Geoffrey L. J. Haig, June 2004
In 1995, when I ﬁrst began to learn Kurdish, my interest was captivated by the feature commonly referred to as ergativity in the past tense of transitive verbs. Although it was familiar to me in an abstract fashion from the linguistic literature, actually using a language with that particular feature is a very different matter. However, at a fairly early stage I came to the conclusion that ergativity in Kurdish was a largely superﬁcial phenomenon, something manifested in the morphology, but without apparent ramiﬁcations for the syntax.
The earlier stages of my thinking on the subject were summed up in Haig (1998). On the analysis embodied in that paper, Kurdish syntax wound up looking very much like that of its close relative, Persian: a fairly unremarkable Indo-European nominative/accusative alignment, but unlike Persian, cross-cut by ergative alignment in morphology in the past tenses. While the analysis offered in my earlier paper is still tenable as a synchronic description of the ‘standard’ versions of Kurmanji (see Section 1.1 on language names), it left a central issue unresolved: How did a language with seemingly unremarkable nominative/accusative syntax acquire morphological alignment bluntly at odds with its syntax? This book represents an attempt to answer that question.
Tackling the issue of ergativity in Kurdish from a diachronic perspective has turned out to be a daunting task. Ideally, it would have involved comprehensive coverage not only of the considerable number of Kurdish languages, but also of the attested earlier stages of these languages, and of the related Iranian languages. To forestall any false expectations, it has not been possible to achieve anything approaching this ideal. In particular, there is a dire lack of systematic evaluation of the attested Middle Iranian languages.
Obviously a more representative corpus of languages is required before ﬁrmer conclusions can be drawn. However, what this study lacks in breadth is partially compensated for by depth. While large-scale check-list typologies are invaluable for certain purposes, I believe that much can be induced through the careful inspection of individual languages, the intra-language variation, and most particularly, through the investigation of the constructions under consideration in running texts. As Allen (1995:452) stresses, one cannot reconstruct syntax merely by “strip mining descriptive studies for facts”. It is one of the main tenets of this study that discourse factors have shaped the development of alignments in various ways, and these can only be observed by investigating connected narrative texts rather than isolated examples in grammars. To this end, I have paid particular attention to analysing text material from Kurdish, rather than merely repeating what is stated in the grammars.
Kurdish provides an excellent starting point for such an undertaking. The various dialects/languages have been comparatively well-documented, and within the Kurdish languages themselves, a broad range of alignment types is attested. That internal variation may provide valuable insights to diachronic change is clearly recognized by Harris and Campbell (1995:12): “A fruitful and often overlooked source of reliable data in diachronic syntax is found in dialectal differences.” In particular, the text material available for Kurdish is far broader than that found in the corpora of older stages of the languages, because it includes extensive documentation of naturally spoken language, and in many cases can be supplemented by information from native speakers. The written records from older periods, on the other hand, often represent highly marked and often conservative varieties and registers, leading to considerable diﬃculties in interpretation. I nevertheless stress that the results presented here are to be considered as hypotheses, to be validated or invalidated against more extensive data from Iranian. Despite their limitations, the value of such hypotheses is considerable. They permit one to deﬁne a research goal in the form of a set of questions, thereby narrowing the scope of the data to be investigated. And they permit the results to be integrated into more general theories of alignment change. In principle, the amount and the nature of data available is unlimited; without some preformulated and testable hypothesis to guide our investigation, we would not progress beyond documentation. In this book then I will be examining a cross-section of past tense alignments, focussing primarily on Kurdish, but supplemented with data from other Iranian (mostly West Iranian) languages and older stages of Iranian, with the aim of formulating some hypotheses on the paths of development that may have led to the current situation. Alignment, and alignment changes, (see Section 2.1 for deﬁnitions) have been the subject of intense investigation in general and historical linguistics over the past three decades. The Kurdish case is, at ﬁrst sight, comparatively well known. In particular, two short papers by Bynon, Bynon (1979) and Bynon (1980), are regularly cited. Although Bynon’s papers are actually based on very little primary data, most historical linguists since have been content to accept her account (see Chapter 8 for references and discussion). Beyond that, there has actually been very little research within general and historical linguistics dedicated speciﬁcally to alignment changes in Iranian.
This is all the more surprising given that the other well-documented case of alignment change in Indo-European, the rise of ergativity in Indo-Aryan, continues to attract intense attention from linguists of all persuasions. 3 Almost 20 years ago Bossong (1985:118) stressed that (my translation) “the problem of ergativity [in Iranian] is in need of thorough analysis”, but little progress has been made in that direction since. It is thus high time that the Iranian case was reassessed against more extensive data, and in the light of more recent theoretical developments regarding alignment shifts. In particular, the claims of Nichols (1992) on alignment as a diachronically stable genetic trait need to be evaluated against the Iranian data.
Alignment in Kurdish:
a diachronic perspective
Geoffrey L. J. Haig
Please read this article in full by downloading the attached PDF file