Download as pdf
Download as pdf
You are on page 1of 52
ARABIC SOCIOLINGUISTICS By JONATHAN OWENS University of Bayreuth rE the 1960'S Sociolinguisties emerged as a semi-autonomous disci- pline, a core methodology and theory developing around the work of William Labov, and adjunct perspectives being added by various scholars.* Arabic sociolinguistics has grown up largely in the wake of this development, though in certain respects was present, as it were, at the founding of the discipline. At the same time, what can, on a nec essarily post hoc basis, be identified as central sociolinguistic issues have heen a part of Arabic linguistics from its very inception. ‘Taken broadly, any aspect of language which correlates with socio- logical categories may be included in sociolinguistics. An exhaustive coverage thus entails not only a consideration of the behavior of specific linguistic forms relative to social constructs, but also the embeddednes of holistic concepts like ‘language’ and ‘dialect’ in a social matrix, via such linkages as language planning and language politics. Rather than say a litle about many topics, [will attempt to focus on what I consider {as of 2000) the core areas of sociolinguistics, namely the association between linguistic forms and social categories. Above all the status of linguistic variants is a central issue, for each variant, supported by a social scaffolding, conveys information about the society it is used in Furthermore, in the Labovian paradigm, understanding linguistic vari- ation ideally gives insight into language change, which is seen not as abrupt, but as spreading outwards from various loci of social innova. tion. Linguistic variation is seen as an index of the spread, mainte- nance or regression of linguistic features, Embedded in the modem linguistic tradition, sociolinguisties has con- centrated on the spoken word. It thus stands in close relationship wo dialectological traditions, though in contrast to dialectology, the main * T would like to thank Pierre Larcher for giving an initial orientation to this article. © Koninklijke Brill NV, Leiden, 2001 Arabica, come XENI 420 JONATHAN OWENS focus of variational sociolinguisties has been on urban areas. ‘These are ecological niches which are hard to define with traditional dialectolog- ical methods, though given the high degree of urbanization in the Arabic world,' indispensable to an adequate coverage of Arabic. In all these respects, Arabie sociolinguistics has followed the broader devel- opments of the discipline. The concentration on the spoken word has had the effect of creating a distance from the traditional (mainly European) Arabicist philological tradition, with its focus on the written text. While there are, doubtlessly, unbridgeable differences between the two foci (e.g, by their very nature, written traditions tend to suppress the type of variation found in spoken language), one can attempt to mediate the differences. I will suggest in section | that sociolinguistic readings of the Arabic grammatical tradition are compatible with the analysis of written texts ‘The rest of the article focuses on the status of variational phenomena in modern Arabic 1. Prote-sociolinguisties While the standard Western grammars of Classical Arabic (e.g. Wright, Reckendorf, Blachére and Gaudrefroy-Demombynes) describe the lan- guage effectively and exhaustively, they approach the language as a product to be packaged, described, and consumed by the interested public. Such is the nature of descriptive grammars. What is singularly ing from them, however, is the recognition that Classical Arabic as it has become known to us today, is itself the result of an evalua- tive process whose input was a considerably wider and more disparate set of linguistic values than is commonly recognized today. Probably this heterogeneity is nowhere more in evidence than in the work of the father of Arabic grammar, Stbawayhi. The question of heterogeneity in Sibawayhi is one which has hardly been broached in the Arabicist literature, and it is hardly the place here to attempt a summary. One illustration may be offered, however, indicating that the Arabic lan- guage which Sibawayhi ‘constructed’ (a grammar by definition is a for- mal construct) was a variable object, one parameter of whose variability was defined by the social categories which Sthawayhi drew on to ori- entate his thinking. In his Aitad, Sibaway! described a good deal of linguistic variation. ARAIIG SOCIOLINGUISTICS 421 This may be broadly divided into variation of two types, internal and external. The internal variation, which is probably the dominant cate- gory, derives from Sibawayhi’s linguistic theory, A topicalized noun, for example, appears in nominative case, but, according to the logic of linguistic rules, it may equally be accusative (e.g. I: 31 ff, see Owens, 19984). The external derives from variation which is legitimized through its association with groups of various sorts. By way of example, in the first 100 pages of the Aid T count 26 such references, which may roughly be grouped into seven different categories, as in (1), (all refer- ences are to Book 1). (la) Areal ~ tribal variation (higizt os. temimi) e.g. ma al-higa (21.20, 94.9) (b) Tribal variants, e.g. Bani Sulaym (51.6)° {c) General group, e.g. “Bedouins”, ‘arab (62.13, 95.15) (d) Unspecified groups, e.g. “some”, “some Arabs”, “someone whose Arabic T trust” (62.14, 97.16) (c) Groups defined by other linguistic characteristics, “those who say x also say y”, eg “those who say “akalant al-baragit” (30.17) {f) Majority groups, common knowledge, e.g. a form is “better known” (a’iaf) or “more common in speech” Vakiar al-kalam) (66.9, 90.10) (g) Stbawayhi’s evaluative criteria, e.g, “good Arabic”, “eloquent Arabs” (21.15, 66.11, 91.15) ipa Each of these in its own way rellects the social embeddedness of Stbawayhi's linguistic thinking. In all instances Stbawayhi legitimizes the citation of a particular form by associating it with various types of units that must have been recognized in the late eighth century Abbasid empire. In (Ia, b) the groups have a geographical or social basis. (Ic) attests to the normative value of Bedouin speech (see Blau 1963). (Id) shows that the ideolect may have normative value and (le) indicates that bundles of features may be associated with another feature stereo typically associated with a particular group (unfortunately in this case, never specified more closely by Sibawayhi). (If) suggests that degree of numerical extension can be a basis for preferring one form or another. Finally, (1g) draws attention to the institutionalized role of the linguist in deciding what is good and bad in the language (see Carter 1972)! © hiasr is a geographical designation; tamint is cbal, though eponyinously may serve as a designation nf an eastern varity. * Here. and olten elsewhere, Sibawayhi's repons are mediated by other language experts, * Canter emphasizes the sovially-orientated metaphorical vocabulary which Sibawayé 422 JONATHAN OWENS mawayhi’s Aitad is a work which a sociolinguistic reading would help elucidate, there are other segments of the Arabic grammatical tra- dition where variation was institutionalized by the grammarians them- selves. Notable in this respect are the different Koranic reading tradi- tions, the gird’ae. The canonization of certain reading traditions, the seven of Ibn Mugahid (d, 924) for instance, represents a formal eate- gorization of a fixed set of variants which in and of itself is mainly of grammatical and perhaps exegetical interest. The existence of the vari- ants, however, reflects a socio-political tension in early Islamic society whose history has been traced in the work of Beck (1946), Kahle (1948), Jeffery (1948) and others, They show that Arabic Koranic readers were subjected to ever tighter strictures in their choice of Koranie variants, culminating in the official proscription of the readings of Ibn Miqsam and Ibn Sanabitd in the early part of the tenth century. Crucially, a scholar like Ibn Sanabid argued that he was, within the limits set by the consonantal mushaf (Koranic text), free (© use certain readings so long as they represented good Arabic. This viewpoint was challenged by the authorities of the time, who forced him to bow to general con- sensus of the Koranic readers (*igma, “amuna; see Gilliot 140) in his choice of variants. Icis thus clear that the Arabic grammatical tradition itself gave explicit recognition to the existence of linguistic variation in the language, a variation which was tolerated, legitimized or proscribed according to social and political institutions with which the variation was associated.” 2. Variation and language history Both in the Arabic grammatical tradition, and in its western reception, variation was either channeled into categories which effectively reduced it to formal linguistic variants, as with the gird’at described above, or ignored. The latter was the favored approach of many Western Arabicists, and Semiticists. Thus, the work of Beck (1946), Kable (1948), Vollers (1906) and Spitaler (1953) among others, which emphasized old tradi- tions parallel to or even interacting with standard, orthodox ones, was i upright”), not the institutionalized social role of Sibawayhi as linguist, a role wl rier in fact would probably deny (Carter 1985) © See Disher (1987 part 1) tor Furth classical period. sociolinguistic interpretations of Arabie i the 423 overshadowed by that of Brockelmann (1908, 1982!) Néldeke (1910), Bergstraber (1928, 197°) and Fick (1950). Although these latter writers recognized the existence of varieties of Arabic, in particular old ‘dialects’, contemporary with Classical Arabic, they never developed a coneep- tual framework explaining the functional co-existence of and linguistic interrelationships between these varieties, nor their historical development. A crucial step in this direction, a step which probably marks the beginning of Arabic sociolinguisties as an academic entity in its own right, was made by Ferguson in his well-known article “Diglo: (1959a}, In the article Ferguson gave theoretical flesh and comparative generalization to a situation which had long been recognized and described, in particular by French linguists working in North Affica e.g. Marcais 1930, 31). In an idealized model diglossia is a socio-polit- ically regulated linguistic situation, where one linguistic variety has a higher status than another (or others), and in which linguistic functions are partitioned between the two in complementary fashion.” In what T term ‘classical diglossia’ (following Fasold 1984), the Hand L varieties belong in some sense to the same language, yet are distinguished by clear structural differences.’ High fictions include the use of the lan= guage in formal occasions, and literary and religious fumetions, while low fimnetions inchide language activity in the home, talk hetween friends, and the marketplace. One important element of the high variety is its link to a valued cultural past, In the instance of Arabic, of course, it is the Standard language, a variety whose grammar is largely derived from Classical Arabic, which is the high variety, and the so-called dialect {sce nm. 7) the low. ‘The appeal of Ferguson’s model is that it provides an explanation for the maintenance of more or less structurally divergent forms in one and the same speech community. It is notable that Ferguson’s model incorporated the spoken language, thus marking a shift away from the philological Arabicist tradition orientated mainly towards the interpre- tation of written texts © Diehy (1994) characterizes the Arabic sittation as “pluriglossie’, understanding this term co relee co different leets of the same language. This allows one to distinguish Arabie from diglosic configurations where different fimetional domains are filled by completely different languages ‘see 1. 7) “Opposed to elasic diglossit would stand both Fishman’s functional diglossia (E97E 74), where different languages may fil different functional niches, and register-basee varieties, where ip both struevural and fianetional terms differences between Hand [. varieties are mot so sharply delineated as in Arabic, | 424 JONATHAN OWENS Although there is not the space here to develop the argument in any detail, it may be suggested that Ferguson’s undeniably elegant syn- chronic characterization of the status of linguistic variants in Arabic has not had desirable consequences for a diachronic interpretation of the language, especially the Labovian integration of modern variation- ist studies with questions of language change. In particular, Ferguson not only described contemporary diglossia, but also projected its ori- gins back into the time of the early Arabic diaspora in the seventh and eighth centuries. Ferguson (1959b) argued that the dialects developed out of a koine which arose in the military camps of the Arab con- querors. Contact (see section 8) produced a degree of simplification and leveling among the dialects, while a variety akin to Classical Arabic continued to be spoken, for a time at least, among Bedouins. Developments of Ferguson’s koine hypothesis, in the sense that the dialects are viewed as being derived from and innovative with respect, to Classical Arabic, are Versteegh’s (1984) proposal that the dislects underwent a stage of pidginization before stabilizing in the form ances- tal to the modern dialects and Blau’s model (1966/67, 1981) whereby the synthetic language of the Bedouins, close to Classical Arabic, gave way to an analytic Middle Arabic, under the influence of widespread Arabicization of non-Arab urbanites.’ In this perspective, the clear diglos- sie dichotomy Standard Arabie vs. dialect is projected back to the early origins of the language, with the modern low variety, the dialect, being seen as derivative from and a corruption of, Classical Arabic (see Versteegh 1983 for origins of this idea). Since the dialects are perceived as being younger than Classical Arabic, whatever interest their study may have for modern Arabie society, they can make only a temporally limited contribution to understanding. proto-Arabie. If, on the other hand, Classical Arabic is itself the endpoint of a development (as argued for by Corriente, 1976), which is shared by other varieties of Arabic (e.g. Rabin’s 1951 dialects), it follows that the modem dialects themselves represent a heritage as old as that of Classical Arabic, In this perspective the ancestors of the modern dialects did not arise diglossically afier ¢. 700 AD, but rather existed contemporane ously with and contributed to the development of Classical Arabic Logically, they derive with Classical Arabic from a common ancestor " Notwithstanding Blau's more recent interpretation of Midlle Arabic as a stylistic “Mischsprache” of Old and New Arabic (1982), his earlier inteepectation contintes te serve as one possible model for the conceptalization of the history of Arabjie ARABIC SOCIOLINGUISTICS 425 in a yet to be reconstructed proto-Arabic. In other words, one must distinguish between the social construct of diglossia and the individual linguistic histories of the entities constituent of diglossia. Tt may be suggested, for lack of space without exemplification (see Owens 1998a), that adopting Corriente’s perspective will have conse~ quences for evaluating modern linguistic variation, particularly that dis- cussed in 4.2 below, for, assuming Labov’s (1994: 21) uniformitarian principle, the variation associated with different contemporary groups may allow insights into the earlier development of the language, even, those prior to the Arabic diaspora of the seventh century 3. Spoken Arabic: Levels and Gradients Ferguson’s diglossia model is valuable not only for its heuristic value in defining language varieties in terms of two idealized types, Standard Arabic and dialect, It is also important for the fact that only with the integration of the two types within a single conceptual framework could serious comparative work between the varieties begin. This took two directions. On the one hand purely structural studies were carried out defining similarities and differences between the two (e.g. Altoma 1969), On the other, a more fruitful line of research was opened up in the attempt to define the use of the two in contemporary Arabic society. Observation of the spoken language quickly revealed that in practice native speakers of Arabic who had access to both the standard lan- guage and the dialect in any given stretch of speech rarely used purely fone or the other variant. The initial approach to this descriptive prob- lem was to posit a series of discrete levels ranging between the ideal Standard/dialect poles. This, for instance, is the approach followed by Blane (1960). He defined five levels of speech, (1) plain colloquial, (2) koineized colloquial, (3) semi-literary, (4) modified classical, (5) standard dlassical. Each of these levels is characterized by linguistic traits. The koineized colloquial, for instance is a “plain colloquial” into which lev- ling features have been introduced, which is to say highly character- istic colloquialisms, such as the f sg object suflix -¢ such as is found in Jordan, southern Iraq and western Arabian varieties, are replaced by common koine forms (see below}, such as f sg -ik “Modified classical” is Classical Arabic with various dialectal admixtures. Badawi (1973) develops a typology along similar lines. Similar characterizations have been applied to the written language as well, notably Blau (1981: 25) who distinguished three main levels of Middle Arabic, Classical Arabic 426 JONATHAN OWENS with Middle Arabic admixture, semi-Classical Middle Arabic and clas- sicized Middle Arabie. ‘The discrete levels approach ran into problems of two sorts (sce Owens and Bani Yasin 1991), One was the sheer methodological problem of dealing with variation across features which ranged over the entire language, from phonology to semantics, As Blane realized from the beginning (1960: 85), assigning a given text to one level or another was difficult simply because it might be variously classified according to the Classificatory criteria used or the stretch of text chosen (Farrier 1991: 7). A second problem pertains (0 the relative status of the two ideal poles, Standard Arabic and dialect. While the two may be presented as structurally opposed to each other, in terms of actual competence they are not of equal status. Whereas the dialect is a native variety of Arabie? and henee, by definition perfectly learned by all Arabs, the Standard language is a variety learned as a second language (see 4.1). ‘The recognition of this state of affairs is of conceptual and descriptive consequence, Diem, for instance, in his description of Lebanese, Syrian and Egyptian Arabic texts operates with a simplified levels classificatory system. He describes the speaker in one text (1974: 76) beginning in dialect with SA interference, then switching over to SA. Given the non- native nature of the Standard, however, it is not clear in what sense one can speak of interference from the Standard in a dialectal text ‘The notion of interference in second language learning usually pertains to the appearance of traits from a speaker's native language in the tar get. Furthermore, while theoretically interference from the dialect in the Standard may be recognized, it is not self-evident what precisely the nature of the Standard is that educated Arabs speak when they target this variety (see below). ‘The problem is not solved by terminology from within Arabic itself While an ideal contrast, analogous to Ferguson’s H-L exists in Arabic (ammiyya ~ dariga etc. vs fusha), Parkinson (1991) points out that there is no consensus within the Arabic world (or Egypt, Parkinson’s locus In fact, L believe the term ‘native Arabic” = “NAY is a hetter designation for variety chan ‘dialect’, and is the term [ henceforth adopt. Conventionally, the regio variant of NA will be indicated with a prefixed letter or letters signifiying the regions E-NA = Egyptian native Arabic, for example “Tn a recent work, Holes (1995: 279 1f) also adopts a levels approach to dhe descrip tion of Gairene Arabic, ‘The fact that Holes’ work has a pedagogical orientation su gests dhat the levels approach is more appropriate as a broad summarizing tool than as a precise categorization of linguistic variation, ARABIC. SOCIOLINGUISTICS 427 of research) as to what, in particular fisha stands for. Parkinson illus- trates his point on the basis of a matched guise he carried out with seven different versions of a nearly identical text, the only differences between them being a gradually decreasing degree of SA features. Versions 1-3, for example had full final short voweling, version | & 4 ¢ = [a5], version 2 = 4 ¢ = [el (note, Egyptian [4]), and so on, each lower version adding, relative 10 the SA. prescriptive rules, more E-NA traits, Asked to rank the versions on a “fisha” scale from 1-7, generally the rankings followed the structurally-planned. sequences. Version 1, for instances, had the “best” isha ranking (though version 7 ranked better than 6, 4 better than 3). Such consistency suggests that atleast according to the variable traits picked out by Parkinson, Egyptians do react consistently to stereotypes of what fishd is. On the other hand, the large majority of all evaluators considered all seven versions to be fushd. The boundaries where fushd end are enigmatic. Ic is in the mid-70’s that an alternative approach to the analysis of linguistic variation in spoken Arabic emerged. In addition to the ide- alized poles, SA and dialect, ‘T. F. Mitchell developed the idea of Educated Spoken Arabic (ESA), the variety of Arabic spoken typically by educated Arabs consisting of elements from both SA and the dialect, and possessing hybrid forms unique to the ESA level.!! Mitchell's main interest was in an interregional koine which he and his co-workers stud- ied in Lebanon, Syria, Jordan and Egypt. The actual model used to describe ESA varied from scholar to scholar. Mitchell himself (1976, 1980, 1986) continued to work with discrete levels, He argued that speakers chose from a range of comparable forms according to para- meters set by stigmatization and degree of formality. His presentations tend to be rather programmatic and at times hard to generalize. The notion of stigmatized forms is not defined, though they appear to be those which are associated exclusively with one dialect, and which are clearly distinetive from SA, Jordanian fad “this m”, for example \ ‘Two earlier works adumbrate Mitchell's approach, Bishai (1966) speaks of “mod fem fimer-Arabie", ostensibly CA without ease endings, though in fact a more mixed variety. The account, however, does not generalize beyond a summary of the cata two texts, Kaye's (1972) characterization of spoken Modern Standard Arable as an defined systert i similar to Mitchell's ESA in the sense that both authors describe system) open to variation along a range of parameters, Talmoudi (1984: 143), basing his study’ om a short text produced by speakers from Tunisia, Algeria and Morocco, speaks of classiczation, interdialectalization and colloquialization of featnres, descrip: Hively adequace terms fiom a stevetural perspective perhaps, but barring more subtle Jon open. to the criticisms directerl against a levels approach, ifr 428 JONATHAN: OWENS Unstigmatized forms themselves are divided between formal and non- formal. Syrian Arabic Aabir “big” (without case ending and hence part of ESA, not SA) for instance is classified as formal, whereas Abir (dele~ tion of a) is informal. ‘The problem with Mitchell's approach is that however much his examples in isolation appear to be intuitively cor- reet (though sce Abdel Jawad and Sulaiman 1990: 296), his classificatony categories are not always readily discernible in complete texts, where forms of potentially different categories frequently stand cheek by jowl. Papers by two of Mitchell’s co-workers, El-Hassan (1979) and Sallam (1980), gave a more detailed and elucidating account." El-Hassan stud- ied the occurrence of demonstratives in about 21 hours of recorded speech by Egyptian, Lebanese, Syrian and Jordanian Arabs. In the m sg category, for instance, five forms were used with a high degree of frequeney, Adda, hiiza, hida, hayda and da “this”. The last two forms are limited to one region, da Egyptian and hayda Lebanese, while hiza is mainly Egyptian. Aida and Adda make up nearly 60% of the forms. El- Hassan began his analysis under the assumption that Adda, neatly iden- tical to SA hada, was the least stigmatized form, hiida medially so, and locally restricted forms, like Jordanian fad (hardly attested), most stig- matized. He hypothesized that among the speech of educated Arabs, the less stigmatized forms would be preferred. His results, however, only partly bear out this assumption, Not only is Aida nearly as fre- quent as ihdda, but there are also presumably stigmatized forms with high degrees of frequency as well. Egyptian de, a form restricted exclu- sively to the Egyptian speakers, makes up nearly 71% (94/113) of all m sg tokens among the Egyptian speakers. Following Mitchell, the study carefully distingnishes between speech situations where a given nation- ality is speaking to his or her own nationality and where the speakers are of mixed nationality (ef, Bell 1984 on variation and audience design). Even when the Egyptians were speaking with other nationalities, how- ever, the percentage of da did not decrease (in fact, it increased slightly. El-Hassan (p. 42) explains this as a desire to “sound Egyptian”, though in fact, the equation: the further one moves away from SA, the greater the stigmatization, is not born out by the facts, which calls into ques- © Meiseles worked within the ‘evels” framework, and suffers from all of its short. comings (see Owens and Bani Yasin 1991 for criticisms). Mitchell continues to employ fa Insie formal/infornial dichotomy in a hargely formal account of cense and modality in Atabie (Mitchell and Et Hassan 199, ARABIC SOCIOLINGUISTICS 429 tion the usefulness of stigmatization as an ordering category. El-Hassan further shows (p. 31) that the variants t not appear to be colloca- tionally determined either. He cites the three examples, all from the same Syrian speaker, hada/haza/hada min nahyi “this is so from one angle”, where the collocation min nalyi, an expression of SA origin, col- locates equally with three different variants. On the other hand, Sallam’s study does show clear instances where the notion of stigmatized form is useful. His study concentrated on the four variant realizations of *y found in the region, namely ¢ ~ & ~ ” ~ q, a set of variants which reappears in many sociolinguistic studies in the area, He notes (1980: 92) that speakers from Beirut, who vary between the local & and the SA pronunciation ¢ when speaking with non-Lebanese show a markedly higher use of g than 4, as if avoiding their local form. ‘The point that emerged from the study of spoken corpuses drawn from a cross-section of speakers is that the classification of variants according (0 a preset scala ranging between SA and NA was problematic. In some instances, as with Sallam’s Beirut speakers, discrete correla- tions do emerge, but with others, as with El-Hassan’s demoustratives, they do not. One of the central themes of Arabic sociolinguistics has been the identification of the parameters defining this. variatios 4. Social Correlations ‘The use of linguistic corpora as the basis for defining Educated Spoken Arabie brought Arabic sociolingu based sociolinguisties, a tradition 4. 1998), and which today is the dominant field of sociolinguisties. In this tradition an attempt is made to explain that part of variation which is not explicable linguistically (e.g. phonological, morphophonological), in terms of sociologically-defined variation, via statistically significant cor relations with various extra-linguistic categories. In the different studies which have been carried out within this framework different points of emphasis have emerged, some inspired by the theoretical orientation, of the researcher, some by empirical tendencies deriving from the data. In this section I will summarize the main parameters which have been shown to correlate with variation in different Arabic-speaking communities. 430 JONATHAN. OWENS 4.1 Education and literacy As already seen in section 3, education is one of the most important elements convibuting to variation in modern-day Arabic. Since World War IT education has expanded enormously in Arabic countries, Because the Arabic used in instruction is, in theory, Standard Arabic, this vari- ety has become accessible to a large segment of the population in a way it has never been in the history of the language. [ts use in edue cation is reinforced by its use in many public forums, including the media, religious contexts and communication between Arabs of different regional origin. [ will adopt Mitchell’s ‘Educated Spoken Arabic’ (ESA) as the name for the variety which results from this variational parameter. ‘The type of variation engendered by literacy in SA has already been illustated in section 3 above. As an idealization, ESA is open to input from its vo constituent varieties, SA and native varieties of Arabic (NA), traditionally called dialects (sce n. 7). However, these two opposed poles are not of equal importance in defining the form of ESA. Rather, NA serves as the basic input, which, depending on any of a number of factors, will be modified in the direction of SA, While this assertion is uncontroversial for most who have experience in living and working (e.g. linguists) in Arabic countries, the distinction between natively learned native Arabic and a formally leaned SA is of such fundamental impor- tance that more clarification is necessary ‘Three reasons may be adduced for this directional interpretation of ESA. First, the high variety, SA, is learned formally in school, whereas NA is learned at home (Ibrahim 1983), In regards to the second point, at least two observed patterns of vari- ation suggest that NA serves as the base variety. One may first observe an asymmetry in the morphemic distribution of the two varieties in the following sense. ‘There tends" to be an implicational relationship between lexical stems and affixes such that SA stems (see next paragraph) co- occur with NA alfixes to a greater degree than do NA stems with SA affixes (Holes 1995: 297). Thus, in the production of hybrid forms a token such as biu-gal “it is said” (Meiseles 1980: 153 on Bgyptian Arabic), with Egyptian Arabic modal prefix bi- prefixed 10 SA prefor- mative vowel + passive verbal stem is more likely than the opposite, (2) puwal “it is said” with SA passive imperfect prefix ju- attached to the Egyptian Arabic passive stem ral, Given that affixal material is © With heavy emphasis on ‘tends’ see Owens and Bani Yasin 1991; 23, ARABIC SOCIOLINGUISTICS 451 generally regarded as crucial in the definition of ‘matrix’ language’ (see e.g, Meyers-Scotton 1993: 98 ff), the wider distribution of NA affixes in ESA points to NA as the basis variety. Further (0 this point, nearly all variationist studies have identified a lexically defined distribution in the occurrence of SA material, For instance, in a detailed study on uvular variation in Amman, ¢ ~ ? ~ g ~ k (see section 3 above), Abd-cl Jawad (1981: 205} shows that the SA phonological variant q strongly correlates with words of SA origin. Moreover, these words with the variant tend to be those which belong, to the ‘cultural domain’ of vocabulary (e.g. qa‘a “lecture hall”, muitaggat derivatives”), words which tend more easily to be borrowed. Those the three NA variants, on the other hand, tend to be either basic vocab- ulary (gadéet “how much”, iz’ “blue”) or belong to the less formal cul- tural domain (ibik “jar”, igid “necklaces” 1981: 367 ff)."' Tt appears possible here to speak of a borrowing of cultural words of SA origin into an NA core (see section 5 below). Thirdly one may adduce emotional and physiological factors. When speakers are excited or angry, they will tend to use NA variants (how often does one hear a curse in SA; see Parkinson 1985: chapter 8), and fatigue tends to favor NA {Tarrier 1993: 104), NA is a refuge in extreme emotional and physical states. Early studies focusing specifically on E tribution of single linguistic features, without defining, other than in Mitchell’s and Meiseles’ rather programatic terms! the range of styles which may be found within it, Parkinson (1994) is probably the first study which attempted to characterize the extent of variation found in texts of speakers who were specifically asked to produce oral texts fushd. Out of a large corpus he chose four speakers with education ranging benween one year of high school and four years of university. While recognizing that a full study must range across a broad spec~ trum of variables, Parkinson concentrated on the presence of SA short final vowels," both lexical (bind’) and inflectional (’rd8), The four speak- ers display a great deal of variation in their use of vowels, the rate A concentrated on the dis- % See also Abdel Jawad and Sulaiman (1990: 298) © Meiseles (1980: 125) distinguished herween oral literary Arabic ithe spoken lan- wuage of educated Arabs attempting w approximate LA) and ESA, though his mai dlistinguishing feature (presence vs absence of case vowels) is, as Parkinson (1994) shows, ‘more programmatic than a reflection of reality. Parkinson counted as SA vowels only those with a characteristic SA form. ‘The epenthesic Egyptian Arabic was exeluded from the couns 452 JONATHAN OWENS ranging between 2 and over 50 per page {his unit of counting). Three of the speakers, however, have only 12 occurrences or less, Generally the use of the vowels was correct.” Parkinson concludes that although there are individuals capable of speaking fluently with SA final short vowels, the lower percentages ate more typical of Cairene ability gen- erally. While Parkinson notes, impressionistically, that degree of vow- cling tends to correlate with other SA features (1994: 184), his study makes no attempt to teeze out correlations, even of a tendential sort, between final voweling and other linguistic structures. Before moving on to the next section, the work of Eid (1982, 1988) may be mentioned as an alternative framework to the analysis of spo- ken Arabic in terms of a medial ESA variety. Eid is concerned with the type of spoken Arabic treated in section 4.1 under the rubric Educated Spoken Arabie. She assumes two opposed varieties, NA and SA, with switching between the two. She herself is aware of the difficulties in this approach, noting that the form =f, for instance (1988: 56), consists of a SA stem, ra’ and an NA suffix, -# (vs. SA -aytu), a typ al ESA hybrid form, Eid, however, must assign each token t one of the two varieties, here, for example, recognizing the stem ra’- as being criterial, hence SA. Eid is interested in defining structural constraints between the two varieties. ‘The interest in her work is not so much in the proposed constraints," but rather in the very application of the codeswitching model to her data. Her own data is based on Egyptian radio and television programs with panel discussions by learned par- ticipants, Such a group will probably come closer to fulfilling a usual prerequisite to codeswitching than do most Egyptians, namely that they have a fluent command of SA. Even if such an approach can be applied to such a group and the example of 1a’é above cautions that com- promises of uncertain theoretical status must he made—it is doubtful that it could be efficiently applied in less formal settings or with speak- ers with a lesser command of SA. © Unfortunately, Parkinson docs not use a precise index, with actual occurrence of vowel-final forms expressed as a percent of all contexts where a final vowel is expected, As Daher (1987: 140) points out, the universality of Eid’s examples ate limited wo her own corpus. Contradictory examples can be found even within the corpus of ESA, as for example her constraint (1982: 69) that the structure “x = SA + ms = NA+ y verb} = SA camnot occur. In Meisetes (1980: 133) one finds noha brnayal “anna “we say that...” with the Egyptian 6i- tense prefix sandwiched berween SA material see 8.1 below and Wilnsert 1996 for recent discussion) ARABIC SOCIOLINGUISTICS 433 ‘These criticisms aside, the codeswitching approach is an interesting one. One may ask, however, that its theoretical underpinnings be more explicitly formulated than hitherto in regards to its application to two varieties of the same language. Walters (1996: 180) points out that Myers-Scotton’s (1993) markedness model of code choice, developed on the basis of codeswitching between discrete languages, appears to make the wrong prediction about the type of code switching (specifically intra- sentential) which one finds in Arabic diglossia. Aside from the invoca tion of such overly general parameters as ‘formal’ vs. ‘informal’ (see section 3 above), there is lacking extensive research on the discourse/ pragmatic and social!” framework of Arabic diglossie ‘switching’ ‘Two approaches to the study of spoken Arabic emerge here. On the one hand one can assume a variety which has been termed ESA and attempt to define the range of structures which occur in it. The codeswitching model, on the other hand, proceeds on the initial basis of only two opposed varieties, Exploring the implications of adopting one approach or the other or a combination of two would undoubt- edly enrich our understanding of spoken Arabic. 4.2 Ethnicity and nationality The work of Mitchell and his colleagues referred to in section 3 above took broad-based nationality differences as an important independent variable. In the data summarized in section 3, nationality is, potentially, one parameter explaining differential language behavior. Syrians, for instance, switched to a far greater degree from their NA hada towards SA hada “this”, than did Egyptians, who by and large retained Egyptian Arabic da, Clearly, El-Hassan’s parameter of stigmatization, the dialectal form being more stigmatized than the SA, does little to explain the dif- ferential behavior, since all speakers in the sample are educated. Rather, one may propose that Egyptians may be less reticent about maintaining their local norms in formal situations than are certain other nationalities However, interpreting variation in terms of nationality is difficult for at least two reasons. First, in the above example, (h aida ~ da) internal linguistic differences may play a role, In purely structural terms it isa smaller step for a Syrian to switch from Syrian Arabic hada 10 The former include factors cuing the ‘codeswitching’, for instance topic, entry of new interlocutor, hesitation, highlighting, emphasis. ec. Social factors include the symbolic value of the codes within the community, role relation between interlocutors and code choice. Walters, oddly, omakes no reference to Kid's work in his discussion of codeswitching, 434 JONATHAN OWENS SA hada, potentially nothing more than a phonetic substitution, than it is for an Egyptian to change from da to hada, involving a complete lexical change." Secondly, and more generally, ‘nationality’ may mask a good deal of internal ethnic variation, a point I turn to now. By ‘ethnic’ [ understand any of a number of social parameters by which, non-national social groupings are distinguished, including reli- gion, shared history, skin color, kinship, lineage and place of origin. The relevant criterion or criteria defining ethnicity may differ from place to place, so the present category is potentially a very large one which may eventually have to be broken up into finer component parts. A classic instance of ethnic difference correlating with linguistic difference is Blanc’s (1964) study of communal dialects of Baghdad, where he showed that Arabic-speaking Muslim, Christian and Jewish communi- ties in Baghdad were marked by sometimes striking linguistic differences. Beyond the purely structural description of these differences, it is of ‘equal interest to know the manner in which they are distributed in the community, and what the distribution tells us of the social relations beuveen the groups. An enlightening study in the respect is Holes’ work on sectarian differences between the ‘Arab and Baharna communities in Babrain, These two groups are differentiated historically, the ‘Arab being more recent immigrants, and confessionally, ‘Arab being Sunni, Bahama Shia. ‘There are found considerable dialectal differences between the ‘Arab and three different Bahama groups, Holes (1987) isolates 19 variables for comparison, On the whole, Tam being brief here and simplify the Bahama groups somewhat, the contrasts mark ‘Arab vs. three different Baharna (Baharna) dialects, eg. ‘Arab 4 dy vs. Baharna ff d, 2" in the realm of phonetics, or ‘Arab drisat “she studied’, pkilbin “they read” vs, Baharna darasat ~ dirsat, ykithin ~ yiktuban®? in the morphophonogical realm. Holes study shows wo tends worthy of mention, First, Bahara tend to move towards ‘Arab phonological variants, whereas the reverse does not occur, Thus, Baharna will switch from their own g to ‘Arab y to some degree (e.g. 94 ~ ga “he came”), whereas ‘Arab make the reverse switch to a far smaller extent (ya, Holes 1985, 1987: $7 ff). This asym- metric movement is explicable through the fact that the ‘Arab are the © Not to mention syntactic differerences as well: both SA Auda and Syrian Arabic ‘hada (Cowell 1964 557; occurs pre or post N. whereas Egyptian Arabie dr is exelu- sively post-nominal, © One of the three Bahama dialects (Baharia Ill) has y as the norm, © Differing for the three Baharna dialectal geonps, ARABIG SOCIOLINGUISTICS 435, dominant political group in the country, and hence their variety com- mands a degree of local prestige (see below). Secondly, variation is governed not only by communal norms, but by SA influence as well. In this respeet both the ‘Arab and Baharna communities display the same bifurcation, namely that educated speak- ers adopt SA forms to a far greater degree than do illiterates. ‘These two tendencies interact in ways which render a simple expla- nation of many variational phenomena difficult. For instance, the ‘Arab have a diachronic guttural epenthesis rule of the form *C,,C + G,yaC, hence *majrib + mganb “west”. The Baharna do not have such a rule, hhence majrib, identical to SA magrib. Holes observes (1987: 172) that whereas there is a slight tendency for ‘Arab to move towards the SA form magrib.® there is also a small tendeney for the Baharna to adopt the ‘Arab mgarib. In respect of morphophonological rules, generally speaking the Baharna move in two directions. Where the ‘Arab and Baharna share a common form which contrasts with SA, the Baharna show a stronger tendency to move towards the SA variant. Where, on the other hand, the ‘Arab and Bahama differ, even where the Baharna variant agrees with the SA form, there is a small though consistent ten= dency for the Baharna to move towards the ‘Arab form (as in the example above)."* ‘One point which emerges from Holes’ study is that the ‘Arab com- munity evinces a greater degree of linguistic security in that ‘Arab speak- ers tend to move away from their native norms to a lesser degree than do the Bahama, and when they do it is in the direction of SA. The Baharna deviance from native norms, on the other hand, goes in two directions, towards ‘Arab and towards SA. Why some Baharna features react in one way and others in another is not obvious, ‘The reasons for it are not discussed by Holes (see below). One effect of the Bahama strategy in those cases where their native forms potentially serve as the prestige target variant (e.g. magnib cited above, where Bahama = SA) is to ensure that their variants do not define communal norms. If this 2 The season this i interpreted ay a movement wards the SA variamt, not the Baharna, is that there are no instances where SA and Baharna forms differ in which the ‘Arab converge towards the Balarna. “Arab nan is realized phonetically as [ged] OA third trend sugested by Holes (1987: 146) i uf potental significance, namely that (Sj iterate speskers of orl communities tene to move towars a sct of cominon variants to fir greater degree than do iterate, particularly in the morphiophonolagial realm of verb stems. Holes argues for a covert SA normalizing factor, though the data +5 to comples to allow for # iiple summary, and may support diffrent conclusions. 436 JONATHAN OWENS is intentional, it would be interesting to know whether they do this «0 maintain the ‘communalness’ of their own forms (see below), or out of deference to the dominant ‘Arab group. A second example of ethnically-based variation is attested in the four reflexes of proto-Arabic *g in Jordan. ‘The variation is particularly wide- spread in urban areas, where speakers of different backgrounds have settled. Abd-el Jawad has studied reflexes of this variation in Amman and Ibid (northern Jordan). The variants are g ~ ? ~ g ~ & (gala ~ val ~ gal ~ kal “he said”), For the most part, q is associated with SA." > with speakers of urban origin, & with central Palestinian rural dialects, and g with other rural Palestinian and Israeli dialects (e.g. Galilee and Negev) and rural Jordanian.” The ’ and & variants were brought to Jordan largely by refugees from the Israeli-Arab wars, Each variant thus has specific socio-political associations, q the standard (see this section below), * urban Palestinian, & rural Palestinian and g largely rural Jordanian, Leaving g aside for the moment, it appears that a bipolar prestige system is developing in urban areas, with ? and g in competition, &, on the other hand, seems to be regressive in urban Jordan, according to Abd-el Jawad’s data at least (see section 7). This emerges from Abd-el Jawad’s corpus-hased study comparing parents and children (1986: 56) ‘where in both Amman and Inbid the following mends are discernible Among & speakers there is a sharp decline in the usage of & from the parental to the children’s generation. Furthermore, no speakers of orig- inal g or ” dialects use & forms. Both original * and g speakers main- tain the forms over wo generations, ” somewhat more than g, though a significant contrast is developing between younger men and women, with women strongly favoring ’, men somewhat favoring g. ‘There is also a geographical contrast, with Amman showing a stronger ? orien tation than Irbid. © A very few dialects in the area have q as the native variety, Nablus town, for instance (Abe! Jawacl 1987: 361), and the few rural Druz communities in Jordan, © Traditionally the ¢ variant is referred to as Bedouin, though this is a. misnomer. Bedouin applies to a very specifi, nomadic lifestyle, though the majority of g speakers in Jordan are in fact sedentary farmers. Moreover, many nomads (eg. eastern Jordan and on into Saudi Arabia) have further conditioned variants of #9, such as ¢¥ and ds Ingham (1976) correlates the Bedouin-sedentary dichotomy with characteristic linguistic variants among different groups in southern Lraq and Khuzistan. He has some degree of success in this exercise, though also describes groups who do not fit into the pro- posed kinguage-lifestyle corvelation at all ‘eg, where long sedentary groups have classi- cal ‘nomaclic” features) ARABIC SOCIOLINGUISTICS. 437 It would appear that ’ vs. g have crosscutting symbolic political and social values. ’ represents Palestinian norms, though is also historically urban and hence may take on overtones of modernity, while g is orig- ally Jordanian, but also tough, slightly macho and rugged (Abd-el Jawad 1981: 176, see section 7 below). Perhaps because it has no strong constituency to represent—Palestinian interests are, as it were repre sented by ? and it never was an urban variety—& has established no we in urban Jordan. ‘Tying together the two types of variation discussed in this and i the previous section, education and ethnicity, in the course of the cor- pus-based study of Arabic a certain puzzle arose, Numerous Western studies have established that standard variants generally are prestigious in various senses: change tends to work towards them; they are the tar get of hypercorrection, they represent the norms of society at large as ‘opposed to local, communal values of the vernacular variants, ete. This may be termed the standard-vernacular model. As early as Schmidt (1974, unfortunately not available to me) a different sort of pattern became evident in Arabic. This is particularly prominent in regards to the present *g variable, In Bahrain, for instance, there are three vale ues for this variant, q ~ g (g) ~ g ~ & (back velar &}. q is the SA value, g and { largely phonologically conditioned “Arab norms, g (no variation with g) the norms of one Bahama community, and & the norm of the largest Baharna community {most ruralites). According to the standard vernacular model one would expect movement towards the standard q variant, Holes, however, encountered a more complex pattern of vari- ation. First, he found that there were some words where g always occurred. These are usually words of SA provenance (Holes 1987: 54; see section 3 above). A second set of words, termed ‘core items’, includes, those where q never replaces the local variant, and a third set shows variation between g and a local variant (see section 6 below). Most telling for the standard-vernacular model, Holes found (1987: 70) that the dominant ‘Arab tended to maintain their own native variants, both g and the marked (in the sense that it is unique to the Arab) § varie ant to a high degree, while those literate Baharna speakers who have as @ native variant, if they switched at all, switched almost categor- ically not to SA g in the core vocabulary, but to ‘Arab ¢ In short, there are two tendential movements, one (0 a pan-native variety g, one to 7. Ibrahim (1986) has given a name to this sort of bi- farcated variation, distinguishing between standard and prestige variant. n 438 JONATHAN OWENS “Standard” refers to the codified SA while ‘prestige’ describes the local norm or norms which, constitute an alternative to the standard. The prestige varieties derive from regional native Arabic varieties, in the present example the g variant. I will return to the standard-prestige dis- ction in 44 and 5 below, ‘As a last topic in this section, 1 would like to switch the geograph- ical and socio-historical orientation to a new focus. Arabs have been in the Lake Chad area since at least 1400. In present-day Borno in NE Nigeria they constitute an important minority of about 500,000. Between 20,000-50,000 of these (modern demographic statistics are impossible to come by) live in Maiduguri, Borno’s capital, where they represent the largest concentration of Arabs in Nigeri Whereas in the previous cases ethnicity was an intra-Arab affair, in the present one it is primarily an Arab-non-Arab one. ‘The dominant political group in the area are Kanuri, and Kanuri and Hausa the main lingua francas. In Owens (1995) it was argued that minority sta- tus played a decisive role in explaining the type of variation observed among Arabs in Maiduguri. The issue may be illustrated on the basis of one particularly clear variable, Maiduguri has an Arabic population which is no more than three generations deep, Generations one and two are comprised almost exclusively of immigrants from rural Nigeria, Cameroon and Chad, and from urban Ndjamena in Chad, These source areas are linguistically distinctive in various ways. One variable per- tains to the realization of the Ipl marker in the imperfect verb, In rural Nigeria it is almost categorically 2-, n-uktud “we write”, while in Chad and parts of Cameroon, where the dialect boundary appears to lie, it ne..-u, nouktub-u. Via migration, both variants have established them- selves in Maiduguri, Migration follows certain patterns, one prominent one being the tendency of Arabs to settle among kinsmen and Arabs from the same source area. One thus finds areas populated largely by immigrants from the eastern Nigerian dialect region, from Ndjamena, and so on. While the linguistic norms from the source area tend to get carried over into the newly-settled urban areas, the correspondence is less than perfect. Thus, in a sample of rural Nigerian speakers (N = 52) 99% of the speakers use exclusively the n= form. While there exist_no corpora-based statistics relating to Chadian Arabic, all grammars of the language which Tam aware of give only the n-... -u form. On the other hand, in the areas of Maiduguri, areas settled largely by Nigerian Arabs, those Arabs (N = 30 in the survey) of Nigerian ori- | | | ARABIC SOCIOLINGUISTICS 439 gin have only 82% use of the m form (18% of n-... 1), Speakers in of Chadian origin in a ‘Chadian’ neighborhood have nearly the mir ror image percentage, 17% n- 83% n-... -t. It thus appears that there is a two-way movement relative to the ancestral norms: speakers of Nigerian ancestry have inereased the non-Nigerian n-... -u form, while speakers of Chadian ancestry have increased the Nigerian n-. In contrast to the other cases examined so far, what this data shows is the lack of a clear defining norm controlling the direction of vari- ants, Speakers are aware of the existence of other variants, though none of these have normative value in the way, for instance, the g variant of #7 has in Bahrain, or the ° and g variants of *y have in urban Jordan. Moreover, in Maiduguri SA has no significant affect on spoken Arabic. It was argued in Owens (1995, also, 1998¢) that the status of Arabs and Arabic in Maiduguri inhibited the formation of a single prestige value of this and other variables. Being a communal language, Arabic is not used widely in Maiduguri in public places, a situation whieh restriets the forums where single norms might be negotiated. Furthermore, Arabs themselves are aware of their minority status, and as such con= trast themselves with other ethnic groups (vs. Kanuri, Margi etc.). It may be surmised that this confrontation with other groups deflects atten- tion away from internal differences in Arabie. It also appears to be the case that Arabic dialectal differences do not correlate with political and social groupings, a fact which further lessens the potential symbolic importance of the different Arabie variants. It may be noted that the variational stats of Arabic in Maiduguri falls outside both the standard-vernacular paradigm, and the diglossic paradigm, which will be introduced and discussed in greater detail in section 5 below. 4.3. Style, context and interpersonal variables Ferguson's original formulation of diglossia described not only an ideal dichotomy between linguistic varieties, but also an ideal socio-functional matrix into which these varieties are distributed. High varieties are used in formal situations, for writing, between strangers, for example, whereas low varieties are for informal contexts, not used in writing, and con- versation nor between friends. Just as the linguistic reality proved to be more complex than could be accounted for by a simple H-L dichotomy, so too do contextual factors correlate with linguistic usage n various ways. 440 JONATHAN OWENS Although Mitchells original conceptualization of ESA defined it primarily as stylistically-controlled variation, studies locating linguistic variation along contextually- stem + V + object pronoun > Isg perfect ~ suffixless imperfect verb © A further parameter Parkinson summuarized was whether or not the chosen vowel was grammatically correct or not, according to the prescribed SA rules "This is pethaps unwarranted extrapolation from Parkinson's data. It is saler to say °a final vowel will oceur to a greater degree in the order” rather than, “added in the order of”, 448 JONATHAN OWENS As they stand such summaries are a necessarily happenstance collec tion of observations, too few studies having been conducted to warrant models accounting for the structural variation leading from NA to SA. Nonetheless, consideration of such hierarchiealizations may find reso- nance in theoretically orientated accounts of second language learning, codeswitching (cf. 4.1 above and 8.1 below), as well as having a home within the general framework of diglossia, Ideally these hierarchies will be established not only in terms of text- based analyses, but also by perceptual tests. For example, a short study by Owens and Bani Yasin (1991) explored the hierarchicalization of features via a matched guise test. Two identical mixed tests were pre~ pared and recorded, where one had the SA variant q and the Horan ipl agreement (sec above), while the other had Horan g and the SA fg agreement pattern (e.g. ‘alaq-at sar-an “relations became” (SA g + Horan pl) vs. ‘alag-at sarat (Horan g + SA fig). Using a modified matched-guise format (see section 7 below), power and solidarity ques. tions were posed. The q variant scored higher among the ‘power’ questions’ whereas the g variant scored higher among the ‘solidarity’ questions. The differing agreement patterns appeared to have no in- fluence on the responses, supporting the conclusion that the speakers responded to the q ~ g variants to the exclusion of the agreement varia- tion, In the Middle East at least, the use of g, and perhaps q alone, appears to move the discourse to a more formal SA level, where other variants, agreement or diphthongs for example, lack such strong symbo- lic character (Gee also Parkinson 1991, discussed in section 3 above)."* There has been no convincing explanation, 10 my knowledge, as to why some features figure more prominently in variation than do others. 6. Variation and Change Before addressing the question of variation in terms of language change, the naive question may be posed, why one of either SA or NA does not simply become the unique norm in the Arabic world, or phrasing the question in terms of language maintenance, why both varieties continue © Parkinson 1991: 58) suggests that phonological variables play particularly: im- portant role in defining text level. Before appealing to global domains. plionology vs, moxphology. however, it appears necessary to define hierarchies within exch as well pronunciation of #7 vs. diphthongs, for instance: see Parkinson 1991: 40 ff tor evalua SA/NA in mixed, written text). ARABIC SOCIOLINGUISTICS 449 to exist. It is easier to explain the maintenance of SA. In the first instance, SA is structurally identical in most respects to Classical Arabic, which is the Janguage of Islam and a link to classical Arabie culture. SA is sustained through this cultural association. Secondly, SA is arguably the most prominent vehicle symbolizing Arabic unity in the modern world. In contrast to the relatively weak and disunited national states, a mod- emized SA symbolizes a uniform standard throughout the Arabic world.** The reasons for the maintenance of the NA varieties are more com plex, as they cannot be subsumed under the rubric of official policy.*" ‘To the contrary, NA (alias dialects) has no official status or recognition, in the Arabic world, Briefly, the diglossia, SA-NA dichotomy has existed at least since the eighth century, when Classical Arabie was codified by the Arabic grammarians (.e. Stbawayhi), producing a variety different from rural and urban dialects. The situation today, though more wide- spread than in the past as access to SA increases, is thus inherited. A second reason may be termed ‘mechanical compatibility’. The basic phonological and morphological structure of SA and NA are very sim- ilar. Lacking compelling reasons for switching to SA, the average Arab could well ask what the necessity is of changing one’s native variety to another, similar one. However, the key reason relates ultimately not to language structure but to general social and political motivation. Should these become impelling enough SA would doubtlessly become the spoken norm throughout the Arabic world. Lacking such motivation, however, and at present there are probably as many reasons for maintaining NA as for adopting SA, diglossia will continue to prevail ‘Tuming to the question whether the variation described in sections 3 and 4 can straightforwardly be interpreted as describing language change, a qualified answer needs to be given, The most important qualification pertains to what the parameters of variation are, whether defined by the introduction of SA material via, primarily, the speech of educated speakers, or by inter-ethnic variation. Additive change may be opposed to replacive change, with Arabic exhibiting both types to a high degree % These two linkages are partially independent variables, As Granckguillaurne (1991 51) points out, the relative weighting given to the Islamic or the modernist legitimiza- tion for the use of Standard Arabic within the modern Arab state is @ significant index for a range of atituces and policies relating to languages other than SA. © Hence maintenance of NA rarely receives the attention from political scientists and historians which Arab policy towards SA does (e.g. Holt 1996, where mainly SA is considered) 450 JONATHAN OWENS Beginning with the first, the degree of influence of SA on spoken Arabic in modern Arabic countries can hardly be understated. Its influence is, however, largely of an additive rather than replacive nature. That is, SA is integrated into spoken Arabic by adjoining it to the NA ase (see 4.1) rather than by displacing it. ‘The linguistic mechanism by which this is accomplished is via the introduction of SA lexical struc~ tures into the NA base, ie. borrowing. If the introduction of SA mate~ rial may generally be characterized as borrowing (Owens and Bani Yasin 1991), it is borrowing of a special kind because the donor vari- ety, SA, is always present among a large population of speakers, and educated Arabs with a reasonable knowledge of it may employ it for oth conceptual and stylistic purposes In the simplest case SA words, usually new concepts, are introduced in their SA guise."’ ga‘a “lecture hall”, for instance, is a word more prevalent in the Inbid region of Jordan since the opening of Yarmouk University in 1976 than before that date (see e.g. Abd-el Jawad 1981 352, Holes 1987: 49, Abd-el Jawad and Sulaiman 1990, Haeri 1991). ‘The introduced lexical structures may also bring entire blocks of rules with them, Bani Yasin and Owens (1987), for example, show that plural non-human nouns of SA provenance almost categorically require fg agreement, as in SA, as opposed to a mixed fig or natural fpl agree ment found in the native Horan Arabic of the region, Moreover, the agreement will often be imposed on words of purely NA form and meaning, yielding a mixed syntagm, with noun of SA origin, SA agree- ment, realized on a word of NA form and meaning. ‘The infiltration of SA into NA is difficult to conceptualize precisely because it proceeds along two dimensions simultaneously. One dimen- sion, at least in its initial impetus, is straightforward, the borrowing of SA lexemes encoding new ideas, described in the previous paragraph. The second, alluded to in sections 3 and 4.3, is stylistically and situa- tionally controlled and hence is always in a state of flux. In part reflecting this multiple motivation, words may appear in a number of different guises, even in the space of a single text fron a single speaker, from fairly pure SA form, co mixed forms having attributes of SA and NA. All studies which have given serious quantitative attention to the mat- {er suggest that the situation can be idealized in the following dichotomy, Sallam (1980: 79), studying the q variable (see section 5 above) being among the first 0 address the question, A given corpus will con! © Bxctuding certain SA featntes, like case endings, ARABIC SOCIOLINGUISTICS 451 words which always occur in a certain SA guise, In his data, qua “villages” always has the SA variant q for example. Other words will alternate between SA and NA, both ga! and “af “he said”, for example. Following Mitchell, for Sallam the q ~ ? alternation is sty- listically-controlled, with g being a formal variant, ’ a casual one. ‘This situation may be represented as in (2), describing the realization of the %q variable in two lexical sets, according to wo contexts Q) Formal — Casual Sec 1 + (gal ~ val? Set 2 + + {always quia) This, however, is an idealization, which rarely, if ever, admits of a cat- egorical classification of lexemes into two discrete sets, Holes, for instance, divides his Bahrain corpus into three lexical categories in respect of g occurrence (1987: 51 ff), these words which only have g (or one of its morphophonological variants), those only with g, and those varying between q ~ g. gal “say” is listed in the first category, mustagbal “future” in the second (only q}. Abdel Jawad (1981: 367 ff) has a mote pre= cise listing, giving the actual percentages of SA q or NA? ~ g~ & in his Amman corpus. Here mustagbal occurs with g in 55% of the wkens, one of the NA variants in 45%; gai occurs in 9% of the tokens as 4, 91% in the NA guises. ‘The comparative Bahrain ~ Amman data is suggestive of different explanations. It may be, for example, that in Amman a ‘nativization’ of certain words like mustagbal, originally from SA, is further along than in Bahrain. The higher percentage of q in gal in Amman may also suggest a greater degree of style shifiing there than in Bahrain, In any case, what this discussion shows is that bor- rowing and stylistic variation may dovetail in such a way that as (mainly) learned words introduced originally in formal contexts become more widespread, they loose their exclusively SA traits. At the same time, most lexemes, of whatever provenance, may be formalized by delive ing them in an SA guise. The logic of this process is such that its inte pretation in terms of linear historical development, such as for instance Labov attempts (1994: 345-6) is by its very nature impossible. ‘The second type of variation, inter-ethnic variation, represents vari- ation which may point to replacive language change. Most of the data, © Whether the g~ ’ (and SA ~ NA alternation in the same lexieal set generally) is 44 phonological or a lexical alternation, or a combination of the two, is a question necel ing wreater research; see e.g. Holes 1987 chapter 8) 452 JONATHAN OWENS discussed in 4.2 may be given a diachronic interpretation in this light It appears, for instance, that the rural Palestinian & variant of *g has not strongly established itself in Jordanian cities, and may one day dis- appear entirely. Such a development would not mean the disappear- ance of the # variant, since it continues in the rural West Bank and in Israel, though its disappearance in one area would represent a loss and regression for this one feature Arabic, with vo different patterns of variation, is unique enough that it deserves a special place in the typology of sociolinguistic varia- tion. Such a conclusion is not one followed by all: sociolinguists. Chambers, for instance, « sociolinguist with no background in Arabic, would have it that“... literary Arabie does not form part of the lin- guistic continuum in Arabic communities but is removed from it by a gap.” (1995: 142). Such simplificatory summaries may make presenta- tion of the Arabic situation easier for the outsider to grasp, though at the expense of distorting the sociolinguistic reality. Without developing the components in detail (Owens, 1998c, 1999}, it may be suggested that a typology of sociolinguistic variation is necessary to account for the differences obtaining between the West and Arabic-speaking coun- tries, In the West (perhaps elsewhere) a single, well-profiled standard (standard English, Hochdeutsch, Classical French) stands opposed to various local vernaculars, what was referred to in 4.2 as the standard vernacular dichotomy. A second position on the typology is represented by Arabic, defined by three parameters, a standard variety = SA, a local prestige variety (in Ibrahim’s 1986 sense, see 4.3, eg. the ‘Arab variety of Bahrain), and other non-prestigious local varieties (e.g. Bahama varieties). Arabic in Nigeria fits neither position, which implies that the typology will require further defining parameters, which I will not go into here (3) standard = prestige vs, vernacular (western countries) standard ys, prestige vs. vernacular (Arab counties) no standard, no prestige (Arabic in NE. Nigeria) 7. Attitudes and usage Language variation is related to a set of attitudes and value judgements pertaining to the different variants, Two approaches to studying this relationship are sometimes termed the direct and indirect. The first asks speakers directly their evaluation of different varieties, frequently in con- ARABIC SOCIOLINGUISTICS 453, junction with questions about domains of language usage, In this vein of research, Diem (1974, 92 ff), provides a good overview about Arab writers’ attitudes towards SA and NA. Examples of the direct approach are Kahtany (1997) and Dweik (1997) who asked a sample of Arab students in America about atti tudes towards and policies concerning the use of SA as opposed to NA, Kahtany’s study concentrating specifically on Damascene Arabic, Gen ally speaking they found an attitude tolerant toward NA, and aware of its broad functional spread, but also one agreeing that SA is a variety of greater prestige whose dominance should be maintained in the media and educat ‘The indircet approach is often used in conjunction with a version of a matched guise test, made famous by Lambert (e.g. 1967). In this test versions of the same text differing only in terms of the features under scrutiny, often read by the same person, are played before an audience, who answer questions about the personality of the speaker. ‘The questions are often arrayed along scalar poles of solidarity (e.g friendliness, honesty, neighborliness of speaker) and power (wealth, edu- cation, respectability of speaker). In addition, subjects are ofien asked to identify the variety being spoken, and under what circumstances (e.g. at home, in a mosque) it would be appropriate to use it, Two of the earlier studies in this second genre of Arabie sociolinguistics were car ried out with Egyptian subjects in Cairo, El-Dash and Tucker (1975) presented subjects with five language guises (Egyptian Arabic, SA, Egyptian, British and American English). Generally the SA guise received te highest scores on all counts (intelligence, likeability, religiousness, leadership) with the British English speaker () receiving the lowest, The Egyptian Arabic guise achieved his highest ranking on the likeability scale. The subjects were asked to rate the suitability of each guise for five domains, The SA. guise scored highest for school, work, radio and television, and formal speeches. Only in the home domain did the SA rank lowest of the five, with the Egyptian Arabie guise scoring highest here. This, the first study of its kind in the Arabic world, showed that Arabic speakers differentiate consistently between different varieties of Arabic according to domain and personality assignment. Herbolich (1979) was interested in defining to what extent Egyptian speakers could identify the speech of other Arab nationals, and what the associated charaeter traits of each were, For this study subjects were played tapes of Egyptian, Libyan, Saudi and Syrian speakers, plus 454 JONATHAN OWENS speakers from these countries trying to speak in an Egyptian manner. Whereas the subjects could identify the pure Egyptian with 86% aceu- racy, they were quite poor in identifying the other three. The Libyan was highest after the Egyptian, though with only 23% correct identi- fication. Despite this, the guises did receive consistent and statistically significant assignment of character traits: Egyptians were generally ranked best, followed by the Syrian guise, Saudi and Libyan, Herbo- lich farther showed that perceived identification of speaker, even if incorrect (¢.g. Saudi speaker perceived as Sudanese) may affect the judgement of personality characteristies. Saudi’s identified as Sudanese, for example, rate slightly higher than Saudis identified as Saudis. Both studies had evaluators from different educational levels, with subjects tending to make more accurate judgements about the speakers’ iden- tity the more education they had. ‘The first two studies inquired about global reactions to different lan- guage varieties, without keying in on which linguistic variables it was that served as identificatory clues to a particular variety. In this respect they are hard to integrate into the detailed, variable-based analysis of Gairene speech which characterizes the corpus-based approach. Two studies in Jordan, on the other hand, concentrated on the question of attinudes towards specific variables, Sawaie (1986) and Hussein and Ek Ali (1989), In both, matched-guise tests were constructed around the variants ¢ ~ ? ~ g ~ 4. Both studies found the SA variant q consis: tently received the most positive scores. The situation is more compli- cated with the other variables, however. Since both studies were carried out among Yarmouk university students in Irbid (Horan dia- lect region), the findings may be compared directly Generally speaking, Sawaie found that among the three NA vari- ties, ° received the most positive evaluation. The speaker of this guise, for example, was more likely to be assigned a higher professional sta~ tus, like doctor or university teacher, than were speakers of the other guises and the speaker was assigned a higher social class, At the same time, other variants did not necessarily have negative associations. g, for example, was associated with a masculine way speaking (see Abd-el Jawad 1981: 335}, and although ° was considered the most “pretty” ‘of the the NA variants), it was also considered the most “pretentious and affected”. It is interesting to note that each variety had strong geo- graphical associations. g was associated with Jordan, & with Palestine, and ’ evenly divided between Jordan and Palestine. On the whole Sawaie’s attitudinal study correlates with the corpus-based study of ARABIC SOCIOLINGUISTICS 455 Abd-el Jawad (1981, 1986). g and ’ appear as standard and prestige varieties respectively, g perhaps a covert prestige form, these the three variants which dominate in Abd-el Jawad’s corpus (see 4.2). Even more so than in Abc-el Jawad's work, ’ emerges as a geographically-neutral urban prestige variety. This happy state of correspondence is disturbed, however, by Hussein and ELAl’s study. In contrast to Sawaie’s, here the g variant emerges as the most positively-evaluated variant, while the ’ guise was ranked wakers of each variety tended t rate their own variety most positively, except the & speakers gave the g variant a higher rating. One can imagine a number of explanations for discrepancies berween the two studies: the actual respondents may have come from different backgrounds, the phrasing of the questions was different and so on Certainly the relatively positive evaluation of the & variants, variants which Abd-el Jawad’s data suggest are regressive in both Irbid and Amman (see 4.2), deserves closer scrutiny. Such discrepencies underscore the importance of conducting inte- grated studies in which the same sample is evaluated according to both self-reported information about language attitude and usage and on the basis of actual textual data, preferably in a range of speech situations. To date, no study has correlated reported language usage with observed, corpus-ased usage Thus far little has been said about Arabie in North Affica. The rea- son for this is the dominant position of French in North Affica, with the result that the majority of sociolinguistic studies in the region include French as one of the language variables. The dominance of French is such that it often overshadows the SA-NA dichotomy," so that a num- ber of studies when contrasting French-Arabie either do not explicitly distinguish between SA and NA, or concentrate on NA, as the spoken variety, alone. There exist, I believe, no corpus-based studies of the type summarized in 4.1 and 4.2 which have investigated the status of three (or four, with Berber) varieties simultaneously, French, SA and NA. As an introduction to this geographical area, 7 studies on language attitudes using both direct and indirect techniques may be briefly sum- marized. The languages involved alone, French and Arabic, and Berber as lowest. § © Tris interesting 16 chserve how sociolinguisties themicelves tend ro be dean towards questions of power in language usage. SA vs. NA figures strongly in the Eastern Mediterrancan, wheteas French-Arabie dominates in the westem, The more recent inter est in Berber ie-g. JJ9L 1997) appears to reflect Berber poswer hase, Previously socion linguistic studies in North Africa rarely included Berber, 456 JONATHAN OWENS in one of the studies give the reader a preview of the very different type of language issues which tend to occupy linguistic studies in this area. In Dichy’s (1994) ‘glossie” terms the relation between the varieties is complex. Ostensibly French and SA constitute H_ varieties, NA and Berber low. The reality is more complex, however, as will be seen presently. Bentahila (1981) collected a number of questionnaires enquiring into the domains of Arabic (usually undifferentiated between NA and SA) and French, Generally speaking, Arabic was the preferred choice in more informal domains (home, grocery store, for joking, speaking with mechanics) whereas French was preferred in formal contexts (speaking to a doctor, attending lectures, filling out a job application, most media). Bentahila similarly found that in matched guise tests using French and Moroccan NA guises, the French guises were generally given a more favorable rating. In one variant of the matched guise test, Bentahila contrasted a French and an Arabic with a codeswitehed version, both the French and Arabic guises receiving a much more positive rating Chebchoub’s (1985) study focusing on French, SA, (irregularly, NA) and mixed French/Arabic in Algiers largely replicated the results of Bentahila’s study. French has a formal profile, both NA and mixed French/Arabic being favored in familiar contexts. In a matched guise there is one notable point of difference from Bentahila’s, Whereas in Bentahila’s Moroccan study there was no significant difference between the French and (Morocean) Arabic guises for the category “patriotic” in Chebchoub’s Algiers study the SA guise achieved a significantly higher rating, perhaps a reflection of the more overtly nationalistic role of SA in Algerian politics (Grandguillaume 1983: 154), Brahimi (1993), using situated scenarios, studied attitudes among Algerian university students in Oran towards SA, NA, and French, Her findings run ac- cording to expectations: French is the preferred language of scientific discourse, modem culture, SA of religious discourse and law, and NA in the market. These tendencies were replicated for two groups of sub- jects, students in the medical faculty (Francophone) and students of Law (Arabophone}, with the former having a relatively more positive orien- tation towards French than the latter, Kidhnel’s study (1995) based nuainly on questionnaires collected at universities in Vez, Oran and Algiers, tend to complement Brahimi’s in that SA emerges as a vari- ety which has (© share its domains, even official ones, with other vari- eties, NA and French, depending on domain and city, are both often, preferred t0 SA ARABIC SOCIOLINGUISTICS 457 Brahimi (1995) adds Berber to the repertoire of North African lan- guages. Brahimi’s study centered on Berbers living in Tizi Ouzu, a Berber-dominated area SE of Algiers (in the Kabylei) and Arabs and Berbers living in Arab-dominated Oran. ‘Two findings of Brahimi were, first, that there was a marked eth- nically-defined difference between Berbers and Arabs to a range of questions about language use and policy. Whereas Algerian Arabs were highly favorable towards SA, Algerian Berbers were not, while attitudes towards Berber took precisely the opposite attitudinal values among the two groups. In terms of positive evaluation, NA and French tok a middle ground, with much more agreement between the two groups. A simple diglossie model thus fails to capture the nuanced symbolie values which each variant may take for the diflerent groups. A second finding was that Berbers living in Oran tended to have more positive values towards Arabic than those in Berber-dominated Tizi Ouzu. A study which stands apart from all others methodologically is that of Lawson-Sako and Sachdev (1996). They considered the phenome- non of convergence and divergence from the perspective of accommo- dation theory. In Sousse, Tunisia, strangers were asked in French and ‘TNA by Europeans, Tunisians and African Blacks for directions to the ain station, Overall it was found that there was a high degree of con- vergence—interpreted as answering in the same language as the ques- tioner—though the greatest degree of convergence was recorded for European questioners, whether they posed their question in T-NA. or French, For Arab questioners, a second question would often. produce a codeswitched ([-NA + French) response. This brief study—it did not 40 beyond a second tum—indicates that factors of personal identity, both of questioner and respondent, are important in language choice, not only language competence. This study adds a new ‘glossie’ factor to those summarized above. The fact that the European questioners commanded the greatest degree of convergence suggests that so long. as there is a dominant foreign group, the language of that culture (in North Africa it could one day be English), will represent a powerful alternative to SA in the H domains (see 8.1). Benrabah (1994) uses a matched guise technique to investigate the relative status of “a” variants in Oran, Unfortunately, the sociolinguis- tic status of the variants, a pharyngealized vs, non-pharyngealized a, the former said to be rural, the latter urban, is too vaguely defined (e.g. via quantitative studies) for the results to be readily interpretable to the outsider, 458 JONATHAN OWENS Finally, in a questionnaire-based study conducted outside of the Arabic world, Shorrab (1986) sketches domains of language use among Arab- American families in Buffalo, New York. Among children, Arabic usage tends to decrease with age, the home environment being the main one where it has currency. 8. Contact Arabie is spread over a huge geographical area, coming into contact with a large number of languages of various types. ‘The synchronic workings of this contact unfortunately have rarely been closely described in descriptive work." This is somewhat surprising, since substrate influence on spoken Arabic has been invoked to explain various phenomena in the history of the language, including the purported development of the Classical language into dialects (see section 2). In this section T will describe contact under two non-parallel categories, code-switching and creolization. A third category which could be included here, Arabie as fa second language, including the role of Arabic in non-Arabie Islamic societies, is too large a topic to be included." 8.1 Code-switching As mentioned in the previous section, the interests of linguists working in North Africa, Tunisia, Algeria and Morocco, have a different focus from that in Middle Eastern countries. Whereas in the Middle East SA. is the undisputed high variety, in North Affiea it is only in post- independence times that SA began achieving parity with Freneh as the language of education and official business." “The lack of interest in this Field in an interesting reflection on the sociology of Arabic linguistic academia, ‘The field of Creole studies, for instance, has been seized upon by European language departments (especially English and French) to expand the range of languages studied within thera, Creole Arabic, on the other hand, receives hardly passing mention in Arabic department curricula and Arabic linguistic journals uy which In Germany, for example. the only two studies initiated from within the cou Tam aware of on Fast African Nubi were carried out by Africanists (Heine 1982, Khamis 1994 © Rouchdy /1992) is a colleetion of articles on Arabic in America, a number focus: ing on the role of Arabic among Americans of Arab descent. I do wot referee most of the articles, as they tend to deal with matters of social history, aspects of Arabie viewed: fio a structoral perspective (e.g, borrowing), oF language teaching, © See Grandguillaume (1983) for an enki cussion of language polities in North Aftica, ARABIC. SOCIOLINGUISTICS. 459 Given its long history in North Africa, French is widely spoken and French-Arabie codeswitching prevalent. It is so ingrained in North African linguistic habits that the codeswitched variety has a name of own, ‘Frarabe’, “Aransiyya’ or ‘Franco-Arabe’. Bentahila (1981), Bentahila and Davies (1983) and Chebchoub (1985) have studied French- Arabic (NA) codeswitching in functional and structural linguistic terms. In their 1983 study Bentahila and Davies examine internal linguis- tic constraints to codeswitching, showing that many structural constraints, proposed up to that time (1983) were found in their data."” It appears that French-Arabie codeswitching is very flexible with few absolute lim- on switch positions. Even the one general constraint which they pro- pose, that there be no violation of the subeateyorization rules of either language (1983: $21) is contradicted in their own data. The adjective in dak Learga blewe “that blue paper”, for instance, lacks the definite article which a definite Arabic head noun requires on it” Chebehoub (1985) includes similar, though less detailed study of French-Arabie codeswitching in Algiers, which largely coincides with Bentahila and Davies’ findings. Atawneh (1993) describes codeswitching among three Arabie chil- dren in the USA, showing, in the manner of Bentahila and Davies, that absolute structural constraints on switch positions are very difficult to define. Eid (1992) considers the same languages in terms of language roles {language structure and competence of speakers), though her con- clusions are exploratory ‘Thus far no studies have explored in structural and interactional terms the coexistence of different diglossic codes available in North Africa, the classic diglossic H-L contrast implicit in the SA-NA dyad and the functional diglossia inherent in the presence of competing H varieties, French and SA. 8.2. Pidginization and Creolization ‘The special sociolinguistic interest of pidgin and creole phenomena lies in the social processes which give rise to pidgins and crcoles, and ensue ing processes of decreolization, all of which imply # intensive contact © Por example, that «PP acting as eomplement to a noun oF verb cannot be in different language from the head, disproved! by the Moroccan, gadviw on sille "going inta town”, {198% 314, similarly Chehchoub 1985: 168 on Algiers eodeswitehingy WAS in dak oage blew. attested in Bentahila 1981 {p. 201; see also Nortier 1904 210 on Dutch-Arabie codeswitching, 460 JONATHAN OWENS between speakers of different languages. Once formed, a creole like Nubi, a creole Arabic of East Africa, is a language as independent of its Arabic source as is Hebrew from Arabic. One of the focuses of this section will thus be socio-historical While there is evidence that simplified Arabic-based trade languages existed as long ago as 1,000 AD (Thomason and El-Gibali 1986), the only welkattested pidgin/creoles (p/c) come out of what is today Sudan, East Africa and Chad.” So far as our knowledge of these varieties go, they arose in the turbulent southern Sudan of the second half of the nineteenth century. Afier 1850 there developed camps of ivory hunters, which within twenty years had developed into slaving camps. The south- em Sudan at this time was broadly divided into a three-class society, two of them found within the camps (Mahmud 1983, Owens 1990, 1997}, At the bottom were the indigenous groups, who had, at best, fleeting contact with the camps, and at worst were the vietims of their predations. At the top was a ruling clite of mixed origin. There were many Arabs from Egypt and the northern Sudan, some European adventurers, ofien in the service of the Egyptian government which nominally ruled the Sudan, and Nubians from the Nile valley in north- em Sudan. In the middle were southern Sudanese who had either vol unteered or been forced to work in the camps. It was in the milieu of the camps that a creole Arabie arose. In the multilingual South there was no single ethnic group dominant in political or economic terms whose language could serve as a lingua franca. Arabic, the prestige lan- guage, had too small a number of native speakers, 20% of the camp population at most, to provide an adequate model for the transmission of a normal form of NA. Moreover, strained social relations between the indigenous camp residents and northerners was not conducive to the transmission of NA. What developed instead was a variety of gin/creole Arabic (see next paragraph), The decisive factor in ensuring the survival of this emenging variety was the attack in 1886 by the forces of the Mahdi on the southern Sudan, and the subsequent retreat in 1888 of Emin Pasha and many of the camp followers into Northern Uganda, Most of the followers never retumed to the Sudan, settling instead in East Affica, where their successors are known as Nubi © Versteegh’s (1984) original and overly criticized thesis that the modem dialects arose via a stage of pidginization inter alia (Owens 1980) suffers from the lack of frst- hand ‘seventh, eighth century) material detailing the pidginization of a classical Arabie~ like variety. ARABIC SOCIOLINGUISTICS 461 Completely cut off from Arabic models, their language of creole ori gin became their native language." Though lexically and phonologically Nubi is derivable from Arabic using classical comparative methods (Owens 1985, Miller 1993), it has almost nothing of Arabic morphology and differs in some respects in its syntax as well. It has lost the Semitic property of deriving grammatical forms via ablaut changes, so that although a word such as airubu “drink” may be derived historically from < *ainb-u “drink imp pI”, it appears only in one segmental form in all aspects and tenses. Verbs have no affixal person marking. ‘The language is not mutually intelligible with any form of Arabic. Nubi thus represents a case of contact producing changes to such a degree that a completely new language results. ‘This language continued to survive in the southern Sudan despite the retreat of the Egyptian government, and in recent years has established itself even as a first language in urban centers. In the Sudan it is known as Juba Arabic,"' and is mutually intelligible with Nubi, having essen- tially the same structure. In the Sudan, however, it exists alongside native Arabic varieties, and comes under its structural influence, Mahmud (1979, see also Miller 1985, to appear for alternative analysis) repre- sents this influence in terms of a creole continuum, Juba Arabic becom- ing assimilated to NA via a series of medial varieties. He outlines a process, for example, whereby Juba Arabic tense/aspeet prefixes, like bi- “future”, gi- “progressive” are replaced by the S-NA verbal prefixes jae, na-, ta- (g-airubu “prog-drinking” -» ya-srab “he is drinking”) Mahmud’s data, though suggestive, does not appear detailed enough to confirm the existence of a relatively stable continuum such as is found in many Caribbean societies (basi-, meso-, acrolect, e.g, Bickerton 1975). Miller and Abu Manga (1992) conducted a similar study in the ‘Takamul squatter settlement of Khartoum North, This is an area pop- ulated largely by non-Arab migrants from the southern and western Sudan, Besides summarizing the results of a survey of language domains, a corpus-based study outlines pretiminary hierarchies for the acquisi- tion of Sudanese NA phonological, morphological and syntactic features ™ Tosco and Owens (1993) argue that Turku, @ pidgin Arabic of Chad, was origi nally implanted in that country by the same social movements whicl produced Nubi in East Attica, "The present authior (so long as his knowledge of Nubi was fluent) always was able to converse with speakers of Juba Atabic. It is thus not clear to me on what basis the ny of SIL ‘SIL Homepage, Ftimolague 1996, the varieties. le Arabic} doubss the mutual intelli 462 JONATHAN OWENS among the non-Arab sample. In contrast to Mahmud, Miller and Abu- Manga do not use the creole continuum model. However, lacking an explicit discussion of the point, one is not certain whether the different approaches are due to different developments in Juba Arabic between the southern Sudan and the North, or due to different theoretical pref erences of the researchers, 9. Prospects Arabic sociolinguistics exists im the trivial sense that Arabic has been studied from sociolinguistic perspectives. It should come as no surprise that the collection of studies summarized here lack the feel of an organic whole, Added to the fact that sociolinguistics itself lends itself to quite diverse approaches in the study of the relation between language and society {see e.g. Figueroa 1994}, is the reality that scholars approach Arabic from the perspective of methodologies and theories developed in western academia, based largely on languages of the West. One may regret this reality (Owens, 1998b), though reality it is, OF course, it need not automatically be assumed that western theories will not be applicable in the Arabic world. In many respects the well-profiled con- trasts between different varieties of Arabic, particularly those diglossi- cally and, in some cases, ethnically defined (4.1, 4.2) find analogies with western counterparts, Standard vs, Black vernacular English, for exam- ple, symbolizes set of cultural and social contrasts different perhaps in content from contrasting varieties in the Arabic world, but not in the saliency with which they are perceived. However, there are socially-defined aspects of Arabic so different in degree from anything found in the West that they represent a type of their own. The typological significance of these was outlined in sec~ tion 6. Put pithily, where western sociolinguistics has to a great degree % Verswegh’s conclusion (1997: 193) that sociolectal varieties of H/L, diglowsic vari ties in Arabic catinot be defined until correlations with socioecononsie class are avail- able, assumes on a prin’ grounds that western, clasebased sociolinguistic parameters will provide the hest of most inportant_or in soine sense significant social parameter for dealing with diglossic variation in Arabic, Without downplaying the danger of reine venting the wheel in ignoring parallels wutside of Arabic Hingnisties, uke above conch sion represents an equally great danger of fiving Arabic diglossia into a mould it does not belong i A similar propensity to search for legitimacy on the basis of consmucts developed within western sociolinguistics is in evidence in Haetis claim (L007 227) 10 have have a change in progress in Arabic sociolinguistes, the palatal been the first 10 docit ARABIG SOCIOLINGUISTICS 463 tion in the can be used ), So too. will been established on the basis of class-based variation, va Arabic world is defined by diglosia. Of course, just as clay as a variable in explaining variation in Arabic (see 4. diglossia be available to western languages (Hudson 1994). Prototypical instances of diglossie or class-based variation, however, are represented by different languages (e.g. Arabic for diglossia, for class-based variation) Tr would be a mistake for Arabie sociolinguistics to stop at diglossia, however, as diglossia is not the only concept distinctive of Arabic socio- linguistics. At this juncture, however, one may register a certain dis- appoinunent in the failure of Arabic sociolinguistics to adequately define further positions on a prospective typological scale. Admittedly, one rea- son for this is the fact that Arabic covers sociolinguistic landscapes whose only coherency at times appears to be the almost accidental fact that the language used in each part happens to be Arabic. Nigerian Arabic, discussed in 4.2, is a case in point. It falls outside the proto- typical range of Arabic diglossia, though dialectically diverse, does not have opposing prestige norms and non-prestige varieties, is not marked by class-based differences, and so on. In North Afvica the social and symbolic values attached to languages (Berber, French, Arabic) and vari- ties of a single language (SA, NA) appear (o shift according to the perspective of the groups using them. As noted in 8.1, the intersection functional diglossia (n. 7) in North Affican societies pre- of clas sents ant interesting research perspeetive. Clearly, a truly general Arabic sociolinguistics will ultimately have to deal with Arabic in all its guises, linking what till today have been regional biases, if necessary develop- 1g models of sociolinguistic interaction for contexts which are unusual or even non-existent in the West. It is not suggested that Arabic will represent a unique type in every respect, The sociolinguistic status of Arabic in Maiduguri will very likely be similar to that of other minority languages of comparable size various areas of Affica, and polyglossic North Afvica will find paral- lels elsewhere.” Orientating Arabic sociolinguistics within a broader ization of dental stops in Cairene Arabic, What she apparendy dacuments is a possi ble correlation between social class and variation. Variation, possibly pointing to. lite suistic change, bas been documented in a number of studies, for instance Abdel Jawad cand d~ g~ “in Amoman; and Holes der alia, { ~ y in Bahrain}. In contrast to westem studies, however, sovial class is not a significant extralinguistic variable here see 42, 6) * For example. dhe multigiosic Language situation in Nairobi bousing estate described by Parkin /1977,, where Hand L. varieties change according 10 concext

You might also like