Our knowledge of the structure of the Arabic language is only to a small extent due to modern research: For most of it we have to thank the Moslem medieval philologists of the three centuries preceding az-Zamakhshari (1075-1143 c.E.). The grammar most widely used today, W. Wright's Arabic Grammar, goes back to C. P. Caspari's Grammatica arabica of 1848, which was based principally on Zamakhshari, with some slight modifications due to the observations of H. L. Fleischer. In twice revising the work, Wright used extensively both the Muslim philologists and the work of European scholars based upon them. The revision by R. Smith and M. J. de Goeje for the third edition left the framework intact, so that we may without exaggeration say that the university student of our days is essentially offered the same course in Arabic grammar as the student of a late Abbasid madrasa.
The treatises on Arabic syntax by C. Brockelmann, I H. Reckendorf,2 and R. Blachere3 depart a good deal farther from the medieval system, without, however, breaking with it. If one considers how radically modern linguistics has abandoned the traditional approach to the grammar of European languages, we cannot fail to be struck by the extent to which the categories established by Sibawaihi and his predecessors in the ninth century C.E. are still felt to be serviceable.
In fact, upon closer examination it appears that the result of the exten sive modern research has been mainly to show certain deviations from the norms set up by the Arab grammarians, grouped around the traditional rules somewhat like the deviations around a statistical curve.
This is not accidental, nor is it entirely due to the thoroughness and acumen of the Arab philologists. It is due to the prominent place the study of the works of these philologists held in the education of those who wrote Arabic, combined with the constant reading of those very texts upon which the philologists had based their analysis: the Koran and the ancient poetry. At least since the end of the Umayyad period, Classical Arabic was not a spoken language, and like all purely literary idioms, naturally conservative, but the place occupied in its acquisition by systematic grammar is, as far as I know, unique among languages. In modern linguistics the question has been much debated by which psychological processes the langue, the system of language which lies behind our individual speech-acts, or langage, functions within our minds. With regard to Arabic there can be little doubt that the langue consisted principally of an articulate set of scientific statements, inculcated in one's youth. Reckendorf observed4 that progress in the study of Arabic syntax was unlikely as long as it produced only average views ("Durchschnittsbilder"), not studies of individual periods or authors. The fact is that the average view is identical with the philologists' system, while the detailed studies provide systems that will always largely cover it, but where certain sets of deviations may in fact give the whole a markedly different structure. Since, however, the langue remains the same throughout, and does not allow any such deviations to become part of itself, a true diachronic treatment of Arabic grammar appears to be an impossibility. Vocabulary and style, on the other hand, though also extensively studied by the philologists, were not fixed by them with the same rigidity. A continuous development is therefore observable, and one hopes that the sound foundations laid for this study in J. Fuck's Arabiya5 will lead to a rapid development of research in this field.
Where structure is concerned, research can hope to reach results independent of the Arab philologists only in the study of that period of the language before their influence became decisive, a period which is, of course, also of particular interest as being the formative stage of the literary idiom. At first sight this seems a manageable task, since we possess literary remains of that period (sixth-eighth century C.E.) that are extensive enough to allow us to examine all aspects of the language and yet not too large to permit a total examination. As soon, however, as we look more closely into these sources, the question arises: What is it that we are investigating?
The material falls into four groups:
(1) Pre-Islamic and early Islamic poetry. With regard to the former, doubts have been raised as to authenticity, and we must, even if we consider the corpus as a whole authentic, reckon with the possibility of any individual verse or poem being a later forgery. Both pre-Islamic and early Islamic poems have been revised by editors, as can be seen not only from the extensive variants, but also from the not infrequent cases where verses are quoted by grammarians for some linguistic oddity, while on looking up the Diwan we find the same line slightly reshaped so that the oddity is eliminated. Nevertheless, we possess here a first-class source for the study of the pre-Islamic language.
(2) The Koran. Here again, variants affecting grammar are frequent, and a certain amount of working-over by philologists is admitted, for example, in the introduction of the hamza-sign into an orthography representing a pronunciation that had eliminated the hamza. On the other hand, variants affect only circumscribed aspects, as the consonant skeleton has, at least since the Othmanic revision in ca. 650 C.E., been carefully guarded from alteration.
(3) The Traditions (hadith). These have been extensively used as a source for syntactic phenomena by Reckendorf, and recently by Bloch,6 who employed mainly examples drawn from this category to represent prose as opposed to poetry. Here the problem of authenticity, raised first by Goldziher, is nowadays generally answered in the negative.? Although even on these modern assumptions many traditions go back to the beginning of the second Islamic century, that is, before the development of philology, an investigation by an Oxford research student, J. L. Pollardwhose untimely death is a great loss to our studies-shows that in the form in which we get them in Bukhari and Muslim, they are considerably changed by the introduction of archaizing and pseudodialectal elements, some taken from poetry, which are still absent from the older versions of the same traditions.
(4) A small number of first-century papyri and documents handed down in works on history. These include treaties and letters said to have issued from the secretariat of the Prophet8 which, if genuine, would be extremely valuable evidence for the language used in writing at the time the Koran was revealed.
Medieval Muslim writers were generally agreed on two points: (1) that the language in which the poems were composed was identical with the spoken language of the bedouins of central and eastern Arabia; and (2) that the language of the Koran was the spoken language of the Prophet, that is, the dialect of Quraish. Since they also held that the language of the Koran was essentially the same as that of the poems, and that it represents Arabic at its best and purest, some9 drew the conclusion that the dialect of Quraish was the most correct of all Arabic dialects. At the same time, the philologists did not obscure the fact that considerable differences were known to have existed between the different dialects; on the contrary, they collected them assiduously, so much so that it has been possible to achieve at least a partial reconstruction of those dialects.
In modern times these statements have been felt to be incompatible, though by no means by all scholars.10 In some cases statement 1 or 2 was accepted at its face value and the other rejected. Thus Ta ha Husain 11 drew from his acceptance of the identity of Classical Arabic and the dialect of Quraish the conclusion that all pre-Islamic Arabic poetry, except that attributed to Hijazi poets, must be forged. K. Vollers,12 on the other hand, accepted the view that the bedouins of Nejd and Yamama spoke a form of Classical Arabic, and concluded that the Koran could not have been conceived in this language, but was only made to conform with it by an extensive process of revision. Its original form Vollers sought to find in the noncanonical readings, which provided a composite picture of a "vulgar tongue" that structurally belonged to the colloquials, not the old dialects, and in particular had lost the case endings (i'rdb).
It is interesting to note that both Vollers and Taha Husain include in their treatment of the question lengthy chapters on the ancient dialects of Arabia, for it is the relation of the classical language to the dialects that really provides the key to the question.
I believe J. G. Wetzstein was the first to claim that Classical Arabic was not the spoken language of the poets who used it for their poetry. This view has been accepted by practically all more recent European writers who discussed the matter at all, whether they believed, with Wetzstein, the classical language to be different from all ancient dialects, or to be based on one or several actual dialects.13 In recent years it has become usual to call it the "poetic koine"-not an entirely happy term, since the Greek koine was, after all, a spoken language, and Classical Arabic, on this view, resembles more closely the status of Homeric Greek. This language is discussed at length in R. Blachere's Histoire de la litterature arabe,14 which also adduces numerous examples of such separate poetic idioms in other societies.15
Apparently independently, H. Fleisch,16 R. Blachere,17 and C. Rabin18 arrived in the forties at the conclusion that the language of the Koran, far from being pure Meccan either subsequently revised (Vollers) or slightly adapted to the poetic idiom,19 was none other than the poetic koine.20 The deviations from the usage of the poems were seen to be due to unconscious backsliding into the Meccan dialect. Some of them may, as Fuck has pointed out,21 be explained as due to the novelty and difficulty of its thought, as well as to the fact that it was perhaps the first attempt to write Arabic prose. Indeed, the latter aspect is put forward with much force, and with apposite parallels from the history of Hebrew, by H. Birkeland,22 whose discussion, though not touching upon the question of idiom, lends much support to the theory just mentioned.
As against this, P. Kahle23 about the same time went in for a spirited revival of Vollers's theory. Noldeke24 had adduced against it that, if the Prophet and his contemporaries had recited the Koran without i1rab, the tradition of it would not have been lost without a trace. Kahle-whose merit in Hebrew studies is to have led research from late and harmonizing sources back to the genuine old traditions-claims that this impression was due to Noldeke's ignorance of certain old sources. He adduces numerous traditions from a tajwid work of about 400 A.H.25-many of them traceable in earlier collections-exhorting people to read the Koran with i`rab, and statements in a fragment by al-Farra', a Kufan grammarian who died in 207 A.H., to the same effect. Thus he comes to the quite correct conclusion that in the second Islamic century the Koran was frequently read without i`rab. It is less easy to follow him in the further conclusion that this proves the Koran to have been recited from the beginning in this manner, and the case endings to have been introduced only by Koran readers who had studied Classical Arabic from poetry and by contact with Bedouin tribes.
The tenor of the traditions quoted, which promise heavenly rewards for reading the Koran with full or even partial i'rab,26 shows clearly that private recitation by the uneducated is intended, not that by trained readers. The opposite of i'rab is called Jahn; hence, we may conclude that the injunctions do not necessarily refer to complete omission of the case endings, but to their wrong use. We have numerous anecdotes proving that even noble Arabs frequently erred in this matter, and that it was the fear of blasphemous meanings by misplaced case endings that caused the insistence on correct i`rab as much as the desire to have the Koran recited beautifully. Indeed, it would matter little if these traditions, instead of being probably invented in the second century, really went back to the time of the Prophet, for there must have been many among his followers whose divergent linguistic background caused them to commit solecisms.
It is even possible that the habit of reading without vowel endings is old. As is well known, the case endings are omitted in Arabic spelling, which writes all words as if they stood in absolute initial and pausal position. This, of course, is due to the habit of dictating slowly, with automatic pausal pronunciation. But if writing was slow, reading of the Kufic polyphonous script cannot have been fast, and the same adaptations may well have been made. While they did not matter in private letters or books, they were naturally discouraged in reading Koran or poetry. In ordinary prose, sentences were generally turned in such a way that no misunderstanding could arise by the omission of the case endings, which thus became something of a luxury.27 Poetry, on the other hand, could permit itself certain types of tmesis ("Sperrung") that depended for their effect entirely upon the presence of case-signs.28 The Koran, too, contains quite a number of phrases that do not make sense unless they were conceived with case-vowels.
The insistence on the presence or absence of case endings ignores the fact that the dialect of Quraish was not simply Classical Arabic minus the i'rdb. It is not possible today to turn a piece of colloquial Arabic into literary by adding case endings. If anything, the dialect of Quraish must have been more unlike the classical than the present-day colloquials, which, after all, are derived from Classical Arabic or from a Vulgarara- bisch closely related to it. Had the Koran been composed in either the dialect of Quraish or in a "vulgar tongue," no amount of revision without altering the consonant outlines could have made it as similar to classical as it is. One need only consider the havoc the one consistent Quraishism, the omission of hamza, has played with the spelling. If, however, the language of the Koran made concessions to the literary koine, the 'Arabiyya, then it must needs have accepted also the case endings, that feature which was felt to be so essential that it was called by the same word as the use of that language itself, i'rdb. Fiick29 may well be right in seeing in the words Sura XVI.103/5 wa-hadhd lisdnun 'arabiyyun mubinun evidence that Muhammad himself was conscious of using the Bedouin 'Arabiyya, since elsewhere in the Koran `arab means "Bedouins."
A question that strangely enough no one seems yet to have asked is this: What reasons caused Muhammad to address his fellow townsmen in a language that originated, and was at the time used, for narrowly circumscribed purposes in Bedouin society, and that mainly in regions fairly remote from Mecca? It has, of course, been long recognized that the acceptance of a standard language has nothing to do with the intrinsic merits of the dialect chosen, but is mainly determined by the social prestige of the group from which the dialect emanates, or, in some cases, by spiritual forces using the dialect for their expression. The biography of the Prophet suggests in some of its episodes that the Quraish valued their genealogical connections with Bedouin tribes, and were in any case interested in attracting Bedouins to their fairs and religious ceremonies. It is possible, therefore, that the self-valuation of the nomad aristocracy was accepted by the townspeople to such an extent that they adopted also the language and the poetry that were the cultural badge of honor in that society. Muhammad's attitude to Hassan b. Thabit and Kalb b. Zuhair rather supports this, though we must remember that both belong to the Medinean period, when the Prophet was interested in extending his sway over Bedouin tribes. Another possibility is that Christian missionaries, starting from Hira, had picked upon the language of the poets, so highly valued at that court, as the vehicle for carrying the Gospel through Arabia. It would be a natural choice, because it was a dialect already widely understood and respected. If indeed these missionaries used a pre-Islamic translation of the Gospels into Classical Arabic30-and they must have had some written material-then the mere existence of a literature written in it would have lent it immense prestige and encouraged those who, like the Quraish merchants, recognized the usefulness of writing to employ the same language. Besides, there can be no doubt of the presence of some Christians at Mecca. The documents handed down as Muhammad's letters and treaties may well have undergone some alterations, but it seems that they provide fairly strong evidence that the use of Classical Arabic in documents was already established in Mecca at the time. However strange the materials on which the Companions had written down the revelations,31 the fact that they were written down shows that here was a society where people already placed more reliance on writing than on their memory, and that the language of the revelations was a recognized written idiom with a more or less established spelling. We may speak of a Meccan "office" language, which was there to be used for the new literary purposes.
This theory would also explain how Muhammad could so violently reject the poets while at the same time using their language: If he received the poetic koine, so to speak, at one remove, he might not have been conscious of the connection. Finally, there may be some kernel of truth in the assertion of the superiority of the language of Quraish, and that it contained all that was best in the other dialects, if these statements refer to the written "dialect," not to their spoken language. After all, a great deal of what we are told about other tribal dialects refers to the special nuances of that tribe's literary usage rather than to their spoken dialect. If this theory is accepted, it would also affect our view of the Hijazi features in the style of the Koran. Many of them might not be due at all to the Prophet's imperfect command of the `Arabiyya, but might have been taken over by him as recognized features of a local linguistic tradition.
I would hasten to add that this suggestion still leaves a large number of loose threads. The theory of Vollers and Kahle, on the other hand, seems to me to founder completely on this very problem. If both Muhammad and the Meccan aristocracy were using their own colloquial exclusively, and if the semi-Bedouin inhabitants of Medina saw nothing wrong in divine revelations being delivered in a "vulgar tongue," what were the reasons for accepting the Bedouin language as an absolute authority in the first and second Islamic centuries, when that Meccan aristocracy was all-powerful and the star of the Bedouins in the descendant? If in the century before Muhammad we are entitled to surmise the working of social forces of which we have no clear record, we have no right to do so in a period fully in the light of history. Unless the prestige of the koine had been securely established before the conquests, I cannot see any way of it having become so established afterward. If, on the other hand, the respect for the rawa, the guardian of correct usage in the accepted literary language, was transmitted by the Arab ruling class to the Mawali, we can understand that the latter turned to people like Khalil b. Ahmed in order to perfect their knowledge of the language, and finally to Bedouins as informants for their incipient philological studies.
Bloch's researches32 throw some additional light on our problem. Though restricting himself on the whole to the order of words, he appears to have established the point that the language of poetry does not, within Classical Arabic, constitute a special poetic variety, in the way, say, that the language of poetry does within biblical Hebrew.33 He shows, however, that a number of words and forms are preferred because they go more easily into meters,34 and that in general poetry freely uses constructions that are distinctly rare in prose. Bloch himself quotes G. Bergstrasser's dictum that the chief characteristic of Arabic syntax is the restriction of the large choice of proto-Semitic constructions to a few standardized types.35 This tendency of development is thus shown to continue into the development of Arabic prose out of the poetic koine. It is obvious that we have to reverse Bloch's approach, and not to treat poetry as a special case of a language principally used as prose but, on the contrary, prose-in Arabic at least-as a special use of an idiom normally associated with poetry. The freer syntax of the poems and of the Koran-and this is likely to apply to most of the Koranic particularities noted by N6ldeke36-is the original state of affairs, while the more regularized constructions of prose style are peculiar to the latter. The only construction Bloch found to be unparalleled in prose, idha clauses with the subject following immediately upon the conjunction,37 appears to be an archaism preserving the emphatic-demonstrative character of the particle.38
Another possible archaism of the poetic language is stated by W. Caskel,39 namely, that in pre-Islamic poetry diptotes and triptotes "are not yet strictly distinguished. 1140 This would be a difference of great interest if it were quite sure that the distinction between diptotes and triptotes developed only within Arabic, or, according to this view, within the `Arabiyya. However, it has been suggested that diptosy existed in Ugaritic;41 and OldAccadian, in denying case inflection to proper names,42 seems to have a similar phenomenon. It may thus be preferable to follow for once the Arab philologists in treating this confusion as a poetic license. It may well owe its wide extension to different usages in this respect in the home dialects of the poets, a factor that is perhaps also responsible for the bewildering variety of case usage in coordinated phrases-43
An interesting problem is raised by Bloch's list of common forms and constructions unsuitable for poetry because they contain sequences of three short syllables that fit into few metres, and of four short syllabes that fit into none, for example, falala, fa`alata, malikuka, fafa'alahu.44 The existence of such forms seems to be a forceful argument against the theory that the language was largely created by the poets;45 the syllabic structure must have existed in the dialect or dialects on which the koine was based, and which the poets did not feel entitled to alter. Even if such forms were never used in a line of verse, they existed virtually in the system, and were at hand when a prose emerged that was not bound to certain rhythmic sequences.
No progress seems to have been made in recent years in solving the problem of the place of origin of the poetic koin-e. We have still no data to place it any more exactly than in the general region of Nejd, Yamama, and the Euphrates. The position of Hira as an early center of poetry, and the fact that the earliest cycle of poems is connected with the War of Basus, which took place in the Euphrates region, would give a certain preference to that part of the area. As against this, Imru'ul-qais, one of the earliest great poets, was of Kinda, and the Kinda empire somehow seems to provide the natural background for the emergence of an Arab national art.
The real clue should, of course, be provided by comparison with the ancient, pre-'Arabiyya, dialects. Interest in these dialects has become widespread in recent years, and between 1940 and 1951 not less than four full-sized studies dealing with the subject have appeared: by H. Kofler,46 1. Anis,47 A. Hammuda,48 and C. Rabin.49 Kofler's is the most complete collection of material and references, though without an attempt at geographical evaluation.50 Ants concentrates on phonetic matters, trying to provide something like an Arabic comparative phonology. Hammuda draws upon an important source not used by the others, the Koran commentary by Muhammad b. Yusuf Abu Hayy5n.51 My own treatment, along the lines of geographical linguistics, is an attempt to recognize the common features of a group of dialects, those along the Western highlands, with particular stress upon the Hijazi dialect.
Between them these four works exhaust pretty well the information that is to be gotten out of Arab philologists concerning the grammar and syntax of the dialects. A good deal of work still remains to be done on their vocabulary, and is likely, as in the case of European studies in geographical linguistics, to provide us with better criteria for dialect geography than is done by grammar. It is disappointing, however, that all this work has brought us no nearer to a solution of the problem of the koine. No dialect or group of dialects within the above-mentioned wide area has emerged with a special claim to be the cradle of the `Arabiyya. On the contrary, precisely those regions in which poetry was cultivated mostthe Euphrates region, Tamim, the areas of Nejd bordering on the Hijazhave turned out to have spoken dialects rather distinct from the poetic idiom. It is unlikely that a study of the vocabulary will give clearer results, for vocabulary is easily borrowed, and it has long been recognized that the poetic idiom has widely borrowed from different dialects.
The mystery is deepened by the epigraphic languages of Arabia. After the systematic presentation of the grammar of South-Arabian by M. Hofner52 there can be no possible doubt that we have here a language completely distinct from Arabic, and the researches of W. Leslau53 show that it forms a group together with Ethiopic and possibly Accadian, so that any genetic connection with Arabic is becoming increasingly unlikely. Some of the results of Rabin,54 as well as a recent study by I. Al-Yasin,55 suggest connections of Arabic with Northwest Semitic. Arabic has generally been considered to form part of a separate branch of the Semitic languages called South Semitic or Southwest Semitic. Among the features distinguishing this branch are the preservation of a wider range of dental consonants and the broken plurals. The first feature is now known also to have distinguished Ugaritic, a Northwest Semitic language known since 1929, and the broken plurals are almost certain to be a late development, which has little value as a genetic criterion. It is, of course, an entirely different question to what extent Classical Arabic contains South Arabian loanwords, or was influenced in its style by South Arabian.
While thus the various South Arabian dialects are not very closely connected with Arabic, there is every likelihood of a close connection between Arabic and several languages known to us only through short inscriptions and graffiti in the northwest of the peninsula and to a smaller extent in the neighborhood of the Persian Gulf. These "proto-Arabic" languages are distinguished as Thamudic, Lihyanic, and Safaitic. In recent years the number of known graffiti has much increased, and grammatical research has been pushed forward.56 From references to various events it has been concluded that graffiti in these languages continued to be incised until well into the third century C.E.,57 perhaps even up to the lifetime of the Prophet.58 They are thus not far removed from the time to which we can ascribe the data recorded as referring to the ancient dialects or the period generally considered to be that of the beginning of Arabic poetry.
These languages resemble Arabic closely in their phonological system. Their vocabulary is similar but distinct. Because of the briefness and stylistic monotony of the texts their grammar is imperfectly known, but it shows some striking differences from Arabic, for example, a definite article h- (sometimes hn-) resembling Hebrew. The cultural connection between them and the later bearers of Arabic poetry is amply demonstrated by the large common fund of proper names, some of which can be traced back as far as Assyrian inscriptions about fights with "Aribi" of the ninth and eighth centuries B.C.E., while some are still borne by presentday Bedouins in Arabia and North Africa.
The strange thing is that these languages or dialects throw no light upon the linguistic development of Arabic. They are a group quite distinct from the dialects mentioned by the Arab philologists, and none of their distinctive traits can be traced among data about those dialects. The reason for this seems to be, partly, that the area covered by them, later called that of the Quda ca tribes, took no part in the poetic activities of the sixth century,59 and was not reckoned as part of the region whose Arabic was correct. Yet there are some data available concerning Quda'a dialects, and none of these fit in with what we know about the three protoArabic languages. As is well known, Arab historical tradition has much to tell about al-'crab al-ba'ida, lost tribes without genealogical connection with those of historical times. Some such tribes were actually still in existence, such as the Jurhum, from whose language a second-century authority60 purports to give twenty-four words. Some of the dialects dealt with by the philologists differ so strongly as to suggest that they belong to an older layer. Indeed, for the Himyar dialect spoken in the uplands of Yemen and the Azd dialect of the isolated coastal areas of Oman in the east and Asir in the south,61 the map suggests that they were remnants of earlier North Arabian expansions surrounded by seas of later, more "Arabic" arrivals.
At one time it was generally assumed that Arabia was the home of the Semitic peoples. The Arabic language, that is Classical Arabic, was likewise assumed to differ but little from the proto-Semitic parent language. It appeared as a quiet pool, opposed to the stormy development of the other Semitic languages. The realization that the camel was domesticated only around the beginning of the first millennium B.C.E., and that human settlement in Arabia is likely to date from that event, is now leading ever wider circles to a view of that country as an area of immigration as well as emigration, a meeting place of ethnic elements coming from various directions. These movements produced language mixtures62 and a checkered map of linguistic boundaries, islets, and isolated remnants of earlier migrations. I have tried to explain the chief linguistic cleavage of preIslamic Arabia, that of West Arabian and the eastern dialects, as being due to the meeting of genetically disparate linguistic groups.63
Classical Arabic is seen to stand at the end of a development, not at its beginning. When this development becomes clearer, as one hopes it will now that the archaeological reconnaissance of the peninsula has begun, its ancestry may well turn out to be a highly complicated one.
The important study "Stress Patterns in Arabic" by H. Birkeland64- which I saw only after the completion of this article-introduces a new problem into the question of the relation between the ancient dialects and the 'Arabiyya. The author shows (p. 12) that the fixed expiratory stress of the modern colloquials of Syria, Iraq, and Egypt was superimposed on a state of language that had long and short vowels where Classical Arabic has them, but had at that time already lost the hamza of words ending in -a'. He further comes to the conclusion that this older state of the language had no fixed expiratory stress at all. Both the absence of such a fixed stress and the loss of hamza, however, are features of West Arabian. On the other hand, we find that the ancient East Arabian dialects had a strong fixed expiratory stress, and that the reductions caused by it show it to have been just in the places where it is in the above-mentioned colloquials. It is reasonably certain that the eastern dialects had preserved the hamza of -a', especially as they are said to have had in a few cases -a' as against Hijazi -a;65 since the Arab grammarians do not know the concept of stress, and the structure of -a' nouns is such that tell-tale contractions cannot occur, we shall never know how these words were stressed in the ancient eastern dialects.
In any case, the colloquial situation presents a curious mixture of eastern and western features. One explanation might be that fixed stress spread in a "wave" movement from the Persian Gulf area, that it reached the home of the standard 'Arabiyya too late to affect its vocalism, and affected the Fertile Crescent and Egypt only after their spoken dialects had lost the final hamza. This theory fits in with Birkeland's discovery that in the Maghribine colloquials the same fixed-stress pattern is superimposed on another fixed-stress pattern, or, in other words, that it reached those colloquials only after a lengthy development bad taken place (p. 28). The situation in the ancient eastern dialects is, as far as I can see, fully consistent with the assumption that fixed expiratory stress was a fairly recent innovation. It may even have been due to contact with speakers of Aramaic, a language in which the effects of stress play a prominent role.
1. C. Brockelmann, Grundriss der vergleichenden Grammatik der semitischen Sprachen, vol. 2 (1908).
2. H. Reckendorf, Die syntaktischen Verhaltnisse des Arabischen (1895-98); Arabische Syntax (1921).
3. In M. Gaudefroy-Demombynes and R. Blachere, Grammaire de l'arabe classique (1937).
4. Reckendorf, Arabische Syntax, p. iii.
5. J. Fuck, cArabiya (Berlin, 1950).
6. Bloch, hers and Sprache im Altarabischen (Basel, 1946), Acta Tropica, Suppl. 5.
7. See especially J. Schacht, Origins of Muhammadan Jurisprudence (Oxford, 1950).
8. Collected by M. Hamidullah, Documents sur la diplomatie musulmane, etc. (Paris, 1934).
9. Cf. C. Rabin, Ancient West-Arabian (London, 1951), pp. 21-23. The theoretical foundation of this view, said ibid., p. 22, to emanate from Ibn Faris, seems in fact to be due to a1-Farra', cf. P. Kahle, JNES 8 (1949): 70 col. 1.
10. For instance, P. Dhorme, Langues et ecritures semitiques (1930), p. 53, and L. H. Gray, Introduction to Semitic Comparative Linguistics (1934), p. 5, accept unquestioningly the identity of Classical Arabic with the Meccan dialect. Even Fuck, Arabiya, pp. 2-3, is rather vague on this point, though he admits that the Meccan dialect may have differed from the Bedouin dialects as much as these last did between themselves.
11. Taha Husain, Al-Adab al-jahili (Cairo, 1927).
12. K. Vollers, Volkssprache and Schriftsprache im alten Arabien (1906).
13. A list of views in Rabin, Ancient West-Arabian, p. 17. An interesting new suggestion is that of W. Caskel, that the 'Arabiyya originated among the settled populations of NW Arabia and was transported into Central Arabia as part of the process of Bedouinization (ZDMG 103 (1953) p. 34; trans. in "Studies in Islamic Cultural History," American Anthropological Association Memoir 76 (April 1954): 43.
14. R. Blachere, Histoire de la Litterature arabe (Paris, 1952), vol. 1, pp. 66-82.
15. Ibid., pp. 80-81.
16. H. Fleisch, Introduction a l'etude des langues semitiques (Paris, 1947), pp. 97-101.
17. R. Blachere, Introduction au Coran (Paris, 1947), pp. 156-69.
18. Fuck, 'Arabiya, pp. 3-4 (the manuscript was sent to the publisher in 1947).
19. So Brockelmann in Grundriss, vol. 1 (1908), p. 25.
20. This was also the view of Brockelmann in 1947, according to a note in Fleisch, Introduction a l'etude des langues semitiques, p. 100.
21. Fuck, 'Arabiya, p. 3.
22. H. Birkeland, Sprdk og religion hos J¢der og Arabere (Oslo, 1949), pp. 35-41. See, however, below.
23. P. Kahle, The Cairo Geniza (London, 1947), pp. 78-84; Goldziher Memorial Volume I (Budapest, 1948), pp. 163-82; JNES 8 (1949): 65-71.
24. F. Noldeke, Neue Beitrage zur sem. Sprachwissenschaft (1910), p. 2.
25. Al-Hasan b. Muhammad al-Malik-1, At-tamhid fi ma`rifat at-tajwid (Chester Beatty MS. no. J. 152).
26. So, and not, as Kahle has it, "whoever reads a part with i`rab and a part with lahn."
27. Cf. O. E. Ravn, Om nominernes bojning, etc. (1909), p. 21.
28. Ibid., p. 117.
29. Fuck, `Arabiya, p. 2.
30. Cf. Violet, OLZ 4 (1901) 384-403; A. Baumstark, Oriens Christianus 18 (1934): 55-66; B. Levin, Die griechisch-arabische Evangelien-Uebersetzung (1938). The numerous reminiscences of the New Testament discovered in the Koran by W. Rudolph (Abhangigkeit d. K. von Judentium and Christentum [Stuttgart, 1922]) and K. Ahrens ("Christliches im Q.," ZDMG 84 : 15-68, 148-90) really presuppose an Arabic written source, not mere oral preaching. Wellhausen also believed that Christians were the first to use Arabic as a literary language (Reste arabischen Heidentums, p. 232).
31. Papyrus scraps, stones, palm leaves, bones, pieces of leather, and pieces of wood; cf. Noldeke, Geschichte des Qordns, 2d ed., vol. 2, p. 13.
32. See above, n. 6.
33. Cf. G. R. Driver, Hebrew Poetic Diction, Supplements to Vetus Testamentum (1953), vol. 1, pp. 26-39.
34. Pp. 7-10.
35. Bloch, Einleitung in die semitischen Sprachen (1928), p. 135.
36. Noldeke, Neue Beitrage zur semit. Sprachwissenschaft, pp. 5-23; French translation by G. H. Bousquet, Remarques critiques sur le style et la syn- taxe du Coran (Paris, 1954).
37. P. 105.
38. Cf., for instance, Rabin, Ancient West-Arabian, p. 38.
39. In the lecture quoted above, n. 13; p. 37 of the English, p. 29 of the German version.
40. For examples, cf. Wright II, pp. 387-88.
41. C. H. Gordon, Ugaritic Grammar, 2d ed. (1947), p. 43.
42. W. von Soden, Grundriss d. Akkadischen Gramm. (1952), p. 81, par. 63 f.
43. E. g., Wright, II, 40C, 97A.
44. Pp. 7-10.
45. This seems to be, among others, the opinion of Brockelmann in EI, I, 408b.
46. "Reste altarabischer Dialekte," WZKM 47 (1940)-49 (1942), altogether 188 pp.
47. Al-lahajat al-`arabiyya (Cairo, ca. 1946), 183 pp.
48. Al-gira'at wal-lahajat (Cairo, 1948), 226 pp.
49. See n. 9.
50. A briefer treatment, partly along geographical lines, based upon Kofler's material was undertaken by E. Littmann in his article, Bagaya 1-lahajat al'arabiyya ft 'l-adab al-'arabi, Majallat Kulliyyat al-Adab, 10, no, 1 (May 1948): 1-44. Another interesting attempt to work out features of a single local dialect is by K. Petracek, "Material zum altarabischen Dialekt von al-Madina," Archiv Orientalni 22 (1954): 460-66.
51. See Noldeke, Geschichte des Qorans, 2d ed., vol. 3, p. 243.
52. M. Hofner, Altsudarabische Grammatik (Leipzig, 1943).
53. "South-East Semitic," JOAS 63 (4-14); "Vocabulary Common to Accadian and S.-E. Semitic," South-East Semitic 64 (1944): 53-58.
54. Cf., e.g., Rabin, "Archaic Vocalization in Some Biblical Hebrew Names," Journal of Jewish Studies 1 (1948): 22-26; "The Ancient Arabic Dialects and Their Relationship to Hebrew," Melilah 2 (1946): 243-55 (in Hebrew); Ancient West-Arabian, pp. 196-69.
55. I. Al-Yasin, The Lexical Relation between Ugaritic and Arabic (New York, 1952), providing 660 equations, not all of them equally certain.
56. Bibliography by G. Ryckmans, Le Museon 61 (1948): 137-213. A. v. d. Branden, Les inscriptions thamoudeennes (1950). Good introduction in E. Littmann, Thamud and Safa (1943).
57. Cf. Rodinson, Sumer 2 (1946): 137-55; a line in Thamudic appears on a Nabataean stela of 267 C.E.
58. F. V. Winnett, JAOS 73 (1953): 41, dates a Safaitic graffito in 614 C.E.
59. Cf. Lammens, L'Arabie Occidentale, p. 308.
60. Cf. Rabin, Ancient West-Arabian, p. 7.
61. Ibid., pp. 42-63.
62. The Tayyi' are said to have adopted the language of the Suhar, whom they had overcome, according to Yaqut, Mu jam al-bulddn, vol. 1, p. 127; the Hudhail dialect betrays strong Eastern Arabic influence, cf. Rabin, Ancient WestArabian, p. 79.
63. Rabin, Ancient West-Arabian, pp. 1-2.
64. Avhandlinger utgitt au det Norske Videnskaps-Akademi i Oslo, II, Hist.- Filos. KI. 1954, No. 3; Oslo, 1954.
65. Cf. my Ancient West-Arabian, p. 141.