Vous êtes sur la page 1sur 24
Arabic Dialect History and Historical Linguistic Mythology JONATHAN Ow! Bayreura UNiversiry ‘s INTRODUCTION Studies of spoken Arabic have been dominated by descriptions of individual dialects, dialect atlases and corpus-based sociolinguistic studies, There have been few attempts to apply prin- ciples of historical linguistics systematically to this relatively rich data base.! Earlier studies such as Cowan (1960) or Kaye (1972) stand out as the exception.? Since these two studies appeared a great deal of progress has been made in Arabic dialectology (e.g., Behnstedt 1985, 1997, Behnstedt and Woidich 1985, many studies of individual dialects). However, the de- scriptive advances remain basically unintegrated into a larger study of Arabic linguistics, ‘There are two prominent pitfalls in relating dialects to the history of the Arabic language. First, to return to Kaye’s study cited above, reconstructions based on dialectal forms even tually have to be melded into the total object, Arabic, and for this their compatibility with the oldest citations of Arabic need to be checked. With regard to ‘jim’, already in the late eighth century various pronunciations are attested in the phonetic description of Arabic (Sibawaih Il: 452). Sibawaih gives two “acceptable” variants of ‘jiim’ and two variants that are not good (Jaa tustahsanu) either in koranic recitation or in poetry.‘ Any discussion of a single proto-form has to consider this situation. It may be asked, for instance, whether Kaye's pro- posed change of *3 > d3 in Aleppo Arabic (1972: 54), doesn’t in fact reflect the use of a Variant already existing before Arabic speakers arrived in Aleppo. In general the problem of Arabic reconstruction must always confront the question of whether dialect forms of the diaspora are due to post-diaspora innovation, or are reflective of pre-existing diversity on the Arabian peninsula. A second problem in many dialect studies revolves around lingui | would like to thank Alan Kaye and Ignacio Ferrando for valuable comments on an eatlier version ofthis paper, 8 well as the detailed and valuable criticisms of an anonymous reviewer: 1. Its, on the other hand, quite typical to find in a description of a particular dialect an historical explanation for what is deemed an odd historical development, presumably odd ia relation to Classical or Standard Arabic, 2. Cowan (1960), believe, is subject to the same criticism that I develop inthis paper, namely that he atempts ‘o create an all-encompassing object, proto-Arabic, without giving due concer to the problem of fist adequately justifying individual reconstructed elements. Kaye’s approach restricting reconstruction to few (in his case a single) features, is sounder in this respect, and is followed here as a general orientation 3. Why there have been so few historically oriented studies isan interesting question in the history of Western oriental studies, a vast topic not to be gone into here, To lst but a few factors: the simple assumption that Arabic dia- lects derive from or even are a corruption of Classical Arabic, the refusal to accord the sociolinguistcaly less pes ‘igious variety the dignity of systematic historical investigation, a deerease of interest in historical linguistics in the wake of the rise first of theoretical, and more recently of cognitive linguistics, Furthermore, with notable exceptions, Semitic studies, of which Arabic is traditionally a part, have not engaged in serious theoretical linguistic reflection Brockelmann (1908: 4~5) sets the tone, saying thatthe reconstruction of a Semitic Urspracke is a chimera, recon. structed forms being mere formulas reflecting our temporary summary of the vatious languages. In one paragraph Brockelmann steers clear of any theoretical or conceptual dficuties entailed in the historical study of languages. 4. The details are not essential here. Sibawaih’s “standard” ‘iim’ appears to be either a dy or a 4. Kaye's re. constructed *5 is apparently included among the forms not appropriate for Koran recitation, if it ean be identified ‘with Sibawaii’s ‘jim’ which is like a “shin”. Sibawaihis socio-aesthetic judgments, of course, are not to be con fused with a proto form, Journal of the American Oriental Society 123.4 (2003) nS 716 Journal of the American Oriental Society 123.4 (2003) features that are attributed a specific historical origin, Closer inspection of these features, however, shows that historical attributions are often made on the basis of contemporary di- alectal distributions rather than on systematic historical reconstruction In this paper I will concentrate exclusively on this second problem, and therefore have chosen a feature which indubitably is of post-diaspora provenance. I examine the well- known first person imperfect (ISG) verb affix, n-, n=... -u, commonly identified as of North African (maghrébin) origin, and attempt a plausible historical linguistic account of its place and, more speculatively, date of origin, 1. BACKGROUND SUMMARY In this first section I present the basic problem and the tools with which it can be tackled. 1.1 Basic Linguistic Problem In Arabic first person singular and plural imperfect verbs fall into one of three paradigms. L write we write (1) b-2-kiob, —men-akiob (2) a-ktib, n-ikith-u (3) niki n-ikitb-w In (1) 1SGis represented by a vowel (a or 2), IPL by n-. It is found in easter Arabic dialects, the present example from Damascene Arabic, most of Egypt, the Sudan and Nigeria, and in Classical Arabic as well. In (3) 1SG is represented by n-, IPL by n=... -u. Alternatively, ‘one could say that 1- has become a marker of first person, -1 of plural. This variant is found throughout North Africa from Morocco to Alexandria, as well as in Chad and, according to Hillelson (1925), in Darfur and some parts of Korodofan in the Sudan. There are no recent studies giving a precise demarcation, and it can be noted that in the Sudan forms like Q) have over the past 150 years tended to migrate east so that they have also become a part of the urban sociolectology in Khartoum and other Nile cities, At the western edge of the west- em Sudanic dialect region (see 4.1 for this terminology) the boundary between (1) and (3) lies, according to Owens (1998: 114), in the small northerly strip of northern Cameroon where Arabic is the dominant language. Nigerian Arabic is basically characterized by (1), Chadian by (3). In this paper, (3) will thus be termed “Chadian Arabic” when reference is made to the dialects of the western Sudanic region. (3) is furthermore characteristic of some Egyptian dialects. It is found in the northwestern Delta area, the so-called biheera (also bu- hayra), and it is found in the area south of Asyut, up to Aswan. At the northern and south- em end of this latter region it is interspersed with villages and cities using (2) (Behnstedt and Woidich 1985: 210-12). (2), a combination of morphemes midway between (1) and (3) is of more limited geographical distribution, being restricted to Egypt.‘ It is found in the west delta area, forming a buffer zone between (1) and (3), and in Upper Egypt occurs interspersed with (1) (see Map 1). I will come back to these forms in section 7 below, in the meantime concentrating on the more widespread (1) and (3) (3), ne... -wis frequently referred to as “North African” (see section 7 below for refer- ences). As a shorthand term this is unobjectionable, It is unwieldy to qualify it as “North 5 Strictly speaking, there is one very small group outside of Egypt where (2) is the dominant pattern. AS re= ported in Owens (1998: 284), a household of seventeen individuals in Maiduguri essentially has (2). The interesting aspect is that this is a family that came to Maiduguri from Ndjamena in Chad. While inthe literature ony (3) is re- ported for Ndjamena, it cannot be ruled out that a pattern Tike (2) is also found there. Owens: Arabic Dialect History and Historical Linguisti 17 To Marsa Matrun cee \ ( \ \ | | — | ; | | | | © ‘El-Farafira Only: Mixed area: aktio - nikitoU *h/S (*laham — *lehem). The raising rule has to be attributed to the pe~ riod when the pharyngeals h/* were still present in the dialect’s predecessor, as the rule is not effective in other contexts (e.g., historical “ha remains ha, as in fa’har ‘month’). Subse quently *h and h fell together in h, giving the present-day contrast fahar/lehem. Bagirmi Arabic is basically spoken in the southern part of the western Sudanic region, the non-Bagirmi in the northern part. The latter is demographically larger, and I leave it un- named. 6 It is plausible to link this north-south dialect division with the Arab civil war cited by Braukiimper. The “losers” moved south, forming the basis of the present-day Bagirmi dialect. What is relevant for the present discussion is that (1) and (3) cross-cut the Bagirmi/ non-Bagirmi dialect line. (1), as briefly described above, is spoken in the extreme west of the western Sudanic dialect region, (3) in the central region. (1) vs. (3) forms an east-west isogloss axis. Relative to the features illustrated in (5) above, (1) occurs in both Bagirmi and non-Bagirmi dialects in the extreme western region. Thus, in Nigeria one has mihalim “we dream’ spoken, for instance, in the village of Kirenawa next to Lake Chad, vs. nihee- lim ‘we dream’ from Banki, a village further southeast on the Nigeria-Cameroon border. These forms illustrate the typical non-Bagirmi/Bagirmi contrast, but they share the IPL imperfect verb form (1). On the other hand, in Chad one has both nahalm-u and niheelm-u 15, Western Sudanic Arabic is characterized by a number of distinctive isoglosses, including antina ‘we’, ki 2FSG object suflix, eu 2MPL object suflix, -i1SG object sufi (stressed, bee ‘my house’, *1/*> hi? (also below). None of these except for aniina are unique among Arabic dialects, and therefore their conjunction in one relatively large dialect area is significant; see Owens 1993a for broad summary. 16. The non-Bagirmi dialect serves asthe basis of the Arabic koine, which is the lingua franca of Chad. 17. In both dialects the vowel after (= *h) is due to the “gahawa complex.” see 2.2 below. In the Bagirmni di- lect the *a not only tums into e, but, inthe imperfect verb i lengthened as well, as in the example, so thatthe im perfect paradigm for this class of verbs falls together with derived stem III verbs c.., of the same class as verbs like niteerib ‘we cultivate’, nPaawin ‘we help’). OwENs: Arabic Dialect History and Historical Linguistic Mythology 725 “we dream’, with the same non-Bagirmi/Bagirmi vocalic contrasts, but also with the typical Chadian imperfect verb form (3). On the basis of contemporary distributions at least, it may be concluded that (1) and (3) were present in the ancestral dialects of both the non-Bagirmi and Bagirmi varieties. This effectively suggests that both (1) and (3) were brought with the earliest Arab immigrants to the western Chadian region.'® ‘Taking all of these points into consideration, we may argue the following. Given that (1) and (3) are morphologically specific features whose chance of independent development is slight (see Hetzron 1973), and given the relative proximity of Egypt and Chad, the proba- bility of (3) developing from (1) (as in [4]) in events independent of each other is slight. Rather it appears that the presence of (3) in Chadian Arabic and in various Egyptian dia- lects argues for a shared innovation, which spread from Egypt to Chad. Just as Egyptian Arabic had both (1) and (3) by at least 1300 a.p., so too were both transferred into western Sudanic Arabic, with (3) eventually spreading throughout Chad, (1) into Nigeria. Since this transfer in all likelihood occurred before the seventeenth century, which is Woidich’s date for the introduction of (3) into Egypt, (1) and (3) must have both existed in Egypt before the seventeenth century as well. Woidich’s dating of the presence of (3) in Egyptian Arabic, as well as his accompanying explanation for the modality of its introduction, namely via migra- tion from North Africa, may be rejected. 4.2 Inner-Egyptian Developments Behnstedt (1998) takes up the debate from an internal Egyptian perspective. His general conclusion (1998: 87) is that (3) is of an early provenance, earlier at least than what Woidich assumes, and, somewhat vaguely, “... due & important element magrébin dans cette ré- gion.” Without specifying where (3) ultimately originated, he suggests that as far as Egypt is concerned (3) first was found in the western delta, perhaps introduced in the course of the thirteenth century, and was transferred in the fourteenth century to Upper Egypt (1998: 88) when the arabicized Berber tribe of the Hawwara, probably from eastern or central Libya, were exiled there (in 1380, Garcin 1976: 469). Note that this mode of dating does fit within the time frame of a further movement of (1) to the western Sudanic region, though only just. Behnstedt adduces various linguistic isoglosses linking the western delta to Upper Egypt, and he relates the spread of these to migrations. The most important is that, as summarized above, besides (3), (2) is also found in both regions. Both regions are marked by the deletion of a short high vowel in an open syllable, resulting in forms such as hnaak ‘there’ (< hi- naak) and yikatbu ‘they write’ (< yiktib-u) and the so-called gahawa-syndrome, character- ized by the insertion of an a in a CC sequence, where the first C is a guttural consonant (gahwa + gahawa “coffee”). Behnstedt, as Woidich before him, looks for bundles of isoglosses linking the western Delta and Upper Egypt. The main linguistic attribute suggesting a relation between the west- ern delta and Upper Egypt are the shared isoglosses (1) and (3). Beyond that, Behnstedt notes (1998: 88) “Malgré cet adstrat maghrébin sensible, les dialectes de la Haute Egypte res- tent tout de méme des dialectes égyptiens, car il leur manque d’ autres traits maghrébins im- portants. ...” That is, beyond this one important isogloss, there is nothing uniquely “North African” about Upper Egyptian Arabic, Note that Woidich (1993) came to a similar conclu- sion regarding the status of the Arabic in the Egyptian oases. While Woidich attributed the 18. leave unanswered the larger question of the linkage of the non-Barigmi-Bagirmi dialects to the Arab mi- gations into Chad, 726 Journal of the American Oriental Society 123.4 (2003) presence of (3) in the two northern oases to late North African influence, in general these dialects have the closest affinity to the Upper Egyptian dialects. I will return to the “Haw- ‘wara hypothesis” in section 5 below. In the context of drawing the Arabic of the Sudanic region into the extended discussion of this paper, parallel to Behnstedt’s summary of affinities between the western Delta and Upper Egypt, it will be helpful to summarize the main isoglosses, if any, linking Upper Egypt dialects with (3) to the dialects of the sudanic region. From the outset it is clear that the presence of (3) in a sudanic dialect like Chadian Arabic does not imply a bundle of isoglosses with the Upper Egyptian dialects where (3) occurs. ‘The general picture here is as with the western Delta and Upper Egypt: there are some spe~ cific isoglosses, but even more specific differences. For instance, a feature Behnstedt used to draw a further affinity between the western Delta and Upper Egypt, the deletion of a short high vowel in an open syllable, is attested only in limited contexts in Chadian Arabic, and western Sudanic Arabic in general. It is attested in the same form in the Sudan, how- ever, in the Shukriyya dialect, located far to the east around the Atbara River. Reichmuth (1983: 276) for instance has ti-kutl-u ‘you M.PL kill”, Shukriyya, however, has first person forms like (1), not like (3). Chadian, along with Western Sudanic Arabic in general, keeps the same stem form throughout, twkrul, tuktul-u (Roth 1978: 3), since the addition of a vowel- initial suffix does not lead (o the deletion of the stem vowel in an open syllable in forms such as these. Schematically the situation can be presented as in (6). (©) Upper Egypt, sudanic area: an isogloss split Upper Egypt -iktib, n-ikirb-u (characteristic first person form) tuktul tukutl-u (characteristic syllable structure) Chad N\ ‘Shukriyya, eastern Sudan neiktib, n-iktib-u (first person) tuktul tukutl-u (syllable structure) tuktulu (syllable structure) aktib, niktib (first person) Behnstedt stuggested that the short high vowel deletion rule supported his claim that Upper Egyptian dialects originated in the western Delta, the presence of (3) correlating with a syl- lable structure type, viz., that deleting a short high vowel in an open syllable. Carrying the linkage further, however, to include dialects of the Sudanic region shows that (3) is not necessarily linked with this syllable structure rule in other dialects of the area. In particular, in the current small data set in (6), the presence of (3) in Chadian Arabic does not co-occur with the present high vowel deletion rule. The short high vowel deletion rule is, however, found in other sudanie area dialects, though not those with (3). In short, one must reckon with the fact that (3) appears to have a history independent of other isoglosses it has been correlated with. In fact, (6) above, though based on only two binary features is already sug- gestive of major dialect divisions. This is typical of the complexity that is found in determin- 1g dialect relations between Egypt and the sudanic region, as will be seen now in greater detail.!? 19. Such complexity is typically met when one tries to link Upper Egyptian dialects to those in other regions. ‘Woidich’s atempt (1993: 351-53) co explain what he terms “innovations” via dialect mixture can be regarded as a deus ex machina to sidestep the difficult question of an adequate historical explanation for the dialect genesis. In a later work (1997: 191) Woidich assumes a more sober position in explaining the origin of present-day Egyptian dialeets, acknowledging that a lack of historical depth in our written records requires the application of standard ‘methods of linguistic reconstruction (including, of course, the factor of contaet-based change). Owens: Arabic Dialect History and Historical Linguistic Mythology 721 Looking at isoglosses linking (or not) Chadian Arabic with Egyptian Arabic four sorts of features may be distinguished: |A. Those shared by Chad and Upper Egypt south of Asyut, but by few or no other dialects B. Those shared by Chad and only a few places in Upper Egypt, usually, though not always isolated occurrences south of Idfu (south of Luxor) C. Those shared by Chad and other Egyptian dialects, but not with Upper Egypt, or only with certain Upper Egyptian dialects D, Features distinguishing Chad from Egypt In the following brief exposition I cite the general Egyptian forms in parentheses in order to elucidate the special nature of the Chad-Egyptian relationship. The cited general Egyptian form, of course, can itself contain many variants, which are not exhaustively listed here. For instance, the non-epenthesis form is cited as bigri ‘he runs’ in (9c) below, though according, to Egyptian dialect this could also be bijri, with the j reflex of ‘jiim’. ‘The first category, A, has the fewest features, though some stand out. Two in particular can be mentioned: Ja, CVCCVC verb used for ‘become a color’ (Behnstedt and Woidich, 253) hammar ‘become red’ (otherwise ihmarr) 7b. Plural verb forms (Behnstedt and Woidich, 350-54, 363) C(V)CaCC: irkabb ‘knees’, Nigeria rikabb *knees"?! CiCiCCae: hisinna *horses’, Chad hunukke ‘chins’ CaCaCa: Upper Egypt bavala ‘stingy’, Chad fayala ‘elephants’ (Roth 1978: 82)*2 Each of these features is highly specific and is not found in the western Delta area. (7) may even be unique to these regions. Under the second category, (B), are found more features: 8a, Indirect object + pronoun, Suffixed to verb in most varieties of Egyptian: on separate word in Aswan (Behnstedt and Woidich, 163) and Chad: general Egyptian gul-r-il-ak ‘I told you’, vs, gult leek 8b. Initial a/. In certain sets of words and in certain word-initial prefixes, either a oF i occurs In most Egyptian dialects the vowel isi (i-beet). In about twenty sites south of Luxor, and in the Daaxila oasis, the vowel is a, al-beer (Behnstedt and Woidich, 77). Chadian Arabic has a in most of these sets (al-beet), though the match is not perfect. For instance, the Egyptian “a” dia- ects have this vowel in personal pronouns, anta ‘you M’, whereas Chad has i, inta (oF ita). $e. Forms 5 and 6 derived verb marked by the prefix al- (Chad), i as the Xarga and Farafira oases (Behnstedt and Woidich, 242) ‘al-kallam “he spoke” (il-kallam Gina villages) (general Egyptian it-kallam, etc.) 3 villages near Gina, as well 8d. Epenthetic vowel in context CVCaC-CV stressed (Behnstedt and Woidich, 53), five sample sites between Idfu and Aswan, universal rule in Chad, ‘isim ‘name’ but i'sim-ha ‘her name” (general Egyptian, ‘isimha/ ismiha, etc.) Though probably less numerous, features under (C) also exist. 20. Others, generally isolated occurrences in Upper Egypt equivalent to the normal form in Chad and in Western Sudanic Arabic, are found in Owens (1993a: 146). 21. ‘The plural forms like hunukke are from my own Chadian Arabic data (Owens 1993s: 136). have not found {or did not try to elicit) forms like rikabb in Chadian Arabic. Neither of these are found in Roth 1978, 22, From Behnstedt and Woidich's examples, this plural coresponds especially to singulars of the form CaCiC. In Chad, on the other hand, it ends to be formed of singulars of CVVC (as here fi!) or CuCu structure 728 Journal of the American Oriental Society 123.4 (2003) 9a. -ki 2FSG object suffix, found in the extreme eastern Delta, the area of south of Luxor, and in the oases (Bebnstedt and Woidich, 152). It is used throughout Chad (general Egyptian -ik) darbi-ki *your E. road’ 9b. Stress shift on CVCV'CVCV words, Eastern Delta region, Chad ‘darabat ‘she hit, dara'bara (general Egyptian has numerous variants here, none Tike this) 9c. bukura syndrome, high vowel inserted in context CCyegoaqe biti — bifir *he runs’. Egyp- tian oases and Bani Swayf region (north of Asyut) (Behnstedt and Woidich, 47), normal form in Chad (general Egyptian no epenthesis, bivigr, etc.) 94. 18G object suffix on noun stressed. Six notations found in the extreme eastern Delta area, normal form in Chad (Behnstedt and Woidich, 150) giddaa'm-i ‘in front of me’ (general Egyp- tian unstressed, giddaam-i, ete.) Probably the most prominent set of features falls under (D). To cite but one trait highly characteristic of Chadian Arabic and, except for the western Sudanic region, unique in the Arabic-speaking world, the 1SG/2MSG perfect C-final verb form is marked by -f only under certain conditions, e.g., when a further suffix is added. Thus one has the alternation, (10) ka'tab ‘I wrote',® vs. ka'tab-t-a ‘I wrote it’ All in all, there are so many isoglosses linking Chad, northern Cameroon, and Nigeria that what I have termed the western Sudanic dialect area when looked at in synchronic dialec- togical terms forms a very coherent dialect area (see n. 15), quite different both from the eastern Sudan (including the Nile valley) and from Egypt in general. 5 A point that emerges here is that there are interesting, in some cases perhaps even unique, isoglosses linking Upper Egypt and Chad, just as there are interesting isoglosses, linking the western Delta and Upper Egypt. (3) is among these. However, there are no mas- sive sets of isoglosses linking Upper Egypt as a whole and Chad, just as there are none linking the western Delta and Upper Egypt. This implies, from the perspective of the first person imperfect verb form, that the presence of (3) in a dialect has little predictive or im- plicative value for what other features are found in 4.3. Migration and Dialect Mixture ‘As a way of introducing the theme of language and migration, dealt with in a more gen- eral way in the next section, it is relevant here to relate the relative non-coherency of dialect, bundles noted in this section to aspects of attested migration patterns. It was seen in sec- tions 4.1 and 4.2 that Upper Egyptian dialects have similarities with both the western Delta and the sudanic region, without forming a coherent dialect area with either. A major theme in the Islamic history of Egypt is the southward drift of Arabs under vari- ous pretences, Many of the original conquerors were nomads (Brett 1978a: 502) who were, it may be assumed, itinerant in Egypt by cultural nature. As early as 727, Qaysi Arabs (appar- 23. The contrast with the 3MSG is maintained via stress, ef. karab “he wrote’. Inthe ISG the stress on the final syllable speaks for an underlying -. ts noteworthy, however, that in Bagirmi Arabic, which is characterized by syl lable final stress on CV'CVC verb and noun forms (also a feature often attributed to North Africa), the form ka'tab is indifferently “Ue wrote’ 24, The only general morphological isoglosses distinguishing Nigerian from Chadian Arabic are (1) and (2) above and the 1SG object suffix on a verb, which is stressed in Chadian Arabic, but notin Nigerian (aafoo ni ‘they saw me’, vs. fadfooni) 25. Writing about the dislectology of the Shukriyya in eastern Sudun, Reichmuth (1983: 29) concludes that there is similarly no direct dependence between Egyptian and Shukriyya Arabic Owens: Arabic Dialect History and Historical Linguistic Mythology 29 ently from the Syrian desert) were settled in Upper Egypt by the Caliph Hisham (Lewis 1970: 176). Other Arabs crossed the Red Sea from Yemen (Garcin 1976: 47).2® There followed a gradual marginalization of the Arab ruling class, and an attendant drift, or forcible eviction to Upper Egypt. Already in the eighth century it is reported that there is disaffection among Arab troops and increasingly rulers rely on foreign mercenaries, Turks, Sudanese, Ethiopi- ans, Berbers, Slavs, and Kurds, to ensure control. Revolts by Arabs against the government occurred in 794, 802, and 830 in the Delta (Brett 1978a: 533-34), and in 869 there was a further revolt by Arabs in Upper Egypt. Central control over the south was marginal.” The displacement of Arabs from the western Delta to Upper Egypt cited by Behnstedt is thus but one episode in a long saga of migration towards Upper Egypt. This provides an initial hypothesis for why he fails to find large numbers of isoglosses between the western Delta and Upper Egypt. The Arabic-speaking population of Upper Egypt was formed from. a constant influx of migrants from various areas of northern Egypt and beyond. They spoke different dialects, these became settled in Upper Egypt, and local dialect contact further in- creased the mix of forms, Outside of the Arabian peninsula, there are few areas so dialectally diverse in the Arab-speaking world as Upper Egypt. This diversity was probably greater in the past than it is today.°* This can be seen in a feature such as (8c) above. That forms like al-kallam are uniformly the norm in Arabic, from Borno in Nigeria to Darfur, suggests that there must once have been a sizeable population of such speakers in Upper Egypt. Though it will remain a matter of pure speculation, itis quite likely that some features which today have to be identified as uniquely sudanic Arabic were first innovations in Upper Egypt that have now disappeared there.”° There is, in fact, good evidence for another innovation that is part of the first person imperfect verb complex given in (13) above, but which disappeared from this area after spreading to the Sudanic region. Besides (1-3) above, in both Shukriyya Arabic and in Ni- gerian Arabic a further pair of forms is attested, namely: (11) mafrab n-afrab ‘I drink’, “we drink? ‘That is, the form nafrab can be used for both the ISG and IPL. In Shukriyya this form is used only after the desiderative particle al-, an-n-afrab ‘I shall drink’ (Reichmuth 1983: 277, here al- assimilated to following n-). In Nigeria, so far as I have been able to determine, n- as 1SG marker is in free variation with ba-. In Owens (1993b: 107-8) a text example is cited where within a sequence of forms the n- prefix form changes from ‘I’ to ‘we’, sawwoo-ni faba l-abid da, ....n-isora be I-bagar.... di bas fagal-an-na aniina n-isawwi fi. “They made me like a slave, . . . [ would herd cattle . .. that’s the work we do’. That the first set of n-’s signify “T” is shown by the suffix -ni ‘me’ on sawwoo-ni, while the switch to the “we” meaning is signaled by the pronoun aniina ‘we’ (Owens 1998: 191-96 for sta- tistical treatment). Both the Shukriyya and Nigerian Arabic varieties are of form (1), ba-, n-. Example (11) is not attested in Egypt. Nonetheless, its origins may be looked for in Upper Egypt, for the 26. Qaysites may be taken as roughly eponymous for Arabs of eastern Arabian peninsular origin, Yemenis for ‘western, Both groups settled early in Upper Egypt. At this point in our study of the history of Arabic dialects, it ‘would be premature to try to link these two broad groupings with specific dialect forms, 27. Holt and Daly (1979: 21) write, “Upper Egypt has been for centuries [i.e., before the Fourteenth century} the refuge of insubordinate Arab wibes, and there were recurrent revols of greater or lesser significance throughout the period and beyond the Mamluk perio 28. Miller (forthcoming) speaks of an Upper Egyptian koine which may level dialect differences inthe atea 29. Like the lexically conditioned, though widespread shift of 6 to d, as in dalla “tame, domesticate’ < dala Woidich (1993: 347) similarly finds “relies” of Upper Egyptian dialects in the oases dialects. 730 Journal of the American Oriental Society 123.4 (2003) following reason." In neither Shukriyya nor Nigerian Arabic is there an analogical moti- vation for (11). So far as I know, these are the only two dialects where n- is used for both. first person singular and plural. Their origin can, however, be situated in Upper Egypt among, those dialects where contact between (1), (2) and (3) occurred. Of course, one cannot deter- ‘mine exactly where this was, as the ancestral migrations to the sudanic area, as noted above, already began in the fourteenth century. It is notable, however, and this simply to indicate that postulating such contact is highly plausible, that today between Idfu and Aswan are found the three forms, (1), (2), and (3) in close proximity. Behnstedt and Woidich’s (p. 213) sample points 764 (2), 765 (3), and 766 (1) are virtually adjacent. Projecting such contact back into time, it is well imaginable that speakers could have formed (11) on the basis of the plural of (1) and singular of (3). I return to this point in section 7. 5. LANGUAGE AND MIGRATION Itis clear from the discussion in section 4 that large-scale migrations have been important in distributing and redistributing varieties of Arabic, Here I shall further highlight the role of migration in bringing different variants into contact, in order to narrow down the time frame within which linguistic changes occurred. This can be elaborated with attention to three aspects of the current puzzle. The first pertains to the possible spread of (3) within Egypt, the second speculates about how and when (3) might have been introduced into Egypt from North Africa, and the third turns the direction of spread around, suggesting how (3) could have spread via migration from Egypt to North Africa. It will be recalled that Behnstedt suggested that (3) was brought to Upper Egypt by the Hawwara. I will assume, for the moment, that this is correct. The Hawwara were originally exiled from the western Delta area to what is present-day Sohag in 1380, but within five years had expanded as far as Aswan (Garcin 1976: 472). [have found no mention (e.g., in Garcin, MacMichael 1967) that the Hawwara played a role in the Arab expansion into the sudanic region, nor am I aware of any tribes or clans of this name. Moreover, their exten sion to Aswan was expressly for the purpose of providing a defense against the attacks of the Banu Kanz, an Arab group that controlled the area south of Aswan, i., of keeping out Arab-speaking groups who were already in the sudanic region. Given this situation, it is, unlikely that the Hawwara can be seen as the purveyors of (3) into Upper Egypt. They 30. Blau (1985: 120) makes the intriguing claim thatthe n- imperfect singular form is attested in various Ara- ban peninsular dialects. For this he cites Socin's (1901: 133, 194) central Arabian texts, However, ail to find ref erence to an n- imperfect in these citations (p. 133 mi/“aal and maf"wt are treated, p. 194 the realization of “jim” and ‘qaaf’). Where Socin treats the 1SG imperfect verb (1901: 162) there is no mention of a 1SG n- imperfect. It could be that Blau took his information directly from Fleisch (1960: $77), without actually having checked Socin, asthe same erroneous information is elliptically mentioned in Fleisch’s article on Arabiya, Blaw also cites Rhodo- kanakis on Dofar Arabic in southern Oman (1911: 166). In his standard paradigm for the imperfect, Rhodokunakis ives the “afTal” form (agiel) as the 1SG form (1911: 166, 168). He does, elliptically, suggest that in his text a single ease of an n- ISG form might oceur, negsidi-ne interpreted as negsid a'na. He himself does not appear cer- tain of the citation, however (1911; 106), and suggests that what appears to be a ISG n- imperfect verb form in fact is the use of the formal “royal we" in the sense of “L" He also suggests (1911: 166 n, 3) that a1- ISG imperfect can result from the contraction of the independent pronoun anaa > 1 ‘Unfortunately, the only intimations for the existence of a n- 1SG imperfect in the Arabian peninsula come from the pre-tape-recorder era, 50 there is no way to recheck what appear to be forms with ambiguous interpretations, So far as I know, there are no recent citations of n- for the 1SG imperfect prefix in Arabian peninsular dialects; e.g, it is not noted in Behnstedt 1985, nor even for another early Oman description (Rheinhardt 1894), describing a north ‘em Oman variety, Blau further notes that n- for ISG occurs in Palestinian Aramaic, though does not explain how this bears on his interpretation of Judaeo Arabic. Lacking better-founded evidence to the contrary, it appears that (14) has to be regarded as an Upper Egyptian innovation, Owens: Arabic Dialect History and Historical Linguistic Mythology TB would have had to have settled in Upper Egypt (beginning in 1380) and been there long enough to have spread (3) to other tribes who would have carried it on into the Sudanic re- gion, since they themselves did not settle there. Recall that Arab tribes had reached the Lake Chad area by the late fourteenth century, about the same time the Hawwara were settling in Upper Egypt. How long it takes a form like (3) to spread among a population is a large im- ponderable. As reported in section 6, in one attested instance of its spread outside its native population of speakers in Maiduguri, its extension in fact is only sporadic. One would have fo assume a number of generations at least for it to take over the entire region, in which case the form could not have been established in Upper Egypt (Sohag to Aswan) until be- tween 1450 and 1500 at the earliest, This is after the time of the first Arab migrations into the Sudanic region. Under Behnstedt’s hypothetical "Hawwara” scenario, one would need to find evidence that (3) was spread into the Sudanie region after 1500. Otherwise it must fol- low that (3) predates the Hawwara in Upper Egypt. Itis also necessary, given an assumed North African origin of (3), to consider the role that large-scale migrations moving from west to cast might have played. The major demographic movement that went in this direction took place during the Fatimid ascendancy in Egypt in the tenth century. A number of attempts were made by the Fatimids to capture Egypt, be- ginning in 915, though only about 970 did they finally gain control of the country. One eth- nic group at the forefront of this movement was the Kutaama Berbers, originally from the ‘Tunisian-Algerian border area, who formed the backbone of the Fatimid expeditionary force. Leaving aside the question of what language they spoke, as soldiers their influence on the language of the country was probably negligible. In any case, shortly after the Fatimid cap- ture of Egypt, they were sent onwards to fight in Syria, while those who stayed in Fust never established an independent power base and within a short time were assimilated to the local population (Lev 1991: 91), The major migratory movement at this time which would probably have had a greater impact on the local population was from the immediately adjacent Cyrenaica (eastern Libya, also referred to as Barga, Barca) and the desert to its south (Brett 1978b: 620, Fisher 1977: 242, Bianquis 1998: 112), whose population, both Arab and Berber (Lawaata) settled in the Delta and in the oases. In the tenth century there was a cyclic movement between Upper Egypt, the Egyptian oases, Cyrenaica, and Alexandria (Garcin 1976: 61). (3) is found in all of these regions, Brett (1974: 42) reports that already in the tenth century the Banu Hilal, who will be treated later this section, were moving from Upper Egypt into the Oases, i.e., an east to west migration. ‘There is, unfortunately, nothing more attested in written sources about the variety of Ara~ bic spoken by these groups than there is about the other varieties discussed in this paper. To my knowledge, Larcher (2001) reports on the earliest eyewitness account of Cyrenaican Arabic, by a Moroccan traveler in the thirteenth century, and this account is silent about (3) (and indeed the account raises more questions than it answers about the Cyrenaican dialect, spoken then). From a contemporary perspective, which faute de mieux has to be the starting point of the discussion, eastern Libyan Arabic (and its extension into the northern coast of Egypt) is, for the region, very much sui generis. It is fundamentally different from Tripoli- tanian Arabic and by implication, other varieties in North Africa (sce Owens 1983), and shares what probably are historically specific syllable structure traits with various Iraqi, Gulf, and Saudi Arabian dialects. In particular, a short low vowel is raised in an open syllable; and ina sequence of three open syllables with low vowels, the first is elided. Disregarding stress, the two forms kitab *he wrote’ and iktibat ‘she wrote’ are equally eastern Libyan and Najdi (Saudi Arabia) 732 Journal of the American Oriental Society 123.4 (2003) ‘Should one speculate that (3) arose during this period (tenth century), there is still no way of pinpointing the geographical source within the large, irregular quadrangle formed by Upper Egypt—the Oases—Cyrenaica—Alexandria. If one should insist on Cyrenaica, in any case, it would be a very easterly origin in the Maghreb, in a dialect that is otherwise missing key North African traits. Turning to the third early population movement, in many respects the best known and most important migration is that of the Banu Hilal, which emanated from Upper Egypt moving to North Africa. According to many scholars (Abun-Nasr 1987: 70), the large-scale Arabicization of North Africa began between 1036 and 1051 when, for various motives, the Banu Hilal were encouraged to leave Upper Egypt for North Africa (Hrbek 1977: 15). Their numbers have been put at between 50,000 and 150,000 (Abun Nasr 1987: 69, Fischer 1977: 242), For linguistic purposes this migration would certainly have been crucial, since entire populations, men, women, and children, units ideally suited for the movement and maintenance of language varieties, migrated. In terms of the present linguistic puzzle this population of speakers can be looked at in two ways. They might have been speakers of (1), and acquired (3) when they came into contact with North African Arabic speakers. If Behnstedt’s dating of the introduction of (3) into Upper Egypt is correct, then it may be assumed that they were indeed speakers of (1), However, it is a priori unlikely that such a large population, one moreover known for its successful, if devastating expansion in large parts of North Africa, should have given up their (salient) trait in favor of that of a minority group.*! Alternatively, they might already have been speakers of (3), and in their exodus to North Africa met other speakers of the same form or speakers of (1) who took over the form of the newly dominant group. In either instance, if this alternative is correct, it is a further piece of circumstantial evidence locating the innovation (3) in Egypt.°? To summarize, beginning from the vantage point of migratory group rather than from lin- guistic form does not, in this instance, solve the question of where and when (3) originated. By working through the linguistic entailments of connecting (3) to one migration or another, it does, however, help delimit when and where the form first arose, and where it might have spread. 6. SALIENCE Before tying the various threads of this discussion together, one more aspect relating to the spread of (3) may be mentioned, Not all linguistic features are equally significant in de- termining relations between dialects. The less likely a feature is reproducible by chance or universal drift, the more indicative it is of a significant relationship, should it appear in two disparate areas. The problem is to identify what a significant feature is. The notion of sa~ lieney may be adduced in this regard. Te was seen above that (3) is characteristic of dialects which in many respects are quite dis- tinct from one another. One explanation for this is that however and wherever (3) originated, 31. What litle influence Classical Arabic might have had on the usage would of course have supported the ‘maintenance of their original form (under this hypothetical scenario). ‘32. The evidence that Blau adduces (1985: 56-62) forthe existence of (3) in twelfth-century Egypt is interest- ing, but too difficult to interpret for it to be reliably integrated into the current discussion. He cites instances of (3) from Maimonides’ (d. 1204) writing, including works that he probably wrote during his residence in Egypt, Since Maimonides lived his early life in Spain, however, i is unlikely that his writing would have reflected an unadulter- ated Egyptian variety. 433, In this instance I obviously find the “Fatimid” solution the better one, though as I note below, I think that even this erais to late for the development of (3) Owens: Arabic Dialect History and Historical Linguistic Mythology 733 once in the language, under the appropriate sociolinguistic conditions (see n. 37), saliency facilitates its spread. Saliency is, unfortunately, an intuitively appealing concept lacking a general definition. I will attempt to give a concrete measure to saliency, using (3) as the focus of discussion. First, it is instructive to look in detail at one situation where (3) competes directly with (1). Maiduguri, Nigeria, as noted above, lies in a dialect area where (1) is the overwhelm- ing norm. (3), however, has been introduced in large numbers as well, via immigration of large numbers of Arabs from Chad. Ndjamena is only one hundred miles (160 km.) from Maiduguri. There exist Maiduguri wards dominated by Chadian Arabs, such as Ruwan Zafi, where (3) is the norm. Thus in Maiduguri one finds quite commonly both: (12) ba-myi, n-amfi ‘I go, we go’ nacmfi, nacmf-u A closer look at the distribution of these forms reveals that both sets are spreading beyond their original population of users. Without citing the statistical basis of this statement (details in Owens 1998: 163, 184),™ one finds speakers who ancestrally have (1) moving towards (3), and simultaneously speakers with ancestral (3) moving towards (1). Many individual, Arabs in Maiduguri thus have both (1) and (3) in their repertoires. At the present time, how- ever, no large-scale shift of (1) to (3) or of (3) to (1) is apparent. The variants are in con- tact, but (1) and (3) are each native to communities large enough to ensure their stability at the present time. This observation becomes significant when the behavior of (1) and (3) is compared to other variables in Maiduguri Arabic. For instance, Nigerian and western Sudanic Arabic in general have two forms for expressing the construction, active particle + object suffix. In one, the object suffix is attached directly to the stem or gender/number suffix, as in (13) kaatib-ha, Kaatb-ce-ha ete. ‘he has written iF, she has written iF” In the other, an intrusive -in is attached between stem and suffix in all forms. Gender and number contrasts are thereby neutralized: (14) kaatb-in-ha *he/she/they have written iF” In terms of rural dislectology, forms like (14) are dominant in “Bagirmi” Arabic, already discussed above, an area beginning approximately fifty miles west of the Cameroon-Nigeria border in Nigeria, running through northern Cameroon and on into Chad south of Najamena. ‘The eastern border has not been determined. The rest of western Sudanic Arabic has (13). Immigrants to Maiduguri come from both rural areas. Unlike (1), however, speakers whose ancestral variety is either (13) or (14) maintain their ancestral form to a very high degree (Owens 1998: 289), Those who use (13) do not use (14), and vice versa. Saliency of features can thus be measured not in absolute terms (note that both intrusive -in and the verbal plural suffix -1 belong to the general class of “suflix"), but rather by the 34, In any case, traditional dialectologists apparently find large-scale statistics “dry” to quote one reviewer's ‘pinion of such data (Behnstedt 2000; 145, German “drdge). However, statistical methodology has become a stan- ard instrument in studies of language contact and change (See, e.g.. Chambers etal, eds. 2002). 35. I argue (1998) that ancestry is one of two Key variables accounting for patterns of variation in Maidugusi Arabie. The time depth of ris shallow, no more than three generations, so the ancestral varieties, those spoken by parents or grandpat relatively easy to identify. These varieties are generally assignable cither to a specific rural area, or to Ndjamena, the only large city from which a sizable population of Arabic speakers has moved to Maiduguri 734 Journal of the American Oriental Society 123.4 (2003) painstaking, quantitative comparison of one variable feature against another, in a popula- tion where the variants are potentially accessible to all.% It is thus the case that some mor- phological variables spread more easily outside of their original population of users, than do others. On an admittedly circular basis, “moveable” features may be identified with those that are salient. In this sense, based on the Maiduguri data, (3) is salient, Generalizing this to the historical spread of (3) in Egypt suggests that the present extension of (3) would not necessarily reflect its area of origin. Explicit manipulation would, depending on the social status of the forms, facilitate either the spread or the recession of a given feature. ‘The question of saliency adds yet another degree of uncertainty to where (3) originated. Whether (3) is a more salient trait, hence more liable to manipulation, than other linguistic features that might be adduced to define a North African vs, Egyptian dialect area requires a basis of comparison in the behavior of different linguistic variables, as sketched above. Should it, in an independent characterization of saliency, turn out to be highly prominent, pinpointing its place of origin becomes all the more difficult. It might be a feature particu- larly prone to migration. Saliency, it should be noted, is a psycholinguistic or cognitive concept: speakers identify a feature and either appropriate or give it up more readily than others because it is percep- tually salient. The notion necessarily has implications for sociolinguistics and language con- tact. However, even if (3) is salient, it does not obviate the need for an historical account of its development. In the context of the overall interpretation to be presented in section 7 below, I would argue that the saliency of (3) facilitated its spread from Egypt to other re~ gions. Furthermore, its saliency might explain another characteristic of the feature, namely that regions as a whole tend either to have it or not to have it It is found throughout North Africa and throughout Chad, for instance, but it is categorically absent in the rural Arabic of northeastern Nigeria or in the Arabic of eastern Sudan. An hypothesis that can be advanced here is that salient forms when in direct competition with other forms (i.c., a paradigmatic choice is implied in using the salient or non-salient forms) will tend to level out, either spreading or disappearing, in language communities.*7 7. THE ORIGIN OF (3) After the above discussion, it should come as no surprise that the provisional conclusion to the question when and where (3) originated is that there is no definitive answer. However, 36, Even in this example the social identity of the users of the different variants is not factored into the prob: Jem, (3) is used predominantly in Maiduguri by speakers of Chadian Arabic. Its a common perception among Mai ‘duguri Arabs thatthe Arabic of Najamena is better than that of Maiduguri, The underlying idea is that that variety is best where itis used by the largest population of speakers. Arabic isa lingua franca in Ndjamen, but a commu- ‘nal language in Maiduguri. The association of (3) with this Chadian immigrant group would thus increase its social salience. The users of (14), on the other hand, area rural population which straddle the Nigerian/Cameroon/Chad border (see above), while users of (13) share essentially the same rural culture, located to the north of speakers of (14), Neither dialect is associated with a particularly prominent, distinctive profile in political, cultural, or geograpl- ical terms, 37, As noted above in this section, saliency does not guarantee survival. To the contrary, it ean, under the appro: priate sociolinguistic circumstances, lead to stigmatization and loss of the feature, This appears to be happening in Cairo, Miller (forthcoming), ina study of the language of Upper Egyptian immigrants to Cairo, notes that ( feature that will tend to be lost not only in second-generation Upper Eayptian speakers, but often in first-generation ‘ones as well. Despite the quite large Upper Egyptian population inthe city, (3) does not migrate well 0 Cairo. By ‘contrast, I argue thatthe coexistence of (I) and (3) in Maiduguri Arabic, described above, is due tothe general lack. of overarching norms of speaking Arabic in Maiduguri, norms that might in the long run lead speakers to adopt either (1) or (3) to the exclusion of the other form. Salieney is a linguistic, perceptual construct whose effects in language change need to be worked out according to the local social configuration. | | | | Owens: Arabic Dialect History and Historical Linguistic Mythology 735 both the geographical and temporal boundaries can be plausibly delimited, The following eight linguistic and historical arguments are relevant. A number of these have been outlined in some detail in previous sections, and will therefore be repeated more perfunctorily here.°8 ‘Argument I. Temporally, as noted above, (3) must have existed in Upper Egypt early ‘enough that we can account for its universal occurrence in Chad. Though our historical knowledge of the Arab settlement of Chad is relatively modest, it can be surmised that this would be no later than 1500, and in all probability was earlier. ‘Argument 2, Sudanic Arabic in general provides a good control for measuring the spread of features into Upper Egypt, since nearly all forms of Arabic in this region were filtered via migrations along the Nile valley. Isolated but historically significant isoglosses such as those. sketched in (8) above, are evidence of features which in Egypt were earlier of more wide~ spread occurrence than they are today. A feature like (8c) can be assumed to have innovated in Upper Egypt and then spread into western Sudanic Arabic. (8d) may have a similar in- terpretation. (3) could equally have been a further Egyptian innovation. ‘Argument 3. As explained in (4) above, I assume a two-step process in the formation of, (3) via (2), which is repeated here as (15) (13) step 1 neuktub : -uktub-u —+ n-wktub-w : tuktub-w step 2 aktub : nuktub-u > n-uktub : n-uktub-u It is assumed that (2) preceded (3). If this is the case it would, on the basis of the known distribution of (1), (2) and (3) mean that (2) and then (3) originated in Egypt, as (2) is firmly attested only in Egypt. Argument 4. Behnstedt’s assumption that (3) spread from northern to Upper Egypt is plausible. However, it is difficult to link it to a specific migration, such as Behnstedt pro- poses. On the basis of the relatively small number of shared isoglosses between the western Delta and Upper Egypt. this suggestion in fact is not well supported. Moreover, the other isoglosses which Behnstedt does name, e.g., short high vowel deletion in an open syllable and the gahawa complex are of great generality in the Arabic-speaking world, and so could have been introduced into Upper Egypt independently of (3). Written sources testify to the 38, The following chart, suggested by an anonymous reviewer, may be helpful in presenting the main clements ‘which are factored into the final analysis. In the linguistic attestation I cite only the IPL form, ‘Synchronie Attestation in written Tinguistic sources Migration(s) oan Enypt (Noni ‘hone Ts century onwards ® west delialex none ‘7th century, later, inter alia, from Cyrenaica ne, ne... Upper Egypt one 7th century onwards, 1380 Hawwara yn. (from Delta) Malta one Pre-Hilalian, from North Africa or Sicily n+... maghrebine (Blau) 1000 Late 7h, later, from Tunisia Beet Andalusia 1125 onwards Post 711 nem (Maimonides, legal deeds) Tunis 1400s 1036-1051 Banu Hilal from UE) meee Sudan none 1300s from UE, later, Darfur to Nile nym Western sudanic __none Late 1300s from UE Reyes 39, Short vowel deletion, for instance, is found in Lebanon, the galst dialects of Irag, rural Jordanian dialects, various Gulf Arabie dialects, ete. The gahawa complex is found inthe gala dialects, various Jordanian dialects (e.g the Bduul of Petra), Najdi Arabic, among others, Indced, both ofthese features are so widespread that their presence in bwo dialects is alone not sufficient to establish a close dialectal relationship. 736 Journal of the American Oriental Society 123.4 (2003) constant influx of Arab speakers into Upper Egypt beginning with the initial Islamic con- quest. Argument 2 above establishes that Upper Egypt has gone through various dialectol- ogies, as it were, different sets of isoglosses being present or dominant at different times. It is quite plausible to suggest that (3) originated somewhere in the Nile Delta (not necessarily in the western Delta) and spread South from there. The original population of innovators may have moved on from where the innovation originally occurred, the innovation being assim- ilated to (1) by later migrations, This is treated further in the next point.“ This innovation must have occurred by the early ninth century at the latest, if, as argued here, the form spread not only into Upper Egypt, but also into North Africa early enough that it was ca ried to Malta around 870. Argument 5. As outlined at the end of section 4.2 (example [11)), it was argued that there is independent evidence that innovation in the first person imperfect verb complex. The forms (16) nafrab n-afrab “I drink’, ‘we drink” found in both Shukriyya and Nigerian Arabic were argued to have their origin in Egypt. Admittedly reconstructing the origin of (16) in Upper Egypt does not directly answer the question of where (3) ultimately arose. It helps in two ways, however. First, according to the current analysis, (16) assumes the previous existence of (1-3), and this in turn again confirms that (1-3) were present in Egypt before Arabic was introduced to the Sudanic re- gion. Secondly, it provides yet another index for the presence of an active dialect contact in Upper Egypt (and by extension, the Nile Delta), contact that led to morphological innova- tion in the verb paradigm. It thus increases the plausibility of my suggestion that (3) origi- nated in Egypt. Argument 6. As discussed in section 5, the migration of the Banu Hilal represented the most important large-scale movement of North Africa, and this moved from Egypt to North Africa. It was further argued that it is most plausible to regard them as speaking a variety with (3) before their movement into North Africa. This of course suggests that (3) is an Egyp- tian innovation. I continue this line of thinking in the next point, Argument 7. Though (3) is customarily known as North African, no one has seriously attempted to locate the time and place of the origin of (3). Assuming that a “North African origin means somewhere west of Tunisia (including Spain), if we take the first written attes- tation of (3), sometime in the eleventh century according to Blau (see section 3 above), the form would have had something like 150 years to spread back eastwards to get introduced into the western Delta in the thirteenth century (assuming Behnstedt’s hypothesis). However, this is precisely the time span during which the Banu Hilal were moving westwards from Upper Egypt. Note further that citations of (3) in Middle Arabic texts increase considerably in the twelfth century. As Blau (1985), for instance, notes, it was fairly common in some Judaeo-Arabic texts. These two observations may be related to the contention developed here that the forms originated in Egypt. It does not involve postulating large-scale migrations moving eastwards (presumably from Tunisia, Morocco or Spain) against the general migra- tory trend during this period, and it explains the attested increasing frequency of occurrence 40. The reverse spread from Upper Egypt to the western Delta also cannot be ruled out, though the lack of losses is equally a problem for this interpretation, Given the well-tested migratory circulation noted above be- ‘oveen Upper Egypt-the oases-Cyrenaica-Alexandria, it could well have been an innovation in Upper Egypt which eventually made ils way into the western Delta, Even multiple movements cannot be ruled out. (1) could have orig tated in Upper Egypt, moved into the northern oases, Cyrenaic, into the western Delta, and from there the Haw- \wara migration would have recycled it back into Upper Eeypt Owens: Arabic Dialect History and Historical Linguistic Mythology 737 of (3) from the more westerly Arabic-speaking world: the Banu Hilal were the purveyors of the (3) from Upper Egypt, and their arrival in the area is reflected in its frequency in texts. The evidence that has been presented is circumstantial. Barring the unlikely discovery of new data sources,“? there probably never will be a definitive answer to where (3) originated. ‘The circumstantial evidence is strong, however, that (3) had an Egyptian, not a North African origin, Yet Arabicists routinely think of (3) as North African. For Ferrando (1998: 60) it is an isogloss linking Spanish with North African (Magrebi) Arabic. Miller (forthcoming) characterizes the Arabic of the Delta and Upper Egypt as having a large number of North African features, Woidich (1993), as seen above, assumes axiomatically that (3) is North African. Versteegh (1997: 162) classifies (3) as North African and so concludes that its presence in Egypt originated from the West. This entire discussion, however, suffers first of all from a methodological fallacy. The North African origin of (3) is assumed, not proved. The reasoning appears to be as follow: since (3) is dominantly associated with North Africa today, it must have had its historical origins there as well. Dialectology is enlisted to do double duty: it describes geographical distribution of forms, and it provides labels to create historical linguistic mythology. Secondly, this approach conceals a problem commented upon by a number of Arabicists, namely that (3) does not correlate well with anything else dialectally. Ferrando (1998: 60) notes that Andalusian Arabic, other than (3), does not share many isoglosses linking it with North African as opposed to eastern Arabic dialects. Versteegh (1997: 162) notes that despite having (3), (northern) Egyptian oases dialects have much more in common with Upper Egyp- tian dialects than with North African. Although Behnstedt suggests that the western Delta, was populated by migrants from the west, he notes that it shares few features with North African dialects. ‘This brings me to an eighth argument, namely, the presence of (3) in areas otherwise lack- ing characteristic North African traits. The present hypothesis, that (3) in fact is an Egyptian innovation, is compatible with all of the observations noted in the preceding paragraph. ‘The Egyptian oases have more in common with Upper Egyptian dialects than they do with North African ones, including (3), which is equally of Egyptian origin. The lack of isoglosses linking western Delta and Upper Egyptian Arabic is due, as noted above, to the extreme overlay of dialect contact that Upper Egypt has experienced since the eighth century, and if (3) originated in Egypt, it would not be expected to carry with it North African traits to An- dalusia, except, perhaps, to the extent that these North African traits themselves originate in or were filtered through Egypt, or to the extent that the purveyors of (3) picked up North African traits on their way to Andalusia. This innovation could have occurred either in the western Delta or in Upper Egypt and spread from there. (1) and (3) are attested in both regions, and it is unlikely that such rare events in Arabic dialects arose independently. The broader distribution of (3) in North Africa 441. The relatively high frequency of (1) in texts from Spain and the West, as opposed to those from Egypt, is ex- plained by the observation that it was in the western region that dialectal forms were sanctioned in vatious sorts of genres. More conservative, normative social pressures explain the lack of (3) in written sources in what is here pos- tulated as the homeland of (3), 42. For example, that (3) is atested somewhere in Saudi Arabia, confirming that itis not North African in origin, Ignacio Ferrando points out (p.c.) that in fact Corriente has suggested a South Arabian origin for (3), with an impor- tant role for Egypt inthe development of the form (Corriente 1996: 56 n, 2). My information comes from Ferrando, Thave not seen Cortiente’s evidence 43. “.. ces dialectes sont essentiellement des dialectes égyptiens,” 1998; 86, » 738 Journal of the American Oriental Society 123.4 (2003) and in Chad is plausibly explained by the migration of speakers of (3) out of Upper Egypt into these regions. 8, WHAT LIES AHEAD. The foregoing discussion has been about the history of one feature, not about a dialect. I would make two final points here. First, one lesson that can be drawn from this exer is that at this point in research on the Arabic language one can reconstruct, even if only in approximate terms, the history of individual features. The history of Arabic dialects, how- ever, is a much broader task which has yet to be embarked upon in a comprehensive way. Consider (6) above. There are but two linguistic features, yet they already suggest the basis for three different dialects. Tens of features are customarily adduced to characterize a single dialect, each feature with its own history. It is only by tracing the intertwining, splitting, re- grouping, partial reforming, transformation and accretion of individual features that a broader historical linguistic panorama will emerge. Secondly, while the totality of the evidence that I have presented is circumstantial, itis multifarious. It encompasses attested variants in written sources, reconstruction based on the distribution of contemporary dialects, the linguistic nature of the form itself, here interpreted in terms of saliency, and the histories of the purveyors of the linguistic form, such as we know them and can plausibly relate them to the language of these speakers. A degree of un- certainty has to be admitted in my proposed solution. I suggest, however, that this is an uncertainty that will have to be tolerated in any Arabic historical dialectology, and that ul- timately the methodological rigor of historical linguistics leads to probablistic rather than de- terministic interpretations. However, Arabic itself is particularly challenging for exploring, the interpretive interplay of sources utilized in such an exercise, offering as it does evidence from a rich array of materials. REFERENCES: Abun-Nast J Press. Aguade, Jordi; Patrice Cressier; and Angeles Vicente, eds. 1998. Peuplement et Arabisation au Ma- ghreb Occidental: Dialectologie et Histoire. Zaragoza: Case de Velazquez. Agius, Dionisius, 1996, Siculo Arabic. London: Kegan Paul International Behnstedt, Peter. 1985. Die nordjementischen Dialekte, Teil 1: Atlas. Wiesbaden: Reichert. 1977. Sprachatlas von Syrien. Wiesbaden: Harrassowitz. 1998. La frontiére orientale des parlers maghrébins en Egypte. In Aguade etal, eds., 85-96, 2000. Review of Owens (1998). Afrika und Ubersee 83: 144-46. Behnstedt, Peter, and Manfted Woidich. 1985. Die deyptiseh-arabischen Dialekte. Wiesbaden: Reicher. Bianquis, Thierry. 1998. Autonomous Egypt from Ibn Tulun to Kafur, 868-969. In The Cambridge History of Egypt, vol. |: Islamic Egypt, ed. Catl Petry. Cambridge: Cambridge Univ. Press. Blane, Haim, 1964, Communal Dialects in Baghdad. Cambridge: Center for Middle Eastern Studies. Blau, Joshua. 1981. The Emergence and Linguistic Background of Judaco-Arabic. Jerusalem: Ben-Zvi Institute forthe Study of Jewish Communities in the East. Brauklimper, Ulrich. (1993), Notes on the origin of Baggara Arab culture with special reference to the ‘Shuwa, Sprache wd Geschichte in Afrika 14: 1346. Brett, Michael. 1974. The Zughba at Tripoli, 429H (1037-38). Annual Report of the Society for Libyan Studies: 41-47 1978a. The Arab Conquest and the Rise of Islam in North Africa, In Fage, ed., 490-555. 1978b, The Fatimid Revolution (861-973) and its Aftermath in North Africa. In Fage, ed., 589-636, 1987. A History of the Maghrib in the Islamic Period. Cambridge: Cambridge Univ. Owens: Arabic Dialect History and Historical Linguistic Mythology 739 Brockelmann, Carl. 19822, Grundriss der vergleichenden Grammatik der semitischen Sprachen. Hil- desheim: Olms. Chambers, Jack; Peter Trudgill; and Natalie Schilling-Estes, eds, 2002, The Handbook of Language Variation and Change. Oxford: Blackwell. Corriente, Frederico. 1976. From Old Arabic to Classical Arabic through the pre-Islamic Koine: Some Notes on the Native Grammarians Sources, Attitudes and Goals. Journal of Semitic Studies 21: 62-98, 1996. Introduccién a la gramética comparada del semitico meridional. Madrid Consejo Superior de Investigaciones cientificas. Cowan, William. 1960, A Reconstruction of Proto-Colloquial Arabic. Ph.D. thesis, Cornell Univ. Fage, J. D., ed. 1978. The Cambridge History of Africa, vol. 2. Cambridge: Cambridge Univ. Press. Ferrando, Ignacio. 2000, The Arabic Language among the Mozarabs of Toledo during the 12th and 13th Centuries, In Arabic as a Minority Language, ed. J. Owens. Berlin: de Gruyter. Pp. 45-87. ‘ischer, Wolfdietrich, 1995, Zum Verhiiltnis der neuarabischen Dialekte zum Kassisch-Arabischen. In Dialectologia Arabica, ed. Tapani Harviainen et al. Helsinki: Finnish Oriemtal Society. Pp. 75-86. Fischer, Wolfdietrich, and Otto Jastrow. 1980, Handbuch der arabischen Dialekte. Wiesbaden: Harrassowitz Fiick, Johann, 1950, Arabiyya. Leiprig: Akademie Verlag. Fisher, Humphrey. 1977. The Eastern Maghrib and the Central Sudan. In Oliver, ed., 232-330. Fleisch, Henri. 1960. Arabiya. £7, Leiden: Brill, Pp. 573-78. Garcin, Jean-Claude. 1976. Un Centre Musulman de la Haute-Egypte médiévale: Quus. Caito: Institut Francais d’archéologie Orientale du Caire. Hetzron, Robert. 1976. Two Principles of Genetic Reconstruction. Lingua 38: 89-108. Hillelson, $. 1925. Sudan Arabic: English-Arabic Vocabulary. London: The Sudan Government. Holt, Peter, and M, Daly, 1979. The History of the Sudan. London: Weidenfeld and Nicolson, Holt, Peter M.; Ann Lambton: and Berard Lewis, eds. 1970. The Cambridge History of Islam, vol. Cambridge: Cambridge Univ. Press. Hopkins, Simon, 1984. Studies in the Grammar of Early Arabic. Oxford: Oxford Univ. Press Hrbek, Ivan, 1977. Egypt, Nubia and the Eastern Deserts. In The Cambridge History of Africa, vol. 3, cd. J. Fage and Roland Oliver. Cambridge: Cambridge Univ. Press. Pp. 10-97 Johnson, H. 1967, The Fulani Empire of Sokoto. Oxford: Oxtord Univ. Press. Kaye, Alan, 1972. Arabic /siiml. Linguistics 72: 31-72. 1976. Chadian and Sudanese Arabic in the Light of Comparative Arabic Dialectology. The Hague: Mouton, Larcher, Pierre, 2001. Le parler des Arabes de Cyrenaique vu par un voyageur marocain du XIIle sid cle, Arabica 48: 368-82, Lewis, Bernard. 1970, Egypt and Syria. In Holt et al., eds., 175-230. MacMichael, H. 1967. The Tribes of Northern and Central Kordofean. London: Cass. Mohammadou, Eldridge. 1978. Tradition historiques des Foulbe de UAdamoua, Tokyo: institute for the Study of Languages and Cultures of Asia and Africa. Miller, Catherine, Between Accommodation and Resistance: Upper Egyptian Migrants in Cairo. Forthcoming in Linguistica Molan, Peter, 1978. Medieval Western Arabic. Ph.D. dissertation, Univ. of California, Los Angeles. Owens, Jonathan. 1983. Libyan Arabic Dialects. Orbis 32: 97-117. - 1993a. Nigerian Arabic in Comparative Perspective. Sprache und Geschichte in Afrika (SUGIA) 14: 85-176. 1993b. A Grammar of Nigerian Arabic. Wiesbaden: Harrassowitz 1998a, Case and Proto-Arabic. Bulletin of the School of Oriental and African Studies 61: 51-73 and 217-27. 19986, Neighborhood and Ancestry: Variation in the Spoken Arabic of Maiduguri, Nige- ria. Amsterdam: Benjamins. Pommerol, Julian. 1990. Da Hayyin (tome 4). Ndjamena: CEFOD. Reichmuth, Stefan, 1983, Der arabische Dialekt der Shukriyya im Ostsudan, Hildesheim: Olms. 740 Journal of the American Oriental Society 123.4 (2003) Rheinhardt, Carl. 1972 (1894). Bin arabischer Dialekt gesprochen in ‘Oman und Zanzibar. Amster ddam: Philo Press, Rhodokanakis, Nikolaus. 1911. Der vulgdrarabische Dialekt im Qofaar (zfaar), 11. Vienna: Holder. Roth, Arletie. 1978. Esquisse grammaticale du parler Arabe d’Abbeché. Paris: Geuthner. Sibawaih, Uthman, Al-Kitaab. Ed. H. Derenbourg. Hildesheim: Olms. Socin, Albert. 1901. Diwan aus Centralarabien Il, Leipzig: Teubner, Vanhove, Martine. 1998. De quelques traits préhilaliens en Maltais. In Aguade et al, ed. 97-108. Versteegh, Kees. 1997. The Arabic Language. Edinburgh: Edinburgh Univ. Press. 2002. Dead or Alive? The Status of the Standard Language. In Bilingualism in Ancient So- ciety, ed. J. N. Adams, Mark Janse, and Simon Swain. Oxford: Oxford Univ. Press. Pp. 52-74. Woidich, Manfred, 1993. Die Dialekte des Agyptischen Oasen: Westliches oder dstliches Arabisch? Zeitschrif fr arabische Linguistik 25: 340-39, 1997. Egyptian Arabic and Dialect Contact in Historical Perspective. In Humanism, Cul- Ture and Language in the Near East, ed. Asma Afsaruddin and A. H. Mathias Zahniser. Winona Lake, Ind.: Eisenbrauns. Pp. 185-97. Zeltner, Jean-Claude. 1993, Premigres migrations arabes du Fezzan au Kanem: Une bréve communiea- tion, SUGIA 14: 81-84,

Vous aimerez peut-être aussi