Himalayan Languages and Linguistics
Brill's Tibetan Studies Library Edited by
Henk Blezer Alex McKay Charles Ramble
Languages of the Greater Himalayan Region Edited by
George L. Van Driem
VOLUME 5/12
Himalayan Languages and Linguistics Studies in Phonology, Semantics, Morphology and Syntax
Edited by
Mark Turin and Bettina Zeisler
BRILL
LEIDEN • BOSTON 2011
This book is printed on acid-free paper. Library of Congress Cataloging-in-Publication Data
ISSN 1568-6183 ISBN 978 90 04 19448 9 Copyright 2011 by Koninklijk.e Brill NV, Leiden, The Netherlands. Koninklijk.e Brill NV incorporates the Imprints Brill, Hotei Publishing, IDC Publishers, Martinus Nijhoff Publishers and VSP. All rights reserved. No part of this publication may be reproduced. translated. stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission from the publisher. Authorization to photocopy items for internal or personal use is granted by Koninklijk.e Brill NV provided that the appropriate fees are paid directly to The Copyright Clearance Center, 222 Rosewood Drive, Suite 910, Danvers, MA 01923, USA Fees are subject to change.
CONIENTS Notes on contributors .........................................................................vii Introduction .......................................................................................... 1
Mark Turin and Bettina Zeisler PART ONE: The Himalayas in history Lost in the sands oftime somewhere north of the Bay of Bengal.. .... 13
George van Driem PART TWO: Phonology and script A key to four transcription systems ofLepcha ................................... 41
Heleen Plaisier Dialectal particularities of Sogpho Tibetan - an introduction to the ''Twenty-four villages' patois" ..................................................... 55
Hiroyuki Suzuki PART TIIREE: Semantics (words and word classes) On the Old Tibetan Term Khrin in the legal and ritual lexicons ........ 77
Brandon Dotson A functional analysis of adjectives in Newar ..................................... 99
Kazuyuki Kiryu PART FOUR: Morphology and syntax The role of animacy in the verbal morphology of Dongwang Tibetan .............................................................................................. 133
Ellen Bartee The Sampang verbal agreement system ........................................... 183
Rene Huysmans Ergativity in Kundal Shahi, Kashmiri and Hindko ........................... 219
Khawaja A. Rehman Kenhat, the dialects of Upper Ladakh and Zanskar .......................... 235
Bettina Zeisler
Vl
CONTENTS
Index ................................................................................................. 303
NOTES ON CONTRIBUTORS ELLEN BARTEE and her husband, Ken Hugoniot, work with SIL and live with their daughter in Shangrila, Yunnan. They are currently working on a project in cooperation with the Yunnan Minority Affairs Bureau to conduct a linguistic survey of the dialects spoken in Diqing Tibetan Autonomous Prefecture. BRANDON DOTSON is a Visiting Researcher at Ludwig-Maximilians-Universitat in Munich, where he leads the research project ''Kingship and Religion in Tibet." He is the author of The Old Tibetan Annals: An Annotated Translation of Tibet's First History (Vienna: Verlag der Osterreichischen Akademie der Wissenchaften, 2009~ GEORGE VAN DRIEM has directed the Himalayan Languages Project since 1993 and has occupied the chair for Language Documentation at Leiden University since 1999. He was with the Department of Comparative Linguistics from 1983 until its dissolution in 2004. He collaborates with colleagues in the Human Genetics department at Leiden and with foreign partners in the international research programme Languages and Genes of the Greater Himalayan Region.
RENE
HUYSMANS studied Italian linguistics at the University of Amsterdam (UvA), where he graduated in 1993. In 1995 he became a member of the Himalayan Languages Project and since then he has been working on Sampang, a Kiranti language spoken in eastern Nepal.
KAZUYUKI KIRYU trained in linguistics at Kobe University and is now Associate Professor at Mimasaka University, Japan. He works on the N ewar language and in 2002 compiled a reference grammar and glossary for an intensive Newar course at the Research Institute of Language and Culture in Asia and Africa. He also works on Meche, a variety of Bodo (Tibeto-Burman) spoken in south-east Nepal. HELEEN PLAISIER holds a PhD from Leiden University and is an expert on Lepcha language, literature and culture. She has published both a grammar of Lepcha and a catalogue of the world's largest collection ofLepcha manuscripts. She is currently completing an analytical edition of a native 19th-century Lepcha dictionary.
V111
NOTES ON CONTRIBUTORS
KHAWAJA A REHMAN teaches at the Post Graduate College Muzaffarabad and is a PhD candidate at Quaid-i-Azam University, Islamabad. He is currently researching Kundal Shahi, an endangered language spoken in the Neelam Valley of Pakistan-administered Kashmir, and is compiling a Kundal Shahi-Urdu-English dictionary. lllROYUKI SUZUKI holds a D.Litt from Kyoto University and is a postdoctoral research fellow at the Japan Society for the Promotion of Science and National Museum of Ethnology. His primary research interests are descriptive and historical linguistics of the languages spoken in the eastern Tibetan cultural area. MARK TURIN is a linguistic anthropologist and holds a PhD from Leiden University where he was a member of the Himalayan Languages Project. He has worked on Thangmi, a Tibeto-Burman language spoken in Nepal, and has supervised the Linguistic Survey of Sikkim. He is presently based at the University of Cambridge where he continues to direct the Digital Himalaya Project BETTINA ZEISLER holds a Dr.Phil. degree from the Free University Berlin. After a seven-year postdoctoral research project in the Collaborative Research Centre 441 'Linguistic data structures' at the University of Tiibingen she is now directing a new project on verb semantics and valency: the Valency Dictionary ofLadakhi Verbs. Her research interests focus on the grammar and history of Tibetan languages, from the th century Old Tibetan to contemporary Ladakhi.
INTRODUCTION MARK TuRIN AND BETIINA ZEISLER
The greater Himalayan region is the principal meeting point for the two largest language families of the world, Indo-European and TibetoBunnan. The same massifs have also been home to two smaller language families (Austroasiatic and Dravidian), and to two language isolates (Burushaski and Kusunda). Despite their physical prominence, the Himalayas constitute not so much an insurmountable barrier but rather a region of interaction between these various language families. Indo-Iranian languages, and perhaps also Burushaski, were in all probability spoken along the northern slopes of the western Himalayas before the spread of Tibetan to this area in the second half of the frrst millennium CE. Much earlier still, Tibeto-Bunnan languages established themselves on the southern slopes, while the speakers of IndoAryan and Iranian languages entered these mountainous regions along the river valleys from the south and, then again, from the northwest Whether such incursions were motivated by conquest or were rather of a more peaceful nature, the resulting linguistic situation was usually characterised by coexistence, with specific niches supporting the survival of specific languages. In those instances where speech forms were abandoned, and where a shift to a dominant language occurred, as in the case of Balti and Shamskat Ladakhi (from Eastern Iranian and/or Dardic languages to Tibetan), the transition took much longer than allowed for by the standard three-generation model for languages under extreme pressure. Multilingualism was surely the norm rather than the exception in many parts of the Himalayas, and this linguistic pluralism left traces in the languages involved. For example, a significant proportion of the lexicon of Newar, a Tibeto-Bunnan language spoken in Nepal, derives from Indo-Aryan; while Ladakhi, spoken in Jammu & Kashmir (India), and Balti spoken in Pakistan, have not only acquired and retained certain traits from Dardic languages, but have even influenced the latter. The geopolitical shifts brought about by the British colonisation of the Indian subcontinent, and the subsequent introduction of Englishmedium education together with the spread of modern mass communi-
2
MARK TURIN & BETIINA ZEISLER
cation, helped Indo-European (whether in the form of Hindi, Urdu or English) to gain ground, furthering the decline of many regional and local languages. While the diversity of the sub-groupings within Tibeto-Burman is on a par with that of the Germanic or Romance languages of Europe, (even if their speaker numbers are relatively low), this richness in variation and the opportunity for original, fimdamental research has not yet translated into a profusion of linguists working in Himalayan areas. Furthermore, while the Himalayan region is the nucleus for languages of the Tibeto-Burman family, and also the area of their greatest diversity, this mountainous territory remains somewhat peripheral for scholars working on other language families. This is reflected in the interests of the participants of the Himalayan Languages Symposia held to date, with the majority of researchers focussing on Tibeto-Burman languages. This collection, too, originating at the 11th Himalayan Languages Symposium, held at Chulalongkom University in Bangkok, Thailand, in December, 2005, brings together a number of papers on the sub-groupings of Tibeto-Burman, such as Kiranti, Lepcha, Newar and Tibetan, one paper on Indo-Aryan, and another which addresses data from Austroasiatic. The present volume contains a selection of six of the best papers frrst presented at the Bangkok meeting, albeit fimdamentally revised and significantly updated with more recent field data, and all situated in a wider conceptual and comparative context. Three additional papers were invited on the basis of topical interest and new research fmdings coming to light (van Driem, Huysmans and Plaisier). Each of these nine contributions highlight the results of primary research, whether from the field or based on comparative material. THE HIMALAYAS IN HISTORY
Focussing on the eastern Himalayas, George van Driem provides a helpful introduction to the background context and macro-history of two Himalayan language families, Austroasiatic and Tibeto-Burman. In so doing, he draws on archaeology, population genetics, and linguistic reconstruction, and what may be termed linguistic paleaontology (the attempt to understand the material culture of a language family on the basis of its inherited lexicon). According to van Driem, these two language families hold the key to understanding the population prehistory of northeastern India and the Indo-Burmese borderlands.
INTRODUCTION
3
The original Tibeto-Bunnan homeland nominates itself on accmmt of the density of its speakers in the northeastern segment of the Indian subcontinent. The location of the ancestral Austroasiatic homeland, however, lying somewhere between South Asia and Southeast Asia, in the area around the northern coast of the Bay of Bengal, may be posited on the basis of linguistic palaeontology, the distribution of modem Austroasiatic language communities and the deep phylogenetic divisions in the family tree. Linguistic palaeontology further identifies the ancient speakers of Austroasiatic as likely being the earliest cultivators of rice, since a rich repertoire of reconstructible roots representing ancient rice agriculture is robustly reflected in all branches of this language family. PHONOLOGY AND SCRIPT
Closer to the ground, the second section opens with a discussion of how spoken and written languages are best transcribed and transliterated. Due to India's colonial heritage and the wide reach of English medium education, many speakers of languages indigenous to the subcontinent rely on conventions of English orthography when giving their tongues written form in Roman script. On occasion, this can result in arbitrary transcriptions signalling contrasts where there are none. A standardised system of transcription, then, based on linguistic criteria and understanding, would be beneficial for both native speakers and concerned fieldworkers. Heleen Plaisier discusses these issues in detail for the case of Lepcha (spoken in Sikkim and K.alimpong), carefully evaluating three established orthographic systems (two introduced by Western authors, the other by a native scholar), each of which have different shortcomings, and contrasts them with her own more systematic approach. Tibetan languages and dialects vary considerably in the degree to which they preserve the original phonological complexity documented in Old or Classical Tibetan orthography. Initial clusters, for example, are best preserved in West Tibetan (Balti and Ladakhi), followed by the Eastern Tibetan Amdo dialects spoken in nomadic regions. The non-nomadic Amdo dialects, as well as the Khams Tibetan dialects, typically occupy an intermediate position between the phonologically conservative dialects and the phonologically most innovative Central Tibetan dialects, where practically all clusters are reduced to a single
4
MARK TURIN & BETIINA ZEISLER
consonant in word-initial position. Such rough classifications will need further qualification and nuancing, however, as more hitherto undescribed Tibetan dialects are documented and analysed. While Sogpho Tibetan, as described by Hiroyuki Suzuki in this volume, is classified as a Khams Tibetan dialect by virtue of its location (it is spoken in Danba County in western Sichuan), it nevertheless shows several conservative features which are otherwise only recorded among the phonologically conservative Amdo dialects. Examples include preaspiration resulting from former clusters with dental and velar preinitials, the retention of clusters with labial stop initials as prelabialised consonants and the partial retention of clusters with a labial initial followed by a palatal glide or alveolar trill. In most Tibetan dialects, these clusters have developed into palatal or alveopalatal affricates and retroflex stops or affricates. Sogpho and the nomadic Amdo dialects are examples of an intermediate stage: while the initial remains unaffected or simply weakened, the post-initial glide or trill undergoes the major change, and the fmal stage is reached only by dropping the initial labial. Such fmdings are extremely important for our understanding of the mechanics and history of sound changes in Tibetan, and this paper is therefore an original and important contribution to this end. SEMANTICS (WORDS AND WORD CLASSES)
Following a more traditional philological approach, Brandon Dotson offers a precise study of an item of the Old Tibetan legal and ritual lexicon: khrin. Philology and linguistics diverged from one another long ago, and while pertinent linguistic issues are almost entirely neglected in Tibetan studies, most modern linguists lack the experience and skills to make sense of Tibetan texts, which are often opaque even to the specialist The rare vocabulary and lack of lexical resources in Old Tibetan add to the difficulties of decoding the syntax of a language that is as yet imperfectly understood. Dotson's contribution is an important step in the re-integration of linguistics with philology, in that he demonstrates the context-sensitivity of lexical items, particularly when used in a technical context As a legal term, khrin should be translated as 'judicial punishment', even though the most basic meaning of the word khrin is 'lead' or 'tether', an instrument for leading (sacrificial) animals along as well as the recipients of legal punishment The noun khrin 'tether' can be
INTRODUCTION
5
analysed as a derivation of the verb 'khrid in Dotson's transliteration 'lead along', with an archaic, 1mproductive derivational sufftx (-d/ -n) added to the root *khri. Beyond the ideological reshaping of concepts and meanings, Dotson's paper also offers insights into the daily practice of ancient Tibetan bureaucrats, who, 1mder certain circumstances at least, decided legal cases by rolling dice. The definition of parts of speech has long been the awkward stepchild of Tibeto-Burman studies. Word classes are typically established rather loosely according to the intuition of mostly non-native scholars, following in broad strokes the classifications fo1md in dominant European languages of which they are native speakers. Adjectivals, namely words describing qualia, constitute a particularly intricate class, oscillating between their more state-like properties, which are linked with nominals, and their more dynamic properties, which are linked with verbs. In some languages, such as Tibetan, adjectivals are split between basic verbal forms used as predicates and derived nominal forms used attributively. Newar, spoken in the Kathmandu valley of Nepal, exhibits an even more complex picture. While some researchers of Newar have denied the existence of a separate class of adjectives altogether, Kazuyuki Kiryu argues for the existence of a distinct word class and, based on morphosyntactic criteria, proposes that Newar adjectives be divided into three subclasses. Discussing thirteen conceptual domains usually associated with adjectivals, Kiryu identifies nine concepts that are realised as adjectives. He further distinguishes between verb-like adjectives, non-verb-like adjectives, and non-predicative adjectives. MORPHOLOGY AND SYNTAX
In many Tibeto-Burman languages, and among the Tibetan languages in particular, the distinction between conjunct information related to an (intentionally acting or controlling) 'informant' (that is, speaker in assertions, addressee in questions) and disjunct information not related to such an informant has become grammaticalised along with the concept of evidentiality, namely the distinction between varying sources of knowledge. Ellen Bartee reports on the Southern Khams Tibetan dialect of Dongwang, spoken in Shangri-la Co1mty in north-west Yunnan, which, like most other modem Tibetan varieties, shows a conjWlct/disjWlct system besides evidential marking. Unlike all other Tibetan dia-
6
MARK TURIN & BETIINA ZEISLER
lects, however, Dongwang, and possibly other Southern Khams dialects also, displays an animacy distinction in existential linking verbs and auxiliaries based on these very verbs. This distinction operates on S and 0 (more precisely possessed) arguments and interacts with the conjunct/disjunct system. Having discussed all possible combinations of conjunct/disjunct and animacy marking and their uses in various non-existential clauses, Bartee investigates the possible origins of the animacy distinction. A similar animacy distinction is found in Lolo-Burmese languages, such as Naxi, and since the speakers of Dongwang Tibetan have been in contact with the Naxi people for several centuries, the innovation in Dongwang most likely results from intensive linguistic contact Bartee briefly outlines the history of interaction between the two communities, and also provides a short description of the animacy distinction in Naxi. According to her analysis, the combination of the Dongwang conjunct/disjunct opposition with the animacy distinction has resulted in something akin to a cross-referencing agreement system, thus distinguishing itself from other evidential systems in modern Tibetan. Nepal's Kiranti languages are known for their intricate verbal agreement systems, perhaps the most complex in Tibeto-Burman. In his paper, Rene Huysmans provides the frrst published analysis of the verbal agreement system of Sampang, a Kiranti language spoken in eastern Nepal. While Sampang intransitive verbs agree with their only argument, transitive verbs show agreement both with the agent, i.e. transitive subject, and the non-agent, i.e. the undergoer, recipient, patient, or beneficiary. Singular, dual, and plural number is marked on all actants, and non-singular frrst person actants are further distinguished for inclusion or non-inclusion of the addressee. The Sampang verbal string is home to eleven afftxal slots (one preftxal, the rest sufftxal) for cross-referencing, tense marking, negation, and for additional copy morphemes. Many of the morphemes in question have several allomorphs, including zero-allomorphs, and, to complicate the picture further, several morphemes have fused into portmanteau morphemes. As a side effect, there is even more complex marking for combinations of frrst person patient with third person agents in past tense, reflecting a higher level of semantic transitivity. After laying bare the elaborate verbal morphology of Sampang, and having discussed several analytical alternatives, Huysmans demonstrates that a large portion of its morphemic inventory can be re-
INTRODUCTION
7
lated in a straightforward manner to the Proto-Kiranti verbal agreement model. Sampang verbal motphology is particularly reminiscent of the verbal agreement system of Kulung, but also shows similarities to the verbal morphology ofLimbu and Lohorung. In short, Kiranti morphological change can be seen to be primarily semantically driven: while the formal systems are indices of meaning, their re-analysis and re-modelling are propelled by ever re-interpreted shades of meaning. While the agreement patterns of Indo-Aryan languages are certainly less complicated than those of their Tibeto-Burman counterparts, there are variations in whether a verb agrees only with the least agent-like argument (always in the absolutive/nominative) or also with further arguments. In his contribution, Khawaja A Rehman discusses three different languages spoken in the Neelam Valley in Pakistanadministered Kashmir: his own native language Kashmir~ the dominant language Hindko and the highly endangered Kundal Shahi. All three languages show full agreement with the least agent-like argument, while Kashmiri, as spoken in the Neelam valley, displays additional agreement in person and number with third person plural and second person agent-like arguments. In contrast to other Indo-Aryan languages, where intransitive verbs may display alternations between absolutive and ergative matking, Neelam Kashmiri further shows obligatory ergative marking for a set of more agentive or volitional one-place verbs, including reflexives such as 'bathe', sound production verbs such as 'laugh', and motion verbs such as 'dance'. Interestingly, most of these verbs do not display gender-based agreement, that is, they only exhibit masculine agreement markers. Dialect classifications are often based on the most evident differences in phonology and lexicon. With respect to the Tibetan varieties, Bettina Zeisler shows that such an approach may be misleading. In the case of the two dialect groups spoken in Upper and Lower Ladakh (Jammu & Kashmir, India), Kenhat and Shamskat, the essential difference is not between the dialects that preserve a conservative phonology with initial clusters, close to 'Jh century Old Tibetan, and those that have lost all clusters, but rather between the dialects that distinguish morphologically between agents and possessors, and those that do not. Other differences at the level of syntax include various ways of marking tense and evidentiality. In terms of phonology, the Kenhat dialects are particularly intriguing as their minor differences provide a model for the progressive stages
8
MARK TURIN & BETIINA ZEISLER
of the loss of initial consonant clusters, in the vocalisation of the fmal sibilant and in the trans-phonologisation from voice to register tone distinctions. More importantly, even though their phonology is less conservative than that of the Shamskat dialects, Kenhat dialects show more lexical and grammatical archaisms, allowing us to trace the morphological developments that both dialect groups, along with other Tibetan varieties, must have undergone. This is an important lesson in itself, since historical reconstruction in general, and that of the Tibeto-Burman languages in particular, tends to be based on somewhat superficial aspects of phonology and neglects the development of grammatical systems. The phonologically innovative Central Tibetan varieties, to date largely disregarded in attempts at reconstruction, may be as rich in lexical and syntactic archaisms as the Kenhat dialects of Upper Ladakh, while the phonologically most conservative dialects might not necessarily represent the oldest layers of the language. Where van Driem focuses on the most southeasterly section of the Himalayas and their prolongation into the Bay of Bengal, Rehman and Zeisler focus on the extreme north-west and its extension into the Hindu Kush. Towards the center of these geographical extremes, Suzuki and Bartee lead us to the northern and southern regions of Tibet, Plaisier to Sikkim, and Huysmans and Kiryu to eastern and central Nepal respectively. Between the two poles of macro-history, as represented by van Driem's overview of tens of thousands ofyears of language development and population migration, and micro-history, as embodied in Dotson's careful reconstruction of the semantics of a specific lexical item used in the bureaucratic jargon of the Tibetan Empire around one thousand years ago, we fmd Bartee's description of the last one thousand years of interaction between two communities speaking TibetoBurman languages, and Zeisler's analysis of the Upper Ladakhi varieties, offering us a glimpse into the prehistory of Tibetan speech forms. This volume has been a long time in the making. Five years have elapsed since the initial conference that brought many of the contributors together and the fmal publication of this edited book. While the delays are regrettable, and remain our responsibility alone, we believe that the collection has matured in the oak and is the better for the wait. In our capacity as editors, we wish to thank Dr Krisadawan Hongladarom, a linguist at Chulalongkorn University and founder of the impressive Thousand Stars Foundation that promotes understanding of
INTRODUCTION
9
Tibetan religion and culture, both for convening the conference and for her support in the early stages of this volume's production. In addition, we are very grateful to the students and staff of Chulalongkom University for helping to make the conference a success; to the former Collaborative Research Centre 441 at the University of Tiibingen for organisational and fmancial assistance; to the anonymous reviewer of the manuscript for such careful reading and helpful comments; to Patricia Radder and the team at Brill for their responsiveness and interest; to Eleanor Wilkinson for her assistance with preparing the index and to all of our contributors for their considerable patience and good cheer.
PART ONE: THE HIMALAYAS IN HISTORY
LOST IN THE SANDS OF TIME SOMEWHERE NOR1H OF THE BAY OF BENGAL GEORGE VAN DRIEM
In the northeastern portion of the Indian subcontinent, the presence of two language families stretches back into prehistory. What light can historical linguistics, linguistic palaeontology, archaeology, palaeoethnobotany and human population genetics shed on the ancient origins of Austroasiatic and Tibeto-Bunnan? Which new questions arise from these interdisciplinary insights?
1. OLD AND NEW LINGUISTIC PHYLA IN THE EASTERN HIMALAYAN REGION
The world's two most populous families of languages meet in the Himalayas. These are the Tibeto-Bunnan phylum, which includes Cantonese and Mandarin Chinese, and Indo-European, to which languages such as English and Bengali belong. In addition to these two great linguistic phyla, Kra-Dai alias Daic, Austroasiatic and Dravidian language communities skirt the eastern Himalayan region. For example, the Austroasiatic language Khasi is spoken in the Meghalaya. Ahom, a now extinct Kra-Dai tongue, was once a prominent cultural language in northeastern India, where scattered Daic language communities are still settled today. The Dravidian tongues Dhangar and Jhangar, dialects of Kurukh or Uraon, are spoken in Nepal's eastern Terai, and Kurukh is also spoken in scattered communities throughout northeastern India by people displaced by British colonial policies in India involving resettlement, not unlike the policy of transmigrasi in the former Dutch East Indies. Yet the Himalayas would appear to be peripheral to our understanding of the prehistory of Dravidian. The crux to the ethnolinguistic prehistory of the eastern Himalayan region are the language families Tibeto-Burman, Austroasiatic, Indo-European and, to a lesser extent, Kra-Dai.
14
GEORGE VAN DRIEM
Tamangic
~o ~
Sinitic
GG 8
0 O Q
Qiangic
sa. ""Raute~epangic D•ganm8~EV~ ~~
Magaric
Rl!i•-
Midzhuim
Dura
Nwgim
~ ~
E:2G~8~
~ ~ ~ e~l';:'\
~e~eeE;8 Figure 1. The fallen leaves diagram for Tibeto-Bmman represents a relatively agnostic view of the internal phylogeny of the linguistic phylum. Some subgroups are wellestablished, whilst others are less so. Brahmaputran may include Kachinic and Dhimalish. For the sake of argwnent, this diagram breaks up the administrative catch-all 'Qiangic' into the Ersfi cluster and a tnmcated 'Qiangic', not to posit a robust phylogenetic hypothesis but with the intent of presenting a challenge, by emphasising the crucial work in this area that has still been left undone. The precise phylogenetic relationships between Erg5ng, Qiang, Mi-fiag (Milyii), Tangut, the diverse rGyal-rong languages, Ersu, Lfisu, Tosu (Duoxu), Namuyi, ShOOng, Guiqi6ng, Choyo (Quey\1), Zhabit and Prinmi (Pfurn") have yet to be demonstrated What is sometimes called 'Northern Qiangic' is supposed to include the rGyal-rongic group recognised by Jackson Sun (Sfin Tiaroon) and Huang BUfan Hopefully scholars working on the TibetoBurman languages of Sichuan and YUnnan will in the coming years shed light on the structure of this portion of the Tibeto-Burman family tree (van Driem 2001, 2006).
The advent of Indo-European languages and Kra-Dai languages to the northeastern portion of the Indian subcontinent is relatively recent and to some extent historically attested. Yet whilst the impact of Kra-Dai today is marginal, the intrusion of Indo-European into the region has been more robust and is represented by expansive languages such as Assamese, Hindi, Bengali and English. At the same time, historical,
LOST IN 1HE SANDS OF liME
15
ethnographic, archaeological, anthropological and linguiltic data all point to an Austroasiatic and Tibeto-Bmman presence in northeastern India which must date back to some hoary period of ethnolinguistic prehistory. Both families, Amtroasiatic and Tibeto-Bmman, therefore hold the key to understanding the population prehistory of northeastern India and the Indo-Bunnese borderlands.
Figure 2. Geographical distribution of the majoc branches of the Tibeto-Bunnan language family. Each damond rep-esents not a language, but a ml!ior subgroup.
The deeper phylogeny of the Indo-Emopean language family was once conceived by August Schleicher as a branching oak tree, but the phylmn has increasingly assmned a rake-like appeamnce in more recent literatme. The cmrently best-informed family tree structure of the Tibeto-Bmman phylmn is likewise essentially rake-like, a situation for which I introduced the metaphor of fallen leaves (Figme 1). The geographical distribution of the major Tibeto-Bmman subgroups saliently shows the densest concentration in the northeastern portion of the Indian subcontinent, with most major subgroups being exclusive to this area (Figme 2). The most authoritative phylogenetic tree for Austroasiatic at present is the model developed by Gerard Diffi.oth (Figme 3). Some novel insights into the phylogeny of Kra-Dai have been put forward by Edmondson and Solnit (1988, 1997) and Ostapirat (2005).
16
GEORGE VAN DRIEM JOOOAD
OAD
JOOOBC
I
I
I
Korlru Kherw arian Kharia-Juang Koraput Khasian Pakan ic Eastern Palaungic Western Palaungic Khmu ic Vietic Eastern Katuic Western Katuic Western Bahnari Northwestern Bahnaric Northern Bahnaric Central Bahnaric Southern Bahnaric Khmeric Peari.c Monic Nerthern Asli Senoic Southern Asli Nicob arese
ZOOOBC
3000BC
I
4000BC
I
I
5000BC
I
Monda
I
I Khasi-Khmuic
I Vieto-Katuic I I
r- -
Khmero-Vietic
Mon-Khmer
KhmeroBahnaric
-
--
Asli-Monic
J-r----1
Nico-Monic
Figure 3. Austroasiatic with Gerard Diflloth's tentative calibration of time depths for the various branches of the language family (modified from Diflloth 2001, 2005). The precise phylogenetic propinquity of Pearic, after Khmeric loan layers have been stripped off, remains Wicertain except that Diflloth observes that Pearic is MonKbmer and not 'Wle espece de vieux khmer', as some scholars once maintained. This diagram arranges in a tree-shaped phylogeny the fourteen recognised branches of Austroasiatic, i.e. North MWida, South MWida, Khasian, Pakanic, PalalDlgic, Khmuic, Vietic, Katuic, Bahnaric, Khmeric, Pearic, Monic, Astian and Nicobarese.
The location of the Austroasiatic ancestral homeland can be argued from a purely linguistic point of view principally on the basis of linguistic palaeontology and on the basis of the geographical centre of gravity of the family based on the distribution of modem Austroasiatic language communities and the deep phylogenetic divisions in the family tree. The distribution of the modem language communities and the geography of the deepest historical divisions in the family's linguistic phylogeny would put the geographical centre of the family somewhere between South Asia and Southeast Asia, in the area around the north-
LOST IN THE SANDS OF TIME
17
em coast of the Bay of Bengal. Whether we asswne that the deepest division in the family lies between Munda1 and the rest, as an older generation of scholars used to suspect, or asswne the veracity of Diffloth's new tripartite division, then the geography of deep historical divisions in linguistic phylogeny would compel us to look for a homeland on either side of the Ganges and Brahmaputra delta, although we would be unable to say whether this homeland would have to have lain to the east or to the west of the delta. When linguists look beyond what linguistic phylogeny can tell them, they must ask which archaeological transition or modem genetic gradient can be related with confidence to an ancient linguistic intrusion or to the prehistorical spread of a language family. When linguists resort to linguistic palaeontology, they must have recourse to the fmdings of palaeoclimatologists and, more particularly, palaeobotanists. Most conceivable theories about the homelands of TibetoBurman and Austroasiatic have already been put forward. The idea of a Tibeto-Burman homeland situated in or nearby present-day Sichuan has been entertained since the 19th century, especially by British scholars in India. Sinocentrists favour a northern Tibeto-Burman homeland in the lower Yellow River basin on the plains of northern China, whereas some have proposed a provenance within the Himalayan region itself. Scholars have sought to situate the Austroasiatic Urheimat as far west as the Indus valley and as far east as the Yangtze delta or insular Southeast Asia. However, the main contenders today for the Austroasiatic homeland are the Indian subcontinent, mainland Southeast Asia and the middle Yangtze. 2. ARCHAEOLOGY, PALAEOETHNOBOTANY AND LINGUISTIC PALAEONTOLOGY
The fimdamental epistemological question will continue to haunt us whether the spread of a recognisable Neolithic and Bronze Age assemblage can actually ever be taken with certainty to reflect the spread of a language and so of a language family. Archaeology reflects what 1 The presence of many speakers of Munda languages in northeastern India is a legacy of resettlement to Assam by the East India Company, an economic policy mentioned above in connexion with speakers of N orthem Dravidian languages who were also thus affected
18
GEORGE VAN DRIEM
we have been able to glean about the material culture of past communities. In fact, how often can we be certain which language was spoken by ancient stone knappers or by the potters behind a particular ceramic culture in some archaeologically attested pre-literate society? Indeed, we must ask whether the modem geographical distribution of the Tibeto-Burman language family correlates with the mute testimony of any single portion of the archaeological record that happens to have been preserved, discovered and studied by archaeologists. A more general issue is time depth. Archaeological transitions are reconstructed at very different times in the past, e.g. the palaeontologically attested spread of anatomically modem humans, the spread of agriculture, and the sometimes well-defined patterns of dispersal of identifiable cultural assemblages in the Neolithic and Bronze Age. At the same time, many known historical transitions and conquests with linguistic consequences have left little or no clear-cut traces whatsoever in the archaeological record. Therefore, we are free to muse, for example, whether the expansion of early Bodie language communities into the Himalayas was associated with the sudden appearance of colonial exponents of the Majiayao Neolithic in eastern Tibet at mKhar-ro and in Kashmir at Burzahom at the same time that the core area in Gansu shrank during a period of climate change between the Majiayao phase (2700-2300 BC) and the Banshan phase (2200-1900 BC) of the Majiayao sequence. This at least is a scenario which I argued in several earlier publications (van Driem 1998, 2001, 2002). Recent palaeoecological evidence indicates that the vast and once heavily forested Tibetan plateau underwent large-scale deforestation precisely during this period of projected Bodie expansion, and the palaeobotanical evidence indicates that this deforestation occurred at the hands of human settlers (Kaiser et al. 2006, Miehe et al. 2006, Wu et al. 2006, Kaiser et al. 2007). Were these people perhaps Bodie colonists from the Majiayao Neolithic core area who introduced to the Tibetan plateau a new lifestyle with deleterious ecological ramifications? Recently, as alternatives to the scenario outlined here, I have presented several alternative scenarios which differently relate the traceable patterns of dispersal of cultural assemblages in the archaeological record with the present geographical distribution of Tibeto-Burman language communities (van Driem 2006). Linguistic palaeontology, a term introduced by Adolphe Pictet in 1859, is an attempt to understand the ancient material culture of alan-
LOST IN THE SANDS OF TIME
19
guage family on the basis of the lexical items which can be reliably reconstructed for the common ancestral language. Linguistic palaeontology, like any attempt to give an ethnolinguistic interpretation to the archaeological record, invariably raises complex issues. Elsewhere, I have discussed at some length the arguments relating to what we can glean about ancient Tibeto-Burman culture and the role played by broomcorn millet Panicum miliaceum and by foxtail millet Setaria italica, the latter reflected in languages as far flung as Old Chinese .fl btsik in the Yellow River basin and Lhokpu2 ca 7 kto 'foxtail millet' in modem southwestern Bhutan (van Driem 2006). At the present time, the earliest archaeologically attested domestic millet dates from before 6000 BC at J1!13f:?M .x-mglooggou near $illt Chif'"eng, where a Neolithic culture without sickles once flourished (Zhao 2005). Linguistic palaeontology strongly qualifies the ancient Austroasiatics as the most likely candidates for the frrst cultivators of rice. Moreover, Diffloth has shown that the reconstructible Austroasiatic lexicon paints the picture of a fauna, flora and ecology of a tropical humid homeland environment, with three salient isoglosses diagnostic for the faunal ecology of the Proto-Austroasiatic homeland reconstructible all the way to the Austroasiatic level and reflected in all branches of the family, i.e. *mra:k 'peacock Pavo muticus', *tn-kuat 'tree monitor lizard Varanus nebulosus or bengalensis' and *tmyu:? 'binturong' or the 'bear cat Arctitis binturong', a black tropical mammal that is the largest of the civet cats (Diffloth 2005: 78). All of these species are not native to areas that currently lie within China, and, to our present knowledge, these species never were native to the area that is today China. Such linguistic palaeontological evidence therefore appears to render the middle Yangtze homeland hypothesis less likely. More reconstructible Proto-Austroasiatic roots indicative of a tropical or subtropical climate are adduced by Diffloth (2005: 78), i.e. *(b~n)jo:l ~ *j(~rm)o:l 'ant eater, Manis javanica', *d~n 'bamboo rat, Rhizomys sumatrensis' (an Austroasiatic root which has found its way into Malay as a loan), *kac~ 'the Asian elephant, Elephas maxi2 The Lhokpu are an inbred and genetically highly distinct group within the Himalayan region as a whole (Kraaijenbrink et al. 2006a, Parkin et al. 2006a). The impact of matrilocality and cross-cousin endogamy is clearly discernible in the genetic signature of this language community. Many of the ancient Tibeto-Bmman groups may have been matrilineal, matrilocal societies with uxorilocal marriage such as 1he modem Lhokpu and Gongduk of Bhutan.
20
GEORGE VAN DRIEM
mus', *kia9 'mountain goat, Capricornis sumatrensis', *rama:s 'rhinoceros, Dicerorhinus sumatrensis' and *tmriak 'buffalo, Bubalus bubalus'. Additionally, Difiloth (2005: 78) points out a fact long noted by scholars of Austroasiatic linguistics, e.g. Osada (1995), namely that a rich repertoire of reconstructible roots representing ancient rice agriculture is robustly reflected in all branches of Austroasiatic, viz. *~)6a:? 'rice plant', *ragko:? 'rice grain', *caJ]ka:m 'rice outer husk', *lGm
LOST IN THE SANDS OF TIME
21
Pal 1990, Agrawal 2002), there are now newer sites with more reliable dates at Lahuradewa (Lahuradeva), Tokuva and Sarai Nahar Rai. At the Lahuradewa site (26°46' N, 82°57' E), the early farming phase, corresponding to period 1A in the site's clear-cut stratigraphy, has radiocarbon dates ranging from ca. 5300 to 4300 BC. Carbonised material from period 1A was collected by the flotation method, yielding Setaria glauca and Oryza rufipogon as well as a morphologically distinct, fully domesticated form of rice 'comparable to cultivated Oryza sativa' (Tewari et al. 2002). More recently, accelerator mass spectroscopy dates were obtained on the rice grains themselves, corroborating the antiquity of rice agriculture at the site. Most recently, new radiocarbon dates for rice agriculture have been coming from the Ganges basin, with the Tokuva site near Allahabad now yielding similar dates (Vasant Shinde [Vasant Sivaram Sinde], personal communication 27 November 2007), and exciting new dates for ancient rice agriculture are also emerging from Sarat Nahar Rat (Manjil Hazarika, personal communication 7 March 2008). Of course, we are living at a time when a more reliable calibration of radiocarbon dates in general has become a matter of great urgency. Further east, at least five species of wild rice are native to northeastern India, viz. Oryza nivara, Oryza officianalis (0. latifolia), Oryza perennis (0. longistaminata), Oryza meyeriana (0. granulata) and Oryza rufipogon, and reportedly over a thousand varieties of domesticated rice are currently in use in the region (Hazarika 2005, 2006a). The different varieties of rice in northeastern India are cultivated in three periods by distinct cultivation processes. In the process of ahu kheti, the rice is sown in the months of Phagun and Sot, i.e. mid February to early April. The seedlings are not transplanted but ripen in just four months in fields which must be constantly weeded. In bau kheti, the rice seedlings are sown from mid March to mid April in ploughed wet fields and likewise do not need to be transplanted. In sali kheti, the rice is sown from mid May to mid June, and the seedlings are transplanted. Sali kheti rice varieties are suspected to derive from the wild officianalis rice still widely found in swampy village areas. The wild rufipogon rice cannot be used for human consumption because the plants shed their seeds before they ripen, so that rufipogon rice is used in Assam and other parts of northeastern India as cattle feed (Hazarika 2006b).
22
GEORGE VAN DRIEM
Whilst claims have been published dating the earliest rice cultivation in East Asia to as long ago as 10,000 BC, the currently available evidence indicates that immature morphologically wild rice may have been used by foragers before actual domestication of the crop, e.g. at Bashidang site (7000-6000 BC) belonging to the ~;~ LlJ the }\ Pengt6ushan culture in the Middle Yangtze and at sites in the Yangtze delta area such as ~mJm Kuahilqiao, .~%~~ Majiabang (5000-3000 BC) and %J~¥1l Hemudu (5000-4500 BC). However, only ca. 5000 BC was the actual cultivation of rice probably frrst undertaken by people in the Lower Yangtze, who at the time relied far more heavily on the collecting of acorns and water chestnuts (Yasuda 2002, Fuller 2005a, 2005b, 2005c, 2006a, 2006b, 2006c, 2007a, 2007b, Fuller et al. 2007, Zong et al. 2007). There is also currently no evidence for the co-cultivation of rice and foxtail millet along the middle Yangtze until around 3800 BC (Nasu eta/. 2006). Today, our understanding of the palaeoethnobotanical picture is more complex. The two main domesticated varieties of rice, Oryza indica and Oryza japonica, are phylogenetically distinct and would appear to have been domesticated separately. Oryza indica derives from the wild progenitor Oryza nivara and was frrst cultivated in South Asia or western Southeast Asia, perhaps in two separate domestication events. On the semi-arid Gangetic plain at the end of the mid-Holocene wet period, habitats for wild rices increasingly shifted to oxbows as palaeochannels dried up and turned into oxbow ponds. This shift favoured monsoonal rather than marshland rice species, including Oryza nivara, the wild progenitor ofOryza indica (Fuller 2006a). Oryza japonica derives from the wild progenitor Oryza mfipogon, and it is currently believed that the rufipogon variety was frrst cultivated to yield early Oryza japonica along the Middle Yangtze. Harvey et al. (2006) have critically reassessed the morphometries of rice fmds associated with various Neolithic sites throughout the Yangtze basin in light of recent genetic fmdings. It appears that the wild progenitor Oryza mfipogon was not fully domesticated in the Lower Yangtze to yield early Oryzajaponica until ca. 4000 BC. Generally, the archaeological record shows a delay of one to two millennia between the beginning of cultivation and the frrst clear evidence of domestication sensu stricto, i.e. genetic modification by selective breeding. Twelve wild forest-margin rice species are known, found mostly in Southeast Asia as well as at old sites of human habitation, e.g. Jiahu
+11
LOST IN THE SANDS OF TIME
23
on the Huai river in Henan in the seventh millennium BC or Hemiidu in the Yangtze delta in the frrst half of the ftft.h millennium BC. Extinct wild varieties of rice also appear to be preserved in the modem japonica genome. Based on the genetics of the officianalis variety, the seasonally wet, puddle-adapted Oryza nivara, and the always wet perennial Oryza rufipogon, there may be evidence for multiple rice domestications in South, Southeast and East Asia. So, maybe the domesticators of Oryza nivara were ancient Austroasiatics, and maybe the domesticators of ancient Oryza rufipogon were ancient Hmong-Mien. O'Connor (1995) and Blench (2001) have argued that irrigated rice agriculture enabled people to seize control of lowlands and flood plains. People were able to move down from upland areas that had hitherto been more favourable habitats after wet cultivation had transformed lowlands from epidemiologically undesirable places into bountiful habitats. But what ifthe frrst cultivators and domesticators of rice already inhabited lowland river basins and flood plains, such as the Ganges or Yangtze basins or even the Brahmaputran flood plains? Turning to northeastern India and the Indo-Burmese borderlands, we must recognise that, notwithstanding the excellent archaeological work conducted in the Ganges and Yangtze river basins, much of the archaeology of ancient rice agriculture is simply not known because no substantive archaeological work has been done on the Neolithic in the most relevant areas, e.g. northeastern India, Bangladesh and Burma. The absence of evidence does not constitute evidence of absence, and the sheer dearth of archaeological research in these areas leaves entirely open the possibility that rice cultivation may have originated in this region. We might expect to find traces of ancient farming communities better preserved in the hill tracts surrounding the Brahmaputran flood plains than on the fertile fields themselves, although the earliest rice-based cultures may frrst have developed on those very flood plains. Perhaps the remains of the frrst rice cultivating cultural assemblages lie buried forever in the silty sediments of the sinuous lower Brahmaputran basin or were washed out by the Brahmaputra long ago into the depths of the Bay of Bengal.
24
GEORGE VAN DRIEM
3. PoPULATION GENETICS, THE LAST ICE AGE AND MORE ARCHAEOLOGY
Very often language seems to be less ambiguously correlated with the geographical distribution of genetic markers in the populations speaking the languages in question. So, can genes and languages generally be correlated and contrasted with each other in a more meaningful way than either can be with the fragments of material culture that happen to have resurfaced unscathed from the sands of time? On the one hand, the linguistic ancestors of a language community were not necessarily the same people as the biological ancestors of that community. At the same time, the genetic picture often shows sexual dimorphism in linguistic prehistory. Some languages appear to be mother tongues, whereas others show up as father tongues. In Baltistan, in what today is northern Pakistan, the phonologically highly conservative local Tibetan dialects appear to correlate with the predominantly Tibeto-Burman mitochondrial DNA, which reflects the Balti community's maternal ancestry (Poloni et al. 1997, 2000, Zerjal et al. 1997, Quintana-Murci et al. 2001, Qamar et al. 2002), whilst the intrusive paternal Y haplogroups from the Near East appear to correlate with the forcible conversion of the area to Islam in the fifteenth century. By contrast, the 'Father Tongue hypothesis' may apply to the spread of Indo-Aryan into the Indian subcontinent and, further east, to the spread of Sinitic during the Han demic expansion. In fact, a likely correlation between the linguistic affmity and the Y haplogroup of a population appears to be a more widespread phenomenon. At many times and in many places in prehistory, the father tongue may have been the guiding mechanism in language shift. The dynamics of a process whereby mothers passed on the language of their spouses to their offspring also has major implications for our understanding of language change. If the language shift giving rise to the Sinitic languages and perhaps also the eastward spread of Indo-Aryan speech across northern India took place in this way, then such languages may have begun as languages belonging to another phylum until they reached the stage currently attained by Michif. In origin at least, Michif is genetically an Alqonquian language that was spoken by women who relexified the language with the French spoken by their husbands to such an extent that the genetic affmity has nearly been obscured (Bakker 1992, 1994, van Driem 2001: 169-173). If the
LOST IN THE SANDS OF TIME
25
process of relexi:fication were to continue beyond the stage attained by Michif, then a language could conceivably change its genetic affmity even though the dynamics of the process would introduce a discontinuity with its past. If such a process took place, could it ever be reconstructed linguistically? Additionally, though these fmdings have not yet reached a wider public, geneticists have increasingly been identifying single nucleotide polymorphisms on the autosomes that are diagnostic for geographical and racial affmity and probably reflect salient episodes of our collective population prehistory. When we turn now to Tibeto-Burman and Austroasiatic, the best correlation with the linguistic affinity of a community to the present state of our knowledge, however, likewise appears to involve the predominant Y haplogroups in the populations tested to date. In other words, both Austroasiatic and Tibeto-Burman may, on the grander scale over time, have preponderantly, albeit not exclusively, spread by the Father Tongue mechanism. By this interpretation, the shared Y haplogroup is assumed to reflect the founding dispersal of the language family. A few of the pioneering genetic assays of Tibeto-Burman populations inside China ventured some plausible claims. The reduced polymorphism of northern populations of East Asia, which represent a subset of the haplotypes found in southern populations, was taken to reflect the peopling of the north after the Ice Age (Su eta/. 1999), whereas the high frequency of H8, a haplotype derived from M122C, was seen as reflecting a genetic bottleneck effect that occurred during an ancient southwesterly migration about 10,000 years ago, suggesting a demic diffusion at the outset of the Neolithic (Suet a/. 2000, Ding et a/. 2000, Shi et a/. 2005). Another study suggested that Han Chinese did not originate in the Yellow River basin but had more recently migrated to this area from southwestern China (Chu eta/. 1998). Comparison of various haplogroup frequencies exhibited by Tibetans vs. Tujia, Bai and Lolo-Burmese groups showed all Tibeto-Burman groups to have a high frequency of the Y-chromosomal haplogroups 03e and 03*, with the average hovering approximately around 40%. These :findings were interpreted as supporting a male-biased infiltration from the Bodish area in Amdo into Yunnan and H1lnan about two and a half millennia ago, though 'the less drastic bias between male and female lineages' suggested that these putative southward migrations 'likely occurred with the involvement of both sexes rather
26
GEORGE VAN DRIEM
than as conquests involving expedition forces primarily consisting of male soldiers' (Wen et al. 2004). Interestingly, genetic studies often appear largely to corroborate the long-standing intuitions of linguists, historians and ethnographers working in the region or even to support their previously published models of ethnolinguistic prehistory. Yet these Chinese studies are limited by the fact that most Tibeto-Bunnan language communities and even most branches of the language family are exclusively represented outside of China. The picture of the Tibeto-Bunnan past has been rendered far more complete by findings of our own research team, which has conducted the most extensive sampling of Tibeto-Bunnan populations in the Himalayan region (Kraaijenbrink et al. 2006a, 2006b, Parkin et al. 2006a, 2006b). These fmdings have allowed us to make novel inferences about the population prehistory of Tibeto-Burman, and has also uncovered some wholly unexpected finds, such as the genetic affmities of the Black Mountain Monpa. Additional fmdings from northeastern India, Tibet and Bunna will enable us to identify the possible molecular correlates corresponding to more episodes in the spread in Asia of ancient Tibeto-Bunnan language communities, giving us a more detailed picture of the past. Recently, population geneticists working in association with rvfrs. Krithika have begun to study Tibeto-Bunnan populations in northeastern India, but these studies have only begun to scrape the surface. Yet these pioneering studies have already shed light on female exogamy amongst the Panggi and adjacent Tani language communities of Upper Siang such as the Pasi and Minyong (Krithika et al. 2005, 2007a, Maji et al. 2007), shown the relative isolation of the Panggi vis-a-vis the Galo, Mishing and Padam (Krithika et al. 2007b, 2008), and held promise for ultimately reconstructing distinct migration routes notwithstanding the effects of geographical contiguity, as perhaps seen in the genetic affinity for twelve microsatellites between Tibeto-Burman populations of Mizoram and Arunachal Pradesh, setting them apart from Manipur, the Garo Hills and Sikkim (Krithika et al. 2006). Recent microsatellite studies have also begun to focus on the population groups in Tibet, whereby all the Tibetan place names have invariably been rendered incorrectly into English (Kang and Li 2005, Kang et al. 2007a, 2007b, Y an et al. 2007). TheY-chromosomal haplogroup 03e (M134), which seems to tag Tibeto-Bunnan language communities within and outside of the Hi-
LOST IN 1HE SANDS OF liME
27
malayan region, may very well have a time depth of at least 10,000 years (Figme 4 ), putting us in a time frame which compels us to consider in which localities the ancient Tibeto-B mmans may have dwelt during the last glacial maximum. Could the southern flank of the Himalayas have served as a vast refuge area during the last Ice Age, or did the early Tibeto-B mmans at this time dwell in some area to the northeast or to the east of the Himalayas? Palaeoclimatologists remain divided into rival camps on the question of the extent of the glaciation in the Himalayas in recent prehistory, e.g. Kuhle (1985, 1986, 1987, 1988a, 1988b, 1990a, 1990b, 1991, 1995, 1997, 1999, 2005), Thompson et al. (1989), Lehmkuhl (1995), Schafer et al. (2002), Owen et al. (2002), Owen et al. (2003), Jin et al. (2005), Lehmkuhl and Owen (2005), Vandenberghe (2007). By virtue of the sheer scale and diversity of the topography, the Himalayas harbom a panoply of climatological enclaves and sheltered areas with their own specific microclimate. Did the Himalayas offer hospitable ice-age refugia to the ancestral Tibeto-Bmmans?
Figure 4. The pmi.oo. of the Y chromosome Jilylogenetic lree relevant to the Father Tongue hypothesis with regard to the Austroasiatic and Tibeto-Burman linguistic Jilyla, provided by Made JoW.ng and Emma Parldn.
In seeking an answer to this question, northeastern India still remains scantily docmnented and poorly mderstood in archaeological tenns. Much of the major work in this region was carried out over a generation ago and has been discussed and referenced in the handbook (van
28
GEORGE VAN DRIEM
Driem 2001). Yet most such work pertains to the Neolithic, a later period which does not help shed light on the issue of possible ice-age refuge habitats. By contrast, a partial answer emerges from the far more complete picture which we have for Nepal thanks to the posthumously published work of Gudrun Corvinus, who, tragically, was murdered in her own flat in Pur).e by the henchman of a local real estate criminal. Ironically, she had chosen to live in Pur:te to avoid the mayhem, extortion and murder that Maoists were inflicting indiscriminately on their fellow countrymen in the regions where she conducted her work. On the basis of a lifetime of palaeontological and archaeological research in the sub-Himalayan tracts of Nepa~ Gudrun Corvinus (2007) developed the theory that the Early and Middle Palaeolithic and microlithic traditions in the western Terai and Sivaliks derive from contemporaneous cultures further south on the Indian subcontinent in a period still characterised by low population density, whereas the Patu culture in the eastern Sivaliks and the Brakhup culture in the western Terai show a later influence emanating westward across the Himalayan foothills from Southeast Asia in the Late Pleistocene and Holocene, with the archaeological record suggesting an increase in population density at the end of the last glacial maximum roughly coincident with the flourishing of the Brakhup culture. Corvinus' comprehensive pioneering work in Nepal has yielded good stratigraphies and optical and infra-red stimulated luminescence datings of key sediment layers based on work at sites along the Rato Khola south of Sindhuli in Mahottari district in the eastern Sivaliks, sites at and near Satpati hill east of Bhairahava in Lumbini district and various sites in the Dang and Deukhuri basin. The archaeological record suggests a long prehistory of human habitation. For example, the alluvial and colluvial hill-wash deposits composing the Gidhiniya and Babai formations in the Tui basin in Dang district have yielded abundant lithic material from the Early, Middle and Late Palaeolithic periods as well as from the Mesolithic and Neolithic periods. An Early Palaelothic presence of the Acheulian tradition in South Asia in the early Middle Pleistocene is indicated by Acheulian bifacials and flake tools industry at Gadari in Dang in the western Terai and at Satpati in central Nepal just west of the NariyaQ.I river. Alluvial terrace deposits at the Atjun site in the badlands on the left bank of the Atjun river show Levallois technology appearing in the Middle Palaeolithic.
LOST IN THE SANDS OF TIME
29
At the same, time the foothills of the central Himalayas show a complex cultural prehistory. Nepal straddles the so-called Movius line and represents a transition zone between two traditions of lithic expression. Sites in the western Nepalese Terai show affmity with other more Occidental technocomplexes as represented by artefacts of the Acheulian and Levallois traditions culminating in late Palaeolithic flaking and microlithic industries. The Patu culture in Mahottarl district in eastern Nepal, however, stands outside of the main subcontinental context and shows clear connexions with coetaneous technocomplexes in the forested habitats of Southeast Asia. Patu technology is characterised by cobble-tools and less distinctly retouched small flake tools and, in the Mesolithic context, with macroliths such as adzes and cobble tools as well as the unifacial, flat-based and steepedged tools called sumatraliths which clearly suggest an affinity with the Hoabinhian. Yet in later strata, the Brakhuti culture in the Tui valley in the Dang and Deukhuri basin preserves core scrapers and unifacial choppers manufactured by stone-knapping techniques that would appear to have been very particular to that area of present-day Nepal and quite distinct from more Occidental technocomplexes of the period. At the same time, the large flake core industry at Brakhuti comprises sumatraliths, some high-crested and some in the shape of a horseshoe, whilst adzes of the Oriental type found at Patu are lacking at Brakhuti. The precise chronological relationship between the Patu culture in the eastern Sivaliks and the Brakhuti culture in the western Terai has not yet been clarified, but heavy-duty stone tools would appear to have been de rigueur in the ancient forests of the Terai and Sivaliks hills. In turning from the archaeological record to the molecular biological legacy, the population genetic data in the Himalayan region correspond with the linguistic divide more sharply than in most other parts in the world. Whilst gradients of biological markers often flow fuzzily across deep linguistic boundaries, in the Himalayas both the genetic and linguistic divide between Tibeto-Burman vs. Indo-European remain sharp. Population geneticists also corroborate what linguists and ethnographers have long known, namely that the Himalayas themselves do not constitute the real geographical divide between TibetoBurman and Indo-European. Rather the divide runs roughly through the sub-Himalayas or the Terai. Similarly, there is a marked discontinuity between Neolithic and Bronze Age traditions up in the hills and
30
GEORGE VAN DRIEM
those down on the plains. Yet such later archaeological assemblages appear, to our current state of knowledge, to be younger than-and therefore posterior in time to--the population genetic divide, and perhaps also to the linguistic one. The genetic divide between Tibeto-Burman and Austroasiatic in the region shows a far more complex structure than the clear line demarcating Tibeto-Burman from Indo-European. To the south, in the Brahmaputran basin and the Indo-Burmese borderlands, some of the spread of Tibeto-Burman may have been at the expense of indigenous Austroasiatic populations whom the Tibeto-Burmans assimilated linguistically. The Y haplogroup 02a (M95) is represented at a frequency of 77% in Austroasiatic groups in India and 47% in Tibeto-Burman groups of northeastern India (Sahoo et al. 2006). This pattern could suggest that Tibeto-Burman paternal lineages may have partially replaced indigenous Austroasiatic lineages in the northeast of the Indian subcontinent in the distant past, and that, wherever their anterior ancestral provenance, ancient Austroasiatic populations preceded the Tibeto-Burmans in this region, as linguists and ethnographers have speculated for over a century and a half (van Driem 2001 ). The geographical extent of the Bodo-Koch languages and the shallow time depth of this sub-branch of Tibeto-Burman might represent the linguistic corollaries of such partial genetic replacement. My earlier arguments regarding the locations and geographical proximity of a more littoral Austroasiatic homeland and a more montane Tibeto-Burman homeland, both within or near the northeastern portion of the Indian subcontinent (van Driem 2006, 2007), continue to be supported by more recent genetic studies involving the Y-chromosomal haplogroups 02a and 03, e.g. Kumar et al. (2007). Austroasiatic is an old language family, and we would expect the population history of this family to be at least as complex as that of Tibeto-Burman, if not more so. In future, more detailed and careful correlation of linguistic and population genetic fmdings based on more fme-mesh population genetic sampling may enable us to reconstruct early language contact situations and ancient cases of language shift and linguistic intrusions that might, for example, account for the phenotypical differences readily observable between Munda speakers as opposed to Khasi-Khmuic and Mon-Khmer language communities as well as between Aslian negrito populations, Aslian non-negrito populations and the Nicobarese. Somewhat in parallel with such soma-
LOST IN THE SANDS OF TIME
31
tological observations by ethnographers, linguists have long observed corresponding typological differences between various branches of Austroasiatic. Donegan and Stampe (1993, 2004) suggest that Austroasiatic spread from the Indian subcontinent to Southeast Asia, but argue that paradoxically the synthetic head-fmal typology of Munda languages resulted from an innovative process of drift which unfolded within South Asia after the linguistic ancestors of modem Khasi-Khmuic and Mon-Khmer language communities migrated towards Southeast Asia. In their view, the typological change in Munda was triggered by a prosodic shift to a falling rhythm, whereas the analytic head-initial typology observed in Khasi-Khmuic and Mon-Khmer languages reflects the more original Austroasiatic state of affairs. By contrast, Zide and Anderson (1999, 2003) have argued that Munda verbal morphology is a conservative retention, and that older Austroasiatic grammatical systems were secondarily lost in the Khasi-Khmuic and Mon-Khmer languages of Southeast Asia. Yet if the Father Tongue hypothesis holds true for the spread of Austroasiatic into South Asia (van Driem 2007), then this outcome would vindicate Robert von Heine-Geldem's view of the Munda as the result of the 'Einwanderung mongolider austrasiatischer Stamme in Vorderindien' and of their 'Mischung mit Dravida und Urbevolkerungselementen' (1928, 1932). In that case, the typological divergence between Munda as opposed to Khasi-Khmuic and Mon-Khmer, lucidly discussed by Donegan and Stampe, may be the result of the adoption of an intrusive paternal tongue by indigenous pre-Austroasiatic populations of the Indian subcontinent The apparent Munda penchant for a falling prosodic rhythm might then be just one residue of a far-reaching action de substrat. If, however, the Father Tongue hypothesis holds true for Austroasiatic and Zide and Anderson are correct, then the Munda descendants of the linguistically assimilated South Asian indigenes may have more faithfully preserved the original Austroasiatic morphology and grammar of the early bearers of the Y-chromosomal haplogroup 02a than is now seen reflected in modem Khasi-Khmuic and Mon-Khmer languages, which underwent divergent areal developments specific to Southeast Asia. On the other hand, the Father Tongue hypothesis may very well not apply in all cases for the biological ancestry of Austroasiatic language
32
GEORGE VAN DRIEM
communities, just as language spreading solely via the paternal line cannot account for the linguistic identity of all Tibeto-Burman populations, e.g. maternal Balti vs. paternal Han. More decisive insights into both the historical linguistics and the historical typology of Austroasiatic and Tibeto-Burman, newly identified single nucleotide polym.otphisms on the autosomes, ethnolinguistically informed fine-mesh genetic assays of Astian negrito populations, Astian non-negrito populations, the Nicobarese and peoples of Tibet and Burma, and insights from other disciplines may help us to retrieve more undiscovered bits and pieces of prehistory that may not have been irretrievably lost in the sands of time.
BIBLIOGRAPHY Agrawal, D.P. 2002. The earliest pottery and agriculture in South Asia. In Yoshinori Yasuda, ed., The Origins ofPottery andAgriculture. New Delhi: Lusre Press and Roli Books, for the International Center for Japanese Studies, 81-88. Bakker, Peter (i.e. Pieter Jan Bakker). 1992. 'A Language of Our Own', the Genesis of Michif, the Mixed Cree-French Language of the Canadian Metis. U ni.versiteit van Amsterdam: doctoral dissertation - - . 1994. Michif, the Cree-French mixed language of the Metis buffalo hunters in Canada. In Peter Bakker and Maarten Mous, eds., Mixed Languages, 15 Case Studies in Language Intertwining. (Studies in Language and Language Use, 13). Amsterdam: Instituut voor Fundamenteel Onderzoek naar Taal en Taalgebruik, 1333. Blench, Roger Marsh 200 1. From the mountains to the valleys: Understanding etlmolinguistic geography in Southeast Asia. In Laurent Sagart, Roger Blench and Alicia Sanchez-Mazas, eds., The Peopling ofEast Asia: Putting Together the Archaeology, Linguistics and Genetics. London: Routledge, 31-50. Chii Jiayou and W. Huang, S. Q. Kuang, J. M. Wang, J. J. Xu, Z. T. Chu, Z. Q. Yang, K. Q. Lin, P. Li, M. Wu, Z. C. Geng, C. C. Tan, R. F. Du and rm Li. 1998. Genetic relationship of populations in China. Proceedings of the National Academy ofSciences ofthe United States a/America 95: 11763-11768. O'Connor, Richard A 1995. Agricultural change and ethnic succession in Southeast Asian states: A case for regional anthropology. Jaurnal ofAsian Studies 54.4: 968-996. Corvinus, Gudrun 2007 [postlmmous]. Prehistoric Cultures in Nepal: From the Early Palaeolithic to the Neolithic and the Quaternary Geology of the Dang-Deokhuri Dun Valleys. (Gisela Freund und Ludwig Reisch, eds., 2 vols.). Wiesbaden:
Harrassowi1z Verlag. Diflloth, Gerard 200 1. Tentative calibration of time depths in Austroasiatic branches, paper presented at the Colloque «Perspectives sm la Phylogenie des Langues d' Asie Orientales» at Pengueux, 30 August 2001. Diflloth, Gerard 2005. The contribution of linguistic palaeontology to the homeland of Austroasiatic. In Lament Sagart, Roger Blench and Alicia Sanchez-Mazas,
LOST IN THE SANDS OF TIME
33
eds., The Peopling of East Asia: Putting Together the Archaeology, Linguistics and Genetics. London: Routledge Curzon, 77-80. Ding Yuan-Clrun, Stephen Wooding, Hemy C. Harpending, Chi Han-Chang, U HaiPeng, Fu Yun-Xin, Pang Jun-Feng, Yao Yong-Gang, Xiang Yu Jing-Gong, Robert Moyzis and Zhang Ya-ping. 2000. Population stru.cture and history in East Asia. Proceedmgs of the National Academy of Sciences of the United States of America 97: 14003-14006. Donegan, Patricia J., and David Stampe. 1983. Rhthym and holistic organization of language structure. In Jolm F. Richardson, Mitchel Marks and Amy Chukennan, eds., Papers from the Parasession of the Interplay ofPhonology, Morphology and Syntax. Chicago: Chicago Linguistics Society, 337-363. - . 2004. Rhythm and the synthetic drift of MlDlda. In Rajendra Singh, ed., The Yearbook of South Asian Languages and Linguistics 2004. Berlin: Mouton de Gruyter, 3-36. van Driem, George. 1998. Neolithic correlates of ancient Tibeto-Burrnan migrations. In Roger Blench and Matthew Spriggs, eds., Archaeology and Language II London: Routledge, 67-102. - - . 2001. Languages of the Himalayas: AnEthnolinguistic Handbook of the Greater Himalayan Region, containing an Introduction to the Symbiotic Theory of Language (2 vols.). Leiden: Brill. - - . 2002. Tibeto-Burman phylogeny and prehistory: Languages, material culture and genes. In Peter Bellwood and Colin Renfrew, eds., Examining the Fanning/ Language Dispersal Hypothesis. Cambridge: McDonald Institute for Archaeological Research, 233-249. - - . 2006. The diversity of the Tibeto-Burman language family and the linguistic ancestry of Chinese. Bulletin ofChineseLinguistics 1.2: 211-270. - - . 2007. Austroasiatic phylogeny and the Austroasiatic homeland in light of recent population genetic studies. Mon-Khmer Studies 37: 1-14. Ech:nondson, Jerold A., and David B. So1nit, eds. 1988. Comparative Kadai: Linguistic Studies beyond Thai. Arlington: Smnmer Institute of Linguistics of the University ofTexas. - - . 1997. Comparative Kadai: the Tai branch. Arlington: Summer Institute of Linguistics of the University of Texas. Fuller, Dorian Q. 2005a. Ceramics, seeds and culinary change in prehistoric India. Antiquity 79: 761-777. - - . 2005b. Formation processes and palaeothenobotanical interpretation in South Asia. Journal ofInterdisciplinary Studies 2.1: 93-115. - - . 2005c. The Ganges on the world Neolithic map: The significance of recent research on agricultural origins in northern India. Pragdharij, Journal ofthe Uttar PradeshArchaeological Department16: 187-206. - - . 2006a. Agricultural origins and frontiers in South Asia: A working synthesis. Journal ofWorldPrehistory 20: 1-86. - - . 2006b. Dung mounds and domesticators: Early cultivation and pastoralism in Kamataka. In Catherine Jarrige and Vincent Lefevre, eds., South Asian Archaeology 2001: Proceedmgs of the Sixteenth lntemational Conference on South Asian Archaeology, EuropeanAssociation ofSouthAsianArchaeologists, Paris, 2-6July 2001. Paris: Editions Recherche sur les Civilisations, 117-127. - - . 2006c. Silence before sedentism and the advent of cash crops: A status report on early agriculture in South Asia from plant domestication to the development of political economies (with an excursus on the proble of semantic shift among millets and rice). In Osada Toshi.ki, ed, Proceedmgs of the pre-Symposium of the
34
GEORGE VAN DRIEM
Research Institute for Humanity and Nature and 7th Ethnogenesis of South and Central Asia Round Table Harvard-Kyoto Roundtable. Kyoto: Research Institute for Hmnani.ty and Nature, 175-213. - - . 2007a. Contrasting patterns in crop domestication and domestication rates: Recent archaeobotanical insights from the Old World Annals ofBotany 2007: 1-22. - - . 2007b. Non-lruman genetics, agricultural origins and historical linguistics in Sou1h Asia. In Michael D. Petraglia and Bridget Allchin, eds., The Evolution and History of Hwnan Populations in South Asia: Interdisciplinary Studies in Archaeology, Biological Anthropology, Linguistics and Genetics. Dordrecht: Springer, 393-443. Fuller, Dorian Q., Emma Harvey and Ling Qin. 2007. Presmned domestication? Evidence for wild rice cultivation and domestication in the fifth millermimn BC of the Lower Yang1ze region. Antiquity. 81: 316-331. Harvey, Emma L., DorianQ. Fuller, R.K. Mohanty and Basanta Mohanta. 2006. Early agriculture in Orissa: Some archaeobotanical results and field observations on the Neolithic. Man and Environment. 31.2: 21-32. Haudricomt, Andre-Georges. 1954. Introduction ala phonologie historique des langues miao-yao. Bulletin de l'i!.cole Fr~aise d'Extreme-Orient 44: 555-574. Haudricomt, Andre-Georges, and Louis Hedin. 1987. L 'homme et les plantes cultiwies. Paris: Editions A.-M. Metaili.e. Hazarika, Manjil. 2005. Neolithic culture ofNortheast India with Special Reference to the Origins ofAgriculture and Pottery. Pw;te (Poona): Unpublished Master's thesis, Deccan College. - - . 2006a. Neolithic culture of northeast India: A recent perspective on the origins of pottery and agriculture. Ancient Asia 1: 25-43. - - . 2006b. Understanding the process of plant and animal domestication in northeast India: A hypothetical approach Asian Agri-History 10.3: 203-212. von Heine-Geldem, Robert. 1928. Ein Beitrag zur Chronologie des Neolithikmns in Siidostasien. In Wilhelm Koppers, ed., Festschrift Pater Wilhelm Schmidt: 76 sprachwissenschaftliche, ethnologische, religionswisenschaftliche, priihistoriche und andere Studien. Vierma: Mechitaristen-Congregations-Buchchuckerei, 809-843. - - . 1932. Urheimat mid friiheste Wandenmgen der Austronesier. Anthropos XXVII: 543-619. Jin Liya, Andrey Ganopolski, Fahu Chen, Martin Claussen and Hui.jun Wang. 2005. Impacts of snow and glaciers over Tibetan Plateau on Holocene climate change: Sensitivity experiments with a coupled model of intermediate complexity. Geophysical Research Letters 32: L 17709. Kaiser, Kmlt, Georg Miehe, Werner H. Schoch, Anja Zander and Frank Schliitz. 2006. Relief, soil and lost forests: Late Holocene enviromnental changes in sou1hern Tibet (under human impact). Zeitschriftfiir Geomorphologie, Neue Folge 142: 149-173. Kaiser, Knut, Werner H. Schoch and Georg Miehe. 2007. Holocene paleosols and colluvial sediments in northeast Tibet (Qinghai province, China): Properties, dating andpaleoenviromnental implications. Catena 69: 91-102. Kang Longli and Shengbin Li. 2005. Allele frequencies of 15 STR loci ofLuoba ethnic group in Tibet (southwestern China). Forensic Science Intemationa/155.2-3: 219-221. Kang Longli, Zhao Jiamnin, Liu Kai and Xiaosong Li. 2007a. Allele frequencies of 15 STR loci of Tibetan lived in Tibet Lassa. Forensic Science Jnternationa/168.2: 236-240. Kang Longli, Yuan Dongya, Yang Fengying, Liu Kai and Za Xi. 2007b. Genetic polymorphisms of 15 STR loci in two Tibetan populations from Tibet Changdu andNaqu, China. Forensic Science Intemationa/169.2: 239-243.
LOST IN THE SANDS OF TIME
35
Kraaijenbrink, Thirsa, and George van Driem, Karma Tshering ofGaselo and Peter de Knij:ff. 2006a. Allele frequency distribution for 21 autosomal STR loci in Blmtan. Forensic Science Jnternational170: 68-72. Kraaijenbrink, Thirsa, and George van Driem, Jean Robert M.L. Opgenort, Nirmal Man Tulaclliar and Peter de Knijfi. 2006b. Allele frequency distribution for 21 autosomal STRloci in Nepal. Forensic Science mtemationa/166.2-3: 176-181. Krithika S., R. Trivedi, V.K. Kashyap, P. Bharati and T.S. Vasulu. 2006. Antiquity, geographic contiguity and genetic affinity among Tibeto-Burman-speak:ing populations of India: A microsatellite study. Annals ofHwnan Biology 33.1: 26-42. Krithika S., R. Trivedi, V.K. Kashyap and T.S. Vasulu. 2005. Genetic diversity at 15 microsatellite loci among the Adi Pasi population of Adi tribal cluster in Arunachal Pradesh, India. Legal Medicine 7.5: 306-310. Krithika S., R. Trivedi, V.K. Kashyap and T.S. Vasulu. 2007a. Genotype profile for fifteen tetranucleotide repeat loci in two Tibeto-Burman speaking tribal populations of Anmachal Pradesh, India. Journal ofForensic Science 52.1: 239-241. Krithika S., S. Maji and T.S. Vasulu. 2007b. Intertribal and temporal allele-frequency variation at the ABO locus among Tibeto-Burman-speak:ing Adi subtribes of Arunachal Pradesh, India. Human Biology 79.3: 355-362. - - . 2008. A microsatellite guided insight into the genetic status of Adi, an isolated hunting-gathering tribe of northeast India. Public Library of Science One 3.7: e2549. doi: 10.1371/journal.pone.0002549. Kuhle, Matthias. 1985. Glaciation research in the Himalayas: a new ice age theory. Universitas 27: 281-294. - - . 1986. The upper limit of glaciation in the Himalayas. GeoJouma/13: 331-346. - - . 1987. The problem of a Pleistocene inland glaciation of the northeastern Qinghai-Xizang Plateau. In J. Hovermann and W. Wenying, eds.,Reports ofthe Qinghai-Xizang (I'ihet) Plateau. Peking: Science Press, 250-315. - - . 1988a. Geomorphological findings on the build-up of Pleistocene glaciation in Southern Tibet and on the problem of inland ice. GeoJouma/17: 457-512. - - . 1988b. Topography as a fimdamental element of glacial systems. GeoJoumal 17: 545-568. - - . 1990a. The cold deserts of high Asia (fibet and contiguous mountains). GeoJournal20: 319-323. - - . 1990b. Ice marginal ramps and alluvial fans in semi-arid mountains: Corwergence and difference. In A.H. Rachocki. and M. Clmrch, eds., Alluvial Fans: A FieldApproack Chichester: Wtley & Sons Ltd, 55-68. - - . 1991. Observations supporting the Pleistocene inland glaciation of High Asia. GeoJouma/25: 131-231. - - . 1995. Glacial isostatic uplift of Tibet as a consequence of a former ice sheet. GeoJouma/37.4: 431-449. - - . 1997. New findings concerning the Ice Age (Last Glacial Maximum); Glacier cover of the East-Parnir, of the Nanga Parbat up to the Central Himalaya and of Tibet, as well as the age of the Tibetan inland ice. GeoJouma/42.2-3: 87-257. - - . 1999. Reconstruction of an approximately Quaternary Tibetan inland glaciation between the Mt. Everest and the Cho Oyu massifs and the Aksai Chin: A new glaciogeomorphological SE-NW diagonal profile through Tibet and its consequences for the glacial isostasy and Ice Age cycle. GeoJournal47: 2-276. Kuhle, Matthias. 2005. The maximum Ice Age (Wiirmian, Last Ice Age, LGM) glaciation of the Himalaya: A glaciogeomorphological investigation of glacier trimlines, ice thicknesses and lowest former ice margin positions in the Mt. EverestMakalu-Cho Oyu massifs (Klmmbu and Klmmbakarna Himal) including inforrna-
36
GEORGE VAN DRIEM
tions on late glacial, neoglacial and historical glacier stages, their snowline depressions and ages. GeoJournal62: 193-650. Kmnar, Vikrant, Arimanda N.S. Reddy, Jagedeesh P. Babu, Tiprisetti N. Rao, Bamida T. Langstieh, Kmnarasamy Thangaraj, Alla G. Reddy, Lalji Singh and Battini M. Reddy. 2007. Y-chromosome evidence suggests a common paternal heritage of Austro-Asiatic populations. BioMed Central Evolutionary Biology 7: 47 (14 pp.) http://www.biomedcentral.com/1471-214817/47 Lehmkuhl, Frank. 1995. Geomorphologische Untersuchungen zum Klima des Holoziins und Jungpleistoziins Osttibets. Gottinger Geographische Ahhandlungen 102: 1-184. Lehmkuhl, Frank, and Lewis A. Owen. 2005. Late Quaternary glaciation ofTibet and the bordering mountains: A review. Boreas 34: 87-100. Maji, Suvendu., S. Krithika and T .S. Vasulu. 2007. Genetic kinship among an isolated Adi. tribe of Anmachal Pradesh: isonymy in the Adi Panggi. Human Biology 79.3: 321-337. Miehe, Georg, Sabine Miehe, Frank Schliilz, Knut Kaiser and La Duo. 2006. Palaeoecological and experimental evidence of former forests and woodlands in the treeless desert pastures of southern Tibet (Lhasa, A.R. Xizang, China). Palaeogeography, Palaeoclimatology, Palaeoecology 242: 54-67. Nasu Hiro, Arata Momohara, Yoshinori Yasuda and Jiejun He. 2006. The occurrence and identification of Setaria italica (L.) P. Beauv. (foxtail millet) grains from the Chengtoushan site (ca. 5800 cal B.P.) in central China, with reference to the domestication centre in Asia. Vegetation History andArchaeohotany 16.6: 481-494. Niederer, Barbara. 1998. Les langues hmong-mjen (miao-yao): Phonologie historique. Miinchen: Lincom Emopa. Osada Toshiki.. 1995. Mundqjin no NOkO Bunka to Shokuji Bun/ca.· Minzoku Gengogakuteki K&atsu. ('The Rice and Food Culture of Munda in Eastern India: An Ethnolingui.stic Study'). Kyoto: Kokusai Nihon Bunka Kenkyii Senta. Ostapirat, Weera 2005. Kra-Dai and Austronesian: Notes on phonological correspondences and vocabulary distribution. In Lament Sagart, Roger Blench and Alicia Sanchez-Mazas, eds., The Peopling of East Asia: Putting Together Archaeology, Linguistics and Genetics. London: Routledge Cmzon, 107-131. Owen, A., Robert C. Finkel and Marc W. Caffee. 2002. A note on the extent of glaciation in the Himalayas dming the global Last Glacial Maximum. Quaternary Science 21: 147-157. Owen, Lewis A., Robert C. Finkel., Ma Haizhou, Joel Q. Spencer, Edward Derbyshire, Patrick L. Barnard and Marc W. Caffee. 2003. Timing and style of Late Quarternary glaciation in northeastern Tibet. Geological Society of America Bulletin 115.11: 1356-1364. Pal, J.N. 1990. The early farming culture of northern India. Bulletin ofDeccan College Post-Graduate and Research Institute 49: 297-304. Parkin, Emma J., Thirsa Kraayenbrink:, Karma Tshering ofGaselo, George van Driem, Peter de Knijffand Mark Jobling. 2006a. 26-locus Y-STR typing in a Bhutanese population sample. Forensic Science .buemational161.1: 1-7. Parkin, Emma J., Thirsa Kraayenbrink, Jean Robert M. L. Opgenort, George van Driem, Nirmal Man Tulaclliar, Peter de Knijffand Mark Jobling. 2006b. Diversity of 26 Y-STR haplotypes in a Nepalese population sample: Isolation and drift in the Himalayas. Forensic Science .buemational166.2-3: 176-181. Poloni, Estella Simone, et al. 1997. Hmnan genetic affinities for Y chromosome P49a,f/Taqi haploptypes show strong correspondence with linguistics. American
LOST IN THE SANDS OF TIME
37
Journal of Hwnan Genetics 61: 1015-1035 (cf. the erratum published in 1998 in theAmericanJournal ofHwnan Genetics62: 1267). - - . 2000. Languages and genes: Modes of transmission observed through the analysis of male-specific and female-specific genes. In Jean-Louis Dessalles and Laleh Ghadakpom, eds., Proceedings: Evolution ofLanguage, 3rd International Coriference 3-6April2000. Paris: Ecole Nationale Supeneure des Telecommunications, 185-186. Pmnell, Herbert C., Jr. 1970. Toward a Reconstruction of Proto Miao-Yao. Ithaca: Cornell University Ph.D. dissertation Qamar, Raheel, Qasim Ayub, Aisha Mohyuddin, Agnar He]gason, Kehkashan Mazhar, Atika Manoor, Tatiana Zerjal, Chris Tyler-Smith and S. Qasim Mehdi. 2002. Ychromosomal variation in Pakistan. American Journal of Human Genetics 70: 1107-1124. Quintana-Mmci, Uuis, C. Krausz, Tatiana Zerjal, S. Hamid Sayar, Michael F. Hammer, S. Qasim Mehdi, Qasim Ayub, Raheel Qamar, Aisha Mohyuddin, U. Radhakrishna, Mark A Jobling Chris Tyler-Smith and Ken McElreavey. 2001. Y-chromosome lineages trace diffusion of people and languages in southwestern Asia. American Journal ofHuman Genetics 68: 537-542. Ratliff, Martha. 2004. Vocabulary of enviromnent and subsistence and in the HmongMien protolanguage. In Nicholas Tapp, Jean Michaud, Christian Culas and Gary Yia Lee, eds.,Hmong/Miao inAsia. Chiang Mai: Silkworm Books, 147-165. Revel, Nicole. 1988. Le riz enAsie du sud-est (3 vols.). Paris: Editions de l'Ecole des Hautes Etudes en Sciences Sociales. Sahoo, Sanghamitra, Anamika Singh, G. Himabindu, Jheeman Banerjee, T. Sitalaximi, Sonali Gaikwad, R. Trivedi, Phillip Endicott, Toomas Kivisild, Mait Metspalu, Richard Villems and V. K. Kashyap. 2006. A prehistcny of Indian Y chromosomes: Evaluating demic diffusion scenarios. Proceedings of the National Academy ofSciences 103.4: 843-848. Schiifer, J.M., S. Tschiidi, Z. Zhizhong, W. Xihao, W., S. Ivy-Ochs, R Wieler, H. Bam, P.W. Kubik and C. Schliichter. 2002. The limited influence of glaciations in Tibet on global climate over the past 170,000 years. Earth and Planetary Science Letters 194: 287-297. Sharma, G.R, V.D. Misra, D. Mandai, B.B. Misra and J.N. Pal. 1980. Beginnings of Agriculture: From Hunting and Food Gathering to Domestication ofPlants and Animals. Allahabad: Abhinav Prakashan. Shi Hong, Dong Yong-li, Wen Bo, Xiao ChlDl-Jie, Peter A Underhill, Shen Peidong, Ranajit Chakraborty, rm Li and Sil s-mg. 2005. Y-chromosome evidence of southern origin of the East Asian-specific haplogroup 03-Ml22. American Journal ofHwnan Genetics 77: 408-419. SU s-mg, Xiao Junhna, Peter Underhill, Ranjan Deka, Zhang Weiling, Joshua Akey, Huang Wei, Shen Di, Dam Lu, Luo Jingclrun, Clm Jiayou, Tan Jiazhen, Shen Peidong, Ron Davis, Luca Cavalli-Sforza, Ranajit Chakraborty, Xiong Momiao, Du Ruofu, Peter Oefner, Chen Zhu and rm Li. 1999. Y-chromosome evidence for a northward migration of modern lmmans into eastern Asia during the last Ice Age. American Journal ofHwnan Genetics 65: 1718-1724. SU s-mg, ChlDljie Xiao, Ranjan Deka, Mark T. Seielstad, Daoroong Kangwanpong, JlDllma Xiao, Daru Lu, Peter Underhill, L uca Cavalli- Sforza, Ranajit Chakraborty and rm Li. 2000. Y chromosome haplotypes reveal prehistorical migrations to the Himalayas. Human Genetics 107.6: 582-590. Tewari, Rakesh, RK. Srivastava, K.K. Singh, K.S. Saraswat and I.B. Singh. 2002. Preliminary report of the excavation at Lalmradewa, District Sant Kabir Nagar,
38
GEORGE VAN DRIEM
U.P. 2001-2002: Wider archaeological implications. Pragdhiirjj, Journal of the Uttar Pradesh Archaeological Department 13: 37-76. Thompson, L.G., E. Mosley-Thompson, M.E. Davis, J.F. Bolzan, J. Dai, L. Klein, T. Yao, X.Wu, Z. Xie and N. Gundestrup. 1989. Holocene-Late Pleistocene climatic ice core records from Qinghai-Tibetan Plateau. Science 246 (4929): 474-477. Vandenberghe, Jef 2007. Permafrost distribution during the Last Glacial Maximum in China and smroundings, paper presented at the symposium Cenozoic Climate and Landscape Evolution and Blrutan and Tibet, University of Amsterdam, 21 May 2007. Wang Fiishi and Mao Zongwii. 1995. Mido-Yao Yu GUyin de GOuni ['Phonological Reconstruction ofHmong-Mien']. Peking: Zhonggu6 Shehui Kexue Chfibanshe. Wen Bo, Xie Xuanhua, Gao Song, Li Hui., Shi Hong, Song Xiufeng, Qian Tingzhi, Xiao Clnmjie, Jin Jianzhong, Su llmg, Lu Darn, Rana jit Chakraborty and Tm Li. 2004. Analyses of genetic structure of Tibeto-Bunnan populations reveals sexbiased admixture in southern Tibeto-Burmans. American Journal of Human Genetics 74.5: 856-865. Wu Yanhong, Andreas Liicke, Jing Zhangdong, Wang Sumin, Gerhard H. Schleser, Richard W. Batterbee and Xia Weilan. 2006. Holocene climate development on the central Tibetan Plateau: A sedimentary record from Cuoe Lake. Palaeogeography, Palaeoclimatology, Palaeoecology 234.2-4: 328-340. Yan Jiangwei, Chumnei Shen, Yuanzhe Li, Xiaoguang Yu, Xin Xiong, Haofang Mu, Yanqing Huang, Yajun.Deng and J1D1 Yu. 2007. Genetic analysis of 15 STRloci on Chinese Tibetan in Qinghai Province. Forensic Science lnternationa/169 .1: e3-e6. Yasuda Yoshinori 2002. Origins of pottery and agriculture in East Asia. In Yoshinori Yasuda, ed., The Origins ofPottery andAgriculture. New Delhi: Lusre Press and Roli Books, for the International Center for Japanese Studies, 119-142. Zerjal, Taljana, B. Daslmyam, A. Pandya, Manfred Kayser, Lutz Roewer, F. R. Santos, W. Schiefenhovel, N. Fretwell, M.A. Jobling, S. Harihara, K. Shimuzu, D. Semjidmaa, Antti Sajantila, P. Salo, M. H. Crawford, E. K. Ginter, Oleg V. Efgrafov and Chris Tyler-Smith. 1997. Genetic relationships of Asian and Northern Emopeans, revealed by Y chromosomal DNA analysis. American Journal of Human Genetics62: 1174-1183. Zhao ZhijUn.. 2005. Zhiwil kaogiixu.e jiqi xin jinzl:Jan ('Palaeobotany and its recent progress'). Kaogii 2005 (7): 42-49. Zide, Norman H., and Gregory D.S. Anderson 1999. The Proto-Munda verb system and some COlUlections with Mon-Khmer. In Working Papers of the International Symposium on South Asian Languages: Contact, Convergence and Typology, Tokyo, December 6-9, 1999. Tokyo: Institute for the Study of Languages and Cultures of Asia and Africa, 400-419.
- - . 2003. Towards an analysis of the South Munda verbal system. In Karen L. Adams, Thomas John Hudak and F.K. Lehman, eds.,Papersfrom the SeventhAnnual Meeting of the Southeast Asian Linguistics Socie~. Tempe, Arizona: Arizona State University Program for Southeast Asian Studies, 243-259. Zong, Y., Z. Chen, J. B. Innes, C. Chen, Z. Wang and H. Wang. 2007. Fire and flood management of coastal swamp enabled first rice paddy cultivation in east China. Nature 449: 459-462.
PART TWO: PHONOLOGY AND SCRIPT
A KEY TO FOUR TRANSCRIPTION SYS1EMS OF LEPCHA HELEEN PLAISIER
1. INTRODUCTION The native Lepcha orthography is believed to have been devised during the reign of the third chOgel of Sikkirn, i41·~12l'~~.:z::~SI'~I2l' Chagii Chado 'Namgii (imperabat 1700-1716). According to Lepcha tradition, the Lepcha script was created by the Lepcha scholar 5~o<-e3 ~<® ThikUngMensal6ng, who is thought to have been a contemporary ofS' S1'1j'~~~·Cf 'Lama Lhatsii.n Chenpo, i.e. §I'SI'1j'~'~'SI~~'il.c;· 'Lama Lhatsii.n Namkha Jimi (1597-1654), the patron saint of Sikkirn. The Lepcha script is written from left to right. Erik Haarh's claim that Lepcha was written in vertical columns from left to right is not corroborated by any evidence and is therefore not convincing (Haarh 1959). In Lepcha, no distinction is made between capital and lowercase letters. Punctuation marks are similar to the ones used in Tibetan orthography, although nowadays full stops, commas and question marks from the Roman alphabet are also used. The native Lepcha orthography is systematically treated in the text ~ laz6ng, the book on the Lepcha alphabet, which is traditionally used to teach Lepcha script (Plaisier 2003: 31-32). When the native Lepcha orthography is not available to or not preferred by someone who wants to write Lepcha words, names or text, a transcription may be used. Most transcriptions of the Lepcha script are romanizations, this is when the Lepcha language is written down using letters of the Roman alphabet. Lepcha is sometimes also transcribed using the Devanagati script. Lepchas themselves may prefer not to use any transcription at all and tend to stick to the native Lepcha orthography as much as possible. If they do need to render a Lepcha name or word in the Roman or Devanagari script, many of them use spontaneous and arbitrary transcriptions. Such transcriptions usually represent a phonetic approximation of Lepcha words. It is common to find many different transcriptions representing single Lepcha words or names. For example, the name '*~ c#' may be found transcribed as dorjee tsering, dorfee tshering, dorji chiring, etc. Likewise, the word «r-€:), meaning 'a Lep-
42
HELEEN PLAISIR
cha ', may be transcribed as r6ngkup, rangcup, raongkaup, etc. In daily life this practice is not a major problem, although it may lead to confusion when one attempts to derive the exact pronunciation of words from a transcription, which may ultimately lead to problems in understanding of the transcribed words. Those people who are not familiar with native Lepcha orthography but have learned to read and write the Roman or Devanagari script might fmd a standardised transcription of the Lepcha language particularly helpful, as they do not have the native Lepcha orthography to refer back to. For scholars engaged in Lepcha studies, such as historians, linguists and anthropologists, a transcription that merely approximates the pronunciation of the Lepcha language is clearly inadequate. It is important for such scholars to use a system of transcription that accurately reflects the original language and enables them to distinguish between different words, names and even different spellings of words. For any scholar of Lepcha language, history, literature or religion, it is also important to be able to derive the spelling in the original Lepcha orthography from a transcription. In order to be able to ascertain precisely how a particular word is written in the native orthography, an unambiguous rendering of the indigenous orthography of Lepcha words is necessary. A transcription of the pronunciation of a word may be called a phonetic or phonological transcription. When a transcription carefully substitutes the native orthographical symbols into the symbols of another writing system, following the same spelling conventions and particularities that are used in the native orthography, this is known as a transliteration. In the discussion below, the four most commonly used transcription systems of the Lepcha language are presented alongside one another for comparison and together with the Lepcha script The phonetic transcription given along with the transcriptions is based on the phonetic description of the Lepcha language in the recently published A Grammar of Lepcha (Plaisier 2007). The four transcriptions discussed here are those by George Byres Mainwaring as introduced in his grammar of Lepcha, Albert Griinwedel as used in his edition of the LepchaEnglish dictionary written by Mainwaring, the native Lepcha scholar
A KEY TO FOUR TRANSCRIPTION SYSTEMS OF LEPCHA
43
erence is given, Lepcha is transcribed according to the conventions of the present author's transliteration (Plaisier 2007).
2. INITIAL CONSONANTS The Lepcha orthography uses 36 consonant symbols to indicate initial consonants. These ld~ &<:o choming .?amo or &:o3 &~ .?amu .?aming 'consonants, mother letters' are given in Table 2 below with their corresponding transcriptions. When the consonant is not marked by any vowel sign, the inherent vowel of the consonant symbol is the vowel transcribed as 'a' In all tables, phonetic transcriptions are given between square brackets. In the Plaisier transliteration of the Lepcha orthography, the symbol& is treated as a consonant symbol that indicates a glottal stop and is transcribed by '?'. This transcription allows for a clear distinction between syllables with an initial glottal stop followed by a vowel or followed by a post-consonantal glide -y- and syllables with an initial y-, and uses a minimum of symbols, as shown in Table 1. The transcription of the consonants in the Plaisier system also differs from the transcription used by George Byres Mainwaring in that Mainwaring uses cha for the consonant symbol "· and chha for the symbol .2C. In his publications on Lepcha language and culture, Kharpu Tamsang uses cha for"· chya for .2C,jya for (G", and shya for~- Albert Griinwedel's approach is slightly different: he uses the apostrophe sign to indicate what might be described as aspiration, using k'a for (S, t'a for.,.., p 'a for J:3, and ts 'a for Ut-. Griinwedel uses ca for"· c'a for .2C, m for,., and sa for~Table 1 Transcription of initial glottal stop Lepcha Mainwaring Griinwedel Tamsling Plaisier
11
l!l
f.
a a a ?a [?:l]
aya 'aya aya ?ya [?j:J]
ya ya ya ya [j:J]
44
HELEEN PLAISIR
Table 2 Transcription of initial consonant symbols of:
(j
(.)
t
ka ka ka ka [ka]
kha
ga ga ga ga [ga]
nga
Lepcha Mainwaring Grtinwedel Tamsang Plaisier
"
"
(II"
~
chya cha [c"a]
Ja ja Jya Ja [3a]
nya nya nya nya [p.a]
Lepcha Mainwaring Grtinwedel Tamsang Plaisier
~
"' tha
*da
(J
da da da [da]
na na
[ta]
t'a tha tha [t'a]
;:J
J:3
9
0
pa pa pa pa [pa]
pha p'a pha pha [pha]
fa Ia fa fa [fa]
ba ba ba ba [ba]
:ij
0'
(JI'
....
rna rna rna rna [rna]
tsa tsa tsa tsa [1sa]
tsha ts'a tsha tsha [1sha]
Lepcha Mainwaring Grtinwedel Tamsang Plaisier
Lepcha Mainwaring Grtinwedel Tamsang Plaisier Lepcha Mainwaring Grtinwedel Tamsang Plaisier
cha ca cha ca [ca]
ta
ta ta ta
k'a
kha kha [IC'a] chha
C'a
na nga nga [IJa]
na
na [na]
za
z.a za za [za]
A KEY TO FOUR TRANSCRIPTION SYSTEMS OF LEPCHA
Table2 Lepcha Mainwaring Griinwedel Tamsang Plaisier Lepcha Mainwaring Griinwedel Tamsang Plaisier Lepcha Mainwaring Grtinwedel Tamsang Plaisier Lepcha Mainwaring Griinwedel Tamsang Plaisier
45
continued
,-
f.
~
\F
[m]
la la la la [l:l]
ha ha ha ha [h:l]
"sha
""sa
sa shya sha [J:l]
sa sa sa [S:l]
wa wa wa wa [w:l]
Jj
~
e;
kla kla kla kla [kl:l]
gla gla gla gla [gl:J]
pla pla pla pla [pl:J]
ya ya ya ya [j:J]
ra
e va va va va [v:l] ;s. a a a ?a [?:l]
ra ra ra
~
~
(')
:J
X
fla fla fla fla [fl:l]
bla bla bla bla [bl:J]
mla mla mla mla [ml:J]
hla hla hla hla [l:J]
3. INITIAL CONSONANT CLUSTERS For initial consonant clusters, the Lepcha term ~ mingth)'U 'conjunct consonants, affixed consonants' is used. As shown in Table 2, the native Lepcha orthography distinguishes a series of special consonant symbols that may be transcribed as initial consonants followed by a post-consonantal glide -1. In traditional Lepcha terminology, this series of '5 Ida, ~ gla, e:. pla, ~ fla, M bla, :1 mla and x hla, is sometimes referred to as the~ lath)'U '1-cluster' or '1-affix'.
46
HELEEN PLAISIR
In Lepcha, an initial consonant may also be followed by the postconsonantal glide -r- or -y- for which the orthography has special symbols, i.e. the n-.3 rathyU. 'r-affix' and the fR3 yathyU. 'y-affix' shown in Table 3 below. An initial consonant cluster consisting of an initial consonant and either a lathyU. or rathyU. may also be followed by a yathyU.. Table 4 lists all initial consonant clusters that are meaningful in Lepcha.
Table 3 Affixed signs
c;,
Lepcha Mainwaring Grtinwedel
Tamsang Plaisier
o)
-y-y-y-y-
-r-r-r-r-
[ -j-]
[-r-]
Table 4 Initial consonant clusters -&kya <» khya
'£)! krya
~gya
~grya
~gla
~glya
;::s, pra
;:J,l prya
t:;;pla
.c::;,plya
s; fra
s, frya
'11 klya
~tya
~qrthya
..., dya ;31 pya
O'phya 611 fya Olbya ':Simya 1"' rya
0) bra
mrya
:oJ mra
:c}o
~hra
-lJ)!hrya
~lya
Vii hya XII hlya ~vya
'81 ?ya
~fla
~flya
Mbla iJ'mla
Mlblya iJI' mlya
A KEY TO FOUR TRANSCRIPTION SYSTEMS OF LEPCHA
47
4. RETROFLEX CONSONANTS The Lepcha language distinguishes three retroflex consonants, which occur mainly in loanwords from Tibetan. The retroflex series are written in the native Lepcha orthography as ii kr for /tr/, -ltj hr for /thr/ and ~ gr for /dr/, as shown in Table 5. In Lepcha orthography, the retroflex consonants are often written with a small dot below the consonant cluster, so as to distinguish these sounds from the unmarked non-retroflex sounds kra, hra and gra, which occur widely in Lepcha. This useful diacritic was introduced by Mainwaring, and is known as 'Mainwaring's dot' (Mainwaring 1876: 10-11). Mainwaring's dot is still widely used to indicate retroflex sounds in Lepcha orthography, although its usage with the letters -t za and 1 ra, as suggested by Mainwaring, is no longer found (Mainwaring 1876: 11).
Table5 Retroflex consonants Lepcha Mainwaring Grtinwedel Tamsang Plaisier
ii
-'6
~
lqa
h{a
g:ra
*kra
*bra
*gra
ta tra [ta]
hta thra [fa]
da dra [eta]
Mainwaring also uses the dot to indicate retroflex sounds in his transcription. As shown in Table 5, the dot is placed below the syllable that contains an initial retroflex consonant. In Griinwedel's transcription, the retroflex sounds are not distinguished from the consonant clusters kra, gra and hra, but Griinwedel does mark each form that he believes to be a Tibetan loan with an asterisk*. Ta.msang's transcription does not distinguish between the orthograpical sequences ~ and iii, both of which he transcribes as ta, nor does he distinguish between and~ both of which he transcribes as da. Tamsang transcribes the orthograpical sequence ~ as hta. The present author distinguishes the retroflex consonants systematically by transcribing '£1 as kra, ii as tra, -lJ} as hra, -ltj as thra, ~ as gra and ~ as dra.
*
48
HELEEN PLAISIR
5. VOWELS
In the Lepcha orthography, vowel diacritics may be added to consonant symbols. The nine vowel diacritics are traditionally known as <.2C~ &-€:) choming Jakup 'vowel signs' or &-€:) &~ Jakup ?inning 'vowel signs, child letters, small letters' and &-€:) ~5~ ?a/cUp thambyin or~ ~5& mingkup thambyin 'diacritical vowel signs'. As illustrated in Table 6, different transcriptions vary considerably in their treatment of Lepcha vowels. The r- ran 'circumflex' sign is a diacritic flourish written over a consonant sign or over a vowel sign 5, as in r- ran, ! ?a or 5! a. The ran sign is often present in closed syllables, in which case the circumflex sign should be written above the final consonant sign. The fimction of the ran sign in the Lepcha orthography remains unclear, it might serve to point to the presence of a fmal consonant diacritic, but it has also been suggested that the function of the ran is to indicate stress or pitch, in order to distinguish stressable syllables from syllables that never appear in a stressed position (Plaisier 2003: 28-29, Sprigg 1983: 316). Nowadays, the usage ofthe circumflex sign appears to be unsystematic and even unpredictable. A closer study of the usage of the ran sign in old texts is likely to shed more light on this matter. Mainwaring does not systematically romanise the ran sign, and he had the Lepcha symbol printed over transcribed words, as in r- ran. In his Lepcha-English Encyclopedic Dictionary, KMrpu Tamsang does not romanise the ran sign either, although from examples in the Lepcha orthography it is evident that he uses the diacritic systematically in closed syllables where no vowel diacritic is used. KMrpu Tamsang sometimes uses the r- ran in open syllables ending in -a, i.e. in syllables where no vowel diacritic is used, which implies that he believes the difference between open syllables ending in -a and -a to be phonological. Griinwedeltranscribes the ran as an acute symbol over the vowel. The transcription proposed by the present author is to transcribe the ran flourish as a circumflex sign, in line with the approach originally suggested by Mainwaring. From Table 6 it can also be seen that the native Lepcha orthography is not wholly phonological. On the one hand, the orthography does not adequately distinguish between the sounds [e] and [e]. On the other hand, the native orthography distinguishes between symbols that do not, or no longer, represent a phonological distinction, i.e. Jo i vs.
A KEY TO FOUR TRANSCRIPTION SYSTEMS OF LEPCHA
49
5o i, and o avs. a. In old manuscripts, the symbols So i and So i are used in a systematic manner: jo i is used only in closed syllables and 5o i only in open syllables. Since Mainwaring's time, and possibly through his influence, the distribution of the orthographic symbols SO i and 5o i has become much more random. Table 6 Transcription of vowels Lepcha Mainwaring Grtinwedel Tamsang Plaisier
Lepcha Mainwaring Grtinwedel Tamsang Plaisier
Lepcha Mainwaring Grtinwedel Tamsang Plaisier
-a
-a -a
-a [:l]
Sn,_, -1 -1 -ae
!..wl
c;(
'i
-a -a -a
-a -a -aa
-e -e -aey
-a
-a
-e
[:l]
[a]
[e-E]
,_, sr--: -1 -i
,_,
""'-o -0
-a1
-ao
-1
-i
-o
[i]
[i]
[o]
Cj -u -ii
,_,
co' -6 -6 -aao
-6 [()]
n3
-u -u
-au
-aoo
-u [w]
-u [u]
6. FINAL CONSONANTS Although all Lepcha consonants may be found at the beginning of a syllable, only a limited number of consonants can occur syllable-finally. The consonants found at the end of syllables are: -k, -t, -n, -p, -m, -r, -1 and -ng. The ~ ~s& tyelbU thambyin or ~ ~s& mingtyel thambyin 'final consonant signs' are given in Table 7. The nyindo sign, i.e. ·~·, transliterated here as -ang, is used to indicate a fmal velar nasal when no specific vowel sign is indicated, as in ~ sang. The ~ lakang sign is used to indicate a final velar nasal
Sm
50
HELEEN PLAISIR
in combination with diacritical vowel signs, as in the syllables """ sang, &;; sing and """ song. The distinction between the nyindo and the lakang is not always made, often the nyindo is used as a general term for a written ftnal velar nasal. Kharpu Tcimsang chooses to avoid the nyindo sign altogether; and where others spell~ sang, he spells ""'·
Table 7 Transcription of fmal consonants (name of symbol) Lepcha Mainwaring Grtinwedel Tamsang Plaisier (name of symbol) Lepcha Mainwaring Grtinwedel Tamsang Plaisier (name of symbol) Lepcha Mainwaring Grtinwedel Tamsang Plaisier
~lakat
~lanyet
~lasam
~t !..wl
!..wl
!..wl
[-k]
-m -m -m -m [-m]
-1 -1 -1 -1 [-1]
~)lanun
.
~lakup
~ladar
Ci
(:j
Ci
-n -n -n [-n]
-p -p -p -p [-p]
-r -r -r -r [-r]
~lakat
~lakang
'l#nyindo
-k -k -k
-k
-n
N
,_, 0'"'
Ci
-t -t -t
-t [-t]
l!.
-ng
-ang
-Ji
-an
-ng -ng [ -l]]
-ang -ang [-at]]
7. MULTISYLLABIC WORDS
In the Plaisier transliteration, syllable boundaries in multisyllabic words are indicated by a hyphen only in those cases where they cannot be predicted, or when it is necessary to separate phoneme symbols which might otherwise be read as a digraph for another phoneme, e.g. ~ 1Uk-hr6ng 'morning' or~ zUk-Jat 'action'. If a syllable ends in a vowel, the syllable boundary is not indicated, e.g. &r .?are 'this' or &S~ .?akrim
A KEY TO FOUR TRANSCRIPTION SYSTEMS OF LEPCHA
51
'bitter' Griinwedel uses hypens to indicate any syllable bmmdary. Mainwaring uses hyphens as well as spaces between syllables in his transcriptions, Tcimsang never uses hyphens, but uses spaces between some syllables and no spaces between other syllables. Both Mainwaring and Tamsang do not explain their conventions in this respect. 8. CONCLUSION
This concludes the discussion of the Lepcha script and the four most widely used transcription systems. It is hoped that this overview may ease comparisons of transcribed Lepcha forms used in the literature and may also help those who wish to use a transcription system to do so in a consistent manner. The following short sentences in Table 8 are transcribed according to the four different conventions treated here and may serve as a further illustration of the transcription systems outlined above. The original Lepcha sentences and the English translations, as well as Mainwaring's transcriptions, are derived from George Mainwaring's grammar (Mainwaring 1876: 143). The remaining transcriptions were created by the present author according to the conventions given by Griinwedel, Tamsang and Plaisier described above. In this article, it is proposed that the transliteration system designed by the present author offers the user the most accurate way of transcribing Lepcha. The transliteration is consistent with and faithful to the way text is written in the traditional Lepcha orthography, so it remains possible at all times to derive the original spelling from the transliteration. The transliteration systems used by Mainwaring, Griinwedel and Tamsang are ambiguous, since none of them offer a straightforward one-to-one correspondence with the native Lepcha orthography.
52
HELEEN PLAISIR
Table 8 Example sentences
(}" v--s~ s!
Lepcha
~~ ~
Mainwaring
tuk-tak ki, man ngo, sa hik.ll tiik-tak lci, man ito, sa-hik. i1 tauktak. ka, man ngao, sahaek ail tuktak ka, man ngo, sahik. ?il
Gtiinwedel Tamsang Plaisier (English) Lepcha
Mainwaring Grunwedel Tamsang Plaisier (English)
'Boil the soup, boil the meat, and roast the fowl. '
~ "" &c-3 .\3 man sa a-chom zUk man sa a-com zuk man sa aachaaom zaook man sa ?acorn .zUk
'Make a meat stew.'
Mainwaring
u3 "".) 63 &3 "" •"".)5-E1 §3 khu su, bUk byilp, sa sung-l.crlayilr
Grunwedel Tamsang Plaisier
khu sii, buk byup, sa sfui-kri 'yur khaoo sau, baook byaoop, sa saung krai ayaoor khu su, bUk byilp, sa sungkri ?yUr
(English)
'Bake the bread and bake the potatoes in the ashes, also the arums in ditto. '
Lepcha
Lepcha
Mainwaring Grunwedel Tamsang Plaisier (English)
""~ (-t &Cf ~ ~ ~) sa-gang zo, a-zorn rna fam mun sa-gang zo, a-zorn rna lam mun sagang zao, aazaaom rna fam maun sagang zo, ?azom rna fam mun
'Eat slowly, do not gobble your food. '
A KEY TO FOUR TRANSCRIPTION SYSTEMS OF LEPCHA
53
BIBLIOGRAPHY Griinwedel, Albert, ed. 1898. Dictionary of the Lepcha Language. Berlin: Unger Brothers. Haarh, Erik. 1959. The Lepcha Script. Acta Orientalia. (ed Societates Orientalia Batavia Danica Norvegica.) 24: 107-122. Mainwaring, George Byres. 1876. A Grammar of the Rong (Lepcha) Language as it Exists in the Dorjeling and Sikim Hills. Calcutta: Baptist Mission Press. Plaisi.er, Heleen. 2003. Catalogue of Lepcha Manuscripts in the Van Manen Collection. (Kern Institute Miscellanea 11.) Leiden: Kern Institute. _ _ . 2007. A Grammar ofLepcha. (fibetan Studies Library 5/5: Languages of the Greater Himalayan Region.) Leiden: Brill. Sprigg, Richard Keith. 1983. Hooker's expenses in Sikkim, an early Lepcha text. Bulletin ofthe School ofOriental andA.fricanStudies46.2: 305-325. Tams8ng, Khlirpll. 1980. The Lepcha-English Encyclopedic Dictionary. Kalimpong: Mayel Clymit Tamsang.
DIALECTAL PARTICULARITIES OF SOGPHO TIBETANAN INTRODUCTION TO 1HE ''TWENTY-FOUR VILLAGES' PATOIS" HIROYUKI SUZUKI
1. INTRODUCTION Sogpho Tibetan is a Khams Tibetan variety spoken in Danba Cowtty in western Sichuan, China, a region that borders the Tibetan and rGyaJrong speaking area and Han China. This paper explores the phonological system and provides an analysis of its peculiar phonetic and lexical features. 1 1.1. Location and cultural background Danba Cowtty is one of the most eastern cowtties in Ganzi Prefecture, western Sichuan, China. This place is traditionally populated by rGyalrong Tibetans, who consider it to be a central spiritual area, where the holy mowttain Moerduo [dMu-rdo] lies. The Tibetans in Danba call themselves rGyalrong, and their culture, traditions, and heritage are, in fact, similar to those of other rGyalrong people who speak rGyalrong languages. 2 The architectural heritage treated in Darragon (2005) clearly indicates some common traits between the rGyalrong and eastern Khams areas, including Danba. 1.2. Languages spoken in Danba The linguistic situation in Danba can best be described as quite complex, since five different languages are spoken by the Tibetan communities of the area: - Khams Tibetan (see 1.3) - Amdo Tibetan, spoken in the nomadic area - Geshitsa, spoken along the Geshiza River, in Niega and Bawang and other villages 3 1 2 3
All data concerning Sogpho Tibetan were collected by the author. For detailed information, see Yang (2005). See rDo-rje (1998) for a comprehensive description of this language.
56
HIRUYUKI SUZUKI
- Situ-rGyaJrong, spoken in Badi and Taipingqiao villages, among others - Chinese (as a regional common language) 1.3. "Twenty-four villages' patois" ''Twenty-four villages' patois" is the local name for a Khams Tibetan dialect group in Danba. This variety is spoken in several villages, such as Suopo (Sog-pho), Gezong (dGu-rdzong), Zhonglu (sPro-snang), Zhanggu (Rong-mi Brag- 'go), and Shuizi (Rwa-tso), all located at the headwaters ofDaduhe River. This speech form is an isolated Khams Tibetan dialect, surrounded by Situ-rGyalrong to the north and the east, Geshitsa to the west and Guiqiong to the south. According to the local Tibetans, the dialect consists of several subdialects, which can be identified according to their location along the river and mountains, namely, Sogpho, dGudzong, sProsnang and Rwatso. While these subdialects differ only minimally at the grammatical level, there are much greater differences at the phonetic level. 4 The speakers of this dialect group only use their native dialect in their own villages, while they use Chinese in other Danba villages on account of the linguistically intricate situation mentioned above. Danba Xianzhi (1996: 174) presents a short description of the ''Twenty-four villages' patois", but few linguistic studies have so far been produced. Suzuki (2005b) provides us with the frrst preliminary description of the phonological system of Sogpho Tibetan. 2. PHONOLOGICAL SYSTEM
2.l.Tone Sogpho shows a five-way distinction in word tone. The following phonemic signs will be used at the beginning of a word: -:high level P5f44] _ : low level FJ1 1]
' : rising f4J3 5] ' : falling [53 J3 1]
A: rising-falling [132]
4 For example, the systematic preservation of the glide /r/ corresponding to rabtags in sProsnang Tibetan only, and a particular sound change in vowel quality attested
only in dGudzong Tibetan
57
DIALECfAL PARTICULARITIES OF SOGPHO TIBETAN
2.2. Vowels Sogpho has a rich vowel inventory. Each vowel has a normal and a nasalised realisation. Moreover, non-nasalised short and long vowels are distinctive.
Table 1 Vowel inventory wu e
0
:>
£
a
o
2.3. Consonants The consonant inventory is similarly rich:
T,bl2S ogplho consonantal.mvenory a e t ph plosive aspirated f' affricate
fricative
nasal liquid semi-vowel
non-aspirated voiced aspirated non-aspirated voiced aspirated non-aspirated voiced voiced voiceless voiced voiceless voiced
p
t
b
d
r
d'
1C'
c
k
q_
j
g
t
tsh ts
tc;:h tc;: ~
dz
cp m
rp
w
sh s z n IJ 1
!
?
xh
~
c;:h c;:
~
~
y
1),
lJ fj
~h
t),
X
h fi
r j
Sogpho shows the following types of initial consonant clusters: - prenasalisation: "C (C =voiced or aspirated occlusives and fricatives) - preaspiration: "C (C =non-aspirated and voiced occlusives) - labialised preaspiration: we (C =non-aspirated and voiced occlusives) - labial plosive: PC (C = occlusives) - glide: C +w orj - triple clusters: preaspiration-C-glide (only in a few words)
58
HIRUYUKI SUZUKI
2.4. Syllable structure The syllable structure of Sogpho can be defined as CC;GVC, to be understood as: c: preinitial, Ci: main initial, G: glide (/w/ and /j/), V: vowel, and the last C: final. In addition, there is a syllable which consists of only a nasal, e.g. I'IJ ce/ 'my, mine.' 3. ILLUSTRATION OF THE PHONEME INVENTORY In this section I provide illustrative examples for the phonemes listed above, and also discuss some phonetic properties, particularly the phonetic features of the tonemes. 3.1. Tonemes and their phonetic features Monosyllabic words have the following six phonetic pitch patterns: high level, low level, high rising, low rising, high falling and low falling: Table 3 level
The high rising tone can occur as a free variant of the high falling tone, for example [iA.135ZjA.153] 'yak' or [:rft)35/:rft) 53 ], an evidential marker. The phonologically distinct pattern consists of the following four types: high level, low level, rising, and falling. Variation within each of the four types is not distinctive. In the case of disyllabic words, the low-level tone is replaced by a rising-falling tone. The same observation applies to a disyllabic word consisting of a monosyllabic word plus dependent element: - high level: [p"A55ji55] 'home land,' [jo55 tvfi 5] 'Yangjin (personal name),' [ts"A' 55
2]
'village'
- rising: [ge13figre51 'teacher,' [n.i13
rcb 3fid.A 2j 'stone,' [lA lA 1
4
13
42]
'house,' [mA' 13IJtii52] 'oil'
DIALECfAL PARTICULARITIES OF SOGPHO TIBETAN
59
Polysyllabic words with more than two syllables are rarely found, one example is the trisyllabic word [ta 12rd 34fi.d 55] 'they.' In general, polysyllabic words can be divided into separate mono- or bisyllabic elements. In this case, each element bears one toneme. On the other hand, from the third syllable onwards, the appropriate toneme is lost and pronounced in a weak low level tone, e.g. [l),i113 'P"w55 !§A11 ~?22] 'twenty-one.'
3.2. Vowels Vowel /e/ has the free variants Pe] and [1e], vowel /o/ the free variants [wa], [ua], and [au]. In the following table, examples are given for the three vowel qualities: normal, nasalised and long. Note that the vowels h.! and /a/ do not have lengthened variants: T,a ble 4 i
e E
'CJ
a
-tc;:%?
0
'rna
()
'p:>? -uo
0
vowe1exampJ1es
nonnal ji? 'de flge
u w 1 e
'shu -ilu
:l
-n:l
-tc;:\ ie m:J
nasalised 'letter' 'Derge' 'rice' 'you' 'wound' 'Tibet' 'five' 'tooth' 'who' 'water' 'leaf 'this'
'Ji 'lltc;:e 's~n£
-ta ba 'naka 'm'""liJO 'so '?a lu 'mw9tsh:J
·:q 'lpe -~
'be' 'urine' 'forest' 'Danba' 'pit' 'blind' 'three' 'breast' 'get dark' 'change' 'official' 'two'
long 'ti: "'dze: 'tc;:E: ba
_flgw:
's/he' 'beautiful' 'bird' 'to not be' 'before' 'pain' 'friend' 'quilt' 'need'
'~'~e:
'gunpowder'
'rna:
-fu:bo 'no: _ro: -ci>u: ge:
3.3. Consonants 3.3.1. Simple initials The palatal stops /ch/, lei and IJI interchange freely with the corresponding palato-alveolar affricate. 5 /ch/ can be found only in the complex initial, /xh/ and /x/ are found in the medial position of the word.
5 Several examples of1he voiced stops can be folmd with high tone as well as with low tone. This phenomenon is attested in o1her Khams Tibetan such as mBathang (sKalbzang 'Gyur-med 1985) and Derge (sKal-bzang 'Gyur-med & sKal-bzang dByangscan 2002: 91-111). The conditions for this split are still unsolved.
60
HIRUYUKI SUZUKI
1'1a ble 5 Consonant exampJ1es high tone 'pig' 'hood'
tY' p b
-pha? 'pa:rje
f'
-f'amu 't:) tY'o
'mother-in-law' 'cloths'
'fa?
'blood' 'get angry' 'snake'
t d
r t
'tu Hi
q_ cf'
'
c j
'ce 'j:)
lC'
-~?
k g ? tgh
'komo
'meteor' 'Han Chinese' 'needle' 'throat'
'?a h_ko -ts% lw 'ts:) ru:
'elder brother' 'rubbish' 'slim'
tc;:h tc;:
'tc;:ho -tc;:a bo
~
-~amo
'wine' 'thing' 'weigh' 'quilt' 'place' 'food' 'summer' 'know' 'umbrella' 'bottom' 'tree' 'milk cow' 'year before last'
ts
not attested
sh s z
-cpu: ge: -sha c;:ho -se 'no lw 'zi
~h
-~~:
~
-~ ndu? b -zp? 'c;:IJ pho -c;:\1: b:) ffi:) -~:) ni du !axe
'mogmog'
'yu: narn
'understand' 'snuff
'mi 'rpan "l>u
'eye' 'doctor'
'no
'sky'
~
c;:h c;: ~
X
xh
y h fi m rp n
'pwmo 'bo
low tone 'homeland' 'daughter' 'broad'
"tari 'do rn
'today' 'face'
't:)
'barley' 'chain'
'
not attested
dz
cp
'p~ju
'kara 'gapo !ats\ "tsha? \:e
'tail' 'peach' 'belt' 'oldman' 'grand-mother' 'half
'dza~w
'feather'
'tc;::)
'tea' 'rGyalrong'
'juma
1C'a bu
·~aro
's~ bo 'sa bu
'zo:nm 'ra ·~a
'dew' 'bridge' 'end of month'
'c;:ha t~ Ac;:oju
'delicious' 'change' 'hound' 'milk tea'
Asaxhu
'kitchen'
Ahe zo 'b ro fiO 'nm
'pepper' 'they' 'husband'
'no
'and'
·:q
'ta:kw
DIALECfAL PARTICULARITIES OF SOGPHO TIBETAN
61
Table 5 continued high tone I),
'I].:> -I)J bo
1\
'1\i
lJ (I 1
'lJi -(J£'1 'Iii
! r
'!a shu -rn?
'nose' 'old' 'heart' 'silver' 'pillow' 'wind' 'Lhasa' 'ask'
'ja?
'yak'
I}-
~'~i?
'IJalpU '~),a mo
w
j
low tone 'nose hole' 'sun'
_lJO
'I'
1a
'road'
'ra 'w:) 'juju
'goat' 'fox' 'potato'
3.3.2. Complex initials The number of consonant clusters is quite large. The ftrst element of a consonant cluster tends to be pronounced weakly word-initially, while at the morpheme boundary it may be pronounced more clearly. Initial weak pronunciation tends to be stable and cannot be freely omitted. Nevertheless, in a few cases, particularly in number words, clusters are observed only word-medially. 6 Table 6 Prenasalisation 'worm' ~ ~wlii -ndo 'Dar-rtse-mdo "d (Kangding)' 'rice' '~let£ ~let 'tremble' -l)i o:t Ug 11go 'go' ndz
-ndzu
'mdzo'
I'~
'~'~allxa
'shoe'
'9ph -'9pho ~
""9fu
~
-~'}
~~
-~~'}
~
-~j
9tsh '9tsha ~c;:h
-~c;:hwrw
tJgh
"t'sho pu
~h
-~~
'raise' 'high' 'drag' 'hug' 'town' 'smoke' 'lip' 'knit' 'beat'
6 The preservation of clusters at 1he morpheme boundary, while they are lost wordinitially, is widely attested in 1he modern Tibetan varieties. For some recent descriptions and discussion, see Shirai (1999) for Lhasa Tibetan, and Zeisler (2005) and 1his volume for Ladakhi.
62
HIRUYUKI SUZUKI
1:a ble 7 Preasprra . tion
tp
·tpn
ht 1
"hta wu '1o Amahce -~a cb;i 'htsa ~a
he ~
hts ht~,:
'ht~,:u '%~ mo
hgh hg
'%o?
~h
'~ha:ma
~
-~:;
tx
'~'cb;a txa
1
'1::>?
'grassland' 'walnut' 'hair' 'jaw' 'mountain' 'root' 'steal' 'elder sister' 'life' 'hunter' 'wrap' 'shoe' 'teach'
lb
-lba?
\:1 nq_ ~ ~
"\ianii
\iz ~
"'z ~ l'lJ.
"'n "n ~
~opltt
'carry on back' 'moon' 'fly (insect)' 'back' 'nine' 'learn' 'willow'
""zo "'zo -~
'many' 'four'
lo n1a1 , ~,:::>"'me '"hamo -~
'right side' 'flood' 'daughter-in-law' 'two' 'fifteen' 'left side'
'flq_::> na'} ~e?~a
-~
_\izo
~
-lj& l'lJ.a?
~!:lh
-~!:!he:
'know'
~~
-~~'}
'talk'
bq_ bcb; wd wj wz
-bq,a 'no bcb;o? -wde ba 't~,:o wja? '~,:t;: wzu
-t~,:wfJ:Jo
Table 8 Labia1ised types
Pf' 'Pfl.
't
ptgh
P!:s Pi:~,:
"""tL
'P!:siJ 'Ptso
p(,:h
'uaP!:~,:w 'P~,:he'}
~
-~'}
i.
'i.w
'k:
'p::>? 'k:::>? -~
~s ~(,:
-~o
%
-~san5
-~~,:a?
'bed' 'cloud' 'open' 'cock' 'fifty' 'splash' 'seed' 'wash' 'cover' 'sell' 'cutoff 'idea'
w~ w~
wm wl
'bite' 'at last' 'honest' 'eighteen' 'carpenter' 't~,:ww~ 'fourteen' -w~e:ze 'forget' "1Jga wmo ~lee? po 'wild ox' 'thigh' ~argo
1:able 9 C1usters WI"th gJlideand tripJe 1 c1usters kw gw !:IW xhw b
""kwozu 'wo 'gwofi '!:!we%6 'xhwa OJC'o: 'ba
'ywa do '?amje 'pa: rje
'hard' 'smile' 'forty-three'
yw mj Ij
'turn' 'tired'
~~:)
'participate' 'grandfather' 'hood' 'love'
DIALECfAL PARTICULARITIES OF SOGPHO TIBETAN
63
4. PECULIARITIES OF SOGPHO TIBET AN
In this section I compare the phonological and lexical features of Sogpho Tibetan with Written Tibetan (hereafter WrT) and other Tibetan dialects, especially those spoken in the so-called Ethnic Corridor of West Sichuan, in order to illustrate the peculiar nature of Sogpho Tibetan. My main putpose in this section is not to produce a monocausal historical analysis of Sogpho Tibetan, but to consider the development in a typological perspective. Due to its complex ethnic, historical and linguistic background, the ''Twenty-four villages' patois" wtderwent many influences through language contact and we cannot therefore expect a regular pattern of sowtd changes. In other words, it is necessary to compare the Sogpho Tibetan dialect with other dialects in order to explain its outstanding phonetical and lexical features. Recent research reveals the existence of other typologically peculiar Tibetan dialects, such as Zhongu7 (Sun 2003) and gSerpa8 (Swt 2005). I refer closely to Swt's (2003) analysis in the following discussion. The data presented here are based primarily on my own research and on Swt (2003, 2005). 9 I discuss the consonants, vowels and special word forms of Sogpho Tibetan, but will not consider the tonal patterns, because my analysis differs from previous analyses. 10
4.1. Features shared with other East Tibetan varieties I will not address the most general consonantal features, such as the three-fold opposition of aspirated, non-aspirated, and voiced articulation, concentrating instead on the initial consonant clusters and vowelfmalforms.
7 Spoken in Rewugou district, Songpan Cowrty. As yet not identified with respect to its dialectal affiliation 8 Spoken in Seerl>a district, Seda Cowrty. As yet not identified with respect to its dialectal affiliation 9 See Figure 1 below for the geographical location and distribution of the dialects mentioned here. 10 With respect to the Tibetan varieties, tone is generally treated as a feature of the syllable, not as a feature of the word. Word tone has been established only for Lhasa Tibetan.
64
HIRUYUKI SUZUKI
4.1.1. Consonant clusters Sogpho Tibetan shows an extended set of consonant clusters, which is more complex than in the agricultural varieties of Amdo Tibetan and most Khams Tibetan dialects. In some aspects, Sogpho resembles the nomadic varieties of Amdo Tibetan. Attention is given to the manner of articulation of consonant clusters, 11 whose ftrst element is always pronounced very weakly with the exception of clusters with glides. This trait, as well as the presence of prenasalisation and preaspiration, is standard in almost all of the Khams Tibetan dialects. 12 It is possible to establish the following correspondences between the patterns of consonant clusters in Sogpho and WrT (here and in the following the WrT forms will be given in italics): 1. Prenasalisation typically corresponds to WrT initial m- and 'a-: I_JJgul 'head' mgo
t~EI
~ 'high' mtho
/'lllbm la/ 'worm' 'bu
'rice' 'bras
Exception: /~sho pu/ 'knit' WrT sla 'twist, plait'
2. Labialised preaspiration, initial labial fricatives and stops correspond to WrT b- (another origin for clusters such as /'Ptso/ 'cock' bya po will be discussed in 4.2.3.): word-initially:
l'+t_ml 'wash' bkru /""'¥!: w 'forget' brjed
word-medially: /'t~ wja?./ 'eighteen' bco brgyad
fiJa Pt9ml 'fifty' lnga bcu
3. Preaspiration typically corresponds to WrT initial d-, g-, r-, 1-, and sword-initially:
word-medially:
/_6gml
rt(jlm lirjo/ 'ftfteen' bco lnga
'nine' dgu
t~ 'two' gnyis
/'?.a "til 'monkey' sprel
11 See Suzuki (2005a) for a detailed discussion of the relation between syllable structure and manner of articulation in consonant clusters. 12 Nevertheless, these two characteristics have not been recorded in previous studies, such as Jin (1983), sK.al-bzang 'Gyur-med and sKal-bzang d.Byangs-can (2002), and others. See, however, Causemann (1989) for a discussion of prenasalisation, prelabialisation, and preglottalisation in Nangchen. Prenasalisation and preaspiration are very common in the dataset I have collected.
DIALECfAL PARTICULARITIES OF SOGPHO TIBETAN
65
/"'tsa hkaf 'root' rtsa ba
tt'ka?./ 'language' skad 4. WrT la-btags and sr- correspond to either a complex or a simple initial: complex initial:
simple initial:
laj/ 'flute' gling rwla mol 'monk' bla ma J"'ida f),i5/ 'moon' zla ba rhso?./ 'life' srog
flU/ 'wind' rlung r1o1 'easy' sla13 /'15/ 'bull' glang r~o/ 'hard' sra
/' 6
4.1.2. Vowel-finalforms Similar to other Khams Tibetan dialects spoken in western Sichuan, syllable or word fmals appear to be quite reduced in Sogpho Tibetan.
5. Nasalised vowels typically correspond to nasal finals in WrT, but in some cases the nasals have been dropped leaving no trace: nasalised vowel:
plain vowel:
riO/ 'bull' glang ripe! 'official' dpon rna! 'sky' gnam
/""lea bol 'foot' rkangpa f""Ptl{, fi'tl{ 'cloud' sprin
6. A glottal stop corresponds to occlusive endings in WrT, but at the morpheme boundary, the glottal stop can be omitted: glottal stop:
omitted glottal stop:
rhtva?./ 'iron' lcags rhka?./ 'language' skad rlr!'o?.l 'needle' khab
l'f.u! 'ZpJI 'sixty' drug cu
rpo mal 'Tibetan' bodpa
7. A lengthened vowel corresponds to WrT final continuants, but a WrT final continuant does not in all cases correspond to a lengthened vowel:
13 The correspondences with WrT la-htags are, wi1h the exception ofWrT zl, quite straightforward and the original contrast of the initials is preserved: all combinations lead to high tone, prelabialisation to WrT bl, the simple voiced forrn to WrT gl and rl, and the simple unvoiced form to sl.
66
HIRUYUKI SUZUKI
lengthened vowel: rmo:l 'low' dma'
normal vowel: l_'l).d'Qdi 'rainbow' 'ja'
flicli:J 'snake' sbrul
r~el 'gold'
gser
8. In a few exceptional cases, words with an initial nasal show a nasalised vowel, where the corresponding WrT word has a stop ftnal:
rmil 'eye' mil4 fiJi! 'silver' dngul 4.2. Features not commonly shared with other dialects Sogpho Tibetan shows some peculiar phonetic features that are rarely attested in other Tibetan dialects. 4.2.1. The phonemic opposition /rl and 12(" Sogpho shows a phonological opposition between lrl and I'Z/., which is very rarely found in other Tibetan dialects. In most cases, Sogpho lrl and I'Z/. correspond to WrT r and zh respectively, but not all WrT words with zh correspond to I'Z/. in Sogpho Tibetan:
I_r~ 'mountain' ri I_liz;i 'four' bzhi
r~
l),i
lui
'year before last'
gzhes ning lo Retroflex realisation of WrT zh is also found in Zhongu and Nyishe, 15 but while Zhongu, like Sogpho, allows for a non-retroflex realisation in some words, Nyishe does not: - Zhongu: I~
'mountain' ri
I'Zf'l 'four' bzhi
ls'P::j./ 'field' sa zhing
- Nyishe:
/'ra/ 'goat' ra r'l1i 'four' bzhi f21_e l),i/ 'day after tomorrow' gzhes nyin 14 This example likely originated from dmig/dmyig in Old Tibetan, as in many other dialects. 15 Spoken in Nixi village, Xianggelila (Shangri-La) County, Yurman. Belonging
to Khams Tibetan.
DIALECfAL PARTICULARITIES OF SOGPHO TIBETAN
67
Corresponding to the different realisation of WrT zh as either retroflex or non-retroflex sibilants, the unvoiced counterpart WrT sh may likewise be realised as either retroflex (/~) or alveopalatal/postalveolar sibilant (/9/, IJI) in the above-mentioned dialects of Sogpho and Zhongu, but in Nyishe only as retroflex: - Sogpho: r~Ita/ 'die'
r ¢of 'meat' sha
shi
- Zhongu: /~a/
'deer' shwa ba
!Ji! 'firewood' shing
- Nyishe: r~hal 'meat'
sha
r~he f/U?./ 'east'
shar phyogs
It is observed that in Sogpho Tibetan, one and the same morpheme may possess the variants of I'Z/. and /r/, such as/~~ 'forty' bzhi bcu and 1'11. h~w <'>~we hsO/ 'forty-three' bzhi bcu zhe gsum. A similar realisation of WrT zh as /r/ is also found in gDongsum Tibetan, 16 for ~wf [~ 55 ht(pW 55 ] 'forty' bzhi bcu, but this dialect has no example distinction between /r/ and /'Z{, it possesses only /r/.
rrn
4.2.2. Palatal plosives ld'/, ld, ljl versus palata-alveolar affricates I (f/'1, I ttl, I c4/ In Sogpho Tibetan, the phonemes /ch/, /c/ and /jl correspond to WrT Ky-, such as /ce/ 'of' kyi, !jol 'Han Chinese' rgya, etc. However, not all instances ofWrT Ky- correspond to a palatal plosive, some ofthem are realised as palato-alveolar affricates, such as r~l 'dog' khyi. More generally, palatal plosives are comparatively rare phonemes in the Khams Tibetan dialects, and only few dialects possess them: - Chamdo: 17 palatal plosives corresponding to WrT c, ch,j - Lhagang: 18 f-lija 6ge?./ 'Chinese' rgya rgod, fcho?./ 'you' khyod
Spoken in Dongsong village, Xiangcheng County. Spoken in Changdu, Tibetan Autonomous Region. Based on sKal-bzang 'Gyurmed & sKal-bzang dByangs-can (2002). 18 Spoken in Tagong village, Kangding County. 16 17
68
HIRUYUKI SUZUKI
Accordingly, free variation is found between palatal plosives and palato-alveolar affricates in Sogpho Tibetan, for example thee?./, fhtq,e?./ 'happy' skyid.
4.2.3. Realisation ofWrT Py Quite unusually, WrT py, phy and by correspond to alveolar affricates in Sogpho Tibetan, which are furthermore frequently preceded by a labial stop, such as /'ts~ mol 'sand' bye ma, fPfi'i./ 'open' phye (second and fourth stem of 'byed), f'Ptsof 'cock' bya and /'?.a ts~, /'?.a Pts\ I 'grand-mother' a phyi. 19 The correspondence of WrT Py with alveolar affricates is extremely rare, but it is also found-without a preceding labial-in Zhongu Tibetan, cf. /t~e/ 'flour' phye and /ts'B/ 'bird' bya. Apart from Zhongu and the ''Twenty-four villages' patois", no examples are found in other Khams Tibetan dialects. The retention of the labial, on the other hand, is shared with other dialects. 20 - Lhagang: r•~al 'cock' bya, fPt~?. ka/ 'spring' dpyid ka - Rangakha: 21 /'Pt~ 'cock' bya, rtq,e mal 'sand' bye ma It is worth mentioning that Tibetan loanwords in nDrapa22 show a realisation with alveolar-palatal affricates, similar to Sogpho Tibetan: I""~ a! 'cock'< WrT bya and /'Ptsa rAI 'coral'< WrT byu ru.
4.2.4. Realisation offinal WrT a WrT fmal a corresponds to final /o/, /~or /a/ in Sogpho Tibetan. Note that the phoneme /o/ may be realised as a diphthong: r~1. [u~] or [~u], cf. /_rjo/ [IJua22] 'I' nga and tsho/ [~u~ 52] 'earth' sa. Similar diphthongs are found in gSerpa Tibetan, cf. /rjwo/ 'Chinese' rgya and /ptJ~ mwa/ 'sand' bye ma.
A few words have a palato-alveolar, such as /'f¥e: ba/ 'bird' bye 'u pa (7). The labial stop may also be preserved in clusters corresponding to Wff Pr, cf. I~ besides fi:l! 'cloud' sprin. The retention of the labial stop, while the following palatal glide or the retroflex approximant have undergone a considerable sound change, is quite typical for the nomadic Amdo dialects, see R6na-Tas (1966: 179-182). Other varieties show this feature only at the morpheme bmmdary of compounds, cf. Lhasa Tibetan "chiipcA. 'water bird"' chu bya., "chXp¢u 'baby chicken"' bya phrug (Shefts & Chang 1967: 519 no. 7, 521 no. 26). See also Zeisler (this volume) for Ladakhi. 21 Spoken in Xinduqiao town, Kangding County. 22 One of 1he Qiangi.c languages, spoken in Zhaba district of Daofu County and Zhamai district ofYajiang County. The dialect mentioned here is Ngwirdei, spoken in Hongding village. 19
20
DIALECfAL PARTICULARITIES OF SOGPHO TIBETAN
69
The realisation of WrT fmal a as /o/ corresponds phonologically to its realisation in Sagong23 and gDongsum Tibetan. Other dialects generally realise WrT fmal a as I a/: - Sagong: /'so:/ 'cock' bya /'rwo/ 'hom' rwa - gDongsum:
r!C'ol 'mouth' kha rt'ro/ 'horse' rta - Lhagang:
thtril 'horse' rta
t 'Val 'nose' sna 4.2.5.lnterchange ofalveo-palatals and retroflexes
In Sogpho Tibetan, initial alveo-palatal affricates corresponding to WrT Ky or c, ch or j may interchange with retroflexive stops or affricates as well as with postalveolars. Suzuki (2005b) proposes that this results synchronically from the property of the apical (retroflexed) vowel IV_.
chu nyi shu kh
'water' 'twenty'
55
'ou'
The example 'you' above does not possess the vowel h/. but the interchange of the initial is attested, so two base forms for this word may be proposed. A similar phenomenon appears in Guiqiong, spoken to the south of Danba. Some other Tibetan dialects, such as Nyishe and Bud~ show retroflex initials for WrT c, ch and j: - Nyishe:
rtl'w/ 'water' chu t"@i?/ 'one' gcig 23 24
Spoken in Shagong village, Xiangcheng County. Spoken in Badi village, Weixi County.
70
HIRUYUKI SUZUKI
- Budy: f't~"a
pal 'rain' char pa rfllhe bo/ 'big' che ba 4.2.6. Additional remarks on some particular lexemes Sogpho Tibetan attests some words which are not found in neighbouring dialects, but found back rather in dialects spoken in places geographically distant from Danba, such as Budy, rGyalthang, 25 Phrengme26 and gDongsum, most of which are in southern Khams. These expressions may or may not show a straightforward correspondence with WrT. A few words seem to be of different origin: they neither show similarity with WrT, nor can they be found in other Tibetan dialects. 1. Expressions with possible WrT correspondences, also found in distant dialects:
-
r~OI 'light' : r~O/ (Budy), cf. WrT sangs pa 'clean, clear'
- rna khaf 'drought' :rna ka/ (Budy), cf. WrT gnam skarn 'dry sky' - /'tJgo/ 'go' : /'tJgo/ (rGyalthang), cf. WrT 'gro 'go' without ra-btags - rwuji?/ 'eleven' : /'t
- rrpl 'evidential marker' :I'm/ (rGyalthang), rr:Jol (Budy), cf.
WrT
snang 'appear, be perceived (as)' - r"l:J?/ 'eagle' vulture'
rlo?/ 'falcon' (Rangakha), cf. WrT glag 'eagle,
2. Expressions without WrT correspondences, but found in other dialects: - /'Jg:J rn/ 'rye' rn rnl (PhrengmelgDongsum), no corresponding words in WrT, but Jiang (2002: 253) shows krungs 'ear of barley' in Old Tibetan 3. Expressions without correspondences in WrT, so far not attested in other dialects:
- /'sa I)£1 'forest', but WrT nags
- 0tS'a/ 'smoke', but WrT du ba 25 26
Spoken in Jiantang town, Xianggelila Cowrty. Spoken in Qingmai village, Xiangcheng County.
71
DIALECfAL PARTICULARITIES OF SOGPHO TIBETAN
- t"ta: pal 'skin', but WrT lpags pa - /'llcb{ 'wheat', but WrT gro (the corresponding realisation f't-;J/ is used in Sogpho for 'barley'). 4.3. Results
Comparing the phonological and lexical particularities of Sogpho Tibetan with some dialects spoken in the Ethnic Corridor of West Sichuan, we may conclude that the phonological features of Sogpho Tibetan are partially shared with these dialects, but not in a systematic manner. With respect to the lexical features, Sogpho Tibetan shows a partial similarity to some distant southern Khams Tibetan dialects, such as Phrengme and rGyalthang. The results of the typological analysis of Sogpho Tibetan are summarised in the following table:
T,b/JJT "1 compans on27 .ypo1ogtca a e WrTzh /zf WrTPy Its/ etc. WrTKy /c/ etc. WrT bC : fPC/, fiCI WrT final a /of WrT c: retroflex
Bu 0
So
Zh
Lh
Ra
Sa
gD
Ny
0 0 0 0 0 0
0 0
-
-
-
-
0
-
-
-
-
-
-
-
-
-
-
-
-
-
0 0
0
-
-
-
0
-
-
-
0
0
-
-
-
-
-
-
-
0
0
5. CONCLUSION In this paper, I have attempted to initiate a linguistic study of Sogpho Tibetan, a variety of the ''Twenty-four villages' patois." Although the subdialects of the ''Twenty-four villages' patois" show some differences at the phonetic level, the features described in this paper arealmost all shared. Sogpho Tibetan, and more generally the ''Twentyfour villages' patois", do not show similarities with the dialects of the immediate neighbourhood, i.e. Lhagang or Rangakha Tibetan, but with dialects located farther away, such as rGyalthang or Zhongu Tibetan in southern Khams. Despite its geographical proximity, Sogpho Tibetan does not belong to Minyag Tibetan.
'2:1 The following abbreviations are used: So: Sogpho; Zh: Zhongu; Lh: Lhagang; Ra: Rangakha; Sa: Sagong; gD: gDongsum; Ny: Nyishe; Bu: Budy.
72
HIRUYUKI SUZUKI
While the typological similarity among Sogpho, rGyalthang and Zhongu Tibetan can be easily demonstrated, these varieties are notrelated to each other in every respect. The historical relationship between these Tibetan speech fonns is still1D11"esolved, as is the degree and dmation of language contact. These dialects share a central common feature, their location: the frontier of the Tibetan-speaking area facing other etlmic and cultmal areas such as the Han Chinese, rGyalrong, Bairna, Qiang and Naxi. • Zhongu
Dege---o
• gSerpa
o - - - - - Maerkang ·Sopo • Lhagang
Batallg---0
-k-Chengdu
Rangakba 0 - - - • Sagong .Phrengme
•gUongsum
1001an Figure 1. Location of the Tibetan dialects dscussed
Kangding
DIALECfAL PARTICULARITIES OF SOGPHO TIBETAN
73
BIBLIOGRAPHY
Causemann, Margret. 1989. Dialekt und En:iihlungen der Nangchenpas. (Beitriige zur tibetischen Erziihlforschung, 11.) BOlUl: VGH Wissenschaftsverlag. Danha xianzm = Sichuansheng Danba xi.anzhi bianzuan weiyuanhui. 1996. Danha xianzm (The Annals ofDanha Cmmty). Beijing: Minzu Chubanshe. Darragon, Frederique. 2005. Ximalaya de shenmi gudiao (Misterious ancient towers in the Himalaya). Shenzhen: Shenzhen Baoye Jituan Clmbanshe. rDo-Ije. 1998. Daofoyu Geshizahua yanjiu ?4. study on Gesmtsa dialect of sTau lan-
guage). Beijing: Zhongguo Zangxue Chuba.nshe. Jiang, Di. 2002. Zangyu yuyinsm yanjiu ?4. Study on the phonological history of the Tibetan language). Beijing: Minzu Chubanshe. Jin, Peng (ed.) 1983. Zangyu jianzm ?4. short description of the Tibetan language). Beijing: Minzu Clmba.nshe. sK.al-bzang 'Gyur-med. 1985. Zangyu Batanglma de yuyin fenxi. (Phonological analysis ofTibetanmBathang dialect). Minzu Yuwen 2: 16-27. sK.al-bzang 'Gyur-med and sK.al-bzang dByangs-can 2002. Zangyu fangyan gailun ?4-n introduction to the Tibetan dialects). Beijing: Minzu Clmba.nshe. R.Ona-Tas, Andnis. 1966. Tibeto-Mongolica. The Tibetan loanwords ofMonguor and the development of the archaic Tibetan dialects. (Indo-Iranian Monographs, 7.) The Hague etc.: Mouton. Shefts, Betty and Kun Chang. 1967. Spoken Tibetan morphophonemics: p. Language 43: 512-525. Shirai, Satoko. 1999. Chibettogo gendai Rasa hogemo 2 onsetsukarmi. arawareru -bnitsuite (On the intersyllabic -h- in Modern Lhasa Tibetan). Gengogaku Kenkyu (Linguistic Research) 17-18: 59-72. Sun, Jackson T.-S. 2003. Phonological Profile of Zhongu: A New Tibetan Dialect of Northern Siclman. Language and Linguistics 4.4: 769-836. - - . 2005. Special Linguistic Features of gSerpa Tibetan Unpublished marruscript presented at 38th ICSTLL (Xiamen). Suzuki, Hiroyuki. 2005a. Tibet-go onsetukoozoo no kenkyuu (A Study of the syllable structure in Tibetan). Ajia-A.forika Gengo Bunka Kenkyuus 69: 1-23. - . 2005b. Chibettogo Danba-Suopo [Sogpho] Hogen no Onsei Bunseki (Phonetic analysis ofDanba-Sogpho dialect ofTibetan). Nidaha 34: 96-104. Yang, Jiaming. 2005. Jiedu Jiarong (Exploring rGyalrong). Kangding Minzu Smfan Gaodeng ZhuankeXuexiaoXuehao 3: 1-5. Zeisler, Bettina. 2005. Skaddi gyurcantiiJ rdohafzi mentogtsogsla luskanni S1JOn!yug hafzi rdzessiskorla-Language change and the fossilisation of the Old Tibetan hprefix in Ladakhi and Balti. Paper presented at the 12111 Colloquium of 1he International Association for Ladakh, Kargi1 (Ladakh), 12111-15111 July 2005. http://www.sfb44l.uni-tuebingen.de/bllllanguageArchaeology.pdf --.This volume. Kenhat, the dialects ofUpper Ladakh andZanskar.
PART THREE: SEMANTICS (WORDS AND WORD CLASSES)
ON THE OLD TIBETAN TFRMKHRIN IN THE LEGAL AND RITUAL LEXICONS BRANDON DOTSON
During the period of the Tibetan Empire (c.608-c.866), Tibet developed a complex administrative and legal system that supported the paired trends of centralisation and the assimilation of subjugated territories and peoples. This legal and bureaucratic system also facilitated the levying of troops, the collection of taxes and the legislation ofthe empire in general. While conquered territories were required to pay tribute, the subjects of the Tibetan emperor were liable to pay numerous taxes and duties. Among the duties required of Tibetan subjects were corvee labour and military service. Failure to comply with any of these taxes or duties resulted in legal proceedings and punishment (chad). With such bureaucracy and legal methodology, of course, came a new vocabulary containing a certain degree of jargon and formulaic expressions. In this paper I will examine one particular term, khrin, from the legal lexicon as it appears in numerous OT sources, and consider its employment in ritual settings as well. 1. PREVIOUS THEORIES ON KHRIN
The ftrst modern scholar to remark on the meaning of the tenn khrin was Giuseppe Tucci, who treated the tenn as it appeared in the Skar chung Edict ofKhri Lde srong btsan preserved in Dpa' bo Gtsug lag's Mkhas pa'i dga' stan (hereafter abbreviated KhG). Tucci translated khrin as 'protection', but stated in a note that his translation was '1d]oubtful" (Tucci 1950: 53, 83, n. 115). Soon after, Richardson, in his treatment of the inscriptions at Zhwa'i Lha khang, took khrin to be ''the equivalent of khrims", and translated it as 'legal proceedings' (Richardson 1952: 34, n. 6). Richardson maintained this definition of khrin in his later work, though with a bit of uncertainty (Richardson 1985: 25, 51, 161). F.W. Thomas translated the term as 'judgment' in TLTD3: 118, citing the compound khrin sa from the "Annals ofthe 'A zha Principality" (TLTD2: 8). Not long after, Thomas translated khrin as 'import' without further comment (AFL: 39). Rona-Tas essentially
78
BRANDON DOTSON
followed Richardson's conclusions in his treatment of a passage from the inscriptions at Zhwa'i Lha khang, but added that khrin stood for khrim, which he took to mean 'right, administration of justice' (R6naTas 1955: 253, n. 12). More recently, W.S. Coblin outlined the problems posed by the term khrin as it appears in the Zhol Pillar Inscription and the inscriptions at Zhwa'i lha khang (Li and Coblin 1987: 184-85). Coblin hesitantly surmised that khrin was "a legal judgment or sentence", but was uncertain of the term's relationship to khrims (Li and Coblin 1987: 185). Coblin asked the following questions: ''Is kln"in a dialectical variant of khrims ~ khrim? Or is it a morphologically different form with a difference in meaning or nuance? Or a totally independent word of similar or identical meaning?" (Li and Coblin 1987: 184). Coblin (1991a: 73) returned to the matter of khrin in a subsequent article on rje bias, another problematic term from the legal and bureaucratic lexicon. There he stated that that khrin means 'judgement, sentence', but gave no detailed justification for his translation, nor did he attempt to solve his earlier questions regarding the term. I will turn to these questions after examining the context in which the term khrin appears in various OT documents.
2. KHRININ OLD TIBETAN SOURCES 2.1. Khrin in previously translated Old Tibetan documents The two examples from the pillar edicts both involve royal grants awarded to the lineage ( bu tsha) of a prominent Tibetan minister. Both edicts reward not only the honoured minister himself, but extend privileges to his close relatives by projecting the grants backwards one or two generations. 2.1.1. Khrin in the Zhol Pillar Inscription Lines 61-65 of the north face of the Zhol Pillar, which dates to c.764, read: If it becomes apparent that one among the increasing lineage [ofNgan lam Zla gong] truly becomes disloyal to the presence of the emperor, the accusation shall be pronounced against whoever is the guilty party [lit. ''the door of whoever is guilty" 1]. I grant that other clansmen will not be attached to the khrin and their life and livelihood will not be destroyed. 1
See Coblin's note on this phrase in Li and Coblin (1987: 388).
ON THE OLD TIBET AN TERM KHRIN
79
(bu tsha 'pheld gyl nang nas la la zhigll btsan po zha sngar glo ba rings yang dag par gyurd na gang nyes pa'l sgor I bka' gyod rma'oll pu nu po gzhan I khrin la my! gdags srog srid la my! dbab par gnang ngo II). 2
2.1.2. Khrin in the west face inscription at Zhwa'i Lha khang Lines 37-40 of the west face inscription at Zhwa'i Lha khang, which dates to the early part of the reign of Khri Lde srong btsan, between c.798 and c.800,3 read: Indeed if one among Snang bzang 'Dus kong's increasing lineage is disloyal to the emperor or the polity, or commits another crime, whoever is guilty will be investigated. Khrin will not be pronounced against other guiltless [Myang] clan lineages, they will not be attached to the charge (read gyoci) and no harsh order (bka' nard) will be made. (snang bzang 'dus kong gi bu tsha 'phel rgyud I la la zhlg gis sku dang chab srid la snylng rings sam/ nyes pa gzhan zhlg byas na yang I gang gls nyes pa la rma/ phu nu bu tsha ma nyes gzhan la khrln myi rma/ rgyod la myi gdagslbka' nard my! mdzaci). 4
2.1.3. Khrin in the ''Annals ofthe 'A zhaPrincipality" The passage Thomas cites in his treatment of the term khrin comes from the Annals of the ~ zha Principality (TLTD2: 8), but is so unclear that no judgment may be made from it other than the supposition
2 For transliteration and translation, see U and Coblin (1987: 150, 173). Coblin (1991: 73) used this passage as his example for his translation of khrin as 'judgment' or 'sentence', stating that khrin Ia ma gdags means "shall not be involved in the judgement" Two of the terms in this passage have been the subject of scholarly articles: Li (1959) treated the phrase glo ba rings/ glo ba 'dring and its opposite glo ba nye, and R6na-Tas (1955) treated the kinship termphu nu. The phrase bka' gyod nna is found in Btsan lha's Brda dkrol gser gyi me long (in its negated form, bka' gyod mi nna), where it is glossed as 'to reprimand' (bka' bkyon 'bebs), 'to dispute' or 'make a lawsuit' (gyod byefi) (Btsan lha 1997: 23). He also treats the phrase bka' gyod Ia chags, which, citing the Old TibetanAnnals' entry for the year of the hare (727-728), he renders as 'to be reprimanded' (bka' bkyon babs pa) or 'tried by law' (khrims Ia thogs pa) (Bacot et al. 1940-1946: 24, 47-48). This agrees also with the definition fmmd in the dictionary in production at 1he Bavarian Academy of Sciences: "Streit, ProzeB, Anklage"; see Franke et al. (2007a: 191 ). In my translation of the Old Tibetan Annals, I render this "[he] was the subject of an accusation" (Dotson 2009: 116). Btsan lha (1997: 663), citing the present passage, glosses nna as 'to investigate' (rtsad gcod pa'i don Ia Jug ste). As noted by Li and Coblin (1987: 439), the primary meaning ofthe verb nna is 'to inquire' and 'to say' They also list its secondary meaning as 'pronounce judgement or puni.slunent' 3 On the validity of these and other dates at the tum of the ninth century, see Dotson (2007b). 4 For transliteration and translation, see U and Coblin (1987: 266, 278-279).
80
BRANDON DOTSON
that khrin sa is a corruption of khrims sa. 5 Remarkably, Yamaguchi ( 1970: 61) reconstructs the same passage, which Thomas read as khrin sa, as khri s[rong brtsan gyi sra]s. This was, of course, in the context of Yamaguchi's radical rereading-cum-revision of this text. 2.1.4. Khrin in the OTritual text!TJ 731, "End ofthe Good Age and Tragedy ofthe Horse andYak" The tenn khrin appears in the second part of ITJ 731, which concerns funerary practices. One passage contains a discussion between Gyim po Nyag gcig and the six Gyim po brothers (Gyim po spun drug) regarding the proper funeral rites (shid, rmang) for their father, Sten rgan Nyer pa. This concerns the proper types of sacrificial animals, in particular the 'beloved horse' (do ma thugs dagsl do ma snying dags) and the psychopomp sheep (skyibs lugs), which serve as guides to the land of the dead. 6 After enumerating the hundreds of horses and sheep to be killed in order to constitute these psychopomp animals and make for a traditional funeral rite, the six Gyim po brothers wish to shirk their filial obligations to their father. A couplet in lines 21 to 22 reads: snga khrin che'i khrin ma blangs smra dos drag ki lcag ma btab (AFL: 17). Thomas translates this "Take not away the authority of what formerly had great authority! Deal not a blow at saying of great weight!" (AFL: 29). In an explanatory note, Thomas writes of khrin in this passage: "perhaps a din form of khrid, 'instruction', 'tutelage"' (AFL: 37). In her reading of this passage, Zeisler (2004: 440-441) translates this couplet: "Do not assume the great authority [which comes from being] ancient authority. Do not apply words [like] the whip for compulsory service!" We will return to this passage below in the discussion of alternate uses of khrin and khrims in Old Tibetan sources. 2.1.5. Khrin in the ShangshuParaphrase Another citation of khrin, this time by Coblin, proves to be fruitless. In his initial discussion of the term khrin, Coblin cites its appearance in the Tibetan paraphrase of the Chinese Shangshu as a variant of khrims 5 In his dictionary of Old Tibetan, Rnam rgyal Tshe ring (200 1: 61-62) glosses khrin sa as 'courthouse' (khrims sa or khrims /chang). 6 For more on the 'beloved horse' (do ma thugs dags/ do ma snying dags) and the sacrificial sheep (skyihs lugs) that guide the deceased to the land of the dead, see Stein (1970: 168-169).
ON THE OLD TIBET AN TERM KHRIN
81
(Li and Coblin 1987: 184). In his subsequent, detailed treatment of this text, however, Coblin (1991b: 307, 1. 49) transliterates the term in question as khrims, and khrin appears nowhere in the text 2.1.6. Khrin in the "Dialogue Between Two Brothers" Cob lin is not alone in reading khrin as a variant for khrims. Chab spel Tshe brtan Phun tshogs translated into modern Tibetan the first part of PT 1283, a dialogue between an elder and younger brother that contains a good deal of folk wisdom. In lines 357-358, we find khrin in the following statement from the elder brother: If my young son, who does not distinguish between right and wrong and what to do and what not to do, is not loving towards me, it may indeed come about that he will wander and be attached to a khrin. (bdagi bu
gzhon bas// legs nyes gnyis bye brag myi phyed la/1 bya myi bya drod myi zin pas// bdag la byams pa nyid ma yin nail 'khyams shing khrin la thogs par yang 'ong) (PT 1283, 11. 357-358; Imaeda and Takeuchi et al. 2007: 173).
Replacing khrin with khrims, Chab spel (1997: 64) translated this last phrase, "it may indeed come about that he will become wanton and be attached to a legal case" ('chal shing khrims Ia thogs pa'ang 'ong srid). 2.1.7. Khrin in Khri Lde srong btsan 's Skar chung Edict The passage from the edict of Khri Lde srong btsan, translated by Tucci, proves to be more useful for determining the meaning of khrin. The edict accompanied the Skar chung Pillar Inscription of Khri Lde srong btsan, in which he reiterates his father's oath to support Buddhism as the official religion of Tibet This likely dates to the latter half of his reign, and Richardson (1998 [1977]: 71) opts for c.812. The edict is preserved in KhG, and the passage in question deals with the privileges of Buddhist monks, and demonstrates the close relationship between khrin and 'accusation' (gyod). Monks shall not be awarded as the bondservants of others. They will not be punished with suppression, they will not be attached to the accusation, [as if] attached to a household's khrin, and ... (rab tu byung ba'i mams gzhan gyi bran du mi sbyin/ nan gyis mi dhabi khyim pa'i khrin la gtags te gyod la mi gdags shing;) (KhG: 411 ).
This passage is not properly an OT source, but an OT source preserved in a post-dynastic history, so it is of less value than the passages from the pillar edicts and OT documents cited above. There is, however, lit-
82
BRANDON DOTSON
tle reason to doubt the authenticity and the orthography of the passage in question. 2.1.8. Khrin in!TJ 753, an OT legal fragment on theft The frrst time a text containing the word khrin was published in a scholarly article is, to my knowledge, in Thomas' treatment of ITJ 753, a fragmentary legal document dealing with theft (Thomas 1936). The passage in question concerns a scenario in which someone purchases a stolen item from a thief, but then confesses to having done so and returns the item: The law (khrims) concerning one who purchased goods from a thief, but later confesses to having purchased them: when an item of wealth (dkor zhig) is stolen by a thief, then sold to an outsider, and, with neither the thief nor the buyer being disputed or investigated by the owner (dbang po), [his] friends (glo zla) or thugs nyen [nyin], the buyer later confesses, if he gives back [the item], as his lot for having confessed ('/hold pa'i 'gel), grant him [exemption] from the khrin for whatever item of wealth he bought. Concerning the item of wealth itself, return it Seal. (rlrun po las dkor nyos pa las/ nyo nyo bas/ slar 'lhol ba'i khrims la/ dkor zhig rkun pos rk:us pal las rkun po des/ myi pa rol po la/ btsongs pa las// glo zla dang thugs nyin dbang pos rkun po dang nyo nyo ba la/ ma brgal ma rmas par/ nyo nyo bas/ slar 'lhol te sladu pul na 'thold pa'i 'gel du/ dkor ji nyos pa'i khrin las gnang 'o Ill dkor dngos ni slar pul chig phyag rgya'o/11) (ITJ 753, 11. 43-47)?
In his treatment of this passage, rather than translating khrin, Thomas glossed khrin las as phrin las, citing as justification the variation between khrom and phrom (Thomas 1936: 283-284, 286, n. 33). If we discount this gloss and take khrin at face value in this passage, then it is a valuable example in which the terms khrims and khrin are employed as separate words. It would be premature, however, to attempt to address the relationship between khrin and khrims based on this evidence alone.
7 For Thomas' transliteration and translation, see Thomas (1936: 280). The translation of the terms glo zla and thugs nyen are uncertain. On the latter term, see Dotson (2009: 113, n 265).
ON THE OLD TIBET AN TERM KHRIN
83
2.2. Khrin in a previously untranslated OT legal text,
"Replies Concerning the Dice Statutes from the Tiger Year Dice Edict" (ITJ 740) Fortunately, another OT legal document, only recently published, is unequivocal in its use of both khrims and khrin. The text is ITJ 740, which consists of a long scroll (849cm X 26cm) containing the Chinese Suvarl)aprabhasottamasU.trendrarlijasiitra in 473 columns on the recto and two separate but related Tibetan texts on the verso. The frrst text, which comprises the frrst 23 7 lines, concerns mo divination, and the second consists of 122 lines containing various replies given to legal questions arising from the implementation of a new legal edict I have treated the document in detail elsewhere, and analysed the use of dice in legal cases (Dotson 2007a). Here it is only the legal text that concerns us. 8 The text, entitled ''Replies concerning the dice statutes from the tiger year dice edict" (stagllo'l bka'l sho byung be'i sho tshlgs gyl zhus Zan) (ITJ 740 (2), 1. 238),9 is divided into eleven clauses pertaining to par8 An edited transliteration of the entire document, including the mo divination manual, can be fmmd on the website of Old Tibetan Documents Online (OTDO) and in Imaeda and Takeuchi et al. (2007: 334-335). My own transliteration of the second text, along with photographic reproductions, can be found in Dotson (2007a), and demonstrates the patterned use of the double and single tsheg in the text. I have edited the text as lightly as possible, adding glosses only where they seemed absolutely necessary. Heavier editing can be found in the OTDO transliteration. My relevant editing conventions are as follows: I Reverse gi gu. i Indiscriminate gi gu. [abc] Intentional deletions in the original. [abc] Reliable reconstruction of text missing in the original. Text intercalated below line. abc abo abc 9 In fact, the term bka' sho poses some problems for translation It is a dice edict that comes from the centre of the empire, but it also carries with it a sense of something approaching 'anmesty' It is not exactly a reprieve or a pardon, but rather extends the possibility of such to certain cases where punishment would otherwise be a certainty, and thus offers defendants the possibility of being let off the hook by a roll of the dice. This benevolent quality of the term bka' sho appears to be the proper context in which to lDlderstand Khri Srong Ide btsan's gift of a bka' sho to his subjects just after his coronation, as recorded in the Old Tibetan Annals' entry for 756-757; see Dotson (2009: 128-129). It is not clear from ITJ 740 (2) exactly how dice were employed to decide cases, since the relevant procedure is always taken for granted in the communications between the judges of the court retinue and the ministers of the exterior (phyi blon). In my work on this topic, I tentatively assumed, based largely on the fact that IT J 740 (2) is immediately preceded on the same scroll by a divination manual-ITJ 740 (1)-
84
BRANDON DOTSON
ticular cases where local magistrates have petitioned the judges of the court retinue (pho brang 'khor gyi zhal ce pa) to clarify whether or not recourse to dice is appropriate (Dotson 2007a: 32-33). Before moving on to the numerous instances of the term khrin in ITJ 740 (2), let us frrst consider the two cases in which the term khrims appears. In clause Vlll, which concerns the proper procedure for punishing a messenger who loses horses, wages, etc. entrusted to him, lines 296-297 read, "according to the official law, debts up until the present having been repaid ... " (bka' khrlms gyis skyin ba' 'da' bar 'byung pas sbyangste). The ftnal clause of the document, clause XI, concerns the proper manner in which estates provision soldiers, and punishment for failing to send the required provisions. It also employs the term 'law' (khrims) in lines 343-344: "concerning the provisions sent by the estate holder (gzhi bu.), according to the law, the soldiers own them." (gzhl pus 10 brdzangs pe 'i rdzangs Ita bo khrims gyls dmag myis dpangste P). These examples reveal that khrims and bka' khrims were employed in their usual sense to indicate 'law' and its honorific, 'official law'. Having established this normative usage of khrims as 'law' in this text, we can now consider its use of the term khrin. The term appears fourteen times. In seven of these instances, khrin appears as part of a standard phrase, so there are therefore only eight discrete examples of the term's employment. that recomse would be made to a divination manual, which would then give a prognosis that would decide the defendant's fate; see Dotson (2007: 30). Ai Nishida, who is undertaking a study of Old Tibetan divination texts, has since made clear to me the problems arising from such an assumption In particular, a prognosis might be clear enough when it says that it is 'good' or 'bad', but there are also those that are 'average' ('bring), leaving one at a loss as to how to decide, other than to roll the dice again and hope for a prognosis that can more easily translate into a 'yes' or 'no'. More likely, it seems, is that the accuser and accused each rolled dice, with the higher roll wirming, as is attested in later Tibetan legal tradition; see, e.g., Cassinelli and Ekvall (1969: 176). Obviously this is also eminently relevant to the relationship between legal and rituallexicons, since here the scene of accused and accuser rolling the dice with possibly severe punishment hanging in the balance would seem to inform the similar mock scenes enacted between the glud 'gong and a representative of the Dalai Lama at the Glud 'gong rgyal po ceremony instituted under the Fifth Dalai Lama and performed up until the Chinese invasion of Tibet in 1950; see Richardson (1993: 60-73). The fateful use of dice in a legal context, alongside their use in divination, also infonns the appearance of dice in the iconography ofDpalldan 1ha mo; see R6na-Tas (1956). 10 Readbus. 11 Read dbangste.
ON THE OLD TIBET AN TERM KHRIN
85
2.2.1. In seven cases, khrin appears in the context of the phrase, "as for the khrin, it can be decided by means of dice" (khrin ni shos chod). This phrase occurs within the context of a legal document that prescribes the use of dice to resolve numerous matters within various legal cases. In order of appearance, the examples are as follows: 1) Concerning debt khrin, shall it be carried out (dgumi 2 by means of dice or not? How do you command? (skyin khrin lta po shos dgum 'am myi dgum llj1ltar 'tshal;) (ITJ 740,
1. 266).
2) As for the khrin, it can be decided by means of dice. (khrln ni shos chod;) (IT J 740, 1. 274). 3) If one can decide by means of dice the khrin for the men who forcibly stole [another man's wife] and abruptly sold her ... (my! dpanl 3 par
brk:us pa dang I thugs thubdu btsongSpa'i rnams I khrln ni shos chod na;) (ITJ 740, 11. 281-282). 4) In earlier dice edicts, the khrin of judicial confiscation14 of wealth borrowed from the church (dkon [bkon] mchog; lit ''the [three] jewels"Y~ could be decided by means of dice. (bka' sho't sngun roldu bkon mchogl dkor pa las zhal ce brlsad bzung pa'I khrin ni shos chod) (ITJ 740, 11.
303-304). 5) Concerning something borrowed (bskyls pa) from the wealth of the church and interest on a loan (bun skyeci), the 'gift' (btang) itself should not be decided by means of dice. Concerning interest on a loan and fmes and khrin, they shall be decided by means of dice. (bkon mchog gsum gyl dkor las bskyls pa dang bun skyed btang dngos ni shos my! gcado I
bun skyed dang chad khrln ni shos gcado;) (ITJ 740, 11. 314-315). 12 In separate studies of the Old Tibetan contract PT 1098, both Coblin (1991 a: 86-87, 90) and Takeuchi (1995: 264-66) arrive at similar translations of this verb as 'appropriately done, responsibly carried out', and 'fulfil', respectively. See also Franke eta/. (2007b: 207). 13 Read dhang. 14 The Old Tibetan term zhal eel zhal ees means 'law' and sometimes 'lawsuit' Here zhal is likely not the honorific for mouth (kha), but a word related to Jail heal/ gzhal/ Jol, 'weigh, assess, ponder, judge'; see Dotson (2007 a: 35, n 39). 15 The 'three jewels' refers to the Buddha, his teachings (Dharma) and the religious commmrity (Sangha), which together comprise the Buddhist religion. Here the term indicates the latter group and their property. The choice of the term 'church' is therefore dictated by the context, and, despite its disadvantages, not all of its Christian baggage is irrelevant. In particular, it recalls the metaphorical use of church in the dyad 'Christ and his Chmch', where the chmch and its property, by developing systems of inheritance that bypassed ties of kinship, gave rise to legal corporations in Emope. This finds its rough parallel in the Tibetan bla brang, which were properties held in trust by an incarnation lineage.
86
BRANDON DOTSON
6) Taking as an example the dice statutes of the horse year, accrued interest and khrin pertaining to the wealth of the church shall be decided by means of dice. (rla'llo'i sho tshlgs gyl 4Je 'ang gsol pa lagste I bkon chogi dkor gyi [d]gyur skyed dang khrin ni shos gcad;) (ITJ 740, ll. 316-317). 7) According to the dice statutes, it being said that all crimes previous and up to the present shall be decided by means of dice, the khrin can also be decided by means of the dice edict. . . (sho tshlgs las sngan cad nyes byungo chog shos gcado zhes byung bas khrin yang bka shos chocl) (ITJ 740, ll. 327-329).
While these phrases are all similar in that the khrin is always decided by means of dice, they help to locate khrin within the context of similar tenns. In the above examples, khrin occurs in conjunction with debt (1. 266), kidnapping (11. 281-282) and judicial confiscation (11. 303-304). Further, the tenn is used in conjunction with interest (bun skyed), punishments/ fmes (chad) (ll. 314-317), and crimes (nyes). 2.2.2. Moving on to other instances of khrin in the text, we fmd the following passage concerning a dispute between lender and borrower: the lender (lit. 'the one pursuing the interest'; gyur dedpaY 6 requests that the interest should be included within those [cases of] loans that should not decided by means of dice, and that not constituting a khrin, it therefore should not be decided by means of the dice edict. .. (gyur ded pe'l mchid nas ni gyur yang bu lon shos my! gcad pe'i nang 'du 'du I khrin ma lags I pas bka' shos my! gcad par gsol ces mchl;) (ITJ 740, ll. 252-253).
2.2.3. The next example is from clause VITI (ll. 286-303), which concerns the punishment of a man who has lost goods entrusted to him, apparently as part of his corvee labour obligations. . . . Concerning faulty dispatch, it is said that one undergoes (thebs) fmes/ punishments and khrin, and this is so. (bkye bskyon lta bu chad khrln thebs lags zhes bgyi pa' 'dl lags;) (ITJ 740, 1. 290).
Here again we note the close relationship between khrin and fmes/ punishments (chad).
16
On this and other terms for lender and borrower, see Dotson (2007a: 28, n 34).
ON THE OLD TIBET AN TERM KHRIN
87
2.2.4. A subsequent phrase from the same clause employs khrin in conjunction with debt (skyin). Although it came about that, according to the replies of the horse year, the khrin from debt of what he borrowed and was given should be decided by means of dice, ... (phor phog pa dang g.yar pe'i skyln khrlnl rta'i lo'i zhu lan las shos gcad par 'byung I gls gyang ;) (ITJ 740, 11. 295-
296).17
2.2.5. Clause IX, which concerns the legal status of Buddhist estates and the clergy, reveals the connection between khrin and legal cases (zhal ce). Any legal cases and so forth pronounced as khrin should be decided by means of dice for [cases concerning] the church on downwards. (zhal ce
lastsogs pa khrin du rma'o 'tshaf 8 bkon mchog man cad gyi shos gcado;) (ITJ 740, 11. 313-314).
2.2.6. Clause XI, which concerns the manner in which men were conscripted as soldiers and provisioned for war, employs the term khrin three times. The ftrst is as follows: Concerning military punishment itself, if not pronouncing the khrin also arises from the statutes of the dice edict, ... (dmag chad dngos ni khrin myi
rma bar bka' sho'i tshigs las gyang 'byung na;) (IT J 740, 11. 333-334).
17 On the use of the word pho in this passage, we find it in the terminative case in the following constructions, 'entrusted to him' (phor mnos; 1. 287, 1. 292), 'left to him' (por htang; 1. 293) (Dotson 2007a: 64-65; Imaeda and Takeuchi et al. 2007: 343). ln parallel with these phrases, I read phog as the past tense of 'hog 'to give, bestow' 18 Here, as in the Old Tibetan legal docmnent PT 1071 and the Shangshu paraphrase, 'o 'tshal appears to ftmction as a plmaliser similar to 'o cog. In the fonner text we find two distinct social groups referred to as 'military royal subjects' (rgyal 'hangs rgod do 'tshal) and 'civilian royal subjects' (rgyal 'hangs g.yung 'o 'tshal), respectively (PT 1071, 11. 277, 287; Dotson 2004: 82; Dotson 2007a: 8-9; Imaeda and Takeuchi et al. 2007: 104). See too Richardson (1998 [1990]: 154 and 164, n 33), where he translates 'o 'tshald with 'any' In treating this tenn in the Shangshu paraphrase, Coblin points out that it co-occurs with the plmaliser 'o cog in the Prayers of De ga G.yu mtshal, and reasons that there must therefore be some difference between 'o 'tshald and 'o cog ( Coblin 1991 b: 320). Cf. Wangdu and Diemberger's translation of the Dha' hzhed, where hzang ngo 'tshal is translated as 'considered good' (Wangdu and Diemberger 2000: 75, 18b).
88
BRANDON DOTSON
2.2.7. The second example from clause XI again reveals the close relationship of khrin with punishment (bka' chad). In this instance, it relates to the execution or banishment of an entire family. when the military punishment (dmag chad) is death or banishment, the official punishment that is khrin19 extends to one's entire family ... (dmag chad 'gum 'am spyugs na I bu smad kun yang khrln gyl bka' chad la I thug pas I) (ITJ 740, 11. 346-348).
2.2.8. The fmal example is quite similar to the one that precedes it, and reinforces what should by now be quite evident: khrin, as Coblin surmised, indicates the punitive aspect of law . . . . Concerning the military punishment, which has become like the official punishment, extending also to banishment and death, the greatkhrin can be carried out (khums) by means of the dice edict. (bka' chad dang 'dra bar gyur pa dmag chad lta bo 'gum spyugs la thug pa yang khrin chen po yang bka' shos/ khums I) (ITJ 740, 11. 354-355).
It is evident from all of these examples that khrin, used in its normative sense, means 'legal or judicial punishment'. Its closest synonym is chad, meaning 'punishment', but sometimes 'fme '. Passages such as 2.1. 7 suggest that chad is subordinate to khrin as a particular to a generality, but this may oversimplify their relationship. In ITJ 740 (2), legal punishments are meted out for unpaid interest, debt, kidnapping, loss of property and evasion of soldier tax. The harshest judicial punishments were for the latter, which could result in the execution or banishment of an entire family. This definition fits easily with nearly all of the above examples from the inscriptions and other OT sources. 2.3. Alternate uses ofkhrin andkhrims in OT sources
Having established the normative use of khrin as 'legal punishment' or 'judicial punishment' within Tibet's legal lexicon, it is now necessary to consider the term's a-normative uses, along with a-normative uses of khrims. In this context, we fmd khrin in Old Tibetan ritual texts. We have seen this already in the example of 2.1.4, where the term ap-
19 Here khrin qualifies hka' chad, and is used adjectivally in the same manner that one might expect of the term khrims, meaning 'legal'
ON THE OLD TIBET AN TERM KHRIN
89
peared in the context of a funerary text In that passage, and in those below, the translation 'judicial punishment' makes little or no sense. 2.3.1. Khrin in the OT fonerary text PT 1042 In PT 1042, a funerary document, we fmd an interesting passage that includes the term khrin. It concerns the preparation of the sacrificial sheep (skyibs lug), which serves as one of the psychopomp animals that guide the deceased to the land of the dead. As for its ornaments, its right horn shall be wrapped with a golden
khrin, and its left horn shall be wrapped with a silver khrin. (rgyan ni rwa g.yas pal gser gyi khrin bas bskri/1 rwa g. yon pa dngul gyi khrin bas bskri/1) (PT 1042, 11. 107-108). 20
This passage reveals that khrin are 'ornaments' (rgyan) that may be wrapped around the horns of a sheep. 2.3.2. Khrin in the OT ritual text PT 1289 Another OT ritual text, PT 1289, employs khrin in much the same way as it is found in ITJ 731 in the example of 2.1.4: When you have killed the horse by means of khrin khrin as the beloved (psychopomp) horse ... (do ma snying dagsu rta khrin khrin gyis bsad na /) (PT 1289, v. 3, 1. 7).
In his treatment of this passage, Thomas read khrin as a variant for khrid, meaning 'to lead'. In this case, with the verbal duplication this could be translated ''when you have killed the horse as the beloved (psychopomp) horse by leading and leading it ... " This gloss is partially supported by the appearance of khrid in a similar phrase further on in the same passage: ''having killed [the horse] with the khrid, the great khrid... " (khrid cen khrid kyis bsad de) (PT 1289, 1. 608). In this case, however, khrid does not seem to be a verb. Given the occurrence of khrin in a similar passage in ITJ 731, it is perhaps incorrect to read khrin here as a variant for the verb 'khrid. We will now return to the passage treated above at 2.1.4. 2.3.3. Khrin in the OT ritual text ITJ 731, revisited ''Do not take the khrin of the previous great khrin" (snga khrin che'i khrin ma blangs) (AFL: 17). In his translation of the line snga khrin che'i khrin ma blangs, Thomas, working presumably from context, translated khrin with 'authority' (AFL: 17). We have established from 20
See also Lalou (1952: 346, 357).
90
BRANDON DOTSON
the above two examples from two other OT ritual texts, however, that a khrin can be an ornament that may be wrapped around a sheep hom. It is also somehow implicated in the killing of the sacrificial horse, though how exactly is not clear. Still, this is a rather specific meaning, and it probably rules out Thomas' translation of khrin. It also appears that khrin in these passages from three separate Old Tibetan ritual texts does not refer to legal punishment This complicates the picture of khrin, and suggests that the term also had a specific role in funerary rites. I cannot here resolve the matter completely, but can tentatively offer that khrin in the above passages from OT ritual texts is a noun derived from the verb 'khrid, 'to lead', and indicates something to lead with, i.e. 'a lead'. This may also apply to khrid in PT 1289 based on din suffix alternation. In the first example from PT 1042, the lead is made of silver and gold, and wrapped around the sheep's horns. In PT 1289 at 2.3.2, the horse is killed with the lead as (read: in order to create) the beloved (psychopomp) horse. In the second phrase in PT 1289, khrid cen khrid kyis bsad de, we have a similar meaning if one accepts here the equivalence of khrid and khrin. Here khrid chen khrid is similar to khrin che'i khrin in ITJ 731. The latter passage is perhaps the most difficult to resolve. In saying, ''Do not take the lead of the previous great lead" (snga khrin che'i khrin ma blangs), I take it that the great lead refers metonymically to the sacrifice of hundreds of horses required for the funeral and for the creation of the psychopomp horse. It is this, and the expense required, that the brothers oppose. There remains the problem of how a lead can be used to kill a horse. Here I can only suppose that it refers to suffocation in the same manner that livestock are killed before being butchered in many parts of Tibet: after being hobbled, a rope is wound tightly around the animal's mouth, after which it suffocates in about half an hour (Thargyal 2007: 87). This way of killing the sacrificial horse has the advantage ofleaving the animal's body intact for its duties as a guide to the Beyond. It also recalls principles of sympathetic magic, with the animal that 'leads' the deceased being killed by means of a lead. I should stress that this is a tentative conclusion, and one that I hope will be confmned (or corrected) with recourse to further instances of khrin in a ritual context. Khrin is not, in fact, the only term from the legal and administrative lexicon found in ritual texts. The passage quoted above in the funerary text PT 1042 is immediately preceded by instructions concerning tal-
ON THE OLD TIBET AN TERM KHRIN
91
lies or tally sticks (khram), implements used for record-keeping during the imperial period, but employed here as part of the funeral ceremony. 21 Notably, one of these tallies is attached to the psychopomp sheep (skyibs lug): The rgyaf-2 cut the scroll and made two tallies. 23 One the [rgyafs] servant entrusted to the lord [the deceased] as the large[r half of the] scro11. 24 [He] entrusted the other tally to the psychopomp sheep. (rgyal gyis thang bead de khram/ gnyis bgyis tell gcig ni zha 'bring rje bo thang chen po la gtad/1 khram gcig ni skyibs lug la/ gtad/1) (PT 1042, 11. 105107).2~
21 On tally sticks (khram/ khram hu) and wooden slips (byang hu), see R6na-Tas (1956), Takeuchi (2003), Dotson (2007a: 35-36), and Uebach (2008). zz The rgyal appears nine times in PT 1042 where it is lDl.equivocally an actor in the funeral rite. In one place in the same text (1. 92) we also find the more well-known term rgyal thag brgyad, usually described as a structure associated with ftmerary rites that marks off an "aire delimi.tee"; see Stein (1971: 494, n 44). Here the rgyal, which seems to be a person, must be distinguished from the rgyal thag brgyad. 23 The term thang, often translated with 'authority', but also meaning 'jurisdiction' and 'rights and duties', is discussed in detail in Uebach (2008), and also in Dotson (2009: 52-53, n 73). Above I have followed the recent and illuminating discussion of the term by Zeisler (forthcoming), who argues corwincingly that its original point of reference was an lDl.rolled scroll. The passage above would seem therefore to retain the vocabulary of the tally stick, a writing technology that predates the introduction of paper to Tibet and which was used alongside paper for over one hundred years: though it is paper that is cut, its two parts are referred to as 'tallies' This is not lDl.expected, and is also folDl.d in the term 'pale tally' (khram skya), which is a tally kept on paper (U ebach 2008: 62-63). Z4 In Old Tibetan funerary texts the term lord often refers metaphorically to the deceased, usually in the pair 'lord (rje) and servant (kho/)', which most often indicates the deceased and a psychopomp animal, usually a horse; see Stein (1971: 490, n 41, 496). Here I have taken 'lord' to refer to the deceased, while lDl.derstanding the zha 'bring to be a servant of the rgyal, probably identical with the zha 'bring pa at 1. 66 and possibly also with the zha 'bring nang pa at 1. 32 and 1. 126. This is a problematic passage, and my translation has ass1Dl.led a change of agent after the semi-final particle te, and the absence of the usual syntax of the verb gtad, according to which one would hope to find zha 'bring marked in the ergative, and the recipient marked in the dative. One alternative, in which the rgyal remains the agent, would be to read zha 'bring rje bo, 'lord of the servants', as the recipient of the larger half of the tally. Such a construction with a verb of giving in which the beneficiary is in the absolutive and not the dative is also found in the Old Tibetan Annals; Uray (1960: 50, n 57) notes three instances, one of which is discussed in slightly more detail in Dotson (2009: 116, n 2 77). There are other such examples of the receiver being marked in the absolutive in the Annals, the Old Tibetan Chronicle, and other Old Tibetan documents, and this construction will be discussed in greater detail in Zeisler (forthcoming). 25 See also Lalou (1952: 346, 357); and Uebach (2008: 58, n. 8).
92
BRANDON DOTSON
This is the same sheep that had the silver and golden 'leads' (khrin) wrapped round its horns at example 2.3 .1. One can see here a bureaucratisation of the afterlife reminiscent of Chinese religion, with the employment of technologies from administrative practice, and with the experience of the deceased paralleling in many ways that of a man brought before a judge. Relating the use of khrin in :ftmerary texts to its appearance in legal texts, it is probably safe to assume that the legal use of the term khrin as 'legal punishment' comes from its literal meaning as 'a lead' or 'tether', which was either used figuratively or metonymically, referring to the physical conditions of those undergoing legal punishment 26 Specifically, the lead may have been attached to the condemned man or the one facing legal punishment. Similarly, other legal and administrative terms such as 'cotvee horse/ stage horse' (slungs rta) and 'military division banner/ horn banner' (ru dar) are also employed in a ritual setting. Moreover, the latter two terms appear to have undergone a sort of apotheosis to become, respectively, prayer flags (rlung rta) and the banner of a wrathful deity (ru dar) only after the collapse of the Tibetan Empire. 27 On the other hand, khram and khrin were employed in the ritual lexicon at the same time as they were used in the legal and bureaucratic lexicon during the period of the Tibetan Empire itself. The relationship between these two lexicons is very complex, and can therefore not be reduced to the mere apotheosis of legal and administrative structures into the ritual lexicon and wrathful pantheon after the collapse of the Tibetan Empire, once their original points of reference had presumably become opaque. 28
2.3.4. Khrims and grims Turning now to khrims, in the Old Tibetan Chronicle and in other OT texts, khrims is frequently spelled grims, and such alternation in spelling is quite common in OT. It is worth noting here that grims covers such meanings as 'concentrated', 'focused', 'tight', 'contracted', 'compressed', 'constricted' and 'cramped'. Coupled with the use of khrin in 26 I am indebted to Bettina Zeisler for this line of reasoning, and indeed for several other insightful comments on drafts of this chapter. Zl On 1his explanation of 1he terms rlung rta and slungs rta, see Uray (1984: 358359, n 56). Ontu dar as the flag of a wrathful deity, see Pommaret (2007: 146). 28 Such a process might account, however, for the 'Zhang zhung' tenn khrin, which indicates "a class of non-hmnan spirits" (mi ma yin gyi rigs zhig) (Pasar Tsultrim Tenzin et al. 2008: 27).
ON THE OLD TIBET AN TERM KHRIN
93
the above examples as the nmm 'lead', this calls to mind the metaphors used to describe Tibet's legal system as a golden yoke or a silken knot 29 For example, in the Dpyid kyi rgyal mo glu dbyangs, a Buddhist history of Tibet composed under the auspices of the Fifth Dalai Lama (1617-1682), we fmd the following metaphor: ''like a golden yoke, the royal law is heavy, and presses down on the napes of all, high and low" (rgyal khrims gser gyi gnya' shing /tar lei bas mchog dman kun gyi gnya' ba gnon par mdzad dol) (DLV: 64). 3. CONCLUSIONS
In response to Coblin's initial question, we can now state that the normative use of the term khrin in ITJ 7 40 (2) alongside khrims (law) reveals that it is not in fact a dialectical variant of khrimsl grims, but a separate term with a related meaning. The employment of khrin in this and other OT documents agrees with Coblin's statement that khrin represents ''the penal or punitive aspects of khrims". This may be easily seen from the contexts in which khrin appears, and justifies its translation as 'legal punishment' or 'judicial punishment'. In the Old Tibetan ritual lexicon, the khrin is both an 'ornament', made of gold or silver, that may be wrapped around a sheep's horns, and an instrument capable of killing a horse. Accordingly, I have provisionally translated it with 'a lead' or 'tether'. In its dual employment in both the administrative and ritual lexicons, khrin is not alone, and takes its place alongside such terms as 'tally/ tally stick' (khram), 'corvee horse/ stage horse' (slungs rta) and 'military division banner' (ru dar). The meaning of khrin as both a technology that plays a role in leading the deceased to the afterlife and as a legal punishment--the latter meaning derived most likely from the practice of binding a prisoner and leading him to judgement-recalls the classic metaphors that describe Tibet's royal and religious legal systems as, respectively, a golden yoke and silken knot that bind the Tibetan people and lead them to liberation. These are slightly oppressive metaphors, but they pale in comparison to the 'lead' or 'tether', and its ominous connotations, both punitive and funerary. At the same
29 Equally, it is reminiscent of the words 'ligature' and 'religion', each from the latin ligare, 'to bind'
94
BRANDON DOTSON
time, the tenn khrin gives us a fascinating alternative vision of the 'union of religion and state' (chos srid zung 'brel) in an early Tibetan context
BIBLIOGRAPHY
Cited Old Tibetan Documents Annals ofthe ~ zhaPrincipality (fLDT 2) ITJ 731 Fragmentary Old Tibetan funerary text, "End of the Good Age and Tragedy ofthe Horse and Yak" ITJ 7 40 (2) Old Tibetan legal document, "Replies Concerning the Dice Statutes from the Tiger Year Dice Edict" Old Tibetan legal docmnent dealing with theft. ITJ 753 PT 1042 Old Tibetan funerary text. PT 1071 Old Tibetan legal text PT 1283 "A Dialogue Between Two Brothers" PT 1289 Old Tibetan ritual text. Skar chung Edict Edict dating to c.812 and accompanying the Skar clrung Inscription It confirms Buddhism as the official religion ofTibet. Zhol Pillar Inscription Inscription dating to c.764, records grants to an aristocratic lineage in recognition of their service during Tibet's brief conquest of the Chinese capital. Zhwa'i Lha khang Inscriptions The earlier, west face inscription dates to between c.798 and c.800. The inscriptions record grants made to Myang Ting nge 'dzin Bzang po and
his clansmen
Abbreviations AFL CD2 DLV ITJ KhG OTDO
PT TLTD
AncientFolkLiteraturefrom Northeastern Tibet, Thomas (1957) Choix de documents tihetains, 2, Spanien and Imaeda (1979) Rgyal ba Lnga pa Chenmo (Fifth Dalai Lama)(ed 1980 [1957]) Indicates that the text is an Old Tibetan docmnent from the India Office Library, kept in the British Library in London. The letter and rrumbers following 'IT' indicate a document's shelf mark. Dpa' bo Gtsug lag Phreng ba (ed 1985) Old Tibetan Documents Online, http://www.aa.tufs.ac.jp/ Pelliot tibetain. The number following indicates the Old Tibetan document's shelf mark at the Bibliotheque nationale de France. Tibetan Literary Texts andDocwnents, Thomas (1935/1951/1955/1963)
References Bacot, Jacques, Frederick William Thomas and Gustave Charles Toussaint 19401946. Documents de Touen-Houang Relatifs a l'Histoire du Tibet. (Arm.al.es du Musee Guimet, Bibliotheque d'etudes, tome 51.) Paris: Libraire Orientaliste Paul Geutlmer. Btsan lha Ngag dbang Tshul khrims. 1997. Brda dkrol gser gyi me long. Beijing: Mi rigs dpe skrunkhang.
ON THE OLD TIBET AN TERM KHRIN
95
Cassinelli, C.W. and Robert Ekvall. 1969.A Tibetan Principality: the Political System
ofSa skya. Ithaca: Cornell University Press. Chab spel Tshe brtan Plum tshogs. 1997. T1m hong gi gna' yig
zhes pa'i rtsa 'grel. Beijing: Mi rigs dpe sknm khang. Coblin, Weldon South 199la. Notes on Old Tibetan Ije-blas. In Steinkellner, Ernst, ed., Tibetan History and Language: Studies Dedicated to Uray Geza on his Seventieth Birthday. (Wiener Stu.di.en zur Tibetologie Buddhismuskunde, Heft 26.) Wien: Arbeitskreis fur Tibetische mtd Buddhistische Studien: 63-110. - - . 199lb. A study of the Old Tibetan Shangshu paraphrase, part I. Journal of the American Oriental Society 111.2: 303-322. Dotson, Brandon. 2004. A note on zan: maternal relatives of the Tibetan royal line and marriage into the royal family. JoumalAsiatique 292.1-2: 75-99. - - . 2007a. Divination and law in the Tibetan Empire: the role of dice in the legislation of loans, interest, marital law and troop conscription. In Kapstein, Matthew T. and Brandon Dotson, eds., Contributions to the Cultural History ofEarly Tibet. (Brill's Tibetan Studies Library 14.) Leiden: Brill, 3-77. - - . 2007b. 'Emperor' Mu rug btsan and the 'Phang thang rna Catalogue. Journal of the Jntemationa/Associationfor Tibetan Studies 3: 1-25. - - . 2009. The Old Tibetan Annals: an Annotated Translation of Tibet's First History. With an Annotated Cartographical Documentation by Guntram Hazod (Philosophisch-Historische Klasse, Denkscluiften, 381. Band. V eroffentliiclrungen zur Sozi.alantbropologie Nr. 12.) Vienna: Verlag der Osterreichi.schen Akadernie der Wissenschaften. Dpa' bo Gtsug lag Phreng ba (1504-1566); Dam pa'i chos kyi 'Ichor lo bsgyur ba mams kyi byung ba gsal bar byed pa mkhas pa'i dga' stan. Beijing: Mi rigs dpe sknm khang, 1985. Franke, Herbert, Jens-Uwe Hartmann, and Thomas 0. Hollmarm, eds. 2007a. W6rterbuch der tihetischen Schriftsprache. 3 Liefenmg, kau stu bha - bka ' brgyud Bearbeitet von Petra Maurer und Johannes Schneider. (Kommissi.on fiir zentral- und ostasiatische Studi.en) Miinchen: Verlag der Bayerischen Akademie der Wissenschaften in Kommission Beim Verlag C. H. Beck. - - . 2007b. W6rterbuch der tihetischen Schriftsprache. 4 Liefenmg, bka 'bsgo - skad silan pa. Bearbeitet von Petra Maurer und Johannes Schneider. (Kommission fur zentral- und ostasiatische Stu.di.en.) Miinchen: Verlag der Bayerischen Akadernie der Wissenschaften in Kommission Beim Verlag C. H. Beck. Imaeda, Yoshiro, Tsuguhito Takeuchi, Izumi Hoshi, Yoshimichi Ohara, Iwao Ishikawa, Kazushi Iwao, Ai Nishida, and Brandon Dotson, eds. 2007. Tibetan Documents from DW'lhuang Kept at the Bibliotheque Nationale de France and the British Library. (Old Tibetan Docmnents Online Monograph Series, 1.) Tokyo: Research Institute for Languages and Cultures of Asia and Africa. Lalou, Marcelle. 1952. Rituel Bon-po des fimera.illes royales. Journal Asiatique 240: 339-361. Li, Fang Kuei. 1959. Tibetan glo-ba-'dring. In Egerod, Smen and Else Glahn, eds., Studia Serica Bernhard Karlgren Dedicata. Sinological Studies Dedicated to Bernhard Karlgren on his Seventieth Birthday, October fifth, 1959. Copenhagen: E. Munksgaard, 55-59. Li, Fang Kuei and Weldon South Coblin 1987. A Study of the Old Tibetan Inscriptions. (Special Publications No. 91.) Taipei: Institute of History and Philology, Academia Sinica. Pasar Tsultrim Tenzi.n, Changru Tritsuk Namdak Nyima, and Gatsa Lodroe Rabsal (ed. Yasuhi.ko Nagano and Samten Karmay, trans. Heather Stoddard). 2008. A
96
BRANDON DOTSON
Lexicon of Zhang zhung and Bon po Terms. (Semi Ethnological Reports 76. Bon Studies 11.) Osaka: National Musemn of Ethnology. Ponunaret, Fr811.9oise. 2007. Estates and deities: a ritual from central Bhutan. The bskang gso of 0 rgyan chos gling. In Ardussi, Jolm and Fran9oise Ponunaret, eds., Bhutan: Traditions and Changes. Proceedmgs of the Tenth Seminar of the International Association for Tibetan Studies, Orford 2003. Volmne 5. (Brill's Tibetan S1ndies Library 10.) Leiden: Brill, 135-158. Rgyal ba Lnga pa Chen mo (Fifth Dalai Lama, Ngag dbang Blo bzang Rgya mtsho; 1617-1682); Bod kyi deb ther dpyid kyi rgyal mo gju dhyangs. Lhasa: Mi rigs dpe sk:run khang, 1980 [1957]. Richardson, Hugh Edward 1952. Ancient Historical Edicts at Lhasa. London: Royal Asiatic Society. - - . 1985. A Corpus ofEarly Tibetan Inscriptions. (James G. Forlong Series XXIX.) London: Royal Asiatic Society. - - . 1993. Ceremonies of the Lhasa Year. London: Serindia. - . 1998 [1977]. Ministers of the Tibetan kingdom. In Aris, Michael V., ed, High Peaks Pure Earth. London: Serindia: 56-73. - . 1998 [1990]. H1D1ting accidents in early Tibet. In Aris, Michael V., ed, High Peaks Pure Earth. London: Serindia, 149-166. Rnam rgyal Tshe ring, ed. 2001. Bod yig brda rnying tshig mdzod. Beijing: Knmg go'i bod rigs dpe sk:run khang. R6na-Tas, Andras. 1955. Social terms in the list of grants ofthe Tibetan Tun-Huang Chronicle. Acta OrientaliaAcademiae Scienliarum Hungaricae 5: 249-270. - - . 1956. Tally stick and divination dice in the iconography ofLha-mo. Acta OrientaliaAcademiae ScientiarumHungaricae 6: 163-179. Spanien, Ariane and Yoshiro Imaeda. 1979, eds. Choix de documents tibetains conserves a Ia Bibliotheque nationale: complete par quelques manuscrits de 1'India Office et du British Museum. Tome 2. Paris: Bibliotheque Nationale. Stein, RA. 1970. Un docmnent ancien relatif aux rites funeraires des Bon-po Tibe1ains. JournalAsiatique 258: 155-185. - - . 1971. Du recit au ri1nel dans les manuscrits Tibetains de Touen-houang. In Macdonald, Ariane, ed, Etudes Tibetaines Dediees a Ia Memoire de Marcelle Lalou. Paris: Adrien Maisonneuve, 479-547. Takeuchi, Tsuguhito. 1995. Old Tibetan Contracts from Central Asia. Tokyo: Daizo Slruppan
- - . 2003. Military administration and military duties in Tibetan-ruled Central Asia (81h-91h Century). In McKay, Alex, ed, Tibet and Her Neighbours: a History. London: Edition Hansjorg Mayer: 43-54. Thargyal, Rinzin. 2007. Nomads of Eastern Tibet Social Organization and Economy of a Pastoral Estate in the Kingdom of Dege, ed. Toni Huber. (Brill's Tibetan S1ndies Library 15.) Leiden: Brill. Thomas, Frederick William. 1935, 1951, 1955, 1963. Tibetan Literary Texts and Documents Concerning Chinese Turkestan. Vols. I-IV. London: The Royal Asiatic Society. - - . 1936. Law of theft in Chinese Kan-su: a Ix'h-xt" century fragment from Tunhuang. Zeitschriftfiir vergjeichende Rechtswissenschaft 50.3: 275-287. - - . 1957. Ancient Folk Literature from Northeastern Tibet. (Abhandlungen der Deutschen Akademie der Wissenschaften zu Berlin, Klasse fiir Sprachen, Literatur und Kunst, Jahrgang 1952, Nr. 3.) Berlin: Akademie Verlag. Tucci, Giuseppe. 1950. The Tombs of the Tibetan Kings. (Serle Orientale Roma 1.) Roma: Is.M.E.O.
ON THE OLD TIBET AN TERM KHRIN
97
Uebach, Helga. 2008. From red tally to yellow paper: the official introduction of paper in Tibetan administration in 7441745. Tibetan Studies in Honow- of Samten Karmay, Part 1 Revue d'Etudes Tibetaines 14: 57-69. Uray, Geza. 1960. The fom horns ofTibet according to the Royal Annals. Acta OrientaliaAcademiae Scientianlm Hungaricae 10.1: 31-57. - - . 1984. The earliest evidence of the use of the Chinese sexagenary cycle inTibetan In Ligeti, Louis., ed., Tibetan and Buddhist Studies Commemorating the 200thAnniversary of the Birth ofAlexander Csoma de Kor6s. Volmne II. (Bibliotheca orientalis Hungarica vol. 19/1-2.) Budapest: Akademi.ai Kiad6, 341-360. Wangdu, Pasang and Hildegard Diemberger. 2000. Dba' Bzhed: The Royal Narrative Concerning the Bringing of the Buddha's Doctrine to Tibet. (PhilosophischHistorische Klasse, Denkschriften, 291. Band. Beitriige zur Kultur- Wld Geistesgeschichte Asiens Nr. 37.) Wien: Verlag der OsterreichischenAkademie der Wissenschafl:en. Yamaguchi, Zui.ho. 1970. Matrimonial relationship between the T'u-fan and the T'ang Dynasties (part 2). Memoirs of the Research Department of the Toyo Bunko 28: 59-100. Zeisler, Bettina. 2004. Relative Tense andAspectual Variations in Tibetan Languages. A Comparative Study. (frends in Linguistics, Studies and Monographs, 150.) Berlin: Mouton de Gruyter. - - . forthcoming. For the love of the word: a new translation ofPt 1287, the Old Tibetan Chronicle, chapter I. In Imaeda, Yoshiro, Matthew T. Kapstein, and Tsuguhitu Takeuchi, eds., Studies on Old Tibetan Documents. (Old Tibetan Documents Online Monograph Series, 3.) Tokyo: Research Institute for Languages and Cultmes of Asia and Africa, Tokyo University of Foreign Studies.
A FUNCTIONAL ANALYSIS OF ADJECTIVES IN NEWAR KAzUYUKI KIRYU•
1. INTRODUCTION This paper deals with adjectives in Newar, a Tibeto-Bunnan language spoken in the Republic of Nepal. 1 In previous studies, adjectives in Newar have been loosely defmed based on semantic aspects of words without any rigorous criteria clearly distinguishing them from nouns and verbs. Since many words that may be identified as adjectives in Newar show similarities with verbs in that they inflect for tense/aspect, some scholars consider Newar to have no adjectives (Kolver 1977, Shakya 1999). According to the national census of 2001, the number of speakers of Newar is around 840,000, and the majority of the speakers are located in the Kathmandu Valley. The genetic affiliation of the language continues to be disputed. There are a few varieties of the language other than the one spoken in Kathmandu City, although this is usually regarded as the standard variety. Within the Kathmandu Valley, there are two dialectal variations: Kathmandu/Patan dialects and Bhaktapur dialect Another well-documented variety is Dolakha Newar, studied by Genetti (1994, 2007). Genetti (2006) recently reported another variety spoken in the Thautali area, close to Dolakha. She regards the Dolakha and Thautali varieties as consisting a linguistic subgroup of 'Newaric'. Newaric also includes the Kathmandu Valley dialects as a peer subgroup. * This is a revised and extended version of my paper presented at the 11th Himalayan Language Symposium held at Chulalongkom University, Thailand in 2005. I'd like to express my gratitude to my two Newar consultants, Manik R. Shakya and Lata Shakya, from Patan. In the process of revising the draft of this paper, I received valuable comments and suggestions from Mark Turin and Bettina Zeisler. My special thanks to them. It goes without saying that any remaining mistakes are mine. 1 This language has been called 'Newari' in the literature, although this is now regarded as inappropriate since the denomination is based on the Nepali word At present, many scholars use the term 'N ewar' instead, although the formal denomination of the language is Nepal BhD,sii and the local denomination in the language is newii: bhiie. My transliteration of Newar basically follows the traditional transliteration scheme for the devanagari script, except that nasalisations are phonetically represented and that ae and iie represent the sounds le: I and Ire: I respectively.
100
KAZUYUKI KIRYU
Typologically speaking, Newar is a head-fmal language with dependent-marking: the basic word order is SOV and modifiers precede heads. The case system is of the ergative-absolutive type and verbs agree with the subject, showing a conjunct and disjunct dichotomy (Hale 1980). The main purpose of this paper is twofold: frrst, to establish clear criteria for verbs, nouns and adjective classes in Newar, arguing for a class for adjectives and proposing three distinct sub-classes based on morphosyntactic criteria along the lines proposed by Dixon (2004); second, to provide a detailed morphosyntactical and semantic description of these adjectives. The data used in the analysis were elicited from two consultants living in Patan, and are all drawn from my Newar story database. 2. ADJECTIVE CLASS Unlike nouns and verbs, it is sometimes difficult to define a prototype for adjectives. Typologically speaking, adjectives are apt to formally share properties found in either nouns or verbs, constituting an ambivalent category between the two. This gives rise to the naming of types of adjectives based on how much they resemble nouns or verbs, such as 'adjectival nouns', 'adjectival verbs' and so forth. Some scholars even dispute the existence of an adjective class. Furthermore, it has been said that some languages lack adjectives at all, and that the concepts that adjectives convey, for example, those in English, are lexicalised as verbs or nouns. However, Dixon (1982, 2004) argues for a distinct class of adjectives in every language. He suggests that 'there are always some grammatical criteria-sometimes rather subtle-for distinguishing the adjective class from other word classes' (Dixon 2004: 1). He goes on to discuss the primary divisions of adjectives in terms of grammatical properties, as in ( 1) below. The first parameter relates to the possibility of functioning as a predicate itself (1111). The second parameter relates to the morphological possibilities of adjectives within a NP (AlB). (1) Types of adjectives by grammatical properties (Dixon 2004: 14-15) I.
Adjectives which can function as intransitive predicate. These take some or all of the morphological processes and/or syntac-
A FUNCfiONAL ANALYSIS OF ADJECfiVES IN NEWAR
101
tic modifiers which can apply to a verb when it functions as intransitive predicate. They may be called 'verb-like adjectives' II. Adjectives which may fill the copula complement slot. These can be called 'non-verb-like adjectives'.
A When it functions within a NP, an adjective may take some or all of the morphological processes that apply to a noun. These can be called 'noun-like adjectives'. B. In a language where nouns show a number of morphological processes, none of these apply to adjectives. These can be called 'non-noun-like adjectives'. The parameters (1111) are related to the predicate position, and the parameters (AlB) are related to the noun modification. Dixon also summaries two canonical functions of adjectives and two further possible properties: (2) Two canonical functions of adjectives (Dixon 2004: 28) (a) in a statement that something has a certain property, coded through the adjective functioning either as intransitive predicate or as copula complement;
(b) as a specification that helps focus on the referent of the head noun in a NP, the adjective functions as modifier to the head. (3) Possible properties of adjectives (Dixon 2004: 28) (a) as the parameter of comparison in a comparative construction;
(b) as modifiers to a verb, in adverbial function.
In the following sections, the morphosyntactic criteria of adjectives and the semantic concepts for adjectives in Newar are discussed. 3. MORPHOSYNTACTIC PROPERTIES OF ADJECTIVES IN NEWAR Unlike languages with distinct morphology for the adjective class, there are no unique morphological properties that distinguish adjectives from nouns and verbs in Newar. The class of adjectives in Newar, however, can be determined by morphosyntactic criteria as in (4), and some examples illustrating them are presented in (5).
102
KAZUYUKI KIRYU
(4) Motphosyntactic criteria for adjectives: 1. An adjective ftlls in the complement slot of the copulaju:; 2. An adjective is negated by ma-ju:; 3. An adjective takes one of the adnominal sufftxes -mha, -pi: and -gu in noun modification; 2 4. An adjective may enter into the se cwane construction; 5. The reduplicated form before the adnominal sufftxes -pi: and -gu expresses plurality. (5)
a. tara saiberiya-pi: dhu thupi: swayll: ta:dhika: but Siberia-PL tiger these than big 'But the tigers in Siberia are larger than these.'
ju:. cop.ST
b. thikae ma-ju: expensive NEG-COP.ST 'That's not expensive.' c. basi-gu Ia rotten-ADN meat 'rotten meat' d. mha yau-se body light-ADV 'I feel fme.'
cwll: stay.ST
e. palu-palu-gu nasa spicy-spicy-ADN food 'spicy foods' N ewar adjectives can be divided into three sub-classes by virtue of the grammatical properties they exhibit, as in (6). (6) (a) Some function as intransitive predicate (verb-like adjectives); (b) Some function as copula complement of ju: (non-verb-like adjectives);
2 Dixon notes that noun modification 'involves just apposition of adjective and no1m, in others a relative clause (or similar) marker may be required' (Dixon 2004: 15). Newar is clearly a language of the latter type.
A FUNCfiONAL ANALYSIS OF ADJECfiVES IN NEWAR
103
(c) Some :ftmction only as a modifier, taking an adnominal suffix, -mha, -pi: or -gu when they modify nouns (non-predicative adjectives). The properties (6a) and (6b ), related to the frrst parameter (1111) in Section 2, suggest that Newar is a language with at least two adjective classes. The first adjective class are verb-like adjectives, so called because they behave like verbs, taking tense-aspect suffixes. The second class of adjectives are 'non-verb-like' adjectives. The property (6c), related to the second parameter (AlB), distinguishes nouns and adjectives. Nouns always require the genitive marker -ya in modifying another noun and take an adnominal suffix as an option, while adjectives do not take the genitive marker and an adnominal suffix is obligatory. In this sense, regarding the second parameter, adjectives in Newar are 'non-noun-like adjectives' In noun modification, all adjectives must take one of the adnominal suffixes that agree with the head nmm in terms of animacy and nwnber, which are also required by verbs to form a relative clause. Newar adjectives cannot directly modify nouns, unless the adjective and noun as a whole constitutes a compound noun, for instance, hyau: maca 'baby', as in (7a). The three kinds of adnominal suffixes are illustrated in (7b-d). Mha is for an animate singular noun head, pi: is for an animate plural noun head, and gu is for inanimate objects, regardless of nwnber.
(7) a. hyau:-maca red-child 'a baby' b. hyau:-mha maca red-ADN child 'a child who is in red' c. khusi ju:-pi: happy COP.ST-ADN.PL 'happy people'
d. thiiku-gu
Jya
di:fficult-ADN work 'difficult work'
manu:-ta man-PL
104
KAZUYUKI KIRYU
In Newar, a very limited nwnber of adjectives, such as nhu: 'such', me: 'other' and thwajya: 'such', may not stand in the complement position and are limited to noun modification. Although they might possibly belong to a different word class, I will label this class 'non-predicative adjectives'. The sole reason for their treatment as such lies in the fact that they exhibit a plural sense in reduplication in a manner that adjectives do. (8) a. me:-me:-gu other-other-ADN 'other foods'
b. nhu:-nhu:-gu new-new-ADN 'new books'
nasa food saphu: book
I now discuss the criteria for verbs and nouns. It is usually easy to recognise word classes for nouns and verbs in any given language. In Newar, too, the following sets of criteria can be established for the word classes of verbs and nouns. The criteria for verb and noun classes are summarised in (9) and ( 11) below. (9) Motphosyntactic criteria for verbs: 1.
A verb ftlls in the predicate slot of a clause by itself;
2.
A verb takes tense/aspect inflectional suffixes and has the stative form characterised by a lengthy vowel;
3.
A verb is negated by the ma- affix in the past form and by makhu in the future form.
Verbs appear in the predicate slot of a clause and inflect for tense and aspect, as in (lOa). There are three inflectional forms for tense/aspect: future, past and stative (or non-future). Past and stative forms are negated by ma- as in (lOb) and future forms by makhu as in (lOc).
ye-e (10) a. ji lSG.ABS Kathmandu-LOC 'I went to Kathmandu.'
b. mhiga:
wan-a. go-PC
wa ma-wa:. yesterday rain.ABS NEG-come.ST 'It didn't rain yesterday.'
A FUNCfiONAL ANALYSIS OF ADJECfiVES IN NEWAR
105
wa-i-makhu. c. kanhae ram tomorrow Ram.ABS come-FD-NEG 'Ram will not come tomorrow.' Following the criteria for nouns (11), examples are provided in (12). ( 11)
Morpho syntactic criteria for nouns: 1. 2.
A noun takes plural suffixes ( -ta or -pi:); 3 A noun may function as the complement of the copula verb
khaye; 3.
A noun takes the genitive marker -ya in noun modification;
4.
A noun takes case clitics. 4
(12) a. manu:-ta man-PL
'men' b. waeka: daktar 3SG.HON.ABS doctor.ABS 'He is a doctor.'
kha:. cop.ST
c. thwa saphu:-ya mu: this book-GEN price 'the price of this book'
3 It is compulsory for animate nOWIS in the plural to take either of the two plmal suffixes. See Hale and Shrestha (2006) for details. Although it is usually taken for granted that inanimate nOWIS do not take plural suffixes at all, the possibility of their taking a plmal suffix -ta is not necessarily rejected. In reality, such cases are often seen in publications and are accepted by many, if not all, speakers ofNewar. To some extent, however, their plmal marking may be said to be optional. 4 As suggested in Hale and Shrestha (2006), Newar case morphemes may be regarded as clitics, rather than inflections or particles, because they appear as the final element of a noun phrase, including a floated classifier phrase.
(i) cha-mha taJdhikaJ-mha
one-CL big-ADN 'to a big elephant' (ii) taJdhikaJ-mha
khisi
big-ADN elephant 'to one big elephant'
khisi-yii:ta elephant-OAT cha-mha-esita
one-CL-DAT
As (ii) shows, the case marker appears as the last element of the entire NP.
106
KAZUYUKI KIRYU
In the copula complement slot, nouns do not take the copula, ju:; they take a different copula, kha:. Newar has two copulas, kha: andju:. In a fmite copulative sentence, the complement nowt does not take ju: but takes kha:, while non-verb-like adjectives cannot take kha:, but must take ju: alone. 5
(13)
kha: ju:
NOUNS Yes No
ADJECTIVES No Yes
Verbs and nouns never fail to fulfil the criteria above and they are mutually exclusive. Newar adjectives are, on the other hand, not unique in morphological and syntactic properties. Some adjectives behave like verbs, taking inflectional suffixes that verbs take; some are like nowts; and others are different from verbs and nouns. For instance, an adjective haku 'black' can take the tense/aspect suffixes as in (14b) and bura 'aged male' can be the head of a NP, as in (15b).
khwa: taska: (14) a. Mira-ya Mira-GEN face.ABS very 'Mira's face is very black. '
haku. black
b. Mira-ya khwa: taska: haku-la. Mira-GEN face.ABS very black-PD 'Mira's face became very black.'
(15) a. wa: cha-mha bura-mha manu: khan-a. 3SG.ERG one-CL aged-ADN man.ABS see-PD 'He saw an old man.'
5 In subordinate clauses the noun copula kha: carmot occur and the inchoati.ve copulaju: replaces it.
(i) riim daktar ju; guli; I bale ... Rarn.ABs doctor coP because I when 'Because/When Ram is/was a doctor '
The word kha: can appear in a subordinate clause. However, it does not fimcti.on as a copula, but rather as an 'adjective-like verb', meaning 'to be true'
(ii) kha:-gu
khii be.true.ST...mvE-ADN story 'A true story'
A FUNCfiONAL ANALYSIS OF ADJECfiVES IN NEWAR
ale b. wa bura that old.man.ERG then
kanhekhunhu next. day
107
he ana EMPH there
hakanll: sima cha-ma pina bi-la. again tree.ABS one-CL plantCP give-PD 'On the next day, the old man then planted another tree.' As stated above, verbs and nouns never fail to fulfil all of their criteria. On the other hand, the criteria for adjectives are not necessary conditions but sufficient conditions, that is, all adjectives do not always satisfy all the criteria above. We may classify a word as an adjective if it fulfils a number of criteria. Most adjectival words can be negated by ma-ju:, although some may not be, such as machalapu 'not ashamed', garke 'odd', me: 'other', thwajya: 'like this', and so forth. Machalapu 'not ashamed' and garke 'odd' are exclusively used in the se-cwane construction and the negation is similarly formed. Me: 'other' and thwajya: 'like this' do not have a predicative usage and are limited to pre-nominal modification with an adnominal suffix. The weak quality of the criteria for adjectives suggests that the class of adjectives as a whole may not be a homogeneous category in Newar. The class itself is situated in the middle of a continuum between nouns and verbs. Giv6n (1984) proposes a time stability scale, as in (16) below, in which he argues that adjectives are located on a scale in which nouns are on the most time stable side and verbs on the rapid change side ( 1984: 55). (16)
Time Stability Scale NOUNS-----------------ADJECTIVES--------------------VERBS most time stable intermediate states rapid change
Stassen (1997) proposes the Adjective Principle, shown in (17). (17)
Predicate adjectives have no prototypical encoding strategy of their own. In their predicative encoding, adjectives will align themselves either with verbs (and, as a result, have a verbal strategy), or with nominals, or with locationals. In no language is it possible to have an encoding strategy which is used exclusively for predicative adjectives (Stassen 1997: 30).
Although his principle does not refute adjective class, he claims that adjectives do not have any unique predicative encodings. When we look
108
KAZUYUKI KIRYU
at adjectives in Newar, they conform to the principle in large part, except that many adjectives may appear in the predicate slot without a copula or tense/aspect sufftx, as in (14a) above. Formally, this is a case of copula omission, but the two are functionally different. In the following section, words that carry adjectival concepts in tenns of semantic classes and morphosyntactic properties are discussed. 4. CODING OF ADJECTIVAL CONCEPTS IN NEWAR Dixon (1982, 2004) proposes thirteen concepts that adjectives prototypically denote: DIMENSION, AGE, VALUE, COLOUR, PHYSICAL PROPERTY, HUMAN PROPENSITY, SPEED, DIFFICULTY, SIMILARITY, QUALIFICATION, QUANTIFICATION, POSITION and CARDINAL NUMBERS. In the following sections, each concept is addressed in relation not only to adjectives but also to related words that carry these concepts, so as to clarify the semantic range of the adjective class in Newar. 4.1. Dimension Words that denote DIMENSION in Newar are unique from a typological point ofview. Almost all the dimensional words are formed with one of two vector morphemes, ta:- and ci:-, and are followed by a numeral classifier by which the noun is classified. 6 To denote concepts of 'big' and 'small', for example, there are as many words as there are numbers of types or kinds referred to by Newar classifiers. These are listed in the following table (18). (18) LARGE
SMALL
DESCRIPTION
ci:dha: ci:dhi(ka): ci:ga: ci:ma ci:gwa: ci:pa:
inanimate things* animate beings* vehicles, containers plants round objects flat objects
SIZE ta:dha: ta:dhi(ka): ta:ga: ta:ma ta:gwa: ta:pa:
6 Some scholars regard ta: and ci: as adjectives by themselves (Hale and Shres1ha 2006). However, they can never function as a lexeme on their own For this reason, and as they do not confonn to the criteria for adjective class by themselves as proposed in this paper, I treat them as bound stems which must be joined with a classifier to form an adjective.
A FUNCfiONAL ANALYSIS OF ADJECfiVES IN NEWAR
ta:ku: ta:kha
109
ci:ku6· ci:khii
rooms houses, buildings
ci,phi ci,pll:
layered objects layered objects
ci:byii:
any object
ci:hiika: ci,pu
any object long objects
ci:jii:
houses, buildings, geographical objects
THICKNESS
ta,phi ta,pll: WIDTH
ta:byii: LENGTH
tii:hiika: ta,pu HEIGHT
ta:jii:
The adjectives with an asterisk by the description are not mmphologically analyzed as head plus classifier. Dhll: and dhilca: are not classifiers, but bound stems that relate to the size and height of animate beings respectively. The classifier for inanimate thing is -gu:, while that for animate beings is -mha. In reality, it is possible to have ta:gu: and ta,mha which convey a sense of 'many.' (19) a. nasii dha-yii-gu ta:gu:-kathll:-yii du. food.ABS say-PC-ADN many-kind-GEN exist.ST 'There are many kinds of that which is called food.'
b. cha-mha jaka ma-kha:-se ta,mha sasudya:-ta one-CL only NEG-COP-ADV many god-PL.ABS 'There isn't only one god, but there are many.'
du. exist.ST
All ta:- versions of the dimensional words may carry the meaning of 'many'. Dimensional adjectives can fill the complement position of the copula juye, and are negated only by ma-ju:.
(20) a. thumi-gu mha 60cm nise: 90cm taka tii:hiika: jui. this-GEN body 60cm from 90cm upto long cop.FD 'Their body is between 60cm and 90cm long.' b. ji-gu che uli ta:kha lSG-GEN house.ABS that.much big 'My house isn't that big.'
ma-ju:. NEG-COP.ST
110
KAZUYUKI KIRYU
The possibility that dimensional adjectives appear in the se cwane construction seems to be low. However, native speaker judgments about whether DIMENSIONAL adjectives enter into the se cwane construction could vary. For instance, both of my consultants accept ta:dhika: and ci:dhika: in the se cwane construction in a special context and reject other dimensional adjectives in the se cwane construction. The interpretation of the two dimensional adjectives in the se cwane construction is such that the things or person described are judged with respect to their size from a more subjective point of view. For example, one of my consultants explains that ta:dhi:-se cwii: expresses the speaker's idea that a child referred to shows adult-like behaviour. However, this use of a dimensional adjective in these cwane construction seems to be rarely used in non-elicited speech. In my database, I have found no instances of DIMENSIONAL adjectives in the se cwane construction. See § 5.5. for details.
4.2.Age In Newar, there are two types of words that denote AGE. The frrst are words that denote a person's age such as bura 'old (male)', buri 'old (female)', lyaemha 'young (male)' and lyase 'young (female). 7 These are the only adjectives to show gender distinctions. There are also two words denoting social status: thakiili 'higher in social status' and kwakali 'lower in social status'. These can be used as nouns, as in (21 ). (21) a. wa bura ale kanhekhunhu that old.man.ERG then next.day
he ana EMPH there
hiikanii: sima cha-ma pina bi-la. again 1ree.ABS one-CL plantCP give-PD 'On the next day, the old man then planted another tree.' b. wa lyaemhii: dha-la. that young.ERG say-PD 'The young man said.' While these words might be regarded as nouns rather than adjectives, they differ from nouns in that they take the adnominal suffix in noun modification, which is not possible for nouns. This fact alone, how7 Bura and buri are originally Nepali adjectives bul/Uz and butfzi. Since Newar lacks retroflex consonants, the Nepali voiced retroflex stops are pronounced as voiced alveolar flaps.
A
FUNCfiONAL ANALYSIS OF ADJECfiVES IN NEWAR
111
ever, does not serve as clear evidence for their status as adjectives. It may be argued that they are in origin nouns which may be used as adjective-like words. However, these words satisfy the criteria for adjectives: they can appear in the complement position of the copula juye, and can be negated by ma-ju:. It is therefore possible to claim that they are adjectives. Furthermore, when a noun modifies another nowt, it must take the genitive marker. These words, in nowt modification, do not take the genitive marker but an adnominal suffix. Since nowts cannot take an adnominal sufftx directly, these words are formally different from nouns and can be viewed as adjectives. The words pula: 'old' and nhu: 'new' indicate the age of things. The term nhu: is mainly used in noun modification, and not at all in predication in its bare form, so it should be thought of as a non-predicative adjective. It often occurs with an adnominal sufftx in the copula complement of khaye, as in (22c). (22) a. thwa saphu: taska: pula:. this bookABS very old 'This book is very old.' b. *thwa saphu: nhu:. this bookABS new 'This book is new.' (intended interpretation)
c. thwa saphu: nhu:-gu kha:. this bookABS new-ADN cop.ST 'This book is new.' (lit. This book is a new one.) On the other hand, pula: may stand in the complement position ofju:. 4.3. Colour
Native Newar colour words have the morphological ending and are listed in (23 ). (23)
hyiiil
tuyu
wacu siyu
'red' 'white' 'blue' 'brown'
wiiil
hiiku mhasu bhuyu
-u or -u,
'green' 'black' 'yellow' 'faded colour'
These words may be regarded as adjectives on the following grounds: they can appear in the complement of the copula verb ju:, not in that
112
KAZUYUKI KIRYU
of kha:; they can enter into the se cwane construction; and they are negated by ma-ju:. It is interesting to note that these colour adjectives cannot be in the complement slot of the copula juye when it is in the future/past form ju-il-la, which denotes inchoative sense. However, they may instead take a verbal inflection to express the inchoative sense.
(24) a. hari-ya khwa: hyau-la. Hari-GEN face.ABS red-PD 'Hari' s face turned red.' b.
*har~ya
Hari-GEN
khwa: face.ABS
hyau red
ju-la cop-PD
The inchoative sense can also be expressed in the se cwane construction. (25)
hari-ya khwa: hyau-se Hari-GEN face.ABS red-ADV 'Hari 's face turned red.'
cwan-a. stay-PD
The subtle difference between the inflected form and the se cwane construction is discussed in §5.5 below. 4.4. Value
The following list shows some VALUE adjectives in Newar.
(26)
yaipu cwatu cipa paechi thikae nhyaipu mhaipu
'lovely gyanapu 'perfect' garke 'impure' nipa 'matched' thik 'expensive' 'interesting, interested' 'boring, bored'
'scary' 'odd, strange' 'pure' 'appropriate'
Motphologically speaking, four of the above adjectives are derived from verbs with the adjectival suffJX -pu. Yaipu is derived from yaye 'to be fond of', gyanapu from gyaye 'to get scared', nhyaipu from nhyaye 'to be willing to/like' and mhaipu from mhaye 'to dislike'. Adjectives with the -pu ending can take verbal forms to convey an inchoative sense. Like colour adjectives, they can also enter into the se cwane construction. The other value adjectives in the list, except for garke, cannot be used with this construction, but they can appear in
A FUNCfiONAL ANALYSIS OF ADJECfiVES IN NEWAR
113
the complement position of the copula ju: and with its inchoative form ju-i/-la. Garke is an unusual adjective. It is used only in these cwane construction and never appears in the complement position of the copula. Some verbs denote a value sense in their stative forms. These are shown in (27) below. (27)
ba:la: 'beautiful'
bhi: 'good'
dll:
'cheap'
These forms are described as verbs because they fulfil the criteria for inclusion in the verb class: they take tense/aspect inflections and are negated by ma-. However, they can be negated by ma-ju: as well, which may be because they are semantically more like adjectives. Since the above forms are verbs, they do not enter into the se cwane construction and cannot be the complement of the copula ju:. Interestingly, however, ba:la: can. Ba:la: fulfils all the adjective class criteria: it can enter into the se cwane construction, and can appear in the complement of stative ju: and inchoative ju-i/-la. In negation, it can take ma- or ma-ju:. The negative affix ma- is placed within the word, as in M:-ma-la:. These two negative forms indicate a semantic difference: M:mala: can be translated as 'bad', while the simple negation is expressed by ba :Ia: ma-ju:. Ba:la: allows three forms in predicate complement position, ba:la:, ba:la: ju: and ba:la:-se cwll:. The difference between the three is quite subtle, with M :Ia: simply used to describe the property of something, while ba:la: ju: is more often used in the context of compartson.
'Iapan phish' Lapan ftsh
dhaka: na: taya-ta:-pi: thupi: COMP name put.CP-put.ST-ADN.PL these
nya swa-e bale ati ba:la: ju:. ftsh.ABS lookat-FC when such beautiful.ST cop.ST 'It is true that the ftsh called 'Lapan ftsh' are extremely beautiful when looked at.' In the above example, the 'Lapan ftsh' is explained in a context in which it is beautiful but poisonous. Since these two characteristics of the ftsh are compared, ju: is used.
114
KAZUYUKI KIRYU
Ba:lii:-se cwii: is used when the speaker has some personal evidence to indicate that the thing in question is 'beautiful'. This perception may be based on recognition or experience.
4.5. Physical Property There are a large class of PHYSICAL PROPERTY adjectives in Newar. In tenns of morphological properties, we recognise three sub-types. Many of the words that express PHYSICAL PROPERTY end with -ii., -ku, -cu, -tu, -nu, -pu, -mu, -yu, -lu or -su. These typically enter into the se cwane construction.
(28)
khwiiii. ciku yecu khwiitu tyiinu khiiyu hwiilu ciisu
'cold' 'cold' 'clean' 'dense' 'tired' 'bitter' 'baggy' 'itchy'
yiiii. tiiku picu jhyiitu lumu niiyu piilu chwiisu
'soft' 'strong (of taste)' 'smooth' 'heavy' 'lukewarm' 'soft' 'pungent' 'loose'
These adjectives may take verbal tense/aspect suffJXes to denote an inchoative sense, but cannot take the inchoative copula ju-i/-la. One exception is lwaManiipu 'attractive', which is derived from a verb, lwaMane 'to be attracted to', in combination with the suffJX -pu. This form is used only in the se cwane construction and may not take the verbal suffJX. PHYSICAL PROPERTIES concepts are also expressed by verbal stativeforms.
(29)
chwa: caka: sii: chii: ba:lii: kwa:
'liberal' 'open-spaced'
'tasty' 'hard' 'strong' 'hot (of liquid)'
chi: kaekii.: lila: syii: tiimuwa:
'comfortable' 'shrunk'
'free' 'painful' 'hot'
On account of being verbs, these terms cannot be used with the se cwane construction or in the complement slot of ju:, but may be negated by ma-ju:. The fmal set of PHYSICAL PROPERTY adjectives have no common morphological characteristics. These adjectives do not enter into these
A
FUNCfiONAL ANALYSIS OF ADJECfiVES IN NEWAR
115
cwane construction and may not take the verbal suffix either, but can nevertheless be regarded as adjectives, for they appear in the complement slot of the copula juye, are negated by ma-ju: and can modify nouns taking adnommal suffJXes. (30)
jabar ta:mi basi
'durable' 'rich' 'rotten'
kara kaci khusi
'strong (of medicine)' 'raw' 'happy'
Except ta:mi and kaci, all of the above terms are borrowed from Nepali. Ta.mi is morphologically analyzable into ta:- and -mi. Ta:- is a bowtd adjectival stem which also constitutes dimensional adjectives, meaning largeness, and -mi is a bowtd noun stem meaning 'man'. 8 4.6. Human propensity
Newar appears not to have a variety of HUMAN PROPENSITY adjectives, and most that exist are borrowings from Indo-Aryan languages such as Nepali and Sanskrit. Other HUMAN PROPENSITY concepts are expressed by a property adjective or a verbal idiom with an adnominal suffJX. (31) a. Indo-Aryan borrowings ca:k 'clever' 'lazy' alsi dukhi 'unhappy' b. Newar adjectives 'talkative' cwaltl: mach a: 'ashamed' hulya: 'of mobster'
irsa dayalu sukhi
'jealous' 'kind' 'happy'
kwamali cwalu asya:
'humble' 'of big mouth' 'mischievous'
c. Other expressions yau-se cwii-mha light- ADV stay. sr -ADN mati bi:-mha heart give.ST-ADN
'kind, generous' 'generous'
The adjectives in (32a-b) appear in the complement of juye, and are negated by ma-ju:, but do not enter into the se cwane construction. They can modify nouns with an adnominal suffJX. It is often the case,
8 Although ci: - is the opposite of ta: -, the form ci: mi 'poor' is not possible. But one consultant offered the idiom ta: mi-ci: mi 'rich and poor'
116
KAZUYUKI KIRYU
however, that they appear with the adnominal sufftxes, -mha or -pi:, in the nowt complement construction to describe a trait
(32)
wa manu:
tasaka: ca:k-mha
the man.ABS very 'The man is very clever.'
kha:.
clever-ADN COP
4.7. Speed In Newar, SPEED is not expressed through adjectives. Rather, the concepts in Newar that correspond to the English adjectives 'fast' and 'slow' are expressed through adverbs or verbs. (33) a. bwaewan-i:-gu gadi khawa-e, buluhu: wan-i:-gu go.fast-FD-ADN car.ABS left.side-LOC slowly go-FD-ADN
gadi
jawa-e
wan-e
ma:.
car.ABS right.side-LOC go-INF need.ST 'Fast cars must go on the left side and slow cars must go on the right side. '
b. Porshe dhaka: dhai-gu gadi taska: Porsche COMP say.FD-ADN car.ABS very 'A Porsche is very fast.' (lit. goes very fast)
bwaewa:. go.fast.ST
In Newar, the concept of SPEED will more often than not be expressed by adverbs of manner and be accompanied by a verb that characterises the event in which the speed is mentioned, such as a verb of motion. In passing, Newar has another verb that conveys SPEED, particularly in the context where someone's speed of work is fast The verb is cwaye and it is often used in the stative form. (34)
lha: cwa:-mha manu: hand be.fastST-ADN man 'someone whose work is fast' (lit someone whose hand is fast)
4.8. Difficulty There are two common adjectives in this group, a:pu 'easy' and thaku 'difficult'. Although these adjectives do not enter into these cwane construction, they can appear in the complement position of the copula ju:, and may also take the verbal sufftx to convey inchoative sense.
A FUNCfiONAL ANALYSIS OF ADJECfiVES IN NEWAR (35)
a:pu
'easy'
thiiku
117
'difficult'
These adjectives can behave like verbs, taking the verbal suffiX to express inchoative sense. Since they are not adjectives, however, they may not be negated by ma-, but by ma-ju:. (36)
bura juya: ;ya yaye thiiku-la. old become.CP work.ABS do.INF di:fficult-PD 'Since I am old, it is difficult for me to work' 4.9. Similarity
There are a few words that pertain to the concept of SIMILARITY. (37)
jwa:la: milae ju: thwajya:
'similar' 'similar' 'like this
pa: me aja:
'different' 'other' 'like that'
The frrst two are stative forms of verbs, and they stand alone as predicates. Although they convey adjective-like concepts, they are not negated by ma-ju: and must be negated by ma- as injwa:-ma-la: and mapa:. Milae ju: is an intransitive compound verb whose origin is the Nepali verb, milnu. Another word meaning 'different' is me:, but this is always followed by an adnominal suffiX, as in me:-gu, conveying the sense 'another/other', which cannot be used predicatively. Thwajya: and aja: do not appear in the copula complement slot. To have such a meaning in the predicate position, these forms must be nominalised by -gu and appear in the complement of the noun copula, kha:. The words may be regarded as adjectives as they satisfy the ftfth criterion: they can be reduplicated to express plurality. 4.10. Qualification Concepts of QUALIFICATION are not typically expressed by adjectives in Newar. I have found only one such adjective,paechi 'correct'. This appears only in the complement slot of the copula and in noun modification with an adnominal suffiX. (38) a. paechi ma-ju-la. correct NEG-COP-PD '(No, your answer) is not correct'
118
KAZUYUKI KIRYU
b. piiechi-gu lisa: correct-ADN answer 'a correct answer' The concept of 'true' is expressed by the copula verb khaye and it usually appears in the stative form. Being a verb, it can be negated by ma-, but the form is irregular, makhu.
(39) a. kha:-gu kha: true.ST-ADN story 'a true story'
b. ma-khu-gu
kha: NEG-true.ST-ADN story 'a lie/an untrue story' 4.11. Quantification
In Newar, words that convey the sense of QUANTIFICATION are not adjectives, even though they behave like adjectives in some ways by appearing in the complement of the copula juye. However, they are not adjectives because they appear after the head where an adjective may not, and do not appear before any adnominal suffiXes in noun modification. Terms of quantification are shown in (41) below. (40)
yakkwa dakwa yiika: kam
'many' 'all' 'only one' 'little'
apwa: iipii: mha bhati:cii
'more' 'many' 'little, less' 'little'
In noun modification, yakkwa, apwa: and dakwa may appear either before or after the head noun. Other quantifiers appear only after the head noun. (41) a. yakwalapwa:/dakwa manu:-ta many/more/all man-PL.ABS 'Many/MorelAll men came.'
b. manu:-ta
yakwalapwa:/dakwa man-PL.ABS many 'Many men came.'
wa-la. come-PD wa-la. come-PD
A
FUNCfiONAL ANALYSIS OF ADJECfiVES IN NEWAR
119
c. dhyebii mha/bhati:cii/kam jaka du. money.ABS little only exist.ST 'I have only a little money.'
These quantifiers may appear in the inchoative form of the copula.
yakwalapwa:liipii: ju-la. (42) a. bidyiirthi-pi: student-PL.ABS many/more/many become-PD 'We now have many/more/many students.' b. nhiipii swaya mhalbhati:cii/kam before than little 'It became less than before.'
ju-la. become-PD
Other than quantifiers, the sense of quantification can be conveyed by a verb. (43)
gii:
'enough'
Gii: is the stative form of the verb gaye 'to be/become sufficient'. The negated form is ma-gii:, meaning 'not sufficient'. 4.12. Position While there are four expressions conveying the concept of POSITION, these are not adjectives. (44)
tiipii: jawa
'far' 'right side'
sa:ti: khawa
'near' 'left side'
The concepts of 'far' and 'near' position are expressed by the stative form of verbs, tiipiiye and sa.1iye respectively. Their past disjunct forms are tiipiita and sa.1ita, and can be negated by ma-, placed before the fmal syllable, as in tii-ma-pii: and sa:ma-ti:. Despite their verbal status, they can fill the complement slot of ju: and ma-ju:. In this sense, they are less verb-like and may conceptually be regarded as adjectives. The directions 'left' and 'right' are nouns, and in expressions such as 'right lane', the relative construction is employed, as in (46) below: (45)
jawa-e lii:-gu right-LOC be.located.ST-ADN 'the right lane'
Ill road
120
KAZUYUKI KIRYU 4.13. Cardinal numbers
CARDINAL NUMBERS in Newar may be considered to constitute a distinct class, and cannot modify nouns without classifiers. The entire numeral and classifier phrase functions like a quantifier and can appear either before or after the head noun. The sense of 'first' and 'last' are expressed by nhtipa: and lipa. These are typically used as adverbs, but can modify nouns by taking an adnominal suffix. 4.14. Summary As we have seen, the majority of the thirteen adjectival concepts are encoded as adjectives. Only SPEED, POSITION, SIMILARITY and QUANTIFICATION are not adjectives, with the first three concepts encoded as verbs and the last one as a quantifier. CARDINAL NUMBERS are not expressed by the adjective class in Newar, and constitute a different word class. Concepts that are coded as adjectives are only DIMENSION, AGE COLOUR, HUMAN PROPENSITY and DIFFICULTY, while VALUE, PHYSICAL PROPENSITY and QUALIFICATION include quite a few verbs as well as adjectives. Morphologically speaking, adjectives that end with -u or -ii. may take a verbal suffix. Adjectives of COLOUR, VALUE, PHYSICAL PROPERTY and HUMAN PROPENSITY that end with -u or -ii. can enter in the se cwane construction. In the following section, I discuss functional characteristics of adjectives.
5. SUB-CLASSES OF ADJECTIVES In this section, based on the criteria proposed in Section 2 and the semantic concepts of adjectives described in the previous section, three sub-classes of adjectives are discussed in detail. It has been proposed that Newar has three adjective classes: verb-like adjectives, non-verblike adjectives and non-predicative adjectives. In addition to the three adjective classes, adjectival verbs, which express adjectival concepts in the stative forms and formally behave like adjectives in some respects, have also been identified. In the following subsections, these classes are addressed in turn.
A FUNCfiONAL ANALYSIS OF ADJECfiVES IN NEWAR
121
5.1. Verb-like adjectives
Verb-like adjectives are words which may take a verbal inflectional suffix and which behave formally like verbs. The adjectives in this class invariably take Class III verbal suffixes. Verbs are classified into five classes in terms of their inflectional suffixes. (46)
Class 1: the stem-final syllable ends with -n; Class II: the stem-fmal syllable ends with a vowel, and the past disjunct suffix is -ta; Class ill: the stem-final syllable ends with a vowel, and the past disjunct suffix is -Ia; Class IV: the stem-fmal syllable ends with -1; Class V: the stem-fmal syllable ends with -k, -por-t.
The verbs in all classes have a stative form, which is distinct for length of the stem-fmal vowel, conveying a habitual or stative meaning. Although they take Class ill inflections, verb-like adjectives end with -u or -U, and never end with a lengthy vowel. Furthermore, they cannot be negated with the negation affix ma-. Even if they take a verbal sufftx, it is therefore better to treat them separately from verbs, as otherwise another verb class for these verb-like adjectives that is less consistent with the five verb classes would need to be proposed and established. The semantic types that adjectival verbs cover are COLOUR, VALUE, HUMAN PROPENSITY and PHYSICAL PROPERTY. (47) COLOUR tuyu 'white',
hyau 'red', mhasu 'yellow', etc.
PHYSICAL PROPERTY
sicu 'cool',
caku 'sweet',
khwiitu 'tightened', etc.
VALUE
nhyiiipu 'amusing', mhaipu 'boring', etc. DIFFICULTY
a:pu 'easy',
thiiku 'difficult'
Syntactically, verb-like adjectives can be the stative copula complement ofju:, but do not appear as the complement ofjula orjui, which
122
KAZUYUKI KIRYU
expresses inchoation. The inchoative sense is expressed when they take the verbal sufftx.
(48) a. wa-ya khwa: 3SG-GEN face.ABS 'His face is red.'
hyiiil
ju:.
red
cop.ST
khwa: hyiiil-la. b. wa-ya 3SG-GEN face.ABS red-PD 'His face became red.' Difficulty adjectives also take Class III verbal inflections, but they are different from other verb-like adjectives in that they do not enter into the se cwane construction but can be the complement of the copula in the inchoative forms. 9
( 49) a. nhapa swaya: a.pu-la. before than easy-PD 'It became easier than before. '
b. nhapa
swaya: a.pu ju-la.
before than easy become-PD 'It became easier than before. ' 5.2. Non-verb-like adjectives Non-verb-like adjectives do not take any verbal sufftxes nor do they enter into the se cwane construction. This class of adjective is larger in size than the verb-like adjective class, and loan words also belong to this class. The adjectives of DIMENSION, HUMAN PROPENSITY and AGE, one QUALIFICfiON adjective, and some adjectives of VALUE and PHYSICAL PROPERTY belong to this class. As discussed in the previous sections, AGE adjectives are closer to nouns because they can be the head noun and take case markers directly. On the continuum, these adjectives are closer to nouns. DIMENSION adjectives seem to have an ambivalent status between verb-like adjectives and non-verb-like adjectives. They do not take verbal inflectional sufftxes and they cannot occur in the se cwane con-
9 The difference between (50a) and (50b) is very subtle. Although (50b) is acceptable to some speakers, (50a) is used more frequently in texts.
A FUNCfiONAL ANALYSIS OF ADJECfiVES IN NEWAR
123
struction, except ta:dhika: and ci:dhika:. According to one consultant, the se cwane form expresses a more subjective judgment (cf. §5.5). 5.3. Non-predicative adjectives The adjectives which belong to this class are those of AGE, nhu: 'new', and SIMILARITY, me: 'another' and thwajwa: 'such'. These adjectives are limited to noun modifications, yet they cannot modify nouns directly but require an adnominal suffix.
(50) a. nhu:-gu wasa: mal-a dha-i-gu? new-ADN clothes.ABS need-PD say-FD-ADN 'Do you say that you want some new clothes?' b. wa me:-gu khii ka. that other-ADN story.ABS PART 'It is another story, you know.' c. thwajwa:-mha ju-yti:, such-ADN be-CP 'Since he was such a man, ... ' Dixon (2004) states that adjectives may have restricted functional possibilities. As presented in Section 2, adjectives have two canonical functions: functioning either as an intransitive predicate or as a copula complement (2a), and functioning as a modifier to the head (2b). In some languages, however, adjectives have only one of the two functions. Non-predicative adjectives in Newar lack the function of (2a). In noun modification, they must always take adnominal suffixes and cannot directly modify the head. Adnominal suffixes are also required when verbs modify nouns. On the other hand, when a noun modifies another noun, it must take the genitive marker. An adnominal sufftx can follow a modifying noun after the genitive marker but it cannot follow immeadiately after the modifying noun.
(51) a. rtim-yti tasbir Ram-GEN photo 'a photo ofRam' b. rtim-yti-gu Ram-GEN-ADN 'Ram's photo'
tasbir photo
124
KAZUYUKI KIRYU
The difference between the two examples in (52) above will be clear from the translations. The adnominal version puts more emphasis on possession or property interpretation (see Kolver 1977 for details). In Newar, when a noun is modified by another word, it cannot be directly modified by it When a noun precedes another noun, it must take the genitive marker, which in itself fimctions as a connector in modification. Verbs and adjectives, on the other hand, never take the genitive marker but must take an adnominal sufftx. 10
5.4. AdJectival verbs Some verbs in the stative form function like adjectives in that they can be negated by the negative copula ma-.fu: and can fill in the complement slot of the copula .fu:. Since they may be negated by the negative ma- and take tense/aspect sufftxes as regular verbs, they may be considered to be a sub-class of verb and viewed as adjectival verbs. Since they are essentially verbs, they do not take the inchoative copula .fuil.fula to denote a change of state. Their meaning is derived from their inflected forms. (52) a. mu: dii:. price.ABS be.cheap.ST 'The price is cheap.'
b. mu:
dan-a.
price.ABS be.cheap-PD 'The price became cheap.' Adjectival verbs may be negated either by the negative afftx ma- or by ma-.fu:, but how do these negative forms differ? Some adjectival verbs do not show a large difference in meaning in either negation, while others are entirely different. In the latter case, negation by the negative affix ma- is lexical negation, that is, it negates the concept denoted by the adjective itself. Negation by ma-.fu: is a negation of proposition, paraphrased as 'it is not the case that such and such.' Lexical negation negates the concept of the adjective itself, which in turn results in a total negation or at least in a denotation of the opposite concept For instance, ma-sii: literally means 'not tasty', but this is
10 The necessity of adnominal suffixes in noWl modification seems to be a recent development. In Classical Newar, verbs and adjectives could modifY noWlS directly.
A FUNCfiONAL ANALYSIS OF ADJECfiVES IN NEWAR
125
not a simple negation, and rather means 'tasteless'. The negation by the copula ma-ju:, on the other hand, is the negation of the proposition, hence sii: ma-ju: means 'it is not the case that something is tasty.' (53) a. sii:. tasty.ST 'It's tasty.'
b. ma-sii:. NEG-tasty. ST 'It's tasteless/yucky.' c. sii: ma-ju:. tasty. ST NEG-COP 'It isn't tasty.' More examples of adjectival verbs that express the opposite concept in the ma- negation are provided below. Those without opposite concepts are cases of total negation. (54)
INFINITIVE STATIVE
QUANTITY giiye PHYSICAL
ba:laye
PROPERTY bhine
liliiye ba:liiye
OPPOSITE CONCEPT
gii: 'sufficient'
magii: 'insufficient'
ba:za: 'good/beautiful'
ba:mala: 'bad/ugly'
bhi: 'good (of quality)' lilii:
'free'
mabhi: 'bad (of quality) limalii: 'busy'
ba:lii: 'strong'
ba.malii: 'weak'
Only adjectival verbs exhibit the property of denoting the opposite concept in negation. Negation of regular verbs is simply propositional negation.
5.5. These cwane construction The se cwane construction pertains to adjectival concepts, and many verb-like adjectives may enter into the construction. Verbs do not appear in the construction, except one adjective-like verb, as discussed
126
KAZUYUKI KIRYU
in Section 4. The primary function of the construction is to exhibit the speaker's subjective judgement on a situation where an adjectival concept is expressed. On the other hand, the adjectives in bare form express an objective description. The se cwane construction can be motphologically analyzed as an adverbial suffix -se and the existential verb cwane, the combination of which means 'to be in the state of'. The construction has two patterns, -se cw/1: for stative and -se cwana/cwani: for past inchoative/future inchoative respectively. Verb-like adjectives are a class of adjectives that appear in these cwane construction. These adjectives end with the motphemes -u or -U, and, except the DIFFICULTY adjectives, can enter into these cwane construction. Verb-like adjectives may therefore appear in three patterns in predicative sentences, as follows: (55) a. aelii: task/1: piilu. liquor very pungent b. aelii: task/1: piilu ju:. liquor very pungent cop.ST c. aelii: task/1: piilu-se cw/1:. liquor very pungent-ADV stay.ST 'The liquor is very strong.' All the examples in (56) have the same meaning in terms of propositional value, but they remain functionally different. It was sometimes difficult for my consultants to clearly describe the differences without being given an appropriate context in which one is more natural than the other. In the frrst pattern, where the adjective constitutes a predicate by itself, it describes the general nature of aelii: (Newar traditional liquor). When the adjective is in the complement position of the copula ju:, there is a sense of an emphatic assertion or an implicit comparison to other kinds. The se cw/1: expression implies a subjective judgment on the part of the speaker. It is, then, quite natural to use this form in a context where the speaker is actually tasting liquor or talking about liquor in general, based on his or her own experience.
A FUNCfiONAL ANALYSIS OF ADJECfiVES IN NEWAR
127
The subjective sense of the se cwane construction is also observed in an inchoative sense. As discussed above, there are two patterns for the inchoative sense with verb-like adjectives.
khwiiU-i. (56) a. ciya tea.ABS cold-FD 'The tea will get cold.' b. ciya khwiiU-se cwan-i:. tea.ABS cold-ADV stay-FD 'The tea is going to get cold.' Functioning as a verb with a verbal suffiX, khwau- i describes the change of state from an objective point of view. For example, (57a) is most appropriately uttered in a science class, taJking about the change in temperature of the tea when it is put in a fridge, with the sense of, ''When you put the tea in the fridge, it will get cold." On the other hand, (57b) is most appropriate in a context where a host urges a guest to drink the tea before it gets cold. In this situation, the focus of attention is not on the physical change of the temperature itself, but on the taste of the tea when it is cold. This is more subjective in that the taste is judged by one's own sense. The difference between the verb-like pattern and the se cwane construction may be based on aspect. Since the suffix -se is imperfective in nature, se cwana in the past form denotes a temporal or transient state that the speaker feels. On the other hand, the verb-like pattern shows a complete realisation of the state denoted by the adjective, and it is hence perfective in aspect This is illustrated by (58) below.
(57) a. ji-ta: tyanu-se lSG-DAT tired-ADV 'I'm feeling tired.'
cwan-a. stay-PD
b. ji-ta: tyanu-la. lSG-DAT tired-PD 'I'm tired (lit I got tired.).'
Tyanuse cwana conveys what the speaker is feeling at the moment, indicating that he is not completely tired, but that he can now sense the intensity of fatigue. Tyanula, on the other hand, conveys the complete realisation of the state, implying that the speaker is too tired to work
128
KAZUYUKI KIRYU
any more. The semantic difference discussed above holds for all verblike adjectives. As discussed in §4.1, two of the DIMENSION adjectives, ta:-dhi(ka): and ci:-dhi(ka): may enter into these cwane construction. (58)
wa ta:-dhi:-se cwCi:-mha mhasyu: la? that big-CL-ADV stay.ST-ADN know.ST Q 'Do you know the man who looks tall?'
The adjectives in the construction may be used to indicate the person in question with reference to his height In this situation, the height of the person is judged relatively and so it is not necessarily the case that he actually is tall. This function is based on the subjective judgement ofthe speaker. As discussed in Section 4, a few adjectives cannot appear in the complement position of the copula juye, and appear only in the se cwane construction. These are machalapu 'shy', garke 'strange' and lwa:wanapu 'attractive' The senses they express are subjective rather than objective, conforming to the sense conveyed by the se cwane construction. 6. CONCLUSION Since some verbs can convey concepts that are often expressed by adjectives, and some verb-like adjectives behave motphosyntactically like verbs, it is sometimes thought that Newar lacks an independent adjectival class. A detailed investigation into motphosyntactic properties that exclude nouns and verbs, however, refutes this position. In this article, I have discussed criteria for adjective classes, through which three types of adjective class based on semantic concepts and motphosyntax can be identified.
A FUNCfiONAL ANALYSIS OF ADJECfiVES IN NEWAR
129
ABBREVIATIONS
ABS ADV ADN ADV CL
COMP COP CP
DAT EMPH ERG FC FD
absolutive adverbial sufflx adnom inal suffix adverbial sufflx classifier complementiser copula conjunctive participle dative emphasis ergative future conjunct future disjunct
GEN HON INF
LOC NEG PART PC PD PL Q SG ST
genitive honorific infinitive locative negation particle past conjunct past disjunct plmal question marker singular stative
BIBLIOGRAPHY Dixon, R.M.W. 1982. Where have all the atJjectives gone?: and other essays in semantics and syntax. (Jarrua Linguarum. Series Maior, 107.) Berlin: Mouton de Gruyter. - - . 2004. Adjective classes in typological perspective. In R W.W. Dixon and A.Y. Aikhenvald, eds.,AtJjeclive Classes: A Cross-linguistic Typology. (Explorations in Linguistic Typology.) Oxford Oxford University Press, 1-49. Genetti, C. 1994. A Descriptive and Historical Account of the Dolakha Newari Dialect. (Monumen1a Serindica, 24.) Tokyo: Institute for the Study of Languages and Cultures of Asia and Africa. - - . 2006. Preliminary notes on the Tauthali dialect of Newar, paper presented at 11th Himalayan Languages Symposium. - . 2007. A Grammar of Dolakha Newar. (Mouton Grammar Library, 40.) Berlin: Mouton de Gruyter. Givon, T. 1984. Syntax: A Functional-Typological Introduction, Vol. 1. Amsterdam: John Benjarni.ns. Hale, A 1980. Person markers: finite conjunct and di.sjmtct forms in Newari. In R. Trail, ed, Papers in Southeast Asian Linguistics. (Pacific Linguistics Series A, no.53.) Canberra: Australian National University. Hale, A and K.P. Shrestha. 2006. Newar (Nepal Bhiisa). Miinchen: Lincom Europa. Kolver, U. 1977. Nominalization and Lexicalization in Modern Newari. (Arbeiten des Kolner Universalien-Projekts, 30.) Koln: Institut fiir Sprachwissenschaft. Shakya, D. R. 1998/1999. In naming a language. Newiih Vijnana. The Journal afNewar Studies)2: 38-42. Stassen, L. 1997./ntransitive Predication. Oxford Oxford University Press.
PART FOUR: MORPHOLOGY AND SYNTAX
THE ROLE OF ANIMACY IN THE VERBAL MORPHOLOGY OF DONGWANG TIBETAN1 ELLEN BARTEE
1. INTRODUCTION 'Conjunct/disjunct' (c/d) are tenns ftrst coined by Austin Hale in 1980 based on his observations of a cross-referencing-like system in the Newar language of Nepal. This pattern has also been reported in the Bodie Branch of Tibeto-Bunnan, in Akha2 of the Lolo-Bunnese Branch of Tibeto-Bunnan (Thurgood 1986) and Minhe Mangghuer, 3 a Mongolic language spoken in Western China, north of Tibet (Slater 1998). In recent years, there have also been reports of similar systems in other parts of the world. Dongwang, a Southern Khams dialect of Tibetan, also has this pattern, but has important characteristics not reported in any other language, namely, the existential verbs and auxiliaries exhibit an animacy split. This paper constitutes the ftrst report in which an animacy split and conjunct/disjunct patternings co-occur in the verbal motphology of a language. I will ftrst describe the general features of the c/d system drawing on data from Standard Tibetan.4 I will then present an overview of the c/d system in Dongwang Tibetan, detail the animacy distinction observed in the existential verbs and auxiliaries, and suggest that this is an areal feature which developed in Dongwang, possibly as a result of contact with speakers of Naxi, a Lolo-Bunnese language.
1 This research was supported mainly by a Fulbright grant during 2002 and 2003. I am grateful to many people for their comments and help on earlier versions of this paper, in particular, Carol Genetti, Sandy Thompson, Bernard Comrie, Timothy Curnow and Alexis Michaud. I am also greatly indebted to Tashi Tsering for introducing me to his family in Dongwang and to the many Dongwang friends that I worked with, es~cially Danzen Chumpi, Tashi Nonbu and Yishi Droma. 2 Akha is known as Haniyu (Pit .IE i~) in China. 3 Mangghue is known as Tuzuyu (±~~) in China. 4 Tournadre and Sangda Dorje (2003) use 'Standard Tibetan' to refer to the lingua franca for most of the Tibetan Autonomous Region
134
ELLEN BARTEE
1.1. C/d systems cross-linguistically Before Austin Hale (1980) described the c/d system ofNewar(i), little was known about this type of system. In the last twenty years there has been much research describing similar systems in Tibeto-Burman, the language family in which this pattern is most prevalent. 6 Several linguists undertook further description of Newar, most notably Genetti (1986, 1994) and Hargreaves (1990, 1991, 2005). Most of the other Tibeto-Burman languages which have been reported to have c/d marking are Tibetan languages, in which all major dialect groups are represented. For example: Lhasa (DeLancey 1985, 1986, 1997, 2001 and Tournadre 2001), Lende (Huber 2000), Lhomi (Vesalainen and Vesalainen 1980), Shigatse and Themchen (Haller 2000), Ladakhi and Purik (Zeisler 2004), Sherpa (ShOttelndryer 1980); rGyalthang (Hongladarom 1996), Dege (Hasler 1999), Dzongkha7 (van Driem 1998), Dongwang (Bartee 2007) and Amdo (Sun 1993). Slater (1998) reports a similar system for Minhe Mangghuer, a Mongolic language which has been heavily influenced by Tibetan. Other Tibeto-Burman languages with c/d patterns include Akha (Thurgood 1986) and Sankhong8 (Matisoff 1993). An earlier understanding that c/d systems were limited to a small cluster of Himalayan languages was dismissed with the description of similar systems in far-flung languages such as Awa Pit (Curnow 2003) and Guambiano (Vasquez de Ruiz 1988) in Latin America and Karol Rawa (Toland and Toland 1991) in Papua New Guinea.
1.2. Characteristics ofc/d systems The term 'conjunct' refers to the canonical situation in which a particular form is used in frrst-person declarative clauses, in second-person interrogative clauses, or in reported speech in which the agentive par5 As many would agree, the tenns 'conjunct/disjunct' are ambiguous and inadequate to capture both the details of any one language or the diversity found cross-linguistically. I agree that there has long been a need for 'more self-evident labels' (Sun 1993: 955-6), but it is not my goal to introduce new labels here. In 1he absence of any terminology that is cmren1ly in widespread use, it is my intention that 1he reader be able to understand the overall system referred to in this paper wi1hout becoming sidetracked by the unfortunate lack of adequate terminology. 6 Most of the sources listed here were taken from Timothy Ctmtow's working bibliography (n.d.), a partial result of his three-year study of a typology of 'first/non-first' systems throughout the world 7 I consider Dzongkha, spoken in Bhutan, to be a Southern Tibetan language. 8 Sankhong is spoken in Yunnan province.
ANIMACY IN DONGWANG TIBET AN
135
ticipant in the complement clause is coreferential with the agentive participant in the matrix clause. The term 'disjunct' refers to a particular form that occurs in all other clauses. Hargreaves outlines the necessary conditions in which the conjunct forms appear: 1) The clause is fmite, and 2) the event being described is interpreted as involving an intentional action by the actor and [/od 3) the speech act is either a) declarative/first person, .QL b) interrogative/second person, .QL c) reported speech when the matrix clause subject and complement clause subject are coreferential (Hargreaves 2005: 5, 6)
The logic of such a system, at least for Tibetan, seems to revolve around the concept that "only the perpetrator of an act can possibly have direct knowledge of the act ofvolition which led to it" (DeLancey 2001: 372). Thus, conjunct forms occur in utterances in which the speaker has personal knowledge, intention, and/or ability to control an action. While this distinction does not occur in Classical Literary Tibetan (fournadre and Dorje 2003: 93), 10 it plays an important role in modern dialects. In some ways, the c/d system looks like a system of cross-referencing, but strictly speaking it is not Speakers' ability to manipulate the system for pragmatic purposes and restrictions on the co-occurrence of conjlDlct forms with certain verbs prevent it from being so. It appears that most modern dialects of Tibetan express the c/d distinction in copula/existential verbs or in auxiliary verbs of active clauses. Many of the auxiliaries in various Tibetan dialects have grammaticalised from verbs, in particular, from copular and existential verbs.
9 Hargreaves' definition, as pointed out by Cmnow (personal communication), applies mostly to conjunct auxiliaries. In this paper I extend Hargreaves' definition to include clauses with first person S/A arguments that express copular, existential and possessive relationships. 10 There are many terms for varieties of written Tibetan. For the purposes of 1his paper, I will use 1he term 'Written Tibetan' as a catchall for what is sometimes called 'Uterary Tibetan', 'Written Tibetan', 'Classical Tibetan' or 'Qassical Uterary Tibetan'.
136
ELLEN BARTEE
The following data which illustrate the basic patterns of c/d in Standard Tibetan, 11 are taken from Toumadre and Dorje (2003) and Hu ( 1986). Examples ( 1) through (6) illustrate copular constructions and examples (7) through (10) illustrate existential constructions. Examples (1) and (2) are equative clauses. (1)
Conjunct-Interrogative: 2nd person, with copula yin
s
cc
khyedmng -tsho slob.grwa. ba gsar.pa yin
pas
student new 'Are you all new students?' (Hu 1986: 36)
Q
2S.HON
(2)
- PL
COP.CONJ
Conjunct-Declarative: 1st person, with copula yin
s nga 1s
cc -tsho
tshang.ma
slob.grwa. ba gsar.pa yin
all student 'We are all new students.' (Hu 1986: 36) -PL
new
COP.CONJ
In ( 1) the conjunct form co-occurs with a second-person pronoun in an interrogative clause. In (2) the conjunct form co-occurs with a firstperson pronoun in a declarative clause. Hale suggests that the conjunct form in second-person interrogative clauses demonstrates the speaker's anticipation of the expected answer rather than the fact that the argument is second person (Hale 1980: 99). Tournadre and Dorje refer to this as ''the rule of anticipation" (2003: 94). As Hale points out, second-person declarative clauses and firstperson interrogative clauses are ''tricky" (Hale 1980: 99) and only occur in a few highly marked contexts. Utterances such as (3) and ( 4) in which a speaker asks a question about himself or tells an addressee something about herself are rare. (3)
Disjunct-Interrogative: 1st person, attributive, w/ copula red
s
cc
nga
smyonpa
red
pas
1s crazy DISJ Q 'Am I mad?' (Toumadre and Dorje 2003: 95) 11 Both DeLancey and Tournadre, who use the tenn 'Standard Tibetan', refer primarily to those dialects which belong to the Central Tibetan dialect group. Standard Tibetan is mutually unintelligible with Dongwang Tibetan.
ANIMACY IN DONGWANG TIBET AN
(4)
137
Disjunct-Declamtive: 2nd person, attributive, w/ copula red
s
cc
khyedmng
smyonpa
(ma)
red
crazy (NEG) DISJ 'You are (not) mad.' (Toumadre and Dorje 2003: 95)
2S.HON
Languages such as English allow speakers to make obvious statements about second person addressees. For example, Oh, you got your hair cut!, You lost weight!, or You are in this class, tool are perfectly acceptable. In Tibetan (and perhaps all languages which have c/d systems), second person declarative clauses are rare as speakers defer to the "epistemic source for the action to which the utterance refers" (Hargreaves 2005: 5). That is, speakers rarely utter second-person declarative clauses as the addressee would already have certain knowledge of the content of the utterance. Similarly, speakers rarely ask a frrst-person question as the speaker is the one most knowledgeable of the answer. There are exceptions however, 12 but this deference to epistemic source makes it difficult and unnatural to elicit second-person declarative clauses and frrst-person interrogative clauses. The motivation for this 'gap' is discussed in more detail below. Examples (5) and (6) show that the disjunct form is used in interrogative and declarative clauses that contain a third-person S argument (5)
Disjunct-Interrogative: Jrd person, with copula red
s
cc
blo. bzang
dge.rgan
red
pas
Lobsang teacher DISJ Q 'Is Lobsang a teacher?' (Tournadre and Dorje 2003: 83)
12 Some examples include compliments (You are pretty) or accusations (You stole 17U1 purse). Most Tibetan dialects also have a complicated system of evidentials, which allow speakers to make second-person statements based on visual evidence. In such cases, the epistemic somce might be the speaker observing something the addressee might not see (You are bleeding). In such cases, an evidential marker indicating the visual source of knowledge would be used
138
(6)
ELLEN BARTEE
Disjunct-Declarative: Jrd person, with copula red
cc
cc
dge. rgan ma
red
slab.grwa. ba red
teacher NEG DISJ student DISJ '(He) is not a teacher. (He) is a student.' (Tournadre and Dorje 2003: 84) In naturally-occurring Tibetan data, arguments are freely omitted as in (6) above. C/d patterns in clauses that express existence, and its extension to location (Jam here, The book is on the table) or possession (J have a ball), operate in the same way as copular clauses. In clauses which express possession, c/d forms index epistemic source with respect to the possessor argument (POSR) and not to the possessed argument (POSD). (1)
Conjunct-Interrogative: 2nd person, with existential yod POSD
deb yod pas book EX.CONJ Q 'Do (you) have any books?' (Tournadre and Dmje 2003: 105) (8)
Conjunct-Declarative: 1st person, with existential yod POSD
deb mang.po yod book many EX.CONJ '(I) have many books.' (Tournadre and Dmje 2003: 105) As in the equative clauses described earlier, the use of the conjunct existential in (7) anticipates the expected reply given in (8). Although both (7) and (8) have unexpressed possessor arguments, the conjunct forms prevent ambiguity. In (9) and (10) the disjunct form is used for interrogative and declarative clauses because the third person is the possessor argument.
(9)
Disjunct-Interrogative: Jrd person, with existential yod.red POSR
POSD
khong
=fa
3S.HON
=DAT
sngon.po yodred pas hat blue EX.DISJ Q 'Does s/he have a blue hat?' (Hu 1986: 60) zhwa.mo
ANIMACY IN DONGWANG TIBET AN (10)
139
Disjunct-Declamtive: Jrdperson, with existentialyod.red POSR
khong
=fa
phalcher yod.red
maybe EX.DISJ 'Sihe might have (a blue hat).' (Hu 1986: 61)
3S.HON
=DAT
As discussed above, the 'gap' produced by the relatively infrequent occurrence of declarative clauses with second person actors or interrogative clauses with :first person actors reflects a grammatical constraint in Tibetan. Attempting to account for what determines grammaticality in the Tibetan c/d system, Garrett (2001: 16) employs the term "privileged access" which acknowledges that there are facts that one can know in a certain way' and other facts that one cannot know in that same way. Why would a speaker tell the addressee what the addressee surely must know? Or conversely, why would a speaker ask a question about herself which she surely must already know? Such sentences do occasionally occur, 13 but they are rare. For such contexts, speakers of Tibetan have a full array of hedges (looks like, appears that, seems like) and evidentials (indicating visual, inference or hearsay knowledge sources) to accompany such utterances. Unfortunately, a complete description of all the auxiliaries is beyond the scope of this paper. A ftnal distinctive characteristic of the c/d system is that of reported speech. Conjunct forms occur in the complement clause when the S/A argument is coreferential with the A argument in the matrix clause as shown below. ( 11)
Conjunct-Reported speech: coreferential A khong
=gisx naax
emchi yin
3S.HON
=ERG
doctor
zer
COMPL
lab
1S
CONJ
song
say VIS.PFV 'Hex said lxam a doctor.' (Tournadre and Dotje 2003: 214)
QUOT
13
This is particularly true in contexts when the speaker is joking or challenging
the addressee.
140
ELLEN BARTEE
( 12) Disjunct-Reported speech: non-coreferential A khong 3S.HON
COMPL =gisx JJB8y =ERG 1S
emchi ma doctor NEG
red DISJ
gsung byung say.HON EGO 'Hex said (to me) 'ly am not a doctor'.' (Tournardre and Dorje 2003: 215) It is important to observe in ( 11) and ( 12) above, the choice of the conjunct or disjunct form, and not the pronoun, triggers the referential distinction. This is not due to direct/indirect speech, as the distinctions hold with subordinate third person S/A arguments as well. What makes the c/d system even more interesting and complex in Tibetan is the speaker's freedom to manipulate the system for pragmatic pwposes which further supports the epistemic function of the system, rather than strictly cross-referencing (Slater 1998: 354). Katrin Hasler's analysis based upon an "empathy hierarchy" is helpful for showing that speakers' choice of form can be motivated by "deictic criteria (temporal distance, direction of an action towards or away from the speaker), personal closeness or personal involvement of the speaker in an action/event (even if this involvement is not a direct one and the speaker is not herself a participant in the utterance), evidential criteria which also express the perspective of the speaker in relation to an action/event, and discourse motivated criteria such as backgrounding or politeness" (Hasler 2001: 27). The following examples from Denwood (1999) illustrate one simple aspect ofthis: 14
(13) Conjunct-Declarative: 1st person, expected
s
cc
nga bslab.grwa. ba yin 1s student CONJ '(Personally) I'm a student' (Denwood 1999: 120)
14 In these examples, I use the tenn 'expected' to refer to the correlation of S/A arguments with c/d fonns as described above, and 'unexpected' when the S/A arguments do not correlate with c/d forms described thus far.
ANIMACY IN DONGWANG TIBET AN
141
(14) Disjunct-Declamtive: 1st person, unexpected
s
cc
nga
bslab.gzwa.ba red
1s student DISJ '(It's a fact that) I'm a student ' 15 (Denwood 1999: 120) (15) Disjunct-Declamtive: Jrd person, expected
s
cc
khong bslab.gzwa.ba red 3S.HON student DISJ 'He's a student.' (Denwood 1999: 121)
(16) Conjunct-Declamtive: Jrd person, unexpected
s
cc
khong bslab.gzwa.ba yin 3S.HON student CONJ 'He's a student (of mine)' or 'He's my fellow student.' (Denwood 1999: 121)
The use of the conjunct form in (13) and the disjunct form in (15) correlate as 'expected' with the active participant of the clause. The surprising use of the disjunct form with a ftrst person participant in (14) and the conjunct form with a third person participant in (16) can be explained with Hasler's analysis. In (14) the speaker distances herself from the statement (either in time or subjectivity), while in (16) the speaker covertly inserts her intimate knowledge of the third person participant. Such interplay as this is what Aikhenvald (2003a) has called an "evidential strategy' in which non-evidential categories can be 'skewed' in such a way to express evidential-like notions. Such evidential strategies are different from grammatical evidentials. The primary function of grammatical evidentials is to indicate 'source of information" (Aikhenvald 2003a: 2). 16
15 The most common use of this type of clause relates to temporal distance, as in 'I used to be a student' 16 It is important to keep these categories separate as speakers of Tibetan can use categories that are not primarily evidential to express evidential-like notions. In addition, speakers of Tibetan may use one of several grammatical morphemes whose primary function is to express evidential notions.
142
ELLEN BARTEE
Finally, copular and existential verbs also function as auxiliary verbs in clauses. In such clauses, the distribution of c/d forms follows the same pattern of grammatical constraints and pragmatic freedom as demonstrated above. (17) Conjunct-Declarative: 1st person, yod A 0 nga bodyig sbyang gi yod 1S Tibetan study FUT EX.CONJ 'I am studying Written Tibetan.' (Hu 1986: 74)
( 18) Disjunct-Declarative: Jrd person, yod.red
s khong -tsho deng. sang 3 S.HON -PL recently slobs.byong.gnang gi yodred study.HON FUT EX.DISJ 'Recently they have been/are studying.' (Hu 1986: 74) In (17) and (18), the occurrence of the auxiliaries yod (conjunct) and yod.red (disjunct) correlate with the ftrst and third person participants respectively. 17 The conjunct forms (gi.yin and pa.yin) can only occur with controllable verbs of which the frrst person is an intentional agent. These same restrictions apply to gi.yod when it is used in a present context. When the frrst person is not an intentional agent, the disjunct form is used to signal lack of intention and control. (19) Conjunct-Declarative: 1st person, past, intentional, control verb,
v!n S OBL nga nang =fa bsdad pa yin 1S inside =LOC stay PST CONJ 'I stayed inside (at home).' (Tournadre and Dorje 2003: 129)
17 In addition to the functions described above, auxiliaries also contribute, not surprisingly, to the tense-aspect infonnation of a clause. In general, gi.yod and gi.yod.red indicate imperfective aspect, which can either be presently occuning or have already occmred. The copular verbs yin and red combine periphrastically with gi andpa to express future (gi.yin and gi.ret!) and past (pa.yin andpa.ret!).
ANIMACY IN DONGWANG TIBET AN
143
(20) Disjunct-Declamtive: Jrd person, red
S nyi.ma Nyima 'Nyima 129)
OBL
nang =fa bsdad pa red inside =LOC stay PST CONJ stayed inside (at home).' (Tournadre and Dorje 2003:
(21) *Conjunct-Declamtive: 1st person, fi.Jture, intentional, non-
control vero, yin *nga na gi.yin/
pa.yin/
gi.yod
*1S ill FUT.CONJ PST.CONJ IMPFV.CONJ *'I am going to be ill', *'I was ill', *'I am being ill.'
Since the verb bsdad 'to sit' in (19) and (20) is a verb over which the first person can exert control, all three conjunct auxiliaries discussed so far can be used (bsdad gi.yod 'I am sitting', bsdad gi.yin 'I am going to sit', and bsdadpa.yin 'I sat'). However, the verb na 'to be ill' in (21) is a non-controllable verb, thus the co-occurrence with conjunct forms is ungrammatical. 18 When a first person participant does not have control over an action, a disjunct form can be used. In (22) the disjunct auxiliary occurs in the clause with a first person participant because the actor of brjed 'forget' is not an intentional instigator of the action. (22) Disjunct-Declamtive: 1st person, non-control vero, red
s bljtxi pa red forgot PST DISJ 'I forgot.' (Hu 1986: 150)
ngas
1S.ERG
18 There are other auxiliaries which I have not mentioned, due to considerations of space. For example, the 'ego' auxiliary byung expresses 'it happened to me' and occurs in clauses such as 'I got sick' There are also ways to express that one is intentionally acting ill, sneezing, etc. In such cases, the most common way would be to nominalise the non-control verb and then 'verbalise' that construction using a 'phrasal verb' One interpretation of the following example would indicate the speaker is acting sick:
(i)
ngas na.pa bycd pa lS.ERG sick.NMLZ do PST 'I did a sick' or 'I acted sick'
yin CONJ
144
ELLEN BARTEE
Grammatical restrictions which constrain frrst person participants, depending upon such factors as intention and control, illustrate the epistemic nature of the c/d system which points to the speaker as the 'default' epistemic source. In this section, I have provided a general description of the c/d system in Tibetan by examining the correlation of conjunct forms in fmite declarative clauses with frrst person, intentional participants and in interrogative clauses with second person participants, and disjunct forms in all other contexts. I also illustrated how c/d forms are used in coreferential and non-coreferential complement clauses of reported speech. Finally, I discussed the ability of speakers to manipulate the system for evidential, empathy and discourse pwposes, and illustrated how notions such as intention and control can restrict the occurrence of conjunct forms. In the following section, I frrst give a brief outline the c/d system in Dongwang Tibetan, before presenting the unique characteristics of existential verbs and auxiliaries in detail. 2. DONGWANGTIBETAN There are three major Tibetan dialects within China: Central, Amdo, and Khams, each of which has many sub-dialects. The Khams group contains the most internal diversity and many 'Khams dialects' may be distinct dialects in their own right (Sun 2001). For the purposes of this paper, I distinguish 'Southern Khams dialects' from other Khams dialects. Diqing Tibetan Autonomous Prefecture (JJ!!.ECJt~ § ¥it1'H), where Southern Khams dialects are spoken, is situated in the northwest comer of Yunnan Province in the southwest ofthe People's Republic of China. It is comprised of three counties: Dechen (t\M.X), Wcixi <*il§) and Shangri-la (ifm:EI:JJL). Almost all of the inhabitants ofDechen County (closest to the border of the Tibetan Autonomous Region) are Tibetans, while Lisus are the majority population in Weixi County. Shangri-la County (ifm:EI:JJL!l), where Dongwang is spoken, lies on the 'frontier' of Tibet, and is bordered by Tibetan counties to the north, east, and west, 19 Yulong Naxi Autonomous County (::Ji:/t~il§~ § ~!!) to the 19 Ywman's Dechen Autonomous County (.fttx §7&£.) to the west and Sichuan's Derong (~5R), Xiangcheng (~ }jj(;) and Daocheng (~}jj(;) counties in Gyantse Tibetan Autonomous Prefecture to the north and east.
ANIMACY IN DONGWANG TIBET AN
145
south, and shares a brief border with the Weixi Lisu Autonomous County <*Jl§il-11~§ 1a!l) to the southwest. Dongwang Tibetan is spoken by approximately 6,000 speakers who live in 57 villages scattered along the steep sides of the Dongwang River in eastern Shangrila County. The data for this paper are drawn from ten short texts which include frrst-person narratives, third-person narratives and procedural texts. I also have a database of over 1,000 elicited clauses. All of the speakers come from Pongding Village, 20 located roughly in the center of the Dongwang Valley. Three of the speakers are men, 21 and four of the speakers are women. The speakers range from twenty-six to fifty-three years old. In addition to data collected in Dongwang, I conducted preliminary interviews and elicited data from speakers in other parts of Southern Khams. Preliminary research suggests that the features I describe for Dongwang in this paper might be characteristic of Southern Khams dialects in general. 2.1. An overview ofthe c/d system in Dongwang
While Dongwang seems to have a more complicated c/d system than that observed in Standard Tibetan described above, the patterns and functions of the system remain quite similar. Although I am mainly interested in the animacy distinction observed in the existential forms, it is important to present these as part of the overall c/d system. In this section, I frrst present the most frequent forms, 22 leaving the existential forms for last. Since there are so many forms, I will not give examples of every form. 2.1.1. Copulas The copulas in Dongwang Tibetan are closely related to those found in other dialects of Tibetan. Table (1) shows the Dongwang forms:
20 I also have texts and elicited clauses from another village, Shengli, at the end of the Dongwang valley, but there are enough differences in the auxiliaries that including them would not contribute to clarity. Further studies and comparisons are in progress. 21 In fact, all three are brothers. 22 A full description of the c/d system will be covered in another paper.
146
ELLEN BARTEE
Table 1 Copulas in Dongwang Tibetan Affmnative Interrogative:.~." as3- a:ssro Conjunct zT Disjunct re are ro
Negative me mare
Table (1) shows the copular verbs found in Dongwang Tibetan. The forms are nearly identical to those already presented for Standard Tibetan with two notable exceptions. First, the reflex of Written Tibetan 24 is [j] in Standard Tibetan, but [z] in Dongwang. So [zT], rather than yin, is a reflex of Written Tibetan . Second, the interrogative fonns are quite different from Standard Tibetan. In Standard Tibetan, the interrogative fonns are quite transparent, 25 unlike Dongwang. This is likely due to the fact that question particles occur before the copula in Dongwan~6 and that the two fonns appear to have fused into one form. 27 2.1.1.1. Copulas, conjunct The following set of examples28 illustrate the basic pattern of c/d marking in declarative, interrogative and reported speech clauses. As discussed above for other languages with c/d patterns, conjunct forms are used in ftnite clauses in which frrst person is the S or A argument and in interrogative clauses in which second person is the S or A argument (23) Conjunct-Declarative: equative, 1st person, zT
s
cc
gai3 hmrei3fxrss zl 1s doctor CONJ 'I am a doctor.'
23 Dongwang has several other interrogative fonns and constructions. The interrogatives listed in Table (1) are yes/no question fonns. 24 I am using <> to bracket fonns which are attested in Written Tibetan. 25 In Standard Tibetan, the conjunct copula is yin and the question is yin bas. 26 This may not be much different from the Eastern Khams dialects. I believe that 1he standard practice for Khams dialects is to put an interrogative marker such as a before the verb or auxiliary. It is interesting that other adjacent languages also fonn questions in this way. Z1 Possibly: a 53+ ji= a5~ If this is indeed the case, it seems to indicate that these fonns merged prior to the ~ [z] sound change. 28 Unless otherwise noted, all the examples were elicited from Traba, Yishi Droma and Danzen Chumpi, Pongding Village, Dongwang.
ANIMACY IN DONGWANG TIBET AN
(24) Conjunct-Interrogative: equative, 2nd person,
s
147
a53
cc
~55 pe13
ii53
Tibetan CONJ:Q 'Are you Tibetan?'
2S
(25) Conjunct-interrogative: equative, 2nd person.
s
~55
a53ro
cc t655wiJ53
)XJ13
2s Dongwang person 'Are you Dongwangi?'
(26) Conjunct-Declarative: equative, negative, 1st person, me
s
cc
gai3
pei3
me
ls Tibetan CONJ:NEG 'I am not Tibetan.' Examples (23) through (26) illustrate expected patterns of c/d marking. The conjunct form is used in declarative clauses with a frrst personS and in interrogative clauses with a second personS. The alternate form in (25) seems to signal politeness to 'soften' a direct question. Recall that a crucial characteristic of c/d systems discussed above is that when the S or A argument of a complement clause of reported speech is coreferential with the A argument in the matrix clause, the conjunct form is used. Examples (27) and (28) show that this characteristic holds for Dongwang Tibetan as well. (27) Conjunct-Reported speech: core.ferential, zi A
COMPL
khui53x
kb:J5~
pei3
3S Tibetan 'Hex said hex is Tibetan.'
3S.ERG
zl
s:JI3
d:{a?
CONJ
say
MOD
148
ELLEN BARTEE
(28) Conjunct-Reported speech: coreferential, zi (RabbitA26/27) A tli}ss piis78 53
=Jix rr
COMPL ya1~
me
then rabbit
=ERG
ls
NEG.CONJ
8:}13
FOC
d¥i?
say DISJ 'Then the rabbitx said '(It) was not mex. " 29 Example (28) is taken from a folktale in which a shepherd suspects that the rabbit who has come to get butter from him has been there once before. The rabbit, lying to the shepherd, disputes this and says 'It wasn't me'. The S arguments ofthe complement clause in (27) and (28) are coreferential with the A argument of the matrix verb s:J 13 'to say', so the conjunct form is used in both examples even though there is a third person form in the matrix clause and a ftrst person form in the complement clause in (28). Clearly the nominal/pronominal forms are not what indexes the referent in Dongwang. This pattern is the same as that shown above for other Tibetan varieties. C/d forms are not obligatory in all subordinate clauses, and are in fact, rare. When they do occur in non-complement clause constructions, however, the c/d distinction is neutralised and only the conjunct form is used. This is true for other dialects of Tibetan as well. This neutralisation seems to be more frequent in the existential verbs and auxiliaries, and is discussed in Section 2.3 below. 2.1.1.2. Copulas, disjunct As we have seen, when the single argument, or most agentive argument, of declarative clauses is non-frrst person, the disjunct form is most frequently used. (29) Disjunct-Declarative: equative. Jrd person, re
s yen
cc ~53
rr
nd~p13gu
lS.GEN dog RI mastiff 'My dog is that black Mastiff.'
29
n:J
1iJa 53
black
t:J
re
DET
DISJ
The first person pronoun :ga 13is the S argument of the complement clause, con-
trary to what is reflected by the English translation.
ANIMACY IN DONGWANG TIBET AN
149
(30) Disjunct-Declamtive: negative, equative, Jrd person, re
s
cc
.k:IKJss
pei3
ma
Tibetan NEG 'Sihe is not Tibetan.'
3S
re DISJ
In interrogative clauses when the single argument, or most agentive argument, is third or ftrst person, the disjunct form is commonly used. (31)
Disjunct-Interrogative: eauative. Jrdeerson. a 55re ro
s
cc
.k:IKJss losssr
as+ero
teacher DISJ.Q 'Iss/he a teacher?'
3s
(32) Disjunct-Interrogative: equative, 1st person, a55 re ro
s
cc
ga13
pe13
as+e ro
1s Tibetan DISJ.Q 'Am I Tibetan?'
While it is easy to fmd situations in which utterances such as (31) might occur, it is less likely to fmd situations in which (32) would be an information-seeking question. As has been discussed for other dialects of Tibetan, such occurrences are rare and the paucity of ftrst person interrogatives as well as second person declaratives points to the important role that epistemic source plays in grammatical choices in Dongwang as in other varieties of Tibetan. Examples of clauses with predicate adjectives in my data are too scarce to make a fmn statement regarding their patterns of use. 30 However, (33) below illustrates how it is possible to use a disjunct form of the copula rein an attributive context
°
3 Clauses which express states such as I am hungry, I am tired, and I am happy use a particular 'ego' auxiliary which indicates that the speaker is a non-volitional patient of the clause. In most of my examples of clauses wi1h third person participants, the auxiliaries used are those which grammatically mark the speaker's inability to claim knowledge of the person's internal state. Thus speakers tend to use hedges like It appears that, It seems tha~ He looks like, etc. There are at least 10 of this category of auxiliary.
150
ELLEN BARTEE
(33) Disjunct-Jrd person, attributive, w/ copula re
s tlij55
ph:J pmpmph:J
then
urn
cc ga55tff53
guei3~53
re
extremely cunning DISJ 'Then he, urn, was extremely cunning.' (RabbitB003) In (33) above, the narrator is introducing the rabbit and describing his character. Finally, when the main participant of a complement clause of quotation is not coreferential with the main participant of the matrix clause, the disjunct form is used, as shown in (34): (34) Disjunct-Reported speech: non-coreferential, re
A
COMPL
khui53x
kb:J55Y
pei3
3S.ERG 3S Tibetan 'Hex said hey is Tibetan.'
re
s:JI3
c4_a'l
DISJ
say
EVI.HS
In (34) above, the disjunct form occurs in the complement clause, which indicates that the two third person pronouns are not coreferential. This is also the case in other dialects of Tibetan, and c/d systems in general. In this section, I have illustrated patterns of c/d marking by giving examples of copular constructions in Dongwang. Before proceeding to existential constructions, I show in 2.2.2 how c/d distinctions hold when the copulas are used as auxiliaries. 2.1.2. Auxiliaries, general In Dongwang, copular c/d forms also combine with other forms to function as auxiliaries. 31 While the forms for Standard Tibetan discussed above are transparent, they are less so in Dongwang. There appear to be several reasons for this: periphrastic forms in Standard Tibetan are fused forms in Dongwang; Dongwang has more tense-aspect distinctions; and there are more auxiliaries in Dongwang which appear to have more refined functions that those in Standard Tibetan. 31 There are some additional auxiliaries which are not derived from existentials or copulas.
ANIMACY IN DONGWANG TIBET AN
151
In Sections 2.2.2.1 and 2.2.2.2 below, I present a partial, non-exhaustive list of auxiliaries that represent the more canonical cases of c/d marking. Then in Section 2.2.3 I present the existentials, which are the main focus of this paper. 2.1.2.1. Auxiliary, conjunct
Conjunct auxiliaries include the following affirmative forms: future imperfective zi; past or present imperfective dii and past perfective ji. zl is used as an auxiliary in future declarative clauses with firstperson intentional participants, and occurs with control verbs only. (35) Conjunct-Declarative: 1st person, zi
A
0
ge13
pa13 =jte
ji53
~13
cow =DAT feed go 'I am going to feed the cowls.' 1S.ERG
zl CONJ
(36) Conjunct-Declarative: 1st person, zi
s
ge 13
pa 13wc5je zl marry CONJ 'I am [going] to get married.' 1S.ERG
(37) Conjunct-Declarative, negative, tsi me
s ga13
~13
tsi
me
1S go IPFV CONJ.NEG 'I am not going.' (GetMar037)
The main participants are frrst person in declarative clauses and second person in interrogative clauses, as in (38) and (39) below. (38) Conjunct-Interrogative: 2nd person, A
0
¢ss
ka1~ess
je13 de
what do 'What are you doing?'
2S.ERG
PROG
dZi dii CONJ
152
ELLEN BARTEE
(39) Conjunct-Declarative: 1st person, @ A
10
yei3
ai3kaS3 =jre
0
child =DAT 'I am nursing the baby.' 1S.ERG
ji53
mi 53
breast feed
de
#i
PROG
CONJ
The conjwtct form is used with second person in interrogative clauses as in (38) and with frrst person in declarative clauses as in (39). While zl and dji are both imperfective, the primary difference between the two is that zl indicates future events and #i indicates past or present events. Both occur with control verbs. The auxiliary ji co-occurs with control verbs in perfective clauses with frrst person agentive participants in declarative clauses and second person agentive participants in interrogative clauses. ( 40) Conjunct-Interrogative: 2nd person. ji
A
0
~iss
pas~0 S3
na I3dzi jei3 ji
marry when do 'When did you get married?'
2S.ERG
CONJ
(41) Conjunct-Declarative: 1st person, ji
A
0
ye13
~
=ji
gai3
wood =GEN trunk 'I made a wooden trunk.' 1S.ERG
t¢ 53
zu 13
ji
one
made
COP.CONJ
(42) ConjWJCt-Declarative: Negative, 1st person, j i
s
ye 13
0
kha 13tso55 ji 13
ma
yesterday work NEG 'Yesterday I didn't work.' 1S.ERG
ngli1 53 ji do COP.CONJ
I have no examples of ji occurring in present contexts such as I am getting married, I am working, etc., as all occurrences in my data are restricted to perfective contexts. 2.1.2.2. Auxiliary, disjwtct In this section, I present an example of each auxiliary parallel to those presented in 2.1.2.1. Recall that disjunct forms occur in declarative
ANIMACY IN DONGWANG TIBET AN
153
clauses with second and third person participants, and in interrogative clauses with frrst person and third person participants. Disjunct auxiliaries include the following fonns: future imperfective re, past or present imperfective d~'l and past perfective thi. Like zl, re functions as an auxiliary in future clauses, but most commonly with second person in imperative clauses, third person participants in declarative clauses and with frrst and third person participants in interrogative clauses. Unlike z~ redoes not mark the intention of the participant or the controllability of the action in declarative clauses with second and third person participants. 32 That is, the grammar ofDongwang, as in other Tibetan varieties, does not allow speakers to claim knowledge of the intention of second or third person participants. (43) Disjunct-Declamtive: 2nd person, re
s ~ 55
ne13di
je 2s careful do 'You be careful.'
re DISJ
(44) Disjunct-Declamtive: Jrd person, re
s
nc4u13
tsi re go IPFV DISJ 'Sihe is going/will go.'
.kb:J55
3S
(45) Disjunct-Declamtive: Negative, Jrd person, rna re A
0 tshe 53 pt155 tsi 3S.ERG vegetables f>uy IPFV 'Sihe will not buy vegetables.'
khui 53
ma
re
NEG
DISJ
(43) is an example of the type of context in which a speaker might utter a clause with a disjunct form and second person actor. (44) and (45) have third personS and A arguments. (43) through (45) all cooccur with the disjunct auxiliary. 32 This does not mean that .reis not relevant to issues such as intention or control. When :rc occurs in clauses with first person S or A arguments, it often does indicate lack of intention or control. For example, when the main participant in a clause is involved in a non-controllable act such as 'I fell', the disjunct form is used
154
ELLEN BARTEE
Like dji, d~'l occurs in imperfective declarative clauses, but with second person and third person participants, and in interrogative clauses with frrst person and third person participants. ( 46) Disjunct-Declarative: 2nd person. @? 10
COMPL
tHJss .ga13
=jre
nassw0 53
~13
then
=DAT
bride
go
lS.DAT
gui ~'l se MOD DISJ say 'Then (they) said to me, '(you) should go be a bride'.' (GetMar035) ( 47) Disjunct-Declarative: Jrd person. @?
s t:J13r£ss =ji
.ga5:t853 rr
past
story
=GEN
RI
cc pii53 n:J =ji za'l353 se d~'l grassland NA =GEN yak say DISJ 'A story from the past called 'the yak of the grassland'.' (RabbitAOOl) In (46), the S argument of the complement clause, although realised as zero, is second person and thus co-occurs with a disjunct auxiliary. What is important is not that it is a second-person referent but that it is not coreferential with the A referent of the matrix clause. Like ji, thi occurs in perfective clauses, but with second and third person participants in declarative clauses and with frrst and third person participants in interrogatives. Although not directly relevant to this paper, thi also includes a visual evidential meaning, indicating that the speaker saw the whole event ( 48) Disjunct-Declarative: Jrd person. thi
A khui53
0
k:ha13ts653
a13~yss
(f:i
3 S.ERG yesterday monkey one 'S/he killed a monkey yesterday.'
se53 thi kill
DISJ
155
ANIMACY IN DONGWANG TIBET AN
I do not have any examples of thi occurring alone with second person participants in declarative clauses, but it does occur with me, apparently to soften a direct observation. (49) Disjunct-Declamtive: 2nd person, thi rm: A ¢sjJaS3 k!Jai3JaS3 tt;i 2PL
all
=ji
one =ERG
0 W05
-kJ
lPL
-PL
Pa S3
tht1353 thi me see DISJ CONJ.NEG 'You all saw us, right?' My consultant explained that (49) would occur if the speaker had seen that the addressees saw the speaker and those the speaker was with. 33 Examples such as (49) highlight that is that the scarcity of declarative clauses with second person actors and interrogative clauses with first person actors is a real gap and not an oversight 2.2.
Existentials
In the previous sections I introduced the notions of conjunct and disjunct, provided examples from Standard Tibetan and then presented the c/d system in Dongwang Tibetan. A quick comparison between the two Tibetan speech forms reveals much functional similarity in spite of the obvious motphological differences. In this section, I show that the existential clauses are different in that they are motivated by semantic notions not seen in other dialects of Tibetan. Existential clauses in Dongwang are constructed with a single unmarked S argument and an existential verb. Locational clauses can be seen as an extension of existentials in that they share the same template structure but additionally contain a locative NP or adverb. Possessive clauses are constructed with a dative-marked possessor (POSR) and an unmarked possessed argument (POSD). These can be schematicised as:
33 I am not exactly sure of the function of me as examples in my database are too scarce. Speakers say it "sounds better". Another example in my database also involves the visual evidential, but in present tense: f;i55 ~53 do 13 de hn5me '(I see) you are hit-
ting the dog'
156 Existential: NPs Locational: NP8 (LOC) Possessive: NPposrDAT
ELLEN BARTEE
Exist 'X exists' Exist 'X exists in/at/on Y' NPposn Exist 'Y has X'
In Dongwang, existential verbs reveal an animacy split which is also extended to the auxiliary forms derived from the existential verbs. This means that while the c/d pattern indexes the complex of factors such as person, intention, control, tense-aspect and knowledge source of the agentive participant, the animacy distinction indexes either the animacy of the single argument in an intransitive clause or the possessed argument in a possessive clause. To my knowledge, this has not been observed in any cld system of any language in the world. Further, although no other variety of Tibetan has been reported to have such a split, this feature occurs in several, if not many, of the Southern Khams Tibetan dialects. For the remainder of this paper, I will discuss this category of animacy, frrst giving examples from Dongwang, and then present language contact as a possible motivation for this development in Dongwang. 2.2.1. Existential verbs Recall that there are two existential verbs in Standard Tibetan: yod and yod.red. In Dongwang there are four existential verbs: 34 ndo(conjunct, animate) and ze (conjunct, inanimate); and ndo d$i.? (disjunct, animate) and ze d$i".?(disjunct, inanimate). In single argument clauses, a speaker's choice is conditioned not only by cld considerations but also by the animacy of the S argument In clauses expressing possession, a speaker's choice is conditioned by c/d considerations as to the possessor on the one hand, and by the animacy of the possessed on the other. Thus there are two different indices for each existential. The interrogative, affmnative and negative forms are shown in the table below:
34 This does not include those forms which I consider to be primarily evidential in nature. In imperfective clauses, these are 'dug for Central Tibetan and hn5 for Dongwang Tibetan. While these fimction primarily as evidentials, they can have existential values as well. While the semantics and distribution of evidentials are beyond the scope of this paper, I will briefly mention these forms below when discussing the etymology of the animate morphemes.
157
ANIMACY IN DONGWANG TIBET AN
T.bl2 E"t 1 etan a e xtsentil" asm D ongwang Tb Interro~ative Affmn ass ze -AN Conj ze ass ndo +AN ndo -AN Disj ze ~?are (ro) ze~? +AN ndo ~? a re (ro) ndod~?
Ne~ative
.ne? mando zed~?
mare
ndo~?mare
Table (2) shows the existential forms that code both c/d and animacy. We can see that ze is used for inanimate forms and ndo is used for animate forms. The Written Tibetan conj1mct existential has become ze in Dongwang/5 while ndo comes from the verb <'dug> 'to sit' or 'to dwell', which in Dongwang also f1mctions as a full verb. It is interesting that in many dialects of Tibetan this verb has grammaticised into some sort of evidential marker. 36 In Dongwang, however, hn5 is a visual evidential (most likely from the WrT verb <snang> meaning 'to appear', 'to be visible') and ndo contributes animacy meaning. It is not surprising that a verb meaning 'to sit' might acquire animate meanings. The fact that it is <'dug> suggests two possible scenarios: the grammatical marking of animacy occurred prior to the grammatical marking of evidentiality, or vice versa. It is quite possible that more research on neighbouring dialects will reveal the parallel paths of grammaticisation that <'dug> and <snang> have followed. In Sections 2.3.1.1 and 2.3.1.2 below, I illustrate each of these with examples from Dongwang. 2.2.1.1. Conj1mct The existential ndo occurs in declarative clauses with ftrst person animate participants or in interrogative clauses with second person animate participants.
35 Throughout the Tibetan dialects, the coronals in coda position affect the previous vowel in different ways. Frequently, the vowel preceding a coronal is fronted, while retaining roundedness. In Dongwang, however, the vowel is fronted and wtrmmded. 36 Usually a direct evidential, or a 'mirati.ve' marker when used in first person contexts.
•
158
ELLEN BARTEE
(50) Conjunct-Interrogative: 2nd person +an. ndo
S
LOC
¢ss =m 2S home =LOC 'Are you home?' ~ess
8
ss ndo
Q
EX.AN.CONJ
(51) Conjunct-Declarative: 1st person +an. ndo S LOC gai3 ¢ss =m ndo 1S home =LOC EX.AN.CONJ 'I am at home.' Both (50) and (51) above are frequently overheard in telephone conversations. In fact, it is unlikely that these clauses would be uttered in a face-to-face context as it would be implausible for this exchange to take place between speech act participants when they clearly know where the other is. It is difficult to fmd a clause in which the inanimate conjunct existential ze is the main verb of a non-possessive construction. This is because one would obviously not question, nor expect a reply from, an inanimate object. However, there are occasionally utterances such as (52) below, which expresses location rather than existence. (52) Conjunct-Declarative: (Jrd person) -an, ze
S !Ji13
LOC t:J13kjSS t:JSpJJa:53
1S.GEN coat over there 'My coat is over there.'
Ze EX.IN.CONJ
(52) is an interesting example of how the empathy hierarchy can interact with both c/d and animacy in Dongwang. Given the freedom according to the concepts of the 'empathy hierarchy' as discussed above, the speaker could either use the conjunct or the disjunct existential in the clause with the third-personS argument 'my coat'. But it appears that the marking of animacy is more constrained and can only correlate with the actual referent, the inanimate head of the noun phrase 'my coat'. As noted, in possessive clauses the c/d distinction holds for the possessor, but the animacy distinction holds for the possessed. In (53) and (54) below, the possessed arguments are animate.
159
ANIMACY IN DONGWANG TIBET AN
(53) Conjunct-Interrogative: 2nd person posr. +an posd, ndo POSR
POSD
a13ka53 a ndo child Q EX.AN.CONJ 'Do you have a child?' fl855
2S.DAT
(54) Conjunct-Declarative: 1st person posr, +an posd, ndo POSR
POSD
wo5j1a53 J1CU53
=jte
a13ka53
two =DAT child 'We two have many children.' lPL
ma1311XJ55 ndo
many
EX.AN.CONJ
In (53) and (54), conjunct existentials correlate with the second person in interrogatives and with the ftrst person in a declarative clause, but the animate existential correlates with the animate possessed argument a 13ka 53 'child/children'. Conjunct forms are also used in (55) and (56) below, but the inanimate forms are selected based on the inanimacy of the possessed argument (55) Conjunct-Interrogative: 2nd person posr. -an posd, ze POSR
POSD
fl855
zi13gi55 a
Ze
book Q EX.IN.CONJ 'Do you have a book?'
2S.DAT
(56) Conjunct-Declarative: 1st person posr, -an posd, ze POSR
POSD
1]813
a13r155
=ji
bao53zhi
lS.DAT today =GEN newspaper 'I have today's newspaper.'
Ze EX.IN.CONJ
The relevant difference between examples (53) and (54) and examples (55) and (56) is not one of c/d status, as the conjunct form is chosen in all four examples. The difference is in the animacy of the possessed argument In Dongwang this split holds for disjunct environments as well. Disjunct When the single argument of an existential clause is third person, the disjunct form of the existential is used, and the animacy of the S argument determines the form of the disjunct. 2.2.1.2.
160
ELLEN BARTEE
(57) Disjunct-Declarative: Jrdperson +an, ndo d~?
s dess wrssz;, re
kh:Js7xf$i53
= ji
P:Jss
then that
3PL.GEN
=GEN
boy
SUB LOC
t~:JI3wuss
rr fliii3 =n:J
ndo c#i?
s
older RI field =LOC EX.AN.DISJ QUOT 'Then at that time, the older son of the household was in the field.' (Prod058) (58) Disjunct-Declarative: Jrd person -an, ze #? LOC
tc5sswfi53
rc5
¥CI30053
00 s5Ja53
Dongwang
and
rGyalthang between
s
r:J 13
sii53 ze d$i mountain three EX.IN.DISJ 'There are three mountains between Dongwang and rGyalthang.' Examples (57) and (58) are both locational clauses, but the single argument of (57) 'the older son' is animate, while the single argument of (58) is inanimate. This animacy distinction is reflected in the choice of an animate or inanimate form of the disjunct existential. As we have seen, in possessive clauses the choice of c/d is determined by the person of the possessor, but the choice of animate/inanimate forms is determined by the animacy of the possessed argument (59) Disjunct-Interrogative: Jrdperson posr. +an posd, ndo #? POSR
POSD
.khw:J 57xf¥a 53 ~ 53
ndo d$i? a rerc5 dog EX.AN.DISJ Q 'Do they (their household) have a dog/dogs?'
3PL.DAT
ANIMACY IN DONGWANG TIBET AN
161
(60) Disjunct-Interrogative: 1st person posr, +an posd. ndo d:;@? POSR
POSD
ai3kaS3 ndocf¥jP child EX.AN.DISJ 'Do I have a child?'
gai3
are
1S.DAT
Q
The possessor arguments in (59) and (60) trigger the disjwtct forms. Remember that ftrst person questions are somewhat wtusual and there are several ways that such a question could be asked. My consultant said that a question as in (60) above could occur in a context in which someone has told the speaker that s/he has a child (either in joking or accusation), and the speaker responds with Do I (really) have a child? The important thing to note here is that since both possessed arguments are animate, the animate disjwtct existential verb is chosen. The question in (61) below contains an inanimate existential verb, reflecting the inanimate nature of the possessed argument (61) Disj-Interrogative: Jrd person posr. -an posd, ze d:i@? POSR
POSD
zi 13gi55 ze cf¥jP are ro book EX.IN.DISJ Q 'Does s/he have a book/books?'
khua 53 3S.DAT
Both (62) and (63) below contain third person possessor arguments, but the difference in animacy of the possessed arguments triggers different disjwtctforms. (62) Disjunct-Declamtive: Jrdperson posr, +an posd. ndo d:;@? POSR
POSD
]KJ 13 t¢ = jte ]XJ 13 J1CU 53 ndo d¥iP person one =DAT son two EX.AN.DISJ 'A man had two sons.' (ProdOO 1)
(63) Disjunct-Declamtive: Jrd person posr. -an posd. zed:;@? POSR
POSD
zii3giss mai3m:Jss Ze d¥iP book many EX.IN.DISJ 'Sihe has many books.' khuaS3
3S.DAT
162
ELLEN BARTEE
Recall that in reported speech the choice of c/d form can indicate coreferentiality of S or A arguments in the matrix and embedded clauses; when the S or A arguments of both clauses are coreferential, the conjWlct form is used. This holds for existentials as well, with choice further conditioned by the animacy of the possessed argument of the complement clause. The examples below illustrate the animate c/d forms:
(64) Conjunct-Reported speech: Corekrential +an posd, ndo A khuiss
COMPL
khua53
ai3ka53 ndo
3S.ERG 3S.DAT child EX.AN.CONJ 'Slhex said slhex has a child.'
s:J
c4a
say
QUOT
( 65) Disjunct-Reported speech: Non-coreferential +an posd. ndo ~
A khuiss
COMPL
khua53
ai3ka53 ndo ~p
3S.ERG 3S.DAT child EX.AN.DISJ 'Slhex said s/hey has a child.'
s:J
cf7IJ
say
QUOT
Based on my present data set, a preliminary hypothesis is that whatever is capable of volitional movement is animate and everything else is inanimate. The animate category thus includes small and large animals (including insects), demons, 37 children and adults. (66) Conjunct-Interrogative: Joe +an. ndo POSR ~iss
=go
POSD
d.ps5wa53 a53 ndo 2S.GEN =LOC flea Q EX.AN.CONJ 'Do you have (a) flea/s?' (='Is/are there (a) flea/s on you?') Conversely, body parts seem to be treated as inanimates.
TI Tibetans have many classes of gods and demons. I assume that Dongwang speakers would treat all demons and gods as animate, but my data are too sketchy to confirm that. Once I was watching a movie with a friend In the movie one of the characters was seized by a demon. I asked my friend how to express that in Dongwang and she said kh{)55h855dztj ndo @7 'He has a demon', using the animate existential form.
163
ANIMACY IN DONGWANG TIBET AN
(67) Disjunct-Declamtive: Jrd person posr. -an posd, ze @? POSR
POSD
khua 53 kha 55
t~:J 13t~ho
mouth small 'S/he has a small mouth.'
3S.DAT
4;i
zecf¥jP
one
EX.IN.DISJ
(68) Conjunct-Declamtive: Joe -an, ze POSR
yei3
POSD
khi 5~ba53
=go
foot =LOC 'My foot has a wound.'
lS.GEN
m:J 5*:ha 53 t¢
wound
ze
one EX.IN.CONJ
Both (66) and (68) have an overt locative marker, which suggests that these are single argument clauses. The use of the conjunct form is conditioned by the involvement of the second person as the physical location of the 'flea' and the 'wound'. Additionally, the semantics of (67) is more attributive than possessive, but the syntax is as in other possessive clauses (possessor argument is in the dative case; possessed argument is an unmarked NP). In my data, Dongwang speakers treat 'flea' as grammatically animate and 'mouth' and 'wound' as grammatically inanimate. In the following section, I illustrate the distribution of animate and inanimate forms in the auxiliaries.
2.2.2. Animacy split in non-existential clauses In the previous section I have shown how animacy plays a primary role in locational and possessive existential clauses. In this next section, I show how some of the same morphemes reflect animacy in four other aspects of the grammar: in attributive constructions, in nominalised and relative clause constructions, as auxiliaries in intransitive and transitive clauses, and in subordinate clauses.
Attributive constructions In my data, the copulas zl or re are most frequently used when the copula complement is attributive and conveys an inherent quality. Examples of the existential morphemes being used in attributive constructions are infrequent However, when they do occur, the single ar-
164
ELLEN BARTEE
gument of the clause controls the c/d38 and animacy choice. Expressing an attribute construction using the existential form seems to convey a more transitional quality. ( 69) Disjunct-Jrd person attributive -an. w/ existential, ze @? 0 ~IIIJ53
wf55z:s.o53 ma J3m:J55 ma
te53
ni
rr
pigfat
like this
give NI
RI
tHJ55
rr
(f;ha13
much ~55
NEG
rr
f:JJ3m:J55
ze ~'I
then RI eat when RI tasty EX.IN.DISJ 'If (you) don't put too much fat on it, then when (one) eats (pig sausage) (it) is tasty.' (K.il1Pig063) In the context of ( 69) above, the speaker is describing how to make sausages from pig's intestines in just the right way. Although it is not mentioned here, the inanimate S argument, z:1 13w6 53 'intestine', is mentioned several clauses previous to this. In (70) below, the S argument is animate and the animate existential is used as expected. (70) Disjunct-Jrd person attributive +an. w/ existential, ndo c;bi?
s Jo55S:J
ph:Jph:Jph:J
JJa13tiJa53
t¢
ndo ~'l
teacher urn hardworking one EX.AN.DISJ '(The/our) teacher was extremely hardworking.' (MyLife136) The animacy distinction is maintained within the pragmatic nature of the c/d system. That is, speakers are allowed to express various degrees of empathy by using an unexpected c/d form, but they are required to maintain the animacy distinction. Thus in (71) below the speaker is describing school days that took place more than forty years earlier:
38 Taking into account the speaker's potential for pragmatic skewing to convey intimate knowledge or distance as elaborated by the 'empathy hierarchy' mentioned above.
ANIMACY IN DONGWANG TIBET AN
165
(71) Disjunct: 1st person attributive +an. w/ exististential, ndo d¥?
s
t:JI31Wss wo5jJa53 -kl rr then 1PL -PL RI
w
5tse 53 tsh657rii ndo ~p ma re like that smart EX.AN.DISJ NEG COP.DISJ 'At that time, we were not smart like that.' (MyLife047) The single argument of (71) is frrst-person plural, yet the speaker chooses a disjwtct auxiliary. This is a common strategy in which a speaker shifts empathy based on temporal distance. The existential still reflects the animacy of the S argument in spite of the shift in empathy.
Relative clauses Relative clauses in Dongwang are formed when a nominalised verb or clause modify a head nowt. The relative clause occurs before the head nowt. It is interesting that the animacy distinction is reflected when existential verbs ze or ndo are used to form a relative clause. Further, it is the head of the relative clause that triggers the choice of animate or inanimate forms.
(72) Relative Clause: -an head w/ existential ze A P:JI3 t~h:JI3t~h6ss son yowtg.small ZE
n:J
EX.IN
NMLZ
=ji =GEN
0
=ji =ERG
khua 53 3S.DAT
se 5 ]xr53 things
k!Jai3la53 all
tsh:Jss ro53 tht:enss !Wss SUB. WHEN hither gather PFV 'When the yowtger son had gathered all the possessions that he had... ' (Prod007) (73) Relative Clause: +an head w/existential ndo
s khua 53 ndo
n:J
~53
3S.dat EX.AN NMLZ dog 'The dogs/he has is good.'
t:J
DET
a I:poss hn3 good EX.EVI
166
ELLEN BARTEE
In (72) the head of the relative clause is inanimate: se 5]la 53 'things'. This is reflected in the choice of the inanimate existential form ze. (73) is also a relative clause, but the animacy of the head of the relative clause, ~ 53 'dog', in (73) triggers the animate auxiliary. Notice that while the c/d distinction is neutralised, the animacy distinction is maintained. Thus the animacy of the referent within the relative clause controls the choice of animate versus inanimate selection.
Intransitive and transitive clauses In my data, when ndofunctions as an auxiliary in an intransitive clause, it carries additional aspectual meaning, specifically occurring with events, past or present, that are happening within a specified time frame: 39
(74) Con[unct-Interro!J!!.tive: intransitive1 2nd f!E.ISO!JJ. +an
s
re13 ka13 nc4u13 ndo 2s now where go EX.AN.CONJ 'Where are you going now?' ~ess
(75) Con[unct-Declarative: intransitive1 1st f!E.ISon
s
gai3 1s
LOC
n: .. :J
jit(X)t~olu
1996
RI
gai3 tsh:J 55d2;6 1s middle school zjliJcfEe
J1E year
bi 5peP53 jei3 ma graduate do NEG
tlNiJ55 ni PFV
NI
ndo
study EX.AN.CONJ 'I, uh, in 1996 having not yet graduated from middle school, I was studying.' (GetMarOOl/002) In (75) the speaker is taJking about a specific span of time during which her parents urged her to get married. It was during that time that she was studying. I have few examples of ze or ze d$i"Pfunctioning as auxiliaries in intransitive clauses. This supports the cross-referencing function of animacy as one would not expect to fmd inanimate arguments serving 39
Tense and aspect in Dongwang are not discussed in detail in this paper.
ANIMACY IN DONGWANG TIBET AN
167
as actors in clauses. The only clauses with an inanimate auxiliary form I have found are those in which the argument is a nominalised verb as in (76) below: (76) Disjunct-Declamtive: 1st person, intransitive, -an, ze @?
s ga13
nc4P13
1s go 'I can't go.'
ts:J NMLZ
ze ~?
ma
re
EX.IN.DISJ
NEG
DISJ
In (76), although the S argument is animate and would be expected to take a conjunct form, the auxiliary is inanimate and disjunct This suggests that it is the nominalised verb, rather than the animate actor, which is being indexed. 40 This hypothesis is further supported by the fact that while the verb nc4P 13 'to go' is controllable, the nominalised construction indicates that it is not controllable by the S argument It would be nice if it turned out that in transitive clauses which contain an auxiliary constructed from an existential, the 0 argument controls the auxiliary with respect to animacy. At ftrst glance, this in fact appears to be the case. (77) Conjunct: transitive clause. 1st person A. -an 0. ze A
0
bie 53 ze again one frre up bum EX.IN.CONJ ' ... and again one (of us) kept the frre burning.' (Hardship 110) zc513
f(:i 53 ni 13 z:J 13
(78) Conjunct.· transitive clause, 1st person A. -an 0, ze 0 tsa 5~ba 53
t¢
f(:ha 13
ze
tsampa INDEF eat EX.IN.CONJ ;(We) ate some tsampa. ' 41 (MyLife035) Both (77) and (78) have animate A arguments, but inanimate 0 arguments. It happens that in my data the most frequent auxiliary choice is 40
This could also be analyzed as a possessor clause ('I do not have a go'), but to
the best of my knowledge never occurs with a conjunct auxiliary. More explicit analysis of this construction will be provided in a future paper. 41 Tsampa, a staple for many Tibetans, is made from barley that is roasted and then finely ground It is often mixed together with butter tea to form a cookie-doughlike ball.
168
ELLEN BARTEE
the inanimate form, perhaps for the reason that all such examples have inanimate 0 arguments. The reason for this is related to the animate form being chosen for intransitive clauses: all the examples in my database which co-occur with existential auxiliaries have inanimate 0 arguments.42 Indeed, the fact that there are no counter-examples (e.g., animate 0 arguments with an inanimate existential form) strongly supports the observation that the animacy of the 0 argument possibly controls the choice of auxiliary. However, this is only a tentative analysis.
Subordination The use of an auxiliary in subordinate clauses is optional. When an auxiliary derived from an existential is used in non-complement clauses, only conjunct forms are found, thus neutralizing the c/d distinction, but the animacy distinction is still maintained. This suggests that the conjunct form is the default form as to epistemic source, but that there is no default form where animacy is concerned. (79) Conjunct: sub-clause, exist, Jrd person +an. ndo
s
de 13 IJ:} ndo raJ 55 sit NMLZ EX.AN.CONJ when
}a 13 n:J me ana good NA CONJ.NEG MOD 'When there is someone who lives in Zhongdian (it) is good, right?' (for our business purposest3 (DCWormGrass074) (80) Conjunct: sub-clause1 exist, Jrd 125ZISon -{!g ze (obl) na 53 sky
s
=n::~
=NA
fii53 clouds
ze
rress
EX.IN.CONJ
when
thii353 hn3 moon NEG see EVI.VIS 'When there are clouds in the sky, you can't see the moon.' J::~55Jl:e53
ma
42 While I have many clauses with transitive verbs such as 'kill', 'hit', 'look at', 'listen to', etc., none of them co-occur with the existential auxiliary. This is probably due to the relatively small size of my database (2,200 clauses) rather than any particular restrictions on such co-occmrences. 43 The 'subject' of (79) is the whole nominalised clause .$C 13d055 di 13 dc 13n{) 'live here in Zhongdian (one)'
169
ANIMACY IN DONGWANG TIBET AN
(81) Conjunct· sub-clause, 2nd person posr. -an, posd, ze POSR
POSD
pJxJ13cf¥e53
~55
~i13w0 ss
ze
ne 55
hereafter
2s
question
EX.IN.CONJ
SUB
!]813
tsi53
~
look for IMP 'In the future when/if you have problems, come find me.' 1S.DAT
In (79) through (81) above, there is no c/d distinction in the subordinate clause even though the clauses involve non-frrst-person arguments. However, the animacy of the S argwnent in each of the subordinate clauses is still reflected in the speakers' choice of an animate or inanimate conj1mct form.
Other subsystems I have shown that animacy is a crucial parameter in existentials and in the periphrastic auxiliary forms which are constructed from existentials. This might lead one to propose that animacy may be extended through the language in other subsystems as well. This appears to be the case in a few classifiers. For example, there are two classifiers used for 'pair' or 'two of something': (f:ha 53 -AN, and gu 55+AN. There is also some indication of an animacy distinction in the pronouns, but my examples are too few to say for certain. Given the fact that animacy has not been described as a component in any other c/d system, or for any other Tibetan dialect, the question arises as to why it plays such an important role in Dongwang Tibetan. The conspicuous absence of any mention of an animacy split in descriptions of historical Tibetan grammar indicates that this innovation is most likely not due to genetic inheritance. Tournadre (2001) does not mention this feature of animacy in his overview of Literary Tibetan and contemporary spoken dialects of Tibetan.44 Nor does Beyer (1992) mention this feature in his description of Classical Tibetan. In the next section, I will suggest that this feature is a result of language contact.
44 Tournadre has collected data from a vast mnnber of dialects, but as far as I know, he did not collect data in 1he Sou1hern Kham area of Yunnan Province.
170
ELLEN BARTEE
3. PoSSIBLE ORIGINS OF ANIMACY IN DONGWANG 3.1. Contact-induced change It should not be surprising that Dongwang would have unique characteristics distinct from other varieties of Tibetan. 45 One would expect that languages which are spoken far away from population centers and communication arteries might well develop differently than those within population centers. As part of the southern Khams Tibetan group, Dongwang speakers have lived far away from the 'centers' of Tibetan political and religious influence, yet have been in frequent contact with other language groups in northwest Yunnan. Thomason and Kaufinan (1988: 35) state that it is ''the sociolinguistic history of the speakers, and not the structure of their language, that is the primary determinant of the linguistic outcome of language contact. Purely linguistic considerations are relevant but strictly secondary overall". Dixon ( 1997) argues that languages might have long periods of relatively slow change, punctuated occasionally by dramatic changes due to a complex bundle of factors such as language attitudes, degree of isolation or contact, degree of bilingualism, and structural differences between languages. In order to investigate the 'history' ofDongwang speakers, it might be beneficial to ask certain questions: What possible source languages are there? What social factors might be involved (intermarriage, language attitudes, dominant culture, etc.)? Is the animacy split likely to be a result of borrowing or shift? What other subsystems have evidence ofcontact-induced change? The identification of a source language requires a plausible account of enough contact with another language in order to hypothesise that innovations could have developed through contact. Obviously, the suggested source language should have, or have had, the features that are under consideration. Social factors such as intermarriage, bilingualism and the nature of the contact must be considered as well.
45
Some have suggested that varieties which exhibit great differences may not even
be Tibetan. Sun Hongkai has suggested that Baima, spoken in Amdo, exhibits enough pronounced differences that, in spite oflexical similarity, it should not be considered a dialect of Tibetan. However, as Jackson Sun pointed out, and as is true for Dongwang,
such distinctive characteristics should be expected from a dialect spoken in an area of great linguistic diversity.
ANIMACY IN DONGWANGTIBETAN
171
Although Thomason acknowledges that predictions of language change are probabilistic, not deterministic, she has observed that in most cases of borrowing, there is initially more lexical borrowing than structural borrowing. Contrastively, in most cases of shifting, phonological and syntactic changes are common while vocabulary lags behind (Thomason 2001: 76fl). She concludes: "if we can establish significant structural interference, but there are few or no loanwords, then the interference must have come about via imperfect learning of a target language during shift, not through borrowing" (Thomason 2001: 80). LaPolla discusses the important role that culture and cognition play when speakers begin to learn the language of another culture. When speakers learn some aspect of another language, they may begin to construe the world around them in the same way as the other group, which in turn can lead to the spread of certain constructions or patterns in the speakers' frrst language (LaPolla 1999: 15, 16). Thomason and Kaufman stress that they have found no instances of completely isolated structural interference in only one subsystem of a language (1988: 62). Hopper and Traugott point out that although independent words and morphemes are the most typical items to be borrowed (citing Weinreich 1953), ''very occasionally" whole paradigms can be borrowed (Hopper and Traugott 1993: 210). They cite the Mednyi Aleut case discussed in Thomason and Kaufman (1988: 20) in which Mednyi Aleut had Russian "ftnite verb morphology but with other largely Aleut grammar and vocabulary" (Hopper and Traugott 1993: 210). Thomason agrees that an implicational hierarchy of interference (ftrst borrow words, then grammar) is not the case in instances where ''the people responsible for the innovations are second-language learners of the receiving language: in these cases ... the ftrst and most significant interference features are structural, not lexical" (Thomason 2001: 64). In the following section I argue that it is quite possible that the animacy distinction found in Dongwang is an areal feature which has developed as a result of contact with Naxi speakers. The history of Southern Khams Tibetans is one of frequent and often tumultuous contact primarily with the Chinese and the Naxi. There are Naxi villages in Tibetan areas whose inhabitants are bilingual in Tibetan and Naxi. Crucially, the language of Naxi speakers with whom Tibetans had contact appears to be the only one with the type of animacy features discussed in this paper.
172
ELLEN BARTEE
3.2. Tibetans ofNorthwest Yunnan 3.2.1. Southern Khams While there have been no descriptions of Tibetan dialects which have the animacy split I have described for Dongwang, it is not the case that other related varieties do not exhibit this split. In fact, there are other dialects in the Southern Khams region which distinguish animacy in this way as well. Preliminary research indicates that the animacy distinction I have described for Dongwang is also reflected in the c/d system of Dechen Tibetan,46 with one crucial difference: animate referents can trigger animate or inanimate forms, but inanimate referents can only trigger inanimate forms. 47 This suggests that the split in Dechen Tibetan might be animate versus neutral, rather than animate versus inanimate. While the disjunct morpheme is different from that of Dongwang, the motpheme which distinguishes animacy is from the same historical source. It also is the case that rGyalthang Tibetan auxiliaries reflect an animacy split. Researchers categorise the varieties of Tibetan spoken in NW Yunnan as part of 'Khams', one of the three large dialect groupings of Tibetan spoken in China, but very little research has been conducted to describe this area. 48 In fact, 'Khams' dialects are usually understood to be those varieties spoken in Dege or Chamdo, or maybe the area around Bathang. Due to the lack of descriptive work available, it is impossible to say how far this feature is spread, but it is nearly certain that Eastern Khams, at least those varieties which usually are cited as representative of Khams (in particular Dege, and Chamdo ), do not mark animacy in this way. For example, Gesang Jumian, in Zangyu Fangyan Gailun, 49 includes several Tibetan counties in SW Sichuan province with those spoken in Diqing Prefecture as representing southern Khams dialects (2002: 72). However, he does not mention an animacy distinction and is content to draw examples from Dege dia-
46
Because there has been no sociolinguistic assessment in that area, it is difficult
to ascertain where dialect boundaries may fall. For this reason, I will refer to this dialect as 'Dechen dialect' based on the name of the region. 47 Further study of this dialect may reveal that the forms used for animate referents are not totally interchangeable. 48 Exceptions that I am aware of are Wang Xiaosong and Krisadawan Hongladarom who have written several papers to describe the rGyalthang (Zhongdian) dialect. 49
Jlff9;/j"l!fj618(0verview ofTihetanDialects).
ANIMACY IN DONGWANGTIBETAN
173
lect to typify all of the Khams dialects (2002: 73). Dege distinguishes cld existentials by two forms: }.a and gge (Gesang Jumian 2002: 131fl). There are several possible avenues to explore the origins of animacy indexing in the cld system described for Dongwang Tibetan and other varieties of Southern Khams. If the sociolinguistic history of speakers is the primary determinant of the linguistic outcome in a contact situation as Thomason and Kaufman suggest, then there are questions which must be addressed in order to adequately explore the unique development observed in Dongwang Tibetan, concerning not only the history of contact in NW Yunnan, but also the duration and nature of this contact Crucially, features such as those described here must be (or have been) observed in the language of those with whom contact was made. 3.2.2. Briefhistory ofcontact in NW Yunnan The history of contact in NW Yunnan is fragmentary and tangled, but it is indisputable that Naxi, Chinese and other ethnicities frequently intermingled with Tibetans, often under hostile conditions. Diqing Autonomous Prefecture, located along the historical southern tea route, was the site of many political contests. The tea trade began during the Tang dynasty (618-907) and the 'Tea and Horse Caravan Road' stretched from India and Tibet to SW China, to Lijiang, which has been and still is, a major Naxi city, and up through Diqing Tibetan Autonomous Prefecture (Yang 2004). During the Tang Dynasty, the Yi (Lolo) established the Nan Zhao kingdom which ruled until the Chinese conquered it in the ninth century and the Dali kingdom was established. The Mongols (1206-1367) then conquered the Dali Kingdom with the help of the Naxi king. From then on, the Naxi served as a vassal to help obtain and maintain any ruler's control, all through the Qing dynasty (1644-1911). In 1382, Imperial China used the Naxi king, headed by the famed Mu family, to obtain control of the area and Lijiang was established as the provincial capital. The Mu family continued to rule through the Ming dynasty (1368-1644), occupied rGyalthang for a time, and advanced through NW Yunnan to Tibet (Wang 1995: 55). When the threat of the Mu family to Tibet grew, the Dalai Lama extended Tibet's rule to NW Yunnan (1662-1723) at which time the Mu family sent troops as far north and east as Markham, Bathang and Lithang. Throughout the Qing Dynasty (1644-1911), the Naxi ruled through a
174
ELLEN BARTEE
hereditary system of chieftains. After the Qing Dynasty gained control of the area, Eastern Kham was divided: Bathang and Lithang were annexed to Sichuan and Dechen and rGyalthang were annexed to Ywman. Due to the decaying power of the empire, Chinese control over Tibetan areas had weakened considerably. Many of the Tibetan areas were ruled locally by chieftains, many of whom openly derided Chinese authority (Coales 1919: 230). After a Chinese official was killed on his way to Lhasa, General Chao Erh-feng was sent to help reign in the Kham areas. Van Spengen details General Chao's ruthless attacks on several monasteries and regions of Kham. This, combined with renegade soldiers, Tibetan warlords and Naxi factions, helped to contribute to the making of NW Yunnan into a notorious robber haunt at the beginning of the 20th century. It was during this time that ''the dreaded Tongwa" (Dongwang) terrorised neighbouring populations, raiding Naxi territory, and at one point even controlling Zhongdian (van Spengen 2002: 16-19). There is evidence that the Naxi intermarried and settled in Tibetan areas. There are Naxi villages scattered throughout the Tibetan areas and some of these near the Tibetan border are fully bilingual in Tibetan. I interviewed one Naxi man, a fully bilingual speaker of the Dechen dialect of Tibetan who lives in a predominantly Naxi village in the far NW Yunnan near the Tibet border in Dechen County on the West side of the Lancang Jiang (Mekong River). He told me that there are three villages near him that are known as 'Naxi' villages. 50 Residents of these villages speak Naxi at home, but Tibetan outside of the home. Most are Tibetan Buddhists who live in Tibetan-style houses, wear Tibetan-style clothes and name their children Tibetan names rather than Naxi names. All are fully bilingual, but his impression is that the younger generation seem to be losing Naxi. Finally, when asked if he knew when Naxi immigrants ftrst arrived in his village, he responded that no one knows for sure, but "it is said that perhaps" they came during the period of conquest by the Mu family. According to the brief history just outlined above, this would be sometime in the early 17th century.
50
Personal interview, March 3, 2005.
ANIMACY IN DONGWANG TIBET AN
175
I examined grammatical descriptions of Yi, Bai, Nu, Lisu, Mongolian and Qiang as possible source languages. 51 None of these have the features that I have described for Dongwang Tibetan. While animacy is one parameter that determines how 0 arguments are treated in Qiang (LaPolla and Huang 2002), the patterns are quite different52 and it is highly unlikely that Qiang would exert influence on Southern Khams while failing to do so on its immediate Eastern Khams neighbours in Sichuan. Naxi and Pumi (Primi), however, make very similar distinctions as those attested in Dongwang. In the following section, I discuss briefly the Naxi and Pumi data.
3.2.3.Naxi Naxi, which is traditionally classified as close to Loloish in the LolaBurmese branch, 53 has about 290,000 speakers who live mainly in the Yulong Naxi Autonomous County to the south of Shangri-la County. Naxi does not have a c/d system, but existential morphemes show a four-way contrast. One contrast correlates to the animacy of the referent, whether in a one-argument clause (correlating with the S argument) or in a possessive clause (correlating with the possessed argument). This is indicated by tonal and consonantal contrasts. The following data comes from the Naxiyu Jianzhi, 54 a sketch of the Naxi language, ( 1985):
51 These are all languages which have had some contact with the Tibetans of Shangri-la. Mandarin is also a possible somce language, but clearly lacks the featmes described in this paper. 52 There are five existential verbs in Qiang. Their use depends on the semantics of where and how the referent exists (e.g., contained, moveable, strength). Animacy is one of the semantic parameters of existential verbs, but the main difference is that the animacy of the 0 argwnent is indicated by casemarking on the NP. 53 I am indebted to Alexis Michaud (personal communication) for directing me to David Bradley's 1975 finding that Naxi is significantly different from Lolo languages and may not even be part ofLoloish, even though they are spoken in roughly the same location It is fair to say, as he points out, that Naxi and Pumi may not be genetically distant. 54 He Jiren and Jiang Zhuyi, eds. The full title is l?ii!ffilf/ifjlf (A sketch of the Naxi language). China has published sketches of all the minority languages. Each one typically contains a brief phonological and grammatical description.
176
ELLEN BARTEE
(82) Naxi: Decl, Jrd person posr. +an posd, ndF 1
s
$i33gy33dy3I ¢33 ni33w?I ggv33 mcu33 nd$Y3I Lijiang person twenty nine thousand EX.AN 'Lijiang County has 290,000 people.' (Naxiyu Jianzhi: 91)
(83) Naxi: Decl, Jrdperson posr. +an posd, ndF 1 POSR
POSD
thai3 Jcul:iyw33 J1i33 phuss ncipr3I 3PL cow two head EX.AN POSR
POSD
ga33 Ja33 dw33 phuss ncipr3I 1s also one head EX. AN 'Their household has two head of cows; our 5 household also has one head.' (Naxiyu Jianzhi: 52)
(84) Naxi: Decl, Jrd person posr. -an posd, POSR
clF3
POSD
thcu33 ba33Ja3I ~ws dcu33 thoss cipr33 3s clothes new one set EX.IN 'She has a set of new clothes.' (Naxiyu Jianzhi: 58)
(85) Naxi: Interrogative, 2nd person posr, -an posd, POSR
drr
3
POSD
u33
the33yw33 ze33tcJ3I cipr33 Je33 2s book many EX.IN Q 'How many books do you have?' (Naxiyu Jianzhi: 102)
In (84) and (85) the existential form ~1 is used when the single argument or the possessed argument (Jcu 13yw33 'cow' and ~i 33 'person') is animate, but in (86) and (87) the existential form ~3 is used when the single argument or the possessed argument ( ba 33la 31 'clothes'
55
First and third person singular pronmms ir1 Naxi are .ya 33and thr.u 33 (Naxiyu Ji-
anzhi: 51).
ANIMACY IN DONGWANG TIBET AN
177
and tkl 3-jr£U33 'book') is inanimate. 56 This parallels the Dongwang Tibetan system presented above. 57 I found no examples of locational statements in the Naxiyu Jianzhi, but apparently the animate form would be used for a sentence such as 'My wife is in the house', but a different verb would be used for a sentence such as 'My bike is at home' For these sentences, one would have to say 'My bike is placed/parked/stored at home'. 58
3.2.4.Pumi Pumi belongs to the Qiangic sub-branch of the Tibeto-Bunnan language family. It is spoken by about 35,000 speakers59 mainly in the SE of Shangri-la County. Little is recorded of their history with Tibetans. The data I have on Pumi indicates that the locational and possessive existentials reflect an animacy split: bbonggf 'to have -AN' and xxiuf'to have +AN'. There is no indication of a c/d distinction. 60 4. DISCUSSION Given the above data, the historical background and the current context, it is quite possible that the animacy distinction observed in Dongwang and certain other Southern Khams Tibetan dialects has developed as a result of language contact The fact that both Naxi, a LolaBurmese language, and Pumi, a Qiangic language, reflect this distinction suggests that this could be an areal feature. The differences in the morphological shape of the animacy markers between Dongwang and Naxi and Pumi indicate that Dongwang developed its animacy markers as a result of contact, but clearly did not borrow the morphological 56 When discussing dialects, the authors of the No:xiyu Jianzhi list the animate existential form for the Eastern variety of Naxi as ti$Y1~ so the animacy distinction is triggered by a tonal contrast alone (p. 112). 57 A finther contrast which Naxi makes, not observed in Tibetan, is between contained and non-contained referents. This seems to be a finther delineation of inanimate arguments, but the authors do not expressly say so. All of the examples in the No:xiyu Jianzhi used the locative particle lo 31 'inside':
(i)
khaJJ
Jd1
fla31
nza33
aJJ
ditch inside water NEG EX.IN.CONTAINED 'There is no water in the ditch.' (N axiyu Jianzhi: 52) 58 Thomas Pinson, personal communication 59 This information comes from the Grimes and Grimes (2004). 60 Jo Chan, personal comrmmication The spelling system is her personal practical orthography which she was using for transcribing her data.
178
ELLEN BARTEE
forms from Naxi. Aikhenvald and Dixon (2003b) suggest that when a significant number of speakers of one language have increasing competence in another language, the languages start to resemble each other. For example, if a language with no noun classes were to come into contact with a language that has noun classes, ''then it is likely to develop its own set of noun classes; most frequently it will achieve this not by borrowing the forms for marking noun classes from a neighboring language, but by developing them from its own internal resources" (Aikhenvald and Dixon 2003b: 2). It appears that the Dongwang animacy distinction may have arisen in this way. The fact that this animacy feature has been incorporated into the c/d system is interesting for several reasons. While animacy in Dongwang is expressed in the c/d system, it is unrelated to typical features of c/d systems. C/d systems typically correlate in some way with the intentionality, volition, control and evidence of S and A arguments. By contrast, Dongwang indicates that speakers are indexing the animacy of S and 0 arguments. This supports the contention that the c/d system in Tibetan is functioning in a cross-referencing capacity; in spite of its pragmatic nature where S and A arguments are concerned, animacy seems to consistently hold for the S or POSD argument. 5. CONCLUSION I began this paper with a descriptive overview of c/d systems in general and in Tibetan in particular. I then outlined how Dongwang is different from other varieties of Tibetan, focusing particularly on the animacy split While animacy is observed mainly in the existential verbs and their functions as auxiliaries, it is also present in other aspects of the language including relative clause construction, subordination and pragmatic skewing. Additionally, there is some indication of animacy spreading to other subsystems, such as classifiers and pronouns. This depth of integration indicates that the animacy split may have developed some time ago. The lack of such a system in other dialects of Tibetan combined with the presence of a similar system in nearby languages suggest that this is an areal feature which may have developed initially as result of contact with Naxi speakers. Finally, the presence of such divergent features as the animacy split in the c/d system of Dongwang points to the possibility that the c/d system is func-
ANIMACY IN DONGWANG TIBET AN
179
tioning in more of a cross-referencing fashion than has been reported for other dialects.
ABBREVIATIONS flrst person second person 3 third person A most agent-like argument in transitive clause AN animate cc complement of copular clause COMPL complement CONJ conjunct COP copula DAT dative DET determiner DISJ disjunct EGO egophorique ERG ergative EVI evidential EX existential FOC focus FUT future HON honorific HS hearsay evidential IN inanimate 1 2
10 IPFV
LOC MOD NEG NMLZ
0
OBL PFV PL POSD POSR PST Q QUOT
s
s VIS
WrT
indirect object imperfective locative modal negative nominaliser most patient-like argument in transitive clause oblique argument perfective plural possessed argument of possessive clause possessor argument of possessive clause past question word or particle quotative single argument in intransitive clause singular visual evidential Written Tibetan
BIBLIOGRAPHY Aikhenvald, Alexandra Y. and R.M.W. Dixon, eds. 2003a. Evidentiality. (Studies in Evidentiality). Amsterdam: John Benjamins. Aikhenvald, Alexandra Y. and R.M.W. Dixon, eds. 2003b. Areal di.f/Usion and genetic Inheritance: problems in comparative linguistics. Amsterdam: John Benjamins. Anderson, Lloyd B. 1986. Evidentials, paths of change, and mental maps: typologically regular asymmetries. In Wallace L. Chafe and JohamJ.a Nichols, eds., Evidentiality: The Linguistic Encoding a/Epistemology. NOIWood, NJ: Ablex, 273-312. Bartee, Ellen. 2007. Dongwang Tibetan. PhD dissertation presented to the Department of Linguistics, UC Santa Barbara. Beyer, Stephen, V. 1992. The Classical Tibetan language. New York: SUNY Press. BracD.ey, David 1975. Nahsi and Proto-Bmmese-Lolo, Linguistics of the TibetoBurmanArea 2.1: 93-150.
180
ELLEN BARTEE
Chafe, Wallace L. and Nichols, Joharma, eds. 1986. Evidentiality: The Linguistic Encoding ofEpistemology. Norwood, NJ: Ablex. Coales, Oliver. 1919. Eastern Tibet. The Geographical Journa/53.1: 228-249. Connie, Bernard. 1998. Rethinking the typology of relative clauses. Language Design 1.1: 59-86. Curnow, Timothy Jowan 2001. Why 'first/non-first person' is not grammaticalized mirativity. In Proceedings ofALS2k, the 2000 conference of the Australian Linguistic Society. http://www.als.asn.au, 1-10. - - . 2002. Evidentiality and me: The interaction of evidentials and first person. In Proceedings of the 2001 Conference of the Australian Linguistic Society. http://www.als.asn.au. 1-10. - - . 2003 Verballogophoricity in African languages. In Proceedings of the 2002 Coriference of the Australian Linguistic Society, 1-8. DeLancey, Scott. 1985. Lhasa Tibetan evidentials and the semantics of causation Berkeley Linguistic Society 11: 65-72. - - . 1986. Evidentiality and volitionality in Tibetan. In Wallace L. Chafe and Johanna Nichols, eds., Evidentiality: the linguistic encoding of epistemology. Norwood, NJ: Ablex, 203-13. - - . 1997. Mirativity: the grammatical marking of unexpected infonnation Linguistic Typology 1: 33-52. - - . 2001. The mirative and evidentiality. Journal ofPragmatics 33: 369-82. Denwood, Philip. 1999. Tibetan. Jolm Benjamins. Dixon, RM.W. 1997. The Rise and Fall ofLanguages. Cambridge University Press. Epstein, Lawrence, ed 2002. Khamspa histories: visions ofpeople, place and authority. Proceedings of the ninth seminar of the IATS. Vol. 4 Leiden: Brill. Garrett, Edward. 2001. Evidentiality and assertion in Tibetan. PhD dissertation presented to the linguistics department at UCLA. Genetti, Carol. 1986. The development of subordinators from postpositions in Bodie languages. Paper presented at The twelfth annual meeting of the Berkeley Linguistic Society, Berkeley. - - . 1994. A descriptive and historical account of the Dolakha Newari dialect. Tokyo: Institute for the Study of Languages and Cultures of Asia and Africa. Gesang Jumian and Gesang Yangjing, eds. 2002. MiH;/J7!fMi:8 Zangyu fangyan gailun ~n overview ofTibetan dialects). Beijing: Minorities Publishing House. GivOn, Talmy. 1982. Evidentiality and epistemic space. Studies inLanguage 6: 23-49. Gordon, Lyrm. 1986. The development of eviden1ials in Maricopa. In Wallace L. Chafe and Johanna Nichols, eds., Evidentiality: the linguistic encoding of epistemology. Norwood, NJ: Ablex, 75-88. Grimes, Barbara F. and Joseph E. Grimes, eds. 2004. Ethnologue: languages of the world 14th edition Dallas: SIL. www.ethnologue.corn Hale, Austin 1980. Person markers: finite conjmtet and di.sjmtet verb forms in Newari. In Ronald Trail, ed, Papers in South-East Asian Linguistics 7, Canberra: Pacific Linguistics, 95-106. Haller, Felix. 2000. Verbal categories of Shigatse Tibetan and Themchen Tibetan. Linguistics ofthe Tibeto-BunnanArea 23.2: 174-188. Hargreaves, David 1990. Indexical ftmctions and grammatical subsystems in KathmanduNewari. Chicago Linguistic Society 26: 179-193. - - . 1991. The conceptual structure of intentional action: data from Ka1hmanduNewari. Proceedings ofthe seventeenth annual Berkeley Linguistics Society 17: 379-389.
ANIMACY IN DONGWANG TIBET AN
181
- - . 2005. Agency and intentional action in Kathmandu Newar. In Himalayan Linguistics 5: 1-48. http://www.lingui.stics.ucsb.edu/HimalayanLinguistics/articles/2005/ HLJ05_Hargreaves_agency. pdf Hasler, Ka1rin Louise. 1999.A grammar of the TihetanDege (Sde dge) dialect. Ziirich: Selbstverlag. _ _ . 2001. An empathy-based approach to the description of the verb system of the Dege dialect ofTibe1an. Balthasar Bickel (ed.). Person and evidence in Himalayan languages. Special issue oflinguistics ofthe Tiheto-Bwman area. 2. 24.1: 1-34. He Jiren, Jiang Zhuyi, eds. 1985. l?fi!fffG{Jfjlf Naxiyu Jianzhi (A sketch of the Naxi language). Beijing: Minorities Publishing House. Hongladarom, Krisadawan. 1996. rGyalthang Tibe1an ofYurman: a preliminary report. Linguisticsofthe Tiheto-BurmanArea 19.2: 69-92. - - . 2000. Indexical categories in Kham Tibe1an and Central Tibe1an. Paper presented at The 6th Himalayan Languages Symposium, University of Wisconsin, Milwaukee, June 15-17. - - . N.d. The Khampas of Tibet's eastern frontiers: language, identity, and etlmohistory. Paper presented at the 2nd Amtual Meeting of the Asian Scholarship Foundation, 1-2 July 2002, Bangkok. Hopper, Paul J. and Elizabeth Cross Traugott. 1993. Grammaticalization. Cambridge: Cambridge University Press. Hu Tan, ed. 1986. Lasa kouyu duhen. (colloquial Lhasa reader) Beijing: Minorities Publishing House. Huber, Brigitte. 2000. Preliminary report on evidential categories in Lende (Kyirong) Tibe1an. Linguistics ofthe Tiheto-BwmanArea 23: 155-174. LaPolla, Randy. 1999. The role of migration and language contact in the Sino-Tibe1an language family. In Aikhenvald, Alexandra Y. and R.M.W. Dixon, eds., Areal diffosion and genetic inheritence: case studies in language change. Oxford: Oxford University Press. LaPolla, Randy and Chenglong Huang. 2002. The copula and existential verbs in Qiang. Paper presented at Workshop on copula clauses and verbless clauses held at the Research Centre for Linguistic Typology. Matisoff, James A. 1993. Sangkong ofYwman: secondary ''verb pronominalization" in Southern Loloish. Linguistics ofthe Tiheto-BurmanArea 16.2: 123-142. Sch.Ottelndryer, Bmkhard. 1980. Person markers in Sherpa. In: Ronald Trail, ed, Papers in South-East Asian Linguistics, No.7. Canberra: Pacific Linguistics, 125-130. Slater, Keith W. 1998. A Grammar ofMangghuer: A Mongolic language of China's Qinghai-Gansu sprachhund. PhD Dissertation presented to the department of linguistics at UC Santa Barbara. Sun, Jackson. T.-S. 1993. Evidentials inAmdo-Tibe1an. Bulletin ofthe Institute afHistory and Philology, Academia Sinica 63-64: 945-1001. - - . 2001. Phonological profile of Zhongu: a new Tibe1an dialect of Nor1hern Sichuan. Paper presented at The Workshop on Tiheto-BwmanLanguages, UC Santa Barbara. Timrgood, Graham. 1986. The nature and origins of the Akha evidential system. In Wallace L. Chafe and Johanna Nichols, eds., Evidentiality: the linguistic encoding ofepistemology. Norwood, NJ: Ablex, 214-221. Timrgood, Graham and Randy J. LaPolla, eds. 2003. The Sino-Tibetan languages. London: Routledge. Thomason, Sarah Grey and Terrance Kaufman 1988. Language contact, creolization, and genetic linguistics. Berkeley: University of California Press.
182
ELLEN BARTEE
Thomason, Sarah Grey. 2001. Language contact. Washington, D.C. Georgetown University Press. Toland, Norma R. & Donald F. Toland 1991. Reference grammar of the Karo!Rawa language (Data Papers on Papua New Guinea Language 38). Ukarumpa: Smnmer Institute of Linguistics. Toumadre, Nicolas. 2001. Final auxiliary verbs in Literary Tibetan and in the dialects. Linguistics ofthe Tibeto-BunnanArea 24.1: 49-110. Toumadre, Nicolas and Sangda Dorje. 2003. Manual ofStandard Tibetan. New York: Snow Lion Publications. van Driern, George L, with Karma Tsering (Collaborator). 1998. Dzongkha. Languages ofthe Greater Himalayan Region, 1 Leiden: CNWS. van Spengen, Wim, 2002. Frontier history of Southern Kham: Banditry and war in the multi-ethnic fringe lands ofChatring, Mili, and Gyethang. 1890-1940. In Epstein, Lawrence, ed., Khamspa histories: visions of people, place and authority. Proceedings ofthe ninth seminar ofthe IATS. Leiden: Brill, 4.7-30. Vasquez de Ruiz, Beatriz. 1988. La predicacwn en Guambiano (Lenguas Aborigenes de Colombia, Descripciones 2). Bogota: Centro Colombiano de Estudios en Lenguas Aborigenes, Universi.dad de los Andes. Vesalainen, Olavi and Ma.Ija Vesalainen. 1980. Clause patterns inLhomi. Pacific Linguistics B, 53. Canberra: Australian National University. Wang Hengjie. 1995. i!h&M/i!ti:f
THE SMAPANG VERBAL AGREEMENT SYS1EM RENE HUYSMANS
Sampang is a Kiranti language spoken in eastern Nepal, in the northeastern part of Khot:aflg district and in a few pockets in Bhojpur district. The morphology of Sampang simplex verb forms clearly indicates that Sampang is a Kiranti language, where transitive verbs show agreement both with the agent, i.e. transitive subject, and the patient, i.e. undergoer, recipient or beneficiary. Sampang intransitive verbs agree with their only argument, the subject, i.e. intransitive or reflexive subject Within Tibeto-Burman, the Kiranti languages stand out on account of their complex verbal paradigms, a feature which explains the epithets of 'pronominalised' or 'complex pronominalising' commonly used to characterise them. The term describes a process whereby verbal afftxes derive from ancient independent pronouns. Pronominalisation in Kiranti languages was first attributed to substratal influence from an Austroasiatic Munda language. Most linguists accepted the Munda substratum hypothesis until the question was finally settled by Bauman (1975), who convincingly showed that the complex transitive conjugations of the Kiranti languages do not correspond in any way to the relatively straightforward conjugations attested in Munda languages. Furthermore, any Munda substratal influence on Tibeto-Burman must have been at a sufficiently remote time, probably dating back to Proto-Austroasiatic, while the incorporation of pronouns into the Munda verb is considered to be a much more recent development Today, the Munda substratum hypothesis has been largely discarded and the pronominalisation attested in Tibeto-Burman languages has been reconstructed to the Proto-TibetoBurman level by van Driem (1993a). A Sampang simplex is a non-periphrastic indicative verb form marked for tense, person, number and grammatical relations of the arguments to the verb. A simplex may or may not be negated and may be marked for reflexive voice. The morphology of Sampang simplicia is elaborate and contains several fused or portemanteau morphemes. As in most other Kiranti languages, the Sampang verb shows copy morphemes in a number of
184
RENE HUYSMANS
forms. In total, there are eleven pronominal categories in Sampang, as is common for a Kiranti language. There is singular, dual and plural nwnber for frrst, second and third person actants and, in addition, nonsingular frrst person actants can be inclusive or exclusive. Many morphemes have different allomotphs and sometimes zero allomotphs. For the reader's comfort, instead of providing the full allomotphy each time I mention an affix, I shall choose one allomotph as the basic one. Unless the discussion at hand concerns a specific allomotph, the reader is referred to Section 3 for a detailed description of the allomOtphic variation of each and every affix in the Sampang verbal agreement system. In this article, the following abbreviations are used: 1, 2, 3 A AUX
d e 1
NG NPT
nls ns np
first, second, third person agent auxiliary dual exclusive inclusive negative non-preterite not first person singular non-singular non-plural
p
p pf PT RFL
s s sf ---+
:L
patient plural prefixal slot preterite reflexive subject singular suffixal slot indicates the direction of a transitive relationship verb stem
1. CONJUGATIONS OF SAMPANG SIMPLICIA Tables 1, 2 and 3 show the affirmative intransitive, reflexive and transitive conjugations of Sampang simplicia. Reflexive forms are expressed with the reflexive discontinuous m01pheme or simulflx <-n -ici> (RFL) occurring immediately after the stem. In reflexive forms with frrst person singular and with second and third person nonplural subjects, the reflexive simulflx <-n -ici> (RFL) is felt to be totally acceptable by Sampang speakers. In all other forms, viz. the reflexive 1di, 1de, 1pi, 1pe, 2p and 3p forms, language consultants tolerate the use of the reflexive simulflx <-n ... -ici> (RFL), but prefer to use the transitive 1di---+3, 1de---+3, 1pi---+3, 1pe---+3, 2p---+3 and 3p---+3 forms instead. The reflexive 1di form and the reflexive 3sd form are homophonous.
185
THE SAMPANG VERBAL AGREEMENT SYSTEM
Table 1
Sampang affurnative intransitive and reflexive simplex conjugations
(in each box the preterite form is listed below the corresponding non-preterite form.) SUBJECf
INTRANSITIVE
REFLEXIVE
ls
~-a
~-nacug
~-a
~-nacug
ldi
~-ici
~-nici
~-aci
~-naci
~-icika
~-nicib
~-acib
~-nacib
lpi
~~ ~-c
~-ncci ~-ncci
lpe
~-cka
~-nccib
~-cka
~-nccib
Ide
2s
~-ina ~-ana
~-nicina
2d
~-icina
~-nacina
~-acina
2p
~-ini
~-nicini
~-ani
~-nacini
~-i
3s
~-a
~-nici
3d
~-ici
~-naci
3p
~-imi
~-nicimi
~-ami
~-nacimi
~-aci
Table2
Sampang affurnative transitive simplex conjugation: ftrSt person patients
(in each box the preterite form is listed below the corresponding non-preterite form.) p a t e n t ls 2s
a 2d
ldi.
Ide
lpi
~-cna
~~kana
~-ana
~-icikana
~-ckana
~~ina
~-acikana
~-ccikana
~-icikanum
~-cbmum
~-Bcli!Jila
g 2p
~-ccikana
~-cni
~ -acikan.um
~-anum
e 3s
n 3d t 3p
lpe
~-cbmum
~~
~-ici
~-icika
~~
~-a
~-aci
~-acika
~~
~~ka
~-ccimug
~-ccikamug ~-ccikamug
~~i
~~ka
~-Bcug
~-icimug
~-icikamll{}
~-ccimug
~-cmi
~-acimug
~-acikatnll{}
~-cmi
~-clamJi
~-cmug
~~katnll{}
~-.imug
186 Table 3
RENE HUYSMANS
Sampang affmnative transitive simplex conjugation: second and third person patients
(in each box the preterite form is listed below the corresponding non-preterite form.) p
ls
a 2s
2d
2p
3
n 3ns
}":.-ni }":.-na
}":...nici }":.-naci
}":.-nini }":.-nani
}:-a }":...UIJ
}:-ina }:-ana
}":.-icina/-nici }":.-acina/-naci
}":.-ini/-nini }":.-ani/-nani
}":.-oCUIJ }":.-u1JCUIJ }":.-ucu }":.-acu }":.-ucuka }":.-acuka }":.-amcwn }":.-umcum/-acwn }":.-amcumka }":.-umcumka }":.-:Jcina }":.-ucina }":.-ucuna }":.-acuna }:-unum
e
t
a ldi lde
g lpi lpe
}:-ina }:-ana
}":.-icina/-nici }":.-acina/-naci
}":.-ini/-nini }":.-ani/-nani
e 2s
}:-lim }":.-um/-a }":.-limka }":.-umka }":.-.:ma }:-una
2d
n 2p
t
}:-anum
3s t
3d 3p
}:-ina }:-ana
}:-imina }":.-amina
}":.-icina }":.-acina
}":.-ini }:-ani
"}":.-~
}":.-~ci
}":...u
}":.-uci }":.-ucu }":.-acu }":.-:Jci }":.-uci
2. COMPETING ANALYSES: TENSE, NUMBER AND COPY MORPHOLOGY
The intransitive 3s form in the non-preterite tense shows a morpheme <-i>, about which two alternative hypotheses can be formulated: the morpheme <-i> can be interpreted as a non-preterite tense marker (analysis A), or it can be interpreted as a number morpheme which marks singular number of a second or third person actant (analysis B). In analysis A, represented in Table 4, the morpheme <-i> is analysed as a non-preterite tense marker (NPT). The advantage of analysis A is that the number morphemes and the second part of the simul:fix can be segmented as <-ci> (d), <-ci> (3nsAP), <-ni> (2p), <-m.i> (3nsAS) and <-ci> (RFL), with relatively straightforward allomorphic rules, subsumed under the phonological rule that vocalis ante vocalem corripitur, which governs deletion of a vowel before another vowel.
THE SAMPANG VERBAL AGREEMENT SYSTEM
187
The disadvantage of analysis A is that it is more common for a Kiranti language to lack a non-preterite tense marker, as is the case in Limbu, Hayu, Thulung, Bahing and Kulung. Moreover, most Kiranti non-preterite tense markers are a stop, such as Dumi <-t> (van Driem 1993b) and Lohorung <-k> (van Driem 1992), whereas the putative Sampang non-preterite tense marker <-i> is vocalic and difficult to relate to the reconstructed Proto-Kiranti non-preterite tense mmpheme *<-k> (see Section 5). From a Kiranti perspective, analysis B has, therefore, greater verisimilitude because it does not postulate the existence of a non-preterite tense marker. Finally, analysis B is more parsimonious in positing just two suffixal slots, whereas analysis A must presume three functional positions for the same set of suffiXal morphemes.
Table4
Analysis A
-ru 2p
-C1
d -mt
-1
NPT
3nsAS
... -C1
RFL -C1
3nsAP sf-a
Table5
sf-c
sf-b
Analysis B -1
-m1
23s
2p
-1C1
-tm1
d
3nsAS
... -1C1
-1C1
RFL
3nsAP
sf-a
sf-b
According to analysis B, the motpheme <-i> is intetpreted as a second or third person singular number marker (23s ), and the dual morpheme <-ici> (d), the third person non-singular agent or patient morpheme <-ici> (3nsAP), the second person plural morpheme <-ini> (2p), the
188
RENE HUYSMANS
third person non-singular agent or subject morpheme <-irni> (3nsAS) and the second part of the reflexive simulftx < ... -ici> (RFL) must all be segmented as containing an initial vowel /jj, which cannot be detached and interpreted as a separate morpheme (see Section 3.5 for the allomorphy of these sufftxes ). The second or third person singular number suffix <-i> (23s) can be related to the Dumi non-frrst person singular number sufftx <-a> (s23) (van Driem 1993b: 145-6). On semantic grounds, the second or third person singular number marker <-i> (23s) is assigned to sufftxal slot sf5 together with the other number sufftxes. The next issue concerns the existence of a third person patient copy suffix in Sampang verbal morphology. This issue is closely tied up with the expression of tense in third person patient forms. In ten out of fifteen third person patient tense pairs, the strategy for the expression of tense is to mark the non-preterite forms with the non-preterite third person patient morpheme <-o> (3PINPT) and leave the preterite forms unmarked for tense. The non-preterite third person patient morpheme <-o> (3P/NPT) has the allomorph <-a> before /-m/ and the allomorph <-0> before the zero allomorph of the frrst person singular suffix <-IJ> (1s) (see Section 3.4). But in the 1di--+3, 1de--+3, 2d--+3, 2p--+3, 3d--+3 forms and the alternative 1pi--+3 and 1pi--+3ns forms -a and -acum, another strategy emerges, whereby the preterite form contains the preterite tense marker <-a> (PT) and the non-preterite forms contain no morphological expression of tense. All third person patient forms with a dual agent contain a sequence /cui, which can be analysed as a regular allomorph of a portemanteau morpheme <-icu> occurring after any sufftx consisting of, or ending in, a vowel, expressing a transitive relationship between a dual agent and a third person patient (d--+3). The allomorph <-icu> of this morpheme would be found with a small number of irregular verbs. This is analysis C, which is represented in Table 6.
Table 6
Analysis C -a PT -lCU
-u 3P
d--+3
sf-a
sf-b
THE SAMPANG VERBAL AGREEMENT SYSTEM
189
The problem with analysis C is that the putative portemanteau morpheme <-icu> (d---+3) is too easily segmented in the dual number morpheme <-ici> (d) with a regular allomorph <-c> in intervocalic position, followed by a third person patient suffix <-u> (3P). If this segmentation of the sequence /cui is accepted, then the non-preterite third person patient forms with a dual agent contain two occurrences of a third person patient motpheme <-u>, and it is likely that one of these two occurrences is a copy morpheme. In analysis D, it is assumed that the frrst occurrence of the third person patient suffix <-u> is an anticipatory copy and that the second occurrence is the original. In analysis E, the reverse is assumed to be the case: the frrst occurrence of the third person patient suffix <-u> is the master which is followed by a copy sufftx. Copy motphemes are indicated by an apostrophe (') in the gloss.
Table 7
Table8
Analysis D -u 3P'
-1c1 d
-u 3P
sf-a
sf-b
sf-c
-u 3P
-1c1 d
-u 3P'
sf-a
sf-b
sf-c
Analysis E
Although analysis D has the advantage that the master third person patient sufftx <-u> (3P) is always present, whereas the third person patient copy sufftx <-u> (3P') is only present when the preterite tense sufftx <-a> (PT) is absent, it must be discarded on the basis of the preterite 2s---+3ns and 3s---+3ns/3p---+3 forms -ucina and -uci, which clearly indicate that the first instance of the third person patient motpheme <-u> (3P) must be the original: a form cannot contain a copy when its master is absent. Analysis E is then the only alternative to analysis C. Absence of the third person patient copy sufftx <-u> (3P') in the preterite 2s---+3ns and 3s---+3ns/3p---+3 forms can be regulated by a special condition, which stipulates that the third person patient copy sufftx <-u> (3P') can only be inserted after the dual number motpheme
190
RENE HUYSMANS
<-ici> (d). The question whether or not the ls--+3ns, lpi--+3ns, lpe--+3ns and 2p--+3 forms also contain a third person patient copy suffix <-u> (3P') will be dealt with below. Suffice it to say at this point that these forms all end in a syllable-final nasal /m/ or /rjl. However, analysis E presents the same problem as analysis D: how can it be that, in the presence of the third person patient copy <-u> (3P'), the original third person patient morpheme <-u> (3P) is absent in the preterite third person patient forms with a dual agent? But what if the preterite tense suffiX <-a> (PT) were in fact to be located after the original third person patient suffiX <-u> (3P)? In preterite third person patient forms, then, with a dual or a second person plural agent and the alternative preterite lpi--+3 and lpi--+3ns forms -a and -acum, the original third person patient morpheme <-u> (3P) would be reduced to a zero allomorph through the rule that vocalis ante vocalem corripitur, as in analysis F.
Table 9
Analysis F
-a PT
-lCl
-u
3P
d
3P'
sf-a
sf-b
sf-c
sf-d
-u
In analysis F, selection of the third person patient copy <-u> (3P') occurs before application of the rule that vocalis ante vocalem corripitur, which, at the surface, deletes the original third person patient morpheme <-u> (3P) before the preterite tense suffix <-a> (PT). But even analysis F is not without problems, for in preterite third person patient forms with a first or third person non-dual and second person singular agent, the suffixal string begins with the third person patient suffiX <-u> (3P) and not the preterite tense suffiX <-a> (PT). A possible solution to this enigma is to assume that in these forms the preterite tense suffix <-a> (PT) is not selected at all, so that there is no sequence of vowels upon which the rule vocalis ante vocalem corripitur can operate. And, in fact, in corresponding non-preterite third person patient forms with a first or third person non-dual and second person singular agent, we observe the presence of the non-preterite third person patient morpheme <-o> (3PINPT). This suggests an overarching principle in the Sampang verbal agreement system against redundant tense marking, whereby only one form in a tense parr is marked by a
THE SAMPANG VERBAL AGREEMENT SYSTEM
191
tense morpheme or a tensed portemanteau morpheme and the other form is left unmarked. But what factor determines the selection of either the preterite tense suffix <-a> (PT) or the non-preterite third person patient morpheme <-o> (3P/NPT) in a third person patient tense pair? As already indicated, the default strategy for marking tense in third person patient forms is to use the non-preterite third person patient morpheme <-o> (3P/NPT) and leave the corresponding preterite form unmarked for tense. Linked to the default tense marking strategy is the possibility to express non-singular number of the third person patient through the third person non-singular agent or patient suffiX <-ici> (3nsAP). If the default strategy for tense marking would be applied in the 1di---+3, 2d---+3 and 3d---+3 forms, the third person patient system would be crippled by the homophony of these forms with the 3s---+3ns/3p---+3 and 2s---+3ns forms. Hence, in these forms, the tense marking strategy with the preterite tense suffiX <-a> (PT) emerges, which leaves the non-preterite forms unmarked for tense. In the 1de---+3 forms as well, the preterite tense morpheme <-a> (PT) is found, suggesting that exclusive and inclusive forms can only differ in the presence of an overt exclusiveness suffix <-ka> (ePS) or <-ka> (eA) or a zero inclusiveness suffix <-0> (iPS) or <-0> (iA). Thoughout the sessions of elicitation of transitive conjugations, all Sampang language consultants insisted upon the alternative forms -a for the preterite 1pi---+3 scenario and -acum for the preterite 1pi---+3ns scenario. These alternative forms are exceptional for three reasons. First, in spontaneous speech, the preterite 1pi---+3 and 1pi---+3ns scenarios are never expressed by the alternative forms -a and -acum, but always by the forms -um and -umcum. Second, there are no alternative forms for the non-preterite 1pi---+3 and 1pi---+3ns scenarios and the 1pe---+3 and 1pe---+3ns scenarios in both tenses. And, third, in all third person patient forms which contain the third person nonsingular agent or patient marker <-ici> (3nsAP), preterite tense is left unmarked, but not so in the alternative forms -a and -acum. I shall, therefore, interpret the alternative forms -a and -acum as remnants of an earlier historical stage in the development of the Sampang verbal agreement system. The third person patient copy morpheme <-u> (3P') may not be the only copy morpheme in the Sampang verbal agreement system. In the reflexive 1s forms and the transitive 2d---+ 1s, 3d---+ 1s and 1s---+3ns forms, we observe each time two occurrences of a velar nasal suffiX
192
RENE HUYSMANS
<-lJ>, the first one of which is almost always revealed by the assimilatory nasalisation of the preceding vowel, except in the preterite ls---+3ns form -UIJCUIJ. And in each lp---+3ns form, there are two instances of a bilabial nasal motpheme <-m>. Again, it is likely that in each of these forms, one of the motphemes <-lJ> or <-m> is a copy and one a master. In analysis G, I explore the hypothesis that the posterior suffixes <-lJ> and <-m> are copy suffixes and are grouped together with the third person patient copy motpheme <-u> (3P').
Table 10
Analysis G -o 3PINPT -u 3P
-a PT
-m lpA
-lCl
-g
... -lCl
ls
RFL
-u 3P'
d -lCl
3nsAP
-m lpA' -g
ls' sf-a
sf-b
sf-c
sf-d
sf-e
sf-f
Both suffixes <-lJ> express a first person singular actant ( ls) and the preterite 2p---+ ls form -anwn clearly identifies the anterior occurrence of the first person singular suffix <-lJ> ( ls) as the original suffix, in accordance with analysis G, because it contains a single instance of the suffix <-1)> and a copy cannot be present in the absence of its master. In addition, analysis G is attractive from an aesthetic point of view because all three copy motphemes are clustered together in a single copy slot I shall not intetpret the vowel /u/ in the non-preterite ls---+3ns and lp---+3ns forms and the second vowel /u/ in the preterite ls---+3ns and lp---+3ns forms as the third person patient copy motpheme <-u> (3P'), but rather as the final vowel of a regular allomotph <-cu> of the third person non-singular agent or patient suffix <-ici> (3nsAP) before the syllable-final nasal matphemes <-m> and <-lJ>, where it underwent a process of back assimilation. There are two advantages to analysing the vowel /u/ in this way. First, in the reflexive ls forms and a myriad of first person patient forms, the vowel /u/ does not have a third person patient meaning at all. Second, the selection criteria for the third
THE SAMPANG VERBAL AGREEMENT SYSTEM
193
person patient copy morpheme <-u> (3P') are greatly simplified if this copy morpheme is not present in the ls--+3ns and lp--+3ns forms. Copy morphemes are then selected if their master occurs in the same form, and if a certain number suffJX is present In other words, the third person patient copy morpheme <-u> (3P') needs the dual number morpheme <-ici> (d), which is why it does not occur in the 2s--+3ns and 3s--+3ns/3p--+3 forms, and, as we shall see, the frrst person plural agent copy <-m> (lpA') is supported by the third person non-singular agent or patient suffJX <-ici> (3nsAP), whereas the frrst person singular copy morpheme <-IJ> (ls') can do with either one of these two number morphemes <-ici> (d) and <-ici> (3nsAP), or with the second part of the reflexive simulfJX <-ici> (RFL). But analysis G fails in one essential respect: the hypothesis that the posterior suffJX <-m> is a frrst person plural agent copy suffJX (lPA') must be discarded, because the morpheme <-m> in the 2p--+3 forms -unum and -anum must be identified on economic grounds with the putative copy suffix <-m>, which, therefore, must express a frrst or second person plural agent (12pA). Following the maxim that there can be no copy without its master, it follows that the 2p--+3 forms identify the frrst or second person plural agent suffix <-m> ( 12pA) as the original morpheme and the frrst occurrence of the suffJX <-m> as the copy. Table 11
Analysis H -o
3PINPT -u 3P
-a PT
-m lpA'
-lCl
-m1
d
2p
-g
... -lCl
-lCl
ls
RFL
3nsAP
-m 12pA -u 3P' -g
ls' sf-a
sf-b
sf-c
sf-d
sf-e
sf-f
In analysis H shown above, the frrst person plural agent copy <-m> (lpA') can only occur in the lp--+3ns forms. As indicated by its gloss, the frrst person plural agent copy <-m> (lpA') is absent in the 2p--+3 forms, despite the presence of its master, the frrst or second person
194
RENE HUYSMANS
plural agent marker <-m> (12pA). Apparently, the frrst person plural agent copy <-m> (1pA') can only be supported by the third person non-singular agent or patient suffix <-ici> (3nsAP) and the second person plural suffix <-ini> (2p) cannot perform that function. Analysis H abandons the aesthetically pleasing single copy slot, but analysis G can only be saved by an unmotivated allomorphic rule which, before the second person plural suffJX <-ini> (2p ), deletes a putative frrst or second person plural agent master suffJX <-m> (12pa), a gloss for which the Sampang transitive conjugation provides no grounds whatsoever. 2.1. The Sampang morphological template The remaining affJXes of the Sampang verbal agreement system can be allotted without much further ado. Tables 12 and 13 provide a schematic overview of the results of the morphological analysis of the conjugations presented in tables 1, 2 and 3. All identified morphemes of Sampang simplex morphology are distributed over one prefJXal slot and ten suffJXal slots.
Table 12
Initial portion of the Sampang affJXal string (prefixal slot pfl, stem and suffixal slots sfl to sf3)
-n ... RFL
-a
-n
1sPS
1~2
man- L NG
-u 3P
-a Pf
stem sfl
-lCl
-g
d
1s -e 1pPS
-o 3P/NPf pfl
-e 1sP/NPT
-1
-m 23s 1pA'
sf2
sf3
... -lCl
RFL
sf4
sfS
As can be seen from Table 12, the frrst slot is prefixal slot pfl, which contains the negative prefJX <man-> (NG). After the stem, the suffixal cham begins with sufftxal slot sfl, which contains the portemanteau morpheme <-n> (1 ~2) expressing a transitive relationship between a frrst person agent and a second person patient, and the frrst element
THE SAMPANG VERBAL AGREEMENT SYSTEM
195
<-n> of the reflexive simul:fix <-n ... -ici> (RFL ), which is distributed over suffixal slots sfl and sfS, and, in addition, the third person patient morpheme <-u> (3P) and the non-preterite third person patient morpheme <-o> (3PINPT). This :ftmctional position can be called the reflexive, 1---+2 and third person patient slot The :ftmction of suffood slot sf2 is the expression of tense and it contains the tense morpheme <-a> (PT). Sufftxal slot sf3 is the ftrst person patient or subject slot and is ftlled by the frrst person singular patient or subject morpheme <-a> (lsPS), the non-preterite frrst person singular patient sufftx <-e> (1sPINPT) and the ftrst person plural patient or subject <-e> (1pPS). The next sufftxal slot is the ftrst person slot sf4, and contains the frrst person singular morpheme <-IJ> (1s) and the copy suffix <-m> (1pA'), which expresses a transitive relationship between a frrst person plural agent and a third person patient. The :ftmctional position sfS is the number and reflexive slot and contains in addition to the second element of the reflexive simulfJX <-n ... -ici> (RFL), the sufftx <-i> (23s) expressing singular number of a second or third person and the dual number morpheme <-ici> (d) indexing dual number of any person. Table 13
Final portion of the Sampang afftxal string (suffixal slots sf6 to sflO)
-m 12pA -tnl
-ka ePS
2p
-g
3nsA -nm
-0 iPS
3nsAS
-g
-0 iA
-no NG
1s' -lCl
sf6
-ka eA
3nsAP
-u 3P'
sf7
sf8
-na 2 s:f9
sflO
Suffixal slot sf6 is the exclusive patient and subject slot, which contains the exclusive patient or subject morpheme <-ka> (ePS) and the zero inclusive patient or subject suffix <-0> (iPS). Suffixal slot sfl is the non-singular number slot and is ftlled by the third person nonsingular agent or subject marker <-imi> (3nsAS), the third person non-
196
RENE HUYSMANS
singular agent or patient sufftx <-ici> (3nsAP) and the second person plural morpheme <-ini> (2p). Sufftxal slot sf8 is the non-singular agent and copy slot, and contains the ftrst and second person plural agent sufftx <-m> (12pA), the third person non-singular agent morpheme <-IJ> (3nsA), the frrst person singular copy sufftx <-IJ> (1s') and the third person patient copy sufftx <-u> (3P'). The exclusive agent and second person slot sf9 contains the exclusive agent morpheme <-ka> (eA), the zero inclusive agent marker <-0> (iA) and the second person sufftx <-na> (2). Finally, sufftxal slot sf 10 is the negative slot and contains the negative sufftx <-m> (NG). 3. AFFIXAL SLOTS 3.1. Pre.fzxal slot pfl: negation
<man->
NG
Negative forms are derived from affirmative forms by simple rules. The negative preftx <man-> (NG) negates the lexical verb tuma (vi) 'to be' in all its meanings, i.e. the existential, attributive and locational meanings, and is used for the negative forms of twna (vi) 'to be' in its function as an auxiliary in the perfect tenses. The negative preftx <man-> (NG) is also used for the negation of all non-fmite verb forms. The negative sufftx <-m> (NG) negates any other indicative verb form, including indicative forms of the homophonous verb twna (vi) 'to sit', which is never marked by the negative preftx <man-> (NG). 3.2. Suffzxal slot sfl: reflexive voice, the 1---+2 scenario marker and third person patient
<-n> <-n ... -ici - -ci - -cu> <-o- -6- -a> <-u--0>
1---+2 RFL
3PINPT 3P
The transitive scenario marking motpheme <-n> ( 1---+2) indicates a transitive relationship between a ftrst person agent and a second person patient As can be seen in Table 3, second person patient forms constitute a subsystem in the Sampang transitive simplex conjugation where the expression of an agent is limited to forms with a frrst person agent and
THE SAMPANG VERBAL AGREEMENT SYSTEM
197
the 3p---+2s form. The portemanteau motpheme <-n> (1---+2) and the second person motpheme <-na> (2) are in complementary distribution. The limited expression of an agent in second person patient forms causes homophony of the intransitive 2s forms and the transitive 1e/3np---+2s forms, of the intransitive 2d forms and the transitive n1s---+2d, and of the intransitive 2p forms and the transitive n1s---+2p forms. The motpheme <-n> ( 1---+2) shows an asymmetrical distribution in the transitive paradigm. The suffix <-n> (1---+2) occurs in all 1s---+2 forms, but scenarios with a ftrst person non-singular agent and a second person dual or plural patient can be expressed by 1---+2d and 1---+2p forms with the sufftx <-n> (1---+2) or by n1s---+2d and n1s---+2p forms, which do not contain the sufftx <-n> (1---+2). The distinction between a singular agent and a non-singular agent in 1---+2 forms is more important when a singular second person patient is involved than when a non-singular second person patient is involved. Sufftxal slot sfl also contains the frrst nasal element <-n> of the discontinuous marker or simulftx <-n -ici> (RFL), which expresses reflexivity. The second element <-ici> of this motpheme can be found in sufftxal slot sf5. As can be seen in Table 1, the reflexive simulftx <-n ... -ici> (RFL) occurs in intransitive forms. In addition to the 1---+2 scenario marker <-n> (1---+2) and the frrst part <-n> of the reflexive simulftx <-n ... -ici> (RFL), sufftxal slot sfl also contains the non-preterite third person patient motpheme <-o> (3PINPT) and the third person patient sufftx <-u> (3P). The non-preterite third person patient motpheme <-o> (3P/NPT) has the allomotph <-a> before /-m/ and the allomotph <-0> before the frrst person singular sufftx <-IJ> (1s). After the assimilatory nasalisation of the nonpreterite third person patient motpheme <-o> (3PINPT), the frrst person singular sufftx <-IJ> ( 1s) elides, leaving its zero allomotph. In preterite 1di---+3, 1de---+3, 2d---+3, 3d---+3, 2p---+3 and the alternative preterite 1pi---+3 and 1pi---+3ns forms, the third person patient marker <-u> (3P) has a zero allomotph <-0> before the preterite tense motpheme <-a> (PT) through the rule vocal is ante vocalem corripitur.
3.3. Suffzxal slots sf2 and sj3: tense and first person patient or subject sf2: <-a~-0>
PT
198 sf3: <-a> <-e> <-e>
RENE HUYSMANS
lsPS lsPINPT lpPS
Time is expressed in Sampang by two tenses, preterite tense and nonpreterite tense, but denoted by a single tense morpheme, the preterite tense morpheme <-a> (PT). The existence of a single tense sufftx follows from the fundamental principle against redundant tense marking in the Sampang verbal agreement system. The exact rules for the interpretation of Sampang tense will be discussed in my forthcoming grammar of Sampang. In this article, I concentrate on the morphological expression of tense. The preterite tense morpheme <-a> (PT) has a zero allomorph <-0> before any vowel, as regulated by the rule vocalis ante vocalem corripitur. The overt allomorph of the preterite tense morpheme <-a> (PT) occurs in transitive forms with a ftrst person dual or second person patient and in most intransitive and reflexive forms. In third person patient forms, tense is expressed either by the non-preterite third person patient morpheme <-o> (3PINPT) or by the preterite tense marker <-a> (PT), as described in Section 2. However, in intransitive and reflexive forms with a ftrst person singular subject, the preterite tense morpheme <-a> (PT) is reduced to its zero allomorph <-0> before the frrst person singular patient or subject marker <-a> ( lsPS), resulting in homophony of the two tenses. The morpheme <-a> ( lsPS) is also used to index a frrst person singular patient in the transitive conjugation and this causes homophony of the non-preterite 3s~ 1s forms and the intransitive 1s forms in the preterite and non-preterite tenses. In intransitive forms with a frrst person plural subject, the zero allomorph <-0> of the preterite tense morpheme <-a> (PI') occurs before the frrst person plural patient or subject morpheme <-e> (lpPS) from sufftxal slot sf3, again resulting in homophony of the preterite intransitive lpi and lpe forms on the one hand and of the non-preterite intransitive lpi and lpe forms on the other. These observations also apply to the reflexive conjugation. In transitive froms with a frrst person singular patient, the preterite tense marker <-a> (PT) is redundant and therefore not selected because transitive forms with a frrst person singular patient show a morphological differentiation of tense through the non-preterite portemanteau
THE SAMPANG VERBAL AGREEMENT SYSTEM
199
morpheme <-e> (lsPINPT) in non-preterite forms and the ftrst person singular patient or subject marker <-a> (lsPS) in preterite forms. This strategy is sufficient for distinguishing the two tenses. There is a side-effect related to the tense distinction in frrst person singular patient forms, which consists of expressing the patient or the agent more explicitly in the preterite forms than in the corresponding non-preterite forms. The preterite 2d---+ ls and 3d---+ ls forms are augmented with the frrst person singular copy morpheme <-IJ> ( ls ') after the allomorph <-cu> of the dual number morpheme <-ici> (d), the preterite 2p---+ ls form contains the frrst and second person plural agent suffix <-m> (12pA) following the allomorph <-nu> of the second person plural morpheme <-ini> (2p) and the preterite 3p---+ls form shows the third person non-singular agent marking sufftx <-IJ> (3nsA) after the allomorph <-mu> of the third person non-singular agent or subject marker <-imi> (3nsAS). The morphological transitivity of the preterite forms with a frrst person singular patient and a non-singular agent is a sign of the higher level of semantic transitivity associated with preterite tense. From a purely economical point of view, however, increased morphological transitivity is redundant in the frrst person singular patient forms, since the tenses are already distinguished through the portemanteau morpheme <-e> (lsP/NPT) and the frrst person singular patient or subject marker <-a> (lsPS). The increased morphological transitivity marking a higher level of semantic transitivity in the preterite forms can also be found in the 3p---+lpi and 3p---+lpe forms, where it has developed into a full-blown tense marking strategy. The preterite 3p---+lpi and 3p---+lpe forms are augmented with the third person non-singular agent morpheme <-IJ> (3nsA) after the allomorph <-mu> of the third person non-singular agent or subject marker <-imi> (3nsAS) but not the non-preterite 3p---+lpi and 3p---+lpe forms, which end in the allomorph <-mi> ofthe third person non-singular agent or subject marker <-imi> (3nsAS). In all other transitive forms with an inclusive or exclusive frrst person plural patient and in intransitive forms with a frrst person plural subject, the preterite tense morpheme <-a> (PT) is reduced to its zero allomorph <-0> before the frrst person plural patient or subject morpheme <-e> (lpPS). Evidently, the expression of tense with a ftrst person plural patient or subject is less important, except when the agent is third person plural. The formal similarity of the non-preterite ftrst person singular patient morpheme <-e> ( lsPINPT) and the sufftx <-e>
200
RENE HUYSMANS
(lpPS) causes homophony of the non-preterite 3s---+ls form and the 3s---+lpi forms, and the intransitive lpi form in both tenses, and of the non-preterite 3p---+lpi and 3p---+ls forms in both tenses. 3.4. Suffzxal slot sf4: first person <-l) ~ -0>
<-m.>
ls lpA'
The morpheme <-lJ> ( ls) indexes a first person singular actant, irrespective of its grammatical role and has a copy, the frrst person singular copy morpheme <-lJ> ( ls '), in sufftxal slot sf8. The copy suffiX <-m> (lpA') expresses a first person plural agent and is the echo of the first and second person plural agent marker <-m> (12pA) from suffJXal slot sf8. After the frrst person singular patient or subject suffiX <-ii> (lsPS) and the non-preterite third person patient morpheme <-o> (3PINPT), the frrst person singular morpheme <-l)> ( ls) elides at the surface, leaving a zero allomorph <-0> and assimilatory nasalisation of the preceding vocalic sufftx as the only trace of its former presence. The nasal quality of the frrst person singular patient or subject suffiX <-ii> (lsPS) is explained by the fact that it is always followed by the frrst person singular morpheme <-lJ> ( ls ). 3.5. Suffzxal slots sf5 and sj7: number and reflexive voice sf5: <-i ~ -0> <-ici ~ -ci ~ -cu ~ -c> < ... -ici ~ -ci ~ -cu> sf7: <-ini ~ -ni ~ -nu> <-imi ~ -mi ~ -mu> <-ici ~ -ci ~ -cu ~ -su>
23s d RFL
2p 3nsAS 3nsAP
The morphemes of suffJXal slots sf5 and sf7 are described in a single section because they share many allomorphic rules, and most of them also share function. The second or third person singular number morpheme <-i> (23s) elides after any vocalic suffiX and after /g/. The four number morphemes <-ici> (d), <-ici> (3nsAP), <-ini> (2p) and <-imi> (3nsAS), and the second part of the reflexive simulftx <-n -ici>
THE SAMPANG VERBAL AGREEMENT SYSTEM
201
(RFL) have the regular allomorphs <-ci>, <-ci>, <-ni>, <-mi> and <-ci> respectively after any suffJX consisting of, or ending in, a vowel and after /g/ or /m/. In this postion, the final vowel /if of these four suffJXes is assimilated to lui before any syllable-fmal bilabial or velar nasal /m/ or /g/, resulting in the allomorphs <-cu>, <-cu>, <-nu>, <-mu> and <-cu> respectively. The dual number morpheme <-ici> (d) has the regular allomorph <-c> in intervocalic position. For some native speakers, the third person non-singular agent morpheme <-ici> (3nsap) has the allomorph <-su> after the first person plural agent copy suffJX <-m> (1pA') and before the first or second person plural agent suffix <-m> (12pA). Other native speakers use the allomorph <-cu> of the third person non-singular agent morpheme <-ici> (3nsAP) in this position. The third person non-singular agent or patient morpheme <-ici> (3nsAP) has the allomorph <-ici> with a small number of irregular verbs, e.g. the non-preterite 3s---+3ns/3p---+3 form ci-ici (eat-3nsAP) 'he/she/it/theyos will eat himlher/it/themns• of the irregular verb ca,ma (vt) 'to eat'. With these irregular verbs, the morpheme <-o> expresses a transitive relationship between a frrst person agent and a third person patient in the non-preterite tense (1---+3/NPT), instead of the regular third person patient meaning in the non-preterite tense (3P/NPT). Singular number of a second or third person actant is indexed by the number morpheme <- i> (23s ). The overt allomorph of the second or third person singular number suffJX <-i> (23s) is found in the nonpreterite intransitive 2s and 3s forms and in the non-preterite forms with a second person singular patient In accordance with the limited expression of agents in second person patient forms, I shall assume that the second or third person singular number morpheme <-i> (23s) can only express singular number of the second person actant here. The dual number morpheme <-ici> (d) expresses dual number of frrst, second or third person actants. The number suffJX may be inserted only once into suffJXal slot sf5 and can refer to only one actant The presence of only one dual number sufftx in the Sampang simplex template makes ascertaining with which actant the dual number morpheme <-ici> (d) agrees impossible in transitive forms which express a scenario with a dual agent and a first person dual patient, viz. in the 2ns---+ 1de, 3ns---+ 1di and 3d---+ 1de forms. Agents in forms with a second person patient can only be expressed by the portemanteau morpheme <-n> (1---+2) or by the third person non-singular agent or sub-
202
RENE HUYSMANS
ject morpheme <-imi> (3nsAS), which means that any occurrence of the dual number suffJX <-ici> (d) in these forms indexes unequivocally the number of the patient. The morpheme <-imi> (3nsAS) expresses a third person nonsingular agent or subject and is treated in more detail below in Section 3.7, together with the third person non-singular agent morpheme <-IJ> (3nsA). The plural second person number morpheme <-ini> (2p) expresses a second person plural subject in the intransitive conjugation and a second person plural agent or patient in the transitive conjugation. The second person suffix <-na> (2) is redundant in a form marked by the plural second person number morpheme <-ini> (2p). All forms with a second person actant share a common format, with number being expressed by the second or third person singular morpheme <-i> (23s), the dual morpheme <-ici> (d) and the second person plural morpheme <-ini> (2p). In the third person patient system, the 2p~3 forms follow the alternative tense marking strategy with the preterite tense suffJX <-a> (PT) as in 2d~3 forms because, according to the present analysis, the plural second person number morpheme <-ini> (2p) blocks selection of the third person non-singular agent or patient suffix <-ici> (3nsAP), use of which appears to be an integral part of the default tense marking strategy as with 2s~3 and 2s~3ns forms. Whereas the morphology of Sampang simplicia differentiates between dual number and plural number with frrst and second person patients, third person patients can either have no number or non-singular number. The morpheme <-ici> (3nsAP) expresses a third person nonsingular agent or patient The third person non-singular agent or patient morpheme <-ici> (3nsAP) supplants the third person non-singular agent or subject morpheme <-imi> (3nsAS) in the third person patient system, so that there is a single 3p~3/3s~3ns form, and is used in any form where the number of a third person patient is specified, i.e. in the ls~3ns, lpi~3ns and lpe~3ns, 2s~3ns and 3s~3ns forms. Expression of the number of the third person patient by means of the third person non-singular agent or patient morpheme <-ici> (3nsAP) is in all cases optional. This is analogous to the optional Sampang nominal non-singular number suffJX <-ci> (ns) suffJXed to a noun (Huysmans, forthcoming). SuffJXal slot sf5 also contains the second element of the reflexive simulfix <-n -ici> (RFL). In reflexive forms with a dual subject, I shall assume that the element /ici/ is to be identified with the second
THE SAMPANG VERBAL AGREEMENT SYSTEM
203
-ici> (RFL) and not element <-ici> of the reflexive simulfJX <-n with the dual number suffJX <-ici> (d). The formal similarity of the second element <-ici> ofthe reflexive simulfJX <-n ... -ici> (RFL) and the dual number suffJX <-ici> (d) makes it impossible for forms expressing the reflexive 2s and 2d scenarios, and the reflexive 3s and 3d scenarios, to distinguish between singular and dual number of the reflexive subject. Furthermore, reflexive 3ns forms are homophonous with 1~2d forms. 3.6. Sujf'vcal slots sf6 and sf9: exclusiveness and second person sf6: <-ka> <-0>
ePS iPS
sf9: <-ka> <-0> <-na>
eA iA 2
On the basis of the third person patient forms with an exclusive agent and the forms with a frrst person exclusive patient, one can deduce the existence of two exclusive markers. The exclusive patient or subject marker <-ka> (ePS) is located in front of the third person non-singular agent or subject morpheme <-itni> (3nsAS) and the second person plural morpheme <-ini> (2p), and is reserved for the expression of exclusiveness of a frrst person patient or subject The exclusive agent suffJX <-ka> ( eA), however, must be located after the first and second person plural agent suffix <-m> (12pA). Absence of either the exclusive patient or subject suffJX <-ka> (ePS) or the exclusive agent suffJX <-ka> (eA), indicates inclusiveness of a frrst person actant and is encoded through the zero inclusive patient or subject suffix <-0> (iPs) and the zero inclusive agent suffJX <-0> (iA) respectively. Inclusive and exclusive forms are structurally identical, but for the presence of either a zero inclusive marker or an overt exclusive marker. The morpheme <-na> (2) refers to a second person actant. The second person morpheme <-na> (2) in combination with the second or third person singular number morpheme <-i> (23s) marks a second person singular actant A second person dual argument is expressed by the combination of the second person marker <-na> (2) and the dual number suffix <-ici> (d), indexing the number of the second person
204
RENE HUYSMANS
actant. The morpheme <-n> (1---+2) and the second person morpheme <-na> (2) are in complementary distribution.
3. 7. Suffrxal slot sf8: non-singular agent and copy slot <-m.> <-lJ> <-lJ> <-U>
12pA 3nsA ls' 3p'
The suffix <-m> (12pA) expresses a frrst or second person plural agent and has a copy suffix <-m> (lpA') in suffixal slot sf4. The frrst or second person plural agent <-m> (12pA) occurs in almost any form with a frrst or second person plural agent, except in the non-preterite 2p---+ ls form, where its absence is motivated by a lower degree of semantic transitivity (see Section 3.3). The frrst or second person plural agent suffix <-m> (12pA) is also absent in the alternative preterite lp---+3 form -a. Likewise, the frrst person plural agent copy morpheme <-m> (lpA') is not selected in the alternative preterite lpi---+3ns form -acum. This is because Sampang morphophonology changes any vowel /a/ to /a/ in closed syllables, e.g. /a/ changes to /a/ in the preconsontal stem allomorph of Mlma (vt) 'to arrange, order' where its prevocalic stem allomorph is identical to the underlying form (see Huysmans, forthcoming). Insertion of the frrst and second person plural agent morpheme <-m> (12pA) in the alternative preterite lpi---+3 form -a and of the frrst person plural agent copy suffJX <-m> (lpA') in the alternative preterite lpi---+3ns form -acum would then render these forms homophonous with the non-preterite lpi---+3 form -am and with the non-preterite lpi---+3ns form -amcum The third person non-singular agent morpheme <-lJ> (3nsA) always follows the third person non-singular agent or subject morpheme <-imi> (3nsAS), unless its absence distinguishes non-preterite tense from preterite tense in forms with a plural agent (see Section 3.3). The third person non-singular agent morpheme <-lJ> (3nsA) takes part in a complex interplay with the third person non-singular agent or subject morpheme <-imi> (3nsAS) and the dual number morpheme <-ici> (d), which-at least partly-seems to have evolved so that frrst person patient scenarios with third person dual and plural agents are expressed by sufficiently distinct forms without awkward cases of homophony.
THE SAMPANG VERBAL AGREEMENT SYSTEM
205
The default expression of a third person dual actant is through the dual nwnber suffix <-ici> (d) alone, and that of a third person plural actant just by means of the third person non-singular agent or subject morpheme <-imi> (3nsAS), as found in the intransitive conjugation. A third person dual reflexive subject is perhaps not expressed at all, if it is asswned that the dual number suffJX <-ici> (d) is blocked by the second element <-ici> of the reflexive simulfJX <-n -ici> (RFL). The third person non-singular agent or subject morpheme <-imi> (3nsAS) can be, but is not, used to mark third person dual reflexive subjects. In this way, homophony with the reflexive 3p forms is avoided, at the lesser cost of homophony of the reflexive 3sd forms with the reflexive Idi forms. Likewise, the third person non-singular agent or subject morpheme <-imi> (3nsAS) is absent in the intransitive 3d forms, because homophony of the intransitive Idi and 3d forms represents a lesser burden for the verbal agreement system. In the 3d---+ Is and 3d---+3 forms, the default strategy for the expression of a third person dual agent or subject using just the dual nwnber suffix <-ici> (d) is followed. The third person non-singular agent or subject morpheme <-imi> (3nsAS) and the third person non-singular agent morpheme <-IJ> (3nsA) are present in 3d---+Ipi forms to distinguish them from the non-preterite 3d---+ Is form. The default strategy for the expression of a third person dual agent or subject using just the dual nwnber suffJX <-ici> (d) would cause the forms expressing 3d---+ Idi and 3d---+ Ide scenarios to be homophonous with the intransitive Idi and Ide forms. This is avoided through the addition of the third person non-singular agent or subject morpheme <-imi> (3nsAS) and the third person non-singular agent morpheme <-IJ> (3nsA), at the cost of 3ns---+ Idi and 3ns---+ Ide forms where third person dual and third person plural agents are not distinguished. Moreover, presence or absence of the dual nwnber morpheme <-ici> (d) cannot distinguish between a third person dual and a third person plural agent in 3ns---+ Idi and 3ns---+ Ide forms, as it does in the forms with a frrst person plural patient, since the dual nwnber morpheme <-ici> (d) can also be used for the expression of a first person dual patient Absence of the third person non-singular agent morpheme <-IJ> (3nsA) generally expresses lesser semantic transitivity of non-preterite frrst person patient forms with a third person plural agent and, theoretically, this strategy could have been applied to create separate forms for the non-preterite 3p---+ Idi and 3p---+ Ide scenarios. But, the morpho-
206
RENE HUYSMANS
logical expression of the degree of semantic transitivity has apparently only spread outside the frrst person singular patient system insofar as it represented a last-resort tense marking strategy, as in the 3p---+ lp forms. In 3ns---+ ld forms, however, tense is already overtly marked by the preterite tense suffix <-a> (PT). Suffood slot sf8 also contains the frrst person singular copy morpheme <-IJ> (ls') and the third person patient copy suffix <-u> (3p'). Copy morphology has been described in detail in Section 2. 3.8. Suffvcal slot sjJO: negation
<-no -
-na - -no -
-na>
NG
The second negative sufftx <-no> (NG) expresses negative meaning and ftlls the last sufftxal slot sflO. The allomorphy of the negative sufftx <-no> (NG) is subject to the following rules: the allomorph <-na> is selected after sufftxal strings ending in the sequence of the ftrst person singular patient or subject suffix <-a> (lsPS) and the frrst person singular sufftx <-IJ> (ls). Sufftxal strings ending in the ftrst person singular sufftx <-IJ> ( ls) preceded by the third person patient morpheme <-u> (3P) or the non-preterite third person patient morpheme <-o> (3P/NPT) result in the allomorph <-nO>, and sufftxal strings ending in the second person morpheme <-na> (2) trigger the selection of the allomorph <-na>. In all other cases, the allomorph <-no> surfaces.
4. PARADIGMS In order to present the results of the analysis of the verbal agreement morphology of Sampang simplicia, the full paradigms of the intransitive verb ghi:ma 'to laugh' and of the transitive verb yemma 'to hit' are presented below. To illustrate the reflexive conjugation, the transitive verb khO:ma 'to look' is used. The Sampang verb has only one stem in its paradigm, which can be adequately described as a set of morphophonologically conditioned allomorphs. Sampang shares this feature with Thangmi, Yamphu and Lohorung, whereas in other Kiranti languages, such as Dumi, Limbu and Wambule, a verb can have several paradigmatically conditioned stems. Most Sampang verbs have an invariable stem used before all sufftxes, but some verbs have a preconsonantal form and a prevocalic form, and some verbs have two or more prevocalic stem forms. The prevocalic forms oftheverbs ghi:ma
THE SAMPANG VERBAL AGREEMENT SYSTEM
207
'to laugh' and yemma 'to hit' are and respectively. Preterite forms are listed below the non-preterite ones. Table 14 1s
1di 1de 1pi 1pe 2s
2d 2p
3s 3d 3p
Table 15 1s/RFL
1di/RFL 1de/RFL 1pi/RFL 1pe/RFL
ghi:ma (vi) 'to laugh' ghis-fi-9 laugh-lsPS-1s ghis-fJ-fi-9 laugh- PT -lsPS-1s ghis-ici-9 laugh-d-iPs ghis-a-ci-9 laugh-PT-d-iPs ghis-ici-ka laugh-d-ePS ghis-a-ci-ka laugh-PT-d-ePS ghis-e-9 laugh-1pPs-iPs ghis-fJ-e-9 laugh-PT-1 pPS-iPS ghis-e-ka laugh-1pPs-ePS ghis-9-e-ka laugh-PT-1 pPS-ePS ghis-i-na laugh-23s-2 ghis-a-9-na laugh-PT-23s-2 ghis-ici-na laugh-d-2 ghis-a-ci-na laugh-PT-d-2 ghis-ini laugh-2p ghis-a-ni laugh-PT-2p ghis-i laugh-23s ghis-a-9 laugh-PT-23s ghis-ici laugh-d ghis-a-ci laugh-PT-d ghis-imi laugh-3nsAS ghis-a-mi laugh-PT-3nsAS khO:ma (vt) 'to look' kho:-n-a-0-cu-lJ look-RFL-1sPs-1s-RFL-1s' kho:-n-fJ-a-fJ-cu-g look-RFL-PT -lsPS-1s-RFL-1 s' look-RFL-RFL-iPS k:ho:-n-ici-9 look-RFL-PT-RFL-iPS k:ho:-n-a-ci-9 khO:-n-ici-ka look-RFL-RFL-ePS kho:-n-a-ci-ka look-RFL-PT-RFL-ePS k:ho:-n-e-ci-9 look-RFL-1pPS-RFL-iPS k:ho:-n-rtJ-e-ci-9 look-RFL-PT-1pPS-RFL-iPS kho:-n-e-ci-ka look-RFL-1pPS-RFL-ePS k:ho:-n-9-e-ci-ka look-RFL-PT-1pPS-RFL-ePS
RENE HUYSMANS
208 Table 15 2sd/RFL
2p/RFL 3sd/RFL 3p/RFL
Table 16 2s---+ ls
2d---+ls 2p---+ls 3s---+ ls 3d---+ls 3p---+ ls 3s---+ ldi 3ns---+ldi 2sd---+lde 2p---+lde 3s---+lde 3ns---+lde 3s---+lpi 3d---+lpi
continued kh&-n-ici-na kh&-n-a-ci-na kh&-n-ici-ni kh&-n-a-ci-ni kh&-n-ici kh&-n-a-ci kh&-n-ici-mi kh&-n-a-ci-mi
look-RFL-RFL-2 look-RFL-PT-RFL-2 look-RFL-RFL-2p look-RFL-PT-RFL-2p look-RFL-RFL look-RFL-PT-RFL look-RFL-RFL-3nsAS look-RFL-PT-RFL-3nsAS
yemma (vt) 'to hit'
yeps-e-0-na yeps-fi-ftJ-fJ-na yeps-e-ci-na yeps-fi-ftJ-cu-.y-na yeps-e-ni yeps-fi-ftJ-nu-m yeps-e-ftJ yeps-fi-fJ-ftJ yeps-e-ci yeps-fi-fJ-cu-.y yeps-e-mi yeps-fi-ftJ-mu-.y yeps-ici-ftJ yeps-a-ci-ftJ yeps-ici-fJ-mu-.y yeps-a-ci-fJ-mu-.y yeps-ici-ka-na yeps-a-ci-ka-na yeps-ici-ka-nu-m yeps-a-ci-ka-nu-m yeps-ici-ka yeps-a-ci-ka yeps-ici-ka-mu-.y yeps-a-ci-ka-mu-.y yeps-e-@-ftJ yeps-ftJ-e-ftJ-ftJ yeps-e-ci-ftJ-mu-.y yeps-ftJ-e-ci-ftJ-mu-.y
hit-lsP/NPT-23s-2 hit-lsPS-ls-23s-2 hit-lsP/NPT-d-2 hit-lsPS-ls-d-ls'-2 hit-lsP/NPT-2p hit-lsPS-ls-2p-12pA hit-lsP/NPT-23s hit-lsPS-ls-23s hit-lsP/NPT-d hit-lsPS-ls-d-ls' hit-lsP/NPT-3nsAS hit-lsPS-ls-3nsAS-3nsA hit-d-iPs hit-PT-d-iPS hit-d-iPs-3nsAs-3nsA hit-PT-d-iPS-3nsAS-3nsA hit-d-ePS-2 hit-PT-d-ePS-2 hit-d-ePS-2p-12pA hit-PT-d-ePS-2p-12pA hit-d-ePS hit-PT-d-ePS hit-d-ePS-3nsAs-3nsA hit-PT-d-ePS-3nsAS-3nsA hit-lpPS-23s-iPS hit-PT-lpPS-23s-iPS hit-1 pPS-d-iPs-3nsAS-3nsA hit-PT-lpPS-d-iPS-3nsAs-3nsA
THE SAMPANG VERBAL AGREEMENT SYSTEM Table 16
3p---+1pi
209
continued
yeps-e-fJ-mi yeps-fJ-e-fJ-mu-g 2s---+1pe yeps-e-fJ-ka-na yeps-fJ-e-fJ-ka-na 2d---+1pe yeps-e-ci-ka-na yeps-IJ-e-ci-ka-na 2p---+1pe yeps-e-ka-nu-m yeps-fJ-e-ka-nu-m 3s---+1pe yeps-e-fJ-ka yeps-fJ-e-fJ-ka 3d---+1pe yeps-e-ci-ka-mu-g yeps-fJ-e-ci-ka-mu-g 3p---+1pe yeps-e-ka-mi yeps-fJ-e-ka-mu-g 1s---+2s yem-n-i yem-n-a-fJ 1e/3np---+2s yeps-i-na yeps-a-fJ-na 3p---+2s yeps-i-mi-na yeps-a-fJ-mi-na 1---+2d yem-n-ici yem-n-a-ci n1s---+2d yeps-ici-na yeps-a-ci-na 1---+2p yem-n-ini yem-n-a-ni n1s---+2p yeps-ini yeps-a-ni 1s---+3 yeps-6-fJ yeps-u-g 1s---+3ns yeps-6-fJ-cu-g yeps-u-g-cu-g 1di---+3 yeps-u-c-u-fJ yeps-fJ-a-c-u-fJ 1de---+3 yeps-u-c-u-ka yeps-fJ-a-c-u-ka
hit-1 pPS-iPS-3nsAS hit-PT-1pPS-iPs-3nsAS-3nsA hit-1 pPS-23s-ePS-2 hit-PT-1pPS-23s-ePS-2 hit-1pPS-d-ePS-2 hit-PT-1pPS-d-ePS-2 hit-1 pPS-ePS-2p-12pA hit-PT-1pPS-ePs-2p-12pA hit-1pPS-23s-ePS hit-PT-1 pPS-23s-ePS hit-1 pPS-d-ePS-3nsAS-3nsA hit-PT-1pPS-d-ePS-3nsAs-3nsA hit-1 pPS-ePS-3nsAS hit-PT-1pPS-ePs-3nsAS-3nsA hit-1---+2-23s hit-1---+2-PT-23s hit-23s-2 hit-PT-23s-2 hit-23s-3nsAs-2 hit-PT-23s-3nsAS-2 hit-1---+2-d hit-1---+2-PT-d hit-d-2 hit-PT-d-2 hit-1---+2-2p hit-1---+2-PT-2p hit-2p hit-PT-2p hit-3P/NPT-1s hit-3P-1s hit-3P/NPT-1s-3nsAP-1s' hit-3P-1s-3nsAP-1s' hit-3P-d-3P' -iA hit-3P-PT-d-3P'-iA hit-3P-d-3P' -eA hit-3P-PT-d-3P'-eA
210
RENE HUYSMANS
Table 16 1pi---+3
1pi---+3ns
1pe---+3 1pe---+3ns 2s---+3 2s---+3ns 2d---+3 2p---+3 3s---+3 3d---+3 3s---+3ns/ 3p---+3
continued yeps-8-m-fJ yeps-u-m-fJ yeps-fJ-a-fJ yeps-8-m-cu-m-fJ yeps-u-m-cu-m-fJ yeps-fJ-a-cu-m-fJ yeps-8-m-ka yeps-u-m-ka yeps-8-m-cu-m-ka yeps-u-m-cu-m-ka yeps-o-fJ-na yeps-u-fJ-na yeps-o-ci-na yeps-u-ci-na yeps-u-c-u-na yeps-fJ-a-c-u-na yeps-u-nu-m yeps-fJ-a-nu-m yeps-o-fJ yeps-u-fJ yeps-u-c-u yeps-fJ-a-c-u yeps-o-ci yeps-u-ci
hit-3P/NPT-12pA-iA hit-3P-12pA-iA hit-3P-PT-iA hit-3P/NPT-1pA'-3nsAP-12pA-iA hit-3P-1pA'-3nsAP-12pA-iA hit-3P-PT-3nsAP-12pA-iA hit-3P/NPT-12pA-eA hit-3P-12pA-eA hit-3P/NPT-1pA'-3nsAP-12pA-eA hit-3P-1pA'-3nsAP-12pA-eA hit-3P/NPT-23s-2 hit-3P-23s-2 hit-3P/NPT-3nsAP-2 hit-3P-3nsAP-2 hit-3P-d-3P' -2 hit-3P-PT-d-3P'-2 hit-3P-2p-12pA hit-3P-PT-2p-12pA hit-3P/NPT-23s hit-3P-23s hit-3P-d-3P' hit-3P-PT-d-3P' hit-3P/NPT-3nsAP hit-3P-3nsAP
5. SAMPANG VERBAL MORPHOLOGY FROM APROTO-KIRANTI PERSPECfiVE
As more information on Kiranti verbal agreement systems becomes available, increasingly accurate inferences can be made about the affJXal morphology of the Proto-Krranti simplex verb. The present comparison of the Sampang verbal agreement system with the simplex morphology of other Kiranti languages is based on the following sources: Michailovsky (1975) and van Driem (1991) for Bahing, Rutgers (1993) for Chepang, van Driem (1993b) for Dumi, Michailovsky (1981) and van Driem (1991) for Hayu, Tolsma (1997) for Kulung, van Driem ( 1987) and van Driem ( 1997) for Limbu, van Driem ( 1992) for Loborung, Turin (2006) for Thangmi, Allen (1975) and van Driem (1990)
211
THE SAMPANG VERBAL AGREEMENT SYSTEM
for Thulung, Opgenort (2002) for Wambule, and Rutgers ( 1998) for Yamphu. The verbal agreement systems of the above languages have been the subject of rigorous synchronic descriptions using the concept of afftxal slot, namely the functional position of a morpheme in the afftxal string. In the pursuit of a model of the Proto-Kiranti verb, it is assumed that the recurring Kiranti morphemes and their positions in the afftxal strings reflect an earlier non-random order of afftxes in ProtoKiranti (van Driem 1990: 27-8). In this section, the Sampang simplex verbal morphemes and their positions in the afftxal string will be compared with the Proto-Kiranti element order as proposed by van Driem (1990, 1991, 1992, 1997). Table 17 reproduces the model of the verbal afftxes of Proto-Kiranti following van Driem (1992) and incorporates the last amendments made by van Driem (1997). Van Driem's Proto-Kiranti model postulates the existence of inflected auxiliaries in Pre-Kiranti followed by a process of univerbation in the Proto-Kiranti stage or shortly after the break-up of Proto-Kiranti (van Driem 1990: 46). Van Driem (1992: 69) adduces the existence of copy morphemes in Limbu, Dumi and Thulung as a possible indication of the former presence of these inflected auxiliary verbs in Pre-Kiranti. The presence in the Limbu, Dumi, Lohorung, Yamphu and Kulung morphological templates of multiple negative morphemes, many of them with specialised meanings, also seems to corroborate the former existence of auxiliaries in the ProtoKiranti verb. The Sampang copy morphemes can also be adduced as evidence for the auxiliary theory. Table 17
Model of the Proto-Kiranti simplex verb
-k 1p -k NPT v•
me- l: -ns1 3pA RFL
-na 2
+AUXt
-U:
-nya 1s---+2
-ru
-ga~-g
1s -C1 d
2p
-ya e
+AUX2 -u 3P
-m -1 12pA 1
PT
-C1 3dP
212
RENE HUYSMANS
The Sampang verbal agreement system follows the split-ergative nature underlying Kiranti conjugational morphology observed in previous comparisons of Kiranti verbal agreement systems (van Driem 1990, 1991, 1992), whereby third person actants are indexed differently in the verb than are frrst and second person actants. Third person actants are marked according to a so-called accusative pattern, in which distinct sets of morphemes index for a third person patient as opposed to a third person agent or subject On the other hand, morphemes denoting the involvement of a first or second person actant follow a so-called ergative pattern in that one set of morphemes indexes for a frrst or second person agent, while another set denotes a frrst or second person patient or subject In Sampang, agreement with a frrst person singular or plural actant exhibits an ergative pattern, as is indicated by the frrst person singular patient or subject marker <-a> (lsPS), the first person plural patient or subject morpheme <-e> ( 1pPS), the exclusive patient or subject marker <-ka> (ePS) and the exclusive agent suffix <-ka> (eA). Agreement with a third person actant exhibits an accusative pattern as is borne out by the third person non-singular subject or agent morpheme <-imi> (3nsAS), the third person patient morpheme <-u> (3P) and the nonpreterite third person patient morpheme <-o> (3P/NPT). A large portion of the Sampang simplex morpheme inventory can be related in a straightforward way to van Driem's model of ProtoKiranti affixes, thereby further increasing its plausibility. Most Kiranti languages have two positions for indicating a plural agent. The frrst one is very anterior, even preftxal in some languages such as Limbu and Dumi, and is represented by the Proto-Kiranti third person plural agent prefiX *<me-> (3pA), in Sampang reflected by the third person non-singular agent or subject marker <-imi> (3nsAS). The Sampang third person non-singular agent or subject marker <-imi> (3nsAS) is cognate with Limbu (3nsAS), Dumi (3ps), Hayu <-me> (3p), Thulung <-mi> (3p~3), Lohorung <-mi> (3p), Bahing <-me> (3p~3/3ps) and <-mi> (3p/1pe), Wambule <-mu> (3~3p) and <-mi> (3ns) and Yamphu <-mi> (3p). Unlike most Kiranti reflexes of the third person plural agent proto-prefix *<me-> (3pA), the Sampang reflex occupies a rather posterior location in the a:fftxal string. The second Proto-Kiranti morpheme for the expression of a plural agent, the proto-suffix *<-m> (12pA), tends to be reserved for first and second person plural agents and is located in a more posterior po-
THE SAMPANG VERBAL AGREEMENT SYSTEM
213
sition in the Proto-Kiranti morphological template. The present-day Sampang first and second person plural agent sufftx <-m> (12pA) can safely be interpreted as a reflex of the Proto-Kiranti first and second person plural agent sufftx *<-m> ( 12pA). The third person nonsingular agent suffix <-lJ> (3nsA) may also be a reflex of this protomorpheme. The Sampang first and second person plural agent sufftx <-m> (12pA) and perhaps the third person non-singular agent sufftx <-lJ> (3nsA) are cognate with other Kiranti agentive morphemes such as Limbu <-m> (pA) and Lohorung <-m> (1peAS) and <-m> (2p---+3). The Sampang reflexive morpheme <-n -ici> (RFL) can be related directly to the Proto-Kiranti reflexive morpheme *<-nsi> (RFL) and is cognate with other reflexive suffixes found in Kiranti languages, such as Limbu <-silJ - -m: - -n> (RFL ), Dumi <-nsi> (RFL ), Hayu <-na- -ntse- -ntsi> (RFL), Bahing <-si> (RFL), Thangmi <-si> (RFL) and Kulung <-nci> (RFL). In almost all Kiranti languages, the reflexive sufftx fills a single slot, mostly immediately after the verb stem. The distribution over two sufftxal slots of the Sampang simulftx <-n -ici> (RFL), namely the first suffixal slot sfl and the number slot sf5, is in all probability a subsequent Sampang reanalysis of the element <-ici> by analogy with the dual number sufftx <-ici> (d). In Kiranti, the exclusive proto-sufftx *<-ya> appears to have been subject to fusion with other sufftxes on more than one occasion. First, the Proto-Kiranti exclusive morpheme *<-ya> (e) may have fused with the second person proto-morpheme *<-na> (2) yielding the ProtoKiranti 1s---+2 morpheme *<-nya>. This tentative reconstruction by van Driem (1990: 45-6; 1991: 350) is prompted by the close formal resemblance between the Kiranti reflexes of the morpheme *<-nya> (1s---+2) and the Proto-Kiranti second person morpheme *<-na> (2). The Sampang 1---+2 morpheme <-n> in sufftxal slot sfl can be related directly to the Proto-Kiranti morpheme *<-nya> (1s---+2) and is cognate to Limbu <-m: - -n> (1---+2), Dumi <-n> (1s---+2), Hayu <-no> (1s---+2), Thulung <-nini> (1s---+2), Bahing <-na> (1s---+2), Wambule <-ni> (1s---+2), Yamphu <-na> (1---+2), Chepang <-ne?> (1---+2) and Kulung <-yan> (1s---+21NPT). On the basis of semantic criteria, the 1s---+2 proto-morpheme *<-nya> is located after the tense morphemes in van Driem's model of the Proto-Kiranti verb, in the same slot as the Proto-Kiranti second person sufftx *<-na> (2) (van Driem 1991: 350). However, most Kiranti reflexes of the morpheme *<-nya> ( 1s---+2), including the
214
RENE HUYSMANS
Sampang reflex, are fmmd in a position anterior to the tense slot The Pre-Kiranti origins of the proto-morpheme *<-nya> (1s---+2) may occupy the second person morpheme *<-na> (2) slot (van Driem 1993a: 320), but perhaps it had already been promoted to a position closer to the stem in Proto-Kiranti times. I suggest (with Rutgers 1993: 120) that the 1s---+2 proto-morpheme *<-nya> in the Proto-Kiranti morphological template be positioned before the tensed auxiliary AUX 1, in the same slot as the reflexive proto-morpheme *<-n8i> (RFL ). The exclusive proto-suffix *<-ya> (e) may also have fused with the Proto-Kiranti frrst person plural suffix *<-k> (1p) (van Driem, 1990, 1991, 1992), yielding the Sampang exclusive patient or subject morpheme <-ka> (ePS) and exclusive agent marker <-ka> (eA), and other cognate reflexes, such as Kulung <-ka> (e) and Lohorung <-ka> (e). The coexistence of two exclusiveness markers is a subsequent development in Sampang. The third possible candidate for a coalescence of the Proto-Kiranti exclusive marker *<-ya> (e) and another proto-marker is peculiar to Sampang. The Sampang frrst person plural patient or subject morpheme <-e> (1pPS) may be a fusion of the Proto-Kiranti exclusive marker *<-ya> (e) and the Proto-Kiranti inclusive suffix *<-i> (i), as is suggested by Kulung, one of Sampang's closest relatives, which has the reflex <-ya> (1pPS/NPI') of the exclusive proto-sufftx *<-ya> (e) in non-preterite frrst person plural patient and subject forms and the reflex <-i> (1pPS/Pr) of the Proto-Kiranti inclusiveness suffix *<-i> (i) in preterite frrst person plural patient and subject forms (Tolsma 1997: 69). If this analysis is accepted, then the Sampang frrst person plural patient or subject <-e> (1pPS) is cognate with the Dumi inclusive marker <-i> (i) and other Kiranti (frrst person) plural markers such as Limbu <-i> (pPS), Thulung <-i> (1pi---+3), Lohorung <-i> (1pPS), Wambule <-i> (1pi---+3), Yamphu <-i> (12p), Thangmi <-i> (1pPS) and the Limbu sufftx <-?e> (ls/NPT). The Sampang non-preterite first person singular patient marker <-e> (1sP/NPT) is cognate with Limbu <-?e> (1s/NPI') and may represent a reflex of the Proto-Kiranti exclusive marker *<-ya> (e). The Sampang preterite tense sufftx <-a> (PT) is cognate with Limbu <-e> (PT), Lohorung <-a> (PT), Yamphu <-a> (PT) and Kulung <-a> (PT), and can be readily derived from the vowel of the ProtoKiranti preterite tense morpheme *<-(T)e> (PT). The anterior position
THE SAMPANG VERBAL AGREEMENT SYSTEM
215
of the Sampang preterite tense morpheme <-a> (PT) is in support of an anterior position of the Proto-Kiranti tense morphemes. The Proto-Kiranti second person marker *<-na> (2) is reflected by the Sarnpang second person morpheme <-na> (2), which is cognate with Thulung <-na> (2), Lohorung <-na> (2), Thangmi <-na> (2s) and Kulung <-na> (2s/PT). In the majority of Kiranti languages, the reflexes of the second person proto-marker *<-na> (2) occur in the initial portion of the sufftxal string. The location of the Sampang second person morpheme <-na> (2) towards the end of the sufftxal string in sf9 is, from a Proto-Kiranti and a Proto-Tibeto-Burman perspective, highly posterior and most certainly a Sampang innovation. The Sampang reflex of the second person proto-marker *<-na> (2) shares its posterior location in the sufftxal string with the Lohorung reflex. The Sampang third person non-singular agent or patient morpheme <-ici> (3nsAP) is in all probability a reflex of the Proto-Kiranti third person dual patient morpheme *<-ci> (3dP). After over-generalisation of the original dual meaning of the third person dual patient protomorpheme *<-ci> (3dP), the proto-sufftx carne to express non-singularity (cf. van Driem 1987: 31-2). The Sampang third person non-singular agent or patient morpheme <-ici> (3nsAP) is cognate with Limbu <-si> (ns), Thulung <-ci> (3dP), Lohorung <-ci> (nsP), Bahing <-si> (3d), Kulung <-ci> (3ns) and Yarnphu <-ji> (3ns). The Sarnpang dual number morpheme <-ici> (d) is likely to be the reflex of the dual proto-morpheme *<-ci> (d) and is cognate with numerous Kiranti dual morphemes such as Limbu <-si ~ -s ~ tchi ~ -tch> (d), Thulung <-ci> (d), Lohorung <-ci> (d), Hayu <-tshe> (d), Wambule <-ci> (d), Yamphu <-ci> (d), Kulung <-ci> (d) and Chepang <-cP (d). The Sampang number morphemes <-ici> (d) and <-ici> (3nsAP) are compatible with the revised model of the Proto-Kiranti verb of van Driem (1997) which posits two distinct number morphemes *<-ci> (d) and *<-ci> (3dP), instead of the three number morphemes proposed in the earlier model of van Driem (1991). The two Proto-Kiranti morphemes for dual number *<-ci> (d) and *<-ci> (3dP) reflect the general tendency for Kiranti languages, also attested in Sarnpang, to have two functional positions for number, one in the initial portion of the sufftxal string, filled by the dual number proto-morpheme *<-ci> (d), after AUXt. and one in the final portion, after AUX:z, filled by the third person dual patient proto-morpheme*<-ci> (3dP).
216
RENE HUYSMANS
The Sampang frrst person singular morpheme <-IJ> ( 1s) in sufftxal slot sf4 reflects the frrst person singular proto-morpheme *<-IJa - -IJ> ( 1s). The single occurrence in Sampang of a first person singular morpheme <-IJ> (1s) is in support of a single Proto-Kiranti first person singular marker <-IJa- -IJ> (1s) as proposed in the frrst model andreintroduced in the fourth model ofProto-Kiranti, instead of the two distinct frrst person singular proto-morphemes of the intermediate models (van Driem 1997: 172). There are numerous Kiranti frrst person singular morphemes which are cognate with the Sampang frrst person singular morpheme <-IJ> (1s), e.g. Limbu <-aiJ - -IJ> (ls), Dumi <-IJ> (1s), Thangmi <-IJa> (1s), Thulung <-IJi> (1sP) and the Lohorung suffiXes <-IJa> (1s) and <-iiJ> (lsPS/PT). The Sampang third person patient marker <-u> (3P) can be related directly to the Proto-Kiranti third person patient morpheme *<-u> (3P), reflexes of which are ubiquitously present in the Tibeto-Bunnan languages (van Driem 1993a: 322). The Sampang non-preterite third person patient marker <-o> (3PINPT) can also be seen as a reflex of the third person patient proto-suffix *<-u> (3P). Kiranti cognates of the Sarnpang third person patient marker <-u> (3P) are present in abundance, e.g. Limbu <-u> (3P), Lohorung <-u> (3P), Yamphu <-u> (3P), Kulung <-u> (3p/PT) and Thangmi <-u> (3P). The Sarnpang second person plural morpheme <-ini> (2p) reflects unequivocally the Proto-Kiranti second person plural marker *<-ni> (2p) and is cognate with Thulung <-ni> (2p), Lohorung <-ni> (2p), Bahing <-ni> (2p), Wambule <-ni> (2p), Thangmi <-ni> (2p) and Kulung <-ni> (2p ). 6. CONCLUDING REMARKS In this contribution, I have provided an analysis of the Sarnpang verbal agreement system. The expression of tense, number and copy morphology in Sarnpang prove to be intimately linked, especially in the third person patient system. Each affJXal string has been analysed in its constituent parts, each morpheme has been assigned a single meaning and its position vis-a-vis the other morphemes has been established with the help of the concept of affJXal slots. This enables a comparison with other Kiranti verbal agreement systems, thereby obtaining a glimpse of the historical development of Sarnpang verbal morphology.
THE SAMPANGVERBALAGREEMENT SYSTEM
217
The results clearly indicate that Sampang has a solid pedigree within Kiranti. Sampang verbal morphology is particularly similar to the verbal agreement system of Kulung, but also to the verbal morphology of Lirnbu and Lohorung. The synchronic analysis of the Sampang verbal agreement system presented here as well as the various reflections on some of its diachronic dimensions illustrate in minute detail how morphological systems operate and are rernoulded as languages evolve. Morphological change, and no doubt language change in general, is seen to be primarily semantically driven, whereby formal systems represent indices of meaning and their reanalysis and remodelling is propelled by meaning or ever reinterpreted shades of meaning.
BIBLIOGRAPHY Allen, Nicholas J. 1975. Sketch ofThulung Grammar: With Three Texts and a Glossary. (East Asia Papers No.6). Ithaca: Cornell University. Bauman, James, Jolm. 1975. Pronouns and Pronominal Morphology in TibetoBurman. University of California at Berkely: Unpublished doctoral dissertation, 20 December 1975. van Driern, George L. 1987. A Grammar ofLimbu. Berlin: Mouton de Gruyter. - - . 1990. An Exploration of Proto-Kiranti Verbal Morphology. Acta Linguistica Hqfniensia 22: 27-48. - . 1991. Bahing and the Proto-Kiranti Verb. Bulletin of the School of Oriental and African Studies 54.2: 336-356. - - . 1992. Le proto-kiranti revisite, morphologie verbale du lohonmg. Acta LinguisticaHqfniensia 24: 33-75. - - . 1993a. The Proto-Tibeto-Burman Verbal Agreement System. Bulletin of the School ofOriental and African Studies 56.2: 292-334. - - . 1993b. A Grammar ofDwni. Berlin: Mouton de Gruyter. - - . 1997. A New Analysis of the Limbu Verb. In David BracD.ey, ed, Papers in Southeast Asian Linguistics No. 14: Tibeto-Burman Languages of the Himalayas. (Pacific Linguistics, A-86.) Canberra: Australian National University, 157-173. Huysmans, Rene. Forthcorning.A Grammar ofSampang. Michailovsky, Boyd 1975. Notes on the Kiranti Verb (East Nepal). Linguistics ofthe Tibeto-BunnanArea 2.2: 183-218. - - . 1981. La langue hayu. Ph.D. thesis. University of California at Berkeley (University Microfilms International). Opgenort, Jean-Robert M. L. 2002. The Wambule Language: Grammar, Lexicon, Texts and Cultural Smvey of a Rai-Kiranti Tribe of Eastern Nepal. Universiteit Leiden: Unpublished doctoral dissertation, 6 June 2002. Rutgers, Roland. 1993. The Chepang Verb. A Synchronic Analysis with some Diachronic Observations.ActaLinguisticaHqfniensia 26: 107-137. - - . 1998. Yamphu: Grammar, Texts and Lexicon. Leiden: Research School CNWS.
218
RENE HUYSMANS
Tolsma, Gerard J. 1997. A Grammar of Kulung. Universiteit Leiden: Unpublished doctoral dissertation, 2 June 1999. Turin, Mark. 2006. A Grammar of the Thangmi Language. Universiteit Leiden: Unpublished doctoral dissertation, 17 May 2006.
ERGATIVITY IN KUNDAL SHAm, KASIDvliRI AND IllNDKO KHAWAJAA REHMAN INTRODUCTION
The present paper focuses on ergative patterns in three languagesKundal Shahi, Kashmiri and Hindko-as they are spoken in the district of Neelam in Pakistan-administered Kashmir. The estimated population of the region is 154,000, 1 while according to the 1998 Census of Pakistan, the population was 124,000. The total length of the valley is about 150 kilometers. Both banks of the river Neelam are populated, with a few villages on the left bank under Indian control. The languages spoken in the area include Hindko, Kashmiri, Kundal Shahi, Gojri, Pashto, Shina of Phulwei and Guresi Shina. Hindko is the predominant language of the district and virtually all members of the other language groups are bilingual in Hindko (Akhtar and Rehman2007). This linguistically rich area has long been inaccessible to linguists. Before partition, the region was inaccessible due to the absence of a road link; after partition, and until the recent devastating earthquake, outsiders (and foreigners in particular) were not allowed access to the region for security reasons. Therefore, no baseline linguistic literature is available on the languages and language varieties spoken in the region, and the area is not covered in Grierson's (1915) or O'Leary's (1992) surveys. Most of the Indo-Aryan languages spoken in the Indian subcontinent are characterised by split ergativity (Butt and Deo 200 1, Masica 1991). In this article, I investigate to what extent the language varieties of the Neelam Valley conform to the patterns of ergativity found in the wider region. In a split ergative system, ergative and accusative alignment alternates within a given language, depending on one of several possible factors (Holmer 2001: 104). Many of the languages spoken in South Asia show morphological ergativity (Butt and Deo 2001) which mani1 Personal communication from officials of the Departmertt of Planning and Development, Muzaffarabad, Azad Janunu and Kaslnnir, March 20, 2006.
220
KHAWAJA A. REHMAN
fests itself through marking the lone argument of an intransitive verb (S) in the same way as the less agent-like argument of a transitive verb (0), namely in the Absolutive case, with different marking for the more agent-like argument of a transitive verb (A), namely in the Ergative case (Dixon 1994: 16). Essentially, when there is a morphological case distinction between ergative and absolutive (Peeters 1995), we may speak of morphological ergativity. Moreover, the verb in an ergative clause is likely to agree with the absolutive argument. However, in northern Pakistan, quite a few languages have ergative marking in which the verb agrees with the A and not with the 0. This is a ftrst attempt to describe and compare the ergative patterns of these languages. We will see that they share the feature of split ergativity based on aspect, that is, A is marked with the ergative case when the verb form is in perfective aspect (simple past and all perfect tenses), while a nominative-accusative pattern occurs elsewhere. However, there are also differences among these languages concerning agreement patterns, the use of the ergative markers, ergative marking with certain intransitive verbs, and alternation between the dative and ergative marking in certain constructions. The data reveal some interesting patterns which are distinct from other languages in the area. KUNDAL SHARI
Kundal Shahi is a village located in the Neelam Valley, 74 kilometers from Muzaffarabad, the capital city of Pakistan-administered Kashmir, and seven kilometers from Athmuqam, the district headquarters. An ethnic group known as the Qureshi, speaking the Kundal Shahi language makes up the majority population in Kundal Shahi village. The language is named after the village (Rehman and Baart 2005). There are also some other communities, but they do not speak the language as their mother tongue. The Qureshi claim to have migrated from the town of Tajjar in Indian-administered Kashmir some three centuries ago (Rehman and Baart 2005, Akhtar and Rehman 2007). However, in summer 2008 I conducted a detailed survey of the community in which some of the members claim that their ancestors either came from Chilas or Muzaffarabad rather than the Valley of Kashmir. No other variety of the language is known. While a frrst publication on the language suggests that the community comprises approximately
ERGATIVITY IN KUNDAL SHARI, KASHMIR! AND HINDKO
221
1,500 to 2,000 people (Rehman and Baart 2005), my current detailed census shows that the total population of the Kundal Shahi community is around 3,371living in 537 households that are scattered throughout eight mohallas 'hamlets', namely: Rait, Graan, Gujhaan, Sinji Nakka, Dollur, Frashian/ Khujhaani, Gheelan and Sattra. Kundal Shahi displays the areal feature of split ergativity, where the split is based on aspect of the verb.
Perfective aspect: Otherwise: la.
ma
ergative nominative-accusative
fer
fad-6:.
lSG.NOM house.F build-PRES.lSG.M 'I am building a house.' lb. ml fer lSG.ERG house.F 'I built a house. '
fad-i build-PST.3SG.F
The ergative sufftx /-an/, homophonous with the analogous sufftx in Kashmiri, is attached directly to the noun stem. lc.
zaman-an
minif khe:gin Zaman-ERG.M food.F eatPST.3SG.F 'Zaman ate food.'
ld. zaman-an khe Zaman-ERG.M eatPST.3SG.F 'Zaman ate food.'
tso_t.
[Kundal Shahi]
[Kashmiri]
food.F
The verb agreement follows the common ergative pattern, i.e. the verb agrees with the 0 in transitive clauses. le.
lf.
fad-i
ml fer lSG.ERG house.F 'I built a house. '
build-PST.3SG.F
haft
fad-i
fer
3PL.ERG house.F build-PST.3SG.F 'They built a house.'
In Kashmiri, all As in perfective clauses take ergative marking. In Kundal Shahi, the second person singular personal pronoun is un-
222
KHAWAJA A. REHMAN
marked for ergativity, while Panjabi (Butt and Deo 2001) and Hindko do not have ergative marking for the ftrst person singular pronoun. 2a.
tu
mist
kra:m
2SG.NOM good workM 'You did a nice job.' 2b.
tu
red
theg. do.PST.3SG.M
Stir
2SG.NOM tomorrow town 'You will go to town tomorrow.'
Ms. go.FUT.2SG
Thus far, the available data from Kundal Shahi suggest that the ergative pattern of the language essentially follows the common patterns of the Indian subcontinent: with many predicates, the A is marked ergative in perfective clauses. KASHMIR! The data below are taken from my own variety of Kashmiri, which is spoken in the district of Neelam in Pakistan-administered Kashmir. Kashmiri is the second largest language of the district, and shares a common pattern of split ergativity as illustrated in examples (ld) above and (3a-3c) below. 3a.
!xJ
ma:r-:J
lSG.NOM kill-FUT.lSG 'I will kill a snake.' mo:r 3b. me lSG.ERG kill.PST.3SG.M 'I killed a snake.'
sarA£
snake.M sarA£
snake.M
ch-o bAt:J 3c. me lSG.ERG be-PRES.3SG.M rice.M 'I have eaten the rice.'
khyomot.
eatPSP.M
This is not the whole story, however, as Kashmiri also has certain unique features that distinguish it from the other languages under discusston. Unlike the other languages of the region, Kashmiri shows a different kind of verb agreement in ergative constructions. In such constructions, the verb always agrees with the 0. In cases where the A is sec-
ERGATIVITY IN KUNDAL SHARI, KASHMIR! AND HINDKO
223
ond person (Koul 2004), the verb must also agree with it, and when the A is a third person plural, then the verb agrees with it only optionally, unless the 0 is frrst person singular, in which case agreement with the 3PL A is obligatory. The verb shows agreement with the A through a suffix attached directly to the verb. The 0 agreement is incorporated into the verb, and the A agreement suffJX also shows 0 agreement. Table 1 below shows the full paradigm of the A agreement suffJXes.
T,a ble 1
. Kashmiri 0 andA agreement mar~ers m 0
lsg A 3pl
2sg 2pl
-has -thas -u:s
lpl
2sg
2pl
-hakh
-rn
f
m -ikh -ith
-Akh
-u
-o
-~th
3sg m -okh -oth
f -~kh
-~th
3pl
f
m -ikh -ith
-~kh
-u
-o
-u
-~th
Below are some examples of the above agreement patterns. 4a.
homyo
woeh-has
3PL.ERG see.M-PST .1 SG.3PL 'They saw me.' (masculine) 4b.
4c.
tohy
wxh-u:s
ba.
2PL.ERG see.F-PST.1SG.2PL 'You saw me.' (feminine)
1SG.NOM
homyo
AS
wxhmat.
1PL.NOM
see.PSP.1PL.M
ch-ikh
3PL.ERG be.PRES-1PL.M.3PL 'They saw us.' (masculine) 4d homyo ch-:1kh 3PL.ERG be.PRES-1PL.F .3PL 'They saw us.' (feminine) 4e.
ba. 1SG.NOM
tse
ch-o-thas
2SG.ERG be-PRES.M-2SG.1SG 'You have seen me.' (masculine)
AS
wxhmats.
1PL.NOM
see.PSP.1PL.F
b:J
woehmot.
1SG.NOM
see.PSP.1SG.M
224
KHAWAJA A. REHMAN
tse
4f.
4g.
4h.
b:J
woehm:Jts.
2SG.ERG be.PRES-F-1SG.2SG 'You have seen me.' (feminine)
ch-:rthas
lSG.NOM
see.PSP.lSG.F
tohy
wxhu
phil:m:J.
2PL.ERG see.F.PST.3SG.2PL 'You saw a ftlm.' (feminine)
ftlm.F
tohy
wxhmats phil:Jm
asu
2PL.ERG be.F.PST.3SG.2PL see.PSP.3SG.F 'You had seen a ftlm.' (feminine)
ftlm.F
In the variety ofKashmiri spoken in the Neelam Valley, 3PL As normally trigger agreement, even if the 0 is not lsG. In other dialects of Kashmiri, A agreement is less common in this context
homyo
Sa.
chi
AS
3PL.ERG be.PRES.lPL.M lPL.NOM 'They saw us.' (masculine) (cf. 4c)
em
Sb. homyo AS 3PL.ERG be.PRES.lPL.F lPL.NOM 'They saw us.' (feminine) (cf. 4d)
wxhmat see.PSP.lPL.M
wxhmats. see.PSP.lPL.F
In ergative constructions with A agreement, the verb agrees in person, number and gender with 0, whereas it agrees only in person and number with A This type of verb agreement exists in Sindhi, Siraiki, Marathi and some Eastern Uttar Pradesh dialects of Hindi. 2 When A is lSG, 3SG or lPL the verb agrees only with 0. The second unique feature of Kashmiri that distinguishes it from other languages in the region is the obligatory ergative marking of the subjects of some intransitive predicates, usually referred to as S. Verbs displaying this property tend to be associated with semantic ftelds such as animal noises, bodily functions and undirected motion. Some such verbs are listed in Table 2, and illustrated in example 6 below.
2
Peter Hook, personal communication 23 May, 2006.
ERGATIVITY IN KUNDAL SHARI, KASHMIR! AND HINDKO
225
ExampJeo 1 f Kashmiri m . 1ranst.tive verb s wt"thErgaf IVe S so:C 'think' gind 'play' 'scratch' 'dance' k:Js natsh 'chew cud' 'romp' dromar kud 'bathe' 'frolic' n:Jmw drak wa.fua:w 'warm oneself' bptsh 'squirm' 'laugh' 'limp' AS Jagg 'weep' 'fart (silent, but...)' wad s:JS 'sob' 'fart (audible)' nek t~1h 'bark' 'urinate' wa:r mAW 'howl' 'defecate' Wli!g char 'whimper' 'pant' czi!k sa:gk 'quarrel' 'say nasty things' tsow wa:r 'bleat' 'boast' thek .ta:nk ifa'Jk 'bray' kr.Jh 'the sound a hen makes, especially after laying an egg'
T,a ble 2
6a. me lSG.ERG 'I laughed.'
laugh.PST.3SG.M
6b.
nots-okh
tiiiJ{)
OS.
dance.PST-3SG.M.3PL 3PL.ERG 'They danced.'
6c. kamal-an gyund. Kamal-ERG play.PST.3SG.M 'Kamal played.' 6d. Su.r-yo motr-okh children-ERG urinate.PST-3SG.M.3PL 'Children urinated on the bed.' 6e.
khar-an
ifo:gk
donkey-ERG bray.PST.3SG.M 'The donkey brayed.'
6f.
me
wa.fuow. lSG.ERG warm.PST.3SG.M 'I warmed myself.'
bistr-as
pe_th
bed-OBL
on
226 6g.
KHAW AJA A.
ba:n¢-o
giw-okh
REHMAN
sarsay doh-as.
mustctans-ERG sing.PST-3SG.M.3PL all 'The musicians sang all day.' 6h.
g:J: myo
day-OBL
tsuw-:Jkh
villagers.ERG quarrel.PST-3SG.F.3PL 'The villagers quarreled.' Note that with the exception of ts'ow, illustrated in example (6h) above, which displays feminine agreement, all of the verbs show masculine agreement, regardless of the gender of the ergatively-marked argument. When the ergative argument is second person singular or plural, or even third person plural, the verb agreement conforms to that shown in example 6 and Table 1 above. Although ergative Ss with certain verbs are found in some South Asian languages, it is only in K.ashmiri that ergative marking is obligatory. This manner of ergative marking also occurs in certain North and South American languages, as well in Caucasian languages and Acehnese (Andreasson 2001, Hook and Koul2004). Dixon (1994) refers to the pattern where the S of an intransitive predicate takes the ergative case as a 'split S' system, although it is also known as 'active alignment'. Languages exhibiting this pattern include Chickasaw, Lakhota, Guarani, Acehnese and Georgian (Andreasson 2001). In Georgian (Hillary 2004), the phenomenon only occurs with verbs in the aorist. In Kashmiri, it occurs with verbs depicting non-continuous situations occurring prior to the speech act, i.e. in the simple past as well as perfect tenses, where the situation depicted occurs prior to some other specified reference point. Ergative marking on intransitive Ss as found in other languages is usually semantically conditioned and varies according to language specific criteria (Andreasson 2001). In common with other languages showing active alignment, ergative marking with a univalent predicate has a semantic basis in Kashmiri. If we look at these verbs (Table 2 above), we can conclude that all the predicates depict controllable and volitional acts which do not involve any kind of achievement, and the acts are not seen as directed towards a specific goal. Ifthis were correct, then verbs such as char'defecate', and mAl':Jr'urinate' would not be expected to fall into this category. In fact, in Kashmiri these verbs also have a counterpart NP+V in com-
ERGATIVITY IN KUNDAL SHARI, KASHMIR! AND HINDKO
227
plex verbs (e.g. mutrun +karun 'to pass urine (lit do urine)'). In short, when an activity is viewed as an achievement, the complex verb is used On the other hand, some intransitive verbs depicting a change of state or location take a nominativeS. Some of these are listed below.
Table3 Sul]g wath beh
atsh .thAmr WAS
let h:Jl khiSAk du:r
K.ashmiri verbs with Nominative 'go to sleep' ts:Ji 'rise I stand up' dr.J,r 'sit down' w:Jif 'enter' tAr 'stop' ner 'descend' sed 'lie down' khas 'move' pakh 'slip away' pher 'fade into the distance' ros
Ss 'run away' 'run' 'fly' 'swim' 'exit' 'straighten up' 'climb I ascend' 'walk' 'return' 'become angry and shun'
Dative As and Ss are a very common feature in the languages of the region. In Hindi-Urdu and Hindko, some verbs allow the A or S to take either ergative or dative marking. K.ashmiri does not exhibit an alternation of this kind. 7a.
7b.
gol-as
cho
gar:J
gatshon
Gol-DAT be.PRES.3SG.M home 'Gol has to go home.'
go.INF.3SG.M
*gal-an
cho
gatshon
Gol-ERG
be.PRES.3SG.M home
gar:J
go.INF.3SG.M
While in K.ashmiri, example (7b) above is not acceptable in any context, in the following section, we may note that this kind of construction is very common in Hindko. HINDKO The dialect of Hindko spoken in the Neelam Valley is known locally as Parimi, Paarim, Pahari and sometimes as Hindko (Akhtar and Rehman 2007). This dialect has traditionally been considered to be a variety of the Pahari language spoken in the Murree Hills, but an analysis of word lists collected in 2004 in collaboration with Joan Baart reveals
228
KHAWAJA A. REHMAN
that the language is in reality closer to the variety of Hindko spoken in the Kaghan Valley, as suggested by Bates (1873: 181). The Kaghan Valley parallels the Neelam Valley immediately to the west in theregion of Hazara. In ergative clauses, Hindko uses the ergative postposition SUIJ. Unless the A is frrst or second person singular, the NP will be in the oblique form. However, use of this ergative marker, unlike in HindiUrdu and Panjabi, seems to be optional in Hindko. My ongoing research indicates that the marker tends to be dropped in informal conversations without any apparent semantic basis. However, the marker is retained in formal or careful conversations and in story telling. The counterpart of SU1J in Hindi-Urdu and Panjabi is ne, but in these languages its use is obligatory and it may not be dropped. 8a. lo:k-a sUIJ ru.ti kha-di. people-OBL ERG food.F eat-PST.3SG.F 'The people ate food.' 8b. lo:k-a ru.ti kha-di. people-OBL food.F eat-PST.3SG.F 'The people ate food.' Both of the above examples are correct in any context The omission of the marker is not limited to the Hindko variety of the Neelam Valley since the same phenomenon is found in Hazara. I speculate that it is in the interest of economy that Hindko tolerates the omission of the ergative marker in clauses of this kind, as the ergativity of the A is fully recoverable from its oblique form. According to Peter Hook, 3 marking of agents with ne is an innovation in Panjabi and Hindi-Urdu. Earlier stages of these languages simply used the oblique forms of the agent nmm or pronmm, and it appears that Hindko preserves the more archaic construction in some contexts. Further research is required to determine the semantic basis, if any, for this alternation. The second person singular agentive tud is the only pronoun in Hindko which is restricted to ergative constructions. The first person singular personal pronoun me occurs in the nominative case in such
3
Personal communication 19 June, 2006.
ERGATIVITY IN KUNDAL SHARI, KASHMIR! AND HINDKO
229
constructions, and neither tud nor me pennit the use of the ergative marker SU1J. 9a.
tud kita:b 2SG.OBL book.F 'You read a book.'
9b.
*tudsUIJ 2SG.OBL ERG
book.F
me
kita:b
9c.
kita:b
1SG.NOM book.F 'I read a book.' 9d.
kita:b *meSUIJ 1SG.NOMERG book.F
par-i read-PST.3SG.F par-i read-PST.3SG.F par-i read-PST.3SG.F par-i read-PST.3SG.F
Whether or not the ergative marker is used, the verb always agrees with the 0. While in Kashmiri, a group of intransitive verbs obligatorily mark the S as ergative, Hindko does not have such a group of verbs. However, in Hindko the verbs hansna 'to laugh', ro:lJB 'to weep', khaggfd 'to cough' and many other verbs appear in ergative constructions when the S performs the action intentionally. This is also true of Hindi-Urdu, with verbs such as khansna 'to cough' (Butt and Deo 2001 ). Hillary (2004) is also of the view that in certain situations, the speaker may choose whether to mark the subject as agentive or 'patientive' for any given intransitive verb, with agentive marking implying a degree ofvolition and patientive marking implying a lack of volition or control. In Hindko, when the activity is viewed as being carried out on purpose or it involves conscious choice on the part of the subject, the speaker uses the ergative construction. A system in which intransitives can optionally take an ergative argument is referred to a 'fluid-S system' (Dixon 1994, Hillary 2004). In Hindko, however, such ergative marking is not possible with all intransitives. Verbs such as turna 'to walk', naSIJfil 'to run away', Aji;Fa 'to sit', ufhfiJ 'to rise',julna 'to go', leiJB 'to 'descend' and CEJf11E1 'ascend' never occur in ergative constructions.
230
KHAWAJA A. REHMAN
lOa. u hasya. 3SG.NOM laugh.PST.M 'He laughed.' lOb. us SUIJ hasya. 3SG.OBL ERG laugh.PST.M 'S/he laughed.' On the basis of these examples, Hindko appears to exhibit a partial fluid-S pattern. In Hindko, dative As and Ss can be replaced by ergative NPs (examples 11a-11d) when the speaker wishes to emphasise the intentionality of the predicate. The same phenomenon is also found in HindiUrdu (Bashir 1999, Butt 2005). This use of an ergative with an infmitive+auxiliary construction is not restricted to transitive clauses, as intransitives can also take aS in the ergative (llc and lld). lla. us SUlJ Aj kita:b par-ni e. 3SG.OBL ERG today bookF read-INF.3SG F be.PRES.3SG 'Today s/he wants to read a book.' llb. us ku A] kita:b par-ni 3SG.OBL DAT today bookF 'Today s/he has to read a book.'
e. read-INF.3SG F be.PRES.3SG
llc. l.l118 SUIJ k:Al ghar jul-na e. 3PL.OBL ERG tomorrow home.M go-INF be.PRES.3SG 'They want to go home tomorrow.' (plan+intention) lld. l.l118 ku kAI ghar Jul-na 3pl.obl dat tomorrow home.m go-inf 'They have to go home tomorrow.' (obligation)
E.
be.pres.3sg
In Urdu, the ergative can appear in infmitive+auxiliary constructions in the Lahori and Delhi dialects (Bashir 1999). The semantic basis for this dative/ergative alternation is explained by Butt (2005) as goal versus control.
zu ]a-na 12a. nadya=ko Nadya.F.Sg=Dat zoo.M.Sg.Loc go-Inf.M.Sg 'Nadya has/wants to go to the zoo.' [Dat- Goal]
he. be.Pres.3.sg
ERGATIVITY IN KUNDAL SHARI, KASHMIR! AND HINDKO 12b. nadya=ne zu ja-na Nadya.F.Sg =Erg zoo.M.Sg.Loc go-Inf.M.Sg 'Nadya wants to go to the zoo.' [Erg- Control] (Butt 2005: 4)
231
he. be.Pres.3.sg
Bashrr ( 1999: 24) analyses the semantics of ne according to grammatical person categories: 1st person: conscious choice by the subject; -
2nd person: questions and conditions: potential conscious choice by the subject; and assertions: action anticipated by the speaker (potential conscious choice by the subject), action desired/ enjoined by the speaker;
-
3rci person: action expected by the speaker.
The use of a dative A or S involves some external obligation. Such an explanation of the alternation between dative and ergative As and Ss is much more plausible, and it is worth exploring whether there is a straightforward correspondence with deontic modality. The dative/ergative alternation in Hindko is similar to that found in the Urdu/Hindi system. CONCLUSION The complex agreement phenomena in ergative clauses in Neelam Valley K.ashmiri require considerable further investigation, particularly in univalent constructions. The semantic basis for alternative case marking strategies in Hindko also needs additional analysis. This is the frrst ever study of the ergative patterns of these language varieties, and it is my aim to introduce these speech forms to the academic community and to invite more extensive studies of these hitherto undescribed languages.
232
KHAW AJA A. REHMAN ABBREVIATION
1SG 2SG 3SG 1PL 2PL 3PL
A
AUX DAT ERG F
1st person singular 2nd person singular 3rd person singular 1st person plural 2nd person plural 3rd person plural more agent-like argument of a transitive verb auxiliary dative case ergative case feminine
INF
M NOM 0
OBL PRES PST FUT PSP
s
infinitive masculine nominative case more patient-like argument of a transitive verb oblique present past future past participle intransitive subject
BIBLIOGRAPHY Akhtar, Raja Nasim and Khawaja A. Rehman. 2007. The Languages of the Neelam Valley. Kashmir Journal ofLanguage Research 10: 65-84. .Andreasson, Daniel. 2001. Active Languages. Stockholm University. Thesis for the degree of Bachelor of Arts in General Linguistics. http://www.ling.su.se/gulkmsmaterial/311_4/active.pdf Bashir, Elena. 1999. The Urdu and Hindi ergative postposition ne: Its changing role in the grammar. In Rajendra Singh, ed, The yearbook of South Asian languages and linguistics. New Delhi: Sage Publications, 11-36. Bates, Charles Ellison. 1873. A gazetteer of Kashmir and af#acent districts of Kishtwar, Badrawah, Jammu, Noashera, Poonch and the valley of the Kishan Ganga. Calcutta: Office ofthe Superintendent of Government Printing. Butt, Miriam and Ashwini Deo. 2001. Ergativity in Indo-Aryan http://www.cogsci.edac.uk/%7Esi.amak:r/Kmdish/KURDICA/2001/31i.a-erg.html.. Butt, Miriam. September 2005. The Dative-Ergative Connection http://ling.uni-konstanz.de/pages/home/butt/cssp05-hnd.pdf Dixon, R.M.W. 1994. Ergativity. Cambridge: Cambridge University Press. Grierson, G.A. 1915. The Linguistic Survey ofIndia, Vol. 8, Part 2. Calcutta. Hillary, P.J. 2004. The Georgian language: An oudine grammatical sununary. http://www.anna.zi..com/Georgi.an/ (via http://www.anna.zi..demonco.uk/). Holmer, Artlrur. 2001. The ergativity parameters. Working Papers 48 {2001), Lund University, Department of Linguistics, 101-113. Hook, Edwin Peter and Omkar N. Koul. 2004. Case as agreement: non-nominative subjects in eastern Shina, non-dative objects in Kashmiri and Poguli, and labile subjects in Kashmiri and Gujarati intransitive inceptives. In P. Bhaskararao and K.V. Subbarao, eds., Non-nominative subjects. Amsterdam: J. Benjamins, 213-225. Koul, N. Omkar. 2004. Kashmiri: A grammatical sketch. http://www.ills.org/pdf)TheKashmiriLanguage.pdf
ERGATIVITY IN KUNDAL SHARI, KASHMIR! AND HINDKO
233
Masica, Colin. 1991. The Indo-Aryan languages. Cambridge: Cambridge University Press. O'Leary, Clare F., ed 1992. Sociolinguistic Survey ofNorthemPakistan. 5 volmnes. Islamabad: National Institute of Pakistan Studies and Summer Institute of Linguistics. Peeters, Bert. 1995. Morphological vs. syntactic ergativity. Department of Modern Languages, (French) University of Tasmania, Australia. http://www.sfs.nphil.uni-tuebingen.dellingui.st/issues/6/6-642.html. Rehman, Khawaja A. and Joan L.G. Baart. 2005. A first look at the language ofKlllldal Shahi in Azad Kashmir. SIL Electronic Working Papers (SILEWP 2005-008). http://www.sil.org/silewpl2005/silewp2005-008.pdf.
KENHAT, TIIE DIALECTS OF UPPER LADAKH AND ZANSKAR1 BETTINA ZEISLER
According to phonetic features alone, the various dialects spoken in Ladakh are presently classified in roughly two main groups: -
Western Archaic Tibetan: the non-tonal 'conservative' dialects of the northeastern and central areas: Baltistan, Purik. Lower Ladakh, Nubra, and Leh. showing initial and final consonant cluster~
-
Western Innovative Tibetan: the 'innovative' dialects of the south-eastern areas: Upper Indus, 3 Changthang, and Zanskar, where the clusters have been reduced and tonal features can be found4
1 I should like to thank Roland Bielmeier, who first alerted me to some phonetic and morphophonemic similarities between what I observed in Cemre and Gya and the Zanskar, Spiti, and Mustang varieties. His student Thomas Preiswerk similarly obliged me by sharing his recent observations on the Zanskar dialects wi1h me. I am even more indebted to Rebecca Norman for all the discussions we have had on the Ladakhi varieties and for her valuable comments on this paper. I should also like to express my gratitude towards all consultants, since without their help and patience, this paper could not have been written. Special thanks go to Mengym Tshomo who, by her own initiative, demonstrated an extraordinary diachronic awareness, which allowed us to catch an important linguistic change red-handed I am also very grateful to my colleague Sam Featherston in the fonner SFB 441 and to Mark Turin for many improvements in style. Finally, I should like to thank the anonymous taxpayer, who supported the fieldwork on which this article is based: 1996 via grants from the state Berlin (NaFoG) and the DAAD, 2004-2008 via the DFG, for a research project within the SFB 441 'Linguistic Data Structures': On the Relation between Data and Theory in Linguistics at the University of Tiibingen (http://www.sfb441.mri-tuebingen.de/b11/), and again in 20 10 by a grant from the DFG. 2 Spoken clusters or their traces in the clusterless dialects will have to be discussed in relation to their Old and Classical Tibetan written cmmterparts. Following conventions of traditional Tibetan grammar, the segments of a written syllable are defined as radical (or root consonant: all30 letters of the alphabet allowed), pre-radical (5 pre-scribed and 3 super-scribed letters),post-radical (4 sub-scribed letters, still preceding the vowel slot), final (1 0 letters allowed at the position after the vowel slot), and post-final (2letters allowed, in combination only with some of the final letters). 3 This designation is used here solely in relation to Ladakh, not in relation to the total comse of the river. More commonly, the designation 'Upper Indus Valley' is applied in relation to Pakistan, referring thus to Gilgit and Chilas. 4 Cf. Bielmeier (2004, Appendix). Western ltmovative Tibetan also comprises the Tibetan varieties of Himachal Pradesh and Uttarakhand. Bielmeier's classification is based on earlier work by the Hungarian scholars Ligeti, Uray, and R6na-Tas (cf. e.g.
236
BETIINA ZEISLER
Representing the allegedly 'most archaic' dialects (cf. Bielmeier 1985: 15), the 'conservative' dialects are taken to be more relevant for thereconstruction of Old Tibetan or even *proto-Tibetan. The classification is highly misleading, and the so-called 'innovative' dialects are for various reasons as relevant or perhaps even more relevant for the reconstruction of the proto-language, not least because of their syntactic properties (see especially sections 4-6 and 8 below). On the basis of my fieldwork in Ladakh and some historical considerations (Zeisler forthcoming, Zeisler in preparation), I propose a somewhat more refmed version of the above classification, distinguishing between -
The group of historically younger, but lexica-phonetically conservative Shamskat dialects (Samskad 'language of Lower' Ladakh) in the west and north of Ladakh: Purik, Sham, and Nubra
-
The historically older, lexica-phonetically partly conservative, partly innovative Kenhat dialects (Gyenskad 'language of Upper' Ladakh) in the centre, south, and south-east of Ladakh5
The Kenhat dialects are closely linked to Tibetan varieties spoken in Himachal Pradesh, with which they probably share a common history. My observations indicate that the Changthang dialects share many grammatical features with the Kenhat dialects, but may show particular lexical traits. The verbal auxiliary motphology shows Central Tibetan influence, and I do not want to preclude that some of the Changthang dialects may be closer to Central Tibetan varieties. 6 However, no historical or linguistic data is presently available to state anything with certainty. The Balti dialects spoken in Pakistan and in the region Rona-Tas 1966: 21f). For critics of this classification see below. In any case, I should prefer the attribute 'lexica-phonetically conservative', since the Ladakhi dialects are highly innovative at the morphological and syntactical level. 5 The Cemre consultant called her own speech 'Gyenskat' (with Leh pronunciation), and the name was also accepted by the Gya consultant. The term is not very common among outsiders, but the alternative term Stot(pa), as given by Koshal (1979: 1, 1990: 14) and Bielmeier ( CDTD: ii "Stot or Ken" for the dialect of "Igu'') is likewise not generally known, nor is it accepted by the speakers themselves. People from Lower Ladakh would call the people of the Upper Indus invariantly Tfaythaypa and 1heir language accordingly lJaythayjxJ skst In contrast to the local designations of Sham and Gyen/K.en, I will use the terms Shamskat and Kenhat as cover terms for a wider range of dialects. For 1he pmpose of the present paper, I will confine them to the region of Ladakh, or more precisely to the two Hill Council districts Leh and Kargil (the latter comprising the regions ofPurik andZanskar). 6 Cf. also LSI (p. 52), stating that the Zanskar dialects agree with Rong, but that in the nomadic area of"Rubshu [... ]a form of Central Tibetan is spoken"
237
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
around Kargil in Ladakh (to a certain extent also in Balti settlements in the Leh district) should perhaps be classified as a special subgroup of the Shamskat dialects, as they show significant differences in the use of verbal auxiliaries.
sound ch
]',a ble 1
an~es
laryngalisation
·Knhe e a m
de-pala- palatatalisation lisation
w > fi
j > fi
j >0
Sham
-
-
0-1,e
LEH
-
-
0-i,e
-
PIP HML
? -
MND CEM
? ?
-
? -
-
+
-
SHA
fi-a,o,u
-
GYS
+ +
+ +
ft-a,o,u
ft
-
NYO
+
-
-
J > nr;: ~ > fij
section
2.1
J > r;: _y
fricativisation of clustel initial vel
medial
(sub-) phonernic tone
vcd
cluster
cluster
lU
cluster
-
cluster
y: xd,xb
cluster
X.,
(all) (all) 11
-
x,
y,ei,p
X.,
11
(all/ 1 vel
X.,
y,p
X. -
-
vel
my
-
-
vcl,rny 12
2.2
-
+ + +
3
7 'Change' is Wlderstood here in relation to the oldest documented language stage, i.e. Old Tibetan (OT), ca. 650-1050 CE. Particularly in the case of 'laryngalisation' and 'palatalisation' it may well be the case that these features reflect an even older, archaic stage of the language. 8 'vel' = voiceless, 'vcd' = voiced, ''x" = preserved pre-radicals. All other abbreviations are explained at the end of this paper. 9 One should, however, note the sporadic Purik form /jak-/ for zag- ' day', which may also appear infrequently in other dialects of western Sham. 10 Note, however, l1Ja ~ /ya/ 'five' 11 Fricativisation appears to be somewhat 1D1predictable or irregular in Manda at the medial position (the position that enhances the preservation of original clusters). I do not have data for the other two dialects, but the pattern may be quite similar. 12 The consultant uttered fricatives every now and then at the morpheme boWldary, but did not accept this consciously as a feature ofhis dialect.
238 T,a ble
BETIINA ZEISLER
Kenha t morpJhopJhonenucs
2
loss of final -s after consonant
vowel
p,m, cluster>o assimil. umlaut of stem k,IJ -IJS -ks
deliniteness
genitive agent -se gen.
Sham
-
-
-
-
-
{po}
-
-
LEH
-
-
-
-
-
{po}
+
-
PIP
+
+
?
-
{po}
+
?-
-
?
+
HML
+
+
+
diphthong diphth.
MND
± 13
-
-
diphth.
-
CEM
+
+
-
+
+
{de}
+
+
SHA
+
+
-
+/t/o
+
{de}
+
+
GYS
+
NYO
+
section
+ -
14
?-
past
future
suk, khantsuk tok, k(j)ak, ok,anok kanok nok, {kak} ?, {kak} kan ? ?,hak
no data available
-
+/t/o
+
{de}
+
+
-
(+)
(-)
{de}
+
+
2.3
evidential marker
7
8
{kak} kan
{tok}, kak, .{kanak}
no data available ak, {kak} {tok}, {kak}, {kanak} kan ak, {kak} CUI), SOt) {kan} tok, kak,? 10
In this article, I present the main features characteristic of the Kenhat dialects, which distinguish them from the Shamskat dialects in so many ways that they may be thought of as two different languages. After describing the geographical setting ( 1), I will present the most obvious phonological features of the Kenhat dialects (2; see also Table 1) and will discuss the problem oftonogenesis (3). Section 4 deals with the preservation, generalisation, and eventual loss of Old Tibetan verbal sufftxes in the two dialect groups. The following sections 5 (Frozen clusters in Gya compounds) and 6 (Substitution offmal-s by 1-t/ or 0) demonstrate the importance of the so-called 'innovative' varieties for the reconstruction of Old Tibetan. The lexical differences will be exemplified with the definiteness marker, remote deixis, and frrst person inclusive plural pronoun (section 7). The more deeply rooted morphological and 13 Although final clusters can be found as alternative forms, the general tendency is for the loss of post-final-s in Hoshi & Tondup Tsering (1978). 14 In a few cases, the coda was reduced to a mere nasalisation of the vowel, cf. /fiO/ h01Js ' threshed grain' or even /si5al somaiJs 'comb', in contrast to !karjl gQI}s ' ice' or /rolarjl roiQI}s ' zombie', but the vocabulary would have to be checked more systematically to find out whether the latter pronunciation is not due to Leh influence or whether the nasalisation could not also result from simple coda consonants.
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
239
syntactical differences are discussed in sections 8 (Bimotphemic case marking and case neutralisation), and 9 (Marking oftense and evidentiality). The two tables above may serve as a guide through the paper.
1. THE GEOGRAPHICAL SETTING
The exact geographical extent and distribution of the Kenhat dialects is as yet unknown. For the Ladakhis, the main centre lies in the vicinity of Hemis Gompa (ca. 45 km south-east of Leh): Kharu, Cemre, Sak:ti on the right side of the Indus; Hemis, Upshi, and Martselang on the left side. The dialects extend to the south-east along the right riverbank over Igoo and Shara, possibly including the Ronghat area (Royskad 'language of the gorges'), 15 with the villages Hemya (located on both sides of the river), Liktse, and Kyunggyam on the right, Tarshit, Tiri, and Kere on the left side of the Indus. The neighbouring dialect of the Nyoma nomads differs mainly in phonology and does not show any fricativisation. The use of the past tense egophoric marker /tfug/ and its counterpart /sog/ indicates a certain proximity to Central Tibetan, but the Nyoma dialect also shares the grammatical feature of genitive agent marking and the use of particular evidential auxiliaries with the Kenhat dialects. To the northwest, the most prominent phonetic feature, fricativisation of initial clusters (see section 2 below and appendix), extends only as far as Thik:tse and upper She on the right bank (about 25km and 15km south-east of Leh respectively), and perhaps up to Matho on the left bank (about 25km south of Leh). But the central dialects around Leh show at least regular fricativisation of clusters with a voiced velar radical. The LSI (p. 52, 54) also mentions fricativisation of clusters
15 For my Gya consultant, the geographic term /Ro:g/ describes a very narrow valley along the Indus. Francke (190 1: 6) mentions the "Rongpa dialect" as situated at the "upper-most Indus valley right bank" The LSI refers to the "dialect to the east of the Leh dialect" (p. 52) or to the "dialect of the upper-most Indus valley'' (p. 53). But the name Rong might be used also for other narrow side valleys and can be found for a place between Gya and Miru (Nest & Wings trekking map, Army Map Series, Washington, U502 India & Pakistan, second edition, NI 43-12 U8589) or for the village Miru itself (sheet NI 43 of the International Series, 8111 edition, Government of Pakistan, Nelles Maps Himalaya, Nelles Mlinchen. and other maps). Zeisler (2004), thus erroneously uses the name Rang for the valley of Gya-Miru in contrast to Stod for the Upper Indus valley.
240
BETIINA ZEISLER
with a voiced labial radical, which I have never observed. While this statement in the LSI possibly reflects the situation of the day, it may rather refer less specifically to the villages in some distance to Leh. 16 In general, the dialects around Leh show a pronunciation closer to Shamskat dialects, but all of them share the grammatical feature of genitive agent marking and the use of particular evidential auxiliaries with Kenhat dialects. While Leh has been an important exchange station on the trade routes from and to Central Asia, Tibet, and Kashmir, from history it is known that the Ladakhi kings repeatedly settled a considerable number of people from Baltistan and Purik in villages around Leh, such as She and Chushot Both of these factors may have led to the observable mixture of what are basically two quite different languages. To the south, the Kenhat dialects comprise the dialect of the eastern side valley ofGya-Miru, Gyahat (Gyaskad) and the central and eastern dialects of the Zanskar valley, Zahare loou (Zanskari zlabo). Along the Indus, there is a clear geographical boundary between the two main dialect regions. Snyemo (or Nyimo), the first village of the Shamskat area, 17 lies in a sharply cut basin, the south-eastern boundary of which is formed by some lines of higher slopes and the very narrow gorge of the Indus river. On the left bank of the Indus rises a forbidding chain of steep mountains. The Zanskar river coming from the south-west and flowing into the Indus at Snyemo could not have served as an access route in the past above Chiling and the Markha valley, except for courageous traders trekking up the ice in midwinter when it was completely frozen. According to an individual from Kaya, 16 In the Kesar narration from Stok, one finds Kenhat features (including the occasional fricativisation of voiced labials) side by side with Shamskat features, but I remain unsure whether this is an idiosyncrasy of the narrator or rather typical of the village dialect. Stok is situated almost opposite Leh on the left riverbank. 17 Zeisler (2004: 607) erroneously placed the dialect boundary to the west "between Basgo (Bahsgo) and Saspol", following Francke's map (1904, plate vii), which shows the dialect boundary as a "line midway between Saspola and Basgo" (LSI p. 52). This could have made sense, since there is another geographical boundary or small pass to be crossed in order to leave the Snyemo-Bazgo basin. I have not yet worked on the dialects of Snyemo and Bazgo, but a young woman from Snyemo told me in 2005 that she always uses the instrumental marker for the agent, thus /khos/, never the genitive /khoi/, and she further described her speech as being quite similar to that of Saspol, except that the people of Saspol would always use the directional marker /-la/ instead of the contracted /-a/ (a feature that the Saspol consultant ascribes to the speakers of Alchi on the other side of the Indus).
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
241
the dialect of the Markha valley would correspond to the Shamskat dialect spoken in Bazgo and Snyemo and the area would have the same climate and economy as does Sham. Although the south-eastern slopes and the sand plains behind Snyemo were only a minor obstacle on the route to Leh, they nevertheless mark a boundary between different climate zones, and thus also between different economic zones: south-east of this boundary, there is no professional horticulture, fields yield only one harvest, and the further one proceeds eastward, the more semi-nomadism or full nomadism become dominant; north-west of the boundary, full nomadism never prevailed, and the economy, especially in the lower areas along the main rivers, is dominated by orchards (apples, apricots and walnuts, and in the lowest areas also peaches and grapes) and fields yielding two harvests. 18 The Zanskar valley itself is connected by old trade routes over Garzha (Lahul) in the south to Spiti and Kulu and by various other passes to the valley of Gya-Miru and then further on to the Changthang. A few trails over the north-eastern mountains connected Zanskar with Central Ladakh and were used for exporting butter, while a very difficult trail led to Kishtwar, used for the import of wood (Thomas Preiswerk, p.c.). To the north-west, the Zanskar valley is connected with the Balti-speaking areas ofPurik via the Doda river (the north-western tributary branch of the Zanskar river) and the Suru valley. The main dialect boundary between the Purik and Zanskar dialects lies after Parkachik (Thomas Preiswerk, p.c. ), the last village of the Purik dialect area. Although quite easy to master, the long ascent 18 I should like to emphasise that there are also notable differences in mentality and culture. Kenpas of the Upper Indus area do not appreciate sarcastic speech and condemn the Shammas for their 'crooked' or 'twisted' speech. Compared to westerners, Shammas still have a rather underdeveloped sense for linguistic jokes, but at least they do enjoy linguistic accidents (e.g. my /kha tshante/ kha tshante 'hot mouth' instead of /!fa khante/ ja khante 'salted tea'), and like to play with opposites (saying 'the door is too high for you' or 'you are too short' when one has bumped one's head). This is exactly the kind of speech that Kenpas detest. While lucky lnmters in Domkhar are (or were) expected to share their prey with the villagers-and would thus try everything to prevent their success becoming known-no such expectation of 'exaggerated' solidarity threatens the candid attitude of hunters in Gya. This contrasting attitude towards lnmting and sharing definitely deserves a detailed anthropological study. It is possible that the obligation to share is related to rites concerning goat-worship and hunting, as reported for the Dards of Pakistan or more generally for the mountain tribes of Iran (Jettmar 1961: 87-92).
242
BETIINA ZEISLER
along the Suru contributory from Parkachik up to Rangdwn Yllldo marks again an economic b01D1da!y: the fonnidable pasture land, bare of any trees, stands in sharp contrast to the rich agricultural land of Parkachik, with its extended fields and copious trees. The handful of meagre fields of Yllldo barely deserves this designation. Only some way after the Pendzila along the Doda river, down to the central plain of Fa<mm (Spa/dum, 'Padwn' on the maps) and in the adjoining valleys, is more substantial agriculture once again fomul, and even then in no way comparable to that of the Suru valley.
Legend: !'fne O}Con'irO!
--
Figure 1. Dialect regions of Ladakh (detail, map not to scale)
The lower part of the Zanskar river is cOIDlected by trails via Wanla and Lamayuru to Lower Ladakh and the dialects spoken there are likely to be either part of the Shamskat group, or to be strongly influenced by that group. 19 The dialect ofYulchung-Nyeraks, for example, is clearly Shamskat. These two villages are situated on opposite sides of the lower part of the river, shortly after the eastward bend near Lingshet. 19 Cf. the LSI (p. 52), accordng to which "the north-western dsbicts" of Zanskar "show traces of the Sham dalect".
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
243
According to the Nyeraks consultant (Tashi Angchuk, 1996), one used to reach there from Wanla in a two or three day walk crossing three passes. There seem to be suitable geographical boundaries within the area, such as several passes on the trail between Fa~um to YulchungNyeraks isolate the populations on each side, but since I have never been in that region this needs further confirmation. According to Thomas Preiswerk, who started his linguistic fieldwork in 2007, the Zanskar dialects are relatively homogeneous, but are classified into four groups by the speakers themselves: 9ot (Stod) or Upper Zanskari along the Doda river, Zhung (Giuy) or Central Zanskari, around the valley of FaOOffi, Sham (G.~am) or Lower Zanskari along the lower course of the Zanskar river, and Lungnak (Lu1Jnag 'Black Valley') along the upper course of the Zanskar river. As we both have noticed, however, there is considerable individual variation among speakers from the same village or even within the same family. I first obtained data from Kenhat in 1996, when I recorded a short version of the Kesar epic in Gya. During my field stay in Ladakh in 2004, I transcribed the narration with the help of a Cemre speaker, and discussed with her in some detail the grammar of Gyahat and her own dialect During my field stays from 2005 to 2010, I worked mainly with a consultant from Gya-Sasoma (in 2009 also with her cousin), but also had the chance to conduct shorter interviews with one speaker each from Shara (Upper Indus), Hameling (Zanskar), 20 Shayok, Laga (Shayok valley), and Nyoma (Changthang). Each of these dialects still requires systematic research. In 2008, I fmally had the opportunity to make a short trip to Zanskar, visiting the villages of FaOOffi, Zangla, and Pipcha together with Thomas Preiswerk. During this visit, I collected some data from a Pipcha speaker. Besides the Comparative Dictionary of Tibetan Dialects (CDTD), comparative Ladakhi data is vailable only for the Manda dialect of Zanskar (Hoshi and Tondup Tshering 1978). 21 The varieties ofHima20 Located on the Doda river, somewhat north-west of Phe, north of 'Padum', AMS NI 43-12 (ca. T7376); also to be found on the Government of Pakistan International Series sheet NI-43. 21 Tondup Tsering hails from Manda village (Hoshi and Tondup Tshering 1978: i), but the authors remain silent about its location. According to the features described below, one might smmise that it is located north-west of Fa6um, somewhat further up than Hameling. But the only village by the name of Manda that I could locate lies between Hameling and Phe (AMS NI 43-12, ca. T7974). The village is not found on other maps.
244
BETIINA ZEISLER
chal Pradesh, namely Spiti (Sham area and Pin valley), Nyamkat, and Tot (''Tod'') have been cursorily surveyed by Sharma (1989 and 1992). 22 Veronika Hein is presently working on the Tabo Spiti dialect and I am much obliged for the information and data which she has generously shared.
2. PHONETIC AND MORPHOLOGICAL FEATURES 2.1. Same minor features The Upper Indus dialects of Cemre and Shara show palatalisation of the simple initial post alveolar fricatives- [D - ['r], which may be blocked, however, for certain morphemes, e.g. the limiting quantifier [-Jik] 'some' (Table 1, column 5). Similar palatalisation is found in Nyoma, with s- [D - [h'r] and z [3] - [~]. Old Tibetan radical f:t and radical y before back vowels are typically realised as voiced fricative laryngal [fi], e.g. /fioma/ vs. Sham and Leh lorna/ f:tama 'milk'; /fiura/ vs. Sham and Leh /jural yurba 'canal'; /fiog/ vs. Sham and Leh /jog/ yoy (CT f:toy ~ yoy) 'come' (Table 1, column 3). In Nyoma, the voiced laryngal is typically related to an original f:t, but may be found in a few words that have an initial palatal in Classical Tibetan, such as /fiura/ yurba 'irrigation channel', /fiotl yod 'exist', /fiop/ yob 'stirrup' and /fiuldak/ ?yulthag 'place to store the grain before threshing'. By contrast, some Kenhat dialects show an initial palatal before front vowels, thus Nyoma, Shara, and Manda /jinl yin 'be' and /jige/ (MND ''yiye'') yige 'letter', Gya /fiinl, /fiige/ vs. Leh and Sham /in/ and /ige/ (Table 1, column4). Nyoma, Shara and Gya (possibly also Cemre) show the realisation of OT w (originally written f:t above v/b, probably for *[yv] or *[fiv], cf. Roerich 1933: %, Uray 1955: 109-112) as [y-] or [fi-], thus [yatse] or [fiatse] for watse in contrast to Leh and Sham /watse/ 'fox' (Table 1, column 2). The voiced laryngal (or its unvoiced but low tone counterpart) is also found in the neighbouring varieties, cf., e.g., Spiti and Nyamkat /fioma/ 'milk'; Spiti /fiatse/ 'fox' (Sharma 1992: 21, 22,
22 It is necessary to point out that a large mnnber of printing and data errors marr Sharma's work, making his books rather unreliable as sources. All finther references are thus given with the proviso that additional research into these dialects is needed
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
245
112);23 Tabo Spiti /fi.oma/ and /fi.atse/ (Veronika Hein, p.c.); but Lahul lorna/ and /uatse/ Roerich 1933: 96, 98). For a more extended connection, see also Drokpa I!,CYIJI ~ fh!CYIJI, Mustang /~o/ wamo 'vixen'; Drokpa /~ ~ /h~ and Mustang /!_si/ wa, watse 'fox'; Drokpa IQagl ~ lhQil]l and Mustang /Q!.D.a/ f:toma 'milk'; Drokpa /h.Q:/ yod 'exist' and lh!!,g/ yo1J/f:t01J 'come' (Kretschmar 1986: 354, 368, 369 for Drokpa, 1995: 469, 489 for Mustang). 24 Uvular or laryngal realisation ofOT b and w is otherwise known only from Eastern Tibet, cf. Kham: Kardze /yQa:/, Batang lyQmal 'milk', Amdo: Rngaba /y;omre/, Mdzorganrabar fy;omal 'milk'; and Amdo: /y;a/ ~fy;~/ ~ /ya/ wa 'fox' (CDTD). 2.2. Fricativisation
ofcluster onsets
The most obvious feature for which the Kenhat dialects around Hemis are known among Ladakhis is the fricativisation of former word-initial clusters of voiceless radical plus non-nasal pre-radical (prescript g-, d-, b- or superscript r-, 1-, s-): [hat] or [x.at] for Leh and Sham /skat/ (skad) 'language', [c)era] for Leh and Sham /spera/ dpesgra 'speech' (see also appendix). 25 Varieties more to the east, such as Nyoma, have lost all pre-radicals and their intermediate traces. Interestingly, the Kenhat dialects vary considerably in which cluster is reduced by fricativisation and which pre-radical is completely lost Gya shows only word-medial fricativisation, but nearly complete loss of the pre-radical at the word onset, the exceptions being [yo] mgo 'head ' 26 and the clusters g-, (b-,) r-, s- + ts - s, 27 Shara shows fricativisation of only the clusters with a voiceless velar radical and loss of 23 Sharma actually writes ''homa" and "hace" But he also uses the letter 'h' in his discussion of aspiration of voiced initials with the example "lorna/ ---+ /homa/'' It appears that he is biased towards the Indian phonetic inventory and does not differentiate between the voiced and the unvoiced laryngal. The transcriptions "c" for [!11 and for [ts] (!)are systematically confused in all of his work. 24 One may also compare Central Tibetan Kyirong /fiQ:/ bog 'below', but /Qk'a:/ bogkhar.J 'ground floor', and /fiyf bu, but /Qra:/ boraTJ 'we (incl.)' (Huber 2005: 32). 25 First mentioned by Francke (190 1: 6) and LSI (p. 52). 26 The realisation /yo/ for mgo 'head' is also attested in the Kesar narration of Stok, cf /ste gozam, stayozam/ rta/:d mgozam rtamgozam 'as big as the head of a horse, as big as a horse-head', /yokak/ mgokag 'head-responsibility (i.e. to warrant with one's one life)' In Hameling, the voiced initial might interchange with a voiced aspirated initial. '2:1 GYS IJaJ]gll! spyaT.Jkhu 'wolf should also be mentioned, but the fricativisation of the clusters+ l[(here frompy) > jis common to all Ladakhi dialects and may even be found in Balti, cf./JaJJko/ beside /spjaJ]ku/ (SPR).
"c"
246
BETIINA ZEISLER
the pre-radicals for all voiced radicals as well as for the voiceless labial and dental radicals, Cemre exhibits fricativisation of the clusters with an voiceless velar and labial as well as a voiced labial radical, and thus loss of pre-radicals for the voiced and unvoiced dental and the voiced velar; whereas Hameling exhibits not only the fricativisation of clusters with voiceless velar and labial radicals and a full set of fricativisation of clusters with voiced velar, dental, and labial radicals, but also the retention of the clusters with a voiceless dental radical. Manda (Hoshi and Tondup Tshering 1978) shows not only fricativisation of clusters with dental radicals (in addition to velars and labials ), but also the alternative retention of clusters with voiced and voiceless dentals (see Table 1, columns 6-8 and the appendix). Similarly, the dialects differ in the realisation of ancient clusters with a palatalised velar: Manda retains it: /xJ.got/ skyod 'go, come (hon)', Hameling and the other Central Zanskari dialects palatalise it completely: /r;ot/; while the other dialects have lost the pre-radical: /kjot/. As in many other Tibetan varieties, the process of onset cluster reduction started at the beginning of the word, whereas the process was delayed in medial position, cf. GYS /kat/ skad 'speech, language' vs. /Gjahat/ Gyaskad 'dialect of Gya', IK~tl Gyenskad 'dialect of Upper' Ladakh. Not only the original syllable onset clusters became fricativised at the motpheme boundary, but simple stops also when the preceding syllable ended in a fricative sonorant (-r, -1, -s). In many such compounds, the original coda of the first syllable, or the coda resulting from a former pre-radical (see section 5 below), was preserved before a voiceless radical of the second syllable, while the following stop became fricativised, cf. GYS and HML [yol-fak] mgo-lpags 'scalp', GYS [kal-fa] bskal-pa 'aeon' and HML [kar-fho] dkar-po 'white'. In such cases, a former pre-radical s- changed into fmal -r before velars, cf. GYS [3ur-hut] giu-skud 'bowstring' and Uar-ham] saskarn 'dried meat'. The regional distribution of this feature suggests that fricativisation was an intermediate state in the process of onset cluster reduction. 28 28 It is not the only possible one. Sun (2003: 785) describes preaspiration, resulting from r- and s- pre-radicals. Balti shows the aspiration of the nominaliser pa by a preceding final -s, cf the examples p. 21 given for the hidden past-tense suffix. Preaspiration as a result of devoicing is found in Leh for nasals with an original pre-radical
and similarly in Leh and some Shamskat dialects for the sub-scribed lateral. The Kenhat fricativisation might thus result from an earlier aspiration In fact, the nominaliser
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
247
However, in view of the original laryngal fricative [h], which remained unaffected, I am somewhat sceptical about the phonetic automatism leading from [x;at] ~[hat] to [kat]~ [qat] simply via de-fricativisation (as a further reduction of phonetic complexity). I would have expected a few mismatches at least, for example a word such as Ladakhi /hampa/ hampa 'courage' becoming */kampa/, and a word like /hat/< skad remaining /hat/. Yet, to my knowledge, no such mismatches have ever been reported for any of the Tibetan dialects that underwent onset simplification. Since Ladakhi shows some allophonic variation between aspirated stops and fricatives, especially labial [ph] ~ [fj, but also velars preceding back vowels: [kha] ~ [qha] ~ [x;a], one might further have expected that fricatives would end up as aspirated stops. The 'correct' reanalysis might thus be due in part to sociolinguistic factors, namely the constant contact with speakers of the less reduced Leh dialect and the prestigious nature and normative pressures of the Leh dialect A second, equally important factor is the delayed development at the medial position. 2.3. Changes in the coda position With the exception of Leh and to a certain extent Manda, all Kenhat dialects have lost post-fmal -s (Table 2, column 2). With respect to the complete loss of the whole velar cluster (OT -gs, -ys, Table 1, columns 3 and 4) one can again observe great variation among the dialects: Manda shows no loss at all, or at least no loss of the final velar; Nyoma (as well as other Changthang dialects) has lost the fmal -s; Hameling (and possibly other Zanskari dialects) has lost both clusters, 29 while the other dialects have lost only the cluster with the nasal. 30 Loss of fmal cluster -gs can also be observed in Tabo Spiti (Veronika of the past participle may be realised either as [-tfa] or as [-<pa]. Fricativisation of clusters with labial radical is also attested in Amdo Tibetan, cf Sun (1986: 119, 129, 143-144) and Haller (2004: 25, 230, only /
248
BETIINA ZEISLER
Hcin, p.c. ), less commonly in Mustang (Kitamura et. al. 1977); while loss of ftnal -1)S is attested in Mustang (Kitamura et. al. 1977, Kretschmar 1995lll: 127), but less regularly in Tabo Spiti (Veronika Hcin, p.c.). A further important difference is found in the treatment of ancient fmal -s after vowel (Table 1, column 5), which is preserved, unaltered, in Shamskat and generally also in Leh. A few words, such as Leh /thorei vs. Sham /thoras/ thoras 'tomorrow' and the ablative morpheme /ne/ nas, nevertheless demonstrate the Kenhat substrate. The Kenhat dialects typically show vocalisation of ancient fmal-s as *i. The Zanskar dialects show a tendency for diphthongisation or fronting after back vowels: o, u + *i- [oe] or [o], [ui] or [ii] and merging of a, e + *i - [e], i + *i - [i]. The other dialects show a complete merger with the vowels [e] (< -as, -es, -os) and [i] (< -is, -us). The latter sound change does not seem to be of a very old age, as it is not yet fully completed in Gya where in a few exceptional cases the diphthong is preserved (e.g. /so/ 'tooth', /soe/ 'of the tooth'). Thomas Preiswerk (p.c.) also observed a shift towards merger in Zanskar, particularly among children and among the Muslim community of Farnun. Final -s appears to be preserved in Manda (with or even without change of the vowel), but is lost in the other Kenhat dialects ( cf. Table 2, column 5). While Thomas Preiswerk (p.c.) suggests that the Manda data is unreliable due to interference with the Leh dialect or a local koine, I would think, in the light of the above observations, that it might represent an earlier stage in a possibly quite accelerated process of sound change. Diphthongisation or vowel merging does not seem to have been pre-eminent in the Kenhat dialects only a hundred years ago, as it went unnoticed by Francke (1901) and the LSI, although diphthongisation with complete loss of fmal -s has been noted by Jaschke (1881: xvii) for an unspecified Ladakhi dialect (perhaps the most eastern one?). But even in this case, vowel merging was not yet observed. Nyoma shows regular fronting of vowel a - [e], frequent, but unpredictable fronting and partial diphtongisation of vowel o - [oe], and very infrequent fronting (and diphthongisation) of u- [iii]. 31 Occa31 This feature has been tested mainly for the past tense suffix -s and has yet to be tested more systematically for noms. It is possible that the non-fronting is due to stem assimilation. On the other hand, the consultant stated that while he could perceive the difference in the vowels, speakers of his dialect would not usually pay attention to it. This would indicate that the vowels or diphthongs [oe] and [ili] are not yet part of the phonemic system.
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
249
sionally, a syllable fmal vowel could be affected by a 'migrating' s- preradical in the following syllable (see also section 5 below): GYS [yefu] < *[yes-fu] < *[yes-pu] < *[yos-pu] < mgo-spu 'hair (on the head)' This sound change goes hand in hand with a dramatic reorganisation of verb stem formation in Gya and the Upper Indus dialects. In the development of the Ladakhi dialects, and similarly also in the Central Tibetan dialects, a certain pressure to eliminate the opaque ablaut system of Old Tibetan can already be observed. With only few exceptions, vowel (and radical) of stem I, the so-called present stem, were assimilated to that of stem II, the so-called past stem (Zeisler 2004, 874-876). The main difference between the two stems was then the presence or absence of the past-tense marker -s (see also section 4 below). While speakers can apparently still ignore the mere fronting or the diphthongisation of the vowel by an immediately following past-tense marker -s, they obviously cannot ignore the new ablaut, resulting from the merger with vowel [e] or [i], particularly since the morphological motivation (past-tense marker -s) is no longer transparent. The pressure to eliminate ablaute must have been very productive, leading thus to a second cycle of verb stem assimilation in the Upper Indus dialects. As in the frrst cycle, assimilation is generally oriented towards stem I (exceptions are typically motivated by semantic disambiguation) and affects also stem IV, the so-called imperative stem. The development can be illustrated with the verb bco 'do': 32 LEH /!fo : !fos/ > :MND /!fo :!foes/> IDv.IL /!fo :!foe/> GYS (grandfather) /!fo !fe/ > GYS, SHA, CEM /!fe/. Stem IV is even affected in those cases where the original verb root had vowel a and the vowel alternation for stem IV: a- o had been preserved in the frrst assimilation cycle, e.g. GYS, SHA, CEM /tel, GYS (grandfather) /ta : te : to/, Sham /ha : has : hos/ < OT Ita: bltas: blta: ltos 'look at' (cf. Table 2, column 6). As the example of Gya shows, the process of verb stem assimilation is of quite recent date, and might still be continuing in other dialects. The grandfather of the Gya consultant died in 2003 at the age of 81. He, as well as his father, hailed from a family that had been living in Gya-Sasoma for centuries. His grandfather was a miikpa 'son-in-
From CT bchos bcos bco chos 'construct, fabricate' The verb has replaced byas bya byos 'do', probably because the sound changes, affecting 1he stems differently, led to an opaque and intolerable clash between stem I *be and stem 32
cr byed II *lfas.
250
BETIINA ZEISLER
law' from Miru, a nearby village on the road to Kharu. His mother came from Gya, his grandmother from Upshi, and his wife from Tiri (Ronghat area). According to the consultant, Ronghat shows particular features, which are quite different from the speech of her grandfather, so that an influence from these dialects can be ruled out. 33 Furthermore, the consultant claimed to have observed the 'archaisms' of her grandfather's speech in the speech of other villagers of the same age group also, whereas the generation of her parents (born 1937 and 1938) would use the new forms almost exclusively. Exceptionally, the verb /so se/ 'look after (children)' (CT ~tshol gso) has retained the two stems in the speech of the younger generation, in contrast to the corresponding verb /se/ 'restart (a frre)' (CT ~tsholgso).
The completion of the morphological change in the Gya dialect can thus be dated to the frrst quarter or frrst half of the 20th century. I would assume that the corresponding changes in the Upper Indus dialects are not very much older. But for the moment, no comparable data are available, and the prospects of getting such data are none too good. The generation of these (great-) grandparents is slowly passing, and only very few consultants have an awareness of the differences in the speech across generations. 34 The Gya consultant (born 1979) is quite exceptional in her diachronic awareness as well as in her passive vocabulary. As a small child, she went with her grandfather to the high pastures, and while attending primary school, she still accompanied him to the winter pastures, thus becoming acquainted with his style of speech.
33 Even more so, as marriages used to be arranged around the age of 13 or 14, cf. the age difference between the consultant's grandfather and father. One would expect that the young girls (or boys) quickly took up the dialect of their in-laws. 34 The consultant's generation were among the first children in remote villages to have access to modern education. Typically, modern education is accompanied by some sort of cultural alienation In Ladakh, it also goes along with a certain degree of standardisation of the dialects towards the speech of Leh and the emergence of an inferiority complex, which does not enhance the memory for non-standard particularities of elders' speech. Aside from their obvious benefits, school education and modern mass media fin1her distract children from the oral traditions transmitted by their grandparents.
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
251
3. TONOGENESIS The Kenhat dialects are particularly informative for the study of tonogenesis in Tibetan. Some of the problems described below for Gya are also known from Tabo Spiti (Veronika Hein, p.c.). Both dialects warrant an in-depth phonological study. The following observations are based primarily on the (self-) perceptions of the consultants. It is not entirely clear which of the Kenhat dialects can be described as having phonemic tone. Three of the consultants denied that there were phonetic tonal distinctions in their dialects. The Leh consultant with whom I worked in 2006 was not at all aware of a tonal difference in Ladakhi dialects, and, confronted with my description and exaggerated pronunciation, explicitly denied that she would produce anything like that My impression, however, was that her devoiced consonants were, at least occasionally, accompanied by a lower pitch and a falling-rising contour. At one occasion, the difference was so striking that I decided to record some of the crucial word pairs, but since the consultant immediately became more selfaware ofher pronunciation, the tonal distinction broke down. The Cemre consultant was not able to hear the differences in the intonation of the Gya narrator, which were absolutely obvious to the Gya consultant. The Cemre consultant grew up with her mother's sister in She, and it is possible that this has influenced her auditory perception. Unfortunately, I could not check her articulation. The Hameling consultant denied that there could be any difference between /put/< bud 'get free, run away' and /put/ < phud 'let free, take off' and her intonation was, in fact, identical (no falling-rising tone contour in the ftrst case). Tone is likewise not reported in neighbouring Manda. Similarly, Thomas Preiswerk (p.c.) could not observe any Zanskari dialect with tonal distinctions. The Hameling consultant used to work for farmers in the Sham areas since her late childhood and picked up the dialect of Khalatse very well. This again might have interfered with her speech production, but less probably with her auditory perception. It turned out that the possible tonal minimal pairs are not very frequent in her dialect and relate mainly to intransitive-nonagentive and transitive-causative verb pairs. In these cases, the ambiguity is easily resolved from context, at least in an ordinary conversation. The same is true for Manda and all other dialects of Zanskar, as well as for Leh.
252
BETIINA ZEISLER
The situation is certainly different in the Gya dialect, where initial fricatives are completely missing and a tonal distinction is apparently more important for understanding. Both the Gya and Shara consultants were able to produce clear minimal pairs, showing both a difference in register or pitch and in intonation, i.e. a falling-rising tone contour on the vowel for the low tone and a level tone contour for high and neutral tones. However, whenever the consultants produced longer example sentences, the difference tended to disappear, and the same could be observed in the Gya narration. In 2006, becoming more aware of her own speech, the Gya consultant stated that the unvoiced low tone stop consonants are actually semi-aspirated, and this could be verified by a primitive test (holding the back of my hand a short distance from her lips), but, again, the feature tended to disappear in free speech. In free speech (including recorded narrations), the falling-rising tone contour appears to be the main distinctive feature, to my ears at least. The Gya consultant was also able to hear a tonal difference in the narration of Stok, and again, after some time, I was able to identify a falling-rising tone contour in those words which she described as being oflow tone, e.g. /palal)/ [palalj]
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
253
these as lower than velars, and these again lower than palatals (t :=::: p > k > tjJ, but the exact ranking could vary somewhat with the type of articulation. Voiced consonants were perceived as being lower than unvoiced consonants originating from a voiced one, these again lower than aspirated consonants, and those lower than an (originally) unvoiced, non-aspirated consonant (g > k (< ~g) > kh > k). But in this case, the ranking could be completely overthrown by the ranking of vowels, which is more or less binary: a and o being perceived as low, i and u as high. 35 The vowel e, if realised as [e], would be likewise classified as high, but the consultant has a strong tendency to pronounce it as [e] or even [~e] in elicited words and sentences, and this pronunciation is in line with its classification as 'low'. Aspirated consonants as well as unvoiced sibilants were initially classified as 'low', later as 'neutral'. Quite surprisingly, and in contrast to the Shara consultant, the Gya consultant did not describe plain nasals as lower than voiced stops, and this relative height (but still in opposition to high tone nasals resulting from clusters) might be the reason why nasals in the coda position or at a following syllable onset interfered with the consultant's perception of tone: syllable final nasals, in particular 1J and m regularly led to a classification as 'high(er)'. The same effect occurred when an open syllable was followed by a syllable with nasal or high tone onset. To give only one example of the complexity and subtleness of the consultant's perception, the low-tone nasals with 'low' vowel a were ranked in between the 'neutral-tone' aspirated velars with 'high' vowel i or u and the low-tone unvoiced velar with 'high' vowel: ga > gu > kha > ka > ku > gi > khi I khu > ma Ina I pa I ya > ki I tri > kii > kil> kf
With all these subtle phonetic distinctions, the binary phonemic tone distinction seemed to be lost completely, and it was only consistent when the consultant repeatedly classified words such as /kore/ kore 'cup' and /tal rta 'horse' as 'low'. Finally, on understanding the various interacting factors and the relativity of each statement, I used ex35 This, at least, is not only due to the consultant's particular musical awareness. A second consultant, with whom I had the opportunity to work in 2009, got several times confused, misclassifying at first low tone words with the 'high' vowels i or u as having high tone and arriving at the correct description only after a lot of thinking.
254
BETIINA ZEISLER
actly these two words as fixed points of comparison to establish a 'high tone' for a level intonation despite a 'low' vowel, particularly when there was no low tone counterpart to be found with the exact match of vowel and final. 36 While the tonal distinction established by this method roughly corresponds to what a Tibeto-linguist would expect, I observed some anomalies, which could indicate that 'tone', whether defined as pitch or register or defmed as tone contour or as a combination of both, is not yet a true phonemic feature in the dialect of Gya, despite the fact that a few minimal pairs can be found. 37 As we would expect, plain 36 We used the terminology /rombo/ rompo 'fat, stout' for low tone and /trhamo/ phramo 'thin' for high tone, alternatively to phoskad 'male voice' for low and moskad 'female voice' for high tones. The expressions rompo andphramo are common among musicians, but obviously also understood by non-professionals (Rebecca Norman, p.c.). Ths terminology was not without disturbing side effects. When we discussed family terms based on the old family 'prefix' a-, the consultant divided 1hese somewhat arbitrarily in /rombo/ and /trhamo/, changing the classification for some of them from week to week or day to day. When I realised that the majority of the terms for females were classified as 'thin' and the majority of the terms for males as 'fat', and joked that especially the lane! 'aunts' were loosing and gaining weight, the consultant started reclassifying all terms along gender lines. Nevertheless, after the long break between my 2005 and 2006 fieldwork, she eventually declared that the differences were artificial and all terms were 'high' tone, in line with linguistic expectations. TI Quite to the smprise of most Sinologists, Beckwith (2005: 10) challenges the concept of phonemic tone in Chinese, since, according to him, a wrong intonation in "actual connected speech" might not necessarily lead to a different meaning for the listener, whereas any change in consonants and vowels does. He describes tone as "at best, an extra, a redundancy feature" that together with other features could enhance clarity. Beckwith, however, does not seem to be aware that different kinds of minimal pairs have different tolerance features, tlrus the wrong choice of articulation place is typically less tolerable than the wrong choice of articulation manner (particularly when the articulation places are quite distant) and the latter might be less tolerable than mistakes in accentuation or tone (I myself owe this insight to discussions with Thomas Preiswerk). Depending on the language, tonal contrasts may be quite articulate or very subtle. In the latter case, speakers certainly make use of all o1her available cues, especially contextual ones. This makes it very difficult to define the stage where a merely phonetic and gradual differentiation of tones turns into a truly phonemic, that is, discrete opposition Given the fact that a phonemic opposition cannot be gradual, by definition, one could perhaps even argue that what is perceived as low and extra low tone in Gya are both concomittant features of two types of voicing, which might perhaps be called 'semi-voiced' and 'enhanced voicing' The former type may be characterised by its additional aspiration feature. The latter type is found with the additional feature of prenasalisation in various Tibetan varieties, e.g. in Zanskari, Tabo Spiti (Veronika Rein, p.c.), and Kyirong (Huber 2005: 15; cf. also note 41 below for her description), sometimes also with fricativisation and/or aspiration as mentioned in note 26 above.
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
255
nasals and laterals were typically classified as being low tone while those with a pre-radical were classified as high. The same effect should have been found with the palatal approximant y as a radical, but quite a few words that have an Old Tibetan pre-radical and that have high tone in the Shara dialect were qualified as low. As these words are found only with the 'low' vowels a and o, high tone is apparently only perceived when the need for differentiation is high, e.g. /~/ *g.yar 'die, croak, perish',38 in contrast to /ja/ gyaf:t 'feel ticklish' and /jar/ gyarcas 'borrow'. The most striking case is perhaps the contrast between high tone /jok/ *g.yog 'twist (threads)', although not corroborated by any classical written form, and the two compounds based on this verb: /jQkto/ *yogto 'coarse thread' and /jQkJig/ *yogsiy 'stick for twisting' with low tone. In several other cases, the consultant was quite confused about the proper classification. 39 Family tenns, starting with a glottalised a in Old Tibetan, were arbitrarily classified as either low or high (see also note 29), while all were unmistakably classified as high by the Shara consultant Some anomalies in the nasal class were corroborated by the Shara consultant and are thus of historical interest: /mak.pa/ 'husband, sonin-law' was unanimously classified as being high tone, thus the alternative spelling dmagpa and the relation to dmag 'army' mentioned by JAK are obviously correct for Ladakhi.40 Another, quite unexpected case of high tone is GYS /moll for CT mol 'speak (hon)', where other tonal varieties of Ladakh and Western Tibet show low tone realisation, e.g. Trangtse and Man-Merak, both in Upper Ladakh, and Tabo Spiti (CDID). However, the Balti form /hmol/likewise indicates that The Domkhar prommciation /hjar/ points to a pre-radical. E.g. /jiik/ g.yag '(male) yak' and/jOkpo/ g.yogpo, because oftheir 'low' vowel, were repeatedly classified as having low tone, although they were consistently uttered with a level tone contour. Highly surprised, I challenged this from time to time, and after a while, the consultant first conceded that it might perhaps be 'neutral' tone, while in 2008 she finally corrected herself, assigning the same tone to /jiik/ as to /jar/ 'borrow' 40 Note, however that other Western Tibetan dialects have a low tone realisation, e.g. Drokpa (Bawa, Bongba Tshogu), Dingri, Mustang/~ 'bride groom, son-inlaw (living in the wife's family)' (Kretschmar 1986: 420, Herrmann 1989: 360; Kretschmar 1995 IV: 143). Low tone realisation is also reported from Ngari and Kyirong, but o1herwise it seems that a greater part of the Central Tibetan dialects, as well as Kham and Dzongka, show high tone realisation (CDTD). By contrast cr mig 'eye' < OT dmyig is found with high tone or even pre-radical in most modern dialects (cf CDTD). The Gya prommciation /nuK/ is no exception 38 39
256
BETIINA ZEISLER
the word must have had (a variant with) a pre-radical: *dmol, as suggested in the CDTD, or *smol, as fmmd in the writings of an Upper Ladakhi scholar (RN). Interestingly, also the spelling contrast between CT giiis 'two' and iiiSu 'twenty' is reflected in a tonal opposition /pi/ and lpiJul, similarly that between gnam /nMnl 'sky' as used in isolation and nam /nyn/ id. as used in compounds and collocations. By contrast, /m,!b3al 'peacock' was unanimously classified as low tone, making the relation to the classical spelling rmabya somewhat dubious. But low rather than high tone was also observed for gnastshul 'news': GYS /n.!tsull ~ lll.!dzull, but SHA /natsul/ and gnaslugs 'condition': GYS, SHA lll.!dluk/, while the dialects of Shayok and Laga showed the 'regular' high tone: SYK /nazluk/, LAG /naluk/. The word rmilam 'dream' was classified as high: /mttam/ by the Gya consultant, in accordance with linguistic expectation, but as low: /milam/ by the Shara consultant. Unexpected low tone was also observed for GYS, SYK, and LAG ll,!D.gjatllhanrgyas 'together' and other words with initiallh in CT. In general, the Shara consultant, although a singer herseJf, never made as subtle distinctions as her colleague from Gya. Independent of the vowel and the final, the Shara consultant distinguished three main tones: high, neutral, and low, with further graduations of low tone: -
'thramo': the originally voiceless, non-aspirated consonants as well as the nasaL palatal approximant and lateral when combined with a pre-radical
-
'normal' (i.e. neutral): aspirated consonants, voiceless sibilants, and the voiceless laryngal
-
'rompo': devoiced consonants
-
·~saJJ
roma' (lower than that): voiced consonants
-
·~saiJ
roma': plain nasals.
Both the Gya and Shara consultants agreed upon the medium or neutral tonal character of aspirated consonants. This is particularly interesting as it matches almost perfectly the traditional grammatical classification of the four consonant rows or articulation types as pho 'male', that is, 'forced' or 'hard' (k), maniT) 'neutral' (kh), mo 'female', that is, 'soft' (g), and sintu mo 'extremely female/soft' (g). The Changthang dialects of the Durbuk and Nyoma blocks are likewise tonal dialects. However in the case ofNyoma, a clear tonal opposition exists only for words having an voiceless non-aspirated radical
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
257
in Classical Tibetan (high tone) and those having a voiced radical, not preceded by any pre-radical. Outside this opposition tonal descriptions turned out to be more or less arbitrary, independent of vowel properties or syllable codas, and the consultant did not perceive tone as a three or four-way distinction and thus lacked the notion of 'neutral tone'. In fact, voiced consonants were often described as higher than their devoiced counterparts. A similar feature has been described for Derge (Hasler 1999: 257275, for high pitched voiced initials see pp. 267, 269 with further references). Although there is a certain tendency for low pitch realisation of voiced consonants, the pitch may vary considerably from word to word (partially in accordance with the degree to which the consonant has been devoiced), but also in different contexts. Hasler thus argues that pitch only becomes a distinctive (phonemic) feature when the consonant has become completely devoiced. But this neither explains the unexpected high pitch realisation of voiced consonants nor the variations in its realisation. The contrast between the Nyoma and Derge data, on the one hand, and the Gya and Shara data, on the other, clearly demonstrates that the stage where pitch contrasts become phonemic can be reached by two completely different developmental paths. On the frrst path, exemplified by Tibetan grammatical understanding as well as by the dialects of Gya and Shara, the actual realisation of voiced consonants follows the tendency of low realisation, so that the phonemic property merely results from a shift of acoustic awareness, triggered by the devoicing process. As a natural result, the devoiced initials also lose part of their low pitch properties. 41 On the second path, as exemplified by the unrelated dialects of Nyoma and Derge, the actual realisation contradicts the natural tendencies and calls for an explanation. This phenomenon could perhaps be understood as a contrastive reaction vis-a-vis devoiced initials or the neu41 This is corroborated quite nicely by the Kyirong data. According to Huber (2005: 19f), the mwoiced consonants, derived from originally voiced initials show middle tone, those derived from clusters of oral stops and voiced radicals show low tone plus breathy voice, those derived from original clusters of nasal and voiced consonants remained voiced, but display a tonal pattern between middle and low tone. What looks rather like a contradiction, finds its explanation in the fact that voiced consonants are realised with a strong prenasalisation, "exclusively used to reinforce the voiced character" However, the "voiced consonant can be almost devoiced after the prenasalization" (p. 15, emphasis added).
258
BETIINA ZEISLER
tralisation process in general: only those initials that are in the process of losing their voiced quality develop an enhanced and marked low pitch realisation. By contrast, all other initials are interpreted as high and, as in the case of Derge, consequently even produced with a higher pitch. It would be interesting to test whether Nyoma speakers in general, or at least the consultant in question, actually produce voiced initials with a high pitch.
4. PRESERVATION, GENERALISATION, AND EVENTUAL LOSS OF THE OLD TIBETAN SUFFIXES -SI-D (STEM II, IV) AND -DI-s (STEM I)
When one looks only at the verbal stems and not at the complex verb forms that are based on these stems, one may easily get the impression that the Kenhat dialects have lost all Old Tibetan sufftxes, while the Shamskat dialects have well preserved the Old Tibetan sufftx -s (~-d) of stem II and to a certain extent also of stem IV. The picture is, however, much more complex, and one fmds traces of the former suffixes for all three stems in the Kenhat dialects. Even more, these traces are, in certain surroundings (such as the directive and question marker), even more frequent than the corresponding overt Shamskat morpheme. I will begin with stem IV, where the data is obvious and uncomplicated, and will discuss the quite problematic fmdings concerning stem I last.
4.1. Stem IV with directive marker l-ay/ In Shamskat and Leh, the imperative sufftx -s is found only in verbs with an open syllable root. In the Kenhat dialects it is preserved and even generalised before the directive marker /-ag/ - /-sag/ (CT day) in the positive command form. This feature has been systematically checked only for Gya. Here the s-form is always optional, but nevertheless frequently used. According to the consultant, the s-less form indicates that the action is quite urgent and should be performed immediately, whereas the s-form signals that it can also be performed somewhat later. The s-form thus co-occurs with all agentive verbs, independent of their shape. The form /-sag/ appears after open syllable roots as well as after closed syllable roots, including those that do not allow a suffix -s in Old and Classical Tibetan and those where the Old
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
259
or Classical Tibetan stem IV lacks a suffix, e.g. /kok-saJ]!/ kog-sIJa1JISa1J besides lkoal)!/ kog-~a1J 'peel [it] offi' (CT ~gog bkog dgog khog), /kok-sag!/ kog-s-~a1JISa1J or lkoaJ]!/ kog-~a1J 'stop [it!]' (CT ~gegs bkag dgag khog), /kor-SalJ !/ skor-s-~a1JISa1J besides /korag!/ skor-~a1J 'tum[it]!' (CT skor bskor*d bskor: skor*d). In the prohibitive, which is based on stem I, the s-form does not occur. For Nyoma it can be said that the directive marker /-sal)/ is obligatory at least for open syllable roots, in which case it may or may not be accompanied with a fronting of back vowels. A likewise cursory survey of the Pipcha verbs showed that the additional -s is obligatory after vowels, optional after fmal -r and -1, and not permissible after all other fmal consonants. Quite interestingly, in Pipcha a final -s morpheme is commonly realised as a dental fricative /6/, but at least the speaker whom I interviewed also accepted the sibilant realisation for the directive marker as an infrequent option. A similar sound change can also be observed for the past tense suffiX -s (see below). 4.2. Stem]] andpast-tenseforms Generally, the Shamskat dialects (except the Nubra dialects) have preserved the past-tense suffiX -s only for agentive verbs and neutralised it for most non-agentive verbs. 42 But the Shamskat dialects have also generalised the suffiX -s in places where Old and Classical Tibetan do not have any suffiX (e.g. /taJ)sl 'gave', OT gto1): bta1): gta1): tho1J), do not allow a suffiX for phonetic reasons (e.g. /sats/ 'killed', OT gsod: bsad: gsad: sod), or would allow only the suffiX -d (/pans/ 'listened', OT (m)Fian: (m)Fian*d: mflan: non). On the other hand, the past-tense suffiX of agentive verbs is often dropped without leaving any trace in Balti and similarly in the western Sham narrations.43 In northern Nubra, post-final -s is generally dropped, but the pasttense suffiX may be preserved in questions (see below) and in nominalised verb forms, e.g. ARA /rgjaJ]pinl ~ /rgjaJ]spin/ brgya1)-(s)-pin 42 The fonner suffix might have been either dropped or lexicalised A few nonagentive verbs, however, still have two stems, but the distinction of the two fonns might depend on the dialect or even the individual speaker. On the other hand, stem neutralisation can also be found with agentive verbs (Zeisler 2004: 620-622). 43 In such cases, my consultants would say that the suffix is there in principle, but carmot be heard due to the speed of speech. A similar careless prommciati.on or drop can also be observed in the Sharnskat dialects with respect to the -s of the ergative marker.
260
BETIINA ZEISLER
( -payin) '(I) stuffed' Very infrequently, the past-tense suffix shows up in nominalised forms of non-agentive verbs in Shamskat, e.g. OOM
!karl 'become white' (CT dkar '(be) white'): /rgunftk karspa zbear!fik SlJOnl dguncig dkar-s-pa dbyarcig syon 'When the winter had been white, the summer will be green' Other traces of a lost suffJX are not found in the Sham dialects, whereas in Balti, the aspiration of a following nominaliser /-pal - /-pha/ gives evidence of a former suffix, similar to the fricativisation in the Kenhat dialects. In a few cases, this trace can also be found with non-agentive verbs, cf. Kharmang /lfhaq/ chag 'break [--ctr]', /lfhaq-pha/ < chag-s-pa 'broken' and /thop/ thob 'fmd', /thop-pha/ < thob-s-pa 'found' besides IJil si 'die' and I Jispa/ si-s-pa 'dead' (Ghulam Hassan Lobsang 1995: 5-6). By contrast, most of the Kenhat dialects have dropped the pasttense suffJX -s, leaving an obvious trace only in the verb stem with the above-mentioned Kenhat vowel change of open syllables and the deletion offmal clusters with velar fmal and suffJX (-gs, -ys). Nevertheless, traces at morpheme boundaries show that the suffJX must have been generalised even for non-agentive verbs at an earlier stage. Altogether there are four indicators: (a) Vowel change or diphthongisation (b) Loss of final clusters -1JS and -gs (c) Aspiration or fricativisation of a following nominaliser pa -- [pha] - [cpa] (d) Retention in (past-tense) questions, cf. LEH /skjotsa?/, GYS llgotsa?/, ARA /skjora?/- /skjodza?/ skyod-da- skyod-s-sa/IJ,a 'did [you/s/he] (already) come?', NYO /arne nU.m trglsa?/ amatzi snum *grul-s-sa/IJ,a 'Did mother apply oil [on the child's face]?'
As far as the Ladakhi dialects are concerned, the combination of features (a), (b), and (c) is restricted to the Upper Indus, Gya-Miru, and Zanskar dialects, whereas (d) has been observed also in northern Nubra, Leh, and the Changthang dialects. As already mentioned, feature (b) can be found in other Western Tibetan varieties, (c) is also observed in Balti, whereas (a) is typical for most of the modern Tibetan varieties and might be found in the Changthang dialects as well. Features (b) and (c) and their combination can be exemplified with the verb GYS, CEM /lalJ : ta ~ ta tog/ btay btays to1) 'give': /khe tal kho!J-i bta1)S 's/he gave', /ge tafin/ ya!J-i btayspin 'I gave' Depending on the dialect, fricativisation can be found with all agentive verbs plus open syllable non-agentive verbs (Cemre), or with all
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
261
agentive and almost all non-agentive verbs, independent of the syllable structure (Gya). In a few cases of non-agentive verbs, the fricativisation is blocked in Gya, but it is unclear when and why. Perhaps such verbs are mere relicts, as yet unaffected by the generalisation rule. The question marker has been systematically checked only for the Gya dialect. In most cases, the question marker can take only the form /-sal (or at least this would be the clearly preferred form), in few cases both forms. When both forms coexist, the s-form might have an admirative function, expressing surprise, a greater emotional involvement, or indignation, e.g. /matosa?/ masto-s-sa 'did [it] really not matter [to him/her]?' but /matoa?/ mastof:ta 'didn't [something] happen [to him/ her]?' Less frequently, the s-less form might have this function. Otherwise, the s-form might refer to a somewhat more distant past and the s-less form to a more recent or more immediate past. Which form of the question marker is chosen appears to be somewhat unpredictable and conditioned more by the context of the utterance than by the semantics of the verb. 44 In Pipcha, the past tense sufftx left its trace in the optional fricativisation of the (voiced) nominaliser as well as in the loss of fmal clusters with the velar nasal (less clearly with velar stop). Additionally, the past tense suffix is preserved in the question marker, following stem II. As in the case of stem IV, the morpheme is realised as a dental fricative/&, alternatively also as /9/. Due to its unstressed position, vowel a of the question marker (and other morphemes) is commonly realised as lei, but the consultant I interviewed also accepted forms with /a/. 45 On the basis of a cursory survey, it turned out that the past tense sufftx /&is obligatory for agentive verbs (except after fmal -t, where it tends to be omitted) and optional for non-agentive verbs. In the latter case, it may convey a notion of surprise. 44 Due to growing language awareness, the consultant accepted fewer s-less fonns in 2006 than in 2005 and claimed more vehemently that such fonns were only used by Leh speakers. Nevertheless, she still accepted s-less fonns for non-agentive and agentive verbs alike. In 2007, I had to remove about 90% of all remaining s-less fonns from the documentation; they were, if at all, only accepted for non-agentive verbs. In 2008, the speaker, with only one or two exceptions, did no longer accept s-less fonns, even for non-agentive verbs. While I am unable to repeat such checks regularly, it is evident that some linguistic features allow a certain variability, but due to various influences the consultants keep changing their minds about the actual extent of the variability. 45 According to another consultant with whom Thomas Preiswerk worked, 1his might be due to intentional switching to the Lungnak dialect.
262
BETIINA ZEISLER
4.3. Stem I with the foture participle or the causative auxiliary
A trace of an over-generalised suffix -s seems to appear also in stem I, or more precisely in a following future participle morpheme cas~ ces: Purik /-!fa/~ Sham /-!fas/ ~ Leh /-!fes/ ~Upper Ladakhi /-!fel, and in a following causative auxiliary /!fuk/ bcug, according to the all-Ladakhi rule: /s/ + /!f/ - I fl. Again, this feature has only been systematically checked for the Gya dialect. Here, the future participle appears as /-!fe/ after final -k, -t, -n, -p, and, -m. Some verbs are also more or less acceptable with the form /-fe/. 46 After vowels, the regular allomorph is /-3e/, due to intervocalic voicing /-!fe/ - (1-<61 -) l-3el (something that does not happen to the causative morpheme /!fuk/, which is thus not fully integrated into the word unit). The allomorph /-fe/ is found under the following conditions: regularly after fmal -r, mostly after -1, and infrequently after -1). In the latter case, the nasal might be dropped. While the change after the dental fricatives -rand -1 could perhaps also be interpreted as an extension of the above mentioned sound change rule, thus /r/, Ill, Is/+ /!f/ - I JI as suggested by Thomas Preiswerk (p.c. ), the loss of the nasal and the optional occurrence after stops, by contrast, demonstrate the former presence of a sibilant. The allomorph thus results from a regular sound change after an original fmal sibilant or, at least, after an original final dental fricative, including the sibilant -s. Theoretically, it is possible that the allomorph /-3el results from intervocalic voicing of* /-se/, and would thus likewise indicate a regular over-generalisation of suffix -s after vowels. The interchange of /qy and /y, however is not uncommon in the Ladakhi dialects, cf. LEH /khar~i/ ~ GYS lkhar3il kharji 'food', Balti IJUI, Ladakhi /~~juan honorific interjection, LEH ~~~. but KHAL, DOM I~JUI ~j~ju 'please', and thus /3el could equally well result from /~e/ (!fe/ due to intervocalic voicing). Occasionally, the future participle as citation form and the complex future tense forms based on the participle show a different behaviour. In a few cases, the s-trace is only positively attested for the future tense forms and, in one case, only for the negated form. A process of dropping the s-trace in the future participle may have started, as the consultant occasionally indicated that the s-trace could be found in the 46 In some cases, this was classified as old people's speech or even more specifically as 'mother's speech', in which case it might be an Upshi feature.
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
263
speech of the older generation. In several cases, she also changed her mind about which form to use, which might be indicative of an ongoing process of change, but could well be due to the constant influences of the Leh dialect on her mother tongue, on the one hand, and an increased awareness due to the investigation context, on the other. The causative motpheme shows a somewhat different development. The form 1-Jukl is in most cases only optional (the preferences vary from verb to verb), but in contrast to the verbal noun it can be found quite frequently after fmal -k, -t, -n, -p, and, -m, cf. llgok!fe/ kyogces 'to change one's direction, turn oneself\ caus. /kjok!fuk/ ~ llgokJukl kyog-(s)-cug, cug; /thritlfe/ ~(less commonly) /trhitJel l:zkhridr(s)-ces 'to be heavy with young', caus. /thrit!fuk/ ~ /thritfukl l:zkhrid-(s)-cug, /traplfe/ bkrabces 'to clean (the crop) from husk or dirt', caus. /traplfuk/ ~ /trapJukl bkrab-(s)-cug. On the other hand, the s-fonn is only rarely found after vowels. This hidden suffiX -s might perhaps go back to the Old Tibetan present tense agentive suffiX -di-s. It could perhaps also be due to a general process of stem assimilation. The feature does not seem to be prominent in the Cemre dialect.47 Nevertheless, one can find alternative forms such as /zerlfel zerces ~ /zefe/ zer-s-ces 'to say'. 48 In the Pipcha dialect, the fricative forms only appear after vowels and nasals in OT/CT roots with fmal -s,49 and after fmal -r, cf. /3en !fhefe/, /0 !fheJukl CT bden ches 'believe', /dra:fe/, /dra:Jukl, cf. DOM /drags/, GYS /dra/ <*!}grays (CT !}gray) 'be satiated', /gorfe/, /gorJukl l}gor 'be late'. It appears thus that the generalisation of the various -s suffixes started in the eastern dialects, where it affected more verbs than in the western dialects. But it also seems that with the regular use of evidential auxiliaries for present tense expressions, the grammatical function of the past-tense suffiX -s became bleached. Possible phonological pressures to avoid final clusters may then have led to its eventual loss in the eastern dialects and to the observable instability in the western dialects. It is quite likely that the loss of the imperative suffiX followed similar lines. Both developments-that of generalisation and that of deletion-may have spread at different speeds. But this may be due to the consultant's socialisation or to normative pressures. Probably via */~:s/ + /!fe/; the final cluster -rs is similarly reduced in the Shamskat dialects. 49 In these cases, the final sibilant may have been part of the original OT root or might have been lexicalised at a later time. 47
48
264
BETIINA ZEISLER
5. FROZEN CLUSTERS IN GYA COMPOUNDS Many Tibetan dialects preserve the former pre-radicals in medial position, i.e. in the second elements of compounds when the frrst syllable ends in a vowel. In that case, the former onset of the second syllable moves to the coda position of the frrst syllable, leading to a shift of the syllable boundary. 50 The Ladakhi dialects are no exception (see Shawe 1894 for an early discussion). As could be expected, frozen clusters are generally more common in the Shamskat dialects than in the Kenhat dialects, but also more common in peripheral dialects. They are much less frequent in the dialects around Leh. Due to the prestigious nature of the Leh dialect and due to an increasing normative influence of the reading style taught in schools, one can observe a tendency to suppress such frozen clusters, especially among the younger generation. Gyahat plays the role of a very peripheral dialect. Of all the dialects surveyed so far, it shows the greatest number of frozen clusters, outnumbering even the western-most western Sham dialect of Achinathang, which comes second before Dornkhar. Some of the rare compounds are only attested in Gya, Domkhar, and/or Achinathang, and in the Purik varieties spoken by the bilingual Dards of Dah and Garkhon (to the northwest of Achinathang) and Dras (in the Kargil district). In the following, I will give only a few examples with the former grammatical b- preftx: -
GYS !Jrap-tri/ skra-bkrus 'hair washing', GYS !Jrap-Jatl skra-bSad 'combing', GYS, GARK !Jrap-JaV skra-bsal'rinsing of the hair'
-
GYS /khap-tril, DOM /khatrhus/- /khap-trhus/ kha-bkhrus- kha-bkrus 'washing of mouth (or face)' (GYS: only for animals), GYS /burfe burbe gop-tri/ sburpatzi mgo-bkrus 'beetles' head-washing' Gokingly for a rainy day, when the beetles come out of the earth)
-
GYS /tap-Jratl, DOM lrhtaJrat/- /rhtap-Jratl rta-bskrad 'chasing away a horse (from the field)', G YS /tap-Jat/, DOM /rhtap-Jat/ rta-bSad 'stroking or brushing of the horse' GYS /dzop-Jrat/ mdzo-bskrad 'chasing away a dzo (from the fields)'
°
5 Cf here and in the following Suzuki (this volume) for Sogpho. The phenomenon of word-medial consonant migration was already observed in the earliest descriptions of Classical Tibetan, beginning with Csoma de Koros (1834: 7-8 and 10). Unfortunately, these works are completely ignored in the cUlTent discussion of the feature, cf Shirai (1999).
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
265
-
GYS /mjp-sat (khip-sat)/ mi-bsad (khyi-bsad) 'one who has killed or murdered several times, 51 professional killer' in contrast to /misat (khisat)/ mi-bsad (khi-bsad) 'one who has killed a single time', 52 ACH /mip-sat/ 'one who has killed, murderer', GARK /mip-sat(pa)/- /mip-sot(pa)/ mibsad(pa) - mi-bsod(pa), DRS lrnip-sat/ 'murderer, killer'
-
GYS lrng>-sat/, DRS lrnep-sat/, DOM /mesat/ (younger generation) /mep-sat/ (older generation), GARK /mesat/- /mep-sat/ me-bsad 'fire-extinguishing,
These examples are not attested with the labial stop or are not attested as compounds at all in the other dialects surveyed (for many more examples also with nasal prefix and lexicalised pre-radicals, see Zeisler 2009 and Zeisler in preparation). In some compounds, the same process applies to the (according to a traditional analysis) radical b followed by the post-radical -y- or -r-. The most prominent might be the all-Ladakhi forms /lab-raJ]/ bla-bray 'residence for monks' or /rib-~~ /rib-3al ri-bya 'mountain fowl'. It cannot be precluded that the status of the initial consonant as a preradical or a radical was irrelevant for its shift. But it is also possible that the traditional spellings represent, or lead to, the wrong analysis. 53 In this context, a very particular set of compounds in Gya, some not attested in any other dialect, might be of special interest The second syllable, if isolated, always starts with a simple r, while the first sylla51 E.g. a mentally disturbed person, who has killed several people in fighting. Now the tenn might also be applied to terrorists. 52 lkhi/ khyi 'dog' is often used as an echo word to /rri/ mi 'man', but of course, lkhi(p)sat/ kyhihsad may also mean 'killing or killer of dogs' 53 Apparently, there was no need to differentiate between the sequences preradical+ radical and radical+ post-radical, as in the case of voiced velar plus palatal approximant (where the former sequence is represented by a horizontal and the latter by a vertical combination). The reason might be that exemplars of one of the two sequences were missing or extremely rare. As the early scribes were not trained linguists, it is not necessarily the case that their spellings were always coherent or always represented the correct analysis in tenns of pre-radicals, radicals, and post-radicals. This is quite apparent in the case of the sub-scribed radical-!-, which is reflected in all varieties as (high tone) I, the only exception being the cluster zl, (Huber 2005: 3lf.). But the regular b-prefix for stem II/III in len: hl01Js: blOIJ: lonll01J(s) 'take' has never been recognised as such by Tibetan grammarians. It is 1hus possible that sequences of super-scribed b- and sub-scribed y or r represent original sequences of pre-radical plus radical. The same seems to be true for at least some of the combinations of g- plus -r- discussed below ( cf. also Beyer 1992: 72, note 5, 74-79. Beyer should better not be quoted, since throughout his 'compilation', he 'cites' other somces without giving appropriate references. Unfortunately, in this case, I have not yet come across the original).
266
BETIINA ZEISLER
ble, if isolated, would end in a vowel, but is, at least optionally, closed by a labial stop in the compound. In one case, the second element is attested as an independent word with a labial cluster onset in the western Sham dialects, e.g. GYS 1-b-rak/ or 1-rak/ '-lining' corresponding to DOM /brak : braks : brok/ or, with metathesis, lrbak : rbaks : rbok/, also lrak raks rok/ 'join together, attach, add'; cf. GYS lftk-rak/ gcig-(X)rag, lpib-rak/ giiis-Xrag, lsum-rakl gsum-(X)rag, l3ib-rak/ bziXrag 'having one, two, three, or four panels or grids (said of a window)',54 cf. DOM lpir-bakl, lsum-brak/, l3ir-bak/, lyar-bak/ ~ lrroarbak/ (younger generation) 11)a-Xrag, ltruk-brak/ drog-Xrag, lrdunbrak/ rdun-Xrag, lgjat-brak/ brgyad-Xrag, lrgur-bak/ dgu-Xrag, llfurbak/ bcu-Xrag. The Domkhar metathesis points to an original form sbrag, 55 and thus we might substitute sb and not only b for X The triple cluster would regularly yield lrak/ in most Ladakhi dialects (including present-day western Sham). 56 We can fmd a similar compound in all-Ladakhi /khib-ragl, KHAL, DOM lkhizbagl 'dog-fly', ultimately derived from khyi-sbra1J (cf. JAK). There are also other compounds where the second element does not appear as an independent word in the western Sham dialects, such as GYS ldorukl ~ ldob-rukl ~ ldom-ruk/, DOM lrdoaruk/ ~ lrdoab-ruk/ rdo-Xrug 'small stones'; GYS, DOM IJab-ruk/, ARA IJaruk/ ~ IJabruk/ (neither form is very common in ARA) sa-Xrug 'little pieces of meat', or where the second element is not attested with labial cluster 54 The consultant could also think of using /J]ii.b-rak/ 11JaXrag, /gub-rak/ dguXrag, and /!jlib-rak/ bcuXrag 'having five, nine, ten grids' for the multigrid windows of modern houses in Leh, but so far the new fashion has not yet fmmd its expression in the Gya dialect. 55 The consultant, from whom the present data is obtained, would accept the metathesis only in case of triple clusters. But some Domkhar and Takmacik speakers of 1he younger generation apply the metathesis also to the double cluster hr. For the moment, I am unable to say whether this is an innovative overgeneralisation or rather typical for the two villages. Apart from Balti, the feature seems to be somewhat exceptional with respect to all other western dialects, where the only form attested so far is Purik /rbi, rbis/ for bbri 'write' 56 Cf cr sbrag sbrags 'lay, put one thing above another' WT /rak/ (JAK; cf also his wr compounds with numerals!), Balti (Skardo) /rbaql 'pile up, join (in activity), follow, come along', Purik (Kargil, Ciktan) /zbraql, eastern Sham (Nurla) /rak/ 'join together' (CDTD), western Sham (DOM) /rbak/ - /brak/ - /rak/ 'join together, attach, add, give dowry' (according to a second consultant /rak/ is used only in connection with the goats to differentiate the verb from /brak/ 'sheer', but for the main consultant there would be usually no need for such dissimilation). For the regular Balti metathesis in triple clusters, cf also Bielmeier (1985: 212).
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
267
onset, such as GYS /doril/ ~ /dob-ril/, DOM /rdoaril/ ~ /rdoab-rill, ARA/rdoril/ ~(less frequently) /rdob-ril/ rdo-Xril 'round stone'. The pronunciation /rill might go back to a verb sgril bsgril with the meaning 'make round'. Similarly, we fmd the compounds GYS ljlirikl ~ /pi"b-rik/ gnis-Xrig 'two rows (of turquoise on the perak, the women's traditional headgear)' and /garik/ ~(as heard) /gab-rik/ l1JaXrig 'five rows, not attested in the other dialects. Here again, the form /rik/ may relate to the verb sgrig bsgrigs bsgrig sgrigs 'put in order, arrange, etc.'. The conclusion would thus be that /ruk/ 'small piece' ultimately relates to either *(b)sgrug or *sbrog. In fact, we do fmd a Classical Tibetan verb sgrog bsgrugs bsgrug sgrugs 'collect, gather, pluck, pick up', an activity one may perform with 'small pieces' in particular. 57 The Gya dialect provides us with another, even longer, series of compounds with the second syllable /-rag/ or /-b-rag/, indicating that the entity described by the frrst syllable is 'alone, single, nothing but itself, not going together with the prototypical counterpart or accessory'. These compounds are only rarely attested in other dialects. Most people would use expressions with /ftkftk/ gcigcig, /ftkpo/ gcigpo 'single' or /khorag/ khora1J 'itself instead. I can offer only a few examples (the full set will be given in Zeisler in preparation) :58 -
ltriratJI- ftrib-rat]/ gri-(X)rat} 'knife without meat'~ 9
-
/tfhub-fat)/ chu-Xrat} 'nothing but water'
-
/p,!!b-rat]/ Pfa-Xrat} 'a single fish'
-
/tiirai]I- /tab-fat)/ rta-Xrat} 'horse without companion, foaL saddle'
-
/daral)/- (rarely) /dab-rat]/ mda}:l-ra1J 'arrow without bow'
-
/m,!!b-rai]I ma-Xrat} 'mother whose child(ren) went to another place'
57 Cf. Balti (Skardo) /rgik/ 'put in order', Purik (Kargil, Ciktan, and Tshangra) /zgrik/ 'fit in (trans.)' and 'arrange, line up, row, repair' as well as Tshangra /'ZJ!,flll
268 -
/m~-raiJ/ - (rarely) without a pot on it)'
BETIINA ZEISLER /m~b-ra.IJ/ me-(X}ralJ
'fire burning for nothing (i.e.
- I3Ura.IJI- (rarely) /JUb-rai]/ iu-ralJ 'bow without arrows' Following the above analysis, one could expect an original form *bsgray or *sbray for the second syllable. The closest or most evident Classical Tibetan equivalent, however, would be hray 'alone' or simply ray 'self, also reybu or reyspo 'alone, separate, not belonging to anything else' (cf. CT pho reypo 'a man without family', Herrmann 1983: 44, folio 266v, line 3). The verb sgray (BRGY: sgroy): bsgrays: bsgray: sgroy (BRGY: sgroys) with the meanings 'enumerate' and 'upbraid, reproach' (JAK) does not at frrst sight seem to be related. JAK, however, paraphrases the frrst meaning as 'reckon up separately', and it might be this 'singling out while counting/in order to count' that underlies the etymon of ray 'self', hray, rey(s)- 'alone', grays 'number', the one-stem verb ~gray, the two-stem verb bgray bgrays, and the four-stem verb sgroy ~ sgray bsgrays bsgray : sgroy(s) 'count'. One may note that there is a similar etymological relation between ril- 'round', hril- 'round, globular, whole', gril 'roll', and the verbs ~gril : gril 'be twisted, wrapped round' and sgril : bsgril 'wind, wrap round, roll, wrap up'. While it could be the case that the Old or Classical Tibetan forms with plain r initial derive from an earlier reduction of the triple clusters sgr and sbr, even this hypothetical development could be indicative that the element r was the semantically most important element and that the other elements were mere derivational elements. Given this possibility, I think it even more likely that the element r set discussed above containing the cluster sgr, at least, is the root consonant. If we assume that two derivational elements cannot be added at the same time, the element g- might have been added at a very early stage of the language, perhaps in regular alternation with b-, and the element s- followed when the derivational character of the element g- was bleached. The forms underlying the compounds may well represent an original alternation g-r b-r for stems 1/IV and 111111. In any case, the compounds are evidently based on stem II and preserve an original grammatical b- prefix.
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
269
6. SUBSTITUTION OF FINAL-S BY 1-T/ OR 0 Looking for an example of word-medial fricativisation of labials in Shara, I suggested, after some futile attempts, the place name Saspol, expecting something like /Sefol/ (if a well known place name could undergo such a dramatic change, at all) or perhaps /Safol/. Quite surprisingly, the village name turned out to be pronounced /Satpoll and the consultant went on to state that she would say /drat/ for Leh /dras/ IJ-bras 'rice', while the expected form would have been /dre/. A short survey through a Ladakhi dictionary showed that the final -s is typically not replaced in verbs (independent of whether the fmal originally belonged to the root or was a past-tense marker). 60 In the case of nominals, however, about 20 to 30% of the lexemes ending in final -s show a replacement by -d. This feature is also attested in Gya, but the consultant herself prefers the form with 'regular' vowel change or the Leh pronunciation, while describing the forms with a substituted final as belonging to the speech of the parents' or grandparents' generation (quite often she gave that form only after several requests). 61 Rebecca Norman drew my attention to the fact that the feature is likewise found in the Changthang dialects, but so far I could verify this only for two dialects of the upper Shayok valley: Shayok and Laga. Again it seems that in the two Changthang dialects, the substitution is more frequent in the speech of the older generation, while the younger generation either uses the Leh form, a form with the 'regular' vowel change, or a form where the fmal -s has been dropped without leading to a vowel change. This kind of drop is also found, but considerably less frequently, in Gya and Shara. In the case of original vowels e and i, vocalisation of fmal -s and drop without vocalisation 60 I have so far found only three exceptions: a) the collocation GYS, SHA /tshe dati tshe bdas 'hon. die', quite probably derived from the adjective GYS ltshedat/ tshebdas
'hon. late'; b) GYS /gja/ -/gjat/ rgyas 'increase, spread, grow, develop [-ctr]'; cf also GYS (elder people) /W'gjat/, SHA /j~gjat/, LAG /Wgjat/ yarrgyas 'development, progress'-interestingly, the verb takes the expected form /gje/ in the collocation GYS /sa.IJ gje/ SQI'}S rgyas 'obtain enlightenment'-; c) the not very common verb GYS /fio:/- /fiot/- /fios/- (perhaps only infrequently) /fie/ 'deserve', which is quite exceptional in that the final-s may be preserved. This may point to a (re)-introduction from the classical language at a rather late stage. 61 As always, there is quite some variation in the perception and acceptance of such forms among individual speakers, e.g. 1he second consultant from GYS was quite astonished that some of the forms given below should exist.
270
BETIINA ZEISLER
cannot be discriminated. 62 In a few cases, one can observe the replacement or alternation with a dental nasal. In the following list I use the index ''S" for the vocalised final -s, ''V" for vocalisation without drop of final -s, ''T'' for the substitution with the dental stop (''N'' for a final nasal), and ''0" for drop without vocalisation. ''L" will be used for the standard Leh forms, which seem to be borrowed. The relevant form is given first, but if a dialect allows alternate forms, the 'regular' form with vocalisation and drop will be given ftrst: replacement by -d: -
-
GYS (mostly) /gjalJre/s - /gjalfrat/T, SHA /gjalfrat/T, SYK /gjalfras/~, LAG /gjalfre/s 'prince (hon)' < ~alsras; GYS /Gjalfrat rimbotjhe/T Gyalsras rinpoche, Ifrat renbotjhe/T Sras rinpoche SHA /gjatpa/T, GYS /gjefals, SYK /gjespalv 'extensive, wide, real' <
rgyaspa -
Gyaik (Rong; RN) /gjutpa/T, Laga (RN), Panggong (RN) /gyunpa/N, but GYS /gjUfa/ 'sinew, tendon'< rgyuspa SYK (only grandparents), GYS, SHA /f<>tma/T, LAG /fc;ma/0 'artificial'
GYS, SHA /!fat/T 'dresses', SHA l!fhat/T 'accessories for dressing', SYK, LAG /!fat/T 'dresses and accessories'
-
SHA /pii]rut/T, GYS /JIIIJrils -ljlii]ru/0 'effort, perseverance, endurance, persistence' < sffl1Jrus
-
SHA /thot/T, SYK, LAG /thewals 'extra' < thos
-
SHA, GYS /thutmi/T, SYK
/thusmi/~,
LAG /thumi/0 'representative' <
flthusmi -
SHA /datgut/T, SYK /dasgu/ [!!] 'respect' < dadgus; cf. G YS /datguttfan/T dadguscan 'very respectful'
-
SHA /ty_t/T, GYS /tgs/L (grandparent's generation!) - (younger generation) /tj/s- /tyt/T63 , SYK /tgs/L; LAG /'qJ/0 'time'< dus
-
SHA, SYK (only grandparents) /dutpa/?T 'imitation' < !zduspa (?)
62 I did not yet have the opportunity to check this feature in the Nyoma dialect, but I would guess from the verb stems that the final -s would have been frequently dropped without vowel change (perhaps via a final -d). 63 In 2005, the consultant generally used /til in free speech, but in 2006 she started using /tyt/, probably because we had discussed 1his feature at length.
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
-
271
GYS /ngtsul/(T) -/ngdzul/1064 'news, condition'
GYS (grandparent's generation, earlier also consultant) /pod/T- (consultant) /pos/L 'incense' < spos
-
GYS /drat]gje/s- /drat]gjat/T, SYK, LAG, SHA /drat]gje/s 'kind of offering'< ?IJbraiJrgyas
-
SHA /drat/T, GYS (generally) /dre/s- (few people of parents' generation) /drat/T 'rice' < J:Wras; cf. also GYS /drethukls - /dratuk/(T)> SHA /dratuk/(T) J:Wrasthug 'rice soup', GYS /dradbu/T- (younger generation) /drasbu/L hhrasbu 'fruit'; but SHA /dribuls; SHA /dratsil/T GYS /dratsil/T (middle generation)- /drasil/0 J:Wrassil 'rice with butter, ~ugar, and apricots'; SHA /sat]drat/T, GYS /Sat)dre/s- (mostly elderly people) /Sat)drat/T bsaiJJ:Wras 'rice (hon)' 6~
-
SHA /m.!l_tpo/T, GYS /m.!l_fo/s- /m.!l_tpo/T, SYK /m.!l_spo/r, LAG lrn.!l_tpo/T (old people) - /mgm/10 < mespo 'grandfather, forefather, ancestor'; cf. GYS /m.!l_tpo Gandi/T 'ancestor Gandhi', /phamefonepharlals - /phametponepharla/T < phamesponaspharla 'since ancient times', SHA /phametpo/T phamespo 'ancestors'
-
GYS /modlam/T < /mozlam/ (LEH) 'prayer' < smonlam
-
GYS (only grandfather's generation) /tshatpotlrr /tshefet/sT- (consultant) /tshefe/ss 'seedling' < tshasspos
-
GYS /tshedat/T, SYK /tshede/s 'late (i.e. dead)' < tshefuias; also as verb: G YS, SHA ltshe dat/T 'die' < tshe IJdas
-
SHA, GYS /fiotpa/T 'worth', SYK, LAG /ospa/L 'deserving' < J:tospa; also as verb: GYS /fio:/- /fiot/- /fios/- (perhaps only infrequently) /fie/ 'deserve'
-
SYK, LAG /jru>het/?T 'discrimination'< ?yaphes
-
SHA /j.!!Jgjat/T, GYS (elder people) /j.!!Jgjat/T - (younger generation) /j.!!Jgjaslr, SYK /j.!!f&ias/L; LAG /ji!fgjat/T- (less frequently) /j.!!f&ia/0 'development, progress' < yarrgyas, also as verb: GYS /gja/ - /gjat/ rgyas 'increase, spread, grow, develop [--ctr]'
-
GYS, SYK (only grandparents) /rudbal/T 'tortoise, turtle'< rossbal
(grandfather)
64 The first form apparently implies a final dental stop. The second form could be due to regular inteiVocalic voicing of simple consonants, or due to a secondary development of the first form. 65 The word bhras is replaced by lbato/ (from Shina or K.asluniri) in SHY and other Changthang dialects. In the comp01Dlds /drastuk/, /drasil/, and /S8J]dras/, the word is clearly a loan from Leh.
272
BETIINA ZEISLER
-
SHA /ret/T, GYS (elder people) /ret/T - (younger generation) /res/L> SYK, LAG /re/0 'turn, alternation'
-
GYS (grandparents) /lu(d)drel/T- llidrellv now becomming replaced by /luzdrel/L < lustzhrel 'sexual relationship' but /ledrel/8 < lastzbrel 'workrelated relationship'
-
SHA!l.Qgjut/T, GYS (father) ll.Qgjut/T- (consultant) /l.Qgju/0 'history, story'<
lorgyus -
GYS /Sat]gje/8 - (some people) /Sat]gjat/T, SYK, LAG, SHA /Sat]gje/8 'Buddha' < Sal)srgyas
-
GYS /sodbu/T, SHA /sibu/8 'adopted child'
-
SHA lhat/T, GYS, SYK, LAG /he/8 'exaggeration'
-
SYK, LAG, GYS /1Nat/T 'all (hon.)', SHA /lank.je/8 'together' < lhan-
rgyas replacement by e: -
SYK, LAG, GYS, SHA/khalen/0 'agreement'
-
SYK, LAG lkQrul/0, GYS /kgul/8 - lkQrul/0, SHA /k~rul/8 'rags'
-
GYS /gjufa/0, SHA /gjitpa/T [!], SYK, LAG /gjunpa/N 'sinew' < rgyuspa
-
LAG /rJQ.a/0, GYS public'
-
LAG, GYS /Dotrup/0 - /.Diirup/0, SYK /Dotrup/0 - /Dorup/0, SHA /.Diitrup/0 < D1}osgrub, a name (common as Durup in Leh)
-
GYS /j1i'tJru/0 besides /pii:]ri/8 sffi1Jrus, see above
-
SYK, LAG /1jho/0, GYS, SHA /1jhe/8 [!] 'religion, religious books' < chos; similarly GYS, SYK, LAG /1jhokjot]/(s) chosskyo1J 'guardian deity', SYK, LAG, GYS, SHA /1jhokhat]/0 choskha1J 'chapel'
-
GYS /J1!!tsull0 gnastshu~ see above
-
LAG /naJ.uk/0 gnaslugs, see above
-
GYS, SHA /lotre/8 - /lotro/0 'intellect', also as name< blogros
-
GYS /drasil/0 - /dratsil/T tzbrassi~ see above
-
SYK, LAG, SHA /tsoodru/0 (also as name), GYS /tsondri/8 'effort' < brtsontzgrus, cf. GYS /Ts6ndru/0 (name); SHA heard ltsondrut/T in Phuktse
-
SHA /mh.on/0, GYS /zehon/8 - /zahon/0 'food scarcity'< zasdkon
/~su/8 -
/rJQ.su/0, SHA /~su/8 < 1JOsla, 1JOSSU 'openly,
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
273
-
SHA /zahorn/0. GYS /zehom/8 - /zahom/0 'food and drinking'
-
LAG /jargja/0 - /jargjat/T yarrgyas, see above
-
SYK, LAG, SHA /rikna/0, GYS /rikne/8 - /rikna/0 'knowledge, science' <
riggnas; SHA heard the Phuktse pronwtciation /riknat/ -
SYK, LAG /rupa/0, GYS lrifa/s- /rufa/0, SHA /rifa/8 'bone' < ruspa, cf. also GYS /trW]rils -ft:rW]ru/0, SHA /t:r"at]rils bralJrus 'chest bone'
-
SYK, LAG, GYS (younger generation) ll.Qgju/0 lorgyus, see above
-
GYS /ja/0, LAG /fe/s, SHA/ye/s 'land rent' <sas
The word dus 'time' behaves quite unpredictably in compounds: -
GYS /tigjUn/8, SHA /tytgjwt!T, SYK, LAG /tygjUn/0 dusrgyun 'always'
-
GYS, SHA lty(t)fhen/m SYK, LAG /tyJen/L (not very common in either dialect) duschen 'festival'
-
GYS /tyf.tJan/T, SHA /t.itJanls- /tytt)an/T, SYK, LAG /tyl:Jan/0 dustjan 'bad, inauspicious times'
-
SYK, LAG /tyzaiJ/0, GYS /tizaiJ/s - /tydzaiJIT, SHA ltizaiJis dusbzQI) 'good, auspicious times, festival, holiday'
-
GYS /tilfhot/s - /ty(t)fhot/m SHA /tilfhot/s, SYK, LAG /tytsot/10 dustshod 'hour, time of the day'
-
GYS /ti3ils- /tyd3i/T, SHA /ti3ils dusbZi 'four seasons'
-
SYK, LAG, GYS /tyrap/0, SHA ft!rab/s < dusrabs 'era'
-
GYS /tu(s)sum/10, SHA /tisum/8 < dussum 'the three 'times': past, present, and future'
The various sound changes concerning final -s seem to have overlapped and it is difficult to establish a relative chronology. Substitution of final -s by a dental stop would typically have happened before the final led to vocalisation and fricativisation, cf. SHA /gjatpa/ < */gjaspal, SHA /metpo/ < /mespo/, GYS, SHA /ftotpa/ < /fi.ospa/. Dropping of final -s without vocalisation must have happened after final -s led to the fricativisation of an unvoiced initial of a following syllable (so far only observed with the nominaliser pa), cf. GYS /gjufa/ < */gjusfa/ and GYS /rufa/ < */rusfa/ besides /rifa/ < */risfa/ < */rusfa/ < ruspa. This would indicate that fricativisation preceded vocalisation. But on the other hand, SHA /gjitpa/ < */gjispa/ < */gjuspa/ demonstrates that the substitution of the sibilant by the dental stop could take place, at
274
BETIINA ZEISLER
least in single cases, after its vocalisation and before fricativisation, and that vocalisation could thus precede fricativisation. It would be swprising if ftnal -s were dropped without leaving a trace in the preceding back vowels against the general development in the Kenhat and Central Tibetan varieties. Furthermore, dropping of the (intermediate) dental stop seems to be quite a recent development, cf. the following pairs: GYS (middle generation, i.e. parents) /dratsil/T ~ (consultant) /drasil/0 , LAG (old people) /m~tpo/T ~ (consultant) /m~o/70, LAG /j~,rgjat/T ~(less frequently) /jMgja/0 , GYS (father, as recorded) /lQgjut/T ~ (consultant) /lQgju/ 0 , 66 and more generally the preference for such forms by the Laga and Shayok consultants who are 10 years younger than the Gya consultant The substitution by a dental stop seem thus to have been an option to avoid the already emerging sound changes triggered by the fmal -s. Second, dropping without vocalisation seems to be a further development of the substitution with dental stop, thus 1-s/ > 1-t/ > 1-01, cf. GYS /sildro/ < bsildrod 'temperate weather' for the occasional deletion of an apparently original fmal-d (but see also further below). The resulting inconsistency (fricativisation of a following nominaliser) seems to be less serious and could possibly be explained as an analogous formation, particularly if the 'regular' forms remained available in the speech community, as in the case GYS /rifa/ ~ /rufa/ < ruspa. Although the substitution of fmal -s by a dental stop seems to be older than the 'regular' Upper Ladakhi sound changes and should have stopped being productive some time ago, it has been applied also to modern loans, such as GYS (middle generation) I Japot/, SHA /r;apot/, SYK, LAG /fapo/, all-Ladakhi /fapos/ 'thick blanket' (source unknown); GYS /get/ for /ges/ 'gas'; SHA lcBatl, SYK, LAG lcBasl (o/ could correspond to a 'regular' sound change. No such objection is possible in the other cases.
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
275
so1md change originally applied only to high register items to avoid the 'regular', but somewhat 'vulgar' change. 67 The same strategy probably lay behind the adaptation of the place name Saspol. The substitution might thus have been, from the very beginning, an option to avoid vocalisation effects, and hence it could remain productive with respect to loan words, particularly because the analogue formation helps to adjust a fmal that does not fit into the phonetic system without changing the so1md of the word too dramatically. But this fact also might indicate that the 'regular' so1md change, which would lead to the pronunciation */be/ for 'bus', is still productive, and thus not of great antiquity. However, the fact that the substitution with a dental stop happened mostly with nominals but typically not with verbs (even though this necessitated a substantial reorganisation process), could perhaps point towards an ancient nominal derivation morpheme -di-n, preserved, but no longer productive in Old and Classical Tibetan, cf. ./ rga: rga rgas 'be, become old', rga-d-po 'old man', rga-d-mo 'old woman', rga-n-(pa) 'old'; ./ che: che ches 'be, become big, great', che-n-po (OT also che-d-po) 'big'; ./Ita: Ita: bltas blta ltos 'look at', lta-dmo 'sight, spectacle, etc.'; ./ dro: dro dros 'be, get warm', dro-d 'warmth, heat' (cf. the above-mentioned bsildrod), dro-n-mo 'warm' (see also Simon 1977).68 In some of these cases, we also fmd no1ms derived with -s, such as rga-s-ka 'old age', lta-s 'omen', dro-s-(chen) 'noon, midday , ' and dro-s-chUT) 'forenoon (?)'. The nominal suffixes -di-n and -s typically formed abstract collective no1ms (see Denwood 1986 and Uebach & Zeisler 2008 for the suffix -s, Zeisler forthcoming, chapter 4 § 2.4.3, notes on smin-(drug), spun, (ma)smad, (pha)spad, and spud for the suffix -di-n). Individual nouns and adjectives are derived by adding a definiteness marker -po or -mo. While the alternation between -d and -n reflects a shift from nasal to oral stops (and back) that affected various Tibeto-Burman and even Indo-European languages, it remains unclear whether the two sufftxes -di-n and -s are similarly related or evolved independently. An occasional interchange 61 While I would not like to go so far as to claim that all these items were merely borrowed from the religious language, their prestigious nature might have prevented an 'ordinary' pronunciation 68 In Kenhat, the dental nasal and the lateral are infrequently also involved in the substitution process, cf. GYS lkynbo/N, SHA lkynpo/ < guspo 'expensive' (see JAK gunpo Lahul 'expensive, dear'), the above mentioned /gjunpa/ from Changthang, as well as SHA /~ralma/ for all-Ladakhi /~ramna/ 'pea' < sradma - sranma 'pea, bean, lentil', possibly related to sra : *sras 'be, become hard, solid, compact'
276
BETIINA ZEISLER
between the two sufftxes can be observed quite early: the Old Tibetan word mye-s-po 'one ofthe group of the forefathers' (Classical Tibetan mespo, cf. also Ladakhi /meme/ 'grandfather, venerable person') appears in one of the lOth century fragments of the R.amayal}a in the form myedpo (A3), while the parallel version has myespo (E53, de Jong 1989: 94f.). Another instance, concerning the same word, is found in a late 11th or early 12th century colophon by lotsiiba Bloldan Sesrab (1059-1109), in which he mentions the temple of Tholing as the 'commitment of the ancestors, uncle and nephew' yabmed khudbongyi thugsdam (Bstanf:tgyur, vol. 123, no. 5719, p. 259; cf. Samten G. Karmay 1998: 23), the expected form beingyabmes. There might also have been other factors involved, as perhaps in the case of the substitution of the past tense and imperative morpheme -s by a dental fricative (perhaps originally a stop) in the Zanskar dialects ( cf. pp. 259 and 261 above). In any case, the substitution cannot be over-emphasised, since Jacques (2006: §2), observing the replacement offmal-s by fmal-t in Rgyalrong nouns and in the aorist form, notes: I1 n'existe pas a notre connaissance de dialecte tibetain ayant confondu -set-ten-t, et i1 doit done s'agir d'une innovation au sein des parlers rgyalrongs.
7. PRONOUNS: DEFINITENESS MARKER {-DE}, REMOTE DEIXIS, AND FIRST PERSON INCLUSIVE PLURAL /ij.OfO/ ~ /ij.OfAI ~ /AH.O/
Not quite unexpectedly, the two dialect groups differ also in their lexicon. However, given the scarcity of lexical resources, it is difficult to ascertain how far the differences go. At least with respect to the verbs, where I have sampled detailed lists for two reference dialects, Domkhar (Shamskat) and Gya-Sasoma (Kenhat), the following picture emerges: disregarding minor and not-so-minor differences in meaning the two representative dialects share about 85% of the vocabulary, that is, about 15% of the verbs of one dialect are not found in a corresponding dialectal form in the other dialect Taking into account the observed differences in meaning, the percentage of verbs and readings not found in the respective other dialect is about 25%. 69 This figure does 69 Actually there is a higher percentage of particular meanings from Gya-Sasoma not foWld in the Domkhar dialect (29%), than the other way roWld (21%), which is
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
277
not include the considerable differences in the argument structure of corresponding verbs,70 for which we do not have any statistics. Differences in the nominal vocabulary can be best demonstrated with various pronouns, including definiteness markers. Defmiteness (or givenness) of a noun is indicated by the marker {-po} (allomotphs: /-po/, /-bo/, /-wo/, /-o/)'1 in the Shamskat dialects and Leh, as well as in Central Zanskar, but by the marker {-de} (allomotphs: /-tel, /-del, /-rei) in the other Kenhat dialects (see Table 2, column 7). The latter marker is certainly related to the Old and Classical Tibetan demonstrative pronoun de 'that'. In Old and Classical Tibetan, demonstrative pronouns, which simultaneously serve as definiteness markers, take the final slot of an (extended) noun phrase (if there is one). In all Ladakhi dialects, the two functions of the Old Tibetan demonstrative pronouns: anaphoric or referential, on the one hand, marking definiteness or givenness, on the other, are split and the referential demonstrative pronouns typically take the initial slot of a noun phrase,72 leaving the original slot to the definiteness markers, which are typically no longer referential. In the LLV (a Lower Ladakhi version of the Kesar epic, written down at the beginning of the 20th century), the remote deixis de is also found in the fmal slot In this case, it is not always referential and functions more like a defmite article. I have also observed the definiteness marker /de/ very infrequently in the speech of an elderly person of Khalatse. In both cases, the use of /de/ instead of {po} might be due to interferences with the classical language, but it might possibly also indicate that an original marker /de/ has been replaced by the innovative marker {po}. due to the fact that we have significantly more readings positively attested for GyaSasoma than for Domkhar, while we still lack the corresponding confirmation or rejection from the Domkhar dialect. 70 To give only one example: in neutral situations verbs of consumption generally do not take an ergative marker for the agent in the Kenhat dialects, but they must take an ergative marker in the Shamskat dialects. 71 This marker might be related to the Old Tibetan pronmm bulbo discussed below (p. 278) as well as to 1he Old Tibetan sentence-final definiteness marker -Xo (i.e. assimilating wi1h a preceding consonant), which showed some nominal properties, such as :Rlural marking, see also note 75 below. 72 Most probably, the pronouns were prefixed to their head wi1h the help of a genitive, as is sti.ll visible in the Gya remote deixis /phail. Stereotypical phrases with a prefixed genitive pronoun, such as debi tshe, debi dus 'that time', often also contracted to compounds detshe, dedus, are very frequent in Classical Tibetan.
278
BETIINA ZEISLER
A trait specific to the Kenhat dialects is that the def"miteness marker {-de} shows assimilation features, indicating that the originally independent marker has become part of the intonational wtit 'word'. Both defmiteness markers are often combined with a demonstrative pronowt or follow an otherwise sufficiently defmite noun, e.g. family terms, cf. for the Kenhat marker: CEM /i kitapte/ IJ-di kitab-de 'this book', /i lamde/ IJ-di lam-de 'this way/ road', /i igere/ IJ-di yige-de 'this letter', /ire/ IJ-di-de 'this', Iare/ !J-a-de 'that', /inarel IJ-dina-de 'this one/ only this/exactly this', /anarel IJ-ana-de 'that one/ only that/exactly that',73 /abaresel aba-de-si 'of/by father'; ILV: de srinpo-de 'that monster' (p. 6, 1. 8), de mgodgu-de 'that nine-headed one' (the monster; p. 6, 1. 9), and kho!J-i no-de 'his yowtger brother' (p. 15, 1. 17). 74 In place of the remote deixis /a/ 'that over there', common in Shamskat and Leh, speakers of the Gya dialect typically use the adjectival root /pha/ 'across, beyond', which is placed in front of the headword with the help of the genitive marker, cf. /phai triigu/ 'that child over there'. A similar substitution for the remote deixis /tel 'that' is fowtd in Spiti and Nyamkat with /phil 'that' (Sharma 1992: 46f., 49, 141, 143), but not in Tot, where again /tel is used (Sharma 1989: 296f. ). Perhaps the most prominent difference between the two dialect groups with respect to pronouns is the expression for the frrst person inclusive plural. While the Shamskat and Leh form /gatag/ and Balti /garag/ yatay are derived from the frrst person singular pronowt /gal ya, many Kenhat dialects make use of a quite different pronowt: GYS /ftoyo/ ~Isao!, SHA /ftoyal, NYO /fto/, CEM and Zanskari /aho/. This word apparently derives from an old pronowt!J-u ~ IJ-o (or directly from IJ-oskol 'we', JAK sub IJ-oskol; this etymology is also current among learned Zanskaris, but not fully motivated by the sowtd laws, which would yield Zanskari */ohol/).
73 Cf Zeisler (2006: 79f, examples 13 to 17) for the emphatic or contrastive use of directional markers with pronmms. 74 Quite similarly, /de/ is used as a definiteness marker in Kyirong (Huber 2005: 58, 7lf). In this context, I should like to mention Isao Honda's paper on 'Grammaticalization of demonstratives and double determination in Tibeto-Burman languages' given at the 36th International Conference on Sino-Tibetan Languages and Linguistics, Melbome 28-30 November, 2003. He shows that the grammaticalisation of demonstratives as definite markers as well as the combination with new pronouns as polymorphemic pronouns is a common feature in Tibeto-Burman. Cf also Honda (2007) more specifically referring to Tamangic languages, but with an example from Kyirong of using the definitive marker de with pronouns (no. 11, p. 106).
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
279
In Old Tibetan texts one can fmd the plural determiner ~ochog, which can be used on verbs, 75 and the (exclusive?) plural forms ~ucag ~ ~ocag, with the diminutive (?) 76 also ~ubu(cag), 'we'. The BRGY further mentions a dialectal diminutive (?}'6 form ~ugu 'I' The sequential discourse markers ~ynas 'and then' and ~uygi-r.fesu 'after that' seem likewise related. Since the form ~uynas alternates with the sequential discourse marker denas derived from the remote deixis de 'that' (the only form in later texts), an interpretation of this as both a dialectal or regional variant of the remote deixis and as a complementary proximate or medial deictic function seem motivated. However, according to BRGY, ~u ~ ~o is an Old Tibetan form corresponding to the proximate deixis ~di 'this'. Similarly, in one of the transcriptions from the Gya dialect we fmd the spontaneous form /.fiuri/ in place of /ire/ 'this one'. The consultant herself, who in this case was also the narrator, was somewhat astonished about her own usage and went on to explain that in the speech of old people one could still hear /.fiu tog/ 'give this' instead of /i tog/, but that nowadays it is mostly replaced by the latter form. This rather inconsistent data allows two interpretations: either the two forms ~ and ~u originally represented two different pronouns, the first distal, the second proximate,77 or that the original ~fl:lu was completely unspecific with respect to distance or nearness. In the latter case, its original function may have been a quite different one and/ or it may have been borrowed with functional shift from an unrelated language. The Kyirong demonstrative pronoun /Q.:dV 'that' (Huber 2005: 71, 73 with n. 79) corresponds very closely to the pronoun /ote/ ode which Koshal (1979: 122f.) describes for the Leh dialect as "non-proximate", 75 Cf. Gzermyig (fol. 78a.4, Francke 1928: 498) dpanlhabis gsurJsna ci gsuiJ manan-no-cog 'When the god-like teacher spoke, we (incl.) have not listened to whatever he said' This usage shows clearly that the sentence final marker bo is itself actually a definiteness marker. The pronominal character of the element bois evident from the use of the plural marker with noun phrases: the element bo takes the exact place of the pronoun or definiteness marker de, cf. the Old Tibetan Chronicle (1. 290, Imaeda & al. 2007: 212) g,Zanmyi bochog 'all the other people' 76 The use of a diminutive for the first person is certainly well-motivated. The diminutive itself shows two variants -bu (assimilating with the preceding consonant) and -hu [vu], rarely also a non-assimilated -gu (cf. Uray 1952). This alternation could result from an original *bvu. 77 In this case, it might be possible that the classical pronoun hdi 'this' results from a contraction of bude; for a corresponding sound change affecting the final vowel, cf. the Kyirong form /Q:dJJ < bo +de.
280
BETIINA ZEISLER
referring to items ''not close either to the speaker or addressee, but within sight". According to Rebecca Norman (p.c.) this pronoun functions as a medial deixis, when pointing to an item near to the addressee and typically located between speaker and addressee. In Domkhar, I otel refers to things that are (a) visible (in contrast to /tel) and (b) near to the addressee but somewhat further away from the speaker (thus not in between). Quite in contrast, the element o merely has an emphatic function in Gya and can be added to any of the pronouns thus Iii, /oil '(exactly) this', /tel, /ote/ '(exactly) that', /phail, /o phail '(exactly) over there', /phai-phai/, and /o phai-phail '(exactly) far over there'. As a frrst person plural inclusive pronoun, /fto/ is also found in neighbouring varieties, e.g. Spiti /ftozakgja/ (''hozakgya'') 'we' (Sharma 1992: 47), possibly also /wogja/ 'you (pl.)' (Sharma 1992: 21)78, Tabo Spiti lhgj ~ /wgj 'you (to elders)', /wQ.aJal 'we (incl.)', /w.Q.jak/ ~ /wQ.akunl 'we (incl. fam.)' (Veronika Hein, p.c.), and Nyamkat /ftoJakl (''h~sak'') 'we (incl.)', but also /fto/, /ftoJakl 'you sg./pl. (extra-honorific)' (Sharma 1992: 138), cf. also Dingri IQJagl, Drokpa IQJagl ~ lhQ.rag/ and lhgj, Mustang /!,rag/~ IQ.rag/ 'we (incl.)' (Herrmann 1989: 45 for Dingri, Kretschmar 1986: 42 for Drokpa, 1995: 523 for Mustang). Cf. also Kyirong /ftu/ ~/ora:/ (Huber 2005: 68). The phenomenon seems to be even more widespread and might be also found in the East: Themchen ( Amdo) shows a first person inclusive plural pronoun based on the element /a/ f:tu plus plural marker /W *chabo: laW f:tuchabo 'we incl.' vs. /gatd'u! yedchabo 'we excl.' (Haller 2004: 50). The use of /fto/ for both frrst and second person in Spiti and Nyamkat indicates that neither use is original, but rather derived from a third person or demonstrative pronoun. The use of a third person pronoun for the frrst person is not uncommon in Tibetan (cf. also CT khobo and khomo 'I' < kho 's/he' plus gender-specific nominaliser). As the speaker distances him- or herself and becomes less prominent, one can interpret the use of a third person pronoun as an expression of humility. It is, however, not immediately clear why this kind of humility should be found in the inclusive and not in the exclusive plural and why not also in the frrst person singular pronoun. 78 This latter form does not appear in the section on pronouns. The labial glide instead of the laryngal or a plain vowel is probably triggered by the rounded vowel. A similar feature can be observed in other Tibetan dialects where one finds /worna/ for boma 'milk'
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
281
The opposite use of a third person pronoun for an addressee of higher status as, e.g., in German, can be interpreted as a strategy of avoiding direct addressing, which would be considered impolite, and it also serves to create a distance in favour of the addressee, making him or her less involved in the situation and thus less responsible. This strategy might also underlie the use of the inclusive pronoun, which implies some sort of address and could therefore be interpreted as you and me. This analysis would at least account for the compound f:toray: f:to 'that' (you) and ray 'self' (me).
8. BIMORPHEMIC CASE MARKING AND CASE NEUTRALISATION The Kenhat dialects show two features that at the first sight appear to be contradictory. On the one hand, some of them show the substitution of a case mmpheme or even the combination of two case mmphemes in place of a simple case morpheme (genitive /-if ~ /-e/ vs. /-se/ (< *-isu), locative-ablative /-ne/ ~ /-nesul). On the other hand, all of them have apparently neutralised the two distinctions, that between ablative and locative case marking (both found as /-nel or /-nesul although the latter form appears to be preferentially used for the ablative function), 79 and that between agent and possessor marking (instrumental vs. genitive, both /-if~ lei and, in some Upper Indus dialects, alternatively /-se/;80 cf. also Table 2, columns 8 and 9). 79 A similar neutralisation can also be observed in the Shamskat dialects where the ablative or the ablative postpositions may be used with a non-dynamic locative meaning in specific contexts, cf ARA /lame-ka -lam-e-kana kbi duk/ lam-mi-ka-(na} kyi bdug 'on the road (PPosLoc- PPosAbl) are dogs' DOM lama gare?-arna naJ]-na jot./ ama gare I ama nfJI'j-na yod I 'Where is mother?-Mother is in the house (Abl)' In such cases, the ablative and ablative postposition narrow the focus onto a more specific location. More particularly, they are used to emphasize an obvious location, implying the rhetorical question 'don't you know?' or 'can't/didn't you see?' The emphasis is strongest when the LCT argument is in the focus position. Cf also the non-elicited KHAL ls:gomne Pod-na tsamtsek !Jhos jot, .. ./ SIJonmabi Bod-na tsamtsaZig chos yod 'As many religious works as there had been in Tibet of olden days (Abl), ' The Shamskat ablative marker I-na! is, of course, formally identical with the OTICf locative marker na. The unmarked rendering for 'in the house' is lnaJ]-al nOIJ-la or, with a locative postposition, lnaJ]-i-a:gl nfJI'j-YJi·nOIJ. 80 Cf. GYA lta-se :gama/ rta-si 17Jatna (Leh lrte roamal rta-bi mama) 'tail of the horse' The substituted genitive marker 1-se/ is also used with postpositions or may even be the only possible form for postpositions as in Shara, cf. GYA lta-se-ka/ or SHA, GYS lta-se-ha/ rta-si-ka, SHA *lte-ha/ (Leh lste-ka/ rta-hi-ka) 'on the horse'
282
BETIINA ZEISLER
The latter statement, however, should be wtderstood in comparison with the situation fowtd in Old and Classical Tibetan, which cannot be taken a priori as the original case system. On the contrary, the Old Tibetan instrumental and ablative markers are obvious bimorphemes, the second element generally surviving as -s in Old Tibetan, thus instrumental-agentive {kyis} = genitive {kyi} + -s; ablative nas = locative na + -s, ablative las = dative-allative la + -s (cf. LSI p. 27, Simon 1940: 385-386, DeLancey 1982: 27, 1984: 61f., Tournadre 1995: 267f. with note 14). It seems that the functional dissimilation occurred not too long before the stage of Old Tibetan. In Classical Tibetan one can still observe the frozen usage of the dative-allative marker la with ablative function (only for particular verbs, cf. JAK sub la IV) and with partitive function (Zeisler, 2006: 70, 75, 77f., examples 7-12). The possessor-agent distinction apparently did not affect all varieties-or started breaking down rather quickly. 81 One can observe the seemingly irregular use of genitive case marking of agents in some Old Tibetan documents as much as in early classical texts (this is often corrected in later editions, occasionally also in cases where a genitive marker would have been correct). While DeLancey (1984: 62, 69) suggests that the additional morpheme of the ablative marker is derived from an unbowtd motion verb *sa 'go, leave', DeLancey (1982: 27) and Toumadre (1995: 267f.) suggest that the additional -s morpheme shared by the ablative and the instrumental represents, as DeLancey puts it, an "abstract superordinate Source category". This argument is put forward by DeLancey and Tournadre in favour of a conceptual relation between ablative and ergative marking in Tibetan. The lack of distinction between agent and possessor at the level of morphology seems to be conceptually deeply rooted. Working with my consultants, it is often extremely difficult to decide for or against an agentive interpretation of the first argwnent in sentences with unclear agentive or valency status, because a possessive interpretation (reducing two arguments to one) is always acceptable and apparently blocks any other interpretation. 81 Although this phenomenon has never been studied in detail, and tlrus no statistics exist, I got the impression that the two markers are more often confounded when 1hey appear in their syllabic form {kyi}, {kyis} after a closed syllable than when they appear after open syllable as -i and -s respectively. The regular spelling -s instead of a linguistically expectable *-is, may indicate some more prominent differences in prommciation These might have resisted the phonetical assimilation operating on the syllabic marker for some time. In that case, the observable confusion in classical texts would clearly be a secondary phenomenon.
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
283
Complex case motphemes combining the ablative marker nas, or more probably the locative marker na with a motpheme su, are found in Classical Tibetan nassu (e.g. in the Btsunmo Bka~i-thayyig, Laufer 1911)82 as well as in Kenhat (locative-ablative /-ne/ ~ /-nesul) and some of the Western Tibetan varieties (Spiti: locative-ablative /-na/ ablative /-nasul, Nyamkat ablative-in the Poo dialect, however, instrumental-/naso/, cf. Sharma 1992: 37-40, 131-133). The complex form or at least its second syllabic element is also attested with the ergative marker in Balti /-(i)si/ and northern Nubra 1-zel (cf. ARA /khoi/ kho~i 'his/her' vs. lkhozel khosi 'by him/her'), Tot/-i/ ~ /-ise/, Nyamkat/-so/ (Sharma 1989: 286, 288f., 1992:131) or /-sui (LSI p. 27), Tabo Spiti /-sui (see below), and, as already mentioned, in the genitive marker of the Upper Indus dialects (Cemre, Shara, Gya) /-i/ ~ /-se/. It was apparently also borrowed into several Shina dialects for the imperfective tenses (cf. Bailey 1924: 11 for Gilgit, Sharma 1998: 60 for Brokskat, Schmidt & Kohistani 2008: 51-53 for Indus Kohistani). 83 If the complex form is a secondary derivation of the genitive marker, as Bielmeier (1985: 90) suggests, it could well have developed out of a motpheme *-su (or *-so) in an unstressed position. One may add the observation that the Balti motpheme I -sil may be dropped, leaving behind the mere genitive marker I -il for the agent (Bielmeier 1985: 90). As Veronika Hein kindly pointed out to me, the Tabo Spiti form of the ergative marker might be either the element /-sui (optionally plus an element /-lui or /-kun/) or the postposition (?) /-ilo/ (or a genitive form plus /-lo/). 84 The ablative marker, also used for instruments, takes the form of the locative marker /-na/ plus genitive /-i/ ~ 1-kil plus the optional elements /-sui (+ /-lui or /-kun/) or /-lo/, thus /-nail (or /-ne/) ~ /-nakil, besides /-naisu/ ~ /-nakisul (/-naisulul ~ /-nakisulul or /-naisukun/ ~ /-nakisukun/) and /-nailo/ ~ /-nakilo/. The short form 82 In this text, one can also find the combination of the nominaliser pa + instrumental marker-s+ su. Although the massive use of the bimorphemic markers is triggered by the metric structure of the text, it must have been licensed by the grammar of the author's (compiler's) dialect. 83 This form is not found in the most archaic forms of Shina, e.g. in the Palula dialect (Schmidt & Kohistani. 2008: 51). The Palula dialect separated from the modern Shina varieties in the early or mid&e 17th centwy (Ruth Leila Schmidt, e-mail communications IV/08). 84 According to Veronika Rein, the function of /-lo/ - /-lui and /-kun/ is quite opaque. These elements may merely add more emphasis to the preceding morpheme.
284
BETIINA ZEISLER
/-nail (or /-ne/) ~ /-naki/ as well as the other Western Tibetan I-na!~ I -nasu/ variants, help to explain the form of the Kenhat locative-ablative marker /-ne/ as being a contraction from the full form /-nesu/ < *naf:tisu (locative plus postposition) and as not being derived from the Old or Classical Tibetan marker nas. The interesting pattern in Tabo shows that the derivational element for the Old Tibetan instrumental and ablative markers could have been a postposition, and thus originally a noun. 85 In the case of the ablative, the element /sui and a preceding genitive marker /i/ or /kil would have been applied, just like a postposition, to a preceding noun, which, however, was modified by a locative marker. In the case of the instrumental and ergative, either the plain 'noun' (as in Tabo) or the postposition (genitive marker plus 'noun', as in Balti) would have been directly applied to the unmodified noun. The motpheme *-su/*-so was also applied to location adverbs, such as ma 'below' and ya 'above', to derive directional adverbs: ma-s 'from below' (i.e., upwards) and ya-s 'from above' (downwards). A syllabic form ya-(s)se86 is attested in two Old Tibetan documents: r.fe Guge Rkayphrangyi giiebo myi brgya rkya brgyazig yasse bywjna (Pt 1136, 1. 47) 'when the go-between of the lord Guge Rkal)phran [together with] hundred men (and) hundred wild asses 87 appeared from above'; f:t[o]naiig re8igna bya thaykar thaynay [=thaynag?] giiissig yase [byuy?] (ITJ 0731, 1. 69) 'Then, once, two birds, a white eagle and a ?black eagle appeared from above'. In both cases, the context leaves no doubt that the go-between and the birds come down, namely from the uplands of Guge and from the sky. Without the evidence from the dialects, these forms could not be analysed. Thomas (1957: 31) leaves yase untranslated, while Uray (1972: 9) describes yasse as "an adverb of unknown meaning". It should be noted that the relational markers OT/CT las and bas used in expressions for non-equative comparisons are likewise bimor85 For Roland Bielmeier (p.c.), however, 1his does not necessarily follow. He thinks that the morpheme *-sui*so could have been a case marker, joined to a case marker that had lost its original fimction. This would mean that the genitive marker originally had an ablative or at least locational fimction. Possessors (and agents) would tlrus have been treated as source of origin. 86 The geminated spelling (yas.se) might reflect an unclear syllable boundary and might thus indicate that the vowel was already about to be lost. ffl rkya. Perhaps merely an error for rkyaiJ, the Tibetan wild ass (equus kiang), but the final -TJ could be a derivational suffix. Possibly an old generic term for equids.
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
285
phemes with a second element -s. The Ladakhi dialects as well as several Western Tibetan dialects show the use of the morpheme say (cf. Hu Tan 1989: 403f. for the dialects of "Ari", i.e. Mya~ris), which could at least be formally analysed as the above element *sul*so plus conjunction ay. What is interesting is that, even if this morpheme is ultimately unrelated, it behaves like a postposition in the Kenhat dialects (including Leh), thus /X-e-sal)/ 'in relation to X', but combines with the morpheme ba in the Sham dialects: /X-ba-sag/. It is not at all clear to me whether the derivation of the complex ablative and instrumental-agentive markers was completely parallel. If so, then the genitive must have been in use for marking the agent as an (animate) possessor of the action. The derivational element *-su would then have been introduced to dissimilate the function of possessor and agent It is interesting in this connection to note that the notions of (inanimate) instrument and (animate) agent are conceptually independent in all Ladakhi dialects: the instrument is in general not expressed by either the genitive marker /-i/ or the ergative marker /-(i)s/, /-(i)si/, 1-zel, but by the comitative marker /-dal)/, 1-fial.JI, /-na(g), or /-oW. This feature is also found in Tot: comitative /-dal)/ (Sharma 1989: 289) and in Nyamkat: comitative /-rag/ or ablative /-naso/ (Sharma 1992: 133), while in Spiti only the ablative marker is used (Sharma 1992: 38 and Veronika Hein, p.c.). However, if the possessor-agent distinction found in Old Tibetan and some of the north-western Ladakhi dialects is an innovation not shared by the Kenhat dialects, why do we fmd the suppletive form /-se/ for the genitive in these dialects at all? The contemporary lack of distinction in these dialects would thus appear to be a secondary neutralisation. On the other hand, the suppletive form of the genitive is not permitted with personal pronouns in Cemre and Gya, whereas the complex ablative form is quite common with demonstrative pronouns (/tenesu/ 'thereafter, and then' for denas). This might point to the fact that the use of the suppletive genitive form was an innovation, on analogy perhaps to the use of the complex ablative. Furthermore, the Old or Classical Tibetan instrumental marker {kyis} (homophonous with the ergative in Shamskat and with the genitive in Kenhat) can be found in Ladakhi with non-agentive verbs to express a (mostly physical) cause of an event 88 In similar contexts, Tabo Spiti 88
There is a strong tendency to replace it by other constructions, cf.:
286
BETIINA ZEISLER
shows the rare use of the ergative marker /-sui. Similarly, and corresponding to Classical Tibetan, the medium-argument of the agentive and non-agentive verbs of filling takes a genitive marker in Kenhat, but also in Shamskat (Zeisler 2007: 407). At first sight, the genitive or the instrumental marker for cause and medium arguments appears to be a remnant of Old Tibetan instrwnent and agent marking. The neutralisation of agent/ cause and possessor marking would then be an innovation. One could accordingly argue that the introduction of the comitative da1J as a marker for the (inanimate) instrwnent was caused by the neutralisation. But in this case, one can but wonder why it should have been more important for the rather peripheral instrument argument than for the agent to find a new marker. Yet again, why should there be a separate marker for the instrwnent in those varieties where the possessor-agent neutralisation did not take place, as in Tot and Nyamkat or in the Shamskat dialects? Likewise, if the neutralisation of the possessor-agent distinction in the Kenhat dialects was due to a process of simplification, why should the same trend not have led to a complete loss of the suppletive form as in the Leh dialect, but, quite the opposite, to an overgeneralisation? For the time being, I cannot offer more than a general suggestion: the possessor-agent distinction might have been superimposed on an early form of Old Tibetan, and the new morpheme *-su might then have spread and taken over ever more functions of the comitative marker da1J, until the latter was more or less completely replaced in Old Tibetan. Before the process was complete, the new possessoragent distinction was also superimposed on the non-Tibetan substrate
(i) SAS
a jura khimsa-s
gaksok canal sweeping-Instr got.blocked 'The canal apparently got blocked because of the sweepings.' b jura khimsa-s-basc gaksok. canal sweeping-PPoslnstr (or: sweeping-Erg done) got.blocked id, lit. 'The canal, the sweepings having done, apparently got blocked'
(ii) a triia anc Aymc ~ A.ymosc hU11111cdla GYS child-Dat aunt Al]mo-Gen(=Instr) suddenly 'The child got suddenly hit because of aunt Al]mo.' b triia anc Aymc-lfcre ~ .Aymosc-lfcre hU11111cdla child-Dat anntA.-PPoslnstr (or: A.-Gen/Erg done) suddenly id., lit. 'The child, aunt Al]mo having done, got suddenly hit.'
thok. got.hit thok. got.hit
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
287
in the western areas. The Kenhat case system might have been reorganised at a later stage according to the old system of the Kenhat substrate, but with a semantic shift concerning the morphemes. In this scenario, the additional motpheme *-su might have lost its grammatical function, but could have been retained as a more prestigious phonetic form. 89 Alternatively, one could suggest that the suppletive form, introduced to mark an agent, was over-generalised to such an extent that the possessor-agent distinction collapsed. That the Shamskat case system was not reorganised could be due as much to a different substrate as to a later contact with the Old Tibetan lingua franca. The Shamskat medium construction is obviously borrowed from Kenhat. It is thus not necessarily a remnant of an earlier Old Tibetan instrumental construction, and the Kenhat construction itself could have as easily been borrowed from neighbouring varieties or the Old Tibetan superstrate. The same could be said about the cause construction. More probably, both constructions constitute a natural development from the available agent marking. For the cause argument in particular, there seems to be a closer conceptual connection with the agent argument than with the instrument argument. As far as the medium argument is concerned, the Gya dialect shows an interesting distribution of genitive and comitative marking. With transitive-causative verbs, the genitive can only be used when the 'subject' slot is filled by a natural-force argument and not by a humanagent argument In this case, the medium argument is apparently conceptualised as a coincidental cause. With a human agent, however, the medium argument is conceptualised as a mere instrument of the agent and thus takes the comitative marker. At this point the Shamskat dialects again differ quite significantly. Although the genitive medium construction is borrowed from Kenhat, the comitative marker can only be used for rather unexpected medium arguments (cf. Zeisler 2007: 412), and the question of whether the sub-
89 This has been nicely confirmed by the GYS consultant, who in 2008 stated that the bimorphemic form is more /deal rdeba 'nice' and thus more honorific. ln the example sentence given by her just before this statement, the reincarnated priest or /rimboche/ received the bimorphemic form as possessor, while the ordinary priests as agents received only the monomorphemic form, to emphasise the contrast in status: /11J!.11ll:J m~elegun-e rimbodhe-se kalowa san./ nanilj memelegun-ni rinpoche-si bkabslobla gsan 'Last year the monks/priests-Erg listened to the preaching of the chief priest-Gen'
288
BETIINA ZEISLER
ject slot is filled with a human agent or a natural force does not play a role. The neutralisation of the location-source distinction might be explained along the same lines as the neutralisation of the possessoragent distinction, although it seems that the complex ablative marker developed independently and perhaps earlier than the suppletive agent marker, as the neutralisation of the locative and ablative markers can also be observed in the Shamskat dialects. It is also interesting to note the complementary distribution of the land the n- forms in Ladakhi. While the 1- ablative is completely lost (or was never developed), the 1- allative is the most common directional marker. The ablative function can thus only be expressed by an n- form. As a result, the locative function is often lost or rather infrequent: in the western Sham dialects, /-na/ is used mostly for the ablative90 (in contrast to northern Nubra, no form /-nas/ is available), while Leh has only /-ne/ (no form /-na/ or /-nas/ available), which again is used mostly for the ablative. On the other hand, /-ne/ nas is generally used in the Upper Indus dialects and Gya to indicate both the location and the source, and to a certain extent this dual function is also found with the complex morpheme /-nesu/. Cemre shows the use of /-na/ for an allative 'into', but since this is somewhat against the non-dynamic character of the original locative marker na, this might be a contraction of either the postposition /-D.al)(a)/ nay(la) or of a dynamic locative-ablative marker */nasu/. Another case of bimorphemic case marking, again in combination with a neutralisation of the location-source distinction, could be observed in Gya. The locative-purposive marker {tu} is rarely used in the Ladakhi varieties, where it is typically restricted to pronouns and place names. In Gya, however, it may appear more frequently after open syllables, where it takes the form /-ru/ or 1-ro/. Most often it is combined with the dative-allative marker I -a! Ia. Despite the original locative values of both markers, the combination may have an ablative function in Gya. In fact, if the context allows an ablative interpretation, as in the case of movement and transfer verbs, this will be the preferred reading, thus /a!fose ~ola khambaru-a kher./ ajos(se) jola khayparola khers 'Elder brother took/ brought
90
But see note 79, p. 281 above.
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
289
the bag( s) thither out of the house (preferred reading)/ to the house (reading depending on contextual support)' The conceptual non-distinction between, or overlap of, locations and sources seems thus to be deeply rooted, and the same can be said about the non-distinction of agents and possessors. The additional morphemes apparently never had a greater conceptual impact and the motivation for adding them fell quickly in oblivion. The Kenhat consultants, at least, interpret the simple forms just as an 'abbreviation' of the complex forms. Nevertheless, as it turned out during new fieldwork in 2010, in case of ambiguity between an agent and a possessor (or rather between a main possessor and a secondary possessor), a speaker may prefer to use the hi-morphemic case marker for the agent (or main possessor) and the mono-morphemic case marker for the (secondary) possessor. This will be indicated by indexing the case marker and underlining its function in the following examples. As one can see in example (2a), however, the hi-morphemic form may also be used for a secondary possessor, particularly when the mono-morphemic form of the genitive-ergative would be more or less homophonous with the absolutive as in the case of /a!fi/ [aft] 'elder sister' and /a!fi/ [aft] ~[aft:] 'ofi'by elder sister' In such cases, it seems that the frrst genitive-ergative marking is by preference interpreted as referring to the agent (main possessor). For more pragmatic usages, see also note 89, p. 287 above. (1) a GYS
Ayme
kelak dress>
khimtsea fol. neighbour-DatLoc entrusted
? 'Agmo's dresses were entrusted to the neighbours [by her].' That is,Aymo entrusted [her] dresses to the neighbours.91
91 Mentioning both possessor and agent, if they are identical, as in situations of reflexivity, is generally avoided When the action is typically performed upon one's own things, impersonal possessive constructions are preferred to a transitive rendering. While the Kenhat informants, as could be expected, have great difficulty in telling the two constructions apart, the same impersonal possessor construction appears also in the Shamskat varieties, where agent and possessor marking are morphologically distinct and where no misconception is possible. Again, when the action is typically performed upon one's own things, the possessor construction is preferred.
290
BETIINA ZEISLER
b .Agmose
kelak
khimtsea
tfol
neighbour-DatLoc entrusted
?
?Aymo's dress(es) 'Agmo entrusted [her] dresses to the neighbours.' ?'Agmo's dresses were entrusted to the neighbours [by her].'
(2) a
Aymose
atfise
GYS
<elder.sister-Gen/Erg-1 dress>
kelak
elder sister's dress(es) ?
elder.sister-Gen/Erg-1> dress
?Aymo 's elder sister ?
elder.sister-Gen/Erg-1
dress>
?Aymo's elder sister's dress(es) khimtsea
tfol.
neighbour-DatLoc
entrusted
'Agmo entrusted [her] elder sister's dresses to the neighbours.' ? 'Agmo 1 's elder sister2 entrusted [her2] dresses to the neighbours.' ?'The dresses of Agmo 1's elder sister2 were entrusted to the neighbours [by her2].'
b .Agme
atfise
kelak
elder.sister-Gen/Erg-1> dress
Aymo 's elder sister khimtsea
tfol.
neighbour-DatLoc entrusted 'Agmo1's elder sister2 entrusted [her 112/their] dresses to the neighbours.'
9. MARKING OF TENSE AND EVIDENTIALITY The two dialect groups also differ in the selection of mmphemes or complementary verbs for complex tense forms, the most prominent being: -
complementary verb /dati (Sham skat and Leh /duk/) as marker of continuous or repeated events
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
291
-
use of the non-contracted combination of nominaliser /-ba/ ba plus auxiliary /-fi.ot/ yod (Shamskat /-et/ or {-bat} and Leh /-at/), at least in combinations with evidential markers, cf. e.g. GYS /saon tab(b)afi.ot(t)ro/ sabon btabbayodJ:zgro (LEH ltab_!Lt(t)ro/) 'might sow' or /saon tab(b)afi.okanak/ btabbayodmldzan*IJ-og 'must be/have been sowing' (LEH ltabetkjak/)
-
future tense {-kan} (kan, ran, han) < mkhan92 for speaker-prominent events (Shamskat and Leh /-in/), 93 also used in Cemre and formerly in Gya for (speaker-prominent) present tense (Shamskat /-et/ and Leh /-at/)
-
generic and inferential marker {-kak}, negation {-kamanak} ?< mkhan *~Jog or rag (Leh /-anok/, Sham skat /-intsuk/- /intsok/)94
-
distance marker {-kanak} (Leh /-kG)ak/) < mkhan /(k(h)antsuk/- /(k(h)antsok/)
-
regular use of inferential past marker {-tok} (tok, dok, rok), not common in Shamskat
+
+ *!Jog (Shamskat
The evidential motphemes /-kan/ and /-ak/ ~ 1-kakl are again found in the neighbouring varieties, cf. the speaker-prominent future I -(k)anl (Pin Spiti), /-kan/ ~/-ken/ (Tot), /-kajinl > /-ken/ (Tabo Spiti) vs. nonspeaker-prominent future /-(k)ak/ (Pin and Tabo Spiti) (Sharma 1989: 315f, 1992: 80-83, Veronika Hein, p.c.). The ftrst motpheme is also found in Kyirong, where it is used in the form {-k~} for habitual events in general and in the form {-l&(ji:)} for future acts of the speaker (Huber 2005: 110, 124). An inferential future marker I -kak/ is also found in Mustang, as well as a future motpheme /-ka/ + /-rak/ or /-nak/ ?< mkhan +rag, expressing certain or generic knowledge (Kretschmar 1995: 145, 149f). The Mustang inferential past marker {tuk} (tuk, ruk; Kretschmar 1995: 156) might likewise be related to the above-mentioned {tok}. This, in turn, might be related to the auxiliary f:ulug, which in many Tibetan varieties has an experiential (eye-witness) function, but is not attested as such in Mustang. Although this does not seem to be immediately intuitive, the experiential function of the auxiliary f:tdug can be conceptually connected Signalling certain knowledge (cf also Zeisler 2004: 652, note 243). ln Nyoma only with 1he awilli.ary /-(J)in/ added to 1he future verbal nolDl. 94 A seemingly related inferential (future) marker is /-(fi.)ak/ *bog; Leh /ok/, westem Sham and Ciktan Purik /-uk/ -l-oki or /-pok/, but northern Nubra {-suk}- {-sok}. Perhaps the marker spread from the Kenhat to the Sharnskat dialects. In the Nyoma dialect, however, the form {-kak} conveys the notion of definiteness and implies intimate knowledge or responsibility for acts of third persons, replacing in this ftmction the morpheme /-in/. 92
93
292
BETIINA ZEISLER
with an inferential marker. In contrast to generally shared knowledge, knowledge based on a singular (visual) experience is less certain. Its content merely 'seems' to be the way it was perceived. The experiential auxiliary f:tdug could thus be interpreted as a reservation on the part of the speaker concerning the truth-value of the statement: 'apparently' or 'as far as I could observe'. This strategy of reservation would explain the possible mirative use of the auxiliary as described by DeLancey (1997).
10. CONCLUSION
As the Kenhat dialects differ in many ways and quite substantially from the Shamskat dialects, it is not appropriate to treat these differences as merely dialectal variations. The differences manifest themselves most obviously at the phonetic level (fricativisation and emerging tone vs. clusters) and at the grammatical level (genitive vs. ergative agent marking, verbal auxiliaries), but also at the semantic level (many Shamskat words are not used or have a different connotation in Kenhat and vice versa). Other differences are less obvious, but are nevertheless important, such as the traces of the past-tense and imperative -s suffix, and the differences in the argument frames of verbs (as in the case of the medium argument). From an exclusively merely phonetic perspective, the differences between the various Ladakhi dialects appear to be gradual, and it may be justified to group the Leh dialect with the phonetically conservative Shamskat dialects. However, this approach does not account for the essential difference at the level of grammar, due to which the Leh dialect can only possibly be grouped with the phonetically innovative Kenhat dialects. The somewhat unexpected mixed character of the Leh dialect itself can be explained by historical facts (Leh as an important point of commercial exchange, repeated settlement of Balti speakers around Leh). Interestingly, it is the historically 'younger' dialects that have exerted the greater phonetic influence on the historically 'older' one (although one can observe some grammatical influences also in the opposite direction). The above fmdings not only show that a classification of dialects cannot be achieved solely on the basis of surface phonetics, which may be more readily influenced by external factors than the grammatical layer, but also show that the terms 'conservative' vs. 'innovative' are
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
293
quite misleading when based exclusively on phonetics. After all, the Kenhat dialects seem to have retained more lexical and grammatical archaisms than the Shamskat dialects. The two dialect groups reflect separate historical developments, and, in fact, different linguistic sub- and adstrates: While the whole of Ladakh and adjacent regions were originally populated by speakers of Eastern Iranian (Scythian), Lower Ladakh (as well as Baltistan) was also subject to several immigration waves oflndoaryan (Dardic) speakers and other groups from Central Asia. Upper Ladakh and the neighbouring regions to the east, by contrast, seem to have been populated additionally by speakers of a non-Tibetan Tibeto-Burman language, namely West Himalayan (Old Zhangzhung; for the complex historical background cf. Zeisler forthcoming). The original ethnic differences between these populations continue to be reflected to this day in mentality and culture. While the gradual Tibetanisation (i.e. the shift to Tibetan as Ll) might not have started in the north-western areas before the end of the lOth or the beginning of the 11th century, when the (possibly merely pretending) descendants of the Old Tibetan imperial dynasty established themselves as kings in Guge and Ladakh (incidentally, this is also the time of the second spread of Buddhism), the interesting development of case marking in the Western Tibetan varieties could be indicative of a very early assimilation process between early Old Tibetan and the languages spoken in Zhangzhung. It cannot be precluded, therefore, that the process of Tibetanisation already took place in (some parts of) this region during the imperial period. One could at least expect that the lingua franca of the empire would have been more dominant in this region.
294
BETIINA ZEISLER
APPENDIX: EXAMPLES OF INITIAL AND MEDIAL FRICATIVISATION
CEMRE
r, s + k--+ [hH:X]-[~] (depending on vowel) [hi],
[~i]
fricativisation blocked with palatalised velars [J.cie] < skyas 'carry, convey'; [kjot] < skyod 'go, come (hon)'; [kerak] < *skyerags 'belt'; [ketpa] < *rkyed 'back' (CT rked, sked)
s + p--+ [f] [fuk] < spug (sbub) 'pitch up' buts, r, I+ t--+ [t] [ta] < rta 'horse'
r + ts--+ [s], r + dz--+ [z] [son]< (b)rtson 'strive'; [zoq] < rdzogs 'finish, vanish' s+b--+[v] [vil] < *sbyil (< (IJ)byit) 'anoint'; [ve] < sbos 'swell' medial fricativisation [sikfon] < (b)rtsigsdpon 'mason'
SHARA [<;ARA]
d, r, s + k--+ [h] [honmo] < dkonmo 'rare'; [hunma] < rkunma 'thief; [honce] < skoncas 'dress (sb)'; but [karma] < skarma 'star' fricativisation blocked with palatalised velars [ki1po] < skyidpo 'happy'
r, I, s + t--+ [t] [taseha] < rtasika 'on the horse'; [tefin] < bltaspin 'I looked at'; [targa] < starga 'walnut'
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
295
d,s+p---+[p] [pera] < dpesgra 'speech'; [pUIJ] < spuy 'heap' medial fricativisation [ka:x;fo] < dkagspo 'difficult'; [-fm] <-s-pin (past marker)
GYA g, (b,) r + ts ---+ [ s] [safo]
+ k---+ [(r)h] [Drohar] < Sgroldkar; [karhw]] < d/skarkhWJ 'window'; ['Yohor] < mgoskor 'deception'; [P.Qthat] - [PQtkat] < bodskat 'Tibetan'; [lfhuhol] < chuskol'boiled water', but [fiomkol] < ~oskol 'boiled milk', [!f~ol] < jaskol 'boiled tea'; [3uhut]- [Jlllhut]
medial fricativisation d, l+ k---+ [h], r, s
medial fricativisation blocked with palatalised velar [-kil] < -dkyil'-middle' medial fricativisation l + p---+ [lf], s + p---+ [f] [kalfa] < bskalpa 'aeon'; UWfa]
HAMELING (ZANSKAR)
d, r, s + k---+ [h] [hon] < dkon 'scarce'; but [karfho] < dkarpo 'white'; [hunma] < rkunma 'thief; [hu] <sku 'statue', [hampo] <skampo 'dry' s + ky ---+ [ c;] (or perhaps [he;]) [c;amo] < skyamo 'lay woman'; [c;ot] < skyod 'go, come (hon)'; [c;itpo] <skyidpo 'happy' d, b, r, s + g---+ ['Y] ['Yamo] < dga~o 'happy'; ['Yofe] < bgocas 'to divide'; ['Yatpo] < rgadpo 'old (man)', ['YO] < sgo 'door'
r + gy---+ [fij] [fijatso] < rgyamtsho 'ocean'; [fijafo] < rgyalpo 'king'
296
BETIINA ZEISLER
no fricativisation and no loss of cluster with radical t [rta] < rta 'horse'; [!hta1fe] < ltacas 'to look at'; [starga] < starga 'walnut'
b, r, z + d--+ [6] [6emo] < bdemo 'nice'; [6UIJ1fe] < (b)rdWJcas 'to hit, strike'; [6uiven] < zduspin 'I gathered'; but [ldUIJma] < rdw;ma 'beam' no fricativisation and no loss of cluster ld [ldemldem] < ldemldem 'swinging'
d, s + p --+ [f] [fao] < dpabo 'hero'; [faiJpo] < dpaypo 'witness'; [fitka] < dpyidka 'spring'; [fera]
s+b--+[v] [ve] <sbas 'hid'; [khevu']
r + k--+ [rh], s + k--+ [h] [jarha] < dbyarka 'summer', [karhUIJ] < d/skarkhWJ 'window'; [leha] < Iaska 'work' r + p--+ [rf], 1, s + p--+ [f] [karfo] < dkarpo 'white'; [Jijafo] < rgyalpo 'king', ['Yolfak] < mgolpags 'scalp'; [safo]
(~)d--+ [r6],
s + [d] (< t)--+ [6]
[phur6uk] < ~phur~dug 'is flying'; ['Yoe6e] < bgos*de (< te) 'divided and'; [pe6e]
s + b,s+[b](
ABBREVIATIONS
Dictionaries, compendia, and texts BRGY Tibetan-Tibetan-Chinese dictionary, Zhang (1993) CDTD Comparative dictionary of Tibetan dialects, Bielmeier (in preparation) JAK Tibetan-English dictionary, Jaschke (1881) LL V Lower Ladakhi version of the Kesar epic, Francke (1905-41) LSI Linguistic Survey of India, Grierson (1909) RN Ladakhi-English dictionary, Norman (in preparation)
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
SPR
297
Balti-English, English-Balti dictionary, Sprigg (2002)
Tibetan: CT Classical Tibetan or Old Tibetan or pre-classical Tibetan (mid 'fh- end of 11th century) WT West Tibetan (Balti and Ladakhi) Ladakhi dialects and consultants ACH Achinathang, western Sham, consultant: SkarmaNamthak ARA Aranu, northern Nubra, consultant: Tsering Youdon CEM Cemre, Upper Indus, consultant: Padma Dohar DOM Domkhar, western Sham, consultants: Thrinlas Chosphel, Tshewang Tharchin DRS Dras, bilingual: Brokskat (Shina) and Purik, consultant (L1): Dr. Saleem Mir GARK Garkhon, bilingual: Brokskat and Purik, consultant (L2): Stanzin Angmo G YA Gya, Upper Indus side valley, narrator (Kesar story, recorded 1996): late Tsewang Norbu GYS Gya-Sasoma, Upper Indus side valley, consultant: Mengyur Tshomo, narrator (Gyapa Co 2005): Urgyen Rigzin (consultant's father) HML Hameling, Upper Zanskar, consultant: Tsering Angmo KHAL Khalatse, western Sham, speaker (personal narrative 2005): Tondrup Tshering LAG Laga, Shayok valley, consultant: Phuntsok Namgyal LEH Leh, Central Ladakh, consultant (among others): Thrinles Wangmo :MND Manda, Upper Zanskar, Hoshi, Michiyo and Tondup Tsering (1978) NYO Nyoma, southern Changthang, consultant: Rigzin Samdup PIP Pipcha, Central Zanskar, consultant: Tsering Samdup SAS Saspol, eastern Sham, consultant Phuntsok Dolma SHA Shara, Upper Indus, consultant: Thukche Dolma SYK Shayok, Shayok valley consultant: Stanzin Dorje Grammatical markers Abl ablative Dat dative-allative Erg ergative Gen genitive Instr instrumental
298 Loc PPos-
BETIINA ZEISLER locative postposition
BIBLIOGRAPHY Bailey, Thomas Graham. 1924. Grammar ofthe Shma (Si:flll) language, consisting of a foil grammar, with texts and vocabularies of the main or Gilgit dialect and briefer grammars (with vocabularies and texts) of the Khohistani, Guresi and Drasi dialects. (Prize Publication F1D1d, 3.) London: Royal Asiatic Society. Beckwith, Christopher I. 2005. Toward a Tibeto-Bunnan theory. In Clnistopher I. Beckwith, ed, Medieval Tibeto-Burman languages II, PlATS 2003: Tibetan stud-
ies: Proceedings ofthe Tenth Seminar of the International .Association for Tibetan Studies, Orford2003. (Brills Tibetan Studies Library, 10.) Brill, Leidenetc.: 1-38. Beyer, Stephan V. 1992. The Classical Tibetan language. New York: State University
of New York. Reprint 1993. (Bibliotheca Indo-Buddhica series, 116.) Delhi: Sri Satguru. Bielmeier, Roland. 1985. Das Marchen vom Prinzen Cobzan. Eine tibetische Erziihlung aus Baltistan. Text, Obersetzung, Grammatik und westtibetisch verg}eichendes Glossar. (Beitriige zur tibetischen Erziihlforsclrung, 6.) St. Augustin: VGH Wissenschaftsverlag. - - . 2004. Lexikalische Variation Wldlexi.kalischer Wandel im Tibetischen am Beispiel einiger K.Orperteilbezeiclrrnmgen. In Wil.trud Mihatsch and Reinhild Steinberg, eds., Lexical data and universals of semantic change. (Stauffenburg-Linguistik, 35.) Tiibingen: Stauffenbmg, 167-202. --.In preparation Comparative dictionary of Tibetan dialects. Vol. 1: Verbs. (Preprint No. I, 1997/pdf-version 02/2008.) Csoma de Koros, Alexander. 1834. A Grammar of the Tibetan language in English. Prepared Wlder the patronage of the govermnent and the auspices of the Asiatic Society of Bengal. Calcutta: Baptist Mission Press. Reprint 1984. (Collected works of Alexander Csoma de K6r0s, 2.) Budapest: Akademi.ai Kiado. DeLancey, Scott. 1982. Lhasa Tibetan: a case study in ergative typology. Journal of linguistic research 2.1: 21-31. - - . 1984. Etymological notes on Tibeto-Burrnan case particles. Linguistics of the Tibeto-BunnanArea 8.1: 59-77. - - . 1997. Mirativity: The grammatical marking of unexpected information Linguistic Typology 1: 33-52. Denwood, Philip. 1986. 'The Tibetan n01Dl final -s.' Linguistics of the Tibeto-Burman Area. 9.1: 97-101. Francke, August Hermann 1901. Sketch ofLadakhi grammar. Journal afthe.Asiatic Society of Bengal 70: 1--63. Reprint as: Ladakhi and Tibetan grammar. Delhi 1979: Seema Publications. - - . 1904. A language map of West Tibet with notes. Journal of the Asiatic Society afBengal13: 362--67. - - . 1905--41. Gsamyulna bsadpa/.zi Kesargyi SgniTJS biugs. A Lower Ladakh; version ofthe Kesar saga. (Bibliotheca Indica, work, 168.), Calcutta 1905--1909 (Fasc. 14), 1941 (Fasc. 5): Royal Asiatic Society of Bengal. Reprint 1981: The Epic of Gesar. Kunzang Tobgyel and Mani Dorje, eds., Vol. 29. Thimphu, Bhutan: Druk Sherig Press.
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
299
- - . 1928. Gzer-myig, a book of the Tibetan Bonpos. Tibetan text according to the Berlin-marruscript (chapter VI). Asia Major 4.4: 481-540. Ghulam Hassan Lobsang. 1995. Short sketch ofBalti grammar. A Tibetan dialect spcr ken in Northern Pakistan. (Universi.tiit Bern, Institut fiir Sprachwissenschaft, Arbeitspapier, 34.) Bern: Universitiit Bern. Grierson, George Abraham, ed 1909. Linguistic Survey of India. Vol. III: TibetoBurman family, Part I: General introduction, specimens of the Tibetan dialects, and the North Assam group. Calcutta: Superintendent of Government Printing. Reprint 1967, Delhi etc: Motilal Banarsidass. Hiisl.er, K.atrin Louise. 1999.A grammar of the TibetanDege {Sde dge) dialect. ZUrich: Selbstverlag. Haller Felix. 2004. Dialekt und En:iihlungen von Themchen. (Beitriige zur tibetischen Erziihlforschung, 14.) Borm: VGH Wissenschaftsverlag. Herrmann, Silke. 1983. Die tihetische Version des Papageienbuches. (Beitrii.ge zur tibetischen Erziihlforschung, 5.) St. Augustin: VGH-Wissenschaftsverlag - - . 1989. En:iihlungen und Dialekt von Dinri. (Beitrii.ge zur tibetischen EIZiihl.forsclnmg, 9.) Borm: VGH-Wissenschaftsverlag. Honda, Isao. 2007. A comparative and historical study of demonstratives and plmal markers in Tamangic languages. In Bielmeier, Roland and Felix Haller, eds., Himalayan linguistics and beyond (frends in Linguistics, Studies and Monographs, 196.) Mouton de Gruyter, Berlin, New York. Hoshi, Michiyo and Tondup Tsering. 1978. Zangskar vocabulary. A dialect spoken in Kashmir. (Morrumenta Serindica, 5.) Tokyo: Institute for the Study of Languages and Cultures of Asia and Africa. Hu Tan 1989. Comparative sentences in Tibetan. Acta OrientaliaAcademiae Scientiarum Hungaricae 43.2-3: 399-408. Huber, Brigitte. 2005. The Tibetan dialect of Lende {Kyirong). (Beitriige zur tibetischen EIZiihlforsclnmg, 15.) BOlUl: VGH Wissenschaftsverlag. Imaeda, Yoshiro & al. 2007. Tibetan Documents from Dunhuang kept at the Bibliotheque Nationale de France and the British Library. (Old Tibetan Documents Online Monograph Series.) Tokyo: Research Institute for Languages and Cultures of Asia and Africa Tokyo University of Foreign Studies. http://otdo.aa.tufs.ac.jp. Jii.schke, Heinrich August. 1881. A Tibetan-English dictionary. With special reference to prevailing dialects. London 4th reprint 1992, Delhi: Motilal Banarsidass. Jacques, Guillaume. 2006. La chute du -s en tibetain de l'est. (XXemes JoUl'Jlf\es de Lingui.stique d'Asi.e Orientale, Paris 22-24 Juin 2006). http://xiang.free.fr/La chute du s.pdf. Jetlmar, Karl. 1961. Ethnological research in Dardistan 1958. Preliminary report. Proceedings ofthe American Philosophical Society 105.1: 79-97. Kitamura, Hajime, K.alsang Namgyal, K.alsang Lhawang, and Michiyo Hoshi. 1977. Glo skad A material of a dialect in 1he Nepal Himalayas. {MOlllliilenta Serindica, 3 .) Tokyo: Institute for the Study of Languages and Cultures of Asia and Africa. Koshal, Sanyukta. 1979. Ladakhi Grammar. Delhi etc.: Motilal Banarsidass. - - . 1990. The Ladakhi language and its regional perspectives. Acta OrientaliaAcademiae Scientiarum Hungariae 44: 13-22. Kretschmar, Monika. 1986. Erziihlungen und Dialekt der Drokpas aus Siidwest-Tihet (Beitriige zur tibetischen EIZiihlforsclnmg, 8.) St. Augustin: VGH Wissenschaftsverlag. - - . 1995. En:iihlungen und Dialekt aus Siidmustang. Vol. III: Untersuchung zur Grammatik des Siidmustang-Dialekts; Vol. IV: WOrterbuch zum Siidmustang-
300
BETIINA ZEISLER
Dialekt. (Beitriige zur tibetischen Erziihlforschung, 12.) St. Augustin: VGH Wissenschaftsverlag. Laufer, Berthold 1911. Btsumno bkahi thaJ]yig. Der Roman einer tihetischen Konigin. Tibetischer Textund Obersetzung. Leipzig: Harrassowitz. Norman, Rebecca. In preparation Dictionary of the language spoken by Ladakhis. Roerich, Georges de. 1933. The Tibetan dialect of Lahul. Journal of the Utusvati Himalayan Research Institute 3: 83-189. Rona-Tas, Andras. 1966. Tibeto-Mongolica. The Tibetan loanwords afMonguor and the development ofthe archaic Tibetan dialects. (Indo-Iranian Monographs, 7.) The
Hague etc.: Mouton. Samten G. Kannay. 1998. An open letter by Pho-brang Zhi-ba-'od. In Samten G. Kannay, The arrow and the spindle. Studies in history myths, rituals and beliefs in Tibet. Kathmandu: Mandala Book Point. Schmidt, Ruth Leila and Razwal Kohistani.. 2008.A grammar afthe Shina language of Indus Kohistan. In collaboration with Mohammad Mama Zarin (Beitriige zur Kennfnis siidasiatischer Sprachen mid Literaturen, 17.) Wiesbaden: Harrassowitz. Sharma, Devi Datt. 1989. Tribal languages of Himachal Pradesh Part I. [TibetoHimalayan dialects of Lahoul: Pattani., Tinani, Gahri, Tod] (Studies in TibetoHimalayanLanguages, 2.) Delhi: Mittal. - - . 1992. Tribal languages of Himachal Pradesh Part II. [Tibeto-Himalayan dialects ofSpiti, Kinnam and Pangi.: Spitian, Nyamkat, Chhitkuli, Kanasi. (Malani).] (Studies in Tibeto-Himalayan Languages, 2.) Delhi: Mittal. - - . 1998. Tribal languages ofLadakh I (Part one. A concise grammar and dictionary ofBrok-skatl). (Studies in Tibeto-Himalayan languages, 6.) New Delhi: Mittal. Shirai, Satoko. 1999. Chibettogo gendai Rasa hogenno 2 onsetsukarmi arawareru -bni.1sui.te (On 1he inter-syllabic -h- in Modern Lhasa Tibetan). Gengogaku Kenkyu (Linguistic Research) 17-18: 59-72. Shawe, F.B. 1894. On the relationship between Tibetan orthography and the original pronunciation of the language. Journal ofthe Asiatic Society ofBengal 63: 4-19. Simon, Walter. 1940. Certain Tibetan suffixes and their combinations. Harvard Journal ofAsiatic Studies 5.3/4: 372-391. - - . 1977. Alternation of final vowel with final dental nasal or plosi.ve in Tibetan Bulletin ofthe School ofOriental and African Studies 40.1: 51-57. Sprigg, Richard Keith. 2002. Balti-English En~ish-Balti Dictionary. London, N.Y.: RoutledgeCmzon. Sun, Jackson T.-S. l986.Aspectsofthephonology afAmdo Tibetan. Tokyo. - - . 2003. Phonological profile of Zhongu: a new Tibetan dialect of northern Sichuan. Language and Linguistics 4.4: 769-836. Suzuki, Hiroyuki, this volmne. Dialectal Particularities of Sogpho Tibetan -An Introduction to the "Twenty-fom villages' patois" Thomas, Frederick William. 1957. Ancient folk-literature from north-eastern Tibet. (AbharuD.mgen der Deutschen Akademi.e der Wissenschaften zu Berlin, Klasse fur Sprachen, Literatur und Kunst, Jahrgang 1952, Nr. 3.) Berlin: Akademi.e-Verlag. Toumadre, Nicolas. 1995. Tibetan ergativity and the trajectory model. Senri Ethnological Studies 41: 261-275. (Originally in: Hajime Kitamura, Ta1suo Nishida and Yasuhiko Nagano, eds., Current Issues in Sino-Tibetan Linguistics. Osaka: The Organising Committee, 637--648.) - - . 1996. L 'ergativite en tihetain. Approche morphosyntaxique de Ia langue parlee. (Bibliotheque de l'informationgrammaticale, 33.) Paris, Leuven: Peeters. Uebach, Helga and Bettina Zeisler. 2008. rJe hlas,pha los and other compounds with suffix-sin Old Tibetan texis. In Brigitte Huber, Mariarme Volkart mid Paul Wich:ner,
KENHAT, THE DIALECfS OF UPPER LADAKH AND ZANSKAR
301
Hgg., Chomolangma, Demawend und Kashek Festschriftfor Roland Bielmeier zu seinem 65. Gehurtstag, Band I: Chomolangma. Halle: International Institute for Tibetan and Buddhist Studies, 309-334. Uray, Geza. 1952. A Tibetan dimirrulive suffix. Acta Orientalia Scientamm Hungaricae 2: 182-220. - - . 1955. On the Tibetan letters ha and wa. Contribution to the origin and history of the Tibetan alphabet. Acta Orientalia ScientammHungaricae 5.1-2: 101-121. - - . 1972. Queen Sad-mar-kar's songs in the Old Tibetan Chronicle. Acta Orientalia Scientarum Hungaricae 25: 5-38. Zeisler, Bettina. 2004. Relative Tense and aspectual values in Tibetan languages. A comparative study. (frends in Linguistics, Studies and Monographs, 150.) Mouton de Gruyter, Berlin, New York. - - . 2006. The Tibetan understanding of /carman: Some problems of Tibetan case marking. In Christopher I. Beckwith, ed, Medieval Tibeto-Burman languages II, PlATS 2003: Tibetan studies: Proceedings of the Tenth Seminar of the International Association for Tibetan Studies, Oxford 2003. (Brills Tibetan Studies Ubrary, 10.) Brill, Leiden etc., 57-101. - - . 2007. Case patterns and pattern variation in Ladakhi: a field report. In Bie1meier, Roland and Felix Haller, eds., Himalayan linguistics and beyond (frends in Linguistics, Studies and Monographs, 196.) Mouton de Gruyter, Berlin, New York, 399-425. - - . 2009. Skaddi IJgyurcanaiJ rdohafzi mentogtsogsla luskanni S1JCJn!yug ha/Ji rdzessiskorla-Language change and the fossilization of 1he Old Tibetan h- prefix in Ladakhi and Balti. In Monisha Ahmed and John Bray, eds., Recent Research on Ladakh2009. Papers from the 12th colloquium ofthe lntemationa/Associalionfor Ladakh Studies, Kargil. Kargi1 & Leh: International Association of Ladakh Studies, 81-96. See also an earlier version (2005) under http://www.sfb44l.uni-tuebingen.de/bllllanguageArchaeology.pdf - . Forthcoming. Ethnic diversity, language contact, and the Old Tibetan lingua franca: early Tibetan history and the development the modern Tibetan dialects. (Asiatische Forsclnmgen.) Wiesbaden: Harrassowi1z. - - . In preparation West Tibetan verbal motphology and the reconstruction of the proto-language: the Sluifer hypothesis revisited Zhang, Yisun [K~ Dhyisun] et. al., eds.,l993. Bod-Rgya tshigmdzod chenmo (The large Tibetan Chinese dictionary). Vol. 1-2. Pecin: Mirigs dpeskrunkha!] (Beijing: Nationalities Publishing House).
of
INDEX
In order to facilitate easy identification in the index, the languages and dialects discussed in this volume are presented below in schematised family trees. Since most classifications continue to be disputed, we ask readers to note that the editors are not committed to any particular theory of genetic relationship and have offered this presentation for convenience only. AUSI'ROASIATIC
INDo-EUROPFAN
...... Proto-Austroasiatic ...... Munda NorthMunda SouthMunda ...... Khasi-Khmui.c Khasian Pakanic Khmuic ...... Mon-Khmer Khmeric
...... Indo-Iranian
Katuic
Balmaric Khmeric Pearle
Aslian Monic Nicobarese DRAVIDIAN
...... Northern Dmvidian Kurukh I Umon ...... Dhangar ...... Jhanger
Iranian
...... Eastern Iranian Northeastern Iranian ..... Pashto Indo-Aryan ...... Old Indic: Sanskrit ...... New Indo-Aryan Cen1ral Group ..... Hindi ..... Rajastani Gojri ..... Urdu
..... Lahori ..... Delhi dialect Dardic
..... Kashmiri ..... Kundal Shahi ..... Shina Brokskat Gilgit Guresi
Indus Kohistani Palula Phulwei (Shina) Eastern Group ..... Assamese ..... Bengali Northern Group ..... Nepali ..... Pah.ari, Parirni, Paarim North-Western Group ..... Lahnda Hindko ..... Panjabi ..... Sindhi
..... Si.raili Sou1hern Group ..... Maratlri
304
INDEX TIBErO-BURMAN I SINO-TlBKrAN
Sinitic
...... Chinese Old Chinese ..... Can1Dnese ..... Han Chinese ..... Mandarin ...... Bai Tani Tibeto-Bunnan ...... Brahmapulran Bodo-Koch Dhimalish Kacbinic ...... Easrem Tibeto-Bunnan Northeastern Tibeto-Bunnan ..... Qiangic Guiqiong nDrapa Qiang rGyalrong ..... Choyo (QueyU)
..... Ersfi
..... Ltisu Tosu ..... NamUyi ..... Primni (Pihni) ..... ShJxi:ng ..... Si1n-rGya1rong ..... Zhiiba Tangut Sou1heastern Tibeto-Bunnan ..... Lolo-Bunnese Akh.a Lisu Naxi Yi Western Tibeto-Bunnan/ Himalayish Bodie ..... Lepcha ..... Monpa Tibetan Proto-Tibetan Old Tibetan Oassical Tibetan Modern Tibetan ..... Sherpa ..... Amdo Tibetan Mdzorganrabar Rngaba
Themchen ..... Central Tibetan Lhasa Tibetan Dechen Tibetan
INDEX ...... Kh.ams I Kh.am Tibetan Northern Kh.ams ..... Batang I Bathang I mBathang ..... Budy
..... Chamdo ..... Dege I Derge ..... gDongsum ..... Kardze ..... Lhagang
..... Minyag Tibetan ..... Nangchen
..... Nyishe ..... Phrengme ..... Rangakha
..... rGyalthang ..... Sagong ..... Twenty-fom villages' patois Gezong, dGudzong (dGu-rdzong) Shuizi (Rwa-tso) Sogpho, Suopo (Sog-pho) Zhanggu (Rong-mi Brag- go) Zhonglu (sPro-smng)
Sou1hem Kh.ams ..... Dongwang West Tibetan (Balli & Ladakhi) Changthang dialec1B ..... Gyaik ..... Laga ..... Nyoma
..... Panggong ..... Shayok
Kenhat, Upper Ladakhi ..... Leh ..... Upper Indus dialec1s Cemre Gya, Gyahat (Gyaslrad)
Ronghat (Ro1JS/rad) Shara ..... Zanskari, Zllh.are 15au (Zanskari zlaho) Sham (GJam), Lower Zanskari Bot (Stod), Upper Zanskari
..... Hameling ..... Manda Zhung (GZu:J), Central Zanskari
..... Pipcha Sham skat ..... Balli Skardo ..... Nubra Aranu ..... Purik Ciktan
Dah (as second language) Dras (as second language) Garkhon (as second language) Kargil Tshangra
305
306
INDEX ..... Sham Eastern Sham ......Nurla Western Sham ...... Achinathang ...... Domkhar ...... Khala1se Western Tibetan ..... Dingri ..... Drok.pa ..... Kyirong ..... Lalrul ..... Mustang ..... Ngari, Ari (M:gal;lris) ..... Nyamkat Poo dialect ..... Spiti PinSpiti Tabo Spiti ..... Tot Zhongu ..... Tamangic Mahakiranti ..... Kiranti Proto-Kiranti Pre- Kiranti (Modem) Kiranti ..... Cen1ral Kiranti Ku1ung Sampang ..... Eastern Kiranti Lohorung Yamphu ..... Limbu Western Kiranti Bahing Chepang Dumi Hayu/Vayu Sunwari Thu1ung Wambu1e ..... Bararn-Thangmi Thangmi
Para-Kiranti ..... Newari I Nepal Bhasa Bhaktapur dialect Dolakha Newar Kathmandu/Patan dialec1s West Himalayan (Old Zhangzhung)
INDEX
307
SUBJECT INDEX ablative ablative case marking 281 ablative fimction 281, 282, 284, 288 ablative marker(s) 282, 283, 284, 285, 288; /-na/ 281; nas 283 ablative postposition 281 complex ablative 285, 288 1- ablative 288 locative-ablative marker *lnasul 288 locative-ablative marker /-ne/ 284 ablaut system of Old Tibetan 249 absolutive 91, 220, 289 absolutive/nominative 7 absolutive marking 7 Acehnese 226 Acheulian traditions 28, 29 Achinathang 264 actant(s) 6, 184, 186, 192, 200-205, 212 active aligmnent 226 active clauses 135 additional -s morpheme, morpheme *-su/*-so 282, 283, 284, 286, 287, 289 *-su/*-so for location adverbs 284 -SDIJ in equative comparisons 285 adjective(s) 5, 99-104, 106-124, 126128, 269, 275 adjectival(s) 5, 100, 128 adjectival concept(s), words 107, 108, 120, 125, 126 adjectival root, stem 115, 278 adjectival suffix-pu 112, 114 age 108, 110, 120, 122, 123 as a comparative construction 10 1 cardinal rrumbers 108, 120 class(es) (of) 5, 100, 101, 103, 107, 108,113,120,122,126,128 classifiers (of adjectives) 108, 109, 120 coding 108 co1om 108, 111, 112, 120, 121 difficulty 108, 116, 120-122, 126 dimension, dimensional adjective(s) 108-110, 115, 120, 122, 128 functioning as an intransitive predicate 100-102, 123 lrumanpropensity 108, 115, 120-122 in adverbial fimction 10 1 non-nom-like 101, 103 nom-like 101
physical property, properties 108, 114, 120-122 position 108, 119, 120 predicate 149 predicative encoding of 107 pre-nominal modification of 107 qualification 108,117,120,122 quantification 108,118,120 similarity 108,117,120123 speed 108, 116, 120 sub-classes of adjectives 100, 102, 120 adjectival noliDS 100 adjectival verbs 100, 120, 121, 124, 125 non-predicative 5, 103, 104, 111, 120, 123 non-verb-like 5, 101-103, 106, 122 se cwane construction 102, 107, 110, 112-116, 120, 122, 125-128 value 108, 112, 113, 120-122 verb-like 5, 101, 102-103, 119-122 125-128 administrative lexicon 90, 93 administrative practice 92 adnominal suffix 103, 107, 110, 111, 115, 117, 118, 120, 123, 124 adstrates 293 adverb(s) 116, 120, 155 directional adverbs 284 location adverbs 284 adverbial suffix -se 126 affixed consonants 45 affixed signs 46 agentive marking 229 agentive participant(s) 134, 135, 152, 156 agentive suffix -di-s 263 agentive rod 228 agent(s) 6, 7, 91, 142, 183, 190, 196, 197, 199,201, 204, 205, 212,240, 277,283-287,289 agent argument 287 agent/ cause, neutralisation of 286 agent-like argument(s) 7, 220 agent marker, marking of agent(s) 228, 281, 285-288, 289, 292 ergative agent marking 292 agent noun 228 agent pronom 228 animate agent 285
308
INDEX
dual agent 188-190,205 exclusive agent 196, 203 genitive agent marking 239, 240, 282, 285,292 lruman agent 287' 288 instrumental agent marker 240 non-agent 6 possessor-agent distinction 282, 285288 possessor-agent neutralisation (see neutralisation) agreement markers/ patterns 7, 220, 223, 223 agriculture 18, 242 of rice 3, 20, 21, 23 Ahom 13 Akha 133, 134 Alchi 240 Allahabad 21 alveolar flaps 110 Amdo (Tibetan) 3, 4, 134, 144, 245, 247,280 agricultural varieties of 64 nomadic varieties of 4, 55, 64, 68 analogous suffix 221 a-normative 88 ancient nominal derivation -di-n 275 anhnacy 103,156-158,163,164,166169,171-173,175,178 animacy as semantic parameters 175 anhnacy distinction 6, 133, 145, 156, 158, 160, 164-166, 168, 169, 171, 172, 177, 178 anhnacy markers, marking 6, 177 anhnacy ofO argumen1(s) 168, 175, 178 anhnacy of possessed argument(s) 159-162 anhnacy of referent 166, 175 anhnacy ofS argument(s) 156, 159, 165, 169, 178 anhnacy split 133, 156, 163, 169, 170, 172, 177, 178 Annals ofthe ~ zhaPrincipality 77, 79 aorist 226, 276 Aranu (dialect; ARA) 259,260, 266, 267,281,283 archaeology 2, 13, 17, 23, 24 Arjmt 28 Arjmt river 28 Anmachal Pradesh 26
Astian 30,32 Astian (dialect) 16 Assam 17,21 Assamese 14 Atbmuqam 220 Austroasiatic 1, 2, 3, 13, 15-17, 19, 20, 25, 27, 30-32, 183 auxiliazy, auxiliaries 6, 133, 135, 139, 142, 143, 144, 145, 146, 148-153, 156, 163, 166-168, 172, 178, 196, 211, 291, 292 animate 163, 166 causative 262 evidential 239, 240, 263 existential 168 inanimate 163, 167 inflected 211 periphrastic 169 verbal 236,237, 292 AwaPit 134 Babai formations 28 Badi village 56, 69 Bahing 187,210,212,213,215,216 Balmaric 16 Bai (Biii) 25, 175 Baima 72, 170 Balti (commmtity) 24, 32,237,292 Balti dialects 1, 3, 236, 241, 245, 246, 255, 259, 260, 262,266,267, 278, 283,284 Baltistan 24, 235, 240, 293 Bangladesh 23 Bimshan phase 18 Bashidimg 22 Bathang, Batang (mBathang) 59, 172174,245 Bawang 55 Bay of Bengal 3, 8, 13, 17,23 Bazgo, Basgo (Bahsgo) 240,241 Bengali 13, 14 Bhairahavii 28 Bhaktapur dialect 99 Bhojpur district 183 Bhutan 19, 134 bilingualism 170 bimorphemic case marking 281, 283, 287,288 bisyllabic elements 59 Black Mountain MOn.pa 26 Bodie 18, 133 Bodo-Koch 30
INDEX Bralunaputran 14 Bralunaputran basin 23, 30 Bralunapulran delta, flood plains 17, 23 Brakhu!i 28,29 Brokskat 283 Bronze Age 17, 18, 29 Buddha 85 Buddhism 81, 94, 293 Buddhist estates 87 Buddhist history of Tibet 93 Buddhist monks 81 Budy 69, 70, 71 Bunna 23,26,32 Bwzahom 18 Cantonese 13 case morpheme 281 bimorphemes, complex morphemes 282, 284, 288 case neulralisation 239, 281 Caucasian languages 226 causative/fuk/ 262,263 Cemre (dialect; CEM) 235, 236, 237, 238,243,244,246,249,251,260, 263,278,283,285,288,291,294 Cemre (village) 239 Central Asia 240, 293 Central Ladakh 241 Central Tibetan dialects 4, 8, 136, 156, 236,239,245,249,255,274 Chamdo 67, 172 Changdu 67 Changfrwng 235,241,243 Changfrwng dialects 236, 247, 256, 260, 269, 271, 275 Chepang 210,213,215 Chickasaw 226 Chileng 19 Chilas 220, 235 Chiling 240 Chinese 56, 254 Choyo (Queyll) 14 Chushot 240 Ciktan 266, 267, 291 circumflex sign 48 Classical Tibetan, CT 3, 135, 169,235, 244,249,250, 255-260,264,266,267, 268, 275, 276,277,280-286,294 closed syllable(s) 48, 49, 204, 258, 282 coda (position) 157, 238, 246, 247, 253, 257,264
309
comitative marking 285-287 complementary distribution 197, 204, 288 complex pronominalising 183 complex tense forms 290 conjunct consonants 45 conjunct/disjunct 5, 6, 133, 134 animate conjunct(s) 156, 169 c/d system(s) 5, 133-135, 137, 139, 140, 144, 145, 147, 150, 155, 156, 164,169,172,173,175,178 conjunct auxiliary(ies) 135, 143, 151, 167 future imperfective zi 151 past imperfective d¥1 153 present imperfective d¥1 151 past perfectiveji 151 conjunct copula 146 conjunct existential(s) 138, 157, 158, 159 conjunct sub-clause 168, 169 conjunct(s) 5, 100, 134-136, 138, 139148, 152, 155, 157, 159, 162, 163 167, 168 ' inanimate conjunct 156, 169 inanimate conjunct existential 158 copula(s) 106, 108, 146, 150 copula complement, complement position/slotofcopula 101,102, 104, 106, 109, 111-117, 119, 121, 123,124,126,128,163 emphatic assertion with adjective in complement position 126 stative copula complement 121 copula omission 108 copulare 149, 150 copula red 136-138 copula verb(s), also copular verb(s) 105, 111, 118, 135, 142, 146 copulayin 136 copula zi 163 copula/existential verbs 135 copular clauses 138 copular constructions 136, 150 finite copulative sentence 106 inchoative (form ofthe) copula 106, 114, 119, 124 negative form of copula 124, 125 noun copula 106, 117 coreferential 135, 139, 144, 147, 148, 150, 154, 162
310
INDEX
conservative dialects 3, 4, 7, 8, 24, 235, 236,292 consonant cluster(s) 47, 61, 64 cluster onset(s) 245, 246,266, 267 final cluster(s) 235, 238, 247, 260, 261,263 contact induced change 170
Dah 264 Daic 13 Dalai Lama 84, 173 fifth Dalai Lama 84, 93 Dali Kingdom 173 Danba Cmmty 4, 55 Diiilg basin 28, 29 Daocheng 144 Dardic 1, 293 dative dative A(s) 227, 230, 231 dative-allative marker Ia 282, 289 dative-allative marker with ablative and partitive function 282 dative/ergative alternation 220, 227, 230,231 dative-marked possessor 155, 163 dative marker, marking 91, 220, 227 dative S(s) 227, 230, 231 Dechen Autonomous County 144 Dechen Cmmty 144, 174 Dechen Tibetan 172, 174 declarative clause(s) 134, 136-139, 144, 146-148' 151-155, 157, 159 conjmct-declarative 136, 138, 140143, 146, 147, 151, 152, 158, 159, 163, 166 declarative/first person 135 disjmct-declarative 137-139, 141143, 148, 149, 153-155, 160, 161, 163, 167 definiteness marker(s) {-po},po, mo 275,277 {-de}, /del, de 276-278,279 -Xo, IJ.o 277, 279 Dege (comty) 172 Dege, Derge (dialect) 59, 134, 172, 173, 257,258 Delhi dialect 230 derivational suffix 5, 284 demonstrative pronoun(s) 277-280, 285 anaphoric functions of 277 referential functions of 277
dental fricative 259, 261,262, 276 dental nasal 270, 275 dental stop 270, 271, 273-275 deonlic modality 231 Derge (see Dege) Derong 144 Deukhu:n basin 28, 29 Devanagatr script 41, 42,99 devoiced (consonants) 251, 256, 257 Dhangar 13 Dharma 85 Dhimalish 14 dialect boundary, bomdaries 172,240, 241 dialectal variations 99, 292 Dialogue Between Two Brothers (see PT 1283) dice 5, 83-88 dimirrutive 279 Dingri 255,280 diphthongisation 248,249,260 Diqing Tibetan Autonomous Prefecture 144, 172, 173 directional adverbs mas andyas 284 directive marker 258, 259 disjmct(s) 5, 100, 119, 135, 137, 138, 14~1~148-1~1~1~1~1~
161,167 animate disjmct 156, 161 disjunct auxiliary, auxiliaries 143, 153, 154, 165 future imperfective re form 153 past imperfective dzi? 153 past perfective thi 153 present imperfective @? 153 disjmct copula 148 disjmct enviromnents 159 disjmct existential 158, 160, 161 disjmct first person attributive +AN 165 disjmct morpheme 172 disjmct third person attributive -AN 164 disjmct third person attributive +AN 164 inanimate disjunct 156 past disjmct 119 past disjmct suffix -Ia 121 past disjmct suffix -ta 121 distal pronom 168, 279 distance marker 291 disyllabic word(s) 58
311
INDEX divination 83, 84 Doab 20 Dodariver 241-243 Dolakha Newar 99 Dollur 221
Domkhar (dialect; DOM) 255,260,262267, 276, 277, 280, 281 Domkhar (village) 241, 264 Dongsong village 67 Dongwang river 145 Dongwang Tibetan 5, 6, 133, 134, 136, 144-150, 153, 155-159, 162, 163, 165, 166,169-175,177, 178
Dongwang valley 145 Dravidian 1, 13,17 Dras 264 rrrokpa 245,255,280 IhDiri 187,188,206,210-214,216 Durbuk (block) 256 Dzongka, Dzongkha 134,255 Early Palaeolithic 28 East India Company 17 eastern Himalayas 2 Eastern Iranian (Scythian) 1, 293 Eastern Kham(s) (area, region) 55, 174 Eastern Khams dialect 146, 172, 175 eastern Nepal 6, 29, 183 eastern Sham 266 eastern Siviiliks 28, 29 eastern Terai 13 eastern Tibet 18,245 Eastern Tibetan 3 Eastern Uttar Pradesh dialects 224 elicited clauses 145 empathy hierarchy 140, 158, 164 deictic criteria in 140 discomse motivated criteria in 140 evidential criteria in 140 emphasis 124, 281, 283 emphatic assertion 126 End of the Good Age and Tragedy ofthe Horse and Yak (see ITJ 731) English 2, 3, 13, 14, 26, 51, 100, 116, 137, 148, 274 epist.emic source 137, 138, 144, 149, 168 ergative 91,212, 219-222, 226,229, 230, 231, 284,285 ergative A 231 ergative clause(s) 220, 228, 231 ergative construction(s) 222,224, 228,229,
ergative marker, marking 7, 220,221, 222, 224,226, 227, 228, 229,259, 27~282,283,284,285,292
ergative marker -su 286 ergative NPs 230 ergative postposition SliiJ. 228, 229 ergative S 225, 226, 229,230, 231 ergative suffix /-an/ 221 fluid-S system 229 split ergative, ergativity 212, 219, 220, 221,222
Erg0ng 14 Ersil 14 Etlmic Corridor ofWest Siclruan 63, 71 ethnolinguistic(s) 19 prehistory of 13, 15, 26 evidentiality, evidential 5, 6, 137, 139, 140, 141, 144, 154-157
evidential markers, marking of evidentiality 5, 7, 58, 70, 137, 157, 238, 239,240,263,290,291
evidential morphemes 291 evidential strategy 141 existential(s) 150, 151, 155. 157, 162, 169 existential auxiliary, auxiliaries 168 existential clause(s) 155, 159 existential construction(s) 136, 150 existential morphemes 163, 175 non-existential clauses 6, 163 father tongu.e(s) 24, 25 father tongue hypothesis 24, 27, 31 FaSmn (Spaldwn, 'Padmn') 242, 243, 248 feminine agreement 226 final consonant(s) 48-50, 259 dental stop, -d 270,271,274 -1 246, 259, 262 nasal 270 -r 246, 259, 262, 263, 296 -s 238, 246, 247, 248, 259, 263, 269, 270, 273, 274, 276 261, 276 velar nasal 49, 50, 261 -1
velar nasal suffix <-lJ> 191 finite declarative clauses 144 folk wisdom 81 Frashian/Khujhaani 221 French 24
fricativisation of (former) clusters 237, 239,245,245,246,254,260,273, 274,292
312
INDEX
as intermediate state 246 blocked with palatalised velars 294, 295 of cluster onsets (see word-initial) of following nominaliser 260,261, 273,274 with labial radical 247 with (voiced) dental radical 246 with voiced labial radical 240, 246 with voiced velar radical 239, 246 with voiceless labial radical 245, 246 with voiceless velar radical 246 word-initial fricativisation 239, 245, 294-296 word-medial fricativisation 245,269, 294-296 fronting 248,249,259 frozen clusters 238, 264 ftmerary 93 ftmerary practices 80 ftmerary rites 90, 91 fimerary texts, documents 89-92 future participle /-!fas/ -/-!fes/ 262 Gadan: 28 Ganges 17, 20, 21,23 Gansu 18 Ganzi Prefecture 55 Garkhon 264, 265 Garo Hills 26 Garzha (Lalml) 241 gDongsum 67, 69, 70, 71 geographical bmmdary, boundaries 240, 243 geographical distribution of genetic markers 24 Kenhat dialects 239 major Tibeto-Bunnan subgroups 15 Tibetan dialects 63 Tibeto-Burman language commmrities 18 Tibeto-Burmanlanguage family 18 generalisation, over-generalisation 215, 258,261-263,266,286 genitive 240,274,277,281-283,285,287 genitive (case marking of) agent 238, 239,240,282,283,292 genitive marker, marking 103, 105, 111, 123, 124,278,281, 283-287 genitive-ergative 289 genetic(s) 23, 26, 29, 30,32
genetic affiliation 99 genetic inheritance 169 genetic markers 24 genetic signature oflanguage community 19 population genetic(s) 2, 13, 24, 29, 30 Georgian 226 German 281 Geshitsa 55, 56 Geshiza River 55 Gezong (dGu-rdzong), dGudzong 56 Gheelan 221 GicD:riniya 28 Gilgit 235 Gilgit (dialect of Shina) 283 glide(s) 4, 62,64 /j/ 57,58 -1 45 1-affix, lathyU 45,46 1-cluster 45 labial glides 280 palatal glide 4, 68 -r-, /r/ 46, 56 r-affix, rathyU 46 /w/ 57,58 -y- 43 y-affix, yathyU 46 glottal stop 43, 65, 247 Gojri 219 Graan 221 Chlbnvedel,AJbert 42,43,47,48,51 gSerpa 63,68 Guambiano 134 Guarani 226 Guge 284, 293 Guiqiong (Guiqi6ng) 14, 56, 69 Gujhaan 221 Gmesi Shina 219 Gya (dialect; GYA, GYS), Gyahat (Gyaskad) 235,236,237,238,240, 243-258, 260-267, 269-280,281,283, 285-291, 295 Gya (village) 239,241,243, 250 Gya-Miru (valley of) 239, 240, 241 Gyaik 270 Gya-Sasoma (village) 249 Gyantse Tibetan Autonomous Prefecture 144 Hameling (dialect; HML) 237,238, 243, 245,246, 247, 249, 251,267, 295
Hameling (village) 243
INDEX HanChina 55 Han (Him.) Chinese 25, 72 haplogroup(s) 24-26, 30, 31 Hayu 187,210,212,213,215 Hazara 228 hedges 139, 149 Remis, Remis Gompa 239, 245 Hemiidil 22 Hemya 239 Henan 23 Himalayan languages 134 flirKti 2,14,224,227-231 ffindko 7,219,222,227-231 historical linguistics 13, 32 Hmong-Mien 20, 23 Hoobinhian 29 Holocene 28 Hmli river 23 Ice Age 24, 25, 27,28 Igoo 239 imperative (stem IV) -s 259, 276 imperative suffix -s (suffix of stem IV) 258, 258, 259,263 inanimate object(s) 103, 158 inclusive plural of the first person 238, 276, 278, 280 India 1, 7, 13, 17, 30, 173,239 Indian-administered Kashmir 220 Indo-Aryanlanguage(s) 1, 2, 7, 24, 115, 219 Jndo-Bmmese borderlands 2, 15, 23,30 Indo-Europeanlanguage(s) 1, 2, 13-15, 29,275 Indus Kohis1ani 283 Indus river 240 inferential marker 291, 292 inferential future marker 1-kakl 291 inferentialpastmarker {tuk} 291 inflectional suffix(es) 104, 106, 121, 122 initial consonant(s) 43-46, 265 initial consonant cluster(s) 45, 46, 57, 63,235 labial plosive 57 labialised preaspiration 57, 64 loss of 8 (see also fricativisation) preasprration 4,57,62,64,246 prenasalisation 57, 61, 64,254,257 triple clusters 57, 62, 266, 268 initial palatal before front vowels 244 initial y- 43 irmovative dialects 3, 8, 235, 236, 292
313
instrumental 281-284 instrument argument(s) 283, 285, 286, 287 ablative used for 285 inanimate instrmnent 285, 286 instrumental construction 287 instrumental marker(s) 240, 282-286 instrumental-agentive (marker) 282, 285 intermediate dental stop 274 interrogative clause(s) 134, 136-139, 144, 146, 147, 149, 151-155, 157 conjunct-interrogative 136, 138, 147, 151, 152, 158, 159, 162, 166 disjunct-interrogative 136-138, 149, 160, 161 interrogative marker 146 interrogative/second person 135 intervocalic voicing 271, 262 intransitive clause(s) 156, 166, 168 intransitive predicate(s) 224, 226 intransitive simplex conjugation(s) 185 ITJ 731, End ofthe GoodAge and Tragedy ofthe Horse andYak 80, 89, 90 ITJ 740 83-88,93 ITJ 753 82
Jhangar 13 Jiantang town 70
Kachinic 14 Kaghan Valley 228 Kangding County 67, 68 Kardze 245 Kargi1 (dialect) 266,267 Kargi1 (town, district) 236, 237, 264 Karo/Rawa 134 Kashmir 1, 7,18,219,240 Kaslunir valley, Valley of Kashmir 220 Kasluniri 7, 219,221-227,229, 231,271 Kathmandu City 99 Kathmandu Valley 5, 99 Kathmandu/Patan dialects 99 Katuic 16 Kaya 240 Kenhat (Gyenskad; see also Upper Ladaklri) 7,8,235,236,239,240,243246,248,251,258, 260, 264, 274-276, 277,285-287,289,291,292,293 bimorphemic case markers 281, 283, 284 borrowing from Kenhat 287
314
INDEX
case neutralisation 281 definitenes marker {-de} 277,278 first person inclusive plmal 278 Kenhat substrate 248, 287 morphophonemics 238 loss of final-s, former suffixes 247, 248,258,260
preservation or traces of former suffixes 258,260 phonemic tone 251 sound changes 237 vocalisation of final-s 248 (see also fricativisation of former clusters) tonogenesis 251 Kere 239 Kesar epic 243, 277 Khalatse (dialect; KHAL) 251,262, 266, 277, 281 Kham(s) (area,province) 174 Kham(s) Tibetan 3, 4, 55, 56, 59, 64-68, 71,144,146,172,173,245,255
Kharu 239,250 Khasi 13 Khasian 16 Khasi-Rluinric 30,31 Khmeric 16 Khmui.c 16 Khotang district 183
Khri Lde srong btsan 77, 79, 81 Kiranti 2, 6, 7, 183, 184, 187,206,210-217 Kishtwar 241 Koldihawa 20 Kra-Dai 13-15 Kuilh1lqiao 22 Kulu 241
Kulung 7,187,210,211,213-217 Kunda! Shahi (dialect) 7, 219, 220-222 Kunda! Shahi (village, community) 220, 221
Kurukh 13 Kyrrong 245,254,255,25~278,279, 280,291 Kyunggyam 239
1- allative 288 Ladakh 235-237,242,243,250,255,293 Ladakhi 1, 3, 61, 68, 134,235,236, 243, 245,247-249,251,255,260,262, 264-266,274-277,285,288,292 Laga (LAG) 243, 256, 269, 270-274
Lahori dialect 230 Lahul 241, 245, 276 Lahmadewa (Lahuradeva) 21 Lakhota 226 Lamayuru 242
late Palaeolithic 28, 29 late Pleistocene 28 Latin America 134 legal cases 5, 83, 85, 87 legal lexicon 77, 88 legal texts 92 Leh (dialect; LEH) 237,238,239, 244249, 251,258,260-264,269,270-272, 277-279,281,285, 286, 288, 290-292 Leh (district) 236, 237 Leh (town) 235, 239-241, 264,266, 292 Leh influence, loans 238, 247, 250, 267, 271,274 Lende Tibetan 134 Lepcha 2, 3, 41-43, 45-49, 51 Levallois teclmology, traditions 28, 29 lexeme(s) 70, 108, 269 lexico-phonetically conservative 236 lexico-phonetically innovative 236 Lhagang 67-69, 71 Lhasa 174 Lhasa Tibetan 61, 63, 68, 134 Lhomi 134 Ujiang 173, 176 Uktse 239 Limbu 7,187,206,210-217 Lingshet 242 linguistic palaeontology 3, 13, 16, 17-19, Lisu 175 Lisus 144 Uthang 173, 174 locational clauses 155, 160 location-source distinction 288 locative (case marker, marking) 163, 281, 282-284, 288 locative fimction, meaning 281, 288 locative NP 155 locative particle Jo 177 locative postposition 281 locative-ablative 281, 283,284, 288 locative-pmposive marker 288 neutralisation oflocative and ablative markers 281,288 Lohorung 7,187,206,210-217 Lolo-Burmese 6, 25, 133, 175, 177 loss of final cluster -TJs, -gs 260, 261
INDEX Lower Ladakh 7, 235, 236, 242,293 Lower Ladakhi 277 lower Yangtze 22 Lumbini district 28 Lungnak (/..u1]111Jg 'Black Valley') 243, 261 Lfisu 14
magic 90 Mahagarha 20 Mahottan district 28, 29 Mainwaring, George Byres 42, 43, 47, 48,51 Miijiii.bii.ng 22 Miijiayao phase 18 Manda (dialect; MND) 237,238,243, 244, 246-249, 251 Manda (village) 243 Mandarin Chinese 13, 175 Manipur 26 Marathi 224 Markha valley 240, 241 Markham 173 Martselang 239 masculine agreement 7, 226 Matho 239 matrix clause 135, 139, 147, 148, 150, 154, 162 mBathang (see Bathang) Mdzorganrabar 245 medial deictic fimction 279 medial position 59, 237, 246, 247, 264 medium argument(s) 286,287,292 Mednyi Aleut 171 Meghiilaya 13 Mekong river, Lancang Jiang 174 Mesolithic 28, 29 mKhar-ro 18 mid Holocene 22 midclle Palaeolithic 28 midclle Yangtze 17, 19, 20, 22 migrating s- pre-radical 249 migration routes 26 military service 77 Ming dynasty 173 Minhe Mangghuer 133, 134 Minyag Tibetan 71 mirative marker 157 Miru(village) 239,250 Mizoram 26 Moerduo [dMu-rdo] 55
315
Mongoliclanguages 133,134 Mongols 173 Monic 16 Mon-Khmer 16, 30,31 mono-morphemic case marker 289 morphological change 7, 217, 250 morphological processes 100, 101 morphological template(s) 194,211, 213,214 morphosyntactic criteria, properties 5, 100, 108, 128 of adjectives 101, 102 ofnouns 105 ofverbs 104 morphosyntax 128 Movius line 29 Mu family 173, 174 multisyllabic words 50 Munda 17, 30, 31, 183 Munda substratum hypothesis 183 Murree Hills 227 Mustang 235,245,248,255,280,291 Muzaffarabad 219,220 Nil.mUyi 14 Nan zhao Kingdom 173 Nangchen 64 Nii.rii.yiiJ:}I river 28 nasalisation(~ 99,192,197,200,238 national census (2001) 99 Naxi 6, 72, 133,171,173-178 Naxi village(s) 171,174 nDrapa 68 Neelam district 219, 222 Neelam river 219 Neelam Valley 7, 219,220, 224, 227, 228,231 negative affix ma- 104, 113, 124, 125 Neolithic Period, Age 17-19,22, 23, 25, 28,29 Nepali 99, 110, 115, 117 neutralisation ofcase 281,285 of c/d distinction 148 oflocation-source distinction 281, 282,288 of possessor-agent distinction 288 ofstems 259 of the voiced-non-voiced opposition 238,258
316
INDEX
N ewar, Newari, Newaric, Nepal Bhii,ya, newa: bhize 1, 2, 5, 99-104, 106-108, 110-112, 114-118, 120, 123, 124, 126, 128, 133, 134 Ngari, Ari (Myabris) 255, 285 Nicobarese 30, 32 Nicobarese dialect 16 Niega 55 nominal suffixes -di-n 275 nominal suffixes -s 275 non-complement clause construc1ion 148
non-contracted combination ofnominaliser ba plus auxiliary yod 291 non-coreferential complement clauses 144 non-retroflex sounds 47 non-speaker-prominent future 291 non-tonal dialects (see tonal) normative pressmes 247,263 normative use 88, 93 North and South American languages 226 North Munda 16 northeastern India 2, 15, 21, 23, 26, 27,30 northern India 24 norlhemNubra 259,260,283,288,291 northern Pakistan 24, 220 northwest Yunnan 5, 170, 172-174 noun(s) 4, 90, 93, 101-106, 108, 111, 123,124,202,277,278,284
animate plmal noun head 103 animate singular noun head 103 class(es) 104, 178 clitics 105 complement construction of 116 compound noun 103 definiteness of 277,278 head noun 101, 103, 118, 120, 122, 165 noun modification 101-105, 110, 111, 117, 118, 123, 124 noun phrase 105, 158, 277, 279 noun stem 115, 221 plural suffixes 105 unmodified noun 284 verbal noun 263,291 Nu 175 Nubra 235, 236 Nubra dialects 259 Nmla 266 Nyamkat 244, 278, 280, 283, 285, 286 Nyishe 66, 67, 69, 71 Nyoma block 256
Nyoma (dialect; NYO) 237-239, 243245, 247, 248, 256-260, 270, 278 291 Old Chinese 19 Old Tibetan, OT 4, 7, 66, 70, 77, 85, 93, 236-238, 244, 245,247,249,252,255, 258-260, 263, 275-277,279,281,282, 284-287 docw:nents, somces, texts 77-83, 83-85, 87, 88-90,91,92,93,279,282, 284 imperial dynasty 293 lingua franca 287,293 superstrate 287
onset cluster reduction, onset simplification 246, 247 open syllable(s) 48, 49, 253, 258-260, 282,288
oral stops 257,275 orthographical sequence 47 orthography 46,48,82,177 English orthography 3 Lepcha orthography 41-43,45,47, 48,51
native orthography 42,48 Tibetan orthography 3, 41 over-generalised suffix -s 262 Padam 26 Pahwd,Parimi,Paarim 227 Pakistan 1,219,235,236,239,241,243 Pakistan-achninistered Kashmir 219, 220, 222
palaeoethnobotany 13, 17 palatal plosive(s) 67, 68 palatalisation 237, 244 palato-alveolar affiicate(s) 59, 67, 68 Palaungic 16 Palula dialect 283 Panggong 270 Panjabi 222, 228 Papua New Guinea 134 Parkachilk 241,242 particles 105 partitive function 282 Pashto 219 past tense (stem II) -s 259, 261,276 past tense egophoric marker 239 past tense suffix (-s) (suffix of stem II) 248,258-261,263,292
Patan 99, 100
INDEX patient(s) 6, 149, 183, 185, 186, 188192, 194-199, 201-206,212,214, 216 patientive 229 Patu 28 Pearic 16 Pend.zila 242 Pengroushan culture 22 Phe 243 phoneme(s) 50, 58, 67 phoneme inventory 58 phonemic opposition 66, 254 phonemic signs 56 phonemic tone 251,253,254 phonetic distinctions 253 phonetic pitch patterns 58 high falling pitch 58 high pitch 257, 258 high rising pitch 58 low falling pitch 58 low pitch 257, 258 low rising pitch 58 phonology 3, 7, 8, 20, 239 phonological system 55, 56 phonological transcription 42 Phrengme 70, 71 Plruktse 272, 273 Pin Spiti (dialect) 291 Pin valley 244 Pipcha (dialect; PIP) 237,238, 243,247, 259, 261, 263 Pipcha (village) 243 plural pronmm. 238, 280 polymorphemic pronouns 278 Pongding village 145, 146 Poo dialect 283 population density 28 population prehistory 2, 15, 25, 26 porkrrnwrneau 183,188,189,191 possessive clause(s) 155, 156, 158, 160, 163, 175 posterior suffix <-m> 193 post-final consonant 235 post-final-s 238, 247, 259 postposition(s) 281, 283-285, 288 post-radical(s) 235, 265 predicate position 101, 117 predicate slot 104, 108 preglottalisation 64 Pre-Kiranli 211,214 pre-radical(s) 235, 237, 245, 246, 249, 255-257, 264, 265
317
present tense agenli.ve suffix -di-s 263 prevocalic stem, form(s) 204, 206 Primni (see Pumi) procedural texts 145 pronoun me 228 Proto-Austroasiatic 19, 183 Proto-Kiranti verb(al), perspective 210, 211, 213,215 Proto-Tibetan 236 Proto-Tibeto-Burman 183, 215 proximate pronouns 279 psychopQ[Op 80,89-91 PT 1042 89-91 PT 1071 87 PT 1283, Dialogue Between Two Brothers 81 PT 1289 89, 90 Pumi (Pfurn"), Prinmi 14, 175, 177 Pune 28 pwiishment 77, 79, 83, 84, 86, 88 judicial/legal punishment 4, 88-90, 92,93 military punishment 87, 88 Purik (dialects) 134, 236,237,241, 262, 264, 266, 267,291 Purik (region) 235, 236, 240, 241
Qiang, Qiangic (Qiii.ng, Qiii.ngic) 14, 68, 72,175,177 Qing dynasty 173, 174 Qingmai village 70 Qureshi 220 question marker 258, 261 radical (root consonant) 235, 249, 265 b followed by -y- or -r- 265 /.z andy before back vowels 244 labial 247 sub-scribed radical -!- 265 voiced 246, 257 voiced dental 246 voiced labial 240, 246, 247 voiced velar 239,246 voiceless 245, 246 voiceless dental, t 246, 296 voiceless labial 246 voiceless non-aspirated 256 voiceless velar 245, 246 y 255 Rait 221 Rangakha 68, 70, 71 Rangdum Yiildo 242
318
INDEX
R.ato Kholii 28 referential distinction 140 reflexive simplex conjugations 185 relational markers 284 relative c1ause(s) 102, 103, 163, 165, 166, 178 relexification (process of) 25 remote deixis 276-279 fimctioning as definite article 277 reorganisation of verb stem fonnation 249 reported speech 134, 135, 139, 144, 147, 162 conjlDlct-reported speech 139, 147, 148, 162 disjunct-reported speech 140, 150, 162 reported speech clauses 146 Republic of China 144 Republic ofNepal 99 retroflex consonant(s) 47, 110 retroflex sowtd 47 Rewugou district 63 nJyalrong 55,72,276 nJyalthang 70-72, 134, 160, 172-174 ritual1exicon 4, 77, 84, 92, 93 ritual texts 88, 90 Rngaba 245 Roman alphabet, script 3, 41,42 Ronghat (Ro1JSkaJ), Rong dialect 236, 239, 247, 250, 270 Rubshu 236 Russian finite verb morphology 171 SIA argument(s) 135, 139, 140
Sattra 221 sacrifice, sacrificial (animal) 4, 80, 89,90 Sagong 69, 71 Sak1i 239 Sampang 6, 7183-188, 190, 191, 194196, 198, 201,202,204, 206, 210-217 Sampang affixal string 194, 195,211, 212, 216 Sampang copy morphemes <-u> (3P') 189-193, 196, 204, 206; <-!]> (ls') 193, 196 199, 200, 204, 206; <-m> (1pA') 200 Sa.mpang morphemes exhibiting accusative pattern <-imi> (3nsAS) 188, 195, 199, 200, 202205, 212; <-o> (3P/NPT) 212 exhibiting ergative pattern <-e> (1pPs) 195, 198-200, 212, 214; <-ii> (1sPS) 195, 198-200, 206,
212; <-ka> (ePS) 191, 195,203, 212, 214 expressing 1~2 scenario <-n> (1 ~2) 194~196,197,201,204,213
expresmg agent <-m> (12pA) 193, 194, 196, 199-201, 203, 204, 213; <-!]> (3nsA) 196, 199, 202, 204, 205,213 expressing exlcusiveness <-ka> (ePS) 191, 195,203, 212, 214; <-ka> (eA) 191, 196, 203, 212,214 expresing negative polarity <man-> (NG) 194, 196; <-no> (NG) 196,206 expressing number <-i> (23s) 187, 188, 195, 200-203; <-ici> (d) 187, 189, 190, 193, 195, 199-205, 213, 215; <-ini> (2p) 187, 194, 196, 199,200,202,203,216• <-imi> (3nsAS) 188, 195, 199, 200, 202-205, 212; <-ici> (3nsAP) 187, 191-194, 196,200202,215 expressing person<-!]> (ls) 188, 192, 195, 197,200, 206, 216; <-na> (2) 197, 202-204, 206 215 expressing reflexive voice (RFL) 184, 188, 195-197, 200, 202, 203, 205, 213 expressing tense, non-preterite <-e> (1sP/NPT) 199; <-o> (3P/NPT) 188, 190, 191, 198; <-o> (3P/NPT), allomorphy of: 197; preterite <-a> (PT) 188-191, 195, 197-199, 202,206, 214, 215 ~~zero morphemes (expressing mc1USiveness) <-0> (iPS) 191, 195, 203; <-0> (iA) 191, 196, 203 Sangha 85 Sankhong 134 Sanskrit 115 Sariii N ahar R.iii 21 Saspol (dialect; SAS) 240,286 Saspol, Saspola (village) 240, 269, 275 Siitpati: 28 scroll 83, 91 secondary neutralisation 285 Seda County 63 Seerba district 63 semi-final particle te 91 sequential discourse marker 279 Shagong village 69 Sham (Gsam, in Spiti) 244
INDEX Sham (Gsam, Lower Ladakh) 241, 251 Sham (Gsam, Lower Zanskari) 243 Sham dialects 236, 237,238, 242,244, 245,248,249,259,260,262,285 Sl:!a.rmkat (Gsamskad) 1, 7, 8, 236-238, 240-242, 246, 248, 252, 258-260,263, 264, 276-278,281, 285-293 Shangri-la County 5, 66, 144, 175, 177 Shangshu paraphrase 80, 87 Shara (dialect; SHA) 237, 238, 243-245, 249, 252,253,255-257,269-274,275, 278, 281, 283, 294 Shara (village) 239 Shayok (di.alec1; SYK) 243,256,269-272, 274 Shayok valley 243, 269 She 239,240,251 Sherpa 134 Shigatse 134 ShUna 219,271,283 ShUna ofPhulwei 219 Shlxing 14 Slruizi (Rwa-tso),Rwatso 56 Siclruan (Sichuan) 17, 174, 175 Siclruan (Sichuan) language 14 Sildciln 3,8,26,41 srrnulfix 184,186,188,193,195,197, 200,202,203,205,213 Sindhi 224 SincD:m1I 28 Sinji Nakka 221 Siraiki 224 Si1n-rGyalrong 56 Sivii.liks 28, 29 Skar chmtg Edict 77, 81 Skar chmtg Pillar Inscription 81,94 Skardo 266, 267 Snyemo (or Nyimo) 240, 241 Sogpho, Sogpho Tibetan, Suopo, (Sogpho), SoFJJho 4, 55-58,63-72,264 soldier(s) 26, 84, 87, 88, 174 somtd change 4, 56, 63, 68, 146, 237, 248, 249,259,262,273-275,279 Songpan Comtty 63 SoutnPl&a 3,16,22,28,31,219 Soutn Mmtda dialect 16 Southeast ~a 3, 16, 17, 22, 28, 29,31 Southern Khams dialect 5, 6, 133, 144, 145, 156, 170,172,173,175,177 southern Khams or Kham (region) 70, 71,145, 169, 171, 172
319
southwestern Bhutan 19 speaker-prominent fu1nre 291 special word forms 63 Sp[ti(dialect) 235,244,278,280,283,285 Spiti (region) 241 splitS system, 226 Standard Tibetan 133, 136, 145, 146, 150, 155, 156 stative 104, 106, 113, 114, 116-121, 124-126 stem(s) adjectival stem 115 bomtd stems 108, 109 nomt stem 115,221 stem-final syllable 121 stem-final vowel 121 verb stem 184, 194,206, 213, 214, 249, 250, 258, 259, 268, 270 stem I (present stem) 249, 258, 259,262,268 stem II or second (past) stem 68, 249,258-261,262,265,268 stem III (future stem) 265, 268 stem IV or fortn (imperative) stem 68,249,258,259,261,268 stem allomorph 204, 206 stem assimilation 238, 248, 249, 263 stem neutralisation 259 verb stem formation, reorganisation of 249 Stok 240, 245, 252 story telling 228 subdialects 56, 71 subordinate clause(s) 106, 148, 163, 168, 169 subordination 168, 178 substrate(s) 248, 286, 287, 293 subsystem(s) 169, 170, 171,178,196 suffixal slot(s) 187, 188, 194-198, 200204, 206, 213, 216 suffixal string 190, 206, 215 suppletive form 287 loss of 286 oftne agent marker 288 oftne genitive 285 Suru valley 241, 242 syllable bomtdary 50, 51,284 shift oftne syllable bomtdary 264 syntactic changes 171 syntax 4, 5, 7, 91, 163
320
INDEX
Tabo Spiti 244, 245, 247,248,251,254, 255,280,283,285,291 Tagong village 67
Taipingqi.ao village 56 Ta.ijar 220 tally sticks (khram) 91, 93 Tamangic 278 T~,Klmrpti 42,43,47,48,50,51 Tang Dynasty 173 Tangut 14 Tani 26 Tarshit 239 taxes 77 tense-aspect 142, 156 distinctions 150 inflections 113 suffixes 103 Terai 29 Thangrrri 206,210,213-216 Thautali 99 Themchen 134, 280 Thi.ktse 239 T~ung 187,211-216 9ot (Stod), Upper Zanskari 243 Tibetan (dialects) 1-8, 24, 47, 55, 61, 63, 66,68-72, 133-135, 137-140, 141, 1~1~14~1~1~1~1~1~
172, 174, 177, 178,246,247,251, 254,260,264,280,282,291,293 Tibetan Autonomous Region 67,133, 144 Tibetan Empire 8, 77, 92 Tibetan grammar, grammarians 169, 235, 257' 265 Tibetan varieties of Himachal Pradesh and Uttarakhand 235, 236, 244 Tibeto-Bunnan 1-3, 5-8, 13-15, 17-19, 24-27, 29, 30, 32, 99, 133, 134, 177, 183,216,275,27~ 293 Tiri 239,250 Tokuva 21 tonal 63, 251 character, features, properties 235, 252,256 dialects 256 non-tonal dialects 235 (phonemic) contrast, opposition 175,
177,254,256 phonetic distinction, difference 251-254 tone 56,63 emerging tone 292 high tone 59, 60, 61, 65,253-257,265
high level 56, 58, low tone 59, 60, 61,244,252-256,257 low level 56, 59 middle tone 257 neutadtone 252,257 pitch 48, 58, 251,252,254,257,258 register 8, 252, 254 tone contour 252, 254 falling 56,58 falling-riSUQg 251,252 level 56, 58, 59, 252,255 risUQg 56, 58 ri8UQg-fallUQg 56, 58 toneme(s) 58, 59 tonogenesis (in Tibetan) 251 topography 27 Tosu (Duoxil) 14 Tot 244, 278, 285, 286, 291 trade routes 240, 241 transcription 3, 41-43, 47, 48, 51 of final consonants 50 of initial consonant symbols 44 of initial glottal stop 43 ofvowels 49 transitive clause(s) 156, 163, 166, 167, 221,230 conjunct: transitive clause 167 transitive simplex conjugation 185, 186, 196 transliteration 5, 42, 43, 50, 51, 79, 82,
83,99 trisyllabic word 59 Tshangra 267 Tui basin, valley 28, 29 Tiijiii 25 Twenty-fom villages' patois 56, 63, 68, 71 univalent construction (in the complex agreement phenomena) 231 univalent predicate 226 unvoiced laryngal 245 Upper Indus (region, valley) 235,236,
239,243 Upper Indus dialects 244,249,250, 260, 281, 283, 288 Upper Indus Valley (Pakistan) 235 Upper Ladakh 7, 8, 236, 246, 255, 293 Upper Ladakhi dialects 8 Upshi 239,250,262 Uraon (see Kurukh) Urdu 2, 227-231
INDEX verb(s) 5, 7, 79, 85, 89, 90, 91, 99-101, 103, 104, 106, 107, 112-114, 116-121, 123-125, 127, 128, 135, 143, 146, 148, 157, 158, 177, 183, 196, 206, 210-213, 215, 227,249,250,255,258,259-263, 266, 267-269, 271, 275-277, 282, 288, 292 adjective-like verb(s) 125, 106, agen1ive verbs 7, 251,258-261,286 auxiliary verbs 135, 142, 178, 211 complementary verbs 290 control verbs 142,143,151,152,167 copular verbs 105, 111, 118, 135, 142,146 existential verbs 6, 126, 133, 135, 142, 144, 148, 155, 156, 165, 175, 178 animate existential verbs 161 inanimate existential verbs 161 irregular verbs 188, 20 1 intransitive verb(s) 6, 7, 183, 206, 220,225,227,229,251 nominalisedverb 143,165, 167,259 non-agentive verbs 251, 259-261, 285,286 non-control(lable) verb 143 non-periphrastic indicative verb fonn 183 preconsonantal verb fonn 206 prevocalic verb form 206 stative fonn(s) 116-119, 121, 124 transitive verb(s) 6, 168, 183, 206, 220,251,287 verb(al) agreement 6, 7, 100, 183, 184, 190, 191, 194, 198, 201,205, 206, 210, 211,212, 216, 217, 220224, 226, 229 verbal affixes 183, 211 verbal inflectional suffix 121, 122 class I 121 class II 121 class III 121, 122 class IV 121 class V 121 verbal suffix 114-117, 120-122, 127 verb-like adjectives (see adjective(s)) verbs with absolutive S 220 verbs with ergative S 224-227, 229 verbs with nominative S 227 Vietic 16 visual evidential 154, 155, 157
321
vocalisation 269,270,273-275 of (ancient) final sibilant-s 8 voiced(fiicative)l~ 244,245 vowel(s) 43, 48, 50, 56, 57, 58, 59, 121, 157, 186, 188, 192, 198,201, 214, 238, 248, 249, 252-254, 256, 257, 259, 261,284,294 alternation 249 apical (retroflex) vowel N 69 back vowels 244,247,248,259,274 deletion of 186 final (forms) 65, 249, 262-264, 266, 279 front vowels 244 'high' vowel 253 initial vowel/if 188 inventory 57 long, lengthy, lengthened vowel(s) 59,65,66,104,121 'low' vowel 253, 254, 255 nasalised vowel(s) 57, 59, 65, 66,238 nonnal vowels(s) 59, 66 plain vowel 65, 280 rmm.ded vowel 280 vowel change 248, 260, 269, 270 diphtongisation 248, 249, 260 fronting 248,249,259 merging 248, 249 vowel sign 43, 48, 49 diacritical vowel sign 48, 50 vowel slot 235 Wambule 206, 211-216 Wanla 242, 243 Weixi ColDl.ty 69, 144 Weixi LisuAutonomous County 145 West Himalayan (Old Zhangzlnmg) 293 West Tibetan (Balti and Ladakhi), WT 3,266 Western Archaic Tibetan 235 Western Innovative Tibetan 235 western Sham dialects 237, 259, 264, 266, 288,291 Western Tibetan dialects 255, 260, 283285,293 western Terai 28, 29 Written Tibetan, Wff 63-71, 135, 146, 157 Xiangcheng County 67, 69, 70, 144 Xianggelila County 66, 70 Xinduqiao town 68 Xingl011ggou 19
322
INDEX
Yamphu 206, 211-216
Yang1ze basin 22, 23 Yang1ze delta 17, 22,23 Yellow river 17, 19, 25 Yi (Lolo) 173 Yulchmtg-Nyeraks 242,243 Yulong Naxi Autonomous County 144, 175 YU11l1a11 (YUnruin) 25, 66, 134, 144, 169, 174 YUl1l1a11 (YUnruin) dialect 14 Zangla 243 Zanskar river 240-243 Zanskar (valley) 235,236,240, 241-243 Zanskari, Ziihare l~au (Zanskari zlaho), Zanskar dialects 235,236,240, 241,
243, 247' 248, 251' 254, 260, 276, 278,295 zero allomorph(s) 6, 184, 188, 190, 197, Zhaba (Zhaba) dialect 14 Zhaba district 68 Zhanggu (ft.ang-mi Brag- 'go) 56 Zhangzhmtg 293 Zhongdian 168, 174 Zhongdian dialect 172 Zhonglu (sPro-snang) 56 Zhongu 63, 66-68, 71, 72 Zhol Pillar Inscription 78 Zlnmg (GZw:J), Central Zanskari 243, 246 Zhwa'i Lha khang Inscriptions 77-79