1.0. Introduction to the Ottoman Cosmopolitan

1Arabic, Charles Ferguson has famously told us, is—like modern Greek—a diglossic language, ‘high’ and ‘low’ varieties of which are used in different and complementary settings. Diglossia differs from bilingualism in that it involves two varieties of the same language; moreover, the high variety lacks native speakers, and is acquired only in formal educational settings, and used only in official and/or written forms of communication. This diglossia, he adds, has proved remarkably resilient and enduring. Since their original publication in 1959, however, Ferguson’s ideas have been modified and refined: varieties of modern Arabic other than the two reified high and low registers have rightly been distinguished; and the diglossic situation in Arabic, and even more in modern Greek, has been shown to be rather less stable and more contested than Ferguson’s irenic picture would have us believe.1

2Here, I would like to suggest that we can fruitfully explore the topic of Arabic diglossia—and of the development of modern Arabic more generally—laterally, and in a comparative and diachronic manner. More concretely, when studied in their broader Ottoman and post-Ottoman settings, the diglossic constellations of Arabic and Greek turn out to be but two very distinct outcomes of a rather broader process of vernacularisation, that is, a shift from written classical to locally spoken language varieties, in which hitherto spoken languages started being used for new literate uses, such as, most importantly, official courtly communication, high literature, and learning. This broader process in fact occurred across virtually the entire the Ottoman Empire; its consequences are still visible in the Empire’s various successor states. Here, however, I will not discuss the case of Arabic in detail; rather, I will briefly sketch the wider pattern of development, and leave discussion of the implications for the study of Arabic to another occasion.

3I take my cue from Sheldon Pollock, who has, famously, identified a number of cosmopolitan orders in the world of Latinity and the Sanskrit-based civilization in and around the Indian subcontinent during the first millennium CE; both of these orders, he further argues, went through broadly similar processes of vernacularisation around the year 1000 CE.2 In Western and Southern Europe, this process yielded written Romance languages like Italian, Catalan, and French; in South Asia, vernaculars like Tamil, Telugu, and Kannada were similarly promoted to written status. Thus, vernacularisation is not specifically or uniquely modern or European; it may occur at different times and in different places.

4The Ottoman Empire, as I hope to show below, knew a cosmopolitan order similar to those of Sanskrit and Latinity; and it, too, went through a massive wave of vernacularisations, in the 17th and 18th centuries CE. These vernacularisations, moreover, paved the way for the new, vernacular language-based ethnic identities and national movements that emerged in the course of the 19th century. In their earlier stages at least, these identities and movements developed largely, if not completely, independently of any cultural, ideological, or political influence or interference from Western or Central Europe. Thus, the widely held but rarely investigated assumption that national identities outside Europe were crucially influenced by European (and, more specifically, German) romantic nationalism and shaped by the categories of philological orientalism would seem to deserve reconsideration.

5Although many discussions of nationalism contrast the multilingualism of premodern empires with the monolingual ideals and the linguistic standardisation of modern nation states, few empires can match the diversity and complexity of the early modern Ottoman linguistic constellation. In the Ottoman Empire, Arabic enjoyed a high status as the language of the revelation of Islam and of Islamic religious learning; but it was not the only written language of prestige even among Muslims, let alone other population groups. Famously, the Ottoman elites recognised ‘the three languages’ (elsine-i selâse) that dominated literate communication: Arabic for religious learning, Persian for poetry, and Ottoman Turkish for administration and official correspondence. The latter, as is well known, was a form of Turkish with a large, if highly variable, proportion of vocabulary items and grammatical constructions borrowed from Arabic and Persian; being virtually incomprehensible to the uneducated masses, and deliberately so, it also served as a marker of social distinction for the Ottoman bureaucratic elites.3

6Christians living in the Empire had a number of classical, or sacred, languages of their own: in theory, Koinè Greek served as the language of liturgy and learning for all Orthodox Christians in the Empire, although some Orthodox communities used other ancient tongues, like Old Church Slavonic in the Balkans or Arabic in the Levant. Armenians, who had had their own church for centuries, used Grabar, or classical Armenian, as a liturgical and learned language; and Eastern Christians of different denominations generally used Syriac, which had been the regional lingua franca in the Fertile Crescent prior to the arrival of Arabic, but by the early modern period had become a dead language, and was used exclusively in formal and/or written communication. The spoken varieties of these languages had a rather lower status—so low, in fact, that, among Greeks and Armenians in particular, one observes substantial language loss and a shift towards locally dominant languages or spoken linguae francae like Turkish, colloquial Arabic, or Kurdish. There is no evidence that this language shift was due to repressive Ottoman policies, as some nationalist historians have claimed; in fact, there is little evidence of any substantial Ottoman language policies prior to the last decades of the 19th century CE.

7Among Ottoman Jews, the ‘Sacred Language’ (leshon haqodesh), a blend of Hebrew and Syriac, was the main written language prior to the arrival of large numbers of Sephardic Jews from the Iberian Peninsula in the late 15th and early 16th centuries. The main written language of this group was ‘Judaeo-Hispanic, ’ grammatically a calque of the sacred language with a large number of Hispanic lexical items; this was distinct from ‘Ladino’ in the strict sense, the commonly spoken variety of Judaeo-Hispanic, which was much closer to colloquial 15th-century Spanish.4

8Apart from these, there were also languages that had little or no written tradition like, most significantly, the Romance varieties spoken by several Orthodox Balkan Christian groups, Albanian, and Kurdish, not to mention a number of mixed languages like the famous ‘Asia Minor Greek, ’ which was almost exactly half Greek and half Turkish in its vocabulary and grammar, and the language varieties spoken by the Dom, or ‘Gypsy’, groups in different parts of the Empire. Although we have rather less information about these spoken vernaculars on the verge of the vernacularisation wave, we are fortunate in having a rich and relatively reliable source of information in Evliya Çelebi’s famous Seyâhatnâme, or ‘Book of Travels’, which was written in the mid- 17th century CE but not published until three centuries later.5 Spoken language is always foremost in Evliya’s mind, with sex a close second. Accordingly, the Seyâhatnâme offers a plethora, not only of basic vocabulary and stock phrases in various Ottoman vernaculars, but also obscene expressions. The care and precision of its transcriptions make this work a precious source for linguists even today.

2.0. Early Modern Ottoman Vernacularisation

9Evliya observes that in the medreses of the Empire’s outlying regions, Arabic and Persian were the main languages of instruction; but he also describes how Muslims in Ottoman Bosnia used a small Turkish-Bosnian lexicon—a vocabulary that has become known, and in fact appears to have gained a rather wide circulation, under the title of Potur shahidiyya (Dankoff et al. 2011, 5/229–30). That is, he points to the vernacularisation of ‘Bosnian’, i.e., the locally spoken South Slavic dialect, which was very close to the varieties that have subsequently become known as Serbian and Croat. This is one of the earliest examples of a much broader pattern of vernacularisation in the early modern Ottoman Empire: between the 17th and the early 19th centuries CE, various Ottoman population groups in different parts of the Empire shifted to new written uses of local vernacular languages.6 The best known, and best documented, examples of this process are probably those among the Empire’s various Christian groups. First and foremost, among Ottoman Greeks, a movement arose in the mid-18th century, pioneered by authors and actors like Iosipos Moisiodax and Adamantios Korais, which propagated the use of language varieties closer to locally spoken dialects than the millennia-old Koinè Greek, with the aim of making Greek-language education easier and less time-consuming. Amidst fierce polemics, Korais—ultimately successfully—argued that a modern, civilised Greek nation should speak and write neither a vulgar dialect nor the old-fashioned Koinè Greek, but a purified form of language (subsequently called Katharevousa), which was free of Turkish loans and enriched with neologisms to express modern concepts. Likewise, among Ottoman Armenians, in early modern times a new, supraregional variety emerged, called K’aghak’akan or ‘the civil language’, which was much closer to— though not identical with—regionally spoken dialects, and hence much easier to learn, read, and write than classical Armenian.7 In the Ottoman Balkans, authors like Dositej Obradovic and Vuk Karadzic encouraged the written use of South Slavic (subsequently labelled ‘Serbian’), against the dominance of both Koinè Greek and Old Church Slavonic; among Ottoman Serbs that had sought refuge in the Austro-Hungarian Empire in the 1690 exodus headed by patriarch Arsenije III, a supraregional language for learned and literate communication emerged that was called ‘Slaveno-Serbian;’ its use was actively encouraged by the Habs-burg authorities, as a way of countering Russian linguistic, religious, and political influences. Further Eastward, in the Danube provinces, mid-18th-century authors like Paisii Hilendarski and Sofronij Vracanski simultaneously preached and practiced the literate use of the Bulgarian, or as they called it, ‘Slaveno-Bulgarian’, vernacular; and already earlier in the century, the famous Dimitrie Cantemir had pioneered the written and printed use of Romance vernacular locally called ‘Wallachian’, but subsequently labelled ‘Romanian.’ Initially, Cantemir appears to have intended this Romance vernacularisation as a way of countering the influence of Old Church Slavonic; but its later proponents emphasised the venerable pedigree of this vernacular in the Latin of antiquity, in an obvious effort to counter the dominance and prestige of Koinè Greek.

10But these developments were not restricted either to the Empire’s European provinces or to its Christian population groups. The Sephardic Jewish communities witnessed (or rather, caused) the emancipation of spoken Ladino as a medium of religious learning in the early 18th century. During the same period, Muslim Albanians started to produce Arabic–Albanian and other vocabularies for educational purposes, and started composing learned divan poetry in an Albanian enriched with Arabic, Persian, and/or Ottoman Turkish expressions, locally called bejtexhi or ‘Bayt poetry’. In the Empire’s Easternmost provinces, Kurdish authors like Ehmedê Xanî started using Kurmanji or Northern Kurdish both for didactic works and learned mathnawî poetry. In Mesopotamia, different denominations of Eastern Christians started using different forms of modern Aramaic, as distinct from classical Syriac, for literate, literary, or liturgical purposes. Even Ottoman Turkish witnessed significant attempts at simplifying the written language of bureaucracy in the 18th century in the direction of the Turkish dialect spoken in Istanbul, to the dismay of some officials, who feared they could no longer show off their social and linguistic distinction.8 This period also witnessed significant linguistic shifts among different Ottoman population groups: in the 18th century, substantial numbers of so-called Romaniotes, or Greek-speaking Jews of the Ottoman Balkans, started speaking Ladino; and many Copts in Egypt and some Eastern Christians in the Mashriq and in Mesopotamia, appear to have become Arabised, largely abandoning their traditional vernaculars in favour of colloquial Arabic.

3.0. Attempts at Explanation: The Role of Vernacular Philologies

11The fact that similar processes of vernacularisation occurred across, and perhaps even beyond, the early modern Ottoman Empire calls for explanation. At present, however, we are at a loss for any such explanatory account. For linguists, it would seem reasonable to suspect some kind of areal convergence or other form of language contact; this would raise the further question of whether such common or converging innovations simultaneously occurred in several languages, or rather started in one language, which then triggered similar changes in others. Such areal explanations, however, may be only part of the story: given that vernacularisation involves written rather than spoken language forms, and literate elites rather than the uneducated masses, such questions of cultural contact may also involve factors that are not strictly or structurally linguistic. To mention but one example: although the spoken varieties of Southern Slavic known today as Serbian, Croat, and Bosnian were mutually intelligible, and were in contact in urban centres like Sarajevo, the written traditions developed by authors writing in each of these three vernaculars were, for all practical purposes, completely independent from one another, if only because they involved, respectively, the Cyrillic, Latin, and Arabic alphabet.

12One obvious level to look for explanations is the Ottoman political economy, in particular the well-known phase of some form of economic ‘liberalisation, ’ coupled with a relative political decentralisation, in the 17th and 18th centuries CE.9 Perhaps, then, we may fruitfully relate early modern cultural and linguistic phenomena to the rise of mercantilism; and indeed, among the Greek and Armenian communities in the major cities of the Western Ottoman Empire, like Istanbul, Izmir, and Salonica, something like a mercantile bourgeoisie had emerged, which had become affluent through trade with Christian powers, especially in the Western Mediterranean and Central Europe. The rise of such new secularised elites may tempt us to see linguistic developments among them as triggered and inspired by the cultural epiphenomena of such commercial contacts, and in particular by imported ideas associated with the Enlightenment and early Romantic nationalism. But quite apart from the question of whether there were any concrete and coherent vernacularising doctrines or tendencies specific to the European Enlightenment, such an explanation overstates Western European influence and downplays local Ottoman dynamics. These vernacularising processes, after all, took place not only among the European-oriented mercantile bourgeoisie in the Empire’s urban centres, but also among different population groups in its more remote and isolated rural peripheries.

13Given these difficulties, we should perhaps first try to isolate and explicate all potentially relevant linguistic, sociolinguistic, and other factors before attempting any explanation. There are several such factors that may help in guiding our explanations; but here, I will discuss only the role of printing and of vernacular philologies. First, it should be noted that some, but by no means all, of these vernacularising movements were accompanied and facilitated by the use of printing. Thus, texts in different varieties of Greek and Armenian were printed in centres like Venice and Vienna, primarily targeting publics living in Ottoman territory and often sponsored by wealthy Ottoman citizens. Even more intriguingly, these foreign presses also produced materials written in Turkish, but printed in Greek or Armenian characters (subsequently called, respectively, ‘Karamanlidiki’ and ‘Armeno-Turkish literature’), indicating that by this time, a substantial part of the affluent reading publics could read these scripts, but had long since shifted to spoken Turkish. The Empire’s Sephardic Jews had known-printing in Judaeo-Hispanic since the 16th century CE; but from the early 18th century on, printed works of religious learning (and, later, increasing numbers of secular texts) in colloquial Ladino started being published as well. Famously, Ibrahim Müteferrika’s government-sponsored press printed a number of works in Ottoman Turkish in the first half of the 18th century; but in the face of protests from scribes and copyists, and more importantly of disappointing sales, it discontinued activity. Other vernacularising movements, however, like those among Albanians, Bulgarians, and Kurds, would not involve printed texts until well into the 19th century. In short, the mere availability of printing technology was in itself neither a causal factor nor a necessary feature of the various Ottoman vernacularisations.10

14A second important if variable aspect of Ottoman vernacularisation is the appearance of vernacular grammars. The writing, let alone printing, of such grammars points to a later stage in the process of Ottoman vernacularisation, which stretches from the mid-18th to the mid-19th century. Until then, grammatical instruction was generally restricted to classical or sacred languages among Muslims, Christians, and Jews alike. In Muslim educational institutions, only Arabic grammar was studied systematically; Persian was acquired not by studying grammatical textbooks, but by reading works like Saʿdî’s Golestan; and Ottoman Turkish, which had no fixed grammatical or stylistic rules or norms to begin with, appears to have been acquired informally, or simply to have been presumed as known. Even less current was any belief that locally spoken dialects were worthy of having their grammars written down and studied—or indeed that they had a system of grammatical rules to begin with. Generally, vernaculars appear to have been seen as deviations from classical norms or rules, rather than as full-fledged languages having rules of correctness of their own.11

15This was to change in the 18th century: during this period, one witnesses the development of what one may call ‘vernacular philologies’, in particular through the writing of grammars and lexica for various vernaculars. To mention but a few: in 1757, Dimitri Eustatievici wrote a Romanian grammar, Gramatica Rumaneasca; but this text would not be printed until well into the 20th century. Likewise, probably around the mid-18th century CE, Elî Teremaxî composed a Tesrîfa Kurmancî or ‘Kurdish morphology’ in Kurmanjî or Northern Kurdish. Written for young Kurdish-speaking medrese pupils taking their first steps in Arabic grammar, this work gained a wide circulation in the rural medreses of Northern Kurdistan, and, in fact, continued to be used clandestinely even after the rulers in the new republic of Turkey ordered the closing down of all medreses in the 1920s. In 1815 Vuk Karadzic wrote a grammar of his native dialect of Serbian, the Pismenica serbskoga jezika, at the request of his friend Jernej Kopitar; this work was to gain rather wider circulation in Northern European historical-comparative linguistic circles thanks to Jakob Grimm’s 1824 translation, the Kleine serbische Grammatik. In 1835, Neofit Rilski had a Bolgarska gramatika printed for use in schools trying to rid themselves of Greek linguistic and cultural dominance. And as late as 1851, Ahmed Cevdet Pasha published a Kavâ’id-i Osmaniyye ‘Principles of Ottoman [Turkish]’, which was to go through numerous printed editions in the following decades.

16It should be emphasised that these new vernacular philologies owe less to modern Western philological orientalism than to local classical traditions. Even in a relatively late work like Cevdet and Fuad Pasha’s textbook, the categories employed are those of traditional Arabic grammar, rather than of modern Western philology. Thus, in its treatment of the locative and ablative case, evidentials, and vowel harmony, the Kavâ’id-i Osmaniyye differs radically from A.L. Davids’s 1832 Grammar of the Turkish Language, which some modern scholars, mistakenly, have seen as a source of inspiration for Cevdet’s work.12 In short, a strong argument can be made that these vernacularising processes, and the emergence of new local vernacular philologies, preceded any influence or hegemony of modern Western orientalist philology.

17The historical and theoretical significance of these vernacular grammars has not yet been assessed. Here, however, I wish to suggest that they not only mark an important dimension of the vernacularisation of various Ottoman languages; they also embody a step in what one may call the governmentalisation of language, that is, in a process that simultaneously turned vernacular languages into objects of knowledge and objects of governmental concern. One of the main aspects of modern nationalism, after all, is that all subjects are to be turned into full-blooded citizens, and into loyal members of the nation, by universal education in a standardised, unified and codified version of what is called ‘the mother tongue’; and that the spread and implementation of this mother tongue through educational systems and institutions is one of the primary responsibilities of the new institution of the nation state. The history of modern nationalism, that is, is also a history of how vernacular languages—or new forms of language much closer to spoken dialects—simultaneously became instruments of mass communication, symbols of identification, and objects of government.

4.0. A Sonderweg for Arabic?

18At first blush, Arabic seems to form the most important, if not virtually the sole, significant exception to this empire-wide process of vernacularisation. Although dialectal or colloquial traces appear in various Arabic-languages of different ages, no authors openly proclaim or propagate either the written use of vernacular forms of Arabic, or the modernisation or purification of the Arabic language prior to the nahda, or literary Renaissance movement, that emerged in the mid-19th century. But perhaps we simply have not looked closely enough, or have been misled to some extent by the pervasive linguistic ideologies concerning the unity and uniqueness of Arabic.

19Considerations of space, and lack of relevant expertise, prevent me from pursuing these questions in greater detail; but here, I would merely like to suggest that the study of Ottoman Arabic may be enriched by a more systematic contextualisation: we can, and perhaps should, ask whether and how the structure, use, and ideologies of Arabic were affected by developing institutions and practices of government, and compare and contrast the development of Arabic with that of other languages in the Ottoman Empire. To take but one example, one may think of so-called ‘Middle Arabic’ typologically as a specific style or register of Arabic between the normative ideal of Classical Arabic and the colloquial realities of local dialects, rather than historically, as a developmental stage or period as was done by earlier scholars.13 In doing so, however, we may come to see the similarities and divergences between Arabic and other Ottoman languages in a rather different light. As discussed above, speakers of several other Ottoman languages also developed supraregional forms that consciously differed from, and acted as intermediaries between, on the one hand, the classical norm and, on the other hand, the regional, ‘vulgar’ dialects. The modern Greeks developed Katharevousa; among Ottoman Armenians, a supraregional ‘civil language’ (K’aghak’akan) emerged; and Serbian exiles produced an educated Slaveno-Serbian. Only Greek and Arabic, however, retained an enduring diglossia, whereas both Civil Armenian and Slaveno-Serbian disappeared in the 19th century.

20There was nothing inevitable about these outcomes. Prior to the 1815 publication of Karadzic’s Pismenica serbskogo jezika, several grammars of Slaveno-Serbian had been written and printed; in fact, Karadzic’s own grammar has been shown to be a calque of one of these grammars, which simply replaced Slaveno-Serbian items and paradigms with dialectal ones.14 And Nichanian (1989) describes how a substantial literature (both translated and original) in Civil Armenian had circulated before being replaced by a variety closer to the dialects spoken in the Empire, called ‘Western Armenian’. Thus, even if the process of vernacularisation occurred throughout the Ottoman Empire, its outcomes varied widely across different languages.

21The brings up the substantial question why only Greek and Arabic retained a relatively stable diglossic constellation, whereas languages that emerged from broadly similar backgrounds, like Armenian and the Slavic languages, did not.

22One crucial factor appears to have been the role of language ideologies: among Greeks and Arabs alike, the belief that, despite all dialectal differences and diachronic developments, their language—like their nation—was and should remain a unitary and unified entity appears to have predominated, and to have created the preconditions for relatively stable and enduring—if by no means uncontested—diglossia. Among Armenians, by contrast, the language-ideological belief that a modern language should be closer to the dialects of ‘the people’ appears to have carried the day. Finally, Slavic languages, and apparently also the various forms of Neo-Aramaic, appear to have been shaped by what has been called ‘fractal logic’ (cf. Gal 2005), which leads to ever-greater linguistic differentiation alongside the proliferation of new ethnic or sectarian antagonisms. In the mid-19th century, attempts at creating a unified ‘Serbo-Croat’ language seemed to be successful, but the two main varieties continued to be written in different scripts; and since the wars of the 1990s, efforts to emphasise the linguistic differences—not only between Serbian and Croat, but also with Bosnian and Montenegran—have been further stepped up. Another South Slavic vernacular, Bulgarian, appears to have followed a similar fractal logic: it came to be seen, and used, as a distinct Slavic language only in the later 18th century, and by the turn of the 20th, a movement had emerged that claimed ‘Macedonian, ’ which hitherto had been classified as ‘Western Bulgarian, ’ as a language in its own right; and the fractalising process may not have ended there. Similarly, in Northern Iraq, among Eastern Christians of different denominations, a bewildering variety of modern and not-so-modern standards of written Modern Aramaic has emerged, without any one variety gaining a wider currency.15

23In short, common processes of vernacularisation have had very different results in different languages, depending in part on linguistic ideologies, on ethnic and sectarian relations, and on vernacular philologies. Most, if not all, of these outcomes, it should be noted, had already been more or less decided (though by no means completed) by the end of World War I, that is, prior to the formation of the Ottoman successor states and the imposition and permeation of national languages through educational institutions and mass media. Thus, they were not dependent on, or decided by, sovereign state power; hence, it may be useful to study Ottoman processes and patterns of vernacularisation neither in purely linguistic terms nor in terms of sovereign state power, but with an eye to the development of vernacular philological traditions as a crucial factor in linguistic governmentalisation.