Introduction
This volume of Kurdish Studies is dedicated to research on the Kurdish language. While language issues have always been integral to academic and popular efforts at staking out Kurdish identities, much of the relevant linguistic research either does not feed into the broader discourse, or is simplified or distorted in various ways. We therefore welcome the initiative of the journal editors in accepting a cross-section of current research in Kurdish linguistics, reflecting fields as diverse as language contact and relevance theory in pragmatics, as a modest attempt at embedding Kurdish linguistics more firmly into the larger context of Kurdish studies.
Although the term “language” is a deceptively familiar item in most people’s daily vocabulary, scientific approaches to “language” tend to highlight different facets, yielding a multiplicity of varied conceptualisations: language can be seen as a complex, self-organising semiotic system, as the repository of cultural memory, an emblem of group identity, or a biologically endowed instinct that triggers acquisition in early childhood, to name but a few. Our perspective, as linguists, is primarily in terms of language as a complex, self-organising system, but we are sympathetic to an approach that sees the linguistic system as embedded in a social matrix, of which it is both product, and producer. In this introductory essay, we will take up both systemic linguistic aspects as well as social ones in an attempt to develop a reasonably coherent account of what constitutes “Kurdish”.
Attempting a definition of Kurdish/the Kurdish language(s) is an undertaking beset with controversy. Our aim in the first part of this introductory chapter is therefore to clarify certain conceptual matters concerning the notions of “language” and “dialect”, before tackling some of the thornier issues that have figured in discourse related to Kurdish, or “the Kurdish language”. We then review some recent approaches to classifying Kurdish, before presenting some proposals of our own. We suggest that a good deal of the confusion arises through a failure to distinguish different kinds of linguistic evidence (in a narrow sense), from the results of socially contracted and negotiated perceptions of identity, rooted in shared belief systems and perceptions of a common history. We do not argue for the precedence of any particular kind of evidence in defining a language and its speech community; on the contrary, a language is always at the nexus of a social construct with a set of linguistic facts. What we emphasise, however, is the analytical importance of distinguishing the findings of different fields. In the second part of the introduction, we present an overview of recent trends in Kurdish linguistics, though we make no claims to exhaustive coverage. Instead, we discuss what we consider to have been the most salient trends, based on a selective cross-section of the literature, and attempt to identify some of the most pressing research desiderata. Our treatment focuses on more recent developments; we refer to Haig and Matras (2002) for a summary of earlier research. Finally, in the third section, we summarise the other contributions to this issue and how they relate to the broader themes identified in this introduction.
On Defining Kurdish
There are two principal inter-related issues in defining Kurdish: (i) which criteria define its current scope, and which speech varieties should be included under the label “Kurdish”; (ii) what is its historical descent, i.e. from which proto-language(s) are the Kurdish varieties considered to have descended? Though related, these issues are logically distinct. In practice, the first issue relates mainly to the question whether Zazaki and Gorani (along with a few other varieties like Laki and sections of what is generally called Luri) are part of Kurdish, or independent languages. The second issue relates, on the one hand, to whether a common ancestor can be postulated for all the varieties to be considered under Kurdish, and, on the other hand, to the relationship of this putative common ancestral Kurdish to a Middle or an Old Iranian language. We begin with a discussion of the conceptual and methodological dimensions entailed in these two broad issues.
Dialect vs. Language: Conceptual Issues
In spite of their apparent simplicity in daily usage, distinguishing between “language” and “dialect” in a technical linguistic sense is very difficult, since linguistic factors are inextricably entwined with sociological, political and ethnic ones. The most widespread diagnostic has been that of “mutual intelligibility”, according to which varieties that are mutually intelligible are defined as dialects of the same language, whereas varieties that are not mutually understandable are assigned to distinct languages. Intuitive as it may seem, it is far from being a reliable diagnostic in several respects. In methodological terms, it is the speakers of the varieties who understand each other or not, not the varieties themselves. Thus speaker attitudes may weigh more than “objective” measures of linguistic similarities/differences. It has also been pointed out many times that this notion fails to account for chains of mutually intelligible dialects (dialect continua), where speakers of contiguous varieties may understand each other, but those at either end of the continua cannot. Where, then, does one language stop and next begin? Furthermore, intelligibility is often a dynamic process: on initial exposure to a different variety, one may understand very little, but within a short space of time, intelligibility in one, or both, directions, may increase dramatically. Again, this poses considerable methodological problems when attempting to assess mutual intelligibility.
In practice, there are numerous examples worldwide where the criterion is simply ignored when languages are being defined. For instance, Swedish and Norwegian are mostly mutually intelligible, both in spoken and written forms, but they are standardly considered to be separate languages. On the reverse side, mutually unintelligible varieties of Mandarin (Standard Chinese) and Cantonese (the variety of Hong Kong, Macau, and Guangdong province), together with five other major varieties, are considered to be dialects of Chinese (cf. Wang, 1997: 55). In these and many other situations around the world (e.g. Urdu-Hindi, Croatian-Bosnian-Serbian, Moldavian-Romanian), what is a “language” and what is a “dialect” are determined on social and political grounds. More recently attempts have been made to relate measurable linguistic differences (in the form of quantified phonetic differences, known as Levenshtein distances) to degrees of mutual intelligibility, undertaken by linguists working at the University of Groningen (see e.g. Gooskens, 2007). The research has been conducted with speakers of Germanic languages such as Danish, Dutch and Swedish, working in the contexts of written, standardised languages in industrialised societies. It is difficult to see how this research paradigm, promising though it may be, can be readily transferred to the Kurdish context. In short, the oft-cited and supposedly “objective” criterion of mutual intelligibility has to date been of little value in distinguishing languages and dialects.
Recognising these shortcomings, a number of scholars have tried to go beyond mutual intelligibility towards more socially-informed definitions of language and dialect. Crystal (1997: 248) has added “common/different cultural history” to “mutual (un)intelligibility”. In a situation where these two criteria do not match, it is considered not possible to decide on the status of the given varieties. Trudgill’s (2000) distinction of “autonomous” vs. “heteronomous” varieties addresses an important insight: an autonomous variety would be an independent code, recognised as such for purposes of media and education without necessary reference to an over-arching variety. A heteronomous variety on the other hand is perceived as a variant of some autonomous code. The intuition here is that when we use the term dialect, there is always the sense of “dialect of X”, with X being some independently recognised linguistic unit of a larger order. While this distinction is moderately useful in the context of languages with state-sanctioned status, it is of restricted relevance for the cluster of varieties that constitute Kurdish, and many other languages with restricted official status.
It is thus a sobering fact that to this day, the science of linguistics has nothing to offer in terms of an operative definition of “language” (cf. Fromkin et al., 2003: 446; Trudgill, 2000: 3–5). Within mainstream generative linguistics, one response to this state of affairs has been to deny the relevance of “a language” entirely, and to focus solely on the abstract linguistic abilities of the individual speaker. Under this conception, linguistics “becomes part of psychology, ultimately biology”, and the notion of “a language” (which implies a community of speakers sharing that language) is simply sidestepped (see Chomsky, 1986: 30–32 for justification of this line of argument). Sociolinguists, however, for whom the notion of speech community, or communities of practice, is indispensable, (e.g. Fasold, 2005; Romaine, 2001), conclude that language and dialect are fundamentally social, and not linguistic constructs: “a language is a language if it has been so socially constructed” (Fasold, 2005: 698). The view that a language is in some sense a tangible, homogenous entity with a more or less fixed form, most suggestively fostered in the case of large national languages, is likewise untenable. As Linell (2005: 45) puts it:
there is no single system of spoken language corresponding to the idea of a unitary national language; instead there are overlapping regional and social varieties as well as partially specific languages tied to communicative activities and genres. The notion of a unitary national language is an artificial social reality attempted at as the result of political actions, including linguists’ standardising efforts.
Let us consider how some of these issues would relate to the Kurdish case. It should by now be evident that there is no simple answer to the question of whether Kurdish is “one language”, and if so, which varieties should belong to it. The criterion of mutual intelligibility is fundamentally flawed, as we have pointed out above, and would almost certainly yield contradictory results if applied to Kurdish. In fact, in the absence of a generally accepted definition of language, the question of whether Kurdish is a language is vacuous. We can, however, meaningfully investigate what speakers of the varieties concerned perceive about their own variety in relation to others. In this case, there seems to be a relatively broad consensus among speakers of Sorani/Central Kurdish, and speakers of Kurmanji/Northern Kurdish that their respective varieties can be identified with a larger-order entity Kurdish/Kurdî. Similar perceptions may hold for speakers of Southern Kurdish (Fattah, 2000), and for some varieties of Gorani (see below). We can then proceed to explore the histories, sources and variations in these self-perceptions: where do they originate, how have they shifted, how they correlate (or not) with other principles of social organisation (tribal, religious, means of subsistence, political). A perception of “Kurdish” is in fact historically well-attested in the Sharafname of Sharaf Khan Bidlisi, who some 420 years ago was developing a definition of Kurdish together with a detailed classification of its dialects on ethnic/cultural grounds. Reference to Kurds, and to Kurdistan, feature regularly in the sources of the Ottoman era; it is evident that a notion of group identity that preceded the modern era was well established, but the precise extension of that group identity is difficult to establish in retrospect.
In practice, there is relatively little controversy with regard to Sorani/Central Kurdish, and Kurmanji/Northern Kurdish. The litmus test for approaches to Kurdish are Zazaki, and Gorani. In the case of Zazaki, the discussion has become regrettably politically charged, and the linguistic arguments are regularly instrumentalised by different political factions, rendering rational debate increasingly difficult. The case is, however, extremely instructive, and therefore worth dwelling on in some detail here (below we take up the arguments from historical linguistics on Zazaki). Reliable information on the historical self-perceptions of Zaza speakers is hard to come by. Evliya Çelebi, the celebrated Ottoman traveller, in his account of his journey from Bitlis to Van, refers to the “Zaza” among a list of Kurdish tribes grazing the alpine pastures of the Suphan Mountain north of Lake Van (Bulut 1997: 221). In this part of his travelogue, Çelebi regularly refers loosely to “Kurdistan” and the “Kurdish people” (qavm-i ekrad), but has otherwise little to say of the Zazas. According to Özoğlu (2004: 34–35), in general Çelebi, like the historian Sharaf Khan mentioned above, treats the Zazas on a par with other tribal (and dialectal) groups among the Kurds, concluding that “since Evliya gathered his information among local sources from the region, one can conclude that the Zaza speakers were considered Kurds by Evliya’s sources”. In more recent times, when Kurds mobilised as a political entity as in the Sheikh Said uprising of 1925, Zaza and Kurmanji speakers were collectively implicated (and the leader of the revolt himself was a Zaza speaker). Finally, it is worth recalling that one of the most widespread traditional self-designations for the Zazas and their language is Kirmanj/Kirmanjkî We can reasonably assume that this is the same word as Kurmanj/Kurmanjî; the difference in the quality of the first vowel is minimal (the short centralised vowels are frequently interchangeable in a number of words), and the suffixes -kî and -î are the regular equivalents of each other in Zazaki and Kurmanji respectively. If this is the case, we can assume a common self-designation for both groups, possibly in the sense of a generic term for people associated with particular kinds of livelihoods, rather than terms targeting ethnic or linguistic identities (see Asatrian, 2009: 28–30 for a discussion of the term “Kurmanj”). Thus although we are far from anything approaching a reliable ethno-linguistic characterisation of premodern identity perception among the Zazas, and although there are undoubtedly considerable local wrinkles that more general statement would fail to capture, there is certainly good evidence for an inclusive perception of “Kurds” which generally subsumed the Zazas.
At the turn of the 20th century, western philologists began to analyse the structure of Zazaki. The most influential scholar in this respect was Oskar Mann. Mann pointed out a number of phonological and morphological differences between Zazaki and Kurmanji, which led him to the claim that the two should be considered quite distinct languages. On his view, Zaza did not belong to Kurdish (cf. discussion in Mann and Hadank, 1932: 19–23, and below). Mann’s views were entirely based on linguistic/philological facts; they actually entail no consequences in terms of speakers’ perceived identities, and initially, the discussion on the position of Zazaki was largely confined to Iranian philology. The speakers themselves were unaware of these evaluations of their languages up until the 1980’s, when Kurdish intellectuals in the diaspora came to be informed about the discussions. Subsequently, a small number of Zazaki-speaking exile intellectuals, applying a positivistic notion to identifications and classifications, began to adopt the idea. As a result, some Zazaki-speaking intellectuals who had previously referred to Zazaki as part of Kurdish, and the Zazas as Kurds, rephrased their discourse in favour of an exclusivist Zazaki language and identity distinct from Kurdish.1 Given the many political stakeholders in the contested arena of Kurdish identities, it is hardly surprising that this debate has long since left the purely academic domain, and the originally purely philological arguments have been entirely misconstrued and instrumentalised by different parties. Our point here is that the arguments from philology, which we take up below, were never intended as statements regarding perceived identities, and are in fact largely orthogonal to that debate.
Reconstruction of Language History
The analysis of the more or less “hard” linguistic facts – the lexicon, the phonology, the morphology – can yield a basis for classification of languages, involving the reconstruction of language history (commonly known as “comparative method”). The comparative method works on the comparison of phonologies in a set of languages, and seeks to identify systematic correspondences as evidence of shared history. In addition, features of morphology and syntax may also be compared which again can yield clues regarding innovations shared among the varieties investigated. The underlying assumption is that certain kinds of change are regular, and the systematic comparison allows one to identify changes shared in some varieties, but absent in others. From this, the analyst reconstructs the most plausible set of historical movements – splits among the investigated language group – that would have yielded the available picture. Thus we can arrive at a relative measure for degree of relatedness among the varieties concerned: the more closely related two varieties are, the shorter the time span that has elapsed between now and the point in time at which the two varieties split from their common source. However, the success of this method depends in part on the time depth of attestation of the languages concerned. Note that this method works solely with observable linguistic data, with little or no relation to the social conditions of the speech communities themselves. Thus the results of the comparative method cannot simply be interpreted as evidence for defining “language” or “dialect” (and indeed, nothing in the comparative method is contingent on these concepts). Of course we would expect a degree of correlation between socially determined identity perception, and degree of linguistic relatedness: a group sharing a close-knit social structure and a common identity is unlikely to speak varieties that are only distantly related, and conversely, we would not expect groups speaking only very distantly related varieties to share a common group identity; but these are only rough tendencies, and there are numerous examples worldwide where degree of social identity of the speakers does not match linguistic relatedness of their speech. For example, many Black African Americans legitimately claim a group identity distinct from, for example, the descendants of the British invaders of North America. Linguistically, however, it is a simple fact that today, members of both groups speak closely – related varieties of a Germanic language. The issue of language origins is logically distinct from the issue of perceived group identity, and should not be confused with it, though in many cases the two do go hand in hand.
Although the classification provided by historical linguistics purports to be “objective”, we should note that it is far from secure, and the available evidence leaves considerable scope for interpretation. Let us briefly consider some evidence from phonology, as it relates to the position of Zazaki vis-à-vis Kurdish. In Zazaki, words such as werd- “eat”, or weš “good, fine” exhibit an initial [w-], whereas Kurmanji and Sorani have the velar fricative [x-] (with additional labialisation in some varieties). In this feature, Zazaki patterns with Hawrami/Gorani (see below). However, although this piece of evidence points towards a different path of development in Zazaki as opposed to Sorani and Kurmanji, Gippert (1996) concludes that on the whole, the evidence from phonology does not suffice to yield a conclusive picture regarding the position and origins of Zazaki.
Potentially more revealing is the evidence from morphology. Perhaps the most salient difference between Zazaki and the rest of Kurdish is the formation of the present indicative. In Zazaki, it is formed with an infixal augment, containing the nasal [-n], which attaches to the present stem: we-n-o (eat-PRES.AUGM-3S.MASC) “he eats”, (cf. Paul, 1998: 74–76 for details). In Kurmanji and Sorani, on the other hand, no stem-final augment is possible in the present tense. Instead, present stems are preceded by certain prefixes, for example di- in most of Kurmanji, as in di-xw-e (IND-eat:PRS-3SG) “s/he eats”, de- or a- in Sorani, and so on. It is generally assumed that the stem-augment of Zazaki goes back to an old participial form, but there is no reflex of this participle in either Sorani or Kurmanji. However, close parallels are found in West Iranian languages of the Caspian region, for example Mazanderani (Haig, in print), or Semnani (Gippert, 1996). These facts, taken at face value, speak for a Caspian origin of Zazaki. However, the interpretation remains controversial; Gippert (1996), and following him Jügel (this volume) suggest that the present-stem formation based on a participle may be a “recent” loan influence, citing North East Neo-Aramaic and East Armenian as possible sources. In the case of North-East Neo-Aramaic, this does not seem particularly plausible, as it is by no means clear where, or when, Zazaki would have been exposed to heavy influence from any variety of this language. East Armenian is a much more likely contender. But why should all varieties of Zazaki, regardless of their geographic setting, have undergone a contact-induced development of this type, whereas none of the surrounding varieties of Kurmanji did, although they were exposed to Armenian and Neo-Aramaic influence? Finally, the simple fact remains that in order to develop a participial-based present stem formation at all, the original participial-forming morphology must have been retained in Zazaki. It is the retention of this morphology in Zazaki that sets it off from the rest of Kurdish, and this can hardly be explained through recent contact influence. Rather, it would seem more probable that precisely this point distinguished the ancestor of Zazaki from the ancestors of the other varieties considered to constitute Kurdish, and the development of the participial-based present tense (and the extension of the nasal augment to the past imperfective) are early innovations, hence found in all varieties of Zazaki. Contact influence is certainly possible, but its source is more likely to be found in the formative stages of Zazaki, possibly prior to its spread to its current location.
It is often suggested that Zazaki is in fact more closely related to Gorani than to either Kurmanji or Sorani. The most prominent advocate of this view was Oskar Mann, who went as far as to claim “near identity” of Zazaki and Gorani (Mann and Hadank, 1932: 25). Hadank himself, however, who was entrusted with posthumously preparing Mann’s work for publication, had already pointed out that his predecessor’s assessment was exaggerated (Mann and Hadank, 1932: 25–26). Above we noted the presence of an initial [w-] in words such as “eat” and “read/study” in Zazaki, which is also shared in Gorani. However, it is not an exclusive factor uniting Zazaki and Gorani; it is also found, for example, in Balochi (Korn, 2005: 122). It is therefore fairly thin evidence on which to base a Zazaki/Gorani group, as is often assumed (see below). Indeed, although we still await a detailed systematic comparison of Gorani and Zazaki (a surprising desideratum), we are currently unaware of a truly convincing historical demonstration of the viability of this subgrouping. It is true that each display features common to other languages outside of Kurdish, and not shared by Sorani and Kurmanji. Zazaki, for example, shows obvious parallels to Iranian language of the Caspian region. But from this it does not follow that they can be meaningfully grouped together within a putative historical group of “Kurdish”. It simply follows that historically, both need to be set off from Sorani and Kurmanji.
The concept of “Southern Kurdish” also raises certain difficulties. One issue concerns where the borders of Southern Kurdish in relation to geographically contiguous West Iranian languages, such as Luri, should be drawn. Anonby (2003), for example, suggests that Luri is part of a language continuum spanning northwest Iranian Kurdish, and southwest Iranian Persian. Such a statement is difficult to reconcile with the traditional view of a northwest vs. southwest Iranian split, and would essentially dissolve Southern Kurdish as a viable genetic group.2 The question of “Southern Kurdish” has been most extensively treated in Fattah (2000), who defends the coherence of the group. Fattah recognises a group which he refers to as kurde du sud, which covers most of what is traditionally included under “Southern Kurdish”, but he subsumes this group under a larger group of dialectes kurdes méridionaux. Within the latter, Fattah includes Laki, but considers it distinct from kurde du sud. The most salient linguistic feature distinguishing Laki from kurde du sud is the presence of ergativity in past tenses in Laki (Fattah, 2000: 61). The position of Luri within this scheme is discussed at some length (Fattah, 2000: 40–55), with the general conclusion that it lies outside of the dialectes kurdes méridionaux, although within Luri itself there are internal divisions that remain problematic for any straightforward classification in terms of Kurdish/non-Kurdish.
Within all classifications known to us, Kurmanji and Sorani are uncontroversially defined as belonging to “Kurdish”. The question of their relationship to each other has, however, seldom been explicitly discussed (see Jügel (this volume) for references). In Haig and Öpengin (forthcoming), it is noted that although there are intermediate varieties exhibiting typical properties of both (e.g. Surči, discussed in MacKenzie, 1961), in general the boundary between Sorani and Kurmanji is relatively clearly delineated. There is a long list of distinctive morphological features that distinguish them (cf. Haig and Öpengin, forthcoming), and for any given variety, it is generally not difficult to assign it to either Sorani or Kurmanji. This fact is rather surprising; it is certainly not what one would expect if Sorani and Kurmanji had originated from a single source, and then gradually spread into their current localities. Such a scenario would have yielded a dialect continuum, with each variety gradually shading into the next. Instead, we find two large and relatively distinct speech zones, sharing a fairly narrow ribbon of overlap in which there are varieties exhibiting the typical features of both. What this suggests is that Sorani and Kurmanji evolved in geographically distinct regions, and later came into contact. This is not to deny the obvious relationship between the two, but it suggests that we require a more sophisticated account of the genesis of these two varieties than is currently available.
In sum, historical linguistics can help unravel the relationships between related languages (or indeed demonstrate their relatedness in the first place). But we would urge caution in interpreting the findings, which are seldom as clearcut as the family trees that are traditionally used to represent them (see Jügel, this volume). But again, the problems here are not restricted to Kurdish, but are endemic to historical linguistics, regardless of the language family concerned. For example, in Oceanic linguistics the term “linkage” is regularly used to refer to geographically contiguous groups of related languages which exhibit certain similarities, but cannot be reliably traced to a common ancestor, hence would not be representable on a traditional family tree (Lynch et al., 2011). We have briefly investigated some of the problems associated with Zazaki and Gorani in this respect, noting that from the perspective of historical linguistics, there is no doubt that these two are less closely related to Sorani and Kurmanji. However, we also note that this does not imply that Zazaki and Gorani ever formed a historical unit. But perhaps the most important point is that the comparative method yields a hypothesis regarding the ancient history of the languages; its results cannot be simply translated into claims regarding social identity constructions of the speakers.
Existent Approaches to Classification of Kurdish
Having discussed two major approaches to identifying language, we will briefly review some existing classifications of Kurdish. It is noteworthy that most classifications are not explicitly justified, but draw on a mix of geographic, socio- historical, ethnic and linguistic criteria. This is particularly true of Hassanpour (1992) and Izady (1992). The classification of Fattah (2000) is based on a more detailed discussion of linguistic and extra-linguistic criteria, with a focus on Southern Kurdish. Consider first the classifications of Hassanpour and Izady:
Even these broad classifications differ in a number of respects, some merely terminological but others more substantial. For example, in Hassanpour’s classification the term Hawrami, which is a highly specific regional variety in Iranian and Iraqi Kurdistan (see MacKenzie, 1961, Mahmoudveysi et al., 2011: 2–4), is intended as a general term encompassing both Gorani and Zazaki. Given the linguistic (see above) and geographical distance between the two varieties, it is at best odd to subsume one of the varieties under the other. Izady’s classification involves an initial division into two major groups, “Kurmanji” vs. “Pahlawani”. The latter name might be intended to reflect the claim that both Zaza and Gurani are often considered rather archaic in the sense that they have preserved certain features found in Parthian. But it is not clear if this is intended, nor is it clear why Laki should be included in this putative grouping. It will be seen that in Izady’s classification, there is no obvious equivalent to Hassanpour’s “Kirmashani”. Of course both classifications were intended as rough guides, and make no claim to exacting scientific rigour. But even this perfunctory comparison suffices to reveal a number of contradictions and unresolved issues.
A classification based on extensive fieldwork and with more detailed justification is proposed by Fattah (2000):
In particular, Fattah discusses in considerable detail the nature of “Southern Kurdish” in relation to Laki, and also to Gurani. His suggestion of a Kurdo- Caspian group is developed at some length (2000: 62–70), and it is worth dwelling briefly on it here. Fattah concedes to the majority view of Iranian philologists, who concur on assigning Zazaki and Gorani to a peripheral status vis-à-vis the rest of Kurdish, but argues that this does not necessarily imply their exclusion from Kurdish as a socio-linguistic unit (Fattah, 2000: 65):
En admettant l’hypothese de l’origine caspienne des Gurân, qui remonterait à des époques très anciennes, et de leur installation, ensuite, d’une part dans la grande région de Kirmânshah-Hamadân, et d’autre part, pour une partie d’entre eux tels que les Zâzâ, vers l’ouest dans le Kurdistan de Turquie, celle-ci n’est pas forcément en contradiction avec leur appartenance et leur implication dans les processus de formation du peuple kurde, ou du moins leur fusion trés ancienne avec eux, et leur participation dans la constitution de l’identité politique, culturelle et sociologique du peuple kurde. [footnote omitted]
This is a view we would generally comply with, though as linguists, we have some reservations regarding the postulation of a distinct sub-group for Zaza and Gurani, for the reasons outlined in the preceding section. Our own, more cautious approach would be the following, which avoids sub-grouping Zazaki and Gorani (at least until positive evidence in favour of such a move is forthcoming) and identifies the five groups, whose approximate localities are shown in Figure 1:
Northern Kurdish (Kurmanji): It is often divided into Badini (spoken principally in Duhok and Hakkari provinces) and Kurmanji (in the rest of Northern Kurdish speech zone) varieties; both include a number of other regional dialects (see Öpengin and Haig, this volume).
Central Kurdish (Sorani): Its main regional dialects are Mukri (Mahabad), Hewlêrî (Erbil), Silêmanî (Suleimaniya), Germiyanî (Kirkuk) and Sineyî (Sanandaj).
Southern Kurdish: It includes the varieties such as Kelhuri, Feyli, Kirmashani, as well as some dialects of what is called Laki, in Ilam and Kermanshah provinces of Iraq and the town of Khaneqin in Iraq (see above on Fattah’s classification).
Gorani: It covers what is known as Hawrami or Hawramani, with the well-known dialects of e.g. Paveh and Halabja, and includes the old transdialectal literary koine, the language of religious rites among some Yaresan groups. In this sense, “Gorani” would include several varieties spoken in present-day Iraq, e.g. Bajalani. (cf. Fattah, 2000: 62–70, and Mahmoudveysi et al., 2012 for discussion of “Gorani”).
Zazaki: Its three main dialects are Northern Zazaki (Tunceli-Erzincan provinces), Central Zazaki (Bingöl-Diyarbakir provinces) and South Zazaki (Diyarbakir province and Siverek town).
Map of language varieties spoken by the Kurds
From Öpengin, 2013
In sum, there is no consensus in the literature when it comes to defining, and classifying “Kurdish”. Again, this is not a particularly surprising, nor by any means unusual state of affairs. A closer look at most of the supposed clear cases of “languages” on a global scale yields a similar picture. The exercise is nevertheless valuable as it serves to highlight precisely those areas where conflicting approaches yield methodological and conceptual challenges. In our view, the crucial point is to avoid conflating the results from distinct methodologies, and to explicitly recognise the limitations of any kind of static taxonomy.
We certainly acknowledge, however, that such classifications are not merely academic exercises, but have quite concrete repercussions. Consider for example the online discussion on whether Zazaki should have an independent entry in Wikipedia, as opposed to a sub-entry under Kurdish.3 The arguments posted provide interesting insights into how ideological arguments impact on supposedly democratic forums. What should have emerged from this section, among other things, is that it is perfectly possible to accept both the conclusions of the historical linguists (Zazaki is historically not closely related to Kurmanji), and the conclusions of many native speakers (Zazaki speakers are Kurds, and their language belongs to a larger-order entity “Kurdish”). There is not necessarily any contradiction here.
Current Trends in Kurdish Linguistics
Taken in the broadest sense, linguistic research on Kurdish runs back several centuries, starting with the grammar of Ali Taramakhi (see Leezenberg, 2014) towards the end of 17th or 18th century and the grammar of Maurizio Garzoni in 1787. However, we are concerned here with work that has been undertaken in an academic context within the disciplines of Iranian philology, or general linguistics, and we consider only work undertaken since 2000. We also exclude the numerous studies conducted in the realm of standardisation, and the large number of pedagogical works on Kurdish that have appeared in the last decade. Within European and North American academia, Kurdish linguistics remains institutionally poorly represented; at best, Kurdish linguistics is an ancillary sub-discipline within another department, as in the department of General Linguistics at the University of Bamberg, or at the Department of Empirical Linguistics at the University of Frankfurt, or it is undertaken within the framework of comparative Iranian philology at the departments of Iranian Studies in Hamburg and Göttingen. The Kurdish Institute in Paris continues a long tradition of invaluable descriptive work on Kurdish, but their main research output is concentrated in political science and sociology. Somewhat paradoxically, Turkey now offers considerable scope for Kurdish linguistics. In particular, the Mardin Artuklu University has recently undertaken moves to establish a section on linguistics as part of its existing Kurdish programme. In Iran work on various varieties of Kurdish has been undertaken within Iranian studies, though much of it remains relatively inaccessible to scholars outside the country. In Iraqi Kurdistan, there are Kurdish departments at the many recently-grounded universities, offering exciting new prospects for international cooperation, though we are not in a position to sketch these developments here.
Despite some positive developments, research on Kurdish linguistics continues to be hampered by a lack of institutional support in domains such as post-graduate or PhD programmes, and a lack of secure teaching positions. Current research on the Kurdish language is thus largely carried out in a piecemeal fashion by individual researchers from a disjoint set of disciplines, without an overarching institutional framework. There is also a lack of dedicated journals or regular conferences that treat the topic. However, following the International Workshop on Variation and Change in Kurdish (August 2013, Bamberg), the organisers have decided to establish the event on a regular annual basis (in 2014 the conference is held at the Mardin Artuklu University).
Theoretical Approaches to Kurdish
Within the framework of what Culicover and Jackendoff (2005) refer to as Mainstream Generative Grammar, a number of studies have appeared on Kurdish syntax, starting with Fattah (1997), and work in the generative tradition is continued for Sorani in a number of papers by Karimi on the ezafe (2007) and agreement (2010), while Karimi-Doostan (2005) discusses complex predicates. For Kurmanji, ongoing work by Gündoğdu (2011, forthcoming) continues the tradition of generative-inspired research on syntax. A different framework (Head-Driven Phrase Structure Grammar, HPSG) is utilised by Samvelian (2006, 2007) in her work on clitics in Sorani, while an alternative approach drawing on prosodic phonology is developed by Öpengin (2013). More theory-neutral approaches are adopted by Franco et al. (2013) on ezafe, Haig (2002) on complex predicates, and Öpengin (2012b) on adpositions and argument structure in Sorani. Pragmatics and relevance theory has been explored with reference to the Badini dialect of Kurdish extensively by Unger (2012, and this volume).
General Descriptive and Historical Studies
Apart from the numerous pedagogical works that have been produced in the past decade (not discussed here), two interesting grammars (Sorani and Kurmanji respectively) have been published on the internet by Thackston (not dated). However, undue reliance on the model of Persian has led to some distortions, particularly in the Kurmanji grammar, which needs to be treated with caution. A short grammatical sketch, intended primarily for linguists, is Aygen (2007). Regrettably, it contains numerous factual and analytical errors, and appears to be based on data gathered from a single native speaker, with some supplementary explanations from Bedir Khan and Lescot (1970); we mention it solely for the sake of completeness, but definitely cannot recommend it. A lengthy overview of “Kurdish” by McCarus (2009) is useful, but treats almost exclusively Sorani. Fattah (2000) is a monumental study of “Southern Kurdish”, though somewhat difficult to navigate through. Relevant to the concept of “Southern Kurdish” are the articles by Anonby (2003 and 2004/2005), which discuss the place of Luri and Laki. An overview of Kurmanji in Turkey is Haig and Öpengin (forthcoming), and an initial classification of Kurmanji dialects is now available in Öpengin and Haig (this volume). Haig and Öpengin (in print) provide an updated synthesis of structural and sociocultural aspects of gender in Kurdish. One area where Kurdish linguistics has benefited from recent developments in general linguistics is research on endangered and under-studied languages. Within the framework of a Volkswagen-Foundation funded project, two dialects of Gorani from West Iran were documented (see Mahmoudveysi et al., 2012, and Mahmoudveysi and Bailey, 2013). In a similar spirit, Öpengin (2013, forthcoming) provides the most comprehensive treatment of a dialect of Central Kurdish available to date in any European language, based on extensive original texts and combined with a theoretically informed discussion of person marking in this variety. Published research on the first-language acquisition of ergativity in Kurmanji is now available in Mahalingappa (2013), while Mohamad (2014) investigates Kurdish-German code-switching among pre-school children in Austria.
The history of Kurdish syntax, with particular reference to alignment, is discussed in Haig (2004a, 2008 and forthcoming b) within the context of West Iranian. An overview of grammatical changes is provided in Jügel (this volume). For Zazaki, since the comprehensive grammars of Paul (1998) and Selcan (1998), very little substantial linguistic research has appeared. An overview of linguistic and socio-cultural aspects of the Zaza is provided in Keskin (2010), and Paul (2009). Our overall impression is that within linguistics, Zazaki remains sorely underrepresented, both in terms of descriptive and more theoretically oriented research.
Kurdish in Contact with Other Languages
Kurdish is nested in a complex multi-lingual context, and the effects of language contact continue to be a key topic in Kurdish linguistics. Following the pioneering work by Dorleijn (1996) on ergativity, a number of topics have been investigated. The vowel system is studied in Özsoy and Türkyilmaz (2006) and evidentiality in Bulut (2000). A perspective on Kurdish as part of a putative Anatolian linguistic area is Haig (2001, 2006, 2007, in print, forthcoming a), Matras (2002, 2007, 2010). While the impact of Turkish, Persian and Arabic on Kurdish has been emphasised in a number of publications, there is also growing interest in Kurdish influence on Neo-Aramaic (e.g. Khan, 2007, and Noorlander, this volume), and on vernacular varieties of Arabic (Talay, 2006/2007). These studies are important as they provide valuable linguistic evidence regarding the role of Kurdish as a lingua franca across large areas prior to the spread and increasing dominance of the national languages.
Sociolinguistics
The highly complex social and political dimensions in which the Kurdish language is evolving have been treated in a number of recent works. Of particular importance is the special issue of the International Journal of the Sociology of Language, in 2012, dedicated to various questions of Kurdish in the countries it is spoken. A number of publications by Salih Akin have created an important body of literature in French, of special interest are Akin (2000, 2002, 2004). Recently, some fieldwork-based sociolinguistic research has investigated the linguistic vitality of Kurdish in Turkey (Öpengin, 2012a, Çağlayan, 2014), the language attitudes of the speakers (Coşkun et al., 2013), but there has been surprisingly little work within variationist sociolinguistics. A large number of articles, on the other hand, have documented the language policy in Kurdistan, to name only a few Zeydanlıoğlu (2012), Sheyholislami (2012), Hassanpour (2012), Haig (2004b and 2012). Hassanpour (2001) has dealt with the gendered language in Sorani Kurdish, while Asadpour et al. (2012) discusses the address terms in the same variety.
As is evident from this short overview, Kurdish linguistics has continued to develop despite the paucity of institutional support. There is an increasing level of theoretical sophistication in much of the more recent work, carried by a growing number of highly-trained younger scholars, which bodes well for the future of the field. Nevertheless, certain areas remain somewhat under- represented, a couple of which we briefly outline here: within sociolinguistics, the dominant paradigm remains a more abstract socio-political one, drawing on identity, nationalism, and ethnicity theories rather than data-driven variationist approaches to sociolinguistics and multilingualism. With regard to the history of Kurdish, given the continued popular interest in the topic, we are surprised to note how little research is actually dedicated to disentangling the linguistic facts; as mentioned, one topic which obviously demands attention is the nature of the historical relationship between Zazaki and Gorani. Currently, most researchers are content to repeat opinions originally formulated 100 years ago, which are sorely in need of verification. Given the advances in historical linguistics over the last decades, we would welcome an application of more recent methodologies to these questions. Finally, there is still a need for well-grounded descriptive work on the numerous regional varieties within Kurdish, which would provide the raw material for a more comprehensive assessment of central issues regarding the nature of Kurdish as a linguistic and socio-cultural entity, and its historical evolution.
Synopsis of the Contributions
In his contribution, Thomas Jügel takes up the question of the relationship of Kurdish to the rest of the West Iranian languages and attempts to develop a relative chronology of the changes that led to Kurdish becoming a distinct group. As Jügel notes, reconstructing the history of Kurdish is hampered by the lack of reliable attestation beyond a couple of centuries. Thus any attempts to trace the history of Kurdish need to rely in part on inferences gleaned from the histories of better-attested, closely related languages, in this case, Parthian and Middle Persian. The other confounding factor in reconstructing the history of Kurdish is the effect of language contact. The state of any particular variety, at any given point in time, is the result of both its inheritance – those features which are simply the continuation of the earlier stages of that variety – and the influence of the languages with which it has been in contact. Kurdish is a particularly challenging case due to the multi-lingual environment in which it is traditionally spoken, and also to the mobility of its speakers over many centuries. Contact effects cannot be easily represented in traditional family trees, so Jügel sets up a modified model of historical relationships which attempts to synthesise both “vertical” inheritance of linguistic features and lateral effects of language contact. Jügel identifies a number of grammatical features that serve to distinguish among different varieties of Kurdish, including the presence versus absence of pronominal clitics, the presence versus absence of oblique case marking on nouns, or gender on nouns. On this basis, he proposes a relative chronology of grammatical changes that led to the current distribution, basing his findings on what is known of parallel developments in the better-attested Middle Iranian languages Parthian and Middle Persian. This is an important contribution as it demonstrates both the potentials, and the limitations, of historical reconstruction in the case of Kurdish.
Regional variation in Kurmanji is the topic of Öpengin and Haig’s contribution. While an awareness of regional variation and the ability to negotiate around it are part of informed native speakers’ competence, to date there is virtually no serious research dedicated to the topic. The authors propose a broad distinction into five regions, and then proceed to apply a combination of established methodologies, targeting the lexicon, the phonology and the morpho-syntax, with the intention of identifying the main linguistic features that serve to differentiate the various regions. Given the size of the region and lack of previous research, they emphasise the tentative nature of the proposed classification, but note that their findings are generally consonant with laypersons’ perceptions, and also reflect the rough geographic distribution of the varieties. The authors identify “Southeastern Kurmanji”, the variety of Hakkari province, and including the Badini variety of Iraqi Kurdistan, as the most distinctive, in that it possesses the largest number of features not shared by any other variety. In some respects, these properties can be related to Southeastern Kurmanji’s proximity to Sorani, but this cannot be the whole story, as some of the features concerned contrast sharply with Sorani. The authors also identify a Northwestern Kurmanji, e.g. of Elbistan, which displays a number of divergent features, many of which have scarcely been documented, to say nothing of being analysed. Along with presenting an initial framework for future research on regional variation in Kurmanji, the authors aim for a reconciliation between dialectal/philological research and more general work on Kurmanji, which has primarily focussed on issues of standardisation. Rather than seeing regional variation as an obstacle to standardisation, it can also be seen as the repository of rich linguistic resources, and a legitimate source for enriching the available register repertoire in Kurmanji.
In Christoph Unger’s contribution, a novel approach to the interpretation of the so-called “future tense” in Badini Kurmanji is adopted, based on Sperber and Wilson’s Relevance Theory (1995). Along with an analysis of the Badini Kurdish facts, the paper also proposes an explanation for the differences between Badini Kurdish and the more widely-known Botan-based “standard” variety of Kurdish with regard to tense and modality marking. Unger notes that the so-called future marker of Badini, dê, is not only used to indicate future time reference, but also for a number of modal nuances. The question is, can these apparently divergent semantic functions be related to each other in a coherent manner, or should we simply accept a disjoint list of temporal and modal meanings. Unger argues for a unified semantics of dê, framed in terms of procedural semantics: an underspecified set of conditions that guides the listener’s interpretation of a particular linguistic item, on the assumption that the listener will apply general default principles of relevance and efficiency in her interpretation. In the case of dê, a procedural semantic account involves the claim that the state of affairs modified by dê lies outside the range of shared verifiable experience, but the speaker nevertheless commits herself to their factuality. This would include future states of affairs which the speaker can reasonably assume to come about, but other kinds of events that are not verifiable but whose plausibility the speaker commits to. The particle dê in Badini contrasts with another modal particle, da (not present in other dialects of Kurmanji), and this opposition also serves to delineate the function of dê in Badini. The Badini system contrasts with most other dialects of Kurmanji, which allow the dê particle (or its variants wê, ê etc.) to combine with different kinds of verb forms, including past subjunctives. Unger relates these formal differences in combinability of dê with different verb forms to differences in the procedural semantics associated with the particle in Badini, and with its counterpart in the rest of Kurmanji. Thus what appears to be a minor, and often overlooked, distinction in the grammars of the different varieties can be plausibly related to distinct underlying semantics of the particles concerned.
The pervasive effects of the national languages Turkish, Persian and Arabic on Kurdish, both in the lexicon and the grammar, have been regularly noted in the literature, and are a recurrent target of disapproval in some circles. What is much less well-known is that Kurdish itself has had a deep impact on another language of the region, Neo-Aramaic. Kurdish influence on Neo-Aramaic is the topic of Paul Noorlander’s richly illustrated contribution, which draws on Matras’ functional-communicative theoretical framework for analysing language contact. Kurdish influence on Neo-Aramaic is most evident in the lexicon, but it runs much deeper than merely the borrowing of a large number of Kurdish words. It is generally agreed among Neo-Aramaicists that the emergence of ergativity in the past/perfective verb system of Neo-Aramaic largely follows a Kurdish model; the structural parallels are so striking, and the development is so unusual within Semitic that Kurdish influence, although not actually provable, can hardly be discounted. This is perhaps the most remarkable indication of the long-standing and intense contact between the two language communities. A number of syntactic parallels include patterns of clause combining, negation, and ezafe constructions, while in the realm of morphology we find the adoption of the Central Kurdish definiteness suffix in some Neo-Aramaic dialects, or the Kurdish comparative suffix for adjectives also being adopted in varieties of Neo-Aramaic. While the history of Kurdish has tended to be cast against the backdrop of the emergence and increasing domination of the national languages in the last century, Noorlander’s article is a timely reminder of the centuries of co-existence between Kurds and members of other stateless minorities in the region prior to the era of nation states, a history which left an indelible mark on both languages.
References
Fattah, M. (1997). A Generative Grammar of Kurdish. (Unpublished doctoral dissertation). University of Amsterdam, Holland.
Franco, L., Manzini, R. and Savoia, L. (2013). Some call it article, some ezafe: “Linkers” as “agreement”. Retrieved from http://www.academia.edu/4393956/Some_call_it_article_some_ezafe_Linkers_as_agreement (Accessed: 22 January, 2014).
Gündoğdu, S. (2011). The phrase structure of two Dialects of Kurmanji Kurdish: Standard Dialect and Muş Dialect. (Unpublished MA Thesis). Boğaziçi University, Turkey.
Gündoğdu, S. (forthcoming). Case and Ergativity in Kurmanji. Dilbilim Araştırmaları.
Haig, G. (2011). Linker, Relativizer, Nominalizer, Tense-particle: On the Ezafe in West Iranian. In F. Yap, K. Grunow-Hårsta, and J. Wrona (eds.), Nominalization in Asian Languages: Diachronic and Typological Perspectives Volume 1: Sino-Tibetan and Iranian Languages (363–390). Amsterdam: John Benjamins.
Öpengin, E. (2013). Clitic/affix interactions: a corpus-based study of the person marking in the Mukri variety of Central Kurdish. (Unpublished doctoral dissertation). Université Sorbonne Nouvelle and Universität Bamberg.
Thackston, W. M. (n.d.). Sorani Kurdish: A reference grammar with selected readings. Retrieved from: http://www.fas.harvard.edu/\~iranian/Sorani/sorani_1_grammar.pdf (Accessed: 21 November, 2013).
Thackston, W. M. (n.d.). Kurmanji Kurdish: A reference grammar with selected readings. Retrieved from: http://www.fas.harvard.edu/~iranian/Kurmanji/ (Accessed: 21 November, 2013).
It is worth noting that this shift (from Kurdish to an exclusivist Zaza(ki) identity) happened in a relatively short period of time, as illustrated in the case of Zülfü Selcan, the author of the book Grammatik der Zaza-Sprache (1998). The author was an activist in pro-Kurdish progressist circles at the turn of 1980s. In an article that he published in the Hêvî magazine, publication of the Paris Kurdish Institute, he freely uses the terms Kurd and Kurdish language to refer to Zaza people and Zazaki, considering the latter under “Kurdish”. This is apparent also in the title of the article, as “Folklorê Kurdî ebe zaravayê dimilkî”, which can be translated as “Kurdish folklore in its Dimilkî dialect”. Some years after, the author adopted the view of Zazaki as a separate language and the Zazas a distinct people from the Kurds. Recently, the author shared his 1983 article on his online academic profile page, however, in the new version of his article he replaced all of the occurrences of the terms “Dimilkî dialect” and “Kurdish” with “Zazaki language” and “Zaza people”, deleting quite a few other references to Kurdish including those in the title of the work. For the original version of the article see Zilfî (1983) and for the modified recent version of the same article see: http://www.tunceli.edu.tr/akademik/fakulteler/edebiyat/bolum/doded/zaded/EdebiyateZazay1.pdf (Accessed: November 25, 2012).
Anonby (2003) combines observations and interviews regarding levels of mutual intelligibility, with lexicostatistics. However, although both methods are of considerable interest in their own right, neither method will reliably yield genetic sub-groupings, and nor will a combination of the two. Anonby (2004/2005) applies more reliable methods to a classification of one variety of Laki.