Citation
Austroasiatic languages

Material Information

Title:
Austroasiatic languages essays in honour of H.L. Shorto
Series Title:
Collected papers in Oriental and African studies
Creator:
Davidson, Jeremy H. C. S
Shorto, Harry L
University of London -- School of Oriental and African Studies
Place of Publication:
London
Publisher:
School of Oriental and African Studies
Publication Date:
Language:
English
Physical Description:
viii, 249 p. : ;

Subjects

Subjects / Keywords:
Austroasiatic languages ( lcsh )
Genre:
festschrift ( marcgt )
Spatial Coverage:
Asia -- Southeast Asia

Notes

Bibliography:
Includes bibliographies.
Statement of Responsibility:
edited by Jeremy H.C.S. Davidson.

Record Information

Source Institution:
SOAS, University of London
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
129342 ( ALEPH )
0-7286-0183-4 ( ISBN )
Classification:
GM410 ( ddc )

Downloads

This item has the following downloads:


Full Text


AUSTROASIATIC LANGUAGES

Essays in honour of
H.L. Shorto



Edited by J.H.C.S. Davidson















School of Oriental and African Studies
University of London
1991




Collected Papers in Oriental and African Studies

AUSTRO ASIATIC LANGUAGES
Essays in honour of H. L. Shorto

Edited by

Jeremy H. C. S. Davidson
Formerly Lecturer in Vietnamese
School of Oriental and African Studies

SCHOOL OF ORIENTAL AND AFRICAN STUDIES
UNIVERSITY OF LONDON
1991


School of Oriental and African Studies, University of London 1991
All rights reserved

Published by

School of Oriental and African Studies

University of London

Thornhaugh Street

Russell Square

London WC1H OXG

British Library Cataloguing in Publication Data

Austroasiatic languages: essays in honour of H. L. Shorto
- (Collected papers in Oriental and African Studies).
I. Davidson, Jeremy H. C. S. (Jeremy Hugh Chauncy Shane)
IT. Shorto, H. L.
TT. Series 495

ISBN 0-7286-0183-4

Typeset by PDQ Typesetting, Stoke-on-Trent, England.
Printed in England by Hobbs the Printers Ltd., Southampton


CONTENTS

CONTRIBUTORS...................................................................................vi

PREFACE.............................................................................................vii

ACKNOWLEDGEMENTS...................................................................... viii

H. L. SHORTO: A BIOGRAPHICAL NOTE J. H. C. S. Davidson......... .............1

PUBLICATIONS OF H. L. SHORTO Helen Cordell.........................................3

AUSTRIC: AN 'EXTINCT' PROTO-LANGUAGE Paul K. Benedict....................7

PALAUNGIC VOWELS IN MON-KHMER PERSPECTIVE G. Diffloth............. 13

COMMUNICATEES, EXISTIVES, AND STATIVES IN
PROTO-SOUTH-BAHNARIC David Thomas.............................................29

A LINGUISTIC ANALYSIS OF SOME SOUTH MUNDA
KINSHIP TERMS, I Norman H. Zide & Arlene R. K. Zide............................43

PROBLEMS AND PITFALLS IN THE PHONETIC INTERPRETATION
OF KHASI ORTHOGRAPHY Eugenie J. A. Henderson..............................61

HU A LANGUAGE WITH UNORTHODOX
TONOGENESIS Jan-Olof Svantesson....................................................... 67

ON AUSTRONESIAN LEXICON IN VIETNAMESE Kenneth Gregerson............ 81

SEVENTEENTH-CENTURY VIETNAMESE LEXICON: PRELIMINARY
GLEANINGS FROM ALEXANDRE DE RHODES'

WRITINGS Nguyn -Dinh-Ho............................................................... 95

THE PHONOLOGY OF KOMPONG THOM CHAM Robert K. Headley...........105

ASPECTS OF INTER-CLAUSAL RELATIONS IN
KHMU Suwilai Premsrirat...................................................................123

AN INSTRUMENTAL STUDY OF CHONG
REGISTERS Theraphan L. Thongkum....................................................141

KEEPING THINGS UP FRONT: ASPECTS OF INFORMATION
PROCESSING IN MAL DISCOURSE STRUCTURE David Filbeck..............161

LES DERIVES DSIDRATIFS EN KHMER Saveros Pou..........................183

A DIACHRONIC SURVEY OF SOME KHMER PARTICLES
(7th to 17th CENTURIES) Judith M. Jacob..............................................193

THE FORM syah IN ANGKORIAN KHMER Philip N. Jenner.......................227

OLD MON s- Christian Bauer.................................................................241

Frontispiece Professor H. L. Shorto

V


CONTRIBUTORS

J. H. C. S. Davidson, formerly Lecturer in Vietnamese, School of
Oriental and African Studies, University of London

Helen Cordell, Sub-Librarian, School of Oriental and African Studies,
University of London

P. K. Benedict, Adjunct Professor of Anthropology, Arizona State
University, Tempe, Arizona, USA

G. Diffloth, Professor, Modern Languages, Cornell University, Ithaca,
New York, USA

D. Thomas, Summer Institute of Linguistics, International Linguistics
Center, Dallas, Texas, USA

N. H. Zide, Professor, Department of South Asian Languages and
Civilizations, University of Chicago, Chicago, Illinois, USA

Arlene R. K. Zide, Department of Foreign Languages, Harold Walsh
College, Chicago, Illinois, USA

|Eugnie J. A. Henderson, Professor Emeritus of Phonetics in the
University of London

J.-O. Svantesson, Professor, Department of Linguistics and Phonetics,
University of Lund, Lund, Sweden

K. J. Gregerson, President, Summer Institute of Lingiustics, Interna-
tional Linguistics Center, Dallas, Texas, USA

Nguyen Dinh-Ho, Professor, Vietnamese Studies Program, San Jose
State University, San Jose, California, USA

R. K. Headley, Senior Research Linguist, US Government, Washington,
DC, USA

Suwilai Premsrirat, Institute of Language and Culture for Rural
Development, Mahidol University, Salaya, Nakhorn Pathom, Thailand

Theraphan L. Thongkum, Department of Linguistics, Chulalongkorn
University, Bangkok, Thailand

D. Filbeck, Christian Mission to the Orient, Chiang Mai, Thailand

Saveros Pou, Directeur de Recherche, CNRS, Paris, France

Judith M. Jacob, formerly Senior Lecturer in Cambodian, School of
Oriental and African Studies, University of London (rtd.)

P. N. Jenner, Director, Department of Indo-Pacific Languages,
University of Hawaii, Honolulu, Hawaii, USA

C. Bauer, Lecturer in Linguistics and Mon, Institute of Language and
Culture for Rural Development, Mahidol University, Salaya, Nakhorn
Pathom, Thailand

vi


PREFACE

The present volume is the fourth in this series to be dedicated to a
retired scholar of the School. Here we honour Professor H. L. Shorto,
whose standing in the field of Austroasiaticespecially Mon-Khmer
Studies is well reflected in the range and quality of the papers offered as a
tribute in this volume.

The arrangement is, I hope, obviousopening with a questioning
paper by Paul Benedict, then moving from the languages of the rims of
the area to focus on the Mon-Khmer heartland of Harry's major interest.
The papers do not call for any comment from me as Editor, nor do their
authors need introduction. Their contributions speak for themselves and
will undoubtedly stimulate further research in the field of Austroasiatic
Studies in coming years.

Jeremy H. C. S. DAVIDSON

vii


ACKNOWLEDGEMENTS

I should like to thank all those who have helped make the task of
editing these seventeen contributions lighter. In particular I wish to thank
the Publications Committee (especially its Chairman, Professor Shackle,
and its Secretary, Mr Martin Daly) for financial support in this
publication of the volume. Thanks are due to the Editorial Secretary,
Miss Diana Matias and to the SOAS photographer, Mr Paul Fox, for
help with its production.

The photograph on the cover is by Hans Hinz and is reproduced by permission from Thai
painting published by the Office du Livre, Fribourg.

viii


HARRY LEONARD SHORTO: A BIOGRAPHICAL NOTE

Professor Harry Shorto is one of the founding scholars of Mon-Khmer
studies, the field which his knowledge and insight helped to make the
active and expanding area of scholarship that it is today.

It is now some forty five years since Harry Shorto completed his MA in
Modern and Mediaeval Languages at Cambridge before moving to
SOAS to take up a training appointment as Lecturer in Linguistics in
1948. From that point he was to pursue the study of the Austroasiatic
and Austronesian language families of his choice: in 1952 he was
appointed Lecturer in Mon; in 1964 Reader in the Languages and
Literatures of South-East Asia; in 1971 to the Chair of Mon-Khmer
Studies in the University of London. During this period he produced his
two magisterial dictionaries, A dictionary of Modern Spoken Mon (1962)
and A dictionary of the Mon inscriptions from the sixth to the sixteenth
centuries (1971), to which he added numerous other stimulating
publications listed in the bibliography of this volume.

Harry Shorto is a private man, given to weighing his words carefully
between frequent re-lightings of his ever-present pipe. His Common
Room colleagues may often have found awesome the ease with which he
negotiated the depths and further reaches of his unfamiliar linguistic
territory, but the learning was always leavened with humour, the delight
in language for its own sake infectious.

The lightness of touch and passion for detail which Harry brought to
the investigation of languages, their histories, constructions, and
interconnections, are in complete harmony with his other absorbing
interest: early musical instruments, in particular the flute, gamba, cornet
and virginal. This interest extends to performance and he was for a time a
member of a consort of viols. Today he continues to perfect his technique
on his 1960 Morley virginal.

To Harry Shorto's many colleagues and former students it must be a
source of great satisfaction that retirement has not put an end to his
encyclopaedic researches, notably his project for a comparative
phonology of the Mon-Khmer languages. We wish him well and offer
this volume in appreciation of his contribution to scholarship over the
years.

Jeremy H. C. S. DAVIDSON




PUBLICATIONS OF H. L. SHORTO
Helen Cordell

1956 Notes on Mon epigraphy. Bulletin of the School of Oriental and
African Studies 18(2), 344-52.

1956-57 Quantification in Mon. Proceedings of the 23rd International
Congress of Orientalists, Cambridge, 1954. London: Royal
Asiatic Society. 278-79.

1958 The Kyaikmaraw inscriptions. Bulletin of the School of Oriental
and African Studies 21(2), 361-67.

1960 Word and syllable patterns in Palaung. Bulletin of the School of
Oriental and African Studies 23(3), 544-57.

1961 A Mon geneaology of kings: observations on the Nidana
Arambhakhatha. In, Historians of South East Asia, (ed.)
D. G. E. Hall. London: Oxford University Press, 63-72.

1962 A Dictionary of Modern Spoken Mon. London: Oxford
University Press, xvi, 280p.

1963 Bibliographies of Mon-Khmer and Tai linguistics, compiled by H.
L. Shorto, Judith M. Jacob and E. H. S. Simmonds. London:
Oxford University Press. (London oriental bibliographies 2) x,
87pp.

Linguistic comparison in South East Asia and the Pacific, (ed.) H.
L. Shorto. London: School of Oriental and African Studies.
(Collected papers in Oriental and African studies 4).

The structural patterns of northern Mon-Khmer languages. In,
Linguistic comparison in South East Asia and the Pacific, (ed.) H.
L. Shorto. London: School of Oriental and African Studies.
(Collected papers in Oriental and African studies 4), 45-61.

The 32 myos in the medieval Mon kingdom. Bulletin of the
School of Oriental and African Studies 26(3), 572-91.

1965 The interpretation of archaic writing systems, illustrated by the
analysis of the phonological systems in early Mon dialects. In,
Indo-Pacific linguistic studies pt. 7, (eds.) G. B. Milner and E. J.
A. Henderson. Amsterdam: North-Holland. 88-97.

1966 Mon vowel systems: a problem in phonological statement. In, In
memory of J. R. Firth, (eds.) C. E. Bazell et al. London:
Longmans, 398-409.

3


The devata plaques of the Ananda basement In, Essays offered to
G. H. Luce by his colleagues and friends in honour of his seventy-
fifth birthday, (eds.) Ba Shin et al.. Ascona: Artibus Asiae. 2,
156-65.

1967 Die Register-unterschiede in den Mon-Khmer Sprachen.

Wissenschaftliche Zeitschrift der Karl-Marx-Universitat Leip-
zig, Gesellschafts- und Sprachwissenschaftliche Reihe, 16 (1/2),
245-48.

The dewatau sotapan: a Mon prototype of the 37 nats. Bulletin of
the School of Oriental and African Studies 30(1), 127-41.

1969 Mon labial clusters. Bulletin of the School of Oriental and African
Studies, 32 (1), 104-14

1970 The Gavampati tradition in Burma. In, R. C. Majumdar
felicitation volume, (ed.) H. B. Sarkar. Calcutta: K. L.
Mukhopadhyay, 15-30.

1971 A dictionary of the Mon inscriptions from the sixth to the sixteenth
centuries, incorporating materials collected by the late C. O.
Blagden. London: Oxford University Press. (London oriental
series; 24), xlii, 406.

1972 The word for two in Austroasiatic. In, Langages et techniques,
nature et socit, (eds.) J. Barrau et al. Paris: Klincksieck. 1, 233-
35.

1973 Three Mon-Khmer word families. Bulletin of the School of
Oriental and African Studies 36(2), 374-81.

1975 Achinese and mainland Austronesian. Bulletin of the School of
Oriental and African Studies 38(1), 81-102.

1976 In defense of Austric. Computational Analyses of Asian and
African Languages 6, 96-104.

Gayo consonant correspondences. In, South-East Asian linguis-
tic studies 2. (ed.) Nguyen Dang Liem. Canberra: Australian
National University. (Pacific linguistics. Series C, 42), 199-217.

The vocalism of proto-Mon-Khmer. In, Austroasiatic studies 2
(eds.) P. N. Jenner, L. C Thompson, S. Starosta. Honolulu:
University Press of Hawaii, 1041-67.

1977 Proto-Austronesian *taqdn: an anomaly removed. Bulletin of the
School of Oriental and African Studies 40(1), 128-9.

4


1978 The planets, the days of the week and the points of the compass:
orientation symbolism in 'Burma'. In, Natural symbols of South
East Asia, (ed.) G. B. Milner. London: School of Oriental and
African Studies (Collected papers in Oriental and African
studies), 152-64.

1979 The linguistic protohistory of Mainland South East Asia. In,
Early South East Asia, (eds.) R. B. Smith and W. Watson.
London: Oxford University Press, 273-78.

1982 The affinities of Kuy. Bulletin of the School of Oriental and
African Studies 45(3), 574-76.

1982 Obituary, Professor Emeritus Walter Simon. Bulletin of the
School of Oriental and African Studies 42(2), 344.

1990 Professor Eugnie J. A. Henderson: a personal note. Bulletin of
the School of Oriental and African Studies 53(3), 501.




AUSTRIC: AN 'EXTINCT' PROTO-LANGUAGE

Paul K. Benedict

A leading Mon-Khmerist (Diffloth 1985) has recently raised the question:
what happened to Austric? One can dilly-dally over this at great length
but the answer is very brief: it became extinct.

The Austric hypothesis was developed shortly after the turn of the
century by Wilhelm Schmidt (1906), a pioneer in the field of comparative
Mon-Khmer linguistics. He had noted a general similarity in morphol-
ogy, with even some resemblances in prefixes and infixes, between
Austronesian (AN) and Mon-Khmer (MK), the latter grouped with the
Munda languages of India to make up the Austroasiatic (AA) stock. True,
he had found nothing in MK to compare with the elaborate suffixial
apparatus of AN but, not surprisingly, he skipped lightly over this
discrepancy. And what of the anticipated common stock of 'core' (basic)
lexical elements? Linguists of that period, and well up into the middle of
this century, laboured under the mistaken notion that lexical elements are
pretty much the last place to look in setting up linguistic relationships;
furthermore, in those pre-Swadesh days, that even when one does get
around to inspecting these elements he need only apply an aphorism that
Gertrude Stein might have expounded: a word is a word is a word. Henri
Maspero, perhaps linguistically the most sophisticated of all the French
sinologists of the first half of the century, predictably followed along these
lines of thought in wrenching Vietnamese out of its native MK setting and
misidentifying it as kindred to the Tai languages. It would have been
equally predictable, given the circumstances, that a linguist such as
Schmidt would seek to establish a genetic linkage between AN and AA.
Thus it can be said that the birth of 'Austric' was expected; rather less
expected has been a certain continuing enchantment with it on the part of
some scholars, a reluctance to accept its demise. A review of the matter, as
attempted in this paper, is thus in order.

The writer, when not busy with actual (Sino-Tibetan [ST], Austro-Tai
[AT]) rather than fantasied language stocks, has from time to time given
some consideration to the basic question here. He at first (Benedict 1942)
en passant expressed a willingness to go along with Schmidt's hypothesis
on the assumption that continued research in that still largely unexplored
field would turn up a respectable body of AN/MK cognate sets for 'core'
lexical elements, thereby 'fleshing out' Austric, so to speak. Much later,
after a sizeable amount of data in the field had become available, he read
a paper on the subject at the First International Conference on
Austroasiatic Linguistics at Honolulu (Benedict 1973). Here he played
the 'devil's advocate' role, first setting up as good a case as possible for

7


PAUL K. BENEDICT

Austric, tying in both phonological and lexical evidence, before finally
and with some reluctancereaching the conclusion:

AT and AA [the earlier levels of AN and MK] do not have a core
vocabulary in common, despite the morphological similarity of the two
language stocks, hence the idea of an 'Austric' superstock must be
abandoned.

The homelands of both stocks must be assigned to adjacent areas in
South-East Asia (SEA) and the similarities in overall patterning can be
ascribed to areal factors. This corner of the Asiatic mainland has, in fact,
over the past decades acquired a certain fame among linguists for the vast
scope of its areal influences, involving even highly specific processes such
as vocalic transfer (Benedict 1979), and there is no reason to suppose that
similar influences were not also operative at an early period.

In the paper presented in 1973 the writer went rather further in
suggesting that a handful of lexical similarities in basic vocabulary
between AT and AA are indicative of an early relationship of substratum
type between the two stocks. It now seems evident that he overplayed his
'devil's advocate' role in this respect, paying insufficient attention to the
alternative explanation in terms of 'look-alikes' or 'comparabilia'
(Matisoff 1976), along with areal influences or 'border phenomena'.
Thus, the classical pair: PAN *mata (as earlier reconstructed); PMK *mat
'eye' are attention-getting, to be sure, but the syllable reduction is
bascially unmotivated (di- as well as mono-syllabic roots occur in MK);
PAN *maCa (with *C a cover symbol for reflexes representing proto-level
palatal or consonant cluster), as now reconstructed by Austronesianists,
weakens the comparison (final *-c is a prominent feature of PMK) while
PAT *mapra (for the earlier cited *mapla; see Benedict 1990) makes
matters even worse: finally, PAA *mst, the likely reconstruction on the
basis of the Munda cognates (see Pinnow 1959), takes us even further
away from PAT *mapra at the early time level at which these proto-
comparisons must be made. We can, in fact, do much better by
comparing the PST root for 'eye': *mysk (as now reconstructed; see
Benedict 1976a), with final *-k > *-t shift after the medial *-y- (lacking in
AA), even obviating the problem of syllabic reduction!

The above is an excellent illustration of the useful linguistic rule of
thumb: 'look-alikes' look less and less alike as we attain more precise
reconstructions at earlier and earlier levels. Thus, the suggested 'twin/two'
set of the Benedict (1973) paper, labelled as 'doubtful' even by Shorto
(1976), must now be discarded in the light of recent MK evidence re the
medial vocalism1 while the discussion there of the suggested set for 'dog'
overlooked Monic final *-r, which vitiates the comparison. Both
comparisons are also burdened with the problem of unmotivated syllabic

1. The recently reported Lai language of Southern China (GungxT) has /bi'/ for 'two',
supporting a PMK reconstruction of *9biar or *9biaar type, in line with Shorto's suggested

*bi9aar > ?bia(a)r rather than the *9baar cited in Benedict (1973). See Benedict
forthcoming, n.6.

8


Austric: an 'extinct' proto-language

reduction, found also in 'eye' (above) as well as in a fourth set: PAN
*busuk; PMK *sook 'hair'.2 A fifth, often cited set: PAN *ikan; PMK *ka
(or *kan ) Tish' has the same syllabic discrepancy along with 'problems' in
both final and initial (the Munda languages point to *q- as the PAA
initial, as reported in Benedict 1973). These difficulties led the writer to
reject the comparison in 1973 and the present reconstruction and analysis
of the PAN form (Benedict 1990) supports that decision: *sikan (for the
cited *isikan), from *si-ka-n, an archaic derivative (> Japanese ika
'squid') of the ubiquitous AT 'core' root: *ka/kan 'eat'.

It is entirely possible, of course, that the borrowing of one or more
basic roots such as 'hair' (above) will eventually be assignable to the AT/
AA 'border phenomena'. If so, the syllabic reduction factor requires that
one set up the direction of the loan(s) as AT > AA. This direction is also
required for the early AT > AA culture-word loans, notable those for
'copper' (> Munda group 'iron') and 'sugarcane', as pointed out in the
1973 paper. The Aslian languages maintain an archaic final *-s (generally
> -h~ 0 in non-Formosan AN languages) in the latter loan and an
extensive corpus of Aslian material of this kind recently collected by
Geoffrey Benjamin (pers. comm.) includes what appear to be parallel
examples of early loans that preserve other archaic phonological features,
including even a final *-l (maintained as a lateral in Formosan languages
only). The analysis of this material is still in an early phase and it is
unclear how many, if any, of these loans can be established as roots at the
PMK (possibly even PAA) level; in any event, they supply additional
evidence for the priority of the AT-speaking peoples in the early
emergence of 'high culture' in South-East Asia (Benedict 1975).

At the 1976 Toronto symposium on AT, Harry Shorto made a spirited
defence of Austric, prompting a 'Comment' by the writer (1976b). The
argumentation there need not be repeated in this paper; much of it
concerns lack of agreement re lexical items. Shorto proposed a number of
new AN/MK cognate sets, falling for the most part in what the writer
termed the ketketbongbong class, involving lexical areas well outside 'core'
vocabulary, with a striking attenuation or gap in this key aspect. As
summed up by Benedict (1976b: 106):

...the point to be stressed is that a score of ketketbongbong do not one
language stock make, since correspondences of this kind have significant
value for comparative purposes only when found in association with a corpus
of core vocabulary items.

2. The Formosan evidence, strongly supported by the Japanese (see Benedict 1990), points to
an underlying trisyllabic root: *busukas or *bosokas (PA medial *o and *u were merged in
PAN *u), which yielded Malayo-Polynesian forms of *busuk > buhuk type. The mid or
mid-low height of the PMK vowelthe cited form reflects the Proto-Semai vocalismthus
can readily be explained as a retention rather than an innovation while the length can be
interpreted through vocalic transfer (Benedict 1979). This root appears to be unrepresented
in the mainland AT languages (Kadai, Miao-Yao); the indicated line of development, with
syllabic reduction on-the-left along with vocalic transfer, is of the sort that characterises the
Kadai family as a whole. PAT *o is also maintained in this family, which may well have been
the source of an early loan (as an alternative to a 'look-alike') for this lexical item.

9


PAUL K. BENEDICT

The final clause was italicised lest the point be overlooked, as
apparently has occurred in the more recent paper by Diffloth (1985).
The nub of the writer's argument here lies in the fact that Shorto, a MK
specialist with good access to a wealth of comparative AN data, failed to
turn up an even halfway decent corpus of 'core' vocabulary correspon-
dences. This is indeed the name of the game, at least in the South-East
Asia region, on the basis of an apparent linguistic consensus at the present
time, and on this basis Austric surely emerged as the loser. And the writer
(1976b: 105), even 'at the risk of appearing whimsical', could quite
honestly report his belief that Shorto's paper had served only to
strengthen the case against Austric!

Almost a decade passed before the appearance of another defence of
Austric, this also by a leading Mon-Khmerist, Grard Diffloth (1985).
Unlike Shorto, however, who proposed a sizeable number of new AN/
MK cognate sets, Diffloth included only two 'possible Austric
etymologies': for 'wood' (-'tree') and 'bone', apparently the prize
members of his collection, preferable to certain 'runts' which he
modestly kept to himself. The pair that he does offer are in fact 'look-
alikes' or 'comparabilia', with Diffloth admitting to serious difficulties in
reconstruction even at the PMK, let alone a PAA or 'Austric', level. It
must also be noted that he has unparalleled MK comparative material at
his disposal along with the most recent advances in the AN field, all of
which he has made use of, hence his failure to do any better than Shorto
simply further strengthens the null case here. Given the present
circumstances, the notion that a presentable corpus of 'core' AN/MK
(or AT/AA) cognate sets lies 'out there somewhere' waiting to be
discovered strikes one as quixotic in the extreme.

Diffloth's citation of 'look-alikes' in an undertaking of this nature
conforms to general linguistic practice, as does his 'sharpening the
reconstructions of promising cases' (Schmidt had proposed the 'wood'
etymology). Quite novel, however, is his attack upon what he labels
'Swadesh's "basic" notions', which is no less than a frontal assault upon
the whole idea of utilising basic or 'core' vocabulary in comparative work.
Granted a certain feeling of frustration on his part in being unable to
uncover 'core' AN/MK cognate sets, one can hardly follow him in this
rash line of thinking. After pointing out that given items of'core' type can
on occasion be replaced through factors such as 'taboo or euphemistic
passe-passe', he makes the following summary statement:

Over the millennia, the recurrence of factors like the above [cit. supra]
could easily wipe out 100 such 'basic words', or even 200 [i.e. both the
standard Swadesh-type word lists], while protecting for us many hidden
gems, such as 'to whittle bamboo strips' or 'scruff, not to mention
'smegma' and the like. Benedict may detest it, but the Austric
hypothesis is still very much alive. (Diffloth 1985)

The writer does not 'detest' Austric; in fact, as an old supporter of the
hypothesis he must admit to a certain sentimental attachment to it and a

10


Austric: an 'extinct' proto-language

sadness at its demise. He must confess, however, that he is dismayed at the
thought of a proto-language with a corpus of reconstructed roots made
up exclusively of 'scruff, 'smegma', and the like. One can hardly visualise
comparativists recognising a Proto-Indo-European language put together
in such a fashion, and there is no reason for them to accept an 'Austric' of
this sort either. South-East Asia has its novelties as a research field, to be
sure, but it is hardly extra-linguistic. The cardinal point, as emphasised
above, is that roots of this kind, whether called ketketbongbong or not
(they need not be reduplicated), can be used in establishing a genetic
relationship only in conjunction with 'core' vocabulary. On theoretical
grounds, of course, all 100 or even 200 'Swadesh list' roots could have
been replaced in a given instance, but how would one prove it, especially
in the case of proto-languages, such as PAT and PAA, that occupy
contiguous territories? Perhaps, in the dim South-East Asian past, AT and
AA not only influenced each other structurally but also exchanged a few
scruffy or smegmatic words in the bargain, along with some 'high culture'
items after AT had made the early advance towards 'civilisation'. None of
this, however, justifies the setting up of an 'Austric' superstock. As
reported at the outset of this paper, Austric is 'extinct'.

REFERENCES

Benedict, R K. 1942. Thai, Kadai, and Indonesian: a new alignment in Southeastern

Asia. Amer. Anthropol. 44, 576-601.

1973. Austro-Thai and Austroasiatic. First Internat. Conf. on
Austroasiatic Linguistics, Honolulu (revised version appears as chap.
II in Benedict 1975).

1975. Austro-Thai: Language and culture. New Haven: Human Relations
Area Files Press.

1976a, Sino-Tibetan: another look. /. Amer. Orient. Soc. 96(2), 167-97.

1976b. ShortoIn defense of Austric.Comment. Comp. Analyses
Asian Afr. Langs. 6, 105-8.

1979. Vocalic transfer: a Southeast Asia areal feature. Acta Orientalia
40, 229-52.

1990. Japanese/Austro-Tai. Ann Arbor: Karoma Press.

(forthcoming). How to tell Lai: an exercise in classification (to appear in
Ling. Tibeto-Burman Area).

Diffloth, G.

1985. What happened to Austric? Paper presented at the 18th Sino-
Tibetan Conf., Bangkok.

Matisoff, J. A.

1976. Austro-Thai and Sino-Tibetan: an examination of body-part
contact relationships. Papers for the 1st Japan-US Joint Seminar on
East & Southeast Asian Linguistics, Tokyo.

Pinnow, H.-J.

1959. Versuch einer historischen Lautlehre der Kharia-Sprache.
Wiesbaden: Harrassowitz.

Schmidt, W.

1906. Die Mon-Khmer-Volker, ein Bindeglied zwischen Volkern
Zentral-Asiens und Austronesiens. Archiv fur Anthropologic 5, 59-109.

Shorto, H. L.

1976. Tn defense of Austric. Comp. Analyses Asian Afr. Langs. 6, 96-104.

11




PALAUNGIC VOWELS
IN MON-KHMER PERSPECTIVE

G. Diffloth

Thirteen years ago, H.L. Shorto pointed to the vowel system of Proto-
Mon-Khmer as being the 'crux' in the historical phonology of this family
(Shorto 1976). His assessment remains as valid today as it was then, even
though some advances in reconstruction have been made; our data base
has considerably expanded and improved, but the mirage of Proto-Mon-
Khmer vowels continues to recede, even as we penetrate further into the
past.

The solution proposed then: vowel variation in the proto-language, is
consistent with certain facts which can be observed in several Mon-
Khmer languages spoken today. In Bahnar, Sre, Khmu and Semai, to
select but a few, whole families of Expressives (Diffloth 1979; in press) are
often built on vowel permutations, and such Expressives occasionally find
their way into the prosaic (Non-Expressive) vocabulary; conversely,
prosaic words often serve as a starting point for building families of
Expressives which differ, for example, only by their major vowel. This has
surely contributed to the formation of word-families such as those
identified in Shorto (1973). This explanation, however, has its limits:
presumably, these processes would have affected a word here and a word
there, at different times, but it is difficult to see how it could have
pervaded the thousands of items which form the non-expressive lexicon of
one language, not to speak of an entire family. Other factors are needed in
order to account for the numerous vowel correspondences which have
been detected so far.

For example, it may well be that the Proto-Mon-Khmer vowel system
reconstructed until now, although sizable, is not rich enough for the
purpose, and that we need to expand it with some additional phonological
dimensions.

Tone has been practically ruled out for Proto-Mon-Khmer since the
simple tone systems of Blng and Riang, and the tone-cum-register
system of Nyah Kur (Diffloth 1980, 1984) can all be explained as
innovations; but the newly recorded Angkuic languages U and Man Met
(see below) have four-tone and six-tone systems respectively, the origins of
which remain partly unknown for the moment. Then again, Haudri-
court's account of Vietnamese (VN) tonogenesis has generally been
accepted, but it leaves out, as tonally irregular or unexplained, a large
number of words which do belong to the indigenous Mon-Khmer stratum
of the language. And the recently discovered Palyu language, called Lai in
Chinese, also has six tones which may, or may not, turn out to be

13


G. DIFFLOTH

recalcitrantPalyu is apparently Mon-Khmer (Liang 1986, Benedict, in
press), but its position in the family is still undecided.

Register is a better candidate. This is, typologically, a well-established
feature of Mon-Khmer languages (Huffman 1976). The general consensus
is that Register is a relatively recent phenomenon, and Shorto, accordingly,
does not reconstruct it for Proto-Mon-Khmer. Ferlus (1979) desribed all
Mon-Khmer register systems found until then as being the result of one
type of evolution: devoicing of initial consonants. This explanation has
long been accepted in the case of Spoken Mon (Blagden 1910) and of
Modern Khmer (which ironically has now lost phonation-type distinc-
tions), and it does account for the registers of several other newly recorded
languages (e.g. Kuy, Bruu, Phalok). But it is inadequate in certain other
languages: the Pacoh register system, admittedly an innovation, has
nothing to do with the process of devoicing which has independently taken
place in this language. It is also inapplicable to the North Bahnaric
languages where no devoicing has taken place, except in Sedang. In Pacoh,
the genesis of register is due to changes in vowel quality, namely, the
fronting or backing of certain proto-central vowels (Diffloth 1982), and a
similar innovation has apparently also taken place in North Bahnaric
(Diffloth 1983). So, we do not have yet a case of reconstructing register as
being ancient in Mon-Khmer.1

However, the Pearic branch might force us to do that: recently,
Huffman (1985) has shown that Chong, a language of the Pearic branch,
had a Clear vs. Breathy distinction, criss-crossing a Plain vs. Glottalised
contrast, giving rise in effect to a four-register system. Theraphan (this
volume) describes in detail the complex bundle of phonetic features these
four registers contain. This phenomenon has no historical explanation,
and Headley himself (1985) has abandoned to the sagacity of future
historical linguists any attempt in this regard.

Gage (1985) has pointed out that certain unexpected occurrences of the
sac tone in Vietnamese seem to find an echo in the register system of
Pearic. The tonogenesis of Vietnamese requires that the sac tone occur
with final proto-stops, and indeed cannot explain the tones of many VN
words which have excellent Mon-Khmer etymologies, such as: bon 'four',
chin 'cooked', gib 'wind', or: ngai 'far'. Cognates to all four of these
words happen to have glottalisation in Pearic. In Chong, as I have
recorded it, the first three have the 'tight' register:2 /phoon/ 'four', /chiin/
'cooked', /kayaay/ 'wind'; the fourth word has a 'breathy-creaky'
register3: /rjaay/ 'far'. Other examples can be found, e.g. VN: cam,
Chong /ksqaam/ 'rice-husk', but there are counter-examples as well, e.g.:
VN: chim, Chong /chiim/ 'bird'. Since the Pearic and the Viet-Mu55ng
branches are only distantly related, the implications of this fact could go
back directly to Proto-Mon-Khmer.

1. Smith's opinion to this effect (Smith 1972) was not based on the establishment of sound
correspondences, but on statistical tendencies within a very small set of possible Mon-Khmer
cognates, which a more thorough comparison does not confirm.

2. Clear voice plus glottalisation in Huffman's (1985) analysis.

3. Huffman's (1985) breathy voice plus glottalisation.

14


Palaungic vowels in Mon-Khmer perspective

There are also less exotic vowel features which have not been fully used
in Mon-Khmer reconstruction. Diphthong systems can be much richer
than the simple *ia and *us usually proposed; I have reconstructed Proto-
Katuic with five proto-diphthongs (Diffloth 1982), and Nancowry
Nicobar (Radakrishnan 1981:25) is described even today as also having
five diphthongs: /i/, /a/, u/, /a/ and /uia/, which seem to correspond
with what we can reconstruct for Proto-Aslian.

Some of these phonological features may have to be reconstructed back
to Proto-Mon-Khmer, and could well explain a number of Shorto's
variations as being regular outcomes of a much richer proto-vowel system.
But then, the number of proposed Proto-Mon-Khmer etyma becomes a
relevant issue, and what has been published so far can be said to represent
only a sample.

In this paper, I will not explore these possibilities, but only prepare the
comparative ground to do so; I will try to clarify some points in the
history of vowel systems in the Palaungic branch, where recently recorded
material allows us to make systematic reconstructions. This may seem at
first to be somewhat irrelevant: if Waic and Palaung are notable for one
thing, it is precisely the poverty of their vowel systems. The old vowel-
length contrast was already lost in Proto-Waic, and the best source of
information on Milne's Palaung ( = Ta-ang)4 appears not to have a
phonemic contrast of this kind.

But the Palaungic branch has an important role to play in reconstruction:
it belongs to a distinct division of the family, the Northern Division, and it
provides us with an independent testimony for the reconstruction of Proto-
Mon-Khmer vowels. Besides, as I will try to show, Palaungic vowel systems
are not as poor as they first seem to be.

1. Proto- Waic

The term 'Waic' covers (1) several Wa languages, e.g. Paraok, Aviia', La
(Zhou & Yn 1984) and their dialects; (2) the Phalok language,5 formerly
referred to as Khalo or Mae Rim Lawa (Flatz 1970); (3) Lawa and its
dialects (Mitani 1972); and (4) the Bulang-Phang complex with its many
dialects6 (Diffloth 1980). Certain Waic languages, Lawa and Paraok in
particular, currently have rich and complicated vowel systems7 but this is

4. Professor Shorto has let me use his own notes from Riang and from the same Palaung
language, Ta-ang, as described in (Milne: 1931); this is the source of the 'Ta-ang' and 'Riang'

words quoted here. Let the present article be a small token of appreciation for his kindness.
5.1 collected the information on Phalok included here in two separate field trips, one in April
1981 with the help ofTheraphan L. Thonkum, and the second by myself in July of the same
year. This research was financed by a grant from the National Science Foundation (NSF)
entitled 'An etymological lexicon of Mon-Khmer'. More information on Phalok will be
made available in the forthcoming volume 'Wa-Lawa-Bulang'.

6. In that study, I called 'Samtao' a language which later turned out to be identical to that
spoken by the Blng National Minority in Yunnan, China. No linguistic information was
available on Blng at the time, as Zhou & Yn (1983) had not yet appeared.

7. Because of a somewhat artificial analysis, Zhou & Yn (1984) describe Paraok as having 50
vocalic nuclei.

15


G. DIFFLOTH

due, in part, to the influx of Tai borrowings8 and in part to recent
processes of vowel warp, conditioned by Registers and final consonants.
Only nine proto-vowels are needed at the Proto-Waic stage9

Proto-Waic Vowel system

i i

evo
s o

a d

This maximum system is found with most final consonants, but there
are certain distributional gaps; for example, with final only eight
proto-vowels are found (all the above except *d); with final *-h, only seven
are found (*y and *d are excluded); and there are no open final syllables
in Proto-Waic.

2. Proto-Palaung-Rumai

The term 'Palaung-Rumai' also covers several languages, the best known of
which is Ta-ang, i.e. the Palaung of Nam Hsan described by Milne (1931).
The Rumai language and its dialects, also belongs here10, as well as the
dialects of Riang;11 it also includes another distinct group sometimes called
'Pale', which contains at least Da-ang and Na-ang; and several other
languages, like Ka-ang and Ra-ang12. Other Palaung-Rumai languages
surely await description in Burma or Yunnan, and they may or may not
belong to one of the seven groups mentioned here.

There is considerable diversity within Palaung-Rumai, but this is not
the place to present all the phonological innovations which can now be
documented. Mitani (1977, 1979) has already reconstructed the Palaung-

8. Lawa has borrowed from Northern Thai and more recently from Standard Thai; Paraok
has borrowed from Shan (referred to in China as 'Dhng Di'), Blng has borrowed from
Lii (referred to in China as 'Xshangbnn Di', or XT Di for short).

9. This reconstruction was presented in (Diffloth 1980); since then, a Paraok-Chinese
dictionary has been published (Yn et al. 1981), and, in 1984,1 was fortunate to collect Waic
linguistic material in China, with the help of Zhou, Z.-Z., under another NSF grant entitled
'Comparison of the Mon-Khmer languages of China with other languages of the Mon-
Khmer family'; with this new material, the number of reconstructed Proto-Waic words has
now more than doubled, and the reconstruction of a few etyma given in Diffloth (1980) has
been modified; this new information confirmed the nature of the Proto-Waic system I had
reconstructed in 1980 with the help of Y. Mitani.

10. All the Rumai examples quoted here were collected in May 1981, with the help of a family
of Rumai emigrants living in Chiang Mai at the time.

11. See note 4 above.

12. The information on Na-ang included here was kindly given to me by Yn, Q.-X. as part
of a research programme in China (see n.9). She is the author of a valuable sketch on the
Benglng language(s) (Yn 1983). I collected myself the Ka-ang data from a native speaker in
Kunming, in the course of the same research programme. The Da-ang and Ra-ang data were
collected in 1981 (see n.5), and 1984, during my stays in Thailand. It was not possible for me
to determine with precision the geographic spread of these languages, especially for those
spoken in the Shan States (Burma) where there seems to be a lot of small-scale migrations.
For China, Svantesson et al (1981), Yn (1983), Zhou & Yn (1983, 1984) provide
geographic and demographic information.

16


Palaungic vowels in Mon-Khmer perspective

Rumai vowel system, using older material; and even though none of the
sources used had indicated the vowel length contrasts which are clearly
present today in most of these languages, he did reconstruct vowel length
contrasts at the Proto-Pal-Ru stage. Remarkably, his reconstructions,
which he termed as tentative, are confirmed by the better material we now
have; this is true at least for the presence of a proto-length contrast, and
for the number of proto-vowels: 10 (although this represents a minimum);
only the reconstructed quality of some of these vowels can now be
improved upon.

Briefly, with only a few rare gaps and notation problems, we find the
following correspondences in words having a pro to final Velar Nasal:13

Rumai Na-ang Da-ang Ra-ang Ta-ang Ka-ang
(1) -aag -aag -aag -aag -ag -aag
(2) -Dg -og (-oog) -og -Dg -og -Ag
(3) -DDn -eeji (-ssji) (?) -vg -uig -og
(4) -ag -ag -Ag -ag -tug -DDg
(5) -uig -yg (-vyg) -uig -vg -ug -ug
(6) -og -oog -oog -oug -og -og
(7) -ssg -sag -8Ag -sag -og (-og) -DDg
(8) -en -eeji -ig -eg -ig -ig
(9) -DDn -88ji (-eeji) -eeg -eg -eg -eeg
(10) -(y)og -iag -iag -iag -eog -iag

Phonetic values for reconstruction of these proto-rimes do not come to
mind immediately, to say the least. But the first four are relatively easy:
Mitani (1977) reconstructs *aa, *a, *uiui, and *ui respectively, for (1), (2),
(3) and (4).

Correspondence (1), includes items such as:
'bone' Ru, Na, Da, Ra: /ka^aag/, Ta: /kan^ag/, Ka: /ka^aag/
'hawk' Ru: /klaag/, Na: /ma-glaag/, Da, Ra: /glaag/, Ta: /lag/, Ka:
/klaar)/

'elephant' Ru: /saag/, Na, Da: /ma-saag/, Ra: /saai]/, Ta: /sai]/, Ka:
/saag/

'house' Ru: /gaag/, Na, Da, Ra: /kaarj/, Ta: /gag/, Ka: /gaag/
'rancid, sour' Ru: /byaag/, Na: /praag/, Ta: /brag/, Ka: /braag/
'torch, lamp' Ru: /raag/, Ta: /rag/.

Riang-Lang cognates have a back /a/ in these words:
/tsn^g/ 'bone', /klg/ 'kite', /s tsg/ 'elephant', /kdg/ 'house, shrine',
/prdg/ 'sour, acid, rancid'.

Waic cognates always have Proto-Waic *a:

*S9?ag (Diffloth 1980: 5) 'bone'; *klag (69) 'hawk'; kssag (79)
'elephant' *?g-gag (12) 'scabbard' (a dervate, from an earlier verb
*gaag ('to cover, to protect'); *rag (53) 'brilliant, bright'.

13. Evidently, some vowels have caused PPal-Ru to palatalise to -ji, and sometimes
further merge with *-n; but both *-ji and *-n must be reconstructed at the proto-Palaung
stage, in contrast with *-rj.

17


G. DIFFLOTH

Lamet cognates have a long /aa/:

/xs^ag/ 'bone', /klag/ 'hawk', /kasag/ 'elephant', /ka g/ 'shelter,
shell, bark, husk';

And in the Angkuic branch of Palaungic, Man Met14 has /-ag/ finals
with a low-falling tone, and U14, which does not have vowel length
contrasts, shows finals in /-ai]/ with a high-falling tone:

MM: /?ag/, U: /s^g/ 'bone'; MM /khag/, U: /khlag/ 'hawk'; MM:
/sag/ 'elephant'; U: /kg/ 'house'; MM: /hag/ (high-rising tone
unexplained) 'bright'.

Outside Palaungic, Mitani's reconstruction of *-aag is confirmed, for
instance, by Khmu:

/cs^aag/ 'bone', /klaag/ 'hawk', /sscaag/, 'elephant', /gaag/ 'house'

Note that we will not be concerned here with the history of initial stops in
these languages, interesting as that may be. In a nut-shell, Rumai and Ta-
ang have preserved the original values of proto-voiced and proto-voiceless
stops; the same state of affairs also exists in Ka-ang, except that initial *p-
and *t- have become implosive 6- and cf- respectively; in Ra-ang, *p- and
*t- have followed the same evolution, and *k- has become a voiced stop g-,
while all proto-voiced stops have become voiceless; Da-ang and Na-ang
have followed the same course as Ra-ang, and, in addition, have lost the
implosion of 6- and cf-, which then become ordinary b- and d-; Da-ang and
Na-ang therefore show a total reversal of the voicing values of PPal-Ru
stops: another illusion of 'flip-flop', with implosiveness as the point of
transition in this game of musical chairs. Angkuic, on the other hand, has
undergone a completely regular 'Germanic'-type of sound change where
proto-voiceless stops are now aspirated, and proto-voiced stops are now
voiceless. But note that Angkuic tono-genesis is mainly due to vowel
qualities and quantities, with some influence from the final consonant; it is
unrelated to the earlier or the present voice features of initials.

Correspondence (2) includes:
'bitter, gall' Ru: /sg/, Na, Da: /sog/, Ra: /sg/, Ta: /sag/, Ka: /sai]/
'thatch-grass' Ru: /plDg/, Na: /bloorj/, Da: /blog/, Ra: /61di]/, Ta: /ptag/
'bamboo-shoot' Ru: /bg/, Na: /poog/, Ta: /bag/, Ka: /bAg/
'house-pole' Ru: /rog/, Na, Da: /rog/, Ka: /rAg/
'horse' Ru: /mbyog/, Da: /mprog/, Ta: /brag/, Ka: /brAg/
Riang-Lang cognates have a front /a/ in these words:
/tsg/ 'bitter', /plg/ 'thatch-grass', /knrg/ 'post, upright', /mrag/
'horse'

14.1 recorded Man Met and U from native speakers, in Ynnn, during the research project
mentioned above (n.9); they were introduced to me as speaking 'dialects', or more exactly
'fangyn', of the Blng language. The Chinese term 'fangyn' corresponds most of the time
to what Western linguists consider to be different languages; this was true of these 'fangyn'
of Blng which do not even belong to the Waic branch of Palaungic, but to the little known
Angkuic branch (Diffloth, 1974). The location of the U language is given in (Zhou & Yn
1973), Man Met is spoken a few miles from Jinghong, Xshangbnn, Ynnn.

18


Palaungic vowels in Mon-Khmer perspective

Proto-Waic cognates have *d:

*sdq (80) 'bitter', *pbr) (73) 'thatch-grass', *tabdr) (44) 'bamboo-
shoot', (56) 'house-pole', *mrDi] (58) 'horse'.
Lamet has a short /a/:
/ci]/ 'bitter'; /plg/ 'thatch grass'; /tspaq/ 'bamboo-shoot'; /mxi]/
'horse'

And in the Angkuic branch, Man Met has /-q/ rimes with a high-rising
tone, and U has /-k/ rimes with mid-rising tone:
MM: /sr)/, U: /chk/ 'bitter'; MM: /phr)/ U: /phlk/ 'thatch-grass';
MM: /hq/, U: /crk/ 'house-pole'; MM /pq/, U: /mbrk/ 'horse'.
Here again, Mitani's reconstruction of a short *a is confirmed, outside
Palaungic, by Khmu which regularly shows cognates with short /a/:
/caq/ 'bitter', /tbaq/ 'small bamboo-shoot', /cndraq/ 'house-pole',
/mbraq/ 'horse'.

The contrast between correspondences (1) and (2) shows that length vs.
shortness has been preserved everywhere in the Palaung-Rumai lang-
uages, at least for this pair of vowels, except in Ta-ang which does not
show any length distinctions anywhere in its system.

It also shows that in the Angkuic branch, tonogenesis is directly due to
vowel-length, and has nothing to do with the proto-voice feature of
initials; this kind of tonogenesis is unique in the Mon-Khmer family, but,
annoyingly, it accounts for only some of the tonal contrasts found in
Angkuic languages. It should also be noted that Man Met does undergo
this kind of tonogenesis even though the older length distinction is
retained; in U however, the length contrast, before disappearing, leaves
another trace in the final consonant; it de-nasalises final nasals after short
vowels.

Another remark: if we only had the Waic and the Riang-Lang material
at our disposal, it would appear that a so-called flip-flop has taken place:
Waic has a front *-a- where Riang-Lang has a back /a/; and vice-versa:
Waic has a back *-d- for Riang-Lang's front /a/. The former presence of a
length contrast shows this apparent flip-flop to be nothing but a
synchronic illusion.

Correspondences (3) includes:
'high' Na: /leeji/, Ta: /hluir)/, Ka: /hloi]/
'to dig (a hole)' Ta: /kuii]/

'yarn' Ru: /sDDn/, Na: /seeji/, Ra: /svr)/, Ta: /sun]/, Ka: /soi]/
'foot' Ru: /jDDn/, Na: /ceeji/, Da: /ceen/, Ra: /cyi]/, Ta: /juir)/, Ka: /jog/
Riang-Lang has an /o/ reflex:
/kor)/ 'to dig', /tsor)/ 'foot'
and Proto-Waic shows *-o-:

*hloi] (77) 'high', *kor) (9) 'to dig', *joij (18) 'foot'
Lamet has /ee/:
/leeq/ 'high, long', /keeg/ 'to dig', /ceeg/ 'foot'.

19


G. DIFFLOTH

In Angkuic, Man Met has long /ee/ with a low falling tone, U has /e/ with
a high-falling tone and a final nasal:
MM: /leg U: /hleg/ 'high, long'; U: /khg/ 'to make with a dibble-
stick'; MM: /ceg/, U: /cg/ 'foot'.
Outside Palaungic, Khmu has cognates with /ia/:
Khmu Yuan (Svantesson, personal notes): /khig/ (aspiration unex-
plained) 'to dig up', Southern Khmu: /jiag/ 'foot'.

Correspondence (4):

'bamboo' Ru: /hrarj/, Na: /hrog/ (Vowel ?), Da: /hrAg/, Ra: /hrag/, Ra:
/hruig/, Ka: /sjDog/

'bed-bug' Ru, Na, Ra: khag/, Ta (Milne, 1931): kong

'stalk, trunk, post' Ru, Na: /tar)/, Ta: /tag/ (Milne: ting, tong), Ka:

/cTddi)/

'meat' Ru, Na: /yag/, Ta: /yuig/, Ka: /yDDg/
The Riang reflex is /a/:

/rag/ 'bamboo', /tag/ 'tree-trunk', /ysg/ 'meat'
Proto-Waic has *-y-:

*hvg (Paraok: /hug/, Drage's Wa: hong, Phalok: /hug/, Phang: /huig/
'bed-bug'
Lamet has a short /a/:

/rag/ 'bamboo', /hag/ 'bed-bug'
In Angkuic, Man Met has a short /a/ with a high-rising tone (but the tone
of 'bed-bug' seems to be low-rising), while U has a /a/ with a mid-rising
tone and de-nasalisation of the final:

MM: /hag/, U: /hrak/ 'bamboo'; MM: /sag/ (tone?), U: /sSk/ 'bed-bug'
Outside Palaungic, Khmu has a short /ui/:
/huig/ 'bed-bug'

Another etymon with the same proto-vowel, unfortunately without
Palaung-Rumai attestations, is:

'horn': Riang /kmrag/, Proto-Waic *?rvg (62), Lamet /kruig/,
Angkuic: MM: /kg/, U: /krak/; Khmu: /cndruig/.

Mitani's reconstruction of a length contrast, *uiui vs. *ui, for these two
proto-vowels is confirmed independently by the evidence of Lamet and
Angkuic (both Man Met and U); in Palaung-Rumai, Na-ang and Rumai
also display the same contrast, which appears to be a retention, not only
from Proto-Palaungic, but from still earlier periods, as the Khmu
evidence indicates.

For all four proto-vowels *aa, *a, *uiui and *ui, Na-ang and Rumai
have thus preserved the older length feature. I depart slightly from Mitani
in terms of vowel-qualities: (3) and (4) were probably central mid-vowels,
and This would explain how (3) became *ee in both Lamet and
Angkuic and *o in Proto-Waic, and why (4) has very open vowel quality
reflexes throughout Palaung-Rumai, except in Ta-ang where (3) and (4)
merged and were pushed higher to /ui/ by the shift of *a to /s/. The
original value of (4) is preserved in Riang, Lamet and Angkuic.

20


Palaungic vowels in Mon-Khmer perspective

For the remaining vowel correspondences, Mitani does not reconstruct
any length contrast. For (5), (6) and (7) he proposes a back-rounded
series: *u *o and *o respectively.

Correspondence (5) includes:
'big village, country' Ru: /kuir)/, Ta: /kurj/
'drum' Na: /grvvg/, Ra: /grvrj/, Ta: /krug/, Ka: /kruig/
'bamboo-strip mat' Ru: /bluig/, Ra: /plvg/, Ta: /blur)/
'love, like' Ru, Da: /?uig/, Ra: /?yg/, Ta: /^uq/
'to bury' Ta: /krpug/, Ka: /ka6ug/ (Vowel ?)
The corresponding PWaic vowel is *i:

*kirj (10) 'wet field, country', *krig (63) 'drum', *krpig (41) 'to
bury'

In Angkuic, Man Met has a short /u/ with high tone, and U has /u/ with a
mid-rising tone and de-nasalisation of the final:

MM: /khg/, U: /khk/ 'wet rice-field'
And outside Palaungic, Khmu has /u/: /kug/ 'village'

Lamet cognates are missing, although there are other etyma with
apparently the same proto vowel, but without Palaung-Rumai attesta-
tions; these show the Lamet reflex to be short /u/:
'to blow': Lamet /pij/, PWaic *pig (40), Man Met: /phg/, Khmu /pug/
'a sprout': Lamet: /plug/ (tone ?), PWaic *blig (Paraok /plaag/), Khmu:
/blur,/

For this correspondence, the evidence for proto-shortness is clear: in
Palaung-Rumai, only Na-Ang has a long vowel (in a single item which
could have been misrecorded); all other languages where a length contrast
exists, Rumai, Da-ang, Ra-ang, Ka-ang, Lamet, Angkuic, Khmu, have a
short reflex.

The other two correspondences (6) and (7), seem, by contrast, to be on
the long side:

Correspondence (6)
'knee-cap' Ru: /gyog/, Da: /kroug/, Ra: /kroog/, Ta: /grog/, Ka: /grog/
'male bird' Ru: /kog/, Ta: /?3-kog/, Ka: /kog/
'Classifier: round objects' Ru: /pog/, Ta: /pog/, Ka: /6og/
'buttocks' Na: /sopoog/, Da: /sopoug/, Ra: /sopoog/, Ta: /sobog/ (my
own recording, cf. Milne sa-bong which would indicate /sobog/)
'far, long' Ru: /dog/~/ndog/, Na: /toog/, Ra: /toog/, Ka: dog/
The Proto-Waic reflex is *o:

*sgrog (61) 'knee-cap', *kog (8) 'peacock'
Lamet has a long /oo/ reflex:
/kxog/ 'knee'

In Angkuic, Man Met has a long /oo/ with a low-falling tone, and U has
an /o/ with high-falling tone but no denasalisation of the final:

MM: /kog/ 'knee'; U: /khog/ 'peacock', U: /phg/ 'round object' (in
the expression: /khk phg/ 'pubis', where /khik/ = 'head')

21


G. DIFFLOTH

Correspondence (7) is well documented and contains well-known etyma; it
also includes some surprising Front reflexes for what is certainly a proto
Back vowel:

'hornet' Ru: /?eeg/, Da /?eAg/, Ta /?og/, Ka: /?DDg/
'rainbow' Ta: /poryog/, Ka: /?oyDDg/

'back (of body, of knife)' Ru: /kyeeg/, Ra: /greag/, Ta: /krog/. Ka:
/krDDq/

'stairs' Ru: /ndeeg/, Na: /nteag/, Ka: /dDDg/
In Riang, the normal reflex is a diphthong /ua/ or / uo/:
/9ag/ 'hornet', /prjiog/ 'rainbow', /rsgdag/ 'stairs'
Proto-Waic regularly shows *o:

*9og (6) 'hornet', *pryog (84) 'rainbow', *krog (60) 'back'
Lamet has a long /oo/ reflex:

/?6og/ 'hornet', /pxyog/ 'rainbow', /kxoog/ 'back'
In Angkuic, the Man Met reflex is a long /oo/ with a low falling tone, and
the U reflex is a diphthong /ua/, with a high-falling tone and no
denasalisation of the final:

MM:/?oog/, U: /9ag/ 'hornet'; U: /phyag/ 'rainbow'; U: /?ag-ghrag/
'backbone'

Outside Palaungic, the Khmu reflex is regularly a long /oo/:

/?oog/ 'hornet', /pryoog/ 'rainbow', /kndroog/ 'back', /rgdoog/ 'stairs'

Mitani was certainly justified, on phonemic grounds, in leaving the
feature of length unspecified in the back vowels series of Proto-Palaung-
Rumai; but we can afford to be more precise now, and say that *u was
probably short, while *o and *o were probably long *oo and *oo. This will
allow for an easier description of the gradual collapse of the older vowel-
length system in languages like Ta-ang and Proto-Waic. And for Front
vowels, the same line of reasoning will help us to solve a curious problem.

The proto Front Vowel system was also left unspecified as to length by
Mitani, who reconstructed *i, *e, and *s for our correspondences (8), (9)
and (10) respectively.

Correspondence (8):
'to sew' Ru: /jen/, Na: /ceeji/, Ra: /ceg/, Ta, Ka: /jig/
'head' Ru: /ken/, Na: /geeji/, Da: /gig/, Ra: /geg/, Ta, Ka: /kig/
'navel' Ru: /koden/, Na: /koteeji/, Da: /kotig/, Ra: /koteg/, Ta: /ksrdig/,
Ka: /kodig/

'bamboo water-container' Ru: /den/, Ra: /teg/, Ta, Ka: /dig/
'husband' Ru: /men/
Riang has an /i/ reflex for this correspondence:

/kig/ 'head', /kndg/ 'navel'
But PWaic has an *e:

*jeg (17) 'to sew', *keg (7) 'head', *krdeg (30) 'navel', *deg
(Paraok: /taji/) 'bamboo water container', *hmeg (47) 'male of
animal'

Lamet normally has a short /i/, except for one case of long /ii/:

22


Palaungic vowels in Mon-Khmer perspective

/cig/ 'to sew', /kig/ 'head', /tig/ 'bamboo-container', /miig/ 'male of
animal, endearing term for son'

This last item may actually be one of the many words Lamet has
borrowed from the neighbouring Khmu; unfortunately, I have not yet
found the expected word /hmiig/ in any dialect of it. But the Khmu reflex
of correspondence (8) is indeed a long /II/:
Khmu Yuan (Svantesson, personal notes): /ktiig/ 'navel', /tiig/
'bamboo water-container'
In Angkuic, Man Met has a short /!/ with a high-rising tone, and U has /!/
with a mid-rising tone and denasalisation of the final Nasal:

MM: /khg/, U /khik/ 'head'; MM: /mtig/ (tone ?) 'bamboo con-
tainer';15 MM: /-tig/ 'the middle of (in a compound: /kag-/)'; U: /hmik/
'male of animal'

The Paluang-Rumai evidence, except for Na-ang, points to a proto-short
vowel *i, and this is confirmed at the Proto-Palaungic level by Lamet and
Angkuic.

Correspondence (9):
'sky' Ru: /plDDn/, Na: /bleep/, Da: /bleeg/, Ra: /61eg/, Ta: /pleg/, Ka:
/61eeg/

'way, path' Ru: /ndDDn/, Na: /nteeji/, Da: /nteeg/, Ra: /?anteg/, Ta:
/(ra)-deg/, Ka: /deeg/

'yellow' Ru: /tDDn/, Na: /deeji/, Da: /teeg/, Ra: /deg/, Ta: /teg/, Ka:
/deeg/

'wheel' Ru: /kalDDn/, Ra: /ksleji/, Ta: /ksnleg/
'equal amount' Ta: /krpreg/
Riang cognates have an /e/ vowel:

/plg/ 'sky', /ragdg/ 'way', /kanleg/ 'wheel', /trkrg/ 'equal amount'
And Proto-Waic, surprisingly, has an *i:

*klig (Paraok: /klig/) 'to spin (yarn)'; *mrig (Bo Luang and Umphai
Lawa: /mbrig/) 'to compare quantities, to match'
The Lamet evidence is, unfortunately, limited to a single item:
/tsmpliig/ 'sky'

In Angkuic, I do not have Man Met cognates, but U has the same reflex
as in correspondence (8): /!/ with mid-rising tone and denasalisation:
U: /phlik/ 'sky'

This merger of (8) and (9) is specific to U and not general in Angkuic:
there are other cognate sets, unfortunately without attestations in
Palaung-Rumai, where this U rime corresponds to Man Met /-eeg/ with
middle tone, (and to Proto-Waic *-ig, as correspondence (9) requires):
'to return home' U: /?k/, PWaic *?ig (1)

'wall, partition' U: /ndhik/, Mok (a close relative of Man Met spoken in
Thailand, Wenk's (1965) 'Ya Ang Lawa'): /theeg/, PWaic *ntig (21)
'ginger' U: /sakhik/, MM: /kheeg/, PWaic *sgkig (Paraok: /ssggig/, La:
/kig/, Avia': /sagkiag/, Phalok: /kmg/, Bulang (Da-Luo): /kkg/, Lawa
(Umphai): /ssceg/

15. The initial /m-/ is probably a trace of the word /?oom/ 'water' with which *dii] often
forms compounds, e.g. Blng pm tg/ 'bamboo water-container'.

23


G. DIFFLOTH

I propose to reconstruct PPal-Ru *ee for correspondence (9), and for
PPalaungic as well. The Palung-Rumai reflexes indicate a long vowel, even
in Rumai, where the strange reflex, /-DDn/, represents a merger with *-3sg
(correspondence 3), itself a proto-long vowel.

This reconstruction also provides a simple explanation for what would
appear to be yet another case of flip-flop: if we kept Mitani's
reconstruction, without length, the correspondences would be:

(8) *-i-: PPal-Ru *-i- =PWaic *-e-

(9) *-e-: PPal-Ru *-e- =PWaic *-i-

If, however, we reconstruct PPal-Ru *ee for (9), head-on collisions are
easily averted, and gradual phonetic change can proceed smoothly.

In addition, we can also explain the Lamet and Angkuic reflexes: in
those two sub-branches, Proto-Palaungic (correspondence 3) was
fronted to a long /ee/, pushing the older *ee (corr. 9) out of the way: in
Lamet, this *ee was raised to /ii/, keeping its long feature and filling a gap
in the vowel system; but in Angkuic, Man Met and U evolved differently:
in Man Met *ee and merged in terms of qualities and length, but seem
to have acquired different tones, whereas in U, *ee simply merged with *i.
If this is correct, the movement of to /ee/ is probably not an
innovation shared by Lamet and Angkuic; it may have happened
independently in these two sub-branches which do not appear to be
especially closely related to each other.

The last correspondence, (10), poses a special problem in that it is
difficult to decide if *se or *ia should be reconstructed, for either PPal-Ru
or PPalaungic. In any event, a short *e seems a very unlikely value for this
correspondence in view of its reflexes:

'to drink' Ru: /cog/, Na, Da: /diag/, Ra: /diag/, Ta: /tesg/, Ka: /diag/
'excrement' Ru: /?yog/, Na, Da: /9iag/, Ta: /?eag/, Ka: /?iag/
'rice straw' Ru: /hroog/, Na: /riag/, Ra: /hriag/, Ta: /hresg/, Ka: /sjiag/
'oily, unctuous, delicious' Ru: /pyog/, Na: /briag/, Ra: 6riag/, Ta:
/preag/, Ka: /priag/

'wrist, ankle; bracelet' Ru: /kyog/, Na: /giag/, Ta: /kesg/
The Riang reflex is a diphthong, usually transcribed /ie/:

/tieg/ 'to drink', /yg/ 'excrement', /rieg/ 'straw', /prieg/ 'delicious'
In PWaic, the reflex is *e:
*?eg (3) 'excrement'

The normal Lamet reflex is a long /ee/, although 'excrement' has
unexpected reflexes, possibly due to euphemistic deformation:

/teg/ 'drink', /?g/~/9g/ 'excrement', /preg/ 'oily', /kog-keeg/ 'elbow'
In Angkuic, the Man Met reflex is a short /e/, usually with a high tone,
whereas in U, we find a diphthong /ia/ with a high-falling tone and a
Nasal final; 'excrement' is irregular only in U, having what seems to be an
otherwise unattested proto-short diphthong:

MM: /thg/, U: /thrng/ 'drink'; MM: ptg/, U: /?eak/ 'excrement', MM:
/mpheg/, U: /phrTag/ '(pig) fat'

24


Palaungic vowels in Mon-Khmer perspective

Only one cognate has been found so far in Khmu16: /kiai]/ 'elbow'.

For the sake of consistency with Mitani's notation, I will arbitrarily
reconstruct *ee for this correspondence at the PPal-Ru level.

The vowel system reconstructed for PPal-Ru now appears as follows:

Proto-Palaung-Rumai Vowel System
*i *u

*ee *oo

*S8 *03

*aa *a

In contrast to the 9-vowel system shown above for Proto-Waic, this 10-
vowel system shows the vowel length contrast to be still operating, but
with a small functional load. I have not attempted here to reconstruct the
Proto-Palaungic state of affairs, because Lamet and Angkuic are still too
poorly known; but it seems likely that the vowel system will be, if
anything, richer at that stage, probably filling some of the gaps in the long
vs. short contrasts.

The Proto-Palaung-Rumai vowel system appears to be, typologically,
half-way between a full South-East Asian system, as found for example in
Khmu or Standard Thai, and a more contracted system where vowel
length has been lost, as found for example in Proto-Waic, and Modern
Mon. This vowel-system contraction seems typical of the Burma-Yunnan
linguistic sub-area. But what Palaung-Rumai shows is that areal pressure
does not work like a stream-roller: Ta-ang did lose the length contrast,
but Rumai, even with two mergers, maintains it systematically and even
innovates in this respect with complete disregard for its more forceful
neighbours.

Something similar can be said of U: while it did lose vowel-length and
acquired tones, thus conforming to its neighbours, it did so in a way
which is competely original since the tono-genesis of U is partly due to
vowel length.

In the perspective of Proto-Mon-Khmer, we may have to cope with as
many upheavals, during that long stretch of history which separates us
from these ancient times, as we can see in the relatively short adventure of
Palanguic vowels.

REFERENCES

Blagden, C. O. 1910. Quelques notions sur la phontique du talain et son volution

historique. J. Asiatique (10me sr.) 15, 477-505.

Diffloth, G. 1974. Austro-asiatic languages. Encyclopedia Britannica (15th ed.) 2,

480-4.

16. Tn spite of its conservative phonology, Khmu is not as useful for reconstructing Proto-
Palaungic as it might appear at first glance; it is true that the Khmuic branch is closer to
Palaungic than to other branches of Mon-Khmer, but the Khmu lexicon has undergone a
great many replacements.

25


G. DIFFLOTH

Diffloth, G.

Ferlus, M.
Flatz, G.
Gage, W. W.

Headley, R. K. Jr.

Huffman, F. H.

Liang, Min

Milne, Leslie
Mitani, V.

Radhakrishnan, R.

1979. Expressive phonology and prosaic phonology in Mon-Khmer.
In Studies in Tai and Mon-Khmer phonetics and phonology in honour of
Eugenie J. R. Henderson (eds.) T. L. Thongkum et al. Bangkok:
Chulalongkorn Univ. Press. 49-59.

1980. The Wa languages. (= Ling. Tibeto-Burman Area 5(2), 1-182.)

1982. Registres, dvoisement, timbres vocaliques: leur histoire en
Katouique. Mon-Khmer Stud. 11, 47-82.

1983. Parthenogenesis of register in North Bahnaric. Paper read at the
9th Annual Meeting of the Berkeley Ling. Soc., Berkeley.

1984. The Dvaravati-Old-Mon language and Nyah-Kur (Monic Lang.
Stud.l). Bangkok: Chulalongkorn Univ. Printing House.

(in press) T big, 'a' small. In Conference on sound symbolism (eds.) J.
O'Hala & L. A. Hinton. Berkeley: Univ. California Press.

1979. Formation des registres et mutations consonantiques dans les
langues mon-khmer. Mon-Khmer Stud. 8, 1-76.

1970. The Khalo or Mae Rim Lawa, a remnant of the Lawa
population of Northern Thailand. J. Siam Soc. 58(2), 87-104.

1985. Vietnamese in Mon-Khmer perspective. Tn Southeast Asian
linguistic studies presented to Andr-G. Haudricourt (eds.) Suriya
Ratanakul et al. Bangkok: Mahidol Univ., 493-524.

1985. Proto-Pearic and the classification of Pearic. In Southeast Asian
linguistic studies presented to Andr-G. Haudricourt (eds.) Suriya
Ratanakul et al. Bangkok: Mahidol Univ., 428-78.

1976. The register problem in fifteen Mon-Khmer languages.
Austroasiatic Stud. 1 (Oceanic Ling. Spec. Publ. 13). (eds.) P. N.
Jenner et al. Honolulu: Univ. Hawaii Press, 575-89.

1985. The phonology of Chong, a Mon-Khmer language of Thailand.
Tn Southeast Asian linguistic studies presented to Andr-G. Haudricourt
(eds.) Suriya Ratanakul et al. Bangkok: Mahidol Univ., 355-88.

1986. On the genealogical classification of the Lai language.
Computational analysis of Asian and African languages 26: 1-14.
[Tokyo.]

1931 .A Dictionary of English-Palaung and Palaung-English. Ran-
goon: Govt. Printing and Stationery.

1972. A short vocabulary of Lawa (ra-wa-go goi shiryo). Tonan Ajia
KenkyU 10(1), 131-68.

1977. Palaung dialects: a preliminary comparison. Tonan Ajia KenkyU
15(2) 193-212.

1979. Vowel correspondences between Riang and Palaung. In Studies
in Tai and Mon-Khmer phonetics and phonology in honour of Eugenie J.
A. Henderson (eds.) T. L. Thongkum et al. Bangkok: Chulalongkorn
Univ. Press, 142-50.

1981. The Nancowry word: phonology, affixal morphology and roots of
a Nicobarese language (Current Enquiry into Lang, and Ling. 37).
Carbondale 111.: Linguistic Research Inc.

26


Palaungic vowels in Mon-Khmer perspective

Shorto, H. L. 1973. Three Mon-Khmer word families. Bull. Sch. Or. Afr. Stud.

36(2), 374-81.

1976. The vocalism of proto-Mon-Khmer. In Austroasiatic Stud. 2
(Oceanic Ling. Spec. Publ. 13). Honolulu: Univ. Hawaii Press, 2.
1041-67.

Smith, K. D. 1972. A phonological reconstruction of Proto-North-Bahnaric. Santa

Ana: Summer Institute of Linguistics.

Svantesson, J.-O. 1981. et al. Mon-Khmer languages in Yunnan. Asie du Sud-Est et
Monde Insulindien 12(1-2), 91-100.

Yn Qi-xiang 1983. Benglng-y gi kung (A brief description of the Benglong

language) Mnz Ywn (1983) 5, 67-80.

Yn Qi xiang 1981 .et al. Pug lai cix ding yiie si ndong lai vax mai lai hox (A concise

Wa-Chinese dictionary). Kunming: Ynnn Mnz Chbnsh.

Zhou, Z.-Z. 1983. Blng-y gi kung (A brief description of Bulang language).

& Yn, Q. X. Mnz Ywn (1983) 2, 71-81.

1984. Wa-yu jin zhi (A concise description of the Wa language).
Beijing: Mnz Chbnsh.




COMMUNICATEES, EXISTIVES, AND
STATIVES IN PROTO-SOUTH-BAHNARIC1

David Thomas

0. Introduction

This paper is a study of some communicative, existive, and stative clause
types in South Bahnaric languages, comparing them, and postulating
some Proto-South-Bahnaric (PSB) forms. The data from the various
languages, as will be apparent from the discussion, are uneven in both
quantity and quality, so that the present paper must be considered
preliminary.

The South Bahnaric languages are the southern section of the Bahnaric
branch of Mon-Khmer (Thomas & Headley 1970), located mostly in
southern Vietnam, with some spilling over into Cambodia. I draw most
heavily on Chrau, Eastern Mnong (Rlam), and Stieng, as representative
languages of the group, with additional data from Koho Sre and Central
Mnong (Bunar and Preh). The three main languages above are
respectively at the south-eastern, north-eastern, and western edges of
the South Bahnaric area, so should give a fairly good picture of the range
of diversity.

In the examples, words whose main significance seems to be as functors
rather than as content words are underlined. Vocabulary items, mostly
nouns and adjectives, whose meaning is not basic to the structure of the
clause, are glossed beside their first occurrence. Functors and central
verbs, i.e. elements that are basic to the clause structure, are listed and
glossed below the set of examples. It would be desirable to list other verbs
that take the same structure, but in most cases I am limited to the
published data sources.

In the reconstructions, an agreement of Chrau, Rlam, and Stieng is
taken as sufficient evidence to reconstruct it for Proto-South-Bahnaric.
An agreement of just Chrau and Stieng is also considered sufficient if
there is no contrary evidence.

1. Communicatives

The talking perceiving quoting informing group of clause types have
a basic SpeakerVAddresseeInformation order in all the daughter
languages.

1. This is a companion article to "Some Proto-South-Bahnaric Clause Grammar", paper
delivered to the 18th Sino-Tibetan Conference, Bangkok, 1985. (Mon-Khmer Stud. 15,
1989,111-24). That article dealt with transitivity and locational clause types. The numbering
of the examples here follows on the previous numbering.

The clause presentation is based on my clause components outline (Thomas 1983: 137-
42). It is semantically based, looking for and comparing the forms which manifest the
desired meanings.

29


DAVID THOMAS

Mnong Bunr ( = B)

B3a: gop ngdoingach 'I speak fast' (Phillips 1963 = MLC. 1.1) (ngach
'fast')

B3k: gop ntay an naao BuNoong ma khon ay 'I will teach the Mnong
language to you' (MLCA.2) (naao BuNoong 'Mnong language',
khon ay 'you f.pl.')
ma 'to'
ngodi 'speak'
ntay an 'teach'

Chrau ( = C)2

C3a: a aai (yuur yuur) 'I speak (slowly')
C3b: a aai sk[ neh 'I talk about him' (neh 'him')

a aai sinlo i heq 'I talk about this house' (i heeq 'this house')
a naai siq sinlo i heeq 'id.'
C3c: a aai bay neh 'I spoke to him'
C3d: a chiih neh 'I scolded him'

C3e: a aai paa neh saaq 'I said he went/I said "He went" '
C3f: a paa neh saaq 'I said he went'

a paa, neh saaq 'I said, "He went"'

2. The Chrau data are my own (see Thomas 1971); the Koho Sre data are from Evans &
Bowen n.d. (indicated as KLC) and Manley (OSS); the Mnong Bunr data are from Phillips
Ms. (MLC); the Mnong Preh data are from Phillips & Kem (1974; CMLL)\ the Mnong
Rlam data are from Tang (1976; MLLL), plus personal communications from Evangeline
Blood (1985; unmarked), and the Stieng data are from Miller 1976 (OSG), plus personal
communications from Ralph Haupers (1985; unmarked). I was not able to recheck any of
the data with native speakers.

Because of varied spelling conventions used in the different sources, T have standardised
the writing of length as VV, the voiceless velar stop as k, and the final glottal stop as q. The
'whiskered' o> and w are rendered here as o and w. The Koho o with the lowered dot is
rendered g.

Four different spelling systems have been used for Koho in the past (Manley 1972:39), so

I have converted the data from the different sources to the so-called SIL system, as it more
closely matches the spelling of the other South Bahnaric languages.

In some cases I have taken the liberty of replacing nouns and place names with other
nouns and places names for reader ease.

In the original sources shortness/length is marked as follows:

Usual Markings Other Markings Environments

short long only short only long

Bunr
Chrau
Koho KLC
OSS

Preh
Rlm
Stieng

v

v, a,
vN, vT
v

v,
v,
v

-i/y, u/o, ti/o
-i/y, u/o
-i/y, u/o
v = vq
-i/y, u/o
-i/y, u/o

v,e,o
v, e, o,
vN, vT
v

v, e, ti, o,
v

vv

A combined phonetic chart of the vowels would be:
Front Central Back

High i u

Low

-h, -?

-h, -?
-h, -?

-0
-0
-0
-0
-0
-0
-0

e, uo, uo are
centralising offglides

30


Clause types in Proto-South-Bahnaric

C3g: a aai ba^ neh paa a saaq 'I told him that I was going'
C3h: a chiih neh paa neh saaq 'I scolded him, saying that he went'
C3i: a aai aan neh gt (paa) a saaq 'I spoke letting him know I
was going'

C3j: a paa neh saaq 'I invited him to go'
a sir neh saaq T invited him to go'
a aan neh saaq 'I allowed him to go'
C3k: a padau neh gih troong Chrau 'I'll teach him the Chrau
language'

aan 'permit, allow'

aan...gt 'let...know'

bay 'with, to'

chiih 'speak, scold'

gt 'know'

aai 'speak'

padau 'teach'

paa 'say, saying, invite'

sier 'invite

siq 'concerning (lit. returning)'
Koho Sre (= K)

K3a: a (ggq) dos 'I (don't) speak' (Manley 1972 = OSS.2M)

a dos adaar adaar 'I speak slowly' (Evans & Bowen, n.d. =
KLC.2) (adaar 'slow'
K3j: a jaaq me saao 'I invite you to eat' (KLC.63) (me saao 'you eat')
dos 'speak'
jaaq 'invite'

Mnong Preh (= P)

P3a: gap ngobi ngach 'I speak fast' (Phillips & Kem 1974 =

CMLL. 18) (ngach 'fast')
P3j: gap jaq may soong sa 'I invite you to eat' (CMLL.2 (soong sa
'eat')

P31: gap nti aan an may g nau BuNoong 'I will teach you Mnong'

(CMLLA6)
an...git 'inform, let know'
jaq 'invite'
ngooi 'speak'
nti aan 'teach'

Mnong Rlm (= R)

R3a: a ngobi broq broq 'I am speaking slowly' (broq broq 'slowly')

R3b, c: a ngobi ta kan 'I spoke to/about him' (kan 'him')

R3f: a lah kan saak 'I said he went/ I said "He went"' (saak 'go')

R3g: (laai) a lah ta kan a saak 'I told him I was going'

R3i: a ngobi aan kan gt a saak 'id.'

R3j: a ndbom kan saak 'I invited him to go'

31


DAVID THOMAS

R3k: a aan kan saah T allowed him to go'
aan 'allow, give, let'
lah 'say'
laai 'past' Preferred in R3g.
nddom 'invite'
ngooi 'speak'
ta 'to, for, etc.'
Stieng ( = = S)
S3a: hey mor (dreet dreet) 'I speak (slowly)'
S3b: hey chhuor baak bu 'I spoke about him' (bu 'him')
S3c: hy lah a bu 'I spoke to him'
S3d: hy lah bu 'I scolded him'
S3e: hy chhudr lah bu han 'I said that he went/I said "He went" (han
'go')
S3f: hy lah bu han 'I said that he went'
S3g: hy lah a bu, lah hy han 'I told him that I was going'
S3i: hy lah aan bu gt (lah) hy han 'I spoke informing him that I
was going'
S3j: hy maa/sir bu han 'I invited him to go'
S3k: hy aan bu han 'I let him go'
S31: hy tti bu gt mor Soding 'I taught him to speak Stieng'
a 'to, for, from, etc.'
aan 'allow, let'
baak 'matter, concerning'
chhuor 'relate, tell'
gt 'know'
lah 'say, scold, tell'
maa 'command'
mor 'speak'
sir 'invite'
tti 'teach'
From the foregoing data one can immediately reconstruct an

intransitive talking clause (3a) as Proto-South-Bahnaric *SV(Adv.),
attested in all six daughter languages. The presence of an adverb or a
negative with this construction seems to be preferred.

For 'talking about' clauses (3b), all three attested languages have S-V-
Link-Content, but the type of Link varies from a preposition (Rlam) to a
generic noun (Stieng, Chrau) or a motion verb (Chrau). The first Chrau
form, with a verbal Link, may possibly be a caique on Vietnamese v, and
it is not attested in my data from the other languages. The second Chrau
form, with a nominal Link, is matched by Stieng; it could of course, also
be a caique on Vietnamese vic, but it seems to be a more general South-
East Asian pattern, and Stieng has been under less Vietnamese influence
than Chrau. The Rlam form is ambivalent for 3b and 3c. The wide
variation in South Bahnaric forms here might seem to indicate some

32


Clause types in Proto-South-Bahnaric

instability or ambivalence at the proto stage, probably not matching any
of the attested current forms.

Addressee clauses (3c in C, R, S) all have the structure S-V-Prep-
Addr., so that structure should be posited for PSB. For the preposition
Stieng and Rlam use their broad-spectrum prepositions a and ta, but
Chrau, lacking such a broad-spectrum preposition uses bay 'with'.
Perhaps a broad-spectrum preposition (*ta?) should be posited for PSB.

In Chrau, the communicative verbs have been split into at least four
classes. VI verbs, like aai 'speak', doom 'converse', are used in 3a, b, c,
etc., and require a Preposition before an Addressee, and require a Quote
Introducer before a Quote. V2 verbs, like chiih 'scold', lah 'scold', payoom
'praise', are used in 3d, h, and require a Quote Introducer before a Quote.
The V3 verb paa 'say' cannot take an Addressee, and it takes no Quote
Introducer since that would be homophonous with it. And paa replaced
PSB *lah as the Quote Introducer. V4 verbs, like sier 'invite', paa 'say,
invite', are used in 3j.

A transitive talking clause (3d in C, S) *S-V-0 should be reconstructed
for verbs like 'scold' or 'praise'.

There was apparently no distinction between direct and indirect
quotatives (3e, f) in Proto-South-Bahnaric. Form 3e S-V1-Quotlnt.-
Quot. is attested in Chrau and Stieng. The dropping of VI gives a simpler
form (3f in C, R, S) in which the Quote Introducer (C. paa, R, S. lah)
becomes the main verb *S-VQuot.-Quot. The verb lah 'speak, scold' is
attested in C, R, S and probably served as both a VQuot. and a Quotlnt.
in PSB, but its Quotlnt. function dropped out in Chrau. All these
languages have many verbs that can function as VI, but only one that can
function as Quotlnt.

Simple quotative addressee clauses (3g) in Chrau and Stieng have the
form S-V l-Prep.-Addr.-Quotlnt .-Quot., in which VI and Quotlnt. are
the same as in the direct and indirect quotatives (3e, f), and the Prep, is
that in 3c. Rlam has S-Vl-Addr.-Quot. The modern forms could be
accounted for by positing a PSB *S-Vl:v/VQuot.: /z/i-Prep.-Addr.-
Quotlnt.: lah-Quot. Stieng seems to have this structure. Mnong Rlam
dropped the redundant second lah. Chrau dropped lah completely out of
this construction by substituting paa in the Quotlnt. slot and by putting
lah into the V2 class, which does not occur in the 3g construction.

The 3h form, used with Chrau class V2 verbs, does not have a
Preposition before the Addressee. Data from other languages are lacking,
so no conclusions can be drawn for PSB.

The longer quotative addressee form (3i) has the form *S-V1 -aan-
Addr.-g/i-Quot., as attested in C, S, R, with aan...git 'let know' functioning
as a benefactive marker. In C there is an optional Quotlnt. before the
quotation; this may be a Chrau idiosyncrasy on the analogy of 3e.

Imperative clauses, with verbs like 'invite, command, permit' (3j) have
the same form as 3f, ie. S-V-Addr.-Quot., attested in C, S, R, so may be
reconstructed for PSB in that form. In Chrau, the 3f and 3j verbs are

33


DAVID THOMAS

homophonous in one instance, paan, yielding an ambiguous clause. The
causative verb aan 'give, allow' can also be used here (C, R, S).

Teaching clauses (3k) have the form S-V-aa-Addr.-g/r-Content
( = 3i) in C and P, perhaps reconstructable for PSB. Stieng lacks the
*aan. Only Bunar has S-V-Content-raa-Addr.

2. Existives

The simple existence identified existence transitive existence (existive
possession) naming becoming group of clause types take a variety of
forms, as may be seen from the following:

Mnong Bunr

B4b: geh ngih ta ri 'There is a house there' (MLC2A1) (ngih 'house',
ri 'there')

B4c: ta ti jng ngih 'Over there are houses' (MLC2A 1)

B4e: buum jng du ntiil ndd sa 'A tuber is a kind of food' (MLC32)

(buum 'tuber', du ntiil ndd sa 'one kind of food')
B4f: gop geh du play ngih 'I have a house' (MLC.2.1) (du play 'one cl.')
B4g: gop NDjreet 'I am Djrt' (MLCAA)
B4k: amoh gop NDaan 'My name is Dan' (MLCAA)
amoh 'name'
geh 'have'
jng 'be'
ta 'at'

Chrau

C4a: geh ni 'There are houses' (ni 'house')

C4b: geh ni u heeq 'There are houses here' (heeq 'this, here')

C4c: u heeq geh ni 'id.'

C4d: ni heeq 'This is a house'

C4e: heeq la ni'id.' (rare < Vietnamese)

C4f: a geh ni 'I have a house'

C4g: GaPe a 'I am GaPe'

C4h: a heeg GaPe 'id.'

C4j: a soq GaPe 'I am named GaPe'

C4k: sa£ a GaPe 'My name is GaPe'

C4m: a tanhya sag neh GaPe 'I named him GaPe' (neh 'him'

C4n: neh jng yaw 'He became a tiger' (yaw 'tiger')

C4o: a qo£ vadaai jng/luh i 'I made the lean-to into a house'

(vadaai 'lean-to')

geh 'be, have'

heeq 'here, this one'

jng 'become'

la 'is' (< Vietnamese)

luh 'appear, become'

oop 'make, do'

saq 'name, be named'

34


Clause types in Proto-South-Bahnaric

tan'hya

saq 'to name'
Koho Sre

K4a: gos hiu There are houses' (hiu 'house')

K4b: gos hiu teeng dg There are houses here' (KLC.9) (dg 'here'

K4e: chi dg la hiu 'This is a house' (KLC.26) (chi dg 'this')

khay lah/jng caw mih 'He is an American' (OSS. 189)
K4f: an gos kroac 'I have oranges' (KLC. 14a) (kroac 'orange')
K4h: chi dg kroac 'This is an orange' (KLC. 14)
K41: sondan an la K'Poh 'My name is K'Poh' (KLC.U)
K4n: khay jng/gos kliu 'He became a tiger' (OSS. 189-190) (khay

'he', kliw 'tiger')
gs 'be, have, become'
jng 'become'
la, lah 'is' (< Vietnamese)
sondan 'name'
teng 'at'

Mnong Preh

P4c: tarn bri geh ne 'In the forest there are rats' (CMLL.2S) (bri

'forest', ne 'rat'
P4e: gap jng BuNoong 'I am a Mnong' (CMLLA6)
P4f: gap geh pe nuyh koon 'I have three children' (CMLL.S) (pe nuyh

koon 'three children')
P4k: moh sak gap NDoong 'My name is Dong' (CMLL.l)
geh 'have'
jeng 'be'
moh sak 'name'
tarn 'in, at'

Mnong Rlam

R4a: geh hih 'There are houses' (hih 'house')

R4b: geh hih ta han 'There are houses there' (han 'there')

R4c: to car Mriik mau hih 'In America there are houses' (MLLL.36)

(car Mriik 'America')
?R4d: hih ho 'This is a house' (ho 'this')
R4e: ho eh jeng hih 'This is a house' (eh 'this')
R4f: an geh hih 'I have a house'
R4g: Jhang a 'I am Jhang'
R4i: a jng Jhang 'id.'
R4k: nan a Jhang 'My name is Jhang'
R4m: a nan to kan Jhang 'I named him Jhang'
R4o: a mhoq tm hn njng hih 'I made that shelter into a house'

(turn hn 'shelter')
geh 'have'
jng 'be'

35


DAVID THOMAS

mau 'be'

mhoq 'make, cause'

nan 'name'

njeeng 'become'

ta, to 'at'

Stieng

S4a: geh ni 'There are houses' (ni 'house')

S4b: geh ni a ny 'There are houses there' (ny 'there')

S4d: au ni 'This is a house' (au this, here')

S4f: hey geh ni 'I have a house'

S4g: hey Gape 'I am GaPe'

S4h: hey au GaPe 'id

S4j: hey chhak GaPe 'I am named GaPe'

S4k: chhak hey GaPe 'My name is GaPe'

S4m: hey chudl bu chhak GaPe 'I named him GaPe' (bu 'him')

S4o: hey loh nom (lh) bt ni 'I made the lean-to into a house' (nom

'lean-to')

a 'at'

but 'become'

chhak 'name'

chudl to name'

geh 'be, have'

loh 'appear'

loh 'make, cause'

A simple existence clause *ExistV-S (4a) may be posited for Proto-
South-Bahnaric on the evidence of C, K, R, S. And a PSB existence verb
*geh is attested by B, C, (K?), R, S.

A located existence clause ExistV-S-Prep.-Loc. (4b) is attested by B,
C, K, R, S with a demonstrative Loc. The locative preposition, however, is
different in all five languages, leading one to suspect that perhaps PSB
had no preposition there; but the need for a preposition (in a preposing
language group) became felt, perhaps to avoid ambiguity with a
homophonous N-Dem. noun phrase. Bunar, Chrau and Preh data (4c)
also include a transposed Prep.-Loc.-ExistV-S form, emphasising the
Location rather than the Subject, which should probably be posited for
PSB. Bunar has different ExistV in 4b and 4c.

The simplest identification clause (4d) is Ident.-S, attested in C, R(?),
or its reverse S-Ident. attested in S. It is not clear what should be
reconstructed for PSB.

A copula-linked identification clause (4e) S-Cop.-Ident. is found in B,
K, P, R. It is absent from Stieng and only borrowed in Chrau, the two
most reliably attested languages in the sample. (Caiques from Vietnamese
or English are very possible in the language lesson books, the sources for
most of the other language data.) The copula jeeng used in B, K, P, R,
however, is a verb of becoming in Chrau (4n) and Rlam (4o), so perhaps

36


Clause types in Proto-South-Bahnaric

PSB should be reconstructed without a copular identification clause form,
then the northern tier of daughter languages expanded the use of jeng
'become' to provide a copula.

A possession clause (4f) *S-Poss.-V-Item is clearly reconstructable
from B, C, K, P, R, S, and the Possessive Verb in all of them is the
existence verb *geh/gos of 4a, b. In PSB, apparently, existence and
possession were parts of a single semantic category; perhaps possession
should be viewed as transitive existence. (This use of have/be is paralleled
by Vietnamese co, Thai mil, Khmer miian, and many other South-East
Asian languages.)

The personal identification clause (4g) is like the simple identification
clause (4d), with Name3-S attested in C, R, and S-Name in B, S.
Reconstruction is not clear. A topicalized form *S-Dem.-Name (4h) is
more widely attested in C, K, S and should be reconstructed for PSB.
Rlam has also a copular form (4i) S-Cop.-Name, not to be reconstructed
for PSB (see 4e).

A name clause (4j, k) is semantically very close to the personal
identification clause. A form *S-NameN-Name (4j) with sak 'body,
name' as the Name Noun can be reconstructed from C and S. And an
alternate form *NameN-S-Name (4k) can be reconstructed from B, C, P,
R, S. The nan 'name' in R is probably a loan from Rade. Koho also has a
copular form (41) with the borrowed Vietnamese copula la.

The naming clause (4m) has different forms in C, R, and S. The history
of these is not clear.

A becoming clause (4n) *S-jeeng-0 is probably reconstructable from
C, K (see 4e). There is no outside support for the Koho variant S-gos-O,
which is homophonous with the possession clause (4f).

The transforming clause (4o) has the form *S-Vl-Forml-V2-Form2. C
and R have jeng as V2, probably reconstructable for PSB. C can also have
luh as V2, but S has luh as VI. PSB *luh 'go out' must have been part of the
PSB semantic field of transforming, but its syntactic use is not clear.

3. States

The ambient stative comparative superlative evaluative group of
clause types tends to be S-V in South Bahnaric languages, though V-S
state clauses are quite common.

Mnong Bunar

B5a: naar aao geh mih 'There will be rain today' (MLCA.2) (naar aao

'today', mih 'rain')
B5b: klaang book naar jeh 'It is noon already' (MLC.2.2) (klaang

book naar 'noon')
B5c: naar aao ji kat 'Today it is cold' (MLCA.2) (ji kat 'cold')
B5d: gop ji ngoot 'I'm hungry' {MLC.2.2) (ji ngoot 'hungry')
geh 'have'
jeh 'already, now'

3. Structurally 'name' is the head of the Subject noun phrase, but semantically T is still the
Subject, as in the two preceding forms.

37


DAVID THOMAS

Chrau

C5a: kd mi 'It is raining' (kd 'sky', mi 'rain')

C5b: nar tamvddp een 'It's noon now' (nar 'day', tamvbop 'middle')

C5c: nar heeq takat 'Today is cold' (heeq 'this', takat 'cold')

C5d: an takat 'I'm cold'

C5e: takat an 'I'm cold' (more common)

C5f: an takat an 'id.' (emphasising 'I')

C5g: an luh takat 'I became cold'

C5h: an takat doong neh 'I'm colder than him'

C5i: neh eq takat ka an 'He's not as cold as me'

C5j: an takat dodng leq 'I'm the coldest of all'

doong 'more than'

een 'already, now'

eq 'not'

ka 'like, as'

leq 'all'

luh 'become, appear'
Koho Sre

K5b: guul ngai rau 'It is noon already' (KLC A 1) (guul ngai 'noon')

K5c: ngai dg noat 'Today is cold' (KLC.38) (noat 'cold')

K5d: an kggp 'I'm sick' (KLC.22) (kggp 'sick')

K5g: khay golbh koop 'He became sick' (OSS. 190)

K5i: go ggq niam be chi dg 'They are not as good as this one'

(KLC.33) (niam 'good', go 'they')

K5j: chi dg buon rlau joh 'This one is cheapest' (KLC.33) (buon 'sell')

be 'as, like'

gblbh 'become'

ggq 'not'

rau 'already, now'

rlau joh 'most, superlatively'

Mnong Preh

P5a: bri mih 'it is raining' (CMLL.24) (bri 'jungle', mih 'rain')
P5c: naar aao duh (ngan) 'Today is (very) hot' (CMLL.23) (nar aao
'today', duh 'hot')

P5d: rpual prah joong 'A melon is long' (CMLL.29) (rpual prah

'melon', joong 'long'
P5h: gap kataang loon ma may 'I am stronger than you' (CMLL.23)

(kataang 'strong')
loon ma 'more than'

Mnong Rlam

R5a: mih 'It is raining' (mih 'rain')

R5c: naar o kokat 'Today is cold' (naar o 'today', kokat 'cold')
R5d: an kokat 'I'm cold'

38


Clause types in Proto-South-Bahnaric

R5e: kokat a 'id.'
R5g: a jng kokat T became cold'
R5h: an kokat hin ta kan 'I'm colder than him'
R5i: kan han ay so kokat blah a 'He is not as cold as me'
R5j: a kokat hin ta leq mot nih 'I am the coldest of all'
ay so 'not'
blah 'as, like'
han 'not' (?)
hin ta 'more than'
jeeng 'become'
leq mot
nih 'all, everyone'
so 'see, perceive' (?)
Stieng
S5a: mi 'It is raining' (mi 'rain')
mi loh 'id.'
S5c: 'lk 'It is cold' (leek 'cold')
nar 'leek 'id.' (nar 'day, sun')
S5d: hey 'leek 'I am cold'
S5e: 'leek, hey aq\ 'Cold, indeed I am!'
S5f: hey 'leek hey 'I am cold'
S5g: hey loh 'leek 'I became cold'
S5h: hey 'leek huos bu 'am colder than him'
hey teq a bu 'leek hey 'I am colder than him/Beside him I'm cold'
S5j: let pal ney, 'lk hy 'I am the coldest of all/Of all of them I'm
coldest'
stative = S + P:Vi (OSG.9)
aq 'exclamation'
but 'become'
huos 'than, more than'
leet pal
ney 'completely, all of them'
loh 'appear, become'
teq a 'place beside, compare with'

There is no agreement on the form of the simple ambient clause (5a).
Koho, Rlam, and Stieng have a simple Amb. structure. Chrau and Preh use
a dummy subject DumS-Amb. structure. Stieng can use a dummy verb
Amb.-DumV structure. And Bunr uses a geh-Amb. existive (4a) structure.
This variety could perhaps be explained by positing a PSB simple *Amb. as
in K, R, S. Chrau (which has been in close geographical proximity to
Vietnamese) and Preh adopted a Vietnamese-like structure, treating the
Ambient as an intransitive verb. The Bunr form and the alternate Stieng
form treat the Ambient as a noun. Positing a simple proto *Ambient would
most easily account for the verb and noun developments. Nominal use of

39


DAVID THOMAS

the Ambient does appear also in Chrau in the form mi sa neh rain-eat-him
'he was heavily rained on'.

A time clause *Time-Adv. form (5b) can be reconstructed on the
evidence of B, C, K.

A time-located ambient *Time-Amb. form (5c) is reconstructable from
B, C, K, P, R.

A stative *S-State (5d), which is similar in form to the intransitive (la),
is reconstructable from B, C, K, P, R, S.

A reversed stative form *State-S (5e), emphasising the state, is
reconstructable from C, R, S. The Stieng form seems to be more
emphatic than the Chrau, and is normally accompanied by an emphatic
final particle. An echo form *S-State-S (5f) is reconstructable from C, S,
giving mild emphasis to the Subject.

An inceptive state *S-IncepV-State (5g) is reconstructable from C, K, R,
S. The Inceptive Verb is luh in C, S Jeeng in R, and golds (from gos ?) in K.
These verbs have other PSB functions in 4a, e, n, o, and it is not clear which
of these verbs should be reconstructed as the PSB Inceptive Verb.

A comparative state *Sl-State-CompMk.-S2 (5h) is reconstructable
from C, P, R, S. But the Comparison Marker is different in each language,
so no conclusions can be drawn concerning the PSB Comparison Marker.
Stieng also has a form Sl-CompMk.-S2-State-Sl.

For the negative comparison (5i) Chrau, Koho, and Rlam have Sl-
Neg.-State-CompMk.-S2, which can probably be reconstructed for PSB.

The superlative (5j) has the form S-State-CompMk.-All, paralleling
5h, in Chrau and Rlam. In Koho the form is S-State-Superl. In Stieng
the form is All-State-S. Reconstruction is not clear.

4. Summary of reconstructions

The reconstructed Proto-South-Bahnaric forms may be summarised, with
sample glosses, as below. Binomial slot : filler formulations are given
when both the functional slot and the actual filler were discussed.

Communicatees

3a: *S-V-(Adv.) 'I speak (slowly)'
3b: *S-V-Link:?-Content 'I spoke about him'
3c: *S-V-Prep.:ta-Addr. 'I spoke to him'
3d: *S-V-0 'I scolded him'
3e: *S-Vl-QuotInt.:/z/z-Quot. 'I spoke saying he went'
3f: *S-VQuot.-Quot. 'I said he went'
3g: *S-V 1: v/VQuot. :/a/z-Prep.-Addr.-QuotInt.-Quot. 'I told him
that I was going'
3h: ??
3i, k?: *S-Vl-BenV:flflfl-Addr.-BenV:g/-Quot. 'I let him know that I
was going'
3j: *S-V:v4/flw-Addr.-Quot. 'I invited him to go'
Existives
4a: *ExistV:ge/z-S 'There are houses'

40


Clause types in Proto-South-Bahnaric

4b: *ExistV:ge/z-S-(Prep.:?)-Loc. 'There are houses there'

4c: *Prep.-Loc.-ExistV-S 'Over there are houses'

4d: ?*Ident.-S / ?*S-Ident.

4e: (PNSB *S-Cop.:yee/7£-Ident. 'That is a house')

4f: *S-PossV:ge/z-Item 'I have a house'

4g: ??

4h: *S-Dem-Name 'I here am GaPe'
4i:

4j: *S-NameN:^/:-Name 'I am named GaPe'

4k: *NameN:^fl/:-S-Name 'My name is GaPe'
41:

4m: ??

4n: *S-V:jng-0 'He became a tiger'

4o: * S-V 1-Form l-V2-Form2 'He made the lean-to into a house'
Statives

5a: *Amb. 'It is raining'

5b: *Time-Adv. 'It is noon now'

5c: *Time-Amb. 'Yesterday it rained'

5d: *S-State 'I am cold'

5e: *State-S 'I am cold'

5f: *S-State-S 'I am cold'

5g: *S-IncepV:?-State 'I became cold'

5h: * S1 -State-CompMk:?-S2

5i: *S1-Neg-State-CompMk-S2 'I am not as cold as he'

5k: ??

REFERENCES

Evans, Helen &
Peggy Bowen

Haupers, R. &
Bieu 'Bi

Manley, T. M.

Miller, Vera G.

Phillips, R. L.

Phillips, R. L. &
Y Kem Kpor

Tang Hmok, Y.

Thomas, D. D.

Thomas, D. D. &
Headley, R. K.

Thomas, Dorothy

41

n.d. (c. 1965) Koho language course. Dalat: Christian & Missionary
Alliance mimeo.

n.d. (c. 1970). Stieng phrase book. Saigon: Summer Inst. Ling.

1972. Outline of Sre structure (Oceanic Ling. Spec. Publ. 12). Honolulu:
Univ. Hawaii Press.

1976. An overview of Stieng grammar. Grand Forks: Summer Inst. Ling.

1963. Mnong language course, (ms.) microfiche Summer Inst. Ling.

1974. Central Mnong language lessons. Saigon: Summer Inst. Ling. &
Min. Educ.

1976. Mnong Lam language lessons. Summer Inst. Ling.

1971. Chrau grammar. (Oceanic Ling. Spec. Publ. 7.). Honolulu: Univ.
Hawaii Press.

1983. An invitation to grammar. Bangkok: Mahidol Univ.
1970. More on Mon-Khmer subgroupings. Lingua 25, 398-418.

M. 1969. Chrau affixes. Mon-Khmer Stud. 3, 90-107. Saigon: Summer Inst.
Ling.




A LINGUISTIC ANALYSIS OF SOME SOUTH
MUNDA KINSHIP TERMS, I

Norman H. Zide & Arlene R. K. Zide

Part 1

1. In this paper we present etymologies for a number of South Munda
kinship terms.1 South Munda (SM), one branch of Proto-Munda (PM),
branches into Kharia-Juang (KJ) or Central Munda, and Koraput
Munda (KM); KM branches into Sora-Juray-Gorum (SJG) and Gutob-
Remo-Gta? (GRG). While we have not done anthropological analyses of
the Munda kinship systems, we have made use of the anthropological
sources in defining and relating kin terms and kin-term sets of the
languages and proto-languages.

Our objective has been to provide a linguistic analysis of the SM kin-
terms, with a view to reconstructing as much of the original (i.e. non-
borrowed) terminology as possible, and to integrate the results into as
coherent and plausible a system as we can. Since KM and SM noun
morphology have not been analysed and described, we will present a
description of SM noun morphology with particular reference to the
derivational morphology of full forms (FF), and combining or
compositional forms (CF) of Munda kin terms. This will come in the
second paper in this series, along with the first full sets of kin-term etyma.

In the first section of this paper we take up and criticise the work of
Bhattacharya (1970) and Parkin (1985) on Munda kin terms, and in the
second section we discuss how relationship and reciprocity are shown in
Munda kin terms.

We have taken our data, the SM kin terms and definitions, from a wide
range of existing sources, and from our own fieldnotes. The anthro-
pological sources are fuller in the coverage of the terms and their uses, but
are poorer in linguistic transcription, and lack morphological analysis.
The linguistic sources provide better linguistic data, but are incomplete
and, for purposes of kinship term analysis, poorly organised. A few
publications (e.g. Deeney, 1975, on Ho) are exceptional in presenting the
kinship system and the kin terms fully and perspicuously in linguistically
well-analysed form.

1. We use the following abbreviations in describing kin terms: M, 'mother'; F, 'father'; P,
'parent'; Br, 'brother'; Si, 'sister'; Sib, 'sibling'; Y, 'younger'; O, 'older'; Hu, 'husband';
Wi, 'wife'; Sp, 'spouse'; So, 'son'; Da, 'daughter'; Ch, 'child'.

The abbreviations for the modern language names are: Sa, 'Santali'; Mu, 'Mundari';
Kher, 'Kherwarian'; Kh, 'Kharia'; Ju, 'Juang'; So, 'Sora'; Go, 'Gorum'; Jr, 'Juray'; Gu,
'Gutob'; Re, 'Remo', and Ga, 'Gta7'. The more common ethnonyms for the KM
languages are: Sora: Saora, Savara; Juray: Juray Sora; Gorum: Pareng, Parenga, Parengi;
Gutob: Gad(a)ba, Gutob Gadba; Remo: Bonda, Bondo; Gta?: Didayi, Didei, Dire.

43


NORMAN H. ZDE & ARLENE R. K. ZTDE

There has been widespread borrowing of kinship terms into SM: from
languages as diverse as the Dravidian Ollari Gadba, and Indo-Aryan
Kotia Oriya, standard Oriya, and standard and dialectal Hindi, Bhojpuri
and Bengali, and English. Undoubtedly earlier borrowings, particularly
from Dravidian, have been missed by us. We hope, for a later paper in this
series, to collaborate with a Dravidianist on Dravidian borrowings, early
and late, of both kinship behaviour and kinship words. This paper does
not deal with kinship terms that are not genetically old in Munda, and
that are not likely to have cognates in Mon-Khmer (MK). We propose to
present the full set of Munda kin terms with Mon-Khmer (and, possibly,
Austronesian) cognates in a later paper.

The only published general treatments of Munda kin terms are those of
Bhattacharya (1970) and Parkin (1985), and the relevant sections of
Pinnow's Ver such (1959), as updated by him in 1960 in his unpublished
monograph on Juang. We show that Bhattacharya's weighting of
semantic similarity at the expense of sound correspondences leads him
to lump together forms that are semantically very close, but not cognate.2
Parkin, an anthropologist who has written a dissertation on Austro-
asiatic kinship, needs to use linguistic evidence. He makes use of
Bhattacharya's material, but is sometimes misled by Bhattacharya's
methods of analysis and presentation of data. Parkin also takes rough
phonetic similarity as indication of genetic relationship, and uses the
usually spurious'cognition' to support anthropological arguments
about kinship. He sometimes ignores Bhattacharya's conclusions (e.g.
about bare (Bhattacharya's Set 66. (1970: 455), Ga bare; ZZ *boHre,3 'a
woman's brother') and mistakenly connects reflexes of this *boHre (e.g.
bo?re, bok're) with *boko, YSib. By doing this he misses the importance of
PM *boHre and the existence in PM of terms restricted to male or female
speakers, e.g. GRG *bo(b)re, 'a woman's brother', *bVlon, 'a woman's
sister', *bVyai] 'a man's brother', and *tonan, 'a man's sister'. We need
the anthropologist to make kinship sense of the meanings of cognate sets
where we cannot reconstruct a properly precise meaning for the PM term,
much less account for the change in meaning in NM and SM and the
modern languages. The example of Northern Munda (NM) *hili, OBrWi,
and KM *hVli SpYBr will be discussed in some detail later in the paper.
Bhattacharya has confused the issues by putting NM *hili and KM *hVli
into different semantic-cognate sets because their meanings are not close
enough.

We should note that anthropological studies of Munda kinship have
flourished in the past fifteen years; we are thinking of the work of S.
Bouez, Deeney, Parkin, Pfeffer and Vitebsky. However, most of these

2. He implies that they are sufficiently cognate for his purposes.

3. ZZ = Zide & Zide; our reconstructions, e.g. PM *boHrs, or, better, *b(4)H(b)rs, differ
in general reliability from proto to proto. SJG reconstructions are more solid than KJ
reconstructions. Tn the *boHre example, the first reconstruction can be considered to be
an abbreviation for the second, which shows more questions and possible answers. Since
the reconstructions given here will not be evaluated and justified, they should be
considered abbreviations of a fuller treatment of the comparative phonology and lexicon.

44


South Munda kinship terms, I

analyses do not bear directly on the linguistic problems with which we are
concerned, and no further reference will be made to most of these.

Bhattacharya's survey paper is very useful in that it presents and begins
to organise his own rich field data. He is less thorough in abstracting the
published literature. His analyses, his semantic-cognate sets, observations
on borrowings, etc., are usually perceptive. In some cases he is more
conservative than he needed to be, e.g. in putting Ho haam4 'old man,
husband' in one set, and the reflexes of GRG *hV-n-dam (e.g. Ga handa,
etc.) in another. Had he noted Mundari haram (same meaning as the Ho
and the GRG), he would have been compelled to put all these forms in
one semantic-cognate set. In other etyma, particularly where he brings in
possible cognates in Mon-Khmer, he is too indiscriminately accepting.
(But then, he is the only Indian Munda scholar who has paid any
attention to Mon-Khmer.) Bhattacharya knows several of the Munda
languages, and he knows Indo-Aryan and some Dravidian. Parkin, when
he disregards Bhattacharya, usually goes wrong.

Parkin is conscientious in examining long lists of kinship terms, and he
turns up interesting forms not earlier appreciated (e.g., Remo N-kwi, YSi,
which Bhattacharya also records). He can be perceptive about borrowings,
e.g. noting that KM mama, MBr (in some languages SpF), is borrowed
(along with kin behaviour) from Dravidian, and not directly from Indo-
Aryan. But his use of rough phonetic similarity as evidence of cognation is
self-defeating. We do not quarrel with Parkin's anthropology. It is true, as
Parkin says (1985: 705) that the absence of studies of comparative Munda
kinship constitutes 'a major gap in south Asian studies', and that Parkin
has done a considerable amount of serious work in filling that gap. We
give three examples of the sorts of things we object to:

(1) his Table of 'Basic forms of typical NM and SM terminologies';

(2) boko and bare; and

(3) erjga, 'mother'.

After this we take up the confused (mostly by Bhattacharya) case of Juang
ini-bo9 HuYBr.

In his Table, Parkin gives 'typical' NM and SM terminologies, and
includes the terms for SpF, SpM, ChCh, same sex SibCh (of the same sex
as Ego), and FSibCh of same sex as Ego, opposite sex SibCh (of opposite
sex to Ego) and of PSibCh, BrWi, HuSib, SiHu and WiSib, etc. What is a
'typical' terminology, and how and why does one compare typical
terminologies? 'Typical' would seem to mean 'well-described', or 'well-
known'. The NM and SM forms given in the table are of little value. To a
linguist, the forms to be compared would be the reconstructed NM and
SM forms. These turn out to be closer to each other than Parkin's typical
forms from typical modern languageswhen they are cognate. NM

4. We havewithout marking these formsretranscribed some of Bhattacharya's
transcriptions in order to indicate morpheme boundaries, and to prune dubious and
excessive phonetic detail. We will give Bhattacharya's transcriptions and comment on
them in the full data sets in later papers in this series. The form transcribed by us haam
was transcribed ham by Bhattacharya, and ham by Deeney.

45


NORMAN H. ZDE & ARLENE R. K. ZTDE

*kankar and NM *kinkar, SpM, are fairly similar whereas Parkin's Mu
hanar and (KM ?) *kinar are less soperhaps because Korku, Kharia
and Juang, which reserve PM *k, are not typical enough. Parkin's jai
'grandchild' (in Mundari-Ho) was selected as the representative NM form
because he wanted to connect it with jia, 'grandmother'. The connection
is dubious. A better word for NM 'grandchild' than jai, found only in the
Kherwarian branch of NM, would be kVrar/kVrar, found in both
branches (Korku kurar, Santali korar). The data can be found in
Bhattacharya.

2. Both in the table and elsewhere Parkin notes the important distinction
between sibling terms where one must know whether the speaker is of the
same or opposite sex to the kin-term referent. What seems importantto
generalise the caseis not same or opposite sex, but whether the set of
terms is sensitive to speaker's and referent's sex. That is, we have in KM a
set of four terms where this feature is marked ('a woman's brother', 'a
woman's sister', 'a man's brother', 'a man's sister'). This set contrasts
with another set of terms where this feature is absent, but where we mark
relative age of the speaker and referent, i.e. OBr, YBr, OSi, and YSi.
Parkin's boko belongs to this second set (and perhaps should be defined
not YBr, but YSib) and the others of the sibling terms in his table tonan,
bokre (from *boHre, Bhattacharya's bare set) and misi belong in the first
set. It is possible (as the NM data suggest) that there were only two sex-
sensitive terms in PM in this first set, and these were both opposite-
marking terms, but this needs to be demonstrated. Parkin takes bokre as
cognate with boko, and not with bare, *boHre which leads him to miss the
one cognate in PM of the sex-sensitive set. Since such terms go back to
PM (however many may have to be reconstructed), we want to know what
the distinctive functions of these two sets of sibling terms werein earlier
times and protosand what they are now. None of the anthropologists
we have read takes up this important problem.

If one wants to use linguistic evidence in kinship arguments, then it is
necessary to be able to recognise the historical depth of an etymology.
*boHre can be reconstructed for PM; this is noteworthy. It is possible that
*boHre at some more distant levelperhaps PAAcan be shown to have
a morpheme in common with *boko, but good MK evidence of this would
be needed to make such a case. One has to be able to recognise that
*V?(g)-le is a good KM etymology for 'grandchild', but not a good SM
or PM etymology.

3. Lastly, Parkin's treatment of'mother'. Parkin writes that 'the standard
NM and KM term for "mother" is eijga... '. eyga is only standard in one
branch of NM, Kherwarian. If one wants an etymological formula for
PM 'mother', it would be V-ya-N, V-yo-N. Korku has ayom in one
dialect, the more archaic ayaij in others; the reduced form of this before
-te? is an-; Santali has eyga and ayo/ayo; Mundari-Ho has eyga, but

46


South Munda kinship terms, I

e(y)ay in the vocative (we do not find much morphological alternation of
this sort in Munda); Juang has bwi-N, Kharia has ayo/ayo-N, Sora has
yaij, Juray ayay and yoy, Gorum has yay, Remo yoy, and iyoy, and Gta?
yay. The basic form is the V-ya-N/V-yo-N. The question is how to relate
eyga and e(y)ay (vocative) in Ho-Mundari. Deeney has anticipated us in
pointing out (Bhattacharya and Pinnow have made partly similar
observations) that the vocative e(y)ay resembles eyga-n/eya-n, i.e. the
non-vocative stem with the first person singular pronoun i.e. 'my
mother'. This suggests that most of the Munda forms for 'mother' were
originally vocatives with first person pronominal enclitic (of possession)
-n/-y, and that eyga represents (how accurately?) the basic form of
'mother'. How we get from eyga to eya/ena where these forms precede the
enclitic -n-y has just been shown in Deeney's data. One could suggest
metathesis, common in Munda, but we have no (other) cases of eyay/eyya,
y/n, metathesis. We reconstruct V-ya which becomes V-na before final
nasal in the vocative. To take eyga, ayo, eay, etc. as obviously cognate is
risky (it was not wrong). In any case, 'eyga is not the standard form of
"mother" in NM and KM'. The point of these criticisms is not that
anthropologists should reconstruct linguistic proto-forms, but that they
should be less free in identifying putative cognates, and using these
largely spuriouscognates to support other arguments.

ini-bo or ini-bau/-bou. Our first comment on Parkin's rejection of
Bhattacharya's connecting ini-bo and hVli was that Bhattacharya's
judgment on cognation was better than Parkin's, and whether the
semantics agreed with Parkin or not, ini and h Vli were almost certainly
cognate. A re-examination of Bhattacharya's sets 81. and 83. (1970: 457)
shows the situation to be more complicated. Bhattacharya misleads his
readers by setting up two semantic-cognate sets, and putting NM *hili in
the second (with Juang kuli), and KM *hVli in the first. (Bhattacharya
does not think with or use reconstructed form; we have constructed
'Bhattacharya's reconstructions' for him.)

The facts are these: Bhattacharya has made two semantic-cognate sets,
81. and 83., these coming in his section of 'Terms for Brother, Sister,
Brother-in-Law, Sister-in-law', one of his more complicatedand
importantsets of terms. In 83. he puts Juang kuli/koli, OBrWi, and
Kherwarian (there is no Korku cognate), *hili, OBrWi. In 81. there are
three sets of words:5 (a) KM *hVli(-boj) (ZZ reconstruction), Vli-boi,
HuYSi, SpYSi; (b) KM *Vrel/*Vrer, HuYBr, SpYBr, and (c) NM *erwel/
*Vrwel, HuYBr. Bhattacharya does not sort these into sets, so that it is
not clear if he thinks Juang ini-bo goes in set (a), or with sets (b) and (c).
The latter two sets clearly reconstruct to PM *Vrwel.6 The meanings of
the 83. forms are fairly close: (a) HuYSi, SPYSi, and (b) (c) HuYBr,
SpYBr. ini-bo means HuYBr. Since 83. (a) is clearly cognate with 81.
which includes Juang kuli, to relate ini to the augmented 83. we either
have to disqualify and reject kuli, somehow to relate kuli and ini, or to
connect ini with 81. Can weas Bhattacharya's array of data suggests

5. See the Appendix for fuller presentation of the data.

6. The ? in Gta? wrwe7 requires some discussion.

47


NORMAN H. ZDE & ARLENE R. K. ZTDE

show ini to be a cognate of PM *Vrwel, HuBYr, SpYBr? We need to show
that the sound correspondences are possible: that Juang n corresponds
with r (or something like r)\ this is possible; that the first vowel i
corresponds to V, where the following, stressed, vowel is *e and i in Juang:
this is possible; the correspondence of Juang stressed i and PM *e: this is
possible; and that a final consonant, Juang /, can be lost compound-
medially: this also fits into the set of expectable, regular, correspondences.
We accept, tentatively, ini as cognate with PM *Vrwel, and rearrange
Bhattacharya's sets as follows: 81. (6), (c) plus ini, HuYBr, SpYBr, and
83. plus 81 (a), *kuli(-boi), OBrWi, HuYSi, SpYSi.

Part II

This second section takes up:

(1) (once more) the two different sets of sibling terms in a number of the
Munda languages;

(2) the old system of pronominal enclitics (Pro,p) marking inalienable
possession. The kin terms of Kherwarian that do not take Proip are:
(i) name-like kin terms, and (ii) conjoined pairs of terms whose
referents are related to each other, not to the speaker or some other
named or pronominally indicated person. The latter, paired
reciprocal terms, are common in Munda, both north and south.
Reciprocal infixes in kin terms (the infix is the same one found as the
verbal reciprocal marker) are commonly used where each of the
referents of a hypothetical pair refers to the other by the same kin
term. The various meanings of this reciprocal infix, NM -/?-, SM -m-,
are examined in several Munda languages.

(3) reciprocal (kinship) terms of address are examined in Santali, Juang
and Sora.

(4) Some kin-term affixes that look like but are not reciprocal -p-j-m- are
described, and traced to PAA.

We noted earlier that in some Munda languagesRemo is the best
example of one with two complete setsthere are two different sets of
sibling terms, one marking relative age (e.g. 'younger brother', 'older
sister'), and the other marking 'relative sex' of speaker and referent (i.e. 'a
man's brother' versus 'a woman's brother'). Relative age is commonly
marked in the Indian area; relative sex is not. So far as we know, none of
the anthropologists who have worked on Munda groups with two
(complete or incomplete) sets of sibling terms has described the distinctive
functions of these sets. Parkin does note these terms, but in his table he
distinguishes between same-sex (of speaker and referent) and opposite-sex
terms, whereas we see the basic difference as between sex-marked (same
or different sex) terms, and age-marked terms. The GRG languages have
two full (four-term) sets, but it is not clear that cognate forms of the sex-
marked set in Sora (e.g. GRG *bVyaq, So boaij, GRG *tonan, So tonan,
these meaning 'a man's brother' and 'a man's sister' respectively in GRG)

48


South Munda kinship terms, I

have the same meanings and sort into similar sets in Sora. In Kherwarian
the sex-marked set has only two terms, both marking opposite sex kin.
The history of these sex-marked terms is still largely obscure. But in any
case the distinction is noteworthy, and an understanding of the functions
of the two sets should be important to the anthropologist studying
Munda kinship.

When studying the Munda languages that preserve the old system of
pronominal enclitics for marking inalienable possession (e.g. Ho, Santali,
Juang, Kharia), we find different definitions of grammatical inalien-
ability. The minimal set of inalienably possessible nouns would be the kin
terms, the next minimal set would add words for body parts. The
pronominal enclitics, Pro,p, are required with almost all inalienably
possessible nouns, with a few exceptions. The common exceptions are: (a)
'name-like' terms; and (b) paired reciprocal terms where the relationship
obtains between the referents of the pair. These paired reciprocal kin
terms are common in Munda.

One Mundari example from Hoffmann (1950: 1303) of a name-like
term has to do with the speech of siblings (i.e. those who have a common
referent for 'mother' and 'father'). In this contextsiblings speaking to
each otherone cannot say 'my mother' without the (erroneous and
offensive) implication that she is not also 'your mother'. What is said
and is preferable to using a Proip for first person dual inclusiveis simply
'mother' (perhaps 'Mother' would better represent this), i.e. the form
erjga without a Proip.

Paired reciprocals of one form or another are characteristic of several of
the Munda languages, North and South. The kinds of pairs found are:

(i) the senior term occurs followed by the 'collective' suffix -ya/-ea; in
Santali, e.g. hili-ya (hili, OBrWi); we discuss the meaning of such
forms below;

(ii) the senior term of the reciprocal relationship (if there is one) is
mentioned followed by an echo word; in Santali, e.g. hili + hali; we
discuss the meaning of these forms below;

(iii) both members of the pair are mentioned, the senior one first; in
Gta9, e.g. nta?+ 2dglx 'grandfather and grandchild'; this means (we,
they etc.) are grandfather and grandchild. Whether, as for similar
forms of Santali (i and ii above), the compound can also mean the
relationship (in this case the grandfather-grandchild relationship) is
questionable.

(iv) the construction with nu- plus (usually junior) kin term in Remo;
e.g. nu-girirj 'I and my wife's younger brother', and probably also
'we two are OSiHu and WiYBr';

(v) where both members of the (hypothetical) pair would refer to each
other by the same term, e.g. GRG *bVyai], 'a man's brother'; in the
GRG languages, e.g. GRG *b-Vm-lon (*bVlon, 'a woman's sister')

49


NORMAN H. ZDE & ARLENE R. K. ZTDE

'each other's sisters', Gutob bumulon, Remo bunlu, Gta9 bumluy.7
The -m- infix (which becomes -n- in Remo) marks this relationship.
There are two seemingly irregular cases in Remo: t-un-una, and g-
Vn-riy (/gindriy/), where tuna is 'a man's sister' (younger sister
according to Bhattacharya) and girirj is WiYBr. These have the same
meanings as the forms nu-tuna and nu-giriy, i.e. T and my younger
sister'; 'we are (older) brother and (younger) sister' and the girirj
form with -n- would be glossed the way the nu- form was. These are
reciprocal pairs, but unless Remo has a common term of reference
(or of address?) for brother and sister, and for OSiHu and WiYBr,
these are not terms for referents who refer to each other by a
common kin term, and we conclude that Remo -n- has been
extended to a few such pairs.

The first definition (in the third volume of Bodding's Santal dictionary
(1935: 116)) of hili + hali is 'a man's wife and younger brothers', then
(that) 'relationship', and then, more literally, 'a hili and those who call her
so', i.e. her erwel(s), HuYBr. That is, hili+hali means 'the hili-erwel
relationship'; hili-and-erwel: the erwel(sy hili, and the hilfs erwel(s). Some
echo words can be defined as 'what goes along with (the preceding N1,
which the echo word echoes)' or N1 'and the like'. What goes with hili,
OBrWi, here is its reciprocal. We find echo words of kin terms used
comparatively infrequently in this way, but the synonymous construction
with -yaj-ea (Bodding's 'collective') is more common, (hili-ya, according
to Bodding, means exactly what hili + hali means.) Bodding in his
Materials (1929: II 41-2) lists more than twenty such kin-term pairs (or
collectives), most of them in common use.

So far as we know, such reciprocals are not common elsewhere in the
Indian area. There are in Indo-Aryan and Dravidian pairs like Hindi
maa + baap 'mother and father' (in more formal speech maataa+pitaa),
and bhaaii + bahan, 'brother and sister', but not twenty other pairs of kin-
term reciprocals in common use.

The other examples of SM -m- are found in Juang, and of NM *-p- in
North Munda fairly widely (we discuss forms in Ho, Santali and Korku).
Pinnow gives two Juang examples of -m-\ ssmelay (from selay, 'grown-up
girl'), 'young woman, woman', and komoygsr (from koyger, 'young man,
youth'), 'master, lord, husband'. In the first case, the affixed form is
slightly more general; in the second the infixed form is a kind of honorific.
The only feature of the reciprocal evident in the second example is its use
of plurality to mark the honorific, something quite common in the Indian
area.

Hothe third of Deeney's small 1975 dictionary that we sampledhas
few examples of -p- in kin terms. It is interesting that the two we found
have precisely the meanings of the Juang forms but are not cognate with

7 A common Indian areal construction repeats the noun (kin term) for this reciprocal
'distributive'meaning. Thus, Hindi ham bhaaii bhaaii haT. (bhaaii, 'brother') 'We are
(like) brothers', 'buddy buddy'. Similar constructions have been noted in some of the
Munda languages.

50


South Munda kinship terms, I

them: h-Vp-anum 'young woman' (there is no hanum), and s-Vp-eed
(Mundari sepered) 'young man'. There is no free seed, but it occurs bound
in the 'nephew' term hon-seed (NM *kon-sered, *kon-serej, FBrSo).

The Santali exampleswe used as a sample the letter H in the third
volume of Bodding's dictionary (1935: 1-184)are hapamun, hapam
'grown-up girl' (there is no hamun or ham); hapram-ko/haprum-ko
'ancestors, forefathers, collection of old men' (haram, harum 'an old
man'); hepsl, along with herel, 'man, male' (no el); and hopon along with
hon. He translates hopon 'offspring, young, child, son daughter'; adj.
'small, little', and hon 'a son, child', now used only in compounds.

For Korku, N. Zide recorded along with kon, kon-jei, kurar, kosered¡
koserej and some others, kopon (with the dual or plural), koponjei, kuparar,
kopsered¡kopserej. The simple forms were much more common. It never
became clear what the meanings of the dual and plural -p- infixed forms
were. The meaning of -p- in NM is still obscure; the plural aspect of the
'reciprocal' (and its development into an honorific), and the notion of a set
or collection can be seen in some of the infixed forms. The forms with
-m-j-p- for 'young girl' and 'young man' in Juang, Santali and Ho have
been noted, but what the function of the reciprocal is here is unknown.

Of mutual appellatives (Bodding's term, i.e. reciprocal terms of
addressin our examples, of persons not of the same generation),
Parkin notes Santali gorom, which Bodding translates 'namesake', used by
grandfather and grandson as terms of address. Parkin also notes the
existence of such forms in Juang. McDougal (1963: 141) writing of
generation sets and the extension of kin terms, says that 'any male of an
odd-numbered ascending or descending generation, regardless of kin
type, may be addressed with the term for "father", and any female with
the term for "mother"... Any male of Ego's own or any even-numbered
generation may be addressed with the term for "brother", any female with
the term of "sister". For example, a man may address his brother, father's
father, and son's son with the term for "brother"...'. In Vitebsky's as yet
unpublished notes on Sora kinship there are also examples of cross-
generation kin addressing each other by the same term. The term tata,
FEBr, is used in address reciprocally, i.e. by YBrSo. Similarly entalaij,
MOSi, can be used reciprocally in address by YoSiDa.

It is likely that more Munda languages have mutual appellatives, but
that they have not been recorded.

There are two more possible infixes found with kin terms: the -m- in
NM k-Vm-on 'nephew, niece', from the simplex kon, 'child', also attested
in SM: Sora amon-sij 'nephew', and amon-sil, 'niece'. This is the only
example of this PM *-m-. G. Diffloth tells us that there is a cognate affix,
-mold but unproductive in Mon-Khmer, with cognate forms in a
number of MK families (Khmer, Bahnaric) meaning 'child' for the
simplex, and 'nephew/niece' for the infixed form.

The other case of a kin term that might be analysed as containing an
infix is PM *kin- or kin-, found only in *kinkar/kinkar 'mother-in-law'

51


NORMAN H. ZDE & ARLENE R. K. ZTDE

(and perhaps Juang kan-dae, 'old woman'), the kin having a cognate in
MK meaning 'woman'. There is another kin term in Munda, kimin,
'SoWi, YBrWi', which might be related morphologically at an early stage
to *kin. Diffloth informed us of an MK -mp- infix, that would account
for this derivation. If we take the *-mp- as going to *mh- in PM, this helps
clarify a previously baffling set of correspondences. We find *kVmin
everywhere in Munda except in SJG, where the cognatesif they are
cognatesare unexpected: Sora koyen, (Vi: koin, kaon), Juray kaun,
[kawun], and Gorum konun for YBrWi. We reconstruct *k-Vmh-in/k-
Vmh-in, and begin to understand how these SJG forms (and presumably
others) developed.

It is difficult to say what in the Munda kinship (and kinship term)
system comes from Austroasiatic and what does not. Of the topics we have
discussed, those peculiar to Munda (in the Indian linguistic area) are: the
sex-marked sibling terms; the use of cross-generational reciprocal terms of
address with some frequency (fuller data on more Munda groups would
probably show more evidence of this); the use of paired reciprocal terms
commonly; and the various infixes and their uses. Indo-Aryan and
Dravidian do have age-marked terms, as has Mon-Khmer. Munda has
some, but a number of the distinctions and terms look like borrowings. It
would take a closer examination of Munda and Mon-Khmer kinship to
speak with authority on this. That age distinctions in kin terms are old in
MK does not, of course, mean they are old in Proto-Austro-Asiatic.
Munda is not the only language family to be influenced by its neighbours,
although it may be true that it is easier to perceive areal influences on
Munda than on Mon-Khmer. The usual assumption, that MK preserves
much more of PAA than does Munda, is probably true, but what is and is
not PAA needs to be demonstrated after sufficient MK and Munda
evidence is in, examined, and evaluated, and the outlines of PAA are
clearer than they are now.8

8. Some of the fieldwork reported in this paper was done under a grant from the National
Science Foundation.

52


South Munda kinship terms, I

APPENDIX9

*kuli

Bhattacharya 83. (abridged) Juang koli, kuli, OBrWi; Kherwarian
*hili, OBrWi

We noted earlier that adding KM *hVli makes the SM evidence more
convincing. Pinnow in his Juang monograph also connected Juang kuli
and Kher *hili, but took the etymology no further. We note the form
kulaya-sini (only in McDougal) which is, apparently, a derivative of kuli-
sini (elsewhere -sel, -sen) is the combining form of (McD.) kon-chalan (Pi.
kon-selarj) 'young woman', kulayasini means HuYSi, and can be roughly
glossed 'kuli-junior', i.e. the 'junior' of OBrWi is HuYSi. Note the use of
-sini in the 'grandchild' terms boko-lap (McD.), 'grandson', and boko-sini,
'granddaughter'. The grandchild terms probably derive from boko-X-lab/
sini, the X perhaps being the -du? found in Kharia (cognate with words
for 'child' in Gorum (adud), and for 'young man' in Gotub (o-rug)).

SM *hVli(-boi), PM *Vrwel, SM *Vrel/*Vrer, NM *erwel

Bhattacharya 81. (abridged, rearranged, and provided with
additional data, the latter in parenthesis)

Set (a). Ga ili-boy (MZ hli-bwe?), HuYSi; Re liboi, HuYSi; So dliboi,
HuYSi (Vi. aliboi, HuYSi, WiYSi); 7? Ju ini-bo (McD. ini-bou, MM ini-
bo), HuYBr. -boil-boyl-boj mean 'woman' in these compounds; -sij/sij
(full form in So pAsij, 'child', in compounds with terms meaning nephew/
niece -sij means 'male person'. This -sij is probably cognate with the -serej,
-serej/sered, -sered mentioned earlier for NM and meaning 'young man,
nephew'. Bhattacharya seems to be suggesting that the -bo in ini-bo is a
combining form of boko, YBr. We have no other examples of thisor
anything elseas a combining form of boko.

Set (b). Ga urve, HuYBr (MZ wrwe?); Re ere (ZZ ere(l)), HuYBr; Gu
erel, HuYBr; Go ilil, HuYBr; So arer-sij-an, HuYBr, erer-sij-dn, WiBr (we
take drer, and erer to be the same), erel-boi, WiYSi (Vi eri-sij, WiYBr,
HuYBr; Sur erer-sij, WiYBr, HuYBr; ali-boy, arrel-boy, WiYSi, HuYSi).

One complication in Sora that needs comment; we find in both
Vitebsky's and Suryanarayana's data that along with ali-boi for SpYSi, we
also find eri- (Vi.) and arrel-(Suv.)-boi. The eri- and arre I- are both from
KM *Vrel, although the eri looks like a portmanteau of dli and Vrel. Since
the marking of dli as feminine, and of Vrel as masculine in Sora is being
lost (this sex-marking role has been taken over by -boi and -sij) the
meanings of Vli and Vrel have come closer. Vitebsky suggests that eri-

9. The data in the Appendix are, if not otherwise marked, from Bhattacharya. Words in
parenthesis not further labelled are from Zide and Zide. The abbreviations refer to: FF,
Ferndandez; MM, Mahapatra and Matson; MZ, K. Mahapatra and N. Zide; McD.,
McDougal, Pi., Pinnow; Sur., Surayanarayana; Vi., Vitebsky; VU, Vidyarthi and
Upadhyay; and ZZ, Zide and Zide.

53


NORMAN H. ZDE & ARLENE R. K. ZTDE

means 'younger', i.e. 'younger SpSib', since it can now be used with either
wife's younger sister (eri-boy along with ali-boy) or younger brother (eri-
sij), and similarly for a female ego for HuYBr and HuYSi. For Vitebsky's
Soras this would work, but not for Bhattacharya's. For Suryanarayana
SpYBr is erer-sij; *Vrel has been generalised, but then split into arrel,
which commutes with ali-, i.e. goes with -boy. The other form *Vrel has
split into is erer-, and this is used only with the masculine -sij.

We are grateful to R. J. Parkin for copies of his papers, and to Piers
Vitebsky for his notes on Sora, and for copies of the extract from
Suryanarayana's dissertation.

Set (c). Sa erwel, HuYBr (Bodding era+ ell); Mu iril, irul, iriul, HuYBr;
Korwa irvil, HuYBr; Koku, ilur, HuYBr. (WiYBr in NM is *teya (i.e.
[teya\).

For set (a)putting aside ini-bo for the momentwe reconstruct
*hVli(-fo/), HuYSi, SpYSi; using Bhattacharya's data only it would be
*Vli-boi. We find this to be cognate with Bhattacharya's 83., *kuli/ *kVli,
OBrWi. The definition of the new PM *kuli would be 'female affine of
ego's generation' (i.e. English 'sister-in-law'), OBrWi, HuYSi, SpYSi. An
alternative interpretation would reject Juang kuli as cognate with either
NM *hili or KM *hVli, and include ini and NM *hili and KM *hVli in
the reconstruction.

Bhattacharya presents sets of forms that are semantically similar. He
would like these semantic sets to be cognate sets as well. But he wants his
semantic-cognate sets to be closer in meaningfor him to accept them
than, we think, such cognate sets in PM will often be. It is true that some
of these kin terms' semantic-cognate sets (e.g. Bhattacharya's bare, ZZ
boHre) show forms that have remained remarkably close in meaning and
in phonetic form in the modern languages. But given the chronological
separation of the Munda languages (at least twenty-five hundred years),
there is no reason to expect this degree of closeness. Just as we have no
hesitation in calling Ga swa and So atjaI, both 'fire, firewood', cognates
because we can see how both developed regularly from KM *sVi]aH1we
should have no trouble in accepting NM *hili and KM *hVli as cognate
despite the excessive (to Bhattacharya) difference between the meanings of
the two forms. We need the anthropologist to reconstruct a more precise
meaning for the proto of these, i.e.if we accept Juang kuli*kVli, and
to account for the changes in meaning between the PM form, and the
forms in the modern lanuguages.

From the forms in set (b) we reconstruct KM *Vrel/*Vrer, HuYBr,
SpYBr. From this and NM *erwel (reconstructed from the forms in set
(c)), we reconstruct PM *Vrwel/*erwel, SpYBr, HuYBr.

It is possible, as Bhattacharya suggests, that *Vrwel is bimorphemic,
and that the second morpheme (of *Vr()-) is -hVl(i) or -kVl(i), but we see
no evidence of this.

54


South Munda kinship terms, I

ini-bo, ini-bou/'bau

What then of ini-bol Since bau/bou is the reciprocal in Sora of KM
*hVli and KM *Vrel derivatives, and a similar situation may obtain in
Juang, let us look at Bhattacharya's Set 62., bau/bou.

PM *bao/*bau

Bhattacharya 62. (edited, and with additional material): So bao-n,
'brother-in-law', (Vi. baon OSiHu, Sur. boung, OSiHu, kinar-boung
SpOBr); (ZZ Go bao, 'brother-in-law'); ZZ Gu -bay, m-baij, OSiHu);
Remo (Bhattacharya, 1968, 111) m-bat;, OSiHu; (ZZ Ga m-bia, OSiHu);
Kh bau HOBr, Roy bao-tang, WiOBr (VU boutang, WiOBr); Ju bov-kar,
HuOBr (Pi. bau/bou, OSiHu, MM bau, SiHu, McD. bou, ESiHu, bokar
(probably from bau-kar), HuOBr); NM: Ho bavo, OBr, SpOSiHu; bau
honjar, SpOBr (Deeney bau, OBr or cousin; bau honyar, SpOBr); Mu bau,
OBr (in address); bau honjar, SpOBr; Sa bahon-har, SpOBr; Korku bao,
WiOBr, (Girard bao, WOBr). The final nasals in some of these words are
frozen first person prenominal enclitics used in terms of address, i.e.
literally, 'my OSiHu', etc.

To return to ini-bo (or -bou), we stated in the body of the paper that
with regard to sound correspondences and semantics, a case can be made
for ini being cognate with PM *Vrwel, SpYBr, HuYBr, if not the
strongest case. What of the -bo or -bout ini-bo or -bou is an inalienably
possessed noun in Juang. It is true that reciprocalsboth terms and the
relationship between thenare often maintained through many linguistic
changes. But that a pair of reciprocals meaning (an inalienably
possessible) HYoBr, this deriving from the usual paired reciprocal
meanings, could have developed in Juang is, for semantic reasons, highly
unlikely, -bo, although we have seen no (other) examples of it as a
combining form of boko seems more likely. Why does ini need a second
morpheme at all to have the meaning HYBr? As far as we can see, it does
not. The spouses of both kulayasini and inibo are both some kind of boko:
the inibo's wife is (McD.) boko-ray (YSi), and kulayasinf s husband is boko
(YBr). HYBr in McDougal is boko-ger (from boko-kerl); elsewhere he is
just boko. Apparently ini too required a second morpheme, -bo from boko
makes better sense than bau/bou.

-kar, 'in-law'

On bou-kar and the use of the affine-marking -kar in Kharia-Juang.
The affinal -kar, 'in-law', is found in PM in *kikar/*kankar, SpM (or
perhaps, SpM, WiOSi), and *ku(X)nkar, SpF (or SpF, HuOBr). The -kar
may be related to what has become the NM word for 'man, person,
member of the community', koro. The Kharia -kar marks the agent in
forms like rema-kar, 'call-er', rema- 'to call' (Malhotra, 1982: 311). The
kar is also used in forming singular pronominal stems from demonstrative

55


NORMAN H. ZDE & ARLENE R. K. ZTDE

bases (e.g., ho-, -), thus ho-kar, u-kar, 'he, she'; i.e. -kar means 'one'. In
KJ we find the following forms: Juang keeps the term for 'father-in-law',
kukar/kwikar/kwiykar, and also has bao-kar, HuOBr, and aji-kar,
SpOSi. Kharia keeps the old word for 'mother-in-law', kinkar, and has, in
addition, boker (probably from boko-ker), SPYBr, and (Roy) aichkar,
WiOSi. (We noted earlier that Juang has boko-gerwith the ger perhaps
from -ker and not from koN-ger, 'young man'.)

In some but not all cases the -kar can be translated simply as 'in-law',
i.e. if X means YSi, then X-kar means YSi-in-law, i.e. SpYSi. This
presupposes that X is not an affinal term. The examples of this in KJ are
aji, ESi, and boko, YBr. One could, simplistically, try to derive the
remaining terms from hypothetical simple terms, not in KJ, but in
Kherwarian. Thus Ho bau, OBr, Juang bau-kar, HuOBr. This would miss
the fact that Juang bau¡bou is an affinal term, OSiHu, and that kar
derivatives of affinal terms are different from derivatives of simplex terms.
Where the fcar-derivative is formed on an affinal term, the simplex refers
to a sibling's spouse, and the kar-derivative to a spouse's sibling. The
other example of this in KJ is (Roy) aji, OBrWi, and aich-kar, WiOSi. The
Juang boko-ger, HuYBr, if regular, should come from bok(V)-(k)er-ker.

The term bok-sel in Kharia, SpYSi (boker is SPYBr) is probably to be
derived from bok-ker-sel. And the Juang 'grandchild' terms, boko-lab,
'grandson', and boko-sini, 'granddaughter' are probably to be derived
from boko-X-lab/sini, the X perhaps being cognate with the du? found in
Kharia in bok-du?, 'grandchild'. The reduction of similar compounds in
Sora from morphemes 1-2-3 to 1-3 was observed by N. Zide in regard to
the Sora numerals.

The Kherwarian uses of *-kar (which, if it were reduced the way it is in
KJ would be *-har): PM *kinkar/*kankar (Kher *hanhar) and
*ku(X)nkar (Kher *honhar): these occur in compounds of the structure
X-hanhar/honyar, which have the meaning hanhar's or honyar"s X, i.e.
kaka-honyar in Ho means honyar"s kaka, i.e. SpFYBr. For X as aji and
bau this is not the case. The construction with aji-hanar, and bau-honyar
in Ho, is like the construction in KJ with non-affinal simplex plus -kar.
Note that (unlike KJ) aji in Ho is OSi, and bau is OBr; aji-hanar means
SpOSi, and bau-honyar means SpOBR. Note that these are 'spouse's
sibling' terms, which, in KJ, are those derived from affinal simplex terms,
not, as here, non-affinal simplex terms.

Santali is much like Ho in its SpF (honhar) and SpM (hanhar) terms:
most of them have the X-hanhar/honhar construction where this means,
e.g. hanhar"s X (where, for instance, X would be goijgo, OBr). For the two
forms cognate with the two just distinguished in Ho, Santali bahohar and
ajhnar, the meanings are parallel to those in Ho: SpOBr, and SpOSi. Note
that baho does not occur as a free form (as does the Ho cognate bau,
OBr), but aji, OSi, does. Note the reduction of honhar to -har in the first
form, and that of hanhar to -hnar in the second.

56


South Munda kinship terms, I

*boko (NM *boko)

It is not surprising that Parkin (with Bhattacharya's help) sees forms
related to *boko almost everywhere. He is right to connect Ga tabo (MZ
ta-bo?) and Re tabuk\ FYBr, MYSi, stepF (Bh. Set 31.) to boko. Cognate
forms in Set 33. should also be included: Bh. mbu, (FF mbuElwin
umbuk-boi), MYSi, FYBrWi, stepM. Bhattacharya is probably wrong in
bringing in bobo (he translates the Gta? form, wrongly, we think, as YBr).
In KJ *boko and possibly related forms cover a good deal of semantic
territory: apart from the basicYSibmeanings there is the Juang word
for 'clan' (Pi. bog, McD. bok), the 'grandchild' terms we have talked about
(Kh (Roy) bok-du(?), Juang (McD.) bokolap and bokosini). Pinnow
records the Juang term 'boko-rad, "Verwandschaft" (relation), Bedeutung
und Form unklar'. The form found in Kherwarian, boko, is unexpected,
since one would expect the intervocalic k to go to h before o. Bhattacharya
gives a form, Ho boho, which is not found in Deeney (but see the other
forms in Bhattacharya's Set 64. (1970: 455)). There do seem to be related
forms in Kherwarian with h, e.g., in Bodding's Santali Materials II (1929:
21) there are boko boeha and bohok bohok boeha, 'cousins any number of
times removed, descendants in the male line'. One wants to know not only
the nouns, the kin terms, but the rest of the kinship vocabularythe
verbs and the constructions. This sort of information can be found in
Bodding and Vitebsky.

As to the rest of the terms with initial b that might be etymologically
connected with boko, a better knowledge of MK cognates would help. For
instance, we would tentatively connect the first piece (etymologically,
presumably, a morpheme) of bVyay, boay, 'brother, a man's brother'
and GRG *bVlon 'sister, a woman's sister', Sora (Vi.) bui-mai, (buj-mai ?
bV-mai ?), 'true sister, usually used by women'. Bhattacharya wants to
connect bVyay with Sora, YBr, ubba-n-u(b)bdy- (Vi. u'ba-). We think it is
probably cognate with boko. To make a case for eitheror, conceivably,
bothconsiderably more synchronic phonological and morphological
analysis needs to be done. For instance, by Bhattacharya's hypothesis,
one would take -bdy as a combining form of bVyay, odd but not
impossible, but it does not fit in with assumption of a morpheme bVn-\
bVl- shared by *bVlon and KM *bVyaN. And we have other forms:
Gorum biy-ger, 'bloodbrothers', Sora birinda, 'clan', i.e. (Vi.) 'exogamous
agnatic localised virilocal kin group' (Vitebsky translates b Vyay as 'male
member of own birinda'), and Sora FaYBr kim-bom, Vi. kin-bom, kin-boy
that come into the picture. The sibling terms and related vocabulary will
be taken up in a later paper, which will take MK cognates into
consideration, and do more synchronic phonological and morphological
evidence than we have done here.

57


NORMAN H. ZDE & ARLENE R. K. ZTDE

REFERENCES
Bhattacharya, S.

Biligiri, H. S.
Bodding, P. O.

Bouez, S.
Deeney, J.

Diffloth, G.
El win, V.
Fernandez, F

Girard, B.

Guha, U., Siddiqui,
M. K. A. &
Mathur, P. R. G.

Hoffmann, J. &
van Emelen, A.

Karve, I.

Macphail, R. M.

Mahapatra, B. P. &
Matson, D. M.

Mahapatra, K. &
Zide, N.

Malhotra, V. B.
Matson, D. M.
McDougal, C. W.
Munda, R. D.
Parkin, R. J.

1968. A Bonda dictionary. Poona: Deccan College & Post-Grad. Res.
Inst.

1970. Kinship terms in the Munda languages, Anthropos 65, 444-65.

1965. Kharia phonology, grammar and vocabulary. Poona: Deccan
College & Post-Grad. Res. Inst.

1929. Materials for a Santali grammar II. Dumka: Santal Mission of
the Northern Churches.

1934, 1935. A Santal dictionary, Vols. 2, 3. Oslo: Jacob Dybwad.

1985. Rciprocit et hirarchie, L'alliance chez les Ho et les Santal de
l'Inde. Paris: Soc. Ethnographie.

1975. Ho grammar and vocabulary. Chaibasa: Xavier Ho Publ.
1978. Ho-English dictionary. Chaibasa: Xavier Ho Publ.

1986. Personal comunications.

1950. Bondo highlander. Bombay: Oxford Univ. Press.

1968. A grammatical sketch of Remo: a Munda language. Unpubl.
dissertation. Ann Arbor: Univ. Microfilms.

1969. A critique of Verrier Elwin's anthropology. In Anthropology and
Archaeology, essays in commemoration of Verrier Elwin. New Delhi:
Oxford Univ. Press.

n.d. Korku-Hindi-English dictionary. Ramkherra, Dertalai: Central
India Baptist Mission.

1968. The Didayi (Memoir 23, Anthropolog. Survey of India). Delhi:
Govt. India Publ.

1950. Encyclopaedia Mundarica, Vol. 2. Patna: Bihar Govt. Press.

1968-9. Kinship organization in India and place of Mundari-speaking
people in it. Adibasi 10 (1), 1-25.

1954. (ed).Campbell's English-Santali dictionary. Benegaria: Santal
Christian Council, 3rd ed.

1963. Juang Vocabulary (Mimeo.) Chicago. [Included in Matson,

1964.]

1970-2, 1978-9, 1982-3. Gta? field notes.

1982. The structure of Kharia. Delhi: Centre Ling. & Engl., Sch. Lang.,
Jawaharlal Nehru Univ.

1964. A grammatical sketch of Juang, a Munda language. Unpubl.
dissertation, Ann Arbor: Univ. Microfilms.

1963. The social structure of the Hill Juang. Unpubl. dissertation. Ann
Arbor: Univ. Microfilms.

1965. Proto-Kherwarian phonology. Unpubl. M.A. thesis. (Mimeo.)
Chicago.

1985. Munda kinship terminologies, Man (N.S.) 20, 705-21.

58


South Munda kinship terms, I

PfefTer, G.

Pinnow, H.-J.

Roy, S. C. &
Roy, R. C.

Stampe, D.

Suryanarayana, M.

Vidyarthi, L. P. &
Upadhyay V. S.

Vitebsky, P.

Zide, A. R. K.

Zide, N.

Zide, N., Das, B. P.
&. DeArmond, R

1982. Status and affinity in Middle India (Beitrage zur Siidasien-
Forschung, Siidasien Institut, Univ. Heidelberg, 73). Wiesbaden:
Franz Steiner.

1959. Versuch einer historischen Lautlehre der Kharia-Sprache.
Wiesbaden: Harrassowitz.

1960. Beitrage zur Kenntnis der Juang-Sprache. Unpubl. typescript.
1937. The Kharias. Ranchi: 'Man in India' Office.

1979. A compendium of Sora vocabularies, with data from H. S.
Biligiri, P. Donegan, B. P. Mahapatra, R. Mahapatra, D. Stampe, and
S. Starosta. (Printout.) Columbus.

1977. Marriage, family and kinship of the Saoras of Andhra Pradesh.
Unpubl. Ph.D. thesis. Andhra Univ.

1980. The Kharia: then and now. New Delhi: Concept Publ. Co.

1979. Field notes on Sora kinship.
1968. A Gorum lexicon. (Unpubl. typescript.)
1970-2, 1978-80, 1983. Field notes on Juray.
1970. Field notes on Sora.

1982. Reconstruction of Sora-Juray-Gorum phonology. Unpubl.
dissertation, Ann Arbor: Univ. Microfilms.

1956-8. Field notes on Korku.

1978. Studies in the Munda numerals. Mysore: Central Tnst. Indian
Langs.

1962-4, 1966-7. Field notes on Gutob.

59




PROBLEMS AND PITFALLS IN THE
PHONETIC INTERPRETATION OF
KHASI ORTHOGRAPHY

tEugnie J. A. Henderson

The growing interest in Khasi in recent years, as the lone representative of
the Austroasiatic family in an area surrounded by Indo-Aryan and
Tibeto-Burman languages, has resulted in a welcome number of serious
studies of the language. A number of otherwise valuable studies have,
however, been flawed by misunderstanding arising from the Khasi spelling
system. This looks relatively simple on the surface but there are pitfalls for
the unwary. Scholars with an English or Indo-Aryan linguistic back-
ground may be led astray if they are tempted to take the spelling at its face
value and to trust their eyes rather than their ears. The sections devoted to
phonetics and phonology in some studies I have seen have contained
phonetic or phonemic 'transcriptions' which are little more than letter-
for-letter replicas of the spelling.1 In these circumstances, it is worthwhile
taking a closer look at problem areas within the spelling system itself.

There have, I believe, been certain attempts during the last few years at
minor revisions of the official orthography, and I cannot claim to have
up-to-date knowledge of all of these. As far as I am aware, however, the
system is substantially still the same as that used in Diengdoh (1965),
Kharkongngor (1968), Bars (1973), and Blah (1974). I shall also refer in
this paper to the older and probably better-known dictionary of U Nissor
Singh (1906), and occasionally to even older forms. The pronunciations
cited are from Cherrapunji speakers, since this variety of Khasi is accepted
as 'standard'.

Vowel length and final consonants

What has to be borne in mind is that the Khasi roman spelling was
devised in the first half of the nineteenth century, not by English or Scots
or American missionaries, as in some other parts of India and Burma, but
by Welsh missionaries, who introduced certain Welsh orthographic
conventions unfamiliar to most English speakers.

To the non-Welsh, one of the most confusing of these conventions
relates to the use of the letters -p, -b, -/, -d, -c, -j2 in word final position.
Linguists accustomed to the phonemic differentiation of voiced and
voiceless stops, in such languages as French and English for example, find
it natural to suppose that Khasi has this sort of phonemic distinction also,

1. This does not apply to Rabel (1961), whose phonetic observations are beyond reproach.

2. On the absence of final -g in Khasi, and on the special distribution of final -k, see
Henderson (1965).

61


EUGNTE HENDERSON

when they are confronted for the first time with spellings such as those in
(1) below:

1.Khasi: dab 'bullock' dap 'full'

had 'to tear' kat 'so much, as much as'

mad 'to taste, try' mat 'joint'
sngab 'gill' sngap 'to hear'

For those whose work has hitherto been mainly with Indo-European
languages, there is nothing very remarkable about this kind of final
alternation, but those who have specialised in the languages of South-East
Asia to which Khasi is related, it would be not a little surprising to find a
phonemic distinction between voiced and voiceless final stops. The
characteristic typological pattern found all over continental South-East
Asia is for the voice: voiceless contrast, which is frequently found in initial
position, to be neutralised in final position. What we usually get is an
unexploded voiceless stop, often accompanied by simultaneous glottal
closure. Listening to the pronunciation of the Khasi words in question
soon makes it clear that Khasi is no exception to the general rulethe
final stops are voiceless, no matter what letter they are written with. What
we also observe is that, in the great majority of cases, where the final letter
in the spelling is b or d the preceding vowel is always long, while where it is
p or it is almost always short.3 The words in (1) above are regularly
pronounced:

2. dab [da:pn] dap [dap-1]
kad [ka:f] kat [kaf]
mad [ma:!"1] mat [mat"1]

sngab [sqa^"1] sngap [srjap-1]

This seems very curious until it is realised that Welsh has just such a
convention: vowels before the letters b, d and g are almost always long,
whereas before /?, t and c are almost always short. (For this rule, and for
the exceptions to it, see Wells (1979).)

3. Welsh: mab [ma:b] 'son' map [map] 'map'

tad [ta:d] 'father' at [at] 'to, towards'

¡log [io:g] 'interest' llac [iak] 'slack'

Phoneticians who have worked on Welsh claim that while the sounds
written -b, -d, -g are voiceless finally, they are also lenis, or lax, as
compared with those written -p, -1, -c, which are voiceless and fortis, or
tense. Wells thus prefers to describe the lenis set as 'devoiced', here
symbolised [b, d, g], following the IPA convention. In Khasi there is no
such tense/lax distinction, any more than there is a voice/voiceless
distinction in final stops: they are all voiceless. It seems clear that the early
Welsh missionaries made use of the their own spelling conventions in

3. See note 2 above.

62


Problems in Khasi orthography

order to mark distinctions of vowel length in Khasi.

This convention works well with stops, where there are two sets of
letters available, but it does not work for words ending in final nasals or
in final r, which are always voiced. In these circumstances, in Welsh, a
circumflex accent is used to mark the long vowel:

4. Welsh: tan [ta:n] Tire' tan [tan] 'under'

ton [to:n] 'tune' ton [ton] 'wave'

glan [gla:n] 'clean' glan [glan] 'bank'

Earlier spellings of Khasi also tried this device. In Nissor Singh (1906)
the circumflex accent is often used to mark vowel length before nasals and
r\ alternatively, a grave accent may be used to show length, or an acute
accent to show shortness, but this is not at all regular, e.g.:

5. Khasi: bam [ba:m] 'to eat' sbai [sba:i] 'cowry'
ker [ke:r] 'to enclose' kpr [kpe:r] 'garden'
her [hs:r] 'to fly'
r [ur] 'to slip, fall' tur [tur] 'to butt'

It appears probable that, in these words, there is some variation in usage
among Khasi speakers, which may account for some of the seeming
irregularities. Thus, we find r 'to slip, fall', marked with a short vowel in
Nissor Singh, but tur 'to butt' written without an accent, although also
pronounced with a short vowel by my Khasi informants. Similarly, we
find the spelling ker [ke:r] 'to enclose', but kpr [kpe:r] for the derived
form meaning 'garden, enclosure'. Her, marked short, was pronounced
with a long vowel by my informants, but it may be that some speakers use
a short vowel here. This irregularity and uncertainty in the use of accents
may be one of the reasons why they seem to have been abandoned
completely in most of the recent dictionaries. This leaves the problem, for
the foreigner, of being unable to tell from the spelling whether the vowel is
long or short in such words. Father Bars' (1973) dictionary, for example,
has to resort to such entries as 'bam (long a)' and 'bam (short a)'.

The vowel 'y'

Another problem for linguists has been the vowel spelt y in Khasi. It is
almost always unstressed and behaves in much the same way at the shwa
vowel [a] in English, as in 'about, again, collapse, suppose', etc. When we
look at Welsh, we find this similarity is no accident, since y is the Welsh
way of spelling the vowel [3]. Depending upon its position in the word, y
may in Welsh also represent a vowel which is pronounced either [1] or [i],
according to dialect, e.g.: mynydd[mani, 'mani] 'mountain', y, with the
[a] value, is also the vowel that is used where a cushion vowel or epenthetic
vowel is felt to be needed to break up awkward or alien consonant
clusters, as in the forms ysgdr~sgdr [asgo:r, sgo:r] 'score'. The rather old-
fashioned Khasi spellings kypa 'father', bydi 'twenty', byni 'moon',
month', etc., show the same process at work. Since clusters like [kp, bd,

63


EUGNTE HENDERSON

bn], and many others, are difficult for Welsh or English speakers to
pronounce without a short vowel in between, the y vowel was inserted.
More recently the convention on the whole is not to write y between 2-
consonant clusters, but to insert it where there are clusters of more than
two consonants, e.g. bna 'to hear', bdi 'twenty', but byndi 'to mortgage'.
There seems still to be some uncertainty about this, however, as Father
Bars still gives spellings like byna. Phonologically speaking, there would
be much to be said for dropping the y in the 3-consonant clusters as well,
simply writing bndi, which is perfectly clear. This is the solution proposed
by Dr. Rabel in her phonemic analysis of Khasi (Rabel 1961).

Final palatals

Khasi has final palatal consonants; Welsh does not. What appears to have
struck the ears of the missionaries who first listened to Khasi was the i-like
glide that precedes the final palatals. We thus have early spellings such as
skin 'fly' n., bysein 'snake, python', etc., and it seems clear that the final
consonant was thought of as plain n in such words, though its palatal
quality was later recognised by the use of the letter : skai, bsei, etc.
Since there is no accepted roman equivalent for the final palatal stop, -d
and -t continued to be used, except after ie, when we find j, e.g.:

6. Khasi skai [ska:ji] 'fly n.' krui [kruji] 'white ant'

kshaid [kjaic1] 'waterfall' kait [kac"1] 'plantain'
btuid [btuiac] 'slippery' buit [buc"1] 'cunning'
miej [mi:3c] 'table'

From the phonological point of view, the vocalic segments preceding
these consonants should be interpreted as plain vowels (or in some cases
as centring diphthongs), with an automatic palatal on-glide to the final
consonant, rather than as phonemes /ei, ai ui/, etc. Father Schmidt,
writing on Khasi as early as 1904, recognised this; his knowledge of the
Austroasiatic family led him to expect that final palatals might be present
and he interpreted the missionaries' spelling in this light.

Representation of the glottal stop

The representation of the glottal stop, which is a consonant phoneme in
Khasi, does not derive from a Welsh convention, but rather from a much
more general missionary usage, found also in the roman orthographies of
Tibeto-Burman languages such as Lushai and Chin. Word-initially, the
glottal stop is not written at all; word-finally it is spelt with -h; which is
not otherwise needed in final position:

7. Khasi khlieh [khle^] 'head' spah [spa*?] 'hundred'

soh [so,?] 'fruit' kseh [kse,?] 'pine'

Problems arise when [>] occurs medially, or when it is the second
consonant of an initial cluster. Devices used in such contexts include
hyphens, apostrophes, or the vowel letter y. In the last case, the y

64


Problems in Khasi orthography

presumably represents the slight vocalic sound that can be heard as the
glottal stop is released:

8. Khasi ha-oid, ho-oid, Woid [ha^ox*1] 'Yes';

sh'iap, shyiap [piap"1] 'sand';

I'er, Iyer [l?e:r] 'wind';

s'iar, syiar [s9iar] 'fowl';

lyoh [1V>] 'cloud';

syiem [s?i:3m, s?e:m] 'king, chief

In the word for 'cloud' above, we find the glottal stop represented by h
finally, and by y in the initial cluster.

Dental I alveolar distinction

It is common for Western linguists to expect that th, d, n will form a
natural class, united by a common place of articulation. In Khasi,
however, as in certain other languages in the area, this turns out not to be
the case, [t] and [th] are dental, [d] and [n] are alveolar. This is not simply a
matter of phonetic detail; it is important in understanding the rules for
permitted sequences of plosives and nasals in Khasi initial clusters. At
first sight it appears that a sequence of virtually any two consonants may
form an initial cluster, but while the possibilities are very rich, there is a
constraint upon sequences of plosives and nasals whereby those with the
same place of articulation may not form a cluster together. For example,
bn, bt, bth, pd, pn, bn, thy, dy, dp, km, kt, etc., are permitted; pb, pm, bm,
ky, khy, are not; td, tn and thn are permitted, and seem to be exceptions
until it is recognised that the sequences here are also hetero-organic, and
hence perfectly regular.

Ambiguous vowel spellings

The letters u, o, and e,A and the digraph ie are sometimes phonetically
ambiguous.

Finally and before a glottal stop, ie represents a pure vowel, a very close
[e]. Before other consonants, pronunciation varies between [i:s] and [e:],
depending upon the speaker:

9. khlieh [khle?] 'head' lieh [le^] 'white'
sdie [sde*] 'axe' kmie [kme*] 'mother'
dieng [dirsq, de:i]] 'tree, wood' ktien [kti:an, kte:n] 'word'

In open syllables, the spelling o represents a very close [o] in some
words, [o] in others. There seems to be variation in usage between speakers
for some words:

10. ro [ro* or ro] 'quicksilver'; to [to] (indicating assent);
-to [to-] 'that (enclitic)'.

In closed syllables, the phonetic value is always [o].
4. For a more detailed account of Khasi vowels, see Henderson (1967).

65


EUGNTE HENDERSON

u before a final glottal stop is [ir] in some words, [o ] in others:

11. ruh [rir?] 'also, indeed' ruh [ro?] 'cage'
duh [do*?] 'lose' shniuh [Jjio 'hair'

Before other consonants, u may represent long [u:a]. [o:] or short [u]:

12. bud [bu:st\ bo:tn] 'follow' thub [thu:9pn, tho^"1] 'reduced in size'
tup [tupn] 'cannon' lum [[lu:3m, lo:m] 'hill'

lm [lum] 'gather'

In open syllables, e may represent either close [e] or open [e]:

13. te [te ] 'then, but' de [ds'] 'also',
me [mc] 'thou' re [re*] 'or'

In closed syllables the value is always [e].

To sum up, with a proper understanding of its origins and background,
Khasi spelling has much to recommend it. Once mastered, the conventions
regarding the use of the letter y, the use of -b, -d, -j to indicate length of the
preceding vowel, and the conventions for the representation of the glottal
stop, are in the main clear and unambiguous. A reasonable complaint might
be that vowel length before nasals is commonly not shown. However, in my
view, the traditional spelling, properly understood, represents Khasi usage
far more accurately than the misinterpretations and attempted 'improve-
ments' that have been suggested by some linguists with an Indo-Aryan
linguistic background, and little or no understanding of the strong Mon-
Khmer legacy that still survives in Khasi phonological structure.

REFERENCES

Bars, E. 1973. Khasi-English dictionary, Shillong: Don Bosco Press.

Blah, V.E. 1974. (comp.) Chapolas Anglo-Khasi dictionary. (3rd ed.) Shillong:

Chapala Book Stall.

Diengdoh, A.K. 1965. Leemuel's Anglo-Khasi pocket dictionary. Shillong: L. Harrison.

Henderson, Eugnie, 1965. Final -k in Khasi: a secondary phonological pattern. Lingua
J. A. 14, 459-66.

1967. Vowel length and vowel quality in Khasi. Bull. Sch. Or. Afr. Stud.
30(3), 564-88.

Kharkongngor, U I. 1968. Ka Dienshonhi (a Khasi-Khasi dictionary) Shillong: Ri Khasi
Press.

Rabel, Lili 1961. Khasi: a language of Assam. Baton Rouge: Louisiana State Univ.

Press.

Schmidt, W. 1904. Grundziige einer Lautlehre der Khasi-Sprache in ihren Beziehungen

zu derjenigen der Mon-Khmer Sprachen. Mnchen: Verlag der K.
Akademie.

Singh, U Nissor 1906. Khasi-English dictionary. Shillong: Eastern Bengal & Assam
Secretariat Press.

Wells, J.C. 1979. Final voicing and vowel length in Welsh. Phonetica 36(405), 344-

60.

66


HU A LANGUAGE WITH UNORTHODOX
TONOGENESIS

Jan-Olof Svantesson

In this article I will describe and analyse a small vocabulary I collected in
September 1984 from a native speaker of Hu /x?/, a Mon-Khmer
language spoken by about 1000 persons in a few villages in the Xio
Mngyng area in Jinghong county, Sipsong Panna (Xishuang Bnn),
Ynnn province, China. The Hu are known among the local Chinese as
Kngg.

Hu belongs to the little-known Angkuic group of the Palaungic branch
of the Mon-Khmer languages. Small Angkuic populations are scattered
over south-western Ynnn province, and in another article (Svantesson
1988) I describe the language of another of these, U, spoken in the village
Paa Xep (Bangxi) in Shuangjiang county.

The place of Angkuic within the Palaungic branch is shown in the
following table according to Diffloth (1982a):

Palaungic

East Palaungic
Waic
Angkuic
Lamet
West Palaungic
Danaw
Riang
Rumai

Although my material is too small to allow a complete synchronic
phonemic analysis, the most important historical developments in Hu can
be inferred from it. From a general phonological point of view, the most
interesting phenomenon is the development of a two-tone system where
the tones are not the reflexes of voiced/voiceless proto-initials, as is most
often the case in Mon-Khmer two-tone (or two-register) languages.
Instead, the tones are the reflexes of the long/short vowel opposition
which existed in Proto-Palaungic (inherited from Proto-Mon-Khmer). As
far as I know, no language with this kind of tonogenesis has been
described before.

67


JAN-OLOF SVANTESSON

Initial consonants

The Angkuic languages are characterised by a 'Germanic' development
of the initial stopsthat is to say, voiceless stops have become aspirated,
and voiced stops have become voiceless unaspirated. This is illustrated by
the following examples:

Hu U Lamet N. Kammu S. Kammu
*voiceless
phip pht pin pip pip 'to shoot'
thp than taaji taap taap 'to weave'
khp khap kap kap kaap 'jaw'
*voiced
pap pan paji - - 'white'
pKp q priP p? brP 'forest'
phltk ?ata? pltaak ktak kdaak 'palm (of hand)'
kij ka - kay gaaij 'house'
kk ktf kk 'to bite'

In Southern Kammu (as recognised from Lindell et al. 1981), the
original Proto-Mon-Khmer voicing contrast is retained. The unaspirated
stops in the Angkuic languages Hu and (Pac Xep) U correspond to
voiced stops in Southern Kammu, and the aspirated stops correspond to
voiceless unaspirated. Original voiced and voiceless stops have merged in
Lamet (Rmet; from Lindell et al. 1978) and Northern (Yan) Kammu,
giving rise to lax and tense register in Lamet, and low and high tone in
Northern Kammu, as is the case in Mon-Khmer and other languages
with 'orthodox' register or tone development*. The examples also show
that the Hu tones are not the result of orthodox tonogenesis.

Hu has a contrast between initial s- (with the allophone [q] before i, and
in the word snfj 'red') and 0an opposition which is not present in U or
other Angkuic languages, but which is found in Danaw (Luce 1965), as a
contrast between ts- and 0-. In Lamet, and in the rest of Palaungic (and
in Kammuic), Hu s- and 6- correspond to s- and h-, respectively. Diffloth
(1977) and Ferlus (1978) reconstruct these as Proto-Mon-Khmer *c- and
*s-, respectively. According to Diffloth (1977), *c- became *ts- in Proto-
Palaungic. Examples are:

Hu U Lamet Danaw
*c- so7 sd so? tso] 'dog'
sal7 sal slee? kl] 'rain'
ns? nch sP tsi1 'louse'

*In the Northern Kammu form ktak, the tone is determined by the voiceless k. (Ed.)

68


Hu A language with unorthodox tonogenesis

Om sp hum Oorx 'to bathe'
paOji set phji pOr4 'snake'
0a?5ji S?5dji 'dry'
6 ama? sama ?ma? 'wind'

U has the regular reflex ch- of *c- after minor syllables, as in the example
louse'. Proto-Palaungic also had *h-, which is retained as such in all
languages but, unfortunately, I failed to elicit any word with *h- in Hu.

There are some words in which Palaungic j- corresponds to c- in
Kammuthey include Hu srj 'bitter', Kammu cay. Here, Ferlus
reconstructs *ts- in Palaungic, Kammuic and Viet-Muong (correspond-
ing to two Proto-Mon-Khmer initials, *ts- and *ts-).

Minor syllables

As in other Northern Mon-Khmer languages, most non-compound
words are monosyllabic or sesquisyllabic, i.e. consisting of a major
syllable preceded by an unstressed minor syllable (cf. Shorto 1960).
Minor syllables have a syllabic sonorant (such as m in Omphp 'lung') or
a (probably) non-contrasting vowel which I have written as a (ka?a
'two').

There is also a contrast between 0- and s- minor syllable initial in Hu
(and in one word, ts- is attested). This opposition is not maintained in this
position in U, Lamet or Proto-Waic ( = PW; Diffloth 1980), where the
most common corresponding initial is s-:

Hu U Lamet PW
Oanat nt snat *snat 'gun'
Oa?aw sa?a s?ar 'sour'
0a?5ji s?5oji 'dry'
Oavy sava 'to ask'
Oathn satht 'old'
Oath? sat ha nta? *snta? 'tail'
6 ama? sama ?ma? 'wind'
Omphp saphp 'lung'
sam? sam *smo? 'stone'
sal? sal ses7 *hls? 'rain'
saply sa/a smplay 'shoulder'
sayy yy yay *?r)ay 'eye'
sykh7 kh *i]ko? 'yesterday'
tsaijal saysn syal *sgal 'blue'

Presumably, 0- and s- are the reflexes of Proto-Palaungic *s- and *ts-
(< *c-), respectively, which have merged into s- in U, as usual. It may be
noted that *s- has (at least in some cases) been retained in this position in

69


JAN-OLOF SVANTESSON

Lamet and Waic. As mentioned above, word initial *s- has usually
become h- in these languages. There are irregularities, which may be due
to more complex initial consonant clusters.

The occurrence of minor syllable initial ts- in tsaijal 'blue' suggests a
different Proto-Palaungic consonant, presumably the reflex of Ferlus'
(1978) *ts-, since the Kammu cognate is cijar with initial c-.

Hu has also retained the contrast between s- (< *ts- < *c-) and 0- (<
*s-) in major syllable initial position after a nasal minor syllable:

Hu U Lamet PW
nsP nchi sP *si? 'louse'
nasdk sii? ydok *hyok 'ear'
nOac ntsht mac *hmac 'sand'
nOim nchip Imhiim *mhem 'claw'

Here, U has s in 'ear', where the minor syllable has disappeared,
otherwise ch or tsh (which are probably allophones of a single phoneme).
After a nasal minor syllable, s never occurs in U, but has developed into
tsh/ch, which accounts for the unexpected occurrence of these consonants
in the words for 'sand' and 'claw'.

Hu also has a number of words which have a nasal minor syllable
followed by a voiceless nasal major syllable initial:

Hu U Lamet PW
nnam sanm nam *hnam 'blood'
nnim sandp mm *nvm 'year'
nnt sana 'comb'
nn? n *n?ne? 'meat'
mml mn kmul *kmil 'silver'
njit jiaS kjtas *kjias 'to laugh'

The Hu forms suggest a *nasal + h initial cluster, while Lamet and Waic
in some cases have clusters with a stop and a nasal. Taken together, this
implies proto-forms with *stop + nasal + h clusters. Medial h has been lost
in most of Palaungic (cf. Diffloth 1977), and in Hu, where h is retained,
the initial stop has disappeared.

U often loses cluster initital stops (see Svantesson 1988), as is the case
in the last three examples above. In the preceding list, the minor syllable
sa in the first three words in U, taken together with evidence from outside
Palaungic, suggests Proto-Mon-Khmer *j- (palatal voiced stop). This
tallies perfectly with Diffloth's (1980:175) reconstruction of 'blood' as
Proto-Mon-Khmer *jnhaam. For the other two words, *jnh- clusters are
also supported by evidence outside Palaungic: Kammu crias (Southern
Kammu jrias) 'comb', Mon criam 'year'.

70


Hu A language with unorthodox tonogenesis

Final consonants

Hu has retained the Proto-Palaungic (and even Proto-Mon-Khmer)
system of final consonants rather intact. An exception is final *-s, which
has developed into -t (merging with original *-t), another innovation
shared by Hu and Danaw (and several other Mon-Khmer languages as
well):

Hu Danaw Lamet
?axt k'r^et3 kriis 'bear'

phot prt3 pos 'sambar deer/barking

deer'

The final palatal stop *-c sometimes appears as -t in Hu (cf. Hu pet 'to
spit', Lamet pee; Hu ?amt 'mosquito', Lamet rmuc 'ant'), but my data
are too limited to reveal the exact circumstances under which this has
taken place.

Initial *r- is realised as a uvular [k] (as is also the case in some Lamet
dialects, in Blngshan Blang, and in some other languages of the area).
In final position, [k] is pronounced as a rather vocalic uvular glide which
can be written [] (incidentally a development which has also taken place
in my own southern Swedish dialect! See Lindau 1985 for different kinds
of r). Thus ?ih 'fowl' is pronounced pi]. After the vowel a, final *-r has
disappeared, or is retained as -w:

Hu Lamet

ka? ?lar 'two'

m mar 'field'

Oa?w s?ar 'sour'

kw kaar 'they (dual)'

Tones and vowels

There are two tones in Hu, namely, high (denoted by over the vowel)
and low ( ).

The co-occurrence of tones, vowels and final consonants is restricted,
as shown in this table:

Final

0 k others

Vowel, i u i

esas
oo

As mentioned above, the general rule is that originally long vowels
have conditioned low tone, and short vowels have conditioned high tone,
and the co-occurrence restrictions probably reflect restrictions on the co-
occurrence of long and short vowels with final consonants by the time

71


JAN-OLOF SVANTESSON

that the tone system was formed. Since Lamet is a Palaungic language
which retains the Proto-Palaungic (and even Proto-Mon-Khmer) length
distinction, I will use Lamet examples for showing the relationship
between Proto-Palaungic vowel length and Hu tones. For non-high
vowels, the relationship is quite clear-cut:

Hu Lamet
*short vowels: yam yam 'to die'
paOn phn 'five'
mj krmkji 'star'
ncen keen 'heavy'
*long vowels: yarn yaam 'to cry'
lk liik 'Pig'
?5m ?om 'water'
nasdk yook 'ear'

Before a final glottal stop, the tone is always high, probably due to
shortening of the vowel in this position before the development of tones.
For comparison, forms from (Northern) Kammu are given. This
language, which has orthodox tonogenesis, is another language where
the length distinction in vowels is lost before a glottal stop. Examples:

Hu Lamet Kammu
so? so? so? 'dog'
kath? kta? pt? 'earth'
W ?oo? ?d? 'I'
phl? ple? pi? 'fruit'

The high vowels i and u always have high tone, except before e or in open
syllables, where both tones occur. Compare the following examples with
long proto-vowels (I have no examples with long *ii):

Hu Lamet
?asim siim 'bird'
phifim priim 'old'
mml kmul 'silver'
?p ?uup 'cooked rice
Bum huum 'to bathe'
?IK ?eer 'fowl'

One possible explanation for the absence of a tone contrast in the high
vowels is that they have higher intrinsic pitch than low vowels, as has
been shown for many different languages (see, e.g., Lehiste 1970: 68-71),
which might have conditioned high tone for both long and short high
vowels when the Hu tone system developed. This explanation does not,
however, account for the occurrence of high vowels with low tone on
open syllables.

72


Hu A language with unorthodox tonogenesis

On the other hand, there are indications that the length contrast was
already lost in the high vowels in Proto-Angkuic, i.e. before the
development of the Hu tone system.1 Thus, judging from the words
given in Diffloth (1982a), there is no length contrast for u and i in Mok,
while the contrast is retained for non-high vowels. In U, the vowel length
contrast has disappeared, although it has left traces in final nasals which
are retained after originally long vowels but have become stops after
originally short vowels. Denasalisation has, however, taken place after
both *long and *short i and w, which are thus treated as if they were short.

The following examples show this development. Lamet cognates are
given because they retain Proto-Palaungic vowel length:

Lamet Hu U Mok
*non-high long: yaam yam ym jaam 'to cry'
?bom ?5m ?m ?oom 'water'
pon ?aphn phn phoon 'four'
*non-high short: yarn ym yp yem 'to die'
ntm nthm nthp tham 'egg'
keen ncn kit kdcsn 'heavy'
*high long: sim ?asm pachip ?a-sim 'bird'
ku/i kh/i kht khun 'male'
prim phxm xip phim 'old'
*high short: ky (khiy) khir khiy 'head'

(The Hu word khiy, which occurs in khiy koy 'knee' may be cognate to
the words meaning 'head' in the other languages.)

The reason for the loss of vowel length in the high, but not in the other,
vowels may be their shorter intrinsic length, something which has been
attested for various languages (see Lehiste 1970: 18-19). Furthermore, the
length contrast in the high vowels seems to have carried a rather small
functional load.

As mentioned above, final -k is more or less vocalic, which may explain
why low tone can occur on high vowels before this final, as it does in
open syllables.

The reason why o and o occur only with low tone (except before 9) in
my data is probably that there were relatively few words with short *o
and *o. This is the case in Kammu, where short and long o do not
contrast (see Svantesson 1983).

1. Proto-Angkuic did not have tones, as is proved by the absence of tones in Mok (Diffloth
1982a). U has a tone system, which is different from that in Hu, and for the other Angkuic
languages it is difficult to know whether they have tones or not, since they are known only
from older and not very reliable sources, which do not give any tones.

73


JAN-OLOF SVANTESSON

Vowel length or tones?

Figure 1 below shows some typical examples of the fundamental
frequency (F0) contours of the two tones on different types of syllables:

Hz yam yam pap khap

300 -

100 -

100 200 300 400 500 ms

Fig. 1.

In these cases, as in many others, there is a co-variation between
fundamental frequency and vowel duration, so that vowels carrying low
tone have longer duration than vowels carrying high tone. One might,
therefore ask whether an analysis in terms of vowel length rather than
tones is possible.

To investigate this, the duration of the vowel and the mean value of the
fundamental frequency over the vowel were computed (using the ILS*
program package) for some words with high and low tone. The words
were said in isolation by the female informant. Since the tone contours
are rather flat, the average frequency value can be used to characterise the
tones. The results are given in Table 1.

As seen in this table, vowels carrying low tone often have longer
duration than those with high tone, as seen in the only recorded minimal
pair, yam 'to die' and yam 'to cry'. On the other hand, some vowels with
high tone are longer than some with low tone, and this overlapping of
duration can be taken as evidence against treating vowel length as
distinctive. Each of the analysed high-tone words also has higher
fundamental frequency than each of the low-tone words.

Thus, fundamental frequency is definitely a consistent phonetic
correlate of the investigated opposition, while vowel length may be
regarded as a concomitant factor. It seems, therefore, reasonable to
analyse the opposition as one consisting of two tones.

* ILS = Interactive Laboratory System (Ed.).

74


Hu A language with unorthodox tonogenesis

Table 1

Mean values:

F0 (Hz) Duration (ms) F0 Duration
High tone: yarn 269 130
247 135
263 120
263 120 260 126
pp 253 115
249 95
242 95 248 102
kk 253 100
252 130
258 120 254 117
Low tone: yarn 214 200
215 175
215 225 215 200
khp 204 115
203 130
208 115 205 120
?ak 205 180
201 205
201 180 202 188
Note: The F0 ranges are 242-269 Hz for high tone and 201-215 for low tone.

Conclusion

The Hu data raise a number of intriguing questions, two of which will be
discussed here.

One concerns the classification of Palaungic. There are some striking
similarities between Hu and Danaw. In particular, both have 0- where
Proto-Mon-Khmer has *s-, and this is found nowhere else in Palaungic.
In the sub-classification of Palaungic given in Diffloth (1977), Danaw is
close to Angkuic, but in Diffloth (1982a), a classification (shown on p.
67), which puts Danaw and Angkuic rather far from each other is given.
Mitani (n.d.), using lexicostatistical methods for classifying Palaungic,
also finds that Danaw and Angkuic are rather distant from each other.
One might ask if my data from Hubeing an Angkuic language which
shares the innovation *s-> 6- with Danawnecessitates a revision of
this. Not necessarily, since it is quite possible that the development *s-

> h-, which has taken place in all Palaungic languages except in Angkuic
and Danaw, is the final result of two different processes: *s- > 6 and 6-

> h-. If that is the case, Hu and Danaw are not languages which share an

75


JAN-OLOF SVANTESSON

early innovation, but rather languages where a phonological rule (6- > h-)
has not applied, and there is no reason to assume that they have branched
off together from the rest of Palaungic at some early time.

Another question concerns tonogenesis. In almost all cases where a
Mon-Khmer language has acquired tones (or registers), the development
of the tone (or register) system is the result of a loss of contrastsusually
voicing contrastsin the consonants, but in Hu, the tones have
developed in connection with the loss of vowel length. The only other
case known to me, in which tonogenesis of a similar kind may have taken
place, is in Estonian, where a tonal distinction (different from that in Hu)
has possibly developed from an earlier vowel length opposition (see
Lehiste 1978). One somewhat similar case in Mon-Khmer is Pacoh which,
according to Diffloth (1982b), has an unorthodox registrogenesis, where
a register difference has replaced an earlier difference in vowel quality.

Both the acquisition of tones and the loss of vowel length are ongoing
processes in the area where Hu is spoken, so it is perhaps not surprising
to find a language that combines both. There might be a phonetic
explanation as to why long vowels have acquired low tone and short
vowels high tone, since there seems to be a general tendency for vowel
duration and fundamental frequency to vary inversely with each other.
As already mentioned, it has been shown for many languages that high
vowels have intrinsically shorter duration and higher pitch than low
vowels. For instance, measurements of the intrinsic pitch and duration in
Standard Chinese vowels (ptongha) have shown that, if other factors
are constant, vowels with relatively high intrinsic pitch also have
relatively short duration (Shi Bo, pers. comm.). Naturally, pitch
differences of 40 Hz or more, as are found in Hu, are not the result of
automatic adjustments, but an originally non-distinctive pitch difference
could have taken over some of the functional load carried by vowel
length, eventually acquiring phonemic status.

Vocabulary

The vocabulary is presented in reverse alphabetical order.

ka?a 'two' ma 'dry field'
man 'thing' p3 'not'
Ki 'to go' ?amd 'one'
?a?5 'monkey' mo 'axe'
lu 'bad'
la? 'leaf 6 ama? 'wind'
rja? 'to itch' man? 'to steal'
Qatha? 'tail' ?? 'we (pi.)'
k? 'they (pi.) sal? 'rain'
phl? 'fruit' m? 'you (sg.)'

76


Hu A language with unorthodox tonogenesis

kam? 'dream' tham? 'new'
nn? 'meat' ph? 'you (pi.)'
Ge? 'tree' th? 'to sit'
kath? 'earth' v? 'left (side)'
katP 'nose' ?? 'person'
rj? 'day' km? see po?~
tJX? 'pestle' px? 'forest, mountain'
pas? 'string' ns? 'louse'
paG? 'blood-vessel' th? 'hand'
jkhd? 'rice (husked)' sijkho? 'yesterday'
sam? 'stone' po? ~kxi?: 'spirit'
?o? T Gaklo? 'bark (of tree)'
so? 'dog' kato? 'banana'
pal? 'salt' x? 'Hu'
nOc 'sand'
lh 'to go down'
?k 'bow' kk 'to bite'
thifk 'buffalo' Gk 'rice (plant)'
phltk 'palm (of hand)' nthk 'tongue'
lk 'Pig' phuk 'ribs'
?aGk 'rat' tk 'small'
Gdk ~wiy: 'flea' nasdk 'ear'
nthok 'head' Gk 'hair'
thk 'to hang'
jl 'fire' t sayal 'blue'
mphl 'mortar' phl 'wing'
ntol 'wine' mml 'silver'
katl 'belly'
nnm 'blood' nthm 'egg'
ym 'to die' ym 'to cry'
nem 'younger brother' ?aGm 'right (side)'
?m 'to live' Km 'village'
phxm 'old (of things)' ?asm 'bird'
nOm 'claw' nnm 'year'
?om 'water' kathom 'liver'
nm 'piss' ttjHm 'under'
Om 'to bathe'
ln 'long (in space)' paGn 'five'

77


JAN-OLOF SVANTESSON

than see noy, ?ji
ncn 'heavy'
?dn 'he, she, it'
phin 'woman'
khon 'child'
ka?ajt 'wasp'
thajx 'to weave'
paOeji 'snake'
6a?oji 'dry'
?a?ji 'father'
matji 'ant'
ka? ai] 'bone'
kaij ~xaw\ 'heaven'
kayy 'iron'
maxy 'horse'
thy 'to kill'
Oavy 'to ask'
ley 'high'
nty 'big'
phdy 'many'
khy 'tooth'
many 'crab'
Oy 'bamboo'
my 'to look'
xoy Mekhong
khy 'wet field'
khp 'jaw'
Qa?ep 'rainbow'
cop 'to run'
Gmphp 'lung'
n* 'fowl'
phaOiK 'bee'
phK 'to fly'
pha?t 'to swell'
Oant 'gun'
njiat 'to laugh'
pt 'to spit'
?axt 'bear'
?ot 'to be at'

78

Oathan 'old'
ten 'low'
mon 'long (in time)'
?aphon 'four'
khon ~thP: 'finger'
pji 'white'
mji 'star'
phifi 'to shoot'
~than: 'grandfather'
khji 'man; husband'
ntji 'mouth'
ky 'house'
saply 'shoulder'
nay 'flower'
sy 'bitter'
pavy 'tomorrow'
cy 'foot'
sxy 'red'
they 'drink'
G5y ie\ 'bedbug'
khy ~koy: 'knee'
wy see 6dk~
koy see khy-
?a?oy 'wasp'
yov 'good'
pp 'to speak'
lep 'blind'
?p '(cooked) rice'
khiK 'moon'
kin 'finished'
klK 'to sew'
kht 'sick'
nnt 'comb; to comb'
thast 'lightning'
khast 'charcoal'
m 'to sleep'

phot 'sambar deer'


Hu A language with unorthodox tonogenesis

Got 'barking deer' ?amut 'mosquito'
tht 'breast'
Ga?aw 'sour' kw 'they (dual)'
?apalaw 'fish' phw 'you (dual)'
vw 'wide' xw see kaij~
?ayew 'cat' khw 'green'
7ay 'we (dual)' khy 'to eat'
?alay 'squirrel' ijy 'sun'
sany 'eye' rj?rjay 'far'
ndy -than: 'grandmother' ?andy 'mother'
khoy 'to have ka?5y 'three'
Koy S 'hundred'

Diffloth, G.

Ferlus, M.
Lehiste, Ilse

Lindau, Mona

Lindell, Kristina,
Svantesson, J.-O. &
Damrong Tayanin

Luce, G.
Mitani Yasuyuki
Shorto, H. L.

Svantesson, J.-O.

1977. Mon-Khmer initial palatals and 'substratumized' Austro-Thai.
In Mon-Khmer Stud. 6 (ed.) P. N. Jenner et al. Honolulu: Univ.
Hawaii Press, 39-57.

1980. The Wa languages. (= Ling. Tibeto-Burman Area 5(2)).

1982a. Subclassification of Palaungic and notes on 'Fuman1. Paper
presented at the 15th Sino-Tibetan Conference, Beijing, August 1982.

1982b. Registres, dvoisement, timbres vocaliques: leur histoire en
katouique. In Mon-Khmer Stud. 11 (ed.) P. N. Jenner. Honolulu:
Univ. Hawaii Press, 47-82.

1978. Reconstruction de /TS/ et /TS/ en Mon-Khmer. In Mon-Khmer
Stud. 7 (ed.) P. N. Jenner. Honolulu: Univ. Hawaii Press, 1-38.

1970. Suprasegmentals. Cambridge, Mass.: MIT Press.

1978. Poly tonicity in the area surrounding the Baltic Sea. In Nordic
prosody (eds.) Eva Grding, Gosta Bruce and Robert Bannert. Lund:
Lund Univ., Dept. Ling., 237-47.

1985. The story of /r/. In Phonetic linguistics: Essays in honor of Peter
Ladefoged (ed.) Victoria Fromkin. Orlando: Academic Press, 157-68.

1978. Two dialects of the Rameet (Lamet) language. Cahiers Ling.
Asie Orient. 4, 5-22.

1981. Phonology of Kammu dialects. Cahiers Ling. Asie Orient. 9,
45-71.

1965. Danaw, a dying Austroasiatic language. Lingua 14, 98-129.

n.d. Problems in the classification of Palaungic. Ms.

1960. Word and syllable patterns in Palaung. Bull. Sch. Or. Afr. Stud.
23 (3), 544-57.

1983. Kammu phonology and morphology (Travaux Inst. Ling. Lund
XVIII). Lund: Gleerup.

1988. U. Ling. Tibeto-Burman Area 11(1), 64-133.

79




ON AUSTRONESIAN LEXICON
IN VIETNAMESE1

Kenneth Gregerson

The sources of Vietnamese lexicon have been much discussed as regards
Chinese, Tai, and Mon-Khmer vocabulary (Maspero 1912; Schmidt 1926;
Haudricourt 1954), and one may now presume that Vietnamese is rather
safely established as an Austroasiatic language. The Austronesian (AN)
subset of Vietnamese lexicon, however, while clearly not having gone
unnoticed (qv. discussion in Benedict 1976),2 continues to invite a good deal
of sorting out. The present paper mentions some well-known Austronesian
forms as well as raising questions about whether certain other Vietnamese
items are ultimately of Austronesian (or Austro-Thai?) origin. The effort
here is to contribute towards a more deliberate Vietnamese perspective on
Austronesian lexical associations with Austroasiatic.

1. Function forms

Vietnamese possesses several very regularly used grammatical or closed-
set function words that appear to have Austronesian counterparts.

1.1. Aspectuals

The following forms function as regular Vietnamese (VN)3 aspectuals:

(l)a. da 'already

Toi da mua xe rdi.
I already buy auto already
'I have already bought a car.'

b. sap 'about to'

Toi sap mua xe.

I about-to buy a car
'I am about to buy a car.'

c. dang 'in process'

Ong ay dang lam viec.

grandfather that in-process do work
'He is working.'

1. Aversion of this paper was presented at the Eighteenth International Conference on Sino-
Tibetan Languages and Linguistics (Bangkok) in 1985. I gratefully acknowledge helpful
comments from Paul Benedict, James Matisoff, Nguyen Dinh Ho, and Bill Gage, though
certain things still survive that are perhaps not to their liking.

2. The ultimate Austronesian origin of Vietnamese is a notion that has existed Jor some time
(cf. e.g. discussion in E. Sauvignet 1922; K. Wulff 1942; as well as by Binh Nguyen Loe, c. 1960
in his Ngudn gdc Malai, so Prof. Ho informs me).

3. Some, though not all, Vietnamese illustrative forms are from Nguyen Dinh Ha, 1971.

81


KENNETH GREGERSON

d. mai 'continuative'

Ho di bo mai den Thu--Dwc.

They go by-foot continue to Thu-Dfrc
'They walked all the way to Thu-Du?c.'

These forms may be compared with the following in Bahasa Indonesia

(BI):4

(2)a. sudah 'already'

la sudah pergi.

he already go
'He has already gone.'

b. siap 'ready'

Mereka siap untuk pergi.
they ready for go
'They are ready to go.'

c. sedang 'while, in process'

la sedang membaca ketika saya datang.
he in-process read when I arrive
'He was reading when I arrived.'

d. masih 'still, yet'

ia masih tidur.
he still sleep
'He is still sleeping.'

As exemplified above in la and 2a, VN da 'already': BI sudah 'already'
both occur in preverbal position to signal Perfective aspect. In VN it will
be observed that a clause-final roi reinforces the same completive meaning
redundantly. This form is perhaps the 'true' Austroasiatic marker (cf.
Rengao (Rg.) hddroi 'before') into whose territory da has intruded.

The tone on da accords well with diachronic expectations clarified
originally by Haudricourt (1954), which may be generally summarised as
follows:

(3)

Original Finals
open final final

syllable stop spirant

voiceless CV(N) CVC CVH high tones

original
initials

voiced CV CYC CVH low tones

4. A number of the sample Indonesian sentences and lexicon are from Echols and Shadily
(1974).

82


On Austronesian lexicon in Vietnamese

Thus, depending on the original voicing status of the initial consonant
(C-) and the closure status of the final (-V(N), -VC, or VH), a particular
contrastive tone has developed in Vietnamese, typically, with a
concomitant loss of most of the original conditioning features. The
High Tones associated with old voiceless or imploded consonants are V
(symbolised in our exposition with V for clarity though it (the macron
is not written in Vietnamese orthography), V, and V. The Low Tones that
originally occurred with voiced initial consonants are V, Y, and V.

At this point, however, when dealing with disyllabic forms, such as BI
sudah, it is crucial to unravel the 'pecking order' that dictates which initial
consonant counts as to voicing status in the selection of the High vs. Low
tone set in (3) above. That is, in sudah does the or the d- take
precedence? Clearly, the voiced initial d- had prevailed, and the V tone has
been appropriately selected, and sudah > da.

To clarify this consonant precedence, it is perhaps worth the digression
to discuss the discoveries of Friberg and Hor (1977) as to initial consonant
'dominance' with respect to the register (+ATR* vs. -ATR) selection of
stressed (main) syllables in Western Cham (Austronesian of southern
Vietnam and Cambodia). Distinguishing register A (-ATR) consonants
from register B ( + ATR), Friberg and Hor (1977: 36) summarise how
precedence is established and register effects determined on the phonation
of vowel quality of the succeeding syllable:

(4) A (-ATR) B (+ ATR)
1. p t c k b d j g
ph th ch kh bh dh jh gh
?
2. s h m n i)

w 1 y r

Register A and B in Western Cham words were accounted for by the
following 'dominance' rules:

(5)a. A +A = A

b. B + B = B

c. Al 4- Bl = second element

d. Bl + A2 = B

e. A + B2 = A

These rules (following Purtle (1969) for Khmer), although neither ordered
with respect to each other nor within the left-hand members, indicate
register of tonic syllable. Quadrant Al and Bl are equally strong; they
both dominate A2 and B2. A2 also dominates B2. This analysis is based
entirely on the consonant 'strength'; the consonant is seen to 'dominate' or
determine the register characteristics of the following vowel. And, based
on the five 'dominance' rules noted above, certain atonic syllable initial

* = Advanced Tongue Register (Ed.).

83


KENNETH GREGERSON

consonants exert their strength over an intervening tonic syllable initial to
determine the 'registerness' of the tonic syllable vowel.

In the following examples, consonant dominance is seen in ^ the
combining of various syllables to form words (a grave accent / / is

added to indicate second ( +ATR) register clearly):
(6) Al+Bl /ka/ + /baw/ /kabaw/ 'buffalo'
BI + Al no examples (historically BI has become Al).
B1+A2 /bal/+/haw/ /bahaw/ 'new'
A2 + B1 /ha/ + /dom/ /hadom / 'how much'
A1+B2 /ka/ + /ro/ /karo/ 'strong'
B2 + A1 /la/ + /kaw/ /lakaw/ 'to step over'
A2 + B2 /ha/ + /nin/ /hanin/ 'bow'
B2 + A2 /la/+/say/ /lasay/ 'cooked rice'

Returning now to sudah > da, one may usefully compare this process
with Friberg and Hor's A2 + B1, in which /ha/ + /dm/ > /hadm/, the
word initial spirant yielding to the main syllable initial stop as the prosody
determining element. Then, of course, Vietnamese ultimately reduces the
form to a monosyllable.

Resuming our discussion of the other aspectuals, it will be observed
that the pair VN sap 'about to': BI siap 'ready' are also preverbal forms.
The tone of sap results straightforwardly from its syllable type as CVC (cf.
1.1.(3) above).

For VN dang 'in process': BI sedang 'in process' one must, as with da,
assume a simplification which ultimately drops the first syllable. As to
tone assignment, one expects either level (unmarked) or with a CVN
syllable (no. (3) above), but with Friberg and Hor's rules an initial
should have yielded to d- and resulted in *dng. Perhaps the situation is
more complex historically and Friberg and Hor-type dominance is further
conditioned by other factors. A prime candidate is perhaps an original
stress difference in Austronesian. Specifically, for example, note:

(7) BI 'sudah 'already'

BI se'dang 'in process'
in which VN da derives from an unstressed syllable while dang
corresponds to a stressed one.

The pair VN mai 'continuative': BI masih 'still, yet' do not function
quite like the other aspectuals above, for mai occurs as a post-verbal while
masih is a pre-verbal. Phonetically, one can assume a reduction masih >
maih (i.e. CVH), after which, given the voicing of the initial, the resultant
mai is completely expected (qv. (3) above).

1.2 Desiderative

The following sentences are instances of desiderative modality in VN or
BI:

84


On Austronesian lexicon in Vietnamese

(8)a. VN: em ny muon v nh.

younger sibling this want return home
'He/she (young one) wants to go home.'

b. BI: Dia mau datang sore ini.

3-sg. want return home this
'She wants to come this evening.'

That is, VN muon 'want' and BI mau 'want' both function as conventional
preverbal desiderative forms. The question is 'Do they have any historical
connection?' The similarity of form is strengthened if one assumes that
mubn derives from the nominalised (ke ...an) form ke-mau-an 'a wish'.

If we again invoke consonant dominance (qv. 1.1 (5) and (6) above), the
initial k- would determine one of the three High tones (V, V, V), but why
V (muon), when CVN would seem to predict V (*muon)l Again, the
hypothesis of a nominalisation source may provide an answer, for suffixes
in Indonesian (cf. also Philippine languages) regularly insert a (?) between
vowel sequences, thus (BI):

ke-mau-?an 'wish (n.)'

which would provide the explanatory feature of stop in the final (-V7N) to
produce an expected muon 'want'.

1.3 Equative

Consider the following sentences in VN and BI:

(9)a. VN: Anh toi la gio su*.

older brother my 'is' teacher
'my older brother is a teacher.'

b. B.I: Bahasa Indonesia ialah bahasa kebangsaan
language Indonesia 'is' language national
'Indonesian is the national language.'

Structurally, VN la and BI ialah (lit. i '3rd sg.' -h lah 'emphatic') operate
in remarkably similar ways. Assuming the reduction to one syllable lah,
the problem for tone would be that CVH predicts a form rather than
la. The indication then, is that the Austronesian form, if the connection is
authentic, must itself have been reduced to la at the time it was given a
tonal interpretation in Vietnamese. This account of la is problematical,
however, for Nguyen Dinh Ho has pointed out (pers. commun.) that in
the old Chw Nom data this copular la turns up as lam 'to do', and his
suggestion is that that is, indeed, its origin. Since, on the other hand, it is
acceptable even in Modern Vietnamese to use lam as well as la in a
copular sense, it is not clear to me that they could not be independent
forms.

85


KENNETH GREGERSON

1.4 Pronouns

There appear in Vietnamese a number of pro-forms of a locative, personal
or interrogative nature that bear a good deal of resemblance to
Austronesian5 forms, among which are the following:6

(10) VN kia; kia 'there': UAN *ija 'he, she, it', BI ia, Chmr. gwida,
SAt. hid

VN nay, nay: PAN *iniH2, BI ini 'this' sini 'here', Rade nei (cf.
similar forms in Tai).

VN no [arrogant] '3-sg.', chng no 'they'; PAN *na '3-sg.', Agta na
'3-sg.', BI sana 'there'.

VN ta '[arrogant] I', chng ta 'we incl.': PAN *(k)ita, *ta 'we incl.',
BI kita.

VN ma 'which' (rel. pron.), mo 'what, where?': BI mana 'where,
which' (interrog. pron.).

1.5 Adverbs

(11) VN lau 'long time', luon 'forever': UAN *laun 'duration', BI laun
'to linger, loiter', Pw. laui 'unfinished portion'.

VN rat 'very': UAN *bavat 'heavy', BI berat 'heavy' (cf. VN ho>i lit.
'vapour' = 'rather').

VN lam 'very': Haroi hlam 'very', cf. BI selama 'as long as', selama
lamanya 'at the most'.

VN xa 'far': PMP *za[h]ouq, BI jauh, Chmr. tlaglo.
2. Content forms

Vietnamese has, in addition to the more 'grammatical' forms above, quite
a large number of general lexical items that also bear enough resemblance
to Austronesian forms to have been noted by a number of investigators.

5. Austronesian citations are in general from Dahl (1977). Rade forms are from Egerod (1978),
other Chamic citations are from Burnham (1976). Waic references are from Diffloth (1980).

6. Abbreviations for languages cited without full text references are:

Ami
Bisayan

Bisayan as recorded
in Encarnacin
Bunun
Chmr. Chamorro
CLi. Central Li
Javanese
Kuvalan
Laqua
Makassarese
Madurese

Am.

Bs.

BsE

Bu.

Jav.

Kv.

Lq.

Mak.

Md.

Mlg. Malagasy
MVN Middle (17th-cent-

ury) Vietnamese
NgD Ngaju Dayak
PAA Proto-Austro-Asiatic
PAN Proto-Austronesian,
DahPs constructions
PMP Proto-Malayo-Poly-

nesian
Pu. Puyama
Pw. Paiwan
Pz. Pazeh

Rg. Rengao
Ru. Rukai
SAt. Squliq Atayal
Sir. Siraya
SLi. Southern Li
TB Toba-Batak
Tg. Tagalog
Th. Thao
UAN Uraustronesisch,
DempwolfPs con-
structions

86


On Austronesian lexicon in Vietnamese

2.1. Body parts

(12) VN vai (MVN Sai) 'shoulder': *bava 'shoulder', BI bahu, Pz. la-
baxa, Truk.jafar (Dyer 1953).

VN hr&i 'tongue': UAN *dilah 'tongue', Jav. dilah, BI lidah, Pw.
lidalid, Rade lah, Haroi caliah, Cham talah 'id'.

VN tri 'left hand': PAN *uiri 'left (hand)', BI kiri, Mak. ka-iri,
Pu. tama-wiri'.

VN tai 'ear': PAN *t2alina, BI telinga 'ear', cf. CLi. thai.

VN dau 'head': PMP *qulu 'head', Tg. ?ulo, BI hulu 'upper end,
head', cf. also Lq. ru, SLi. dau 'head'.

2.2. Humans and body functions

(13) VN mu 'blind': UAN *buta, BI buta 'blind', Pw. ma-vutsa, Rade
buum ala? cf. Thai *'bot.

VN t, tro 'show': UAN *tu(n)duh, Jav. tuduh, BI tunjuk 'point',
Tg. turo' 'instruction', Bs. tolo 'finger', Am. to rol 'point'.

VN i 'defecate': UAN *ia(h) 'urine', Jav. p-ih 'water conduit', Bu.
xisah 'urine'.

VN an 'eat': PMP *ka'sn 'eat', Tg. kaa'in, BI makan, SAt. qan-iq.
cf. Rg. kaq 'eat (meat)'.

VN guc 'bend head down': BI angguk 'nod' cf. BI anggut 'nod,
(ship) pitch'.

VN ngd 'take a look': BI anggul 'tip toward, raise head', Rade angii
'look up'.

VN mwa 'vomit': UAN u(n)tah 'vomit' BI muntah; SAt. m-utaq.
Note that, here, VN retains the Austronesian verbal prefix m~.

VN gai 'scratch': UAN *gatal 'itch', Jav. gatal, BI garit, Md. ghatal,
Pw. gatsdl, cf. also BI kais 'scrape for food'.

VN nghe 'hear': UAN *dzT}3 'hear', BI dengar, Tg. dirjig, Pu. ma-
rngai, Rade kna 'ear' (< *taliqa).

VN da (MVN dea) 'belly': UAN *[t]ijan 'belly' BI tian 'abdomen
of pregnant woman', Am. tiai, Th. tiiya.

VN dwng 'stand': UAN *divi 'stand', Rade dook ddij 'stand' (cf. Li
tsuon, Thai *'yn).

VN (nam) me 'dream': UAN *i(m)pi, BI mimpi, Pw. sapi, Rade epei
'dream'.

VN b 'grandmother': UAN baji 'wife, woman', Tg. ba-bai'i Am.
va-vahi\ BsE bayi 'grandmother', cf. also VN vo> 'wife'.

87


KENNETH GREGERSON

VN thwa 'respect form': UAN *tuha/tuva 'old' Jav. tuwa.

2.3. Flora, fauna

VN cay 'tree': UAN *kaju, PMP *kaS2iu 'tree, wood', Bl kayu, Tg.
kaahoy, Am. kasui.

VN bong 'flower': Bl bunga 'flower', Rade my a. cf. PMunda
*ba[g]a, Mon pkao, Aslian *bskaw, PAA *baka[l].

VN ma 'rice seedling': PMP *qumaH 'field', Bl huma 'field for dry
rice', Bu. humaq, SAt. qumah 'work in the field'.

VN trwng (MVN thing): *UAN *taluY 'egg', Tg. litlog, Bl telur,
Am. lita'uy.

VN swa 'milk': UAN *tutu 'female breast', Bl susu 'milk, breast',
Am. tsotso\ Rade ksau 'breast'.

VN cang 'claw, pincer': UAN *bavan 'molar', Tg. bagan 'molar'
Jav. wan 'jaw', Rade kaay 'chin, jaw'.

2.4. World, weather

VN bui 'dust': UAN *dabuk 'ashes, dust, grey', Bl abu 'dust', Jav.
dawu 'grey', ?bruih 'dust'. But cf. Rg. hdpuih 'to dust', Waic. *pes.

VN dat 'earth': UAN *datav 'flat', Bl rata 'flat, level', Tg. lataq
'carpet', NgD datah 'step, rung'.

VN lo> (of cliff, well) 'collapse, slide': UAN *dabt 'leave suddenly',
Tg. lilis 'wipe off, TB dolos 'glide', Jav. ddtes 'keep away', Rade luh
'fall to the ground', cf. Rg. rdlayh 'cave in'.

VN trng/ging (MVN blng) 'moon': UAN *bulan 'moon', Bl
bulan, Rade mlaan, Jarai blan, Pu. votan, Kv. buu ran.

VN sang 'become bright': Bl terang 'clear, bright', cf. VN trang
'white', rang 'dawn'.

VN dim 'night': UAN *dsm dam 'keep quiet', Bl diam 'quiet', Pw.
dzdm dzdm 'last night before full moon'.

VN mai 'tomorrow': UAN *damav 'resin, torch', Sir. madama
'morning', Ru. damar 'moon', Tg. damag 'night' (cf. with dem
above). Cf. Rg. mar 'morning' mar eh 'tomorrow', also Rg. mang
'night'.

VN dwang/dang 'road': PMP *Zalan 'road, path', Bl jalan, Tg.
daan, Bu. daan, Sir. darang, Rade elaan, Haroi calan 'id'.

88


On Austronesian lexicon in Vietnamese

2.5 Miscellaneous verbs

VN ri, trai 'to sow': BI beras 'rice', Rade rah 'sow rice in wet field'.

VN tra, gia (MVN bla) 'pay back': UAN *b9lah 'split', BI belah
'split, part', Pw. vdlaq 'split'. Cf. also VN bira 'split open' (except ~
tone is expected).

VN. ket 'fasten together': UAN *dak9t 'to stick, BI dekat 'near',
Tg. dikit 'joined'. Pw. ddkdts 'to stick'.

VN t&i 'arrive': UAN *ha(n)tad 'deliver, convey', BI antar
'introduce' Jav., Md. atdr, Pw. satddz 'send'.

VN ddi 'deceive, lie': UAN *putsd 'rotate', BI putar 'turn, be
dishonest', cf. Rg. podar 'spin a top, deceive'.

VN nau 'cook': UAN *tunu 'roast', Jav. tunu 'burn, Md. tono(h)
'roast', Pw. ma-tsulu 'hot', Rade m?dau, Roglai pa?dau 'warm'.

VN mat 'lose': UAN *mataj, BI mati 'die, dead'.

VN kiim 'seek': UAN *ki[l]im 'send', BI kirim 'id.', Am., Bu.,
kilim, Pw. kim 'search for'.

2.6. Descriptives

VN sai 'wrong': PMP *t'alaq, BI salah, Tg. sala 'mistaken', Pw. pa-
talaq 'envy, jealous'.

VN sac 'sharp': PMP *hat'aq 'whet', Chmr. gwasa\ Tg. hasa, Pw.
t-ataq 'id'.

VN bu> 'big': UAN *bsYat 'heavy', BI berat 'id.', Mlg. be 'big, great,
many'.

2.7. Miscellaneous

VN cuoi 'end, least': UAN *likud 'back, behind', BI ekor 'tail', Tg.
likod, NgD ba-rikor, Pw. likudz 'behind'.

VN sng 'gun': UAN hterj, Tg. lusoy, Rade suy, Rg. iisuk 'mortar',
Pz. ludzfj, SAt. luhuq.

VN van 'plank, board': UAN *papan, BI papan 'board', cf. Li. pen
'classifier for people', Thai *peen 'plank'.

VN na 'bow': UAN *panah, PMP *panaq 'bow, arrow, shoot', BI
panah 'bow and arrow', Tg. paana, Rade hna, Am. pana'h. Cf. VN
ban 'shoot', and BI panar 'stunned, dull', but also BI senapan
'weapon'.

89


KENNETH GREGERSON

3. Some Phonetic Patterns

The Vietnamese forms cited above in sections 1 and 3, while presenting
a number of 'irregularities', do, on the other hand, exhibit a number of
likely phonetic associations with Austronesian forms in general, now
summarised in 3.1-10.

3.1. VH high tones: AN voiceless initials

VN ket 'fasten together': UAN *dsk3t 'to stick' BI dekat 'near', Tg.
dikit 'joined', Pw. d'dkdts 'to stick'.

VN trwng ( MVN tlwng) 'egg': BI telur, Tg. itlog

VN ta '[arrogant] I', chng, ta 'we incl.': PAN *(k)ita, *ta, BI kita
'we inclusive'.

VN cay 'tree': BI kayu 'tree'.

VN t, tro 'show': Jav. tuduh 'point', Tg. turo' 'instruction'.

3.2. VN low tones: AN voiced initials

VN ba 'grandmother': UAN *baji 'wife, woman', Tg. ba-baii, Am.
va-vahi, BsE bayi 'grandmother' (cf. VN vo> 'wife').

VN mil 'blind': UAN *buta, BI buta, Rade buum ala? 'blind', cf.
Thai *'bot.

VN ma 'rice seedling': BI huma 'field for dry rice', Bu. humaq, SAt.
qumah 'work in the field'.

VN guc 'bend head down': BI angguk 'nod'.

VA mira 'vomit': UAN *u(n)tah, BI muntah, SAt. m-utaq 'vomit'.

3.3. VN'* or ~ tone: AN -s-

VN mai 'continuative'; BI masih 'still, yet'.

VN swa 'milk': BI susu 'milk, breast', Am. tso tso\ Rade ksau
'breast'.

3.4. VN or ~ tone: AN -s/h

VN ia 'defecate': UAN *is(h), Jav. p-ih 'water conduit', Bu. 'isah
'wine'.

VN rai 'to sow': BI beras 'rice', Rade rah 'sow rice in wet field'.
VN Iwai 'tongue': Jav. dilah, BI lidah, Pw. lidalid, Rade lah.

90


On Austronesian lexicon in Vietnamese

3.5. VN or tone: AN final stop

VN dot 'earth': UAN *data 'flat', BI rata 'flat, level', Tg. lataq
'carpet', NgD datah 'step, rung'.

VN sac 'sharp': PMP *hat'aq 'whet', Chmr. gwasa, Tg. haasa', Pw.
t-ataq 'id'.

VN bui 'dust': UAN *dabuk 'ashes, dust, grey', BI abu 'dust', Jav.
dawu 'grey', cf. PMP *abuh, Rade ?bruih, Rg. hdpuih 'to dust',
Waic. *pes.

VN guc 'bend head down': BI angguk 'nod' (cf. 3.2.).

3.6. VN level (unmarked) or tone: AN open syllable I final nasal

VN tai 'ear': BI telinga. cf. CLi. thai 'ear'.

VN ta '[arrogant] I', chng ta 'we incl.': PAN *(k)ita, *ta, BI kita
'we incl.'.

VN an 'eat': PMP *ka'sn 'eat', Tg. kaa'in, BI makan, SAt. qan-iq
'eat'.

VN ma 'which (rel. pn.)': BI mana 'where, which (interrog. pn.)', cf.
also VN mo 'what, where'.

VN dwang, dng 'road': PMP *Zalan 'road, path', BI jalan, Tg.
daan, Sir. darang.

3.7. VN ch-: AN t/pl-

VN chi 'thread': PAN *taliS' 'rope, cord', BI tali, Tg. taali?, Pw.
tsalis, Pz. sariss 'cord'.

VN chuc 'a collection of ten': UAN/PMP *puluh/puluq Tg. pulo\
pulo\ Ru. porok, Pw. puluq 'ten'.

3.8. VN IP or o> : AN -1/r-

VN lu>o>i 'tongue': UAN *dilah 'tongue', Jav. dilah, BI lidah, Pw.
lidalid, Rade lah.

VN trng (MVN thrng): UAN *tsluY 'egg', Tg. 'itlog, BI telur,
Am. lita'uy 'egg'.

VN dwdng, dng 'road': PMP *Zalan 'road, path', BI jalan, Tg.
daan, Sir. darang, Rade elaan.

VN bw 'big': UAN *b3Yat, BI berat 'heavy', Mlg. be, 'big, great,
many'.

91


KENNETH GREGERSON

3.9. VN -i: AN -1/r

VN cudi 'end, last': UAN *likud 'behind, back', BI ekor, Tg. likod,
NgD ba-rikor, Pw. likudz 'behind'.

VN sai 'wrong': PMP *t'alaq, BI salah, Tg. sala 'mistaken', Pw.
pa-talaq 'envy, jealous'.

VN tai 'arrive': UAN *ha(n)tsd 'deliver, convey', BI antar
'introduce', Jav., Md. atdr, Pw. satddz 'send'.

VN doi 'deceive, lie': UAN *putad 'rotate', BI putar 'turn, be
dishonest'.

VN tai 'ear': PAN *talii]a, BI telinga 'ear', cf. CLi. thai.

3.10. VN tr-/gi-/b\-:AN b(v)l-

VN trng/ging (MVN blang) 'moon': UAN *bulan 'moon', BI
bulan, Rade mlaan, Pu. volan, Kv. buu ran.

VN tra/gia (bla) 'pay back': UAN *balah 'split', BI belah 'split,
part', Pw. vdlaq 'split'. Cf. VN bwa 'split open'.

VN tron 'be round': UAN *baluY. cf. Li. (pjluon, Thai *'duag,
'don (Benedict 1966: 246).

4. Concluding remarks

This brief consideration of possible lexical affinities between Vietnamese
and Austronesian makes no claim to far-reaching conclusions. At the
same time, some of the VN forms observed above would, if valid, seem to
call for at least two sources or periods of Austronesian (or Austro-Thai?)
contact in order to explain their contemporary phonological constitution.
These I will, for present purposes, distinguish simply as Immediate vs.
Remote sources. The following sets of vocabulary are illustrative:

Immediate

VN da 'already': BI sudah 'already'.

VN sap 'about to': BI siap 'ready'.

VN dang 'in process': BI sedang 'in process'.

VN la 'is': BI ia lah 'is'.

VN thwa 'respect': BI, Jav. tuwa 'old'.

VN cay 'tree': BI kayu 'tree'.

VN red 'to sow': BI beras, Rade rah 'sow rice in wet field'.
Remote

VN gai 'scratch': UAN *gatdl 'itch', BI garit, Pw. gatsdl. cf. BI kais
'scrape for food'.

92


Full Text

PAGE 1

25 AUSTROASIATIC LANGUAGES Essays in honour of H .L. Shorto Edited by J .H.C.S. Davidson. School of Oriental and African Studies \ University of London I 1991

PAGE 3

Collected Papers in Oriental and African Studies AUSTROASIATIC LANGUAGES Essays in honour of H. L. Shorto Edited by Jeremy H. C. S. Davidson Formerly Lecturer in Vietnamese School of Oriental and African Studies SCHOOL OF ORIENTAL AND AFRICAN STUDIES UNIVERSITY OF LONDON 1991

PAGE 4

School of Oriental and African Studies, University of London 1991 All rights reserved Published by School of Oriental and African Studies University of London Thornhaugh Street Russell Square London WCIH OXG British Library Cataloguing in Publication Data Austroasiatic languages: essays in honour of H. L. Shorto (Collected papers in Oriental and African Studies). I. Davidson, Jeremy H. C. S. (Jeremy Hugh Chauncy Shane) IT. Shorto, H. L. TIT. Series 495 ISBN 0-7286-0183-4 Typeset by PDQ Typesetting, Stoke-on-Trent, England. Printed in England by Hobbs the Printers Ltd., Southampton

PAGE 5

CONTENTS CONTRIBUTORS ................................................................................... vi PREFACE ............................................................................................ vii ACKNOWLEDGEMENTS ...................................................................... viii H. L. SHOR TO: A BIOGRAPHICAL NOTE J. H. C. S. Davidson ....................... 1 PUBLICATIONS OF H. L. SHOR TO Helen Cordell ......................................... 3 AUSTRIC: AN 'EXTINCT' PROTO-LANGUAGE Paul K. Benedict .................... 7 PALAUNGIC VOWELS IN MON-KHMER PERSPECTIVE G. Dijjloth ............. 13 COMMUNICATIVES, EXISTIVES, AND STATIVES IN PROTO-SOUTH-BAHNARIC David Thomas ............................................. 29 A LINGUISTIC ANALYSIS OF SOME SOUTH MUNDA KINSHIP TERMS, I Norman H. Zide & Arlene R. K. Zide ............................ 43 PROBLEMS AND PITFALLS IN THE PHONETIC INTERPRETATION OF KHASI ORTHOGRAPHY teugenie J. A. Henderson .............................. 61 HU -A LANGUAGE WITH UNORTHODOX TONOGENESIS Jan-O/of Svantesson ....................................................... 67 ON AUSTRONESIAN LEXICON IN VIETNAMESE Kenneth Gregerson ............ 81 SEVENTEENTH-CENTURY VIETNAMESE LEXICON: PRELIMINARY GLEANINGS FROM ALEXANDRE DE RHODES' WRITINGS Nguy'Jn Dinh-Hoa ............................................................... 95 THE PHONOLOGY OF KOMPONG THOM CHAM Robert K. Headley ........... 105 ASPECTS OF INTER-CLAUSAL RELATIONS IN KHMU Suwilai Premsrirat ................................................................... 123 AN INSTRUMENTAL STUDY OF CHONG REGISTERS Theraphan L. Thongkum .................................................... 141 KEEPING THINGS UP FRONT: ASPECTS OF INFORMATION PROCESSING IN MAL DISCOURSE STRUCTURE David Fi/beck .............. 161 LES DERIVES DESIDERATIFS EN KHMER Saveros Pou .......................... 183 A DIACHRONIC SURVEY OF SOME KHMER PARTICLES (7th to 17th CENTURIES) Judith M. Jacob .............................................. 193 THE FORM syan IN ANGKORIAN KHMER Philip N. Jenner ....................... 227 OLD MON sChristian Bauer ................................................................. 241 Frontispiece -Professor H. L. Shorto V

PAGE 6

CONTRIBUTORS J. H. C. S. Davidson, formerly Lecturer in Vietnamese, School of Oriental and African Studies, University of London Helen Cordell, Sub-Librarian, School of Oriental and African Studies, University of London P. K. Benedict, Adjunct Professor of Anthropology, Arizona State University, Tempe, Arizona, USA G. Diffloth, Professor, Modern Languages, Cornell University, Ithaca, New York, USA D. Thomas, Summer Institute of Linguistics, International Linguistics Center, Dallas, Texas, USA N. H. Zide, Professor, Department of South Asian Languages and Civilizations, University of Chicago, Chicago, Illinois, USA Arlene R. K. Zide, Department of Foreign Languages, Harold Walsh College, Chicago, Illinois, USA tEugenie J. A. Henderson, Professor Emeritus of Phonetics in the University of London J.-0. Svantesson, Professor, Department of Linguistics and Phonetics, University of Lund, Lund, Sweden K. J. Gregerson, President, Summer Institute of Lingiustics, Interna tional Linguistics Center, Dallas, Texas, USA Nguy~n Dinh-Hoa, Professor, Vietnamese Studies Program, San Jose State University, San Jose, California, USA R. K. Headley, Senior Research Linguist, US Government, Washington, DC, USA Suwilai Premsrirat, Institute of Language and Culture for Rural Development, Mahidol University, Salaya, Nakhorn Pathom, Thailand Theraphan L. Thongkum, Department of Linguistics, Chulalongkorn University, Bangkok, Thailand D. Filbeck, Christian Mission to the Orient, Chiang Mai, Thailand Saveros Pou, Directeur de Recherche, CNRS, Paris, France Judith M. Jacob, formerly Senior Lecturer in Cambodian, School of Oriental and African Studies, University of London (rtd.) P. N. Jenner, Director, Department of Indo-Pacific Languages, University of Hawaii, Honolulu, Hawaii, USA C. Bauer, Lecturer in Linguistics and Mon, Institute of Language and Culture for Rural Development, Mahidol University, Salaya, Nakhorn Pathom, Thailand Vl

PAGE 7

PREFACE The present volume is the fourth in this series to be dedicated to a retired scholar of the School. Here we honour Professor H. L. Shorto, whose standing in the field of Austroasiatic-especially Mon-KhmerStudies is well reflected in the range and quality of the papers offered as a tribute in this volume. The arrangement is, I hope, obvious-opening with a questioning paper by Paul Benedict, then moving from the languages of the rims of the area to focus on the Mon-Khmer heartland of Harry's major interest. The papers do not call for any comment from me as Editor, nor do their authors need introduction. Their contributions speak for themselves and will undoubtedly stimulate further research in the field of Austroasiatic Studies in coming years. Jeremy H. C. S. DAVIDSON vu

PAGE 8

ACKNOWLEDGEMENTS I should like to thank all those who have helped make the task of editing these seventeen contributions lighter. In particular I wish to thank the Publications Committee (especially its Chairman, Professor Shackle, and its Secretary, Mr Martin Daly) for financial support in this publication of the volume. Thanks are due to the Editorial Secretary, Miss Diana Matias and to the SOAS photographer, Mr Paul Fox, for help with its production. The photograph on the cover is by Hans Hinz and is reproduced by permission from Thai painting published by the Office du Livre, Fribourg. viii

PAGE 9

HARRY LEONARD SHORTO: A BIOGRAPHICAL NOTE Professor Harry Shorto is one of the founding scholars of Mon-Khmer studies, the field which his knowledge and insight helped to make the active and expanding area of scholarship that it is today. It is now some forty five years since Harry Shorto completed his MA in Modern and Mediaeval Languages at Cambridge before moving to SOAS to take up a training appointment as Lecturer in Linguistics in 1948. From that point he was to pursue the study of the Austroasiatic and Austronesian language families of his choice: in 1952 he was appointed Lecturer in Mon; in 1964 Reader in the Languages and Literatures of South-East Asia; in 1971 to the Chair of Mon-Khmer Studies in the University of London. During this period he produced his two magisterial dictionaries, A dictionary of Modern Spoken Mon (1962) and A dictionary of the Mon inscriptions from the sixth to the sixteenth centuries (1971), to which he added numerous other stimulating publications listed in the bibliography of this volume. Harry Shorto is a private man, given to weighing his words carefully between frequent re-lightings of his ever-present pipe. His Common Room colleagues may often have found awesome the ease with which he negotiated the depths and further reaches of his unfamiliar linguistic territory, but the learning was always leavened with humour, the delight in language for its own sake infectious. The lightness of touch and passion for detail which Harry brought to the investigation of languages, their histories, constructions, and interconnections, are in complete harmony with his other absorbing interest: early musical instruments, in particular the flute, gamba, cornet and virginal. This interest extends to performance and he was for a time a member of a consort of viols. Today he continues to perfect his technique on his 1960 Morley virginal. To Harry Shorto's many colleagues and former students it must be a source of great satisfaction that retirement has not put an end to his encyclopaedic researches, notably his project for a comparative phonology of the Mon-Khmer languages. We wish him well and offer this volume in appreciation of his contribution to scholarship over the years. Jeremy H. C. S. DAVIDSON

PAGE 11

PUBLICATIONS OF H. L. SHORTO Helen Cordell 1956 Notes on Mon epigraphy. Bulletin of the School of Oriental and African Studies 18(2), 344-52. 1956-57 Quantification in Mon. Proceedings of the 23rd International Congress of Orientalists, Cambridge, 1954. London: Royal Asiatic Society. 278-79. 1958 The Kyaikmaraw inscriptions. Bulletin of the School of Oriental and African Studies 21(2), 361-67. 1960 Word and syllable patterns in Palaung. Bulletin of the School of Oriental and African Studies 23(3), 544-57. 1961 A Mon geneaology of kings: observations on the Nidana Arambhakhatha. In, Historians of South East Asia, (ed.) D. G. E. Hall. London: Oxford University Press, 6372. 1962 A Dictionary of Modern Spoken Mon. London: Oxford University Press. xvi, 280p. 1963 Bibliographies of Mon-Khmer and Tai linguistics, compiled by H. L. Shorto, Judith M. Jacob and E. H. S. Simmonds. London: Oxford University Press. (London oriental bibliographies 2) x, 87pp. Linguistic comparison in South East Asia and the Pacific, (ed.) H. L. Shorto. London: School of Oriental and African Studies. (Collected papers in Oriental and African studies 4). The structural patterns of northern Mon-Khmer languages. In, Linguistic comparison in South East Asia and the Pacific, (ed.) H. L. Shorto. London: School of Oriental and African Studies. (Collected papers in Oriental and African studies 4), 45-61. The 32 myos in the medieval Mon kingdom. Bulletin of the School of Oriental and African Studies 26(3), 572-91. 1965 The interpretation of archaic writing systems, illustrated by the analysis of the phonological systems in early Mon dialects. In, Indo-Pacific linguistic studies pt. 1, (eds.) G. B. Milner and E. J. A. Henderson. Amsterdam: North-Holland. 88-97. 1966 Mon vowel systems: a problem in phonological statement. In, In memory of J. R. Firth, (eds.) C. E. Bazel! et al. London: Longmans, 398-409. 3

PAGE 12

The devata plaques of the Ananda basement In, Essays offered to G. H. Luce by his colleagues and friends in honour of his seventy fifth birthday, (eds.) Ba Shin et al .. Ascona: Artibus Asiae. 2, 156-65. 1967 Die Register-unterschiede in den Mon-Khmer Sprachen. Wissenschaftliche Zeitschrift der Karl-Marx-Universitiit Leip zig, Gesellschaftsund Sprachwissenschaftliche Reihe, 16 (1/2), 245-48. The dewatau sotapan: a Mon prototype of the 37 nats. Bulletin of the School of Oriental and African Studies 30(1 ), 127-41. 1969 Mon labial clusters. Bulletin of the School of Oriental and African Studies, 32 (I), 104-14 1970 The Gavampati tradition in Burma. In, R. C. Majumdar felicitation volume, (ed.) H. B. Sarkar. Calcutta: K. L. Mukhopadhyay, 15-30. 1971 A dictionary of the Mon inscriptions from the sixth to the sixteenth centuries, incorporating materials collected by the late C. 0. Blagden. London: Oxford University Press. (London oriental series; 24), xlii, 406. 1972 The word for two in Austroasiatic. In, Langages et techniques, nature et societe, (eds.) J. Barrau et al. Paris: Klincksieck. 1, 23335. 1973 Three Mon-Khmer word families. Bulletin of the School of Oriental and African Studies 36(2), 374-81. 1975 Achinese and mainland Austronesian. Bulletin of the School of Oriental and African Studies 38(1 ), 81-102. 1976 In defense of Austric. Computational Analyses of Asian and African Languages 6, 96-104. Gayo consonant correspondences. In, South-East Asian linguis tic studies 2. (ed.) Nguy~n D~ng Liem. Canberra: Australian National University. (Pacific linguistics. Series C, 42), 199-217. The vocalism of proto-Mon-Khmer. In, Austroasiatic studies 2 (eds.) P. N. Jenner, L. C. Thompson, S. Starosta. Honolulu: University Press of Hawaii, 1041-67. 1977 Proto-Austronesian *taqan: an anomaly removed. Bulletin of the School of Oriental and African Studies 40(1), 128-9. 4

PAGE 13

1978 The planets, the days of the week and the points of the compass: orientation symbolism in 'Burma'. In, Natural symbols of South East Asia, (ed.) G. B. Milner. London: School of Oriental and African Studies (Collected papers in Oriental and African studies), 152-64. 1979 The linguistic protohistory of Mainland South East Asia. In, Early South East Asia, (eds.) R. B. Smith and W. Watson. London: Oxford University Press, 273-78. 1982 The affinities of Kuy. Bulletin of the School of Oriental and African Studies 45(3), 574-76. 1982 Obituary, Professor Emeritus Walter Simon. Bulletin of the School of Oriental and African Studies 42(2), 344. 1990 Professor Eugenie J. A. Henderson: a personal note. Bulletin of the School of Oriental and African Studies 53(3), 50 I.

PAGE 15

AUSTRIC: AN 'EXTINCT' PROTO-LANGUAGE Paul K. Benedict A leading Mon-Khmerist (Diffloth 1985) has recently raised the question: what happened to Austric? One can dilly-dally over this at great length but the answer is very brief: it became extinct. The Austric hypothesis was developed shortly after the turn of the century by Wilhelm Schmidt (1906), a pioneer in the field of comparative Mon-Khmer linguistics. He had noted a general similarity in morphol ogy, with even some resemblances in prefixes and infixes, between Austronesian (AN) and Mon-Khmer (MK), the latter grouped with the Munda languages oflndia to make up the Austroasiatic (AA) stock. True, he had found nothing in MK to compare with the elaborate suffixial apparatus of AN but, not surprisingly, he skipped lightly over this discrepancy. And what of the anticipated common stock of 'core' (basic) lexical elements? Linguists of that period, and well up into the middle of this century, laboured under the mistaken notion that lexical elements are pretty much the last place to look in setting up linguistic relationships; furthermore, in those pre-Swadesh days, that even when one does get around to inspecting these elements he need only apply an aphorism that Gertrude Stein might have expounded: a word is a word is a word. Henri Maspero, perhaps linguistically the most sophisticated of all the French sinologists of the first half of the century, predictably followed along these lines of thought in wrenching Vietnamese out of its native MK setting and misidentifying it as kindred to the Tai languages. It would have been equally predictable, given the circumstances, that a linguist such as Schmidt would seek to establish a genetic linkage between AN and AA. Thus it can be said that the birth of 'Austric' was expected; rather less expected has been a certain continuing enchantment with it on the part of some scholars, a reluctance to accept its demise. A review of the matter, as attempted in this paper, is thus in order. The writer, when not busy with actual (Sino-Tibetan [ST], Austro-Tai [AT]) rather than fantasied language stocks, has from time to time given some consideration to the basic question here. He at first (Benedict 1942) en passant expressed a willingness to go along with Schmidt's hypothesis on the assumption that continued research in that still largely unexplored field would turn up a respectable body of AN/MK cognate sets for 'core' lexical elements, thereby 'fleshing out' Austric, so to speak. Much later, after a sizeable amount of data in the field had become available, he read a paper on the subject at the First International Conference on Austroasiatic Linguistics at Honolulu (Benedict 1973). Here he played the 'devil's advocate' role, first setting up as good a case as possible for 7

PAGE 16

PAULK. BENEDICT Austric, tying in both phonological and lexical evidence, before finallyand with some reluctance-reaching the conclusion: AT and AA [the earlier levels of AN and MK] do not have a core vocabulary in common, despite the morphological similarity of the two language stocks, hence the idea of an 'Austric' superstock must be abandoned. The homelands of both stocks must be assigned to adjacent areas in South-East Asia (SEA) and the similarities in overall patterning can be ascribed to areal factors. This corner of the Asiatic mainland has, in fact, over the past decades acquired a certain fame among linguists for the vast scope of its areal influences, involving even highly specific processes such as vocalic transfer (Benedict 1979), and there is no reason to suppose that similar influences were not also operative at an early period. In the paper presented in 1973 the writer went rather further in suggesting that a handful of lexical similarities in basic vocabulary between AT and AA are indicative of an early relationship of substratum type between the two stocks. It now seems evident that he overplayed his 'devil's advocate' role in this respect, paying insufficient attention to the alternative explanation in terms of 'look-alikes' or 'comparabilia' (Matisoff 1976), along with areal influences or 'border phenomena'. Thus, the classical pair: PAN *mata (as earlier reconstructed); PMK *mat 'eye' are attention-getting, to be sure, but the syllable reduction is bascially unmotivated (di-as well as mono-syllabic roots occur in MK); PAN *maCa (with *C a cover symbol for reflexes representing proto-level palatal or consonant cluster), as now reconstructed by Austronesianists, weakens the comparison (final *-c is a prominent feature of PMK) while PAT *mapra (for the earlier cited *mapJa; see Benedict 1990) makes matters even worse: finally, PAA *mgt, the likely reconstruction on the basis of the Munda cognates (see Pinnow 1959), takes us even further away from PAT *mapra at the early time level at which these protocomparisons must be made. We can, in fact, do much better by comparing the PST root for 'eye': *mygk (as now reconstructed; see Benedict 1976a), with final *-k > *-t shift after the medial *-y(lacking in AA), even obviating the problem of syllabic reduction! The above is an excellent illustration of the useful linguistic rule of thumb: 'look-alikes' look less and less alike as we attain more precise reconstructions at earlier and earlier levels. Thus, the suggested 'twin/two' set of the Benedict (1973) paper, labelled as 'doubtful' even by Shorto (1976), must now be discarded in the light of recent MK evidence re the medial vocalism 1 while the discussion there of the suggested set for 'dog' overlooked Monie final *-r, which vitiates the comparison. Both comparisons are also burdened with the problem of unmotivated syllabic 1. The recently reported Lai language of Southern China (Guangxi) has /bi'/ for 'two', supporting a PMK reconstruction of *?biar or *?biaar type, in line with Shorto's suggested *bi?aar > ?bia(a)r rather than the *?baar cited in Benedict (1973). See Benedict forthcoming, n.6. 8

PAGE 17

Austric: an 'extinct' proto-language reduction, found also in 'eye' (above) as well as in a fourth set: PAN *busuk; PMK *s:x,k 'hair'.2 A fifth, often cited set: PAN *ikan; PMK *ka (or *ka" ) 'fish' has the same syllabic discrepancy along with 'problems' in both final and initial (the Munda languages point to *qas the PAA initial, as reported in Benedict 1973). These difficulties led the writer to reject the comparison in 1973 and the present reconstruction and analysis of the PAN form (Benedict 1990) supports that decision: *ikan (for the cited *iikan), from *si-ka-n, an archaic derivative ( > Japanese ika 'squid') of the ubiquitous AT 'core' root: *ka/ka" 'eat'. It is entirely possible, of course, that the borrowing of one or more basic roots such as 'hair' (above) will eventually be assignable to the AT/ AA 'border phenomena'. If so, the syllabic reduction factor requires that one set up the direction of the loan(s) as AT > AA. This direction is also required for the early AT > AA culture-word loans, notable those for 'copper' ( > Munda group 'iron') and 'sugarcane', as pointed out in the 1973 paper. The Aslian languages maintain an archaic final *-s (generally > -h -0 in non-Formosan AN languages) in the latter loan and an extensive corpus of Aslian material of this kind recently collected by Geoffrey Benjamin (pers. comm.) includes what appear to be parallel examples of early loans that preserve other archaic phonological features, including even a final *-1 (maintained as a lateral in Formosan languages only). The analysis of this material is still in an early phase and it is unclear how many, if any, of these loans can be established as roots at the PMK (possibly even PAA) level; in any event, they supply additional evidence for the priority of the AT-speaking peoples in the early emergence of 'high culture' in South-East Asia (Benedict 1975). At the 1976 Toronto symposium on AT, Harry Shorto made a spirited defence of Austric, prompting a 'Comment' by the writer (1976b). The argumentation there need not be repeated in this paper; much of it concerns lack of agreement re lexical items. Shorto proposed a number of new AN/MK cognate sets, falling for the most part in what the writer termed the ketketbongbong class, involving lexical areas well outside 'core' vocabulary, with a striking attenuation or gap in this key aspect. As summed up by Benedict (1976b: 106): ... the point to be stressed is that a score of ketketbongbong do not one language stock make, since correspondences of this kind have significant value for comparative purposes only when found in association with a corpus of core vocabulary items. 2. The Formosan evidence, strongly supported by the Japanese (see Benedict 1990), points to an underlying trisyllabic root: *busukas or *bosokas (PA medial *o and *u were merged in PAN *u), which yielded Malayo-Polynesian forms of *busuk > buhuk type. The mid or mid-low height of the PMK vowel-the cited form reflects the Proto-Semai vocalism-thus can readily be explained as a retention rather than an innovation while the length can be interpreted through vocalic transfer (Benedict 1979). This root appears to be unrepresented in the mainland AT languages (Kadai, Miao-Yao); the indicated line of development, with syllabic reduction on-the-left along with vocalic transfer, is of the sort that characterises the Kadai family as a whole. PAT *o is also maintained in this family, which may well have been the source of an early loan (as an alternative to a 'look-alike') for this lexical item. 9

PAGE 18

PAULK. BENEDICT The final clause was italicised lest the point be overlooked, as apparently has occurred in the more recent paper by Diffloth (1985). The nub of the writer's argument here lies in the fact that Shorto, a MK specialist with good access to a wealth of comparative AN data, failed to turn up an even halfway decent corpus of 'core' vocabulary correspon dences. This is indeed the name of the game, at least in the South-East Asia region, on the basis of an apparent linguistic consensus at the present time, and on this basis Austric surely emerged as the loser. And the writer (1976b: 105), even 'at the risk of appearing whimsical', could quite honestly report his belief that Shorto's paper had served only to strengthen the case against Austric! Almost a decade passed before the appearance of another defence of Austric, this also by a leading Mon-Khmerist, Gerard Diffloth (1985). Unlike Shorto, however, who proposed a sizeable number of new AN/ MK cognate sets, Diffloth included only two 'possible Austric etymologies': for 'wood' (-'tree') and 'bone', apparently the prize members of his collection, preferable to certain 'runts' which he modestly kept to himself. The pair that he does offer are in fact 'look alikes' or 'comparabilia', with Diffloth admitting to serious difficulties in reconstruction even at the PMK, let alone a PAA or 'Austric', level. It must also be noted that he has unparalleled MK comparative material at his disposal along with the most recent advances in the AN field, all of which he has made use of, hence his failure to do any better than Shorto simply further strengthens the null case here. Given the present circumstances, the notion that a presentable corpus of 'core' AN/MK (or AT/AA) cognate sets lies 'out there somewhere' waiting to be discovered strikes one as quixotic in the extreme. Diffloth's citation of 'look-alikes' in an undertaking of this nature conforms to general linguistic practice, as does his 'sharpening the reconstructions of promising cases' (Schmidt had proposed the 'wood' etymology). Quite novel, however, is his attack upon what he labels 'Swadesh's "basic" notions', which is no less than a frontal assault upon the whole idea of utilising basic or 'core' vocabulary in comparative work. Granted a certain feeling of frustration on his part in being unable to uncover 'core' AN/MK cognate sets, one can hardly follow him in this rash line of thinking. After pointing out that given items of 'core' type can on occasion be replaced through factors such as 'taboo or euphemistic passe-passe', he makes the following summary statement: Over the millennia, the recurrence of factors like the above [cit. supra] could easily wipe out 100 such 'basic words', or even 200 [i.e. both the standard Swadesh-type word lists], while protecting for us many hidden gems, such as 'to whittle bamboo strips' or 'scruff, not to mention 'smegma' and the like. Benedict may detest it, but the Austric hypothesis is still very much alive. (Diffloth 1985) The writer does not 'detest' Austric; in fact, as an old supporter of the hypothesis he must admit to a certain sentimental attachment to it and a 10

PAGE 19

Austric: an 'extinct' proto-language sadness at its demise. He must confess, however, that he is dismayed at the thought of a proto-language with a corpus of reconstructed roots made up exclusively of 'scruff, 'smegma', and the like. One can hardly visualise comparativists recognising a Proto-Indo-European language put together in such a fashion, and there is no reason for them to accept an 'Austric' of this sort either. South-East Asia has its novelties as a research field, to be sure, but it is hardly extra-linguistic. The cardinal point, as emphasised above, is that roots of this kind, whether called ketketbongbong or not (they need not be reduplicated), can be used in establishing a genetic relationship only in conjunction with 'core' vocabulary. On theoretical grounds, of course, all 100 or even 200 'Swadesh list' roots could have been replaced in a given instance, but how would one prove it, especially in the case of proto-languages, such as PAT and PAA, that occupy contiguous territories? Perhaps, in the dim South-East Asian past, AT and AA not only influenced each other structurally but also exchanged a few scruffy or smegmatic words in the bargain, along with some 'high culture' items after AT had made the early advance towards 'civilisation'. None of this, however, justifies the setting up of an 'Austric' superstock. As reported at the outset of this paper, Austric is 'extinct'. REFERENCES Benedict, P. K. Diffioth, G. Matisoff, J. A. Pinnow, H.-J. Schmidt, W. Shorto, H. L. 1942. Thai, Kadai, and Indonesian: a new alignment in Southeastern Asia. Amer. Anthropol. 44, 576-601. 1973. Austro-Thai and Austroasiatic. First Tnternat. Conf. on Austroasiatic Linguistics, Honolulu (revised version appears as chap. TT in Benedict 1975). 1975. Austro-Thai: Language and cullure. New Haven: Human Relations Area Files Press. 1976a, Sino-Tibetan: another look. J. Amer. Orient. Soc. 96(2), 167-97. 1976b. Shorto--In defense of Austric.-Comment. Comp. Analyses Asian Afr. Langs. 6, 105-8. 1979. Vocalic trnnsfer: a Southeast Asia areal feature. Acta Orientalia 40, 229-52. 1990. Japanese/Auslro-Tai. Ann Arbor: Karoma Press. (forthcoming). How to tell Lai: an exercise in classification (to appear in Ling. Tibeto-Burman Area). 1985. What happened to Austric? Paper presented at the 18th Sino Tibetan Conf., Bangkok. 1976. Austro-Thai and Sino-Tibetan: an examination of body-part contact relationships. Papers for the I st Japan-US Joint Seminar on East & Southeast Asian Linguistics, Tokyo. 1959. Versuch einer historischen Lautlehre der Kharia-Sprache. Wiesbaden: Harrnssowitz. 1906. Die Mon-Khmer-Volker, ein Bindeglied zwischen Vi:ilkern Zentrnl-Asiens und Austronesiens. Archiv frlr Anthropologie 5, 59-109. 1976. In defense of Austric. Comp. Analyses Asian Afr. Langs. 6, 96-104. 11

PAGE 21

PALAUNGIC VOWELS IN MON-KHMER PERSPECTIVE G. Diffloth Thirteen years ago, H.L. Shorto pointed to the vowel system of ProtoMon-Khmer as being the 'crux' in the historical phonology of this family (Shorto 1976). His assessment remains as valid today as it was then, even though some advances in reconstruction have been made; our data base has considerably expanded and improved, but the mirage of Proto-MonKhmer vowels continues to recede, even as we penetrate further into the past. The solution proposed then: vowel variation in the proto-language, is consistent with certain facts which can be observed in several MonKhmer languages spoken today. In Bahnar, Sre, Khmu and Semai, to select but a few, whole families of Expressives (Diffloth 1979; in press) are often built on vowel permutations, and such Expressives occasionally find their way into the prosaic (Non-Expressive) vocabulary; conversely, prosaic words often serve as a starting point for building families of Expressives which differ, for example, only by their major vowel. This has surely contributed to the formation of word-families such as those identified in Shorto (1973). This explanation, however, has its limits: presumably, these processes would have affected a word here and a word there, at different times, but it is difficult to see how it could have pervaded the thousands of items which form the non-expressive lexicon of one language, not to speak of an entire family. Other factors are needed in order to account for the numerous vowel correspondences which have been detected so far. For example, it may well be that the Proto-Mon-Khmer vowel system reconstructed until now, although sizable, is not rich enough for the purpose, and that we need to expand it with some additional phonological dimensions. Tone has been practically ruled out for Proto-Mon-Khmer since the simple tone systems of Buliing and Riang, and the tone-cum-register system of Nyah Kur (Diffloth 1980, 1984) can all be explained as innovations; but the newly recorded Angkuic languages U and Man Met (see below) have four-tone and six-tone systems respectively, the origins of which remain partly unknown for the moment. Then again, Haudri court's account of Vietnamese (VN) tonogenesis has generally been accepted, but it leaves out, as tonally irregular or unexplained, a large number of words which do belong to the indigenous Mon-Khmer stratum of the language. And the recently discovered Palyu language, called Lai in Chinese, also has six tones which may, or may not, turn out to be 13

PAGE 22

G. DTFFLOTH recalcitrant-Palyu is apparently Mon-Khmer (Liang 1986, Benedict, in press), but its position in the family is still undecided. Register is a better candidate. This is, typologically, a well-established feature of Mon-Khmer languages (Huffman 1976). The general consensus is that Register is a relatively recent phenomenon, and Shorto, accordingly, does not reconstruct it for Proto-Mon-Khmer. Ferlus (1979) desribed all Mon-Khmer register systems found until then as being the result of one type of evolution: devoicing of initial consonants. This explanation has long been accepted in the case of Spoken Mon (Blagden 1910) and of Modern Khmer (which ironically has now lost phonation-type distinc tions), and it does account for the registers of several other newly recorded languages (e.g. Kuy, Bruu, Phalok). But it is inadequate in certain other languages: the Pacoh register system, admittedly an innovation, has nothing to do with the process of devoicing which has independently taken place in this language. It is also inapplicable to the North Bahnaric languages where no devoicing has taken place, except in Sedang. In Pacoh, the genesis of register is due to changes in vowel quality, namely, the fronting or backing of certain proto-central vowels (Diffloth 1982), and a similar innovation has apparently also taken place in North Bahnaric (Diffloth 1983). So, we do not have yet a case of reconstructing register as being ancient in Mon-Khmer.1 However, the Pearic branch might force us to do that: recently, Huffman (1985) has shown that Chong, a language of the Pearic branch, had a Clear vs. Breathy distinction, criss-crossing a Plain vs. Glottalised contrast, giving rise in effect to a four-register system. Theraphan (this volume) describes in detail the complex bundle of phonetic features these four registers contain. This phenomenon has no historical explanation, and Headley himself (1985) has abandoned to the sagacity of future historical linguists any attempt in this regard. Gage (1985) has pointed out that certain unexpected occurrences of the sb.c tone in Vietnamese seem to find an echo in the register system of Pearic. The tonogenesis of Vietnamese requires that the sb.c tone occur with final proto-stops, and indeed cannot explain the tones of many VN words which have excellent Mon-Khmer etymologies, such as: b8n 'four', chin 'cooked', gio 'wind', or: ngai 'far'. Cognates to all four of these words happen to have glottalisation in Pearic. In Chong, as I have recorded it, the first three have the 'tight' register:2 /ph(;nn/ 'four', /chjin/ 'cooked', /byaay/ 'wind'; the fourth word has a 'breathy-creaky' register3 : /rre'i!Y/ 'far'. Other examples can be found, e.g. VN: cam, Chong /blJH-m/ 'rice-husk', but there are counter-examples as well, e.g.: VN: chim, Chong /chjim/ 'bird'. Since the Pearic and the Vi~t-Mmhg branches are only distantly related, the implications of this fact could go back directly to Proto-Mon-Khmer. I. Smith's opinion to this efTect (Smith 1972) was not based on the establishment or sound correspondences, but on statistical tendencies within a very small set or possible Mon-Khmer cognates, which a more thorough comparison does not confirm. 2. Clear voice plus glottalisation in Huffman's (1985) analysis. 3. HufTman's (1985) breathy voice plus glottalisation. 14

PAGE 23

Palaungic vowels in Mon-Khmer perspective There are also less exotic vowel features which have not been fully used in Mon-Khmer reconstruction. Diphthong systems can be much richer than the simple *ia and *ua usually proposed; I have reconstructed Proto Katuic with five proto-diphthongs (Diffloth 1982), and Nancowry Nico bar (Radakrishnan 1981 :25) is described even today as also having five diphthongs: /ia/, /ia/, ua/, /ua/ and /wa/, which seem to correspond with what we can reconstruct for Proto-Aslian. Some of these phonological features may have to be reconstructed back to Proto-Mon-Khmer, and could well explain a number of Shorto's variations as being regular outcomes of a much richer pro to-vowel system. But then, the number of proposed Proto-Mon-Khmer etyma becomes a relevant issue, and what has been published so far can be said to represent only a sample. In this paper, I will not explore these possibilities, but only prepare the comparative ground to do so; I will try to clarify some points in the history of vowel systems in the Palaungic branch, where recently recorded material allows us to make systematic reconstructions. This may seem at first to be somewhat irrelevant: if Waic and Palaung are notable for one thing, it is precisely the poverty of their vowel systems. The old vowel length contrast was already lost in Proto-Waic, and the best source of information on Milne's Palaung ( = Ta-ang)4 appears not to have a phonemic contrast of this kind. But the Palaungic branch has an important role to play in reconstruction: it belongs to a distinct division of the family, the Northern Division, and it provides us with an independent testimony for the reconstruction of Proto Mon-Khmer vowels. Besides, as I will try to show, Palaungic vowel systems are not as poor as they first seem to be. 1. ProtoWaic The term 'Waic' covers (1) several Wa languages, e.g. Paraok, Aviia', La (Zhou & Yan 1984) and their dialects; (2) the Phalok language,5 formerly referred to as Khalo or Mae Rim Lawa (Flatz 1970); (3) Lawa and its dialects (Mitani 1972); and (4) the Bulang-Phang complex with its many dialects6 (Diffloth 1980). Certain Waic languages, Lawa and Paraok in particular, currently have rich and complicated vowel systems7 but this is 4. Professor Shorto has let me use his own notes from Riang and from the same Palaung language, Ta-ang, as described in (Milne: 1931); this is the source of the 'Ta-ang' and 'Riang' words quoted here. Let the present article be a small token of appreciation for his kindness. 5. I collected the information on Phalok included here in two separate field trips, one in April 1981 with the help ofTheraphan L. Thonkum, and the second by myself in July of the same year. This research was financed by a grant from the National Science Foundation (NSF) entitled 'An etymological lexicon of Mon-Khmer'. More information on Phalok will be made available in the forthcoming volume 'Wa-Lawa-Bulang'. 6. In that study, I called 'Samtao' a language which later turned out to be identical to that spoken by the Bulang National Minority in Yunnan, China. No linguistic information was available on Bulang at the time, as Zhou & Yan (1983) had not yet appeared. 7. Because of a somewhat artificial analysis, Zhou & Yan (1984) describe Paraok as having 50 vocalic nuclei. 15

PAGE 24

G. DTFFIDTH due, in part, to the influx of Tai borrowings8 and in part to recent processes of vowel warp, conditioned by Registers and final consonants. Only nine proto-vowels are needed at the Proto-Waic stage9 Proto-Waic Vowel system e E a y D 0 :, This maximum system is found with most final consonants, but there are certain distributional gaps; for example, with final *-?, only eight proto-vowels are found (all the above except *o); with final *-h, only seven are found (*Y and *o are excluded); and there are no open final syllables in Proto-Waic. 2. Proto-Palaung-Rumai The term 'Palaung-Rumai' also covers several languages, the best known of which is Ta-ang, i.e. the Palaung of Nam Hsan described by Milne (1931). The Rumai language and its dialects, also belongs heret0, as well as the dialects of Riang; 11 it also includes another distinct group sometimes called 'Pale', which contains at least Da-ang and Na-ang; and several other languages, like Ka-ang and Ra-ang12 Other Palaung-Rumai languages surely await description in Burma or Yunnan, and they may or may not belong to one of the seven groups mentioned here. There is considerable diversity within Palaung-Rumai, but this is not the place to present all the phonological innovations which can now be documented. Mitani (1977, 1979) has already reconstructed the Palaung-8. Lawa has borrowed from Northern Thai and more recently from Standard Thai; Paraok has borrowed from Shan (referred to in China as 'Deh6ng Dai'), Bulang has borrowed from Lti (referred to in China as 'Xishuangbanna Dai', or Xi Dai for short). 9. This reconstruction was presented in (Diflloth 1980); since then, a Paraok-Chinese dictionary has been published (Yan et al. 1981), and, in 1984, I was fortunate to collect Waic linguistic material in China, with the help of Zhou, Z.-Z., under another NSF grant entitled 'Comparison of the Mon-Khmer languages of China with other languages of the Mon Khmer family'; with this new material, the number of reconstructed Proto-Waic words has now more than doubled, and the reconstruction of a few etyma given in Diflloth (1980) has been modified; this new information confirmed the nature of the Proto-Waic system I had reconstructed in 1980 with the help ofY. Mitani. 10. All the Rumai examples quoted here were collected in May 1981, with the help of a family of Rumai emigrants living in Chiang Mai at the time. 11. See note 4 above. 12. The information on Na-ang included here was kindly given to me by Yan, Q.-X. as part of a research programme in China (see n.9). She is the author of a valuable sketch on the Bengl6ng language(s) (Yan 1983). I collected myself the Ka-ang data from a native speaker in KUnming, in the course of the same research programme. The Da-ang and Ra-ang data were collected in 1981 (see n.5), and 1984, during my stays in Thailand. It was not possible for me to determine with precision the geographic spread of these languages, especially for those spoken in the Shan States (Burma) where there seems to be a lot of small-scale migrations. For China, Svantesson et al. (1981), Yan (1983), Zhou & Yan (1983, 1984) provide geographic and demographic information. 16

PAGE 25

Palaungic vowels in Mon-Khmer perspective Rumai vowel system, using older material; and even though none of the sources used had indicated the vowel length contrasts which are clearly present today in most of these languages, he did reconstruct vowel length contrasts at the Proto-Pal-Ru stage. Remarkably, his reconstructions, which he termed as tentative, are confirmed by the better material we now have; this is true at least for the presence of a proto-length contrast, and for the number of proto-vowels: 10 (although this represents a minimum); only the reconstructed quality of some of these vowels can now be improved upon. Briefly, with only a few rare gaps and notation problems, we find the following correspondences in words having a proto final Velar Nasal: 13 Rumai Na-ang Da-ang Ra-ang Ta-ang Ka-ang (1) -aal) -aalJ -aalJ -aal) -al) -aalJ (2) -DI) -:ll) (-:>:>I)) -:ll) -DI) -::JI) -Al) (3) -oon -eep (-EEJ1) (?) -YI) -WI) -01) (4) -al) -al) -Al) -al) -WI) -DOI) (5) -WI) -YI) (-YYI)) -WI) -YI) -UI) -UI) (6) -01) -001) -:lOI) -:lUI) -01) -01) (7) -EEi) -Eal) -EAI) -Eal) -:ll) (-01)) -DOI) (8) -Ell -eep -ii) -el) -ii) -ii) (9) -oon -EEJ1 (-eep) -eel) -El) -el) -eel) (10) -(y)olJ -ial) -ial) -ial) -e::ilJ -ial) Phonetic values for reconstruction of these proto-rimes do not come to mind immediately, to say the least. But the first four are relatively easy: Mitani (1977) reconstructs *aa, *a, *unu, and *ur respectively, for (I), (2), (3) and (4). Correspondence (1), includes items such as: 'bone' Ru, Na, Da, Ra: /b?aaIJ/, Ta: /bn?arJ/, Ka: /b?aaIJ/ 'hawk' Ru: /klaaIJ/, Na: /ma-glaarJ/, Da, Ra: /glaaIJ/, Ta: /klarJ/, Ka: /klaaIJ/ 'elephant' Ru: /saarJ/, Na, Da: /ma-saaIJ/, Ra: /saaIJ/, Ta: /saIJ/, Ka: /saaIJ/ 'house' Ru: /gaaIJ/, Na, Da, Ra: /kaarJ/, Ta: /gaIJ/, Ka: /gaarJ/ 'rancid, sour' Ru: /byaaIJ/, Na: /praaIJ/, Ta: /brarJ/, Ka: /braarJ/ 'torch, lamp' Ru: /raaIJ/, Ta: /rarJ/. Riang-Lang cognates have a back /a/ in these words: /tsn?aIJ/ 'bone', /klaIJ/ 'kite', /si tsaIJ/ 'elephant', /kaIJ/ 'house, shrine', /prdrJ/ 'sour, acid, rancid'. Waic cognates always have Proto-Waic *a: *s;)?arJ (Diffloth 1980: N5) 'bone'; *klaIJ (N69) 'hawk'; bsarJ (N79) 'elephant' *?Ij-gaIJ (N12) 'scabbard' (a derivate, from an earlier verb *gaaIJ ('to cover, to protect'); *raIJ (N53) 'brilliant, bright'. 13. Evidently, some vowels have caused PPal-Ru *-1) to palatalise to -Jl, and sometimes further merge with *-n; but both *-Jl and *-n must be reconstructed at the proto-Palaung stage, in contrast with *-1). 17

PAGE 26

G. DIFFWTH Lamet cognates have a long /aa/: /x'd?frnrJ/ 'bone', /klaarJ/ 'hawk', /bsaarJ/ 'elephant', /kaarJ/ 'shelter, shell, bark, husk'; And in the Angkuic branch of Palaungic, Man Met14 has /-aarJ/ finals with a low-falling tone, and U14, which does not have vowel length contrasts, shows finals in /-arJ/ with a high-falling tone: MM: /?aarJ/, U: /s'd?arJ/ 'bone'; MM /khaarJ/, U: /khlarJ/ 'hawk'; MM: /saarJ/ 'elephant'; U: /karJ/ 'house'; MM: /haarJ/ (high-rising tone unexplained) 'bright'. Outside Palaungic, Mitani's reconstruction of *-aarJ is confirmed, for instance, by Khmu: /c'd?aarJ/ 'bone', /klaarJ/ 'hawk', /s'dcaarJ/, 'elephant', /gaarJ/ 'house' Note that we will not be concerned here with the history of initial stops in these languages, interesting as that may be. In a nut-shell, Rumai and Ta ang have preserved the original values of pro to-voiced and pro to-voiceless stops; the same state of affairs also exists in Ka-ang, except that initial *pand *thave become implosive 6-and cfrespectively; in Ra-ang, *p-and *thave followed the same evolution, and *khas become a voiced stop g-, while all proto-voiced stops have become voiceless; Da-ang and Na-ang have followed the same course as Ra-ang, and, in addition, have lost the implosion of 6-and cf-, which then become ordinary b-and d-; Da-ang and Na-ang therefore show a total reversal of the voicing values of PPal-Ru stops: another illusion of 'flip-flop', with implosiveness as the point of transition in this game of musical chairs. Angkuic, on the other hand, has undergone a completely regular 'Germanic'-type of sound change where proto-voiceless stops are now aspirated, and proto-voiced stops are now voiceless. But note that Angkuic tono-genesis is mainly due to vowel qualities and quantities, with some influence from the final consonant; it is unrelated to the earlier or the present voice features of initials. Correspondence (2) includes: 'bitter, gall' Ru: /sorJ/, Na, Da: /s-:JfJ/, Ra: /sofJ/, Ta: /S'dfJ/, Ka: /SAfJ/ 'thatch-grass' Ru: /plofJ/, Na: /bb-:JfJ/, Da: /bbfJ/, Ra: /6lofJ/, Ta: /pl'dfJ/ 'bamboo-shoot' Ru: /bofJ/, Na: /p-:J-:JfJ/, Ta: /b'dfJ/, Ka: /bAfJ/ 'house-pole' Ru: /rorJ/, Na, Da: /rJfJ/, Ka: /rAfJ/ 'horse' Ru: /mbyofJ/, Da: /mpr-:JfJ/, Ta: /br'dfJ/, Ka: /brAfJ/ Riang-Lang cognates have a front /a/ in these words: /tsarJ/ 'bitter', /plarJ/ 'thatch-grass', /knrarJ/ 'post, upright', /mrarJ/ 'horse' 14. I recorded Man Met and U from native speakers, in Yunnan, during the research project mentioned above (n.9); they were introduced to me as speaking 'dialects', or more exactly 'fangyan', of the Biilang language. The Chinese term 'fangyan' corresponds most of the time to what Western linguists consider to be different languages; this was true of these 'fangyan' ofBiilang which do not even belong to the Waic branch of Palaungic, but to the little known Angkuic branch (Diffloth, 1974). The location of the U language is given in (Zhou & Yan 1973), Man Met is spoken a few miles from Jinghong, Xishuangbanna, Yunnan. 18

PAGE 27

Palaungic vowels in Mon-Khmer perspective Proto-Waic cognates have *o: *sol) (N80) 'bitter', *plol) (N73) 'thatch-grass', *t;}bDlJ (N44) 'bambooshoot', *?rolJ (N56) 'house-pole', *mrolJ (N58) 'horse'. Lamet has a short /a/: /cal)/ 'bitter'; /plal)/ 'thatch grass'; /t;}palJ/ 'bamboo-shoot'; /mxalJ/ 'horse' And in the Angkuic branch, Man Met has /-al)/ rimes with a high-rising tone, and U has /-ak/ rimes with mid-rising tone: MM: /salJ/, U: /chak/ 'bitter'; MM: /phalJ/ U: /phlak/ 'thatch-grass'; MM: /halJ/, U: /crak/ 'house-pole'; MM /pal)/, U: /mbrak/ 'horse'. Here again, Mitani's reconstruction of a short *a is confirmed, outside Palaungic, by Khmu which regularly shows cognates with short /a/: /calJ/ 'bitter', /tbal)/ 'small bamboo-shoot', /cndralJ/ 'house-pole', /mbral)/ 'horse'. The contrast between correspondences (1) and (2) shows that length vs. shortness has been preserved everywhere in the Palaung-Rumai lang uages, at least for this pair of vowels, except in Ta-ang which does not show any length distinctions anywhere in its system. It also shows that in the Angkuic branch, tonogenesis is directly due to vowel-length, and has nothing to do with the proto-voice feature of initials; this kind of tonogenesis is unique in the Mon-Khmer family, but, annoyingly, it accounts for only some of the tonal contrasts found in Angkuic languages. It should also be noted that Man Met does undergo this kind of tonogenesis even though the older length distinction is retained; in U however, the length contrast, before disappearing, leaves another trace in the final consonant; it de-nasalises final nasals after short vowels. Another remark: if we only had the Waic and the Riang-Lang material at our disposal, it would appear that a so-called flip-flop has taken place: Waic has a front *-awhere Riang-Lang has a back /a/; and vice-versa: Waic has a back *-ofor Riang-Lang's front /a/. The former presence of a length contrast shows this apparent flip-flop to be nothing but a synchronic illusion. Correspondences (3) includes: 'high' Na: /leep/, Ta: /hlUilJ/, Ka: /hlol)/ 'to dig (a hole)' Ta: /kUil)/ 'yarn' Ru: /soon/, Na: /seep/, Ra: /SYlJ/, Ta: /SUilJ/, Ka: /sol)/ 'foot' Ru: /joon/, Na: /cr.r.p/, Da: /ceen/, Ra: /CYlJ/, Ta: /jUilJ/, Ka: /jol)/ Riang-Lang has an /:)/ reflex: /k51J/ 'to dig', /ts61J/ 'foot' and Proto-Waic shows *-o-: *biol) (N77) 'high', *kol) (N9) 'to dig', *jol) (N18) 'foot' Lamet has /ee/: /!eel)/ 'high, long'' /keel)/ 'to dig'' /ceel)/ 'foot'. 19

PAGE 28

G. DIFFLOTH In Angkuic, Man Met has long /ee/ with a low falling tone, U has /e/ with a high-falling tone and a final nasal: MM: /!eel) U: /hlelJ/ 'high, long'; U: /khelJ/ 'to make with a dibble stick'; MM: /ceelJ/, U: /eel)/ 'foot'. Outside Palaungic, Khmu has cognates with /ia/: Khmu Yuan (Svantesson, personal notes): /khial)/ (aspiration unex plained) 'to dig up', Southern Khmu: /jial)/ 'foot'. Correspondence (4): 'bamboo' Ru: /hral)/, Na: /hr':JI)/ (Vowel?), Da: /hrAIJ/, Ra: /hralJ/, Ra: /hrw.lJ/, Ka: /~1001)/ 'bed-bug' Ru, Na, Ra: khalJ/, Ta (Milne, 1931): kong 'stalk, trunk, post' Ru, Na: /ta!)/, Ta: /ta!)/ (Milne: t~ng, tong), Ka: jo.ODI)/ 'meat' Ru, Na: /yal)/, Ta: /yw.1)/, Ka: /yoolJ/ The Riang reflex is /a/: /r~IJ/ 'bamboo', jeJIJ/ 'tree-trunk', /y;JI)/ 'meat' Proto-Waic has *-Y-: *hYIJ (Paraok: /hul)/, Drage's Wa: hong, Phalok: /hul)/, Phang: /hti:cl)/ 'bed-bug' Lamet has a short /a/: /di)/ 'bamboo', /h~IJ/ 'bed-bug' In Angkuic, Man Met has a short fa/ with a high-rising tone (but the tone of 'bed-bug' seems to be low-rising), while U has a fa/ with a mid-rising tone and de-nasalisation of the final: MM: /h~IJ/, U: /hr::ik/ 'bamboo'; MM: /s::ilJ/ (tone?), U: /s::ik/ 'bed-bug' Outside Palaungic, Khmu has a short /w/: /hw.1)/ 'bed-bug' Another etymon with the same proto-vowel, unfortunately without Palaung-Rumai attestations, is: 'horn': Riang /kmdl)/, Proto-Waic *?rYIJ (N62), Lamet /kniJ.1)/, Angkuic: MM: /k~IJ/, U: /kr::ik/; Khmu: /cndrw.1)/. Mitani's reconstruction of a length contrast, *ww vs. *w, for these two proto-vowels is confirmed independently by the evidence of Lamet and Angkuic (both Man Met and U); in Palaung-Rumai, Na-ang and Rumai also display the same contrast, which appears to be a retention, not only from Proto-Palaungic, but from still earlier periods, as the Khmu evidence indicates. For all four proto-vowels *aa, *a, *ww and *w, Na-ang and Rumai have thus preserved the older length feature. I depart slightly from Mitani in terms of vowel-qualities: (3) and (4) were probably central mid-vowels, *aa and *a. This would explain how (3) became *ee in both Lamet and Angkuic and *o in Proto-Waic, and why (4) has very open vowel quality reflexes throughout Palaung-Rumai, except in Ta-ang where (3) and (4) merged and were pushed higher to /w/ by the shift of *a to /a/. The original value of (4) is preserved in Riang, Lamet and Angkuic. 20

PAGE 29

Palaungic vowels in Mon-Khmer perspective For the remaining vowel correspondences, Mitani does not reconstruct any length contrast. For (5), (6) and (7) he proposes a back-rounded series: *u *o and *:, respectively. Correspondence (5) includes: 'big village, country' Ru: /krnTJ/, Ta: /kuTJ/ 'drum' Na: /grYYTJ/, Ra: /grYTJ/, Ta: /kruTJ/, Ka: /krwTJ/ 'bamboo-strip mat' Ru: /blwTJ/, Ra: /plYTJ/, Ta: /bluTJ/ 'love, like' Ru, Da: J?wTJ/, Ra: /?YTJ/, Ta: /?uTJ/ 'to bury' Ta: /krpuTJ/, Ka: /b6uTJ/ (Vowel ?) The corresponding PWaic vowel is *i: *kiTJ (NIO) 'wet field, country', *kriTJ (N63) 'drum', *krpiTJ (N41) 'to bury' In Angkuic, Man Met has a short /u/ with high tone, and Uhas /u/ with a mid-rising tone and de-nasalisation of the final: MM: /khuTJ/, U: /khiik/ 'wet rice-field' And outside Palaungic, Khmu has /u/: /kuTJ/ 'village' Lamet cognates are missing, although there are other etyma with apparently the same proto vowel, but without Palaung-Rumai attestations; these show the Lamet reflex to be short /u/: 'to blow': Lamet /puTJ/, PWaic *piTJ (N40), Man Met: /phuTJ/, Khmu /puTJ/ 'a sprout': Lamet: /pluTJ/ (tone?), PWaic *bliTJ (Paraok /pla;rn/), Khmu: /bluTJ/ For this correspondence, the evidence for proto-shortness is clear: in Palaung-Rumai, only Na-Ang has a long vowel (in a single item which could have been misrecorded); all other languages where a length contrast exists, Rumai, Da-ang, Ra-ang, Ka-ang, Lamet, Angkuic, Khmu, have a short reflex. The other two correspondences (6) and (7), seem, by contrast, to be on the long side: Correspondence (6) 'knee-cap' Ru: /gyorJ/, Da: /kr:,uTJ/, Ra: /kr:,oTJ/, Ta: /gr:,TJ/, Ka: /groTJ/ 'male bird' Ru: /koTJ/, Ta: /?g-bTJ/, Ka: /koTJ/ 'Classifier: round objects' Ru: /poTJ/, Ta: /p:,TJ/, Ka: /6oTJ/ 'buttocks' Na: /sgpoorJ/, Da: /sgp:,uTJ/, Ra: /sgp:,oTJ/, Ta: /sgb:,TJ/ (my own recording, cf. Milne s~-bong which would indicate /sgboTJ/) 'far, long' Ru: /doTJ/-/ndoTJ/, Na: /tooTJ/, Ra: /t:,oTJ/, Ka: doTJ/ The Proto-Waic reflex is *o: *STJfOTJ (N61) 'knee-cap', *koTJ (NS) 'peacock' Lamet has a long Joo/ reflex: /kxoorJ/ 'knee' In Angkuic, Man Met has a long Joo/ with a low-falling tone, and U has an Jo/ with high-falling tone but no denasalisation of the final: MM: /kooTJ/ 'knee'; U: /kh6TJ/ 'peacock', U: /ph6TJ/ 'round object' (in the expression: /kh'ik ph6TJ/ 'pubis', where /kh'ik/ = 'head') 21

PAGE 30

G. DIFFWTH Correspondence (7) is well documented and contains well-known etyma; it also includes some surprising Front reflexes for what is certainly a proto Back vowel: 'hornet' Ru: /?f.f.fJ/, Da /?f.AfJ/, Ta /?:JfJ/, Ka: /?oorJ/ 'rainbow' Ta: /p';Jry:JfJ/, Ka: /?';JyoorJ/ 'back (of body, of knife)' Ru: /kyf.f.fJ/, Ra: /grf.arJ/, Ta: /kr:JrJ/. Ka: /kroorJ/ 'stairs' Ru: /ndf.f.fJ/, Na: /ntrnrJ/, Ka: /doorJ/ In Riang, the normal reflex is a diphthong /ua/ or / m/: /?uarJ/ 'hornet', /prJlU:JfJ/ 'rainbow', /rnrJdUarJ/ 'stairs' Proto-Waic regularly shows *:J: *?:JfJ (N6) 'hornet', *pry:JfJ (N84) 'rainbow', *kr:JfJ (N60) 'back' Lamet has a long j:J:Jj reflex: /?5:JfJ/ 'hornet', /pxy5:JfJ/ 'rainbow', /kx5:JfJ/ 'back' In Angkuic, the Man Met reflex is a long j:J:Jj with a low falling tone, and the U reflex is a diphthong /ua/, with a high-falling tone and no denasalisation of the final: MM:/?:brJ/, U: /?fiarJ/ 'hornet'; U: /phyfiarJ/ 'rainbow'; U: /?arJ-ghrfiarJ/ 'backbone' Outside Palaungic, the Khmu reflex is regularly a long /:J:J/: /?:J:JfJ/ 'hornet', /pry:J:JfJ/ 'rainbow', /kndr:J:JfJ/ 'back', /rfJd:J:JfJ/ 'stairs' Mitani was certainly justified, on phonemic grounds, in leaving the feature of length unspecified in the back vowels series of Proto-PalaungRumai; but we can afford to be more precise now, and say that *u was probably short, while *o and *:J were probably long *oo and *:J:J. This will allow for an easier description of the gradual collapse of the older vowel length system in languages like Ta-ang and Proto-Waic. And for Front vowels, the same line of reasoning will help us to solve a curious problem. The proto Front Vowel system was also left unspecified as to length by Mitani, who reconstructed *i, *e, and *f. for our correspondences (8), (9) and (10) respectively. Correspondence (8): 'to sew' Ru: /jrn/, Na: /ceep/, Ra: /cerJ/, Ta, Ka: /jirJ/ 'head' Ru: /krn/, Na: /geep/, Da: /girJ/, Ra: /gerJ/, Ta, Ka: /kirJ/ 'navel' Ru: /bdrn/, Na: /bteep/, Da: /btirJ/, Ra: /bterJ/, Ta: /brdirJ/, Ka: /bdirJ/ 'bamboo water-container' Ru: /drn/, Ra: /terJ/, Ta, Ka: /dirJ/ 'husband' Ru: /mrn/ Riang has an /i/ reflex for this correspondence: /kirJ/ 'head', /kndirJ/ 'navel' But PWaic has an *e: *jerJ (Nl 7) 'to sew', *kefJ (N7) 'head', *krdefJ (N30) 'navel', *defJ (Paraok: /t~p/) 'bamboo water container', *hmerJ (N47) 'male of animal' Lamet normally has a short /i/, except for one case of long /ii/: 22

PAGE 31

Palaungic vowels in Mon-Khmer perspective /ciIJ/ 'to sew', /kiIJ/ 'head', /til)/ 'bamboo-container', /miiIJ/ 'male of animal, endearing term for son' This last item may actually be one of the many words Lamet has borrowed from the neighbouring Khmu; unfortunately, I have not yet found the expected word /hmiiIJ/ in any dialect of it. But the Khmu reflex of correspondence (8) is indeed a long /ii/: Khmu Yuan (Svantesson, personal notes): /kiitiiIJ/ 'navel', /tiiIJ/ 'bamboo water-container' In Angkuic, Man Met has a short /i/ with a high-rising tone, and U has /i/ with a mid-rising tone and denasalisation of the final Nasal: MM: /khiIJ/, U /kh'ik/ 'head'; MM: /rp.t'iIJ/ (tone ?) 'bamboo con tainer'; 15 MM: /-til)/ 'the middle of (in a compound: /kaIJ-/)'; U: /hm'ik/ 'male of animal' The Paluang-Rumai evidence, except for Na-ang, points to a proto-short vowel *i, and this is confirmed at the Proto-Palaungic level by Lamet and Angkuic. Correspondence (9): 'sky' Ru: /ploon/, Na: /blcEJ1/, Da: /bleeIJ/, Ra: /6lcIJ/, Ta: /pleIJ/, Ka: /6leeIJ/ 'way, path' Ru: /ndoon/, Na: /nteep/, Da: /nteeIJ/, Ra: /?antEIJ/, Ta: /(ra)-deIJ/, Ka: /deeIJ/ 'yellow' Ru: /toon/, Na: /deep/, Da: /teeIJ/, Ra: /oEIJ/, Ta: /teIJ/, Ka: joeeIJ/ 'wheel' Ru: /bloon/, Ra: /blcp/, Ta: /bnleIJ/ 'equal amount' Ta: /krpreIJ/ Riang cognates have an /e/ vowel: /pleIJ/ 'sky', /raIJdeIJ/ 'way', /bnleIJ/ 'wheel', /trkreIJ/ 'equal amount' And Proto-Waic, surprisingly, has an *i: *kliIJ (Paraok: /kliIJ/) 'to spin (yarn)'; *mriIJ (Bo Luang and Umphai Lawa: /mbriIJ/) 'to compare quantities, to match' The Lamet evidence is, unfortunately, limited to a single item: /tampliiIJ/ 'sky' In Angkuic, I do not have Man Met cognates, but U has the same reflex as in correspondence (8): /i/ with mid-rising tone and denasalisation: U: /phHk/ 'sky' This merger of (8) and (9) is specific to U and not general in Angkuic: there are other cognate sets, unfortunately without attestations in Palaung-Rumai, where this U rime corresponds to Man Met /-eeIJ/ with middle tone, (and to Proto-Waic *-ii), as correspondence (9) requires): 'to return home' U: /?1k/, PWaic *?ii) (NI) 'wall, partition' U: /ndhik/, Mok (a close relative of Man Met spoken in Thailand, Wenk's (1965) 'Ya Ang Lawa'): /theeIJ/, PWaic *ntil) (N2l) 'ginger' U: /sakh'ik/, MM: /kheeIJ/, PWaic *sIJkiIJ (Paraok: /saIJgiIJ/, La: /kiIJ/, Avi:a': /saIJkiaIJ/, Phalok: /kwIJ/, Bulang (Da-Luo): /kakiIJ/, Lawa (Umphai): /saceIJ/ 15. The initial /m-/ is probably a trace of the word /?oom/ 'water' with which *di!) often forms compounds, e.g. Bulang /?um teTJ/ 'bamboo water-container'. 23

PAGE 32

G. DIFFLOTH I propose to reconstruct PPal-Ru *ee for correspondence (9), and for PPalaungic as well. The Palung-Rumai reflexes indicate a long vowel, even in Rumai, where the strange reflex, /-oon/, represents a merger with *-gglJ (correspondence 3), itself a proto-long vowel. This reconstruction also provides a simple explanation for what would appear to be yet another case of flip-flop: if we kept Mitani's reconstruction, without length, the correspondences would be: (8) *-i-: PPal-Ru *-i-= PWaic *-e(9) *-e-: PPal-Ru *-e-= PWaic *-iIf, however, we reconstruct PPal-Ru *ee for (9), head-on collisions are easily averted, and gradual phonetic change can proceed smoothly. In addition, we can also explain the Lamet and Angkuic reflexes: in those two sub-branches, Proto-Palaungic *gg (correspondence 3) was fronted to a long /ee/, pushing the older *ee (corr. 9) out of the way: in Lamet, this *ee was raised to /ii/, keeping its long feature and filling a gap in the vowel system; but in Angkuic, Man Met and U evolved differently: in Man Met *ee and *gg merged in terms of qualities and length, but seem to have acquired different tones, whereas in U, *ee simply merged with *i. If this is correct, the movement of *gg to /ee/ is probably not an innovation shared by Lamet and Angkuic; it may have happened independently in these two sub-branches which do not appear to be especially closely related to each other. The last correspondence, (10), poses a special problem in that it is difficult to decide if *EE or *ig should be reconstructed, for either PPal-Ru or PPalaungic. In any event, a short *E seems a very unlikely value for this correspondence in view of its reflexes: 'to drink' Ru: /col)/, Na, Da: /dial)/, Ra: /cI.iaIJ/, Ta: /tegIJ/, Ka: /cI.iaIJ/ 'excrement' Ru: /?yoIJ/, Na, Da: /?iaIJ/, Ta: /?egIJ/, Ka: /?iaIJ/ 'rice straw' Ru: /hrooIJ/, Na: /r,iaIJ/, Ra: /hriaIJ/, Ta: /hregIJ/, Ka: /~1ia1J/ 'oily, unctuous, delicious' Ru: /pyoIJ/, Na: /briaIJ/, Ra: 6riaIJ/, Ta: /pregIJ/, Ka: /priaIJ/ 'wrist, ankle; bracelet' Ru: /kyoIJ/, Na: /giaIJ/, Ta: /kegIJ/ The Riang reflex is a diphthong, usually transcribed /ii:/: /tii::IJ/ 'to drink', /yoIJ/ 'excrement', /riEIJ/ 'straw', /priEIJ/ 'delicious' In PWaic, the reflex is *i::: *?EIJ (N3) 'excrement' The normal Lamet reflex is a long /EE/, although 'excrement' has unexpected reflexes, possibly due to euphemistic deformation: /tEEIJ/ 'drink', /?EIJ/-/?aIJ/ 'excrement', /prEEIJ/ 'oily', /k51)-kEEIJ/ 'elbow' In Angkuic, the Man Met reflex is a short /i::/, usually with a high tone, whereas in U, we find a diphthong /ia/ with a high-falling tone and a Nasal final; 'excrement' is irregular only in U, having what seems to be an otherwise unattested proto-short diphthong: MM: /thEIJ/, U: /thiaIJ/ 'drink'; MM: /?EIJ/, U: /?eak/ 'excrement', MM: /mphi:IJ/, U: /phrfaIJ/ '(pig) fat' 24

PAGE 33

Palaungic vowels in Mon-Khmer perspective Only one cognate has been found so far in Khmu 16 : /kial)/ 'elbow'. For the sake of consistency with Mitani's notation, I will arbitrarily reconstruct *EE for this correspondence at the PPal-Ru level. The vowel system reconstructed for PPal-Ru now appears as follows: Proto-Palaung-Rumai Vowel System *ee *' I *u *oo *EE *aa *a In contrast to the 9-vowel system shown above for Proto-Waic, this 10vowel system shows the vowel length contrast to be still operating, but with a small functional load. I have not attempted here to reconstruct the Proto-Palaungic state of affairs, because Lamet and Angkuic are still too poorly known; but it seems likely that the vowel system will be, if anything, richer at that stage, probably filling some of the gaps in the long vs. short contrasts. The Proto-Palaung-Rumai vowel system appears to be, typologically, half-way between a full South-East Asian system, as found for example in Khmu or Standard Thai, and a more contracted system where vowel length has been lost, as found for example in Proto-Waic, and Modern Mon. This vowel-system contraction seems typical of the Burma-Yunnan linguistic sub-area. But what Palaung-Rumai shows is that areal pressure does not work like a stream-roller: Ta-ang did lose the length contrast, but Rumai, even with two mergers, maintains it systematically and even innovates in this respect with complete disregard for its more forceful neighbours. Something similar can be said of U: while it did lose vowel-length and acquired tones, thus conforming to its neighbours, it did so in a way which is competely original since the tono-genesis of U is partly due to vowel length. In the perspective of Proto-Mon-Khmer, we may have to cope with as many upheavals, during that long stretch of history which separates us from these ancient times, as we can see in the relatively short adventure of Palanguic vowels. REFERENCES Blagden, C. 0. Diffioth, G. 1910. Quelques notions sur la phonetique du talain et son evolution historique. J. AsiaJique (1 ome ser.) 15, 477-505. 1974. Austro-asiatic languages. Encyc/opedia Britannica (15th ed.) 2, 480-4. 16. In spite of its conservative phonology, Khmu is not as useful for reconstructing Proto Palaungic as it might appear at first glance; it is true that the Khmuic branch is closer to Palaungic than to other branches of Mon-Khmer, but the Khmu lexicon has undergone a great many replacements. 25

PAGE 34

G. DIFFIDTH Diffioth, G. Ferlus, M. Flatz, G. Gage, W.W. Headley, R. K. Jr. Huffman, F. H. Liang, Min Milne, Leslie Mitani, V. 1979. Expressive phonology and prosaic phonology in Mon-Khmer. T n Studies in Tai and Mon-Khmer phonetics and phonology in honour of Eugenie J. R. Henderson (eds.) T. L. Thongkum et al. Bangkok: Chulalongkorn Univ. Press. 49-59. 1980. The Wa languages. (=Ling. Tibeto-Burman Area 5(2), 1-182.) 1982. Registres, devoisement, timbres vocaliques: leur histoire en Katouique. Mon-Khmer Stud. 11, 47-82. 1983. Parthenogenesis of register in North Bahnaric. Paper read at the 9th Annual Meeting of the Berkeley Ling. Soc., Berkeley. 1984. The Dvaravati-Old-Mon language and Nyah-Kur (Monie Lang. Stud.1). Bangkok: Chulalongkorn Univ. Printing House. (in press) 'T' big, 'a' small. Tn Conference on sound symbolism (eds.) J. O'Hala & L. A. Hinton. Berkeley: Univ. California Press. 1979. Formation des registres et mutations consonantiques dans Jes langues mon-khmer. Mon-Khmer Stud. 8, 1-76. 1970. The Khalo or Mae Rim Lawa, a remnant of the Lawa population of Northern Thailand. J. Siam Soc. 58(2), 87-104. 1985. Vietnamese in Mon-Khmer perspective. Tn Southeast Asian linguistic studies presented to Andre-G. Haudricourt (eds.) Suriya Ratanakul et al. Bangkok: Mahidol Univ., 493-524. 1985. Proto-Pearic and the classification of Pearic. Tn Southeast Asian linguistic studies presented to Andre-G. Haudricourt (eds.) Suriya Ratanakul et al. Bangkok: Mahidol Univ., 428-78. 1976. The register problem in fifteen Mon-Khmer languages. Austroasiatic Stud. 1 (Oceanic Ling. Spee. Pub!. 13). (eds.) P. N. Jenner et al. Honolulu: Univ. Hawaii Press, 575-89. 1985. The phonology of Chong, a Mon-Khmer language of Thailand. Tn Southeast Asian linguistic studies presented to Andre-G. Haudricourt (eds.) Suriya Ratanakul et al. Bangkok: Mahidol Univ., 355-88. 1986. On the genealogical classification of the Lai language. Computational analysis of Asian and African languages 26: 1-14. [Tokyo.] 1931.A Dictionary of English-Palaung and Palaung-English. Ran goon: Govt. Printing and Stationery. 1972. A short vocabulary of Lawa (ra-wa-go goi shiryo). Ti5nan Ajia Kenkyfl 10(1), 131-68. 1977. Palaung dialects: a preliminary comparison. Ti5nan Ajia Kenkyfl 15(2) 193-212. 1979. Vowel correspondences between Riang and Palaung. Tn Studies in Tai and Mon-Khmer phonetics and phonology in honour of Eugenie J. A. Henderson (eds.) T. L. Thongkum et al. Bangkok: Chulalongkorn Univ. Press, 142-50. Radhakrishnan, R. 1981. The Nancowry word: phonology, ajjixal morphology and roots of a Nicobarese language (Current Enquiry into Lang. and Ling. 37). Carbondale Tll.: Linguistic Research Tnc. 26

PAGE 35

Shorto, H. L. Smith, K. D. Svantesson, J .-0. Yan Qi-xiang Yan Qi xiang Zhou, Z.-Z. & Yiin, Q. X. Palaungic vowels in Mon-Khmer perspective 1973. Three Mon-Khmer word families. Bull. Sch. Or. Afr. Stud. 36(2), 374-81. 1976. The vocalism of proto-Mon-Khmer. Tn Austroasiatic Stud. 2 (Oceanic Ling. Spee. Pub!. 13). Honolulu: Univ. Hawaii Press, 2. 1041-67. 1972. A phonological reconstruction of Proto-North-Bahnaric. Santa Ana: Summer Institute of Linguistics. 1981. et al. Mon-Khmer languages in Yunnan. Asiedu Sud-Est et Monde Jnsulindien 12(1-2), 91-100. 1983. Beng16ng-yu gai kuang (A brief description of the Benglong language) Minzu Yuwen (1983) 5, 67-80. 1981. et al. Pug lai cix ding yiie si ndong lai vax mai lai hox (A concise Wa-Chinese dictionary). Kunming: Yunnan Minzu Chubiinshe. 1983. Billiing-yu gai kuang (A brief description of Bulang language). Minzu Yuwen (1983) 2, 71-81. 1984. Wa-yu jian zhi (A concise description of the Wa language). Beijing: Minzu Chubiinshe.

PAGE 37

COMMUNICATIVES, EXISTIVES, AND STATIVES IN PROTO-SOUTH-BAHNARIC1 David Thomas 0. Introduction This paper is a study of some communicative, existive, and stative clause types in South Bahnaric languages, comparing them, and postulating some Proto-South-Bahnaric (PSB) forms. The data from the various languages, as will be apparent from the discussion, are uneven in both quantity and quality, so that the present paper must be considered preliminary. The South Bahnaric languages are the southern section of the Bahnaric branch of Mon-Khmer (Thomas & Headley 1970), located mostly in southern Vietnam, with some spilling over into Cambodia. I draw most heavily on Chrau, Eastern Mnong (Rlam), and Stieng, as representative languages of the group, with additional data from Koho Sre and Central Mnong (Bunar and Preh). The three main languages above are respectively at the south-eastern, north-eastern, and western edges of the South Bahnaric area, so should give a fairly good picture of the range of diversity. In the examples, words whose main significance seems to be as functors rather than as content words are underlined. Vocabulary items, mostly nouns and adjectives, whose meaning is not basic to the structure of the clause, are glossed beside their first occurrence. Functors and central verbs, i.e. elements that are basic to the clause structure, are listed and glossed below the set of examples. It would be desirable to list other verbs that take the same structure, but in most cases I am limited to the published data sources. In the reconstructions, an agreement of Chrau, Rlam, and Stieng is taken as sufficient evidence to reconstruct it for Proto-South-Bahnaric. An agreement of just Chrau and Stieng is also considered sufficient if there is no contrary evidence. 1. Communicatives The talking perceiving quoting informing group of clause types have a basic Speaker-V-Addressee-Information order in all the daughter languages. I. This is a companion article to "Some Proto-South-Bahnaric Clause Grammar", paper delivered to the 18th Sino-Tibetan Conference, Bangkok, 1985. (Mon-Khmer Stud. 15, 1989, 111-24). That article dealt with transitivity and locational clause types. The numbering of the examples here follows on the previous numbering. The clause presentation is based on my clause components outline (Thomas 1983: 137-42). It is semantically based, looking for and comparing the forms which manifest the desired meanings. 29

PAGE 38

DAVID THOMAS Mnong Bunflr ( = B) B3a: gop ngooi ngach 'I speak fast' (Phillips 1963 = M LC. 1.1) (ngach 'fast') B3k: gop ntay an naao BuNoong ma khon ay 'I will teach the Mnong language to you' (MLC.4.2)(naao BuNoong 'Mnong language', khon ay 'you f.pl.') ma 'to' ngooi 'speak' ntay an 'teach' Chrau (=C)2 C3a: an iiaai (yuur yuur) 'I speak (slowly') C3b: an iiaai jg_ neh 'I talk about him' (neh 'him') an fiaai sinlo iii heq 'I talk about this house' (iii heeq 'this house') C3c: C3d: C3e: C3f: an naai siq sinlo iii heeq 'id.' an fiaai t!_.!!:Jl._ neh 'I spoke to him' an chiih neh 'I scolded him' an fiaai paafi neh saaq 'I said he went/I said "He went"' an paafi neh saaq 'I said he went' an paafi, neh saaq 'I said, "He went"' 2. The Chrau data are my own (see Thomas 1971); the Koho Sre data are from Evans & Bowen n.d. (indicated as KLC) and Manley (OSS); the Mnong Bunar data are from Phillips Ms. (MLC); the Mnong Preh data are from Phillips & Kem (1974; CMLL); the Mnong Rlam data are from Tang (1976; M LLL), plus personal communications from Evangeline Blood (1985; unmarked), and the Stieng data are from Miller 1976 (OSG), plus personal communications from Ralph Haupers (1985; unmarked). I was not able to recheck any of the data with native speakers. Because of varied spelling conventions used in the different sources, I have standardised the writing of length as VV, the voiceless velar stop ask, and the final glottal stop as q. The 'whiskered' a, and Ii' are rendered here as ii and ii. The Koho o with the lowered dot is rendered 9. Four different spelling systems have been used for Koho in the past (Manley 1972:39), so I have converted the data from the different sources to the so-called SIL system, as it more closely matches the spelling of the other South Bahnaric languages. In some cases I have taken the liberty of replacing nouns and place names with other nouns and places names for reader ease. In the original sources shortness/length is marked as follows: Usual Markings Other Markings Environments short long only short only long Bunar V v,e,6 -i/y, u/o, ii/o -h, -? -0 Chrau V, ii, ii V, S, 0, 6 -i/y, u/o -0 Koho KLC vN, vT vN, vT -i/y, u/o -0 oss V V v=vq -0 Preh v, a v, e, u, o, a -i/y, u/o -h, -? -0 Rlam V, ii V -i/y, u/o -h, -? -0 Stieng V VY -0 A combined phonetic chart of the vowels would be: Front Central Back High ii u e 0 6 ie, iiO, u6 are e a 0 centralising offglides Low a 9 30

PAGE 39

Clause types in Proto-South-Balmaric C3g: an naai !zgy_ neh paan an saaq 'I told him that I was going' C3h: an chiih neh paan neh saaq 'I scolded him, saying that he went' C3i: an naai aan neh g!d!_ (paan) an saaq 'I spoke letting him know I was going' C3j: an paan neh saaq 'I invited him to go' an sier neh saaq 'I invited him to go' an aan neh saaq 'I allowed him to go' C3k: an padau neh g!d!_ troong Chrau 'I'll teach him the Chrau language' aan 'permit, allow' aan ... giit 'let.. .know' bay 'with, to' chiih 'speak, scold' giit 'know' naai 'speak' padau 'teach' paan 'say, saying, invite' sier 'invite siq 'concerning (lit. returning)' Koho Sre ( = K) K3a: an (99q) dos 'I (don't) speak' (Manley 1972 = OSS.217) an dos adaar adaar 'I speak slowly' (Evans & Bowen, n.d. KLC.2) (adaar 'slow' K3j: dos anjaaq me saao 'I invite you to eat' (KLC.63) (me saao 'you eat') 'speak' jaaq 'invite' Mnong Preh ( = P) P3a: gap ngooz ngach 'I speak fast' (Phillips & Kem 1974 = CMLL.18) (ngach 'fast') P3j: gap jaq may s88ng sa 'I invite you to eat' (CMLL.2 (s88ng sa 'eat') P31: gap nti aan an may g_J!_ nau BuNoong 'I will teach you Mnong' (CMLL.16) an ... git 'inform, let know' jaq 'invite' ngooi 'speak' nti aan 'teach' Mnong Rtam ( = R) R3a: an. ngooi broq broq 'I am speaking slowly' (broq broq 'slowly') R3b, c: an ngooi ta kan 'I spoke to/about him' (kan 'him') R3f: an !ah kan saak 'I said he went/ I said "He went"' (saak 'go') R3g: (laai) an !ah ta kan an saak 'I told him I was going' R3i: an ngooi aan kan giit an saak 'id.' R3j: an ndoom kan saak 'I invited him to go' 31

PAGE 40

DAVID THOMAS R3k: aan !ah laai ndoom ngooi ta an aan kan saak 'I allowed him to go' 'allow, give, let' 'say' 'past' Preferred in R3g. 'invite' 'speak' 'to, for, etc.' Stieng ( = S) S3a: S3b: S3c: S3d: S3e: S3f: S3g: S3i: S3j: S3k: S31: a aan baak chhu8r gut !ah maan mor sier tti hey mor (dreet dreet) 'I speak (slowly)' hey chhu8r baak bu 'I spoke about him' (bu 'him') hey !ah~ bu 'I spoke to him' hey !ah bu 'I scolded him' hey chhu8r !ah bu han 'I said that he went/I said "He went"' (han 'go') hey !ah bu han 'I said that he went' hey !ah a bu, !ah hey han 'I told him that I was going' hey !ah aan bu gut (!ah) hey han 'I spoke informing him that I was going' -hey maan/sier bu han 'I invited him to go' hey aan bu han 'I let him go' hey tti bu gut mor Sodieng 'I taught him to speak Stieng' 'to, for, from, etc.' 'allow, let' 'matter, concerning' 'relate, tell' 'know' 'say, scold, tell' 'command' 'speak' 'invite' 'teach' From the foregoing data one can immediately reconstruct an intransitive talking clause (3a) as Proto-South-Bahnaric *S-V-(Adv.), attested in all six daughter languages. The presence of an adverb or a negative with this construction seems to be preferred. For 'talking about' clauses (3b), all three attested languages have S-VLink-Content, but the type of Link varies from a preposition (Rlam) to a generic noun (Stieng, Chrau) or a motion verb (Chrau). The first Chrau form, with a verbal Link, may possibly be a calque on Vietnamese vi, and it is not attested in my data from the other languages. The se1.ond Chrau form, with a nominal Link, is matched by Stieng; it could of course, also be a calque on Vietnamese vifc, but it seems to be a more general South East Asian pattern, and Stieng has been under less Vietnamese influence than Chrau. The Rlam form is ambivalent for 3b and 3c. The wide variation in South Bahnaric forms here might seem to indicate some 32

PAGE 41

Clause types in Proto-South-Bahnaric instability or ambivalence at the proto stage, probably not matching any of the attested current forms. Addressee clauses (3c in C, R, S) all have the structure S-V-Prep.Addr., so that structure should be posited for PSB. For the preposition Stieng and Rlam use their broad-spectrum prepositions a and ta, but Chrau, lacking such a broad-spectrum preposition uses bay 'with'. Perhaps a broad-spectrum preposition (*ta?) should be posited for PSB. In Chrau, the communicative verbs have been split into at least four classes. VJ verbs, like iiaai 'speak', doom 'converse', are used in 3a, b, c, etc., and require a Preposition before an Addressee, and require a Quote Introducer before a Quote. V2 verbs, like chiih 'scold', !ah 'scold', payoom 'praise', are used in 3d, h, and require a Quote Introducer before a Quote. The V3 verb paaii 'say' cannot take an Addressee, and it takes no Quote Introducer since that would be homophonous with it. And paaii replaced PSB *!ah as the Quote Introducer. V4 verbs, like sier 'invite', paaii 'say, invite', are used in 3j. A transitive talking clause (3d in C, S) *S-V-0 should be reconstructed for verbs like 'scold' or 'praise'. There was apparently no distinction between direct and indirect quotatives (3e, f) in Proto-South-Bahnaric. Form 3e S-Vl-Quotint.Quot. is attested in Chrau and Stieng. The dropping of VI gives a simpler form (3f in C, R, S) in which the Quote Introducer (C. paaii, R, S. !ah) becomes the main verb *S-VQuot.-Quot. The verb !ah 'speak, scold' is attested in C, R, S and probably served as both a VQuot. and a Quotint. in PSB, but its Quotint. function dropped out in Chrau. All these languages have many verbs that can function as VJ, but only one that can function as Quotint. Simple quotative addressee clauses (3g) in Chrau and Stieng have the form S-Vl-Prep.-Addr.-Quotint.-Quot., in which VJ and Quotint. are the same as in the direct and indirect quotatives (3e, f), and the Prep. is that in 3c. Rlam has S-VJ-Addr.-Quot. The modern forms could be accounted for by positing a PSB *S-Vl:v/VQuot.: /ah-Prep.-Addr.Quotint.: lah-Quot. Stieng seems to have this structure. Mnong Rlam dropped the redundant second !ah. Chrau dropped !ah completely out of this construction by substituting paaii in the Quotint. slot and by putting !ah into the V2 class, which does not occur in the 3g construction. The 3h form, used with Chrau class V2 verbs, does not have a Preposition before the Addressee. Data from other languages are lacking, so no conclusions can be drawn for PSB. The longer quotative addressee form (3i) has the form *S-VJ-aanAddr.-git-Quot., as attested in C, S, R, with aan ... git 'let know' functioning as a benefactive marker. In C there is an optional Quotint. before the quotation; this may be a Chrau idiosyncrasy on the analogy of 3e. Imperative clauses, with verbs like 'invite, command, permit' (3j) have the same form as 3f, ie. S-V-Addr.-Quot., attested in C, S, R, so may be reconstructed for PSB in that form. In Chrau, the 3f and 3j verbs are 33

PAGE 42

DAVID THOMAS homophonous in one instance, paaii, yielding an ambiguous clause. The causative verb aan 'give, allow' can also be used here (C, R, S). Teaching clauses (3k) have the form S-V-aan-Addr.-git-Content ( = 3i) in C and P, perhaps reconstructable for PSB. Stieng lacks the *aan. Only Bunar has S-V-Content-ma-Addr. 2. Existives The simple existence identified existence transitive existence (existive possession) naming becoming group of clause types take a variety of forms, as may be seen from the following: Mnong Bunfir B4b: ngih ta ri 'There is a house there' (MLC.2.11) (ngih 'house', B4c: B4e: B4f: B4g: B4k: amoh geh jeeng ta Chrau C4a: C4b: C4c: C4d: C4e: C4f: C4g: C4h: C4j: C4k: C4m: C4n: C4o: geh heeq jeeng la !uh 88p saq ri 'there') ta ti jeeng ngih 'Over there are houses' (MLC.2.11) buum jeeng du ntiil ndo sa 'A tuber is a kind of food' (MLC.3.2) (buum 'tuber', du ntiil ndo sa 'one kind of food') gop geh du play ngih 'I have a house' (MLC.2.7) (du play 'one cl.') gop NDjreet 'I am Djret' (MLC.4.1) amoh gop NDaan 'My name is Dan' (MLC.4.1) 'name' 'have' 'be' 'at' iii 'There are houses' (iii 'house') iii 1J. heeq 'There are houses here' (heeq 'this, here') 1J. heeq iii 'id.' iii heeq 'This is a house' heeq !!!:_ iii 'id.' (rare < Vietnamese) aii geh iii 'I have a house' GaPe aii 'I am GaPe' aii heeq GaPe 'id.' aii GaPe 'I am named GaPe' aii GaPe 'My name is GaPe' aii tan 'hya saq neh GaPe 'I named him GaPe' (neh 'him' neh jeeng yaw 'He became a tiger' (yaw 'tiger') aii _p_ vadaai jeeng/ !uh iii 'I made the lean-to into a house' (vadaai 'lean-to') 'be, have' 'here, this one' 'become' 'is' ( < Vietnamese) 'appear, become' 'make, do' 'name, be named' 34

PAGE 43

Clause types in Proto-South-Bahnaric tan'hya saq 'to name' Koho Sre K4a: K4b: K4e: K4f: K4h: K41: K4n: hiu 'There are houses' (hiu 'house') hiu teeng dr; 'There are houses here' (KLC.9) (dr; 'here' chi dr; !!!:. hiu 'This is a house' (KLC.26) (chi dr; 'this') khay lah/ieeng caw mih 'He is an American' ( OSS.189) afi kroac 'I have oranges' (KLC.14a) (kroac 'orange') chi dr; kroac 'This is an orange' (KLC.14) sondan aii !!!:. K'Poh 'My name is K'Poh' (KLC.14) khay jeeng/gos kliu 'He became a tiger' (OSS.189-190) (khay 'he', kliw 'tiger') gas 'be, have, become' jeeng 'become' la, !ah 'is' ( < Vietnamese) sondan 'name' teeng 'at' Mnong Preh P4c: tiim bri ne 'In the forest there are rats' (CMLL.28) (bri P4e: P4f: P4k: geh 'forest', ne 'rat' gap jeeng BuNoong 'I am a Mnong' (CMLL.16) giip geh pe nuyh koon 'I have three children' (CMLL.8) (pe nuyh koon 'three children') moh sak giip NDoong 'My name is Dong' (CMLL.7) 'have' 1eeng 'be' moh sak 'name' tiim 'in, at' Mnong Rliim R4a: R4b: R4c: ?R4d: R4e: R4f: R4g: R4i: R4k: R4m: R4o: geh jeeng hih 'There are houses' (hih 'house') hih ta han 'There are houses there' (han 'there') to car Mriik mau hih 'In America there are houses' (MLLL.36) (car Mriik 'America') hih ho 'This is a house' (ho 'this') ho eh jeeng hih 'This is a house' (eh 'this') aii geh hih 'I have a house' Jhang aii 'I am Jhang' aii jeeng Jhang 'id.' nan aii Jhang 'My name is Jhang' afi nan to kan Jhang 'I named him Jhang' aii mhoq tum hun njeeng hih 'I made that shelter into a house' (tum hun 'shelter') 'have' 'be' 35

PAGE 44

DAVID THOMAS mau mhoq nan njeeng ta, to Stieng S4a: S4b: S4d: S4f: S4g: S4h: S4j: S4k: S4m: S4o: a biit chhak chu8l geh l8h !oh 'be' 'make, cause' 'name' 'become' 'at' g!!.f!._ iii 'There are houses' (iii 'house') g!!.f!._ iii~ ney 'There are houses there' (ney 'there') au iii 'This is a house' (au this, here') hey geh iii 'I have a house' hey Gape 'I am GaPe' hey au GaPe 'id.' hey chhak GaPe 'I am named GaPe' chhak hey GaPe 'My name is GaPe' hey chu8l bu chhak GaPe 'I named him GaPe' (bu 'him') hey !oh nom (l8h) biit iii 'I made the lean-to into a house' (nom 'lean-to') 'at' 'become' 'name' to name' 'be, have' 'appear' 'make, cause' A simple existence clause *ExistV-S (4a) may be posited for Proto South-Bahnaric on the evidence of C, K, R, S. And a PSB existence verb *geh is attested by B, C, (K?), R, S. A located existence clause ExistV-S-Prep.-Loc. (4b) is attested by B, C, K, R, S with a demonstrative Loe. The locative preposition, however, is different in all five languages, leading one to suspect that perhaps PSB had no preposition there; but the need for a preposition (in a preposing language group) became felt, perhaps to avoid ambiguity with a homophonous N-Dem. noun phrase. Bunar, Chrau and Preh data (4c) also include a transposed Prep.-Loc.-ExistV-S form, emphasising the Location rather than the Subject, which should probably be posited for PSB. Bunar has different ExistV in 4b and 4c. The simplest identification clause (4d) is Ident.-S, attested in C, R(?), or its reverse S-Ident. attested in S. It is not clear what should be reconstructed for PSB. A copula-linked identification clause (4e) S-Cop.-Ident. is found in B, K, P, R. It is absent from Stieng and only borrowed in Chrau, the two most reliably attested languages in the sample. (Calques from Vietnamese or English are very possible in the language lesson books, the sources for most of the other language data.) The copula jeeng used in B, K, P, R, however, is a verb of becoming in Chrau (4n) and Rlam (4o), so perhaps 36

PAGE 45

Clause types in Proto-South-Bahnaric PSB should be reconstructed without a copular identification clause form, then the northern tier of daughter languages expanded the use of Jeeng 'become' to provide a copula. A possession clause (4f) *S-Poss.-V-Item is clearly reconstructable from B, C, K, P, R, S, and the Possessive Verb in all of them is the existence verb *geh/gos of 4a, b. In PSB, apparently, existence and possession were parts of a single semantic category; perhaps possession should be viewed as transitive existence. (This use of have/be is paralleled by Vietnamese co, Thai mii, Khmer miian, and many other South-East Asian languages.) The personal identification clause (4g) is like the simple identification clause (4d), with Name3-S attested in C, R, and S-Name in B, S. Reconstruction is not clear. A topicalized form *S-Dem.-Name (4h) is more widely attested in C, K, S and should be reconstructed for PSB. Rlam has also a copular form (4i) S-Cop.-Name, not to be reconstructed for PSB (see 4e). A name clause (4j, k) is semantically very close to the personal identification clause. A form *S-NameN-Name (4j) with sak 'body, name' as the Name Noun can be reconstructed from C and S. And an alternate form *NameN-S-Name (4k) can be reconstructed from B, C, P, R, S. The nan 'name' in R is probably a loan from Rade. Koho also has a copular form (41) with the borrowed Vietnamese copula Iii The naming clause (4m) has different forms in C, R, and S. The history of these is not clear. A becoming clause (4n) *S-Jeeng-O is probably reconstructable from C, K (see 4e). There is no outside support for the Koho variant S-gos-O, which is homophonous with the possession clause (4f). The transforming clause ( 4o) has the form *S-V I-Form l-V2-Form2. C and R haveJeeng as V2, probably reconstructable for PSB. C can also have !uh as V2, but S has !uh as VI. PSB *!uh 'go out' must have been part of the PSB semantic field of transforming, but its syntactic use is not clear. 3. States The ambient stative comparative superlative evaluative group of clause types tends to be S-V in South Bahnaric languages, though V-S state clauses are quite common. Mnong Bunflr B5a: naar aao mih 'There will be rain today' (MLC.4.2) (naar aao B5b: B5c: B5d: geh Jeh 'today', mih 'rain') klaang book naar i1f!:. 'It is noon already' (MLC.2.2) (klaang book naar 'noon') naar aao Ji kat 'Today it is cold' (MLC.4.2) (ji kat 'cold') gap Ji ngoot 'I'm hungry' (MLC.2.2) (ji ngoot 'hungry') 'have' 'already, now' 3. Structurally 'name' is the head of the Subject noun phrase, but semantically 'I' is still the Subject, as in the two preceding forms. 37

PAGE 46

DAVID THOMAS Chrau C5a: ko mi 'It is raining' (ko 'sky', mi 'rain') C5b: C5c: nar tamv88p een 'It's noon now' (nar 'day', tamvoop 'middle') nar heeq takat'Today is cold' (heeq 'this', takat 'cold') C5d: ai'i takat 'I'm cold' C5e: C5f: C5g: takat ai'i 'I'm cold' (more common) ai'i takat ai'i 'id.' (emphasising 'I') ai'i !uh takat 'I became cold' C5h: C5i: C5j: d88ng een eeq ka ai'i takat doong neh 'I'm colder than him' neh eeq takat ka ai'i 'He's not as cold as me' ai'i takat doong leq 'I'm the coldest of all' 'more than' leq !uh Koho Sre 'already, now' 'not' 'like, as' 'all' 'become, appear' K5b: guul ngai rau 'It is noon already' (KLC.11) (guul ngai 'noon') K5c: ngai d9 noat 'Today is cold' (KLC.38) (noat 'cold') K5d: ai'i k99p 'I'm sick' (KLC.22) (k99p 'sick') K5g: khay go/oh koop 'He became sick' ( OSS.190) K5i: go 99q niam be chi d9 'They are not as good as this one' (KLC.33) (niam'good', go 'they') K5j: chi d9 buon rlaujoh 'This one is cheapest' (KLC.33) (buon 'sell') be 'as, like' go/oh 'become' 99q 'not' rau 'already, now' rlau joh 'most, superlatively' Mnong Preh P5a: bri mih 'it is raining' (CMLL.24) (bri 'jungle', mih 'rain') P5c: naar aao duh (ngan) 'Today is (very) hot' (CMLL.23) (nar aao 'today', duh 'hot') P5d: rpual prah joong 'A melon is long' (CMLL.29) (rpual prah 'melon', joong 'long' P5h: gfip kataang loon ma may 'I am stronger than you' (CMLL.23) (kataang 'strong') loon ma 'more than' Mnong Rtam R5a: mih 'It is raining' (mih 'rain') R5c: naar o kokat 'Today is cold' (naar o 'today', kokat 'cold') R5d: ai'i kokat 'I'm cold' 38

PAGE 47

R5e: R5g: R5h: R5i: R5j: ay so blah han hin ta jeeng leq mat nih so Stieng S5a: S5c: S5d: S5e: S5f: S5g: S5h: S5j: Clause types in Proto-South-Bahnaric kokat aii 'id.' aii jeeng kokat 'I became cold' an kokat hin ta kan 'I'm colder than him' kan han aysokokat blah afi 'He is not as cold as me' aii kokat hin ta leq mat nih 'I am the coldest of all' 'not' 'as, like' 'not' (?) 'more than' 'become' 'all, everyone' 'see, perceive' (?) mi 'It is raining' (mi 'rain') mi !oh 'id.' 'leek'It is cold' ('leek 'cold') nar 'leek 'id.' (nar 'day, sun') hey 'leek 'I am cold' 'leek, hey aq! 'Cold, indeed I am!' hey 'leek hey 'I am cold' hey 18h 'leek 'I became cold' hey 'leek hu8s bu 'am colder than him' hey teq a bu 'leek hey 'I am colder than him/Beside him I'm cold' let pal ney, 'leek hey 'I am the coldest of all/Of all of them I'm coldest' stative= S+P:Vi (OSG.9) aq 'exclamation' biit 'become' huos 'than, more than' leet pal ney 'completely, all of them' 18h 'appear, become' teq a 'place beside, compare with' There is no agreement on the form of the simple ambient clause (5a). Koho, Rlam, and Stieng have a simple Amb. structure. Chrau and Preh use a dummy subject DumS-Amb. structure. Stieng can use a dummy verb Amb.-DumV structure. And Bunar uses a geh-Amb. existive (4a) structure. This variety could perhaps be explained by positing a PSB simple Amb. as in K, R, S. Chrau (which has been in close geographical proximity to Vietnamese) and Preh adopted a Vietnamese-like structure, treating the Ambient as an intransitive verb. The Bunar form and the alternate Stieng form treat the Ambient as a noun. Positing a simple proto *Ambient would most easily account for the verb and noun developments. Nominal use of 39

PAGE 48

DAVID THOMAS the Ambient does appear also in Chrau in the form mi sa neh rain-eat-him 'he was heavily rained on'. A time clause *Time-Adv. form (5b) can be reconstructed on the evidence of B, C, K. A time-located ambient *Time-Amb. form (5c) is reconstructable from B, C, K, P, R. A stative *S-State (5d), which is similar in form to the intransitive (la), is reconstructable from B, C, K, P, R, S. A reversed stative form *State-S (5e), emphasising the state, is reconstructable from C, R, S. The Stieng form seems to be more emphatic than the Chrau, and is normally accompanied by an emphatic final particle. An echo form *S-State-S (5f) is reconstructable from C, S, giving mild emphasis to the Subject. An inceptive state *S-IncepV-State (5g) is reconstructable from C, K, R, S. The Inceptive Verb is !uh in C, S,jeeng in R, and golos (from gos ?) in K. These verbs have other PSB functions in 4a, e, n, o, and it is not clear which of these verbs should be reconstructed as the PSB Inceptive Verb. A comparative state *S l-State-CompMk.-S2 (5h) is reconstructable from C, P, R, S. But the Comparison Marker is different in each language, so no conclusions can be drawn concerning the PSB Comparison Marker. Stieng also has a form Sl-CompMk.-S2-State-Sl. For the negative comparison (5i) Chrau, Koho, and Rlam have SlNeg.-State-CompMk.-S2, which can probably be reconstructed for PSB. The superlative (5j) has the form S-State-CompMk.-All, paralleling 5h, in Chrau and Rlam. In Koho the form is S-State-Superl. In Stieng the form is All-State-S. Reconstruction is not clear. 4. Summary of reconstructions The reconstructed Proto-South-Bahnaric forms may be summarised, with sample glosses, as below. Binomial slot : filler formulations are given when both the functional slot and the actual filler were discussed. Communicatives 3a: *S-V-(Adv.) 'I speak (slowly)' 3b: *S-V-Link:?-Content 'I spoke about him' 3c: *S-V-Prep.:ta-Addr. 'I spoke to him' 3d: *S-V-0 'I scolded him' 3e: *S-Vl-Quotlnt.:/ah-Quot. 'I spoke saying he went' 3f: *S-VQuot.-Quot. 'I said he went' 3g: *S-Vl:v/VQuot.:/ah-Prep.-Addr.-Quotlnt.-Quot. 'I told him that I was going' 3h: ?? 3i, k?: *S-Vl-BenV:aan-Addr.-BenV:git-Quot. 'I let him know that I was going' 3j: *S-V:v4/aan-Addr.-Quot. 'I invited him to go' Existives 4a: *ExistV:geh-S 'There are houses' 40

PAGE 49

4b: 4c: 4d: 4e: 4f: 4g: 4h: 4i: 4j: 4k: 41: 4m: 4n: 4o: Statives Sa: Sb: Sc: Sd: Se: Sf: Sg: Sh: Si: Sk: Clause types in Proto-South-Bahnaric *ExistV:geh-S-(Prep.:?)-Loc. 'There are houses there' *Prep.-Loc.-ExistV-S 'Over there are houses' ?*Ident.-S / ?*S-Ident. (PNSB *S-Cop.jeeng-Ident. 'That is a house') *S-PossV:geh-Item 'I have a house' ?? *S-Dem.-Name 'I here am GaPe' *S-NameN:sak-Name 'I am named GaPe' *NameN:sak-S-Name 'My name is GaPe' ?? *S-Vjeeng-O 'He became a tiger' *S-Vl-Forml-V2-Form2 'He made the lean-to into a house' *Amb. 'It is raining' *Time-Adv. 'It is noon now' *Time-Amb. 'Yesterday it rained' *S-State 'I am cold' *State-S 'I am cold' *S-State-S 'I am cold' *S-IncepV:?=-state 'I became cold' S l-State-CompMk:?-S2 *Sl-Neg.-State-CompMk-S2 'I am not as cold as he' ?? REFERENCES Evans, Helen & Peggy Bowen n.d. ( c. 1965) Koho language course. Dalat: Christian & Missionary Alliance mimeo. Haupers, R. & Oieu 'Bi Manley, T. M. Miller, Vern G. Phillips, R. L. Phillips, R. L. & Y Kem Kpor Tang Hmok, Y. Thomas, D. D. Thomas, D. D. & Headley, R. K. n.d. (c. 1970). Stieng phrase book. Saigon: Summer Inst. Ling. 1972. Outline of Sre structure (Oceanic Ling. Spee. Pub!. 12). Honolulu: Univ. Hawaii Press. 1976. An overview of Stieng grammar. Grnnd Forks: Summer Inst. Ling. 1963. Mnong language course. (ms.) microfiche Summer Inst. Ling. 1974. Central Mnong language lessons. Saigon: Summer Inst. Ling. & Min. Educ. 1976. Mnong Lam language lessons. Summer Inst. Ling. 1971. Chrau grammar. (Oceanic Ling. Spee. Pub!. 7.). Honolulu: Univ. Hawaii Press. 1983. An invitation to grammar. Bangkok: Mahidol Univ. 1970. More on Mon-Khmer subgroupings. Lingua 25, 398-418. Thomas, Dorothy M. 1969. Chnm affixes. Mon-Khmer Stud. 3, 90-107. Saigon: Summer Inst. Ling. 41

PAGE 51

A LINGUISTIC ANALYSIS OF SOME SOUTH MUNDA KINSHIP TERMS, I Norman H. Zide & Arlene R. K. Zide Part 1 I. Jn this paper we present etymologies for a number of South Munda kinship terms.1 South Munda (SM), one branch of Proto-Munda (PM), branches into Kharia-Juang (KJ) or Central Munda, and Koraput Munda (KM); KM branches into Sora-Juray-Gorum (SJG) and GutobRemo-Gta? (GRG). While we have not done anthropological analyses of the Munda kinship systems, we have made use of the anthropological sources in defining and relating kin terms and kin-term sets of the languages and proto-languages. Our objective has been to provide a linguistic analysis of the SM kin terms, with a view to reconstructing as much of the original (i.e. non borrowed) terminology as possible, and to integrate the results into as coherent and plausible a system as we can. Since KM and SM noun morphology have not been analysed and described, we will present a description of SM noun morphology with particular reference to the derivational morphology of full forms (FF), and combining or compositional forms (CF) of Munda kin terms. This will come in the second paper in this series, along with the first full sets of kin-term etyma. In the first section of this paper we take up and criticise the work of Bhattacharya (1970) and Parkin (1985) on Munda kin terms, and in the second section we discuss how relationship and reciprocity are shown in Munda kin terms. We have taken our data, the SM kin terms and definitions, from a wide range of existing sources, and from our own fieldnotes. The anthro pological sources are fuller in the coverage of the terms and their uses, but are poorer in linguistic transcription, and lack morphological analysis. The linguistic sources provide better linguistic data, but are incomplete and, for purposes of kinship term analysis, poorly organised. A few publications (e.g. Deeney, 1975, on Ho) are exceptional in presenting the kinship system and the kin terms fully and perspicuously in linguistically well-analysed form. I. We use the following abbreviations in describing kin terms: M, 'mother'; F, 'father'; P, 'parent'; Br, 'brother'; Si, 'sister'; Sib, 'sibling'; Y, 'younger'; 0, 'older'; Hu, 'husband'; Wi 'wife' Sp 'spouse' So 'son' Da 'daughter' Ch 'child' The abbreviations rdr th~ mo
PAGE 52

NORMAN H. ZIDE & ARLENE R. K. ZIDE There has been widespread borrowing of kinship terms into SM: from languages as diverse as the Dravidian Ollari Gadba, and Indo-Aryan Kotia Oriya, standard Oriya, and standard and dialectal Hindi, Bhojpuri and Bengali, and English. Undoubtedly earlier borrowings, particularly from Dravidian, have been missed by us. We hope, for a later paper in this series, to collaborate with a Dravidianist on Dravidian borrowings, early and late, of both kinship behaviour and kinship words. This paper does not deal with kinship terms that are not genetically old in Munda, and that are not likely to have cognates in Mon-Khmer (MK). We propose to present the full set of Munda kin terms with Mon-Khmer (and, possibly, Austronesian) cognates in a later paper. The only published general treatments of Munda kin terms are those of Bhattacharya (1970) and Parkin (1985), and the relevant sections of Pinnow's Versuch (1959), as updated by him in 1960 in his unpublished monograph on Juang. We show that Bhattacharya's weighting of semantic similarity at the expense of sound correspondences leads him to lump together forms that are semantically very close, but not cognate.2 Parkin, an anthropologist who has written a dissertation on Austro asiatic kinship, needs to use linguistic evidence. He makes use of Bhattacharya's material, but is sometimes misled by Bhattacharya's methods of analysis and presentation of data. Parkin also takes rough phonetic similarity as indication of genetic relationship, and uses theusually spurious-'cognition' to support anthropological arguments about kinship. He sometimes ignores Bhattacharya's conclusions (e.g. about bare (Bhattacharya's Set 66. (1970: 455), Ga bare; ZZ *b::iHn:,3 'a woman's brother') and mistakenly connects reflexes of this *b::iHn; (e.g. bo?re, bok're) with *b::ib, YSib. By doing this he misses the importance of PM *b::iHn: and the existence in PM of terms restricted to male or female speakers, e.g. GRG *b::i(b)n:; 'a woman's brother', *bVlon, 'a woman's sister', *bVyal), 'a man's brother', and *tonan, 'a man's sister'. We need the anthropologist to make kinship sense of the meanings of cognate sets where we cannot reconstruct a properly precise meaning for the PM term, much less account for the change in meaning in NM and SM and the modern languages. The example of Northern Munda (NM) *hili, OBrWi, and KM *h Vii Sp YBr will be discussed in some detail later in the paper. Bhattacharya has confused the issues by putting NM *hili and KM *hVli into different semantic-cognate sets because their meanings are not close enough. We should note that anthropological studies of Munda kinship have flourished in the past fifteen years; we are thinking of the work of S. Bouez, Deeney, Parkin, Pfeffer and Vitebsky. However, most of these 2. He implies that they are sufficiently cognate for his purposes. 3. ZZ = Zide & Zide; our reconstructions, e.g. PM *b::,Hn:, or, better, *b~H(b)n:, differ in general reliability from proto to proto. SJG reconstructions are more solid than KJ reconstructions. In the *b::,Hr& example, the first reconstruction can be considered to be an abbreviation for the second, which shows more questions and possible answers. Since the reconstructions given here will not be evaluated and justified, they should be considered abbreviations of a fuller treatment of the comparative phonology and lexicon. 44

PAGE 53

South Munda kinship terms, I analyses do not bear directly on the linguistic problems with which we are concerned, and no further reference will be made to most of these. Bhattacharya's survey paper is very useful in that it presents and begins to organise his own rich field data. He is less thorough in abstracting the published literature. His analyses, his semantic-cognate sets, observations on borrowings, etc., are usually perceptive. In some cases he is more conservative than he needed to be, e.g. in putting Ho haam4 'old man, husband' in one set, and the reflexes ofGRG *hV-n-c;lam (e.g. Ga hamja, etc.) in another. Had he noted Mundari haram (same meaning as the Ho and the GRG), he would have been compelled to put all these forms in one semantic-cognate set. In other etyma, particularly where he brings in possible cognates in Mon-Khmer, he is too indiscriminately accepting. (But then, he is the only Indian Munda scholar who has paid any attention to Mon-Khmer.) Bhattacharya knows several of the Munda languages, and he knows Indo-Aryan and some Dravidian. Parkin, when he disregards Bhattacharya, usually goes wrong. Parkin is conscientious in examining long lists of kinship terms, and he turns up interesting forms not earlier appreciated (e.g., Remo N-kwi, YSi, which Bhattacharya also records). He can be perceptive about borrowings, e.g. noting that KM mama, MBr (in some languages SpF), is borrowed (along with kin behaviour) from Dravidian, and not directly from Indo Aryan. But his use of rough phonetic similarity as evidence of cognation is self-defeating. We do not quarrel with Parkin's anthropology. It is true, as Parkin says (1985: 705) that the absence of studies of comparative Munda kinship constitutes 'a major gap in south Asian studies', and that Parkin has done a considerable amount of serious work in filling that gap. We give three examples of the sorts of things we object to: (1) his Table of 'Basic forms of typical NM and SM terminologies'; (2) boko and bare; and (3) e1Jga, 'mother'. After this we take up the confused (mostly by Bhattacharya) case of Juang il:zi-bo, HuYBr. In his Table, Parkin gives 'typical' NM and SM terminologies, and includes the terms for SpF, SpM, ChCh, same sex SibCh ( of the same sex as Ego), and FSibCh of same sex as Ego, opposite sex SibCh ( of opposite sex to Ego) and of PSibCh, BrWi, HuSib, SiHu and WiSib, etc. What is a 'typical' terminology, and how and why does one compare typical terminologies? 'Typical' would seem to mean 'well-described', or 'well known'. The NM and SM forms given in the table are of little value. To a linguist, the forms to be compared would be the reconstructed NM and SM forms. These turn out to be closer to each other than Parkin's typical forms from typical modern languages-when they are cognate. NM 4. We have-without marking these forms-retranscribed some of Bhattacharya's transcriptions in order to indicate morpheme boundaries, and to prune dubious and excessive phonetic detail. We will give Bhattacharya's transcriptions and comment on them in the full data sets in later papers in this series. The form transcribed by us /zaam was transcribed ham by Bhattacharya, and hiim by Deeney. 45

PAGE 54

NORMAN H. ZIDE & ARLENE R. K. ZIDE *kankar and NM *kinkar, SpM, are fairly similar whereas Parkin's Mu hanar and (KM ?) *kinar are less so-perhaps because Korku, Kharia and Juang, which reserve PM *k, are not typical enough. Parkin's jai 'grandchild' (in Mundari-Ho) was selected as the representative NM form because he wanted to connect it with jia, 'grandmother'. The connection is dubious. A better word for NM 'grandchild' thanjai, found only in the Kherwarian branch of NM, would be kVrar/kVrar, found in both branches (Korku kurar, Santali korar). The data can be found in Bhattacharya. 2. Both in the table and elsewhere Parkin notes the important distinction between sibling terms where one must know whether the speaker is of the same or opposite sex to the kin-term referent. What seems important-to generalise the case-is not same or opposite sex, but whether the set of terms is sensitive to speaker's and referent's sex. That is, we have in KM a set of four terms where this feature is marked ('a woman's brother', 'a woman's sister', 'a man's brother', 'a man's sister'). This set contrasts with another set of terms where this feature is absent, but where we mark relative age of the speaker and referent, i.e. OBr, YBr, OSi, and YSi. Parkin's boko belongs to this second set (and perhaps should be defined not YBr, but YSib) and the others of the sibling terms in his table tonan, bokre (from *b:iHn:, Bhattacharya's bare set) and misi belong in the first set. It is possible (as the NM data suggest) that there were only two sex sensitive terms in PM in this first set, and these were both opposite marking terms, but this needs to be demonstrated. Parkin takes bokre as cognate with boko, and not with bare, *b:iHn; which leads him to miss the one cognate in PM of the sex-sensitive set. Since such terms go back to PM (however many may have to be reconstructed), we want to know what the distinctive functions of these two sets of sibling terms were-in earlier times and protos-and what they are now. None of the anthropologists we have read takes up this important problem. If one wants to use linguistic evidence in kinship arguments, then it is necessary to be able to recognise the historical depth of an etymology. *b:iHn: can be reconstructed for PM; this is noteworthy. It is possible that *b:iHn: at some more distant level-perhaps PAA-can be shown to have a morpheme in common with *b:ib, but good MK evidence of this would be needed to make such a case. One has to be able to recognise that *V?(g)-lEii is a good KM etymology for 'grandchild', but not a good SM or PM etymology. 3. Lastly, Parkin's treatment of 'mother'. Parkin writes that 'the standard NM and KM term for "mother" is eljga ... '. e1Jga is only standard in one branch of NM, Kherwarian. If one wants an etymological formula for PM 'mother', it would be V-ya-N, V-y:J-N. Korku has ayom in one dialect, the more archaic ayalJ in others; the reduced form of this before -te? is an-; Santali has e1Jga and ayo/ay:J; Mundari-Ho has e1Jga, but 46

PAGE 55

South Munda kinship terms, I e(y )alJ in the vocative (we do not find much morphological alternation of this sort in Munda); Juang has bwi-N, Kharia has ayo/ayo-N, Sora has yalJ, Juray ayalJ and yolJ, Gorum has yalJ, Remo yolJ, and iyolJ, and Gta? yalJ. The basic form is the V-ya-N/V-y:J-N. The question is how to relate e1Jga and e(y)alJ (vocative) in Ho-Mundari. Deeney has anticipated us in pointing out (Bhattacharya and Pinnow have made partly similar observations) that the vocative e(y )alJ resembles e1Jga-ii/eya-ii, i.e. the non-vocative stem with the first person singular pronoun -ii, i.e. 'my mother'. This suggests that most of the Munda forms for 'mother' were originally vocatives with first person pronominal enclitic (of possession) -Pi/-lJ, and that e1Jga represents (how accurately?) the basic form of 'mother'. How we get from e1Jga to eya/eiia where these forms precede the enclitic -ii-lJ has just been shown in Deeney's data. One could suggest metathesis, common in Munda, but we have no (other) cases of eya1J/ey1Ja, y/n, metathesis. We reconstruct V-ya which becomes V-iia before final nasal in the vocative. To take e1Jga, ayo, ealJ, etc. as obviously cognate is risky (it was not wrong). In any case, 'e1Jga is not the standard form of "mother" in NM and KM'. The point of these criticisms is not that anthropologists should reconstruct linguistic proto-forms, but that they should be less free in identifying putative cognates, and using theselargely spurious-cognates to support other arguments. i,:ii-bo or i,:ii-bau/-b:Ju. Our first comment on Parkin's rejection of Bhattacharya's connecting i,:ii-bo and hVli was that Bhattacharya's judgment on cognation was better than Parkin's, and whether the semantics agreed with Parkin or not, i,:ii and hV/i were almost certainly cognate. A re-examination of Bhattacharya's sets 81. and 83. (1970: 457) shows the situation to be more complicated. Bhattacharya misleads his readers by setting up two semantic-cognate sets, and putting NM *hili in the second (with Juang kuli), and KM *h Vii in the first. (Bhattacharya does not think with or use reconstructed form; we have constructed 'Bhattacharya's reconstructions' for him.) The facts are these: Bhattacharya has made two semantic-cognate sets, 81. and 83., these coming in his section of 'Terms for Brother, Sister, Brother-in-Law, Sister-in-law', one of his more complicated-and important-sets of terms. In 83. he puts Juang kuli/bli, OBrWi, and Kherwarian (there is no Korku cognate), *hili, OBrWi. In 81. there are three sets of words:5 (a) KM *hVli(-boj) (ZZ reconstruction), Vli-b:Ji, HuYSi, SpYSi; (b) KM *Vrel/*Vrer, HuYBr, SpYBr, and (c) NM *erwel/ *Vrwel, HuYBr. Bhattacharya does not sort these into sets, so that it is not clear if he thinks Juang i,:ii-bo goes in set (a), or with sets (b) and (c). The latter two sets clearly reconstruct to PM *Vrwel.6 The meanings of the 83. forms are fairly close: (a) HuYSi, SPYSi, and (b) -(c) HuYBr, SpYBr. i,:ii-bo means HuYBr. Since 83. (a) is clearly cognate with 81. which includes Juang kuli, to relate i,:ii to the augmented 83. we either have to disqualify and reject kuli, somehow to relate kuli and i,:ii, or to connect i,:ii with 81. Can we-as Bhattacharya's array of data suggests-5. See the Appendix for fuller presentation of the data. 6. The ? in Gta? wrwe? requires some discussion. 47

PAGE 56

NORMAN H. ZIDE & ARLENE R. K. ZIDE show i[li to be a cognate of PM *Vrwel, HuBYr, SpYBr? We need to show that the sound correspondences are possible: that Juang fl corresponds with r ( or something like r): this is possible; that the first vowel i corresponds to V, where the following, stressed, vowel is *e and i in Juang: this is possible; the correspondence of Juang stressed i and PM *e: this is possible; and that a final consonant, Juang !, can be lost compound medially: this also fits into the set of expectable, regular, correspondences. We accept, tentatively, i[li as cognate with PM *Vrwel, and rearrange Bhattacharya's sets as follows: 81. (b), (c) plus i[li, HuYBr, SpYBr, and 83. plus 81 (a), *kuli(-b.Ji), OBrWi, HuYSi, SpYSi. Part II This second section takes up: (1) (once more) the two different sets of sibling terms in a number of the Munda languages; (2) the old system of pronominal enclitics (ProiP) marking inalienable possession. The kin terms of Kherwarian that do not take Pro'P are: (i) name-like kin terms, and (ii) conjoined pairs of terms whose referents are related to each other, not to the speaker or some other named or pronominally indicated person. The latter, paired reciprocal terms, are common in Munda, both north and south. Reciprocal infixes in kin terms (the infix is the same one found as the verbal reciprocal marker) are commonly used where each of the referents of a hypothetical pair refers to the other by the same kin term. The various meanings of this reciprocal infix, NM -p-, SM -m-, are examined in several Munda languages. (3) reciprocal (kinship) terms of address are examined in Santali, Juang and Sora. (4) Some kin-term affixes that look like but are not reciprocal -p-/-m-are described, and traced to PAA. We noted earlier that in some Munda languages-Remo is the best example of one with two complete sets-there are two different sets of sibling terms, one marking relative age (e.g. 'younger brother', 'older sister'), and the other marking 'relative sex' of speaker and referent (i.e. 'a man's brother' versus 'a woman's brother'). Relative age is commonly marked in the Indian area; relative sex is not. So far as we know, none of the anthropologists who have worked on Munda groups with two (complete or incomplete) sets of sibling terms has described the distinctive functions of these sets. Parkin does note these terms, but in his table he distinguishes between same-sex (of speaker and referent) and opposite-sex terms, whereas we see the basic difference as between sex-marked (same or different sex) terms, and age-marked terms. The GRG languages have two full (four-term) sets, but it is not clear that cognate forms of the sex marked set in Sora (e.g. GRG *bVyal), So bofiafJ, GRG *tonan, So tonan, these meaning 'a man's brother' and 'a man's sister' respectively in GRG) 48

PAGE 57

South Munda kinship terms, I have the same meanings and sort into similar sets in Sora. In Kherwarian the sex-marked set has only two terms, both marking opposite sex kin. The history of these sex-marked terms is still largely obscure. But in any case the distinction is noteworthy, and an understanding of the functions of the two sets should be important to the anthropologist studying Munda kinship. When studying the Munda languages that preserve the old system of pronominal enclitics for marking inalienable possession (e.g. Ho, Santali, Juang, Kharia), we find different definitions of grammatical inalien ability. The minimal set of inalienably possessible nouns would be the kin terms, the next minimal set would add words for body parts. The pronominal enclitics, ProiP, are required with almost all inalienably possessible nouns, with a few exceptions. The common exceptions are: ( a) 'name-like' terms; and (b) paired reciprocal terms where the relationship obtains between the referents of the pair. These paired reciprocal kin terms are common in Munda. One Mundari example from Hoffmann (1950: 1303) of a name-like term has to do with the speech of siblings (i.e. those who have a common referent for 'mother' and 'father'). In this context-siblings speaking to each other-one cannot say 'my mother' without the (erroneous and offensive) implication that she is not also 'your mother'. What is saidand is preferable to using a Proip for first person dual inclusive-is simply 'mother' (perhaps .'Mother' would better represent this), i.e. the form eJJga without a Pro'P. Paired reciprocals of one form or another are characteristic of several of the Munda languages, North and South. The kinds of pairs found are: (i) the senior term occurs followed by the 'collective' suffix -ya/-ea; in Santali, e.g. hili-ya (hili, OBrWi); we discuss the meaning of such forms below; (ii) the senior term of the reciprocal relationship (if there is one) is mentioned followed by an echo word; in Santali, e.g. hili + h9li; we discuss the meaning of these forms below; (iii) both members of the pair are mentioned, the senior one first; in Gta?, e.g. nta?+ ;egfre 'grandfather and grandchild'; this means (we, they etc.) are grandfather and grandchild. Whether, as for similar forms of Santali (i and ii above), the compound can also mean the relationship (in this case the grandfather-grandchild relationship) is questionable. (iv) the construction with nu-plus (usually junior) kin term in Remo; e.g. nu-giriJJ 'I and my wife's younger brother', and probably also 'we two are OSiHu and WiYBr'; (v) where both members of the (hypothetical) pair would refer to each other by the same term, e.g. GRG *bVyal), 'a man's brother'; in the GRG languages, e.g. GRG *b-Vm-lon (*bVlon, 'a woman's sister') 49

PAGE 58

NORMAN H. ZIDE & ARLENE R. K. ZIDE 'each other's sisters', Gutob bumulon, Remo bunlu, Gta? bumluy.7 The -m-infix (which becomes -n-in Remo) marks this relationship. There are two seemingly irregular cases in Remo: t-un-una, and g Vn-riy (/ginfiriy/), where tuna is 'a man's sister' (younger sister according to Bhattacharya) and giriy is WiYBr. These have the same meanings as the forms nu-tuna and nu-giriy, i.e. 'I and my younger sister'; 'we are (older) brother and (younger) sister' and the giriy form with -n-would be glossed the way the nu-form was. These are reciprocal pairs, but unless Remo has a common term of reference (or of address?) for brother and sister, and for OSiHu and WiYBr, these are not terms for referents who refer to each other by a common kin term, and we conclude that Remo -n-has been extended to a few such pairs. The first definition (in the third volume of Bodding's Santai dictionary (1935: 116)) of hili + h9li is 'a man's wife and younger brothers', then (that) 'relationship', and then, more literally, 'a hili and those who call her so', i.e. her erwel(s), HuYBr. That is, hili+h9li means 'the hili-erwel relationship'; hili-and-erwel: the erwel(s)' hili, and the hi/i's erwel(s). Some echo words can be defined as 'what goes along with (the preceding N1 which the echo word echoes)' or N1 'and the like'. What goes with hili, OBrWi, here is its reciprocal. We find echo words of kin terms used comparatively infrequently in this way, but the synonymous construction with -ya/-ea (Bodding's 'collective') is more common. (hili-ya, according to Bodding, means exactly what hili + h9li means.) Bodding in his Materials (1929: II 41-2) lists more than twenty such kin-term pairs (or collectives), most of them in common use. So far as we know, such reciprocals are not common elsewhere in the Indian area. There are in Indo-Aryan and Dravidian pairs like Hindi m'ii'ii + baap 'mother and father' (in more formal speech maataa + pitaa), and bhaaii + bahan, 'brother and sister', but not twenty other pairs of kin term reciprocals in common use. The other examples of SM -m-are found in Juang, and of NM *-pin North Munda fairly widely (we discuss forms in Ho, Santali and Korku). Pinnow gives two Juang examples of -m-: st:mt:lay (from st:lay, 'grown-up girl'), 'young woman, woman', and bm;yygu (from bygu, 'young man, youth'), 'master, lord, husband'. In the first case, the affixed form is slightly more general; in the second the infixed form is a kind of honorific. The only feature of the reciprocal evident in the second example is its use of plurality to mark the honorific, something quite common in the Indian area. Ho-the third ofDeeney's small 1975 dictionary that we sampled-has few examples of -p-in kin terms. It is interesting that the two we found have precisely the meanings of the Juang forms but are not cognate with 7 A common Indian areal construction repeats the noun (kin term) for this reciprocal'distributive'-meaning. Thus, Hindi ham bhaaii bhaaii huf. (bhaaii, 'brother') 'We are (like) brothers', 'buddy buddy'. Similar constructions have been noted in some of the Munda languages. 50

PAGE 59

South Munda kinship terms, I them: h-Vp-anum 'young woman' (there is no hanum), and s-Vp-eed (Mundari sepered) 'young man'. There is no free seed, but it occurs bound in the 'nephew' term hon-seed (NM *bn-sered, *bn-serej, FBrSo). The Santali examples-we used as a sample the letter H in the third volume of Bodding's dictionary (1935: 1-184)-are hapamfin, hapamfi 'grown-up girl' (there is no hamfin or hamu); hapram-ko/haprum-ko 'ancestors, forefathers, collection of old men' (haram, harum 'an old man'); hcpd, along with herd, 'man, male' (no d); and h:Jp:m along with h:Jn. He translates h:Jp:Jn 'offspring, young, child, son daughter'; adj. 'small, little', and h:Jn 'a son, child', now used only in compounds. For Korku, N. Zide recorded along with kon, kon-jei, kurar, kosered/ koserej and some others, kopon (with the dual or plural), koponjei, kuparar, kopsered/kopserej. The simple forms were much more common. It never became clear what the meanings of the dual and plural -p-infixed forms were. The meaning of -p-in NM is still obscure; the plural aspect of the 'reciprocal' (and its development into an honorific), and the notion of a set or collection can be seen in some of the infixed forms. The forms with -m-/-p-for 'young girl' and 'young man' in Juang, Santali and Ho have been noted, but what the function of the reciprocal is here is unknown. Of mutual appellatives (Bodding's term, i.e. reciprocal terms of address-in our examples, of persons not of the same generation), Parkin notes Santali g:Jpm, which Bodding translates 'namesake', used by grandfather and grandson as terms of address. Parkin also notes the existence of such forms in Juang. McDougal (1963: 141) writing of generation sets and the extension of kin terms, says that 'any male of an odd-numbered ascending or descending generation, regardless of kin type, may be addressed with the term for "father", and any female with the term for "mother" ... Any male of Ego's own or any even-numbered generation may be addressed with the term for "brother", any female with the term of "sister". For example, a man may address his brother, father's father, and son's son with the term for "brother" .. .'. In Vitebsky's as yet unpublished notes on Sora kinship there are also examples of cross generation kin addressing each other by the same term. The term tata, FEBr, is used in address reciprocally, i.e. by YBrSo. Similarly entalaij, MOSi, can be used reciprocally in address by YoSiDa. It is likely that more Munda languages have mutual appellatives, but that they have not been recorded. There are two more possible infixes found with kin terms: the -m-in NM k-Vm-:m 'nephew, niece', from the simplex bn, 'child', also attested in SM: Sora am:m-sij 'nephew', and am:Jn-sil, 'niece'. This is the only example of this PM *-m-. G. Diffloth tells us that there is a cognate affix, -m-, old but unproductive in Mon-Khmer, with cognate forms in a number of MK families (Khmer, Bahnaric) meaning 'child' for the simplex, and 'nephew/niece' for the infixed form. The other case of a kin term that might be analysed as containing an infix is PM *kin-or kin-, found only in *kinkar/kinkar 'mother-in-law' 51

PAGE 60

NORMAN H. ZIDE & ARLENE R. K. ZIDE (and perhaps Juang kan-
PAGE 61

*kuli South Munda kinship terms, I APPENDIX9 Bhattacharya 83. (abridged) Juang b/i, kuli, OBrWi; Kherwarian *hili, OBrWi We noted earlier that adding KM *h Vii makes the SM evidence more convincing. Pinnow in his Juang monograph also connected Juang kuli and Kher *hili, but took the etymology no further. We note the form kulaya-sini ( only in McDougal) which is, apparently, a derivative of kuli sini (elsewhere -st:l, -sen) is the combining form of (McD.) kon-chalan (Pi. k:m-sla1J) 'young woman'. kulayasini means HuYSi, and can be roughly glossed 'kuli-junior', i.e. the 'junior' ofOBrWi is HuYSi. Note the use of -sini in the 'grandchild' terms boko-lap (McD.), 'grandson', and boko-sini, 'granddaughter'. The grandchild terms probably derive from boko-X-lab/ sini, the X perhaps being the -cfu? found in Kharia (cognate with words for 'child' in Gorum (ar;luef), and for 'young man' in Gotub (o-rug)). SM *hVli(-boi), PM *Vrwel, SM *Vrel/*Vrer, NM *erwel Bhattacharya 81. (abridged, rearranged, and provided with additional data, the latter in parenthesis) Set (a). Ga iii-boy (MZ hli-bwe?), HuYSi; Re liboi, HuYSi; So aliboi, HuYSi (Vi. aliboi, HuYSi, WiYSi); ?? Ju ir_zi-bo (McD. ini-bou, MM ir_zi bo), HuYBr. -boi/-boy/-b:Jj mean 'woman' in these compounds; -sij/sij (full form in So pAsij, 'child', in compounds with terms meaning nephew/ niece -sij means 'male person'. This -sij is probably cognate with the -serej, -st:rt:j/-sered, -st:rt:d mentioned earlier for NM and meaning 'young man, nephew'. Bhattacharya seems to be suggesting that the -bo in ir_zi-bo is a combining form of boko, YBr. We have no other examples of this-or anything else-as a combining form of boko. Set (b).Ga uruve, HuYBr (MZ wrwe?); Re ere (ZZ ere(/)), HuYBr; Gu ere/, HuYBr; Go ilil, HuYBr; So arer-sij-an, HuYBr, erer-sij-an, WiBr (we take arer, and erer to be the same), erel-boi, WiYSi (Vi eri-sij, WiYBr, HuYBr; Sur erer-sij, WiYBr, HuYBr; ali-boy, arrel-boy, WiYSi, HuYSi). One complication in Sora that needs comment; we find in both Vitebsky's and Suryanarayana's data that along with ali-boi for Sp YSi, we also find eri-(Vi.) and arrel-(Sur.)-boi. The eri-and arrel-are both from KM *Vrel, although the eri looks like a portmanteau of ali and Vrel. Since the marking of alias feminine, and of Vrel as masculine in Sora is being lost (this sex-marking role has been taken over by -boi and -siJ) the meanings of Vii and Vrel have come closer. Vitebsky suggests that eri-9. The data in the Appendix are, if not otherwise marked, from Bhattacharya. Words in parenthesis not further labelled are from Zide and Zide. The abbreviations refer to: FF, Ferndandez; MM, Mahapatra and Matson; MZ, K. Mahapatra and N. Zide; McD., McDougal, Pi., Pinnow; Sur., Surayanarayana; Vi., Vitebsky; VU, Vidyarthi and Upadhyay; and ZZ, Zide and Zide. 53

PAGE 62

NORMAN H. ZIDE & ARLENE R. K. ZIDE means 'younger', i.e. 'younger SpSib', since it can now be used with either wife's younger sister (eri-boy along with ali-boy) or younger brother (eri sij), and similarly for a female ego for HuYBr and HuYSi. For Vitebsky's Soras this would work, but not for Bhattacharya's. For Suryanarayana Sp YBr is erer-sij; *Vrel has been generalised, but then split into arrel, which commutes with ali-, i.e. goes with -boy. The other form *Vrel has split into is erer-, and this is used only with the masculine -sij. We are grateful to R. J. Parkin for copies of his papers, and to Piers Vitebsky for his notes on Sora, and for copies of the extract from Suryanarayana's dissertation. Set (c).Sa erwel, HuYBr (Bodding era+ el ?); Mu iril, irul, iriul, HuYBr; Korwa irvil, HuYBr; Koku, ilur, HuYBr. (WiYBr in NM is *teya (i.e. [teyaJ). For set ( a)-putting aside i,:zi-bo for the moment-we reconstruct *hVli(-b:Jj), HuYSi, SpYSi; using Bhattacharya's data only it would be *Vli-boi. We find this to be cognate with Bhattacharya's 83., *kuli/ *kVli, OBrWi. The definition of the new PM *kuli would be 'female affine of ego's generation' (i.e. English 'sister-in-law'), OBrWi, HuYSi, SpYSi. An alternative interpretation would reject Juang kuli as cognate with either NM *hili or KM *h Vli, and include i,:zi and NM *hili and KM *h Vii in the reconstruction. Bhattacharya presents sets of forms that are semantically similar. He would like these semantic sets to be cognate sets as well. But he wants his semantic-cognate sets to be closer in meaning-for him to accept themthan, we think, such cognate sets in PM will often be. It is true that some of these kin terms' semantic-cognate sets (e.g. Bhattacharya's bare, ZZ b:JHrt:) show forms that have remained remarkably close in meaning and in phonetic form in the modern languages. But given the chronological separation of the Munda languages (at least twenty-five hundred years), there is no reason to expect this degree of closeness. Just as we have no hesitation in calling Ga swa and So Al)Al, both 'fire, firewood', cognatesbecause we can see how both developed regularly from KM *sVJJAHl-we should have no trouble in accepting NM *hili and KM *h Vii as cognate despite the excessive (to Bhattacharya) difference between the meanings of the two forms. We need the anthropologist to reconstruct a more precise meaning for the proto of these, i.e.-if we accept Juang ku/i-*kVli, and to account for the changes in meaning between the PM form, and the forms in the modern lanuguages. From the forms in set (b) we reconstruct KM *Vrel/*Vrer, HuYBr, SpYBr. From this and NM *erwel (reconstructed from the forms in set ( c) ), we reconstruct PM *Vrwel/*erwel, Sp YBr, Hu YBr. It is possible, as Bhattacharya suggests, that *Vrwel is bimorphemic, and that the second morpheme ( of *Vr( )-) is -h Vl(i) or -k Vl(i), but we see no evidence of this. 54

PAGE 63

South Munda kinship terms, I i,:zi-bo, i,:zi-bou/-bau What then of i,:zi-bo? Since bau/bou is the reciprocal in Sora of KM *hVli and KM *Vrel derivatives, and a similar situation may obtain in Juang, let us look at Bhattacharya's Set 62., bau/bou. PM *bao/*bau Bhattacharya 62. (edited, and with additional material): So bao-n, 'brother-in-law', (Vi. baon OSiHu, Sur. boung, OSiHu, kinar-boung SpOBr); (ZZ Go bao, 'brother-in-law'); ZZ Gu il-bmJ, m-ba!J, OSiHu); Remo (Bhattacharya, 1968, 111) m-ba!J, OSiHu; (ZZ Ga m-bia, OSiHu); Kh bau HOBr, Roy bao-tang, WiOBr (VU boutang, WiOBr); Ju bov-kar, HuOBr (Pi. bau/fou, OSiHu, MM bau, SiHu, McD. bou, ESiHu, bokar (probably from bau-kar), HuOBr); NM: Ho bavo, OBr, SpOSiHu; bau honjar, SpOBr (Deeney bau, OBr or cousin; bau honyar, SpOBr); Mu bau, OBr (in address); bau honjar, SpOBr; Sa bahJn-har, SpOBr; Korku bao, WiOBr, (Girard bao, WOBr). The final nasals in some of these words are frozen first person prenominal enclitics used in terms of address, i.e. literally, 'my OSiHu', etc. To return to i,:zi-bo (or -bou), we stated in the body of the paper that with regard to sound correspondences and semantics, a case can be made for i,:zi being cognate with PM *Vrwel, SpYBr, HuYBr, if not the strongest case. What of the -bo or -bou? i,:zi-bo or -bou is an inalienably possessed noun in Juang. It is true that reciprocals-both terms and the relationship between then-are often maintained through many linguistic changes. But that a pair of reciprocals meaning (an inalienably possessible) HYoBr, this deriving from the usual paired reciprocal meanings, could have developed in Juang is, for semantic reasons, highly unlikely. -bo, although we have seen no (other) examples of it as a combining form of boko seems more likely. Why does i,:zi need a second morpheme at all to have the meaning HYBr? As far as we can see, it does not. The spouses of both kulayasini and i,:zibo are both some kind of boko: the i,:zibo's wife is (McD.) boko-ray (YSi), and kulayasini's husband is boko (YBr). HYBr in McDougal is boko-ger (from boko-ker?); elsewhere he is just boko. Apparently i,:zi too required a second morpheme. -bo from boko makes better sense than bau/ bJu. -kar, 'in-law' On bou-kar and the use of the affine-marking -kar in Kharia-Juang. The affinal -kar, 'in-law', is found in PM in *kikar/*kankar, SpM (or perhaps, SpM, WiOSi), and *ku(X)nkar, SpF (or SpF, HuOBr). The -kar may be related to what has become the NM word for 'man, person, member of the community', kJrJ. The Kharia -kar marks the agent in forms like rema-kar, 'call-er', rema-'to call' (Malhotra, 1982: 311). The kar is also used in forming singular pronominal stems from demonstrative 55

PAGE 64

NORMAN H. ZIDE & ARLENE R. K. ZIDE bases (e.g., ho-, u-), thus ho-kar, u-kar, 'he, she'; i.e. -kar means 'one'. In KJ we find the following forms: Juang keeps the term for 'father-in-law', kuiikar/kwiiikar/kwi1Jkar, and also has bao-kar, HuOBr, and aji-kar, SpOSi. Kharia keeps the old word for 'mother-in-law', kinkar, and has, in addition, baker (probably from boko-ker), SPYBr, and (Roy) aichkar, WiOSi. (We noted earlier that Juang has boko-ger-with the ger perhaps from -ker and not from koN-ger, 'young man'.) In some but not all cases the -kar can be translated simply as 'in-law', i.e. if X means YSi, then X-kar means YSi-in-law, i.e. SpYSi. This presupposes that X is not an affinal term. The examples of this in KJ are aji, ESi, and boko, YBr. One could, simplistically, try to derive the remaining terms from hypothetical simple terms, not in KJ, but in Kherwarian. Thus Ho bau, OBr, Juang bau-kar, HuOBr. This would miss the fact that Juang bau/b:Ju is an affinal term, OSiHu, and that kar derivatives of affinal terms are different from derivatives of simplex terms. Where the kar-derivative is formed on an affinal term, the simplex refers to a sibling's spouse, and the kar-derivative to a spouse's sibling. The other example of this in KJ is (Roy) aji, OBrWi, and aich-kar, WiOSi. The Juang boko-ger, HuYBr, if regular, should come from bok(V)-(k)er-ker. The term bok-sel in Kharia, SpYSi (baker is SPYBr) is probably to be derived from bok-ker-sel. And the Juang 'grandchild' terms, boko-fab, 'grandson', and boko-sini, 'granddaughter' are probably to be derived from boko-X-fab/-sini, the X perhaps being cognate with the (iu? found in Kharia in bok-(iu?, 'grandchild'. The reduction of similar compounds in Sora from morphemes 1-2-3 to 1-3 was observed by N. Zide in regard to the Sora numerals. The Kherwarian uses of *-kar (which, if it were reduced the way it is in KJ would be *-har): PM *kinkar/*kankar (Kher *hanhar) and *ku(X)nkar (Kher *h::inhar): these occur in compounds of the structure X-hanhar/h:Jiiyar, which have the meaning hanhar's or h:Jiiyar's X, i.e. kaka-honyar in Ho means honyar's kaka, i.e. SpFYBr. For X as aji and bau this is not the case. The construction with aji-hanar, and bau-honyar in Ho, is like the construction in KJ with non-affinal simplex plus -kar. Note that (unlike KJ) aji in Ho is OSi, and bau is OBr; aji-hanar means SpOSi, and bau-honyar means SpOBR. Note that these are 'spouse's sibling' terms, which, in KJ, are those derived from affinal simplex terms, not, as here, non-affinal simplex terms. Santali is much like Ho in its SpF (h:Jiihar) and SpM (hanhar) terms: most of them have the X-hanharjh:Jiihar construction where this means, e.g. hanhar's X (where, for instance, X would be go1Jgo, OBr). For the two forms cognate with the two just distinguished in Ho, Santali bah:Jiihar and ajhnar, the meanings are parallel to those in Ho: SpOBr, and SpOSi. Note that bah:Jii does not occur as a free form (as does the Ho cognate bau, OBr), but aji, OSi, does. Note the reduction of h:Jiihar to -har in the first form, and that of hanhar to -hnar in the second. 56

PAGE 65

South Munda kinship terms, I *boko (NM *b::ib) It is not surprising that Parkin (with Bhattacharya's help) sees forms related to *boko almost everywhere. He is right to connect Ga tabo (MZ ta-bo?) and Re tabuk', FYBr, MYSi, stepF (Bh. Set 31.) to boko. Cognate forms in Set 33. should also be included: Bh. mbu, (FF mbu?, Elwin umbuk-boi), MYSi, FYBrWi, stepM. Bhattacharya is probably wrong in bringing in bobo (he translates the Gta? form, wrongly, we think, as YBr). In KJ *boko and possibly related forms cover a good deal of semantic territory: apart from the basic-YSib--meanings there is the Juang word for 'clan' (Pi. bog, McD. bok), the 'grandchild' terms we have talked about (Kh (Roy) bok-(lu(?), Juang (McD.) bokolap and bokosini). Pinnow records the Juang term 'fob-ra(i, "Verwandschaft" (relation), Bedeutung und Form unklar'. The form found in Kherwarian, fob, is unexpected, since one would expect the intervocalic k to go to h before :J. Bhattacharya gives a form, Ho boho, which is not found in Deeney (but see the other forms in Bhattacharya's Set 64. (1970: 455)). There do seem to be related forms in Kherwarian with h, e.g., in Bodding's Santali Materials II (1929: 21) there are fob boeha and foh;:,k foh;:,k boeha, 'cousins any number of times removed, descendants in the male line'. One wants to know not only the nouns, the kin terms, but the rest of the kinship vocabulary-the verbs and the constructions. This sort of information can be found in Bodding and Vitebsky. As to the rest of the terms with initial b that might be etymologically connected with boko, a better knowledge of MK cognates would help. For instance, we would tentatively connect the first piece (etymologically, presumably, a morpheme) of bVya1J, bofia1J, 'brother, a man's brother' and GRG *bVlon 'sister, a woman's sister', Sora (Vi.) bui'-mai, (buj-mai? bVfi-mai ?), 'true sister, usually used by women'. Bhattacharya wants to connect bVya1J with Sora, YBr, ubba-n-u(b)ba1](Vi. u'ba-). We think it is probably cognate with boko. To make a case for either-or, conceivably, both-considerably more synchronic phonological and morphological analysis needs to be done. For instance, by Bhattacharya's hypothesis, one would take -ba1J as a combining form of bVya1J, odd but not impossible, but it does not fit in with assumption of a morpheme b Vfi-/ bVI-shared by *bVlon and KM *bVyaN. And we have other forms: Gorum bi1]-ger, 'bloodbrothers', Sora birin(ia, 'clan', i.e. (Vi.) 'exogamous agnatic localised virilocal kin group' (Vitebsky translates b Vya1J as 'male member of own birinda'), and Sora Fa YBr kim-bom, Vi. kin-born, kin-bo1J that come into the picture. The sibling terms and related vocabulary will be taken up in a later paper, which will take MK cognates into consideration, and do more synchronic phonological and morphological evidence than we have done here. 57

PAGE 66

NORMAN H. ZIDE & ARLENE R. K. ZIDE REFERENCES Bhattacharya, S. Biligiri, H. S. Bodding, P. 0. Bouez, S. Deeney, J. Diffloth, G. Elwin, V. Fernandez, F Girard, B. Guha, U., Siddiqui, M.K.A.& Mathur, P. R. G. Hoffmann, J. & van Emelen, A. Karve, I. Macphail, R. M. Mahapatra, B. P. & Matson, D. M. Mahapatra, K. & Zide, N. Malhotra, V. B. Matson, D. M. McDougal, C. W. Munda, R. D. Parkin, R. J. 1968. A Bonda dictionary. Poona: Deccan College & Post-Grad. Res. Inst. 1970. Kinship terms in the Munda languages, Anthropos 65, 444-65. 1965. Kharia phonology, grammar and vocabulary. Poona: Deccan College & Post-Grad. Res. Inst. 1929. Materialsfora Santaligrammar JI. Dumka: Santai Mission of the Northern Churches. 1934, 1935. A Santai dictionary, Vols. 2, 3. Oslo: Jacob Dybwad. 1985. Reciprocite et hierarchie, L'alliance chez les Ho et les Santai de l'Inde. Paris: Soc. Ethnographie. 1975. Ho grammar and vocabulary. Chaibasa: Xavier Ho Pub!. 1978. Ho-English dictionary. Chaibasa: Xavier Ho Pub!. 1986. Personal comunications. 1950. Bonda highlander. Bombay: Oxford Univ. Press. 1968. A grammatical sketch of Remo: a Munda language. Unpubl. dissertation. Ann Arbor: Univ. Microfilms. 1969. A critique of Verrier El win 's anthropology. In Anthropology and Archaeology, essays in commemoration of Verrier Elwin. New Delhi: Oxford Univ. Press. n.d. Korku-Hindi-English dictionary. Ramkherra, Dertalai: Central India Baptist Mission. 1968. The Didayi (Memoir 23, Anthropolog. Survey of India). Delhi: Govt. India Pub!. 1950. Encyclopaedia Mundarica, Vol. 2. Patna: Bihar Govt. Press. 1968-9. Kinship organization in India and place ofM undari-speaking people in it. Adibasi 10 (1), 1-25. 1954. (ed).Campbell's Eng/ish-Santali dictionary. Benegaria: Santai Christian Council, 3rd ed. 1963. Juang Vocabulary (Mimeo.) Chicago. [Included in Matson, 1964.] 1970-2, 1978-9, 1982-3. Gta? field notes. 1982. The structure of Kharia. Delhi: Centre Ling. & Engl., Sch. Lang., Jawaharlal Nehru Univ. 1964. A grammatical sketch of Juang, a Munda language. Unpubl. dissertation, Ann Arbor: Univ. Microfilms. 1963. The social structure of the Hill Juang. Unpubl. dissertation. Ann Arbor: Univ. Microfilms. 1965. Proto-Kherwarian phonology. Unpubl. M.A. thesis. (Mimeo.) Chicago. 1985. Munda kinship terminologies, Man (N.S.) 20, 705-21. 58

PAGE 67

Pfeffer, G. Pinnow, H.-J. Roy, S. C. & Roy, R. C. Stampe, D. Suryanarayana, M. Vidyarthi, L. P. & Upadhyay V. S. Vitebsky, P. Zide, A. R. K. Zide, N. Zide, N., Das, B. P. &. DeArmond, R South Munda kinship terms, I 1982. Status and affinity in Middle India (Beitriige zur Sildasien Forschung, Sildasien Institut, Univ. Heidelberg, 73). Wiesbaden: Franz Steiner. 1959. Versuch einer historischen Lautlehre der Kharia-Sprache. Wiesbaden: Harrassowitz. 1960. Beitriige zur Kenntnis der Juang-Sprache. Unpubl. typescript. 1937. The Kharias. Ranchi: 'Man in India' Office. 1979. A compendium of Sora vocabularies, with data from H. S. Biligiri, P. Donegan, B. P. Mahapatra, R. Mahapatra, D. Stampe, and S. Starosta. (Printout.) Columbus. 1977. Marriage, family and kinship of the Saoras of Andhra Pradesh. Unpubl. Ph.D. thesis. Andhra Univ. 1980. The Kharia: then and now. New Delhi: Concept Puhl. Co. 1979. Field notes on Sora kinship. 1968. A Gorum lexicon. (Unpubl. typescript.) 1970-2, 1978-80, 1983. Field notes on Juray. 1970. Field notes on Sora. 1982. Reconstruction of Sora-Juray-Gorum phonology. Unpubl. dissertation, Ann Arbor: Univ. Microfilms. 1956-8. Field notes on Korku. 1978. Studies in the Munda numerals. Mysore: Central Inst. Indian Langs. I 962-4, 1966-7. Field notes on Gutob. 59

PAGE 69

PROBLEMS AND PITFALLS IN THE PHONETIC INTERPRETATION OF KHASIORTHOGRAPHY tEugenie J. A. Henderson The growing interest in Khasi in recent years, as the lone representative of the Austroasiatic family in an area surrounded by Indo-Aryan and Tibeto-Burman languages, has resulted in a welcome number of serious studies of the language. A number of otherwise valuable studies have, however, been flawed by misunderstanding arising from the Khasi spelling system. This looks relatively simple on the surface but there are pitfalls for the unwary. Scholars with an English or Indo-Aryan linguistic back ground may be led astray if they are tempted to take the spelling at its face value and to trust their eyes rather than their ears. The sections devoted to phonetics and phonology in some studies I have seen have contained phonetic or phonemic 'transcriptions' which are little more than letter for-letter replicas of the spelling. 1 In these circumstances, it is worthwhile taking a closer look at problem areas within the spelling system itself. There have, I believe, been certain attempts during the last few years at minor revisions of the official orthography, and I cannot claim to have up-to-date knowledge of all of these. As far as I am aware, however, the system is substantially still the same as that used in Diengdoh (1965), Kharkongngor (1968), Bars (1973), and Blah (1974). I shall also refer in this paper to the older and probably better-known dictionary of U Nissor Singh (1906), and occasionally to even older forms. The pronunciations cited are from Cherrapunji speakers, since this variety of Khasi is accepted as 'standard'. Vowel length and final consonants What has to be borne in mind is that the Khasi roman spelling was devised in the first half of the nineteenth century, not by English or Scots or American missionaries, as in some other parts oflndia and Burma, but by Welsh missionaries, who introduced certain Welsh orthographic conventions unfamiliar to most English speakers. To the non-Welsh, one of the most confusing of these conventions relates to the use of the letters -p, -b, -t, -d, -c, -j 2 in word final position. Linguists accustomed to the phonemic differentiation of voiced and voiceless stops, in such languages as French and English for example, find it natural to suppose that Khasi has this sort of phonemic distinction also, I. This does not apply to Rabel (1961), whose phonetic observations are beyond reproach. 2. On the absence of final -g in Khasi, and on the special distribution of final -k, see Henderson ( 1965). 61

PAGE 70

EUGENIE HENDERSON when they are confronted for the first time with spellings such as those in (1) below: I. Khasi: dab 'bullock' kad 'to tear' mad 'to taste, try' sngab 'gill' dap 'full' kat 'so much, as much as' mat 'joint' sngap 'to hear' For those whose work has hitherto been mainly with Indo-European languages, there is nothing very remarkable about this kind of final alternation, but those who have specialised in the languages of South-East Asia to which Khasi is related, it would be not a little surprising to find a phonemic distinction between voiced and voiceless final stops. The characteristic typological pattern found all over continental South-East Asia is for the voice: voiceless contrast, which is frequently found in initial position, to be neutralised in final position. What we usually get is an unexploded voiceless stop, often accompanied by simultaneous glottal closure. Listening to the pronunciation of the Khasi words in question soon makes it clear that Khasi is no exception to the general rule-the final stops are voiceless, no matter what letter they are written with. What we also observe is that, in the great majority of cases, where the final letter in the spelling is b or d the preceding vowel is always long, while where it is p or t it is almost always short.3 The words in (1) above are regularly pronounced: 2. dab [da:p'] kad [ka:C] mad [ma:C] sngab [sIJa:p '] dap [dap'] kat [kaC] mat [maC] sngap [sIJap'] This seems very curious until it is realised that Welsh has just such a convention: vowels before the letters b, d and g are almost always long, whereas before p, t and c are almost always short. (For this rule, and for the exceptions to it, see Wells (1979).) 3. Welsh: mah [ma:l;>] 'son' tad [ta:cj] 'father' /log [io:] 'interest' map [map] at [at] /lac [iak] 'map' 'to, towards' 'slack' Phoneticians who have worked on Welsh claim that while the sounds written -b, -d, -g are voiceless finally, they are also lenis, or lax, as compared with those written -p, -t, -c, which are voiceless and fortis, or tense. Wells thus prefers to describe the lenis set as 'devoiced', here symbolised [l;>, cj, g], following the IPA convention. In Khasi there is no such tense/lax distinction, any more than there is a voice/voiceless distinction in final stops: they are all voiceless. It seems clear that the early Welsh missionaries made use of the their own spelling conventions in 3. See note 2 above. 62

PAGE 71

Problems in Khasi orthography order to mark distinctions of vowel length in Khasi. This convention works well with stops, where there are two sets of letters available, but it does not work for words ending in final nasals or in final r, which are always voiced. In these circumstances, in Welsh, a circumflex accent is used to mark the long vowel: 4. Welsh: tfm [ta:n] ton [to:n] glan [gla:n] 'fire' 'tune' 'clean' tan [tan] ton [ton] glan [glan] 'under' 'wave' 'bank' Earlier spellings of Khasi also tried this device. In Nissor Singh (1906) the circumflex accent is often used to mark vowel length before nasals and r; alternatively, a grave accent may be used to show length, or an acute accent to show shortness, but this is not at all regular, e.g.: 5. Khasi: biim [ba:m] 'to eat' sbiii [sba:i] 'cowry' leer [kE:r] 'to enclose' kper [kpE:r] 'garden' her [hE:r] 'to fly' itr [ur] 'to slip, fall' tur [tur] 'to butt' It appears probable that, in these words, there is some variation in usage among Khasi speakers, which may account for some of the seeming irregularities. Thus, we find itr 'to slip, fall', marked with a short vowel in Nissor Singh, but tur 'to butt' written without an accent, although also pronounced with a short vowel by my Khasi informants. Similarly, we find the spelling ker [kE:r] 'to enclose', but kper [kpc:r] for the derived form meaning 'garden, enclosure'. Her, marked short, was pronounced with a long vowel by my informants, but it may be that some speakers use a short vowel here. This irregularity and uncertainty in the use of accents may be one of the reasons why they seem to have been abandoned completely in most of the recent dictionaries. This leaves the problem, for the foreigner, of being unable to tell from the spelling whether the vowel is long or short in such words. Father Bars' (1973) dictionary, for example, has to resort to such entries as 'ham (long a)' and 'barn (short a)'. The vowel 'y' Another problem for linguists has been the vowel spelt y in Khasi. It is almost always unstressed and behaves in much the same way at the shwa vowel [::i] in English, as in '~bout, ~gain, CQllapse, S:!:!J)pose', etc. When we look at Welsh, we find this similarity is no accident, since y is the Welsh way of spelling the vowel [;i]. Depending upon its position in the word, y may in Welsh also represent a vowel which is pronounced either [1] or [i], according to dialect, e.g.: mynydd ['m;imo, 'm;inio] 'mountain'. y, with the [;i] value, is also the vowel that is used where a cushion vowel or epenthetic vowel is felt to be needed to break up awkward or alien consonant clusters, as in the forms ysg8r-sg8r [;isgo:r, sgo:r] 'score'. The rather old fashioned Khasi spellings kypa 'father', bydi 'twenty', bynai 'moon', month', etc., show the same process at work. Since clusters like [kp, bd, 63

PAGE 72

EUGENIE HENDERSON bn], and many others, are difficult for Welsh or English speakers to pronounce without a short vowel in between, the y vowel was inserted. More recently the convention on the whole is not to write y between 2-consonant clusters, but to insert it where there are clusters of more than two consonants, e.g. bna 'to hear', bdi 'twenty', but byndi 'to mortgage'. There seems still to be some uncertainty about this, however, as Father Bars still gives spellings like byna. Phonologically speaking, there would be much to be said for dropping they in the 3-consonant clusters as well, simply writing bndi, which is perfectly clear. This is the solution proposed by Dr. Rabel in her phonemic analysis of Khasi (Rabel 1961 ). Final palatals Khasi has final palatal consonants; Welsh does not. What appears to have struck the ears of the missionaries who first listened to Khasi was the i-like glide that precedes the final palatals. We thus have early spellings such as skain 'fly' n., bysein 'snake, python', etc., and it seems clear that the final consonant was thought of as plain n in such words, though its palatal quality was later recognised by the use of the letter ii: skaiii, bseiii, etc. Since there is no accepted roman equivalent for the final palatal stop, -d and -t continued to be used, except after ie, when we findj, e.g.: 6. Khasi skaiii [ska:p] kshaid [kJa:c'] btuid [btu::;ic] miej [mbc] 'fly n.' 'waterfall' 'slippery' 'table' kruiii [krup] 'white ant' kait [kac'] 'plantain' buit [buc'] 'cunning' From the phonological point of view, the vocalic segments preceding these consonants should be interpreted as plain vowels (or in some cases as centring diphthongs), with an automatic palatal on-glide to the final consonant, rather than as phonemes /ei, ai ui/, etc. Father Schmidt, writing on Khasi as early as 1904, recognised this; his knowledge of the Austroasiatic family led him to expect that final palatals might be present and he interpreted the missionaries' spelling in this light. Representation of the glottal stop The representation of the glottal stop, which is a consonant phoneme in Khasi, does not derive from a Welsh convention, but rather from a much more general missionary usage, found also in the roman orthographies of Tibeto-Burman languages such as Lushai and Chin. Word-initially, the glottal stop is not written at all; word-finally it is spelt with -h; which is not otherwise needed in final position: 7. Khasi khlieh [khle?] 'head' soh [s:,?] 'fruit' spah [spa?] 'hundred' kseh [ksi:?] 'pine' Problems arise when [?] occurs medially, or when it is the second consonant of an initial cluster. Devices used in such contexts include hyphens, apostrophes, or the vowel letter y. In the last case, the y 64

PAGE 73

Problems in Khasi orthography presumably represents the slight vocalic sound that can be heard as the glottal stop is released: 8. Khasi ha-oid, ho-oid, h'oid [ha?;:i:c'] 'Yes'; sh'iap, shyiap [J?iap'] 'sand'; !'er, Iyer [l?E:r] 'wind'; s'iar, syiar [s?iar] 'fowl'; lyoh [l?;:,?] 'cloud'; syiem [s?i:Jm, s?e:m] 'king, chief In the word for 'cloud' above, we find the glottal stop represented by h finally, and by y in the initial cluster. Dental/alveolar distinction It is common for Western linguists to expect that t, th, d, n will form a natural class, united by a common place of articulation. In Khasi, however, as in certain other languages in the area, this turns out not to be the case. [t] and [th] are dental, [d] and [n] are alveolar. This is not simply a matter of phonetic detail; it is important in understanding the rules for permitted sequences of plosives and nasals in Khasi initial clusters. At first sight it appears that a sequence of virtually any two consonants may form an initial cluster, but while the possibilities are very rich, there is a constraint upon sequences of plosives and nasals whereby those with the same place of articulation may not form a cluster together. For example, bn, bt, bth, pd, pn, bn, thl], di], dp, km, kt, etc., are permitted; pb, pm, bm, kl], khlJ, are not; td, tn and thn are permitted, and seem to be exceptions until it is recognised that the sequences here are also hetero-organic, and hence perfectly regular. Ambiguous vowel spellings The letters u, o, and e,4 and the digraph ie are sometimes phonetically ambiguous. Finally and before a glottal stop, ie represents a pure vowel, a very close [~]. Before other consonants, pronunciation varies between [b] and [e:], depending upon the speaker: 9. khlieh [khle?] 'head' lieh [le?] 'white' sdie [sde] 'axe' kmie [kme] 'mother' dieng [di:JIJ, de:IJ] 'tree, wood' ktien [kti:Jn, kte:n] 'word' In open syllables, the spelling o represents a very close [o] in some words,[;:,] in others. There seems to be variation in usage between speakers for some words: 10. ro [ro or r;:,] -to [to] 'quicksilver'; 'that (enclitic)'. to [b] (indicating assent); In closed syllables, the phonetic value is always [;:,]. 4. For a more detailed account of Khasi vowels, see Henderson (1967). 65

PAGE 74

EUGENIE HENDERSON u before a final glottal stop is [u] in some words, [o] in others: l l. ruh [ru?] 'also, indeed' ruh [ro?] 'cage' duh [do?] 'lose' shniuh [JJ10?] 'hair' Before other consonants, u may represent long [u:;:i]. [o:] or short [u]: 12. bud [bu:;:iC, bo:C] 'follow' thub [thu:;:ip', tho:p'] 'reduced in size' tup [tup'] 'cannon' /um [[lu:;:im, lo:m] 'hill' /um [lum] 'gather' In open syllables, e may represent either close [e] or open [i:]: 13. te [te] 'then, but' de [di:] 'also', me [mi:] 'thou' re [re] 'or' In closed syllables the value is always [i:]. To sum up, with a proper understanding of its origins and background, Khasi spelling has much to recommend it. Once mastered, the conventions regarding the use of the letter y, the use of -b, -d, -j to indicate length of the preceding vowel, and the conventions for the representation of the glottal stop, are in the main clear and unambiguous. A reasonable complaint might be that vowel length before nasals is commonly not shown. However, in my view, the traditional spelling, properly understood, represents Khasi usage far more accurately than the misinterpretations and attempted 'improve ments' that have been suggested by some linguists with an Indo-Aryan linguistic background, and little or no understanding of the strong Mon Khmer legacy that still survives in Khasi phonological structure. REFERENCES Bars, E. 1973. Khasi-English dictionary, Shillong: Don Bosco Press. Blah, V.E. 1974. (comp.) Chapa/a's Anglo-Khasi dictionary. (3rd ed.) Shillong: Chapala Book Stall. Diengdoh, A.K. 1965. Leemue/'s Anglo-Khasi pocket dictionary. Shillong: L. Harrison. Henderson, Eugenie, 1965. Final -k in Khasi: a secondary phonological pattern. Lingua J. A. 14, 459-66. 1967. Vowel length and vowel quality in Khasi. Bull. Sch. Or. Afr. Stud. 30(3), 564-88. Kharkongngor, U I. 1968. Ka Dienshonhi ( a Khasi-Khasi dictionary) Shillong: Ri Khasi Press. Rabel, Lili 1961. Khasi: a language of Assam. Baton Rouge: Louisiana State Univ. Press. Schmidt, W. 1904. Grundzuge einer Lautlehre der Khasi-Sprache in ihren Beziehungen zu derjenigen der Mon-Khmer Sprachen. Miinchen: Verlag der K. Akademie. Singh, U Nissor 1906. Khasi-English dictionary. Shillong: Eastern Bengal & Assam Secretariat Press. Wells, J.C. 1979. Final voicing and vowel length in Welsh. Phonetica 36(405), 344-60. 66

PAGE 75

HU -A LANGUAGE WITH UNORTHODOX TONOGENESIS Jan-Olof Svantesson In this article I will describe and analyse a small vocabulary I collected in September 1984 from a native speaker of Hu /xu?/, a Mon-Khmer language spoken by about I 000 persons in a few villages in the Xiao Mengyang area in Jingh6ng county, Sipsong Panna (Xishuang Banna), Yunnan province, China. The Hu are known among the local Chinese as Kongge. Hu belongs to the little-known Angkuic group of the Palaungic branch of the Mon-Khmer languages. Small Angkuic populations are scattered over south-western Yunnan province, and in another article (Svantesson 1988) I describe the language of another of these, U, spoken in the village Paa: XEp (Bangxie) in Shuangjiang county. The place of Angkuic within the Palaungic branch is shown in the following table according to Diffloth (1982a): Palaungic East Palaungic Waic Angkuic Lamet West Palaungic Danaw Riang Rumai Although my material is too small to allow a complete synchronic phonemic analysis, the most important historical developments in Hu can be inferred from it. From a general phonological point of view, the most interesting phenomenon is the development of a two-tone system where the tones are not the reflexes of voiced/voiceless pro to-initials, as is most often the case in Mon-Khmer two-tone (or two-register) languages. Instead, the tones are the reflexes of the long/short vowel opposition which existed in Proto-Palaungic (inherited from Proto-Mon-Khmer). As far as I know, no language with this kind of tonogenesis has been described before. 67

PAGE 76

JAN-OLOF SVANTESSON Initial consonants The Angkuic languages are characterised by a 'Germanic' development of the initial stops-that is to say, voiceless stops have become aspirated, and voiced stops have become voiceless unaspirated. This is illustrated by the following examples: Hu u Lamet N. Kammu S. Kammu *voiceless phiJt phet Pifl Plfl pi./1 'to shoot' thafl than taafl taafl taafl 'to weave' khap khap kaap kaap kaap 'jaw' *voiced pafl pan paajl 'white' pKf? qi prii? pri? bri? 'forest' phltak ?atar pltaak ktaak kdaak 'palm ( of hand)' kaTJ kaa kaalJ gaalJ 'house' kak kar kak 'to bite' In Southern Kammu (as recognised from Lindell et al. 1981), the original Proto-Mon-Khmer voicing contrast is retained. The unaspirated stops in the Angkuic languages Hu and (Pao Xi:p) U correspond to voiced stops in Southern Kammu, and the aspirated stops correspond to voiceless unaspirated. Original voiced and voiceless stops have merged in Lamet (Rmeet; from Lindell et al. 1978) and Northern (Yuan) Kammu, giving rise to lax and tense register in Lamet, and low and high tone in Northern Kammu, as is the case in Mon-Khmer and other languages with 'orthodox' register or tone development*. The examples also show that the Hu tones are not the result of orthodox tonogenesis. Hu has a contrast between initials(with the allophone[~] before i, and in the word sKelJ 'red') and 0-, an opposition which is not present in U or other Angkuic languages, but which is found in Danaw (Luce 1965), as a contrast between ts-and 0-. In Lamet, and in the rest of Palaungic (and in Kammuic), Hu s-and 0correspond to s-and h-, respectively. Diffloth (1977) and Ferlus (1978) reconstruct these as Proto-Mon-Khmer *cand *s-, respectively. According to Diffloth (1977), *c-became *tsin Proto Palaungic. Examples are: Hu u Lamet Danaw *cs5? so s5? tso1 'dog' safe? sale slie? kale' 'rain' nsf? nchi sf? tsi1 'louse' *In the Northern Kammu form ktaak, the tone is determined by the voiceless k. (Ed.) 68

PAGE 77

Hu A language with unorthodox tonogenesis *s0um sup huum 0:m4 'to bathe' pa0e]l set phijl pa0en4 'snake' 0a?3Jl s?5:J]l 'dry' 0amii? sama ?mfla? 'wind' U has the regular reflex eh-of *cafter minor syllables, as in the example 'louse'. Proto-Palaungic also had *h-, which is retained as such in all languages but, unfortunately, I failed to elicit any word with *hin Hu. There are some words in which Palaungic s-corresponds to c-in Kammu-they include Hu sii1J 'bitter', Kammu ciilJ. Here, Ferlus reconstructs *tsin Palaungic, Kammuic and Viet-Muong (correspond ing to two Proto-Mon-Khmer initials, *tsand *ts-). Minor syllables As in other Northern Mon-Khmer languages, most non-compound words are monosyllabic or sesquisyllabic, i.e. consisting of a major syllable preceded by an unstressed minor syllable (cf. Shorto 1960). Minor syllables have a syllabic sonorant (such as m in 0mphup 'lung') or a (probably) non-contrasting vowel which I have written as a (ka?a 'two'). There is also a contrast between 0-and s-minor syllable initial in Hu (and in one word, ts-is attested). This opposition is not maintained in this position in U, Lamet or Proto-Waic (=PW; Diffloth 1980), where the most common corresponding initial is s-: Hu u Lamet PW 0anat niit snaat *snat 'gun' 0a?aw sa?fl s?iiar 'sour' 0a?3Jl s?5:J]l 'dry' 0avil1J savaa 'to ask' 0athiin sathat 'old' 0athii? satha ntiia? *snta? 'tail' 0amii? sama ?mfla? *?ma? 'wind' 0mphup saphop 'lung' samo? samo *smo? 'stone' safe? sale slf1e? *hi&? 'rain' saplil1J saxaa smplaalJ 'shoulder' SalJGY !Jay 1Jaay *?JJay 'eye' S1Jkho? khit *JJko? 'yesterday' tsa1Jill SalJfo SIJllaf *Sl)aJ 'blue' Presumably, 0-and s-are the reflexes of Proto-Palaungic *s-and *ts( < *c-), respectively, which have merged into s-in U, as usual. It may be noted that *shas (at least in some cases) been retained in this position in 69

PAGE 78

JAN-OLOF SVANTESSON Lamet and Waic. As mentioned above, word initial *s-has usually become h-in these languages. There are irregularities, which may be due to more complex initial consonant clusters. The occurrence of minor syllable initial ts-in tsa1Jill 'blue' suggests a different Proto-Palaungic consonant, presumably the reflex of Ferlus' (1978) *ts-, since the Kammu cognate is c1Jf1ar with initial c-. Hu has also retained the contrast betweens( < *ts-< *c-) and 0( < *s-) in major syllable initial position after a nasal minor syllable: Hu u Lamet PW nsf? nchi sf? *si? 'louse' nasok suf" yook *hyok 'ear' n0iic ntshat maac *hmac 'sand' n0im nchzp lmhiim *mhem 'claw' Here, U has s in 'ear', where the minor syllable has disappeared, otherwise eh or tsh (which are probably allophones of a single phoneme). After a nasal minor syllable, s never occurs in U, but has developed into tsh/ch, which accounts for the unexpected occurrence of these consonants in the words for 'sand' and 'claw'. Hu also has a number of words which have a nasal minor syllable followed by a voiceless nasal major syllable initial: Hu u Lamet PW m;iiim saniim naam *hnam 'blood' m;iim sanap nim *nYm 'year' nlJiit sanii 'comb' mJe? ne *n?ne? 'meat' mY[lul mun kmuul *kmil 'silver' nJ!ilt Jlar-k]liias *kjlaS 'to laugh' The Hu forms suggest a *nasal+ h initial cluster, while Lamet and Waic in some cases have clusters with a stop and a nasal. Taken together, this implies pro to-forms with *stop+ nasal+ h clusters. Medial h has been lost in most of Palaungic (cf. Diffloth 1977), and in Hu, where his retained, the initial stop has disappeared. U often loses cluster initital stops (see Svantesson 1988), as is the case in the last three examples above. In the preceding list, the minor syllable sa in the first three words in U, taken together with evidence from outside Palaungic, suggests Proto-Mon-Khmer *J-(palatal voiced stop). This tallies perfectly with Diffloth's (1980: 175) reconstruction of 'blood' as Proto-Mon-Khmer *Jnhaam. For the other two words, *Jnhclusters are also supported by evidence outside Palaungic: Kammu crlas (Southern Kammu;rias) 'comb', Mon cnZim 'year'. 70

PAGE 79

Hu A language with unorthodox tonogenesis Final consonants Hu has retained the Proto-Palaungic (and even Proto-Mon-Khmer) system of final consonants rather intact. An exception is final *-s, which has developed into -t (merging with original *-t), another innovation shared by Hu and Danaw (and several other Mon-Khmer languages as well): Hu ?axet phot Danaw k'r1'et3 pYt3 Lamet kriis poos 'bear' 'sambar deer/barking deer' The final palatal stop *-c sometimes appears as-tin Hu (cf. Hu pet 'to spit', Lamet pee; Hu ?amut 'mosquito', Lamet rmuuc 'ant'), but my data are too limited to reveal the exact circumstances under which this has taken place. Initial *ris realised as a uvular [If] (as is also the case in some Lamet dialects, in Bulangshan Blang, and in some other languages of the area). In final position, [If] is pronounced as a rather vocalic uvular glide which can be written [a] (incidentally a development which has also taken place in my own southern Swedish dialect! See Lindau 1985 for different kinds of r). Thus ?lK 'fowl' is pronounced [?iii]. After the vowel a, final *-r has disappeared, or is retained as -w: Hu Lamet ka?a ?laar 'two' ma maar 'field' 0a?aw s?aar 'sour' kaw kaar 'they (dual)' Tones and vowels There are two tones in Hu, namely, high (denoted by over the vowel) and low('). The co-occurrence of tones, vowels and final consonants is restricted, as shown in this table: Vowel: 1 u 1 e g a ;i 0 c) 0 Final ? others As mentioned above, the general rule is that originally long vowels have conditioned low tone, and short vowels have conditioned high tone, and the co-occurrence restrictions probably reflect restrictions on the co occurrence of long and short vowels with final consonants by the time 71

PAGE 80

JAN-OLOF SVANTESSON that the tone system was formed. Since Lamet is a Palaungic language which retains the Proto-Palaungic (and even Proto-Mon-Khmer) length distinction, I will use Lamet examples for showing the relationship between Proto-Palaungic vowel length and Hu tones. For non-high vowels, the relationship is quite clear-cut: Hu Lamet *short vowels: yam yiim 'to die' pa0an phan 'five' mejl krmljl 'star' nci:n kcin 'heavy' *long vowels: yiim yiiam 'to cry' Iek llik 'pig' ?3m ?{JOm 'water' nasok yook 'ear' Before a final glottal stop, the tone is always high, probably due to shortening of the vowel in this position before the development of tones. For comparison, forms from (Northern) Kammu are given. This language, which has orthodox tonogenesis, is another language where the length distinction in vowels is lost before a glottal stop. Examples: Hu Lamet Kammu s5? s5? s5? 'dog' kathe? kta? pte? 'earth' ?j? ?jJ? ?o? 'I' phfe? pfee? pie? 'fruit' The high vowels i and u always have high tone, except before Kor in open syllables, where both tones occur. Compare the following examples with long proto-vowels (I have no examples with long *ii): Hu Lamet ?asim siim 'bird' phKfm priim 'old' mY[!ul kmuul 'silver' ?up ?uup 'cooked rice' 0um huum 'to bathe' But: ?)K ?frr 'fowl' One possible explanation for the absence of a tone contrast in the high vowels is that they have higher intrinsic pitch than low vowels, as has been shown for many different languages (see, e.g., Lehiste 1970: 68-71 ), which might have conditioned high tone for both long and short high vowels when the Hu tone system developed. This explanation does not, however, account for the occurrence of high vowels with low tone on open syllables. 72

PAGE 81

Hu A language with unorthodox tonogenesis On the other hand, there are indications that the length contrast was already lost in the high vowels in Proto-Angkuic, i.e. before the development of the Hu tone system. 1 Thus, judging from the words given in Diffloth (1982a), there is no length contrast for u and i in Mok, while the contrast is retained for non-high vowels. In U, the vowel length contrast has disappeared, although it has left traces in final nasals which are retained after originally long vowels but have become stops after originally short vowels. Denasalisation has, however, taken place after both *long and *short i and u, which are thus treated as if they were short. The following examples show this development. Lamet cognates are given because they retain Proto-Palaungic vowel length: Lamet Hu u Mok *non-high long: yaam yam yam Jlaam 'to cry' ?bom ?3m ?bm ?oom 'water' poon ?aphon phon phoon 'four' *non-high short: yam yam yap yem 'to die' ntam ntham nthap tham 'egg' kci:n ncen kit koeen 'heavy' *high long: slim ?asim pach)p ?a-sim 'bird' kUUJl khu]l khut khun 'male' priim phKfm x1p phim 'old' *high short: kb; (khi,;) khzr khilJ 'head' (The Hu word khi1J, which occurs in khi1J ko1J 'knee' may be cognate to the words meaning 'head' in the other languages.) The reason for the loss of vowel length in the high, but not in the other, vowels may be their shorter intrinsic length, something which has been attested for various languages (see Lehiste 1970: 18-19). Furthermore, the length contrast in the high vowels seems to have carried a rather small functional load. As mentioned above, final -K is more or less vocalic, which may explain why low tone can occur on high vowels before this final, as it does in open syllables. The reason why o and :J occur only with low tone (except before ?) in my data is probably that there were relatively few words with short *o and *:). This is the case in Kammu, where short and long :J do not contrast (see Svantesson 1983). I. Proto-Angkuic did not have tones, as is proved by the absence of tones in Mok (Diffloth 1982a). Uhas a tone system, which is different from that in Hu, and for the other Angkuic languages it is difficult to know whether they have tones or not, since they are known only from older and not very reliable sources, which do not give any tones. 73

PAGE 82

JAN-OLOF SVANTESSON Vowel length or tones? Figure 1 below shows some typical examples of the fundamental frequency (F 0 ) contours of the two tones on different types of syllables: Hz yam yam pap khap 300 ,--....._ -.......... 200 __,,.--_____ ---........ 100 100 200 300 400 500 ms Fig. 1. In these cases, as in many others, there is a co-vanat10n between fundamental frequency and vowel duration, so that vowels carrying low tone have longer duration than vowels carrying high tone. One might, therefore ask whether an analysis in terms of vowel length rather than tones is possible. To investigate this, the duration of the vowel and the mean value of the fundamental frequency over the vowel were computed (using the ILS* program package) for some words with high and low tone. The words were said in isolation by the female informant. Since the tone contours are rather flat, the average frequency value can be used to characterise the tones. The results are given in Table 1. As seen in this table, vowels carrying low tone often have longer duration than those with high tone, as seen in the only recorded minimal pair, yam 'to die' and yiim 'to cry'. On the other hand, some vowels with high tone are longer than some with low tone, and this overlapping of duration can be taken as evidence against treating vowel length as distinctive. Each of the analysed high-tone words also has higher fundamental frequency than each of the low-tone words. Thus, fundamental frequency is definitely a consistent phonetic correlate of the investigated opposition, while vowel length may be regarded as a concomitant factor. It seems, therefore, reasonable to analyse the opposition as one consisting of two tones. ILS = Interactive Laboratory System (Ed.). 74

PAGE 83

Hu A language with unorthodox tonogenesis Table 1 Mean values: F0 (Hz) Duration (ms) F0 Duration High tone: yam 269 130 247 135 263 120 263 120 260 126 pap 253 115 249 95 242 95 248 102 kak 253 100 252 130 258 120 254 117 Low tone: yam 214 200 215 175 215 225 215 200 khap 204 115 203 130 208 115 205 120 ?ak 205 180 201 205 201 180 202 188 Note: The F0 ranges are 242-269 Hz for high tone and 201-215 for low tone. Conclusion The Hu data raise a number of intriguing questions, two of which will be discussed here. One concerns the classification of Palaungic. There are some striking similarities between Hu and Danaw. In particular, both have 0where Proto-Mon-Khmer has *s-, and this is found nowhere else in Palaungic. In the sub-classification of Palaungic given in Diffloth (1977), Danaw is close to Angkuic, but in Diffloth (1982a), a classification (shown on p. 67), which puts Danaw and Angkuic rather far from each other is given. Mitani (n.d.), using lexicostatistical methods for classifying Palaungic, also finds that Danaw and Angkuic are rather distant from each other. One might ask if my data from Hu-being an Angkuic language which shares the innovation *s-> 0with Danaw-necessitates a revision of this. Not necessarily, since it is quite possible that the development *s> h-, which has taken place in all Palaungic languages except in Angkuic and Danaw, is the final result of two different processes: *s> 0 and 0> h-. If that is the case, Hu and Danaw are not languages which share an 75

PAGE 84

JAN-OLOF SVANTESSON early innovation, but rather languages where a phonological rule (0> h-) has not applied, and there is no reason to assume that they have branched off together from the rest of Palaungic at some early time. Another question concerns tonogenesis. In almost all cases where a Mon-Khmer language has acquired tones (or registers), the development of the tone ( or register) system is the result of a loss of contrasts-usually voicing contrasts-in the consonants, but in Hu, the tones have developed in connection with the loss of vowel length. The only other case known to me, in which tonogenesis of a similar kind may have taken place, is in Estonian, where a tonal distinction (different from that in Hu) has possibly developed from an earlier vowel length opposition (see Lehiste 1978). One somewhat similar case in Mon-Khmer is Pacoh which, according to Diffloth (1982b), has an unorthodox registrogenesis, where a register difference has replaced an earlier difference in vowel quality. Both the acquisition of tones and the loss of vowel length are ongoing processes in the area where Hu is spoken, so it is perhaps not surprising to find a language that combines both. There might be a phonetic explanation as to why long vowels have acquired low tone and short vowels high tone, since there seems to be a general tendency for vowel duration and fundamental frequency to vary inversely with each other. As already mentioned, it has been shown for many languages that high vowels have intrinsically shorter duration and higher pitch than low vowels. For instance, measurements of the intrinsic pitch and duration in Standard Chinese vowels (putonghua) have shown that, if other factors are constant, vowels with relatively high intrinsic pitch also have relatively short duration (Shi Bo, pers. comm.). Naturally, pitch differences of 40 Hz or more, as are found in Hu, are not the result of automatic adjustments, but an originally non-distinctive pitch difference could have taken over some of the functional load carried by vowel length, eventually acquiring phonemic status. Vocabulary The vocabulary is presented in reverse alphabetical order. ka?a 'two' ma 'dry field' maKa 'thing' pa 'not' Kl 'to go' ?amo 'one' ?a?3 'monkey' m3 'axe' tu 'bad' /ii? 'leaf 0amii? 'wind' l]ii? 'to itch' maKii? 'to steal' 0athfi? 'tail' ?e? 'we (pl.)' ke? 'they (pl.) safe? 'rain' phte? 'fruit' me? 'you (sg.)' 76

PAGE 85

Hu A language with unorthodox tonogenesis kame? 'dream' thame? 'new' nl}e? 'meat' phe? 'you (pl.)' ee? 'tree' the? 'to sit' kathe? 'earth' ve? 'left (side)' katfJ? 'nose' ?? 'person' 1)? 'day' klfi? see po?-1JKI? 'pestle' pKf? 'forest, mountain' pas? 'string' nsi? 'louse' pa0? 'blood-vessel' thi? 'hand' 1jkho? 'rice (husked)' S1]kho? 'yesterday' samo? 'stone' po? -kKi?: 'spirit' ?5? 'I' 0ak/5? 'bark ( of tree)' s5? 'dog' kat5? 'banana' pa/u? 'salt' xu? 'Hu' n0ac 'sand' lih 'to go down' ?ak 'bow' kak 'to bite' thKak 'buffalo' eak 'rice (plant)' phltak 'palm ( of hand)' nthak 'tongue' tek 'pig' phKek 'ribs' ?a0ek 'rat' tek 'small' 0ak -Wi1J: 'flea' nasok 'ear' nth3k 'head' euk 'hair' thuk 'to hang' 1JC1l 'fire' tSa1)a/ 'blue' mphal 'mortar' phel 'wing' nt3l 'wine' mrz1ul 'silver' katul 'belly' nl}am 'blood' ntham 'egg' yam 'to die' yam 'to cry' ntm 'younger brother' ?a0tm 'right (side)' ?fm 'to live' Kim 'village' phKfm 'old (of things)' ?asim 'bird' n0im 'claw' nl}im 'year' ?3m 'water' kath3m 'liver' num 'piss' t1)Kum 'under' 0um 'to bathe' tan 'long (in space)' pa0an 'five' 77

PAGE 86

JAN-OLOF SVANTESSON than see nay, ?[qz 0athan 'old' nctn 'heavy' ten 'low' ?{Jn 'he, she, it' man 'long (in time)' phin 'woman' ?aphon 'four' kh3n 'child' kh3n thf?: 'finger' ka?iqz 'wasp' pilfl 'white' thilfl 'to weave' meJl 'star' pa0efl 'snake' phi_ 'to shoot' 0a?3fl 'dry' ?ufl -than: 'grandfather' ?a?ufl 'father' khilfl 'man; husband' maKUJl 'ant' ntilfl 'mouth' ka?il1J 'bone' kil1J 'house' ka1J -xaw: 'heaven' saplil1J 'shoulder' ka1Ja1J 'iron' K[llJ 'flower' maKalJ 'horse' SalJ 'bitter' thii1J 'to kill' pavalJ 'tomorrow' 0avil1J 'to ask' CelJ 'foot' le1J 'high' SKelJ 'red' nte1} 'big' the1J 'drink' pha1J 'many' 0{J1J -?zK: 'bedbug' khi1J 'tooth' khi1J -ko1J: 'knee' maKilJ 'crab' WflJ see 0ak-0i1J 'bamboo' ko1J see khi1J-molJ 'to look' ?a?31J 'wasp' X31J Mekhong y31J 'good' khu1J 'wet field' khilp 'jaw' pap 'to speak' 0a?ep 'rainbow' tep 'blind' dp 'to run' ?up '(cooked) rice' 0mphup 'lung' ?zK 'fowl' khzK 'moon' pha0zK 'bee' kiK 'finished' phiK 'to fly' kfUK 'to sew' pha?at 'to swell' khiit 'sick' 0anilt 'gun' nf}ilt 'comb; to comb' rwat 'to laugh' thasilt 'lightning' pet 'to spit' khaset 'charcoal' ?axet 'bear' ?ft 'to sleep' ?at 'to be at' phot 'sambar deer' 78

PAGE 87

Hu A language with unorthodox tonogenesis 03t 'barking deer' ?amut 'mosquito' thut 'breast' 0a?aw 'sour' kaw 'they (dual)' ?apalaw 'fish' phaw 'you (dual)' REFERENCES Diffloth, G. Ferlus, M. Lehiste, Tise Lindau, Mona vaw 'wide' xaw see km;-?a,;i:w 'cat' khiw 'green' ?jjy 'we (dual)' khay 'to eat' ?a/ay 'squirrel' l)ay -l]f?: 'sun' sanay 'eye' lJ?l)ay 'far' nay than: 'grandmother' ?anay 'mother' khoy 'to have ka?3y 'three' KJJ 'hundred' 1977. Mon-Khmer initial palatals and 'substratumized' Austro-Thai. In Mon-Khmer Stud. 6 (ed.) P. N. Jenner et al. Honolulu: Univ. Hawaii Press, 39-57. 1980. The Wa languages.(= Ling. Tibeto-Burman Area 5(2)). 1982a. Subclassiflcation of Palaungic and notes on 'P'uman'. Paper presented at the 15th Sino-Tibetan Conference, Beijing, August 1982. 1982b. Registres, devoisement, timbres vocaliques: leur histoire en katouique. In Mon-Khmer Stud. 11 (ed.) P. N. Jenner. Honolulu: Univ. Hawaii Press, 47-82. 1978. Reconstruction de /TS/ et /TS/ en Mon-Khmer. In Mon-Khmer Stud. 7 (ed.) P. N. Jenner. Honolulu: Univ. Hawaii Press, 1-38. 1970. Suprasegmentals. Cambridge, Mass.: MIT Press. 1978. Polytonicity in the area surrounding the Baltic Sea. In Nordic prosody (eds.) Eva Garding, Gosta Bruce and Robert Bannert. Lund: Lund Univ., Dept. Ling., 237-47. 1985. The story of /r/. In Phonetic linguistics: Essays in honor of Peter Ladefoged (ed.) Victoria From kin. Orlando: Academic Press, 157-68. Lindell, Kristina, Svantesson, J.-O. & Damrong Tayanin 1978. Two dialects of the Rameet (Lamet) language. Cahiers Ling. Asie Orient. 4, 5-22. Luce, G. Mitani Yasuyuki Shorto, H. L. Svantesson, J.-O. 1981. Phonology of Kammu dialects. Cahiers Ling. Asie Orient. 9, 45-71. 1965. Danaw, a dying Austroasiatic language. Lingua 14, 98-129. n.d. Problems in the classification of Palaungic. Ms. 1960. Word and syllable patterns in Palaung. Bull. Sch. Or. Afr. Stud. 23 (3), 544-57. 1983. Kammu phonology and morphology (Travaux Inst. Ling. Lund XVIII). Lund: Gleerup. 1988. U. Ling. Tibeto-Burman Area 11(1), 64-133. 79

PAGE 89

ON AUSTRONESIAN LEXICON IN VIETNAMESE1 Kenneth Gregerson The sources of Vietnamese lexicon have been much discussed as regards Chinese, Tai, and Mon-Khmer vocabulary (Maspero 1912; Schmidt 1926; Haudricourt 1954), and one may now presume that Vietnamese is rather safely established as an Austroasiatic language. The Austronesian (AN) subset of Vietnamese lexicon, however, while clearly not having gone unnoticed ( qv. discussion in Benedict 1976),2 continues to invite a good deal of sorting out. The present paper mentions some well-known Austronesian forms as well as raising questions about whether certain other Vietnamese items are ultimately of Austronesian (or Austro-Thai?) origin. The effort here is to contribute towards a more deliberate Vietnamese perspective on Austronesian lexical associations with Austroasiatic. I. Function forms Vietnamese possesses several very regularly used grammatical or closed set function words that appear to have Austronesian counterparts. 1.1. Aspectuals The following forms function as regular Vietnamese (VN)3 aspectuals: (I )a. d?i 'already Toi d?i mua xe roi. I already buy auto already 'I have already bought a car.' b. slip 'about to' Toi slip mua xe. I about-to buy a car 'I am about to buy a car.' c. dang 'in process' Ong ily dang grandfather that in-process 'He is working.' lam vifc. do work I. A version of this paper was presented at the Eighteenth International Conference on Sino Tibetan Languages and Linguistics (Bangkok) in 1985. I gratefully acknowledge helpful comments from Paul Benedict, James Matisoff, Nguyn Dinh Hoii, and Bill Gage, though certain things still survive that are perhaps not to their liking. 2. The ultimate Austronesian origin of Vietnamese is a notion that has existed for some time (cf. e.g. discussion in E. Sauvignet 1922; K. Wulff 1942; as well as by Binh Nguyn L9c, c.1960 in his Ngubn g8c M!ilai, so Prof. Hoii informs me). 3. Some, though not all, Vietnamese illustrative forms are from Nguyn Dinh Hoa, 1971. 81

PAGE 90

KENNETH GREGERSON d. mai 'continuative' Ho di bij mcii dn Thu-fJwc. They go by-foot continue to Thu-Bfrc 'They walked all the way to Thu-Bfrc.' These forms may be compared with the following in Bahasa Indonesia (BI):4 (2)a. sudah 'already' Ja sudah pergi. he already go 'He has already gone.' b. siap 'ready' Mereka siap untuk they ready for 'They are ready to go.' c. sedang 'while, in process' pergi. go Ja sedang membaca ketika he in-process read when 'He was reading when I arrived.' d. masih 'still, yet' ia masih tidur. he still sleep 'He is still sleeping.' saya datang. I arrive As exemplified above in la and 2a, VN da 'already': BI sudah 'already' both occur in preverbal position to signal Perfective aspect. In VN it will be observed that a clause-final r8i reinforces the same completive meaning redundantly. This form is perhaps the 'true' Austroasiatic marker (cf. Rengao (Rg.) hadroi 'before') into whose territory da has intruded. The tone on da accords well with diachronic expectations clarified originally by Haudricourt (1954), which may be generally summarised as follows: (3) Original Finals open final final syllable stop spirant voiceless CV(N) eve ., CVH high tones original initials voiced CV eye CVH low tones 4. A number of the sample Indonesian sentences and lexicon are from Echols and Shadily (1974). 82

PAGE 91

On Austronesian lexicon in Vietnamese Thus, depending on the original voicing status of the initial consonant (C-) and the closure status of the final (-V(N), -VC, or VH), a particular contrastive tone has developed in Vietnamese, typically, with a concomitant loss of most of the original conditioning features. The High Tones associated with old voiceless or imploded consonants are V (symbolised in our exposition with V for clarity though it (the macron -) is not written in Vietnamese orthography), V, and V. The Low Tones that originally occurred with voiced initial consonants are V, Y, and V. At this point, however, when dealing with disyllabic forms, such as BI sudah, it is crucial to unravel the 'pecking order' that dictates which initial consonant counts as to voicing status in the selection of the High vs. Low tone set in (3) above. That is, in sudah does the s-or the d-take precedence? Clearly, the voiced initial d-had prevailed, and the V tone has been appropriately selected, and sudah > da. To clarify this consonant precedence, it is perhaps worth the digression to discuss the discoveries of Friberg and Hor (1977) as to initial consonant 'dominance' with respect to the register ( + ATR vs. -ATR) selection of stressed (main) syllables in Western Cham (Austronesian of southern Vietnam and Cambodia). Distinguishing register A (-ATR) consonants from register B ( + ATR), Friberg and Hor (1977: 36) summarise how precedence is established and register effects determined on the phonation of vowel quality of the succeeding syllable: (4) A (-ATR) B(+ATR) 1. p t C k b d J g ph th eh kh bh dh jh gh ?b ?d ?' J ? 2. s h m n fi I) w y r Register A and B in Western Cham words were accounted for by the following 'dominance' rules: (5)a. A+A=A b. B+ B=B c. A 1 + B 1 = second element d. BI+ A2=B e. A+B2=A These rules (following Purtle (1969) for Khmer), although neither ordered with respect to each other nor within the left-hand members, indicate register of tonic syllable. Quadrant A 1 and BI are equally strong; they both dominate A2 and B2. A2 also dominates B2. This analysis is based entirely on the consonant 'strength'; the consonant is seen to 'dominate' or determine the register characteristics of the following vowel. And, based on the five 'dominance' rules noted above, certain atonic syllable initial *=Advanced Tongue Register (Ed.). 83

PAGE 92

KENNETH GREGERSON consonants exert their strength over an intervening tonic syllable initial to determine the 'registerness' of the tonic syllable vowel. In the following examples, consonant dominance is seen in the combining of various syllables to form words (a grave accent / '; is added to indicate second ( + ATR) register clearly): (6) Al+ B 1 /ka/ + /baw/ /kabaw/ 'buffalo' B 1 +AI no examples (historically B 1 has become A 1 ). B 1 + A2 /bal/ +/haw/ /bahaw / 'new' A2 + B1 /ha/+ /dam/ /hadom/ 'how much' Al+ B2 /ka/ + /ro/ /karo/ 'strong' B2 + A 1 /la/+ /kaw / /lakaw / 'to step over' A2 + B2 /ha/+ /nin/ /hanin/ 'bow' B2 + A2 /la/+ /say/ /lasay/ 'cooked rice' Returning now to sudah > da, one may usefully compare this process with Friberg and Hor's A2 + B 1, in which /ha/+ / /hadom/, the word initial spirant yielding to the main syllable initial stop as the prosody determining element. Then, of course, Vietnamese ultimately reduces the form to a monosyllable. Resuming our discussion of the other aspectuals, it will be observed that the pair VN sZlp 'about to': BI siap 'ready' are also preverbal forms. The tone of sZlp results straightforwardly from its syllable type as eve (cf. 1.1.(3) above). For VN dang 'in process': BI sedang 'in process' one must, as with da, assume a simplification which ultimately drops the first syllable. As to tone assignment, one expects either level (unmarked) or with a evN syllable (no. (3) above), but with Friberg and Hor's rules an initial sshould have yielded to d-and resulted in *dang. Perhaps the situation is more complex historically and Friberg and Hor-type dominance is further conditioned by other factors. A prime candidate is perhaps an original stress difference in Austronesian. Specifically, for example, note: (7) BI 'sudah 'already' BI se'dang 'in process' in which VN da derives from an unstressed syllable while dang corresponds to a stressed one. The pair VN mai 'continuative': BI masih 'still, yet' do not function quite like the other aspectuals above, for mai occurs as a post-verbal while masih is a pre-verbal. Phonetically, one can assume a reduction masih > maih (i.e. eVH), after which, given the voicing of the initial, the resultant mai is completely expected (qv. (3) above). 1.2 Desiderative The following sentences are instances of desiderative modality in VN or BI: 84

PAGE 93

On Austronesian lexicon in Vietnamese (8)a. VN: em nay muon ve nha. younger sibling this want return home 'He/she (young one) wants to go home.' b. BI: Dia mau datang sore mz. 3-sg. want return home this 'She wants to come this evening.' That is, VN mu8n 'want' and BI mau 'want' both function as conventional preverbal desiderative forms. The question is 'Do they have any historical connection?' The similarity of form is strengthened if one assumes that mu8n derives from the nominalised (ke ... an) form ke-mau-an 'a wish'. If we again invoke consonant dominance (qv. 1.1 (5) and (6) pbove), the initial kwould determine one of the three High tones (V, V, V), but why V (mu8n), when CVN would seem to predict V (*muon)? Again, the hypothesis of a nominalisation source may provide an answer, for suffixes in Indonesian (cf. also Philippine languages) regularly insert a(?) between vowel sequences, thus (BI): ke-mau-'ian 'wish (n.)' which would provide the explanatory feature of stop in the final (-V?N) to produce an expected mu8n 'want'. 1.3 Equative Consider the following sentences in VN and BI: (9)a. VN: Anh toi la giao sw. b. B.I: older brother my 'is' teacher 'my older brother is a teacher.' Bahasa Indonesia ialah bahasa language Indonesia 'is' language 'Indonesian is the national language.' kebangsaan national Structurally, VN la and BI ialah (lit. ia '3rd sg.' + /ah 'emphatic') operate in remarkably similar ways. Assuming the reduction to one syllable /ah, the problem for tone would be that CVR predicts a form *la rather than Ja. The indication then, is that the Austronesian form, if the connection is authentic, must itself have been reduced to la at the time it was given a tonal interpretation in Vietnamese. This account of la is problematical, however, for Nguy~n Dinh Hoa has pointed out (pers. commun.) that in the old Chi, Norn data this copular Ja turns up as lam 'to do', and his suggestion is that that is, indeed, its origin. Since, on the other hand, it is acceptable even in Modern Vietnamese to use lam as well as Ja in a copular sense, it is not clear to me that they could not be independent forms. 85

PAGE 94

KENNETH GREGERSON 1.4 Pronouns There appear in Vietnamese a number of pro-forms of a locative, personal or interrogative nature that bear a good deal of resemblance to Austronesian5 forms, among which are the following:6 (10) VN kia; kza 'there': UAN *ija 'he, she, it', BI ia, Chmr. gwid'a, SAt. hia' VN nay, nay: PAN *iniH2 BI ini 'this' sini 'here', Rade nei (cf. similar forms in Tai). VN no [arrogant] '3-sg.', chung no 'they'; PAN *na '3-sg.', Agta na '3-sg.', BI sana 'there'. VN ta '[arrogant] I', chung ta 'we incl.': PAN *(k)ita, *ta 'we incl.', BI kita. VN ma 'which' (rel. pron.), mo 'what, where?': BI mana 'where, which' (interrog. pron.). 1.5 Adverbs (11) VN tau 'long time', luon 'forever': UAN *laun 'duration', BI faun 'to linger, loiter', Pw. lauf 'unfinished portion'. VN rfit 'very': UAN *b;;,yat 'heavy', BI berat 'heavy' (cf. VN hO'i lit. 'vapour'= 'rather'). VN liim 'very': Haroi hlam 'very', cf. BI selama 'as long as', selama lamanya 'at the most'. VN xa 'far': PMP *za[h]ouq, BI jauh, Chmr. t'ag'o. 2. Content forms Vietnamese has, in addition to the more 'grammatical' forms above, quite a large number of general lexical items that also bear enough resemblance to Austronesian forms to have been noted by a number of investigators. 5. Austronesian citations are in general from Dahl (1977). Rade forms are from Egerod (1978), other Chamic citations are from Burnham (1976). Waic references are from Diffloth (I 980). 6. Abbreviations for languages cited without full text references are: Am. Ami Mlg. Malagasy Rg. Rengao Bs. Bisayan MVN Middle (17th-centRu. Rukai BsE Bisayan as recorded ury) Vietnamese SAt. Squliq Atayal in Encarnacion NgD Ngaju Dayak Sir. Siraya Bu. Bunun PAA Proto-Austro-Asiatic SLi. Southern Li Chmr.Chamorro PAN Proto-Austronesian, TB Toba-Batak CLi. Central Li Dahl's constructions Tg. Tagalog Jav. Javanese PMP Proto-Malayo-PolyTh. Thao Kv. Kuvalan nesian UAN Uraustronesisch, Lq. Laqua Pu. Puyama DempwollT's conMak. Makassarese Pw. Paiwan structions Md. Madurese Pz. Pazeh 86

PAGE 95

On Austronesian lexicon in Vietnamese 2.1. Body parts (12) VN vai (MVN cbai) 'shoulder': *baYa 'shoulder', BI bahu, Pz. 'a baxa, Truk. jafar (Dyer 1953). VN /U'5'i 'tongue': UAN *dilah 'tongue', Jav. di/ah, BI lidah, Pw. lirjalirj, Rade /ah, Haroi caliah, Cham ta/ah 'id'. VN trai 'left hand': PAN *uiYi 'left (hand)', BI kiri, Mak. ka-iri, Pu. tama-wiri'. VN tai 'ear': PAN *t2alina, BI telinga 'ear', cf. CLi. thai. VN dflu 'head': PMP *qulu 'head', Tg. ?ufo, BI hulu 'upper end, head', cf. also Lq. ru, SLi. dau 'head'. 2.2. Humans and body functions (13) VN mu 'blind': UAN *buta, BI buta 'blind', Pw. ma-vutsa, Rade buum ala? cf. Thai *'bot. VN to, tro 'show': UAN *tu(n)duh, Jav. tuduh, BI tunjuk 'point', Tg. turo' 'instruction', Bs. tolo 'finger', Am. to ro' 'point'. VN ia 'defecate': UAN *ia(h) 'urine', Jav. p-ih 'water conduit', Bu. 'isah 'urine'. VN an 'eat': PMP *ka'an 'eat', Tg. kaa'in, BI makan, SAt. qan-iq. cf Rg. kaq 'eat (meat)'. VN g7:Lc 'bend head down': BI angguk 'nod' cf. BI anggut 'nod, (ship) pitch'. VN ngo 'take a look': BI anggul 'tip toward, raise head', Rade angii 'look up'. VN mfea 'vomit': UAN u(n)tah 'vomit' BI muntah; SAt. m-utaq. Note that, here, VN retains the Austronesian verbal prefix m-. VN gcii 'scratch': UAN *gatal 'itch', Jav. gatal, BI garit, Md. ghatal, Pw. gatsal, cf. also BI kais 'scrape for food'. VN nghe 'hear': UAN *daIJa 'hear', BI dengar, Tg. dil]ig, Pu. ma rngai, Rade kna 'ear' ( < *taliIJa). VN dr;,, (MVN deg,) 'belly': Uj\.N *[t]ijan 'belly' BI tian 'abdomen of pregnant woman', Am. tiaf, Th. tiiya. VN dung 'stand': UAN *<;iiYi 'stand', Rade dook dalJ 'stand' (cf. Li tsuon, Thai *'yiiiin). VN (n&m) me 'dream': UAN *i(m)pi, BI mimpi, Pw. sapi, Rade ept:i 'dream'. VN ba 'grandmother': UAN baji 'wife, woman', Tg. ba-bai'i Am. va-vahi', BsE bayi 'grandmother'. cf. also VN VQ' 'wife'. 87

PAGE 96

KENNETH GREGERSON VN thU'a 'respect form': UAN *tuha/tuva 'old' Jav. tuwa. 2.3. Flora, fauna VN cay 'tree': UAN *kaju, PMP *kaS2iu 'tree, wood', BI kayu, Tg. kaahoy, Am. kasui. VN bong 'flower': BI bunga 'flower', Rade mlJa. cf. PMunda *ba[g]a, Mon pkao, Aslian *bakaw, PAA *baka[l]. VN mg 'rice seedling': PMP *qumaH 'field', BI huma 'field for dry rice', Bu. humaq, SAt. qumah 'work in the field'. VN tning (MVN thing): *VAN *talmi' 'egg', Tg. 'itlog, BI telur, Am. lita'uy. VN swa 'milk': UAN *iuiu 'female breast', BI susu 'milk, breast', Am. tsotso', Rade ksau 'breast'. VN cimg 'claw, pincer': UAN *baYan 'molar', Tg. bagan 'molar' Jav. wan 'jaw', Rade kaalJ 'chin, jaw'. 2.4. World, weather VN bl:li 'dust': UAN *<;labuk 'ashes, dust, grey', BI abu 'dust', Jav. dawu 'grey', ?bruih 'dust'. But cf. Rg. hapuih 'to dust', Waic. *pes. VN d8-t 'earth': UAN *datay 'flat', BI rata 'flat, level', Tg. lataq 'carpet', NgD datah 'step, rung'. VN 10' (of cliff, well) 'collapse, slide': UAN *<;labt 'leave suddenly', Tg. lilis 'wipe off, TB dolos 'glide', Jav. rj.alas 'keep away', Rade /uh 'fall to the ground'. cf. Rg. ralayh 'cave in'. VN trang/giang (MVN blang) 'moon': UAN *bulan 'moon', BI bulan, Rade mlaan, Jarai blan, Pu. vofan, Kv. buu ran. VN sang 'become bright': BI terang 'clear, bright', cf. VN tr!ing 'white', rljng 'dawn'. VN dem 'night': UAN *<;lam <;lam 'keep quiet', BI diam 'quiet', Pw. dzam dzam 'last night before full moon'. VN mai 'tomorrow': UAN *damay 'resin, torch', Sir. madama 'morning', Ru. damar 'moon', Tg. damag 'night' (cf. with dem above). Cf. Rg. mar 'morning' mar eh 'tomorrow', also Rg. mang 'night'. VN dU'
PAGE 97

On Austronesian lexicon in Vietnamese 2.5 Miscellaneous verbs VN rai, trai 'to sow': BI beras 'rice', Rade rah 'sow rice in wet field'. VN tra, gia (MVN bla) 'pay back': UAN *b;ilah 'split', BI be/ah 'split, part', Pw. valaq 'split'. Cf. also VN bi}a 'split open' (except tone is expected). VN. Mt 'fasten together': UAN *d;ibt 'to stick, BI dekat 'near', Tg. dikit 'joined'. Pw. dakats 'to stick'. VN t&i 'arrive': UAN *ha(n)t;i<;l 'deliver, convey', BI antar 'introduce' Jav., Md. atar, Pw. sa'iadz 'send'. VN d8i 'deceive, lie': UAN *put;id 'rotate', BI putar 'turn, be dishonest'. cf. Rg. podiir 'spin a top, deceive'. VN nfiu 'cook': UAN *tunu 'roast', Jav. tunu 'burn, Md. tono(h) 'roast', Pw. ma-tsufu 'hot', Rade m?dau, Roglai pa?dau 'warm'. VN mfit 'lose': UAN *mataj, BI mati 'die, dead'. VN kim 'seek': UAN *ki[l]im 'send', BI kirim 'id.', Am., Bu., kilim, Pw. kim 'search for'. 2.6. Descriptives VN sai 'wrong': PMP *t'alaq, BI salah, Tg. sala 'mistaken', Pw. pa talaq 'envy, jealous'. VN sac 'sharp': PMP *hat'aq 'whet', Chmr. gwasa', Tg. haasa, Pw. t-ataq 'id'. VN blf' 'big': UAN *b;iyat 'heavy', BI berat 'id.', Mlg. be 'big, great, many'. 2.7. Miscellaneous VN cu8i 'end, least': UAN *liku<;l 'back, behind', BI ekor 'tail', Tg. likod, NgD ba-rikor, Pw. likudz 'behind'. VN sung 'gun': UAN laielJ, Tg. lusolJ, Rade SUIJ, Rg. iisuk 'mortar', Pz. ludzu(J, SA t. luhulJ. VN van 'plank, board': UAN *papan, BI papan 'board', cf. Li. pen 'classifier for people', Thai *peen 'plank'. VN na 'bow': UAN *panah, PMP *panaq 'bow, arrow, shoot', BI panah 'bow and arrow', Tg. paana, Rade hna, Am. pana'h. Cf. VN ban 'shoot', and BI panar 'stunned, dull', but also BI senapan 'weapon'. 89

PAGE 98

KENNETH GREGERSON 3. Some Phonetic Patterns The Vietnamese forms cited above in sections I and 3, while presenting a number of 'irregularities', do, on the other hand, exhibit a number of likely phonetic associations with Austronesian forms in general, now summarised in 3.1-10. 3. I. VH high tones: AN voiceless initials VN Mt 'fasten together': UAN *dabt 'to stick' BI dekat 'near', Tg. dikit 'joined', Pw. dakats 'to stick'. VN trwng ( MVN tlwng) 'egg': BI telur, Tg. itlog VN ta '[arrogant] I', chung, ta 'we incl.': PAN *(k)ita, *ta, BI kita 'we inclusive'. VN cay 'tree': BI kayu 'tree'. VN to, tro 'show': Jav. tuduh 'point', Tg. turo' 'instruction'. 3.2. VN low tones: AN voiced initials VN bil 'grandmother': UAN *baji 'wife, woman', Tg. ba-bai'i, Am. va-vahi, BsE bayi 'grandmother' (cf. VN v~ 'wife'). VN mu 'blind': UAN *buta, BI buta, Rade buum a/a? 'blind', cf. Thai *'bot. VN mg 'rice seedling': BI huma 'field for dry rice', Bu. humaq, SAt. qumah 'work in the field'. VN g7
PAGE 99

On Austronesian lexicon in Vietnamese 3.5. VN or tone: AN final stop VN dfit 'earth': UAN *data 'flat', BI rata 'flat, level', Tg. lataq 'carpet', NgD datah 'step, rung'. VN sfic 'sharp': PMP *hat'aq 'whet', Chmr. gwasa', Tg. haasa', Pw. t-ataq 'id'. VN b;d 'dust': UAN *dabuk 'ashes, dust, grey', BI abu 'dust', Jav. dawu' 'grey'. cf. PMP *abuh, Rade ?bruih, Rg. hapuih 'to dust', Waic. *pes. VN g7:lc 'bend head down': BI angguk 'nod' (cf. 3.2.). 3.6. VN level (unmarked) or 'tone: AN open syllable/final nasal VN tai 'ear': BJ telinga. cf. CLi. thai 'ear'. VN ta '[arrogant] I', chung ta 'we incl.': PAN *(k)ita, *ta, BI kita 'we incl.'. VN an 'eat': PMP *ka';m 'eat', Tg. kaa'in, BI makan, SAt. qan-iq 'eat'. VN ma 'which (re!. pn.)': BI mana 'where, which (interrog. pn.)', cf. also VN mo 'what, where'. VN dieiJ,ng, dang 'road': PMP *Zalan 'road, path', BI jalan, Tg. daan, Sir. darang. 3.7. VN eh-: AN t/pl-VN chl 'thread': PAN *taliS' 'rope, cord', BI tali, Tg. taafi?, Pw. tsalis, Pz. sariss 'cord'. VN chl:lc 'a collection of ten': UAN/PMP *puluh/puluq Tg. pulo', pu'o', Ru. porok, Pw. puluq 'ten'. 3.8. VN U' or O' : AN -1/r-VN lie5'i 'tongue': UAN *dilah 'tongue', Jav. di/ah, BI lidah, Pw. licj.ali<;l, Rade !ah. VN trwng (MVN tlwng): UAN *tdluY 'egg', Tg. 'itlog, BI telur, Am. lita'uy 'egg'. VN dzdmg, dcmg 'road': PMP *Zalan 'road, path', BI jalan, Tg. daan, Sir. darang, Rade elaan. VN bi!' 'big': UAN *bdYat, BI berat 'heavy', Mlg. be, 'big, great, many'. 91

PAGE 100

KENNETH GREGERSON 3.9. VN -i: AN -1/r VN cu8i 'end, last': VAN *liku~ 'behind, back', BI ekor, Tg. likod, NgD ba-rikor, Pw. likudz 'behind'. VN sai 'wrong': PMP *t'alaq, BI salah, Tg. sala', 'mistaken', Pw. pa-talaq 'envy. jealous'. VN t6'i 'arrive': VAN *ha(n)t;:i~ 'deliver, convey', BI antar 'introduce', Jav., Md. atar, Pw. satadz 'send'. VN d8i 'deceive, lie': VAN *put;:i<;l 'rotate', BI putar 'turn, be dishonest'. VN tai 'ear': PAN *taliIJa, BI telinga 'ear'. cf. CLi. thai. 3.10. VN tr-/gi-/bl-:AN b(v)l-VN trang/giang (MVN blang) 'moon': VAN *bulan 'moon', BI bulan, Rade mlaan, Pu. volan, Kv. buu ran. VN tra/gia (bla) 'pay back': VAN *b;:ilah 'split', BI be/ah 'split, part', Pw. valaq 'split'. Cf. VN bfa 'split open'. VN tr?m 'be round': VAN *b;:ilUY. cf. Li. (p)luon, Thai *'dual), 'don (Benedict 1966: 246). 4. Concluding remarks This brief consideration of possible lexical affinities between Vietnamese and Austronesian makes no claim to far-reaching conclusions. At the same time, some of the VN forms observed above would, if valid, seem to call for at least two sources or periods of Austronesian ( or Austro-Thai?) contact in order to explain their contemporary phonological constitution. These I will, for present purposes, distinguish simply as Immediate vs. Remote sources. The following sets of vocabulary are illustrative: Immediate VN dii 'already': BI sudah 'already'. VN slip 'about to': BI siap 'ready'. VN dang 'in process': BI sedang 'in process'. VN Iii 'is': BI ia !ah 'is'. VN thwa 'respect': BI, Jav. tuwa 'old'. VN cay 'tree': BI kayu 'tree'. VN rai 'to sow': BI beras, Rade rah 'sow rice in wet field'. Remote VN giii 'scratch': VAN *gatal 'itch', BI garit, Pw. gatsal. cf. BI kais 'scrape for food'. 92

PAGE 101

On Austronesian lexicon in Vietnamese VN mq, 'rice seedling': PMP *qumaH 'field', BI huma 'field for dry rice', Bu humaq, SAt. qumah 'work in the field'. VN kiJm 'seek': UAN *ki[l]im 'send', BI kirim 'id', Bu. kilim, Pw. kim 'search for'. VN sfic 'sharp': PMP *hat'aq 'whet', Chmr. qwasa', Tg. haasa', Pw. t-ataq 'id'. VN kJt 'fasten together': UAN *d:ibt 'to stick', BI dekat 'near', Tg. dikit 'joined', Pw. dakats 'to stick'. VN dwO'ng diing 'road': PMP *Zalan 'road, path', BI jalan, Tg. daan, Sir. darang, Rade elaan. Thus, some of the forms cited in this paper may well reflect contact between Austronesians or Austro-Thai and Austroasiatic, rather than with Vi~t-MU'&ng proper. Even so, it is suggested that, for example, syllable reduction and tonal effects in the latter contribute significant perspective to historical processes even in early linguistic relationships. REFERENCES Benedict, P.K. Burnham, E.C. Dahl, 0. Chr. Dempwolff, 0. Difnoth, G. Dyen, I. Echols, J.M. & Shadily, H. Egerod, S. Encarnacion, Juan Felis de la Friberg, T. & Kvoeu Hor. Haudricourt, A.G. Maspero, H. Nguy~n Dinh-Hoa 1976. Austro-Thai and Austroasiatic. In Austroasiatic Stud. 1, (Oceanic Ling. Spee. Puhl. 13) (eds.) P.N. Jenner, et al. Honolulu: Univ. Hawaii Press, 1-36. 1976. The place or Haroi in the Chamic languages. Unpubl. M.A. Thesis, Univ. Texas, Arlington. 1977. Proto-Austronesian. (Scandinavian Inst. Asian Stud. Mono graph Ser. 15). Lund: Studentlitteratur. 1934-38. Vergleichende Lautlehre des austronesischen Wortschatzes. 3 vols. Beiheft zur ZJES 15, 17, 19. 1980. The Wa languages. (=Ling. Tibeto-Burman Area 5(2)). 1953. Dempwolrrs R. Language, 29, 359-66. 1974. An Indonesian-English dictionary. Ithaca: Cornell Univ. Press. 1978. An English-Rade Vocabulary. Bull. Mus. Far East. Antiq. 50, 49-104. 1851. Diccionario Bisaya-Espaiiol. Manila: Imp. Amigos del Pais 1977. Register in Western Cham phonology. In Papers in Southeast Asian Ling. 4. Chamic Studies (Pacific Ling. 48) (eds.) D. Thomas, et al. Canberra: Austral. Nat. Univ., 17-38. 1954. De l'origine des tons en vietnamien. J. Asiat., 242, 69-82. 1912. Etudes sur la la phonetique historique de la langue annamite: I: Les initiales. Bull. Ee.fr. Extr.-Orient. 12, 1-127. 1971. Vietnamese-English student dictionary. Carbondale: South Illinois Univ. Press. 93

PAGE 102

KENNETH GREGERSON Purtle, D. Sauvignet, E. Schmidt, W. WullT, K. 1969. Some speculations of the genetic relationships of Sino-Tibetan to the languages of Southeast Asia. Paper read at LSA annual meeting, December 1969, San Francisco. 1922. Les origines de la /angue annamite. Hanoi-Haiphong. 1926. Die Sprachfami/ien und Sprachenkreise der Erde. Heidelberg: Winter. 1942. Uber das Verhaltnis des MalayoPolynesischen zum Indochinesischen. Danske Videnskap Sepskab (His.-fil.-Medd.) 27(2), 1-157. 94

PAGE 103

SEVENTEENTH-CENTURY VIETNAMESE LEXICON: PRELIMINARY GLEANINGS FROM ALEXANDRE DE RHODES' WRITINGS Nguy~n Dinh-Hoa I. Introduction The phonological system of Middle (i.e. seventeenth-century) Vietnamese I has been treated in Gregerson (1969), Haudricourt (1974), and others. This paper takes a look at the Vietnamese lexicon of the same period, as shown in the trilingual (Vietnamese-Portuguese-Latin) dictionary and in the Catechism authored by the Jesuit scholar missionary Alexandre de Rhodes (1593-1660). Only 'full words' (th'c tw) or content words in their archaic forms will be discussed, leaving out function words, known by traditional Sino-Vietnamese grammarians as 'empty words' (hw-tw). The Dictionarium Annnamiticum [sic] Lusitanum et Latinum2 (Rhodes 1651a, hereafter Diet.) consists of 900 numbered columns, two to each page, listing Vietnamese entry words in alphabetical order according to the spelling of the time, each headword and illustrative example followed by its Portuguese equivalent, then by its Latin equivalent. Between the 'b' and 'c' sections, there is a section (cols. 65-74) devoted to words whose initial consonant is transcribed with the symbol dJ; this lenis obstruent was described by de Rhodes (1651 b: 3) as 'almost like Greek f3 as in clJeao "enter", dJ ai ea "fin [offish]" '. This feature is explained within a 31-page statement on Vietnamese grammar bound in the back of the dictionary, following an Appendix, which contains five unnumbered errata pages, and a 171-page index of Latin words: this Linguae Annamiticae seu Tunchinensis Brevis Declarato (hereafter BD) is composed of eight headings. On the other hand, the Catechismus pro iis qui volunt suscipere Baptismum in octo dies divisus, Phep giimg tarn ngay choke mufm chju phep rwa tqi ma efJeao dg,o thanh fJwc Chua Bloi (Rhodes 1651c; hereafter Cat,), constitutes 'the first work appearing in romanized Vietnamese' (tac-pham qu8c-ngi, dflu tien), as the work is referred to in the title Giao-si Dlic-lq va tac-phiim qu8c-ngi,dflu tien, (eds.) Nguy~n Kh~c Xuyen and Ph~m Dinh Khiem (1961 ), a book issued on the 300th anniversary of the death of the missionary 'the evangelical apostle who codified the qu8c-ngi, script, I. Maspero (1912) distinguished five periods in the history of the Vietnamese language: proto-annamite (before and during Chinese rule), preannamite (during independence), annamite ancien (fifteenth century), annamite moyen (seventeenth century), and annamite moderne. 2. For a general description of this trilingual dictionary, see Nguy~n Dinh-Hoa 1985b. 95

PAGE 104

NGUYEN DINH-HOA and the reverend benefactor of the Vietnamese church and the Vietnamese nation.' Since the compilation Diet. systematically listed the seventeenthcentury lexicon, I have first tried to cull out archaic forms used in the Catechismus, then to check their listings in the trilingual dictionary. I also present other interesting items found in Diet., but not in Cat. In this preliminary gleaning process, of necessity not exhaustive, I shall successively discuss nouns, classifiers, adjectives, adverbs, preverbs, verbs, and postverbs. 2. Nouns T9e Diet. !ists several 'taboo words' denoting sexual organs, ~.g. brn, dij.u, doc, doi, Ion, ke (female), boi, c(,.c, dai, /8 (male). The entry dech is glossed as 'semen humanum': cf. 'secretion from female genitalia' and 'female genitalia' (Van Tan 1977, under dach). (an)nan t9i (Diet. 7) 'to eat the bitter grass [called nan (Diet. 504)] 'to show remorse for one's sin'. Cat. 177h. bai, con bai dI bai (Diet. 19) 'prostitute, lustful woman'. bqm, con bqm (53)* 'prostitute', cf. modern VN di bqm. Both Le Van E>ll'c (1970) and Viin Tan (1977) have the compound bqm bai 'deceitful person' (cf. bai, binh bai (also Diet. 19, but separate entry) 'deceitful person'). b~u (30) 'companion, partner'. Cat. 234. Le Viin E>U'c et al. (1970) has both b~u 'you [to wife or girl friend]' and b~u-b~ 'friends' whereas Van Tan (1977) has only biu b~ 'same as be b~, friends'. cai (79) 'head, commander' Cat. 71t, u, cf. modern VN 'foreman'. cang la (85), same as giang la (277) 'wicker basket with handle', cross listed under la (389). This compound can be found in 'Thap-gi6'i c6-h6n qu6c-ngfr viin' ('Ten commandments to lonely souls'), in Thien-nam Dll' hi;t T~p, a collection of poems by Emperor Le Thanh-tong and his co members in Tao-diin Circle (fifteenth century): Song viJt /iJn tay; cang la, non anh. Hom mai h9p m(,.t: nf}i co, vzdm /au. (Cao Hfru L~ng 1983: 12) cat nhan (89) 'incense'. Cat. 165b-c. ~t (90) 'back'. Cat. 185d. Diet. also gives slip cij.t cung ai and blai elj.t (cf. modern VN chung hmg dfiu cij.t and trai cij.t). chS (101) 'pretty large earthenware vessel'. Cat. 18lf-g. chfun cu (95) 'punctuation'. This expression is found in Poem 3, line 4, by Nguy~n Trai (1380-1442): PhiJn sach ngay xucin ng8i chfim cu, although the word cu (136) for 'sentence' is cciu in modern VN. dinh li~u (415, under li~u) 'huge bamboo torch'. doi t6i (228) 'chain, shackle'. Cat. I Of. Also spelled loi t6i, with t6i itself listed separately (822) 'chain'. c6m (130) tlang (804) 'stock made of bamboo or wood to confine a *Unless necessary for clarity, Diet. is dropped before page number to the entry (Ed.). 96

PAGE 105

Seventeenth-century Vietnamese lexicon prisoner's neck or ankles'. Cat. l0f. Cf. modern VN cum and triing, respectively, with the second member of the compound occurring in the expression gong dong trong mang to denote the status of a prisoner restrained by a cangue (gong, Diet. 299) around the neck and by stocks around the ankles. diio (206), i.e. dong 'spear'. Cat. 232q, u. he (318) 'mourning rites lasting three years'. Cf. modern hiji he 'festivals, merry-making occasions'. khoc, (374) 'period, time' < Chinese *k'uii if~, which explains why the variant forms thoc, (776) and thiic, (784), nowadays spelled th{la/thuO', are transcribed in the nom script by the character ~lCat. 150i, 176d, 195h, 196s-t, I 97b-c. khu (376) 'rear end, buttocks'. la da (191 under da and 390 under la) 'stone slab' Cat. 2800,p,r. This archaic form is found for instance in Poem 21, line I by Nguin Trai: Dau ngU'oi di Iii da mon. 3 lfun (396) 'mud' Cat. 195n, 196x. Cf. chan tarn tay bun. mill. tui (489) 'coriander' seeds from the H6-tuy plant (Le Van D&c 1970: 939)-nS in to nS 'ancestors' Cat. 81 x. Cf. t6 toii (817) and both compounds appearing as subentries under toii, i.e. tong (827). pheo (598) 'bamboo': the second member of the synonym compound tre pheo retains this meaning in MU'<'mg. l\lC (632) 'prison, jail'. Cat. tu rq,c l0z, IOe. Diet. also gives the phrase rq,c dja ng{lc 'hell'. tap (725) 'tempest, hurricane', nowadays occurs only in such compounds as biio tap. tang_ (721) 'mourning' provides cultural meaning of the expression di tang dJ toe [taoc] (the bereaved man letting his hair grow on top and in front and the bereaved woman cutting part of her hair), which confirms that tang toe is another synonym compound and not a reduplication. thS (754) 'world'. Cat. Sb, 6i: thJ nay 'this world'. thoi (774) 'short moment'. Cat. 241m. thoa (773) 'brothel; prostitute', with con thoa meaning 'prostitute, harlot' (cf. the adjective di thoci in modern VN). toi (822) 'servant, slave' with the compound toi ta 'servants' Cat. 6t,x,y (cf. modern VN t8i to'). 3. Determiners, Classifiers and Demonstratives Two items are used to denote 'all, the whole': ea and thay thay. The former occurs with vii: ea vii nhii 'the whole house', ea vii thien-~ 'the whole universe' (77), cha ea vii loiii ngll'O'i ta 'for our whole mankind' ( Cat. 88), ea vii nU'O'C 'the whole country' (Cat. 21c). Diet. also lists ea hoa as the equivalent of ea vii under the entry hoa (229). The latter word can be found in Poems 80, 90, 93 and 247 by Nguyen Trai, and it is tempting to posit the following development: hoa/hwa/ > uii/wa/ > vii/va/. One can 3. Theurel (1877: 220) lists la dii 'hail' and mU'a la dii 'it is hailing'. 97

PAGE 106

NGUYEN f>INH-HOA also note va hai 'both' and the use of va as correlative conjunction in va banh va ea 'both bread and fish' (Cat. 184n). The determiner thay truly is found with hh (734) although under hSt (322) the spelling hJt thay thai is presented. In addition to pluralisers cac (78), chiing (121-2), nhil'Dg (559), Diet. lists me, (475), as in m& t8i (475),4 also discussed in BD 12, and pho (601), as in ph8 8ng ph8 ba 'ladies and gentlemen', ph8 thay 'masters' and nhung ph8 8ng 'gentlemen' (BD 12, 19). This last example is also listed under nhih1g2,that is glossed 'only', so nhu-ng ph8 8ng means 'only gentlemen' with chang nhu-ng meaning 'not only' whereas nhil'Dg1 means 'all', and also 'nothing but' as in nhu-ng mong 'only hoping', d8y nhung mu8i 'full of mosquitoes', nhu-ng mu8i la mu8i 'nothing but mosquitoes' in modern VN. Concerning nhung, the sense 'as many as' (e.g. nhu-ng tarn dwa con 'as many as eight kids') seems to have been non-existent in middle Vietnamese, and a dictionary published in the late nineteenth century, (Huinh-Tjnh Cua 1896: 143) does not list this sense, either, under the entry nhii'ng. On the other hand, for the second-Qerson plural pronoun, the two arrogant expressions m{j, bay and m{j, may are noted together with chung may, chung bay and bay 'you guys' (BD 12). The pluraliser :ho (Cat. 197 and passim) had been used in the fifteenih century by Nguyen Trai, transcribed by the n8m character 4lJ (Nguyen Dinh-Hoa 1985a: 471); it was still listed in Theurel (1877: ix, 363), and in Huinh-Tjnh Cua (1896: 200) with the character "t:1f Under the rubric 'classifiers', it is worth mentioning that cai, nowadays used for inanimate, non-living things, appears in the Diet. with such nouns as c6c 'toad', d?n 'spider', &h 'frog', kiSn 'ant' (128,167,249,380, respectively), etc. 5 Beside the demonstratives nfiy, n9, fiy, kia, kia, the Diet. also haste as a synonym of kia 'that... over there': 8ng te (728; BD 20). The latter is still used in Central Vietnamese. nao (507) is shown as occurring following the noun it modifies: sao nao?, thJ nao?, each nao?, dU'ong/nhzdmg nao? (BD 21) to denote 'which ... ?' and as occurring in such idiomatic expressions as nao co ai biJt? 'who knows?', nao co t(ji gl? 'which sin? which offence? which crime?' with the connotation of negativity (508; BD 22). But there is also another usage, in which nao precedes a noun: nao sach? 'where's the book?', niw thay? 'where's the teacher?' (508). This word order can be found in some of Nguy~n Trai's poems in the fifteenth century: nao nO'i 'which place?' (poem 47), nao hoa 'which flower(s)?' (poems 59 and 224), nao cua 'which thing?', nao thu
PAGE 107

Seventeenth-century Vietnamese lexicon 4. Adjectives There is a striking number of synonym compounds among the adjectives, in which one member somehow has lost its original meaning so that in modern VN the compound sounds like a reduplication. d9t, for example, meant 'ignorant, feeble-minded' (178), occurring in the compound ~-d9t 'imprudent, unwise' (155). hoi (334) was a synonym of h~ (321 ), and the compound meaning 'narrow [of mind]' is not an alliterative reduplication. However, the adjective chc:ri ooi 'to indulge in promiscuity' is merely a rhyming reduplication (52), in which b&i is not the homophonous entry b&i (also 52) meaning 'a lot' which Le Trung Hoa (1982: 30) interprets as a modifier of the verb.chO'i 'to play (around)'. kham (360) 'sufficient' with the example chZmg kham 'not enough' (cf. no below). khon (374) 'difficult' [and also 'wise']: this word was used by Nguy@n Trai in poems 1, 6, 7 and 65 (Nguy@n Binh-Hoa 1985a: 468), and also discussed in Theurel (1877: p. xxi) as a synonym of kh6. khong (spelled khou, Diet. 375) is glossed 'vacuus, -a, -um' as in examples an cO'm kh8ng 'to eat plain rice [without meat or vegetables]', tarn (vifc) kh8ng 'to work without pay', nhiI. kh8ng 'empty house', and, in the following entry, kh8ng goa 'widow': cf. nhwng xac kh8ng (Cat. 15x), nhU'ng kh8ng d8i (Cat. 237p.), next to the compound hU' kh8ng (Cat. 54h) which is glossed 'vacuum', i.e. 'nothing, nothingness, void, vacuum' in such collocations as hoa hU'-kh8ng, lgi ra hU'-kh8ng (Diet. 341 ). (On the meaning of khong as negative particle, see the section on verbs below). la ciao (400) or lao nhao 'disordered, chaotic' (Cat. 19y, 87u, etc.) l(!p (425) is given as a synonym ofkjp (384) 'opportune: (dJn cho kfp): with such illustrations as di ( cho) IQ'p 'to go in time', chang IQ'p 'not in time'. Cat. does not use IQ'p (it has only kfp), and Theuret (1877), which lists a number Of illustrative examples, -thz, sanh -, V! -, -chan, VU'a -, chU'a -, dii -, an nan chang -, for kfp (218), does not have the entry IQ'p. mll'a (487) 'more than sufficient, superabundant' (synonymous with dlF (179)) nowadays occurs only in the compound thwa mua (487) and in the expression bo mua 'to leave one's food unused': cf. Theurel (1877; add. 44) bo mua, mua mua 'leave work unfinished'. nat (509) has the sense 'rotten, spoiled, decadent' used in the moral sense in Cat. 70 (also hU' nat). nghl (526) 'easy' (chJng nghl chfu IO'i nb cam d8, nghl d8i dan ba dU'q;c in Cat. 83 n, r; nght chju dU'(!"'C tha vgy in Cat. 900 ); ciI.ng nghl tarn vifc ay in Cat. 118}'.; nghl tarn !en in Cat. 2510) has been found in Chl-nam Ng9c-am Giai-nghia, a dictionary of Chinese and Norn characters, compiled some time after the sixteenth century (Tr~n Xuan Ng9c-Lan 1982; Le Van Quan 1981). In the preface, its author, a monk by the name of Phap-tinh, said: Bay giO' Nam dgy chw dO'n, Cho ngU'O'i mO'i h9c nghl xem, nghl nhu[m. 'The nom script is taught in here as single characters, 'So as to make it easy for beginners to read and to learn them'. 99

PAGE 108

NGUYEN DINH-HOA nhtl> nham (556) 'deceitful, false', synonymous with d8i tra (d8i bla), as in 8ng Adam chfmg co phai lO'i nhi,e nham dfiu (Cat. 88-1). no (562) 'full; sufficient' has such run-on entries as no m9i nO'i, no m9i Slf', than15 no, di'i no mq.t, and also no du, no kham [see khLn above]. Cf. dJn khi 6, the nay di'i no (Cat. 8lx), ta phai kinh 6, no m9i nO'i (Cat. 51s), the latter structure no m9i ... being the equivalent of English 'every ... all ... with the meaning of this lexeme restricted in modern VN to the notion of 'fullness (from eating), satiety'. nc,t (571-2) 'premature', synonymous with sinh non, also has yJu nO'l 'feeble in body and mind'; the form non nM 'tender, immature' in modern VN, made up as a synonym compound, is often considered an alliterative reduplication. (ran may) rfm mq.t under Iin (636) 'obstinate, not amenable to control' (O' rfm mq.t cung ,no, ip Cat. 87r: cf Theuret 1877: 383). The equivalent in modern VN is ran dfiu or cung c8, and involves the head or neck, not the face. tay (716) 'equal' is cited as appearing freely (like its peer bimg) followed by a complement: tay ngU'O'i nen hai mU'O'i tu8i, 'equal in size to a 20-year old person'. Cf. modern VN idiomatic expressions tay amh and tay trO'i as modifiers of ti)i 'offence', or vq 'offence, fine, calamity'. WJ.Y (857) 'slanting, crooked' used with the noun dqo 'way, path, religion' just as in dqo ta 'false or erroneous precept' (712); Cat. 29m, 30p, where dqo vqy refers to Confucianism, Taoism, Buddhism and other local beliefs of the time. 5. Adverbs Only a few adverbs deserve mention: gia giSt (271) is a modifier like lfim, thay or rap 'very': BD 14 gives t8t lf'im, lanh thay, xfm rap, and ml&n, i.e. l&n, gia giJt ,as examples, where such an intensifier follows the modifier. Cf. l&n gia giet hU'n ea (Cat. 62y), dqp gia giJt d8.u may (Cat. 92e), logia giJt lam vgy cho cha m<; (Cat. 120n). Le Van DU'c (1970: 537) also gives the meaning 'severely, sternly' illustrated by the expression phe-blnh gia gih m9t (482) is listed and glossed only as a numeral. However, its use in the sense of 'only' is,foun~ in more than ~me source: mi)t lo Slf' thJ gian ma th8i (Cat. 951); chang thfiy IQ'i, mi)t thfiy hqi ma th8i 'non lucrum sed tantum detrimentum video' (Theur~l 1877: 284, under mi)t); as well as mi)t noi Slf' mlnh ma th8i (in Sach s8 sang chep cac vifc, a diary by an eighteenth-century priest, Philiphe B1nh (1968: 152)). rap (641) 'very' as in t8t rap 'very good' (and in xfiu rap, BD 14). Cf. nq.ng rap (Cat. 221s), die rap (Cat. 217m). se se (684), synonym of d8.n d8.n (268), as in to ra se se (256p), se se co chfu (268x). vc,i, often considered an adverb modifying xa 'far' (872-3) is glossed as a substantive meaning 'high seas' so that xa vO'i really means 'far off the coast, on the high seas'. 100

PAGE 109

Seventeenth-century Vietnamese lexicon 6. Prfverbs chang is used more often than khong (375) as the negative preverb, wherea~ d6ng_ is SY,nonymous with cung 'all, together' as in d8ng J.. m9t tang, dong nhdt the (236). kb.a 'can; appropriately, fittingly' (359) < kha (Cat. l ly-z has Day bin kha bu9c swng trau I Le thfj,t kha phlfC dU'Q'C long ngU'O'i ta, which is a translation of the Sino-ietnamese couplet Kien-thlmg kha ke ngU'u-giac I Ly-ngw nang phlfc nhan-tam). khll'ng (377) 'agree to, consent to, accept to'; chang khrmg vang phep cha mr; (Cat. 181). mlPa (487) is a prohibitive form 'don't, should not' found in fifteenth cent~ry utterances (Nguy~n Dinh-Hoa 1985a: 469); the entry mlf'a noi d8i 'don't lie, should not tell lies' is repeated in BD 24 as mva hJ noi d8i 'should never tell lies'. rinh (649) 'be about to, approach [a condition]' with three examples r)nh chJt, rlnh ngii, rlnh de, the first of which is found in Cat. 67, 69 with the meaning 'dying, about to die', also cited in Theuret (1877: 390) 'morti vicinus'. Cat. 98b also has rinh lift about the Deluge. 7. Verbs The subgroup comprising bi, chiu, dU'Q'C and phai deserves priority treatment simply because these suggest the passive voice of Western languages. The first one, bi (34), not to be confused with the preceding entry, which means 'bag' ('mantica'), is given only one illustration: bi phii [phong] ba 'run into a storm' although in modern Vietnamese (Clark 1974, Nguy~n Dinh-Hoa 1972) its [ -pleasant] or [ + adversative] feature tempts many an analyst to translate it as 'be ... get... ,' in English. The verb ch.iu, with the meaning 'undergo, suffer, experience' must have been used more frequently than bi. Diet. (109) lists chiu tqi, chiu chh, chiu !Bi, chju llfy, chju nQ', chju mifng, chiu thai. Phai, on the other hand, which denotes involvement in an unpleasant or unfortunate situation, is illustrated by no less than twelve examples, in which the 'predicament' can be as serious as thunder and lightning (phai sfim, set), paralysis (phai Ii? t), or as inadvertent as eating meat by mistake (mlam phai thjt), or as committal as falling in love (phai long). (590) The opposite of these three terms, the entry dlPQ'C (243) is often glossed as 'to gain, get, acquire, earn, find, obtain' with the feature [+pleasant]. At any rate, all four lexemes fully qualify as regular verbs. ho V{l (49) 'to slander, calumniate'; Cat. 65, 190. chac (93) 'to purchase', found earlier in Nguy~n Trai's poems. dai (154) 'to fear'; Cat. 59, 91, 221x. Cf. modern VN expression khan cho ngU'O'i ta dai, dai cho ngU'O'i ta thU'O'ng. diu (160) 'to love' with the examples yeu dau con and thu8c dau, the latter still found in bua yeu thu~c dau 'love charm, philtre, love potion'. dS (161) 'to despise' as in dJ ngU'O'i 'to be haughty'. 101

PAGE 110

NGUYEN DINH-HOA da,(191) 'to touch', with co-verb d~n or phai preceding the object: tqy da den cai sang 'His hand touching the coffin' (Cat. 187), Tay t8i da den du--Q'c ch[m ao dwc Chua Jesu 'As my hand touched the hem of Jesus' robe' ( Cat. 185). da (191) 'to heal': some examples in Cat. are ke dpu nq.ng th1 cho da 'curing persons who were seriously ill' (179r), song chang dii cho 'but they could not cure her' (l 85e), th1 t8i dii 'then I was cured' (! 85g), chwa dii tgt ta 'to c!lre our ailments' (186m). Cf. modern saying Thu8c dang dii tgt / Sif thgt mat long. dem [dang] (195) 'to guide, lead, show the way' as in [cai sao] soi cho ma dem diJng '[the star] guiding them and showing them the way' (Cat. 164u), [8ng thanh Ioong Baptista] khi dem diJng cho dwc Chua Jesu 'showing the way to Jesus' (Cat. 177k). de (211) 'to threaten': Cat. 86h has de ch~t, and Theurel (1877: 114) lists three synonymous compounds de m;t, de /oi, de phq,t. doi (228) 'to imitate, copy' with examples theo doi, /iJm doi listed as synonyms of bllt chu--&c: Theurel (1877: 125) has both theo doi and h9c doi. dom (231) 'to add, augment', about which Diet. has next to them dO'm the exrression noi dO'm _dq.t 'to exaggerate'; Cf. thi~u mfjt chiJ, chua, Vl vgy thz phai them dO'm chw ay 'since the word chua is missing, we have to add that word on' (Cat. 16z). gia (272) 'to show gratitude to, thank', as in giii O'J?; cf. co giii nhiJ hay /a giii chu nhiJ (Cat. 14c ), vi himg co ai giii nhiJ ma chJng giii chu nhiJ (Cat. 14e). gu'Q'ID (185) 'to wait, hold it'; Gu--q,m [spelled du--Q'm] dii! (Cat. 306p). I? (407) 'to bj afraid, fear', a lexeme that occurred freely in the fifteenth century (Nguyen 0inh-Hoa 1985a: 468) but later was -and is in modern parlance used only in compounds: for instance e If, SQ' If recorded in Theurel (1877: 233). mllng (452) 'to hear, perceive through the sense of hearing', therefore the equivale_nt of nghe_ thfiy, as o_pposed to n$he 't? listen'. Theurel (1877: 268) lists mang tin, mang tai, mang nghe, mang tieng, and a recent article by Nguy~n B~t-Tvy (1976) points out, using cognates in several Mon K~me~ languages, that the common reading of the n8m character -.Jif;.. as mang 1s wrong. nen (513) 'to become [so many years old]' as in nen mu--O'i tu8i, or nen mfjt, nen hai (Theurel 1877: 297) collocations in which modern VN would substitute /en 'to go up to, reach', Cf. 8ng Noe nen sau tram tu8i (Cat. 99p), ba Sara dii nen chin mu--O'i tu8i (Cat. 125a). phen le (405) 'to envy, begrudge'; cf. phen bz, phen le in Theurel (1877: 360), or phizn bz with the meaning 'to compare' in modern VN. tay (716) 'to, be partial to, favour', for which the example ip Diet. (dwc Chua b/jj,i) chang tay ai is also cited in Cat. 56k: phan xet chang tay ai 'in his judgement, was not partial to anyone'. 102

PAGE 111

Seventeenth-century Vietnamese lexicon 8. Postverbs do\l,ll (226) 'to compl~te, finish'; phq,m tfji doq,n (Cat. 89u), khi de con doq,n (Cat. 16le), khi de doq,n (Cat. 162s, t, u) 'after She gave birth [to Jesus]'. The example th8i doq,n 'after stopping' cited in Diet. shows that the usage has not changed, since modern VN has noi doq,n 'so saying ... comparable to da chep thw rfoq,n in BD 28. lien (413), modern VN lien, as in hqc lien 'to study continuously'; cf. gii1' lien ( Cat. 40g, h), which could be likened to luon. r6i (653) 'to finish, complete' > 'already': the early meaning persists in Saigonese (R8i chwa?-R8i. 'Did you finish yet? -I did'), the equivalent in the Hanoi dialect being Lam xong chwa? Lam xong r8i. Diet. gives the basic meaning of 'leisure' ['otium'], encountered in vo con.ls r_8i nghJ (Theurel 1877: 392) or in the modern expression an khong ngoi roi 'to sit idle, live in idleness'. REFERENCES Blnh, Philiphe Cao Hifo L.;mg Clark, Marybeth Gregerson, K. J. Haudricourt, A. G. Huinh Tjnh (Paulus) Cua Le Trung Hoa LeViin 01.l'c LeViin Quan Maspero, H. Nguy~n B![t-Tvy Nguy~n Dinh-Hoa 1822. Sach sb sang chep cac vifc (ed.) Thanh Lang. Saigon: Vi~n O![ih9c E>al![t, 1968. 1983. Thll' tim hiJu hai chiP 'song vifr trong Qu6c-am Thi-tap cua Nguy~n Trai. Khoa-hQc Xci-hi}i to & 11 (Dec. 1983), 11-15. 1974a. Submissive verbs as adversatives in some Asian languages. In South-East Asian linguistic studies (ed.) Nguy~n E>iing Liem. Canberra: Austral. Nat. Univ., 89-110 1974b. Passive and ergative in Vietnamese. In South-East Asian linguistic studies (ed.) Ngu~n E>iing Liem. Canberra: Austral. Nat. Univ., 75-88. 1969. A study of Middle Vietnamese phonology, Bull. Soc. Etudes Jndochinoises, 44(2), 131-93. 1974. Hai chiP B trong cu6n ti:1'-diJn cua Al~chxan dO' Rot, Ngon-ngll' 22 (Dec. 1974), 37-38, 63. 1895-96. -D(li-Nam qu&-iun t1,1-vj. Dictionnaire annamite. Saigon: Imp. Rey, Curio! & Cie. Tome I: A-L, 608 pages. Tome 2: M-X, 596 pages. 1982. Tim hiJu m9t s6 thanh-t6 m~t nghia trong ciic fa, ghep qua cu6n 'Dictionarium anamiticum [sic) Lusitanum et Latinum' (1651) cua A. de Rhodes. Ngon-ngu-S6 ph\l 2/82, 29-34. 1970. et al. _Vift-Nam tU'-diJn. 2 Vols. S1,1igon: Kh~i-tri. Ph~n thong thm'mg: Phan II: Tc-ngiP, thanh-ngiP, dien-tich, Phan III. Nhan-danh, <1ia-danh). 1981. Nghien-cwu vJ chit' nom. Hanoi: Khoa-h9c xa-h9i. 1912. Eludes sur la phonetique historique de la langue annamite. I. Les initiales, Bull. Ee.fr. Extr.-Orient 12, 1-127. 1976. 'Mang' hay 'm.lng'? Ngon-ngll' 27, 44-52. 1959. ChiP Norn, the demotic system of writing in Vietnam. J. Orient. Soc. 79(4), 270-74. 103

PAGE 112

NGUYEN DINH-HOA Ngu~n Dinh-Hoa 1972. Passivization in Vietnamese. In Langues et techniques. Nature et societe I (eds.) T. Bernot et al. Paris: Klincksieck, 179-87. 1981. Bilingual lexicography in Vietnam: The state of the art. In Papers of the Dictionary Society of North America ( 1979). London, Ontario: Univ. Western Ontario, School of Library Information Science, 149-71. 1982-84. Studies in chiJ' nom: the state of the art, Part 1, Vietnam Culture Journal 1(1) (Dec. 1982), 25-36; Part 2, ibid., Special Issue 2 (I & 2) (1983) and 3(1) (1984), 107-13. 1985a. Some archaic Vietnamese words in NguySn Trai's poems. Tn Linguistics of the Sino-Tibetan Area: The state of the art (ed.) Graham Thurgood et al. Canberra: Austral. Nat. Univ. 463-73. 1985b. Alexandre de Rhodes' Dictionary (1651). Papers in linguistics 18(4) (eds.) W. Frawley & R. Steiner. Edmonton, Alberta: Boreal Scholarly Publishers. NguySn Khiic Kham 1961-62. Ltrc;,c-s& c6ng-trinh bien-so\}n tU'-dii!n Vi~t-ngij, tiP th6-ky thu XVIT, Luijn-dam 1(12), (Dec. 1961), 144-8; 2(1) (Jan. 1962), 61-6; 2(2) (Feb. 1962), 222-9. NguySn Khfu: Xuyen 1960. LU'Q'c-khaov~ cuon tir-diJn Vi~t-B6-La (1651). Bach-khoa77 (3-8); 78 (3-7). Rhodes, A. de 1651a. Dictionarium annnamiticum, lusitanum et latinum. ape Sacrae Congregationis de Propaganda Frde in Lucem edition. Roma. Theurel, J. S. Trin Xuan Ng9C-Lan Van-Tan VU'O'ng Lqc 1651 b. Linguae annamiticae seu tunchinensis brevis dec/aratio. ape Sacrae Congregationis de Propaganda Frde. Roma. [Appendix to the Dictionarium.] 1651c. Catechismus pro iis qui volunt suscipere Baptismum in octo dies divisus. Phep giang tarn ngay cha ke mufm chiu phep r,;.a tiji ma r9eao dq.o thanh-Dwc Chua BlO'i. Reedite a ]'occasion du Tricentenaire de la mort de l'Auteur, avec introduction et notes par Andre Marillier. Saigon: Tinh Vi~t, 1961. 1877. Dictionarium anamitico-latinum (by Rev. Taberd). Ninh-Phu: ex typis Missionis Tunquini Occidentalis. 1982. SO'-bij khao-sai quftn til'-adn Chl-nam Ng<;>c-am Gitii-nghia ( Phien nom. Khao-luijn). Tom t~t lu~n-iin Ph6-ti~n-si Khoa-h9c. Hanoi: Vi~n Ng6n-ngij,-h9c. 1977. Til'-aiJn tiJng Vi~t. (2nd ed.) C6 chlnh-ly va b&-sung. Hanoi: Khoa-hqc Xa-h9i. 1978. V~ quii-trinh bi~n-d&i, (ch)>v. Ngon-ngie 38 (Dec. 1978), 42-4. 104

PAGE 113

THE PHONOIDGY OF KOMPONG THOM CHAM Robert K. Headley 0. Introduction Cham is a member of the Chamic Group of Austronesian. Other Chamic languages include Chru, Haroi, Jarai, Rade, and Roglai. It is classified into two major dialects: Eastern Cham (spoken by about 30,000 people in the Socialist Republic of Vietnam in the vicinity of Phan Rang) and Western Cham (spoken by about 150,000 people in Vietnam in the vicinity of Chau Doc and in Tay Ninh Province, and in Cambodia along the Mekong River and its tributaries). This study describes the synchronic and diachronic phonology of the Western Cham dialect spoken in Kompong Thom Province, Cambodia, 1 (KTC). This dialect differs somewhat from the Western Cham dialect of Chau Doc (CHD) described by Friberg and Hor (1977); it differs substantially from the Eastern Cham of Vietnam (CVN). The Western Chams apparently represent the descendants of immi grants who left Vietnam after the collapse of the Cham Kingdom in the sixteenth century. The majority of Western Chams are Muslims and use an adoptation of the Arabic script-via Malaysia-to write Cham. Recently, a Latin script has been devised to write Eastern and Western Cham. I. The Word The native Cham word-which may differ significantly in pattern from borrowed words-has the following syllabic patterns: MONOSYLLABLES: (C3V2)'C,(C2)V1F CV /la/ ['la:] CVF /saw/ ['sa\!] CVCV /taho/ ['tho:] /pat-:J/ [pa'h] CVCVF /kamlJ/ [ka'm:IJ] 'leaf 'dog' 'large jar' 'teach' 'be angry' I. This study is based on a corpus of data obtained from a male speaker in his late twenties from Phum Baray in southern Kompong Thom province. The Proto-Chamic reconstruc tions are mainly from Lee (1965) with some minor modifications, such as *ca?buay 'mouth' for Lee's *ca?buai, and a few new reconstructions by the author. Additional Western Cham data are from Friberg et al. (1977), and Eastern Cham data are from Blood (1967) and Moussay (1971). Forms between slants are phonemic while forms in italic are generally in the popular orthography of the language. 105

PAGE 114

ROBERT HEADLEY cvcvcv CVCVCVF CVCVFCVF /marasa/ /panataIJ/ /patanrau?/ ["maya'sa:] ["pana'tal)] ["pa tan 'yav?] 'maybe' 'animal' 'make heavy' In the patterns above: C = any consonant, V = a vocalic nucleus, marks primary stress of the following syllable, '' marks secondary stress of a following syllable, and F = any final consonant. Borrowed words from Khmer, Vietnamese, Arabic, Malay, Sanskrit, or European sources may show varied syllabic patterns. In some cases, there has been a reshaping of a foreign word into a Cham syllabic pattern. More recent borrowings are less likely to be reshaped than older ones. /brga/ [s5rga] 'heaven' (Skt. svarga) /fim/ [fi:m] 'film' (ultimately from French or English 'film') 2. Register Western Cham is a two-register language. The effects of register are seen most obviously on the vocalic nucleus. The High Register (HR) vowels are higher (in terms of tongue height), rather 'breathy' in voice quality, and associated with low pitch. Low Register (LR) vowels are lower in tongue height, often with lower on-glides, rather 'clear' in voice quality, and associated with higher pitch. Friberg and Hor (1977: 18-19), following Gregerson (1976), suggest that the physiological basis for register is the advancement or retraction of the tongue root. They describe First Register (with the tongue root retracted, called Low Register in this paper) vowels as generally lower and tense and Second Register (with tongue root advanced, called High Register in this paper) vowels as generally higher and lax with associated lower pitch and slightly breathy quality. The High Register developed in syllables which originally, in an earlier stage of Cham,2 began with voiced consonants. In the dialect of Western Cham described by Friberg and Hor (1977), the High versus Low Register opposition was noted only following stops. Low register vowel nuclei followed originally voiceless stops and high register vowel nuclei followed originally voiced stops. The remaining consonants seemed to be associated with the high register. In the Kompong Thom dialect, original clusters of /h/ + /w, I, y, r, m, n, IJ/3 have lost the /h/4 and have vowel nuclei associated with the low register. In some cases, the original Proto-Chamic (PC) cluster had initial /s/ which must have become /h/ before the Cham Empire broke up, e.g.: *huma 'field' > /mi/ [m"i] cf. CHD hamw and CVN hmu or hamu) *hana? 'asthma' > /ni?/ [n"i?] (cf. CHD hanwc and CVN hanw') 2. Since the traditional Cham script distinguishes between voiced and voiceless stops and it was adopted prior to the eighth century A.D., it is believed that the voiced-voiceless distinction was present in Cham at that time. 3. /h/ + /fi/ has not been found. 4. In careful speech, the /h/ may be pronounced, but it has been lost in the everyday spoken language. 106

PAGE 115

The phonology of Kompong Thom Cham *sana 'roast' > /na/ [na:] (cf. CHD hana and CVN hna) *hure? 'vine' > /rs?/ [ys?] (cf. CHD hrek and CVN hare') *surii? 'writing' > /rii?/ [ya?] (cf. CHD hrak and CVN hra? or sra?) *sula 'leaf > /la/ [la:] (cf. CHD hla and CVN hala) *haway or huway 'rattan' > /wiiy/ [wai] (cf. CHD haway or CVN hawey) The contrast between the reflexes of *w, *l, *y, *r, *m, *n, *IJ and *h(V) + *w, *l, *y, *r, *m, *n, *IJ can be seen in pairs like the following. /ni/ [nai':] 'bee' < *hun1 /ni/ [nt:] 'this' < *?uni /mltf [mit] 'hear' < *hmgt /mit/ [mit] 'always' < *(mit)t /w~?/ [ya?] 'eat' < *hua? /wii?/ [yii] 'pull down' < *(wii?) /y~w/ [iay] 'like' < *(hy.iy) /yiiw/ [iiiy] 'yoke' < *y.iy /rii?/ [ya?] 'writing' < *surii? /r~?/ [yii?] 'vein' < *?uriit 3. Consonants The following consonants occur in KTC: BILABIAL DENTAL PALATAL VELAR GLOTTAL STOPS p t C k ? b d J NASALS m n i'i lJ SONANTS w 1 y r s h /f, f, s, z/ also occur in borrowed words. The consonant phonemes of KTC are described in detail below. /p t c k/ are voiceless, unaspirated bilabial, dental, palatal, and velar stops respectively. They are unreleased in final position. When followed by / /, they are lenis. /paliiy/ 'village' [p~'liii] /cip/ 'Thursday' [cip] /tara? / 'sow rice' [t~'y:?] /matiih/ 'raw' [m~'tah] /cim/ 'bird' [cim] /tacui?/ 'spit' [t~'cui?] /kapal/ 'thick' [k~'pa:l] /paha/ 'thigh' [pha:] /tii?/ 'chop trees' [ta?] /tahiiw/ 'know' [thay] /!Jiit/ 'be careful' [l)at] /cakiiw/ 'cut with scissors' [c~'kay] /ks? J 'to bite' [h?] /take/ 'horn, antler' [t~'ke:] t Asterisked entries in parentheses ( ) are tentative reconstructions (Ed.). 107

PAGE 116

ROBERT HEADLEY /palak/ 'muddy' [p~"lak] /t?J? / 'be at' [t5":?] /ku/ 'kind of taro' [ku:1 /tahu/ 'bat' [thu:1 /para/ 'shoulder' [p~"ya:] /toah/ 'search for' [toah] /patii.r/ 'remind' [p~'t:lw] /kara?/ 'rough' [k~'ya:?] /t::J?j 'bottom' [t::J:?] /ku/ 'Cambodian' [ku:] /tahu/ 'dry' [thu:] /pa/ 'lead' [pa:] /tapa/ 'bland' [t~'pa:] /tar:lp/ 'of [t~'y:lp] /kiu/ 'oven' [ki:u] /caru/ 'medicine' [c~'yu:1 Modern KTC initial/pt ck/ are reflexes of Proto-Chamic *p, *t, *c, *k and *b, *d, *j, *g. The four voiced PC stops conditioned following high register vowel nuclei. *pat;:iw 'stone' > /pataw/ *tub;:iw 'sugarcane' > /tap:lw/ *pagi 'tomorrow' > /pake/ *?ad;:iy 'younger sibling' > /t:ly/ *cim 'bird' > /cim/ *ju? 'black' > /cu?/ *tupa? 'straight' > /tapa?/ *dur;:iy 'thorn' > /truay/ *bala 'tusk' > /pla/ *tian 'abdomen' > /tean/ *ke? 'bite' > /kf.?/ *groh 'to bark' > /kr3h/ The clusters *diand *tl become /kl-/. *tlaw 'three' > /klaw/ *dleh 'tired' > /kl~h/ Voiceless stops are retained in words borrowed from Khmer. Khm. /ph;:iu/ 'inch' > /ph~rJ/ Khm. /ckiel/ 'gouge' > /cakial/ Khm. /kmaoc/ 'ghost' > /kamot/ Khm. /tnaot/ 'sugar palm' > /tanot/ Khm. /ta;:i/ 'only' > /t;:i;:i/. Some words which contained voiced stops in Middle Khmer were apparently borrowed before these stops were devoiced in Khmer and then subsequently devoiced in KTC. The vowel nuclei in KTC that follow originally voiced Khmer stops are those asociated with the high register. Khm. /peil/ (be[) 'time' > /pel/ Khm. /koum/ (gom) 'lamp' > /kom/ Khm. /toap/ (da,pa) 'army' > jt5p/ Khm. /tup/ (da'pa) 'stop' > /t5p/ There are several problems with the reflexes of the final PC stops. Generally, KTC has/-?/ as the reflex for *-p *-t *-c *-k and*-?. Final *-p 108

PAGE 117

The phonology of Kompong Thom Cham becomes /-u?/;5 final *-t becomes/-?/; final *-c becomes /-i?/ or/-?/ after front vowels, and final *-k and *-?become/-?/. Examples of the normal reflexes are given below. *lap 'fold'> Mu?/ *gap 'each other' > /k~u?/ *?asap 'smoke' > /siiu?/ *chiap 'wing' > /ceau?/ *hadip 'alive' > /tiu? / *?diap 'sticky rice' > /diau? / There are numerous examples of final /-p/ in all phonetic environments in KTC. Many of these words are clearly borrowings, but others are not identifiable as such. /kadp/ 'solid' cf. Chru kO'jap, S. Roglai kojap, and Rade kjiip. /kiimsup/ 'dawn' cf. CHO masup and N. Roglai m8sup. /radap/ 'accustomed' cf. Chru lO'dap and S. Roglai lodap. /pak5p/ 'compare' cf. Chru pO'gap. /pap/ 'evil' cf. Khm. /baap/ ( < Skt. plipa). /tanap/ 'low' cf. Old Khm. dnlip 'lowland' < dlipa 'low' /l5p/ 'erase' cf. Khm. /lup/ (lu'pa). /r.p/ 'accompany' cf. Khm. j?aep/ ('epa) 'take shelter' /rup/ 'body' cf. Skt. rflpa. *haget 'what' > /ke?/ 'jhit 'sew' > /chi?/ *tu?ut 'knee' > /ta?ii?/ *lal)it 'sky' > /lal)i?/ Lee (l 965) has two reconstructions of PC words with final *-t that have kept /-t/ in Cham. *capat 'squeeze' > /capt/ *kawat 'wire' > /kawEt/ There are other words with final /-t/ for which reconstructions are not available, but which have cognates in the other Cham dialects. /kliit/ 'stuck, jammed' cf. CHO glut and CVN kliit (
PAGE 118

ROBERT HEADLEY There are also a fair number of words with final /-t/ in KTC for which no cognates ~ave been found: /pi!J 'of, /pr~t/ 'run fast', /cawet/ 'erratically', /cut/ 'small bamboo', /mit/ 'run', /padit/ 'make up a story', /padl':t/ 'make a fin_gerprint', /prit/ 'to set limits', /sr5t/ 'collapse', /wet/ 'shake the head', /yit/ 'keep on doing'. Of interest is the fact that, in all of these words, /-t/ is preceded by a short vowel and in most of them the vowel is a front vowel. Borrowed words with final /-t/ are numerous: /ahat/ 'Sunday' (Malay ahat, ahad), /pf.t/ 'hospital' (Khm. jpEEt/), /b5t/ 'turn' (Khm. /bat/, /carit/ 'cricket' (Khm. /caIJr'Jt/), /makl':t/ 'female' (Malay megat ?). There may be some modifications of the vowel: /pit/ 'surround' (Khm. /poat/ ba' ta), /t~t/ 'kick with the toe' (Khm. /toat/ da' ta). Final /-c/ in Khmer loans becomes /-t/ in KTC: /khut/ 'broken' (Khm. /khooc/), /kamot/ 'ghost' (Khm. /kmaoc/). *bile 'pull up' > /pui? / *pruec 'intestine' > /proi? / *hue 'whistle' > /hui?/ *?amrec 'pepper' > /mre? / *briik 'rotten' > /prfr?/ *bru~? 'work' > /prfr?/ *katiik 'flatus ventrus' > /katu?/ *pa? 'four' > /pa?/ *hua? 'eat rice' > /hoa?/ *duac 'run' > /duai?/ *sac 'shake out' > /sai?/ *huac 'afraid' > /huai? / *piriik 'silver' > /prea?/6 *manak 'oil' > /mafii?/ *pitu? 'star' > /patu?/ *IJ6? 'upgrade' > M5?/ *tasi? 'sea' > /tasi? / Usually, following a nasalised vowel in PC and in borrowed words, /-k/ and j-?/ become /-k/: *fiii? 'dive' > /nuk/ *masak 'brave' > /masiik/ *tulok 'disk shape' > /tal5k/ but, in some cases, no evidence of a nasalised vowel has been found: *kutok 'grasshopper' > /kat::>k/ *jiik 'clever' > /dk/. KTC also retains final stops intact in borrowings from Khmer. Khm. /sroop/ 'absorb' > /srop/ Khm. /aep/ 'move' > /cp/ Khm. /book/ 'hump' > /bok/ Khm. /lout/ 'jump' > /lot/ /b d j/ are voiced, fortis, optionally preglottalised, bilabial, dental, and palatal stops. The preglottalisation is especially noticeable in /j/ which appears to vary between [?j] and [?i]. These stops do not occur in final position. 6. There may have been metathesis here: *piriik (I am unsure of the length here, it may have been *pirak which would explain the Cham form better) > *priak > Cham /prea?/. 110

PAGE 119

The phonology of Kompong Thom Cham /beah/ 'until' (?beah] /birJ/ 'eat' [?birJ] /kad;}Tj/ 'bell' [ka'?d;}TTJ] /pajuan/ 'move' [pa'?juan] /dEh/ 'sputter' (?dsh] /labiih/ 'miserable' [l;\'?bah] /jau?/ 'must' (?ja11?] The KTC preglottalised series reflects preglottalised consonant in PC: *?barJ 'eat' > /birJ *ca?buay 'mouth' > /cabuay/ *?duac 'run' > /duai?/ *?di? 'climb' > /di?/ *?jau? 'must' /jau? / *?jiih 'firewood' > /jiih/ Preglottalised stops are retained in words borrowed from Khmer: Khm. /book/ 'hump' > /bok/ Khm. /samba;}m/ 'great' > /sab;}m/ Khm. /dael/ 'which' > /del/ /m n ii rJ/ are bilabial, dental, palatal, and velar nasals respectively. All but /ii/ have been found in final position. /mit/ 'hear' [mi't] /m5h/ 'place' [m:,h] /piim/ 'kind of pancake' [pam] /ni/ 'bee' [n~i':]] /bn/ 'group' [b:n] /rJin/ 'noun' [rJan] /maiium/ 'drink' [ma'iiu:m] /mit/ 'always' [mi"t] /m5h/ 'hate' [m::,h] /ni/ 'this' [ni":] /narirJ/ 'bow' [naYi:rJ] /nom/ 'dye' [iio:m] The PC nasals *m, *n, *ii, and *rJ are retained unchanged in KTC. *mata 'eye' > /mata/ *tama 'enter' > /tami/ *maiiak 'oil' > /mafii?/ *?ina 'main' > /ni/ *rJii? 'do'> /rJi? / *sarJ 'house' > /sarJ/ *maliim 'night' > /malam/ *nu 'he' > /nu/ *naw 'go' > /naw/ *jalan 'road' > /calan/ *larJa 'sesame' > /larJi/ The nasals are also retained in words borrowed from other languages. Khm. /mei/ 'chief > /me/ Khm. /maak/ 'type' > /mak/ Khm. /nien/ 'addicted' > /nian/ Khm. /aerJ/ 'self > /erJ/ Mai. hina (ult. from Skt. hina) 'inferior' > /hina/ 'difficult' Mai. saman (ult. from Ar. zaman) 'era' > /samiin/ or /zamiin/. /w/ is a bilabial semi-vowel. It occurs in C1 and F positions, and, in rare occasions, in C2 position; and it is associated with high register nuclei. Original *hVwclusters have become /w/ following by low register nuclei. /wi?/ 'pull down' [11a?] /lawiy/ 'stir' [la'11ai] 111 /yaw/ 'like' [ia11] /iw/ 'left (side)' (?i:11]

PAGE 120

ROBERT HEADLEY In at least one case, the diphthong /oa/ which is a reflex of PC *ua and which seems regularly to become /wa/ in Eastern Cham, formed a cluster with /w/ as C2 This forms alternates with the expected /oa/ form. /hoa?/ and /wa?/ ( < *hwa? < PC *hua?) 'eat rice' /w/ is a reflex of PC *w. In borrowings from Khmer, /v/ is reflected by KTC /w/ in all positions. *war 'stable' > /war/ *pat;}w 'stone' > /pataw/ *lawa or jawa 'soul' > /nawa/ *?iaw 'left (side) > /iw/ *haway (or) huway 'rattan' > /way/ Khm. /vie!/ 'field' > /wal/ Vn. chiJu 'mat' > /ciaw/ Khm. /sievphi:v/ 'book' > /saphaw/ or /saphaw/ /1/ is a lateral sonant. It occurs in C1 C2 C3 and F positions. /Ja?u/ 'coconut' [la'?u:] /lan-i:ilJ/ 'eel' [la'nulJ] /blay/ 'buy' [?blai] /ta!/ 'until' [ta!] /1/ is a reflex of PC *I; the clusters *h(V)land *s(V)lare reflected in KTC by /1/ followed by a low register nucleus. /1/ in borrowed words is retained. *lu 'much, many' > /lo/ *le? 'fall' > /IE/ *glo 'brain' > /kl3/ *tulalJ 'bone' > /talalJ/ *lupan 'centipede' > /Japan/ *jal 'net' > /ea!/ *kapal 'thick' > /kapal/ *ta! 'until' > /ta!/ *hVJa? 'rice dust' > /Ja?/ *hVlua? 'sharp' > /lua?/ *hulun 'slave' > /Jin/ *sula 'leaf > /la/ Khm. /lout/ 'jump' > /lot/ Khm. /peil/ 'time' > /pel/ Khm. poolih/ (ult. from French, police) > /plih/ 'police' French cha/and 'ferry' > /chalal)/ Skt. dhuli 'dust' /thdi/ Khm. /r::ibuah/ (spoken: /l;}buah/) 'wounded' > /labuaih/ /r/ has two allophones. In all positions but Fit is a voiced velar fricative [y] which sometimes seems to approach a voiced stop [g]. In C2 position preceding high central vowels, /r/ shows a tendency to be lost; in F position, it is a voiced, high-mid, unrounded semi-vowel [w], while in C3 position, there is some alternation between /r/ and /1/. /ruh/ 'shake out' [yuh] /prfJ.y/ 'give' ['payai] /krim/ 'bamboo' ['kayi:m] or [ki:m] /par/ 'to fly' [paw] /rani:IJ/ or /lani:IJ/ 'wide' [ya'ni::I)] or [lani:I)] /r/ in all positions is a reflex of PC *r, which in F position following a high back rounded vowel or a diphthong is lost in KTC. /r/ in borrowed words has several treatments. In most borrowings from Sanskrit and Khmer, it has the same allophonic distribution as it has in native Cham words, but in some Sanskrit and Arabic loans, /r/ is a flap [r]. 112

PAGE 121

The phonology of Kompong Thom Cham *ruah 'choose' > /roah/ *kram 'bamboo' > /krim/ *cur 'lime' > /cu/ *kuar 'gather up' > /kua/ Skt rupa 'body' /rup/ *?ariaIJ 'crab' > /rial)/ *?usar 'seed' > /sar/ *bier 'dwarf > /pla/ Skt. nagara 'fortress' > /lakir/ 'country' Skt. naraka 'hell' > /na'fka/ Skt svarga 'heaven' > /s':lrga/ Malay murtabat or martabat (ult. from Arabic) 'grade, step' > /miiftabat/ 'knowledge' Khm. /riep/ 'level' > /rap/ Khm. /sroop/ 'absorb' > /srop/ /y/ is a palatal semi-vowel. In C1 position preceding a vowel it may have considerable friction, approaching [j ]. /yaw/ 'like' [ia11] /yiw/ 'yoke' [ia.11] /hatiy/ 'after' [hi'i'tai] /kayaw/ 'tree' [ki'i'ia11] /y/ in all positions reflects PC *y. The cluster *hVygives /y/ followed by low register vowel nuclei. *yi'iw 'yoke' > /yiw/ *hayuak 'harvest' > /yoa?/ /y/ in borrowed words remains /y/. *hayi'iw 'like' > /yaw/ *tuay 'stranger' > /tuay/ Khm. /cla::Jy/ 'answer' > /chl::Jy/ Mai. dunia (ult. from Arabic) 'world' > /tiinya/ /s/ is a voiceless dental spirant. It occurs in C1 C3 and C2 positions. It occurs in F position only in some recently borrowed words. /sit/ 'small' [si:t] /paseh/ 'castrate' [pi'i'seh] /sabo/ 'colour' [s;'i'?bo:] /kumiis/ 'dictionary' [kumus] /s/ is a reflex of PC *s. Final s in early borrowings is usually reflected by /-ih/; final PC *s has become /-h/. *su::Jy 'slow' > /suay/ *?usar 'seed' > /sar/ *sruh 'nest' > /sriih/ *bras 'husked rice' > /prih/ *tikus ( or) *tukus 'rat' /takiih/ *kakas 'scale' > /kakah/ Khm. /saak/ 'attempt' > /sak/ Skt. do:fa 'fault' > /tuih/ Skt. miinu:ja 'man' > /manuih/ Khm. /cbah/ < (cbiis) 'clear' > /cabaih/ Khm. /cumnuah/ < (jamnuas) 'substitute' > /camnuaih/ PC *s in C3 position may be lost before stops or retained before other consonants. In the latter case, *s sometimes becomes *h which in turn is lost after causing a change in the register of the following syllable. 113

PAGE 122

ROBERT HEADLEY *sagor 'drum' > /kir/ *sara 'salt' > /sra/ *sidor (or) *sid;:ir 'remember' > /tir/ *sidOm (or) *sudOm 'ant' > /t5m/ *samu (or) *sumu 'flat' > /samu/ *surii? 'writing' > *hara? > /ra?/ *sula 'leaf > *hala > /la/ *sana 'roast' > *hana > /na/ /h/ is a voiceless glottal spirant. It occurs in all positions. /hoa/ 'pull tight' [ho;:i] /hawa/ 'dissatisfied' [h::\'1Ja:] /pahi?/ 'bitter' [phi:?] /plih/ 'split' [plah] /h/ in all positions reflects PC *h. Final PC *hand *shave merged, and are reflected by /-h/. In borrowed words, /h/ remained /h/. Arabic l;i merged with Cham /h/. *ha 'yes' > /hi/ *hia 'cry' > /hea/ *hue 'whistle' > /hui?/ *bahr;:iw 'new' > /phiw/ *hadum 'how many' > /hatom/ *?abih 'all' > /pih/ Mal. hakem 'judge' (ult. from Ar. bakim) < /hakem/ Khm. /cneah/ ( /cani /f s z/ appear to be limited to borrowed words especially those from Arabic through Malay. /f/ is a voiceless bilabial spirant; /s/ is a voiceless palatal spirant, and /z/ is a voiced dental spirant. In casual speech, /s/ and /z/ usually become /s/. /bfga/ 'heaven' < Skt. svarga /fal/ 'memorise' < Ar. fa/ /israt/ (or) /israt/ 'show' < Mal. isharat (ult. from Ar.) 'signal' /izin/ 'permission' < Mal. idzin (or) izin (ult. from Ar.) There seem to be two kinds of initial consonant clusters in KTC. One may be represented as /C1C2 / and consists of two consonants in close juncture without any intervening vowel. C2 in these clusters is limited to /1 r h/. The following examples have been noted in this corpus: SECOND MEMBER I r h p X X X t X C X k X n (X) The second type of cluster may be represented as /C1aC2/. Jn clusters of this type, there is an intervening short vowel. Some of these show considerable variation. One speaker, in slow, careful speech may pronounce a cluster with the short vowel, e.g., [t::\hun] 'year', while the 114

PAGE 123

The phonology of Kompong Thom Cham same speaker in more casual, rapid speech may pronounce the same cluster without any perceptible intervening vowel, e.g. [thun]. The C2 in these clusters can be/pt ck? j m n fi IJ w 1 yr s h/. As described in the second type above, there has been some simplification of PC clusters of *hVC and *sVC-. In at least one case, the historical treatment of clusters differs. PC *tland *di-> CHD /kl-/ but *tVI-> KTC /tal-/, *tlaw 'three' > /klaw/, *tuHtIJ 'bone' /talaIJ/. Cham also has consonant clusters in polysyllabic words. In these, the common pattern is CVCCV(F), as in: /cakl~/ 'lightning' [ci)'kla:] /paplay/ 'to sell' [pi)'plai] /kanram/ 'trap' [ki)n'yam] /masriih/ 'war' [mi)s'yuh] /tapli':t/ 'scraped off' [ti)'plEt] Another pattern has a short vowel between the two middle consonants. The third syllable receives the main stress while the first receives secondary stress. /marasa/ 'maybe' ["mayi)'sa:] /masatah/ 'from a distance' ["masi)'tah] /palahiw?/ 'to destroy' ["pali)'hiw?] Finally, there is the pattern CVCVFCVF as in: /patanrau?/ 'make heavy' ["pati)n'ya1p] 4. Vowels KTC has the following vowels: FRONT CENTRAL HIGH i i i MID-HIGH e MID ;:) g MID-LOW E i'; LOW a a GLIDES: /ia, i{!l, ea, au, ua, oa/ BACK u ii 0 6 :, 5 All vowels that precede /-h/ are short, but they are still marked with the breve ( ~) in this article. First or low register vowels are generally lower than their second or high register counterparts and may have on-glides. The vowels are now described in detail: /i i/ are high front vowels. First register /i/ is [ai':] /ni/ 'this' [ni:] /Jin/ 'I' [li'n] /ni/ 'bee' [nai':] /padi?/ 'hurt' [pi)?di:?] 115

PAGE 124

ROBERT HEADLEY KTC /i/ and fi/ are reflexes of PC *i, *1, *i, and *i. *cim 'bird' > /cim/ *jhit 'sew' > /chi?/ *?aIJin 'wind' > /IJin/ *phi? 'bitter' > /phi?/ *?abih 'all' > /pih/ *gil) 'stove' > /kil)/ *laIJ1?; 'sky' > /laIJi?/ KTC /i/ and fi/ may reflect /i/ and /::i/ in borrowed words. Pali bhikkhu > Khm. /phi?kho?/ > /phi?/ 'Buddhist monk' French police > Khm. /pooliih/ > /plih/ 'police' Khm. /bn/ > /kin/ 'to mill rice' Skt. sak:;in > Malay saksi > /saksi/ 'witness' /e/ is a long mid-high front vowel; no short /e/ has been found in the present material except preceding /-h/. /me/ 'chief [me:] /pel/ 'time' [pe:l] /?eIJ/ 'self [?e:IJ] /seh/ 'student' [seh] /take/ 'horn, antler' [t~'ke:] KTC /e/ is a reflex of PC *I. *pagi 'tomorrow' > /pake/ *tuki 'horn, antler' > /take/ It also occurs in words borrowed from Khmer in which it is a reflex of /ei/ and /ae/. Khm. /peil/ 'time' > /pel/ Khm. /?aeIJ/ 'self > /?eIJ/ Khm. /mei/ 'chief > /me/ Khm. /dael/ 'have ever' > /del/ /E './ are mid-low front vowels; first register /E/ is [aE] in open syllables. /It?/ 'fall into' [JE?] /sarn/ 'debt' [s~'yaE] /cth/ 'insert' [cEh] PC *e and *e and *e are reflected by KTC /El and /El. *ke? 'bite' > /kr,?/ *bube 'goat' > /papE/ *?eh 'excrement' > /?'.h/ *sre 'debt' > /sarn/ /ii/ are high central unrounded vowels; first register /i/ is ["iT:] in open syllables. /bil)/ 'eat' [?biIJ] /mi/ 'field' [m"i':] /mi?/ 'get' [mi-?] /ni-/ 'chief, main' [n+:] KTC /i/ and Ji/ are reflexes of PC *a and *:i and *~ and *~-*tan:i? 'cook' > /tani?/ *lima 'five' > /Jami-/ *lan:iIJ 'worm' > /laniIJ/ *luman 'elephant' > /Jamin/ 116

PAGE 125

The phonology of Kompong Thom Cham *lumia? 'prepare' > /lamiq_i:?/ or /rami~?/ *ha 'yes' > /hi/ *?b:il) 'eat' > /bi!)/ *c:i? 'mountain' > /ci?/ *d;ilJ 'stand' > /t'iIJ/ KTC /i/ may reflect /a/ in ancient borrowings. Khm. /cneah/ (jhnah) > /canTh/ 'win' The word finIJau?/ 'verb?' is probably a loan from Malay. /;i :i/ are mid-central unrounded vowels. The perceived difference between first register /ii and second register /:i/ is often very slight. /y;i/ final emphatic particle [i;i:] /p;il)/ 'blown away' [p;i:IJ] /pg?/ 'small dam' [p;"J?] /p;i?/ '(banana) sheath' [p;i:?] /cal:iIJ/ 'very tall' [c:i'l;iIJ] KTC /;i/ and /:i/ are reflexes of PC *;) and *;i. They also derive from Khmer /;i;i/ and /a;i/ in borrowed words. *p:il) 'to nail, pound' > /p:iIJ/ *p:ih 'open' > /p:ih/ Khm. /cla;iy/ 'answer' > /cby/ *w;i? 'instead' > /wg?/ *( ... )t;i? 'tremble' > /cata?/ Khm. /somba;im/ 'important' > /sab;im/ Khm. /pr:,h;i;in/ or /p;ih;i;in/ 'arrogant, insolent' > /ph;'Jn/ There is one instance of KTC /;i/ from Khmer /a;i/. Khm. /tae/ 'only' > /t;i;i/ (cf. p.108 above). /a a/ are low central unrounded vowels. /a/ in unstressed presyllables represents a very short mid-central vowel [:i]; when it occurs before /m/ in presyllables, it is backed and rounded to [3]. /saIJ/ 'house' > [sa:IJ] /nan/ 'that' [nan] /tiy/ 'younger sibling' [tai] /kapa?/ 'walk' [k:i'pa:?] KTC /a/ and /a/ are reflexes of PC *a and *a. PC *:i preceding *r and *I is also reflected by /a/. *blah 'split' > /plih/ *dalam 'deep' > /tal~m/ *pha 'thigh' > /pha/ *pa? 'four' > /pa?/ *?ula 'snake' > /la/ *t:il 'until' > /ta!/ *haIJ 'spicy' > /haIJ/ *katal 'itchy' > /katal/ *tulaIJ 'bone' > /talaIJ/ *hujan 'rain' > /can/ *g:ir 'handle' > /k~r/ KTC /a/ may reflect /a/ m borrowed words as well as /ie/ m some borrowings from Khmer. 117

PAGE 126

ROBERT HEADLEY Pali akkhara > /akhiir/ 'letter' Pali kapplisa > Khm. /kapbaah/ > /kapiih/ 'cotton' Malay aka! (ult. from Arabic) > /akiil/ 'intelligence' Skt. rlijaklira > Khm. /rieccakaa/ > /racaka/ 'government' Khm. /riep/ (rlipa) > /rap/ 'level' /u ii/ are high back rounded vowels; first register /u/ and /ii/ have lower on-glides [0u] or [u]. /hu/ 'have' [h0u] /kufJ/ 'who' [kufJ] /nu/ 'he, she, it' [nu:] /ciiriih/ 'fall down' [c;)'yu"h] KTC /u/ and /ii/ are reflexes of PC *u and *u and PC *u and *ii. *tub 'pour' > /tub/ *pitu? 'star' > /patu?/ *tupufJ 'flour' > /tapufJ/ *thun 'year' > /thun/ *manii? 'chicken' > /manu?/ sriih 'nest' > /sruh/ *nus 'blow nose' > /nuh/ *ju? 'black' > /cu? / *cur 'lime' > /cu/ *ja?bu 'dry' > /cabu/ *jamuk 'mosquito' > /camu?/ *nu 'he, she, it' > /nu/ KTC /u/ and /u/ reflect /u/ in borrowed words. Skt. dhuli > Khm. /thuulii/ > /thiil/ 'dust' Malay dunia (ult. from Arabic) > /tiinya/ 'world' Khm. /puun/ (buna) > /pun/ 'to mound up' Skt. guru > Khm. /kruu/ > /kru/ 'sorcerer' Malay kubur (ult. from Arabic) > /kupur/ 'tomb' /o o/ are mid-high back rounded vowels. /?o/ final negative particle [?o:] /pahofJ/ 'group' [pho:fJ] /p6h/ 'fruit' [p6"h] /lo/ 'very' [lo":] KTC /o/ and /o/ are reflexes of PC *u, *u and *ii. In loanwords, KTC /o/ may also reflect Khmer /oo/, /ou/, and /ao/. *?buh 'see' > /boh/ *hiifJ 'papaya' > /lahorJ/ *hadum 'how long' > /hatom/ *tagu? 'arise' > /tako?/ *Ju 'many' > /lo/ *tu? 'receive' > /to?/ Khm. /tnaot/ 'sugar palm' > /tanot/ Khm. /youfJ/ 'two' > /yofJ *plum 'leech' > /plom/ Khm. /sroop/ 'absorb' > /srop/ Khm. /lout/ 'jump' > lot/ /'J 5/ are mid-low back rounded vowels. /lam'J 'cow' [l;)'m'J:] /15?/ 'to be located at' [fr?] /t5p/ 'hit with the flat of the fist' [t'Jp] 118

PAGE 127

The phonology of Kompong Thom Cham /k5h/ 'knock' [k5h] KTC /'J/ and /5/ are reflexes of PC *o, *6, *o, and *6. *?ak6? 'head' > /k5?/ *gr6h 'to bark' > /kr5h/ *pato 'teach' > /pat'J/ *roIJ 'nourish' > /dIJ/ *lok 'to peel' > /15?/ *IJ6? 'on' /IJ5?/ *hol) 'wasp' > /h'JI)/ Lee's (1965) *0 sporadically becomes KTC /5/. *?brOm 'arrow' > /br5m/ *sudOm (or) *sidOm 'ant' > /t5m/ Khmer /oa/ and /'J-J/ become /5/ in KTC; /oa/ becomes // before /-t/. Khm. /toap/ 'army' > /t5p/ Khm. /cr'J-Jk/ 'pickle' > /cr5k/ There is abundant evidence in KTC for a split in the PC high and low vowels: *i > /i/ and /e/, *u > /u/ and /o/, and *a > /a/ and /i/. The conditioning factor for the split of *a is apparently nasalisation: *a > /a/, while *a > /i/. The conditioning factor for the split of the high vowels has not been discovered. Lee (1977: 89-92) relates similar vowel splits in Haroi to: (1) the voicing feature of a preceding obstruent, and (2) the height feature of a preceding vocoid. Both reflexes of *i and *u have been found in apparently identical environments in KTC. It is always possible that the /e/ and /o/ reflexes, which are statistically much rarer than the /i/ and /u/ reflexes, are limited to borrowings from other dialects or from related languages, but so far, in most cases, no sources have been found. The following glides occur: HIGH MID LOW /sia/ 'near' [si-J] FRONT /lahiw?/ 'lose' [l;)'hiq_i:?] /ceaIJ/ 'craftsman' [ceaIJ] /kua/ 'embrace' ]ku-J] /toa/ 'two' [to-J] CENTRAL BACK /hapian/ 'when' [h;)'pi-Jn] /seam/ 'good' [seam] /yuan/ 'Vietnamese' [iibn] /roah/ 'seek' [Yo-Jh] /sau?/ 'smoke' [sa~?] 119

PAGE 128

ROBERT HEADLEY KTC /ia/ is a reflex of PC ie, *ia, and *io. It also occurs m borrowings from Khmer and Vietnamese. *bier 'dwarf > /pia/ *palia 'hail' > /plia/ *pioh 'keep' > /mapiah/ Kh. /ckiel/ 'scrape' > /cakial/ Vn. chiJu 'sleeping mat' > /ciaw/ Khm. /nien/ 'addicted' > /fiian/ KTC /iw./ occurs only before /-? / where it is a reflex of PC *iii?. *lumia 'put away' > /lamiw.?/ *lahia? 'lose' > /lahiw.?/ *ma?ia? 'urinate' > /ma?iw.?/ KTC /ea/ is a reflex of PC *ia/ *hia 'cry' > /hea/ *liah 'lick' > /leah/ *pa?dia? 'hot' /padea?/ *tian 'abdomen' > /tean/ *biak 'true' > /pea?/ KTC /ua/ is a reflex of PC *u::, and *u:}. It also occurs in loanwords from Khmer. *ru::,y 'fly' > /ruay/t *lu::,n 'to swallow' > /luan/ *su::,y 'slow' > /suay/ *kamu:}n 'nephew' > /kamuan/ Khm. /phuay/ 'blanket'> /phuay/ Khm. /sua/ 'ask' > /sua/ Kh. /t::itual/ 'accept' > /tatual/ KTC /oa/ is a reflex of PC *ua and *ua. *dua 'two' > /toa/ *?dua 'carry on the head' > /doa/ *caIJua 'basket' > /caIJoa/ *cuah 'sand' > /coah/ *jua? 'step on' > /coa?/ KTC /au?/ is a reflex of PC *-ap and *5y?.7 (See also the treatment of PC final *-p above p. 109). *cap 'tie' > /cay?/ *tr5u? 'heavy' > /tray?/ Presyllable Vowels *?anap 'in front' > /tanay? / *hau? 'sweat' > /hay?/ Lee (1965), mainly on the basis of evidence from Roglai, reconstructed three presyllable vowels: *i, *u, and *a. In KTC these have all merged to /a/, which is [3]. When the presyllable began with *?-, *h-, and sometimes *s-, the presyllable has been lost in KTC. tcf. VN ru8i 'housefly, fly' (Ed.). 7. Although I suspect that *:\11? derives ultimately from *iip. 120

PAGE 129

The phonology of Kompong Thom Cham *?ina 'female animal' > /ni/ *dilah 'tongue' > /talih/ *sula 'leaf' > /la/ *bilan 'season' > /plan/ *?uni 'this' > /ni/ *lima? 'five' > /Jami/ *?adh~y 'forehead' > /thiy/ *pat::iw 'stone' > /pataw/ *kubaw 'water buffalo' > /kapaw/ *haway 'rattan' > /way/ There may also be some evidence of the presyllable vowel in cases of metathesis. *bul)a 'flower' > /pa!Ju/ *durgy 'thorn' > /truay/ *piriik 'silver' > *pria? > /prea?/ *man1 'bathe' > *mnay > /manay/ REFERENCES Blood, D. L. Friberg, T., Friberg, Barbara & Kvoeu-Hor Friberg, T. & Kvoeu-Hor Gregerson, K. J. Lee, E.W. Moussay, G. 1967. Phonological units in Cham. Anthrop. Ling 9(8), 15-32. 1977. Weslern Cham thesaurus vocabulary. Huntington Beach, CA.: Summer Institute of Linguistics. 1977. Register in Western Cham Phonology. In Papers in South East Asian Ling. 4, Chamic Studies (Pacific Ling. A. 48), (eds.) D. Thomas el al. Canberra: Austral. Nat. Univ., 17-38. 1976. Tongue-root and register in Mon-Khmer. In Auslroasiatic Stud. 1. (Oceanic Ling. Spee. Pub!. 13). (eds.) P.N. Jenner, et al. Honolulu: Univ. Hawaii Press, 323-69. 1965. Proto-Chamic phonological word and vocabulary. PhD. Dissertation, Indiana Univ. 1977. Devoicing, aspiration, and vowel split in Haroi: evidence for register (contrastive tongue-root position). Papers in South East Asian Ling. 4 Chamic Studies (Pacific Ling. A. 48). Canberra: Austral. Nat. Univ., 87-104. 1971. Dictionnaire cam-vietnamien-franr,:ais: T]!-die7n Chiim-Vi~t Phiip. Centre culture! ciim. 121

PAGE 131

ASPECTS OF INTER-CLAUSAL RELATIONS IN KHMU Suwilai Premsrirat 0. Introduction This paper attempts to investigate certain mechanisms involved in the manifestation of inter-clausal relations in Khmu, 1 and will discuss the Khmu that is spoken in Huoyen village, Chiengkhong district, Chiengrai province, Thailand. This variety is also found widely spoken in the Park Bang area of northern Laos. Khmu is a Mon-Khmer language spoken by a group of hill tribesmen found primarily in northern Laos, northern Thailand-especially the area along the Thai-Lao border-northern Vietnam and southern China. They are believed to have previously occupied a larger area than at present, their arrival to this area predating that of Thai speakers and other more recent hill tribes such as Miao, Yao, Lisu, Karen, and others. A. General syntactic structure2 Words in Khmu are mainly monosyllabic, though disyllabic words are still numerous. It is possible that Khmu once had a very complex system of affixation even though nowadays many affixes are found only in fossilised form and have often lost their meanings. In text materials, although most of the items can obviously be defined as belonging to the natural word classes such as noun, verb, adjective, adverb, preposition, demonstrative, etc., there are a number of 'syntactic' particles that do not seem to have meaning by themselves but do have certain functional meanings. They have a possible word shape, Cv(C), but they are never stressed. A number of them occur before the main verb. Many seem to be semantically related to the verb morpheme while in other cases they are related to the whole clause or even to a higher level. Without a knowledge of functional meanings of the particles (PT) it is almost impossible to understand the subtlety of the expressions and thus to follow a conversation, story, or talk thoroughly, even though the meanings of all the major words are clearly known. In order to give a clearer view of this curious syntactic feature which may lead to the understanding of inter-clausal syntax in Khmu, an overview will be given here, including a discussion of the verb phrase. I. I am grateful to my Khmu friends who patiently gave me the data discussed here and for David Thomas' comments and suggestions. 2. Further information on Khmu structure (phonology and syntax) can be found in Suwilai Premsrirat (1987c). Detailed information on phonology (different dialects) and morphology will also be found in Smalley (1961) and Svantesson (1983). 123

PAGE 132

SUWILAI PREMSRIRAT The normal word order in Khmu is SVO. In case of emphasis the OSV order is very common. Examples of normal clause patterns are: Intransitive clause -k5:n jam3 child cry 'The child is crying.' Transitive clause m: pa? mah they eat rice 'They are having a meal.' Bi-transitive clause ?o? ?iwn ma? j011 te? kmu:l I give parents take money 'I gave my parents some money.' Semi-transitive clause ja? jat ta ka:1) grandmother stay at home 'My grandmother is at home.' Equative clause ?o? mah l<'ru: I am teacher 'I am a teacher.' Existive clause ta ka:1) ?ah h?far, ?ah s5?, ?ah sia'} at home have hen, have dog, have pig 'There are hens, dogs and pigs at home.' Quotative clause jo1J pra:k law j:Jh la? la? father say before say go well well leaving 'My father wished them a safe trip.' Modifiers (Mod.) in a noun phrase normally follow the head noun (HN). HN+Mod. s5? hiau pa:r (to:) kam nam kana:i dog black two (class.)* which big that 'those two big black dogs' However, it is noticeable that the reversed orders of clause pattern, (O)VS, and of noun phrase, Mod.+ HN, have also been found in text materials. (1) lam mah lam delicious rice delicious 'Have a good life.' ma:r salt 3. refers to the clear tense register of this Khmu variety, whereas the breathy lax register is unmarked. *=classifier (Ed.). 124

PAGE 133

Aspects of inter-clausal relations in Khmu (2) puh kma) fall rain (n.) 'It is raining.' The Khmu verb phrase normally consists of one or more main verbs (MV) and their modifiers (auxiliaries and preverbal adverbs) occurring mainly in the pre-verbal positions. The pre-verbal modifiers (preV) can be classified into 5 categories according to their positional occurrences away from the main verb. preV 5 preV 4 preV 3 preV 2 preV 1 MV PreV 1 indicates state of existence like /p+::m/ 'able', /jat/ 'to have been happening, still', /h6:c/ 'already', /cu?/ 'want to', /?ik/ 'not want to', /b:j/ 'slowly, gradually', /lak/ 'completely', /ph5:n/ 'never,' etc. PreV 2 indicates negative meaning like /pe/ 'not', /pl5?/ 'not any more'. PreV 3 indicates intention, determination and future action like /ce/ 'will'. PreV 4 indicates tense and aspect like /ha/ 'already (past action)', /n5:TJ/ 'still, will'. Some of the pre-verbals of category 1 and category 2 like /pi':'ln/, /h6:c/, /lak/, and /pl5?/ may sometimes be shifted to the post-verbal position. Examples of verb phrases containing pre V 1, 2, 3, and 4 functioning in the clause are na: ph5:n ?i':'lk pu:c 1 MV she never drink wine 'She has never drunk wine.' (3) sna: ha 4 r:J:t MV they (two) already arrive 'They have come.' (4) ?o? Pi pian .2l!_ taju? 2 1 MV I not able go to forest 'I did not have a chance to go to the forest.' (5) ?o? n5:v ce .2l!_ m:J: t kmu:l ka:l 4 3 MV I still will go take money before 'I still have to go to take the money first.' (6) n:J: ha 4 ce pl5? 3 2 pian ?{Jh ja:1J ki: 1 MV they already will not any more able do like this 'They will not be able to act like this any more.' 125

PAGE 134

SUWTLAT PREMSRTRAT PreV 5 indicates modality such as the actor's volition, obligation, necessity or ability in doing something. Since it involves a relationship between a causer which may be stated or unstated and the actor and the action it is more like a clause rank notion than a phrase one. In Khmu it contributes a good deal to the inter-clausal relationship. This category includes /tarJ/, /le/, /cen/, mak/, /kaj/ and /tha/.4 These will be dealt with in detail in the next section. Examples of Khmu utterances with various pre-verbal particles including the category 5 pre-verbals5 are (1) prap bh !as ?)? mak pj_ pian El!:_ 5 2 1 MV dress body quickly we afraid not able go 'Dress quickly otherwise we will not be able to go.' (2) naj ?o? law ce pj_ El!:_ ?o? l(!IJ_ ha ce 3 2 MV 5 4 3 a moment ago I say will not go I certainly already will pe j:Jh jt:? j:Jh ?o? le j:Jh 2 MV MV 5 MV not go you go I therefore go 'I said that I would not go so I should not go. But you will go so I will too.' (3) p:J: j:Jh ka:t ?z? l[!IJ_ ce n5:y.t!l!:_ 5 3 4 MV you go before I certainly will still go 'You go first and I will follow you.' It should be noted here that the preV 3 /ce/ which indicates the future and the speaker's intention and determination can also function at the inter-clausal level (as will be seen in the next section) whereas the preV 5 /tarJ/ which also indicates the speaker's determination or obligation (though in a higher degree) can be manifested at the clause level. B. Inter-clausal Relations In the context of speech, a semantic sentence represents a mm1mum speech act consisting of one or more predications (semantic clauses). A structural sentence is therefore composed of at least one main independent clause with or without subordinate clauses and occurs with a distinct illocutionary force (mood) which may be a statement, question or command. In Khmu, two clauses or series of clauses whose meanings are related are structurally joined by various devices. The formal linguistic devices used may be conjunctions, pre-verbal particles, and reversed word 4. Since these pre-verbal particles do not have any content meaning, the explanation here is presented without giving the meaning of each item. However, in the following examples the approximate meaning of each item has been attempted. 5. The category 4 pre-verbals can appear after the category 3 in certain cases. It is noticeable that consciously and off context the co-occurrence between the category 4 pre-verbals /ha/ and /n5:TJ/ and the category 3 pre-verbal /ce/ is considered as 'possible' but 'not quite right'. The /ce/ seems to be the main item. 126

PAGE 135

Aspects of inter-clausal relations in Khmu order in the subordinate clauses. Apart from this, there are structurally independent clauses that are just juxtaposed (strung together within a common intonation contour but without any formal conjunction markers). I shall discuss inter-clausal relationships in terms of temporal, logical, and compounding sequences. 1. Temporal sequence relations: these have an initial action followed by one or more succeeding actions. They are normally joined by the medial conjunctions /h6:c/ 'after that, already, then' /pa:m/ 'at the time, when, time', /ka:l 'before, in front of and /khni?/ 'after, at the back'. However, in certain contexts the succession of actions may appear without any conjunction. The structural manifestation of this kind of relation is: SVO (conj.) SVO (1) pe? mah h6:c ?Jh wiak eat rice then do work 'I will eat rice first then I will go to work.' (2) ?o? ?Jh rna: h6:c ka.j sih (3) (4) (5) I do rice field then come sleep 'When I finish working on the rice field I will come to sleep.' ka: j3h J1,a:m ?o? n5:1J sih he go when I still sleep 'He went while I was sleeping' ?o? mu:m ?am kli:l pa? mah I bath water before eat rice 'I take a bath before having a meal.' ?o? rah ta sih k"nf? ma? j:Jh ta re? I get out from sleep after mother go to field 'I got out of bed after my mother had gone to the field.' Sometimes the meaning of words chosen indicates the temporal sequence. (6) je? j:Jh ka:l ?o? ce j3h nam ta k"nf? you go before I will go follow at the back 'You go first. I will follow you.' In some contexts a series of verbs following one another indicates the sequence of actions that happen one after the other. They may or may not have the conjunction /h6:c/ in between. The /h6:c/ normally occurs before the last verb of the utterance. (7) pa:t plu? 0 chrth ( hb:c) kbh cut off leg remove flesh then chop 'Sever and bone the leg and then chop it up.' 127

PAGE 136

SUWILAI PREMSRIRAT 8. pok 0 pliut 0 pen tJ:n 0 trafl 0 wah ]7,t::r cut off cut short to be piece cut square split small ]7,t::r ( ho:c) h5:c small then cut smooth 'Cut down (a tree) cut it into (short) pieces, smooth off the rough edge, split it into small pieces, then smooth them.' The pre-verbal particle category 5 /tha/ or its variants /tada-sa/ has also been found optionally co-occurring with the pre-verbal category 3 /ce/ and the conjunction /h6:c/ in the following clause to indicate that the event happens after another. Its structure can be manifested as SVO (h6:c) S (tha) ce VO (I) ?o? rJ:t /or jJ? pa: ho:c ?o? (tha) ce jJh pe? mah I come talk with you after that I then will go eat rice 'I will come to talk to you first, then I will go to eat rice.' (2) kma? ha phrial) ho:c ?o? (tha) ce jJh ?{Jh wiak rain already stop after that I then will go do work 'When the rain has already stopped I will then go to work.' (3) ?o? n5:1J pa? mah ho:c ?o? ( tha) ce hian tJ: I still have to eat rice after that I then will learn continue I have to have a meal first then I will continue learning.' ( 4) ?8? mu:m ?om ho:c ?o? ( th a) ce jJh sih I bath water after that I then will go sleep 'I will go to sleep after I have taken a bath.' 2. Logical sequence relations: these will be discussed mainly in terms of various 'conditional' sequences that are different in the degrees of determination, possibility, certainty, expectation and counter-expectation, including various shades of overtone. This type of sequence normally consists of a main clause stating 'condition', 'cause' or 'environment' and one or more dependent clauses stating 'consequence'. Notions like 'causal', 'resultant' and 'co-varying' will also be included in the discussion of this type of relationship. 2.1. Conditional relations: six different pre-verbals /ce/, le/, /ta[J/, cen/, /mak/, and /kaj/ are found attached to the consequence clause distinguishing six different types of conditional relations. The first particle has been discussed as pre-verbal category 3, whereas the last five belong to category 5. They can be manifested structurally as SVO S preV VO [ t1=[J1 cen mak kaj 128

PAGE 137

Aspects of inter-clausal relations in Khmu 2.1.1. /ce/ normally occurs within a clause to indicate futurity and the actor's determination to do something. Occurring in the consequence clause, it indicates that the action it is attached to will normally happen as the possible consequence of the preceding event. It is the most common type of conditional relationship. There is more than one way of manifesting this type of conditional relationship. Apart from the fact that the verb in the consequence clause is introduced by /ce/, the conditional clause may be introduced by the Lao loan conditional conjunction /khan/ or the Thai loan /tha:/, both meaning 'if. The structural manifestation of this relation can be summarised as: (conj.) SVO S preV VO (1) (k"an) ?o? pup r6.j ?o? ce tar If-I meet spirit I will run 'If I see the spirit I will run.' (2) (khan) kma? rxt 1J5? ce pian woh If-rain arrive paddy will get plenty 'If it rains we will get plenty of paddy.' (3) (khan) ?o? pe ?{Jh wlak pomma.j ?o? ce cu:r If I not do work the second day I will go down after tomorrow r:J:t rac pa: arrive visit you 'If I do not work on the second day after tomorrow I will go to see you.' (4) (tha.-J ?o? ?ah sna:t ?o? ce PIJI si:m IfI have gun I will shoot bird 'If I have a gun I will shoot the birds.' (5) (tha.-J pa j:Jh ?i? ce j:Jh dt If you go we will go together 'If you go so will we.' Some speakers put the word /?uan/ 'give' before the first clause of a sentence with /ce/. (6) ?{wn mar t6k m: m: ce ha:n give snake bite then they will die 'If they are bitten by a snake they will die.' (7) ?uanjolJ rxt ?o? ce j:Jh give father arrive I will go 'If my father comes I will go.' (8) ?uan ?o? ?ah pnfr ?o? ce ti:r mian si:m give I have wing I will fly like bird 'If I had a wing I would fly like like a bird.' 129

PAGE 138

SUWJLAI PREMSRIRAT 2.1.2. /le/ indicates that the action it is attached to has naturally happened or has to happen as the consequence of the preceding action. It is information which is being reported to somebody. (1) ?o? mu:m ?6m h6:c ?o? le j:,h sih I bath water already I therefore go sleep 'I go to sleep after taking a bath.' (2) ma? cth 'lo? 'lo? le ja:m mother scold me I therefore cry 'My mother scolded me so I cried.' (3) hie 1J5? ?uan m:Jk ho:c k_ ku:m crush paddy cause small piece already therefore winnow 'Crush the paddy until it becomes small pieces then winnow it.' (4) }OIJ ka: 'lik ?fak pu:c ka: k_ tt:I) te: pa.J1 father he not want drink wine he therefore make himself drunk 'My father did not want to drink wine so he pretended to be drunk.' (5) ?a: jt:? klam pian le mxt ?5:r j:Jh ?am6 init. part. you carry able therefore take lead go fin. part. 'If you can carry it then take it.' 2.1.3. /tarJ/ indicates that the action it is attached to certainly and obligatorily happens as determined by the event or circumstances which may be expressed in a clause, or even by a non-verbal behaviour or situation. It implies the speaker's expectation and determination. It may also co-occur with /cej. (I) n:J: }at pian ?j? !!!Jl jat pian they live able we certainly live able 'If they can stay so can we.' (2) l)ki:n 'lo? j:Jh ta ju? mi: ski: 'lo? !!!Jl ce j:Jh me? yesterday I go to forest today I certainly will go new 'Yesterday I went to the forest, today I will go again.' (3) ma wat !!!Jl wat init. part. fight certainly fight 'Let's fight.' ( 4) ten ta ki: !!!J}_ ])fan sit at here certainly able 'Have a seat here.' (5) 'lo? cu? lu.j 'lo? !!!Jl j:Jh b:J:k pa: I have a pain stomach I certainly go tell you 'If I have a stomach-ache I will tell you.' 6. fin. part. is abbreviated from 'final particle'. The other related abbreviation is init. part. (=initial particle). 130

PAGE 139

Aspects of inter-clausal relations in Khmu 2.1.4. /cen/ indicates that the action it is attached to is unavoidable and would logically happen as a result of a specific action or state. It normally has an overtone of 'sarcasm', 'unconcern', or sometimes 'annoyance'. In some cases it indicates that the action it is attached to is an unreal situation. It normally occurs with the final particle /(?u) ?Eh/. (I) jt:? na:JJ pri? jt:? cen pe ?ah ja:JJ kani ?t:h you know the right thing you definitely not do like this fin. part. 'If you know what is right and wrong you will definitely not do like that.' (2) ka.j cen ka.j pa: na jat ?ah mah ta ki: go back definitely go back you part. live do what at here 'If you want to go (home) then go. Why continue living here?' (3) ha:n cen ha:n ?t:h pa: na prial ?ah mah die definitely die fin. part. you part. alive do what 'If you want to die then die. Why stay alive?' (4) si:m ka: ?ah pnir ka: ti:r cen pian ?u ?t:h (5) bird it have wing it fly definitely able fin. part. (jt:? pe ?ah pnir jt:? ti:r ce pian mah) you not have wing you fly will able what 'Birds have wings so they can fly. (You don't have wings, how can you fly?)' kt:? pah ja:1) kani? ka: cen la:c ?th you leave like this it definitely fin. part. lose 'You left it here like this so (of course) it got lost!' (6) S5:n, jt:? cen sih rah sruat ?ef1 j;1h ?ah wiak Sorn, you definitely sleep get up early fin. part. go do work t"a ce tan then will be in time Sorn, if you had got up early you would have been able to go to work in time.' (7) na: cen tar fas ?u ?t:h na: n5:1J nim she definitely run quickly fin. part. she still young 'She is still young, of course she can run quickly.' 2.1.5. /mak/ indicates that the action to which it is attached is undesirable, and will (probably) happen as a result of another action. It has the overtone of 'warning' or 'concern'. It may occur with the conjunction /hi?/ 'in a moment, otherwise'. (I) ha ta: !or re:fJ (hf?) ka: mak mec already not talk loud otherwise he probably hear 'Don't talk loudly or he will hear what you say.' 131

PAGE 140

SUWILAI PREMSRIRAT (2) ta: /or ja:IJ bni nJ: mak mxr jt:? (3) not talk like this they probably fine you 'Don't talk like this or they will fine you.' kir ?fh ?i? ka.j !as mak puh thunder loud noise we go back quickly probably be soaked in kma? ka.j mak pe tan rain go back probably not be in time 'It is thundering so we have to go quickly, otherwise it will rain and we will not be able to go back.' (4) ta: jJh st:t hf? mak la:c not go alone otherwise probably get lost 'Don't go alone or you will get lost.' (5) ta: h5.j wd hf? mak ?oh not play knife otherwise probably cut oneself 'Don't play with the knife, otherwise you will cut yourself.' (6) ta: jJh krip ka? te: jJ? nJ: mak pi1J pJ: not go seize fish other people because they probably shoot you mak ha:n lc:1J probably die free 'Dont' go to catch others' fish or they will shoot you and you will die for nothing.' 2.1.6. /kaj/. In concessive or adversative sequence, the main independent clause and the subordinate clauses that follow are contrasted in content. The subordinate clause normally posits an exception or 'concession' to that which was or will be stated in the main clause. In Khmu this relationship is expressed by the occurrence of the pre-verbal particle /kaj/ in the subordinate clause. /kaj/ indicates that the event it is attached to happens in a contra-normal sequence, as an unexpected consequence of the preceding event, or even as a kind of conflict with the condition. (I) ?e: jat ta kulJ jt:? fsgj ?ah kmu:l woh ?o? kam init. part. live at village you become have money a lot I who jJh Cll.'IJ fsgj pe ?ah go be employed become not have 'How come you who stay at home are rich but I who go to work outside am not?' (2) bJ:k ?uan pa: jJh ?fJh wiak hb:c pa: fsgj jat ?fJh mah tell give you to do work already you become stay do what 'I told you to go to work. Why are you still here?' (3) m: jat pian ?i? fsgj jat pe pian they live able we become live not able 'They can stay; how come we can't?' 132

PAGE 141

Aspects of inter-clausal relations in Khmu (4) mi: ski: ?o? pe ?ah wiak ta su? ?o? '5!!1 ?!Jh wiak ta ki: today I not have work there I become do work here 'I come to work here because there was not any work over there.' (5) skja? ?o? pha:k kmu:l ?uanj:J? ka: ?uan a long time ago I ask to take money give to him give ka: ?ufm j:J? JolJ hb:c ka: '5!!1 pe ?uan j:J? }olJ '5!!1 him give to father then he become not give to father become j:Jh ?uanj:J? tmb? te: go give to wife his 'A long time ago I gave him the money (asking him) to give (it) to my father but he didn't. He gave it to his wife instead.' (6) ka: ce srblJ k5:n te: ku:t ro:l)rian ka: '5!!1 law he will send child his enter school he become say pe ?ah ma? p5:s ht:m te: not have who look after younger sibling his 'Instead of sending his child to school he said that there was no one else to look after his younger sibling.' 2.2. For 'causal' sequences the conjunction /j::J?/ 'because' is used between the consequence and its conditional cause. The structural manifestation of this type of relationship can be written as: SVO conj. SVO (1) ka: ha:n E!_ mar tok he die with snake bite 'He died of snake bite' (lit. 'The snake bit him to death') (2) 0mu? hial) E!_ puh wa:r Khmu black because to be in the sun hot 'The Khmu are dark because they work in the sun.' (3) na: tal) ?ik }e E!_ Jt:? hii ?ah tmb? she certainly not want you because you already have wife 'She must not want you because you already have a wife.' (4) ?o? j:Jh taju? E!_ te: ce ?{Jh h?{J? I go to forest because myself will do firewood 'I will go to the forest to get the firewood.' 2.3. Co-varying sequence or a sliding-scale conditional sequence consists of two clauses that are grammatically balanced but semantically opposing, with a free variable and a conditioned variable. It indicates the increasing degree in the event of the second clause. The conjuctions normally used are the Thai loan /j+::ilJ ... j+::iIJ/ and the Lao loan /hE:IJ ... hE:IJ/. The structural manifestation of this type of relationship can be written as: conj. SVO conj. SVO 133

PAGE 142

SUWILAI PREMSRIRAT (1) jiay ?{Jh jiay ?ah kmu:l woh the more do the more have money a lot 'The more (he) works the more money he has.' (2) ht:y trket ht:y cu? mpo1J the more think the more pain head 'The more I think the more I have a headache.' (3) ht:y l1JiJi ht:1] 1]:J? the more dark the more fear 'The darker it gets, the more I feel afraid.' 2.4. In certain conditional constructions the subordinate clause which states the result has the normal clause order reversed. svo ovs (1) ?o? sr?e:1J ma? jo1J lian ?om mat I miss mother father come out water eyes 'I miss my parents so much that the tears fall.' This characteristic has been extended to other sequences such as certain contrastive compounding (which is treated in the next section) and also to an immediate answer to the question in a question-answer sequence. (2) ma? ce ra: sntt:h ra: ?i? who will clean dishes clean we 'Who will do the dishes? We will.' (3) ma? ru1J pu:c ka ki: lam lak pu:c ru1J ?o? who cook make wine this delicious a lot wine make I 'Who made this delicious wine? I did.' (4) ma? pi1] s?:J:1] s?x1J pi1J ?o? who cut down tree tree cut down I 'Who cut the tree down? I did.' However, it must be noted that this kind of syntactic reversal occurred only in certain contexts. The effect is probably topicalisation. 3. Compounding sequence relations: these refer to the co-ordination of two independent clauses acting as a single unit in the higher grammatical level or sentence level. I shall discuss this type of relation in Khmu in terms of additive, contrastive and alternative sequences. There is no formal conjunction between clauses. The clauses are just juxtaposed. However, they normally occur within a common intonation contour. The structural manifestation of this relation is svo svo However, in contrastive sequence the subordinate clause may have a reversed order of the elements. svo ovs / ovs svo 134

PAGE 143

Aspects of inter-clausal relations in Khmu 3.1. Additive compounding consists of clauses with only either the subject, predicate or the object in contrast. (1) na: i:Jh de: ;a:m di;: she go also cry also 'She was crying while going.' (2) ta kill] ?o? ?!Jh re? ?!Jh pra:; liay s5? liay siay at home I do field do trap feed dog feed pig 'At home I worked on the rice field making traps, feeding dogs and pigs.' (3) nak y5? seh nak kla:y seh heavy paddy put in heavy rock put in 'It is heavy because both paddy rice and rocks have been put in there.' (4) na: ku? m: woh na: pe 16JJ m: she love them a lot she not forget them 'She loves them a lot and she will not forget them.' (5) co:n la:c 05:y ia:n nx ha:n thief steal thing kill them die 'The thieves took the things and killed them.' 3.2. Contrastive compounding consists of two contrasting clauses m subject and predicate (verb or object). (1) jt:? nam ?o? nt:? you big I small 'You are big (but) I am small.' (2) ?o? tuk pa: rma:y I poor you rich 'I am poor (but) you are rich'. (3) ?o? ht:t ka: pe ;as I call he not answer 'I called (him) (but) he did not answer.' (4) ?o? ce i:Jh ta mok na ce i:Jh ta ?bm I will go to mountain she will go to water 'I will go to the mountain (but) she will go to the river.' (5) mt::w pe pty s?xy khmit? pty s?:J:fJ Meo not cut down tree Khmu cut down tree 'The Meo do not cut the tree down (but) the Khmu do.' (6) ?o? pe cha:; t!J:m ka: cha.'j I not know how to sing he know how to 'I do not know how to sing Khmu traditional song (but) he does.' 135

PAGE 144

SUW!LA! PREMSR!RAT The order of the subordinate clause in the last two examples is naturally reversed in the text. h (5)a mc::w pe ptlJ s?J:IJ s?J:IJ ptlJ k mu? Meo not cut down tree tree cut down Khmu 'The Meo do not cut the tree down (but) the Khmu do.' or (5)b s?J:IJ ptlJ ~mu? mc::w pe ptl] tree cut down Khmu Meo not cut 'The Khmu cut down the tree (but) the Meo do not.' (6)a ?o? pe cha.j ta:m cha_-j ka: I not know how to sing know how to he 'I do not know how to sing Khmu traditional song (but) he does'. 3.3. Alternative compounding consists of two clauses with alternative predicate or alternative participants. (1) pa: ce jJh (pa:) ce pe jJh you will go you will not go 'You will go or not.' (2) pa: ce jJh pa: ce jat tal] pian pa? pii:r you will go you will stay certainly able both two 'You will go or you will stay.' (3) pa: /or 'mo.j cum' tal) /a? 'mo:j ku:' tal] fa? you say 'mo.j cum' certainly right 'mo.j ku:' certainly right 'You can say either 'mo.j cum' or 'mo.j ku:'.' (4) !or rJIJ !or h5:j speak real speak play 'Are you just joking or not?' (5) pa: ?ah kmu:l woh pa: ce wt::t lot pa: ce ro? ka:I) you have money a lot you will buy car you will build house 'If you have a lot of money will you buy a car or will you build a house?' The correspondences between various types of semantic relations and syntactic devices is summarised in Fig. 1 on the following page. 136

PAGE 145

Aspects of inter-clausal relations in Khmu semantic relations syntactic devices I 2. logical +.-2.2 causal ~2.3 resultant---,.-<---~~---3. reversed word order '--2.4 co-varying ,-3.1 additive 3. compounding--,-3.2 contrastive L...._------'=:c~ 4. juxtaposition 3.3 alternative Fig. I Conclusion Inter-clausal relations in Khmu have been discussed both semantically and structurally. Three main types of semantic relations, temporal, logical and compounding sequences, including various sub-types, have been discussed. The syntactic devices used in manifesting these semantic relations are conjunctions, pre-verbal particles, reversed word order and juxtapositions or a mixture of these. It is obvious that in Khmu conjunctions are not in predominant use, only a few examples having been found. Some are Thai or Lao loanwords, for instance, those initial, or initial and medial, conjunctions discussed in 2.1.1 and 2.4. The common Khmu medial conjunctions are /h6:c/ 'then, after that', fr:,?/ 'because', and /hi?/ 'otherwise'. It is noticeable that their occurrences are not obligatory. They may be left out, or some other way of expressing the relationship may be taken instead. They also occur normally in positions of other word classes such as /j::ih ho:c/ 'to have already gone', /ck6? hi?/ 'this coming evening', /pa? jJ? ti?/ 'to eat with hand', /j::ihj;,? jJ?/ 'to go with others'. The reversed word order has been found used mainly in special or emphasised cases. The use of pre-verbal particles and juxtaposition is quite clear and obligatory. Pre-verbal particles are mainly used for subordinative constructions manifesting temporal and various logical (conditional) sequences, whereas juxtaposition is mainly used for co ordinative constructions manifesting various compounding sequences. The systematic use of different pre-verbal particles for distinguishing different conditional relations in Khmu is unique. It also shows certain things about how the speakers of this language view and systematically categorise their 'conditional (cause and consequence)' experiences. The young Khmu consider the pre-verbal particles as the conservative 137

PAGE 146

SUWTLAJ PREMSRTRAT forms of Khmu. Those who work in urban areas among the Thais tend to lose them gradually and unconsciously adopt the Thai pre-verbal particles /k5:/ 'then' /t5fJ/ 'must' and final particles /s'i./ and /ki:j/, even though they are not exactly equivalent, whereas the people at home, especially the old and the women, still make quite a lot of use of the old forms. They sometimes adopt Thai conjunctions, /tha:/ 'if, /phr5? (wa:)/ 'because', /IE?/ 'and', /t&:/ 'but', and /d:/ 'or'. However, they said that they still found it difficult and unnatural to express the contrastive, alternative or additive compoundings using the conjunctions /t&:/, /rt/ and /IE?/, respectively. Looking at the text materials of another Khmu variety which is spoken in the Nam Pan, Huay Moy, and Nam Lu villages in Chiengklang district, Nan province, Thailand, it is noticeable that the syntactic constructions in that area have been simplified quite a lot and become more like those of Thai. However, unconsciously, in certain contexts, most of the pre-verbal particles mentioned above are still retained, though the pronunciation may be slightly different. The exception is /cen/ which is generally replaced by / ce/, the equivalent of the Thai /ea?/ which indicates the speaker's intention and future action. This pre-verbal seems to be very common and can sometimes replace other pre-verbals such as /mak/ and /tafj/. REFERENCES Gregerson, K. Manley, T. 1979. Predicate and argument in Rengao grammar. Arlington: Summer Inst. Ling. and Univ. Texas. 1972. Outline of Sre structure. (Oceanic Ling. Spee. Publ. 12). Honolulu: Univ. Hawaii Press. Nguyen Dang Liem 1975. Cases, clauses and sentences in Vietnamese. (Pacific Ling. B, 37.) Canberra: Austral. Nat. Univ. Peck, C. 1984. A survey of grammatical structure. Arlington: Summer Inst. Ling. and Univ. Texas. Pike, K. L. 1977. Grammatical analysis. Norman, Oklahoma: Summer Inst. Ling. & Pike, E. V. Smalley, W. A. 1961. Outline of Khmu? structure. New Haven: Am. Or. Soc. Smith, K. 1979. Sedang grammar. (Pacific Ling. B, 50). Canberra: Austral. Nat. Univ. Premsrirat, S. 1986. Lokkathat naj Phasaa Khamu (World view in Khmu), J. Thammasat Univ. 1: 94-113. 1987a. Khmu medical conversation book. Nakhorn Pathom: Mahidol Univ., Institute of Language and Culture for Rural Development. 1987b. Praj medical conversation book. Nakhorn Pathom: Mahidol Univ., Institute of Language and Culture for Rural Development. 1987c. Khmu, a minority language of Thailand. In Papers in Southeasl Asian Ling. 10. (Pacific Ling. A, 75). Canberra: Austral. Nat. Univ. 138

PAGE 147

Premsirat, S. Svantesson, J.-O. Thomas, D. Aspects of inter-clausal relations in Khmu 1988. Khmu life style in 50 conversational lessons. Nakhorn Pathom: Mahidol Univ., Institute of Language and Culture for Rural Development. 1990. (Forthcoming) Thai-Khmu-Eng/ish Dictionary. 1983. Kammu phonology and morphology. (Travaux Inst. Ling. Lund XVIII). Lund: Gleerup. 1977. Chrau Grammar. (Oceanic Ling. Spee. Pub!. 7). Honolulu: Univ. Hawaii Press. 1975. Notes and queries on language analysis. (Language Data, Asian Pacific Series, 10). Huntington Beach, California: Summer Inst. Ling. 1982. The Vietnamese preverb auxiliary system. In Linguistics across continents (ed.) A. Gonzalez & D. Thomas. 1983. An invitation to grammar. Bangkok: Mahidol Univ. 139

PAGE 149

AN INSTRUMENTAL STUDY OF CHONG REGISTERS1 Theraphan L. Thongkum 0. Introduction The Chong language belongs to the Pearic branch of the Mon-Khmer language family (Thomas & Headley 1970; Diffloth 1974; Huffman 1976, 1985; Headley 1977, 1978, 1985). Both amateur and professional linguists who have worked on Chong (eg. Baradat 1941; Martin 1974; Huffman 1985; Gainey (personal communication); Suphanphaiboon 1982) seem to recognize the 'glottal feature' or 'glottalization' which occurs in some Chong words. Some of them (Huffman, Gainey, and Suphanphaiboon) hear phonation types-normal voice vs. breathy voice, etc.-and pitches. On the basis of linguistic descriptions, there is no doubt that Chong is a register (R) language. In 1983, Gerard Diffloth and I made several linguistic field trips to Chong communities in Makham District, Chanthaburi Province; we also visited Chong villages in Pong Nam Ron district, Chanthaburi Province, in Bo Rai District, Trat Province. During 1983-85, two female Chong informants from Krathing Village, Phluang Sub-district, Makham District, were brought to Bangkok many times for the purpose of checking language data and making high quality recordings and instrumental studies. 2 In spite of my training as a phonetician and my knowledge of MonKhmer languages such as Bru, Nyah Kur (Chao Bon), Mon, Kui (Suai), and Mia Bri, I still think that Chong is very exotic. In my opinion, the cause of the complexity lies in the process of Chong becoming a tone language. In fact, some dialects of Chong, such as the one spoken in Chamkhlo' Village, Takhianthong Sub-district, Makham District, have already become tonal: presyllables are dropped; phonation types are less prominent and in some cases disappear; and pitch differences can be heard clearly, especially in slow speech. Our Chong informants also describe their language as having high, higher, mid and low tones. The instrumental analysis presented in this paper is based on the speech of four Chong informants from Krathing Village. For the sake of convenience, they will be addressed as MA (first male speaker), MB (second male speaker), FA (first female speaker), and FB (second female I. This report is part ofmy research project on 'Registers in Chong, Mon and Kui (Suai): a phonetic study'. I should like to express my gratitude to Chulalongkorn University for providing the research funds and to thank Professor Arthur S. Abramson for his valuable advice. Many thanks go to Jerry W. Gainey, Suraphon Wongthongwatthana and Sitthichai Sisukhon for their assistance in many different ways. 2. Gerard Diffloth and I have made an agreement that he will be responsible for the comparative and historical aspects of Chong whereas the phonetics will be my responsibility. 141

PAGE 150

THERAPHAN L. THONGKUM speaker). The Krathing dialect was chosen because its register phenomena suit my major interest-the acoustical measurements of the register complexes which involve several phonetic parameters. Moreover, the place where it is spoken is easy to reach, and the villagers are also very co operative. The phonetic instruments used in the study are as follows: -Kay Sono-Graph 6061-B; -Fundamental Frequency Meter, type FFM 650 (F-J); Intensity Meter, type IM 360 (F-J); Electro-glottograph, type EG 830 (F-J); Electro-aerometer, type EA 510/4 (F-J); -Mingograf 34 T (Siemens AB). I Definitions of 'register' The term 'register' has been used in many different ways. As a result, many definitions can be found in the literature depending upon who uses it-music and voice specialists, phoneticians, linguists, or language teachers. 1.1. Voice register Music and voice specialists describe the rate of vocal-fold vibration in terms of registers. Garcia (1855) recognises three voice registers or ranges of pitch: the chest register, the mixed or middle register, and the head register. More often, only the chest and head registers are used. Others have tried to clarify the problems of voice-register terminology: 'The terminology with regard to voice pitch level, i.e. "registers", suffers from the existence of an abundance of terms and an ambiguity of their use.' (Morner, Fransson & Fant 1963: 18). They therefore define a register by means of its range on the musical scale, suggesting five basic registers, namely: deepest range, deep level, mid level, high level, and highest level. The approximate ranges and boundary limits of these registers are illustrated, and some synonyms are listed, for example: Deepest range Deep level Mid level High level Highest range Rayon profond Chest register Falsetto I Falsetto II Pipe register Chest voice Medium Falsetto voice Flute Long-reed Mid voice Short-reed Whistle Site grave Site moyen Site aigu Rayon eleve A particular mode of vocal-fold vibration is usually confined within a pitch range. Zemlin says that when an individual reaches the upper limits of his normal pitch range, the mode of vocal-fold vibration may be modified. He states: This modification of the mode of vocal-fold vibration may be regarded as an operational definition of voice register. Thus, as a person transcends the limits of a particular vocal register, the voice may undergo an abrupt modification of quality. This vocal quality is often the primary characteristic of voice register. (Zemlin 1968: 193) 142

PAGE 151

An instrumental study of Chong registers Also, according to Zemlin (op.cit.: 206-9), besides normal or acceptable vocal quality, there are three types of unacceptable vocal-quality: breathiness (incomplete blockage during the closed phase results in a continuous flow of air during the entire vibratory cycle), harshness (irregular vocal-fold vibration), and hoarseness (combination of the features harshness and breathiness). 1.2. Register vs. Contour When Pike discusses the types of tone languages, he defines registers as contrastive level phonemes. A language can have two, three, or four registers. The labels for two-, three-, and four-register systems are as follows: Language A Language B Language C high-high-high-midmid-norm-low-low-low-Thus, a register tone language is a tonal language that has a register-tone system, and a contour tone language is the one in which gliding tonemes are basic to the system (Pike 1948: 5-9). 1.3. Designative register Register can also be regarded as part of tone of voice. In some languages, changes of register may be used to express different emotional states and attitudes of the speaker. The same register might not carry the same affective indices in different cultures (Abercrombie 1967: 101). This is a paralinguistic use of register. 1.4. Register vs. Tone Register as used by Henderson (1952) is a phonological concept. It is a cover term not only for laryngeal activity but also for a cluster of activities in the vocal tract. She states: The Cambodian 'registers' differ from tones in that pitch is not the primary relevant feature. The pitch ranges of the two registers may sometimes overlap, though what I shall call the Second Register tends to be accompanied by lower pitch than the First Register. The characteristics of the first register are a 'normal' or 'head' voice quality, usually accompanied by a relatively high pitch. The characteristics of the second register are a deep rather breathy or 'sepulchral' voice, pronounced with lowering of the larynx, and frequently accompanied by a certain dilation of the nostrils. Pitch is usually lower than that of the first register in similar contexts. The register of a syllable is closely bound up with the vowel nucleus of that syllable, the two being mutually interdependent in a way that will be shown hereafter. 143

PAGE 152

THERAPHAN L. THONGKUM In sentences the word registers are modified according to intonation and by emotional factors. Register may be used, as in many other languages, to express emotion, and when this happens the emotional register may overlie the lexical register, much as in many tone languages intonation may overlie lexical tone. (Henderson 1952: 151-2). This new concept of register which was introduced into the field of South-East Asian linguistics by Henderson was adopted by Shorto (1966) and also by linguists of later generations, including myself. Abercrombie established the term. In his book Elements of general phonetics, besides mentioning Henderson's work on Cambodian, he also points out (I 967: 101-20) that Gujerati, Danish, some dialects of Scots Gaelic, and various West African languages make use of register contrast. He finally concludes: 'It is to be expected that future research will disclose many more examples of the linguistic use of register.' (op. cit., 102). In this sense, a register language may be defined as a language that has a lexically contrastive register complex (a combination of vowel quality, pitch, phonation type, etc.), whereas a tone language has lexically contrastive pitch.3 2. Brief sketch of Chong phonology Consonant system Initial consonants Consonant clusters Final consonants Vowel system Monophthongs Diphthongs Register system p t c k ? ph th eh kh b d m n J1 lJ s h w r I j pr tr kr phr khr pi kl phi khl ml mr kw ptck?hmnJll)Wj ieEUIYauo::, ii ee EE UIUI YY aa uu oo ::,::, J:l Ul:l U:l Static registers R1 (clear voice, higher pitch, more open or on-gliding vowel) R3 (breathy voice, lower pitch, raised vowel) Dynamic registers R2 (clear-creaky voice, high-falling pitch, more open vowel) R4 (breath-creaky voice, low-falling pitch, raised vowel) The co-occurences of registers with initial consonants, final consonants and vowels are given in Charts I, 2 and 3 below. 3. Di moth does not like this definition. He thinks that a register language should be defined as 'a language that has contrastive phonation type' (personal communication). Certainly, this definition is more specific. In my view it is also problematic. Both pitch and phonation type can be heard clearly in all register languages that I have come across. Without doing perception testing, I do not think that we can make a definite claim. Native speakers of register languages might hear both or more phonetic features at the same time. 144

PAGE 153

p t C k ? ph th eh kh b d m n J1 1J s h w r l j pr tr kr phr khr pl kl phl khl ml mr kw R1 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + --+ R2 + + + + + + + + + --+ + + + + + + + + + + -+ -+ + + + + R3 + + + + ------+ + + + --+ + + + + -+ --+ + --+ + + R4 + + + + -----+ + + + --+ + + + + -+ ---+ --+ + -Chart I: Co-occurrence of registers with initial consonant. p t C k ? h m n J1 1J w j R1 + + + + + + + + + + + + R2 + + + + --+ + + + + + .I::,. R3 + + + + + + + + + + + + V, + + + + + + + + --+ + Chart 2: Co-occurrence of registers with final consonants. e E w y a u 0 :) ii ee EE ww YY aa uu oo :,:, fa w~ u~ R1 + + + + -+ + + + + + + + + + + + + + + + R2 + -+ + -+ + -+ + + + + + + + + + R3 + + + + + + + + + + + + + + + + + + + + + + + -+ + + + + -+ + + -+ + + + + Chart 3: Co-occurrence of registers with vowels.

PAGE 154

THERAPHAN L. THONGKUM 3. Acoustic measurements In this section, an acoustic analysis of Chong vowels in respect of the formant frequency, power spectra, fundamental frequency, duration, and overall intensity will be given. The results of the measurements will indicate the characteristics of the register complex in Chong. Wide band spectrograms were used for measuring formant frequencies and power spectra, and narrow band spectrograms for fundamental frequencies. 3.1. Formant frequency The frequencies ofF I and F2 of the following 17 clear vowels, 15 clear-creaky vowels, 18 breathy vowels, and 15 breathy-creaky vowels were measured: Clear ( R1 ) Clear-creaky ( R2 ) Breathy ( R3) Breathy-creaky ( R4) e e e E E E w w w w y y a a a a u u u u 0 0 0 :) :) :) 11 11 ii 11 ee ee ee ee EE EE EE EE WW WW WW yy yy yy yy aa aa aa aa uu uu uu uu 00 00 00 00 :):) :):) :):) :):) Five test words said in isolation by the two male speakers (MA, MB) were used for each vowel: there were altogether 650 ((65 x 5) x 2) test tokens. The mean values of F1 and F2 were plotted separately for each speaker on vowel charts (see Figures 1, 2, 3 and 4). In general, the results of the measurements confirm what can be perceived auditorily: breathy and breathy-creaky vowels are higher than clear and clear-creaky vowels. The two male speakers do not seem to diphthongise their clear vowels. This finding supports Gregerson's (1976) hypothesis that in most Mon-Khmer languages first register (clear) vowels which are produced with retracted tongue-root are always more open than second register (breathy) vowels which are produced with advanced tongue-root. The four vowel formant charts exhibit obvious patterns: clear voice and semi-clear voice vowels of the first and second registers are more open than the breathy voice and semi-breathy voice vowels of the 146

PAGE 155

An instrumental study of Chong registers 2500 2000 1500 I 1000 ~<--1-ol 2 Hz Fig. I: I ,, I ,, ,_+.,,.., Formant frequencies ( F1 and F2 ) (speaker MA) + clear vowel (R1 ) A breathy vowel (R3 ) 1 200 F, ,. I I w Ji,.\ + I ,.. -,1 / ( y 4-'' / ..,. '-.1 / '( Ji,. u \ ''\ +., ,_/ r;_-' I o \ 'I '\ +I /-;.. \ / I / '/ \ + / -300 l 400 f500 600 f700 f800 of 1st, 2nd, 3rd and 4th register short vowels clear-creaky vowel (R2 ) breathy-creaky vowel (R4 ) 2500 2000 1500 1000 ""<---F2 H 600 2 EE 'I + I I ,, / Ji,. \ I I / I I/+ I / aa/ / I / ---~ / ,. ... _, + I / / \ ._:,'l ~--(. ...... \ uu + J ---J / --I A\ I oo / I / /1 200 F1 > 3()() l 400 500 600 700 800 -------------------------900 Fig. 2: Formant frequencies ( F1 and F2 ) ( speaker MA) + clear vowel (R1 ) A breathy vowel (R3 ) of 1st, 2nd, 3rd and 4th register long vowels clear-creaky vowel (R2 ) breathy-creaky vowel (R4 ) 147

PAGE 156

THERAPHAN L. THONGKUM 2500 2000 I 1500 I /;;: \ / / / / I E / I \ + ) ..... -' / ) / w ( / +,,. / \ / /' /. /" / '-,,,. / ..... / a Ji./ ,. I \. + / .... / iE: F2 Hz 1000 700 200 F1 I /-300 l I "\ I I u I I I 400 I / ..... / /. "/ 500 I+ o I ,. /'i) I // / ,,.., 600 / 0 / I I + / I _, 700 800 ,__ ___________________ ___,900 Fig. 3: Formant frequencies ( F1 and F2 ) of 1st, 2nd, 3rd and 4th register short vowels 2500 (speaker MB) + clear vowel (R1 ) .._ breathy vowel (R3 ) 2000 / \ / ii / I + ,,. I I ,. I /I"\ ,..., 1+ ', \ ee '-!_J_.J 4' I I /EE I I \ \ + I ..... ., 1500 I clear-creaky vowel (R2 ) breathy-creaky vowel (R4 ) 1000 I ---F2H 700 z 200 F1 -, l ,~ -11 -300 -./ / i":.: I ~oo I -400 f \ ... / / / 00 / I+ I I I \~ / 500 600 700 800 ____________________ __.900 Fig. 4: Formant frequencies ( F1 and F2 ) (speaker MB) of 1st, 2nd, 3rd and 4th register short vowels + clear vowel (R1 ) .._ breathy vowel (R3 ) 148 clear-creaky vowel (R2 ) breathy-creaky vowel (R4 )

PAGE 157

An instrumental study of Chong registers third and fourth registers. However, it is still too early to accept Gregerson's claim. Although it works well for Chong, unfortunately it cannot explain the register phenomena in Nyah Kur and Kui (Thongkum 1982, 1985). 3.2. Power spectra Kirk and others point out that the power spectra enable phoneticians to quantify the relative amount of energy in different harmonics. For their study of phonation types in Jalapa Mazatec, they measured the difference in dB between the intensity of the fundamental and the intensity of the first formant. They conclude, 'There is considerable variation from speaker to speaker in the three phonation types; but for each speaker on this measure the value for creaky voice is less than that for modal voice, and the value for modal voice is less than that for breathy voice' (Kirk et al. 1984: 109). Following their recommendations I did the same measurements for Chong. For the measurement of power spectra, each of the two male speakers said 325 test words. The results of the measurements may be summed up as in Table 1: Table 1: Mean values and standard deviation ( SD) of the relative amplitude of F0 and F1 ( in dB). Speaker MA Short vowel Long vowel Register R1 R2 R3 R4 R1 R2 R3 R4 Number of 40 30 45 35 45 45 45 40 test token Mean -8.09 -8.35 -8.27 -8.61 -9.18 -9.28 -8.86 -8.95 so 2.01 1.75 1.98 1.76 1.25 1.83 1.57 1.58 Speaker MB Short vowel Long vowel Register R1 R2 R3 R4 R1 R2 R3 R4 Number of 40 30 45 35 45 45 45 40 test token Mean -8.38 -7.99 -8.20 -8.82 -9.31 -8.54 -8.97 -8.83 so 1.98 2.21 1.89 1.92 1.37 1.69 1.44 1.67 It is unfortunate that the results do not meet my expectation; i.e. the measure does not seem to separate out successfully in that the value for modal voice is higher than that for creaky voice and less than that for breathy voice. Why? The reasons that I can think of are as follows: (i) The measurements were done by hand because I did not have a Kay digital sound spectrograph, and so they could be less accurate. (ii) As stated by Kirk (op. cit. 109), 'This measure can be used for comparing phonation types only in cases in which the vowels being 149

PAGE 158

THERAPHAN L. THONGKUM compared have similar formant frequencies; as the relative intensity of each formant is a function of its frequency.' It is true that in Chong the vowels of each register have specific quality as described in 4. l above. Moreover, R2 and R4 vowels have dynamic or combined phonation types: clear followed by creaky phonation and breathy followed by creaky phonation; this can actually cause problems for the measure. 3.3. Fundamentalfrequency (F0 ) The word list used for F0 measurement consisted of 71 minimal or analogous sets; for example, Set 20 pwJ I 'to get pregnant' pum2 'rash' pu1J3 'the entrails of animals' pu1J4 'water melon' 284 meaningful items were said by each of the two female speakers (FA, FB). Narrow band spectrograms were made and measured at ten points starting from the onset to the end of vowel. For plotting the results of F 0 measurements, the 284 test words were divided into 20 sets based on register differences and different types of syllable:4 Clear Clear-creaky Breathy Breathy-creaky Number of Test Token CVN CVN CVN CVN 40(10x4) CVS CVS CVS CVS 16 (4x4) CVVN CVVN CVVN CVVN JOO (25 X 4) cvvs cvvs cvvs cvvs 88 (22 X 4) CVH CVH 12 (6x2) CV? CV? 8 (4 X 2) The mean values of F0 (in Hz) are shown in Figures 5 and 6. From the data presented there, the following points can be made: (i) In CVN, CVS, CVVN and CVVS types of syllable, clear-creaky (R2 ) vowels have the highest fundamental frequency. Only the clear part of R2 and R4 vowels could be measured because of the irregularities of the creaky part. However, it is possible to claim that R2 and R4 vowels have high rise-fall and high fall F0 contours, respectively. The insertion of laryngealisation or creakiness seems to be the cause of falling F0 contour. This may be an explanation of how falling tones are acquired in some tone languages. The narrow band spectrograms of sets 5, 16, 17, 18 and 32 (speaker FB) illustrate the absence of creaky phonation. Thus, clear-creaky (R2 ) vowels become clear vowels with higher fall F0 contour, and breathy-creaky (R4 ) vowels become breathy-clear vowels with lower fall F0 contour (see Figure 7 below). 4. N: represents nasals and semi-vowels (m n J1 lJ w j); S: stops (pt ck); H: glottal fricative (h); and ?: glottal stop (?), respectively. 150

PAGE 159

Hz 230 210 190 170 150 130 Hz 230 210 190 170 150 130 Hz 230 210 190 170 150 ,,,. ,,,. CVVN --R2 ,,,. R1 'R3 DURATION 100% CVVS -..... 'R2 R -~-i R1 '-'\ \R DURATION 100% CVH R3 R1 3 An instrumental study of Chong registers Hz 230 210 190 170 150 / CVN --R2 ---130 .__ __________ Hz 230 210 190 170 150 DURATION 100% CVS 130 '-----------DURATION 100% Hz CV'I 230 210 190 170 ....._ ....._ -R3 150 130 '------------130 Fig. 5 DURATION 100% DURATION 100% Mean F0 values (in Hz) of 1st, 2nd, 3rd and 4th register vowels in 6 syllable types ( speaker FA) clear vowel (R1 ) breathy vowel (R3 ) 151 clear-creaky vowel (R2 ) breathy-creaky vowel (R4 )

PAGE 160

THERAPHAN L. THONGKUM Hz 260 CVVN -240 / / \ 220 R4 _:_.:_....:.....:.--:-:--200 1.-----180 160 ~---------DURATION 100% Hz 260 240 220 200 180 cvvs ---.... 160 '------------Hz 260 240 220 200 180 DURATION 100% CVH R3 / / R1 / 160 ~----------DURATION 100% Hz 260 240 220 200 180 CVN 160 '------------DURATION 100% Hz 260 240 220 200 180 CVS -..... / 160 '-----------DURATION l00% Hz CV'I 260 240 220 200 180 160 DURATION 100% Fig. 6 Mean F0 values (in Hz) of 1st, 2nd, 3rd and 4th register vowels in 6 syllable types ( speaker F B) clear vowel (R 1 ) --breathy vowel (R3 ) 152 clear-creaky vowel (R2 ) breathy-creaky vowel (R4 )

PAGE 161

Hz 180 160 140 120 100 -.... / -.. / An instrumental study of Chong registers ' -.. -.. -.. ---------R4 DURATION 100% Fig. 7: Mean F0 values (in Hz) of 2nd and 4th register vowels in CVVS syllable type when creaky phonation disappears and is replaced by falling F0 contour (speaker FB) (ii) Although breathy vowels can be perceived auditorily as having the lowest pitch, their Fa does not start low at all. According to the tracings (from 0% up to 30% of the vowel duration) breathy vowels can have even higher Fa than clear vowels. The point where the higher pitch of clear vowels and the lower pitch of breathy vowels can be differentiated is the Fa from 30% up to the end of vowel duration. (iii) There is a tendency for R 1 R2 R3 and R4 vowels in CVN and CVS types of syllables to have higher Fa than those in CVVN and CVVS syllable types. (iv) Only clear and breathy vowels can occur in CVH and CV? types of syllable. In CVH syllable structure, both types of vowel have rising Fa contour; however, breathy vowels do not always have lower Fa than clear vowels. For speaker FA, the Fa tracings of clear and breathy vowels do not exhibit much difference; i.e. both types of vowel have rising Fa contour. In CV? syllable structure, clear vowels have rising Fa contour with an abrupt fall at the end, whereas breathy vowels have falling Fa contour. 3.4. Duration The word-list used for Fa measurement was used again for the measurement of vowel duration and overall intensity. The test words were said by two female speakers of Chong. The audio recordings used as input were made in the recording studio of the Linguistics Research Unit (LRU) at the Faculty of Arts, Chulalongkorn University, and the results 153

PAGE 162

THERAPHAN L. THONGKUM of the measurement can be found in Tables 3 and 4. It is noticeable that breathy-creaky (R4 ) vowels seem to be shorter than the other kinds of vowels no matter what the types of syllable. However, there is one exception; i.e. in CVN syllable type, R4 vowels are somewhat longer.5 I have earlier commented that differences in duration caused by differences in phonation types might not be important when the language in question possesses distinctive vowel length (Thongkum 1985: 12). The results of the measurements of Chong vowels give me more confidence. Although breathy or murmured vowels in Gujerati (Fischer-10rgensen 1977) and Jalapa Mazatec (Kirk et al. 1984) have longer duration than clear vowels, we still cannot claim that it is a universal phonetic characteristic. Regarding clear-creaky (R2 ) and breathy-creaky (R4 ) vowels, the creaky part of R4 vowels is longer than that of R2 vowels, although the overall duration of R4 vowels is shorter than that of R2 vowels. This quality can also be perceived auditorily. The proportion (in percentage) of the duration of the first and second parts of vowel pertaining to dynamic registers (R2 and R4 ) is given in Table 2 below. Table 2: Mean duration ( in percentage) of the clear and creaky parts of the 2nd register vowels and of the breathy and creaky parts of the 4th register vowels .... SPEAKER FA +-' u, o CVN CVS CVVN cvvs R2 V y V y V y V y 51.95 48.05 71.43 28.57 53.55 46.45 57.61 42.39 R4 y y y y y y y y 47.89 52.11 62.50 37.50 51.79 48.21 52.10 47.90 .... SPEAKER FB +-' u, o CVN CVS CVVN cvvs R2 V y V y V y V y 52.72 47.28 68.84 31.96 53.08 46.99 57.60 48.40 R4 y y y y y y y y 51.58 48.42 58.33 41.67 49.23 50.77 45.25 54.75 5. Since the segmentation was done by hand without any computer aids, defects may have been caused by inaccurate segmenting. The beginning part of laryngealised or creaky voice nasals and semi-vowels could have been segmented as part of clear-creaky voice and breathy-creaky voice vowels. 154

PAGE 163

An instrumental study of Chong registers 3.5. Intensity Regarding overall intensity, clear vowels and clear-creaky vowels seem to have higher amplitude than breathy and breathy-creaky vowels. The loss of intensity is due to leaking glottis during breathy phonation (Fischer forgensen 1977: 119), and the acoustic energy is lost by the damping effect of the general relaxation of the muscles of the whole vocal system in lax voice (Laver 1980: 135). The intensity curves of clear and breathy vowels look similar (i.e. look more bell-shaped), whereas those of clear creaky and breathy-creaky vowels look different (i.e. more cone-shaped), which is due to a sudden drop of intensity when clear and breathy vowels become creaky caused by the abrupt closing of the vocal folds. The results of the measurements are given in Tables 3 and 4. 1-5 = overall vowel duration 1-3 = distance from the onset of vowel to the peak of intensity curve 3-5 = distance from the peak of intensity curve to the end of vowel 1-a = amplitude at the onset of vowel 2-b = amplitude at the half distance of the onset of vowel and the peak of intensity curve 3-c = peak of amplitude 4-d = amplitude at the half distance of the peak amplitude and the end of vowel 5-e = amplitude at the end of vowel 3.6. Manifestation of R2 and R4 vowels In the recorded material, clear-creaky (R2 ) and breathy-creaky (R4 ) vowels behave in many different ways. 6 The following are the most obvious manifestations of clear-creaky and breathy-creaky vowels: (i) Complete change of the oscillatory pattern in the middle or at of vowel duration; (ii) Jagged oscillations into the following consonant; (iii) Brief change of waveforms in the middle of vowel with a decrease in intensity and fundamental frequency; (iv) Slight change of waveforms which indicates laryngealisation or creakiness throughout the vowel, not found very often. 6. Danish, it is maintained, is a register language: 'Tn Danish two such words as hun "she" and hund "dog" are pronounced alike except for a difference of register, the second having creaky voice.' (Abercrombie 1967: IOI). Peterson (1973) did an instrumental investigation of the Danish 'stod', which is regarded as a phonologically distinctive element. Tt is amazing to see that the results of my instrumental study of clear-creaky and breathy-creaky vowels in Chong more or less agree with these of Petersen who also refers to another instrumental investigation of the stod which was done by Smith (1944). Smith, who included electromyographic, oscillographic and kymographic registrations, thinks that it is a stress accent, concluding that it often appears to be three-phased: (I) A ballistic contraction of the expiratory muscles; (2) Cessation of this activity which causes a lack of balance in the reaction of the vocal folds; (3) A new activity in the expiratory muscles. 155

PAGE 164

THERAPHAN L. THONGKUM Table 3: Mean amplitude, mean duration, and standard deviation of 1st, 2nd, 3rd and 4th register vowels in six types of syllable (speaker FA) .... 2l sl, Amplitude (dB) Syllable type 1-a 2-b 3-c 4-d Rl 23.52 36.32 39.52 36.60 {5.34) (3.63) (3.59) (3.73) (C)VV(N) R2 17.20 36.48 40.80 34.68 (7.71) (3.55) (3.27) (3.89) R3 21.08 33.12 35.96 33.48 (6.35) (4.09) (3.43) (3.51) R4 19.29 33.17 37.38 28.72 (5.62) (4.76) (3.98) (5.91) Rl 21.82 34.64 38.68 34.09 (8.68) (5.04) (3.55) (6.04) R2 19.55 35.27 39.50 33.05 (C)VV(S) (10.55) (5.05) (4.13) (5.21) R3 16.59 30.86 34.14 31.32 (8.88) (5.44) (5.20) (5.66) R4 14.05 32. 18 35.59 28.00 (7.67) (6.81) (6.41) (5.75) Rl 19.50 32.40 35.50 27.00 (7.09) (2.94) (2.33) (3.29) R2 14.40 32.40 37.30 28.30 (C)V(N) (9.67) (5.04) (2.49) (4.61) R3 9.30 26.30 29.80 26.10 (5. 18) (6.60) (5.95) (6.53) R4 10.80 26.50 30.80 21.20 (8.11) (5.22) (6.14) (6.16) 28.50 34.75 38.50 39.00 5-e 27.60 (4.67) 25.80 (6.42) 24.96 (3.52) 18.63 (5.54) 13.14 (6.45) 14.95 (9.13) 10.09 (7.73) 11.59 (7.46) 23.10 (3.96) 19.30 (4.38) 19.30 (3.58) 10.80 (4.09) 17.25 Rl (4.09) (4.60) (3.77) (5.09) (I 1.03) R2 13.75 31.00 38.50 29.75 11.25 (C)V(S) (8.32) (2.23) (3.5) (2.05) (6.83) R3 15.75 27.00 30.00 25.25 3.00 (5.85) (6.08) (4.84) (3.70) (4.12) R4 19.25 26.50 32.00 27.00 6.50 (6.83) (2.60) (4.74) (4.58) (8.17) Rl 20.17 37.17 40.17 34.00 19.67 (C)V(H) (9.5) (4.59) (5.75) (5.36) (7.34) R3 11.33 35.83 37.00 29.83 10.83 (7.09) (4.71) (5.42) (4.71) (8.13) RI 29.00 39.75 42.50 40.50 31.25 (C)V(?) {5.20) (1.48) (2.69) (2.18) (5.36) R3 16.75 35.25 38.50 35.50 27.00 (10.52) (2.38) (2.06) (2.60) (4.74) 156 Duration (msec) 1-3 3-5 1-5 166 181 347 (10.45) (10.02) (1.87) 122 216 338 (6.19) (6.42) (2.11) 115 220 335 (6.00) (5.62) (2.17) 118 201 319 (4.45) (4.87) (2.81) 145 176 321 (9.23) (9.48) (2.96) 095 214 309 (3.45) (3.39) (3.14) 096 238 334 (5.83) (5.24) (2.57) 104 182 286 (4.47) (3.50) (2.90) 046 171 217 (2.69) (4.66) (2. 76) 058 173 231 (4.31) (4.65) (1.76) 063 148 211 (3.35) (4.51) (2.34) 070 143 213 (3.55) (3.16) (2.53) 453 090 143 (2.27) (1.87) (0.83) 072 103 175 (2.58) (2.58) (0.5) 068 090 158 (3.63) (1.58) (2.58) 040 088 128 (1.87) (I. 79) (1.09) 080 132 212 (2.58) (1.34) (1.57) 102 117 229 (2.03) (2.43) (1.57) 098 087 185 (0.83) (0.83) (0.50) 082 098 180 (2.86) (3.70) (1.22)

PAGE 165

An instrumental study of Chong registers Table 4: Mean amplitude, mean duration, and standard deviation of 1st, 2nd, 3rd and 4th register vowels in six types of syllable ( speaker FB) ... 1:! Syllable type s, i::! I-a Rl 11.00 (6.58) (C)VV(N) R 2 9.96 (6.31) R3 10.28 (5.18) R4 9.48 (5.55) RI 7.82 (6.23) R2 6.68 (C)VV(S) (7.09) R3 6.27 (5.58) R4 7.73 (5.48) Rl 7.30 (5.42) R2 6.50 (C)V(N) (6.95) R3 4.40 (4.43) R4 5.20 (5.13) Rl 7.50 (7.63) R2 9.00 (C)V(S) (7.28) R3 3.75 (4.49) R4 3.50 (4.37) Rl 3.67 (C)V(H) (5.06) R3 3.67 (3.14) Rl 10.00 (C)V(?) (7.07) R3 8.25 (2.86) Amplitude (dB) 2-b 3-c 4-d 24.92 27.36 25.00 (3.19) (3.54) (3.37) 26.72 31.32 22.48 (6.28) (5.24) (7.76) 21.28 24.40 21.44 (3.70) (3.05) (3.47) 22.08 28.52 16.16 (3.78) (3.77) (5.61) 23.09 27.41 22.91 (5.68) (3.28) (4.92) 25.00 30.05 21.68 (4.49) (2.85) (5.06) 20.00 22.77 19.36 (4.15) (3.32) (3.89) 21.18 26.14 16.41 (5.25) (4.63) (5.27) 21.70 26.50 21.70 (3.85) (2.46) (2.61) 24.90 30.60 17.60 (4.66) (5.04) (5.35) 19.00 23.00 17.50 (5.67) (5.69) (6.55) 19.70 24.90 8.40 (5.16) (5.26) (7.36) 21.25 28.50 23.50 (2.95) (2.69) (4.15) 21.75 30.50 27.50 (6.57) (3.84) (2.87) 15.00 23.75 19.00 (5.92) (1.09) (1.12) 19.25 24.50 19.00 (6.61) (4.56) (7.00) 26.33 31.17 25.33 (3.09) (3.48) (4.11) 25.17 28.83 24.33 (7.84) (5.08) (6.49) 24.75 31.50 29.50 (6.38) (2.5) (2.06) 24.50 28.50 26.50 (2.64) (2.60) (3.20) 157 Duration (msec) 5-e 1-3 3-5 1-5 17.72 141 150 291 (3. 73) (6.35) (6.69) (2.57) 14.92 105 162 267 (6.05) (4.28) (4.19) (2.78) 13.88 096 189 285 (5.17) (4.71) (5.80) (2.50) 12.68 108 153 261 (4.75) (2.93) (3.11) (2.58) 11.55 119 181 300 (7 .33) (6.66) (6.71) (2.29) 10.82 090 191 281 (5.98) (5.06) (4.87) (2.28) 4.50 080 215 295 (4.19) (4.74) (4.27) (2.18) 7.50 807 176 263 (5.95) (3.08) (2.52) (2.56) 12.90 057 122 179 (4.35) (I .62) (3.22) (2.26) 7.30 053 131 184 (2.24) (1.10) (2.62) (2.15) 10.90 065 116 181 (3.91) (2.73) (4.15) (2.07) 4.40 081 109 190 (4.45) (2.74) (1.45) (2.93) 10.75 060 078 138 (3.27) (3.00) (1.09) (2.17) 16.50 080 058 138 (8.20) (4.12) (2.77) (1.79) 4.00 055 083 138 (4.00) (2.29) (1.09) (2.28) 3.50 070 050 120 (3.57) (2.45) (1.58) (1.87) 10.33 060 140 200 (3.54) (1.29) (2.00) (2.58) 7.67 075 108 183 (5.93) (2.57) (0.89) (2.13) 24.50 080 098 178 (2.96) (4.30) (5.21) (1.29) 17.50 121 042 165 (6.38) (2.68) (1.32) (3.5)

PAGE 166

THERAPHAN L. THONGKUM 4. Notes on laryngeal waveforms and airflow For the investigation of laryngeal vibration and airflow, two female speakers of Chong were brought to the Phonetics Laboratory at Chulalongkorn University. No quantitative measurement was attemp ted. However, it can be seen clearly that the laryngograms of the four sets of Chong vowels look different. At the onset of clear and clear-creaky vowels, there is a relative rise which is produced by the rapid closing of the vocal folds for normal voicing,7 and on the other hand, there is a relative fall at the onset of breathy and breathy-creaky vowels which is caused by a more open glottis during breathy phonation. There is also a relatively sharp rise after the release of final consonant following clear-creaky and breathy-creaky vowels. Regarding airflow, the most prominent characteristic of breathy vowels is strong airflow. 8 During the creaky part of R2 and R4 vowels, there is a sudden drop of airflow caused by a rapid closing and tightening of the vocal folds.9 6. Conclusion The instrumental investigation of Chong vowels supports what can be perceived auditorily: that is, there are four sets of register complex, namely, R1 R2 R3 and R4 The phonetic correlates of the four registers may be summed up as follows: Register I: Register 2: higher frequency of F1 (more open vowel); rather level F0 contour (level pitch); regular audio and laryngeal wave forms; higher amplitude; bell-shaped intensity curve; lower amount of airflow in comparison with R3 ; normal voice phonation; higher frequency of F1 (more open vowel); high rise-fall F0 contour; regular followed by irregular audio and laryngeal waveforms; highest peak of amplitude; cone-shaped intensity curve; sudden decrease of intensity and funda mental frequency; sudden drop of airflow; normal voice followed by creaky voice phonation; 7. In normal voicing there are three distinct parts of the waveform. First, there is a relatively sharp rise which is produced by the rapid closing of the vocal folds which is so characteristic of their normal vibration and is associated with the interval of greatest acoustic excitation of the vocal tract. Second, there is the more gradual fall which is associated with the parting of the vocal folds as the sub-glottal pressure is increased; and third, there is the flatter base of the waveform which corresponds to the interval during which the glottis is open and the vocal folds are out of contact. (Fourcin & Abberton 1972: 165-66) 8. Fischer-forgensen measured the airflow of clear and murmured or breathy vowels in Gujerati by means of aerometer. She concludes (1967: 153) that murmured vowels are characterised by a strong airflow which is due to the presence of a small opening in the rear part of the glottis, and/or by an increased activity of the expiratory muscles. 9. When a laryngealised or creaky voice sound is made, the arytenoids are held tightly together while a small length of the ligamental vocal cords is vibrating (Ladefoged 1971: 8). 158

PAGE 167

Register 3: Register 4: REFERENCES Abercrombie, D. Baradat, R. Diffioth, G. An instrumental study of Chong registers lower frequency of F 1 (more closed vowel); gradual fall Fo contour; regular audio and laryngeal waveforms; lowest amplitude; bell-shaped intensity curve; strong air flow; breathy voice phonation; lower frequency of F 1 (more closed vowel); high fall Fo contour; regular followed by irregular audio and laryngeal waveform; rather low amplitude; cone-shaped intensity curve; sudden decrease of intensity and fundamental frequency; sudden drop of airflow: breathy voice followed by creaky voice phonation. 1967. Elements of general phonetics. Chicago: Aldine Publ. Co. 1941. Les dialectes des tribus samre. Paris: MS, Ee. fr. Extr.-Orient. 1974. Austro-asiatic languages. In Encyclopedia Brit., 480-84. Fischer-forgensen, E. 1977. Phonetic analysis of breathy (murmured) vowels in Gujarati. Fourcin, A.J. & Abberton, E Garcia, M. Gregerson, K.J. Headley, R.K. Henderson, Eugenie J.A. Huffman, F.E. Kirk, P.L. Ladefoged, P. Laver, J. A Rf PUC 2 (repr. Annual Report Inst. Phonetics, Univ. Copenhagen) 105-56. 1972. First applications of a new laryngograph. The Volta Review 74(3), 161-76. 1855. Observations of the human voice. London, Edinburgh and Dublin Philosophical Magazine and Journal of Science 10, 511-13. 1976. Tongue-root and register in Mon-Khmer. In AustroasiaLic Stud. 1 (Oceanic Ling. Spee. Pub. 13) (eds.) P. N. Jenner, et al. Honolulu: Univ. Hawaii Press, 323-69. 1977. A Pearic vocabulary. Mon-Khmer Stud. 6, 69-149. 1978. An English-Pearic vocabulary. Mon-Khmer Stud. 7, 61-94. 1985. Proto-Pearic and the classilication of Pearic. In Southeast Asian linguistic studies presented to Andre-G. Haudricourt (eds.) Suriya Ratanakul, et al. Bangkok: Mahidol Univ., 42878. 1952. The main features of Cambodian pronunciation. Bull. Sch. Or. Afr. Stud. 14(1), 149-74. 1976. The register problem in lifteen Mon-Khmer languages. Austroasiatic Stud. 1 (Oceanic Ling. Spee. Pub. 13) (eds.) P. N. Jenner, et al. Honolulu: Univ. Hawaii Press, 575-89. 1985. The phonology of Chong, a Mon-Khmer language of Thailand. In Southeast Asian linguistic studies presented to Andre-G. Haudricourt (eds.) Suriya Ratanakul, et al. Bangkok: Mahidol Univ. 355-88. 1984. et al. Using a spectrograph for measures ofphonation types in a natural language. UCLA Working Papers in Phonetics 59, 102-13. 1971. Preliminaries to linguistic phonetics. Chicago: Univ. Chicago Press. 1980. The Phonetic description of voice quality. Cambridge: Cambridge Univ. Press. 159

PAGE 168

THERAPHAN L. THONGKUM Lee, T. Martin, Marie A. Morner, M., Fmnsson, F. & Fant, G. Petersen, Pia R. Pike, K.L. Shorto, H. L. Suphanphaiboon, Surekha Thomas, D., & Headley, R. K. Thongkum, Themphan L. Zemlin, W.R. 1983. An acoustic study of the register distinction in Mon. UCLA Working Papers in Phonetics 57, 79-96. 1974. Remarques genemles sur Jes dialects pear. Asie du Sud-Est et Monde Insulindien 5(1), 25-37. 1963. Voice register teminology and Standard pitch. Quarterly Progress and Status Report 4, 17-23. Stockholm: Speech Transmission Lab., Royal Inst. Technology. 1973. An instrumental investigation of the Danish 'st0d'. AR/PUC1 (Annual Report Inst. Phonetics Univ. Copenhagen), 195-234. 1948/76. Tone languages. Ann Arbor: Univ. Michigan Press. 1966. Mon vowel systems: a problem in phonological statement. In In memory of J. R. Firth (ed.) C. E. Bazell, et al. London: Longmans, 398409. 1982. The phonological system of Chong language, Muban Takhian thong, Tambon Takhianthong, Amphoe Makham, Changwat Chanthaburi. M.A. Thesis, Srinakharinwirot Univ., Bankgok. 1970. More on Mon-Khmer subgroupings. Lingua 25, 398-418. 1982. Register without tongue-root in Nyah Kur (Chao Bon). Paper presented at the 15th International Conference on Sino-Tibetan Languages and Linguistics. August 1982, Beijing. 1985. An acoustic study of the register complex in Kui (Suai). Paper presented at the 18th International Conference on Sino-Tibetan Languages and Linguistics. August 1985, Bangkok. 1968. Speech and hearing science: anatomy and physiology. New Jersey: Prentice-Hall. 160

PAGE 169

KEEPING THINGS UP FRONT: ASPECTS OF INFORMATION PROCESSING IN MAL DISCOURSE STRUCTURE David Filbeck 0. Introduction Mal is a Mon-Khmer language spoken in Pua and Chiang Klang Districts of Nan Province, Thailand. It is also one of the many dialects spoken by the Tin tribal people who are located in this part of Thailand and across the border into Laos (Filbeck 1978). The number of Mal speakers probably does not exceed 6-7000 people but an accurate census is not available since the Thai government includes all Tin people, regardless of dialect, in a single census. The purpose of this paper is to describe discourse-level phenomena found in the Mal language. Related phenomena occur in other Tin dialects, but it appears that only in Mal are they so extensively used. Since Mal is the most conservative of the Tin dialects, retaining more of Proto Tin than other dialects (Filbeck 1978), we may presume that these phenomena were part of Proto-Tin as well. However, there is internal evidence showing that at least some of these discourse features are developments unique to Mai. For example, the emergence of a rising tone in Mal (Filbeck 1972) has evidently been used to signal at least one discourse feature in Mai. (See discussion in 3. below.) The particular viewpoint taken here in describing these aspects of Mal discourse structure is information processing within a discourse.1 As a descriptive approach to language, information processing differs from the more traditional linguistic approach. The latter views language as composed of different hierarchies or systems, e.g. phonology, morpho syntax, semantics, and discourse, each of which is then described as a more or less autonomous part, and when all the hierarchies or systems have been described separately for a particular language, the task of the linguist is considered to be complete. While this 'divide and conquer' approach has been productive, it has also had its costs. Logically such hierarchies or systems appear to exist, and in the way that linguists say they exist. However, in terms of process, not everything that linguists claim to exist-or to exist in the way that they claim-plays a part in the main function of language, viz. communicating or transmitting meaning. Not everything is needed or used to process, i.e. I. At this stage, information processing should not be confused with information theory. The latter is basically a form of statistical theory dealing with the number of alternatives available in a code and the results of choosing one such alternative. On the other hand, as will be seen in various parts of the following discussion, information processing does use a number of insights drawn from the more statistically oriented information theory. 161

PAGE 170

DAVID FILBECK encode and decode/interpret, meaning by means of language. At least, not everything is used at all times. In short, processing in language follows its own 'rules' which, though drawing upon the traditional categories of linguistics, describe language from a different perspective and with different goals in mind. Information processing-viewed here in perhaps a narrower sense than that which is found in the literature-is a specialised part of the encoding/ decoding aspect of language. On the one hand, there is content that is encoded in a text, whether written or oral, for transmission and decoding. Yet, content is not just encoded and decoded; it is also 'processed' in various ways by means of linguistic structure-more precisely, the categories used in linguistic description. The linguistic structure of a language, therefore, plays a 'mediating role' in the encoding, transmitting, and decoding, of meaning. In this view of linguistic structure, processing may be considered a part of the total semantic structure of language, or, to state it another way, meaning as transmitted by language is composed of content plus processing.2 Content is processed in two ways. First, it is distributed or mapped linearly along a syntactic code. For example, content may be 'tightly packed' into a few sentences (as a technical exposition), or it may be distributed over many sentences (as in popular exposition). Second, content may be processed globally as theme, prominence, topic, focus, old and new information, etc., in the course of a text or discourse. In either way, what is processed is additional information 'about' the content that is similarly encoded in the code. Moreover, just as the speaker processes content for additional information while encoding content, so the hearer processes the incoming code for both content and information in the act of decoding. That is, the hearer (or reader of a text) comes to certain judgements about the content he or she has received on the basis of the code, thus providing (additional) information 'about' the content. For Mai, this paper discusses information processing in this second way only. More specifically, Mal employs various linguistic means to signal and/or differentiate i.e. process various types of information flowing globally through discourse. As will be seen in the following discussion, such processing forms a crucial dimension of the Mai language. A description of such, then, is necessary for a full linguistic description of Mal. Moreover, as a by-product, it is hoped that this paper will make a contribution to describing the nature of discourse phenomena in Mon Khmer in general. 3 Data for this paper are taken from both recorded texts that I have collected and from the recently completed translation of St. Mark's 2. Or, in terms of information theory: content, plus information about content, constitutes the message of a text. 3. Some speak also of a pragmatic level of information processing (qv. Smith & Wilson 1979), where information is processed according to shared knowledge, assumptions, or beliefs, but not necessarily according to the characteristic of the linguistic code itself. This pragmatic aspect of information processing is also not discussed in this paper. 162

PAGE 171

Information processing in Mai discourse structure Gospel into Mal. Examples will also be taken from the first two chapters of the Book of Genesis, which is currently being translated into Mal. Illustrations from the latter two sources are especially revealing in that they show how productive discourse-level phenomena are in Mal. Greek, English and Thai obviously do not have the same type of formal means for dealing with such phenomena. Consequently, it was fascinating to observe the translator (translating from Thai into Mal in this case) determine from the context how best to structure the text formally according to Mal discourse structure. 1. A discourse characteristic For this paper, a deductive approach, i.e. from the 'top down', to informational processing in Mal is assumed. That is, we begin by observing a general characteristic-a 'flavour'-typical of Mal discourse, whether interpersonal communication or story-telling. When we have established what this discourse characteristic is, we will then illustrate how it is realised formally or linguistically in discourse. In analysing the following data we can observe a general 'theme' operating throughout the discourse structure of Mal, namely, the theme of inclusivity vs. exclusivity, or 'this/that-including-others' vs. 'this/that excluding-others'. Of course, inclusivity dominates discourse activity in Mal; on the other hand, exclusivity can be forcefully and unambiguously marked with formal linguistic mechanisms. That it is so marked quite often gives Mal discourse a distinct character. In this regard, then, we may classify exclusivity as the 'marked' process and inclusivity as the 'unmarked' process. Also, a text may be 'more or less' highly marked according to the number of marked processes it contains. Moreover, when considering the Mai language from a discourse perspective, we can observe that Mai is a comparatively 'highly marked', i.e. more complex, language with respect to the inclusive/exclusive processing of information. For, characteristically, this distinction is a real option in everyday speech, and it is an alternative that is chosen often enough in interpersonal communication as well as in other forms of discourse. The communicative function of this inclusive/exclusive dichotomy in Mai is, as the title of this paper suggests to 'keep things up front', i.e. to keep track in a formal way of what is and is not important during the flow of discourse. It is a formal way of processing information, both for the speaker and hearer, in communication, whereby a piece of content is 'kept up front', or made prominent, in discourse. Conversely other pieces of content are 'shuttled' to the background but, should any of these need to be made prominent, it is possible to do so by 'bringing it up front', as it were, by formal linguistic means. 2. Pronouns: 'we vs. they' The dichotomy between inclusivity and exclusivity in Mai discourse is nowhere better observed than in the pronouns, which form a well-163

PAGE 172

DAVID FILBECK developed system of exclusive-inclusive relations. Singular /:Jfi./ 'I' /mah/ 'you' /nam/ 'he, she' Dual /ia?/ 'we (excl.)' /aa? / 'we (incl.)' /paa?/ 'you' /paam/ 'they' Plural /ii/ 'we (excl.)' /EE/ 'we (incl.)' /pEE/ 'you' /ah/ 'they' The unmarked pronouns include the singular pronouns ('I, you, he, she') and the inclusive plural. These are used the most often in conversation and in recorded texts, while the marked pronouns-the duals and exclusives-occur more rarely in those environments. Indeed, the marked pronouns in Mai occur not just in one degree of markedness but two: marked and highly marked. The marked pronouns are the dual and/or exclusive plurals, e.g. /ia?/ 'we (dual excl.)', /paa?/ 'you (dual)', /paam/ 'they (dual)' and /ii/ 'we (pl. excl. '). These occur often enough in conversation and in texts to lend a flavour of exclusivity to Mai communication. Only /aa?/ 'we (dual incl.)' is a highly marked pronoun. It is classified here as an inclusive dual pronoun while /ia?/ is classified as its exclusive counterpart. The reason for this is that /ia? / is used in third-party conversation where the listener is not included, i.e. is 'excluded' from the dual grouping. For example, in a recent conversation two young boys requested leave to go somewhere. One of the boys, who was the spokesman, used the pronoun /ia? / to include himself and the other boy while at the same time excluding me as the listener in the request. By contrast, the other pronoun, /aa?/ 'we (dual incl.)', is used to include each partner of the dyad in the conversation. Consequently it is rarely used, being limited to conversations between two people who are alone together. In the Mai story 'The Magic Trail' (which we shall refer to again in this paper), a husband and wife set out alone to seek a fortune. In the version I have recorded, the narrator relates the story using a great number of direct quotes, especially when the husband speaks to the wife. Since the couple are poor and own no livestock, they are seeking in the forest for animals which can be taken back and raised as domesticated stock in the village. One of the first creatures they find is a wild boar, at which the husband exclaims to his wife: oo, aa'i hal) ac! Oh, we-two are-rich already He then instructs her to take the boar back to the village and pen it up. Whenever a different forest animal is found, the husband makes the same exclamation in addressing his wife, each time using the pronoun /aa?/. The other dual pronoun, /ia?/, is not the correct pronoun here as there is no third person whom the husband is addressing. Presumably, on 164

PAGE 173

Information processing in Ma! discourse structure returning to the village with his wife and his entourage of newly-found animals, he could announce to the villagers that /ia? haIJ ':JC/ 'we-two are now rich', since he would be excluding the villagers, in their role as listeners, from any participation in this new-found wealth. A major role of pronouns in language is to keep track of participants, especially on the discourse level in a way that is not excessively redundant. Or, from the standpoint of information processing, pronouns serve to process information about participants that figure in a discourse. For example, to repeat a noun or a person's name continually in a discourse or text is tediously redundant. Substituting pronouns for such items allows redundancy to be reduced while at the same time allowing the processing of essential information regarding the participants. Obviously, then, the more complex a pronominal system the better a language is able to keep track of, or to process, information about participants in a text or discourse. Because of a well-developed system of duals, and inclusive-exclusive relations, pronouns in Mai are able to fulfil the above role in perhaps a more refined way than that which is possible in some other languages. This can be seen through the use of the dual pronouns. Examples are found in the Mai translation of Chapter 14 of Mark's Gospel in which Jesus selects two of his disciples to go and prepare a place for the Passover meal: And he sent two of his disciples, and said to them (dual), 'Go into the city, and a man carrying a jar of water will meet you (dual); follow him, and wherever he enters, say to the householder, 'The Teacher says, Where is my guest room, where I am to eat the passover with my disciples?' And he will show you (dual incl.) a large upper room furnished and ready; there prepare for us (pl. incl.)' And the disciples set out and went to the city, and found it as he had told them (dual); and they (dual) prepared the passover. (Mark 14:13-16, RSV). These instructions were originally clearly addressed to the two disciples, but of course there are no dual pronouns in either Greek or English. Consequently, the information about the 'duality' of the participants (i.e. the two disciples) is processed in a more refined way in Mai; once it is established in the text, dual pronouns are used thereafter in order to keep track of them. Viewing the process another way, one might say that-just as in the morphology of certain languages-in Mai discourse structure a 'principle of informational agreement' is in operation. That is, once two participants have been established as a 'dual unit' in a textual discourse, subsequent anaphoric reference to them must be kept 'in agreement informationally' by employing the appropriate dual pronoun(s). Use of other plural pronoun(s) would not be in agreement and would transmit incongruous information. In Mai, then, it is necessary to use dual pronouns in order to process correctly certain types of information about participants who figure in a text or discourse. 165

PAGE 174

DAVID FILBECK Even in the Mai story about the magic trail, mentioned above, we can see how certain information regarding the participants is processed through the use of the pronoun /aa?/ 'we (dual, listener inclusive)'. At the time of recording, the story was told by a narrator to a group of people. By casting the husband's exclamation ('We are now rich!') to his wife in direct quotes, the narrator was able to use the pronoun /aa? / in his narrative, thus transmitting to his listeners that not only were there two participants but that they were also alone together in their adventures along the magic trail. Again we can see the 'principle of informational agreement', mentioned above, in operation here. In the story, the actual situation of a husband and wife starting out alone on the magic trail is established immediately. Consequently, any anaphoric reference to the two of them must necessarily involve the pronoun /aa?/; the information transmitted by any other pronoun would not be in agreement with the established context. Exclusive pronouns in Mai also function to process information about participants in a text or discourse. An example of this may be seen in Mark 9: 28, which tells how nine of Jesus' disciples were unable to cast out a spirit from a little boy. Jesus did so and, after they had entered the house, the disciples asked him: GfJ phi? naa ii yaap ay p:J;m why we (excl.) cast un-able 'Why could we (excl.) not cast it out?' For Mai, the pronouon 'we' is here translated as /ii/ 'we (pl. excl.)', which both refers unambiguously to the nine disciples and excludes Jesus; that is, the information that these participants are mutually exclusive is processed in a straightforward manner. 3. Demonstratives: 'keeping things up front' Mai parallels other languages in that it has demonstratives, relative pronouns and nominalisers, i.e. words or particles that turn verbs and clauses into nominals. The Phrase Structures underlying these categories are as follows (see also Filbeck 1976): (for demonstratives and re!. clauses) NP N-X-DEM X = {Adj. + Adj. +No.+ Class Rei. Pro.+ S (for nominalization) NP A NOM -S 166

PAGE 175

Information processing in Mai discourse structure However, this is where the parallel with other languages often ends, for the information processed by these grammatical categories and structures in Mal frequently turns out to be different. To begin with, Mal demonstratives also function like determiners (e.g. 'the' in English) to introduce old information as well as to specify (e.g. 'this' and 'that'). Moreover, there are two sets of the above categories for demonstratives, relative pronouns and nominalisers, which include also a null element,4 as follows: Demonstratives Rei. pron. Nominaliser: Set I /nee/ 'this, here' /een42/ 'that, there' /0/ 'null' /ta-/ 'that which is' Set 2 /;ir42/ 'this, here' /ee42 / 'that' /ee/ 'who, which' /ee/ 'that which is' These differ in that Set I may be considered the unmarked set with its items being used to signal old information (in the case of 'this' and 'that'), or to process information internally on the sentence level (as in the case of the relative pronoun and nominaliser). Set 2, then, is the marked set whose items are used on the discourse level to process information on a more global scale across sentence boundaries. Also, in Set 2, /ee/ may be either a relative pronoun or nominaliser. In historical items, /ee42 / 'that (dem.)' and /ee/ 'who, which (rel. pron.)' may have been the same word but, because of the development of the rising tone in Mal,5 it may have been split in order to process in an unambiguous manner the difference in information (demonstrative vs. relative pronoun/nominaliser) that the two words now convey. The demonstratives may also be used as locatives, with the exception of /ee42/ in Set 2. Yet even this word may have a locative sense, meaning 'there', under certain circumstances, so that it can correspond to the /ee42 / 'there' of Set I. However, this locative sense, due to the special or marked role of Set 2, is clearly secondary when used in discourse. That is, the special role of Set 2 takes precedence in the usage and interpretation of /ee42/, as will be seen below. The special role of Set 2 is to mark the focus of a discourse. More specifically, Mal employs several linguistic markers for singling out and keeping track of this focus, which may be a noun (in which case it is modified by a demonstrative), or a verb or even a clause (in which case the relative pronoun/nominaliser is used). If the focus changes during the discourse, these same linguistic markers are used to indicate that change. They stand in opposition to 'non-markers' which signal that other words are not foci. Stated another way, Mal employs certain morphemes to 'mark' certain pieces of content as more salient in discourse; other pieces 4. I am indebted to Hermann Janzen for this particular way of describing the phenomenon. 5. This single tone is signalled here and above by the raised numerals [42] showing that the tone starts on a [4] or low-level and rises to a [2] or high-level pitch. 167

PAGE 176

DAVID FILBECK are then left 'unmarked' and thus are not as salient.6 In terms of information processing, the function of the words contained in Set 2 is to 'keep things up front', i.e. to keep track in a formal way of what is, and what is not, the focus of a discourse. It is a way of processing information 'about' content in discourse, viz. to signal which piece of content is the focus; other pieces of content not so marked are not foci but rather form background information. But should any of these need to be made prominent, this can be achieved by 'bringing them up front', as it were, by means of these same linguistic devices. We now turn our attention to a more detailed description of these two sets of demonstratives/relative pronouns/nominalisers. Most of the discussion will focus on Set 2 since it is the more interesting with regard to discourse structure in Mai. 3.1 /nee/ and /ar42 /: 'this vs. this' The basic meaning and usage of /nee/ is that of the demonstrative 'this', e.g. /phyam nee/ person this 'this person' When combined with the noun meaning 'place', however, it has the locative meaning, eg. /nam at taanee/ he lives place-this 'he lives here' In the former example, /nee/ would, in a discourse, signal old information, i.e. that the person discussed has already been introduced and is now being referred to again. In the latter example, it is not apparent that /nee/ necessarily signals old information each time it is used in this construction. As a locative, /taanee/ 'here' could very well be new information introduced for the first time. The difference between /nee/ and /-;ir42 / may be summarised as follows: /nee/, either as 'this' or 'here', is general and non-specific, while /-;ir42/ is much more specific, especially in focusing, in both meanings. Normally, according to their respective functions, both words are used interchangeably in a discourse. However, I have one short Mai text describing a healing ceremony in which a piglet was sacrificed. The reason for this was probably the brevity of discourse and the fact that it described sequentially the events that took place during the course of the ceremony. In other words, there was no hierarchical structuring of the description there were no diversions to fill in background information, flashbacks, 6. Other dialects of Tin have cognates to /ee/ but, as far as I have been able to determine, their usage does not include the special functions that /ee42/ has in Mai. This leads me to believe that Set 2 words in Mai may be a later development, perhaps a more explicit development of such a discourse tendency in Proto-Tin. 168

PAGE 177

Information processing in Mai discourse structure loops, etc. Hence nothing (noun or clause) turned out to be the focus or foci of the description. Conversely, everything in the description turned out to be a ( or 'the') focus. In any event, there was no need to mark anything specifically; consequently, only /nee/ was used throughout. As a demonstrative meaning 'this', the specificity of /'dr42 / may be seen in the Mai translation of Mark 8:32. In this verse, Jesus has just finished foretelling his coming Crucifixion. It was not a veiled foretelling, for, as this verse states: 1]] ar42 yeesuu mpfe? alJ mac word this Jesus show to see 'Jesus revealed this plainly .. .' so that his disciples would not misunderstand. The demonstrative /nee/ was rejected by the Mai translator; it was not specific enough and thus did not 'cohere' with the adverb 'plainly'. /'dr42/, on the other hand, processed the clarity or 'plainness' of the prediction in a much more straightforward manner and it also marked the foretelling of the Crucifixion as the focus of the discourse, thus turning other pieces of content into background information, a function that /nee/ could not have fulfilled. However, it is in the locative meaning of 'here' that the difference between /nee/ and /'dr42/ is more apparent. Two examples will help illustrate this. /at taa-'dr42/ 'stay here!' is the statement of choice with a Mai parent in commanding a child to 'stay put' and not to wander off as children often do. In such a pragmatic context the parent means 'here in the immediate vicinity', such as the child's home and/or compound where the house is located and not in a wider area which may include the whole village. The command /at taanee/ would amount to permission for the child to wander and play in a much wider, less specififed area. On the other hand, in answer to the question 'Where does (someone) live?', one answer may be /nam at taanee/ 'he lives here', i.e. this general vicinity (normally a village) is where he lives. But in answer to the more specific question of 'Where is (someone)?', one answer could very well be /nam at taa-'dr42/ 'he is here', i.e. at the time of the answer it is affirmed that the person is physically located in the immediate vicinity (e.g. inside the house where the questioning took place) and nowhere else. In these examples the locative /taa-'dr42/ 'here' marks the location of the conversation as a specific focus of the conversation as opposed to /taanee/ 'here', which leaves the location of the conversation unmarked as unspecified or background information. So, the importance or non importance oflocation in the conversations above is processed formally in Mai with the use of these two Iocatives. In texts, the difference between the locative meanings of /nee/ and /'dr42/ 'here vs. here' takes on an added dimension, especially when the latter is 169

PAGE 178

DAVID FTLBECK used in direct quotations. The use of /:Jr42/ 'here' in a direct quote in a text reflects the view of the location from the perspective of the participant(s) in the story. Two examples will help illustrate this dimension. In a recorded story about a hunting party composed of Mai men, the narrator, who was a member of the party, recounted the instructions given to each regarding his particular place in the party as the men fanned out to 'sweep' an area of the forest in order to flush out any wild animal that might be hiding in the underbrush. The instruction to each member was given as a direct quote in the story. 7 For the instruction given to him, the narrator stated that he was told to /at taa-:Jr42/ 'stay here'. By using /ta:Jr42/, and phrasing it as a direct quote, the narrator was telling the story from his own perspective or position in the hunting party. Positional instructions to other men were given in relation to the position of the narrator. The second example is taken from Mark 6: 36 during the account of Jesus feeding the 5000. Before the miracle took place, however, the disciples asked Jesus to dismiss the people so they could go to buy food: tarn IJUal. at tarn ar42 at village ... locate around here ... at the surrounding villages' The expression 'surrounding villages' was not translated directly into Mai; rather, the Mai translator chose the expression 'villages that are located /t:Jm :Jr42 / around here'. In Mai, the choice of /:Jr42 / achieves the same meaning but from a different perspective, viz. the perspective of the disciples who were in the same vicinity in which the villages were also located. In both examEles, we should not forget that the location here as expressed by /:Jr / still specifies a focus of the text. In the former, the position of the narrator in the hunting party was the focus since it served as the point from which other positions were pinpointed and described. In the latter, the area containing the surrounding villages was a focus because it was the area in which the people would have to search for food. 3.2. /een42/ and /ee42/ 'that vs. that' The demonstratives /een42/ and /ee42/ parallel /nee/ and /:Jr42/ with the exception of differences in meaning. The former in each case function to process old information while the latter in each case function to mark the focus or foci of discourse. For example, in the following phrase: 7. It is interesting to note that certain sub-dialects of Prai (the other branch, along with Mai, in Tin) contain the two sets apparently only in the area oflocatives, i.e. there are two words meaning 'here' and two words meaning 'there'. The first set is used to denote location only, while the second set is used to establish the locational focus or foci of the discourse. More investigation is needed to pinpoint the difference between the two sets more adequately in the Prai dialect. 170

PAGE 179

Information processing in M al discourse structure phyam een42 person that 'that person ... /een42 / modifies the preceding noun and also signals that the noun is old information which has already been introduced. In addition, when combined with the noun meaning 'place', this demonstrative functions as the locative meaning 'there'. nam at taa-een42 he lives place-that 'he lives there' However, as in the previous section, it is /ee42/ that holds more interest for this discussion, as the demonstrative that marks a noun as the focus of a discourse. This is true in conversation or dialogue as well as in other types of discourse. For example, in a conversation with a Mai man, whom I had never met before, I asked him what village he was from. For some reason his answer was not clear. On my asking again he expanded his answer in order to help me to understand. His expanded answer did help, for when I asked if he lived in a certain village (whose name I also gave), his face lit up as he exclaimed: l)Ua[ village ee42 that = 'That's the village!' In this dialogue the noun /TJual/ 'village' with a certain name was the obvious focus of the interrogation and, when this village was at last named, it was then affirmed formally as the focus. In longer texts, whose content may include several nouns or items, /ee42/ 'that' has a more complicated role. It not only serves to mark the focus of the text but it also reintroduces the focus (e.g. after background information has been discussed) as well as serving to mark a new focus (i.e. when the focus has changed in a text). To illustrate this, let us look again at the story about the 'Magic Trail'. In this story a husband and wife are looking for wild animals, each of which is to be taken back to the village and domesticated. Four such trips are related in the story. After each trip to the village, to take back a wild animal, the husband and wife resume their journey along the trail, which is termed: y:XJl) trail ee42 that = 'That trail' 171

PAGE 180

DAVID FILBECK The use of /ee42 / at these junctures of the story signals that the trail under discussion is the 'Magic Trail', the focus of the story, and not some other trail. However, after some time the focus shifts to other items in the story. For example, a magical snake is introduced and, after some background information is added, the snake is referred to as mar ee42 snake that = 'that snake .. .' to mark the snake as the focus of this part of the story. After this episode, the husband and wife travel on to a magical pond. Again, after background information is added, the pond is referred to as ee42 nh:xJ,/2 pond that = 'that pond .. .' to signal that the focus has changed again. The snake instructs the man to splash the pond dry, after which a single fish is found. This fish takes on a human body and becomes a person. Since this fish-turned-human-body now plays a role in the story, subsequent reference to it in the story takes the form na11 ee42 body that = 'that body .. .' to mark it as the new focus of the story. As pointed out at the beginning of this discussion, focusing in Mai discourse structure also entails a contrastive function, of signalling 'this/ that-and-no-other' regarding an item of content in the discourse. This function is clearly seen in the translation of a parable in Mark 4: 26-29. In this, Jesus compares the growth of the Kingdom of God upon earth to a farmer who broadcasts his seed upon the ground. Jesus explains that the farmer does not know how the seed grows; nevertheless, he waits patiently for the ground to germinate and grow the seed so that he may harvest the grain. In the parable, the two main participants are the seed and the ground/ earth, and both are introduced in 'the farmer went out to sow seed upon the earth', hence any subsequent mention of these two participants is now old information. In verse 27 'seed' is indeed mentioned again and in the Mai translation it is tagged as: khlualJ een42 seed that 'that seed .. .' It is also stated that the farmer does not understand how seed sprouts and grows. In verse 28 'earth' is similarly reintroduced; moreover, in contrast, the verse specifies the earth as the cause of the seed's sprouting 172

PAGE 181

Information processing in Mai discourse structure and yielding its fruits. To convey this contrast, 'earth' is translated into Mai as: thu: ee42 earth that 'that earth .. .' As the real cause of the seed's sprouting, earth, then, becomes the proper focus of the parable as well as the focus for its correct interpretation as a parable about the Kingdom of God. Seed, furthermore, by being modified by /een42/, is relegated to a background or supportive role in the structure of the discourse. The demonstratives /een42/ and /ee42/ interact in still another way in Mai discourse. This interaction is associated with the word /kyaw/ 'to be equal to'. In the expression /kyaw een42/ 'only', the combination is used as a sentence-level modifier without regard to any broader context: nam uuy piar if:? kyaw een42 he has two things equal-to that 'He has only two of them' On the other hand, the expression /kyaw ee42/ is often used as a discourse level modifier in the sense of 'That's the way it is!', or 'That's the summation of the matter!' In a short recorded text, a Mai shaman explained why he would be unable to convert to Christianity. He explained that his most important role was to be a priest interceding for the villagers before the village deity. In this explanation, he gave a list of religious duties that only he as the village shaman could perform. If he converted, there would be no one to perform these duties for the villagers. Therefore, he could not become a Christian. He then ended his explanation with the final kyaw ee42 ac Equal-to that complete 'and that's the complete summation of the matter!' Here /ee42 / relates not to any single noun as its reference but to the whole discourse or, more properly, to the total content of the discourse. Thus the content is the focus of the discourse and not just a noun or item in that content. The demonstrative /ee42/ also has an idiosyncratic but interesting usage at the beginning of discourses which demonstrates its global effect over a discourse. If, for example, a discourse is about an event that took place in a time prior to its narration, it may begin with the time expression: 42 naam ee time that 'at that particular time .. .' 173

PAGE 182

DAVID FTLBECK in order to set the time frame for the narration that follows. This express ion assumes that the Iistener(s) already know the time to which it refers, and, as such, the time of the event also becomes a focus of the discourse. 3.3. /ta-, OJ and fee/ "Clause vs. Clause" In this section, I shall discuss two common grammatical structures-the relative clause and nominalisation and, more specifically, shall limit the discussion to the words or morphemes that constitute the heading to this section because they are the items that process the difference between focus and non-focus information for relative clauses and nominalisations in Mai. The items /ta-/ and the null or zero element /0/ may both be used as relative clause markers e.g. phyam ta-cak een42 kayh aw person who goes that is father 'That person who is going is my father'. phyam {3 cak een42 kayh aw person goes that is father 'That person (who is) going is my father' In addition, /ta-/ may be used before a verb or before a clause, turning it into a nominal.8 In this structure, it can be understood as 'that which is'. nam a;m ta-luh 'he does wrong (that-which-is-wrong)' On the other hand, the null element /0/ may be used only as a relative clause introducer in Mal,9 while the word Jee/ may be used both as a relative pronoun and as a nominaliser. The difference is that /ta-, 0/ are used to process non-focused information while Jee/ is used more globally to mark the focus or foci of a discourse. It should be noted at the outset that this difference between /ta-, 0/ and /ee/ as relative pronouns in Mai does not necessarily parallel the distinction between restrictive and non restrictive relative clauses found, for example, in English, or between /thii/ (falling tone, restrictive re!. pron.) and /sirJ/ (falling tone, non-restrictive 8. Direct quotes are a favourite strategy of story-telling among the Mai. While this is probably more a function of their being members of a pre-literate society, it is nevertheless interesting from the viewpoint of information processing. That is, the content of a Mai story is more often than not processed from the viewpoint of a participant and not from the viewpoint of content as autonomous information unconnected with a narrator. As more Mai become literate it may be expected that content-as-autonomous-information will be used more often as a strategy of story-telling. 9. /ta-/ appears to be the same word as /taa/ 'place' which figures in the discussion in previous sections. In this regard it parallels the Thai word /thii/ (falling tone) which also means 'place' and can be used as a relative pronoun and as a nominaliser. Indeed, this usage in Thai may have influenced that of /taa/ in Mai, extending it to be used as a relative clause marker and nominaliser, too. Whatever the facts in this case, I have chosen to keep /ta-/ and /taa/ as separate morphemes for the purpose of this discussion. 174

PAGE 183

Information processing in Mai discourse structure rel. pron.) in Thai. The information that is processed by way of /ta-, 0/ and /ee/ is different in nature from restrictive vs. non-restrictive information. /ee/ as a relative pronoun introducing a relative clause functions to mark the focus of a discourse. In this case, however, it is not the noun ( or item modified by the relative clause) that is the focus; rather, it is the relative clause itself or, more properly, the content contained in the relative clause, that is the focus. /ee/ shows that the content of the relative clause, even though encoded in a subordinate syntactic construction, is still a focus in its own right. In Mark 11: 21 the disciples pass by a fig tree which Jesus had earlier cursed. The disciples draw Jesus' attention to the tree by exclaiming n:J:Jy lam phft:? khyur ee mah thun al) een42 look tree fig that you curse that 'Look at the fig tree that you cursed!' The focus at this point in the discourse is not the fif tree, as it is modified by the unmarked sentence level demonstrative /ee 2 / in Mai. The focus is on the cursing, i.e. the embedded relative clause that begins with/ ee/ 'that', thus showing that the content of the subordinate clause is a focus in its own right. Earlier, we stated that Set 2 demonstratives in Mai also serve an exclusive function, that of signalling the information of this-and-no-other regarding an item in a discourse. /ee/, as a relative pronoun, performs this function as well. This can be seen in the translation of Mark 3: 13-19 which relates the time when Jesus chose his twelve disciples. In this list of twelve disciples, two have the same name: Simon. The first Simon Jesus renames Peter. The second Simon is distinguished from the first by a relative clause: siimoon ee at phiah mphualJ phyam yaay Simon who live side-of group person dangerous 'Simon who was a terrorist (Zealot).' The inclusion of /ee/ above clearly processes the information that this second Simon is to be distinguished from the first Simon. /ee/ may also be used as a nominaliser. In this construction /ee/ occurs before a verb or verbal clause thereby turning it into a nominal, which in turn may function as a subject or object of another verb. The special function of /ee/ as a nominaliser, however, is to signal that the content of the nominal is a focus of a discourse. In Genesis 2: 17, a tree of the knowledge of good and evil plays a prominent part in the account of the Creation. Of course, the focus of the account is not the 'tree', but on the 'knowledge of good and evil' that arose after the creation of man. To process this information in a translation of this verse into Mai, /ee/ is used three times. 175

PAGE 184

DAVID FTLBECK lam ualj ee mac mphi? ee b? ha? ee ay b? tree knowing that-which-is good and that-which-is not good 'The tree (of) knowing (that which is) good and (that which is) not good.' The first occurrence of /ee/ turns the whole of the following construction into a nominal, and as a nominal directly following a noun ('tree' in this case) it stands in a genitive relationship to the noun (i.e. the 'of relationship). The following nominals (lee/ plus verb) are also genitives. In each case, the /ee/ processes the information that the content of the subsequent verb is a focus, more precisely a part of the total focus contained in the construction. In a usage closely related in meaning to the nominal usage above, /ee/ may stand alone, in which case it functions as a 'pro-nominal' in Mai. That is, /ee/ functions something like a pronoun, except that its reference is not an 'object, place or thing' but rather an event or state of existence; it refers to, and includes, all that happens in an event or to the salient characteristics found in a state of existence. It is as though the /ee + verb/ nominal has undergone an ellipsis, which deletes the /verb/, thus leaving only the /ee/ to communicate the full scope of the nominal. In the story about the 'Magic Trail', the husband and wife came to a pond. They splashed all the water out and found a single fish lying in the mud. This fish turned into a human body. At this point the fish-turned into-human-body becomes a focal point in the story: it (or he) helps the husband and wife in the search for fortune. Consequently, in subsequent reference to this creature the 'pro-nominal' /ee/ is used. In this context, /ee/ could be translated as 'it', but its complete reference is to the (new) state of existence of the participant, viz. the fish-turned-into-human-body. However, /ee/ in this usage is not restricted to events. In another story about magic, a brother and sister left home to find their fortune in a similar manner. A snake told them of a tree whose bark had magical properties. The children took this bark and eventually arrived at the king's palace in the city. There they learned that the princess had died. They made their way into the palace and sprinkled some of the bark on the princess. At this, the princess came back to life-and at this point in the story she becomes a focal participant. Not only does the princess return to life but the magical bark causes her to become very pretty. The narrator of the story next states that: ee at ta-ee42 she lived like that 'She lived in that (pretty) condition [for the rest of her life].' /ee/, translated as 'she', refers not only to the princess but also to the event of returning to life and becoming pretty. The full meaning would be something like: 'she-who-came-back-to-life-and-became-pretty'. In this sentence, the existential state of being pretty is signalled by /ta-ee42/, an 176

PAGE 185

Information processing in Mai discourse structure expression that also means 'there' (see above). However, the full scope of the expression in this context conveys the information that it was this state-and-no-other-state or condition that the princess existed in for the rest of her life. In each story above /ee/ as a pro-nominal marks the focus. However, it is not merely the 'person, place or thing' that is the focus. More properly, the focus is the event or existential state of which the person, place or thing is a part and, being the focus, it excludes other events or states that could be associated with the participant in the discourse. 3.4. Special combinations From the above data, two combinations deserve special comment. They both involve the nominalisation usage of /ee/ plus a demonstrative. The two combinations are: /ee .. [verb] ... een42/ 'that [Nominal]' /ee .. [verb] ... ee42/ 'that [Nominal]' From textual analysis it is difficult to discover any real difference between these two constructions; they each seem to process the same information of focus and/or exclusivity regarding the content contained in the nominal. Both, for example, have been used in translating Mark into Mai. Mark 10: 35-40 records the time when two of the disciples requested seats of honour at the time when Jesus assumed complete power in his kingdom. One asked to sit on Jesus' right and the other on his left. Jesus' answer to this request begins with a nominal: ee paa? soap ee42 that-which you-two ask that '(Regarding) that request of yours ... after which he explains that such positions involve suffering and not honour. When the two disciples affirmed their willingness to suffer, Jesus explains that: ee khuyh phiah sam phiah wi? aii een42 to sit side right side left I that '(But) that sitting on my right or left .. .' was in reality not his to bestow but God's. It is difficult to detect any difference in focusing or exclusivity between these two constructions, unless the former is a higher degree of focusing, etc., than the latter. 4.0. The /i-/ clitic: 'This one exclusively Mai has one clitic, /i-/, that occurs as a prefix to a few words. As a prefix /i-/ is stressless but it can be given stress and thus stand as a free morpheme in a sentence; normally, however, it occurs as a (bound) prefix 177

PAGE 186

DAVID FILBECK in the course of speech. Semantically, though, /i-/ carries no definite meaning in the traditional morpheme sense; rather it functions more as an 'exclusive enhancer' for the words on which it may occur as a prefix. Consequently, its productivity in Mai is limited to occurring with only those words that already contain a component of exclusivity as a major part of their semantic composition. On the other hand, given the exclusivity characteristic of Mai discourse, it is not surprising to see these few words plus /i-/ occurring often in both speech and narratives of the Mai people. /i-/ occurs as an exclusive enhancer of two types of words in Mai, demonstratives and interrogatives. Words in both categories already contain a semantic component of exclusivity, for to modify a noun or nominal with a demonstrative is to exclude other nouns or nominals, and to ask a question often involves a decision to choose one alternative to the exclusion of others. The addition of /i-/ in these cases enhances the exclusive property of such words. 4.1 /i-/ plus Demonstratives In the case of demonstratives, however, /i-/ occurs only with the sentence level demonstratives and not with the discourse-level demonstratives, i.e. with Set I Demonstratives and not with Set 2 Demonstratives as listed in 3 above. The reason for this is that in processing information, /i-/ as an exclusive enhancer in effect 'raises the informational level' from that of old information to that of 'focused information' in the discourse. For, with the addition of /i-/, there is a more explicit exclusion of some other referent, whether in text or conversation, thus throwing more attention on the referent modified by the demonstrative in question. There is therefore no semantic or informational reason for combining /i-/ with the Set 2 Demonstratives since this set already performs the same task in discourse. On the other hand, even though they perform the same task, there is a difference. Consider these examples. ( a) kayh khyaak inee is buffalo this one 'It's this particular buffalo.' (b) kayh khyaak ar42 is buffalo this 'It's this buffalo.' Example ( a) contains a Set 1 Demonstrative, /nee/ 'this', in combination with /i-/. Example (b) contains the Set 2 Demonstrative /ar42/ 'this'. The difference between the two is that the former is more negative or constrastive while the latter is more positive and affirmative. Both, of course, make the NP (/khyaak/ 'buffalo' in this case) a focus. Yet /inee/ turns /khyaak/ into a focus by means of excluding other possible buffaloes while /ar42/ makes it a focus by means of highlighting. 178

PAGE 187

Information processing in M al discourse structure In Mark 10: 17-20 a young man came to see Jesus to ask what must be done to inherit eternal life. Jesus told him to keep the following commandments: do not kill, do not commit adultery, do not steal, do not bear false witness, do not defraud, honour (your) father and mother. To these instructions the young man answered: lJcclJ s;xm42 inee word teach these ones 'These particular teachings ... (I have observed from my youth).' The demonstrative /inee/ in this discourse gives a 'particularising' flavour to the discourse: it was these particular commandments, in contrast to possible others, that he had especially observed in his life. This contrastive characteristic is further brought out in the discourse by what Jesus said next: 'You lack one thing; go, sell what you have and give it to the poor ... and come, follow me.' In this context, the demonstrative / inee/ processes the contrastive nature of the conversational exchange between the younf man and Jesus in a better way than the more positive demonstrative /'Jr 2/. /i-/ does not occur with /een42/ 'that' as an exclusive enhancer. The reason for this, however, appears to be phonological and not semantic. As will be seen below, when /i-/ combines with a following vowel of a stressless syllable (which means that the vowel is automatically short in duration), the vowel is assimilated with the /i/. But in this particular case /i-/ would occur with a long vowel of a stressed syllable, /i + een42 /, hence no assimilation takes place. That is, */ien42/ or */iin42/ are both unacceptable. However, this does not mean that the exclusive enhancer marker /i-/ does not occur with 'that' in Mai. For such occasions, Mai speakers will use the Thai demonstrative /nan/ 'that' (high or falling tone) in combination with /i-/ to mean 'that particular one'. In Mark 4: 10-20 Jesus explains the parable of the sower who went out to sow seed on various types of ground. Some seed fell among thorns which choked out the plants that grew from the seed. Jesus explained that this part of the parable referred to those people who upon hearing the Gospel accept it but soon fall away because of delights in riches and so? p:xm inee inan want get this that 'Wanting this particular thing and that particular thing.' In Mai the demonstratives /inee inan/ may occur without accompany ing NPs as an idiom in much the same way that 'this and that' is used in English to refer to miscellaneous things or to 'odds and ends' in one's possession. The addition of /i-/ in Mai, therefore, enhances the particularising and constrastive nature of the 'things' referred to. Before leaving the above example, we should take note of an evident 179

PAGE 188

DAVID FILBECK relation between phonology and information processing that can be seen in the case of /inan/ in Mai. As in the case of /inee/ 'this one', it is also possible, informationally and logically, to have 'that one also. However, a phonological rule blocks the processing, more properly the encoding, of this information using the corresponding Mai demonstrative. Thus the Thai demonstrative /nan/ 'that', which has a phonological structure allowing the stressless /i-/ to occur as a prefix, is used to process the enhanced exclusive nature of 'that one in Mai. /i-/ also occurs as an exclusive enhancer to the word /soo/ 'other'. In such occurrences the combination /isoo/ further enhances the exclusive nature of the 'other' (thing, person) that is referred to. In Mark 4: 10, after Jesus had told the parable of the sower, many people returned to their homes, but ah isJJ at YJJp42 yeesuu they others locate around Jesus 'Still another group gathered around Jesus .. .' to hear his explanation of the parable. The combination /isoo/ enhances the 'otherness' of this latter group in contrast to the former group of people who returned home. 4.2. /i-/ plus Interrogatives /i-/ also occurs with the interrogative /ehee/ 'who?', becoming /ihee/ under the influence of the phonological rule mentioned above. The difference in meaning between /ehee/ and /ihee/ parallels that described above, for example, for /nee/ and /inee/. That is, /ehee/ is an unspecified 'who?' while /ihee/ is a more selective, particularising 'who?', resembling perhaps the now archaic 'whosoever'. In Mark 8: 27 Jesus asked his disciples ah SJJ khay an kayh ihee they other say I am who? 'What particular person do others say I am?' The enhanced or particularising /ihee/ is more appropriate for this context since there were rumours that Jesus was John the Baptist, or maybe Elijah, or perhaps one of the prophets. The unspecified /ehee/ in the above quote would not process this (particularising) type of information for the context in question. A second interrogative that /i-/ occurs with is /naa/ 'which?'. In the same manner as above, /naa/ is an unspecified 'which?' while /inaa/ is an enhanced 'whichever one it is'. In Mark 12: 28, a scribe asked Jesus a question that had caused much argument among scholars of the Torah. 180

PAGE 189

Information processing in M al discourse structure 1Jf:f:1J sJ;m42 inaa kayh sak lhia mphua1J word teach which is big above group 'Which particular teaching is more important than all the others.' Since there were several possible answers, the interrogative /inaa/ processes the particularising nature of the question, something which the unspecified /naa/ is unable to do. /i-/ occurring with an interrogative in effect throws the focus of the discourse on the question or, more properly, on the cognitive process of choosing among alternatives for the answer. And in choosing an alternative, others are excluded or eliminated. Hence, with the addition of the exclusive enhancer, /i-/, 'things are kept up front' in yet another way in Mai discourse structure, i.e. by eliminating or excluding other possibilities. 5. Keeping things up front: 'That's the way we talk' In this paper, I have surveyed several aspects of information processing in Mai discourse, comprising both conversation and recorded narratives. In surveying these aspects we saw where pronouns, demonstratives, relative pronouns, nominalisers, and a clitic, function to process certain types of information regarding content in a discourse or text. Moreover, I stated that this processing gives a particular characteristic to Mai discourse, a characteristic of this-and-no-other, or of 'keeping things up front' vs. 'keeping other things in the background'. We also included examples from Christian scripture to show how productive this characteristic is even in translating from another language into Mai. After revising a certain text from the Gospel of St. Mark, for example, to include more occurrences of /ee/ and /ee42/ according to the discourse principles described in this paper, the text was read aloud to others for reactions. On hearing the text for the first time a Mai woman responded with the comment: 'That's sounds just like we talk'. Among others, the revisions met with greater acceptance and satisfaction. In short, 'keeping things up front' is a favourite style of communication among the Mai. Of course, as was also seen above, a complete text or discourse may be unmarked with regard to focus and/or exclusivity, but such a way of communicating appears not to be very exciting. Such an unmarked discourse will transmit content well enough, but for 'real performance' in Mai, a discourse should also process information about content. REFERENCES Fil1,eck, D. 1972. Tone in a dialect of T'in. Anthropol. Ling. 14, 111-18. 1976. Toward a grammar of relative clauses in T'in. Tn Austroasiatic Stud. 1 (Oceanic Ling. Spee. Pub!. 13) (eds.) P. N. Jenner et al. Honolulu: Univ. Hawaii Press. 181

PAGE 190

DAVID FILBECK Fillieck, D. Smith, N. V. & Wilson, Deidre 1978. T'in: A historical study (Pacific Ling. B, 49). Canberra: Austral. Nat. Univ. 1979. Modern linguistics: The results of Chomsky's revolution. Bloomington: Indiana Univ. Press. 182

PAGE 191

LES DERIVES DESIDERATIFS EN KHMER Saveros Pou 0. Le mot 'desideratif est emprunte a la grammaire sanskrite pour des raisons de commodite et de simplification terminologique. L'on evoque par exemple sk. cikitsli 'medicine' < cit 'observer, penser', ou mimlirrzsli 'un systeme de philosophie' < man 'penser'. La langue cambodgienne possede des derives semblables, mais en verite ils ne procedaient pas de la meme fa~on qu'en sanskrit tant dans Jes concepts que dans Jes formes. 1. Ils sont attestes dans l'usage des Jes premiers documents epigraphiques khmers. Relevons quelques exemples avant d'en analyser le type: cliy 'disperser' > cicliy 'detruire' dab 'heurter' >padab 'enclore un espace par des murs'1 darrz 'se percher, se tenir' > phdarrz 'dormir' (Jang. princ.) dap 'barrer' >pandap 'enfermer' dau 'aller, avancer' > pradau 'instruire' day 'porter, soutenir' > phdliy 'renforcer le soutien, servir un ma1tre'2 aul-dval 'porter sur la tete, >dadval 'recevoir' parler a un etre sacre' > pandval 'commander' garrz 'joindre, reunir' > pangarrz 'joindre Jes mains d'adoration' glip 'convenir, plaire' > phglip 'desirer obtenir Jes faveurs de qq'un' pat 'perdre, dispara1tre' > parrzpat 'detruire completement' rac 'use, mine' > vrac '(d'un elephant) tuer sauvagement' ( sra)sir 'ecrire' > prasir 'ecrire dans un but elogieux'3 flip 'etre bout a bout' >stlip 'desirer suivre, ecouter' tin 'savoir' > tantin 's'informer, s'enquerir' tval 'tomber a la renverse' > phtval 'tuer dans un combat' vat 'faire un geste circulaire' > cvat, chvlit'borner une terre' I. D'ou moderne phda!J 'un espace enclos de cette fa~on: la maison'. TI convient de rappeler que certains cas de derivation sont de nature exclusivement semantique, l'affixation ne jouant aucun role dans la production des derives, tels que: crat' 's'appuyer du bras sur' > crat' 'canne, baton' (des le vieux khmer), ou bien dap' 'barrer, enclore' > (*dap') note maintenant dab 'enclos, camp de soldats, armee'. 2. Le derive secondaire en est panday 'endroit de refuge; fort, fortification'. Cf. Jes toponymes Banteay Meas, Banteay Chmar, etc. 3. Cf. khmer moyen praser et moderne prasoer /p:isa:i/ 'digne d'eloge, admirable'. 183

PAGE 192

SAVEROS POU ven 'long' vyat 'vrai' > sven 'parcourir un long chemin; rechercher' > parrzvyat 's'adonner a la veracite; confirmer' Au premier abord, ii semble qu'on ait affaire a des derives de types couramment definis, a savoir des frequentatifs, des intensifs et similatifs, et des causatifs. Or, en plus de ces traits grammaticaux, nos derives ont un caractere fondamental commun tres net du fait qu'ils expriment un 'desir' de la part du Sujet, ou !'on decele desir proprement dit, souhait, intention et meme pretention. Ce trait qui apparait deja comme distinctif va s'illustrer de fa9on copieuse dans le lexique moderne, y compris le lexique moyen (XV"-XVIIIe siecle), et de fa9on tres originale dans ce dernier. Aussi le regroupement de ces derives s'impose pour un nouvel examen (a la fois dans le cadre de la langue ecrite et orale). 2. En morpho-phonologie, ils ne presentent aucun signe distinctif qui les separe des derives de tout le systeme. 2.1. Les mots de base sont en general des verbes. Par ex.: gat' 'mesure, ferme' > phgat' 'prendre une attitude kat' parrzn 'noter' 'barrer' reservee, pratiquer une observance' > sm;kat' 'presser, opprimer' > pa,rzparrzn 'empecher de voir, rendre invisible' Mais des substantifs ont aussi servi de base, bien que plus rarement, comme: bot 'le ma1s' cor 'voleur, brigand' ruot 'couche, etage' > pa,rzbot 'faire cueillir son ma1s par un naff, Jui ravir son ma1s, duper qq'un'4 > pai'icor 'traiter qq'un (surtout une femme) de voleur, l'injurier vi lai nemen t' > pruot 'conjuguer des efforts dans une entreprise' 2.2. Les exemples du vieux khmer (sup. I) revelent trois modes de derivation, qu'on va retrouver dans toute l'histoire de la langue, a savoir: redoublement: cica.y, dadval, tantin ... 4. Pour saisir cette derivation assez insolite et la semantique mirifique du derive, ii faut tenir compte de !'anecdote qui en etait l'origine. Une jeune fille coquette fit cueillir le ma"is de son champ par un sou pi rant naff contre promesse de sa main, promesse qu'elle ne chercha point a tenir aussitot la tiiche finie. Un tel gan;on est dit prus pal)tfil bol 'gan;on rane de ma"is', et le derive pa,r,bol signifie 'tromper, duper' (cf. Aymonier, 1984: 22, 93). Cf. d'autres details, inf, n. 6. 184

PAGE 193

Les derives desideratifs en Khmer prefixation labiale: pada/:z, phdiiy, phtval ... prefixation de /s-c-/: sven, cvat. Dans le cas precis des derives desideratifs, je regroupe ces deux phonemes, me basant simplement sur leur frequente alternance rencontree en khmer comme dans tout le groupe mon-khmer. 3. Jusque la, il semble qu'on puisse classer ces derives dans les categories deja etablies, et evoquees a sup. I, comme des sortes de sous-types; par exemple cicliy comme un frequentatif de cliy 'eparpiller', ou bien sven comme un intensif de ven 'long'. Mais cela ne suffit pas en fin de compte quand on procede a un examen global de ces derives dans le contexte derivationnel entier, et surtout quand on avance dans le temps vers le khmer moyen. 3.1. Par exemple, tantilJ, derive de tin 'savoir', signifie 'desirer savoir, s'informer', de fa9on univoque. De dau 'aller', pradau, sur les bas-reliefs d'Angkor Vat ou il apparait, n'est pas un simple causatif: 'faire aller' certes, mais dans un sens special, intellectuel et moral, d'ou 'desirer le progres, instruire, eduquer'. De vat 'faire un mouvement circulaire', cvat n'est ni un intensif ni un similatif, car, dans les inscriptions, il signifie 'delimiter un terrain de fa9on deliberee', dans un but determine.5 Et enfin, cicliy n'est pas simplement comparable a notre moderne kakliy 'gratter a petits coups', derive de kliy 'fouiller'. 3.2. Le cas du derive a prefixe labial panhoer 'faire voler' (b) une construction syntactique schematisee par: oy + Verbe ou dhvoe oy + Verbe Ainsi: phjk 'boire' > A oy B phjk, 'A fait boire B'; tandis que le derive parrzphjk est desideratif, 'faire boire qq'un pour le rendre ivre'. De meme: khuc 'abime, mourir' > A oy B khuc, 'A abime B' sans premeditation; tandis que pankhuc est un desideratif, 'debaucher, calomnier' (cf. inf 3.3.). 3.3. C'est en khmer moyen qu'on voit les traits particuliers de ces derives ressortir de fa9on indiscutable et decisive. Prenons d'abord quatre 5. Le derive secondaire canvat est devenu en siamois ciinhviit 'district'. 185

PAGE 194

SAVEROS POU exemples typiques de derives a prefixe labial /pa-/, /baN-/, repondant aux formes graphiques pa-, pra-ou pa + Nasale-.6 Les valeurs generalement reconnues par Jes Iinguistes a ces prefixes sont soit causative soit reciproque, ce qui n'est pas avere dans Jes exemples qui nous preoccupent et que voici: cor 'voleur, brigand > pancor (cf.2.1) 'traiter qq'un (surtout une femme) de voleur; l'injurier vilainement' da,:i ptandli 'souhaiter le chatiment a qq'un, Jui souhaiter beaucoup de ma!' dlisli 'esclave' > ptlisli 'souhaiter la servitude a qq'un; le maudire' dos 'faute, peine' > pandas 'imputer une faute a qq'un; le reprimander' II est difficile de parler ici d'une expression de causatif car on deformerait la realite Iinguistique sur Iaquelle tous les documents s'accordent sans aucune exception. 3.4. Et que dire de quelques exemples de redoublement qui d'une part prouvent la productivite de certains mots de base, et d'autre part constituent de tres beaux specimens de polysemie? Tels: cel:z 'savoir > cacel:z 'se croire savant, fort; etre -caces obstine' -dfli 'en travers' > dandfli 'se tenir en travers du chemin, i.e. attendre l'arrivee de qq'un avec impatience' foe 'au-dessus' > /aloe 'se croire superieur; se prendre bctement au serieux' !iii 'etre au courant' > fa,:itjn 's'enquerir; s'informer de la situation d'une jeune fille, la demander en mariage' foem 'le debut, principal' > fa,:ztoem 'se croire le premier; disputer qqch. a qq'un' 6. La n.4 a laisse pressentir le role de la litterature (ecrite et orale) dans la formation des mots, d'ou le rapport etroit entre litterature et grammaire. Et cela n'est que trop vrai dans le cas des derives. Beaucoup d'entre eux furent crees par des ecrivains, des poetes et artistes qui-est-il besoin de l'expliciter-maniaient la stylistique et le folklore avec dexterite pour exprimer leurs idees, leurs emotions, parfois au detriment de certaines regles: cas de licences (litteraires et poetiques). Certaines creations plurent rapidement au public, qui Jes adopte et les propage. Le mouvement part ainsi d'une sphere reservee de locuteurs, mais un consensus etait necessaire pour faire passer les creations dans !'usage national. Dans ces conditions, on saisit que la structure de tels mots echappe parfois aux regles grammaticales qu'on souhaiterait appliquer en linguistique. Pra-n'est pas necessairement un prefixe de reciprocite, ni paN-un prefixe de causatif. Ils sont confondus dans noire sujet, et c'est le role dominant des arts, au sens large, qui est responsable de cette confusion des formes et d'autres irregularites. Voir des exemples dans la nomenclature, inf, 4 186

PAGE 195

Les derives desideratifs en Khmer Aucun doute n'est possible sur la nature et la valeur de ces derives. Ce qu'ils ont de commun avec vieux-khmer cicay 'detruire', c'est leur expression fondamentale d'un desir oriente vers un but precis, de souhait d'une action a accomplir par soi ou par un autre. Ce trait distinctif clarifie, nous pouvons maintenant examiner des exemples modernes, parfois semantiquement complexes, qui sont toujours en usage a l'heure actuelle.7 4. Plusiers modes de classification sont possibles et aussi defendables l'un que l'autre. A cote du critere morphologique proprement parrzbarrzn /b:impeaTJ/ 'calfater'8 ban /pi:in/ 'grimper sur' > parrzban /b:impi:in/ 'pietiner volontairement; bar /pi:i/ 'se cogner' can' /coTJ/ 'vouloir' carrz /cam/ 'attendre' cap' /cap/ 'saisir' ceb /ceh/ 'avoir des connaisances' citt /cyt/ 'le coeur' cies /ci:ih/ 'eviter' CUC /coc/ 'mettre le doigt sur' -cilv /cW/ 'aux levres mobiles' chit /chYt/ 'raser de pres' chot /chaot/ 'nai'f > parrzbar /b:impi:i/ > phcan' /pcoTJ/ > pracarrz /pxam/ > pancarrz /b:ij1cam/ > pracap' /pxap/ > caces /cxeh/ > paiiceb /b:iJ1ceh/ > pracitt /pxYt/ > pancies /baj1ciah/ > phcuc /pcoc/ > cae cilv /cxyw/ > panchit /b:Jj1ChYt/ > panchot /baj1chaot/ blesser, abuser de' 'manquer d'egards, heurter, blesser' 's'appliquer entierementa une tache' 'etre poste a un endroit, s'attacher a un objet' 'mettre un objet en gage' 'jeter une faute sur qq'un, l'accuser' 'se croire fort, etre obstine' 'inciter, pousser a mal agir' 'se soucier de' 'parler de fa<;on indirecte' 'questionner sans detour ni laisser d'issue' 'se livrer aux commerages; servir d'entremetteuse' 'parler a mots couverts' 'se jouer d'un naiJ' 7. Des exemples presentes plus haut ne seront pas rappeles sauf en cas de multi-derivation. D'autres, en particulier des derives par redoublement, avaient deja ete examines par moi meme (Pou 1977: 126). 8. A gloser ainsi: 'desirer boucher une entree; empecher une impregnation ou une infiltration, d'oii calfater Jes joints, Jes trous .. .' On procedera de la meme fa,;:on dans l'analyse des exemples qui suivent, a savoir qu'il faut attributer a chaque derive un premier degre, explicite ou implicite, de semantique represente par 'desirer, vouloir, souhaiter' dans un sens particulier, lequel se developpera le cas echeant dans d'autres directions par metaphore ou par metonymie. 187

PAGE 196

SAVEROS POU dan' /bn/ 'doU)(, tendre' > dandan' /t':mbn/ 'se montrer modeste; s'afTaiblir, decliner' > sandan' /s:inbn/ 'adopter une maniere douce, tendre' dap-diip' 'aplati contre' > dandap /t:inb:,p/ 'marcher en catimini' /bp/ dar /b:,/ 'frapper' > pandar /b:inb:,/ 'accompagner en cadence' dau /t:iw/ 'aller, avancer' > dildau /t:it:iw > tut:iw/ 'repandre partout, generaliser, general' >pra{au(3.1) diik' /teak/ 'poser un piege' > sdiik' /steak/ 'poser une embuscade' diirtZ /b:im/ '(des coups) > dandiirtZ /t:inb:im/ 'onduler gracieusement nombreux et le corps et Jes membres' repetes' diip /ti:ip/ 'bas' > dandiip /tmti:ip/ 's'abaisser doucement, se montrer modeste' den-doen 'desirer > danden dandoen 'se montrer hautain, /t\!\!IJ-t331) / avidement' /t:in t\!\!IJ -t:int:i:iIJ/ fier, al tier' > pandoeiz /b:int:i:iIJ/ 's'enthousiasmer, s'exalter' den /tl!Jl/ d'une cadence > danden /t:int\!Jl/ 'reciter rapidement pour rapide' memoriser' die -dee /tic/ 'petit, inferieur'9 > pradee /p:itic/ 'avilir qq'un, le couvrir de maledictions' djm /twm/ 'atteler, > dandjm /t:intwm/ 'se mettre cote a cote' 10 conjuguer' > phdjm /ptwm/ 'placer cote a cote (rituellement); comparer' does /t:i:ih/ 'se heurter a un > dadoes /t:it:i:ih/ 'faire obstruction' obstacle' -duy, -thuy etre en saillie' > prathuy /pothoy/ 'foncer en avant; tenter la /tuy, thoy/ chance' -due /tuuc/ 'attirer' > dadile /t:ituuc/ 'demander avec insistance, supplier' > sdile /stuuc/ 'prendre au moyen d'engin: pecher' -dur /tuu/ 'le sommet'* > dadur /t:ituu/ 's'envelopper la tete d'une etolTe' gaiz' /bi)/ 'qui phgaiz' /pbIJ/ 'subvenir a, pratiquer permanent' une observance' }lit' /c:,:it/ 'filtrer l'eau' > Jan}iit' /c:iJlc:,:it/ 'repeter ce geste au moyen d'un panier pour pecher, ou pour ramener Jes esprits vitaux'11 9. Cf. le congenere m6n signifiant 'esclave'. 10. On notera aussi la forme dialectale sandjm. 11. Le corps humain heberge 19 esprits vitaux (bra/jn) qui s'en echappent partiellement en cas de sommeil ou de maladie, d'ou la necessite de Jes 'ramener' en vue de la guerison. Allomorphe de du/ dval ( cf. sup., 1) 188

PAGE 197

Les derives desideratifs en Khmer jik /ciik/ 'creuser' > jajfk /cxiik/ 'interroger de fai;on insinuante' jo!J /~h/ 'piquer avec une > panjo!J /b;;iJl~h/ 'riposter ii des critiques de pointe' fai;on acerbe, blessante' }or /c99/ 's'enfler' > paiijor /b;;iJ1c99/ 'flatter, flagomer' Jun /cuun/ 'offrir' > jarfjfln /c;;iJlCUUn/ 'deplacer, transporter Jes chases' > parfjfln /b;;ijlCUUn/ 'envoyer vers une destination precise ou dans un but determine' kak' /kok/ 'tremper, etre > pankak' /baIJkok/ '(rite) ondoyer' trempe' klic' /kac/ 'casser, couper' > panklic' /baIJkac/ 'calomnier, honnir' klin' /kan/ 'tenir' > praklin' /pakan/ 's'attacher aux formes, etre ii cheval sur les principes' klir /kaa/ 'proteger' > panklir /b:nJkaa/ 'parer ii une eventualite' klit' /kat/ 'couper, > sklit' /skat/ 'prendre un chemin court trancher' pour retrouver; interrompre un interlocuteur' kin /hn/ 'ecraser, > sankin /saIJkYn/ 'opprimer, oppresser' moudre' kin /kYIJ/ 'rigide' > prakin /pakYIJ/ 'exiger severement sans compromis' la /b'J/ 'essayer, tenter' > Jan/a /c';Jn]'J'J/ 'eprouver, taquiner' >sla /sloo/ 'bouillir plusieurs matieres ensemble'12 laen /lr.r.IJ / 'lacher' > plen /pleeIJ/ 'noyer en usant d'un lest' > banlaen /panlr.r.IJ/ 'brutaliser par des coups ou des mots'13 -lak' /bk/ 'plonger dans > bhlak' /pbk/ 'tremper ses levres; un liquide' gouter ii un aliment' -Ian' /Ion/ 'egare, > plan' /pion/ 'attaquer ii l'improviste, desoriente' piller' Ian' /bIJ/ 'se noyer' > ban/an' /p;;inbIJ/ 'noyer qq'un deliberement' Ian' foe 'loin-brule' > Ian/an' lanloc 'se consumer de nostalgie' /b1J-l99c/ /J';JnbIJ ]';Jn]99c/ lap /b'Jp/ 'agir doucement, > jhlap /cl'J'Jp/ 'espionner' secretement' llik' foe 'en bas-en haut' > lallik' !aloe 'se comporter de maniere /leak-laa/ /laleak ]';J]a';J/ legere, ecervelee' larrz 'aux traits > panlarrz /b;;inlom/ 'se camoufler, duper' /b';Jmlom/ indecis' 12. On peut gloser par: 'faire un essai sur plusieurs matieres, obtenir une solution a partir de produits dissous ensemble'. C'est la le sens premier de notre derive sla, qu'on retrouve dans le fameux derive de celui-ci, samla /s;imloo/, generalement rendu par 'la soupe'. 13. A entendre: 'lacher des coups dans le desir de brutaliser, d'intimider'. 189

PAGE 198

SAVEROS POU lee /lr;c/ 'apparaitre' > jhlee /clr;c/ 'presser de questions pour faire sortir la verite' fen /lr;r;TJ/ 'jouer, s'amuser' > panlaen /b:mlaeTJ/ 'distraire par des jeux ou de la musique' lie /lie/ 'etre immerge' > banlie /p'Jnlic/ 'enfoncer dans l'eau de force, vilipender, detruire' foe jl'J'Jj 'au-dessus' > !aloe (cf. 3.4.) > bhloe /pl'J'Jj 'se croire important: etre niais, bete' foes /l'J'Jh/ 'depasser, > phloes, banloes 'tenir des propos exageres, surpasser' /pla'Jh, p'Jnl'J'Jh/ contraires aux faits' 14 fut /luut/ 'surgir, croitre' > banlut /p'Jnluut/ 'operer un avortement' /ap' /lop/ 'faire un tour, > panlap' /banlop/ 'confondre qq'un; egarer, retoumer' detoumer' moel /maal/ 'regard er' > pramoel /pamaal/ 'regarder a distance, observer de loin' man /mian/ 'avoir, posseder' > braman /p'Jmi'Jn/ 'donner un avertissement' mue /muc/ 's'immerger' > pramue, eramue 'enfoncer qq'un dans /pamuc, camuc/ l'eau' noey /TJ'J'Jyj 'lever la tete' > branoey /parJaay/ 'se montrer indifferent, insouciant' pan' /boTJ/ 'jeter, lancer' > pa,ripan' /b'JmboTJ/ 'abandonner, se debarrasser de' pa,rin /baTJ/ 'barrer la vue' > parr,pa,rin /bambaTJ/ 'empecher de voir, rendre invisible' 15 pit /hvt/ 'fermer, clore' > pa,ripit /bambvt/ 'cacher soigneusement; empecher de sortir et circuler' po!J /boh/ 'jeter' > parr,po!J /bamboh/ 'lancer des paroles blessantes' pos /baoh/ 'balayer' > papas /babaoh/ 'tapoter affectueusement, caresser' pul--puol 'appeler, > papuol /b'Jbu'Jl/ 'inciter a une action, /boo!bual/ conjurer' proposer' pa!J /pah/ 'heurter, cogner' > parr,pa!J /bampah/ 'heurter, se cogner, avec intention blessante' rat' jr'Jtj 'courir, fuir' > banrat' jp'JT]r'Jtj 'enlever, ravir une jeune fille' ran /ri'JTJ/ 'forme' > bran, pa,ribran 'donner une autre forme; /priaTJ, bampriaTJ/ falsifier' 16 14. On trouve egalement dans !'usage parr,phloes /b;impla;ih/. Cf. aussi, brlin et parr,brlin. 15. Exemple deja relate (sup., 2.1) mais qui gagnerait a etre compare avec parr,blirr,n (premier ex. de la liste). Les mots de base, quasi synonymes 'barrer la vue', produisent de faux synonymes dans Jes derives, en !'occurrence notre parr,plirr,n est courant en magie pour 'se rend re invisible'. 16. Les textes en khmer moyen montrent le meme usage de ces deux derives. 190

PAGE 199

rap' /roap/ ray /riay/ rit /rrut/ riep /riap/ roep /raap/ rut /ruut/ sap /saap/ sol;i /soh/ fal]'I /dam/ fen /deJl/ toy /daoy/ fuc /dooc/ Les derives desideratifs en Khmer 'compter, > prap' /prap/ conter' 'repandre' > banray /parJriay/ > sray /sraay/ 'serrer' > prit /prrt/ 'ranger' > priep /pri.ap/ 'remonter, se > sroep /sraap/ remettre' 'tirer sur un fil > srut /sroot/ dans une coulisse' 'sans saveur' > pansap /bansaap/ 'fini, epuise' > sasol;i /sasoh/ 'planter dans le > phtal]'I /pdam/ sol' 'poursuivre' > pa,:tten /bandeJl/ 'suivre, accepter > pa,:ttoy /bandaoy/ une proposition' 'ressembler' > pratuc /padooc/ 'informer' 'faire du desordre' 'delier, defaire un noeud, resoudre' 'se montrer severe, ri.goureux' 'placer ensemble; comparer' 's'exciter pour qqch.' 'accelerer le pas pour arriver vite' 'neutraliser un poison, une attaque malicieuse' 'demeler; parler pour ne rien dire' 'faire entrer dans l'esprit; faire des recommandations' 'chasser, expulser d'un endroit, congedier' 'laisser faire, laisser toute liberte' 'comparer une chose a une autre' vaen /wf.f.TJ/ 'long' > panvaen /barJwaerJ/ 'prolonger un trajet; detourner du but, egarer' > svaen (cf. 1.) Cette nomenclature n'est pas exhaustive pour une raison evidente, a savoir l'espace limite d'un article. Et le temps aussi constitue un facteur important, car ii reste a approfondir l'exploration de tout l'appareil derivationnel du khmer si complexe et, je dirais meme, sophistique, en interrogeant tous Jes textes et en ecoutant simplement Jes locuteurs. Mais ces premieres notes succinctes, je Jes offre deja, ainsi que mes voeux Jes plus sinceres, au grand linguiste-philologue qu'est Monsieur H. L. Shorto. REFERENCES Aymonier, E. Pou, Saveros 1984. Notes sur les coutumes et croyances superstitieuses des Cambodgiens (commente ... par S. Pou). Paris: Cedoreck. 1977. Etudes sur le Ramakerti (XVI" XV/I" siecles) (Publ. Ee. fr. Extr.-Orient, H 1). Paris: Ee. fr. Extr.-Orient. 191

PAGE 201

A DIACHRONIC SURVEY OF SOME KHMER PARTICLES (7TH TO 17TH CENTURIES) Judith M. Jacob Introduction In modern spoken Khmer, grammatical relationships between the clauses of a sentence may be understood without the use of clause particles. Intonation and pause, together with the context in which the sentence is spoken, make the meaning clear. Thus, when a speaker says /mtay dau phsar, kun dau phan/, 1 intonation and pause mark off two separate phrases and the hearer will know from the context whether he is being told that it is 'because', 'if or 'when' the mother is going to the market that the child will go too. Particles to express 'because', 'if and 'when' are available but are not felt to be necessary. In modern written Khmer, the roles of intonation and pause are played by punctuation, in the form of full stops and of spaces between phrases; context still plays an important role, more so than in European languages, and particles occur frequently. When we turn to the inscriptions and the earliest non-inscriptional texts, punctuation is minimal and the contexts are unfamiliar to us in comparison with the communities for whom the texts were composed. On the inscriptions, punctuation occurs rarely, usually in the form of a circle indicating the end of a phrase. In poetry, the only formal pauses indicated are at the ends of verses and stanzas. The function of particles is, therefore, very important to our understanding. This paper is the result of the writer's interest in the form, meaning, grammatical usage and provenance of the particles occurring in Old and early Middle Khmer. The period under review saw great change and development in the written Khmer language. The early texts have been regarded as struggling attempts to write prose; the complexities of Angkorian sentences seemed to show how difficult it was for the language to express the growing complications of Angkorian life. It is only on the Middle Khmer inscriptions that an easy style was found. In order to consider the use of particles in these ten centuries, the following texts, (given here with an abbreviation which will be used throughout), have been used: P.A. Pre-Angkorian inscriptions, seventh-ninth centuries A.D.; A. Angkorian inscriptions, ninth-fourteenth centuries A.D.; I. The transliteration used for all Khmer citations is that of Lewitz (1969). For modern pronunciation the writer's transcription is used (Jacob 1968). 193

PAGE 202

JUDITH M. JACOB Mid. Middle Khmer inscriptions. Special attention was paid to texts dated between A.D. 1574 and 1630 (qqv. Lewitz 1970-72, nos. 1-16c). 2 R. Reamker, the Khmer literary version of the Rlimiiya,;a. Special attention was paid to the first five Parts or 2542 stanzas, much or all of which was probably composed during the sixteenth century A.D. (qv. Pou 1977a: 59). L. Lpoek Angar Vat. Poem about the creation of Angkor Vat, dated A.D. 1620.3 C. The old Cpiip'. Didactic moral poetry, regarded by Pou and Jenner (1979: 134) as dating from before the mid seventeenth century. Mod. Modern, twentieth century. Examples are either colloquial or from twentieth-century writing. The P.A., A., and Mid. texts of the seventh to seventeenth centuries are pre-eighteenth-century dated inscriptions on stone. They constitute the whole body of Khmer literature in prose. Although all are associated with religious foundations, the subject matter and style develop from the terse pre-Angkor texts chiefly concerned with lands, duties and provisions, to the Angkor texts in which more information is given about individuals or about historical events, territorial rights, disputes, etc., and to the Middle inscriptions in which the good works and Buddhist fervour of royal and other persons are described in fluent prose. The Reamker represents, along with the old Cpiip ', the earliest extant Khmer poetry. Though undated, it may be assigned to the early Mid period (at least as far as the early part of it is concerned). The oldest stanzas may well precede the earlier Mid. inscriptions. The first five Parts (i.e. the first 2542 stanzas) have been examined thoroughly and the results used for this paper because it was felt that the text would supply useful additional information, being a continuous narrative with many colloquial or semi-colloquial passages and having a more varied content and style than the inscriptions. It was not expected that the Lpoek Angar Vat would produce as great a variety of syntactic constructions as the Reamker, because insofar as it has a story, the narrative style is very simple and the non-narrative part of the poem is, on the whole, descriptive. The reason for its inclusion here as a text for examination is that it is now dated and the date, A.D. 1620, falls within the Middle period chosen for this study. The old Cpiip', though assigned to a suitably early period, have a restricted range of syntactic constructions owing to the gnomic character. They were included for the sake of completeness. The results of this research are presented in three sections. The first consists of comment, under various headings, on all the particles studied; 2. The so-called "Middle" Khmer period is from the 16th to the 19th centuries A.D. Only the early Middle inscriptions (for which see Lewitz I 970-72; Pou 1977b; Khin Sok 1978, 1980a, i980b) are relevant to this paper. 3. See Pou (1975a) for dating, Aymonier (l 878) for text. 194

PAGE 203

A diachronic survey of some Khmer particles the second comprises six tables which show the grammatical usages of all the particles at various periods; the third gives citations from a text for each particle with each grammatical usage in each period. The indications of word categories are made in accordance with the writer's analysis of Khmer (Jacob 1968: 330-2). SECTION I: Comment Forms of the particles As may be seen from looking at the examples, many Khmer particles have operated in the language constantly from the seventh to the twentieth centuries. Some, eg. nel:z, nol:z, nai and sot are still spelled exactly as they were in the P.A. period. Others have undergone spelling changes in accordance with regular phonetic or orthographic developments.4 However, not quite all the changes of spelling can be explained away. The form njn would not be anticipated from old Khmer nu. It has developed alongside nu nil. There was confusion over the word already in pre-Angkor Khmer, when it was written nau occasionally. Modern Khmer uses the form njn, pronounced according to modern rules, except in formal circumstances, when nilv may be used for 'with' instead of njn. nilv is, however, then pronounced exactly like the verb and particle nau. The particle pi also develops in a puzzling way. The modern forms of the combinations foemp[ 'so as to', sump[ 'even (with following noun)', gapp['it behoves (one) to', are as expected (all are heralded in Mid. or C.) but, alongside pi in the combination dol:z pi in Mid. (See Table 1: Condition), is the form peh, with the same meaning 'if. In the Mid. period, open syllables were spelled with final -h. The form we are considering therefore is pe, Mod. poe 'if. Were the forms pi and pe, future p[ and poe, variants of the same word? If so, why did pi have such an unprecedented variant? Lewitz (1972: p. 110, n.3) does not comment on the relationship, if any, between the two words pi and pe, but relates pe to Mid. poe. Compounded particles Various combinations of particles occur, seeming to answer a need for reinforcement or greater precision, e.g. the relative pronoun man occurs in the P.A. period in combination with ta, another particle which fulfils the 4. Improved devices, copied from the Thais, for representing different vowels, produced the changes from fey to /oey, dep to doep, tempi to /oempi, hey to hoey and tel to /ael. Vowels in open syllables were probably always pronounced long but were rarely written so; the more modern spellings of nu, ra, ru, pi, and a,pvi mark the long vowel. The diphthongisation of vowels in Angkorian Khmer, followed by a return to a pure vowel, is shown in the spellings of vo111 -vva111-bu111, sya,i sin, lot:, -/vat:, -fut:,. Old Khmer initial I became an imploded, voiced dental consonant, spelled with /. The diacritic was added in modern times to mark the characters I and k ofT as complete words. v was frequently written for the initial consonant which was later written b. Finally, the vowel o became an inherent vowel in many modern words and was shortened before h. Hence, phon pha,i and daha-dot:,. Old spellings with final h instead oft:, account for the transcription daha. The transcription kala is due to omission of the killer sign in the Khmer spelling. 195

PAGE 204

JUDITH M. JACOB relative pronoun function. man also occurs in combination with gi, both as a relative pronoun and as initiator of discourse, 'i.e., then .. .'. gi also seems to reinforce pi in its occurrences in which it may express either purpose or result. By the Mid. and Modern periods, however, pi is combined with tem and clearly then expresses purpose. In literary Khmer, pf'as it were' is much used to introduce attributive verbs. Gradually pfin this use was combined with hiik' and file. hiik' p[ file became current in modern times as a literary and poetic way to introduce a comparison. In C. silmpf 'when; even' and gappf 'it behoves' are attested. The use of nau as a final phrase particle, (f.) meaning 'still' is in evidence in modern Khmer only in combination with /oey and usually in negative statements. One sequence of particles which must not be construed as a combination of constituents which reinforce each other is vvarrz tel -burrz fael 'never'. Here, both particles contribute different meanings 'not' and 'one who', respectively. Changes of word-order category A sphere in which changes seem to have taken place both within one period and from one period to another is that of the word order in which some forms occur. In the case of particles5-which are catalysed according to their position in relation to verbs, nouns and phrasesthis amounts to a change of word category.6 Five areas of change (with a possible sixth) have been observed: 1. Position of the demonstrative particles nel;i 'this, these' and nol;i 'that, those' Tables 3 and 4 and the examples show these particles preceding the noun in the pre-Angkor period but following it from the Angkor period onwards. Although the movement from pre-Angkorian to Angkorian Khmer involves a change of area as well as of time, such a reversal as this does seem unusual (and the P.A. order is uncharacteristic of MonKhmer). Jenner (1982) discusses the possibility that nebfnob (and often gi gui nebfgi gui nob) placed before the noun are, in fact, the 'phrasal head', while the noun is attributive to it. neb sre or gi neb sre would then be 'These are the ricefields'. It is tempting to see some such explanation of the word order, though one has also to explain sentences such as oy gui nob sre sot (K.79.18) 'gives this ricefield also' where sre is required as the object of the verb oy. Another possibility would be that sre, if given modern punctuation, would be in parentheses or follow a colon: 'gives this (the ricefield) also'. Whatever theory is offered, however, it is very difficult to explain why there is no occurrence at all in P.A. of what we would call the normal order. 5. A change of word order has also taken place in Khmer in the sequence of numeral and classifier. (See Jacob 1965: I 61-2). 6. That is to say: a recognisably consistent form with a recognisably consistent meaning has to be regarded as performing more than one grammatical function. 196

PAGE 205

A diachronic survey of some Khmer particles 2. Position of ru-rilva-ril 'like, as' Tables 3 and 4 and the examples show that this particle occurs within the pre-Angkor period both preceding and following the noun. Perhaps the English use of the word 'like (such as)' which may also occur before and after a noun, e.g. 'like the moon' and 'flower-like', is comparable? 3. Position of ni-ni 'with reference to; at' It seems possible that ni -n[, which occurs in both Tables 3 (P.A. and A.) and 4 (A. and Mid.) with meanings which could be equated, is another example of changed position. Occurrences of ni following a noun are rare and very restricted. They involve the words mu!J 'nose' and vnek 'eyes'. It is to Pou (1976: 340-1) that is owed the interpretation of the phrases mufJ ni and vnek ni as 'before' and 'in the presence' respectively. 4. Position ofhey-hoey 'already; and then, and now' In both Mid. and Mod. Khmer, as Tables 1 and 2 show, this form carries out two functions: that of marker (m.; occurring clause-initially), and that of final phrase particle. 5. Position of nai 'of' The particle nai has the same spelt form, the same translation 'of and the same pre-nominal use throughout Khmer, as Table 3 shows, but it also occurs phrase-finally in R., L., and C. with back reference to the subject of the sentence: 'of him/her/them', as is illustrated in the examples. In Khmer poetry, sequences of alliterating words which do not convey much additional meaning to the sentence often occur at the end of a verse or stanza, where they are useful as rhymes or to provide the required number of syllables (qv. Jacob 1979: 124-5). Such sequences are ban' bek 'very', krlls' krael 'very', ne!J ,:ill /no!J ,:ill 'this, these'/'that, those'. nefJ nai/ nofJ nai also occur in poetry. There are such occurrences with nai in R., L. and C. 6. Position of ai (?) R. also provides an example of ai, normally a pre-nominal particle (pre n.p.), in phrase final position: R.1.29 .1-2. doep an jzoe bra!J staen thlai, bans bra!J nllriiy( 1J) ai, isilr nai lok traitll 'Then will I believe you, honoured sir, to be in the line ofNarayal).a, lord of the world, in the traita'. Foreign borrowings and Thai comparisons ukk < Skt. ukra 'also' occurs phrase-finally in Khmer and has, therefore, been entered on Table 2 as f. It is attested only in the pre-Angkor period, whereas sot performed a similar role throughout all periods. kala, kal < Skt. kllla 'time' is current throughout Khmer as a noun but also may be catalysed as a marker 'when' in the Angkor and modern periods. 197

PAGE 206

JUDITH M. JACOB ta, fa is a general dependent particle (g.) which may occur before words of any category, and may also act as a marker (m.). It seems just possible that ta was related to Skt. tii, the base of the demonstrative pronoun, as the present writer suggested on an earlier occasion (Jacob 1977: 166-7). kiir, kli. Occurrences on P.A. inscriptions of the form ka show no similarity to the particle kiir, ka. Some may be construed as ka 'to construct'. Some may be a shorthand form of kiiurri 'serf. In A. Khmer, the writer has noted occurrences of ka on only one inscription, K 34, of the tenth century; several very clear examples occur there, however. It seems extremely likely that it was borrowed by the Thais; it is attested on the Sukhothai inscriptions (Ishii et al. 1977). The particle has been treated here as both marker and pre-verbal particle (pre v.p.), both of which functions it performs fully in Modern Khmer. For the Mid. inscriptions under review there is only the one occurrence cited in examples where it is m. In R. it occurs frequently and seems always to be immediately pre verbal. In L. it occurs pre-verbally, but does not in fact occur very frequently. ktl. (f.) 'whether ... or'. ktl appears first in Mid. Khmer and was already well established in Sukhothai Thai. There is nothing in its form to prevent the theory that it came from Thai to Khmer. Having co-existed with the Khmer form of the same meaning, lab, during the Mid. period, ktl supplanted it completely. ti. Table I and the examples show ti 'earth; direction; focus' in use as a marker. It has often been described as an indicator of the passive voice but in the opinion of the present writer the passive voice occurs only in the translation and there is no passive voice in Khmer. Clauses following ti have a form no different from other statements translated by the active voice. It seems rather to convey 'person or thing in respect to which'. The idea that in this usage ti indicates the focus of attention immediately provokes comparison with Mod. df in its use, for example, at the beginning of a letter: mak A.fad[ raljk 'To A. whom I miss' (lit. come A. being subject of regret-absence). Lewitz (1971: pp. 115-16, n. 8.) noted the similar usage when editing IMA 4. Thai /thii/ is also used to introduce the focus of attention and was well established at the Sokhothai period. Did the Khmers in the Mid. period borrow Thai /thii/? If so, either it was a literary loan based on spelling or Thai written initial dwas not then aspirated when pronounced. (Similar parallels occur with Khmer gf and Thai khuu 'that is', Khmer dan' and Thai /than/ 'catch up', but for these pairs there are Old Khmer occurrences with the expected initial consonants (gi and dan) and one assumes the loan was from Khmer to Thai. Whatever the direction of the borrowing of dl, the Khmers acquired from the Thais the use of it to form an ordinal numeral system. In R., d[ seems rather like a new borrowing. It occurs with dis 'direction', des 'country', uttar 'north', and not alone. Although the Khmers may have borrowed dlfrom Thai /thii/, did the Thais borrow the focus usage from the earlier Khmer ti? Lewitz also suggests the possibility that Thai /thii/ was itself the borrowed form of Angkorian ti. 198

PAGE 207

A diachronic survey of some Khmer particles tae. This particle seems clearly to be a loan from Thai since the parallel Thai form tiia was well established in the Sukhothai period and tae appears for the first time on the Mid. Khmer inscriptions. It is entered as a general particle on Table 6. Its limiting effect on meaning requires many different translations according to context: 'only; just; exactly; simply (and no more)'. It seems to the present writer that tae has very much the same effect on meaning which the final phrase particle gu!J gus had during the A. and Mid. periods; in this case of replacement, however, a different word category and grammatical role are involved too. tae and gu!J occurred alongside each other in early Mid. Khmer but gus then became obsolete. However, there is a need to present a case for the comparison of gu/J with tae. Lewitz (1970: p. 104, n.10) when translating a very difficult passage, at the beginning of her editions of the Mid. inscriptions, said that, with the negative particle, it means 'not... at all, absolutely not'. For gu/J in affirmative contexts, in IMA 3, 4 and 6, she gave no translation. There are A. period occurrences of gus -gu/J where the translation 'only, exactly, just' suits very well. Two are cited in the examples. And later, Lewitz herself translated gu!J as 'seul' and 'seulement', when editing the Mango Grove inscription. She explains (Pou 1978: p. 354, n.3) that she is taking 'l'adverbe gul;i "exclusivement"' with the preceding verb. (This reference back to the preceding verb is a function characteristic of all particles catalysed as final phrase particles.) bro!J 'because'. This word is mentioned here because there are reasons for supposing that it also may have been borrowed from Thai. (In recent times it is also frequently combined with the verb 'to say', as it is in Thai.) Although Thai phrq /phro?/'due to, because or is attested on the Sukhothai inscriptions (Ishii 1977), it does not occur in Khmer until the modern period. (It does not, for example, occur on any of the Inscriptions modernes d'Angkor nor in the old Cplip'.) Nacaskul (1962: 185-7) held that the direction of borrowing for the comparable words Thai ko/ko?/ and Cambodian ko!J /bh/ 'island' must have been from Cambodian to Thai because otherwise the Cambodian form would have had the final glottal stop -k. However, it seems possible to the writer that the loan may have been of a more literary and less colloquial nature. It may have been the spelled rather than the spoken form which was borrowed; both forms have the same written vowel and the final consonant which goes back to the same Indian symbol, the visarga. Jenner and Pou (1980-81) construe brob as < ro!J 'form, shape, manner', thus suggesting a Mon-Khmer origin. Did the Thais originally acquire bro!J from Khmer? The use of particles in R. (The Reamker Parts 1-5) The R. occurrences have not been entered on the tables except when they are unusual or constitute the only mid-Khmer entry but they are given in the examples. It was possible to find every particle and usage for which 199

PAGE 208

JUDITH M. JACOB there was a Mid. inscriptions occurrence except: gi pi. pi occurs in R., as it does on the Mid. inscriptions, expressing purpose. Perhaps it is by chance only that gi pi was not also found to occur in the stanzas examined. gi does occur with {a several times, eg. 1.7.5-6. nu nuv. nuv does not occur as a marker initiating discourse. Owing to the nature of the text, speech is initiated in many other ways: by exclamations of surprise or woe, for example, and frequently by the use of the title or name of the person addressed. bl. No occurrence of b[ meaning 'at (past time)' was found, but its absence may be due to chance. bek. This is a non-occurrence which is not surprising because many other ways of expressing superlatives are used in R. and the poetry which followed it, e.g. kanlan 'surpassing', krai 'very much', foes 'over and above'. bek does, however, occur in L. and C. man occurs only with gi, and then not as an initiator of discourse. man was gradually being less used; one might expect that it would be used in combination. man gi in R. may be understood as 'in that it is'. rob was soon to disappear from the language. Perhaps, too, it was rather formal for poetry? tarn and {oy 'in accordance with' were soon to replace it. lab-lob 'whether ... or'. The newly borrowed kd(see p. 20 above) occurs several times in R. but not lab. vin occurs as a verb 'to return' but not as f. 'back again'. There are, on the other hand, occurrences in Parts 1-5 of the Reamker of particles or usages which are not be found in the Inscriptions modernes d'Angkor l-l 6c (Lewitz 197072): 7 kuv, ,:ia f. (Emphasis). The first belongs to poetry, the second, in Mod., to the colloquial. nil'v, nai. These forms fulfil a function, as a link between verb and object, which is new to them and in which they begin to replace ta. (See Table 3.) nuv continues to behave in this way in Mod. literary Khmer. nai. f. 'of'. This is the use with back reference to the sentence subject, discussed under 'Changes of word order category' (p. 19 above). nau. f. 'still, continuously'. R., like L. and C. (see below), provides occurrences in Middle Khmer, a link between the A. period and Mod. nau /oey.8 7. Further particles which are not on the Mid. inscriptions l-16c, or in Mod. Khmer, e.g. lgjk in lgjk poe, lelJ. (related to Mod. mlelJ.. See Pou 1977a: 120-4) occur in R. R. occurrences are not varied or numerous enough to make it possible to establish their word categories. 8. Did this use of the verb 'to stay, live, remain' as a final particle meaning 'still' go to Thai from Khmer? cf. Thai use of yuu. 200

PAGE 209

A diachronic survey of some Khmer particles phail. pre n.p. 'together with'. An unusual usage but one which is attested elsewhere.9 ra, vvaf?l fael. It seems surprising that they are in R., and not in the Mid. texts but this may just be by chance. In addition to the above R. usages, the particles ka and fa have further uses in R. which are not in evidence in the other texts under investigation (i.e. up to A.D. 1630) and which the writer does not remember meeting in later literature. Examples of these uses are given separately here and they are not entered in the tables since it is felt that they are on the border-line between grammar and poetic style or licence. ka. the normal Mid. use of ka to introduce a slight consequence 'and so' occurs passim. The following two citations illustrate the extra use: 1.14.1-5. yoeil git smlin jli randab ... neb [luv dut du! an thli foem brab rlimadhiraj ka loek rddhi dhanu 'We thought it was claps of thunder ... (but) now the envoy tells me about Prince Ram, how he raised the mighty bow'. the cause: Prince Ram, who 2.2.1 sury saeil clif?lil caeil e ambar fuc brab candr car tracab ka ben -pun:zamz The light of the Sun shining in the heavens was like that of the bright Moon, when it/which is full'. ka seems to be a link, like a relative pronoun, between the clauses. No consequence is expressed. fa is very much in evidence in R. Apart from its use as relative pronoun, as general dependent particle linking an attribute to a noun (the attribute including possessor or a noun in apposition) and as link between verb and object-all of which are illustrated in the examples and entered on the tables-fa also occurs between two nouns or two verbs of similar meaning, e.g. 2.23.1. 1.1.1. kal' fa mliyli 'tricks, wiles' prasoer fa uttam 'admirable, superior'. Evidence from Lpoek Ailgar Vat The following usages which were found in R. and not on the Mid. inscriptions occur in L.: the use of nai as a final phrase particle and as a pre-nominal particle linking verb to object; the use of ,:za as a final phrase particle adding emphasis and the use of nau as a final phrase particle. (See examples.) L. also corroborates the two further uses of ka and fa just illustrated above from R.: 9. I am grateful to Miss A. Compain, a missionary relief worker, for the information that speakers from Battambang use phan in this way. She also drew my attention to similar usages in the translation of the Bible, where phan is used pre-nominally in conjunction with rapas'. 201

PAGE 210

JUDITH M. JACOB ka as a kind of relative pronoun. 282.6. chlak' rup maha yaksa vlrad muhima ka chak' seta naf!l rat'. '(They) had sculpted a relief of the mighty yak:ja Biradh who/as he was seizing hold of Seta to run off with her'. between two nouns. 288.6 asur ta dahan 'demon soldiers'. between two verbs. 289.14 graviis fa gravl 'swinging-up-their-arms and brandishing (them)'. It is noticeable, however, in reading L. that both fa and pi 'as it were' are used much less frequently than in R. Evidence from the old Cpap' The composers of the Cpap' did not use a great variety of syntactical constructions. They tended to repeat 'do's' and 'don'ts' and to prognosticate that 'if certain actions were taken, a good or bad result would ensue. They used a familiar style of discourse which gives a modern colloquial impression. The texts do, in fact, furnish some very interesting occurrences of particles and in one case a non-occurrence of an expected one: fa is occasionally used to attach an attribute to a preceding noun. Otherwise it has one, perhaps fossilised, use with lok or paralok, e.g. Hai Mahajan st. 2. oy pan kti" gap'fa lok paralok 'and so gain advantage in this world or the next.' toe. m. (Introduces a question). This word, familiar in Mod., was not found elsewhere in the early texts. It occurs in Kun Cau st. 5. die toe jjunman 'It stings-say, how much?' min. C. and L. provide what seems to be the earliest evidence for this colloquial negative particle, much used in Mod. pl does not occur alone, as it does so frequently in R.; kuf!l p['do not' is very common. Certain combinations with pl, not found elsewhere in the early texts, occur in C.: sump[, translated as 'lorsque' by Pou but now meaning 'even', and gappl 'it behoves ... to'. Both dof:z pl and do{i poe occur in Hai Mahajan (st. 49 and 80 respectively) which strengthens the theory mentioned in Section I above that they are forms of the same variant word. hetu. m. 'because'. The only earlier occurrences which have been noted of hetu as a clear marker of cause were those in A., where hetu is combined with man. In C. hetu is also used in combination with tpid, and, as in Mod., with tae. ka most interestingly does not occur at all in the old Cpap ', but it does occur, for example, in the later Cpap' Prus, st. 65. Summary It will have been observed that, not unexpectedly, some forms occur on more than one table, eg, hey hoey, which functions as both m. and f. (Tables I and 2). However, there are also some forms, such as pi and man 202

PAGE 211

A diachronic survey of some Khmer particles (Table 1) and ta (Table 3) which occur in more than one section of a particular table; different meanings, semantic or syntactic, are indicated for the different sections in the left-hand column. It might be argued that two sections should be combined if they contain the same Khmer word, performing the same function, and the English 'translation' is the only difference between them. However, justification for the separation is claimed as follows: pi may introduce a clause of Cause or Purpose, which should surely be regarded as syntactically separate; man occurs in one section with nu and in another with tel, though nu and tel do not overlap. ta, when following a verb of giving, buying or speaking, links it to the indirect object, but in other contexts provides a link with the direct object. The apparent confusion in P.A. and A. concerning cause and purpose (both introduced by pi) leads one to the conclusion that the meanings which Coedes worked out by reference to the whole inscription and to information from other inscriptions, etc., despite the lack of clarity in the text, were easily understood at the time because people were familiar with the whole context. We therefore return to what was said in the opening paragraph of this article, that Khmer syntax may be inexplicit unless there is a need for precision. With regard to the borrowing of particles from Thai, Huffman concluded (1973: 491-502) when comparing modern Thai and Cambo dian syntax, that Cambodian syntax was strongly influenced by Thai. It is significant too that, at the same early Mid. period considered here, a variety of Thai vocabulary was being borrowed by the writer or writers of the Reamker as a whole. R. and C., and even L., offered likely material, heralding Mod. in one way or another. It was interesting to find that R., though probably pre dating Mid., had no rob or lab; that in L., while fa and p[ occur, particularly in the passage describing the Angkor bas-reliefs, linking attribute to noun, their use is much less than in R.; that in L. and C., but not in R., bek and min are attested; and that ka does not occur in C. It is possible that some help towards dating the Reamker and early Cplip' may come from an examination of the grammatical forms but great circumspection is needed. 203

PAGE 212

JUDITH M. JACOB SECTION II: Tables ( R., L. and C. occurrences are entered in the tables only if there is no Mid. occurrence from the inscriptions.) TABLE I: Markers (m.) Indication P.A. A. Mid. Mod. Time 'when (past)' kala R. kal kal na 'when (by the time)' luQ luQ (lit.) 'until' luQ na [ta!' (kal i:ia)] Condition 'if daha daha doQ poe dOQ pi peh (=pe) 'whether' doQ ( + kti f.) doh do~ p"i doQja Cause 'the reason ... was because' pi ... gi pi... [pan ja ... b"i broQ] 'because' hetu man (ta) pad !Pit C. hetu, [broQ] hetu tae [toy] hetu tae Purpose 'so as to' pi 'so that...(not)' pi pi kaf!lpi (so kuf!l ... oy that...not) gi pi gi pi tempi toempi (niil) tempi nu Result 'and so, and gi pi then ... dep dep dep doep 'with the ka kar ka result that' [pan ja] Relative pronoun ta ta ta ta man man man mangi tel tel R. tel tael na 204

PAGE 213

A diachronic survey of some Khmer particles Indication P.A. A. Mid. Mod. Focus of attention 'with respect to ti ti dI which' Reported speech 'that...' man man tha Initiation of discourse 'Now .. .' nu nu nu tpit (lit. and ginu man gi old-fashioned) man R. tpit Exhortation and Optation 'let...(not) .. .' karppi karppi kurp oy 'may ... (not) .. .' 'do ... not .. .' R., C. kurp kurp 'may .. .' Jen leli oy Co-ordination 'and (then)' hey hoey TABLE 2: Final phrase particles(!) Translation P.A. A. Mid. Mod. 'already; by now, lhey hey hoey by then' 'also' ukk sot sot sot sot 'and ... too' phoil phail 'back, again, re' viii viii viii (Emphasis) ra ra R. ra ra holi hoil hoil R.,C. kTiv R.,L.,C. !)a !)a 'only' gus gu}:! (Possession) R.,L.,C. nai 'still, continuing' anau R.,L.,C. nau nau ( + joey) 'utterly; (not) ley ley ley joey at all' 'very much; too much' beg bek 'whether ... or .. .' lah lah lo}:! ktI ktI 205

PAGE 214

JUDITH M. JACOB TABLE 3 Pre-nominal particles (pre-n.p.) Translation P.A. A. Mid. Mod. 'according to' ro):i ro):i ro):i [tarn, toy] 'all' iss is as' 'As to .. .' e e ri riy rI rI (lit.) 'at, in (place or ai, ay ai, ay, ta i, e. ai ta, ta e (place) time), with na na na regard to' anau, nov no nau 'by (the time lo):i, lva):i lva):i ta gi lu):i [tal'] that)' 'from' arpvi arpvi arpbI arpbI, bI C. bI 'in' karpluil karpluil knuil knuil 'like, as, similar to' ru ruva rU (Link between ta R. ta verb and object) R.,L. nai, nuv nuv (lit.) 'near' iiyail iiyail LJit] 'or nai nai nai nai (lit.) [rapas'] 'this, these/that, ne):i/no):i those' 'to (a place)' lo):i, lva):i lva):i [dau] 'to, for, from (re ta ta ta giving, speaking ai ta, ay, ta [dau] and buying)' 'towards (N.S.E.W.)' ti ti [khan] 'with, and' nu nu nu nuv (lit.) niil niil R.phail 'with, by means or nu nu nu nuv (lit.) njil 'with reference to' nI ni 206

PAGE 215

Indication 'all' Interrog.: 'which?' 'like, as' Location, Reference 'this, these/that, those' Indication Future time, possibility Negation 'never' (Joins plural subject to verb) 'all' Result Indication 'all, complete with, as well' (Link between noun and attribute) 'only' A diachronic survey of some Khmer particles TABLE 4: Post-nominal particles ( post n.p.) P.A. A. Mid. phoii phoii phoii na na ru ruv m ni nel:J/noI:i neI:i/noI:i TABLE 5: Pre-verbal particles (pre-v.p.) P.A. A. Mid. nu nu nu VOfTl VVafTl bvum L.,C. min vvafTl tel R.,L.,C. bufTl tael syaii ta syiii, syiii ta ka kar TABLE 6: General particles ( g.) P.A. A. Mid. daii daf!1Ii ta ta ta tae 207 Mod. phaii Mod. niii bufTl mm bufTl tael (sin= almost) ka Mod. dafTIIi ta (lit.) tae

PAGE 216

JUDITH M. JACOB SECTION III: Citations from texts The particles entered in the tables are listed here in syllabary order under the heading of the oldest spelling, with citations illustrating the uses. P.A. and A. citations are given with their Khmer (K.) reference number, as given in Coedes, face and line. Mid. citations are given with their Lewitz (197072) number, face and line. R. citations have a reference to part, page, and line in the Reamker text of the Institut Bouddhique; L. to page and line; C. to the Cplip' title and the stanza. L. and C. citations are given only if no Mid. inscriptions occurrence has been found. Where Mod. particles are related to the older forms, an example is given for comparison. ka, kar, ka [P.A. occurrences of the form ka show no similarity to the particle, so may be construed as > ka 'to construct'; some may be a shorthand form of kiiurri 'serf'.] 'and so; accordingly': Khmer m. or pre-v.p. A. 349 sud 15. ka gi nob phdai krorri nob ta jli daksina hon 'and so it was these very territories presented as gifts' Mid. 8.30 kar A, B, C, etc.10 (40) jli saks[ 'And so A, B, C, etc. were witnesse~ R.1.9.11-12 stec moel as' bhudhar loek ,:ddh[ dhanfl ka broes bral:z bhaktr prim priy. 'As he watched all the kings (trying to) raise the mighty bow, an eager look came into his handsome face'. Mod. nlin njn khjn knliii' njn khiiurri hoey khiiurri ka khuc khlit mittabhlib rapas' yoen. 'You will be angry with me and I shall lose our friendship'. Mod. poe dau ka dau. 'If we're going, let's go.' klila, klil 'when (past time)' :Khmer m. A.669.B.3. klila samrlic homa ... 'When he had completed the sacrifice, .. .'. R. 2.67.12. klil brab rlem stec mak phgan as' maha ,:s[ phan sjn ar daduol bra!J ang. 'When your brother came here and greeted all the great sages and paid them homage, they were delighted to receive him'. Mod. e nandaka: ho,:i, klil sarrirec ktl prlithnli khluon hoey, ka krlip lli ... 'As for Nandaka, when he had achieved his own wish, he took his leave respectfully .. .'. kflv (emphasis): Khmer f. R.2.65.8. burri !fli foem brab bhirut jli cneb kflv ,:ili. 'did not know Bhirut was in this mood!' C. Kun cau. 62. tae anak muoy kuv ra-ae ra-fiv 'Now another person, constantly complaining'. 10. In the citations and their translations, A,B,C, etc. are space-savers used to replace the individual names of persons mentioned. 208

PAGE 217

A diachronic survey of some Khmer particles ku,ri (exhortation): Khmer m. R.2.50.2. ang an sot kum prahaes. 'And as for me, let me not be negligent'. C. Hai Mahajan. 58. khjn prakaek jhlob. 'Don't be angry, arguing and quarrelling'. Mod. kum dhvoe. 'Don't do it'. ka,ripi, ku,ri pi, ku,rimpi (i) (Exhortation): Khmer m. P.A.724. kampi tve agara le guha. 'Do not build houses on the grotto'. Mid. 6.B.8. kampi oy mok pyet pyen kambujadesa. 'Let them not come to oppress Cambodia'. R. 1.39.6 kum pi pa git can dos. 'Do not think of initiating a wrong'. (ii) (Purpose): Khmer m. Mid. 3.A.31. lek yas brab rajasantan ... kurrzmpi man dharmmantarliy ley 'to increase the glory of the royal line ... so that there should not be any calamity (for them)' R. 1.25.6 kum p[ Ian' 'so as not to be a long time'. ka,rilun 'in': Khmer pre-n. p. P.A. 560.11. sre kamlun tnal 'ricefield within (the boundary of) the road'. A. 291.22 sre kamlun vrah mand[ra 'ricefield within the royal palace (grounds)'. kd 'even (if)', 'whether ... or': Khmer f. Mid. 5.2. dob kfln li.n kd kmfly an kd 'whether it be my child or a nephewor-niece or mine'. R. 2.75.7-8 ac dap' dal' khyal' da,rin 8 dis pok pak' kd. 'can even withstand the winds from all 8 directions'. R. 2. 73.6-7. dob brab bhirut ras' lokl( y), lub paralok kd'whether Bhirut is alive in this world or has gone to the next worid-:-:-7 Mod. dob prus kd sd kd 'whether male or female'. knun 'in': Khmer pre-n. p. Mid. 8.7. knun pa,rimros 'in freedom' (lit. in (state of) affranchisement). R. 2.61.9 knun brai 'in the forest'. Mod. knun sastra rfoen ram kerti 'in the manuscript of the Reamker'. gi pi 'and that was why'; 'being an occasion for', 'and so': Khmer m. A. 222.16 man khnu,ri nob rat dau gj_j_ Tan oy Tai B ... 'That servant ran away; that was why Tan A. gave Tai B .. .'. A. 450.14-15. pre guruyaga gj_pj_ vrab oy daksi!Ja 'ordered (someone) to perform-the-sacrifice-to-the-guru so that (i.e. on that occasion,) His Majesty might give offerings'. Mid. 6.A.11-12. rantap kriya puja sa,rirap gj__pj_ niman brab siigh phon 'prepare offerings in readiness for inviting all the monks'. gus, gub 'just, only; even (if)': Khmer f. A. 56.C.3 l. rata 1 ca,rinay f5!:!!. gi ta bhflmi A. 'at exactly the distance of 100 (measures of distance) is the territory A'. 209

PAGE 218

JUDITH M. JACOB A. 211.6 /en oy carrmam kalpa!Ja neh /en rohh kalpana vrah kamraten aii... 'let (them) provide only these provisions, letting them be in accordance with the provisions for the god .. .' Mid. 3 A 78-9. doh pi anak phon vanven do Ian/yin ai ta caturapaybhum ~-'Even if all these people err and go and fall into the four domains-of-suffering ... R.3.31.1. khnuf!l cul paJ?lroe prayoj(n) yak kusal pavar mangal 'I went to them simply to serve them for the sake of attaining the highest merit'. iiyan 'near': Khmer pre-n.p. P.A. 155.18. sre ai iiyan travan poii 'ricefield near Pofi's reservoir'. A. 239.33. sre x iiyan vrah kamraten aii 'x (number of) ricefields near the sanctuary'. ta. See ta. ta dai. See dai. f;empl. See tempi. fael. See tel. !Ja. See na. ta, fa (i) 'who, which (rel. pronoun)': Khmer m. P.A. 561.20. ge ta sak gui 'anyone who spoils (things) here'. A. 56.B.27. anakta taf!l aJ?lvau 'persons who plant sugar-cane'. Mid. 2.5. kamraten ta ja bvuJ?lnak 'the lord who is our refuge'. R. 2.49 .1. brah ram rzem ratn ta ja raj putr cpan 'the beloved older Prince Ram who is His Majesty's oldest son'. (ii) 'the one who, the one which, the': Khmer g. (linking an attribute to a noun) P.A.30.C.9. kiiuf!l ta si 'serfs, the males'. A.19 .21. dravya phon ta damnepra all the following goods'. Mid. 6.B.3. is kal ta lungh hon 'for a long time indeed' (lit. whole-of time particle, long indeed). R.1.1.3 stambh dvar fa bicitr 'decorated pillars and door'. Mod. (lit.) aramm(l))ta ascary neh 'this remarkable idea'. (iii) 'to, for, from, (re giving, buying, speaking)'; 'at, in (re time, place)'; (a link between verb and object" Khmer pre-n. p. P.A. 90.10. ge jon pie jhe ta vrah kamratan aii 'persons offering fruit to the god'. P.A. 689 A.11 ku srac ta punya 'Ms. Achieved good works'. A.259 sud 8. dun bhumita Vap A. 'buy a territory from Yap A'. A. 249 .1 ta rajya vrah pada kaJ?lmraten aii 'In the reign of King .. .' 11. Jenner (1981) demonstrates, by illustrating the Old Khmer uses of ta, that this particlethough found in a variety of contexts which suggest in translation a wide range of grammatical functions in European languages-functions consistently in Khmer as a linking particle. While agreeing that the role of ta is always to be a mere link between other words, I separate the examples here into three groups because for me ta is classified according to whether it links a clause or an attribute or forms an untranslatable connection between a verb and its direct or indirect object. 210

PAGE 219

A diachronic survey of some Khmer particles Mid. 2.4. du/ ta bral:z 'humbly-inform the Lord'. Mid. 6.A.36 sarrmak ta vral:z siisthii 'resorting to the Lord's teaching'. Mid. 3.B.2 cul ta pad moks 'go on to the way of deliverance'. Mid 8.24. kiit ta ktl nel:z jii 'decide the case as being'. R. 1.59.5. pramiid !E_ bral:z ang attack you, my lord'. ta pad. See pad. ta man 'who, whom, which' (rel. pronoun): Khmer m. P.A. 562 26. kFiurrz ta man klofi A. oy 'serfs which Klofi A. gave'. ti. (i) 'earth; place; direction': Khmer pre-n. p. P.A. 688.4 !J. tnai luc sruk 'to the West of the village'. A. 158.28 !J. purvva ... daksil)a ... paccima ... uttara 'to the East... South ... West... North'. (ii) 'focus of action; in respect to which': Khmer m. P.A. 927.2. sre ta tel ti kuriik ... tve 'ricefield which Kurak ... worked'. A. 246-8. 3-4. dravya-::_ ti afi yok 'goods ... which I take'. tempi, tempi nu, foempl njn 'so as to': Khmer m. Mid. 6.A.15 (prepare a sacrifice) tempi panlub do ta brab Fiat 'to send (merit) to kin'. Mid. 3 A.28. tempi nu siin siisnii bral:z tathiigat 'so as to build up the teaching of the Tathagat'. R. 1.5.3-4 prae basudhii toempl njn dhvoe oy ksem fa prajiiriistr 'turn the soil in order to bring peace to the people'. Mod. kharrz rlen toempl (njn) piin carrzneh vijjii /-a 'study hard in order to achieve a good education'. tel, fael (i) 'who, whom, which' (rel. pronoun): Khmer m. P.A. 561.33 ge tel pofi A. pre 'personnel whom Pon A. orders'. A. 958. Nord. 29. ri bhumi tel sruk ta jmal:z karrznun sruk 'as for the territory which (is in) the locality known as Kaqmuri Sruk .. .' R. 4.46.5. tree tran' pan' mrjk tae/ tin sarrzfau dau paficaling. 'He went his way, casting aside the animals, conscious that (lit. who be-conscious that) he was heading straight towards Paficaling'. Mod. cor !E!!l. rat' dau truv ge clip'. 'The thief who ran off was caught'. (ii) '(not) ever': Khmer pre-v.p. A. 235.D.27 kule nel:z phon vvarrz tel cek mu/a. 'the relatives never shared out (the functions of) the original branch'. R. 4.11.3. bral:z burrz tae/ miin hman 'the lord who has never had any imperfection'. L. 269.10. burrz !E!!l. iik' 'never failing (to do so)'. C. Hai Mahajan 40. burrz tael khus 'is never wrong'. 211

PAGE 220

JUDITH M. JACOB Mod. gat' buf?1 tael dau angar. 'He has never been to Angkor'. taeh, tae 'only, but, just, exactly and no more': Khmer g. Mid. 9.46. oy paeii taeh anak ta reh paf?1mros neb 'strike only those who have taken these freed people'. R. 1.29.4. stec yak tae brab hasth chven bhudhar loek silp dhanu. 'With only his left hand, the supporter of the earth raised the magic bow'. Mod. nau sal' tae p[ de. 'There are only three left'. tpit. See pad. dan. See don. daha, dob, dob pi, dob p[ 'if: Khmer m. P.A. 1004.7. daha mlin kliryya vrab kamratlin an 'if there was work to be done for the lord'. A. 843.C. l 3-4. daha ayat kule aii 'if there-is-no descendent of mine'. Mid. 4.C.5. doh anak af?1mplil neb vanven dau ai ta catarlipliyabhum gub oy anak af?1mplil neb oy rruc lae[n] 'if all these persons lose their way in the four domains of suffering, even so let them escape'. R. 3.12.9. doh brab pad stec fa!' oy brab dasabal ... bidagdh asflr 'if the prince comes here, let that lord of the ten powers ... destroy the demon'. Mod. doh /-a kd akrak' kd 'Whether (it is) good or bad .. .' Mid. 6B.4. doh pi mlin rlijasatrfl ... 'If there should be (any) enemies of the king .. .' R. 1.61.6 doh p[ pli en mlin dukkh, an buf?1 bram sranuk 'If you had troubles, my friend, I could not bring myself to be content'. Mod. doh p[anak gman klir av[, anjoen mak !en phdab khnuf?1. 'Even if you do not have any business, do come and visit me at home'. d[dai. See dai. dep ( depv, doep) 'and so; and then': Khmer m. P.A. 726.A.8. fkp_ ge crip ti 'Then (i.e. after the purchase of the ricefield) they reserved the land'. A. 207.7. man khmi sthapanli lisanli kamraten jagat, fkp_ thve caf?1na1?'l kalpanli 'He wished to establish the seat of the god and so he made a foundation, furnishing .. .'. Mid. 17.35-8. gappi yen khiiaf?1m paf?1pils anak A ... debv yen khiiaf?1m oy rantap phlaeh jheh ... 'It seemed a good idea to us to place Anak A. in a religious community ... and so we prepared fruit... (i.e. for the ceremony)'. R. 4.66.2-4. yal' jlik' jliti ja brab lohit ceii mak nob doep khiiuf?1 prlim proe yak as' parbat mak ... 'We saw clearly what it was like: it was your blood flowing out. And so I gave the order to bring rocks .. .' Mod. truv pralan jap, doep lie cul rlen pan. 'You have to pass the exam, then you may go to the school'. 212

PAGE 221

A diachronic survey of some Khmer particles don, dan, daun, diiffln 'all, complete with, as well': Khmer g.12 P. A. 18.2. oy sre dan ki'iuffl 'give ricefields complete with serfs'. Mid. 8.6. pros khnuf(lm damn gruv 'free slaves, a complete family (of them)'. R. I .17. I 0. bir ksatri ksatrii 'both princes'. Mod. kun pi niik' 'all three children'. doh, doh pi, doh p1~ See daha. diiffln See don. nii, ,:iii (i) 'at, in (place or time)': Khmer pre-n. p. P.A. 79.9. sruk niijlan kaol 'village at Jlari Kao!'. A. 235 passim. nii kamraten jagat ta riija 'before the lord of the world who is king'. Mid. 6.A.6. mok nii sthiin brah bisnulok 'come to the place Bra}:i Bisnulok (=Angkor Vat)'. R. 2.49.3. su rat' dau nau nii brai 'grimly-determined to flee and live in the forest'. (ii) 'with regard to': Khmer pre-n. p. A. 23 I.I 2. nii vrah jaf(lnvan 'with regard to the royal gifts'. (iii) 'who, when': Khmer m. A. 878.2. dak$inii mratiii'i khloi'i A. nii choln priisiida 'honorarium for Mratafi Khlofi A. on-the-occasion-of inaugurating the tower'. Mid. 6. A.33. mahiiksetr phon nii gron pranipat brah 'the gods, makiiksetr, who protect (the religion of) the Buddha'. R. 1.25.12. nii brah staen mak ta/' sthiin an neh brah staen man af(lboe paf(lnan priithnii 'As you approach my encampment here, you have many demands!'. (iv) (cf. Mod. ,:iii) 'which, any, some': Khmer post-n. p. A. 348 nord 22. nu sre nii man 'with any ricefields (which) they had'.13 Mid. 3.B.10. no kiil nii pi 'at such time as'. 4.30 saf(lner neh nii muy 'one of the novices'. R. 2.31.5. it narnai anak @ raf(lbfn 'without any one thinking of...' Mod. anak @ tjn? 'Who (which person) knows?' (v) (emphasis): Khmer f. R. 4.34.3-4. buffl diin' sabv say saffltl neb hon /oey r@_. 'before this idea should spread at all'. L. 269 .1. priisiid neb en nii 'This very palace .. .' 12. Since 1968, I have realised that dlif!1n, which may occur immediately preceding numerals (dlif!1n blr 'both') and verbs (dliflln nay 'easily') as well as nouns, should have been classified as g. 13. The A. example here might be considered, like the A. example in the previous paragraph, to illustrate nli (m.). However, nli -,:,a seems like natural Khmer and was the way Georges Coedes took it in editing 348. 213

PAGE 222

JUDITH M. JACOB C. Hai Mahajan 5. /arr,plik nas' !!Q. 'It will be very hard indeed (for you)'. Mod. praylitn !!QI 'Look out!' ni, n[ niy (i) 'with reference to': Khmer pre-n. p., m. P.A. 44.A. 7. lijiia vrab kamratan aii n[ vrab kamratlin aii rri ... 'Edict of His Majesty with reference to the God <;ri .. .' A.470.11. phtyan ni pre ... 'announced concerning the order to .. .' (ii) (Location reference): Khmer post-n. p. A.245.91. dau vnek ni aii 'before my very eyes'. Mid. 2.29. prlikat tile sa,rinilm slitaprlinidhan muhi:!]_ vin hon 'exactly in accordance with my earlier prayer'. R. 8.5. (Pou 1979, st. 3663.) b[ mub n[ 'before, in times past'. nin, njn (i) 'with': Khmer pre-n. p. Mid. 9.5. brom '1i!i uk iia A. 'were-in-agreement with Uk Na A'. Mod. khiiurr, '1i!i mitt sarr,/aii' khiiurr, 'my friend and I'. (ii) 'shall, will': Khmer pre-.v.p. Mod. khiiurr, !Y!l dau 'I shall go'. nu, nil, nilv, gi nu (i) (Initiates discourse): Khmer m. Some P.A. and A. occurrences are spelt nau, eg. 154.A.16; 957.18. P.A. 493.20-1 ... canlek yuga/a yau 3 nu man poii A. ktin sre ... ... 3 yau of double cloth. Now Poff A. releasedthe ricefield .. .' A. passim (e.g. 925, 14) after date, preceding rest of text. (gi nu also occurs similarly.) Mid.4.B.25. nil tejab pha/anisan neb camren ... 'Now the power of the fruit of this merit may give prosperity .. .' (ii) 'with, by means of: Khmer pre-n. p. P.A. 424.B.6. man duii nu sru vrab 'which (they) bought with paddy from the foundation'. A. 239.24 sre neb kula A. ti duii nu krapi 2. 'This ricefield was bought by the family/descendents of A. for 2 buffaloes'. Mid. 4.C.9. banarliy nil tribidhasarr,mplit 'resplendent with the three-fold fortune'. R. 1.2 hob hoer nilv {ej p[ banljk 'go through the air with astonishing power'. Mod. aiijoeii ... karr,sant nilv kl/a pra1Jli,rin kan' 'invite (you) to amuse yourselves with a cycling contest'. (iii) 'and': Khmer pre-n. p. P.A. 134.20 sre nu pdai karorr, nu darr,rin 'ricefields and lowlands and orchards'. 214

PAGE 223

A diachronic survey of some Khmer particles A. 215-6.10. yok sruk sre nu khiiurrz phoit 'took the villages, ricefields and all the serfs'. Mid. 6.A.5. ok hluit A. nu anak cov bhikkh B. 'Ok Hluon A. and the bhikkhu B'. R. 1.3.11. narr, raj fa pabitr nilv brab anuj laksa 'brought the august prince and Laksmax:i his young brother'. Mod. (lit.) stl va nilv gil kan darr,it punman 'addressed them and all their playmates'. (iv) 'shall, will, for the future purpose of: Khmer pre-v.p. P.A. 341.nord 11. ge gi ta nu pi niraya 'they are the-ones-who will be for Hell'. A. 246-8.3-4. dravya ta robh ti an yok nu 0ras 'the goods according to (the above list) were taken by me for a living'. Mid. 2.6-7. sarr,bau ja rat run utuit ta ja nu narrz chloit 'a boat, splendid, magnificent, to take (us) across'. R. 1.43.4-5. file nilv cheb chur chap vinas 'as if it would burn furiously and be destoyed'.-(v) (connector betwen verb and object): Khmer pre-n. p. R. 1.9 .10. sarr,taeit nilv fej brab dhanil 'demonstrate the power of his bow'. L. 289.14. khlab kan' nilv tarr,pait 'Some held batons'. Mod. (lit.) daduol nilv dukkh {al' khluon viii 'meet with suffering themselves in return'. neb (i) 'this, these': Khmer pre-n. p. P.A. 79.12 sre sin pan/ass neh sre 'a further ricefield in place of this ricefield'. (ii) 'this, these': Khmer post-n. p. A. 291.32 puja neh 'this cult'. Mid. 6A.20.jun kusal-phal puny neh 'offer the fruits of this work-of-merit'. R. 5.20.1-2. mak fa!' sthan brai neh 'come to this forest-place'. Mod. kiln neh 'this child'. nai, naiy (i) 'of: Khmer pre-n.p. P.A. 30.26. nivandha ta nai vrab karr,mra[tait] an 'provisions of the god'. A. 207 .15. thvay dravya ... neb ja nai karr,mrateit jagat 'offer these goods ... to be the possession of the lorcl of the world'. Mid. 6.A. 7. sthan jarr,nurrzm naiy, debvata 'meeting place of the gods'. R. 1.16.9. sthan nai traitrjnsa 'the dwelling of the thirty-three'. Mod. (lit.) knuit nam nai bra{i raj-raf{habhipal 'in the name of the royal government'. (ii) 'of that (with back reference)': Khmer f. R. 2.63.1-2. man bat prarr,-muoy hmTn nai,. 'He has a force of sixty 215

PAGE 224

JUDITH M. JACOB thousand (of him)./There is a force of sixty thousand of his'. L. 290.22-291.1.ja sthan bra!J maha isl en nai, 'being the dwelling-place of a great anchorite (of him)'. C. Kun cau 19. heu citt can' krep janjap klin nai,. 'because it wanted to taste and drink in the flavour of (it)'. (iii) Connector between verb and object): Khmer pre-n. p. R. 3.1.4. pl {uc 0nak {ut nai aggl 'as if someone had lit a fire (in him)'. L. 283.18. ka phtuol nai yakkh mara,:za '(He) then felled the yakkha to death'. no, nov, nau, 0nau (also spelt nuv in R.) (i) 'at': khmer pre-n.p. A. 249.14-5. vnek vra!J 0nau stuk vryan 'before (lit. eyes) the god at Lake Vryari'. Mid. 4.B.10. do no caturapayabhum 'go and dwell in the four domains of suffering'. R. 1.1.6. pansap bis nuv sela 'dilute the poison on the stones'. Mod. din phda!J nau bhnarrz ben 'buy a house in Phnom Penh'. (ii) 'continuing, still': Khmer f. A. 235.C.7. pamre ta vrah pada A. ru no!Jh 0nau 'continue serving King A. in that way'. R. 1.4.6-7. bra/J thlaen asur kakanasur ksay j[vit ka mar(lJ) {uol nau. 'The lord shot at the demon Kakanasur to take her life and so she fell dying and remained (fallen)'. R. 1.62.11-63.1. jal netr dhara sasrak' sanjap' sup nau. 'His tears flowed freely and his face was downcast still'. R. 2.6.10. foemjhuk sat' makjarrzbak' darrzn dan doer nau. 'The lotuses drifted along and, their stalks entangled, remained resting-on-each-other'. R. 2.45.1-2. bra!J pad ... nau man bra!J janm man nau loey? 'Is the King still alive?'. L. 283.19-20. prada!J sugrlb yarrz nau 'came across Sugrib still weeping'. C. Hai Mahajan 87. kal b!Ten nau 'while it is still raining'. Mod. Only the use with foey is current and this is characteristically in the negative, e.g. min dan' dhvoe nau loey 'has still not done so'. no!J (i) 'that,those': Khmer pre-n.p. P.A. 44.B.9. parrznos ta pos gi noh vnam 'religious personnel m that sanctuary. (ii) 'that, those': Khmer post-n. p. A. 216.10. thve kuti noh puja kamraten jagat 'build that cell to carry out the cult of the lord of the world'. Mid. 8.11. pamros phon !!2fl. 'that the whole (matter of the) freeing-( of slaves)'. R. 1.14.1. kal noh 'that time'. 216

PAGE 225

Mod. thnai noh 'that day'. A diachronic survey of some Khmer particles nau. See no, etc. pad, ta pad, tpit (i) 'for the reason that': Khmer m. Mid. 8.3-6. ok hlun A. mimantr bra/:t ariyasans phon jarrinurri ta pad okk hlun B ... pros khii.urrim 'Ok Hluoil. A. invited several noble monks to meet together because Ok HluoI.1 B. .. had enfranchised slaves'. R. 2.62.3-4. nef:t pad nlin kaikesl fa!Jfoem rlijy b[ braf:t rlim mak oy bhirut. 'This is because Kaikesi has taken the kingdom from Ram to give it to Bhirut'. Mod. iluv njn dau fak srflp viii. e !Jli ka plin !J!}!_ klir vli huos dau hoey. 'And now, how can one undo what is done (lit. pull out and swallow again), for the matter has gone too far'. (ii) tpit (Initiates statement): Khmer m. R. 3.6.9. ang aii. [/uvne/:t nai !J!}!_ mlin sansay amba/' njn bhariyli hoey. 'As for me at present, now there is a problem: I already have a wife to occupy my attention'. Mod. In undated texts of folktales, printed in the twentieth century, tpit introduces the 'hero's' tale of woe passim, when he meets Judge Hare. pi, p[ 'for the purpose of, so that; because, the reason why; as it were': Khmer m. P.A.21.2 tmo f!.!. vnlik 'gems for the decoration'. P.A. 451 sud 6. va cap J!.!. hau 'Mr. Catch (him) so as to employ (him)'. A. 207.2. oy sron kamraten jagat fri A. f!.!. jvan dravya ta arripa/1 neb 'arranged for the ablutions of Lord <;ri A. so as to ( on the same occasion) give all these goods'. A. 219.8. J!.!. ayat kvan cau fey syan ta tlic santlina dau phon g_j___J!.!. vrafi plida A. oy praslida bhumi 'Because there were no children or grandchildren, their line being extinct, that was why His Majesty graciously gave land ... '. A. 207 .32 yen vvarri kapata f!.!. thve apavlida tyan 'we shall not deceive in such a way as to make a denial of our knowledge'. Mid. 2.25-6. pan paii.cakhandh arribijlit f!.!. ando/ do mok ai ta saslirlibhabbh 'depart this life so as to move to and fro in the state-of-transmigration'. R. 5.63.6-7. nflv bhjurri ba/ blinarli J!.!. mak thvliy ksatr 'will assemble a force of monkeys to offer to the Prince'. R. 3.1.11. yaksli f!.!. dranan' 'A fierce yak~a'. Mod. (Only occurs in compounds. See under 'Compounded particles', p. 17 above). peh (=pe), poe 'if,when (in future)': Khmer m. Mid. 8.43. l!!d!:. mok srati 'if (such a person) should come and complain'. R. 1.57.7. I!!!!.. parapflrn yoen ylitrli vin mak ya/' pli. 'When this is completed we shall come back and see you'. 217

PAGE 226

JUDITH M. JACOB Mod. 122!!. iin kurrz kan' dos doep khiiurrz han prap'. 'If you won't hold it against me, then I'll dare to tell you'. poe. See peh. phan. See phon. phon, phan (i) 'all': Khmer post-n. p. P.A. 127 .12 arrzpall kule ge phon 'all the relatives of the personnel'. A. 699.5. rajakule phon 'all the royal family'. Mid. 3.A.25. nu braf:i nati phon 'and the whole royal family'. R. 1.8.9. diit deb ksatr phon 'all envoys, gods and kings'. Mod. anak jit khan phan 'all the neighbours'. (ii) 'both ... and': Khmer f. Mid. 10.4. kat don satpakar darrzn 7 hey eh/on phniis phon sot 'cut banners for the 7 Books (of the Abhidhamma) and also made (the people) enter the religious life'. R. diil bhlan yarrz phan 'speaking and weeping at the same time'. Mod. pos phdaf:i hoey pok ut phan 'sweeps the house and does the washing and ironing'. (iii) 'together with': Khmer pre-n. p. R. 1.53.4. anak ylitrli phan braf:i jesfhli 'you journey together with your elder'. bI See arrzvi. bek, beg 'too much, very much': Khmer f.14 Mid. 2.15. tryak ar anumodana nu stac braf:i rajaputr hon 'I most delightedly participated in what my son did'. Mod. dhnan' bek 'too heavy'. burrz, bvurrz. See vorrz burrz fael. See tel. man (i) 'who, whom, which': Khmer m. P.A. 561 A.27. knurrz man pon A. oy 'servants whom Pofi A. is giving'. A. 207.29. nli bhiimi nef:ih man vraf:i kamraten aii A. dun ta yeiz 'with regard to this territory which my Lord A. bought from us'. (ii) 'that' (reported speech): Khmer m. A. 843.C.13 kule an tyan man bhiimi nil upliya no(1h phon viii ayatta ta santlina an. 'My family know that all those territories and their subsistence are returning to the authority of my line'. Mid. (?) 2. 7-8. sramaddh bol man gi bhabbh jat 'the ocean: say that it is (taking gj_ as a verb but see man gi below) existence'. 14. This is also a change of mind about the classification of a particle as compared with Jacob (1968) when bek was assigned to the adverbial particle class. 218

PAGE 227

A diachronic survey of some Khmer particles (iii) (Initiates discourse): Khmer m. A. 207.15 man srlic thvliy dravya neb ... dep dau 'Now, having finished the consecration of those goods ... he then went...' man gi (i) 'who, whom, which': Khmer m. Mid. 2.10. jlit zluneb man gi lub ta 1499 raka 'this present life, which is in 1499 qaka'. R. 5.35.5. neb yal' man gi kal' hon 'this (which) one may indeed see as being a trick' (ii) (Initiates discourse): Khmer m. Mid. 3.A.16. man gi arrzbi klil jli pratharrzm samtec ... plin thlen svey rlijasampatt ... 'Now from the beginning of His Majesty's ascending the throne ... min 'not': Khmer pre-v.p. L. 273.3 min oy yilr fun 'not being a very long time'. C. Hair Mahajan 62. min thlas' dhloy 'not making careless mistakes'. Mod. glit' min dau de 'He is not going'. ra, rli (Adds emphasis): Khmer f. P.A. 726.C.8-9 kiiurrz vnlik pon yajamlina dai, ~tel oy ta liframa, klah ra, tel oy ta vrab 'other servants of the section of Pon, some of whom he gives to the aframa, some to the god'. A. 205.8. pek sapp kule ta gi bhliga didai ra 'divided (it) among all the relatives there, each having his share'. R. 2.41.1. ob o brabjanm skal' skap' rli plin yal' hna pw:zy putr cpan. 'Alas! Your life was completed, then, when you had the chance to see the merit of your eldest son'. Mod. mak naeb rli. 'Come here!'. riy, rl 'as for': Khmer pre-n. p. P.A. 518.D. l. ri anak ta sok 'as for anyone who corrupts .. .'. A. 33.35. dJ!. anak ta parrzplit dharmma ... dJ!. ta pariplila ... 'as for persons who destroy the foundation ... ; as for those who look after (it) .. .' Mid. 4.23-4. ,:Jjz dlin pz neb 'as for these 3 young men'. R. 5.3.4-6. klil nob hetu pliramztli d brab suriyli burrz lie astangat dau. 'Then, owing to his perfection, the Sun (lit. as for the Sun) was unable to set'. Mod. rz cau adhiklir viii ... 'Now as for the abbot...' ru, rilva, ril (i) 'like': Khmer pre-n. p. P.A. 341.sud 5. pre thve piljli vrab karrzmratlin an ru lingapurvvli 'ordered that the cult of the god should be carried out as (at) Lingapura'. 219

PAGE 228

JUDITH M. JACOB A. 235.71 riiva nohh anau 'continuing like that'. Mid. 2.6. fii bvurrz~ak trey trlin pradhlin ru sarrzbau 'is a refuge for (our) protection like a boat'. R. 5.31.3. sabd sraek rii randab 'the sound of roars like thunder'. (ii) 'like': Khmer post-n. p. P.A. (Occurs frequently at the end of names.) 689.A.13. ku al ru 'Ms. In a-rush-like'. A. 231.35 neb ruv 'like this'. rob, robh 'in accordance with': Khmer pre-n. p. P.A. 904.B.25. gi neb ta rohh nebh phon tel .. .'It is all these (things), in accordance with this (list just given), which .. .' A. 216.10. roh man steii vrab Jrai ta aji dai kalpm;li 'in accordance with what the other Sten of the Sacred Fig-tree, ancestors ( of this Sten) provided'. Mid. 3.B.20. laen sf hit sfhir is klil ta langh roh brab carrznan prlithnli. 'May it remain firmly established for ever in accordance with the king's wishes'. !ah, lob 'whether. .. or': Khmer f. P.A. 51.14. kon prasli !ah cau prasli !ah 'whether it be son-/daughter-in law or grandson-/daughter-in-law .. .' A. 842.24 caturthanra !ah !fa!jthanra !ah 'whether it be a quarter or a sixth'. Mid. 3.A.37. dob purus !oh dob rrT filJ 'whether it be male or (further) female .. .' lub, lob, lvab, lub nli (i) 'to': Khmer pre-n. p. P.A. passim arrzvi ... !2f2. 'from (a place) ... to (a place)'. A. 190.5-6. lvah chok khadira 'as far as Chok Khadira'. A. 457.8 b!!!fl:. ta gi camkli 'as far as the garden'. (ii) 'by the time that, by (a time)': Khmer pre-n. p. P.A. 259.III 21. !oh ta gi rlijya vrab kamratlin aii 'by the time of/in the reign of Her Majesty'. A. 855.5. lvah ta gi 888 raka 'by 888 c;aka'. Mid. 2.9-10. luhjlit f/uv 'until this present life' .. R.1.5.2. !uh bislikh 'in April-May'. (iii) 'until, by the time': Khmer m. Mid. 6.29. !uh pan cul ta pad moks 'until (they) manage to enter the way of deliverance'. R. 2.5.1. bi/lip !E!J:. lat' ralliy rarrzsliy sok 'wept until sorrow melted away'. Mid. 3.A.43. !uh nli pan pus 'until he would be able to enter the religious life'. R. 2.81.6-7. !uh sanlap' prab prlilJ 'until they fainted, writhing'. 220

PAGE 229

A diachronic survey of some Khmer particles Mod. fuf1 tra/ap' dau than devata viii ... 'When they arrived back in the dwelling of the gods .. .' /en, laen (Exhortation): Khmer m. A. 219.5. lenja bhumyupaya nai sruk stuk ransi 'let the subsistence-of-the territorybelong to (be of) Bamboo Lake'. Mid. 6.A.36 laen pan pus sarrinak ta bral:z sastha 'May he manage to become a monk, staying in the religion of the Buddha'. R. 4.8.8. laen pan khmoc aii phi'ioe bhudhar bTr bral:z dhanfl 'let it be possible for my body to be there for the two supporters of the earth'. fey, /oey 'totally, utterly, at all': Khmer f. P.A. 557.Est 4. va ahvan arrive kI_ 'Mr. Avoid all action'(?). A. 989 .B.10-11. daha 0yat santana ta purw;a kJ!. 'if there is no male descendent at all'. Mid. 4.B.11-12. kurrippi pros anak nol:z pan kJ!. 'let (them) not be able to save those persons at all'. R. 2. 7 .2. paksT khluon khlau burri tin toem dan jhuk /oey 'the foolish bird had no idea that they were lotus stems'. Mod. citt anak karrilol:z ... min nay njn rasay pan loey 'the boy's feelings ... were not at all easily appeased'. lol:z. See lul:z lval:z. See lul:z. /hey. See hey. vin, viii 'again': Khmer f. A. 697.B.18. cat sruk A. vin 'restore the village A.'. Mid. 2.13. san ru puran vin 'reconstruct as of old'. Mod. tra/ap' dau phdal:z viii 'return home'. vorri, vvarri, bvurri, burri 'not': Khmer pre-v.p. P.A. 154. A.16. ge ta oy gi 'persons who do not give (things) here'. A. 85.2. nau 0nak ta~ thve toy kalpana 'now persons who do not act in accordance with the arrangements ... Mid. 2.17. bvum d[en 'not certain'. R. 1.8.10. bum ac njn lock dhanfl pan 'were unable to raise the bow'. Mod. sTevbhau nel:z burri maen ja pravatti panlarri rapas' jan rilp ,:za /oey. 'This book is not in fact the story in disguise of any (real) person'. vvarri tael. See tel. sot 'in addition': Khmer f. P.A. 561.38. oy antyanti sot 'finally gives in addition'. A. 67.C.6. dar prasada .. ~ar prasada sot 'acquired the royal favour. .. acquired the further royal favour'. Mid. 16a.8. ri e anak tajhar lek khmoc nol:z sot man A., B., etc. 'As for those who took charge of the funeral, one may add, there were A., B., etc.' 221

PAGE 230

JUDITH M. JACOB R. 2.2.7. camjfii sot man phka 'and the Michelia too has a flower'. Mod. rloen bren rT pad camrlen sot ... 'legends or songs, too ... syan, syan ta, syin ta (i) 'all':15 Khmer pre-v.p. (P.A. The rare occurrences of sin are either in ambiguous contexts or are instances of sin 'to officiate'.) A. 207.27 anak nef:z phol) syan ta samayuga yok .... 'all these people together jointly took'. Mid. 6.A.9. syinjarrznwrim sumukkata ( = sam mukha gata) 'all met face to face'. Mid. 3.A.64-5. syin ta jarrznurrz samuggata ( = sam mukha gata) 'all met face to face'. R. 1.27.10 ksatr sabv prades hon cuf:z cfll 'the princes of every country all submitted to him'. (ii). 'almost: more or less: generally speaking': Khmer pre-v.p. Mod. man babak ... mun t arrzpiln sa ruoc praphaef:z khmau 'there were clouds, white at first, then grey, almost black'. hetu, hetu man, het tae 'because': Khmer m. A. 348.nord.2. hetu man man apavada 'because there were objections'. L. 282.10. luon lorn car ea hetu can' smof:z sman braf:z ang 'she cajoled (them) with her chatter because she wanted to make friends with them'. C. Kun cau 30. hetu min sralaii' 'because you do not like it'. Mod. hetu tae man sangram 'because there was war. .. hey, !hey, hoey (i) 'and (then)': Khmer m. Mid. 6.1. anak cov bhikkh A. san !!:!!J!.. cah (=Mod. Khm. car) ca,:k phon. 'The monk A. reconstructed (this) and then wrote the inscription as well'. R. 1.41.9-11. file kiln koet bl udar hon prae pras brat' phan hoey mak prasab yal' vjn 'as if (they were) their own children (who) had been separated from them and were reunited, seen once again'. Mod. as' karrzlarrzn hoeyjhf grun phan 'exhausted and feverish too'. (ii) 'having already: already; by then: by now; and so': Khmer f. A. 413.II 53. adhi:fthan pvas !hey dep dran braf:z carat cuf:z arrzbi subarl)aprasad 'Being now ordained (lit. established in the religious life), he descended from the golden tower'. Mid. 6.A.11. surec !!:!!J!.. debv yen khnurrzm rantap kraya pilja. 'This done (lit. finished already), we prepared the things-needed for the offering'. 15. I feel convinced, after observing for many years that pre-modern occurrences of sin syan are preceded by a plural subject (or, rarely, singular but if so needing the translation 'wholly'), that sfn -syan has the function of linking a plural subject to the verb. Frequently, e.g. in R., the verb is separated from the subject by several words and sfn-syan is also useful in marking the main verb. 222

PAGE 231

A diachronic survey of some Khmer particles R. 1.57.8-9. kukhan stap' bral:z pandfil phan srec hoey carrzlan darrzn p[ bral:z ang lul:z fa/' troey. 'Kukhan, having listened to the lord's reply, took the three royal persons across to the other side'. Mod. diii ivan' hoey kli tralap' dau phdaf:z. 'Having bought the things, (he) returned home'. hon (Emphasis): Khmer f. A. 215.35 tac santana yen hon 'our line is completely extinct'. Mid. 8.51. sum pan dauv ket na tussidd hon 'pray to be reborn in the Tusita'. R. 1.57.4. guor hon aii jhap' foy pa. 'It would indeed be good for me to stay with you'. Mod. (Lit.) kurrz jrlet jraek gna file kal gra mun nol:z hon. 'Do not all push in front of each other like the previous time'. /oey. See fey. as'. See is. ay. See ai. i. See ai. is, iss, as' 'all': Khmer pre-n. p. A. 207 .27 yok iss dravya nol:z phon ta ja thlai bhfimi. 'take all those goods as the price of the territory'. Mid. 4.A.31. is iiat gat 'all his relatives'. R. l.13.12. as' da,rzn manuss ay fa lok 'all the men in the world'. Mod. nissitdamn as' 'all the students'. Also, however, as'= 'to have exhausted (the supply)', eg. as hoey 'That's the lot' as' karrzlarrzn 'tired' ( out of strength). ukk 'also': Khmer f. P.A. 493.24. sre A., pradana poii ukk, man jahv... 'the ricefield A., a further gift of Pofi, which was acquired-by-exchange .. .' e See ai. ai, ay, i, e, ai ta, ay ta, ai fa (i) 'to (re gifts)': Khmer pre-n. p. P.A. 44.A.8-9 arrznoy tan aii kloii ran[kjo don pofi A. ai ta vral:z kamratan aii 'gifts of Tan Afi, chief of the husked rice, and of Pofi A. to the god'. P.A. 561.7. pie tdaiy ?!]: ta vral:z ka,rzmratan aii 'other revenues for the god'. (ii) 'at, in': Khmer pre-n.p. P.A. 38.4. sre ai travail 'the ricefield at the reservoir'. P.A. 416.I.5. sre ?!]: stu[k] anrok 'ricefield at Anrok Lake'. A. 933.4. ai hari haralaya 'at Hariharalaya'. A. 235.D.37. ?!]: vi.yayajen vnarrz 'in the district of Jeri Vnarp'. Mid. 2.15. sucarit sarddha ai ta bral:z sasna bra/:z tathagat 'perfect faith in the teaching of the Tathagat'. Mid. 4.B.26. { !eh (=e foe) 'on high'. 223

PAGE 232

JUDITH M. JACOB R. 1.13.12. as' darrzn manuss ta lok 'all the men in the world'. Mod.~ bhnarrz befi 'at Phnom Penh'. (iii) 'as to': Khmer pre-n. p. Mid. 9.32. e sagh sotr man A., B., C., etc. 'as to the monks, there were A., B., C., etc.'. R. 1.37.11 e nan kaikes[ ,:za hetu citt drjsya ditthi drust kac kan' khjap '. 'As for Kaikes~ out of jealousy she wickedly persisted in her wrong attitude'. Mod. khiiurrz viii, khfiurrz min cul citt. 'As for me, I do not like (it)'. arrzvi, arrzbl, bl 'from': Khmer pre n.p. P.A. 44.A.11 dik hera /oh vrai tarrzponn 'from the spring to Cane Wood'. P.A. 44.A.9. kala 'from the time'. P.A. 149.2. dar ta vralJ, kamratan afi 'claim from the god'. A. 56.A.35. travan trapek dau ti uttara 'from the Trapek Reservoir to the north'. Mid. 4.8. ambl chnam 'from the year ... '. R. 1.58.6-7. pabitr carrznay phlflv ,:za stec njn yatra ambl nel;i dau diiragam 'August One, the length of the road that you will travel from here on! It is far to go!'. C. Kram 31. bl tflc 'from childhood'. Mod. bl khan joen dau khan tpfln 'from north to south'. REFERENCES Aymonier, E. Coedes, G. Huffman, F.E. Ishii, Yoneo, Akahi, Osamu & Endo, Noriko. Jacob, Judith M. anau. See nov. 1878 Textes khmers. Publies avec une traduction sommaire. I"" Ser. Saigon (Lithographed). [Contains trans., pp. 68-84, text, pp. 267-97, of the poem Lpoek Aligar Vat 'Edification d'Angkor Vat'.) 1937-66. (ed. & trans.) Inscriptions du Cambodge (Coll. Textes et Documents sur l'Tndochine 3). Hanoi & Paris: Ee. fr. Extr.-Orient & Boccard. 8 vols. 1973. Thai and Cambodian: a case of syntactic borrowing? J. Amer. Orient. Soc. 93(4), 488-509. 1977. A glossarial index of the Sukhothai inscriptions Kyoto: Shoukadoh Pub!. 1965. Notes on the numerals and numeral coefficients in Old, Middle and Modern Khmer. In Indo-Pacific Linguistic Studies (eds.) G. B. Milner & Eugenie J. A. Henderson. Amsterdam: North Holland Pub!. Co., 2, 143-62. [=Lingua 15] 1968. Introduction to Cambodian. London: Oxford Univ. Press. 1977. Sanskrit loanwords in pre-Angkor Khmer. Mon-Khmer Stud. 6, 151-68. 1979. Observations on the uses of reduplication as a poetic device in Khmer. In Studies in Tai and Mon-Khmer phonetics and phonology in 224

PAGE 233

Jenner, P. N. Jenner, P. N. & Pou, Saveros Khin Sok Lewitz, Saveros A diachronic survey of some Khmer particles honour of Eugenie J.A. Henderson (eds.) Theraphan L. Thongkum et al. Bangkok: Chulalongkorn Univ. Press, 111-30. 1981. The role of ta in pre-Angkorian Khmer. Asiedu Sud-Est et Monde Insulindien 12(1-2), 75-90. 1982. Pre-Angkorian gnil:z and gnol:z and the syntax of gi. Asiedu Sud-Est et Monde lnsulindien 13(1-4), 143-53. 1980-81. A lexicon of Khmer morphology, Mon-Khmer Stud. 9-10. Honolulu: Univ. Press Hawaii. 1976. Les cplip' ou 'codes de conduites' khmers: TT Cplip' Prus. Bull. Ee. fr. Extr.-Orient 62, 313-50. 1978. Deux inscriptions tardives du Phnol)l Bakheri: K.465 et K 285. Bull. Ee.fr. Extr.-Orient 65, 271-80. 1980a. L'inscription de Vatta Romlok K 27. Bull. Ee. fr. Extr. Orient 67, 125-31. 1980b. L'inscription de Prah Thom du Kulen K 715. Bull. Ee. fr. Extr.-Orient 67, 133-4. 1969. Note sur la translitteration du cambodgien. Bull. Ee.fr. Extr. Orient 55, 163-9. 1970 Textes en Khmer moyen. Inscriptions modernes d'Angkor 2 et 3. Bull. Ee. fr. Extr.-Orient 57, 99-126. 1971. Inscriptions modernes d'Angkor 4, 5, 6 et 7. Bull. Ee.fr. Extr. Orient 58, 105-23. 1972a. Inscriptions modernes d'Angkor I, 8 et 9. Bull. Ee.fr. Extr Orient 59, 101-21. 1972b. Inscriptions modernes d'Angkor 10, 11, 12, 13, 14, 15, 16a, 16b, et 16c. Bull. Ee.fr. Extr.-Orient 59, 221-49. Nacaskul, Karnchana 1962. A study of cognate words in Thai and Cambodian. M.A. Pou, Saveros Pou, Saveros & Jenner, P. N. Thesis, Univ. London. 1975. Note sur la date du poeme d' Arigar Vat. J. Asia/. 263, 119-24. 1976. Recherches sur le vocabulaire cambodgien (IX). J. Asia/. 264, 333-55. 1977a. Etudes sur le Rlimakerti (XVf'-XVIf' siecles). (Puhl. Ee. fr. Extr.-Orient. 111). Paris: Ee. fr. Extr.-Orient. 1977b. Inscriptions en khmer moyen de Vat Athvea (K.261). Bull. Ee.fr. Extr.-Orient 64, 151-66. 1978. Inscription dite de Brai Svay ou 'Bois des manguiers' de Sukhoday. Bull. Ee. fr. Extr.-Orient 65, 333-59. 1979. Ramakerti (XVf'-XV/f' siecles). Texte Khmer publie. (Puhl. Ee. fr. Extr.-Orient. vol. 117). Paris: Ee. fr. Extr.-Orient. 1975. Les cplip' ou 'codes de conduite' khmers: I. Cplip' kerti kal. Bull. Ee. fr. Extr.-Orient. 62, 369-94. 1977. Les cplip' ou 'codes de conduite' khmers: TIT. Cplip' kiln cau. Bull. Ee. fr. Extr.-Orient. 64, 167-215. 1978. Les cplip' ou 'codes de conduite' khmers: TV. Cpap' Rlijaneti ou Cplip' Bral:z Rlijasambhlir. Bull. Ee.fr. Extr.-Orient. 65, 361-402. 1979. Les cplip' ou 'codes de conduite' khmers: V. Cplip' kram, Bull. Ee. fr. Extr.-Orient. 66, 129-60. 225

PAGE 235

THE FORM SYAN IN ANGKORIAN KHMER Philip N. Jenner In at least one respect the study of ancient languages known only from written monuments is like the study of contemporary languages which have never been reduced to writing: in both cases the investigator encounters as a matter of course lexical items which are not readily amenable to analysis. What is sometimes forgotten is that the linguist working on a living language normally has recourse to informants, whereas the linguist working on a dead language can have no such guidance. Hence, if most of the ancient lexicon is known or knowable, certain forms prove resistant to identification, and alternative means must be brought into play in attempting to account for them. It is not my intention to claim that the difficulties confronting us in Old Khmer are as numerous or as perplexing as they seem to be, or to have been, in many another ancient language. Since the first inscriptions were published by Aymonier a century ago we have, thanks to the painstaking work of a few dedicated French scholars, seen the gradual elaboration of a tolerably good understanding of Old Khmer. By 1966, when the eighth and last volume of Credes's monumental Inscriptions du Cambodge [ = C] made its appearance, most of the texts in the growing corpus of Khmer epigraphy had been analysed and explicated with an acceptable degree of reliability. By that year the bulk of the lexicon was fairly well understood, but a good many specialised terms had not been worked out to our satisfaction. This is still the case today, when a good many lexical and other problems continue to elude us. Among these is a small class of forms which appear to perform grammatical functions. 1 Typical of this class is syan, which I propose to discuss here. Not found in the pre-Angkorian inscriptions recovered so far,2 this orthographic form is attested in Angkorian Khmer well over a hundred and fifty times. In addition, its life is extended through the Middle Khmer period, where it appears variously as syan (A.D. 1560-77), syin (1560-1701 ), and sin (1587-1706), with the apparent nonce-forms sin (1620), siny (1696) and sin (1701). As to the meaning of the Angkorian form, Aymonier (1883: 494, n. l) took it as synonymous with Sanskrit kevalam 'solely, entirely' and as the source of modern Khmer 'sin' by which he I. There were no published studies of Old Khmer grammar until the valuable sketch by Saveros Pou (1979). The enquiry presented here goes hand in hand with my earlier Asie du Sud-Est et Monde lnsulindien studies (1981, 1982). 2. The language of the pre-Angkorian period, conventionally ending with the founding of Angkor in A.D. 802, is manifested in the form of two dialects. The principal or A dialect, among other criteria, lacks the high falling diphthongs [i:;i), [w:;i], [u:;i] and their short counterparts. The lesser or B dialect, conforming with the phonology of the Angkorian period, has these diphthongs. The orthographic form syan is ambiguous but must have been realized with a high falling diphthong: [si:;iIJ), [si;iIJ], [sw:;iIJ), [sw;iIJJ. It could, therefore, have occurred in pre-Angkorian as a dialect B form, but would take some other form in dialect A. 227

PAGE 236

PHILIP N. JENNER evidently meant siha /syTJ/ [SYTJ] 'nearly all, almost'.3 In 1913 Credes (Parmentier 1913: 14) understood it in the same sense as Aymonier and rendered it 'sans exception'. In 1915, however, Finot (1915: 106) re-defined it as a pronoun, glossing syah ta as 'qui'. For most of the half-century down to 1966, Credes and his contemporaries treated the form in various ways which are impossible to reconcile. Thus in the largest number of cases, representing about 35% of the total, we find it expressed by appropriate forms of tout. Yet in roughly 29% of its occurrences it does not seem to be rendered at all, at least overtly,4 even in those fairly numerous cases where the idea of tout would not appear out of place. Several times it is expressed by the related notion of ensemble. In other passages, representing a little over 15% of the total, we find it rendered by 'comme', 'tel', 'reellement' and, in one instance, 'voila:. It must be said, however, that in none of these latter cases is it absolutely certain that syah is being expressly rendered at all. Indeed, one has the impression that Credes and his contemporaries allowed themselves in some instances to be led into the translation they give under pressure from the supposed contexts. This uncertainty is greatest in passages in which figure appropriate forms of etre: one simply does not know whether syah is being so rendered or is being passed over in silence. To this diversity of interpretations may be added the treatment of syah and its variants in Middle Khmer texts. During the 1970s my learned teacher Saveros Pou redefined syah as an anaphoric 'pronoun or particle' (sic), holding at the same time that the Middle Khmer form or forms were the source of modern Khmer siha /syTJ/ 'nearly'. As far as I am aware, however, it is nowhere rendered as an anaphoric pronoun in any of her superb translations. Of the fifty-seven Middle Khmer occurrences of syah and its variants collected for the present study, twenty-seven (48%) are not overtly rendered at all. Twelve, or 21 %, mostly in combination with what we are no doubt justified in now calling the equational copula Ja /Jaa -> cii;}/, are conceivably assigned a copular function or, if this is not so, are also unrendered. In only three cases is syah represented by 'comme'. On the other hand, in eight cases (14.5%), French and my own English translations read as if syah marked the apodosis of an 'if' or 'when' clause, or as an unexpressed 'then' introducing the consequence of a prior clause. We also have a small number of instances in which syah is, as I might say, plainly none of the foregoing but seems to express a contrastive ('nevertheless') or instrumental ('thereby') idea. In only two cases, both suspicious, does it appear to be taken as a pronoun. 5 I may as well mention 3. As will be seen, he may have had in mind modern siha /s?:JQ/ [SYrJ] 'to rest'. 4. In addition to the occurrences tallied here, in over 9% of its instances the passages in which syah occurs are left untranslated for one reason or another. The Angkorian corpus also includes twelve passages with syah which are too garbled to be used in the present discussion. 5. I might mention in this connection that I have considered the possibility of a loan relationship, in one direction or the other, between Middle Khmer syah sih and Thai /syrJ/ (for the corresponding Lao form see Kerr 1972: 493b), usually treated as a relative pronoun 'used in reference to a person, an animal, or an inanimate object... in the nominative, objective, or possessive case' (Sethaputra 1965: I, 349ab, who adds that it is also 'used in a literary context to introduce a noun in the objective case after a transitive verb, where in ordinary language no preposition (sic) is required'; cf. Haas 1964: 157b; McFarland 1944: 309b). Since it seems to yield no useful results, my consideration of this possibility is not included in the present discussion. 228

PAGE 237

The form syan in Angkorian Khmer here that none of the Old Khmer or Middle Khmer occurrences of syaiz allows interpretations suggestive of modern siiza /syIJ/ 'nearly'; as far as I am able to determine, this sense is unattested in the older language. In view of such radically different interpretations of the meaning and function of syaiz, .we have no choice except to re-examine the data. It will be appreciated that the task of doing so has something of the character of a trial at law in which the evidence, abundant though it may be, is entirely circumstantial. This, taken with the diversity of views expressed by a number of eminent scholars, demands that all of the usable evidence be adduced and weighed. With this in mind I have screened all of the Old Khmer inscriptions available to me and collected every occurrence of syaiz with a view to ensur ing that no usage escaped notice. For Middle Khmer I have gone through all of the so-called 'modern' inscriptions of Angkor (Lewitz 1970-72);6 some seventeen cpli'pa /cbap/ or ethical texts;7 the Lp
PAGE 238

PHILIP N. JENNER opportunity of weighing their judgements and mine for himself. I therefore give the published translations of each passage cited; my own alternatives to those translations are given later. The simplest structural contexts we are concerned with are those in which syah follows a demonstrative pronoun (Dern.) and is itself followed by a noun (N) designating a metal-though there seems to be no good reason to take the metallic nature of the noun as in any way obligatory. In each case the Dern.+ syah + N sequence is preceded by a more or less lengthy list, which I abbreviate, of objects forming part of an endowment: (1) vaudi mviiy svok mviiy... 'arghya piidya mviiy tarripar nef:z syah priik (K.171: 7-8), 'Un vaudi, un plateau, ... quatre vases pour le lavage des pieds (iirghya piidya). Tout8 cela en argent' (C VI: 166). (2) cancyiin 1 ratna ta gf 1 naupura 2 khse chdviil 1 nef:z syah mas (K.669C: 10), ... 1 bague avec 1 joyau, 2 anneaux de cheville, 2 (sic) chaines; tout cela en or' (C I: 182). Only slightly less simple are structures such as the following in which syah, still following a demonstrative pronoun, is itself followed by a noun phrase (NP). The sequences Dern.+ syah + N and Dern.+ syah + NP may of course be considered equivalent. (3) ... me 'yak me narri me des me siin me dvat nef:z syah sruk 'amariilaya (K.598B: 29), 'Les me Tak, Naip, Des, San, Dvat, tous (sic)9 du sruk Amaralaya' (Finot 1928: 77). (4) ... ta duk prasa~ta nef:z mratiiii sri satyayudha nu mratiiii sri ripumatha nef:z syah kvan mratiiii sri prathivinarendra ... (K.956: 58-9), ... Ceux qui conservent cet acte inscrit sont Mratafi (:ri Satyayudha et Mratafi (:ri Ripumatha(na), enfants de Mratafi (:ri Prthivinarendra' (C VII: 135). (5) kamrateh 'aii yogf ta pviis ta nef:zh phye phlu purvvottara tfrthodyiinapu~piiriima nef:z syah dharmma kamrateh 'an didai ra ... (K.139B: 7-10), 'Les seigneurs Yogin qui sont entres en religion ici confient le chemin du nord-est, le bain, le pare, le jardin fleuri: ce sont les reuvres pies de chacun des seigneurs' (C III: 179). In such passages as the following we see that the place occupied by Dern. in the preceding sequences may be filled by an NP: (6) ... patigraha raupya 2 khliis 2 vodi priik 2 bhiijana dramvah 1 bhiijana khpac 1 syah hanfra bhiijana pralvah 3 bhiijana ta madhyama 6 ... (K.669C: 15-6), ... 2 crachoirs d'argent, 2 agrafes, 2 vodf d'argent, 1 recipient dramvah, 1 recipient decore tout en hanfra, 3 grands recipients, 6 recipients moyens, .. .' (C I: 183). (7) ... dep ref:z ta dai ti syah dak~ilJO. (K.263D: 44), ... ensuite on en choisit d'autres comme offrande (dak~i1Jii)' (C IV: 138). (8) karristeh siintilak~mi pahket chloii haridiitta chloii somasarmma syah 8. Here and hereafter, those forms which T take to be intended to express syah are italicized. 9. One would expect toutes. 230

PAGE 239

The form syan in Angkorian Khmer bhagavata pa,rire (K.989B: 16), 'Ka111sten (:antilak~mi donna naissance a Chlofi Haridatta et a Chlofi Somayarman, tous (deux) bhagavata serviteurs' (C VII: 183). (9) ... vaudi 3 katliha 5 svok 10 syaiz tap prli,ri jaizjyaiz padiga{1 4 ... (K.263D: 14), ... 3 vaudi; 5 bassines, 10 plateaux, soit 15 jaizjyaiz; 4 crachoirs; .. .' (C IV: 137). A further step toward structural complexity is seen in cases of the following type in which the NP following syaiz is introduced by the subordinating conjunction ta without a grammatical head. As far as the data show, this type is not common. (10) kamrateiz sivlisrama nu steiz 'ai'i vna,ri kansli yok kanmvliy 3 strijana syaiz ta sahodara 'a,rivi sruk kuti ... (K.235D: 24-5), 'Le seigneur du (:ivayrama et le steiz ai'i de Vna111 Kansa prirent trois femmes, leurs nieces, toutes de meme mere, provenant du sruk de Kuti .. .' (Credes & Dupont 1943-46: 117). (11) neb syaiz ta gi kalpana dau ta vrab kamrateiz 'ai'i sri campesvara sap chna,ri (K.99S: 11-12), ... voila ce qui doit etre fourni a V.K.A. (:rI Campeyvara tousles ans' (C VI: 112). (12) nau 'a[,rijpall punyasrama kuti sruk sre bhujylikara ki'iu,ri (sic) dravya phoiz xxxxx dravya syaiz ta vrab rajapunya (K.19: 17-8), 'toutes les fondations, monasteres, cellules, villages, rizieres, revenus des terres, esclaves, biens de toute sorte ... ces biens sont des fondations royales' (C VI: 146). With the foregoing type the NP following syaiz ta is realised as N, as Dern.+ N, and as N + N. The next structural type, which is the most abundantly represented of all in the Angkorian data, consists of syaiz ta followed by a Verb phrase (VP). This is the same as saying that syaiz is followed by an NP consisting of ta+ VP; viz., a 'headless ta phrase' in which the ta subordinates the VP not to the preceding syaiz but to some such unmanifested headword as 'nak 'person' (qv. Jenner 1981). (13) steiz 'an sivasoma nu steiz 'ai'i vlimasiva syaiz ta cat sivasrama sthlipanli vrab nob (K.235D: 7-8), 'Le steiz ai'i (:ivasoma et le sten ai'i Vamayiva, ensemble, etablirent le (:ivayrama, y fonderent un sanctuaire' (Credes & Dupont 1943-46: 112). (14) gramavrddha syaiz ta sapatha kathli ruva bhumi 'anin ta nirmula krau go! (K.598B: 39), 'Les anciens des villages preterent serment et dirent que cette terre d'Anin etait sans maitre et hors des bornes'. (15) 'nak neb phoiz syaiz ta samayuga yok iss dravya nob phoiz ta jli thlai bhumi (K.207: 27-8), 'Tous ces gens ensemble ont pris tous ces biens comme prix de la terre' (C III: 21). (16) neb syaiz ta dau dlir bhumi nli thve vrab ca,rinli,ri (K.425: 8), 'Tels ont ete ceux qui sont alles demander le terrain pour instituer la prestation' (C II: 144). 231

PAGE 240

PHILIP N. JENNER (17) teiz bhava x teiz k,:~ teiz rudrllf}i teiz ke ni teiz x te (sic) so teiz sa[ rajsvati syaiz ta dar vralJ karu1Jliprasli[ daj ... (K.61B: 7-9), 'Teti. Bhava, Teti. Kr~. Teti. RudraQi, Teti.-, Teti.-, Teti. Sarasvati obtinrent toutes de la faveur royale -.. .' (C VII: 22). (18) ... nau 'nak ta 'aizgvay ta gi sruk nelJ nu dharmma sre nolJ syan ta oy piijll [ka]mraten 'an ta gi dvadasr phon pratipaksa ... (K.100: 3-4), 'Les gens installes dans ce pays et dans Jes fondations et sur ces rizieres, off rent la piijll au K.A. le douzieme jour de chaque quinzaine .. .' ( C VI: 215). (19) ... vol ekavllkya man xxxx [ten hyan]n vasanta nu Ion ney vra/J chpllr syan 'yat santllna fey... (K.208: 53-4), '(Ceux-ci) declarerent unanimement que Teti. Hyati. Vasanta et Lon Ney des saints jardins etaient reel/ement sans descendance .. .' (C VI: 292). A related structural type is seen in a few cases in which syan, preceded as usual by an NP, is followed by an NP manifested as the complementiser man+ a VP. In all cases, the man serves as a relative pronoun in the objective case, its antecedent being either animate or inanimate. (20) khnuffl sata mvliy JO sakarma pvlln phoiz ne/J syan man jvan ta vralJ kamraten 'an sivaliizga thvliy ta vralJ plida kamraten 'an sri siiryyavarmmadeva (K.212A: 11-5), 'Cent dix esclaves et quatre employes (sakarma), tous ces gens sont offerts au V.K.A. <;ivaliti.ga et remis a S.M. le roi <;ri Suryavarmadeva, .. .' (C III: 32). (21) gi no!J sre no!J syan man oy ta vralJ [kal'[lmrateiz 'a]n sivalinga nu vra/J kal'[lmraten 'an sivapada (K.353S: 26-7), 'Ces rizieres ont ete donnees a V.K.A. <;ivaliti.ga et a V.K.A. <;ivapada' (CV: 139). This leaves us, finally, with a number of passages in which syan, still preceded by an NP, is followed directly by a VP without an intervening ta. Note that in five of the following cases the verb following it is passivised by the marker ti. (22) xx khnul'[l vralJ nelJ phon ta daiy ti /en xx vralJ tapasvi bhagavat pllda vralJ kal'[lmrateiz 'an ta gurujvan klila sthlipaka syaiz codita 'an tajma(1 vralJ tapasvi vidyllspada gi pi man ta upakal[ pa j ka jvlln ta vralJ sivalinga ... (K.523D: 18-24), '-tous ces esclaves du dieu et Jes autres, -Vra}:i. Tapasvi Bhagavat Pada V.K.A. ta Guru me Jes a offerts au moment ou ii fit la fondation, en m'incitant, moi qui ai nom Vral). Tapasvi Vidyaspada, a faire Jes preparatifs pour Jes offrir au saint <;ivaliti.ga .. .' (C III: 141). (23) lvolJ ta 1035 saka pi vralJ plida kamraten 'an sri siiryyavarmmadeva ta jll vra/J cau mllt,:pak~a vra/J pllda karaten 'an sri jayavarmmadeva nu vra/J pllda kamraten 'an sri dharal}indravarmmadeva svey vra/J dharmmarlljya 'anjen bhagavat puda kamraten 'an ta guru sri divllkarapaf)<}ita jli vralJ guru gi ta thve rlljllbhi~eka man vralJ pllda kamraten 'an syan thve vra/J dik~li ryyan iss siddhlinta phoiz ... (K.194: 26-9), 'En 1035 yaka ... lorsque S.M. <;ri Suryavarmadeva, petit-232

PAGE 241

The form syali. in Angkorian Khmer neveu en ligne maternelle de S.M. <;ri Jayavarmadeva et de S.M. <;ri DharaQindravarmadeva, acceda a la sainte royaute, ii invita le venerable seigneur Guru <;ri DivakarapaQ<;lita a remplir Jes fonctions de Vral). Guru pour celebrer le sacre royal. Alors Sa Majeste accomplit la sainte initiation (vra}J dik!;a), etudia toutes Jes sciences (siddhanta), .. .' (O:edes & Dupont 1943-46: 146). (24) nef:i syan ti thvay ja vraf:i rajadharmma (K.33: 30), 'Tous ces (dons) sont offerts a titre de fondation royale (rajadharma)' ( C III: 152). (25) sril ta kh 'val nef:i phon phle chpar ta nof:ih phonn syan ti jvan ta vraf:i ka,rzmraten 'aii ekiidasamukha ... (K. I 68: 11-3), 'Tout ce paddy ... et tous Jes fruits de ces jardins sont offerts a V.K.A. Ekadac;amukha .. .' (C VI: 169). (26) ... nu 'angiisa nu kaiije chniin kaiije kalpita khjen phon nu duk pay syan ti yok dau uk nu sthali ce{1 dlaf:i (K.353N: 33-4), ... nu a. distribuer la nourriture, paniers en forme de marmite, paniers kalpita, kf:ijen, pour mettre la nourriture, tout cela est aussi emporte avec Jes sthiili, jarres et dlaf:i' (CV: 142). (27) ... ti pascima vayavya 'a,rzvi !en sthapana sre lvaf:i travan xx dau lvafijen x vin uttara prasap vraf:i phlu nob phon syan ti jau ta cak sviiy ... (K.353S: 31-2), 'A l'ouest et au nord-ouest, depuis Len Sthapana Sre jusqu'au bassin .... revenant jusqu'au bassin au pied de ... ; au nord, touchant ces chemins sacres; tout cela a ete achete a Cak Svay; .. .' (C V: 140). (28) nef:i syan ti cii,rz ca,rznii,rz ta ka,rzmraten jagat srt vrddhesvara (K.33: 23-4), 'Ces (terres) sont affectees au service des fournitures pour le
PAGE 242

PHILIP N. JENNER agony of being led through a critical review of the data, and simply assert here that few of the interpretations given above stand up under such a simple test. For example, the idea of 'solely, entirely' or 'sans exception' would not appear out of place in examples (I), (2), (5), (6), (7) and (11); it would not do in the remaining citations. All other imaginable adverbial ideas have been tried but none seems to fit. On the other hand, the notion of tout has been particularly beguiling. It is seen in nine or more of the above citations and, as has been said, with these we may group the related idea of ensemble seen in (13) and (15). One thing that seems fairly clear is that in most of these examples the immediate constituents are interpreted as neI:i syaii. I prak (I), nel:i syaii. I mas (2), and so on, in which syaiz ( whether construed as a noun, a pronoun, or a verb) is seen as attributive to ne/J. by virtue of its position. However this may be, it is curious that tout would have been no less plausible in nine other of our citations, namely (4), (5), (8), (11), (13), (18), (21), (25), and (28). The fact that it is not used in these cases seems to reflect uncertainty as much as inconsistency. The main point to be considered, however, is that in examples (7), (9), (14), (22) and (23) the idea of tout is pretty clearly excluded. The fact that syaiz has been taken in senses other than tout, or has been left unexpressed, cannot be ignored, and the more one weighs those cases in which it is understood as tout the more one believes that this idea is contextually derived and that such dubious cases as example (8) were influenced by the apparent preponderance elsewhere of the tout idea. The same kind of review must be made in weighing the possibility that syaiz is a relative pronoun, as Finot thought. In this case a cursory examination suffices. While one might be tempted to take it as a pronoun in as many as fifteen of the above citations, to do so does violence to each passage. We now know enough about the role of all the other elements in most of these examples to be on our guard against such interpretations, though our knowledge is still imperfect. I have recently demonstrated the possibility of 'double' demonstratives in Old Khmer (Jenner 1982) and this might suggest that syaiz man could be something of the kind. But consider the following: (29) ne/J. bhumi ta ro/J.h ne/J.h man vap 'amarananta dun syaiz man }van ta vra!J. no!J. ... (K.693B: 20-1 ). A close rendering of this would be 'These lands, aforesaid, which vap Amarananda bought syaiz what (he) offers to the sanctuary .. .' 10 No conclusions can be drawn from this or from our two other syaiz man citations, (20) and (21 ), since all three passages might have a zero copula with syaiz, conceivably, duplicating the office of man. But all the other evidence indicates that this is not so. The claim that syaiz fulfils an anaphoric function is as difficult to disprove as it is to prove. Re-examination of our citations yields mixed I 0. Cf. CV: 208: ... telles sont les terres que Yap Amarananta a achetees et qui sont offertes ace temple .. .' 234

PAGE 243

The form syan in Angkorian Khmer results: in six cases the possibility seems very unlikely, while in the remaining cases the possibility is present. In citations (I) to (5) plus (13), (16), (20), (24) and (28) together with citation (21), in which we have respectively ne!J syah and no!J syah conceivably standing to the left of a binary cut, one wonders why the syah and not the ne/J or no!J should be anaphoric, and also why the alleged anaphor should be attributive to the demonstrative, as it would be, when syah alone or with a following ne!J or no!J would be expected. 11 The claim for syah as anaphoric pronoun ( or particle) is too weak to be applied to all of our citations, let alone most of the data collected but not given here, and is hence unconvincing. When we turn to consider the remaining examples given above, we find them nearly equally divided between those in which syah is not expressly rendered at all and those in which it seems to be rendered by appropriate forms of etre. The former comprise (4), (14), (18), (22), (23), and probably (21) and (28); the latter comprise (5), (12) (19), and probably (21), (25), and possibly (28). If none of my predecessors has advanced the claim that syah is a copula or copula-like verb, it has not been made entirely clear earlier in this discussion that such a claim is almost implicit in a good many of their interpretations both of Old Khmer and of Middle Khmer texts. In a remarkable number of cases it is as if each context forced them into such a position without their being aware of it. Indeed, the moment we take a new look at our twenty-nine citations and consider them apart from their renderings, we are obliged to allow that many of them could equally well, considering the nature of our enquiry, be analysed as nelj I syan prak (!), nelj I syan mas, and so on, in which syah could be functioning as a copula identifying ne!J or another subject with a predicate. To test this hypothesis, I give here and now my own fairly close versions of our twenty-nine examples, my tentative equivalents of syah being shown in italics: (1) 'One vaudi, one svok, four. .. (to hold) water for washing the feet, these consisting of silver.' (2) ... one ring, one jewel (belonging) thereto, 2 anklets, one khse chdval, these being of gold.' (3) ... me 'Yak, me Naqi, me Des, me San, me Dvat-these being of the land of Amaralaya.' (4) ... Those having this edict in their safekeeping (are) the lord Sri Satyayudha and the lord Sri Ripumatha, these being sons of the lord Sri Prrthivinarendra.' (5) 'Our high lords the yogin who have been ordained here give over the northeast road, the garden on the tfrtha [bathing-place] (and) the flower garden, these being their several pious works .. .' (6) ... 2 silver cuspidors, 2 clasps, 2 silver vaudi, I dramvah vessel, I figured vessel consisting of hanfra [an unidentified alloy], 3 vessels with spouts, 6 vessels of medium size .. .' 11. The sequences syaiz ne(1 and syaiz nof:, are not attested in Old Khmer. This is hardly the place to take up the question of anaphora and its ramifications. Those interested should consult Patricia A. Lee (1981), especially 6.5 on 'Reference, Anaphora, Deixis.' 235

PAGE 244

PHILIP N. JENNER (7) ... (they) then selected others (to serve) as a fee'. (8) 'The kan:zsten Santilak~mi begat the eh/on Haridatta (and) the eh/on Somaarman, (who) were bhagavata in (divine) service.' (9) ... 3 vaudi, 5 kat"iiha (with) 10 svok totalling fifteen, an upright panel, 4 cuspidors .. .' (I 0) 'The high lord of the Sivasrama and the sten 'an of Vnarp. Kansa took three female nephews/nieces (who) were co-uterine (and) from the land of Kuti .. .' (11) 'These (items) constitute the endowment gong to the vraf:i.12 Our High Lord Sri Campevara each year.' (12) 'All (these) gifts of lisramas, cells, villages, ricelands, rents, slaves (and) objects ... (all this) valuable property comprises the vrab the royal gift.' (13) 'The sten 'an Sivasoma and the sten 'an Vamaiva are the ones who founded the Sivarama (and) set up the image therein.' (14) 'The village elders are the ones who declared under oath that the land of 'Anin (was) uninhabited (and) outside (anyone's) bounds'. (15) 'These individuals are ones who joined together to take all of these possessions as equal to the value of the land.' (16) 'These are the ones who went and claimed the land (as) a place on which to establish the vraf:i. the foundation.' (17) 'The ten Bhava .. the ten Kr~, the ten Rudrai:ii, the ten Ke Ni (?), the ten ... te So (?), (and) the ten Sarasvati are the ones who claimed the vrab the royal favour...' (18) ... the individuals who are settled on this land and on its gift of ricefields are ones who (shall) offer a sacrifice to Our High Lord on the twelfth days of each fortnight...' (19) ... (they) declared in one voice that... the ten hyan Vasanta and the Ion Ney of Vral). Chpar were ones who had no family whatever. .' (20) 'One hundred and 10 slaves (and) four helpers, -these are what (he) offered to the vrah Our High Lord of the Sivalinga (and) presented to His Majesty Our. High Lord Sri Suryavarman.' (21) 'Those same _ricefields are what (they) gave to the vra,.f:i. Our High Lord of the Sivalinga and to the vrab Our High Lord Sivapada.' (22) ... these slaves of the sanctuary and others, ... the vraf:i. the tapasvin bhagavat piida (and) the vraf:i. Our High Lord the Guru offered (them) on the occasion of (his) consecration of the image (and) enjoined me, named the vrab the tapasvin ,Vidyaspada, to make arrangements to offer (them) to the vrab Jhe Sivalinga .. .' (23) 'In the year 1035 of the Saka era, when His Majesty Our High Lord Sri Suryavarmadeva-who w~s the vraf:i. the maternal grandchild of His Majesty Our High Lord Sri Jayavarmadeva and of His Majesty Our High Lord Sri Dharal)indravarmadeva-(began to) exercise the just kingship, (he) invited the bhagavat plida Our High Lord the Guru Sri Divakarapai:i<;lita to serve as the vraf:i. the guru, he (being) the one 12. Tn these close translations, vra!J is a noun usually but not always marking any divine or royal being or object, and functions as a headword with which the following NP is in apposition. (Note that the purpose of the closeness of the translations is to show the structure of the Khmer rather than the sense.) 236

PAGE 245

The form syari in Angkorian Khmer to celebrate the king's consecration, (and) His Majesty Our High Lord made his preparations (and) learned all of the siddhanta .. .' (24) 'These were offered (to serve) as a vra!J a royal good work.' (25) 'The paddy in these granaries (and) the yield of these fields are offered to the vra!J Our High Lord Ekadasamukha .. .' (26) ... ahgasa vessels, vessels (in the form of) kettle baskets, kalpita baskets and khjeh, (and) vessels in which to keep (cooked) rice are also taken away together with sthali, ce!J (and) dla!J.' (27) ... westward (and) northwestward from Leri Sthapana Sre to the reservoir. ... (and) on to the outskirts of ... (and) back northward to meet the vra!J the roads (which) were acquired by exchange from Cak Svay .. .' (28) 'These (lands) are assigned to the endowment of Our High Lord of Creation Sri Vrddesvara.' (29) 'These lands, aforesaid, which vap Amarananda bought are what (he) offers to the sanctuary. The first twelve of these new versions, together with my interpretation of the three examples with man-(20), (21) and (29)-seem plausible enough; but my restatement of examples (13) to (19), with syah ta+ VP, and of examples (22) to (28), with syah + VP, seems less plausible at first glance. Having considered these two types for more years than I care to admit, however, I have come to believe that the difficulty lies not in the value I assign to syah but in the circumstance that syah forms part of two patterns of expression one of which seems to lie just beyond our comprehension. The seven examples with syah ta+ VP strike me as not especially exotic variations played upon the simple VP. Thus I suspect that example (13) does not mean precisely what it would mean in fluent English, but is a weak expansion of 'The steh 'aii Sivasoma and the steh 'aii Vamasiva founded the Sivarama (and) set up the image therein.' The seven examples with syah + VP are sharply divided into two groups. The last five-syah ti thvay (24) 'were offered', syah ti jvan (25) 'are offered', syah ti yok dau (26) 'are taken away', syah ti jau (27) 'were acquired by exchange', and syah ti carrz (28) 'are assigned to'-seem quite admissible until it is remembered that my English equivalents would more naturally be expressed by ti thvay, tijvan, and so on, without syah. These five cases, then, are as perplexing as the two examples of the remaining group, syah codita (22) 'enjoined' and syah thve (23) 'made'. For all that, serial verb constructions are so widely used during all periods of the language for which we have documentation that, for the moment at least, I am unwilling to reject the hypothesis on the grounds that I cannot fathom these seven cases. It seems more prudent to allow that we are concerned here with a periphrastic construction the effect of which cannot yet be determined. This situation, it will be remembered, was alluded to in my first two paragraphs above. Rather than leaving the matter here, however, I should, because of its 237

PAGE 246

PHILIP N. JENNER importance to our understanding of Old Khmer, attempt to account for the form and function of syan. If its function is indeed copular, as I claim, how does it happen that the form itself suddenly appears in A.D. 895 and enjoys a life of nearly seven centuries, only to vanish into thin air? If the function it serves is an innovation, we may suppose that the form itself either has a source in the earlier lexicon or is a loan from some other language. I have not found any cognates of syan in Mon or other Mon Khmer languages and have, as I believe, considered all the possibilities outside the latter family. In fact, the only light thrown on the question comes from the claims of Aymonier and Pou. As has been explained, Aymonier opined that Angkorian syan is the source of a modern 'sin', without specifying whether he meant modern sina 'nearly' or modern sina 'to rest'. On the other hand, Pou has stated categorically that Middle Khmer syan led to modern sina 'nearly'. Whether we consider sina or sina, these claims are ones of which I was justifiably sceptical when they first came to my attention, since their authors nowhere troubled to explain the radical line of semantic development involved. In either case, the leap was one that I could not imagine. Yet there is a way of linking modern sina 'nearly' with modern sina 'to rest', and of linking both with Angkorian syan-provided that a copular or quasi-copular function is assigned to the latter. It can be pointed out, first of all, that the semantic range of modern sina is exceedingly narrow: if dictionaries define it as more than 'nearly, almost', it is only to ring the changes on this same idea. This is enough to suggest that the modern meaning is a restriction of some broader one. What is more, forms having only adverbial senses in Khmer are rare. This circumstance permits us to suppose that the adverbial sense of sina represents a reduction from a more general verbal idea. In attempting to reconcile sina and sina therefore, we may posit an earlier transitive meaning for sina 'to be near to, just short of. As to modern sina 'to rest', this form is attested once in pre-Angkorian (K.44B: 10) with the sense of 'to dwell in'. In Angkorian it occurs thirty three times, most often with the meaning 'to officiate' but also four times with the meaning 'to dwell in' (K.754:13; K.413/IV: 12; K.56C: 37; K.70: 16), and once with the meaning 'to depend on' (K.369: 6,7). It persisted through Middle Khmer and into the modern language, where its sense is restricted to '(of monks) to sleep' and '(of magicians) to perform (a rite)'. The Middle Khmer form is unquestionably the source of Thai /s'ifJ/ 'to stay, enter and inhabit, possess, (of spirits) haunt', which tells us much about the range of the Old and Middle Khmer forms. These attested meanings are all we need to show that the idea of 'nearly' represented by modern sina /syfJ/ [sYfJ] is an entirely orthographic specialisation of the Old and Middle Khmer verbal idea dimly recognised in modern sina /s:JfJ/ [sYfJ] as defined above. This recognition, in turn, forces upon us the insight that Angkorian copular syan is probably a doublet of Angkorian sin /sifJ/, as defined above, and that all of the forms in question here make 238

PAGE 247

The form syan in Angkorian Khmer up a fairly compact semantic cluster which can be set up as: 1. (intr.) to stand, hold still, be at rest; (a) to rest, lie; to repose, sleep; (b) to rest, remain, continue; (c) to be inherent or present; to exist, be. 2. (tr.) to remain in or at, inhabit; to dwell in; (a) to lie or reside in, be in the presence of, be near to; (b) to consist of, comprise, constitute; (c) to be present at, preside over, officiate at; to perform or celebrate (rite). The meaning of Angkorian siiz and of the single occurrence of pre Angkorian siiz having been given above, it is not unreasonable to suppose that the syaiz doublet (if such it is) of Angkorian siiz was restricted to the remaining semantic field. All of the data collected show no overlap between Angkorian sin and syan and suggest, rather, that the latter was narrowly limited to the intransitive sense of 'to remain, be' and to the transitive sense of 'to consist of, comprise'. It is hardly necessary to add that we cannot at this late date know the genesis of the syaiz doublet of Angkorian sin. The correspondence between pre-Angkorian and Angkorian sin and what we may now represent as modern sin sin is normal, while that between pre-Angkorian sin, Angkorian syan, and modern sin sin is seen in only a handful of cases-most notably in Angkorian tyaiz /01arJ/ : modern tina /dyTJ/ 'to know' and Angkorian 'yat /qiat/ : modern it /q::it/ 'to lack'. If the doublet relationship is tenable, Angkorian syan is therefore short /si~TJ/ or, possibly, /smaTJ/. That the hypothesis of a doublet relationship explains so much overrides, as I believe, the apparent impossibility of our ever knowing how and why these doublets arose. REFERENCES Aymonier, E. Ccedes, G. Ccedes, G. & Dupont P. Finot, L. 1878. Textes Khmers. Publiees avec une traduction sommaire. 1' Ser. Saigon (Lithographed). 1883. Quelques notions sur Jes inscriptions en vieux khmer. J. Asiat. 1883 (1), 441-505; (2), 199-228. 1937-66. (ed. and trans.) Inscriptions du Cambodge (Coll. Textes et Documents sur l'Indochine 3). Hanoi & Paris: Ee. fr. Extr.-Orient & Boccard, 8 vols. [ = C] 1943-46. Les steles de Sd6k Kiik Tho1]1, Phnol]1 Sandak et Pra]:i Vihar. Bull. Ee. fr. Extr.-Orienl 43, 56-154. 1915. Notes d'epigraphie. Bull. Ee.fr. Extr.-Orient 15, 1-211. 1928. Nouvelles inscriptions du Cambodge. Bull. Ee.fr. Extr.-Orienl 28, 43-80. 239

PAGE 248

PHILIP N. JENNER Haas, Mary Jenner, P. N. Kerr, A. D. Lee, Patricia A. Lewitz, Saveros McFarland, G. B. Parmentier, H. Pou, Saveros Sethaputra, So Supina 1964. Thai-English student's dictionary. London: Oxford Univ. Press. 1981. The role of ta in pre-Angkorian Khmer.Asiedu Sud-Est et Monde Jnsu/indien 12 (1-2), 75-90. 1982. Pre-Angkorian gnil;, and gnol;, and the syntax of gi. Asiedu SudEst et Monde Jnsu/indien 13 (1-4), 143-53. 1972. Lao-English dictionary. Washington: Consortium Press, 2 vols. 1981. Topical bibliography in linguistic pragmatics. Working papers in Linguistics (Dept. Ling., Univ. Hawaii at Manoa, Honolulu) 13(1), 1-89. 1969. Note sur la translitteration du cambodgien. Bull. Ee.fr. Extr. Orient 55, 163-9. 1970 Textes en Khmer moyen. Inscriptions modernes d'Angkor 2 et 3. Bull. Ee.fr. Extr.-Orient 57, 99-126. 1971. Inscriptions modernes d'Angkor [=IMA] 4, 5, 6 et 7. Bull. Ee.fr Extr.-Orient 58, 105-23. 1972a. Inscriptions modernes d'Angkor 1, 8 et 9. Bull. Ee. fr. Extr. Orient 59, 101-21. 1972b. Inscriptions modernes d'Angkor 10, 11, 12, 13, 14, 15, 16a, 16b, et 16c. Bull. Ee.fr. Extr.-Orient 59, 221-49. 1973. Inscriptions modernes d'Angkor 17, 18, 19, 20, 21, 22, 23, 24, et 25. Bull. Ee.fr. Extr.-Orient 60, 163-203. 1974. Inscriptions modernes d'Angkor 26, 27, 28, 29, 30, 31, 32 et 33. Bull. Ee.fr. Extr.-Orient 61, 301-37. 1975. Inscriptions modernes d'Angkor 34 et 38. Bull. Ee. fr. Extr. Orient 62, 283-353. 1944 [ < 1941]. Thai-English dictionary. Stanford: Stanford Univ. Press. 1913. Complement a l'inventaire descriptif des monuments du Cambodge. Bull. Ee.fr. Extr.-Orient 13, 1-64. 1977. Etudessur le Ramakerti(XV-XVIII" siecles)(Publ. Ee. fr. Extr. Orient 111 ). Paris: Ee. fr. Extr.-Orient. 1979. Une description de la phrase en vieux-khmer. Mon-Khmer Stud. 8, 139-69. 1965. New model Thai-English dictionary. (library ed.). Samrong, Samud Prakan: So Sethaputra's Press, 2 vols. 1965. Rialia supina. Phnom Penh: Selia-Nwna-Hwta 240

PAGE 249

OLD MON1 s-Christian Bauer The Old Mon (OM) inflectional prefix s-, referred to by Shorto (1971: xxii) as the 'hypothetical', marking in particular 'futurity', has not received the attention it deserves. It has been extensively commented upon by Duroiselle in 192 l, while translating the Ananda plaques and a number of Mon nissaya, and has again been mentioned in Jacob's comparative study of 1963, albeit briefly. The analysis of OM s-is especially challenging for three reasons: (i) tense, aspect and modality cannot clearly be separated; (ii) cognates in other Mon-Khmer languages have so far not been identified; (iii) given the fact that at later stages of Mon the prefix s-has become obsolescent, with no such prefix left in the spoken language today, it can only be analysed on the basis of the attested OM and Early or Epigraphic Middle Mon (EMM) corpus, and must thus be recon structed internally. The semantic complexity of OM s-would go unnoticed were it not for the fortunate circumstance that a number of s-inflected forms occur in glosses accompanying frescoes (the Myinkaba Kubyauk-gyi, for instance) or glazed plaques (at the Ananda). While occurrences of s-in the Shwezigon inscription (EB #1)* would strongly suggest the identification of a tense/modal prefix 'future/irrealis', the glosses, on the contrary, would imply the aspect 'ingressive'. This is especially clear in plaque #565 (Duroiselle #28) and plaque #721 (Duroiselle #189), where the action referred to in the verb (#565 scis 'to descend', s-, and #72 l slop 'to enter', s-, from the bases cis and lop respectively) has not yet been accomplished, but has already been initiated. It would then appear that art-historical evidence such as frescoes, plaques, or illuminated manuscripts, can no longer be ignored by linguists. The prefix s-was fully productive in OM. It was prefixed to bases with simple and complex initials. Because of this low selectivity of its host, characteristic for inflectional affixes, it had no allomorph. Whenever syllabic, the vowel in the minor syllable showed allophonic variation a i in its orthographic form, i being confined to base-initials /s-, k-/, the I. Sources for this study: inscriptions have been quoted from transcriptions as published in Epigraphia Birmanica (EB), and, in the case of inscriptions from Thailand, from my own notes. Complete bibliographical references can be found in Shorto (1971), with addenda in 1990 in Bauer (in press) which also includes a glossary. # indicates the number of the inscription or gloss, as listed in the source. It also indicates the syntactic boundary noted in the inscriptions; see Fig. 4-A. (Ed.) 241

PAGE 250

CHRISTIAN BAUER former conditioning, no doubt, the front-colouring of the vowel. Base types and their inflected forms are given in Fig. 1; all but the last are canonical (native) forms in OM. OM s-is found attached to Indo-Aryan (IA) loans. In these cases the loans are invariably nouns in the source language, except Pali (P) nimanteti, but are to be analysed in OM grammar as verbs; non-inflected forms are attested as verbs. s-inflected IA loans are given in Fig. 2. In one instance, OM s-inflection has led to re-analysis: sthapana 'to erect, establish', Sanskrit (Skt.) sthapana 'founding, storing', is re-analysed as a simple form OM thapana.2 OM s-may be combined with other verbal affixes such as the causative p-and -a-, and the frequentative -N-. It may, however, not be combined with the attributive infix -m-, yielding non-predicative forms. Typical OM paradigms are set out in Fig. 3. OM s-inflected forms may occur in declarative sentences, commands, and questions. In serialised constructions only the head-verb is inflected, as in
PAGE 251

Old Mons-base inflected form gloss IA form prathana sprathana 'to pray' Skt. prlirthanli 'prayer' byapar sbyapar 'to render service' Skt. vyliplira 'service' prak:Jt sprak:Jt 'to be clear' Skt. prlikatya nirban snirban 'to attain NirvaQa' Skt. nirvlifJa cuti? scufi? 'to die' P. cuti 'death' nim:m snim;;n 'to invite' P. nimanteti Fig. 2: sinflected IA loans in OM base frequentative causative causative hypothetical hypothetical gloss (base) frequentative frequentative causative frequentative -N-p-, -ap-, -a/ -N-s-I -N-s-I p-I -Ndwk randwk *pcfurk pandwk spandwk 'to be completed' ywk panywk spanywk 'to raise' *cbh canbh scanbh 'to be apparent' *rleh ran/eh raleh 'to dance' Fig. 3: OM non-attributive verbal paradigms Noteworthy is the absence of suppletive forms in negated contexts, a common feature in modern Mon. The OM verbal negative ka!J may be inflected for the 'hypothetical', OM ska!J (as in #7, Fig. 4), or the verbal negative kum ( < Khmer), OM skum (as in #10, Fig. 4). If the verbal negative is inflected for the 'hypothetical', the following verb is not. Problematic are, however, cases where it is not the verbal negative which is inflected but the following verb (as in #9, although there it is sak which is never attested as an inflected form) or the preceding clause or sentence (as in #8). Purposive clauses ('in order to, so as to') are introduced by the clause head OM tnas or OM dna!J; in the latter case, the head is linked to the clause by na, which in some instances is attested without the clause-head (Ananda plaques #715, #780, #798; Kubyauk-gyi gloss #164; see Figs. 5, 7). Whenever the clause is linked by na the following verb must be inflected for the 'hypothetical'. In Middle Mon (MM) clause-linkage also occurs with ma, but, except for tna!J as clause-head, s-inflection is not obligatory. However, as in OM, na clause-linkage requires s-inflection. Rules of linkage and inflection are listed in Figs. 8, 9. The exact details of clause-subordination in Mon need not concern us here; suffice it to say that s-inflection is obligatory when na linkage occurs in OM and MM. To conclude: An examination of OM glosses supports the interpreta-243

PAGE 252

N -I'>-' Translation of OM gloss 565. (Temiya) descending on to the ground 601. They (the two attendants) bring the alms-bowl (and) give it (to the Mahajanaka) 621. Syama informs (his parents) that he is going to enter the forest 721. Mahosadha enters the king's house 756. (Bhuridatta) asks his father for leave to go and observe the silas 816. Vessantara goes to his father 715. Mahosadha tells (the king) that he is going (to Uttarapancala) 780. Vidhura tells (Puvva the Yakkha) he is ready to go (with him) 798. (Vessantara) asks for gold to give alms Comment EB Il.l (text) ... Temiya doubts whether he has still the control of his limbs and gets down from the chariot to ascertain the prefix s here denotes the future, the immediate future (ad # 755) he informs his parents of his intention Evening fell while the Bodhisatta was giving alms; he then returned to the palace, to see his father and mother, before departing in the morning and he tells the king of his intention announces to the Yakkha he is now ready to proceed with him wither he desires Comment EB //.2 (plates) ... he is in the act of descending; his left leg is already outside, his foot resting on a stone ... bids the two servants bring him yellow robes and an alms bowl. In the casket held by the first servant are the robes. The other is holding the bowl. he informs his parents that he intends going to the forest He is asking from his father permission to observe the silas. The king and the queen are seated, he is standing in front of them. When everything is over, Vessantara, accompanied by his wife and children, goes to visit his father. Vessantara, then nm, his son, and near the boy, Maddi carrying Kavhajina on her hip in front of the king, Mahosadha with, being who he is, three attendants Fig. 6. Duroiselle's translations, comments, and annotations to the Ananda glosses; text presented in Fzg. 5.

PAGE 253

Old Mons-Kubyauk-gyi Luce/Ba OM gloss ( Myinkaba) Shin page ref 28 365 tarley taw rjey 'anupiya 'amba XXXXXX barbeh slop rjuh rajagrif:z 131 380 kyek buddha tarley spluf:z das kyek # 164 386 # smih 'asur # wepacit # sucit # paharata # 'asurinda rahu # taw dirse' na s'ar dwan tawattih # 185 395 # rjey sthan nagawanuyyan # 'ut kindok ma taw rjey lanka smic cinlef:z kyek buddha tarley tutaw 'antul 'akas ptit kyal dainlu buhic rjef:z t-ef:z 'ac bhey # # 189 395 #
PAGE 254

CHRISTIAN BAUER clause-head clause-linkage main verb m;}-n;}sAUX* JCT* JCT* OM tn:,s V OM dl)ah nas-V OM nas-V MM tn:,h nas-V MM tn:,h mas-V MM tn:,h s-mik V MM tn:,h na s-mik V MM hot ma-(s-) V MM swak ma-(s-) V MM ma-(s-) V LM** tn:,h na-swak V LM skw? V SM** k!J? V Fig. 8: Rules for clause-linkage and inflection in Mon *AUX = auxiliary complex; JCT = junctural complex. (Ed.) **LM = Modern Literary Mon; SM = Modern Spoken Mon. (Ed.) OM EMM LM SM n;}s-V m;} (s-)V AUX JCT tn:Js tn:,h tn:,h EMM LM dl)ah OM EMM LM/SM na ( ) OM hot EMM swak swak EMM MM/LM Fig. 9: Mon clause-heads ('in order to, so as to') and linkage-types 248

PAGE 255

Fig 4-A scis s'or sgilr sgo' spa