The Arabic sources appear to be unanimous about the outstanding role played by Bedouins in the development of Classical Arabic. For them, Classical Arabic is identical with the spoken language of the Bedouins. Whereas the Arabs dwelling in towns came in contact with the indigenous population, so that their language deteriorated, the Bedouins, it is claimed, preserved their speech free from contamination and continued to speak pure Classical Arabic, just as they had done in the days of the djahiliyya. Not only did a real Bedouin not understand the gibbering of the lower strata of the settled population, I he could not even constrain himself to utter any locution offending the rules of Classical Arabic. The Arabic sources bristle with anecdotes that show how much the linguists of the first Islamic centuries depended on Bedouins in ascertaining the proper rules of Classical Arabic.2
Modern scholars have realized for some time that this is not borne out by the facts as related by the same Arabic sources. These tell us expressly that considerable differences separated the tribal languages from one another and thus confirm our a priori assumptions about dialects scattered over such a wide area as the Arabian peninsula. These various dialectal features were assembled in a comprehensive paper by H. Kofler3 and independently collected again and evaluated for the West Arabian language area by C. Rabin, who applied the criteria of dialectal geography.4 These works clearly prove that the dialectal differences were so considerable that the various tribal languages must, to some extent at least, be considered as different idioms. Consequently, it cannot be maintained any longer that Classical Arabic is identical with the "Common Bedouin Language," simply because no such common Bedouin language existed, and, as is borne out both by general considerations and by the traditions transmitted by Arabic sources, the Bedouin local dialects differed from one another. Hence, there is substantial agreement among modern scholars5 that to most or all of those who employed Classical Arabic for writing poetry it was to some extent a foreign idiom which had to be acquired. We may in this connection disregard the hackneyed question whether or not Classical Arabic emerged from the dialect of a single tribe. For our purpose it is enough to state that as early as the second half of the sixth century C.E., Classical Arabic showed every sign of being a common tribal language, being open to (i.e., freely borrowing from) the various Bedouin dialects in lexicography, and, as it seems, even in phonology, morphology, and syntax. Thus, it necessarily became something quite different from the everyday language of one tribe. Moreover, the elevated nature of poetic language and the subject matter of pre-Islamic poetry inevitably resulted in a language rather different from everyday speech. Typologically, however, Classical Arabic and the various ancient Arabic dialects were closely akin, being all of them languages of the synthetic type, that is, tending to express several notions by the same word, so that the need to change over from the dialects to Classical Arabic caused no real difficulties.
It is somewhat surprising that although modern research almost unanimously agrees about the differences between Classical Arabic and the individual Bedouin dialects before Islam, scholars continue to accept stories that presuppose a common spoken Bedouin language identical with Classical Arabic in the first Islamic centuries. Rabin, it is true, has stated his disbelief in clear terms,6 but generally European scholars accept such stories. Even the most important recent publication on the development of Classical Arabic, J. Fuck's Arabiya,7 wholeheartedly accepts the thesis of a common Bedouin language, for all practical purposes identical with Classical Arabic. Being well aware of the dialectal differences between the Arab tribes before Mohammad, he arrives at the ingenious solution that this diversity was blurred during military expeditions and by the common life in the camps. Here the various Bedouin dialects intermin gled and a common Bedouin language arose, which in the first centuries of Islam constituted the basis of Classical Arabic. Thus, though in the preIslamic centuries dialectal differences had occurred and no common spoken Bedouin language had existed, a common spoken Bedouin language emerged from the military camps. Consequently, it is suggested, we have no reason to doubt the reliability of stories depicting the dependence of the grammarians on the Bedouins, who, according to this view, were the arbiters in linguistic questions.
Despite its ingenuity, however, Fuck's solution seems to us contrary both to the general trend of linguistic development and the express statements of our sources. It is clear that the centrifugal forces exceeded the centripetal. There was, no doubt, some general leveling of the dialects in the military camps, but this was only ephemeral. New dialects arose everywhere, the determining factor in each instance being the ancient dialect that happened to be prevalent.8
We have to assume that the great conquests and migrations affected the ancient Arabic dialects, but they resulted only in a variety of dialects and did not produce any common spoken Bedouin language. This is attested by the contemporary sources: Fuck himself quotes Hamdani's description of the complicated linguistic situation in South Arabic-already, it is true, in the tenth century-but we possess also an express statement of al-Djahiz9 that in every province the dialect of the tribe that had settled there prevailed. According to al-Djahiz, this accounted for the difference between the dialects of Kufa, Basra, Syria, and Egypt. Furthermore, even the great variety of traditions about the most correct tribal language assume that the dialects were different from one another. So our contention that no common Bedouin language existed either before or after the Islam is corroborated both by general considerations and express attestation.10
Since no common Bedouin language existed, the stories in which every ordinary Bedouin poses as an expert in every fine shade of Classical Arabic must be considered as apocryphal. But how, then, did these apocryphal stories arise? There must have been a background favorable for them. Rabin, in our opinion, goes too far in maintaining that these stories were only "justified by the rich speech of the Bedouins and his natural rhetorical ability and by the fact that a tradition of Classical Arabic poetry still continued among the tribes for some centuries, as proved by the Diwan of Hudhail." Of these points, only the last is of real importance. It seems evident that the tradition of Classical Arabic poetry continued among the Bedouins largely because their conditions of life, in contradistinction to those of the settled population, had not changed, and still, as before Islam, constituted the background to Bedouin poetry. Another important point, adduced by Rabin, is that the native assistants of the schools were probably ruwat. 1 1 These ruwdt, we may add, were experts in Classical Arabic poetry, as they continued to live among the same conditions as their ancestors and, therefore, were inclined to write in the same way. But besides these sociological reasons there was also a linguistic consideration that not only facilitated the survival of Classical Arabic poetry among the Bedouins, but, in our opinion, also contributed to the emergence of these apocryphal stories. This was the dichotomy of the Islamic Arabic dialects, divided into Bedouin and urban vernaculars. In an analysis of early Islamic papyri, the author has attempted12 to prove that immediately after the great conquests the dialects of the settled population, which came in contact with foreign elements, became contaminated, so that vernaculars arose that may already be justly considered Middle Arabic dialects. These Middle Arabic dialects, though separated from one another by many dialectal differences, emerging, developing, and disappearing again in the way envisaged by the wave hypothesis, nevertheless exhibit a certain unity when compared with Classical Arabic and the Bedouin dialects. If we may reduce the different and intricate features that typify these dialects to a common denominator, the most conspicuous feature that characterizes them is that they diverge from the synthetic type, that is, they do not tend any more, like Classical Arabic and Bedouin dialects, to express several notions by one word, and instead approximate to the analytic type, generally expressing each notion by a separate word. The most obvious sign of this phenomenon is the disappearance of the cases and moods. Because of the chasm between Middle Arabic dialects and Classical Arabic, the urban speakers had to overcome considerable difficulties when they tried to use Classical Arabic, whereas even ordinary Bedouins, speaking, as in the Djdhiliyya, synthetic dialects closely akin to Classical Arabic, could do so relatively easily and were less apt than the urban population to make mistakes. It was therefore much easier for a rawi of Bedouin stock to transmit Classical Arab poetry. Moreover, even an ordinary Bedouin, speaking his own dialect, may have appeared to speakers of Middle Arabic vernaculars of the lower strata of the town population to be speaking some kind of Classical Arabic, since he used case endings, the most conspicuous outward sign of the literary language. Against this background, the emergence of stories extolling the linguistic faculties of Bedouins becomes quite understandable. Sometimes we may assume that the heroes of these stories were not ordinary Bedouins, but ruwat, referred to, as it seems, by expressions like 'arabun turda `arabiyyatuhum ("Bedouins whose Arabic is agreeable"), al-`arabu al-mawthuqu bihim ("reliable Bedouins"), fusaha' u-l-'arabi ("correctly speaking Bedouins"),13 al-`arabu al fusaha'u al-`ugala'u ("correctly speaking and intelligent Bedouins").14 Other stories, however, illustrating the linguistic faculties of ordinary Bedouins, must be considered apocryphal, arising against the background of the somewhat paradoxical Bedouin superiority in linguistic matters, which depended on Bedouin ruwat and the kindred character of Bedouin dialects and Classical Arabic. The Bedouins were also extolled, as Rabin his pointed out, as a part of the general idealization of early Islamic society and in accordance with the romantic hankering after the primitive in other urban societies, and, we may add, sometimes also from a taste for paradox, since the superiority of a common Bedouin to the refined citizen was not without a paradoxical touch. In some cases, too, Bedouin boasting has to be taken into consideration. Only in one field does the early Arab philologist seem to have relied on the ordinary Bedouin, that is, in lexicography. Apparently, in the early period the Arab philologists excelled each other in collecting every lexical detail from every source available, that is to say, Classical Arabic remained at this time lexicographically "open" to the Bedouin dialects, just as it had done before Islam, for this fitted well the Arab flair for rariora (nawadir).15 On the other hand, the earliest philologists already seem to have restricted the range of "correct" orthographical, morphological, and syntactical usage, apparently adjusting the sources transmitted by them to the patterns they demanded. The fact that Koranic orthography was revised in accordance with the rules of Classical Arabic and that features like the omission of the glottal stop were excised,16 presupposes a language with fixed rules already firmly established. The result is that Classical Arabic, as transmitted by the early philologists, is rich in vocabulary, but much more restricted in orthography, morphology, and syntax than one would have expected of a language "open" to a great number of tribal dialects.
We thus acquire new criteria for examining the reliability of stories about the linguistic superiority of the Bedouins. Some of these stories are, in our opinion, historical, since they refer to Bedouins who acted as ruwat and were therefore masters of Classical Arabic. Sometimes even ordinary Bedouins were questioned on matters of lexicography. On the other hand, in matters of phonology, morphology, and syntax the ancient philologists were much stricter and appear to have restricted the old usages more and more; it is therefore unlikely that they relied on ordinary Bedouins in these fields. It must always be borne in mind that even the ruwat did not generally speak Classical Arabic as their natural language. Thus, all the stories about Bedouins who simply could not speak anything but "correct" Arabic are to be regarded as apocryphal.
In the light of these criteria, it will be worthwhile to analyze an account of one of the most famous linguistic disputes, the so-called mas'ala az-zunburiyya. This story exists in several different versions. According to all of them,17 the dispute took place in Baghdad between the famous grammarians al-Kisa'i and Sibawayhi about the use of the accusative in two expressions. In the first, the problem was whether the continuation of the sentence kuntu azunnu anna -l-'agraba ashaddu las'atan mina -z-zunburi ("I thought that the scorpion stung more severely than the wasp" [az-zunbur, whence the name of the dispute, "the question concerning the wasp"])-should be only fa'idha huwa hiya or also fa'idha huwa 'iyyaha ("And behold, the one is [like] the other"). The other problem concerned the sentence kharajtu fa'idha 'Abdo -llahi -l-ga'imu/a ("I went out, and behold, there was A., who was standing"), and whether it was correct to use al-ga'ima as well as al-ga'imu. Whereas al-Kisa'i considered the use of both nominative and accusative in each case as correct, Sibawayhi would only approve of the use of the nominative. It was agreed to refer the question to Bedouins,18 who decided in favor of al-Kisa'i. There were rumors that this decision had been brought about by bribery.
The grammatical issue in the dispute was thoroughly analyzed by A. Fischer. In the wake of H. L. Fleischer, Fischer demonstrated that the use of 'iyya in a phrase like fa' idha huwa 'iyyahd is attested and may be regarded as correct. The second accusative, however, in a phrase like fa'idha `Abdu-llahi al-ga'ima,19 is, according to Fischer, rather harsh, being used on the assumption of the Kufic grammarians that a circumstantial phrase may also be definite.
The historicity of this story is generally asserted. Fischer, while taking account of the different versions, nevertheless inclines to regard the dispute as historical. Fuck, who assumes the existence of a common Bedouin language that provided the basis of later Classical Arabic, and who therefore accepts the traditional view of the unlimited linguistic superiority of the Bedouins, adduces our story as proof that Bedouins acted as arbiters in linguistic questions.20 Blachere apparently regards even the rumors about the bribe as historical, accepting them as proof of the low moral standards prevailing among Bedouins.21 However, this last point, at least, seems to be apocryphal;22 this part of the story is always attached to the main body rather as an afterthought. It is generally only connected by wayugalu ("and it is said"), and the like, as, for example, in the Mughni-l-labib of Ibn Hisham.23 Very interesting is the passage quoted by Ibn aPAnbari24 in the name of Basran authorities: 'amma ma rawauhu `ani -l-`arabi ... famina -sh-shadhdhi -lladha la yu`ba'u bihi ... `ala 'annahu qad ruwiya 'annahum 'u`tu `ala mutaba`ati -l-Kisa'iyyi dju'lan fala takunu ft i qaulihim hudjdjatun litatarrugi -t-tuhmati fi-l- muwa fagati ("And the traditions of the Bedouins [permitting the use of the accusative in the above-mentioned expressions] ... are so isolated that one must not take them into account. And there is a tradition that they received a bribe in order to agree to the opinion of al-Kisa'i, and if so, there is no proof to be adduced from their judgment, as it may be mistrusted.") This passage deserves special attention, as it proves that even in Basran circles the tradition concerning the bribe was not well founded. It is therefore not surprising that in some versions the bribe is omitted altogether, as by Suyuti. In other sources, for instance one source in the Mughni -1-labib and especially Ibn Khallikan,25 the traditions concerning the bribing of the Bedouins are apocryphal when they depict the Bedouins as unable to utter an incorrect sentence and therefore forced to say only "al-Kisa'1 is right," instead of repeating the "incorrect" phrases approved by him. So one may claim that the acceptance of bribes was quite common among the ruwat and, as Blachere himself perhaps intended to do, regard this part of our story as typical rather than historical.
Even the gist of our story, concerning the role of the Bedouins, seems to us unhistorical. It is so, no doubt, according to the version of Ibn Khal- likan, which maintains that a single Bedouin acted as umpire. Certainly, Bedouin dialects were different from one another so that an expression like fa'idha huwa 'iyydha may have been familiar in one, but relatively unknown in another vernacular. Therefore, to apply to an unknown Bedouin, whose only qualification was his pure Bedouin speech, is rather like casting lots, since by chance this man may or may not have known the construction in question. Men like al-Kisa'i and Sibawayhi, no doubt, were well aware of this fact and it is unlikely that they would have agreed to such a way of deciding the dispute. Even the tradition that several Bedouins acted as arbiters is not well founded. According to the tradition preserved by Ibn al-Anbari,26 it seems that all the Bedouins, of whom four are known by name, agreed with al-Kisa'i, though it is not probable that Sibawayhi did not know a turn of phrase known to all or almost all the Bedouins. The tradition in Ta'rikh Baghdad,27 the smoothest and so, in our opinion, the latest version, relates that some of the Bedouins supported Sibawayhi, but eventually the majority decided in favor of al-Kisa'i. This may have worked well, theoretically, in the case of fa'idha huwa 'iyyahd, which is attested in Classical and Middle Arabic, but this version fails to explain how the majority of the Bedouins decided in favor of fa'idha 'Abdu -lldhi -l-ga'ima, which is described by Fischer as rather harsh and based only upon the theory of the Kufan grammarians that a circumstantial phrase may be definite as well. Could it, then, be imagined that the majority of the Bedouins would have supported the possibility of using such a construction, which, except in rhyme, could not be demanded even by darurat ash-shi'r? It seems, therefore, that the whole story that the mas'ala az-zunburiyya was decided by Bedouins is apocryphal.
We have accordingly reached the conclusion that stories enhancing the role of Bedouins as arbiters in linguistic questions must be taken with a grain of salt. It was the dichotomy of the Arabic language as it emerged after the great conquests that produced these stories. On the one hand, there existed the Middle Arabic dialects of the urban population, analytic in type and typologically closely akin to the modern Arabic dialects, and on the other hand, the local Bedouin tongues, which had remained unchanged in their main features, and which were thus still close to Classical Arabic so that their speakers could readily change to speaking Classical Arabic. Since these dialects preserved, in contradistinction to the urban vernaculars, the most conspicuous outward sign of the synthetic trend, the case endings, townsmen listening to them might have been given the superficial impression that they were hearing Classical Arabic. If we also remember the fact that Classical Arabic poetry continued among the Bedouins for both linguistic and sociological reasons, the general idealization of early Islamic society, the romantic hankering after the primitive, the delight taken in the paradox that the Bedouins excelled the cultivated urban population, and also Bedouin boasting, we shall understand how these stories developed. When examining them, one has to bear in mind the basic fact that Bedouin dialects, however close to each other and to Classical Arabic, differed from one another and from Classical Arabic in many respects, so that unknown Bedouins, as distinguished from ruwat, could not be asked questions concerning morphology and syntax, whereas in the field of lexicography Classical Arabic remained open to Bedouin speech for a long while. The linguistic situation of the first centuries of the Islam, thus revealed to us, is complicated, as Middle Arabic and Bedouin dialects, which had developed along complex lines, mingled, absorbed each other and often converged, whereas Classical Arabic, typologically akin to the Bedouin dialects, served as the literary language.
1. See Djahiz, quoted by J. Fuck, 'Arabiya (Berlin, 1950), p. 66 n. 8.
2. See, for example, R. Blachere, "Les savants iraquiens aux IIe-IVesiecles de 1'Hegire," Melanges ... W. Marcais (Paris, 1950), pp. 37-48, and Fuck, 'Ara- biya, pp. 29 ff., 66 ff.
3. "Reste altarabischer Dialekte," WZKM 47-49 (1940-42).
4. In his important work, Ancient West-Arabian (London, 1951).
5. See Rabin, ibid., p. 17; Studia Islamica 4 (1955): 19 ff.; El, 2d ed., vol. 1, p. 565a.
6. Rabin, Ancient West-Arabian, pp. 17 ff.; it seems to us that he went too far in his skepticism.
7. Fuck, 'Arabiya.
8. See W. Fischer, Die demonstrativen Bildungen der neuarbischen Dialekte ('s-Gravenhage, 1959), p. vii; see also the balanced judgement of C. Pellat, Le milieu basrien et la formation Gahiz (Paris, 1953), p. 125, on the formative factors of the dialect of Basra. Though allowing for the general leveling influence of the koine in the use of the troops in the military expeditions, he rightly takes into consideration the dialects of the principal tribes of Basra in the formation of a special dialect.
9. Ed. Sandubi, 1926/7, I, 33, quoted by Fuck, 'Arabiya, p. 65, and Pellat, Le milieu basrien et la formation Gdhiz, p. 125, where the page quoted is mistaken.
10. For similar reasons Ferguson's (Language 35 : 616 ff.) contention that the modern Arabic dialects emerged from a koine, has to be rejected. In a Hebrew paper, published in Tarbiz 30 (1961), the author has dealt briefly (pp. 133-35) with the features claimed by Ferguson to represent this koine. A special paper devoted to this subject is still unpublished.
11. Blachere, "Les savants iraquiens aux II`-IV` siecles de 1'Hegire," p. 39, maintains that most of the Bedouins mentioned in the sources were also poets.
12. In his paper "The Importance of Middle Arabic Dialects for the History of Arabic," in Scripta Hierosolymitana of the School of Oriental Studies (Jerusalem, 1961).
13. Used by Sibawayhi, see Fuck, `Arabiya, pp., 29-30.
14. Used by Djahiz, see ibid., p. 66.
15. Cf., for example, the story quoted ibid., p. 69 n. 1, which is, in our opinion, applicable to such rariora.
16. For details see C. Vollers, Volkssprache and Schriftsprache im alten Arabien (Strasburg, 1906), without accepting his theory that variants in the Koran necessarily reflect ancient features. See also Rabin, Ancient West-Arabian, p. 9.
17. Deviations from this, it seems, have to be regarded as mere mistakes, see Fischer, Die demonstrativen Bildungen der neuarbischen Dialekte. See Fischer's account in A Volume of Oriental Studies presented to E. G. Browne (Cambridge, 1922), pp. 150 ff.; see also Fuck, 'Arabiya, p. 30 n. 11.
18. See below on these variae lectiones.
19. Fischer, A Volume of Oriental Studies, p. 155, rightly rejects the view that Sibawayhi disapproved even of a phrase like fa'idha `Abdu -lldhi ga'iman, ga'iman being indefinite.
20. Fuck, `Arabiya, p. 30.
21. Blachere, "Les savants iraquiens aux II'-IV, siecles de l'Hegire," p. 48.
22. We do not even deal here with a source like Ta'rikh Baghdad, which, being in favor of al-Kisa'i, of course rejects the mere possibility of bribery.
23. Quoted by de Sacy, Anthol. gram. arab., pp. 199-201.
24. Ibn al-'Anbari, Die grammatischen Streitfragen der Basrer and Kufer, ed. G. Weil (Leiden, 1913), p. 294.
25. Ed. Bulaq (1299 A.H.), I, 487.
26. Apparently, Suyuti's version, and even that of Ibn Hisham, Mughni -l- labib, are abridgements of Ibn al-'Anbari, though it may be claimed that Ibn al-'Anbari's version, which contains so much detail, is complementary to the shorter tradition, as preserved by Suyuti and Ibn Hisham.
27. Khatib (Cairo, 1349 A.H.), XII, 105.