Linguistics in Oceania 9783111418827, 9783111054445


205 105 61MB

English Pages 886 [892] Year 1971

Report DMCA / Copyright

DOWNLOAD PDF FILE

Table of contents :
Editor's Introduction
Contents
Master List Of Abbreviations
Part One: Indigenous Languages
Austronesian
The Austronesian Languages And Proto-Austronesian
Indonesia And Malaysia
Tagalog And Other Major Languages Of The Philippines
Minor Languages Of The Philippines
The Austronesian Languages Of Formosa
The Chamic Languages
Malagasy
The Austronesian Languages Of Australian New Guinea
Languages Of The New Hebrides And Solomon Islands
New Caledonia And The Loyalty Islands
Fijian And Rotuman
Micronesian Languages
The Languages Of Polynesia
Papuan
History Of Research In Papuan Languages
The Papuan Linguistic Situation
Australian
History Of Research In Australian And Tasmanian Languages
Classifications Of Australian Languages, Including Tasmanian
Lexicographic Research In Aboriginal Australia
The Indo-Pacific Hypothesis
The Indo-Pacific Hypothesis
Recommend Papers

Linguistics in Oceania
 9783111418827, 9783111054445

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

CURRENT TRENDS IN LINGUISTICS VOLUME 8 *

CURRENT TRENDS IN LINGUISTICS Edited by THOMAS

A.

SEBEOK

VOLUME

8

Linguistics in Oceania •

Associate Editors: J. DONALD BOWEN - ISIDORE DYEN GEORGE W . GRACE - STEPHEN A . WURM

Assistant Editor: GEOFFREY N . O ' G R A D Y

Assistants to the Editor: ALEXANDRA RAMSAY - LUCIA HADD ZOERCHER

El 1971

MOUTON THE H A G U E • PARIS

© Copyright 1971 in The Netherlands. Mouton & Co. N.V., Publishers, The Hague. No part of this book may be translated or reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publishers.

LIBRARY OF CONGRESS CATALOG CARD NUMBER: 64-3663

Printed in The Netherlands by Mouton & Co., Printers, The Hague.

EDITOR'S I N T R O D U C T I O N

As the observant English anthropologist Geoffrey Gorer pointed out in 1948, 'money and prices figure greatly and consistently in American writing and conversation; ... the cost of an article is a necessary and useful piece of social information; to inform a stranger on such points, when he is likely to be ignorant, is an act of neighborliness' (The American people-. A study in national character, p. 175). In my Introduction to Vol. 7, Linguistics in Sub-Saharan Africa, I judged it illuminating, and perhaps comforting as well, to discuss several of the tangled financial strands of the series as a whole, and offered answers to three of the salient questions I am most often asked about sources and destination in the flow of funds. There, I detailed the support of the U.S. Office of Education, which has contributed the lion's share of over a quarter of a million dollars that it has cost to produce half the volumes in the series. The second largest contributor has been another one of our federal agencies, the National Science Foundation, which assumed full responsibility for four volumes, and major responsibility for a fifth, as follows: Volume 1 2 3 8 10

Title Soviet and East European Linguistics Linguistics in East Asia and South East Asia Theoretical Foundations Linguistics in Oceania Linguistics in North America

Recipient

Amount

Indiana University Foundation

$ 14,950

Indiana University Foundation

$ 19,550

Indiana University Foundation Center for Applied Linguistics Center for Applied Linguistics

$ 12,200 $ 47,502 $ 36,400 (partial)

Total NSF (Grand total, USOE and NSF

$130,602 $395,788)

The NSF grant that made it possible to ready the present volume for press bore the identification GS-1522, and is hereby particularly, and most gratefully, acknowledged. (In addition, the articles by Isidore Dyen were prepared with the Foundation's support, through grant GS-1468.) Lesser amounts were also allocated by two other agencies of the U.S. government

VI

EDITOR'S INTRODUCTION

and by The Canada Council; this additional help for the series will be acknowledged in Vols. 9 and 10, respectively. With the publication of Linguistics in Oceania — the third volume to appear this year — two-thirds of the series has emerged into the light of day. Many reviews, including some very valuable ones, have appeared in linguistic journals the world over, and I have tried to keep track of them. In the Introduction to Vol. 6,1 listed reviews of Vols. 1, 2, or 3 that have come across my desk and that I have profited from; three important references that must now be added are to Mary R. Haas' perceptive judgment of Vol. 2, in the American Anthropologist 72.188-90 (1970), and Jean Dubois' thorough evaluation of Vol. 3, published in the Revue Beige de Philologie et d'Histoire for 1968, No. 4, pp. 1453-58, as well as E. HajiCova's brief review in Jazykovedne aktuality 3.34-5 (1967). Reviews of Vol. 4 (1968) are also beginning to come into public view; see P. Russell-Gebbett's significant (though highly selective) account, in The Year's Work in Modern Language Studies 30.311-18 (1969), and Peter Boyd-Bowman's constructive comments in Language 46.189-92 (1968). This is the state of the volumes yet to appear: Vol. 9, Linguistics in Western Europe, has largely been typeset, and we are currently busy checking the first proofs. The Canadian chapters in Vol. 10, Linguistics in North America, have all been received and are being edited. The U.S. chapters are due in the spring. The entire book will go to press before the summer of this year. Vol. 11, Diachronic, areal, and typological linguistics, will follow in the fall. A few manuscripts are already in hand, with all the rest expected over the next two months. Then the three successive tomes of Vol. 12, Linguistics and adjacent arts and sciences, will begin to flow to the printer. Several outstanding articles have been received at this time, much to our gratification. Some of the substantial as well as economic problems of Vol. 13, Index to Current Trends in Linguistics, Vols. 1-12, which were alluded to in some of my previous Introductions, have, in the meanwhile, been widely explored and are receiving sympathetic consideration. Indeed, alternative ways of producing the indexes discussed do appear feasible, and I am sanguine that a decision can be announced soon. The Editorial Board of Linguistics in Oceania has participated in unprecedented depth in the realization of the volume entrusted to its care. In addition to performing the customary chores of selection and supervision, Dyen and Wurm themselves wrote four of the chapters each, and Bowen, Grace, and O'Grady also each contributed a chapter. Furthermore, Dyen, O'Grady, Wurm (and Zisa) are collaboratively responsible for the contents of the Checklist, and all maps were prepared under O'Grady's superintendence. Uhlenbeck's article is a thoroughly reworked version of a previous contribution, under the same title, to the section on 'Linguistics in South East Asia' in Vol. 2. The article of Lavondes was translated from French by K. M. Fenton (Indiana University), and that of Barthel from German by John F. Davis (University of California, Los

EDITOR'S INTRODUCTION

VII

Angeles). The materials in the Appendix were checked and assembled by Charles A. Zisa (Center for Applied Linguistics). The technical preparation of this book for press was accomplished jointly by Alexandra Ramsay and Lucia Hadd Zoercher. The sections on Indonesian, Papuan, and Australian were the responsibility of Mrs. Zoercher, and the rest of Miss Ramsay. The Master List of Abbreviations, Index of Languages, and Index of Names are the result of teamwork between the two. To the Editorial Board and staff, to the twenty other authors (two of them with multiple contributions), and to the publisher's crew, my thanks — and Aloha kakoul Bloomington, February 23, 1970

THOMAS A . SEBEOK

CONTENTS

EDITOR'S INTRODUCTION

v

MASTER LIST OF ABBREVIATIONS

xi

PART ONE: INDIGENOUS LANGUAGES Austronesian

The Austronesian Languages and Proto-Austronesian, by Isidore Dyen . . . Indonesia and Malaysia, by E. M. Uhlenbeck Tagalog and Other Major Languages of the Philippines, by Ernesto Constantino Minor Languages of the Philippines, by Howard McKaughan The Austronesian Languages of Formosa, by Isidore Dyen The Chamic Languages, by Isidore Dyen Malagasy, by Isidore Dyen The Austronesian Languages of Australian New Guinea, by A. Capell. . . . Languages of the New Hebrides and Solomon Islands, by George W. Grace . New Caledonia and the Loyalty Islands, by André G. Haudricourt Fijian and Rotuman, by George B. Milner Micronesian Languages, by Byron W. Bender The Languages of Polynesia, by Bruce Biggs

5 55 112 155 168 200 211 240 341 359 397 426 466

Papuan

History of Research in Papuan Languages, by D. C. Laycock and C. L. Voorhoeve The Papuan Linguistic Situation, by Stephen A. Wurm

509 541

Australian

History of Research in Australian and Tasmanian Languages, by A. Capell.

661

X

CONTENTS

Classifications of Australian Languages, including Tasmanian, by Stephen A. Wurm Lexicographic Research in Aboriginal Australia, by Geoffrey N. O'Grady . .

721 779

The Indo-Pacific Hypothesis The Indo-Pacific Hypothesis, by Joseph H. Greenberg

807

PART TWO: SELECTED TOPICS

Intrusive Languages English and Other Germanic Languages, by D. C. Laycock French in the Pacific, by K. J. Hollyman Hispanic Languages and Influence in Oceania, by J. Donald Bowen

877 903 938

Asia The Influence of Indian Languages, by J. Gonda Japanese in Taiwan, by T. Sugimoto

955 969

Languages in Contact Pidgins, Creoles and Lingue Franche, by Stephen A. Wurm

999

Socio-Linguistic Problems Language Policy, Language Engineering, and Literacy, by Stephen A. Wurm (New Guinea and Australia), Bonifacio Sibayan (the Philippines), Gregory J. Trifonovitch (Trust Territory of the Pacific Islands), S. T. Alisjahbana (Indonesia and Malaysia), and Henri Lavondes (French Polynesia) . . . . 1025 Style Contrasts in Pacific Languages, by John L. Fischer 1129 Pre-contact Writing in Oceania Pre-contact Writing in Oceania, by Thomas S. Barthel

1165

APPENDIX

Checklist of Oceanic Language and Dialect Names, edited by Geoffrey N. O'Grady and Charles A. Zisa 1189 BIOGRAPHICAL NOTES

1279

INDEX OF NAMES

1287

INDEX OF LANGUAGES

1306

M A S T E R LIST OF A B B R E V I A T I O N S

JOURNALS AND PUBLICATIONS AEH AION-O AJPh AJS AL AmA AnL AnnLat Anthropos AO Archive ArchR AS AUMLA AuÜ BaBu BEFEO BIHPAS Bijd. BijdrTLV BP BS BSEIC BSEO BSL BSOAS BT Bull. Soc. Amer. CAnthr CHM De Gids EiA EM

Acta Ethnographica Academiae Scientiarum Hungaricae (Budapest). Annali, Istituto Universitario Orientale di Napoli, Sezione Orientale (Naples). American Journal of Philology (Baltimore). American Journal of Sociology (Chicago). Acta Linguistica ( = Revue internationale de linguistique structurale, Copenhagen). American Anthropologist (Menasha, Wisconsin). Anthropological Linguistics (Bloomington, Indiana). Annali del Pontificio Museo Missionario Etnologico già Lateranensi (Vatican City). Anthropos ( = Revue internationale d'ethnologie et de linguistique!Internationale Zeitschrift für Völker- und Sprachenkunde, Freiburg, Switzerland). Archiv Orientalni (Prague). The Archive. A Collection of Papers pertaining to Philippine Linguistics (Manila). Archiv für Religionswissenschaft. Heidelberger Akademie der Wissenschaften; Religionswissenschaftliche Gesellschaft, Stockholm (Leipzig & Freiburg). American Speech. A Quarterly of Linguistic Usage (New York). Journal of the Australasian Universities Language and Literature Assoc. (Christchurch, N.Z.). Afrika und Übersee. Sprachen, Kulturen. Folge der Zeitschrift für EingeborenenSprachen (Berlin). Bahsa dan Budaja. Bulletin de l'École Française d'Extrême-Orient (Saigon). The Bulletin of the Institute of History and Philology. Academia Sinica (Shanghai). Bijdragen tot de Taal-, Land- en Volkenkunde van Nederlandsch-Indië (The Hague). Bijdragen tot de Taal-, Land- en Volkenkunde. Uitgegeven door het Koninklijk Instituut voor Taal-, Land- en Volkenkunde (The Hague). Bijdragen van de Philosophische en Theologische Faculteiten der Nederlandsche Jezuieten. Bibliographical Series. Koninklijk Instituut voor Taal-, Land- en Volkenkunde (The Hague). Bulletin de la Société des Études Indochinoises (Saigon). Bulletin de la Société des Études Océaniennes (Papeete, Tahiti). Bulletin de la Société de Linguistique de Paris (Paris). Bulletin of The School of Oriental and African Studies, University of London (London). The Bible Translator. Periodical for the assistance of Bible translators (London). Belge. Bulletin de la Société des Américanistes de Belgique (Brussels). Current Anthropology. A World Journal of the Sciences of Man (Chicago). Cahiers d'Histoire Mondiale (Paris). De Gids (Amsterdam). English in Australia. Australian Association for the Teaching of English (Melbourne). Études mélanésiennes. Société d'études mélanésiennes (Noumea, New Caledonia).

XII

ESE Ethnology E&W FAE FE FM Folk FR GL Globus Homme HZM IAE IG IJAL Indonésie Intisari IPLS IUPAL IZAS JA JAF JAOS JAS JASt JEAS JEL JIA JMBRAS JOS JPNGS JPS JRAI JRAS JRSNSW JRSWA JSBRAS JSOc JSocI JWH KS KSINA KSIV LCC LCC Bull Lg LGU

MASTER LIST OF ABBREVIATIONS

Ergebnisse der Südsee-Expedition 1908-1910 (Hamburg). Ethnology (= International Journal of Cultural and Social Anthropology, Pittsburgh). East and West. Quarterly published by the Istituto Italiano per il Medio ed Estremo Oriente (Rome). Fundamental and Adult Education. United Nations Educational, Scientific and Cultural Organisation (Paris). Folia Ethno-glossica. Blätter für Völkerkunde, Sprachwissenschaft, Verwandtes (Hamburg). Le Français Moderne (Paris). Folk. Dansk Etnografisk Tidsskrift. (Copenhague). The French Review (Baltimore). General Linguistics (Lexington, Kentucky). Globus ( = Peter mann's Geographische Mitteilungen, Hildburghausen, Brunswick). VHomme (= Revue française d'anthropologie, Paris and The Hague). Handelingen van de Zuidnederlandsche Maatschappij voor Taal- en Letterkunde en Geschiedenis (St. Pieters-Jette). Internationales Archiv für Ethnographie (Leiden). De Indische Gids (Leiden). International Journal of American Linguistics (Baltimore). Indonésie (= Tijdschrift gewijd aan het indonesisch cultuurgebied, The Hague). Intisari. The Research Quarterly of Malaysia (Singapore). Indo-Pacific Linguistic Studies, Part I ( = Lingua 14, 1965). Indiana University Publications in Anthropology and Linguistics (Bloomington and The Hague). Internationale Zeitschrift für allgemeine Sprachwissenschaft (Leipzig). Journal Asiatique (Paris). Journal of American Folklore (Philadelphia). Journal of the American Oriental Society (New Haven, Connecticut). Journal of Austronesian Studies (Victoria, British Columbia). Journal of Asian Studies (New York). Journal of East Asiatic Studies (Manila). Journal of English Linguistics (Bellingham, Washington). Journal of the Indian Archipelago and Eastern Asia (Singapore). Journal of the Malayan Branch of the Royal Asiatic Society (Singapore). (Formerly JSBRAS.) Journal of Oriental Studies. Hong Kong University Press (Hong Kong). Journal of the Papua and New Guinea Society (Port Moresby). Journal of the Polynesian Society (Wellington, New Zealand). Journal of the Royal Anthropological Institute of Great Britain and Ireland (London). Journal of the Royal Asiatic Society of Great Britain and Ireland (London). Journal and Proceedings of the Royal Society of New South Wales (Sydney). Journal of the Royal Society of Western Australia (Perth). Journal of the Straits Branch of the Royal Asiatic Society (Singapore). Journal de la Société des Océanistes (Paris). Journal of Social Issues. Society for the Psychological Study of Social Issues (Ann Arbor, Michigan). Journal of World History (Paris) ( = CHM). Koloniale Studien. Vereeniging voor Studie van Koloniaalmaatschappelijke vraagstukken (Weltevreden, Batavia). Kratkie Soobscenija Instituta narodov Azii (Moscow). Kratkie Soobscenija Instituta Vostokovedenija (Moscow). Linguistic Circle of Canberra, Publications. Series A, Occasional Papers; Series B, Monographs; Series C, Books (Canberra). Linguistic Circle of Canberra Bulletin (Canberra). Language ( = Journal of the Linguistic Society of America, Baltimore). Leningradskij gosudarstvennyj universitet (Leningrad).

MASTER LIST OF ABBREVIATIONS

Lingua

XIII

Lingua ( = International Review of General Linguistics/Revue internationale de linguistique générale, Amsterdam). Linguist The Linguist. The Linguists' Club, 20 Grosvenor Place S.W.I (London). Linguistics Linguistics. An International Review (The Hague). LL Language Learning (Ann Arbor, Michigan). LPosn Lingua Posnaniensis (Posen). LT Levende Talen (Groningen). Man Man. A Record of Anthropological Science (London). MBA Micro-Bibliotheca Anthropos. Posieux (Freiburg) Anthropos Institut. MDS Mitteilungen aus den Deutschen Schutzgebieten (Berlin). MeBa Medan Bahasa. Department Pendidikan, Pendijaran dan Kebudajaan (Djakarta). MellPe Medan Ilmu Pengetahuan. Madjelis Ilmu Pengetahuan Indonesia. Index of Indonesian Learned Periodicals (Djakarta). Micronesian Reporter Micronesian Reporter. Trust Territory Publications Office (Saipan, Mariana Islands). MIO Mitteilungen des Instituts für Orientforschung. Deutsche Akademie der Wissenschaften zu Berlin (Berlin). MISI Madjalah Ilmu-Ilmu Sastra Indonesia [Indonesian Journal of Cultural Studies] (Djakarta). Mission des Isles Mission des Isles. Missions Maristes de l'Océanie (Paris). MLN Modern Language Notes (Baltimore). MNZG Tijdschrift voor zendingswetenschap 'Mededeelingeii (Formerly Mededeelingen van wege het Nederlandsch Zendelingsgenootschap, Rotterdam.) MPhon Le Maître Phonétique. Organe de l'Association Phonétique Internationale (London). MSLL Monograph Series on Languages and Linguistics, Georgetown University (Washington, D.C.). MSOS Mitteilungen des Seminars für Orientalische Sprachen (Berlin). NGS Nieuw Guinea Studien (Den Haag). Nova Guinea Nova Guinea. A Journal of Botany, Zoology, Anthropology, Ethnography, Geology, and Paleontology of the Papuan Region (Leiden). NTS Norsk Tidsskrift for Sprogvidenskap (Oslo). Oceania Oceania. A Journal devoted to the Study of the Native Peoples of Australia, New Guinea and the Islands of the Pacific Ocean (Sydney). OE Oriens Extremus (Wiesbaden). OL Oceanic Linguistics. Special Publication. Pacific and Asian Linguistics Institute, University of Hawaii (Honolulu). OLM Oceania Linguistic Monographs (Sydney). Orbis Orbis. Bulletin international de documentation linguistique (Louvain). Pacific Speech Pacific Speech. Pacific Speech Association, Inc., Department of Speech, University of Hawaiian (Honolulu). PAPhilosS Proceedings of the American Philosophical Society (Philadelphia). Papua Annual Report Papua Annual Report. Government Printer (Canberra). PCLS Monographs, Philippine Center for Language Study (Quezon City). PemBalnd Pembina Bahasa Indonesia (Djakarta). PhilJSci Philippine Journal of Science (Manila). Phonetica Phonetica ( = Internationale Zeitschrift für Phonetik!International Journal of Phonetics, Basel and New York). PICAm Proceedings of the International Congress of Americanists. PICL IX Proceedings of the 9th International Congress of Linguists, Cambridge, Massachusetts, August 27-31,1962. Ed. by Horace G. Lunt. Janua Linguarum, Series Maior 12. The Hague, Mouton, 1964. PJEd Philippine Journal of Education (Quezon City). PJL Philippine Journal for Language Teaching (Quezon City). PL Pacific Linguistics, Series A, B, C (Canberra) (Continuation of LCC.)

XIV

PM PMLA PSR PSSR QJSp RBMIS RNAE RO RomPh Runa SarawakMJ SbBAW SG SIL SJA SL Sociologus SovAA SovEtn Straits Times Annual SPCQB SPiH SSR TBB TBG Temenos Te Reo TeT TFS TNG TNI TPhS Tribus TRSSA UCPAAE UCPSP VBG Vest. ist. mir. kul't. VJa

MASTER LIST OF ABBREVIATIONS

Petermanns [Geographische] Mitteilungen (Gotha). See Globus. Publications of the Modem Language Association of America (New York). Philippine Sociological Review (Manila). Philippine Social Sciences and Humanities Review (Manila). The Quarterly Journal of Speech (Columbus, Missouri). Renward Brandsetter's Monographien zur Indonesischen Sprachforschung (Luzern). Reports of the Norwegian Archeological Expedition to Easter Island and the East Pacific (= Monographs of the School of American Research and the Kon-Tiki Museum, No. 24) (Oslo). 1965. Rocznik orientalistyczny wydaje Polskie towarzystwo orientalistyczne (Krakôw). Romance Philology (Berkeley and Los Angeles). Runa. Archivo para las Ciencias del Hombre, Universidad de Buenos Aires, Instituto de Antropologia (Buenos Aires). Sarawak Museum Journal for the Promotion of Scientific Knowledge and Study of the Natives and Natural History of the Island of Borneo (Kuching). Sitzungsberichte der Bayerischen Akademie der Wissenschaften, Philosophisch-historische Klasse (Munich). Studium Generale (Berlin, Göttingen and Heidelberg). Studies in Linguistics (Buffalo, New York). Southwest Journal of Anthropology (Albuquerque, New Mexico). Studio Linguistica (= Revue de linguistique générale et comparée, Lund). Sociologus (= Zeitschrift für empirische Soziologie, Sozialpsychologische und Ethnologische Forschung, Berlin). Soviet Anthropology and Archaeology. A journal of translations from Soviet sources. International Arts and Sciences Press (White Plains, New York). Sovietskaja Etnografija (Moscow-Leningrad). The Straits Times Annual. The Straits Times Press (Malaya) Ltd. (Singapore). South Pacific Commission Quarterly Bulletin (Sydney). Social Progress in Hawaii (Honolulu). Sociology and Social Research (Los Angeles). Tijdschrift voor het Binnenlandsch Bestuur (Batavia). Tijdschrift voor Indische Taal-, Land- en Volkenkunde/Madjalah untuk Ilmu bahasa, ilmu bumi dan kebudajaan Indonesia. Lembaga kebudajaan Indonesia (Djakarta). (Formerly: Bataviaasch genootschap van kunsten en wetenschappen, Batavia). Temenos. Studies in Comparative Religion presented by scholars in Denmark, Finland, Norway and Sweden. Institute of Folklore and Comparative Religion, University of Turku, Finland. Te Reo. Proceedings of the Linguistic Society of New Zealand (Auckland, New Zealand). Taal en Tongval ( = Driemaandelijks tijdschrift voor de Studie van de Nederlandse volks- en streektalen, Bosvoorde, Belgique). Transactions of the Fijian Society (Suva, Fiji). Tijdschrift Nieuw Guinea (The Hague). Tijdschrift voor Nederlandsch Indië (Batavia). Transactions of the Philological Society (Oxford). Tribus. Veröffentlichungen des Linden-Museums (Heidelberg). Transactions of the Royal Society of South Australia (Adelaide). University of California Publications in American Archaeology and Ethnology (Berkeley). University of California Publications in Semitic Philology (Berkeley). Verhandelingen van het Koninklijk Bataviaasch Genootschap van Kunsten en Wetenschappen (Batavia and The Hague). Vestnik istorii mirovoj kuVtury (Moscow). Voprosy Jazykoznanija (Moscow).

MASTER LIST OF ABBREVIATIONS VKI VKNA VL VSISUH Wending Word WZKM ZAOS ZDMG ZES ZEthn ZKS ZPhon

Verhandelingen van het Koninklijk Instituut voor Taal-, Land- en Volkenkunde (The Hague). Verhandelingen van de Koninklijke Nederlandsche Akademie van Wetenschappen, afdeling Letterkunde. Nieuwe reeks (Amsterdam). Vie et langage (Paris). Veröffentlichungen des Seminars für Indonesische und Südseesprachen der Universität Hamburg (Berlin). Wending. Maandblad voor evangelie en cultuur (The Hague). Word (= Journal of the Linguistic Circle of New York, New York). Wiener Zeitschrift für die Kunde des Morgenlandes (Wien). Zeitschrift für Afrikanische, Ozeanische und Ostasiatische Sprachen (Berlin). (Formerly: Zeitschrift für Afrikanische und Ozeanische Sprachen.) Zeitschrift für Deutschen Morgenländischen Gesellschaft (Wiesbaden). Zeitschrift für Eingeborenen-Sprachen (Berlin). Zeitschrift für Ethnologie (Brunswick). Zeitschrift für Kolonialsprachen (Berlin). Zeitschrift für Phonetik, Sprachwissenschaft, und Kommunikationsforschung (Berlin).

OTHER ABBREVIATIONS ACLS A.I.A.S. ANZAAS ETV F.I.D.E.S. HRAF INSEE LMS NAEB NDEA PCLS PICS S.I.L. TESL

XV

American Council of Learned Societies. New York. Australian Institute of Aboriginal Studies. Canberra. Australian and New Zealand Association for the Advancement of Science. Educational Television. Fonds d'Investissement et de Développement Économique et Social. [F] Human Relations Area Files. New Haven, Connecticut. Institut National de la Statistique et des Études Economiques. [F] London Missionary Society. National Association of Educational Broadcasters. [US] National Defense Education Act. [US] Philippine Center for Language Study. [Phil.]. Pacific Islands Central School. Summer Institute of Linguistics. Headquarters: Santa Ana, California. Teaching English as a Second Language.

PART O N E

INDIGENOUS LANGUAGES

AUSTRONESIAN

THE AUSTRONESIAN LANGUAGES A N D PROTO-AUSTRONESIAN

ISIDORE DYEN

The Malayopolynesian or Austronesian family of languages has a membership which forms a significant portion of all of the languages in the world.1 The number of languages is probably between 300 and 500 and these form about ten per cent of the languages of the world. The great number of languages in this family is no doubt due to its being distributed mainly over islands: Madagascar, Indonesia, the Philippines, Formosa, Micronesia, New Guinea, Melanesia, New Zealand, and Polynesia. It is represented on the continent of Asia, but only by Malay (on the Malay Peninsula) and by the Chamic languages of Vietnam and Cambodia. It is the only well-established language family whose membership has an extensive and predominantly insular distribution. In its island area, non-Austronesian languages are found on Formosa (Chinese), Halmahera, Timor, Alor, New Guinea, New Britain, and some of the Solomon Islands. The similarity between distant members of this family was probably first observed by Hadrian Reland (1706) in linguistic materials from Madagascar, the East Indies, and Cocos Island. The definite establishment of the Malayopolynesian family was due to Lorenzo Hervas y Panduro in his Catalogo delle lingue (1784) although the name appears first in Wilhelm von Humboldt Über die Kawi-Sprache auf der Insel Java (1836-39). The name Austronesian was given the family by W. Schmidt in his article "Die sprachlichen Verhältnisse Ozeaniens ..." (1899). The two names Malayopolynesian and Austronesian are now in current use. THE AUSTRONESIAN HOMELAND

Like Hervas other scholars interested in this family of languages such as William Marsden and Wilhelm von Humboldt tended to regard the family as the remains of a language which extended over the island area. This interpretation, if taken literally, implies that in antiquity there could have been a language extending over thousands of miles of islands in a chain of mutually intelligible dialects. The factors in the past favoring the spread of a language must have acted to produce slow rather than 1 This paper represents partial results obtained with the support of the National Science Foundation by Grant No. GS-1468, and the Concilium on International and Area Studies of Yale University.

6

ISIDORE DYEN

rapid advances partly because it was normally difficult to move far rapidly and survive, and partly because human beings by and large tend not to migrate at all except when it is necessary. It is not likely that the forces for migration could have acted at so early a time over all of so many islands successively to produce so rapid a migration that the integrity of the language would have been preserved. Crawfurd suggested that the common words of the Malayopolynesian languages all derived from two languages, Malay and Javanese, and were spread by conquest and commercial intercourse over the area conceived to be inhabited by distinct indigenous tribes (1852: cclxxxiii). He called the two languages Malayan, but only for convenience, for he claimed that there was no basis for saying they are derived from the same source. His discussion appears to confuse unity of language with common origin (cf. vi), for he finds Welsh and Irish to show the same relation to each other as Malay and Javanese. Even if one disregarded Crawfurd's view of the relationship of the two languages to each other, his explanation of the spread of common words would require the implausible stipulation of a military or commercial power with an incredible facility of movement at a time preceding the Indian penetration of the East Indian Archipelago, for no words of Indie origin are represented in the areas east of New Guinea. A hypothesis regarding the homeland of a language family depends for its persuasiveness on the data it has been constructed to explain. There seem to be essentially two independent types of data which concern the homeland hypothesis, both of which depend on the subclassification of the languages for their interpretation. One consists of the lexical reconstructions whose meanings limit the possible homelands. The other consists of the geographical distribution of the subgrouped daughter languages. There are examples of less sophisticated hypotheses even among those proposed recently. C.E. Fox (1947:59) seems ready to consider one in saying that 'the original speakers of Austronesian were the far off ancestors of the present Polynesians.... From where they came we do not know. It is possible that MacMillan Brown may prove after all to be right and that the original home was in Micronesia; that it subsided, or was overwhelmed by the sea; and that it was this which drove them west, south, and then finally east.' Of course the ancestors of the present Polynesians spoke ProtoAustronesian in the same sense that the ancestors of other Austronesian groups spoke Proto-Austronesian. But the selection of Micronesia as a homeland appears to be proposed casually. Its appeal is limited to the fact that it could be looked on as a center of dispersal and, if it could be imagined as subsiding, could provide a motive for migration. It is evident however that such a hypothesis is of little use or concern. Perhaps the most influential article on the subject of the Austronesian homeland was H. A. Kern's "Taalkundige gegevens ter bepaling van het stamland der MaleischPolynesische Volken" (1889). He argued that linguistic evidence points to the presence of words for the following in Proto-Austronesian: sugar cane, coconut, banana, bamboo (in a number of varieties), reed, rice (unhusked, husked, dry-field), cucumber, pandanus, yam, stinging nettle, taro, vegetable fish poison, shark, octopus, lobster/

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

7

shrimp, ray, sea-turtle, boat, mosquito, fly, house, nit, moth/mite, spider, rat, dog, pig, chicken, heron, crocodile, carabao, dark-haired langur, ape, iron. He infers that the area of origin must be either in Indonesia or on the east coast of Farther India, with the Tropic of Cancer as the northern limit and Java as the southern limit. He concludes that the Indo-China coast is a better hypothesis because (a) it would explain the common appearance in Austronesian languages of the word for 'coast' as one fixed direction and the use of the word for the opposite direction that also means 'interior' or 'highland'; (b) this widespread Austronesian practice could not be associated with a large island (like Borneo) because we could not then explain why such an area, 'hardly to be described as unfruitful', has remained so lightly populated, and why the original people would not have cultivated the island before having set out in such large numbers to take over other areas; (c) to explain the drive to migrate, there is needed an external pressure which can be simply supplied by the hypothesis that the ProtoAustronesians were gradually pressed back and lost ground in a struggle against more powerful opponents, as might have happened on the Indo-China coast; and (d) there are present many Proto-Austronesian words in Cambodian, Vietnamese, and Thai. It is somewhat surprising that Kern failed to consider that his argument that the act of migration could only be explained if the Austronesians were placed where pressure could be exerted from the hinterland, makes it difficult for his hypothesis to explain the migration farther eastward from the islands off the Indo-China coast once the people had reached them. Lexical reconstructions form the basis of Kern's point of departure. The assignment of etyma to the earliest proto-language depends on the appearance of the evidence in at least two immediate subgroups of that proto-language (in our case, ProtoAustronesian), for otherwise there is no evidence that the reconstruction is to be attributed to that proto-language. This consideration can produce noteworthy changes in hypotheses that have not taken it into account. For example, the words associated with rice (dry-field, unhusked, husked) which Kern attributes to the proto-language lack cognates in eastern Austronesian languages. Similarly the attribution of an origin in 'ancient prehistoric times' to the source of Old Javanese wrai 'ape' depends not only on the fact that there are cognates, but on the chances that the most remotely related cognate, said to be Ambonese rua 'ape', belongs to a language in a different immediate subgroup of the earliest proto-language. Kern regards as the crucial argument for assigning the 'rice' terms to the proto-language the fact that it would otherwise be difficult to explain how so many languages had chosen the same names. Kern is no doubt right in saying that borrowing is not an available explanation in view of the systematic correspondences exhibited by the words concerned. He has failed, however, to consider the possibility that these words might have originated in a proto-language of later date than Proto-Austronesian, say, that of the western languages that arose after the dissolution of Proto-Austronesian. This counter-hypothesis faces the difficulty that words which at this time seem to be plausible cognates appear in languages of eastern Indonesia which are remotely related to

8

ISIDORE DYEN

those of western Indonesia. If these words are not borrowed and thus prove to be cognate, and the relevant languages of eastern Indonesia (generally those assigned by Dyen to the Moluccan Linkage with the exception of Nufors, a member of the Geelvink Hesion, see below) are found to involve different immediate subgroups of ProtoAustronesian, then this counter-hypothesis, despite its strong intuitive appeal, would fail. If, however, one adopts the hypothesis that the immediate descendants of ProtoAustronesian were Proto-Hesperonesian (or Proto-Indonesian, i.e. the putative ancestor of western Austronesian) and Proto-Oceanic (i.e. the putative ancestor of eastern Austronesian) the absence of a cognate in the east would eliminate the possibility of attributing rice-culture to the Proto-Austronesians. However in a tripartite subgrouping such as suggested by Haudricourt (1965:315) in which there was added to the other two a third immediate descendant, Proto-Formosan (i.e. the putative ancestor of the Austronesian languages of Formosa), the words for 'rice' would be restored to the lexicon of Proto-Austronesian by virtue of their appearing in Formosan languages, provided only that the Formosan cognates did not prove to be borrowings. Furthermore, for a complete lexical reconstruction — that is, the assignment of the meaning to the reconstruction — not only must cognates appear in at least two immediate subgroups of the ultimate proto-language, but the same meaning must appear for cognates in at least two immediate subgroups of the ultimate proto-language. This important prerequisite, which can only be regarded as a natural, conservative safeguard, can likewise produce noteworthy changes in hypotheses when it is taken into consideration. For example, Kern bases his conclusion (1889:119) that iron was already known in the Austronesian homeland on the hypothesis that Fiji vesi 'name of a spear', vesivesia 'hard' are cognate with Malay besi, Javanese wesi, Mongondow watoi and words in other languages of western Indonesia for 'iron' as well as Bisayan wasay, Ibanag watay 'axe'. Timor and Roti besi 'iron' are most simply interpreted as loanwords from Malay. Yapese wasai 'iron' (Elbert) is also to be considered here, but is likely to be a loanword from a Philippine language (such as perhaps Bisayan). Although the inclusion of the Fiji word as cognate has some appeal and is adopted by Dempwolff, there is no other evidence from eastern languages that bears on an early knowledge of iron. Kern accepts the direct consequence of his hypothesis in concluding that the Austronesians lost the art of working iron upon finding themselves in an area where the metal was not known (119). However it appears clear now that there is no fundamental evidence for the early knowledge of iron among Austronesians. This ignorance of iron in the east could just as well have been due to an original ignorance of it in the homeland as to having lost the knowledge in an unfavorable environment. Furthermore the meaning of Fiji vesi does not compel us to regard it as cognate with the other Austronesian words. Capell's gloss (1943:194) is 'a very hard tree, Afzelia bijuga, Leguminosae...; the same also of spear made from this wood. Vesivesia, hard, of the heart of a tree...' The gloss suggests that the meaning 'kind of spear' which

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

9

appears in Dempwolff and Kern need be in no way connected with 'iron', but rather is dependent on the use of the wood of that particular tree to make the spear. From this point of view the meaning of Fiji vesi does not particularly favor its treatment as cognate to the words meaning 'iron'. It is thus reasonable to entertain the hypothesis that the words for 'iron' in the western languages represent the dispersal of a word for this material that accompanied the dispersal of the material itself. It is interesting that there is some evidence that the dispersal took place at a time early enough for many of the words for 'iron' in the various languages to look like cognates, perhaps at the time of a common proto-language which postdated Proto-Austronesian. The words for 'axe', though at first glance conceivably also cognate, probably have a different etymon in view of Kankanay wa:say 'axe' which, along with SeBs. wa:say 'axe', contraindicates an initial *b. The assignment of 'cucumber' to Proto-Austronesian is unsupported at present because no eastern cognates of Malay timon 'cucumber' with the same meaning have been cited: cf. Fiji timo 'melon', Futuna tsimo 'watermelon'. It should be noted that Wilkinson (1932:589) mentions that Malay timon 'watermelon' occurs in certain localities; if not secondary, the latter would be evidence that the Proto-Austronesian word meant 'melon'. The possibility that borrowing is somehow involved in the eastern languages also should not be ignored; however, even if borrowing is not the explanation, the particular plant concerned cannot be securely identified. Similarly, Kern's assignment of 'dark-haired langur' to Proto-Austronesian is unsupported in the absence of an eastern cognate. The acceptance of Kern's assignment of 'carabao' to PAN depends on Fiji karavau 'cow', which not only does not offer the necessary meaning, but is suspect of being a loanword because of its form and because the animal to which it refers is newly introduced. There is a noticeable number of words connected with the sea (and water) among the words assignable to Proto-Austronesian, and these are evidence of a sea-associated homeland (for abbreviations of names of languages, see List at end of this article, p. 50): 'boat': Tobaparau, Mai. pera(h)u, Jav. perau, Sam. folau. 'giant clam': Toba hima, Mai. kima, Jav. kimo, Mer. hima, Saa ime. 'crocodile': Tag. buwa:ya, (dial.) buqa:ya, Mai. bu(h)aya, Jav. boyd, Mer. vuai, Saa huasa. 'eel': Mer. tuna 'large eel', Fiji nduna, Tonga, Fu., Sam. tuna 'eel'. 'fish-poison': Tag., Toba, Mai. tuba, Jav. tubo, Ngaju tuwe, Fiji tuva, nduva, Saa uhe. 'fish trap': Tag. bu:bo, Toba, Mai. bubu, Jav. wuwu, Ngaju bowo, Mer. vuvu, Fiji vuvu. 'octopus': SeBs. kugi:ta, Jav. keritd, Mer. hurita, Fiji kuita. 'oyster': Mai. tiram, Jav. tirem, Fiji ndio, Saa iilo, Tonga tsio, Sam. tio. 'paddle': Ngaju besei, Mer. fi-vei, Fiji i-voce, Saa hote, Tonga fohe, Fu., Sam .foe.

10

ISIDORE DYEN

'ray': Tag. pa:gi, Mai. pari, Jav. pe-pe, Ngaju pahi, Mer. fai, Fiji vai, Saa hali, Tonga, Fu., Sam. fai. 'shark': Mer. a-kiu, Fiji ygio, probably related to Toba iu, Mai. hiu. 'shrimp': SeBs. qu:laij, Toba udarj, Mai. huday, Jav. uraij, Mer. urana, Fiji ura, Ulawa ura; perhaps also for 'lobster' and other crustaceans, cf. Tag. qulay, Tonga quo 'lobster', Sam. ula 'crustacean'. 'Triton (shell)': Tag. tambu:liq, Fiji ndavui, Saa ahuri. 'sea-turtle': Toba ponu, Mai., Jav. penu, Fiji vonu, Saa honu, Tonga, Fu.fonu. Similarly there is a number of words that can be reconstructed as occurring in ProtoAustronesian that mean particular tropical plants. They are: 'alocasia': Mai. birah, bira, Fiji via. 'bamboo': Mai. (h)aor, Saa du. 'banana': Mer. akundru afuntsi 'a kind of large banana', untsi 'a kind of banana', Fiji vundi, Saa hutsi, Tonga, Fu. futsi, Sam. futi. 'barringtonia': Mai. buton, Fiji vutu, Saa huu, Tonga, Fu., Sam. futu. 'casuarina equisetifolia': SeBs. qagu:ho, Fiji cau, Saa salu. 'coconut': Tag. niyog, Toba niur, Mai. nior, nior, Fiji, Saa, Tonga, Fu., Sam. niu. 'erythrina': Tag., Toba dapdap, Mai. dadap, Jav. dadap, Fiji, Saa rara. 'ginger': SeBs. luyqa, Mai. halia, Saa lie. 'hibiscus': Mai. baru, Jav. waru, Mer. varu, baru, Fiji vau, Tonga, Fu., Sam. fau. 'mangrove': Mai. terjar, Fiji ndorjo, Saa orjo, Tonga, Sam. torjo. 'pandanus': Tag., Toba, Mai. pandan, Jav. pandan, Mer. fandrana, Fiji vandra, Tonga fala,fa, Fu., Sam. fala. 'sugar cane': Tag. tubo, Toba tobu, Mai., Jav. tebu, Ngaju tewu, Fiji ndovu, Saa oohu. 'taro': Minangkabau tolas, Jav. tales, Fiji ndalo 'calladium', Tonga, Fu., Sam. talo 'arum'. 'terminalia': SeBs. tali:say, Saa alite, Fu., Sam. talie. 'yam': Tag. qubi, Mai. ubi, Jav. uwi, Mer. uvi, Fiji uvi, Saa uhi, Tonga, Fu. qufi, Sam. ufi. The large number of Proto-Austronesian words for things connected with living near the sea and for particular tropical plants strongly suggests that the homeland was in the tropics and was either insular or, if continental, coastal. It is interesting that the strong indication of a tropical homeland militates somewhat against the hypothesis of a Formosan homeland. Formosa became an island after the end of the last glacial period about 8,000 B.C. Thus all tropical plants now found on Formosa must have been introduced after this time. Casuarina (*[]aRuha[]) was almost certainly introduced in historic times. The means by which many tropical plants (because of their own character or the size of their seeds) could reach Formosa are limited to ocean currents and man. Erythrina (*DapDap), pandanus (*panDan), and terminalia (*YtT^\ali[cs]ay) occur only coastally and clearly arrived by ocean current. On the other hand ginger, banana, taro, melon, sugar cane, and yam could not

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

11

reach Formosa by ocean current and therefore must have been brought by man. Some plants like the coconut and perhaps sugarcane were probably introduced by man about 2,000 B.C., since their pollen count increases remarkably around that period (Tsukada 1947:49 f. and personal communication). Since Proto-Austronesian can hardly have persisted as a unified language after 3,000 B.C., if lexicostatistical evidence in this respect is reliable, Formosa could hardly have been the homeland of a people who knew these plants. No doubt basing himself on some such homeland hypothesis as Kern's, Georg Friederici in his "Untersuchungen über eine melanesische Wanderstrasse" (1913) posited two migrations into Melanesia: (1) from the Moluccas to coastal east New Guinea, New Britain, South Solomons, and New Hebrides; (2) from the South Philippines and Northeast Celebes to the Admiralties, New Ireland, North Solomons, merging there with the other migration. It should be observed here that the assumption that languages mix or merge that is implicit in Friederici's migrational hypothesis is one that is not accepted by many comparatists in addition to the writer. The two migrations until their merger are separated by a line which bounds the appearance of the 'preposed genitive'. Friederici's construction of this line is an extensive modification of one originated by Brandes. In his Bijdrage tot de vergelijkende klankleer der westersche afdeeling van de MaleischPolynesische taalfamilie (1884: following p. 184), J.L.A. Brandes offers the thesis that the Malayopolynesian languages of the Indonesian archipelago can be subdivided into an eastern and a western branch and that 'the boundary between them is a line drawn between Sawu and Rotti, Flores and the Solor Islands, east of Buton, west of the Sula Islands, east of Minahasa, the Sangir and Talaut Islands, and the Philippines'. East of this line is found a different 'genitive-construction' from that found west of the line. He cites as an example Kisar manu-keer 'bird egg' (attribute plus head) as compared with Malay (Mai.) telor ayam '(egg chicken) chicken egg' (head plus attribute). Friederici has in effect accepted the Brandes line and extended it in a southeasterly direction from just north of Morotai, so that it passes between the Admiralties and the small islands off New Guinea and roughly over the eastern end of New Britain, proceeding southward between the Solomons and the New Hebrides on the one side and the small island groups near the southwestern tip of New Guinea on the other. Jonker, in his "Kan men bij de talen van den Indischen Archipel eene westelijke en eene oostelijke afdeeling onderscheiden?" (1914), raises a number of objections to the line. One is that though the 'genitive-construction' isogloss had come to be regarded as uniting the eastern division, not the western division, it should not be regarded as necessarily applying to the languages of Melanesia and Micronesia, that also are east of the line. Another is that Banggai, which is actually west of the line, belongs with the languages east of it. Finally he raises a number of well-justified objections to using such a criterion for subgrouping when the feature utilized could have appeared independently in different areas. In a reversal of the usual manner of regarding languages Sidney Ray, in A compara-

12

ISIDORE DYEN

five study of the Melanesian island languages (1926), suggested that in the Melanesia area at any rate the Papuan (i.e. non-Austronesian) element in the languages is 'aboriginal, the Melanesian immigrant' and that the immigrants were Indonesian (1926:25). The statement as formulated seems to confuse language and race, for there is no comparative linguistic evidence of a Papuan substratum cited. Though the Melanesian languages may prove to have been immigrant, the hypothesis that they were is not necessary to explain the presence of Austronesian languages in Melanesia since they might conceivably have always been there. Arthur Capell studied the vocabularies of a large number of languages in The linguistic position of southeastern Papua (1943). Capell, who followed Ray, believed that his study substantiates Friederici's posited migration into Melanesia from the Philippines. However he rejects the migration posited as originating from the Moluccas because 'the reversed genitive ... has been shown to be a common heritage of the Papuan languages, and only found in eastern IN as a relic of pre-IN days in that region also' (Capell 1943:266). Instead, using arguments based on vocabulary, he finds two other migrations into southeastern Papua. He arranges the three migrations chronologically as follows: (1) from Borneo, directly or indirectly (this being the earliest migration except for the possibility of 'a still earlier movement represented by the Big Nambas and their related peoples in Malekula, the people of Eromanga, some of the Loyalty Is. and New Caledonia which often preserve earlier forms than any in Oceania' [270] after the colonization of Madagascar [?400 A.D.; 276]); (2) from Central Celebes and the Philippines; (3) from the Java-Sumatra area, via the Philippines; perhaps about the thirteenth century A.D. (276). It is difficult to accept such late dates for significant migrations from the indicated areas in the absence of pairs of languages closely related to each other and distributed over the separated areas; arguments based on vocabulary alone are not sufficient here unless it is also claimed that the associated vocabulary establishes the prerequisite closer relationship. Capell's conclusions were not reached inductively from his vocabulary study alone. Instead he assumed that if, for example, a word found in a language of southeast Papua had a cognate only in Borneo, it constituted evidence of a migration from Borneo. This type of hypothesis takes common origin into account, but obviously has assumed the point of origin of the word since no evidence is adduced that necessarily makes Borneo the place of the common ancestor that used the proto-word.

SUBGROUPING

The other evidence bearing on the homeland is the distribution of the subgrouped daughter languages. It is evident that the simplest hypothesis of a homeland in terms of the required migrations is that area, if any, in which are found the subgroups of the highest order of diversity, provided of course that there is no reason to believe that the representative of any subgroup has wandered back into that area. Under

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

13

the widely accepted Eastern-Western dichotomy as exemplified in F. N. Finck's Die Sprachstamme des Erdkreises (1909:47) and E. Kieckers' Die Sprachstamme der Erde (1931:117), this would imply as the area of highest probability an area approximating the seam between the two divisions, say eastern Indonesia and western New Guinea. Under Haudricourt's tripartite subgrouping (see below) there would be a temptation to include areas farther north among the possible homelands. There are however few areas to the north that can reasonably be considered except perhaps the MindanaoCelebes complex and Formosa itself. Dyen's investigation, reported in A lexicostatistical classification of the Austronesian languages (1965b), indicated that when the Austronesian languages were compared by the Swadesh proceduie there were three areas of the highest order of diversity, namely the (1) New Guinea-Melanesian area, (2) Formosa, and (3) Sumatra and the islands along its western flank. These three areas differed in the number of languages and language groups which seemed to show only very low percentages with others. The New Guinea-Melanesia area contains by fai the largest number of languages and language groups of this type, whereas the numbers in Formosa and in the Sumatra area respectively are relatively very small. The large number in the New GuineaMelanesia area seemed least likely to prove to be due to error, so that lexicostatistical evidence seemed to point strongly to this area as that of the highest order of diversity and thus the Austronesian homeland. Nevertheless the strong conviction of scholars that the languages of Melanesia and eastern New Guinea are part of a single subgroup with all of the languages in the east should be taken into consideration, particularly since this conviction is associated with some qualitative evidence. This evidence concerns the merger of labial voiced and voiceless stops (*p, *b), which Milke wishes to extend to 'mergers of *g and *k, of *d and *D, of *n and *n, and of the whole series of palatals' (1965: 331). It seems clear, however, that Bugotu (Ysabel, Central Solomons) retains the distinction *n=£*fi in namu 'mosquito' (*namuk), manu 'bird' (*manuk) (Ray 1926: 523). Grace (1964:367f.), after considering both the lexicostatistical and the qualitative evidence, favors the traditional hypothesis in concluding that Austronesian first differentiated 'into at least two subgroups, one of which was Eastern Austronesian'. However, Dyen (1965b:55f.) points out that the qualitative evidence, if it should be sufficient to substantiate a Proto-Oceanic, would serve only to move the implied homeland somewhat farther west because the area of the highest order of diversity would still remain in the neighborhood of New Guinea. The languages which merge *p and *b reach no farther west than Biak (which does not) in the New Guinea area. If one takes into account this qualitative evidence, and in effect assigns to error the large number of more eastern language groups with low critical percentages, there remain a striking number of highly diverse language groups in the area in and around western New Guinea. In this sense at least the inference that can be drawn from the lexicostatistical analysis as modified by qualitative considerations yields essentially the same conclusion as might be drawn from the Eastern-Western dichotomy. How-

14

ISIDORE DYEN

ever, further study is necessary to determine the strength of the qualitative arguments adduced here for subgrouping. It is not impossible that other explanations for the evidence may be found even if it is substantiated. Dyen's lexicostatistical study seemed to indicate that certain Formosan languages, there called the East Formosan Hesion and now regarded as members of the Paiwanic subgroup, seemed to be somewhat more closely related to other Austronesian languages than were the members of the Atayalic subgroup, Atayal (At.) and Seediq (Se.). Other evidence however seems to favor regarding the Austronesian languages of Formosa as forming a single subgroup. This hypothesis is discussed further in the paper entitled 'The position of the Malayopolynesian languages of Formosa' (Dyen 1964). The consequences of an hypothesis of an extra-Formosan mesolanguage (i.e. the proto-language of the putative extra-Formosan which we can now refer to as ProtoExtra-Formosan) would be revolutionary. It would imply that Proto-Extra-Formosan (PEF) was a tropical subgroup of a family that ultimately originated in Formosa and that many of the reconstructions currently assigned to Proto-Austronesian as the proto-language of highest order would then have to be reassigned to PEF wherever a Formosan cognate was lacking. These considerations emphasize how much remains to be done to determine the relation between the languages of Formosa and the other Austronesian languages. In his 'Formosan evidence for some new Proto-Austronesian phonemes' (1965a) Dyen points out that a number of distinctions (*t/T and *n/N among others, see below) assignable to Proto-Austronesian from the evidence of Formosan languages appear to be merged outside of Formosa in the same way. On qualitative grounds these would suggest that all of the extra-Formosan languages should be assigned to the same subgroup, presumably coordinate with the Formosan subgroups which give evidence of these distinctions or more closely related to one of them — e.g. Paiwanic — than to the others. This subgrouping could then be regarded as evidence for a Formosan homeland, for Formosa would then be the area of the highest order of diversity, disfavoring a subgroup in which all of the Formosan languages are united and more closely related to the Philippine languages than to others, and offering serious obstacles to the generally accepted explanation of the evidence presented above from lexical reconstruction in its relation to Formosan climatic changes. It is worth noting that the extra-Formosan mergers seem to constitute as strong an argument for an extra-Formosan unity as the mergers in eastern Austronesian languages are for an eastern Austronesian unity. It is not impossible, of course, that the extra-Formosan mergers first occurred dialectally in Proto-Austronesian rather than in a daughter language or that the mergers represent independent though similar changes in different daughter languages. But such hypotheses would also be available to explain the eastern Austronesian mergers without constructing Proto-Oceanic. In view of the evidence above it is not surprising that a tripartite immediate subgrouping of Austronesian has been suggested. Haudricourt (1965:315) expresses the view that in the present state of our knowledge

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

15

it seems premature to reconstruct common Austronesian and that reconstruction must stop short within the following three subdivisions: (1) Western Austronesian, which stretches from Madagascar in the South West and from the Mergui Archipelago (Burma-Thailand) in the North West to the Palau Islands, Guam and Botel-Tobago in the North East. In that area one can reconstruct 'common' Indonesian. (2) Northern Austronesian, made up of the Highland languages of Taiwan (Formosa): ... (3) Eastern or Oceanic Austronesian, which embraces all the Melanesian and Polynesian languages, together with most of the languages of Micronesia. In that area one can reconstruct 'common' Oceanic (Urmelanesisch, Ozeanisch). It appears to be implicit that Haudricourt regards these three groupings as the coordinate highest subdivisions of Proto-Austronesian. Unfortunately he does not present the reasoning which leads him to this conclusion.

F U R T H E R SUBCLASSIFICATION

The further subclassification of the Austronesian languages is still uncertain in many points. There seems to be general agreement that the group formed by the languages of the Philippines is coordinate with the group formed by those of West Indonesia, and the two together form a subgroup which I shall call Hesperonesian here (as the equivalent of the more common West Indonesian). Malagasy, originally regarded as a coordinate immediate member of Hesperonesian, is now widely accepted as being more closely related to Maanyan than to the others, largely through the work of Dahl (1951). The languages of eastern Indonesia likewise form a subgroup for which I suggest the name Moluccan, though its membership may extend into the lesser Sundas. Dyen (1956b: 39) includes in it Sikka of eastern Flores, a language which is clearly west of the Brandes line. Many scholars believe that the remaining languages with the exception of the Formosan languages, Chamorro, and Palau — which they assigned to Hesperonesian — constitute a subgroup often called, as here, Oceanic. There is now no general agreement on whether Eastern Indonesian and Hesperonesian form a single subgroup, nor just where the Formosan languages fit. It is furthermore hard to justify the placement of Palau among the Philippine languages, as is commonly done. Dyen's lexicostatistical classification does not support the Oceanic hypothesis. Instead his classification presents a large number of languages and language groups of the highest order of lexicostatistical divergence concentrated in the east. This suggests at least that the putative Oceanic is probably much more internally divergent than has hitherto been assumed. There is also the side indication that some parts of the area occupied by the so-called Oceanic languages have been settled for longer periods than has been imagined. However if one excludes the area covered by the Oceanic hypothesis, there remains

16

ISIDORE DYEN

a sizable number of highly divergent languages that fall between Hesperonesian and the putative Oceanic. For example, it has been claimed that Moluccan falls in with Melanesian as against Hesperonesian. Finally, the Austronesian languages in and around the western end of New Guinea have not yet been fully reported. It is obvious at this point that qualitative studies and further lexicostatistical investigation are needed to determine whether the hypothesis of a Proto-Oceanic is needed to explain those phenomena of eastern Austronesian which are offered as common innovations. There is a clearly marked Polynesian subgroup. In his well-known work, The position of the Polynesian languages within the Austronesian (Malayo-Polynesian) language family (1959), George Grace attempted to show that its nearest relatives were Rotuman and Fiji. Dempwolff had subgrouped Fiji and Saa together and apart from Polynesian, but at the same time had concluded that his Melanesian, represented only by Fiji and Saa, and Polynesian constituted a single subgroup against his Indonesian (1943-38:2.193), represented by Tagalog, Javanese, Malay, Ngaju, and Merina and thus not essentially different from Dyen's Hesperonesian. Grace agreed with Dempwolff in assigning Fiji and Polynesian to the same subgroup, but reached the conclusion that Fiji was closer to Polynesian and Rotuman than to Saa. Dyen's lexicostatistical classification did not include Saa in its purview, but showed that lexicostatistically, at any rate, there were Melanesian languages (Efate among others) that were as close to Polynesian and Fiji as they were to each other. These languages were grouped as the Heonesian Linkage. The suggestion seems to follow that Fiji, Rotuman, and Polynesian left approximately the same Melanesian area at about the same time and had already begun to differentiate at the time of departure or began to do so immediately thereafter. Whether this hypothesis is substantiable remains to be seen. Perhaps the most interesting construction of a subgroup was that attempted by Dahl in Malgache et Maanjan (1951) on the basis of a suggestion by Aichele (Dahl 1938:230ff.). Dahl's work was reviewed by Dyen (1953c). On the basis of a thorough phonological, morphological, and syntactic comparison, Dahl concludes that the nearest relative of Malagasy, the Austronesian language of Madagascar, is Maanyan, a language of Southeast Borneo. Dahl suggests that Madagascar was colonized by merchants coming from the Hinduised kingdom existing in Borneo in the fourth century A.D. This hypothesis is preferable to the older hypothesis of van der Tuuk (1865a: 427) that the Malagasy were most closely related to the Toba Bataks and came 'from the west coast of Sumatra after having mixed with a tribe resembling the Nias people ...'. It should be added, however, that this great scholar did not have available the Maanyan data. W.H. Goodenough in his "Migrations implied by relationships of New Britain dialects to Central Pacific languages" (1961b) concluded that the Nakanai dialects of the north coast of Central New Britain represent a migration from the central Pacific. Grace (1959) had set up a subgroup on the basis of shared innovations called Eastern

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

17

Austronesian which includes Fiji, Saa, Rotuman, and Polynesian. Renaming the group formed by these four as Central Oceanic, Goodenough brings evidence pointing toward Nakanai's being a fifth member. He then concludes that Nakanai must have reached New Britain from the area of the other four, namely the central Pacific.

THE EXTERNAL RELATIONSHIP OF AUSTRONESIAN

A number of different external relationships have been proposed for Proto-Austronesian. Perhaps the most widely known is that of W. Schmidt who in his Die MonKhmer-Völker (1906) proposed those languages as a link between the Austronesian languages and those of Central Asia. He constructs the Austroasiatic family from the Mon-Khmer and other languages of southeast Asia and eastern India and asserts that this family is related to Austronesian on the basis of an extensive comparison of phonological and grammatical features and a long list of proposed cognates. This view had been originally proposed by A. H. Keane in his "On the relations of the Indo-Chinese and inter-Oceanic races and languages" (1880), whose evidence Schmidt finds insufficient, and also touched on by G. von der Gabelentz who attempted to connect Nicobar with the Malayopolynesian family (1885) with very few comparisons that Schmidt found attractive. Despite his greater care, there is little in Schmidt's collection of evidence that is not reasonably attributed to chance or borrowing. Similarly negligible, and also less pretentious in content, is the connection with Japanese proposed by Van Hinloopen Labberton in his "Preliminary results of researches into the original relationship between the Nipponese and the MalayoPolynesian languages" (1924). K. Wulff, in his "Über das Verhältnis des MalaioPolynesischen zum Indochinesischen" (1942), despite the title of his work, argues for a relationship between his Proto-Thai-Chinese (argued for in Wulff 1934) and Proto-Austronesian chiefly on the basis of vocabulary similarities. He utilizes Brandstetter's hypothesis that Austronesian polysyllabic — mainly dissyllabic — roots originate from monosyllabic elements, as the result of derivation by composition or duplication. P. Benedict in "Thai, Kadai, and Indonesian" (1942) proposed a coordinate relationship between Thai, Austronesian (under the alias Indonesian), and four languages (Laqua, Li, Kelao, Lati) classified under the constructed name Kadai. The Kadai languages are presented as transitional between the Thai languages and Austronesian, whereas in fact it can easily be demonstrated that the so-called Kadai languages are more closely related to Thai than to Austronesian. Thus the difference between his hypothesis and Wulff's is the Chinese element in the latter's. Benedict has more recently continued his thesis in "Austro-Thai" (1966), "Austro-Thai studies: material culture and kinship terms" (1967a) and "Austro-Thai studies: 3. Austro-Thai and Chinese" (1967b). In the last Benedict suggests that there was lending and that the lending direction may have been from Austro-Thai to Chinese.

18

ISIDORE DYEN

Of the two, Benedict's argument is clearly the more persuasive and has achieved wider acceptance. For example it is followed with approval by J.Greenberg (1957: 40). Wulff dilutes his argument by following Brandstetter's monosyllabic hypothesis for Austronesian. On the other hand Benedict has achieved an equivalent flexibility by permitting the Thai or Kadai monosyllable to be compared to either one of the two Austronesian syllables. Some of his comparisons are striking indeed; it is impressive to find Proto-Thai tai 'die' beside PAN (Dempw.) matay 'die' and PT ta 'eye' beside PAN mata 'eye'; PT vai 'fire' beside PAN apuy and PT van 'tooth' beside PAN ipen,YT näm 'water', Li nöm ~ nam beside PAN danum 'water' and S.Li nom '6' beside PAN enem. Although these are impressive pairs of comparisons, it should not be forgotten that the chances of finding such pairs were increased by ignoring one syllable of PAN. One can increase the number of putative cognates by allowing the additional syllable to be optionally not lost as in the following: PT kin 'eat', PAN kaPen and PT ku 'I (pejorative)' beside PAN aku 'I'; PT blüän 'moon' beside PAN bulan and PT blok 'flower' beside PAN burak 'flower'. Despite the rather remarkable agreements here, it should be recognized that at least part of the evidence is phonetic similarity and that the number of words covered by correspondences in the strict sense is very small. Our reconstructive hypotheses are changing in both families and it is not inconceivable that changes in one or two words might alter the impressiveness of the comparisons significantly. For example, the effect of Benedict's (1967a: 240 f.) new reconstructions *mapra 'eye', *maprai 'die' (instead of *mata, *matai as before) for Austro-Thai, necessitated by new comparisons on the Thai-Kadai side (though these languages lack the initial syllable ma- which is apparently attached only to account for the Austronesian words), can only be regarded as weakening or nullifying the rather powerful appeal this set of matchings had when compared with PAN mata 'eye', matey 'die'. Only a minor reinvigoration can be obtained by associating the sequence *pr directly with the PAN mat- sequence. In his 'Die Verwandtschaft des Indonesischen mit dem Indo-germanischen' (1937), Brandstetter collects what he regards as the common property of Proto-Indonesian (PIN) and Proto-Indoeuropean (PIE). He believes that they establish the relationship between the Indonesian language family and the Indo-European language family. Basic to his etymological procedure are the terms word-kernel (WK), base, primary formant, and secondary formant. For example Ngaju (Ngj.) has a word sir/kap 'grasp' in which the WK is kap and siy- is a primary formant. The combination of the two is a base. Affixes by which other forms are derived from a base are secondary formants. In this attempt he is continuing the effort of Franz Bopp in his "Über die Verwandtschaft der malayisch-polynesischen Sprachen mit den indisch-europäischen" (1840) which Brandstetter regards as failing because of the scarcity and unreliability of the available material. As an example of his own method he cites the WK lap 'shine, light, burn' to be found in Mai. gilap 'polish', Karo arlap 'shine', Pampangan aklap 'shine through', Old Javanese (OJav.) gurilap 'flicker', Madurese kolap 'kindle', Bali

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

19

hulap 'dazzled', etc. There is a PIE WK lap (Walde 1930-32, 11:383) which appears in Lettish lapa 'pine torch'. Brandstetter concludes that the WK lap [lap] is a common proto-possession of PIE and PIN and is a piece of evidence of the relationship between the two. After presenting the phonemic inventories of the two proto-languages, Brandstetter gives some morphological comparisons. Among these he identifies the p that appears in Sanskrit sthapayami 'I put' (cf. stha 'stand') as a causative formant with the causative prefix pa- of PIN. This is fairly typical of his method in this area. He deals however not only with secondary formants, but also with numerals, personal pronouns, demonstratives. His demonstration is not convincing. Quite independently of Brandstetter's attempt but rather in response to the challenge offered by the Thai-Austronesian hypothesis, Dyen in his "Background 'noise' or 'evidence' in comparative linguistics: the case of the AustronesianIndoeuropean hypothesis" (1970) has collected what appears to be impressive evidence for a conceivable, though at least at first blush hardly likely, Indoeuropean-Austronesian unity. The basis of his argument is that as the number of strict demands on an etymology are weakened, the remaining strict demands will tend to produce etymologies in greater number. Thus, for example, Dyen ignored the vowels and final consonant of a Proto-Austronesian reconstruction. By so doing he was able to amass a large collection of Austronesian-Indo-European etymologies while rather strictly observing standard comparative procedure in all other respects. This collection was noticeably larger and in many respects (geography aside), more appealing than the Thai-Austronesian hypothesis. Nevertheless, despite the fact that ignoring the vowels would conform to current theories about pre-Indoeuropean, the precedure is probably to be evaluated as allowing too much freedom in the selection of cognates to be acceptable. However, it seems almost naturally to follow that if such is the case for the Austronesian-Indoeuropean hypothesis, the same can be said for the Thai-Austronesian hypothesis. RECONSTRUCTION

The reconstruction of Proto-Austronesian has advanced most in phonology and relatively little in other areas of structure, so that the following discussion is essentially restricted to Proto-Austronesian phonology. Even in phonology, problems are still multiplex and many remain unsolved because of a lack of reliable data. In reaction to this situation Haudricourt (1965), as we have seen, has suggested that it is (? tactically) better to aim at the phonological reconstruction of the proto-languages of the highest order subgroups. However the impression of backwardness in the field is a direct consequence of the large number of Austronesian languages, their wide geographical distribution, and their past relative unimportance in the increasingly complex interrelation of the politico-economic activities of western culture. No work has been done on PAN morphology and syntax except casually. The

20

ISIDORE DYEN

vast numbers of languages make such an enterprise one not to be undertaken lightly. Brandstetter, who restricted himself essentially to Hesperonesian languages, did do some work on the morphology and syntax. This appears in such of his works as Die Beziehungen des Malagasy zum Malaiischen (1893), Tagalen und Madegassen (1902), Wurzel und Wort (1910), Gemeirtindonesisch und Urindonesisch (1911a), Sprachvergleichendes Charakterbild eines indonesischen Idiomes (1911b), and Das Verbum (1912). Charles Blagden has translated three of these (see Bibliography) in his Renward Brandstetter, An introduction to Indonesian linguistics (Brandstetter 1916). The manner of citation of language material often leaves much to be desired. In any case even the field of Hesperonesian morphology and syntax has been barely touched in a systematic manner. The first attempt at a coherent reconstruction of the phonology of Proto-Austronesian was made by Otto Dempwolff in his Vergleichende Lautlehre des Austronesischen Wortschatzes (1934-38). His work was the culmination of years of study of Austronesian languages of both the eastern and western areas and of the works of his predecessors van der Tuuk, Kern, Brandstetter, and Conant. Although he was familiar with many other Austronesian languages, Dempwolff limited the languages he used in his great work to a small number that he felt were relatively representative and at the same time in each case offered a satisfactory amount of reliably described material. The languages he chose, as he subgrouped them, were: Indonesian: Tagalog (Tag.), Javanese (Jav.), Toba-Batak (i.e. Toba), Malay (Mai.), Ngaju-Dayak [i.e. Ngaju (Ngj.)], Hova [i.e. Merina (Mer.), the standard dialect of Malagasy]. Melanesian: Fiji (Fi.), Saa (with occasional citations from closely related Ulawa, the latter perhaps being simply another dialect). Polynesian: Tonga (To.), Futuna (east) (Fu.), Samoa (Sam.). However instead of dealing with all of these languages at once, he divided them into two groups. He selected Tagalog, Toba, and Javanese for the first step. From the data of these three languages he drew inductive inferences regarding Proto-Indonesian, the putative closest common ancestor of the three languages concerned. His results are set forth in the first volume. The second volume deals with the deductive application of the Proto-Indonesian reconstructions to the other languages. It is a matter of some surprise that he was as successful as he was in view of the fact that non-Indonesian languages were included. In any case his final step was to construct the dictionary of Proto-Austronesian reconstructions that constitutes the third volume. Dempwolff reconstructed the following phonological elements (1.64): Syllabics: Non-syllabics:

i a a u Bilabial: Alveolar: Retroflex:

p t t

b d d

m n

v 1 1

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

Palatal

{frontal [dorsal

Velar Laryngal

t' k' k

d' g' g

ñ

j

g

Y h

21

He posited a 'weicher Vokal-An-, Zwischen-, und Absatz' (the laryngal described 1.15 as 'der lockere Verschluss der Stimmritze') listed separately in the text, as above, but which is not listed separately in the Wörterverzeichnis, and is there cited only proposed to the vowels : (a, (e, % (u (3.9). It could however be regarded as no more than a signal of the beginning or end of a vowel or the glide from one vowel to another. It is perhaps a necessary consequence of Dempwolif's way of adapting the Lepsius phonetic system since he introduced it in the notation of every language compared. No vowels appear final since in the absence of any other non-syllabic, a word ends in c. Furthermore no vowels appear initial since if a word does not begin with another non-syllabic, it begins with Dempwolif did not however introduce ' in medial position in the compared languages. Surprisingly enough he used it to transcribe initial Malay /h/ which he specifies as lacking except in obvious loanwords (2.15), though in fact it is observable and distinctive in Malay speech. Brandstetter in Gemeinindonesisch und Urindonesisch (1911a) had preceded Dempwolif in reconstructing in the Austronesian field, but limited himself to languages found in Formosa, the Philippines, Madagascar, and the Indonesian archipelago and was thus reconstructing a proto-language mainly based on Hesperonesian languages with some languages from eastern Indonesia. He drew his material from seven large island areas : Philippines, Celebes, Borneo and Java, together with Madura and Bali, Sumatra, the Malay Peninsula with its nearby islands, and finally Madagascar. In addition he drew from three border areas: in the north, the Batan Islands and Formosa; in the east, the islands from Lombok to New Guinea, particularly the languages Bimanese, Kambera, Sawunese, Rotinese, Tettum, Masarete; and in the southwest, the row of islands west of Sumatra — Simalur, Nias, Mentawai. By comparing these languages, though not in as thoroughgoing a fashion as Dempwolif, Brandstetter reached the following phonological reconstructions (1911a:14, 45), which are essentially followed in Gabriel Ferrand's chapter 'Indonésien' in Les langues du monde (Ferrand 1924): a i k g c j t d p b y r (i) s h

u ë ù ñ n m r (2) 1

e

w

o

22

ISIDORE DYEN

Actually Brandstetter presents the table above with a single r symbol (191 la: 14), and as the sound system of Gemeinindonesisch. He regards as Gemeinindonesisch that language material found unchanged, or changed in accordance with strict sound laws, in the languages which he regards as Indonesian. However when he turns to dealing with Urindonesisch (45), he finds that the latter differs from the former by having two distinct r sounds, one of which is to be identified with the RGH-series formulated later by Dempwolff as *y and symbolized by Ferrand by p. [Ferrand adds ? to the inventory, but in his commentary shows that he regards it as a later development in the daughter languages from *h.] There are thus the following direct equivalences between the symbols used: Dempwolff a g k' d' ö 1 Y

Brandstetter e n c j n r (i)

t'

s

j

y

V

r

Ferrand e n ts 6i n r

P s

C2)

y W

W

Dempwolff makes it clear that each one of his reconstructions is not merely a formula for a correspondence, but involves a hypothesis about the phonetics. Thus, for example, he concludes that the correspondence of voiceless sibilants in Tagalog, Toba, and Javanese reflects a voiceless frontal palatal stop (1.39) because the reflexes are voiceless and alternate with a nasal (in prefixation) in all three languages and in particular with /n/ in Javanese, where alone /n/ is distinguished from /n/: Tag. su:lat (base) 'write', panu:lat 'pen'; Toba surat (base), manurat (active) 'write'; Jav. surorj (base), nuror) (active) 'push'. He believes this hypothesis is supported by the observation that nasal clusters with /s/ appear in Tagalog and Javanese, such clusters being otherwise found only with stops. There is however perhaps good reason to be cautious at this time. Hypotheses as to the phonetics of proto-phonemes depend not only on the phonetics of the corresponding phonemes but also, like all hypotheses other than those regarding the protophonemes and subgrouping, on the subgrouping of the languages. The protophoneme hypotheses, however, are dependent only on the distribution of the correspondences and the likelihood that one or another correspondence contains a matching phoneme that is due to an analogical change or is indicative of a loanword. Subgrouping and morphemic reconstruction constitute simultaneous operations which assign an etymon and a distribution over languages to every cognate set. The etyma of cognate sets with a distribution among the most independent members are assigned

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

23

to the earliest proto-language. By a similar procedure etyma are assigned to the subproto-languages. The consequence is that the determination of the phonetic nature of a protophoneme depends increasingly on the subgrouping as the phonetic variation of the correspondences increases. Since the subgrouping of the Austronesian family still is in large part indeterminate, particularly at the highest levels, it is perhaps incautious to attempt to deal with the phonetics now whenever the phonetics of the reflexes show great variation, except in the most tentative way. With this in mind it is desirable and has become customary to use a more convenient set of symbols in place of those used by Dempwolff, but without phonetic prejudice. The equivalents are as follows: Dempwolff

Equivalent

1

1

3

a u

e a u

P

P

b m V

b m w

t

t

d n

d n

1

1

t

T

d

D r s z n

1 t'

d' n j k'

y

g' k

j k

g

g

r)

D R

Y h c

c

q h

24

ISIDORE DYEN THE VOWELS

Phonetically *i was undoubtedly high front unrounded, *u high back rounded, and *a low unrounded. All three probably had variants according to environment. The fourth vowel *e was almost certainly mid-central, but not necessarily unrounded like the Javanese 'pepet'. Those languages that reflect *e distinctively from the other inherited vowels generally do so with a mid vowel. These are distributed among front unrounded [e.g. Malagasy, Ngaju, and some dialects of Ilocano (Ilk.)], central unrounded (e.g. Malay, Javanese, and other dialects of Ilocano), and back rounded (e.g. Ifugao, Polynesian, and Trukese). The conclusion that *e was mid-central is hardly escapable. But the evidence as to rounding cannot be regarded as conclusive; it is reflected distinctively both as rounded and unrounded as indicated above and in each case with a distribution among remotely related languages. Furthermore, it has merged not only with the reflexes of *i (e.g. Tagalog), but also with those of *u (e.g. Bisayan). In reconstructing the four vowels Dempwolff followed the presentation of the correspondences that was employed by Conant in his 'The pepet law in Philippine languages' (1913) for the Philippine languages. The recognition of the proto-phoneme *e first appears in Brandes (1884:19f.). He inferred the quality to be the same as that of the Javanese mid-central vowel which is called the 'pepet' and hence gave the reconstructed vowel the same name. He was followed in this by nearly all writers including Dempwolff. The formulation of the so-called 'Pepet Law' was first made by Brandstetter in Mata-Hari (1908:52) who dealt with the reflexes in Old Javanese (OJav.), Tontemboan, Buginese (Bug.), Karo, Minangkabau, Makassarese, Bisayan, Toba, Tagalog, Ngaju, Malay, and Malagasy. Conant (1913) studied the reflexes of *e in other Philippine languages as well as Tagalog and Bisayan. There can be little doubt about the need to reconstruct at least four vowels. Dyen, in "On the history of the Trukese vowels" (1949), tested the four-vowel reconstruction for its sufficiency to explain the nine-vowel inventory of Trukese and found it satisfactory. However not all problems in the reconstruction of vowels can be said to have been solved. The origin of e and o in open syllables (as distinct from i and u) in Malay as well as in other languages, and of Javanese e [e], e [e], o [o], in open syllables and of o [o] in closed syllables (where not from *a) is not yet clear. There is, however, a reasonable chance that these vowels are to be attributed to secondary developments including dialectal and interlinguistic borrowing rather than to distinctions in Proto-Austronesian. Superficially less susceptible to such an explanation are the Sundanese (Snd.) vowels written eu and e respectively, both of which correspond to what is currently reconstructed as PAN e: *peZem, Snd. peureum 'close eyes'; *peRes, Snd. peres 'squeeze'. B. Nothofer in a personal communication has presented strong arguments for regarding Javanese loanwords as the main source of instances of Snd. e. Brandstetter, in determining Gemeinindonesisch, felt there was evidence for two additional vowels *e (distinct from pepet) and *o (191 la: 10), which however he felt

25

THE AUSTRONESIAN LANGUAGES AND PROTOAUSTRONESIAN

should be 'nur zögernd' explained as 'common Indonesian'. The evidence for *o is negligible. Although the etymology cited for his *e is stronger, it does not necessarily apply to Proto-Austronesian since it has no known cognates outside of languages of Indonesia and perhaps does not even apply to his 'common Indonesian' since it lacks cognates in the Philippines. Ferrand (1924) assigns these two vowels to ProtoIndonesian and they are clearly assigned to Urindonesisch without qualification in Brandstetter's Die Verwandschaft des Indonesischen mit dem Indogermanischen (1937:9). T H E SEMIVOWELS

The semivowels *y and *w are generally like other non-syllabics. There is, however, at least one instance of interchange between syllabicity and non-syllabicity which suggests that the semivowels may not have been distinct from *i and *u respectively. This interchange occurs in the comparison of Tag. ka:hoy 'wood', with Mai. kayu, Kankanay ka:qiw, Thao kawi? 'wood'. For reconstruction there is the problem of the order of the vocoids and that of their syllabicity. Not all of the data is clear or reliable as to syllabicity. Those languages for which the evidence seems clear follow a rule that if there is a consonant after the a, then the next vocoid is syllabic and the one following that is non-syllabic. In the absence of such a consonant the order of syllabicity is reversed. The four possibilities are exemplified in the following table, the cases uncertain as to syllabicity being marked with (?): *-uy Tagalog Pazeh (?) Saisiyat (?) Ami

ka:hoy kahuy kahui kasui

*-iw

*-wi

Kankanay ka:qiw (?) Paiwan kasiu

Thao kawi? Tsou evi (?) Puyuma kaui

*-yu Malay Ilocano Kanabu Saaroa

kayu ka:yu kalu kiyuPu

Sundanese kai 'wood' is now most easily interpreted as from *kaCiw (C = consonant) since *kaCuy is not available in view of Snd. tuluy 'go forth', Tag. tuloy 'forward!' from an earlier * tuluy. Ulawa Pai 'tree', on the other hand, is assignable to either *-uy or *-iw, and not to *-wi nor to *-yu. The independent appearance of the different orders in the daughter languages suggests that the order was not uniform in the proto-language of the languages showing the variation. The distribution of the different sequences of the two high vowels in the word for 'tree' is associated with a subgrouping only to the extent that the order *ui is confined, as far as is known, to Formosa and the Philippines except perhaps for Ulawa and, again as far as is known, the entire indeterminacy is restricted to the western languages, with the exception again of Ulawa. The other eastern languages all appear to show

26

ISIDORE DYEN

reflexes of the type *-yu. At the same time it must be admitted that the matter has not been exhaustively investigated. There thus appears to be no reasonable solution at this time which does not posit that at some time in the history of Proto-Austronesian there was syllabicity in some of the semivowels hitherto reconstructed. Dempwolff reconstructed *y and *w in all of the positions of other non-syllabics except postconsonantal. The reconstruction of *y is exemplified in *layaR, Tag. la:yag, Mai. layar, Mer. lai, Fi. laca, To. laa, Fu., Sam. la 'sail'. The only instance of *y in initial position is in *yuyu[ ], Jav. yuyu 'fresh-water crab', Saa a-susu 'coconutcrab, birgus latro', To. quuquu, Fu. ququ, Sam. uuuu 'giant land-crab, birgus latro'. His reconstruction of *w is exemplified in *wawaq, Tag. wa:waq 'mouth (of river)', Mer. vava, Saa wawa- 'mouth'. The reconstruction of semivowels in final position will be exemplified below. In his article "Some new Proto-Malayopolynesian initial phonemes" (1962), Dyen proposed that there was evidence for a semivowel-like phoneme with labiovelar articulation that was different from *w; to this he assigned the symbol * W. Furthermore in his 'Formosan evidence' (1965a) Dyen reported that the word for 'right' should apparently be regarded as having a different initial from that of the word for 'eight'; to these he assigned the respective symbols *w1; and *w2. The correspondences involved are as follows: *W: * Wasu[ ], Tag. qa:so, Kavalan (Kv.) wasuP, Pazeh (Pz.) watsPuP, Bunun (Bu.) asu, Ami watsu, Paiwan (Pa.) vatu, Rukai (Ru.) adu 'dog'. *w1: *w1anaN[ ], Mer. ha-vanana, Kv. ma-wanan, Pz. Panan, Ami ka-wanan, Pa. ka-naval (metathesis), Ru. wanal, Saaroa (Sr.) halanu (methathesis) 'right'. *w2: *w2alu[ ], Tag. walo, Kv. Pu-waloP, Bu. vau, Ami valu, Pa. alu, Ru. walu, Sr. k-walu 'eight'. What is most interesting about the last correspondence is that it fits with certain reflexes in Toba and Javanese. Of the words hitherto reconstructed with initial *vf only Toba ualu '8' shows a reflex of the *w: e.g. *waRi[], Toba ari 'day'. This difference of reflex in the word for 'eight' can now be attributed to *w2. Similarly only the Old Javanese wwalu 'eight' shows such an initial: cf. *waRi[~\, OJav. we 'sun'. It would appear then that we must regard the distinction between these three initials as relatively well established; however the phonemic and phonetic interpretations are perhaps best postponed for the time being. With these exceptions however *y and *w are like other non-syllabics with the exception that Dyen does not reconstruct them after homorganic vowels in the manner adopted by Dempwolff. Dyen thus permits vowel sequences in reconstructions such as *buaq, Tag. buwaq 'additional growth in fruit', Mai. buah, Jav. wdh, To. fua 'fruit' where Dempwolff reconstructed *buwaq. It is obvious that Dyen's mode of reconstruction ignores the transition from the high vowel to a different vowel. Dropping the transitional semivowel was originally associated with the explanation of the con-

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

27

traction of vowel sequences in Javanese (e.g. 1945a: 13), but is not necessary after the fuller development of the hypotheses concerning the 'laryngeals' (cf. 1953a:46, and below). Rather the essential matter now is the distinctiveness of the semivowels associated with non-syllabicity that was thrown into question by the problems connected with reconstructing the word for 'wood'. One should add at this point that the individual analysts of two different Formosan languages agree in treating their respective languages as lacking syllabicity distinction in higher vocoids: T. H. Tung in A descriptive study of the Tsou language, Formosa (1964) and S. Egerod in his "A statement of Atayal phonology" (1966). This is not to say that the higher vowels are not sometimes impressionistically non-syllabic, but rather that the authors did not feel compelled to note this difference as a distinction. The implications of their findings have yet to be assessed. However it should also be noted that initial and intervocalic *w, *y appear as Tsou (Ts.) v, z respectively: *w2alu, Ts. voeu 'eight'; *[ ]ayam[ ], Ts. zomu 'bird'. In any case it is somewhat more convenient in the present state of affairs to continue the notation which involves what may be an otiose distinction in syllabicity in *y and *w; consequently we reconstruct both *kaS2iw and *kaS2uy for 'wood': for *S2 see below. Dyen reconstructed *-ey (1949:421 fn.) to account for correspondences such as that of Tag. matay 'die' with Mai. mati, To. mate 'die' (i.e. *-ey) as against that of Tag. qa:nay, Mai. anay-anay 'termite', To. arte 'moth' (i.e. *-ay). On the same pattern he reconstructed *-ew (1953b: 363, fn. 18) for the comparison of Tag. bu:gaw 'drive away', Toba mamuro (base -buro) 'scare birds away from fields', Mai. buru 'chase', Jav. buru 'try to catch, hunt', Ngj. ba-bohaw 'run away', mam-bohaw 'cause to run away', Saa huru 'run' (i.e. *buRew). The appearance of Saa final -u instead of the anticipated *-o was attributed to assimilation to the preceding vowel. Rufus S. Hendon in "The reconstruction of *-ew in Proto-Malayopolynesian" (1964) reconstructs *-ew instead to account for Tagalog -oq in comparison with Ngaju -aw (Tag. la:loq 'more, excessive', Ngj. lalaw 'past, exceed, too'), Fiji -o (Tag. la:koq 'door-to-door peddling, things to sell', Fiji lako 'go'), and Old Javanese o (Tag.pa:koq 'nail', OJav. pako 'nail', Fiji vako 'to nail (something)', i-vako 'nail'; Tag. ba:hoq 'foul odor', OJav. am-bd 'smell', Ngj. ewaw 'odor', bew-an 'stink', Fiji mbo 'have a smell', from *behew). Hendon's suggestion is to reassign *-ew from the correspondence to which Dyen had assigned it. Admittedly Dyen's correspondence is more closely patterned with that of *-ey, but there are no other comparisons to support it. Consequently Hendon prefers to divide Dyen's comparison into *buRaw 'drive away' (Tag., Toba, Ngj.) and *buru 'try to catch' (the remainder), thus essentially supporting Dempwolff's treatment though, as Hendon points out, one must also consider the chances that Tag. bu:gaw is a borrowing (and thus excludable from consideration), so that the remainder of the comparison could then be assigned to *buRew under Hendon's correspondence. There is some non-conforming evidence. A number of Tagalog words end in -oq

28

ISIDORE DYEN

where the other languages show reflexes contradicting Hendon's formula. Thus Tag. pa:loq 'beat' can hardly be dissociated from Fiji valu 'make war on', OJav. a-palu 'beat'. Furthermore even for some of the comparisons basic to Hendon's formula there is contradictory evidence: Ngj. paku 'nail' for his *pakew, OJav. lalu 'go past' for his *lalew, and laku 'go' for his *lakew. All of the last three cases could be explained as borrowings from Malay, most reasonably the Ngaju word. If the Old Javanese words are not Malay loans, they could be due to manuscript corruptions. Hendon would however accept an analogically developed q (such as is suggested by Dyen [1953a:20ff.]) as the best available explanation for the final of Tag. pa:loq 'beat'. Hendon has obviously brought together material which bears on the solution of a number of rather difficult discrepancies. It remains to be seen whether the particular phonological assignment of the correspondences will continue to prove attractive with the accumulation of further evidence.

THE LABIALS

Among the reconstructed phonemes with a labial articulation there is little doubt that there was a simple labial series *p, *b, *m such as is reconstructed by Dempwolff. However Haudricourt in his "Variations parallèles en Mélanésien" (1951) raised the question whether there is not evidence for a separate Proto-Austronesian labiovelar series generally reflected in the western (Indonesian) languages as simple labials and therefore not separately reconstructed by Dempwolff. The evidence concerns in part reconstructions which have hitherto been assigned with some degree of awkwardness to initial clusters (Dyen 1951:fn. 9). Dempwolff (1934-38:2.135 ff.) concluded that the Fiji intervocalic -mb-, -nd-, -ndr-, -rjg-, actually unit phonemes, corresponded to nasal clusters of Proto-Indonesian: e.g. Fi. tumbu 'grow', PIN tumbuq; Fi. lombi-a, PIN lempit; Fi. mundu 'cut off', PINpuntuy ; Fi. vandra 'pandanus', PINpanDan ; Fi. warjga 'boat', PIN waijkaij. He interpreted like initial correspondences in the same way: Fi. mboyi 'night', PIN berji', Fi. mbune 'wild dove', PIN punay; Fi. ndovu 'sugar cane', PIN tebu; Fi. ndrau 'leaf', PIN dahun; Fi. ygari-a 'scratch', PIN garis; Fi. rjgili 'rub in hands', PIN giliij 'roll'. He drew this conclusion despite the fact that it seemed to imply the possibility of an initial cluster in Proto-Austronesian. He made a similar conclusion for the evidence of Saa (2.155 ff.) and the Polynesian languages (2.179ff.). Similarly interpretable phenomena were found by Dyen in his treatment of Trukese (1949:424 passim), who however offered the explanation that all had developed by analogical change (425). Under such a hypothesis the assignment of initial clusters to Proto-Austronesian couldbe avoided. Another way of avoiding this assignment would be by positing the development of the implied initial clusters in a proto-language after Proto-Austronesian, say, Proto-Oceanic. Haudricourt (1951) argues that there is evidence for New Caledonian (NC) phonemes symbolized as *kp, *"gb, *"m and *kph. His *kp and *kph are not known to have cog-

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

29

nates elsewhere and thus do not concern us further. Irfthe case of *"gb he finds that Dempwolff has reconstructed the same proto-phoneme as for cases for which there are different reflexes in New Caledonian. For example for DempwolfT's reconstructions *batu 'stone', *bulu 'hair', *betji 'night', *baTuk 'skull', we find the following New Caledonian reflexes: Poapoa Moaeke Haeke

'stone' paie vai vai

'hair' puni fudufudu-

'night' g en m w b en m w b en A w

'skull' "gwam w b am w b a-

Haudricourt shows that this difference also appears medially in NC: e.g. Poapoa ncfguc, Moaeke nambu 'mosquito' but such words have not been associated with cognates in western languages. If the hypothesis of a Proto-Austronesian labiovelar stop is thus restricted in its necessary application to the explanation of initial reflexes, it cannot be evaluated as superior to the hypothesis of an initial cluster of analogical origin. Haudricourt finds a similar failure in DempwolfT's reconstructions to account for the difference in the reflexes of DempwolfT's *lima 'five', *manuk 'bird' on the one hand and *Rumaq 'house' and *mu(n)taq 'vomit' on the other, as reflected in NC: Poapoa Moaeke Haeke

'five' nim nim nim

'bird' manic mani

'house' wa mwa mwa

'vomit' nuja muja muja

The differences in the NC reflexes for 'house' and 'vomit' would require a subsidiary hypothesis attributing their differences in reflex to another factor, say the difference in their respective environments. Goodenough in treating "The Willaumez languages of New Britain" (1961a) suggests the presence of a proto-phoneme *mw in ProtoAustronesian. The evidence generally involves etymologies lacking cognates in languages of the Philippines or of Indonesia. Reliable exceptions are few: Fiji ta-gane 'male', Trukese mwaan, Balinese anak-moani 'man, male'; Trukesejimwa-n 'his house', Malay rumah 'house'. Although this maldistribution does not by itself necessarily weaken the argument for the presence of this phoneme in Proto-Austronesian, it is not yet unlikely that the available correspondences can be explained otherwise. The argument for a labiovelar nasal can perhaps be regarded as stronger than that for a labiovelar stop. Nevertheless it should be pointed out that the appealing etymology for the Poapoa word for 'house' demands an explanation for the absence of a reflex of the *u of the reconstruction. Furthermore it is not impossible that the n of the Poapoa word for 'vomit' (and also of Poamei nujai 'vomit') could have another explanation than a labiovelar origin. There can be little question that the distinction observed by Haudricourt in NC languages is represented in other eastern languages, as for example in Tr. wyyn 'hair', faaw 'stone', pwiin 'night', niim 'five', maan 'animal, bird',y'H'mw 'house', mwus 'vomit'.

30

ISIDORE DYEN

The following comparisons can also be cited: Lau kwou, Saap w au 'head' (assigned to *baTuk); Lau urja, Gilbertese umwa 'house'; Lau ifaa, Saa mwaa 'snake', Gilbertese myata 'worm'. Haudricourt's hypothesis cannot be ignored, but is not compelling. The hypothesis that the initials arose from clusters by analogy will explain the contrast in the stops. Since the contrast in the nasals is limited to the neighborhood of *u and otherwise to words not attested in the west, the hypothesis that the labiovelar nasal arose from an allophone of the normal nasal is still a not unreasonable hypothesis. It is perhaps for this reason that he has more recently expressed uncertainty about 'the ancient status of an Oceanic labiovelar order' (1965:325, fn. 17). THE APICALS AND FRONTALS

There can be little doubt that the apicals and frontals offer great difficulties in reconstruction. For the time being the need appears to be in the direction of expanding the number of proto-phonemes, or at least the provisional proto-phonemes, in order to account for the correspondences. There is good evidence for reconstructing two voiceless apical stops, namely *t and *C, instead of Dempwolff's *t: e.g. *t:

*batuP[ ], Tag. bato, At. batu-nuh, Se. butu-nuh, Saisiyat (Ss). batu, Kv. vatoP, Pz. batu?, Thao (Th.) fatuP, Bunun (Bu.) batu, Ts. fatu, Kanabu [ = Kanakanabu] (Kn.) batu, Sr. batuPu 'stone'. *C: *maCaP, Tag. mata 'eye', At. ma-masa 'eyeball', Ss. masa, Th. mdaP, Bu. matta, Ami mata, Puyuma (Pu.) mata, Pa., Ru. matsa, Ts. mtso 'eye'. The evidence for this distinction was discovered by Ogawa and Asai (1935) and restated by Dyen (1965a). The evidence for Dempwolff's *Tis chiefly offered by Javanese {in correspondence with other reflexes otherwise associable with his *t: e.g. Tag. kv.ta 'see', Jav. was-kito 'clear insight, foresight', Ss. k-um-ita, Sr. k-m-akita 'see'. Additional evidence appears in the following comparison: Tag. batok 'nape of neck', Jav. bafuq 'forehead', Saa pwau 'head'. If Trukese faat 'eyebrow' (Jety-n 'his eyebrow') is associated with this etymology, it would suggest that Trukese t in similar correspondences reflects *T and then Trukese futuk 'flesh' (futuke-n 'flesh-of') might imply *biTuka with Tag. bituka, Toba butuka 'intestines'. The vocalism of the Trukese word would then perhaps be best explained from a form with *u in the first syllable by assimilation to the second (Dyen 1949:432) as in the Toba cognate. Although the evidence for *T is extremely limited, it should not be ignored (but cf. Haudricourt 1965:320). In this connection it should be observed that Graged (Gr.), a language of Astrolabe Bay (Territory of New Guinea), unaccountably exhibits two different reflexes for each of both *t and *C: thus, Gr. pat 'stone' (*batuP[ ], see above), pal 'four' (*x2epat[ ], Se. sepat, Ss. spat, Kv. Pu-sapat, Pz. supat, Th., Bu. pat, Ami spat, Pa. spat, spac, Ru.

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

31

spat, Ts. suptu, Kn. so:patu, Sr. pa:tu 'four'), and on the other hand Gr. mat 'die' (*maCey, *paCey, Tag. patay, Kv. patay, Th. maday, Bu. ma-patad, Ami mi-patay, Pu. p-in-a{ay, Pa. pa-patsay, Ru. wa-patsay, Ts. o-ptsoyi, Kn. miya-patsay, Sr. pa-patsi 'die') and mal 'eye' (*maCaP, see above). The explanation for these correspondences is not immediately apparent. It has been suggested orally to me by Milke that it might be due to borrowing between a language showing t for *t and *C and another showing / (perhaps from *r of an intervening stage) for both proto-phonemes. Since the implications of this suggestion have not yet been investigated, the Graged distinction has not been introduced into the reconstructions. The evidence for two apical nasals *n and *N (both different from *rj [the last originally symbolized as *N in the Laryngeals]) was also first observed by Ogawa and Asai (1935) and reviewed by Dyen in Formosan evidence (1965a). The distinction is exemplified by: *n: *panaq[ ], Tg. pa:naq 'arrow', Mal. panah, To. kau-fana 'bow', Ss. manaP, Pz. pa-panaP, Ami mi-panaP, Pu. p/irt/anaP, p/in/anaq, Ru. wa-Pana, Ts. puno, Kn. mwa-panaPu 'shoot'. *N: *tuNa, Mer. tuna 'large eel', To. tuna, At. tala-qi, Ss. tula, Ami tula, Pa. tula, cula, Ru. tula 'eel'. The evidence for the distinction is thus far only from the Formosan languages. There remains however the possibility that it is an instance of *N that accounts for the / of Ilk. qaldaw 'day', and the n of Ngaju andaw 'day'. It should also be noted that the chances are good that *N was probably a lateral and probably also non-nasal, for the languages that reflect this phoneme distinctively from *n do so with a lateral [and this suggests that a phonetically more appropriate symbol might be *L]. One way of attacking the problem of the voiced apical stops is to consider those cognate sets which include widely diverse languages and to see how many different reconstructions are necessary to account for them. Among the sets which involve voiced correspondents there are two among those reconstructed by Dempwolff with *D which provisionally we shall call *D1 and *D2. There is an ambiguous reconstruction symbolized by Dempwolff as *[dD] which as far as I know does not show distinguishing reflexes in Formosan languages. *D1 is typically represented by the initial of the word for 'two' and *D2 by the words for 'interior' taken together with the word for 'water': *£>!: *D1ewS3a, Tag. da-lawa 'two', Jav. pin-do 'twice', Mal. dua, At. rusa, Se. daha, Ami tusa, Pu. dua, Pa., Ru. dusa, Ts. yuso 'two'. *D2: *D2aya, Tag. qi-la:ya 'interior', Jav. atjen daya 'offshore wind', Mal. barat daya 'southwest', Se. daya 'east', Pu. i-zaya, Pa. i-zaya 'mountain direction', Ru. dala, 'east'. *D2aNum[ ], Ilk. danum, Ami nanum, Pu. zanum, Pa. zalum, Ts. chumu 'water'.

32

ISIDORE DYEN

As an instance of*D 2 in medial position one can compare Tag. qu:lay 'intestinal worm', Jav. w-ude 'Aftermaden', To. ule 'penis'. (See Note, p. 54.) Similarly there is evidence that not all of the instances assigned to *Z in agreement with van der Tuuk's Third Law (Dyen 1951) will conform to each other. I have provisionally called these *ZU typified by the word for 'rain', and *Z 2 , typified by the word for 'needle': * Z t : *quZ1aN[ ], Tag. qulan, Jav. udan, Mai. (h)ujan, At. qwal-ax, Se. Putal-ah, Pu. hudal, Pa. qudal, Ru. udalu, Ts. m-uchu 'rain'. *Z2: *Z2aRum, Ilk. da:gum, Jav. ddm, Mai. jarum, At. raum, Pu. daum, Pa. daum 'needle'. Regardless of one's views about the distinction between *D1, *D2, and *D3 on the one hand and between *Zi and *Z 2 on the other, there can be little question as to the distinction between the correspondences associated with van der Tuuk's Second Law (1865b:443f.; called the RLD-law by Brandstetter [e.g. 1906:13], but actually ambiguous for Dempwolff's *D and */'; see below) and those associated with his Third Law (reconstructed as *Z). Indeed since Dempwolif's *D was essentially reconstructed to account for a Tagalog I in correspondence with a Javanese d (or d if that occurred) or a Javanese d in the absence of a Tagalog cognate, all other correspondences being the same as those for his *d, the position is now reached that the evidence for a separate *d needs to be reexamined for probative value. The characteristic of Dempwolff's *d is the correspondence of Tag. d (and intervocalic r) with Jav. d. But if, as was done above, the correspondence of Tag. / with Jav. d is not assigned with Dempwolff to *D (but to *D2), then the appearance of Tag. d in correspondence with Jav. d is not sufficient to distinguish *d from *D2. If the d of the Tagalog word originally alternated with /, it would be assignable to *D2. Thus for example comparison of Tag. dagas 'haste' with Jav. deres 'swift (of river)' could reflect *D2 as well as *d. On the other hand the correspondence of initial Tag. d with Jav. d could reflect *D± against Dempwolff who assigned this correspondence to *d. Thus Tag. da:hon, Jav. ron-don 'leaf could reflect *Dt or *D3 and the initial correspondence need not be assigned to *d as was done by Dempwolff. The consequence is that direct evidence for the *d inferred by Dempwolff requires the appearance of Tag. r in positions in which it did not alternate with d, for in alternating positions it could be due to analogy. Very few of such instances are associated with eastern (Melanesian, Polynesian) cognates. [It is perhaps this lack of eastern evidence that suggested to Haudricourt (1965:323) the collapsing of Dempwolff's *d and *£>.] Our interest in such cognates is their implication that the distinction under consideration is to be assigned to ProtoAustronesian. In the absence of such cognates, it is not at all impossible that if Tagalog (and other Philippine languages) and Javanese (and other southwestern languages) form the subgroup constructed by Dempwolff under the name Indonesian,

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

33

then the correspondence could perhaps reflect a phoneme not traceable beyond the proto-language of that subgroup. However this concern is perhaps premature in view of the vast problems that face work in Proto-Austronesian. There can be little question that Tagalog often shows intervocalic r in correspondence with voiced stops in other languages: e.g. Tag. ha:riq 'ruler', Jav. adi 'excellent, splendid', radén (*raadi-an) 'prince'. This comparison is perhaps the best example of those in which Javanese shows d. There are however other instances in which Javanese exhibits d contrary to Dempwolff's correspondence: e.g. Tag. darak, Jav. dédaq 'bran'. Such a correspondence is normally treated by Dempwolff under the hypothesis that the Javanese word is a loanword from Malay and in fact there is a Malay dédaq 'bran'. Undoubtedly Dempwolff would do the same with Tag. ha:raij 'attack', Jav. adarj 'wait for' (Mai. adarj 'lie in wait'), and Tag. harap 'front', Jav. arép-arép-an, adép-adép-an 'front' (Mal. hadap 'front'). However these instances in which the Javanese words belong to the standard ngoko status-style (used to inferiors or to those in the same lower social status) suggest that these correspondences should not be summarily dismissed. If Tag. r is permitted to correspond with Jav. d, one can approach reconciling Tag. tiris 'press flat', Jav. teres 'ringed (of dying tree)' with Jav. tiijidés 'press' (Mai. tindas 'press', interpretable along with the Javanese word as from a form with infix -in- and loss of interconsonantal *-e-: i.e. from a form like *t/injedes). There thus appears some reason to assign the reconstruction dx to those instances for which Javanese offers d in correspondence with Tag. r and to assign *d2 to those instances in which Javanese offers d for Tag. r. Although the Formosan languages suggest that our *Z perhaps applies to two different reconstructions *Zj and *Z 2 , other problems still remain. The typical correspondence for Dempwolff's *z appears in Tag. qa:ral, Jav., Mai. ajar 'learn'. However there are instances in which Tagalog appears to exhibit /: Tag. talim 'cutting edge', Jav. tajem, Mai. tajam 'sharp'; Tag. hilaw, Jav. ijo, Mai. hijaw 'green'. There is thus reason to speak of *z1 with Tag. r reflex and *z2 with Tag. / reflex. An instance of initial *z2 is perhaps to be found in the comparison of Tag. la:qit 'insult' with Jav. jahat 'burst', Mai. jahat 'bad', although the second vowel of the Tagalog word offers difficulties. There are a number of other problems that remain in the matter of voiced apical stops, but the few that have been brought up here will suffice to show that the work is far from complete. The reconstruction of *l and *r is based on comparisons such as the following: *latjit, Tag. la:rjit, Mai., Jav. layet Fi., To., Fu., Sam. larji 'sky'; *garis, Tag. galis 'scabies', Mai., Jav. gales, Fi. kari 'scratch', Saa Pa-Pari 'tear off, splinter off'. Dempwolff (1.39) argued that *s was probably a stop. This proto-phoneme was undoubtedly in alternation with a nasal, presumably *«, which he classified as a palatal nasal, as exemplified in Jav. surotj 'push', ñurorj (active) 'push'. On these grounds he assigned a 'forward' palatal articulation to *s, for which I would prefer the description

34

ISIDORE DYEN

'frontal' in terms of the tongue-position involved. However, since there is no language which exhibits a stop articulation, there is the danger that Dempwolff is here reconstructing the phonetic characterization of the phoneme in pre-Proto-Austronesian times rather than its characterization in Proto-Austronesian. In any consideration of the implications of the nasal alternants of *s for its phonetic character, it is necessary to observe the absence of eastern evidence for a nasal alternant. For the *s reconstructed by Dempwolff there were perhaps two different sibilants indicated by the difference in treatment in Malagasy and Maanyan. This difference was first noted by Dahl (1951:40ff.) and attributed to dialect mixture (1951:42f.). This explanation is at first glance not implausible particularly if the change is restricted to these two languages. Under a hypothesis that the two languages continue the same private proto-language, we could avoid the necessity of reconstructing two different proto-phonemes. Since however there is some evidence in Saaroa (a Formosan language) of a similar distinction we make the provisional reconstructions *slt and *s2 : : *les1uy, Sr. lotja, Mny. lehung, Mer. laona 'mortar'. *s2: *s2iwa[ ], Sr. k-siya, Mny. suai (metath.), Mer. sivy 'nine'. THE DORSALS

The evidence for the dorsal stops is generally clear, but not as conclusive in all cases as one might like. For *k, *g, *y, there are the following: *kaS2iw, *kaS2uy, Tag. ka:hoy, Toba hau, Jav., Mai. kayu, Mer. hazu, Fi. kau, To. qa-kau 'tree, wood', At. kahu-niq, Se. qahu-ni, Ss. kahui, Pz. kahuy, Th. kawi? 'tree', Ami kasui 'firewood', Pu. kaui, Pa. kasiu, Pasiu, Ts. evi, Kn. kalu, Sr. kiyuPu 'tree'; *ge(l)gel, Tag. gilgil 'cut', Jav. gegel 'notch', To. koko 'notch'; *gilay, Jav. gilarj 'twinkle', Mai. gilay, To., Fu. ki-kila, Sam. qi-qila 'glitter, sparkle'; *Caijis[ ] (by a slip incorrectly reconstructed as *tarjis in Dyen 1965a:290), Tag. ta:yis, Toba tarjis, Jav., Mai. tarjes, Ngj. tayis, Mer. tani, Fi. tarji, To., Fu., Sam. tayi, Th. 8-m-anit, Bunun, southern district (BuS.) tayis, Ami t-im-ayits, Pu. f-im-a-(ayis, Kn. t-um-a-tayi, Sr. t-um-a-tayi-i 'weep'. Goodenough (1961b) has suggested that a Bulu (New Britain) distinction of velars must be assigned to Proto-Austronesian. Thus Bulu exhibits kuli 'skin', but xutu 'louse' for *kulit and *kutu respectively. Since the evidence for this distinction as Proto-Austronesian is perhaps still not too strong, the distinction is best indicated by using his provisional reconstructions *kt for the correspondence in 'skin' and *k2 for that in 'louse' until further evidence accumulates. The other proto-phonemes assigned by Dempwolff (cf. 1.64) as palatals to this articulation are *c, *j, *n, and *R. The strongest evidence for *c is to be found perhaps in the following: *cabut, Mai., Jav. cabot, Fi. cavut-a 'pull up, out'. The proto-phoneme *c is assigned a dorsal articulation by Dempwolff who finds that it is 'der stimmlose palatale Verschlusslaut mit vorwiegender Beteiligung der Hinterzunge' (1.61). He rightly found it necessary

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

35

to reconstruct this new proto-phoneme in addition to *s, despite the fact that most languages offer the same reflex as for *s. No language offers a reflex with a predominantly rear-of-the-tongue articulation; it is only some of the languages of western Indonesia that offer different reflexes and these are frontal stops (e.g. Malay) or affricates (e.g. Javanese). It thus seems difficult to see why one should conclude that *c had a dorsal articulation. One consideration is the articulation of *s itself. Above, as we have seen, Dempwolff reached the hypothesis that *s was a palatal stop with a 'forward' tongue-blade, i.e. frontal articulation. It seems natural to conclude that if *c also had a palatal articulation, parts of both the palate and the tongue involved in the articulation of *c would be back of that of *s. However, if *s was not actually a stop in Proto-Austronesian (whatever it was in pre-Proto-Austronesian) then this argument for a dorsal as opposed to a frontal tongue articulation would disappear. The remaining argument for a dorsal articulation of *c lies in its proposed similarity to *j, which does show evidence of a dorsal articulation. It may perhaps be too early to conclude that these two consonants do in fact share this feature of articulation in the absence of any similarity in their behavior in the morphology to match the natural temptation to assign them to the same row in a display of the proto-phonemes. The proto-phoneme *j almost certainly had a dorsal articulation. It is, thus far at least, reconstructed only in intervocalic and final positions. This proto-phoneme is one of those satisfying van der Tuuk's Second Law (van der Tuuk 1872:143 [and less explicitly 1865b:443]: so called by Brandes [1906:139]). Among the western languages the Batak languages of Sumatra and Ibanag in the Philippines exhibit intervocalic g and final k for *j, thus continuing the distinction from *D which was lost by merger in many of the other western languages. These velar reflexes were noted by Kern (1906:11 and passim) and also by Brandstetter who, like van der Tuuk, did not distinguish the two proto-phonemes, but rather included both under the RLD-law, noting the Batak and Ibanag reflexes above as surprising (1906:61; 1908:22,26). Conant however seems to regard the velar reflexes as their only reflex of the 'consonant of the RLD-series' (1911:83) and he was followed in this respect by Dempwolff (1934-38:1.55). The determination of *j as distinct from *D is thus to be attributed to Dempwolff. Examples are: ^q^ijutj, Tag. qilorj, Toba igutj, Jav. irurj, Mai. hidurj, Fi. ucu (assim.), To. ihu, Fu., Sam. isu; *belaj, Tag. bilad 'spread in sun', Toba bolak 'broad', Mer. velatra, Saa hola, To., Fu., S a m . f o l a 'spread out'. The reconstruction of the proto-phoneme *n is based on comparisons like the following: *hamuk, Tag. namok, Mai. hamoq, Fi., To., Fu., Sam. namu 'mosquito'; *penu[ ], Mai., Jav. penit, Fi. vonu, Saa honu, To., Fu.fonu 'sea turtle'. That it persisted as a separate phoneme into the post-Proto-Austronesian history of some eastern languages is indicated by Bugotu (Ysabel Is., Solomons) hamu 'mosquito', vono (assimilation) 'sea turtle'. Like *c, *n is not reconstructed in final position. The proto-phoneme *R is described by Dempwolff as a voiced velar spirant (1934-38:1.54). Its correspondence was first formulated by van der Tuuk (1865:443

36

ISIDORE DYEN

and elsewhere) and was given the name of van der Tuuk's First Law by Brandes 1884:30) and referred to as the RGH-law by Brandstetter (1910:12). Its reflexes in the Philippine languages were exhaustively treated by Conant (1910) along with those in the so-called RLD correspondence. However Dyen (1953b) collected comparisons which suggest that perhaps more than one proto-phoneme is concealed in Dempwolff's *R. Four correspondences were distinguished, each assigned to a provisional reconstruction marked by a numerical subscript. is distinguished from *R2 by the Merina reflex: *R1abut, Tag. ga:but 'tear out', Mai. rabot 'tear off', Mer. avutra 'tear out'; *R2atus, Tag. gatos 'trillion', Jav. atos, Mai. ratos, Mer. zatu 'hundred'. *R3 is distinguished from the preceding by the Javanese reflex: *qaR3us, Tag. qa.gos, Jav. aros, Mai. (h)aros 'current', Ngj. asoh 'flow off'. *i?4 is distinguished from all the others by the Merina reflex: SeBs. kugi:ta, Jav. kerito, gerito, Mai. gurita 'octopus', Mer. hurita 'squid'.

THE LARYNGEALS

As was indicated above, Dempwolff reconstructed a proto-phoneme *h which we symbolize by *q. He inserted in his reconstructions a symbol c which we call smooth transition and symbolize *h. It is in effect a proto-phoneme, but Dempwolff's treatment leaves its status as he saw it unclear. I shall here treat it as though it had been proposed as a proto-phoneme. Although generally its positions are like those of non-syllabics, it does not cluster with them. This proto-phoneme is a smooth transition (1) from pause to a vowel, (2) between vowels, or (3) from vowel to pause. It is convenient to consider Dempwolff's *q, *h (in their new values) and his *y, *w after *i, *u respectively, an area of reconstruction that we call laryngeals. The term is to be regarded as a cover term, not in its etymological sense. Dyen's The Proto-Malayo-Polynesian laryngeals (1953a) aimed at suggesting a different relation between Dempwolff's initial laryngeal correspondences on the one hand, and his medial and final correspondences on the other. Dempwolff distributed the laryngeal correspondences as follows: 1. Initial: *h: Tag. q, Mai. 0, To. 0: Tag. qa:nay, Mai. anay-anay 'termite', To. ane 'moth'. *q: Tag. h, Mai. h, To. 0 : Tag. hu:nos 'moult', Mai. hunos 'pull out', To. unuh-i 'draw out'. 2. Intervocalic: *h: Tag. h, Mai. 0, Jav. contraction: To. 0, Tag. da:hon Mai. daon, Jav. ron, ron-don, To. lau 'leaf.

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

37

*q: Tag. q, semivowel, Mai. h, 0, Jav. h, To. q: Tag. taqon, Mai., Jav. tahon 'year', To. taqu 'season'. Mai. tua 'old', To. ma-tuqa 'parents'. *semivowel: Tag. semivowel, Mai. 0, Jav. semivowel, To. 0: Tag. luwaq 'emission', Mai. luah 'spit out', To. lua 'vomit'. Toba ma-tua, Jav. tuwo 'old'. 3. Final: *h: Tag. 0, Mai. 0, Jav. 0: Tag. pito, Toba, Jav.pitu, Mer.fitu, Fi. vitu, Saa hiu, To., Fu., Sam. fitu 'seven'. *q: Tag. q, Mai. h, Jav. h: Tag. mamaq, Mai., Jav. mamah, To. mama 'chew'. Dempwolff's treatment suffered from a number of defects. For example it produced a certain number of doublets, one of which is illustrated in the words for 'old' among the intervocalic correspondences listed above. Actually the number of doublets produced in initial position is rather noticeable (1953a: 49). The chief objection to Dempwolif's treatment is not in the phonetic interpretation (for otherwise the substitution of symbols would have been misleading), but in the fact that initial Tagalog h was needlessly assigned to a different phoneme from Tagalog h elsewhere as was similarly initial Tagalog q. Dyen suggested that three elements were needed: *q, *h, and a zero element, the last of which could be equated phonetically with Dempwolff's smooth transition, or another phoneme, or regarded as nothing. Dyen chose the last alternative as the simplest. In effect then in the Laryngeals Dyen, among many details, simply reassigned Dempwolff's initial correspondences and those of *nothing in all positions as follows: *q: Tag. q, Mai. h, To. q: Tag. qa:gos, Mai. (h)aros, To. qau 'stream'. This comparison was divided by Dempwolff into *qaRus and *haRns. *h: Tag. h, Mai. 0, To. 0 : Tag. hu:nos 'moult', Mai. (h)unos 'pull out', To. unuh-i 'draw out'. *nothing: Tag. q-, -0- or -q-, -0 (with q or n before suffix), Mai. 0, To. 0: (initial) Tag. qa:nay, Mai. anay-anay 'termite', To. ane 'moth'; (intervocalic) Tag. luwaq 'emission', Mai. luah 'spit out', To. lua 'vomit'; Tag. ka:qin, Mai. ma-kan 'eat', To. k-in-a 'meat-dish'; (final) Tag. mata 'eye', mata:-qan 'be looked around for', Mai. mata, To. mata 'eye'. One of the hypotheses formulated by Dyen appears in the following: 'The Malay material is quoted from Wilkinson. An important feature of his orthography is that he writes h between a and a different vowel in certain words and not in others: e.g. bahu "shoulder", bau "smell". In the speech of my informants these words were not normally distinguished. As far as I know through haphazard checking, his orthography is accurate for the transcription of the h in the Arabic orthography of Malay. The occurrence of a graphic h between vowels is significant for the reconstruction of

38

ISIDORE DYEN

PMP q ... This finding implies of course that the Arabic orthography has preserved a real distinction which occurred in Malay earlier than the speech of my informants. Occasionally Wilkinson records words both with and without h. This fact is noted by parentheses around the h: thus tu(h)a "old" means that both tuha and tua occur' (1953a:7). I follow the same practice here throughout in respect to the citation of Malay (and Javanese) words as in the Laryngeals. Not everyone agreed that these laryngeal hypotheses were correct. In a review of Dyen's Laryngeals, Uhlenbeck (1955-56) felt that his own 'rather extensive review of the sources used by the author demonstrates that the reconstructions of the laryngeals which the author gives ... are not based on a solid foundation. It is hardly doubtful that at most one of the three languages compared, viz. Tonga, possesses a glottal stop in initial and intervocalic position, with the result that the reconstruction of PMP q in these positions is highly uncertain. In addition, the relation between the languages spoken in Polynesia and those in Indonesia make it improbable that Tonga — and the Polynesian languages in general — would have retained a glottal stop in these positions whilst this stop had disappeared elsewhere. In any case it is difficult to arrive at the reconstruction of PMP initial q within the narrow limits of Dyen's investigation. 'A second doubtful point with likewise fundamental consequences for Dyen's reconstructions is the uncertainty concerning Javanese and Malay h. It is difficult to accept without further evidence that in words like jait and pait (Dyen writes jahet and pahet), which occur both in Malay as in Javanese, h was ever actually pronounced, as at present no h is heard. Amongst Dyen's material only the word tuha might tend to show that a sound-change took place from intervocalic h to the semivowel corresponding to the preceding vowel, because in this case a real change in spelling seems to have taken place. In modern Javanese and Malay the word tuwa is always written and pronounced with w, although in the archaizing Javanese poetry the spelling tuha still occurs. It is far from impossible that eventually the occurrence of h may be proved also in initial and intervocalic position in an earlier stage of Javanese. Careful research may reveal perhaps other significant changes in spelling, and comparison with e.g. Sundanese may demonstrate the presence of initial h in ancient Javanese words, e.g. Jav. yan, god: Snd. hiyan, a comparison indicated by Esser already in 1929.' The reviewer seems to have missed the point that Tagalog, like Tonga, has an intervocalic glottal stop. He furthermore failed to take into account the rather broad phonetic characterization of both *q and *h stated in the introduction of the Laryngeals (p. 1, and particularly fn. 2). The possibilities for *q are there summed up as not only (1) a glottal stop (as the reviewer seems to conclude), but also (2) a glottal spirant, or some other type of post-velar stop or spirant. The reviewer's caution about the uncertainties regarding Javanese and Malay h is somewhat exaggerated. Malay h will be dealt with below; as for Javanese, it can only be noted that Old Javanese

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

39

generally supports Dyen's inferences, though unfortunately the evidence cannot be presented here. The reviewer's concern for the dangers involved in the comparative method suggest that it is perhaps worthwhile to record here some evidence that applies to the occurrence of intervocalic Malay h. For the reconstruction of *q in the Laryngeals (1953a) Dyen used the evidence of orthographic h in Malay in positions in which, as in /paet/ 'bitter' — actually written pahit, but under the special conventions in the Laryngeals, there written pahet — it did not correspond to /h/ in pronunciation. However there does exist a Malay dialect in which /h/ appears in nearly all of the instances in which a formerly pronounced /h/ was inferred from the evidence. This dialect is Bandjarese (Bjr.) as spoken by Abdulgani Madjedi, born in Amuntai, Kalimantan (Borneo), not far from Bandjarmasin. What was most striking in his speech was a well-marked voiceless vowel (or glottal spirant), which I will call /h/, occurring between unlike vowels in words cognate with Malay words with only orthographic h. The following is a list of the critical words with the reconstructions in the Laryngeals to which they were attached: *q\ Bjr. jahit, Mai. ja{h)et, Tag. daqit, daqi.ti 'unite', To. haqi 'bind together' 'sew', *zaqit. Bjr. tuhaq, Mai tu(h)a, To. ma-tuqa 'parents' 'old', *tuqa(h). Bjr. dahiq, Mai. da{h)i, To. laqe 'forehead', *(dD)aqey. Bjr. pahit, Mai. pa(h)et, Tag. paqit 'bitter', *paqit. Bjr. tahun, Mai. ta(h)on, Tag. taqon 'year', To. taqu 'season', *taqun. Bjr. buhayaq, Mai. bu(h)aya, PgTg. buqa:ya 'crocodile', *buqaya[ ]. *h: Bjr. darn, Mai. daon, Tag. da:hon, To. lau 'leaf', *dahun. Bjr. kait, Mai. kaet 'catch with a hook', Cebu Bisayan (SeBs.) kahit 'hook', *kahit. Bjr. lauk, Mai. laoq 'food eaten with rice', Tag. lahok 'mixed', *lahuk. Bjr. naik 'climb', Mai. naeq, Tag. pa-nhik 'mount', *nahik. ^nothing: Bjr. banua 'town and surrounding area', Mai. benua 'continent', HIBs. banua 'town', To. fonua 'land', *banua. Bjr. buah, Mai. buah, To. fua 'fruit', *buaq. Bjr. laut, Mai. laot 'sea', Jav. lor 'north', Bis. lawod 'sea', *laud. It is perhaps also interesting that Bandjarese shows initial (and unmistakable) /h/ in the following: *q: Bjr. Bjr. Bjr. Bjr.

harus, Mai. {h)aros, Tag. qa:gos, To. qau 'stream', *qaRus. hanut, Mai. hanot, Tag. qanod, To. ma-qanu 'float', *qanud. hati, Mai. hati, Tag. qatay, To. qate 'liver', *qaiey. hatap, Mai. {h)atap, Tag. qatip 'roof-covering', To. qato 'cover r o o f , *qatep.

40

ISIDORE DYEN

Bjr. hubi, Mai. ubi, Tag. qubi, To. qufi 'yam', *qubi(h): Bjr. ba-humbi 'have a taproot', Mai. (h)umbi 'taproot', *qumbi(h). Bjr. hundarj 'shrimp, cray-fish', To. quo, *qunDarj; Mai. (h)uday 'lobster, shrimp', Tag. qulay 'lobster', *quDatj. Bjr. hidup, Mai. hidop (both with metathesis), Jav. udep, urep 'live', *quDip; To. mo-qui 'live', *qunDip. Bjr. hutan, Mai. (h)utan 'forest', To. quia 'interior', *qutan. Bjr. hujan, Mai. (h)ujan, Tag. qulan, To. quha 'rain', *quZan. Bandjarese agrees with Malay in having initial vowel (varying freely with initial q [glottal stop]), in correspondence with Tagalog initial q and Tonga initial vowel. These were originally reconstructed with initial vowel in the Laryngeals and this reconstruction is cited; where a new reconstruction has been reached, it is indicated in parentheses: *nothing: Bjr., Mai. aku, Tag. qako, To. au 'I', *aku. Bjr., Mai. anay-anay, Tag. qa:nay 'termite', To. ane 'moth', *anay. Bjr., Mai. api, Tag. qapoy, To. afi 'fire', *apuy (now *x1apuy). Bjr. anam, Mai. enam, Tag. qa:nim, To. ono 'six', *enem (now *Penem[ ]). Bjr. ampat, Mai. empat, Tag. qa.pat, To. fa 'four', *empat, *epat (now *x2epat[ ]). The appearance and non-appearance of Bandjarese /h/ nearly always agrees with that of fifth) in Malay. In initial position some discrepancies were observed: e.g. Bjr. ampar, Mai. (h)ampar 'spread out' (despite To. qe-qepa 'spread out'). Thus it can be inferred that some of the difficulties dealt with in the Laryngeals that involve the Malay correspondence will continue to be problems even after the evidence of Bandjarese and dialects like it have been considered. A most surprising turn of events is the consequence of reexamining the reflexes assigned to initial *h in view of the Bandjarese evidence. If it be estimated that Bandjarese initial h is more reliable evidence than the Malay orthographic (and real) h, then it is possible to reconsider all those instances in which there is evidence from Malay, Bandjarese, Tagalog (and other Tagalic languages), and Tonga. The surprising fact is that I have found only a few instances of the correspondence Bjr. h, Mai. (ih), Tag. h, To. 0. There are in fact four different correspondences. Here must be taken into account the vagaries of comparative evidence, for Bandjarese evidence is not always available when desired. For this reason we adopt the following evidentiary rule: Malay evidence in respect to *h is accepted in the absence of or in conformity with Bandjarese evidence, but Bandjarese evidence prevails when it disagrees with Malay. The four correspondences are given below, distinguished by subscripts, with the original correspondence now assigned to *hi:

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

41

*ht: *ha(m)pir, Mai. (h)amper, Tag. hampil, To. ofi 'near'. *h1unus, Mai. (h)unos 'pull out', Tag. hu:nos 'moult', SeBs. hu:nos 'pull out', To. unuh-i 'draw out'. *h2: *h2arjin, Bjr. agin, Mai. atjen, Tag. ha:tjin 'wind', To. ayi 'blow'. *h2i(y)say, Bjr. iysarj, Mai. isarj, Tag. si:hay (metathesis) 'gills'; *h2a(rj)saij, Jav. atjsarj, Tag. ha:say] 'gills', To. aha 'meaty places on fish-head.' *h3: *h3a(n)dep, Bjr. ma-hadap, Mai. (h)adap 'to face', Jav. arep, Tag. harap 'front', To. qalo 'belly', qao 'front'. *h3a(m)pit, Mai. ([h)apet, Tag. ha:pit 'pressure', To. qapi-qapit-a 'press into something'. *h3anta(D), Mai. ter-hantar 'prostrate', Tag. hantad 'visible', To. qata 'appearance, vision, shadow'. *h4: *h4atur, Bjr. atur, Mai. (h)ator 'arrange', Tag. ha:to! 'counsel', To. qotu 'row, rank'. The very fact that so large a number of different correspondences has been found in this position exemplified by a small number of comparisons suggests that ultimately we can expect some reduction as further evidence is brought to bear. Dyen's reassignment of the 'laryngeal' correspondences agrees with the evidence of Formosan languages (in which q indicates a postvelar [and not a glottal] stop). That this was so had already been indicated in his Laryngeals (1953a: 50, fn. 2). However when these languages were examined more systematically, they were found to offer new problems. Dyen treats the Formosan reflexes in his "Formosan evidence for some new Proto-Austronesian phonemes" (1965a). Formosan languages generally reflect the reassigned *q in the same way initially and medially: *qaCey, Tag. qatay, Mai. (h)ati, To. qate, Pz. Pasai, BuS. xattad, kmiatai, Pu. ha{ai, Pa. Patsai, qatsai, Ru. atai 'liver'. *CaqiP, Tag. ta:qe, Mai. tahi, To. taqe, Ss. saPi, Th. OaqiP, Ami taPiP, Pu. (ahi, Pa. tsaPi, tsaqi, Ts. tPe, Kn. tai, Sr. ti:?i 'excrement'. *bituqen, PgTg. bitu:qin, To. fetuPu, Pu. vituhin, Pa. vituPan, vituqan, vicuPan, vicuqatt; *bintuqen, Ss. bintuPan, BuS. bintuxan 'star'. *[ ]utaq, Jav. Utah, At. m-utaq, Kv. m-otiP, Pz. m-utaP, Pa. q-em-ucaq, Ru. PotaP 'vomit'. *panaq[ ], Tag. pa:naq 'arrow', Mai. panah, To. kau-fana 'bow', Ss. manaP, Kv. maniP, Pz. pa-panaP, Ami mi-panaP, Pu. pjinjanaP, Pa. p/in/anaP, p/injanaq, Ru. wa-pana, Ts. puno, Kn. mwa-panaPu 'shoot'. The following discrepant reflexes were observed and given the provisional reconstructions *QU *Q2:

42

ISIDORE DYEN

*2i: *paQliC[ ], Tag. paqit, Bjr. pahit, Mai., Jav. pahet, BuS. ma-paPis 'bitter', Kn. ma-paPitsi 'sour'. *paQ1paQ1, Tag. plalla:paq 'spiny leaf', Mai. p/el/apah 'flattened bamboo', At. pahpah, Se. pappah 'flower', Ami papah 'leaf'; *pnQiepaQ1, BsE. pjaljaopa 'pulpy leaf', Se. paPeppah 'flower'. *Q2: *[dD]aRaQ2, Mai. darah, Se. da:ra, Pu. dalaP 'blood'. Apart from the suggestions in Dyen 1965a it is now possible to deal with a discrepancy observed in Dyen 1953a. In a number of comparisons otherwise assignable to *q Tongan exhibits nothing (or transition) instead of the anticipated q. Heretofore in initial position the only evidence for *q was the variable Malay h, and this h could be regarded as a secondary acquisition under the influence of doublet forms that arose in Malay from dialect mixture between an initial /z-retaining dialect and an initial hlosing dialect (cf. Dyen 1953:38). However in at least one of such comparisons (the first below) the Formosan languages agree in showing the same reflexes as for *q. Consequently we now assign to all correspondences showing Tongan q, as above, and to *q2 all correspondences showing Tongan nothing (or transition), as follows: *q2abuP[ ], Tag. qabo, Bjr. habu, Mai. (h)abu, To. efu, Ss. abo, Pz. Pabu Th. qafuP, Pa. Pavu, qavu, Pu. abu, Ami avu, BuS. xabu, Ru. abu, T s A u , Kn. abu, Sr. PabuPu 'ashes'. [It is difficult to dissociate this etymology from At. qabuli, Se. qabulit 'ash' whose initials are also compatible with *q.] *q2ijuy, Tag. qiloy, Bjr. hidurj, Mai. (h)idot), To. ihu 'nose'. *q2ipon, Tag. qi.pon, Jav. ipon 'gather'; and *qimpun, Mai. (h)impon, Jav. import 'gather', To. ipu 'container, cup'. *peraq2u, Mai. pera(h)u, Jav. pSrahu 'boat', To. folau 'flotilla, travel'. *suq2a(ttN), Jav. suwan 'digging stick', To. huo 'hoe, clear the field'. The proto-phoneme hitherto reconstructed as *h exhibits sibilant reflexes in Formosan languages. However not all of the languages agree in showing the same reflexes in all instances. To deal with this circumstance Dyen (1965a) suggested making a number of provisional reconstructions generally entitled *S: i.e. *Slt *S2, etc. [However it should be noted that the *S 4 was there reconstructed in error and is not different from *Si. The error apparently arose because of erroneous entries in Table 3 under *S4 for nonexistent Rukai, Mantauran dialect (RuMn.) cognates. The consequence in any case is that there are now only four (instead of five) other *5-type correspondences. Since *SA would be unoccupied, we install here as *S 4 what is there *S6. Furthermore, because of the manner in which the evidence has accumulated, it has not yet proved useful to amalgamate these *S reconstructions with the *h reconstructions above despite the fact that some of the different reconstructions are almost certainly the same phoneme. The difficulty is in making the assignment.] We give here the essentials of the available evidence:

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

43

*St: *qaSielu, Tag. ha:lo (from *haqlu after metathesis and loss of *e), SeBs. alho (metathesis), Mai. (h)alu, At. qasayu, Se. se.ru, Ss. Pasuyu, Ami asulu, Pa. Pasulu, Ru. asulu 'pestle'. *CaqiS1[ ], Tag. tahiq, At. slem/aqis, Se. sjmjaisi, Ss. s/um/a?is, Kv. t/omjaPis, Pz. sa-sais, Th. s/mlaqis, Bu. ma-taqqais, Ami mi-taPis Pu. t/im/aPi, Ru. wa-tsais, Ts. t/m/ePesi, Kn. t\um\ata-isi 'sew'. *[ ]iS^q, Tag. qi:hiq, Bu. isax, Ami isiP, Pu. isiP, Pa. isiq, Ru. isi 'urine'. *Siinaw, {? and/or *S1enaw), Tag. hinaw, HIBs. hinaw, hunaw 'wash', Se. s/im/inaw 'wash (clothes)', Th. sinaw-an 'wash (other than clothes)', Bu. ma-sinaw, Pa. slim/maw 'wash (utensils)', Ru. wa-sinaw 'wash (clothes). *tuqaS1, Mai. tu(h)a 'old', To. ma-tuqa 'parents', Th. tuqa-tuqas 'old'. *buS1ek[ ], Tag. buhok, Ss. bukis, Kv. boqas, Pz. bekos, Th. hukis, Ami vukis, Ts.fuPusu, Kn. bukusu, Sr. bukuu 'hair', cf. To.fuk-a 'cut hair'. *S2: *kaS2iw[ ], *kaS2uy, Tag. ka:hoy, Mai. kayu, At. kahu-niq, Se. qahu-ni, Ss. kahui, Pz. kahuy, Th. kawiP 'tree', Ami kasui 'firewood', Pu. kaui, Pa. kasiu, Ts. evi, Kn. kalu, Sr. kiyuPu 'tree'. *kuS2kuS2, Tag. kuko, Mai. kuku, Se. kukkuh, Ss. ka-k-l-ukuh 'nail', Th. kuku? 'claw', Bu. kuskus, Pa., Ru. k-al-uskus-an, Sr. P-al-uku 'nail'. *S3: *D1ewS3a, Tag. da-lawa (for *-wah from *-wha), Bis. duha, Mai. rua, At. rusa, Se. daha, Ss. rasa, Kv. Pu-zusaP, Pz. PdusaP, Th. tusaP, Bu. dusa, Ami tusa, Pu. dua, Pa. dusa, rusa, Ts. yuso, Kn. tsu:sa, Sr. suuwa 'two'. *SA: *SAriRup, Tag. hi:gop 'sip', Bjr. hirup, Mai. (h)irop 'lap up', Ulawa ilu, Ami herop, Pu. s-m-irap (with unexplained a) 'sip'. *S5: *Sseyup, Bis. hu:yop, At. yup, Kv. s-m-yop, Th. m-yop, Ami yof, mi-Piyup 'blow'. Not only did the hitherto *h show different reflexes in Formosan languages, but so also did some of what would have hitherto not been reconstructed at all, the so-called 'non-phoneme'. These reconstructions are divided into "^-reconstructions and the *H and *P reconstructions. They are exemplified as follows: : *xiapuy[ ], Tag. qapoy, Mai. api, At. pu-rtiq, Se. pu-niq, Ss., Pz. hapui, Th. Papuy, Bu. sappud, Pu. apui, Pa. sapui, Ru. apui, Ts. puzu, Kn. apulu, Sr. (h)apulu 'fire'. *x2: *x2epat[], Tag. qa:pat (from *x2a-x2epat), SeBs. upat, At. spayat, Se. seppat, Ss. spat, Kv. Pu-sapat, Pz. supat, Th. pat, Bu. pa:t, Ami spat, Pa. spat, spac, Ru. spat, Ts. suptu, Kn. soipatu, Sr. pa. tu 'four'. *X: *PiXu, Tag. qiyo 'thy, thee', At. isu, Se. issu, Kv. Pa-ysoP, Pz. Pisiu, Th. PihuP, Bu. suu, Ami k-isu, Pu. yu-yu, Pa. t-isu-n, Ts. su, Kn. i:ka-su 'thou'. [The other instances of *X cited in Formosan evidence are ambiguous with *S'1.] *H: *beffi[ ], Tag. ba-ba:Pi, ba-ba:yi 'woman', Ami va-vahiP 'wife', Pu. ba-

44

ISIDORE DYEN

bay-an 'wife, woman', Pa. va-vay-an 'woman', Ru. Pa-ba-bay 'woman, wife', cf. Mai. bjinji 'wife', Ngj. ha-b/in/ay 'feminine', To. //w/e 'woman'. *qumaH, Mai. huma 'cultivated field', At. qama-yah, Ss. um-umah, Ami umah, Pu. huma, Pa. Puma, quma, Kn. uma, Sr. (h)umu-uma 'field'. *baRaH[ ], Tag. ba.ga, Mai. bara 'ember', At. bagah, Ss. bayah, Kn. bara, Sr. baraPa 'charcoal'. *P: *kaPen[ ], Tag. ka:Pin, Mai. ma-kan, Se. mu-Pekkan, m-kan, Kv. k-dm-art, Pz. mAkAit, Th. ma-kan, Bu. maun, Ami k-im-aPn, Pu. mi-Pkan, Pa. k-im-an, Pim-an, Ru. wa-kan, Kn. k-um-a-unu, Sr. um-au 'eat'. *Penem[ ], Tag. qa.nim (from *Pa-Penem), Kv. Pu-nam, Bu. num, Ami, Pu., Pa., Ru. inim, Ts. nomu, Kn. num, Sr. k-numu 'six'. *kiTaP, Tag. ki:ta, At., Se. m-ita, Ss. k-um-ita, Kv. ma-ytaP, Pz. kitaP, Sr. k-um-a-kita 'see'.

SYLLABIC STRUCTURE

From the point of view of syllabic structure the words of Proto-Austronesian can be described as consisting of multiples of the sequence (C)V(C) where C is consonant (including semivowel), where V is vowel, and parentheses mean 'optionally present'. The formula implies: (1) vowel sequences are permitted; (2) initial and final vowels are permitted; (3) a cluster consists of no more than two consonants; and (4) clusters occur only between vowels. There is evidence for double consonants that has not yet been taken into account. For example one can compare Ilocano tekken 'pole' with Buginese tikkerj 'stick' and Ilocano panni:ki 'fruit bat' with Makassarese panniki 'bat'. All of these languages distinguish double from single consonants. Furthermore there are many other instances of similar comparisons between these languages that are geographically remote from one another. They suggest the hypothesis that double consonants existed in their common proto-language. However most of the evidence for double consonants is restricted to the position after *e. Thus one must consider the possibility that those comparisons which show double consonants after vowels reflecting *e do so by independent development of the double consonants after this vowel. In fact one can argue that Ilocano alternations support such an explanation: thus Ilocano punno 'full' (Bug. pSnno 'full') has the syncopated form na-pno 'full' after a vowel-final prefix and such an alternation is unlikely to have developed if it would have brought three consonants together. This argument would lay the burden of proof on those instances in which the evidence for a double consonant was not after *e. Since such instances (like 'fruit bat/bat' above) are quite rare, and may have a different explanation, the matter should perhaps be left indeterminate for the time being.

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

45

Furthermore, the evidence is restricted to the languages called Indonesian (or Hesperonesian). Thus even if the double consonants did not have an independent origin, the origin may have been in a change or changes that postdated Proto-Austronesian. It can be shown that the Tagalog contrast between long and short penult vowels before a single consonant corresponds long-to-long and short-to-short to the same contrast in Ilocano to a significant degree. An examination of the agreements in length of the penult vowel before single consonants showed an overwhelming majority of agreements over disagreements — 39 agreements to 6 disagreements in clear cases in a rough count that I made in the 200-word Swadesh basic vocabulary list; interestingly enough, nearly all of the disagreements seemed to lend themselves to explanations which would assign their difference to regular change (e.g. *baqeRu[ ], Tag. ba.go [from *baqgo], Ilk. baru (presumably with the loss of *q and contraction of the two vowels, though other developments would also be possible) or to analogical changes. Furthermore, this length or its absence plays a role in the morphology. Thus in both languages there is the following behavior illustrated from Tagalog. The penult length of a root shifts to a penult length of the suffixed form in the derivation of transients: e.g. bi:latj 'count', bila:rjin 'be counted'. Similarly such a penult length is absent from the suffixed form if it is not present in the unsuffixed one: qalis 'go away', qalisan 'be removed from'. Such an arrangement in a paradigm does not fit well with a hypothesis that the length was directly due to the loss of a phoneme and thus would suggest that the correspondence might continue a Proto-Austronesian distinction. However there is little or no evidence known or easily available outside of the Philippines that fits with the Tagalog-Ilocano penult length correspondence. Thus this agreement may be no more than an innovation of some Philippine languages. There is little question that the matter deserves thorough investigation, but there is no reason yet to assign the feature of penult length distinction to Proto-Austronesian. Dempwolff's Wortschatz (1934-38) is testimony to the fact that he found the overwhelming majority of reconstructed morphemes to be dissyllabic. Essentially this dominance led Chrétien in his exploratory statistical study of the Proto-Austronesian 'morph' (1965) to limit his investigation to dissyllables. There is however evidence that seems to point to the presence of an additional syllable at the end of many reconstructions that now appear to be dissyllabic. Dyen (1965a) used formulas which took this possibility into account to explain additional vowels appearing in Formosan words. Similar additional vowels appear in Sangir and in Makassarese and Buginese and are reported by Capell (1962:382) for the western Solomons, but no connection has been shown between these different phenomena. Milner (1965:428) has proposed that Proto-Austronesian may have had 'prenasalized voiced stops in free distribution and non-prenasalized voiceless stops also in free distribution', so that prenasalization would then be a 'phonetically distinctive feature' in initial as well as medial positions. In support of this hypothesis he quotes DempwolfF (1934-38:2.95): 'The Proto-Austronesian initials that appear in Ho(va)

46

ISIDORE DYEN

with reduced nasal clusters are generally here not augmented into facultative ( = optional) nasal clusters because a nasal cluster in initial position is found only in Ho., but otherwise hardly in any other language of Indonesia though it is found in Melanesian and Polynesian'. Dempwolff was there referring to his practice in reconstruction. When one cognate differs from another by reflecting a single consonant against the same consonant preceded by a nasal, it is Dempwolff's practice to enclose the nasal in parentheses in the reconstruction: e.g. *ka(m)pit, to be read *kapit and/or *kampit. When Merina ( = Hova) exhibits for example a -p- or -k- for *p and *k respectively, instead of the anticipated -/- or -h-, Dempwolff posits 'reduced nasal clusters' which he feels are needed 'since these sounds have not appeared hitherto as the regular correspondences of any sound of Proto-Austronesian' (2.91). He is forced to posit other 'reduced nasal clusters' as well. Not only that, but he finds that these are to be posited in initial as well as intervocalic position. At this point Dempwolff, in the sentence quoted by Milner, is stating that he will not introduce optional nasals (i.e. nasal in parenthesis) in the reconstructions concerning Merina cognates reflecting 'reduced nasal cluster'. The reason is given by Dempwolff a few paragraphs later where he shows that analogical changes from sandhi forms in compounds could explain the 'reduced nasal clusters'. Hence Dempwolff did not really see the Merina material as Milner suggests. Milner (1965:416), like Haudricourt (1956:328, see below), seems to be concerned about the 'proliferation' of Proto-Austronesian phonemes and additions to 'Dempwolff's already formidable inventory'. Actually Dempwolff's inventory is a very small one and can be regarded as large only in comparison to the languages of Oceania. There is no virtue in an inventory that is too small any more than in one that is too large. The consideration of simplicity is useful only in the comparison of hypotheses, not in intuitive judgments. Therefore one can only applaud Milner's inconsistency in proposing the addition of prenasalized stops to the Proto-Austronesian inventory even if one cannot agree that this is necessary. As Milner himself indicates, the presence of prehomorganic 'nasalization by accretion' in Jabem (1965:427 fn., attributed to Milke) strongly militates against his interpretation. THE APPLICABILITY OF THE COMPARATIVE METHOD TO THE AUSTRONESIAN LANGUAGES

The question has been raised as to whether the Austronesian languages are fit subjects for the application of the comparative method. Gonda (1940, 1952) suggests that there may be an essential difference between the application of the comparative method to families like the Indoeuropean or Semitic on the one hand, and to a family of languages like 'Indonesian' (== Austronesian) on the other — granted the temporarily almost unavoidable restriction of comparative study to vocabulary — because the languages belong to 'generally uneducated and illiterate peoples and therefore there is not as sharp a distinction between the affective and non-affective categories of words as can

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

47

be found in the other languages'. In principle the basis of the argument is that in 'the Indonesian field, where cultural and other circumstances are to a high degree uniform and languages are often closely related, the chances are that parallel developments have, comparatively speaking, taken place on a large scale' (1952:89). Without this basic assumption our concern with the division between the two types of vocabulary would not be great, for the simple restriction of the study to cognates associated by systematic correspondences and by non-affective meanings would be sufficient. Gonda appears to suggest, however, that if parallel developments are rife and words pass easily from the affective to the non-affective sphere, then pseudo-cognates might result in numbers which would invalidate the reconstructive inferences. His concern is based on an over-estimation of the chances of such events, but is no doubt affected by strong objections to the hypotheses of other scholars. In his discussion Gonda mainly attacks the etymological efforts of two scholars, Brandstetter and Dempwolff, but chiefly the former. Dempwolff receives only two glancing blows: the first is for associating with Toba topap 'hand-breadth', the Malay, tempap which Dempwolff glosses 'hand-breadth' and Gonda (presumably only) 'flapping against, clapping the hand on'; Wilkinson gives both glosses and one must conclude that Dempwolff was correct in associating the two words. The remaining objection could only be that the pair of words underwent a like, but independent semantic development from independently originated onomatopoeic words to a pair that mean 'hand-breadth' and now, because of the fit with other correspondences, constitute a pseudo-cognate pair. As a matter of general practice one could be ready to admit the possibility of an error in a cognation, but in estimating the probability of cognation in this case, it would appear difficult to conclude that the two words were more likely to be a pseudo-cognate pair than a cognate pair. Gonda's other reference to Dempwolff is to observe that Dempwolff's association of the two reconstructions *taykap and *tarjgap as 'Nebenformen' can be called 'variation', but not in the same sense as that found so commonly in derivations in languages such as French. His objection appears to concern only a matter of terminology. Gonda's main target appears to be Brandstetter's quite free-wheeling association of roots with similar meaning and one syllable in common under the rubric 'rootvariation': e.g. Mai. bintit, bintat, bintul, bentil, bintil 'pustule, pimple, gnat-bite, etc.'; bincut 'slight swelling on forehead'; bincul 'bump on forehead'; binjut 'inflamed bump on face'; benjol 'big inflamed bump' etc. Gonda is quite right in pointing out that 'to call these phenomena "root-variation" is not the same thing as to explain their origin and history'. Gonda's stricture was observed by Dempwolff (1934-38:1. 119ff.) before its enunciation. Dempwolff lists examples leading in the direction taken by Brandstetter, but excludes such comparisons because the hypotheses would then be based on abstractions derived from divided roots, and thus unlike the ones that he had been constructing to explain the data of the three languages of his first book. However, although it is evident that Brandstetter's method is probably too loose, there is some evidence for root-formation by repeating monosyllables and by affixing

48

ISIDORE DYEN

to monosyllables. One need only cite the following pairs reconstructed by Dempwolff with their meanings to show that it would be premature to abandon this line of investigation in view of how much is at stake: (1) *bunbun 'Sammeln', *tabun 'Haufen'; *kubkub 'Bedecken'; *ta{ij)kub 'Überstülpen, Zudecken, Deckel'; *pakpak 'Schlagen, Flügel Schlagen', *tapak 'Handfläche, Fusssohle'; (2) *qis{ae)p 'Aufsaugen', *sepsep 'Saugen'; *iket 'Drehen, Knüpfen', *keket 'sich Festhalten'; (3) *kelem 'Dunkelsein', *lemlem 'Düstersein'; *kembarj 'sich Blähen, Aufblühen', * bay bay 'Weitsein, Ausgebreitet' ; *kepel 'Kneten', *pelpel 'Anfüllen'; *ke(m)pit 'Zusammenhalten, Klemmen', *pitpit 'Klemmen'. Implicit here is the hypothesis that there was a period in pre-Austronesian that perhaps continued into Proto-Austronesian when monosyllabic roots were much more common than is evident in the daughter languages. Such a hypothesis would explain the monosyllabic root-forms that appear in Fiji mi 'urinate' (associated with *miRmiR) and vuk-ä 'mouldy' (associated with *bukbuk) much more simply than Dempwolff's hypothesis of the loss of a (repeating or reduplicating) syllable (2.140). However this type of reconstruction is internal rather than merely comparative and needs to be approached with great care if one is to reduce the false starts. It is not enough to show word families with the same syllable in single languages, as Brandstetter did. One procedure that suggests itself is to find sets of roots such that all parts are found to recur in combination in the manner exemplified above. Haudricourt (1965:328) finds that 'the blind application of the principle of regular correspondence leads to the proliferation of reconstructed phonemes and hence to the proliferation of accidental coincidences. Thus in 1951 I had thought that Dempwolff had confused under b- two separate phonemes that Javanese distinguished as w- and b- and that one should compare them with New Caledonian mb-, mbwand ygw- and so reconstruct labio-velars in Austronesian'. However it would be too much to hope that we shall not make errors; we can only direct our research so as to reduce the amount of error. The so-called proliferation of proto-phonemes is more apparent than real if one keeps in mind that reconstruction carries with it a probability factor that it in fact maps reality. We have not reached the stage of being able to estimate this probability factor in each case, but scholars do have an intuitive estimate by which they rate and compare different reconstructions. This intuition is dependent on the type and amount of data. Thus there are different degrees of confidence associated with the reconstruction of PAN *a and of *S3 regardless of the fact that both apply to particular correspondences. This does not result from the 'blind application of the principle of regular correspondences'. The structural interpretation of the history of languages — as exemplified in Haudricourt 1965 — is dependent on the gains made by the application of the comparative method. The structural interpretation gives a more elegant appearance, and is in this sense perhaps more appealing and should not be ignored, but it is perhaps more practical to apply this interpretation to comparative interpretations which are secure or at least relatively secure. For structural interpretation

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

49

must shift — even if the shift is extremely abrupt — to accommodate any change in the comparative inductions. The reason is simple. Comparative induction makes use of the law of regular phonetic change by which words can be mapped from any earlier stage from which they persist. Structural interpretation merely provides the simplest description of the changes in the system, but these are necessarily ad hoc constructions bounded only by competitive simplicity, not by any law of language change. Haudricourt's (1965:315) suggestion that it is perhaps premature to attempt to reconstruct Proto-Austronesian and that we should only attempt to reconstruct the proto-languages of the subgroups of highest order — Pro to-Indonesian, Proto-Formosan, Proto-Oceanic — is a counsel of despair. However like many desperate counsels it ignores part of the danger. Although following this suggestion will reduce the amount of language material that a scholar, if he restricts himself to a single subgroup, would have to deal with, he would have to work with some degree of uncertainty as to the proper subcollection that he should deal with. The subgroups are actually not well enough defined at this time to allow such a plan to be carried out systematically. Rather subgrouping and reconstruction should be carried on simultaneously. The lexicostatistical procedure can be used to obtain a likely subgrouping which, however, must be regarded as an approximation. Some of the errors in this approximation can be discovered in the course of reconstructing the phonology, for this will lay bare the mergers necessary to account for the changes from the phonological system of the proto-language to that of the daughter languages. As soon as it becomes feasible to deal with the various morphological-syntactic systems in a comparative way, most of the subgrouping will become well-determined. This is not to discourage attempts to reconstruct sub-proto-languages or mesolanguages. Though there may be disagreements as to membership in subfamilies, the process of collecting cognates is a prolonged one and every contribution to this process is useful. It would be quite difficult to imagine a scholarly endeavor to reconstruct a subproto-language that did not at the same time make a real contribution toward the reconstruction of the proto-language of highest order. At the same time however the meticulous accounting for reliable Proto-Austronesian correspondences can only promote our understanding of the developments which have resulted in the daughter languages. Highest order correspondences imply corrections or justifications of the results of the original lexicostatistical procedure and thus tend to make that data useful for more accurate subgrouping and for glottochronology on the one hand, and on the other imply discriminating innovations useful for subgrouping not only in phonological developments, but also in all other structural areas. These implications justify the major effort in this field. The solutions to the difficult and frustrating problems that arise serve not only to illuminate the history of the great Austronesian family itself, but also to deepen our understanding of genetic comparative linguistics and the comparative method which is its instrument.

ISIDORE DYEN

50

ABBREVIATIONS F O R N A M E S O F L A N G U A G E S

At. - Atayal; Bis. - Bisayan; Bjr. - Bandjarese; BsE. - Bisayan as recorded in Encarnación (1885); Bu. - Bunun; Bug. - Buginese; BuS. - Bunun, southern district; Fu. Futuna; Gr.-Graged; HIBs. - Hiligaynon; Ilk. - Ilocano; Jav. - Javanese; Kn. Kanakanabu, Kanabu; Kv. - Kavalan; Mai. - Malay; Mer. - Merina; Mny. Maanyan; NC - New Caledonian; Ngj. - Ngaju (-Dayak); OJav. - Old Javanese; Pa. - Paiwan; PaF. - Paiwan (as in Ferrell 1969); PAN - Proto-Austronesian; PgTg. Pagsanghan Tagalog: PIN - Proto-Indonesian; PMP - Proto-Malayopolynesian; Pu. - Puyuma; PuF. - Puyuma (as in Ferrell 1969); Pz. - Pazeh; RiPu. - Rikavong Puyuma; Ru. - Rukai; Sam. - Samoa; Se. - Seediq; SeBs. - C e b u Bisayan; Snd. Sundanese; Sr. - Saaroa; Ss. - Saisiyat; Tag. - Tagalog; Th. - Thao; To. - Tonga; Tr. - Trukese; Ts. - Tsou. BIBLIOGRAPHY 1 9 4 2 . Thai, Kadai, and Indonesian: a new alignment in southeastern Asia. AmA 4 4 . 5 7 6 - 6 0 1 . . 1966. Austro-Thai. Behavior Science Notes 1.227-61. New Haven, Conn. . 1967a. Austro-Thai studies: material culture and kinship terms. Behavior Sciences Notes 2.203-44. . 1967b. Austro-Thai studies: 3. Austro-Thai and Chinese. Behavior Science Notes 2.275-336. BOPP, FRANZ. 1 8 4 0 . Über die Verwandtschaft der malayisch-polynesischen Sprachen mit den indisch-europäischen. Abh. Preuss. Akad. Wiss. Phil. u. Hist.

BENEDICT, PAUL.

Abh.,

169-246.

J. L . A . 1 8 8 4 . Bijdrage tot de vergelijkende klankleer der westerse afdeeling van de Maleisch-Polynesische taalfamilie. Utrecht. BRANDSTETTER, RENWARD. 1893. Die Beziehungen des Malagasy zum Malaiischen. Luzern. . 1902. Tagalen und Madegassen (Eine sprachvergleichende Darstellung als Orientirung für Ethnographen und Sprachforscher). Luzern. . 1906. Ein Prodomus zu einem vergleichenden Wörterbuch der malaio-polynesischen Sprachen für Sprachforscher und Ethnographen. Luzern. . 1908. Mata-Hari. Luzern. . 1910. Wurzel und Wort in den indonesischen Sprachen. Luzern. . 1911a. Gemeinindonesisch und Urindonesisch. RBMIS 8. . 1911b. Sprachvergleichendes Charakterbild eines indonesischen Idiomes. Luzern. . 1912. Das Verbum: dargestellt auf Grund einer Analyse der besten Texte in vierundzwanzig indonesischen Sprachen. Luzern. . 1916. An introduction to Indonesian linguistics. Translated by C. O. Blagden. London.

BRANDES,

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

51

. 1937. Die Verwantschaft des Indonesischen mit dem Indogermanischen. Wir Menschen der indonesischen Erde, II. Luzern. CAPELL, ARTHUR. 1943. The linguistic position of southeastern Papua. Sydney. . 1962. Oceanic linguistics today. CAnthr 3.371-428. CHRÉTIEN, C . DOUGLAS. 1965. The statistical structure of the Proto-Austronesian morph. IPLS 243-70. CONANT, CARLOS EVERETT. 1910. The RGH law in Philippine languages. JAOS 31. . 1913. The pepet law in Philippine languages. University of Chicago dissertation. Also in Anthropos 7.920 ff. (1912). CRAWFURD, JOHN. 1852. Grammar and dictionary of the Malay language. London. DAHL, OTTO CHR. 1938. Le système phonologique de proto-malgache. N T S 10.189235.

. 1951. Malgache et maanjan: une comparaison linguistique. Studies of the Egede Institute 3. Oslo. DEMPWOLFF, OTTO. 1934-38. Vergleichende Lautlehre des austronesischen Wortschatzes: 1. Induktiver Aufbau einer indonesischen Ursprache, ZES 15 (1934); 2. Deduktive Anwendung des Urindonesischen auf austronesische Einzelsprachen, ZES 17 (1937); 3. Austronesisches Wörterverzeichnis, ZES 19 (1938). Berlin. DYEN, ISIDORE. 1947. The Tagalog reflexes of Malayo-Polynesian D. Lg 28.227-38. . 1949. On the history of the Trukese vowels. Lg 25.420-36. . 1951. Proto-Malayo-Polynesian *Z. Lg 27.534-40. . 1953a. The Proto-Malayo-Polynesian laryngeals. Baltimore. . 1953b. Dempwolff's *R. Lg. 29.359-66. . 1953c. Review of Dahl (1951). Lg 29.577-90. . 1962. Some new Proto-Malayopolynesian initial phonemes. JAOS 82.214-15. . 1964. The position of the Malayo-Polynesian languages of Formosa. Asian Perspectives 7.261-71. Tucson, Arizona. . 1965a. Formosan evidence for some new Proto-Austronesian phonemes. IPLS 285-305. . 1965b. A lexicostatistical classification of the Austronesian languages. IUPAL Memoir 19, supplement to IJAL. . 1970. [See p. 54.] EGEROD, SOREN. 1966. A statement of Atayal phonology. Artibus Asiae, Luce Volume, 120-30. Ascona. ELBERT, SAMUEL. 1946. Yap-English and English-Yap wordlists: with notes on pronunciation and grammar, (mimeo.) n.p. ENCARNACIÓN, J. F . FÉLIX DE LA. 1885. Diccionario Bisaya-Español. Tercera edición. Augmentada ... por J. Sanchez. Manila. ESSER, S. J. 1938. Talen. Atlas van Tropisch Nederland, p. 9. FERRAND, GABRIEL. 1924. Les langues malayo-polynésiennes. Les langues du monde, ed. by A. Meillet and M. Cohen, 405-55. Paris.

52

ISIDORE DYEN

FERRELL, RALEIGH.

1969. [ S e e p . 5 4 . ]

FINCK, FRANZ NIKOLAUS. 1909. Die S p r a c h s t ä m m e des Erdkreises.

Leipzig.

Fox, C.E. 1947. Phonetic laws in Melanesian languages. JPS 56.58-118. FRIEDERICI, GEORG. 1912-13. Wissenschaftliche Ergebnisse einer amtlichen

For-

schungsreise nach dem Bismarck-Archipel im Jahre 1908. 2. Beiträge zur Völker- und Sprachenkunde; 3. Untersuchungen über eine melanesische Wanderstrasse. Mitteilungen aus den deutschen Schutzgebieten, Ergänzungshefte 5 (1912), 7 (1913).

GABELENTZ, G. VON DER. 1885. Einiges über die Sprachen der Nicobareninsulaner. Berichte der kgl. sächs. Ges. d. Wissensch., phil.-hist. kl. 37.296-307. GONDA, J. 1940. Opmerkingen over de toepassing der comparatieve methode op de Indonesische talen, voornamelijk in verband met hun woordstructuur. Bijd. 99.397-466. . 1952. The comparative method as applied to Indonesian languages. Lingua 1.86-101. GOODENOUGH, W . H . 1961a. The Willaumez languages of New Britain. Presented to Xth Pacific Science Congress, Honolulu, (mimeo.) . 1961b. Migrations implied by relationships of New Britain dialects to Central Pacific languages. JPS 70.112-36. GRACE, GEORGE W. 1959. The position of the Polynesian languages within the Austronesian (Malayo-Polynesian) language family. IUPAL, Memoir 16, supplement to IJAL. . 1964. The linguistic evidence. Movement of the Malayo-Polynesians : 1500 B.C. to A.D. 500, by Kwang-Chih Chang, George W. Grace, and William G. Solheim II. CAnthr 5.359-406. GREENBERG, JOSEPH. 1957. Essays in linguistics. Chicago. HAUDRICOURT, ANDRÉ G . 1951. Variations parallèles en Mélanésien. B S L 4 7 . 1 4 0 - 5 3 .

. 1965. Problems of Austronesian comparative philology. IPLS 315-29. HENDON, RUFUS S. 1964. T h e reconstruction of *-ew in Proto-Malayo-Polynesian.

Lg 40.372-80. HERVAS Y PANDURO, LORENZO. 1784. Catalogo delle lingue. Vol. 17 in his I d e a

dell'Universo, 21 vols. In Cesena, completed 1787. HINLOOPEN LABBERTON, D. VAN. 1924. Preliminary results of researches into the original relationship between the Nipponese and the Malayo-Polynesian languages. J P S 33.244-80. HUMBOLDT, WILHELM VON. 1836-39. U e b e r die Kawi-Sprache auf der Insel Java.

Berlin. JONKER, J. C.G. 1914. Kan men bij de talen van den indischen Archipel eene westelijke en eene oostelijke afdeeling onderscheiden? Verslagen en Mededeelingen der Kon. Akad. v. Wetenschappen. 4de reeks. 12.314-417. KEANE, A. H. 1880. On the relations of the Indo-Chinese and inter-Oceanic races and languages. JRAI 9.254-89.

THE AUSTRONESIAN LANGUAGES AND PROTO-AUSTRONESIAN

53

1889. Taalkundige gegevens ter bepaling van het stamland der MaleischPolynesische volken. Verslagen en Mededeelingen der kon. ak. van wetenschappen, afdeeling Letterkunde. 3de reeks, dl. 6 . 2 7 0 - 8 7 (reprinted in Verspreide geschriften 6 . 1 0 4 - 2 0 , 's-Gravenhage, 1917). . 1906. Taalvergelijkende verhandeling over het Aneityumsch, met een aanhangsel over het klankstelsel van het Eromanga. Amsterdam. KERN, R. A. 1940. De Soendasche ö-klank. Bijd. 99.111-18. KIECKERS, ERNST. 1931. Die Sprachstämme der Erde. Heidelberg. MARSDEN, W M . 1830. Miscellaneous works. London. MILKE, WILHELM. 1965. Comparative notes on the Austronesian languages of New Guinea. IPLS 1.33(W8. MILNER, G.B. 1963. Liquid consonants and the relationship of Polynesian to Austronesian languages. BSOAS 2 6 . 6 2 0 - 3 1 . . 1965. Initial nasal clusters in Eastern and Western Austronesian. IPLS 1.416-30. OGAWA, N . , and E . ASAI. 1935. The myths and traditions of the Formosan native tribes. Taihoku. RAY, SIDNEY HERBERT. 1926. A comparative study of the Melanesian island languages. Cambridge. RELAND, HADRIAN. 1708. Dissertatio de linguis insularum quarundam orientalium. Hadriani Relandi dissertationum miscellanearum pars tertia et ultima, 55-139. Trajecti ad Rhenum. SCHMIDT, WILHELM. 1899. Die sprachliche Verhältnisse Oceaniens (Melanesiens, Polynesiens, Mikronesiens und Indonesiens) in ihrer Bedeutung für die Ethnologie. Mitt. anthrop. Ges. Wien 2 9 . 2 4 5 - 5 8 . . 1906. Die Mon-Khmer-Völker ein Bindeglied zwischen Völkern Zentralasiens und Austronesiens. Braunschweig. . 1926. Sprachfamilien und Sprachenkreise der Erde. Heidelberg. TSUKADA, MATSUO. 1967. Vegetation in subtropical Formosa during the Pleistocene glaciations and the Holocene. Palaeogeography, Palaeoclimatology. Palaeoecology 3.49-64. Amsterdam. TUNG, T'UNG-HO. 1964. A descriptive study of the Tsou language, Formosa. Taipei. TUUK, H. N. VAN DER. 1865a. Outlines of a grammar of the Malagasy language. JRAS 1.419-42. . 1865b. Note on the relation of the Kawi to the Javanese. JRAS 1.442-6. . 1872. 't Lampongsch en zijne tongvallen. Tijdschrift v. Indische Taal-, Land-, en Volkenkunde 18.118-56. UHLENBECK, E . M . 1955-56. Review of Dyen 1953a. Lingua 5.308-18. VANOVERBERGH, MORICE. 1933. A dictionary of Lepanto Igorot or Kankanay: as it is spoken at Bauco. St. Gabriel, Mödling bei Wien. KERN, H . A .

54

ISIDORE DYEN

1930-32. Vergleichendes Wörterbuch der indogermanischen Sprachen, ed. and reissued by Julius Pokorny. 3 vols. Berlin and Leipzig. WILKINSON, R. J. 1932. A Malay-English dictionary (Romanized). Mytilene. WULFF, KURT. 1934. Chinesisch und Tai: Sprachvergleichende Untersuchungen. D. kgl. Danske vidensk. selskab. Hist.-fil. Medd. 20.3. . 1942. Über das Verhältnis des Malaio-Polynesischen zum Indochinesischen. Det kgl. Danske vidensk. selskab. Hist.-fil. Medd. 27.2. (posthumous). WALDE, ALOIS.

Note: There is furthermore evidence for a D 3 (drawn to my attention by Shigeru Tsuchida). Ferrell (1969), using sources dialectically different from those cited above, lists PuF. rua?, RiPu. towa, 30\va, PaF. rusaP 'two'. The two languages involved show different reflexes in the following comparison: *D3 anaw, Tag. da:naw, la:naw 'pool', Jav. danu (with the difficult final vowel), Mai. danau, Sam. lano 'lake', RiPu. danao 'pond', PaF. dänau 'lake'. ADDITIONAL BIBLIOGRAPHY

'Background noise' or 'evidence' in comparative linguistics: The case of the Austronesian-Indo-European hypothesis. Indo-European and Indo-Europeans: papers presented at the Third Indo-European Conference at the University of Pennsylvania, ed. by G. Cardona, H . M . Hoenigswald, and A. Senn, 4 3 1 - 4 0 . Philadelphia. FERRELL, RALEIGH. 1 9 6 9 . Taiwan aboriginal groups: Problems in cultural and linguistic classification. Inst, of Ethn. of Ac. Sinica Monograph 17. Taipei.

DYEN, ISIDORE.

1970.

I N D O N E S I A A N D MALAYSIA

E. M. UHLENBECK

1.

INTRODUCTION

Although the history of the study of Indonesian languages is still to be written, some features of its development are clearly discernable. As in other parts of the non-Western world the study of the languages of Indonesia was initiated by foreigners. First mainly out of mercantile interests, later out of missionary fervor, still later also led by political motives, Europeans reached the Indonesian Archipelago from the sixteenth century onwards in gradually increasing numbers. The nineteenth century witnessed an enormous spread and intensification of Western influence over large parts of Africa and Asia. In that century the study of language also became a concern of trained scholars. It is necessary to bear in mind, however, that as far as Indonesia goes this holds true only for relatively few languages. For many of them, data collected by travellers, traders, missionaries, or administrative officials to this very day remain the only material available. As the area where Indonesian languages are spoken came within the sphere of influence of different Western powers, different traditions of description, terminology, and spelling developed. A clear instance of this is the British and Dutch traditions of the spelling of Malay. Different regions of linguistic study came into being between which there was sometimes little scientific contact. It was thought natural that it was the French who occupied themselves with the Indonesian languages of Madagascar, that the British turned to Malaya and to the languages of Sarawak and North Borneo, that the Dutch kept to the languages of the Dutch East Indies, while the eastern part of the isle of Timor was tacitly assumed to be the concern of the Portuguese. Each colonial power seemed to have its own linguistic hunting grounds. Accordingly, it was thought a matter of course that the subsequent results of the various linguistic activities were published in the language of the colonial power in question. As the Dutch occupied a large share of the territory where Indonesian languages are spoken, most of the publications on these languages were written in Dutch. As this language is accessible only to a restricted number of scholars, this practice jeopardized international communication and cooperation. It is not an exaggeration to state that owing to this language barrier up to the Second World War much of the often valuable work done in the field of Indonesian linguistics remained largely unknown. Only within a very

56

E. M. UHLENBECK

small circle of non-Dutch specialists was it recognized that a reading knowledge of Dutch was a prerequisite for anybody planning to enter the study of Malay or any of the other more or less intensively studied Indonesian languages. European centers such as the School of Oriental and African Studies in London or the École des langues orientales vivantes in Paris required such a knowledge, but only very few scholars reached the point that they could fully utilize Dutch sources. Aside from the fact that many and important data were gathered by the admirable endeavors of amateurs, and aside from the relative isolation in which especially the Dutch tradition of Indonesian language study grew up, there are other characteristic features which ought to be mentioned here briefly. In Western Europe, in the whole of the nineteenth century and in the first four decades of the twentieth century the center of the study of language, the emancipation of linguistics as a science with its own aims distinct from philology began around the middle of the nineteenth century. On the periphery, this emancipation took place much later. Far into the present century the study of what was often vaguely called Oriental languages was rarely pursued for its own sake. It usually served a wider purpose. It often formed only a part, albeit an important and even essential part, within a broader context of scholarly activity. The attention of many scholars was directed towards what may be called a general historically-oriented understanding of non-Western peoples and cultures. In other words there was a much stronger tendency towards areal specialization than towards specialization by discipline. The boundaries of one's scientific work were largely geographically determined. In general one did not become a linguist or an archaeologist, but one strove to become an Orientalist, that is, an all-round scholar who considered it his duty to keep abreast of what was done within his area in a great variety of disciplines: linguistics as well as philology, but certainly also history, sometimes also archaeology and ethnology, or if circumstances demanded it, even folklore and musicology. A good example of this type of scholar is Winstedt, of whom John Bastin in his introduction to the Festschrift for Sir Richard Winstedt, 'the last and greatest of British colonial scholars', correctly remarks: Winstedt's 'interests are not limited to one or two fields, but cover almost the whole range of Malay culture: language, history, economics, arts and crafts, law and religion' (Malayan and Indonesian studies, Essays presented to Sir Richard Winstedt, p. 9 [1964]). To a certain extent this attitude was reflected by the nature of the preparations deemed necessary in order to become a specialist in Indonesian studies. These preparations as they would normally take place in Leiden, from 1875 onwards the main center of Indonesian studies in The Netherlands, consisted in the first place of three years' study of Sanskrit and Arabic. Because of the Indian influence and the no less important Islamic influence on the cultural history of Java and of some islands in its immediate vicinity, these two languages were considered to furnish an indispensable foundation for the subsequent study of Indonesian culture which took up another three or four years. Within the years between the two world wars, the period of the

INDONESIA AND MALAYSIA

57

greatest flowering of Dutch scholarship in Indonesian studies, it had become the rule that after the completion of his academic training the young Orientalist got a research position in the Dutch East Indies, either by entering into the service of the Dutch colonial government as language officer or by being engaged by The Netherlands Bible Society or other missionary societies. Within the Dutch East Indies the Batavia Society of Arts and Letters, already founded in 1778 and thereby being the oldest learned society in the southern hemisphere, acted as the center of Dutch Indonesian scientific activities by publishing a learned journal and a rapidly growing series of transactions and by building up a central library. Regular meetings were held where progress reports of research conducted in various parts of the Archipelago were discussed. Fully in accord with this development of special areas of research was the fact that only exceptionally did scholars from other nationalities take part in Indonesian studies. Not only did the Dutch keep to their own territory, only rarely trespassing into areas occupied by the British or the French, but also very few scholars from other nations attempted to undertake linguistic research in the Dutch East Indies. It is significant to note that in 1942 only two institutions in the whole of the United States offered a course in an Indonesian language (Malay), the University of Michigan (see Senstius 1943) and Yale University (personal communication, Professor Dyen). Although in general among Indonesians a certain interest in their own languages seems to be a not uncommon phenomenon, it was only in the twentieth century that important contributions to the scientific study of Indonesian linguistics and philology were made by Indonesians (Hoesein Djajadiningrat, Poerbatjaraka, Prijono, Prijohoetomo, Tjan Tjoe Siem). They became fully steeped in the Dutch Orientalistic tradition, receiving their training in The Netherlands, as nowhere in Indonesia did any possibility exist for obtaining academic training in this field. Only shortly before the outbreak of the Second World War was this deplorable situation amended somewhat by the founding of a faculty of letters in Batavia. Within the last twenty-five years the picture given above has markedly changed. There are two developments which stand out very clearly: in the first place the internationalization of Indonesian studies, and, secondly, an increasing interest of Indonesians in the study of their national language, the Bahasa Indonesia, and to some degree also in the Indonesian regional languages. The Second World War had a profound effect on the study of foreign languages in the United States. Intensive language courses were established virtually overnight. Within the framework of the Army Specialized Training Program alone hundreds of Americans got practical instruction in strategically important Far Eastern languages. Indonesian languages, and among them in the first place modern Malay, received a share of this interest which arose out of the exigencies of war. From about 1950 onwards American students came to Indonesia sometimes as teachers of English but at the same time trying to do linguistic and ethnographical fieldwork on Indonesian

58

E. M. UHLENBECK

languages and cultures. Conversely, a number ©f Indonesian students went to centers of language study in the United States and in this way became acquainted with modern trends in linguistics. Within the last fifteen years the universities of Yale, Cornell, and Berkeley have become prominent centers of study of Indonesian languages, culture, and history, but also at other universities (Indiana, Michigan) there is a growing interest in this field. Other countries followed suit. In Australia the National University in Canberra, the University of Sydney, and Monash University in Clayton, Victoria developed programs of Indonesian studies, while other universities started to offer courses in Bahasa Indonesia. In spite of Japan's complete loss of political influence after the war, Japanese scholars did not diminish their attention to Indonesian languages and culture. In India a heightened interest can be observed especially in those sections of Old Javanese literature where Indian influence is of paramount importance. Also, scholars from Germany, Russia, and allied countries of Eastern Europe turned to the study of Bahasa Indonesia and in a few cases to other Indonesian languages as well. The position of Dutch scholarship was seriously affected by the war. About onethird of those engaged in fieldwork in Indonesia lost their lives between 1940 and 1945. Owing to the continuously deteriorating relations between the Republic of Indonesia and The Netherlands fewer and fewer Dutch scholars were in a position to remain in Indonesia, although a few of them continued to hold teaching posts at the University of Indonesia up till the middle fifties. Finally, except for a few missionaries, no Dutch scholars were able to do linguistic fieldwork in Indonesia. In 1955 the Batavia Society ceased publication. A few years later it was dissolved. Since 1967, the future for Dutch scholarship in Indonesia seems to have brightened as new ways and means are being devised for renewed economic and cultural cooperation between the two countries. In The Netherlands, however, a remarkable renaissance took place around 1950. Various circumstances made this possible. In the first place nearly all Dutch scholars who had previously been engaged in Indonesian studies and who had survived the hazards of war, gradually returned to The Netherlands, sometimes bringing with them still unpublished results of fieldwork done during their stay in Indonesia. Several of them got positions in The Netherlands as well as abroad. Secondly, there was a general feeling among them of being under the obligation not only to make the past results of Dutch scholarship in Indonesia internationally known, but also to continue the Dutch tradition of Indonesian studies. More and more the opinion gained ground that in spite of the temporary difficulty of access to Indonesia, Dutch scholarship was still in a unique position because of its rich libraries, its unsurpassed manuscript collections, and its easy access to the existing scientific literature. Moreover, a number of scholars with long experience gained in the field were able to transfer much of their knowledge to a younger generation. The venerable Royal Institute of Linguistics and Anthropology founded in Delft in 1851, took the lead. From 1950 onwards it displayed markedly increased activity.

INDONESIA AND MALAYSIA

59

Its journal was modernized and publication in English was stimulated. Thanks to generous support from the Government the Institute was able to embark upon an ambitious program of translation which was aimed at making the most important results of Dutch Indonesian studies more internationally accessible. To this same end a series of critical bibliographies was initiated in 1955 which offered expert guidance to those newcomers in the field who lacked the opportunity of consulting the existing literature. At the same time its series of transactions was rapidly expanded. In the years which followed the Japanese occupation linguistic interest among Indonesians was of necessity focused on the many and diversified problems related to the further development of their national language, the Bahasa Indonesia, a term for Modern Malay introduced in the Dutch East Indies around 1930. Technical vocabularies had to be created, dictionaries and grammars for use in school had to be written, and textbooks and manuals for a great variety of subjects had to be either newly composed or translated from other languages. These highly urgent tasks of applied linguistics doubtless led to a certain awareness of the importance of the scientific study of language. Several journals devoted to practical problems of language were founded in Indonesia: Medan Bahasa (1951-59), Bahasa dan Budaja (1952-), Pembina Bahasa (1948-57); and in Malaya, where a parallel development could be observed: Dewan Bahasa (1957-) and Bahasa (1957-). In Brunei the journal Bahana was founded (1966). Of course they were all written in the national language. However, as few trained linguists were available, the general level of the contributions stayed well below that of a scientific journal. Only recently have successful attempts been made in Djakarta to found a scientific journal devoted to the study of Indonesian culture, Madjalah Ilmu-IImu Sastra Indonesia (1963-), which may be considered a continuation of the journal of the former Batavia Society. While some of the journals previously mentioned are entirely devoted to Bahasa Indonesia, such as Pembina Bahasa, other journals also accept articles on other Indonesian languages. The term Indonesian languages is ambiguous. It may refer to languages spoken in the Republic of Indonesia; it may also be taken in a (rather vague) linguistic sense. In that case it refers to the Austronesian languages spoken in Madagascar, Formosa, the Philippines, and in Indonesia. In the present survey we will use the term in a somewhat different and more restricted sense. For practical reasons we will not discuss the languages of Madagascar, Formosa, and the Philippines, nor the languages of New Guinea. On the other hand we will include in the discussion of Malay and Indonesian ( = Bahasa Indonesia) the Malay spoken in the Malayan Peninsula. This implies that we will treat successively: 1) Malay and Bahasa Indonesia, 2) Sumatra languages, 3) Borneo languages, 4) the languages of Java and Madura, 5) Celebes languages, 6) the languages of the islands east of Celebes and Java. Attention^will also be^paid|to the non-Austronesian languages in the same area. These include: non-Austronesian languages of Malaya, of North Halmahera, Alor, and Timor. In conjunction with the term 'studies', the word Indonesian is also ambiguous in a

60

E. M. UHLENBECK

different way. Indonesian studies may also refer to studies by Indonesian scholars. As hardly any linguistic studies by Indonesians related to other languages than those spoken in Indonesia have appeared thus far, we do not have to bother with this interpretation of 'Indonesian studies'. We have mostly confined ourselves to studies which appeared after World War II. In some cases it appeared useful to mention older publications as well, especially if more recent work stays far behind in importance compared to results published earlier. The following sections could not have been written if we did not have at our disposal the many bibliographies on South-East Asian studies which have recently appeared and which also testify to the heightened international interest in that part of the world. A great help have been, in the first place, four 'critical bibliographies' issued by the Royal Institute of Linguistics and Anthropology: P. Voorhoeve on Sumatra (1955), A. A. Cense and E.M. Uhlenbeck on Borneo (1958), A. Teeuw on Malay and Bahasa Indonesia (1961), and E.M. Uhlenbeck on Java and Madura (1964). For Celebes it was possible to make use of the materials collected by J. Noorduyn for the bibliographical survey of the languages of that island which will appear in the same series. That the linguistic bibliography published by CIPL from 1947 onwards has frequently been consulted, goes without saying. In some cases the bibliography of Kennedy (2nd ed. 1955), the bibliographical notes regularly published in Oceanic Linguistics (1963-), Damais' bibliographical notes (1955-64), and the second edition of Capell's linguistic survey of the South-Western Pacific (1962) proved useful. For Malaya we consulted Cheeseman's bibliography of Malaya (1959), with the supplements which have appeared since the year of its publication. For the linguistic publications of Flores and adjacent islands we have profited from Goris' survey in Bahasa dan Budaja (1956).

2.

MALAY — BAHASA INDONESIA

Since 1945, the year that the Bahasa Indonesia was officially declared the national language of the Indonesian nation, the study of the variegated language phenomena for which the terms Malay, Malay dialect, and Bahasa Indonesia have come into use, has made considerable progress. Within the compass of this article it hardly seems possible to list and to evaluate all the work that has been done in the field of Malay studies within the last two decades. Moreover, it is unnecessary, because Teeuw's bibliography (1961) provides an extremely detailed and virtually complete survey from the seventeenth century to the present day. However, while his guide will remain an indispensable aid for all future students of Malay and Bahasa Indonesia, there seems to be room for a more succinct exposé of the main linguistic advances recently made, omitting everything of secondary value and adding what has been published since Teeuw's survey appeared. We will discuss these advances under the following five headings: 1) the history of Malay, 2) Malay

INDONESIA AND MALAYSIA

61

dialects, 3) grammatical studies, 4) lexicography, 5) the growth and development of the national language in Indonesia and in Malaysia.

2.1

The History of Malay

Our knowledge of the historical development of Malay is still very limited. A certain knowledge has been gained concerning a language which probably correctly has been called Old Malay, but the relation between this Old Malay and the oldest literary Malay known thus far, is still far from clear. Moreover the position of this literary Malay among the rich proliferation of Malay dialects and Malay-like languages spoken not only in Malaya but also in large areas of central Sumatra and of Borneo is yet another unknown chapter in the history of Malay. The term Old Malay has come into existence for the language used in a few inscriptions discovered in Sumatra and dating from the second half of the seventh century. After Coedes' fundamental publication on these inscriptions (1930) further contributions on this Old Malay were made mainly by Aichele (1942-43), De Casparis (1956), Teeuw (1959), Coedes (1964), and Boechari (1966). Aichele devoted an imaginative, if on some points rather speculative, article to the relation between this inscriptional Malay and Old Javanese, suggesting possible influence from Old Batak. De Casparis materially widened the scope of the study of Old Malay by his ingenious decipherment of several Old Malay inscriptions of a somewhat later date, one of them found not in South-Sumatra but in Kedu (Java). Teeuw gave a valuable appraisal of their work, surveying the scant materials (i.e. other Malay inscriptions from the fourteenth century) at present available for the description of the historical development of Malay. By its informative character and balanced judgment his article forms an excellent introduction for anyone who wants to get acquainted with this branch of Malay studies. Coedes gave a new interpretation of the Kedukan Bukit inscription (1964), while Boechari reported on an Old Malay inscription found at Sodjomerto in East-Java (1966). A contribution to our knowledge both of sixteenth century Malay and of Malay influenced by Arabic was made by Drewes, who prepared an edition of a sixteenth century manuscript containing a Malay version of an Arabic panegyric (1955). His short article on the history of the Malay conjunction bahwasanja should also be mentioned in this context (1950). Mention should also be made of Bausani's successful attempt (1960) to furnish a more correct interpretation of several items of the famous wordlist of Pigafetta of 1522 by a renewed study of the only Italian manuscript. Hamilton (1947) discussed the first Dutch-Malay dictionary printed in The Netherlands in 1598. Linehan (1949) provided a survey of Old Malay wordlists and dictionaries. Several studies were devoted to Malay loanwords. Bausani carefully examined Persian loanwords (1964) while Santa Maria did the same for Malay borrowings from

62

E. M. UHLENBECK

Portuguese (1967). Skinner wrote on Arabic loanwords (1964), while Fähnrich compared Arabic loanwords in Indonesian and Georgian (1966). 2.2 Malay Dialects Malay dialectal studies have been hampered for a long time by normative tendencies observable among British as well as among Dutch scholars. Instead of accepting the data as they were, they applied the in itself already doubtful yardstick of so-called Classical Malay (the Malay of the Sejarah Melayu and other writings), condemning instead of studying all 'deviations' from that norm. Of the dialects of Perak, Kelantan, and Trengganu spoken in the Malay Peninsula, some new data were provided by Brown (1956), although his book mainly consists of reprints of three dialect studies previously published by him before the war. Firth (1946), Hill (1951,1952), Berwick (1953), Taib Osman (1964), and Nik Safiah bt. Hj. A. Karim (1966) furnished a few new data on the dialect of Kelantan. The most important advance in the knowledge of the Malay dialects in the Malay Peninsula is Hendon's monograph on the Ulu Muar Malay, spoken in the Kuala Pilah District in Negri Sembilan (1966). For the dialects of Malay spoken in Sumatra and Borneo, I refer to the paragraphs devoted to these islands and for a more detailed survey to the critical bibliographies by Yoorhoeve (1955), and Cense and Uhlenbeck (1958). For the so-called Djakarta Malay, a very valuable publication is Muhadjir's structural description in the journal Madjalah Ilmu-Ilmu Sastra Indonesia (1964), one of the still very rare publications written by linguistically well-informed Indonesian scholars. No less welcome is Kähler's dictionary of the same dialect (1966). On Chinese Malay hardly anything of interest has been published. Nio Joe Lan's article (1960) is of little value. Of Malay as spoken in East Indonesia very few sources can be given. The dictionary by Van Pernis (1950) contains a number of words in Amboina Malay. Teutscher (1954) published some material on the Malay spoken in West New Guinea, while the Kapauku dictionary of Miss Doble (1960) also contains here and there Malay words in use in the same area. Suhartono (1956) commented on Bahasa Indonesia as spoken in Kupang (Timor), while Taulu (1957) furnished some data on Bahasa Indonesia in Menado, and Udinsah (1958) on Indonesian as spoken in the Batjan area. 2.3

Grammatical Studies

Since the well-known extensive grammars of Gerth van Wijk (1889), Spat (1900-01), and Van Ophuysen (1910), a long series of grammars, grammatical introductions, and textbooks have been published. It is doubtful whether these publications really contributed much to a better insight into the structure of Malay or Bahasa Indonesia.

INDONESIA AND MALAYSIA

63

As to the more recent grammars written in Dutch, those by Fokker (1941, 4th ed. 1950), Emeis (1946,4th ed. 1950), and Mees (1927,4th ed. 1946) were most widely used. Of these three the one written by Fokker is easily the best, being least blinded by traditional suppositions and being based on excellent materials. Kahler's grammar of Bahasa Indonesia (1956), meant as a scientific description, is rather disappointing, since it does not take into account the progress in descriptive techniques made in the centers of linguistic research since the war. On Verguin's recent attempt at a functional and structural analysis of Indonesian (1967) it is also not possible to comment favourably. Kahler's and Verguin's works have been severely criticized by Teeuw (1959 and 1968). Grammatical introductions of various scope and aims were written by DempwolfF (1941) in German and by Teselkin and Alieva (1960) in Russian. Textbooks and manuals written in Dutch, in Bahasa Indonesia, and in other languages as well appeared in great quantities, such as in Dutch by Croes, Duin, and Van Dijck (1949), and Rambitan (1949), in English by Dyen (1945), Mrs. Pino (1950), Schmidgall Tellings (1957), and more recently by Lie (1965, 1966), Sarumpaet and Mackie (1966), and Poerwanto Danoesoegondo (1966), in German by Mrs. HilgersHesse (1956), and Poetzelberger (1965), in Czech and English by Oplt (1960, 2nd ed. 1966), in Indonesian by Takdir Alisjahbana (1949-50), Poedjawijatna and Zoetmulder (1955), and Slametmuljana (1956-57), of which the last one is undoubtedly the most original although serious objections were raised against it by Teeuw in a very critical review (1960). A list of the more elementary textbooks is found in Teeuw's bibliography (1961:158-71). The main linguistic contributions by Dutch scholars since 1945 were made by Gonda, Roolvink, Fokker, and Teeuw. Of the many articles by Gonda on Indonesian linguistics several are of importance for Malay in spite of the drawback that they are not the result of fresh observation. They are mostly based on existing sources. Perhaps Gonda's most important contribution to our knowledge of Malay is his article on the parts of speech in Malay in which he attempted to apply the structural approach developed by De Groot (1949). Gonda's short article on the Malay and Javanese verbal auxiliaries (1943) written in opposition to the views of Berg on the same subject, is also worth mentioning. Roolvink's monograph on the prepositions (1948) is mainly of value because of its rich collection of material. Its theoretical basis is questionable. His more recent analysis of the verbal forms with the prefixes per-, ber- and memper- (1965) is of importance also for the problem of the development of Malay to Bahasa Indonesia. Fokker's introduction to syntax (1951) is a courageous attempt to free himself of preconceived ideas, but from the point of modern structural linguistics his theoretical approach is not beyond reproach, as was pointed out by Anceaux in a review in BijdrTLV (1952). Teeuw's short article in the Festschrift De Groot (1962) devoted to the morphology of the Indonesian adjective, together with his study on duplication in Bahasa (1962-

64

E. M. UHLENBECK

63), may be considered as a first step towards a morphological description along structural lines. The publications of Emeis are of much less importance. His thesis on the Malay verb (1945) must remain inconclusive, being based on superseded linguistic notions. The same may be said of his later articles (1948, 1953, 1955). Bijleveld's thesis on doubling phenomena (1943) based on written sources is of value only as a collection of material. Some contributions by Indonesian scholars have appeared which show the impact of modern structural thinking. Beside the recent study by Muhadjir of DjakartaMalay (1964) already mentioned in the previous section, Samsuri's description of the Indonesian sound system (1960) is probably the most interesting; but also the short article by Kamil and Sugeng Sukarsono (1960) on some productive morphemes in Bahasa Indonesia, Kamil's balanced review of the semantic analysis of the system of personal pronouns by Krupa and Altmann (1961), Tang Tjia Han's discussion of the prefix ter- (1960), and more recently the contributions by Kamil (1964), Ramlan (1964, 1965) and Harimurti Kridaleksana (1964) are promising signs of modern Indonesian scholarship in the field of linguistics. Of particular interest is the set of principles for language study in Indonesia developed by Kamil and Moeliono, more or less modeled on Bloch's set of postulates for phonemic analysis, published in the journal Medan Ilmu Pengetahuan (1961). Another very productive Indonesian scholar in the field of Indonesian linguistics is Umar Junus. The quality of his writings, however, does not yet meet international standards, in spite of the fact that they show a certain familiarity with the international linguistic literature. Recent studies by other scholars which deserve to be mentioned are the short publications by Verguin (1955) on stress in Indonesian, by Von Essen (1959) and Bjelkina (1965) on sentence intonation, by Altmann on the phonemic structure of the morpheme in Indonesian (1967) and on the structure of the Malay pantun (1963, with Stukovsky, 1965). Mention should also be made of Denzel Carr's short article on expressive, quasi expressive, and non-expressive words in Indonesian (1966). In our opinion by far the most important recent advances in the description of Bahasa Indonesia have been made by the Russian scholar Andrejev (1957) and by the British linguist Payne (1964). Andrejev's very succinct description of the phoneme system seems to us essentially correct, although a few points merit further investigation. Payne's monograph on Malay syntax satisfies modern demands of structural linguistics. It makes use of freshly collected data (obtained from an informant from Negri Sembilan), and it is based on an explicitly formulated theoretical framework of description. It is to be hoped that this thesis, written under the direction of Robins, which is now available only in stencilled form, will soon become more widely accessible in a printed edition. Judging from recent surveys by Spitzbardt (1964a) and by Parnickel and Sirk (1965), several Russian scholars have made valuable contributions to the study of Indonesian: Teselkin (1959), Alieva (1961a+b, 1963a-c), Sirk (1964), Andrejev and Uchanova

INDONESIA AND MALAYSIA

65

(1959). After her unpublished thesis on the verb system of Indonesian of 1963 (to which we had no access) Alieva continued to publish on various aspects of Indonesian morphology and syntax (1964a and b, 1967). More recent contributions by Russian scholars are Pavlenko's report on phonetic research on Indonesian consonants (1967) and Demid'uk's article on reduplication (1967). That recent developments in linguistic theory are also reflected in Malay and Indonesian studies is shown by some articles by Spitzbardt (1964a and b, 1967), which show the influence of generative grammar, and probably also by the article of Andrejev and others (1958) to which we had no access.

2.4

Lexicography

A great number of dictionaries of widely different degrees of reliability has appeared since 1945. Valuable efforts in this field have been the Indonesian-Dutch dictionaries by Van Pernis, and by Poerwadarminta and Teeuw which were both published by the same publisher in the same year (1950). These two dictionaries may be said to complement each other in several respects, Van Pernis paying more attention to dialectal terms, Poerwadarminta and Teeuw's dictionary supplying the vocabulary used in literary Malay. The best Indonesian-English dictionary is the second edition of the dictionary compiled by Echols and Shadily (1963), but the dictionary by Mrs. Pino and Wittermans (1953) is also quite reliable. I have not had access to an Indonesian-English dictionary prepared by B.K. Lathief, apparently published in Japan, according to Teeuw (1961). Useful is the handy-sized Malay-English and English-Malay dictionary compiled by Abdul Rahman bin Yusop (1964). Winstedt's many dictionaries (see 1948, 1949, 1952, 1958, 1960, 1963, 1966), which are partly reprints or later editions of older work, are mainly based on the Malay as spoken in the Malayan Peninsula, but in later editions material from Bahasa Indonesia is also included. I am not in a position to have an opinion on the large Malay-Japanese dictionaries which appeared in Japan during the Second World War, Miyatake (1941, 1943) and Taketomi (1942). The extensive Indonesian-German dictionary recently published by Karow and Mrs. Hilgers-Hesse (1962) adds little to what is found in other dictionaries. The Indonesian-German dictionary by Kahlo and Barwinkel (1963) is far from reliable. Poerwadarminta's monumental Kamus Umum Bahasa Indonesia (1952, 3rd ed. 1961) is the best Indonesian-Indonesian dictionary in existence, easily surpassing the dictionary of Muhammad Zain (1954) which was severely criticized by Prijono in the journal Bahasa dan Budaja (1958). A fourth edition of Poerwadarminta is in the process of being published; the first part came out in 1967. An Indonesian-Russian dictionary based on Poerwadarminta's dictionary was compiled by Korigodskij,

66

E. M. UHLENBECK

Kondraskin, and Zinov'ev (1961). For information on the meaning of the hundreds of abbreviations now in current use, the books by Koesnodiprodjo (1954-55), Schmidgall Tellings (1955), and Adinegoro (1965) are indispensable. Poerwadarminta also published a Bahasa Indonesia-Javanese and a JavaneseBahasa Indonesia dictionary (1942, 3rd ed. 1945-48) while Satjadibrata compiled a Sundanese-Bahasa Indonesia dictionary (1944, 2nd ed. 1950).

2.5

Growth and Development of the National Language in Indonesia and Malaya

Since October 28,1928, the day on which the famous slogan 'one country, one people, one language' was raised, the development of Malay in Indonesian society, its capacity as a general vehicle of communication for all Indonesians, its position in relation on the one hand to the regional languages ('bahasa daerah' now called 'bahasa nusantara') and on the other hand to Dutch, became hotly debated issues. Much of what has been written on these problems is now completely antiquated. However, part of it retains its value and deserves to be mentioned here, because it may serve as a useful introduction to the highly complicated linguistic situation of Indonesia. In the first place, then, the inaugural lectures of Drewes (1948) and Teeuw (1955) are of importance. Drewes' lecture is an excellent, highly informative essay on the many varieties of Malay and the development of the ideal of one language for the whole of Indonesia. Teeuw's somewhat more technical lecture dwells longer and in a little more detail on the developmental problems of the Bahasa Indonesia. Fokker's inaugural lecture delivered in Djakarta (1951) discussed the relation between Bahasa Indonesia and the regional languages. His plea for a multilingual society with Indonesian as a general vehicle of communication mainly for intellectual purposes drew forth a strong and negative reaction in the journal Pembina Bahasa Indonesia from Takdir Alisjahbana, the brilliant and indefatigable champion of the Bahasa Indonesia as the only acceptable language for modern Indonesian society. Fokker's reply was again answered by Takdir Alisjahbana, who a few years later discussed the same subject again (1954). Other publications on the position and the development of Indonesian are the inaugural lectures of Berg (1951) and Slametmuljana (1959). Useful for the general reader not acquainted with Dutch or Indonesian are several introductory essays by Takdir Alisjahbana (1949, 1956, 1962) and Mrs. Hilgers-Hesse (1960, 1965). At the international conference on the modernization of languages in Asia organized by the Malaysian Society of Orientalists in the fall of 1967 several papers were read on various aspects of the problems of standardization and modernization of Indonesian. A printed edition is planned of the proceedings of this conference, at present only available in stencilled form. In one of his three papers delivered at this conference, Takdir Alisjahbana repeated his plea for a prescriptive grammar of Indonesian, a plea

INDONESIA AND MALAYSIA

67

previously made at the conference on linguistic problems of the Indo-Pacific Area held in London in 1965. To design the host of new technical and specialist terminologies needed for scientific and professional discourse and, further, to introduce them into society, was one of the most pressing and gigantic tasks of applied linguistics which faced the Indonesian authorities. It is true that some spadework had been done during Dutch rule in this respect, especially by the very meritorious efforts of Balai Pustaka (the Bureau for popular literature founded in 1908 in the capital), but what was needed now was language engineering on a quite different scale and with much more ambitious aims. A short history of what has been done during the period of Japanese occupation and in the following years can be found in section 31 of Teeuw's bibliography (1961). It stands to reason that the creation of new terms, a task for which much inventiveness and linguistic insight is needed, did not follow a carefully mapped-out course. The circumstances demanded that the urgent tasks of applied linguistics had to be solved by people whose linguistic training had not prepared them for such work. It cannot be denied that the many lists of terms {istilah) which have been published thus far (the first five as supplements to the journal Medan Bahasa, Nos. 6-59, the last thus far published, as supplements to the journal Bahasa dan Budaja) represent an enormous effort on the part of the sixteen committees which have been entrusted with this task. A number of technical dictionaries (Ramus Istilah) have been published and many more are in the making. At the symposium on Indonesian language and literature organized by the Lembaga Bahasa dan Kesusastraan in Djakarta in 1966 Harimurti Kridaleksana gave a paper on the coining of technical vocabularies in Indonesian preceded by a short survey of what was done in this field since the Second World War. The proceedings of this symposium have been edited by Miss Achadiati Ikram and Lukman Ali (1966). A step towards the unification of the spelling of Indonesian to be used in Indonesia as well as in Malaysia was the publication (in stencilled form) of the detailed recommendations made by the committee on spelling of the Lembaga Bahasa dan Kesusastraan headed by A. Moeliono (Edjaan Baru Bahasa Indonesia 1966). Problems of spelling still keep attracting the interests of many scholars as may be seen from the Djakarta symposium of 1966 and from the proceedings of the Kuala Lumpur conference of 1967, both mentioned above. Problems of bilingualism or rather of multilingualism in Indonesia have not attracted much interest from linguists thus far. The only publications in this field so extremely important for Indonesian linguistics seem to be the short articles by Winarno Surachmad and Murray Thomas (1961), and Le Page (1964). More extensive research on bilingualism in West-Java is being conducted by the Dutch linguist C. D. Grijns. Several journals have been founded in which problems of grammar and spelling are discussed. Some of them have been discontinued, others still appear regularly. They are clearly focused on Bahasa Indonesia, on its grammar, on its position in relation to Malay, on problems of standardization and language-policy and — as always and

68

E. M. UHLENBECK

everywhere — on problems of spelling and etymological questions. There is much interest in practical problems of language, problems which are of the highest importance for the further development of Indonesian society, but the linguistic ability to discuss them is often absent. It is not surprising that the scientific value of the contributions to the three journals Bahasa dan Budaja (1952), Medan Bahasa (1951), and Pembina Bahasa (1948) the only journal completely devoted to Indonesian, is on the whole not impressive. Nevertheless they often offer interesting reading because they testify to a keen and general interest in problems of applied linguistics. As we already mentioned in our introduction, the new journal Madjalah Ilmu-Ilmu Sastra Indonesia (1963) is of an altogether different level. From the eight issues published thus far, one may conclude that it aims at becoming the scientific journal where problems of language, culture, and history can be discussed on a more detached plane and according to modern scientific standards. The development of Malay to the status of national language of Malaya, a status which it had reached in 1967, is on the whole parallel with what took place in Indonesia, although there are some important differences. In the first place, in Malaya the contrast Malay: regional languages does not exist. Secondly, in Malaya the competition of English and Chinese still constitutes a problem, while in Indonesia Dutch has ceased to be a serious rival. The best way to become informed on current opinions on language problems in Malaya is to study the reports of the Third Malay Language Congress in Singapore in 1956, and the proceedings of the conference of 1967 previously mentioned, and still better to read the journal Bahasa founded by the Association of Malay students (1957), and the journal Dewan Bahasa which started publication in Kuala Lumpur in the same year. Both journals contain a number of short introductory articles on various aspects of linguistics, mainly written for the interested layman. In some others grammatical problems of Indonesian are discussed, see for instance the valuable series of articles by Asmah Haji Omar (1962, 1962-63, 1963-64, 1967) and Liaw Yock Fang (1967, 1967-68). Both scholars gave papers at the International Conference on the modernization of languages in Asia (1967). Asmah Haji Omar discussed 'the role of dialects in the modernization of languages', while Liaw Yock Fang introduced 'the transformational-generative approach to Malay syntax'. Asmah Haji Omar also began to contribute to journals outside Malaya (1968). Each issue of Dewan Bahasa gives a list of Malay equivalents of English technical and scientific terms (Istilah). The Dewan Bahasa dan Pustaka which is responsible for these lists, has also started publishing a series of technical English-Malay dictionaries (Siri Istilah DBP, I960-). A useful introduction for the literary development of Malay in Malaya is Taib bin Osman's contribution to Wang Gungwu's recent monograph on Malaysia (1964). As in Indonesia, the predominant linguistic interest is in the national language. However a few studies devoted to the non-Austronesian languages have recently become available. Iskandar Carey (1961) published data on the Temiar language,

INDONESIA AND MALAYSIA

69

according to Pinnow (1963) one of the dialects belonging to the Northern Sakei or Senoi-group in the jungle areas of Perak and Kelantan, while a stencilled research report on the Bahasa Semang was recently made by Asmah Haji Omar (1963). The monograph of Mrs. Lewis (1960) gives the scant data on the Selung language spoken by sea gypsies who frequent certain islands of the Mergui archipelago (some texts, a wordlist). This language is classified by Salzner (1960) as one of the Orang Laut dialects of Malay, although he also lists Mergui Malay among the Malay dialects of the Riau-Lingga area. Its position is still uncertain, although there seems to be general consensus among specialists as to its Austronesian character.

3. SUMATRA LANGUAGES

The languages of Sumatra are Achehnese, Gayo, Batak (in a great variety of quite different dialects), Minangkabau, Malay (also many dialects), Rejang, and Lampung. On the islands east of Sumatra (Rapat, Bengkalis, Padang, Rangsang and Tebingtinggi, the Riau-Lingga archipelago, and finally Bangka and Belitung) Malay dialects are spoken. The languages of the islands west of Sumatra are: Simalur and Sikule or Sichule, both on the island of Simalur, Nias on the island of the same name and on the Batu Islands, Mentawai on the Mentawai Islands, and finally much farther south Enggano on the island of Enggano and on four smaller islands in its vicinity. Since 1940 no new contributions of any importance have been made to our knowledge of ACHEHNESE after the fundamental work of Snouck Hurgronje (1893, 1900, and 1906) and Hoesein Djajadiningrat (1934). Cowan's article on the relation of Achehnese to the Mon-Khmer languages (1948), being of a comparative nature, will be discussed in the article in this volume devoted to comparative linguistics. Two unpublished master theses recently defended at the Faculty of Language and Art of the Institute of Educational Science of Malang, Indonesia, by Soemadi Nitiwijoto ("Atjeh (Takengon) phonemes") and by A. Siahaan ("Phonemic analysis of Atjeh (Pidie)") were not accessible to me. For GAYO the situation is hardly different. Hazeu's large dictionary with short grammatical introduction (1907) is still the only reliable source. Plans to publish the Gayo folktales collected by Hazeu and kept in the Leiden University Library have not yet been realized. As to the numerous BATAK dialects, the most important postwar publications are Neumann's revised edition of the Karo-Batak dictionary by M. Joustra (1951) and Wismar Saragih's translation of the New Testament in Simalungun-Batak (1953). H.G. Tarigan published some texts in Karo-Batak (1965), and in Simalungun-Batak with translation in Bahasa Indonesia in the Indonesian journal Bahasa dan Budaja (1958, 1958-60, 1959, 1960a).pln the same journal J.D. Poerba wrote a short article on this dialect (1960). In the journal Medan Bahasa a few grammatical articles of little importance appeared on Simalungun, Karo-Batak by Tarigan and Unggas (1956,

70

E. M. UHLENBECK

1957) and Edmont-Tes (1957). Texts in the Toba-Batak dialect were given by De Boer (1947) and Bartlett (1952), while Fischer and Van Renselaar wrote on Batak kinship terminology (1959, 1966). K. Percival tried to give a concise description of the Batak sound system (1960). Tarigan continued his work on Karo-Batak. After his Karo-Batak phonology (1960b) he published (in stencilled form) a short grammar (1966a) and a description of the Karo-Batak root morpheme (1966b). To the unpublished Malang thesis of Silitonga ("Some transformations in Batak") we had no access. In spite of these publications, Van der Tuuk's Toba-Batak grammar, his dictionary, and his reader which appeared a full century ago (between 1860 and 1867) have lost nothing of their value. Therefore, the English translation of Van der Tuuk's grammar, unquestionably his chef d'oeuvre, which is scheduled to appear in 1970 in the Translation Series of the Royal Institute of Linguistics and Anthropology, will be extremely welcome, especially as the original Dutch edition is now hard to obtain. Since the dictionary of M.Thaib (1935) nothing has been published on MINANGKABAU, except for two short articles by Umar Junus, one on the pronominal system (1956) and another on the spelling of Minangkabau (1959) and an introduction to the language by Roesli written in Indonesian (1967). Two unpublished Malang theses by Koestimah ("The phonemes of Minangkabau") and by Rapilus ("Minangkabau transformations") and the University of Chicago thesis of G. E. Williams ("Colloquial Minangkabau") were not accessible to us. In this case too, two nineteenth century publications (Van der Toorn 1891, 1899) are still indispensable sources for this language. According to Voorhoeve (1955), a case might be made for considering Minangkabau a language of its own, distinct from Malay, but there seems no sufficient reason to assume that the dialects for which the term Middle Malay has been coined constitute a language different from Malay. Although there are considerable gaps in our knowledge of these dialects, it seems the best policy to assume that over a large area in Central Sumatra Malay dialects are spoken which more or less closely resemble each other. Since the days of Helfrich (1904, 1915, 1921, 1927, 1933) hardly any new data have become available. On the Malay as spoken in Deli, Roolvink gave a short description (1953). The same scholar edited some Kubu texts collected by Saragih, that is, texts in another Malay dialect (1950-51). The Rejang Ka-ga-nga texts recently published by Jaspan (1964,1967) are not texts written in the Rejang language, but texts in dialectal Malay written in Rejang script. Three unpublished Malang theses by Soeseno Kartomihardjo ("Komering phonemes"), by Amin Rianom ("Some transformations in Komering") and by A. R. Sofjan ("Phonemic analysis of Ogan") could not be consulted by us. As to LAMPUNG, the only known publications of the last twenty-five years are a few short articles by Honzali (1955), Noormali (1956), and M. W. El Lampuny (1958a + b, 1959). The scholar who did much to further the knowledge of the island languages west and east of Sumatra is the German H. Kahler. In his rather traditional descriptive work Kahler appears little influenced by structuralist thinking. Initially he followed

INDONESIA AND MALAYSIA

71

the tradition established by Dempwolff. In his more recent publications he mainly acknowledges his indebtedness to Jespersen's Philosophy of grammar. Most of his descriptive publications are the result of fieldwork conducted shortly before World War II. After his SICHULE texts (1940a) he gave a grammatical sketch (1955) and a wordlist of the same language in which Sichule words are compared mostly with their Nias and Simalur equivalents. This practice is related to Kahler's view that Sichule is a 'Mischsprache' (mixed language), at least from a diachronic point of view, showing features partly shared with Nias, partly with Simalur. On his concept of a mixed language Kahler wrote short articles in Anthropos (1942-45) and in Lingua (1965). O n Simalur Kahler published a Simalur-German dictionary with German-Simalur index (1961) and a collection of texts with German translation preceded by a short grammatical introduction (1963). After the war, Steinhart continued the publication of his extensive collection of texts from Nias (1950-51, 1954), which he had begun in 1934. Two short articles on the grammatical structure of Nias were recently published by Patsch (1964a+ b). The language of Enggano has probably become extinct recently. Kahler published a short grammar (1940b) and a collection of texts in this language (1954-64). A pupil of Kahler, Willms (1956) attempted an analysis of MENTAWAI using the older description of Adriani (1928) and the textual materials previously furnished by Loeb (1929a + b) and Morris (1900). The data on various Malay dialects spoken on the islands east of Sumatra, collected by Kahler in 1939, were published by him first in two articles in Anthropos (1946-49) and later in a more extended form in a monograph (1960).

4. BORNEO LANGUAGES

The present linguistic situation of Borneo is still very imperfectly known, even to the point that it is not yet possible to list all the languages and to determine their exact location. This is caused in the first place by the paucity and the poor quality of the existing data. Only very few qualified linguists have occupied themselves with the languages of Borneo (Kalimantan). It is missionaries and civil servants to whom we owe most of our data. In the second place, the data were collected in quite different periods. As the linguistic situation is highly fluid owing to constant migration and immigration, it is often not quite possible to find out how far data from older, nineteenth century sources are still relevant to the present situation. Moreover, confusion is often created by the use of general descriptions such as 'river people' etc., and by the indiscriminate introduction of a variety of language names. This situation is further complicated by the following features: (1) The penetration of Malay dialects over a large area, not only along the coastal areas but also along the rivers deep into the interior. These dialects include: BANJARESE spoken in and around Banjarmasin and in the Hulu Sungai area, and probably also

72

E. M. UHLENBECK

Sea Dayak or Iban, a language which has spread in recent times over all five divisions of Sarawak, and is also spoken in Brunei and in an extensive adjacent area of Indonesian Borneo, and Mualang spoken along the Ayak and Betilang Rivers, right hand affluents of the Kapuas River. Other languages sometimes indicated by a separate name (such as Labu, the language of a group dwelling on the Lasan River) may turn out upon closer inspection to be better classified among the Malay dialects. (2) The increasing influence of the Bahasa Indonesia owing to its present status as the national language of Indonesia. (3) The presence in Northern Borneo of Philippine languages (such as Illanun, Bulud Upi, and Sulu) and languages of a Philippine-like type (such as North Borneo Murut and Dusun). (4) The presence of large Buginese-speaking settlements along the coasts, especially in Kutai, Pasir, and Pegatan in the southeast corner of Borneo. (5) The presence of Chinese dialects in some coastal areas and along a few of the great rivers. Since in our survey (Cense and Uhlenbeck 1958) the existing literature on the Borneo languages has been examined exhaustively, it is superfluous to give again a detailed account of the data available for each of the languages which probably have to be distinguished. Here we will restrict ourselves to indicate the languages which according to Borneo standards have been best studied and on which fresh information has become available. The order in which the languages are discussed is largely the same as the one followed in our survey of 1958. Of the various MALAY DIALECTS Banjarese, Sea Dayak, and Mualang are the ones which are relatively the best known. Our knowledge of BANJARESE mainly rests on some unpublished texts noted down by the late W. Kern and on an unpublished, rather extensive wordlist made by the same scholar, a list supplemented by Van Pernis and Ras. The list is kept at the Leiden University Library. In the periodical Medan Bahasa, M. Sanjoto and Is. M. Jusran furnished some grammatical notes (1957), while Rachmat Marlin published a page of Banjaresepantun in the same journal (1955). Ras' excellent edition of the Chronicle of Bandjarmasin (1968) furnishes many new data on Banjarese. SEA DAYAK, or IBAN, has become known through the dictionary of Howell and Bailey (1900 and 1909), to which a supplement recently was published (1961). This dictionary is completely superseded by the recent dictionary compiled by the British linguist and phonetician N. C. Scott (1956), who also contributed a number of observations on the phonetics of the language (1957). Recently Asmah Haji Omar gave as an M.A. thesis a short grammatical description (1964), while Saudin gave the text with translation of the Iban mourning chant (1966). To the Iban reader published by Burr Baughman (1963) we had no access. MUALANG has been studied by Father Donatus Dunselman ( 1 9 5 4 , 1 9 5 5 , 1 9 5 9 , 1 9 6 1 ) . Especially his edition with Dutch translation and notes of Kana Sera, a long sacral chant of pregnancy, deserves to be mentioned not only as a linguistically important source, but also because of its literary value.

INDONESIA AND MALAYSIA

73

For KENDAYAN DAYAK Father Dunselman also furnished extensive data (1949-50). The closely related LAND DAYAK has become somewhat better known by a short article by Scott on the phonetics of the final nasal consonants in that language (1964), and by Geddes' survey (1954) and his eminently readable Nine Dayak nights (1957), which, although in the first place an ethnographical account, contains some textual material with translation. R. Nyandoh deserves to be mentioned here because of his publications of LAND DAYAK texts (1956, 1959). It is to be hoped that the translation of the seven Land Dayak texts published by him in 1963 will be followed by the texts themselves. New information concerning the languages previously indicated by the name LAND DAYAK was provided by Court in two short articles (1967a + b). His description of the phonological structure of the word in Mentu Land Dayak, a language distinct from the Bukar-Sadong Land Dayak studied by Scott, will be published in the series of Oceania Linguistic Monographs. The collection of folktales which according to A.A.Hill (see JMBRAS 28.148) was published by Father P. Howes in 1935 was not accessible to me. On the many MELANAU dialects no fresh information has been published since we finished our survey. The same is true for BISAYA (not to be confused with Philippine Bisaya!). A first attempt was made by Mrs. Shirley P. Lees to give a preliminary phonemic analysis of LUN DAYA or LUN DAYE, one of the SARAWAK MURUT dialects spoken in the Pa Kemaloh or Pa Kemalu area (1959). The only new material of the North Borneo Murut published since 1938 is Prentice's recent article on the verb system of the Timugon dialect of Sabah Murut (1965). Some new data on the language and the location of the TAMBUNAN DUSUN has been furnished by Williams (1962) who has done fieldwork in Sensuran, a Dusun community in the Tambunan-district of the North Borneo Interior Residency. His article of 1962 contains a phoneme inventory and a list of kinship terms. A detailed report on the structure of Tambunan Dusun dialects may be expected to be published by Williams in the near future (see note 1 to his article of 1961 in the Journal of American Folklore). On another Dusun dialect some data are given by E. Majusim bin Majinal (1947). By far the most important recent publication is Father Antonissen's Kadazan dictionary, which is preceded by a short grammatical introduction (1958). His dictionary and grammar are based on data collected in Penampang and adjacent villages. The Dusun reader published by the Borneo Evangelical Mission (1964) as well as the field dictionary made by G.N. and L.W.B. Appell (1961) were not accessible to me. Some comparative data on a number of Dusun dialects were recently furnished by Miss B.M.Clayre (1966-67a), who also made a valuable contribution to the study of Dusun syntax (1966-67b). On BAJAU, a text with a few grammatical observations was published by Abdul Ghani bin Bagul (1950). More recently Sather published some new data on the language (1964-65,1967). The results of fieldwork conducted by Mr. and Mrs. Pallesen of the Summer Institute of Linguistics (1965) have not yet been published.

74

E. M. UHLENBECK

No new data of any importance have come to my notice on UBIAN, ILLANUN, and BULUD-UPI, nor on the TIDUNG and BASAP dialects or on LABU. Hardaker published an introduction to SULU(K) (1963) and a wordlist of that language (1964). As to the Central Borneo languages a short BARAM KENYAH text with English translation was published in the Sarawak Museum Journal by its indefatigable editor T. Harrisson (1961). Galvin gave some Baram Kenyah songs with English translation (1966). A first description of the phonemic system of the Uma Pu dialect was provided by Cubit (1964) and of the Lepu' Tau dialect by Lees (1965). In the same issue of the Sarawak Museum Journal (1961) Urquhart commented on the wordlist of Douglas (1911) comparing his vocabulary items of SARAWAK KENYAH with his own. Two years earlier he had supplied a comparative list of thirty-five kinship terms in KAYAN UMA POH (PU), noted down with the assistance of 'Baram District People and the sepengs of Kapit district' (1959). In Leach's report on Social Science Research in Sarawak (1950), some Kenyah kinship terms are found. Of all the languages of Central Borneo BUSANG is the best known owing to the valuable and extensive dictionary of Barth (1910) which is preceded by a short grammatical introduction. On PUNAN BA, PUNAN, and PENAN Needham tried to shed some light in a series of short publications which appeared between 1954 and 1959. Recently Tuton Kaboy published a wordlist (1965). No new information was published on MODANG and EMBALOH. MAANJAN (also called SIHONG), a language spoken in the region of Tamianglayang, has become internationally known as the name of a Borneo language owing to Dahl's monograph (1951), in which he tried to make plausible an especially close genetic relation between this language and Malagasy. To this end Dahl tried to describe Maanjan as accurately as the relatively scarce data supplied by Sundermann in 1913 and found in the existing translations of Christian literature permitted. Dahl did not make use, however, of the translation of the New Testament made by the missionary Gerlach, which is now kept in the office of the Netherlands Bible Society in Amsterdam. Recently Kiting and Dullah supplied some data on the language (1958). Of the other languages of South Borneo, LAWANGAN, TABUYAN, NGAJU, and OT DANUM, Ngaju is by far the best known. It may be even considered the best known of all the languages of the island. There is a long tradition of study of this language which goes back to 1852, when the German linguist H.C. von der Gabelentz compiled a short grammar, making use of translations of Biblical stories which at that date had already been published. The Protestant mission has tried to make Ngaju the language of the Church. Still the most important sources for the language are Hardeland's extensive grammar and dictionary which came from the press more than a century ago (1858, 1859). Dempwolff, after having published an article on nasals and'nasal clusters in Ngaju (1922), included Ngaju in his well known comparative study (Dempwolff 1937:52ssq.). Dempwolff's assumption of an old speech stratum in Ngaju was rightly refuted by Dyen (1956). On the so-called sacred 'language', the Basa Sangiang, many data are found in

INDONESIA AND MALAYSIA

75

Scharer's highly valuable thesis (1940) of which an English translation made by R. Needham was recently published by the Dutch Royal Institute of Linguistics and Anthropology (1963). It appears that in the years he spent in Borneo Scharer noted down an enormous amount of textual material. Most of it is now kept at the Leiden University Library. Part of it, an extensive (about 800 printed pages) monograph on the death cult of the Ngaju, has been published by the same Dutch Institute (1966). The journals Medan Bahasa and Bahasa dan Budaja contain some short contributions on Ngaju by Sjamsji (1958) and Kiting (1958, 1959a,b + c). They are of little scientific value. Finally, mention should be made in this section of the recent extensive and interesting publications of Harrisson on various types of communication systems, marker devices, writing boards and forms of 'writing' in use among the Iban and other Borneo peoples (1965, 1966). In the field of Borneo comparative linguistics an important advance was made by Hudson (1967). On the basis of fieldwork conducted in South Borneo he was able to set up a Barito family of languages comprising three subgroups (Barito-Mahakam, West Barito, and East Barito). Within the West Barito and East Barito subgroups further minor subgroups are distinguished.

5. THE LANGUAGES OF JAVA AND MADURA

The languages which will be treated in this section are Sundanese, Javanese, and Madurese. The Malay spoken in Djakarta has been discussed in the section devoted to Malay and Bahasa Indonesia. SUNDANESE is the language of West Java, with the exception of a strip along the north coast west and east of Djakarta where Javanese dialects are spoken, while in the capital, Djakarta, Malay is used. Since the Second World War several Indonesian as well as European scholars have enriched our knowledge of Sundanese language and literature. Of the Indonesian scholars, R. Satjadibrata must be mentioned first. Although his treatise on Sundanese metrics (1931, reprinted 1953) and on ceremonial word usage (1943, reprinted 1956) are not without value, his main contribution lies in the field of lexicography (Sundanese-Malay dictionary 1944,2nd ed. 1950; SundaneseSundanese dictionary 1948, 2nd ed. 1954). Other Indonesian scholars are R. Momon Wirakusumah who together with I. Buldan Djajawiguna wrote a short grammar (1957), A. Perdanakusumah who contributed a series of elementary articles on Sundanese grammar in the journal Medan Bahasa (1956a, 1956-57), and Ukun Surjaman who gave a description of Sundanese kinship terms (1956). The master thesis of Sumarto on Sundanese phonology defended at Malang was not accessible to me. A special Sundanese section of Medan Bahasa is completely devoted to the Sundanese language and literature. While the grammatical contributions by Indonesian scholars cannot be said to meet

76

E. M . U H L E N B E C K

present standards of modern structural description, Robins' five articles on Sundanese phonology and morphology (1953a + b, 1957, 1959, 1965) leave nothing to be desired on this score. His work is of the best recently done in the field of Indonesian linguistics. Other recent work more or less tinged by modern linguistic views includes Van Syoc's rather disappointing M.A. thesis on Sundanese phonology and morphology defended at the University of Michigan in 1959 and Rukmantoro Hadi Sumukti's M.A. thesis on the same subject defended at Indiana University to which I had no access. A Russian introduction to Sundanese was recently published by Pavlenko (1965). Valuable not only from the point of view of Sundanese literature but also for our knowledge of the language is Eringa's edition (1949) of a part of the Lutung Kasarung. Especially do his many lexicological observations and his discussion of ceremonial word usage deserve to be mentioned. Eringa is now engaged in bringing out a completely new and very copious Sundanese-Dutch dictionary to be published by the Dutch Royal Institute of Linguistics and Anthropology. The basis for this dictionary was laid by the late R. A. Kern, but after his death in 1958 Eringa, with the help of A. A. Fokker, started to revise the then existing manuscript, at the same time adding to it many new lemmata. It is to be expected that the author will be able to finish his dictionary in the course of 1971. Another Dutch scholar who specializes in Sundanese is J. Noorduyn. After having worked for some years at a revision of the Sundanese Bible translation (part of which was published by him with the help of N. Titus in 1962), he has concentrated on the study of Old Sundanese. In 1959 he published several corrections of the transcription of the Batu tulis inscription which probably dates from the fourteenth or fifteenth century. In 1962 he succeeded in giving a better interpretation of the first part of the Tjarita Parahyangan, a mythological-historical text which seems to be written in the same idiom as the Batu tulis inscription. Some results of his study of the original manuscript in Djakarta were recently published by him (1966). The study of J A V A N E S E , the language of the whole of Central and East Java, with the exception of the Madurese-speaking areas of East Java, can boast of a long tradition. Since Pigeaud's lexicographical publications (Javanese-Dutch dictionary 1938, Dutch-Javanese dictionary 1948) and those of Poerwadarminta (Javanese-Javanese dictionary 1939, Indonesian-Javanese dictionary 1942, reprinted in 1943 and 1945), no advances have been made in this field. The enormous amount of lexicographical data collected by Pigeaud and now kept in Djakarta still awaits publication. A structural description of modern standard Javanese phonology and morphology was attempted by Uhlenbeck (1949), who also published extensively in the field of Javanese morphology (between 1953 and 1962). Recently he published an article on Javanese syntax (1965) and on Javanese proper names (1971). In two articles (1947, 1948) De Casparis paid attention to two striking characteristics of Javanese, viz. the high frequency of nonaffixed words containing two vowels and the abundance of rhyming binomials. On the phenomenon of the special ceremonial vocabularies several articles appeared

INDONESIA AND MALAYSIA

77

(Loeb 1944, Gonda 1948, Uhlenbeck 1950, Damais 1950), of these four contributions the one by Loeb is without any value. Slametmuljana, in his thesis of 1954, tried to explain the development of a special polite vocabulary (krama) by certain exigencies of Javanese metrics, an explanation proposed earlier. Besides Slametmuljana's book, which also discusses Javanese metrics, Teeuw's inaugural lecture of 1952 discussed the same subject within the framework of the general problem of the relation between language structure and metrics. For anyone interested in a more or less complete introduction to Javanese metrics, Mas Hardjowirogo's introduction written in Javanese (1926, reprinted in 1952) may be recommended. Gonda's many publications on Indonesian linguistics (see bibliography) often contain Javanese data mostly taken from existing Javanese descriptions. This circumstance limits their value. His monumental work on Sanskrit loanwords in Indonesian languages (1952) is unquestionably his most important contribution to the study of modern as well as Old Javanese. Recently scholars from other countries have begun to pay attention to Javanese. Mrs. Horne published a more or less complete introductory course for Englishspeaking students, with a Javanese-English and English-Javanese glossary (1961, 1963). Teselkin wrote two very concise introductions in Russian, one to the study of modern Javanese and another to Old Javanese (1961, 1963). Most recently Herrfurth presented an introduction to the language in German (1964) which does not reflect our present knowledge of the language. In the United States two as yet unpublished M.A. theses were written by M. Samsuri on Javanese phonology and by Sutardijah Kromowidarso on the Javanese verb system, in the course of their stay at Indiana University. It appears that since the Second World War not much attention has been paid to the study of Javanese in Indonesia. The journals such as Bahasa dan Budaja, Medan Bahasa, Medan Ilmu Pengetahuan, and Madjalah Ilmu-Ilmu Sastra Indonesia only rarely contain articles on Javanese. Mention may be made of short articles by Hadisartama (1958) and Ismail (1959) on the Javanese dialect of Tjirebon, by Prijanggana on the Javanese of Banjuwangi (1957), Tardjan Hadidjaja on the reciprocal influence of Javanese and Bahasa Indonesia (1951, 1951-52), Soetikno on Javanese script (1953), Ukun Surjaman on Javanese kinship terminology (1956), and Trajono on certain marriage ceremonies in Tjilatjap (1959). The most valuable publication is without question Samsuri's article on Javanese phonology (1961). Other elementary publications on Javanese are Prawiroatmodjo's short study on the similarity of Javanese and Bahasa Indonesia (1956), and Padmosoekotjo's Sarine basa Djawa (1956) (the essentials of the Javanese language). Of real scientific value is only Kamil's review of Uhlenbeck's monograph on the Javanese pronominal system (1960) published in BijdrTLV (1962). The M.A. theses of Ramelan ("Bagelen Javanese phonology"), of Noertati ("A phonemic analysis of Semarang-Central-Javanese"), Poernomodjati ("An introduction to the classification of Javanese morphemes and words"), Hartono

78

E. M. UHLENBECK

("Javanese syntax"), and Soepeno ("Geographical boundaries of the Surabaja dialect of Javanese"), all defended at Malang, Indonesia, were not accessible to me. O L D JAVANESE is a term in current use among students of Javanese philology to indicate the language of (1) inscriptions on stone, copper, or gold which are found in Java from the end of the eighth century onwards, and of (2) texts written between the ninth and fifteenth centuries mainly emanating from Indianized kingdoms of East Java. Old Javanese is a term of literary history rather than a purely linguistic term. In general it may be stated that the study of the older phases of Javanese has now reached the point that the data necessary for developing a picture of the historical development of the language have become available. Especially between the two world wars, but also after 1945, much careful philological spadework has been done mainly by Indonesian and Dutch scholars. A great variety of texts became available in reliable editions often with good translations and copious grammatical and lexicological notes, while inscriptional Javanese has also yielded a rich harvest of linguistic data. After Kern's first attempt at giving a grammatical description of Old Javanese (between 1898 and 1906; also found in a slightly revised form in his collected works Vol. 8 and 9) on the basis of the relatively few texts available at that time, Zoetmulder gave a very useful grammatical analysis of the Old Javanese (1950) as found in the Javanese prose version of the Adiparwa, the first book of the Mahabharata epic, which dates from the end of the tenth century. A few years later he published together with I. R. Poedjawijatna a grammatical introduction in Indonesian under the title Bahasa Parwa ( = The language of the Parwa's) which in many respects is similar to his monograph of 1950. Together with his anthology Sekar Sumawur (1958, 1963) this introduction forms a useful manual for Indonesian students. Other elementary introductions to Old Javanese appeared after the war in Indonesia as well as abroad (S. Wirjosuparto 1952, A. Prawirasuganda and S. Sauni 1951, A. Teselkin 1963, A. S. Broto 1963, Soewadji Sjafei 1966). Recently Trajono Sastrowardojo devoted a short article to 'linking elements' in Old Javanese comparing the viewpoints of Zoetmulder and Poerwadarminta (1964). Most important advances were made in the field of Old Javanese epigraphy mainly owing to De Casparis (1950, 1956,1958), Goris (1954), Damais (1951-58) and Pigeaud (1960-63). The two volumes published by De Casparis in the series Prasasti Indonesia contain texts with translations and commentary of a number of Old Javanese, Old Malay, and Sanskrit inscriptions. In his two volumes in the series Prasasti Bali Goris published inscriptions of the island of Bali written in Old Javanese and probably in Old Balinese. Damais extensively reviewed both volumes (1955). Mention should be made of Gonda's articles on Old Javanese most of which appeared in BijdrTLV in the period 1946-60. They are mainly devoted to problems of Old Javanese morphology and syntax, such as his article on the Javanese ife«-forms (1946), his remarks on Old Javanese syntax (1949), his discussion of the relations between syntactical and metrical units in Javanese poetry, and his article "On Old

INDONESIA AND MALAYSIA

79

Javanese sentence structure" (1959). Many of his articles on Indonesian linguistics deal with Old Javanese facts. In particular we ought to mention in this respect his extensive study on verbal forms in Javanese (1949c) in which Modern Javanese as well as Old Javanese morphological facts are discussed. Of the high quality of Gonda's Sanskrit in Indonesia (1952) we have already spoken in the preceding section. Recently Uhlenbeck published two articles on Old Javanese morphology and Old Javanese syntax respectively, both based on the Old Javanese translation-prose of the tenth century (1968, 1970). In the present stage of Javanese studies one cannot limit oneself to the few linguistic publications mentioned above. For the older phases of the language, philological and linguistic studies are so closely intertwined that one cannot escape consulting the existing text editions in Javanese. This implies that, at least for the study of Old Javanese language and literature, a reading knowledge of Dutch is still a prerequisite, as nearly all the translations of the texts and also the existing dictionaries are in Dutch. An exception is Pigeaud's recent translation of the Nagarakrtagama, the most important Javanese text of the fourteenth century of which he has prepared a complete new edition with extensive commentary (1960-63). I have to refrain from giving here a detailed survey of the Old Javanese philological studies. I refer the interested reader to the chapter devoted to Old Javanese and Javanese literature of my bibliographical survey (1964). MADURESE is the language of the island of Madura and of the coastal area of Java opposite Madura. Also in and around Surabaya and in the regency of Banjuwangi Madurese is spoken. Since the Second World War Van Stappershoef supplied a collection of pantun with Dutch translation (1955), while Wirjo Asmoro wrote a series of short articles on Madurese grammar (1951-52, 1952). Recently Stevens published two articles on Madurese, the first on special ceremonial vocabularies and speech styles (1965), the second on Madurese reflexes of Proto-Malayo-Polynesian (1966). His valuable Yale dissertation on Madurese phonology and morphology, the result of recent fieldwork, has recently become available in print (1968). Kiliaan's reliable grammar (2 Vols. 1897, reprinted in 1911) and extensive dictionary (1904-05) together with Berg's description of the Madurese sound system (1941) remain important sources for this language. The unpublished M.A. theses of Endang Soetartinah ("Phonemic analysis of Sumenep") and of R. A. Kamar ("Madurese transformations") were not accessible to me.

6. CELEBES LANGUAGES

In comparison with Borneo considerably more progress has been made with the linguistic exploration of Celebes (Sulawesi). Although the linguistic situation is very intricate in certain areas, the language map made by Esser (1938) gives a fairly reliable, if still on certain points preliminary, list of the languages and their location. It is not

80

E. M. UHLENBECK

surprising that not all the many Celebes languages have been investigated to the same extent. Some of them, such as Bare'e and Mori, have been very well studied by trained linguists; for others we have to content ourselves with a short wordlist or some textual material. A bibliographical survey of the languages of Celebes is being prepared by J. Noorduyn. In his linguistic map which rests on Adriani's pioneering work, Esser arrived at a classification of the languages of Celebes and adjacent islands in eight groups of related languages. Within some of them he made a number of subdivisions. The criteria on which he based his classification of the Celebes languages are nowhere explicitly stated, but it is safe to assume that he mainly relied on genetic considerations. 1. The Philippine group: To this group belong the following languages: 1) Sangirese and Talaud, la) Bantik, lb) Bentenan, 2) Mongondow (with Ponosakan), 3) Tombulu-Tonsea'-Tondano, 4) Tontemboan-Tonsawang. 2. The Gorontalo group: 1) Bulanga, 2) Kaidipan, 3) Gorontalok, 4) Buol. 3. The Tomini group. 4. The Toradja group -. 1) Kaili, 2) Kulawi, 3) Pipikoro, 4) Napu, 5) Bada'-Besoa, 6) Leboni, 7) Bare'e, 8) Wotu. 5. The Loinang group: 1) Loinang, 2) Bobongko, 3) Balantak. 5a. Banggai. 6. The Bungku-Laki group: 1) Bungku-Mori, 2) Laki, 3) Laiwui, 4) Landawe, 5) Mapute. 7. South Celebes languages: 1) Makasarese, 2) Buginese, 3) Luwu' language, 4) Sa'dan, 5) Pitu-ulunna-salu, 6) Mandarese dialects, 7) Seko. 8. The Muna-Butung group: 1) Muna-Butung, 2) Southern-Butung, 3) Language of the Tukangbesi islands, Kalao-toa, Karompa, and Bonerate, 4) Wolio and Lajolo, a language spoken in the southern part of the isle of Salayer. The languages of the first group are considered by Esser to be a separate subgroup of the Philippine languages. Of these languages Sangirese, (Bolaang-) Mongondow, and Tontemboan have been well studied. In the middle of the nineteenth century SANGIRESE, like Ngaju-Dayak in Borneo, was made the language of the Church by the German missionaries working in the archipelago, which explains why so much Christian literature has been translated into Sangirese. By far the most important recent addition to our knowledge of Sangirese since Adriani published his Sangirese grammar (1893), is the large Sangirese-Dutch dictionary with Dutch-Sangirese index by Steller and Aebersold, the result of nearly sixty years of lexicological effort (1959). This dictionary is based on the dialect of Manganitu, a dialect which had acquired a certain preeminence, but the dictionary contains data from other dialects as well. In addition, new textual material with translation was published by Aebersold (1952, 1957, 1959). Three articles on the Sangirese as spoken on Sarangani island were contributed by Kenneth Maryott (1961, 1963) and Alice Maryott (1963). Our knowledge of BOLAANG-MONGONDOW mainly rests on the lifelong linguistic activity of the missionary W. Dunnebier. Within the period 1911-38 he published

INDONESIA AND MALAYSIA

81

a great number of ethnographical and linguistic studies, and translated the Old Testament, the Acts of the Apostles, and other parts of the New Testament. In 1929 and 1930 his grammar of over 300 printed pages appeared. After the war he continued his activity, bringing out a very extensive dictionary (1951) and much textual material with translation (1948, 1953a and b). The value of Dunnebier's œuvre derives from the fact that he had a very intimate factual knowledge of the language, the result of a prolonged stay among the speakers of Bolaang-Mongondow. TONTEMBOAN has been studied mainly by the missionary Schwartz (texts 1907, dictionary 1908) and by Adriani and his wife, who composed a grammar on the basis of the not quite reliable material collected by Schwartz (1908). After the war the French linguist J. Verguin published a few articles on Tontemboan (1960, 1961). Recently F. S. Watuseke published a series of short articles on TONDANO (1956-59), while Wantalangi made some remarks on morphology and word-formation (1957, 1960). Two master theses on Tonsea' by J. G. Rumbajan ("Tonsea' morphology") and J. F. Pattiasina ("Phonemic analysis of Tonsea") defended at the Faculty of Language and Art Teachers of the Institute of Educational Sciences of Malang, Indonesia were not accessible to me. On Esser's second group, the GORONTALO languages, no new publications appeared after the Second World War, except Van Wouden's article on myths and society in Buol (1941), which contains text and translation of the tribal myth of the Buol, or Bwuolo, and a few short articles by Taulu (1955, 1956a + b) and Baga (1955). Hardly any information is available on the third group, the TOMINI languages (Petapa, Tinombo, Kasimbar, Umalasa, Tomini, Bolano, and Tontoli). Much better is the situation for the TORADJA languages, that is the languages of Central Celebes. This we owe mainly to the work of two linguists, N. Adriani and S.J. Esser. The latter, a linguist with an exceptionally wide and intimate knowledge of Indonesian languages, had, before his untimely death in 1944, done much and valuable work on the languages of the first of the three subgroups of the Toradja languages. An extract of his reports to the Batavia Society of Arts and Sciences was recently published by Noorduyn together with a biographical notice (1963). Before the Second World War Esser made descriptions of LEDO, the chief Kaili dialect (1934) and of KULAWI (with Adriani, 1939). His Kulawi description partly rests on material collected by Adriani. His sketch of the UMA ( = Pipikoro) language which is closely related both to Ledo and to Kulawi, was posthumously published by Noorduyn (1964) who added a wordlist based on the texts noted down by Esser. BARE'E has been studied extensively by N. Adriani. His main linguistic work appeared after his death. Within the period 1928-33 a dictionary of over 1000 pages, a grammar of nearly 500 pages, and a collection of texts of nearly 700 pages came from the press thanks to the expert assistance of Esser, and as to the texts, with the help of Mrs. Adriani. After the Second World War a second edition of the monumental ethnography by Adriani and Kruyt of the Bare'e speaking Toradja of Central Celebes (1950-51) came out completely revised by Kruyt. The third volume of this

82

E. M. UHLENBECK

monograph contains a sketch of Bare'e and a linguistic survey of Celebes and Halmahera. A short farticle by Verguin (1959) is based on the material collected by Adriani. No publications on the LOINANG languages have appeared since 1935. On the closely related BANGGAI language a grammar was published by the missionary Van den Bergh (1953). His grammar which contains a wordlist, some texts, and a linguistic map furnishes data on the two dialects spoken in the Banggai archipelago. Of the various languages which belong to the BUNGKU-LAKI group, only MORI has been described. In spite of its traditional framework, Esser's detailed grammar of this language (1927, 1933) unquestionably belongs to the best of what has been written on the languages of Indonesia before the Second World War. Based on firsthand knowledge of the language, it gives rich and reliable information throughout on many phonological and morphological characteristics of the language. Esser distinguished many dialects within Mori. His grammar is mainly based on Mori as it was spoken in Tinompo, but it also contains data on the other Mori dialects. Since the days of Matthes (middle of the nineteenth century) hardly any linguistic publication has appeared on MAKASARESE and BUGINESE, the two chief languages of the South Celebes group. Kinship terms and short texts in Makasarese can be found in Chabot's thesis (1950) and in an article by Fischer (1951). A short sketch of BUGINESE is found in Noorduyn's important thesis (1955a) in which he gave text and translation of an eighteenth century chronicle of Wadjo'. A few articles of dubious value were published by R.A.Kern (1939, 1940, 1950). Noorduyn wrote a short note on Buginese phonetics (1955b). Before 1942 fieldwork on both languages was done by the linguist A. A. Cense. Part of his extensive collection of data was lost during the war. Nevertheless it seems that enough has been saved to enable him to publish a new Makasarese dictionary which certainly will supersede B. F. Matthes' work. Preparation of this dictionary is nearly completed but it will take several years before a printed edition will be available. It is strongly to be hoped that Dr. Cense will find occasion to make his intimate knowledge of the South Celebes languages generally available. Texts written in Makasarese or Buginese script can be used only by those who have become thoroughly familiar with the Makasarese and Buginese language. Therefore it may be useful to mention here the textual materials available in Latin transcription: Abd. Rahim's small collection of Makasarese poems (1954), the Buginese maritime code and regulations published by Tobing with an Indonesian translation (1961), and the Makasarese history of Goa by Wolhoff and Abd. Rahim ( ± 1959). In Makasar a new journal called Bingkisan was founded in 1967. This journal which is published in stencilled form twice a month contains some short Makasarese and Buginese texts usually with Indonesian translation, and also a few introductory articles on both languages. The Indiana University thesis of Samsuri ("An introduction to Rappang Buginese grammar", 1965) was not accessible to me. On MANDAR not much is known. In 1955 Tenriadji and Wolhoff supplied a text

INDONESIA AND MALAYSIA

with translation in Bahasa Indonesia. A wordlist compiled by Cense and enlarged by him with the help of Tenriadji still awaits publication. TAE' or SA'DAN-TORADJA has been studied by the missionary and linguist Van der Veen. His main linguistic publication is his copious dictionary (1940), which also contains a short grammatical introduction. An attractive feature of his dictionary is that the various lemmata are explained in a very detailed manner. The many other publications of Van der Veen include a translation of the New Testament (1951) and more recently of the whole Bible (1960) and other Christian literature. A great number of Tae' texts were supplied by him, mostly with Dutch, Bahasa Indonesia, or English translation and notes (1950, 1965, 1966). Finally some articles by Pakan (1957, 1958, 1959a-c) and Watuseke (1960) containing some textual material with Indonesian translation have to be mentioned. Of Esser's last group, the MUNA-BUTUNG languages, only one has been studied by linguists. First the linguist Van den Berg did fieldwork on Wolio, the official language of the Buton sultanate, but his collection of materials was lost during the war. An extract of his linguistic reports to the Batavia Society of Arts and Sciences has been published by Cense (1954). After Van den Berg (who died during the Second World War) the linguist Anceaux gave a valuable if too short analysis of Wolio (1952). His concise outline on many points gives interesting information on syntactical constructions and intonational features in Wolio. A short Wolio dictionary will be published by the same author in the near future.

7.

THE LANGUAGES OF THE ISLANDS EAST OF JAVA A N D CELEBES

The languages which will be discussed here belong according to the classification of Esser (1938) to five groups: 1) The Bali-Sasak group, 2) The Bima-Sumba group, 3) The Ambon-Timor group, 4) The Sula-Batjan group, and 5) The South-HalmaheraWest New Guinea group. Outside this classification remain the non-Austronesian languages which are spoken on Alor, on North Halmahera, and in some places in the interior of Timor. These will be discussed at the end of this section. The languages of the Bali-Sasak group have long been neglected. It is particularly surprising that so little attention has been paid to BALINESE, the language of an internationally so widely known culture. One of the few grammatical descriptions is Kersten's Balinese grammar (1948). A short note on Balinese was published by Sutardjo (1958). It is to be expected that Kamil, who is preparing a thesis on Balinese, will give us the first structural description of the language. Two master theses of the Faculty of Language and Art Teachers in Malang by Kartini ("Balinese phonemes") and by Moeljono ("Balinese transformations") were not accessible to me. Textual material has been furnished by C. Hooykaas (1948a+b), J. Hooykaas-van Leeuwen Boomkamp (1955, 1961), and Ktut Ginarsa (1959). Ktut Ginarsa's article contains a list of Balinese affixes and some more grammatical information. Very concise

84

E. M. UHLENBECK

Balinese-Indonesian and Indonesian-Balinese dictionaries were published by I Gusti Ananda Kusuma (1956). Swellengrebel, before the war engaged in the preparation of an extensive Balinese-Dutch dictionary, lost most of his material during the war. From his pen a number of articles appeared on Balinese incantation formulas and sayings (1949, 1950-51) and on problems of Bible translation in Balinese (1950a+ b, 1963). Great progress has been made in the study of Inscriptional Balinese thanks to Goris (1954, 1957) and Damais (1960, 1963). Teeuw pointed out the importance of this Inscriptional or Old Balinese for comparative purposes (1965). In 1961 Ktut Ginarsa gave text and translation of an inscription discovered in Gianjar in 1957. SASAK, the language of the island of Lombok although previously only studied by Goris (1938) and Hooykaas (1948b), has the rare distinction to be the first language of Indonesia to which modern dialect-geographical methods have been applied. In 1951 Teeuw published his dialect atlas, in 1958 the interpretation of the maps of this atlas. Some preliminary results of his fieldwork are found in a series of earlier articles (1950, 1953a+b, 1954,1957). To the Malang master thesis of Salehudin ("Sasak phonemes") I had no access. On SUMBAWARESE no new information has become available since Voorhoeve published the textual materials collected by Jonker (1935), except for a Sumbawarese translation of a Malay religious tract (Voorhoeve 1942-48). The languages of the Bima-Sumba group are relatively well known, BIMA was studied by Jonker who published a grammar, a dictionary, and a volume of texts (1894, 1896a + b). The only publication after the war is a short article by Sjamsuddin (1959). The Malang master thesis of Soepardi on the morphology of Bima was not accessible. MANGGARAI, a language spoken on the western part of the island of Flores, was described by the missionary Burger (1946) who a few years earlier had supplied some textual material (1940-41, 1941). An extensive dictionary compiled by Yerheijen was recently published (1967). Some lexicological and grammatical articles also from the hand of Verheijen deserve to be mentioned here (1941, 1942^8, 1950, 1963). The Malang master theses of Suharno on the phonemes of Manggarai and of Soebowo Tjitrowidjojo on phonemic analysis in Maumere were not accessible to me. As to SIKKA, NGAD'A, and LIO, three closely related languages spoken on Central Flores, they have been studied by the missionary Arndt. After his prewar publications (grammar of the Sikka language 1932, Lionese-German dictionary 1933b, "Grammatik der Ngad'a-Sprache" 1933a), he recently brought out a collection of texts (1960) and a dictionary of the Ngad'a language (1962). The two dialects of Sumba, KAMBERA and WEWEWA, have been studied for many years by the linguist and missionary Onvlee. Beside some articles (1925, 1929, 1933, 1949, 1950) he gave a complete translation of the New Testament not only in the Kambera dialect but also in Wewewa. The Kambera translation came out in 1955; the edition of the Wewewa translation is in preparation. In the journal Bahasa dan Budaja he wrote a short survey of his research (1956). At present Onvlee is preparing a dictionary of Kambera, which will also contain information on Wewewa. A biblio-

INDONESIA AND MALAYSIA

85

graphy of linguistic work done on Flores and adjacent islands was supplied by Goris (1956). Of HAWU or SAWU not much is known. Onvlee made some phonetic observations (1950). A description of Hawunese phonology formed the subject of an unpublished M.A. thesis of M.C. Radja Huba at Indiana University in 1958. Only a few languages of Esser's Ambon-Timor group have been the subject of recent linguistic fieldwork. Father Arndt wrote a grammar of Solor (1937). His monograph on the religion of Eastern Flores, Adonare, and Solor (1951) contains textual material. A few data on the languages of the islands of Kedang, Alor and Pantar are found in the ethnographical studies of Nicolspeyer (1940) and Cora Du Bois (1944). A valuable survey of the languages spoken on the island of Timor was given by Capell (1943-45), who was the first to recognize that non-Austronesian languages are also spoken on the island. Within the Portuguese part of the island, Capell distinguishes the following languages: a) Austronesian languages: 1. Tetum, 2. Mambai, 3. Tukudede, 4. Galoli, 5. Idate; b) non-Austronesian (or Papuan): 1. Bunak, 2. Makasai, 3. Waimaha, 4. Kairui. In the now Indonesian western part of the island Capell distinguishes three languages: Kupang, Vaikenu (spoken in the Portuguese enclave in West-Timor), and West-Timorese. Of all these languages or interrelated dialects TIMORESE has become rather well-known through the efforts of the missionary Middelkoop, who published an extensive collection of texts (1939, 1958, 1959, 1963) and a grammar (1950). A dictionary is still in preparation. A collection of texts in TETUM with Portuguese translation and wordlist were supplied by De Sa (1961). In the publications by Vroklage (1952-53) and Wortelboer (1955) there are hardly any data on the language. Data on BUNAK or BUNA' were recently furnished by Berthe (1959, 1963, 1964-65). His opinion that Buna' is a kind of mixed language has been refuted by Cowan (1963) who assumes that the language is related to the western group of what he calls the West-Papua-phylum. The Malang thesis of Marian Manuputty ("Phonemic analysis of Dawan") was not available. After Jonker's important and voluminous studies on ROTINESE (a dictionary [1908], a grammar of over 700 large, closely printed pages [1915], a volume with texts [1911], and several long articles [1906, 1913]), nothing of any value has been published on this language. The short notes by Gyanto (1957), Mardiwarsita (1957), and Manafe (1959, 1960) are hardly of any importance. The languages of the islands of Wetar and Kisar became better known owing to fieldwork done by De Josselin de Jong. In 1937 he published a short grammatical sketch with textual material and English translation and a vocabulary of the OIRATA language. This language is spoken in Oirata on the island of Kisar, and according to De Josselin de Jong is closely related with the Timorese dialect of Loikera in northeast Timor. In a second volume (1947) he gave texts, grammatical notes, and a wordlist on ERAI (Wetar). On the language of Led, apparently very closely related with Moa, no new data

86

E. M. UHLENBECK

became available since Jonker's texts and grammatical notes were posthumously published by Voorhoeve (1932). On the rest of the languages of the Ambon-Timor group little work of linguistic importance has been done. On the languages of the Tanimbar islands the missionary Drabbe published elementary grammars of limited value, namely of Fordate (1926a), Jamdena (1962b, 1935), and Selaru (1932c). Of all these languages he also supplied vocabularies (1932a, b, c). On the language spoken on the island of Kai nothing has been published since the missionary Geurtjens published his grammatical introduction and wordlist (1921 a + b), except for a short note by Farfar in the journal Medan Bahasa (1959). On the languages of West and East-Ceram (Seran) Stresemann (1918) and Niggemeyer (1951-52) are the only sources of any importance. A few specimens of poetry in the languages of Rana (Central Buru) and Wetan (Babar archipelago) were published by De Josselin de Jong (1941). On the Ambon languages only Stresemann's study is worth mentioning (1927). No reliable information is available on the languages of the Sula-Batjan group. Of the languages of South Halmahera (Buli, Jawanli or Wajamli, Maba, Bitjoli or Ingli, Patani, Sawai, Weda or Were, Gane) and of the adjacent Makian-islands (East and West Makian) only BULI is rather well recorded owing to the missionary Maan, who in 1940 published an extensive wordlist, followed in 1951 by a grammar. A few data on Makian are found in two short articles by Ali Mahmud Abd. Rahim/ Amra (1958, 1960). The non-Austronesian languages of North Halmahera (Galela, Tobelo, TobeloBoeng, Tobelo-Dodinga, Modole, Tabaru, Sahu'u, Loda, Ternate, Tidore) have become known by the thesis of Van der Veen (1915). The hypothesis already proposed by Schmidt in 1902 that the languages of North Halmahera were related with certain Papuan languages of New Guinea was later confirmed by Cowan (1957-58).

BIBLIOGRAPHY ABD. RAHIM.

1954. Tamalate, 20 Poesi Mangkasara'. 45 pp. Makassar. 1950. Notes on the Bajau language. SarawakMJ 5.196-

ABDUL GHANI BIN BAGUL.

200. 1964. Collins Malay Gem dictionary, Malay-English: English-Malay. 702 pp. London, Glasgow. ACHADIATI IKRAM, NJ., and LUKMAN ALI, eds. 1966. Simposium Bahasa dan Kasusastraan Indonesia 25-28 Oktober 1966. v i + 2 6 + 61+ 6 9 + 9 7 + 13+ 15 pp. Djakarta. 2 ADINEGORO. 1965. Kamus-Singkatan kata Indonesia sekarang. 74 pp. Djakarta. ADRIANI, N. 1893. Sangireesche spraakkunst. x + 284pp. Leiden. ABDUL RAHMAN BIN YUSOP.

INDONESIA AND MALAYSIA

87

. 1928a. Spraakkunstige schets van de taal der Mentawai-eilanden. BijdrTLV 84.1-117. . 1928b. Bare'e-Nederlandsch woordenboek met Nederlandsch-Bare'e register. xv +1074 pp. Leiden. . 1931. Spraakkunst der Bare'e-taal. YBG 70. xvi + 481 pp. . 1932, 1933. Bare'e verhalen. 2 vols. Vol. 1, vm + 427 pp.; vol. 2, xi + 247 pp. The Hague. ADRIANI, N . , and M . L . ADRIANI-GUNNING. 1 9 0 8 . Hoofdstukken uit de spraakkunst van het Tontemboansch. 259 pp. The Hague. ADRIANI, N . f , and S. J. ESSER. 1939. Koelawische taalstudien. Bibliotheca Celebica 1-3. Vol. 1, VIII+91 pp.; vol. 2, 90 pp.; vol. 3, 113 pp. Bandoeng.

N., and ALB. C. KRUYT. 1950-51. De Bare'e sprekende Toradjas van Midden-Celebes (De Oost-Toradjas). 3 vols., 4 parts. VKNA 44,45, and 46/1-2, x n + 478 pp., vni-f 557 pp., vm + 484+11 pp. +159 plates. New revised edition. Amsterdam. (First ed. 1911 and 1912). AEBERSOLD, WALTER E. 1952. Het verhaal van Himbawo, een Sangirees heldendicht. BijdrTLV 108.265-97. . 1957. Het verhaal van Makäampo. BijdrTLV 113.122-67. . 1959. Sasahola laanang manandu (De lange Sasahola). BijdrTLV 115.372-89. AICHELE, W. 1936. Eine neu erschlossene frühindonesische Literatursprache in ihrem Einflusz auf das Altjavanische. ZDMG 90. N.F. 15. *18*-*19*. . 1942-43. Die altmalaiische Literatursprache und ihr Einfluss auf das Altjavanische. ZES 33.37-66. . 1954. Sprachforschung und Geschichte im Indonesischen Raum. OE 1.107-22. ALI MAHMUD ABD. RAHIM/AMRA. 1 9 5 8 . Sedikit tentang bahasa Tahani dipulau Makian. MeBa 8 / 4 . 2 7 - 3 2 . . 1960. Seni Dontji dipulau Makian. BaBu 9.238^16. ALIEVA, N . F . 1961a. Indonezijskij jazyk na sovremennom etape. Problemy vostokovedenija 1.85-93. . 1961b. Main theses of lecture. Ponjatie aggl'utinacii i aggl'utinativnogo tipa jazykov (tezisy dokladov), pp. 51-2. Leningrad. . 1963a. Sootnosenie slovoobrazovatel'nykh i sintaksiöeskikh funkcii glagol'nykh affiksov v indonezijskom jazyke. VJa 103-11. . 1963b. Vyrazenie nekotorykh modal'nykh znacenij v sovremennom indonezijskom jazyke. Jazyki Kitaja i Jugo-Vostoönoj Azii, 59-74. Moscow. . 1963c. O susönosti glagol'noj affiksacii v indonezijskom jazyke. Jazyki Kitaja i Jugo-Vostocnoj Azii, 75-80. Moscow. . 1964a. K voprosu o klassifikacii kornevykh morfem v indonezijskom jazyke. Spornye voprosy stroja jazykov Kitaja i Jugo-Vostocnoj Azii, 163-73. Moscow. . 1964b. Nekotorye osobennosti indonezijskoj aggljutinacii. Morfologiceskaja tipologija i problema klassifikacii jazykov, 292-7. Moscow. , 1967. K kharakteristike glagol'nykh kategorij indonezijskogo jazyka. IstoADRIANI,

88

E. M. UHLENBECK

riko-filologiceskie issledovanija Sb. statej k 75-letiju Akad. N. I. Konrada, 25-9. Moscow. ALTMANN, G. 1963. Phonic structure of Malay pantun. AO 31.274-86. . 1967. The structure of Indonesian morphemes. Asian and African Studies 3.23-36. , and R. STUKOVSKY. 1965. The climax in Malay pantun. Asian and African Studies 1.13-20. AMIN, MUH. 1958. Akar kata lo'dalam bahasa Bugis. BaBu 7.29-31. . 1959. Akar kata lo' dalam bahasa Bugis. MeBa 9/2.43-6. ANANDA KUSUMA, I GUSTI. 1956. K a m u s Bali-Indonesia. (Mimeographed). 57 p p . ;

Kamus Indonesia-Bali. (Mimeographed). 70 pp. Denpasar. ANCEAUX, J.C. 1952. The Wolio language, outline of grammatical description and texts. VKI 11. 93 pp. ANDREJEV, N. D. 1957. Some problems of Bahasa Indonesia phonology. SL11.44-6. , V . P . GOLOVANOV, L . I . IVANOV, a n d A . K . OGLOBLIN. 1958.

Kornevydelitel'-

naja programma indonezijskogo algoritma masinnogo perevoda. Materialy po masinnomu perevodu coll. 1.88-97. Leningrad. , and L. I. UCHANOVA. 1959. Struktura indonezijskogo sloga. Uëebnye zapiski L G U 282, series of Oriental Sciences, 11th issue, 185-8. Leningrad. ANTONISSEN, A. 1958. Kadazan-English and English-Kadazan dictionary. 274 pp. Canberra. APPELL, G.N. and L.W.R. 1961. A provisional field dictionary of the Rungus Dusun language of North Borneo. [Mimeographed.] ARNDT, P. 1931. Grammatik der Sika-Sprache. 52 pp. Ende, Flores. . 1933a. Grammatik der Ngad'a-Sprache. YBG 72.353 pp. . 1933b. Li'onesisch-Deutsches Wörterbuch. 555 pp. Ende. . 1937. Grammatik der Solor-Sprache. 124 pp. Ende. . 1951. Religion auf Ostflores, Adonare und Solor. Studia Instituti Anthropos 1. xii + 248 pp. + 1 map. Wien, Mödling. . 1960. Mythen der Ngadha. AnnLat 24.9-137. . 1962. Wörterbuch der Ngadha-Sprache. Studia Instituti Anthropos 15. ASMAH HAJI OMAR. 1962. Serba sedikit tentang Tatabentok dan Tatakalimat dalam Bahasa Melayu. D e w a n Bahasa 6.440-52, 502-13, 549-60.

. 1962-63. Apakah Bahasa Melayu mempunjai unsor fleksi. Bahasa 3.18-22. . 1963. Bahasa Semang, dialek Kentakbong. (Stencilled). 75 pp. Kuala Lumpur. . 1963-64. Unsor Bawahan Terdekat dalam Tatakalimat Bahasa Melayu. Dewan Bahasa 7.545-53 (1963); 8.9-15 (1964). . 1964a. Bahasa Iban. (Stencilled). 49 pp. Kuala Lumpur. . 1964b. Pemindahan kalimat. Dewan Bahasa 8.105-15, 158-65. . 1964c. Chiri 2 Khusus Untok Gatra. Dewan Bahasa 8.499-509. . 1965. Kata-kerja dengan awalan di-. Bahasa 7.100-6.

INDONESIA AND MALAYSIA

89

. 1967. The role of dialects in the modernization of language. International Conference on the Modernization of Languages in Asia. 7 pp. . 1968. Word classes in Malay. AnL 10/5.12-22. BAGA, H . IDUNA. 1955. Asalnja Gorontalo. MeBa 5 / 1 0 - 1 2 . 2 8 - 3 1 . BARTH, J.P.J. 1910. Boesangsch-Nederlandsch woordenboek. xxvm+343 p p . + map. Batavia. BARTLETT, H.H. 1952. A Batak and Malay chant on rice cultivation, with introductory notes on bilingualism and acculturation in Indonesia. PAPhilosS 96.629-52. BAUGHMAN, BURR. 1963. Speaking Iban. Sarawak Information Service Publication, i v + 76 pp. Brunei. Review: N.C. Scott, BSOAS 28.461 (1965). BAUSANI, ALESSANDRO. 1960. The first Italian-Malay vocabulary by Antonio Pigafetta. EW NS 11.229^8. . 1964. Note sui vocaboli persiani in Malese-Indonesiano. AION-O 14.1-32. BERG, C.C. 1941. Beschouwingen over de grondslagen der spelling. TBG 81.96-174. . 1951. De problematiek van het Bahasa-Indonesia-experiment. 36 pp. Groningen, Djakarta. BERTHE, L. 1959. Sur quelques distiques buna' (Timor central). BijdrTLV 115.336-71. . 1963. Morpho-syntaxe du Buna' (Timor central). Homme 3.106-16. . 1964-65. La Terre, l'Au-Dela et les thèmes maritimes chez les Buna' (Timor Central). L'ethnographie NS 58-9, 72-87. BERWICK, E.J. H. 1953. Some Kelantan bird names. JMBRAS 26/1.140-4. BIJLEVELD, B. J. 1943. Herhalingsfiguren in het Maleisch, Javaansch en Soendaasch. Diss. Utrecht. vm + 118pp. Groningen, Batavia. BJELKINA, E.S. 1965. Intonacija osnovnykh kommunikativnykh tipov predlozenij v indonezijskom jazyki. Phonetica 12.134-6. BOECHARI. 1966. Preliminary report on the discovery of an Old-Malay inscription at Sodjomerto. MISI 3 . 2 4 1 - 5 1 . Borneo Evangelical Mission. 1964. Dusun reader. Hilo, Hawaii, University of Hawaii Peace Corps Training Center. BROTO, A.S. 1963. Bahasa kawi I. HI+ 151 pp. Djakarta. BROWN, C.C. 1956. Studies in Country Malay, xi + 259 pp. London. BURGER, P.A. 1940-41. Manggaraische Tiermärchen mit grammatischen Erlaüterungen. ZES 31.249-77.

. 1941. Manggaraise verhalen over het ontstaan van de rijst en de mais. TBG 81.411-23. . 1946. Voorlopige Manggaraise spraakkunst. BijdrTLV 103.15-265. CAPELL, A . 1943-45. Peoples and languages of Timor. Oceania 14.191-219, 3 1 1 - 3 7 ; 15.19-48.

. 1954 (19622). A linguistic survey of the South-Western Pacific. South Pacific Technical Paper 70. ix+210 pp. Noumea, New Caledonia. (2nd ed. South Pacific Technical Paper 136. xiv + 258 pp. 1962).

90

E. M. UHLENBECK

1961. Tengleq Kui Serok, A study on the Temiar language, with an ethnographical summary. v + 195pp. Kuala Lumpur. CARR, DENZEL. 1966. Homorganicity in Malay/Indonesian in expressives and quasi expressives. Lg 42.370-7. CENSE, A.A. 1954. Mededelingen uit de verslagen van Dr. E.J. van den Berg. BijdrTLV 110.154-84. , and E. M. Uhlenbeck. 1958. Critical survey of studies on the languages of Borneo. BS 2.82 pp. CHABOT, H . T H . 1950. Verwantschap, stand en sexe in Zuid-Celebes. xii+277 pp. Groningen, Djakarta. CHEESEMAN, H. R. 1959. Bibliography of Malaya, vn + 234 pp. London, New York, Toronto. Supplementary Addendum and Corrigendum 1-6 (1959-64) pub. in Malaysia, The Journal of the British Association of Malaysia (London). CLAYRE, B . M . 1966-67a. A comparison of some dialects of Dusun. Sabah Society Journal 3.3-12. . 1966-67b. The two favourite clause types in use in Dusun. Sabah Society Journal 3.113-27. COEDÈS, G. 1930. Les inscriptions malaises de Crivijaya. BEFEO 30.29-80. . 1964. A possible interpretation of the inscription at Këdukan Bukit (Palembang). Malayan and Indonesian studies, Essays presented to Sir Richard Winstedt on his eighty-fifth birthday, ed. by John Bastin and R. Roolvink, 24-32. Oxford. COURT, CHRISTOPHER. 1967a. Some areal features of Mëntu Land Dayak. OL 6.46-50. . 1967b. Pointing and asking: a note on deixis in Mëntu Land Dayak. BijdrTLV 123.520-21. The Hague. COWAN, H . K . J . 1948. Aanteekeningen betreffende de verhouding van het Atjehsch tot de Mon-Khmer-talen. BijdrTLV 1 0 4 . 4 2 9 - 5 1 4 . . 1957-58. A large Papuan language phylum in West New Guinea. Oceania 28.159-66. . 1963. Le Buna' de Timor: une langue 'Ouest-Papoue'. BijdrTLV 119.387400. 2 CROES, H . C . , H. M. DUIN, and A. VAN DUCK. 1949 . Bahasa Indonesia. Eenvoudig leerboek voor practijk en schoolgebruik. 2nd ed. 187 + 34 pp. 4th, 5th ed. 216 + 48 pp. Groningen, Djakarta. (19514, 1953s). CUBIT, L.E. 1964. Kayan phonemics. BijdrTLV 1 2 0 . 4 0 9 - 2 3 . DAHL, OTTO CHR. 1951. Malgache et Maanjan: Une comparaison linguistique. Studies of the Egede Institute 3.406 pp. Oslo. DAMAIS, L. C H . 1950. Les formes de politesse en javanais moderne. B S E I C 2 5 . 2 6 3 - 8 0 . . 1951-58. Études d'épigraphieindonésienne. BEFEO 45.1-63 (1951); 46.1-105 (1952); 47.7-290 (1955); 49.1-257 (1958). . 1955-64. Bibliographie Indonésienne. BEFEO 47.621-54 (1955); 48.607-98 CAREY, ISKANDAR.

INDONESIA AND MALAYSIA

91

(1957); 49.679-746 (1958-59); 50.417-518 (1960-62); 51.535-94 (1963); 52.183240 (1964). . 1960, 1963. Etudes Balinaises. BEFEO 50.133-60 (1960); 51.125-42 (1963). DE BOER, D . W . N . 1947. Het Toba-Bataksche verhaal van Si Tadjom Bolak en Si Radja Mebangebang. BijdrTLV 104.45-88. DE CASPARIS, J. G. 1947. L'importance de la disyllabie en Javanais. India Antiqua (Festschrift Vogel), 63-76. Leiden. . 1948. De betekenis van het rijm in de structuur van het Javaans. TBG 82. 401-537. . 1950. Inscripties uit de Cailendra-tijd. Diss. Djakarta-Prasasti Indonesia I. 204 pp. Bandung. . 1956. Selected inscriptions from the seventh to the ninth century A.D. Prasasti Indonesia II. 395 pp. Bandung. . 1958. Short inscriptions from Tjandi Plaosan-Lor. Berita Dinas Purbakala 4. 36 pp. + 9 plates. DEGROOT, A . W . 1949. Structurele syntaxis. 289 pp. The Hague. DE JOSSELIN DE JONG, J . P . B . 1937, 1947. Studies in Indonesian culture. 1. Oirata, a Timorese settlement on Kisar, 2. The community of Erai (Wetar) (Texts and notes). V K N A new ser. 39. 289 pp.; 50/2. 152 pp. . 1941. Oost-Indonesische poézie. BijdrTLV 100.235-54. DEMID'UK, L . N . 1967. Udvojenije kak sposob slovoobrazovanija i formoobrazovanija v sisteme glagola v indonezijskom jazyke. Jazyki Jugo-Vostoönoj Azii, 249-56. Moscow. DEMPWOLFF, O . 1922. Entstehung von Nasalen und Nasalverbindungen im Ngadju. Z E S 13.161-205.

. 1934, 1937, 1939. Vergleichende Lautlehre des austronesischen Wortschatzes. 3 vols. ZES 15, 17, 19. . 1941. Einführung in die Malaiische Sprache. ZES Beih. 22. vin + 86 pp. DE SA, A. B. 1961. Textos em Teto da Literatura Oral Timorense. Estudos de ciéncias políticas e sociais. 45/l.xxxm + 266 pp. Lisboa. DJAJADININGRAT, HOESEIN. 1934. Atjéhsch-Nederlandsch Woordenboek met Nederlandsch-Atjéhsch register door G.W.J. Drewes. Vol. 1, xvi+1011 pp.; vol. 2, 1349 pp. Batavia. DOBLE, MARION. 1960. Kapauku-Malayan-Dutch-English dictionary. vn+156pp. The Hague. DOUGLAS, R. S. 1911. A comparative vocabulary of the Kayan, Kenyah, and Kalabit languages. SarawakMJ 1/1.75-119. DRABBE, P . 1926a. Spraakkunst der Fordaatsche taal. V B G 67/1.75 pp. . 1926b. Spraakkunst der Jamdeensche taal. VBG 67/2.93 pp. . 1932a. Woordenboek der Fordaatsche taal. VBG 71/2.118 pp. . 1932b. Woordenboek der Jamdeensche taal. VBG 71/3.122 pp.

92

E. M. UHLENBECK

. 1932c. Beknopte spraakkunst en korte woordenlijst der Slaroeéesche taal. VBG 71/4.32 pp. . 1935. Aanvulling van de spraakkunst der Jamdeensche taal. TBG 75.625-33. DREWES, G.W. J. 1948. Van Maleis naar Basa Indonesia. 39 pp. Leiden. . 1950. De herkomst van het voegwoord bahwasanja. Bingkisan Budi (Festschrift Van Ronkel), 104-16. Leiden. . 1955. Een 16de eeuwse Maleise vertaling van de Burda van Al-Bùsirl (Arabisch lofdicht op Mohammad) VKI 18.100 pp. Du Bois, CORA. 1944. The people of Alor. A social-psychological study of an East Indian island, xvi + 654 pp. Minneapolis. (2nd enlarged ed., 1960). DUNNEBIER, W. 1929-30. Spraakkunst van het Bolaang-Mongondowsch. BijdrTLV 85.297-468 (1929); 86.42-177 (1930).

. 1948. Het bemedicineeren van een dorp in het landschap Bolaang-Mongondow (Noord-Celebes). BijdrTLV 104.201-35. . 1951. Bolaang-Mongondowsch — Nederlandsch woordenboek met Nederlandsch-Bolaang Mongondowsch register, xn + 635 pp. The Hague. . 1953a. Bolaang Mongondowse teksten. vm + 238pp. The Hague. . 1953b. Ouman bo Tonggina. 82 pp. The Hague. DUNSELMAN, P. DONATUS. 1949-50. Bijdrage tot de kennis van de taal en adat der Kendajan Dajaks van West-Borneo. BijdrTLV 105.59-105 (1949); 106.321-73 (1950).

. 1954. Kana Sera of zang der zwangerschap. Een sacrale hymne der MualangDajaks. BijdrTLV 110.52-63. . 1955. Kana Sera. Zang der Zwangerschap. VKI 17.283 pp. + plates and map. . 1959. Uit de literatuur der Mualang-Dajaks. 79 pp. The Hague. . 1961. Ngebau tadjau, een kosmogonie der Mualang-Dajaks. Anthropos 56.409-37. DYEN, I. 1945. Spoken Malay. 2 vols., XIV + VII + 517 pp. . 1956. The Ngaju-Dayak 'Old speech stratum'. Lg 32.83-7. ECHOLS, JOHN M . , a n d HASSAN SHADILY. 1961. A n Indonesian-English dictionary,

xvi + 384 pp. ; 2nd ed. xvin + 431 pp. New York. (2nd ed. 1963). EDMONT-TES. 1957. Onomatope dalam bah. Karo. MeBa 7/1.23-4. EL LAMPUNY, M.W. 1958a. Sedikit tentang Bahasa Lampung. MeBa 8/3.36-40. . 1958b. Bidal Lampung. MeBa 8/12.32-4. . 1959. Tulisan Lampung. BaBu 8.45-8. EMEIS, M.G. 1945. Vorm en functie in klassiek en modern Maleisch. De verbale constructies. 168 pp. Utrecht. . 1946. Inleiding tot de Bahasa Indonèsia. (1st ed. Inleiding tot het algemeen beschaafd Maleisch) 1st ed. 136 pp. ; 2nd ed. 216 pp. ; 4th ed. 243 pp. Groningen, Batavia. (2nd ed. 1948; 4th ed. 1950). . 1948. De inversie in het Moderne Maleis. BijdrTLV 104.605-19. . 1953. Het Maleise flexie-type. BijdrTLV 109.321^41.

INDONESIA AND MALAYSIA

93

. 1955. Vocaal-harmonie in de Maleise kern van de Indonesische vocabulaire. BijdrTLV 111.191-201. ERINGA, F. S. 1949. Loetoeng Kasaroeng. Een mythologisch verhaal uit West-Java. Diss. Leiden ( = VKI 8). v n + 3 9 4 pp. ESSER, S. J. 1927, 1933. Klank- en Vormleer van het Morisch. 2 vols. VBG 67.3 and 67.4. 434 pp. . 1934. Handleiding voor de beoefening der Ledo taal. VBG 72.2. 90 pp. . 1938. Languages. Atlas van Tropisch Nederland, sheet 9, 9b. Amsterdam. . 1964. De Uma-taal (West Midden-Celebes), spraakkundige schets en teksten, uitgegeven en van een woordenlijst voorzien door Dr. J. Noorduyn. VKI 43. x i v + 144 pp. . See: Adriani and Esser. FÄHNRICH, H. 1966. Arabische Lehnwörter in der Bahasa Indonesia und in der georgischen Sprache. MIO 11.451-62. FARFAR, J . J .

1959. B a h a s a K a i . M e B a 9/1.44-7.

FIRTH, RAYMOND. 1946. Malay fishermen: their peasant economy, xn + 354 pp. London. FISCHER, H.TH. 1951. Makassaarse verwantschapstermen. BijdrTLV 107.61-6. . 1966. Toba Batak kinship terms. Oceania 36.253-63. , and H. C. VAN RENSELAAR. 1959. Over enkele Batakse verwantschapstermen. BijdrTLV 115.40-55. FOKKER, A.A. 1940. Over verbale vormen in het Maleisch. TBG 80.395-424. . 1941. Beknopte grammatica van de Bahasa Indonesia (lst-3rd ed.: Beknopte Maleise grammatica). 2nded. 154 pp., 4th ed. xi + 160 pp. Groningen, Djakarta. (2nd ed. 1946, 4th ed. 1950). . 1951. Bahasa Indonesia dan Bahasa Daerah. 27 pp. Groningen, Djakarta. . 1950-51. Mempersoalkan Bahasa Indonesia dan Daerah. PemBalnd 3.166-8. . 1951. Inleiding tot de Studie van de Indonesische syntaxis. x + 242pp. Groningen, Djakarta. Review: Anceaux, BijdrTLV 108.298-300 (1952). GALVIN,A.D. 1966. Some Baram Kenyah songs. SarawakMJ 14.6-14. GEDDES, W.R. 1954. The Land Dayaks of Sarawak. Colonial Research Studies 14.113 pp. London. . 1957. Nine Dayak nights. xxxi + 144pp. Melbourne. GERLACH. See: Habar Salamat. GERTH VAN WIJK, D. 1889. Spraakleer der maleische taal. (3rd ed. 1909). 303 + ix pp. Batavia. GEURTJENS, H. 1921a. Spraakleer der Keieesche taal. VBG 63/2.86 pp. . 1921b. Woordenlijst der Keieesche taal. VBG 63/3.196 pp. GONDA, J. 1941. Enkele opmerkingen over woordvorming. BijdrTLV 100.127-72. . 1942. Inwendige nasaal- en liquidaverbindingen in Indonesische talen. BijdrTLV 101.141-206.

E. M. UHLENBECK

94

. 1943. De zg. hulppraedicaatswoorden in Maleis en Javaans. BijdrTLV 102. 95-103. . 1946. De Javaanse den-vormen. BijdrTLV 103.267-305. . 1948. The Javanese vocabulary of courtesy. Lingua 1.333-76. . 1949a. Opmerkingen over Oud-Javaanse zinsleer. BijdrTLV 105.25-57. . 1949b. Prolegomena tot een theorie der woordsoorten in Indonesische talen. BijdrTLV 105.275-331. . 1949c. Over Indonesische werkwoordsvormen. BijdrTLV 105.333-421. . 1949-50. The functions of word duplication in Indonesian languages. Lingua 2.170-97. . 1949-52. Indonesian linguistics and general linguistics. Lingua 2.308-39 (1949-50); 3.17-51 (1952). . 1950a. Een type van syntaktische spaarzaamheid. BijdrTLV 106.47-53. . 1950b. Observations on ordinal numbers. Bingkisan Budi (Festschrift Van Ronkel), 135-45. Leiden. . 1951. Universele tendenzen in de Indonesische syntaxis. BijdrTLV 107.179200. . 1952. Sanskrit in Indonesia. Sarasvati Vihara Series 28. xxix + 456 pp. Nagpur. . 1953. Varia over Indonesische telwoorden. BijdrTLV 109.23-31. . 1958. Some notes on the relations between syntactic and metrical units in a Javanese kidung. BijdrTLV 114.98-116. . 1959. On Old-Javanese sentence structure. OE 6.57-68. GORIS, R . 1 9 3 8 - 3 9 . Beknopt Sasaksch-Nederlandsch Woordenboek. Mededelingen van de Kirtya Liefrinck-Van der Tuuk, 7 - 1 0 . 3 5 2 pp. Singaradja, Bali. . 1954. Inscripties voor Anak Wungçu. Prasasti Bali. 2 vols. 353 p p . + 10 plates. Bandung. . 1956. Penjelidikan Bahasa-bahasa dipulau Flores. BaBu 4/4.36—47. . 1957. Dinasti Warmadewa dan Dharmawangça dipulau Bali. BaBu 5/3.18-31. GYANTO. 1957. Hukum M. D. didalam bahasa Roti. MeBa 7/8.28. Habar salamat manjoe kasaksian Loekas. (Inside:) Lucas Maanjan-Dajak. 99 pp. (No place, no year). HADIDJAJA, TARDJAN. 1951. Pengaruh Bahasa Indonesia kepada Bahasa Djawa. MeBa 1 / 3 . 5 2 - 7 . . 1951-52. Pengaruh Bahasa Djawa kepada Bahasa Indonesia. MeBa 1/4.40-4 ( 1 9 5 1 ) ; 2 / 5 . 2 2 - 5 (1952).

S. 1958. Serba-serbi tentang peladjaran bahasa Daerah (Djawa) di Indramaju. MeBa 8 / 9 . 3 0 - 7 . HAMILTON, A. W . 1947. The first Dutch-Malay vocabulary. JMBRAS 2 0 / 2 . 2 0 - 5 . HARDAKER, M. B. 1963. An introduction to Suluk. SarawakMJ 1 1 . 1 3 8 - 7 3 . . 1964. An introductory vocabulary of Suluk. SarawakMJ 1 1 . 5 7 8 - 9 6 . HADISARTAMA,

INDONESIA AND MALAYSIA

95

A. 1858. Versuch einer Grammatik der Dajackschen Sprache. vni + 374 pp. Amsterdam. . 1859. Dajacksch-Deutsches Wörterbuch. vm + 638 pp. Amsterdam. HARDJOWIROGO, M . 1 9 2 6 . Patokaning njekaraken, BP 7 2 5 . 1 0 2 pp., Jav. char. Weltevreden. (Reprinted as BP 1 8 0 5 . 7 0 pp., Lat. char. 1 9 5 2 ) . HARIMURTI KRIDALEKSANA. 1 9 6 4 . Perhitungan leksikostatistik atas delapan bahasa nusantara barat serta penentuan pusat penjebaran bahasa 2 itu berdasarkan teori migrasi. M I S I 2 . 3 1 9 - 5 2 . . 1966. Pembentukan Istilah dalam Bahasa Indonesia. Simposium Bahasa dan Kesusastraan Indonesia, published by Lembaga Bahasa dan Kesusastraan, Direktorat Djenderal Kebudajaan Departemen P. dan K, B 43-B 57. Djakarta. HARRISSON, TOM. 1961. A Sarawak Kenyah journey through death. SarawakMJ 10.191-213. . 1965. Borneo Writing. BijdrTLV 121.1-57. . 1966. Three communication systems among Borneo nomads (and their dogs). JMBRAS 38/2.67-86. , and B. SAUDIN, eds. 1966. Borneo writing and related writings. SarawakMJ, special monograph, XII + 426 pp. HAZEU, G.A.J. 1 9 0 7 . Gajösch-Nederlandsch woordenboek met NederlandschGajösch register, xx+1148 pp. Batavia. HELFRICH, O . L . 1904. Bijdragen tot de kennis van het Midden Maleisch (Besemahsch en Serawajsch dialect). YBG 53.IV + 284 pp. . 1915. Nadere aanvullingen en verbeteringen op de Bijdragen tot de kennis van het Midden Maleisch (Besemahsch en Serawajsch dialect). VBG 61/3-4.1-28. . 1921. Supplement op de in deel LXI, 3e en 4e stuk, der Verhandelingen gepubliceerde 'Nadere aanvullingen en verbeteringen' op de Bijdragen tot de kennis van het Midden Maleisch (Besemahsch en Serawajsch dialect) (versehenen in deel LIII der Verhandelingen). VBG 63/1.66 pp. . 1927. Nadere bijdragen tot de kennis van het Midden Maleisch (Besemahsch en Serawajsch dialect). VBG 68/3.128 pp. . 1933. Bijdragen tot de kennis van het Midden Maleisch (Besemahsch en Serawajsch dialect). Supplement op de 'Nadere Bijdragen' (1927). VIII + 80 pp. The Hague. HENDON, RUFUS S. 1966. The phonology and morphology of Ulu Muar Malay (Kuala Pilah District, Negri Sembilan, Malaya). Yale University Publications in Anthropology 70, xvi+160 pp. Review: James N.Sneddon, Oceania 37. 316-17 (1967). HERRFÜRTH, H . 1964. Lehrbuch des modernen Djawanisch. 259 pp. Leipzig. HILGERS-HESSE, I. [1956]. Indonesisch. 101 pp. Köln. . 1960. Uber die Position der Regionalsprachen (als Träger alter Kulturen) gegenüber der Bahasa Indonesia im heutigen Indonesien. Tribus 9.147-52. . 1965. Entwicklungsgeschichte der Bahasa Indonesia. 115 pp. Heidelberg. HARDELAND,

96

E. M. UHLENBECK

1951. Kelantan padi planting. J M B R A S 24/1.56-76. . 1952. Some Kelantan games and entertainments. JMBRAS 25/1.20-34. HONZALI, H S . 1955. Menoleh ke Lampung. MeBa 5/7.20-9. HOOYKAAS, C. 1948a. Balische verhalen van den Halve. 96 pp. The Hague. . 1948b. Enige Sasakse volksvertelsels (Lombok). 93 pp. The Hague. HOOYKAAS, JACOBA. 1955. A journey into the realm of death. BijdrTLV 111.236-73. HOOYKAAS-VAN LEEUWEN BOOMKAMP, J. 1961. Ritual purification of a Balinese temple. VKNA NS 68.81 pp. HORNE, C. ELINOR. 1961. Beginning Javanese. Yale Linguistic Series 3. xxxm + 560 pp. New Haven, London. Reviews: Denzel Carr, JASt 21.568-9 (1961-62); Karl V. Teeter, Lg 39.146-51 (1963); E.M. Uhlenbeck, Lingua 12.69-86 (1963); R.S. Hendon, AmA 65.482-4 (1963); C. Hooykaas, OLZ 59.291-4 (1964). . 1963. Intermediate Javanese. Yale Linguistic Series 4. xxi + 505 pp. New Haven, London. HOWELL, W. f 1961. Supplement to Howell and Bailey's 'Sea Dyak dictionary' (with notes by T. Harrisson). SarawakMJ 10.127-69. , and D.J.S. BAILEY. 1900. A Sea Dyak dictionary. x n + 1 8 6 + 2 4 pp. Singapore. . 1909. An English-Sea Dyak vocabulary. 175 pp. Kuching, Sarawak. HUDSON, ALFRED B. 1967. The Barito isolects of Borneo: A classification based on comparative reconstruction and lexicostatistics. x + 1 1 2 pp. (Data Paper 68 Southeast Asia Program.) Ithaca, New York. ISKANDAR, T. 1958. De Hikajat Atjéh. YKI 26.205 pp. ISMAIL, J. 1 9 5 9 . Dialek Indramaju. MeBa 9 / 4 - 5 . 7 0 - 2 . Istilah Jawatan dan Jabatan. Siri Istilah DBP 1. iv + 175 pp. (1961); 3rd ed. x + 232 pp. (1964). Istilah Pertadbiran. Siri Istilah DBP 2, iv + 196 pp. (1962). Izui, HISANOSUKE. 1949. Études comparatives des langues du Sud. vin + 304 pp. Tokyo, Osaka. Review: E. Gaspardone, JA 240.259-62 (1952). JASPAN, M. A. 1964. Redjang Ka-ga-nga Texts. 92 pp. Canberra. . 1967. Symbols at work; aspects of kinetic and mnemonic representation in Redjang ritual. BijdrTLV 123.476-516. JONKER, J.C.G. 1894. Bimaneesche texten. VBG 48/2.127 pp. . 1896a. Bimaneesch-Hollandsch woordenboek. VBG 48/1.134 pp. . 1896b. Bimaneesche spraakkunst. VBG 48/3. xv + 487 pp. . 1906. Over de eind-medeklinkers in het Rottineesch en Timoreesch. BijdrTLV 59.263-343. . 1908. Rottineesch-Hollandsch woordenboek. 806 pp. Leiden. . 1911. Rottineesche teksten met vertaling. 135 pp. Leiden. . 1913. Bijdrage tot de kennis der Rottineesche tongvallen. BijdrTLV 68.521622. HILL, A . H .

INDONESIA AND MALAYSIA

97

. 1915. Rottineesche spraakkunst. xxiv + 714pp. Leiden. . 1932. Lettineesche taalstudien. VBG 69.VU + 252 pp. (With a preface by P. Yoorhoeve). . 1935. Soembawareesche teksten met vertaling. BijdrTLV 92.211-336. KAHLER, H. 1940a. Sichule Texte. VBG 74/4.41 pp. . 1940b. Grammatischer Abriss des Enggano. ZES 30.81-117, 182-210, 296320. . 1942-45. Uber Mischsprachen in Indonesien. Anthropos 37-41.889-90. . 1946-49. Ethnographische und linguistische Studien von den Orang laut auf der Insel Rangsang an der Ostküste von Sumatra. Anthropos 41-4.1-31, 75785. . 1954-64. Texte von der Insel Enggano (Westküste von Sumatra). AuÜ 39.89-94 (1954-55); 41.153-6 (1957); 42.179-89 (1958); 43.299-312 (1959); 44.41-71 (1960); 45.205-25 (1961); 46.127-^15 (1962-63); 48.46-58 (1964). . 1955. Die Sichule-Sprache auf der Insel Simalur an der Westküste von Sumatra. AuÜ Beih. 27.75 pp. Review: A. Teeuw, BijdrTLV 115.397-9 (1959). . 1956. Grammatik der Bahasa Indonèsia. Porta Linguarum Orientalium, N.S. 2.VII + 307 pp. Wiesbaden. Review: A. Teeuw, OLZ 54.67-74 (1959). . 1959. Vergleichendes Wörterverzeichnis der Sichule-Sprache auf der Insel Simalur an der Westküste von Sumatra. VSISUH l.m + 90 pp. ——. 1960. Ethnographische und linguistische Studien über die Orang darat, Orang akit, Orang laut und Orang utan im Riau-Archipel und auf den Inseln an der Ostküste von Sumatra. VSISUH 2.VI + 294 pp. ——. 1961. Simalur-Deutsches Wörterbuch mit Deutsch-Simaluresischem Wörterverzeichnis. VSISUH 3.IX + 281 pp. . 1963. Texte von der Insel Simalur. VSISUH 4.xxxi + 3 8 7 pp. . 1965. Dialect and language (investigated with some examples from Indonesian languages). Lingua 15.497-514. . 1966. Wörterverzeichnis des Omong Djakarta. VSISUH 5.xxn + 351 pp. Review: A. Teeuw, BijdrTLV 122.480-3. KAHLO, G . , and R. BÄRWINKEL. 1963. Indonesisch-Deutsches Wörterbuch. xm + 510 pp. Leipzig. KAMIL, T . W . 1 9 6 4 . Perbandingan beberapa pandangan tentang konsepsi morfem dan saran 2 mengenai adaptasi konsepsi tersebut dalam bahasa 2 Nusantara. MISI 2.301-18.

, and A . M . MOELIONO. 1961. Beberapa Patokan dan Saran untuk Pelaksanaan Linguistik di Indonesia. MellPe 2.9-19. , and SUGENG SUKARSONO. 1960. Beberapa morfem jang produktif dalam Bahasa Indonesia. BaBu 9.58-78. KAROW, OTTO, and IRENE HILGERS-HESSE. 1 9 6 2 . Indonesisch-Deutsches Wörterbuch. Kamus Bahasa Indonesia-Djerman. xix + 483 pp. Wiesbaden. KENNEDY, R . 1 9 5 5 . Bibliography of Indonesian peoples and cultures. 2nd ed.

98

E. M. UHLENBECK

revised by Th. W. Maretzki and H.Th. Fischer. 2 vols, XXVIII + 663 pp. New Haven. KERN, H. 1898-1906. Bijdragen tot de Spraakkunst van het Oudjavaansch. BijdrTLV 49.635-54 (1898); 50.96-110, 231-46, 401-4 (1899); 51.263-71 (1900); 53.161-83, 512-31 (1901); 54.173-94 (1902); 55.345-57 (1903); 59.229-62 (1906). = VGH 8.135-323 (1918); 9.1-28 with an appendix (1920). KERN, R. A. 1939. Gender in Buginese. AO 17.228-34. . 1940. Boegineesche en Makassaarsche taalkundige bijdragen. BijdrTLV 99. 295-338. . 1950. De conjunctief in het Boeginees. Bingkisan Budi ( = Festschrift Van Ronkel), 162-9. Leiden. KERSTEN, J. 1948. Balische Grammatica. 102 pp. The Hague. KILIAAN, H.N. 1897. Madoereesche spraakkunst. Vol. 1, 172 pp.; vol. 2, 192 pp. Batavia. (Reprinted Semarang, 1911). . 1904-05. Madoereesch-Nederlandsch woordenboek. Vol. 1, vn + 462 pp.; vol. 2, 384 pp. Leiden. K I T I N G , C H . 1958. Imbuhan awalan ba dan ha dalam bahasa Kahajan (Dajak Ngadju). MeBa 8/4.23-6. . 1959a. Bahasa Kahajan ( = Dajak Ngadju) selajang pandang. MeBa 9/3. 34—40. . 1959b. Tatabahasa bahasa Dajak Ngadju (Kahajan). MeBa 9/4-5.61-6. . 1959c. Perbandingan bahasa. BaBu 8.41-4. , and W . C O R . D U L L A H . 1958. Sedjemput tatabahasa Dajak Siang. MeBa 8/9.38-40. KOESNODIPRODJO. 1954. Kamus singkat dari singkatan-kata dalam-negeri dan luarnegeri, dengan terdjemahan dan pendjelasannja. Rev. and enl. ed. 497 pp. Djakarta. (2nd ed. 1955).

Kongres Bahasa dan Persuratan Malaya 3d Singapore. Laporan tentang bahasa Melayu Malaya dan bahasa Indonesia, dikemukakan oleh Persekutuan bahasa Melayu University Malaya (1956). KORIGODSKIJ, R.N., O . N. KONDRASKIN, and B . I . ZINOV'EV. 1961. Indonezijskorusskij slovar. 1171 pp. Moscow. K R U P A , V., and G. ALTMANN. 1961. Semantic analysis of the system of personal pronouns in Indonesian language. AO 29.620-5. Review: T.W. Kamil, AO 3 0 . 6 3 2 - 3 (1962).

I Gunawati. BaBu 7 . 1 2 9 - 8 6 . . 1961. Prasasti Baru Radja Marakata. BaBu 9.3-17. LEACH, E. R. 1950. Social science research in Sarawak. Colonial Research Studies 1. 93 pp. LEES, SHIRLEY P . 1959. Lun Daye phonemics. SarawakMJ 9 . 5 6 - 6 2 . • . 1965. An introduction to the sound system of L6pu' Tau. SarawakMJ 12.179-84.

K T U T GINARSA.

1959.

INDONESIA AND MALAYSIA

99

LEPAGE, R.B. 1964. Multilingualism in Malaya. Symposium on Multilingualism held in Brazzaville (16-21 July 1962) = Publication 87 of the Scientific Council for Africa, 133-46. London. LEWIS, M. BLANCHE. 1960. Moken texts and word-list, a provisional interpretation. Federation Museums Journal 4 N.S. XII + 102 pp. LIAW YOCK FANG. 1967. A transformational approach to Malay syntax. International Conference on the Modernization of Languages in Asia. 19 pp. . 1967-68. Ilmu Bahasa dan Penggolongan Jenis Kata dalam Bahasa Melayu. D e w a n B a h a s a 11.304-16, 5 5 2 - 9 (1967); 12.33-45 (1968).

LIE,T. S. 1965-66. Introducing Indonesian. Book I, 115 pp. (1965); Book II, 184 pp. (1966).

Sydney.

LINEHAN, W. 1949. The earliest word-lists and dictionaries of the Malay language. JMBRAS 22/1.183-7. LOEB, EDWIN. 1929a. Mentawei myths. BijdrTLV 85.66-244. . 1929b. Mentawei religious cult. UCPAAE 25/2.185-247. . 1944. Javanese word formation, high and low. JAOS 64.113-26. MAAN, G. 1940. Boelisch-Nederlandsche woordenlijst met Nederlandsch-Boelisch register. VBG 74/3. vi + 1 8 7 pp. . 1951. Proeve van een Bulische spraakkunst. VKI 10.160 pp. MAJUSIM BIN MAJINAL, E. 1949. Some notes on the Dusun language. SarawakMJ 5.125-9. MANAFE, MINGGUS D. 1959. Hukum M. D. dan bunji sengau dalam bahasa Roti. MeBa 9/4-5.67-9. . 1960. Tiga buah tjeritera dari Roti Lelenuk. BaBu 9.247-53. MARDIWARSITA, SMD. 1957. Hukum M. D. didalam Bahasa Roti. MeBa 7/10.20-1. MARYOTT, ALICE. 1963. The nuclear predication in Sangir. PhilJSci 92/1.111-20. MARYOTT, KENNETH R. 1961. The phonology and morphophonemics of Tabukang Sangir. PSSHR 26/1.111-26. . 1963. The substantive phrases of Sangir. Papers on Philippine languages 1.50-62. Manila. MEES, C. A. 1927. Maleise spraakkunst. [ 1 st ed.: Beknopt overzicht van de Maleische grammatica, 2nd, 3rd ed.: Beknopte Maleische grammatica]. 1st ed. 124 pp., 4th ed. 291 pp. Santpoort. (2nd ed. 1931, 3rd ed. The Hague 1938, 4th ed. 1946). More recent enlarged and revised editions written in Indonesian appeared since 1951 under the title Tatabahasa Indonesia. MIDDELKOOP, P. 1939. Amarasisch-Timoreesche teksten. VBG 74/2.105 pp. (Texts of Prof. J. C. G. Jonker annotated and translated by Middelkoop.) . 1950. Proeve van een Timorese grammatica. BijdrTLV 106.375-517. . 1958. Four tales with mythical features characteristic of the Timorese people. BijdrTLV 114.384-405. . 1959. A Timorese myth and three fables. BijdrTLV 115.157-75.

100

E. M. UHLENBECK

. 1963. Head hunting in Timor and its historical implications. OLM 8 a, b, c. 423 + 5 pp. MIYATAKE SEIDO. 1941. Konsaisu Marai-go shinjiten. Kamoes baroe bahasa Indonesia-Nippon. 3 3 4 pp. Tokyo. (5th ed. 1943). . 1943. Hyöjun marai-go daijiten. Kamoes Bahasa Melajoe (Indonesia)-Nippon jang lengkap. 1774 pp. Tokyo. MOMON WIRAKUSUMAH, R . , and I . BULDAN DJAJAWIGUNA. 1957. Tatabasa Sunda. 82 pp. Bandung, Djakarta. MORRIS, MAX. 1900. Die Mentawai-Sprache. vm+356 pp. Berlin. MUHADJIR. 1964. Dialek Djakarta. MISI 2.25-52. MUHAMMAD ZAIN, ST. 1954. Kamus moderen bahasa Indonesia. 1090 pp. Djakarta. Review: Prijana, BaBu 7.32-44 (1958). NEEDHAM, R. 1954a. The system of teknonyms and death names of the Penan. S J A 10.416-31. . 1954b. A Penan mourning usage. BijdrTLV 110.263-7. . 1955a. PunanBa. JMBRAS 28.24-36. . 1955b. More concerning the Punan of Borneo. AmA 57.131. . 1959. Mourning terms. BijdrTLV 115.58-9. NEUMANN, J. H. 1951. Karo-Bataks-Nederlands woordenboek. 343 pp. Medan. NICOLSPEYER, M. M. 1940. De sociale structuur van een Aloreesche bevolkingsgroep. Diss. Leiden. vi +180 pp. + 2 maps. Rijswijk. NIGGEMEYER, H. 1951-52. Alune-Sprache. Texte, Wörterverzeichnis und Grammatik einer Sprache West-Cerams. ZEthn 76.50-69, 288-300 (1951); 77.116-32, 238-50 (1952). Nio JOE LAN. 1960. Bahasa Melaju-Tionghoa. BaBu 9.201-15. NOORDUYN, J . 1955a. Een Achttiende-eeuwse kroniek van Wadjo', Buginese historiografie. Diss. Leiden. 332 pp. The Hague. . 1955b. Een lange pepet in een Buginees dialect. BijdrTLV 111.415-16. . 1959. Batu tulis di Bogor. Indonesia, madjalah kebudajaan. 10/11.504-7. . 1962a. Over het eerste gedeelte van de Oud-Soendase Carita Parahyangan. BijdrTLV 118.374-83. . 1962b. Het begingedeelte van de Carita Parahyangan. BijdrTLV 118.403-32. . 1963. Mededelingen uit de verslagen van Dr. S. J. Esser, taalambtenaar voor Celebes 1928-44. BijdrTLV 119.329-70. . 1966. Enige nadere gegevens over tekst en inhoud van de Carita Parahyangan. BijdrTLV 122.366-74. , see Perdjangdjian Anjar. NOORMALI. 1956. Seni Sastra Lampung. MeBa 6 / 4 - 5 . 1 - 6 . NYANDOH, R. 1956. The story of KumangRuwai. Sirituh Kumang Ruwai. Sarawak MJ 7.208-20. . 1959. Lingagat and the magical faeces (a Land Dayak bird story). SarawakMJ 9.49-52.

INDONESIA AND MALAYSIA

101

. 1963. Seven Land Dayak stories of Sarawak and Kalimantan. SarawakMJ 11. 114-31. ONVLEE, L. 1925. Eenige Soembasche vertellingen. Diss. Leiden. Lxxix + 139pp. Leiden. . 1929. Palatalisatie in eenige Soembasche dialecten. Feestbundel Kon. Bataviaasch Genootschap 11.234-45. . 1933. Na Hoeri Häpa, eenige regelen en zegswijzen betrekking hebbend op het sirihpruimen van den Soembanees. TBG 73.476-94. . 1949. Naar aanleiding van de stuwdam in Mangili (Opmerkingen over de sociale structuur van Oost-Soemba). BijdrTLV 105.445-59. . 1950. Over de mediae in het Soembanees en Soewanees. Bingkisan Budi ( = Festschrift Van Ronkel), 215-34. Leiden. . 1956. Keterangan tentang bahasa dan kebudajaan suku bangsa Sumba. BaBu 4/4.48-50. . 1961. Na Parändingu Bidi. 689 pp. Djakarta. OPLT, M . 1960. Bahasa Indonesia. Uöebnice Indonöstiny-Indonesian language. 357 pp. Prague. (2nd rev. ed. 395 pp. 1966). PADMOSOEKOTJO, S. 1956. Sarine basa Djawa. 148 pp. Djakarta. PAKAN, L. 1957. Memperkenalkan kebudajaan dan kesenian daerah Toradja kepada masjarakat Indonesia. BaBu 5/5.25-6. . 1958. Aneka ragam tentang bahasa Toradja. MeBa 8/4.20-2. . 1959a. Memperkenalkan beberapa tarian suku Toradja-Sa'dan. BaBu 7.123-6. . 1959b. Kebudajaan orang Toradja. BaBu 7.199-211. . 1959c. Njanjian badong, berupa puisi. MeBa 9/3.29-33. PARNICKEL, B . , and Ü . SIRK. 1965. Austronesian philology in the Soviet Union. BijdrTLV 121.245-58. PÄTSCH, GERTRUD. 1964a. Verbale und nominale Fügungen im Nias. ZPhon 17.597608.

. 1964b. Reste einer personalgegliederten Konjugation im Indonesischen. MIO 10.171-81. PAVLENKO, A . P . 1965. Sundanskij jazyk. 83 pp. Moscow. . 1967. Predvaritel'nye dannye intonografiöeskogo issledovanija zvukov indonezijskogo jazyka. Jazyki Jugo-Vostoönoj Azii, 1 1 7 - 2 5 . Moscow. PAYNE, E.M.F. 1964. Basic syntactic structures in Standard Malay. Unpublished thesis Univ. of London. 143 pp. PERCIVAL, KEITH. 1960. A sketch of Toba Batak phonology. MPhon 28-30. PERDANAKUSUMAH, A . 1956a. Pendahuluan tentang paramasastra Sunda. MeBa 6/1.22-6. . 1956b. Pramasastera Bahasa Sunda. MeBa 6/2-3.16-25. . 1956c. Sambungan Pramasastera Sunda. MeBa 6/7.23-9,

102

E. M. UHLENBECK

. 1956-57. Paramasastra Ba(ha)sa Sunda. MeBa 6/10.23-9, 6/11.17-27 (1956); 7/4.31-4, 7/9.24-32 (1957). Perdjangdjian Anjar. 1962. 651 pp. Djakarta. (The New Testament, translation revised by J. Noorduyn and N. Titus.) PIGEAUD, TH. 1938. Javaans-Nederlands handwoordenboek. XII+624 pp. Groningen, Batavia. . 1948. Nederlands-Javaans handwoordenboek. VII+663 pp. Groningen, Batavia. . 1960-63. Java in the fourteenth century, a study in cultural history. Translation Series of the Royal Institute of Linguistics and Anthropology 4. Vol. 1, xvi + 125 pp.; vol. 2, xvi + 153 pp.; vol. 3, xin + 177 pp.; vol. 4, x + 552 pp.; vol. 5, v i n + 451 pp. The Hague. PINNOW, HEINZ-JÜRGEN. 1963. The position of the Munda-languages within the Austroasiatic language-family. Linguistic comparison in South East Asia and the Pacific, ed. by H . L . Shorto, 1 4 0 - 5 2 . London. PINO, E. 1950. Bahasa Indonesia, The national language of Indonesia. 152+32 pp Groningen, Djakarta. , and T. WITTERMANS. 1953. Kamus Inggeris. English-Indonesian ix +468 pp., Indonesian-English v m + 170 pp. Djakarta, Groningen. POEDJAWIJATNA, I.R., and P . J . ZOETMULDER. 1955. Tatabahasa Indonesia untuk Sekolah Landjutan Atas. 2 vols. 135 and 137 pp. Djakarta. POERBA, J . D . 1960. Simalungun dan bahasanja. BaBu 8 . 2 3 8 - 4 7 . POERWADARMINTA, W. J . S. 1942. Baoesastra Indonesia-Djawi. B P 1450. 3rd ed. 203 pp. Djakarta. (2nd ed. 1943, 3rd ed. 1945). . 1952. Kamus Umum Bahasa Indonesia. BP 1866,903 pp. Djakarta. (2nded. 1954; 3rd ed. 1961). , with the assistance of C . S. HARDJASOEDARMA, J . CHR. POEDJASOEDIRA. 1939. Baoesastra Djawa. VIII + 670 pp. Groningen, Batavia. , and A . TEEUW. 1950. Indonesisch-Nederlands woordenboek. xi+369 pp., 2nd ed. 383 pp. Groningen, Djakarta. (2nd ed. 1952). POERWANTO DANOESOEGONDO. 1966. Bahasa Indonesia for beginners. Vol. 1,160 pp. Sydney. POETZELBERGER, H. A. 1965. Einführung in das Indonesische. 147 pp. Wiesbaden. PRAWIRASUGANDA, A., and S. SAUNI. 1951. Kitab peladjaran bahasa Djawa-Kuna. Vol. 1, 97 pp. (3rd ed. 104 pp.) Bandung. (2nd ed. 1953, 3rd ed. 1954). Only the first volume is available in The Netherlands. PRAWIROATMODJO, S. 1953. Konklusi paramasastra beserta persamaannja DjawaIndonesia. Surabaya. Not available in The Netherlands. PRENTICE, D.J. 1965. Form and function in the verbs of Sabah Murut: A preliminary analysis. OL 4.127-56. PRIJANGGANA, M.S. 1957. Sedikit tentang bahasa Banjuwangi. BaBu 6/2.32-6. RACHMAT MARLIN. 1955. Beberapa untai pantun dari kampung. MeBa 5 / 5 . 2 7 - 8 .

INDONESIA AND MALAYSIA

103

1949. Bahasa Indonesia. Eenvoudige leergang voor de Indonesische taal. 2 vols. 106 and 152 pp. Batavia. RAMLAN, M. 1964. Immediate constituents (IC) dalam morfologi dan sintaksis Indonesia. MISI 2.219-24. . 1965. Kata-madjemuk dalam bahasa Indonesia. MISI 3.39-48. RAS, J.J. 1968. Hikajat Bandjar, A study in Malay historiography, diss. Leiden = Bibliotheca Indonesica I. xm + 651 pp. The Hague. ROBINS, R . H . 1953a. Formal divisions in Sundanese. TPhS 109-42. . 1953b. The phonology of the nasalized verbal forms in Sundanese. BSOAS 15.138—45. . 1957. Vowel nasality in Sundanese, a phonological and grammatical study. Studies in Linguistic Analysis. Special vol. Philol. Soc., 87-103. Oxford. . 1959. Nominal and verbal derivation in Sundanese. Lingua 8.337-69. . 1965. Some typological observations on Sundanese morphology. Lingua 15.435-50. ROESLI. 1967. Peladjaran 'Bahasa Minangkabau'. 74 pp. Djakarta. ROOLVINK, R . 1948. De voorzetsels in Klassiek en Modern Maleis. Diss. Utrecht x i x + 230 pp. Dokkum. . 1950-51. Kubuse teksten. TBG 84.10-32. . 1953. Dialek Melaju di Deli. BaBu 3/1.3-40. . 1965. The passive-active per/ber//per-/memper- correspondence in Malay. Lingua 15.310-37. SAFIAH BT. H J . A . KARIM, N I K . 1966. Loghat Kelantan, Huraian Fonoloji dan Chatetan Pendek mengenai Sifat2 Umum-nya. Dewan Bahasa 1 0 . 2 5 8 - 6 4 . SALZNER, R. 1960. Sprachenatlas des Indopazifischen Raumes. 2 vols. vol. 1, VIII + 138 pp.; vol. 2, Karten 1-64. Wiesbaden. SAMSURI, M. A . 1960. Sistem Fonem Indonesia dan Suatu Penjusunan Edjaan Baru. MellPe 1 . 3 2 3 - 4 1 . . 1961. Javanese phonemes and their distinctive features. MellPe 2.4. SANJOTO, M., and ISM. JUSRAN. 1957. Bahasa Bandjar sepintas lalu. MeBa 7 / 3 . 1 5 - 9 . SANTA MARIA, LUIGI. 1967. I Prestiti Portoghesi nel Malese-Indonesiano. v + 130 pp. Naples. SARUMPAET, J.P., and J. A . C . MACKIE. 1966. Introduction to Bahasa Indonesia. 115 pp. Melbourne. SATHER, CLIFFORD A. 1964-65. Bajau Numbers and Adjectives of Quantity. Sabah Society Journal 2, 194-197. . 1966. Bajau Laut Phonology and Grammar. Sabah Society Journal 3, 205-224. SATJADIBRATA, R. 1931. Rasiah tembang Soenda, BP 940. 79 pp. Batavia. (2nd augmented ed. 99 pp. 1953). . 1943. Oendak-oesoek basa Soenda, BP 1506. 52 pp. Djakarta. (2nd ed. 72 pp. 1956). RAMBITAN, M . H .

104

E. M. UHLENBECK

. 1944. Kamoes Soenda-Indonesia (1st ed. : Kamoes Soenda-Melajoe), BP 1561, 1st ed. 379 pp., 2nd ed. 414 pp. Djakarta. (2nd ed. 1950). . 1948. Kamoes basa Soenda, BP 1692. 448 pp. Djakarta. (2nd ed. 479 pp. 1954). SAUDIN, B. 1966. A Saribas Iban death dirge. SarawakMJ 14.15-80. SCHÄRER, H. 1946. Die Gottesidee der Ngadju Dajak in Süd-Borneo. VII + 236 pp., plates and map. Leiden. . 1963. Ngaju religion, the conception of God among a South Borneo people. x v + 229 pp., plates and map. The Hague. (Translation of Schärer 1946 by Rodney Needham with a preface by P. E. de Josselin de Jong). . 1966. Der Totenkult der Ngadju Dajak in Süd-Borneo. VKI 51/l-2.xv + xi + 963 pp. SCHMIDGALL TELLINGS, A . E D . 1 9 5 5 . Kasip. Kamus singkatan Indonesia pertama. 417 pp. Djakarta. . [1957]. Indonesian for today. A practical course for English speaking people. 216 pp. Djakarta. SCHWARTZ, J. ALB.T. 1907. Tontemboansche teksten. 3 vols. Vol. 1, 340 pp.; vol. 2, 513 pp.; vol. 3, 175 pp. Leiden. . 1908. Tontemboansch-Nederlandsch woordenboek, met Nederlandsch-Tontemboansch register. 690 pp. The Hague. SCOTT, N.C. 1956. A dictionary of Sea Dayak. xi + 218 pp. London. . 1957. Notes on the pronunciation of Sea Dayak. BSOAS 20.509-12. . 1964. Nasal consonants in Land Dayak (Bukar Sadong). In honour of Daniel Jones, 432-6. London. SENSTIUS, MAURICE W. 1943. A survey of aids for the study and teaching of the Malay language. 64 pp. Ann Arbor. SHORTO, H. L. (ed.). 1963. Linguistic comparison in South East Asia and the Pacific. 159 pp. London. SIRK, Ü . C H . 1 9 6 4 . Ob aktual'nom clenenii i formal'no-grammatiöeskoj strukture predlozenija v indonezijskom jazyke. KSINA 68.174-84. Moscow. SJAMSJI, D.N. 1958. Saling mengisi kekurangan dalam bahasa Dajak Ngadju. MeBa 8/10-11.49-54. SJAMSUDDIN, H. 1959. Memperkenalkan sedikit bahasa Bima. BaBu 8.22-34. SKINNER, C . 1 9 6 4 . The influence of Arabic upon modern Malay. Intisari 2 / 1 . 3 4 - 4 7 . SLAMETMULJANA, R. B. 1954. Poëzie in Indonesia. Een literaire en taalkundige Studie, thesis Leuven. Bibliothèque du Muséon 36.xvi + 248 pp. Leuven. . 1956-57. Kaidah Bahasa Indonesia. 2 vols. Vol. 1, xn + 207 pp.; vol. 2, x v i + 256 pp. Djakarta. Review: A.Teeuw, BijdrTLV 116.485-8 (1960). . 1959. Politik Bahasa Nasional. Inaugural lecture. 52 pp. Djakarta. Review: A. Teeuw, BijdrTLV 116.488-9 (1960). SNOUCK HURGRONJE, C. 1893. Studien over Atjèhsche klank-en schriftleer. TBG

INDONESIA AND MALAYSIA

105

35.346-442. Also published in the collected works of C. Snouck Hurgronje ( = Verspreide Geschriften 5.169-247). . 1900. Atjehsche taalstudieen. TBG 42.144-262. Also published in the collected works of C. Snouck Hurgronje ( = Verspreide Geschriften 5.285-372). . 1906. The Achehnese, translated by A. W. S. O'Sullivan, with an Index by R.J. Wilkinson. Vol. 1, xxi + 439 pp.; vol. 2, vi + 384 pp. Leiden. SOETIKNO. 1953. Serat pitedah. Bab panjeratipun basa Djawi ngangge aksara Latin. MeBa 3/6.33-6, 3/7.31-7. SOEWADJI SJAFEI. 1966. 'Pürwa$ästra', Kitab Peladjaran Bahasa Kawi. 170 pp. Djakarta. SPAT, C. 1900-01. Maleische taal. 2 vols., 247 and 223 pp., 3rd ed. of vol. 1,275 pp., 5th ed. 214 pp. Breda. (3rd ed. 1916, 5th ed. 1931.) SPITZBARDT, HARRY. 1964a. Automatische Synthese indonesischer Syntagmata. M I O 10.3-56. -——. 1964b. Review of N. F. Alieva, Glagol w indonezijskom jazyke. Problemy affiksacii (Moskva, 1963). Deutsche Literaturzeitung 85.629-33. . 1967. An algorithm for syntactic permutations in Indonesian. Asian and African Studies 3.13-22. STEINHART, W . L . 1934, 1937. Niassche teksten met Ned. vertaling en aanteekeningen. TBG 74.326-75, 391-440 (1934); VBG 73.vm + 316 pp. (1937). . 1938. Niassche priesterlitanieen opgeteekend op de Batoe-eilanden, met vertaling en aanteekeningen. VBG 74/1.X + 63 pp. . 1950-51. Niasse teksten. TBG 84.33-109. . 1954. Niasse teksten, vertaald en van aantekeningen voorzien. 264 pp. Den Haag. STELLER, K . G . F . , and W . E . AEBERSOLD. 1 9 5 9 . Sangirees-Nederlands woordenboek met Nederlands-Sangirees register, xi + 622 pp. 's-Gravenhage. STEVENS, ALAN M. 1 9 6 5 . Language levels in Madurese. Lg 4 1 . 2 9 4 - 3 0 2 . . 1966. The Madurese reflexes of Proto-Malayopolynesian. JAOS 86.147-56. . 1968. Madurese phonology and morphology. American Oriental Series Vol. 52. v i i + 2 1 5 pp. STRESEMANN, E. 1918. Die Paulohi-Sprache. Ein Beitrag zur Kenntnis der Amboinischen Sprachengruppe, xxm + 243 pp. The Hague. . 1927. Die Lauterscheinungen in den Amboinischen Sprachen. ZES Beih. IO.XII + 2 2 4 p p . + 1

map.

1956. Bahasa Indonesia Kupang. MeBa 6/10.11-15. H. 1913. Der Dialekt der Olon Maanjan (Dajak) in Süd-Ost-Borneo. BijdrTLV 67.203-36. SUTARDJO, K S . 1 9 5 8 . Huruf 'r' bahasa Indonesia dan huruf 'h' bahasa Bali. MeBa SUHARTONO.

SUNDERMANN,

8/5.18-20.

1949. Baiische bezweringsspreuken. TBG 83.159-76. . 1950a. A literal translation of Mark 1 in Balinese. BT 1.75-8.

SWELLENGREBEL, J . L .

106

E. M. UHLENBECK

. 1950b. Bible translation and politeness in Bali. BT 1.124-30. . 1950-51. Balische zegswijzen. TBG 84.159-208, 280-325. ——. 1963. Politeness and translation in Balinese. BT 14.158-64. TAIB BIN OSMAN, MOHD. 1964a. Trends in modern Malay literature. Malaysia, a survey, ed. by Wang Gungwu, 210-24. London. . 1964b. A text on the rules of the Kelantan bullfight. JMBRAS 37/2.1-10. TAKDIR ALISJAHBANA, ST. 1949. The Indonesian language, by-product of nationalism. Pacific Affairs 12.388-92. . 1949-50. Tatabahasa Baru Bahasa Indonesia. 2 vols. 96 and 115 pp. Djakarta. . 1950-5la. Prof. Fokker tentang Bahasa Indonesia dan Bahasa Daerah. PemBalnd 3.161-5. . 1950-5lb. Tentang Bahasa dan Perguruan Tinggi Indonesia. PemBalnd 3.169-72. . 1954. Bahasa Indonesia dan bahasa daerah. Konfrontasi 1.18-34. Also separately: 20 pp. Djakarta. (1954). . 1956. Le développement de la langue et de la littérature indonésiennes. CHM 2.682-703. . 1962. Indonesian language and literature: Two essays. Yale University Southeast Asia Studies. Cultural Report Series 11.II + 40 pp. New Haven. . 1965. New national languages: a problem modern linguistics has failed to solve. Lingua 15.515-30. TAKETOMI MASAKAZU. 1942. Marai-go daijiten. (Large dictionary of the Malay language) 1075 pp. Tokyo. TANG TJIA HAN. 1960. Awalan kata kerdja 'ter' dalam bahasa Indonesia. BaBu 8.136-68. 1956. Awalan'ter'dalam bahasa Karo. MeBa 6 / 1 1 . 1 3 - 1 5 . - — . 1958. Tjerita2 dari Simalungun. BaBu 7.46-51. . 1958-60. Beberapa folklore dari Simalungun. BaBu 7.52-75 (1958), 114-22 (1959), 212-34 (1959); 9.134-96 (1960). . 1959. Serba-serbi dari Simalungun. BaBu 7.240-57. . 1960a. Serba-serbi dari Karo dan Simalungun. BaBu 8.99-106. . 1960b. Fonologi. Bandung. . 1965. Nure-nure di Karo, Seri Budaja Karo I. vi + 87pp. Bandung. . 1966a. Katatugas dalam Bahasa Karo. v + 30pp. Bandung. . 1966b. Katadasar Bahasa Karo. vi + 65 pp. Bandung. , and T.M. UNGASSP. K. 1956. Awalan 'er' dalam bah. Karo. MeBa 6/4-5. 45-7. . 1957. Awalan 'mar' dalam bahasa Simelungun. MeBa 7/1.26-8. TAULU, H.M. 1955. Bahasa Gorontalo. MeBa 5 / 1 0 - 1 2 . 3 2 - 8 . . 1956a. Sjair2 Persatuan Gorontalo. MeBa 6/2-3.10-13. . 1956b. Suara Rakjat dalam adat istiadat Gorontalo pada menobatkan radja. MeBa 6/4-5.27-36. TARIGAN, H . G .

INDONESIA AND MALAYSIA (

107

1957. Bahasa Melaju-Manado. MeBa 7 / 5 . 3 3 - 7 . A. 1950. Rijst en zo op Lombok. Bingkisan Budi ( = Festschrift Van Ronkel), 281-90. Leiden. . 1951. Dialectatlas van Lombok. vn + 78maps. Djakarta. . 1952. Taal en Versbouw. 27 pp. Amsterdam. . 1953a. lets over een historisch geschrift van Lombok. BijdrTLV 109.193-226. . 1953b. Over een dialectgeografisch onderzoek in Indonesie. TeT 5.145-63. . 1954. Balische invloed op het Sasaks. BijdrTLV 110.116-31. . 1955. De Bahasa Indonesia, De Wereld en Nederland. 23 pp. Amsterdam. . 1957. Over E en O in Sasakse dialecten. BijdrTLV 113.32-49. ), TOMOHON.

TEEUW,

. 1958. Lombok.

Een dialect-geografische Studie. VKI 25.VI + 247 pp. and

maps. . 1959. The history of the Malay language. BijdrTLV 115.138-56. (with the assistance of H. W. Emanuels). 1961. A critical survey of studies on Malay and Bahasa Indonesia. BS 5.176 pp. . 1962. Some problems in the study of wordclasses in Bahasa Indonesia. Lingua 11.409-21. . 1962-63. Tentang Pergandaan Dalam Rangka Morfoloji Bahasa Indonesia. Bahasa 3/1.37-49. . 1965. Old Balinese and comparative linguistics. Lingua 14.271-84. TENRIADJI, A . , and G . J . WOLHOFF. 1955. Lontar Mandar. BaBu 3 / 3 . 7 - 3 1 , 3 / 4 . 2 5 - 3 8 . TESELKIN, A.S. 1959. O 'sprjagaemykh formakh glagola' v indonezijskom jazyke. KSIV 29.26-38. . 1961. Javanskij jazyk. 73 pp. Moscow. . 1963. Drevne-Javanskij jazyk (Kawi). 79 pp. Moscow. , and N . F . ALIEVA. 1960. Indonezijskij jazyk. 59 pp. Moscow. TEUTSCHER, H . J . 1954. Wat gaat men straks op Nieuw-Guinea spreken? Wending 9.113-29.

M. 1935. Kamoes Bahasa Minangkabau-Bahasa Melajoe-Riau. 271 + v n p p . Batavia. TOBING, P H . O. L. 1961. Hukum Pelajaran dan Perdagangan Amanna Gappa. 202 pp. Makassar. TRAJONO. 1959. Dua upatjara, waktu ada perajaan perkawinan di Tjilatjap. BaBu 7/5.189-98. TRAJONO SASTROWARDOJO, M . 1964. Tjatatan tentang Kalaganti penghubung terutama dalam Bahasa Djawo Kuno. M I S I 2.187-94. TUTONKABOY. 1965. Punan vocabularies. SarawakMJ 1 2 . 1 8 8 - 2 0 0 . UDINSAH, R.D. 1958. Bahasa Indonesia daerah Batjan. MeBa 8 / 1 . 2 9 - 3 1 . UHLENBECK, E. M. 1949. De structuur van het Javaanse morpheem. VBG 78.240 pp. . 1949-50. The structure of the Javanese morpheme. Lingua 2.239-70. . 1950. De Tegenstelling Krama: Ngoko. Haar positie in het Javaanse taalsysteem. 28 pp. Groningen, Djakarta.

THAIB GL. ST. PAMOENTIAK,

108

E. M. UHLENBECK

. 1953a. The study of wordclasses in Javanese. Lingua 3.322-54. . 1953b. Woordverdubbeling in het Javaans. BijdrTLV 109.52-61. . 1953c. De systematiek der Javaanse telwoorden. BijdrTLV 109.342-75. . 1954. Verdubbelingsprocédé's bij het Javaanse werkwoord. BijdrTLV 110. 369-87. . 1955. Over woordvorming in het Javaans. BijdrTLV 111.286-307. . 1956. Verb structure in Javanese. For Roman Jakobson, 567-73. The Hague. . 1959. Die mit javanisch rasa morphologisch zusammenhängenden Wörter. Ein Beitrag zur javanischen Lexikologie. OE 6.104-15. . 1960. De systematiek der Javaanse pronomina. VKI 30.vm+63 pp. Review: T.W. Kamil, BijdrTLV 118.277-81 (1962). . 1962. Limitations of morphological processes, some preliminary remarks. Lingua 11.426-32. . 1964. A critical survey of studies on the languages of Java and Madura. BS 7. vili+ 207 pp. Review: F.S. Eringa, Lingua 19.99-104 (1967). . 1965. Some preliminary remarks on Javanese syntax. Lingua 15.53-70. . 1968. Personal pronouns and pronominal suffixes in Old Javanese. Lingua 21.466-82. . 1970. Position and syntactic function of the particle iß in Old Javanese. Studies in general and Oriental linguistics presented to Shirö Hattori, ed. by R. Jakobson and Shigo Kawamoto, 648-58. Tokyo. . 1971. Systematic features of Javanese personal names. Word, Festschrift André Martinet (in press). U K U N SURJAMAN. 1956. Tempat pemakaian istilah klasifikasi kekerabatan pada orang Djawa dan Sunda dalam susunan masjarakat. BaBu 5 / 2 . 7 - 3 2 . UMARJUNUS. 1956. Sistem pronomina Bahasa Minangkabau. BaBu 4 / 4 . 3 3 - 5 . . 1959. BeberapatjatatantentangtranskripsibahasaMinangkabau. BaBu8.3-19. . 1962. Persoalan disekitar posisi kalimat pasif dalam Bahasa Indonesia. MellPe 3.610-42. . 1964. Pleonasme. MellPe 2.195-200. . 1965. Sedjarah dan Perkembangan kearah Bahasa Indonesia dan Bahasa Indonesia. 96 pp. (stencilled). Malang. . 1966. Struktur Bahasa Indonesia. 73 pp. (stencilled). Malang. . n.d. Dasar-dasar Interpretasi Sandjak (Pendekatan Linguistik). 68 pp. (stencilled). Malang. URQUHART, I . A . N . 1959. Some Baram kinship terms. SarawakMJ 9 . 3 3 ^ 6 . . 1961. Vocabulary comparisons, 1911-61. SarawakMJ 10.120-4. VAN DEN BERGH, J. D. 1953. Spraakkunst van het Banggais. 170 pp. The Hague. VAN DER TOORN, J. L. 1891. Minangkabausch-Maleisch—Nederlandsch woordenboek. x i + 392 pp. The Hague. . 1899. Minangkabausche Spraakkunst. xxiv + 227pp. The Hague. V A N D E R T U U K , H. NEUBRONNER. 1860-62. Bataksch leesboek, bevattende stukken

INDONESIA AND MALAYSIA

109

in het Tobasch, Mandailingsch en Dairisch. 4 vols. Vol. 1, 326 pp.; vol. 2, 295 pp. ; vol. 3, 287 pp. ; vol. 4, 222 pp. Amsterdam. . 1861. Bataksch-Nederduitsch woordenboek. vm+551 pp.+xxx pl.+5 pp. text. Amsterdam. . 1864, 1867. Tobasche spraakkunst. 1st pt. viii+84 pp. (p. 1-83); 2nd pt. v + 347 pp. (p. 85—431). Amsterdam. VAN DER VEEN, H. 1915. De Noord-Halmahera'se taalgroep tegenover de Austronesiese talen, vni + 226 pp. Leiden. . 1940. Tae' (Zuid-Toradjasch)-Nederlandsch woordenboek met register Nederlandsch-Tae'. xxm+930 pp. The Hague. . 1950. De samenspraak der beide priesters, de woordvoerders van bruid en bruidegom bij de huwelijksplechtigheid der Sa'dan Toradja's. Bingkisan Budi ( = Festschrift Van Ronkel), 291-306. Leiden. . 1951. Papasan Ba'ru, het Nieuwe Testament in het Tae' (Zuid-Toradja). 996 pp. Amsterdam. . 1960. Sura' Madatu. 984+ 346 pp. Djakarta. . 1965. The Merok-feast of the Sa'dan Toradja. VKI 45.196 pp. . 1966. The Sa'dan Toradja chant for the deceased. VKI 49.vm + 91 pp. VAN OPHUIJSEN, CH. A. 1910. Maleische spraakkunst. 260 pp., 2nd ed. 407 pp. Leiden. (2nd. ed. 1915). VAN PERNIS, H.D. 1950. Woordenboek Bahasa Indonesia-Nederlands. xii + 317 pp. Groningen., Djakarta. VAN STAPPERSHOEF, E . H . W . 1955. Koñeq ghunon. BijdrTLV 111.71-91. VAN SYOC, W.B. 1959. The phonology of the Sundanese language. Diss. Michigan. 181 pp. Ann Arbor. VAN WOUDEN, F . A . E . 1941. Mythen en maatschappij in Boeol. TBG 8 1 . 3 3 3 - 4 1 0 . VERGUIN, J . 1955. L'accentuation en malgache-merina et en malais. Orbis 4 . 5 2 2 - 8 . . 1959. Structure morpho-syntaxique du bare'e. BSL 54/1.162-75. . 1960. Expression du procès en tontemboan. BSL 55.270-6. . 1961. Faits de syntaxe et de détermination en tontemboan. Homme 1.92-7. . 1967. Le Malais, Essai d'analyse fonctionnelle et structurale. 146 pp. Paris. Review: A. Teeuw, BijdrTLV 124.552-3. VERHEIJEN, J . 1941. Assonantie in het Manggarais. T B G 81.455-83. . 1942-48. Woordherhaling in het Manggarais. TBG 82.258-301. . 1950. De stem der dieren in de Manggaraise folklore. BijdrTLV 106.55-78. . 1963. Bird names in Manggarai, Flores, Indonesia. Anthropos 58.677-718. . 1967. Kamus Manggarai I: Manggarai-Indonesia. xxvm + 772 pp. + map. 's-Gravenhage. VON DER GABELENTZ, H.C. 1852. Beiträge zur Sprachenkunde. Erstes Heft: Grammatik der Dajaksprache. 48 pp. Leipzig. VON ESSEN, O . 1959. Die sprechmelodische Gestaltung der Aussage und Frage in der Bahasa Indonésia. OE 6.49-56.

110

E. M. UHLENBECK

1942-48. Voorschriften voor de sëmbahjang in het Soembawarees. TBG 82.380-400. . 1955. Critical survey of studies on the languages of Sumatra. BS 1.55 pp. VROKLAGE, B.A.G. f 1952-53. Ethnographie der Belu in Zentral-Timor. 3 vols. Vol. 1, x i + 688 pp.; vol. 2, 228 pp.; vol. 3, 426 pp., 106 tab. Leiden. WANTALANGI, A . 1957, 1960. Seluk beluk bahasa Tondano/Minahasa. BaBu 5/5. 37-42, 9.216-25. WATUSEKE, F. S. 1956. Bahasa Tondano. BaBu 4/5.3-14. . 1957a. Sumbangan landjutan tentang bahasa Tondano. BaBu 5/6.3-42. . 1957b. Suatu tjeritera India-Kuno dan suatu tjeritera Minahasa-Kuno jang bersamaan. BaBu 6/2.21-9. . 1957c. Salangkew, een mythische vogel in een volksverhaal uit Tondano (Minahasa). BijdrTLV 113.341-60. . 1958a. Tjatatan pada 'Kamus Djolong (Sederhana) Bahasa Tondano-Indonesia'. BaBu 6/4.19-24. . 1958b. Arti nama-nama negeri dalam wilajah bahasa-Tondano. BaBu 7.3-8. . 1958c. Sekitar Nama Minahasa. BaBu 7.25-8. . 1959. Teks-teks Bahasa Tondano II. BaBu 7.88-113. . 1960. Sebelas Bidadari turun mandi di Pantilang (Daerah Toradja-Sa'dan). BaBu 8.248-61. WILLIAMS, THOMAS RHYS. 1961a. A Tambunan Dusun origin myth. J A F 74.68-73. . 1961b. A survey of native peoples of North Borneo. Sociologus 10.170-4. . 1962. Tambunan Dusun social structure. Sociologus 12.141-57. WILLMS, A . 1956. Lautliche und Syntaktische Untersuchungen über die MentawaiSprache. AuÜ 4 0 . 1 - 2 4 . 4 9 - 7 2 . WINARNO SURACHMAD, and R . MURRAY THOMAS. 1961. Penggunaan Bahasa oleh murid-murid sekolah di Bandung. MellPe 2.1.20-31. WINSTEDT, R . 1913. An English-Malay dictionary (roman characters). 3rd ed. m + 524 pp. Singapore, London. (2nd ed. 1920; 3rd ed. 1949). . 1920. Dictionary of Colloquial Malay. 175 pp. Singapore. (2nd ed. London, VOORHOEVE, P .

1945).

. [1952]. A practical modern Malay-English dictionary. 204 pp. Singapore. . 1958. An unabridged English-Malay dictionary. 398 pp., 2nd ed. 431 pp. Singapore. (2nd ed. 1960). . 1960. Kamus Bahasa Mëlayu. 338 pp. Singapore and Kuala Lumpur. WIRJO ASMORO. 1951-52. Sepatah dua patah kata tentang bahasa Madura. MeBa 3.60-4 (1951); 4.45-8 (1951); 5.29-30 (1952). . 1952. Sedikit tentang Bahasa Madura. MeBa 6.33^0. WIRJOSUPARTO, S. 1952. Candakaranîka Âdiparwa. Djakarta. Not available in The Netherlands. WISMAR SARAGIH, J. 1953. Padan na baru. Na mangalimkonsi hubagas sahap Simalungun. 584 pp. Djakarta.

INDONESIA AND MALAYSIA WOLHOFF, G . J . ,

111

and ABD. RAHIM, eds. n.d. Sedjarah Goa. 88pp. Makasar. 1955. Zur Sprache und Kultur der Belu (Timor). Anthropos

WORTELBOER, W .

50.155-200. P. J. 1950. De taal van het Adiparwa, een grammaticale Studie van het Oud-Javaans. VBG 79.vi + 261pp. Review: E.M. Uhlenbeck, BijdrTLV 108. 305-11 (1952). . 1958, 1963. Sekar Sumawur. 2vols. Vol. 1,117 pp.; vol. 2,182pp. Djakarta. , and I. R. POEDJAWIJATNA. 1954,1961. Bahasa Parwa. Tatabahasa Djawa Kuno. Vol. 1, 140 pp. (2nd ed. 138 pp.); vol. 2, 171 pp. Djakarta.

ZOETMULDER,

TAGALOG A N D OTHER MAJOR LANGUAGES OF THE PHILIPPINES* ERNESTO

1.

CONSTANTINO

INTRODUCTORY

This paper 1 has two main parts. The first part gives a general introduction to the major languages of the Philippines. The second part is a review of the developments in the study of the major languages since the beginning of the twentieth century. 2. THE MAJOR LANGUAGES** 2.1

Number

The enumeration of the Philippine languages and dialects is until now incomplete and is largely done by guesswork. The first survey of the Philippine languages and dialects in the twentieth century was made by William E. W. MacKinlay (1902). H e counted 'eight tongues spoken by the civilized races of the country, and about sixty dialects of the savage mountain tribes,' besides 'a dialect of Spanish spoken in and around Zamboanga' and 'two or three dialects spoken by small half-civilized tribes' and the 'Negrito dialects'. In 1917 the anthropologist H. Otley Beyer (1917) listed 43 lan* This work was supported in part by a grant from the Social Science Research Council, University of the Philippines, and in part by a grant from the Institute of Advanced Projects, East-West Center. ** Abbreviations of language names: Akl. = Aklanon; Bag. = Bagobo; Bis. = Bisayan; Bkl. = Bikol; Bon. = Bontoc; Hil. = Hiligaynon; Ibg. = Ibanag; Iby. = Ibaloy; Ilk. = Ilokano; Isn. = Isinai; Ivt. = Ivatan; Kan. = Kankanai; Kap. = Kapampangan; Kin. = Kinaray-a; Kim. = Kalamian; Kuy. = Kuyonon; Mag. = Magindanao; Mar. = Maranao; Orln = Original Indonesian; PA = Proto-Austronesian; PMP = Proto-Malayo-Polynesian; Png. = Pangasinan; Rom. = Romblomanon; Seb. = Sebuano; Smb. = Sambal; Sml. = Samal; Tag. = Tagalog; Tau. = Tausug; Tir. = Tirurai; War. = Waray. 1 It is my pleasant duty to express my thanks to Dr. Howard P. McKaughan for generously allowing me to use his private library ; to Dr. J. Donald Bowen for suggestions on the contents of this work ; to Dr. Cecilio Lopez for information about his researches and certain linguists ; to Mrs. Consuelo J. Paz, Misses Rosario Perez, Emilita Cruz, Erlinda Petilla, Erlinda Abad, and Mr. Dan Solis for sending me books and articles as well as language data from thè Philippines ; to Mr. Ernesto H. Cubar for xeroxing for me some articles at the UCLA library; to Mrs. Lan Hiang Char, Mr. Shoso C. Sueda, and Mrs. Hannah S. Timbol for assistance in getting some books, articles, and microfilms; to Fr. Teodoro A. Llamzon for giving me a list of his works and a microfilm of one of his latest ones; to Drs. John U. Wolff and Alan M. Stevens for information about their research works; and to Mrs. Hazel Tatsuno and Mrs. Arline Uyenten for getting the body of the manuscript typed.

THE

PHILIPPINE

ISLANDS

114

ERNESTO CONSTANTINO

guages and 87 dialects. Beyer's colleague, Marcelo Tangco (1951), revised the list for 1942 and he listed 55 languages and 137 dialects. In 1952 another anthropologist, Harold C. Conklin (1952), made a tentative outline of the native Philippine ethnic and linguistic groups in which he listed 75 'main linguistic groups', 32 of which had 113 'subgroups'. The most recent census of the Philippines taken in 1960 lists 75 Philippine 'mother tongues'.2 A still unfinished project on the Philippine languages and dialects had collected by the end of 1967 data from more than 300 Philippine dialects which were tentatively grouped into more than 70 main linguistic groups.3 2.2

Major Ethnic Groups

Eight of the Philippine languages, chief among which is Tagalog (Tag.), have been called major languages. The other major languages are Sebuano (Seb.), Ilokano (Ilk.), Hiligaynon (Hil.), Bikol (Bkl.), Waray (War.), Kapampangan (Kap.), and Pangasinan (Png.). These languages were designated major languages because they were the native languages of the eight largest ethnic groups in the Philippines. All eight groups were (and still are) Christian groups. They are commonly referred to now as the major cultural-linguistic groups in the Philippines, and they generally have the same names as their native languages. The eight major cultural-linguistic groups are until now the first eight largest ethnic groups in the Philippines. Together they comprise more than 85 percent of the total population of the Philippines.4 However, not one of them constitutes more than a quarter of the population. The largest of these groups, whose members constitute about 25 percent of the population, is Seb., followed by Tag., Ilk., Hil., Bkl., War., Kap., and Png. in that order. Five of these groups, namely Tag., Ilk., Bkl., Kap., and Png., are found principally on the island of Luzon; the three others, the Bisayan (Bis.) groups, are the principal inhabitants of the Bisayan islands and the northern part of the island of Mindanao. 2.3

Dialects

Not one of the eight major Philippine languages is homogeneous. As a matter of fact, every one of them is broken up into several distinct dialects whose differences from each other are mainly phonological and lexical. Bkl. and Seb. seem to have the greatest number of known dialects; Bkl. has at least ten and Seb. at least seven. Several of the dialects commonly assigned to the Bkl. or Seb. language do not seem to be mutually intelligible with the other dialects of the language.5 2

See the Census of the Philippines: I960, vol. 2, pt. 1, table 15, p. 14. This research project, entitled 'Structural and comparative studies of Philippine languages and dialects', has been going on since 1963 supported by grants from the Social Science Research Council of the University of the Philippines. E. Constantino is the principal investigator. 4 Census of the Philippines: 1960, vol. 2, pt. 1, pp. xxi, xxiii. 5 This observation is based primarily on the field survey made in connection with the project described in fn. 3, above. 3

TAGALOG

115

The Bis. languages, i.e. Seb., Hil., War., and also Kinaray-a (Kin.) and Romblomanon (Rom.), are regarded by some linguists and anthropologists as dialects of one language, called the Bis. language or simply Bisayà or Binisayà. However, no one, to the knowledge of this writer, has clearly shown this as being actually the case. 2.4

Names

Each of the major Philippine languages has several names other than the one used here. 6 Some of the names are, however, mere spelling variants. Tag. has the other names 'Tagal', 'Tagalo', and 'Tagala', which are not often used anymore. Seb. has the following other names : 'Cebuano', 'Sugbuhanon', 'Sugbuanon', 'Cebu', and 'Cebuan'. Ilk. is also known as 'Iloko', 'Iloco', 'Ilocano', 'Ilukano', or 'Ilokan'; in formal literary writings it is sometimes referred to as 'Samtoy'. The other names of Hil. are 'Ilongo', 'Hiligayna' or 'Hiligaina', and 'Panayan' ; the last term usually includes Kin. and Aklanon (Akl.). War. is also known as 'Waray-Waray', 'Waraywaray', 'SamarLeyte', 'Leytean', and 'Samaron'. Bkl. is also spelled 'Bicol' or 'Vicol'. Kap. is also known as 'Pampango', 'Pampanga', and 'Pampangan'. Png. is still referred to by a few people (somewhat derisively) as 'Pangalato'. 2.5

Writing System

Each of the major Philippine languages has extensive written literature, some dating from the sixteenth century, with that of Tag. being the oldest and the most extensive. 7 U p to the sixteenth century, i.e. before the colonization of the Philippines by the Spaniards, these languages, as well as other Philippine languages, had a writing system which was based on a syllabary. This syllabary, which was probably of Indian origin, consisted of seventeen symbols, of which three were for vowels and the rest for consonants. 8 The Spaniards replaced this syllabary with the Roman alphabet which has continued to be used until the present time with only a few modifications. 2.6

Classification

The eight major Philippine languages are all related. They, as well as the other Philip* Conant (1909:1069) says that this 'existing variety of usage arose from a haphazard adaptation in the different [Indo-European] languages of the Spanish forms' for the names of the Philippine languages and dialects. He proposes (p. 1073) that 'all Philippine languages and dialects be designated by their native names without inflectional endings, and that in their spellings all peculiarities of Spanish orthography be eliminated.' ' See Beyer (1917:27); Blake (1911c). 8 See Marcilla (1895). Conklin (1949:272-3) reports that 'in the Philippines today three linguistically distinct tribal groups numbering roughly between 15,000 and 16,000 individuals (Buid, 3,500 and Hanunôo, 6,000 on Mindoro; Tagbânuwâ proper, 6,000 on Palawan) are still acquainted with two forms of pre-Spanish writing of Indie origin.'

116

ERNESTO CONSTANTINO

pine languages, except perhaps the Spanish-based Creoles, belong to the same language family, the Austronesian or Malayopolynesian language family. They show many lexical and grammatical similarities, but they also show significant differences. Tag., Bkl., and the three Bis. languages are usually regarded as belonging to the same subgroup. Ilk. and Png. seem to belong to another subgroup, while Kap. seems to constitute a subgroup by itself. The relationships of the eight languages to each other and to the other Philippine languages will be discussed further later in this study.

2.7

Number of Speakers

As previously mentioned, the native speakers of Seb. outnumber those of any one of the other linguistic-cultural groups in the Philippines. However, the speakers of Tag., both native and non-native, are considerably greater than those of Seb. or of any one of the other Philippine languages. Seb., of course, ranks next to Tag. in number of speakers (both native and non-native), followed by Ilk., Hil., Bkl., War., Kap., and Png. in that order. 2.8

Distribution

The eight major Philippine languages have been the dominant languages in almost all parts of the Philippines. Their native speakers constitute the majority of the inhabitants of all the provinces, except six, and all the cities, except three.9 Ilk.10 is the dominant language in all the provinces of Northern Luzon (except of course in the Batanes province where Ivatan (Ivt.) is the dominant language), in Tarlac, in Pangasinan (except in the central area), in Zambales, and in some towns of Occidental Mindoro, Oriental Mindoro, and Cotabato. Png. is the dominant language only in the central area of Pangasinan. Kap. is the dominant language in Pampanga, in four towns of Tarlac (Bamban, Concepcion, Tarlac, and Capas), and in two towns of Bataan (Dinalupihan and Hermosa). Tag. is the dominant language in Greater Manila, and in the following provinces: Bataan, Batangas, Bulacan, Cavite, Laguna, Marinduque, Nueva Ecija, Occidental Mindoro, Oriental Mindoro, Quezon, and Rizal. It is also the dominant language in the northern half of Camarines Norte, in some towns of Zambales, and now even in the cities of Davao, Cotabato, and General Santos. Bkl. is the dominant language in Albay, Camarines Sur, Catanduanes, Sorsogon, the southern half of Camarines Norte, and in some towns of Masbate. Seb. is the dominant language in Augsan, Bohol, Bukidnon, Cebu, Davao, Lanao 9

The six provinces are Aklan, Antique, Batanes, Lanao del Sur, Palawan, and Sulu; the three cities are Basilan, Zamboanga, and Marawi. 10 This determination has been based on data taken from the Census of the Philippines: 1960, vol. 1, Report by province: Abra to Zamboanga del Sur, table 15.

TAGALOG

117

del Norte, the western half of Leyte, Misamis Occidental, Misamis Oriental, Negros Oriental, Surigao, Zamboanga del Norte, Zamboanga del Sur, and in some towns of Cotabato. Hil. is the dominant language in Capiz, Iloilo, Negros Occidental, Romblon, and in some towns of Cotabato, Occidental Mindoro, and Oriental Mindoro. War. is the dominant language only in Samar and the eastern half of Leyte.

2.9

The Spread of Tag.

The most remarkable linguistic development in the Philippines in this century has been the rapid spread of the use of Tag. throughout the country. Before the Second World War, in 1939, only 25.4 percent of the population of the Philippines spoke Tag. in contrast with the 26.6 percent who spoke English,11 the other language which was (and still is) in widespread use in the Philippines as the medium of instruction in the schools. A few years after the last war, in 1948, the speakers of Tag. increased to 37.1 percent of the population while the speakers of English increased to 37.2 percent or only 0.1 percent more than those of Tag.12 But in the 1960 Census of the Philippines, the speakers of Tag. not only exceeded those of English but they also increased to almost 50 percent of the population. The 1960 Census of the Philippines shows that 44.4 percent of the Philippine population spoke Tag. while only 39.5 percent spoke English, and 2.1 percent spoke Spanish.13 By the end of 1967 it is safe to estimate that at least 60 percent of the population of the Philippines could speak Tag. Tag. was chosen as the basis of the Philippine National Language, now called Pilipino, in 1937 when it was spoken by only 25 percent of the population. Since then the percentage of its speakers in relation to the total population of the Philippines has more than doubled. In most provinces and cities where its native speakers are a small minority, more than half of the people can speak Tag.14 Tag. serves as the lingua franca in places where several languages are used even when its native speakers are very few or are lacking.15 The phenomenal growth of Tag. can be explained partly by its favored position among all the languages of the Philippines. Being the dominant language in Manila, Tag. enjoys the highest prestige among the languages. Tag. is also the third official language of the Philippines, Spanish and English being the first two. It is taught as a subject in all schools throughout the country, and in places where it is the local vernacular, it is used as the medium of instruction in the first two grades.16 11

See the Census of the Philippines: 1939, vol. 2, chap. 7, pp. 321, 343. See the Census of the Philippines: 1948, vol. 3, pt. 1, chap. 7, pp. 302-3. See the Census of the Philippines: 1960, vol. 2, pt. 1, pp. 15-16. 14 This determination has been based on data taken from the Census of the Philippines: 1960, vol. 1, Report by province: Abra to Zamboanga del Sur, table 14. 16 This is based on actual field survey conducted in the project described in fn. 3, above. 18 Aguilar (1968:14-A) reports that a 'sub-committee of the Board of National Education [of the Philippines] recently conducted a series of forums on the plan to make Filipino (sic!) the chief medium 12

13

118

ERNESTO CONSTANTINO

The spread of Tag. has also been aided by the mass communications media, such as newspapers, weeklies, comics, movies, radio, and now even television. Also, the improved means of transportation, especially to and from Manila, has brought more people in contact with Tag. 2.10 Philippine Linguistics Tag. and the other major Philippine languages are not only the most widely spoken languages in the Philippines, but they are also the most frequently studied. In fact, from the sixteenth century until very recently, they had been about the only Philippine languages studied. Thus, most of the important works in Philippine linguistics are on these languages. Indeed, one can say without exaggerating that the history of Philippine linguistics is largely the history of the study of the major Philippine languages, especially Tag. 3.

REVIEW OF LINGUISTIC WORKS

3.1

The Three Periods

The study of the Philippine languages may be divided into three sharply defined and contrasting periods : the Spanish Period, the American Period, and the Independence Period. The Spanish Period extends from the sixteenth century when the Spaniards colonized the Philippines to the end of the nineteenth century when the United States, after defeating the Spanish Armada in the Battle of Manila Bay, took over the Philippines as her colony. The American Period started in 1900 and ended after the Second World War. The Independence Period may be said to have started after the Independence of the Philippines in 1946. This review will cover only the last two periods. However, a brief survey of the nature and the significant results of the linguistic activities during the Spanish Period seems to be in order. 3.2 Spanish Period The study of the Philippine languages during the Spanish Period was done exclusively by Spanish missionaries who were usually stationed in the Philippines for some period of time. The linguistic work of these missionary-linguists was ancillary to their mission of Christianizing the natives. Since the natives were not taught the Spanish tongue, the missionaries had to learn the native languages so as to be able to preach of teaching Grades 1 to 6 and eventually through high school in all public high schools' in the Philippines.

TAGALOG

119

to the natives and prepare religious instructional materials for them. As Phelan (1942:194) says, 'The friars did enough linguistic research in order to enable them to discharge their sacerdotal obligations.' It is not surprising then that the bulk of Philippine imprints during the Spanish Period are linguistic studies: grammars, dictionaries, catechisms, confessionals, and Doctrinas Christianas (Phelan 1942:158). The Spanish missionaries emphasized the study of the major languages, most especially Tag. Much of their work, like that of the European missionaries who went to Oceania, 'rests upon the normative preconceptions and traditional pigeon-holes of conventional European grammar' (Milner 1963:64).

4.

THE AMERICAN PERIOD

4.1

Army

Beginnings

The American occupation of the Philippines by the end of the nineteenth century started a new era in Philippine linguistics. The Spanish missionaries almost immediately abandoned their study of the Philippine languages and this task fell into the hands of the Americans who apparently had only a vague picture of the enormous number and discreteness of the languages in the Philippines. Except for a few, the linguistic studies produced during this period of more than forty years were in English. And most of these studies, like those of the Spanish Period, were on the major languages, especially Tag. 4.1.1

The Army

linguists

The United States Army, being the first to arrive in and govern the Philippines, inaugurated the study of the Philippine languages and dialects by the Americans. The Army officers, like MacKinlay, immediately saw the need of works on the Philippine languages, especially Tag. and the other major ones, written in English. The task of running a government in the newly acquired colony made them feel that 'he is best equipped and prepared to do his work, especially a government official, who can meet them [the Filipinos] on their own ground' (Swift 1909:13), that is to say, speak their own languages. This need was expressed cogently by an Army officer, Henry Swift (1909:13), who said: 'The English and Dutch, as masters in the art of colonizing, make the knowledge of the languages of the natives a necessary part of the curriculum for students in the civil service.' It was for this reason that the United States Army offered incentives to its officers to 'spur them on to the mastery of the various languages of the Archipelago' (Swift 1909:13). The linguistic works produced by the Army officers are few. Among these are two survey articles on the languages and dialects of the Philippines by MacKinlay (1902, 1904) and three books: a handbook and grammar of Tag. by MacKinlay (1905), a grammar of Ilk. by Swift (1909), and a primer and vocabulary of Magindanao (Mag.)

120

ERNESTO CONSTANTINO

by R.S. Porter (1903). We must add to this list C.C. Smith's translation (1906) into English of the Mag. grammar written in Spanish by J. Juanmarti and O. T. Mason's translation from the original German into English of F. Blumentritt's survey (1899) of the native tribes of the Philippines and of the languages spoken by them. All the books enumerated above were published by the United States Government. Two other grammars of Tag. belong to this period. One is the grammar written in French by A. Marre (1901), and the other one written in English by C. Lendoyro (1902), 'a Spanish gentleman of more than twenty years' residence in the Philippines' (MacKinlay 1905:12). The interest of the United States Army in the Philippine languages seems to have lasted only a few years. It faded away soon after 1901 when the military government was replaced by the civil government. The publications of the Army linguists were limited to the first few years of the occupation; cf. Blake (1922). 4.1.2 The chief linguists The study of the Philippine languages and dialects reached firmer grounds under the civil government. Almost all the linguists at that time studied the Philippine languages and dialects as a purely scientific or academic concern. Except for one or two, they were employees of the government in the Philippines or were professors at universities in the United States or at the University of the Philippines (which was established in 1908). Most of the results of their researches were published in the Ethnological Survey Publications which became the Division of Ethnology Publications and later in the Philippine Journal of Science in the Philippines; in the Journal of the American Oriental Society, American Journal of Philology, and American Anthropologist in the United States; and in Anthropos in Europe. The chief linguists during the American Period were one Filipino, Cecilio Lopez; two Germans, Otto Scheerer and H. Costenoble; one Belgian missionary, Morice Vanoverbergh; and three Americans, Carlos Everett Conant, Frank R. Blake, and Leonard Bloomfield. All of these linguists, except Blake and Bloomfield, did field work in the Philippines. The studies on the Philippine languages and dialects during the American Period covering more than forty years make a fairly long and impressive list. Those dealing with the major languages will be discussed here under two main headings: Historical and Descriptive. The works on loanwords and the language situation in the Philippines, purely pedagogical works, dictionaries, vocabularies, phrasebooks, readers, and the like will not be included in the discussion.

4.2 Historical Works The three historical linguists during the American Period were C. Conant, H. Costenoble, and O. Scheerer. The most outstanding of the three was probably Conant,

TAGALOG

121

though his publications extended only from 1908 to 1916. Scheerer's publications appeared between 1905 and 1929, and Costenoble's between 1937 and 1940. Scheerer studied Philippine and Formosan languages, though it seems that he never studied any of the major Philippine languages, except in one article (Scheerer and Pablo 1925-26) of which he is the principal author. This article investigates the uses of two particles in Ilk., ti and iti, and compares them with their equivalents in Tag. and Png. 4.2.1 Conant Conant is known to have authored twelve articles, of which ten are on the Philippine languages, one on the Palau language, and another one on the Chamorro language. He was in the Philippines as a government translator from at least 1903 to 1906 where he compiled six wordlists; cf. bibliography of Conant (1912). These wordlists contain from 50 to 200 words which he collected from native speakers of some seven minor Philippine languages. He also compiled in collaboration with two native speakers of Seb. a Seb.-English dictionary (1906) containing about 5500 words. The wordlists and the dictionary have remained unpublished until the present time. Of the ten articles of Conant on the Philippine languages, seven (1908, 1910a, 1910b, 1911a, 1911b, 1912, 1916) are historical and comparative studies of certain Philippine sounds. These articles clearly establish Conant as a pioneer not only in the historical reconstruction of the sounds of the Philippine languages but also in the genetic classification or subgrouping of the languages through the discovery of exclusively shared (non-accidental) innovations. Conant (1908), in his first historical paper, investigates the origins of the / and v sounds in the Philippine languages. He observes in this paper that the great majority of the Philippine languages and dialects, including the eight major ones, have neither / nor v sound, which he says is one of the most obvious characteristics of Malayan phonology as contrasted with Malagasy and certain groups of Polynesian languages. His last article (1916) traces the evolution of the Original Indonesian (Orln) / in the Philippine languages. He says that though Orln / has remained unchanged in the majority of the Austronesian languages and also in the majority of the Philippine languages, it has been lost in intervocalic position in the 7-cockney speech of Bisaya of Cebu City and vicinity', with the resulting contraction of like vowels, and in Tag. without any vowel contraction. He also notes that in Tag. the / is often lost after i (p. 195). Conant's best known articles, "The RGH Law in Philippine languages" (1910a) and "The Pepet Law in Philippine languages" (1912), deal with two of the most important sound changes affecting the Philippine languages. In the first article, Conant investigates the r-g-h-y correspondence, otherwise known as the RGH Law or the first van der Tuuk Law, in seventeen Philippine languages, among which are the eight major languages. His investigations show that the majority of the Philippine languages have g, while others have r, /, or y for the RGH consonant ( = PA *R). He then classifies the languages according to the sound into which the RGH consonant has normally

122

ERNESTO CONSTANTINO

developed. Thus, according to this scheme of classification, Tag., Bkl., and the Bis. languages are ^-languages (together with Ibanag (Ibg.), Mag., Tausug (Tau.), and Bagobo (Bag.)); Ilk. is an /--language (together with Tirurai (Tir.)); Png. is an /language (together with Kankanai (Kan.), Ibaloy (Iby.), Bontoc (Bon.), and Kalamian (Kim.)); and Kap. is a ^-language (together with Ivt. and Sambal (Smb.)). In tracing the origins of the Philippine g, Conant investigates briefly the r-l-d-g correspondence in the same languages and groups them again according to the sound into which the RLD consonant ( = PA *j) developed in the languages. War. and Bkl. are r-languages (together with Tir., Ivt., Kuyonon (Kuy.), and Kim.); Tag., Kap., Seb., and Hil. are /-languages (together with Tau. and Mag.); and Ilk. and Png. are ^-languages (together with Ibg.). Conant concludes that the g's of the Philippine languages may be divided into three classes according to their origin, namely original g, the g of the R G H series, and that of the RLD series. Conant's study of the Pepet Law (1912) is lengthy and includes data from some thirty-four Philippine languages and more than ten other Austronesian languages. His presentation of the evolution of the pepet vowel ( = PA *e) is very systematic and thorough. He ingeniously traces the evolution of the vowel in the Philippine languages in five types of environment. Then he classifies the languages according to the principal reflex of the pepet vowel in them. Each of the classes has among its members at least one of the major languages. Tag. is an /-language; Ilk. and Png. are e-languages, Kap. is an a-language, and the Bis. languages and Bkl. are «-languages, except that Bkl. has u or o in the ultima and a in the penult except in the pp-class where it has u. The pp-class stands for the class of words where the penultimate and ultimate vowels originated from the pepet vowel. Conant wrote an article each on Kap. (1911a), Seb. (1910b), and Tir. (1911b). Tir. is outside the scope of this paper. Conant's article (1911a) on the monosyllabic roots in Kap. describes how as a result of the sound changes in the language a considerable number of Indonesian dissyllabic roots have been reduced to monosyllables. He counted thirty-five of these roots, leaving out some twenty others. 4.2.2 Costenoble It seems that Costenoble wrote only five articles dealing with the Philippine languages, of which the last one was written in German and was published in Europe. The other four earlier ones were written in English and published in the Philippines. One of these four articles (1937a) is a clear and simple presentation of the pepet and R G H sound correspondences in the Philippine languages which, as we have just seen, Conant had discussed in two earlier articles. Another of the four articles in English (1937b) discusses the very interesting and controversial question of the existence of monosyllabic roots in the Philippine languages. This article is especially interesting for its reconstruction of the numerals for 'one', 'four', and 'six' in the Philippine languages as monosyllabic forms. He claims that these numerals, as they appear in the Philippine languages today, were 'enlarged' either by reduplication or by the

TAGALOG

123

addition of 'formative elements'. As they appear now in Tag., Seb., Hil., and Ilk., these numerals are enlarged by formative elements. Costenoble explains his reconstruction of the numerals, as follows: Students of Philippine languages who do not recognize the existence of monosyllabic roots in them [the numerals] always find themselves in a quandary to explain these three numerals. The original form of these numerals is supposed by them to be asa, apat, ansm; the vowels with which these numerals begin in the present-day languages are explained to be evolved from the peppet (sic!), but the fact that they seldom conform to the peppet (sic!) rule does not seem to concern these people, or if they notice the discrepancy they explain it lightly as a case of exception; which is no explanation at all (1937b:76). Costenoble's last article (1940) deals with the sound changes in Tag. He compares here the results of his own investigations of the development in Tag. of certain ProtoIndonesian sounds with those of Dempwolff's in the latter's article entitled Die R-, L-, und D-Laute in austronesischen Sprachen (1925-26). Costenoble says that many Tag. words in which the sound changes deviate from the regular ones are borrowed from, or influenced by, Kap. with which Tag. was (and still is) contiguous. 4.2.3 Other works In this review we should mention Dempwolff's three-volume work (1934-38) on the reconstruction of the phonology of PA in which Tag. was one of the basic languages used. This monumental work was the subject of a monograph by Lopez (1939a).17 We should also mention here that Blake (1906b: 329-37), in his "Notes on the phonology of the Philippine languages", reconstructed the sounds of the primitive Philippine language by comparing representative words from the various Philippine languages. The reconstructed sounds consist of: 3 vowels (a, i, o or u); 13 consonants which are divided into 4 labials (j>, b, m,w); 5 dentals (?, d, n, /, j) ; and 4 palatals and gutturals (k, g, y, y). In addition to these sounds, the primitive Philippine language has 2 diphthongs, ai and aw, and probably the sounds e, r, h, and the glottal catch. All in all, the primitive Philippine language, as reconstructed by Blake, has 20 simple sounds and 2 diphthongs. 4.3

Descriptive Works

The descriptive study of the Philippine languages and dialects during the American Period was dominated by two American linguists, Blake and Bloomfield, both of whom never visited the Philippines, and one Filipino linguist, Lopez. 4.3.1 Blake Blake was among the first persons during the American occupation of the Philippines to become interested in the study, as well as the teaching, of the Philippine lan17

This work was mimeographed by the S.I.L., Philippine Branch, about 1961 with an English translation of Dempwolff's wordlist appended to it.

124

ERNESTO CONSTANTINO

guages and dialects, especially the major ones. This interest of his started as early as 1900 when the Johns Hopkins University, in the hope that the United States would develop a colonial policy which would make provisions for the study of the various Philippine languages, decided to start a program of giving instruction in the Philippine languages. Blake, who was a graduate of the university and a student of the Semitic and Sanskrit languages, was asked to study the Philippine languages and teach courses in some of them. In the academic year 1901-02, Blake conducted a course in elementary Tag. which was attended by eight students. The following year, he offered a course in Seb., as well as an elementary and a more advanced course in Tag. (Blake 1926a). The United States Government, however, refused to be interested in the work on the Philippine languages at the Johns Hopkins University. In spite of this discouraging outlook, Blake continued his study of the Philippine languages, especially of Tag., which extended to more than fifty years. As late as 1943, Sebeok (1943:154) reported that Blake would be willing to offer a course in Tag. at the Johns Hopkins University if requested. Between 1902 and 1950, Blake wrote a total of twenty-seven articles dealing directly with the languages and dialects of the Philippines.18 He also wrote a book on Tag. grammar and a monograph on Tag. accent. The latter work has until now remained unpublished.19 Most of Blake's articles deal with the general features of the Philippine languages, especially those features which unite them or define them into groups. Most of these features were extracted from Tag. and the other major languages, especially Seb., since all his data came from printed sources which were mostly in those languages.20 On the basis of these features, Blake was able to classify the Philippine languages into two main groups: the Northern Group, to which Ilk. and Png. belong; and the Central Group, to which Tag., Bkl., and the Bis. languages belong. Kap. lies between these two main groups. In the south, Mag. and Mar. belong together, while Bag. and Tau. constitute two independent groups. Blake's most important work in Philippine linguistics is undoubtedly his grammar of Tag. (1925). This book, which was the result of his extended study of the language, has a very interesting publication history. Completed in 1902, it was not published until 1925. In the years between, Blake published several articles, chief among which 18

The topics which Blake wrote on are (1) the similarities and differences of the Bis. languages, and those of Bis. and Tag.; (2) the Sanskrit loanwords in Tag.; (3) Philippine literature; (4) the study of the Philippine languages at the Johns Hopkins University; and (5) the role of the United States Government in the development of Philippine linguistics. He also reviewed four works, including Bloomfield's grammatical analysis of Tag. Lastly, he compiled two bibliographies of works on the Philippine languages and dialects which until now have not been updated; but see Asuncion-Lande (1965) and Lopez (1965b). 19 Lopez in a personal communication says that Blake 'gave me his ms. on Tag. accent to do whatever I want with it.' ao Blake (1906b: 223) says that his 'materials for the study of the Philippine languages consist of texts, collections of conversational phrases, grammars, dictionaries, and vocabularies.'

TAGALOG

125

were his two long articles on comparative Philippine grammar (1906b, 1907). He also revised his Tag. grammar bringing it to its present form in 1910. Like most of the previous grammarians, Blake did not gather the data on which he based his Tag. grammar from native speakers of the language. Instead, he based it 'on the works of the best Spanish grammarians, checked and verified by the reading of numerous Tagalog texts' (Blake 1925: vii). Because of this, his grammar, like most of his other works, contains Tag. words and expressions that are either archaic or not used in present-day Tag. 21 Blake's purpose in writing his Tag. grammar is firstly scientific and secondly practical. As he states in the preface, he wants to give the student of Tag. 'a clear and concise presentation of all the facts of the language in an arrangement in which scientific principles are observed, but which offers at the same time some of the advantages of a practical treatment' (p. ix). Moreover, he states that his grammar 'will serve as an introduction to the study of the Philippine languages in general and as a model for other Philippine grammars' (p. x).22 Blake's presentation of Tag. grammar shows the unmistakable influence of the 'best' Spanish grammarians, particularly Totanes (1865). His grammatical categories and terms are those used by the Spanish grammarians who, as early as the sixteenth century, adopted them from European traditional grammars. Blake's grammar is divided into four parts: (1) Orthography and phonology, (2) Morphology, (3) Syntax of combinations, and (4) Special syntax. In the syntax of combinations, 'the material of the language is treated mainly from the standpoint of that word of the combination which is modified by the other', while in the special syntax, 'all the uses of the various elements and principles of the language are discussed from the standpoint of the individual element or principle' (1925:201). In his study of Tag. and other Philippine languages, Blake showed a great interest in three interrelated grammatical units: verb, voice, and case. He states that as a general rule practically any word in Tag. belonging to any part of speech may be verbalized. He concludes that the 'verbalizing power' of Tag. and 'generally speaking of other Philippine languages, and indeed of Malayo-Polynesian languages in general ... has so to speak run wild' (191 la:439-40). This conclusion seems exaggerated especially since it is based on incomplete studies. The construction of verbs in Tag. and in the other Philippine languages has not 'run wild' and can be formulated in an orderly manner. 23 Also, it is not the case, as it seems implied in Blake's concluding 21 A noted Filipino historian made the following remarks on how the Spanish friars studied the Philippine languages: 'The friars, in studying the Filipino languages, continually compared them with the Latin and the Castilian, to the grammar and genius of which they molded, whenever they could, those of the new language which they were learning. As a result, the grammars of the Filipino languages which they soon made created an artificial language very different from that actually spoken by the islanders' (Blair and Robertson 1903-09:147). 22 Blake also intended this work to be an exposition of a new method of syntactical treatment which he described earlier; cf. Blake (1921). ,s See, for example, Gonzales (1962) and Silverio (1962).

126

ERNESTO CONSTANTINO

statement, that ANY word in Tag. or any of the other Philippine languages can be verbalized by ANY verbal affix. Nevertheless, the 'power of verbalization' of Tag. and the other Philippine languages, as earlier described by Wilhelm von Humboldt (1838: 347), is clearly the most outstanding characteristic of the Philippine languages, one that distinguishes them from the other Austronesian languages. A verb in Tag., according to Blake, may either be in the active or passive voice. The passive verbs have three forms according to the affix with which they are formed; these are the in-, /-, and aw-passives. As to the uses of the voice forms, Blake (1925:140) says that 'in any given sentence the voice of the verb depends upon the relative importance of the various elements, the most important or most emphatic idea being made the subject of the sentence. If this is the agent of the action expressed by the verb, the active voice is used; if it is any other element of the sentence, then one of the three passives is employed.' Blake (1925:140-1) summarizes the uses of the three passives as follows: 'In general the in passive is used when the object of an action towards the agent (e.g. to take) is made the subject; the i passive when the subject is the object of an action away from the agent (e.g. to give), or the instrument or cause of the action; the an passive, when a place or anything regarded as place stands as subject.' Blake (1925:141) adds that the passive constructions occur more frequently than the active; he says that the passive constructions may in fact be the rule, and the active constructions the exceptions. 24 The Tag. verbs, according to Blake, express not only voice but also case relations. He says that the four voice forms of the verb express four different cases. He (1906a:188), however, adds that these cases 'do not conform exactly in their scope to any of the cases commonly recognized in Indo-European grammar; sometimes two forms are used to express what is ordinarily considered one case, sometimes one form expresses two or more cases.' He (1906a: 189) summarizes the cases expressed by the four verb forms as follows: 'The active expresses the case of the agent; the in passive, the case of the non-alienated object; the an passive, all local relations of whatever character; while the i passive bears the burden of the remaining case relations.' The expression of case relations in Tag. is not confined to the verbal forms alone. To Blake, the forms of the article or demonstratives are case-marking forms too. There are three of these case forms to which Blake gives the traditional terms NOMINATIVE, GENITIVE, a n d OBLIQUE.

4.3.2 Blake's students We should mention here the works of two students of Blake, W.G. Seiple and L.B. Wolfenson, whom Blake encouraged to study Tag. Seiple wrote an article (1903) on the Tag. numerals, and another one (1904) on the Tag. polysyllabic roots with initial p. Wolfenson, on the other hand, wrote one article (1906) on the infixes la, li, lo in Tag. ,4 This writer made a count of the occurrences of the different types of Tag. simple sentences in two Tag. short stories and in two dialogs in Bowen (1965). Of the 500 sentences counted 41 percent were active sentences, 30.4 percent passive sentences, 22.6 percent nonverbal sentences, and .06 percent nonpredicative sentences.

TAGALOG

127

4.3.3 Bloomfield By far the most well-known and most influential work on any Philippine language is Bloomfield's grammatical analysis of Tag. (1917) with the accompanying texts from the language. This grammar is until now considered by many to be the best grammar of Tag. 25 The publication of Bloomfield's book Language in which some of the important grammatical features of Tag are summarized (1933:173, 200-1, 221) and the subsequent rise of the Bloomfieldian type of linguistics especially after the Second World War have made Bloomfield's Tag. grammar the model or the starting point of almost every modern descriptive study of any of the Philippine languages and dialects. Bloomfield's Tag. grammar created a revolution in the study of the Philippine languages and dialects in two important ways: (1) It made the use of informants (minus their intuitions) a must in the gathering of language data; and (2) it started the now established practice in Philippine linguistics of using new grammatical terms in place of the traditional or more familiar ones to describe the same grammatical categories and relations in order to emphasize the difference (not the similarities) of the language with respect to other languages, especially the Indo-European languages. 26 As a result of Bloomfield's influence, Philippine linguistics has become thoroughly descriptive and classificatory or taxonomic, and to a certain extent non-intuitive. Leaving the examples out, Bloomfield's Tag. grammar is not as long as Blake's, nor is it as elaborate. Bloomfield's grammar differs from Blake's not in scope or coverage, but in the way the grammatical units and relations are arranged and labeled. Bloomfield's grammatical analysis of Tag., unlike Blake's, is a purely linguistic piece of work. The author rightly claims that it is 'the first scientific analysis of the structure of Tagalog' (1917:10). The analysis is based on the speech of an educated speaker of Tag., though it is obvious that the author must have consulted the works of earlier Tag. grammarians, especially the Spanish ones. Bloomfield's Tag. grammar is divided into three parts. Part I contains the Tag. texts in broad phonetic transcription with English translation. Part II contains the grammatical analysis of the language; it is subdivided into Phonetics, Syntax, and Morphology. Part III is the list of formations and glossary. Only Part II, the most important part of the work, will be discussed here. Under the section of Phonetics (pp. 134-45), Bloomfield describes the distinctive sounds of Tag., Tag. syllabification, and accentuation. He finds the variation between / and e and u and o never distinctive, except perhaps in borrowed words. He fails, " Lopez (1965a:3) says that Bloomfield's grammatical analysis of Tag., 'as a treatise on any single Philippine language, remains unmatched.' 26 Blake (1919:89), in his review of this work, gives as one of its chief defects 'the great number of peculiar and unusual grammatical terms used without good reason for perfectly familiar and suitable designations.' He adds (p. 93): 'Any grammarian is, of course, thoroughly in sympathy with the invention of new terms and the setting up of new categories in the study of a new and peculiar form of speech, but familiar terms and categories should not be thrown overboard, as they are here, without good and sufficient reason, especially when the new terms and categories offer no special advantage over the old, or are in many cases decidedly inferior to them'.

128

ERNESTO CONSTANTINO

however, to note a number of native words where e and o may never vary with i and u, respectively.27 He describes two phonetically distinct word-accents, primary and secondary, which never occur in the same environments except in a few words. He also describes the pitch rise and vowel length that accompany each accent. Bloomfield's treatment of Tag. syntax (pp. 146-209) is short and sketchy. It is divided into the following four headings: (1) Sentence and word, (2) Subject and predicate, (3) Attribute, and (4) Serial relation. Each of these syntactic relations may be expressed by particles. Bloomfield classifies the words syntactically into two parts of speech: FULL WORDS and PARTICLES. The particles either express the syntactic relations between full words or act as attributes of full words. The full words, on the other hand, act not only as attributes but also as subject or predicate. The full words are classified into TRANSIENT and STATIC words. Transient words are those which express an element of experience; those words which are not transient are static. The transient words 'fall into four classes according to the four relations which a subject may bear to them when they are used as predicate' (pp. 153-4). Bloomfield gives these classes the following names: 1. active: the subject is viewed as an actor: Sumusulat sya narj liham. He (subject) is writing a letter, letters. Sya y pumutol narj kahoy. He (subject) cut some wood. Umalls sya. He went away. 2. direct passive: the subject is viewed as an object fully affected or produced: Sinulat nya ari liham. Was-written-by-him the letter (subject), i.e. The letter was written by him. He wrote the letter. Pinutol nya aq kahoy. Was-cut by-him the wood (subject), i.e. He cut the wood. 3. instrumental passive', the subject is viewed as a means, an instrument, something given forth or parted from: Insinulat nya ar) kwento. Was-written-down by-him the story (subject), i.e. He wrote the story. Ipinutol nya arj gulok. Was-cut-with by-him the bolo (subject), i.e. He used the bolo for cutting. He cut with the bolo. 4. local passive: the subject is viewed as an object partly or less fully affected, as a place or sphere: Sinulatan nya ako. Was-written-to by-him I (subject), i.e. He wrote me. Pinutulan nya arj kahoy. Was-cut-from by-him the wood (subject), i.e. He cut a piece off the wood (1917:154). Bloomfield's four transient classes are the same four verb classes which the Spanish grammarians and Blake called the active, the /«-passive, the /-passive, and the anpassive, respectively. Bloomfield distinguishes words (transients or statics) with or without their modifiers which are preceded by an article or a demonstrative from those which are not. He calls the former OBJECT CONSTRUCTIONS or EXPRESSIONS and the latter OPEN CON27

It is now generally accepted that Tag. e and i and o and u are separate phonemes. The vowels e and o occur not only in borrowed words but also in native words. For example, e and / contrast in the following native words: he.to 'here it is' and hi:toq 'catfish', e:wan 'I don't know' and t:wan 'leave'. On the other hand, o and u contrast in the following forms: a:patnapuq 'forty' and a:patnapoq 'four already sir', isu:si mo 'lock it' and isd:li mo 'return it'.

TAGALOG

129

or EXPRESSIONS. The pronouns with or without an article are object constructions or expressions. Bloomfield says that most Tag. sentences consist of a subject and a predicate. These are the PREDICATIVE sentences, or predications. The other sentences are NON-PREDICATIVE, i.e. they lack the subject-and-predicate structure. The subject may precede the predicate or it may follow the first accented word of the predicate. When the subject precedes the predicate, the particle ay (y after a vowel) stands between them. This particle expresses the predicative relation. The subject is always an object expression, except in complex sentences where an entire predication or quotation may stand as subject. The predicate may be (1) a transient word with its modifiers, (2) a static word with its modifiers, or (3) an object expression. In his book, Language (1933:173), Bloomfield sets up five types of predicative sentences in Tag., an EQUATIONAL type and four NARRATIVE types. These sentences are identified by their predicates. The predicate of the equational sentences is either a static word with its modifiers or an object expression. The predicate of the narrative sentences is always a transient word. The narrative sentences are further subdivided into four subtypes according to the class of the transient predicate (or the relation of the subject to the transient predicate): STRUCTIONS

( 1 ) ACTOR-ACTION, ( 2 ) GOAL-ACTION, ( 3 ) INSTRUMENT-ACTION, a n d ( 4 ) PLACE-ACTION.

Bloomfield distinguishes four types of attribution, as follows: 1. Conjunctive attribution: the attribute is joined by means of the particle na or r|: isa i) tao a person, one person; 2. Disjunctive attribution: the attribute, which is always an object expression, stands in a special disjunctive form, ai), for instance, being changed to nag; ar) puno nar) ui) go? the tree of the monkey, 3. Local attribution: the attribute, which is always an object expression, stands in a special local form, aq, for instance, being always changed to sa: nallh'go sa flog bathing in the river; 4. Absolute attribution: the attribute merely precedes or follows: hindi tama? not correct (1917:160). Bloomfield (1917:161) recognizes three case-like forms of the object constructions; these forms are indicated by the proclitic particles ay, nay, and sa. He calls these forms SUBJECTIVE, DISJUNCTIVE, a n d LOCAL.

Finally, Bloomfield (1917:205) defines the serial relation as being expressed 'in part by particles and in part by parataxis, parenthesis, and anacolouthic constructions.' Bloomfield's treatment of Tag. morphology (1917:210-316) emphasizes the uses of the active and passive affixes, their relations to each other and to other affixes which produce (1) gerund-like words which he calls 'abstracts of action' or (2) nominals which he calls 'special static words'. Except for the treatment of the role of accent and accent-shift in word-formation (which no grammarian before him had done) Bloomfield's Tag. morphology adds little to what the Spanish grammarians and Blake had already said about Tag. morphology. In 1942, more than twenty-five years after the publication of his grammatical analysis of Tag., Bloomfield (1942) published a much abbreviated syntax of Ilk. The

130

ERNESTO CONSTANTINO

chief features of this syntax are as follows. All syntactic expressions in Ilk. are divided according to their functions into open expressions and object expressions. (Bloomfield does not elaborate on this. Presumably he means here the subject, predicate, and attribute functions.) Words fall into three main classes: (1) pronouns, which are always object expressions; (2) full words, which are open, but may be objectified; and (3) particles, which are always open. The object expressions (i.e. the pronouns and the full words and phrases objectified by certain proclitic particles) are divided into two categories (or cases): STRAIGHT and OBLIQUE. The straight category is subdivided into NOMINATIVE and ATTRIBUTIVE. The enclitic pronouns appear in straight form only, i.e. in both nominative and attributive forms. The nonenclitic pronouns appear in nominative and oblique forms only. Most sentences have a subject and a predicate. The chief types of attributes are object expressions in the three categories, open expressions with the proclitic ya, enclitics, and phrasal attributes introduced by various particles and full words. There are a few significant contrasts between Bloomfield's Tag. and Ilk. syntax. In his Ilk. syntax, he sets up three parts of speech, while in his Tag. syntax he has only two. The third part of speech consists of the pronouns, which are listed in their different forms. There is no mention at all in his Ilk. syntax of transient and static words, nor of nouns, verbs, and adjectives. There is no mention here too of the types of sentences, especially the four narrative types. 4.3.4 Lopez The Philippines produced her first native Filipino linguist, Lopez, before the Second World War. Lopez began as a student of Scheerer at the University of the Philippines. From there he was sent to the University of Hamburg where he finished the doctorate degree in linguistics in 1928. For his dissertation, he wrote a comparison of Tag. and Ilk. (Lopez 1928) under the direction of Dempwolff. Blake (1929) wrote a review of this work and that started a long correspondence between the two. When Blake wanted to revise his Tag. grammar, he asked Lopez, a native speaker of Tag., to help him 'by checking the statements of older Spanish authorities with modern Tagalog usage' (Blake 1929:191). The early works of Lopez were stimulated and influenced by the writings of Blake. Among these are those dealing with the general features of the Philippine languages (1937c), Tag. affixation (1937b), and Spanish loanwords in Tag. (1937a). The most important work of Lopez to date is his Tag. grammar, A manual of the Philippine national language (1940a). When this grammar was written, Tag. had just been proclaimed as the basis of the Philippine national language which the Philippine Government, through the Institute of National Language, 28 wanted to develop. Lopez intended his Manual to be a scientific grammar of Tag. that could be used in the classroom. 28

See Frei (1949-50:182ff.).

TAGALOG

131

Lopez's Manual shows the influence of Bloomfield's grammatical analysis of Tag. and Jespersen's The philosophy of grammar. It is divided into four sections, one on phonetics, two on morphology, and one on syntax. The treatment of Tag. syntax is new and is more extensive than Bloomfield's. The other works of Lopez published before the end of the Second World War include a psycho-morphological analysis of Tag. (1940b), a comparison of Tag. and Malay lexicographies (1939b), an article written in German on the relationships between Tag. and Malay (1931a), and an article on the language situation in the Philippines (1931b). Lopez has continued his researches in Philippine linguistics after the Second World War. We shall take up his post-war works when we come to that period. Lopez was able to encourage a number of Filipinos to study Philippine languages before the last war. Among the studies written under his guidance are one article on the infixes la, li, lo, and al in the Philippine languages (Viray 1939), another article on prenasalization in the Philippine languages (Viray 1941), and a third one (GamboaMendoza 1940) on the phonological peculiarities of Kap. based on the phonology of PA as reconstructed by Dempwolff. Also, Lopez directed the compilation of the preliminary studies on the lexicography of some Philippine languages of the Institute of National Language. 29 4.3.5 Vanoverbergh Another linguist who deserves mention is the Belgian missionary, Vanoverbergh, who has done extensive work on Ilk. and the Negrito and Igorot languages. Vanoverbergh's publications on the Philippine languages and dialects date as far back^as 1917 and as recent as 1956. However, he did most of his linguistic work before the Second World War. His distinct contribution to the analysis of the Philippine languages and dialects seems to be his rejection of the terms active and passive for^the voice-forms of the Ilk. verb; for these terms he substituted ADJECTIVAL and SUBSTANTIVAL, respectively. He (1931:486) claims that 'the change was not only useful, but that it was a practical necessity, and that the new terms are just what we have claimed that they should be.' It seems that he justifies the change on the basis of how the Ilk. verbal sentences are translated (by him) into English. 4.3.6 Other works on Philippine verbs The verbs of the Philippine languages have always attracted great interest among students in Philippine linguistics. The linguists who worked on the Philippine languages during the American Period recognized fully the important role of the verbs in the languages. In addition to the works which we have already mentioned, two articles on the Philippine verbs appeared during this period, one by Scheerer (1924) and the other one by Costenoble (1937c). This emphasis on the verbs, especially in 29 These studies constitute the first volume of the publications of the Institute of National Language which were published when Lopez was the secretary and executive officer of the Institute.

132

ERNESTO CONSTANTINO

connection with their role in the syntax of the Philippine languages, has continued until today. 4.4

Summary

The preceding discussion has shown us that during the American Period, only six or seven linguists made serious studies on the Philippine languages and dialects, especially the eight major ones. These linguists were geographically divided into two groups: those who were in the Philippines and those who were in the United States. The first group consisted of Conant, Scheerer, Costenoble, Vanoverbergh, and Lopez. The second consisted of Blake and Bloomfield. All of these linguists, except Vanoverbergh and Lopez, are no longer living. Vanoverbergh is still in the Philippines but apparently is no longer active. Lopez, on the other hand, is still very active having published only recently an article on the origins of the Philippine languages. As Professor Emeritus at the University of the Philippines where he started his career in linguistics, he is at present finishing his two major research projects on the comparison of the Philippine languages. 30 Though the study of the Philippine languages and dialects started as early as the sixteenth century, Philippine linguistics became scientific only during the American Period. In spite of the lack of interest of the United States Government in the study of the Philippine languages and dialects (cf. Blake 1922), Philippine linguistics made some real progress. The following statements summarize what we consider to be the significant achievements in the study of the eight major Philippine languages (and also for that matter the other Philippine languages) during the American Period: 1. Pioneering studies on the comparative phonology of the Philippine languages were made, primarily by Conant; 2. A preliminary comparison of the chief grammatical features of the Philippine languages, especially the eight major ones, was made by Blake. This comparison, however, needs a thorough re-doing primarily because of the quality of the data used; 3. Three scientific grammars, all of Tag., were written by Bloomfield (1917), Blake (1925), and Lopez (1940a). The first was based on data supplied by a native speaker of the language; the second was based on data taken from printed sources; the third was written by a native speaker of the language. Bloomfield's grammar is considered superior to the other two, but not one of them is now considered adequate. In short, the single most significant achievement in the study of the eight major Philippine languages before the outbreak of the Second World War was the grammatical analysis of Tag. by Bloomfield. This fact became very clear immediately after the Second World War, that is during the Independence Period.

80

These projects are entitled 'Comparative Philippine wordlist' and 'Comparative Philippine syntax'.

TAGALOG 5.

133

THE INDEPENDENCE PERIOD

5.1

Introduction

The Independence Period began after the Second World War, in 1946, when the Philippines became an independent state. This period has been characterized by a great increase in the number of linguists working in the Philippines which has resulted in a great increase in the number of the Philippine languages and dialects which have been studied. The amount of studies on the languages and dialects, especially the minor ones, has likewise increased considerably. This portion of the study will survey the developments in the study of the eight major Philippine languages after the last war. 5.1.1

Significant

developments

The progress of Philippine linguistics after the Second World War has been influenced tremendously by three significant developments. The first of these was the establishment of the Philippine branch of the Summer Institute of Linguistics (S.I.L.) in 1953. From that year on many missionary-linguists belonging to that organization went to the Philippines and worked on the languages and dialects in the field. These linguists put to use the great advances made in descriptive linguistics in the United States before and after the last war, especially in phonetics and phonemics and later in syntax, in the analysis of many of the Philippine languages and dialects, particularly the minor ones. The second development was the application of the linguistic approach to the teaching of English to Filipinos which generated widespread interest in the study of the Philippine languages, especially the major ones. This development culminated in the establishment in 1957 of the Philippine Center for Language Study under the Department of Education of the Philippines and the University of California, Los Angeles (UCLA). The third significant development affecting the progress of Philippine linguistics after the last war has been the gradual increase in the number of Filipino linguists, especially after 1960. Until almost the end of 1950 there was only one Filipino linguist, qua linguist. At present, there are about four or five new ones, all of whom studied in the United States. There are also several Filipinos pursuing graduate studies in linguistics in American and Philippine universities and a few in Canadian universities. The three developments described above have served to make Philippine linguistics an extension of American linguistics, and the Philippines an overseas field station and laboratory for many American linguists. Indeed, a number of the Philippine languages and dialects have been used to test or exemplify the workings of some grammatical models, theories, or methods that have been developed by American linguists in the past decade.31 81

See, for example, Pike (1963), Austin (1966), and Kess (1967),

134

ERNESTO CONSTANTINO

5.1.2 Linguists The study of the Philippine languages and dialects at present is being carried on mostly by several groups of linguists in the Philippines. The largest and most widely distributed of the groups is the Philippine branch of the S.I.L. The first director of this S.I.L. branch was Richard S. Pittman; the second was Howard P. McKaughan. This group has been concentrating on the lesser-known (minor) Philippine languages and dialects. The studies made by this group comprise a long list; see Roe (1967). Another group of linguists in the Philippines today is found in the Department of Oriental Languages and Linguistics of the University of the Philippines. 32 For the past five years, the members of this group have been conducting a linguistic survey of the Philippines, and have been collecting extensive grammatical and lexical data from all the Philippine languages and dialects. The data, which are transcribed and taperecorded, are intended to be used in making structural and comparative studies of the Philippine languages and dialects, and in compiling dictionaries of a selected number of the languages and dialects.33 Other groups of linguists in the Philippines are found at the Language Study Center of the Philippine Normal College, the Ateneo de Manila University, the University of San Carlos, and the Interchurch Language School. The New Tribes Mission in the Philippines has also one or two linguists. Most of the post-war studies on the Philippine languages and dialects have been made by linguists who are in the Philippines. However, a substantial number of studies have also been made by linguists and graduate students at American universities, especially at the University of Hawaii, Yale University, the University of Michigan, and UCLA. In the succeeding sections, the works dealing with the eight major languages of the Philippines which have been written after the Second World War are discussed in three groups: (1) those which deal with the classification of the Philippine languages, (2) the historical works, and (3) the descriptive works. 5.2

Classification

Since the Second World War, there have been some attempts to classify the Philippine languages and dialects. The first of these seems to be that of Conklin (1952) which on the basis of 'systematic sound correspondence and other linguistic features' groups some of the Philippine languages and dialects into the ILOKO-TYPE and the TAGALOGTYPE. Ilk. and Png. belong to the Iloko-Type group, while Tag., Bkl., Hil., and probably Seb. and War. belong to the Tagalog-Type group. Kap. does not seem to • 2 This group is actually the oldest surviving group in the Philippines. The Department was founded in 1923jfor the purpose of making a 'scientific comparative study of the various dialects' of the Philippines; cf. Frei (1949-50:168). 88 More than 2,000 sentences illustrating different morphosyntactic patterns and more than 4,000 stems have been collected from most of the 300 dialects surveyed. The manuscript and tapes are stored in the Department's Archives of Philippine Languages and Dialects.

135

TAGALOG

belong to either group. Dyen's TAGALIC GROUP (Dyen 1953a) has essentially the same composition as Conklin's Tagalog-Type group. Three attempts have been made to classify the Philippine languages and dialects on the basis of lexicostatistical comparison. The first of these was made in 1953 (Fox, Sibley, and Eggan 1965) for the Northern Luzon Type of languages which include Ilk. and Png. and some other minor languages. The comparison yielded four divisions: NORTHERN, CENTRAL, SOUTHERN, a n d SOUTHEASTERN. I l k . b e l o n g s t o t h e N o r t h e r n

Division, and Png. to the Southern Division. Thomas and Healey (1962) made another lexicostatistical comparison of some thirty-seven Philippine languages and dialects, among which are the major languages, except War. and Hil. The results of this comparison with respect to the major languages only show that (1) Tag., Kap., Seb., and Bkl. are more closely related with each other than with Ilk. and Png. which in turn seem to be closely related with each other; and (2) Tag. and Kap. are more closely related with each other than with Seb. or Bkl. Dyen's monumental lexicostatistical classification of the Austronesian languages (1965) includes some 60 Philippine languages; among these are the major languages, except War. and Png. The results of Dyen's study agree generally with those of Thomas and Healey's, except for one difference. Dyen considers the close relationship between Tag. and Kap. as being due to UNDETECTED borrowing. He says that Tag. is more closely related with Seb. and Kuy. than with Kap. (Dyen 1963:63). A classification of twenty-one Philippine languages on the basis of shared vocabulary has been made by Chretien (1961). The results of this study show two wellm a r k e d divisions, t h e LUZON SEQUENCE a n d t h e MACRO-BISAYAN GROUP, with Bkl. as a

transition language between them. All the Luzon languages, except Bkl., fall into the Luzon sequence, and all the Bis. languages into the Macro-Bisayan group. Perhaps we should mention here a recent article of Lopez (1965a) which the author hoped would suggest a way of grouping the Philippine languages on the basis of shared syntactic peculiarities. In this article Lopez compares twelve Philippine languages, among which are the eight major languages, on the basis of the three syntactic relations identified by Bloomfield for Tag. Lopez, however, does not make any grouping of the languages at all. A subgrouping of nine Philippine languages by the comparative method, i.e. by the discovery of exclusively shared innovations, has been made by Llamzon (1965). Of the nine languages, six are major languages, namely Tag., Seb., Hil., Ilk., War., and Bkl. Finally, the same author, Llamzon (1966c), wrote a review of the various attempts to subgroup the Philippine languages by applying the three subgrouping procedures described by Dyen (1953c), namely (1) judgment by inspection; (2) discovery of exclusively shared (non-accidental) innovations; and (3) lexicostatistic dating. 5.3

Historical

Works

One of the characteristics of post-war Philippine linguistics is the neglect of historical

136

ERNESTO CONSTANTINO

studies. However, some comparative studies involving two or three groups of the Philippine languages, especially the major ones, have been made by a few linguists. The foremost of these linguists is Dyen of Yale who has published one monograph and several articles on the reconstruction of some phonemes and words in PA (which he called Proto-Malayo-Polynesian (PMP) before 1965). In one of his articles, Dyen (1947a) discusses his theory that PA *D became Tag. d in initial and postconsonantal positions, and not / as Dempwolff stated. In his monograph and other articles in which he reconstructs the PA laryngeals (1953a) and a number of additional PA phonemes and words (1947b, 1951, 1953b, 1962a), Dyen uses among his witnesses the Tagalic languages, namely Tag., Bkl., Seb., Hil., and War. Another linguist at Yale, Rufus S. Hendon, has written at least two articles on ProtoMalayopolynesian (1964a and b) which include Tag. A few more historical studies have been made by other linguists. E. Verstraelen has written at least three articles (1960, 1961, 1962) in which he investigates the origins and correspondences of some sounds in a number of Philippine languages, chief among which are the eight major ones. Llamzon (1966b) has written an article in which he reinvestigates the Tag. reflexes of PA *e. He comes u p with the conclusion that PA *e regularly became Tag. i and irregularly a or u, the same conclusion Costenoble (1940) arrived at earlier. In a recent paper, Lopez (1967) makes an attempt to trace the origins of the Philippine languages through a comparison of the phonological, morphological, and syntactic features of the languages with those of the other Austronesian languages. He concludes his study by saying that he favors the theory which places the origins of the Philippine languages 'somewhere in Asia'.

5.4

Descriptive

Works

The most outstanding feature of post-war Philippine linguistics has been the great number of descriptive studies. In the early years after the last war, phonemic studies constituted the bulk of the linguistic works. However, after 1960 the number of grammatical studies, especially syntactic studies, began to increase until at present phonemic analyses are made almost only on newly 'discovered' Philippine languages or dialects, or those which present special problems in phonetics and phonemics. 5.4.1 Descriptive models Three descriptive models are discernible in post-war Philippine linguistics. The first is of course the Bloomfieldian model as exemplified by Bloomfield's analyses of Tag. and Ilk. The second model is the tagmemic model developed by K . L . Pike which is used chiefly by the S.I.L. linguists. The most recent model is the transformationalgenerative model of the Chomskyan type which is favored by the linguists at the University of the Philippines and UCLA. These three models, especially the first two,

TAGALOG

137

do not differ so much on how they analyze the data; they differ more on how they present or arrange and label the results of the analysis.34 Of the Philippine languages, Tag. has continued to engage the attention of most of the linguists. However, some of the major languages have been neglected. Indeed there have been very few or no significant studies at all on some of the major languages, as we shall see next. We will take up first the general works and then the works on each of the major languages. 5.4.2 General works The Australian linguist A. Capell (1962) wrote a lengthy article on the Philippine verbal systems in which he charged some American linguists with making the analysis of the Philippine languages 'very much more difficult' by calling the passive forms 'pure verbs'. He says that these passive forms are not verbs but are 'verbal nouns'. His main argument is that the pronouns which occur with the passive forms are not nominative pronouns but are possessive pronouns which occur with noun stems in possessive constructions. 35 E. Constantino (1965) wrote a paper on the sentence patterns of twenty-six Philippine languages among which are the eight major ones. This paper presents an immediate constituent (IC) analysis followed by a transformational-generative analysis of the sentences in the languages. Constantino's syntactic analysis of the Philippine languages differs in some basic points from those of the other linguists, particularly Bloomfield and the S.I.L. linguists. He groups the predicative sentences according to the structure of their ICs into three types: DEFINITE, INDEFINITE, and SITUATIONAL. Each of the ICs of the definite sentence is marked, i.e. preceded by a particle or marker. Only the second ICs of the indefinite and situational sentences are marked. Constantino analyzes the first ICs of the definite and indefinite sentences and the second IC of the situational sentence as the subjects of the sentences, and the second ICs of the definite and indefinite sentences and the first IC of the situational sentence as their predicates. Thus, his analysis of the subject and the predicate of the Philippine languages differs markedly from the traditional one which Blake, Bloomfield, Lopez, and other linguists have followed. In the traditional analysis, the first IC of any predicative sentence in the normal order is the predicate, and the second IC the subject irrespective of the type of the sentence. Constantino has several reasons for departing from the traditional analysis of the subject and the predicate, which are as follows: 1. The centers (or heads) of the first ICs of the definite and indefinite sentences 34

One other model, the stratificational model, has been used to analyze and describe the structure of a Philippine language; see Austin (1966). The S.I.L.-Pike tagmemic model is described in McKaughan's article on the minor languages of the Philippines, pp. 155-67 in this volume. 86 Capell may not be entirely right in assuming that the passive construction and the possessive construction are syntactically the same. For example, a free (or full) possessive pronoun may substitute for a possessive construction, but not for a passive construction. Thus, in Tag. Akin ito. (mine this) 'This is mine.' may be a substitute for Bahay ko ito. (house my this) 'This is my house.', but not for Binili ko it6. (bought I this) 'I bought this'. See also McKaughan (1962:fn. 8).

138

ERNESTO CONSTANTINO

and of the second IC of the situational sentence are the same; i.e. they are all noun phrases. 36 On the other hand, the centers of the second ICs of the definite and indefinite sentences and of the first IC of the situational sentence are also the same; they are any of the following: verb, noun, adjective, or prepositional phrase. Thus, if the first IC of the situational sentence is analyzed as the predicate (as is done by all linguists), then the second ICs of the definite and indefinite sentences must also be analyzed as the predicates of these latter sentences. 2. The same marker (or class of markers) occurs in the first IC of the definite sentence and the second IC of the situational sentence; Constantino calls this marker the SUBJECT or NOUN (or SUBSTANTIVE) MARKER. On the other hand, another marker occurs in the second ICs of the definite and indefinite sentences; Constantino calls this the PREDICATE MARKER. This predicate marker is not the same as the subject marker, contrary to what most linguists have assumed. This fact is shown clearly by languages like Mar. and Samal (Sml.) where the two markers are morphemically different. 37 3. The nominative pronouns occur as the second IC of the situational sentence and the first IC of the definite sentence. Thus, in Constantino's analysis these pronouns always occur as subjects. In the traditional analysis, the nominative pronouns are subjects in situational sentences and predicates in definite sentences. 4. The first ICs of the definite and indefinite sentences and the second IC of the situational sentence have a transformational relation to each other. Two of the said ICs can be derived from the third one by simple transformation rules that operate on the kernel sentence (which may be the definite or situational sentence). On the other hand, the second ICs of the definite and indefinite sentences and the first IC of the situational sentence are transformationally related to each other. Thus the grouping of the ICs of the sentence types into subject constituents and predicate constituents is a consequence of their structural relationships to each other and not merely a consequence of their positions in the sentences.38 38

This statement properly applies to the 'kernel' sentences, which are the 'active' and the 'nonverbal' definite sentences. [Definite sentences and not situational sentences were chosen to provide the kernel sentences for the reason that a definite goal may not occur in situational sentences, but it does in definite (and indefinite) sentences.] An adjective which is either the goal or locative complement of a verb may become the 'transform' subject of an indefinite sentence, but not of a definite or situational sentence; cf. Constantino (1965:87). An adverb or adverbial phrase may also become a transform subject, as in Ngayong gabi ang alis ng bapor (now night the departure of ship) 'The ship leaves tonight'; cf. Aalis ang bapor ngayong gabi. 'The ship leaves tonight.' Also, the subject of a complex sentence may be a sentence; cf. Constantino (1965:107-8). See the last sentence of fn. 39 for the revised definition of subject. 87 The predicate marker in Mar. is i. So wataq i tuminindig (the child the stood-up) 'It was the child who stood up.' In Sml. the predicate marker is ya. Anak-anak itu ya bay paragon, (child this the past run-away) 'It was this child who ran away.' In Dibabawon (Forster 1964:39) the predicate marker is te. Kan umey te pigtanem ni Juan diyaq te pasak (the child the planted by John there the ground) 'The rice was the one which John planted in the ground*. 88 It seems that in the traditional analysis, the sole structural basis for identifying the subject and the predicate is their position in the sentence. The subject normally occurs after the first accented word of the predicate, or in initial position before a particle (e.g. Tag. ay) or a pause. It should be added

TAGALOG

139

In the transformational analysis of the sentences, Constantino derives the passive sentences from the active sentences. 39 The transformationalists at U C L A do not do this. They assign both the active and the passive sentences to the kernel. Unlike Bloomfield and the SIL linguists, Constantino distinguishes more than three passives. One obvious justification for this is the fact that many Philippine languages distinguish in the morphological shapes of their verbs more than three passives. For example, in Ilk. there are at least five passives which are indicated by the affixes -en, i-, -an, i-an, and pag-. Tag. uses only three affixes, -in, i-, and -an, to indicate the grammatical relations expressed by the five affixes of Ilk. Constantino does not say, as Bloomfield and the S.I.L. linguists do, that an affix indicates one and only one grammatical relation. It is clear that in many Philippine languages two or more affixes may be used to indicate the same grammatical relation between the subject and the verb. 40 On the other hand, one affix may indicate two or more different grammatical relations, not only with different stems but also with the same stems. 41 Finally, Constantino does not identify or define the passives simply on the basis of the verb affixes. He considers a sentence a goal-focus sentence (i.e. a sentence whose subject is the goal of the action expressed by the verb) not because of its affix but because its subject can be shown by simple transformation rules to have originated from the substantive expression that is marked as goal complement in the corre-

here that in the Constantino model any one of the verb complements may become the subject of the verbal sentence, and the predicate head of one sentence may not occur as the subject of another structurally-related sentence regardless of the sequencing of the constituents. See fn. 39. 39 This model has since been slightly revised, especially for the verbal sentences such that both 'active' and 'passive' sentences are derived from the formula: PV + C, in which Predicate) V(erb) is rewritten as P(redicate) M(arker) plus V(erb) or V only, and C(omplements) as one or more of the following depending on the subcategory of V: A(ctor) C(omplement), G(oal) C(omplement), Locative) C(omplement), B(enefactive) C(omplement), Instrumental) C(omplement), R(eciprocal) A(ctor) C(omplement), C(ausative) C(omplement). AC is rewritten as A(ctor) M(arker) plus N(oun) P(hrase); GC as G(oal) M(arker) plus NP or Adj(ective); LC as L(ocative) M(arker) plus NP or Adj; BC as B(enefactive) M(arker) plus NP; IC as I(nstrumental) M(arker) plus NP; RAC as R(eciprocal) A(ctor) M(arker) plus NP; and CC as C(ausative) M(arker) plus NP. NP is rewritten as Art(icle) plus N(oun) or N only. A transformation rule is introduced which does the following: It obligatorily deletes the marker of any one of the C and all the voice or case-marking 'features' of V, except the one which corresponds to the deleted marker; and it optionally deletes PM, except when the C whose marker is deleted is not an ArtN. The C whose marker is deleted becomes the 'subject' of the sentence. Thus, only complements may function as subject. 10 For example, in Tag. i- indicates that the subject is the goal in the first sentence and the instrument in the second sentence. (1) I-bigay mo ito sa kaniya. (give you this to him) 'You give this to him.' (2) I-putol mo ito naq kdhoy. (cut you this particle wood) 'You cut wood with this.' 41 For example, in Seb. the suffix -an indicates that the subject is either the location or the beneficiary of the action expressed by the verb. Palit-an ko siya-g dulsi. (will-buy I she-particle candy) 'I will buy for her some candy.' or 'I will buy from her some candy.' This sentence may also be 'disambiguated' by 'expanding' it with either: (1) a locative phrase, e.g. sa dalaga 'from the unmarried woman', in which|case the subject is unambiguously the beneficiary of the action; or (2) a benefactive phrase, e.g. para sa dalaga 'for the unmarried woman', in which case the subject is unambiguously the location of the action.

140

ERNESTO CONSTANTINO

sponding active sentence. Thus, Constantino says that the occurrence of the affixes -in, i-, and -an in the Tag. verb is determined not by the focus relation of the subject to the verb but by the subclass of the verb stem.42 Constantino (1967) collaborated with two others in describing the morphology and syntax of the personal pronouns of Tag., Ilk., Kap., and Isinai (Isn.). One of the significant contributions of this study is its 'discovery' in some Philippine languages, like Ilk. and Isn., of two enclitic pronouns of the third person plural nominative which behave differently in syntax. One occurs as the subject of passive sentences only, and the other as the subject of active and nonverbal sentences only. The existence of these two subject pronouns gives additional structural support to the basic dichotomy between active and passive sentences in the Philippine languages. This dichotomy has been rejected by the S.I.L. linguists. The Interchurch Language School (ILS) in the Philippines has described at least four of the major Philippine languages using the same descriptive model. This model seems to be a modified version of the S.I.L.-Pike tagmemic model; we will call it the 'ILS (tagmemic) model'. The first form of the ILS model is presented in the outline guides for Tag. (Larson 1963a), Ilk. (Larson 1963b), and Seb. (Larson 1963c). These guides are identical except for the examples from the languages. They were written when D. Larson was the director of the ILS. Larson's descriptive model for the three major Philippine languages (and presumably for most of the other Philippine languages as well) has two main divisions: Phonology and Grammar. The phonological and grammatical units are described in hierarchical order, from the smallest to the largest. We will discuss here only the grammatical units, especially as they are presented in the Tag. guide (Larson 1963a). There are three classes of grammatical units: words, phrases, and sentences. The words are either free forms or bound forms. The free forms are divided into bases and substitutes, and the bound forms into affixes, PREPOSITIVES, and POSTPOSITIVES. The bases are either unaffixed (nouns) or affixed. The affixed bases are either verbs or descriptives. The verbs are divided semantically into STATIC and DYNAMIC verbs, and morphologically and syntactically into four types: SOURCE, INERTANT, TERMINAL, and AUXILIARY. The source verbs are equivalent to our active verbs, and the other types to our passive verbs. The four verb types are distinguished from each other morphologically by their affixes. In Tag., -um- and mag-, -in and i-, -an, i- are the principal affixes of the source, inertant, terminal, and auxiliary verbs, respectively (Larson 1963a: 29). The corresponding affixes in Seb. are mo- and mag-, -on and i-, -an, i- (Larson 1963c: 47-8); and in Ilk., -um- and -ag, -en and /-, -an, i- (Larson 1963b: 28). These verb types determine the different roles of the subjects of the sentences. 42

Thus, in Tag., when the goal is the subject, -in is used with a group of verb stems, like bili 'buy'; i- with a second group, like bigay 'give', and -an with a third, like bukas 'open'. See Constantino 1968:29 30.

TAGALOG

141

Like McKaughan (1958:10-11), Larson divides the phrases into CENTERED and UNCENTERED. Roughly speaking, centered phrases are endocentric, and uncentered phrases exocentric. Larson sets up four positional classes in the sentence: TOPIC, COMMENT, modifier, and particle. The topic is identified (perhaps by mistake) with the predicate, and the comment with the subject. The topic is usually 'unfocused' (i.e. not preceded by an article), while the comment is usually 'focused'. Larson's unfocused comment (1963a: 130-1) is what we call indefinite goal or object, or passive actor complement or agent, and not subject. Both topic and comment may be focused in the same sentence. Finally, the sentences are divided into two types: MINOR and MAJOR. A sentence without a topic is a minor sentence; all other sentences are major sentences (1963a: 142).43 When M. Eyestone became the director of the ILS, he modified Larson's model for Ilk.; see Eyestone (1965). He divides the Ilk. sentences into two parts or slots. The second slot is the subject slot, and the first slot the predicate slot. However, Eyestone says a sentence may have two subjects (and no predicate) or no subject at all (only a predicate). It must have become very obvious to Eyestone that Larson's four verb types do not fit the Ilk. verbal system. He therefore divides the Ilk. verbs into six focus types: ACTOR f o c u s , INERTANT f o c u s , LOCATION foCUS, INSTRUMENT foCUS, ACCESSORY foCUS,

and BENEFICIARY focus. The principal affixes of these focus types are -um- and ag-, -en and /-, -an and pag-an, pag-, i-, and i-an, respectively. The latest modifications on the ILS model have been made under the present directorship of F. Flores; see Muyargas et al. (1967). The first part of the Tag. sentence is called comment and the second part topic. There are two classes of verbs according to voice: ACTOR TOPIC verbs, which indicate 'the action done by the Topic'; and GOAL TOPIC verbs, which indicate 'the action done to, at, on, for, or with the Topic' (Muyargas et al. 1967:221). The goal topic verbs are divided into three types according to their affixes. The first type of goal topic verbs occur with -in or ma-, the second type with i- or mai-, and the third type with -an or ma-an. 5.4.3 Works on Tag. At least three studies have been made on the phonology of Tag. since the end of the Second World War. The first of these was written by R. Cayari (1956). Cayari ignores the existence of the phonemes /e/ and /o/ in Tag. although she cites borrowed words as well as native words in Tag. in which the two sounds contrast. The two other phonological studies on Tag. are by R. Stockwell (1957) in the first part of his contrastive analysis of Tag. and English and by Llamzon (1966a). In these two studies, the authors identify and describe the phonemes of Tag., which include 48

Larson (1965) wrote a simultaneous grammar of Tag., Ilk., and Seb. as his Ph.D. dissertation at the University of Chicago. Unfortunately, this writer has not been able to get hold of a copy of this unpublished work.

142

ERNESTO CONSTANTINO

/e/ and /o/, as well as the consonant clusters and the intonation patterns of the language. There have been more studies on the grammar of Tag. One of these is a 'restatement' of Tag. grammar by E. Wolfenden (1961) whose aims are (1) to 'realign' Bloomfield's grammatical categories, especially the verbal ones; and (2) to modify Bloomfield's nomenclature. The principal differences between Wolfenden's grammatical analysis of Tag. and that of Bloomfield's are terminological. It seems that many of the new terms introduced by Wolfenden have not gained wide currency. The S.I.L. tagmemic analysis of Tag. grammar is described briefly by Waterman (1960) in an article dealing with syntactic problems in the translation of the Scriptures in the Philippine languages. The article discusses how the S.I.L. tagmemicists analyze the Tag. sentences, substantive constructions, pronouns, and verbs. A transformational analysis of a segment of Tag. grammar has been made by P. Schachter (1962). The same author (1961) has written an article on structural ambiguity in Tag. In this article he shows how the structural ambiguity in two Tag. sentences can be accounted for by tracing the transformational histories of the sentences. His fourteen constituent-structure rules for Tag., especially rules 8 to 11, show that he considers the Tag. passive sentences as kernel sentences. Four master's theses dealing with different aspects of Tag. grammar have been written at the University of the Philippines in the Department of Oriental Languages and Linguistics. The first one (Silverio 1962) examines the passive verbs and the passive sentences in Tag., and the second (Gonzales 1962) the active verbs and the active sentences. These two theses classify the verbs as well as the verbal sentences, affixes, and stems. The sentences are classified according to their subjects; the verbs according to their stems, affixes, and complements; the affixes according to their subjects; and the stems according to their affixes. The third thesis (Cayari 1963) classifies the Tag. time adverbs on the basis of their distribution. The fourth thesis (Paz 1967) is a study of the morphology and syntax of the Tag. nouns and adjectives. Three recent works on Tag. grammar are (1) a study of the syntactic features of the Tag. verbs by J. Kess (1967), (2) a syntactic analysis of the basic sentence types in Tag. by A. Buenaventura (1967), and (3) a Ph.D. dissertation entitled 'Modern Tagalog: a functional-structural description with particular attention to the problem of verification' by Llamzon (1968). Kess's study was his Ph.D. dissertation at the University of Hawaii. Buenaventura's analysis of the basic Tag. sentence types, which is intended for the use of non-native speakers of Tag., is essentially the same as Bloomfield's. Four articles on Tag. morphology and syntax have appeared recently. One of these (Buhain 1958) describes the uses of the Tag. ligature na. Another one, written by Pittman (1966), describes and compares the different uses or meanings of the two most commonly used affixes in Tag., namely -um- and mag-. The two others are both by Llamzon (1966b, 1966c). One of them is on predication in Tag., and the other on the main transient (verb) formations in Tag. Both articles of Llamzon are based on Bloomfield's Tag. grammar.

TAGALOG

143

At least five contrastive analyses of Tag. and English have been made. Two of these (Stockwell 1957; Castelo 1964) present an analysis of Tag. that is basically the same as that of Bloomfield. The third one (Guanco 1963) is a Ph.D. dissertation which describes the structure of Tag. in terms of the tagmemic framework. The other two works are the Ph.D. dissertations of E. Pascasio (1960) and F. Otanes (1966). There have also been several practical courses in Tag. which contain explanations about the structure of the language. The most well-known of these are the Intensive Tagalog conversation course by E. Wolfenden and R. Alejandro (1957) and Beginning Tagalog, edited by J. Donald Bowen (1965). The analysis of Tag. presented in the first course is based on the Bloomfieldian model; that of the second is similar to the S.I.L. tagmemic model. 44 Finally, Conklin (1956) has described an interesting form of linguistic play in Tag. which he calls 'speech disguise'. 5.4.4 Works on Seb. The first linguistic work on Seb. published after the last war is the short article on Seb. syntax by Percy and Kathleen Meiklejohn (1953). This article analyzes the Seb. sentence structure on the basis of the relationships of the verbal prefixes to the other parts of the sentence. The verbal prefixes are grouped into PRIMARY (or active) and SECONDARY (or passive). The substantives may occur in primary (or subject), secondary (or goal or agent), or TERTIARY (or locative or referent) relation to the verbal stem depending on the verbal affix used. The syntax and the affixes of Seb. have also been described in a monograph by Bergh (1958). This work, which is intended to help young priests learn the language, is based on the analysis of some 4000 lines of Seb. text. The author uses many new and unfamiliar terms, and many of his examples are not used in ordinary conversations in the language. The Cebuano reference materials by V. Morey (1961) seems to have been based chiefly on this work, although the terminology is from McKaughan's work on Mar. (1958). Two doctoral dissertations on the contrastive analysis of Seb. and English have been written. In the first one (Flores 1963), the analysis of Seb. is based on a modified version of Pike's tagmemic model. The second one (Anderson 1965) describes the syntax of Seb. using Chomsky's transformational model. John Wolff (1966, 1967) has produced two volumes of Seb. lessons in which the grammar of Seb. is described in accordance with the Bloomfieldian model. 45 5.4.5 Works on Ilk. The first grammatical description of Ilk. after the last war is presented in the inten41

Bowen adds two focuses to the Bloomfieldian and S.I.L. four focuses, namely the BENEFACTIVE and the CAUSATIVE. 46 An article by Wolff on the history of a Seb. dialect is coming out in the next number of OL. Also, Wolff's Ph.D. dissertation (1965) is a Seb. Visayan syntax.

144

ERNESTO CONSTANTINO

sive Ilk. course by H. McKaughan and J. Forster (cl952). This Ilk. grammar was intended to serve 'as a suggestive guide or model for the production of similar grammars in many of the other Philippine languages in the years to come' (p. ii). It is based on Bloomfield's "Outline of Ilocano syntax" (1942). A transformational-generative grammar of Ilk. was written by Constantino (1959) as his Ph.D. dissertation. The model for this work is the grammar presented by Chomsky in his Syntactic structures. Constantino (1963b) has also written a short article on the Ilk. pluralizers, their allomorphs, and their distribution. Three analyses of the semantic components of the Ilk. personal pronouns are discussed in an article by Thomas (1955). In this article, the nominative and possessive pronouns are analyzed as allomorphs. Also, the third person pronouns are analyzed as mere number indicators and allomorphs of the prefix of the plural pronouns. A Ph.D. dissertation contrasting the Ilk. segmental phonemes with those of English has been written by B. Sibayan (1961). 5.4.6 Works on the other languages Not much has been written about the other major Philippine languages. In fact, this writer has not been able to find a single linguistic work on Bkl. published after the last war.46 At least three works on Kap. have been written since the last war. The first of these is a master's thesis (Castrillo 1955) written at the University of the Philippines which describes the structure of the Kap. sentences using the Bloomfieldian model. The second work is a Ph.D. dissertation (Clardy 1958)47 which describes the phonemes of Kap., their allophones, and their distribution. The third work is an article (Tabasondra 1962) in which the Kap. phonemes are listed and compared with the English sounds. The structure of Hil. has been described and contrasted with that of English in at least two doctoral dissertations (Juntado 1961; Ruiz 1963). There has been only one work on Png., Schachter's Ph.D. dissertation (1959), which is a contrastive analysis of English and Png. This work presents a small segment of Png. grammar cast in the transformational model promulgated by Chomsky in his Syntactic structures. Schachter, who says there are no narrative predications in Png., assigns the Png. passives to the kernel. War. has been described in only one work, Beginning Waraywaray, by J. and I. Wolff (1967). Their grammatical description of the language is based on the Bloomfieldian model.48 46

Stevens has written a paper on the morphophonemics of the suprasegmental in Bkl. which is due for publication. 47 A 'shortened version' of this dissertation was published in 1959 (Clardy 1959). 48 Wolff has an article on the schwa in War. which is coming out in Samar-Leyte Studies, a publication of the Divine Word University, Tacloban, Leyte, Philippines. A traditional grammar of War. has been written by Romualdez (1908).

TAGALOG

5.5

145

Summary

The foregoing review of the works on the major languages of the Philippines shows that the study of these languages after the Second World War has followed the trends in post-war American linguistics. This is not surprising for two reasons. First, about half of those who have been studying the major languages are Americans. Second, the Filipinos working on the languages have all studied at American universities or at universities in the Philippines with American professors or Filipino professors who studied in the United States.49 Most of the works on the major languages are descriptive, with little emphasis on phonology. The majority of the linguists are more interested in analyzing and describing the morphological and syntactic features of the languages. This interest is primarily due to the development in the United States of two new grammatical models: the tagmemic model and the transformational-generative model. Most descriptive linguists working on the major Philippine languages today use either one of these two models; only a few still use the Bloomfieldian model. Most of the post-war works on the major Philippine languages have been produced by graduate students, both Americans and Filipinos. This is reflected by the fact that almost half of the linguistic works are doctoral dissertations and master's theses most of which have not been published. The majority of the dissertations are contrastive analyses of English and the major languages, except Bkl., War., and Kap. The participation of the Filipinos in the study of the major Philippine languages, not as informants only but as both informants and analysts, has particularly become noticeable after the Second World War, especially after 1960. Almost half of those studying the major languages during this period are Filipinos who have contributed about half of the published works. Due to recent developments in linguistic theory, especially with regard to the crucial role of the native speaker's linguistic intuitions and competence in the scientific study of language, the future study of the major Philippine languages will demand a much closer and deeper cooperation between Filipino linguists and other linguists.

REFERENCES 1968. National identity through language teaching assessed. The Manila Times, March 4, 1968, 14-A-15-A. ANDERSON, TOMMY RAY. 1965. A contrastive analysis of Cebuano Visayan and English. Ph.D. dissertation, University of California, Los Angeles. ASUNCION-LAND^, NOBLEZA. [ 1 9 6 5 ] . Bibliography of Philippine linguistics (preliminary edition). [Mimeo.]

AGUILAR, JOSE V .

" Even Lopez spent a year in the United States shortly after the Second World War, after which he started teaching the American type of linguistics at the University of the Philippines.

146

ERNESTO CONSTANTINO

Attention, emphasis and focus in Ata Manobo. Hartford Studies in Linguistics, 20. Hartford, Connecticut, Hartford Theological Seminary. BERGH, J. D. v. D. 1958. Analysis of the syntax and the system of affixes in the Bisayà' language. Translated by G. Trienekens. Surigao, Sacred Heart Missionaries. BEYER, H . OTLEY. 1 9 1 7 . Population of the Philippine Islands in 1 9 1 6 . Manila, Philippine Education Co. BLAIR, EMMA, and JAMES ALEXANDER ROBERTSON. 1 9 0 3 - 0 9 . The Philippine Islands, 1 4 9 3 - 1 8 0 3 ... vol. 1. [Cleveland, Ohio, A.H. Clark Co.] BLAKE, F R A N K R . 1 9 0 2 . Study of Philippine languages at Johns Hopkins University. AmA (n.s.) 4 . 7 9 3 - 4 . . 1903a. Sanskrit loanwords in Tagalog. Johns Hopkins University Circulars 22.63-5. . 1903b. Analogies between Semitic and Tagalog. Johns Hopkins University Circulars 22.65-6. . 1904. Differences between Tagalog and Bisayan. JAOS 25.162-9. . 1905. The Bisayan dialects. JAOS 26.120-36. . 1906a. Expression of case by the verbs in Tagalog. JAOS 27.183-9. . 1906b. Contributions to comparative Philippine grammar. JAOS 27.317-96. . 1907. Contributions to comparative Philippine grammar, II. JAOS 28.199253. . 1908. The Tagalog ligature and analogies in other languages. JAOS 29.227-31. . 1910a. Expression of the ideas 'to be' and 'to have' in the Philippine languages. JAOS 30.375-91. . 1910b. Review of The first grammar of the language spoken by the Bontoc Igorot, by C.W. Seidenadel. AJPh 31.339-42. . 1911a. Tagalog verbs derived from other parts of speech. AJPh 32.436-40. . 1911b. Review of The RGH law in Philippine languages, by C.E. Conant. AmA 13.472-3. . 1911c. Philippine literature. AmA 13.449-57. . 1916a. The Tagalog verb. JAOS 36.396-414. . 1916b. Construction of coordinated words in the Philippine languages. AJPh 37.466-74. . 1917. Reduplication in Tagalog. AJPh 38.425-31. . 1918. Review of A grammar of Lepanto Igorot as it is spoken at Bauco, by M. Vanoverbergh. AJPh 39.417-20. . 1919. Review of Tagalog texts with grammatical analysis, by L. Bloomfield. AJPh 4.86-93. . 1920. A bibliography of the Philippine languages. JAOS 40.25-70. . 1921. A new method of syntactical arrangement. JAOS 41.467-71. . 1921-25. Instructions for Philippine linguistic workers. [Mimeo.]

AUSTIN, VIRGINIA.

1966.

TAGALOG

147

. 1922. The part played by the publications of the U.S. government in the development of Philippine linguistic studies. JAOS 42.147-70. . 1925. A grammar of the Tagalog language. American Oriental Series, 1. New Haven, Connecticut, American Oriental Society. . 1926a. The study of Philippine languages at Johns Hopkins University and its bearing on linguistic science. Johns Hopkins University Alumni Magazine 14.389-400. . 1926b. Review of The language of education of the Philippine Islands, by N . M . Saleeby. JAOS 46.61-71. . 1929. Review of Comparison of Tagalog and Iloko, by C. Lopez. Lg 5.190-2. [ ]. 1930a. Philippine languages. The New International Encyclopaedia. 2nd ed., vol. 18, pp. 501-3. New York, Dodd, Mead & Co. . 1930b. A semantic analysis of case. Curme volume of linguistic studies, ed. by James T. Hatfield et al., pp. 34-49. Baltimore, Waverly Press. . 1938. The study of language from the semantic point of view. Indogermanische Forschungen 56.241-55. . [1947]. Selected list of materials for the study of the Tagalog language. Southeast Asia Institute. Language Series, 4. New York, Southeast Asia Institute. . 1950. Tagalog noun formation. JAOS 70.271-91. BLOOMFIELD, LEONARD. 1917. Tagalog texts with grammatical analysis. University of Illinois studies in language and literature, 3.2-4. Urbana, Illinois, University of Illinois. . 1933. Language. New York, Holt. . 1942. Outline of Ilocano syntax. Lg 18.193-200. BLUMENTRITT, FERDINAND. 1899. List of the native tribes of the Philippines and of the languages spoken by them. Translated with introduction and notes by O. T. Mason. Annual report of the Smithsonian Institution for 1899, pp. 527-47. BOWEN, J . DONALD, ed. 1965. Beginning Tagalog. Berkeley and Los Angeles, University of California Press. BRANDSTETTER, RENWARD. 1902. Tagalen und Madagassen. Luzern, J . Eisenring. . 1908. Mata-Hari. Luzern, E. Haag. BUENAVENTURA, AMPARO S. 1967. A syntactic analysis of the basic sentence types in Tagalog. San Carlos publications, series A: Humanities, 6. Cebu, University of San Carlos. BUHAIN, AMPARO. 1958. The Tagalog particle na. PhilJSci 87.473-8. CAPELL, ARTHUR. 1962. Oceanic linguistics today. Current Anthropology 3.371428. . 1964. Verbal systems in Philippine languages. Philippine Journal of Science 93.231-49. CASTELO, LUTGARDA MENDOZA. 1964. Structural differences between English and Tagalog verbs; a study to improve the teaching of English to advanced Filipino students. Ed.D. dissertation, Columbia University.

148

ERNESTO CONSTANTINO

1955. Pampango syntax. Master's thesis, University of the Philippines. CAYARI, REMEDIOS M. 1956. The phonemes of Tagalog. PhilJSci 85.249-55. . 1963. Tagalog time adverbs. Master's thesis, University of the Philippines. Census of the Philippines: 1939. 1941. Vol.2. Manila, Bureau of Printing. Census of the Philippines: 1948. 1954. Vol. 3, pt. 1. Manila, Bureau of the Census and Statistics. Census of the Philippines: 1960. 1963. Vols. 1 & 2. Manila, Bureau of the Census and Statistics. CHRETIEN, C. DOUGLAS. 1962. A classification of twenty-one Philippine languages. CASTRILLO, MARIA LUISA Y.

PhilJSci 91.485-506. CLARDY, CATHERINE JANE.

1958. Pampango phonology.

Ph.D. dissertation, Uni-

versity of Texas. . 1959. Pampango phonology. Phonetica 3.118-44. CONANT, CARLOS EVERETT. 1906. A Bisaya-English dictionary, prepared at Cebu, P.I. with collaboration of Vicente Sotto and Juan Villagonzalo. [Ms.] . 1908. 'F' and 'V' in Philippine languages. Division of Ethnology Publications 5.135-41. Reprinted in Galang (1938:169-78). . 1909. The names of Philippine languages. Am A 4.1069-70. . 1910a. The RGH law in Philippine languages. JAOS 31.70-85. . 1910b. The Bisaya language: its evolution in the last decade. Ang Suga (Cebu), June 16, 1910, 1. . 1911. Monosyllabic roots in Pampanga. JAOS 31.389-94. . 1912. The pepet law in Philippine languages. Anthropos 7.921-47. Reprinted in Galang (1938:126-68). . 1913. Notes on the phonology of the Tirurai language. JAOS 33.150-7. . 1915a. Grammatical notes on the Isinai language (Philippines). JAOS 35. 289-92. . 1915b. The language problem in the Philippines. Report of the Thirty-third Annual Lake Mohonk Conference on Indian and other Dependent Peoples, pp. 98-102. . 1916. Indonesian I in Philippine languages. JAOS 36.181-96. CONKLIN, HAROLD C. 1949. Preliminary report on field work on the islands of Mindoro and Palawan, Philippines. Am A 51.268-73. . 1952. Outline gazetteer of native Philippine ethnic and linguistic groups. [Mimeo.] . 1955. Preliminary linguistic survey of Mindanao. [Mimeo.] . 1956. Tagalog speech disguise. Lg 32.136-9. CONSTANTINO, ERNESTO. 1959. A generative grammar of a dialect of Ilocano. Ph.D. dissertation, Indiana University. . 1963a. Some problems in Philippine linguistics. Asian Studies 1.23-30.

TAGALOG

149

. 1963b. Ilukano pluralizers. PSSR 28.409-15. . 1965. The sentence patterns of twenty-six Philippine languages. Lingua 15. 71-124. . 1968. Models for the description of the Philippine languages. Papers of the Linguistic Society of Hawaii 1.22-30. . 1970. The deep structures of the Philippine languages. (Mimemo). , CONSUELO J . PAZ, and MARIETTA N . POSONCUY. 1967. The personal pronouns of Tagalog, Ilukano, Isinai and Kapampangan. Studies in Philippine anthropology (in honor of H. Otley Beyer), ed. by Mario D. Zamora, pp. 567-91. Quezon City, Philippines, Phoenix Press. COSTENOBLE, H. 1937a. Tracing the original sounds of the languages of today. Philippine Magazine 34.24, 38-9. Reprinted in Galang (1938:111-16). . 1937b. Monosyllabic roots. Philippine Magazine 34.76, 82-4. Reprinted in Galang (1938:116-25). . 1937c. The Philippine verb. Philippine Magazine 34.169-70, 180. . 1938. Language notes. In Galang (1938:107-11). . 1940. Die Lautwandlungen des Tagalog. Bijd. 99.65-87. DEMPWOLFF, OTTO. 1 9 2 5 - 2 6 . Die L-, R - und D-Laute in austronesischen Sprachen. Z E S 15.19-50, 116-38, 223-38, 273-319.

. 1934-38. Vergleichende Lautlehre des austronesischen Wortschatzes. Beihefte zur ZES 15, 17, 19. DYEN, ISIDORE. 1947a. The Tagalog reflexes of Malayo-polynesian D . Lg23.227-38. . 1947b. The Malayo-polynesian word for 'two'. Lg 24.50-5. . 1951. Proto-Malayo-polynesian *Z. Lg 27.534-40. DYEN, ISIDORE. 1953a. The Proto-Malayo-polynesian laryngeals. William Dwight Whitney series. Baltimore, Linguistic Society of America. . 1953b. Dempwolff's *R. Lg 29.359-66. . 1953c. Review of Malgache et maanjan: une comparaison linguistique, by Otto Chr. Dahl. Lg 35.577-90. . 1962a. Some new Proto-Malayopolynesian initial phonemes. JAOS 82.214-15. . 1962b. The lexicostatistical classification of the Malayo-polynesian languages. Lg 38.38-46. . 1963. Lexicostatistically determined borrowing and taboo. Lg 39.60-6. . 1965. A lexicostatistical classification of the Austronesian languages. IJAL supplement vol. 31, no. 1. Baltimore, Waverly Press. EYESTONE, MAYNARD M . , ed. 1965. Ilocano grammar and vocabulary. 3 vols. Manila, Interchurch Language School. FLORES, FRANK G. 1963. Analysis of selected clause types in Cebuano and English. Ed.D. dissertation, University of Michigan. FORSTER, JANNETTE. 1964. Dual structure of Dibabawon verbal clauses. OL 3 . 2 6 - 4 8 . Fox, ROBERT B . , WILLIS E . SIBLEY, and FRED EGGAN. 1965. A preliminary glottochronology for Northern Luzon. Asian Studies 3 , 1 0 3 - 1 3 ,

150

ERNESTO CONSTANTINO

J. 1947. Tagalog as the Philippine national language : the inception and development of the idea of a national language, final choice of Tagalog, and the history and character of its grammatical treatment and promotion. Ph.D. dissertation, Hartford Seminary Foundation. . 1949-50. The historical development of the Philippine national language. PSSR 14.367-400(1949); 15.45-79, 163-94(1950). GALANG, ZOILO, ed. 1938. Filipiniana, Vol. 2: Linguistics. Manila, Philippine Education Co. GAMBOA-MENDOZA, VIRGINIA. 1940. Phonological peculiarities of Pampangan (based on the OrAus. as constructed by Dempwolff). Publications of the Institute of National Language, Bulletin no. 4. Manila, Bureau of Printing. GONZALES, LYDIA FER. 1962. The active sentences and active verbs in Tagalog. Master's thesis, University of the Philippines. GUANCO, NELIA R. 1963. A descriptive-contrastive analysis of English and Tagalog verbs. Ph.D. dissertation, University of Michigan. HENDON, RUFUS S. 1964a. The reconstruction of *-ew in Proto-Malayopolynesian. Lg 40.372-80. . 1964b. The Proto-Malayopolynesian word for 'ladder, staircase'. JAOS 84.259-64. HUMBOLDT, WILHELM VON. 1838. Über die Kawi Sprache auf der Insel Java. Zweiter Band. Berlin, Druckerei der Königlichen Akademie der Wissenschaften. JUNTADO, LORETO GRAJO. 1961. Number concord in English and Hiligaynon. Ph.D. dissertation, University of Michigan. KESS, JOSEPH FRANCIS. 1967. Syntactic features of Tagalog verbs. Ph.D. dissertation, University of Hawaii. [LARSON, DONALD N., ed.]. 1963a. Tagalog outline guide. Part 1 : description. Manila, Interchurch Language School. [ , ed.]. 1963b. Ilocano outline guide. Part 1 : description. Manila, Interchurch Language School. , ed.]. 1963c. Cebuano outline guide. Part 1 : description. Manila, Interchurch Language School. . 1965. Toward a simultaneous multilingual grammar of Tagalog, Cebuano and Ilocano. Ph.D. dissertation, University of Chicago. LENDOYRO, CONSTANTINO. 1902. The Tagalog language. Manila, Imprenta 'EL Mercantil'. LLAMZON, TEODORO A . 1965. A subgrouping of nine Philippine languages. [Ms.] . 1966a. Tagalog phonology. AnL 8.30-9. . 1966b. Tagalog reflexes of PMP *e. AnL 8.13-23. . 1966c. The subgrouping of Philippine languages. PSR 14.145-50. . 1966d. A note on predication in Tagalog. PSR 14.150-5. , 1966e. Main transient formations in Tagalog. PhilJSci 95.143-57. FREI, ERNEST

TAGALOG

151

. 1968. Modern Tagalog: a functional-structural description with particular attention to the problem of verification. Ph.D. dissertation, Georgetown University. LOPEZ, CECILIO. 1928. Comparison of Tagalog and Iloko. Hamburg, J . J . Augustin. . 1931a. Das Verwantschaftssystem der Tagalen und der Malaien. Archiv fiir Anthropologie XII/3. . 1931b. The language situation in the Philippine Islands. Manila. . 1937a. Tagalog words adopted from Spanish. PSSR 8.223-46. . 1937b. Preliminary study of the affixes in Tagalog. Publications of the Institute of National Language, vol. 2. Manila, Bureau of Printing. . 1937c. General features of Philippine languages. PSSR 9.201-7. . 1939a. Studies on Dempwolff's 'Vergleichende Lautlehre des austronesischen Wortschatzes'. Publications of the Institute of National Language, Bulletin no. 1. Manila, Bureau of Printing. . 1939b. A comparison of Tagalog and Malay lexicographies (on phoneticosemantic basis). Publications of the Institute of National Language, Bulletin no. 2. Manila, Bureau of Printing. . 1940a. A manual of the Philippine national language. 3rd ed. Manila, Bureau of Printing. . 1940b. The Tagalog language: An outline of its psychomorphological analysis. Publications of the Institute of National Language, Bulletin no. 5. Manila, Bureau of Printing. . 1941a. Studies on Tagalog lexicography (Tagalog words for Spanish loanwords) no. 1. Publications of the Institute of National Language, Bulletin no. 6. Manila, Bureau of Printing. . 1941b. Reduplication in Tagalog. Publications of the Institute of National Language, Bulletin no. 7. Manila, Bureau of Printing. . 1949. Studies on Sugbuhanon affixes. PSSR 14.29-35, 183-226. . 1965a. Contributions to a comparative Philippine syntax. Lingua 15.3-16. . [1965b]. A critical survey of works on the Philippine languages, 1900-1963. [Typescript.] . 1967. Origins of the Philippine languages. Philippine Studies 15.130-66. MACKINLAY, WILLIAM E. W . 1902. Memorandum on the languages of the Philippines. Journal of the Anthropological Institute of Great Britain 31.214-18. . 1904. Some minor languages of Luzon. JAOS 24.170-4. . 1905. A handbook and grammar of the Tagalog language. Washington, Government Printing Office. MARCILLA Y MARTIN, CIPRIANO. 1895. Estudio de los antiguos alfabetos filipinos. Malabon, Tipolitografia del Asilo de Huerfanos. MARRE, ARISTIDE. 1901. Grammaire tagalog, composée sur un nouveau plan. Bijd. 53.547-92,

152

ERNESTO CONSTANTINO

1958. The inflection and syntax of Maranao verbs. Publications of the Institute of National Language. Manila, Bureau of Printing. . 1960. Review of Analysis of the syntax and the system of affixes of the Bisaya' language, by J.D. v.d. Bergh. Philippine Studies 8.648-55. MCKAUGHAN, HOWARD, and JANNETTE FORSTER. 1952. Ilocano: an intensive language course. Grand Forks, North Dakota, Summer Institute of Linguistics. . 1962. Overt relation markers in Maranao. Lg 38.47-51. MEIKLEJOHN, PERCY and KATHLEEN. 1953. Notes on Cebuano syntax. Folklore Studies 12.96-100. MILNER, G.B. 1963. Oceanic linguistics. Trends in modern linguistics, ed. by Christine Mohrmann, F. Norman and Alf Sommerfelt, pp. 62-94. Utrecht, Spectrum Publishers. MOREY, VIRGINIA. 1961. Cebuano reference materials. Edited by Howard P. McKaughan. A joint publication of the Summer Institute of Linguistics and the Philippine Association for Language Teaching, Philippines. MUYARGAS, WILFREDO, ROSA SOBERANO, FRANK FLORES, and the ILS staff. 1967. Tagalog for non-Filipinos. Vol. 1, units 1-4. Manila, Interchurch Language School. OTANES, FE. 1 9 6 6 . A contrastive analysis of English and Tagalog verb complementation. Ph.D. dissertation, University of California, Los Angeles. PASCASIO, EMY M. 1960. A descriptive-comparative study predicting interference and facilitation for Tagalog speakers in learning English noun-head modification patterns. Ph.D. dissertation, University of Michigan. PAZ, CONSUELO J . 1967. The morphology and syntax of Tagalog nouns and adjectives. Master's thesis, University of the Philippines. PHELAN, JOHN LEDDY. 1955. Philippine linguistics and Spanish missionaries. Mid America 3 7 . 1 5 3 - 7 0 . PIKE, KENNETH L. 1963. A syntactic paradigm. Lg 3 9 . 2 1 6 - 3 0 . PITTMAN, RICHARD. 1966. Tagalog -urn- and -mag-. LCC, A, 8. PORTER, R.S. 1903. A primer and vocabulary of the Moro dialect (Magindanau). Washington, Government Printing Office. ROE, G. RICHARD. 1967. Current research of the Summer Institute of Linguistics in the Philippines. OL 6.52-61. ROMUALDEZ, NORBERTO. 1908. A Bisayan grammar (Samaro-Leytean). Takloban (Leyte), Pagpahayag Co. Ruiz, MACARIO B. 1963. Weighting and sequencing English tense-aspect modifications for Hiligaynon speakers. Ed.D. dissertation, University of California, Los Angeles. SCHACHTER, PAUL. 1959. A contrastive analysis of English and Pangasinan. Ph.D. dissertation, University of California, Los Angeles. [Mimeo.] . 1961. Structural ambiguity in Tagalog. LL 11.135-45. MCKAUGHAN, HOWARD.

TAGALOG

153

. 1962. Rules for a segment of Tagalog grammar. NDEA Tagalog project. Los Angeles, California, University of California, ditto ed. SCHEERER, OTTO. 1924. On the essential differences between the verbs of the European and the Philippine languages. Philippine Journal of Education 7 . 1 - 1 0 . , and E. PABLO. 1925-26. The use of ti and iti in Iloko compared with Tagalog and Pangasinan equivalents. The Archive, paper no. 4. SCHURMANN, JACOB GOULD. 1901. Report of the Philippine Commission to the President. Vol. 3. Washington, Government Printing Office. SEBEOK, THOMAS A. 1943. The languages of Southeastern Asia. Far Eastern Quarterly 2.349-456. SEIPLE, WILLIAM G. 1903. Tagalog numerals. Johns Hopkins University Circulars 22.79-80. . 1904. Roots with initial p in Tagalog. JAOS 25.287-301. SIBAYAN, BONIFACIO PADILLA. 1961. English and Iloko segmental phonemes. Ph.D. dissertation, University of Michigan. SILVERIO, ANTONIA C. 1962. The passive verbal sentence constructions in Tagalog. Master's thesis, University of the Philippines. SMITH, C . C . , transi. 1906. A grammar of the Magindano tongue. Washington, Government Printing Office. STOCKWELL, ROBERT P . 1957. A contrastive analysis of English and Tagalog. Los Angeles, California, University of California. [Mimeo.] SWIFT, HENRY. 1909. A study of the Iloko language. Washington, Government Printing Office. TABASONDRA, ILUMINADA S. 1962. Pampango consonants and vowels and their influence on English as spoken by Pampangos in Tarlac. PJL 1.1-6. TANGCO, MARCELO. 1951. The Christian peoples of the Philippines. Natural and Applied Science Bulletin 1 1 . 9 - 1 1 5 . THOMAS, DAVID. 1955. Three analyses of the Ilocano pronoun system. Word 11. 204-8. THOMAS, DAVID, and ALAN HEALEY. 1962. Some Philippine language subgroupings : a lexicostatistical study. AnL 4 . 2 1 - 3 3 . TOTANES, S. DE. 1865. Arte de la lengua tagala y manual tagalog. 4th ed. Binondo, Miguel Sanchez y Ca. VANOVERBERGH, MORICE. 1928. Notes on Iloko. Anthropos 2 3 . 1 0 2 9 - 5 0 . . 1931. Iloko substantives and adjectives. Anthropos 26.469-88. . 1933. Pronouns and numbers in Iloko. Anthropos 28.683-720. . 1937. Some undescribed languages of Luzon. Publications de la Commission d'Enquête Linguistique, 3. Nijmegen, Dekker & van de Vegt N.V. . 1938. 'To have'and'To be'in Iloko. PhilJSci 66.417-38. . 1939. The Iloko adjectival voice. PhilJSci 69.223-56. . 1940a. The Iloko substantival voice. PhilJSci 72.9-37.

154

ERNESTO CONSTANTINO

. 1940b. The verbal prefixes mang and ma and defective verbs in Iloko. PhilJSci 72.421-49. . 1941. Supplementary notes on the Iloko verb. PhilJSci 75.197-225. . 1949. Adverbs and prepositions in Iloko. PhilJSci 78.167-205. . [1955], Iloko grammar. [Baguio, Catholic School Press.] VERSTRAELEN, EUGENE. 1960. An essay towards a historical description of Tagalog and Cebuano Bisaya. Philippine Studies 8.491-514. . 1961. Some further remarks about the L-feature. Philippine Studies 9.72-7. . 1962. Sound shifts in some dialects of the Philippines. Anthropos 57.826-56. VIRAY, FELIZBERTO B. 1939. The infixes la, li, lo and al in Philippine languages. Publications of the Institute of National Language, Bulletin no. 3. Manila, Bureau of Printing. . 1941. Prenasalization in Philippine languages. PSSR 13.119-47. WATERMAN, G. HENRY. 1960. Problems of syntax in the translation of the Scriptures in Philippine dialects. The Bible Translator 11.162-72. WELSH, DORIS VARNER, compl. 1950. Checklist of Philippine linguistics in the Newberry Library. Chicago, The Newberry Library. WOLFENDEN, ELMER. 1961. A restatement of Tagalog grammar. Manila, Summer Institute of Linguistics and Institute of National Language. WOLFENDEN, ELMER, JR., and RUFINO ALEJANDRO, eds. 1957. Intensive Tagalog conversation course. Manila, Summer Institute of Linguistics and Institute of National Language. WOLFENSON, LOUIS B. 1906. The infixes la, li, lo in Tagalog. JAOS 27.142-6. WOLFF, JOHN U. 1965. Cebuano Visayan syntax. Ph.D. dissertation, Yale University. . 1966. Beginning Cebuano, 1. New Haven and London, Yale University Press. . 1967. Beginning Cebuano, 2. New York and London, Yale University Press. . 1968. The history of the dialect of the Camotes Island, Philippines, and the spread of Cebuano. OL 6 (in press). WOLFF, JOHN and IDA. 1967. Beginning Waray-Waray. 4 vols. [Mimeo.]

MINOR LANGUAGES OF THE PHILIPPINES

HOWARD McKAUGHAN

0.

INTRODUCTION

This paper discusses trends in current linguistic studies of the minor languages of the Philippines. It summarizes what has been done, and indicates some of the research needs for the future. The languages treated are Austronesian, and exclude the major ones (Bikol, Cebuano, Hiligaynon, Ilocano, Pampanga, Pangasinan, Tagalog, and Waraywaray). 1.

RELATIONSHIPS

The exact number of the Austronesian languages spoken in the Philippines and their relationships are not known. As many as 135 dialects and subdialects were attributed to an estimated population of 1,500,000 near the end of the eighteenth century (Tangco 1951:13). Beyer (1917) lists 43 languages divided into 87 dialects for early in the current century. Conklin (1952) divides the various languages geographically into 75 main groups, 113 subgroups out of 32 of the main groups, with 43 main groups without subdivision. His total number of separate groups for the Philippines is thus 156. There are now some 25,000,000 people in the Philippines, and about 10% of these speak minor languages. A practical consideration in determining the number of languages in the Philippines is the yardstick to be used to distinguish language from dialect. The Philippines present an excellent field for the establishment of criteria for this determination. The two measures currently being used, mutual intelligibility and lexicostatistical percentages, are not yet sufficiently rigorous to be dependable. Although boundaries between languages are not sharp in the Philippines in many places, and although the languages are closely related to each other, one cannot move from the north to the south going from language to language in a chain of mutually intelligible dialects in the same way that can be done say in Germany (Hockett 1958:321-9). Yet in an area like the province of Bikol, there are thirty or more varieties of speech communities, all referred to as Bikol, some of which may very well be distinct languages. For the most part, Philippine languages have been classified by lexicostatistical measures. Such classifications have been suggested by Chretian for 21 languages in-

156

HOWARD MCKAUGHAN

eluding 14 minor ones (1961), Thomas and Healey for 37 including 31 minor languages (1962), Dyen with 84 Philippine lists, though representing a number of dialects rather than all distinct languages (1962, 1965, 1966), and Fox, Sibley, and Eggan for 17 northern languages, 13 minor (1965). C. F. and F. M. Voegelin (1964) and Grace (1964,1966) have summarized some of these findings and commented on them. Suffice it to say here that much still remains to be done. Although grouping geographically remains fairly accurate, more of the minor languages must be taken into consideration with better data. Basically the languages may be grouped into Northern (Ilocano-like), Central (Tagalog-like), and Southern (Bilaan, Tagabili, Tiruray). Currently, Elkins and Reid of the Summer Institute of Linguistics are gathering wordlists for a more careful survey of minor languages. Constantino of the University of the Philippines, through a project entitled 'Structural and comparative studies of Philippine languages', is gathering extensive wordlists and sentences for many of the languages. We hope to have a better understanding both of the number of languages extant, and of the language relationships from these researchers, before long. However, it should be indicated with much emphasis that there is a dire need for the application of the comparative method to these languages. Newell (1953) has given some sound correspondences for six languages, but little else has been done recently along these lines. Some of the comparative work done in the early 1900's like that of Conant (1908, 1910, 1912, 1916) needs to be extended to the minor languages. It is to be hoped that sound correspondences will be established with more and more of these languages considered, and that grammatical comparisons will also be made as time goes on. 2. DESCRIPTIVE STUDIES BY S.I.L.

Most of our recent descriptive linguistic knowledge of the minor languages of the Philippines comes from research published by members of the Summer Institute of Linguistics.1 This organization started its work in the Philippines in 1952. It currently has researchers in 42 minor languages (or dialects in some cases).2 Members have 1 There are a number of unpublished MSS on Philippine minor languages both by S.I.L. members and by students in U.S. and Philippine universities. I have not referred to such, nor to M. A. theses or Ph.D. dissertations unless they seem especially pertinent to current trends. 2 Summer Institute field workers are doing linguistic research on the following languages (location by province in parentheses): Agusan Manobo (Agusan), Agta (Cagayan), Ata Manobo (Davao), Atta (Cagayan), Balangao (Mountain Province), Batak (Palawan), Bilaan (Koronadal and Sarangani, Cotabato), Binukid (Bukidnon), Bontoc (Mountain Province), Cotabato Manobo, Dibabaon (Davao del Norte), Dumagat (Casiguran and Umirey, Quezon), Gaddang (Mountain Province), Ifugao (Amganad, Antipolo, and Batad, Mountain Province), Ilianen Manobo (Cotabato), Inibaloi (Benguet), Isneg (Kalinga-Apayao), Itneg (Abra), Ivatan (Batanes), Kalagan (Cotabato), Kalinga (KalingaApayao), Kankanay (Mountain Province), Mamanwa (Agusan), Mansaka (Davao del Norte), Maranao (Lanao del Sur), Samal (Sulu), Sambal (Zambales), Sangil and Sangir (Cotabato and Davao del Sur), Sarangani Manobo (Davao del Sur), Subanon (Sindangan, Zamboanga del Sur and Siocon Zamboanga del Norte), Tagabili (Cotabato), Tagbanwa (Aborlan and Kalamian, Palawan), Tausu (Sulu), Tigwa Manobo (Bukidnon), Western Bukidnon Manobo.

MINOR LANGUAGES OF THE PHILIPPINES

157

published some 87 linguistically oriented articles on minor languages Of these, 36 deal with some aspect of phonology covering 27 different languages. Among the more recent are Lee (1964a), J. and H. Miller (1964), de Vries and Roe (1966), Post (1966), Roe (1966), Shetler (1966), and Headland and Wolfenden (1967). These papers are data oriented for the most part and documented with many examples. Their framework is in the S.I.L.-Pike tradition, a taxonomic approach which identifies the significant sound units and their major alternants. Usually illustrations show obvious contrast, and environments causing variation are noted. Other articles by Institute members treat some phase of the grammar such as the pronoun system (Austin 1961, Lusted, Whittle, and Reid 1964, Weaver 1964), verb morphology (A. Maryott 1963, Lee 1964b, Shand 1964, S. Hussey 1965, Austin 1966), grammatical relations (Dean 1958, Reid 1964, Kerr 1965), phrase structure (Forsberg 1966, J. Headland 1966, J. Hussey 1966), clause structure (Forster 1964, Newell 1964, J. Miller 1964), and word (verb) classification (Abrams 1961, J. Miller 1964, Kerr 1965, Forster and Barnard 1967, and Ward and Forster 1967). A complete bibliography of the published works of the Philippine branch of this organization appears in Oceanic Linguistics (Wolfenden 1963, Roe 1967).

3.

PHONOLOGY

The articles by Institute members and others 3 show that segmental inventories of Philippine languages are quite similar. Consonants usually include the voiceless and voiced series of stops: p, t, k, q (glottal), b, d, g; the nasal series: m, n, rj; and the liquids, fricatives, and resonants: l/r, h, s, w, y. Vowels include the basic three — high front and back and low central (/, u, a), and often the pepet (a central relatively high vowel symbolized as /). Probably due to contact with Spanish, many languages now add the mid front and back vowels to their systems (e and o). Bilaan and Tagabili of southern Mindanao add to these six vowels a seventh, which is a low back rounded vowel (J. and G. Dean 1955, Maryott, Forsberg, and Healey 1959). Jo Shetler (1966) reports seven vowel phonemes for Balangao spoken in northern Luzon in the Bontoc subprovince. These vowels include the front high and mid (i, e); the back (u, o); and the central high mid and low (/, a, a). She cites the following contrasts for the latter: qd'nan 'to increase', qVnim 'six', qanaq 'child of', qa'nad 'earlier', qi'nib 'door', qamah 'to bathe'. A seven vowel system for northern languages is unusual. Most systems in the Philippines have either a four or five vowel inventory. Syllable patterns throughout the area are relatively simple — either CV or CVC almost everywhere. Perhaps the relative ease with which a western investigator imitates words or phrases in Philippine languages has led to somewhat incomplete phonological studies. At any a

See, for example, Yamada's phonology statement for Itbayatan (1965).

158

HOWARD MCKAUGHAN

rate very little has been published on the prosodic features of Philippine minor languages. An important article in this respect was published by Shetler and Hopkins, entitled "The obscuring of word accent in Balangao" (1964). The authors indicate that Balangao accent is characterized by vowel length, a slight rise in pitch, and intensity 'which is actualized as the gentle peak of a subtly crescendoing decrescendoing sequence' (p. 102). Words in Balangao carry accent as a contrastive feature on either the ultimate or penultimate syllable. Morphophonemic shifts of accent in words depend upon the addition of suffixes. When words are combined into larger units of phrases and sentences, intonation contours tie them together. The pitches of contrastive contours either highlight or obscure the pitch component of word accent. For example, pitch rise as a feature of accent is obscured on syllables adjacent to high pitch in an intonation contour. Length as a feature of accent does not occur on a closed syllable. Intensity as a feature of accent is never diagnostic for Balangao, since when pitch rise and vowel length are obscured either by syllable patterns or by intonation contours, intensity is not aurally distinguishable. When both pitch and vowel length components of accent are obscured, then contrastive accent is obscured on the word level. Accent may not be determined in these environments and can only be determined by changing the environment. This article gives insight to the problems of the prosodic features of the Philippine languages. Other languages need similar or even more extensive treatment. The eifect of intonation on stress, pitch, and length on the word level is no doubt important in most Philippine languages. Investigators should be encouraged, then, to give us a more complete picture of the prosodic features on all levels of analysis — the word, phrase, clause, and sentence.

4.

GRAMMAR

Grammatical studies of minor languages in the Philippines often concentrate on verb morphology and the relation of the verb to other parts of the sentence.4 The most prominent grammatical feature, at least to the western observer, is that which indicates grammatical relations between the verb and the topic of the sentence. To mark these relations, the verb contains 'case-marking' morphemes indicating that the topic is the actor, goal, indirect referent, or instrument of the action denoted. Early writers referred to these relations as active and passive, with three kinds of passive involved (Bloomfield 1917:154, Blake 1925:38ff.). Summer Institute researchers prefer the term focus to denote these relations, though voice has been used and is still preferred by this writer (McKaughan 1958, 1962). 4

Paz (1965) has published an interesting article on -ad in Isinai. This is an almost unique syntactic device in Philippine languages co-occurring with di and si to mark subject in some cases and complement in others.

MINOR LANGUAGES OF THE PHILIPPINES

159

Kerr (1965) has given an excellent treatment of the use of the verbal affixes by discussing the case-marking and classifying function of Cotabato Manobo voice affixes. He maintains, with good reason, that the verbal affixes involved mark case relations just as do certain particles which introduce substantive phrases (or different sets of pronouns replacing these phrases). Thus case-like relations are marked by particles, and by verbal affixes, the latter only related to the topic of the sentence. Topic, in turn, is designated by particular particles, or by a particular set of pronouns. Perhaps the most striking thing to the westerner is the fact that the passive-like constructions are much more favored in Philippine languages than the passive is in English. Descriptions to date of the grammatical relations obtaining between the predicate and topic (voice/focus) are quite representative for Philippine minor languages. McKaughan (1958), Twedell (1958), Healey (1960), Pike (1963), Reid (1965,1966) and Elkins (1967) have given fairly extensive treatments of Maranao, Iraya, Yogad, Bilaan, Bontoc, Ivatan, and Western Bukidnon Manobo. The voice/focus system for all Austronesian languages in the Philippines is carried by verbal prefixes, infixes, and suffixes. The four basic voices/focuses have been called subjective, objective, referential, and instrumental (some languages are believed to have more than the basic four). The first is 'active', the last three 'passive'. Native speakers' reaction is that the first takes an indefinite object, if any, and that the others always have a definite specific object in view. The objective has only the direct object of the verb in view while the other two, referential and instrumental, have both a specific direct object and an indirect object (again specific), which is an accessory to the action either in that it benefits from it, is used to bring it about, or in some way is indirectly involved. If the accessory to the action is in grammatical focus, then either the referential or instrumental voice is used. The subjective voice is marked by an infix {-um-), usually occurring after the first consonant of the word base (many stems having as initial consonant the glottal stop which traditionally has not been written), and by a number of prefixes (ma-, maka-, mag-, etc.). Differences of meaning between the infix and the prefixes may be in part due to the meaning of the word base and in part to the kinds of action desired (aptative-abilitative, distributive, etc.). The objective is marked by the suffix -in or -en (zeroed out with the past tense). It should be noted here that though most languages seem to require some noun phrase related to the verb predicate, most give an option as to what substantive may occur. In other words the topic of the sentence, whether related to the verb as the subject, object, referent, or instrument, need not be overtly made manifest in a given sentence. But every verb has an 'understood' relation to a topic whether manifested or not. The -en suffix 'understands' a definite direct object as topic. Thus when manifested, we may have 'It is the dog that he hit', and when not manifested 'It is something understood by context that he hit'. The referential is usually marked by suffix -an, probably the most universal in phonological shape and having the least variants of any of the voice/focus affixes from

160

HOWARD MCKAUGHAN

language to language. In this case, the topic benefits from the action, or is the place where an action occurs, and that action is also directed toward an object. In this relation there is a dual 'specificity' — 'It is for the man (topic) that I hit something'. The one for whom the action is performed, and the goal of the action need not be overtly manifested, but the relations are there. We could say 'He hit something for someone' where only 'he' and 'hit' are morphemically present, the 'hit' occurring with -an to indicate that the one benefitted is the topic if named, and that a direct object is also involved, named or not. Instrumental voice/focus is usually indicated by the prefix /-. Here, that which is used to bring about the action is the topic of the sentence — 'It is with the knife (topic) that he cut the paper.' Again, substantives may or may not be present, but two, the instrument and the object, are always 'understood'. The verbal predicate in all of these kinds of sentences is obligatory. One substantive phrase (or pronoun replacing it) is usually present (obligatorily in many cases). The minimal meaning of an instrumental voice action then is always 'something is used to cut something'. Nominals are related to the predicate in these same dramatis personae roles by 'case-marking' particles which introduce the substantive phrases. Some languages such as Maranao have a rather full set of particles for this purpose — one to mark the actor when not topic (o), one to mark the object when not topic (sa), one to mark referent or instrument when not topic (ko), and one to mark topic (so), with the case relations for the latter indicated by the verbal affixes. Other languages (usually northern, though Tagabili of the south also shows this characteristic) have the same particle introducing substantive phrases whatever the role, with differences indicated by position or some other means. All of the languages have at least three different sets of pronouns to mark differences in dramatis personae roles. Usually one set marks the topic, another the actor when not the topic, and a third set is used for the other roles when not topic. Some languages affix pronominal morphemes to the verb while others do not, though the affix morphemes for this purpose mark only topic, and non-topic actor. Verb affixes also occur to indicate tense (past, progressive, future, general, or present), aspect or kind of action (aptative, distributive, causative, etc.), and mode (obligatory or indicative). The causative has been called voice by some (Forster 1964) with the word focus reserved for topic-verb relations rather than voice. The causative system is operative with all voice/focus, tense, aspect, and mode affixes. New insights into this area are being made, and are important (Ward and Forster 1967). Relations between constructions not mentioned above are marked by particles often referred to as ligatures. These may indicate apposition or coordination and may be on the phrase level (within phrases) or between clauses. Other grammatical features are less prominent than those mentioned, and are left to individual language descriptions for discussion. Coupled with the linguistic research on major languages, we now have an excellent data oriented understanding of the grammatical relations existing and of

MINOR LANGUAGES OF THE PHILIPPINES

161

the meanings of affixes and particles. Given a page of text material in a Philippine language described to this degree, we can translate quite effectively. The real problem, however, comes when we try to generate text material rather than analyze or describe material already produced. This becomes all too apparent when a learner of one of these languages tries to generate his own utterances. He has a difficult time from extant descriptions knowing what he can say and what he cannot. He has a difficult time knowing what affixes word bases can take unless the particular word base has been described with its affixes. This problem is compounded because, characteristic of all the languages, the same word bases are freely used in predicate or substantive functions. However, an excellent beginning has been made in the right direction for various of the minor Philippine languages. This beginning is coming through attempts to classify verbs (usually defined as the word base with its affixes). Abrams (1961) uses potential clause expansions and clause transformations as criteria supporting a primary criterion of clause focus when a base occurs unaffixed in the predicate to establish Bilaan verb word base classes. Miller (1964) classifies Mamanwa verb stems by their occurrence or nonoccurrence with specific voice affixes. Barnard (1965) classifies Dibabaon verbs on the basis of obligatory situational slots. One class requires an obligatory actor, another requires an obligatory actor and site, etc. Reid (1966) classifies for Ivatan with criteria of potential clause expansion of nuclear grammatical slots, and potential clause transformation. Abrams thus cites three classes for Bilaan, Miller eight classes for Mamanwa, Barnard locates seven for Dibabaon, and Reid has twelve for Ivatan. Ward and Forster write on this subject (1967:30): 'In highlighting the distinctive feature of verb focus, and making extensive use of it in classifying verbal constructions, there has been a tendency to overlook the importance of verbal classes, particularly in identifying clause types.' Thus establishing verb classes becomes all important in description. We may ask, which verbs prefer non-object or general object relations with grammatical focus on the actor-action relation? Which verbs prefer dual 'specificity' and require say the instrumental or referential voice/focus? Can the verbs be grouped by semantic criteria? It is this writer's strong hope that more native speakers will become linguistically sophisticated (because this is a prime necessity) so that more can be indicated in the realm of generative rules for these languages. Here we are faced with the purpose of grammars brought sharply to our attention by the transformationalists. Grammars must tell us how to generate. Even for practical purposes, our descriptions of the minor languages of the Philippines must move to greater power in this direction. We need to put into the hands of learners of these languages the key to what is possible. Perhaps Philippine verbs have principal parts as do the Latin verbs — particular morphemes which will give clues to a verb's full paradigm. The criterion of potential occurrence of voice/focus affixes marking case relations with the topic of a sentence does not seem to be sufficient for this purpose, though certainly a move in the right

162

HOWARD MCKAUGHAN

direction and part of the answer. The possibility of adding semantic co-occurrence potential and transformational possibilities to the focus/voice potential has also helped toward solving this problem. This writer believes that further light will be shed on the problem as we move into the semantic area. An immediate objective is to design practical dictionaries that will tell us what the generational possibilities are for a given stem, at least for the voice/focus system. Most descriptions of minor Philippine languages to date have been couched in taxonomic descriptive terms. A number of them have used the tagmemic model with various modifications. The importance of transformations in these descriptions has been recognized (Reid 1966, Elkins 1967), but the transformational-generative model has yet to be used for a full description of one of the minor languages. Maryott (1963) and Austin (1966) have attempted studies of focus/voice and its function in Sangir and Ata Manobo through the stratificational model. Every encouragement should be given to the application of this model in linguistic research, and to the continued application of the taxonomic approach with attention to generative detail. The application of the generative-transformational model to minor languages is a must, even as it has been applied to the major languages in a number of instances. Summer Institute of Linguistics personnel under the direction of Kenneth Pike and Robert E. Longacre are currently studying structural levels above the sentence. Roe recently wrote: 'We are discovering what well-formed paragraphs are in terms of the types of sentence margins (e.g. elements introduced by because, when, so that) which occur to link the tagmemes of the paragraph. Different kinds of links occur in different types of paragraphs, which in turn are determined largely by the type of discourse involved.' We look forward to the publication of results from this project. Manuscripts that have been completed so far in the Longacre workshops between September 1967 and January 1968 include sentence level analyses in 11 minor dialects (Agta, Atta, Koronadal Bilaan, Binukid, Ata Manobo, Mansaka, Maranao, Bolotan Sambal, Sangir, Tagabili, Tausug); paragraph level analyses in six languages (Atta, Koronadal Bilaan, Ata Manobo, Mansaka, Maranao, Botolan Sambal); and discourse level analyses in three languages (Atta, Maranao, and Bolotan Sambal).

5.

LEXICON

Another important area of linguistic research for minor languages of the Philippines is that of lexicography. Conklin (1953, 1955, 1962, 1967a, 1967b), Frake (1961), Yamada (1966), McKaughan and Macaraya (1967), and Elkins (1968) have published lexical articles and monographs. 5 Frake's diagnosis of disease among the Subanon 6

There are a number of early wordlists, some grammatical notes, and some general statements about various of the minor languages of the Philippines (see bibliography in Tangco 1951). This article is concerned with the last 10 to 15 years only. No attempt has been made to refer to early works or vocabularies.

MINOR LANGUAGES OF THE PHILIPPINES

163

and Conklin's lexicographical treatment of folk taxonomies are important steps in a very needy direction. They suggest semantic systems (more or less closed) and extend the treatment of word study beyond mere translation equivalents. Much more along this line is needed. Conklin's earlier work on Hanunoo-English (1953) is an excellent contribution. Elkins gives a careful determination of semantic values of individual words in his Manobo dictionary. McKaughan and Macaraya have given an extensive glossary (about 20,000 words) of Maranao which, though a good step, needs extension now in depth. Reid and Elkins are currently extending dictionary research for Bontoc and Western Bukidnon Manobo by the use of a computer generative elicitation technique suggested by Carroll (1966). The possibility of a much greater degree of completeness for the number of word bases in less research time through this technique is an important advance. The technique is to program a computer to turn out strings of phonemes up to a certain predetermined number, programming in any sequence restrictions so that theoretically the printout represents words in a given language. Both Elkins and Reid indicate that this method is eliciting far more word bases than they had previously been able to obtain in 8 to 12 years of field study. Word and morph concordances have been made from texts in 30 minor languages on the IBM 1410 computer at the University of Oklahoma by the Linguistic Information Retrieval Project of the Summer Institute of Linguistics and the University of Oklahoma Research Institute. These concordances are being used for the most part for analytical purposes rather than dictionary results, although the possibility of the latter is important. Perhaps some programming for ready-made computer printouts of dictionaries can be made from texts if the English equivalents can be added to the input. The computer should be used increasingly for dictionary construction. Carroll's printout method augers well for completeness. Valuable time may be saved by having materials sooner. This writer found almost by accident an important factor in determining which words to illustrate in the Maranao dictionary (McKaughan and Macaraya 1967). A computer printout of the English to Maranao was programmed from the Maranao-English input. Both McKaughan and Macaraya were surprised to find the extent of semantic overlap in words as indicated by English glosses. For example, 53 Maranao words were found to have been glossed by the English word 'stop' (with other words also of course). Maranao words glossed by the same English word need illustration to show differences of semantic content. Only when the computer gave the English-Maranao printout did the overlap become readily (and strikingly) apparent. Without the computer it would have been many months before the English-Maranao section of the dictionary would have been available. The relation of lexicographical research to the domain of semantics is important. Perhaps an extension of dictionary research will give a breakthrough in this area. Some of the investigations carried on by Summer Institute of Linguistics personnel related to translation techniques should give insight here. Professor Conklin of Yale is

164

HOWARD MCKAUGHAN

currently carrying out extensive research with the Ifugao, collecting linguistic and ethnographic data. He has been particularly interested in lexical problems. Professor Frake of Stanford has been engaged in language and culture research in the southern Philippines. He is doing linguistic analysis that includes semantics and sociolinguistics for Yakan, a member of the Samal subgroup spoken by Moslems of Basilan Island. Recently he has extended his research to include the Zambuangeno, a Spanish based creole in Zamboanga City.6 Research projects such as these coupled with those of Summer Institute field workers should result in important observations of a semantic nature. Perhaps subcategorizational rules and co-occurrence potential as well as restrictions will be worked out, and thus give us further insight into how to write into our descriptions (or our lexicons) the needed generative rules.

6.

CONCLUSION

We are fortunate in the amount of descriptive material we have for Philippine minor languages. Progress in the application of recent theoretical advances is excellent. If we can add to the descriptive materials more of the needed generative power, move further into comparative study, and extend our semantic insight, our understanding of this branch of the Austronesian languages will be unexcelled.

REFERENCES ABRAMS, NORMAN.

1961.

AUSTIN, VIRGINIA MOREY.

Word bases in Bilaan. Lingua 10.391-402. 1961. Some particles and pronouns in Batak. PhilJSci

90.263-70.

. 1966. Attention, emphasis and focus in Ata Manobo. (Hartford Studies in Linguistics 20.) Hartford, Connecticut, Hartford Theological Seminary. BARNARD, M Y R A L. 1965. Dibabaon transient verbs. Unpublished MS. Manila. S.I.L. BEYER, H . OTLEY. 1917. Population of the Philippine Islands in 1916. Manila, Philippine Education Co. BLAKE, FRANK R. 1925. A grammar of the Tagalog language. New Haven, American Oriental Society. BLOOMFIELD, LEONARD. 1917. Tagalog texts with grammatical analysis. Urbana, University of Illinois.

• Chabacano (Zambuangeno) in the south and Caviteno and Ternateno in the north need further study. Whinnom (1956) and McKaughan (1954) have made a few suggestions. Currently Michael L. Forman is writing a dissertation for Cornell on Chabacano. Frake's project should give us much formalized linguistic information.

MINOR LANGUAGES OF THE PHILIPPINES

165

CARROLL, VERN. 1966. Generative elicitation techniques in Polynesian lexicography. O L 5/2.59-70.

CHRETIEN, C. DOUGLAS. 1961. A classification of twenty-one Philippine languages. PhilJSci 91.485-506. CONANT, CARLOS EVERETT. 1908. ' F ' and 'V' in Philippine languages. Bureau of

Science Division of Ethnology Publications no. 5. vol. 5, part 3.135-41. . 1910. The RGH law in Philippine languages. JAOS 31.70-85. . 1912. The pepet law in the Philippines. Anthropos 7.920-47. . 1916. Indonesian T in Philippine languages. JAOS 36.181-96. CONKLIN, HAROLD C. 1952. Outline gazeteer of native Philippine ethnic and linguistic groups. (Southeast Asia Studies.) New Haven, Yale University. . 1953. Hanunoo-English vocabulary. Berkeley, University of California Press. . 1955. Hanunoo color categories. SJA 11.339-44. . 1962. Lexicographical treatment of folk taxonomies. IJAL 28(2) part 4.11941. . 1967a. Ifugao ethnobotany 1905-1965: the 1911 Beyer-Merrill Report in perspective. Economic Botany 21.243-72. . 1967b. Some aspects of ethnographic research in Ifugao. Transactions of the New York Academy of Sciences, Series II, 30.99-121. DEAN, JAMES. 1958. Some principal grammatical relations of Bilaan. OLM 3, ed. by A. Capell and S. Wurm, 59-64. Australia, University of Sydney. , and GLADYS. 1955. The phonemes of Bilaan. PhilJSci 84.311-22. DYEN, ISIDORE. 1962. The lexicostatistical classification of the Malayopolynesian languages. Lg 38.38-46.

. 1965. A lexicostatistical classification of the Austronesian languages. IJAL supplement vol. 31, no. 1. Baltimore, Waverly Press. . 1966. Comment. OL 5/1.32-49. ELKINS, RICHARD E. 1967. Major grammatical patterns of Western Bukidnon Manobo. Ph.D. dissertation, University of Hawaii. . 1968. Western Bukidnon Manobo dictionary. University of Hawaii Press. FORSBERG, VIVIAN. 1966. Phrases in Tagabili. LCC A, 8, ed. by S. A. Wurm, 21-32.

Canberra, The Australian National University. FORSTER, JANETTE. 1964. Dual structure of Dibabaon verbal clauses. OL 3.26-48. , and MYRA L. BARNARD. 1967. A classification of Dibabaon active verbs. Lingua (in press). F o x , ROBERT B., WILLIS E. SIBLEY, a n d FRED EGGAN. 1965. A preliminary glotto-

chronology for Northern Luzon. Asian Studies 3/1.103-13. FRAKE, CHARLES O. 1961. The diagnosis of disease among the Subanon of Mindan a o . A m A 63/1.113-32.

GRACE, GEORGE. 1964. Movement of the Malayo-Polynesians: the linguistic evidence. CAnthr 5,361-8.

166

HOWARD MCKAUGHAN

. 1966. Austronesian lexicostatistical classification: a review article. O L 5/1. 13-31. HEADLAND, JANET. 1966. Case-marking particles in Casiguran Dumagat. PJL 4/1-2.58-9. HEADLAND, THOMAS N . , a n d ELMER P . WOLFENDEN.

1967. T h e v o w e l s o f C a s i g u r a n

Dumagat. Studies in Philippine anthropology (in honor of H. Otley Beyer), ed. by Mario D. Zamora, 592-6. Quezon City, Alemar-Phoenix Publishing House. HEALEY, PHYLLIS M. 1960. An Agta grammar. Manila, The Institute of National Language and the Summer Institute of Linguistics Philippine Branch. HOCKETT, CHARLES F. 1958. A course in modern linguistics. New York, The Macmillan Company. HUSSEY, JEAN. 1966. Noun phrase markers in Aborlan Tagbanwa. LCC A, 8, ed. by S.A.Wurm, 33-8. Canberra, The Australian National University. HUSSEY, STEWART C. 1965. Aborlan Tagbanwa verbal system and related topics. (Hartford Studies in Linguistics 19.) Hartford Connecticut, Hartford Theological Seminary. KERR, HARLAND B. 1965. The case-marking and classifying function of Cotabato Manobo voice affixes. OL 4.15-47. LEE, ERNEST W. 1964a. Magiundanao /l/. G L 6.24-6. . 1964b. Nonfocus verbs in Maguindanao. O L 3.49-57. LUSTED, RUTH, CLAUDIA WHITTLE, a n d LAWRENCE A . REID. 1964. T h e u s e o f m a t r i x

technique in an analysis of Atta personal pronouns. OL 3.138-60. MARYOTT, ALICE. 1963. T h e nuclear predication in Sangir. PhilJSci 92.111-20. , VIVIAN FORSBERG, a n d ALAN HEALEY. 1959. T h e p h o n e m e s of Tagabili. Phil-

JSci 88.225-43.

MCKAUGHAN, HOWARD P. 1954. Notes on Chabacano Grammar. The University of Manila Journal of East Asiatic Studies 3/2.205-26. . 1958. The inflection and syntax of Maranao verbs. Manila, The Institute of National Language. . 1962. Overt relation markers in Maranao. Lg 38.47-51. , and BATUA A. MACARAYA. 1967. Maranao Dictionary. Honolulu, University of Hawaii Press. MILLER, JEANNE. 1964. The role of verb stems in the Mamanwa kernel verbal clauses. OL 3.87-100. , and HELEN. 1964. Mamanwa phonemes and orthography. PSSR (in press). NEWELL, LEONARD E. 1953. Some sound correspondences in six Philippine languages. Folklore Studies 12.105-7. . 1964. Independent clause types of Batad Ifugao. OL 3.171-99. PAZ, CONSUELO S. 1965. -Ad in Isinai. Asian Studies 94.373-93. PIKE, KENNETH L. 1963. A syntactic paradigm. Lg 30.216-30.

POST, URSULA. 1966. The phonology of Tiruray. PhilJSci 95/2 (in press).

MINOR LANGUAGES OF THE PHILIPPINES

167

REID, LAWRENCE A. 1964. A matrix analysis of Bontoc case-marking particles. O L 3.116-37.

. 1965. A formal analysis of the clause structure of Central Bontoc. Unpublished M.A. thesis, University of Hawaii. . 1966. An Ivatan syntax. OL Special Publication no. 2. Honolulu, University of Hawaii. . 1967. On redefining transitivity for Philippine languages. PJL (in press). ROE, G. RICHARD. 1966. Isneg spelling. P J L 4/1-2.72-4.

. 1967. Current research of the Summer Institute of Linguistics in the Philippines. OL 6/1.52-61. SHAND, JEAN. 1964. Categories and markers of tense, focus and mode in Ilianen Manobo. OL 3.58-68. SHETLER, Jo. 1966. Balangao phonemes. LCC A, 8, ed. by S.A.Wurm, 1-7. Canberra, Australian National University. , and ANNE FETZER HOPKINS. 1964. The obscuring of word accent in Balangoa.

OL 3.101-9. TANGCO, MARCELO. 1951. The Christian peoples of the Philippines. (National and Applied Science Bulletin 11, No. 1.) University of the Philippines, Quezon City, Philippines. THOMAS, DAVID, a n d ALAN HEALEY. 1962. Some Philippine language subgroupings:

a lexicostatistical study. AnL 4/9.21-33. TWEDDELL, COLLIN ELLIDGE. 1958. The Iraya (Mangyan) language of M i n d o r o ,

Philippines: phonology and morphology. Washington. VRIES, VIRGINIA H . DE, a n d G . RICHARD ROE.

Ph.D. dissertation, University of 1966.

Semivowels in the

Cuyono

alphabet. PhilJSci 95.267-74. VOEGELIN, C.F. and F . M . 1964. Languages of the world: Indo-Pacific fascicle one. A n L 6/4.88-100.

WARD, ROBERT G . , a n d JANNETTE FORSTER.

1967. V e r b stem classes in M a r a n a o

transitive clauses. AnL 9/6.30-42. WEAVER, DANIEL and MARILOU. 1964. R a n k i n g of personal p r o n o u n s in Agusan M a n o b o . O L 3.161-70.

WHINNOM, KEITH. 1956. Spanish contact vernaculars in the Philippine Islands. Hong Kong University Press. WOLFENDEN, ELMER. 1963. Report on the work of the Summer Institute of Linguistics in the Philippines. OL 2.1-15. YAMADA, YUKIHIRO. 1965. Phonology of Itbayatan. PhilJSci 94.373-93. . 1966. A preliminary Itbayatan vocabulary. Institute of Asian Studies. University of the Philippines. (Mimeo.)

THE AUSTRONESIAN LANGUAGES OF FORMOSA*

ISIDORE DYEN

THE LANGUAGES

The Austronesian languages of Taiwan have been receiving increased attention. There has not been a complete survey of the Formosan languages though an approach to one is found in Naoyoshi Ogawa and Erin Asai's The myths and traditions of the Formosan native tribes (1935: hereafter referred to as OA). Attempts to classify the languages and dialects are found in A. Nikigawa's "A classification of the Formosan languages" (1953: hereafter referred to as Nik.), Cestmir Loukotka and Peter A. Lanyon-Orgill's "A revised classification of the Formosan languages" (1958: hereafter referred to as LL) and P. A. Lanyon-Orgill's translation of a small part of OA under the title "Comparative vocabulary of Formosan languages" (1961: hereafter referred to as LO). One of the typical problems in dealing with the Formosan languages is the variant names for the same language or dialect. The convention is followed below of selecting one form of a name as standard and enclosing in parentheses the other forms that have been applied to the same set of speakers. The following is a listing of the languages. In the first group are the extant languages with more than 1000 speakers according to the estimated number given in the census of 1960 as listed in Chen Kang Chai's Taiwan Aborigines (1967:39). Ami (Amis, Pangtsah): 87,345 speakers on the eastern coastal plain from Sincheng to south of Taitung (Pinan). Its reported dialect names are Baran (Falanao, Maran), Kibi (Kiwit), Taparon (Tavalon), Tauran (Nataoran, Ami, Amis), and Sabari. Atayal (Tayal, Ataiyal, Taiyal): combined with Seediq, 49,406 speakers, with about 36,000 for Atayal according to Chen Chi-lu (1958) in the mountains of the north (south of the Ketagalan area). The reported dialect names for Atayal are, aside from the language name, Taihyoo (Mekbengetseq, Bonotsek), Takonan, Yukan, Taroko, Sqolyeq. Seediq (Seedek, Seedik, Sedek, Sedeq, Shedekka, Sazek) estimated at about 14,000 (in terms of the census combined figure with Atayal taken in conjunction with * This paper represents partial results obtained with the support of the National Science Foundation by Grant No. GS-1468.

FORMOSAN

LANGUAGES

170

ISIDORE DYEN

Chen's estimate for Atayal above) in the northern mountains south of the Atayal area. The reported dialect names other than the language name are Wu (Hogo, Gungu, Gunu), Taroko, Iboho (Ilutso, Tlukko), Buhwan, Paran, and Tangao (a thoroughly sinised dialect). Bunun (Vunun, Bunum, Vonum): 21,440 speakers over a wide area on the east and south sides of the central plain south of the Seediq area. The reported dialect names, aside from the language name, are Tamaroan (Tamado'wan, Tamaroau, Northern Bunun), Katoguran (Katongulan, Ketanganau, Central Bunun), Ibaho (Ivaxo, Southern Bunun), Sivukun (Sibukon), Kagi (? extinct), and Tsuihwan (Chibora: ? extinct). Paiwan: 36,775 speakers in the mountains in the southernmost part of the island. The reported dialect names other than the language name are Tokubun (Tokovul, Tokuvul), Shimopaiwan (Ka-paiwan-an), Kapiyan (Kaviangan, Kipiyan), Raisha (Tsala'avus), Kunanau (Kulalau), Naibun (Tsa'ovo'ovol, Tsa'ovo'ovul), Kachirai (Tsoa-qatsilai), Rikiriki (Raokrik, Laliklik, Lilisha), Tachaban (Patsavan, Patsaval), Tamari (Tsaovali). Puyuma (Pyuma, Pilam, Pelam): 8,687 speakers along the east coast south of Taitung (Pinan: i.e. south of Ami) and inland. The reported dialect names other than the language name are Chipon (Kata-tipol, Pyuma), Hinan, Pinan (Hinan, Pilam), Sanwhai, Tsarisen (Tsalisen), Saprek, Katsausan, and Kale-whan. Rukai: 5,871 speakers in the south central mountains, west of Puyuma. LO divides Rukai into two groups without commitment as to their connection: Rukai (Tsarisen, Tsalisen) with the reported dialect names Dainan (Taromak, Taromaki), Taramakau (Talamakau, Takamakau); and Rukai (Dukai) with the reported dialect names Maga (Tordukana), Tona (Kongadavanu), and Mantauran (Upunuhu, Opunoho). The relation between LO's division and the population figures is not known. Saisiyat (Saisiat, Saisiet, Saisett, Saiset, Saisirat): 3,394 speakers on the west side of the island in the mountains west of the Atayal area. The reported dialect names other than the language name are Taiai (Sairakis), Buiok, Arikun. LL also reports Pazzehe (Pazehe) and Taokas, but these are provisionally regarded as different languages: see Pazeh, Taokas below. Tsou (Tsoo, Tsu'u, Tsu-u, Tsuou, Northern Tsou): 3,223 speakers in the west central mountains southeast of Kagi around Mt.Ali. The reported dialect names other than the language name are Luhtu (Dufutu, Namakaban, Namahabana, Namakabau, Ruftu), Tfuea (Tufuja, Tfuya, Chibora, Tsuihwan), Arisan (Alisan), Tapangu, Iimutsu (extinct). Yami: 1,957 speakers on Botel Tobago Island (also called Lanyu Island). With it is associated the dialect name Imurut (Imulud).

Nik. and LL list Tsarisen (Tsalisen; see Rukai above) as an additional language; the

THE AUSTRONESIAN LANGUAGES OF FORMOSA

171

latter places it in the interior of the south, south of Takao (according to Nikigawa) and west of the Paiwan area. This language is reported by many writers, but its area is clearly included by OA as the western part of their Paiwan area. Furthermore LO distributes the dialects assigned by LL to Tsarisen among other languages as follows: Tsarisen, Dainan, and Takamakau to Rukai (Tsarisen); Tona to Rukai (Dukai); Tipun (Chipun) to Puyuma (?); Kapaiwanau (if it is LO's Ka-paiwan-an); Lilisha (Raokrik), Tamari (Tsaovali), and Kapiangan (if it is LO's Kaviangan) to Paiwan. The languages Saaroa (Saarua, La'alua, La'aroa, Sisyaban) and Kanakanabu (Kanabu, Kanakanavu) are together sometimes called Southern Tsou against Tsou proper (then called Northern Tsou). Their territory lies contiguous and just to the south of the Tsou area. Separate population figures for them are not available. For Saaroa there is the reported dialect name Haisen (Paitsiana, La'aluwa) and for Kanakanabu there is Nagisaru (Nagisaran). There are three languages which in 1930 were used at home by speakers who elsewhere used Taiwanese. These are Kavalan (Kuwarawan, Kiwarawa, Kuvarawan, with the additional dialect name Kareovan) in the northeast, Pazeh (Pazzehe, Pazehe) near the west coast just north of 24° N. lat., and Thao (Sao, Shao, Sau, Chuihwan) with about 160 speakers near Lake Suisha in the center of the island. The following languages are no longer spoken or are found spoken only by bilinguals: Ketagalan, formerly across the northern end of the island; reported as occurring in one village in the Kavalan area. Taokas, northwest coast. Papora, northwest coast south of Taokas. Babuza (Babusa, Favorlang), west coast in middle of the island around 24° N. lat.; associated with it is the dialect name Poavosa reported still spoken in 1908 in a village south of the Papora area. Hoanya, west coast south of Babuza; associated dialect names are Loaand Arikun (and perhaps Taokas, listed above). Siraya (Sideia, Sideis), southwest coast; associated dialect names are Sideia, Siraya (still spoken in 1908 in a village in the Ami area), Makatao, Tamsui (extinct), and Sinkan (Sakam: extinct), and perhaps also Kongana (extinct) and Baksa (extinct).

RECENT HISTORY

The first visit to Formosa by westerners was in 1498 by Portuguese who gave it its name. The population at that time was chiefly Austronesian; there are reports of 'little black men', presumably Negritos, distributed widely on the west side of the Central Mountains, who disappeared about 100 years ago. There was a large migration of Chinese from Fukien in 1626 and another of thousands of civilians and servicemen after the end of World War II. The consequence of these migrations has

172

ISIDORE DYEN

been the gradual restriction of the Austronesians to the mountains of the eastern half of the island and to the eastern coastal plain. Despite occupation by the Dutch and Spanish in the early seventeenth century and by the Japanese from 1895 to 1945 there are practically no immigrants or descendants of immigrants of these nationalities living as civilians on the island. The numerous Chinese are the chief and practically only non-Austronesian group on the island. They total about 12,000,000 and are distributed mainly over the western plain in the south and in the eastern plain areas. The Dutch period saw the compilation of the famous dictionary of the Favorlang (now Babuza) language in 1650 by Gilbertus Happart, the Woordboek der Favorlangsche Taal, republished by W. R. van Hoevell (1842) and translated by W. H. Medhurst (1840) under the title Dictionary of the Favorlang dialect of the Formosan language. C.J. van der Vlis (1842a) published his Vocabularium Formosanum (Woordenlijst en gespreken in't Sideisch-Formosaansch en Hollandsch) based on Daniel Gravius' (1662) Formulier des Christendoms met de verklaringen van dien, inde Sideis-Formosaansche tale. The latter was republished in facsimile by Erin Asai (1939). Vlis (1842b) also published a Formosaansche woordenlijst volges een Utrechtsch handschrift based on Daniel Gravius' (1661) Het hey lige euangelium Matthei en Johannis ... overgeset inde Formosaansche tale, voor de inwoonders van Soulang, Mattau, Sinckan, Bacloan, Tavokan, en Tevorang. Vlis' (Sinkan) vocabulary was republished in 1933 with two other vocabularies in the Memoirs of the Faculty of Literature and Politics of the Taihoku Imperial University. In the modern period the most important contributions have been made in Japanese, principally by N. Ogawa who published, aside from the joint work with Asai mentioned above, a Japanese-Paiwan vocabulary (Paiwan Goshû, 1931), a Japanese-Atayal vocabulary (Atayaru Goshû, 1932; selections translated by Ferrell 1967), and a Japanese-Ami vocabulary (Ami Goshû, 1934) as well as a number of articles on features of these languages. CLASSIFICATION

The first identification of the tribal languages of Formosa as Austronesian was made by J. H. Klaproth in his "Sur la langue des indigènes de l'île de Formose" (1822) as far as C. Douglas Chrétien could discover when preparing his article "Oceanic material in Bu-hwan, a hill dialect of Formosa" (1951). H.C. von der Gabelentz (1859) discusses the relation between the Formosan languages and other members of the Austronesian family. Others who have discussed the same subject are T. L. Bullock in his "Formosan dialects and their connection with the Malay" (1874-75) and Terrien de Lacouperie in his "Formosa notes on Mss., races, and languages" (1887:477ff.). It is a commentary on how recent and relatively superficial our knowledge of the Formosan languages is that Chrétien in his article correctly identifies Buhwan as Seediq on the basis of information supplied by Erin Asai — who later identified the material as from

THE AUSTRONESIAN LANGUAGES OP FORMOSA

173

the Paran dialect (Asai 1953:1) — but nevertheless states (Chrétien 1951:166) that he 'seeks to apply standard comparative techniques to one dialect of Formosa, now perhaps extinct, using a word list compiled in the winter of 1873-1874', published separately by T. L. Bullock (1874-75), and J. B. Steere in "The aborigines of Formosa" (1879). He concludes however that his material 'does not permit a positive conclusion...', but only demonstrates "that there is an Oceanic [ = Austronesian — I. D.] element in Bu-hwan, and ... that this element may run deep. Beyond this we cannot go..." (Chrétien 1951:75f.). Nevertheless he later says (76) 'the fact that most of the vocabulary available is Oceanic, more particularly Indonesian, in form, argues for a more pervasive influence than the borrowing of a small number of words'. The classification of the Formosan languages has been the subject of some discussion. Yami is widely recognized to be closely related to Itbayat and Ivatan of the Philippines and is therefore generally disregarded in this connection. Nik. constructed the following groups : Atayal Group, including Atayal, Seediq, and Saisiyat. Tsou Group, including Tsou, Saaroa, Paiwan, Pyuma, Tsarisen, Rukai. Bunun Group: Bunun. Ami Group: Ami. LL form no groups beyond including Pazeh in Saisiyat and Kanakanabu in Saaroa. They do however report that Rukai is closely related to Puyuma and that Tsarisen is closely related to both. In his "Lexicostatistical classification of the Austronesian languages" (1965a) Dyen treated only six Formosan languages (Atayal, Seediq, Ami, Paiwan, Bunun, Thao) and determined that Atayal grouped with Seediq (Atayalic) against the other four which formed a second group. In Dyen's "The position of the Malayopolynesian languages of Formosa" (1963) a division into three groups is presented: 1) Atayal, Seediq. 2) Tsou. 3) Bunun, Rukai, Paiwan, Puyuma, Ami, Pazeh, Kavalan. It was there also suggested (1963:263) that since Ogawa and Asai say that Kanakanabu and Saaroa should be regarded as dialects of Tsou (1935:36), these two languages should be added to group 2. In an unpublished paper "The position of Favorlang among the Austronesian languages of Formosa" presented at the Eleventh Pacific Science Congress (August 1966) Dyen reported that Tsou, Saaroa and Kanakanabu exhibited lexicostatistical percentages with each other that implied strongly that they constituted a subgroup to which the name Tsouic has been given. Lexicostatistical percentages likewise weakly indicate the presence of a third subgroup formerly called by Dyen the East Formosan Hesion (1965b) and Catenate Formosan (1966). Ferrell (1966) has suggested the more convenient name Paiwanic, which is here adopted. The percentages presented were as follows:

174

ISIDORE DYEN TABLE 1

At

49.0 16.9 16.2 12.1 12.7 17.9 15.9 15.7 16.5 14.2 17.1 H.9 10.3

Se 18.2 17.8 16.4 13.8 17.5 19.5 16.1 16.3 14.8 19.9 12.4 10.2

Ts 39.7 34.5 18.7 22.4 23.7 21.9 22.8 19.5 20.6 16.4 17.5

Sr 37.2 16.9 19.4 20.0 20.7 20.9 19.5 20.5 20.7 14.8

Kn 18.8 17.6 22.4 20.7 20.7 19.4 22.7 18.1 14.9

Kv 21.8 21.3 19.5 23.9 21.0 22.8 17.6 14.2

Pz 28.3 21.8 23.5 26.3 24.3 15.8 25.2

Th 34.8 24.7 25.2 25.2 18.1 25.3

Bu 28.9 21.6 25.0 17.9 16.8

Am 29.9 Pu 30.2 31.3 Pa 21.4 29.0 31.5 Ru 17.9 24.4 21.3 16.2 Fv

Key to Abbreviations: Atayal (At), Seediq (Se), Tsou (Ts), Saaroa (Sr), Kanakanabu (Kn), Kavalan (Kv), Pazeh (Pz), Thao (Th), Bunun (Bu), Ami (Am), Puyuma (Pu), Paiwan (Pa), Rukai (Ru), Favorlang (Fv).

The closer relation between the Atayalic (Ate) languages (Atayal, Seediq) and between the Tsouic (Tsc) languages (Tsou, Saaroa, and Kanakanabu) is evidenced by the high percentages that these sets of languages respectively share with their co-members. Upon averaging their respective percentages with non-member Formosan languages we obtain the following table: TABLE 2

Ate 16.3 13.3 17.7 17.5 15.9 16.4 14.5 18.5 12.2 10.3

Tsc 18.1 19.8 22.0 21.1 21.8 19.5 21.3 18.4 15.7

Kv 21.8 21.3 19.5 23.9 21.0 22.8 17.6 14.2

Pz 28.3 21.8 23.5 26.3 24.3 15.8 25.2

Th 34.8 24.7 25.2 25.2 18.1 25.3

Bu 28.9 21.6 25.0 17.9 16.8

Am 29.9 Pu 30.2 31.3 Pa 21.4 29.0 31.5 17.9 24.4 21.3

Ru 16.2

Fv

The tables indicate relatively closer relations between the members of the Paiwanic group by percentages linking the pairs that range from 28.3% Pazeh-Thao to 34.8% Thao-Bunun. The position of Kavalan and Tsouic in relation to the Paiwanic group is somewhat unclear, for Ami is linked to the Paiwanic by its 30.2% with Paiwan and this is not significantly greater than Kavalan's 23.9 % with Ami. Similarly Pazeh is

THE AUSTRONESIAN LANGUAGES OF FORMOSA

175

linked to Paiwanic by its 28.3 % with Thao and this is not significantly different from Thao's 22.0 % with Tsouic. On the other hand Tsouic's association with Paiwanic is not significantly different from its association with Atayalic (16.3%). However the percentages linking the so-called Paiwanic languages are significantly different from their percentages with Atayalic. It thus follows that either Paiwanic is connected with Atayalic by a linking through Kavalan and/or Tsouic or it is discrete from Atayalic. Provisionally we continue with the appellation of Paiwanic. The position of seventeenth century Favorlang (the more modern, but extinct Babuza) is indicated to be lexicostatistically closer to Pazeh, Thao, and Puyuma (Paiwanic languages) than to Atayalic, Kavalan, or Tsouic (despite its failing to be so from Tsou). Raleigh Ferrell has treated the subject of the classification at great length in his Taiwan aboriginal groups: Problems in cultural and linguistic classification (1969), which has just appeared. He accepts the tripartite classification of the languages into Atayalic, Tsouic, and Paiwanic. He indicates (p. 426) that Ogawa has treated the subject of the place of the Formosan languages in the Indonesian languages in Indoneshia-Go ni okeru Taiwan Takasago-Go no Ichi (1944). Ferrell's work contains extensive wordlists for nearly all of the Formosan languages (Atayal, Seediq, Tsou, Kanakanabu, Saaroa, Rukai, Favorlang, Thao, Pazeh, Saisiat, Paiwan, Puyuma, Ami, Bunun, Kavalan, Siraya) and includes Yami. It also has a thorough bibliography. In an unpublished paper S.Tsuchida compared Siraya with Paiwan, Tsou, and Atayal. His percentages indicated that Siraya favored the Paiwanic group strongly against Atayalic, and only slightly against Tsou. Furthermore in his count Tsou favored Paiwanic more strongly against Atayalic than it did in Dyen's paper above. Using Tsuchida's materials my count showed an even stronger bias for Paiwanic. My figures are given below with Tsuchida's in parentheses: Siraya 26.7 (25.9) 24.1 (22.8) 15.0 (14.7)

Paiwan 26.0 (22.4) 15.0 (14.9)

Tsou 15.0 (14.0)

Atayal

These figures can be interpreted as indicating that either Dyen's original lists for the Tsouic languages were not full enough to bring this favoring out or the difficult reflexes of Atayalic and particularly Tsouic tended to mask this difference. In any case the lexicostatistical evidence now seems to point to no more than four different Formosan subgroups even after one has considered the hitherto uncalculated languages Favorlang and Siraya: i.e. Atayalic, Paiwanic, Tsouic, Kavalan. There is now evidence that points to the grouping of Tsouic with Paiwanic against Atayalic with Kavalan's position somewhat ambiguous, but generally favoring Paiwanic against Atayalic. To this extent quantitative evidence suggests two main groups on Formosa. Furthermore in Dyen's original calculation Atayalic did not seem notably to favor

176

ISIDORE DYEN

the other Formosan languages as against some non-Formosan languages. This suggested that the number of highest order (or prime) subgroups on Formosa was at least two. Since Atayalic's percentages with other Formosan languages are now higher, further inferences along these lines must await a new determination of the percentages of Atayalic with extra-Formosan languages. However the indication that there may have been no more than two prime subgroups on the island may be of some importance. On the basis of Dyen's (1965a) original lexicostatistical calculations there seemed to be three subgroups. A hypothesis of a non-Formosan homeland of the Austronesian languages would then imply three separate migrations into Formosa and would correspondingly reduce the likelihood of such a hypothesis. The reduction of the number of independent subgroups to two would correspondingly act to raise the likelihood of a non-Formosan origin. In the unpublished paper mentioned above Dyen gave evidence, also lexicostatistical, that Favorlang was probably at least associated with, and perhaps more likely a member of the Paiwanic subgroup. Tsuchida has now supplied evidence that the same can be said for Siraya, another extinct language. Thus the lexicostatistical evidence as a whole now seems to point toward the presence of no more than two independent subgroups on Formosa no more closely related to each other than to outside languages. On the basis of a non-Formosan homeland this would imply two independent migrations. Dyen (1963) also studied the vocabularies of Formosan languages for the presence of cognate sets restricted to Formosan languages, both of the type which crossed subgroup lines and those which were limited to subgroup members. He found some evidence pointing to a closer relation between Saaroa and Kanakanabu not indicated lexicostatistically. Furthermore there seemed to be a noticeable number of words in cognate sets with basic meanings restricted to Formosan languages with cognates in both Atayalic and other languages. A few of these words have since turned out to have extra-Formosan cognates, as was to be anticipated (e.g. Ilocano isbu 'urine' for [p. 266, 1.28] Se. rebu, Ts. sifu, Kn. ibu, Sr. i:bu 'urine', among a few others), but a significantly large number have not or at least not yet. Thus there remains considerable, though not necessarily decisive evidence for a hypothesis that the Formosan languages constitute a single family. The reconstruction of a Proto-Formosan must in principle await the adduction of additional evidence. However at the present time there seem to be only a few phonological elements of Proto-Austronesian which could not be reconstructed from the evidence of the Formosan languages. If it were true that all of them could be so reconstructed, it would be evidence for believing that the first division of ProtoAustronesian took place on Formosa. This is not very likely however because ProtoAustronesian was most likely a language spoken in the tropics and the last glacial period of Formosa probably ended too late for Proto-Austronesian, granted that it was on Formosa, to develop in a tropical climate. This point is discussed also in "The Austronesian languages and Proto-Austronesian", pp. 5-54 in this volume.

THE AUSTRONESIAN LANGUAGES OF FORMOSA

177

COMPARATIVE TREATMENT OF THE FORMOSAN LANGUAGES

The impact of the data of the Formosan languages has not yet been fully felt. Some of its effects were originally indicated in the introduction to OA (5-17), and later in Dyen's "Some new Proto-Malayo-Polynesian initial phonemes" (1962) and"Formosan evidence for some new Proto-Austronesian phonemes" (1965b). Most of these results have been reported in the chapter in this volume entitled "The Austronesian languages and Proto-Austronesian". Needless to say, much remains to be done.

DESCRIPTION

The widest coverage of the Austronesian languages of Formosa appears in OA. In this book there are twelve sections, each dealing with a language, the whole being preceded by an introduction consisting almost wholly of a valuable comparative treatment. At the end of the volume is an extensive wordlist. The languages treated are Atayal (Taihyoo dialect), Saisiyat (Taiai dialect), Paiwan (a number of dialects are reported), Puyuma (Pinan dialect), Rukai (Dainan and Shimosan dialects), Ami (Kibi dialect), Seediq (Musha and Taroko dialects), Bunun (three dialects), Tsou (Dufutu and Arisan dialects), Saaroa, Kanakanabu, and Yami. The last is spoken on Botel Tobago and unlike the Formosan languages is very closely related to Itbayat and Ivatan of the Philippine languages. Each section consists of a grammatical sketch, and a collection of texts, each of the latter being accompanied by literal and free translation and notes. The grammatical sketch is very limited. The phonology consists chiefly of a list of symbols, unfortunately semi-phonetic, and their values. The morphology generally consists of a list of affixes and their meanings, pronouns, demonstratives, verb inflections, and lists of particles and their functions. The material for the languages varies considerably in quality, but for many of the languages the account in this book is the only one available. The expository part and the glossing is in Japanese and the work is thus not easy of access. Fortunately I have had the advantage of a translation prepared by Masako Lounsbury. At the end of the volume there is an extensive wordlist covering about 285 meanings in the following languages and dialects: Atayal (Taihyoo, Takonan), Saisiyat (Taiai), Paiwan (Tokubun, Shimopaiwan, Kapiyan, Raisha, Kunanau, Kachirai, Rikiriki, Tachaban, Tamari), Rukai (Dainan, Taramakau), Puyuma (Chipon, Hinan [Pinan]), Ami (Baran, Kibi, Taparon, Tauran), Yami (Imurut), Bunun (Tamaroan, Katonguran, Ibaho), Tsou (Namakaban [Duftu], Chibora), Kanakanabu (Nagisaru), Saaroa (Haisen), Shimosan Rukai (Maga, Tona, Mantauran), Seediq (Hogo, Taroko [Iboho]). A translation of part of the wordlist appears in Naoyoshi Ogawa and Erin Asai's "Comparative vocabulary of Formosan languages" (1961). The only fuller treatments of Formosan languages are N. A. Nevskij's Materialy po

178

ISIDORE DYEN

govoram jazyka Cou (1935), now largely replaced by Tung T'ung-ho's posthumously published "A descriptive study of the Tsou language, Formosa" (1964), and Seren Egerod's series of articles on Atayal. The only other languages to have received more than cursory treatment are Seediq and Thao. What is striking is how structurally different the remotely related Formosan languages are. TSOU

Tung T'ung-ho's work on Tsou includes a large collection of texts from three dialects: Tapangu, Tfuea, and Luhtu. These are followed by a glossary. The texts and glossary are preceded by a phonological and grammatical treatment which is extensive and promising. However under the circumstances of its publication it is not surprising that occasionally crucial points are left undecided. A brief presentation of his conclusions is given here. Phonemes

The consonants are: /p b m f v 11 n c s z k ij ? h/. The associated phonetics are the usual ones except as in the following: /c/ is an affricate with the blade of the tongue articulated against the palate with sibilant release. Both /b/ and /I/ are preglottalized. The vowels are /i e a a u o/ among which /u/ is high central and the others have common phonetic values. In the Luhtu dialect there is an apical vowel /r/ contrasting with /e/ but varying freely with /e/ in some words. The vowels /i e o u/ are sometimes impressionistically non-syllabic when standing next to a different vowel of the same height or lower. Alternations

Most alternations concern the alternants of roots appearing before the passive suffixes /-a/ and /-i/. Among the consonants only /v/ and /z/ are involved in alternations: /v/ /v/ /v/ /z/ /z/

~ ~ ~ ~ ~

Iu/, as in /toaluqu/ (active) 'to angle': /toaluqva/ (passive). /o/ as in /eoeaso/ (active) 'to play': /eoeasva/ (passive). /«/ as in /toPtohui]«/ (active) 'to think': /toPtohurjva/ (passive). /i/ as in /ahoi/ (active) 'begin': /ahoza/ (passive). /e/ as in /mateoue/ (active) 'change clothes': /pateouza/ (passive).

However other alternations involving consonants also appear: /?/ ~ 0 as in /nacPo/ (active) 'distressed': /nacva/ (passive), /h/ ~ /kI as in /s-m-uhnu/ (active) 'send to do': /skuna/ (passive). /mj ~ /q/ as in /tumu/ (active) 'to ponder': /tiirjva/ (passive).

THE AUSTRONESIAN LANGUAGES OF FORMOSA

179

/mI ~ /p/ as in /msipgi/ (active) 'insert': /psipga/ (passive). /m/ ~ 11/ as in /maPsonu/ (active) 'guess': /taPsona/ (passive). /z/ ~ 0 as in /tatae/ 'admire': /tataepza/ (passive: suffix elsewhere only /a/). /v/ ~ 0 as in /rjoe/ (active) 'carry a burden': /qoevi/ (passive). All or nearly all of the vowels alternate with 0 under some condition. The following are typical: /«/ ~ /u/ ~ /o/ ~ /iI ~

0 as in /kaebu/ (active) 'to like': /kdeba/ (passive). 0 as in /soupu/ (active) 'to put together': /soupa/ (passive). 0 as in/titho/ (active) 'to use': /titha/ (passive). 0 as in /poccrji/ (active) 'to watch': /poceqa/ (passive).

whereas rarely: /a/ ~ 0 as in /bo-hiehti/ (active) 'cut branches': /hiaehtia/ (passive). A different type of alternation appears in many roots, presumably associated with shift of stress: /pono/ (active) 'shoot': /pnaa/ (passive), /t-m-opsa/ (active) 'read and write': /tposi/ (passive). Associated with the change of stress is the presence before a suffix of a final vowel which alternates with 0 in that position in other morphemes: /«/ as in /el«/ (active) 'search': /eMa/ (passive). /u/ as in /etamaku/ (active) 'smoke': /etamakua/ (passive). /o/ as in /coqo/ (active) 'hurt': /cogoa/ (passive). /i/ as in /esmi/ (active) 'pass by': /esmia/ (passive). Associated with this type is the relatively rare alternation between two vowels: /e/ ~ /i/ as in /soepe/ (active) 'stick in': /soepia/ (passive). It is interesting that alternation with 0 accompanied by constancy of stress does not affect the suffix /-a/ that is associated with the causative prefix /poa-/: /poasupeohua/ 'to let fall'; cf. /supohu/ (active) 'fall': supeohi (passive). The formation of actives where an affix is involved is generally with one of a set of alternants with /b/ or one of a set with /m/. Those with /b/ (1) have the alternant /b/ before a root with initial vowel: /bucumu/ (active) 'pluck a thorn': /ucuma/ (passive). (2) consist also of an additional vowel before a cluster that appears between members of the cluster in a different alternant: /bohqa/ (active) 'see clearly': /hoija/ (passive).

180

ISIDORE DYEN

and (3) otherwise have the alternant /bo/: /bohiciinu/ (active) 'cut in': /hiciina/ (passive). The affix /m-/ has two major sets of alternants: one prefixal /m-/ and the other infixal /-m-/. For the most part the affix /m/ has alternants like those of the prefix /b-/ in the same environments: /mimo/ (active) 'drink': /ima/ (passive), /maeo/ (active) 'take': /eaa/ (passive), /moteovaho/ (active) 'repair': /teovahi/ (passive). Like these are the instances of this alternant accompanying a replacement of /p/ or /t/ (see above). However in some cases (1) the alternant has a vowel like the first vowel of the root: /mahafo/ (active) 'take': /hafa/ (passive). (2) has the form /mo/ in: /mochi/ (active) 'throw away': /ciha/ (passive). and (3) /mu/ in: /musieu/ (active) 'to salt': /sieva/ (passive). The alternant /-m/ appears after the first element of the uninterrupted root and (1) appears after a single consonant in many instances: /tumamzo/ (active) 'ask': /tamzi/ (passive). and (2) has the alternant /-mo-/ before the second member of a cluster appearing in the uninterrupted alternant of the root: /smokoeacpo/ (active) 'sandwich meat between leaves': /skoeacpa/ (passive). There are also more complex types of alternation including suppletive alternation: e.g. there is a particle which under some conditions appears as /i/ and under others as /os/: see below. Syntax The syntactic discussion which follows is intended to sketch the main features of Tung's view of the syntax without claim to being exhaustive, exactly representative, or tight, even though his manner of presentation is generally followed. There are three major sentence patterns: (a) standard, (b) equational, and (c) /pan/, or casual (or existential). The standard clause The standard clause consists minimally of a core. The core is often a single word or a core phrase:

THE AUSTRONESIAN LANGUAGES OF FORMOSA

181

/timcoza/ 'He was killed by his hand.' Beginners A core word is commonly preceded by a beginning part. The beginning part consists minimally of a beginner. A beginner is (1) a /m/-beginner, (2) a minus-/m/-beginner, or (3) a /te/-beginner. The /m/-beginners are (meanings are interpreted approximately as explained by Tung; I follow his use of majuscules for morphemes with alternants): /mio/ 'durative' /MOI /mo/ ~ /mi/ 'punctual' /MOSO/ /moso/ ~ /mo/ ~ /moh/ 'remote past' The /m/-beginners occur with non-passive cores: e.g. /mio eiisuhrju/ '(He) was sitting.' i i /mo tueuI '(They) are three.' i i /moso ozomu/ '(They) went head-hunting.' i 1 1 The minus-/m/-beginners are: /I/ ¡ij ~ /os/ (corresponds to /MO/). /O/ /o/ ~ /oh/ (corresponds to /MOSO/). The minus-/m/-beginners are followed both by a post-beginner of the /si/-group (enclitic pronouns) and a passive core. The post-beginners of the /si/-group are: singular: 1 /o/ ~ /u/, 2 /ko/, /su/, 3 /si/, /ta/; plural: 1 inclusive /to/, exclusive /mia/, /mza/, 2 /mu/, 3 /he/, /ho/. For example (with the minus-/m/-beginner /O/): /o si eainca/ (lit. was-said by-it) 'It said.' For examples of the alternation /i/ ~ /os/ of /I/, see below. The /te/-group beginners are: /te/ 'anticipated', /nte/ 'imagined, desired', /nto/ 'imagined in error', /tena/, /ta/ 'new situation', /la/, /lea/ 'general'. The /te/-group beginners are followed by either a passive or a non-passive: /te ehoza/ '(I) will start speaking.' /te o eusbusbuta/ (lit. will by-me be-told) 'I will tell you it.' Post-beginners The class of post-beginners is made up of (1) the /si/-group, (2) the /cu/-group, (3) the /c?o/-group, (4) /nana/ (quotative), (5) /so/ ~ /s?o/ 'surely', (6) /la/ 'repeatedly, for a long time'. No more than one member of each group occurs and when members of different groups occur they appear in the relative order of their group. For the /si/group, see above. The /cu/-group consists of:

182

ISIDORE DYEN

/CU/ /cu/ ~ /c?u/ 'already' /av?a/ 'don't' The second alternant of the first member appears after a /te/-beginner, otherwise the first appears: /tena c?u mcoi/ 'It will then die.' /te c?u pnia/ 'He was about to shoot it.' /te o cu mcoi/ 'I'm going to die.' /o he cu poaeuova/ 'They sent him back.' and before /CU/ the alternant /mi/ of /MO/, /moh/ of /MOSO/, and /oh/ of /O/ appear: /mi cu zomu/ 'It became a bird.' /moh cu pkadko/ 'They fled.' /oh cu opcoza/ 'They killed it.' The alternant /os/ of /I/ appears before the post-beginners /o/ and /ko/: /os o cu aiti/ 'I saw it.' The post-beginner /av?a/ appears only after /te/ with which it sometimes contracts: /te av?a pnda/ 'Don't shoot it.' /tav?a s?entuha/ 'Don't cut it.' The members of the /c?o/-group are /c?o/ 'just', /s?a/ 'eventually, indeed', /n?a/ 'right away, on the point of.' /os o n?a ciiga/ 'I took it off right away.' /la u c?o taakukunueva/ 'I just don't believe it.' The post-beginner /nana/ indicates that the event is based on a report and not witnessed: /mi cu s?a nana aumtu peaeofu/ 'They, it is said, eventually (really) had a race.' 1

2

3

4

2

1

3

2

The post-beginner /SO/ has the alternant /?so/ after a te-beginner and /so/ otherwise: /la si ?so poceija/ 'He surely keeps looking it over.' The post-beginner /la/ is exemplified as follows: /tena to la eupabaito/ 'We shall be seeing each other.' Core phrases

The core phrase that consists of more than one word is serially constructed as in:

THE AUSTRONESIAN LANGUAGES OF FORMOSA

183

/fiho eoha/ 'went together to hunt' 1 2 1 1 2 or subordinately constructed as in: /o?te titho/ 'do not use' 1 2 1 1 2 /eurja naa/ 'took off first' 1 2 1 1 2 /k&eba emaa/ 'like to plant' 1 2 1 2 /akei muchu/ 'rain a little' 2

1

1

2

In the latter the attribute sometimes agrees in voice with the head: /mo i?vaho buerju/ 'He baked again.' (active). 1

2

2

1

/i si cu i?vaja eaa/ 'He took it again.' (passive: lit. by-him already again taken). 1

2

2

1

Conjunctive phrases Conjunctive phrases, when they occur, follow the core. They are of two kinds distinguished by particles: first conjunctive phrases and second conjunctive phrases. First conjunctive phrases are introduced by the particles (with the meanings given by Tung): /ta/ 'visible', /to/ 'invisible, but seen before by speaker', /no/ 'invisible and never seen by speaker', /ne/ 'invisible, but known to both speaker and hearer', /nca/ 'invisible, but nearby', /ci/ 'non-existent'. Second conjunctive phrases introduced by the particles: /e/ 'visible and near', /ta/ 'visible, but distant', /o/ 'invisible, but seen before by speaker', /na/ 'invisible and never seen by speaker', /co/ 'invisible, but nearby'. After a basic form (i.e. active) either type can occur, but if both occur, their order is indicated by their name. In any case a first conjunctive phrase with an active means the goal or place, whereas a second conjunctive phrase means the actor: /aomotuu no hicu/ 'talks to the god (first)' 1

2

1

2

/mivei ta tiskova/ 'reluctant-to-leave the world (first)' 1

2

1

2

/mi cu p£eo o hie/ 'The men (second) appeared.' 2 1 1 2 /moso nana uh ne taibueanu na okosi/ 'His son (second) went, I hear, to 3 2 4 1 1 2 3 Taibueanu (first).' 4 /moh cu ei?mi ta qeesdqsi na vcoqasi/ 'His wife (second) came from the plain 2 3 1 1 1 3 (first).' A n indefinite goal appears without a particle: /la dacni mooeai emi/ 'One always makes wine (first).'

184

ISIDORE DYEN

Two first conjunctives can follow an active with different meanings: /la tmaaeazoen to zoma to cconu/ 'One listens to the bird (first) on the road (first).' 2 3 1 2 3 1 With a passive the subject appears as a second conjunctive and various attributes such as agent, additional goal, place, etc. appear as first conjunctives: /sia to teova o keupu/ (lit. was-put in-the hunting-hut the net) 'One put the i I net (first) in the hunting-hut (second).' Passives Five types of passive are distinguished. Examples follow: 1) The /-a, -i/ or simple type (exemplified above and by): /i si cu nana fihi no mamespirji na teesi/ (lit. by-he it-is-said was-followed the 1 2 3 woman [first] the cord [second]) 'The woman followed the cord.' 1 3 2 2) The independent /-i/ (or local) type which appears with roots that have an /-a/ passive: /sii no taihi na chana/ (lit. is-a-place-of-putting' the compost the field) 'Put the 1 2 3 1 compost (first) in the field (second).' 2 3 For the use of the simple passive /sia/ see above. 3) The independent /-neni/ (or instrumental or benefactive) type which appears with roots that have a simple passive: /i si eobakneni no ceoa na sPofusi/ (lit. by-him was-used-to-strike the ground 1 2 3 [first] his stick [second]) 'He struck the ground with his stick.' 1 2 3 /te mu teaineni cucu na ao/ (lit. by-you be-made-for a meal-package [first] I [second]) 'You will make a meal-package for me.' 4) The /poa-, pa?-/ or causative type in which the subject is made to do what the root indicates: /poamainePe ne qeesaqsi e puutu/ (lit. be-made-to-return to the plain [first] the 1 2 3 Chinese [second]) 'Let the Chinese return to the plain.' (/mainePe/ 'return') /paPcohivi na eoifo/ (lit. be-fed the wizard) 'inform the wizard (second)' (Jcohivi/ 'know')

THE AUSTRONESIAN LANGUAGES OF FORMOSA

185

5) The causative-instrumental (/poa-, pa?-/ plus /-neni/) for which the subject (i.e. the second conjunctive) is the means used in causing an action: /nte si cu poaaneni no oko na cou/ (lit. by-him was-to-be-made-eaten, the cub 1 2 3 4 1 2 3 [first] the man [second] 'He (the bear) was about to feed the man to the cub., 4 /la si paPcohieni na haqu/ (lit. by-him made-known the enemy [second]) 'He let 12 3 1 2 them know about the enemy.' 3 Extended conjunctive phrases There are three types of extended conjunctive phrases. The first type consists of a conjunctive phrase of either type followed by a first conjunctive phrase: /to feou no iia/ 'the skin (first) of the deer {first)' 1

2

1

2

/ogko no emoo/ 'a name (first) of the clan (firsty 1

2

1

/os o cu tieopi o oko no 1 2 3 [first]) 'I held the small 3

2

hopi/ (lit. by-me be-held the small [second] the ladle 4 ladle.' 4

The second type consists of a head and attribute in either order, the second member being preceded by /ci/. This is most common when the attribute is a clause: /na ooko ci si topuei no trjoo/ 'the children (second) ci covered by the pot 1 2 3 1 2 3 Cfirst)' /na i si topuei no tgoo ci ooko/ (lit. the by-it covered by-the-pot ci children) 2 3 1 'the children covered by the pot.' 3 1 2 The third type consists of a head and attribute in which the attribute is a particle immediately preceding the head. The particles involved are exemplified by /nia/ 'former, ancient, past', /maezo/ 'same': /na nia oko/ 'the passed-away child' /ci maezo tuthusana/ 'the same Tuthusana' Tail Certain sentence attributes often appear after the core and conjunctive phrases if any. These attributes constitute the tail. Examples are /mai tan?e/ 'today, now', /ne noanao/ 'once, in the past':

186

ISIDORE DYEN

/moso la maica ne noanâo/ 'It was like this in the past.' 1 2 1 1 2 Preceders Certain particles, called preceders, can precede a beginner; these are such as /oa/ 'not', /oamocu/ 'not yet', /cima/ 'as it happens', /ma/ 'so!', /ci/ 'because, since', /a/ 'in reply': /oamocu mi o emomcovhi ne emoo/ 'I was not yet far away from home.' 23 4 2 1 1 3 3 4 1 The equational clause The equational clause type consists of two second conjunctive phrases, the first of which is a second conjunctive phrase in independent form, that is, one in which the particle appears extended by a prefixed /i-/. The independent form is the predicate: /ina éni na huphina no oko/ 'This (predicate) is the price of the child.' 1 2 3 1 2 3 Iie nia rjohôo e i to opcoza/ (lit. the former Ngohoo the [beginner] by-us killed) 1 2 3 4 'It is old Ngohoo whom we killed.' 1 2 3 4 This independent form can be predicate (i.e. a core phrase) by itself : /io la he n?a koa akea taauiva/ 'It is what they still rather have reason to 1 2 3 4 5 1 2 4 3 3 believe.' 5 Otherwise the independent form occurs in an absolute phrase preceding the core: I io teezo la boemi to meemeno ho poamaenoa / (lit. the fishing-fork uses [active'] 1 2 3 4 5 iron [first] and be-made-sharp) 'As for the fishing-fork, one uses iron and 1 2 3 4 sharpens it.' 5 /ina oko zou nana oko no poepe/ 'As for the child it was in fact, they say a 1 2 3 1 4 1 2 child of the wind.' 1 4 This is likewise true when two independent forms occur: /ina i si seoisi no teesi ma ina la noepohu ci kuhku/ 'As for what was tied with 1 2 3 4 5 1 the cord, why it was the wild-cat she was-living-as-a-pair with.'

THE AUSTRONESIAN LANGUAGES OF FORMOSA

187

The Ipanl clause

The casual (or existential) clause type has two sub-types. The first is with /pan/ 'there is'. This is followed by a first conjunctive phrase, but it is not definite in meaning, so that it could be identified with the second part of the extended conjunctive phrase type one (see above). /pan no nia ndna oko no tapdqu/ 'There was an ancient boy of Tapangu.' 1 2 3 4 1 1 2 3 4 /pan no moso eohu ci luhtu/ 'There were some Luhtus who went hunting.' 1 2 2 1 1 The second type has /uk?d/ 'there is not' and /uk?dna/ 'there is no longer' connected by /ciI to the following: /uk?d ci uachumu/ 'There are no buffalo.' i i /uk?a ci i si nana umnua/ 'There was not anything she liked.' i i /uk?ana ci snufusi/ 'There was no longer any skin.' i

i

ATAYAL

Among the works on Atayal that have recently appeared is the privately published Atayal vocabulary (Sqolyeq dialect), a selective translation of N. Ogawa's Atayaru Goshu (Japanese-Atayal word book [1932]) by Raleigh Ferrell (1967). Another interesting work dealing with Atayal is Otto Scheerer's "Sagen der Atayalen auf Formosa" (1931-32). It contains a number of texts with free and literal translations and extensive notes with mainly grammatical content. Soren Egerod has done the latest work on the phonology and grammar of Atayal and it replaces that done by Scheerer and by Ogawa. He has published the following four articles: "An English-Atayal vocabulary" (1965a), "Verb inflexion in Atayal" (1965b), "A statement on Atayal phonology" (1966a), and "Word order and word classes in Atayal" (1966b). Egerod's grammatical terminology changed somewhat between the articles 1965b and 1966b; equivalences will be indicated wherever pertinent. Phonemes

Egerod found the following consonants: / p b m t r n s c z l k g g x q h ? / . Of these /b/ and /g/ are sometimes stops, but often freely voiced spirants [P, y]. /c/ is an affricate with s-release before /i/, with s- or s-release before consonant plus /i/, and otherwise with s-release. /q/ is a uvular stop and /h/ is a pharyngealized spirant, /r/ is flapped, /z/ is [z] and occurs before only /i/ and rarely before consonants and is palatalized before /u/.

188

ISIDORE DYEN

Final vowels occur and only a single consonant can follow the last vowel. The consonants /b, r, z, g/ do not occur in final position. The other consonants do, and all occur in prevocalic position alone and in clusters. /?/ regularly occurs before initial vowels, but is not indicated in that position. The vowels are: /i e a o u/. /e/ is [e] and /o/ is [a]. A vowel in the penult is long: /nanu?/ [na:nu?] 'what'. On the other hand /i/ except before /i/ and /u/ except before /u/ are nonsyllabic in both prevocalic and postvocalic position. Only syllabic /i/ occurs before (nonsyllabic) /i/ and only syllabic /u/ occurs before (nonsyllabic) /u/: /iiat/ [?i:jat]. However vowel sequences with /u/ or /i/ first and standing before a last vowel are of varying syllabicity: /mian/ 'our' is [mian], [m'jan], [mi:jan] (i.e. /miian/); /spiun/ 'dream' is [s 3 piun], [Sgp'jun], [s a pi:jun] (i.e. /spiiun/); /nbuan/ 'drinking vessel' is [n3(3uan], [n 3 P u wan], [n„Pu:wan]. As in the last examples between the successive consonants of a prevocalic cluster there occurs [ J .

Alternations The alternations in the phonology in Atayal are quite remarkable; they are dealt with chiefly in Egerod 1965b. There are surprisingly few words with a syllabic in the antepenult or earlier in a word. The only instances that I have observed have been in the local passive past (i.e. Egerod's definite passive perfect 1965b = first passive perfect 1966b): e.g. ginlegan 'was led', inuahan 'was come to'. Thus a root that appears with two syllables in the absence of a suffix will have the first syllabic alternating with 0 when the root stands before a suffix; the suffixes are all monosyllabic. The root appears in unsuffixed form in the active imperative and this form is regularly cited as the independent form of the root e.g. kugus (i.e. 'scrape!') 'scrape'. The corresponding local passive (l.p.) is kgusan. Penult syllabic thus alternates with 0 . In a few instances a preceding nonsyllabic vowel is also dropped: siaqeh 'hate!': l.p. sqehan; kualai 'make!': d.p. (direct passive) klaiun. In a few roots a final /?/ remains upon the addition of a suffix: suaP 'do how': d.p. suaPun. In roots however the final /?/ of the root is lacking before the suffix and a preceding /a/ contracts with the vowel of the suffix: kitaP 'see': l.p. ktan; nagaP 'wait': d.p. ngon; but hiliP 'accuse': l.p. hlian; biruP 'write': bruan. It is important in this connection that, as was seen above, hlian and bruan are ambiguously [h 3 li:ian, b 3 ru:uan] or [h 3 ljan, b 3 ruan] respectively, for the former pronunciation conforms to the type of passive with suffix that has a long penult, whereas that with a short or no syllabic does not. A few roots with a penultimate syllabic lose the final vowel as well before a suffix: tehak 'arrive': l.p. thkan; qianux 'to live': l.p. qnxan. Nearly all, if not all roots with /i/ or /u/ immediately before the last vowel have suffixes as though they had a syllabic penult: quax 'wash things': l.p. quaxan. Otherwise roots lacking a syllabic penult somewhat arbitrarily belong to one of two

189

THE AUSTRONESIAN LANGUAGES OF FORMOSA

classes (or in a few instances to both) which differ in whether the final vowel of the root is retained before a suffix: kut 'cut': l.p. kutan\ but qhut 'arrange, crowd': d.p. qhtun. Certain final consonants that appear before a suffix have an alternant in the root before pause. Thus /g/ alternates with /u/ after /a/ or /u/ in the independent form: d.p. htagun: hutau 'fall'; l.p. ksiugan: ksiuu 'borrow'; d.p. phogun: phau 'to fine'; l.p. kogan: kuu 'too tired'; l.p. htgan: htuu 'come out'. In the single case after a different vowel /g/ alternates with /?/: d.p. nogun: gnoP 'joke'. Similarly /b/ alternates with final /p/ in d.p. gobun: mgop (active) 'share one cup'; and /r/ with final /i/ as in l.p. piaran: pgiai 'run away, desert'. As the last few examples illustrate, there are a number of irregular alternations to be found in the roots. In some instances this involves vowel variation: e.g. puij 'hear': l.p. poyan; qualax 'rain': l.p. qluaxan. But many involve the presence or absence of a consonant: e.g. sPurux 'stand'; l.p. srxan; lux 'insist': d.p. xun (for *lxuri); kziup 'enter': d.p. iopun, kiopun; ¡jlwj 'think': l.p. lluyan (both for *lr/luij, cf. the active l-m-tjluij). Syntax Egerod gives a brief outline of Atayal syntax in 1966b. He uses the term 'construct' for the central word or expression of a predicate (i.e. predicator). This can be followed by 'complements' and preceded by units in precentral 'exposure'. The construct commonly is a verb defined morphologically as a conjugatable word followed by one or more pronouns: muqi? sakuP. 'I forget.' 1

2

2 1

ktan sakuP niaP. '(lit. be-seen I by-him) He sees me.' i i Sometimes the construct, or rather quasi construct, consists of a noun and a pronoun or two nouns: Ptaial sakuP. 'I am an Atayal.' 1 2 2 1 squliq iutjai. 'The monkey is a human being.' 1 2 2 1 ita? quP (nominalizer) uahan niaP cisal. 'It is we that he came to play with.' 1

2

3

4

5

1

2

3

4

5

Units in exposure and complements are nouns, nominalized pronouns, or nominalized phrases. They indicate subject, object, or circumstance (time or place) without differentiation in form. Exposed units are of two types: (1) unmarked and (2) ending in ga?:

190

ISIDORE DYEN

ima? mosaP mluu iabaP 'Who will go with father?' 1

2

3

4

1

2

3

4

psianai ga? ini pkziap uzi 'Brothers-in-law do not wrestle either.' 1

2

3

4

1

2

3

4

Generally the pronoun sequence consists of one primary (nominative) and one secondary (genitive, possessive) form in that order: ktan saku? nia?. '(lit. be-seen I by-him) He sees me.' 1

2

3

3

1

2

ksiagun su? nha? kmal. '(lit. be-cheated you by-them speaking) They lied to 1

2

3

4

1

2

3

4

you.' However in one instance there is a complex pronoun misuP 'you (sg.) by-me': biqun misu? lukus. '(lit. be-given you-by-me clothes) I'll give you clothes.' 1 2 3 1 2 3 Preverbs (quasi verbs) supplement the conjugation of the verbal paradigm. They occur normally first in their clause except for exposed units and are followed immediately by primary and secondary pronouns associated in the construct. The occurrence of preverbs is exemplified by the following most common ones: 1) With indicative, niux 'close to speaker progressive', ciux 'away from speaker progressive', ual 'past', aki 'will': niux kuP mamu glanun. '(lit. ... I by you [pi.] be-neglected) You neglect me.' 1

2

3

4

1

2

3

4

ciux suP nbuu nanuP '(lit. ... you drink what) What are you drinking?' 1

2

3

4

1

2

3

4

ual nia? pson stunux rumaP. 'Some were sent by him to Stunux.' 1

2

3

4

5

5

1

3

2

4

aki magal kneril. 'He wanted to take a woman.' 1

2

3

1

2

3

2) With negatable indicative: iiat 'do not want to' inP 'not': iiat nha? kiaqeh babau nia. '(lit. not-want by-them have-bad-luck afterwards) 1

2

3

4

1

2

3

4

They did not want to have bad luck afterwards.' nhaP si mrhuu hiaP '(lit. 1not by-them put chief he) 'They did not make him ini? 1 2 3 4 5 2 3 4 5 chief.' 3) With imperative, laxi 'don't': laxi usa? qmaiah. 'Don't go to the field.'

191

THE AUSTRONESIAN LANGUAGES OF FORMOSA

There are a number of final (sentence) particles: ha 'formerly', la 'already, by now', ma 'I am told, I hear', ra 'of course': nuai ziau makuP skaial ta? ha. 'Let us first discuss my affair.' 1

2

3

4

5

6

1

5

6

4

3

2

niux baq mqianux la. 'Now he knows how to live.' 2

1

3

3

1

2

muci uzi ijarux ma. 'Thus also said the bear, I am told.' 1

2

3

4

1

2

1

3

4

ual suP slaqux squliq qasa ra. 'You won over that man, didn't you? 1

2

1

3

4

5

2

1

4

3

5

There are a number of preposition-like particles: tePi ~ ci 'direction', kiP 'inclusion', naP 'agent, possession', sa 'locative': mosaP sakuP te?i tjasal. 'I go (toward) home.' 1

2

3

4

2 1 3

4

msaiuP sami ki? kneril makuP. 'I (lit. we) had an argument with my wife.' 1

2

3

4

5

2

1

1

3

5

4

ktan sakuP na? squliq qasa. '(lit. be-seen I by man that) That man sees me.' 1

2

3

4

5

1

2

3

4

5

hkqiun niaP sa tjuruu. '(lit. was-found by-him in nose) He found it in the nose.' 1

2

3

4

1

3

4

Unmarked complements serve in various functions: 1) Direct goal with active: mitap sakuP squliq. 'I see the man.'

1

2

3

2

1

3

2) Local goal: psau misuP liuq cikui. '(lit. shall-be-sent thou-by-me river small) I shall send 1

2

3

4

1

2

3

4

you to the small river.' 3) Second goal with passive: biqun sakuP lukus. '(lit. be-given I clothes) You will give me clothes.' 1

2

3

1

2

3

4) Subject complement: ciux mhoqil iaba? maku?. 'My father is dead.' 1

2

3

3

2

1

[It is to be noted that since only pronouns are in the construct with a verb, a subject noun phrase is regarded as a complement. However the difference in position between the subject noun phrase and a pronoun may actually be due to the enclitic behavior of the pronoun which requires that it be placed after the first word of its clause: miux saku? nbuP. 'I am ill.']

192

ISIDORE DYEN

The so-called 'particle of adjugation' na? connects head and attribute in either order (when the attribute is not agentive or possessive): psaniq na? ziau; or ziau na?psaniq 1 'a tabooed matter.' 2 2 1 Unmarked constructs occur also as preceding and following attributes: nita? makuP ual hoqil squliq. 'I saw a dead person.' 1 2 3 4 2 1 3 4 milaP taP puniq ciux ptlapa?. 'We see a fire blazing.' 2 3 4 5 2 1 3 5 1 A construct also appears as complement: niux taP sqasan ual ta? maqux. (lit. by-us be-celebrated we won) 'We are 1 2

3 4

1

2

3 4

celebrating our victory.' Casual (existential) clause The words niux 'there is (here)', ciux 'there is (there)', and uyat 'there is not' as predicates form existential or casual clauses: niux pilaP mu. '(lit. there-is money my) I have money.' 1 2 3 1 3 2 ugat qaiaP makuP. '(lit. not-exist thing my) My things are not here.' 1 2 3 1 2 3 Morphology Egerod presents essentially the following table for verb forms in 1966b: Active Indicative Negatable Imperative Future Perfect

mPzero P(m)n-

Subjunctive

(m-) -a

-mzero

-mn-

zero zero zero

Third Passive

Second Passive

san san szero nsanai s-

-un

-an -i -i

(-)n -au

(-) n-an

First Passive

-ai

A reduced stem is used for the forms outside the heavy lines. Formation of the reduced stem sometimes entails ablaut and/or consonantal modifications. His presentation of the table in 1965b differs from the one here only in that there he used the terms relational passive ( = third passive), indefinite passive ( = second passive), and definite passive (=first passive). These terms probably can be further equated with the respective terms used by Bloomfield for Tagalog: instrumental passive, direct passive, and local passive.

193

THE AUSTRONESIAN LANGUAGES OF FORMOSA

However in both tables Egerod lists four types of active. I have listed only three. The second of Egerod's types is omitted here; it has a prefix k- indicated for the negatable forms and the only two instances cited with accompanying pronouns have agentive (not nominative) pronouns: iiat nha? kiaqeh babau nia?. '(lit. not-want by-them be-bad afterwards) They 1

2

3

1

2

3

do not want bad luck afterwards.' For the first type of active one can cite: makiP mkut uzi, makiP iniP pkut. 'There are some who cut, some who did not 1

2

1

3

4

1

1

2

1

3

cut.'

4

whereas for the second: aki magal kneril. 'He wanted to take a woman.' 1

2

3

1

2

3

ual iniP agal tunux kmukan. 'He has not taken a Chinese head.' 1

2

3

4

5

1

2

3

5

4

Examples of other forms of the active follow: (imperative) agal qaiaP sup. 'Take your things!' 1

2

3

1

3

2

(future) iiat simu p?agal. 'You will not get anything.' 1

2

3

2

3

1

3

(perfect) mnalax kneril likui qasa. 'That man has given up his wife.' (cf. 1

2

3

4

4

3

1

1

1

2

imperative alax 'give up!'), (subjunctive) hqila iabaP makuP. 'My father may die.' (cf. imperative hoqil 1

2

3

3

2

1

'die!'). The second ( = indefinite, Bloomfield's direct) passive is exemplified as follows: (indicative) biqun misuP lukus. 'I will give you clothes.' (cf. biq 'give!'). 1

2

3

2

1

2

3

(perfect) bniq mrhuu smiP lihui gasa. 'That right to apply tatooing has been 1

2

3

4

5

5

3

given by the chief.' (cf. biq 'give!'). 1 2 (subjunctive) spgau tap lukus suP. 'Let us measure your clothes.' (cf. sputj 1

2

3

4

1

2

1

'measure!'). Examples of the first ( = definite, Bloomfield's local) passive are: (indicative) hbgan makuP. 'I put a spell on him.' (cf. hgup 'put a spell!'). 2 21 1 1 1 (negatable) iniP sakuP habi naP iziuP. 'You have not given me an injection.' 1

2

3

4

(cf. hop 'stab', haban 'be-stabbed')

1

2

3

4

194

ISIDORE DYEN

(imperative) laxi sakuP kuci. 'Do not cut me.' 1

2

3

1

3

2

(perfect) iiat balai snian bcziux. 'Actually no dowry was given.' (cf. sip 'put!', 1

2

3

4

2

1 4

3

sian 'be-put'). (subjunctive) lgai makuP '(lit. would-be-accompanied by-me) will come along' 1 2 1 2 (cf. gluu 'Accompany!', Igan 'be-accompanied'). Examples of the third ( = relational, Bloomfield's instrumental) passive are: (indicative) ual nhaP sPagal pagai piataP. 'They used the bowl to scoop rice.' 1

2

3

4

1

2

4

2

3

(cf. agal 'take!'). (negatable) si nhaP an ssuliq btunux. 'So they just heat stones.' (cf. sully 'heat'). 1 2

3

4

2

1

3

4

(imperative) an skaial lakuP niaP. 'Please tell me his name.' (cf. kaial 'talk!'). 1

2

3

1

3

2

(future) qalup makuP lalau qani. '(lit. be-used-for-hunting by-me knife this) I 1

2

3

4

1

2

3

4

will hunt with this knife.' (perfect) inshoni? 'bewitched' (cf. honiP 'bewitch!'). (subjunctive) anai spquas cikai iaquP quas niaP '(lit. should-please-be-sung a 1

1

2

3

4

5

1

little that song of his) Please sing his song for a bit.' (cf. quas 'sing!'). 2

3

4

5

SEEDIQ

Ralph R. Co veil in his "Sediq syntax as related to problems of Bible translation" (1956) gives a brief statement of the phonemes before making a number of points about Seediq syntax. He says there are the four vowels /a i o e/, the last of which is [a]; unfortunately he almost immediately cites /buqa/ 'sweet potato' and the reader is left confused. He lists seventeen consonants: / p b m w t d n s l r k g g x y g h / . Of these /g/ is a spirant, /k/ is a palatal stop, /q/ is a back velar stop, and /r/ is an alveolarpalatal flapped vibrant (i.e. a flapped '1'). Unfortunately Covell deals with only a few points of Seediq syntax because of the limitations of his subject. These concern two common sentence types, a goal-marking suffix -an, and the frequent addition of messa 'it is said' in conversation where the one responsible for the original statement is not known or at least not emphasized. Erin Asai's "The Sedik Language of Formosa" (1953) attempts to describe chiefly the dialects of Paran and Gungu in the Musha district. Covell's material was from Terowan in the same district. Asai's description is extensive and includes a brief section on Seediq correspondences with other Austronesian languages. The description deals with the phonology, the morphology, and the pronouns and verbs followed by a few texts. Sentences as well

THE AUSTRONESIAN LANGUAGES OF FORMOSA

195

as texts are generally accompanied by literal as well as free translations. Although it can be said to replace the brief description that appears in OA, it only broaches the problem of describing this language. The phonology is given in a quasi-phonetic form. Although some attempt at phonemicization is made, the continuation of the use of the quasi-phonetic notation is sometimes confusing. From Asai's presentation one concludes that the consonants in the Paran dialect are about the same as those of Terowan except for the presence of/c/, an affricate with [s] release, in Paran. Furthermore /d/ is [d]. Words with long or double consonants are commonly cited throughout the book: e.g. laqqi 'child'. Unfortunately one is unable to tell whether this is distinctive because Asai says that consonants are sometimes lengthened: e.g. mita, mitta; miusa or mwssa. Furthermore Asai cites mekkan 'eat' whereas Co veil has mekan. Certain alternations like those noted for Atayal appear in Seediq. Thus the change of quality to /a/ in the pre-penultimate vowel appears in makila 'to know': kalaon 'known'. However Asai cites many words with pre-penultimate vowels other than /a/: e.g. taggiyan 'place of sleeping'. Likewise the alternation involving /g/ before a suffix appears in instances like: taba 'to rear animals': tdbagi 'rear'. The morphology as a whole shows many similarities to Atayal. This is particularly evident in the pronouns and verbs. The material presented for the verbs is in rather preliminary stages of organization with many details given, but not treated systematically. This is no doubt in part due to lack of sufficient material on which to base a treatment of the syntax.

OTHER L A N G U A G E S

Saisiyat Shigeru Tsuchida in his "Preliminary reports on Saisiyat: phonology" (1964) presents a careful treatment of the phonemics preceded by some information on the dialects and followed by a text with literal and free translation. He finds seventeen consonants / p b t k P r s z s h m n q l ' y w / and seven vowels /i e a o o o a/. Most of the phonemes have phonetic characteristics that would normally be associated with the symbols. However /b/ is a stop in initial position, freely a stop or spirant medially, and a voiced spirant in final position, /t s z/ are dental and /s/ is not interdental as it has been described elsewhere, /r/ is a voiced, post-alveolar, retroflex affricate [di] varying freely with a spirant [j] in all positions with the affrication more common in syllable-initial position and the spirant more common elsewhere. /'/ is smooth onset when initial (only before vowel), a weakening of voice in intervocalic position, and a smooth vanish (only after a vowel) in final position. When initial it is accompanied by a half-lengthening of a following non-prepenultimate vowel

196

ISIDORE DYEN

and in final position by a half-lengthening of the preceding vowel. It occurs also between consonant and vowel, the preceding phoneme being half-lengthened: /ba'la?/ is [ba-la?] and /mon'asab/ is [mon-asap]. Among the vowels /a/ is [as] and /o/ is [oe]. There are two kinds of syllables: CV and CVC, of which only the latter occurs in final position. Stress is automatically on the final syllable of a word with the exception of the words /bonaq/ 'betel-nut' (perhaps a borrowing) and /kigkayni?/ 'not like' (perhaps more than one word).

Kanakanabu and Saaroa Margaret M. Y . Sung in her "Phonetic and phonemic system of the Kanakanavu language, Formosa" (1966) finds thirteen consonants: / p m v t n l r s c k q P h / . The values are the common ones except as follows: /t/ and /n/ are alveolar, /r/ is a voiced alveolar trill in most positions, however resembling [z] in initial position and otherwise varying freely with a single tap [j]. /v/ is usually labiodental, but is bilabial before /u/. /sI and the affricate /c/ (with s-release) are palatalized to [s, e] and [tf, te] respectively before /i/. /?/ is a glottal stop intervocalically, but initially it is somewhat softer, resembling a voiced glottal spirant [h]. The nasals /m n r)/ become syllabic according to Sung when the vowels following the nasals are dropped in word-formation: e.g. /caniI 'one' + /panta/ 'bottle', /canpanta/ 'one bottle'. The vowels are /i e a A o a u/. /«/ and /A/ are respectively high and mid back unrounded. Although /e/ and /i/ vary freely in many words, there are some in which they contrast; a similar partial free variation was noted for /o/ and /u/ and for /u/ and /A/ as in /AVU/ 'knee', /uvu/ 'smoke'. Mrs. Sung distinguishes a pitch-accent and a stress-accent. The stress-accent generally falls on the same syllable as the highest pitched syllable. A few words bear a stress, but no high pitch; in such cases she uses the acute: /cau/ 'people, man', /tapurou/ 'sparrow', /siona/ 'place-name'. Otherwise she uses the overline for the highest pitched syllable. Such overlines are found on the antepenult (/itumuru/ 'many'), on the penult (/muciri/ 'stand') and on a last vowel (at least before a final consonant): /muqutur)/ 'to back'. Vowels appear in sequences of up to four vowels with any permitted member bearing the accent: /lumluu/ 'to catch fish'. Consonant clusters are listed only intervocalically and according to Mrs. Sung are the result of either (1) the dropping of post-nasal vowels in word-formation or (2) the obligatory loss of the vowel /«/ after /s/, /c/, or /k/ in rapid speech, but this point is not demonstrated. O n l y /m/ and /n/ appear doubled: /tammi/ 'sweet potato', /pakAnnanu/ 'how did it proceed?'.

A syllable is formulated as consisting of a vowel preceded by an optional consonant: i.e. (C)V. An alternative type is presented by the syllabic nasal: i.e. Mrs. Sung finds that there are 140 syllables attested in her material.

THE AUSTRONESIAN LANGUAGES OF FORMOSA

197

Mrs. Sung finds that in her material there is only one final intonation and that is falling. In the preceding part of the utterance a rising intonation sometimes occurs. Miss Yan (the later Mrs. Sung) had preceded her article on the Kanakanavu by another called "Languages of the Kanakanavu and LaPalua: a preliminary comparison" (1964). It is in Chinese and thus not available to me. From the brief English resume it can be learned that it deals with three topics: (1) comparison of the phonetic system; (2) comparison of words; and (3) comparison of the morphological and syntactic features. It also contains a small section dealing with sound correspondences between the two languages.

Thao Li Fang-kwei and his associates Chen Chi-lu, and Tang Mei-chun published their "Notes on Thao language" (1956). This too is in Chinese. It is the result of a five day recording session at the Thao village on the edge of Sun M o o n Lake. Though clearly preliminary in nature, it contains a brief phonology, morphology, and vocabulary, and two short texts. The words in the vocabulary are accompanied by English glosses. This vocabulary is for the time being the main source for Thao lexical material.

REFERENCES

ASAI, ERIN. 1939. Gravius' Formulary of Christianity in the Siraya language of Formosa. Facsimile edition of the original in 1662. Memoirs of the Fac. of Lit. and Pol., Taihoku Imper. Univ. 4/1.1-301. BLOOMFIELD, LEONARD. 1917. Tagalog texts with grammatical analysis. University of Illinois Studies in Language and Literature III.2-4. BULLOCK, T.L. 1874-75. Formosan dialects and their connection with Malay. China Review 3.38-46.

. 1876. Formosa. Journ. of the Am. Geogr. Soc. of New York 6.302-34. CHEN CHI-LU. 1958. A cultural configuration of the Island of Formosa. Bulletin of the Ethnological Society of China 2. CHEN KANG CHAI. 1967. Taiwan aborigines. Cambridge. CHRÉTIEN, C.DOUGLAS. 1951. Oceanic material in Bu-Hwan, a hill dialect of Formosa. Semitic and Oriental Studies, UCPSP 11.65-77. COVELL, RALPH R. 1956. Sediq syntax as related to problems of Bible translation. BT 73.171-85.

DYEN, ISIDORE. 1962. Some new proto-Malayopolynesian initial phonemes. JAOS 82.214-15.

. 1963. The position of the Malayopolynesian languages of Formosa. Asian P e r s p e c t i v e s 7.261-71.

ISIDORE DYEN

198

. 1965a. The lexicostatistical classification of the Austronesian languages. IUPAL, Memoir 19, supplement to IJAL 31. . 1965b. Formosan evidence for some new Proto-Austronesian phonemes. Lingua 14.285-305. . 1966. The position of Favorlang among the Austronesian languages of Formosa. Paper presented atXIth Pacific Science Congress in Tokyo (mimeographed). EGEROD, S0REN. 1965a. An English-Atayal vocabulary. AcOr 29.203-20. . 1965b. Verb inflexion in Atayal. Lingua 15.251-82. . 1966a. A statement on Atayal phonology. Artibus Asiae, Supplementus 23 (Luce Volume), 120-30. Ascona. . 1966b. Word order and word classes in Atayal. Lg 42.346-69. FERRELL, RALEIGH. 1967. Atayal vocabulary (Sqolyeq dialect). Taipei. . 1969. Taiwan aboriginal groups: Problems in cultural and linguistic classification. Inst, of Ethn. of Ac. Sinica Monograph 17. GABELENTZ, H. C. VON DER. 1859. Ueber die formosanische Sprache und ihre Stellung in dem malaiischen Sprachstamm. ZDMG 1859.59-102. GRAVIUS, DANIEL. 1661. Het heylige euangelium Matthei en Johannis... overgeset inde Formosaansche tale, voor de inwoonders van Soulang, Mattau, Sinckan, Bacloan, Tavokan, en Tevorang. Amsterdam. . 1662. 't Formulier des Christendoms met de verklaringen van dien, inde Sideis-Formosaansche tale. Amsterdam. (See Asai 1939). HAPPART, GILBERTOS. See Hoevell, Medhurst. HOEVELL,W.R.VAN. 1842. Woordboek der Favorlangsche taal zusammengestellt von Gilbertus Happart im jähre 1650 mit linguistischen erläuterungen. Verh. v. h. Bataviasch Genootsch. 18. KLAPROTH, J . H . 1822. Sur la langue des indigènes de l'île de Formose. Asia Polyglotta, 380-2. Paris. LACOUPERIE, TERRIEN DE. 1887. Formosa notes on Mss., races and languages. J R A S 19.413-94.

A. 1961. Naoyoshi Ogawa and Erin Asai : Comparative vocabulary of Formosan languages. JAS 2/2.5-32. LI FANG-KWEI, CHEN CHI-LU, and TANG MEI-CHUN. 1956. Notes on Thao language. Bulletin of the Department of Archaeology and Anthropology, National Taiwan Univ. 7 . 2 3 - 5 1 . LOUKOTKA, CESTMÍR, and PETER A . LANYON-ORGILL. 1958. A revised classification of the Formosan languages. J A S 1/3.56-63. MEDHURST, W. H. 1850. Dictionary of the Favorlang dialect of the Formosan language, by Gilbertus Happart, written in 1650, translated from the Transactions of the Batavian Literary Society. Batavia. NEVSKIJ, N . A . 1935. Materialy po govoram jazyka cou. Trudy Instituía Vostokovedenija 11. Moscow. NIKIGAWA, A . 1953. A classification of the Formosan languages. J A S 1 / 1 . 1 4 5 - 5 1 . LANYON-ORGILL, P .

THE AUSTRONESIAN LANGUAGES OF FORMOSA OGAWA, NAOYOSHI.

1931.

199

Paiwan Goshu (Japanese-Paiwan vocabulary). Taihoku

(Taipei). . 1932. Atayaru Goshu (Japanese-Atayal vocabulary). Taihoku (Taipei). . 1934. Ami Goshu (Japanese-Ami vocabulary). Taihoku (Taipei). . 1944. Indoneshia-Go ni okeru Taiwan Takasago-Go no Ichi (The place of the Formosan languages in the Indonesian language). Taiheiyo-Ken (Pacific Areas), 451-502. Tokyo. OGAWA, NAOYOSHI, and ERIN ASAI. 1 9 3 5 . The myths and traditions of the Formosan native tribes. Taihoku. . 1961. Comparative vocabulary of Formosan languages, prepared with introductory note by Peter A. Lanyon-Orgill. JAS 22.5-32. SCHEERER, OTTO. 1931-32. Sagen der Atayalen auf Formosa. ZES 22.81-114 (1931), 179-215 (1932). STEERE, J . B . 1 8 7 4 . The aborigines of Formosa. China Review 3 . 1 8 1 - 4 . SUNG, MARGARET M . Y . 1966. Phonetic and phonemic system of the Kanakanavu language, Formosa. BIHPAS 36.783-800. TSUCHIDA, SHIGERU. 1 9 6 4 . Preliminary reports on Saisiyat phonology. Gengo Kenkyu [Journal of the Linguistic Society of Japan] 4 6 . 4 2 - 5 2 . . 1968. Lexicostatistic study of four Formosan languages, (typescript). T U N G T ' U N G - H O , with the assistance of S . H . W A N G , T . K . K U A N , T.F. CHENG, and MARGARET M . YAN. 1964. A descriptive study of the Tsou language, Formosa. Taipei. VLIS, C . J . VAN DER. 1842a. Vocabularium Formosanum (Woordenlijst en gespreken i n ' t Sideisch-Formosaansch en Hollandsch). Verh. v. h. Bataviasch Genootsch. 18. Reprinted in alphabetical rearrangement and with an English translation. Memoirs of the Fac. of Lit. and Pol., Taihoku Imp. Univ. 2/1.154-202. . 1842b. Formosaansche woordenlijst volges een Utrechtsch handschrift. Soeracarta. Y A N , MARGARET M . 1964. Languages of the Kanakanavu and LaPalua: a preliminary comparison. BIHPAS 35.135-54.

THE CHAMIC LANGUAGES*

ISIDORE DYEN

The Chamic group of Austronesian languages are found in Vietnam and Cambodia. They are according to David Thomas, "Vietnam minority languages" (1968): Cham: two main dialects (1) about 50,000 speakers mainly in Phanrang and Phanri areas with scatterings in Binhthuan, Binhtuy and elsewhere; (2) about 100,000 speakers mostly in Cambodia, but with several thousand near Chaudoc and in Saigon-Cholon. Chru (Churu, Cado, Chrau): about 15,000 in southeastern Tuyenduc province. Hroy (Bahnar Cham): small group (76-10,000) in Binhdinh and Phuyen provinces. Jarai (Djarai; with subgroups Puan, Hodrung, Hrue, Arap): about 150,000. An influential tribe found mainly in Pleiku and Phubon provinces. Rade (Rade, Raday, Rde, Ede, Rhade; with subgroups Mdhur, Adham, Bio, Kadrao, Bih, Krung, Rde Kpa): 80-100,000. One of the most influential tribes found throughout Darlac and part of Khanhhoa provinces, centered around Banmethuot. Rai (Seyu): 5-10,000 speakers in Binhtuy and Binhthuan provinces, with a few in Tuyenduc province. Roglai: 37-47,000. There are three dialects: (1) Northern (Radlai, Aadlai) spoken by a little more than half in the mountains west and south of Nhatrang and by a few people near Dalat; (2) Southern, very closely related to Northern, spoken by somewhat less than half, found mostly in Ninhthuan and Binhthuan provinces, with a few in Tuyenduc; (3) Cac Gia, considerably different from the other two dialects and spoken by about 2,000 northeast of Phanrang. The Cham language continues that of the great kingdom of Champa which dominated southern Indo-China for many centuries. Its language appears in some 75 inscriptions, the earliest dated one from 829 A.D. (though there is an undated one thought to be from the 4th century) and the latest dated one from 1401 though there are several younger ones without dates. The inscriptions are dealt with in many sources, nearly all French, by scholars such as Etienne Aymonier, A. Barth, Antoine Cabaton, George Coedes, E.-M. Durand, Louis Finot, Edward Huber, and H. Parmentier. * This paper represents partial results obtained with the support of the National Science Foundation by Grant No. GS-1468.

THE CHAMIC LANGUAGES

201

The great work in the Chamic field is that of Etienne Aymonier and Antoine Cabaton, Dictionnaire cham-français ( = Publications de l'École Française d'ExtrêmeOrient, VII) (1906). Large dictionaries of equal scholarship are lacking for the other languages. There are however the following noteworthy dictionaries: Nicolle, Français-jaray et jarayfrançais (1942?) reported by Pittman (1959:59); Père Davias-Baudrit, Dictionnaire rhadé-français (1966). The Chamic languages were classified by Schmidt as a southeastern mixed group of the Austroasiatic languages (1926:140). In this he followed K. Himly who in his "Sprachvergleichende untersuchung des wôrterschatzes der Tscham-sprache" (1890) recognized them as mixed languages, but expressed himself in favor of their fundamentally belonging to the Austroasiatic family. This view has been generally rejected. The Chamic languages on the basis of lexicostatistic evidence probably belong to the West Indonesian group of languages (Dyen's West Indonesian Cluster in his classification of the Austronesian languages (1965)). Their nearest relatives are thus the languages of Borneo, Sumatra, Java, and Bali. No one has questioned their belonging together as a single subgroup within the Austronesian family. A somewhat rough lexicostatistical count for Rade, Jarai, Chru, Cham, and Roglai indicate beyond question that these languages are at least very closely related and in some cases may even be dialects of the same language; the latter is strongly suggested for Rade and Jarai. The percentages obtained (by A. Duffek) are : Rade 83.5 68.5 64.5 60.0

Jarai 71.5 66.5 66.0

Chru 73.0 73.0

Roglai 68.0

The relationship to other languages is however not entirely uncontroversial. Thomas and Healey in "Some Philippine language subgroupings" (1962) state that Rade, Jarai, Chru, and Cham form a subgroup that is coordinate with Malay within the Malay Stock; their Malay Stock is coordinate with Southern Mindanao on the one hand and a Philippine Superstock (including the other languages of the Philippines) on the other. This is roughly in agreement with Dyen's placing Chru (the only Chamic language he classified) in the West Indonesian Cluster, the latter being coordinate with the Philippine languages grouped in the Northwest Hesion. The suggestion has been made by G. K. Niemann in his "Bijdrage tot de kennis der verhouding van het Tjam tot de talen van Indonesie" (1891) that Cham (and so its Chamic congeners) is more closely related with Achinese among the Sumatran languages. Niemann saw agreements in the phonology, grammatical forms, and particularly in the lexicon. The phonological and grammatical arguments are too difficult to assess at this time and are difficult to present because in general Niemann does not select the arguments for this closer relationship from the comparisons with other

202

ISIDORE DYEN

languages. I was able to construct the following list of lexical comparisons from his collection. The lack of a question-mark implies that Niemann indicates that as far as he knows no other Indonesian language has a cognate with the same peculiarity. For those with a question-mark, such a remark is lacking: C. atuk 'joint, node', Ach. atut 'id.' (?) C. ba 'carry', Ach. ba 'carry' C. dom 'all', Ach. dum 'all' (?) C. galam 'carry on shoulder', Ach. gulam 'id.' (?) C. ing 'loins', Ach. kiéng 'id.' (?) C. kang 'chin', Ach. king 'id.' (?) C. koh 'hew', Ach. koh 'cut, chop, hew' C. krdtj 'orange tree', Ach. kéruét 'aromatic citrus tree' C. krong 'river', Ach. krung 'id.' (?) C. lapei 'dream', Ach. lumpoi 'id.' (?) C. mwnei 'bathe', Ach. manoi 'id.' (?) C. patao, putao 'king', Ach. pauté 'king, prince'. (?) C. poh 'strike', Ach. poh 'id.' (?) C. throng, srong 'draw water', Ach. tjérong 'id.' (?) C. takwai 'throat', Ach. takué 'throat, neck' C. tjagau 'honey-bear', Ach. tjagé 'id.' C. tjim 'bird', Ach. tjitjim 'id.' (Annam. chim). (?) C. tram 'kick', Ach. trom (?) C. trei 'satiated', Ach. troi Niemann explains these agreements by the hypothesis that a large part of the population of Acheh originated from Champa and that many of those who used to be regarded as immigrant Hindus could very likely have moved there from Champa. Of course he sets this thesis forth with caution. Blagden in his "Achínese and Mon-Khmer" (1929) collects evidence supporting Niemann's hypothesis in words like the Achinese presumed cognates of C. krdtj, tjim above. He finds the following Achinese words that can be compared with Mon or Khmer words: Ach. é? 'dung', Mon ik 'id.' Ach. cicém 'bird', Mon gacem (pronounced kecim) 'id.' Ach. po 'fly', Mon paw 'id.' Ach. kruéñ (see C. krong above), Mon kruñ 'river' Ach. keumuén 'nephew, niece' (C. kamwori), Mon kmin 'id.' Blagden believes that although like words occur in Nicobaran dialects, that is not a probable source since the Nicobarese are people of relatively lower culture. He suggests instead that Indonesians probably lined the whole Southern and Western coast of Indo-China in contact with Mon-Khmer inland. The Achinese might have been

THE CHAMIC LANGUAGES

203

living on the mainland somewhere between Kra and Penang opposite their present home in Sumatra. Here they may well have been in contact with Mon-Khmer people and have derived a few words from them. The Achinese then (following Kern's hypothesis of an Austronesian homeland on the continent) would have been the last of the Indonesians to leave the mainland for the area they now occupy. The most extensive description of a Chamic language is E. Aymonier's Grammaire de la langue chame (1899). There is an extensive treatment of the phonetics and of their orthographic representation which occupies a little less than one-half of the grammatical discussions. This part contains a great deal of information including remarks about the phonetics of the dialects. It is followed by short sections dealing with numerals, dialectal differences, word formation, pronouns, adjectives and demonstratives, various types of particles and a brief syntax (about three pages). All of this is followed by a transcription, lexicon, and translation of the Royal Chronicle. David L. Blood in his article "Phonological units in Cham" (1967) gives the most modern treatment of a Chamic language, which is outlined with some modifications in the following. A word as a potential free form can be assigned to one of three types distinguished by their constituent syllables, phonetic pattern, and the phonemic constituency of the syllables. The types are (1) (pa)pan ['plank'] or single-stressed words; (2) ta.hpat ['to bow'] or trisyllables, and (3) ci.cih ['to be clean'] or reduplications. Both of the latter two types involve a secondary stress marked by period: the trisyllables on the first syllable and the reduplications on the syllable before the seam (indicated by Blood with space). The marking of secondary stress suggested here will provide for the possible contrasts between *tam.lay and *ta.mlay and one between *pan.mitjtay andpan.miij.tay. Blood indicates that a word with secondary stress is not distinguished from a sequence of words in which the first has secondary stress. The (ps)pan type consists of an obligatory, or main, syllable that may optionally be preceded by a preliminary syllable. Dissyllabic (ps)pan words have a weakly stressed preliminary syllable of very short duration followed by a heavily stressed main syllable as in [pa'pan]. There is considerable variation in the preliminary syllable usually between [a] and another vowel or no vowel at all, but sometimes also involving a nasal coda lj as in [miq'ta ~ ma'ta ~ m'ta] 'eye'. The vowels of a preliminary syllable are limited to /i i a a u/. The middle syllable of a trisyllable is unstressed and its vowel is limited to /i i a a/. Some words show a variation between 3 and a different vowel: ta.hpat ~ ta.lipat 'to bow'. Trisyllables generally compete with a fused form in which the medial syllable is replaced by a nasal consonant; the latter becomes the coda of the altered first syllable, which, however, retains its secondary stress: pan.miytay ~pamatay ta.hpat ~ ta.lipat tha.hp3n ~ t"a.lip3n ka.limirj

pam.tay 'to kill' tam.pat 'to bow' tham.p3n 'nine' kan.miq 'spider'

204

ISIDORE DYEN

Reduplications are either full (e.g. karurj.kararj 'to be in anguish' [karuy 'worried']) or partial (e.g. tari.tareij 'industrious' [tarey 'diligent']). Repetitions like myet.myet 'forever' could be considered a special case of reduplication. The consonant phonemes are: /p p h b m w t t h d n 1 r c c h s ñ y k k h q ' h/. The vowels are: /i e e i a a u o o/. The values of the phonemes are common ones except as indicated below. /p t c kI and /p h t h c h k h / are respectively corresponding aspirated and unaspirated voiceless oral stops. Aspirated stops do not occur in final position. /p h / has the variants [p h , f], /1 t h / are alveolar. Retroflex [t, t h ], which occur only before stressed vowels, are interpreted as /tr t h r/ respectively (presumably despite the absence of variation with */tsr-/, */t h ar-/ respectively), /c c h / are palatal to prepalatal [?, Qh] respectively in non-final positions, and in final position /c/ is usually a glottal stop with prepalatal coarticulation [-j']> very rarely varying with [5]. /b d/ are voiced oral stops, sometimes preglottalized, everywhere except that in final position /b/, the only one to occur here, is [-u']. /s h/ are voiceless fricatives; the first generally alveopalatal or prepalatal and the latter isoglottal. Final /s/ is a prepalatalized fricative [-ih]. A retroflex [s] has been interpreted as /sr/ (? occurring only in the freely varying [sam ~ sam] = /sram/ ~ /sam/ 'a lesson') despite the occurrence of phonetic [sr], also interpreted as /sr/ in sra" 'a writing'. /m n ñ T)¡ are nasals with /ñ/ occurring only before stressed vowels and /g/ only before stressed vowels and finally. /I r w y/ are liquids: lateral, flap and (two) semivowels respectively. There is free variation between phones assignable to /n/, /I/, and /r/ in final position. Since there are some contrasts between a final nasal and the free variation between /I/ and /r/ (e.g. pian 'when', piar, pial 'low') whereas there are none involving final /l/ and /r/, Blood speaks of the contrast between these two phonemes as being neutralized in final position. This matter is treated more fully in David Blood's "A problem in Cham sonorants" (1962). Homosyllabic consonant clusters are biconsonantal and triconsonantal. Biconsonantal clusters have a liquid as their second member and triconsonantal clusters have a liquid as the second member and a semivowel as third; e.g. mlay 'demon', Iwa' 'entrance', blwa' 'more than'. Among the vowels /¿/ is high central unrounded. Before /w/ or /y/, /a/ is [a - ] and /a/ is [a]: i.e. they differ only in length, as inpataw [pata'w] 'king',pataw [pataw] 'stone'. Elsewhere they differ in height. Along with the nine simple vowels there are also two sequences [i3] and [u 3 ] which Blood finally assigns to /i 3 /, /u 3 / after much discussion; he unfortunately does not make it plain why these could not be /ia/ and /ua/; perhaps it is because otherwise vowel sequences do not occur. Low pitch /V contrasting with unmarked non-low pitch occurs with the voiceless oral stops' káh 'direction', kah 'fish scale'; mata 'rich', mata 'eye';para' 'north',para* 'silver'.

THE CHAMIC LANGUAGES

205

Blood states that phonetically syllables with initial consonants other than voiceless oral stops have non-low pitch (1967:29). He does however add that a prosodeme of pitch may stretch over both syllables of a (pd)pan word, regardless of which syllable owns the phonemic pitch: tehla' [tehla'] 'I' (30). Doris Blood, in "Reflexes of ProtoMalayo-Polynesian in Cham" (1962:19), implies that the latter case is the rule, for she goes so far as to mark the pitch on the main syllable of dissyllabic words even when the stop is initial in the preliminary syllable: her /perew/ 'new' is his /paraw/. Throughout his article Blood gives examples of shortened forms. Such shortened or fused forms are found for the (pa)pan type as well as the ta.hpat (trisyllabic) type. The fused forms of the (pa)pan type generally involve a dissylable with monosyllabic fused form: ?ikan ~ kan 'fish'. Occasionally a simple variation between /a/ and a different vowel is involved: pHraw ~pihraw 'new'. Since ci.cih words are reduplicated (pa)pan words, fusion would affect the individual members of the reduplication. Thus ci.cih 'to be clean' is a reduplication of (hd)cih 'clean'. There are certain characteristics that occur predominantly or wholly in women's speech. Some men also speak with these characteristics, particularly illiterates. There are also some women who do not speak with these characteristics. The phenomenon is treated by Doris Blood in her article on "Women's speech characteristics in Cham" (1961). She finds that the principal differences in the phonology concern (1) the phoneme r, (2) retention of preglottalization, (3) phonetic changes in the preliminary syllable and some in the word base, and (4) the g segment following velar nasal. The r in initial consonant clusters is replaced by y: e.g. hray 'day' is hyay. Furthermore initial r is also often y: e.g. roy 'crawl' may be pronounced yoy. When the word base begins with a velar nasal, the preliminary syllable is often dropped and a velar stop release g follows the nasal: taijin 'arm' is fused to rjgin. This is the only condition under which g may be part of the alternation; the word rja? 'to work' is never *rjga In some words in men's speech /n/, /l/, and /r/ may alternate freely in final position, but in women's speech only /n/ occurs: e.g. 61 ~ or~ on 'to remember fondly' is only on in women's speech. It seems evident that the speech of men is on the whole more affected by the Cham script, showing fewer instances of fusion and a greater tendency toward spelling pronunciation. Since fewer young men are now becoming proficient in the use of the script, Mrs. Blood suggests that the so-called women's speech is perhaps a preview of the Cham language of the future. Doris W. Blood (1962) treats the correspondences between Dempwolff's reconstructions of Proto-Malayopolynesian and the Cham words assignable to them. She uses Dyen's orthography and her usage will be followed here. She follows David Blood's analysis of Cham phonology, but uses a slightly different symbolism in which her /e/ is his /E/; her /!/ is his /i/; her /e/ is his /a/; her /o/ is his /o/; her /o/ is his /o/; her /cs/ is his /c h /; her /q/ is his /'/; and her /ng/ is his /q/.

206

ISIDORE DYEN

The PMP voiceless stops (*p, *t, *c, *k) in non-final position generally appear unchanged in Cham (C.) except as indicated below: *pusej, C.pethaq 'navel'; *taqun, C. thun 'year'; *ciyum, C. cum 'smell'; *hikan, C. qikan 'fish'. However *c is dissimilated to k before another *c: *cecak, C. kecaq 'salamander'. Furthermore *t becomes & in a cluster with */: *teluh, C. klew 'three'. PMP b, d, D, j, g, z, Z are reflected in Cham by the voiceless stops p, t, t, t, k, c, c respectively accompanied by a low pitch on the syllables in the same stress unit following the stop: *habuh, C. hepew 'ash'; *dilah, C. telah 'tongue'; *quDay, C. hetang 'crustacean'; *hijutj, C. qatung; *gulurj, C. kelung 'roll'; *zaRum, C. cerum 'needle'; *Zalan, C. celan 'road'. However when the second syllable of a dissyllable has initial *t, the low-pitch does not appear: *batuk, C. petuq 'cough'. Inexplicably there are a number of instances in which *b is reflected by /b/: *buhuk, C. buq 'hair'; similarly in one instance *(dD) appears as /d/: *pe(dD)iq, C. pediq 'to pain'. *q generally became h, but as initial of a main syllable fused with the initial unaspirated stop of a preliminary syllable to form an aspirated stop: *quZan, C. hecan 'rain'; *taqun, C. thun 'year'; *puluq, C. pluh 'ten'. In intervocalic position *h is lost: *buhuk, C. buq 'hair'. Doris Blood does not distinguish between etyma beginning with *h from those with initial vowel. Nevertheless some words with PMP initial vowel and others with initial *h (according to Dyen 1953) appear with C. initial q: e.g. *hikuR, (Dyen *ikuR), C. qiku 'tail'; *hiyup, C. qayuq 'blow'. There are in addition some words with PMP initial vowel that appear with C. initial h: *habu(h) [Dyen *abu(h)], C. hepew 'ash'. In a few instances she somewhat awkwardly assigns m (of a prefix) as an outcome from *h; for example C. menum 'to drink' should rather be interpreted as from *-inum plus a prefix m- rather than directly from *hinum. *m, *n, *tj are retained in non-final positions: *manuk, C. menuq 'hen'; *minak, C. meniq 'oil' * hay in, C. qangin 'wind'. *n is generally retained in main positions, but becomes I in preliminary positions: *henem, C. nem 'six'; *nupih, C. lipey 'dream'. *n, appear as n respectively in *inum, C. menum 'drink', and *hiyus, C. huh 'blow nose'. *s in main position becomes C. th: *baseq, C. pethah 'wet'. In preliminary syllables it appears as h before a stop, t before *s and as s before r: *sakit, C. hekiq 'sickness'; *sira(q), C. sera 'salt'; *susuh, C. tethew 'breast'. *r and *R are reflected as r in non-final syllables: *ribuh, C. ripew 'thousand'; *Rusuk, C. rithuq 'rib'. */ appears as / in non-final position: *qulej, C. helaq 'worm'; *lima{h), C. limi 'five'. *w in Mrs. Blood's material is found intervocalically only after *u in PMP words. It is reflected either as h or q in preliminary positions: *waRi(h), C. hrey 'day'; *wayeR, C. qya 'water'. The last example is perhaps the best for *y, otherwise poorly attested. Final *p, *t, *d, *k, and *j are reflected as Cham q: *(q)as(ae)p, C. qathaq 'smoke'; *uRat, C. qaraq 'vein'; *(t)uhud, C. tequq 'knee' (first q unexplained); *batuk, C. petuq 'cough'; *pusej, C. pethaq 'navel'. For reflex of final *q see above.

THE CHAMIC LANGUAGES

207

*h, *R are lost as finals: *taquh, C. thew 'know'; *hikuR, C. qiku 'tail'. The actual reflex of final *r is not certain, for the only available instance is one that hitherto has been reconstructed ambiguously as (*/•£>). On the assumption that *D would probably become C. q (like *d), *sisi(rD) could be corrected to *sisir on the basis of C. tethi 'comb'. Final *m, *n, *rj are retained: *Dalem, C. telem 'depth'; *papan, C.pepan 'board'; *quDat7, C. he tang 'crustacea'. Final *s is reflected as h: *beRas, C. prah 'polished rice'. Final */, *w, *y are retained: *kapal, C. kepal 'thick'; *danaw, C. tenaw 'lake'; *hapuy, C. qapuy 'fire'. Mrs. Blood finds that the vowels of the preliminary syllable have not been retained nor reflected consistently and therefore discusses only the vowel reflexes in the main syllable: 1) *i, *e, *a, *u generally appear in Cham as i, e, a, u respectively before a final consonant other than h: *kulit, C. keliq 'skin'; *henem, C. nem 'six'; *babaq, C. pepah 'mouth'; *labuq, C. lipuh 'fall'. However in a number of instances C. a appears for *e: *qulej, C. helaq 'worm'. Likewise C. o appears for *u: *DukDuk, C. toq 'sit'. 2) *a before final *h is generally C. a: *DaRa(h), C. tera 'maid'. Main syllable *a after *m, *n, *n becomes C. i: *hamah, C. qami 'father'; *hanak, C. qaniq 'child'; *minak, C. meniq 'oil'. After *tj, *a appears as i: *taijan, C. tengin 'hand'. 3) Final *u, *i before final *h or *(h) are reflected by C. ew, ey respectively: *batuh, C. petew 'stone'; *belih, C.pley 'buy'. Richard S. Pittman (1959) presents a large number of Jarai words that reflect PMP etyma. He notes a generally close equivalence between the Jarai and the PMP vowels with the following two shifts. Only a and e appear in the first syllable of dissyllabic Jarai word bases; e replaces all original vowels in this syllable (— the preliminary syllable of Cham) except when the vowel is initial in Jarai: *hikuR, J. aku 'tail'; *hamah, J. ama 'father'; *quDip, J. hedip 'live'; *kapas, J. kepaih 'cotton'; *limah, J. rema 'five'. Word-final *-i, -ih becomes J. -ai; word-final *ey becomes J. ai or J. ei (as yet indeterminably); word-final *ih becomes J. ei; and word-final *uh becomes J. ao after *q, */, *r and J. do after any other consonant: *kaki, J. takai 'foot' (Pittman's own formulation); *qatey, J. hetai 'heart'; *beRey, J. brei 'give'; *qubih, J. hebei 'sweet potato'; *taquh, J. thao 'know'; *batuh, J.petao 'stone'. *q becomes J. h: *quZan, J. hejan 'rain'; *paqit, J. phiq 'bitter'; *baseq, J. sah 'wet'. Pittman like Mrs. Blood does not distinguish between the initial vowel and the initial *h of Dyen (1953). Thus using Dempwolff's formulae he speaks of wordinitial *h in words such as *hikan, J. akan 'fish' where he is simply transcribing Dempwolff's smooth onset. Pittman gives a list of Jarai words with initial vowel assigned to etyma with initial *h in the etymology and rightly indicates that Dyen reconstructed nearly all with initial vowel. See Note p. 210. Intervocalic and final *h disappear: *buhuk, J. buq, buk 'hair'; *kutuh, J. ketao 'louse'.

208

ISIDORE DYEN

Most stops remain unchanged. The following are the exceptions. Word-final *t and *k sometimes become J. q: *zaqit, J. siq 'sew'; *hanak, J. anaq 'offspring'; but *huRat, J. arat 'vein' and *buhuk, J. buq, buk 'hair'. Among some other miscellaneous changes, Pittman notes: 1) Word-initial *w becomes J. h: *waRih, T. hrei 'day'. 2) Word-initial *z becomes J. s: *zaqit, J. siq 'sew'; 3) *Z becomes J.j: *quZan, J. hejan 'rain'; *Zalan, i.jelan 'road'. The appearance of J. jelah (cf. Ngaju jela) 'tongue' is worth noting, since it does not agree with C. telah which, as Mrs. Blood indicates, points to a different initial like Malay lidah (with metathesis). 4) Word-initial *D becomes J. d: *Dalem, J. dlam 'deep'. The words reconstructed with *(dD) likewise show J. d: *(dD)uRi(h), J. drei 'thorn', however *{dD)aqan is probably reflected by J. than 'branch'. 5) *j becomes J. d: *pajey, J. pedai 'unhusked rice'; however in the instance *hajetj, J. hedang 'charcoal', the result is d. In her "Proto-Malayo-Polynesian reflexes in Rade, Jarai, and Chru" (1963) Mrs. Thomas gives a carefully drawn up listing of the cognate sets of the three languages by their reflexes. She precedes this by some general rules of phonetic change. All three languages — Rade, Jarai, and Chru — appear to have qa- for initial *hor * V other than *e, except that the vowel is sometimes lost in Rade, and if so, the q sometimes as well: *hapuy, R. qapuy,puy, J., Cr. qapuy 'fire'; *hiyup, R. qayuq, qyuq, J. qayup 'blow'; *hasuh, R. qasew qasaw, J., Cr. qasew 'dog'. The first vowel of dissyllables is regularly lost in Rade, but becomes Rade e after an initial PA voiced apical consonant (*D, *Z, */, *r, *R), which becomes *q, except when followed by *q or *w: *kutuh, R. ktSw, J., Cr. katSw 'louse'; *Danaw, R. qenaw, J., Cr. danaw 'lake'; *Zalan, R. qelan, J., Cr.jalan 'road'; *limah, R. qema, J. rama, Cr. lama 'five'; *ribuh, R. qebSw, J., Cr. rabSw 'thousand'; *Ratus, R. qetuh, J. ratuh, Cr. ratiih 'hundred'. In all three languages an *e that is the first vowel of the dissyllable is dropped: *hepat, R. paq, J. paq, Cr. paq 'four'. In Jarai and Chru the first vowel in PA dissyllables becomes a under most conditions, but is always dropped if it stands before a prevocalic *r, *R, *q, or *h, and usually dropped if before */ when the initial is *m, *p, *k, *g, or *q (see *kutuh above): *sira(q), R., J. hra, Cr. sra 'salt'; *buRuk, R., Cr. bruq, J. bruq 'rotten'; *(dD)aqan, R. qadhan, R., J., Cr. dhan 'branch'; *bahuh, R. baw, J., Cr. baw 'stench'; *malem, R., J., Cr. mlam 'night'; *palaj, J., Cr.plaq 'palm (of hand)'; *kulit, R. kilt, J. kllq, Cr. kliq 'skin'; *bulan, R. mlan, J, Cr. blan 'moon'; *gulutj, J. glurj 'roll'; *qulej, R. hlwdt, J. hlat, Cr. halaq 'worm'. In Rade m often appears for initial *b: *basaq 'wet', R. msah, J. sah, Cr. pasah 'wet'; *batuk, R. mtuk, i.patuk, Qx.patuq 'cough'. So occasionally for *p: *puntih, R. mtey, J. patay, Cr. patSy 'banana'. Furthermore in Rade and Chru the high vowel of the first syllable before *R, *l is

THE CHAMIC LANGUAGES

209

sometimes reflected by a semivowel after the reflex of the following consonant: *(dD)uRih, R. qêrwê, J. dray, tray, gray, Cr. drway, drway 'thorn'; *hiRa(q), R. hrah, J., Cr. m-ryah 'red'; see also *qulej above. In addition to those features which have appeared in the preceding discussion which resemble and differ from Cham's, there is the noteworthy fact that none of the three languages exhibits the change of PA voiced stops to voiceless ones with associated low tone that appears in Cham. Furthermore most of the changes are not only compatible with, but tend to favor an association with the languages of western Indonesia (Dyen's West Indonesian Cluster, perhaps equivalent to Thomas and Healey's Malay Stock). A notable agreement between the Chamic languages and Malay appears in showing resemblant but distinctive reflexes for the same proto-phoneme as hitherto reconstructed: *waRih, Malay (h)ari, C. hrey, R. hrwêq, J., Cr. hray 'day'; *wayeR, Malay aer, C. qya 'water'. It is clear that we must now consider the possibility that different initial proto-phonemes may be involved, yet it remains a remarkable fact that the difference in reflex is limited, as far as I know at this time, to Chamic and Malay. In the preceding discussion we adhered to the approach that the authors of the articles employed. There is now accumulated evidence that Dempwolff's reconstructions are insufficient to account for the systematic correspondences that can be found among the Austronesian languages. The authors of the comparative articles reviewed here have done a valuable service in presenting their interpretation of the Chamic reflexes of Dempwolff's reconstructions. It is clear however that much remains to be done both in reconciling the cognates with the new reconstructions and in pursuing the problem of the immediate relationship of the Chamic languages with other Austronesian languages. REFERENCES

Grammaire de la langue chame. Saigon. , and ANTOINE CABATON. 1 9 0 6 . Dictionnaire cham-français ( = Publications de l'École Française d'Extrême-Orient VII). Paris. BLAGDEN, C.O. 1929. Achinese and Mon-Khmer. Feestbundel uitg. d. h. Kon. Bat. Gen. v. Kunsten en Wetensch. 1.39-48. BLOOD, DAVID L . 1962. A problem in Cham sonorants. ZPhon 15.111-14. . 1967. Phonological units in Cham. AnL 9.15-32. BLOOD, DORIS WALKER. 1 9 6 1 . Women's speech characteristics in Cham. Asian Culture 3 . 1 3 9 - 4 3 . . 1962. Reflexes of Proto-Malayo-Polynesian in Cham. AnL 4.11-20. DAVIAS-BAUDRIT, PÈRE. 1 9 6 6 . Dictionnaire rhadé-français (mimeogr.). DYEN, ISIDORE. 1 9 5 3 . The Proto-Malayo-Polynesian laryngeals. Baltimore. . 1965. A lexicostatistical classification of the Austronesian languages. IUPAL, memoir 19, supplement to IJAL 31. AYMONIER, ÉTIENNE.

1899.

210

ISIDORE DYEN

K. 1890. Sprachvergleichende Untersuchung des Wörterschatzes der Tschamsprache. SbBAW, 322-456. NICOLLE. 1 9 4 2 ? Fran?ais-jaray et jaray-fran?ais. NIEMANN, G . K . 1891. Bijdrage tot de kennis der verhouding van het Tjam tot de talen van Indonesie. Bijd. 40.27-44. PITTMAN, RICHARD S. 1959. Jarai as a member of the Malayo-Polynesian family of languages. Asian Culture 1.59-67. SCHMIDT, WILHELM. 1926. Sprachfamilien und Sprachenkreise der Erde. Heidelberg. THOMAS, DAVID. 1968. Vietnam minority languages, (mimeogr.). THOMAS, DAVID, and ALAN HEALEY. 1962. Some Philippine language subgroupings: a lexicostatistical study. AnL 4 . 2 2 - 3 3 . THOMAS, DOROTHY M. 1963. Proto-Malayo-Polynesian reflexes in Rade, Jarai, and Chru. SIL 17.59-75.

HIMLY,

Note (see p. 207): Again, like Mrs. Blood, Pittman finds a change of initial *h to J. m in instances like *hasem, J. mesam 'sour', where however the initial m is undoubtedly the residue of an earlier prefix.

MALAGASY*

ISIDORE DYEN

Malagasy (Mlg.) is the major language of Madagascar and the standard dialect, Merina, is the official language of the Malagasy Republic. In 1951 it was spoken by about 6,500,000, according to the 1967 census. It has a number of named dialects. There is according to Dahl (1951:5f.) a western group consisting of Sakalâva (Sak.), Vézo, Mahafâly, Tanalâna, and Tandroy on the west coast and also Bâra Imamôno. There is an eastern group consisting of Tanosy, Taimanambôndro, Taisâka, Sahâfatra, Taifâsy, Taimoro, Tambahoaka, and Betsimisâraka on the east coast, Sihânaka, Bezanozâno, and Tanâla on the eastern watershed, and Bâra, Betsiléo, Merina, and Tsimihéty on the high plateaus. Tankârana in the north is a transitional dialect between the eastern and western groups and so is to some extent Tanosy in the south. There seems to be no question in regard to mutual intelligibility. Van der Tuuk in his "Outlines of a grammar of the Malagasy language" (1865:427) says that 'varieties of dialect exist ... but are not so numerous that people residing in different parts cannot understand each other, some practice enabling them to sustain a conversation'. Dahl (1951:6) says that 'les différences dialectales ne sont nulle part très grandes. Les individus parlant des dialectes différents arrivent toujours à se comprendre'. The standard dialect, Merina, is the dialect of the area of Tananarive, the capital of the Malagasy Republic. It is also called Hova (as consistently by Dempwolff) and Ambaniândro. The first contact by literate foreigners was that of Moslems about the middle of the twelfth century on the east coast. The Taimoros learned writing from them and continue using an Arabic orthography to this day. The current alphabet is Latin and is believed to have been constructed by the Welshman, David Jones, who arrived in 1820 as an early missionary of the London Missionary Society. His orthography was adopted in 1823. O.C. Dahl in his "Étude de phonologie et de phonétique malgaches" (1952) describes the articulatory and acoustic phonetics of Merina and of other dialects by * This paper represents partial results obtained with the support of the National Science Foundation by Grant No. GS-1468. I wish to express here my appreciation of the suggestions and additions made by L.M.S. Slawecki.

PRINCIPAL

DIALECTS

OF

MALAGASY

213

MALAGASY

the use of glossograms and kymograph tracings. In this he continues and adds to the work of Rousselot "Phonétique malgache" (1913). Published work on the Malagasy language has been otherwise mainly traditional in character, the most useful products being Malzac's Grammaire malgache (3rd ed. 1950) and the fine dictionary that Abinal and he published, Dictionnaire malgachefrançais (hereafter referred to as AM). J. Richardson's Malagasy-English dictionary (1885) is of equal merit and has recently been republished; it is prefaced by a short grammatical treatment. Some of the dialects are represented by dictionaries : H. Dubois' Essai de dictionnaire betsileo (1917) and R. Decary's Lexique français-antandroy (1928). In both of these the dialect words are accompanied by the Merina equivalent. Other works of interest are Gabriel Ferrand's Essai de grammaire malgache (1903), and Jacques Faublée's Introduction au malgache (1946). In addition there is G. Cousins' Gramera malagasy (1960; in Malagasy) and the school-text, R. Rejemisa-Raolison's Grammaire malgache (1966). The modern study of the grammar has just begun with the appearance of Catherine J. Garvey's Malagasy introductory course (1964). However her treatment is necessarily hampered by the need for an uncomplicated presentation suited to her pedagogical purposes. Another more extensive, but less scientifically oriented practical book is E.L. Stark's Malagasy without moans (1967). In his Malgache et Maanjan (1951) Dahl treats many points of grammar as he develops his argument that the closest relationship of Malagasy is with Maanyan, a language of southeastern Borneo. A presentation of many of the facts of the grammar is given by V. D. Arakin in his Maljgasskiy Yazik (1963) and in Arakin's appendix to Lev Korneev and Frederik Rakotoson's Diksionary Malagasy-Rosiana = Maljgassko-russkiy slovarj (1966). The latter is the most recent dictionary, but is not as complete or as satisfactory for many purposes as Abinal and Malzac's, though undoubtedly more modern. A completely untrustworthy treatment of words of Sanskrit origin in Malagasy is found in Dama-Ntsoha's Les mots dérivés des apports sanscrits (1951). PHONOLOGY

According to Dahl (1951:32f.) the phonological system of Malagasy is as follows (with some modification) : Vowels: i e a o u

Diphthongs: ai au ui voiceless

labial dentalveolar alveolar alveolar retroflex velar

P t ts tr k

Consonants:

Stops voiced prenasalized b d dz dr g

mb nd ndz ndr qg

nasal m n

D

Spirants voiceless

voiced

V

f

1 z r

s h

214

ISIDORE DYEN

rj is in complementary distribution with n and is thus not distinct from it. The phonemes v a n d / are labiodental. The alveolar and retroflex alveolar stops have affricate release. The prenasalized stops occur initially as well as medially. Although there are orthographic initial mp and nt, the corresponding pronunciation lacks a nasal; there are no voiceless counterparts to the voiced prenasalized consonants listed in the chart. Orthography In the orthography it is customary to write j for dz, nj for ndz, n for ij (see above), o for u, and y for i final in a word. The diphthongs are written respectively as follows: ai medially, ay in final position; ao; oi medially, and oy in final position. Since both Malzac and Garvey find it convenient to use the standard orthography, I shall follow that practice here. Garvey indicates that only in a few particular cases, except for the differences in forms of different tempo, does the orthography fail to represent the phonemes. It should be noted however that among other things Garvey writes ianao 'thou', ianereo 'ye' for Malzac's hianao, hianareo respectively, the latter being obsolescent, and indicates that isika 'we (inc.)' is pronounced as though it were written itsika (Garvey 1964:19), and that words written with clusters of nasal and voiceless stops lack the nasal in pronunciation in normal speech, so that e.g. the written tompoko is generally pronounced topoko, though the nasal may be heard in deliberate speech. Although orthographically there are such initial clusters as well, their nasal never appears in pronunciation. Accent There is a distinctive accent: tânana 'hand', tanâna 'village'; milâika 'is dancing solemnly', milaikâ 'dance solemnly!'. It has been described as a 'longueur vocalique' by Joseph Verguin in his "L'accentuation en malgache-mérina et en malais" (1955: 525, 528), 'entraînant une augmentation de force qui n'est en elle-même nullement caractéristique.' The accent is limited to occurrence on one of the last three syllables of a word: manasâ 'wash!', manâsa 'is washing', sâsatra 'tired'. One can refer to the words as being oxytone if they have an accent on the ultima, paroxytone if the accent is on the penult, and proparoxytone if it is on the antepenult. All (or nearly all) proparoxytones end in one of the syllables ka, tra, na, ko, tro, ny : bétsaka 'numerous', zâvatra 'thing', tânana 'hand', tânako 'my hand', zâvatro 'my thing', zâvany 'his thing'. The final a and the consonant preceding it enter into special alternations and so we use the term a-proparoxytones for those ending in a. The a-proparoxytones include words with ai or ao in the antepenult with accent on the first element: kâikitra 'bite', tâolana 'bone'. Interestingly enough the sequence oi, called a diphthong by Dahl, has not been found in this position : cf. tôina 'response to

MALAGASY

215

a call', fanoinana 'action of responding'. Furthermore Malzac indicates that the sequences ai and ao formed by the passive forming prefix a do not form a diphthong with the i or o that follows: aidina 'be made to go down' (idina 'descend'), aôrina 'be built' (orina 'build'). The difficulty that these words offer is that there now appears to be little reason to regard even ai and ao as diphthongs, except the fact that not to do so above will produce words accented on the fourth syllable from the end. In the following discussion however such words will be regarded as a-proparoxytones. Vowel variation Voicelessness of vowels is apparently very common in Merina and other Malagasy dialects. The vowels y, o or a are pronounced 'very faintly or whispered' (Garvey 1964:2) if unstressed following a consonant at the end of a word. Furthermore within a word the vowels i and o, when unstressed between consonants excepting h, are similarly very faint or even whispered. Malzac reports [o] not only in the interjection written ô but as the pronunciation of the orthographic diphthong ao in a number of French loanwords: baoty 'boot', laodivy 'brandy'. Ferrand states (1903:10) unequivocally 'ao se prononce comme Vo français dans hôte'. Dahl says (1952:161): 'En mérina l'assimilation réciproque des deux éléments de ao aboutit quelquefois en la voyelle simple o ouvert. C'est surtout lorsque l'accent est affaibli mais, à titre d'exception, même lorsque le phonème porte l'accent principal.' Here Dahl seems clearly to be referring to the pronunciation of ao under any circumstances, not limited to occurrence in French loanwords. Garvey (6) however says that ao corresponds in deliberate speech to the final sound of English 'endow' [au], and in normal speech to that of English 'below' [ou]. However her description of non-final ao more closely approximates Dahl's description, for she speaks of its being similar to 'the English sound in "owe" except the lips and mouth are a little more open and relaxed throughout the duration of the sound' (3.12). Dahl says that in Betsimisaraka ao is [o] everywhere and ai has fallen together with e. It is interesting to note that Malzac describes the pronunciation of the first part of non-final ai as being [e] most generally, [a] appearing only when final (i.e. ay). In the other dialects these diphthongs are maintained. Alternations There are few automatic alternations in Merina. An example is the contraction of unstressed like vowels in sequence: Mbavereno. 'Please repeat' (mba + avereno). Another is e in normal speech for the sequence i followed by a in deliberate speech : de 'then' for deliberate dia. There is however a great variety of alternations that are conditioned by the morphemes concerned. Alternations are of two types : initial and final. Initial alternations

216

ISIDORE DYEN

affect the first consonant of a base, whereas final alternations affect the end of the base and then often more than one of the phonemes. There are two major types of initial alternation. In the first type the initial alternation affects the initials of bases in combination with certain active prefixes -an-, -ampan-, -ifan-. We will call it here the N-alternation. The N-alternations will be exemplified by citing the present forms of the actives with -an- and post-pausal forms of the base. Alternations involve the initials of all bases except those with initial vowel, d, g,j, and a few bases with initial b. The nasal of the prefix also shows the alternation n ~ m, the second alternation appearing if the base has an initial labial, whether present or not. The alternations are as follows: 1) h, k, n, s, t, ts, b,f, m,p, v ~ 0 : hantona, manantona 'hang'; kolikoly, manolikoly 'cajole'; nenina, manenina 'repent'; sasa, manasa 'wash'; toro, manoro 'point to'; tsentsina, manentsina 'plug'; b&bo, mamabo 'capture'; fantsika, mamantsika 'nail'; mosavy, mamosavy 'bewitch'; potraka, mamotraka 'throw down'; velona, mamelona 'nurse, raise'. 2) h ~ g: hataka, mangataka 'request'. 3) l~d: lany, mandany 'spend'. 4) r~dr: resy, mandresy 'win'. 5) z ~ j: zaitra, manjaitra 'sew'. Examples of non-alternating initials are: adina, manadina 'examine'; didy, mandidy 'order'; gina, mangina 'keep silent'; jono, manjono 'fish with bait'. The second type of initial alternation involves a spirant as one member which alternates with the non-nasal stop of the same voicing: i.e. f ~ p , v ~ b, I ~ d, s ~ ts, z~j,r ~ dr,h~k. We will call this the spirant alternation. The stop alternant appears typically in the following member of a compound or doubling, but under certain conditions the spirant alternant appears there. The base alternant that appears as the following member of a compound or doubling is called the posterior combining alternant. The spirant alternant alone appears postpausally and in other positions and will be referred to as the postpausal alternant. There are two major types of final alternation. The first type concerns the alternations appearing in prior elements of (what I shall here call) close compounds and doublings; the base alternants that appear in this position are called the prior combining alternants. The second major type of final alternation is that of different forms of a root and is called root alternation. There are other types also such as that in the active affixes illustrated above. Prior combining alternation is of two types. The first type involves bases ending in a vowel. The second affects chiefly cr-proparoxytones, but also a few paroxytones that show the same alternation. Since in the latter case the alternation is typically between two alternants differing by at least a final vowel, the bases showing this alternation are called apocopating bases. In a close compound or doubling, the prior and posterior members appear in their combining alternants whenever this varies from the alternant (prepausal or postpausal)

MALAGASY

217

relative to pause. The posterior member is in its posterior combining alternant form unless it follows the prior member directly and the form of the latter is prepausal. As the prior member of a close compound a base ending prepausally with vowel is linked to a posterior member by a nasal n ~ m, with the alternant m appearing before a labial: (rano 'water' + afero 'gall-bladder') rano-n-afero 'gall', (loha 'head' + piso 'cat') loha-m-piso 'cat's head', (volo 'hair' + vava 'mouth') volo-m-bava 'mustache', (reny 'mother' + lalana 'road') reni-n-dalana 'highway'. This linking nasal is lacking in a doubling and the posterior combining alternant is the postpausal alternant: lazo 'withered', lazolazo 'slightly withered'. The longer alternant of apocopating bases is prepausal; shorter alternants appear in close compounds. Alternants lacking only the final vowel occur as the prior members if the posterior member has initial vowel: (kotroka 'thunder' + orana 'rain') kotrok-orana'thunder presaging rain', (foitra 'navel' + ampinga 'shield') foitr-ampinga 'shield-boss', (taolana 'bone' + orona 'nose') taolan-orona 'nasal cartilage'. Similarly with doublings: (antoka 'nod') antok-dntoka 'nod slightly', (ampatra 'stretch out') ampatr-ampatra 'often stretched out', (alona 'wave') alon-alona 'small waves'. However the prior combining alternant of an apocopating base is shorter by the last vowel and a preceding tr or k if the posterior member has an initial consonant: (hoditra 'skin' + maso 'eye') hudi-maso 'eyelid', (lelaka 'lick' + molotra 'mouth') lelamolotra 'lick the lips', (tandroka 'horn' + vy 'iron') tandro-by 'iron cap for horns', (efatra 'four' + zoro 'corner') efa-dzdro 'quandrangle', ( f i r a k a 'bad' + fotsy 'white') fira-potsy 'tin', (vadika 'reverse side' + lamba 'clothing') vadi-damba 'innerside of clothing', (safotra 'cover' + rano 'water') safo-drano 'flood', (hoditra 'skin' + hazo 'wood') hodi-kdzo 'bark'. So also with doublings: (bitsika 'whisper') bitsi-bitsika 'whisper frequently', (tahotra 'fear') taho-tahotra 'be slightly fearful', {velatra 'spread out') velabelatra 'somewhat spread out', (fotitra 'turn around') foti-potitra 'often turned around', {hindzaka 'jump') hindza-kindzaka 'jumped often', (sesika 'stuff') sesi-tsesika 'often stuffed'. If the consonant preceding the last vowel of an apocopating base is n, the n alternates with m before a labial: (randrana 'braid' + volo 'hair') randram-bolo 'curl', (hanina 'eat' + lohy 'man') hanin-dahy 'intoxicating drinks', (lalana 'way' + rd 'blood') lalandra 'vein', (fahana 'food' + basy 'gun') faham-basy 'cartridge'. So likewise with doublings : (velona 'live') velom-belona 'somewhat lively', (lalina 'deep') lalin-dalina 'somewhat deep', (zavona 'cloud') zavon-javona 'somewhat cloudy', (hodina 'roll') hodinkodina 'often rolling'. Root alternants The second major type of final alternation concerns the alternant forms of the root. Many roots exhibit two forms, one that appears before pause and the other that appears before certain suffixes: the active imperative suffix -a ~ 0 and the passive suffixes -ina ~ -na, and -ana ~ -na. These are most conveniently presented in the

218

ISIDORE DYEN

context of the relation between the root as it appears in its prepausal form and the passive forms. Malzac finds the formation of passives with a suffix quite puzzling. He says (1950: 59) 'Il n'est pas possible d'établir des règles fixes permettant de passer infailliblement d'une racine au participe à suffixe correspondant. Soit le suffixe lui-même, soit surtout les modifications subies par la racine avant de s'adjoindre le suffixe, échappent à toute classification rigoureuse.' One can however obtain a large class if one starts from the non-prepausal alternant of the root, i.e. the form it has with the passive suffix and if one then derives the prepausal form from it instead of the reverse as is commonly done. For the most part in the following I shall cite AM's heading word to exemplify the prepausal form. The passive suffix is regarded as having the form -na after a stressed penult vowel (e.g. teré-na 'squeezed') and -ana or -ina if the passive is proparoxytone (i.e. has antepenultimate accent: e.g. tapâh-ina 'what is cut', tdpaka 'cut'). The prepausal forms that are members of the large or regular class have retracted accent relative to the non-prepausal form wherever possible and have i (orthographically y y final) for the last e of the non-prepausal form, if any. One can formulate rules like the following: 1) If the non-prepausal alternant ends in a vowel, the prepausal alternants are regular and exemplified as follows : vahâ-na 'be untied', vdha 'untie' ; iri-na 'be desired', iry 'desire'; angolé-na 'be deceived', angoly 'deceive'; lazd-ina 'be said', lâza 'say'; ari-ana 'be thrown', dry 'throw'; torotorô-ina 'be crushed', torotôro 'crush'; solô-ana 'be replaced', sôlo 'replace'. 2) If the non-prepausal alternant ends in a consonant, the prepausal alternant differs from it in that : (i) a final s, z, or v is dropped : lâv-ina 'be denied', Id 'deny' ; rdis-ina 'be taken', ray 'take' ; gidz-ana 'be squeezed', gia 'squeeze' ; todv-ina 'be submitted to', tôa 'submit' ; fohdz-ina 'be awakened', fôha 'wake'; nofis-ina 'be dreamed', nôfy 'dream'; tohiz-ina 'be continued', tôhy 'continue'; refés-ina 'be measured', réfy 'measure'; tetéz-ina 'be passed over', téty 'pass over';fonôs-ina 'be enveloped', fôno 'envelope'; tsinjôv-ina 'be looked at from above', tsinjo 'look at from above'; arôvana 'be protected', dro 'protect' ; tovôz-ina 'be drawn (water)', tôvo 'draw (water)'. (ii) a final h,/is replaced by ka: tsdoh-ina 'be adored', tsâoka 'adore'; rdof-ina 'be picked up', rdoka 'pick up' ; tapâh-ina 'be cut', tdpaka 'cut' ; hohôf-ana 'be lain on', hôhoka 'lie down'; atréh-ina 'be faced', dtrika 'face'. (iii) a final r, t is replaced by ira: tsôr-ina 'be simplified', tsôtra 'simplify'; rdot-ina 'be picked', râotra 'pick' ; sorât-ina 'be written', sôratra 'write' ; avôt-ana 'be redeemed', âvotra 'redeem' ; atér-ina 'be brought', dtitra 'bring' ; kaikér-ina 'be bitten', kâikitra 'bite'. (iv) a nasal is replaced by na: tdnana 'be held', tdna 'hold'; sitrdn-ina 'be cured', sitrana 'cure' ; ambdn-ana 'be threatened', dmbana 'threaten' ; ambén-ana 'be watched', dmbina 'watch'; andrdm-ana 'be tasted', dndrana 'taste'; tandrém-ana 'be paid attention', tândrina 'pay attention to' ; velom-ina 'be given life', vélona 'live'.

MALAGASY

219

There are of course some alternations that do not fit the above scheme. A few are given here: omé-na 'be given', orné 'give'; lalaovi-na 'be played with', laláo 'play'; tetév-ana 'be dripped into', teté 'drip'; vovóz-ina 'be barked at', vovó 'bark'; voizi-na 'be rowed', vóy 'row'; dokáf-ana 'be flattered', dóka 'flatter'; tratràr-ìna 'be pursued', trátra 'caught'; tadiáv-ina 'be sought', tády 'seek'; endás-ina, endázina 'be roasted', éndy 'roast'; fafáz-ana 'be sowed (field)', f á f y 'sow'; sokáf-ana 'be opened', sókatra 'open'. Other forms than the passive show the non-prepausal alternant. The imperative of the passive adds o ~ y to the non-prepausal alternant. The alternant y depends on the presence of o in the preceding form, for otherwise the alternant is -o : irí-o, irina 'be desired'; lotó-y, lotó-ina 'be dirtied'; roáh-y, roáh-ina 'be hunted'; ario, ariana 'be thrown'; sasá-o, sasá-na 'be washed'. The imperative of actives likewise shows the non-prepausal form before their suffix -a: maniría 'desire!', iri-na 'be desired'; mandotó-a 'dirty!', lotó-ina 'be dirtied', etc. If the pre-pausal form ends in a, the suffix -a coalesces with it: manasá 'wash', sasá-na 'be washed'. GRAMMAR

Predications The subject normally follows the predicator : tsara ny andrò, '(lit. good ny day) The weather is nice.' However the reverse order also occurs: ny andrò tsara. 'The day is nice.' In a more explicit form of the reverse order the particle dia is inserted: ny andrò dia tsara. Garvey permits a third order called emphatic with the particle no. She contrasts: Nipetraka tao ny namako. '(lit. lived there ny friend-my) My friend lived there.' Ny namako no nipetraka tao. '(lit. ny friend-my no lived there) It was my friend who lived there.' She calls ny namako an emphatic subject, presumably on the grounds that (1964:82) 'the choice of pattern P S or (S) no P(S) with subject before or after P, is a matter of emphasis.' One can consider the possibility that ny namako is rather the predicator with the no-phrase as subject. This would be in keeping with the general rule that the predicate is normally first, and that dia is (probably) not possible if this is regarded as reverse order. There is also an interesting interchange of no and ny. Thus in the normal order: Ny vehivávy no manasa lamba, '(lit. ny woman no wash laundry) It is the women who do the laundry.' For this there appears to be no normal order for I have not yet found a no phrase in initial predicate position. What is interesting is that a reverse order does occur with ny instead of no and obligatory dia: ny manasa lamba dia ny vehivavy. 'The ones who do the laundry are the women.' It is furthermore interesting that the following sentence also occurs: Vehivavy no

220

ISIDORE DYEN

manasa lamba. 'It is women that do the laundry.' Thus with a no phrase as a subject at any rate there is a choice of a predicator with or without ny according to definiteness in equational clauses. It has proven difficult up until now to ascertain the limits of such contrasts when one considers the different possible subjects. A subject is a pronoun (personal or demonstrative), a demonstrative phrase, an /-phrase, a «^-phrase, or a no-phrase. A pronoun as subject is exemplified by: Mangetaheta izy. '(thirsty he) He is thirsty'; Bokiko ity. '(lit. book-my this) This is my book.' A demonstrative phrase consists of a demonstrative (usually repeated before and after the head expression) and a head expression, usually a noun (and its attributes): Mahafinaritra ity boky ity. '(lit. pleasant this book this) This book is pleasant.' An /-phrase consists of the particle i or ry with a proper noun: Miverena i Rabe. 'Rabe is returning'; A vy ry Rabe. 'The Rabes (i.e. Rabe and those associated with him) are coming.' This particle i is also used with certain familiar nouns: Aiza i neniko? '(lit. where mother-my) Where is my mama?' A nj-phrase consists of the particle ny and a common noun: Avy ny olona. '(iscoming ny man) The man is coming.' The ny-phrase is not definite here, i.e. it does not necessarily imply that the hearer knows the object: Mendrika hovalian-tsoa ny mpianatra mazoto. '(lit. deserve will-be-rewarded student diligent) A diligent student deserves to be rewarded.' A no-phrase consists of the particle no and an expression that can be a predicate, as was exemplified above. Thus far a no-phrase has occurred as subject, but has not been observed to occur as predicate, either at the beginning of a sentence or after dia. Consequently it is possible to regard a n_y-phrase, when it is a preceding subject, as being a form associated with the no-phrase, i.e. the alternation ny ~ no with manasa lamba 'washes laundry' depends on whether the predicator precedes or follows: in normal order Ny vehivavy no manasa lamba; and in reverse order ny manasa lamba dia ny vehivavy. 'The ones who do the laundry are the women.' Thus there is reason to consider the possibility that there are two particles: ny1 that is constructed with nouns and does not alternate with no; and ny2 that constructs with predicative expressions and alternates with no. The difference between no and nyx is illustrated by the following: Tsara no mampandeha azy. '(lit. good no making-leave him) The one making him leave is good.' Tsara ny mampandeha azy. '(lit. good ny make-leave him) It is good to make him leave.' In the latter instance mampandeha is a noun, though it retains enough of its verbal character to take a direct complement. The casual predication Thus far the predications have been of a narrative type with a verb as predicator or an

MALAGASY

221

equational type, usually with a noun or «y-expression as predicator. There is a third type which I will call casual', I have also called it existential. It occurs with the verbs misy 'exist' and be 'be many' : misy taratasy ato '(lit. exist paper here) There is some paper here'; be mpianatra aty '(lit. many student here) There are many students here.' Some attributive expressions Most expressions that function as subject in narrative or equational predications enter, sometimes with modifications, into two other constructions as attributes: as genitive (Malzac 'régime indirect') and as oblique (Malzac 'régime direct'). As genitive it appears chiefly in constructions in which it has meanings such as possessor, partitive, agent, etc. As oblique it appears most commonly as direct complement of a verb. A genitive expression follows its head. Certain differences in form accompany the appearance of expressions as genitive attribute. A personal pronoun has an enclitic form : e.g. trano-ko 'my house' (aho 'I'). It is not clear whether a demonstrative pronoun can serve as a genitive attribute. A proper noun expression that has the particle i has as a genitive expression (orthographically) the same form as when subject. However a preceding word that is prepausally proparoxytone (i.e. with a stress on the antepenult) appears without its last vowel (orthographically replaced by an apostrophe) : ny zanak' i Koto \zanaka 'child') Koto's child' ny mpiasan' i Koto \mpiasana 'servant') Koto's servant' nalefan' i Koto \nalefana 'was sent') sent by Koto' A preceding word that is not a proparoxytone appears with an additional n (followed orthographically by an apostrophe) : ny vadin' i Koto '(vady 'wife') Koto's wife' However if the particle is ry or there is no particle, as is permitted before a name beginning with ra-, then after n the alternant dr of r appears (and a hyphen instead of the apostrophe) : ny mpiasan-dry Rabe 'the Rabes' servant' ny mpiasan-dRabe 'Rabe's servant' nalefan-dRabe 'sent by Rabe' A (definite) common noun expression as genitive has the same form as when subject (thus consisting of ny plus noun) and the preceding noun shows the same forms as before ry (except that the apostrophe is written) : ny mpiasan' ny namako 'my friend's servant' ny vadin' ny namako 'my friend's wife'

222

ISIDORE DYEN

ny zanaky ny namako 'my friend's child' ny zavatry ny namako 'my friend's thing' In these cases a different analysis is possible than that indicated by the orthography. One could analyze ny vadin' i Koto as ny vady ny Koto ; ni vadin ny namako as ny vady nny namako; and ny zanaky ny namako as ny zanak' iny namako. In this analysis the variations in form would be assigned to the particle except that apocopated forms would be permitted to occur not only before the alternant i (of i ~ ny with personal nouns, but also before the alternants ny and iny (of ny ~ nny ~ iny with common nouns). Malzac regards the second part of an instance like zana-bitro 'rabbit young' (zanaka 'child', bitro 'rabbit') as an example of a 'régime indirect indéterminé' which should then be the same thing as an indefinite common noun expression in genitive function. However one could consider regarding it as an instance of a compound. He does the same for instances like leon-toaka 'drunk with rum' (leo, leony 'drunk', tôaka 'rum'), but these too can be regarded as compounds and dissociated from instances of a following genitive like sasatry ny dia 'tired by the trip' (sâsatra 'tired', dia 'trip'). Furthermore there are no instances of a 'régime indirect indéterminé' as agent despite the general rule that a genitive expression otherwise appears both as possessor and as agent. There are a number of problems. There would then appear to be two types of compounds exemplified by trano vato (lit. house stone) 'stone house' (if a non-proparoxytone is followed by a noun meaning 'material', there is simple juxtaposition [Malzac 1950:101]) and tranon-kazo '(lit. house wood) wood-shed, shed for wood'). Formally the trano vato type could be called an open compound and the tranon-kazo type a close compound. However the open type is uncommon. Dahl (1951:136) also notes satroka miaramila 'soldier's cap' beside satromiaramila, but does not indicate the difference in meaning, if any. Malzac (1950:101) also reports the interesting lanja na voamena 'gram's weight' beside lanjam-boamena 'idem' and Dahl (1951:136) reports tranona hazo 'wood-shed'. The oblique occurs as direct object complement ( = Malzac's régime direct) with an active or passive and with preposition. A personal pronoun has a special form : e.g. ahy 'me' {aho 'I'), anao 'thee' (hianao 'thou') : miandry anao aho '(lit. waiting thee I) I am waiting for you'; omeko anao ity '(lit. given-by-me you this) I gave you this.' A demonstrative is preceded by an': aza mamono an' io '(lit. don't strike art that) Do not strike that.' So also a demonstrative phrase: misotro an' io kafe io izy. '(lit. drinking an' that coffee that he) He's drinking that coffee.' Similarly a proper noun expression is preceded by an' or an- : miandry an' i Sahondry. 'I am waiting for Sahondry.' miandry an-dRabenja. 'I'm waiting for Rabenja.' miandry an-dry Rajaona. 'I'm waiting for the Rajaonas.' An oblique common noun expression is definite if ny precedes and otherwise is indefinite:

MALAGASY

223

nividy ny boky aho. '(lit. bought ny book I) I bought the book.' nividy boky aho. '(lit. bought book I) I bought a book.' Some verbs are constructed with two complements. Only one of the complements has appeared as an indefinite common noun expression: manome vola ny mahantra izy. '(lit gives money the poor he) He gives money to the poor'; mampianatra gramara ny mpianatra aho. '(teach grammar ny student I) I teach grammar to the students.' Under certain conditions the indefinite complement can be the second one: nanome ahy vola be izy. '(lit. gave me money much he) He gave me a lot of money.' It is not clear whether both complements can be indefinite. It is furthermore not entirely clear which types of complements can be indefinite. There are however the following examples: Nameno vary ny kitapony izy. '(lit. filled rice his-bag he) He filled his bag with rice.' Natafy azy jaky izy. '(lit. dressed him purple he) He dressed him in purple.' There is a preposition with a wide range of meaning which is given the absolute form amy and is constructed with the genitive: Hilaza izany aminao aho '(lit. will tell that amy-thee I) 'I will tell you that.' It is different from other prepositions in that, like the demonstrative adverbs, it is inflected with a i-prefix for past: Milaza izany taminao aho. 'I told you that.' [Cf. Eto an-tranoko izy. '(lit. here in-my-room he) He's here in my room'; Teto an-tranoko izy fahiny. '(lit. was-here in-my-room he before) He was in here in my room before.'] However amy is not marked for past if it is in a construction with a past-marked adverb: Avy tao amin' ny restauranta aho. '(lit. come there-past amy restaurant I) I came from that restaurant (there).' There is an interchange between oblique forms of the pronoun and the constructions with amy: e.g. Nambarako azy (or taminy) ny zavatra. '(lit. reported-by-me to-him the thing) I have reported the matter to him.' Nampiseho ahy sary mahafinaritra izy. '(lit. showed me pictures pleasant he)' or Nampiseho sary mahafinaritra tamiko izy. '(lit. showed he pictures pleasant amyme he) He showed me some pretty pictures.' Nanolotra azy voninkazo aho. '(lit. offered him flower I)' or Nanolotra voninkazo taminy aho. '(lit. offered flower amy-him I) I offered him some flowers.' It then becomes difficult without further information to analyze a sentence like the following that appears in Malzac: Hampanalavitra anao an' Andriamanitra ny ataonao. '(lit. will-make-be-far thee God ny done-by-you) What you do estranges you from God.' for one cannot tell whether the anao would or would not interchange with an amyconstruction. In the above alternation between oblique and am^-construction the oblique appears

224

ISIDORE DYEN

first after the verb and the amy-construction appears after the oblique. Some amyconstructions follow the verb immediately even if there is another oblique following: Nilaza taminy ny zavatra reko aho. '(lit. told past-awry ny thing heard-by-me I) I told him what I heard.' There are a few instances of what appears to be a pronominal oblique after what seems to be a complement. One should perhaps consider the sequence of verb and complement as a compound (or idiom): Tsy namintana andro ahy hianareo. '(lit. not miss time me you) You did not miss our rendezvous.' Hamaly soa anao aho. '(lit. will-return reward thee I) I will reward you.' In a very striking agreement with Philippine languages the independent possessive is identical in form with the oblique. One can compare Tagalog qa:kin 'mine, me'. The form ahy which appears as the oblique also appears as the independent possessive: Ahy io. 'That is mine'; Tsara ny ahy. 'Mine is good.' Similarly there are the following examples: An-dRakoto io. 'That is Rakoto's.' An i Koto io. 'That is Koto's.' An ny vadiko io. 'That is my wife's.' The actives There are a number of different types of active in Malagasy. They are characterized by an inflection m- 'present', n- 'past', h- 'future', and an imperative with the suffix -a. The root is in its prepausal form except in the imperative. Malzac lists the following as active prefixes (1950:16): m, ma, man, mana, manka, mampa, mampan, mampana, mifan, mifana, mifanka, mampifan, mi, mifampi, maha, mampaha, miha. The m- that appears in these prefixes as they are usually cited is the present tense prefix. Thus it is possible to regard the form cited less the tense-prefix m- as the affix. Some of the active prefixes have a specifiable meaning: aha 'be able', e.g. mahatohitra 'able to resist'; iha 'gradually become', e.g. miha-tsára 'gradually become better'; ampa, ampan, ampana, ampanka, ampaha, ampi, ampiha 'causative', e.g. mampa-tóry 'put to sleep' (tóry 'sleep'), mampam-bóly 'have planted' (vóly 'plant'), mampana-lávitra 'estrange' (lávitra 'far'), mampanka-láza 'cause to exalt' (laza 'speak'), mampaha-tézitra 'cause to anger' (tézitra 'become angry'), mampi-réhitra 'cause to light' (réhitra 'to light'), mampiha-rátsy 'cause to become bad' (rátsy 'bad'); if an, ifana, ifanka 'reciprocal', e.g. mifanóro 'show themselves to each other' (toro 'show'), mifanarátsy 'speak ill of each other', mifankahála 'detest each other'; ampifan, ampifanka 'reciprocal causative', e.g. mampifanátrika 'make face each other' (átrika 'face'), mampifanka-tia 'cause to love one another' (tía 'like').

MALAGASY

225

The meanings of the other formants are less easily specified. These are 0 - ~ a-, e.g. m-áka 'take', má-tahotra 'fear'; í-, e.g. mi-tády 'seek'; an-, e.g. man-iry 'desire'. The formants ending in n- are usually associated with alternation affecting the first consonant of a following morpheme other than a vowel or d, g, j. For the sake of simplicity the illustrations have been selected only from those with an-: 1) The initial h of a root alternates with 0 in some cases, or g in others: manántona 'suspend' (hántona 'hang'), mangátaka 'request' Qiátaka 'request'). 2) k, n, s, t, is alternate with 0 : manolikóly 'cajole' (kolikóly 'cajolery'), manénina 'repent' (nénina 'repenting'), manása 'wash' (sása 'washing'), manóro 'point to' (toro 'pointing'), manéntsina 'cork' (tséntsina 'plug'). 3) / alternates with d: mandány 'consume' (látiy 'used up'). 4) r alternates with dr: mandrésy 'conquer' (résy 'beaten'). 5) z alternates with j: manjáitra 'sew' (záitra 'needlework'). The n of the prefix itself alternates with m before a root that begins with f,p, and the latter alternate with 0 : mamantsika 'nail' (fántsika 'nail'), mamotraka 'throw down' (pótraka 'fallen'). Similarly with b and v, but here in some cases an initial b remains and an initial v is replaced by b: mamábo 'capture' (bábo 'prisoner'), mamélona 'nourish' (velona 'living'), mambéta 'carry in hand' (béta 'carrying in hand'), mambóly 'plant' (vóly 'plantation'). Examples of the non-alternation of d, g,j are: mandidy 'order' (didy 'order'), mangina 'remain silent' (gina 'silent'), manjóno 'fish with bait' (Jóno 'bait'). The passives A passive is constructed with a genitive as agent, though the appearance of an agent is optional. Passives are of two types: (1) those which are inflected only for the future, taking ho; and (2) those which have 0 'present', no ~ n 'past', ho ~ h 'future'; the vowelless alternants appear before vowels. Passives inflected only with ho are perhaps better called quasi-passives. The membership consists of root verbs and prefixed verbs. Root passives are exemplified by résy 'conquer': (ho) resiko izy (lit. [will] be-conquered by-me he) 'I (will) conquer him.' The imperative is active in form: resé 'be conquered'. Cf. the active imperative (Malzac 1950:53) mijeré 'look!' associated with mijéry 'is looking'. Two prefixes that appear in the formation of passives are voa-, tafa-. The first of these has the sense of 'completion', whereas the meaning of tafa- as a passive is not specifiable: Ho voavaliko rahampitso io taratasy io. '(lit. will have-been-answered by-me tomorrow this letter) I will have answered the letter tomorrow'. Tafavoako ny omby. '(lit. be made-go-out by-me ny cow) I've let the cows out.' Passives with full inflection have an imperative with the suffix o ~ y. They are of two

226

ISIDORE DYEN

types: (1) with prefix a- or (2) with suffix -ana ~ -na or -ina ~ na. The shorter alternant appears immediately after a stressed vowel. The fully inflected passives with prefix a- are exemplified by: atao 'being made' (root tao, cf. manao 'make'), n-atao 'was made', h-atao 'will be made', ataovy 'be made!'; afoy 'being yielded up' (Joy 'yielded'), n- 'past', h- 'future', afoizo 'yield-up!'. A few have a prefix an (with the same alternations as for man above): e.g. anampy 'being added', cf. manampy 'add'. Generally a root forms only one suffixed passive, either with -ina or with -ana. In a few instances both are formed with the same root with different uses. Thus the subject of aterana 'be brought' is the person to whom it is brought whereas that of aterina 'be brought' is the thing brought. In the same way a few roots have both a passive with a suffix and one with a prefix a- with different uses. Thus toroana 'be shown' has as subject the person to whom an object is shown whereas atoro has as subject the thing shown: toroy lalana aho. '(lit. be-shown-to [imperative] road I) Show me the road.' atoroy ahy ny tranonao. '(lit. be-shown me ny house-thy) Show me your house.'

The 'relative form' It seems possible to divide the actives into two groups depending on their association with a 'relative form'. Malzac lists only relative forms (1950:69f.) associable with verbs containing the prefix a ~ 0 (present m ~ ma), i- (present mi-), and an- (present man-). Provisionally we call active verbs that have a 'relative form' associated with them 'relatable verbs' and the others 'non-relatable verbs'. The relative form consists of the active prefix plus the root in non-prepausal form plus the suffix -ana ~ -na (only after a): matory 'sleep', rel. atoriana; maka 'take', rel. akana. It is not yet clear how relatability is associated with the active prefix. Thus AM list a relative form akana for maka 'take'; atahorana for matahotra 'fear', but none for m-arina 'right', ma-loto 'dirty', which may therefore be non-relatable (lending themselves to an adjectival class). Similarly AM do not cite a relative form for miha-tsara 'grow better', maha-tohitra 'be able to resist', and they too may be non-relatable. However AM fail to cite relative forms for derived forms like the causative mampaka '(lit. make get) send to get' (aka 'take') and this omission could be due to its obvious derivation. Thus one may suppose that the non-relatable actives are of the type already indicated, for we find: (a) for an-, maniry 'desire' dry 'desire'), rel. aniriana; for ana-, manaratsy 'speak ill of' (ratsy 'bad'), rel. anaratsiana; for anka-, mankahala 'hate' (hala 'hate'), rel. ankahalana; for if anka-, mifankatia 'be in love with one another' (tia 'love'), rel. ifankatiavana; (b) for /-, midera 'praise' (dera 'praise'), rel. iderana; (c) for aha, mahatahotra 'frighten' (tahotra 'fear'), rel. atahorana. For the remainder of

MALAGASY

227

the actives corresponding relative forms have not been located. Among these is the active with the prefix aha- 'be able' as in maha-tohitra 'be able to resist'. The relative form has an imperative like a passive (see 'Root alternants' above). In meaning it is somewhat more polite than other imperatives: anaovy vary aho. '(lit. please-be-made rice I) Please make rice for me.' Here anaovy is an imperative of the relative anaovana associated with manao 'make'. andraiso kely sakafo ny vahiny. '(lit. please-be-served some food ny stranger) Please serve the stranger some food.' Here andraiso is the imperative of andraisana, the relative form associated with mandray 'hand to'. Except in its imperative, the relative form appears as predicator in an independent clause in Malzac (1950:73) in a partitive meaning: nanalako ny entanao. '(taken by-me [rel. analana plus -ko, cf. manala 'take'] ny burden thy) I've taken part of your burden.' nahaverezako dimy ny ombiko. '(lit. was-lost by-me [rel. ahaverezana plus ko, cf. mahavery 'lose'] five ny cows my) I lost five of my cows.' and taking as subject the one to whom something is done: nilazako izany izy. '(lit. was-told-to-by-me [nil&zana + ko, cf. milaza 'tell'] that he) He was told that by me'. nanolorako voninkazo izy. '(lit. be-offered-to-by-me [nanolarana + ko, cf. manolotra 'offer'] flower he) He was offered flowers by me'. It is very common as the predicator in a subordinate clause introduced by no and lacking a subject. In such a clause the relative form can be associated with the 'time, place at which; the means, material by which; the cause, motive for which; the manner in which the action is done'. When there is both an agent and direct complement they are respectively semantically equivalent to actor and direct goal: time: rahampitso no hamangiako azy. '(lit. tomorrow no when-will-be-visited-byme him [hamangiana + ko, cf. mamangy 'visit']) Tomorrow is when I will visit him.' place: io tokotany io no hanorenako ny tranoko. '(lit. this site this no where-willbe-built-by-me ny house-my) This site is where I will build my house.' means: ny famaky no anaovany hazo fisaka. '(lit. ny axe no with-which-is-madeby-him [anaovana + ny, cf. manao 'make'] plank [i.e. wood flat]) It is the axe with which he made planks.' material: ny vy no anaovana ny antsy, '(lit. ny iron no of-which-is-made [see preceding] ny knife) It is (the) iron of which knives are made.' cause: izao no namoriako anareo. '(lit. this no for which-was-assembled [anoriana + ko, cf. mamory assemble'] by-me you) This is what I assembled you for.' manner: toy izao no hitondrako anareo. '(lit. like this no how-will-be-by-me treated [itondrdna + ko, cf. mitondra 'treat'] you) (Like) this is the way that I will treat you.'

228

ISIDORE DYEN

benefaction : ny zanany no amoriarC ny ray harena. '(lit. ny child-his no for-whomis-amassed [cf. vorinâ 'be collected'] by-father wealth) It is his child for whom a father amasses wealth.' With some relative forms there is no direct complement (particularly for those that lack one with the active) : roa volana no hitoérako eto. '(lit. two months no when-will-be-stayed [hitoérana + ko, cf. mitoétra 'stay'] by-me here) Two months is as long as I will stay here.' Some relative forms in a no-clause are very much like other passives in the absence of a direct complement: Ny bibilava no atahôrako indrindra '(lit. ny snake no is-feared [atahôrana + ko, cf. matâhotra 'fear'] most) The snake is what I fear most.' Compare this with: ny hasiahany no atahôrako azy. '(lit. ny cruelty-his no for-which-is-feared by-me him) It is his cruelty that I fear him for.' An even more marked difference appears in relative forms with the prefix aha. In the following the direct complement is lacking in the no-clause : Teo no nahalavôako omaly. '(lit. there no where-was-fallen [nahalavoana + ko, cf. mahalâvo 'be able to turn over, fall'] by-me yesterday) There's where I fell yesterday.' With the direct complement: ny mahazdka no nahalavdko ny ômby. '(lit. ny rope no with-which-was-turned-over by-me ny cow) It was the rope that I turned the cow over with.' Relative forms, like other verbs, also occur as attributes: io ny fantsika hanantonao ny akanjonao. '(lit. this ny nail where-will-be-hung by-thee ny clothes-thy) This is the nail for you to hang your clothes on.' COMPARATIVE STUDY

There has been important activity in the comparative field. Dempwolff used Merina, under the name Hova, taken from the dictionary of Abinal and Malzac for his work on the reconstruction of Proto-Austronesian (PAN). Otto Dahl wrote a significant article entitled "Le système phonologique du proto-malgache" (1938) at a time when he apparently had available only Dempwolff's first volume and his work was therefore independent of Dempwolff's treatment of Hova in his second volume. Although there is general agreement on developments, there are occasionally interesting divergences of opinion. The following mergers of initial and intervocalic consonants took place in my terms : *b, *w > Mlg. v *d, *D, *Z, */, *r, *Ra > Mlg. r *z, *R2, *R3, *y > Mlg. 2 *g, *k > Mlg. h *q, *h, »iî^Mlg. 0 *t, *T> Mlg. t and for Merina:

MALAGASY

229

*n, *r) > Mer. n The following phonetic changes occurred: *c > Mlg. ts and */ before *i > Mer. (and other eastern dialects) d. As for the vowels, *e became Mlg. i after the stress, otherwise Mlg. e, the other vowels persisting. The nasal clusters offer similar results. The following intervocalic mergers occurred: *nd-, *-nD, *yj- > Mlg. ndr *-Vg-, *-yk-> Mlg. tjg *-nt-, *nT- > Mlg. nt *nc-, *-ns- > Mlg. nts and the following phonetic changes occurred: *-nz- ~ Mlg. ndz The Malagasy reflexes of consonants in final position have been a source of confusion. Merina offers only three consonants that can be associated with the final consonants reconstructed by Dempwolff and then (as in most dialects) only if a vowel follows, which Dempwolff ignored. The three consonants involved are: tr (in other dialects: is), k, n. The following discussion will limit itself to the Merina phenomena. Notes referring to other dialectal phenomena can be found in Dahl (1938). Both Dahl (1951:62ff.) and Dempwolff (2.77ff.) have expressed themselves as regarding the following table as showing the regular reflexes: *-d, soratra *-g, 'child'. *-m,

*-D, *-_/', *-r, *-t > Mer. tra: e.g. *()ulej, Mer. olitra 'worm'; *surat, Mer. 'written'; *luar, Mer. loatra 'extract'. *-k > Mer. ka: e.g. *kuDug, Mer. kotroka'thunder'; *()anak, Mer. z-anaka *-n,

> M e r . na: *?enem, Mer. enina 'six', *bulan, Mer. volana 'moon'.

Furthermore *-p and *-b in this interpretation are reflected sometimes by Mer. ka, but often by Mer. tra: *ta-heyup Mer. tsioka 'breeze', tsiof-ina 'be blown over'; *tutup, Mer. tototra 'cover', totof-ana 'be covered', *kubkub, Mer. hohoka 'turn over', hohofana 'where one turns over', *taykub, Mer. takotra 'cover', takof-ana 'be covered'. It is apparent that t h e / i n the forms with suffix is regular if from *p, but not from *b where v would be expected. Dahl makes the point (1938:201) that there are no instances of forms with suffix exhibiting Mer. v from final *b. This suggests that there might have been a loss of voicing in final position and that/appears instead of the anticipated v by analogical change. Dahl attributes the development of the additional a found in proparoxytones to a Bantu substratum (1951:113), since there are, he asserts, very few Bantu languages that have final consonants. Since speakers would have difficulty in pronouncing final consonants, they could very well have introduced the additional vowel in attempting to do so. However not all PAN final consonants have a corresponding consonant in Malagasy. Thus *-s always and -/ often are simply lost: *lepas, Mer. lefa 'loose'; *paijkal,

230

ISIDORE DYEN

Mer. faka 'root'. However when there is a corresponding consonant for *-/, it is Mer. na: *gatel, Mer. hatina 'itch'. For completeness we should add that generally *R has the same effect in final position as *y: *x1apuy, Mer. afo 'fire'; *timuR, Mer. a-tsimo 'south'; *matey, Mer. maty 'die'; *()ibeR, Mer. ivy 'saliva'; *salay, Mer. saly 'dry grill'; *damaR, Mer. ramy 'white incense'. Furthermore *iR became Mer. i: *kaiR, Mer. hay 'heat'; and *aw became Mer. u: *qanjaw, Mer. andro 'day'. Thus in effect final *R, *y, *w disappear or combine with the preceding vowel. Dempwolff inferred that the Malagasy consonants were regular reflexes because of their frequency (1934-38:7.83) relative to other outcomes. This argument is useful only in the absence of other information. It is based on the general proposition that in most instances the regular reflex will be the most frequent one, but this principle can lead to error, even serious error, as it did in his interpretation of the so-called 'old speech-stratum' of Ngaju-Dayak (cf. Dyen 1956). Dahl likewise regards these consonants as regular reflexes, but does so for different reasons. He recognizes (1938:200f.) a tendency toward loss of final consonants that is manifested in the loss of final *w, */, *y, *R, *h, and *s. He points out that as a consequence the most weakly articulated consonants were lost. He reasons however that if all of the final consonants had disappeared, there would have resulted many homophones as in the Polynesian languages. He indicates that the consequence would have been that for example ana 'out of breath', anaka 'child', anana 'possession', anana 'legume', anatra 'admonition', all would have become *ana. He concludes that undoubtedly it was a reaction against this homophony that brought about the preservation of the stops and r in final position. The argument that a phonetic change could be held back by the danger of homophony is not strong. It is evident that there must be a limit to the amount of homophony since a language is governed by the need to communicate. On the other hand the avoidance of homophony is more reasonably attributed to the acts of individual speakers needing to make themselves understood. For this reason if avoidance of homophony explains the presence of these final consonants, then this hypothesis would fit better with an explanation by analogical change. But if the source of these consonants was analogical change, the regular change involved could be the loss of all final consonants. Neither Dempwolff nor Dahl offers a satisfactory explanation for the fact that there are only three final consonants represented. It is clear that under their hypotheses one must suppose that the finals were first reduced to a single final stop which became -ka, a single final non-nasal continuant which became -tra, and a single final nasal which became -na. However the correspondences to final *p and *b militate against the hypothesis of regular change. Under this hypothesis it is difficult to explain both tr and k from both *-p and *-b; see above. Of course if the consonant were restored by some analogy an explanation would be reached. But a restoration would imply that the consonant was not there! And

MALAGASY

231

interestingly enough there are instances of the absolute loss of *-p and *-b: *( )alap, Mer. ala 'take'; *( )arjap, Mer. ana 'out of breath'; *dibdib, Mlg. R.* tritry 'suck'; *( )urub, Mer. oro 'burn'. It is now reasonable to posit the loss of *-p, *-b in pre-Malagasy or early Malagasy, presumably after their having merged as voiceless phonemes. There is other evidence that could be explained by positing the loss of additional final consonants. There are instances of the loss of *-k in the following: *ka(m)pak, Mer. kapa 'cut'; *zwjkuk, Mer. joko 'bow'; *repuk, Mer. refo 'fragility'. For final *-? a reasonable example is *( )arjkat, Mer. aka 'take, fetch, lift', cf. Malay ankat 'lift up'; the Merina word was overlooked by Dempwolff who picked Mer. akatra 'ascent' as cognate instead, but the latter need only be a doublet, if it is related. Similarly final -r, though commonly represented by -tra, is not only represented by -na, but also by loss: *sulur, Mer. solo 'shoot', beside *kembar, Mer. kambana 'turn', and the more common type exemplified by *luar, Mer. loatra 'extract'. The hypothesis that final nasals were lost would be supported by the fact that loss of a final nasal appears also in Merina though rarely: *( )intem, Mer. (Sak.) ma-inty 'black'; *piggan, Mer. finga 'plate'; *laraij, Mer. rara 'prohibit'. Furthermore the loss of a final nasal is regular in Sakalava and is represented in Betsimisaraka by n, na, y, tja without rule: *?enem, B.sar. enin *six'; *( )inom, B.sar. m-inon 'drink'; *Dalem, B.sar. lalitj 'deep'; *tatjan, B.sar. tarjaii 'hand'; *bulan, B.sar. volana 'moon'; *diydiq, B.sar. rindriya 'wall'; *lesuy, B.sar. leon 'mortar'; *bakuy, B.sar. vahona 'a plant'. Of course such a distribution could be brought about by dialect borrowing, but it could also be due to the restoration of consonants analogically in a position from which they had been lost. Both Dahl and Dempwolff agree that one must allow for 'false analogy' to explain the finals in such instances as Mer. sisina 'edge' from *sisiq; Mer.fasika 'sand' from *pasiR; Mer. vavatra 'opened' from *wawaq 'wide' and many others. Such analogical explanations would be facilitated if final consonants were lost before pause, but retained under certain sandhi conditions. Analogy would however no longer be necessary to explain Mer. ahy 'mine' from *aken; Mer fofo-fofo 'light breeze' (beside fofotra 'blow') from *puput, or Mer. ma-inty 'black' from *( )intem, for these would now be evidence that final consonants had been lost. On the other hand the root form appearing in maintis-ana 'be blackened' must be attributed to a different analogy based on the alternation 0 ~ s exemplified above. A suggestion of the manner in which the consonants might have come to be restored is now in order. It can be supposed that originally at least some words had an initial consonant that formed clusters with final consonants in sandhi and that these clusters later simplified into single phonemes. Some of these initial consonants may have disappeared (if e.g. from *q) or continued in their initial occurrence otherwise. Let us call these initial consonants X and suppose that they were voiceless. There would then 2

Mlg.R. indicates a Malagasy word in Richardson's dictionary (1885) after Prov. (Provincial).

232

ISIDORE DYEN

be a variation between certain final consonants C, (from *-Cj) according to their position before pause, before vowel, and before X. Suppose the following changes were to occur: 1) Ci remains or assimilates to X with the result that some consonants merge and there are now relatively fewer consonants that occur before X. This could happen for example if the paired voiced and voiceless stops were to merge into voiceless stops before X: e.g. the reflexes of *-p and *-b merge into *-p. We will now call the fewer varieties C}. 2) X ceases to occur after Cj either by coalescing with Cj into a single consonant X} or by disappearing. It remains elsewhere however, say after pause and after vowels. The consequence is that now Xj alternates with the final stop Cj in the positions in which coalescence occurred, and not necessarily anywhere else, for example before vowels. On the other hand X outside of the position of coalescence continues to occur as X, or simply disappears. Thus if X is *q, we can imagine that it might either (1) disappear after consonants without trace, thus producing no alternation, or (2) it might coalesce with a preceding consonant, the result of the coalescence then merging with another consonant, thus producing an alternation between the new outcome and the old final consonant. Thus for example a sequence of *-r before a *-q would result in a cluster *-rq- which might result in perhaps *-r- now occurring before a vowel after the loss of *q or might become a new consonant say *tr (perhaps through a voiceless r) occurring intervocalically before the vowel of the next word. Thus the intervocalic tr of ampatr'ampatra, the doubling of ampatra 'extend' ( < *hampa[r]), could have originated in a coalescence of the initial and final consonants. 3) Final consonants are lost. The consequence is that new alternations appear between 0 prepausally and the old consonants continuing in sandhi. Essentially this alternation now obtains between the Merina prepausal alternant lefa 'escaped', and its presuffixal alternant as in a-lefas-o 'let go'. 4) Like vowels in any sequence coalesce. This actually appears in sasa 'wash', sasana 'washed' (taken to contain sasa + ana), but this event is now posited to have occurred in any sequence of like vowels in a sentence. This is currently the case for sequences of unstressed like vowels. 5) By analogy to the relation between sasana to sasa, a form like *ampatrampa (from *hampa[r]hampa[r]) could be interpreted as * ampatra-\-*ampa\ the prior form could then replace the anticipated *ampa particularly because *ampatr would also be appearing before other words of a sentence with other vowels as well as initial a. Such sequences are posited to be the basis of the compounding rules as they apply to the cr-proparoxytones and words like them. 6) Further analogical changes of various and less general types would bring the observable situation into being. The development indicated is indeed long and complicated. But it is one way to explain all of the phenomena without the contradictions that face the attempt to treat the last consonant of a-proparoxytones as regular reflexes. The model of the analogy

MALAGASY

233

that produced the so-called final consonants would then be the alternation between a prepausal vowel and its coalescence with a like vowel in sentence sandhi. The analogy itself would tend to remove an alternation between 0 (prepausal) and a consonant (prevocalic) by replacing the 0 with the consonant, a perhaps not too surprising replacement. This would explain at once why some original final consonants are not represented at all whereas others are represented irregularly. Nor need it be true that in the analogical restoration of a consonant to a position corresponding to the PAN final consonant only those factors indicated were operative. Other factors could have cooperated to produce the rather varied results that appear. Thus one must also include the possibility of a second analogy to produce Mer. tenom-ina 'be woven' from tenona *kk (see above). But this would imply some analogy again. Words that originally ended in *-r could perhaps offer a model; e.g. hihitro 'my avarice' from *kikir + *ku if the cluster *rk passed through a stage *rh because *k became *h after *r as well as elsewhere. This *rh could have become a voiceless *r such as might have become tr. [Of course it is not excluded that *rk might have become tr, but this hypothesis is less appealing.] The same hypothesis in any case could apply to words originally ending in *-d, *-D, *-j, and *R4, since all became r in other positions. Under this hypothesis the form soratro 'my writing' would be analogical to some such form as hihitro 'my avarice', because of the similarity of soratra 'writing' to hihitra 'avarice'. This hypothesis suggests solutions for other problems. The origin of Mlg. tr in intervocalic position is puzzling. Dempwolff posited 'reduced nasal clusters' to account for the intervocalic and initial Mer. k instead of h from *k or *g; intervocalic and initial tr instead of the regular r from *d and *D; intervocalic and initial p from *p. It is evident that Dempwolff's hypothesis of 'reduced nasal clusters' is unsatisfactory. It provides for two types of reflexes from the same proto-sequence without difference in condition. Thus he assigns Mer. titjgi 'high' to *tiygi{) and Mer. takuna 'carry together' to *tayguy; similarly Mer. vurjgu 'pile' to *burjkus 'bundle' and Mer. takotra 'cover' to *tarjkub\ Mer. fatrana 'clearing' to *panday and Mer. tandruka 'horn' to *tanduk; Mer. sampi 'hang' to *sampay and Mer. toko 'tripod' to *twjku{). This hypothesis is objected to by Dahl (1951:115-19) insofar as it applies to tr, but not so in the case of k and p. However for the latter two he offers an alternative hypothesis. Dahl had concluded (1938:218) that in the interior of a word *i]k had become *kk (as in Toba) and then k. But in order to do so, he would have to ignore Demp-

234

ISIDORE DYEN

wolff's instances of *yk > rjg and he does not do so explicitly. Nevertheless Dempwolif's examples can be disregarded on the grounds that they are weak. In the case above the cognates associated with *buykus generally have a meaning quite different from that of the Merina word: *burjkus, Toba bukkus, Jav. burjkos = wutjkos, Mai. burjkos, Ngj. buijkus 'bundle'. The same poor semantic fit appears in Dempwolff's other example: *suykuk, Jav. sdtjkdq, Mai. sorjkoq 'velvet cap', Mer. sutjgu 'place on head with long hairs'. If these instances are ignored, then *rjk always becomes Mer. *k. But the instances involving *g (i.e. *>jg > rjg and *rjg > k) seem to be semantically sound. Simply put, the problem still remains that there are double reflexes assigned to *tjg (rjg, k), *nd, *nD (ndr, tr), and *mp (mp, p) without condition. It is natural under the circumstances to consider the possibility of clusters with a consonant other than a nasal as origins for the voiceless stops as reflexes. Thus one could associate Mer. takotra 'cover' with Tag. taklob as reflecting *taklub, or even perhaps *talkub under a hypothesis of metathesis in Tagalog. Such a hypothesis would simplify the explanation of intervocalic *p since it permits the derivation of Mer. tapaka 'cut' from a reconstruction with an internal cluster of *p with a preceding element not identical with the reconstructed nasal. But the other member was presumably voiceless since all of the so-called 'reduced nasal clusters' are voiceless even when as in the case of *g and *d, *D, the proto-phonemes involved were presumably voiced. In fact the other member could even have been a voiceless nasal since it would then have been sufficiently different to act as a separate conditioning factor producing a single voiceless consonant in Merina. To simplify our further discussion it is necessary to assign a symbol to the required, but only posited consonant. There could have been a number of different consonants in different cases. Thus the symbol chosen can have only a cover value. The symbol we choose is X and assign it to a position before a consonant. We can now say that Mer. takotra is from *taXkub\ Mer. tapaka is from *taXpak, and Mer.fatrana 'clearing' is from *paXday as opposed to Mlg. R.fandra 'plain' from *pandarj. The hypothesis involving is necessary only for the intervocalic reflexes. Initial k (instead of h), tr (instead o f f ) , and p (instead o f / ) can be explained as resulting from a clustering with various voiceless final consonants, later reduction of the cluster, and analogical reassignment. Even the occurence of initial b (as in Mer. bako 'pretty', *bagus) could explained in this way. However the intervocalic d of Mer. bodo 'infantile' if assigned to *buDu(), Jav. boDo, Mai. bodoh, Ngj. ba-bodo, bodo-bodo 'foolish' would need a different hypothesis. Since thus far at any rate it is unique, it can perhaps be disregarded. One might have expected some of the doubled monosyllables to confirm the hypothesis indicated above. The following two instances can be cited: Mer. kakas-ana 'scraped' from *kaskas (with analogical initial k) and Mer. tritry 'suck' from *dibdib (probably through a stage *dipdip > *riprip with analogical initial tr). Other words associated with doubled monosyllables are either ambiguous for the difference between

MALAGASY

235

a doubled monosyllable (C a VC b C a VC b ) and a reduplicated monosyllable (C a VC a VC b ), or point to a reduplicated monosyllable (cf. Dyen 1947:235 ff.). Thus this area of evidence tends to be neutral.

The linking nasal The nasal that is added in close compounds after a non-proparoxytone ending in a vowel (e.g. volo-m-bava 'mustache': volo 'hair' + vava 'mouth', see above) was explained by Dempwolff as originating from the abbreviation of an attribute-forming particle ni (cf. Tag. possessive particle ni, 1934-38:5.118) and the appositive-forming particle na (cf. Tag. apposition-determinant na, 5.118). Dahl finds this explanation too complicated. He believes (1951:138) 'Tout s'explique sans la moindre difficulté si l'on suppose qu'à l'époque ou naissait le sandhi, la langue exprimait le possessif de la même manière que le fait le ngadju actuellement.' He illustrates this by the examples from Hardeland (1858:83-5) sokah olo ta (lit. stick man that) 'the stick of the man' and huma-n olo ta (lit. house of man that) 'the house of the man'. Here the nasal n appears after the head word of the construction if it ends in a vowel. Moreover Hardeland (1858:84) draws attention to a contrast between huma papan 'a board house' and human papan 'a house for boards' which suggests that the former is semantically equivalent to the Malagasy open compound and the latter to the Malagasy close compound. On the other hand the latter could be a phrase with the enclitic nasal functioning as a relation in a relation-axis construction. The appearance of n after an attribute of the head, as in huma tahetan keton '(lit. house new-of [tahetâ 'new'] you) your new house', agrees with this analysis. Since no nasal can occur after the head expression if it ends in a consonant, the Ngaju sequence lampat hedjan (lit. rung ladder) is apparently ambiguously 'ladder rung' or 'rung of a ladder': i.e. it can belong to either the class of huma papan or of human papan above. In Malagasy the additional n occurs in compounds. There would seem no point to regarding a word like volo-m-bava 'mustache' as anything but a compound. It might then not be precise to say, as does Dahl (1951:136f.), that Ngaju has exactly the same manner of forming the 'possessive' (the second part of what is here called a compound) as Malagasy. At best one can say that the additional n that appears in the Malagasy compounding could be due to a fixation in the compounding of what was once a syntactic construction like that which appears in Ngaju. There is however another system of compounding that is relevant here. Bloomfield (1917:210) says of Tagalog that in compounds 'if the first member ends in syllabic n, or glottal stop, the particle IJ appears before the second member' and otherwise there is a simple succession : e.g. bu:ya-y-ka:hoy '(fruit tree) tree-fruit', bu:tas-kara:yom '(hole needle) needle's eye'. The single difference between the two ways of compounding lies between the Tagalog ij and the Malagasy n, for the latter must be presumed to appear in dialects which unlike Merina have the n:y distinction. However it is

236

ISIDORE DYEN

not impossible that the apparent difference is due to an analogical change in Malagasy so that both the Tagalog and the Malagasy system of compounding could be related to each other and perhaps even to the corresponding Ngaju construction. Such a hypothesis would be supported by the y of Sakalava eney-andro 'six days' (ene 'six' + andro 'day'). Unfortunately other examples have not been found.

The date of settlement and the Malagasy-Maanyan

hypothesis

The position of Malagasy among the Austronesian languages cannot fail to be a stimulating problem because of its isolation. One writer, M. E.-F. Gautier in an article entitled "Les hova sont-ils des malais?" (1900), concluded that they were Melanesian and probably Papuan. It is now generally accepted that the Malagasy are to be associated with some subgroup of the western Austronesian languages, but the identity of its immediate relatives is likely to be a matter of some controversy. Dahl (1951) has suggested that it should be immediately subgrouped with Maanyan and that the settlement by people from Borneo occurred about 400 A.D. Dyen in a review of Dahl's work (1953) has indicated that although there is some unevenness in his arguments, Dahl has presented a case worth taking seriously. Dyen concluded that the arguments that Dahl advanced on morphological and syntactic grounds seemed not to be persuasive because it was difficult to distinguish between inheritances from Proto-Austronesian and common innovations. However a crude lexicostatistical study seemed to confirm Dahl's hypothesis because Malagasy scored markedly higher percentages with Maanyan than with Malay or Toba. Nevertheless although Malagasy scored higher with Maanyan than with Ngaju, the difference was relatively small. Still a lexicostatistical comparison of Antandroy and Marina by P. Verin and C. Kottak reported in the Bulletin de Madagascar (Dec. 1967, No. 259, p. 950) yielded an estimate of fifteen centuries for the divergence of the two dialects, a figure that fits well with the date 400 A.D. inferred by Dahl as the date of the settlement of Madagascar by Indonesians. Consideration of the vocabulary shared by Malagasy exclusively with Maanyan (Mny.) tends to be persuasive as the number is large. It is hardly likely that borrowing is to be taken seriously as a factor. The following is a selection from among the Malagasy-Maanyan comparisons offered by Dahl that have no known cognate in other languages (though undoubtedly one will turn up for a few) and are probably cognate: Mny. Mny. Mny. Mny. Mny.

tenga, Mer. tena 'body'. lapeh, Mer. lafy 'side'. mahamoi, Mlg. mamo 'drunk'. butoh, Mer. boto 'youngster'. walenon, Mer. lavenona 'ashes'.

MALAGASY

237

Mny. ng-anro, Mer. ma-handro 'cook'. Mny. ondan, Mer. ondana 'pillow'. Mny. tonrak, Mer. tondraka 'spread'. Mny. a-limoken, Mer. domohina, Old Betsimisaraka (VB.sar.) demohin 'pigeon', Sak. limohe 'turtle-dove'. Mny. hanai, Mer. akany 'nest'. For similar prefix in Mlg. cf. Sak. akalo 'pestle' with Mai. (h)alu 'pestle'. Mny. puhi, Mer. foy 'hatched'. Mny. puya 'kind of crustacean', Mer. foza 'crab'. Mny. mamo, Tandroy mamo 'mosquito'. Mny. putut, Mer. fototra 'foot of tree or plant'. Mny. wasi, Mer. vatsy 'provisions'. Mny. kariwe, Mlg. hariva 'afternoon, evening'. The Mny. -e offers difficulties. Mny. itah, 'to pass', Mer. ita 'water passage'. Mny. ng-uteh, Mer. oty 'gather fruit'. Mny. mi-amben, Mer. mi-ambina 'protect oneself'. Mny. wintan, Sak. vinta, Mer. fintana (with / by false analogy from mamintana 'fish with line') 'hook'. Mny. nipet, Mer. mi-tifitra 'shoot', tifir-ina 'be shot at' (with i for expected e by analogy). Mny. nanga 'open', Mer. tana-tana 'opened'. Mny. ng-ungkar, Mer. voa-okatra 'make appear'. Mny. wakut, Sak. vahor-a 'bound'. Mny. tuing 'answered', Mer. manoina 'answer', mi-toina 'resound'. Mny. umbon 'company', Mer. ombona 'community, society'. Mny. ng-alat 'steal', Mer. halatra 'theft'. Mny. aret, Mer. aritra 'endure, tolerate'. Mny. law it, Mer. lavitra 'distant'. Mny. mape, Mer. mafy 'hard'. Mny. bidik, Mer. ma-vitrika 'agile'. Mny. tena, Mer. tena 'really'. Mny. ma-riip 'dry', Mer. ritra 'dried'. Conclusion The reflexes of Malagasy offer striking problems. They present an interesting challenge to the comparativist and it is to be hoped that some progress has been made here toward their solution. A more thorough comparative study of the morphology and syntax begun in a very small way by Brandstetter in his Tagalen und Madegassen (1902) and taken up in a somewhat restricted fashion by Dahl (1951) is highly desirable. Above all the analysis of the language hitherto attracting French scholarship almost

238

ISIDORE DYEN

exclusively has begun to interest a more international group including scholars in the United States and the Soviet Union. It is to be hoped that we have seen thus far only the forerunners of more extensive and deeper studies.

REFERENCES* ABINAL,

P., and P.

MALZAC.

1888.

Dictionnaire malgache-français. Paris (5th ed.

1955).

Maljgasskiy yazik. Moscow. 1917. Tagalog texts with grammatical analysis. University of Illinois Studies in Language and Literature III.2-4. BRANDSTETTER, REN WARD. 1902. Tagalen und Madagassen: eine sprachvergleichende Darstellung als Orientierung für Ethnographen und Sprachforscher. Luzern. COUSINS, G. 1960. Gramera malagasy. Tananarive. DAHL, OTTO CHR. 1 9 3 8 . Le système phonologique de proto-malgache. N T S 10.

ARAKIN, V . D .

1963.

BLOOMFIELD, LEONARD.

189-235.

. 1951. Malgache et Maanjan: une comparaison linguistique. Oslo. . 1952. Étude de phonologie et de phonétique malgaches. NTS 16.148-200. DAMA-NTSOHA. 1 9 5 1 . Dictionnaire étymologique de la langue malgache. Première partie: Les mots dérivés des apports sanscrits. Tananarive. DECARY, R . 1928. Lexique français-antandroy. Tananarive. DEMPWOLFF, OTTO. 1934-38. Vergleichende Lautlehre des austronesischen Wortschatzes. ZES Beihefte 5 (1934), 7 (1937), 9 (1938). DUBOIS, H . 1917. Essai de dictionnaire betsileo. Tananarive. DYEN, ISIDORE. 1947. The Tagalog reflexes of Malayo-Polynesian D. Lg 23.277-38. . 1953. Review of Malgache et Maanyan by O. Dahl. Lg 29.577-90. . 1956. The Ngaju-Dayak 'old speech-stratum'. Lg 32.83-7. FAUBLÉE, JACQUES. 1 9 4 6 . Introduction au malgache. Paris. FERRAND, GABRIEL. 1 9 0 3 . Essai de grammaire malgache. Paris. . 1909. Essai de phonétique comparée du malais et des dialectes malgaches. Paris. GARVEY, CATHERINE J. 1 9 6 4 . Malagasy introductory course. Washington. GAUTIER, M. E.-F. 1900. Les hova sont-ils des malais? JA 278-96. HARDELAND, AUGUST. 1 8 5 8 . Versuch einer Grammatik der Dajakschen Sprache. Amsterdam. KORNEEV, LEV, and FREDERIK RAKOTOSON. 1966. Diksionary Malagasy-Rosiana = Maljgassko-Russkiy slovarj. With a brief sketch of Malagasy grammar by V. D. Arakin. Moscow. * This treatment of Malagasy benefited from having access to the "Short bibliography of Malagasy", a study prepared by Mary Lu Joynes for the Center of Applied Linguistics.

MALAGASY

239

P. 1950. Grammaire malgache. 3rd ed. REJEMISA-RAOLISON, R. 1966. Grammaire malgache. Tananarive. RICHARDSON, J. 1885. A new Malagasy-English dictionary. Antananarive. ROUSSELOT, P. 1913. Phonétique malgache. Revue de Phonétique 2.364-88; 3.5-49, 119-51. STARK, E . L . 1967. Malagasy without moans. Tananarive. TUUK, H. N. VAN DER. 1865. Outlines of a grammar of the Malagasy language. MALZAC,

J R A S 2 d ser. 1 . 4 1 9 - 4 2 . VERGUIN, JOSEPH.

1955.

L'accentuation en malgache-mérina et en malais. Orbis

4.522-8.

. 1957. Deux systèmes de vocabulaire parallèle à Madagascar. Word 13.153-6.

Abbreviations B.sar — Betsimisaraka; Jav. — Javanese; Mal. — Malay; Mer. — Merina; Mlg. — Malagasy; Mlg. R. — Malagasy cited from Richardson 1885; Mny. — Maanyan; Ngj. Ngaju-Dayak; PAN — Proto-Austronesian; Sak. — Sakalava; Tag. — Tagalog; VB. sar. — Old Betsimisaraka cited in Dahl 1951.

THE AUSTRONESIAN LANGUAGES OF AUSTRALIAN NEW GUINEA

A. CAPELL

L INFORMATIONAL

0. Introduction: The Area and the Languages 0.1 The study of the language family called 'Austronesian' finds a key area in Australian New Guinea. Western New Guinea is a part of the same problem, but to a somewhat lesser degree because there are fewer Austronesian languages there. Farther west again, the eastern islands of Indonesia possess a similar importance. The Austronesian ( = AN) languages of Western New Guinea are considered in another chapter of this book, so that the present chapter is limited to the AN languages of the Australian part of the island. The geographical extent of these languages is defined approximately as that between 141° east long, to 156° east long, and 2.8° south lat. to 12° south lat. This is the region known as the political unit of Australian New Guinea. In it are included the Territory of Papua, which has been under Australian control since 1883, and the Trust Territory of New Guinea. The latter came into Australian hands during World War I and was assigned as an Australian mandate by the League of Nations; at the end of World War II it became a Trust Territory under the United Nations. While the Territory of Papua is a geographical unity, the Trust Territory of New Guinea has a number of obvious subdivisions: (1) the mainland, including the Admiralty Group on the north, (2) New Britain and New Ireland, with a number of smaller island groups; and (3) Buka and Bougainville, separated from the remainder of the Solomon Islands by the fact that before World War I they were German possessions, parts of German New Guinea. 0.2 Not all the languages of this area are the concern of this chapter, but only those which are classed as 'Austronesian'. Even before the beginning of the present century it had been recognised that in the apparent chaos of New Guinea languages there were some that shared vocabulary and basic grammatical structure. In New Guinea these were a minority group, but in the islands eastwards there was a majority of such languages, and when the New Hebrides was reached there were no languages that did not belong to it. This group had already been called 'Melanesian' in 1861

AUSTRONESIAN LANGUAGES OF AUSTRALIAN NEW GUINEA

241

by H.C. von der Gabelentz in the first volume of his Die Melanesischen Sprachen, and the name had gained general acceptance. In 1885 R. H. Codrington in The Melanesian languages gave the first detailed account of the island section of the languages, embracing the central and eastern Solomon Islands, Santa Cruz, the Banks Islands, and New Hebrides. In 1895 Sidney H. Ray, an outstanding English student of the Pacific Island languages, published a Comparative vocabulary of the New Guinea dialects, in which Melanesian and non-Melanesian groups had been clearly separated. The work of Wilhelm Schmidt made a similar distinction in the languages along the north coast of the then German New Guinea. 0.3 Distribution of Austronesian and non-Austronesian Languages Within the New Guinea area as thus defined, two linguistic groups are found. One is negatively definable as non-Austronesian (NAN) or 'Papuan', and this negative definition is of use, in that it sets the two main groups of languages apart without expressing any a priori theory of the nature of either group. The NAN languages cover the larger part of mainland New Guinea; in the other larger islands they are minority languages. In New Ireland there is only one such language — Panaras — on which there is still no published material at the present time; while the number of the NAN languages in New Britain is still somewhat uncertain, depending on the classification accepted for Mengen and one or two others. In Bougainville the bulk of the languages are NAN, but there are a few AN languages in the extreme north, and scattered along the east and west coasts, although still very poorly recorded. The northern and southern AN languages of the Buka-Bougainville area belong to different subgroups, and the Torau-Uruava subgroup has affinities with languages in the western part of the British Solomon Islands Protectorate from which they seem to be quite recent immigrants. 0.4 Subdivisions of New Guinea Austronesian The Austronesian languages of New Guinea (abbreviated NGAN) also fall into two subgroups, a fact which has not been pointed out explicitly before, but which is helpful in the present discussion. This subdivision is based on a typological feature manifested in two contrasting types of syntax. Austronesian languages in general, apart from New Guinea, have the order subject + verb + object in the simple declaratory sentence. In part of the N G A N field this order holds good, but in part the order is subject + object + verb 1.e. SOY as against the more normal SVO. This typological feature permits a first dichotomy of the languages into those which place the verb before the object and those which place the object before the verb. The former are in the majority, and the

5

o * 3 £ _) Q PL, Z >n vci r-I oo

256

A. CAPELL

Is. Farther away to the northeast there is the Nuguria group (or more exactly Nukuria, seeing that there are no voiced plosives in the language). This is Polynesian-speaking, and although, of course, Polynesian is part of the AN family, it is isolated in this area and will not be treated, except to add that Taku (Ta'u, Mortlock Is.) and Nukumanu or Stewart 1. speak practically the same form of Polynesian, which links with Ongtong Java and Sikayana in the British Solomon Islands Protectorate as one group of the Polynesian outliers. The various linguistic groups involved are shown on Map 3, which covers New Britain and New Ireland. Most of the islands are linguistically very inadequately known. The vocabularies of all this group are set out in a group of 25 words in Table IV. Most of the material has been gathered by the present writer; some is drawn from other sources acting on the writer's behalf. Grammatical material has been collected similarly. TABLE I V

New Ireland

Vocabularies

English PAN

sun *a(n)dav

moon *bulan

water *wayey, *danum

Musau E Mira Lavongai Tigak Kara Nalik Notsi Panaras (NAN) Lamasong Lelet

sinaka sinata si nag ias gas ejas siat mwaba'ri

ulana ulana ulen ulen fiilén filen ulirj ulaq

ateo ateu laman dum rarem danim dan borenam

vao laqit lerjit ráfúl davur laqit farat

taumatu oroi towan anu matá ramti tamat teima

gagankig limerenlayankig akin kasakes (nata)nas pisi wasu kamis kamis pisar matanias mataniaso

rjgolirj laggalir)

dan lendan

wubu lembet

randi leredi

tege teka kalaij funil kemben kalan kalan bialak wura wura

ari tak dan dan malun malun malum laur, godom da da

bera bata rein

tódl

Barok Pala Sursurunga Muliama Konomala Siar Lambel Nisan Tabar N. Tabar C.

rain *quzan

US

daf, tai bat kear usan bat bata

man *tavu± matah

tunan kalou kaltu misa barsan barsan tamat ka ka

AUSTRONESIAN LANGUAGES OF AUSTRALIAN NEW GUINEA

Lihir N.

257

Tanga Anir

matan gandiek pisi ligas

houwa

lindon

wos

tomat

funil funil

dan dan

mbat mbat

kaltu kaltu

English PAN

woman *(ba)binay

canoe *wagka

banana *punti

Musau E Mira Lavongai Tigak Kara Nalik Notsi Panaras (NAN) Lamasong Lelet Barok Pala Sursurunga Muliama Konomala Siar Lambel Nisan Tabar N. Tabar C. Lihir N. Tanga Anir

bause vause aina ulina tefen rafin aina mayambun

ale anua lu lui llfú val anua luán

laka tepe tape bi ruei kio kain

olimo olimo vul bul kaleu yaleu bul obinam

uri uri uri ur

une latkin wtine hahin wak fifin fafni fa'in hane kuah vevin vevine wahien fifin fifin

anu lorjgu lagúnón hala rum fel umai rumai pal um vono vanua luom fel fel

kasakis layavi ràt, kis kayau kas sara sara rat kas tolah like lake keke tag tag

zim lisim sim wagga takup tagup tagup waga tinih tsim tsim tagoup wag wag

udu loun un hudu hun komo komo fun hun pos win wundi huin komo komo

English PAN

sugarcane *tebu

fish *ikan

crocodile *buwaya

dog *asu

pig *babi

Musau E Mira Lavongai Tigak Kara Nalik

tou tou to tu taf ruf

koko koko yen ien in yen

ava waiya ue wuj vuye

pawa pawa kauvek piu più viu

mosu mosu vogo boyo wìàk beri

house basket *yumah, *balay, •bakul *banuwa

vudu ude

258

A. CAPELL

Notsi Panaras (NAN) Lamasong Lelet Barok Pala Sursurunga Muliama Konomala Siar Lambel Nisan Tabar N. Tabar C. Lihir Tanga Anir

tu sombok

ko ambulema

lawa fuye

kapune kawuna

bui kumarot

verun vatu tu tuh tuh tuf tuf

lowa lagip yip

yandi lagau wuluwun pap pap pul pap pul pap wilhire kopwen kapune poi pul pul

bo lombo

tou tovu toh tuf tuf

vamat lemat in kirip hisu kok wusif sis sis ian i ia matsien kok kok

English PAN

butlerfly *(mbembe)

bird *manuk

egg *teluy

head *ulu, *batu

ear *teliga

Musau E Mira Lavongai Tigak Kara Nalik Notsi Panaras (NAN) Lamasong Lelet Barok Pala Sursurunga Muliama Konomala Siar Lambel Nisan Tabar N.

kurubebe loulou be pepe pepe veve be kalalalambun bebe leveves roro tota toto fen fembe bem hambe

gila gila mani manui mànù mani man oyaratun

otolu itolu katui kateluk kàtèluk yaturo katul sag gara

uru noqo patu patu pat vata kuan boyom

taleqa talirja talirja telaga tàrja lirjga talirjga kiginan

viga lapike pun man man man sika mani mani

varjgatli layahilik-piga karalo tulura pasin katlu kolurna

bebe

ma

katuru

vatlak vatlak hoyo ulu lulupakeluulu puklupukulu lukowa

rintirja laltirjateliqi talirja taiga balmbalu taiga pitalgatalgatalgateterirja

toh

UW3Ì

fuai fuai uwai otiot lova lova kaidam

kotol

bore bor bo boi bori boroi suat bor boro bwal bo bo

2

AUSTRONESIAN LANGUAGES OF AUSTRALIAN NEW GUINEA

Tabar C. Lihir Tanga Anir

beabea komukobe tultul fem

ma mon man man

katoru katuru katluk kotlu

kowa kaho paklupaklu-

tatariija tena belbelubelbelu-

English PAN

hand *lima

I *aku

you (sg.) *kaw

one *esa

two * CT* -O 43 O 60 9O cd «

s .3

o G •S

a . . D M Cl cd Ul

P

6

a 4>

" f t

S 2

T3

2cd

0

« S cd o S '-B S oa Ci M r* çd 3

C 8

* S3 cd a Poyo, *kamu > Pomi; while the third person forms are the same as in (b). These languages also form duals and trials in the same way as group (b): Dobu dual, ta-te-rua, a-te-rua, wa-te-rua, si-te-rua, respectively. In the trial the suffix is to < PAN *telu, 'three'. Kiriwina ( A N J uses the taw-forms in the third person only, and PAN reflexes in the other two, but the situation is complicated by the noun classification and concord discussed previously (p. 274). There are no such subdivisions of these languages in regard to verb subject markers, possessive or object suffixes. These all follow normal AN patterns with reflexes of PAN roots not limited to NGAN. The double possessive — pronoun preceding the possessive — appears in a number of mainland languages, e.g. Motu lau egu ruma, 'I my house'. This seems to have been first an emphatic form. It reappears in Graged on the north coast of New Guinea (Capell 1943:231-2) and would seem to be an NAN feature either borrowed or inherited.

AUSTRONESIAN LANGUAGES OF AUSTRALIAN NEW GUINEA

293

ii. Verbal systems (a) The classification of verbs in some of these languages has been cited and will arise again in the theoretical part of this essay (p. 294). (b) The use of transitive suffixes to verbs, characteristic of island AN and of PN, is not universal in the Papuan section of NGAN, or, for that matter, very frequent in the N G section. In Motu a direct object is simply added to the verbal root: ita-gu, 'see me' in Motu, and in Wedau inanai-u, 'see me'. In Motu, however, there is a 'remoter transitive' suffix as in gwau-rai-a, 'talk about it'. In Wedau a rather elaborate set of nominalisations seems to be based on a similar kind of suffix. Formally, though not phonologically, these suffixes are parallel to Fijian -taka in dambe-taka, 'climb with (something in one's hands)'. In other languages a transitive suffix does appear, usually -i, which has Indonesian AN parallels: Mukawa bere, 'give' < PAN *beyay: beriku, 'give me', bere-i, 'give him'. Dobu shows similar formations (Arnold 1931 :§ 123). Other languages again embody the PAN *aken preposition: Tubetube -aga-u, '(do it to) me'. (c) Tense and aspect systems vary greatly but can be harmonised in certain area groups. The Motu aspect and tense forms were given in detail on p. 286. Three future formatives occur: na,pa, and ka. Where na between subject and verb marks a future, ta or da generally marks an Irrealis; where pa < PAN *panav, 'go', marks a future, the Irrealis is ma; where ka < PAN *lakav, 'go, walk', marks a future, ma serves again as an Irrealis marker. The na forms occur in Mekeo, Wedau-Mukawa group, Ferguson, and Normanby Is., except for Dobu, which differs but uses da as an Irrealis, and in the Misima area. The Irrealis ta occurs in Wedau not only as 'may, might', but with non-future negatives: eya a ta inanai, 'I did not see him'. The negative is regarded as unreal in that the action did not take place, but not all the languages make such a use of it. Futures with pa are found in the western Papuan area, and the central as exemplified earlier in Motu is really a different treatment of the same theme. It appears in what would seem to be an earlier stage in the Markham Valley Wampar as ban < PAN *panav before the entire verb phrase: gean ban i mu garaweran, 'he will become chief'. This language also has a set of gerunds that recall the Wedau structures very closely. The ka forms are found in Roro; elsewhere they appear chiefly in AN t areas such as Nakanai and outside the NGAN region in the southeastern Solomons, Malekula, and some PN outliers, as well as in parts of Indonesia. A vaguer system of tense with more use of Irrealis forms occurs in the Suau area and is picked up in the north in Yabem (p. 288). Past tenses are either not specifically distinguished, the forms really being nonfuture, or used with particles that are so variable as to suggest NAN origins. iii. Numeral systems In the SEP languages the numeral systems show a deal of diversity: (a) decimal only in Sudest and Misima — and interestingly enough in the NAN Yele of Rossel Island. Quinary systems, usually with a special word for 'ten' (normally of PAN origin) are the commonest types. They are illustrated in Capell (1943:260-4).

294

A. CAPELL

2.1b The Bougainville Section of AN 2 Languages These comprise the Torau-Papapana-Uruava group, scattered along the east coast of the island. They have been described in other connections in earlier parts of this paper. Some specimen sentences in Uruava and Torau were given on p. 279 and the main points of contrast were noted in the description of the A N j Banoni language on p. 281. The vocabulary specimens have also been given (pp. 277 if.). The phonology will be dealt with in Part II (p. 300) because its special features have some importance for the historical study of Oceanic AN (or 'proto-Oceanic (PO)' as Milke first called the eastern half of the AN languages), and there is therefore little need for further general comment now. II. ANALYTICAL 1.

Introduction

The historical questions involved in the N G A N languages are obvious from the contents of Part I. Linguists concerned in this field have been active in analysis, discussion, and advancing of theories over a long period of years. These cannot all be discussed here, but the chief theories must be mentioned and indications of their merits and demerits given. The activity in this field gains in importance because the N G A N area is regarded by some as holding the key to much of the history of the whole family of Oceanic languages. G.W. Grace (1964) has written in this regard some words that by themselves should be sufficient to justify not only the present essay but further developments of it: 'Dyen's study indicated that the area of greatest diversity was New Britain which, in the framework of his study, seemed to suggest New Britain as the possible homeland of the Austronesian family. I am also impressed by the linguistic diversity in that part of Melanesia, although I would define a more general area consisting of northeastern New Guinea and the neighboring islands. It is my present inclination to suggest very hesitantly that the Proto-Eastern-Austronesian language was spoken, and first differentiated, somewhere in that general area.' Although this is not proven, yet, if there is the least possibility of it, the study of NGAN becomes a key to the whole problem of Oceanic linguistic origins. In the present space hardly anything more than an outline of the problems can be given. Hence the most important of the historical questions will be passed in review in Part 2, which consists of three sections: phonology, lexicon or wordstore, and morphology-syntax, the grammatical structures of the languages. Examination in terms of lexicon alone is not sufficient, because words are more easily borrowed or lost than grammar. Lexicostatistics alone cannot pass any final judgment on the history of languages; it can only give directions for deeper research. The procedures called by Capell 'Structure statistics' (Capell 1962a:l-ll) carry the study a stage further, but these also would nearly require a book length publication.

AUSTRONESIAN LANGUAGES OF AUSTRALIAN NEW GUINEA

295

As each level of language cannot be dealt with in detail here, only indications can be given as to current research and present trends. 1.1 Present Position of Linguistic Research on NGAN Work carried out by S. H. Ray, Schmidt and others in the sphere of NGAN was limited by the linguistic methods of the times, which were based on an adaptation of the classical philological methods by J. C. H. Kern and R. Brandstetter. By the time Dempwolff's work was carried out, these earlier investigators were no longer working. Ray tried to apply Dempwolff's findings as far as his materials permitted. His use of the term 'Indonesian' in the sense of 'Proto-Austronesian' created continual ambiguities and mars what he wrote about the languages. His chapter in Melanesian Island languages (1926) is particularly misleading in this way, although his suggestions are good. Capell's thesis (1943) made perhaps the first large scale attempt to apply Dempwolff's thought to NGAN, but this work was limited to Southeastern Papua (SEP). He put forward a four-strata theory of migrations or movements. Later scholars have weighed this theory in the balance and declared it wanting, Chrétien (1956) in the first place and also Milke (1958 and 1961). The latter has given a good deal of attention to NGAN, and a résumé of his opinions is desirable at this point, for they provide a basis on which further study can then build. These views are set out in three articles : Milke (1958, 1961, and 1965). Milke first set up an Oceanic subgroup of the AN languages, to which he referred as PO, 'Proto-Oceanic', and linked it generally with Dyen's 'Heonesian linkage' (Dyen 1962), in spite of Dyen's own objections to the concept of PO. Within this postulated PO subgroup of AN Milke then distinguished a 'New Guinea cluster' (Milke 1965:331), thus reviving a discarded theory of Schmidt (1900-01) and Friederici (1912-13). This subgroup he found chiefly in New Guinea itself. He specifically excluded Nakanai of northeastern New Britain (Goodenough 1961b). Neither Milke nor Goodenough has studied the languages of the Admiralty Islands, New Ireland and the smaller dependent groups, New Britain as a whole, Buka and Bougainville. As parts of Australian New Guinea, these languages are relevant to the present undertaking. Practically no comparative work has been done as yet on these areas, and the suggestions to be made here are therefore extensions of the published work of the earlier authors, arising largely from field work and other studies by the writer. 1.2 Plan of Following Sections The general plan of the work here set out may be expressed as follows. Milke's work was concerned in the main (especially his 1965 paper) with northern NGAN. If that work can now be coordinated with Capell's (1943) earlier work, then the bulk of the N G A N on the mainland will have been covered. Extensive work by Goodenough, both the 1961b article and an unpublished paper (1961a) presented at the Pacific Science Congress in Hawaii in the same year, on languages of the north coast of New Britain, will carry the study as far as the northern Solomon Islands. To

296

A. CAPELL

this extended material will then be added the Admiralty and New Ireland areas, together with the Buka area and the AN languages of Bougainville, and so take in the remainder of Australian New Guinea.

2.

AN Phonology in NGAN

2.1 The study of phonology provides the foundation of linguistic comparison, even if structure statistics are used at a later stage. The starting point here is Dempwolff's basic work, allowing for Dyen's restatements (1953). It may not be necessary to use the entire sound system to establish diagnostic tests for subgrouping. Milke held that the really diagnostic developments in P O are those concerned with original palatals. These will therefore be examined first, other groups of sounds being added as needed. A few modifications of Dyen's symbols have been made by practically all subsequent writers, and Dyen has introduced an extra symbol, Z (Dyen 1951), for which there is no earlier equivalent, to cover the cases in which Malay j corresponds to Javanese d, r, and Hova r. In most of the words involved, Capell used d, and this usage is retained here. The palatal series in the present work therefore presents the following correspondences : Dempwolff Dyen Milke Capell

d' s z z

(D') Z (d)

g' j j r

k' c c e

t' s s s

On the basis of P A N palatals, Milke (1965) arranged N G A N into the following subgroups, presumed to be genetic: 1. P O *z, both > / d / : AN 2 languages of western Papuan section: Mekeo-Motu-Hula areas. 2. P O *s, *z, unified as /s/: Geelvink Bay languages: Tumleo-Ali group; Yabem-Taemi-Kawa? group; South Cape and Dobu group. 3. P O > /s/; P O > IT/: Manam-Wogeo; northern New Britain; western New Britain (Bariai, Kove, Bulu, Nakanai, and Vitu); Misima area of SEP. 4. P O *s > /s/; P O *z > 0 : Graged group; Atsera; Mukawa group. Some of these groupings are especially interesting, chiefly perhaps Group 4, for Milke was inclined to see direct genetic links between Graged and the Mukawa region via the Markham Valley, and this has been commented on several times in Part 1. Here, of course, Wampar and Watut, the latter as given by Fischer (1963: 207-86), will need to be taken into account. Milke adduces as common to all these languages the grammatical feature of verb classification by prefixes. He points out

AUSTRONESIAN LANGUAGES OF AUSTRALIAN NEW GUINEA

297

that this had been one of the main grammatical traits of Capell's postulated 'northeastern Regional Language' (Milke 1965:347; Capell 1943:169), as a predecessor of the modern languages of the northeast coast of Papua. This possibility will be considered later (p. 323). It is the only case in which any evidence other than phonological has been previously produced for one of the suggested classifications, but the present paper will provide others. Moreover, all these languages belong to the AN 2 group, with its SOV syntax, postpositions, and other characteristic features. On the other hand, all the members of 3 are ANX in every respect, except ManamWogeo, and this exception shows that the particular phonetic characteristic is independent of subgroupings made on other grounds. Milke's 2 includes also both AN t languages (Geelvink Bay, Tumleo-Ali) and AN 2 (Jabem-Tami-Kawa? and South Cape-Dobu). Here again the syntactic features distinguishing AN^ and AN 2 appear to be independent of subgroupings made on other bases. The AN 2 languages thus share a feature which belongs to the NAN languages about them, and such a sharing would suggest a 'pidginization' process. Milke's 1 represents a special local development for which he gives a fully satisfactory theoretical diagram (1965:337), and all its members are AN 2 , though they differ markedly among themselves in grammar. Entirely separate from these are the New Ireland and northeast New Britain languages as far as Rabaul and the Duke of York Islands. These Milke had already separated out (Milke 1938:60). According to the 1938 article, the N G A N languages formed a unified group subdivided into languages from Geelvink Bay to Humbolt Bay (the present West Irian north coast) and the 'New Guinea group in the stricter sense', which included 'all other M N languages of New Guinea including southwest New Britain.' These are contrasted to the 'New Ireland group — languages of New Ireland itself and the outlying islands, Tuna (Gazelle Peninsula), MN languages of Buka and Bougainville, New Georgia and Choiseul'. This second grouping as it stands is not satisfactory. It subdivides into quite a number of smaller groups which differ in grammatical structure from each other and from New Guinea. It is quite impossible to group together with Rabaul-New Ireland the structurally quite different languages of Buka. In fact the languages of Buka present a verbal system without parallel elsewhere in Oceania. Milke's diagnostic trait, that PAN *y and *c?fall together as r, and original */is kept separate seems to be valid, at any rate for Pala and Tuna. This phonological development is held in common, but it seems almost a minor matter in relation to the structural differences in this rather large area. The subgrouping suggested here will be examined below. However, a footnote in Milke 1965 (330, fn. 1), after a revision of the substance of the article, which he was unable to embody in the published paper itself, taking into account the Watut languages that are AN, explains that 'this revision goes far to invalidate my original hypothesis, namely that the different patterns of reflexes of PO *s and *z might represent genetic groupings'. In this case, what has been written

298

A. CAPELL

earlier in the present paper may itself not be true. The reflexes shown in the immediately preceding Table (p. 296) are in themselves correct; the only point at issue is whether they represent a genetic group or not. That they occur as stated is not in question. 2.2 The Reflexes of the Palatals In view of these facts, and the possible change of opinion that Milke would have made had he lived, it is desirable to tabulate afresh the actual reflexes of the palatals in the various languages included in this paper. Languages in areas not tested by Milke must be added, continuing the testing eastwards and northwards. The test words used are taken as far as possible from among those used by Milke in the same context, and the *s and *z represent his PO forms so that comparisons may legitimately be placed alongside his. Non-AN roots are bracketed. In the extended examination, the Bougainville languages may be taken first: PO PO PO PO PO PO PO

*susu, 'breast' *qasu, 'smoke' *zai, 'who?' *zaba, 'what?' *quzan, 'rain' *Zalan, 'path' *zaqat, 'bad'

Ban. Ban. Ban. Ban. Ban. Ban. Ban.

vi-suu dzaaso? se sua (yarau) 'sanana sara

Ur. Ur. Ur. Ur. Ur. Ur. Ur.

susu aao (taka) n-ava wana 'arana (rori)

To. To. To. To. To. To. To.

susu aasu sea sai sua (tauna) sakaala

In summary, Banoni retains initial *s but loses medial *s, and merges *z with *s under some conditions. Torau merges *s and *z but retains medial *s. Uruava keeps initial *s but loses medial *s and all *z. Thus Banoni fits into Milke's Group 2 (and is an ANX language) as does Torau (AN 2 ), but Uruava into Group 4 (and is also AN 2 ) but like Banoni it loses medial *-s-. The Buka languages used in this essay are chiefly Taiof, Sailo, Saposa, Petats, Halia, and Kilinailau (Carteret Is.), from south to north. Most of the material is based either on the writer's own research or on data collected by the Tri-Institutional Programme of 1955 (TRIPP). These languages in some regards stand widely apart from the other AN languages, especially in the conjugation of the verb; in the treatment of the sibilants they show retention of both *s- and *z- throughout the languages here used, but the treatment of *z shows considerable variation: *z- > s- in Hahon, Saposa; *z- > h- in Timputs; *-z- > -s- in Taiof, Saposa, Petats, and Halia; -It- in Timputs. Difficulties arise in finding test words, because, curiously enough, comparatively few of the P O *z words have come into these languages. The Teop language of northeast Bougainville must also be taken into account, but this belongs to a very different tradition. This variation in the treatment of the sibilants in a group of dialects (rather than languages) seems to bear out Milke's second thoughts, i.e. there are no genetic factors involved.

AUSTRONESIAN LANGUAGES OF AUSTRALIAN NEW GUINEA TABLE I X

Reflexes of PO Palatals in Buka PO *susu, 'breast'

PO *tasik, 'sea' PO *esa, 'one'

PO »sa-puluq, 'ten'

PO »zaqat, 'bad'

PO *boze, 'paddle'

PO *azar), 'name'

PO *quzan, 'rain'

PO *biza, 'how many'

sus: Taiof, Sailo, Saposà, Kilinailau sisibi: Timputs cicinó: Hahon lumaluma: Petats, Halia tasi : Taiof, Petats, Halia, Kilinailau teis: Saposà sen :Taiof isen: Saposà sea: Petats tua: Sailo, Kilinailau toa: Halia safunu: Taiof, Saposà sawun: Hahon havun: Timputs towan: Sailo maloto: Halia, Kilinailau malot: Petats sata: Hahon hat: Timputs nomi: Petats orni: Halia, Kilinailau irig : Saposà, Taiof fos: Taiof, Saposà hose: Halia, Kilinailau hos: Petats asag : Taiof asagan: Saposà e sànana:Hahon e hnan : Timputs solo: Halia, Kilinailau sol: Petats usan:Hahon ?uhan: Timputs rarjit: Kilinailau lagits: Petats, Halia rarjits: Taiof ( = PAN *lagit, 'sky') fis: Taiof, Saposà to-his: Timputs sohis: Petats lahisa: Halia, Kilinailau

300

A. CAPELL

Table IX shows reflexes of PO palatals in the Buka languages. These examples highlight the extreme variation of AN and NAN vocabulary in the N G A N languages, e.g. words for 'bad' in Table IX, where in addition to the AN root, two NAN roots appear, iriy and nomi; in the words for 'rain', roat and ruesa do not appear to be reconcilable with AN although they are with each other; alongside *quzan, 'rain', appears the equally AN * lag it 'sky', with a change of meaning. Such changes of meaning are not uncommon in some areas and with some words, e.g. words for 'head': *ulu occurs, but *batu is not infrequent, although its primary meaning is 'stone'. Of course it is possible to reckon two homonymous roots *batu and there is no evidence either for or against this decision. 2.3 Retention of Final Consonants Another fact begins to draw our attention: the equivalents of both *quzan and *larjit retain the final consonants. Final vowels are often lost, but where final consonants are permitted in the structure of the morpheme, they are retained. The logical conclusion is that the PAN vocabulary when it passed outside the Australian New Guinea region still had its morphemes phonologically complete. Moreover, they remained complete still farther east, some as far as the Southern New Hebrides: *zaqat, 'bad', is a clear example, for this is sat in Eromanga, as it is hat in Timputs and even gifted with a supporting vowel as sata in Hahon. Eromanga deals largely in final consonants, e.g. nomurep 'life' < *ma + hudip (with n- article); menuoy, 'bird' < manuk, and many others. In Tanna also examples are frequent, e.g. manuk, 'bird' < PAN *manuk; nemuk, 'mosquito' < PAN *namuk; -arat, -elat, 'bad' < PAN *zaqat. The Uruava and Torau examples in this instance are of special interest, in view of the fact that final PAN consonants have been adapted to local rules, e.g. /t/ supported by /i/ has become /s/ where the combination ti- is not acceptable. This means that in languages in which a final consonant was no longer permitted, such finals in this area were supported rather than rejected. East of the New Georgian Archipelago this does not happen: the PAN form *ma-\-takut, 'feat' > Roviana matayutu > Nggela matayu, and in the last case the transitive suffix does not carry the thematic consonant t, but the form -ni in matayu-nia 'fear him'. In the western dialects of Fiji, however, where the word occurs (it does not occur in Bauan), the transitive form is mataku-dia; the original t was still extant when the word reached Western Fiji, and was changed according to local laws. In the Bougainville area the examples that come to hand include: Banoni manugu 'bird', sanana 'road', boroyo 'pig'; Uruava arana 'road', nasi 'four', onomo 'six' (Roviana also onomo), ranisi, 'cloud' < *larjit 'sky' (becoming 'rain' in Taiof, Halia and Kilinailau), tanisi 'weep' < *tayis, karasi 'bite' < kayat, for which Torau has karati, while it has for 'four' evati and for 'six' onomo. There are many other examples of such retention of final consonants by means of a supporting vowel throughout the AN vocabulary of this area, as well as farther east.

AUSTRONESIAN LANGUAGES OF AUSTRALIAN NEW GUINEA

301

TABLE X

Retention of-C (q)ayus 'current' atep 'roof' azan name (qa)bayat 'northwest monsoon' (m)bimbiy 'lip' d(d)ayah 'blood' degey 'hear' empat 'four' ikan 'fish' inum 'drink' kasaw 'rafter' kuden 'pot' kulit 'skin' kulul 'breadfruit' lumut 'moss' manuk 'bird' matakut 'fear' nanaq 'pus' namuk 'mosquito' niuy 'coconut' pulut 'glue' quzan 'rain' sabuy 'scatter' tal 'chop' taneh 'ground'

in SEP1

Tub. karusi; Kir. kum 2 , contrast Motu group unu forms Kir. katuva; Suau Patohi; Bas. gatoa; Dob. Patoa Dob. Pesana Pan. yaras; Bwd. yavalata Hula pipiya Bwd. dayagi Bwd. nogalina; Muk. Wed. nonori Kir. vasi; Dob. ata; Muk. bata; Bas. Keh. pari Kir. iena; Dob. iana; Tav. Muk. iana; Bwd. igana. Tav. Dob. numa Bwd. auga Bas. Puroni 'cook' Pan. kunis Suau Punuri Tub. dumosi Bwd. manuga; Dob. manua Bwd. Dob. matauta; Tav. matouta; Tub. matausi Suau nanahi Dob. namua Tav. nirua Tub. posori Kir. kuna; Dob. Pusana; Muk. kusana Bwd. yavula Tub. tari Dob. tanoa 'cultivated land'

The abbreviations used in this Table for names of languages are: Bas. = Basilaki Kir. = Kiriwina Tav. = Tavara Bwd. = Bwaidoga Muk. = Mukawa Tub. = Tubetude Dob. = Dobu Pan. = Panayati (Misima) Wed. = Wedau. a The Kir. kum draws attention to an interesting phonetic phenomenon of local development in certain of the languages. In the Muk. and Kir. groups only -m is permitted as a final consonant. When by development other consonants would become final, they are changed to -m, as in the examples here, to which *ka(e)n-i 'eat' may be added: Kir. Muk. kam; Wed. Ubir am. In all cases the addition of the final -i of the transitive suffix restores the proper consonant, e.g. Muk. kani; Wed. ani 'eat it*. 1

302

A. CAPELL

tebu 'sugarcane' telen 'swallow' uley 'worm' unap 'fish scale' wa(h0)ey 'water'

Muk. Wed. torn Dob. tonena; Muk. tonani Bwd. uloga Tub. kunai (-p > -hi > -0i) ; Motu unahi 'to scale a fish' Tav. goira

If the theory of eastward movements (one or more) of PAN speakers is correct, it might be expected that such retentions of final consonants, with or without supporting vowels, would be found in the mainland New Guinea languages, perhaps to an even greater degree. The occurrence of the phenomenon was pointed out by Capell (1943: 31, 34, 48, 51, 63, 66, 72, 84, 94, 102), and the instances are gathered up here for discussion in terms of Table X, alphabetical according to the PAN root. The area of concentration as regards the retention of final consonants, with or without support, appears to be centred largely on the southeast Papuan area known to anthropologists as the Massim area. Table X, of course, does not imply that the roots given are found nowhere else in southeastern Papua, but that only in these instances are the original final consonants preserved. Thus *matakut 'fear' appears in Pokau and Kuni makau, Kabadi mekau in the western area of Papuan AN, agreeing with Oceanic abraded forms much farther east than NGAN. So also Mekeo rorjo 'hear', Guregureu (Duau) nono, Bunama (Duau) nonoi, as against Bwd. nogalina, Muk. and Wed. nonori. Within the SEP area itself, the centre is about Bwaidoga, Duau to South Cape and the opposite mainland. This would be linked with Milke's Groups 1 and 2. Where else does it occur in NGAN? Examination to the north and west finds sporadic examples in Manam, where final consonants are allowed, yet wakara 'root' < (y)akal is the only case in a fairly extensive vocabulary. Graged and Yabem also permit final C, but they both abbreviate PAN roots to such an extent that sometimes only one syllable remains. The bulk of New Britain does not allow final consonants at all, but Tuna (Kuanua, Rabaul district) shows a number of the words under consideration still retaining final C: *(v)akal 'root' okor *atep 'roof' (ni)etep *degey 'hear' *ikan 'fish' •kulit 'skin' *lumut 'moss' *(m)pulut 'glue' *tebu 'sugarcane'

va-loqor-e en kulit limut bulit tup

In Manus, although final C is permitted, the PAN words do not seem to have preserved the final consonants anywhere. In New Ireland, Pala, which allows final C, produces only bulbulut 'glue', and tuh 'sugarcane'. Consonant retention appears

AUSTRONESIAN LANGUAGES OF AUSTRALIAN NEW GUINEA

303

next in the Bougainville languages, where it has already been illustrated. The areas of retention, then, in NGAN, whether a new final vowel is added or not, are: N.E. New Britain, Suau-Dobu and mainland, with Kiriwina, Bougainville AN, leading to certain New Hebrides languages of a seemingly archaic nature. In Malekula, Pangkumu shows (ir)ir 'hear' < *deqey; Baki has kulukuti 'skin' < *kulit, and for 'sugarcane', Pangkumu na-tuv, Bieri sob. In Tanna, 'hear' is -ateteliy, presumably from *taliqa 'ear'. In all instances only certain words receive this treatment, and these seem to form a fairly regular pattern. The retention of final C with a supporting V was noticed by Ray (1926:46-8, 49-50) but his observations have not been followed up. He illustrated from Mota of the Banks Is. and from Fijian; in the central Solomons, however, it is almost absent: he mentions only Sa'a lumute 'moss', but here also lumu is found. It will be remembered that Schmidt linked the central Solomons with Polynesia in many of its traits. In the remainder of the New Hebrides the final C is lost, and throughout Santo and Malekula (apart from the Pangkumu examples above) frequently the whole final PAN syllable is lost. Ray (1926:53) states Kern's conclusions regarding final consonants in PO as follows: 'Kern has pointed out the importance of these changed finals in studying the history of the languages. The IN ( = PAN) *takut must have become taku or mataku before it reached those places in which another consonant took the place of final t, when a transitive termination is added.' This is misleading if taken as implying such early loss of final consonant that when the system of transitive endings required replacement of a thematic consonant there was no memory of what the final consonant had been and some other was substituted. There do seem to be some instances of this procedure, e.g. *deyey becomes rorjo-mi in the Solomons and rorjo-da in Fiji, but this is not usual. If the final consonant is kept even sporadically, as the above examples have shown, as far east as the southern New Hebrides, then there cannot have been a complete loss of memory of the original consonantal ending. The case of *derjey is that of what might be called a 'weak' consonant — one that is frequently lost in any position. The matter is rather one of retention, loss or thematic revival in specific areas, which could not have taken place had the final consonant not still been present even at the end point of the movement — southern New Hebrides, Fiji, or wherever it might be. This point could well be of immense importance in theories of the AN peopling of the Pacific. Coupled with the equally sporadic accretion of initial consonant, to be discussed below, the two phonetic phenomena could provide guidance as to the movements of words. There would seem to be here a fairly regular progression, but what type of subgrouping can be based on it, or historical sequence derived from it, has yet to be investigated. Something will appear in subsequent sections of this paper. 2.4 Accretion of Initial Consonant A further phonological feature of certain areas of NGAN is the development of

304

A. CAPELL

initial consonants in such instances as Motu l-au < *aku 'I', and Tubetube karusi < *(g)ayus 'current'. This process has often been noticed as it were in passing, but it has not been studied. The quite unexpected appearance in Banoni of dzai 'fire', dzaso 'smoke', and dzorjo 'net bag' (the spellings are those of the S.I.L. recorders) recalls attention to the feature, and the following data will show that here, as in the matter of final C, there are links between SEP and Bougainville. The complex written dz by the S.I.L. recorders occurs only initially in Banoni words, taken in Matsungki village, and in a list of 170 words only in those here quoted, i.e. there is no objection to initial vowels as such. This implies that the first hearers (or bearers) of the PAN words heard some sort of consonantal initial in words of this type. Dempwolff and subsequent workers have all been hesitant about the initials of quite a number of words which in modern Indonesian languages have initial vowels. Dempwolff used a rough breathing, others have used h or q or provided for a glottal stop or some sort of hiatus. In regard to the specific Banoni words here under discussion, the lists show dzai 'fire' in the dialect (Nagarige) spoken in Tengeropaia and Koromaketo villages (presuming an initial e to be an article, as is frequently the case in this part of Oceania). In the northern dialect of Amun, vi- alone occurs in place of this dz- (e viavi, e viaso, e viotjo) in these words. The vi- prefix is by no means limited, however, to these three words in this northern Banoni; it is found also in e viyotsu < *kutu 'louse' and in 25 other words out of the 170, including adjectives and numerals, and it is not limited to AN words. The material is insufficient to provide clear information as to the value and origin of this prefixial vi-. Of the three cases in the (southern) Banoni list, Saposa provides only oij 'net bag'; for 'fire' it has ku-af < *apuy, to which Taiof w-af answers. These aberrant prefixes, for lack of data, must be disregarded here and some attention paid to the general matter of initial accretions. In SEP the development is I- in Motu l-au < *aku 'I', l-ahi < *apuy. Moreover, where PAN apparently had this still rather uncertain initial, whether V or C, and other areas have developed other accretions, /- is the normal reflex in Motu. In addition, Capell (1943:27) gave: 1-ada 'name' < PO *azan (PAN *ag'an), 1-ada 'gills' < PO *asar) (PAN *ha(n)t'ar>), but a-iha 'centipede' < PAN *ha-lipan. This phenomenon is local in Motu within the subgroup; the neighbouring languages (Mekeo, etc.) do not have it. In the Normanby Is. and mainland areas farther east, the accretion is k/P according to the dialect: Dobu Pusi, < PAN *utin 'penis'. Sinaugoro has both initial and internal accretion where word or syllable initial vowels might occur: yaro 'sun', yeba 'mat', yima 'hand', yio 'spear', yua 'fruit', sinaye 'bowels', nuya 'the inside', vanuya 'village', mayuri 'life' (Capell 1943:37). Phonetically all these are much abraded forms; mayuri 'life' is commonly in Oceanic AN mauri, and represents a far call from Eromangan nomurep, in spite of the latter's being so far east. In Sinaugoro also the accretion is still retained when a word is reduplicated: yauyua 'fruit'. The accretion affects AN and NAN roots alike.

AUSTRONESIAN LANGUAGES OF AUSTRALIAN NEW GUINEA

305

A digression at this point will show that the accretion has not begun in this part of AN territory. It was noticed by Stresemann in Amboina-Seran. His statements on the problem are: 'Some Seran and Amboyna languages regularly placeJ;he spirant j (= y) before an older initial a. It developed from the spiritus lenis of Orig. Ambon. This phenomenon is widespread in the eastern archipelago (sc. Moluccas). We meet it, for example, in Bonfia, where the J has often developed further to dj, in Kor, Teor, and in the languages of the Kei Islands. Also in a very noteworthy way in the language of the so-called Papuans of Aiduma, where y has further become / ('I' is laku < jaku < *aku; 'fire' is lawi < jawi < PAN *apuy). The examples given hitherto are those in which initial a is Orig. Ambon. On the other hand, if the initial a has arisen through loss of a consonant, the addition of j does not take place. This can only be explained by the fact that a consonant does not vanish without leaving a trace, but has left a glottal stop behind. In Bonfia, where the same rule holds, the tenuis k develops from this glottal stop, when a word ending in a consonant precedes it, e.g. Pasawais, "cassowary", tuin kasawais, "to shoot cassowaries".' (Capell 1943:37, translating from Stresemann 1927:114). In eastern Indonesia, then, similar phenomena are found. Stresemann's explanation may be right, but in any case there are some awkward exceptions. Motu l-ada 'name' points back to PAN *ag'an, not to rj-aran, a form of quite wide occurrence, in which the y- has been assumed to be an article, a facile but unenlightening explanation; but the similar *rj-ipen 'tooth' appears in Motu as ise, not *lise. Here again, certain developments seem to be present over a wide area, but it is not possible yet to account for them historically. It may well be, of course, that these areal similarities present convergent development rather than derivations from a single source. Certain of them occur in eastern Melanesia, such as *yaran forms for 'name', but accretions as a whole do not. Unfortunately Ray (1926) did not discuss the matter of initial accretions at all, but from the above it will be plain that at intervals along the way between Amboyna and Bougainville the process is visible in many of the languages. In Suain, for example, there is djau 'you' < *kaw > *(k)au > dzau; in Megiar, inland west of Madang, *k-as < *asu 'smoke'; in Manam k-urita < *quyita 'octopus' (but Wogeo urita); in Graged l-at < *atas 'on top', in Sio k-awa < *awaq 'mouth' (but here Graged awa); in Wampar dzif; Atsera dzaf < *apuy, 'fire'; in Taemi dz-ilo < PO *izun < PAN *ig'un 'nose'. These are examples which may be linked with those given above from other parts of the NGAN area. They seem to be random, and it is not possible in most cases to lay down a regular change such as 'accretion of consonant to initial vowel'. Perhaps it may even be said that the PAN originals are not quite certain in all cases. If it were better known just which was the initial sound of V-, q-, and h- in some of the PAN examples, the subsequent events might become clearer. Some of the words, such as *apuy 'fire' tend to develop an accretion fairly regularly. Others, such as *uyita (PAN form) 'octopus', do so less regularly.

306

A. CAPELL

2.5 Loss of Consonants Commoner than initial accretions is the reverse process, loss of consonants, and this may help to distinguish subgroups. Such losses occur in both initial and final positions, and Tables XI and XII show the types of loss that take place in the Southeastern Papuan area. This is intended simply as a specimen. Such phonetic changes have not yet been worked out in detail for all the areas. A full study for one area is included in Goodenough ( 1 9 6 1 a + b) for the Willaumez Peninsula languages and the Nakani subgroup in New Britain. In the subjoined Tables XI and XII, the PAN consonants are listed down the left hand side of the page, first as initial (Table XI) and then as medial (Table XII). The final position is not involved, as in most of the languages final consonants are lost, and the situations that arise when they are retained have been discussed already. Such finals are usually treated as medials in the historical process, and the conditions of their retention are then much the same as though they were medial. Where a medial consonant is changed but not lost in a language, the same type of change takes place for a final consonant, e.g. if a medial */ is lost before i or u, then a final */ is treated similarly. In Tables XI and XII a plus ( + ) in a column right of the name indicates a loss of that particular consonant in the given language. Thus -k- with + in the Motu column is to be read as 'medial -k- is lost in Motu'. Where no plus is inserted, the consonant is kept or replaced by some other—whether there is change, or what the change is, does not matter, as the nature of the change is irrelevant in the present instance. Only losses are being studied, not changes; therefore only such consonants as have been subject to loss are entered in the columns. Highly exceptional losses (i.e. apparently irregular changes) have been disregarded for the present purpose, and the aforementioned development of final -m in the Mukawa area is treated as retention of the consonant. It will be immediately apparent that more medial than initial consonants are lost. In some cases, for instance, there is loss of initial bilabial plosives, and in these instances the medial bilabials are also generally lost. In general where a language is used as representative of the one subgroup, the one with fewest losses is chosen (e.g. Mukawa rather than Wedau which loses /k/ and /s/, both retained in Mukawan) and fewest changes (again, e.g. Mukawan, which keeps /p/ where Ubir changes it to /f/). Where two entries have been given, as Mukawa w, marked as + ( < k), the entry is to be read as 'accretion of k to the initial vowel that results from loss of w'. Although the Tables set out to show the retentions and losses for SEP, some other languages where the information was readily available have been added, so that an idea might be given of what similarities and contrasts will be shown when other areas are studied in detail. A few comments on the Tables may be made. A full examination of all the N G A N languages has not yet been carried out, so that it is not possible at this stage to make a final judgment on all the sub-classifications made by Milke and others.

307

AUSTRONESIAN LANGUAGES OF AUSTRALIAN NEW GUINEA TABLE X I

Loss of Initial Consonants, chiefly in SEP

Language

Mekeo Motu Hula Suau Tubetube Dobu Bwaidoga Mukawa Kiriwina Yabem Taemi Atsera Wampar Sio Biliau Graged Manam Wogeo Suain Sabon Papitalai Ninigo Tabar Lihir Nisan E Mira Musau Lavongai

Consonant p-

b-

+ +

+

t-

z-

k-

+

+ + + + + +

+

g-

rj-

1-

+

+ +

+

Y-

q-

+

+

+ >k

+ + >k

> k

(+)>b (+)

(+)>g

+ + +

+

w-

+ + + +

y-

308

A. CAPELL TABLE X I I

Loss of Medial Consonants, chiefly in SEP Consonant

language -PMekeo Motu Hula Suau Tubetube Dobu Bwaidoga Mukawa Kiriwina Yabem Taemi Atsera Wampar Sio Biliau Graged Manam Wogeo Suain Sabon Papitalai Ninigo Tabar Lihir Nisan E Mira Musau Lavongai

-b-

+ + +

+ + + +

+ +

+

+

+

+

+ +

+ + + +

-k-

-3-

?

+

(+) + + + + +

+

-1-

+

+

+ (i,u)

+ +

+ +

+ +

+ (i)

+

+

+ +

+ (i)

+ +

+

+

+

+

-Y- -q- -w- -y-

+

+

+ + +

-r-

+ (i) + (U)

+

+

+

+

+ +

-z-

+

+

+

-t-

+

+

+ +

+

+ + +

+

+ + + +

+ + +

+ + +

+ + + + + + + + + +

+ +

+ +

AUSTRONESIAN LANGUAGES OF AUSTRALIAN NEW GUINEA

309

The necessity for these Tables has sprung out of the special conditions of SEP, where such losses and retentions help to establish subgroups. This can be seen from the fact that Kiriwina, an A N j language, loses only initial *b, whereas the AN 2 languages west of it lose more. Those on the islands again show more agreements with the mainland languages than with Kiriwina, and also amongst themselves there is considerable agreement. Tubetube and Dobu agree with Motu and Mekeo in the loss of initial *p, but go a stage farther in losing initial *b also. This is of interest, as Dyen (1965:305) has stated that there is a 'phenomenon observed among the languages of New Guinea, Melanesia, Micronesia and Polynesia. This is the merging of *p and *b.' If such a merging had taken place, as he suggests, at some stage before the movements east of Formosa (with which the article quoted is concerned), it is hard to see how initial b could have been retained in Motu and Mekeo, because it would not have been there at that period. Yet *banuva 'land, settlement', which in Dobu is anua 'house', appears in Mekeo as parjua and in Motu as hanua 'village, place'; *buwah 'fruit' appears in Dobu as ua, but in Mekeo as pua, in Roro and Kuni as bua, and Pokau as vua. Yet it ought not to have been present in the western languages if it had been lost in an earlier part of the movement. As for *p-, examples appear such as Dobu e-isa < *pira 'how many', but Mekeo pika, Roro bita, Pokau and Kabadi vida, and also Kuni bida. Of the medials, 'sugarcane' *tebu appears in Dobu as tou, in Motu as tohu, Aroma as obu. Medial *-p- is kept in Mekeo, e.g. *apuy 'lime' is Mekeo apu, while in the only example available in Dobu *puluh 'ten' seems to be represented by a simple u in sa-na-u < *{e)sa-Tja-puluh. Such a generalization as Dyen has made in the article mentioned needs to be checked against the details of the individual languages. This is a large task, and one of the reasons that no full treatment of them has been undertaken in Tables XI and XII is just that in such an article as this they would be out of place. There is still great need for minute examination of these N G A N languages individually. Some of the data is still inadequate for such an examination, and research is called for. Summary findings on the phonology of NGAN to date would seem to be somewhat as follows: 1. Some consonants of PAN suffer little or no change — but these are few, chiefly /m, n/; in some cases /s/ (see 4 below); *w and *y being glides are ambiguous as regards interpretation. 2. Some consonants of PAN undergo special types of change in different areas: *t > k *t > r *t kept *p > v *rj > n, 0 *j > r *y > r

Mekeo Markham Valley New Ireland, New Britain SEP, Tuna (kept in Bariai), Torau SEP; kept in New Britain Central SEP, but /s/ in eastern SEP and Torau Most of SEP; Bougainville (y lost); r/1 New Ireland and New Britain

310

A. CAPELL

3. Nasal combinations produce fairly uniform changes in different areas, and this indicates that they may have begun to change even in the PO stage. Examples (not exhaustive) are: *mp > p, b

New Britain and New Ireland; in some cases -mp-

*-mb- as well as *mb- > b *nd-, *-nd- > d-, -d*rjk-, *-qk- > g-, -g-

-P(SEP), Torau, New Ireland; in Tuna -mb- > -pin New Ireland and New Britain in SEP, New Britain, New Ireland

4. The changes of PO sibilants (as reconstructed by Milke) allow four sets of results: *s, *z > /d/ *s, *z > /s/ *s > /s/; *z > /r/ *s kept; *z lost

SEP, western area N G north coast except Graged Manam, Tuna, west New Britain, Misima Graged, Markham Valley, Mukawa

5. Loss of consonants, chiefly p-, b-, t-, q-, -p-, -b-, -t-, -k-, -I- as below under (6), -y-, -q- as shown in Tables XI and XII. 6. Final consonant retention is subdivisible into two kinds of phenomena: (a) simple retention, where there is no change at all, and (b) retention by addition of a supporting vowel. The types occur in the following areas: (a) simple retention mostly in areas where final -C is allowable in any case: Tuna, parts of New Ireland, Graged. (b) retention by addition of final -V, usually harmonising with the -V- before final -C, as in Roviana of the British Solomon Islands. In N G A N this occurs in SEP (Dobu, Bwaidoga, Mukawa), and in Bougainville in Banoni, Uruava, and Torau. Retention by the addition of a final 'weak' vowel, such as -z, with sound changes to ~t to -si (SEP south coast and islands), Bougainville (Uruava and Torau), Buka. 7. Accretion of initial consonant replacing original q-, which is completely lost in other areas, leaving the word with initial V. The consonant developed varies: Malol and Suain (dz), Bougainville {dz), Markham Valley and Taemi (dz), Megiar, Manam, Mukawa (k), Motu (/). 8. Loss of the entire final syllable, often reducing the word to a monosyllable or at most a disyllable. This is seen largely in Malol-Suain, Admiralty Islands, New Ireland and dependencies, and it is picked up farther east in Malekula of New Hebrides and in the southern dialects of Efate. In such cases there may be local and parallel development, seeing that final consonants are retained even in Eromanga and Tanna. The occurrence of final consonants in the southern New Hebrides shows that PAN roots still retained them when they arrived here (if one reckons on the theory of a movement from South East Asia). If one accepts a theory of distribution of languages from some already Oceanic centre, as is presupposed in the theories of Dyen and Grace, the matter is easier, but in any case some stages of movement are indicated,

AUSTRONESIAN LANGUAGES OF AUSTRALIAN NEW GUINEA

311

though they will be different in each case. The discussions on this point are still far from finalised. 3.

NGAN

Vocabulary

Study of the phonological phenomena amongst the AN languages leads to a study of the shared vocabulary. Each language in the AN regions, whether within or without New Guinea, consists of two kinds of vocabulary — N A N and AN. The AN vocabulary is in the minority, at any rate in all areas except perhaps the central Solomons; N A N words appear to have a majority, and in many parts a very high percentage majority. Dempwolff reconstructed about 2,000 words, to which some others have been added since; at the outside, acknowledged PAN vocabulary may consist of about 2,500 words. This number of words does not make a language of full efficiency for daily life. Fauna and flora limited to an area may be expected to bear local names, though these may spread. In N G Pidgin many such words are of Tuna origin. Ethnobotany deals with this particular linguistic phenomenon and needs much more study yet. It would seem that AN speakers have brought into the Pacific almost as large a proportion of food items as Europeans brought to Australia, or at least have imposed their names for them if they did not bring them. These two strata of words need to be examined in the N G area. The present concern is with N G A N only. Something will have to be said about N A N vocabulary also, because this may have considerable importance; in fact it may be just as illuminating as the AN vocabulary. It is extraordinary that students of Oceanic origins have been so long content to neglect the N A N factor. The use of lexicostatistical methods may seem to have produced a blind spot among students as far as 'whole language' evidence is concerned. Future study will need to pay more attention to words of C. A. Schmitz, who argued from the anthropological angle for substrata to be studied in Melanesian languages (Schmitz 1961). He says in this connection: 'The proof of belonging to the Austronesian family is based on lists of 200 to 400 words, which in the admittedly large vocabulary and richness of forms of the Melanesian dialects is after all only the smaller part of the evidence. It is then to be considered throughout that in the Melanesian dialects we have to do with an independent language family ... that has been strongly modified by Austronesian elements, belonging to the southern group of the Austronesians.' The linguist, approaching the matter in a way different from the anthropologist's, may find fault with the manner of expressing the ideas in the above passage, but he can have none to find with the viewpoint and the call. The first subsection of this section will say something briefly about the PAN vocabulary as a whole, as far east as Fiji; then the recognition that there are regional vocabularies, both within N G A N and beyond it. Of the words indicating, e.g. 'sleep, lie down', one may appear in one area and the other in different areas, while both are P A N ; of three terms for 'house' all are PAN, though one originally did not mean 'house' at all. This is a phenomenon which has not received enough attention hitherto,

312

A. CAPELL

but it may prove to be valuable. Thirdly, there is shared NAN vocabulary within NAN and outside it. Consideration needs to be given to whether this is accidental or not. On these vocabulary occurrences, then, some of the subgroupings may be based. They can be tested later by examining the incidence of grammatical phenomena to decide whether these also group in the same way. If they do, there is strong evidence of a special influence in such an area. 3.1

AN Vocabulary in the NG Area

Out of the ± 2,500 PAN roots accepted, less than 400 are found commonly occurring in NGAN. Capell recognised 320 in SEP (Capell 1943:109-15), together with some words which he called 'unplaced'. There were 36 of these, all occurring in some parts of Oceania, but not included in Dempwolff because they did not seem to have reflexes in Indonesia. Grace (1964) comments on this situation: 'It has furthermore been suggested that a considerable number of probably lexical innovations (vocabulary items restricted to the Eastern Austronesian languages but widely distributed among them) will eventually be compiled (Capell 1943:164-8)', and he goes on: 'There are perhaps certain other changes that may be recorded as Eastern Austronesian innovations, e.g. the irregular loss of PAN *-k- in PAN *aku "I"; the irregular loss of PAN *R in PAN *maRi "hither", and, following Milke (1961), the development of Proto-Eastern Austronesian *suRi irregularly from PAN *(dD)uRi "thorn". I personally must also confess to being motivated by a general impression that most of the languages which I attribute to the Eastern Austronesian subgroup differ grammatically in similar ways from the languages of Indonesia and particularly from those of Dyen's Hesperonesian group.' Some of the above-mentioned words have been accepted since by Milke, Grace, and others, and they may well be admitted as at least PO (if Milke's intermediate stage of AN is itself accepted). The difficulty with them is that they seem to be very rare in the Indonesian region. There are also many others that do not come eastwards, and they raise questioning also. It is not relevant here to examine these in terms of the ultimate status, though some of them do occur in NGAN as a whole and not only in SEP. They should form the subject of a future study. The number of PAN words in NGAN will be about 350 at the most — about 1/7 of the 2,500 words acknowledged as PAN. While this is not in itself a great number, some others appear within the bounds of geographical Melanesia, so that perhaps a total of some 600 of Dempwolff's vocabulary appears. Others, again, seem to be limited to Polynesia. A considerable number do not seem to occur outside Indonesia, as mentioned. Thus the whole question of the validity of Dempwolff's list is raised; if that list passes the test, how can the fact be accounted for that some of the words appear in certain areas only? The test question itself is, why should so many words be limited to Indonesia, especially if the theory of a spread of AN from somewhere in the present Melanesia, e.g. New Britain-Huon Peninsula, be accepted? The theory seems to be condemned by the vocabulary distribution, in spite of its strong lexicostatistical base.

AUSTRONESIAN LANGUAGES OF AUSTRALIAN NEW GUINEA

313

S.H. Ray (1926:595-7) discussed the nature of the AN vocabulary which appears in island Melanesia, and much the same distribution of groups of words is to be seen in NGAN. Ray stressed the uniformity of the PAN vocabulary 'as it appears in the comparative tables' — and also pointed out that these tend to exaggerate the PAN element in the M N island languages. This also is true for New Guinea languages as well, in fact still more so, for their over-all PAN content is in many cases less. Ray mentioned that 'the articles and actions named are those of common intercourse, and that as a rule out of doors'. There has been much discussion about the occurrence of words for the betel nut and the accessories used by betel nut chewers. W. H. Rivers in his History of Melanesian society (1914) posited a special migration of betel nut people, as he called them. This suggestion has been dismissed by later anthropologists; but nothing has replaced it. Betel nut is chewed in similar ways and with similar accessories over a certain area, and many of the terms involved are AN: how did they reach there? Terms for spiritual concepts also are widespread. They have been studied by Fischer (1965) both from the viewpoint of the anthropologist and from that of the linguist, and his findings are very instructive, although strictly speaking they lie outside the present paper. The question of how certain'practices reached their present locations is admittedly not a linguistic problem; but the linguistic answer to the origins of the words used should not conflict with results reached by other disciplines. One example may be given of what is being argued. The word given for 'ghost' (Geist) by Dempwolff is *sumaijes. This word has no cognates outside Indonesia. A word that is of very wide occurrence in Polynesia is vaerua, which is not accounted for in Dempwolff. It occurs also in NGAN, e.g. Tabar, where virua is 'the souls of those dying unnatural deaths, e.g. suicide, warfare, drowning, sharks. They"'stay in the clouds in the upper air, descending to the ground by the rainbow, and are invoked in rain-making and fishing.' Is this a word that was taken over ready-made from some Polynesian source? Fischer (1965:339) treats it as such, and gives examples from most of Polynesia, SEP, Tuna, Bariai, Tabar, Mono (Solomon Is.), and Manus. Its other relatives appear to be in Indonesia, and Fischer seems to incline to the meaning given by Capell (1943:270) that its root meaning is 'second self'. This depends on the relationship of the -rua element to PAN *duwa 'two', and the first presumably with PAN *bayi- 'reciprocal', the total compound being 'mutual relationship to each other' of soul and body. Linked with this in Capell's account is huarj, a term used for the 'seat of feeling' in various Bornean languages, which then would appear to link with Eromanga n-owa-n 'his heart' in the same sense; the root nua-, nuanua- is common in the SEP region for 'heart' as seat of feelings, 'inside' in general. Problems of this nature arise in semantic study of the AN languages, both in N G and elsewhere, and they require other types of studies — such as Fischer's — in addition to purely linguistic studies, to produce the best results.

314

A. CAPELL

3.2 AN Vocabulary in the NGAN Area Vocabulary of practically every semantic group mentioned by S. H. Ray is found in some part of the N G area. N o itemization has been carried out that would show what recognized PAN vocabulary is not found at all, nor what words are of areal occurrence, i.e. what might be called 'strata' levels. To such a problem there are two approaches. One is the examination of the occurrences of groups of P A N words which may be associated with each other in different areas, but absent in an intermediate area, as an interconnected group. Such examination might yield useful results, in showing the spread of word groups, but in many of the languages the available dictionaries are insufficient to prove the difficult negative — that a word is really absent and not found perhaps in rare deeper levels of the language. Another question not yet studied is that of doublets, one being a 'surface' word, the other a more technical and specialised word. An example of what is intended here may be given from outside the N G area. Atchin is a small island off the northeast coast of Malekula, New Hebrides; the culture of the area was described very fully by Layard (1942). Here the word tsali 'rope' is clearly < PAN *!ali; but there is another word tali, a special term referring to the rope by which a pig is tied to a post ready for sacrifice in the Maki rites. This word breaks the Atchin rule that PAN *t-> Atchin *ts-. The most probable explanation of the phonetic divergence would be that tali is a special word introduced by the people responsible for the Maki rites. Similarly in SEP and elsewhere, where the betel nut is chewed, the word for betel nut is usually bua, no matter what b- ought to become according to local sound laws. In Motu *b- > h, as in PAN *bulan > Motu hua; but betel nut is still bua in Motu. It looks like a foreign word throughout the entire region. Such sociological examination of lexicon in N G A N — or indeed in other regions of the Pacific — have not yet been undertaken, except for a specific field such as that of Fischer (1965) in terms for the soul or Milke (1938) in terms for kinship relations. The second approach to vocabulary identification is the purely linguistic approach, where types of sound change are established as in classical philology, with no reference to any field outside linguistics itself. This is the method normally employed, e.g. by Capell (1943) and Milke (1938,1961, 1965). It is, obviously, the method best calculated to attain the results desired. However, there can be the exceptional cases, as in those cited in the previous paragraph where non-linguistic factors enter, factors referred to by de Saussure as 'outside' or 'external factors'. For a full examination of the lexicon of a language both types of research are desirable. One of the notoriously unsatisfactory sections of Dempwolff's work is the number of doublets he left unexplained, principally the facultative occurrences of prenasalised stops. One cannot help feeling sometimes that some at least of these are not phonological matters at all, but due to extralinguistic causes such as those illustrated here. A particularly interesting case is that of words for 'house' in N G A N , which will be discussed below (p. 318). The apparently arbitrary occurrence of three different roots for 'house' surely has some

AUSTRONESIAN LANGUAGES OF AUSTRALIAN NEW GUINEA

315

connection with the living habits of the people who brought them, or with living habits that were theirs in an earlier home. It is recognized that different derivatives from a single root may characterize different periods of borrowing — as in the case of Latin regalis, Engl, 'royal' and 'regal'; even where sound changes are regular, however, there may be different origins if they can be traced. 3.3 AN and NAN in Area! Occurrence The first question to be taken up is the occurrence of AN roots in certain areas of New Guinea. Some AN roots undergo remarkable change throughout the N G area, and their links with Eastern Austronesian seem, at least on the surface, to be haphazard also. A few examples (Capell 1943:19) will show that some roots undergo variations within different languages of one subgroup, and in some cases also an AN root will be found in one language but not in another in close geographical contact with it, e-g. 1. 2. 3. 4. 5. 6.

English bowels evening flesh hear lime smoke

7. 8.

thin water

PAN *tinahi *yabi (PO *pinsiko) *deqey *apuy (PO *apu, 'ashes') *nipis *wayey, (*(dd)anum)

Mekeo ige-na rjapi pikio rogo apu aku

Motu (bogarau) (ado)rahi hidio (kamonai) ahu (kwa)lahu

ma-gipi-na vei

(severasevera) ranu

This is a particularly striking set of words, about which the following facts may be noted: 1. A number of the PAN roots are found in both Motu and Mekeo, the relevant sound changes taking place in the expected manner. This applies to No. 3 (where difficulties arise in another way) and to No. 5. 2. In some cases one language has an NAN term while the other has an AN; this applies to Nos. 1, 4 and 7. The point of interest here is that it is Mekeo and not Motu which has the larger number of PAN roots. 3. In some cases Motu has a PAN root combined with an NAN prefix which cannot be identified. In No. 2 ado- is probably PAN *a(n)dav 'sun, day', so that two PAN roots are conjoined. This is not without parallel elsewhere in Motu, e.g. hanuaboi 'darkness' > PAN *banuva 'land' + *beyi 'dark'. In No. 6, however, the compound has a first element that does not seem to be PAN, if a breakdown into kwa-l-ahu is right; the -/- is the frequent Motu accretion of /- before an initial vowel (pp. 303 ff.), whether this be original or represent a lost glottal stop. This means that the word came into Motu as ahu, and has later been compounded. Although the Motu

316

A. CAPELL

dictionary gives many words beginning with kwa-, there is nothing that would throw any light on the present instance. 4. Word No. 3 is an example of words not occurring in IN, and so not recognized by Dempwolff. A possible original is suggested by Grace at the P O stage, in a "Proto-Oceanic finder list" issued by him in mimeograph form in 1965. For his *pinsiko he says, 'the vowels are doubtful'. It may be rather that the vowels are right, as the word seems always to occur with these vowels, but the doubt rests on the internal consonant cluster, for which the writer can see no reason. The finally approved form of the root should certainly be accepted into PO at least, whatever the situation regarding PAN may prove to be. 5. Word No. 8 is of considerable interest. Two words for 'water' are recognized by Dempwolff: *wayey and *(dd)anum. Their spheres of occurrence in Eastern Austronesian overlap. Other examples of this phenomenon will be studied in the following subsection. While the second of the words is not documented in Mekeo, and ranu is the normal word in Motu, Motu also has sina-vai 'river', a compound of *tina 'mother' and the first root for 'water'. P N shows terms for river such as Samoan and Tongan vai-tafe 'water-flow', but 'mother of water' is an Asian combination, as in Thai mee-nam. The nearest Oceanic approach to this term seems to be the Florida (British Solomon Is.) mbeti-tina 'water-mother', where the elements are in the opposite order. N o explanation of this combination of roots, one of which does not occur independently in Motu, while the other does not appear in Mekeo, is forthcoming. The whole structure of the SEP vocabulary in the languages east of Motu is of interest; something has been shown of it in the section on Phonology here, and other aspects appear in Capell (1943:39-41). In this latter section the occurrence of many P A N roots in N A N languages such as Dimuga is noticeable. Here they are obviously loanwords, as the structures of the languages are NAN. The best known N A N language in this region is Mailu, which also has numbers of PAN loanwords, though its grammar and basic vocabulary is entirely NAN. Throughout SEP the P A N element is very clearly a veneer and not a foundation part of the languages. This feature is clearer in all the AN 2 languages than in the AN 1 ( and it is fairly evident that the AN 2 languages are not PAN derivatives in the same way as those of A N t both in New Guinea and in the eastern parts of the 'Melanesian' groups. The statement just made can be tested by an examination of the vocabulary of a language such as Graged, for which Mager (1952) gives P A N roots wherever he has been able to find them. The number is not proportionally large. Whether those that occur in SEP also occur here, or whether Graged has any appreciably different set of roots has not yet been tested, but it could easily be done. Dempwolff's Sio grammar, appended to the Ms. Sio dictionary, lists only 80 PAN roots representing what he could identify in the language. Sio is an A N i language, so that it looks as though examination might show that the N G A N as a whole is less in quantity than that of AN further nelara 'the menfolk', using a suppletive form, or Atsera rosa, prefix for pluralizing kinship terms: rama-y rosa-t] 'my fathers'. Of more general interest are the systems of forming nouns of various types. Some nouns carry general markers that seem to indicate simply that they are nouns: this is the case chiefly in SEP. Others carry prefixes showing their functions, as in the appended diagram: Noun general markers -a

-na

-ra

functional markers abstract

agent

local instrument

I I he- bai-

I tau-

I kaba-

I kai-

These are scattered in different parts of SEP and provide interesting study of origins, which has not been fully carried through as yet. Thus, Kiriwina kuria- 'pot' points back to PAN *ku(dD)en 'pot', but what is the -a? Brandstetter pointed out the IN usage of a suffixed -a to mark definiteness, and this may be a similar case. The functional markers are particularly SEP, and are exemplified in such cases as Motu kaha 'to help' > hekaha 'help' (n.). Mukawa beyebeye- 'show' > bai-beyebeye-na 'manifestation'; for the agent: Motu kaha 'to help' > he-kaha tau-na 'help its man'; Dobu gimwani 'trade' > togimwani 'trader'. The place prefix kaba- is of special interest because forms of it occur as free forms in other parts of Melanesia at least: Misima mina 'stay in a place' > aba-mina 'dwelling place'. This occurs as Manam kaba 'place', and in Micronesia aba 'land', 'place'.

330

A. CAPELL

Instrumental prefixes are true PAN elements and are found in the Indonesian area and eastward in Bank Is., New Hebrides, and Fiji, especially for /-. Sometimes a considerable degree of complication is reached in parts of SEP, e.g. 'a companion', from mata 'face' and gamo Mukawa bai-bai-mata-gamo-gome-na 'voice'. This is less usual in the ANX languages, and is at its maximum in SEP. All the languages, however, possess methods by which these functions are indicated; only the more outstanding ones are mentioned here. There is need for deeper study of these formations, the materials for which (at least) are of PAN origin, and their spheres of occurrence may serve to highlight directions of movement. The presence or absence of such prefixes and suffixes does not seem to characterize either A N t or AN 2 . This is another indicator of the probability that the distinction has developed locally in each area, and been added to a common PAN element in the languages. The lexemes commonly called 'articles' are absent from wide areas, especially N G A N 2 group. Yabem and Graged lack them, as do Atsera and Wampar and the languages west of Manam. Overt indication of syntactic function, such as Motu ese, na (p. 327), seems to be limited to Motu itself, though Tuna a and ra serve to mark such function in a different way. The a of Mengen (and the N A N Sulka) does not serve such a purpose. In insular M N farther east, however, na is almost universal as a noun marker. In Manam, the formative -rja which forms nouns from verbs is particularly noticeable, and links with languages farther east, and also with Polynesia. O n the other hand, possessive formations are shared with insular M N in general; in a few cases there are classificatory forms which recall Micronesian (e.g. Truk). The Mindiri and Sio possessives recall Fijian very strongly: this fact has been remarked on for Mindiri earlier (p. 253). There is not in N G A N the subdivision of independent possessives into smaller classes as they are found in Mota and Fijian amongst other eastern languages; where any subdivision at all is found in N G A N , it is a matter of dividing food and drink off from general possession. Notes on this fact will be found on p. 286. In the same place attention was drawn to the repetition of the cardinal pronoun redundantly before the noun as in Motu lau egu ruma 'my house', and Graged r/a inag ab is the same. This is paralleled in the N A N languages and would seem to come from that source. ii. Verbals The various patterns of the verbal complex in N G A N sentences may be diagrammed as follows, using s = subject marker, in the sense previously defined (p. 327) and t = tense. 1. Intransitive. V == s ± t + v expandable t o : S = (NP) + ( s ± t + V ) ± 0 The second pattern points out that when a noun subject occurs, the subject marker

AUSTRONESIAN LANGUAGES OF AUSTRALIAN NEW GUINEA

331

must still be repeated before the verb. In a few cases this does not happen, but it is rare. In Manam the pattern is S + V ± 0 + t and in Matukar (AN 2 , hinterland of Madang) S + V ± O ± t, while the neighbouring Megiar has S + V ± O + t, i.e. the tense marker is essential in Megiar, while in Matukar it is not. The details of the sentence arrangement have been shown diagrammatically for each of ANX and AN 2 groups (p. 327). Where an object (O) occurs as a pronoun it is generally appended to the verb itself, as shown above by ± O . The pattern then is AN 2 : S = N i + N 2 + sm i t + V i O, and A N , : S = N , + sm ± t V ± O, where O is either N 2 or a pronoun. The verb phrase thus contains generally 1) a subject marker; 2) a verbal root or base; 3) a marker of mood or tense; and 4) a marker of object, which is optional. Roots may be simple or compounded, and there are prefixes and suffixes which are in large part PAN material, and form a normal part of the lexical content studied in section 3 above. In a paper of this nature they lie outside the sphere of examination, though it can hardly be said that as yet their occurrence in N G A N has become a 'current trend in linguistics' apart from what light they may have to throw on phonological developments. The additive components to verb stems commonly found in this area and identifiable as PAN in origin are: PAN *pa-, *paka-, 'causative': most commonly in this area pa- or its phonetic derivatives. There is undoubtedly linguistic history involved in the distinction of areas where pa- and where paka- are found. In Manam the form is aka-, which can be traced into Eastern Austronesian and Polynesian; pa- is much the commoner in both north and south: Motu he- as in he-diba 'cause to know, inform'; Tuna va- in va-loyore 'cause to hear'. However, many of the languages lack it, and have local causatives. In Dobu, for instance, the transitive form of a naturally intransitive verb acts as a causative: i Peno 'he lay down' > i Penoigu 'he made me lie down'. The Kiriwina group, which classifies its nouns and its verbs, has a set of causative prefixes which point to a classificatory system of some sort there also, but details do not seem to have been worked out. In Kiriwina causative prefixes are ka-, katu-, kata-, ko-, ki-, vaka-, io-, no-, lo-, mitu-, vitu-, va-, vi-, to-, si-. Some of these can be assigned special uses, e.g. ki- 'make by hand', lo- 'cause by going', vitu- 'do by way of example', va'come and do', to- 'make to stand', si- 'make to sit'. PAN *bayi-, 'reciprocal' is preserved in Tuna as vara, prefixed to verbals: vara-gire 'see each other', and bar-, prefixed to kinship terms, bar-tama-na 'father and child'. This usage extends eastward into the Solomon Islands at least, but is not shared in any other part of NGAN. In Motu bayi- has become homonymous with pa-, each giving he- as a prefix, but in the reciprocal sense he- ... heheni, as in he-diba-heheni 'give each other information', is the normal usage. The more normal phonetic form is bai- (as in Roro), vei-, hei- in different areas. In AN outside New Guinea there is usually a set of transitive suffixes to verbs. These occur in N G A N but not so regularly either in form or use. In Tuna the suffix

332

A. CAPELL

-e is found: gira 'look' > gire 'look at'; alir- 'swim' > alire 'swim off with something'; burnt 'be afraid' > burutane 'fear something'. This is parallel to (though it is hard to say that it originated from) the -i suffix found in parts of IN, and C +i in parts of Eastern Austronesian. Manam again has the clearest set of suffixes: mule 'come or go back' (also found in the Banks Is.) > muleaka 'bring back'; boli 'tell lies' > boliak(i) 'slander'. This is another aspect in which Manam is very similar to Eastern Austronesian languages. Markers of the subject, also called verbal pronouns, where they occur, may be traced back to a hypothetical set as follows: 1. incl. 1. excl. 2. 3.

Singular — *ya, *a *ku, *u *i, *e

Plural *ta *ma *kwa, *wa si, *se ^ G1

These are slightly different from those found in more easterly regions, but not radically so. A full tabulation is given in Capell (1943:219-20) for all SEP, but not for the northern regions. In the IN area these forms are either 'outliers', e.g. islands west of Sumatra, or 'central', e.g. Toradja areas of Celebes, and to some extent eastern IN. Capell remarks that 'In SEP, and indeed in MN generally, the shortened forms (of the pronouns) are derived, not from IN (i.e. PAN) forms, but from the local pronouns.' That seems to suggest that the system of subject markers is not PAN; it arose at some subsequent point, but apparently before there was any great separation of the individual languages. This is a matter for further enquiry. The subject markers combine in some cases with tense indicators into compounded particles, illustrated for SEP languages in Capell (1943:220ff). In northern New Guinea and the islands to the east, similar systems are found, but the details vary quite a lot. Similar principles are often disguised by the practice of sometimes writing them as parts of the verb, sometimes as separate particles. Markers of tense and mood may either be combined with the subject markers or be separate particles, which may or may not immediately precede the verb stem. The languages vary greatly in this regard. The distinction of aspect systems from mood and tense systems has already been mentioned (p. 286). The languages do not as a rule have a passive voice; an impersonal 'they' is in general used to express an act whose agent is not known. Tuna has a very clear tense system, with no trace of aspect. The forms are as follows, in each case a particle placed between subject marker and verb: yau gire 'I see it' — present; yau ga gire 'I saw it' — past as in the corresponding English; i na gire 'I shall see it' — a future. The only form which may be regarded as perhaps aspectual is tar, ter, in yau tar gire 'I have seen it'. The aspectual languages are illustrated in Kiriwina. In this group, three markers are used:

AUSTRONESIAN LANGUAGES OF AUSTRALIAN NEW GUINEA

333

i. person markers only, denoting acts which are independent of time. They mark a fact without expressing the time of its occurrence: i sisu 'he is', exists independently of time; i sisu Kitava 'he lives in Kitava'. Completion of act, past time, is marked by bogwa, which is rather a marker of state attained: bua bogwa i yagi 'the betel has now ripened'. ii. A root I- may be prefixed to the subject markers, and marks a definite past. Malinowski says of it, 'At times it places the action into a regular past, accomplished state; at times it gives only emphasis.' It may be combined with bogwa. iii. Another particle b- is more or less Irrealis; it serves to indicate a future or an imperative, a potential or a habitual present or future. The particular systems of Motu, Yabem, and Graged have been discussed already (pp. 286-9 resp.). As remarked in a forthcoming book (Capell 1969), 'the general feature of the AN languages seems to be that the farther west along the north coast of New Guinea, the simpler the morphological structure of the verb — and this applies to the island groups such as the Admiralty Islands as well as to the mainland. In the far west, Windesi and Biak represent propositions reduced to the lowest terms of simplicity. It should be added, however, that some of the north coast languages are very poorly documented. This is true for practically all between Manam and Biak, except for Waropen ... however, where information is to hand, as in Tumleo, the verbal system seems to be extremely simple in structure.' iii. Particles In a general discussion such as this it is not possible to go into detail about the particles in NGAN languages. The most noticeable feature, which is not to be found in the languages of the Solomon Islands and eastwards, is the use of postpositions in the AN 2 languages, where the AN 2 languages, like those farther east and in western AN (Indonesia) have propositions. In Graged, for instance, one says, ab-lon i madoime 'he was in (Ion) the house'. Coupled with the SOV order, this feature helps to give a very un-AN appearance to these languages. It is a very common feature in the A N j group: Manam larj-lo 'in the sky', exactly like Graged did-lon. In southwest New Britain this -lo occurs again (p. 268). This usage is noticeable particularly where the NGAN languages are in close contact with NAN neighbours, and it appears also in Torau and Uruava. There is no information for Banoni. In general it seems to accompany SOV word order in the syntax of a language. The languages in which postpositions prevail (there may be an occasional preposition) in contrast to prepositions, may be tabulated as follows: Postpositions', in SEP, from Mekeo in the west to boundary on northeast coast; in the islands from Misima to Sudest; Eastern Bougainville; Graged, Manam, Suain on north. Prepositions: In SEP, the Kiriwina group northwards, Wampar, Sio-Mindiri; New Ireland and New Britain generally, but western Bougainville is not illustrated.

334

A. CAPELL

The languages of the Markham Valley area are not fully recorded, but Atsera has both prepositions and postpositions: ini da rain rosan 'he said to his brothers' (da = to); if an nam mana 'he went place outside', i.e. he went out of the place. The coastal languages from about Kela on the Morobe coast to Yabem at Finschhafen, use some postpositions, but more frequently verbalised expressions which are unmatched in other parts of the NGAN area. Many expressions with reference to places and even times are in verbal form. There are many directional verbs in these languages, e.g. Yab. -sep 'to go down', ey kesep ayga loPtepoe gemey 'he camedown from the hill', yapale sepa katapa ayga salty 'the boys have cut boards in the forest'. Localisation is left unexpressed: yapaleo geP me sauy (girl lies bed small), 'the girl is lying on a small bed' (see Dempwolff 1939:33-40). Among postpositions there is a locative -ya, seen in a compounded form with andu nem-ya (house frontat), 'in front of the house'; Laukanu here has tsa yaoy similarly. Still further to complicate the pattern, there is a preposition ayga 'locative, movement from'; gemey ayga tenane 'he came from his mother', Laukanu minan gey sanaygway ya, with postposition ya. The examples given in the various sections of Part 4 do nothing more than point the way to a comparative grammar of the N G A N languages. They show the different types of grammatical structures to be found in the area, limited as it is, and highlight the clear dichotomy of the ANX and AN 2 subgroups, and the basically NAN structure of the latter. The AN 2 languages would seem to be at root NAN languages with a veneer of AN lexicon and less grammar. Facts of this nature suggest that here at least if not also elsewhere in AN linguistic territory there has been pidginization. As far as New Guinea is concerned this does seem to be beyond dispute. As far as the areas farther east are concerned, the difficulty felt about this theory of origins is that there appears to be very little common matter in the substrata of the languages once the PAN element is accounted for, and it is hard to think that every little area in preAN times had a completely unrelated language. Close study of the coastal regions of New Guinea, however, makes it seem at least possible that this really was so. On the other hand, study of the NAN vocabulary of the island languages may show a greater agreement in NAN elements than has been thought. The exemplification by Capell of possible pre-AN 'regional languages' in SEP raises this possibility. Within the New Guinea area itself, the only full examination on the grammatical level is that given in Capell (1943: Part III). Detailed study of the New Ireland and New Britain areas along with the Buka languages has not yet been carried out, but information in the preceding sections from the present author's field notes and other sources set out the main features of these languages. Subdivisions within the groups are obvious even from the little that it has been possible to say here. The publication of fuller information needs to be undertaken as soon as possible. Until this is done, it is impossible to reach firm conclusions concerning the history of the region. If Grace's previously quoted statement (p. 294) together with Goodenough's and the general theory presented by Dyen have any validity at all, the N G A N area is probably

AUSTRONESIAN LANGUAGES OF AUSTRALIAN NEW GUINEA

335

the most important subdivision of Eastern Austronesian from the historical point of view. This further study, publication, and analysis is the next call on Oceanic linguists, but their studies need to be of whole languages, not just of lexicostatistical lists.

5.

Conclusions

These pages have aimed at giving a general view of NGAN both descriptively and analytically. They have tried to presume no initial acquaintance with the special area concerned. The matter contained in Part I, deliberately called 'Informational' because it seeks to provide an orientation in NGAN — its extent, what is known of it and what still has to be found out in terms of actual empirical information — is merely primary data; Oceanic linguists desire to know the history that lies behind the present-day linguistic facts. Part II, 'Analytical', therefore summarises the analyses carried out by various workers, and indicates outstanding problems in all sections of the linguistic study: phonology, vocabulary, and semantics, as well as grammatical structure. If it should seem that undue use has been made of an earlier work of the author (Capell 1943), the only apology that can be offered is that this was a first effort in its field, a sort of pilot undertaking, and therefore contains much basic information that cannot be repeated in full here. At the same time, further theoretical problems are arising as the scope of AN reconstruction widens. Dyen (1962, 1965) has felt a need to posit further PAN phonemes as a result of examining evidence from Formosa, and these may have more than one reaction on the N G A N field. In his 1965 paper he discussed in the light of the new phonemes proposed for PAN the matter of the homeland from which the languages emerged. He states (p. 304): 'As the density of the mergers with the same distribution increases, so increases the probability that the mergers are common innovations and reflect a common protolanguage ... These mergers thus tend to confirm the hypothesis of a Formosan homeland ... The Formosan homeland hypothesis will also facilitate the explanation of a phenomenon observed among the languages of New Guinea, Melanesia, Micronesia and Polynesia. This is the merging of *p and *b\ These two consonants are kept separate only west of and including Biak (Numfor); elsewhere they are unified. However, Dyen remains doubtful: 'there are a number of reasons for caution before regarding the Formosan homeland hypothesis as a necessary implication of the mergers involved in the correspondence discussed in this paper. Among these is the lexicostatistical evidence for a large number of highly divergent languages in the New Guinea-Melanesian area' (1965:305). The suggestion that such mergers may point to a Formosan homeland is completely at variance with earlier theory, and against Dyen's own earlier theory (1963), and it can rightly be described as a 'current trend' that finally may not be substantiated. In the presentation of the background to all these current discussions some repeti-

336

A. CAPELL

tion has been unavoidable, but in general cross-references have been given in all cases. It may seem, finally, to some linguists, that too much emphasis has been placed on factors that are rather sociological than linguistic, and are not amenable to computer or other exact assessment. The author is aware of this possible reaction. Linguistics claims autonomy as a science. This can readily be granted where purely linguistic processes are involved, as in the laws of sound change and many other matters. The weakness of the claim to absolute autonomy is that it may — and has tended to — lift linguistics out of its life situation into something as impersonal as mathematics itself. But mathematics is by nature impersonal, unaffected by who performs mathematical calculations or where or when they are performed. Language, on the other hand, lives only in a human and social setting, and many non-computable factors enter into it; indeed, some almost imponderable factors. Hence the writer has tried consistently to draw attention to these sociological aspects, to stress the fact that language occurs always in society, and to recall the further fact that movements of people for historical and social reasons may invalidate arguments perfectly cogent in themselves, if they are isolated from life and kept to the computer. The result has perhaps been a picture that not all linguists will entirely approve, yet it seems to the writer necessary to call attention to it. In the process, he has tried to make explicit many features of NGAN that require further study, in the hope that linguistic field workers will be attracted to the problems that remain to be solved.

BIBLIOGRAPHY

and C . H U R D . 1963a. Languages of the Bougainville District. Summer Institute of Linguistics, New Guinea. 56 pp. and map. . 1963b. Languages of the Cape Hoskins Patrol Post Division of the Talasea Sub-District, New Britain. Port Morseby, Papua. 21 pp. and map. . 1965. Halia language course. Summer Institute of Linguistics, Ukarumpa. ARNOLD, J.K. 1931. A grammar of the Edugaura language. Port Moresby. 60 pp. AUFENANGER, A . 1939. Wetterzauber auf den Yabob Inseln. Anthropos XXXIV. 277-91. BALDWIN, B . 1945. Usituma! A Song of Heaven. Oceania XVIII.201-38. . 1950. Kadaguwai: Songs of the Trobriand Sunset Isles. Oceania XXIV.26385. BLEY, B . 1912. Praktisches Handbuch zur Erlernung der Nordgazelle-Sprache (Neupommern). 238 pp. BRANDSTETTER, R . 1 9 1 6 . An introduction to Indonesian linguistics. Four essays, translated by C.O. Blagden. London. 351pp. BROWN, G. and B . DANKS. 1882. A dictionary of the Duke of York language, New Britain Group; also a grammar of the same and an introduction by the Rev. G. Brown. Mss. in Mitchell Library, Sydney, lxx + vi + 328 pp. ALLEN, J . ,

AUSTRONESIAN LANGUAGES OF AUSTRALIAN NEW GUINEA

337

A. 1943. The linguistic history of Southeastern Papua. Sydney, Mepical Publishing Company. . 1962a. The techniques of structure statistics. Oceania XXXIII/1.1-11. . 1962b. Oceanic linguistics today. CAnthr 3/4.371-428. . 1964. Verbal systems in Philippine languages. PhilJSci 93/2.231-49. . 1968. A lost tribe in New Ireland. Mankind 6/10.499-509. . 1969. A survey of New Guinea languages. Sydney, University Press. CARTER, G . G . 1952. Some grammatical notes on the Teop dialect. JPS 6 1 . 2 3 0 - 4 2 . CHINNERY, E . W . P . 1925. Notes on the natives of certain villages of the Mandated Territory of New Guinea. Anthropological Reports, Territory of New Guinea. No. 1. . 1926. Certain natives of South New Britain and Dampier Straits. Anthropological Reports, Territory of New Guinea. No. 3. CHOWNING, A . 1968. Languages of New Britain. PL, A 2 1 . CHRETIEN, C.D. 1956. Word distribution in Southeastern Papua. Lg 32/1.87-108. CHURCHILL, W. 1916. Sissano: movements and migrations within and through Melanesia. Washington, Carnegie Institute. 181 pp. and maps. CODRINGTON, R.H. 1885. The Melanesian languages. COSTANTINI, A. 1907. Theoretisch-praktischer Lehrgang von der Neupommerschen Sprache. Arch. Stadium. Dtsch. Kolonialsprachen 5 . 1 - 2 2 1 . DEMPWOLFF, O. 1905. Beiträge zur Kenntnis der Sprachen von Deutsch Neuguinea. MSOS 8.182-254. . 1911. Sagen und Märchen aus Bilibili. Bässler Archiv 1.63-102. . 1924-25. Der Verwandschaftsystem der Kate. ZES 10.65-73. . 1928-29. Ethnographische Schilderungen aus Graged (Neuguinea). ZES 18.205-36, 309-15. . 1934-38. Vergleichende Lautlehre des Austronesischen Wortschatzes. ZES Beihefte 15, 17, 19. . 1939. Grammatik der Jabem Sprache auf New Guinea. Abh. aus dem Gebiet der Auslandskunde, Hamburg. Bd. 50. DYEN, I. 1953. The Malayo-Polynesian laryngeals. Baltimore. . 1962. Some new proto-Malayopolynesian initial phonemes. JAOS 82.214-15. . 1963. The lexicostatistical classification of the Austronesian languages. New Haven. . 1965. Formosan evidence for some new Proto-Austronesian phonemes. Lingua 15.285-305. FELLOWS, S.B. 1901. Grammar of the Kiriwina dialect. Brisbane, Government Printer. 28 pp. (See also article by same title in Annual Report on British New Guinea, 1 9 0 0 - 0 1 , 1 7 1 - 7 . ) FISCHER, H . 1963. Watut: Notizen zur Kultur eines Melanesierstammes in Nordost Neuguinea. Brunswick. 289 pp. and maps and illustrations. . 1965. Studien über Seelenvorstellungen in Ozeanien. Brunswick. 432 pp.

CAPELL,

338

A. CAPELL

FRIEDERICI, G. 1912-13. Wissenschaftliche Ergebnisse einer Amtlichen Forschungsreise: I (1912) Beiträge zur Völker- und Sprachenkunde von Deutsch Neu Guinea. II (1913) Untersuchungen über eine melanesischen Wanderstrasse. Berlin. GABELENTZ, H. C. VON DER. 1861-71. Die Melanesischen Sprachen nach ihrem grammatischen Bau under ihrer Verwandschaft unter sich. 2. vols. GOODENOUGH, W. H. 1961a. The Willaumez languages of New Britain, (typescript). 39 pp. . 1961b. Migrations implied by relationships of New Britain dialects to central Pacific languages. JPS 70/1.112-26. GRACE, G.W. 1964. Movement of the Malayopolynesians 1500 B.C. to A.D. 500: The linguistic evidence. CAnthr 5/5.361-83. HAMBRUCH, P. 1907. Wuvulu und Aua. Mitt. Mus. Völkerkunde. Hamburg, Beih. Jb. hamburg wiss. Anst. 4/25.1-156. HARDING, T . G . 1967. Voyagers of the Vitiaz Strait. 282 pp. HEES, F. 1915-16. Ein Beitrag aus den Sagen und Erzählungen der Nakanai (NeuPommern). Anthropos 10-11.34-6, 562-85, 861-87. KLAFFL, J., a n d

F . VORMANN. 1905. D i e

Sprachen

des

Berlinhafenbezirkes

in

Deutsch-Neuguinea. MSOS 8.1-138. KLEINTITSCHEN, A. 1924. Mythen und Erzählungen eines Melanesierstammes aus Paparatava, Neu Pommern. Anthropos Ethnologische Bibliothek, Band II, Heft 4. KLIENEBERGER, H. R. 1957. Bibliography of Oceanic linguistics. London Oriental Bibliographies, Vol. I. London, Oxford University Press. 143 pp. KRÄMER, A. 1908. Wuvulu und Aua: Besprechung nebst einigen Beobachtungen von Dr. Augustin Krämer. Globus 43.254-7. LANYON-ORGILL, P. 1942. A Polynesian settlement in New Britain. JPS 51.87114.

LAYARD, J. 1942. Stone men of Malekula. London. LISTER-TURNER, R., and J. B. CLARK. 1930. Revised Motu grammar and vocabulary. Port Moresby. 192 pp. MAGER, JOHN. 1952. Gedaged-English dictionary. Columbus, Ohio. 353 pp. MALINOWSKI, B. 1917-20. Classificatory particles in the language of Kiriwina. B S O S 1/4.33-78.

. 1935. Coral gardens and their magic. 2 vols. London. MANNERING, G. (ed.). 1967. English-Kuanua Dictionary, (mimeographed). 120 pp. MAYR, E., a n d O . DEMPWOLFF.

1929.

Nisan-Deutsches Wörterverzeichnis.

ZES

21.252-6. MEIER, J. 1906. Berichtingen zu Dr. Schnee's Mitteilungen über die Sprache der Moanus. Anthropos 1.201-18, 472-82. . 1907-08. Mythen und Sagen der Admiralitäts-inseln. Anthropos 2.646-67, 933-41; 3.193-206, 651-71.

AUSTRONESIAN LANGUAGES OF AUSTRALIAN NEW GUINEA

MEYER, H. 1932. Wunekau, oder Sonnenverehrung in Neuguinea.

339 Anthropos

27.423-55, 819-54.

MEYER, O. 1910. Mythen und Erzählungen von der Insel Vuatom. Anthropos 5.711-33. . 1911. Die Schiffahrt bei den Bewohnern von Vuatom. Bässler Archiv 1.25768.

MEYER, P. 1909. Mythen und Sagen der Küstenbewohner der Gazellehalbinsel. Anthropos Inst. Bibliothek. 281 pp. MILKE, H. 1938. Die Benennungen der Geschwister in den AN Sprachen Ozeaniens. ZES 70.51-66. . 1958. Zur Inneren Gliederung und geschichtlichen Stellung der ozeanischAN Sprachen. ZES 83.58-62. . 1961. Beiträge zur ozeanischen Linguistik. ZES 86.162-82. . 1965. Comparative notes on the AN Languages. Lingua 14.330-48. MÜLLER, H. 1907. Grammatik der Mengen Sprache. Anthropos 2.80-99, 241-54. NEUHAUSS, K. 1962. Beiträge zur Ethnographie der Pala, Mittel Neu Irland, Ed. by C. Laufer and C. A. Schmitz. 432 pp. PARKINSON, R. 1907. Dreissig Jahre in der Südsee. Stuttgart. 876 pp. (Linguistic section, pp. 721-87.) PEEKEL, S. 1908a. Grammatik der neu-mecklenburgischen Sprache, speziell der Pala Sprache. Arch. Studium, deutsch. Kolonialsprachen 9.216 pp. . 1908b. Die Yerwandtschaftsformen des mittleren Neu-Mecklenburgs. Anthropos 3.456-81. ——. 1929-30. Grammatische Grundzüge und Wörterverzeichniss der Labelsprache im Süden von N e u - M e c k l e n b u r g . Z E S 20.10-34, 92-120.

RAUSCH, P.

1912.

Die Sprachen von Südost Bougainville. Anthropos 7.105-34,

585-616, 964-94.

RAY, S. H. 1895. Comparative vocabulary of the New Guinea dialects. London. 40 pp. . 1926. The Melanesian Island languages. Cambridge University Press. 598 pp. RICKARD, R. H. 1889. Dictionary of the New Britain dialect and English, and English and New Britain, also a grammar. Sydney, Mitchell Library, (mimeo.). SALZNER, R. 1960. Sprachenatlas des Indopazifischen Raumes. 2 vols. SCHMIDT, P.W. 1899. Die sprachlichen Verhältnisse Ozeaniens. Mitt. der Antrop. Ges. in Wien. Bd. XXIX, 245-58. . 1900-01. Die sprachlichen Verhältnisse von Deutsch Neuguinea. ZAOS 5.354-84, 6.1-97. SCHMITZ, C.A. 1960. Historische Probleme in Nord-Ost Neuguinea. Wiesbaden. 4 4 1 pp. and map. . 1961. Das Problem der Austromelaniden Kultur. Acta Tropica 18.97-141. SCHNEE, H. 1901. Beitrag zur Kenntnis der Sprachen im Bismarck-Archipel. MSOS 4/3.229-79.

340

A. CAPELL

L. 1911. Zur Kenntnis der Melanesischen Sprache von der Insel Tumleo. Jena. 96 pp. STRESEMANN, E . 1927. Lauterscheinungen in den amboinischen Sprachen. TREVITT, J . W . 1940. New Britain dictionary. 143 interleaved pp. WATERHOUSE, J . H . L . 1920(1939). New Britain language book. ZWINGE, HERMANN. (1953?). A grammar of the Gunantuna language. Bound typescript. SCHULTZE,

L A N G U A G E S OF THE NEW H E B R I D E S A N D SOLOMON I S L A N D S GEORGE W. GRACE

The area with which we are concerned in the present paper consists of the British Solomon Islands Protectorate (BSIP) and the New Hebrides Condominium. The BSIP includes all but two of the principal islands of the Solomons chain. Those two, Bougainville and Buka, are administered as part of the (Australian) Trust Territory of New Guinea. The political boundary passes close to the southeast tip of Bougainville, with all of the larger islands in the vicinity of that part of the Bougainville coast falling within the BSIP. The Protectorate includes a number of islands outside the Solomon chain. Most important are the Santa Cruz and Reef groups to the east. Also included are a number of islands populated by Polynesians. Notable among these are Rennell and Bellona, to the southwest of the Solomon chain, and — on the north and east of the chain — Ongtong Java (the northernmost populated group of the Protectorate), Sikaiana, Anuta (the easternmost inhabited island), and Tikopia (the southernmost inhabited island of the Protectorate). The New Hebrides Condominium, administered jointly by the United Kingdom and France, lies immediately to the south of the BSIP. To the south and west lies New Caledonia, a French overseas territory, and the terminus of the Melanesian chain of islands. The New Hebrides islands extend in a roughly north-south direction. The Condominium includes, in addition to what are often thought of as the New Hebrides proper, the Banks Islands to the north, and to the northwest of the Banks Islands, the group of small islands known as the Torres. Directly north of the Torres Islands are found the Santa Cruz Islands of the BSIP. Precise figures as to size and population are not available. The ninth edition of the Pacific Islands Year Book (Sydney 1963) gives the land area of the British Solomons variously as 12,000 and 11,500 square miles. That of the New Hebrides is given as 5,700 and again 5,000 square miles. The population of the British Solomons is estimated at around 130,000; that of the New Hebrides at about 60,000. In both cases the great majority of the population consists of indigenous islanders. Both territories present the characteristic Melanesian picture of great linguistic diversity within a restricted area. English, and, in the New Hebrides, French, are being taught, but to date only a small proportion of the population has been significantly affected. The principal lingua franca in the New Hebrides, and again in all but the western portion of the Solomons is Pidgin English. The Pidgin in each of these

342

GEORGE W . GRACE

territories is clearly related to the better-known Pidgin of New Guinea, for which the name 'Neo-Melanesian' has been proposed. However, in each of the three territories Pidgin has come to develop its own local characteristics — in particular there has been considerable borrowing from influential local languages — and the mutual intelligibility is reportedly not of a high order. Certain of the indigenous languages of the area are excluded from the scope of this paper. These consist of (1) the non-Austronesian languages of the Solomons, which are to be treated in the chapter on Papuan Languages, and (2) the Polynesian Outliers, which are to be treated in the chapter on Polynesian languages. The Papuan languages are: (1) Bilua, the language of Vella Lavella, (2) Baniata, spoken on the southern end of Rendova, (3) apparently Kazukuru, formerly spoken on New Georgia (see notes in Lanyon-Orgill 1953 on Guliguli and Dororo, spoken in the vicinity of Kazukuru, which may represent dialects of Kazukuru, or closely related languages), (4) Savosavo, of Savo Island, (5) the language of the Main Reef Islands (north of Santa Cruz) for which Davenport (1962) proposes the name 'Reef Island-Santa Cruz', (6) the languages of Santa Cruz Island. Although there are village-by-village dialect differences which make the determination of language boundaries somewhat arbitrary, Davenport proposes that the dialects be classified into three languages, which he labels, 'North-West-Santa Cruz', 'South-Central-Santa Cruz', and 'South-East-Santa Cruz', respectively. The Polynesian languages are: A. In the Solomons, (1) the aforementioned Rennellese, with Bellonese as a slightly divergent dialect, (2) Luangiua of Ongtong Java (apparently dialectally close to the languages of Nukumanu, Takuu, and Nukuria in the Territory of New Guinea), (3) Sikaiana, Stewart Islands, east of Malaita, (4) the language of the Outer Reef Islands and the Duff Islands (with two distinguishable dialects), (5) the language of Tikopia, and, apparently only dialectally different, the little known language of Anuta. B. In the New Hebrides, (1) Mae (Emwae) of the Three Hills Island south of Epi, (2) the language of the islands of Mele and Fila in Vila Harbor, Efate, (3) the language (with some dialect difference) of Aniwa and Futuna to the east of Tanna.

SOURCES OF GENERAL INFORMATION

A survey of the linguistic situation in the area is considerably facilitated by several useful bibliographic treatments in recent years. Especially relevant is the revised edition (1962a) of Arthur Capell's A linguistic survey of the south-western Pacific, which contains chapters on the Solomons and the New Hebrides. This work reflects the great store of information accumulated by Capell over his many years of interest in the languages of the Pacific. Much of this information has been obtained by first hand contact with the languages. It has been supplemented in communication with persons, particularly missionaries, in the field. In addition to extensive bibliographic

LANGUAGES OF THE NEW HEBRIDES AND SOLOMON ISLANDS

343

references and the best set of linguistic maps available for the area as a whole, this work contains an assortment of information on aspects of language structure, linguistic relations, and numbers of speakers and other matters relating to the current status of various of the languages. Unfortunately, the value of this book as a reference work is somewhat diminished by the presence of a number of errors and inconsistencies, and by what appears to be a rather haphazard organization (which, however, is due in considerable part to the spottiness of the available information). A second very useful work is H. R. Klieneberger's (1957) Bibliography of Oceanic linguistics. This work is generally very complete and accurate. Also to be mentioned are the Bibliographie des Nouvelles-Hébrides, by Patrick O'Reilly (1958), A Pacific bibliography, by C.R.H. Taylor (second edition 1965), and the bibliographies appearing periodically in the Journal de la Société des Océanistes and in Oceanic Linguistics. There is also much useful information, including grammatical sketches of a considerable number of languages, in Codrington 1885 and Ray 1926. Both of these will be referred to again. Hollyman 1960 is a useful guide through the maze of language names which have been or still are in use. LINGUISTIC RELATIONS

The Concept 'Melanesian' The (non-Polynesian) Austronesian languages of Melanesia are traditionally referred to as the 'Melanesian' languages. However, one is tempted to characterize Fijian and the languages of much of the Solomons and New Hebrides as the Melanesian languages par excellence. These languages were almost the only examples available at the time that the notion of Melanesian languages was achieving currency and substance, and they still often appear to be regarded as the classical examples. However, they represent only a fraction of the languages that are grouped under the rubric. To define in extension, the term 'Melanesian' refers to Fijian (all dialects), Rotuman, and (except for Polynesian Outliers) all Austronesian languages in the Melanesian chain, including New Guinea. Some Austronesian languages in the westernmost part of New Guinea are excluded by some authors, and some of these will probably be more generally excluded once they are better known. Many authors, particularly the Germans, include as well all of the languages of Micronesia except the Polynesian Outliers and Palauan and Chamorro (the latter two being classified as 'Indonesian', or more specifically, 'Philippine'). There are probably no specialists who would hold that Melanesian so defined is a valid genetic grouping. However, there is one school of thought (to which the present author subscribes) which holds that there is probably a valid genetic subgroup (usually called 'Oceanic') of Austronesian which includes the Melanesian languages, but also includes Polynesian as a relatively recent offshoot.

344

GEORGE W . GRACE

A variety of factors seem to have contributed to the tradition of grouping these particular languages together. First, the area which they occupy (or more precisely, share with Papuan languages and a few Polynesian Outliers) is geographically delimitable. Second, it was the last of the major areas occupied by Austronesian languages to become known to the outside world. Although specimens of vocabulary from the New Hebrides and Solomons had been collected as early as the voyages of Mendana and Quiros in the late sixteenth and early seventeenth centuries, the languages of Melanesia remained virtually unknown until the latter half of the nineteenth century. By that time the existence of the family called Malayo-Polynesian (later, Austronesian), embracing the Indonesian (or 'Malayan') languages to the west of Melanesia and the Polynesian languages to the east, had already been recognized. The Melanesian languages, when discovered to be Malayo-Polynesian but to be neither Malayan nor Polynesian, therefore naturally assumed the status of a tertium quid. A third factor was the assumption that the peoples of Melanesia were racially distinct from the Polynesians and Indonesians. This assumption was firmly established before any significant information regarding their languages was available. It is interesting to observe that the linguistic relations between Polynesia and Indonesia were noted very early, and were adduced as evidence that the Indonesians and Polynesians belonged to the same great race. The physical differences between Indonesians and Polynesians thus tended to be played down. Perhaps, as the other face of the same coin, the differences between the supposed two branches of the great race, on the one hand, and the intervening 'Oceanic Blacks', on the other, tended to be further accentuated. Thus, at the middle of the nineteenth century the greatest gap in the linguistic map of Oceania had become identified with 'that other race'. It was into this picture that the first substantial reports on Melanesian languages had to be fitted. It should be noted, by the way, that this popular dichotomy of Pacific races, not, by far, always free of evaluative connotations, has persisted to the present day. It is frequently adverted to in works on linguistic classification and migrations. A fourth factor supporting the Melanesian concept is the fact that it has seemed possible to characterize a Melanesian linguistic type which contrasts with similar characterizations of Polynesian and Indonesian types. However, it is uncertain just how representative such a characterization of the Melanesian type is. The Melanesian languages number in the hundreds. The vast majority are either totally undescribed or very poorly described. Moreover, it has become customary to regard some Melanesian languages as more typical than others. Descriptions of the Melanesian type therefore deliberately give more weight to some languages than to others, are drawn up in the expectation that some languages — in particular, the less typical ones — will probably not conform with respect to any given particular in the description, and must perforce omit many languages from consideration for lack of evidence. The single language which is drawn on most frequently in exemplifications of Melanesian characteristics is undoubtedly Fijian. Fijian was the first Melanesian language to become known; the first grammar and dictionary were published in 1850

|5S°

IB««

187®

IB«»

BRITISH

IS**

SOLOM

!»••

OMON

i«o«

ISLANDS

fei»

!•»•

!•»•

LANGUAGES OF THE NEW HEBRIDES AND SOLOMON ISLANDS

345

(Hazelwood 1950a, 1950b). Subsequent works have maintained its status as by far the best described of the Melanesian languages. It is the most important of the Melanesian languages in terms of use; more is published in it, and more people speak it than is the case of any other Melanesian language. Finally, it is well-suited for exemplification because it seems to have a relatively large number of widely-shared features as compared with a small number of idiosyncratic ones. If there is frequent recourse to many Solomons and New Hebrides languages for examples of Melanesian characteristics, the reasons are much the same. However, none of these languages is nearly so widely known and used as Fijian.

The Position of the Melanesian Languages With respect to the relationship of the Melanesian languages, as defined, to the rest of the Austronesian family, it is possible to distinguish three schools of thought. One school (cf. for example, Ray 1926, Capell 1943, 1962a, 1962b) regards the Melanesian languages as a rather special kind of Austronesian, less representative of the Protolanguage than are, for example, the Indonesian languages. This school assumes that there is a Melanesian race, represented by the speakers both of Papuan and Melanesian languages, but that none of the Melanesians originally spoke Austronesian languages. The present Melanesian languages are believed to represent pidgin languages resulting from contact between Melanesians and colonies of Indonesians. The Melanesian languages, themselves, appear to represent the only surviving traces of the postulated colonies. Each pidgin language is conceived of as involving two components: an Austronesian component due to the language of the Indonesians, and an indigenous component due to a particular Papuan language (different in each instance) which has since become extinct. According to this view some Melanesian languages are more Austronesian, i.e. have a more substantial Austronesian component, than others. Moreover, adherents of this school believe that other Melanesian languages with further reduced Austronesian components arose through secondary pidginization — the pidginization of an already established pidgin. Linguists of this persuasion are, therefore, prepared to find that some Melanesian languages have a much more prominent Austronesian element than do others. Moreover, since it is only the Austronesian element, presumably, that is comparable from one language to the other, those languages with a larger Austronesian element present a greater number of features that are apt for comparison. They, therefore, are the best examples of the Melanesian type. Although many linguists (including the present author) do not subscribe to these views, it is a striking fact that some Melanesian languages do appear to be more idiosyncratic than others. That is, substantially more cognates with other Austronesian languages (both languages in Melanesia and those elsewhere) can be found in some Melanesian languages than in others. Likewise, there are some Melanesian languages in which the sound correspondences with other Austronesian languages are easier to

346

GEORGE W. GRACE

formulate and where exceptions that resist explanation are fewer. There are, finally, languages which more closely approximate what appears to be the typical Melanesian grammatical pattern than do others. The fundamental assumptions of the Ray-Capell school would lead one to expect these variables to be linked. That is, we would expect that if a language is an exemplary Melanesian language from one point of view, it would be so from others as well, and if it is idiosyncratic with regard to one variable, it would be so for all. With regard to grammatical structure it is hard to say how far this expectation is realized. However, on the basis of numbers of cognates alone, one almost inevitably comes to think of Melanesian languages as occupying various points on a scale ranging from very idiosyncratic at one extreme to (let us choose a neutral term) very 'typical' at the other. It appears to hold true quite generally that a very idiosyncratic language (say, Zabana of Ysabel) will show fewer cognates with any other Austronesian language, except possibly one or two close neighbors, than will a 'typical' language such as Fijian. Moreover, we will find far more cognates between Fijian and a remote Austronesian language such as Malay than between Fijian and an idiosyncratic Melanesian language. It is this state of affairs that has inspired a second view which has recently emerged regarding the position of the Melanesian languages. The advent of lexicostatistics has formalized and given prominence to the familiar assumption that cognate density tends to be inversely proportionate to divergence time. If this assumption is valid, it would follow as suggested by Ward Goodenough (1962:408) that, 'The role of Melanesia in the history of the Austronesian languages appears to be central and ancient, not peripheral and recent'. Specifically, Melanesia would emerge as a likely candidate for the Austronesian homeland. This argument is made more concrete by Isidore Dyen's (1965) lexicostatistical study. In this study lists from more than 200 Austronesian languages were compared. In the resulting classification Austronesian consists of 40 branches. One of these branches, which is named 'Malayopolynesian' (note the departure from traditional usage, in which 'Malayo-Polynesian' and 'Austronesian' are synonymous) contains more than half of the languages in the study. In Malayopolynesian are included all of the Polynesian languages, almost all of the Indonesian languages, and eight languages which are usually counted as Melanesian. The eight Melanesian languages are Fijian, Rotuman, two New Hebrides languages (Mota and Efate), three Solomons languages (Kerebuto of Guadalcanal and Lau and To'abaita of Malaita), and Motu of the south coast of Papua. These eight Melanesian languages, together with Polynesian, constitute the Heonesian Linkage, one of the seven branches of Malayopolynesian. Of the 85 languages comprising the remaining 39 branches of Austronesian, at least 76 would normally be counted as Melanesian (we include 7 languages in Micronesia in this count). We may note here that there are four New Hebrides and eleven Solomon Islands languages included in this figure. All eleven of the Solomons languages are from the western Solomons, while no western Solomons language is included in Malayopolynesian.

LANGUAGES OF THE NEW HEBRIDES AND SOLOMON ISLANDS

347

These results tend to confirm the suggestion made above that the New Hebrides and southeastern Solomons contain more of the 'typical' languages than do most parts of Melanesia. They also present a very clear argument that the Melanesian languages do not all belong to a single subgroup of Austronesian. The third view of the position of the Melanesian languages has been alluded to above. In this view the Polynesian languages and the Melanesian languages (including the languages of Micronesia with the exception of Palauan, Chamorro, and the Polynesian Outliers) constitute a single subgroup of Austronesian. This subgroup is usually called 'Oceanic'. This conclusion regarding the linguistic relations has been expressed by Wilhelm Schmidt (e.g. 1899a, 1899b, 1926), Otto Dempwolff (e.g. 1931:162; 1937:165, 193), George W. Grace (e.g. 1964:365-6), Wilhelm Milke (e.g. 1958, 1961, 1962), André Haudricourt (e.g. 1965). However, these writers do not generally conceive of Melanesian and Polynesian as coordinate branches of Oceanic. Most assume, on the contrary, that Polynesian is a relatively recent offshoot, and that it should be possible to identify a subgroup of Oceanic which would include the Polynesian languages along with just some of the Melanesian languages. A number of considerations seem to have contributed to the development of this view. First, the relationships among the Polynesian languages are very close ; it has been suggested at various times that they are no more than dialects of the same language. This fact indicates that their differentiation is of relatively recent date. Second, it was felt that the Melanesian languages showed a structural homogeneity as opposed to Indonesian and Polynesian. This had been argued by Robert H. Codrington (1885), for example. Third, in spite of this, Hendrik Kern's important comparative treatment of Fijian (1886, expanded version in Kern 1913-28) had represented the relationship of this Melanesian language to Polynesian as being very much closer than the relationship of either to Indonesian. Finally, and this has become the most important argument in recent years, the studies of comparative phonology made by Otto Dempwolff in the 1920's and 1930's point to a series of phonological innovations shared by all of the Melanesian and Polynesian languages and by no other Austronesian language.

Classification of the New Hebrides and Solomons Languages Most references to the subgrouping of the Melanesian languages of the Solomons and New Hebrides are of one of two types. There are statements regarding limited local relationships scattered through the literature, particularly the ethnographic literature. Although these are unquestionably of interest, they are of very uneven reliability corresponding to the great variation in linguistic competence and perspective of the authors. Much of the information made available in such sources is summarized in Capell 1962a and 1962b. Otherwise, since the Melanesian languages of the New Hebrides and the Solomons do not constitute a single genetic grouping, we find their interrelationships treated only

348

GEORGE W . GRACE

incidentally or within a larger framework. Schmidt (1899a, 1899b, 1926, 1940-41) maintained that the languages most closely related to Polynesian were Bugotu of Ysabel, Nggela of Florida Island, and Vaturanga of Guadalcanal. These languages represent the westernmost limit of what is here referred to as the southeastern Solomons. After those three, the languages most closely related to Polynesian are held to be (with some variations from one of his works to another) the remaining languages of the southeastern Solomons, Fijian, Rotuman, the languages of the central (sometimes northern) New Hebrides, and the Melanesian languages of Papua. Grace argues (1959) that the languages most closely related to Polynesian are Fijian and Rotuman, with Tongoa, Sesake, Makura, Nguna, and the languages of Epi and Efate, all in the central New Hebrides, only slightly more distant (1961:364). Milke (1958) classifies Oceanic into three subgroups on the basis of the treatment of Proto-Oceanic /, d, and y. One group consists of the southeastern Solomons alone. A second includes the western Solomons and the Banks and Torres Islands and, except for the Admiralties and Western Islands, all of the Melanesian languages of the Bismarck Archipelago and New Guinea. The third group contains the Polynesian languages, the Micronesian languages, Fijian, Rotuman, the New Hebrides (except for the Banks and Torres Is.), New Caledonia and the Loyalties, and the Admiralties and Western Islands. Bruce Biggs (1965) takes as a working hypothesis the existence of a subgroup which he calls 'Eastern-Oceanic'. This group is tentatively assumed to include at least Polynesian, Fijian, Rotuman, a number of languages of the southeastern Solomons (from the languages named it seems probable that all languages of the southeastern Solomons, except probably Bugotu, would be included if sufficient data were available), and Mota and Efate of the New Hebrides. Although these classifications may appear disturbingly diverse, this is due in part to the sparcity of data that underlies them and the different samplings of data available to different observers. One thread of consistency, however, deserves to be pointed out. There are certain languages which tend to be singled out, on the one hand by scholars such as Sidney Ray (1926:37, 595) as languages with a relatively large Austronesian component, and on the other by Schmidt, Grace, Milke, and Biggs as well as by Dyen's lexicostatistical classification as being among those Melanesian languages which are most closely related to Polynesian. These are: (1) Fijian, (2) Rotuman, (3) Efate and often neighboring languages of the central New Hebrides, (4) Mota and sometimes neighboring languages of the Banks Islands, (5) Nggela and often neighboring languages, especially Bugotu and languages of Guadalcanal, (6) only slightly less frequently all other languages of the southeastern Solomons, (7) often Motu and sometimes neighboring languages of southeastern Papua, (8) sometimes other languages of the northern New Hebrides. It is noteworthy that the languages of the southern New Hebrides (Eromanga, Tanna, and Aneityum) and of the western Solomons are almost invariably excluded as are the majority of Melanesian languages outside the Solomons-New Hebrides

LANGUAGES OF THE NEW HEBRIDES AND SOLOMON ISLANDS

349

area. The contrast that has been drawn here is, therefore, the contrast between what we have referred to as the typical and the idiosyncratic languages, whatever its historical significance may be. Sound Correspondences There are a considerable number of works in which some mention is made of the reflexes in one or more languages of the Solomons or New Hebrides of ProtoAustronesian phonemes. More detailed treatments are fewer. However, some should be mentioned here. The earliest such treatment is Hendrik Kern's (1906) study of Aneityumese, containing also a brief examination of a language of Eromanga. This study antedates the more systematic reconstructions of Proto-Austronesian phonology and vocabulary by Renward Brandstetter and Otto Dempwolff, but does rest on a considerable accumulated knowledge of comparative Indonesian due to T. N. van der Tuuk, Kern himself, and others. Ray (1926) applied Brandstetter's principles and results to an array of Melanesian languages, mostly in the Solomons and New Hebrides. The book contains a number of grammatical sketches, but in addition for many of the languages attempts to determine the reflexes of Brandstetter's proto-Indonesian phonemes. There are also chapters dealing with Melanesian grammar in general as viewed from the vantage point of Proto-Indonesian. In a series of studies, Otto Dempwolff investigated the reflexes in various languages of the Proto-Austronesian phonemes. At the outset, he used Brandstetter's ProtoIndonesian reconstruction, but subsequently introduced certain modifications. Solomons and New Hebrides languages which receive significant treatment in these studies are: in Dempwolff 1920, M o t a ; in Dempwolff 1924-25, Mota, Sa'a, and Aneityumese; in Dempwolff 1937, Sa'a. Dempwolff 1927 summarizes the situation with regard to the Melanesian languages in general. Although these works do not involve Proto-Austronesian, Biggs 1965 and Fox 1947 may be mentioned here. Biggs, in the study previously referred to, employs data from several Solomons and New Hebrides languages in reconstructions attributed to ProtoEastern-Oceanic. Fox discusses sound correspondences among a number of New Hebrides and Solomons languages and provides a list of putative cognates.

LINGUISTIC DESCRIPTION

Linguistic Structures The principal point to be made here is that in the present state of our knowledge there is no satisfactory way to give a picture of the structural characteristics of the Solomons and New Hebrides languages. Most statements that could presently be made would,

350

GEORGE W . GRACE

particularly if some organizing principle more demanding than that of the 'old curiosity shop' were adhered to, be either so narrow as to apply only to a very few languages or so broad as to apply to very many languages outside the area. Of course, a structural description of even a single language could be illuminating if the language were chosen on a suitable basis. However, no satisfactory choice is available. Let us point out that the number of languages in the area is impressively large. At present it is impossible to give an exact figure, but the order of magnitude is indicated by the language lists at the ends of the respective chapters in Capell 1962a. A total of 190 languages are listed; 91 in the British Solomons and 99 in the New Hebrides. None of them could be singled out as particularly important languages. Although several are known by considerable numbers of people outside their home territories, there is still no language with more than a few thousand speakers. Moreover, the languages of the area do not constitute a natural linguistic grouping. Since they do not, it is impossible to select a single language within the area as being more representative of the linguistic structures of them all than is some language outside the area. Nor is it possible to identify any sufficiently small number of natural groupings that a few languages might be selected as a representative sample. Even if such a sample could be identified, however, we would be confronted with the problem of inadequate data, for by present-day world standards all existing descriptions of individual languages of the Solomons or New Hebrides must be judged to be very imprecise and incomplete. That the descriptions should not conform to modern descriptive standards is not surprising. There have been very few instances where any significant amount of work with informants from this area has been carried out by persons with modern linguistic training. Most of these instances are very recent, and have not yet led to significant published results. Moreover, those individuals such as Ray, Capell, and Ivens who have acquired some measure of expertise extending over much or all of the area, have been heavily dependent on indirectly acquired information. A long-established means of learning about these languages has been the analysis of texts, the overwhelming bulk of which have been Bible translations. Although the information thus extracted has often been supplemented through correspondence with the missionaries in the field, the understanding of the structure of the languages which can be acquired by these means is obviously limited by the missionary's own understanding. Most of the existing descriptive literature either has been written by missionaries themselves or has been based on missionaries' knowledge of the languages. Since the missionaries have in general had no linguistic training, it is not surprising that phonological statements are generally vague, or, for that matter, that many of the Bible translations are reputedly very poor. The kind of information that figures most heavily in the existing descriptive literature relates to aspects of morphology (which is mainly agglutinative) and phrase structure. In particular, we often find fairly detailed inventories of affixes and particles. There have been a number of generalizing treatments of Melanesian grammar or

LANGUAGES OF THE NEW HEBRIDES AND SOLOMON ISLANDS

351

aspects thereof. Most of the more detailed studies of this sort emphasize the languages of the Solomons and New Hebrides. Some perspective both on the structure of the languages and on the kind of descriptive approach which characterizes the existing literature can be obtained from them. Also useful for the same purpose are the collections of grammatical sketches of individual languages in Codrington 1885 and Ray 1926. The most important of the general treatments is undoubtedly Chapter III, "Short comparative grammar of the Melanesian languages", in Codrington 1885 (100-92). Other chapters in the same work are also relevant. Likewise to be mentioned are articles by Walter G. Ivens. His "Melanesian modes of speech" (1939-42) and "Melanesian demonstratives" (1937—39b) are relevant, as is perhaps also his "Certain suffixes in Oceanic languages" (1914). We may perhaps also include Fox 1948a, "Passives in Oceanic languages", and 1948b, "Prefixes and their functions in Oceanic languages", and the more wide-ranging paper by Capell (1949), "The concept of ownership in the languages of Australia and the Pacific".

Descriptive Efforts So Far The preceding remarks might appear to imply that most of the work so far done in the New Hebrides and Solomons has been misguided and is linguistically irrelevant. If one were satisfied with the simple prescription (in the words of Rulon Wells 1963:38), '... that the first task of linguistics is to well describe languages, one by one', it might seem so. Still, if we consider the virtual terra incognita that confronted the missionaries at the outset of their Melanesian activities, it is hard to feel that 'to well describe languages, one by one' is a proper characterization of the linguistic needs of that time. Surely, the best allocation of effort would have included some kind of inventory taking, if only to make possible a linguistically motivated choice of successively smaller samples of languages for successively more thorough description. For certainly the thorough description of more than a small sample exceeds any foreseeable means. If we conceive of the initial task as one of giving some kind of accounting of the linguistic whole, it is possible to conclude that the efforts were expended in a fairly rational way, particularly in the beginning, and that substantial progress has been made. It was in the middle of the nineteenth century that missionary activity got underway in eastern Melanesia. In 1861 Hans Conon von der Gabelentz was able to publish the first volume of his Die melanesischen Sprachen (1861-73). This work provided grammatical sketches and vocabularies of a number of languages, mostly in the New Hebrides and Solomons. They were based on materials made available by the missionaries. Bishop John Coleridge Patteson of the Melanesian Mission showed a particular interest in the languages of the area, and collected quantities of data. It appears that he prepared grammars and phrase books of some of the languages, and

352

GEORGE W. GRACE

that several of these were published in 1866. However, they have never been widely available. Bishop Robert Henry Codrington, Patteson's successor, continued the latter's linguistic interests. In 1877 he published a grammatical sketch (Codrington 1877) of the Mota language of the Banks Islands, the lingua franca of the Melanesian Mission. In 1882, a dictionary and grammatical sketch (Inglis 1882) of the language of Aneityum, the southernmost of the New Hebrides islands, appeared. Shortly afterward, Codrington's (1885) important book, The Melanesian languages, was published. It was here that he made his case that Melanesian was a homogeneous entity distinct from Polynesian and Indonesian. As has been mentioned, he also presented generalized sketches of Melanesian grammar and Melanesian phonology. There are also comparative vocabulary lists with a lengthy discussion and 34 grammatical sketches. Of the grammatical sketches, that of Mota is by far the most extensive, being over 50 pages in length. That of Nggela (Florida), although shorter, is also highly regarded. The following years saw a number of additional publications of some significance. Donald MacDonald published in 1889 a volume entitled Three New Hebrides languages, which became, ex post facto, the first volume of a two volume work entitled, South Sea languages (MacDonald 1889-91). This work contains sketches and vocabularies of a number of languages in the New Hebrides. Sidney Ray's report on the languages of the New Hebrides (Ray 1893) followed. In 1894 MacDonald published a dictionary of the various languages of the island of Efate (central New Hebrides). However, its usefulness is seriously impaired by his failure to identify the specific languages from which the various words were taken. The book is devoted to and shaped by the thesis that the Oceanic languages derive from Semitic. A revised edition, entitled in this instance, The Oceanic languages (MacDonald 1907), is more frequently referred to. The last major publication before the turn of the century was a dictionary of Mota (Codrington and Palmer 1896). There were, in addition to the aforementioned works, a number of vocabularies and separately published grammatical sketches during this period. In terms of the linguistic complexity of the area and the state of descriptive linguistics at the time, the accumulation of information on the languages of the New Hebrides and the southeastern Solomons during the last few decades of the nineteenth century is rather impressive. Linguistic mapping and sampling had progressed to the point that most New Hebrides languages and a fair proportion of southeastern Solomons languages were represented in print by at least a short wordlist. Grammatical notes on a number of them had been published, and in a few privileged instances longer grammatical treatments or dictionaries had appeared. Treatments of phonology and grammar were, of course, not up to mid twentieth century standards, but the twentieth century lay before them. As has been previously indicated, the nineteenth century literature on Melanesian languages concerned primarily the New Hebrides and southeastern Solomons, although the Loyalties and New Caledonia were somewhat involved, especially at the

LANGUAGES OF THE NEW HEBRIDES AND SOLOMON ISLANDS

353

beginning, and Fijian was always a point of reference. The western Solomons and the Santa Cruz islands as well as the entire New Guinea area were almost totally excluded. Where there were descriptions of comparable scope of Melanesian languages outside the area (e.g. van Hasselt 1868 and subsequent works, Lawes 1885, Schellong 1890), they seem isolated and separate from the mainstream of Melanesian linguistics. From about the beginning of the twentieth century languages in other parts of Melanesia, particularly New Guinea, began to receive more attention. However, the published literature on the languages of the Solomons and New Hebrides which has appeared during the twentieth century does not present a very sharp contrast with that of the preceding period. There are no works that substantially surpass the quality of Codrington's Mota grammar (1885:253-310) or Mota dictionary (Codrington and Palmer 1896). However, there are several works of comparable quality. There has been somewhat less emphasis on filling in the map of the area. As a result the publication of very brief notes has become less frequent, and except for Ray 1926 (which contains 28 grammatical sketches — 26 of languages in the New Hebrides or Solomons) no important collections of descriptive sketches have appeared. Perhaps the best grammatical descriptions available besides that for Mota are Ivens 1911 (Sa'a of Small Malaita), Ivens 1931-14 (Ulawa of Ulawa, or Contrariété, Island), and Deck 1933-34 (Kwara'ae of Malaita). In each of these cases the author had extended experience with the language. Ivens subsequently published grammatical sketches of a number of languages. The first few dealt with languages that he had been able to study somewhat at first hand. These are Oroha of Small Malaita (Ivens 1926-28), Lau of various coastal areas of Malaita (last version, Ivens 1928-30a), and Marau Sound of Guadalcanal (1928-30b). The later ones were based primarily on the analysis of texts, and these are reputedly less accurate. The languages treated are Kwara'ae (1930-32a), Bugotu of Ysabel (1933-35a), Vaturanga of Guadalcanal (1933-53b), Longgu of Guadalcanal (193335c), Nggela of Florida Island (1935-37), all in the Solomons, and Lamalanga of Raga (Pentecost) Island (1937-39c), Lobaha of Omba (Lepers') Island (1940^2a), and Lotora of Maewo (Aurora) Island (1940-42b) all in the northern New Hebrides. Other grammatical sketches which may be mentioned are that by Bouillon (191516) of Sugu, and that by Capell (1930) of Inakona. Both languages are spoken on Guadalcanal. C.E. Fox (1950) published some notes on Nggela. He recommends that these in conjunction with Codrington's (1885) sketch should be used in preference to Ivens 1935-37. The best dictionaries are probably the Codrington and Palmer (1896) Mota dictionary and those of Sa'a and Ulawa (Ivens 1929) and Nggela (Fox 1955). Of somewhat lesser scope are dictionaries of the Roviana language of New Georgia (Waterhouse 1949, original edition 1923), Lau (Ivens 1934), and Bugotu (Ivens 1940). Ivens also published moderately lengthy vocabularies of Marau Sound (1930-32b) and Longgu (1937-39a). The most urgent need, obviously, is for more thorough and sophisticated gram-

354

GEORGE W. GRACE

matical descriptions. One of the major obstacles to this work has been the virtual absence of speakers of these languages from places where they would be easily accessible to trained linguists. However, there are some hopeful signs. Informants for some languages can be found in Auckland, where an active group of linguists exists (cf. Kasarhérou 1962, Walsh 1962 and 1966, Whaley 1962, and Hewitt 1966). Moreover, some fieldwork has been done recently on Nguna (Albert J. Schütz 1968, 1969a and b), Southeast Ambrym (Gary Parker 1968a and b, 1969), and Zabana (Glenn Seglern). REFERENCES BIGGS, BRUCE.

1965. Direct and indirect inheritance in Rotuman. Lingua 14.383-

415. P. 1915-16. Étude sur le dialecte de Sugu (Guadalcanal, Solomon Is.). Anthropos 1 0 - 1 1 . 7 5 8 - 8 0 . CAPELL, A. 1930. The language of Inakona, Guadalcanar, Solomon Islands. JPS BOUILLON,

39.113-36.

. 1943. The linguistic position of south-eastern Papua. Sydney, Australasian Medical Publishing Company Limited. -——. 1949. The concept of ownership in the languages of Australia and the Pacific. SJA 5.169-89. . 1962a. A linguistic survey of the south-western Pacific. New and revised edition. (South Pacific Commission Technical Paper 136.) Nouméa, South Pacific Commission. . 1962b. Oceanic linguistics today. CAnthr 3.371-428. . 1968. Lexicostatistical study of the languages of Choiseul, British Solomon Islands. PL, A 15.1-25. CODRINGTON, ROBERT H. 1877. A sketch of Mota grammar. London, Gilbert and Rivington. . 1885. The Melanesian languages. Oxford, Clarendon Press. , and JOHN PALMER. 1896. A dictionary of the language of Mota, Sugarloaf Island, Banks' Islands. London, Society for Promoting Christian Knowledge. DAVENPORT, WILLIAM. 1962. Comment on Capell 1962b. CAnthr 3.400-2. DECK, NORMAN C. 1933-34. A grammar of the language spoken by the Kwara'ae people of Mala, British Solomon Islands. JPS 42.33-48,133^14,241-56; 43.1-16, 85-100, 163-70, 246-57. DEMPWOLFF, OTTO. 1920. Die Lautentsprechungen der indonesischen Lippenlaute in einigen anderen austronesischen Südseesprachen. (Beihefte zur ZES 2.) Berlin, Dietrich Reimer. . 1924-25. Die /-, r- und ¿/-Laute in austronesischen Sprachen. ZES 15.19-50, 116-38, 223-38, 273-319. . 1927. Das austronesische Sprachgut in den melanesischen Sprachen. Folia ethnoglossica 3.32-43.

LANGUAGES OF THE NEW HEBRIDES AND SOLOMON ISLANDS

355

. 1931. Einige Probleme der vergleichenden Erforschung der Südsee-Sprachen. A n t h r o p o s 26.157-70.

. 1937. Deduktive Anwendung des Urindonesischen auf austronesische Einzelsprachen. (Vol. 2 of Vergleichende Lautlehre des austronesischen Wortschatzes.) (Beihefte zur ZES, 17). Berlin, Dietrich Reimer. DYEN, ISIDORE. 1965. A lexicostatistical classification of the Austronesian languages. (IUPAL 19.) Baltimore, Waverly Press. Fox, CHARLES ELLIOT. 1947. Phonetic laws in Melanesian languages. JPS 56.58118, 322-3. . 1948a. Passives in Oceanic languages. JPS 57.2-29. . 1948b. Prefixes and their functions in Oceanic languages (ma, ngä). JPS 57.227-55. . 1950. Some notes on Nggela grammar. JPS 59.135-69. . 1955. A dictionary of the Nggela language. Auckland, The Unity Press. GABELENTZ, HANS CONON VON DER. 1861-73. Die

melanesischen

Sprachen

nach

ihrem grammatischen Bau und ihrer Verwandtschaft unter sich und mit den malaiisch-polynesischen Sprachen. (Abhandlungen der philologisch-historischen Classe der Königlich Sächsischen Gesellschaft der Wissenschaften, 3, 7.) GOODENOUGH, WARD H. 1962. Comment on Capell 1962b. CAnthr 3.406-8. GRACE, GEORGE W. 1959. The position of the Polynesian languages within the Austronesian language family. (IUPAL 16.) Baltimore, Waverly Press. . 1961. Austronesian linguistics and culture history. AmA 63.359-68. . 1964. The linguistic evidence. (Movement of the Malayo-Polynesians: 1500 B . C . t o A . D . 500.) C A n t h r 5 . 3 6 1 - 8 , 4 0 3 - 4 .

HASSELT, J. L. VAN. 1868. Allereerste beginselen der Papoesch-Mefoorsche taal. Utrecht, Kemink en Zoon. HAUDRICOURT, ANDR£ G. 1965. Problems of Austronesian comparative philology. Lingua 14.315-29. HAZLEWOOD, DAVID. 1850a. A compendious grammar of the Feejeean language; with examples of native idioms. Vewa, Wesleyan Mission Press. . 1850b. A Feejeean and English dictionary: with examples of common and peculiar modes of expression, and uses of words. Vewa, Wesleyan Mission Press. HEWITT, HELEN-JO JAKUSZ. 1966. Aneityum of the southern New Hebrides: Anejom

segmental phonology and word list — a preliminary report. Te Reo 9.1-43. HOLLYMAN, K.J. 1960. A checklist of Oceanic languages. (Te Reo Monographs.) Auckland, Linguistic Society of New Zealand. INGLIS, JOHN. 1882. A dictionary of the Aneityumese language. London, Williams and Norgate. IVENS, WALTER GEORGE. 1911. G r a m m a r of the language of Sa'a, Malaita, Solomon

Islands. Anthropos 6.755-73, 926-40. . 1913-14. Grammar of the language ofUlawa, Solomon Islands. JPS 22.28-35, 96-103, 134-40, 219-24; 23.21-7.

356

G EORGE W. GRACE

. 1914. Grammar of the Lau dialect of the island of Mala. Norfolk Island, Melanesian Mission Press. . 1926-28. A study of the Oroha language, Mala, Solomon Islands. BSOAS 4.587-610. . 1928-30a. A grammar of the Lau language, North East coast of Big Mala, Solomon Islands. BSOAS 5.323-43. . 1928-30b. A study of the language of Marau Sound, Guadalcanal, Solomon Islands. BSOAS 5.345-58. . 1929. A dictionary of the language of Sa'a (Mala) and Ulawa, South-East Solomon Islands. London, Oxford University Press. . 1930-32a. A grammar of the language of Kwara'ae, North Mala, Solomon Islands. BSOAS 6.679-700. . 1930-32b. A vocabulary of the language of Marau Sound, Guadalcanal, Solomon Islands. BSOAS 6.963-1002. . 1933-35a. A grammar of the language of Bugotu, Ysabel Island, Solomon Islands. BSOAS 7.141-77. . 1933-3 5b. A grammar of the language of Vaturanga, Guadalcanal, British Solomon Islands. BSOAS 7.349-75. . 1933-35c. A grammar of the language of Longgu, Guadalcanal, British Solomon Islands. BSOAS 7.601-21. . 1934. A vocabulary of the Lau language, Big Mala, Solomon Islands. (Polynesian Society Memoir Supplement, 11.) . 1935-37. A grammar of the language of Florida, British Solomon Islands. BSOAS 8.1075-110. . 1937-39a. A vocabulary of the language of Longgu, Guadalcanal, Solomon Islands. BSOAS 9.165-93. . 1937-39b. Melanesian demonstratives. BSOAS 9.385-405. . 1937-39c. A grammar of the language of Lamalanga, North Raga, New Hebrides. BSOAS 9.733-63. . 1939-42. Melanesian modes of speech. JPS 49.579-94, 50.10-40. . 1940. A dictionary of the language of Bugotu, Santa Isabel Island, Solomon Islands. London, Royal Asiatic Society. . 1940-42a. A grammar of the language of Lobaha, Lepers' Island, New Hebrides, Melanesia. BSOAS 10.345-63. . 1940-42b. A grammar of the language of Lotora, Maewo, New Hebrides, Melanesia. BSOAS 10.679-98. KASARHÉROU, J. 1962. Les changements vocaliques de trois préfixes en Motlav. Te Reo 5.32-4. KERN, HENDRIK. 1 8 8 6 . De Fidjitaal vergeleken met hare verwanten in Indonésie en Polynesië. Verhandelingen der Koninklijke Akademie van Wetenschappen, afdeeling Letterkunde 16.1-242.

LANGUAGES OF THE NEW HEBRIDES AND SOLOMON ISLANDS

357

. 1906. Taalvergelijkende verhandeling over het Aneityumsch, met een aanhangsel over het klankstelsel van het Eromanga. VKNA 8.1-146. . 1913-28. Verspreide geschriften. 's-Gravenhage, Martinus Nijhoff. KLIENEBERGER, H. R. 1957. Bibliography of Oceanic linguistics. (London Oriental Bibliographies, 1.) London, Oxford University Press. LANYON-ORGILL, PETER A. 1 9 5 3 . The Papuan languages of the New Georgia archipelago, Solomon Islands. JAS 1 . 1 2 2 - 3 8 . LAWES, W . G . 1 8 8 5 . Grammar and vocabulary of language spoken by Motu tribe (New Guinea). Sydney, Thomas Richards, Government Printer. MACDONALD, DONALD. 1889-91. South Sea languages. Melbourne, Melbourne Public Library. . 1894. The Asiatic origin of the Oceanic languages: etymological dictionary of the language of Efate (New Hebrides). London, Melville, Mullen and Slade. . 1907. The Oceanic languages. Their grammatical structure, vocabulary, and origin. London, Henry Frowde. MILKE, WILHELM. 1958. Zur inneren Gliederung und geschichtlichen Stellung der ozeanisch-austronesischen Sprachen. ZEthn 83.58-62. . 1961. Beiträge zur ozeanischen Linguistik. ZEthn 86.162-82. . 1962. Comment on Capell 1962b. CAnthr 3.415-16. O'REILLY, PATRICK. 1 9 5 8 . Bibliographie des Nouvelles-Hébrides. (Publications de la Société des Océanistes, 8.) Paris, Musée de l'Homme. Pacific Islands Year Book and Who's Who. 1963. 9th ed. Sydney, Pacific Publications. PARKER, GARY J. 1968a. Southeast Ambrym verb inflection and morphophonemics. PL, A 15.27-40b. . 1968b. Southeast Ambrym phonology. OL 7.81-91. . 1969. Morphophonemics of the inalienable nouns in Southeast Ambrym. University of Hawaii Working Papers in Linguistics 7.97-113. RAY, SIDNEY HERBERT. 1893. The languages of the New Hebrides. J R S N S W 27. 101-67. . 1926. A comparative study of the Melanesian island languages. Cambridge, University Press. SCHELLONG, OTTO. 1890. Die Jabim-Sprache der Finschhafener Gegend. Leipzig, Heitz. SCHMIDT, WILHELM. 1899a. Über das Verhältniss der melanesischen Sprachen zu den polynesischen und untereinander. Sitzungsberichte der Kaiserlichen Akademie der Wissenschaften zu Wien, philosophisch-historische Classe 141.1-93. . 1899b. Die sprachlichen Verhältnisse Oceaniens (Melanesiens, Polynesiens, Mikronesiens und Indonesiens) in ihrer Bedeutung für die Ethnologie. Mitteilungen der Anthropologischen Gesellschaft zu Wien 29.245-58. . 1926. Die Sprachfamilien und Sprachenkreise der Erde. Heidelberg, Carl Winter's Universitätsbuchhandlung.

358

GEORGE W. GRACE

. 1940-41. Das Verhältnis der melanesischen zu den polynesischen Sprachen. Anthropos 35-36.379-80. SCHÜTZ, ALBERT J. 1968. A pattern of morphophonemic alternation in Nguna, New Hebrides. PL, A 15.41-7. . 1969a. Nguna texts. (OL Special Publication 4.) . 1969b. Nguna grammar. (OL Special Publication 5.) TAYLOR, C.R.H. 1965. A Pacific bibliography. 2nd ed. Oxford, Clarendon Press. WALSH, D . S. 1 9 6 2 . The phonemes of Raga. Te Reo 5 . 5 7 - 6 0 . . 1966. The phonology and phrase structure of Raxa. M.A. thesis, University of Auckland. WATERHOUSE, J.H.L. 1949. A Roviana and English dictionary. Sydney, Epworth Printing and Publishing House. WELLS, RULON. 1 9 6 3 . Some neglected opportunities in descriptive linguistics. AnL 5/1.38-49. WHALEY, OWEN

G. 1962. Babatana segmental phonemes. Te Reo 5.60-2.

NEW CALEDONIA A N D THE LOYALTY ISLANDS

ANDRÉ G. HAUDRICOURT

We shall class the languages of New Caledonia in five groups: Far North, North, Center, South, and Deep South (see Haudricourt 1964; Hollyman 1964),1 to which should be added the four languages of the Loyalty Islands: one on Maré, one on Lifu, and two on Uvéa. For each of these nine groups we shall examine the history of research and the results obtained, then in a general discussion section we shall see what can be determined from these results from a comparative point of view. I.

SOURCES

Documents which we have concerning the languages of New Caledonia and the Loyalty Islands may be classed in three categories : (1) Documents deficient both phonologically and grammatically, which can be utilized only with great caution. In this category are those produced by the Cook and D'Entrecasteaux expeditions, by the officials G. Glaumont in 1888 (OR # 1903)2 and J. Bernier in 1899 (OR # 1881), by the deportee Louise Michel in 1885 (OR # 1805), and even the work of a German university man, Hans Neverman, in 1935 (OR # 1927). The latter, for example, in a Paaci wordlist transcribes the 1st incl. pi. suffix /"fe/ sometimes as ndje, sometimes nye, and translates it as 'mein'. (2) Documents phonologically deficient but grammatically usable. This category is comprised of works by missionaries, who had to learn the languages in order to preach and to make translations — prayers and catechisms for the Catholics, New and Old Testaments for the Protestants ; however, they were little prepared for studying these difficult languages. The orthography was fixed very rapidly and in 1859 a proclamation by Governor Saisset was translated into 14 dialects. But then in 186? 1

These languages are not always called by the native names in the various sources, but are aiso known by the name of the tribe (locally called 'chefferie') or by the name of the village (locally called 'tribu') where the mission is located. We shall give all these names, then indicate the languages by abbreviations: the first letter designates the group (F, N, C, S, D for Far North, North, Center, South, and Deep South respectively), the second the name of the main tribe, and the last letter, following a hyphen, indicates the relative geographical position (-n 'north', -s 'south', -e 'east', -w 'west'). The names of tribes often persist as names of colonial centers. This is the case for Gomen, Koumac, Voh, Koné, Hyenghène, Touho, Bayes, Houailou, Canala, Thio, Boulouparis, and Nouméa. » Of the four bibliographies by Leenhardt (1946), O'Reilly (1955), Loukotka (1957), and Klieneberger (1957), that of O'Reilly is the most complete. We shall cite printed sources by the letters OR followed bv the item number in O'Reilly's bibliography.

360

ANDRÉ G. HAUDRICOURT

Governor Guillain forbade the missionaries to teach in the native languages and their works remained unpublished; the publication of catechisms began only in 1888. On the other hand, the English Protestant missionaries in the Loyalty Islands did not obey, leading to prolonged conflicts which did not subside until around 1895. At this latter date, on the west coast of New Caledonia native pastors from Maré, and on the east coast from Lifu and Uvéa, began transcribing the New Caledonian languages on the basis of their customary Loyalty Islands orthographies. The Catholics produced catechisms in eight New Caledonian and four Loyalty Islands languages, whereas the Protestants produced biblical translations in three Loyalty Islands languages and in Houailou of New Caledonia. The only extended grammatical works by the Catholics were the Wagap grammar published in 1889 (OR #1886) and the unpublished works of J.M. Dubois on Nengone. Those of the Protestants are represented by the publications of Maurice Leenhardt in 1935 (OR #1916) and 1946 (OR #1913). The works of H.C. von der Gabelentz (OR # 1893-1900) and of Sydney Ray (OR # 1934-1941) are indirect works based on the study of translations published by missionaries. The thesis sustained at Canberra by D.T. Tryon (1966), which in no way improves our knowledge from the viewpoint of phonology, should also be included in this category. (3) Documents phonologically valid. Two conditions were necessary before such documents could be collected: (a) phonemics had to be taught in France, and this began only with the teaching of André Martinet in 1937, and (b) persons other than policemen and missionaries had to be able to contact the natives, which was not possible until after the Second World War. In 1946 the pharmacist Maurice Lenormand, who specialized in Lifu, arrived, and in 1947 Jean Guiart, ethnologist with the Office de la Recherche Scientifique et Technique d'Outre-Mer, began collecting texts. In 1955 George W. Grace came to conduct a study on behalf of the Tri-Institutional Pacific Program. In 1958 Jacqueline de la Fontinelle, a linguist from the Centre National de la Recherche Scientifique who had married her informant, Hippolyte Kasarhérou, left for New Caledonia to study the Houailou language and neighboring dialects. In 1959 I made my first stay, investigating primarily the languages of the north. In 1962 I made a survey of all the languages of New Caledonia, supplemented in 1966. A third researcher from the Centre National de la Recherche Scientifique, Jean-Claude Rivierre, remained from 1965 to 1967 to investigate tonal languages. II.

NEW CALEDONIAN LANGUAGES

1. Far North Group :

Language of the Balade Mission, called Niélaiou (in Leenhardt 1946, yalasu /jalaju/): 3 •

The present pronunciation will be given between slant lines /

/ i n IPA transcription.

NEW CALEDONIA AND THE LOYALTY ISLANDS

As spoken by the Pouma /puma/, Paiak /paaak/, and Obat /ovac/ tribes Dialect of the Belep tribe, islands of Pott /pooc/ and Art / h a a t / . . . . Language of the Pouébo Mission, called /caaac/ : Mwelebeng tribe Dialect of La Conception, called/cawac/, separate since 1855 . . . .

361

Fb-s Fb-n

Fp-n Fp-s

Language of the tribe of Gomen, called nua (Leenhardt 1946) or youanga /jiuaaga/ Fg-s Dialect of the Bondé Mission, Pabwa tribe Fg-n Language of the tribe of Koumac /kumak/, called /fwa-yumak/ (Grace 1955) Fk-s Dialect of the Nénéma tribe Fk-n

a. History of Research (1) Fb-s. In September 1774, 30 words were taken down in English orthography by James Cook (OR #52), and 38 by the Forsters (OR #51). These were compared by Sydney Ray (OR #1934:12-3) with those of Fabre (OR #1891), and were also studied by Hollyman (Haudricourt and Hollyman 1960). In April and May 1793 the D'Entrecasteaux expedition, in search of La Pérouse, collected a sizable vocabulary at the same location; 284 words were published by E.P.E. De Rossel (OR # 6 5 ) and 251 by J. J.H. Labillardière (OR #64). Some years later a Russian sailor, S. Samounov, collected 24 words and the unpublished document was published in 1951 by P. Lanyon-Orgill (OR # 1912). In 1843 Commander J. Laferrière, who brought the first Catholic missionaries, furnished a vocabulary of 139 words (OR # 8 6 ) . The missionaries tried to learn the language and in 1845 the naval surgeon C. Fabre was able to collect a vocabulary of 451 words, translations of prayers by Father Viard, and grammatical information (personal flexion of nouns) by Mgr. Douarre (OR #1891). Another naval surgeon, E. Vieillard, in 1863 (OR #1946) furnished the grammatical notes used in 1873 by von der Gabelentz (OR #1900). In 1855 Father Chapuy lithographed three booklets at Balade: Ba comuli kristiano nan nielaiu (iv + 140 pp) containing catechism, prayers and hymns; Vendiamo ka tin e uendo a pundia malaimipuamendan (84 pp.), a résumé of the Old Testament; and Meditasio (35 pp.), exercises for self-examination. The manuscript vocabularies found at the Pouébo Mission by Hollyman (1962:41) are probably by Father M. Gagnère and anterior to the preceding publications, since the dictionary is entitled 'Puma-français' and not 'Niélaiou'. The notation used was defective on three counts : vowel length was not noted, nor

362

ANDRÉ G. HAUDRICOURT

Vi

Z

Q

¿y



a



a o Oí

-

«