle (1999; 2006) gave information of a vehicular variety of Arabic used by ‘Afar and Saho bilingual speakers in Eritrea. In all the before mentioned geographical contexts, Arabic as a non-native language reproduces the main grammatical features of local dialectal varieties, albeit with a number of phonological and morphological peculiarities due to language contact (Miller 2006: 61). As far as Kadugli and the Nuba Mountains are concerned, Arabic is still used as a vehicular language by both local and outsider non-Arab groups. In this context, the choice of the informants representing non-native speakers of Arabic has been made according to three main criteria. First of all, as in a typical interethnic medium, the native language of the speakers plays an important role in shaping the phonology of Arabic as a vehicular language. On that account, selected informants are native speakers of genetically unrelated languages with different phonological systems, i.e. Korongo and Miri (Nilo-Saharan, Kadu), Logorí (Nilo-Saharan, Daju), Fulfulde (Niger-Congo, Fula). Secondly, individual variation in non-native Arabic can also be due to different degrees of proficiency. For instance, Nuba speakers who resided in Kadugli for a long time, and who have been thus persistently in contact with a majority of native speakers of Arabic, are less subject to phonological variation and morphological simplification than other non-native speakers. In view of that, the selected informants are representative of different periods of urbanization. Thirdly, variation in Arabic as a non-native language may also be related to exposure to different target varieties. If Nuba groups coming form the western mountains (i.e. Miri) seem to have been more exposed to Sudanese Arabic, Nuba groups coming from eastern and southern mountains (i.e. Logorí and Korongo) clearly have Baggara Arabic as their target language. Accordingly, the choice of the informants also took into account the different contact modalities between Arabic dialects and minority languages of the Nuba Mountains. Table (8) summarizes the metadata for non-native speakers of Arabic in Kadugli.

38

Stefano Manfredi

Table 8. Non-native Arabic – Speakers profile

Age Gender Ethnicity Urbanization Education inf.1c inf.2c inf.3c inf.4c inf.5c

42 48 63 55 42

F M F M M

Reika Korongo Miri Logorí Fellata

7 years 10 years 35 years 15 years 1 year

Pattern of bilingualism none Korongo/Arabic primary Korongo/Arabic none Miri/Arabic primary Logorí/Arabic religious Fulfulde/Hausa/ Arabic

In the same way as the emerging urban dialect of Kadugli (Section 4.1.), Arabic as a non-native language is characterized by the absence of pharyngeal consonants. This notwithstanding, the outputs of this phonological simplification are rather variable. More in detail, the voiceless pharyngeal fricative ? can be realized either as a voiceless laryngeal fricative [h] (ex. 48, ?ajar al=makk > hajar al=makk ‘Lajar al-Makk’; ex. 50 ?ukūma > hukūma ‘government’) or as a voiceless velar occlusive [k] (wa?id-īn > wakid-īn ‘some of them’) depending on the nature of the surrounding vowels. Besides, ? is often elided when occurring in final position (ex. 49, mil(e)? > mile ‘salt’). (48) ana sakan-ta hini fi hajar al=makk 1SG inhabit-1SG here in Lajar al-Makk ‘I installed myself here in Lajar al-Makk’ (inf.2c) (49) zaman=dāk mile māfi time=DIST.SG.M salt NEG.EXS ‘At that time, there was no salt.’ (inf.3c) (50) inʃalla ta-lga wakid-īn ʃaggāl hukūma God_willing 2SG.M-find one-PL.M working government ‘You can also find someone who works in the administration.’ (inf.5c) The voiced pharyngeal fricative ȥ, on its part, can be realized either as a glottal stop [ʔ] when occurring in initial position (ex. 51, ȥasīde > ʔasīde ‘porridge’; ex. 52, abū ȥidʒāl > abū ʔidʒāl ‘Abū ‘Ijāl’; ex. 53 ȥarabi > ʔarabi ‘Arabic’) or it can be elided in middle and final position (ex. 51,

Native and non-native varieties of Arabic in Kadugli 39

bitāȥ > bitá ‘of’; ex. 52, gaȥd-na > gáad-na ‘we sat down’, maȥ > ma ‘with’; ex. 53 gāȥid > gāid ‘sitting’). (51) na-ml-u ʔasīde wa na-ml-u mulā bitā corōro 1-do-PL porridge and 1-do-PL sauce POSS Cororo ‘We prepare the porridge and we also prepare the Cororo sauce.’ (inf.1c) (52) kān

zamān gáad-na ma abū ʔidʒāl long_time_ago sit-1PL with Abū_‘Ijāl ‘Long time ago we used to live along with the Abū ‘Ijāl Arabs.’ (inf.4c) ANT

(53) al=ʔarabi ana gāid na-laggit ʃweya~ʃweya DEF=Arabic 1SG sit\ACT.PTCP.SG.M 1SG-fail few~few ‘As for the Arabic language, I am not very fluent.’ (inf.5c) It should be stressed that the tendency to drop pharyngeal consonants has been equally observed in other non-native varieties of Arabic spoken in Sudan (Miller and Abu Manga 1992: 78; Abu Manga 2006: 255) and in Chad (Jullien de Pommerol 1997: 125). As a further matter, all non-native speakers of Arabic in Kadugli lost etymological pharyngealized consonants. Like in Arabic-based pidgins and creoles (Tosco and Manfredi Fc) and other vehicular varieties of Arabic (Jullien de Pommerol 1997: 126, Simeone-Senelle 1999: 157), pharyngealized consonants are ge-nerally realized as their non-pharyngealized counterparts (ex. 54 -arʃān > tarsān ‘deaf’; ex. 55 abyaE > abyad ‘white; ex. 56 na-guFF-u > na-guss-u ‘we cut’). Given the overall absence of ‘emphatic’ consonants in the indigenous languages of the Nuba Mountains, this process of phonological convergence is clearly due to substrate interference. (54) wild-ó tarsān keda bear-3PL\3SG.M deaf like_this ‘He was simply born deaf’ (inf.4c) (55) sokol abyad mitil=da stuff white like=PROX.SG.M ‘A white stuff like this’ (inf.2c)

40

Stefano Manfredi

(56) na-guss-u al=gēs wa na-rdʒ-u fi l=hille 1-cut-PL DEF=hay and 1-come_back-PL in DEF=village ‘We cut the hay and then we come back to the village.’ (inf.3c) As regards the voiceless velar fricative x, it can be realized either as a laryngeal fricative [h] (ex. 57 xallē-t=a > hallē-t=a ‘I let it’; ex. 58 xarbān > harbān ‘destroyed’) or as a voiceless velar plosive [k] (ex.57 xalāF > kalās ‘definitively’). This does not concern Logorí speakers (inf.4c), given that in Daju languages x is a phonemic segment (Thelwall 1981: 45; Manfredi Fc-c). (57) ana hallē-t=a kalās 1SG let-1SG=3SG.F definitively ‘I definitively let it.’ (inf.1c) (58) al=balad=da harbān DEF=country=PROX.SG.M ruined ‘This country is ruined.’ (inf.3c) The voiced velar fricative M, on its part, can be realized as a voiced velar plosive [g] (ex. 59 Māba > gāba ‘forest’), as a voiceless velar plosive [k] (ex. 60 na-zaMrat > na-sakrat ‘we cry aloud’), or as voiceless velar fricative [x] among Logorí speakers (ex. 61 ni-ʃtaMal > ni-staxal ‘I work’). (59) ana na-ʃīl=u min gāba 1SG 1SG-bring=3SG.M from forest ‘I bring it from the forest.’ (inf.5c) (60) na-msi bes na-skarit fōg=a 1PL-go only 1PL-cry_aloud on=3SG.M ‘We go and we cry aloud for him.’ (inf.1c)) (61) ana ni-staxal bes 1SG 1SG-work only ‘I just work.’ (inf.4c) Probably as a consequence of his religious education, inf.5c displays several instances of hypercorrection in which the voiceless velar plosive k

Native and non-native varieties of Arabic in Kadugli 41

is realized as velar fricative [x] (ex. 62 kādūgli > xādūgli ‘Kadugli’; ex. 63 henāk > henāx ‘there’). (62) ana dʒī-t fi xādūgli sana 1SG come-1SG in Kadugli year ‘I arrived in Kadugli one year ago.’ (inf.5c)

alfāt last

(63) garē-t fi l=xalwa henāx fi nidʒērya study-1SG in DEF=religious_school there in Nigeria ‘I studied in the religious school, there, in Nigeria.’ (inf.5c) Among all non-native speakers of Arabic in Kadugli the voiceless palatoalveolar fricative ʃ tends to be realized as a voiceless alveolar sibilant [s] (ex. 64 ʃuffaȥ > suffa ‘children’; ex. 65 ʃunú > sunú ‘what’). This is another phonological feature in common with both Arabic-based pidgins and creoles and other non-native varieties of Arabic in Africa (Simeone-Senelle 1999: 160; 2006: 270; Tosco and Manfredi Fc). (64) ana dʒib-ta saba suffa 1SG bring-1SG seven children ‘I bore seven children.’ (inf.1c) (65) fi

sunú tāni ? EXS what second ‘What else?’ (inf.3c)

Given the absence of a voiced alveolar sibilant in Korongo (Schadeberg 1981: 298), inf.1c and inf.2c tend to realize z as a voiceless sibilant [s] (ex. 66 zōl > sōl ‘man’; ex. 67 nazal-ta > nasal-ta ‘I got down’). (66) as=sōl=da masak=ní kalās DEF=man=PROX catch\3SG.M=1SG definitively ‘This man finally married me.’ (inf.1c) (67) nasal-ta hini ind=i asara sene get_down-1SG here at=1SG ten year ‘I arrived here ten years ago.’ (inf.2c)

42

Stefano Manfredi

Like in the emerging urban dialect of Kadugli (Section 4.1.), certain speakers tend to realize the voiced palatal stop dʒ as a palatal glide y [j] in the proximity of central vowels. (68) mimkin yāb-o hāya possible bring-3PL thing ‘It’s possible that they brought something new.’ (inf.3c) As already mentioned, Korongo and Logorí speakers have been historically exposed to the Arabic dialect spoken by Baggara nomads in rural areas of the Nuba Mountains. On that account, inf.1c, inf.3c, and inf.4c display typical features of Baggara Arabic such as the feminine marker -e after front vowels (ex. 70 medīne ‘city’) and the form =a for the 3SG.M bound person pronoun (ex. 71 bēt=a ‘his house’, Section 4.2.). (69) al=medīne ahsan min al=kala DEF=city better from DEF=desert ‘The city is better than the savannah’ (inf.3c) (70) bēt=a baīd sadīd house=3SG.M far very ‘His house is very far.’ (inf.4c) Considering the morphosyntax of Arabic as a non-native language, it should be stated that, differently from Arabic-based pidgins and creoles, it largely preserved two series of personal pronouns (independent and bound), personal indexes on the verbs, as well as internal plural patterns. Among Nuba speakers the 1PL independent personal pronoun is nīna. (71) nīna akal-na sabā bedri 1PL eat-1PL morning early ‘We ate early in the morning.’ (inf.4c) inf.5c instead uses the 1PL independent pronoun nānu which clearly derives from the Old Arabic form na?nu. (72) nānu fellāta kullu wāhid 1PL Fellata all one ‘We as Fellata, we are all the same.’ (inf.5c)

Native and non-native varieties of Arabic in Kadugli 43

Similarly to the vehicular variety of Arabic spoken in Eritrea (SimeoneSenelle 1999: 173), all non-native speakers of Arabic in Kadugli tend to generalize the use of the 1PL bound pronoun also for the 1SG person (ex. 73–74, indi=na ‘I have’): (73) ána indi=na itnēn wa 1SG at=1(SG) second and ‘I am forty-two years old.’ (inf.5c)

arbaīn forty

sana year

(74) ma=indi=na kalām tāni NEG=at=1(SG) speech second ‘I have nothing more to say.’ (inf.1c) In the same manner, the 1SG person marker of the imperfective often coalesces with the 1PL person (ex. 75 na-ganni ‘I sing’; ex. 76 na-mʃi ‘I go’; ex. 77 na-kubb ‘we pour’): (75) ita dāir na-ganni 2SG.M want\ACT.PTCP.SG.M 1(SG)-sing ‘Do you want me to sing for you?’ (inf.1c)

lē=k ? to=2SG.M

(76) ana dāir na-mʃi al=umura 1SG want\ACT.PTCP.SG.M 1(SG)-go DEF=pilgrimage ‘I want to do the pilgrimage (to Mecca).’ (inf.5c) (77) na-ml-u serībe na-kubb ēs fōgo 1-do-PL enclosure 1(PL)-pour wheat inside ‘We build an enclosure and we put the wheat inside.’ (inf.3c) Independently of their ethnolinguistic background, late urbanized nonnative speakers of Arabic tend to elide the personal indexes of both 3SG and 3PL persons in the prefixed conjugation of three consonantal verbs (ex. 78 ø-amul ‘he does’, ø-adba ‘he slaughters’; ex. 79 ø-aml-u ‘they do’). (78) kan sōl dāir Ø-amul karāma Ø-adba tōr if man want\ACT.PTCP.SG.M Ø-do sacrifice Ø-slaughter bull ‘If someone wants to do a sacrifice, he slaughters a bull.’ (inf.2c)

44

Stefano Manfredi

(79) nās Ø-aml-u merīsa katīr people Ø-do-PL beer much ‘People make a lot of beer.’ (inf.5c) All non-native speakers living in Kadugli present an irregular use of the preverbal marker bi=. For instance, in example (80) bi= marks an imperfective after the deontic modal verb lāzim ‘must’. (80) lāzim bi=ta-kul hāja must IND=2SG.M-eat thing ‘You have to eat something.’ (inf.3c) As in most vehicular languages (Miller 2006: 62; Simeone-Senelle 2006: 272), agreement in gender and number is highly irregular even between subject and auxiliary verbs. In example (81) the progressive auxiliary gāid does not agree in gender with the female subject-speaker. In a similar way, in example (82) the auxiliary gāid does not agree in number with the 1PL pronoun nīna. (81) ana ja hini ana gāid murtá 1SG come here 1SG sit\ACT.PTCP.SG.M happy ‘I came here (in Kadugli) and I am happy.’ (inf.1c) (82) fi sibir nīna gāid na-mul in ritual 1PL stay\ ACT.PTCP.SG.M 1(PL)-do ‘During the ritual, we organize the wrestling.’ (inf.4c)

surá wrestling

Regarding the nominal morphosyntax we can note that non-native speakers of Arabic are largely characterized by an irregular use of the definite article al=. In the following example, the indefinite adjective tegīl ‘heavy’ modifies the definite noun al=kalām ‘the speech’: (83) al=kalām tegīl=da ana ma=a-fham=u DEF=speech heavy=PROX.SG.M 1SG NEG=1SG-understand=3SG.M ‘As for this difficult language, I do not understand it.’ (inf.3c) Like in non-native varieties of Arabic spoken by rural migrants in Khartoum (Miller and Abu Manga 1992: 73), in Kadugli we can also

Native and non-native varieties of Arabic in Kadugli 45

observe the generalized use of the preposition fi ‘in’ for introducing indirect arguments of motion verbs. (84) rajáa-na fi kūdūgli come_back-1PL in Kadugli ‘We come back to Kadugli.’ (inf.2c) Non-native speakers of Arabic tend to generalize the use of analytic possessive constructions (Miller 2006: 64). In Kadugli, it seems that they mainly use the Egyptian possessive particle bitāȥ, which is realized either as bitā (ex. 85) or as ta (ex. 86) as in the Sudanic Arabic-based pidgins and creoles (Tosco and Manfredi Fc). (85) ahal bitā=i sākin family POSS=1SG inhabit\ACT.PTCP.SG.M ‘My family lives in Umm Bres’ (inf.4c)

fi umm brēs in Umm Bres

(86) hille ta=na begādi al=hayar village POSS=1PL by_that_side DEF=mountain ‘Our village is behind the mountain.’ (inf.3c) As a final remark, we note that, differently form the vehicular variety of Arabic spoken in Eritrea (Simeone-Senelle 1999: 170), the numerous substrate languages of the Nuba Mountains do not affect the syntax of nonnative varieties of Arabic in Kadugli, which always display a basic SVO syntactic order.

5. Conclusion Kadugli is of particular interest for Arabic dialectology because of the present-day coexistence of both native and non-native varieties of Arabic. This atypical sociolinguistic situation gives us the possibility to enlarge our understanding of the impact of urbanization in arabophone countries. As a general remark, in consequence of its relatively late process of urbanization, Kadugli still does not possess a shared dialectal norm. At the same time, the strong prestigious status occupied by the Sudanese national koiné undoubtedly inhibits the affirmation of an innovative urban dialect and reduces the maintenance of regional rural dialects (i.e. Baggara

46

Stefano Manfredi

Arabic). The high degree of individual variation affecting non-native Arabic impedes a comprehensive evaluation of its phonological and morphosyntactic features. Nevertheless, it is clear that non-native varieties of Arabic used as vehicular means of communication are distinct from both native Arabic dialects and Arabic-based pidgins and creoles. Although it is true that a certain amount of morphological generalization is the rule, differently from Arabic-based pidgins and creoles, verbal inflection and derivation are largely preserved. Likewise, non-native Arabic generally retains two series of personal pronouns (independent and bound), and a certain amount of Arabic plural patterns. The study on non-native varieties of Arabic gives evidence of the direct relation between the different socio-economic contexts of language contact and their linguistic outcomes. As a further matter, non-native varieties of Arabic may have theoretical relevance for historical dialectology if we consider that new dialects always develop in contact environments. On that account, the most striking phonological feature of the native variety of Arabic spoken by local non-Arab groups (i.e. the absence of pharyngeal consonants) is evidently related to the previous diffusion of Arabic as a non-native language. Table (9) summarizes some of the main phonological and morphosyntactic divergences between native and non-native varieties of Arabic found in Kadugli (note that the chevron > signals the effects of the ongoing dialect levelling towards the Sudanese koiné). Table 9. Arabic varieties in Kadugli

*?, *ȥ *-, *E, *F, *E *x, *M *z, *ʃ SG.F *-a 1PL pronoun 1SG, 1PL marking preverbal markers possessive marker

Kadugli Arabic h > ? , ʔ/Ø > ȥ -, E, F, G x, M z, ʃ -a anīna a- 1SG n- 1PL bi= imperfective ha= future hagg > bitā

Baggara Arabic ?, ȥ -, E, F, E > G x, q > M z, ʃ -a/-e > -a anī?na a-/n- 1SG > an-...(-u) 1PL > nbi= indicative hān, hūl > ?agg

Non-native Arabic h/k/Ø, ʔ/Ø t, d, s, z/s k/h, g/k/x z/s, s -a/-e nīna, nānu n- 1SG, 1PL bi= irregular values bitā, ta

References Abu Absi, Samir 1995 Chadian Arabic. Munich: Lincom Europa. Abu-Manga, al-Amin 2006 Hausa. In Encyclopaedia of Arabic Language and Linguistics, Vol. II, Kees Versteegh (ed.), 250–256. Amsterdam: Brill. Barth, Fredrik 1969 Introduction. In Ethnic Groups and Boundaries. The Social Organization of Culture, Fredrik Barth (ed.), 9–38. Bergen: Universitets Forlaget. Bell, Herman 1995 The Nuba Mountains: Who Spoke What in 1976? Paper presented to the Third Conference on Language in Sudan (Language Situation in Sudan), University of Khartoum 3–5 December 1995. Dickins, James 2006 Khartoum Arabic. In Encyclopaedia of Arabic Language and Linguistics, Vol. II, Kees Versteegh (ed.), 559–571. Amsterdam: Brill. Ferguson, Charles 1970 The role of Arabic in Ethiopia: A sociolinguistic perspective. In Languages and Linguistics Monograph series 23, 355–68. Georgetown: Georgetown University Press. Ille, Enrico 2011 Tracing golden past. Historical narratives about Shaybun and Shawabna in the Nuba Mountains, Sudan. Leipzig: Ille and Riemer. Jullien de Pommerol, Patrice 1997 L’arabe tchadien. Émergence d’une langue véhiculaire. Paris: Karthala. 2006 Chad Arabic. In Encyclopaedia of Arabic Language and Linguistics, Vol. I, Kees Versteegh (ed.), 360–368. Amsterdam: Brill. Manfredi, Stefano Fc-a Kordofanian Baggara Arabic. Wiesbaden: Harrassowitz. Fc-b One tribe, one language. Ethnolinguistic identity and language revitalization in the Nuba Mountains. The case of Logorí. In Reshaping Livelihoods, Political Conflict and Identities in Contemporary Sudan, Barbara Casciarri, François Ireton, and Monzoul Assal (eds.). London: Berghahn. Fc-c Arabic borrowings in Laggorí (Eastern Daju). In Acts of the 1st Nuba Mountains Language Conference, Thilo Schadeberg, and Roger Blench (eds.). Cambridge: Kay Williamson Educational Foundation.

48

Stefano Manfredi

Miller, Catherine 2002 The relevance of Arabic-based pidgins-creoles for Arabic linguistics. In Al-lugha, Gerda Mansur and Madiha Doss (eds.), 7–46. Cairo: Arab Development Center. 2004 Variation and change in Arabic urban vernaculars. In Approaches to Arabic Dialects: Collection of Articles presented to Manfred Woidich on the Occasion of his Sixtieth Birthday. Martine Haak, Kees Versteegh, and Rudolf De Jong (eds.), 177–206. Amsterdam: Brill. 2006 Lingua Franca. In Encyclopaedia of Arabic Language and Linguistics, Vol. III, Kees Versteegh (ed.), 58–64. Amsterdam: Brill. 2007 Arabic urban vernaculars: development and change. Arabic in the City. Issues in Dialect Contact and Language Variation, Catherine Miller, Enam al-Wer, Dominique Caubet, and Jannet Watson (eds.), 1–30. London: Routledge. Miller, Catherine, and Al-Amin Abu Manga 1992 Language Change and National Integration. Rural Migrants in Khartoum. Khartoum: Khartoum University Press. Owens, Jonathan 1993 Nigerian Arabic in Comparative Perspective. Sprache und Geschichte in Afrika 14: 85–175. Quint, Nicolas 2006 Do you speak Kordofanian? Paper presented at 7th International Sudan Studies Conference, 6–8 April 2006, University of Bergen, Norway. Reichmuth, Stefan 1983 Der arabische Dialekt der Shukriyya im Ostsudan. Zürich: Olms. Roth-Laly, Arlette 1979 Esquisse grammaticale du parler arabe d’Abbéché. Paris: Geuthner. 1994 Dialectologie comparée au Tchad et au Sudan. De quelques matériaux et discriminants. In Materiaux Arabes et Sud-Arabiques 6: 69–102. Schadeberg, Thilo 1981 The classification of the Kadugli language group. In Nilo-Saharan, Thilo Schadeberg, and Marvin Bender (eds.), 291–305. Dordrecht: Foris Publications. 1989 Kordofanian. In The Niger-Congo Languages. A classification and description of Africa's largest language family, John BendorSamuel, and Rhonda Hartell (eds.), 66–80. Lanham: University Press of America. 1994 Comparative Kadu Wordlist. Afrikanistische Arbeitspapiere 40: 11– 48.

Native and non-native varieties of Arabic in Kadugli 49 Simeone-Senelle, Marie-Claude 1999 L’arabe véhiculaire parlé en Érythrée sur la côte sud de la Mer Rouge de Massawa à Rahayta. Oriente Moderno XIX (LXXX): 153–181. 2006 Horn of Africa. In Encyclopaedia of Arabic Language and Linguistics, Vol. II, Kees Versteegh (ed.), 268–275. Amsterdam: Brill. Stevenson, Roland 1984 The Nuba People of Kordofan Province, An Ethnographic survey. Khartoum: Khartoum University Press. Thelwall, Robin 1981 The Daju Language Group. Systematic Phonetics, Lexicostatistics and Lexical Reconstruction. Ph.D. thesis, New University of Ulster. Thelwall, Robin, and Thilo Schadeberg 1983 The linguistic settlement of the Nuba Mountains. Sprache und Geschichte in Afrika 5: 219–231. Tosco, Mauro, and Stefano Manfredi Fc Pidgins and Creoles. In The Oxford Handbook of Arabic Linguistics, Jonathan Owens (ed.). Oxford: Oxford University Press. Zeltner, Jean-Claude, and Henry Tourneux 1986 L’arabe dans le bassin du Tchad. Paris: Karthala.

Reinventing negation patterns in Moroccan Arabic Mena Lafkioui 1. Introduction The present chapter examines how new negation patterns are created in Moroccan Arabic in the Oujda region, North East Morocco, through contact with Berber (Tarifit language, North Morocco, see Figure 1).1 The discussed cases concern contact-induced innovation processes in Moroccan Arabic negation, through which the morphological data as well as the syntactic structuring and semantic functioning have been modified by analogy with Berber negation. The negation system of North African Arabic is fundamentally built on the morphosyntactic opposition between verbal negation and non-verbal negation. Where in the former negation type a verbal component functions as the negation incidence point (generally the verbal predicate or its auxiliary), in the latter type, the negation refers to various nominal elements, such as nouns and prepositions, that generally play the predicative role. Verbal predication is negated by means of the discontinuous marker ma ___ s− (s− i/s− ay) and its optional or conditioned (modal and expressive) variants. The verbal negation structure mainly expresses existential values, whereas the non-verbal negation structure is used for both existential and attributive semantic purposes. The existential negation marker is similar in verbal and non-verbal negation, while attributive negation is mainly marked by the continuous morpheme mas− i ___ or mus− ___ (or their variants). With respect to this negation system, the Moroccan Arabic variety of Oujda distinguishes a new discontinuous marker, ma ___ bu, whose second element is borrowed from Tarifit, which is the only Berber language where this morpheme occurs (Lafkioui 1996, 1999: II/Chapter 2, 2007: 234–236, 2011a: 62–69). They do not only have this morphological particularity in common, but also its morphosyntactic combinatorial restrictions and syntactic functioning (structural analogy). The hybrid negator ma ___ bu occurs in the following innovated negation patterns: [ma + verbal predicative 1.

The dark grey zone on the map corresponds to the Berber speaking area of the Rif that distinguishes this phenomenon.

52

Mena Lafkioui

syntagm + bu + noun] and [ma + non-verbal predicative syntagm + bu + noun]. Moreover, this Moroccan Arabic variety basically shares the same semantic properties with Tarifit, that is, it is also employed for existential descriptive negation. The analyses referring to Moroccan Arabic and Berber in this chapter are based on data from my fieldwork in Morocco from 1992 onwards. As for the negation in Tarifit, I also refer to Lafkioui (1996, 1999: II/ Chapter 2, 2007: 234–236 and 2011a: 62–69). The personally gathered data exposed here were transcribed according to the phonological system of the languages at hand, in which only the distinctive value and length of the vowels and consonants are represented. Quoted data, however, are reproduced as such in this contribution.2 This introduction is followed by an examination of the negation system in North African Arabic languages (Section 2). In Section 3, the case of Moroccan Arabic is considered in detail by distinguishing between verbal and non-verbal negation and by analyzing their respective morphosyntactic patterns and semantic values. Section 4 addresses the contact-induced innovated negation marker ma ___ bu and its combinatorial and distributional implications on the present negation structure and functioning in Moroccan Arabic. The same section is dedicated to the question of borrowing and its development on the basis of the examined data. Finally, this chapter ends with a number of conclusions about contact-induced language variation and its connection with language diffusion and evolution.

2.

The following abbreviations are used: AFF ‘affix’, COMPL ‘complement’, COP ‘copula’, DEF ‘definite’, DET ‘determined’, DIR ‘direct’, DIST ‘distal’, EA ‘emphasis accent’, F ‘feminine’, Fo ‘pitch’, I ‘intensity’, IMP ‘imperative’, IMPERF ‘imperfective’, INDIR ‘indirect’, INTRG ‘interrogative’, INVOC ‘invocation’, M ‘masculine’, MAX ‘maximum’, MIN ‘minimum’, NEG ‘negation’, NVPS ‘nonverbal predicative syntagm’, O ‘object’, P ‘predicate’, PART ‘particle’, PERF ‘perfective’, PL ‘plural’ PREP ‘preposition(al)’, PROP ‘proposition(al)’, PROS ‘prosody’, PROX ‘proximal’, PTCP ‘participle’, REF ‘referential’, S ‘subject’, SG ‘singular’, SUB ‘subordination’, SUBcompl ‘completive subordination’, TOP ‘topic(alization)’, UNDET ‘undetermined’, VPS ‘verbal predicative syntagm’, ↑ ‘rise in pitch’, ↓ ‘fall in pitch’.

Reinventing negation patterns in Moroccan Arabic 53

Figure 1. Geographical distribution of negation with marker NEG___bu

2. North African Arabic negation The North African Arabic negation system is basically composed of the morphosyntactic opposition between verbal negation and non-verbal negation. In the verbal negation structures, the incidence point generally corresponds to the verbal predicative syntagm (V + S) or a directly related element, whereas diverse nominal elements functioning as the syntactic nucleus are referred to in the non-verbal negation structures. Negation does not engage any morphological modification of the verbal stem of the concerned head. With the exception of Hassaniyya (South Morocco and Mauritania), which uses the morpheme ma ___, verbal negation is expressed in North African Arabic through the discontinuous marker ma ___ s− (s− i/s− ay) and its optional or conditioned variants.3 The same marker is used for nonverbal negation in prepositional predicative configurations and when occurring in a conjunctive expressive context, it mostly signifies adverseness; otherwise, non-verbal negation is rendered by means of the continuous marker mas− i ___ or mus− ___ (or their variants), whether evoking existential semantic values (equative, situative, possessive) or attributive ones. As regards the discontinuous marker ma ___ s− (s− i/−say), both elements are required to negate the verbal predication when no modal, discursive or

3.

A similar development is attested in the Arabic variety of the Egyptian Dakhla-Oasis (Woidich 1995–97).

54

Mena Lafkioui

expressive charge is involved (except Hassaniyya, example 3), as is shown in the next examples: (1)

ma

y-ākul

š

#awwa. NEG IMPERF-3MSG-eat NEG now ‘He does not eat now.’ (Tunisian Arabic; Lafkioui)

(2)

mā smä&-t-nī-š [NEG PERF-2MSG-hear+1SG+NEG] ‘Tu ne m’as pas entendu.’ (Algerian Arabic; Elhalimi 1996) [‘You did not hear me.’]

(3)

mā rg(d-t [NEG PERF-1SG-sleep] ‘Je n’ai pas dormi.’ (Hassaniyya; Taine-Cheikh 1995–96) [‘I did not sleep.’]

The first element ma is the head negator;4 it has a generalized and highly frequent usage in the Arabic spoken varieties, especially in comparison to Classical Arabic, where it is more confined to certain syntactic and semantic contexts. It cannot be omitted under any circumstances but can be substituted by its conditioned variant la.5 Besides its basic function of an isolated sentential negation particle,6 the latter is also employed for (1) iterative negation (la... la...), (2) constituent negation (e.g. la-bās lit. ‘no harm’ > ‘it is okay’), (3) predicative negation of coordinated structures with u-la (‘or not’) or wƽ-lla (‘or not’), (4) expletive negation with subordination, (5) negation of adverbial subordinate clauses denoting purpose and result, and (6) as a composite negative nominal marker (e.g. b-la lit. ‘with no’ > ‘without’). Most of all, though, it appears in semantically and pragmatically marked constructions (e.g. oath, interdiction, order, prohibition, admonition and categorical expressions), such as the next examples: 4.

5. 6.

This morpheme is not only employed for negation but also as an interrogative, a relative, an exclamative, and even an indefinite pronoun, which may point to a common origin (Pennacchietti 1967; Wehr 1953). This does not hold for some marginal cases like that of Maltese, where ___ š is used in constructions engaging modality (Vanhove 1994). Hassaniyya is an exception to this generalized practice because it draws on the particle bdê (*abadan ‘never’) in order to respond with ‘no’ (Taine-Cheikh 1995–96).

Reinventing negation patterns in Moroccan Arabic 55

(4)

allah-la-t-abb.(-k (Tunisian Arabic; Chaâbane 1996) [INVOC-NEG-IMPERF-3FSG-succeed+2SG] ‘J’implore Dieu pour que tu ne réussisses pas.’ [‘By God, I wish you will not succeed!’]

(5)

la tgīs hum! (Hassaniyya, Morocco; Iaaich 1996) [NEG IMPERF-2MSG-go 3PL] ‘Ne vas pas chez eux!’ [‘Do not go to them!’]

So, assertions – actions and states connoted by the realis mood – are negated by different morphemes that are mainly based on the element ma (examples 1 to 3). Other moods, on the other hand, such as the irealis are rather expressed through morphemes containing the element la, often in combination with 0 as the second constituent (examples 4 and 5) but also as la ___ š (and free variants).7 In the latter case, certain aspects of the expressed modal process or state are highlighted or elaborated (see also Harrell 1962: 153; Elhalimi 1996). However, modality can also be conveyed through the negation morpheme based on ma (or an allomorph), as in example (6).8 Apart from modality purposes, the second element of the discontinuous marker may be absent for reasons of discursiveness and heightened expressiveness (example 7), and may be replaced by other elements (i.e. adverbs, indefinite pronouns, relative markers, interrogative markers) that are considered to be semantically or pragmatically more adequate or more expressive (examples 8 and 9). In fact, the second element š, derived from šayʔ ‘thing’ of Classical Arabic, was originally an element of intensification (emphasis) – and still is to a different degree in various Arabic varieties – that has lost some of its marking strength with time and therefore had to appeal to other forms to complete this function. Its grammaticalization has reached different stages according to region, and its precise functional roles differ from one variety to another. Compared to the first element, it has a certain distributional flexibility; its position in the syntagm depends on the variable scope of the negation. This innovation phenomenon is absent from Classical Arabic but abundantly observed in the spoken varieties, 7.

8.

I have some reservations about Taine-Cheikh’s (2000) statement concerning the marginal frequency of la ___ 0 in North Africa, especially when it comes to its western part. Interestingly, Maltese Arabic displays the continuous variant ___ š in modal contexts.

56

Mena Lafkioui

mainly of North Africa (Marçais 1977; Taine-Cheikh 1995–96), where the data show a remarkable parallelism with the observed phenomena in the current Berber languages, which might have triggered this extended transformation process in North African Arabic (Brugnatelli 1987). Examples: (6)

w-allāhi hu mo rāgud (Libyan Arabic; Owens 1984: 162) [INVOC 3MSG NEG PTCP-MSG-sleep] ‘I swear he is not asleep!’

(7)

l-xdem ma ta-i-tts3ra-u (Moroccan Arabic; Youssi 1992: 109) [TOP NEG PART-IMPERF-3MPL-buy] ‘Les esclaves ça ne s’achète pas.’ [‘Slaves, they are not for buying.’]

(8)

ma sm(&-na .add (Algerian Arabic; Lafkioui) [NEG PERF-1PL-hear anybody] ‘We did not hear anybody.’

(9)

ma fh(m-ti .ta z(fta. (Moroccan Ar.; Lafkioui) [NEG PERF-2SG-understand even a piece of tar] ‘You did not understand a thing.’

The second element s− is principally used for basic negation and thus serves grammatical purposes, whereas its variants s− i, s− ay and s− ayn also play the role of negation intensifier, indicating modality and expressiveness on the enunciative (utterance) level as well as the discursive level (example 10). The element s− ayn may also behave like an indefinite morpheme signifying a focused ‘nothing’ (Vanhove 1994). For example: (10) la! āna ma nā&(s šāy! (Moroccan Arabic; Caubet 1996) [NEG 1SG NEG PTCP-MSG NEG] ‘Mais non! Je ne dors pas du tout!’ [No! I am not sleeping at all!]

Reinventing negation patterns in Moroccan Arabic 57

The discontinuous marker ma ___ s− (s− i/s− ay) is also attested in non-verbal predication structures, mainly when these refer to existential values: (11) ma-fi-ha-š taȥb (Algerian Arabic; Boucherit 2002: 64) [NEG-in+3FSG+NEG effort] ‘Il n’y a pas de peine.’ [‘It is no big deal.’] (12) ma &and-u š kalma. (Tunisian Arabic; Lafkioui) [NEG at+3MSG NEG word] ‘He does not keep his word.’ Two major exceptions to this rule are observed. The first is Moroccan, where this morpheme is applied to attributive constructions based on an adjectival predicate, as in the following examples: (13) ma zwīna š (,) had ---(šma.(Moroccan Arabic; Lafkioui) [NEG pretty NEG PROS PROX DEF-ornament] ‘It is not pretty, this ornament.’ (14) la! xu-ya ma qbi.-š (Moroccan Arabic; Adila 1996) [NEG brother+1SG NEG mean+NEG] ‘Non! mon frère n’est pas méchant.’ [‘No! My brother is not mean.’] Discontinuous negation is used here for the purpose of expressiveness, especially when it indicates an adversative meaning and hence a contrastive positioning in the conversational exchange; otherwise, the continuous marker mas− i ___ would be employed (see below in this section).9 The second exception is detected in Libyan Arabic, which allows for the negation of attributive predication by means of the discontinuous marker. It does so, though, on the condition that the predicate is composed of one single term (even a substantive term):

9.

In fact, diverse nominals and nominal groups may be negated by the discontinuous marker when they form a kind of ‘negative echo’ to the preceding utterance (Marçais 1977: 278).

58

Mena Lafkioui

(15) āna mo #ālib š (Libyan Arabic; Owens 1984: 157) [1SG NEG student NEG] ‘I am not a student.’ Furthermore, the morpheme ma ___ s− (s− i/s− ay) can be combined with personal pronouns, as affixes or as independent forms, in order to negate all sorts of nominals. While in Tunisian and Maltese mixed structures with affixed and independent pronouns are attested (example 16), in Algerian Arabic, only the amalgam with affixes occurs (example 17), and in Moroccan Arabic, solely the independent pronoun variant is observed (example 18). Examples: (16) m-īni( š (Maltese Arabic; Vanhove 1994) [NEG-1SG NEG] ‘I am not...’ (17) mā-hī-š m-ē9a (Algerian Arabic; Elhalimi 1996) [NEG-3FSG+NEG ill] ‘Elle n’est pas malade.’ [‘She is not ill.’] (18) ma-.iyya-š .amqa baš t-&āwƽd-ha.(Moroccan Ar.; Lafkioui) [NEG-3FSG-NEG stupid SUB IMPERF-3FSG-repeat+3FSG] ‘She is not so stupid to repeat it.’ It should be noted that in Libyan Arabic a distinction is made between the male and female for the plural forms: man-kam š (2MPL) ~ man-kan š (2FPL) and man-hum š (3MPL) ~ man-hun š (3FPL). As to the continuous marker mas− i ___ (Moroccan, Algerian) or mus− ___ ( ‘never’), which is always followed by a personal affix that fully agrees with the subject of the predication (examples 37, 38, 39, and 40). The negation marker ma succeeds the latter locution (example 37), precedes it (example 38) or even occupies both syntagmatic positions (example 39); in the last option, more semantic definiteness and energy is added to the negated message. In cases like (37) and (39) – and even (38) with a proper intonation – the syntagm [&ammə-- + personal affix] is inevitably subject to topicalization, which is mainly marked by specific intonation features, such as an intonation rupture and pitch dominance.13 Examples: (36) wəCCah

fi-h! in+3MSG ‘I swear I did not talk about him!’ INVOC

ma

h9ə--t

NEG

PERF-1SG-talk

(37) &ammə--ha ma mšā-t l-.əmmām! never+3FSG NEG PERF-3FSG-go DEF-hammam ‘Never, she went to the hammam!’ (38) ma

&ammə--ha mšā-t l-.əmmām! never+3FSG PERF-3FSG-go DEF-hammam ‘She never went to the hammam!’

NEG

(39) ma

&ammə--ha ma mšā-t l-.əmmām! never+3FSG NEG PERF-3FSG-go DEF-hammam ‘She never ever went to the hammam!’

NEG

Even though Moroccan Arabic displays a slight combinatory preference for the Perfective in negative oaths with ma ___ 0, constructions with the Imperfective (without TAM marker) are also attested; example (36) would be ‘wəCCah ma n(-h9ə- fi-h!’ (‘I swear I will not talk about him!’). It should also be mentioned that Moroccan Arabic makes it possible to denote the 13. These findings are based on my project in progress on “Intonation and Topicalization” (Lafkioui 2002, 2010, 2011c).

Reinventing negation patterns in Moroccan Arabic 65

basic meaning of examples (37) to (39) without inserting any negation particle, albeit in a more subsidiary way. This is illustrated in the following example: (40) &ammə--ha mšā-t l-.əmmām! never+3FSG PERF-3FSG-go DEF-hammam ‘She never went to the hammam!’ A diachronic change scenario that could account for this phenomenon is the following: the constituent &ammə--, originally a noun grammaticalized into an adverb of frequency, is grammaticalized into a negation adverb. This process is favoured by its topicalized form, which is prosodically highly marked and so produces its detachment from the rest of the utterance.14 Put differently, this means that the topicalization of a specific adverbial use could have sustained the development of a new pre-head negation marker (&ammə--) in contexts where the initial negators (ma + ma ___ 0 and ma ___ 0) have vanished for reasons of semantic weakness. Consequently, as &ammə-- combines with ma ___ 0, it might have been reanalyzed as a dislocated forclusive before its grammaticalization as a negation adverb. Another topicalized adverbial determinant that frequently occurs in negated structures with ma ___ 0 is gā& (‘all’, ‘at all’), used to signify the meaning ‘not at all’, like for instance in example (41): (41) gā& ma bqa təm. all NEG PERF-3MSG-stay there ‘He did not stay there at all.’ The same marker recurs in negation constructions containing the verb &-af (‘to know’) as a predicate if it is an Imperfective without preverbal TAM markers:15

14. I am grateful to Vermondo Brugnatelli, who drew my attention to the Italian adverb affatto (‘absolutely’, ‘utterly’) that has recently undergone a somewhat similar development in spoken Italian, viz. the negation adverb meaning ‘no’ or ‘not at all’ generally used as a negative response in configurations without a negation marker. 15. Mostly for the 1S; otherwise, the presence of the second part of the discontinuous negator is preferred.

66

Mena Lafkioui

(42) ma

māl-hum. what about+ 3PL ‘I do not know what is wrong with them.’

NEG

n-ə&-af

IMPERF-1SG-know

Another combinatorial context in which this negator is regularly detected is [ma + VS + ^ + O (= SUBcompl)]. For example: (43) ma

dər-ti &la-aš t-ərža&. PERF-2SG-do on what IMPERF-2MSG-go back ‘You did not do anything to fall back on.’

NEG

Besides its use in verbal constructions with a completive subordinate clause as the direct complement (example 43), this marker is also brought into play in verbal negation structures with an indefinite direct complement, as in the next example: (44) ma

dər-ti wlād? NEG PERF-2SG-do children ‘You haven’t made any children?’ = ‘You haven’t got any children?’

Interestingly, this predicative structure is compatible with the second element š (or allomorphs) if there is a contrastive conjunction in which the second part is explicitly uttered or just presupposed. For example, if utterance (44) contained the element š, a conjunctive part such as ‘dər-ti wālu’ (‘you did nothing’) would implicitly be present. Furthermore, this morpheme adds some referentiality to the dereferentialized nature of this negation type that is supported by the undetermined direct object; (44) would thus mean ‘You haven’t got any children?’ with a focus on ‘any’. As regards the discontinuous marker la ___ 0, it is employed in the context of negative prohibitions, injunctions and oaths; the former may also be expressed through the markers la ___ š and ma ___ š, like in example (46). Yet, prohibitions with 0 as the second negation element, such as the one presented in example (45), are more energetic. An appropriate exclamative intonation is of course a conditio sine qua non for the syntactic adequacy

Reinventing negation patterns in Moroccan Arabic 67

of this utterance, as well as for its semantic and pragmatic function (prohibition). In addition, an asymmetrical relationship is observed in Moroccan Arabic – as well as in Arabic in general – between the negative prohibition structure, based on the Imperfective, and its matching positive configuration, i.e. that of the Imperative, which draws on a different conjugation paradigm (only 2S and 2P) and employs dissimilar personal affixes. Examples: (45) la

t-əh-əb! IMP-2MSG-run

NEG

away

‘Do not run away!’ (46) la NEG

t-əh-əb IMP-2MSG-run

š! (= ma t-əh-əb š!) away NEG

‘Do not run away!’ (47) wəCCah INVOC

la

klīti-h!

NEG

PERF-2SG-eat+3MSG

‘I swear you will not eat (of it)!’ Concerning the negation of oaths and wishes (example 47), like most North African Arabic varieties,16 Moroccan Arabic has conserved the flip-flop phenomenon detected in Classical Arabic. This phenomenon consists of using the Perfective – usually indicating the idea of completion – in contexts situated or projected in the future. Accordingly, this type of negation conveys the inexistence of a process or a state positioned in the future.17

3.1.3. Verbal negation patterns As shown in the data examined in the previous sections and some of the examples that I replicate here, verbal negation follows a pattern in which the discontinuous negation markers generally encircle the predicative syntagm or one of its components.

16. Yet, this feature is often absent in other Arabic speaking parts of the world. 17. About this phenomenon in North African Arabic, see also (among others) Aguadé and Elyaacoubi (1995: 148) and Taine-Cheikh (2000).

68

Mena Lafkioui

Examples: (21) ma NEG

kla

š.

PERF-3MSG-eat

NEG

‘He has not eaten.’ (22) ma

ka y-ākul š. NEG PART-IMPERF-3MSG-eat NEG ‘He does not eat.’

(41) gā& ma bqa təm. all NEG PERF-3MSG-stay there ‘He did not stay there at all.’ (45) la NEG

t-əh-əb! IMP-2MSG-run

away

‘Do not run away!’ The first element always precedes the predicative syntagm, even if the second element is 0 (examples 41 and 45). General verbal negation pattern: (COMPL+) NEG + VPS + NEG (+ COMPL) However, it should be specified that when the VPS contains a verbal auxiliary, the negation markers may merely enclose the latter instead of the whole predicative syntagm, depending on the negation scope (example 24 repeated below, but also 28 and 29 in Section 3.1.1.). The fact that the auxiliary of these examples is fully conjugated in agreement with the head verb underpins its predicative function. In contrast, when the auxiliary is a grammaticalized form (invariable unit or group of units), usually playing the role of an adverbial determinant that occupies the initial position of the sentence, the negation proceeds according to the general pattern, as in examples (33) and (35).

Reinventing negation patterns in Moroccan Arabic 69

Examples: (24) ma

bqa

š

NEG

AUX

NEG

(ka) y-ākul. (PART)-IMPERF-3MSG-eat ‘He does not eat anymore (like he used to).’

(33) bāqi

ma

sālī-ti AUX NEG PERF-2SG-finish ‘Haven’t you finished yet?’

(35) -ā-ni PART+1SG

š? NEG

mazāl

ma

bdī-t

š.

AUX

NEG

PERF-1SG-start

NEG

‘I have not started yet.’ So, the second element of the negation marker sets the limits of the negation scope and hence determines specific aspects of the actualized or projected process or state, which may be highlighted by means of prosodic devices (intonation and accent). With respect to the precise negation scope, the marker š may follow the indirect affix, as in (48), but may also precede it when the direct object is emphasized: (48) ma

gāl-u-ha PERF-3PL-tell+3FSG ‘They did not tell him that.’

NEG

l-u š to+3MSG NEG

Alternative verbal negation pattern I: (COMPL+) NEG + VPS + DIR COMPL (= AFF1) + INDIR COMPL (= AFF2) + NEG Alternative verbal negation pattern II: (COMPL+) NEG + VPS + DIR COMPL (= AFF1) + NEG + INDIR COMPL (= AFF2)

3.2. Non-verbal negation Non-verbal predication in Moroccan Arabic is negated by means of both continuous and discontinuous markers. The continuous markers, on the one hand, are principally used in descriptive contexts, that is, to identify or to

70

Mena Lafkioui

distinguish and to localize the discursive objects described by means of the nominal elements that constitute the utterance. The discontinuous markers, on the other hand, serve more expressive goals, which often relate to modality.

3.2.1. The continuous negation marker Non-verbal negation in Moroccan Arabic is generally expressed through the continuous marker mas− i ___, as is revealed in the following examples: (49) maši

d-drāri l-kƽddāba. DEF-children DEF-liars ‘The children are not the liars.’

NEG

(50) maši

HHɀa-h. NEG father+3MSG ‘It is not his father.’

(51) l-wərqa (,) maši rqīqa DEF-pastry dough PROS NEG thin ‘The pastry dough, it is not very thin.’

bəzzāf. very

The nominal predicate of these utterances points to a referential identification (examples 49 and 50) or a qualitative attribution (example 51) of the discourse object. The predication in example (52), on the other hand, indicates its localization in the inter-subjective space: (52) maši

hna. here ‘It is not here.’

NEG

The marker mas− i ___ thus precedes the syntagmatic constituent to which the negation relates, which can be the subject of the predication (example 49), its predicate (examples 50 and 51), a complement, or even the whole predicative configuration (example 62, see infra). In case of an identification relationship between the predicative constituents (examples 49 and 50), the negator maši also functions as a predicative auxiliary (i.e. negative copula), a role that is played by an independ-

Reinventing negation patterns in Moroccan Arabic 71

ent personal pronoun in the corresponding positive nominal configuration (example 53). The pronoun in question is restricted to the third person; i.e. huw(w)a (3MSG), hiy(y)a (3FSG) and huma (3PL) in agreement with [S-P] or [P]. Examples: (53) d-drāri huma l-kƽddāba. DEF-children 3PL DEF-liars ‘The children are the liars.’ (54) *d-drāri l-kƽddāba DEF-children DEF-liars ‘the lying children’ The personal pronoun huma, which separates the subject (the first term) from the predicate (the second term) in the positive structure (example 53) may appear in the negation structure, not as a predicative auxiliary but as a deictic emphasizing certain aspects of the referential process; in the present case, the subject is emphasized:18 (55) maši

d-drāri huma l-kƽddāba. DEF-children 3PL DEF-liars ‘It is not the children who are the liars.’

NEG

As regards attributive negation, the predicate can be focalized by inverting the syntagmatic order and hence by positioning the subject after the predicate. These are then both separated by an intonation rupture, as for instance in the cleft structure in (56), based on example (51): (56) maši

rqīqa bəzzāf (,) l-wərqa. NEG thin very PROS DEF-pastry dough ‘It is not very thin, the pastry dough.’

The intonation features are highly demarcative in the case of a focalized negative configuration – the intonation break in (56) is much more distinctive than in (51) – as well as in the case of a basic positive configuration

18. This implies a corresponding focalization.

72

Mena Lafkioui

(example 57), where these are essential to distinguish the subject from the predicate: (57) l-wərqa (,) rqīqa bəzzāf. DEF-pastry dough PROS thin very ‘The pastry dough, it is very thin.’ Furthermore, in adversative contexts, the negation of an attributive predication may proceed through the discontinuous marker (see also Section 3.2.2.). For instance, if the interlocutor contests the quality attributed to the pastry dough described in example (57), he or she would utter one of the following options: (58) lwərqa (,) ma rqīqa DEF-pastry dough PROS NEG thin ‘The pastry dough, it is not very thin.’

š NEG

bəzzāf. very

(59) ma

rqīqa š bəzzāf (,) lwərqa thin NEG very PROS DEF-pastry dough ‘It is not very thin, the pastry dough.’

NEG

Regarding the negation of locative predications, Moroccan Arabic makes use of the continuous marker maši if its components are determined (example 52 supra and 60 infra). Otherwise, existential markers such as the invariable marker kayn (‘there is’) and the variable kān (‘there was’) and kaykūn (‘there is’) are employed, as in example (61): (60) l-.līb maši f t-t(llāža DEF-milk NEG in DEF-fridge ‘The milk is not in the fridge.’ (61) ma

.līb f t-t(llāža milk in DEF-fridge ‘There is no milk in the fridge.’

NEG

kayn

š

COP

NEG

Besides predicative negation, which has a particular term of the predication as the incidence point (usually the predicate), there are also numerous cases of propositional negation (sentential negation) – with the whole proposition, behaving like a nominalized syntagm, as the incidence point.

Reinventing negation patterns in Moroccan Arabic 73

For example: (62) maši

gāl-ha l-u. PERF-3MSG+3FSG to+3MSG ‘It is not that he told him that.’

NEG

It is the asserted proposition gāl-ha l-u (‘he told him that’) that is denied in this utterance and not its verbal head gāl (‘he told’), which would generally be negated by means of a discontinuous marker if it were the predicate: (63) ma

gāl-ha š l-u. NEG PERF-3MSG+3FSG NEG to+3MSG ‘He did not tell him that.’

3.2.2. Discontinuous negation markers Adjectival predicates are negated by means of the continuous marker mas− i ___ (example 51 in Section 3.2.1.) as well as by the discontinuous marker ma ___ š (or its more expressive variants ma ___ ši and ma ___ šay), as presented previously (examples 13 and 14 in Section 2) and also in the next non-verbal utterance: (64) ma

rqīq š. thin NEG ‘He is not thin.’

NEG

It is the quality of ‘being thin’ that is negated in this example; this feature is asserted as being invalid with respect to the referential object, that is, a male person about whom the interlocutors are talking. This case of predicative negation could therefore be glossed as follows: ‘The quality of being thin is not attributed to the person about whom we are speaking’. But when the same adjective is negated by the continuous marker, it signifies ‘It is not that he is thin’: (65) maši

rqīq. thin ‘It is not that he is thin.’

NEG

74

Mena Lafkioui

Discontinuous negation of the type presented in (64) also entails the idea of disagreement, as it were the first part of a bipartite opposing structure, of which the second part is implied or explicitly uttered, like in the following example: (66) ma

rqīq š , ġlī9. thin NEG PROS big ‘He is not thin; he is (rather) big.’

NEG

Predication based on prepositional constructions uses the discontinuous marker ma ___ š for negation (examples 67 and 68). This marker surrounds the predicative syntagm, composed by the preposition and its personal affix, which is sufficient for a valid assertion when supported by the referential context (example 68). This example may be used, for instance, as a negative response (initiating with la... ‘No.’) to the question Kū-k, &and-u #umubīCa? (‘Your brother, does he have a car?’). However, this utterance lacks the object complement (or its topicalized variant), which is essential to the grammaticality of this type of non-verbal structure (example 67): NVPS (=PREP + AFF) + Oobligatory. (67) ma

&and-u š at+3MSG NEG ‘He hasn’t got a car.’

NEG

#umubīCa. car

(68) ma

&and-u š. at+3MSG NEG ‘He hasn’t got (it).’

NEG

In addition, the syntagmatic structure of example (68) is ambiguous: on the one hand, it represents the syntactic core of a possessive assertion with an elliptic object complement ([P + S] order). This is for example the case when it refers to (67). On the other hand, it also stands for a locative assertion in which the prepositional group functions as the predicative syntagm with an elliptic subject ([S + P] order). If there is a complete structure, it could look as follows: (69) #-#umubīCa (,) ma &and-u š. DEF-car PROS NEG at+3MSG NEG ‘The car, it is not at his (place).’

Reinventing negation patterns in Moroccan Arabic 75

However, in case of a locative structure, continuous negation with mas− i ___ is largely preferred, which is the principal negator of non-verbal structures, especially when the first term is explicit and hence determined. Conversely, with a possessive structure, the discontinuous marker ma ___ š, typical of verbal negation, is favoured. This evidence corroborates the verbalization hypothesis of Cohen (1984), which is also supported by the fact that possessive non-verbal predication displays a word order that is distinctive for verbal predication in Arabic ([P + S]), in contrast with the common non-verbal word order ([S + P]; for example, locative predication). Thus, Cohen’s (1984: 585) idea that nominal structures may evolve to “quasi-verbal” structures in some Semitic languages and hence exhibit syntactic features characteristic of verbal structures also applies to Moroccan Arabic and Hassaniyya (Taine-Cheikh 1995–96). Interestingly, this phenomenon is also widely attested in Berber (Lafkioui 1999, 2011a: Chapter 1), which offers an interesting field for comparative study that could extend to the Afro-Asiatic level. Moreover, the morpheme ma ___ s− (or variants) is compatible with independent personal pronouns in order to negate various nominals and nominal groups, such as in example (18) (see Section 2). The nominal head may also contain the particle -ā, which does not only place the expressed notion (process or state) in the Present Continuous (aspectual-temporal particle), but above all refers to the inter-subjective space between the interlocutors and their respective place and discursive commitment.19 Here again, Berber displays a remarkable similarity, as it provides accounts of specific but similar particles that indicate how the interlocutors are involved in their interactions (Lafkioui 1999, 2011a: Chapter 1). It should also be noted that diverse nominals and nominal combinations are negated by means of this discontinuous morpheme when they constitute a negative response that repeats the core content of the directly preceding question: (70) fīn

huwa ? Kār(ž ? 3MSG PROS PTCP-MSG-go out PROS ‘Where is he? Out?’ INTRG

19. For more about this particle, see Taine-Cheikh’s chapter in this volume. Another interesting contribution regarding -ā in Moroccan Arabic is Caubet (1992).

76

Mena Lafkioui

(71) la NEG

,

ma

PROS NEG

Kār(ž š. PTCP-MSG-go out NEG

‘No, he is not out.’ Like verbal negation (examples 44 and 43, respectively, of Section 3.1.2.), nominal constructions with an indefinite noun (example 72) or a completive subordinate clause (example 73) as the object complement require the discontinuous marker ma ___ 0 for negation.20 Examples: (72) ma

m&a-kum h(9-a. with+2PL talk ‘There is no sense in speaking with you.’

NEG

(73) ma

&and-na fīn n-ƽmši-w. NEG at+1PL where IMPERF-1PL-go ‘We haven’t got anywhere to go.’

But, as mentioned before for verbal configurations, if the interlocutor wishes to focus on a particular constituent of the predication by adding some referentiality to the expressed content, he inserts the element š (or a variant) just at the end of the negation scope; in examples (72) and (73), this insertion could only occur after the prepositional syntagms.

3.2.3. Non-verbal negation patterns Non-verbal negation distinguishes between two distributional patterns: the first one concerns continuous negation and assigns a pre-predicative position to the single negator maši, which is the principal marker of non-verbal negation. The second one is discontinuous and corresponds to the general morphemic order detected for verbal negation, that is, the first element ma precedes the syntactic head while the second element š follows the final constituent of the negation scope.

20. This is another point that supports the idea that certain nominal structures have been subject to a verbalization dynamic.

Reinventing negation patterns in Moroccan Arabic 77

Non-verbal continuous negation pattern: NEG + NVPS For example: (74) huwa maši HHɀa-h. 3MSG NEG father+3MSG ‘He is not his father.’ The basic word order of predicative negation built on maši is [S + NEG + P], when S is uttered in sentences with an identification (example 74), attributive or locative value. The S term of locative predication should be determined. If not, a copula is employed, which brings discontinuous negation into play. Discontinuous negation is also used in attributive and locative constructions for adversative purposes. Focalization of P is established by inverting this basic syntagmatic order and marking the focus by prosodic means (e.g. intonation break) that at the same time separate the subsequent S from P (example 75 here below and also example 56 in Section 3.2.1.): [NEG + P + PROS + S]. For example: (75) maši

HHɀa-h (,) father+3MSG PROS ‘It is not his father.’

NEG

huwa 3MSG

Propositional negation follows the general continuous negation pattern, i.e. [NEG + PROP]. As regards discontinuous negation, the marker ma ___ š is detected in specific contexts when applied to non-verbal predication. Apart from contexts indexing referential and adversative(-conjunctive) meanings, it mainly appears with adjectival predicates (example 64 of Section 3.2.2.) and prepositional predicative syntagms (examples 67, 68, and 69 of Section 3.2.2.). The basic word order of this negation type is [S + NEG + P + NEG]. Only predication expressing the idea of possession has a different (inverse) order, namely [NEG + P + S + NEG]. Moreover, when O is represented by an indefinite nominal or by a completive subordination, non-verbal negation is marked by means of ma ___ 0, if no referential connection is involved.

78

Mena Lafkioui

Non-verbal discontinuous negation pattern: NEG + NVPS + NEG (+ COMPL)

4. Innovations in Moroccan Arabic Arabic negation The mutual influence between Arabic and Berber in North Africa has been attested in diverse linguistic domains (i.e. phonetics, morphology, syntax and lexicon). It varies with respect to different parameters, of which the geographical, historical and social ones are fundamental, in addition to the system-based parameters. Negation is one of the major research fields that is subject to contact-induced change in North Africa. An important aspect of this field is the emergence of discontinuous negation markers in North African Arabic, an innovation that is largely generalized over North Africa’s Arabic speaking zones and that could have been created by contact with Berber – where these markers are common and highly developed in many languages. However, the influence might also have occurred in the other direction; that is, Berber might have borrowed post-head negation marking from Arabic, as this feature is also attested outside Berber speaking areas, mainly in some Egyptian and Levantine varieties.21 Even so, the case investigated in this chapter, Moroccan Arabic of Oujda, clearly shows specific borrowing phenomena from Berber into Arabic, as will be demonstrated in this section and in the subsequent one. But before going into detail, a presentation of the Berber negation system is needed in order to better understand the contact phenomena explained. Like North African Arabic, the Berber negation system basically draws on a distinction between verbal negation, mainly marked by discontinuous morphemes, and non-verbal negation, characterized by continuous marking procedures. The principal distinctive feature between these two negation systems is that most Berber languages also mark negation by means of verbal stem alternation, in which the Negative Perfective is more prevalent than the Negative Imperfective. Furthermore, the presence of preverbal negators generally leads to a syntagmatic position change of the postverbal affixes (fronting); they precede the verbal head (but follow the negator), 21. The hypothesis that discontinuous negation in Berber has been borrowed from Arabic (Lucas 2007) is interesting but unconvincing for different reasons, the main reason being that the Berber data on which the analysis is based are too scarce on a comparative level and therefore inconclusive as evidence.

Reinventing negation patterns in Moroccan Arabic 79

usually without modifying their respective order, namely [indirect affix + direct affix]. The predominant variant of the first element of the discontinuous negator is presumably of Berber origin and is derived from *w(r, a grammaticalized verbal form composed of the negation element *w or *u and the verbal root *r (expressing modality): *w(r = NEG = [NEG *w or NEG *u + V *r].22 While this part of the negator is compulsory in most languages – with the exception of some cases where only the postverbal element is used, like in western Tarifit (Lafkioui 2007: 234–236)23 –, the second part that usually follows the syntactic head may be optional – as an intensifier – or required, depending on the negation context. The discontinuous marker is also employed in non-verbal predication, even in a more extended way than in North African Arabic, usually in contexts indexing existential values. However, the Berber languages make use of continuous markers too, which are based on Berber elements as well as on Arabic elements.24

4.1. A new discontinuous negator The Moroccan Arabic variety of Oujda (MAO) provides abundant evidence of the existence of a new discontinuous negation marker which was created under contact influence from Berber, i.e. the hybrid morpheme ma ___ bu. Its second constituent bu is borrowed from Tarifit (Rif Berber, North Morocco; see Figure 1 in Section 1), which is the only Berber language in which this morpheme occurs (Lafkioui 1996, 1999: II/Chapter 2, 2007: 234–236, 2011a: 62–69).25

22. Different hypotheses have been proposed regarding the etymological origin of the two components of the negation operator in Berber, for example Galand (1995), Chaker (1996) and Brugnatelli (2011). 23. It is, however, in free variation with preverbal negation. For other cases with enclitic negation, see Brugnatelli (1987). 24. Grammaticalized amalgams are frequent in Berber. 25. Concerning the origin of this marker, no conclusive explanation is available. However, two options may be envisaged: the first one relates to the Tuareg Berber verb iba (and variants) signifying meanings such as ‘there is no’, ‘cessation of’ and ‘lack of’, of which the derived form ăba is used in optative constructions (Prasse et al. 2003: 2). Texts from the Ayer region (Niger) prove the existence of this verb as a negation adverb, i.e. ebƽw ‘no, really!’ (Petites sœurs de Jésus 1974: 65). The second option has to do with the Arabic nomi-

80

Mena Lafkioui

Indeed, this element (or its spirantized variant Nu) is part of all sorts of frequently used negators in Tarifit, such as u ___ bu, ur ___ bu, wa ___ bu and war ___ bu, which perfectly correspond to the innovated hybrid morpheme of MAO on a structural as well as a functional level. Examples: (76) u

awar K-as. word on+3SG ‘They did not say a word about him/her.’

NEG

nna-n

Nu

PERF-3MPL NEG

(77) u

ġā-s Nu Oam##uO. at+3SG NEG women ‘He has not got a wife.’

NEG

These Berber utterances demonstrate verbal predication (example 76) and non-verbal predication (example 77), both negated by means of the morpheme u ___ bu. The second element is compulsory and is necessarily followed by a noun (in its free state) functioning as an object complement. Data from MAO do not only reveal the presence of the innovated morpheme containing exactly the same element bu, but also prove that it shares even the structural and functional features with Tarifit (see also Section 4.2.). Examples: (78) ma

šra-w bu l-.awli had l-&ām. NEG PERF-3PL NEG DEF-sheep PROX DEF-year ‘They did not buy a sheep this year.’

(79) ma

&and-na bu 9-9ƽw f d-dāl-qdīma. at+1PL NEG DEF-electricity in DEF-house DEF-old ‘We do not have electricity at the old house.’

NEG

As in Tarifit, the post-head element bu is associated with a determined subsequent object, regularly marked by the definite article in MAO. The neganal modifier and determinant bu (e.g. bu -ās ‘him with the head’ > ‘big headed one’) that occurs in both Arabic and Berber data.

Reinventing negation patterns in Moroccan Arabic 81

tor ma ___ bu is used for basic negation in descriptive contexts. Compared to its major functional contender ma ___ š, this marker is much more restricted when it comes to its combinatorial options, its syntagmatic distribution and its functional roles. Since it needs to be combined with a determined object, its structural potential is limited to certain verbal configurations and to prepositional predicative configurations or some other quasi-verbal structures. These limitations are reflected in the distributional order, as the discontinuous marker is required to surround the verbal or non-verbal predicative head. Therefore, its second element does not function as a tool that demarcates the negation scope and may occupy different positions in the syntagm. These structural restrictions are related to functional roles that are of a descriptive nature, supported by highly referentialized notions conveying existential values. Expressive negation with ma ___ bu only occurs when an appropriate intonation backs up the whole structure. Other examples confirming the structural and functional correspondence between the negation system of MAO and of Tarifit with respect to this contact-induced phenomenon are these Berber equivalents of example (78) and (79):26 (80) u

ssġi-n Nu izmər asəggɀas-a. PERF-3MPL NEG sheep year+PROX ‘They did not buy a sheep this year.’

NEG

(81) u

ġā-nəġ Nu OfawO gi OaddāO at+1PL NEG electricity in house ‘We do not have electricity at the old house.’

NEG

OaqPinO. old

4.2. Innovated negation patterns The adoption of the new negation marker ma ___ bu in MAO had some interesting implications on its negation system. It triggered a restructuring of its existing morphosyntactic patterns by introducing a new general pattern:

26. Besides examples (76) and (77) and numerous records from Lafkioui (1996, 1999: II/Chapter 2, 2007: 234–236, 2011a: 62–69).

82

Mena Lafkioui

New general negation pattern: ma + PS + bu + O (DET) This innovated pattern is characterized by its fixed morphosyntactic order in which the determined post-head noun as an object complement is a prerequisite. It enfolds two sub-patterns, one applying to verbal predication (examples 78, 82, and 83) and another one related to non-verbal predication (examples 79, 84, and 85): New verbal negation pattern: ma + VPS + bu + O (DET) (82) ma

lqā-t

l-.əll li-ha. NEG PERF-3FSG NEG DEF-solution for+3FSG ‘She did not find the solution for her (problem).’ = ‘She did not find a way to deal with her.’

(83) ma

ġadi i-rža&

NEG

bu

bu

l-flūs

AUX IMPERF-3MSG-return NEG DEF-money

l mulā-h. to owner+3MSG

‘He won’t give back the money to its owner.’ New non-verbal negation pattern: ma + NVPS + bu + O (DET) (84) ma

fi-ya bu ž-žəhd baš in+1SG NEG DEF-strenght to ‘I do not have the energy to stand up.’

NEG

(85) ma

&and-ƽk bu lakart kima at+2SG NEG (DEF-)carte like ‘Don’t you have a card like this one?’

NEG

n-nū9. IMPERF-1SG-stand

up

had-i? PROX+FSG

In both predication types, this new negator can be substituted by the discontinuous marker ma ___ −s without altering their distributional properties. Even on the semantic level, no distinctive values are observed if no prosodic markers are involved. However, as a basic negator, ma ___ bu is in complementary distribution with ma ___ −s (and variants) in all contexts where the latter is incompatible with directly ensuing objects. Functionally speaking, this means that, compared to the centring latitude of the discon-

Reinventing negation patterns in Moroccan Arabic 83

tinuous marker ma ___ −s, and more particularly its second constituent, ma ___ bu is restrained to descriptive negation in which a precise referential object is asserted to be nonexistent. Moreover, since the new negator is strictly reserved for configurations with a determined (mostly definite) object, it has a particular paired but opposite relationship with ma ___ 0 that is used in configurations with an undetermined object. On a functional level, this implies that ma ___ bu is specialized in highly referential negation, in contrast with [ma + HEAD + 0 + O (UNDET)], which is dedicated to the negation of dereferentialized notions. Consequently, the generalized adoption of this new marker has caused a restructuring of the negation system of MAO: it introduced a specialized negation domain characterized by its high referentiality, which is morphosyntactically represented by structures containing determined objects. To a certain extent, this incorporation took away some of the functional domains of those discontinuous markers that are associated with variable referentiality degrees of the actualized or projected process or state. In other words, certain domains of the original system are reanalyzed as preferential for basic negation by means of the new operator; the same domains are reanalyzed as optional but not preferential with respect to the other discontinuous operators. ma ___ s− (and variants) ma + PS + š (+ O) variable referentiality: {^ < REF ≤ MAX}

ma ___ s− (and variants)

ma ___ bu

ma + PS + š (+ O)

ma + PS + bu + O (DET)

variable referentiality: {^ < REF < MAX}

fixed referentiality: {REF = MAX}

Figure 2. Reinvented negation system of MAO

84

Mena Lafkioui

4.3. Who borrowed from whom? The city of Oujda and its surroundings, originally Berber speaking (mainly Zenati language group), is of economic and political importance because of its strategic location on the border between East Morocco and West Algeria (see Figure 1, Section 1). Its role as an ancient and intensely active trading passage towards all directions is mostly responsible for its linguistic and cultural diversity. Yet, this diversity is still characterized by a considerable presence of Berber speaking people. In the last three decades, they have become even more prevalent due to, inter alia, students coming from Berber speaking areas outside Oujda (e.g. Berkane, Nador, Alhoceima, Figuig) who want to study at the relatively new University and often even settle there afterwards. Until now, the negation with ma ___ bu has only been attested, at least in a systematic way, in the region of Oujda (particularly in the city). At the same time, the diverse corresponding Berber variants are displayed in most of the regions of the large Rif area. Therefore, it is most likely that the borrowing direction occurred from Berber into Arabic (see Figure 1: dark grey zone). Especially since the users of this new negator also, and even mostly, speak Arabic and have no command of any Berber language. Furthermore, Tarifit has provided regular diachronic accounts of this phenomenon for at least a century, whereas for MAO, the informants have not always been able to confirm its existence prior to their own language practices (e.g. that of their parents and grandparents), which are mainly limited to the first and second generation. Even though no studies have reported on this phenomenon for other parts of North Africa (not even for Algeria), it would be of great interest to extend the fieldwork area across the Moroccan-Algerian border in order to verify whether it exists there too. If it does occur in the Arabic speaking parts on a considerable scale and in a regular way, the borrowing direction could be from Arabic into Berber. A valid explanation could be, for instance, that ma ___ bu was adopted and then reanalyzed into Berber by Tarifit speaking people who immigrated to this region during the colonial period for economic reasons and who came back after the independence.27 Then again, Berber locutors could also have introduced this variant in the area through migration. Still, this is not supported by the actual data, which point to Berber as the borrowing source of the negation phenomenon at hand. These data also suggest that its adoption, acquisition 27. I have numerous oral narrative testimonies of this economic migration.

Reinventing negation patterns in Moroccan Arabic 85

and diffusion are chiefly motivated by system-internal properties (see Section 4.1. and 4.2.), since social factors such as prestige or community solidarity nearly do not apply to this case. Tarifit u ___ bu (and variants) NEG + PS + bu + O (DET)

MAO ma ___ s− (and variants) NEG + PS + NEG (+ O)

MAO ma ___ bu ma + PS + bu + O (DET)

Figure 3. Contact-induced negation patterns in MAO

5. Conclusion The addressed negation phenomena engendered by contact between Moroccan Arabic (MAO) and Berber (Tarifit) are examples of how language is constantly modulated in the form of innovations that may emerge in structurally layered and causal formations dictated by system-based properties. The several cases examined in this chapter show that even though innovations are induced by contact (i.e. extra-linguistic parameters), their diffusion and development can also be linguistically motivated (i.e. formal and functional parameters). Whether or not the new negation variants (form and structure) are widely distributed across the Moroccan Arabic varieties of the studied area is largely determined by system-internal factors such as structure adequacy and systematization of specific complementary combinatorial and distributional patterns connected with defined enunciative functions. Therefore, it is important to consider language change and diffusion not solely from a social perspective (“propagation” as in Croft 2000: 38, 178), but also in terms of how the variants are formally and functionally integrated into ever-changing linguistic structures (Lafkioui 2009, 2011b).

86

Mena Lafkioui

Another interesting finding is that Moroccan Arabic – and even North African Arabic in general28– proves that there is a significant relationship between negation, modality and expressiveness, which I would explain as follows: the full occurrence of the discontinuous negation marker in verbal configurations mainly relates to descriptive predication with a variable negation scope. This variability depends on the position of the postverbal negator, which at the same time adds a certain degree of referentiality ({MIN < REF < MAX}) to the selected discursive object. This is what I label as basic negation, unless expressive cognates of the postverbal intensifier š (or variants) are employed (e.g. ši, šay) for emphatic purposes. Conversely, marked negation chiefly engages the absence of the postverbal element (NEG ___ ^), a procedure through which all kinds of modal and expressive values, in addition to existential ones, are asserted. However, referentiality is reduced here, even to ^ in some cases (e.g. [V + S + O (= UNDET)]): {^ ≤ REF < MAX}. Pragmatically speaking, this implies that the absence of the postverbal negator allows for the fine-tuning of the discursive investment of the enunciators. A distinction between basic versus marked is also displayed with respect to the nature of the preverbal negator: while the item ma is generally related to descriptive and hence basic negation – especially when combined with the postverbal š – the item la is usually restricted to modal and expressive use (i.e. marked negation). Compared to verbal negation, the inverse of the opposition [basic negation versus marked negation] is observed in non-verbal configurations. In other words, discontinuous negation marking stands for modality and expressiveness, while continuous negation marking signifies descriptive meanings that are connected with semantic values indicating an identification of the referential object, its localization or its attribution to a defined semantic category. The only exception is that of quasi-verbal constructions in which the nominal head tends to behave like a verbal predicate and hence makes use of the discontinuous negator for basic negation. Cases such as the hybrid negator ma — bu in MAO, occurring in specific complementary morphosyntactic configurations referring to assertoric values, suggest that it is basic negation that is predominantly subject to change in the asymmetrical negation system of North African Arabic. Accordingly, as far as the linguistic level is concerned, the evolution dynamics of negation in this area are significantly motivated by functional needs regarding descriptiveness and referentiality.

28. Except Bedouin varieties such as Hassaniyya.

Reinventing negation patterns in Moroccan Arabic 87

In contrast with the negation system of Classical Arabic, North African Arabic has probably undergone the following transformation stages with respect to basic verbal negation: Transformation process A: mā ___ , lā ___ , lam ___ (stage I) ⇒ ma ___ 0 (stage II) ⇒ ma ___ −s(i) (stage III) Stage II is merely attested in some southern varieties such as Hassiniyya and Skura in Morocco. It is common, though, in the Arabic varieties spoken in sub-Saharan Africa (e.g. Sudan and Chad) and in the Middle-East. From this stage, a discontinuous marker was created (stage III), which currently functions as the principal negator in North Africa. It is beyond the scope of this chapter to explain the motivation behind these diachronic changes, in particular the progression from stage II to stage III. Nevertheless, it should be mentioned that Brugnatelli’s (1987, 2002, 2005) hypothesis about the Berber origin of this latter innovation in Arabic is appealing as a basis for further investigation, especially when it comes to the Arabic part. He draws on a compelling parallel between the comparative Berber data and the Arabic case of the Dakhla-Oasis (Woidich 1995–97). Basically, he suggests that Berber had a postverbal negator at a proto-Berber level, which provoked the emergence of negative aspectual forms (at practically a panBerber level) due to accent dislocation and vowel reduction. For reasons of redundancy, some of the Berber languages have lost this postverbal item. Those which conserved it, reanalyzed it as a negation marker similar to stage III in Arabic. They did so at a later time through contact with those Arabic varieties that had already undergone this transformation at the time in question. Given that Berber already had a discontinuous marker before its postverbal development, i.e. preverbal negator + negative aspectual marker, it is problematic to assume a borrowing from Arabic while ignoring this important fact (Lucas 2007). As a matter of fact, many Berber languages are characterized by a triple negation marking: preverbal negator (particle) + mesoverbal negator (apophonic aspectual form) + postverbal negator (particle). Consequently, it is reasonable to assume that Berber was a borrowing source for stage III in North African Arabic – even when this theory is based on a case of double negation marking in Berber (SchmittBrandt 1979) – or that both Berber and Arabic have been the subject of an independent evolution.

88

Mena Lafkioui

Additionally, the Arabic variety investigated in this study has developed a new negator that is in complementary distribution with the primary negator ma ___ −s, as described in the following diagram: Transformation process B: mā ___ , lā ___ , lam ___ (stage I) ⇒ ma ___ 0 (stage II) ⇒ ma ___ −s(i) (stage III) ⇒ [ma ___ −s(i) ~ ma ___ bu] (stage IV) The diachronic change in MAO from stage III to stage IV was triggered by a borrowing process from Berber, in which not only a formal property was (partially) borrowed but also its corresponding constructions (“replica grammaticalization”; Heine & Kuteva 2003: 539; see Section 4). At the basis of the diffusion of this phenomenon are system-based factors such as structural adequacy and generalization. An interesting point that should also be noted in this conclusion, but definitely calls for additional research, even on a general comparative and typological level, is the existence of negative structures with none of the usual negation markers (i.e. apparent zero negation marking). These structures comprise nominals in topical position in the phase of grammaticalization (by means of adverbialization) towards new pre-head negation markers (Section 3.1.2., example 40). Furthermore, this remarkable development also accounts for the bipartite basic structure of verbal negation in Arabic, which is constantly adjusted to both linguistic and extra-linguistic needs. As regards marked negation (see diagram below), again stage II is only observed in a limited number of Bedouin varieties, whereas stage III is quite widespread in North Africa. On the other hand, stage I as well as stage II is frequently employed elsewhere in the Arabic speaking areas. Transformation process A: la ___ 0 (stage I) ⇒ ma ___ 0 (stage II)⇒ ma ___ −s(i) (stage III) Moreover, the most western parts of North Africa (especially Morocco) give evidence of a discontinuous marker directly derived from the first stage, as is shown in the next diagram: Transformation process B: la ___ 0 (stage I) ⇒ la ___ −s(i) (stage II)

Reinventing negation patterns in Moroccan Arabic 89

As this evolution process is simultaneously attested with the process exposed in A, the negator ma ___ −s(i) may also have been created by means of the following course of action: Transformation process C: la ___ 0 (stage I) ⇒ la ___ š(i) (stage II)⇒ ma ___ −s(i) (stage III) Why and when all these transformations occurred in Arabic and how they challenge Jespersens’ Cycle model (Jespersen 1917: 4) is definitely outside the realm of this mainly dialectological contribution. Yet, it is a topic that is certainly in need of further investigation, not only from a historical (AfroAsiatic) and typological viewpoint but also from an areal perspective that takes into account the contact situations and phenomena that led to the language change and diffusion.

References Adila, Aziz 1996

La négation en arabe marocain. In La négation en berbère et en arabe maghrébin, Salem Chaker, and Dominique Caubet (eds.), 99– 116. Paris: L’Harmattan. Aguadé, Jordi, and Mohammad Elyaacoubi 1995 El dialecto arabe de Skūra (Marruecos). Madrid: Consejo Superior de Investigaciones Cientificas. Boucherit, Aziza 2002 L’arabe parlé à Alger. Aspects sociolinguistiques et énonciatifs. Paris/Louvain: Peeters. Brugnatelli, Vermondo 1987 La negazione discontinua in berbero e in arabo magrebino. In Atti della 4a Giornata di studi camito-semitici ed indoeuropei, Giuliano Bernini, and Vermondo Brugnatelli (eds.), 53–62. Milan: Unicopli. 2002 Les thèmes verbaux négatifs du berbère: quelques réflexions. In Articles de linguistique berbère. Mémorial Werner Vycichl, K. NaïtZerrad (ed.), 165–180. Paris: L’Harmattan. 2005 Voyelles et accents dans l’histoire du berbère. In Proceedings of the 10th Meeting of Hamito-Semitic (Afroasiatic) Linguistics, Pelio Fronzaroli, and Paolo Marrassini (eds.), 371–380. Firenze: Università di Firenze. 2011 Négations, participes et figement en berbère: nouvelles hypothèses. In “Parcours berbères”. Mélanges offerts à Paulette Galand-Pernet et Lionel Galand pour leur 90e anniversaire, Amina Mettouchi (ed.), 521–532. Köln: Köppe. Caubet, Dominique 1992 Deixis, aspect et modalité, les particules hā- et -ā- en arabe marocain. In La deixis, Marie-Annick Morel, and Laurent Danon-Boileau (eds.), 139–49. Paris: P.U.F. 1996 La négation en arabe maghrébin. In La négation en berbère et en arabe maghrébin, Salem Chaker, and Dominique Caubet (eds.), 79– 97. Paris: L’Harmattan. Chaâbane, Nadia 1996 La négation en arabe tunisien. In La négation en berbère et en arabe maghrébin, Salem Chaker, and Dominique Caubet (eds.), 117–134. Paris: L’Harmattan. Chaker, Salem 1996 Quelques remarques préliminaires sur la négation en berbère. In La négation en berbère et en arabe maghrébin, Salem Chaker, and Dominique Caubet (eds.), 9–22. Paris: L’Harmattan.

Reinventing negation patterns in Moroccan Arabic 91 Cohen, David 1984 La phrase nominale et l’évolution du système verbal en sémitique. Etude de syntaxe historique. Paris: Société de Linguistique de Paris. Croft, William 2000 Explaining Language Change: An Evolutionary Approach. Harlow: Longman. Elhalimi, Brahim 1996 La négation dans le parler arabe de Mazouna (Algérie). In La négation en berbère et en arabe maghrébin, Salem Chaker, and Dominique Caubet (eds.), 135–162. Paris: L’Harmattan. Galand, Lionel 1995 La negation en berbère. Matériaux arabes et sudarabiques (GELLAS, nouvelle série) 8: 169–181. Harrell, Richard S. 1962 A Short Reference Grammar of Moroccan Arabic. Washington D.C.: Georgetown University Press. Heine, Bernd, and Tania Kuteva 2003 On contact-induced grammaticalization. Studies in Language 27: 529–572. Iaaich, Jamal 1996 La négation en hassaniyya de Tan-Tan (Maroc). In La négation en berbère et en arabe maghrébin, Salem Chaker, and Dominique Caubet (eds.), 135–162. Paris: L’Harmattan. Jespersen, Otto 1917 Negation in English and Other languages. Copenhagen: Host & Son. Lafkioui, Mena 1996 La négation en tarifit. In La négation en berbère et en arabe maghrébin, Salem Chaker, and Dominique Caubet (eds.), 49–77. Paris: L’Harmattan. 1999 Syntaxe intégrée de l’énoncé non-verbal berbère. PhD, INALCO (Paris). 2002 L’intonation et ses fonctions syntaxiques en rifain. In Articles de linguistique berbère. Mémorial Werner Vycichl, Kamal Naït-Zerrad (ed.), 253–281. Paris: L’Harmattan. 2007 Atlas linguistique des variétés berbères du Rif. Köln: Köppe. 2009 Synchronic and diachronic linguistic variation as an indicator of language change and diffusion in Tarifit. Studi Maġrebini (Nuova Serie) 6: 65–84. 2010 La topicalisation en berbère: formes et structures. In Études berbères V – Essais sur des variations dialectales et autres articles, Dymitr Ibriszimow, Rainer Vossen, Harry Stroomer, and Maarten Kossmann (eds.), 121–132. Köln: Köppe.

92

Mena Lafkioui 2011a

Etudes de la variation et de la structuration linguistiques et sociolinguistiques en berbère. Köln: Köppe. 2011b How system-internal linguistic factors indicate language change and diffusion. A geolinguistic analysis of Berber data. Dialectologia et Geolinguistica 19: 62–80. 2011c Intonation et topicalisation en berbère. In Mélanges offerts à Paulette Galand-Pernet & Lionel Galand pour leur 90ème anniversaire, Amina Mettouchi (ed.), 387–397. Köln: Köppe. Lucas, Christopher 2007 Jespersen’s Cycle in Arabic and Berber. Transactions of the Philological Society 105(3): 398–431. Marçais, Philippe 1977 Esquisse grammaticale de l’arabe maghrébin. Paris: Maisonneuve. Owens, Jonathan 1984 A Short Reference Grammar of Eastern Libyan Arabic. Wiesbaden: Harrassowitz. Pennacchietti, Fabrizio 1967 Sull’origine della particella negativa araba mā. Annali dell’Istituto Orientale di Napoli 18: 15–23. Petites soeurs de Jésus 1974 Contes touaregs de l’Aïr. Paris: SELAF. Prasse, Karl-G, Ghoubeïd Alojaly, and Mohamed Ghabdouane 2003 Dictionnaire touareg-français (A-L). Copenhagen: Museum Tusculanum Press. Schmitt-Brandt, Robert 1978 Berberische Adstrateinflüsse im maghrebinischen Arabisch. Folia Linguistica 13: 229–235. Taine-Cheikh, Catherine 1995–96 Trois points de vue sur la negation mā dans le dialecte arabe de Mauritanie. Matériaux arabes et sudarabiques (GELLAS, Nouvelle Série) 7: 11–61. 2000 Les emplois modaux de la negation lā dans quelques dialects arabes. Comptes rendus du GLECS 33: 39–86. Tesnières, Louis 1959 Eléments de syntaxe structurale. Paris: Klincksieck. Vanhove, Martine 1994 Sur le fonctionnement de la négation en maltais. Matériaux arabes et sudarabiques (GELLAS, Nouvelle Série) 6: 141–167. Wehr, Hans 1953 Zur Funktion arabischer Negationen. ZDMG 103: 27–39.

Reinventing negation patterns in Moroccan Arabic 93 Woidich, Manfred 1995–97 Negation in the Egyptian Arabic dialect of the Dakhla-Oasis: a case of rule morphologization. Mediterranean Language Review 9: 13–28. Youssi, Abderrahim 1992 Grammaire et lexique de l’arabe marocain moderne. Casablanca: Wallada.

The prosody of Juba Arabic: split prosody, morphophonology and slang Shuichiro Nakao 1. Introduction The typological account of Juba Arabic (abbreviated: JA) and Nubi, closely related Arabic Creoles spoken in Africa, has long been controversial where their prosodic features are concerned. Both languages have been described as tone languages, pitch-accent (tonal accent) languages, and even stressaccent languages in previous researches, given in chronological order in what follows: (1)

Nhial (1975: 81) Owens (1977: xiv–xv) Heine (1982: 26–27) Watson & Ola (1985: 4) Nyombe (1986: 71) Kaye (1991) Chol (2005: 79) Wellens (2005: 53–55) Luffin (2005: 74) Manfredi (2005: 76) Gussenhoven (2006) Miller (2006: 520)

JA & Nubi Nubi Nubi JA JA JA & Nubi JA Nubi Nubi JA Nubi JA

Tone (for both) Stress-Accent Tone Stress-Accent Tone Tone (de facto, for both) Stress-Accent (“nabr”) Stress-Accent & Tone Stress or Pitch-Accent Pitch-Accent Pitch-Accent Pitch-Accent (“tonal stress”)

This chapter is an attempt to offer an alternative account of the prosody of modern basilectal Juba Arabic, concluding it would be best described as a “split prosodic” language, a new typological category proposed by Good (2004a, b). It is organized as follows: Section 2 describes the characteristics of lexical prosody and prosodic split in Juba Arabic lexicon, which consists of Pitch-Accent (PA)-type words and Tone (T)-type words. Section 3 presents PA-type morphology and Section 4 investigates T-type morphology

96

Shuichiro Nakao

emerging in youth slang. A historical account of the development of split prosody is given in conclusion.1

2. Split prosody: the coexistence of pitchpitch-accent with tone 2.1. Characteristics of prosody in Juba Arabic First, we examine some basic word-prosodic facts of Juba Arabic on the basis of the (semi-)minimal pairs shown in (2). These types of pairs show Juba Arabic utilizes pitch to distinguish each word, distinguishing “high tone” (H [ɮ]: marked with an acute accent), “low tone” (L [ɲ]: unmarked) and “falling tone” (F [\Ơ]: marked with a circumflex) at least in surface representation. As (2a) suggests, the TBU of Juba Arabic is basically the syllable. F tone only appears word-finally and triggers the optional lengthening of simple vowels, which is phonologically redundant. At least in basilectal Juba Arabic, intensity is not stable enough to consider its distinctiveness. In addition, there are some redundant pitch representations in Juba Arabic, e.g. middle tone [ɰ] as an allotone of L in non-word-final syllables, as in (2b), and rising tone [/Ơ] in a closed syllable, as a result of phonetic syncope/apocope, as in (3).

1.

This chapter is based on the author’s M. A. dissertation (Nakao 2011a) and his research supported by the Japan Society for the Promotion of Science, Grantin-Aid for JSPS Fellows (23•6924). Both are based on the author’s fieldwork in Juba (South Sudan) during 2009 and 2010. All the data are provided mainly by Tenet and Päri people living in Juba. The segmental orthography of Juba Arabic used in this chapter is given below. However, the segmental inventory is unstable with a diversity of idiolects, as is ordinary with Pidgins and Creoles (e.g. /sh/~/s/, /j/~/z/, /h/~Ø, and /p/~/f/). Basilectal Juba Arabic has no distinction in length (duration) of vowels or consonants. Consonants: /b/, /p/, /m/, /f/, /v/, /w/, /d/, /t/, /n/, /l/, /r/, /j/ [ǯ], /ch/ [c] /sh/ [ʃ], /ny/ [Ȃ], /y/ [j], /g/ [ɡ], /k/, /ng/ [ŋ], /’/ [ʔ], /h/ Vowels: /i/, /e/, /a/, /o/, /u/, /ai/, /ei/, /au/, /ou/ Additional characters are used to describe the segmental representations in Sudanese Arabic (/T/ [dɩ], /V/ [tɩ], /W/ [zɩ], /X/ [sɩ], /Y/ [lɩ], /Z/ [rɩ], /gh/ [ȑ]/, /kh/ [χ] , /‛/ [ȥ], /_/ [ħ]).

The prosody of Juba Arabic

(2) a. [saɮ] sá ‘right’ vs. [saɚ\Ơ] [domɮ] dóm ‘blood’ vs. [joɚm\Ơ] [dulɮ] dúl ‘shade’ vs. [fuɚl\Ơ] [haiɮ] hái ‘alive’ vs. [hai\Ơ] [ǯouɮ] jóu ‘weather’ vs. [nou\Ơ] [auɮ] áu ‘or’ vs. [wau\Ơ] b. [keɲ.niɮ] kení ‘second wife’ vs. [keɰ.niɲ] [boȂɲ.ǯoɮ] bonjó ‘pumpkin’ vs. [boȂɰ.ǯo ɲ] [paŋɮ.ɡaɲ] pángga ‘machete’ vs. [daŋɮ.ɡaɮ] [ǯaɮ.maɲ] jáma ‘university’ vs. [ǯaɲ.maɚ\Ơ] [zaɮ.manɲ] záman ‘time’ vs. [zaɲ.maɚn\Ơ] [aɲ.i ɮ.waɲ] aíwa ‘yes’ vs. [aɲ.iɲ.waɮ] [kiɮ.bi ɰ.raɲ] kíbira ‘experience’ vs. [kiɰ.biɰ.raɲ] [aɮ.fu ɰ.raɲ] áfura ‘to dig’ vs. [baɮ.fuɮ.raɮ] [buɮ.kuɰ.ra] búkura ‘tomorrow’ vs. [koɮ.koɮ.raɲ] (3)

anína tanína giníta degíga galí derú yaú

[aɲ.niɮ.naɲ] ~ [an/Ơ.naɲ] [taɲ.ni ɮ.naɲ] ~ [tan/Ơ.naɲ] [ɡiɲniɮ.taɲ] ~ [ɡin/Ơ.taɲ] [deɲ.ɡiɮ.ɡaɲ] ~ [deɚɡ/Ơ.ɡaɲ] [ɡaɲ.liɮ] ~ [ɡaɚl/Ơ] [deɲ.ruɮ] ~ [deɚr/Ơ] [jaɲ.uɮ] ~ [jaɚ/Ơ]

97

sâ ‘hour’ yôm ‘day’ fû ‘(broad) bean’ hâi ‘eek’ (intj.) nôu ‘kind’ wâu‘Wau (town)’ keni‘church’ (slang) bonjo ‘kid’ (slang) dánggá ‘bow’ jamâ ‘people’ zamân ‘in old times’ aiwá ‘I see’ kibira ‘forest’ báfúrá ‘cassava’ kókóra ‘redivision’

‘we, us’ (also nína, ína) ‘our, ours’ (also teína) ‘anus’ ‘minute’ ‘that’ (quotative) ‘to be wanted’ ‘here is/are’

The function of pitch in Juba Arabic is not limited to lexical distinction, but it also marks grammatical contrasts. We have pairs of functional words only distinguished by pitch (See Section 3 for morphological minimal pairs). (4)

gúna tákum ‘your your (pl.) song’ vs. gúna takum ‘sing (impera (imperative pl.)’ pl.) ma gúna ‘with with a song’ vs. mâ gúna ‘not not a song’ fi gúna ‘in in a song’ vs. fí gúna ‘there there is a song’

Furthermore, in youth Juba Arabic, pitch seems to bear a higher functional load. Juba Arabic, as described in most preceding works, has possessed two imperfect TAM markers: bi for irrealis and gí for realis, however, they are becoming bi ~ gi for irrealis and bí ~ gí for realis in youth speech (Nakao 2011b), as in (5).

98

(5)

Shuichiro Nakao

ána bi béredu (~ ána gi béredu) ‘I will take a bath.’ ána gí béredu (~ ána bí béredu) ‘I am taking a bath.’

2.2. Split prosody in Juba Arabic 2.2.1. Pitch-accent in Juba Arabic Given that most words have only one H or F, as paradigmatized in (6), it is too early to qualify Juba Arabic as a tone language. The words in question are mainly of Arabic origin, and in most cases the pitch patterns can be diachronically explained by the stress accent system of the lexifier (Sudanese Arabic), where the position of stress is calculated according to syllable weight.2 Synchronically, though, the position of H/F in each word is lexically idiosyncratic; therefore, it must be given in the underlying representation.3

2.

3.

Polysyllabic words with a surface word-final H mainly represent an inheritance from Sudanese Arabic “high tone” (Dickins 2007), which occurs in some lexical items, e.g. Sudanese Arabic yātú > Juba Arabic yatú ‘which’. Surface H of monosyllabic words seems to have developed through tonogenesis. Monosyllabic words with word-final gemination in Sudanese Arabic usually correspond to monosyllabic words with H in Juba Arabic as below. Usually long vowels of Sudanese Arabic correspond to F in Juba Arabic. SA. ,a-- – JA. sá ‘right’ cf. SA. sā‛a – JA. sâ ‘hour’ SA. jaww – JA. jóu ‘weather’ cf. SA. nō‛ – JA. nôu ‘kind’ SA. damm – JA. dóm ‘blood’ cf. SA. yōm – JA. yôm ‘day’ SA. 0ull – JA. dúl ‘shadow’ cf. SA. fūl – JA. fûl ‘bean’ SA. nayy – JA. néi ‘row’ SA. simm – JA. sím ‘poison’ SA. murr – JA. múr ‘bitter’ SA. sirr – JA. sír ‘secret’ SA. -agg – JA. hák ‘right’ SA. aw – JA. áu ‘or’ Free variations on the place of H/F are also observed; e.g. méderesa ~ mederésa ‘school’, rúmla ~ rumlá ‘sand’, ála ~ alâ ‘God’, dúniya ~ duníya ‘world’, gurgúr ~ gurgûr ‘snail’.

The prosody of Juba Arabic

(6)

1 syllable sá ‘right’ sâ ‘hour’

2 syllables watá ‘ground’ jamâ ‘people’ gára ‘gourd’

3 syllables gamará ‘moon’ usubû ‘week’ kadára ‘spike’ bágara ‘cow’

99

4 syllables ateresá ‘dumb’ musuturâ ‘toilet’ terebéza ‘table’ melériya ‘malaria’ méderesa ‘school’

In interpreting this system, it is worth noticing the prosodic system of Nubi, a sister language of Juba Arabic, of which Gussenhoven (2006) proposed that its phonological word can be characterized as [+obligatory +obligatory] +obligatory (at least one H per word) and [+c +culminative +culminative] ulminative (at most one H per word), differing from typical pitch-accent languages, such as Tokyo Japanese, in lacking a class of “unaccented” word.4 This explanation for Nubi is inapplicable to Juba Arabic, since it cannot explain the distinctiveness of H and F in word-final position (at least in surface representation). If we analyze it in a way parallel to Tokyo Japanese, by regarding words in the first line as “unaccented”, i.e. [-obligatory], the underlying representation of (6) could be given as in (7), using H* for pitch-accent (surface H in word-middle, and F at word-final), and Ø for underspecified (surface L or H#). (7)

Ø H*

Ø.Ø Ø.H* H*.Ø

Ø.Ø.Ø Ø.Ø.H* Ø.H*.Ø H*.Ø.Ø

Ø.Ø.Ø.Ø Ø.Ø.Ø.H* Ø.Ø.H*.Ø Ø.H*.Ø.Ø H*.Ø.Ø.Ø

Using this analysis, Juba Arabic can be categorized as a typical pitchaccent language with [--obligatory] +culminative] obligatory and [+culminative +culminative (at most one H*). The surface representation of /Ø/ can be explained by post-lexically given boundary tones,5 such as %L (drops when word-initial H*), H% (if no H*) and L% (elsewhere), as in (8). 4. 5.

See Hyman (2006) for Gussenhoven’s view on the pitch-accent system in the prosodic typology. Though it is not discussed here in detail, it should be mentioned that Juba Arabic has an intonational system. For example, H% at sentence-final to indicate a polar question; dé bágara [deɮ # baɮ.ɡ raɲ] ba ɡaɰ.ra ra (with L%) ‘This is a cow’ vs. dé bágara [de deɮ raɮ] de # baɮ.ɡ ba ɡaɮ.ra ra (with H%) ‘Is this a cow?’ Gussenhoven (2006: 195) reports that Nubi is devoid of an intonational system.

100

(8)

Shuichiro Nakao

ga.ma.rá

u.su.bû

%L

%L H* L%

H%

bá.ga.ra (%L) H* L%

As will be described in Section 3, Juba Arabic has de-accenting (delinking H*) and accent-shift (delinking H*, and linking to a syllable other than the original one) in word formation, which are typically seen in pitch-accent languages.

2.2.2. Tone in Juba Arabic Though the pitch-accent system accounts for most of the prosodic appearance of lexical items, it is still not enough for describing the whole system. Juba Arabic, as a contact language influenced by Nilo-Saharan languages of South Sudan, has a number of borrowings from these languages, whose prosodic behaviour does not conform to the account above.6 They could be interpreted as necessary to require assignments of H, L or F tone to each syllable in the underlying level.7 (9)

6.

7.

All High: dánggá ‘bow’ kwátá ‘wrestling’ dóngóng ‘back of head’ lókílíng ‘elbow’ kórókórók ‘moneybox’

mányáng ‘monitor lizard’ báfúrá ‘cassava’ bánggírí ~ gwánggírí ‘cheek’ jújú ‘shrew’ múnúkí ‘Munuki (area in Juba)’

Cognate loanwords are observed in the Arabic-based Pidgins spoken in colonial East Africa. It may indicate that the split prosody already had existed at that time. For example, Jenkins (1909) provides kibera ‘forest’ < Luganda ekibira (JA. kibira), fūtukū ‘leg (of beef, mutton, etc.)’ < Madi putuku (JA. putuku ~ putukuru), lobolo ‘banana’ < Bari rabolo (JA. rabolo ~ laboro ~ labolo). See Nakao (2012) for etymological explanations of some of these words. Some words show free variations concerning the prosodic class, where tonelike pitch patterns are considered more basilectal (or rural), and pitch-accentlike ones are considered more acrolectal; e.g. gwedegwéde (PA) ~ gwedegwédé (T) ‘African spinach’, mundukúru (PA) ~ mundúkurú (T) ‘Arab’, tenêt (PA) ~ ténét (T) ‘Tennet (Surmic people)’, salva-kîr (PA) ~ sálva-kíír (T) ‘Salva Kiir’, jon-gárang (PA) ~ jôn-garáng (T) ‘John Garang’.

The prosody of Juba Arabic

101

All Low: kibira ‘forest’ putuku ‘hoof (food)’ laboro ‘banana’ kwete ‘white beer’ dwar ‘hunting’ jenggeli ‘leper’ jokjok ‘spirit-possession’ kau ‘cowpea’ nyakuron ‘Nyakuron (area in Juba)’ Mixed: alíwárá ‘second-hand’ mútukelí ‘sandal made of tire’ kapáparât ‘butterfly’ adúnggú ‘Acholi harp’ wótisí ‘sneeze’ ayúyúi ‘ululation’ tuútuút ‘poisonous ant’ kúrikúri ‘eagle, hawk’ gulunggúlúng ‘round’ giringgíríng ‘spotted’ This type of words basically has a minor place in Juba Arabic lexicon and lacks morphological processes (e.g. pluralization by suffixation). We name this prosodic class “Tone (T)-type words”, as opposed to the “Pitch-Accent (PA)-type words” in 2.2.1., which are characterized as in Table (1). Table 1. Features of PA-type words and T-type words

in JA lexicon origin8 phonological word morphology

PA-type words majority mainly Arabic at most one H* yes

T-type words minority mainly vernaculars no limitation (basically) no

In fact, this kind of prosodic split in the lexicon originating from language contacts was first reported in Saramaccan by Good (2004a, b). He states that “Saramaccan, an Atlantic Creole whose lexifier languages are Portuguese and English, has a “split” prosodic system wherein the majority of its words are marked for pitch-accent but an important minority are marked for tone” (Good 2004b: 11). This phenomenon is peculiar to contact languages as other cases have already been reported in English-based contact languages, e.g. Krio (Finney 2002, 2004), Pichi (Yakpo 2009), and some Asian Englishes (Lim 2009). 8.

The origin of a lexical item does not always match the prosodic class (PA or T). Some T words are etymologically derived from Arabic, e.g. lebenlébén ‘milk bush’ < lében ‘milk’. Also mútukelí ‘sandal made of tire’ is from Arabic mútu ‘to die’ and kelí ‘to remain’ (i.e. ‘it is strong enough to remain after your death’). On the contrary, some clear borrowings from vernacular are treated as PA words, e.g. korófo ‘leaf’ < Bari koropo, kení ‘second-wife’ < Bari köyini, which have PA-type morphology (see Section 3).

102

Shuichiro Nakao

For the time being, Juba Arabic is the only example of this type of nonEuropean-based contact language.

3. PitchPitch-accentaccent-type morphology The morphology of Juba Arabic is, as is typical of Pidgins and Creoles, reduced and full of re-analyses when compared to the lexifier. We will deal with most morphological processes of Juba Arabic below, i.e. compounding, suffixation, and reduplication.

3.1. Compounding Compounding has been developed as a way of productive word formation, where the compounds are right-headed and H* of the head noun is delinked. Compounds vary freely with a general genitive construction with a proclitic genitive marker ta, e.g. móyo ta éna (‘water’ GEN ‘eye’) ~ moyo-éna (water-eye) ‘tear’. This process is faithful to the culminativity constraint (at most one H*), so each compound is interpreted to constitute one (PA-type) phonological word, and the phonological derivation can be depicted as in (10). (10)

mo.yo-é.na = %L H* H* L%

Delinking H* is obligatory only in compounding (11), but inapplicable to a simple noun phrase (12), which constitutes two (phonological) words. The minimal pairs in (13) show that numerals (e.g. wáhid ‘one’) bear ordinal sense only when compounded with head nouns such as sâ ‘hour’, sháhar ‘month’ and nímira ‘number’. (11) mahâl ‘place’ & zára ‘plant’ > mahal-zára ‘field’ sídi ‘owner’ & mahal-zára ‘field’>sidi-mahal-zára ‘owner of a field’ râs ‘head’ & júwa ‘house’ > ras-júwa ‘roof’ korófo ‘leaf’ & bonjó ‘pumpkin’ > korofo-bonjó ‘pumpkin leaf’ árabi ‘Arabic’ & júba ‘Juba’ > arabi-júba ‘Juba Arabic’

The prosody of Juba Arabic

(12) móyo ‘water’ & ketîr ‘a lot’ > mahâl ‘place’ & tanína ‘our’ > (13) sâ wáhid ‘an hour’ sháhar wáhid ‘a month’ nímira wáhid ‘a number’

103

móyo ketîr ‘a lot of water’ mahâl tanína ‘our place’

vs. sa-wáhid ‘1 o’clock’ vs. shahar-wáhid ‘January’ vs. nimira-wáhid ‘no. 1’

The T-type words are excluded from head nouns, but they can be dependent nouns of compounds as in (14). Theoretically, these “chimaeric” examples should be treated as T-type words since they violate the culminativity constraint. (14) fâr ‘mouse’ & jújú ‘shrew’ > far-jújú ‘shrew’ jébel ‘mountain’ & kórók (place) > jebel-kórók ‘Korok mountain’ hái ‘quater’ & nyakuron (place) > hai-nyakuron ‘Nyakuron square’

3.2. Suffixation Juba Arabic has only a few suffixes for (i) pluralization, (ii) passivization, (iii) verbalization of nouns and adjectives, and (iv) nominalization of verbs. Suffixation generally requires prosodic shift, though the shift patterns are dependent on suffix types. There are some prosodic morphemes (tonal morphemes), which are also treated here. At first, we will simply describe each phenomenon mentioned above, after which we will try to give a possible analysis in Section 3.2.5.

3.2.1. Pluralization/dualization In what follows, we present examples of plural suffixes: (15a) -ât only for nouns (including some compounds), (15b) -în for adjectives and a few personal nouns, and (15c) the dual suffix -ên/tên, which is limited to a few nouns. Word-final vowels are sometimes dropped, though the circumstances are synchronically unpredictable. (15) a. sâ > sa-ât jéres > jeres-ât ganamóya > ganamoy-ât

‘hour’ ‘bell’ ‘goat’

104

Shuichiro Nakao

korófo kení mahal-zára ras-júwa b. kwês tabân giyáfa mudéris láji c. sháhar yôm ída sâ séna usubû

> > > > > > > > > > > > > > >

korof-ât keni-yât mahal-zar-ât ras-juw-ât kwes-în taban-în giyaf-în muderis-în laj-în shahar-ên yom-ên id-ên sa-tên sena-tên usubu-(t)ên

‘leaf (food)’ ‘second-wife’ ‘field’ ‘roof’ ‘good’ ‘tired’ ‘beautiful’ ‘teacher’ ‘refugee’ ‘month’ ‘day’ ‘hand, arm’ ‘hour’ ‘year’ ‘week’

3.2.2. Passivization The passive form of a verb is formed by suffixing -ú, as in (16a). The passive of a few exceptional verbs in (16b) are formed only by prosodic shift. (16) a. rúdu séregu ásurubu ágara kóre nesítu b. káti nadí wodí ligó

> > > > > > > > > >

rud-ú sereg-ú asurub-ú agara-ú kore-ú nesit-ú kat-í nad-í wod-í lig-ó

‘to accept, like’ ‘to steal’ ‘to drink’ ‘to read, study’ ‘to cry out’ ‘to forget’ ‘to cover’ ‘to call’ ‘to give’ ‘to find, get’

3.2.3. Verbalization Nouns and adjectives are verbalized by suffixing -u and shifting H* to the word-initial syllable of the stem, as in (17a). If the stem-final is an open

The prosody of Juba Arabic

105

syllable, the segmental suffix -u is sometimes lacking (cf. passive guna-ú ‘to be sung’), as in (17b): (17) a. dusmân ‘fight’ gurbâl ‘sieve’ kebîr ‘big, large’ towîl ‘long’ waskân ‘dirty’ tabân ‘tired’ hílim ‘dream’ ilâj ‘treatment’ hujûm ‘attack’ zirára ‘button’ b. gúna ‘song’ shákila ‘fight’

dúsman-u ‘to fight’ gúrbal-u ‘to sieve’ kébir-u ‘to enlarge’ tówil-u ‘to live long’ wáskan-u ‘to dirty’ tában-u ‘to make tired’ hílim-u ‘to dream’ ílaj-u ‘to treat’ (also áliju) hújum-u ‘to attack’ (also hájimu) zírar-u ‘to button’ gúna ‘to sing’ (pass. guna-ú ‘to be sung’) shákila ‘to fight’

3.2.4. Nominalization Juba Arabic has several patterns of nominalization (as listed below), though the productivity of each pattern seems to differ in each idiolect. It would be best to begin with an introduction of existing descriptions of Nubi by Wellens (2005) and Gussenhoven (2006) before we categorize those of Juba Arabic. Wellens (2005: 54–55, 184–193) reports that Nubi has two nominalization patterns (“gerund” and “infinitive” in her terms), differing in the prosodic representations. “Gerund” has H in the penultimate syllable of the verb stem and requires the genitive marker ta before the object noun (Gerund ta Object), while “infinitive” has H in the word-initial two syllables and is immediately followed by an object (Infinitive Object). These constructions are critically revised by Gussenhoven (2006), who claims her “infinitive” is a bound form of gerund in a compound, cf. kásulu ‘to wash’ in Table (2). Wellens also mentions that a few verbs turn word-final vowel -u into -a in nominalization (e.g. séregu ‘to steal’ > serégu ~ seréga), seréga or have a suppletional form (“irregular gerund” in her term, e.g. ágara ‘to read’ > agára ~ garáya). garáya

106

Shuichiro Nakao

Table 2. Interpretations of Nubi nominalizations

Wellens (2005) HLL (kásulu): bare verb LHL (kasúlu): gerund HHL (kásúlu):9 infinitive

Gussenhoven (2006) HØØ (kásulu): bare verb ØHØ (kasúlu): gerund ØØØ (kasulu-N):compounded gerund

Patterns of verb nominalization in Juba Arabic look similar to Nubi. However, the identical analysis leaves some problems. As the next examples show, the nominalization pattern differs in each verb. Some verbs have only “gerund” forms (18a) and some have suffixed forms (-a) (18b) or suppletional forms (18c), or both (18d) along with a “gerund” form, while a few relatively new verbs have only suppletional forms (18e). (18)

9.

Finite a. ákulu ‘to eat’ > kúruju ‘to cultivate’ > ásurubu ‘to drink’ > jére ‘to run’ > nesítu ‘to forget’ > wodí ‘to give’ > ligó ‘to find’ > b. séregu ‘to steal’ > kárabu ‘to destroy’ > bérjilu ‘to mess up’ > wéledu ‘to bear’ > nyákamu ‘to attack’ > sítimu ‘to insult’ > kútu ‘to put’ > c. dóuru ‘to walk’ > kóre ‘to cry’ > ágara ‘to read, study’ > nédifu ‘to clean’ >

Gerund akúlu kurúju asurúbu jére nesítu wódi lígo serégu karábu berjílu welédu nyakámu sitímu kútu doúru kóre agára nedífu

Suffixed/Suppletion ───── ‘eating’ ───── ‘cultivating’ ───── ‘drinking’ ───── ‘running’ ───── ‘forgetting’ ───── ‘giving’ ───── ‘finding’ serég-a ‘stealing’ karáb-a ‘destruction’ berjíl-a ‘mess’ weléd-a ‘birth’ nyakám-a ‘attack’ sitím-a ‘insult’ kút-a ‘putting’ douríya ‘walking’ korarâk ‘crying’ giráya ‘study’ nadáfa ‘cleaning’

The author supposes that Wellens’s transcription /HHL/ may be the Nubi equivalent of a redundant “Middle tone”, i.e. [MML]. See phonetic representation of (2b) kibira ‘forest’.

The prosody of Juba Arabic

107

d. wónusu ‘to talk’ > wonúsu wonús-a ~ wánasa ‘talk’ fékiru ‘to think’ > fekíru fekír-a ~ fékira ‘tought’ e. istákdam ‘to use’ > ───── istikdâm ‘use’ istemîr ‘to colonize’ > ───── istimâr ‘colonizing’ istágrab ‘to be surprised’> ──── istigrâb ‘surprise’ In addition, there is another method of nominalization which resembles the compounding process, which would correspond to Wellens’ “infinitive”, as is shown in the examples in (19a). Here, gurûsh ‘money’ and kitâb ‘book’ are object nouns. (19) a. b. c. d.

‘stealing money’ seregu--gurûsh seregu serégu ta gurûsh seréga ta gurûsh *serega-gurûsh

‘reading a book’ agara--kitâb agara agára ta kitâb giráya ta kitâb *giraya-kitâb

(tentative namings) (“infinitive”) (“gerund”) (suffixed/suppl. “gerund”)

The examples in (19d) clearly indicate that the suffixed form (serég-a) or the suppletional form (giráya) cannot be compounded. Would it then be best to consider the examples in (19a) to be compounded forms of a “gerund” following Gussenhoven’s analysis of Nubi in Table (2), or compoundded forms of a bare verb? (20) a.

se.re.gu-gu.rûsh = or %L H* H* L%

b. se.re.gu-gu.rûsh = %L H* H* L%

In fact, verbs which lack “gerund” forms (18e) have “infinitive” forms as in (21), and this fact would rule out the former explanation. Thus, the author would be for (20b), considering construction (19a) a different type of nominalization. (21) istakdam-gurûsh ‘using money’ We can add the sociolinguistic information that the “gerund” form is losing productivity, especially among younger people (20s–30s), in whose idiolects nominalization requires an object noun. For example, one speaker did not accept agára ‘reading’, but accepted suppletional giráya ‘reading’ and agara-kitâb ‘reading a book’. The author suspects this is because of an

108

Shuichiro Nakao

influx of suppletional gerunds borrowed from Sudanese Arabic, which are common in mesolectal Juba Arabic.

3.2.5. How to analyze prosodic shift in suffixation We still have a problem in analyzing the phonological derivation of suffixed forms, i.e. whether the prosodic patterns of suffixed forms could be analyzed in the parallel way with compounding, or another principle, say, accent-shift is needed. Both analyses explain plural/dual-type suffixation, as in (22), though passive, “gerund” and verbalization patterns are inexplicable. (22)

je.re.s-ât = %L H* H* L%

or

je.re.s-ât = %L H* L%

Considering all derivation patterns that we have observed in Section 3, the author typologizes the suffixes as in Table (3) and (23). That is to say, each suffix has information (rather abstract compared to lexical morphemes) of an idiosyncratic pattern of de-accenting or accent-shift observed in the word formation. Table 3. Derivation patterns of suffixing

Delinking H* H* to ultimate H* to penultimate H* to word-initial (23)

se.re.g-ú = %L H* H%

Pass. -ú Yes No No No je.re.s-ât = %L H* L%

Pl. -ât Yes Yes No No

Ger. -u/-a Yes No Yes No

Verb. -u Yes No No Yes

se.ré.g-a ké.bi.r-u = = %L H* L% (%L) H* L%

The prosody of Juba Arabic

109

3.3. Reduplication 3.3.1. Nominal reduplication Nominal reduplication is a process of semantic derivation categorized as below. This type of reduplication is faithful to the culminativity constraint, and has no problem to be analyzed in the same manner as compounding. (24) a. Reciprocity sabí ‘friend’ kení ‘second-wife’ dúfa ‘age-mate’ gurûp ‘group’ gerîb ‘near’ sendûk ‘box’ akúlu ‘eating, dining’ berédu ‘bathing’ jére ‘running’ b. Time sabâ ‘morning’ nahâr ‘day-time’ bilêl ‘night’ c. Children’s games jéna ‘child’ júwa ‘house’ gátar ‘train’ kalâs ‘finished’

sabi-sabí ‘frends each other’ keni-kení ‘socond-wives each other’ dufa-dúfa ‘age-mates each other’ gurup-gurûp ‘groups each other’ gerib-gerîb ‘near each other’ senduk-sendûk ‘rotating credit’ akulu-akúlu ‘dining together, mess’ beredu-berédu ‘bathing each other’ jere-jére ‘running race’ saba-sabâ ‘early in the morning’ nahar-nahâr ‘midday’ bilel-bilêl ‘midnight’ jena-jéna‘playing mam’s (with a doll)’ juwa-júwa ‘playing house (made of sand)’ gatar-gátar ‘playing train (with a rope)’ kalas-kalâs ‘playing tag’

3.3.2. Adjectival/adverbial reduplication However, reduplicated adjectives/adverbs violate the culminativity constraint, as the following examples show. It should be interpreted to count two “phonological words”. In other words, they are examples of inconsistency between “phonological words” and “lexical words” in Juba Arabic. Yet, there are some items which freely alternate with nominal reduplication-like prosodic representation, as in (25b).

110

Shuichiro Nakao

We can confirm they are not simply repetitions (which also exists in Juba Arabic, e.g. shedîd ‘very’ vs. shedîd shedîd shedîd ‘very very very’), since the meanings of the original items change when reduplicated. (25) a. móyo ‘water’ béled ‘country’ rutân ‘vernacular’ itnîn ‘two’ seî ‘really’ kêf ‘how’ b. kúlu ‘all, every’ barâu ‘alone’ birâ ‘slowly’ guwâm ‘fast’

móyo-móyo ‘watery’ béled-béled ‘provincial’ rutân-rutân ‘in vernacular accent’ itnîn-itnîn ‘in twos’ seî-seî ‘very (much)’ kêf-kêf ‘terrible’ kúlu-kúlu ~ kulu-kúlu ‘never’ barâu-barâu ~ barau-barâu ‘different’ birâ-birâ ‘little-by-little’ guwâm-guwâm ~ guwam-guwâm ‘very fast’

3.3.3. Verbal reduplication Verbal reduplication is completely productive, and adds iterative sense. The prosodic representation of verbal reduplication is unique in the Juba Arabic prosodic system. It not only violates the constraint of culminativity, but also requires an ad hoc explanation due to the lack of a parallel phenomenon.10 As the examples in (26) show, all syllables in the first reduplicant are represented as surface H and those in the second reduplicant are represented as surface L. The derivation can be tentatively depicted as in (27). (26) séregu ‘to steal’ kóre ‘to cry’ ligó ‘to find, get’ nesítu ‘to forget’

sérégú-seregu ‘to steal repeatedly’ kóré-kore ‘to quarrel’ (also kóró-koro) lígó-ligo ‘to find, get repeatedly’ (also ligo-ligó) nésítú-nesitu‘to forget repeatedly’ (also nesitu-nesítu)

10. However, a diachronic explanation may be possible. The verb gálígali ‘to hassle’ and its free variant gálgal, are obviously an inheritance of Sudanese Arabic galgal ‘to hassle’. Similar, free variations are seen in gurgúr ~ gurugúrú ‘snail’ < Sudanese Arabic gargūr, and dúl ~ dúlú ‘shade’ < Sudanese Arabic 0ull. These pairs of free variations may suggest the phonetic spread of H to an epenthetic vowel (which corresponds to the first reduplicant) was analogically interpreted as the spread of H* to all syllables of the first reduplicant.

The prosody of Juba Arabic

(27)

sé.ré.gú-se.re.gu = (%L) H* H* L%

111

né.sí.tú-ne.si.tu = (%L) H* H* L%

This prosodic pattern also applies to underlying reduplicated verbs, with their morphologically re-reduplicated forms, as in (28): (28) gálígali ‘to hassle’ múnúmunu ‘to cheat’

gálígálí-galigali ‘to hassle repeatedly’ múnúmúnú-munumunu ‘to cheat repeatedly’

When a reduplicated verb is suffixed with passive -ú or nominalized, this spread of H is not applied, as shown in (29): (29) Passive sereg-ú vs. seregu-sereg-ú “Gerund” serégu vs. seregu-serégu “Infinitive” seregu-N vs. seregu-seregu-N

4. Emerging tonetone-type morphology in Juba Arabic slang 4.1. Sociolinguistic situation of Juba Arabic slang Miller (2004) first reports on the slang in Juba Arabic, based on her data collected in the 1980s. Today, the slang has radically changed; many slang words and morphological processes reported in Miller (2004) were not observed during our research period in 2009 and 2010 (Nakao 2011b). A current version of this slang described below is elicited from Juba youth (20s–30s). The slang is mainly used by males,11 and has a lot of lexical and morphological borrowings mainly from Bari. This may be because Bari, as the dominant vernacular of Juba, has obtained a “covert prestige” among the speakers. The usage and knowledge of the slang vary according to each speaker’s idiolect, thus the description below reflects that of my informants to some extent.

11. Probably this male-dominant usage is why there are more than a dozen of homonyms meaning a ‘girl, girlfriend’ (JA. biníya) in the slang, e.g. bámba, logóya, logó, motéma, vite, cíci, larkí, ánsha, anggóla, gúndi, jíha, jáha and gúti.

112

Shuichiro Nakao

4.2. Borrowings Slang has many borrowings from vernaculars (mainly from Bari),12 as in (30). It enhances the relative amount of T-type words in youth’s lexicon. (30) JA words gurûsh ákil jalába harámi kûs gumâs hájer kára sharmúta júbur ketîr kurâ waskân ájila

Slang equivalents nyíng, bôb, jîs lagá, chámá, chúmá lóndó lotúle, lotu nyet, kweret bong mungga ninga chali logot díp kada, keréng longinya jiling

‘money’ ‘food’ ‘Arab (Jallaba)’ ‘thief’ ‘vagina’ ‘clothes’ ‘stone’ ‘faeces’ ‘prostitute’ ‘penis’ ‘a lot’ ‘foot, leg’ ‘dirty, filthy’ ‘bicycle’

Basically slang words are used when substituting one or more original words of a sentence (31a) into a slang equivalent (31b). In addition, new genitive markers borrowed from Sudanese Arabic (e.g. hagí ‘my’ < Sudanese Arabic -aggī ‘my’ hágu ‘his/her’ < Sudanese Arabic -agg-u ‘his’) are substituting original Juba Arabic forms (e.g. taí ~ bitaí ‘my’, tô ~ bitô ‘his/her’). (31) a. sabí dé jínu waskân fi gumâs tô áswid dé. b. jêk dé jínu longinya fi bong hágu áswid dé. guy DEM used.to filthy LOC clothes 3SG.POSS black DEM ‘This guy is used to be filthy, (always) wearing his black clothes.’

12. Though the etymologies of each lexical item are usually unknown to the speakers. There are many borrowings from English, Swahili, and notably Rendók, a slang (or “youth secret language”) popular in Sudan Republic (Manfredi 2008), e.g. sân ‘guy’ < Rendók sān < Sudanese Arabic nās, jáha ‘girl’ < Rendók jā-a < Sudanese Arabic -āja ‘thing’. However, its metathetic morphology itself is unproductive in Juba Arabic slang.

The prosody of Juba Arabic

113

4.3. Neologism 4.3.1. PA-type suffixation There are three suffixes for coining slang words, namely -îs/ês, -ê, and -e. It is not predictable which suffix is attached to a word. The suffix -îs/-ês has been reported by Miller (2004), though it is currently less productive and probably retained only by middle-aged people (30s–40s). The few examples shown in (32) are almost fossilized and no longer considered as slang words. The suffix -ê (33) is more productive. As we can see in the examples, phonological derivations of -îs/ês and -ê seem identical to the pluralization-type suffix (see Section 3.2.4.). The suffix -e, as in (34), is limited to only a few words, and the phonological derivation may be identical to the “gerund”-type. All these suffixes can be regarded as PA-type and not so novel in Juba Arabic morphology. (32) yúma yába andáya mára

> > > >

yum-ês yeb-îs anday-îs mar-îs

‘mom’ ‘dad’ ‘bar’ ‘woman’

(33) itnîn míya bêt júwa júba árnab râs akû

> > > > > > > >

itnin-ê miy-ê bet-ê juw-ê jub-ê arnab-ê ras-ê akuw-ê

‘two’ ‘a hundred’ ‘house’ ‘house’ ‘Juba’ (town) ‘rabbit’ ‘head’ ‘brother’

(34) díkin wáhid béled lében úkut

> > > > >

dikín-e wahíd-e beléd-e lebén-e ukút-e

‘beard’ ‘one’ ‘country’ ‘milk’ ‘sister’

114

Shuichiro Nakao

4.3.2. Back-clipping: root-and-pattern morphology Back-clipping is another process of coining slang words side by side with the suffixes above. It is interesting that back-clipping apparently coins T-type words, where the first two syllables are left and L tone is assigned to each: (35) méderesa kenísa bakána bonjôs bontolôn sondowîsh merísa ishirîn teletîn kamsîn sebeîn temenîn tizeîn ‘smoke’

> > > > > > > > > > > > > >

mede keni baka bonjo bonto sondo meri ishi tele kami sebe teme tize simo

‘school’ ‘church’ ‘toilet’ ‘young child’ ‘trousers’ ‘panino’ ‘beer’ ‘twenty’ ‘thirty’ (also tulu) ‘fifty’ (also kumu) ‘seventy’ (also subu) ‘eighty’ (also tumu) ‘ninety’ (also tuzu) ‘cigarette’

This process is considered as the only instance of root-and-pattern morphology in Juba Arabic, which can be depicted as in (36). Other evidence will be introduced in Section 4.4, in order to declare that these are T-type words. (36)

me.de.[re.sa]

son.do.[wish]

σ σ

σ σ

L

L

Some exceptions to the generalization above are found, as in (37): (37) kásuma bátuna emánuel

> kasú ‘mouth’ > batú ‘belly’ > emá ‘Emmanuel’ (personal name)

The prosody of Juba Arabic

115

4.3.3. Prefixation The slang has a prefix lo/ló/lú- ‘one whose characteristic is…’ which produces personal nouns. A few examples are shown in (38), though more data are needed to define the circumstance of each allomorph. This prefix generates T-type words. It seems to be a borrowing from Bari masculine genitive lo/lu (Spagnolo 1933: 55; Yokwe 1987: 248–284), though the functions are different from those of Bari. (38) pómbe ‘beer’ (slang) simo ‘cigarette’ (slang) nyíng ‘money’ miyáng ‘alcohol’ (slang) beléde ‘country’ (slang) bonjo ‘young child’ (slang)

> > > > > >

lo-pómbé ‘drunkard’ lo-símó ‘smoker’ lo-nyíng ‘stingy’ lo-míyáng ‘drunkard’ lú-beléde ‘provincial (person)’ lú-bonjo ‘young child’

4.4. Plural suffix As previously noted in Section 2.2.2, T-type words lack morphological processes in Juba Arabic. However, this description must be revised in the account of the slang, since the slang demonstrates unique morphological processes which generate T-type words (= “T-type morphology”). Before ending this section, we examine the only inflectional T-type morphology, which is a novel plural suffix, used only for T-type words. It is expected to consider that the influx of T-type words triggered the adoption of this suffix. As shown in (39), -jín jín / H ___ or -jin / L___ (with j > g / ng___) is suffixed to T-type slang words, including forms resulting from back-clipping (39b) and prefixing (39c): (39) a. lóndó lagá chámá mungga jiling b. mede bonto sondo

lóndó-jin lagá-jin chámá-jin mungga-jín jiling-gín mede-jín bonto-jín sondo-jín

‘Arab’ ‘food’ ‘food’ ‘stone’ ‘bicycle’ ‘school’ ‘trousers’ ‘panino’

116

Shuichiro Nakao

baka c. lo-pómbé lo-míyáng

baka-jín ‘toilet’ lo-pómbé-jin ‘drunkard’ lo-míyáng-gin ‘drunkard’

It is surprising that this morpheme is characterized as having a “polar tone” system, which is typical to African tone languages. Furthermore, it must be noted that users of the slang also utilize this suffix for pluralizing non-slang (relatively old) Juba Arabic words, including PA-type words of Bari origin (40b): (40) a. dánggá lókwílíli mányáng alíwárá kwete b. kúri nyalamá

dánggá-jin lókwílíli-jín mányáng-gin alíwárá-jin kwete-jín kúri-jín nyalamá-jin

‘bow’ ‘bat’ ‘monitor lizard’ ‘second-hand clothes’ ‘sorgum/millet beer’ ‘eagle, hawk’ (JA. kuri-yât) ‘toothless’ (JA. nyalam-ât)

This suffix is obviously borrowed from a Bari suffix with an identical polar tone, -jín/-jìn (Yokwe 1987: 156–159). However, Bari has more types of lexically idiosyncratic suffixes (e.g. -a, -at, -et, -an, -ki, -jik, etc.), which are not borrowed into the slang. Thus, the plural forms of the slang sometimes mismatch those of Bari (see example 40), indicating that this process is a new development in the slang: (41) dang lukululi manya kuri nyalama

dang-in lukululy-et manya-jin kuri-lön nyalama-jin

‘bow, arch’ (Spagnolo 1960: 22) ‘bat’ (Spagnolo 1960: 167) ‘Varanus lizard’ (Spagnolo 1960: 179) ‘African black kite’ (Spagnolo 1960: 120) ‘large potsherd’ (Spagnolo 1960: 212)

5. Conclusion In this chapter we have argued that Juba Arabic has a phonological split in the lexicon, which is broadly classified as PA-type words and T-type words. In the former class, each word has at most one H* and, in the latter class, each syllable of a word has idiosyncratic tone information (H, L or F).

The prosody of Juba Arabic

117

PA-type words have morphological processes (compounding, suffixing, and nominal reduplication), which are mostly faithful to the culminativity constraint of pitch-accent, though there are still problems in explaining adjectival/adverbial and verbal reduplication. The significance of this morphological development must be stressed here, since most of these processes are not found in the lexifier, Sudanese Arabic. In other words, PAtype morphology has been developed independently in the process of creolization, presumably in the proto-stage of Juba Arabic and Nubi. On the other hand, T-type words did not have morphological processes, probably because they formed only a minor part of the lexicon. However, as a result of the development of the slang, the number of T-type words increased in the speakers’ lexicon, which motivated the adoption of T-type morphology from Bari. This process could be interpreted as a relexification of Arabic words by Bari borrowings. This phenomenon could be pivotal in the history of Juba Arabic, because it could announce the advent of a second stage in language crystallization, following the former stage of PA-type morphology, as depicted in Table (4): Table 4. Historical development of Juba Arabic morphology

PA-type words

T-type words

PA-type words PA-type morphology

T-type words

PA-type words PA-type morphology

T-type words T-type morphology

This argument also supports Good’s (2009) hypothesis that the prosodic split in Saramaccan was formed during two stages of language mixings, namely pre-Sranan creolization, followed by a relexification of the European lexicon by African lexicon in order to strengthen the linguistic identity of the nascent community of Saramaccan. There have been some theses in favour of the de-creolization of Juba Arabic in the future, e.g. Versteegh (1993), but this historical change of the prosodic system, as a counterexample, tells us that the situation is more complex than has been assumed.

References Chol, Bol Deng 2005 Lahja Jūbā al-‛Arabīya [The Arabic Dialect of Juba]. Al-KharVūm: al-Dār al-Sūdānīya li al-Kutub. Dickins, James 2007 Khartoum Arabic. In The Encyclopedia of Arabic Language and Linguistics. Vol. 2, Kees Versteegh (ed.), 559–571. Leiden: Brill. Finney, Malcolm Awadajin 2002 Compounding and Reduplication in Krio. Journal of West African Languages 29(2): 23–34. 2004 Tone Assignment on Lexical Items of English and African Origin in Krio. In Creoles, Contact, and Language Change: Linguistics and Social Implications, Genevieve Escure, and Armin Schwegler (eds.), 221–236. Amsterdam: Benjamins. Good, Jeff 2004a Tone and Accent in Saramaccan: Charting a Deep Split in the Phonology of a Language. Lingua 114: 575–619. 2004b Split Prosody and Creole Simplicity: The Case of Saramaccan. Journal of Portuguese Linguistics 3: 11–30. 2009 A Twice-mixed Creole? Tracing the History of a Prosodic Split in the Saramaccan Lexicon. Studies in Language 33(2): 459–498. Gussenhoven, Carlos 2006 Between Stress and Tone in Nubi Word Prosody. Phonology 23: 193–223. Heine, Bernd 1982 The Nubi Language of Kibera: An Arabic Creole. Berlin: Reimer. Hyman, Larry 2006 Word-Prosodic Typology. Phonology 23: 225–257. Jenkins, Edward V. 1909 An English-Arabic Vocabulary with Grammar & Phrases, Representing the Language as Spoken by the Uganda Sudanese in the Uganda and British East Africa Protectorates. Kampala: The Uganda Company. Kaye, Alan Sebastian 1991 Nilo-Saharan Influence on Ki-Nubi. In Proceedings of the Third Nilo-Saharan Linguistics Colloquium Kisumu, Kenya, August 4–9, 1986, Franz Rottland, and Lucia Ndong’a Omondi (eds.) 123–129. Hamburg: Helmute Buske.

The prosody of Juba Arabic Lim, Lisa 2009

119

Revisiting English Prosody: (Some) New Englishes as tone languages? English World-Wide 30(2): 218–239. Luffin, Xavier 2005 Un créole arabe: le kinubi de Mombasa, Kenya. München: Lincom. Manfredi, Stefano 2005 Descrizione grammatical dell’Arabo Juba (Sudan Meridionale) con riferimenti sociolinguistici alla comunita Sudanese del Cairo [Grammatical description of Juba Arabic (Southern Sudan) with sociolinguistic references on the Sudanese community of Cairo]. M. A. dissertation, Universita degli Studi di Napoli “L’Orientale”. 2008 Rendók: a Youth Secret Language in Sudan. Estudios de dialectología norteafricana y andalucí 12: 113–129. Miller, Catherine 2003 Reduplication in Arabic-based Contact Languages. Twice as Meaningful: Reduplication in Pidgins, Creoles and Other Contact Languages, Silvia Kouwenberg (ed.), 289–300. London: Battlebridge. 2004 Un parler ‘argotique’ à Juba, Sud Sudan. In Sociolinguistique urbaine: parlers jeunes ici et là-bas, Dominique Caubet, Jacqueline Billiez, Thierry Bulot, Isabelle Leglise, and Catherine Miller (eds.), 69–90. Paris: Harmattan. 2006 Juba Arabic. In Encyclopedia of Arabic Language and Linguistics. Vol. 2, Kees Versteegh (ed.), 517–525. Leiden: Brill. Nakao, Shuichiro 2011a Juba Arabia-go no Purosodī [The Prosody of Juba Arabic]. M. A. dissertation, Kyoto University. 2011b Gendai Jakunensō Juba Arabia-go ni kansuru Yobiteki Hōkoku [A Preliminary Report on Modern Youth Juba Arabic]. Chikyūken Gengo Kijutsu Ronshū 3: 59–83. 2012 Revising the Substratal/Adstratal Influence on Arabic Creoles. In Challenges in Nilotic Linguistics and More, Phonology, Morphology and Syntax, Osamu Hieda (ed.), 127–149. (Studies in Nilotic Linguistics 5.) Tokyo: Research Institute for Language and Cultures of Asia and Africa. Nhial, Abdon Agaw Jok 1975 Ki-Nubi and Juba Arabic: A Comparative Study. In Directions in Sudanese Linguistics and Folklore, Herman Bell, and Sayed Hamid Hurreiz (eds.), 81–93. Khartoum: Khartoum University Press. Nyombe, Bureng George Vincent 1986 Juba Arabic from a Bari Perspective. In Current Approaches to African Linguistics. Vol. 3, Gerrit J. Dimmendaal (ed.), 71–78. Dordrecht: Foris.

120

Shuichiro Nakao

Owens, Jonathan 1977 Aspects of Nubi Grammar. PhD thesis, University of London. Spagnolo, Lorenzo 1933 Bari Grammar. Verona: Missioni Africane. 1960 Bari English Italian Dictionary. Verona: Missioni Africane. Versteegh, Kees 1993 Leveling in the Sudan: from Arabic Creole to Arabic Dialect. International Journal of the Sociology of Language 99: 65–79. Yakpo, Kofi 2009 A Grammar of Pichi. Berlin/Accra: Isimu Media. Yokwe, Eluzai Moga 1987 The Tonal Grammar of Bari. PhD thesis, University of Illinois. Watson, Richard L., and Louis Biajo Ola 1985 Juba Arabic for Beginners. Entebbe: Summer Institute of Linguistics. Wellens, Inneke Hilda Werner 2005 The Nubi language of Uganda: An Arabic Creole in Africa. Leiden: Brill.

Grammaticalized uses of the verb a(a) in Arabic: a Maghrebian specificity? Catherine Taine-Cheikh 1. Introduction In Arabic, particularly in dialects, cases of verb grammaticalization are highly common. The grammaticalized verbs are used to express temporal, aspectual and modal values, such features often characterizing more or less extended dialectal areas. This chapter aims to study the grammaticalization of a verb which originally had the meaning ‘to see’. This meaning is preserved in literary Arabic (where its pattern is raɉā) and in a number of modern dialects, but the word most commonly used in this sense is now šāf. The issues raised when a verb is replaced by another have played an important role in the study of the formation and history of modern dialects.1 However, in parallel, there are also issues raised by the uses of grammaticalized a(a). When the fixed form of the verb raɉā is followed by a pronoun affix, it serves, among other functions, to introduce clauses. In his pioneer work on demonstratives, Fischer (1959) suggests that uses of rā (fixed imperative form) were parallel to the uses of particles stemming 1.

For Ferguson (1959: 629), this is one of the fourteen characteristic features of the military koiné, from which all the urban dialects are said to have derived. However Cohen believes that this feature can absolutely not be related to the koiné, given the persistence of raɉā in Andalusian, Maltese and Jewish dialects. The prevalence of šāf in the Bedouin dialects even leads Cohen to an entirely different conclusion: “L’hypothèse peut donc être faite que ce verbe est d’origine post-hilālienne et que son expansion est encore en cours, de façon particulièrement lente dans les parlers juifs qui sont les moins bédouinisés.” (1962: 141). [One may therefore posit that this verb has postHilalian origins and that its expansion is still ongoing, this expansion being extremely slow in the Jewish varieties which are the least Bedouinized dialects.] On the verb šāf ‘to see’, its geographic distribution and its etymology, see also respectively Kaye (1976: 157 and sq., 1986) and Bloch (1993).

122

Catherine Taine-Cheikh

from demonstratives derived from hā or haiʢ. He underlines that this evolution, facilitated by the adoption of the verb šāf for ‘to see’, was more particularly characteristic of North African dialects. A few years later, when Marçais described the western dialects in E. I., he identified the presentative ā as one of the typical traits of those languages: [...] le Maghrib, du Maroc à la Libye, utilise en propre un présentatif de l’idée verbale qui associe l’impératif du verbe ‘voir’, ā, aux affixes personnels, avec le sens de ‘me voici, te voici, etc.’ ou ‘voici que je, tu, etc.’, ānī, āk, āh, āha (ou āhi), āna, ākum, āhum, pour constater la réalité d’un état ou d’une action, dans le présent ou dans le passé, aussi bien devant un verbe à l’accompli ou l’inaccompli), āni jīt ‘voici que je suis venu’, āh yebkī ‘le voilà qui pleure’, qu’en proposition nominale, āk m ī# ‘(c’est que) tu es malade’, āhum l-temm ‘les voilà là-bas’ (1975: 599). ‘Maghrebian languages, from Morocco to Libya, are alone in using a presentative of the verbal idea which combines the imperative of the verb ā ‘to see’ and a person affix, with the meaning ‘here I am, here you are, etc.’ or ‘so here I am, so here you are, etc.’, ānī, āk, āh, āha (or āhi), āna, ākum, āhum, to denote the reality of a state or action, in either the present or the past, preceding a verb in either the perfective or the imperfective. āni jīt ‘Here I am, I have come’, āh yebkī ‘there he goes crying’, or a noun clause āk m ī# ‘(it is that) you are sick’, āhum l-temm ‘there they are over there’.’

While generalization of n- to mark the first-person singular (through analogy with the plural marker n-) can be found in all North African dialects, other common traits are rare. If the grammaticalization of ra(a) is an African or at least a Maghrebian specificity, it is important to know if this trait is really shared by all Maghrebian dialects and only by them. We will also question if the semantics are always identical from one dialect to the next, or if, as Cohen affirms (1924: 90): “L’emploi de ā est délicat; il demande à être étudié pour chaque parler”.2

2. ra(a) in the Maghreb ra(a) (or a(a) with the emphatic consonant r) is sometimes used on its own,3 but is more usually followed by a pronoun affix – either a clitic pro2.

The use of ā is complex; it has to be studied for each individual dialect.

Grammaticalized uses of the verb Ja(a) in Arabic

123

noun or, in the 3rd person, the abbreviated form of the independent personal pronoun. Other relatively minor variations are found. For example, in certain dialects such as Mauritanian Kassāniyya, the nasal consonant -n-, which regularly appears in the 1SG has spread to the 1PL, the 2SG and the 2PL: i.e. 1SG ā=ni, 2M.SG ā-n=ak, 2F.SG ā-n=ək, 3M.SG ā=hu, 3F.SG ā=hi, 1PL ā-n=nä, 2PL ā-n=kum, 3PL ā=hum (following the pattern of the negative copula 3M.SG mā-hu, but 2M.SG mā-n=ak...). Among the variants identified by Fisher, there are also the significantly abbreviated forms of the Debdou Jewish dialect (M.SG rā, F.SG rī, PL rū), which seem to have been created through analogy with the demonstrative pronouns (ibid.: 190). Let us now discuss in more detail the uses of ra(a) in North African dialects.

2.1. ra(a) in Morocco It is not in the Morrocan dialects that the use of ā is thought to be the most frequent, nevertheless, it is in these dialects that the two particles ha and a have been often compared. In order to understand the specificities acquired by ā in at least part of the Moroccan dialects, the easiest place to begin is with the various possible analyses described by Fisher. Contrary to Harrell (“The two particles ha and a have approximately the same meaning ‘here is, here are...’”, 1962), Colin thought, as early as 1948, that there was a need to make a clear distinction between the particle with deictic origins and the particle with verbal origins: Alors que la particule hā- d’origine interjectionnelle, sert à la ‘monstration’ concrète, ā- sert à la ‘présentation’, plus ou moins affective, insistée: ‘voici que (actuellement, sûrement, bel et bien, etc)’ [...]. La ‘présentation’ peut porter sur le sujet ou sur le prédicat. La notion présentée peut être non seulement une définition, une qualité, une circonstance, mais aussi un procès: ce n’est donc pas exclusivement un outil de proposition nominale (1963: 6–7).4 3. 4.

As in El-Kâmma in Gabès (Marçais and Farès 1932: 222). Colin adds that this particle “est surtout courante dans les régions à substrat berbère; et c’est vraisemblablement dans cette langue qu’il faudra rechercher l’origine de son emploi (see berbère āqəl-, āqqā-, etc. [...])” (ibid.: 7) [is especially common in regions with Berber substrates; that is certainly the language where the origins of the particle’s use is to be sought] (see Berber

124

Catherine Taine-Cheikh

‘Whereas the particle hā-, which originated as an interjection, serves for concrete ‘monstration’, ā- is used for ‘presentation’, which can be more or less emotionally charged, insistant: ‘and here we (currently, certainly, indeed, etc.) have [...]’. Both subjects and predicates can be ‘presented’. The notion presented can be not only a definition, but also a quality, a circumstance, a process: thus it is not a tool restricted to noun clauses.’

“Deixis, aspect and modality, the particularities of hā- and ā- in Moroccan Arabic”, Caubet’s (1992) article defends the same analysis, but further delves into the opposite possibility, only outlined by Colin. Caubet’s conclusion well summarizes the differences: Avec la particule hā-, le point de départ est une absence; on est dans le domaine de la deixis à l’état pur, du constat: on a une identification directe au centre énonciatif, c’est une valeur de découverte [...]. Par contre, avec ā, il faut qu’il y ait d’abord une autre valeur, une position décentrée (associée au co-énonciateur) pour, dans un deuxième temps, avoir recentrage sur la bonne valeur (celle de l’énonciateur); on retrouve l’idée de ‘forçage’ pour amener l’autre sur ses positions (1992: 149). ‘With the particle hā-, the starting point is absence; this is deixis on a very pure level, on the level of stating a fact: there is direct identification at the center of the discourse, denoting discovery [...]. In contrast, with ā, there must first be some meaning, a decentralized position (associated with the hearer) so that, subsequently, the focus is brought back to the speaker’s meaning; this is a way of forcing the hearer to come around to the speaker’s own view.’

The following examples (1 to 10) are from Caubet (1993, II: 26 and sq.). Emphasis is the author’s; we have only replaced the underlining by bold face.5 We will further refer to Caubet to briefly comment on the examples. a) Example (1) illustrates focus on the predicative relation, e.g. when the previous speaker’s assertion is called into question or not well understood. The reaffirmed term is preceded by ā.

5.

āqəl-, āqqā-, etc. [...]) – an idea to which he returns in a later session of the GLECS (Colin 1966: 174). We are responsible for the morphosyntactic analysis and the English translation and transliteration of most of the examples cited here. We have minimally harmonized transcriptions, whenever there was no risk of altering the meaning.

Grammaticalized uses of the verb Ja(a) in Arabic

(1)

/məd Ahmed

ā=h ā=PR.3M.SG

hūwa he

125

əl=mūdīr DEF=director.M.SG

f=hād=əs=sāɭa, ma ɭand=ək ma t-dîri! in=DEM.PROX=DEF=time NEG with= PR.2SG NEG 2.IPFV-can.SG ‘C’est Ahmed le directeur en ce moment, tu n’y peux rien!’ [‘It is Ahmed the director right now, there is nothing you can do about it!’] b) Focus can also be on the predicative marker which, in the next two examples, is the 1SG pronoun. Here, the pronoun affix agrees with the independent pronoun: (2)

(3)

ā=ni ni hīya āna ā=PR.1SG I she ‘C’est moi la directrice!’ [‘II am the Director (F)!’] əl=mudīr-a, ā=ni ni DEF=director-F.SG ā=PR.1SG ‘La directrice, c’est moi!’ moi [‘The director (F) is me!’] me

əl=mūdīr-a! DEF=director-F.SG

āna I

hīya! she

When this type of utterance is negated, the negation is often discontinuous and generally surrounds the unit ā-PR. This is the case in the following example, where the reassertion is conveyed by the 1SG pronoun: (4)

ma

ā=ni š āna hīya əl=mūdīr-a! NEG ā=PR.1SG NEG I she DEF=director-F.SG ‘Ce n’est pas moi la directrice!’ [‘The director (F) is not me!’] me

c) The particle ā is used much more often for differenciation than for identification (Caubet 1993, II: 35). It serves to focus on localization, glossable by ‘is indeed localized in reference to X’, as in (5) and (6). Here the particle can have actualizing meaning.

126

Catherine Taine-Cheikh

(5)

ɭazzūz ā=h f=fās Azzuz ā=PR.3M.SG in=Fez ‘Azzouz est (bien) à Fès en ce moment.’ [‘Azzouz is (indeed) in Fez at the moment.’]

(6)

fäyn ā=k? ā=ni f=əs=s5a/! ā=k where ā=PR.2SG ā=PR.1SG on=DEF=terrace.SG ‘Où es-tu?’ ‘Je suis sur la terrasse!’ [‘Where are you?’ – I’m on the terrace!’]

d) With attributive relations, the particle ā is used only when the located term is clearly determined. Beyond focusing on the relation, the presence of ā can, again, have actualization meaning (Caubet ibid.: 42). (7)

ā=ni m ē#-a ā=PR.1SG sick-F.SG ‘Je suis malade.’(en ce moment, ou vraiment, ou tu ne vois pas que...) [‘I am sick (F).’ (‘right now’, or ‘really’, or ‘can’t you see that...’)]

If the adjectival predicate is fronted, ā-PR follows. (8)

ā=h bärəd, hād=əl=xobz! ā=PR.3M.SG cold.M.SG DEM.PROX=DEF=bread.M.SG ‘Il est (vraiment) rassis, ce pain!’ [‘It is (really) stale, this bread!’]

When negation is inserted between ā-PR and the located term, it places emphasis on the latter, see examples (9) and (10).6 (9)

6.

ā=ni ma_ši 5bēb-a ā=PR.1SG COP.NEG doctor-F.SG ‘Mais, je ne suis pas médecin, moi!’ moi [‘But I am not a doctor, me!’] me

With discontinuous negation surrounding ā-PR, as in (4), the focus is on the relationship (Caubet ibid.: 45).

Grammaticalized uses of the verb Ja(a) in Arabic

127

(10) /məd ā=h ma_ši fəllā/! Ahmed ā=PR.3M.SG COP.NEG farmer-M.SG ‘Ahmed Ahmed, Ahmed il n’est pas paysan! Il n’est pas paysan, Ahmed!’ Ahmed [‘Ahmed Ahmed, he is not a farmer! He is not a farmer, Ahmed!’] Ahmed e) The above examples are all utterances with a non-verbal predicate. However, ā is also used with verbal predicates. In such cases, “[t]he use of ā is especially common when a sort of logical connexion or consequent result is implied”, as underlined by Harrell who gives the following example (see Caubet 1992: 143): (11) xud 5aksi u ā=h iwə99:=ək take.IMP.SG taxi and ā=PR.3M.SG 3.IPRV.get.SG=PR.2SG ‘Take a taxi and he’ll get you (SG) there.’ A similar example was found by Fischer (ibid.: 189) in Colin (1939: 13): (12) u=5:əq əs=slūgi, ā huuʢwa ɭauʢwām and=release.IMP.SG DEF=dog.M.SG ā he swimmer.M.SG ‘und mach dann den Hund los, denn dieser ist ein Schwimmer.’ ‘and then release the dog, because this one is a swimmer.’ f) Based on this connective use of ā, Fischer (ibid.: 193) links it to the use of ā ā ‘come one! give!’. Although the origins of these two forms are probably distinct, in certain Maghrebian dialects, especially Moroccan, ā ā is highly charged emotionally, a trait which makes it almost equivalent to what the particle ā is used to express elsewhere. Thus, in Tangiers (Marçais 1911: 220, 305) a a is very commonly used (with specific suffixes: a a (1) ALLATIVE (2) PASSIVE.

154

Catherine Taine-Cheikh

Their meanings also show differences, which might partially be explained – albeit only partially – by the probable mixed origins of (a)raɭ and ta(a)ri (origins contaminated for the former by the root rāɭa ‘watch, guard’, and for the latter by the root ɉaar ‘trace (of)’). Whereas (a)raɭ, and more particularly šaɭ, are essentially (or exclusively) presentatives, ta(a)ri and (ya) rayt serve mostly as discourse particles. ra(a) and tara(a) tend to be used in very similar ways. In some dialects their use is widespread, including as simple copulas. In others, they only serve a limited number of the possible uses, and are usually used as presentatives or as modals, leaving one of the other series to cover the other uses (see e.g. the case of ā vs. āɭi in Kassāniyya, or tara vs. arɭa in the Negev). Given that ra(a) is found throughout the Maghreb (and only there), one could consider it a characteristic trait of Northern African Arabic. However, the diversity of the trait’s manifestations in the Maghrebian dialects on the one hand, and the similarities between Maghrebian ra(a) and ‘MiddleEastern’ tara(a) on the other hand, show that this trait is far from useful for purposes of distinction. The current study, which builds on previous studies and would certainly benefit further from more in-depth studies on specific points, highlights not a faultline, but rather several continua (in terms of both form and meaning). Some convergences demonstrate kinship among geographically contiguous areas whereas others (such as the use of raɭ) can apparently only be explained through much older kinships of a socio-historical nature.

References Altoma, Salih J. 1969 The Problem of Diglossia in Arabic: a Comparative Study of Classical and Iraqi Arabic. London: Oxford University Press. Badawi, El-Said, and Martin Hinds 1986 A Dictionary of Egyptian Arabic. Arabic-English. Beirut: Librairie du Liban. Bauer, Leonhard 1926 Reprint. Das Palästinische Arabisch. Die Dialekte des Städters und des Fellachen. Leipzig: Hinrichs. Original edition, 1910. Bettini, Lidia 2006 Contes féminins de la Haute Jézireh syrienne. Matériaux ethnolinguistiques d’un parler nomade oriental. Firenze: Università di Firenze. Blau, Joshua 1960 Syntax des Palästinenischen Bauerndialekts von Bīr Zēt. Auf grund der “Volkserzählungen aus Palästina” von Hans Schmidt und Paul Kahle. Walldorf-Hessen: Verlag für Orientkunde Dr. H. Vordran. Bloch, Ariel 1993 Verbs of topographical elevation, the case of šāf ‘to see’ in colloquial Arabic. Zeitschrift für Arabische Linguistik 25: 100–107. Borg, Alexander 2004 Comparative Glossary of Cypriot Maronite Arabic (Arabic-English). With an Introductory Essay. Leiden/Boston: Brill. Boris, Gilbert 1958 Lexique du parler arabe des Marazig. Paris: Klincksieck. Boucherit, Aziza 2002 L’arabe parlé à Alger. Aspects sociolinguistiques et énonciatifs. Paris/Louvain: Peeters. Cantineau, Jean 1936-37 Etudes sur quelques parlers de nomades arabes d’Orient. I. AIEO 2: 1–118. II. AIEO 3: 119–237. Caubet, Dominique 1992 Deixis, aspect et modalité, les particularités hā- et ā- en arabe marocain. In La deixis, Marie-Annick Morel, and Laurent Danon-Boileau (eds.), 139–49. Paris: P.U.F. 1993 L’arabe marocain. Louvain: Peeters. Cohen, David 1962 ‘Koinè’, langues communes et dialectes arabes. Arabica IX/2: 119–144. 1963 Le dialecte arabe kassānīya de Mauritanie. Paris: Klincksieck.

156

Catherine Taine-Cheikh

La phrase nominale et l’évolution du système verbal en sémitique. Etudes de syntaxe historique. Paris: Société de Linguistique de Paris. 1989 L’aspect verbal. Paris: Presse Universitaire de France. Cohen, Marcel 1924 Le système verbal sémitique et l’expression du temps. Paris: Leroux. Colin, Georges S. 1939 Chrestomatie marocaine. Textes citadins en transcription latine. Paris: Maisonneuve. 1963 Quelques calques syntaxiques et sémantiques sur le berbère dans les parlers arabes du Maghreb (séance du 22 décembre 1948). Comptes rendus du GLECS (1960-1963, i.e. 1948-1951) V: 4–8. 1966 La proposition nominale en arabe marocain (séance du 11 mai 1966). Comptes rendus du GLECS (1963-1966) X: 173–175. Cowell, Mark W. 1964 A Reference Grammar of Syrian Arabic. Georgetown: Georgetown University Press. De Jong, Rudolf 2000 A Grammar of the Bedouin Dialects of the Northern Sinai Littoral: Bridging the linguistic gap between the eastern and Western Arab world. Leiden: Brill. Driver, Godfrey R. 1925 A Grammar of the colloquial Arabic of Syria and Palestine. London: Probsthain. Ferguson, Charles 1959 The Arabic Koinè. Language 35(4): 616–630. Fischer, Wolfdietrich 1959 Die demonstrativen Bildungen der neuarabischen Dialekte: Ein Beitrag zur historischen Grammatik des Arabischen. s’-Gravenhage: Mouton de Gruyter and Co. Grand’Henry, Jacques 1995 Conservation and change in Algerian Bedouin Arabic. In Proceedings or the 2nd International Conference of L’Association Internationale pour la Dialectologie Arabe, Joe Cremona, Clive Holes, and Geoffrey Khan (eds.), 51–57. Trinity Hall (University of Cambridge), 10–14 sept. 1995. Harrell, Richard S. 1962 A Short Reference Grammar of Moroccan Arabic. Georgetown: Georgetown University Press. Heine, Bernd, and Tania Kuteva 2002 World Lexicon of Grammaticalization. Cambridge: Cambridge University Press. 1984

Grammaticalized uses of the verb Ja(a) in Arabic

157

Henkin, Roni 2010 Negev Arabic. Dialectal, Sociolinguistic, and Stylistic Variation. Wiesbaden: Harrassowitz. Hillelson, Sigmar 1929 Songs of the Baggara. Sudan Notes and Records 12: 73–83. Holes, Clive 2001 Dialect, Culture and Society in Eastern Arabia. Volume 1. Glossary. Leiden: Brill. Jullien de Pommerol, Patrice 2000 Dictionnaire de l’arabe tchadien. Paris: Karthala. Kaye, Alan S. 1976 Chadian and Sudanese Arabic in the light of comparative Arabic dialectology. The Hague/Paris: Mouton de Gruyter. Kaye, Alan S. 1986 The Verb ‘See’ in Arabic Dialects. In The Fergusonian Impact: In Honor of Charles A. Ferguson on the Occasion of His 65th Birthday, I: From Phonology to Society; II: Sociolinguistics and the Sociology of Language, Joshua A. Fishman, Andree Tabouret-Keller, Michael Clyne, Bhadriraju Krishnamurti, and Mohamed Abdulaziz (eds.), 211–222. Berlin: Mouton de Gruyter. Khan, Geoffrey 2008 Presentatives. In Encyclopedia of Arabic Language and Linguistics, Versteegh (ed.), 703–705. Leiden: Brill. Landberg, comte Carlo de 1909 Etude sur les dialectes de l’Arabie Méridionale. II. Daaînah (2). Leiden: Brill. Lethem, Gordon J. 1920 Colloquial Arabic, Shuwa dialect of Bornu, Nigeria and of the region of Lake Tchad. London: The Crown Agents for the Colonies. Manfredi, Stefano 2010 A Grammatical Description of Kordofanian Baggara Arabic. PhD thesis, Università degli Studi di Napoli “L’Orientale”. Marçais, Philippe 1956 Le parler arabe de Djidjelli (Nord Constantinois, Algérie). Paris: Maisonneuve. Marçais, Philippe 1975 ɭArabiyya III. – Les dialectes arabes. 3. – Les dialectes occidentaux. Encyclopédie de l’Islam (2nd ed.) 597–601. Leiden/Paris: Brill/ Maisonneuve/Larose. Marçais, William 1911 Textes arabes de Tanger. Paris: Leroux.

158

Catherine Taine-Cheikh

Marçais, William, and Farès Jelloûli 1932 Trois textes arabes d’El-Kâmma de Gabès (suite). Journal Asiatique CCXXI (1932 oct.-déc.), 193–269. Marçais, William, and Abderrahmân Guiga 1958-61 Textes arabes de Takroûna, II. Glossaire. Paris: Geuthner. Owens, Jonathan 1984 A short reference grammar of Eastern Libyan Arabic. Wiesbaden: Harrassowitz. Palva, Heikki 1980 Characteristics of the Arabic Dialect of the Bani faxar Tribe. Orientalia Suecana 29: 112–139. Palva, Heikki 1992 Artistic Colloquial Arabic. Traditional narratives and poems from alBalqâɭ (Jordan): transcription, translation, linguistic and metrical analysis. Studia Orientalia 69: 191 p. Pereira, Christophe 2010 Le parler arabe de Tripoli (Libye). Zaragoza: Instituto de estudios Islámicos. Estudios de Dialectología Árabe. Reinhardt, Carl 1972 Reprint. Ein arabischer Dialekt gesprochen in ɭOmān und Zanzibar. Amsterdam: Philo Press. Original edition, 1894, Stuttgart/Berlin: Spemann. Rhodokanakis, Nikolaus 1908 Der vulgärabische Dialekt im Dʟofâr (xfâr). I. Prosaische und poetische Texte. Wien: Hölder. Sowayan, Saad Abdullah 1992 The Arabian Historical Narrative. An Ethnographic and Linguistic Analysis. Wiesbaden: Harrassowitz. Stewart, Frank H. 1990 Texts in Sinai Bedouin Law, Part 2. Texts in Arabic. Wiesbaden: Harrassowitz. Stowasser, Karl, and Mokhtar Ani 1964 A Dictionary of Syrian Arabic: English-Arabic. Georgetown: Georgetown University Press. Stumme, Hans 1896 Grammatik des Tunisischen Arabisch nebst Glossar. Leipzig: Hinrichs. Stumme, Hans 1898 Märchen und Gedichte aus der Stadt Tripolis in Nordafrika. Leipzig: Hinrichs. Taine-Cheikh, Catherine 1988-98 Dictionnaire Hassaniyya-Français. 8 volumes. Paris: Geuthner.

Grammaticalized uses of the verb Ja(a) in Arabic

159

Talay, Shabo 2003 Der arabische Dialekt der Khawētna. Teil II: Texte und Glossar. Wiesbaden: Harrassowitz. Tauzin, Aline 1993 Contes arabes de Mauritanie. Paris: Karthala. Trimingham, J. Spencer 1946 Sudan colloquial Arabic. London: Oxford University Press. Vanhove, Martine 1995 Les particules qad et raɭ dans un dialecte arabe de la région de Yâfiɭ (Yémen). In Proceedings or the 2nd International Conference of L’Association Internationale pour la Dialectologie Arabe, Joe Cremona, Clive Holes, and Geoffrey Khan (eds.), 243–52. Trinity Hall (University of Cambridge). 2010 Deixis, information structure and clause linkage in Yafiɭ Arabic (Yemen). In Clause Linking and Clause Hierarchy. Syntax and pragmatics, Isabelle Bril (ed.), 333–354. Amsterdam/Philadelphia: Benjamins. Woidich, Manfred 2006 Das Kairenisch-Arabische. Eine Grammatik. Wiesbaden: Harrassowitz. Woodhead, Daniel R., and Wayne Beene 1967 A Dictionary of Iraqi Arabic: Arabic-English. Georgetown: Georgetown University Press.

Some new information information about Bongor Arabic Xavier Luffin 1. Introduction Bongor Arabic (BA) is locally called árab haná bóngor, túrku or túrgu, though the two last glossonyms seem to be less used than the first. It is a pidginized form of Arabic spoken today in South Western Chad, in the Mayo-Kebbi area, more specifically in Bongor, a city which is close to the border with Cameroon. Bongor Arabic is used as a lingua franca mainly between the Masa (or Masana) and the Tupuri (two populations who speak respectively a Chadic language and a Niger-Congo language) on the one hand, and the Arabic speaking traders coming from the North on the other hand. There is no data concerning the number of speakers. A pidginized form of Arabic, called Turku (Tourkou) was already in use in Southern Chad and even in Nigeria and what is now Central African Republic in the last decades of the 19th century. In the beginning of the 20th century, Derendiger (1912) says that various forms of Arabic are spoken in Chad, including a pidginized form. Turku was mentioned by Lethem (1920), describing a form of Arabic spoken in Bornu, Nigeria. He attributed its origin to Rabeh, a Nubian trader who established himself in Chad in 1879 (Lethem 1920). This language has been described by a French officer, Muraz (1930), together with two Sara dialects, in a booklet already analyzed by Tosco and Owens (1993). Not mentioning the name of the language itself and the area where it was spoken, many parallels may be observed in the phonology, the morphosyntax and the lexicon of the two languages. I have already published a general description of Bongor Arabic in the Encyclopedia of Arabic Languages and Linguistics (Luffin 2008: 634– 639). At that time, my main source of information was a series of recordings made in situ in 2004 by our colleague Madeleine Somte, a Chadian linguist, as well as a series of elicited data given by a Chadian informant, Mahamat Litassou. In 2009, Miss Somte made a series of new recordings in Bongor, and she kindly gave me the possibility to use them in order to complete my survey. This new information has enabled me to enrich my description of Bongor Arabic with new points and new examples that I will examine in this chapter.

162

Xavier Luffin

2. BA as a pidgin 2.1. Main features A pidgin language is usually seen as a speech-form which does not have native speakers (the main feature distinguishing a Pidgin and a Creole), and is therefore primarily used as a means of communication among people who do not share a common language. The pidgin is a simplified form of language compared to the lexifier language, regarding especially the morphology and the lexicon. Pidgins are born and used in specific social situations: interethnic contacts, contacts among traders, contacts between colonial people and local workers (Arends1994). The lexifier of BA is obviously Chadian Dialectal Arabic (CDA), seen as a dialectal group, not as one specific variety of Arabic spoken in Chad, since there is more than one Arabic dialect in the country (Jullien 1997, 2006). Chadian Arabic itself belongs to Western Sudanese Arabic (Kaye 1976). The main features that permit us to see BA as a pidginized form of CDA are the following: i. the long vowels are mainly replaced by stressed vowels. ii. some phonemes tend to disappear (kh > k in kalí ‘to let’) or to be modified (f > p in pi ‘in, to’). iii. lack of gender distinction. iv. lack of derivational morphology. v. the article al- is replaced by the demonstrative da and dol. vi. the genitive is almost always analytic. vii. the inflection of the verb is marked by the independent pronoun alone. viii. the negation particle stands at the end of the sentence.

2.2. BA compared with other Arabic Pidgins and Creoles At least two other Arabic Pidgins and Creoles are spoken nowadays in Africa: Juba Arabic (JA), which is spoken in Southern Sudan, and Ki-Nubi (KN), which is spoken in Kenya and Uganda. I present here the main common features and differences between BA on the one hand, and JA and KN on the other hand:

Some new information about Bongor Arabic

163

Common features: i. Many phonological processes are alike. ii. There is no consonant gemination. iii. Long vowels are replaced by stress. iv. The 2nd person plural independent pronoun is built on the singular. v. Nominal and verbal derivational morphology is very poor. vi. There is no gender distinction. vii. The definite article has disappeared and is replaced by demonstrative pronouns. viii. Verbal inflection is marked by TMA markers. ix. The negation stands at the end of the sentence (like in Ugandan KN, not in Kenyan KN). x. A wide part of the lexicon is common. xi. The verbal roots are mainly built on the imperative forms of the lexifier. Some of these features may be explained by the fact that the lexifiers of BA and the other Arabic Pidgins and Creoles belong to the same group, which is the Sudanese Chadian Arabic Dialects. Other features cannot be explained by this common origin, but are found in many other processes of pidginization or creolization, like the lack of gender distinctions and the use of TMA markers. But BA also shows differences with KN and JA. The main differences are: i. some specific phonological processes, probably linked to the substratum, like f > p and t > d. ii. TMA Markers: where JA and KN use 0, gi- and bi- as main TMA markers, BA use 0 and gay-. iii. The lexicon. Many lexical items are different, showing another dialectal lexifier: ána (BA), bitá (JA) and tá (KN) ‘of’, orú and kélem (KN, JA) ‘say’, yátu (BA) and munú ‘who’(KN, JA), almé (BA) and móyo (KN, JA) ‘water’, antú (BA) and wedí (KN, JA) ‘give’, dár (BA) and béle (KN, JA) ‘country’. iv. BA does not use abú as relater. v. The adstrate. While BA lexicon is influenced by CDA and by French, JA is influenced by English and Sudanese Arabic, KN by Swahili and English.

164

Xavier Luffin

3. Linguistic description 3.1. Phonology 3.1.1. Consonants labials plosives

voiceless voiced fricatives voiceless voiced nasals vibrants laterals semivowels

p b f

dentals palatals velars t d s z n r l

m

w

č č š

k g

ny

ng

glottals

h

y

There is no consonant gemination. Some speakers tend to pronounce f > p, t > d.

3.1.2. Vowels i

u e

o a

These vowels may be long too, but the vowel length is not distinctive.

3.1.3. The syllable The patterns of the syllable are: v, vC, Cv and CvC, like in á-na ‘I’, al ‘that, who’ (relative pronoun), dá ‘this’ and dár ‘country’.

Some new information about Bongor Arabic

165

3.1.4. Accent There is a tonic accent (which in most cases corresponds to the long vowel or the stressed vowel in the lexifier: kalám ‘word’, dérib ‘way’), which may be distinctive for some words, like ána ‘I’ and aná ‘of’.

3.2. Morphology 3.2.1. Pronouns BA distinguishes between independent and suffix pronouns, though the pronouns of the 3rd person are similar:

1st sg. 2nd sg. 3rd sg. 1st sg. 2nd sg. 3rd sg.

Independent pronouns

Suffix pronouns

ána ínti/índi hú anína índukum (h)úman

-(yi) -k(i) -hú -na -kú(m) -úman

The independent pronouns are used in four cases: i. Subject of nominal sentences: (1)

hú wilét ngambáy 3SG child Ngambay ‘He is a Ngambay.’

ii. Subject of verbal sentences: (2)

ána orú le 1SG say to ‘I told him.’

hu 3PL

166

Xavier Luffin

iii. Objects of verbs: (3)

ána kúlu dóru orí 1SG too want say ‘Me too I want to tell you.’

ínti 2SG

iv. After a preposition: (4)

hú god ma ána fi Bongor 3SG stay with 1SG in Bongor ‘He stayed with me in Bongor.’

The suffix pronouns are used after the possessive marker aná and the preposition le: (5)

usm aná y Kaltúma name GEN 1SG Kaltuma ‘My name is Kaltuma.’

(6)

anína gáy só lé 1PL TMA do for ‘We will do this for you.’

ku 2PL

sókol thing

da DEM

The relative pronoun has a unique form, al: (7)

sána al gáy ğá dá ána gáy mási France year REL TMA come DEM 1SG TMA go France ‘Next year I’ll go to France.’

Demonstrative pronouns are da (sg.) and dólda (pl.): (8)

dá gum gáy kóre DEM rise stay cry ‘Then he started to cry.’

Some new information about Bongor Arabic

167

They are used as definite markers too: (9)

Sára dólda Sara DEM.PL ‘the Sara’

3.2.2. Nouns and adjectives There is no productive marker of word classes, the distribution of the words being their main indication. a. Gender There is no morphological gender distinction: (10) uléd sakár boy young ‘a young boy’ (11) bináya sakár Girl young ‘a young girl’ b. Number There is no dual; the suffix –én is still present in the numerals, but is not productive: tinén ‘two’, mitén ‘two hundred’. The majority of nouns have no morphological plural, the plurality being marked by the demonstrative dólda or by quantifiers like katir ‘many’ or kul ‘all’: (12) mer dólda mayor DEM.PL ‘the mayors’, (13) Sára dólda Sara DEM.PL ‘the Sara’

168

Xavier Luffin

(14) kalám katír katir thing many.2 ‘(many) questions’ However, a few nouns may bear the suffix –in as a plural marker: malán, malan-ín ‘full’, wáy, wad-ín ‘some’, gáid, gaid-ín ‘staying’. Some other plural markers may occur, though less commonly: internal plural like ráğil ‘man’, ruğál ‘men’, kebír, kubár ‘big, and suppletive plural, like nadum ‘person’, nas ‘people’, mára ‘woman’ awín ‘women’, wilét ‘boy’, iyál ‘boys’.

3.2.3. Numerals Etymologically, all the numerals (e.g. wáy, tinén, taláta, árba, kámsa) are related to the Arabic lexicon. However, the unity always follows the ten – a feature which also appears in other varieties of Chadian Arabic: ásara u kámsa ‘fifteen’. The speaker may switch to French, especially to express high numbers, like trente mille, quarante mille ‘thirty thousand, forty thousand’, though this code-switching occurs even for small numbers: (15) húman gáy nantiní sáğan áskar quinze ans. 3PL TMA give pension army fifteen years ‘They give me a military pension (after) fifteen years.’

3.2.4. The verb a. Verbal form All the verbs have a unique form, which is the verbal stem. There is no derivational morphology, except for some compound verbs with só ‘to do’, like só sarági ‘to steal’, só súk ‘to trade’. The noun associated with the verb may be a borrowing from French, like in só mer ‘to become a mayor’ or só mersí ‘to thank’. Some verbs, when they precede another verb, acquire a new meaning or function, like a TMA marker or an inchoative verb. For instance, gáy ‘to stay’ is used as a TMA marker:

Some new information about Bongor Arabic

169

(16) ínti gáy fi Bongor 2SG stay in Bongor ‘You live in Bongor.’ (17) ambákir dá ána gáy mási tomorrow DEM 1SG TMA go ‘Tomorrow I’ll go to the market.’

súk market

Gum ‘to rise’ may be used as an inchoative verb, meaning ‘start to’: (18) nas gum kúrbu úman People start take 3PL ‘People started to them.’ (19) dá gum gáy kóre DEM rise stay cry ‘Then he started to cry.’ Kalási ‘to end, to finish’ is also used to emphasize the perfect (see below): (20) da kalási DEM end ‘It was over.’ b. Inflection The perfect is marked by the use of the non-inflected verb stem: hú mási ‘he went’, while the imperfect is marked by the prefix gáy: (21) índukum gáy mási 2PL TMA go ‘Where are you going?’

wén ? where

The marker kalás may reinforce the perfect: (22) kalás úman máso TMA 3PL go ‘They are gone.’

170

Xavier Luffin

All markers are in preverbal position. The perfect 1st sg. 2nd sg. 3rd sg. 1st sg. 2nd sg. 3rd sg.

ána mási índi mási hú mási anína mási índukum mási úman mási

The imperfect 1st sg. 2nd sg. 3rd sg. 1st sg. 2nd sg. 3rd sg.

ána gáy-mási índi gáy-mási hú gáy-mási anína gáy-mási índukum gáy-mási úman gáy-mási

From a tense perspective, the non-inflected verb mainly renders the past, while gáy may be used for the present as well as for the future: (23) ána gáy gatá kalám fi kásm aná ki mafí 1SG TMA cut speech in mouth GEN 2SG NEG ‘I don’t interrupt you.’ (24) ambákir dá ána gáy mási tomorrow DEM 1SG TMA go ‘Tomorrow, I’ll go to the market.’ (25) sána al gáy ğá dá ána year REL TMA come DEM 1SG ‘Next year, I’ll go to France.’

súk market

gáy mási France TMA go France

However, gáy may also occur to render the duration of the action, even in the past:

Some new information about Bongor Arabic

171

(26) human gáy dowúsu híni 3PL TMA fight here ‘They were fighting here.’ Stative verbs may be used without gáy, even in the present: (27) ána dóru árge 1SG want alcohol ‘I want (to drink) alcohol.’ (28) ána dorú máši fi dar aná úman máfi 1SG want go in land GEN 3PL NEG ‘I don’t want to go in their country.’ (29) be kalám árab dá With word Arabic DEM ‘You don’t know it in Arabic?’

índi 2SG

árpu máfi? know NEG

However, stative verbs may also occur with a TMA: (30) índi gáy árfu Bongor máfi? 2SG TMA know Bongor NEG ‘Don’t you know Bongor?’ The imperative The imperative is rendered by the use of the verbal root alone: orí ‘say’, asérbi ‘drink’. For some verbs, there is a suppletive form of imperative: ámsi ‘go’ (mási ‘to go’). However, the verb is often preceded by the 2nd person pronoun: (31) kalám da índi orú ána ‘Explain me the problem.’ (32) sopér, índi ágip ‘Driver, stop!’

172

Xavier Luffin

The subjunctive The verb introduced by an auxiliary is always non-inflected: (33) almé aná ku al gáy ğá kasurú bét aná na water Gen 2PL REL TMA come destroy house GEN 1PL ‘Your water which comes to destroy our houses.’ (34) ána gáy dorú 1SG TMA want ‘I want to say this.’

orú say

kalám dá thing DEM

3.3. Syntax 3.3.1. Noun phrase a. Definiteness The demonstrative pronouns are used as definite markers: (35) dérib da road DEM ‘the road’ (36) journaliste dolda journalist DEM.PL ‘the journalists’. Indefiniteness may be marked by the omission of the definite marker, as in: (37) hú gáy adulú lé na dérib 3SG TMA make for 1PL road ‘He will build a road for us.’ or by the use of the numeral wáy ‘one’, as in:

Some new information about Bongor Arabic

173

(38) mára wáy woman one ‘a woman’ (39) ána gáy orú kalám 1SG TMA say thing ‘I will say something.’

wáy one

b. Modifiers The noun modifier – demonstrative/definite marker, adjective, numeral – always stands after the noun: (40) iyál da children DEM ‘the kids’ or ‘these kids’ (41) bináya sakár girl small ‘a young girl’ (42) ruğál tinén men two ‘two men’. c. Number agreement Since many words have no specific plural marker, agreement is not systematic: (43) bakán iširín place twenty ‘twenty places’ (44) anína ma masás 1PL NEG witch ‘We are no witches.’

174

Xavier Luffin

However, we also find: (45) ruğál tinén men two ‘two men’ (46) bakán wad-ín place some ‘some places’. d. Genitive The genitive marker is aná. It may be followed by a suffixed pronoun: (47) hasáy aná stick GEN ‘my stick’

y 1SG

(48) kalám aná word GEN ‘your speech’

ki 2SG

(49) bet aná house GEN ‘his house’

hu 3SG

or by a noun: (50) kalám aná depité dá speech GEN deputy DEM ‘the deputy’s speech’ (51) nádum aná Sar man GEN Sarh ‘a man from Sarh’ (52) gúrs aná petról money GEN oil ‘the money of the oil’.

Some new information about Bongor Arabic

175

The use of a synthetic genitive may also occur, but is very rare and often ren-ders specific expressions or compounds: (53) kalám nasára speech European ‘French’ (54) kalám árab speech Arab ‘Arabic’ (55) kášam bár mouth river ‘river bank’ The synthetic genitive is also used with the word wilét ‘child’ or its plural form iyál, ‘children’, in order to name nationality: (56) wilét ngambáy child Ngambay ‘a Ngambay’ (57) wilét Banana child Banana ‘a Banana’ (58) iyál tšad children Chad ‘the Chadians’ However, it seems to be used in combination with the analytical genitive structure, since we also find for instance: (59) wilét aná child GEN ‘a Ngambay’

ngambáy Ngambay

176

Xavier Luffin

e. Comparison There is no elative form; comparison is marked by fútu after the adjective: (60) Mamát dá tawíl fútu Ali Mamat DEM big ELAT Ali ‘Mamat is taller than Ali’. f. Relative clauses The relative clause is introduced by al: (61) hu kalám al fí gélb aná k almé 3SG thing REL in heart GEN 2SG water ‘This is the thing that you keep in your heart.’ (62) aná ku al gáy ğá kasurú bét aná GEN 2PL REL TMA come destroy house GEN ‘Your water which comes to destroy our houses.’

na 1PL

g. Reduplication Reduplication may be used to emphasize the meaning of an adjective or an adverb: (63) nas kubár kubár people big PL2 ‘very important persons’ (64) lámpa dólda al fóg fóg da lamp DEM.PL REL above.2 DEM ‘the lamps which are above’ (65) kalám katír katir word many.2 ‘(many) questions’.

Some new information about Bongor Arabic

177

3.3.2. The verb phrase a. Phrase order The phrase order is SVO: (66) ána súfu sókol I see thing ‘I saw this.’

dá DEM

(67) hú gáy adulú bét aná 3SG TMA make house GEN ‘He builds his house.’

hú 3SG

However, topicalization is quite common: (68) dérib dá bes anína dóru road DEM just 1PL want ‘We want this road.’ (Literally: ‘This road, we want it.’) b. Indirect discourse After the verbs gál and orú, orí ‘to say’, the particle ké is often used: (69) índukúm bas gal ké potról pí dár aná 2PL just say that oil in country GEN ‘Now you say that there is oil in your country.’

kum EPL

However, one can also find the same verbs without this particle: (70) hú gal gomáš ana hú gáy kásar 3SG say shirt GEN 3SG TMA break ‘He said that his shirt will be torn.’ c. Existential and possessive sentences Existential sentences are usually marked by the use of pi or fi (mapí or mafí in a negative sentence):

178

Xavier Luffin

(71) hopital fi fi Bongor hospital EX in Bongor ‘There is a hospital in Bongor.’ (72) dérbe máfi road NEG ‘There is no road.’ But fi may also be omitted: (73) índukúm bas gal ké potról pí dár aná kum 2PL just say that oil in country GEN EPL ‘You are saying that there is oil in (the soil of) your country’ The possessive sentence may be rendered by various expressions: i. Possessor + gáy me (literally ‘to be with’) + Possessed, like in: (74) ána gáy be wadír ğedíd 1SG TMA with car new ‘I have a new car.’ ii. Possessor + me + Possessed, like in: (75) ána be akú aná 1SG with brother GEN ‘I have two brothers.’

y tinén 1SG two

iii. Possessor + Possessed + Existential word, like in: (76) inti da livre 2SG DEM book ‘You have a book.’

way bas ana one only GEN

ki fi sé 2SG EX thing

In the last two expressions, possession is emphasized by aná + possessive pronoun.

Some new information about Bongor Arabic

179

3.3.3. Negation The marker mafí (or mapí) is used to negate both nominal and verbal sentences. It occurs usually in sentence-final position: (77) dérbe máfi road NEG ‘There is no road.’ (78) ana luti mafi 1SG stupid NEG ‘I am not stupid.’ (79) ána árifu mafí 1SG know NEG ‘I don’t know.’ If the verb is followed by an object or a complement, the negation comes after the whole sentence: (80) aná árfu maktúb máfi 1SG know letter NEG ‘I do not know how to write.’ (81) aná dorú máši fi dar aná 1SG want go in land GEN ‘I do not want to go in their country.’

úman 3PL

máfi NEG

Another marker, ma, is sometimes used. It is placed before the predicate or the verb. However, it is rare and in most cases it seems to be an influence of CDA: (82) ána ma dorú 1SG NEG want ‘I do not want to.’ It is particularly obvious in the following sentence, when the informant starts to speak in Chadian Arabic and then comes back to BA, using the two different negations:

180

Xavier Luffin

(83) wilét Banana sowá ké ké ma gídir, ké ké bágdar máfi child Banana do so so NEG can so so can NEG ‘The Banana moved so and so, but he couldn’t, he couldn’t (escape them).’ We also find the expression ma... batán to express ‘not anymore’: (84) ána má uléd batán 1SG NEG kid anymore ‘I am not a child anymore.’ The word batán means ‘again’ when it is used without má: (85) gomáš aná y gáy kásar batán shirt GEN 1SG TMA break again ‘My shirt will be torn again.’

3.3.4. Conditional sentence The main word used to express a conditional sentence is kan, like in: (86) kan lámpa fí dá anína gáy só sarági mafí if lamp EX DEM 1PL TMA do stealing NEG ‘If there are lamps here, then we will not steal.’ (87) kan só sarági mafí dá gáy ligí mára mafí if do stealing NEG DEM TMA find woman NEG ‘If he doesn’t steal, then he cannot find a wife.’

3.3.5. Depidginization A BA speaker may be influenced by his knowledge of other forms of Arabic – mainly Chadian dialect(s) – which leads to a partial depidginization of his speech. Here are some examples found in data from different informants, who use conjugated verbs or synthetic genitive:

Some new information about Bongor Arabic

(88) anína rikíb na wotír 1PL take 1PL car ‘We took the car together.’

dá DEF

181

sáwa together

(89) ána kúlu gáy ni dóra ni orí 1SG too TMA 1SG want 1SG sai ‘Me too, I would like to say something.’

séi thing

(90) ána ma n árfa 1SG NEG 1SG know ‘I don’t know.’ (91) índi gáy árifu úsum 2SG TMA know name ‘You know my name.’

i my

(92) ána gamét díkta 1SG start 1SG laugh ‘I started to laugh.’ Sometimes, the informant understands that he is supposed to speak in BA, not in Chadian Arabic, so he repeats the same sentence in both dialectal varieties, like in: (93) dá gáma gáy yébki, dá gum gáy kóra DEM start TMA cry 3SG DEM start TMA cry ‘Then he started to cry, and then he started to cry.’

3.4. Lexicon The lexicon is clearly derived from the vocabulary of CDA: e.g. kikéf ‘how’, orú ‘say’, ana ‘of’, dóru ‘want’, yátu ‘who’, almé ‘water’, dár ‘land, country’, hártu ‘to cultivate’. None of these words are found in the other Arabic Pidgins and Creoles spoken in Africa. However, many loanwords come from French, which is the official language of Chad, a phenomenon which can be observed in Chadian Arabic as well. They can be nonintegrated borrowings, keeping their original phonology, like maire ‘mayor’, député ‘deputy’ and infirmier ‘nurse’. At a second level, some of

182

Xavier Luffin

these loanwords are phonologically integrated: sofér for chauffeur ‘driver’, presdán for president ‘president’, wotír for voiture ‘car’, fantalon for pantalon ‘trouser’, boté for voter ‘to vote’, and apermé for infirmier ‘nurse’. The level of integration depends on the knowledge the informant has of French. These loanwords are mainly nouns and adjective, though one can also find some verbs, like komandé (from commander) ‘to command’, and even prepositions, like zuska (from jusqu’à), ‘when, during’: (94) ána máši komandé garağ 1SG go command garage ‘I went to work as a chief in a garage.’ (95) anína só kídma dá ké zuska anína gáy fi hinák 1PL do work DEF that when 1PL stay in there ‘We worked during the whole time we were there.’ A few other loanwords come from local languages – supposedly Massa, like bursdiya ‘monday’ and kadánga ‘axe’.

References Arends, Jacques, Pieter Muysken, and Norval Smith (eds.) 1994 Pidgins and creoles. An introduction. Amsterdam/Philadephia: Benjamins. Derendiger, Robert 1912 Notes sur le dialecte arabe du Tchad. Revue Africaine 56: 341–358. Jullien de Pommerol, Patrice 1997 L’arabe tchadien, émergence d’une langue véhiculaire. Paris: Karthala. 2006 Chad Arabic. In Encyclopedia of Arabic Language and Linguistics I, Kees Versteegh (ed.), 360–368. Leiden/Boston: Brill. Kaye, Alan S. 1976 Chadian and Sudanese Arabic in the light of comparative Arabic Dialectology. The Hague: Mouton de Gruyter. Lethem, Gordon J. 1920 Colloquial Arabic, Shuwa Arabic dialect of Bornu, Nigeria and of the region of Lake Chad. London: Crown Agent for the Colonies. Luffin, Xavier 2008 Pidgin Creole: Bongor Arabic. In Encyclopedia of Arabic Language and Linguistics III, Kees Versteegh (ed.). 634–639. Leiden/Boston: Brill. 2011 Arabic based Pidgins and Creoles. In Semitic Languages, S. Weninger (ed.), 990–1000. Berlin/New York: Mouton de Gruyter. Muraz, Gaston 1930 Vocabulaire du patois arabe tchadien ou ‘tourkou’ et des dialectes Sara-Madjinngaye et Sara Mbaye. Nancy: Charles-Lavauzelle. Tosco, Mauro, and Jonathan Owens 1993 Turku: a descriptive and comparative study. Sprache und Geschichte in Afrika 14: 177–267.

Strata on loanwords from Arabic and other Semitic languages in Northern Northern Somali Giorgio Banti* 1. Introduction The term Somali (af-Soomaaliga ‘the Somali language’) is frequently used for referring to two different things that should not be confused: (a.) the written Somali language and the corresponding spoken koiné and (b.) the Northern Somali (NS) dialects and their most closely related varieties. After Heine (1978), Ehret and Mohamed Nuuh Ali (1984), and Lamberti (1986, 1988) there has been a general consensus that the following subgroups should be included in the Somali cluster:1 (i.a.) the NS dialects spoken in Djibouti, present-day Somaliland and Puntland, eastern and southern Ethiopia, north-eastern Kenya and the adjoining regions of southern Somalia west of the Juba river (ca. 14.000.000 speakers together with the Benaadir subgroup); (i.b.) the Benaadir dialects spoken in Somalia from the southern Mudug region to the left bank of the Shabelle river and the area of Merka (ca. 14.000.000 speakers together with the NS subgroup);

*

1.

The author is grateful to Alessandra Avanzini, Peter Behnstedt, Cabdalla Cumar Mansuur, Riccardo Contini, and Mena Lafkioui who, at different times and in different ways, helped him in preparing this chapter. He also thanks Bertrand Hirsch and François-Xavier Fauvelle-Aymar for allowing him to reproduce as Figure 1 one of their maps. Obviously, any mistake in this contribution is only due to the present author. His special compliments go to the team of Corpus of South Arabian Inscriptions of the University of Pisa for having prepared such a rich and useful database. Somali is spelt here with its official orthography; its main peculiarities are: ’ for glottal [>], c for [µ] (i.e., Arabic cayn), dh for retrofrex [Ö] and its allophone [∞], and x for [ê]. Notice Ar.= Arabic, ASA= Ancient South Arabian, Har.= Harari; MSA= Modern South Arabian. The distribution of the subgroups indicated below does not take into account the worldwide Somali diaspora in several Arab countries, India, eastern and southern Africa, Australia, Europe, and North America.

186

Giorgio Banti

(i.c.) the so-called Ashraaf dialects spoken mainly for in-group communication in several families of Mogadishu, Merka and some smaller settlements along the intervening coast; (ii.) the Maay dialects spoken by almost 2.000.000 pastoralists and farmers in the interriverine area of Somalia from Wanle Weyn to the eastern bank of the Juba River; (iii.a.) a cluster of rather diverse dialects spoken by ca. 130.000 pastoralists and farmers partly in the same area as Maay, from Baidoa to the mouth of the Juba River. It includes Tunni, Dabarre (and Irroole), Garre and Jiiddu. The Garre who live near Moyale, on the border between Kenya and Ethiopia, speak a southern Oromo dialect; (iii.b.) the Boni dialects spoken by ca. 8000 hunter-gatherers from the southern tip of Somalia to the eastern bank of the Tana River in southeastern Kenya, and (iv.) Rendille, spoken by ca. 35.000 pastoralists between Marsabit and Lake Turkana in northern Kenya. Boni has been shown by Tosco (1994) to be closely linked to Garre, and thus to be actually a southern extension of subgroup (iii.). Abdirachid Mohamed Ismaïl (2011) has suggested on the basis of mainly phonological and lexical isoglosses that subgroups (ii.) and (iii.) should be regarded as a single subgroup of dialects, that he called entre-deux-fleuves, i.e., interriverine. Notice that this is basically a subgroup formed by language contact and convergence, rather than a genetic taxon. The ensuing major articulation of the Somali cluster is thus: 1) Somali strictu sensu: Northern Somali + Benaadir + Ashraaf; 2) Inter-riverine: Maay + Tunni + Dabarre + Garre‒Boni + Jiiddu; 3) Rendille. Small groups of speakers such as the Bayso and the Girirra use languages related to the Somali cluster in southern Ethiopia. Together with Arbore, its close relative Elmolo, and Dhaasanač, the Somali cluster is regarded by linguists (cf. Sasse 1979: 3 f.) as forming the Omo-Tana subgroup of East Cushitic, one of the major branches of the Cushitic language family. NS, the Benaadir and the Ashraaf dialects are characterized by the following historical features:

Strata on loanwords from Semitic in Northern Somali

187

c

‒ they preserve * and *

as separate phonemes, e.g., NS mágac ‘name’ vs. Maay magəʴ, cAfar migác, Oromo maqáa; NS madáx ‘head’ vs. Maay madəʴ, Oromo matáa, etc.; ‒ they develop *z > d in all positions, e.g., NS wádne ‘heart’ vs. Tunni wésna, Rendille wéyna (regularly with *z > y), Northern Saho and Kambata wazana, etc.; ‒ they develop *m > n in syllable codas, e.g., NS nín sg. ‘man’ but pl. nimán (still namán in the southern Mudug region) vs. Jiiddu lám, Oromo namá, cAfar núm, etc.

NS, the Banaadir dialects and, in most but not all position, the Ashraaf ones also develop *k > g and *t > d after vowels as in: ‒ NS ilk-ó pl. ‘teeth’ vs. ílig sg. (< *ilk-) ‘tooth’, cf. Oromo ilk-áan ‘teeth’, etc.; ‒ NS madáx ‘head’ vs. Oromo matáa, etc. It should also be remarked that dh, i. e., [Ö], whatever its origin, evolves into r after a vowel in the Ashraaf and Benaadir dialects and in presentday eastern NS, but is either preserved or weakened to [∞] in the rest of NS, e. g., Benaadir and eastern NS yiri ‘he said (it)’ vs. other varieties of NS yidhi [ihÖh] ~ [ih∞h] ‘id.’ from the prefix-conjugated verb *-eÖħ‘say’, cf. Rendille yiÖaħ ‘he said’, Saho-cAfar yeÖħe ‘id.’, southern Oromo yed’e ‘id.’

2. Avenues of contact of Northern Somali with Semitic Now the main NS-speaking area is in contact with the following languages and language (sub-)groups: a) cAfar (East Cushitic) in the north-west; b) Oromo (East Cushitic) in the west and south-west; c) Harari (Southern Ethiosemitic) in the west; d) inter-riverine Somali dialects in the centre and south; e) minority Bantu languages (Mushunguli, and the two northernmost Swahili dialects, i.e., Mwiini and Bajuni) in the south; f) Amharic, widely spoken as a second language in the Somalispeaking areas of Ethiopia;

188

Giorgio Banti

g) Arabic (Ar.), known by religious people, educated Somalis, Somali traders and business people, and by Somalis who have been living and working in the Arab countries (see also Simeone-Senelle 2006). This picture, however, has been changing through time, and cannot be naively projected into the past. For instance, it is reasonable to assume that the protracted wars between Christian and Muslim polities in central Ethiopia during the late Middle Ages caused considerable changes in the distribution of languages after the 16th century in the central and eastern Horn (see later in this section). On the other hand, present-day NS includes a considerable amount of loans from the European colonial languages that came in contact with it since the end of the 19th century: Italian in former Italian Somalia, English in Somaliland and north-eastern Kenya, and French in Djibouti. Kenyan NS also borrowed more heavily than elsewhere from Swahili, whose role as a lingua franca in Kenya has grown considerably from colonial times till now. In addition to this, Cardona (1988) and Cabdalla Cumar Mansuur (2008: 26) identified a number of Persian and Indian loanwords, many of which entered NS through the lingua francas used across the Indian Ocean “among seamen, traders, soldiers” (Cardona 1988: 24). Cardona writes on the same page that such lingua francas have been in use “since the Middle Ages”, but it is likely that some forms of them were already used on the northern Somali coast in classical antiquity. Indeed the Periplus of the Erythraean Sea, a description of the sea lanes and trading opportunities from Roman Egypt to India probably written in the middle of the 1st century, explicitly states that the τοU πWραν [µπ]ρια “far-side ports of trade”, i.e., the ports that lay beyond the Bab-el-Mandeb strait on the coast of northern Somalia and immediately to the south of Cape Guardafui traded at that time both with Egypt and with the Indian port of Barygaza, which is usually identified with present-day Bharuch in Gujarat, at the mouth of the Narmada river. These “far-side ports of trade” were, e.g., Avalites (_ Αaαλcτης), Malao (f ... Μαλαh), Mundu (f Μοiνδου), Tabai (τl ... Τnβαι) and Opone (’Οπhνη). Πλεsται δt εuς πnντα ταUτα τv τοU πWραν [µπ]ρια wπl µtν Αuγiπτου περy τlν ’Ιοiλιον µ{να, | [στιν ’Επι•c. ’Εξαρτcζεται δt συν‚θως καy wπl τ†ν ‡σω τ]πων, τ{ς ’Αριακ{ς καy Βαρυγnζων, εuς τv αaτv τv τοU πWραν [µπ]ρια γWνη προχωροUντα wπl τ†ν τ]πων, σsτος καy Šρυζα καy βοiτυρον καy ‡λαιον σησnµινον καy ‹θ]νιον, Œ τε µοναχ• καy f σαγµατογ‚νη, καy περιζhµατα καy µWλι τl καλnµινον τl λεγ]µενον σnκχαρι. Καy ο• µtν

Strata on loanwords from Semitic in Northern Somali

189

προηγουµWνως εuς ταUτα τv [µπ]ρια πλWουσιν, ο• δt [κατv] τlν παρnπλουν wντι•ορτcζονται τv [µπεσ]ντα. Οa βασιλεiεται δt _ τ]πος, wλλv τυρnννοις uδcοις καθ’ •καστον [µπ]ριον διοικεsται. ‘Departure from Egypt to all these far-side ports of trade is made around the month of July, that is Epiphi. To these far-side ports of trade it is also common to ship in from the inner regions of Ariaca and Barygaza goods from those places: wheat, rice, clarified butter, sesame oil, cotton cloth, both the monache and the sagmatogene kinds, and girdles, and honey from the reed called sacchari. Some make the voyage especially to these ports of trade, and others exchange their cargoes while sailing along the coast. This area is not ruled by a king, but each port of trade is ruled by its separate chief.’ (Periplus of the Erythrean Sea, 14.1 f.; translation adapted from Casson 1989)

It is reasonable to assume that the lingua franca used along this trade route was at that time heavily lexified with Indian, South Arabian, Coptic and Greek terms. Old Ar., i.e., the older form of what is now called Ar., probably played just a minor role at that time. A little later, while speaking of the ports that lie farther south along the African coast of the Indian Ocean, such as Rhapta (τv ‘Ραπτn), the author of the Periplus wrote: ΜWγιστοι δt ([ν) σhµασι περy ταiτην τ•ν χhραν ’νθρωποι _ρατοy κατοικοUσι καy κατv τ]πον •καστος _µοcως τιθWµενοι τυρnννοις. ΝWµεται δt αaτ•ν, κατn τι δcκαιον wρχαsον ”ποπcπτουσαν τ• βασιλεc– τ{ς πρhτης γινοµWνης ’Αραβcας, _ Μο•αρεcτης τiραννος. Παρv δt τοU βασιλWως ”π]•ορον αaτ•ν ‡χουσιν ο• wπl Μοiζα, καy πWµπουσιν εuς αaτ•ν [•]λκια τv πλεcονα κυβερν‚ταις καy χρειακοsς ῎Αραψι χρhµενοι τοsς καy συν‚θειαν καy [πιγαµβρcαν ‡χουσιν [µπεcροις τε ο™σι τ†ν τ]πων καy τ{ς •ων{ς αaτ†ν. ‘Very big-bodied men inhabit the region, under separate chiefs for each place. The region is under the rule of the governor of Mopharitis, since by some ancient right it is subject to the kingdom that has become first in Arabia. The people of Muza receive tributes from it through a grant from the king, and send out to it many large ships, using Arab captains and crews, who are familiar with the natives and intermarry with them, and who know the places and their language.’ (Periplus of the Erythrean Sea, 16.5 f.; translation adapted from Casson 1989)

190

Giorgio Banti

At that time, thus, part of the East African coast to the south of Cape Guardafui had political and commercial links with the port of Muza,2 in present day south-western Yemen not far north from the Bab-el-Mandeb, and with the country of Mopharitis (aka Mapharitis, possibly to be identified with present-day Macāfar), that lay inland of it. It is difficult to assume that the ships from Muza sailed directly to the region of Rhapta without stopping at the “far-side ports of trade” mentioned earlier, in Periplus 14.1 f. As a consequence, even though it is possible that Malao, Mundu or Opone were not paying tributes to Muza like the more southern region of Rhapta, they had to have contacts with its sailors and merchants. However, one should beware of assuming that these “Arab captains and crews” spoke Old Ar. The Arabes of classical antiquity were generically the inhabitants of the Arabian Peninsula, whatever language they spoke or ethnical group they belonged to. In particular, Muza and the country of Mopharitis were on the western margin of the area of the Himyarites, already known to the author of the Periplus as Homērítai (‘Οµηρcται), who generally used south and late Sabaic in their inscriptions. There are different opinions about the language they actually spoke: Robin (2007) insists upon the differences between pre-Islamic Himyaritic and Sabaic, and seems to regard it as a Semitic language altogether different from ASA, whereas Stein (2008, 2012) stresses their similarities and regards Himyaritic as a development of Late Sabaic. Himyaritic was still known to medieval Arab scholars as a separate language, and later converged into Ar. leaving even morphological relics in some contemporary Yemeni dialects, such as the article (V)n- ~ (V)m- (< Himyaritic ) and the 1.st and 2.nd persons with -k- in the perfect. In any case, the language(s) spoken by the seamen and traders from Muza can be regarded as a source of Semitic loanwords in the language(s) of the communities that lived on the southern coast of the Gulf of Aden. As a working hypothesis, Stein’s suggestion will be followed here, and possible cognates of NS terms will be looked for in what is known about Middle and Late Sabaic. There is no direct evidence about the language(s) of the northern Somali coast at the time of the Periplus, beyond the obscure place names it mentions such as Malao (f.), Mundu (f.), Tabai (neuter), etc. But there is no compelling reason for thinking that it could not be an older form of NS. Indeed, even if Mohamed Nuuh Ali (1985: 82 f.) suggests that “a North 2.

Possibly corresponding to Mawzac, which now lies several miles inland because the coastline has receded.

Strata on loanwords from Semitic in Northern Somali

191

Lowland Eastern Cushitic”, i.e., Saho-cAfar “language could have been spoken both in the trading ports and in the hinterland”, he concedes that the evidence for this is very scarce. The group of items he lists as Saho-cAfar loanwords in NS (1985: 144 ff.) has to be taken with great care because (i.) they may have been borrowed much later, when NS was indisputably in contact with cAfar as it is today, and (ii.) for many of them there is no evidence that the direction was from Saho-cAfar into NS. For instance, he regards NS magáalo ‘town’ as borrowed from cAfar magaalá ‘town, village’, because of its “distribution” (1985: 148). However, the word also occurs in Harari magāla ‘market’ and eastern Oromo magaláa ~ magaalá ‘market, town’. It is thus attested in the four main languages spoken today in the eastern Horn. But since it lacks an obvious etymology in any of these four languages, it may also have been borrowed from a fifth, unidentified language. The only clear fact is that there is no compelling evidence that it has been borrowed from cAfar into NS! A second wave of Semitic loanwords reached the region where NS is spoken today through the spread of Islam and the ensuing growing number of people who learnt some Ar. because of their religious studies and practice, as well as the growing role of some form of Ar., both in many areas of the southern Arabian Peninsula where South Arabian and Himyaritic were spoken previously, and as a trade language along the sea routes that connected the eastern Horn with Egypt, the southern Arabian Peninsula, and Mesopotamia. The oldest known Ar. inscription from Somalia lay on the tomb of Abū c Abdallāh bin Rāyā an-Naysābūrī, found by Cerulli to the north-east of the old district of Shingaani in Mogadishu (Cerulli 1957: 2 f.). This Persian from Nishapur in the northeastern province of Khorasan died in June-July 1217: ‫ا‬ -

‫ﻣ‬ ‫را‬ ‫ﷲ‬ ‫ﷲ ا‬ ‫ وار‬, ‫ ﻣ‬+ ‫ ر ' ا*ول ﻣ‬%&

‫ا ر‬ ‫ ث‬#$ ‫م ا‬

‫ا ا‬ ‫ا‬

‫ا‬ ‫ا‬ ‫رى ا‬ ‫ا‬ ٦١٤ .

‘The weak slave, hoping in God’s mercy, Abū cAbdallāh bin Rāyā bin Mu ammad bin A mad an-Naysābūrī al-¡urāsānī died. He died on the day of Tuesday in the month of Rabīc al-Awal; six hundred and fourteen years 614 had passed (since the Hijra).’

Since then, Ar. has been the only language written by the Somalis until the end if the 19.th century, when the first examples of Somali written in Ajami, in new local alphabets such as the Osmaniya, and in Latin script start to appear (see, e.g., Banti 2011b).

192

Giorgio Banti

One important avenue through which the knowledge and use of Ar. spread to areas here NS is now spoken were the above-mentioned sea routes that connected the northern and eastern coast of the Horn with the Arabian Peninsula, Egypt and beyond. But commercial, religious and political relationships with the Islamic polities in the interior of the Horn were an important and often disregarded avenue for the spread of Ar. loanwords into the NS-speaking communities. In recent years, Fauvelle-Aymar and Hirsch (e.g., 2008 and 2011) have published the results of several archaeological campaigns and historical studies aimed at shedding more light on such polities during the Middle Ages. One of their maps, reproduced here as (Figure 1), shows a major cluster of early Islamic sites pointing south from Berbera, and south-east from this town and Zeyla towards the Chercher Mountains and the lakes of the Rift Valley. Many of these sites are towns with stone buildings, that were already abandoned and in ruins when the first European travelers saw them in the 19th century. Harar and the Argobba villages to the south-west of it, like Koromi, are the only ones that are still thriving today. In the first half of the 16th century, speakers of NS were already present in this region. The Futū al-2abaša, an eye-witness account of the wars between Christian Abyssinia and the Islamic polities led by A med the Left-handed (1506–1543 ca.), mention the Somali as one of the major elements in his army, and name among them several now NS-speaking tribes such as the Harti, the Marreexaan, and the Habar Magadle. Harar has been a major political and religious centre also for many NS areas at least from that time. Nowadays, several Somali clans still use for their traditional chiefs Har. titles such as garâad and malâaq, and the Har. word âw ‘father’ as a religious title. More recently, direct political, educational, and cultural relations with the Ar.-speaking countries have increased during the 20th century, especially after independence was reached. Since then, Ar. has been one of the three languages of the country. Many educated Somali’s have studied in Arab universities, and Somalia joined the Arab league during the seventies. Not only Ar. has been used in Harar. Its inhabitants speak a Southern Ethiosemitic language, Harari, which has been written in Ajami for several centuries. The oldest dated manuscript is from the very beginning of the 18th century, but it has to be assumed that Ajami Harari was already written at least during the second half of the 16th century (cf. Cerulli 1936: 439; Banti 2005a: 77 etc., 2010: 153 f.). The Somalis who fought in the

Strata on loanwords from Semitic in Northern Somali

193

armies of A med the Left-handed were certainly in contact with speakers of Harari in the first half of the 16th century. It is not known when Southern Ethiosemitic spread into central Ethiopia, and we can only suggest hypotheses about the languages that were spoken in the medieval Muslim towns along the northern and southern sides of the Rift valley. What appears to be certain is that: i.) the town of Harar has been a major political, religious and economic centre at least from the 16th century; ii.) there are no obviously Oromo names in medieval Christian and Muslim Ethiopian sources before the XVI century, nor in the Futū al-2abaša (cf. Banti 2011a); iii.) the Muslim Argobba to the south-east of Harar now speak Oromo and some Harari, but they still preserved in the late 19th century songs in their separate Ethiosemitic language; iv.) other Muslim Argobba groups, on the western escarpment of the Rift Valley, speak several varieties of Argobba, a Southern Ethiosemitic language closely related to Amharic; v.) Harari is closely related to Eastern Gurage languages such as Silt’e, now separated from it by a wide stretch of Oromo-speaking territory; vi.) medieval Ethiopian sources, the Futū al-2abaša, and local oral traditions mention the ¤arla as an important ethnic group in Muslim eastern Ethiopia, but it is not known what language or language(s) they spoke at that time (cf. Banti 2005b; Chekroun et al. 2011). Now there are tiny groups calling themselves Xarla, i.e., [¤arla], among the north-western Somalis and 2aralla among the c Afar, and speaking the language of the people among whom they live. Fact (ii.), together with the Abyssinian historical traditions, is generally construed as an indication (a.) that Oromo ethnic identities and the Oromo language reached their present distribution in Ethiopia only after the first half of the 16th century and, (b.) that the protracted wars between Christian and Muslim polities in central Ethiopia during the late Middle Ages caused considerable changes in the distribution of languages and ethnic identities after that century. For instance, fact (vi.) shows that even major groups vanished after that time, and that their remnants merged with the new dominant groups and languages. On the other hand, facts (iii.) to (v.) above are seen as evidence of a much wider distribution of Southern Ethiosemitic in the

194

Giorgio Banti

eastern Horn before the spread of Oromo, as Wagner (2009: 112) repeated recently. It is thus quite likely that there were substantial numbers of speakers of some variety of Southern Ethiosemitic also in the Muslim towns whose ruins lie in presently NS-speaking territory, along the old caravan routes between Harar and the coastal towns of Zeila and Berbera. Just like Harar is still now a place where NS is in contact with Southern Ethiosemitic, also these other towns probably must have played a similar role as places where speakers of NS met speakers of Southern Ethiosemitic languages. The roles of Harar and, possibly, also of speakers of other Southern Ethiosemitic language varieties in an Islamic setting in the eastern Horn of Africa, make it likely that Southern Ethiosemitic not only transmitted Ethiosemitic terms to NS, but also Ar. words it had already borrowed.

3. South Arabian (SA) loanloan-words In this section and in the two that follow it, NS items have been drawn mainly from the two major monolingual dictionaries that have been published till now: Yaasiin Cismaan Keenadiid (1976), and Saalax Xaashi Carab (2004).3 Although the Somali spelling does not mark pitch accent, tone marks have been added here for ease of pronunciation. An Ancient South Arabian (ASA) loanword in NS was identified by Cerulli (1959: 119), i.e., NS addôon m. ‘slave’. But a systematic study of the contacts between ASA, Himyaritic, and NS has not been done yet. Nevertheless, a few possible ASA (or Himyaritic) loanwords can be identified in NS. NS addôon m. and addóon f. ‘slave’ (pl. addoomó m.) has -n < -m in syllable coda, a regular development in NS (cf. § 1.) Cerulli (1959: 119) connected it with Sabaic ’dm ‘servants, vassals, religious servants’. This particular semantic development of the well-known Semitic root ’DM still survives in Modern Yemeni Ar. ’awādim ‘domestics’ (Ricks 1989: 5). Notice that -oo- is here probably from *-a- or *-ā-, a development that took place in the donor language, not in NS. Modern South Arabian (MSA) languages have several instances of accented *-āɹ- > -ōɹ- (> -ūɹ-).

3.

A third monolingual Somali dictionary, edited by Annarita Puglielli and Cabdalla Cumar Mansuur, was recently published but not yet available to the author while preparing this chapter.

Strata on loanwords from Semitic in Northern Somali

195

NS gêel m. ‘camel’ has -ee- < -aa-, that is still preserved, e.g., in NS compounds like haruub-gâal m. ‘vessel for milking camels’ or daba-ggáalle m. ‘ground squirrel’ (lit. ‘having a camel tail’), etc. East Cushitic *gaala ‘camel’ is from Semitic *gamal- ‘id.’ and the most likely intermediary has been a language spoken in Southern Arabia when camels were introduced to the eastern Horn ca. 2000 years ago, as argued in Banti (2000: 49 f.). Gml ‘camel’ is well attested in Sabaic. However, the loss of -m- is not regular in ASA, nor in NS or any known East Cushitic language, and points to a different language: loss of -m- between vowels occurs several times in Southern Ethiosemitic and in MSA. NS jádar m. ‘incense from Boswellia frereana’ (cf. Axmed Cartan Xaange 1984: 23 ff.; Saalax Xaashi Carab 2004: 295a “pure incense”). The root occurs in Middle Sabaic mq=rt ‘incense altar’ and in Ar. miq=ara ‘censer’, but Ar. q never gives NS j. However, if ASA (and Himyaritic) had an ejective realization of q like MSA and Ethiosemitic, i. e., [k’], *q- > j in NS jádar parallels *k’- > j- in NS jéb-i ‘break (tr.)’ < *k’eb-, a word with a good East Cushitic etymology, as shown by Dhaasanech g’éb ‘break (tr.)’, Oromo č’ab-s- ‘id.’, Konso qep-š- ‘id.’, etc. (cf. Sasse 1979: 49; Banti and Contini 1997: 173 f.) In several other instances, words that have been mentioned in the literature as loans from Ar. are already attested in ASA, and may thus have entered NS at a much earlier date than usually assumed: – NS luubáan f. (also lubáan and lubaanád) ‘frankincense’ either directly from ASA lbny (attested in Sabaic and Qatabanian), or through Ar. lubān. – NS berkéd f. ‘cistern, artificial pond’ from Sabaic brkt ‘id.’ (Beeston et al. 1982: 31) either directly or through Ar. birka ‘cistern’, ‘pool’. According to Behnstedt and Woidich (2011, Map 272) this word is used only in the mountains of southern Yemen and the Tihama, with some further isolated attestations in Jordan, northern Syria, the Gulf and southern Oman. Notice that a direct borrowing from Ar. would have been expected to be *birkad or *birko, as in Har. birka. – NS gidâar m. ‘wall’ either directly from ASA gdr ‘id.’ (e.g., Qatabanian), or from an Ar. dialect that had a velar realization of LJīm in Ar. LJidār ‘id.’, cf. Har. LJidār ‘thickness of wall’ with the palatal affricate. – NS qáalin m. and qaalín f. (pl. qaalm-ó) ‘young camel or calf when it approaches sexual maturity’ (nêef gêel áh ama ló’ oo dá’ yár Yaasiin Cismaan Keenadiid 1976: 333b), with ġ > q and -m > -n in

196

Giorgio Banti

syllable coda, that are regular developments. From the Semitic root ĠLM, attested in Sabaic ġlm ‘boy, child’, or directly from Ar. ġālim (participle of ġalima) or ġalim both meaning ‘excited by lust’, from the same root Ar. has ġillīm ‘(he-camel) excited by lust’ and ġulām ‘a young man, youth’. The present author already pointed out (Banti 2000) that it also occurs in Rendille kháalim m. ‘male camel calf’ and khaalím f. ‘female camel calf’ and is thus unlikely to be a recent loanword. The ASA hypothesis assigns qaalin to the oldest core of camel-related terms that entered the eastern Horn, together with *gaala ‘camel’ > NS gêel; in this case, the extension to bovine calves is a secondary development. The fragmentary nature of what is known about ASA and, even worse, about Himyaritic makes it difficult to identify many other terms as old loanwords from these languages. But in the four above cases, there is no compelling evidence for excluding a pre-Islamic origin. A more careful examination of the NS lexicon related to camels, building, and incense may yield some other findings.

4. Arabic loanwords Present-day NS and Benaadiri varieties are full of Ar. loanwords, even though their occurrence may vary in individual texts or dictionary sources. For instance: − the two above-mentioned monolingual dictionaries for native speakers of Somali don’t include many of the more obvious Ar. loanwords, because their authors regarded them as being “just plain Arabic”, as one of them explained almost thirty years ago; − instead, Agostini et al. (1985), a Somali-Italian dictionary prepared by a team of Italian and Somali linguists, includes a higher amount of Ar. loanwords, because many of them are not obvious for European users who are not familiar with this language; − sermons and written texts about Islam are teeming with Ar. words; − newspaper articles about contemporary politics display a considerable number of terms of Ar. origin; − traditional narratives display lower percentages of Ar. loanwords.

Strata on loanwords from Semitic in Northern Somali

197

The issue of Ar. loanwords in NS and their phonological and grammatical adaptation have been discussed by several authors, with different methodologies, e. g., Reinisch (1903: 1–23), Zaborski (1967, 2008), Callegari (1988), Cardona (1988), Soravia (1994), Cabdalla Cumar Mansuur (2008: 25 ff.), Vasaturo (2012), etc. Callegari (1988) indentified 1297 Ar. loanwords in her mainly NS corpus, and classified them in the following meaning groups: 1) religione (‘religion’); 2) utensili, vestiti, attrezzi (‘household implements, garments, tools’); 3) cultura, insegnamento, grammatical (‘culture, education, grammar’); 4) amministrazione, politica (‘public administration, politics’); 5) relazioni sociali (‘social relationships’); 6) commercio, economia (‘trade, economy’); 7) corpo umano, medicina (‘human body, medicine’); 8) sentimenti, comportamenti umani (‘emotions, human behaviors’); 9) legge (‘law’); 10) calendario, tempo, giorni della settimana (‘calendar, time, days of the week’); 11) piante, frutta, profumi (‘plants, fruits, perfumes’); 12) cibi, bevande (‘foods, beverages’); 13) edifici, abitazioni (‘buildings, dwellings’); 14) navigazione (‘seafaring’); 15) professioni, mondo del lavoro (‘crafts, work’); 16) geografia, natura (‘geography, nature’); 17) parentela, rapporti familiari (‘kinship, family relations’); 18) gioco, prostituzione, alcolismo (‘games, prostitution, alcoholism’); 19) spostamenti, viaggi (‘trips, travels’); 20) animali (‘animals’); 21) minerali, gioielli (‘minerals, jewels’); 22) misure (‘measures’); 23) insulti (‘insults’); 24) nazionalità (‘nationalities’); 25) varie (‘miscellaneous’, i.e., words that had not been listed above); 26) avverbi, preposizioni, particelle, esclamazioni (‘adverbs, prepositions, particles, exclamations’). Soravia (1994: 201) identified 1436 Arabic loanwords using Agostini et al. (1985) as his corpus. As mentioned above, this dictionary tends to record more Arabic loanwords than the two monolingual ones. In addition to this,

198

Giorgio Banti

it should also be pointed out that it includes some Benaadiri items beside the NS lexicon, because it aimed at recording “Somali” in a broader sense. Soravia (1994: 202) remarks that he assembled a 100-item Swadesh list of basic lexicon on the basis of his corpus, and that it included just one item borrowed from Ar., i.e., ákhdar ‘green’. Yet even this loanword co-occurs with the non-Ar. word cagâar ‘green’ and ‘verdure, lush vegetation’. In other words, he claims that the core vocabulary of Somali and of NS as well, is almost devoid of Ar. loanwords, even though they are extremely frequent in more culturally-bound areas of the lexicon. There also are several syntactic constructions in NS that can be regarded as grammatical and syntactic borrowings from Arabic, e.g.: – iláa Hargeysá ‘until Hargeisa’, ‘from Hargeisa till here’ replacing the inherited type tán iyo H., lit. ‘this and H.’, that has parallels in Saho etc. (an intermediate type is iláa iyo H. lit. ‘to and H.’); notice that the only two prepositions of Somali, iláa and mín, are borrowed from Ar.: otherwise Somali uses local nouns and oblique particles before predicates where most other Semitic languages use prepositions; – in as a generic complementizer for several types of sentential complements, from dialectal Arabic in ~ inn (cf. Banti 2011b: 39 f.); – the type isága oo óoyayá ‘while he cries, crying’ (lit. ‘he and he cries’) from the Ar. wāw al- āl construction wa-huwa yabkī, dialectally u-hū bi-yibki (lit. ‘and he cries’.) Many of the above-mentioned studies list how Ar. sounds are rendered in the receiving language. Frequently they focus only on the written forms, without taking in consideration that educated speakers may tend to follow an Ar. pronunciation on several occasions. For instance, NS lacks z, and Ar. wazīr ‘minister’ is spelt wasiir in NS, yet educated speakers frequently realize it as wa[z]iir with a voiced fricative. There is thus a considerable degree of variation according to the speaker and to the register of his speech. Nevertheless, some Ar. sounds or clusters are rendered in NS in several cases in different ways that require a more detailed discussion. One of these is Ar. F (Fād, ‫)ض‬: – NS d as in: mucáarad m. ‘opponent’ (< mucāriF ‘opponent’ and mucāraFa ‘opposition’); fadéexo f. and fadeexád f. ‘great shame, scandal’ (< faFī a ‘id.’); cudúd f. ‘the upper arm between the shoulder and the elbow; strength’ (< caFud ~ cuFud ‘id.’); etc.

Strata on loanwords from Semitic in Northern Somali

199

– NS l ~ ll as in: qaallí m. ‘cadi, Islamic judge’ (< qāFin ~ al-qādī ‘id.’); árli m. ~ árlo ‘earth’ (< arF ‘id.’); líd m. and líddi m. ‘person who is against something’ (< Fidd ‘against’); etc. – NS s in: wéeso ‘ritual ablution’ (< wuFū’ ‘id.’). – Possibly NS dh [Ǧ ~ ȍ] in: wádhaf m. ‘sling’, cognate of Yemeni Ar. waFāfa ‘id.’ according to Behnstedt and Woydich (2011: map 270); however, they record also waGaf ~ waGif ‘id.’, awƽl=of ‘id.’, and developments of these forms in Maltese and Ar. dialects from Tunisia, northern Algeria and Marocco, and medieval Andalusia, while Piamenta (1990–91: 526a–b) has both wa=uf ‘id.’ and miGāfa ‘id.’; bádh m. ‘half’ if it really is from Ar. bacF ‘part’, ‘some’. The most frequent developments are d and l ~ ll, the second one in more common and everyday words (but with exceptions like the above cudúd f. ‘the upper arm between the shoulder and the elbow; strength’), the fist one seemingly in more specialized terms, and in contemporary political language. I > l ~ ll reflects a lateral articulation of this phoneme, that is now lost in most contemporary varieties of Arabic. It is thus an old pronunciation that was established in NS quite early. Interestingly, Ar. F never occurs as l in loanwords in Har. (Leslau 1956) and cAfar (Leslau 1995), nor in Swahili (Baldi 2008; Vasaturo 2012): its most frequent rendering is d, but dh and sometimes z in Swahili. Amharic has either of d, z or = (Leslau 1957a, 1988), while Argobba has a few occurrences of d according to Leslau (1957b). The isolated rendering as s in wéeso, a technical term of Islamic ritual, is also isolated in Har. wussa, where Ar. F generally gives d as in bidā’a ‘merchandise’ < biFāca ‘id.’ cAfar has wadó with d. A possible explanation for this irregularity in NS and Har. is that it is a word that entered the Horn during the early diffusion of Islam, and then spread to Har. and NS through a third language where Ar. F > s was a regular development. Leslau (1963: 162) regards wussa as a loanword from Somali in Harari, without taking into account the fact that it is also irregular in Somali. On the other hand, wádhaf is an old word in the Horn, as attested by the Geez verb waFafa ~ waFFafa ~ waJafa ‘hurl with a sling’. Akkadian already had waJpu > aJpu ‘sling’, but the occurrence of both F, G and = in its Yemeni Ar. cognates is highly irregular. Har. has woč’āfa ‘id.’, Amharic.

200

Giorgio Banti

wanč’əf ‘id.’ with a non-etymological -n- that also occurs in cAfar wanÖef ‘id.’ The irregular correspondences in dialectal Arabic and the isolated [Ö] in Somali and cAfar indicate that this word, even though it is likely to be of Semitic origin, had a complex history, and is not necessarily a loanword from regional Arabic into Ethiosemitic and East Cushitic. More likely, it re-entered Ar. from the Horn, and the occurrence of [Ö] in NS and cAfar has to be explained in the context of the bilateral history of East Cushitic and Ethiosemitic. Also NS bádh ‘half’ is suspect as a loan from Ar. bacF ‘part’, ‘some’, because cayn is usually retained in NS loanwords from Ar. It thus has probably a different etymon. In some cases, one Ar. word has produced two different NS words. For instance, there are from Ar. zakā ‘zakat alms, tithe, the giving of 2.5% of one’s surplus wealth once a year to charity, generally to the poor and needy’: – NS dágo f. 1. ‘the first month of the Muslim year’; 2. ‘domestic animals that are given away as alms once every year’. – NS sáko ~ séko f. 1. ‘zakat alms’; 2. ‘the first month of the Muslim year’. The first term develops Ar. z- and -k- in the same way as inherited East Cushitic *z and postvocalic *k, as seen above in § 1. It is not used as a technical term for the Islamic zakat. The second term devoices Ar. z into s and seems to preserve Ar. -k- after the first vowel. However, one should not forget that inherited East Cushitic long *-tt- and *-kk- occur as -t- and, respectively, -k- in NS. Since the Har. term for zakat is zakka, the occurrence of -k- in NS sáko is best explained as a direct loanword from Harari, that is, from the language of the town that has been the major centre of Islam in the eastern Horn for the last five centuries. Ar. z merges with d, i.e., is treated like inherited East Cushitic *z, only in a few other words that belong to common everyday life, and that probably were borrowed quite early. Otherwise it is just devoiced. Ar. z: – NS d in: múndul m. (> múddul ~ muddúlle) ‘round dwelling with a thatched conical roof’ (< manzil ~ manzal ‘house, dwelling’); etc. – NS s as in: wasaarád f. ‘ministry’ (< wizāra ‘id.’); sáman m. (~ sémen ~ sében) ‘time’ (< zaman ‘id.’); etc.

Strata on loanwords from Semitic in Northern Somali

201

That múndul is an old loanword is also indicated by the complete change of its vocalic pattern, from -a-a- into -u-u-, probably through labialization triggered by the initial m- and subsequent assimilation of the second vowel to the first one, in order to have full vowel harmony. A similar development can be observed in NS musqúl f. ‘washroom, toilet’ < Ar maġsil ‘washroom, lavatory’, with the regular development Ar. ġ > NS q, metathesis -qs- > -sq-, and -a-a- > -u-u-. Ar. vowels are otherwise generally well preserved, even though there can be some changes in vowel length, usually with V > VV, while Ar. a-a sometimes becomes e-e. Finally, Ar. words with a final consonant cluster are rendered in NS in two different ways: – NS -CCi in: dígri m. ‘dhikr’ (< Ar. Nikr ‘id.’); shárci m. ‘law’ (< Ar. šarc ‘id.’); dérbi m. ‘wall’ (< Yemeni Ar. darb ‘wall’, cf. Piamenta 1990–91: 147a); etc. – NS -CVC with echo vowel insertion as in: kídib m. ‘lie’ (< Ar. kiNb); méher m. ‘dowry’ (< Ar. mahr); dúhur m. ‘noon’, ‘noon prayer’ (< Ar. Guhr); etc. It is interesting to notice that echo vowel insertion is an active process in NS nominal and verbal phonology, e.g., ílig ‘tooth’ vs. ilk-ó ‘teeth’, árag ‘see (sg.) it!’ and wây arag-tay ‘she saw it’ vs. árk-a ‘see (pl.) it!’ from the roots /ilk-/ and /ark-/. Instead adding a final -i is an active process in Har. phonology for avoiding final consonant clusters, as in the simple imperfect Har. 3sg.m. yilabsi /yilabs/ ‘he dresses’, Har. karsi ‘belly’ vs. Geez karś ‘id.’, Argobba and Gurage kärs ‘id.’, etc. Accordingly, Ar. words in -CC are generally adapted with the -CCi pattern in Harari, e.g., zikri ‘dhikr’, Old Har. šarci > Modern Har. šar’i ‘law’, ma ri ‘dowry’, zuhri salāt ‘noon prayer’, etc.; occasionally they lose their second consonant as in kiz ‘lie’. A likely explanation of the occurrence of the two patterns in NS is thus that the kídib type is an internal development of NS, while the dígri type spread into NS through Harari, possibly through words like zikri > dígri and šarci > sharci whose pattern was extended analogically to other loanwords from Arabic.4 4.

It is interesting that also Ar. loanwords in cAfar systematically display -i for avoiding final consonant clusters, e. g., sárci ‘doctrine, theology’, máhri ‘alimony given to a divorced wife’, etc. Yet, differently from Har., cAfar phonology either degeminates final long consonants as in enged ‘deny (it)!’ (from

202

Giorgio Banti

5. Ethiosemitic loanwords Not much attention has been devoted to the bilateral contacts between Ethiosemitic and NS, even though several authors have addressed the issue of the impact of Cushitic, and thus also of Somali, upon Ethiosemitic, cf. recently Appleyard (2012), who also mentions the most recent literature on this topic. But borrowing from Ethiosemitic into NS has been discussed, at least to the present author’s knowledge, only by Mohamed Nuuh Ali (1985: 150 ff.) and Banti (1988: 57). The former identifies some Har. loanwords in NS, but commits some astonishing oversights such as considering Northern and Benaadir Somali sitimâan ‘week’ as a borrowing from Har. sātti ‘seven, week’ with “added suffixation” (ibid.: 153), whereas its Italian origin (< settimana ‘week’) is obvious. Banti, on the other hand, points out that ugâas ‘tribal chief’ and agáasin ‘good government, good management’ have Southern Ethiosemitic etymologies: NS agáas-in m. ‘good management, orderly arrangement’, and ugâas m. ‘tribal chief’ are from the Ethiosemitic root GZ’: Geez gaz’a ‘dominate, master’ and ’əgzi’-ənnā ‘dominion, sovereignty’, Har. gaza’a ‘govern, own’, Gurage languages gaza ‘id.’ (Chaha and Endegeň gasa ‘id.’ with their regular devoicing of etymological z), Amh. gazza ‘id.’. The patterns uCaaC- and aCaaC- are typical of NS derivation from prefixconjugated verbs, as shown in Banti (1988). A few other Ethiosemitic loanwords can be pointed out here, in a list that is in no way exhaustive: – NS maalín f. (pl. maalmó m.) ‘day’ < *maal-im cf. máan-ta ‘today’ with *-lt- > -nt- partial assimilation (cf. Ar. al-yawm ‘today’, lit. ‘the day’), from Ethiosemitic: Geez. mocālt ~ macalt ‘day’, Har. ma’altu ~ māltu ‘day’ < WcL ‘spend the day’. The suffix -t and -tu was lost also in several Southern Ethiosemitic palatalized forms like Har. mōy ‘day’, Zway māy-a ‘id.’, etc.; yet NS *maal-ta ‘the day, today’ can also be regarded as a reanalysis of a form like Har. māltu, because the f. article in NS is -ta, with Nominative case -tu.

/engedd/, cf. yengedde ‘he denied it’), or inserts an echo-vowel as in subaáħat ‘with butter’ (from /subaáħ + t/), cf. Parker and Hayward 1985: 216). Is there also here a Har. model, even though cAfar is not in direct contact with Har. today?

Strata on loanwords from Semitic in Northern Somali

203

– NS badán ‘be much, be many’, badí f. ‘being much, being many’, bádi v2 ‘be much, be abundant, be strong in’, from Ehiosemitic: Geez bazUa ~ baz a ‘be abundant, be much, be many’, Har. baza a ‘be numerous, be abundant’ and biz a ‘abundance’. Bazza ‘be abundant’ with loss of occurs in most Gurage languages and in Amharic. Astonishingly, these are words from the basic lexicon, i.e., from the very core of the lexicon. For both of them, the donor language was most likely a Southern Ethiosemitic language. A few other Ethiosemitic loanwords in NS are listed below, five of them most likely from Har., the last one from Amharic during the last decades: – NS géyi m. (~ gáyi ~ gíyi) ‘land, country’, from Southern Ethiosemitic: Har, gē ‘Harar’ (lit. ‘the Country’), several Gurage languages ge ‘country village’ (with possible parallels in other Semitic languages). – NS gówrac v. and n.m. ‘slaughtering an animal by slitting its throat’ from Southern Ethiosem.: Har. gōra’a v. ‘id.’, several Gurage languages gorā v. ‘id.’, Geez gwarcaya ‘id.’ denominative verb from Geez gwərce ‘throat, neck’ with several other Semitic cognates. The development of ō after the first root consonant only occurs in Southern Ethiosemitic. – NS arabbikhí f. ‘maize, corn’ from Har. arab iUi ‘sorghum of the Arabs’, with Har. loss of cayn in arab < carab (cf. NS Caráb-ta ‘the Arabs’) and the typically Har. inverted genitive construction. – NS âw m. religious title, similar to shêekh, from Har. aw ‘father, elder, religious title’ (already mentioned above in § 2). – NS malâakh m. ‘tribal chief’ (already mentioned above in § 2) from Har. malāq ‘official who is responsible for the welfare of each locality connected with one of the five gates of the city of Harar’, from Har. mala ‘ways and means’ + āqa ‘know’ (cf. Leslau 1963: 107b). – NS kílil m. ‘region’ in Kílil Shán ‘Region 5’, i.e, the Somali region of Ethiopia, from Amharic kƽllƽl ‘zone’. Notice that NS garâad ‘tribal chief’, which was mentioned above in § 2, does occur also in Har. with the same meaning, but doesn’t have a clear etymology. As noted by Gori (2005) it already occurs as a title of Muslim rulers in medieval Abyssinian sources, and has been used subsequently also by some Christian rulers. It has thus to be regarded as a title that has been

204

Giorgio Banti

used since the Middle Ages in Muslim polities of the Horn, whatever its linguistic origin, but there is not enough evidence for regarding it as a specifically Ethiosemitic loanword. Indeed, Leslau (1963: 75a) regards it as borrowed from Cushitic into Harari, but also for this there is very little evidence.

Strata on loanwords from Semitic in Northern Somali

200 km l____________l

Massawa

205

Dahlak

Phase 1 : route septentrionale Dahlak-Massawa-Est Choa ?-10e-13e siècle

Zeyla Berbera

H

Phase 2 : route orientale Zeyla/Berbera-Harar-Lacs ?-13e-16e siècle

Séquence des voies de pénétration musulmane dans la Corne et en Éthiopie Sites musulmans Inscriptions funéraires arabes

Figure 1. Penetration routes of Islam into the Horn and Ethiopia (from FauvelleAymar and Hirsch 2011: 20.)

References Abdirachid Mohamed Ismaïl 2011 Dialectologie du somali: problématiques et perspectives. PhD thesis, INALCO/LLACAN (Paris). Agostini, Francesco, Annarita Puglielli, and Moxamed Siyaad Ciise (eds.) 1985 Dizionario somalo-italiano. Rome: Gangemi. Appleyard, David 2012 Semitic-Cushitic/Omotic Relations. In The Semitic languages: an international handbook, Stefan Weninger (ed.), 38–53. (Handbücher zur Sprach- und Kommunikationswissenschaft 36.) Berlin/Boston: Mouton De Gruyter. Axmed Cartan Xaange 1985 Dalkii udgoonaa. Mogadishu. Baldi, Sergio 2008 Swahili. In Encyclopedia of Arabic language and linguistics, vol. 4 (Q-Z), Kees Versteegh (ed.), 381b–387a. Leiden/Boston: Brill. Banti, Giorgio 1988 Reflections on derivation from prefix-conjugated verbs in Somali. In Proceedings of the Third International Congress of Somali Studies, Annarita Puglielli (ed.), 43–59. Rome: Il Pensiero Scientifico. 2000 Notes on Somali camel terminology. In “Mehr als nur Worte...”: Afrikanistische Beiträge zum 65. Geburtstag von Franz Rottland, Rainer Voßen, Angelika Mietzner, and Antje Meißner (eds.), 45–62. Cologne: Köppe. 2005a Remarks about the orthography of the earliest ‘aLJamī texts in Harari. In Scritti in onore di Giovanni M. D’Erme, vol. 1, Michele Bernardini, and Natalia Tornesello (eds.), 75–102. Naples: Università degli Studi di Napoli “L’Orientale”. 2005b Harla. In Encyclopaedia Aethiopica, vol. 2, Siegbert Uhlig (ed.), 1034a–1034b. Wiesbaden: Harrassowitz. 2010 The literature of Harar until the end of the 19th century. Civiltà del Mediterraneo 16–17: 149–181. 2011a Oromiffaa. In Encyclopaedia Aethiopica, vol. 4, Siegbert Uhlig (ed.). Wiesbaden: Harrassowitz. 2011b Somali language. In Encyclopaedia Aethiopica, vol. 4, Siegbert Uhlig (ed.), 693a–696b. Wiesbaden: Harrassowitz. 2011c Internally-headed relative clauses in literary Somali? In A country called Somalia: culture, language and society of a vanishing state, Mara Frascarelli (ed.), 32–47. (Studi Somali 14.) Turin: L’Harmattan Italia.

Strata on loanwords from Semitic in Northern Somali

207

Banti, Giorgio, and Riccardo Contini 1997 Names of aromata in Semitic and Cushitic languages. In Profumi d’Arabia, Alessandra Avanzini (ed.), 169–192. Rome: “L’Erma” di Bretschneider. Beeston, A. F. L., Ghul, M. A., Müller, W. W., and J. Ryckmans 1982 Sabaic dictionary (English ‒ French ‒ Arabic). Louvain-la-Neuve/ Bayrouth: Peeters/Librairie du Liban. Behnstedt, Peter, and Manfred Woidich 2011 Wortatlas der arabischen Dialekte. Band II: Materielle Kultur. (Handbook of Oriental Studies, Section 1 The Near and Middle East 100/2.) Leiden/Boston: Brill. Cabdalla Cumar Mansuur 2008 Taariikhda iyo luqadda bulshada Soomaaliyeed. N.p.: Daabacaaddii Yuusuf X. Cabdullaahi Xasan “Iftiinka Aqoonta”. Callegari, Laura 1988 I prestiti arabi nella lingua somala, Vols. I-II. Dissertation, Istituto di Orientalistica, Università degli Studi di Torino. Cardona, Giorgio R. 1988 Somalia and the Indian Ocean: cultural and linguistic contacts. In Proceedings of the Third International Congress of Somali Studies, Annarita Puglielli (ed.), 22–35. Rome: Università degli Studi “La Sapienza”. Casson, Lionel 1989 The Periplus Mari Erythraei: text with introduction, translation and commentary. Princeton/Guildford: Princeton University Press. Cerulli, Enrico 1936 Studi etiopici I. La lingua e la storia di Harar. Rome: Istituto per l’Oriente. 1957 Somalia: scritti vari editi e inediti, vol. 1. Rome: Amministrazione Fiduciaria Italiana della Somalia. 1959 Somalia: scritti vari editi e inediti, vol. 2. Rome: Amministrazione Fiduciaria Italiana della Somalia. Chekroun, Amélie, François-Xavier Fauvelle-Aymar, Bertrand Hirsch, Hailu Zeleke Deresse Ayenachew, Olivier Onezime, and Shewangizaw Addisu 2011 Les Harla: archéologie et mémoire des géants d’Ethiopie. Proposition de séquence historique pour les sites du Čʖärčʖär. In Espaces musulmans de la Corne de l’Afrique au Moyen Âge: études d’archéologie et d’histoire, François-Xavier Fauvelle-Aymar, and Bertrand Hirsch (eds.), 75–102. (Annales d’Ethiopie, hors-série 1.) Paris/Addis Ababa: Editions De Boccard/Centre Français des Etudes Ethiopiennes.

208

Giorgio Banti

Ehret, Christopher, and Mohamed N. Ali 1984 Soomaali classification. In Proceedings of the Second International Congress of Somali Studies, vol. 1, Thomas Labahn (ed.), 201–269. Hamburg: Buske. Fauvelle-Aymar, François-Xavier, and Bertrand Hirsch 2008 Etablissements et formations politiques musulmans d’Ethiopie et de la corne de l’Afrique au Moyen Age: vers une reconstruction. Annales Islamologiques de l’Institut Français d’Archéologie Orientale 42: 339–75. 2011 En guise d’introduction: sur les traces de l’Islam anicien en Ethiopie et dans la Corne de l’Afrique. In Espaces musulmans de la Corne de l’Afrique au Moyen Âge: études d’archéologie et d’histoire, François-Xavier Fauvelle-Aymar, and Bertrand Hirsch (eds.), 11–26. (Annales d’Ethiopie, hors-série 1). Paris/Addis Ababa: Editions De Boccard/Centre Français des Etudes Ethiopiennes. Gori, Alessandro 2005 Gärad. In Encyclopaedia Aethiopica, vol. 3, Siegbert Uhlig (ed.), 696b–97b. Wiesbaden: Harrassowitz. Heine, Bernd 1978 The Sam languages: a history of Rendille, Boni and Somali. Afroasiatic Linguistics 6: 23‒115. Lamberti, Marcello 1986 Die Somali-Dialekte. (Kuschitische Sprachstudien 5.) Hamburg: Buske. 1988 Die Nordsomali-Dialekte. (Studia Linguarum Africae Orientalis 1). Heidelberg: Winter. Leslau, Wolf 1956 Arabic loanwords in Harari. In Studi orientalistici in onore di Giorgio Levi Della Vida, vol. 2, 14–35. Rome: Istituto per l’Oriente. [Reprinted in Arabic loanwords in Ethiopian Semitic, Wolf Leslau, 119– 140. Wiesbaden: Harrassowitz.] 1957a Arabic loanwords in Amharic. Bulletin of the School of Oriental and African Studies 19: 221–244. [Reprinted in Arabic loanwords in Ethiopian Semitic, Wolf Leslau, 1–24. Wiesbaden: Harrassowitz.] 1957b Arabic loanwords in Argobba. Journal of the American Oriental Society 77: 36-39. [Reprinted in Arabic loanwords in Ethiopian Semitic, Wolf Leslau, 48–57. Wiesbaden: Harrassowitz.] 1963 Etymological dictionary of Harari. Berkeley/Los Angeles: University of California Press. 1988 Additional Arabic loanwords in Amharic. In Collectanea Aethiopic, 87–109. (Äthiopistische Forschungen 26). [Reprinted in Arabic loanwords in Ethiopian Semitic, Wolf Leslau, 25–47. Wiesbaden: Harrassowitz.]

Strata on loanwords from Semitic in Northern Somali 1995

209

Arabic and Amharic loanwords in Afar. Rassegna di Studi Etiopici 39: 131–154. Mohamed Nuuh Ali 1985 History of the Horn of Africa, 1000 B.C.‒1500 A.D.: aspects of social and economic change between the Rift Valley and the Indian Ocean, Ph.D. thesis, University of California. Parker, Enid M., and Richard J. Hayward 1985 An Afar-English-French dictionary (with grammatical notes in English). London: School of Oriental and African Studies. Piamenta, Moshe 1990–91 Dictionary of post-classical Yemeni Arabic, vols. I-II. Leiden/New York/København/Köln: Brill. Reinisch, Leo 1903 Die Somali-Sprache: III, Grammatik. Vienna: Hölder. Ricks, Stephen D. 1989 Lexicon of inscriptional Qatabanian. (Studia Pohl 14.) Rome: Pontificio Istituto Biblico. Robin, Christian 2006 ¤imyaritic. In Encyclopedia of Arabic Language and Linguistics, vol. 2, Kees Versteegh (ed.), 256–261. Leiden/Boston: Brill. Saalax Xaashi Carab 2004 Qaamuus - Ereykoobe. Djibouti: Machadka Affafka/Xarunta Cilmibaadhista. Sasse, Hans-Jürgen 1979 The consonant phonemes of Proto-East-Cushitic (PEC): a first approximation. Afroasiatic Linguistics 7: 1–67. Simeone-Senelle, Marie-Claude 2006 Horn of Africa. In Encyclopedia of Arabic Language and Linguistics, vol. 2, Kees Versteegh (ed.), 268–275. Leiden/Boston: Brill. Soravia, Giulio 1994 Gli imprestiti arabi in somalo. In Sem - Xam - Iafet: atti della 7.a Giornata di Studi Camito-Semitici e Indoeuropei, Vermondo Brugnatelli (ed.), 199–224. Milan: Centro Studi Camito-Semitici. Stein, Peter 2008 The “¤imyaritic” language in pre-Islamic Yemen: a critical reevaluation. Semitica et Classica 1: 203–212. 2012 Ancient South Arabian. In The Semitic languages : an international handbook, Stefan Weninger (ed.), 1042–1073. (Handbücher zur Sprach- und Kommunikationswissenschaft 36). Berlin/Boston: Mouton De Gruyter. Tosco, Mauro 1994 The historical reconstruction of a southern Somali dialect: Proto-Karre-Boni. Sprache und Geschichte in Afrika 15: 153–209.

210

Giorgio Banti

Vasaturo, Laura 2012 I prestiti arabi in somalo e swahili. Dissertation, Facoltà di Lettere e Filosofia, Università degli Studi di Napoli “L’Orientale”. Wagner, Ewald 2009 Harari und Ostgurage. Aethiopica 12: 111–125. Yaasiin Cismaan Keenadiid 1976 Qaamuuska Af-soomaaliga. Mogadishu: Guddiga Af-soomaaliga/ Akademiyaha Dhaqanka/Wasaaradda Hiddaha iyo Tacliinta Sare. Zaborski, Andrzej 1967 Arabic loan-words in Somali: preliminary survey. Folia Orientalia 8: 125–175. 2008 Somali. In Encyclopedia of Arabic language and linguistics, vol. 4, Kees Versteegh (ed.), 272a–275a. Leiden/Boston: Brill.

SubSub-Saharan lexical influence in North African Arabic and Berber Lameen Souag

1. Introduction From the ninth to the nineteenth century, a trans-Saharan trade network carrying gold, salt, slaves, and other commodities linked North Africa with the Sahel, playing an important role in both regions' history. An estimated 5,000 slaves a year reached the Maghreb from the Sahel in this period (Wright 2006); many more remained in the Sahara (McDougall 1992), notably as agricultural labourers. These involuntary immigrants brought their languages with them; some ex-slaves around Timimoun (Algeria) spoke Bambara into the twentieth century (Bouchemit 1951). In one case, a Songhay language (Korandjé) became predominant in an oasis a thousand kilometres north of the Sahel, Tabelbala in Algeria (Cancel 1908; Souag 2010). Conversely, during this period North African merchants regularly travelled south for long trips, forcing them to learn local languages; in nineteenth century Ghadames (Libya), almost all men learned Hausa, and sometimes other Sahelian languages too (Haarmann 1998: 16). We therefore expect some degree of sub-Saharan influence on Arabic and Berber varieties of North Africa, particularly in the Sahara. However, this impact remains nearly invisible: it is indicative that the Encyclopedia of Arabic Language and Linguistics (Versteegh 2005) does not even mention the existence of such loans into Arabic, despite devoting articles to ‘Wolof’, ‘Fulfulde’, ‘Hausa’, ‘Kanuri’, ‘Moroccan Arabic’, ‘Algerian Arabic’, ‘Tunisian Arabic’, and ‘Libyan Arabic’. This chapter is intended to help fill that gap. I exclude here the more extensive, better-known influence of sub-Saharan languages on Arabic and Berber varieties of the Sahel, eg Hassaniya (Heath 2004; Taine-Cheikh 1988), Nigerian Arabic (Owens 2000), and Tuareg (Kossmann 2005); my focus is on loans that have

212

Lameen Souag

reached the northern Sahara or the Maghreb proper, excluding proper nouns. Hausa words are quoted from Bargery (1934)1; Ghadames Berber from Lanfry (1973); Kanuri from Cyffer and Hutchison (1990); Manga Kanuri from Jarrett (2007); Djenné, Timbuktu and Gao Songhay from Heath (1998); Zarma from Bernard and White-Kaba (1994); Dogon from Heath (2010); Benghazi Arabic from Adam Benkato (personal communication); Moroccan Arabic from de Prémare (1993); and Siwi from the author's AHRC-funded fieldwork, unless otherwise noted. 2. Plants and animals For reasons of climate or history, some crops were more commonly grown in the Sahel than in the Maghreb; and, particularly along the northern fringes of the Sahara, slaves often worked in agriculture. A number of plant names reached North Africa as a result. 2.1. Pumpkins The most widespread sub-Saharan loan encountered is a word noted without etymological comment in Behnstedt and Woidich (2010, no. 163): kābūya ‘pumpkin’ (Cucurbita pepo / maxima). While this plant is of New World origin (Paris 2001), its names in many Old World languages derive from ones previously applied to similar indigenous plants; it appears to have reached West Africa as well as Europe relatively early (Blench 1997). In Maghrebi Arabic, Cucurbita pepo in general (including courgettes etc.) is usually called qə ȥa. However, another form is widely attested, particularly in Algeria and Libya: − kābū, sg. kābuww-a: o Arabic: Tunisia: Nafzāwa (Bin Murād 1999; Boris 1958) − kābūy-a: o Arabic: Libya: kabuia Fezzan (Nachtigal 1987: 122; Rohlfs 1881: 479) Tunisia: Marazig (Boris 1951), Sfax (Zuwārī 1998), Takrouna (Marçais 1958) 1.

This source is unreliable on final vowel length (Jaggar, personal communication), but final vowel length is not contrastive in any North African language.

Sub-Saharan lexical influence in North African Arabic

213

Algeria: Algeria (Ben Sedira 1886)/(Beaussier 1958), Sidi Bel Abbes (Madouni-La Peyre 2003), Laghouat, villages near Dellys, Bechar (author's fieldwork) Morocco: Morocco Berber: Algeria: Mzab kabuya, pl. id kabuya (Delheure 1984) − &a-kabuy-&: Berber: Morocco/Algeria: Beni Iznacen, Beni Snous (Destaing 1914) − kabiwa: Berber: Algeria: Ouargla (Delheure 1987) − kƮbewa: Berber: Algeria: Tamahaq (Foucauld 1951); not reported in other Tuareg varieties – contrast Alojaly (1980), Sudlow (2009) − t-kabiwa-t: o Berber: Algeria: Timimoun (Boudot-Lamotte 1964) o Berber: Morocco: Figuig (Sahli 2008) − k(a?)bāwa: Arabic: Algeria (Beaussier 1958) − t-,abaw-&: Berber: Algeria: Metmata (Destaing 1914) − kabau-t: Berber: Algeria: Chaoui (Huyghe 1907) − bkēwa: Arabic: Libya: Benghazi − lə-kdewa: Berber: Siwi No similar forms are attested elsewhere in the Arab world (Behnstedt and Woidich 2010, no. 163) and the ūy/īw makes an Arabic origin extremely implausible. A Berber etymology would be less absurd, but still unlikely. Relevant forms are limited to a single subfamily of northern Berber, Zenati, except for Tamahaq, where it exactly matches the sub-Saharan forms discussed below, and Siwi, where it shows a prefix lə- (< Arabic al-) normally restricted to Arabic loans. The variation between -a and -t is irregular within Berber, but normal for Arabic feminine loans, usually either borrowed with original -a or Berberized with the feminine t-...-t. Bin Murād (1999) derives this word from Romance capo < Latin caput, for its shape, but this fails to explain the forms with uy or iw, sequences atypical for Arabic and unlikely to be introduced into a loanword not already containing anything similar. Moreover, the semantic shift would have taken place within Romance, and no Romance language is known to use capo for ‘pumpkin’. Due to the back vowel, Turkish kabak ‘pumpkin’ would be reflected in Arabic as unattested *qābāq, which also leaves the ending unexplained. Persian kadu ‘pumpkin’ matches Siwi ləkdewa, assuming an unattested Arabic diminutive *al-kudayw-ah; but this cannot explain the other forms.

214

Lameen Souag

The similarity with Japanese kabocha is coincidental, since the latter derives from ‘Cambodia’ via Portuguese (Schmidt 2009). Looking further south reveals a much better match: Hausa kàbēwā3 ‘pumpkin’, segmentally identical to Ahaggar kƮbewa. In Algeria and Morocco unstressed vowels are phonetically short, and there is no phonemic ē; kābīwa is thus the expected reflex there. Adding the Berber feminine circumfix directly yields tkabiwat (Timimoun, Figuig); first excluding the apparent Arabic feminine ending -a from the stem and harmonizing the vowel yields t,abaw& (Metmata), or kabawt with the common Chaoui loss of pre-consonantal t-. The ending -īwa is unusual in Arabic, and metathesis yields a commoner sequence (cp. xū-ya ‘my brother’); thus Algerian kābūya. Substituting the Berber feminine for -a yields &akabuy& (Beni Iznacen, Beni Snous.) The expected form in Libyan Arabic is *kabēwa, but deleting -a- allows this to be reinterpreted as an Arabic diminutive on the CCēCa template, *kbēwa (like iznēga ‘small path’, ij6ēma ‘small sip’ – Owens (1984: 75); the i- is epenthetic, cf. Owens 1984: 13). The attested Libyan form may be taken as the result of metathesis, and the Siwi form (in which lə- regularly reflects the Arabic article) as the result of irregular b > d. Musical loans (discussed below), and Tremearne (1914), confirm that Hausa-speaking slaves were found in the larger towns of Algeria and Libya, whereas Algeria and Libya have very different trans-Mediterranean contact patterns – Spanish and French in Algeria, Italian and Greek in Libya. A Hausa source thus accounts for all observed variants. Ghomara Berber tagnawt ‘pumpkin’ (Hannouche 2010: 266), ie ‘Negro (f.)’, independently confirms sub-Saharan associations. There is evidence for this form's antiquity in Hausa – most importantly Gwandara kàmbúwã (Matsushita 1974), since Gwandara, Hausa's closest relative, split from Hausa before 1500 (Matsushita 1992). It is difficult to trace the form further back, but there are indicators of a longer history. The branch of Chadic to which Hausa belongs, West Chadic, is divided into two subgroups, A and B (Newman 1977). Hausa belongs to A, but the only two attestations of related forms for ‘pumpkin’ are in B: Bade kā3biyón/kā3bìyau (Schuh 2009a) and Ngizim gā3ɓiyƮu (Schuh 2009b). Since both directly border Hausa, borrowing cannot be excluded, but if so, the difference in form suggests an old borrowing. A similar form survives in A: Karekare gwàbò ‘large round gourd’ (Schuh 2009c). Cucurbita pepo includes many very different cultivars, often given separate names (thus English distinguishes ‘courgette’ from ‘pumpkin’).

Sub-Saharan lexical influence in North African Arabic

215

Given the intensity of trans-Mediterranean contact, the species must have reached North Africa from Europe; but the kābūya cultivar may have arrived via Nigeria's Atlantic coast. 2.2. Sorghum and maize An unproblematic loan widely distributed in the eastern/central Sahara is (n)gafūli ‘sorghum/maize’: − gafuli: o Arabic: Libya: Fezzan ‘sorghum’ (Rohlfs 1881: 494; Despois 1946) Libya: Tripolitania, Fezzan ~ mosri ‘maize’ (Laoust 1920: 266; Vogel 1854; Rohlfs 1881: 494) − ngafûli: Arabic: Tripolitania: ‘sorghum’ (Laoust 1920: 266), (Nachtigal 1987: 122) − əngafuli: Berber: Awjila ‘maize’ (Paradisi 1960) − əngafule: Berber: Algeria: Tamahaq Tuareg ‘maize’ (Foucauld 1951; Nicolaisen 1963) − : Berber: Libya: Tamahaq (Ghat) ‘maize’ (Nehlil 1909) (not reported in other Tuareg varieties – contrast Alojaly (1980), Sudlow (2009) These forms derive from Kanuri, as already noted by Rohlfs (1881: 494) and more recently discussed by Kossmann (2005: 62); cp. Kanuri ŋgavəBǽi ‘Sorghum vulgare’ (Lukas 1937), ŋgawúli ‘guinea corn’, as well as Tubu ŋailí, ŋahéli, ŋgáfola ‘Sorghum vulgare’ (Lukas 1953). Maize is an American plant; but it resembles sorghum in size and shape, and throughout a wide area of north-central Africa, ‘maize’ is literally ‘Egyptian sorghum’, including gafuli mosri above (Pasch 1983). Its association with sub-Saharan Africa is confirmed by a different series noted by Laoust (1920: 266): Tunisian Arabic ȥabidiʤa, Nafusi tȥabîdit ‘maize’, literally ‘of slaves.’ 2.3. Cayenne pepper As observed by Lanfry (1973), Ghadames has borrowed ‘Cayenne pepper’, bƮrkano, from Hausa bàrJkKnō.

216

Lameen Souag

The Modern Standard Arabic term for the same plant is šaNNa/šaNīNa = Capsicum conicum (Wehr 1974). However, this form is not classical; it is absent from Lisān al-ȥarab (Ibn Manđʖūr 1955). Dialectally, it appears limited to northeastern Africa: Egyptian šaNNa/(formally diminutive) šiNēNa (Hinds and Badawi 1986), Sudanese šaNNa (Qāsim 1972), Chadian šatte (Jullien de Pommerol 1999), Wadai šīNa (Roth-Laly 1969), Fezzan (Rohlfs 1881), and Benghazi šəNNa ‘harissa’. It is absent from other Arabic dialect dictionaries consulted, such as Levantine (Denizeau 1960), Moroccan, southern Tunisian (Boris 1958), or Yemeni (’Iryānī 1996). At first sight, an Arabic-internal etymology seems conceivable. In Marazig (southern Tunisia; Boris 1958), we find šaNN ‘faire souffrir’ [make suffer], eg šaNN ȥaláy elwƮžaȥ “je souffre énormément” [I suffer enormously]; in Egyptian (Hinds and Badawi 1986), šaNN is ‘to strike (a match)’. Pain and fire both seem natural associations for Cayenne pepper. However, neither adequately accounts for other data. The Hausa term for this pepper is cìttā (from which Kanuri njittá). According to Blench (1998: 178), this formerly referred to the Malagueta pepper, Aframomum melegueta – indigenous to West Africa (Pickersgill 2005) – but it now refers exclusively to Capsicum varieties, native to the Americas, a semantic shift common in Nigeria. Blench (1997; 1998) treats this as an Arabic loan into Hausa, but this is problematic: Hausa has /š/, so it should not substitute c for Arabic š; and the Arabic form is stressed on the first syllable, leading us to expect a high tone followed by a low, not vice versa. If the term was borrowed from Hausa, neither problem applies: Arabic has no c, and stress assignment in Arabic is not lexical. Moreover, it seems unlikely that Hausa would have borrowed an indigenous plant's name from the language of a region lacking it. This suggests that Rohlfs (1881: 486) was correct in identifying as a “name certainly originating from the Sudan”, and makes it unsurprising that in the Maghreb “the chilli became associated with West Africa” (Blench 1998: 177). Other eastern Maghrebi terms corroborate a southern origin: in Tunisia, Cayenne peppers are fəlfəl barr-aȥbīd ‘slaveland pepper’. 2.4. Acacia Benghazi Arabic uses the obviously non-Arabic word brisanya ‘acacia’.

Sub-Saharan lexical influence in North African Arabic

217

This is a Songhay loan: the suffix ñâ ‘mother’ forms tree names from fruit names throughout Songhay, and Zarma has bílsà ‘arbre à tanin, Acacia nilotica’ [tree for tannin]. A similar form without the l is proto-Songhay – cp. Timbuktu, Gao bisow ‘Acacia tortilis sp. raddiana’, Kaado bís-óẁ ‘Acacia raddiana et Acacia dichrostachys ninerea’ (Ducroz and Charles 1978), and Zarma bísàw ‘Acacia raddiana, tortilis’. Acacia pods are used in tanning, an unpleasant job involving smelly substances and likely to have been assigned to slaves. 2.5. Other plants Timimoun Berber dəmši ‘peas’ (Boudot-Lamotte 1964) derives from Songhay *dámsí ‘Bambara ground-nut’ (Souag 2012), which closely resembles the pea (cp. English ‘peanut’ < ‘pea’ + ‘nut’). Ghadames amāNig ‘peanut’ (Lanfry 1973) derives ultimately, as the author notes, from Bambara tìgá ‘peanut’; its immediate source is probably Songhay, cp Gao maatiga. Songhay *gá:sú is reflected in Tidikelt Arabic āgāUu ‘gourd’ (Anonymous 2008) (= Malian Hassaniyya (Heath 2004); probably via Tamasheq a-ǯášu (Heath 2006)), and in Ghadames tagazut ‘gourd’. According to Rohlfs (1881: 471), the lote-tree (Zizyphus spina-christi) is in Fezzan generally called (Kanuri kúrna) although its fruit keeps an Arabic name. Likewise, the Sodom-apple (Calotropis procera) is referred to in Fezzan by a name of Hausa origin, tintafia (Rohlfs 1881: 483) from Hausa tùnfāfìyā, and what he calls a Kanuri name, krunka (although dictionaries give kayôu.) 2.6. Animals dagəl ‘monkey’ in Sokna Berber (Libya) is a Kanuri loan (Jakobi and Kossmann Fc); cp. Kanuri dágəl ‘monkey’. Ghadames Berber āku ‘parrot’ is Hausa àku (= Kanuri akú), as Lanfry notes. Ouargla Uafo ‘stork’ (Delheure 1987), isolated in Berber, might be compared to Hausa tsōfō ‘old person or thing’, for the stork's white head; but no such usage is attested in Hausa itself. The dəmman, a western Saharan sheep breed, requires brief discussion.

218

Lameen Souag

Kossmann (2005: 66), while rejecting previous efforts to derive pan-Chadic *tmk ‘sheep’ from this name, notes its similarity to Kanuri dími ‘sheep’. Since the breed is popularly considered to be of Sudanic origin, and is similar in its colour range and hairy fleece to West African breeds (Mason 1980), the idea of a sub-Saharan etymology is not implausible. However, its geographic distribution militates against this. In Morocco, this breed is found mainly in the southeast (Mason 1980); in Algeria, mainly in the southwest (GREDAAL 2005, via Zouyed 2005). In Tunisia, its introduction dates to 1994 (Saidi et al. 2011). If it had been taken from Kanuri, we would expect a more easterly distribution. Kossmann’s alternative proposal that it derives from a Berber colour term is therefore preferable. 3. Material culture With the exception of musical instruments – discussed separately – few loans in this domain have been encountered except in Ghadames, where they include: − klabo ‘grande peau de bovin’ [big cow-hide] < Hausa kìlābò ‘tanned ox-hide’ (= Kanuri kəláwo); Ghadamsi merchants imported these from the Sahel for an international market, also using the word in their Arabic (Haarmann 1998: 38). Also into Awjila Berber (Paradisi 1960): klâbu ‘pelle di animali molto erta’ [very thick animal hide] − kure ‘grande tunique de Kano’ [big tunic from Kano] < Hausa kōrè ‘gown of very dark indigo-dyed (almost black) material rendered glossy’ – also into Tamahaq kora ‘étoffe indigo en tissu de Kôra’ [indigo material in cloth of Kora] (Foucauld 1951) − kinda ‘grande corbeille de sparterie’ [big woven basket] < Hausa kìndaì ‘finely woven basket’ (= Kanuri kəndâi ‘small round basket’; Gao kokondo ‘large palm-leaf basket’) − kibya ‘arc, (arme de jet)’ [bow, (arrow)] < Hausa kibiyà ‘arrow’ − sese ‘pièce de monnaie soudanaise’ [Sudanese coin] < Hausa sīsì ‘sixpence’ − mazamne ‘coussin de cuir (rond)’ [(round) leather cushion], derived from Hausa zamnà/zaunà ‘sit’, as noted by Lanfry − lugde ‘grande louche en bois’ [big wooden ladle] < Hausa lūdàyī ‘ladle or spoon’. The g is original; by Klingenheben's Law, syllablefinal velar stops have weakened to u in Hausa (Jaggar 2001: 27).

Sub-Saharan lexical influence in North African Arabic

219

The Arabic-speaking Dawwada of Fezzan (Pauphilet 1953) use tari ‘fishing-net’ < Timbuktu Songhay taari (cp. Hausa tārū). Further west, Gourara Berber offers one possible case: kurru, ‘petite boite fait avec de la peau séchée de chameau’ [small box made from dried camel-hide] (Bellil 2000a: 225). Given the material, this word may be compared to proto-Songhay *kù:rù ‘skin, hide’; however, the specificity of the meaning and the presence of gemination present problems. 4. Ideophones Many Sahelian languages have a category of lexicalized ideophones with highly specific meanings; no such category exists in most North African languages. The Berber language of Ghadames is an exception; Lanfry (1968: 374) gives a list. Not only is the category itself of sub-Saharan origin – so are a fair number of the ideophones themselves. The probable source is Hausa, and some have been noted only there: − Gh.: inna frit, ‘D D’un bond, frit yƮrwƮl bond il s’éclipsa.’ [‘In n one bound, bound he slipped away.’] Hausa: firit (emphasizes suddenly leaving place) eg ya fita firit ‘He left firit!’ − Gh.: Hausa:

yəmmūt kəff yā ƙare kaf

‘Il est mort, plus de doute. doute.’ [‘He is dead, no more doubt doubt.’] ‘It is absolutely finished.’

More frequently, however, the ideophones have a wider areal distribution, including Kanuri and Songhay: − Gh.: mƮlləl fərr ‘Il est d'un blanc éclatant.’ éclatant [‘It is dazzlingly white.’] Hausa: fari fat ‘snow snowsnow-white’ Kanuri: bûl fók ‘very very white’ Gao: a kaarey far! ‘It’s very white.’ far − Gh.: Hausa:

sƮNNƮf çlə çlək tilik, tilik, tìlikà eg baki tilik

‘Il est noir, on ne peut plus.’ plus [‘It is black, it couldn't be more.’] more (exclamation at deep blackness) ‘black tilik!’

220

Lameen Souag

Gao: − Gh.: Hausa: Kanuri: − Gh.:

tirik!

(intensifier for bibi ‘black’)

tu^əft zərr

‘de la laine absolument pur’ pur [‘absolutely absolutely pure wool’] ‘wholly wholly guinea-corn, unmixed with any other grain’ grain ‘girls only’ nly

dāwa zar ferowá zúr yƮrrə_ qƮšš

Hausa: yā ƙarye ƙâs Kanuri: kás namgátə

‘Il est cassé net.’ [‘It broke cleanly.’] cleanly ‘It snapped with a loud noise.’ noise ‘It broke cleanly in two.’ two

Nigerian Arabic has likewise acquired ideophones under the influence of surrounding languages; cf. Owens (2004). Ghadames appears to be the northernmost example so far reported in Africa. 5. Curses and insults The possibility of a Songhay origin for Gourara Berber (Algeria) abəngu ‘head’ was first discussed in Kossmann (2004); my recent fieldwork has clarified the situation. In the Berber of Timimoun (Gourara), certain body parts have synonyms used primarily in maledictions of the form “God give you a (pain in the)...” In this context, fus ‘hand’ is replaced with kambi, and taməgna ‘head’ with abəngu. In Timimoun abəngu can also be extended to other negative contexts like abəngu-nu la ihəlk-i “my head is destroying me” (ie: “I have a headache”); in three villages it has become the unmarked word for ‘head’ (Bellil 2000b: 48; Basset 1959). Both are of Songhay origin; cp. respectively proto-Songhay *kàmbè ‘hand’ and *bòŋo ‘head’ (Souag 2012). (Not all curse-synonyms are Songhay; contrast kukku ‘eye’.) This phenomenon has not been documented elsewhere; however, closely related Ouargla Berber (Algeria) uses the term mo66o ‘gros mangeur, goinfre, baffreur’ [big eater, greedy pig, glutton]. This could speculatively be derived from Bambara mǤ3gǤB [mǤ_`Ǥa] ‘person, human’, plus comparable pejoration.

Sub-Saharan lexical influence in North African Arabic

221

6. Music, dance, and healing Ex-slaves from West Africa and their descendants have a strong corporate presence in certain traditional musical genres of North Africa; this is reflected in the distribution of loanwords. 6.1. Musical instruments 6.1.1. Percussion The name of the big double-sided drum used by black musical organizations in North Africa is widely loaned: − ganga: o Arabic: Algerian: ‘grosse caisse, (tambour)’ [big drum, (drum)] (Beaussier 1958) Moroccan: gānga: ‘tambour des Gnāwa comportant une caisse en bois assez longue et deux peaux’ [Gnawa drum consisting of a rather long wooden case and two skins] Libyan: ganga ‘tambour traditionnel’ [traditional drum] (Le Quellec 2003) o Berber: Morocco: Tashelhiyt: ganga ‘tambour des nègres’ [Negro drum] (Destaing 1920) Libya: Ghadames: gangan ‘tambour cylindrique à deux peaux’ [cylindrical drum with two skins] (Lanfry 1973) Algeria: Tuareg: Tamahaq: ganga ‘petit tambour plat’ [small flat drum] (Foucauld 1951) Niger: Tuareg: Tamajeq: gƮnga ‘tambour’ [drum] (Alojaly 1980) Burkina Faso: Tuareg: Tamasheq: tegƮngƮnt ‘drum’ (Sudlow 2009) − banga: o Arabic: Tunisian: bǻnga: ‘gros tambour à deux faces; individu stupide (insulte à un noir); séance diurne d’exorcisme’ [big two-sided drum; stupid individual (insult to a black person); diurnal exorcism session] (Boris 1958)

222

Lameen Souag

Algerian (Annaba): (Azarnia 2007) Algerian (Constantine): (Pâques 1964: 558) Aguadé (2002) points to the Hausa word gàngā as the source, but the word is areally widespread; cp. Kanuri gangá, eastern Songhay (Kaado gàŋgà, Zarma gàŋgá, Dendi gámgàn), Dogon (Nanga gàŋ'gƮŋ, Tabi gàŋá) as well as Gourma and Baguirmi (Skinner 1996). The Tunisian variant with a b refers to the same item, but its form reflects the influence of another Hausa word: bàngā ‘a 'royal' small drum, tapped with the fingers’ (Bargery 1934). A smaller drum is called sara at Constantine (Pâques 1964: 558); cp. Hausa sārā ‘A particular rhythm played on a drum by the Emir of Kano's drummers’. A yet smaller drum is called kulu at Constantine (Pâques 1964: 558); cp. Hausa kōlō ‘small drum’, Kanuri kóló = Manga Kanuri kólló ‘very small drum beaten with one or two sticks and making a loud rattling noise’, Zarma kólló ‘petit tam-tam’ [little tom-tom], Timbuktu koloo ‘simple, onesided drum’. An even smaller single-sided drum is called kurk(u)tu in some areas, including Tunis (Rahal 2000; CMAM; Jankowsky 2010), Annaba (Azarnia 2007), Constantine (Pâques 1964: 558), and Algiers (Dermenghem 2002); cp. Hausa kurkutu ‘small drum consisting of a piece of wood hollowed out like a basin and covered with skin’, a term used in Katsina, Sokoto, and Daura (Bargery 1934). Related verbs are attested in Songhay, eg Timbuktu kurkutu ‘beat drums’. A long barrel-drum formerly used in Tunis is called the dūndūfa (Jankowsky 2010: 106); cp. Hausa dùndufā3 ‘long, narrow drum which stands on the ground while being played’. 6.1.2. Stringed The most important stringed instrument in the repertoire is the Moroccan gəmbri/Gourara bəngri (Boudot-Lamotte 1964)/Tunis gumbri (Jankowsky 2010). The term is used in Morocco to refer to two distinct lute types, one specific to North Africa played mainly by Arabs, and one with sub-Saharan affinities played by ex-slaves (Schuyler 1979) (via Charry 2000: 129); similar instruments are referred to with an Arabic diminutive of this term, gunībrī (Farmer 1928), or in Tunis a feminine form, gambara (Jankowsky 2010). A West African source has been suggested by Delafosse (1955),

Sub-Saharan lexical influence in North African Arabic

223

who compared Manding konibara ‘four-stringed guitar’, and by Charry (1996), who, as well as citing Delafosse, compares Soninke gànbárè/ gùnbàllé, treated by Diagana (2011) as an Arabic loan. Songhay cases may be added: Djenné kumbur, Timbuktu kubur, Gao kurbu, whence Dogon (Walo) kùmbúrù. The first attestation of this word, according to Farmer (1928) (plural qanābir), is in Ibn Baccūcah's description of the Mali royal court, which would greatly strengthen the case for it being a sub-Saharan African loan. However, Charry (1996) casts doubt on this, noting that it might be explained as a copyist's error for tanābīr, a word for ‘lutes’ attested in medieval sources. The question thus remains open, since Arabic qunbur ‘lark’ also provides a possible etymology. A deeper examination falls beyond this paper's scope. A one- or two-stringed fiddle is termed gūgēy/gūgāy in Tunis (Zawadowski 1942; Jankowsky 2010), and in Morocco gūgọ ‘sorte de très grande mandoline des Gnāwa’ [a kind of very big Gnawa mandolin]. As Jankowsky (2010: 107) notes, it derives from Hausa gKgē, whence also Zarma gKjé. 6.2. Ritual implements and medicines The UNambēli brotherhoods of Tunisia call a ritual potion used in initiation (Rahal 2000: 31). This is Hausa māganī ‘medicine, magic, charm’. A ritual drink of soured milk, sorghum, water, and sugar is called dindiri (Jankowsky 2010); this must be Hausa kìndirmK ‘sour milk with or without cream, but with no added water’ = Kanuri câm kə3ndəBrmú ‘sour milk’, although the initial consonant is troubling. Among the brotherhoods of Algiers, a ritual whip was called bulala (Pâques 1964: 520) = Hausa būlālà ‘a whip of hide or plaited leather’. 6.3. Dances The kəUka of Libya, glossed by Mitchell (1957) as ‘Negro dance’, is a type of men's stick-dance (Ali, Rollow and Wood 1998: 666). As such, its name is probably Kanuri kəská ‘tree, wood... wooden pole or post’. The ba-hore dance of Ghadames (ba = ‘master’) is ‘un genre de réunion dansante en rangs parallèles avec musique et chant’ [a type of gathering for

224

Lameen Souag

dancing in parallel rows with music and song]; cp. Gao Songhay horey ‘...singing and dancing, celebration, festivity’. In Siwa (Egypt), stilt-walking is called NguNgu. Stilt-dancing is wellattested in West Africa, and comparable terms (certainly not the immediate source) are found in Hausa (tā3kwarākwarà ‘stilt’) and Dogon (Perge tèŋgè-táŋgá ‘stilt dancers’); but further data is needed. 6.4. Organizations Jankowsky (2010: 19) points out evidence that the sNambēli music of Tunisia was previously termed sambeli, and plausibly compares this to Hausa sambalē ‘a dance of youth and maidens’. Less plausibly, he also compares Zarma sambeli, given by Stoller (1989: 233) as ‘to shoot the magic arrow’. In any case, as he notes, the intrusive t reflects the name of Istanbul, to which folk etymology links the term. In Tripolitania and Fezzan, “les Noirs d'origine esclave s'appellent Sambani” [Blacks of slave origin call themselves Sambani] (Pâques 1964: 473), a name given to their organizations (Pâques 1964: 499). Jankowsky (2010: 19, 203) implicitly compares this to sNambēli, discussed above, and to Hausa sàmbā3nī ‘a kind of castanets or metal clappers’, commonly played by such groups; the former appears preferable, but the latter could have influenced the change l > n. 6.5. Titles Much of the internal terminology of these organizations derives from Sahelian languages. In the “seven springs” cult of Algiers (Andrews 1903), the chief of the women's dar was termed . This looks like Songhay huu ñaa ‘house's mother’, but a broader meaning for is suggested by the title of an official of the male assembly who beat the drum, ; ganga is clear, but the remainder is unexplained. The room where religious paraphernalia were kept was termed ; Hunwick (2004) plausibly derives this from Manding (Bambara só ‘house’.) The treasurer of the men’s dar, , which Pâques (1964: 567, 641) gives for Saïda and Algiers as ‘singer’, poses a few possibilities. For the first word, Andrews' version suggests Timbuktu Songhay kaaya

Sub-Saharan lexical influence in North African Arabic

225

‘inheritance’, while Pâques' rather suggests the name of a dance, (Pâques 1964: 567). In either case it is tempting to interpret the last element as Songhay boŋo ‘head’, but ‘head’ > ‘chief’ would be odd in Songhay, where one would expect -koy ‘master’. If correct, this might reflect a calque from Turkish, the language of the pre-French ruling class, which uses baş ‘head’ to mean ‘master’ (Hony, Iz and Alderson 1992). In In-Salah, the title of is rendered by Pâques (1964: 642) as ‘the donkey’, a “servant of the brotherhood”; cp. Bambara jǤ3nkk ‘slave, captive’ (Dumestre 2011). In Tunisia, the head of a dar bears the title of galadīma kbīr (Rahal 2000: 25; Jankowsky 2010: 37); apart from Arabic kbīr ‘big’, this is Hausa gàlā3dīmà ‘an official title or position... the Emir's principal minister’, itself a loan from Kanuri galtíma/galadíma ‘title given to the official next in rank to the Shehu’. The attendant and cook was given the title serki samari (Jankowsky 2010: 59), to be compared to Hausa sarkin samrī/saurī ‘an official position and title of one in whom is invested absolute authority over the young folk of the town for communal work, for games, and the immorality consequent on games’. A more difficult problem is posed by yenna/yinna (Rahal 2000: 32; Jankowsky 2010: 37), given in Tunisia to the leader of the band. A term meaning ‘father’ or ‘uncle’ would seem appropriate, but the best matches are in fact with senior female relatives – Hausa innà ‘mother’ and Manga Kanuri yìnná ‘mother's younger sister’ – although the position in question is apparently always occupied by a man. Some support for gender reversal is perhaps provided by the (Arabic) term ȥarīf-a, with a feminine ending, used in sNambēli to refer to male as well as female healers; but this position was typically held by women (Jankowsky 2010: 59). 6.6. Group-internal argot Almost in passing, Berjaoui (1997: 156–7) documents an argot used by the Gnaoua of El Jadida in Morocco, substituting non-Arabic words into an Arabic matrix. His data has not previously been analyzed. Most forms derive from Songhay: noro ‘money’ < nooru; kuy ba kuy ‘let's go’ < koy ‘go’; sindi ‘fatigue’ < cindi ‘rest’; glo ‘good’ < gumo ‘right, much’; kaNihari ‘(bring) water’ < kate hari ‘bring water’; ^angi mamatšin ‘be quiet’ < dangey ‘be quiet’ + baa ‘be better’ + ma (subjunctive) + čiiney ‘whisper’ (Heath 1998). Some of these forms reflect

226

Lameen Souag

historically recent sound changes: nooru derives from proto-Songhay *nògrú, as a result of the post-14th century regular loss of coda g [`] in most varieties (Souag 2012). Bambara/Manding has also contributed a few words (Dumestre 2011; Bailleul 1981): ^ominika ‘food’ < dúmuni kkB ‘eat’; sġo ‘meat’ < sògo; go minti ‘good’ perhaps from bk3n ‘be pleasant’ + -nci (superlative); and probably, šo o o ‘tea’ < shyǤBǤlǤ ‘filter, pour slowly, drip’ + perfective -la. A couple of forms may be Fulani: nyama ‘bread’ < ñaam- ‘eat’; dawkri/ kokri ‘man’ probably < gorko (Niang 1997). The preponderance of Songhay is surprising, since at present there are far fewer speakers of Songhay than of Hausa or Manding. Is this a longterm effect of the Moroccan conquest of Songhay in 1591, or just a result of Songhay’s regional role as a lingua franca? 7. Conclusion While there is no reason to believe that the list of loans above is exhaustive – quite the contrary – the loans observed so far give us a general picture of sub-Saharan lexical influence on North Africa, falling under three main headings. The most significant impact for North Africa as a whole relates to the spread of new crops: the widespread Hausa loans kābūya ‘pumpkin’ and šaNNa ‘Cayenne pepper’ suggest that Nigeria was a secondary source for the dispersal of certain cultivars of New World plants. The more localized case of gafūli ‘sorghum/maize’ may relate to regional crop preferences rather than to the introduction of new varieties. Along the northern edge of the Sahara – Touat, Ghadames, and Fezzan in particular, and to a lesser extent Cyrenaica – the lexical impact of subSaharan Africa was greater, chiefly from Songhay, Hausa, and Kanuri respectively. The direct impact of slavery is reflected in the borrowing of agricultural and tanning terminology; the impact of trade, in the borrowing of terms for material goods and exotic animals, especially in Ghadames. However, this contact also introduced functionally unmotivated loans serving stylistic purposes, such as ideophones in Ghadames and body part curse-words in Timimoun. Among the oases in question, good dictionaries exist only for Ghadames and Ouargla; many relevant varieties are barely documented (for example, Touat Arabic), and further work will likely uncover more loans there.

Sub-Saharan lexical influence in North African Arabic

227

The organizations established by (ex-)slaves carved out a position for themselves in North Africa as musicians, dancers, and healers; in this capacity, they not only preserved a large number of sub-Saharan terms internally, but even gave some such terms (notably musical instruments: ganga and possibly gumbri) currency in wider society. Three words used in this context in Tunis – the substitution of banga for ganga, the title galadīma, and the insertion of t in sNambēli – suggest an overt effort to associate this practice with royalty, a phenomenon not noted elsewhere. The terminology of these organizations is eclectic in origin, but we may discern a strong Hausa influence everywhere, particularly in Tunisia; a more limited Kanuri influence in Libya; Songhay and Manding influence in Morocco and Algeria; and possible traces of Fulani influence in Morocco. More work on these organizations’ argots – if others exist – would help in identifying loan sources. While sub-Saharan lexical influence in North Africa is real, it should not be exaggerated. Only four non-music-related loans (‘pumpkin’, ‘sorghum/ maize’, ‘Cayenne pepper’, and ‘acacia’) are observed on the coast, and even there, they remain largely restricted to Bedouin or Zenati varieties. Many northern dialects use none of these. Irrespective of domain, the parts of North Africa influenced by each language are generally those due north of its speakers; only Hausa’s influence goes well beyond these bounds.

References Aguadé, Jordi 2002 La langue des gnâwa. In Aspects of the Dialects of Arabic Today: proceedings of the 4th Conference of the International Arabic Dialectology Association (AIDA, Marrakesh, Apr. 1-4. 2000), Abderrahim Youssi, Fouzi Benjelloun, Mohamed Dahbi, and Zakia Iraqui-Sinaceur (eds.), 405–411. Rabat: Amapatril. Ali, Aisha, Mardi Rollow, and Leona Wood 1998 North Africa. In International Encyclopedia of Dance: A Project of Dance Perspectives Foundation Inc., (Ed.) Selma Jeanne Cohen, George Dorris, and Thomas F Kelly. New York: Oxford University Press. Alojaly, Ghoubeïd 1980 Lexique Touareg-Français/Ăwgălel Təmajəq-Təf ənsist. Copenhagen: Akademisk Forlag. Andrews, James B. 1903 Les Fontaines des Génies - Seba Aioun. Croyances soudanaises à Alger. Algiers: Jourdan. Anonymous 2008 ‘Agāgū. AbbesTidikelt.com. http://abbestidikelt.ahlamountada.com/ montada-f20/topic-t121.htm (16 April, 2010). Azarnia, Abdelkader 2007 Ma Biographie. Benga System Gnawa. http://annabagnawa.voila.net/ page4/index.html (7 March, 2012). Bailleul, Charles 1981 Petit Dictionnaire Bambara-Français, Français-Bambara. Amersham: Avebury. Bargery, George P. 1934 A Hausa-English Dictionary and English-Hausa Vocabulary. London: Oxford University Press. Basset, André 1959 Articles de Dialectologie Berbère. Paris: Klincksieck. Beaussier, Marcelin 1958 Dictionnaire Pratique Arabe-Français contenant tous les mots employés dans l’arabe parlé en Algérie et en Tunisie, ainsi que dans le style épistolaire, les pièces usuelles et les actes judiciaires. Algiers: La Maison des Livres. Behnstedt, Peter, and Manfred Woidich 2010 Wortatlas der Arabischen Dialekte. Band I: Mensch, Natur, Fauna und Flora. Leiden: Brill.

Sub-Saharan lexical influence in North African Arabic

229

Bellil, Rachid 2000a Les Oasis du Gourara (Sahara algérien) III : Récits, contes et poésie en dialecte tazenatit. Leuven: Peeters. 2000b Les Oasis du Gourara (Sahara algérien) II : Fondation des ksour. Leuven: Peeters. Ben Sedira, Belkassem 1886 Dictionnaire Français-Arabe de la Langue Parlée en Algérie. Algiers: Jourdan. Berjaoui, Nasser 1997 Parlers secrets d’el-Jadida: Notes préliminaires. Estudios de Dialectología Norteafricana y Andalusí 2: 147–158. Bernard, Yves, and Mary White-Kaba 1994 Dictionnaire Zarma-Français (République du Niger). Paris: Agence de coopération culturelle et technique. Bin Murād, Ibrāhīm 1999 Al-Kalim al-AɄjamīyah fī ɄArabīyat Nifzāwah (bi-al-Janūb al-Gharbī al-Tūnisī) [Foreign Words in the Arabic of Nefzaoua (in Southwestern Tunisia).] (Silsilat al-Lisānīyāt 10). Tūnis: al-Wizārah al-Ūlá, Kitābat al-Dawlah lil-Bajth al-ɄIlmī wa-al-Tiknūlūjiyā, Markaz al-Dirāsāt wa-al-Bujūth al-Iqtigādīyah wa-al-IjtimāɄīyah. Blench, Roger 1997 A history of agriculture in Northeastern Nigeria. In L’Homme et le Milieu Végétal dans le Bassin du Lac Tchad, Daniel Barreteau, René Dognin, and Charlotte von Graffenried (eds.), 69–112. Paris: Office de la recherche scientifique et technique outre-mer. 1998 The Introduction and Spread of New World Crops in Nigeria: a historical and linguistic investigation. Plantes et Paysages d’Afrique: une histoire à explorer, 165–210. Paris: Karthala. Boris, Gilbert 1951 Documents Linguistiques et Ethnographiques sur une Région du Sud Tunisien (Nefzaoua). Paris: Imprimerie Nationale de France. 1958

Lexique du Parler Arabe des Marazig. Paris: Imprimerie Nationale de France. Bouchemit, Rachid 1951 Le Kouriya du Gourara. Bulletin de Liaison Saharienne 5: 46–47. Boudot-Lamotte, Antoine 1964 Notes ethnographiques et linguistiques sur le parler berbère de Timimoun. Journal Asiatique CCLII: 487–558. Cancel, Lt. 1908 Etude sur le dialecte de Tabelbala. Revue Africaine 270–271: 302–347.

230

Lameen Souag

Centre de Musiques Arabes et Méditerranées n.d. Kourktou. Centre de Musiques Arabes et Méditerranées. http://www. Cmam.nat.tn/fo/fr/dynamique/instrument_popup.php?id=29 (7/03/20 12). Charry, Eric S. 1996 Plucked Lutes in West Africa: An Historical Overview. The Galpin Society Journal 49: 3–37. doi:10.2307/842390 (11 March, 2012). 2000 Mande Music: Traditional and Modern Music of the Maninka and Mandinka of Western Africa. (Chicago Studies in Ethnomusicology). Chicago: University of Chicago Press. Cyffer, Norbert, and John Hutchison 1990 Dictionary of the Kanuri Language. (Publications in African Languages and Linguistics 13). Dordrecht: Foris. Delafosse, Maurice 1955 La Langue Mandingue et ses Dialectes (Malinké, Bambara, Dioula). (Bibliothèque de l’Ecole Nationale des Langues Orientales Vivantes). Paris: Geuthner. Delheure, Jean 1984 Agraw n Yiwalen Tum_abt t-Tfransist/Dictionnaire Mozabite-Français. Paris: SELAF. 1987 Agerraw n Iwalen Teggargarent-Tarumit/Dictionnaire OuargliFrançais. Paris: SELAF. Denizeau, Claude 1960 Dictionnaire des Parlers Arabes de Syrie, Liban et Palestine; Supplément au dictionnaire arabe-français de A. Barthélemy. (Études Arabes et Islamiques. Série 3; Études et Documents 3). Paris: Maisonneuve. Dermenghem, Emile 2002 The Black Diwans of Algeria. In The African Diaspora in the Mediterranean Lands of Islam. (Princeton Series on the Middle East), John O. Hunwick, and Eve Troutt Powell (eds.). Princeton: Wiener. Despois, Jean 1946 Mission Scientifique au Fezzan, t. III: Géographie Humaine. Algiers: Chevalier. Destaing, Edmond 1914 Dictionnaire Francais-Berbère: Dialecte des Beni-Snous. (Publications de la Faculté des Lettres d’Alger 49). Paris: Leroux. 1920 Etude sur la Tachel…ît du Sous. Vol. I: Vocabulaire français-berbère. Paris: Leroux. Diagana, Ousmane M. 2011 Dictionnaire Soninké-Français (Mauritanie). Paris: Karthala.

Sub-Saharan lexical influence in North African Arabic

231

Ducroz, Jean-Marie, and Marie-Claire Charles 1978 Lexique Soŋey (Songay)-Français: parler kaado du Gorouol. Paris: L’Harmattan. Dumestre, Gérard 2011 Dictionnaire Bambara-Français, suivi d’un index abrégé françaisbambara. Paris: Karthala. Farmer, Henry G. 1928 A North African Folk Instrument. Journal of the Royal Asiatic Society (New Series) 60(1): 25–34. Foucauld, Charles E. de 1951 Dictionnaire Touareg-Français: dialecte de l’Ahaggar. Paris: Imprimerie Nationale de France. GREDAAL (Groupe de Recherches et d’Études pour le Développement Durable) 2005 Une première lecture des résultats préliminaires du recensement relatif aux élevages en Algérie. GREDAAL. http:/www.gredaal.com/ ddurable/agricelevage/obselevages/publications/autres/Elevage-Algérie-2005.pdf. Haarmann, Ulrich 1998 The Dead Ostrich Life and Trade in Ghadames (Libya) in the Nineteenth Century. Die Welt des Islams 38(1): 9–94. Hannouche, Jamal El2010 Arabic Influence in Ghomara Berber. MA thesis, University of Leiden. Heath, Jeffrey 1998 Dictionnaire Songhay-Anglais-Français. (Langues d’Afrique 4–6). Paris: L’Harmattan. 2004 Hassaniya Arabic (Mali)-English-French Dictionary. (Semitica Viva 33). Wiesbaden: Harrassowitz. 2006 Dictionnaire Touareg du Mali: tamachek-anglais-français. Paris: Karthala. 2010 Dogon and Bangime Linguistics. http://dogonlanguages.org/. Hinds, Martin, and El-Said Badawi 1986 A Dictionary of Egyptian Arabic: Arabic-English. Beirut: Librairie du Liban. Hony, Henry C., Fahir Iz, and Anthony D. Alderson 1992 Oxford Turkish Dictionary. Oxford: Oxford University Press. Hunwick, John O . 2004 The Religious Practices of Black Slaves in the Mediterranean Islamic World. In Slavery on the Frontiers of Islam, Paul E. Lovejoy (ed.). Princeton: Wiener. Huyghe, P. G. 1907 Dictionnaire Chaouia-Arabe-Kabyle & Français. Alger: Jourdan.

232

Lameen Souag

Ibn Manđʖūr, Mujammad 1955 Lisān al-ɄArab [Tongue of the Arabs]. Beirut: Dār nādir. ’Iryānī, Mucahhar ȥAlī al1996 Al-Muȥjam Al-Yamanī fī al-Luġah wa-al-Turāθ [The Yemeni Dictionary of Language and Heritage]. Damascus: Dār al-Fikr. Jaggar, Philip J. 2001 Hausa. (London Oriental and African Language Library 7). Amsterdam: Benjamins. Jakobi, Angelika, and Maarten Kossmann Fc Alleged Berber loan words in Nubian. Sprache und Geschichte in Afrika. Jankowsky, Richard C. 2010 Stambeli: Music, Trance, and Alterity in Tunisia. Chicago: University of Chicago Press. Jarrett, Kevin 2007 A dictionary of Manga, a Kanuri language of Eastern Niger and NE Nigeria. (Ed.) Roger Blench. http://www.rogerblench.info/Language /Nilo-Saharan/Saharan/Manga%2520dictionary%2520Unicode.pdf. Jullien de Pommerol, Patrice 1999 Dictionnaire Arabe Tchadien-Français : suivi d’un index françaisarabe et d’un index des racines arabes. Paris: Karthala. Kossmann, Maarten 2004 Is there a Songhay substratum in Gourara Berber? In Nouvelles Etudes Berbères: Le verbe et autres articles, Rainer Vossen, and Dymitr Ibriszimow (eds.), 51–66. Köln: Köppe. 2005 Berber Loanwords in Hausa. (Berber Studies v. 12). Köln: Köppe. Lanfry, Jacques 1968 Ghadamès I: Textes : Notes philologiques et ethnographiques. Fort-National (Algérie): Le Fichier Périodique. 1973 Ghadamès II: Glossaire (parler des Ayt Waziten). Fort-National (Algérie): Le Fichier Périodique. Laoust, Émile 1920 Mots et Choses Berbères : notes de linguistique et d’ethnographie: dialectes du Maroc. Paris: Challamel. Lukas, Johannes 1937 A Study of the Kanuri Language: Grammar and Vocabulary. Oxford: Oxford University Press. 1953 Die Sprache der Tubu in der Zentralen Sahara. (Veröffentlichung, Deutsch Akademie der Wissenchaften zu Berlin. Institut für Orientforschung Nr. 14). Berlin: Akademie. Madouni-La Peyre, Jihane 2003 Dictionnaire Arabe Algérien-Français: Algérie de l’ouest. Paris: Langues & Mondes - L’Asiathèque.

Sub-Saharan lexical influence in North African Arabic

233

Marçais, William 1958 Textes Arabes de Takroûna: II Glossaire: Contribution à l’étude du vocabulaire arabe. Paris: Imprimerie/Bibliothèque de l’École des Languages Orientales Vivantes. Mason, Ian L. 1980 Prolific Tropical Sheep. Rome: Food and Agriculture Organization of the United Nations. http://www.fao.org/DOCREP/004/X6517E/ X6517E00.htm. Matsushita, Shuji 1974 A Comparative Vocabulary of Gwandara Dialects. (African Languages and Ethnography). Tokyo: Institute for the Study of Languages and Cultures of Asia and Africa. 1992 When did Gwandara split from Hausa? African Urban Studies 2: 43–53. McDougall, Ann 1992 Salt, Saharans, and the Trans-Saharan Slave Trade: Nineteenth Century Developments. In The Human Commodity: Perspectives on the Trans-Saharan Slave Trade, Elizabeth Savage (ed.). London: Frank Cass. Mitchell, Terence F. 1957 Some properties of Zuara nouns. Memorial André Basset (1895– 1956). Paris: Maisonneuve. Nachtigal, Gustav 1987 Reprint. Sahara and Sudan. Translated by Allen B. Fisher and Humphrey J. Fisher. London: Hurst. Original edition, 1971. Nehlil, Mohamed 1909 Étude sur le Dialecte de Ghat. Paris: Ernest Leroux. Newman, Paul 1977 Chadic classification and reconstructions. Afroasiatic Linguistics 5 (1): 1–42. Niang, Mamadou O. 1997 Pulaar-English/English-Pulaar. (Hippocrene Standard Dictionary). New York: Hippocrene Books. Nicolaisen, Johannes 1963 Ecology and Culture of the Pastoral Tuareg: With Particular Reference to the Tuareg of Ahaggar and Ayr. (Nationalmuseets Skrifter 9). Copenhagen: National Museum of Copenhagen. Owens, Jonathan 1984 A Short Reference Grammar of Eastern Libyan Arabic. Wiesbaden: Harrassowitz. 2000 Loanwords in Nigerian Arabic: a quantitative approach. In Arabic as a Minority Language, Jonathan Owens (ed.), 259–346. Berlin: Mouton de Gruyter.

234

Lameen Souag

2004

Remarks on Ideophones in Nigerian Arabic. In Approaches to Arabic Dialects: A Collection of Articles Presented to Manfred Woidich on the Occasion of His Sixtieth Birthday, Martine Haak, Rudolf Erik de Jong, and Kees Versteegh (eds.), 207–220. Leiden: Brill. Pâques, Viviana 1964 L’Arbre Cosmique dans la Pensée Populaire et dans la Vie Quotidienne du Nord-Ouest Africain. Paris: Institut d’Ethnologie. Paradisi, Umberto 1960 Il berbero di Augila, Materiale lessicale. Rivista degli Studi Orientali XXXV: 157–177. Paris, Harry S. 2001 History of the cultivar-groups of Cucurbita pepo. Horticultural Reviews 25. Pasch, Helma 1983 Verbreitung des Maises in Afrika. Sprache und Geschichte in Afrika 5: 177–218. Pauphilet, Didier. 1953 Les Daouada. In Mission au Fezzân 1949, Didier Pauphilet, and Pierre Bellair (eds.), 115–132. Tunis: Imprimerie officielle de la Tunisie. Pickersgill, Barbara 2005 Spices. In The Cultural History of Plants, Ghillean Prance (ed.), 153–172. London: Routledge. Prémare, Alfred-Louis de 1993 Dictionnaire Arabe-Français. Paris: L’Harmattan. Qāsim, ‘Awn al-Šarīf 1972 Qāmūs al-Lahjah al-‘Āmmiyyah fī al-Sūdān [Dictionary of the Colloquial Dialect in the Sudan]. Beirut: Dār al-Fikr. Le Quellec, Jean-Loïc 2003 Yalla: Méthode d’Arabe Libyen (Tripolitaine et Fezzân). Paris: Harmattan. Rahal, Ahmed 2000 La Communauté Noire de Tunis: Thérapie initiatique et rite de possession. (Collection La Bibliothèque d’Africultures). Paris: L’Harmattan. Rohlfs, Gerhard 1881 Kufra. Reise von Tripolis nach der oase Kufra. Leipzig: Brockhaus. Roth-Laly, Arlette 1969 Lexique des Parlers Arabes Tchado-Soudanais. Paris: CNRS.

Sub-Saharan lexical influence in North African Arabic Sahli, Ali 2008

235

Mu`jam Amāzīγī-`Arabī (xāUU bi-lahjat ‘ahālī Fijīj) ya^ummu qawā`id hāđihi l-lahjah wa-jāniban min turāŧihā l-’adabī [ArabicAmazigh Dictionary (specific to the dialect of Figuig) including the grammar of this dialect and an aspect of its literary heritage]. Oujda: El Anouar El Maghribia. Saidi, C., M. Mahouachi, N. Atti, and N. Mathlouthi 2011 Etude de la croissance, la qualité de la carcasse et de la viande des agneaux de deux génotypes. Options Méditerranéennes A (97): 81–85. http://ressources.ciheam.org/om/pdf/a97/00801451.pdf (8 March, 2012). Schmidt, Christopher K. 2009 Loanwords in Japanese. In Loanwords in the World’s Languages: A Comparative Hand-book, Martin Haspelmath, and Uri Tadmor (eds.), 545–574. Berlin: Mouton De Gruyter. Schuh, Russell 2009a Bade-English-Hausa Dictionary. Yobe Language Research Project. http://www.humnet.ucla.edu/humnet/aflang/Yobe/Dictionaries/Karek are_english_hausa_2009.pdf. 2009b Ngizim-English-Hausa Dictionary. Yobe Language Research Project. http://www.humnet.ucla.edu/humnet/aflang/Yobe/Dictionaries/Karek are_english_hausa_2009.pdf. 2009c Karekare-English-Hausa Dictionary. Yobe Language Research Project.http://www.humnet.ucla.edu/humnet/aflang/Yobe/Dictionaries/Karekare_english_hausa_2009.pdf. Schuyler, Philip 1979 A repertory of ideas: The music of the Rwais, Berber professional musicians from southwestern Morocco. PhD thesis, University of Washington. Skinner, Neil 1996 Hausa Comparative Dictionary. (Westafrikanische Studien, Frankfurter Beiträge zur Sprach- und Kulturgeschichte 11). Köln: Köppe. Souag, Lameen 2010 Morphosyntactic borrowing in the Sahara. PhD thesis, School of Oriental and African Studies (University of London). 2012 The Subclassification of Songhay. Journal of African Languages and Linguistics 33(2): 181–213. Stoller, Paul 1989 Fusion of the Worlds: An Ethnology of Possession Among the Songhay of Niger. Chicago/London: University of Chicago Press.

236

Lameen Souag

Sudlow, David 2009 Dictionary of the Tamasheq of North-East Burkina Faso. (Berber Studies 24). Köln: Köppe. Taine-Cheikh, Catherine 1988 Dictionnaire ”assaniyya-Français. Paris: Geuthner. Tremearne, Arthur J. N. 1914 The Ban of the Bori: Demons and Dancing in West and North Africa. London: Heath. Versteegh, Cornelius H. M. (ed.). 2005 Encyclopedia of Arabic Language and Linguistics. Leiden: Brill. Vogel, Eduard 1854 Dr. E. Vogel and Dr. B. Seemann - Murzuk, October 8, 1853. Bonplandia II(1): 2–5. Wehr, Hans 1974 A Dictionary of Modern Written Arabic. Beirut: Librairie du Liban. Wright, John L. 2006 The Trans-Saharan Slave Trade. London: Routledge. Zawadowski, Gaston 1942 Le rôle des nègres parmi la population tunisienne. En Terre d’Islam 1942: 146–152. Zouyed, Ilhem 2005 Engraissement des ovins : Caractéristiques des carcasses et modèle de classification. Constantine: Université de Mentouri. http://bu.umc. edu.dz/theses/veterinaire/ZOU4251.pdf. Zuwārī, ɄAlī 1998 MuɄjam al-Kalimāt wa-al-Taqālīd al-ShaɄbīyah bi-•afāqis [Dictionary of Words and Popular Traditions in Sfax]. nafāqis: al-MuɅallifayn.

Lexical aspects aspects of Maghrebi Arabic Peter Behnstedt

1. Introduction The differences between western (Maghrebi) and eastern Arabic dialects are considerable and the problems of mutual understanding were explained mainly by lexical differences and illustrated with nineteen maps from the ‘Wortatlas der arabischen Dialekte’ vols. I and II (= WAD I, WAD II). For publication, the fully colored maps which were presented, some of them with more than fifty different symbols, had to be transformed using black, white, grey tones and hatchings, and had to be reduced in format and considerably simplified. Some of the maps are too complicated to be presented here in a simplified and reduced form. Problems of understanding may result from the so-called “false friends” like xalla ‘to pay’ in Maghrebi Arabic, meaning ‘to finish’ in eastern Arabic dialects; bisbās ‘fennel’ in Maghrebi Arabic, signifying ‘macis’ in eastern dialects and even ‘hot peppers’ in Yemeni Arabic. ma allāt being used by an Iraqi may mean ‘quarters (of a town)’ and may be completely misunderstood by a Maghrebi as ‘shops’. Maghrebis often do not understand the Eastern ašāyir ‘tribes’ or muxtā ‘mayor’, and vice versa an Easterner would hardly understand an Algerian būši ‘butcher’ from French boucher. There are words and expressions to be avoided in the Maghreb like the Levantine ya īk ƽl-āfye ‘thank you! (May God give you health!)’, or at least in Morocco ābūn ‘baking oven’, āfye being an euphemism for nā ‘fire (hell fire)’, and ābūn a taboo word (i.e. ‘pudendum mulieris’). The list of “false friends” could be continued with many other examples, but to my knowledge the subject has not yet been dealt with systematically. Other examples are observed in internet forums, such as Egyptian Arabic dulāb ‘cupboard’, which in Levantine Arabic is ‘wheel, tyre’. The reasons for the lexical bipartition Maghreb – Mashreq are multiple. Amongst them has to be mentioned the role of different linguistic substrata, superstrata and adstrata. As for the substrata, there is Berber in the Maghreb, Coptic in Egypt (the role of which is rather insignificant), Aramaic in the Levant and Iraq, South Arabic and, last but not least, African substra-

238

Peter Behnstedt

ta in Chad, Sudan, Mauritania, etc. On the other hand, perhaps the role of different colonial languages as superstrata is as important as substrata languages. Borrowings from different languages in Arabic dialects are mentioned in various articles in the Encyclopedia of Arabic Language and Linguistics, but not in a systematic way. Berber, French and Italian loanwords in Maghrebi Arabic are dealt with, but Spanish as a source of borrowing is ignored. The most comprehensive article about loanwords is Vollers' (1886) study of Egyptian Arabic. However, no general investigation of loanwords in Arabic dialects exists until now. If we take Classical Arabic (= CA) as a point of historical reference (not as an origin of the dialects!), we deal with a period of some 1400 years of linguistic history and, as for the geographical distances between actual dialect areas, they count thousands of kilometers. It is therefore quite natural that changes have happened. Still with reference to CA, one can observe that the Maghreb often has made another choice out of the many synonyms or pseudo-synonyms to be found in the dictionaries of CA than the Mashreq. Perhaps many of those synonyms at the time of their compilation by the lexicographers had already been regional variants. This was illustrated by maps with the designations for ‘sesame’, ‘threshing place’, ‘yoke’, and ‘bee-hive’, and it was shown that some Maghrebi forms hark back to Yemeni dialectal forms 2. Mediterranean lexical contact effects on Maghrebi Arabic The Maghreb is part of this big cultural area: the Mediterranean. It is therefore natural that its dialects contain vocabulary from the other Mediterranean languages. It is different from other cultural areas like Iraq and the Gulf or South Arabia with much more lexicon from other linguistic adstrata like Persian and Urdu for example. Especially in the domain of “the things” the Maghrebi dialects are very Mediterranean with loanwords from the Vulgar Latin then spoken in Northern Africa, from the Romance languages such as Spanish, Italian, French, the lingua franca, or from Greek and Turkish. Often those loanwords are indirect ones, e.g. the Spanish terminology of the “fes” makers of Tunisia, which has been introduced by the Moriscos expelled from Spain (cf. Singer 1979). This is true for other vocabulary, too. De Prémare derives Moroccan qāmƽžža, qmƽžža, qmƽLJLJa ‘shirt’ from

Lexical aspects of Maghrebi Arabic

239

Spanish camisa, which for phonological reasons is rather improbable. An Andalusian Arabic form camija is more obvious. Some of the lexical differences may be explained by semantic change such as extension and restriction or specialization. The first example (map 1, map 115 in WAD I) is ‘fish’. In CA samak is the generic term for ‘fish’, so is ūt. ūt in CA may refer to small fish as in the sura al-Kahf, or to a big fish as in the sura a -.affāt, where Jonah and the whale are mentioned (‘Then the big Fish did swallow him’). Due to that legend, ūt by restriction has come to mean ‘whale’ in Eastern dialects whilst in Maghrebi Arabic it has kept its generic meaning. From the fact that ūt is also used generically in Southern Arabia and in the LiLJāz, one may conclude that originally ūt was a regional variant and that this form was introduced into the Maghreb by tribes from Western and Southern Arabia. The homeland of the Bani Hilāl, indeed, was NaLJd and the LiLJāz, and amongst the tribes of the first Arabic invasion of the Maghreb were definitely Yemeni tribes. Map 2 (WAD II map 192): Another example of different semantic evolutions is a loanword from Persian pingān ‘clepsydra’, a kind of bowl-shaped outflow for measuring time. pingān > finLJān almost everywhere is nowadays ‘a cup, a bowl’, but pingān in Tunisia has given mungāla ‘clock, watch’ and similarly māgāna in Morocco. As for Tunisian and Moroccan Arabic, the form must have been borrowed a second time, but now as fƽnžāl (Singer 1984: 545 ‘Kaffeetäßchen’ = ‘small cup for coffee’) and fƽnžān (De Prémare 1998: 164) as a poetical form for fƽnžƽl ‘godet, tasse, gobelet, verre’ = ‘jar, tumbler, cup, glass’. The Lassaniyya dialect has derived its designation waqqāta from waqt ‘time’ and Maltese has borrowed its form arloġġ from Italian orologio. A third example which was presented with two maps is mūs, which as in CA everywhere designates ‘a razor blade’ or ‘a razor’ but which in the Maghreb by extension denotes also other types of knives, namely ‘a knife with a short blade’, ‘a table knife’, ‘a folding knife’, rarely ‘a bigger knife such as that of a butcher’. On the other hand, the generic oriental form sikkīn ‘knife’ in the Maghreb is ‘sabre’ or ‘dagger’ (Algeria, Morocco) with a shift of meaning. The two maps cannot be reproduced here. See WAD II maps 213a and 213b (cf. also Heath 2002: 109–111). It was postulated above that many synonyms of Classical Arabic at the time of their compilation by the lexicographers may have been regional

240

Peter Behnstedt

synonyms. This was illustrated by maps with the designations for ‘sesame’, ‘threshing place’, ‘yoke’, ‘bee-hive’, and it was shown that some Maghrebi forms hark back to Yemeni dialectal forms. Map 3: There is a neat regional distribution of the two CA designations for ‘sesame’ LJulLJulān and simsim in the dialects. In the Maghreb, there exist only reflexes of LJulLJulān, in the East reflexes of simsim. Yemen is the exception. One conclusion may be that LJulLJulān originally was more frequently used in Yemen and brought to the Maghreb by Yemeni tribes. Map 4 (map 273 in WAD II): That Arabic dialects already existed at the time of the Islamic conquests is a well known fact; sometimes the lexicographers of CA even indicate a form as dialectal. This is the case for the denominations of ‘threshing place/floor’. The form LJarīn is classified as Yemeni, LJurn as Egyptian, andar as Syrian and baydar as Iraqi. It has to be kept in mind that ‘Iraq’ originally meant ‘Mesopotamia’, thus including the Eastern part of actual Syria. The strange thing is that the “Syrian” form is not to be found in Syria anymore. All over the Levant the “Iraqi” baydar is used. andar is maintained in Maltese and with a change of scheme as mandra in Libya and Tunisia, as nādƽr in Algeria and Morocco where it also designates a ‘heap of wheat’ on the threshing place. Besides, there are cases of specialization or shift of meaning as CA ra ba ‘court, open area’, qā‘a ‘paved courtyard’, Lassaniyya barza ‘flat, public place’, blayda diminutive of blād ‘open, empty space’. Map 5 (map 268 in WAD II): Amongst the correspondences between Yemeni and Maghrebi Arabic are designations for ‘yoke’. In the Western part of Yemen and the South-west of Saudi-Arabia (Asir and Jazan province) it is called d3umād or mad3mid, in CA mi4mada ‘a sort of yoke’ (LANE 1980 s.v.). Reflexes of this form are to be found in Egyptian Arabic as 4an4 in the Fayyūm and Bani Swayf area where at the time of the conquest mainly Yemeni tribes had settled. Furthermore, they are found in Andalusian Arabic as 4amd, ma4mad, mi4mad, in Maltese as madmad, in Algeria (Djidjelli) as mo mo , in Beaussier (2006) as 4md, m4amd, and in Morocco mƽ4mƽ4. The latter, according to dialect, may refer to a modern vineyard plough, too, or as a diminutive form m4ēmda ‘sous-ventrière en bois’, a kind of pole attached under the belly of the animal where the towing chains are attached and which is used instead of a yoke. This device is called in some dialects zāglu or zāylu from Berber azaglu, a borrowing from Latin iugulum ‘small yoke’, still attested for the dialect of Skoura as zāglu

Lexical aspects of Maghrebi Arabic

241

‘yoke’. Designations for ‘yoke’ are not very frequently attested in Maghrebi Arabic, simply because of the fact that normally a collar is used for attaching the plough, not a yoke. Map 6 (map 277 in WAD II): For ‘beehive’ the MSA designations given in Wehr (1980) are man al and kuwāra, in Corriente (2010) xaliyya, ‘assāla, qafīr, kuwāra, LJab . Al-Selwi (1987: 56) supposes that LJib is a Yemeni form, but the Lisān al-‘Arab does not define it as such. A look at the map shows a clear lexical relation between the Maghreb (including Andalusian Arabic, Malta) and Yemen. As for Maltese ġabsala, it is a combination of two words meaning ‘hive’, namely LJab and ‘assāla. Map 7 (map 270 in WAD II): Another Yemeni contribution already discussed by various scholars (Theodor Nöldeke, Ewald Wagner and others) are forms for a slingshot, namely wa4af and wa4āfa in Andalusian Arabic (Dozy 1968 s.v.), forms which are not atttested in CA but in Ethiopian. In Yemen, the forms are wud3uf, wud3af, wad3if, wačfah; mīd3af, mīd3afah, in south-western Saudi Arabia, wačfah, mīd3afah and n-doublets in LaWramawt, nad3af and mind3afah. In Tunisia, wad3af is attested as well as ud3af (in Takrouna); in Maltese, wadab ~ wadaf ~ waddafa; in Djidjelli (Algeria), awƽl of via a Berber form *l-wƽl of > wƽl of > awƽl of; in the Jbala dialects of Morocco, wu āf. The original map shows some 50 forms which cannot be presented here for technical reasons. Amongst them are loanwords from French in the Maghreb like gomme carrée > gōnkārē, élastique > lƽstāk, stīk, tire-boules/boulettes > tīrbūla, tīrbūlāt, and in Maltese, vleġġa, żbandola from Sicilian Italian fileccia ‘dart’ and sbandola ‘slingshot’. This leads to another theme, namely loanwords from almost all the Mediterranean languages in the Maghrebi dialects. Map 8: stork. For ‘stork’, except the Lassaniyya, Maghrebi Arabic uses reflexes of Greek πελαργ`ς: bƽllārƽž, bƽrrārƽž. The way this word became part of Maghrebi vocabulary is not very clear. At the time of the Islamic conquest of Africa Greek was spoken in towns in Egypt, Greek colonies existed in the Cyrenaica and the Byzantines had dominated actual Tunisia from 533 to 698. Perhaps the word was borrowed in Ifrīqiya, where in Cairouan, according to Singer (1991), a Maghrebi koiné originated and then was spread to other towns in North Africa. In Tunisia, this bird is also called āžž Qāsim and in Egypt agg ’āsim, i.e., it is compared to a ‘pilgrim’. A certain similarity between bƽllārƽž with Latin peregrinus

242

Peter Behnstedt

‘stranger, traveler’, Late Latin pelegrinus is probably a hazard. For Tlemcen (Algeria), an onomatopoetic form like the Eastern forms laqlaq, laklak, laglag is attested, namely bu šƽqšāq; also in some Berber dialects of the area as abšƽqšaq. The Lassaniyya of Mauritania form sounds like a Berber loanword, but designates another bird in the Lassaniyya of Mali, a ‘little egret’. Map 9 (map 91 in WAD I): Surprisingly, there are quite many loanwords for animals in Maghrebi Arabic, not only those from Berber like fƽkrūn ‘turtle’, žrāna ‘frog’. The origin of the form ‘atrūs ‘billy-goat’ for some is enigmatic, for others it is Greek τράγος. The simplified map shows that Libya for this word has the Eastern form and does not go with the rest of the Maghreb. Malta is apart with bodbod which most probably is onomotapoetic. ‘atrūs is not attested for Andalusian Arabic. Map 10 (map 106 in WAD I): Also for another form typical for Maltese, Tunisia and parts of Algeria, namely sardūk, serduq ‘cock’, no satisfactory etymology has been proposed until now. The almost pan-Arabic dīk has been borrowed by some Berber dialects of Northern Morocco and then passed again to Arabic as adkōk. The same is true for fƽ ūž which in oriental dialects normally designates a young cock or a roast chicken. As a Berberized form, ābƽrrūg is to be found in some Moroccan Jbala dialects. Map 11 (map 107 in WDA I): As for ‘chicks’, the etymology of Maghrebi fƽllūs, however, is clear. It is Latin pullus. Most probably the word was borrowed via Berber afƽllus (not mentioned in Schuchardt 1918) in which loanwords from Latin still have the nominative ending -us like asnus from asinus ‘donkey, small donkey’, ifires, lfiras from pirus ‘pear’. By the 7th century, the nominative in the proto-romance languages normally had been replaced by the accusative. The Moroccan Jbala dialects have the onomatopoetic forms čwīčo, čōčo, to be compared with Libyan šawšaw, Sudanese siwsiw and Levantine ū . The Lassaniyya form farkūs (varkūs) in Eastern Mauritania is ‘young guinea fowl’, in Morocco and Algeria ‘young partridge’ or ‘young quail’. It is compared in WAD I: 316 to the Arabic roots fqs and fq ‘to break an egg’ or ‘to hatch from an egg’. Perhaps a certain similarity with farrūLJ ‘young cock’ should have been mentioned, and why not Latin perdix ‘partridge’?

Lexical aspects of Maghrebi Arabic

243

Map 12 (map 111 in WAD I): For ‘rabbit’, Algerian and Moroccan Arabic use forms like qunn, qniyya, gunn, gniyya, gnūn, gnīn most probably via Berber (cf. agnīn, ?agnīnt, ?aqnenneç? and the like) from Latin cuniculus which itself is a loanword from an Iberian language. Forms like qunn may reflect a form without the diminutive ending, i.e. *cun-, others like qniyya Latin cunicula. Beaussier (2006 s.v.) mentions a partly vocalized form qllīn for Morocco and Tunisia. The form was overlooked and is not to be found on the original map nr. 111 in WAD I. Beaussier’s form is not attested in De Prémare for Morocco, only qlāin for Tangiers, qlayna in our own materials. On the map, the symbol for this form in Tunisia is arbitrarily placed in the centre of the country. Map 13 (map 116 in WAD I): The ichthyonomy of the Mediterranean is well documented by the work of Giovanni Oman (1966). The names of shrimps and the like quite nicely illustrate the links of Arabic with its neighbouring languages. On the Arabian peninsula, except the Levant, a word of Persian origin is used, i.e. rubyān, in the Levant we have forms like qraydes, ’rēdes from Greek γαρίδες via Turkish karides. From Italian gamberi are derived Egyptian Arabic gambari, Libyan gƽmbri, gambari, Tunisian gambri, gambli and the like, Maltese gamblu, gambru, Eastern Algerian qumbri, Northern Morocco Spanish gambas. Except for Libya, French crevettes is used all over the Maghreb. Oman (1966 passim) gives other forms for members of the familiy like (bū) qamrūn for Algerian Arabic from Spanish camarón ‘palaemon serratus’, matsangūn from Italian mazzancollo ‘penaeus caramote’. Map 14 (map 200 in WAD II): For ‘kitchen’ the original map presented shows more than fifty designations. It is impossible to represent them all in the given format. The main contrast is the Maghreb with loanwords from French, Spanish and Italian, namely kučīna, kūzīna and similar forms, and the Mashreq with an Arabic form ma bax and many others especially in the south-west of the Arabian Peninsula. The forms of Maltese kċina, Libyan Arabic kučīna, kūžīna, Tunisian Arabic kšīna, kužīna are derived in the sources from Italian cucina, the Algerian and Morrocan forms of the type kūzīna from French cuisine, also Western and Southern Lassaniyya kuzīn, kwayzīn. As for Moroccan keššīna, kuššīna, kutšīna, De Prémare s.v. derives it from Spanish cocina, which is problematic since /šš/ represents a Spanish [θ]. So is also an older pronunciation of |c|, namely [ts] (till the 16th century), since this normally corresponds to /n/ in Arabic like in

244

Peter Behnstedt

zapato > ubbā ‘shoe’. Besides these borrowings there are other forms in the Maghreb often referring to an annex used as a kitchen: a tent or a hut and the like, such as nuwwāla, dwīra, xyāma, žīr, amongst them also loanwords from Berber like āsfālu in Skoura/Morocco or afalli, valli, ta nīt in Mauritania. French potager ‘vegetable garden’ may be used in some Algerian dialects as ‘kitchen’, too. Those forms are neglected on the map. The lexical impact of the colonial languages, especially French, on Maghrebi Arabic is indeed enormous. This fact is well known and need not be discussed any further (Cf. Kropfitsch 1978; Paradis and LaCharité 2007). Map 15 (maps 214a, 214b in WAD II): As for loanwords from Turkish, due to a longer colonial domination in the East, obviously they are more frequent in oriental Arabic, but even in Morocco without any Ottoman domination Turkish words can be found. This was illustrated by the designations of ‘spoon’. Reflexes of Turkish kaşik are dominant on the Arabian Peninsula as xāšūqa, xāšūga, except the south. In Libya, the reflex is xāšīk. The Moroccan ‘āšƽq (an older form is ġāšƽq) normally denotes a small spoon and might have been influenced by √‘lq for ‘spoon’. The map shows many other borrowings like kūčā a used in Northern Morocco and the Western Sahara from Spanish cuchara, kƽddu partly in Mauritania from Wolof, tasūkālƽt from Berber. The mainly Tunisian form ġanžāya is considered by some as being of Berber origin (cf. aġenža ‘ladle’), a Romance etymology gancho, gancio is discussed in WAD II: 134. The form muġ ƽf, attested mainly for Algeria, is elsewhere a ‘ladle’. Map 16 (map 235 in WAD II): Turkish, Persian, or Persian via Turkish, have furnished most of the designations for ‘socks’, e.g. LJōrāb, šu āb, etc. in the East. There is no Arabic form for it attested in the dialects, and the forms used have undergone shifts of meaning. The designations of the Maghreb, tƽqšīra, normally used in the plural tqāšƽr, in Eastern Algeria also šƽqšīr and in Libya šaxšīr, originally refer to large trousers with light shoes sewed on. Especially in Syria, qalšīn is used from Italian calzino via Turkish kalçin ‘long, thick felt trousers’. The form of the Gulf and Oman dlāġ (more rarely zlāġ) derives from Turkish dolak ‘ankle gater’, rāna, rānat in Eastern Syria from Persian rānin ‘trousers’. The Levantine forms kalse and the Tunisian/Maltese kalsī a, kalzetta from Italian calza, calzetta. For more forms see the original map.

Lexical aspects of Maghrebi Arabic

245

3. Lexical homogeneity of Maghrebi Arabic Another aspect is that of a greater homogeneity of Maghrebi Arabic vis-àvis Eastern dialects. This was compared to the “dialectal” situation in the USA and Great Britain. There is more dialectal variation in the homeland than in the ex-colony which is explained by leveling and mixing of dialects. “Colonial Arabic” – to use a term of Otto Jastrow – is more uniform in many respects. This applies for instance to morphology: there is only one form for the 1st person perfect: ktabt or ktƽbt, vis-à-vis katabt, ktabt, katabtu, katabta, čatabt, etc. in Eastern dialects, not to mention Yemeni forms like katabk, katubk, katbuk. As an example of the lexical homogeneity, the designations for ‘plough’ were chosen. All over North Africa, including Egypt, the dominating form is mu āt with little variation. In the East, besides mu āt (rather rare) some 50 different designations are to be found. The reader is referred to the original maps 266a and 266b in WAD II.1 In Morocco, besides mu āt, mƽ4mƽ4 is used for a modern French plough. As in the East, mainly Palestine, partly Syria, ‘ūd is attested sporadically in Morocco and Algeria. For Dellys (Algeria), the form šā yu was indicated, probably a contamination of French charrue with chariot ‘cart’. Some forms were overlooked in Beaussier, namely Eastern Algeria kābša for ‘hoe’ or a ‘plough for ploughing trees’ (‘charrue pour labourer les arbres’), sƽkka ‘plough’, a pars pro toto, since it normally denotes the plough-share. This form is also attested in Syria and Palestine. A third form is LJābda ‘pair of ploughing oxen’, ‘plough’ from √LJbd ‘to pull’. 4. The Berber substratum substratum of Maghrebi Arabic Finally the role of the Berber substratum is discussed here. According to Thomason and Kaufman (1988: 20) loanwords are simply “borrowings” and have nothing to do with substratum which is defined as “imperfect group learning during a process of language shift” (p.38).

1.

On the original map 266a the symbol for the Anatolian form kōtān is to be found again in Eastern Algeria. This is an error which occurred when copying the symbol from the legend to the map itself.

246

Peter Behnstedt

An effect of the Berber substratum is, however, the well-known example mā ‘water’ which is considered a plural as in Berber: mā suxnīn ‘hot water’ vs. mayy suxne e.g. in Syria. Map 17 (map 224 in WAD II): As for the number of Berber loanwords, the impression was uttered that their amount decreases from West to East and that it is most pronounced in Lassaniyya and in Moroccan Arabic. This was first illustrated with a map for ‘key’, where only Moroccan Arabic and partly Lassaniyya use Berber sārūt. Map 18 (map 53 in WAD I): The map for ‘elbow’ shows a more restricted area for a Berber loanword, namely ġumra, ġummā a and tāġummā t in the North Moroccan Jbala dialects and partly in Eastern Algeria. Schuchardt (1918: 43) considers this as a borrowing from Latin camur(us) ‘krumm’ = ‘crooked, bent’. For publication, this map had to be radically simplified. Next to the predominant Maghrebi form mƽrfƽq (and the like) an isolated form kūd for Algiers was interpreted as a loan from French coude. Meanwhile the form kū is attested for Dellys (Algeria; in a questionnaire for volume I submitted after publication), it is also to be found in Beaussier (2006 s.v.) and had been overlooked. Schuchardt (1918: 43) considers this form, too, as French coude. Another neglected form was qub āl attested in the dictionaries (De Prémare 1998, Beaussier 2006, Corriente 1997, Dozy 1968, also Heath 2002: 87), namely for Fes (Morocco), Algeria (there also qu bāl, kub āl), Andalusian Arabic qub āl, from Latin *cubitellus and according to Schuchardt (1918: 43) influenced in form by Spanish cobdal, codal from cubitalis. As for Fes, none of the younger and older persons asked recently about ‘elbow’ again ever had heard it. The current form elicited for Fes was mƽrfƽq. This is corroborated by Heath (2002: 87): “qb al ... very rare, recorded from one speaker each from Fs [Fes] and Sf [Sefrou] (and not recognized by young Fs speakers)”. In WAD II: 500, kū is compared with mkawwat ‘hunch-backed’ in Lassaniyya (Heath 2004: 120) belonging to the root √kwt ‘to bend’. As for the predominant Eastern form kū‘, it refers in Maghrebi Arabic to another joint, namely the ‘metacarpus’, in CA to “the extremity of the radius, or bone of the fore-arm, next the thumb” (Lane 1980 s.v.). Maltes minkeb is related in Aquilina (1990: 913) to CA mankab ‘shoulder of men and beasts’ and a root nkb ‘to incline’. Also in the Lassaniyya dialect mƽnkƽb is ‘shoulder’, in Jewish Arabic of Djerba (Tunisia), however, it is attested as ‘elbow’.

Lexical aspects of Maghrebi Arabic

247

Map 19 (map 63 in WAD I): A last map shows that awrƽz for ‘heel’ is mainly to be found in the Jbala dialects of Northern Morocco, where the Berber substratum, indeed, is very pronounced as illustrated, for instance, by pronunciations like zī? ‘oil’ or dīç ‘cock’. The original map (WAD I: 174) ignores awrƽz in Algerian Djidjelli (Marçais 1956: 303). The commentary refers to a detailed map for Northern Morocco in Behnstedt (2007: 52). There is no such map, only the commentary was published there (p. 56), not the map itself due to a confusion of two maps. The forms elicited were awrƽz, wrƽz, wƽrz; Heath (2002: 89) writes awƽrz ~ awriz (read awrīz, also in Berber of Ghomara awriz). For Berber forms other than awrƽz see Lafkioui (2007: 253, map 312). awrƽz on this map is limited to the West and adjacent to the Arabic area with this form. The predominant Maghrebi form gdƽm with variants like gidam, partly in central Egypt gadam, shows a shift of the CA meaning of qadam “the human foot from the ankle downwards” (Lane 1980: 2985). 5. Conclusion As stated in Behnstedt and Woidich (2005: 26), the formation of the modern Arabic dialects has to be explained by polygenesis. This is particularly true for the lexicon. Or as Singer (1984: 5) has put it: the bulk of the Maghrebi lexicon is of Arabic stock, but the Western Arabic dialects have made another choice of synonyms of CA than in the East. Moreover, Western Arabic mainly borrows from Berber and the Romance languages, whereas Eastern Arabic has more loans of Greek, Aramaic, Persian or Turkish.

+

Map 1.

Fish

001

Fish

˙la πabb† mahiyye

ßayd Ù ˙awt (Mehri)

bali∆a

ßayd, ß„d samak, sima≥, etc. ßayd Ù ˙et (ÁarsÂsi)

Niger

bal∆

Mali

Algeria

˙Ât

Mauritania

Morocco

Rabat

Algiers

Libya

Malta

Tripoli

Cameroon

Nigeria

Tunis

Chad Sudan

Uganda

Southern Sudan

Egypt

Syria

Kenya

Mecca

Medina

Iraq

Da⁄nah

Saudi Arabia

Riad

Yemen

Iran

Dubai Abu flabi

Oman

5

+

12

Afghanistan

Uzbekistan

248 Peter Behnstedt

Map 2.

Watch

002

Clock

arlo∆∆

agøgo, agøga

waqqåta

mågåna

mungåla

så©a

Mauritania

Mali

Morocco

Rabat

Algeria

Algiers

Niger Chad

Libya

Malta Tripoli

Cameroon

Nigeria

Tunis

Uganda

Southern Sudan

Sudan

Egypt

Syria

Kenya

Mecca

Medina

Iraq

Da⁄nah

Saudi Arabia

Riad

Yemen

Iran

Dubai Abu flabi

Oman

5

12

Afghanistan

Uzbekistan

Lexical aspects of Maghrebi Arabic 249

Map 3.

E F G

A B C D

Sesame

003

Mali

G

Sesame

≤»l≤lån, Û»lÛlån Û»lÛlåna ∆ul∆lien Ù ∆in∆lien ≤»n≤lån, Û»nÛlån

julånu ≤ul≤ulån, ≤il≤ilån, ≤ul≤ilån

gulgul, ≤il≤il, yilyil

simsim, sumsum, etc.

Mauritania

G,H,I

Morocco

Rabat

A

H I

Algeria

w»lÛlån

Tunis

D

Cameroon

D

Chad

Libya

Malta

Tripoli

F

Nigeria

D

Niger

E

z»nÛlån

D

Algiers

Uganda

Southern Sudan

Sudan

Egypt

Syria

Kenya

Mecca

Medina

Iraq

Da⁄nah

A,B,C

Saudi Arabia

Riad

Yemen

Iran

Dubai Abu flabi

Oman

5

12

Afghanistan

Uzbekistan

250 Peter Behnstedt

+

barza, blayda

Mauritania

Map 4.

+

+

+

+

Mali

Morocco

Rabat

Threshing floor

ra˙ba

qå©a, gå©a, qiegÓa

baydar, b„dar

andar, nåd»r, mandra

Threshing floor

004

Algeria

+

Algiers

Niger

Cameroon

madagg

Chad

Libya

Malta

Tripoli

+

Nigeria

Tunis

+

Uganda

Southern Sudan

taga

Sudan

≤urn

Egypt

≤urn

Iraq

Mecca

Medina

Kenya

≤urn

Syria

Da⁄nah

≤urn mi≤rån

≤arn

Saudi Arabia

Riad

Yemen

Iran

Dubai Abu flabi

5

Oman

gannÂr

12

Afghanistan

Uzbekistan

Lexical aspects of Maghrebi Arabic 251

Map 5.

Yoke

005

zåglu

other forms, or not used

Yoke

+

Mali

Algeria

Niger

∂amd, ‹umåd, ma‹mid, madmad, etc.

Mauritania

+

Morocco

Rabat

Algiers

Chad

Libya

Malta Tripoli

Cameroon

Nigeria

Tunis

Uganda

Southern Sudan

Sudan

Egypt

Syria

Kenya

Mecca

Medina

Iraq

Da⁄nah

Saudi Arabia

Riad

Yemen

Iran

Dubai Abu flabi

Oman

5

12

Afghanistan

Uzbekistan

252 Peter Behnstedt

Map 6.

beehive

006

Mali

Beehive

other forms

kwåra, kÂra

∆absala

≤ab˙, ≤ba˙, Û»b˙ etc.

Mauritania

Morocco

Rabat

Algeria

Algiers

Niger Chad

Libya

Malta Tripoli

Cameroon

Nigeria

Tunis

Uganda

Southern Sudan

Sudan

Egypt

Syria

Kenya

Mecca

Medina

Iraq

Da⁄nah

Saudi Arabia

Riad

Yemen

Iran

Dubai Abu flabi

Oman

5

12

Afghanistan

Uzbekistan

Lexical aspects of Maghrebi Arabic 253

Map 7.

Titel

000

Mali

Algeria

Niger

Slingshot

other forms

wa‹af, wa∂åfah, m‹åfah, min‹åfah, etc.

Mauritania

Morocco

Rabat

Algiers

Chad

Libya

Malta Tripoli

Cameroon

Nigeria

Tunis

Uganda

Southern Sudan

Sudan

Egypt

Syria

Kenya

Mecca

Medina

Iraq

Da⁄nah

Saudi Arabia

Riad

Yemen

Iran

Dubai Abu flabi

Oman

5

12

Afghanistan

Uzbekistan

254 Peter Behnstedt

Map 8.

Stork

008

Mali

Stork

other forms

adanåy

laqlaq, etc.

b»llår»Û, b»rrår»Û

Mauritania

Morocco

Rabat

Algeria

Algiers

Niger

kun≤i

Chad

Libya

Malta Tripoli

Cameroon

Nigeria

Tunis

Uganda

Southern Sudan

kul≤a

Sudan

Egypt

Syria

Kenya

Mecca

Medina

Iraq

Da⁄nah

Saudi Arabia

Riad

Yemen

Iran

Dubai Abu flabi

Oman

5

12

Afghanistan

Uzbekistan

Lexical aspects of Maghrebi Arabic 255

Map 9.

Mali

Morocco

Rabat

Billy-goat

bodbod

tays, t„s and others

©atrÂs

Mauritania

Billy-goat

009

Algeria

Algiers

Niger Chad

Libya

Malta Tripoli

Cameroon

Nigeria

Tunis

Uganda

Southern Sudan

Sudan

Egypt

Syria

Kenya

Mecca

Medina

Iraq

Da⁄nah

Saudi Arabia

Riad

Yemen

Iran

Dubai Abu flabi

Oman

5

12

Afghanistan

Uzbekistan

256 Peter Behnstedt

Mali

Morocco

Rabat

dk, etc.

f»®®ÂÛ

sardÂk, serduq

Mauritania

Map 10. Cock

Cock

010

Algeria

Algiers

Niger Chad

Libya

Malta Tripoli

Cameroon

Nigeria

Tunis

Uganda

Southern Sudan

Sudan

Egypt

Syria

Kenya

Mecca

Medina

Iraq

Da⁄nah

Saudi Arabia

Riad

Yemen

Iran

Dubai Abu flabi

Oman

5

12

Afghanistan

Uzbekistan

Lexical aspects of Maghrebi Arabic 257

+

Mali

Algeria

Algiers

others

ßÂß

≥w≥o, ≥ø≥o, πawπaw, siwsiw

farkÂs

+

Morocco

Rabat

f»llÂs, fellus

Mauritania

Map 11. Chick

Chick

011

Niger

+

Chad

Libya

Malta Tripoli

Cameroon

Nigeria

Tunis

+

Uganda

Southern Sudan

Sudan

katkÂt

Egypt

Syria

Kenya

Mecca

Medina

Iraq

Da⁄nah

Saudi Arabia

Riad

Yemen

Iran

Oman

5

sq„w

Dubai Abu flabi

12

Afghanistan

Uzbekistan

258 Peter Behnstedt

Mali

Morocco

Rabat

Algeria

Algiers

arnab, etc.

nayrab

qunn, gunn, qniyya, gnÂn, etc.

Mauritania

Map 12. Rabbit

rabbit

012

Niger Chad

Libya

Malta Tripoli

Cameroon

Nigeria

Tunis

Uganda

Southern Sudan

Sudan

Egypt

Syria

Kenya

Mecca

Medina

Iraq

Da⁄nah

Saudi Arabia

Riad

Yemen

Iran

Dubai Abu flabi

Oman

5

12

Afghanistan

Uzbekistan

Lexical aspects of Maghrebi Arabic 259

Mauritania

Mali

Morocco

Rabat

Algeria

Algiers

Niger

Map 13. Shrimps

crevettes

gambari, gamblu, gambru, gambas, etc.

qr„des

rubyån

Shrimps (prawns)

013

Chad

Libya

Malta Tripoli

Cameroon

Nigeria

Tunis

Uganda

Southern Sudan

Sudan

Egypt

Syria

Kenya

Mecca

Medina

Iraq

Da⁄nah

Saudi Arabia

Riad

Yemen

Iran

Dubai Abu flabi

Oman

5

12

Afghanistan

Uzbekistan

260 Peter Behnstedt

9

10

7

4,7

Mali

Morocco

Rabat

ku≥na . kcina, kπna kÂπna kutπna, kuππna kÂÛna kÂÛ„na kÂzna kwzna kÂzn kwayzn

Mauritania

7

Map 14. Kitchen

1 2 3 4 5 6 7 8 9 10

Küche

014

mat∫ax and others

Algeria

Algiers

77 7 7 7 7 2 5

Cameroon

Chad

Libya

Malta

Tripoli

52 5

Tunis

Nigeria

6

5

Niger

3,8

1

Uganda

Southern Sudan

Sudan

Egypt

Syria

Kenya

Mecca

Medina

Iraq

Da⁄nah

Saudi Arabia

Riad

Yemen

Iran

Dubai Abu flabi

Oman

5

12

Afghanistan

Uzbekistan

Lexical aspects of Maghrebi Arabic 261

k»ddu

mu∆®»f + ∆anÛåya

kÂ≥å®a

©åπ»q

xåπÂqa, xåπk

Mali

Morocco

Rabat

tasÂkål»t

Mauritania

ma©»lqa

ma©»lqa, mu∆®»f

Map 15. Spoon

Spoon

015

Algeria

mu∆®»f

Algiers

Niger

Cameroon

Nigeria

kiy„r

Chad

Libya

Malta Tripoli

kiy„r

Tunis

Uganda

Southern Sudan

ma©laqa

Sudan

ma©laqa

Egypt

ma©laqa

Syria

Kenya

Mecca

Medina

Riad

kam≥a

gafπa

Da⁄nah

ma©laqa

Saudi Arabia

ma©laqa

Iraq

Dubai

kam≥a

5

Oman

gafπa

Abu flabi

gafπa

Yemen

Iran

12

Afghanistan

Uzbekistan

262 Peter Behnstedt

Mali

Morocco

Rabat

Algeria

Algiers

kalse, kals†a, kalzetta, qalπn

dlå∆

råna

Reflexes of pers.-turk. ≤Âråb, çorap

Reflexes oft turk. çakÏ°r

Mauritania

Map 16. Socks

Socks

016

Niger Chad

Libya

Malta Tripoli

Cameroon

Nigeria

Tunis

Uganda

Southern Sudan

Sudan

Egypt

Syria

Kenya

Mecca

Medina

Iraq

Da⁄nah

Saudi Arabia

Riad

Yemen

Iran

Dubai Abu flabi

Oman

5

12

Afghanistan

Uzbekistan

Lexical aspects of Maghrebi Arabic 263

fakkåk, mafakk

dåyir

kald, kalíd

miglad, igld, agld

kuπir . cavetta

sårÂt

Algeria

mansa≤eh

Mali

Morocco

Rabat

Algiers

muftå˙, etc.

Mauritania

Map 17. Key

Key

017

Niger Chad

Libya

Malta Tripoli

Cameroon

Nigeria

Tunis

Uganda

Southern Sudan

Sudan

Egypt

Syria

Kenya

Mecca

Medina

Iraq

Da⁄nah

Saudi Arabia

Riad

Yemen

Iran

Dubai Abu flabi

Oman

5

12

Afghanistan

Uzbekistan

264 Peter Behnstedt

Mali

Morocco

Rabat

Algeria

Algiers

minkeb, mankib

kÂd, k†

qub†ål

k©, kars©, ©aks, ≥isa© and others

∆ummå®a, tå∆umma®†

marfaq, m»rf»g

Mauritania

Map 18. Elbow

Elbow

018

Niger Chad

Libya

Malta Tripoli

Cameroon

Nigeria

Tunis

Uganda

Southern Sudan

Sudan

Egypt

Iraq

m»rf»g Fes

Larache

Riad

Da⁄nah

Saudi Arabia

Tangiers

Mecca

Medina

Rabat

Syria

Oman

Taza

m»rf»g

tig»mm»rt lm»rf»g

Yemen

Iran

Afghanistan

Uzbekistan

Lexical aspects of Maghrebi Arabic 265

Mali

Morocco

Rabat

ka©b and others

awr»z

gd»m

Mauritania

Map 19. Heel

Heel

19a

Algeria

Algiers

Niger Chad

Libya

Malta Tripoli

Cameroon

Nigeria

Tunis

Southern Sudan

Sudan

Egypt

Syria

gd»m

Mecca

Medina

Iraq

awr»z

Da⁄nah

Saudi Arabia

Riad

Yemen

Iran

Dubai

Oman

5

(i)n»rz, etc.

Abu flabi

gd»m

12

Afghanistan

Uzbekistan

266 Peter Behnstedt

References Al-Selwi, Ibrahim 1987 Jemenitische Wörter in den Werken von al-Hamdâni und Našwân und ihre Parallelen in den semitischen Sprachen. Marburger Studien zur Afrika- und Asienkunde, Serie B. Asien/10. Berlin: Reimer. Aquilina, Joseph 1990 Maltese-English Dictionary. I, II. Malta: Midesea Books Ltd. Beaussier, Marcelin, Ben Cheneb Mohamed, and Albert Lentin 2006 Reprint. Dictionnaire Arabe - Français. Paris: Ibis. Original edition, Alger: Bouyer, 1871. Behnstedt, Peter 2007 Materialien für einen Dialektatlas von Nordost-Marokko. I: Tiernamen, Teil 2: Kommentare, Bibliographie. II: Mensch: Körperteile, Teil 1: Karten, Teil 2: Kommentare. Estudios de Dialectología Norteafricana y Andalusí 11: 7–57. Behnstedt, Peter, and Manfred Woidich 2005 Arabische Dialektgeographie. Eine Einführung. Leiden/Boston: Brill. 2011 Wortatlas der arabischen Dialekte. Band I: Mensch, Natur, Fauna, Flora. Leiden/Boston: Brill. 2012 Wortatlas der arabischen Dialekte. Band II: Materielle Kultur. Leiden/Boston: Brill. Corriente, Federico 1989 South Arabian Features in Andalusi Arabic. In Studia Linguistica et Orientalia Memoriae Haim Blanc Dedicata, 94–103. Wiesbaden: Harrassowitz. 1997 A Dictionary of Andalusi Arabic. Leiden/Boston/Köln: Brill. Corriente, Federico, and Ahmed Salem Ould Mohamed Baba 2010 Diccionario avanzado árabe. Barcelona: Herder. De Prémare, A.-L. 1993-99 Dictionnaire Arabe–Français. I–XII. Paris: L’Harmattan. Dozy, Reinhart 1968 Reprint. Supplément aux Dictionnaires Arabes. I., II. Beyrouth: Librairie du Liban. Original edition, Leiden: Brill, 1881. Heath, Jeffrey 2002 Jewish and Muslim Dialects of Moroccan Arabic. London: Routledge/Curzon. 2004 Hassaniya Arabic (Mali)–English–French Dictionary. Wiesbaden: Harrassowitz.

268

Peter Behnstedt

Kropfitsch, Lorenz 1978 Der französische Einfluss auf die arabische Schriftsprache im Maghrib. Zeitschrift der deutschen morgenländischen Gesellschaft 128: 39–64. Lafkioui, Mena 2007 Atlas linguistique des variétés berbères du Rif. Köln: Köppe. Lane, Edward William 1980 Reprint. An Arabic-English Lexicon in eight parts. Beirut: Librairie du Liban. Original edition, London: Williams and Norgate, 1863– 1893. Marçais, Philippe 1956 Le parler arabe de Djidjelli (Nord constantinois, Algérie). Paris: Publication de L’Institut d’études orientales d’Alger. Oman, Giovanni 1966 L’Ittionimia nei Paesi Arabi del Mediterraneo. Quaderni Dell’Archivio Linguistico Veneto 3. Firenze: Olschki. Paradis, Carole, and Darlene La Charité 2007 French Loanwords. In Encyclopedia of Arabic Language and Linguistics. Vol. II Eg–Lan. Leiden/Boston: Brill. Schuchardt, Hugo 1918 Die romanischen Lehnwörter im Berberischen. Kaiserliche Akademie der Wissenschaften in Wien Philosophisch-historische Klasse. Sitzungsberichte, 188. Band, 4. Abhandlung 188. Wien. Singer, Hans-Rudolf 1979 Das arabische und das romanische Element in der Fachsprache der tunisischen Fesmacher. Zeitschrift für arabische Linguistik 3: 28–46. 1984 Grammatik der arabischen Mundart der Medina von Tunis. Berlin/New York: Mouton de Gruyter. 1991 Die Beduinen als Träger der Arabisierung im islamischen Machtbereich. In Gedenkschrift Wolfgang Reuschel, Akten des III. Arabistischen Kolloquiums, Leipzig 21.–2. November, 263–274. Stuttgart: Franz Steiner. Thomason, Sarah Gray, and Terence Kaufman 1988 Language Contact, Creolization and Genetic Linguistics. Berkeley/ Los Angeles/London: University of California Press. Vollers, Karl 1886 Beiträge zur Kenntniss der lebenden arabischen Sprache in Aegypten. II. Ueber Lehnwörter. Fremdes und Eigenes. Zeitschrift der deutschen morgenländischen Gesellschaft 50: 607–657. 1887 Beiträge zur Kenntniss der lebenden arabischen Sprache in Aegypten. Ueber Lehnwörter. Fremdes und Eigenes. VII. Semitische Entlehnungen. Zeitschrift der deutschen morgenländischen Gesellschaft 51: 2–364.

Lexical aspects of Maghrebi Arabic

269

Wehr, Hans 1980 Reprint. A Dictionary of Modern written Arabic. J. Milton Cowan (ed.). Beirut: Librairie du Liban. Original edition, London: Macdonald and Evans, 1960.

ArabArab-Berber contacts in the Middle Ages and ancient Arabic dialects: new evidence from an old Ibā Ibā ite religious text Vermondo Brugnatelli 1. Introduction The modern Berber languages of North Africa contain a quantity of borrowings from Arabic, as a result of substantial contacts over many centuries. The analysis of these data provides a rich source of information for studies on linguistic contact from a typological point of view as well as for research on the early history of the Arabic dialects in a diachronic perspective. This field of research has not yet been methodically investigated ‒ though some studies already exist, concerning general issues (Chaker 1984; van den Boogert and Kossmann 1997; Ameur 2011) as well as some specific linguistic areas: Tuareg (Pellat 1962; Prasse 1986; Brugnatelli 1995), Jerba (Brugnatelli 2002), Rif (Kossmann 2009), Mauritania (Taine-Cheikh 2008). The study of ancient sources might enhance the historical research, inasmuch as it would provide evidence much older than contemporary languages, which have been known only since a couple of centuries at most. Unfortunately, the number of ancient Berber texts published so far is small and limited to Morocco (van den Boogert 1997), but our knowledge of old Eastern Berber is improving thanks to new texts from the Ibā ite area (Southern Tunisia, Northern Libya, Eastern Algeria).

2. The ancient text One century ago, Motylinsky (1907) announced the discovery of a substantial commentary in medieval Berber on the Mudawwana, a compendium of Ibā ite jurisprudence by Abū Ġānim Bišr b. Ġānim al-Aurāsānī and offered 16 short excerpts of it. The untimely death of the French scholar prevented any further publication on the “Moudaououana d’Ibn

272

Vermondo Brugnatelli

R’anem”1 and the whereabouts of the sources he was working with have remained unknown for a long time. After a long pause, some manuscripts have been newly (re-)discovered and some more information has been published by Ould-Braham (2008, 2009), U-Madi (n.d.) and Brugnatelli (2010, 2011a).2 The linguistic material contained in the Mudawwana allows for a preliminary research on some aspects of language contact between Berber and Arabic in the Middle Ages. The date of composition of this text is not known but it most probably goes back to a very early period ‒ at an undetermined time between the 10th and 15th centuries. According to the first lines of a hand-written glossary of ancient words extracted from this text (Bossoutrot 1900), the author was a certain Abū Zakarīyaɇ al-ɇIfrānī who is unknown from other sources, while the glossary was established by order of the sheikh “Messaoud ben Salah ben Abd el Ala”, whom Ould Braham (2008: 56, 58) identifies with MasɈūd b. JāliK al-Jamumnī, a sheikh who ruled in Jerba during the 16th century. If its language was so archaic that a glossary was needed to explain the meaning of many vocables, the text must have been composed at a much earlier date.3 A short poem quoted in the manuscript (f. 278a, l. 10-13) mentions a Hārūn ibn Yūnus who probably coincides with Abū Mūsā Hārūn ibn Yūnus al-Izāyī of the Kutāma tribe of the Masalta, the ‘supreme shaykh’ (šay al 1.

2.

3.

This is the title Motylinski consistently used in his works (1905: 146 and 1907: 68 and 69). Other designations are “commentaire de la Medawana d’Ibn Ghanem” (Bossoutrot 1900: 489) and “Modawana d’Ibn Ghanem” (R. Basset 1907: 540). The catalogue of the National Library in Tunis wrongly records a manuscript of this text under another author, Abdallāh b. ɈUmar b. Ġānim. On the contrary, both Ould-Braham (2008 and 2009) and U Madi (n.d.) refer to the author by his kunya Abū Ġānim instead of his nasab Ibn Ġānim. The manuscript I use for my analyses is the longest one, bearing the title Kitāb al-Barbariyya. It is located in the National Library of Tunis (black and white photographs, Ms.Or. 2550). A microfilmed copy is also kept in the University Library of Aix-en-Provence (n° 125.3–6 in the catalogue of Stroomer and Peyron 2003). The manuscript is described in Brugnatelli (Fc). It should be noted, though, that the colophon of a manuscript studied by U Madi (n. d.: 4) states that it was copied in 1208 AH (1794 AD) by MasɈūd b. al-UāLJLJ Jālih b. Sulaymān b. al-UāLJLJ ɈAbd ar-RaKīm b. al-UāLJLJ Idrīs b. alUāLJLJ Abū l-Qāsim b. al-MurābiX ɈAbd al-ɈAlâ b. Yūnis b. Mūsâ b. YaZlif b. Sufyān al-MaɈqilī. The sheikh named in Bossoutrot’s glossary seems more likely to correspond to this individual than to the ruler of the 16th century.

Arab-Berber contacts in the Middle Ages 273

mašāyi ) at the time of the Fatimid Mahdi (beginning of the 10th century), and this establishes a terminus post quem for the composition of the text. Some philological considerations too, such as the spelling of [g], with and not with , point to a period prior to the Hilālī invasion or very close to it.4 Bossoutrot (1900: 489) suggests that the Mudawwana was composed “some time before the 9th Century AH”, without telling where this information comes from.

3. Contact phenomena: borrowings The evidence which permits to detect contact phenomena lies chiefly in the numerous Arabic loanwords found in the manuscript. They are especially frequent in the field of religion, even if some native vocabulary (sometimes of Latin origin) is still preserved. According to the time of the borrowing, loanwords are more or less adapted to Berber phonology and morphology. In some cases, above all in everyday vocabulary, borrowings tend to reflect a spoken variety of Arabic rather than the literary language, and this reveals some early features of North African dialects. An interesting remark concerns the names of the Arabs and of their language, which are not expressed by means of borrowings (Arabic endonyms), but through a xenonym formerly widespread among the Byzantines and in Europe: aser!in, pl. iser!inen ‘Saracen, Arab’; taser!int ‘the Arabic language’. In present-day Berber we still find these terms in the easternmost oases of Augila (ašer!în) and Siwa (a#er!ên).5 In the manuscript, aser!in is used not only as a noun (many quotations are preceded by yenwa user!in ‘an Arab said’, or nnan iser!inen ‘the Arabs say’) but also as an attribute, as we can see in the opposition: yur aser!in/

4.

5.

In my transcription, I always transcribe with LJ, since in many cases it is impossible to determine whether it represents [LJ] or [g]. Moreover, when I do not quote other authors, I tend to use for Berber the standard orthography of Kabyle (whenever appropriate), which differs in some points from the standard transcription of Arabic: < c > stays for < š >, < ` > for < ġ >, < x > for < Z >, < b > for < Ɉ > (moreover, vowel length is not marked, and < e > renders [c]). This is intended to help the reader discern what I consider Berber or Arabic in this composite text (in the examples, the two languages will also be differentiated by means of boldface vs. regular letters). From Coptic sarakēn-os according to Vycichl (2005: 192).

274

Vermondo Brugnatelli

yur a'eLJmi ‘Arab, i.e. Islamic, lunar month’ vs. ‘non-Arab, i.e. indigenous, North African, Julian month’ (f. 244b, l. 12). Both languages, Berber and Arabic, are referred to in the following passage: ula t_taser!int aLJLJ_eLJLJull ula t_tamazi!t [= ula d taser!int ay yeLJLJull ula d tamazi!t] ‘whether he took an oath in Arabic or in Berber’ (f. 133b, l. 6).

3.1. Borrowings in the domain of religion The vast quantity of borrowings in the spiritual and religious domain is not surprising. It is a well known fact that even the Tuareg lexicon, which is the least affected by Arabic influence, displays the highest rate of loanwords in this domain.6 In any case, the great antiquity of this text is also proved by the preservation of a rich indigenous lexicon, not yet supplanted by Arabic, in the spiritual and religious sphere. A number of these terms are the remnants of an older Christian vocabulary containing also Latin borrowings, sometimes of Greek or Hebrew origin. Some examples of this pre-Arabic religious lexicon: Yuc and Bab-enne! ‘God, Allāh’ along with ababay, pl. ibabayyen ‘(a) god, deity’ (Brugnatelli 2010); aykuzen ‘Islām’; anLJlusen ‘malāɇika, angels’ (< Latin < Greek); adaymun, pl. idaymunen ‘aš-šay0ān, the devil’ (< Latin < Greek); tira ‘the Book, the Qurɇān’ (Ar. gloss: al-kitāb, ay al-Qurɇān, f. 300b, l. 8); iser ‘nabī ɇ , rasūl, prophet, messenger’; the daily prayers: tizzarnin ‘4uhr, noon prayer’, tuqzin ‘ Ɉa#r, afternoon prayer’ tin wučču ‘maġrib, evening prayer’, tin yi8es ‘ Ɉišāɇ , night prayer’, tin wezečča ‘#ub9, dawn prayer’;7 tifellas ‘people of the book’ (Ar. glosses: ahl al-LJizya wa al-Ɉahd, f. 299b, l. 12; ahl kitāb yaɈ0ūna al-LJizya, f. 299b, l. 12); imusnawen ‘fuqahāɇ, experts in Islamic Law’; tafe#ka, pl. tife#kawin ‘ Ɉīd, religious feast’ (< Latin < Hebrew); tir meLJLJu; and abekka8u (< Latin) ‘sin’; taLJerzawt ‘nadam, repentance’; amerki