In this slim volume, Jonathan Owens sets himself two ambitious aims: to reconstruct in outline some of the principal characteristics of both Proto-Arabic, the ultimate parent of all later varieties of Arabic, and pre-diasporic Arabic, a cover term for the varieties of the language spoken in the Arabian peninsula at the time of the Islamic conquests and “exported” by migrations over the succeeding century or so. The first of these objectives has rarely been attempted, while the second, to which Owens devotes rather more of his attention, has been the subject of controversy in Western scholarship for over a hundred years. Following an introductory chapter, the meat of the book consists of seven chapters, three of them published in similar form before, all of them variations on the theme of historical reconstruction.
The data Owens adduces is of two main types: attestations of “Old Arabic” (OA) and evidence from the modern dialects. OA is not to be identified with Classical Arabic (CLA); the OA data consists mainly of the reports of the early grammarians such as Sibawayhi (d. 793), who, in the larger context of establishing a set of “best practice” rules, which subsequently became reified as CLA, made many detailed observations on the spoken language usage he observed around him. A second ancient source of information on OA, especially relevant to the issue of whether there were case-less varieties of early Arabic, comes from the quranic reading traditions (qirā’āt) of Sibawayhi's time and the generation immediately before. Excluded from consideration is the language of pre-Islamic poetry because it probably represents a poetic koiné rather than ordinary usage, and may in any case have undergone “correction” in the process of its transmission via the back-projection onto it of the subsequently codified rules of CLA. The other source of data for reconstruction are the modern dialects of Arabic. Data from some forty-nine separate dialects covering the whole of the Arabic-speaking world are adduced, including several from the vast Western Sudanic Area (WSA) – Sudan, Chad, Nigeria and Cameroon – which until recently have been little investigated, and tiny Sprachinseln such as the Arabic-speaking community of Uzbekistan, a region which was settled in the early eighth century by Arabic speakers, but was soon cut off from contact with Arabic speakers from other areas.
The strength of the author's approach lies in synthesizing in a sustained manner data of these disparate types. As a source, Sibawayhi is often tricky to interpret: his obscure technical terminology (what, for example, did he mean by rawm and ’ishmām in describing pausal phenomena in the spoken Arabic of his time?); the analogical, ahistorical style of his reasoning, in which an internal logic, however seemingly far-fetched, had to be found to account for every competing form; and the only rudimentary information he gives on the social/tribal distribution of the linguistic usages he describes. Owens is to be commended for his close and thorough reading of Sibawayhi and al-Farrā’ (sources all too rarely encountered among Arabic dialectologists); for his dogged deployment of the comparative method in finding unlikely links between spatially and chronologically widely dispersed modern dialects, and relating his findings, usually in a highly plausible fashion, to the evidence from OA; and, most of all, for his unblinkingly critical stance towards the received, but in his view wrong-headed conceptualizations of Arabic that have for so long dominated the field. These, he observes, are often cast in the form of overly neat dyadic oppositions – (diachronically) Old Arabic v. Neo-Arabic, (typologically) synthetic v. analytic, (ecologically) bedouin v. sedentary – or in terms of geographical blocs such as “Mesopotamian” Arabic, “Egyptian” Arabic, “Maghrebi” Arabic and so on. To be sure, these labels can often serve as handy reference points, but the danger is that they become thought of as primary classificatory categories, closing off research avenues that might reveal, inter alia, historical linkages between their speakers that predate their current locations.
Owens’s basic premise is that the phenomenon of a dialectal feature that is widely geographically dispersed, especially if it is morphological, and especially if it is unusual, is to be explained by the early contact of those who have the feature at the pre-diasporic stage, rather than by polygenesis. Pre-diasporic Arabic is not conceived of as variation-less monolith, but comprising an unknowable number of sub-varieties. Certainly, the early sources, which are full of examples of tribal variations, suggest this, but they do not give us the full picture. Sibawayhi, after all, never ventured into the Arabian peninsular, and lived the whole of his life in Basra. However widely he cast his net for informants, he cannot have captured the full diversity of the Arabian dialects. Hence the reason to scrutinize the contemporary dialects and the patterning of isoglosses, along with what is known about the history of tribal migrations at the time of the conquests and immediately after.
In chapter 2, “Old Arabic, Neo-Arabic and comparative linguistics”, the Western Arabicist tradition is subjected to a sustained critique, in particular the ideas that: (1) Neo-Arabic (NA) is “chronologically younger” than OA (first claimed by the Finnish Orientalist Wallin in the mid-nineteenth century, on the basis of Najdi texts); (2) Classical Arabic is the starting point for the development of all later varieties, literary or dialectal. Owens regards both these assumptions as “fundamental impediments to a linguistic history of Arabic” (p. 40). Furthermore, Middle Arabic, proposed by Blau as the “missing link” in this chain, is more plausibly viewed not as a chronological stage, but as a range of styles in which the writer aims at writing a form of Arabic more elevated than his normal speech, but misses the target (p. 47). Texts with characteristics identical to those of medieval Middle Arabic were still being produced in the seventeenth (and, I would add, the nineteenth) centuries. The claimed crucial differences between OA and NA (W. Fischer and O. Jastrow, Handbuch der arabischen Dialekte, Wiesbaden: Harrassowitz, 1980) are not always, or even generally, consistently found. For example, unstressed short vowel deletion in open syllable is a characteristic of some but by no means all NA dialects, and such vowels were not always maintained even in OA: Sibawayhi reports vowel elision in sequences of short open syllables in Hijazi, for example. Secondly, the data are (inevitably) incomplete: for example, dialects described since Fischer and Jastrow's work disprove the claim that, e.g., the glottal stop is completely lost in the dialects; that there is never a fricative/stop contrast in the alveolar-dental emphatic; and that mā “what” does not exist as an inherited original form in any modern dialect. All three features have since been found by Behnstedt in the Yemen (and, I would add, mā also occurs in Oman, e.g. mu (<ma hu) ismak? “What's your name?”). Fischer and Jastrow leave the dialects of the entire Sudanic region out of account, even though Carbou (1913) had described some of the basic features of Chadian Arabic. The effect of these over-generalizations and omissions is to reduce the number of differences between OA and NA that Fischer and Jastrow list to the point where one may ask whether the distinction is really justified. Among these, Owens identifies only four unchallengeable differences, and in one of these, the dual, Jastrow and Fischer themselves argue that the simpler NA dual system, whose scope is only nominal, is in fact the older one, and similar to that of ancient Semitic languages like Akkadian. According to them, the more elaborate OA system is a later development, and “wohl nur eine Kunstform der altarabischen Dichtersprache” (Fischer and Jastrow, p. 46). Rather than represent a decisive break with OA, brought about, so the theory goes, by the export of OA to the conquered territories, and its learning by non-Arab populations, could not, Owens asks, NA really be just a continuation of trends already present in one or more of the pre-diasporic varieties and, earlier, Proto-Arabic? If so, what basic features characterized them?
These questions are answered in the next six closely argued chapters, which focus on the following issues: Was there, as Vollers first claimed in 1906, a caseless variety of pre-diasporic Arabic (or one on the way to becoming that)? And, if there was a variety with case, are there any vestiges of a case system left in the modern dialects that might argue for an ancestor language from which these vestiges were inherited (ch. 3 and 4)? Which features of modern Arabic dialects do an extended cross-dialectal comparison suggest might have been present in the pre-diasporic varieties (ch 5–8)? A collection of forty-nine variables are inspected and subjected to a statistical analysis in ch. 5; the origins of the shape of the imperfect verb in Nigerian Arabic are investigated in ch. 6; the incidence, conditioning and history of imāla (vowel-raising) in ch. 7; and the shape of suffix pronouns in Proto Arabic in ch. 8.
The direct evidence for case in some pre-diasporic varieties seems overwhelming, if Sibawayhi's observations are to be taken at face value. However, his lengthy discussion of what seem to be variable case-endings in the same syntactic environment, as well as variants associated with one tribe rather than another, suggest that the system may have been in flux; and in both the Quran and early collections of lore and poetry like the Kitāb al-Aghāni (F. Corriente, “Marginalia on Arabic diglossia and evidence thereof in the Kitab al-Aghani”, Journal of Semitic Studies 20, 1975, 38–61) there are neutralized nominative/genitive contrasts, inflectionally invariable duals, mixed up cases, or none at all. The eighth-century Abu Amr's quranic reading tradition (al-idghām al-kabīr), slightly before Sibawayhi but quoted extensively by him, involves the suppression of word-final vowels, morpho-syntactic or lexical, and the progressive assimilation of the first C in the resulting C#C cluster. But it could equally be argued from the description Sibawayhi gives that it results from there being no final vowel there to delete. This may suggest the co-existence of a case-less (Hijazi?) Arabic alongside (eastern Arabian?) varieties with case, the latter being represented in the other reading traditions. As for evidence for vestiges of case in the modern dialects on which a pre-diasporic case system might be reconstructed, the potential candidates are two: what is often described as the vestiges of tanwīn, and the vowels which occur between nouns and suffixed pronouns, either now morphologized as the vowel of a vowel-initial suffix pronoun, or as an epenthetic vowel before a consonant-initial one. Neither, however, turns out to be a convincing candidate. Dialectal tanwīn is both formally and functionally different from the CLA system. Formally, it is generally in the form of a suffix (usually -in) that may be attached to a noun when it is followed by an adnominal modifier of some kind. This noun can, and often is, in a class that did not have tanwīn in CLA (e.g. muslimīn-in liŧāf “bad Muslims”, p. 104). It is always fixed in form (unlike the -un/-an-in CLA system), always optional, and almost always a marker (Owens calls it a “linker”) of noun modification. It has no function at all as a marker of case. In other words, it does not look like the descendant of the CLA case system, or, put another way, one would not reconstruct the CLA case system on the basis of the modern data. Dialectally, it is not a widespread feature, but its distribution is far-flung, chronologically and geographically, being attested with virtually identical forms and functions in the eighth-century Arabic of Andalusia, and in the modern dialects of Najd, the Tihama, Sudan, Afghanistan and (I would add) the Gulf States and Oman. Given its unusual nature, it looks therefore like an ancient shared feature of pre-diasporic Arabic, possibly predating Sibawayhi (who makes no mention of it). The evidence for a vestige of case vowels in the dialects is equally weak. The post-nominal vowel in forms such as galb-a-na, and in forms like raas-ak in which the a of the -ak suffix was claimed by Cantineau to be a vestige of inflection, can be explained more plausibly as the output of general rules of epenthesis that applied to clusters, however formed.
A further striking illustration of what can be revealed by Owens’ approach is in ch. 5, where he compares statistically the reflexes of some sixty-five (mainly morphological) variables in the dialects of two large areas, the WSA area and Mesopotamia, and includes Uzbekistan (near the latter) and Shukriyya, eastern Sudan (near the former) as “controls”. Despite its vast area and relatively sparse population, WSA is shown to be much more dialectally compact than the smaller and more heavily populated Mesopotamia. Even more surprisingly, Uzbekistan Arabic turns out to be quite close to that of WSA, and Shukriyya and Mesopotamia are also similar to one another. Owens gives a number of possible reasons for the dissimilarity of Mesopotamian dialects: divergence through different patterns of contact, independent innovation, different characteristics of the founding dialects – and, one might add, sociolinguistic reasons (compare, for example, the deep and long-standing Arab-Baharna dialectal cleavage on the small island of Bahrain, maintained by lack of contact because of confessional differences). But what of the similarity between very distant dialects, which Owens’ data also reveals? This is explained as a consequence of their common origin, and Owens provides another striking piece of evidence for this claim to go with that of the distribution of the noun–adnominal linker described earlier. This is an intrusive -in- infix that is inserted between an active participle and an object pronoun, as in kaatb-in-ha “I/you(m)/he has written it”, kaatbat-in-ha “I/you(f)/she has written it”. Of the three alternative possibilities which are found in Arabic dialects to express this participial-object structure, the one with the -in- infix is by far the rarest, and it is most unlikely that independent development would be the explanation for its observed distribution. It is attested only in isolated dialects that are widely dispersed: western Hadramawt, Bahrain (Baharna only), parts of the Emirates, Oman, WSA, Khorasan, and (though in a much changed form) Uzbekistan. That these dialects may be from a common pre-diasporic source finds support in what is known about early tribal migrations: “Qaysites” are known to have formed part of the eighth-century Arab army that conquered Khorasan, and others of the same tribal confederation migrated at roughly the same time from Arabia to Upper Egypt, which was the area from which the WSA was arabized beginning in the fourteenth century. On Owens's argument, both groups were originally, in pre-diasporic times, from a location in north or north eastern Arabia, and took with them to their new abodes a dialect of which this participial construction, and the adnominal linker construction, formed part. Neither of these rare features is attested as an old tribal variant, and neither can be plausibly derived from the descriptions of OA we have. But, as Owens observes, the fact they are not attested by Sibawayhi does not mean they did not exist, and it is also hard to advance any explanation other than Owens's to explain their current distribution. In subsequent chapters, Owens details further examples of widely dispersed modern dialectal variants, and processes (e.g. imāla) whose shared form and distribution suggests that post-diaspora polygenesis is an unlikely explanation, but there is insufficient space to go into them here.
This book is a highly stimulating read for anyone with an interest in the history of Arabic. Owens’s close reading of the old sources, married to his deployment of the classic comparativist approach, succeeds in giving a good shaking to the comfortable theoretical structures of contemporary Arabic dialectology. It may be that occasionally he goes too far in his claims and interpretations of individual variables, but his basic argument that the present-day Arabic dialects are originally descended from a set of pre-diasporic dialects not identifiable with Classical Arabic is, to this reviewer at least, pretty convincing. It is not too much of a stretch to suggest, as Owens does in his conclusion, that this book resurrects the arguments of Vollers, Landberg and later Kahle concerning the linguistic history of Arabic, against the arguments of Nöldeke, Brockelmann and Fück, but with much more modern data to back them up than was available in the early twentieth century. It deserves a warm welcome.
