A Grammar of Warrongo

Mouton Gratntnar Library

53
Editors
Georg Bossong
Bernard Comrie
Matthew Dryer
De Gruyter Mouton
A Grammar of Warrongo
by
Tasaku Tsunoda
De Gruyter Mouton
ISBN 978-3-11-023876-1
e-ISBN 978-3-11-023877-8
ISSN 0933-7636
Library of Congress Cataloging-in-Publication Data
Tsunoda, Tasaku.
A grammar of Warrongo I by Tasaku Tsunoda.
p. em. - (Mouton grammar library; 53)
Includes bibliographical references and index.
ISBN 978-3-11-023876-1 (alk. paper)
1. Warungu language - Grammar. 2. Aboriginal Australians -
Australia, Northern - Languages. 3. Australia - Languages.
I. Title.
PL7101.W386f78 2011
499' .15 -dc23
2011038671
Bibliographic information published by the Deutsche Nationalbibliothek
The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie;
detailed bibliographic data are available in the Internet at http://dnb.d-nb.de.
2011 Walter de Gruyter GmbH & Co. KG, Berlin/Boston
Typesetting: Frank Benno Junghanns, Berlin
Printing: Hubert & Co. GmbH & Co. KG, Gottingen
oo Printed on acid-free paper
Printed in Germany
www.degruyter.com
Preface
What follows is a description of the Warrongo language, once spoken in the upper
Herbert River area of north Queensland, Australia. I conducted fieldwork on sev-
erallanguages in north Queensland from 1971 to 1974. They were all endangered
at that time. I worked mainly on Warrongo, recording it from perhaps about ten
people. But almost all the data on it were provided by the late Mr. Alf Palmer
(Warrongo name: Jinbilnggay), the last fluent speaker of the language. A very
small amount of information -nonetheless invaluable -was recorded from the
late Mr. Alec Collins (Warrongo name: Wolngarra) by Peter Sutton and by me.
Alf Palmer passed away in 1981. But towards the end of the 20th century, a
movement to revive the ancestral languages of the area (including Warrongo)
started, and I have been conducting Warrongo language lessons in Townsville
since 2002 (Tsunoda (2002, 2004, 2005: 212-213) and Tsunoda and Tsunoda (2006,
2007, 2008, 201 0)).
The first outcome of the fieldwork in the early 1970s was my M.A. thesis
(Tsunoda 1974a), submitted to Monash University. Subsequently I wrote about
various aspects of Warrongo: Tsunoda (1976a, 1976b, 1984, 1987, 1988a, 1988b,
1990, 1992, 1995, 1997, 1998, 1999, 2006, 2007, 2008), among others. The main
points of these papers are incorporated into the present work. With permission
from Mrs. Rachel Cummins (nee Rachel Wilson), who is Mr. Alf Palmer's
daughter's daughter, I conducted a course on the Warrongo language at the
University of Tokyo in the academic years 2003, 2005, 2007 and 2009. The pre-
sent work is almost entirely based on the lecture notes for that course. Some of the
comments from the students are incorporated into the present work.
The present grammar is not a revision of Tsunoda (1974a). It is an entirely new
work. It contains the following parts: Chapter 1 The language and its speakers,
Chapter 2 Phonology, Chapter 3 Word classes and morphology, Chapter 4 Syntax,
and Texts. This format is based on the guideline Handbook of Australian Lan-
guages prepared by R. M. W Di.xon and Barry J. Blake. Inevitably there are
overlaps among the chapters. For example, Chapter 3 contains a large amount of
syntactic information. There is no separate chapter for semantics, discourse, or
comparative-historical linguistics. Nonetheless some information on semantics is
included in Chapters 1, 3, and 4; some information on discourse in Chapter 4; and
some information on comparative-historical linguistics in Chapters 1 to 4.
The present work is a description of a language that was already endangered
when the fieldwork was conducted. In view of this, in each chapter, I make an
effort to provide as detailed a description as possible, accompanied by as many
examples as possible. I even report a phenomenon that may not have existed
when the language was thriving; see 2.11. This is because, when working on an
endangered language, we can never tell what may turn out to be important for
vi Preface
the linguistic academia and/or the community concerned. It is important to bear
in mind that 'the data one will be collecting may well be all that there will be of
documentation of the language' (Grinevald 2007: 43), and that 'any record made is
likely to be used for other purposes later on' (Mithun 2001: 34).
Furthermore, in works on endangered languages, the 'reliability of the data
needs to be assessed, and care must be taken in order to provide a reliable docu-
mentation' (Tsunoda 2005: 244). In view of this, for each sentential example, I
indicate whether it was given by AlfPalrner ('AP') or by Alec Collins ('AC'). For
many (though not all) of the examples given by AlfPalrner, I indicate whether it is
cited from a text, it was uttered spontaneously during a conversation between Alf
Palmer and me, or it is a Warrongo tmnslation of an English sentence. For many
(though not all) of the examples cited from a text or our conversation, the context
in which that sentence was uttered, is presented.
The examples that were composed by me too late to be checked by AlfPalrner
are marked with '(TT)' The vast majority of the examples marked by '(AP') were
uttered by Alf Palmer either in a text or during our conversation. However, they
(marked with '(AP)') include a very small number of examples that were composed
by me and approved by Alf Palmer; the abbreviation '(TT, AP)' is used when it is
deemed useful to emphasize that the example in question is such an example. In
addition it is important to mention that maximum caution is exercised in presen-
ting Table 3-14 Verbal paradigm (1).
As can be seen, the present work is an example of what can be done, and what
cannot be done, when only the last single speaker is available.
In addition to north Queensland (where I conducted field work from 1971 to
1974), I have been carrying out fieldwork in Kimberley, Western Australia since
1975, working on Djaru, Wanyjirra and a few other languages; one of the out-
corn es is Tsunoda (1981 a) on Djaru. A similarity or a difference between Warrongo
and Djaru/Wanyjirra will be sometimes be pointed out, for this may be useful to a
reader who is not familiar with Austmlian languages.
August 2011 Tasaku Tsunoda
National Institute for
Japanese Language and Linguistics
Tachikawa City, Tokyo, Japan
Acknowledgements
The present work is dedicated to the memory of the late :Mr. Alf Palmer, to whom
my most sincere words of gratitude are due. He was an extremely intelligent and
helpful consultant. Also he was entertaining and humorous, as can be seen in a fair
num her of sentential examples given below. He was deeply concerned about the
fate of his language, and was eager to have his language recorded for posterity. He
used to say to me, 'I'm the last one to speak Warrongo. When I die, this language
will die. I'll teach you everything I know, so put it down properly' (Tsunoda 2005:
vii). Indeed. he made admirable efforts to each me everything he knew (Tsunoda
2004: 273-274). In retrospect, it was Alf Palmer who taught me the importance of
documenting endangered languages. It is truly unfortunate that I was not compe-
tent enough to fully document his knowledge. It is also unfortunate that there was
no opportunity to visit the Warrongo country with him to check placenames,
etc., or just to see the country where he was born and grew up. (See 1. 8.4.1 for his
biography and an account of his linguistic knowledge.)
Alf Palmer named me after his uncle no doubt, mother' brother (MB), and
not father's brother (FB) (see 1.5.4.2-[2]). This has an important implication.
According to Berndt and Berndt (1965: 87-88), in the traditional Aboriginal socie-
ties, MB and father's sister (FZ) played pivotal and crucial roles, involving 'special
obligations and responsibilities'. In view of this, it is an honour and privilege to
be named after Alf Palmer's uncle (no doubt :MB). At the same time, his inten-
tion may have been to assign me an obligation and responsibility to document his
language to pass it on to posterity. In Rachel Cummins' words, he entrusted the
Warrongo language to me.
The Warrongo language revival movement is making a slow but steady pro-
gress (see Tsunoda and Tsunoda (2010), among others). Alf Palmer's dedicated
efforts to have his language documented have proved to be truly worthwhile. What
he sowed four decades ago is now beginning to be harvested by his descendants.
Also in the 1970s, the following people each provided a limited amount of-
nonetheless invaluable- data on Warrongo: Alec Collins (1.8.4.2.), and also Harry
Bunn, Vera Smallwood, :Mr. Williamson, :Mrs. Morgenstern, Denny Hoolihan,
:Mrs. Cassidy, Tommy Murray, Tommy Springcart, and Ado Cashmere (1.8.4.3).
The then Australian Institute of Aboriginal Studies (now Australian Institute of
Aboriginal and Torres Strait Islander Studies) and Monash University financed my
research into the Warrongo language in the 1970s.
Barry J. Blake supervised my M.A. study (and also Ph.D. study) at Monash
University, and he has been continuing to give me advice.
Peter Sutton made available all his unpublished data on Warrongo recorded
from Alec Collins, and also his unpublished data on Gugu-Badhun and Gujal. I
recorded a very small amount of data from Alec Collins (see 1.8.4.2), but all the
viii Acknowledgements
examples provided by Alec Collins that are given below were recorded by Peter
Sutton.
Gavan Breen made available his data on Gujal recorded from George Reid.
R. M. W. Dixon made available his unpublished data on Warrongo recorded
from Alf Palmer, and also provided a photocopy of a word list of the Ngaygungu
language (1.4.1): 'Tribes of Aboriginal met with around Atherton' by Department
of Commissioner of Police, Brisbane, dated 4 November 1898.
In the 2000s, during my participation in the Warrongo language movement,
Rachel Cummins (Alf Palmer's daughter's daughter) provided a large amount of
information, in particular, on the biography of AlfPalmer (1.8.4.1). She also issued
permission to conduct the class on Warrongo at the University of Tokyo, and to
publish the present volume.
Willy Santo (Gujal group) and Shirley Johnson (Wulgurugaba group), both
interviewed in 2001, supplied information on the mythology of the region. (See
1.5.3.)
Leigh Pentecost and Rod Nielson provided information on aspects of the geo-
graphy and history of the region. (See 1.7.)
The students of the class on Warrongo that I conducted at the University of
Tokyo supplied helpful comments.
The comments on my previous works on Warrongo were gratefully acknow-
ledged therein.
Tatsuyuki Mimura and Kan Sasaki each read a draft of Chapter 2 and furni-
shed comments and references. Kan Sasaki detected a fair number of typos in 4.11.
Meaghan Cummins (Rachel Cummins' daughter), Kazuhiro Imanishi, Yasuhiro
Kojima, Yusuke Nakamura, Ken Shibushita, Mie Tsunoda, and John B. Whitman
each provided comments on, or gave advice relevant to, portions of the drafts of
the present work.
Yasunari Imamura checked the numbering of examples. Minoru Yamaizumi
examined the contents and pagination. Hatsue Ishii prepared the basis of :Maps 1,
3 and4.
The National Institute for Japanese Language and Linguistics (Tachikawa City,
Tokyo, Japan), where I serve as the Director of the Department of Crosslinguistic
Studies, has supported my research, including the preparation and the publication
of the present work. Parts of the present work are an outcome of the research I car-
ried out for the collaborative research project that I am conducting, as the project
leader, at the Institute: 'Modal and speech-act constraints on clause-linkage'
The anonymous reviewer of the manuscript and Bernard Comrie, one of the
editors of Mouton Grammar Library, furnished very detailed and helpful comments.
The late Ursula Kleinhenz, Julie Miess and Kirstin Borgen at De Gruyter
Mouton supplied much-needed editorial assistance and advice.
My wife, Mie Tsunoda, and her mother, Suzu Mori, provided full support
during the preparation of the present work, as always.
Contents
Preface v
Acknowledgements. Vll
List of tables and figures XlX
List of maps and photos XXl
List of abbreviations and symbols XXll
Chapter 1: The language and its speakers . 1
1.1. Linguistic type . 1
1.2. Names of the language and people 2
1.3. Dialects. 3
1.4. Territory and neighbouring languages 4
1.4.1. Territory 4
1.4.2. Neighbouring languages and their classification 6
1.4.2.1. Proposed classifications. 6
1.4.2.2. Studies on the neighbouring languages. 6
1.4.2.3. Warrongo, Gugu-Badhun and Gujal as a linguistic unity. 8
1.4.2.4. Problems with the proposed classifications. 10
1.5. Environmental and socio-cultural background . 14
1.5.1. Environmental setting 14
1.5.2. Archaeological and anthropological accounts. 15
1.5.3. Mythology 15
1.5.4. Names of groups, individuals, and places. 19
1.5.4.1. Names of groups 19
1.5.4.2. Names of individuals. 21
1.5.4.3. Names of places 22
1.5.5. Sections and totems. 25
1.5.6. Marriage rules 27
1.5.7. Kinship system. 28
1.5.8. Other topics 32
1.6. Special styles of speech and songs. 33
1.6.1. Jalngoy: the avoidance style of speech 33
1.6.2. Songs 34
1.6.3. Curses and expressions for abuse 37
1.7. Post-contact history 38
1.8. Studies on the Warrongo language 40
1.8.1. Introductory notes 40
1.8.2. Early studies on Warrongo. 40
1.8.3. Modem studies on Warrongo 41
X Contents
1.8.4. Speakers interviewed for the present study. 43
1.8.4.1. Alf Palmer (Warrongo name: Jinbilnggay). 43
1.8.4.2. Alec Collins (Warrongo name: Wolngarra) 47
1.8.4.3. Other speakers . 48
1.9. Present-day situation . 51
Chapter 2: Phonology. 53
2.1. Phonemes and their realizations 53
2.1.1. Phoneme inventory 53
2.1.2. Minimal pairs/sets. 55
2.1.3. Allophones of consonants and sem ivowels. 59
2.1.3.1. Nasals, rhotics, lateral, and sem ivowels 59
2.1.3.2. Stops. 60
2.1.3.2.1. Stop voicing (1): in terms of places of articulation. 60
2.1.3.2.2. Stop voicing (2): in the word-initial position. 61
2.1.3.2.3. Stop voicing (3): in the second syllable. 65
2.1.3.2.4. Stop voicing ( 4): effect of C2 nasal on Cl stop. 67
2.1.3.2.5. Stop voicing (5): effect of Vl on Cl stop, and of V2 on C2 stop 68
2.1.3.2.6. Stop voicing (6): in consonant clusters 71
2.1.3.2.7. Stop voicing (7): concluding remarks 74
2.1.4. Allophones of vowels. 75
2.1.4.1. Ia! 75
2.1.4.2. lui 76
2.1.4.3. Iii 76
2.1.4.4. Iii and lui 80
2.1.5. Problems with /j/ and lw/ 84
2.1.5.1. ljl 84
2.1.5.1.1. Introductory notes 84
2.1.5.1.2. Allophones of /jil 84
2.1.5.1.3. Allophones of ljl that is not followed by a vowel 88
2.1.5.2. lwl. 92
2.2. Phonotactics. 94
2.2.1. Structure of words . 94
2.2.2. Characterization of enclitics 95
2.2.3. Syllable structure of roots, suffixes, enclitics, and words 97
2.2.3.1. Sy liable structure of roots 97
2.2.3.2. Syllable structure of suffixes and enclitics. 99
2.2.3.3. Syllable structure of words 100
2.2.4. Syllable types in roots, suffixes, enclitics, and words 101
2.2.5. Distribution of consonants and semivowels 105
2.2.5.1. Consonants and semivowels in roots. 105
2.2.5.2. Consonants and semivowels in suffixes 107
2.2.5.3. Consonants and semivowels in enclitics. 108
Contents Xl
2.2.5.4. Consonants and semivowels in words. 108
2.2.6. Consonant clusters. 108
2.2.6.1. Intra-root consonant clusters. 108
2.2.6.2. Inter-morphemic consonant clusters. 112
2.2.6.3. Comparison and analysis of intra-root and inter-morphemic
consonant clusters 114
2.2.7. Syllables: onset and coda. 119
2.3. Statistics. 121
2.3.1. Statistics (1 ): dictionary count. 121
2.3.2. Statistics (2): text count. 123
2.4. Alternation between phonemes. 124
2.5. Natural classes. 128
2.6. 'Phonological processes' 130
2.7. Stress 133
2.8. Pitch. 141
2.9. Connected speech 142
2.9.1. Intonation. 142
2.9.2. Other phenomena. 145
2.10. Careful speech. 147
2.10.1. Deletion of a phoneme. 147
2.10.2. Isolation of suffixes . 149
2.11. Women's pronunciation(?) 150
2.12. Ideophones. 151
2.13. Adaptation of foreign words 152
2.13.1. Words from other Australian languages . 152
2.13.2. Words from English. 152
2.14. Practical orthography 155
Chapter 3: Word classes and morphology 156
3.1. Word classes 156
3.1.1. Outline of word classes 156
3.1.2. Additional comments on individual word classes. 157
3.2. Noun morphology (1): case inflection. 164
3.2.1. Case inflection of nouns and three demonstrative words. 164
3.2.1.1. Regular nouns 165
3.2.1.2. Vowel-final proper and kin nouns. 169
3.2.1.3. Irregular nouns 171
3.2.1.4. Ngani 'what', yarro 'this, here', ngona 'that, there', and ngoni
'there, that' 173
3.3. Pronoun morphology 174
3.4. Adverb morphology (1 ): case inflection . 176
3.4.1. Types of adverbs 176
3.4.2. Wanyja 'where' 177
Xll Contents
3.4.3. Comparison of adverbs and nouns: case inflection 178
3.5. Notes on case inflection 183
3.5.1. Silverstein's NP hierarchy 183
3.5.2. Genitive, comitative and ablative cases. 184
3.5.3. Nouns with no case inflection (?) 184
3.6. Meanings and functions of cases. 185
3.6.1. Nominative case. 185
3.6.2. Ergative and accusative cases 187
3.6.3. Locative case 190
3.6.4. Dative case. 195
3.6.5. Genitive case 201
3.6.6. Ablative-2 case. 205
3.6.7. Ablative-1 case. 205
3.6.8. Comitative case 210
3.7. Noun morphology (2): derivation 222
3.7.1. Noun-stem-forming suffixes. 222
3.7.2. Reduplication. 238
3.7.3. Compounding 241
3.8. Adverb morphology (2): derivation. 245
3.8.1. Adverb-stem-forming suffixes 245
3.8.2. Reduplication. 251
3.8.3. Compounding 252
3.9. Verb morphology (1): introduction and inflection 253
3.9.1. Verb roots and transitivity 253
3.9.2. Verbalization. interrogative verbs, and demonstrative verbs 254
3.9.3. Conjugational classes. 255
3.9.4. Verbal paradigms. 259
3.9.4.1. Verbal paradigm (1): L-class, Y-class and ZERO-class 259
3.9.4.2. Verbal paradigm (2): derived stems 264
3.10. Meanings and functions of conjugational categories 266
3.10.1. Past. 266
3.10.2. Nonfuture-1 267
3.10.3. Nonfuture-2 271
3.10.4. Nonfuture-3 274
3.10.5. Future. 276
3.10.6. Comparative notes on tenses: past, present and future . 277
3.10.7. Stative 277
3.10.8. General imperatives. 278
3.10.9. Mfirmative imperative 284
3.10.10. Negative imperative. 285
3.10.11. Apprehensional 286
3.10.12. Four purposives 288
3.10.13. Purposive- I and purposive-2. 291
3.10.14.
3.10.15.
3.11.
3.11.1.
3.11.1.1.
3.11.1.2.
3.11.1.3.
3.11.1.4.
3.11.1.5.
3.11.1.6.
3.11.1.7.
3.11.2.
3.11.3.
Purposive-3
Purposive-4
Verb morphology (2): derivation.
Verb-stem -forming suffixes
-rra-(.v)-mbi-L '(song style ?)'
-nda-Y 'repeatedly', 'plural actors'
-ya-L 'all', 'all the time'
-rV-L 'pluml' ('all', 'many').
-yi-? 'future'
-1 and -i.
-rrand -i.
Reduplication of verb roots .
Compounding
Contents xm
296
296
298
298
298
300
301
301
303
304
311
315
316
Chapter 4: Syntax. 318
4.1. Introductory notes 318
4.2. Types of sentences. 318
4.2.1. Sentence types (1): declarative, imperative, interrogative, and
exclamatory sentences. 318
4.2.1.1. Imperative sentences. 318
4.2.1.2. Interrogative sentences 320
4.2.1.3. Exclamatory and declamtive sentences. 324
4.2.2. Sentence types (2): affirmative and negative sentences . 324
4.2.3. Sentence types (3): simple and complex sentences. 324
4.3. Types of clauses. 325
4.3.1. Clause types (1 ): intransitive and transitive clauses . 325
4.3.2. Clause types (2): basic and derived clauses 326
4.3.3. Clause types (3): in terms of valency 326
4.3.3.1. Introductory notes. 326
4.3.3.2. Three-place clauses 327
4.3.3.3. Two-place clauses 336
4.3.3.4. One-place clauses 337
4.4. Constituents of clauses 338
4.5. Types of phrases 340
4.5.1. Introductory notes 340
4.5.2. Noun phrases. 341
4.5.2.1. Non-configurational NPs 342
4.5.2.1.1. Constituents of non-configurational NPs 342
4.5.2.1.2. Types of non-configurational NPs 342
4.5.2.1.3. Order of the constituents of non-configurational NPs 347
4.5.2.1.4. Additional remarks on non-configurational NPs 351
4.5.2.2. Configurational NPs 352
4.5.2.2.1. Genitive in NPs 353
XlV Contents
4.5.2.2.2. 'One's camp' (1). 354
4.5.2.2.3. 'One's camp' (2). 355
4.5.2.2.4. 'One's own camp' 358
4.5.2.2.5. 'One's/this camp' 361
4.5.2.2.6. 'His/her father' 361
4.5.2.2.7. Additional examples 363
4.5.2.2.8. Discussion and summary of configurational NPs. 364
4.5.3. Adverb phrases 365
4.5.4. Verb phrases 366
4.5.4.1. Verb phrases consisting of two verbs 367
4.5.4.2. Verb phrases consisting of a verb and a noun or adverb. 370
4.6. Order of constituents . 372
4.6.1. Constituent order in words 372
4.6.2. Constituent order in phrases 374
4.6.3. Constituent order in clauses 374
4.6.3.1. Introductory notes 374
4.6.3.2. Constituent order in those independent clauses and main clauses
which are used as affirmative declarative sentences 375
4.6.3.3. Constituent order in those independent clauses and main clauses
which are used as imperative sentences 380
4.6.4. Constituent order in sentences. 382
4.7. Verb formation: -bi-L 'INTR', -nga-L 'TR', and -(m)ba-L 'TR' 383
4.7.1. Intransitive-stem-forming suffix -bi-L 'be, become'
(state, change/inchoation) 384
4.7.2. Transitive-stem-forming suffix -nga-L 'make', etc.
(causation, creation, etc.). 391
4.7.3. Transitive-stem-forming suffix -(m)ba-L 408
4.8. Complex sentences involving purposive subordination 412
4.8.1. Introductory notes 412
4.8.2. Corn bination of clauses 413
4.8.3. Coreference and syntactic ergativity 414
4.8.4. Coreferential deletion in purposive subordination 415
4.8.5. More on purposive subordination 422
4.9. Antipassive construction. 427
4.9.1. Case frames and verbal morphology 427
4.9.2. Transitivity. 430
4.9.3. Coreference: antipassives and syntactic ergativity 431
4.9.4. Purposive subordination 432
4.9.4.1. Purposive subordination without an antipassive. 432
4.9.4.2. Purposive subordination with an antipassive 433
4.9.5. Coordination and sentence-sequence 438
4.9.6. Summary of coreference. 443
Contents XV
4.9.6.1. Comparison of purposive subordination, coordination, and
sentence-sequence. 443
4.9.6.2. Crosslinguistic comparison of syntactic ergativity. 445
4.9.7. More on antipassives in purposive subordination. 445
4.9.8. Behaviour of pronouns 452
4.9.8.1. Voice distinction 452
4.9.8.2. Coreferential deletion and retention 453
4.9.8.2.1. Performance errors 453
4.9.8.2.2. Deletion and retention 455
4.9.9. Case frames and semantics 459
4.9.9.1. Semantic types of antipassive verbs 459
4.9.9.2. Case ofthe undergoerNP 464
4.9.9.2.1. Introductory notes. 464
4.9.9.2.2. Animacy of the undergoer in terms of Silverstein's NP hierarchy 464
4.9.9.2.3. Affectedness of the undergoer in terms of verb semantics. 467
4.9.9.2.4. Affectedness of the undergoer of action antipassives in terms
of the conjugational category 469
4.9.9.3. Perception and knowledge antipassives 476
4.9.9.3.1. Semantics and case of the undergoer NP 476
4.9.9.3.2. N,.vaga-L and nyaga-gali-ZERO 'see, look at, etc.' 478
4.9.9.3.3. Nyaga---nyaga-L 'look at, look after, mind, watch out' and
nyaga---nyaga-gali-ZERO 'look at, watch out' 479
4.9.9.3.4. Jaymba-L 'find' andjaymba-gali-ZERO 'find, search for' 480
4.9.9.3.5. Ngawa-L andngawa-gali-ZERO 'hear, listen to' and 'understand' 481
4.9.9.3.6. Additional observations on perception and knowledge antipassives 483
4.9.9.4. NOM-GEN and NOM-ACC for antipassives 483
4.9.9.5. Case frames of three-place antipassives 485
4.9.9.6. Case ofNPs for instrument, location, and destination 491
4.9.9.6.1. Case ofNPs for instrument. 491
4.9.9.6.2. Case ofNPs for location 496
4.9.9.6.3. Case ofNPs for destination. 497
4.9.9.6.4. Case ofNPs: discussion. 497
4.9.10. Aspectual and modal meanings 498
4.9.11. Pragmatics and discourse 502
4.9.11.1. Introductory notes. 502
4.9.11.2. Continuous mentions of referents 502
4.9.11.3. Focus 505
4.1 0. Subordination with the enclitic =ngomay 'after, if, because' 506
4.1 0.1. Introductory notes 506
4.10.2. Involving an intransitive verb or an adjective-like noun 507
4.10.3. Involving an antipassive verb 509
4.10.4. Involving a transitive verb. 512
4.10.5. Discussion 513
XVI Contents
4.11.
4.11.1.
4.11.2.
4.11.2.1.
4.11.2.2.
4.11.3.
4.11.3.1.
4.11.3.2.
4.11.4.
4.11.5.
4.12.
4.12.1.
4.12.2.
4.12.3.
4.12.4.
4.12.4.1.
4.12.4.2.
4.12.5.
4.12.5.1.
4.12.5.2.
4.12.5.3.
4.12.6.
4.12.6.1.
4.12.6.2.
4.12.6.3.
4.12.6.4.
4.12.7.
4.12.8.
4.12.9.
4.12.9.1.
4.12.9.2.
4.12.9.3.
4.13.
4.13.1.
4.13.2.
4.13.2.1.
4.13.2.2.
4.13.3.
4.13.3.1.
4.13.3.2.
4.13.4.
4.13.4.1.
Reflexive and middle constructions
Introductory notes
Vt-gali-ZERO 'reflexive' (AP)
Vt-gali-ZERO 'true reflexive' (A&O d-S) (AP)
Vt-gali-ZERO 'anticausative' (0 d-S) (AP)
Vt-li-ZERO 'reflexive' (AP, AC).
Vt-li-ZERO 'true reflexive' (A&O d-S) (AP, AC)
Vt-li-ZERO 'anticausative' (0 d-S) (AP, AC)
Vi-gali-ZERO and Noun-gali-ZERO 'middle' (AP).
Vi-Ii-ZERO, Vi-1-Vi-li-ZERO, Noun-li-ZERO and
Adverb-li-ZERO 'middle' (AP) .
Reciprocal construction.
Introductory notes .
Reciprocal verbs
Syntax
Semantics.
Completely coreferential and partly coreferential.
Simultaneous and alternate
Lexicalization
Reciprocals with a singular d-S.
Higher frequency of reciprocal verbs
Idiom-like expressions.
Non-reciprocal uses.
Sociative 'together'
Participating in a group action
Reflexive
Anticausative .
Expressions of the d-S and the S.
More on the semantics of reciprocals
More on the formation of reciprocal verbs
Morphological restrictions.
Reciprocal verb based on a noun.
Etymology
Iterative construction.
Iterative verbs
Syntax
Intransitive-garra-Y
Tmnsitive-garra-Y
Semantics.
Aspectual meanings: the imperfective
Obligatory vs. non-obligatory use of -garra-Y
More on the formation of iterative verbs.
Iterative verbs based on an adverb and a noun
515
515
516
516
523
524
525
530
533
535
537
537
538
540
546
547
548
550
550
554
554
555
555
556
557
558
559
559
560
560
561
561
562
562
566
566
567
570
570
573
575
575
4.13.4.2.
4.14.
4.14.1.
4.14.2.
4.14.2.1.
4.14.2.2.
4.14.2.3.
4.14.2.4.
4.14.2.4.1.
4.14.2.4.2.
4.14.2.4.3.
4.14.2.4.4.
4.14.2.4.5.
4.14.2.5.
4.14.3.
4.14.3.1.
4.14.3.2.
4.14.3.3.
4.14.3.3.1.
4.14.3.3.2.
4.14.3.3.3.
4.14.3.3.4.
4.14.3.4.
4.14.3.5.
4.14.3.5.1.
4.14.3.5.2.
4.14.4.
4.14.4.1.
4.14.4.2.
4.14.4.3.
4.15.
4.15.1.
4.15.1.1.
4.15.1.2.
4.15.1.3.
4.15.2.
4.15.2.1.
4.15.2.2.
4.15.2.3.
4.15.2.4.
4.15.3.
4.16.
4.17.
-n-garra-n
Applicative construction
Introductory notes
Verbal locative construction: Vi-ri-L
Locative verbs
Case frames
Transitivity
Use
Frequency.
Use in purposive subordination: syntactic ergativity
Lexicalized locative verbs .
'ADJECTIVE to VERB' construction
Word order: topic (?)
Locative verbs with the causative meaning.
Verbal instrumental construction: Vt-ri-L
Instrumental verbs.
Case frames and valency
Use
Syntactic possibilities
Use in purposive subordination: syntactic ergativity
Lexicalized instrumental verbs .
'X is for VERBing with' construction.
:More on the morphology
Restrictions (?) on the use of instrumental verbs
Inflectional suffixes.
Pronouns for the DAT
Restrictions on the use of applicative verbs
Derivational suffixes .
Animacy of the A NP
Meaning of the 'basic' NP
Participial subordination
Participle-I -nyo.
Introductory notes
Participle-I used in a subordinate clause.
Participle-I used in a simple sentence.
Participle-2 -nji
Introductory notes
Participle-2 used in a subordinate clause.
Participle-2 used in a simple sentence.
Participle-2 used in songs
Participle-3 -njirra .
Relative clause
Apprehensional construction 'lest ... should'
Contents XVll
575
576
576
576
576
578
581
581
581
581
583
584
584
585
586
586
588
590
590
595
597
597
598
598
598
600
600
600
601
601
602
602
602
604
607
608
608
609
611
611
612
613
614
xvm Contents
4.17.1.
4.17.2.
4.18.
4.18.1.
4.18.2.
4.18.3.
4.18.4.
4.18.5.
4.18.6.
4.18.6.1.
4.18.6.2.
4.19.
4.20.
4.20.1.
4.20.2.
4.20.3.
4.20.4.
4.20.5.
4.20.5.1.
4.20.5.2.
4.20.5.3.
4.20.5.4.
4.20.6.
4.20.7.
4.20.8.
4.21.
4.22.
4.23.
4.24.
4.25.
4.26.
Texts.
Introductory notes
Comparison of apprehensional construction and purposive
subordination.
Subordination without any morphosyntactic marker
Emotion (1): 'X is afraid that ... '
Emotion (2): 'X is glad that
Knowledge: 'X knows that ...
Perception: 'X sees/hears Y doing
Speech activity 'X tells/asks
Condition 'if' and future time 'when, while'
With a subordinate clause containing a verb.
With a subordinate clause containing no verb
'And', 'but' and 'for'
Expression of possession and existence
Introductory notes
Possession cline
Genitive case
Dative, ablative-1, and comitative cases
Apposition of possessor/whole and possessee/part.
Introductory notes
Apposition (1): accusative, ergative, dative, locative, and
ablative-1 cases
Apposition (2): nominative
Apposition (3): discussion .
X Y nyawa 'NEG': 'X has no Y', 'There is no Y in X'
Subordination involving a body part noun
Transitive verb ganyji-L 'carry'
Ngani 'Whatsisname'
Sentence-topic word
Yamanyon 'similar, like' word and enclitic(?).
Adverbs of modality and the like
Enclitics
Interjections.
Text 1. An excerpt from Tape 72/26.
Text 2: An excerpt from Tape 72/28.
Text 3: An excerpt from Tape 72/23.
References.
Index of subjects.
Index of languages
Index of names
614
617
618
619
621
622
622
623
629
629
633
634
636
636
636
637
642
644
644
645
650
657
658
661
664
665
669
671
673
682
698
700
700
713
719
723
736
746
748
List of tables and figures
Table 1-1. Wurm and Hattori's (1981) classification 7
Table 1-2. Dixon's (2002) classification 7
Table 1-3. Lexical sharing among Warrongo, Gugu-Badhun and Gujal 8
Table 1-4. Oppositions in laminal consonants 9
Table 1-5. Selected pronouns of Warrongo and Jirrbal. 11
Table 1-6. Phonological correspondences. 12
Table 1-7. Warrongo section terms (1) (AlfPalmer's Warrongo). 25
Table 1-8. Warrongo section terms (2) (AlfPalmer's Warrongo). 26
Table 1-9. Totems 27
Table 1-10. Marriage rules 28
Table 2-1. Consonant and semivowel phonemes 53
Table 2-2. Vowel phonemes. 53
Table 2-3. Voicing tendency of stops 61
Table 2-4. Voicing of stops in word-initial positions (C1) 62
Table 2-5. Voicing of stops in the second syllables intervocalically (C2) 65
Table 2-6. Effect of C2 nasal on Cl stop. 67
Table 2-7. Effect of V1 on C1 68
Table 2-8. Effect of V2 on C2 70
Table 2-9. Stop voicing in liquid-plus-stop clusters (1) 72
Table 2-10. Stop voicing in liquid-plus-stop clusters (2) 73
Table 2-11. Lowering effect of lrl 78
Table 2-12. Iii and lui 80
Table 2-13. Allophones of /ji/ 85
Table 2-14. Allophones of ljl 89
Table 2-15. Evidence for word-final/j/ 90
Table 2-16. Syllable structure of words, roots, enclitics, and suffixes. 105
Table 2-17. Phonemes in structural positions. 106
Table 2-18. Intra-root consonant clusters 109
Table 2-19. Inter-morphemic consonant clusters 113
Table 2-20. Phoneme frequency (1): dictionary count 121
Table 2-21. Frequency of vowels. 122
Table 2-22. Phoneme frequency (2): text count 123
Table 3-1. Case labels: ergative, nominative, and accusative. 158
Table 3-2. Regular nouns. 166
Table 3-3. Vowel-final human proper nouns and kin nouns. 169
Table 3-4. Irregular nouns. 172
Table 3-5. Ngani 'what',yarro 'this, here', ngona 'that, there', and
ngoni 'there, that' 173
Table 3-6. Pronouns, including 'who' 174
XX List of tables and figures
Table 3-7.
Table 3-8.
Table 3-9.
Table 3-10.
Table 3-11.
Table 3-12.
Table 3-13.
Table 3-14.
Table 3-15.
Table 3-16.
Table 3-17.
Table 4-1.
Table 4-2.
Table 4-3.
Table 4-4.
Table 4-5.
Table 4-6.
Table 4-7.
Table 4-8.
Table 4-9.
Table 4-10.
Table 4-11.
Table 4-12.
Table 4-13.
Table 4-14.
Table 4-15.
Table 4-16.
Table 4-17.
Table 4-18.
Table 4-19.
Table 4-20.
Table 4-21.
Table 4-22.
Table 4-23.
Table 4-24.
Table 4-25.
Table 4-26.
Table 4-27.
Table 4-28.
Table 4-29.
Wanyja 'where'
Comparison of adverbs and nouns.
Silverstein's NP hierarchy
-ngomay 'ablative-I' plus another case
Comitative plus another case .
Meanings of nouns when used in compounds.
Conjugational classes and transitivity
Verbal paradigm (1 ): L-class, Y-class and ZERO-class
Verbal paradigm (2): reciprocal verbs; iterative verbs; and
antipassive verbs and reflexive/middle verbs
Past, present and future
General imperatives and purposives
Case frames of goyba-L 'give'
Case frames of ngonbaynga-L 'show, teach'
Case frames of mayga-L 'tell'
Case frames of ganyji-L 'carry, take, bring'
Locative and dative of genitive singular pronouns
Locative and dative of 'pronoun-LINK'
Cases of 'one's camp'
Cases of 'one's own camp'
Cases of 'his/her father'
Head and dependent.
Repetition of Vi in intransitive sentences
Combination of clauses involving purposive subordination .
Ideal coreferential deletion patterns involving purposive
subordination
Basic transitive clauses and antipassive clauses
Purposive subordination without an antipassive .
Purposive subordination with an antipassive.
Summary of coreference
Coreferential deletion and retention in purposive subordination
Hierarchy of two-place predicates .
Transitive verbs and voice expressions
Case of undergoer NP (1 ): animacy hierarchy.
Case ofundergoer NP (2): verb semantics.
Case of undergoer NP (3): conjugational category of action
antipassives.
Perception antipassives
'See, look at', 'find', and 'search for'
Goyba-L 'give' and 'throw' transitive and antipassive.
Mayga-L 'tell': transitive and antipassive
Ganyji-L 'carry, bring, take': transitive and antipassive
Birra-L 'tell, talk': transitive and antipassive
177
179
183
205
210
242
256
260
265
277
280
327
330
332
335
354
356
357
359
362
373
380
413
418
427
432
434
443
455
459
460
464
467
469
477
480
486
489
490
491
List of maps and photos XXl
Table 4-30. Case ofNPs for instrument 492
Table 4-31. Case ofNPs for location. 496
Table 4-32. Continuous mentions of referents 505
Table 4-33. Transitive and reflexive clauses (1): instrument ERG. 518
Table 4-34. Transitive and reflexive clauses (2): body part noun. 519
Table 4-35. Transitive and anticausative clauses 523
Table 4-36. Two-place transitive and corresponding reciprocal clauses 540
Table 4-37. Goli waga-L 'get angry' and goli waga-wa-Y 'get angry-RECP' 540
Table 4-38. Goyba-L 'give' andgoyba-wa-Y 'give-RECP' 541
Table 4-39. Mayga-L 'tell' and mayga-wa-Y 'tell-RECP' 541
Table 4-40. Birra-L 'tell' and birra-wa-Y 'tell-RECP' 542
Table 4-41. Transitive and reciprocal clauses involving a body part noun. 545
Table 4-42. Verbal locative construction and basic clauses 578
Table 4-43. Verbal instrumental construction and basic clauses 589
Table 4-44. Various types of the verbal instrumental construction. 590
Table 4-45. Possession cline 638
Table 4-46. Words combined with the enclitic =wa 'focus' 694
Figure 1-1. Pedigree (1): a male EGO. 29
Figure 1-2. Pedigree (2): a female EGO 30
Figure 4-1. Ergativity and accusativity 414
Figure 4-2. Intra-linguistic distribution of syntactic ergativity 445
List of maps and photos
Map 1 Australia. XXV
Map2 Herbert-Burdekin area (1): placenames XXVl
Map3 Warrongo territory xxvn
Map4 Mari subgroup. XXVlll
Map5 Herbert-Burdekin area (2): languages XXlX
Photo 1 Alf Palmer (July 1972) XXX
Photo 2 AlfPalmer and Tasaku Tsunoda (September 1974). XXX
List of abbreviations and symbols
The following abbreviations follow 'The Leipzig Glossing Rules' (www.eva.mpg.de/
lingua/pdf/LGR09 _ 02 _ 23.pdf; accessed on the 21 February 2011) wherever possible.
A
a
ABL
AC
ACC
Adj
Adv
AD NOM
AFFIMP
atum
ANTIP
AP
APPL
APPR
B
B&G
BI
b
B.T.L.
c
c
CAUS
CAUSE
CF
COM
D
d
DAT
DD
DH
DNK
DS
(a) male member of a section, (b) so-called transitive subject
female member of a section
ablative
Alec Collins
accusative
adjective
adverb
adnominal
affirmative imperative
animate
antipassive
(a) AlfPalmer
(b) word, sentence, etc. that was:
(b-i) given by Alf Palmer spontaneously (in a text, during elicitation,
or during our conversation), or given by him as a Warrongo translation
of an English sentence, or,
(b-ii) suggested by Tasaku Tsunoda and approved by Alf Palmer
applicative
apprehensional
male member of a section
Bidyara and Gungabula
Bidyara
female member of a section
Blue Tongue Lizard
(a) male member of a section, (b) consonant
female member of a section
causative
cause
counter-factual
comitative
(a) daughter, (b) male member of a section
female member of a section
dative
daughter's daughter
daughter's husband
'I do not know' (gloss for the enclitic =bajon (4.25-[2])
daughter's son
DU
d-S
EB
ERG
EZ
F
FB
FEM
FF
FM
FOC
FUT
FZ
GEN
GNG
GNY
H
HB
hum
IMP
man
INS
INTR
ITER
L
LINK
Lit.
LOC
M
MA
MB
MF
:MID
MM
MZ
N
NEG
NEG IMP
NF
NOM
NONINT
0
PTCP
p.c.
dual
derived S
elder brother
ergative
elder sister
father
father's brother
feminine
father's father
father's mother
focus
future
father's sister
genitive
Gungabula
Gunya
(a) husband, (b) high pitch
HarryBunn
human
imperative
inanimate
List of abbreviations and symbols xxm
instrumental
intransitive-stem-forming suffix
iterative
low pitch
linking interfix
literal translation
locative
(a) mother, (b) middle-level pitch, (c) man
Margany
mother's brother
mother's father
middle
mother's mother
mother's sister
nasal
negation
negative imperative
non future
nominative
non-interference
so-called transitive object
participle
personal communication
XXlV List of abbreviations and symbols
PI..
PR.OH
PST
pron
PURP
Q
RECP
REFL
s
SD
SG
ss
sp.
STAT
sw
TAGQ
TR
TT
TT,AP
v
Vi
Vt
w
W-GB-G
WM
YB
YZ
1
2
3
[ ]
[ TT]
?
plural
prohibition
past
pronoun
purposive
question
reciprocal
reflexive
(a) son, (b) so-called intransitive subject, (c) stop
son's daughter
singular
son's son
spec1es
stative
son's wife
tag question
transitive-stem -forming suffix
This indicates (a) words, sentences, etc. that were made up by Tasaku
Tsunoda or (b) information, words, etc. supplied by Tasaku Tsunoda.
This emphasizes that the item in question was suggested by Tasaku
Tsunoda and approved by AlfPalmer.
vowel
intransitive verb
transitive verb
(a) wife, (b) woman
Warrongo, Gugu-Badhun and Gujal
wife's mother
younger brother
younger sister
first person
second person
third person
This indicates (a) words that were supplied by Tasaku Tsunoda or (b)
so-called deletion.
This indicates words that were supplied by Tasaku Tsunoda.
This precedes an enclitic.
This precedes a suffix or indicates a morpheme boundary within a
word.
A question mark following a verb indicates that the conjugational
class membership of this verb is not known.
This indicates that one or more words have been deleted by Tasaku
Tsunoda from the original sentence, for the purpose of exposition.
e.g. 'S 0' indicates that 'the S corresponds to the 0'
This indicates (a) reduplication or (b) alternation.
1
2
3
4
5
6
7
8
9
10
WESTERN
AUSTRALIA
Guugu Yimidhirr
Kugu Nganhcara
Kuuk Thaayorre
Kayardild
Kalkatungu
Yalarnnga
Warluwarra
Pitta-Pitta
Bandjalang
Gumbaynggir
Map 1. Australia
21 18
20
19
17
NORTHERN
TERRITORY
16
5
6
7 8
Maps XXV
QUEENSLAND
See
Map2
15
SOUTH

14
11 Dharuk 19 Kukatja
12 Woiwurrung 20 Gooniyandi
13 Narrinyeri 21 Djaru (also spelt Jaru)
14 Ngiyambaa 22 Wunambal
15 Diyari 23 Warrwa
16 Pitjantjatjara/ 24 Gaagadju
Yankunytjara 25 Djapu
17 Warlbiri
(also spelt Warlpiri) The location of the languages
18 Wanyjirra is only approximate.
XXVI Maps
Einasleigh e

Chudleigh Spnngs
Parke e
Map 2. Herbert-Burdekin area (1): placenames
See
Map3
Burdeldn Falls


ltMt.Gamet
. lnnot
:" Springs

.

\

.
Kinrara
Valley of
Lagoons
.
.
.
.
.
.
.
.

.

.
.
.
JIRRBJ\L
Mt.Fox

Kangaroo
Hills

. . ...

Map 3. VVarrongo territory
Maps xxvii
xxviii Maps
I Jirrbal
2 Girramay
3 Warrgamay
4 Nyawaygi
5 Warrongo
6 Gugu-Badhun
.
.
.
9
QUEENSLAND
.
.
.
.
.
.
.
.
.
.
.
.
. .
: ro :
. .
. .
. .
: 13
11
12 :
: 14 :
. .

NEW SOUTH WALES
7 Gujal
8 Biri
9 Koa (also spelt Guwa)
10 Koamu
(also spelt Guwamu)
11 Bidyara
Map 4. Mari subgroup
The broken line indicates the
very approximate extent of the
Mari subgroup (based on Wurm
and Hattori (1981) and Dixon
(2002)).
12 Gungabula
13 Margany
14 Gunya
The location of the languages
is only approximate.
15 11
17
16
10 9
Ml. Gamctfl.o.Ravcnshoe
. .


.

0
.
.
18

.
.


.
19
20
.
.
.
.
.
.
.
.
.
.


.
.



.
.


.
.
.

.
.
.

.
.
.

.
.
1Wulguru
2 Gabilgaba
3 Buluguyban
4 Nyawaygi
5 Biyay
6 Wangamay
7 Jirru
8 Girram.ay
9 Mamu
10 Jirrbal
Charters Towers
11 Ngajan
12 Yidiny
13 Dja:bugay
(also spelt Djabugay)
14 Gugu Ya.J.argi
(also spelt Gugu-Ya.J.argi
and Kuku Yalanji)
15 Ngaygungu
16 Mbabaram
17 Wagaman
Map 5. Herbert-Burdekin area (2): languages
Maps XXIX
The broken line indicates the very
approximate boundary between the
Mari subgroup and other groups
(based on Wurm and Hattori (1981)
and Dixon (2002)).
3
D Palm Island
18 Agwamin
19 Wamin
20 Mbara
21 Warrongo
22 Gugu-Badlrun
23 Gujal
24 Biri
24
The location of the languages
is only approximate.
XXX Photos
Photo 1. Alf Palmer (July 1972, Palm Island, Queensland, Australia)
Photo 2. Alf Palmer (r.) and Tasaku Tsunoda (I.) (September 1974, Palm Island, Queensland,
Australia)
Chapter 1
The language and its speakers
1.1. Linguistic type
Warrongo is a fairly typical Australian language except for the existence of
syntactic ergativity, a phenomenon that is unique not only among Australian lan-
guages, but also among the world's languages, being mainly found in less than ten
languages of a small area in northeast Australia.
The following phonemes can be set up for Warrongo: /b, d, J, g, m, n, Jl, IJ, r,
J, 1, j, w, a, i, u, aa/, i.e. four stops (labial, apical, !aminaL dorsal; voicing is not
distinctive), four nasals each corresponding to a stop, two rhotics, one lateral, two
semivowels, three short vowels, and one long vowel. Stress and pitch are not dis-
tinctive.
The practical orthography employed for the present work is as follows: b, d j,
g, m, n, ny, ng, rr, r, L y, w, a, i, o, aa. The rhotic lr!, written with rr, is generally
an alveolar tap, while the other rhotic !JJ, written with r, is generally a retroflex
approximant. Note that the stop IJI is written withj, and the semivowel /j/ with
y. This practical orthography is identical with many other practical orthographies
used for Australian languages except that lui is written with o, and not u. But the
letter u, and not o, will be used when citing from other sources.
Warrongo is almost entirely suffixing, although there are elements which may
be considered interfixes (not as suffixes): linking interfixes (cf. comments on, e.g.,
Tables 3-3, 3-5 to 3-7) and epenthetic phonemes (2.6-[2]). Warrongo is agglutina-
tive, except that the singular pronouns of the first person and the second person
exhibit fusion.
Five word classes can be set up: (personal) pronoun, noun (including 'adjec-
tive-like' noun), adverb, verb, and interjection. As a ru1e each stem belongs to only
one word class, although there are a very small number of stems that may possibly
have dual class membership of noun and verb. There are two productive processes
for deriving verbs from nouns, etc., but there is no productive process for deriving
nouns from verbs. In addition to words, there are a fair number of enclitics, whose
function is largely modal.
Pronouns have three num hers (singu1ar, dual, plural) and three persons (first,
second, third), without an inclusive/exclusive distinction. Nouns in the main lack a
number distinction. Nouns and pronouns have eight cases, while some of adverbs
have a limited set of cases. Roughly speaking, nouns have the A=i=S=O pattern (the
ergative-absolutive pattern), and pronouns the A=S=i=O pattern (the nominative-
accusative pattern).
2 The language and its speakers
Verbs have three conjugational classes. One (L-class) is predominantly tran-
sitive, while the other two (Y-class and ZERO-class) are entirely intransitive.
Verbal inflectional suffixes have a temporal, aspectual and/or modal meaning.
Subordination is productively fanned by the pan-Australian purposive and also by
a few other conjugational categories.
There are four voice-related constructions: antipassive, reflexive/middle, re-
ciprocal, applicative, and one construction that is dominantly aspectual: iterative.
Syntactic ergativity is attained predominantly by means of the antipassive con-
struction.
Modal qualification of sentences can be achieved by enclitics and/or adverbs of
modality. Polar questions can be formed by the use of the question marker wayi or
simply just a rising intonation.
Word order is fairly free, but there are a small number of restrictions or ten-
dencies. For example, interrogative words and the question marker wayi tend to
occur sentence-initially. Also, word order is fixed in a few types of noun phrases
and probably in a few types of verb phrases, and Warrongo is configurational re-
garding these phrases, although it is in the main non-configurational. Warrongo
discourse is often highly elliptical.
1.2. Names of the language and people
In Aboriginal Australia the name of a language is often used to refer to the people
in question as well (cf. Dixon 1980: 41).
The name of the language under study is [ waroiJo] /warui]UI The rhotic is an
alveolar tap, and not an approximant. This name has been spelt in various ways,
e.g. 'War-oong-oo' (William Craig's letter dated 24th July 1898; see 1.8.2-[2]),
'Warungu' (Oates and Oates 1970, Tindale 1940, 1974, Tsunoda 1974b), 'WarUIJU'
(Dixon 1970, 1972, Tsunoda 1974a), ''WaruiJU' (Tindale 1940, 1974) (the apos-
trophe indicates stress (Tindale 1974: 2)), and 'Warrungu' (Tsunoda 1988b). The
present work employs 'Warrongo' (the reason for this choice is given in 2.14).
The meaning of the word warrongo is not known. Tindale (1974: 188) states
that ['waruiJu] means 'woman'. However, in this language, the word for woman
is [wariJo] /wa.riJU/, and not [waroiJo] lwatUI]U/. Girramay, immediately east of
Warrongo (see Map 5), has the noun wammgu 'evening star' (Dixon 1982: 1).
Yidiny, about 80 km northeast of Warrongo, has the compound verb warrungu-
gifju-L 'dream about'; the verb gifju-L means 'bite' (Dixon 1991a: 286, 299). But
the meaning of warnmgu is not given. It would be expected to mean 'dream' (a
noun). It is not known if these two words (warnmgu) are related to the name
Warrongo.
Names other than [wacoiJo] and its various spellings have been given. Oates
and Oates (1970: 175) list Bara as an alternative to Wanmgu. According to Sutton
(1973: 9), there was a language called Mbara, southwest of Gugu-Badhun and west
Dialects 3
of Gujal, around Gregory Springs and Chudleigh Park Stations (see Map 2). Bara
is possibly a mishearing for Mbara. The initial/m/ of Mbara would have been dif-
ficult to detect; see 1.4.2.2.
Tindale (1974: 188) lists Warungu, 'WaruJ]u, and Warrialgona. But he com-
ments: 'The suggested Warrialgona equation is doubtful'. The meaning of
Warrialgona is not known- if such a word exists at all. See 1.8.2-[1] for more on
the word Warrialgona.
Alec C.ollins (cf. 1.8.4.2) referred to his language as Warrongo and also as
Gobobarra. The noun gobo means 'lear, and the suffix -barra 'denizen of, be-
longing to, associated with' (cf. 3.7.1-[1]). Therefore, literally gobobarra means
'person(s) of a place with [lots of] leaves'. Tommy Murray, a Jirrbal speaker (cf.
1.8.4.3-[7]), independently stated that there were a group of people and a language
called Gobobarra around Mount Garnet (see Map 3). But Alec Collins seems to
have come from Herbert Gorge area (cf. 1. 8.4.2), and not from :Mt. Garnet. It is pos-
sible that Goboba"a refers to the Warrongo language and people of an area that
includes :Mt. Gamet and Herbert Gorge. However, the vegetation of Mt. Garnet is
an open forest, while that of Herbert Gorge is a dense rainforest (cf. 1.5.1). In view
of this, the name Goboba"a seems appropriate for the language and the people of
Herbert Gorge, rather than of :Mt. Garnet.
Alf Palmer (cf. 1.8.4.1) referred to Warrongo people as Warrongo and also as
Gonarribarra 'basalt person(s)' (gonarri 'basalt'). In addition, Alf Palmer gave
the following words: (i) barribiri 'lava', 'lava country', 'Valley of Lagoons' (see
Maps 2 and 3), 'person of/from lava country', e.g. a Warrongo person, a Gugu-
Badhun person; and (ii) barribiriba"a 'person of/from Valley of Lagoons'. There
was volcanic activity in the region, which produced much lava and basalt (cf. 1.5.1).
1.3. Dialects
Warrongo probably had dialects. However, they were impossible to investigate
already in the early 1970s, when I conducted fieldwork on the language. There
appear to have been at least two dialects. One is presented by the data obtained
from Alf Palmer, a central dialect (?) (1.8.4.1), and the other by the data obtained
form Alec Collins, an eastern dialect(?) (1.8.4.2).
The data from Alf Palmer are extensive, but the data from Alec Collins are se-
verely limited. Nonetheless, it is clear that these two dialects were mutually intelli-
gible perfectly. They are identical regarding their phoneme inventory, and presum-
ably phonotactics as well. Morphologically Alec Collins' Warrongo in the main
does not differ from Alf Palmer's, although it exhibits a small number of differ-
ences (see 3.2.1.1 and 4.9.1). Regarding syntax, again Alec Collins' Warrongo does
not seem to differ from Alf Palmer's. As for basic vocabulary, the two dialects
share 90%. (This figure is in terms ofK. L. Hale's core vocabulary for Australian
languages, cited in Sutton and Walsh (1979: 38), which comprises 99 items. Since
4 The language and its speakers
Alec Collins' data are limited, only 80 items were available for comparison.) The
shared vocabulary consists of cognates- both identical and non-identical ones. For
example, for 'mother's brother', AlfPalrner used /galr.Janal (and /gab:.Jal) only, but
Alec Collins used /galr.Jinru (the second vowel is /if) (see 1.8.4.2), in addition to
/galr.Janal (and /galr.Ja/) (see 3.2.1.2 and (4-257)). (The suffi..x -na is used with vowel-
final kin nouns. See 3.7.1-[16].)
1.4. Territory and neighbouring languages
1.4.1. Territory
The exact eA'tent of Warrongo territory is not known. Tindale (1940: 174) gives
the following as Warrongo territory 'Head-waters of Burdekin River, south prob-
ably to about Clarke River; west to Dividing Range; east to inland foot of Coast
Range' (See Maps 2 and 3.) Tindale (1974: 188) gives 'Headwaters of Burdekin
River, southeast to near Charters Towers and southwest along the Clarke River;
west to the Dividing Range; east to the inland foot of the Coastal Range and to the
big southern loop of the Burdekin'. Oates and Oates (1970: 175) give 'Headwaters
of Burdekin River, south to Clarke River, to Dividing Range, to foot of Coastal
Range'
Dixon (1970: 662) states:- 'WaruiJu was spoken over a long tract on top of the
range, in contiguity with Wargamay, Giramay and Dyirbal and also, to a lesser
extent, with Nyawigi [sic], Mbabaram and probably Wagaman' (cf. also Dixon
1972: 26), and Warrongo territory eA'tends to 'just north of Mount Gamet' (Dixon
1991b: 350).
Sutton (1973: 14) states that Warrongo territory includes an area from Mount
Gamet southeast along the Herbert River and it has its border with Gugu-Badhun
at Meadowbank, Glenharding and Wairuna Stations.
R. S. Atkinson- a member of the Atkinson family who own many cattle sta-
tions in the region - stated in his letter to me dated 12th September 1974 that
there was a man called Wambino who was, according to R. S. Atkinson, the king
of the tribe. They used to travel as far as Wairuna, Abergowrie, Tully River and
Ravenshoe. Perhaps that was, R. S. Atkinson suggested, the rough outline of the
tribal territory. (Wambino is Alec Collins' father; cf. 1.8.4.2. The etymology of the
name Wambino is given in 1.5.4.2-[1] and 3.7.1-[12].)
The language northeast of Warrongo is Jirrbal. According to Dixon (1970: 658),
the latter's territory includes Ravenshoe.
South of Jirrbal and east of Warrongo is Girramay. There was a cattle station
called Kirrama Station. According to Peter Sutton (p.c.), that station was named
after Girramay. (Like most of the Australian languages (cf. 1.1), Girramay has no
phonemic contrast between /g/ and lk/ and, and no doubt the initial consonant of
Tenitory and neighbouring languages 5
this word was perceived as [k] by the person who named that station.) According
to Dixon ([1983] 1989a: 78), Kirrama Station is in the Girramay country. In March
2001, Leigh Pentecost (p.c.), an archaeologist, stated that Kirrama Station is on
the border among Jirrba1, Girramay and Warrongo. This view is shared by Rachel
Cummins, Alf Palmer's daughter's daughter (e-mail message of 25 July 2006).
(Jirrbal, Girramay, Mamu, etc., are collectively called Dyirbal by Dixon (1972).)
South of Girramay and east of Warrongo is Warrgamay. According to Dixon
(1981: 0, 2), its territory extends from just west of Ingham up along the Herbert
River, past Abergowrie, and beyond Niagara Vale and Yamanie Creek, and it in-
cludes Wallman Falls and Peacock Siding. However, according to Alf Palmer,
Warrongo territory extends close to Abergowrie.
South of Warrgamay and east of Warrongo is Nyawaygi. According to Dixon
(1983: 432), the Stone River is in Warrongo territory.
West of Nyawaygi and south of Warrongo is Gugu-Badhun. According
to Sutton (1973: 14), Gugu-Badhun territory reaches as far as Meadowbank,
Glenharding and Wairuna Stations in the north (as noted above), and Clarke River
in the south, about where it joins the Burdekin River.
South of Gugu-Badhun is Gujal (also spelt Gudjal; also known as Gujala and
Gurjal). According to Sutton (1973: 14), its approximate southern boundary is
Charters Towers.
According to Sutton (1973: 9), west of Jirrbal and northwest of Warrongo is
Mbabaram. South of Mbabaram and west of Warrongo is Agwam in. South of
Agwamin and west of Gugu-Badhun is Wamin. Mbara (mentioned in 1.2) is south
ofWamin and west of Gujal.
According to Dixon (1991b: 348, 350), west of Jirrbal and north of Warrongo
is Ngaygungu, its territory including Atherton but excluding Mount Garnet.
West of Ngaygungu and north of Warrongo is Mbabaram, its territory including
Irvinebank. West of Mbabaram and northwest of Warrongo is Wagaman, its ter-
ritory including Almaden and upper reaches of Tate River. South of Wagaman
and west ofWarrongo is Agwamin, its territory including Lynnbrook [sic], Mount
Surprise, and Einasleigh. (The road maps I have show the place name 'Lyndbrook',
not 'Lynnbrook'.)
Rachel Cummins stated (e-mail message of 3 April 2007) that it was agreed
upon by Warrongo people and Mbabaram people that the border between their ter-
ritories is south of Mount Garnet.
Taking all this into account, the approximate border of Warrongo territory
seems to be (i) Mount Garnet, Innot Hot Springs, and close to - but excluding -
Ravenshoe on the north, (ii) close to Kirrama Homestead, down the Herbert
River Gorge close to .Abergowrie, and the Stone River on the east, (iii) Wairuna,
Glenharding, and Meadowbank Stations on the south, and (iv) the Great Dividing
Range on the west. Warrongo territory seems to have an oblong shape; see Map
3. The information available does not say whether Warrongo territory includes
Kangaroo Hills.
6 The language and its speakers
In view of the above, the location ofWarrongo in Oates and Oates (1970: 180),
Tindale's (1974) map, and Map 23 of Wurm and Hattori (1981) appears to be in-
correct. Warrongo is located in Gugu-Badhun territory (and Gugu-Badhun in
Warrongo territory).
As mentioned at the beginning of 1.4.1, the exact extent of Warrongo territory
is not known. The preceding account ofWarrongo territory is highly tentative, and
it is by no means definitive.
1.4.2. Neighbouring languages and their classification
1.4.2.1. Proposed classifications
As examples of classification of the languages mentioned above and those men-
tioned below, two of the more recent classifications are cited: Wurm and Hattori
(1981) in Table 1-1, and Dixon (2002: xxxii-xxxiii) in Table 1-2. (The languages
that are not mentioned in the present work are not listed in these tables. The spell-
ings of the language names are left intact.)
The label 'Pam an' is taken from the word pama 'man, person'. (This word is
written bama in the practical orthography employed in the present work; recall the
absence of voicing opposition (1.1).) This word is common in the languages to the
north ofWarrongo, e.g. Yidiny (Dixon 1977b: 547), Kuku Yalanji (Oates and Oates
1964: 81), Kuuk Thaayorre (or Thaayorre) (Map 1) (Foote and Hall 1992: 101). It
also occurs in Warrgamay (Dixon 1981: 124), east ofWarrongo. Similarly, the label
'Marie' is based on the word mari 'man'. This word (or its cognate) is common in
the languages to the south ofWarrongo, e.g. Biri (my data), Bidyara and Gungabula
(Breen 1973: 206) andMargany and Gunya (Breen 1981: 362) (Map 4.)
1.4.2.2. Studies on the neighbouring languages
Available descriptions of the neighbouring languages were mentioned in 1.4.1. For
Gujal, south of Gugu-Badhun, Sutton's unpublished data (see 1.8.4.3-[9]) are avail-
able. For the lexical comparison given in the second paragraph below, my Jirrbal
data recorded from Tommy Springcart and Tommy Murray in 1974 (see 1.8.4.3-
[7]) are included.
Ngaygugu (north of Warrongo) and the languages to the northwest and west
of Warrongo- i.e. Wagaman, Agwamin, Wamin, and Mbara- are poorly docu-
mented. Except for Ngaygugu, they and Mbabaram underwent a series of drastic
phonological changes (Dixon 1991b, Sutton 1973: 60-67). One of the results is
the existence of word-initial consonant clusters, e.g. the names of the languages:
Mbabararn and Mbara. Such clusters are unusual among the languages of the
region. (Alf Palmer knew the name Mbabararn, but he pronounced it [ba:ba.y_arn.].
He dropped the initial nasal lrn/ and replaced the short vowel [ba] with the long
Tenitory and neighbouring languages 7
vowel [ba:]. This vowel lengthening is an instance of compensatory lengthening;
see 2.6-[4].) As mentioned in 1.2, the putative language name 'Bara' may be a mis-
hearing for ':tvfbara'. For Warnin, I recorded twelve words from Freddy Fulford
in 1974, but I am unable to analyze the data. In his pronunciation, the name of
his language is [wamen]. (Alf Palmer knew this name, and he, too, pronounced it
[wamen]. This word occurs in the example (4-348).) Almost certainly these lan-
guages are mutually unintelligible with Warrongo.
Table 1-1. Wurrn and Hattori's (1981) classification
Pama-Nyw1gan family:
Yalandyic group: Kuku Yalanji
Yidinyic group: Yidiny
Parnan group:
Western Parna subgroup: Thaayorre
Southern Pama subgroup: Agwarnin, Mbabararn, Ngaygungu, Warnin, Mbara
Dyirbalic group:
Dyirbal: Dyirrbal, Girramay
Warrgarnay
Nyawaygic group: Nyawaygi, Wulguru
Marie group:
Mari subgroup: Warungu, Gugu Badhun, Gudjala, Biri, Bidyara, Gungabula,
Margany, Gunya
Table 1-2. Dixon's (2002) classification
E Western Cape York Peninsular areal group: Kuuk Thaayorre
F Kukll-Yalanji: Kukll-Yalanji, (Kukll-)Wakarnan
G Cairns subgroup: Yidiny
H Herbert River Group:
Hl Dyirbal: Jirrbal, Girrarnay
H2 Warrgarnay
H3 Nyawaygi
H4 Manbarra: Buluguyban, Wulgmukaba
J Greater Marie Group:
Ja Marie Proper subgroup:
Jal Bidjara [i.e. Bidyara- TT], Gungabula, Marganj [i.e. Margany- TT],
Gunja [i.e. Gunya - TT]
Ja2 Biri
Ja3 Warungu, Gugu-Badhun, Gudjal(a)
Ja4 Ngaygungu
Jb Mbabararn/Agwami.n group:
Jbl Mbabararn
Jb2 Agwarnin (or Wami.n)
8 The language and its speakers
The languages of Dixon's Herbert River Group are fairly different from
Warrongo. For example, in terms of K. L. Hale's 99-item core vocabulary for
Australian languages (cf. 1.3), Warrongo shares 46% with Jirrbal, 39% with
Girramay, 42% with Warrgamay, and 27% with Nyawaygi. (In contrast, it shares
around 90% with Gugu-Badhun and Gujal; see Table 1-3.)
No doubt Warrongo and the languages of Herbert River Group are mutually
unintelligible. Alf Palmer stated that Jirrbal is 'hard'. Reciprocally, Dixon (1970:
662; cf. also Dixon 1972: 26) states: 'speakers of Dyirbal [i.e. Jirrbal, Girramay,
etc. - TT] always refer to Warui]U as a very "difficult" language, far harder for
them to speak and understand than Wargamay ...
Despite this, among the speakers interviewed for Warrongo, bilingualism in-
volving Warrongo and one language (or two) of Herbert River Group (e.g. Jirrbal)
was common. See 1.8.4.1 to 1.8.4.3.
1.4.2.3. WarTongo, Gugu-Badhun and Gujal as a linguistic unity
Hereafter the names of these three languages will be often abbreviated as 'W-GB-
G' (They are referred to as 'Herbert-Burdekin Languages' in Tsunoda (1984).)
The classifications cited in Tables 1-1 and 1-2- and virtually all other classifica-
tions, e.g. O'Grady, Voegelin and Voegelin (1966: 51-52), Oates and Oates (1970:
169-177), Wurm (1972: 141)- considerW-GB-G as the most northerly members of
a large group of closely related languages that 'stretched from the southern edge
of the Cairns rain forest region, north Queensland, to the border of New South
Wales' (Sutton 1973: 4, cf. also Breen 1981: 275), whose most southerly members
are Margany, Gunya, Bidyara and Gungabula. (See Map 4.) They are Wurm and
Hattori's Mari subgroup, and Dixon's Marie proper subgroup. Hereafter, this group
will be referred to as Mari subgroup.
In contrast with Warrongo, the data on Gugu-Badhun are limited, and even
more so on Gujal. Nonetheless, it is clear that these three languages are very sim-
ilar to each other, and no doubt they are mutually intelligible.
Table 1-3. Lexical sharing among Warrongo, Gugu-Badhun and Gujal
Warrongo
90 Gugu-Badlum
94 88 Gujal
Lexically, W-GB-G share around 90% of basic vocabulary (in terms ofK. L. Hale's
99-item vocabulary); see Table 1-3. (For Warrongo, both the data recorded by me
from Alf Palmer, and those recorded by Peter Sutton from Alec Collins are used.)
Morphologically, too, they are very similar. The syntactic information on
Gugu-Badhun and that on Gujal are limited, and it is difficult to make any signifi-
cant comparison.
Tenitory and neighbouring languages 9
In phonology, W-GB-G show a small difference. Gujal has the lamina! con-
trast in each of stops and nasals (i.e. kJI versus IJI; frJ/ versus /p./) (Sutton 1973:
34). Warrongo has no lamina! contrast, either in stops or nasals. Gugu-Badhun,
which is geographically between Gujal and Warrongo, has the !aminal contrast in
stops only, and lacks it in nasals (Sutton 1973: 34). See Table 1-4. See (i-i) and
(i-ii) of Table 1-6 for examples. (Dixon (1980: 141) shows the distribution of the
!aminal contrasts on the continent. The border between the languages with !aminal
contrast(s) and those without runs between Gugu-Badhun and Warrongo.)
Table 1-4. Oppositions in lamina! consonants
Warrongo
Gugu-Badlnm
Gujal
stops nasals
!pi
/pi
l:rJI!pl
W-GB-G are very similar to each other, so that they can be regarded as 'di-
alects' of the same 'language' Their similarities were noticed not only by their
speakers but also by speakers of other languages, for example, by Tommy Murray
(a Jirrbal speaker; see 1.8.4.3-[7]).
Sutton (1973: 14) notes that the peoples of W-GB-G 'formed something of a
unity', and he cites Eric Gertz, a Gugu-Badhun speaker (cf. 1.8.4.3-[8]), who said:
'Our mob used to run all the way from the [Atherton- TT] Tableland to Charters
Towers' The southern end of Atherton Tableland is around Mt Garnet (http://
www.athertontableland.com/, accessed on 30 March 2007). Therefore, the area
mentioned by Eric Gertz (i.e. from :tvft. Garnet to Charters Towers) coincides with
the territories ofW-GB-G.
W-GB-G seem to have been sometimes collectively referred to by one single
label, as follows.
(i) In 1971, when Alf Palmer started working with me, he called his language
[ko.{i1fal] (Gurijal), but in 1972 and 1974 he called it Warrongo (cf. 1.8.4.1).
(ii) Harry Bunn called his language Gujal (in 1974), despite the fact that his
language appears to be Warrongo (see 1.8.4.3-[1]). He stated that the outline of
Gujal territory is Mt. Garnet, Cashmere Station, Herbert Gorge (excluding Glen
Ann and Kirrama Stations), Wairuna, :tvft. Fox, Blue Range and Hill Grove Stations
(excluding Charters Towers) and Mount Surprise. He did not know if Innot Hot
Springs was included in the territory. (See Maps 2 and 3.) This area roughly coin-
cides with the W-GB-G territories.
(iii) Pompy Clurnppoint (a Jirru speaker from the coast) stated (in 1974) that
Gurjal was spoken from Mt Garnet to Charters Towers. This area coincides with
the W-GB-G territories.
The remarks cited above suggest that W-GB-G were collectively referred to by
one single label- Gurijal or something similar.
10 The language and its speakers
1.4.2.4. Problems with the proposed classifications
As noted above, Wurm and Hattori (1981), Dixon (2002), and also virtually all
other classifications assign W-GB-G to the Mari subgroup. However, this classi-
fication is not free from problems. The problems are discussed in Tsunoda (1984),
and the main points are summarized below. We shall look at the following groups
of languages.
(a) Dixon's Herbert River Group:
Jirrbal, Girramay, Warrgamay, and Nyawaygi.
(b) Mari subgroup:
(b-1) Most northerly members:
Warrongo, Gugu-Badhun, and Gujal.
(b-2) Biri.
(b-3) Most southerly members:
Bidyara and Gungabula; and Gunya and Margany.
The sources of information are as follows. (i) Jirrbal: Di.xon (1972), and my data
recorded from Tommy Springcart and Tommy Murray, (ii) Girramay: Dixon
(1972), (iii) Warrgamay: Dixon (1981), (iv) Nyawaygi: Dixon (1983), (v) Warrongo:
my data recorded from Alf Palmer, (vi) Gugu-Badhun: Sutton (1973), (vii) Gujal:
Peter Sutton's data recorded from Freddy Toomba and Ranji Pope, (viii) Biri: my
data recorded from Harry Johnson, Eddy Barker, and Reggie Dodd, (ix) Bidyara
and Gungabula: Breen (1973); and (x) Gunya and Margany: Breen (1981). For
Warrongo, only the data from Alf Palmer (and not from Alec Collins) are cited.
See 1.8.4.1-[1] for the reason.
The languages of (a) are W-GB-G's northern and eastern neighbours. (As noted
in 1.4.2.2, the languages to the northwest and the west of Warrongo are poorly
documented, and consequently difficult to assess.) The languages of (b-3) are
more than 1,000 km away from Warrongo. The languages between (b-1) and (b-3)
are poorly documented, and there is no published grammar available, except for
Terrill (1998) on (b-2) Biri. (Terrill (1998) incorporates my data on Biri.)
A comparison of these languages shows that there are contradicting pieces of
evidence regarding the genetic classification ofW-GB-G.
There are three pieces of evidence that support the classification ofW-GB-G as
Mari languages, and not as those of Herbert River Group.
Evidence 1: Pronouns
The pronouns of (b) differ from those of (a). The differences are most clear in '1SG',
'2SG', '2DU', and '2PL'. Their NOM/ERG forms are cited in Table 1-5, with the
Warrongo forms as the representative of (b), and the Jirrbal forms as the represen-
tative of (a). There are slight variations among the languages of each of (a) and (b).
Tenitory and neighbouring languages 11
In Table 1-5, the two items in each pair are cognates. Note the three correspon-
dences: (i) the palatal stop IJI and the semivowel /jl in 'lSG', (ii) the velar nasal /rjl
and the semivowel /j/ in '2SG', and (iii) the nasal /pi and the semivowel /j/ in '2DU'
and '2PL'. (The Jirrbal forms are conservative, and the Warrongo forms are inno-
vative.)
Table 1-5. Selected pronouns ofWarrongo and Jirrbal
'lSG'
'2SG'
'2DU'
'2PL'
Jirrbal
ngaja (IIJJ.jal)
nginda (l:ginda/)
nyubalaji (/pubalaji/)
nyurraji (lpuraJi/)
Evidence 2: Vero roots
Warrongo
ngaya (/I]aja/)
yinda (ljinda/)
yobala (/jubala/)
yorra (/jural)
In terms of the 22 verbs from K. L. Hale's 99-item list, W-GB-G share a higher
percentage with (b-2) Biri and (b-3) than with (a). This is despite the fact that (a)
is immediately contiguous with (b-1), in contrast with Biri, which is about 500 .km
away from (b-1), and with (b-3), which is about 1,000 km away from (b-1). Thus,
Warrongo shares:
(a) 36% with Jirrbal, 38% with Girramay, 41% with Warrgamay, 52% with
Nyawaygi;
(b-1) 95% with Gugu-Badhun, 94% with Gujal;
(b-2) 52% with Biri;
(b-3) 55% with Bidyara and Gungabula, 55% with Gunya, and 45% with Margany.
For examples, see the verb roots in Table 1-6: 'to bite', 'to cook', 'to see', and 'to
enter'
Evidence 3: Regular phonological correspondences
See Table 1-6. (The abbreviations of the names of the languages are as follows:
BI - Bidyara, GI- Girramay, GNG - Gungabula, GNY - Gunya, JI - Jirrbal, MA-
Margany, NY - Nyawaygi.)
(b-1) shares at least three sets of regular phonological correspondences with
(b-3). The three sets involve (i) a lamina!, (ii) a retroflex, and (iii) a stop-plus-stop
sequence.
(b-1) shares at least one set of regular phonological correspondence with (b-2)
Biri. This involves (i) a lamina!. (The words with an asterisk - /jwi/* 'meat' and
/barbi.Ia/* 'echidna' -are identical to those of Warrongo. The word with two aster-
isks- /badbi.Ia/** 'echidna' -is an exception to the correspondence in question.)
For each set, selected examples are given.
12 The language and its
(b-1) does share correspondences with (a) (see (vi) below, Table 1.5, 2.4-[2] to
-[6], and 3.11.1.3), but these correspondences are neither so systematic nor so reg-
ular as those between (b-1) and (b-2)-plus-(b-3).
Table 1-6. Phonological correspondences
(a) (b-1)
Warrongo Gugu-Badlnm Gujal
(i) laminal
(i-i) ljl-lg!
'head'
'to bite'
'to cook'
(i-ii) ITJI-Ip!
lbaja/
(JI)
/gaQa/
lbaQa/
/waQ.ul
'to see' /pa:J (NY) /paga/ /paga/
(ii) retroflex
l:rl-I<V_
'kangaroo' /jurl/
(JI, GI)
(iii) rhotic+stop - stop+stop
/jwil /jwi/
'echidna' lbarbi.Ia/ lbarbira/
'to enter'
(i) lamina!
(i-i) ljl-lg!
'head'
'to bite'
'to cook'
(i-ii) ITJI-Ip!
(b-3)
/gaQa/ (MA)
lbaQa/
/ja.xga/
(BI, GNG, GNY, MA)
/waQ.ul
(BI, GNG, GNY, MA)
'to see' /uaga/ (GNY)
(ii) retroflex
lli-I<V_
'kangaroo'
/juc:l.il
'meat, animal'
(BI, GNG, GNY, MA)
(iii) rhotic+stop- stop+stop
'echidna' lbadbiqa/
(BI, GNG, GNY, MA)
'to enter' (GNY, MA)
/gaQa/
lbaQal
/waqu/
/uaga/
/jwi/
'meat'
/barbi.Ia/
(b-2) Biri
/gaQa/
lbaQal
/waqu/
/IJagal
/jurl/*
'meat'
/barb:i.Ia/*,
lbadbila/* *
Territory and neighbouring languages 13
The correspondences in (i) involve a laminal consonant. In (i-i), kJI of (b-2) and
(b-3) corresponds to kJI of Gujal and Gugu-Badhun, and to ljl of Warrongo. In (i-
ii), /J)I of (b-2) and (b-3) corresponds to IQ/ of Gujal, and to !pi of Gugu-Badhun
and Warrongo. In (ii), the retroflex stop of (b-3) corresponds to the retroflex ap-
proximant of (b-1). (Here, (b-2) Biri shares /jmil'kangaroo'/'meat' with (b-1), and
not with (b-3). That is, in this respect, (b-1) and (b-2) should be grouped together.)
In (iii) a stop-plus-stop sequence of (b-3) corresponds to a rhotic-plus-stop se-
quence of (b-1). (Here again, (b-2) Biri shares /barbiia/ 'echidna' with (b-1), and
not with (b-3). In this respect, too, (b-1) and (b-2) should be grouped together. The
form lbadbiia/ is an exception to this correspondence.)
(b-1) and (b-3) share these regular phonological correspondences, although
they are about 1,000 km away from each other. (b-1) and (b-2) exhibit regular
phonological correspondences in terms of (i). They share /jmil 'kangaroo'/'meat'
and /barbiia/ 'echidna'. In contrast, as shown in Table 1-6, (a) often lacks a cor-
responding cognate where (b-1) and (3)- and also (b-2) in terms of (i)-exhibit a
correspondence. This is despite the fact that (a) is contiguous with (b-1).
Thus far, we have seen three pieces of evidence that indicate that (b-1) (W-
GB-G) should be grouped with (b-2) (Biri) and (b-3) (southernMari languages).
Now, there is one piece of evidence that suggests that (b-1) should be assigned
to (a) (Herbert River Group), and not to Mari subgroup.
Evidence 4: Verbal inflectional morphology
(b-1) shares at least six verbal inflectional suffixes with (a), to the exclusion of
(b-2) and (b-3). They are listed below. In contrast, (b-1) shares no such suffix with
(b-2) and/or (b-3), to the exclusion of (a). This strongly indicates that (b-1) (W-
GB-G) should be assigned to (a) (Herbert River Group), and not to (b) (Mari sub-
group).
(i) (a) -ya 'imperative' in Warrgamay. (b-1) -ya 'imperative' in W-GB-G.
(ii) (a) -y 'unmarked aspect' in Warrgamay. (b-1) -y 'past/present' in Warrongo,
-y 'present/future' in Gugu-Badhun, -y 'past/present(?)' in Gujal.
(iii) (a) -n 'past/present' in Jirrbal and Girramay. (b-1) -n 'past/present' in
Warrongo, -n 'past' in Gugu-Badhun and Gujal.
(iv) (a) -nyu 'past/present' in Jirrbal and Girramay, -nyu 'perfect' and 'subordi-
nate' in Warrgamay, -nya 'unmarked' ofNyawaygi. (Dixon (1983: 476) sug-
gests that the Nyawaygi -nya derives from -nyu.) (b-1) -nyu (also written
-nyo) 'subordinate' in Warrongo and Gujal.
(v) (a) -ngu 'subordinate' in Jirrbal and Girramay. (b-1) -ngu (also written -ngo)
'subordinate' in Warrongo and Gujal.
(vi) (a) -1-jay and -n-jay 'future' in Girramay. (b-1) -yay 'future' in Warrongo and
Gugu-Badhun. (Note that these suffixes exhibit an alternation (j- y) that
parallels the alternation between ngaja '1SG' of (a) and ngaya '1SG' of (b);
see Table 1-5.)
14 The language and its speakers
To sum up, the evidence regarding pronouns, verb roots and regular phonolog-
ical correspondences supports the classification of W-GB-G as Mari languages.
However, the evidence regarding the verbal inflectional morphology indicates that
they should be assigned to Herbert River Group.
One possible scenario to account for this puzzle is as follows. W-GB-G were
(and still are?) Mari languages, but they heavily borrowed verbal inflectional
suffixes from Herbert River Group. (This is despite the claim that verbal inflec-
tional morphology is very resistant to borrowing; cf. Sapir [1921] 1949: 201-206,
Weinreich [1953] 1974: 32, Bynon 1977: 189, 253, Heath 1978: 68-71.) That is,
(all?) Warrongo verbs have retained the Mari root but use a Herbert-River-Group
inflectional suffix rather like the Copper Island dialect of Aleut (Comrie 198lb:
253), which has retained native verb roots but uses Russian inflectional suffixes.
One possible cause for this massive borrowing may be the fact that Warrongo,
Jirrbal and Girramay groups had frequent contacts with each other in the pre-con-
tact times (see 1. 7) and that they lived together at Kirrama Station in the post-
contact times (see 1.7, 1.8.4.1-[1] and 1.8.4.2).
It is relevant to note in this connection that the word for 'man' in W-GB-G is
bama, and not marl. This is despite the fact that they are generally classified as
Mari languages. That is, the distribution of the word marl 'man' and the territory
of Mari languages do not coextend. (I owe this observation to the anonymous re-
viewer.) At least AlfPalmer, and almost certainly all the other W-GB-G speakers,
knew the word marl. As noted in 1.4.2.1, the word bama 'man' is common in the
languages to the north ofWarrongo.
1.5. Environmental and socio-cultural background
1.5.1. Environmental setting
It seems that Warrongo territory can be roughly divided into two areas. (i) The
eastern area: 'The precipitous eastern [area is TT] characterized by water-
falls, deep gorges and valleys, the most notable of which is the Herbert Gorge'
(Brayshaw 1990: 2), and it is covered by a dense rainforest. (ii) The western area is
flatter, and it is drier, consisting of open forests.
There was volcanic activity in the region, and its remnants still remain, such as
lava, basalt (e.g. Great Basalt Wall), volcanic craters (e.g. Herberton Crater), and
hot springs (e.g. Innot Hot Springs). (See Maps 2 and 3.) There are Dream Time
stories associated with the volcanic activity; see 1.5.3. Also there are words based
on the word for lava or basalt that refers to Warrongo and Gugu-Badhun people
(1.2).
Environmental and socio-cultural background 15
1.5.2. Archaeological and anthropological accounts
Archaeological evidence indicates that humans reached this continent more than
53,000 years ago (Flood [1983] 1995: 32). They may have arrived at Atherton
Tableland about 38,000 years ago (Flood 1995: 95).
There appears to be no archaeological or anthropological work that exclusively
deals with Warrongo people. But there are works that refer to them. For example,
Brayshaw (1990: 38), in a detailed survey of archaeological and anthropological
works of the Herbert-Burdekin River area, refers to Warrongo people: 'Warungu
of the Herbert River had much in common culturally with the rainforest peoples
[e.g. Girrarnay and Warrgarnay - TT], although linguistically they had more in
common with the Burdekin group of languages [i.e. Gugu-Badhun and Gujal- TT].
In 1.5.3 and the subsequent sections, we shall look at some selected topics re-
garding the socio-cultural background of these peoples.
1.5.3. Mythology
The mythology explains the origin and formation of the universe that surrounds
the people. Thus, a story narrated (in English, in 1974) by Reggie Palm Island, a
speaker of Buluguyban of Palm Island (Map 3), tells that the islands off the eastern
coast were once connected to the mainland. This story, repeated from Tsunoda
(1996), is roughly as follows. (See Map 2.)
Long ago this area was dry. From somewhere north, a carpet snake carne, through
Hinchinbrook Channel [between Hinchinbrook Island and the main land - TT], to
Palm Island and arrived at what is now called Cannon Bay. It went up the hill and
carne to Bamboo Creek. There was a big pool, where some children were swim-
ming. The carpet snake swallowed up one of them. It travelled over the hills, carne
down to near Butler Bay (?), and then went over to Magnetic Island. When the
child's parents carne back from lnmting, they learned what had happened. They
walked over to Magnetic Island, found the carpet snake, and cut it open. There
inside the snake they found their child [alive?- TT].
The word for 'carpet snake' is gabul in the languages of the region, including
Warrongo. Phonetically it is [gabol] in Reggie Palm Island's pronunciation. (It
is [kabol] in Alf Palmer's. It is spelt gabol in the orthography employed in the
present work.) According to Rachel Cummins (e-mail message of 6 February
2009), the child in question is a young girl.
The pool at Bamboo Creek existed in 1974, when Reggie Palm Island nar-
rated this story. However, in 1976 or 1977 (Rachel Cummins, e-mail of 6 February
2009), the Queensland Government built a darn there, and completely destroyed
this extremely important site -despite the local Aboriginal people's vehement pro-
test (Unfortunately, this is not an isolated incident It is just one of numerous de-
structions of Aboriginal sites.)
16 The language and its speakers
In the story narrated by Reggie Palm Island, Carpet Snake's travel ends on
Magnetic Island. However, in the myth of people south of Palm Island, Carpet
Snake continues its travel beyond Magnetic Island. Thus, (in March 2001) Shirley
Johnson, of the Wulgurugaba group of the Townsville-Magnetic Island area, nar-
rated roughly as follows.
The track of Carpet Snake goes from the bottom [i.e. mouth?- TT] of Herbert River
to Hinchinbrook [Channel - TT], past Palm Island. past Magnetic Island. up Ross
River, up Hervey Range, where it rests. There is a rock painting of [gabol] [on
Hervey Range?-TT].
(In Shirley Johnson's pronunciation the word for 'carpet snake' is [gabol].) See
Map 2 for Ross River and Hervey Range.
There is a word based on gabul (or gabol) 'carpet snake': gabulbarra 'carpet
snake person(s), person(s) associated with carpet snake' (For the suffix -barra
'denizen of, associated with' (mentioned in 1.2), see 3.7.1-[1]. The rhotic is a tap,
and not an approximant.) This word seems to refer to the (original) people of
Townsville-Magnetic Island area. The Aboriginal reference group of which Shirley
Johnson is a member, has the name 'Gubalbarra Reference Group'. No doubt,
'Gubalbarra' refers to gabulharra 'carpet snake person(s)'
In the myth of people west of Townsville, Carpet Snake seems to travel farther
west. (In March 2001) Willy Santo, a Gujal person, narrated a story about how the
Great Basalt Wall, west of Charters Towers, was formed. It is summarized as fol-
lows.
Fire Serpent and Water Serpent fought at Echo Hole. Water Serpent killed Fire
Serpent. Then, at the Burdekin Falls, Water Serpent fought Rainbow Serpent,
which came from Palm Island.
Echo Hole appears to be at the foot of the Great Basalt Wall. (For the locations
of these places, see http://www.pacificislandtravel.com/australia/queensland/
charteredtowers.asp; accessed on 4 May 2011.) Presumably the Great Basalt Wall
was formed during the first battle. Willy Santo stated that the name of Rainbow
Serpent is [gaboba.y,a]. He said that [gabo] means 'snake' (note that the lateral [l]
is missing) and that [ba.y,a] means 'people'. (In his pronunciation, the rhotic is not
a tap, but an approximant- an interference from English.) No doubt, this word
corresponds to [gabolbara] gabul-barra 'carpet snake-denizen'. Since Rainbow
Serpent carne from Palm Island, almost certainly it corresponds to Carpet Snake
of Reggie Palm Island's and Shirley Johnson's stories. That is, in Willy Santo's
story, this snake travels farther west than Townsville.
A dam was built over the Burdekin Falls in 1987 and the falls no longer exist
(Rachel Cummins, e-mail of 6 February 2009; also http://en.wikipedia.org/wiki/
Burdekin_Darn; accessed on 9 February 2009). This is another instance of the de-
struction of an Aboriginal site.
Environmental and socio-cultural background 17
The word gabul 'carpet snake' is found as far south as southern Queensland
and northern New South Wales- more than 1,000 km away from Warrongo, e.g.
Bidyara and Gungabula (Breen 1973: 199) and Margany and Gunya (Breen 1981:
366). Also, Watkin and Hamilton (1887: 224) list kabool for 'Carpet snake' in
the languages of Stmdbroke and Moreton Islands, immediately east of Brisbane.
(See Map 4.) E. Ross (1887: 290) lists cobl.e for 'Snake (carpet)' in the language
of Ballina (New South Wales). About 50krn north of Brisbane, there is a shire
whose name is 'Caboolture'. It is pronounced [kabllilt;P], with stress on [bu:], and
not on [ka] (again an interference from English). According to a tourist brochure
(Bribie Island & Caboolture Shire Tourist Map & Iriformation Guide), this name
means 'the place of the carpet snake'. Obviously cabool means 'carpet snake'. The
meaning of -ture [tfQ] is not known. It may possibly be the comitative suffix ('with,
having'). (For the comitative or the like, Warrongo has C-jiN-yi (cf. 3.6.8), and
Djaru of Western Australia has C-jaroN-yaro (Tsunoda 198la: 227). Similar co-
mitative suffixes occur in many other languages (Di.xon 2002: 170).) If this is the
case, 'Caboolture' means '[a place] with [many?] carpet snakes'. The occurrence
of the word gabul 'carpet snake' in these southern languages suggests that there
may be versions of the story in which Carpet Snake traveled farther south than the
Burdekin Falls.
As mentioned in 1.5.1, there was volcanic activity in the W-GB-G and nearby
regions, and there are many stories that concern it and its results, e.g. lava, basalt,
volcanic craters, and hot springs.
Vera Smallwood (1.8.4.3-[2]) narrated a story (in 1974, in English) that de-
scribes the origin of the Herberton Crater (Map 2), roughly as follows.
Once upon a time there lived a greedy snake. It collected all the scrub turkey eggs.
The scrub turkeys got angry, and chased the snake. The earth, too, got angry and
fire came out of the ground, which created a big crater there. The snake, which had
been chased, fled into the crater and was trapped in it. Nowadays there is a creek
running into the crater. This is the route by which the snake fled into the crater.
Going down south, there is a story that describes the origin of Innot Hot
Springs (Map 3). Alf Palmer narmted two versions of this story (in English). The
story line is as follows.
One night, three men went to spear the gigantic eel Yamani, with torches in their
hands. They came to Nettle Creek, at Innot Hot Springs. They fonnd the eel and
one of them speared it. He then grabbed the spear and tried to pull it out, but the
spear did not come out. The eel began to drag the man, and he dropped the torch
from his hand. Suddenly the area became hot, and hot water came out of the gronnd.
This is what is now Innot Hot Springs. The eel dragged the man into a cave. The
other two men tried to chase the eel and the man, but they couldn't, for the area
aronnd the cave was too hot. So they just waited for him outside the cave. They
waited and waited, but the man never returned. They went back to their camp and
told other people what had happened. They searched for the man again A few days
18 The language and its speakers
later, they came to Blunder Swamp and found the eel dead and lying/resting on a
log. They cut the eel open, but they did not fmd the man inside it. One of the men
present was so overwhelmed by this incident that he decided to name his children
after it. One boy was named Wambino, for the eel was fmmd lying/resting on a log.
[There is an intransitive verb wambi-L 'rest (on something)' - TT.] Another boy
was named Babino, for the eel was cut open. [There is a transitive verb bahi-L 'cut' -
TT.] A third boy was named Bayornbirri, for an eel waves its tail. [There is a transi-
tive verb bayombi-L '[fish] wave [its tail].) A fourth boy was named Wagaygorro
after wagay 'wishbone' of the eel. One girl was named Jilanggo after the fig trees
(jilanggo) that were found in the swamp. Another girl was named J albino after the
water lilies (jalhino) in the swamp.
Nettle Creek flows southwards by the township of Innot Hot Springs (the map
in Glenville Pike [1976] 1990: 217). No doubt, 'Blunder Swamp' refers to the
swamp south of Blunder Park Station (see Map 3). The nounjilanggo exhibits in-
teresting morphological behaviour. See 3.2.1.2. Also, the name Bayombirri will be
discussed in 3.2.1.2.
According to Tommy Murray, a Jirrbal speaker (1.2, 1.8.4.3-[7]), in the story
about Innot Hot Springs, a snake (not an eel) swallowed up humans.
Going further south, there is a Gujal story that concerns the formation of the
Great Basalt Wall, west of Charters Towers; it was cited above.
Thus far, we have looked at stories that in the main describe the formation
of geological features. In addition, the mythology explains the origin of fire, of
water, etc. Thus, a story about bangga"a 'blue tongue lizard' concerns the origin
of water. The version narrated by Alf Palmer tells how Blue Tongue Lizard used
to hide water at Cameron Creek (almost certainly in Warrongo territory; see Map
3), but two mice rolled Banggarra over, so that the water flowed again. This story
is in Text 1 in this volume, repeated from Tsunoda (1988b). (Another version of
this story, narrated by Chloe Grant, who is a speaker of Jirrbal and Girmmay, is
in Dixon (1989a: 32).) Also, Alf Palmer narrated a story about bajinyjila (bird sp.)
regarding the origin of fire. (A Jirrbal version narrated by Chloe Grant is in Dixon
(1989a: 32-33), and a Girramay version narrated by George Watson is in Dixon
(1989a: 187-188). Dixon's gloss for bajinyjila is 'spangled drongo'.)
Dixon (1972: 29, 1989a: 153-154, 295) suggests, regarding the myths he re-
corded, that some of them possibly describe events that actually took place. The
same may apply to some of the stories cited above. For example, the story about
the Herberton Cmter may be a description of a volcanic eruption. The Carpet
Snake Story may concern the ice age when the sea level was low and it was pos-
sible to walk to Magnetic and Palm Islands.
The volcanic craters may have been formed about 10,000 years ago (cf. Dixon
1989a: 154). Similarly, the last ice age ended 10,000 years ago (Flood 1995: 313).
In view of the above, these stories may possibly have been handed down for about
10,000 years (Dixon 1989a: 155).
Environmental and socio-cultural background 19
1.5.4. Names of groups, individuals, and places
Some names have a known etymology, while others do not. Many of the known
etymologies have a mythological origin, but others may not. In terms of structure,
names can be roughly classified as follows.
(a) One word: (a-1) one root, (a-2) two roots (i.e. compounding), (a-3) one root and
one suffix (i.e. suffixation).
(b) One noun phrase.
(c) One sentence.
1.5.4.1. Names of groups
Most of the names of groups, e.g. Warrongo, seem to consist of just one root. There
are two suffixes used in the names of groups: -barra and -gaba.
[1] -barra 'denizen of, associated with' (3.7.1-[1])
This suffix is added to place names, nouns describing characteristics of places (e.g.
plants), and adverbs indicating cardinal directions, among others. The resultant
stems refer to person(s) who belong(s) to, or, who is(/are) associated with, or who
come(s) from, the place or the like. They often refer to a group of people, but they
can also refer to its individual members. They refer to humans, and there is no ex-
ample that refers to animals or plants. This suffix is very common in languages of
Queensland (Tindale 1974: 143, Di..xon 1980: 325), including Warrongo. Examples
include (i) gobo-barra 'person(s) of a leafy country' (gobo 'leaf') (this probably
refers to a group within the Warrongo group; see 1.2), (ii) gonarri-barra 'basalt
person(s)', i.e. 'Warrongo person(s)' (gonarri 'basalt') (1.2); and (iii) gabol-barra
'Carpet Snake person(s)', i.e. 'person(s) of the Townsville-Magnetic Island area'
(1.5.3).
[2] -gaba
This suffix appears to mean 'people' (Sutton 1973:103) or 'denizen of, and to
denote 'an intrinsic relationship, one of belonging or identity' (Peter Sutton, e-mail
message of 26 May 2003). It appears to be non-productive, being attested in the
following names only.
(a) Gabilgaba 'Townsville'
Alec White Sr. stated in 1972 that his language is [kabilgqa] Gabilgara (with
-gara, and not -gaba) and that it was spoken in Townsville and on Magnetic Island.
In 1974 he stated that the name of Townsville is [kabilgaba] Gabilgaba (with
-gaba, not -gara). (See Tsunoda 1996.) In Alf Palmer's Warrongo, too, Townsville
is called Gabilgaba. It seems almost certain that gabil is related to gabul 'carpet
20 The language and its speakers
snake' (the vowel changed from u to z). If this is the case, gabil-gaba means
'person(s) of Carpet Snake' (The meaning of -gara in Gabilgara is not known.)
(b) Wulgurugaba 'person(s) of many canoes'(?)
According to Tindale (1940: 175; 1974: 190), Oates and Oates (1970: 180, 182), and
Sutton (1973: 9, 44), there was a language called Wu1guru or Wulgurugaba (also
spelt Wulgurukaba) in the Townsville area. (This language may be the same as
Alec White Sr.'s Gabilgara.) The rhotic is an approximant, and not a tap (Sutton,
e-mail message of 26th May 2003). Tindale states that wulgum means 'man'.
However, Shirley Johnson, who is a descendant of the group in question, said in
March 2001 that the word Wulgurugaba or Wulgurukaba means 'canoe people'.
(According to Donahue (2007: 48), Tindale recorded the word wulgurm (the
rhotic is a trill, not an approximant) with the meaning 'canoe' (not 'man').) The
word wolgo or wulgu occurs in Warrongo (meaning 'boat'), Nyawaygi (meaning
'boat'; Dixon 1983: 511), and Warrgamay (meaning 'bark canoe'; Dixon 1981: 120).
Warrongo has the suffix -rV 'many', whose vowel copies that of the preceding syl-
lable (see 3.7.1-[16] and 3.13.1.4). (Again, the rhotic is an approximant, and not a
tap.) The names Wulguru and Wulgurugaba may contain -rV If this is the case,
wulgu-nl means 'many canoes' and wulgu-m-gaba means 'person(s) of many
canoes'. The latter word is used in the name of an Aboriginal organization on
Magnetic Island, off Townsville: Wulgurukaba Aboriginal Corporation. (There is
a suburb called Wulguru in Townsville. No doubt it was named after the language
Wulguru. Judging by the phonetics of the languages of this region, the name of the
language was most probably pronounced [6lko.J.o] (with stress on the initial syl-
lable). However, according to Rachel Cummins, the name of the suburb Wulguru
is pronounced [wa1ga.n1:]. This pronunciation is a spelling pronunciation and also
it is heavily anglicized.)
(c) Nhawalgaba
According to Sutton (1995, and e-mail message of 16 June 2003), in 1970 and
1973 Reggie Palm Island referred to the Townsville people as Nhawalgaba. The
meaning of nhawal is not known. The letters nh represents an interdental nasal
(Peter Sutton, p.c.).
(d) Birigaba
The language of the Bowen area is Biri (Maps 4 and 5). It is also called Birigaba.
The meaning of biri is not known.
Environmental and socio-cultural background 21
1.5.4.2. Names of individuals
Selected examples are given below.
[1] Some of the individuals' names have a mythological origin. As seen in 1.5.3,
the story about Yamani, the gigantic eel, yielded six personal names: Wambino
(cf. wambi-L Vi 'rest'), Babino (cf. babi-L Vt 'cut'), Bayombirri (cf. bayombi-L Vt
'[fish] wave [its tail]'), Wagaygorro (wagay 'wishbone'), Jilanggo (jilanggo 'fig
tree'), and Jalbino (}albino 'water lily'). The first two names are based on verbs
and contain the suffix -no. (See 3. 7.1-[12] for -no. It is probably accidental that the
name Jalbino contains no; this name is not based on a verb.) The name Bayombirri
contains the (non-productive) suffix -rri, and the name Wagaygorro contains the
(non-productive) suffix -gorro. The suffixes -rri and -gorro are not attested else-
where in Warrongo. But see 3.11.1.7 regarding -rri.
There was at least one Warrongo person whose name is taken from the Yamani
story, and it was Wambino (mentioned in 1.4.1), who is the father of Alec Collins
(mentioned in 1.2 and 1.3). Further information on them is in 1.8.4.2. There is a
cattle station called 'Wombinoo Station' in Warrongo territory (Map 3). No doubt
it was named after Wambino.
There are two other names that do not seem to be related to this myth, but have
the suffi.x -no. See 3.7.1-[12].
According to Atkinson (1979: 36), Wambino had a brother whose name is
'Warkagaroo'. Possibly this is the name Wagaygorro ([wakaigoro]).
[2] There was a man with the name Gonira ([g6ni.J.a]). He is Alf Palmer's uncle
(also mentioned in 1.5.5 and Table 3-3). This name is based on the verb goni-L Vt
'fight' (cf. (4-546), (4-547)), and it means 'fighter' The suffix -ra means 'someone
who does ' (3.7.1-[11]), like the English suffix -er, cf. fight-er. (As noted in
Acknowledgements, AlfPalmernamedme Ganim. See also 1.8.4.1-[3].)
[3] There are names whose etymology is not known, but for which a tentative sug-
gestion may be made. Examples include the following, which are all Alf Palmer's
children's names: (i) Jinabarro 'Maurice Palmer' (cf. jina 'foot', barro 'bent,
crooked'), (ii) Nganyiri 'Edith Lenoy' (cf. nganyi 'face'), and (iii) Gayangara 'Alf
Palmer, Jr.' (cf. gaya 'father', ngara 'it is not easy to do' (4.24-[10])). If the name
Jinabarro consists ofjina 'foot' and barro 'bent, crooked', it means 'crooked/bent
foot', and it is an instance of compounding.
[4] There are names for which no suggestion may be made, e.g. Dolobo 'Joyce
Palmer', Wilbanyo 'Norris Palmer' (both Alf Palmer's children), and Wolngarra
'Alec Collins' (data from Peter Sutton).
22 The language and its speakers
1.5.4.3. Names of places
:Most of the placenames consist of one word. But some of the placenames have the
form of a sentence. (Sentential place names appear to be common in some other
parts of Australia as well; see Baker (2002: 113-118).) There is at least one place-
name that has the form of a noun phrase.
[1] In the myth that concerns the origin of water (cf. 1.5.3, given in Text 1), two
mice (gala 'mouse') rolled over a blue tongue lizard (banggarra), which had been
lying on top of a spring (?) and hiding water from other animals. Alf Palmer said
that this place (Cameron Creek) has three names: (i) Banggarra ('blue tongue
lizard), e.g. (4-65-b), Text 1, Line 28, (ii) Galo Cmouse'), and (iii):
(1-1) galo-nggo balba-n banggarra-flJ.
mouse-ERG roll-NF B.T.L.-ACC
'Mice rolled Blue Tongue Lizard.'
The placenarne (1-1) has the form of a sentence. The placenames Banggarra and
Galo are instances of those consisting of one word.
[2] In the myth that describes the origin of Innot Hot Springs, a burning torch hit
the ground, upon which hot water gushed out of the ground (1.5.3). Alf Palmer
gave the name of Innot Hot Springs as follows.
(1-2) nyalbay-flJ jido-flJ galngga-n.
story-NOM torch-NOM fall-NF
Tentative translation: '[In] a story a torch fell down.'
This name, too, has the form of a sentence. But it is not known how to analyze the
syntactic status of the first word.
[3] A Gugu-Badhun story, narrated by Richard Hoolihan and recorded by Peter
Sutton (in the early 1970s), describes the formation of a turtle-shaped lava. Alf
Palmer (a Warrongo speaker) said that there is a place that he called 'Turtle Jump'
in English. For its Warrongo name, he gave three alternative forms. They all have
the form of a sentence.
(1-3) banggorro-flJ jolba-n.
turtle-NOM jump-NF
'Turtle jumped.'
(1-4) banggorro-flJ jolba-1.
turtle-NOM jump-NF
'(As above.)'
Envirorunental and socio-cultural background 23
(1-5) banggorro-0 jolba-y.
turtle-NOM jump-NF
'(As above.)'
Alf Palmer said that this place is in Wairuna area (that is, around the border be-
tween the Warrongo and Gugu-Badhun territories; cf. 1.4.1), through which the
Burdekin River flows. This placename probably refers to the event depicted in the
Gugu-Badhun story mentioned above. (The verbjolba- seems to belong to L-class
in (1-4), but toY-class in (1-5). See 3.9.3 for the significance of this dual member-
ship of conjugation.) The version (1-5) occurs in Text 1, Line 17.
The locative form of the version (1-5) is attested: banggorro jolba-ja 'turtle
jump-LOC'; see (2-63). Note that the final y is deleted. (See 3.2.1.1 for the locative
forms.)
In passing, there is something like a sentential place name that is followed by
the ablative-1 suffix (-ngomay): garri-0 galngga-n-ngomay-0 'sun-NOM fall-NF-
ABL-NOM', i.e. '[a person who comes] from [the place where] the sun sets'.
These are the only two instances in which a sentential place name or the like is
followed by a case suffix.
[4] Another Gugu-Badhun story, narrated by Harry Gertz and recorded by Peter
Sutton in 1974, tells how a woomera (wumbun) slipped down in the river. (See
Goetz and Sutton (1986) for details. The narrator's surname is spelt in two dif-
ferent ways: Gertz and Geotz.) This place, whose exact location is not known, is
called Wumbunbarra in Gugu-Badhun. (The use of the suffix -barra for a place
name is unusual. It is generally used for names of groups/persons (1.5.4.1-[1] and
3.7.1-[1]).) Alf Palmer stated that there is a place that has the following name (an-
other sentential place name):
(1-6) wombon-0 joyora-n.
woomera-NOM slip-NF
'A woomera slipped.'
AlfPalmer stated that this place is two miles down the Burdekin River from Valley
of Lagoons. Therefore, it must be in Gugu-Badhun territory. Then the Gugu-
Badhun name Wumbunbarra and the Warrongo name Wombon Joyoran appear to
refer to the same place. (Recall that the practical orthography employed for the
present work uses o in place of u.)
[5] There is another sentential placename:
(1-7) warmgo-0 jana-n.
woman-NOM stand-NF
'A woman stood.'
24 The language and its speakers
Alf Palmer gave a place name that appears to be in the Warrgamay language.
Consulting Dixon (1981), it is tentatively analyzed as follows (employing the prac-
tical orthography adopted for the present work).
(1-8) ngolmboro jowa"a-ngo"a
worn an stand-?
Tentative translation: 'A woman stood.'
The meaning of -ngorra is not known. AlfPalrner said that the place is on Herbert
River, below Abergowrie. This place is in Warrgamay territory (1.4.1). It is pos-
sible that (1-7) is the Warrongo name of this place.
[6] The place name jolay nyon.gol 'tree one' has the form of a noun phrase. Alf
Palmer said that its English name is 'One Tree Plain' and that it is on Herbert River.
It is possible, though by no means certain, that this name is a Warrongo translation
of the English placenarne 'One Tree Plain'.
We have seen place names that have the form of a sentence or a noun phrase. In
terms of stress, they do not seem to constitute one single unit. Thus, where the name
consists of two words, e.g. (1-3) to (1-7) andjolay nyon.gol ('tree one'), the primary
stress seems to fall on the first syllable of the second word, with the secondary
stress on the first syllable of the first word. If they constituted one single unit, the
primary stress would fall on the first syllable of the first word. See 2.7 for stress.
[7] There are placenarnes which appear to involve a suffix but whose etymology is
not known. For example, the placenarne Jalnyjanbara 'Cashmere Station' (where
AlfPalmer grew up; cf. 1.8.4.1) appears to contain the suffix -bara (not -ba"a; cf.
1.5.4.1-[1]). The suffix -bara is attested in many words; see 3.7.1-[8]. The meaning
ofjalnyjan is not known.
[8] There is possibly one placenarne that consists of a noun and an enclitic. As
mentioned in 1.5.3, in the story given by Tommy Murray, a Jirrbal speaker, a snake
(not an eel) swallowed up humans at Innot Hot Springs. According to him, the
name of Innot Hot Springs is jambalgoli. This may be a Warrongo placenarne.
First, Tommy Murray stated that it is a Warrongo name. Second, Innot Hot Springs
appears to be in Warrongo territory (1.4.1). Third, the word for 'snake' is wadam in
Jirrbal (Dixon 19TI: 407), but it is jambal in Warrongo. If this is a Warrongo name,
it may be analyzed as follows. It will consist of the nounjambal-0 and the enclitic
=goli 'only'
(1-9) jambal-r2J=goli
snake-NOM/ACC=only
Tentative translation: 'only the snake'
[9] There are many placenarnes that appear to be based on the geography, flora,
fauna, etc. that characterize the place, e.g. (i) gaba 'white clay' and 'place on Stone
Environmental and socio-cultural background 25
River', (ii) dogal 'fern sp., whose root is edible' and 'place on Herbert River (a
swampy area where many dogal-fems grow)', and (iii) bigal 'death adder (snake
sp.)' and 'Henrietta Creek (near Abergowrie)'.
[10] There are also many placenames whose etymology is not known One example
is bajobala ([batfobala]) 'Kirrama Station'. This station appears to be on the border
among Warrongo, Jirrbal and Girramay territories (1.4.1), but the name is known
to, and used by, many other groups. Kirrama Station has played an important role
in the post-contact history of the region; see 1.7, 1.8.4.1, and 1.8.4.2.
1.5.5. Sections and totems
In the Warrongo group, as in the neighbouring groups of the region, its mem-
bers are divided into four sections. (The sections were sometimes called 'skin' in
the English of Aboriginal people; cf. Sharp 1939: 442.) The information on the
Warrongo section terms is incomplete, but complete information is available for
Gugu-Badhun (Sutton 1973: 106) and Warrgamay (Dixon 1981: 5). These three
groups appear to have an identical section system, and virtually identical section
terms- apart from a few differences in the two rhotics, i.e. an alveolar tap (written
with rr) versus a retroflex approximant (written with r); see below.
The section terms of Alf Palmer's Warrongo are shown in Table 1-7. Those
terms marked with '[TT]' have been reconstructed on the basis of (i) the other
terms in Warrongo and (ii) the equivalent terms in Gugu-Badhun and Warrgamay.
The terms of female members contain the feminine suffix -gan (3.7.1-[17]). In ad-
dition, they all involve the insertion ofy andng (2.6-[2]-(c-3), -(d)). Three of them
involve the change of the root-final vowel o to a, and one of them (won.go-rra-y-
ng-gan) involves the insertion of -rra.
Table 1-7. Warrongo section terms (1) (AlfPalmer's Warrongo)
male members
gorgorro
gorrgila
won. go
wojorro
female members
gorgorra-y-ng-gan
gorrgila-y-ng-gan [TI]
won.go-rra-y-ng-gan [TT]
wojorra-y-ng-gan [TI]
Harry Bunn (a Warrongo speaker (?); cf. 1.4.2.3 and 1.8.4.3-[1]) gave the fol-
lowing terms: gorrgila, gorrgoro, wojorro, and won.go. No term for female mem-
bers was given.
In my data and the sources consulted, there is an alternation between anal-
veolar tap (written with rr) and a retroflex approximant (written with r) in the fol-
lowing terms (see 2.1.3.1 for these phonemes). (In each of these languages, the two
rhotics are in opposition.)
26 The language and its speakers
(a) (i) gurrgila or gorrgila in Warrgamay, Alf Palmer's Warrongo (and also Alec
Collins' Warrongo, Harry Bunn's Warrongo (?), Girramay, Jirrbal (Dixon
1972: 31), Nyawaygi (Dixon 1983: 433)), and (ii) gurgila in Gugu-Badhun.
(b) (i) gurrguru in Gugu-Badhun (and also Harry Bunn's Warrongo (?), Nyawaygi
(Dixon 1983: 433)), (ii) gorgorro inAlfPalmer's Warrongo (and also Girramay
and Jirrbal (Dixon 1972: 31)), and (iii) gurguru in Warrgamay.
Note that the correspondences between rr and rare irregular. (Only the term
wojorro or wujurru contains a tap consistently, in my data and all the sources cited
above.) This fluctuation between the two rhotics is possibly due to language obso-
lescence. (See 1.8.4.1 to 1.8.4.3 for a discussion of the nature of the data obtained
for the present study.) Some of the speakers may have found the rhotic contrast
difficult to retain since English lacks this opposition (see Schmidt 1985: 192-193).
Alternatively, this fluctuation may be due to errors on the part of the linguist(s)
concerned. Finally, the forms cited above may be traditional ('correct') words, and
this fluctuation may reflect the situation of the pre-contact times.
It seems that, in the traditional society, although each person had his/her name
(e.g. Jinbilnggay for Alf Palmer), people often called someone by his/her section
name. This practice among Warrgamay people is described by Lumholtz ([1889]
1980: 218, 303), the Norwegian zoologist who lived among Warrgamay people
from August 1882 to July 1883 (see 1.5.8-[3] and 1.7). An analogous practice is
still now observed in Kimberley (Tsunoda 1981a: 8). No doubt it was common
among Warrongo people as well. Thus, someone would call out to a gorgorro man,
'Gorgorro!'
Table 1-8. Warrongo section terms (2) (Alf Palmer's Warrongo)
SetA SetB
gorgarro goynba
gorrgila woragaja
won. go wolmirri
wojarro yawonya
Alf Palmer gave two sets of equivalent terms. See Table 1-8. Note that the cor-
responding terms from the two sets are not just phonological variations. Dixon
(1989b: 265) reports regarding Jirrbal, Girramay and Mamu that 'each section has
associated with it "polite" terms for referring to male and female members respec-
tively. These are used for address and seem to be in the nature of terms of en-
dearment' The 'polite' terms for referring to male members are wurany, wulmirri,
yaww:zya and guymba. They are identical to the Warrongo terms except for wurany
(its Warrongo equivalent will be woragaja) and except that Dixon lists guymba
(the nasal is bilabial) where I list goynba (the nasal is alveolar). In view of this, Set
B of Warrongo seems to be a set of 'polite' terms for referring to male members.
Environmental and socio-cultural background 27
AlfPalmer said that Gonira, his uncle (1.5.4. 2-[2]), was a wolmirri (a term from
Set B). It seems that Alec Collins, too, had these two sets. Thus, he said he was a
goynba (Set B), although the term gorrgila (Set A) occurs elsewhere in his data.
R.anji Pope, a Gujal speaker (1.8.4.3-[9]), said he was a won.go (Set A). It is not
known if he had the two sets. Alf Palmer's use of the term wolmirri (Set B) to refer
to his mother's brother fits in Dixon's characterization of 'polite' terms. But Alec
Collins' use of goynba (Set B) to refer to himself does not seem to.
Each section has totem(s). Attested totems are given in Table 1-9. Harry Bunn
said his totem was garrgay 'sparrowhawk'. Therefore, his section would be gorrgoro.
Table 1-9. Totems
section totem
AlfPalmer
HarryBmm
gorgorro
gorrgila
won. go
wojorro
gorrgoro
gorrgila
won. go
wojori'O
'carpet snake' (probably gabol- TT)
wqjagan 'crow'
'echidna' (probably barrbira- TT)
'eaglehawk' (probably go17'ijala - TT)
'carpet snake' (probably gabol- TT)
garrgay 'sparrowhawk'
garrijala 'eaglehawk'
garraga 'dove' (HB not certain)
gondolo 'emu' (HB not certain)
Sharp (1939: 450) gives the following words as section names for Gugu-Badhun
and several other languages: Kuparu, Kurkila, Wunggu, Banbari. Sutton's (1973:
106) list of Gugu-Badhun section terms contains gurgila (Sharp's Kurkila) and
wungu (Sharp's Wunggu), but it does not contain any term like Kuparu or Banbari.
My data onAlfPalmer's Warrongo contain another section term, namely, bambari,
and its totem seems to be possum (gajarra). This suggests that Alf Palmer knew
at least one more section term and its totem. (The bilabial nasal may be due to my
mishearing, and the correct form may be banbari. Cf. Sharp's Banbari. The rhotic
in bambari is an approximant, and not a tap.)
1.5.6. Marriage rules
Many aspects of the social life were conducted according to the section system.
One example is marriage arrangement Ideally, a person must choose his/her
spouse from the right section; he/she cannot marry just anyone. Their child's sec-
tion membership is determined automatically. Unfortunately, I failed to obtain in-
formation on marriage rules of Warrongo. However, as seen in 1.5.5, Warrongo
shares with Gugu-Badhun and Warrgamay an identical section system and virtu-
ally identical section terms. Fortunately, complete information on marriage rules is
available for Gugu-Badhun (Sutton 1973: 106-107) and Warrgamay (Dixon 1981: 5).
28 The language and its speakers
It is almost certain that Warrongo had the same marriage rules. They would be as
shown in Table 1-10. In accordance with Radcliffe-Brown's (1930: 38) convention,
the labels 'A, B, C, D' are used for male members, and 'a, b, c, d' for the corre-
sponding female members.
Table 1-10. Marriage rules
A man must many and their child is:
who is: a woman
who is: son: daughter:
gorgorro won.gorraynggan wojorro wojorraynggan
(A) (b) (D) (d)
won. go gorgorraynggan gorrgila gorrgilaynggan
(B) (a) (C) (c)
gorrgila wojorraynggan won.go won.gorraynggan
(C) (d) (B) (b)
wojorro gorrgilaynggan gorgorro gorgorraynggan
(D) (c) (A) (a)
Table 1-10 exhibits interesting regularities. For example, regarding male mem-
bers, Pls son is D, D's son is A, A's son is D, D's son is A, and so on. That is, we
have an alternation between A and D (hence, 'A - D'). This will continue indefi-
nitely (if the marriages conform with these rules). Similarly, B's son is C, C's son
is B, B's son is C, C's so is B, and so on (i.e. B - C). Concerning female mem-
bers, b's daughter is d, d's daughter is b, b's daughter is d, d's daughter is b, and
so on (i.e. b - d). Finally, a's daughter is c, c's daughter is a, a's daughter is c, c's
daughter is a, and so on (i.e. a- c).
As alluded to above, this system of marriage rules presents the 'ideal' arrange-
ments. In real life, however, there were probably marriages that deviated from
them, as was the case, for instance, in Kimberley, Western Australia (Tsunoda
1981a: 10-11).
1.5.7. Kinship system
The kinship system in Alf Palmer's Warrongo seems to be as follows, although
there are points that are not certain. (Those kin terms that are relatively more certain,
include those that concern parents, uncles, aunts, grandparents, grandchildren, and
spouses.)
Figure 1-1 concerns a male EGO. The letter 'D' has two uses. 'D' below a kin
term, e.g. galbin, refers to (a male member of) a section, while 'D' above a kin
term, e.g. galbin, means 'daughter' Vowel-final kin terms generally have two
forms: one with the suffix -na (3.7.1-[18]) and the other without. ngaji-na was not
Environmental and socio-cultural background 29
attested, but almost certainly it is a bona fide Warrongo word. ('ngqji(na)*' indi-
cates that 'ngaji' is attested, but that 'ngajina' is not.) In the following discussion,
the forms without -na will be used.
For a male EGO, his father (FG) isgaya, and his mother (M) yanga. Among his
grandparents, his father's father (FF) is bolo, father's mother (FM) gami, mother's
father (MF) ngaji, and mother's mother (MM) babi. His wife (y.l) is birgo. His
son (S) and daughter (D) are galbin, that is, 'child'. Among his grandchildren, his
son's son (SS) and son's daughter (SD) are bolo, while his daughter's son (DS) and
daughter's daughter (DD) are ngaji. (As seen above, galbin may mean 'man's son'
and 'man's daughter' In (4-551), galbin was translated 'niece' by Alf Palmer. In
view of the kin term system observed in Kimberley, Western Australia (Tsunoda
1981a: 11-13), probably galbin can also mean 'man's brother's son' (i.e. 'nephew')
and 'man's brother's daughter' (i.e. 'niece').)
We shall look at Figure 1-1 in connection with the marriage rules (cf. Table
1-10). Take FF of Figure 1-1, for instance. For the purpose of exposition, the section
A is arbitrarily assigned to him. His son (F) is D, whose son (EGO) is A, whose
son (S) is D, whose son (SS) is A That is, we have an alternation between A and D,
exactly as stated regarding Table 1-10.
FF X FM MF X MM
bolo(na) gami(na) ngqji(na)* babi(na)
A b B a
F X M
gaya(na)
I
yanga(na)
D c
EGO X w
I .... .. ;
A b
'
SW X S D x DH


SS SD DS DD
bolo(na) bolo(na) ngqji(na)* ngqji(na)*
A a B b
Figure 1-1. Pedigree (1): a male EGO
Figure 1-2 concerns a female EGO. For the purpose of exposition, in Figure 1-1,
a male EGO was assigned to A, and his wife is to b. Likewise, in Figure 1-2, a
30 The language and its speakers
fern ale EGO is assigned to b, and her husband to A (In fact, the statements about
the kin relationship are applicable irrespective of an EGO's section membership.)
For her parents and grandparents, a female EGO uses the same terms as those
used by a male EGO. But the terms for her spouse, children and grandchildren
differ from those used by a male EGO. Her spouse, i.e. husband (H), is gornggal.
Her son (S; a 'D' man) isjowana. (The formjowa, without -na, is hardly ever used.
See 3.7.1-[18]. For jowana, Alf Palmer gave the gloss 'nephew' as well. Probably
;owana can mean 'woman' son' and 'woman's sister's son' (i.e. 'nephew').) A
female EGO's daughter (D; a 'd' woman) may be yindala, but this is not certain.
Alf Palmer gave yindala, indicating that it meant (i) 'female EGO's daughter (D)',
i.e. 'woman's daughter', and (ii) 'niece' (probably 'woman's sister's daughter'
and 'man's sister's daughter'). But later he denied the meaning (i). (Alf Palmer's
gloss for yindala is, in effect, 'niece from sister's side') Therefore, it is not certain
if there is a word for 'female EGO's daughter' If yindala really means 'female
EGO's daughter', there are separate words for 'son' and 'daughter' for a mother
(though not for a father; see Figure 1-1). (Mamu, north of Jirrbal (Maps 4 and 5),
has the kin term yindala. Its referents seem to include 'female EGO's son and
daughter' (Dixon 1980: 109).)
FF x FM MF x MM
bolo(na) gami(na) ngaji(na)* babi(na)
B a A b
F
gaya(na)
c.
X
H x EGO
A b

M
yanga(na)
d
SW x S D x DH
I
I jowa(na)
D d C.
I I
c
ss SD DS DD
gami(na)
A
gami(na) babi(na) babi(na)
a B b
Figure 1-2. Pedigree (2): a female EGO
Regarding Figure 1-1, we saw an instance of the alternation between two sec-
tions, involving men. Alternations occur concerning women as well. Take Mt\.1, for
Environmental and socio-cultural background 31
example. She is b. Her daughter (M) is d, whose daughter (EGO) is b, whose
daughter (D) is d, whose daughter (DD) is b. That is, we have an alternation of 'b- d'.
The terms for the grandparents' generation and the grandchildren's mean 'X
and reciprocal' (cf. Dixon 1972: 399). For instance, bolo may mean 'male EGO's
FF' (Figure 1-1), and also 'female EGO's FF' (Figure 1-2). In addition, it means
'male EGO's SS and SD' (Figure 1-1). That is, 'FF and reciprocal'. Similarly, ngaji
may mean 'male EGO's MF' (Figure 1-1) and also 'female EGO's MF' (Figure
1-2). In addition, it means 'male EGO's DS and DD' (Figure 1-2). That is, 'MF and
reciprocal'. The same applies to gami 'FM and reciprocal', and to babi 'MM and
reciprocal'.
The use of some of the kin terms is remembered still now. For example, in
March 2002, Rachel Cummins (Alf Palmer's DD) stated that Alf Palmer used
to refer to Raymond Palmer and Roderick Palmer (they are his SS) as bolo.
(Reciprocally, the latter would refer to Alf Palmer (their FF) as bolo.) She also
stated that Alf Palmer (her MF) used to call her (his DD) ngaji. (Reciprocally, she
would call him ngaji.)
Other attested kin terms include the following. A few of the terms listed above
are repeated for convenience.
(a) For both a male EGO and a female EGO: (i) F and FB: (ii) FZ:
bimo(na), (iii) M and MZ: yanga(na), (iv) MB: galnga(na).
Note that gaya refers to F and also FB. Similarly, yanga refers to M and also
MZ. There is no single term for 'uncle'. It has to be either gaya 'FB' or galnga
'MB' Similarly, there is no single term for 'aunt'. It has to be either bimo 'FZ' or
yanga 'MZ'. (As noted in 1.3, for MB, Alec Collins used galngina in addition to
galngana.)
For a male EGO, bimo can mean WM, in addition to FZ. Then, galnga will be
expected to mean WF, in addition to MB. But it is not known if galnga can mean
WF.
(b) For a male EGO: (i) EB: mogina (mogi not attested), (ii) EZ: bolgo (bolgona not
attested), (iii) YB: yabojana (yaboja not attested), (iv) YZ: banina, bolgo (bani
not attested. Cf. bani 'stone'.)
Bolgo seems to mean both EZ and YZ. If this is the case, the opposition be-
tween 'elder' and 'younger' is neutralized in the case of bolgo for a male EGO (i.e.
the opposite sex to EZ and YZ). It will be interesting to see if this neutralization
occurs between EB and YB for a female EGO. Unfortunately no relevant data are
available.
(c) Alf Palmer gave the following kin nouns and glossed them as shown: (i) }or-
gana 'elder sister' (jorga not attested), (ii) woribo 'son-in-law' (woribo-na not at-
tested), (iii) marga 'brother-in-law', and (iv) marga-yan 'daughter-in-law' (-yan may
be a feminine suffix. There is a more productive feminine suffix: -gan (3.7.1-[17].)
Also he gave (v)joway 'brother-in-law' (probably WB, and not HB), (vi)jowayngg-
orro 'term by which brothers-in-law call (address? refer to?) each other' (later, Alf
32 The I anguage and its speakers
Palmer cancelled this meaning and gave the gloss 'big mob ofjoway'), (vi) mogayo
'sister-in-law' (probably HZ, and not WZ), and (vii) mogoynggon 'term by which
to refer to one's deceased wife in avoidance of her name' A song by Alf Palmer
(see (1-10)) yieldedgorramara 'YB, YZ'. See 3.8.1-[7] for its possible etymology.
1.5.8. Other topics
We shall look at some other topics regarding the socio-cultural background -
mainly those topics which are relevant to the language itself, those which are ex-
pressed in the language or those which occur in the examples cited.
[1] A person's language seems to have been a marker for self-identification at an
inter-tribal encounter (cf. Clarke 2003: 39). See Text 1, Line 7.
[2] The ear is regarded as the seat of intelligence and memory (cf. Dixon 1972: 30,
Tsunoda 1981a: 7). There are a noun bormw 'deaf, and two verbs derived from it:
borrmo-bi-L Vi 'become deaf and also 'forget' (4.7.1-[1]-(a)), e.g. (3-88), (4-243),
(4-276), and borrmo-nga-L Vt 'make [someone] forget' (4.7.2-[1]-(a).
[3] Message sticks were used for conveying a message. According to Leigh
Pentecost (e-mail message of 26 March 2009), a message 'stick was their "pass-
port" that would allow them to enter others' territory without being killed'. Carl
Lurnholtz ([1889] 1980: 327-328) (mentioned in 1.5.5; cf. also 1.7) describes mes-
sage sticks used by Warrgarnay people. In Warrongo, figures on message sticks
are called jin.gorr, and message sticks are known as jin.gorr-ji 'message figure-
COM', i.e. '[wood] with!having message figure'. See 3.6.8 for the comitative -ji.
[4] There are a nounjigobina 'shooting star', and an intransitive verb derived from
it: jigobina-bi-L (4.7.1-[1]-(a)). This verb literally means 'be/become a shooting
star', but Alf Palmer used it to describe a young man who was getting married
soon. The same noun and the intransitive verb derived from it occur in Jirrbal
(Henry 1967: 55, Dixon 1972: 401, 1997: 23), but no connection with a man's mar-
riage is known.
[5] As elsewhere in Australia, there was a taboo on the use of the name of a de-
ceased person (Berndt and Berndt [1964] 1968: 112, 297, 389, Clarke 2003: 49,
Dixon 1980: 28-29, Tsunoda 1981a: 7).
[6] A deceased relation's skull was carried for safety; it would warn its carrier of
an approaching danger. See (4-764).
[7] Cannibalism was practiced in this region (Lurnholtz 1980: 294, Dixon 1972:
28). Dixon states: 'Anyone who has persistently broken the social code may be
killed by some of the senior men of the tribe, his flesh eaten and his blood offered
to younger men to drink.' Dixon continues: 'Some men attain the status of gubi
("wise man" or "doctor"). They generally have a thorough knowledge of their en-
Special styles of speech and songs 33
vironment and of the customs and beliefs of the tribe. A necessary prerequisite
for being a gubi is that one has drunk the cannibalistic victim The word gobi
occurs in Warrongo, too; see (4-698), and Text 3, Line 30. Alf Palmer seemed to
believe that a gobi ate human flesh (}algor). See (4-6). He dreaded a gobi.
[8] People believed in the existence of two types of imaginary hairy human-like
beings, who live in lava scrub: (i) gandaro, smaller type, and (ii) gangaligan,
larger type. At least gandaro - possibly gangaligan as well - was feared by Alf
Palmer and presumably by other people, too. The word gandaro occurs in (3-68).
See also (4-401).
[9] Alf Palmer described a technique for possum hunting as follows. When
a possum climbs a tree (e.g. a bloodwood tree), it scratches its bark. The hunter
blows on its trunk. If the trunk has a fresh track left by a possum, then barks chips,
dust, etc. caused by the possum, come off the trunk, indicating that the possum is
likely to be still on that tree. (See also Clarke 2003: 130.) (4-165) contains the verb
boya-L Vt. 'blow with mouth' and describes this technique. There are two words
derived from boya-L: (i) the noun boyal---boyal 'bark chips that come off a trunk
caused by a possum's scratches (for instance, on a bloodwood tree)', and (ii) the
verb Vi 'be/become [such] bark chips' (3.11.1.6-[5], -[6]).
[10] The noun mandija means (i) 'grub in its early stage of growth' and (ii) 'tooth
ache' Possibly tooth ache was believed to be caused by this grub.
[11] A type of vine poison (marra) and a type of fruit poison (mangga) were used
for fishing, to intoxicate fish. See Text 2, Lines 5, 6.
[12] The noun dombil means 'knot of a forest ti-tree' The knot contains water,
which was used as drinking water by Aboriginal people. There are a verb dombil-bi-L
Vi 'have a knot (which contains water)' (4.7.1-[1]-(a)), e.g. (4-900) (second B), and a
place name dombil-bolo (3.7.1-[5]).
[13] There are at least three terms for seasons: bi"gil, bin-gi-bara (the latter
without l; see 2.6-[3]) 'cold weather, winter', woyolo 'summer, beginning of
summer', andgarrimal 'summer'
Alf Palmer produced running texts of about six hours. They contain valuable
information on the traditional ways of life and beliefs. They need to be analyzed.
Three excerpts are included in the present volume.
1.6. Special styles of speech and songs
1.6.1. Jalngoy: the avoidance style of speech
Like many other Australian languages (cf. Dixon 1980: 58-65, Tsunoda 1981a:
14-15, 215-220), Warrongo had a special style of speech that was used between
34 The language and its speakers
taboo kin. Its name is Jalngoy, which Alf Palm referred to as 'big word' in English.
There appears to be no special term for the ordinary style of speech. and probably
it was simply called gogo 'speech, language'
Information on Jalngoy is scanty. Alf Palmer had an excellent knowledge of
the ordinary style, but he admitted that he could not speak Jalngoy. (In a language
death situation, a special style of speech seems to be among the first to cease to
be transmitted, and it is lost earlier than the ordinary style; see Tsunoda 2005:
97-98, 107-108.) Nonetheless, Dyirbal, too, had an avoidance style of speech (with
the same label 'Jalnguy'), and detailed information is available on it (Dixon 1972:
32-34, 292-327).
In Dyirbal, for a male EGO, one of the primary taboo relatives is his wife's
mother (Wlv.f). For example, he had to use Jalnguy whenever his WM was within
earshot. Probably the same applied to Warrongo as well.
Warrongo has two intransitive verbs that mean 'speak in Jalngoy' (i) jalngoy-
bi-L (:Jalngoy-INTR') literally 'be Jalngoy' (4.7.1-[1]-(a)), e.g. (3-139), and (ii)
jalngony-ji-bi-L ('Jalngoy-COM-INTR') literally 'be with/having Jalngoy' See
3.6.8-[3] for the comitative case. Note the alternation betweeny and ny (3.2.1.1-[4]).
In Warrongo, the word for WM is bimo (1.5.7). Alf Palmer gave a Warrongo sen-
tence that means 'When you are at your bmw's house, you must speak in Jalngoy':
(3-139). See also (4-544).
In Dyirbal, Jalnguy and the ordinary style (called Guwal) are identical in
phonology, and almost the same in grammar, but entirely different in vocabu-
lary (Dixon 1972: 32). Presumably the same was true of Warrongo. Alec Collins
(Sutton's data) and Alf Palmer independently gave many nouns and verbs as
Jalngoy words. Unfortunately these words were not crosschecked. Nonetheless,
there are two words that coincide: wombal 'dog' (the ordinary style equivalent is
gando) and the transitive verb ngarrambi-L 'hear' (the ordinary style equivalent
is ngawa-L, e.g. (3-305)). It seems certain that these two words are Jalngoy words.
For ngarrambi-L 'hear', see (1-10-d, -e) and 3.11.1.1.
Dixon (1970: 667, 1972: 320) suggests that Jalnguy of a given language was
built up by borrowing words from the ordinary style of nearby languages. Indeed,
this may apply to some of the words given by Alf Palmer and/or Alec Collins as
Jalngoy. Thus, the noun wombal 'dog, dingo' occurs in (the ordinary style, no
doubt) of Buluguyban of Palm Island (Tsunoda 1996: 82). The verb ngambi- Vt
'hear' occurs in (the ordinary style, no doubt) of Freddy Toomba's Gujal (Peter
Sutton's data). It may be related to the Warrongo ngarrambi-L 'hear' (For Freddy
Toomba, see 1.8.4.3-[9].)
1.6.2. Songs
A nurn ber of songs by Alf Palmer were recorded. They concern ga"gay 'sparrow
hawk', a funeral, a sick person, e.g. (4-70), love, e.g. (1-10), and sexual relationship,
Special styles of speech and songs 35
e.g. (1-11), etc. The ex. (1-10) is cited from the end of Tape 72/25. In this story, after
living on Palm Island for many years, Alf Palmer returned to Cashmere Station
(where he grew up; see 1.8.4.1-[1]), and sang this song: the lines c to f. The song
is presented by IPA syrn bois and the practical orthography adopted for the present
work. I am unable to indicate melody, rhythm, voice quality, etc.
(1-1 0) a. ngaya borrgoman-da nyina=ngomay-0
1SG.NOM(S) Palrn.Island-LOC stay=after-NOM
ngaya ngoni-ngomay-0 ya"o-wo golmi yani-0 yamba-wo
1SG.NOM(S) there-ABL-NOM here-DAT back come-NF camp-DAT
ngaygo wara-yi wara-yi-d-go.
1SG.GEN (error) one's.own-COM-LINK-DAT
'After I had stayed on Palm Island, I carne back here.' (S=S)
b. ngona-ngomay ngaya yarro-n-da nyila=gol jangala-0
that-after lSG.ERG here-LINK-LOC today=only Jangala-ACC
ngaya baya-lgo.
1SG.ERG sing-PURP
'Then I will sing Jangala here, soon [or, for the first time].'
c. [kaorai IJaja .J.ai:olJ1d3o:
gawo"a-yi-0 ngaya rayiwoblJ?io-wo
love-COM-NOM 1SG.NOM very.good.looking.girl-DAT
maborao gorama.J_ao IJaja]
maborra-wo gorramara-wo ngaya
good.looking.girl-DAT last.one-DAT lSG.NOM
'I [am] in love with a very good-looking girl, a good-looking girl[, and]
the last one.'
d. [IJawai: IJawala IJapa
nga-wayi ngawa-la nganya
nga-Q hear-PSI 1SG.ACC
'Did you hear me [singing (?)]?'
IJaiJaram be lao]
nga-ngarrambi-la-wo I
nga-hear-PST-wo
e. [wai: IJawala IJapa IJaiJarambela]
wayi ngawa-la nganya nga-ngarrambi-la
Q hear-PSI 1SG.ACC nga-hear-PST
'Did you hear me [singing(?)]?'
f. [.J.ai:olpd3o: maborao gorama.J_ao
rayiwob'l)'jo-wo maborra-wo gorrarnara-wo
very.good.looking.girl-DAT good.looking.girl-DAT last. one-DAT
IJaja mic)3iraii]go]
ngaya mijirray-nggo I
1SG.NOM promised.one-DAT
'I [am in love] with a very good-looking girl a good-looking girl, the last
one[, and] the promised one.'
36 The language and its speakers
g. jalnyjanbara-ngga ngaya wa wara-yi-da
Cashmere.Station-LOC lSG.NOM(S) (error) one's.own-COM-LOC
baya-gali-0
sing-ANTIP-NF
yinda nganya ngawa-yal nyila=gol
2SG.ERG(A) 1SG.ACC(O) hear-PURP today=only
[ngaya TT] golmi yani-ngo-0. (AP)
[lSG.NOM(S)] back come-ADNOM-NOM(S)
'At Cashmere Station, my own [country], I sang [Jangala] so that you would
hear me [sing], who had just come back today.'
The song is from the lines c to f. Alf Palmer said that this song is jangala.
(Jangala means 'a singing style' in Jirrbal and Girrarnay (Dixon 1972: 401), and
'obscene song style' in Warrgamay (Dixon 1981: 114).) In the lines c to f, a slash
indicates a pause. There is no pause between the lines c and d, and between e and f.
In the line c, the last vowel of each word is somewhat longer than usual, but this is
not shown in the above presentation.
Etymologically, rayiwolnyjo contains the noun rayi 'young girl' (used in the
ordinary style), but rayiwolnyjo seems to be used in songs only. The meaning of
-wolnyjo is not known. The meaning of gorramara is not clear. Alf Palmer's gloss
for it is 'last one' In this context, this may mean 'least or less good-looking girl'.
He also said it meant 'younger brother, younger sister' (1.5.7). (See 3.8.1-[7] for a
possible etymology of gorramara.) Mijirray means 'fiancee promised by a pro-
spective father-in-law or mother-in-law'
This portion of the text cited in (1-1 0) contains useful examples for the discussion
ofWarrongo morphology and syntax. They will be referred to in relevant sections.
A song about sexual relationship:
(1-11) (A man was having sex with a woman.)
a. wajila-nggo galoworo-nggo gombo-0 jingga-la.
testicles-ERG testicles-ERG buttocks-Ace hit-PST
'The [man's] testicles hit the [woman's] buttocks.'
b. gombo-0 ngarba-la ngarba-la. (AP)
buttocks-NOM jurnp-PST jurnp-PST
'[Her] buttocks jumped with a fright.'
Songs exhibit a few differences from the ordinary speech, as follows.
The suffix -la 'past' (3.10.1) is quite common in songs, e.g. (1-10-d, -e) and
(1-11), but it rarely occurs in the ordinary style of speech, e.g. (3-192). According
to Alf Palmer, ngawa-la is a song word. In contrast, the nonfuture form ngawa-n
'hear-NF', i.e. 'hear, heard', occurs in the ordinary style.
Songs contain a nurn ber of words that Alf Palmer considered as Jalngoy. They
also contain words that occur in songs only. For example, (1-10-d, -e) contains
Special styles of speech and songs 31
ngarrambi-L Vt 'hear' (this may be a Jalngoy word; cf. 1.6.1) - in addition to
ngawa-L Vt 'hear' of the ordinary style. (1-11) contains wajila 'testicles' (which
Alf Palmer regarded as Jalngoy) and galoworo 'testicles' (which occurs in songs
only). (The ordinary-style word for 'testicles' is galon, e.g. (4-904).)
The syllable [IJa] nga is added to the beginning of wayi and ngarrambi-la
in (1-10-d). Consequently, all the resultant 'words' begin with nga. (This is
an instance of alliteration.) Also, in (1-10-e), nga is added to the beginning of
ngarrambi-la. Furthermore, in (1-10-d), the syllable [o] wo is added to the end of
ngarrambi-la. The addition of such syllables may be for the purpose of rhyming or
adjusting the number of syllables (or morae). Such elements, used in songs without
any clear meaning, are called vocables by Hinton (1994: 145). (Similar phenomena
occur in some of the songs ofDjaru and Wanyjirra of the northwest Australia.)
See 3.11.1.1 for a comparative account of ngarrambi- Vt 'hear' (a verb that Alf
Palmer and Alec Collins considered as Jalngoy) and two verbs that occur in songs
only.
1.6.3. Curses and expressions for abuse
Alf Palmer gave four expressions that appear to be curses or expressions for abuse.
One of them appears to be a Warrgamay sentence. (Alf Palmer knew Warrgamay;
see 1. 8.4.1-[1]. He worked as a consultant for Dixon's study of Warrgamay (Dixon
1981: 13).) Its phonetic presentation and its tentative phonernicization are given.
Also, tentative morpheme-by-morpheme glosses and translation, based on Dixon
(1981), are added.
(1-12) [Wi:ki IJinba 6panbolo]
lwigi IJinba wupan-bulu/
no.good 2SG.NOM lustful-very
'You are no good. You are very lustful.'
The other three are as follows.
(1-13) [mako.J.aiJgo IJani gapd3indii:lena]
(1-14) [wapd3anda:laiJgo IJani makO.J.aiJgo gapd3inda:lena]
(1-15) [.J.aimondo -1a:d3an gowa1fi]
It is not known what these expressions mean. (1-13) and (1-14) are probably variants
of the same expression. Phonetically and phonologically, these expressions do
not deviate from the usual pattern of Warrongo (see Chapter 2). Indeed, there are
Warrongo words that resemble (part ot) them: ngani 'what', ganyji-n carry-NF', and
wanyja 'where'. Nonetheless, they do not seem to be Warrongo. Indeed, for a reason
that is not understood, (1-15) sounds a little exotic, i.e. non-Warrongo, to my ears.
38 The language and its spealrers
It is interesting to note that these expressions are either not Warrongo or do not
seem to be Warrongo. Possibly people thought that cursing/abuse was more effec-
tive when uttered in a foreign language.
Other styles of speech, e.g. initiation style and secret style, and also sign lan-
guage (cf. Kendon 1988, Tsunoda 1981a: 15) may have existed, but no information
on any of them was obtained.
1.7. Post-contact history
Allingham (1993), Bolton ([1963] 1975), Jones ([1961] 1970) and Glenville Pike
([1976] 1990), among others, provide accounts of the post-contact history of a
region that includes Warrongo territory. The post-contact history of Warrongo peo-
ple's neighbours is described by Dixon (199lb: 350) on Mbabaram, Dixon (1972:
34-37) on Dyirbal (i.e. Jirrbal, Girramay, etc.), Dixon (1981: 6-9) on Warrgamay,
Dixon (1983: 434) on Nyawaygi, and Sutton (1973: 20-21) on Gugu-Badhun.
Cattle and gold - preceded by eA-plorers - decimated the original inhabit-
ants and destroyed their traditional life, as was the case in Kimberley, Western
Australia (Tsunoda 1981a: 16).
The first Europeans who entered Warrongo territory appear to be the Prussian
explorer, Ludwig Leichhardt, and his party, who in 1845 followed Burdekin River,
crossed the Great Dividing Range, and went on to Lynd River (Leichhardt [1847]
1996). They seem to have passed around the western border of Warrongo territory.
Leichhardt details his encounters with the local Aboriginal people. Some of them
may have been Warrongo.
In 1848, Edmund Kennedy's eA-pedition left Cardwell and headed northwest.
Jones (1970: 50) suggests that the party travelled near Ravenshoe. Judging by
the map in Carron ([1849] 1996), they passed somewhere near Innot Hot Springs
and Mount Garnet. Mount Garnet appears to be around the northern border of
Warrongo territory (1.4.1). So they may have entered Warrongo territory.
The following expeditions did not enter, or do not seem to have entered,
Warrongo territory: Augustus and Francis Gregory's expedition in 1856 (Gregory
and Gregory [1844] 1968: 180-184), George Dalrymple's in 1859 (Farnfield 1968:
17), and Frank and Alexander Jardine's in 1864 (Hiddins 1998).
Leichhardt and the Gregorys reported on the promising potential of this region
for pastoral industry, and this attracted the attention of pastoralists and people
with similar interests. Soon afterwards, pastoral industry was introduced into this
region. In probably 1860 the Valley of Lagoons, which is no doubt the most attrac-
tive tract of land for pastoral industry, was taken up by Dalrymple and his partners
(Jones 1970: 65). This was followed by the establishment of other cattle stations.
'Between 1873 and 1877 Cashmere and Gunnawarra stations were carved
from the outlying blocks of the Valley of Lagoons' (Bolton 1975: 92). In 1872-73
John Atherton established Cashmere Station (Jones 1970: 149). Cashmere and
Post-contact history 39
Gunnawarra are in Warrongo territory. Kirrama Station was taken up by Edgar
Collins in the 1880s (Atkinson 1979: 31).
Carl Lumholtz (1.5.5 and 1.5.8-[3]), a Norwegian ethnographer, lived among
Warrgamay people from 1882 to 1883 (Lumholtz 1980). He was based at Herbert
Vale, but in February 1883 he visited the Valley of Lagoons Station. Jones (1970:
241) suggests that he visited Cashmere Station. No doubt he travelled through
Warrongo territory.
Gold rushes broke out in the 1860s and 1870s, and one of the largest gold
rushes occurred at Charters Towers in 1872 (Bolton 1975: 44-49, 52, Jones 1970:
136). The gold rushes occurred outside Warrongo territory, and no gold seems to
have been found inside it.
Gold was not the only metal that affected the life of the Aboriginal people. In or
near Warrongo territory, tin was found in 1880 at Wild River (Map 3), a tributary
of Herbert River (Bolton 1975: 116, Glenville Pike 1990: 68). Silver was found in
1881 at Silver Valley, on Dry River (a branch of Wild River) (Bolton 1975: 117-118,
Glenville Pike 1990: 134-135). CopperwasfoundatMountGametin 1883 (?)(Bolton
1975: 118). On the basis of Glenville Pike's (1990: 217) map and Dixon's (1991b:
348) map, Wild River may be in Ngaygungu territory or in Warrongo territory, and
Silver Valley, southeast of Irvinebank, appears to be in Ngaygungu territory.
The influx of miners and fossickers, together with the establishment of cattle
stations and the introduction of cattle, destroyed the traditional pattern of life.
Then followed the railway, which extended from Cairns and reached Mt. Garnet in
1902 (Glenville Pike 1990: 84).
The clashes between the original inhabitants and the intruders led to many
massacres, which were carried out by local property owners and the police from
Cardwell and Herberton all around the district (Allingham 1993; Bolton 1975;
Dixon 1972: 34-35; Jones 1970). As an example, in March 2000, Rachel Cummins
told me that, at Blencoe Falls in the Herbert River Gorge (in Warrongo territory),
a group of Aboriginal people were pushed over and fell into the gorge. One of the
people killed was Lucy, Alf Palmer's mother, i.e. Rachel Cummins' great-grand-
mother (see 1.8.4.1-[1]). AlfPalmer gave two sentences that describe his mother's
drowning: (3-31) and (3-291). Local Aboriginal people say that still now the place
is haunted. (This massacre occurred probably around 1900.)
Poisoning of Aboriginal people by white settlers was common (Allingham
1993: 125, Dixon 1972: 34). Thus, according to Rachel Cummins, Alf Palmer
became ill from eating poison flour supplied by a white man.
As seen above, the life of the original inhabitants of this region was drasti-
cally affected. Nonetheless, according to Bolton (1975: 95) 'until1914 some of the
Aborigines between the Herbert and the Atherton Tableland maintained their tradi-
tional way oflife in a form at least comparable with their ancestors' It seems that a
fair number of people from different tribes, including Warrongo, lived at Kirrarna
Station, which is on the border among Jirrbal, Girramay and Warrongo (cf. 1.4.1).
This probably explains why people like Alf Palmer and Alec Collins knew both
Warrongo and Jirrbal, among other languages. See also 1.8.4.1 to 1.8.4.3.
40 The language and its speakers
No doubt these groups had frequent contacts with each other in pre-contact
times as well. (This explains, at least partially, the massive borrowing from the
Herbert River Group into Warrongo, Gugu-Badhun and Gujal (1.4.2.4).) Thus, ac-
cording to Leigh Pentecost (e-mail messages of 26 March 2009 and 6 April 2009),
prior to European settlement, Girramay, Jirrbal and Warrongo shared the initiation
site and the ceremonial ground.
There was more to befall the Aboriginal people and destroy their life. In 1914
the Hull River Aboriginal Settlement, southeast of Tully, started and it 'finally
broke up the tribes' (Jones 1970: 304). Aboriginal people from different tribes
were collected and sent there. Some of them were 'handcuffed and chained to be
driven to the [settlement- TT] which seems to have been a penal settlement'
(Jones 1970: 305). Mixing of people with different language had a deteriorating
effect on the maintenance of the traditional languages (see Tsunoda 2005: 58-59).
In March 1918 the Hull River Aboriginal Settlement was destroyed by 'the worst
cyclone in living memory to strike the coast' (Jones 1970: 308). 'Houses, per-
sonnel and aborigines were removed to Palm Island' (Jones 1970: 310). Nowadays
Palm Island, too, is known as a penal settlement. Rosser (1978) gives an account of
the situation on Palm Island, from an Aboriginal perspective.
1.8. Studies on the Warrongo language
1. 8.1. Introductory notes
Studies on the Warrongo language can be divided into two groups: early studies
(1.8.2) and modern studies (1.8.3).
Ludwig Leichhardt's (1996) journal (cf. 1.7) does not contain any information
on any language of the region. Sharp (1939: 256-257, 439-441) mentions the
names of 'tribes' of northeast Australia, but he does not mention Warrongo. Also,
the location of some of the tribes does not match that shown in 1.4.2.2.
1.8.2. Early studies on Warrongo
Early works on Warrongo were by people who were not trained in linguistics. They
used spellings that were based on the English orthography. Some of the words are
recognizable, but others are not
[1] E. M. Curr 'circulated among squatters, policemen and magistrates across the
continent a list of 125 English words, for which he requested the local Aboriginal
equivalents; he published the resulting 300 vocabularies as a four-volume work'
(Dixon 1980: 13). The book is Curr (1886). Unfortunately, it does not contain any
word list that was collected in Warrongo territory. The word lists that were collected
nearby include the following. The pages are as in Vol. 2 ofCurr (1886). (i) No. 112.
'The LyndRiver', by W. 0. K. Hill (pp. 400-401). (ii) No. 114. 'Near the head of the
Studies on the Warrongo language 41
Walsh River', by John Atherton (pp. 408-411). (iii) No. 117. 'Head of Gilbert River',
by Edward Curr (pp. 416-417). (iv) No. 119. 'Herbert River', by William S. Stephen
(pp. 422-423). (v) No. 122. 'Clarke River', by Gresley Lukin (pp. 436-437). (vi)
No. 123. 'Top of the range near Dalrymple', by W. E. Arm it (pp. 440-442).
Dixon (1970: 676) briefly assesses ten word lists, and Sutton (1973: 16-20) pro-
vides a detailed assessment of nineteen word lists. There is not much more to add,
except to note the following two points.
First, Dixon (1970: 676) suggested that Nos. 114, 117 and 122 might be in
Warrongo, and that No. 123 was either in Warrongo or some other closely related
dialect. However, subsequent research by Sutton (1973: 16-20) and myself indicate
that they are not Warrongo.
Second, in No. 112, Hill states that the tribe in question was surrounded by
three tribes, one of them being Warrialgona. As mentioned in 1.2, Tindale (1974:
188) gives Warrialgona as an alternative name of 'Warungu', but he doubts if
Warrialgona is really 'Warungu'. Indeed, the language recorded by Hill is clearly
not Warrongo, and consequently Warrialgona does not seem to be an alternative
name ofWarrongo.
[2] As mentioned in 1.2, William Craig's letter dated 24th July 1898 to A W. Howitt
(cited by Dixon 1981: 11) lists the names of the five tribes of the region, one of
which is "War-oong-oo" This may be the first written record of the name of the
language and people under study. The letter contains eight Warrongo words, seven
of which are recognizable. My comments, preceded by 'TT', are added.
Yuln-gun 'Sun' (TT: This is not recognizable. The Warrongo word for 'sun' is
/ga.ri/. Warrgamay has /jug an/ 'rain' (Dixon 1981: 121).)
Ballanoo 'Moon' (TT: This is Warrongo: lba1anu/ balano 'moon'.)
Boor-ee 'Fire' (TT: This is Warrongo: lbwil bori 'fire'.)
Com-oo 'Water' (TT: This is Warrongo: /gamu/ gamo 'water'.)
Nowa 'No' (TT: This is Warrongo: /Jlawa/ nyawa 'no'.)
Yae-oo 'Yes' (TT: This is Warrongo: /juwu/ yowo 'yes'.)
Wan-ja 'Where' (TT: This is Warrongo: lwaJ1Jal wanyja 'where'.)
yan-ulgoo 'go'
Yan-ulgoo is not attested in modern Warrongo. It will be discussed in 3.9.4.1.
1.8.3. Modern studies on Warrongo
Modern studies on Warrongo have been carried out by trained anthropologists
or linguists (Peter Sutton and Tasaku Tsunoda were postgraduate students in the
early 1970s), with the exception of Jack Doolan.
[1] La Mont West and E. F. Aguas worked on Warrongo briefly (cf. Oates and
Oates (1970: 175-176)).
42 The language and its speokrs
[2] Jack Doolan is a white man who worked on Palm Island reserve in the 1960s as
'a clerk in the settlement office' (Dixon 1989a: 116). He recorded Warrongo from
Alf Palmer. Peter Sutton kindly made his transcripts of Doolan's tapes available to
me.
[3] 'Norman B. Tindale [ ... ] An obituary' by Philip G. Jones (http://www.anu.edu.
au/linguistics/nash/aust/nbt/obituary.html, accessed on 19 April 2007) states as
follows: 'During an eighteen month period through 1938-39, Tindale led a data-
gathering expedition'. In 1938 on Palm Island, Tindale recorded Alf Palmer's ge-
nealogy, and photographed him. Tindale's 'Wordlists in North Queensland lan-
guages' (1938, unpublished) (whose photocopies were made available to me by
Peter Sutton) contain about 100 words for ''WaruiJu'. Less than half of them are
recognizable. Furthermore, they appear to include a fair number of words that are
not Warrongo, e.g. jara (in Tindale's spelling) 'man', i.e. yara (in the orthography
adopted for the present work). For 'man', Warrongo has bama, while Jirrbal,
Girramay and Mamu have yara (Dixon 1972: 408). Tindale (1974: 45-46) lists five
words from 'Warungu'. Most of these words are fairly accurately transcribed and
glossed. However, it seems that Tindale did not distinguish the two rhotics: anal-
veolar tap and a retroflex approximant
[4] Since 1963, R. M. W. Dixon has been working on many languages of North
Queensland; see 1.4.1 for his works. He obtained some data on Warrongo from
Alf Palmer and Alec Collins. (Dixon 1989a describes his fieldwork in North
Queensland, and contains accounts of Alf Palmer and Alec Collins. There is a
photo of Alf Palmer, taken in 1964 on Palm Island, on p. 112. See p. 78 for Alec
Collins.) R. M. W. Dixon kindly made his data from Alf Palmer available to me.
As for the data from Alec Collins, Dixon (p.c.) stated that there is nothing in them
that is not in Sutton's Warrongo data from Alec Collins ([5] below). The ex. (3-232)
describes linguistic work by AlfPalmer and R. M. W. Dixon.
[5] Peter Sutton conducted a survey of a vast area in North Queensland in 1970,
salvaging what was left of a large number of languages, such as Gugu-Badhun,
Warrongo, Gujal, Wamin/Agwamin, Mbara, Buluguyban, and Biri. He worked on
Gugu-Badhun twice in 1970 and again in 1974; one of the outcomes is his M.A.
thesis (Sutton 1973). In 1970 Peter Sutton interviewed Alec Collins and obtained
some materials on Warrongo. He generously made all his Warrongo data and
also his unpublished data on Gugu-Badhun and Gujal- available to me.
[6] I carried out fieldwork mainly on Palm Island and also in the adjacent areas
on the mainland, three times in all, in 1971, 1972 and 1974. My main focus was
on Warrongo, but I also obtained a small amount of data on Biri, Gujal, Gabilgaba,
Buluguyban, Nyawaygi, Wagaman, and Wamin, among others. The outcome of
the work on Buluguyban and Gabilgaba is Tsunoda (1996). Sutton's and my data
on Biri are incorporated into Terrill (1998). My previous works on Warrongo are
my M.A. thesis (Tsunoda 1974a) and a number of papers that deal with aspects of
Studies on the W anongo language 43
the language (mentioned in the Preface); they are all incorporated into the present
work.
[7] According to Dixon (1991b: 353), in 1977 'a man who called himself Alec Chalk
(or Col Stephens) and who purported to speak Mbabaram', was recorded at the
then Australian Institute of Aboriginal Studies; the materials were transcribed by
Michael Walsh. 'Of the 200 words or so words he gave, about 20 are recognisable
as Mbabaram ... ; the remainder are largely Warungu, ' I have no expertise to
comment on Mbabaram, but at least I recognize a number of words that are iden-
tical with the corresponding words recorded from Alf Palmer, e.g. wangal 'boo-
merang', mongan 'mountain' (in the orthography adopted for the present work)
1.8.4. Speakers interviewed for the present study
Already in the early 1970s Warrongo was on the verge of extinction, and there
were only two fluent speakers left: Alf Palmer and Alec Collins. Dorian (1973:
437) notes: 'the last speakers of a dying language can be a very misleading source
of information about the language' In view of this, it is vital to describe the
nature of the data obtained, regarding factors such as the speaker's proficiency,
his/her biography, and the conditions of the work (Tsunoda 2005: 234). Comments
on the speakers interviewed for the Warrongo language are provided in the fol-
lowing.
1.8.4.1. AlfPalmer (Warrongo name: Jinbilnggay)
[1] Biography
Alf Palmer's Aboriginal name is IJinbili]gaj/ Jinbilnggay. (According to Dixon
(1981: 13, 115), his name is !Jim.bib:.Jaj/.)
The evidence provided by Rachel Cummins (AlfPalmer's daughter's daughter)
suggests that Alf Palmer was born around 1880. Also he told me that he remem-
bered the Boer War (1899-1902). This indicates that by 1899-1902 he was old
enough to know about a war that had broken out overseas.
Rachel Cummins stated that Alf Palmer told her that he was born at a cattle
station called Silver Valley, south of Mount Garnet. (I have been unable to find the
location of this station.)
According to the information obtained by Norman B. Tindale in 1938 on
Palm island (cf. 1.8.3-[3]), Alf Palmer's mother is a 'Djiru' person from 'Clump
Pt', her English name is Lucy, and his father is an Englishman whose name is
Palmer. Indeed, Alf Palmer told me that his father was an Englishman. (Rachel
Cummins, too, stated that Alf Palmer's mother's name is Lucy.) 'Djiru', i.e. Jirru,
is the language of Clump Point, north of Tully; see Dixon (1972: 25, 1989a: 110).
(Photocopies of Tindale's records are lodged at Aitkenvale Library in Townsville.)
44 The language and its speakers
According to 'Identification Cards', Alf Palmer was born in 1891 (not around
1880) at Herberton (not at Silver Valley). His father is Tom Palmer and his
mother Kitty Palmer (not Lucy). (The printouts of the computer-typed version
of 'Identification Cards' I obtained at Aitkenvale Library have the following
title: 'Card Index held by the Community and Personal History Section of
the Department of Aboriginal and Torres Strait Islander Policy in Brisbane'.)
According to Rachel Cummins, Kitty was a Warrongo woman, who looked after
Alf Palmer at Gunnawarra Station, after his mother (Lucy) had been killed in a
massacre at Blencoe Falls (cf. 1.7).
Alf Palmer told me that he grew up at Cashmere Station (Warrongo name:
Jalnyjanbara). (See (1-10-g).) According to Rachel Cummins, he worked around
Gunnawarra Station and Cashmere Station (in Warrongo territory- TT), and also
in various places such Valley of Lagoons Station (in Gugu-Badhun territory- TT).
In my view, the fact that he mainly worked around Gunnawarra and Cashmere
strongly suggests that the language he acquired first is Warrongo, rather than any
other Aboriginal language. And that he spoke a central dialect of Warrongo.
Alf Palmer married a Jirrbal woman whose English name is Elizabeth and
who was born at Kirrarna Station (on the border among Jirrbal, Girramay and
Warrongo; see 1.4.1 and 1.7). Elizabeth worked at Kirrarna Station while Alf was
working around the area. But they were moved to Hull River Mission, and when
the mission was destroyed by the 1918 cyclone, they were moved to the Palm
Island settlement (cf. 1.7). Alf Palmer lived on Palm Island all the while until his
death in 1981. (The information provided in this paragraph is largely from Rachel
Cummins.)
[2] Linguistic knowledge
As noted above, Alf Palmer's first language was probably Warrongo. In addition,
he knew many other Aboriginal languages, and English. (Alf Palmer himself said
that he knew many languages. This is also seen in Text 1, Lines 2 to 6.) Dixon
(1989a: 112) states: 'Besides Warungu, Alf also knew Jirrbal and Girrarnay, and
Warrgamay' According to Rachel Cummins, Alf Palmer told her that he spoke to
Elizabeth in Jirrbal. Also, he acted as a consultant for Dixon's study ofWarrgamay
(Dixon 1981: 13-14).
Alf Palmer was interviewed by R M. W. Dixon on Palm Island in 1964, and
produced some Warrongo materials (Dixon 1989a: 112-113).
Alf Palmer was interviewed by me during three spells of fieldwork in 1971-72,
1972 and 1974 on Palm Island. (The examples (4-234) to (4-236) and (4-681) de-
scribe our linguistic work.) In 1971, he was about 90 years old. But he was strong
and active. (I sometimes saw him rowing a dinghy and going out to the sea for
fishing, to supply fish to his family and friends. The exx. (3-146), (4-318), and
(4-319) and (4-765) describe how he maintained his fishing net or made a new one.)
He was alert, and he retained a very good memory. He produced extensive data,
including about six hours of running texts, about 1,500 words, and details on mor-
Studies on the W anongo language 45
phology and synta.x. This is comparable to the kind of data that I obtained on a
living Aboriginal language, i.e. Djaru of Kimberley, Western Australia (Tsunoda
1981a). This is truly remarkable at least for two reasons. First, already in the early
1970s most of the languages in this area were extinct or facing extinction. Alf
Palmer's linguistic knowledge was exceptional. Second. according to Alf Palmer,
he had not spoken Warrongo for 50 years (apart from linguistic interviews), i.e. ap-
proximately since he came to Palm Island.
In my experience of fieldwork in North Queensland and also Kimberley, people
who have a good knowledge of their language, generally speak the kind of English
very different from that of white people. Alf Palmer was exceptional in another re-
spect: he had a very good knowledge of Warrongo, and at the same time he spoke
a variety of English very close to that of white people. This indicates that he had a
talent for language.
Despite his excellent knowledge of (the ordinary style of) Warrongo, Alf
Palmer admitted that he could not speak Jalngoy, the avoidance style. (See 1.6.1.)
This suggests that, when he was acquiring the ordinary style, probably before 1900,
Jalngoy was no longer used actively.
When Alf Palmer started working with me in 1971, he called his language
[k.o.U,1fal] (Gurijal), but in 1972 and 1974 he called it Warrongo. See 1.4.2.3 for a
possible implication of this.
As mentioned above, Alf Palmer knew many other languages than Warrongo -
to varying degrees, including Jirrbal and Warrgamay. The data obtained in 1971
and 1972 contained many non-Warrongo elements largely lexical but they were
checked with him in 1974.
It is interesting to see how Alf Palmer pronounced the names of other lan-
guages. Examples follow. (i) Gugu-Badhun was pronounced [kokobad3on]. Clearly
Alf Palmer lacked the laminal contrast of IQI versus IJI (cf. Table 1-4.). (ii) For
Yidiny (Dixon 1977b), Alf Palmer gave [itip]. He correctly recognized the word-
finallamino-palatal nasal. (iii) Alf Palmer knew the name of the Wamin language
(west of Gugu-Badhun) and pronounced it [wamen] (1.4.2.2). (iv) As seen in
1.4.2.2, for Mbabaram, Alf Palmer gave [ba:ba.ram]. He dropped the initial nasal
/m/ and replaced the short vowel [ba] with the long vowel [ba:]. In Warrongo, no
consonant cluster is allowed word-initially. (This vowel lengthening is an instance
of compensatory lengthening; see 2.6-[4].) (v) An interesting case concerns the
language whose name is written by Di.xon (1983) as Nyawaygi. This name ap-
pears to contain the comitative suffix -gi 'with' (Dixon 1983: 433, 458). North of
Nyawaygi and east of Warrgamay is Biyay, which is, together with Warrgamay,
described by Dixon (1981). According to Dixon (1981: 3), the people 'could refer to
themselves as Biyaygiri, involving the productive derivational suffix -giri "with"'
The word biyay means 'no', and Biyaygiri literally means '[person] with the word
biyay "no"' Now, for Nyawaygi, Alf Palmer gave [JlawagjLLT.il This may be tenta-
tively written 'Nyawagiri'. It is possible that AlfPalmer was confused between the
Nyawaygi suffix -gi and the Biyaygiri suffix -girl and that he used the Biyaygiri
46 The language and its speakers
-giri for the name of Nyawaygi. However, it is also possible that the suffix -giri
once existed in Nyawaygi as well and that it survived in the name 'Nyawagiri' See
also 3. 7.1-[22] for -giri.
The reliability of the Warrongo data I recorded from Alf Palmer has been the
target of some criticism, e.g. Dixon (1981: 13, 1994: 180). However, this criticism
is misguided.
Regarding morphology, Alf Palmer's data are reliable, for he remembered de-
tails in morphology, and he even remembered irregularities in it See, for example,
(i) the ergative suffix -lo (3.2.1.2), (ii) the comitative case (Table 3-11), and (iii) the
suffix -nyjarr 'kinship' (3.7.1-[19]). (Language obsolescence often causes leveling
of paradigms, and therefore, ifthere is any irregularity in a given paradigm, and if
it occurs consistently, then it is likely to constitute reliable data; cf. Tsunoda 2005:
242-243.)
As for vocabulary, Alf Palmer's data initially contained intrusions from other
language(s). However, when confronted with the data, he was careful to correct
any mistake he might have made. For an example, see Text 2, Line 6.
Furthermore, Alf Palmer's vocabulary exhibits regular phonological corre-
spondences with those of southern J\1ari languages, which are more than 1,000
krn away (1.4.2.4). This indicates that AlfPalmer's vocabulary is reliable. (It is in
order to show this point that, for Warrongo of Table 1-6, only the words from Alf
Palmer (and not from Alec Collins) were given.)
Dixon (1994: 180) states that many of the Warrongo texts that I recorded from
Alf Palmer 'are conversations in Warrungu between him and Tsunoda'. This is not
correct. The six hours of running texts are entirely monologues by Alf Palmer.
Excerpts of the texts are given in Tsunoda (1974a: 610-651, 1988: 643-645), and
also in the present volume. All my field tapes (or their copies) are lodged with the
Australian Institute of Aboriginal and Torres Strait Islander Studies in Canberra.
Di.xon (1981: 13) adds: 'Although Palmer's parents were Warrungu, he has lived
most of his life in Wargamay- and Dyirbal-speaking territories' Again, this is not
correct; see Alf Palmer's biography given above.
There are reported instances in which a language with no written tradition has
been influenced by one with that tradition. For example, Mithun (1988: 351) reports
many instances of borrowing of conjunctions, in a contact situation, from a language
with that tradition to one without. (See Olson and Astington (1990: 706-708) on
the impact of writing on language change.) However, it seems certain that Alf
Palmer's Warrongo was not influenced by a language with that tradition- English,
in this case -except for about one dozen nouns (2.13.2) and possibly the tag question
marker /aj/ [ai] (2.2.3.1). This is because Warrongo has not had a written tradition.
AlfPalmer neither read nor wrote English. The same is almost certainly true of the
people from whom Alf Palmer learned Warrongo (probably before 1900).
[3] Letter from AlfPalmer written by Rachel Wilson
Probably in 1972 (or 1974?), Rachel Wilson (who is Alf Palmer's DD, about six-
teen or seventeen years old at that time, and who was to become Mrs. Rachel
Studies on the W anongo language 41
Cummins) wrote a letter to me on behalf of AlfPalmer. It was a reply to my letter.
It starts as follows.
(1-16) Dear Gunira,
Nyah yaroo younangoo coco birin. I hope you understand my spelling of
the language. I am AlfPalmers granddaughter and he has asked me to write
this letter to you.
The first word, Gunira, is the name that Alf Palmer gave me (1.5.4.2-[2]). At that
time, I used the letter 'u' for the phoneme lui (cf. 1.1, and also 2.14). Clearly Rachel
Wilson adopted my spelling. The other words are analyzed as follows.
(1-17) [IJaja jaro inongo koko birin]
ngaya yarro-0 yinon-n-go gogo-0 birri-n
ISG.ERG this-ACC 2SG-LINK-DAT word-ACC send-NF
'I am sending this word to you.'
This letter also contains the word 'Djinbilnggay', i.e. Alf Palmer's name.
Again, Rachel Wilson adopted my spelling of that time. This letter contains two
more Warrongo words. In the letter, Alf Palmer says that they are the names of
his grandchildren. One is "Japanoo" This is no doubt Jabino and the name of
Roderick Palmer (AlfPalmer's SS; see 1.5.4.2-[1]). The other is "Indinoo" It is not
recognizable. According to the practical orthography adopted for the present work
(1.1, 2.14), it may be written as Yindino. Rachel Cummins believes that it is the
Warrongo name of Raymond Palmer, who is Roderick's EB.
This letter, written by Rachel Wilson, is probably the very first written record
of the language ever made by a Warrongo person.
1.8.4.2. Alec Collins (Warrongo name: Wolngarra)
Atkinson (1979: 31) states: 'In the 1920's there was still a small tribe of blacks
living in their wild state in the Herbert River Gorge. The head of this tribe was
Wambino. He had seven sons and a daughter', including Patrick Hooligan and Alec
Collins. (Atkinson uses the spelling 'Alick', and not 'Alec'). This suggests that
Alec Collins grew up in the Herbert River Gorge. (See 1.5.4.2-[1] and 3.7.1-[12]
for the name Wambino.) Alec Collins' Warrongo name is Wolngarra (data from
Peter Sutton). Atkinson (1979) contains accounts of Warnbino and his family, and
a photo of Wambino on p.32. The photo on p.40 shows some of Wambino's sons,
including Alec. Both photos were taken in 1925 at Kirrama Station. At that time,
they were living at Kirrarna Station (in Girramay territory'?; cf. 1.4.1).
Since Alec Collins appears to have grown up in the Herbert River Gorge, prob-
ably he spoke an eastern dialect of Warrongo. (How Alf Palmer and Alec Collins
referred to their language was discussed in 1.2.)
48 The language and its speakers
Alec Collins was interviewed by R M. W Dixon in 1963 (7) at Kirrama
Station. He was also interviewed by Peter Sutton at Kirrama Station in 1970, and
produced some material, including some words and sentences that he said were
Jalngoy ('avoidance style'). Subsequently, however, his whereabouts were not
known. After a long search. in September 1974 I found Alec Collins- alas! -lying
in bed at Herberton Hospital. He was already too weak to do any linguistic work.
Nonetheless, I talked to him in the Warrongo language that I had learned from Alf
Palmer. It seems certain that he understood my Warrongo perfectly. Also, I asked
him a fair number of questions in Warrongo, which he answered in Warrongo. For
example, I asked him about Alf Palmer as follows.
(1-18) ngani-0 nyongo gogo-0? (TT)
what-NOM 3SG.GEN language-NOM
'What is his language?'
Alec Collins replied, saying '[warOI]o]' Furthermore, I narrated a hunting story
in Warrongo (the kind of story Alf Palmer had narrated to me many times). Alec
Collins enjoyed it very much. There were no signs of incomprehension on his part.
This suggests that the Warrongo language as recorded from Alf Palmer is reliable,
being perfectly interpretable by another speaker of the language. (My interview of
Alec Collins is reported in Tsunoda 2005: 242.) Alec Collins was probably almost
as proficient as Alf Palmer was. He passed away in 1975, before adequate data
were recorded form him.
Alec Collins said that Alf Palmer was his [kab]ina] /gab]ina/ ':MB'. It is not
known if he meant 'actual 1\.ffi', or 'classificatory :MB'. (For ':MB', Alf Palmer
used /gab] ana/ and /ga1IJa/; the second vowel is /a/, and not /i/. This is one of the
small lexical differences between their idiolects (1.3).)
1.8.4.3. Other speakers
[1] Harry Bunn was interviewed by me in Townsville in September 1974. He
was born at Innot Hot Springs (on the northern border of Warrongo territory) to
a mother from Valley of Lagoons (a Gugu-Badhun person?) and a father from
Stanley Hill, Irvinebank (a Mbabaram person?). Ranji Pope, a Gujal speaker (see
below), is his half-brother and Richard Hoolihan, a Gugu-Badhun speaker (see [8]
below), is his uncle (presumably 1\.ffi, not FB; in Aboriginal people's English. the
word 'uncle' often means '1\.ffi'). He called his language Gujal, but it was prob-
ably Warrongo, for, like Warrongo but unlike Gugu-Badhun and GujaL it lacked a
laminal opposition (Table 1-4). He was able to produce only limited data, mainly
lexical. The section terms and totems given by him are cited in Table 1-9.
[2] Vera Smallwood was interviewed by me in Ingham in September 1974. She
told me that she was born in 1922 at Mount Gamet to a Russian father. (Mt. Garnet
Studies on the W anongo language 49
is on the northern border of Warrongo territory.) According to Govor (2001: 141,
221, 343, 383), Vera Smallwood was born on the 31st :March 1922 in Spring Gully
(Queensland), her father is a Russian and his name is Leandr Illin, her maiden
name is Vera Araluen Illin, and her mother's maiden name is Kitty Clarke, who is
a Ngajan person (spelt 'Ngadjon' by Govor). (See Map 5 for Ngajan.) In addition to
the story about the Herberton Crater (cited in 1.5.3), she gave 21 words. She was
unable to identify the language, but most of the words she gave coincide with Alf
Palmer's Warrongo.
[3] Mr. Williamson was interviewed by me in Ingham in September 1974. He gave
si.x words. He, too, was unable to identify his language, but five words agree with
AlfPalmer's Warrongo. Mr. Williamson himself suggested the words he gave were
from a language close to Warrongo.
[4] Mrs. Morgenstern was interviewed by me in Ingham in September 1974. She
gave one word: /wagal/ 'eel'. This coincides with Alf Palmer's Warrongo. (My
fieldnotes show the name 'Morgenstern', but this may be my error, and the name
may possibly be 'Morgensen'. Atkinson (1979: 37) describes a white man whose
name is Billy Morgensen. It is possible that some Aboriginal people adopted this
white man's name. As of the early 2000s, there are Warrongo people whose family
name is 'Morganson'. This name may derive from 'Morgensen' (Leigh Pentecost,
e-mail message of26 :March 2009).)
[5] Denny Hoolihan was interviewed by me in September 1974, probably in Mt.
Garnet. He told me that his father is Patrick Hooligan. (The latter is Wambino's
son and Alec Collins' brother; see 1.8.4.2.) He gave about six phrases or short
sentences. Since his father is Wambino's son, the materials he gave would be ex-
pected to be Warrongo. However, while some portions appear to be Warrongo,
others do not seem to be. He himself said that he knew Jirrbal, but hardly knew
Warrongo. (This suggests that Jirrbal was more dominant than Warrongo in the
early 1970s in Mt. Gamet.)
[6] Mrs. Cassidy was interviewed by me in Ingham in September 1974. She told me
that she was born at Gunnawarra Station. She gave five words. Since Gunnawarra
is in Warrongo territory, the words she would be expected to be Warrongo, but
they do not seem to be.
The materials given by Vera Smallwood, Mr. Williamson, Mrs. Morgenstern,
Denny Hooligan, and Mrs. Cassidy are reproduced in their entirety and assessed in
Tsunoda and Tsunoda (2007).
[7] In addition, the following three people were interviewed by me. They are
second-language speakers (or maybe third-language speakers) of Warrongo.
Tommy :Murray (mentioned in 1.2, 1.5.3 and 1.5.4.3-[8]), a Jirrbal speaker, was
interviewed in Mount Garnet in September 1974. Tommy Springcart (mentioned
50 The language and its speakers
in 1.4.2.2), who was a consultant for Dixon's study of Jirrbal (Dixon 1972: 28,
1989a: 119-120), was interviewed (by me) in August and September 1974 on Palm
Island. Ado Cashmere, a Warrgamay speaker, was interviewed in Mount Garnet
in September 1974. (His wife Cheepa is a daughter of Patrick Hooligan, who is
Wambino's son.) The data from Tommy Murray and from Ado Cashmere in the
main agree with Alf Palmer's Warrongo. The data from Tommy Springcart seem
to be a mixture ofWarrongo and Jirrbal.
The consultants for languages other than Warrongo are listed below.
[8] For Gugu-Badhun (Sutton 1973), the following people were interviewed by
Peter Sutton: (i) Richard Hoolihan in 1970, and some of his children (Margaret
Gertz, Janet Eaton, Dennis Hoolihan, Bully Hoolihan), and (ii) Harry Gertz in
1970 and 1974, and some of his children (Henry Gertz, Eddy Gertz, Eric Gertz,
Mollie Gertz). Atkinson (1979) contains accounts of Harry Gertz, and a photo of
him on p.ll6. Peter Sutton kindly made his unpublished Gugu-Badhun available to
me, and a story narrated by Richard Hoolihan and one by Harry Gertz are cited in
1.5.4.3-[3], and -[4], respectively.
[9] For Gujal, Freddy Toomba (who called his language Gujal) was interviewed
by Peter Sutton in August 1974 on Palm Island. Ranji Pope (who called his lan-
guage Gurjal) was interviewed by Sutton in August 1970 in Charters Towers. He
was born in Wondecla, near Herberton (which appears to be in Ngaygungu terri-
tory, according to Di.xon 199lb: 348), and was a half-brother to Harry Bunn. I con-
tacted him in March 1972 in Charters Towers, but I was able to record very little.
(Sutton kindly made the data from Freddy Toomba and from Ranji Pope available
to me.) George Reid (who called his language Gujal) was interviewed by Gavan
Breen in Charters Towers in August 1970 (Breen, p.c.) and by me on Palm Island
in September 1974. (Gavan Breen kindly made the data from George Reid avail-
able to me.)
[10] Reggie Palm Island, whose language is Buluguyban of Palm Island, was inter-
viewed by me on Palm Island in September 1974. His Carpet Snake story is cited
in 1.5.3. Rosser (1978: 47-48) gives an account of Reggie Palm Island.
[11] Alec White Sr. was interviewed by me on Palm Island in 1972 and 1974.
He was born in Townsville, and he called his language is Gabilgara and also
Gabilgaba. See 1.5.4.1-[2]-(a) for its etymology.
All the information recorded from Reggie Palm Island and Alec White Sr., to-
gether with George Reid's story about a canoe-shaped stone on Palm Island, are in
Tsunoda (1996).
[12] In the 2000s, during my participation in the language revival movement (cf.
Preface), I have obtained data from the following people. (i) Rachel Cummins (nee
Present-day situation 51
Wilson), Alf Palmer's DD, who has been interviewed since 2000 and who is often
cited in Chapter 1. (ii) Shirley Johnson, a Wulgurugaba person of the Townsville-
Magnetic Island area, was interviewed in March 2001 and her Carpet Snake story
is cited in 1.5.3. (iii) Willy Santo, a Gujal person, was interviewed in March 2001
and his Water Serpent story is cited in 1.5.3.
All the interviews by me mentioned in 1.8.4.1 to 1.8.4.3 were conducted in
English.
Regarding the Warrongo data, more than 95% was obtained from Alf Palmer.
His data are extensive and include running texts of about six hours, about 1, 500
words, and details on morphology and syntax. In contrast, the data obtained from
Alec Collins are severely limited. They contain about 250 words, a very small
amount on morphology, and very little on syntax. The Warrongo data obtained
from the other people are even more severely limited and largely lexical In view of
the above, the description of Warrongo that follows concerns Alf Palmer's idiolect,
although, where appropriate, reference will be made to Alec Collins'.
As noted in the Preface, what follows is an example of what can be done, and
what cannot be done, when only the last single speaker is available. (My sugges-
tions for fieldwork on endangered languages are in Tsunoda 2005: 216-247.)
1.9. Present-day situation
It is difficult to estimate the number of people who identify themselves as
Warrongo. They are scattered, living in Mount Gamet (in Warrongo territory?),
and in places outside Warrongo territory, such as Palm Island, Townsville, Ingham,
Cardwell, and Cairns.
No Warrongo person speaks the language fluently. But many people, including
children, know a number of Aboriginal words, although they are unable to distin-
guish Warrongo words from those of other languages. The extent of the linguistic
knowledge that has been retained is discussed in Tsunoda and Tsunoda (2007).
There are instances of semantic change. For example, the word gambi has un-
dergone the following change:
(1-19) gambi
(i) 'possum cloak' -. (ii) 'clothes' -+ (iii) '(ladies') underwear'
(i) The word list 'No. 133 Jv.lt. Black', by The Government of Queensland (Curr
1886, Vol. 2: 486), lists 'kumbi' (i.e. /gambil) 'possum cloak'. Mt. Black is west of
Townsville. See Map 2. (ii) In 1971-1974 Alf Palmer gave gambi 'clothes' (gener-
ally referring to European cloths), e.g. (4-18-a/-b). (iii) In 2006 Rachel Cummins
and her daughters said that gambi means '(mainly ladies') underwear'.) Another
example of semantic change is:
52 The language and its speakers
(1-20) bori
(i) 'fire' (AlfPalmer, 1971-1974)-+ (ii) 'light' (in the 2000s)
Most of the present placenames are English. But a very small num her of them
are taken from Aboriginal languages. One example is the name of a suburb in
Townsville: Wulguru, named after the language of Townsville (1.5.4.1-[2]-(b)). A
childcare centre in Townsville is aptly named Galbiri. The word galbiri 'children'
occurs in Warrongo (3.7.1-[16]). But it is not attested, say, in Gugu-Badhun (see
Sutton 1973: 213), in my data of the language of Townsville (Tsunoda 1996) or
in Donahue's (2007) study of the languages of Townsville, Magnetic Island and
Palm Island. It seems that the person who named this childcare centre knew this
Warrongo word.
In 1972 (?) I translated a hymn into Warrongo at the request of people of Palm
Island and in consultation with AlfPalmer, but this did not develop into a language
revival movement. Recently the Warrongo language revival movement started (cf.
Preface and Acknowledgements). Its central figure is Rachel Cummins. Upon re-
quest, I started Warrongo lessons in March 2002 in Townsville, and as of 2010 the
lessons have been conducted five times, about four or five days each time. The
Warrongo language revival activities are reported by Tsunoda (2002, 2004, 2005:
212-213) and Tsunoda and Tsunoda (2006, 2007, 2008, 2010).
Chapter2
Phonology
2.1. Phonemes and their realizations
2.1.1. Phonern e inventory
The Warrongo phoneme inventory comprises eleven consonants, two semivowels
and four vowels. It is one of the smallest phoneme inventories among Australian
languages (cf. Dixon 1980: 140, 143). See Tables 2-1 and 2-2. The phonemes are
written with IPA symbols. The symbols used in the practical orthography em-
ployed in the present work are shown in pointed brackets (see 2.14). (The use of
pointed brackets for letters used in a pmctical orthography is adopted from Harvey
2002: 17.)
Table2-1. Consonant and semivowel phonemes
bilabial a pi co- retroflex larnino- dorso-
alveolar palatal velar
stop b <b> d<d> .f <j> g <g>
nasal m<m> n <n> Jl IJ <ng>
rhotic c <rr> l <r>
lateral 1 <1>
semivowel j <y> w <w>
There are a stop phoneme and a nasal phoneme for each of the following places
of articulation: bilabial, apico-alveolar, !amino-palatal, and dorso-velar. For stops,
voicing is not distinctive, and both voiced and voiceless allophones occur. There
are two rhotics (i.e. r-sounds): lei (an alveolar tap [r]) and /lf (generally a retroflex
approximant [.(), and infrequently a retroflex tap [r]); one lateral; and two semi-
vowels. The vowel system basically comprises three members, and length is sig-
nificant for the pair /a/ and /a:/ only and to a very limited degree (2.1.2).
Table 2-2. Vowel phonemes
close
open
front
i <i>
a, a1 <a, aa>
back
u <o>
54 Phonology
The practical orthography employs <y> for the semivowel/j/, and <j> for the
lamino-palatal stop /J/. Compare:
(2-1) [jamba]
/jambaJ
<yamba>
'camp'
[d3ambal]
!Jam ball
<jambal>
'snake'
The phoneme lui is written with <o>, and not <u>, e.g. (2-2). This is partly be-
cause phonetically /u/ is almost always [o], and not [u] (see 2.1.4.2 and 2.14).
(2-2) [ Wa01]0]
/wacuiJuf
<warrongo>
'the name of a language'
[goman]
/gum an/
<goman>
'other'
The name of this language is written 'Warrongo' in this practical orthography.
A dot is used to distinguish a sequence of /n/-plus-/g/ from the single phoneme
lrjl, e.g.:
(2-3) [bangall
/bangall
<ban.gal>
'little girl'
[jaiJa]
/jaiJal
<yanga>
'mother'
[bongo]
lbungu/
<bon.go>
'maggot'
[boiJgo]
lbUI]gu/
<bonggo>
'knee'
The IPA symbols are used in the remainder of Chapter 2, but from Chapter 3
onwards, the practical orthography will be used.
As is generally the case in Australian languages, there is no fricative phoneme.
There are two interjections that start with the fricative [h]: [haha] or [ha:ha:] 'ex-
clamation of surprise, etc.' and [hai] 'Hi!' But these expressions are marginal, and
the phoneme /h/ is not set up for Warrongo.
Also, there is no affricate phoneme. It is a common practice in Australian lin-
guistics to refer to the lamino-palatal obstruent phoneme (e.g. !J! ofWarrongo) as
a stop (cf. Dixon 1980: 135). However, in Warrongo, Djaru (Tsunoda 1981a: 25)
and all other Australian languages I have listened to, phonetically this phoneme is
not a stop ([J] or [c]). Rather, it is an affricate ([1f] or [d3]); see (2-1). (Blake (1979:
16) on Kalkatungu and Patz (2002: 19) on Kuku Yalanji note that the phoneme in
question has an affricate characteristic.) Nonetheless, in accordance with the usual
practice, the present work considers it a stop phoneme; recognition of an additional
'manner of articulation' (i.e. affricate) has been avoided.
/bl, /gl, 1m/, and /1]1 are labeled 'peripheral' (cf. Dixon 1980: 139). In terms of
articulation, they involve 'articulation at the extreme back or front of the mouth'
(Dixon 1980: 139). Acoustically, the peripheral phonemes share the property of
Phonemes and their realizations 55
'grave', as against 'acute', of Jakobson, Fant and Halle (1967: 30). The feature of
'peripheral' plays an important role in setting up natural classes of phonemes; see
2.2.6.3-[2] and 2.5.
2.1.2. Minimal pairs/sets
Minimal or near-minimal pairs/sets for consonants and sernivowels are given.
[1] Intervocalically (1)- in terms of manner
(a) stops /b, d J, g/
[baba] [bata] [bad,3a] [baga]
lbaba/ lbada/ /bajal lbagal
'spear (IMP)' 'downstream' 'bite (IMP)' 'dig (IMP)'
(b) nasals /m, n, Jl, I]/
[gama] [kana] [gapa] [kaiJalegan]
/gama/ /ganal /gapa/ /kaiJaliganl
'dance type' 'below' 'hut' 'hairy being' (1.5.8-[8])
See 2.13.1 for the etymology of /gapal 'hut'. The difference between !nil and .J!il is
difficult to detect, but it is shown by the following pair:
[I]ani] [IJapi]
/I]ani/ /IJapi/
'what' 'face'
(c) liquids /r, l, V
[maca] [ma.r.a] [malan]
/maca! /mrua/ /malanl
'vine poison' 'hand' 'creek'
(d) semi-vowels /w, j/
[kawa] [gaja]
lgawa/ /gaja/
'stuffvomited' 'father'
[2] Intervocalically (2)- in terms of place
(a) bilabials lb, m/
[baba] [bama]
/babal /hamal
'spear (IMP)' 'man'
(b) apico-alveolars /d, n, r, V
[bata] [banan]
/bada/ /bananl
'downstream' 'break (NF)'
[maca]
/maca/
'vine poison'
[malan]
/malanl
'creek'
56 Phonology
(c) !amino-palatals IJ, Jl, j/
[kalfa] [gaJla] [gaja]
/gajal /gaJla/ /gaja/
'head' 'hut' 'father'
(d) dorso-velars /g, IJ, w/
[mogan] [moiJan]
/mugan/ lmUIJanl
'get (NF)' 'hill'

/muwajljina]
'white chalk'
[3] Word-initially (1)- in terms of manner
(a) stops lb, d, j, g/
[baba] [daba] [kaban]
/babal /dabal /jabali/ /gaban/
'spear (IMP)' 'black palm' 'whip tail kangaroo' 'grub sp.'
(b) nasals /m, n, Jl, IJ1
[malba] [nali]in] [Jlalmo] [IJaliJa]
/malba/ /nab.]inl
/IJab.Jal
'dance type' 'shake (NF)' 'fighting stick' 'PROH'
As another set of examples:
[miJla] [newan]
/miJlal /niwan/
'meat' 'louse egg'
[Jlinan] [IJLJ.iJl]
/jlinan/ /I]i.Iijll
'sit (NF)' 'little finger'
Compare [ne] !nil in [newan] and [Jli] /pi/ in [Jlinan]. (They are in the word-initial
position.) In these words, Iii is [e] when preceded by In/ ([newan]), while it is [i]
when preceded by /pi ([p.inan]). (This phonetic difference in Iii provides a clue for
distinguishing /nil and However, as seen in [1]-(b), in [IJani] /IJanil 'what' and
[IJajli] h.JaJlil 'face', Iii is invariably [i]: [ni] and [Jli]. (They are in a word-medial
position). This phonetic difference regarding Iii may possibly due to the difference
in the position in which it occurs, i.e. word-initial as against word-medial.
(c) liquids /r, J, 1 I
An example of /J! is [.J.era] !Jira/ 'tooth'. /r/ and /11 do not occur word-initially (see
2.2.5.4).
(d) semivowels /w, jl
[walfa] Ualfa]
/wajal /jaja/
'crow' 'later on'
[4] Word-initially (2)- in terms of place
(a) bilabials /b, m/
[hagan] [makan]
/hagan/ /magan/
'dig (NF)' 'false'
(b) apico-alveolars /d, n, r, V
[dodam] [nopa]
/dudam/ /nuba/
'wild peanut' 'bark water-carrier'
lr/ and N do not occur word-initially.
(c) !amino-palatals IJ, Jl, j/
[d3awa] [Jlawa] [jawa]
/jawa/ /p.awa/ /jawa/
'mouth' 'NEG' '(placename)'
(d) dorso-velars /g, IJ, w/
[gama] [IJama] [wama]
/gama/ /IJamal /wama/
'dance type' 'handle of shield' 'mad'
[5] Word-finally (1)- in terms of manner
(a) stops /b, d, j, g/
Stops do not occur word-finally (2.2.5.4).
(b) nasals /m, n, Jl, IJ1
[d3alamd3alam] [d3alan] [d3alaJ1]
/jalam.Jalam/ /jalan/ /jalap/
'big waves' 'greedy' 'tongue'
II]! does not occur word-finally.
(c) liquids /c, l, V
[bakor] [bako.I]
/bagur/ /bagw/
'short' 'fighting stick'
(d) semivowels /w, jl
[makol]
/magul/
'work (noun)'
Phonemes and their realizations 51
An example of /j/ is [bar.J9ai] lbaiJgaj/ 'spear' lw/ does not occur word-finally
(2.2.5.4).
[6] Word-finally (2) - in terms of place
(a) bilabials /b, m/
58 Phonology
An example of /m/ is [dodam] /dudam/ 'wild peanut'. Stops (e.g. /bf) do not occur
word-finally.
(b) apico-alveolars /d, n, r, V
[mad3on] [bakor] [makol]
/majtml /bagur/ /magul/
'quiet' 'short' 'work (noun)'
Stops (e.g. /dl) do not occur word-finally.
(c) lamino-palatals /j, J1, j/
[d3alap] [bai]gai]
/jalap/ lbaiJgaj/
'tongue' 'spear (noun)'
Stops (e.g./jf) do not occur word-finally.
(d) dorso-velars /g, IJ, w/
Stops (e.g. lgf), 11]1 and /w/ do not occur word-finally.
Minimal (or semi-minimal) sets of vowel phonemes (ji. a, u/) are given below.
The long vowel /a:/ will be treated separately.
(a) Word-medially
[gima] [gama]
/gima/ /gama/
'grub' 'dance type'
[goman]
lgrnnanl
'another'
(To be more precise, gima refers to the stage of the growth of a grub in which the
grub is in a cocoon.)
(b) Word-initially
As a rule, vowels do not occur word-initially (2.2.3.3).
(c) Word-finally
[gami]
/gamil
'father's mother'
[gama] [gamo]
/gama/ /gamu/
'dance type' 'water'
The long vowel phoneme /a:/ ([a:]) is attested in six words only (i) /a:/ ([a:])
'exclamation of admiration, surprise, etc.' (4.26), (ii) /ja:/ ([ja:]) 'exclamation of
displeasure, surprise, etc.' (4.26), (iii) /bam! ([bam]) 'whip' (a noun), (iv) II]aua/
([IJa:.ra]) 'a bushy country on Herbert River', (v) /IJali.Ja:l ([IJali.Ja:]) an onomatopoeic
expression to describe the way a baby cries, and (vi) a loan from English: /ja:r/
([ja:r]) 'yard'. There is one minimal pair: h]a:Jal '(placenarne)' and fi]aJal ([IJqa])
'it is not easy to do' (adverb) (see 4.24-[10]). There is one near-minimal pair: /bam!
'whip' and lbananl ([banan]) 'break (NF)'.
Phonemes and their realizations 59
Phonetically, [i:] occurs, but it is interpreted not as /iii, but as fiji/ or as /jil. See
2.1.5.1.2. Similarly, [o:] occurs, but it is interpreted as /uwu/, and not as /uu/. See
2.1.5.2.
The status of /a:! as a separate phoneme is weak, for four reasons. (i) It is attested
in six words only. (ii) /I]a:IJa:l is onomatopoeic. /bam! 'whip', too, may be onomato-
poeic: the cracking sound of a whip. Onomatopoeia is marginal in the phonology
of a given language, and it may be justifiably excluded from its phoneme inven-
tory. (iii) Warrgamay, east of Warrongo, shows contrastive vowel length (Dixon
1981: 16-17): /i, i:, a, a:, u, u:! The location of the place /I]aua/ is not known. But,
since it is on Herbert River, it may be in Warrgamay territory, and if that is the
case, the name /I]aua/ may be a Warrgamay word, and not a Warrongo word. (This
placename is not listed in Dixon's (1981) Warrgamay vocabulary.) (iv) /ja:r/ ([ja:r])
'yard' is a loan from English, and loan words are not central in the phonology of a
given language. (See 2.13.2 for loans from English.)
2.1.3. Allophones of consonants and sem ivowels
Presentation of examples employs the practical orthography in Chapters 1, 3 and 4,
and phonetic and phonemic notations (both using IPA symbols) in Chapter 2. For
readers, particularly for those people who participate in the Warrongo language
revival movement (see 1.9), it is important to be able to know the phonetic value
of a given Warrongo form. However, it is very difficult to accurately describe the
distribution of allophones of phonemes, and it will often virtually impossible for
these people to know the phonetic value of a given form. In view of this, the fol-
lowing discussion of the allophones of phonemes will be as detailed as possible -
to the extent that my knowledge of the Warrongo phonology permits.
2.1.3.1. Nasals, rhotics, lateral, and semivowels
The nasal (/m, n, JI, Ij/), rhotic (/r,J/), lateral (/1/), and semivowel (/j, w/) phonemes
are always voiced. Examples were given in 2.1.2.
lrl is an alveolar tap, e.g. [baro] /baru/ 'bent, crooked', and [bakor] /bagur/
'short' Unlike /r/ of Djaru (Tsunoda 1981a: 25), it is not accompanied by a frica-
tive-like sound.
Ill has two allophones: a retroflex approximant([.[]) and a retroflex tap ([r]).
[.(] is the predominant allophone. ill is always[.(] word-initially, e.g. [.J.aba] !Jabal
'fork', and inter-vocalically, e.g. [ba.J.o] /barul 'lower part of the back'. Word-finally,
and as the first member of an (inter-vocalic) consonant cluster, it is[.(] (generally)
or [r] (infrequently), e.g. (i) [ba.J.IJan] [bat1Jan] /baJI]anl 'waterrat', (ii) [bao.(] [baor]
/baww/ 'rock wallaby', (iii) [bi.u.Jga] [bit1]ga] lbill]ga/ 'grey hair', (iv) [winga.J.]
[wingar] /wingaJ! 'fish'. (Alternatively, it may be possible to say that Ill is always
60 Phonology
[.(1 syllable-initially, and that it is it is [.(1 (generally) or [r] (infrequently) syllable-
finally.) However, [r] does not always occur in these environments. Thus, /bilgu/
'wife' was always(?) pronounced [bi.J.go], and not [birgo]. It is possible that a real-
ization of !JJ as [r] is 'lexeme-specific' (Lass [1984] 1991: 310).
There is a word that contains both /J! and /r/: [.f. era] /lira! 'tooth'.
The lateral (/1/) is alveolar. It is always a 'clear 1', and never a 'dark r -in contrast
with English.
2.1.3.2. Stops
Voicing is not distinctive for stops, and both voiced and voiceless allophones
occur, as in many other Australian languages. It is one of the basic assumptions
in phonology that the distribution of allophones of a given phoneme is predictable.
Indeed, many grammars of Australian languages provide rather simple rules for
stop voicing. See, e.g., Peter Austin (1981: 16), Blake (1979: 15), Breen (1981: 24),
Dixon (1977b: 32), Donaldson (1980: 21), Evans (1995: 52), and Patz (2002: 20).
However, as is the case with Djaru (Tsunoda 1981a: 24) and Gooniyandi (McGregor
1990: 49), it has proved extremely difficult to describe precisely the conditioning
factors of stop voicing in Warrongo. Possibly they are 'at least partly lexeme-specific'
(Lass 1991: 310). (Dixon (1991b: 356) on Mbabaram makes a similar observation
regarding stop voicing under certain conditions.) What follows is merely a state-
ment of the general tendencies that I have noticed.
Regarding phonetic realizations of stops, there is one word that needs a special
mention; it is the noun that means 'shingle back lizard' and that may be tentatively
phonemicized as /buifljull. The stop ljl is realized as in [boipgjo1], i.e. [g] with an
[i]-like off-glide. Unlike the other allophones of ljl (affricate [d3] or [1fJ), it sounds
a little like the stop [j].
2.1.3.2.1. Stop voicing (1): in terms of places of articulation
[1] General tendency
As a general tendency, the more front a stop is, the more likely it is to be voiced,
and the more back a stop is, the more likely it is to be voiceless; see (2-4). (This
tendency seems to have to do with the space above the glottis during the articula-
tion of the stops; see Ladefoged (1975: 127). That is, the more front the stop is, i.e.
the larger the space above the glottis is, the more likely it is to be voiced. And the
more back the stop is, i.e. the smaller this space is, the more likely the stop is to be
voiceless. Dixon (1991b: 355) notes for Mbabaram: /b/ is more likely to be voiced
than/g/.)
Phonemes and their realizations 61
(2-4) /b/ /dl /g/
more likely voiced more likely voiceless
[2] Position within a word
Word-initially, /b/ is almost always voiced (e.g. 2.1.2-[1]-(a)), but there are at
least two exceptions, in which it is voiceless: (i) [pirgil] /birgill and [pirgiba.J.a]
/birgibaJa/ both 'cold weather', and (ii) the loan [pigipigi] /bigibigil 'pig'. /d/, too,
is always voiced, except for [tokal] /dugall 'fern sp.'; see (d-5) of Table 2-4. !JI and
/g/ each have voiced and voiceless allophones.
Inter-vocalically (i.e. word-medially), all the stop phonemes have voiced and
voiceless allophones, and both are fairly frequent. See Table 2-3. (In Table 2-3, a
parenthesis means 'not common'.)
(Word-finally, stop phonemes appear to be prohibited. See 2.2.5.4.)
Table 2-3. Voicing tendency of stops
fbi ldl ljl /g/
word-initially
voiced + + + +
voiceless (+) (+) + +
inter-vocalically
voiced + + +
+
voiceless + + + +
2.1.3.2.2. Stop voicing (2): in the word-initial position
In 2.1.3.2.1, we looked at the word-initial position very briefly. Now we shall ex-
amine this position in more detail. See Table 2-4. It is impossible to provide min-
imal sets, and the table contains near-minimal and not-so-minimal sets. Also, the
table is incomplete. In particular, /d/ is infrequent (2.3.1, 2.3.2), and relevant ex-
amples involving it are very difficult to find. Almost the same applies to Table
2-5 through Table 2-10. In the following discussion of stop voicing, the relevant
allophone of a stop is shown in bold face. ('#' indicates a word boundary (Lass
1991: 33).)
62 Phonology
Table 2-4. Voicing of stops in word-initial positions (Cl)
#fbi- #ldl- #ljl- #/g/-
C2 is a nasal.
(a-1) [bimona] [dimanip] [d;siman] [gima]
lbinnma/ /d.imanip/ IJiman/ lgimal
'FZ' 'flea' 'firestick' 'grub'
(a-2) [bama]
[43ama.I]
[gama]
lbama/
I Jarmul
/gama/
'man' 'web' 'dance type'
C2C3 is a nasal-plus-stop cluster.
(a-3) [bongo] [dongon] [d;songoi] [gonga]
lbliD.gul /dliD.gliD./ /jliD.guj/ /gungal
'maggot' 'ripe' 'tendon' 'raw'
(a-4) [bomba] [domban] [d;somban] [kombala]
lbumba/ /dumbanl /jumban/ /gumbala/
'dust' '(placename)' 'touch (NF)' 'grub'
(a-5) [bamba] [d;sambal] [kamba]
lbamba/ /jamball /gamba/
'guts' 'snake' 'cover (IMP)'
(a-6) [bOIJQO] [ d;soJ]gora] [koiJgari]
lbUIJgul ljUIJQUC"al /gUI]9aci/
'knee' 'fish sp.' 'north'
(b): C2 is a semivowel.
(b-1) [bajan] [ d;saja.ti] [gajana]
lbajan/ /jaja.rl/ /gajanal
'sing (NF)' 'horse' 'father'
(b-2) [d;sawa] [kawa]
IJawa/ lgawa/
'mouth' 'matter vomited'
(c): C2 is a liquid.
(c-1) [bari] [darabila] [ d;sarlba.[a] [karl]
lbaril /dacabilal /jaribrual /garil
'stone' 'neck' 'good' 'sun'
(c-2) [ba.tin] [43a.{in] [ka.pl]
/bruin! ljruinl /gaJill
'twist (NF)' 'woomera' 'type ofboomerang'
(c-3) [bolo] [dolobo] [ d;solo] [kolobo]
lbulu/ /dulubul /julul /gulubul
'belly' 'Joyce Palmer' 'hip' 'wind'
---->
Phonemes and their realizations 63
(d): C2 is a stop.
C2 isfb/.
(d-1) [baba] [daba] [d5abale] [kaban]
/babal /dabal /jabali/ /gabanl
'stab (IMP)' 'black palm' 'kangaroo sp.' 'grub sp.'
(d-2) [babina] [d5abino] [kabilgaba]
/babina/ /jabinul /gabilgabal
'MM' 'Roderick Palmer' 'Townsville'
- - -- - - - - - - - - - - - -
C2 is/dl.
(d-3) [dodam] [d5ota] [kota]
/dudam/
IJUdal
/guda/
'wild peanut' 'dilly bag' 'later'
C2 is/g/.
(d-4) [bigoJI] [degon] [d5igobina] [gekobal]
lbigup/ /digllllf /jigubinal /gigubal/
'nail' 'dig (NF)' 'shooting star' 'Rosevale'
(d-5) [bog an] [tokal] [1foka] [koka]
/bug an/ /dugal/ /jugal /guga!
'bushfire' 'femsp.' 'smoke' 'skin'
(d-6) [bag an] [1fakai] [kakal]
/bag an/ /jagaj/ /gag all
'dig (NF)' 'sand goanna' 'big'
(It might look as if [degon] /digu-nf 'dig-NF' in (d-4) is a loan from English.
However, probably this is not the case. There is no other instance of an English
verb root combined with a Warrongo inflectional suffix. See 2.13.2 for loans from
English.)
The syllable structure of these words may be shown as 'C1V1C2V2... or
'C1V1C2C3V .. .' 'Cl' indicates the consonant of the first syllable, and 'V1' the
vowel of the first syllable. And so on.
In (a-4), (c-1), (c-3) and (d-1), among the C1 stops, only /g/ is voiceless, and all
the others are voiced. That is, there is a cut-off point between ljl and /gl. Probably
the same applies to (a-5), (a-6), (b-2), (c-2), (d-2) and (d-3), although these sets are
incomplete.
In (d-5). among the C1 stops, only /b/ is voiced, and all the others are voiceless.
That is, there is a cut-off point between /b/ and /d/.
In (d-6), among the C1 stops, /b/ is voiced, while /j/ and/g/ are voiceless. There
is no suitable example involving /d/ as Cl. The cut-off point may be either between
lbl and /d/, or between ljl and /g/.
64 Phonology
That is, there are at least four sets that have a cut-off point between /j/ and /g/.
(This is shown by four slashes in (2-5).) There is at least one set that has a cut-off
point between /b/ and /d/. (This is shown by one slash in (2-5).)
There is no evidence to set up a cut-off point between /d/ and /j/.
In (a-1), (a-2), (a-3), (b-1) and (d-4), all the C1 stops are voiced, and there is no
cut-off point. (But (a-2) and (b-2) are incomplete.)
Note that, while no cut-off point can be set up for (a-3), there is a cut-off point
in (a-4) (between /j/ and /g/) despite the similarity between the two sets in terms of
the phonemic environments. The reason for this difference is not known.
These facts can be summarized as in (2-5).
(2-5) C1 stop: /b/ I /d, jl /Ill /g/
more likely voiced more likely voiceless
Consider Table 2-4 again. In (a) (which involves a C2 nasal or a C2C3 nasal-
plus-stop cluster), in (b) (where C2 is a semivowel), and in (c) (where C2 is a liquid,
i.e. a rhotic or a lateral), if there is a cut-off point, it lies between /j/ and /g/. In con-
trast, in (d) (where C2 is a stop), if there is a cut-off point, it lies between /j/ and /g/
(e.g. (d-1) and possibly (d-2) and (d-3)) or between /b/ and /d/ (i.e. (d-5) and possibly
(d-6)). That is, when C2 is a nasal, semivowel or liquid, it is only /g/ that may be
voiceless. In contrast, when C2 is a stop, the cut-off point moves towards the front
and, at least in (d-5), /d/ and ljl as well as /g/ are voiceless.
The above may be restated as follows: C1 stop is more likely to be voiced when
C2 is a nasal, semivowel or liquid than when C2 is a stop, and inversely, C1 is
more likely to be voiceless when C2 is a stop than when C2 is a nasal, semivowel
or a liquid.
(2-6) C2: nasal, semivowel, liquid stop

C1 stop: more likely voiced more likely voiceless
Phonemes and their realizations 65
2.1.3.2.3. Stop voicing {3): in the second syllable
We now turn to stops used intervocalically as C2, in the second syllable. See Table
2-5. The situation is not so neat as that in Table 2-4, which concerns Cl stops.
Table 2-5. Voicing of stops in the second syllables intervocalically (C2)
-fbi- -!d!- -ljl- -/g/-
(a): Cl is a nasal.
---- -------------------------------
(a-1) [maba]
/maba/
'milky pine'
(b): Cl is a semivowel.
[macf5an]
/rna jan/
'corky apple tree'
..................... . ................................................................... .
(b-1) [waban]
(b-2)
/waban/
'look for (NF)'
[jabala]
/jabala/
'fat (not thin)'
(c): Cl is a stop.
Cl is lbl
(c-1)
(c-2)
Cl is /g/.
(c-3)
(c-4)
(c-5)
(c-6)
(c-7)
[baba]
/baba/
'stab (IMP)'
[babina]
/babina/
'MM'
[kaban]
/gaban/
'grub'
[kaba]
/gaba/
'white paint'
[koba.I]
/guba.I!
'(placename )'
[kopo]
/gubu/
'leaf'
[geba]
/giba/
'scratch (IMP)'
[wadalen]
/wadalin/
'nm(NF)'
[jada.I] ('?)
/jadaJ/
'stingray'
[wa1fa]
/waja/
'crow'
[ja1fa]
/jajal
'later'
[bata] [bacf5a]
/bada/ lbaja/
'downstream' 'bite (IMP)'
[batiga.ran]
lbadigaran/ lbaji]ljilal
'cry (ITER-NF)' 'bird sp.'
[kadala]
/gadala/
'dry'
[kata]
/gata/
'possibly'
[kota]
/guda/
'later'
[geto]
/gidu/
'cold'
[ka1fan]
/gajan/
'big grass'
[ka1fa]
/gaja/
'head'
[kotfa]
/guja/
'nose'
[kecf5oroi]
I gi j1111lj/
'centipede'
[makan]
/mag an/
'false'
[wagan]
/wag an/
'rise (NF)'
[jakai]
/jagajl
'Ouch!'
[baga]
/bag a/
'dig (IMP)'
[kakal]
/gag all
'big'
[koka]
/guga/
'skin'
[koko]
/gugu/
'language'
[gekobal]
/gigubal I
'Rosevale'
66 Phonology
(It is not definitely certain that the stop in /jada.II 'stingray' in (b-2) was voiced
([d]).) Cut-off points can be recognized as follows.
(a) Between /dl and /j/ in (b-2) (not definitely), and (c-3).
(b) Between /b/ and /dl in (c-5) and probably in (c-4) (but (c-4) is incomplete).
There is no evidence to set up a cut-off point between IJI (and other stops) and
/g/. It may look as if (a-1) provides evidence for this differentiation, but this is not
conclusive. (yVe shall return to this shortly.)
These facts show a tendency shown in (2-7). Note that (2-7) (which is based
on Table 2-5 and concerns C2 stops) differs from (2-5) (which is based on Table
2-4 and concerns Cl stops) in that, while /j/ is grouped together /dl in (2-5), it is
grouped together with /g/ in (2-7).
(2-7) (b-2) (?), (c-3), (c-4) (?), (c-5):
C2 stop: /b/ /dl lj, g/

more likely voiced more likely voiceless
Table 2-5 differs from Table 2-4 in two important respects: the placement of /dl
and that of /j/.
First, in (c-1) of Table 2-5, among the C2 stops, only /dl is voiceless. This sug-
gests the relative position of the stops as shown in (2-8). In (c-7) of Table 2-5,
among the C2 stops, /g/ as well as /dl are voiceless, and the others are voiced. This
suggests (2-9).
(2-8) (c-1):
C2 stop: /b, j, g/ ldl
voiced voiceless
(2-9) (c-7):
C2 stop: /b, j/ /d, g/
voiced voiceless
It is not known if (c-2) behaves like (c-1) or (c-7). As noted above, it may look as
if (a-1) provides evidence for the cut-off point between /j/ (and other stops) and /g/.
However, it is possible that (a-1) behaves like (c-7), and in that case (a-1) does not
supply this evidence.
Other examples in which a C2/dl is voiceless, include (i) [dotegaran] /dudigacan/
'swing (ITER-NF)', (ii) [d3ito] /jidu/ 'torch', (iii) [kota] /guda/ 'later', (iv) [gote] /gudi/
'back (Adv)', and (v) [jutin] /judin/ 'swim (NF)'
Second, in (b-1), among the C2 stops, only ljl is voiceless. This suggests (2-10).
Phonemes and their realizations 61
(2-10) (b-1):
C2 stop: /b, d, g/
voiced voiceless
The placement of /d/ in (2-8) and (2-9) and that of !JI in (2-10), regarding voicing,
deviate from the general tendency shown in (2-4). The reason for this deviation is
not known. (But see (2-11.)
2.1.3.2.4. Stop voicing (4): effect ofC2 nasal on Cl stop
As seen in (2-6), when C2 is a nasal, a semivowel or a liquid, the Cl stop is more
likely to be voiced than when the C2 is a stop. Indeed, when C2 is a nasal, the
Cl stop is voiced in perhaps the vast majority of cases. See (a) of Table 2-6, and
also 2.1.2-[1]-(b). There are, however, exceptions, in which the Cl stop is voiceless.
They are of two types.
(a) The C2 nasal is lrj/. There is just one example: [kaiJa1egan] /gaiJaliganl 'hairy
being' (cf. 1.5.8-[8]). See (b) of Table 2-6.
(b) The C2 nasal is /n/. There are a fair number of examples, e.g. [kana] /gana/
'below', and [kani] /gani/ 'far' See (b) and (c) of Table 2-6.
Table 2-6. Effect of C2 nasal on Cl stop
-lml- -In!- -/pi- -/zy-
(a) [gom.an] [gona] [gopaga]
/guman! lgllll.a! /gup.agal
'other' 'faeces' 'bird sp. (?)'
(b) [gama] [kana] [gap.a] [kaJJalegan]
/gama/ /gana/ /gap.a/ lgaJJaligan/
'dance' 'below' 'lrut' 'hairy being'
(c) [gami] [kani.] [d5apin]
/gamil /gani/ /japin/
'FM' 'far' 'wake up (NF)'
The reason for the voicelessness of the Cl stop /g/ of [kai.Jalegan] /gai.Jaligan/
is not known - except to note that /g/ is more likely voiceless than the other stops.
See (2-4), among others.
In the examples of (b), it looks as if the C2 nasal /n/ has a 'de-voicing' effect on
the Cl stop /g/ - an effect that may be termed 'anticipatory de-voicing' However,
to complicate the matter, in this environment, /g/ is not necessarily voiceless, e.g.
[gona] /guna! 'faeces'. The conditioning factor is not known.
68 Phonology
Additional examples are given. (Note that the Cl stop /g/ may be voiceless
when it is followed by not just In! (C2), but also a /n/-plus-stop cluster (C2C3).)
(i) The Cl /g/ is voiceless: [kanibaca] /ganibaca/ 'dingo', [kanamale] /gana-
malil 'underneath', [kanda.r.oJ /ganda.ru/ 'hairy being' (1.5.8-[8]), [kangabo]
/gangabu/ 'trident spear' (ii) The Cl /g/ is voiced: [ganima.r.a] /ganima.IaJ 'very
far', [ganbama.J.a] /ganbama.Ia/ 'long ago', [gando] /gandu/ 'dog'.
As seen above, in the only relevant example in which the C2 nasal is /rj/, the Cl
/g/ is voiceless (i.e. [kar.Jalegan]). When the C2 nasal is In/, the Cl /g/ is voiceless
in some words, and voiced in others. Also as stated above, when the C2 nasal is 1m/
or /p/, the Cl /g/ is always voiced. These facts suggest (2-11).
(2-11) C2 nasal: /m, pi In/ /rjl

Cl/g/: always voiced voiced or always (?) voiceless
voiceless
To sum up, when C2 is a nasal, the Cl stop is voiced in perhaps the vast ma-
jority of cases. However, in a fair number of examples with the C2/n/ for C2, the
Cl stop lg/ is voiceless.
Now, compare (2-11) with (2-8) and (2-9). (2-8) and (2-9) jointly show that,
when used as a C2 stop, the apico-alveolar /dl is more likely to be voiceless than
the bilabial/hi and the lamino-palatal/j/. (2-11) shows that, when used as C2, the
apico-alveolar /n/ is more likely to 'devoice' the Cl stop /g/ than are the bilabial
1m! and the lamino-palatal /pl. It may not be accidental that both ldl and In/ are
apico-alveolar. It is possible, though by no means certain, that apico-alveolar-ness
facilitates 'stop de-voicing'.
2.1.3.2.5. Stop voicing (5): effect of VI on Cl stop, and ofV2 on C2 stop
The quality of the VI seems to have an effect on the voicing of Cl stops. Consider
Table 2-7. Since /b/used as Cl is almost always voiced (2.1.3.2.2), it is not included
in Table 2-7.
Table 2-7. Effect ofVl on Cl
(a) Cl is /d/.
C2is a stop.
(a-1)
-/if-
[degon]
/di.gun/
'dig(NF)'
-Ia/-
[daba]
/daba/
'black palm'
-lui-
[tokal]
/dug all
'femsp.'
Phonemes and their realizations 69
(b) Cl is /j/.
C2 is a nasal.
(b-1)
ld3ama.tl
[<i3omobQ{o]
ljimanl ljarnaJl /junrubwu/
'firestick' 'web' 'bullock'
(b-2) [d,5ina] [<i3ana] [d3ono]
ljinal ljanal /junul
'foot' '3PL.NOMIERG' 'lust'
C2 is a liquid-plus-stop.
(b-3) [d3algo.(l [1folki]
/jilbaj/ /jalgwl /jUigi/
'knowing' 'meat' 'scrub'
C2 is a stop.
(b-4) [1fakai] [1foka]
/jigubina/ /jagaj/ /juga/
'shooting star' 'sand goanna' 'smoke'
(c) Cl is /g/.
C2 is a nasal.
(c-1) [gima] [gama] [goman]
/gima/ /gama/ /guman/
'grub' 'dance type' 'other
(c-2) [gimbin] [gambi] [kombo]
/gimbinl /gambil /gmnbul
'blow(NF)' 'clothes' 'buttocks'
C2 is a semivowel.
(c-3) [gija] [gaja] [kojo]
/gija/ /gajal /gujul
'hook' 'father' 'bean'
C2 is a stop.
(c-4) [giban] [kaban] [kobad3i]
/giban/ /gaban/ /gubajil
'scratch (NF)' 'grubsp.' 'crocodile'
(c-5) [geto] [kata] [kota]
/gidul /gada/ /gudal
'cold' 'possibly' 'later'
(c-6) [ked3oroi] [katfa] [kotfa]
/gijW"oj/ /gajal /gujal
'centipede' 'head' 'nose'
70 Phonology
Cut-off points can be set up as follows.
(a) Between Iii and /a/ in (b-4), (c-4), (c-5).
(b) Between Ia! and /u/ in (a-1), (b-3), (c-2), (c-3).
The Cl stop is invariably voiced in (b-1), (b-2), (c-1), and invariably voiceless in (c-6),
and no cut-off point can be set up in them. These facts can be shown as in (2-12).
(2-12) VI: Iii Ia! lui
Cl stop: more likely voiced more likely voiceless
Probably this tendency has to do with the space above the glottis during the ar-
ticulation of these vowels (2.1.3.2.1). That is, the more front the VI is, i.e. the larger
the space above the glottis is, the more likely the Cl stop is voiced. And the more
back the VI is, i.e. the smaller this space is, the more likely the Cl stop is voiceless.
Table 2-8. Effect ofV2 on C2
-Iii- -Ia!- -lui-
(a) C2is /b/.
(a-1) [kabilgaba] [kaban] [kabo1]
/gabilgaba/ /gaban/ /gabull
'Townsville' 'grubsp.' 'carpet snake'
(a-2) [kobi] [kobacJ5i] [kopo]
/gubi/ /gubajil /gubu/
'native doctor' 'crocodile' 'leaf'
(b) C2 is /j/.
(t>-1) [kad5i] [kalfa] [kalfo]
/gajil /gaja/ /gajul
'maybe' 'head' 'white tree ant'
(t>-2) [waf.t.la] [watj"agan] [wad5on]
/wajila/ /wajagan/ /wajUD/
'testicles' 'crow' 'cook(NF)'
(c) C2 is lgl.
(c-1) [bigin] [bigal] [bigop]
fbi gin/ /big all /bagu.Jlf
'shield' 'death adder' 'nail'
(c-2) [mogena] [mogan] [moko]
/mugina/ /mug an! /mugu/
'EB' 'get(NF)' 'shin'
Phonemes and their realizations 71
It should be noted in passing that Table 2-7 conforms to (2-6) in that, in the
main, when C2 is a nasal (i.e. (b-1), (b-2), (c-1), (c-2)) or a semivowel (i.e. (c-3)) or a
liquid (i.e. (b-3)), the Cl stop is more likely voiced than when C2 is a stop (i.e. (a-1),
((b-4), (c-4), (c-5), (c-6)).
Just as the VI may influence the voicing of the Cl stop, the V2, too, seems to
influence the voicing of the C2 stop. See Table 2-8. The tendency seems to be in
accordance with (2-12). (But there is at least one set of exceptions. See below.) A
cut-off point can be set up as follows.
(a) Between /i/ and Ia! in (b-1).
(b) Between /a/ and /u/ in (a-2) and (c-2).
The stop in question is voiceless when it is to the right of the cut-off point (as in Table
2-7). C2 is invariably voiced in (a-1) and (c-1), and no cut-off point can be set up.
The set of exceptions occurs in (b-2), where the C2 to the right of the cut-off
point is voiced, and the C2 to its left is voiceless. The reason for the existence of
this exceptional set is not known.
2.1.3.2.6. Stop voicing (6): in consonant clusters
Consonant clusters (2.2.6) can be classified as follows. (They occur word-medially,
and not word-initially or word-finally.)
[1] Nasal-plus-stop: (a-1) homorganic, and (a-2) heterorganic.
[2] /j/-plus-stop (/b/, !J! or /g/).
[3] Liquid (/c/, Ill or /11)-plus-stop.
[4] Liquid (lei, Ill, Ill) or lj/-plus-nasal-plus-stop (the nasal and the stop are homor-
ganic).
[5] /dg/
Each type will be discussed in the following.
[1] Nasal-plus-stop clusters
The stop is always voiced. There is no exception in my data. Nasals are always
voiced (2.1.3.1), and the voicing of the nasal continues throughout the stop. This
applies irrespective of whether the cluster is homorganic, e.g. (2-13), or heteror-
ganic, e.g. (2-14).
(2-13) [jamba]
/jamba/
'camp'
[band an]
/bandan/
'emerge (NF)'
[waJ14Ja]
/WaJlJaf
'where'
[mar.Jga]
/mar.Jga/
'flower'
n Phonology
(2-14) [gonban]
/gunban/
'cut (NF)'
[gong a]
/gunga/
'raw'
[pongol]
/pongoll
'one'
[2] /j/-plus-stop (lbt, ljl or /g/) clusters
Here, too, the stop is always voiced. There is no exception in my data. Sem ivowels
(e.g. /j/) are always voiced (2.1.3.1), and the voicing of the semivowel continues
throughout the stop. (There is no example of /jd/.)
(2-15) [boibon]
lbujbun/
'spring water'
[koi<f5ari]
/gujJaril
'scrub turkey'
[baiga.J.i]
/bajga.Iil
'fig sp.'
Table 2-9. Stop voicing in liquid-plus-stop clusters (1)
liquid-lb/ liquid-/d/ liquid-/j/
stop
voiced + +
voiceless (+) ?
[3] Liquid (/r/, Ill or /11)-plus-stop clusters
liquid-/g/
+
+
Stop voicing in liquid-plus-stop clusters exhibits the tendency shown in Table 2-9.
See Table 2-10 for examples. This tendency in the main conforms to that shown
in (2-4). Note in particular that, in Table 2-10, lb/ is almost always voiced. But
there is one exception: [IJarpan] /I]arban/ 'fast'. It occurs in (2-100). (In Tables
2-9 and 2-10 there is no example involving /d/. There is no cluster of liquid-plus-/d/
within a root (Table 2-18). There is one across a morpheme boundary (Table 2-19):
/dii]gil-dii]gill 'bell'. But it is onomatopoeic, and it is not useful for the present
purpose.)
[4] Liquid (/r/, /:II, Ill) or /j/-plus-nasal-plus-stop
The nasal and the stop are homorganic. Here, too, the stop is always voiced (there
is no exception). The liquids and sem ivowels (e.g. /j/) are always voiced (2.1.3.1),
and their voicing continues throughout the stop.
(2-16) [bo.rmbon]
lbu.rmbun/
'breathe upon
[someone] (NF)'
[k3.ij1<\5 a]
/ga.IJljal
'spit (noun)'
[kali]gan]
/gali]gan/
'fall(NF)'
[d3aimban]
/jajmban/
'find(NF)'
Phonemes and their realizations 13
Table 2-10. Stop voicing in liquid-plus-stop clusters (2)
liqui.d-/b/
(a) Liquid-stop-/a/.
(a-1) The liquid is /r/.
[gorba.[a]
/gurba.xal
'three'
(a-2) The liquid is II/.
[d;5a.[ban]
/ja.Iban!
'fish sp.'
(a-3) The liquid is /1/.
[balban]
lbalban/
'bone'
[kalbin]
/gal bin/
'child'
(b) Liquid-stop-/if.
(b-1) The liquid is /r/.
[bar hi.{ a]
/barbi.Ial
'echidna'
(c) Liquid-stop-lui.
(c-1) The liquid is 11/.
[d;5olbon]
/julbUD!
'spouse'
[tfolki]
/julgil
'scrub'
liquid-/d/ liquid-/j/
[warlf5an]
/war jan/
'raft'
[bald;5i]
lbaljil
'empty'
liqui.d-/g/
[karkai]
/gargaj/
'sparrow hawk'
[d;5a.rnan]
/ja.xgan/
'enter (NF)'
[d;5i.rka]
/ji.Ikal
'semen'
[balgan]
lbalgan/
'hit(NF)'
(jalka]
/jalgal
'road'
[pirgil]
lbirgill
'cold weather'
[ <fisalgo.[]
/jalgw/
'meat'
74 Phonology
[5] /dg/
There is only one instance of a stop-plus-stop cluster: /dg/ (Table 2-19). This occurs
in one of the three forms of the dative of the word /wa.rajil 'one's own'. The stops
are always voiceless, and never voiced.
(2-17) [ wa.y_aetko]
/wa.raji-d-gu/
'one's.own-LINK-DAT'
/dl is a linking interfix. Etymologically, /wa.rajil 'one's own' seems to contain the
comitative case suffix, whose allomorphs are as follows.
IJil [ci3i] following a consonant
/jil [i] following a vowel
It is in view of this that [e] in [wa.y_aetko] is analyzed as /ji! See 2.1.5.1.2-[3]-(b-2)
and3.6.8.
2.1.3.2. 7. Stop voicing (7}: concluding remarks
From 2.1.3.2.1 to 2.1.3.2.6, we have looked at a fair number of tendencies re-
garding stop voicing (a few of which are exceptionless). They work well in some
instances. For example, as seen repeatedly, among the stops, /g/ is the most likely
to be voiceless. Also, as seen in Tables 2-7 and 2-8, and (2-12), among the vowels,
lui is the most likely to make the preceding stop voiceless. On the basis of these,
/gugu/ 'language', for example, will be expected to involve the voiceless allophone:
[k]. Indeed, Alf Palmer pronounced it as [koko]. Also, it was perceived by Rachel
Wilson as involving the voiceless allophone, and was written down by her as 'coco'.
See (1-16) and (1-17). Another example is /gubul [kopo] 'leaf'; even /b/ is voiceless.
However, despite all this, the account given above is far from complete, and it
is still impossible to predict voicing (or de-voicing) of all instances of stops. For
example:
(2-18) [ba<f!an] [ba1faba1faiJan] [ba1fobala]
lbaJanf /baJabaJaiJanf /baJubala/
'bite (NF)' 'finish up (NF)' 'Kirrama Station'
(2-19) [gando]
/gandu/
'dog'
[kanda.{o]
/gandaru/
'imaginary hairy being (1.5.8-[8])'
Factors such as the following may be involved: (i) the length of the word, (ii) the
number of the syllables that follow the stop in question, and (iii) the location of
stress or pitch.
Phonemes and their realizations 15
Degree of voicing varies among Australian languages. For example, Biri
(recorded from, e.g., Harry Johnson and Eddy Barker in the early 1970s) and
Buluguyban (recorded from Reggie Palm Island in the early 1970s) are more
heavily voiced than Warrongo (at least, in Alf Palmer's pronunciation). For in-
stance, as noted in 1.5.3, /gabul/ 'carpet snake' was pronounced [kabol] by Alf
Palmer, and [gabol] by Reggie Palm Island. In contrast, Djaru (Tsunoda 1981a:
24-25) appears to be voiced roughly to the same extent as Warrongo. Thus,
Warrongo has [makan] /mag an/ 'false' and Djaru has [makan] /mag an/ 'morning'.
2.1.4. Allophones of vowels
Vowels have 'clear' and 'pure' articulation, and they are never diphthongized
(cf. Patz 2002: 21). Nor are they pronounced as a schwa, in contrast with certain
vowels of English.
Conditioning factors regarding the allophones of vowels seem less complicated
than those for stop voicing, and they have been worked out more clearly than those
for stop voicing.
Vowel allophones will be generally written in a broad notation, except for the
allophones of Ia! (2.1.4.1). Also, the vowel in question will be shown in bold face
where deemed useful.
2.1.4.1. Ia/
Ia! is generally an unrounded open vowel. It is somewhere between an unrounded
front open vowel [a] and an unrounded back open vowel [a]. (These two allo-
phones are written in a narrow notation.) But there appears to be no suitable IPA
symbol for this vowel, and the IPA symbol for an unrounded front open vowel, i.e.
[a], is used in the present work. (This is a broad notation.) See 2.1.1 and 2.1.2 for
examples. This vowel seems virtually identical to the major allophone of the vowel
phoneme /a/ of Tokyo Japanese, and it is very similar to, but probably slightly
more front than, the vowel of the Australian English but, cut, etc. (i.e. [A] in a
narrow notation).
When /a/ is preceded by a palatal consonant or semivowel (/J, Jl, j/), it is an un-
rounded front (not back) open vowel [a] (in a narrow notation) e.g.:
(2-20) [d3ami] [d3awa] [Jlando] [Jlawa] uaiJa] [jaro]
/jamil /jawa/ /pandu/ /pawal /j aiJal /j aro/
'fat (noun)' 'mouth' 'ignorant' 'NEG' 'mother' 'here'
76 Phonology
2.1.4.2. lui
lui has two allophones: a close-mid [o] and a close [u]. (They do not involve a signifi-
cant lip rounding.) Its phonetic value varies depending on the preceding consonant:
[o] or [u]:
[o] only:
/bW, /mul, /jul.
/dW, IJU!, /gW, /nW, lpul, IIJU/, lrul, /rul, /lui, lwul.
That is, lui is almost always [o], and it is [u] only in exceptional cases. See 2.1.1
and 2.1.2 for examples of [o]. As noted in 2.1.3.2.7, the word /gugu/ (phonetically
[koko]) 'language' was perceived by Rachel Wilson as involving [o], and was
written 'coco'. The exceptional cases will be discussed in 2.1.4.4. (/wu/ is [o], and it
does not contain [w]. See 2.1.5.2.)
Grammars of Australian languages generally do not mention the existence of
the allophone [o] of the vowel phoneme lui. However, in Warrongo, [o] is clearly
the predominant allophone, as in Djaru (Tsunoda 1981a: 26).
2.1.4.3. Iii
Iii has two major allophones: a close [i] and a close-mid [e]. But /i/ is often realized
somewhere between [i] and [e], and it is not easy (and possibly not significant) to
distinguish between them.
A close [i] is more frequent than a close-mid [e]. Recall that lui is almost always
a close-mid [o], and not a close [u]. That is, the phonetic realizations of lui and /i/
are not parallel in terms of height.
The phonetic value of Iii varies depending on the preceding consonant.
[1] [i] only, and never [e]:
[2] [i] or [e]:
ljil, /pi!, /I]i/, /wil.
/bil, /dil, /gil, /mil, /nil, /ril, !Jil, /li/.
see [1] below and 2.1.5.1.2. [3] lji/:
We shall look at [1] and [2] in the following, and [3] separately, in 2.1.5.1.2.
[1] /ji/, /pi/, II]i/, and !wi1
Here, Iii is always [i], and never [e]. There is no instance of [d3e] (or [1fe]), [pe],
[IJe] or [we] in my data. Examples:
(2-21) /jil:
/pi/:
/gil:
/wil:
[d3iba] /jibal 'liver', [wa1f1la]) /wajila/ 'testicles'
[pinan] lpinan! 'sit (NF)'
[IJ4.iJ1] h]i.Iip/ 'little finger'
[winga.J.] lwinga.tl 'fish'
ljil (i.e. the semivowel plus the vowel) is [i], [e] or [i:]. See 2.1.5.1.2. But it is
never fje]. In this respect, it resembles /ji/, /pi!, IIJi/, and lwil.
Phonemes and their realizations 11
IJ/, !pi and /jl are lamino-palatal, and lrj/ and /w/ are dorso-velar. They do not
allow the following /il to be [e]. /g/ is an exception here. It is dorso-velar, but there
are [ge] as well as [gi]. See [2]-(e) below.
[2] /hi/, /di/, /gil, /mil, /nil, /ri/, lli/, /lil
[2]-(a) /lil and /iV
IV has what may be termed 'lowering effect' on Iii that follows (i.e. /lil) or that pre-
cedes (i.e. /ill). This effect is fairly pervasive. We shall look at llil and /ill.
First., /li/. It is almost always [le], rather than [li]. The lowering effect of /11 is
clearly seen in the following minimal pairs:
(2-22) [IJad3i]
/I]aJi!
[IJale]
'mother's father'
/I] alii
'IDU.NOMIERG'
(2-23) [kobi] [gole]
/gubil /gulil
'native doctor' (1.5.8-[7]) 'angry'
But there are a few exceptions that occur under a definable condition; see 2.1.4.4-
[2], -[3].
Second, /ill. It is almost always [ el], rather than [il].
(2-24) [ba.y_in]
/ba.ri-nl
'twist-NF'
[ba.y_elen]
/ba.ri-li-n/
'twist-REFL-NF'
(The morpheme boundaries are included in the phonemic presentations in (2-24).
This is in order to facilitate the comparison.) Note that., as expected, /ba.rilin/ has
[el], and not [il].
There are at least three exceptions. (i) The noun [dii]gildii.Jgil] /dii]gil-dii]giV
'bell'. It is an ideophone, and ideophones may possibly be exempt from this low-
ering effect; see 2.12. (ii) The noun [bilaiJger] lbilaiJgir/ 'blanket'. It is a loan
from English, and it has not been completely adapted to the Warrongo phonology
(2.13.2-[1]-(a)). (iii) No explanation is forthcoming for the third exception: the
placename [waiJgondila] fwaiJgundilru. It is possible, though by no means cer-
tain, that this placename is non-Warrongo (see 2.2.6.1), and that consequently it is
exempt from the lowering rule.
As mentioned in 2.1.4.2, lui is almost always [o]. Therefore, the issue of the
lowering effect of IV is irrelevant.
Furthermore, IV seems to have a 'long distance' effect in certain instances.
Thus, compare:
78 Phonology
(2-25) [kobi]
/gubi/
'native doctor'
[kobembolo]
/gubimbulu/
'native doctor'
[kobembolo] has [be], and not [bi]. The lowering effect of IV seems to have
reached the V2 /if, extending over the intervening /mbu/. (The lowering effect of
IV on the Iii that follows, is observed in Djaru as well, e.g. [kale] /gali/ '(man's)
sister's child'.)
[2]-(b) /ri/ and fir!
lei, too, has a lowering effect. But there are exceptions to it; see (2-31) and (2-32).
Its effect does not seem so forceful as that of IV. We shall look at /ri/ and fir!.
First, fir! This effect is clearly seen in the near-minimal pairs (shown in col-
umns) listed in Table 2-11. The C2 of the words in (b-1) and (b-2) is /r/, i.e. Iii that
precedes /r/, is 'lowered', i.e. [e], and not [i].
Second, /ri! The lowering effect of /r/ seems to be even weaker than is the
case with fir! Indeed, there are a small number of examples of /ri/ in which Iii is
'lowered' to [e], e.g. [ka1foren] /gajorinl Vi 'go past'. However, there are a larger
number of examples in which Iii is not lowered, e.g.:
(2-26) [bari] [d.3aribqa] [d.3ori]
/bari/ /jariba.ra/ ljUCil
'stone' 'good' 'alongside'
Table 2-11. Lowering effect of /r/
(a-1) [bigoJl] [bimona]
/bigup/ lbimuna/
[kari] [koiJgari]
/garil /gUIJgaril
'sun' 'north'
[dimani]l]
/dimanip/
'nail' 'father's sister' 'flea'
(b-1) [berogai] [becawan] [ decabigal]
lbirugaj/ lbicawan/ /dirabigall
'wOIDb' 'talk-RECP-NF' '(placenarne )'
(a-2) [.J).ban]
/.Ii.ban!
'heated stone'
(b-2) [.J.eca]
/.Ii.cal
'tooth'
[gima]
/gima/
'grub'
[gecamai]
/gicamaj/
'Girrarnay'
The 'lowering effect' of consonants (not necessarily liquids) has been reported
for Kuku Yalanji (Patz 2002: 21-22) and Kayardild (Evans 1995: 58-59). According
to Tatsuyuki Mimura (p.c.), r-sounds in Danish and Norwegian have the effect of
vowel lowering.
Phonemes and their realizations 19
[2]-(c) !Ji/ and IW
First, IW. lil appears to be always [i], and IJI does not seem to have a lowering effect.
(2-27) [bi.[O]
/bi.m/
'gully'
[bi.J.go]
/bi.Igu/
'wife'
Second, !Ji/. In most of the instances, /il is [i], and !J./ does not have a lowering
effect. See (c-2) of Table 2-4 for examples. However, there are at least two instances
in which /if is [e], and /J! has a lowering effect; see (2-28). The conditioning factor
is not known. It may, or may not, be relevant that these two examples involve the
applicative suffix lli/. (See 4.14 for the applicative construction.)
(2-28) [rrina.{en] [janiy,en]
/pina-li-nl /j ani -li-n/
'sit-APPL-NF' 'go-APPL-NF'
Needless to say, /il is [e] when it is within the scope of a lowering effect For
example, (i) lowering effect of /11: [ba.J.elen] 'twist-REFL-NF' in (2-24), and (ii)
lowering effect of /r/: [{era] 'tooth' in (b-2) of Table 2-11.
[2]-(d) !nil
Here, /il is almost always [i], e.g. (i) [kani] /gani/ 'far', (ii) [Jlani] /panil 'ground',
(iii) 'what' [IJani] l.ganil 'what' However, lil is [e] when it is within the scope of a
lowering effect. There is at least one instance:
(2-29) [IJoni]
II]Uilil
'there'
[IJonero]
l!.]uni=ru/
'there=again'
Furthermore, there is at least one instance of [e] that is not within the scope
of a lowering effect (the reason for this is not known); compare [newan] /niwanl
'louse egg' and [Jlinan] i}linanl 'sit (NF)' (2.1.2-[3]-(b)).
[2]-(e) fbi/, /di/, /gi/ and /mil
Here, even where /il is not in the scope of any lowering effect, [ e] as well as [i] are
observed. It is difficult to discern any tendency. Examples are in Table 2-4: (a-1)
and (d-4), among others. Recall ([1] above) that !gil is exceptional in that, despite
its dorso-velar-ness, [e] as well as [i] are observed. Impressionistically, [e] appears
to be more frequent than [i] for /gil, e.g.:
(a) [e] for /gil: (i) [geba] /gibal 'liver', (ii) [mogena] /muginal 'elder brother', (iii)
four examples in (c-7) of Table 2-5.
(b) [i] for /gil: (i) [gima] /gima/ 'grub', (ii) [gija] /gijal 'hook'.
80 Phonology
[2]-(f) Summary oflowering effect
As seen above, the lowering effect is very strong with IV (a liquid), less so with lei
(another liquid), very weak with the third liquid !J./, and apparently non-existent
with the other consonants (i.e. stops and nasals) and the semivowels. That is:
(2-30) Lowering effect:
IV > lrl > !JJ > stops, nasals, semivowels
2.1.4.4. Iii and lui
It is useful to discuss Iii and lui jointly. See Table 2-12. The asterisk means 'not at-
tested'. 'C' stands for a consonant or a semivowel. To be precise, in Table 2-12, C
may be a cluster: CC or CCC. Comments on Table 2-12 follow.
Table 2-12. Iii and/ul
/CiCi/ /Cu. Cui /CiCul /CuCi/
(a) [CiCi] *[Cu.Cu] [CiCu] [Cu.Ci]
[bigin] [bll.J.i]
/big:inl /bujinbucarn/ lbwil
'shield' 'cockroach' 'fire'
(b) [CiCe] *[Cu.Co] [CiCo] [Cu.Ce]
[im.erin] [bigo]l] [bajumberi]
/jimi.rin/ /bigup/ /bajumbiril
'be glad (NF)' 'nail' '(name)'
(c) [CeCil *[CoCu] *[CeCu] [CoCi]
[bajumberl] [kolbila]
/bajumbiril /gulbila/
'(name)' 'south'
(d) [CeCe] [CoCo] [CeCo] [CoCe]
[ba.{elen] [kopo] [ked;5oroi] [mogena]
/ba.Ii -li-nl /gubul /gijuruj/ /mug ina/
'twist-REFL-NF' 'leaf' 'centipede' 'elder brother'
[1] ICuCul
/CuCu/ is always [CoCo], e.g. [kopo] /gubul 'leaf in (d). (In Warrongo, lui is
alrn ost always [ o] (2.1. 4 .2).) The height of the vowel is constant and it is close-mid.
It is never [CuCu] (cf. (a)), [CuCo] (cf. (b)), or [CoCu] (cf. (c)). There are many
other examples: (i) [bolo] lbulul 'belly', (ii) [boiJgo] lbUIJgul 'knee', (iii) [dolobo]
/dulubul 'Joyce Palmer', (iv) [koko] lgugu/ 'language', (v) [moko] lgugu/ 'shin'. As
repeatedly noted, lgugu/ 'language' was perceived as [koko] and written 'coco' by
Rachel Wilson.
Phonemes and their realizations 81
[2] /CiCil
Likewise, /CiCil is predominantly [CiCi], e.g. [bigin] lbiginl in (a), and the height
of the vowel is constant and it is close. There are many other examples: (i) [bibi] I
bibil 'clitoris', (ii) [bi.y,i] lbi.Iil 'the Biri language', (iii) [gimbin] /gimbinl '[wind]
blow (NF)'
Another example of [CiCi] is [d3id9] !JiJil 'a sore'. It is possible to say that !JiJil
is [d3id3i] because /ji/ is never [cf3e] (2.4.1.3-[1]).
Also consider:
(2-31) [cf9li] [bilicf3i]
/jilil /hili Jil
'eye' 'dawn'
Note that, in (2-31), all the occurrences of Iii that precedes or follows IV are [i], and
not [e]. This suggests that the tendency for /CiCil to be [CiCi] outranks the low-
ering effect of /11. These are exceptions to the lowering effect of /11 mentioned in
2.1.4.3-[2]-(a).
There are, however, complications. Consider:
(2-32) [pirgil]
/birgill
'cold weather'
[biri] [birin] [d3igird9gir]
/biril /birinl /jigir--jigir/
'near' 'send (NF)' 'willy wagtail (bird sp.)'
(2-33)
lbijl Jil-bijljili -n/
'feel good (NF)'
(2-34) [imerin]
/jimiri-n/
'be glad (NF)'
(2-35) [bajumberi]
lbajumbiril
[imerimeren] [mericf9n]
/jimir--jimiri-n/ /miri Ji-n/
'be glad (NF)' 'feel itchy (NF)'
[o1meri]
/wulmiril
'(name) (1.5.4.2-[1]) '(one of the sections)' (Table 1-8)
(See 3.11.1.6-[7] and 3.11.1.7-[3], -[7] for the formation of the words in (2-32) and
(2-33), respectively, except for /miriji-n/ Vi 'feel itchy' (L-class).) In (2-31), Iii is
consistently [i]. Here the tendency for /CiCil to be [CiCi] outranks the lowering
effect of /11 or that of /r/ (2.1.4.3-[2]-(b)), and these words are exceptions to these
lowering effects. In contrast, the examples in (2-32) to (2-35) do not consistently
follow this tendency (/CiCi/ [CiCi]). Nor does the lowering effect of Ill or /r/ seem to
operate consistently. These words appear to be exceptions to these lowering effects.
No factor is known that may condition the occurrence of [e] and [i] in them.
The word [birin] /birinl 'send (NF)' was perceived by Rachel Wilson as in-
volving [i], and was written as 'birin'. See (1-16) and (1-17).
82 Phonology
We have said that /CiCi/ is predominantly [CiCi]. Elsewhere, i.e. [CiCe],
[CeCi], and [CeCe], one of the vowels or both vowels are lowered.
/CiCi/ is [CiCe] when the second Iii is lowered. In [imecin] /jimirin/ 'be glad
and smiling (NF)' of (b) of Table 2-12, Iii of /mil is lowered by lei.
/CiCil is [CeCi] when the first Iii is lowered. In [imerin] /jimirin/ of (c), Iii of
/mil is lowered by lei. (But it is not known why /il of lei! is not lowered.)
/CiCil is [CeCe] when the two occurrences of Iii are lowered. In [ba.J,elen]
lba.Ii-li-nl 'twist-REFL-NF' of (d) (of Table 2-12), they are lowered by IV. Here, the
height of the vowel is constant and it is close-mid.
Note that !CuCu! is always [CoCo], whereas /CiCi/ is predominantly, but not
always, [CiCi]. That is, phonetically they are not parallel in terms of vowel height.
[3] /CiCu!
This is generally either [CiCo] (see (b)) or [CeCo] (see (d)). That is, /Cu! is gener-
ally [Co]. (Recall that, in Warrongo, lui is almost always [o].)
Examples of [CiCo] include [bqo] /biru/ 'gully', (ii) [d3ito] /jidu!'torch', (iii)
[ d3igobina] /jigubinal 'shooting star'
There are many examples of [CeCo], and most of them involve [ge] /gil. As
mentioned in 2.1.4.3-[2]-(e), for /gil, [e] appears to be more frequent than [i].
Examples: (i) [ked,3ocoi] /gijuruj/ 'centipede' in (d). (ii) [becogai] /birugaj/
'womb', (iii) [geto] /gidul 'cold', (iv) [gekobal] /gigubal /'Rosevale', (v) [ge.J,oge.J,o]
/gi.mg:i.Iu! 'large intestines (?)'
There is no example of [CeCu] (see (c)).
There are at least four examples of [CiCu] (see (a)). They are [bud3inbucam]
/bujinbucam/ 'cockroach' in (a) and:
(2-36) [bud3ibud3ibin]
lbuji-buji-bi-nl
'fart-fart-INTR-NF'
(2-37) [bud3ilbud3il] [bud3ilbud3ilen]
lbujil-bujill lbujil-bujili-nl
'taking no notice' 'do not care (NF)'
(/Ji! of these words is [d3i] and not [d3e]. That is, the following IV does not lower
the preceding Iii. Recall that ljil is never [ d3e]. See 2.1. 4.3 -[ 1]. This is an exception
to the lowering effect of Ill mentioned in 2.1.4.3-[2]-(a). See 3.11.1.6-[5], -[7] for the
formation of the words in (2-36) and (2-37).)
[4] /CuCil
[CuCi] is not common. Examples include [bu.J,i] lbwi/ 'fire' in (a), and:
(2-38) [mu.J.i] [bundiJl] [bud3inbucam] Uutin]
lmmil /bundip/ /bujinbucam/ /judi-nl
'greedy' 'grasshopper' 'cockroach' 'swim-NF'
Phonemes and their realizations 83
(2-39) [ju.J.i] [ju.J.i:]
/jwi/ /jwi-jil
'kangaroo' 'kangaroo-COM'
(2-40) [bud_9] [bud_9:] [bud_9bud.3ibin]
/buJil /buji-ji/ /buji-buji-bi-n/
'fart' 'fart-COM' 'fart-fart-INTR-NF'
(2-41) [bud_9lbud_91] [bud_9lbud_9len] [bud_91mban]
/bujil-bujill lbujil-bujili-n/ lbujil-mba-n/
'taking no notice' 'do not care (NF)' 'take no notice of (NF)'
(See 3.11.1.6-[2], [5], -[7] for the formation of the words in (2-40) and (2-41).)
[CuCe] is not common. There is only one example, i.e. [bajumberi] lbajumbiril
'(name)' in (b).
[CoCi] is common. There are many examples (recall that lui is almost always
[o]), including [kolbila] /gulbila/ 'south' in (c), and:
(2-42) [bobin]
/bub in/
'pluck (NF)'
[d.3ombi] [d.3o.ri]
/jUIDbil /jUri/
'penis' 'alongside'
(2-43) [kobi] [kod.3ilamila] [koi:]
/gubi/ /gujilamila/ /guji/
'native doctor' '(placename)' 'hungry'
([ d.3ori] IJUCil is an exception to the lowering effect of /r/.)
[CoCe], too, is common. There are many examples, including [mogena]
/mug ina/ 'elder brother' in (d), and:
(2-44) [bogeba] [kobembolo] [gole] [ka1foren]
/bugiba/ /gubimbulul /gulil /gajorinl
'old woman' 'native doctor' 'angry' 'go past (NF)'
([gole] is an example of the lowering effect of /11; see (2-23). [ka1foren] is an ex-
ample of the lowering effect of /r/.)
As noted above, in Warrongo, lui is almost always [o], and not [u]. All the at-
tested examples of lui [u] are listed in (a) to (c) of Table 2-12, and in (2-36) through
(2-41). The conditioning factor for the occurrence of [u] is not known. At least, the
following can be stated regarding the words that contain lui [u].
(a) Most of them consist of, or contain, a root that contains /CuCi/. (IC/ may be a
cluster.)
(b) The first Cis either a labial consonant (lb/ or /m/) or the palatal /j/. It is notal-
veolar or velar.
84 Phonology
Except for [bajwnberi] /bajumbiri/ '(name)' (in which /i/ is lowered by the following
lei and which has [CuCe]):
(c) the second Cis /d/, lj/, Ill or /nd/, i.e. it is not peripheral, and;
(d) the phonetic value of /Cil is [Ci], i.e. /CuCi/ is [CuCi].
2.1. 5. Problems with /j/ and /w/
The analysis of the semivowel phonemes ljl and /w/ is problematic. In particular,
the analysis of /j/ adopted below is inconsistent, and its phonetic value in a given
phonemicization will be sometimes very difficult to know.
2.1.5.1. lj/
2.1. 5.1.1. Introductory notes
/ja! and /ju/ are straightforward. /ja! is [ja], and /ju/ is [jo] or [ju].
(2-45) [jamba] [jomal] [jutin]
/jamba! /jumall /judin/
'camp' 'body' 'swim (NF)'
Problematic are the following two cases. According to the analysis adopted,
they have the following allophones.
(a) /ji/: [i] (frequent), [e] (infrequent), [i:] (infrequent).
(b) /j/C, /j/#: [i] (frequent), [e] (infrequent).
In (b), lj/ is followed by a consonant (/jC/) or by a word boundary (/j#/) (i.e. /j/
in a word-final position). As noted in 2.1.4.3-[1], ljil is never [je]. Nor is it [e:].
Phonetically, there is no [e:] in Warrongo. [e:] occurs in two loan words, and it is
tentatively analyzed as /iji/: (i) [.J.obe:.r] /rubijir/ 'Rosevale', and (ii) [be:ta] lbijid-a!
'bed-LOC' (see 2.13.2-[2]-(d), -(t)).
2.1.5.1.2. Allophones of,(ji/
The distribution of the allophones of /ji/ (except for [e:]), with examples, is shown
in Table 2-13. An asterisk means 'not attested'.
[1] Word-initial position
For a word-initial position, many gmmmars of Australian languages report the ex-
istence of [ji], in addition to [i]. However, in Warrongo, Djaru (Tsunoda 1981a: 33),
Phonemes and their realizations 85
and all other Australian languages I have listened to, in this position phonetically
[ji] does not occur (and not in any position of a word, for that matter), and only [i]
is observed. In the present work, a word-initial [i] is analyzed as /ji/, i.e. ev, rather
than as Iii, i.e. /VI; see (a) of Table 2-13. This is in order for this sound to conform
to the general phonotactic patterns of Warrongo, as shown below. Thus, consider:
(2-46) [inda] '2SG.NOMIERG'
/jinda/ (adopted)
eve ev
linda! (not adopted)
ve ev
(2-47) [igara] 'freshwater lobster'
/jigara/ (adopted)
ev ev ev
/igara/ (not adopted)
vevev
(Syllable boundaries are indicated by an underbar.) An analysis that adopted linda/
and /igarara/ would have to recognize a vowel phoneme in the word-initial posi-
tion. It would also set up the syllable patterns ve and 'V' The syllables ve and
syllable 'V do occur, but they are extremely uncommon (2.2.3.1-[1]). It is in view
of these facts that a word-initial [i] is analyzed as /ji/, and not Iii. For the e, the
homorganic semivowel/j/ has been selected. (Crosslinguistically, syllables starting
with a vowel seem less common than those stating with a consonant; see Jakobson
1971: 526.)
Table 2-13. Allophones of /ji/
(a)
(b)
(c)
word-initially
#[i]-
[inda]
/jinda/
'2SG.NOWERG'
*#[e]
*#[i:]-
word-medially
-[i]-
[ma.Jld3aita]
/mapja-ji-da/
'food-COM-LOC'
-[e]-
[wa.raetko]
/wa.Iaji -d-gu/
'one's.own-LINK-DAT'
-[i:]-
[mai:gan]
/majigan/
'cut(NF)'
word-finally
-[i]#
[map4Jai]
/mapja-ji/
'food-COM'
[bu4Ji:]
lbuji-ji/
'fart-COM'
*-[e]#
-[i:]#
[wai:]
/waji/
'(question marker)'
86 Phonology
[2] Word-final position
Here, /ji/ has two allophones: [i] and [i:].
(a) The allophone [i]
The comitative case suffix (3.6.8) has the following allomorphs.
[ <i3i] /jil following a consonant.
[i] /jil following a vowel.
(2-48) [moran] [mar.J9ai]
/m.uran-jil /mai]ga-jil
'illness-COM 'flower-COM
The allophone [i] /ji/ occurs in the allomorph that follows a vowel. Here, analysis
of [i] as /ji/, and not Iii, sets up a neat morphophonemic alternation between ljil and
/jil; both have the 'CV' pattern. In contrast, recognition of /il, rather than /jil, would
make the alternation asymmetrical; one allomorph would have 'CV', and the other
would have 'V The comitative [i] /jil often occurs in a word-final position.
Now, consider:
(2-49) [bu<i3i:] [ d.3ami:] [jUJ.i:]
/buji-jil /jami-jil /jwi-jil
'fart-COM' 'fat-COM' 'kangaroo-COM
Here, [Ci:] is analyzed as /Ci-jif. These are additional examples of the comitative
[i] /jilin a word-final position.
(b) The allophone [i:]
A word-final [i:] that follows a vowel is analyzed as /jil. (The examples in (2-49)
involve a word-final [i:] that follows a consonant, and not a vowel.) There are at
least five examples. See (2-50) and (2-65). The ex. (2-51) may be considered another
instance. (In (2-50) and (2-65), stress is indicated. A phonetically long vowel is often
(though not always) stressed. See 2.7-[7].)
(2-50) [koii]
/gujil
'hungry'
(2-51) [i:]
/jijil
'Yes!'
[IJai:] [ wai:] [ai:]
/IJajil /wajil /ajil
'Really?' '(question marker)' 'Yes!', 'Well'
Phonemes and their realizations 81
[3] Word-medial position
Here, /jil has three allophones: [i], [e] and [i:].
(a) The allophone [i]
(a-1) The allophone [i] (Part 1). The comitative allomorph [i] /jil (following a
vowel) may occur word-medially, followed by another case suffix. See (a) of
Table 2-13.
(a-2) The allophone [i] (Part 2). Consider (2-52) and (2-53). [oi] is analyzed as fuji!.
(2-52)
(2-53)
[boin] 'eyelash'
/bujin/ (adopted)
cvcvc
[moin] 'stone tomahawk'
/mujin/ (adopted)
cvcvc
lbujnl (not adopted)
cvcc
lmujnl (not adopted)
cvcc
An alternative analysis is to set up a consonant cluster in a word-final position, as
shown above, e.g. /bujnl CVCC However, elsewhere in the language, consonant
clusters are absent word-finally (2.2.6), and this analysis has been avoided.
(a-3) The allophone [i] (Part 3). [i:] plus the preceding consonant, i.e. [Ci:], is ana-
lyzed as /Ciji/. The only example is:
(2-54)
IJalrJgijili/ 'Cameron Hill' (i.e. Mt. Cameron?) (Map 3)
(a-4) The allophone [i] (Part 4). Consider:
(2-55) a. [imerin]
/jimiri-nl
b. [imer-imeren]
/jimir-jimiri-nl
[imerimeren]
/jimir-jimiri-nl
Both words mean 'be glad and smiling (NF)' The reduplicated form was pro-
nounced as [imer-imeren] at least on one occasion; [r] and [i] were pronounced
separately. However, almost always it is pronounced as [imerimeren]; it contains
[ri] at the morpheme boundary. The segments at the morpheme boundary are ana-
lyzed as /r-ji/ (not /ril). See 3.11.1.7-[4], -[5], -[6] for more examples of reduplica-
tion involving /r-jil.
(b) The allophone [e]
(b-1) The allophone [e] (Part 1). Consider:
88 Phonology
(2-56) [ba_rin]
/ba.ri-n/
[ba_relba_rel]
/baiil-ba.IiV
'twist, turn-NF' 'difficult' [e.g. language]
[ba_relen]
/ba.ri-li-nl
'turn-REFL-NF'
[bain]
/baji-nl
'twist-NF'
[baelbael]
/bajil-bajiV
'cunning'
[baelen]
/baji-li-nl
'turn-REFL-NF'
In lbaji-n/, /ji/ is [i]. In /bajil-bajiV and /baji-li-nl, /ji/ is [e] (cf. the lowering effect
of IV (2.1.4.3-[2]-(a)). Nonetheless, it is convenient to set up /ji/ here. This will rec-
ognize an alternation between IJi/ and /j if. See 3.11.1.6 for IV and /li/.
(b-2) The allophone [e] (Part 2). This occurs when the (adjective-like) noun
lwa.raji/ 'one's own' (mentioned in 2.1.3.2.6-[5]) is followed by a (non-zero) case
suffix. Compare:
(2-57) [wa_rai]
lwa.raji/
'one's own'
[wa_raeta]
/wa.raji-da/
'one's own-LOC'
[wa_raetko]
/wa.raji-d-gu/
'one's.own-LINK-DAT'
Note that, when it is not followed by a (non-zero) case suffix, /wa.raji/ ends in [i],
and not [e]. Etymologically, this word seems to contain the comitative suffix /ji/
(3.6.8), and it is in view of this that this [e] is analyzed as /ji/.
Compare [wa_raeta] /wa.raji-da/ 'one's own-LOC' with [mapd3aita] /mapJa-
ji-da/ 'food-COM-LOC' in (a) of Table 2-13, where the vowel in question is [i],
and not [e]. Apparently it is only the putative /ji/ of /wa.rajil that is [e], and not [i],
when followed by a (non-zero) case suffix.
(c) The allophone [i:]
There is only one example:
(2-58) [mai:gan]
/m.ajiganl
'cut(NF)'
Note that, in (2-57), [i:] follows a vowel, i.e. [Vi:] (analyzed as Vljil), while in
(2-53) it follows a consonant, i.e. [Ci:] (analyzed as C/iji/).
2.1.5.1.3. Allophones of /j/ that is not followed by a vowel
We now turn to the allophones of /j/ that is not followed by a vowel, i.e. followed
by a consonant (/j/C) or by a word boundary (/j/#) (i.e. /j/ in a word-final position).
See (a-1) to (c-1) of Table 2-14. For comparison, the allophones of ljl that is fol-
lowed by a vowel are included. See (a-2) to ( c-2).
Comments on, and discussion of, Table 2-14 follow.
Table 2-14. Allophones of /j/
(a-1)
(b-1)
(c-1)
(a-2)
(b-2)
word-medially
-C/aj/C-
[maigan]
/m.ajganf
'tell (NF)'
[baegon]
/bajgun/
'bash(NF)'
-C/uj/C-
[goiban]
/gujban/
'give (NF)'
*-C/ij!C-
-C/ajN-
[baja]
/baja/
'sing (IMP)'
-C/uj/V-
[boja]
/buja/
Phonemes and their realizations 89
word-finally
-C/aj/#
[ba1]9ai]
lba:IJgaj/
'spear'
[<f3olbai]
/julba-j/
'jump-NF'
-Cluj/#
[makoi]
/m.aguj!
'bigger type of carpet snake'
*-Ciijl#
'blow [breath, nose], smoke [tobacco] (IMP)'
(c-2) -C/ij/V-
[gija]
/gija/
'hook'
[1] [Vi] and [Ve] as V/j/
In (a-1) and (b-1), [Vi] and [Ve] are analyzed as V/j/, that is, /aj/ or /uj/. [baegon]
lbajgun/ 'bash (NF)' is one of the two examples available that has [e] for /j/. (The
other example is [koenba] /gujnba/ (a section term') (Table 1-8).)
There is a minimal pair of /j/ and /ji/ word-medially:
(2-59) [maigan]
/majgan/
'tell (NF)'
[mai:gan]
/majigan/
'cut (NF)'
Here, [ai], which involves a short [i], is analyzed as /aj/ (i.e. V/j/), and [ai:], which
involves a long [i:], is analyzed as /aji/ (i.e. Vlji!); cf. (2-58).
Furthermore, there is a minimal pair that shows this opposition word-medially
and word-finally:
90 Phonology
c2-6o) [koikoil
/guj-guj/
'mosquito'
[kof:kof:]
/guji-guji/
'wrinkle'
[2] Word-final /j/ in (a-1) and (a-2)
There are many instances of this - in particular, of -C/ajl#. This /j/ is phonetically
[i]. However, there are two pieces of strong evidence to analyze it as /j/, rather than
Iii or /ji!. They concern the choice ofthe allomorphs of case suffixes (cf. Table 3-2).
We shall use [baiJgai] lbaiJgaj/ 'spear (noun)' as an example.
(a) The dative suffix
The dative suffix for vowel-final stems is [o] /wu/, and that for consonant-final
stems is [go] /gu/ (See 2.1.5.2 for the analysis of /wu/ for [o].) For the dative,
[bai]gai] takes [go] /gu/, and not [o] /wu/.
(2-61) [moiJango]
!mUIJan-gu/
'hill-DAT'
[bai]gaigo]
/baiJgaj-gu/
'spear-DAT'
[bamao]
lbama-wu/
'man-DAT'
(b) The ergative, locative and com itative suffixes
For the vowel-final stems, the allomorphs of the ergative, the locative and the co-
rnitative case suffixes are /I]gu/ [IJgo], II]ga/ [IJga], and /ji/ [i], respectively. See (a)
of Table 2-15. For the /pi-final nouns the relevant allomorphs are IJUI [d3o], ljal
[d.3a], and !Ji/[d.3i], respectively. See (b) of Table 2-15. For [bai]gai], when followed
by the ergative, locative or cornitative suffix, the final vowel [i] is replaced by [p.]
and the relevant allomorphs are the same as those for /pi-final nouns. See (c) of
Table 2-15.
Table 2-15. Evidence for word-final/j/
NOWACC ERG LOC COM
(a) 'man' [bama] [bamaijgo] [bamai]ga] [bamai]
lbarna/ lbarna-l]gul /bama-1J9al /bama-ji/
(b) 'nrussel shell' [gejap] [gejap4Jo] [gejap4Ja] [gejaJ14Ji]
/gijap/ /gijap-jul /gijap-jal /gijap-ji/
(c) 'spear [baiJgai] [baiJgap4Jo] [baiJgap4Ja] [baiJgap4Ji]
fbaiJgaj/ fbaiJgap-jul lbaiJgap-jal
lb81J9aJ1-Jil
The facts presented in (a) and (b) indicate that phonologically the vowel [i] of
[bai]gai] does not function as a vowel. Consequently it is analyzed as /j/
As another example of word-final of /aj/, consider the following placename:
(2-62) (same as (1-5)). As mentioned in 1.5.4.3-[3], its locative form is attested: (2-63).
(2-62) [baiJgoro d3olbai]
/baiJguru julbai/
turtle-NOM jump-NF
'Turtle jumped.'
(2-63) [baiJgoro d3olbad3a]
/baiJguru julba-jal
'turtle jump-LOC'
Phonemes and their realizations 91
In (2-63), the locative allomorph is [d3a] /jal, as is the case with /gijap-jal 'mussel
shell-LOC' and /baiJgap-jal 'spear-LOC' (The root-final /j/ is deleted.) This indi-
cates that [ d3olbai] should not be analyzed as ending in /j/.
Now, compare:
(2-64) [baiJgai]
/baiJgaj/
'spear'
[d3olbai] [mar.J9ai]
/julba-jl lm.aiJga-jil
'jump-NF' 'flower-COM
All of these words involve a word-final [ai]. Its [i] is analyzed as /j/ in the first two
words, while it is analyzed /ji/ in the third word. Note also that [ai] for /aj/ may
cross a morpheme boundary; see [d3olbai] /julba-j/.
[3] [ ai] as /aj/ or /ajil
In [2] above, [ai] is analyzed as /aj/. There is at least one case where [ai] is ana-
lyzed as /ajil, and not as /aj/. Consider:
(2-65) [.{af:]
flajil
'young girl'
(2-66) [{a mJ
!Jajilil
'many young girls'
[-taf:olpd3o]
flaji-wulpjo/
'good-looking girl'
[.[ainba]
!Jajinbal
'many young girls'
(flaji-wulpjo/ occurs in (1-10-c, -f). See 3.7.1-[15] for the suffixes /Iii and /nba/
'many'.) It is possible to analyze [-taili] as flajlil, and [.J.ainba] as !Jajnba/. However, as
is the case in [2], a morphophonological consideration favours !Jajili/ and !Jajinbal,
and [ ai] is analyzed as /aj/. (Note that Iii in !Jajili/ is not lowered despite the presence
of /11. That is, the tendency for /CiCil to be [CiCi] outranks the lowering effect of
Ill (2.1.4.3-[2]-(a)).)
In the phonological analysis of [maiJgai] (cf. (2-48)), [wa-taetko] (cf. (2-56)),
[d3olbai], and also [-taili] and [.J.ainba], information from grammer (and addition-
ally from the etymology in the case of [wa-taetko]) was employed. This indicates
that phonological analysis cannot be separated from grammar. See K. L. Pike
(1973: 115) and Lass (1991: 25, 29, 32).
92 Phonology
[4] -luj/C- and -/uj/# [oi] or [oe]
Phonetically, -/uj/C- and -/uj/# are almost always [oi]; see (b-1) of Table 2-14. (/ul
is predominantly [o], rather than [u] (2.1.4.2).) There are many other examples, e.g.:
(2-67) [boibano] [ci3oimagaran] [koici3ari] [koii]gan]
/bujbano/ !JUjma-g ara-n/ /gujjaril /gujiJganl
'(Alf Palmer's 'crawl-ITER-NF' 'scrub 'female ghost'
MZ's (?)name)' turkey'
(2-68) [bo1ogoiban] [joroimbi] [ked.3oroi]
/bulugujbant /jurujmbil /gijuruj/
'language of Palm Island' 'Yarnanie Creek' 'centipede'
There is, however, at least one exception. The section term /gujnba/ (Table 1-8) is
[koenba], involving [oe]. (This word is one of the two examples that contains [e]
for /j/; see [1] above.) There is no example of [ui] for /uj/.
[5] -C/aj/-, -Cluj!- and -C/ij/-
They occur when they are followed by a vowel; see (a-2) to (c-2) of Table 2-14.
When not followed by a vowel, i.e. (i) when followed by a consonant word-medi-
ally, or (ii) when occurring word-finally, -C/aj/- and -Cluj/- occur; see (a-1) and
(b-1). However, according to the analysis adopted, -C/ij/- does not occur in either
of these two positions; see (c-1). This can be summarized as follows. (An asterisk
indicates 'does not occur'.)
(i) When followed by a vowel: -C/aj/-, -Cluj!-, -C/ij/-
(ii) When not followed by a vowel: -C/aj/-, -Cluj/-, *-C/ij/-
That is, /ij/ does not occur except before a vowel. In this respect, -C/ij/- is excep-
tional, as is the case inDyirbal (Dixon 1970: 665) and Warrgamay (Dixon 1981: 21).
2.1.5.2. /w/
The analysis of /w/ is much easier than that of /j/. /wa/ and /wil are straightforward.
/wa/ is [wa], and lwil is [wi] (never [we]); see 2.1.4.3-[1]).
(2-69) [wama] [winga.()
/wama/ /winga.II
'mad' 'fish'
lwul is invariably [o] (never [u]), unlike /jil (which has three allophones: [i], [e],
[i:]; see 2.1.5.1.1). All the instances of [o] that are not preceded by a consonant
Phonemes and their realizations 93
are analyzed as /wu/. This is in keeping with the general phonotactic patterns of
Warrongo, and it, for instance, avoids a word-initial vowel (as is the case with /ji/;
see 2.1.5.1.2-[1]). For the 'C', the (near-(?)) homorganic semivowel/w/ has been
chosen. Just as [ji] and [je] do not occur, [wu] and [wo] do not occur. Examples of
[o] !wu! follow.
(2-70) Word-initially:
[oka] [oma] [o.J.on]
/wuga/ lwumal lwwunl
'asleep' 'shade' 'beefwood'
(2-71) Word-medially:
[bao.J.] [kaora]
lbawwl /gawural
'rock wallaby' 'love' (noun)
(2-72) Word-finally:
[barao]
lbarawul
'bark water container'
There is a word-final [o] /wul that involves the dative case suffix, e.g. [bamao]
/bama-wul 'man-DAT' in (2-61). Here, [o] is analyzed as !wu/, in parallel with the
alternation of the comitative suffix !Ji!-/jil; see (2-448).
The sequence of 'C[o:]' (i.e. a long [o:] following a consonant) or a word-initial
[o:] occurs in at least three roots. It is analyzed as 'C/uwul':
(2-73) [bo :iJ [ko:iJ
/buwujl /guwuj/
'OK' 'male ghost'
[o:iJ
lwuwuj/
'Hi!'
A parallel [i:] /ijil, following a consonant, is observed, in just one word:
[d3ab:.Jgil{i] !Jab:.Jgiji.ri/ 'Cameron Hill'; see (2-54). That is, phonetically both !uwul
and fiji/ present a long vowel. (However, there is one exception; see (2-77).)
Such a long vowel- [o:] /uwul or [i:] /ijil- may occur across a morpheme
boundary:
(2-74) [gamo:] [d3ami:]
lgamu-wul /jami-ji/
'water-DAT' 'fat-COM'
Also, compare the following near-minimal pair:
(2-75) [ko!i] (stress on [o:])
/guwuj/
'male ghost'
[kol:] (stress on [i:])
/gujil
'hungry'
94 Phonology
The first word contains 'C[o:]' analyzed as C/uwu!.
In passing, consider the following pair of words:
(2-76) [kO:i]
/guwuj/
'male ghost'
[kohJgan]
/guj-1]-gan/
'ghost-LINK-FEM'
Etymologically [koJI]gan] no doubt contains [kO:i] /guwuj/. However, synchronically
it does not contain the long vowel [o:]. Consequently it is best analyzed as /guj-IJ-
g an/, as shown above, rather than as /guwuj -IJ-g an/. (See 3. 7.1-[17] for the fern inine
suffix /gan/)
The only exception to the generalization that phonetically both luwul and /iji/
present a long vowel, is the following word.
(2-77) (kokOOJl]
/guguwup/
'bird sp.'
This word is an ideophone; see 2.12. In Alf Palmer's pronunciation, [kooJl] con-
tains a sequence of two [o], (although there is no pause between them), and not a
long [ko:].
For /j/, the present work recognizes (a) and (b). See (a-1) and (b-2) of Table 2-14.
In contrast, for /w/, it sets up neither (c) nor (d) (indicated by an asterisk).
(a) Word-medial -CV/j/C-, i.e. -C/aj/C- and -C/uj/C- (not *-C/ij/C-).
(b) Word-final-CV!j/#, i.e. -C/aj/# and -Cluj!# (not *-C/ij/#).
(c) Word-medial *-CV/w/C-.
(d) Word-final *-CV/w/#.
I have found no evidence to set up (c) or (d). In this respect, /j/ and /w/ are not
parallel.
It is interesting to note that *-C/ijl- (i.e. *-C/ij/C- and *-C/ij/l:f) behaves:
(i) like *-CV/w/ (i.e. *-CV!w/C- and *-CV!w/#), but,
(ii) unlike -C/aj/- (i.e. -C/aj/C- and -C/aj/#) and -Cluj/ (-C/uj/C- and -Cluj/#)
in that it does not occur either word-medially or word-finally.
2.2. Phonotactics
2.2.1. Structure of words
Roughly speaking, the structure of Warrongo words is as shown in (2-78).
(Enclitics are preceded by an equation symbol ('='), while other morpheme bound-
aries are indicated by a hyphen.)
Phonotactics 95
(2-78) stem (-inflectional suffix) (=enclitic)
That is, a word must contain a stem. A stem may be followed by an inflectional
suffix, and a stem or an inflectional suffix may be followed by an enclitic.
Examples:
(2-79) [jalkaiJgagol]
/jalga-IJga=guV
road-LOC=only
'only on the road'
[IJaruobad3on]
/I]ani-wu=ba .JUI11
what-DAT=DNK'
'I do not know what for.'
To be precise, there are cases that the template in (2-77) does not accommodate:
(i) fusion, as against suffixation, in certain forms of pronouns (Table 3-6), and (ii)
placenames that have a sentential form (1.5.4.3).
Stems are of five types.
(a) A root only, e.g./hamal 'man', /gamu/ 'water'
(b) A root plus a stem-forming suffix, e.g. /guni-laJ 'kill-agent' (1.5.4.2-[2]),
/balga-wal 'hit-RECP' (see 4.12 for the reciprocal construction).
(c) A root plus some other root (i.e. compounding), e.g. IJulaj-JlunguV 'tree-one'
(a placename; see 1.5.4.3-[6]).
(d) A repetition of a root (i.e. reduplication), e.g. /milgan-miigan/ 'hill-hill', i.e.
'many hills, a hilly area'.
(e) A fused form of a pronoun and its case suffix; see 3.3.
Formation of stems of (b), (c) and (d) is discussed in 3.7, 3.8, 3.11.1, and 4.7.
As for inflectional suffixes, they indicate case for nouns, pronouns, and some of
the adverbs (Tables 3-2 to 3-7). For verbs, they show tense or mood, or they may
mark subordination (Table 3-14). Suffixes for voice (e.g. antipassive, reciprocal)
and for aspect (i.e. iterative) are considered as stem-forming suffixes (e.g. Table
3-15). Enclitics are characterized in 2.2.2 and listed in 4.25.
There are a very large number of roots, but suffixes are not numerous, and en-
clitics are even less so -less than a dozen.
2.2.2. Characterization of enclitics
According to a widely accepted view, clitics are intermediate between affixes
(which are bound forms) and words (which are free forms) (see Zwicky 1994). The
present work sets up the category of clitics for Warrongo, to be precise, enclitics.
They are characterized as follows.
96 Phonology
(a) In terms of phonology
(a-1) Enclitics are like suffixes in that there is no pause preceding them. See (2-7(}).
(a-2) Enclitics are like suffixes (and unlike words) in that they do not bear stress
(cf. Zwicky 1994: 572) or pitch. At least., there is no example of an enclitic
with prominent stress or pitch. Again, see (2-79).
(a-3) The shape of some of the suffixes varies depending on the last phoneme
of the stem to which they are added (cf. Zwicky 1994: 572, 574); see, for
instance, Table 3-2 (cases of nouns). In contrast., the phonological shape
of enclitics is invariable. At least there is no putative enclitic whose shape
varies depending on the last phoneme of the stem to which it is added.
(a-4) Enclitics are intermediate between words/roots and suffixes in terms of
syllable structure and syllable types. See Table 2-16.
(b) In terms of distribution
(b-1) Enclitics follow an inflectional suffix if there is one, i.e. they occur word-
finally (cf. Zwicky 1994: 576). Examples include (2-79).
(b-2) Enclitics may appear to be attached to the last word of a phrase, but they
can be considered as attached to the whole phrase, as in English the king
ofEngland's crown (cf. Zwicky 1994: 574).
(b-3) Enclitics may appear to be attached to the first word of a phrase, but they
can be considered as attached to the whole phrase (cf. Zwicky 1994: 574).
(b-4) Enclitics may appear to be attached to the last word of a clause, but they
can be considered as attached to the whole clause.
In the present work, any element that has any one property of (b-1) to (b-2)
is regarded as an enclitic. However, this distinction is not clear-cut., and no doubt
enclitics and suffixes form a continuum. For example, the putative dative case
suffix is considered a suffix when it is attached to a noun, a pronoun, or an adverb
(Tables 3-2 to 3-7), e.g. (2-61), (2-74), (2-7(}). But there are instances where it has
the property of (b-2) (see (4-119)) or that of (b-3) (see (4-120) and (4-121) to (4-123)),
in which case it is regarded as an enclitic. Similarly, /IJUlllaj/ 'from, after' is ana-
lyzed as a suffix (the ablative case suffix) when it is attached to a noun, a pronoun,
or an adverb (Tables 3-2 to 3-7). But it is considered as an enclitic when it has the
property of (b-4); see 4.10.1. (In most of the examples discussed in 4.10, /IJumaj/ is
used as a conjunction, and it may be termed 'an enclitic conjunction'.)
We shall now examine the Warrongo phonotactics in terms of the syllable
structure of roots, enclitics, and suffixes (2. 2. 3), the types of syllables that occur in
roots, enclitics, and suffixes (2.2.4), and the distribution of phonemes in structural
positions in roots, enclitics, and suffixes (2.2.5).
Phonotoctics 91
2.2.3. Syllable structure of roots, suffixes, enclitics, and words
We shall look at the structure of roots, suffixes, enclitics, and words in terms of
mainly the number- and also the types- of syllables that occur in them.
The present work posits the long vowel /a:/ (2.1.1). Languages such as Djaru
(Tsunoda 1981a: 34, 44) yield evidence that /a:/ is phonologically disyllabic.
However, in Warrongo there is no evidence to decide whether this long vowel is
phonologically disyllabic or monosyllabic. As noted in 2.1.2, the status of /a:/ as
a separate phoneme is weak. In view of this, it is tentatively considered monosyl-
labic phonologically.
2.2.3.1. Syllable structure of roots
Attested patterns of roots in terms of syllables are the following. For the purpose of
presenting syllables, e refers to semivowels as well as consonants.
[1] Monosyllabic roots
There are five patterns. Very few examples are attested for each of them.
eve /guj-guj/ [koikoi] 'mosquito' (and also possibly IIJ1ID-IJUI11 [I]OIIIJOn] 'milk'
AlfPalmer suggested that/I]Un-I]UI11 is not Warrongo.)
ev: II]a:-IJa:l [IJa:IJa:] (an ideophone to describe babies' cries)
/ja:/ (interjection) '(exclamation of displeasure, surprise)'
ev:e /ba:n/ 'whip' (cf. (4-895)), /ja:c/ ([ja:c]) 'yard' (cf. (4-735))
ve /aj/ [ai] '(tag question marker)'
V: /a:./ (interjection) '(exclamation of admiration, surprise)'
Tentatively, /guj-guj/, !IJa:-IJa:/, and /IJun-IJUill are considered as reduplicated
forms of a monosyllabic root (/guj/, /IJUil/, /I]a:/), respectively. They exist only in
the forms in which the root is reduplicated (3.7.2-[6]). (The words are dissyllabic.)
lba:nl (mentioned in 2.1.2) may be an ideophone (2.12), describing the cracking
sound of a whip. The variety of English spoken in Queensland has tag questions
that end in something like [ai], and possibly the Warrongo tag question marker /aj/
is a loan from English. /ja:c/ ([ja:c]) 'yard' (mentioned in 2.1.2), too, is a loan from
English (2.13.1-[2]-(d)). Ideophones, interjections and loan words are marginal in
the phonology of a given language. That is, most of the monosyllabic patterns are
marginal in the Warrongo phonology.
[2] Disyllabic roots
The following patterns are attested. There are many examples of them, except for
the last two patterns.
98 Phonology
cvcv
cvcvc
cvccv
cvccvc
cvcccv
cvcccvc
CV:CV
vcv
e.g. /babil 'mother's mother'
e.g. /bagw/ 'fighting stick'
e.g. /bambui 'egg'
e.g. /balban/ 'bone'
e.g. lbalmbuf 'bundle'
e.g. /gUJI]gaV 'husband'
/I]aua/ 'a bushy country on Herbert River'
/aji/ 'Yes' (possibly the only example)
[3] Trisyllabic roots
The following patterns are attested. There are fairly many examples, except for the
last pattern.
CVCVCV e.g.lbadala/ '(placename)'
CVCVCVC e.g. /burubaj/ 'boil [on skin]'
CVCVCCV e.g. /bajundal 'to go past'
CVCVCCVC e.g. /I]UliJljil/ '(placename)'
CVCCVCV e.g. lb31Jgara/ 'blue tongue lizard'
CVCCVCVC e.g. /bii]ga.IaV 'flat'
CVCCVCCV e.g. /burbanba/ 'lizard sp.'
CVCCCVCV e.g. /bab]ga.Ial 'Stony Hill'
VCVCCVC /abirbil/'Herbert Vale' (a loan from English) (2.13.2-[2]-(c))
CVCVCVC '(name of a language)' (1.8.4.1-[2])
The vast majority of verb roots are disyllabic, e.g.lbabal 'spear', lba jal 'bite', lbalga/
'hit', /bajgu/ 'bash', /galiJgal 'fall', and /majga/ 'tell' But there are at least three
(and possibly more; cf. 3.9.1) verb roots that are trisyllabic: /majiga/ Vt 'cut' (see
(2-58)), lbajumbi/ Vt '[fish] wave [its tail]' (1.5.4.2-[1]), and /miriji/ Vi 'feel itchy'
As for nouns, there are many disyllabic roots as well as trisyllabic ones, as
shown above.
[4] Quadrisyllabic roots
The following patterns are attested, with fairly many examples.
cvcvcvcv
cvcvcvcvc
cvcvccvcv
cvcvcvccvc
cvccvcvcv
cvccvcvcvc
cvccvccvcv
cvcccvcvcv
e.g. /bajubala/ 'Kirrama Station'
e.g. /dirabigaV '(placename)'
e.g. /ba.JiJljila/ 'satin bird'
e.g. /biga.I3I]gaV '(placename)'
e.g. /balbaba.Ial 'bundle of spears'
e.g. /balbirigan/ 'shark'
e.g. /bundilbunal 'place near Kirrama'
e.g. /jab]giji.Iil 'Cameron Hill'
Phonotactics 99
[5] Pentasyllabic roots
Two patterns are attested. Only one example is attested for each pattern.
CVCVCVCVCV /gujilamila/ '(placename)'
CVCVCVCCVCV ljirijindwul 'butcher bird'
It is possible that etymologically some of what appear to be trisyllabic or longer
roots do not really consist of just one single root. Thus, lbundilbuna/ 'place near
Kirrama' may contain a putative root /buna/; there is a noun /bunabuna/ 'couch
grass'. As another example, /balbaba.Ia/ 'bundle of spears' may contain the suffi.x
lba.Ial (3.7.1-[8]). Also /guJilamila/ '(placename)' may contain the root /gujilal
'large type of bandicoot' Furthermore, /guj-gujl 'mosquito' may be a reduplica-
tion of the putative root /guj/ (as tentatively decided in [1]), and /gujiguji/'wrinkle'
may be a reduplication of the putative root /guji/ (i.e. /guji-guji/).
Roots longer than quadrisyllabic roots are extremely uncommon, but two in-
stances of pentasyllabic root are attested; see [5]. (But /gujilamilal '(placename)'
may contain the root /gujilal 'large type of bandicoot').
Note the existence of hi-consonantal and tri-consonantal clusters. Bi-
consonantal clusters are attested in [2] disyllabic roots, e.g. /bambui 'egg', [3]
trisyllabic roots, e.g. lbaJUilda/ 'go past', [4] quadrisyllabic roots, e.g. /baJi.Jvila/
'satin bird', and [5] pentasyllabic root, i.e. /jiriJindwu/ 'butcher bird'. In contrast,
tri-consonantal clusters are not attested in quadri- or penta-syllabic roots. They
are attested only in [2] disyllabic roots, e.g. lbalmbul 'bundle', and [3] trisyllabic
roots, e.g. lbalQga.Ial 'Stony Hill'. This difference may possibly have to do with
phonological heaviness. That is, quadri- and penta-syllabic roots are already heavy,
and the increase of their heaviness by the presence of a tri-consonantal cluster is
avoided. (Consonant clusters in roots will be further discussed in 2.2.6.1.)
2.2.3.2. Syllable structure ofsuffi.y;es and enclitics
[1] Suffixes
Nine patterns are attested.
V e.g. lui 'ergative', /a/ 'locative' (Table 3-4)
C e.g. In! 'nonfuture', IV 'nonfuture' (Table 3-14)
CV e.g. /qui 'genitive' (Table 3-2)
CVC e.g. IIJaV 'towards' (3.8.1-[2])
CVCV e.g. /mali/ 'to/on the side of '(3.8.1-[3])
CVCVC e.g. /guwaj/'all my (ofkin)' (3.7.1-[20])
CCV e.g. !IJgul 'ergative', /I]ga/'locative' (Table 3-2)
CCVC one example attested: /Jljar/ 'kin' (3.7.1-[19])
CCCVCV one example attested: /lnbwul (/bUIJgulnburu/ 'on knees' (3.8.1-[10]).
Cf. /bUIJgul 'knee'.)
100 Phonology
[2] Enclitics
Only four patterns are attested.
cv
eve
cvcv
cvcvc
e.g. /ru! 'too' (4.25-[5])
e.g. /gull 'only' (4.25-[4]), e.g. (2-79)
e.g. /guli/ 'only' (4.25-[3])
e.g. /baJUilf 'I do not know' (4.25-[2]), e.g. (2-79)
In terms of syllable structure, roots have the widest range of possibilities (28 pat-
terns attested, excluding onomatopoeic roots, interjections and loans from English),
followed by suffixes (nine patterns), which are in turn followed by enclitics (four
patterns). This parallels the numbers of attested roots, suffixes, and enclitics. As
noted in 2.2.1, there are many roots, but suffixes are far smaller in number, and
enclitics are less than a dozen.
In terms of syllable structure, enclitics are in the main different from roots,
and similar to suffixes. They exhibit four patterns: CV, CVC, CVCV and CVCVC.
Among them, CVCV and CVCVC are attested in roots (2.2.3.1-[2]) and suffixes as
well ([2] above). CV and CVC are attested in suffixes as well, but not in roots.
However, enclitics differ from roots and suffixes in that there is no vowel-ini-
tial enclitic. They all start with CV. Also, like roots and unlike suffixes, there is no
enclitic which starts with a consonant cluster.
Suffixes differ from roots and enclitics as follows.
(a) They may consist of one vowel only (see 'V' above).
(b) They may consist of one consonant only (see 'C' above).
(c) They may start with a consonant cluster (see 'CCV', 'CCVC' and 'CCCVCV').
There is even one (and only one) example of tri-consonantal cluster: /lnbwul
(3.8.1-[10]).
(JM:ost of the facts sated in the preceding two paragraphs are shown in Table 2-16.)
2.2.3.3. Syllable structure of words
The structure of words is shown in (2-78). The syllable structure of words is es-
sentially the same as that of roots (2.2.3.1), except that words are often longer than
roots. It does not seem worthwhile to list all the attested patterns.
Pentasyllabic and hexasyllabic words are not uncommon in texts; see Texts 1
and 2. The longest word attested in my Warrongo data is /gaJIDbujmu-galrn.bujmu-
gaJIDbujmu/ 'grandfather's grandfather's grandfather's grandfather' (Alf Palmer's
gloss), which has nine syllables. (Alf Palmer's gloss for this word contains four oc-
currences of the word 'grandfather' See 3.7.2 and 3.8.1-[4] for the formation and
the meaning of this word.)
Phonotactics 101
The following types of words (and roots) are marginal in the Warrongo pho-
nology.
(a) Words consisting of any one of the roots listed in 2.2.3.1-[1]. They are mono-
syllabic.
(b) Vowel-initial words (2.2.3.1)
(c) Words containing a long vowel /a:/. This long vowel itself is marginal (2.1.2).
Excluding these marginal words (and roots), the syllable structure of Warrongo
words can be shown as follows (cf. Dixon 2002: 554):
(2-80) Syllable structure of words:
CN1Cz(CJ(CA))Vz(C5) where Cz(CJ(CA))Vz can be repeated.
That is, words (and roots) (i) begin with a consonant or semivowel, (ii) are disyl-
labic or longer, (iii) may contain hi-consonantal or tri-consonantal clusters, and
(iv) end in a vowel or a consonant (or a semivowel). This schema covers all the
attested patterns. However, it also generates many patterns that are not attested.
Furthermore, there are restrictions on the occurrence of certain phonemes in cer-
tain structural positions. See 2.2.5.
The words (and roots) referred to in (a) above are monosyllabic, and therefore
they are exceptions to the generalization that words (and roots) are disyllabic or
longer. Note, however, that they each contain a long vowel, and that consequently
they consist of two mora e. They will be no longer exceptions if the generalization
is stated in terms of mora, and not in terms of syllable.
(2-81) Words (and roots) are birnoraic or longer (excluding marginal words and
roots).
That is, there are monosyllabic words, but there appears to be no monomoraic word
(as is the case with Djaru (Tsunoda 1981a: 46)).
It is a common practice in Australian linguistics to describe phonotactics in
terms of syllable. However, the above shows that the concept of mora may be
sometimes more useful than that of syllable.
2.2.4. Syllable types in roots, suffixes, enclitics, and words
We have looked at roots (2.2.3.1), suffixes and enclitics (2.2.3.2) and words (2.2.3.3)
in terms of their syllable structure. We shall now examine the types of syllables
that occur in them.
102 Phonology
[1] Syllable types in roots
The following syllable types are attested.
ev e.g.lba_bi/ 'mother's mother'
eve e.g. /ba_gw/'fighting stick'
ev: e.g.fr.Ja:_la/ '(placename)'
eve lba:nl'whip' (one example only)
ev and eve are common, but ev: and ev:e are extremely uncommon.
There are at least three roots that phonetically begin with [a]. (They were men-
tioned in 2.2.3.1.) Their syllable structure is analyzed as follows.
(2-82) [ai:] [abe.rbEl] [a:]
/ajil /a bir bill /a:!
vev v eve eve V:
- -
'Yes' 'Herbert Vale' '(exclamation of surprise)'
Unlike /ji/ (2.1.5.1.2) and /wu/ (2.1.5.2), there is no suitable semivowel to set up
before [a] (or [a:]). That is, we will have to recognize syllables that consists of a
vowel only: 'V' and 'V:'. However, syllables consisting of a vowel only are marginal
in the Warrongo phonology./abirbiV 'Herbert Vale' is a loan from English, and the
other two words are interjections.
[2] Problem in syllabification
There are tri-consonantal clusters (they are listed in Tables 2-18 and 2-19). They
present a problem for syllabification. (This is due to the principle that 'No syllabi-
fication should yield syllables that are not canonical monosyllables in the language
in question' (Lass 1991: 265). This problem has been noted by Dixon (1972: 274)
for Dyirbal and by Dixon (2002: 556) for Austmlian languages in general.)
Warrongo examples of tri-consonantal cluster include /balmbu/ 'bundle',
/gwQgal/ 'husband', and lbaiQgaJal 'Stony Hill'. For example, lbalmbu/ may be
segmented as lbal_mbu/ or lbalm_bu/. The first analysis yields CCV, a syllable
type not attested elsewhere in the language. The second produces evcc, again
a syllable type not attested elsewhere in the language. (evee does not occur in
Warrongo unless [boin] 'eyelash', for example, is analyzed as lbujnl evcc; see
(2-52).)
As Dixon (2002: 656) admits: 'it can be difficult (or impossible) to segment
such words into syllables on a principled basis'. One alternative is to recog-
nize 'that there are segments that do not belong unambiguously to either of their
flanking syllables' (Lass 1991: 265), i.e. that they are ambisyllabic; see Spencer
(1996: 97-98). According to this view, /m/ in lbabnbu/ is ambisyllabic.
[3] Syllable types in suffixes
The following syllable types are attested.
V e.g. /uJ 'ergative', /a/ 'locative'
C e.g. In/ 'nonfuture'
CV e.g.lrju/ 'genitive'
eve e.g. h:Jall 'towards'
CCV e.g. h]gu/ 'ergative', /I]ga/ 'locative'
CCVC one example attested: IJ!Jar/ 'kin'
Phonotactics 103
CCCV one example attested: llnbwu/ (tbUIJgulnbwu/ 'on knees'. Cf (tbUIJgu/ 'lmee')
Suffixes may start with a consonant, like roots and enclitics. However, as noted
in 2.2.3.2, unlike roots and enclitics, they may
(a) start with a consonant cluster;
(b) consist of just one single vowel, or;
(c) consist of just one consonant.
Consequently, they may deviate from roots and enclitics in terms of syllabification.
(Most of the consonant clusters in suffixes are hi-consonantal, but there is one tri-
consonantal cluster: llnbwuJ:)
(a) Where a suffix starts with a consonant cluster, the preceding root is always
vowel-final (in the examples available), and the first consonant of the cluster is
assigned to the preceding syllable. Examples follow. (Suffixes are preceded by a
hyphen, while an underbar indicates syllable boundaries.)
(2-83) !IJgul (CCV) 'ERG'
tbama-IJgu/
'man-ERG'
lba _ ID3IJ_gu/
cv eve cv
IJ1Jarl (CCVC) 'KIN'
lbabi-Jljar/
':t\.1Jvf-KIN'
lba _ bip. _jar/
cv eve eve
(b) Where a suffix consists of a vowel only, the preceding root is always consonant-
final (in the examples available), and the vowel constitutes a syllable together with
the preceding consonant. Examples:
(2-84) lui 'ERG'
lbabi-J1Jar-u/
':t\.1Jvf-KIN-ERG'
lba _ bi]l_Ja _ rul
cv eve cv cv
- - -
/a/ 'LOC'
tbabi-J1Jar-a/
':t\.1Jvf-KIN-LOC'
lba _ bip. _ja _cal
cv eve cv cv
- - -
104 Phonology
(c) Where a suffix consists of a consonant only, the preceding root is always vowel-
final (in the examples available), and the consonant forms a syllable together with
the preceding vowel. Examples:
(2-85) In! 'nonfuture'
/balga-n/
'hit-NF'
lbal_gan/
eve eve
IV 'nonfuture'
/balga-11
'hit-NF'
/bal_gal/
eve eve
As (2-83) to (2-85) show, syllable boundaries do not always coincide with a
morpheme boundary.
There is a suffix that begins with a tri-consonantal cluster, i.e. /lnburu/ in
lbui]gulnburu/ 'on knees' (/bUI]gu/ 'knee') (2.2.3.2-[1]). Ill will be assigned to the
preceding syllable, while lb/ will be part of the following syllable. Here again, we
are left with In!. It is difficult to decide to which syllable to allocate it. This In! may
be considered ambisyllabic.
Needless to say, there are suffixes that maintain their syllable status intact, that
is, no phoneme is separated from the rest of the suffix in question for the purpose
of syllabification. (Here syllable boundaries coincide with a morpheme boundary.)
Examples:
(2-86) lr]aV 'towards'
!bada-IJaV
'downstream-to'
lba _ da _ IJall
ev ev eve
/mali/ 'on/to the side of
/IJara-m ali/
'above-side'
/I] a _ra _rna _lil
ev ev ev ev
- - -
[4] Syllable types in enclitics
Only two types are attested.
ev e.g. /ru/ 'too'
eve e.g.lba_Junl (eV_eVC) 'I do not know'
These types are attested in roots and suffixes, and they do not deviate from either
of them.
Like roots but unlike suffixes, enclitics maintain the syllable status, e.g. /gull
eve 'only' and /ba JUn! ev _eve 'I do not know' in (2-79).
[5] Syllable types in words
They are the same as those attested in roots.
Phonotactics 105
[6] Comparison of words, roots, enclitics, and suffixes
As seen in 2.2.3.1 to 2.2.3.3 and in [3] above, enclitics and suffixes may consist
of CV only, or CVC only, but words and roots do not (except for the marginal
ones). Suffixes may consist of one vowel only, or one consonant only, but words,
roots and enclitics do not. Suffixes may start with a consonant cluster, but words,
roots and enclitics do not. Suffixes may lose one of their phonemes for the purpose
syllabification, but words, roots and enclitics maintain their syllable status. These
facts are shown in Table 2-16. This clearly shows that, in terms of phonotactics,
enclitics are intermediate between words/roots and suffixes.
Table 2-16. Syllable structure of words, roots, enclitics, and sufflxes
CV,CVC only
one vowel only
one consonant only
initial cluster
syllable status maintained
[7] Vocables
word, root enclitic
+
always always
suffix
+
+
+
+
not always
In songs a syllable with no clear meaning, i.e. 'vocable', is sometimes added to a
word. See 1.6.2.
2.2.5. Distribution of consonants and semivowels
We shall look at the phonotactic possibilities of consonants and semivowels, in
roots (2.2.5.1), suffixes (2.2.5.2), enclitics (2.2.5.3), and words (2.2.5.4). The result
is shown in Table 2-17. Marginal or exceptional phonemes in a given position or
in Warrongo at large are shown in a parenthesis. Some aspects of the phonotactics
of the semivowels were discussed in 2.1.5.1 to 2.1.5.2, and those of the vowels in
2.1.5.1 to 2.1.5.2, and in 2.2.3.1 to 2.2.4.
2.2.5.1. Consonants and semivowels in roots
We shall consider three structural positions: root-initial, intervocalic (and root-
medial), and root-final. Examples were given in 2.1.2.
[1] Root-initially, all of the consonants and semivowels are attested- except for the
(apico-)alveolar liquids: lrl and Ill. Probably, /r/ and /11 are prohibited root-initially,
as in many other Australian languages; see Dixon (1980: 159-167). A root-initial
106 Phonology
[i] is analyzed as /ji/ (2.1.5.1.2), and a root-initial [o] as lwul (2.1.5.2). A root-initial
[a] (or [a:]) is analyzed asIa! (or /a:f). It is marginal (2.2.3.1-[1], -[2], 2.2.4-[1]).
Root-initially consonant clusters are prohibited. Recall that the language name
'Mbabaram' was pronounced [ba:batam] by Alf Palmer (1.4.2.2 and 1.8.4.1-[1]).
[2] Intervocalically, all of the consonants and semivowels are attested. There are
many consonant clusters (some involving a semivowel). They will be discussed in
2.2.6.1 and 2.2.6.2.
[3] Root-finally, the following phonemes are attested: /(d), m, n, J1. r, J, I, j, a, i,
ul. The following phonemes are not attested, and probably they are prohibited: (i)
stops (/d/ has two exceptions), and (ii) the velar sonorants: the nasallrj/, and the
semivowel/w/ (see 2.1.5.2).
The two exceptions involving a root-final /d/ are loan words from English:
/rubijidl 'Rosevale' and /bijidl 'bed' (both mentioned in 2.1.5.1.1).
Table 2-17. Phonemes in structural positions
initial intervocalic final
words b, d, j, g, m., D,Jl, IJ, l, j, w, all consonants and m., n, Jl, .r, l, 1, j,
(a, a1) a, i, u
consonant clusters
roots b, d, j, g, m., D,Jl, IJ, l, j, w, all consonants and (d), Ill, D,Jl, .r, l, 1, j,
(a, a1) a, i, u
consonant clusters
suffixes all consonants, semi- b, .r, l, 1, j, w; n, .r, 1, j, a, i, u
vowels and vowels; no consonant cluster
mb, nd, Jlj, IJ9, nJ, lnb
enclitics b, j, g, Ill, D,Jl, .r, w j, m, 1, w; n,l,a,i,u
no consonant cluster
Compare (i) [.J.obe:r] /rubijir/ 'Rosevale (NOM/ACC)', (ii) [.J.obe:rgo] /rubijir-gu/
'Rosevale-DAT', and (iii) [.J.obe:ll]omai] !Jubijid-IJLUllaj/ 'Rosevale-ABL' (cf. Table
3-4). We can set up the roots /rubijidl as well as /rubijir/. Here, /dl occurs root-
finally. (But it does not occur word-finally.) (The Aboriginal name of this place is
/gigubaV. See (d-4) of Table 2-4.)
The root lbijidl 'bed' occurs in [be:ta] lbijid-al 'bed-LOC' (But [be:t] lbijidl
'bed-NOM/ACC' is not attested.) Again, /dl occurs root-finally. (But it does not
occur word-finally.)
Since these roots are loans from English. the root-final /dl is marginal in the
Warrongo phonology. Also, since !dl is not attested in the final position of these
words (or any other word), the restriction on the final stop is stronger word-finally
than root-finally.
2.2.5.2. Consonants and semivowels in suffixes
[1] Suffix-initial position
The following phonemes are attested.
(a) /a, i., u, b, d, J, g, m, n, p, IJ, c, l, 1, j, w/
(b) /mb, nd, JlJ, IJg, nJ, lnb/
Phonotactics 101
The set (a) consists of all the phonemes except for /a:/, while (b) comprises all the
homorganic nasal-plus-stop clusters, /nj/ and /Inb/. Suffixes may consist of a vowel
only, e.g. (2-84), or a consonant only, e.g. (2-85). In such a case, the suffix-initial
phoneme is also the suffix-final phoneme.
Suffixes differ from words, roots and enclitics in that they may start with:
(a) a vowel, e.g. (2-84);
(b) /11, e.g. /gaja-na-lu/ 'father-KIN-ERG' (fable 3-3) and /wapu-lu/ 'who-ERG'
(Table 3-6), or;
(c) a consonant cluster, e.g. (2-83).
As noted in 2.2.5.1, root-initially (and also word-initially and possibly enclitic-
initially too), Ill does not- and probably cannot- occur.
Unlike words and roots, but like enclitics, suffixes may start with lei, e.g.
/waJlja-cu/ 'where-to' (see 3.8.1-[1] for /ru/).
[2] Intervocalic position
lb, c, l, 1, j, w, ndl are attested. They are much fewer than, and included in, the
possibilities for words and roots. They are slightly more numerous than those for
enclitics. But unlike enclitics, /m/ is not attested intervocalically in suffixes. There
is no example of an intervocalic consonant cluster. (Most of the suffixes are mono-
syllabic (2.2.3.2-[1]) and consequently unable to contain an intervocalic consonant
cluster.)
[3] Suffix-final position
In, c, 1, j, a, i, u/ are attested. (This may be accidental, but the consonants In, c, 1/
are apico-alveolar sonorants.) They are far fewer than, and are included in, the
possibilities for words, roots and suffixes. (Recall that in the case of suffixes that
consist of a vowel only or a consonant only, the suffix-initial phoneme is also the
suffix-final phoneme.) Unlike enclitics, lei is attested suffix-finally, e.g. lpjacl 'kin'
in (2-83).
108 Phonology
2.2.5.3. Consonants and semivowels in enclitics
There are only about ten enclitics (4.25).
[1] Enclitic-initially, /b, J, g, m, n, JI, r, w/ are attested. That is, unlike words and
roots but like suffixes, there is at least one enclitic that starts with /r/: /ru/ 'again,
too'. Like words and roots and unlike suffixes, enclitics do not start with a conso-
nant cluster.
[2] Inter-vocalically, IJ, m, 1, w/ are attested, and the patterns of enclitics are in-
cluded in those of words and roots. There is no example of an intervocalic conso-
nant cluster. (Most of the enclitics are monosyllabic (2. 2.3.2-[2]) and consequently
unable to contain an intervocalic consonant cluster.)
[3] Enclitic-finally, /n, 1, a, i, ul are attested. In this respect, the patterns of enclitics
are included in those of words, roots and suffixes.
2.2.5.4. Consonants and semivowels in words
Word-initially and word-medially, the possibilities are the same as those for the
root-initial and root-medial positions, respectively. As noted in 2.2.5.1-[3], /dl
occurs root-finally, but not word-finally. In this respect, word-final possibilities
are different from (and narrower than) the root-final possibilities. (This indicates
that words and roots need to be distinguished.)
2. 2.6. Consonant clusters
Here, 'consonants' include semivowels. Consonant clusters are not attested word-
initially, word-finally, root-initially, root-finally, enclitic-initially, enclitic-finally
or suffix-finally. But suffix-initially there are two consonant clusters (2.2.3.2-[1]).
There are many consonant clusters in intra-root positions (2.2.6.1) and inter-mor-
phemic positions (2. 2.6.2).
2.2. 6.1. Intra-root consonant clusters
:Most of the intra-root consonant clusters are hi-consonantal, but others are tri-con-
sonantal. See Table 2-18.
Intra-root consonant clusters can be classified as shown in Table 2-18. For each
group in Table 2-18, a generalization about its membership is given, together with
a list of unattested combination(s), where relevant.
Phonotactics 109
Table 2-18. Intra-root consonant clusters
Last member of
the cluster is: bilabial apico-alveolar lamino-palatal dorso-velar
hi-consonantal
(a) mb nd
Jlj
IJ9
(b) rb ij rg
lb lj lg
nb nj ng
(c) rm
riJ
1m ljl
liJ
(d) rw
lw
(e) nrn
DIJ
(f) Jb lj rg
jb jj jg
(g) nn
liJ
jm
iP
jiJ
tri -consonantal
(h) rmb
1Jlj
I1J9
.IIIlb ljlj
liJ9
1mb l]lj lijg
jmb jp.j jiJg
(i) rnb
lnb
jnb
We shall look at hi-consonantal clusters first.
(a) Homorganic nasal-plus-stop clusters: /mb/, lnd/, /p.J/, /Ijg/
All possibilities are attested, e.g. lbambaial 'white', /banda/ 'to come out', /warua/
'where', lbalaggaV 'dugong'.
(b) Apico-alveolar sonorant (/c/, IV or /n/) plus non-apico-alveolar stop (/b/, !JI or /g/)
All possibilities are attested, e.g. /barbila/ 'echidna', /waqan/ 'raft', lbirgil/ 'cold
weather', /balba/ 'to roll', /ba\fi/ 'empty', lbalga/ 'to hit', /banba/ 'fig sp.', lwarua/
'hole', /wangall 'little girl'.
(c) Apico-alveolar liquid (jc/ or IV) plus non-apico-alveolar nasal (/m/, lp.l or /rf)
lcp/ is not attested. Examples: lbucmu/ 'deaf, /bacga/ 'light (not dark)', /gulmi/
'back (Adv)', /IJalflil 'kind (not unkind)', /galga/ 'MB'
(d) Apico-alveolar liquid (jc/ or IV) plus semivowel (/j/ or /w/)
lcj/ and /lj/ are not attested. Examples: IJicwil 'namesake', /walwa/ 'bad'. Dixon
(1972: 287) states that in Dyirbal the clusters /lj/ and /nj/ are prohibited. In
110 Phonology
Warrongo, it is possible to set up /cj/ across a morpheme boundary; see 2.2.6.2-(d).
But within a root, !cj! and /lj/ appear to be prohibited.
(e) Apico-alveolar nasal (In!) plus peripheral nasal (jml or lrjl)
Both possibilities are attested, e.g. /gunma/'to break'. There is only one example of
frn:J/: lwanQaJal 'bony bream'
(f) Retroflex approximant (jJI) or palatal semivowel (jjl) plus non-apico-alveolar
stop (jbl, IJI or lgl)
All possibilities are attested, e.g. 'ice', lgwja/ 'native bee', 'bush
rat', lbujbun/ 'spring water', lgujjaci/ 'scrub turkey', /bajgaJi! 'fig sp.'
(g) Retroflex approximant (/J!) or palatal semivowel /j/ plus non-apico-alveolar
nasal (jml, !pi or lfJ/)
!J:pl is not attested. Examples: '(place on Herbert River, near Cashmere,
towards Cameron Creek)', lba.IQanl 'kangaroo rat', IJujma/ 'to crawl', lmajpall
'dirty thing (e.g. penis)' There is not a good example of lj:gl. One possible example
is lgaJujQa/ 'to hide' Etymologically it may contain the transitive-stem-forming
suffix /gal (4.7.2), but there is no evidence for this etymology, and tentatively
/garujfJa/ is treated as a root.
We shall now look at tri-consonantal clusters.
(h) Liquid (/c!, Ill or Ill) or palatal semivowel lj! plus a non-apico-alveolar nasal-
plus-stop homorganic cluster (jmb, JlJ, I]g/)
All possibilities are attested, e.g. lbucmbul 'to breath upon', lmacJlJal 'sore (noun)',
lwicQga/ 'tail of fish', /ga.tmbujmu/ 'grandfather's grandfather', /bl.LIJl.fu/ 'elbow',
lhiJ:Qga/ 'grey hair', /balmbi/ 'to smell', /ba)Jlja/ 'to come out', lbaiQgaJa/ 'Stony
Creek', !Jajmbaj/ 'hither', /bujJljall'fish smell', lbajQgica/ 'sweat'.
(i) lei, Ill or /jl plus lnb/
It is difficult to generalize about these three clusters. Examples: /mUIJambaJal
'many', /bilnbiran/ 'crimson rosella (?)', /gujnbun/ 'sorry' Etymologically, lrn.UIJarn-
baJa/ 'many' may contain the suffix lbaJa/ (3.7.1-[8]). If this is the case, it is not a
root. (As noted below, it is not certain if there is a contrast between lnbl and lmbl
in tri-consonantal clusters.)
Roots generally do not contain more than one consonant cluster. However, there
is at least one root that contains two consonant clusters: /waggundila/ 'a place on
Herbert River, where there is a big cliff' This may be irrelevant, but both clusters
are homorganic nasal-plus-stop ('N-plus-S') clusters. Since the place is on Herbert
River, it may be in Warrongo territory or in Warrgamay territory. Warrgamay has
waiJgu 'small goanna' (Dixon 1981: 125). It is possible that lwaiJgundila/ contains
two roots, including /waggu/. A root with three clusters is not attested.
Phonotactics 111
When I first heard the word fwai]gundilal in a teA1: (see Text 1, Line 19), on the
24th August 1972, after having worked on Warrongo with Alf Palmer for about
six months, this word sounded very 'foreign', i.e. 'non-Warrongo', to my ears. The
cause for this is not known. At least, it seems certain that the cause is not the
presence of two (not just one) N-plus-S clusters. This is because many Warrongo
words (though not roots) contain two N-plus-S clusters, e.g. /gunda-IJga/ 'night-
LOC' and /jamba-I]gaf 'camp-LOC' The cause may possibly be the location of
stress in this word; stress falls on the third syllable, i.e. [waiJgondila], despite the
general tendency in Warrongo for stress to fall on the word-initial syllable (see
2.7-[2]). (In passing, I note that !waiJgundila/ [waiJgondila] is an exception to the
lowering effect of Ill (2.1.4.3-[2]-(a)).)
[jld3] IJ1jl and [nd3] /nj/ seem to contrast. Thus, AlfPalmer said, 'When you are
lying on top of a woman at night, and if you can't find it, you say':
(2-87) [waJld3a wand3a] [AP]
/waJlJa wanJal
where hole
'Where is the hole [i.e. vagina]?'
Alf Palmer stated to the effect that these two words are distinct. My observation of
his pronunciation (in particular, the position of his tongue) supported his statement.
Another minimal pair is [maJld3a] lrnapJal 'vegetable food' and [mand3a] /manJal
'to send (IMP)'.
Also, lngl and lrjg/ are in opposition, e.g. (i) [bongo] /bungul 'maggot' and
[boiJgo] /bmjgu/ 'knee' (cf. (2-3)), and (ii) [gonga] /gunga/ 'raw' and [koiJgari]
/gUIJg ari/ 'north'.
It is possible that certain nasals in tri-consonantal clusters are neutralized and
do not contrast. (i) In Table 2-18, /lmb/ (e.g. /balmbi-n/ 'smell-NF') and /lnb/ (e.g. I
bilnbi.Ianl 'crimson rosella (?)') are listed separately. This is based on my observa-
tion of Alf Palmer's pronunciation: the closure/non-closure of his lips. However,
it is not certain if phonologically /lmb/ and /lnb/ contrast. (ii) Also /jmb/ and /jnb/
may not contrast. (In my observation of Alf Palmer's pronunciation, the name of
one of the sections (Table 1-8) is [koenba] /gujnbal, and not [koemba] /gujmba!)
These generalizations about the consonant clusters are based on the attested
patterns. There are many gaps in Table 2-18. Some of the unattested patterns may
be genuinely prohibited, while some others may simply accidental 'gaps' in the
language or they may be due to the incompleteness of the data. The same applies
to the generalizations about inter-morphemic clusters (2.2.6.2) and the comparison
ofbi-consonantal and tri-consonantal clusters (2.2.6.3).
112 Phonology
2. 2. 6.2. Inter-morphemic consonant clusters
Attested inter-morphemic clusters are listed in Table 2-19. To facilitate comparison,
they are classified roughly in the same way as in Table 2-18. For each group in
Table 2-19, a generalization about its membership is given, together with a list of
unattested combinations, where relevant.
We shall look at hi-consonantal clusters first.
(a) Homorganic stop-plus-nasal clusters
Examples: !JUJam-bu/ 'locust-ERG', /galbin-dUJ 'child-ERG', /gija.p-Jul 'mussel
shell-ERG' /rj-g/ is not attested. (lrj/ does not occur morpheme-finally (Table 2-17),
and probably /1]-g/ is prohibited. Hence the four dots in Table 2-19.)
(b) Apico-alveolar sonorant (/n, c, Ill) plus stop (/b, d, J, g/)
Examples: /gumun-bil 'quiet-INTR', /bigin-Ji! 'shield-COM', /mUIJan-gu/ 'hill-
DAT', /jamur-bil 'hurry-INTR', lbagur-Jil 'short-COM', /gaJar-gaJacl Adv 'going
past', /gagal-bil 'big-INTR', /dii]gil-dii]gill 'bell' (the only example of /1-d/),
/gagal-Jil 'big-COM', /waiJal-gu/ 'boomerang-DAT' The inter-morphemic ln-J/,
e.g. lbigin-Jil 'shield-COM', presumably involves [nd.3], not (Jl.d.3], phonetically,
although this is not certain. led! is not attested. /nd/ is already listed as a homor-
ganic nasal-plus-stop cluster, and therefore it is parenthesized in (b) of Table 2-19.
(c) Apico-alveolar liquid (lc, 11) plus peripheral nasal (/m, I]/)
All possibilities are attested, e.g. /gulmbur-maJ 'to set up', lbiJir-ga! 'wide-TR',
lbaba-1-mu/ 'stab-NEGIJ\.1P', !Jlungul-ga! 'one-TR'.
(d) Apico-alveolar liquid (lc, 11) plus semivowel (/j, w/)
Examples of /c/ plus /j/: /jamur-jamuri-nl [jamocjamocen] Vi 'hurry', and /jimi-
jimiri-nl [imerimecen] Vi 'be glad' (see (2-34)). (In /jimir-jimiri-n/, /jl is phoneti-
cally almost always zero (2.1.5.1.2-[3]-(a-4)). /c-ji/ is pronounced as one syllable:
[ri]. See 3.11.1.7-[6] for the formation of these verbs.) The other three possibilities
(le-w/, 11-j/, /1-w/) are not attested.
(e) Apico-alveolar nasal (/nl) plus a nasal (/m, n, I]/; !pi is not attested.)
Examples: /milgan-milganl 'hill-hill', /waga-n=na! 'rise-NF=really' (/=na/
'really' is an enclitic; see (4-1018)), /mag an-gal 'false-TR'.
(f) Retroflex approximant (/J/) or palatal semivowel (/j/) plus non-apico-alveolar
stop (/b, J, g/)
All possibilities are attested. Examples: lrnunJw-bi/ 'painful-INTR', lwinga.t-Jil
'fish-COM', /winga.t-gu/ 'fish-DAT', /wamaj-bil 'good-INTR', !Jawuj-Jul 'hot-ERG',
/baiJgaj-gu/ 'spear-DAT'
Phonotactics 113
Table 2-19. Inter-morphemic consonant clusters
The last member of
the cluster is: bilabial apico-alveolar lamino-palatal dorso-velar
hi-consonantal
(a) mb nd
Jlj
(b) nb (nd) flj ng
rb ij rg
lb ld lj lg
(c) em
.ri.J
lm
liJ
(d) .rj
(e) nm nn
Ill]
(f) Jb lj
19
jb jj jg
(g)
liJ
jm
.iJl
jiJ
others illj mg
IDIJ
Jlb
PIJ
dg
diJ
nw
ii
tri -consonantal
(h) nnb
Jmb
1mb
lnb*
jmb
jiJg
(i) jnb
(g) Retroflex approximant (/J.!) or palatal semivowel /j/ plus non-apico-alveolar
nasal (/m, J1, 1]1)
Examples: lbil]gi.I-1Jal 'hot-TR', /gujaj-mali/ 'across-side', /.JUI.aj-punguV 'tree-one'
(a placename; cf. 1.5.4.3-[6]), /gujaj-IJal 'across-TR'. /Irnl and !Ipl are not attested.
'Others'
It is difficult to generalize about these clusters. Examples: (i) /jalam-Jalam/
'wave-wave', (ii) /gwim-gwim/ 'busy and not looking', (iii) /jalam-Jalam-Qal
'wave-wave-TR', (iv) lbuWUJl-buwup/ (Jalngoy?) 'butcher bird', (v) I.JUI.bi.rlJl-Qal
114 Phonology
'skinless-TR', (vi) /wa.Iaji-d-gu/ 'one's own-LINK-DAT' (cf. (2-57)), (vii) /rubijid-
QUID.aj/ 'Rosevale-ABL' (cf. 2.2.5.1-[3]), (viii) /jwaj-jwaj/ 'quiet-quiet'
We now turn to tri-consonantal clusters.
(h) Liquid (/r/, !JJ or Ill) or palatal sem ivowel/j/ plus nasal plus stop
Examples: (i) /gulmbuc-m-baJ 'to set up [a fence]' (see 4.7.3 for the transitive-stem-
forming suffix /m-ba/), (ii) /gujnbw-m-ba/ 'to feel sorry for', (iii) /bujil-m-ba/ 'to
take no notice of (see (2-36) for lbujill), (iv) /baraj-m-barajl 'in the morning', (v)
lbaiJgaj-Q-gu/ 'spear-.y-DAT' There is no example of /1-n-b/ across a morpheme
boundary (indicated by an asterisk). But there is one example of /lnb/ next to a
morpheme boundary: lbUIJgulnbwul 'on knees' (cf.lbUIJgul 'knee') (mentioned in
2. 2.3.2-[1 ]).
(i) Palatal semivowel /j/ plus nasal plus stop
An example: /wamaj-n-bi/ 'good-LINK-INTR'.
2.2.6.3. Comparison and analysis of intra-root and inter-morphemic consonant
clusters
There is only a very small difference, in terms of number, between intra-root clus-
ters (Table 2-18; 48 patterns) and inter-morphemic ones (Table 2-19; 45 patterns).
(It is interesting to note that, in Djaru, inter-morphemic clusters far outnumber
intra-root ones (Tsunoda 1981a: 38-40).) In the main they behave similarly. As
far as the attested clusters are concerned, the following observations can be made.
(A detailed discussion of intra-root and inter-morphemic consonant clusters of
Warrongo is in Tsunoda (2008).)
[1] Tendencies and restrictions
The following tendencies or restrictions apply to both types of clusters, unless
stated otherwise.
(a) Cluster-initially, stops and the semivowel /w/ are not attested.
(b) Cluster-finally, the following phonemes are not attested: the apico-alveolar
nasal In! and liquids lr, J, 1/. The inter-morphemic In-n! is an exception. (To be
precise, /n=n/. See ( 4-1018).)
(c) There is no example of 'nasal+liquid', such as /nl/. (In Djaru, the interfix /gul
has to be inserted between a nasal and a lateral at a morpheme boundary.)
(d) The clusters that contain the apico-alveolar stop /dl are almost completely
absent. There is only one within a root (/ndl), and only two across a morpheme
boundary (ln-dl and /1-d/). (This explains the absence of liquid-plus-/dl clus-
ters within a root (mentioned in 2.1.3.2.6-[3]).) The only example of /1-d/ is I
c:fu]gil-dii]giV 'bell', but it is an ideophone, and consequently it is marginal in
the Warrongo phonology.
Phonotactics 115
(e) Nasal+stop ('N+S') clusters are fairly common. Furthermore, every tri-conso-
nantal cluster contains a N+S cluster.
(f) Among the N+S clusters, homorganic ones are fairly common. (Even the
apico-alveolar stop /d/ occurs here, i.e. /nd/.) Furthermore, they occur in most
of the tri-consonantal clusters. This is no doubt due to an articulatory reason.
(Homorganic N-plus-S clusters appear to be among the most common conso-
nant clusters in some other Australian languages, e.g. Dyirbal (Dixon 1972:
272), and Djaru (Tsunoda 1981a: 38), and this may be the case crosslinguisti-
cally as well.)
There are clusters that are attested within a root only, and also those that are
attested across a morpheme boundary only. The following points are worth men-
tioning.
(g) One intra-root cluster starts with /rjl: /I]g/, e.g. /bUQgu/ 'knee'. (See (a) of
Table 2-18.) But there is no such inter-morphemic one. This is because /rjl
does not, and probably cannot, occur in the final position of roots or suffixes
(Table 2-17).
(h) The following types of clusters are attested across a morpheme boundary
only, and not within a root.
(h-1) The cluster /cj/, e.g. /jarn.ur-jarn.uri/ 'to hurry' (See (d) of Table 2-19.) Within
a root, /cj/ and /lj/ appear to be prohibited (2.2.6.2-(d)).
(h-2) The cluster Inn/, e.g. /wagan-n=na/ 'rise-NF=really' in (4-1018)). (See (e) of
Table 2-19.)
(For (h-3) to (h-7) below, see 'other' of Table 2-19.)
(h-3) The following three clusters that start with /m/, i.e. (i) lmJI (/Jalam-Ja1am/
'wave-wave'), (ii) /mgt (/gwim-gwim/ 'busy and not looking'), and (iii) /m.rjl
(/Jalarn.-jalam-Qal 'wave-wave-TR').
(h-4) The cluster /ph/ (/bu\Vl.ljl-buwup/ (Jalngoy?) 'butcher bird').
(h-5) The two clusters that start with the stop /d/: (i) /dg/ (/wa.Iaji-d-gu/ 'one's own-
LINK-DAT' (3.6.8)), and (ii) /drj/ (/Jubijid-Qumaj/ 'Rosevale-ABL'). They are
the only clusters in the language that start with a stop. (/d/ in /wa.Iaji-d-gu/ is
a linking interfix. It is the only linking interfix that consists of a stop. Recall
also that bona fide Warrongo roots do not end in a stop. IJubijid/ 'Rosevale'
is a loan.) Therefore, /wa.Iaji-d-gu/ contains the only S-plus-S cluster in the
language, and /rubijid-IJumaj/ contains the only S-plus-N cluster, in the lan-
guage.
(h-6) The cluster /nw/ (/wannan-wannan/ 'busy').
(h-7) The cluster /jj/, in which the same phoneme 1s repeated: /jwaj-jwaj/
'quiet-quiet'.
Also, all inter-morphemic tri-consonantal clusters involve nasal insertion, e.g.
/baraj-m-baraj/ 'in the morning'; see 2.6-[2]-(a).
116 Phonology
[2] Natural classes
The various groups of consonants and semivowels listed in 2.2.6.1 (intra-root clus-
ters) and 2.2.6.2 (inter-morphemic ones) may each be considered as natural classes.
Then, intra-root clusters and inter-morphemic ones have almost the same natural
classes. Examples follow, presented mainly in terms of place of articulation.
(a) Apico-alveolar. This is the most recurrent feature: (i) apico-alveolar sonorants
(jr, 1, nJ) (2.2.6.1-(b), 2.2.6.2-(b)), and (ii) apico-alveolar liquids (/r, 11) (2.2.6.1-
(c), -(d), 2.2.6.2-(c), -(d)).
(b) Non-apico-alveolar: (i) non-apico-alveolar stops (/b, J, g/) (2.2.6.1-(b), -(f),
2.2.6.2-(f)), (ii) non-apico-alveolar nasals (/m, JI, IJI) (2.2.6.1-(c), -(g), 2.2.6.2-(g)),
and (iii) non-apico-alveolar homorganic clusters (/mb, Jlj, I]g/) (2.2.6.1-(h.)).
(c) Peripheral: (i) peripheral nasals (/m, IJI) (2.2.6.1-(e), 2.2.6.2-(c), -(e)).
(d) Semivowel (/j, w/): (i) (2.2.6.1-(d), 2.2.6.2-(d)).
(e) Possibly, the retroflex approximant/.II and the palatal semivowel /j/ (2.2.6.1-(f),
-(g), 2.2.6.2-(f), -(g)) constitute a natural class. See 2.5-[4].
That is, mainly in terms of place of articulation, 'apico-alveolar' is the largest
natural class, followed by its opposite: non-apico-alveolar. Other classes are much
smaller.
As seen above, when apico-alveolars function as a natural class, they consist of
sonorants only (jr, 1, n/), i.e. liquids and the nasal, and exclude the stop (jdl). When
non-apico-alveolars function as a natural class, they consist of stops only, nasals
only, or a nasal and a stop. When peripherals function as a natural class, they corn-
prise the nasals only (/m, 1]1), and exclude the stops (/b, g/) and the semivowel (/w/).
This indicates that, when a given group of phonemes function as a natural class,
there are preferences between places and manners of articulation as shown below.
(2-88) Preferences:
a. apico-alveolars
b. non-apico-alveolars
c. peripherals
sonorants
nasals, stops
nasals
There appears to be no place of articulation which prefers stops, to the exclusions
of sonorants. (See 2.5 for more on natural classes.)
[3] Relative order of members and syllable contact
Intra-root and inter-morphemic clusters exhibit very similar orders of the members.
[3-1] Intra-root clusters (Table 2-18)
Bi-consonantal clusters exhibit the following orders (repeated from 2.2.6.1).
Phonotactics 117
(2-89) Intra-root hi-consonantal clusters:
(a) homorganic N-plus-S clusters (/mb, nd, JlJ, IJg/).
(b) apico-alveolar sonorant (/r, 1, nl) + non-apico-alveolar stop (/b, J, g/).
(c) apico-alveolar liquid (jr, 11) + non-apico-alveolar nasal (jm, Jl, IJ/).
(d) apico-alveolar liquid (jr, 11) +semivowel lw/
(e) apico-alveolar nasal (In/) + peripheral nasal (jm, I]/).
(t) liquid !J./ or semivowel ljl + non-apico-alveolar stop (/b, J, g/).
(g) liquid Ill or semivowel lj/ + non-apico-alveolar nasal (lrn., J1, I]/).
In each combination, the first member is a nasal, a liquid, or the semivowel lj!
They are all sonorants. (But lw/, which is a sonorant, does not occur here.) The
second member is often, though not always, a stop. But the second member may
be a nasal (jm, J1, IJ) or the semivowel /w/
Intra-root tri-consonantal clusters exhibit the relative order of its mem hers
shown in (2-90).
(2-90) Intra-root tri-consonantal clusters:
liquid (/r, 1, 11) or ljl +nasal (lrn., n, Jl, IJ/) + non-apico-alveolar stop (/b, J, g/)
[3-2] Inter-morphemic clusters
Bi-consonantal clusters exhibit the following orders (repeated from 2.2.6.2).
(2-91) Inter-morphemic hi-consonantal clusters:
(a) homorganic stop-plus-nasal clusters (/m-b, n-d, ]1-J/, not IIJ-gl).
(b) apico-alveolar sonorant (In, r, Ill)+ stop (/b, d, J, g/).
(c) apico-alveolar liquid (jr, 11) +peripheral nasal (jm, I]/).
(d) apico-alveolar liquid (/r, 1/) +semivowel (/j, w/).
(e) apico-alveolar nasal (In/)+ nasal (lrn., n, IJ/).
(t) liquid Ill or semivowel (/j/) + non-apico-alveolar stop (/b, J, g/).
(g) liquid Ill or semivowel ljl + non-apico-alveolar nasal (/m, J1, I]/).
others: lrn.J, mg, IDIJ, pb, JliJ, dg, di], nw, jjl.
In (a) to (g), the first rnem her is a nasaL a liquid or the semivowel lj/ The second
member is a stop, a nasal, or a semivowel. It is difficult to make any generalization
about clusters of 'others'.
Tri-consonantal clusters show the following order.
(2-92) Inter-morphemic tri-consonantal clusters:
liquid (/r, 1, 11) or ljl + nasal (/m, n, I]l, but not 1]1/)
+ non-apico-alveolar stop (lb, gl, not IJ/)
Compare (2-92) with (2-90). Unlike intra-root clusters, inter-morphemic ones lack
the nasal I pi for the second member, and the stop IJI for the third. This is another
difference between intra-root and inter-morphemic clusters.
118 Phonology
The relative order of consonants and semivowels in clusters can be shown very
roughly in (2-93). Vowels are added for comparison. '>' means 'precedes'. (See
below for sonority, strength, and syllable contact.)
(2-93) vowels> liquids> semivowels >nasals> stops
more sonorous
weaker
syllable-final
less sonorous (sonority)
stronger (strength)
syllable-initial (syllable contact)
The evidence for this ordering is as follows.
(i) 'Nasal> stop' is supported by (2-89-a, -b), (2-90), (2-91-a, -b), (2-92).
(ii) 'Liquid> stop': (2-89-b, -f), (2-90), (2-91-b, -f), (2-92).
(iii) 'Liquid> nasal': (2-89-c, -g), (2-90), (2-91-c, -g), (2-92).
(iv) 'Liquid> semivowel': (2-89-d), (2-91-d).
(v) 'Semivowel> stop': (2-89-f), (2-91-f).
(vi) 'Semivowel> nasal': (2-89-g), (2-90), (2-91-g), (2-92).
(Also, there is some evidence to group vowels and liquids as against nasals. See 4.7.3.)
Similar rankings of phonemes have been proposed by other scholars. These
rankings are claimed to show the degree of (i) sonority of these phonemes, in-
cluding vowels, and (ii) inversely, the strength of semivowels and consonants. On
this view, the first phoneme in a cluster is more sonorous (or weaker) than the seg-
ment that follows.
In (2-93), liquids are considered more sonorous (or weaker) than semivowels.
In this respect, it is similar to the rankings proposed by Hankamer and Aissen
(1974) and by Suzuki (1989), and it differs from those set up by Hooper (1976),
:Murray and Vennemann (1983), Zwicky (1972) and many others, where the rela-
tive ranking of liquids and semivowels is reversed.
As for the nasals, (2-89-e) and (2-91-e) indicate '/n/ > /m, !]I' Furthermore, /mr.]l
of 'others' of Table 2-19 points to 'lml > lrj/', and 1'1!]1 of 'others' shows '1If < 1!]1'
Jointly they suggest:
(2-94) In! > /m, pi 1!]1
There is no evidence to relatively rank /m/ and /pl.
As for stops, /dgt of 'others' suggests:
(2-95) /dl > /gl
There is no evidence to relatively rank other stops. The cluster /dg/ occurs in
/wa.Iaji-d-gu/ 'one's own-LINK-DAT' (2.2.6.2). It is the only stop+stop cluster in
Warrongo.
Phonotactics 119
Note that In/ of (2-94) and /dl of (2-95) are apico-alveolar, and that lrjl and lgl
are (dorso-)velar. That is, apico-alveolars may be the most sonorous (or weakest)
and velars the least sonorous (or strongest).
There are two exceptions to the generalization of (2-93): lnwl (nasal > semi-
vowel) and ldijl (stop >nasal) of 'others'. Also, in each of the clusters Inn! and ljjl,
the two members are the same phoneme and consequently they are equal in terms
of sonority or strength. At least these clusters are not exceptions to (2-93). These
four clusters occur across a morpheme boundary, and not in roots. In this respect,
the restrictions are weaker on inter-morphemic clusters than on intra-root clusters.
Note that In/ of /nw/ and /d/ of /dij/ are apico-alveolar, and lwl and lrjl are velar.
Taking (2-94), (2-95), lnwl and ldijl into consideration, it is possible to set up the
following hierarchy:
(2-96) apico-alveolar > bilabial, lam ino-palatal > velar
more sonorous
weaker
syllable-final
less sonorous
stronger
syllable-initial
If this is the case, lnwl and ldijl conform to (2-96), although they are exceptions to
(2-93).
The rankings of consonants and semivowels cited above also concern the phe-
nomenon of syllable contact. For example, Murray and Vennernann (1983: 514)
state that there is 'a universal preference for syllable onsets [i.e. cluster-final con-
sonants/semivowels- TT] that are at least as strong as the preceding syllable offset
[i.e. cluster-initial consonant/semivowel - TT]' On this view, (2-93) (manners of
articulation) and (2-96) (places of articulation) summarize the 'syllable contact
laws' ofWarrongo.
2.2.7. Syllables: onset and coda
In 2.2.4, we looked at the syllable types that occur in words, in roots, in suffixes,
and in enclitics. In the following, we shall consider syllables in words and roots,
and not those in suffixes alone or enclitics alone.
As noted in 2.2.4-[2], tri-consonant clusters present a problem for syllabification.
The second member of a given cluster may be assigned to the preceding syllable
or the following syllable. In the following, we shall be concerned with hi-conso-
nantal clusters only.
As seen in 2.2.4, words and roots exhibit the following four types of syllables:
ev, eve, ev:, and eve. Among them, ev and eve are very common, while
the latter two are eA1:remely uncommon. We shall look at each of these four types.
Many examples ofeV and eve were given in 2.1.2, among others.
120 Phonology
[1] cv
All of the consonants and sem ivowels are attested in the position of C.
[2] C1VC2
This has yielded at least the following patterns (and no doubt there are more).
(a) C1 is any stop and C2 is a nasal, a liquid or the semivowel /j/ (but not /w/;
see below). This pattern is by far the most frequent, e.g. /baQ_gaj/ 'spear',
lba.r_bi_la/ 'echidna', lbal_banl 'bone'
(b) C1 is the semivowel /j/ and C2 is the stop /dl: /wa_Ja_ji-d-gu/ 'one's own-
LINK-DAT', /.ru_bi_jid-I]u_majl 'Rosevale-ABL'. Syllables ending in a stop are
unusual in Warrongo. (As noted in 2.2.6.3, /dl is the only linking interfix that con-
sists of a stop. Bona fide Warrongo roots do not end in a stop. /.rubijidl 'Rosevale'
is a loan.)
(c) C1 is a nasal and C2 is a liquid or the semivowel/j/ (but not /w/), e.g. /mar_ga/
'thin', /ma.J_bUJ 'louse', /mal_ba/ 'dance style', /gi_ca_maj/ 'Girramay', /nai_I]il
'to shake' /Jla.f_ga/ 'down', /Qa.r_gun/ 'blood', lba_da_Qall 'downstream-to',
/ja1_Quj/ 'avoidance style'.
(d) Cl is a liquid or a semivowel and C2 is a nasal, e.g. II]a_cam_bil 'to hear',
/ju_di_ga_ca-n! 'swim-ITER-NF', lba:_ba_.Jaml 'Mbabaram', /I]U_.J\p_Jill '(place-
name)', /ja_lam/ 'wave', /jam_ba/ 'camp', /jan_ba_ca/ 'long spear', /gi_jl\fl/
'mussel shell', ljaQ_ga/ 'to look for', /wam_bi_rrul 'Wambino' (1.5.4.2-[1]), /wan_Jal
'hole', /wa.Jl_Jal 'where', /waQ_gun_di_1a/ '(placename)'
(e) C1 is a liquid and C2 is the semivowel /jl (but not /w/), e.g. /gi_JU_cuj/ 'centi-
pede', /ju_.1aj/ 'quiet'.
(f) C1 is a semivowel and C2 is a liquid, e.g. /jal_ga/ 'road', /wa.r_ga_maj/
'Warrgamay',lba_wi.J/ '(placename)', /wal_wa/ 'bad'.
(g) C1 and C2 are nasal, e.g. /mam_bu/ 'back bone', /gu_mun/ 'quite', /llil\fl_Jal
'food', /maQ_ga/ 'flower', /I]a_nim_ba_cil 'how many', /ga_nan/ 'again',
'ignorant', /Qam _ba/ 'half', /Q1Ql_Ji_la/ 'ant nest'.
Stops as a rule do not occur syllable-finally, although there are two exceptions;
see (b) above. According to the analysis adopted (2.1.5.2), /w/ does not occur syl-
lable-finally; there is no exception.
[3] CV:
This is attested in only two instances: /Qa:_Ja/ '(placename)', /ja:_c-a/ 'yard-LOC'.
C1 is the nasal/1]1 or the sern ivowel /j/.
Statistics 121
[4] ClV:C2
This, too, has only two examples: /ba:n/ 'whip', /ja:.r/ 'yard'. Cl is the stop lb/ or
the semivowel /j/, while C2 is the nasal /n/ or the liquid /r/.
Regarding Australian languages, Di.xon (1980: 188) states that syllable-initially
peripherals are the most favoured, followed by lam inals, while syllable-finally api-
cals are the most favoured. This generalization applies to Warrongo fairly well, as
can be seen in the examples above. See also (2-96).
2.3. Statistics
The statistical tendency of occurrence of phonemes will be examined, by means of
a dictionary count (2.3.1) and by a text count (2.3.2).
2.3.1. Statistics (1): dictionary count
This dictionary count is based on approximately 1600 roots in Tsunoda (2003). The
figures for root-initial and root-final positions are given in Table 2-20. The four
dots indicate that the phoneme in question is (or appears to be) prohibited in this
position. There are two roots that end in the stop ldl (2.2.5.1-[3]). But they are both
loans from English. They constitute only 0.2% of root-final phonemes. Also, root-
initial /a/ and /a:/ are exceptional (2.2.3.1-[1], -[2]). These roots will be ignored in
the following discussion of statistics.
Table 2-20. Phoneme frequency (1): dictionary connt
root-initial root-final
lbl 23%
} }
ldl 1%
62o/
0.2%
0.2o/
ljl 13%
/g/ 25%
lml 10%
}
0.3%
}
In/ 0.4%
18.4o/
7%
8.3o/
/pi 2% 1%
/zy 6%
ld
}
1%
}
Ill 2% 2% 1% 7%
Ill 5%
/j/ 7%
}
17%
4%
}
4%
/w/ 10%
Ia/ 0.2%
}
58%
}
Iii
0.2%
11%
81%
lui 12%
/a;/
122 Phonology
[1] Root-initial position
Stops are the most frequent (62%), followed by nasals (18.4%), and semivowels
(17%). /:r/ is the least frequent (2%). This situation can be shown as follows.
(2-97) Phoneme frequency in root-initial position:
l:r/ semivowels, nasals stops
least common most common
The relative frequency of /:r/ (a liquid), semivowels, nasals and stops is exactly
the same as the relative order of consonants and semivowels in consonant clusters
shown in (2-93)- except that the other two liquids k, 11 do not occur root-initially
(Table 2-17).
Among the stops, peripherals (/b/ 23%, /g/25%: 48% in all) are the most fre-
quent, followed by the lamino-palatal/JI (13%). The apico-alveolar /dl is the least
frequent (1%). The same applies to nasals. Peripherals (/m/ 10%, /rjl 60/o) are the
most frequent, followed by the !amino-palatal /p/ (2%). The apico-alveolar In!
(0.4%) is the least frequent.
[2] Root-final position
Vowels (81%) are overwhelmingly the most frequent. Among them, /a/ (58%) is
by far the most frequent, and /u/ (12%) and /il (11%) are almost exactly equally
frequent. This is shown in Table 2-21. Consonants and the semivowel /j/ are un-
common, and it is difficult to generalize about them. At least, among the conso-
nants, In/ (7)1o) and Ill (5%) (both apico-alveolars) are the most common.
[3] Vowels
In both root-initial and root-second syllables, again /a/ is by far the most frequent,
followed by lui, and /il is the least frequent. See Table 2-21. According to Chomsky
and Halle (1968: 409), among vowels, /a/ is the least marked, followed by u/,
and there is no difference in terms of markedness between /i/ and /u/. Among the
Warrongo vowels, /a/ is by far the most frequent in each of these three positions,
exactly as expected. Now, since lui and /il have the same degree of markedness
according to Chomsky and Halle's analysis, they would be e"-'Pected to be equally
frequent. However, this does not apply to Warrongo. lui is consistently more fre-
quent than Iii (although the difference is very small root-finally). As far as their
frequency is concerned, /u/ may be considered less marked than Iii in Warrongo.
Table 2-21. Frequency of vowels
Ia! lui Iii
in root-initial syllable 52% 34% 14%
in root-second syllable 57% 27% 17%
root-finally 58% 12% 11%
Statistics 123
Table 2-22. Phoneme frequency (2): text connt
word-initial word-final
lbl 13%
} }
/dl 0%
32o/ Oo/
ljl 6%
/g/ 13%
lml 6%
}
0%
}
In/ 0%
33%
5%
5%
/pi 2% 0%
/rj/ 25%
ld
}
1%
}
!JJ 0% Oo/ 0% 7o/
Ill 6%
/j/ 27%
}
36%
9%
}
9%
/w/ 9%
Ia!
OCJ/
}
38%
}
/if
Oo/
20%
78o/
lui 20%
/all 0% 0%
2.3.2. Statistics (2): text count
This text count is based on the first 100 words of the texts in Tape 72/26; this por-
tion of the texts is included in Text 1. The dictionary count (2.3.1) concerns roots.
However, this text count deals with words. This is because, in texts, nouns and
verbs (and verbs in particular) are generally combined with an inflectional suffix.
Here, by words, we mean 'word' as shown in (2-78), i.e. an enclitic is considered
to be part of the word in question. This text count looks at only 1 00 words, but it
does not seem worthwhile to expand the sample. An expanded sample is likely to
produce much the same result. Consider Table 2-22.
In the main, Table 2-22 shows a similar tendency to that in Table 2-20, except
for two notable differences in the initial positions.
First, /rj/ is 6% by the dictionary count., but it is 25% by the text count., about four
times more frequent. This is due to the frequent occurrence of h:Jaja/ '1SG.NOMIERG'
and ll]una-IJUIDaj/ 'that-after' ('after that; and then'; 3.8.1-[9]) in the texts.
Second, /j/ is 7% by the dictionary count, but it is 27% by the text count., again
about four times more frequent. In the texts, /jani/ 'to go/come' occurs very fre-
quently. See Texts 1 and 2 for examples.
124 Phonology
2.4. Alternation between phonemes
Alternations between a consonant and another consonant (or a semivowel) are at-
tested. All of them occur in Alf Palmer's data. One of them occurs in Alec Collins'
data, as well; see [5].
[1] 11/-/r/
In AlfPalmer's speech, Ill is sometimes pronounced as /r/. For example, in TeA1: 2,
Line 41, !Jilbaj/ 'knowing' sounds like [d.3irbai]. This may be a phonetic variant
of Ill. However, there are cases where the pronunciation of [r] occurs frequently
enough for it to be recognized as manifesting the phoneme /r/. See (2-98) and (2-
99). In the case of /gimbira/ in (2-100), only [r] occurs, and not [1], hence /r/.
(2-98) lbiJil/, /bijir/ lbiJil-I]a/, /biJir-IJal
'wide' 'wide-TR'
(2-99) /I]alban/
'fast'
II] arb an/
'fast'
(2-100) /gimbil /gimbil-gimbiV /gimbiJ-....gimbir/ /gimbiral
Vt, Vi 'blow [of wind]' 'windy' 'windy' 'fart'
Regarding (2-100), the verb /gimbil 'to blow' appears to belong to L-class (cf. Table
3-14); this is reflected in its derivative word /gimbil-gimbiV 'windy' Regarding
(2-99), II]arban/ 'fast' seems to be used for emphasis. An example from the texts,
in which a man says to his wife, 'Walk fast!':
(2-101) II]arban jinda IJarbanl
nganvan-0! yinda ngarrban-0!
fast-NOM 2SG.NOM fast-NOM
'[Walk] fast! You [walk] fast!
Taking these facts into account, the forms involving IV may be the older ones. See
3.11.1.7-[11] for more on Ill and/r/.
Phonetically II]arban/ in (2-101) is [IJarpan]. That is, /b/ is voiceless. As noted
in 2.1.3.2.6-[3], II]arbanJ is the only instance in which /b/ is voiceless in a liquid-
plus-stop cluster. In other instances, lbl is voiced consistently.
[2] /pi- /j/
In this alternation and those discussed below, relevant forms occur in some other
languages. In (2-1 02) and those that follow, the languages are ordered roughly from
north to south. A broken line indicates the boundary between Dixon's Herbert
River Group (which includes Dyirbal (i.e. Jirrbal Girrarnay, etc.), Warrgamay
Alternation between phonemes 125
and Nyawaygi) and the Mari subgroup (which includes Warrongo, Gugu-Badhun,
Gujal and Biri). Warrongo neighbours the Herbert River Group. See Table 1-1 and
Map 5. (The data on Alec Collins' Warrongo and on Gujal are from Peter Sutton.)
There are two sets of alternations that involve '/pi- /j/' First, consider:
(2-102) 'sand goanna'
Jirrbal of Dyirbal (my data from
Tommy Springcart)
Warrgamay (Di.xon 1981: 114)
Nyawaygi (Dixon 1983: 506)
AlfPalmer's Warrongo
Alec Collins' Warrongo
Gugu-Badhun (Sutton 1973: 210)
Gujal
/jagapi
/jagapi 'big goanna'
/jagapi, /jagaj/
/jagajl
/gagaj/
/gagaj/
(The correspondence between the palatal /j/ and the dental /Q/ is regular; see Table
1-4.) Second, consider:
(2-103) 'fingernail'
Dyirbal (Dixon 1972: 399)
Alec Collins' Warrongo
AlfPalmer's Warrongo
Gugu-Badhun (Sutton 1973: 208)
/bigup/
lbigupl
/bigup/, /biguj/
/biguj/
In Warrongo, the ergative and the locative forms of /pgaj/ 'sand goanna', for
instance, are /jagap-JU! and /jagap-ja/, as is the case with lbaiJgaj/ 'spear'; see
Table 2-15. The same is true of /biguj/ 'finger nail'. This suggests that, with the
forms 'sand goanna' and those for 'finger nail', the form ending in /pi is the older
form and that ending in /j/ is an innovation. That is, regarding this change from /pi
to lj/, in the main the Herbert River Group is conservative, and the Mari subgroup
is innovative.
[3] /g/- ljl
(2-104) 'liver'
Jirrbal of Dyirbal (my data from
Tommy Springcart)
Warrgamay (Dixon 1981: 123)
Nyawaygi (Dixon 1983: 508)
AlfPalmer's Warrongo
Gugu-Badhun (Sutton 1973: 196)
/jib a!
/giba/
/giba-giba/
/giba/ ([geba]),
/jibal ([ <f3iba])
/jib a/
126 Phonology
It seems likely that the change is /g/ -> ljl; cf. Latin genus (the initial consonant is
a velar stop) and English genus (the initial consonant is a palatal affricate). That
is, /giba/ is the older form and !Jiba/ is an innovated form. (See 2.1.4.3-[1], -[2]-(e)
for the phonetic realizations of /giba/ ([geba]) and /jiba/ in Alf Palmer's
speech.)
[4] !J/-Ij/
Consider (2-55) and the following forms in Alf Palmer's Warrongo.
(2-105) lbula.Ii! /bula.Ii-bula.Ii!
'two-two (i.e. four)'
*/bulaji/ lbulaji-bulaji/
'two-two (i.e. four)'
(Phonetically, /ji/ of /bulaji-bulaji/ is [i] (2.1.4.3-[1], 2.1.5.1.2-[3]-(b-1).)
(2-106) lbunbul/ /bunbul-baJa/ /bunbu-baJa!
'smoke' 'big smoke' 'big smoke'
(Jalngoy?) (Jalngoy?)
--------------------------------------------------------------------
/bunbul-baja/ /bunbu-baja!
'smoke' 'smoke'
(Jalngoy?) (Jalngoy?)
(2-107) !Jam. if /jami-ba.Ia! /jami-baja/
'fat (not 'fat (not thin)' 'fat (not thin)'
meat)'
(2-108) /guli/ /guli-ba.Ia/ /guli-baja/
'angry' 'angry' 'angry'
(2-109) /gunda! /gunda-ba.Ial /gunda-baja/
'darkness' 'sky', 'bright (not dark)' 'dark'
(The meaning of /gunda-ba.Ial seems unpredictable.) Clearly a suffix is involved in
these forms, and it shows the alternation of 'lba.Ial- /baja/' (But its function is not
clear; see 3.7.1-[8], -[9].) What appears to be a cognate is attested as follows.
(2-110) lba.Ial
Dyirbal (Dixon 1972: 226)
Warrgamay (Dixon 1981: 34)
lba.Ial 'comparative'
lba.Ial 'comparative'
AlfPalmer's Warrongo lba.Ial, /baja/ (meaning unknown)
Gugu-Badhun (Sutton 1973: 103-104) lba.Ial (meaning unknown)
Alternation between phonemes 121
Only Alf Palmer's Warrongo exhibits the alternation of -/baj af
There is a phonetic similarity between Ill and /j/ Recall that phonetically l:r/ is
realized generally as a retroflex approximant and occasionally as a retroflex flap
(2.1.3.1). In each set of the relevant forms in (2-56) and (2-105) to (2-110), l:r/ seems
to be (always?) realized as a retroflex approximant. Now, Jakobson, Fant and Halle
(1967: 22) note: 'As for the so-called "continuant [i.e. approximant-TT] r", it is ac-
tually a non-syllabic vowel' That is, the Warrongo l:r/, when realized as a retroflex
approximant, is a non-syllabic vowel, like lj/.
Dixon (2002: 576, 627) reports the diachronic change of l:r/--+ lj!- and also that
of lrl --+ ljl- from a few other Australian languages. In view of this, the change in
Warrongo seems to be l:r/--+ lj!, and not ljl--+ l:r/.
[5] /bl -!ml
This alternation occurs not only AlfPalrner's data, but also Alec Collins'.
(2-111) 'to cut'
Dyirbal (Dixon 1972: 403)
Warrgarnay (Dixon 1981: 117)
Nyawaygi (Dixon 1983: 508)
AlfPalmer's Warrongo
Alec Collins' Warrongo
Gugu-Badhun (Sutton 1973: 216)
Ranji Pope's Gujal
Biri (my data)
/gllllbal
lgunbal
/gllllbal
lgllllbal, lgunma/
/gllllbal, /gunma/
lgunma/
lgunma/
/gulmal
AlfPalmer's data exhibit the alternation of 'lbl -lmf (/gunbal is far more frequent
than lgunmal.) A possible line of development is as follows: lgulma/ -+ /gunma/
--+ /gunba/ The first change (/1/ --+ In!) involves assimilation (to the following
nasal), and the second change (!ml --+ lbl) involves dissimilation (away from the
preceding nasal). The alternative scenario is the opposite: lgunba/--+ lgunma/--+
lgulma/. There is no evidence to prefer one hypothesis to the other.
[ 6] Ill and In!
(2-112) 'to squeeze'
Dyirbal (Dixon 1972: 402)
Warrgarnay (Dixon 1981: 115)
AlfPalrner's Warrongo
IJUimal
IJUillll a/
IJUimal, IJUillllal
A possible line of change is: IJUlmal -+ IJunmal The alternative is: IJUillllal --+
IJUimal. Again, there is no evidence to prefer one hypothesis to the other.
Note that [5], too, exhibits the alternation between IV and In!. In [5] the form
involving Ill (/gulma/) occurs in the Mari subgroup. In contrast, in [6] the form
128 Phonology
involving IV (IJU.Imal) occurs in the Herbert River group. That is, this correspon-
dence is irregular.
In Alf Palmer's data, there is also one instance of each of the following alterna-
tions: ldl -lrl and Ill- In!; see Table 3-4. There are two instances of IJ!- In!; see
Table 3-4 and 4.7.3-[4].
2.5. Natural classes
Natural classes may be recognized on the basis of phonotactics, synchronic alter-
nations, diachronic changes, and allomorphy, among others. In Warrongo, the syn-
chronic alternations discussed in 2.4 appear to be results of diachronic changes.
Natural classes proposed for Warrongo are listed in [1] (phonotactics), in [2] (syn-
chronic alternations), in [3] (allomorphy of case suffixes) and discussed in [4].
Some of the classes are very small, consisting of just two members, while others
are somewhat larger. It is sometimes very difficult to make a 'natural' generaliza-
tion. Such classes are 'unnatural' (Lass 1991: 82).
[1] Natural classes (1): in terms ofphonotactics
In terms of the possibilities in consonant clusters (2.2.6.1, 2.2.6.2), it is fairly easy
to set up 'natural' classes. They can be classified as follows.
(i-1) Homorganic nasal-plus-stop clusters (/mb, nd, JlJ, IJg/) (2.2.6.1-(a)).
(i-2) Non-apico-alveolar nasal-plus-stop homorganic clusters (/mb, JlJ, IJg/)
(2.2.6.1-(h)).
(i-3) Non-velar nasal-plus-stop clusters (lrnb, nd, JlJ/) (2.2.6.2-(a)).
(ii-1) Apico-alveolar sonorants (/r, I, n/) (2.2.6.1-(b), 2.2.6.2-(b)). Furthermore, they
are alike in that they seem to be prohibited word-initially and root-initially.
(ii-2) Apico-alveolar liquids (/r, 1) (2.2.6.1-(c), -(d), 2.2.6.2-(c), -(d)).
(iii-1) Stops /b, d, J, gl (2.2.6.2-(b)).
(iii-2) Non-apico-alveolar stops (/b, J, g/) (2.2.6.1-(b), -(f), 2.2.6.2-(f)).
(iv-1) Non-apico-alveolar nasals (lm, JI, I]/) (2.2.6.1-(c), -(g), 2.2.6.1-(g)).
(iv-2) Peripheral nasals (/m, I]/) (2.2.6.1-(e), 2.2.6.2-(c), -(e)).
(v) Semivowels (/j, w/) (2.2.6.1-(d), 2.2.6.2-(d)).
(vi-1) Liquids (/r, l, 11) and palatal semivowel ljl (2.2.6.1-(h), 2.2.6.2-(h)).
(vi-2) Retroflex approximant IJ! and palatal semivowel ljl (2.2.6.1-(f), -(g), 2.2.6.2-
(f), -(g)).
(vii) 11, r, jl (2.2.6.1-(i)). This class does not seem 'natural'.
The classes listed above are recognized on the basis of the occurrence of phonemes
in consonant clusters. All of them are 'natural', except for (vii).
Furthermore, in terms of the possibilities in structural positions (Table 2-17),
the following classes can be set up. However, they do not seem 'natural', in con-
trast with the classes listed above.
Natural classes 129
(viii-1) Stops (lb, d, J, g/) and velar sonorants (/IJ, w/): prohibited word-finally.
(viii-2) Non-apico-alveolar stops (/b, J, g/) and velar sonorants (II], w/): prohibited
root-finally.
(ix) /b, r, J, 1, j, w/: attested suffix-medially and inter-vocalically in suffixes.
(x) /b, J, g, m, n, p, r, w/: attested enclitic-initially.
(xi) !J, m, 1, w/: attested enclitic-medially.
(xii) Vowels (/a, i, u/), apico-alveolar sonorants (In, r, l/), and palatal semi-vowel
/j/: attested suffix-finally.
(xiii) Vowels (/a, if, u/) and In, 11: attested enclitic-finally.
[2] Natural classes (2): in terms of synchronic alternations
The following alternations between phonemes are observed in Alf Palmer's
Warrongo (2.4-[1] to -[6]).
(i) Apico-alveolar: lateral Ill - rhotic lr/ (2.4-[1])
(ii) Palatal: nasal /pi- semivowel /j/ (2.4-[2])
(iii) Velar stop /g/- palatal stop !JI (2.4-[3])
(iv) Retroflex approximant !J/- palatal semivowel/j/ (2.4-[4])
(v) Bilabial: stop /b/- nasal/m/ (2.4-[5])
(vi) Apico-alveolar: lateral /11- nasal/n/ (2.4-[6])
[3] Natural classes (3): mainly in terms of the allomorphy of case suffixes (3.2.1.1)
(i) Liquids: /r, J, 11 (3.2.1.1-[2]-(c))
(ii) Palatal non-nasals: /j, J! (3.2.1.1-[ 4]). See (2-47).
(iii) Palatal sonorants: /j, Jl! (3.2.1.1-[2]-(d-2), -[4]-(b-2))
(iv) Peripheral non-nasals: /g, w/ (3.2.1.1-[5]). See (2-60).
[ 4] Discussion
We have looked at 'natural' classes in terms of[l] phonotactics, [2] synchronic alter-
nations, and [3] mainly in terms ofthe allomorphy of case suffixes.
Regarding place of articulation, the most recurrent feature is 'apico-alveolar'
(referred to eight times), followed by 'non-apico-alveolar' (seven times), and 'pe-
ripheral' (three times). The fact that the feature 'apico-alveolars' is recurrent as
against all the other places of articulation supports Dixon's (1980: 188) generaliza-
tion for Australian languages that apicals are the least marked.
As for manner of articulation, the recurrent features are 'semivowel' (seven
times), 'nasal' (six times), 'stop' (four times), 'liquid' (four times), 'sonorant' (three
times), and 'nasal-plus-stop homorganic' (three times).
As seen in [1]-(ii-1), -(ii-2) above, and also noted in 2.2.6.3-[2], where the feature
'apico-alveolar' is invoked, the stop member /d/ is always excluded.
The term 'liquids' refers to 'the laterals (1-sounds) and the various intermittent
r-sounds' (Jakobson, Fant, and Halle 1967: 19; cf. also Ladefoged 1975: 79 and
130 Phonology
Lass 1991: 83), i.e. 11, r, Jl in the case of Warrongo. This term is useful, as seen
above. However, there is evidence to divide them into /1, r/ and /:r/. The evidence
for grouping of 11, r/, to the exclusion of Ill, is in [1]-(ii-2) (four times) (and also
perhaps [1]-(ii-1) (twice). In contrast, /J.! often behaves together with the palatal
semivowel /jt, not /1, r/; see [1]-(vi-2) (four times). This shows that /JI should be
grouped with /j/. Recall that an approximant /J.! is a non-syllabic voweL like /j/.
It is a controversial issue in Australian linguistics whether to group /J.! with
lr! under the rubric of 'rhotics', or to group it with semivowels (/j, w/) (McGregor
1988, Dixon 2002: 574-575). As far as Warrongo is concerned, it is best to group /J.!
with /j/, in contrast with the group of /L r/.
2.6. Phonological processes'
A small number of phonological processes can be recognized. Except for the ap-
prehensional suffix for verbs (see [2]-(c-1) below), they are all sporadic and non-
productive, and have a limited range of application. Except for the apprehensional
suffix, all (or most) of the attested instances are listed.
[1] Vowel copying
The suffix !JVI is added to a noun root or a verb root, and the root-final vowel is
copied in the vowel of the suffix. The suffix generally seems to mean 'many' See
3.7.1-[16] (nouns) and3.11.1.4 (verbs).
[2] Epenthesis
Almost all the epenthetic rules involve insertion of a nasal, with one exception (in-
sertion of /j/). The epenthetic segments are considered interfixes.
(a) Insertion of /m/
Examples:
(2-113) /gubi/
'native doctor'
lbaraj-m-baraj/
'in the morning'
/dab a/
'black palm'
/gubi -m-bu1u/
'native doctor'
/daba-bulu/
lrjani/
'what'
/IJani-m-bari/
'how many'
'Black Palm (a place on Stone River)'
For comparison, an example in which this rule does not apply (i.e. /daba-bulul) is
included. Similarly for (2-113) and (2-115). See 3.7.1-[5] for /bulu/, and 3.7.1-[6]
for lbari/. In Warrongo, the meaning of /bulul is not clear, but Warrgamay has
'Phonological processes' 131
the noun-stem-forming suffix /bulul 'very, lots of' (Dixon 1981: 34). See (1-12).
The meaning of the placename /daba-bulul may be '[a place with] many black
palms' Etymologically, /gubi-m-bulul may mean 'true native doctor' or the
like. (Stone River appears to be in Warrongo territory (1.4.1)). See 3.8.2-[1] for
/baraj-m-baraj/. The non-reduplicated form lbarajt is not attested in Warrongo,
but it occurs in DyirbaL with the meaning 'next week' (Dixon 1972: 399).
/m/-insertion is also attested in the transitive-stem-forming suffix /(m)bal
(4.7.3).
(b) Insertion of /n/
A set of examples involving /wamaj/ 'good':
(2-114) /wamaj/ /wamaj-n-bil /wamaj-n-bi.Iil
'good' 'good-n-INTR' 'good-n-?'
/wamaj-n-biii-bil /wamaj-bil /wamaj-IJal
'good-n-?-INTR' 'good-INTR' 'good-TR'
See 4.7.1 for the intmnsitive-stem-forming suffix /bil, and 4.7.2 for the transi-
tive-stem-forming suffix /I]al. /wamajt by itself is not attested in Alf Palmer's
Warrongo, but it occurs in Alec Collin's data, in Gugu-Badhun (Sutton 1973: 222),
and in Ranji Pope's Gujal (Sutton data). (The verb wamay-n-bi-n 'good-LINK-
INTR-NF' occurs in Text 3, Line 32.)
/n/-insertion also occurs in certain inflected forms of certain nouns (Table 3-3),
demonstmtives (Table 3-5), and pronouns (Table 3-6). It is possibly involved in the
plural suffix /nbal; see 3.7.1-[15].
(c) Insertion of lrjl
This is attested in the following.
(c-1) The apprehensional forms of verbs. Verbs are of three classes: L-class,
Y-class, and ZERO-class. The apprehensional forms involve the suffix /gal. In
the apprehensional forms, II]! is absent in L-class, but it is obligatorily inserted in
Y-class and ZERO-class (Table 3-14). Examples:
(2-115) L-class
/balga-1-ga/
'hit-L-APPR'
Y-class
/pina-j-g-ga/
'sit-Y-IJ-APPR'
ZERO-class
/wadali-g-ga/
'run-(ZERO-)IJ-APPR'
As noted above, /I]!-insertion in apprehensional forms is productive.
(c-2) The dative of one of the two groups of y-final stems (Table 3-2), e.g. /baiJgaj-
g-gu/ 'spear-IJ-DAT' There is also a dative form that does not involve lrj/: /baiJgaj-
gu/ 'spear-DAT'
132 Phonology
(c-3) The feminine suffix /gan/ almost always follows /jl (there are three excep-
tions) and, when /gan/ follows /j/, /rjl is obligatorily (?) inserted between /jl and
/g/. All the instances, including the exceptions, are listed in 3.7.1-[17]. For example,
compare:
(2-116) [ko:i] [koii.Jgan] [jab:.]gai] [jab:.]gaii.Jgan]
/guwuj/ /guj-g-gan/ /jalngaj/ /jalngaj-g-gan/
'male.ghost' 'ghost-IJ-FEM' 'single.man' 'single.man-IJ-FEM'
/bimu/ /bimu-gan/ !Julbun! !Julbun-gan/
'WM' 'WM-FEM' 'spouse' 'spouse-FEM'
(The suffix /gan/ seems redundant in /bimu-gan/ 'wife's mother'.) Additional ex-
amples are shown in Table 1-7. Sections terms for feminine members involve the
epenthetic /rjl followed by the fern inine suffix /g an!.
(c-4) !Juwaj-g-guru/ 'term by which brothers-in-law call (address? refer to?) each
other'; cf. /juwaj/ 'brother-in-law' (probably, 'WB', and not 'HB') (1.5.7). However,
lrjl is not involved in /wagaj-guru/ '(man's name)' (1.5.4.2-[1]); cf. /wagaj/ 'wish-
bone'
(d) Insertion of /j/
All the epenthetic rules discussed above involve insertion of a nasal. But there is
one that involves the insertion of the semivowel /j/. Section terms for female mem-
bers involve the insertion of /j/, followed by another epenthetic element /rjl and the
feminine suffix /gan/ (see (c-3)). (In addition, two of them exhibit the root-final
vowel alternation of lui and /a!.) There is another example of /jl-insertion; compare
/wapJal 'where' and /wapJa-j-mu/ 'where-j-from' (3.8.1-[5]).
It seems, on the basis of (a) to (c), that nasal insertion tends to occur following
a phonetic [i] or [e], i.e. /j/ or /il. For example, compare /gubi-m-bulu/ [kobembolo]
'native doctor' and /daba-bulu/ [dababolo] 'Black Palm' Also, all the verbs of
Y-class involve /j/ [i] before the apprehensional suffix, e.g. /pina-j-IJ-gat [JlinaiiJga].
Verb stems of ZERO-class end in /i/ ([i] or [e]), e.g. /wadali-1]-ga/ [wadaleiJga].
This suggests that position immediately following [i] or [e] (jjl or Iii) is a preferred
environment for nasal insertion. Then, it is possible to say that, in the section
terms for female members, ljl is inserted in order to create a preferred environ-
ment for nasal insertion (i.e. the insertion off /rjl).
All of these epenthetic segments are referred to as linking interfixes in the
present work.
[3] Insertion or deletion of Ill
Consider the following examples.
(2-117) a. lbirgil/
'cold weather'
b. lbilllbul/
'smoke'
(Jalngoy?)
c. /gagal/
'big, hard
(not soft)'
/gag a-brua/
'big-'?', i.e.
'big'
/gaga-Jaril
'big-'?', i.e. 'big'
Stress 133
lbirgi-ba.Ial /birgi-ba.Ia-bil
'cold-?', i.e. 'cold-?-INTR
'cold weather'
lbilllbul-ba.Ial
'big smoke' 'big smoke'
(Jalngoy?)
lbilllbul-baj a/ /billlbu-baj a/
'smoke' 'smoke'
(Jalngoy?) (Jalngoy?)
/gagal-bi/ /gag al-i/
'big, hard-INTR' 'hard, strong-INTR'
/gag a-ba.Ia-bi/ /gaga-ba.Ia-IJal
'big-?-INTR' 'big-?-TR'
/gag a-lii
'big-APPL' (4.14.2.5-(b))
(See 3.7.1-[7], -[8], and -[9] for /gaga-Jaril, /balal and lbaja/, respectively. See
3.11.1.6 for the intransitive-stem -forming suffix /if.) Two analyses are possible: (i)
to say that the roots contain /11, and that Ill is deleted in certain words, and (ii) to
say that the roots do not contain /11, and that Ill is added in certain words. But the
conditioning factor for the presence/absence of this Ill is not known. Compare, for
example, lbilllbul-ba.Ial and lbunbu-ba.Ial.
[ 4] Compensatory lengthening
There is just one example. The language name 'Mbabaram' (1.4.1) was pronounced
[ba:ba.ram] by Alf Palmer (1.8.4.1-[2]). The initial [m] was dropped and the short
vowel [ba] was replaced with the long vowel [ba:]. This vowel lengthening is an
instance of compensatory lengthening. (The rhotic in this language is 'back rhotic',
and it may be 'a tap, a trill, or a rhotic continuant' (Dixon 199lb: 355-356).)
2.7. Stress
Stress is not distinctive in Warrongo. Nonetheless, it does exist phonetically. It ap-
pears that in Warrongo, as in many other languages, a 'stressed syllable is pro-
nounced with a greater amount of energy than an unstressed syllable' (Ladefoged
1975: 222). Impressionistically speaking, stress in Warrongo is generally accornpa-
134 Phonology
nied by a 'rise in pitch' (Ladefoged 1975: 223). However, this is not always the case.
See the first word of (2-126)-A, for instance. The first syllable ([wa)l]) bears stress
(to be precise, secondary stress), and yet it has low pitch. (The subsequent syl-
lables are totally unstressed, but they have middle-level pitch.) That is, stress is not
always accompanied by a rise in pitch. Also, again impressionistically speaking,
stress in Warrongo does not seem to lengthen the vowel concerned.
It has turned out to be very difficult to generalize about stress in Warrongo.
The following account is highly tentative, and it is not exhaustive. Also, I have
never investigated Warrongo stress instrumentally, and the following account is
highly impressionistic, based on my observation. The following general tendencies
may be noted.
[1] Three degrees of stress
It is possible to set up three degrees of stress: primary stress (shown by an acute
accent), secondary stress (shown by a grave accent, where necessary or useful) and
the absence of stress. However, the difference among them is not always clear-cut.
[2] Word-initial syllable
In the main, stress tends to fall on the word-initial syllable. This is the most domi-
nant tendency. Examples include the following. Disyllabic: /bama/ [bam a] 'man',
/m.aguj/ [makoi] 'big carpet snake', /waJal [wa1fa] 'crow', /winga.I! [Winga.J.]
'fish'. Trisyllabic: /dulubu/ [d6lobo] 'Joyce Palmer', /I]a.]1i.ri/ [IJaJli..J.i] 'Edith Lenoy',
!jaiJanal UaiJana] 'mother', lwaJaganl [wa1fagan] 'crow', /wafUIJu/ [waroiJo]
'Warrongo' Quadrisyllabic: (2-78) (/jalga-IJga=gul/). Pentasylalbic: (2-78) (h:Jani-
wu=baJUnl).
Furthermore, in quadrisyllabic and longer words, in addition to the primary
stress on the first syllable, secondary stress is sometimes observed on the third
syllable. But it is not easy to distinguish this secondary stress from the absence of
stress. The putative secondary stress is shown (by a grave accent) in the following
examples: /baJUbala/ [ba1fobala] 'Kirrama Station' and 'imaginary
hairy man'.
Stress seems never to fall on an enclitic or a suffix.
[3] Reduplication
If a given word involves reduplication (3.7.2, 3.8.2, 3.11.2), stress falls on the first
syllable of each root. There appears to be no exception. (The stress on the first
syllable of the second root may be considered secondary, and not primary, but this
distinction is not easy to make.) Examples: lbuna-buna/ [b6nab6na] 'couch grass',
!Jigif-.-.Jigir/ [d3igird31gic] 'willy wagtail', !Juda-JUdal [d,tad.36ta] '(placenarne)',
/gaJar-gaJarl [ka1farka1far] '(Adv) going past', lm:iigan-m:iigan/ [mLJ.ganmiJ.gan]
'hill-hill', /I]arban-IJarbanl [IJarpaniJarpan] 'fast-fast', /jwaj-jwaj/ U6.J.aij6.J.ai]
'quiet-quiet', and /wuma-wuma/ [6ma6ma] 'shade-shade' The same applies
to those words which exist in the reduplicated form only, e.g. /guj-guj/ [k6ik6i]
Stress 135
'mosquito', and h:Ja:-IJa:l (an onomatopoeic word to describe babies' cries)
(2.2. 3 .1-[ 1 ]).
As another example, compare the following:
(2-118) /baJanl
[Md.3an]
'to bite (NF)'
/ba Jab a JaiJanl
[ba1faba1faiJan]
'to finish up (NF)'
Etymologically, it is not certain if these two verbs are related. Nonetheless,
lhaJabaJaiJanl clearly involves reduplication of /baJa/, and stress falls on the first
syllable of each of the two occurrences of this root. (The stress on the third vowel
of /baJabaJaiJanl may be secondary stress.)
[4] Compounding
If a given word involves compounding (3.7.3, 3.8.3, 3.11.3), (in my observation) pri-
mary (not secondary) stress falls on the first syllable of the second root, and second-
ary (not primary) stress on the first syllable of the first root. There appears to be no
exception. Examples: IJUlaj-JlunguV[d_laiJ16ngol] 'tree-one (placename)' (1.5.4.3-[6],
3.7.3-[1]), and /ma.Ia-guna/ [ma.y_ag6na] 'hand-faeces' ('lazy person') (3.7.3-[1]).
There are instances where the 'heaviness' of the root or suffix involved, appears
to influence stress assignment. This will be discussed in [5] and [6].
[5] Bi-consonantal and tri-consonantal clusters
Bi-consonantal and tri-consonantal clusters behave differently in terms of stress as-
signment. When the cluster is hi-consonantal, stress falls on the first syllable of the
word, e.g. /bacbi.ra/ [bacbqa] 'echidna', /bacgan/ [bacgan] 'swamp wallaby', /balban/
[balban] 'bone', /bambui [barnbo] 'egg', /bujbun/ [b6ibon] 'spring water', /gandul
[gando] 'dog', !rn.ajgan/ [maigan] 'to tell (NF)', and/wambinu/ [wambino] 'Warnbino'.
However, when the cluster is tri-consonantal, (although I have not measured
stress instrumentally) my impression is that stress generally (though not always)
falls on the cluster-initial phoneme (not on the first syllable of the word). Compare
/balmbul [ba1rn.bo] 'bundle' with /bambui [bambo] 'egg' Other examples in-
clude (i) /IJU.fiilbun/ [IJofmbon] 'noise that one does not other people to hear', (ii)
lrnUIJamba.Ial [moiJainbqa] 'many', (iii) /bi.n:J9al [b4.IJga] 'grey hair', (iv) /gUllJgal/
[ko{I]gal] 'husband', (v) /jinbili]gaj/ [d.3inbih.19aj] 'Alf Palmer', (vi) /jali]gail
[jah.mai] 'single man', (vii) /bilnbi.Ian/ [bilnbqan] 'crimson rosella (?)', (viii)
/jajmbanl [d.3a1mban] 'to find (NF)', (ix) /jajJlJanl [d.3a1Jld.3an] 'to copulate with
(NF)', (x) /gujiJgan/ [koiiJgan] 'female ghost'
The cluster-initial phoneme of tri-consonantal clusters is a liquid (/c, .J., 11) or
the semivowel /j/ (Tables 2-18, 2-19). They can receive stress probably (i) because
liquids and /j/ are high in sonority, like vowels (see (2-93)), and (ii) because, fur-
thermore, /j/ is phonetically a vowel ([i] or [e]) (2.1.5.1.1).
136 Phonology
As seen above, in certain instances a tri-consonantal cluster attracts stress, but
a hi-consonantal one does not. That is, the 'heavier' cluster attracts stress, but the
'lighter' one does not.
However, there are exceptions, in which stress does not fall on initial phoneme
of a tri-consonantal cluster. The reason for this is not known. Examples: /bu.iJl.Jul!
[boipgj61] (see 2.1.3.2), /warn.aj-n-bi-nl [wamainbin] 'good-n-INTR-NF', and
lwulaj-IJ-gal [61aii]ga] 'die-Q-APPR'.
[6] 'Stress shift'
Affixation of certain (though not all) suffixes causes 'stress shift'; the stress shifts
to one of the subsequent phonemes. The data are far from complete, and details are
not known. Nonetheless at least the following tendencies can probably be recog-
nized. In all the attested examples, the root is disyllabic. Also recall that generally
stress falls on the word/root-initial syllable.
(a) The root contains no consonant cluster.
(a-1) The resultant stern/word contains no consonant cluster. Stress shift does
not occur: (2-119)-(a-1), (2-120)-(a-1), (2-121)-(a-1), (2-123)-(a-1).
(a-2) The resultant stern/word contains one hi-consonantal cluster (CC). Stress
shift does not occur: (2-122)-(a-2).
(b) The root contains a hi-consonantal cluster.
(b-1) The resultant stern/word contains one hi-consonantal cluster (CC).
(b-1-1) The suffix is CV Stress shift does not occur: (2-123)-(b-1-1).
(b-1-2) The suffix is CVCV Stress shifts to the next vowel: (2-120)-(b-1-2),
(2-121)-(b-2-3).
(b-2) The resultant stern/word contains two hi-consonantal clusters (CC, CC).
(b-2-1) The suffix is CCV. Stress shift does not occur: (2-122)-(b-2-1).
(b-2-2) The suffix is CVC. Stress shifts to the next vowel: (2-119)-(b-2-2).
(b-3) The resultant stem/word contains one hi-consonantal cluster and one tri-
consonantal cluster (CC, CCC). Stress shifts to the first phoneme of the
tri-consonantal cluster: (2-119)-(b-3).
(c) The root contains a tri-consonantal cluster.
(c-1) The resultant stern/word contains one tri-consonantal cluster (CCC).
Stress remains in the first phoneme of the first tri-consonantal cluster:
(2-120)-(c-1).
(c-2) The resultant stern/word contains two tri-consonantal clusters (CCC, CCC).
Stress shift to the initial phoneme of the second tri-consonantal cluster:
(2-119)-(c-2).
That is, the following two factors seem relevant.
Stress 137
(i) The absence/presence of a consonant cluster.
(ii) The heaviness of the cluster in question (i.e. hi-consonantal vs. tri-consonantal).
(iii) The heaviness of the suffix in question (i.e. CV, CCV vs. CVC, CVCV).
The general tendency seems to be for stress to be more likely to shift if there is a
heavy material towards the end of the stem/word. Examples follow.
(2-119) The suffix /gan/ 'feminine' (3.7.1-[17])
(a-1) The resultant stem/word contains no consonant cluster. Stress shift
does not occur.
/bimu/ /bimugan/
[hlmo] [hlmogan]
'WM 'WM-FEM'
(b-2-2) The resultant stem/word contains two hi-consonantal clusters (CC,
CC). The suffix is CVC. Stress shifts to the next vowel.
!Julbun/ !Julbun-gan!

'spouse'
(b-3) The resultant stem/word contains one hi-consonantal cluster and
one tri-consonantal cluster (CC, CCC). Stress shifts to the first
phoneme of the tri-consonanta1 cluster.
/gwguru/ /gwgura-j-IJ-gan!
[g6.J.goro] [gO{gorall]gan]
'(male memberof a section)' '(female member of the same)'
(c-2) The resultant stem/word contains two tri-consonantal clusters (CCC,
CCC). Stress shifts to the initial phoneme of the second tri-conso-
nantal cluster.
/jaliJgaj/ /jaliJgaj-IJ-gan/
[ja1r.J9ai] [jali]ga"ll]gan]
'single man' 'single woman'
(There are not many examples of /gan/ 'feminine'. /julbun-gan/ means 'woman
who runs away from her husband'. Both /bimu/ and /bimu-gan/ mean 'wife's
mother', and /gan/ in /bimu-gan/ seems redundant.)
138 Phonology
(2-120) The suffix /gall/ 'antipassive' suffix (Table 3-15)
(a-1) The resultant stem/word contains no consonant cluster. Stress shift
does not occur.
/baba-n/ /baba-gali-n/
[baban] [babagalen]
'stab-NF' 'stab-ANTIP-NF'
lm.uja-n/ lm.uja-gali-n/
[m6d3an] [m6lJ3aga1en]
'eat-NF' 'eat-ANTIP-NF'
lwaju-n/ /waju-gali-n/
[wad3on] [wad3oga1en]
'cook-NF' 'cook-ANTIP-NF'
(b-1-2) The resultant stem/word contains one hi-consonantal cluster (CC).
The suffix is CVCV Stress shifts to the next vowel
lba1ga-n/ /balga-gali-n/
[balgan] [balgagalen]
'hit-NF' 'hit-ANTIP-NF'
/ga]lji-n/ /ga]lji-gali-n/
[gaJln] [gaJ1d3J.galen]
'carry-NF' 'carry-ANTIP-NF'
/jar.Jga-n/ /jar.Jga-gali-n/
[jaiJgan] [j3IJgagalen]
'search-NF' 'search-ANTIP-NF'
(c-1) The resultant stem/word contains one tri-consonantal cluster (CCC).
Stress remains in the first phoneme of the tri-consonantal cluster.
/ja]lja-n/ IJa.iJlJa-gali-n/
[ d3a1J1lJ3an] [ d3a1J1lJ3ag alen]
'copulate.with-NF' 'copulate.with-ANTIP-NF'
/jajmba-n/
[d3al.mban]
'find-NF'
/jajmba-gali-jall
[d3a"'mbagaleja1]
'find-ANTIP-PURP'
(2-121) The suffix lbara/ 'denizen' (3.7.1-[1])
(a-1) The resultant stem/word contains no consonant cluster. Stress shift
does not occur.
/gubul /gubu-baral
[k6po] [k6pobara]
'leaf 'leaf-denizen'
Stress 139
/gani/ /gani-bara/
[kani] [kanibara]
'far' 'far-denizen'
(b-2-3) The resultant stem/word contains one hi-consonantal cluster (CC).
The suffix is CVCV Stress shifts to the next vowel.
/jamba/ /jamba-bara/
[jamba] [jambabara]
'camp' 'camp-denizen'
/bomb a/
[b6mba]
'dust'
lbomba-bara/
[bombabara]
'dust-denizen'
(2-122) The locative suffix /Ijga/ (following a vowel; 3.6.3).
(a-2) The resultant stem/word contains one hi-consonantal cluster (CC).
Stress shift does not occur.
/gamu/ /gamu-IJga/
[gamo] [gamoiJga]
'water' 'water-LOC'
(b-2-1) The resultant stem/word contains two hi-consonantal clusters (CC,
CC). The suffix is CCV Stress shift does not occur.
/jamba/ /jamba-I]ga/
[jamba] [jambaiJga]
'camp' 'camp-LOC'
(2-123) The dative suffix /wu! (following a vowel; 3.6.4).
(a-1) The resultant stem/word contains no consonant cluster. Stress shift
does not occur.
/gamu/ /gamu-wu!
[gamo] [gamo:]
'water' 'water-DAT'
(b-1-1) The resultant stem/word contains one hi-consonantal cluster (CC).
The suffix is CV Stress shift does not occur.
/jamba/ /jamba-wu!
[jamba] [jambao]
'camp' 'camp-DAT'
[7] Long vowels
Long vowels may cause stress shift. We shall look at both phonologically long
vowels and phonetically long vowels.
140 Phonology
[7-a] Monosyllabic words/roots
Stress falls on the only syllable, e.g. /bam/ I [ba:n] 'whip', /ja:r/ [ia:r] 'yard' (a loan
from English), and II.Ja:-IJa:l [IJaliJa:] (an onomatopoeic word to describe babies'
cries).
[7-b] Disyllabic or longer words, excluding those which involve reduplication, such
as /I] a:-IJa:l
In the word-initial position, a (phonologically or phonetically) long vowel attracts
stress in all the attested instances. (This is in keeping with the general tendency in
Warrongo for the initial syllable to receive stress; see [2] above.) However, this is
not always the case non-initially.
[7-b-1] The long vowel /a:/ [a:]
Examples involving the word-initial position: II.Jaual [IJfu.ta] '(placenarne)', /ba:baJam/
[ba:bay,arn] 'Mbabaram', and /ja:r-ru [ioi:ra] 'yard-LOC' (There is no example in
which /a:! [a:] occurs non-initially.)
[7-b-2] The phonetically long vowel [i:]
The only example involving the initial position is /jiji/ [1:] 'Yes!' Non-initially [i:]
attracts stress in a few instances (e.g. /guji/ [ko1:] 'hungry', /jali]gijili/ [d.3ab:.Jg1qi]
'Cameron Hill'). But it does not attract stress in other instances, e.g. /majiganl
[mai:gan] 'to cut (NF)', /jarni-ji/ [d.3ami:] 'fat-COM (see (2-74)), and the following
interjections (see (2-50) and (2-51)): II.Jaji/ [IJai:] 'Really?', /waji/ [wai:] 'How is it?',
and /aji/ [ai:] '(exclamation of surprise)'. The conditioning factor on this difference
is not known.
Note that /jaliJgijilil [d.3aliJg1qi] contains a tri-consonantal cluster, and yet stress
falls not on the cluster-initial consonant (see [5] above), but on the long vowel [i:].
This suggests that, in terms of stress assignment, the long vowel [i:] outranks a tri-
consonantal cluster.
[7-b-3] The phonetically long vowel [o:]
Examples involving the initial position: [b6:i] /buwuj/ 'OK', [k6:i] /guwujl 'male
ghost', and [6:i] lwuwuj! 'Hi!' (repeated from (2-73)). Non-initially [o:] does not at-
tract stress in the examples available, e.g. [gamo:] /garnu-wu/ 'water-DAT'
So far we have listed the general tendencies observed in stress assignment.
There are deviations, or what may appear to be deviations, from them. There may
be some explanation for some of them.
Pitch 141
[8] Deviations (1)
Stress falls on the third syllable in certain words, e.g. lbiga.IaiJgaV [biga.rfu]gal]
'fish hawk', lblllldilblllla/ [bondelb6na] '(placename)', /waiJglllldila/ [waiJgondila]
'(placename)'. Etymologically, these words may involve compounding, and this
may possibly be the reason for the occurrence of stress on the third syllable (i.e.
the initial syllable of the putative second root). For example, lbundilbuna/ may
contain the root lbllllal; cf. lblllla-blllla/ [b6nab6na] 'couch grass'. Warrgamay has
/waiJgu/ 'small goanna' (Dixon 1981: 120). /waiJglllldila/ may contain this word.
[9] Deviations (2)
Despite all this, there are instances for which no explanation seems forthcoming,
e.g. IJalpJanba.Ia! [ci3alpd.3anba.ra] 'Cashmere Station', /guriJala/ [gorid3ala]
'eaglehawk', /ganda.ru/ [kandil.J.o] 'imaginary hairy man', and /gangabu/ [kangabo]
'trident fishing spear'; stress falls on the third or second syllable. Compare /ganda.Iu!
[kandil.J.o] with/gandu/ [gando] 'dog'; in the latter, stress falls on the initial syllable
(as is generally the case in Warrongo).
There are a handful of words that do not seem to have stress. They will be dis-
cussed in 2.8, in connection with pitch.
2.8. Pitch
Pitch is not distinctive in Warrongo. Impressionistically speaking, three levels
of pitch can be perceived: high (H), middle (M), and low (L), although it is not
always easy to distinguish the three levels. The vast majority of words start with
Hand end with L. In trisyllabic and longer words, the middle-level pitch (M), too,
may be recognized.
(2-124) lbama!
[bam a]
HL
'man'
/gamu/
[gamo]
HL
'water'
/balanu/
[baJ.ano]
HML
'moon'
/jaiJa-na!
[jfu]ana]
HML
'mother-KIN'
(To be precise, the fall or rise in pitch is not abrupt, but is gradual.)
As mentioned in 2.7, there are a handful of words (and maybe more) that do not
seem to have stress. Impressionistically, they seem to have the pitch contours as
shown in (2-125).
(2-125) (a) /Jaji/
[.J.ai:]
LMM
'young girl'
/Jaji-nba!
[.J.ainba]
LMM
'y.g.-many'
haji-lil
[.J.aili]
LMM
'y.g.-many'
!Jaji-wulpJU!
[.J.ai:olpci3o]
LMMMM
'good-looking girl'
142 Phonology
(b) /buwuj/
[bo:i]
:Mtv.lJ\!1:
/guwujl
[ko:i]
:Mtv.lJ\!1:
'OK, I'll go.' (interjection) 'male ghost'
(c) /IJtma-IJumaj/ !IJUgala-malil
[IJOnaiJomai] [IJokalamale]
LMMMM LMMMM
'that-after' 'outside-side'
(See (2-65) and (2-66) for the analysis of /jif and /j/ in the words in (2-125)-(a).)
As shown above, these words do not seem to have stress (at least, in my observa-
tion). They have the following pitch contour: the pitch height is M and the contour
is flat, except that the words in (a) and (c) start with L. They are exceptions to the
general tendency shown in (2-124). Ladefoged (1975: 222-223) points out a close
relationship between stress and pitch. In view of this, the absence of high pitch and
pitch peak in them (their pitch contour is flat) seems responsible for the absence
of stress in the words in (2-125). (To be precise, their absence seems to make me
perceive no stress in these words.)
(Unlike !Jajil [.J.ai:] 'young girl', /guji/ [koi:] 'hungry' has stress, and the stress
is on the long vowel; see also (2-50) and (2-75). Its pitch contour is LHL or LHH;
both are attested.)
2.9. Connected speech
2.9.1. Intonation
So far it has been impossible to provide a reasonably coherent account of intona-
tion of sentences. This is by far the most difficult aspect of Warrongo phonology/
phonetics. There are many variations, which are no doubt caused by discourse fac-
tors, among others. (Hershberger and Pike (1970) and Patz (2002: 35-41) provide
detailed accounts of sentence intonation of Kuku Yalanji. At least some portions
of their analysis seem to apply to Warrongo as well.) The following account is not
exhaustive. It mainly lists phenomena that have caught my attention.
As samples, Text 1, Lines 3-4 and 7-8 are cited. An impressionistic presenta-
tion of stress (primary stress by acute accent and secondary stress by grave accent)
and intonation (by means of H, M and L, and also an intonational curve) is indi-
cated. The stress is assigned in terms of each line, and not in terms of each word.
The glosses are simplified. In this story, AlfPalmer pretends to be visiting an area
outside Warrongo territory.
(2-126)
(2-127)
Connected speech 143
Text 1, Lines 3 and 4:
A (The local people ask Alf Palmer about a language that he has learned
somewhere else.)

B
/wapjar.JUID.aj jinda IJlllla gugu muga-n/
[wapd3aiJomai inda IJOna koko m6gan]
L M M:M:M MM MM MM H L
where-ABL 2SG.NOM that language catch-NF
'Where did you catch [i.e. learn] that language from?'
(AlfPalmer answers.)
--.......____
/jamu IJaja=guli (pause) muga-gali-n
[cJ3amo IJajagole mogagalen
HL HMLL MLLL
just ISG.NOM=only catch-ANTIP-NF
'I have just learned this [language] by myself.'
'-..._...___----;-
gugu-I]gu/
kokoiJgo]
ML L
language-ERG
Text 1, Lines 7 and 8:
A (The local people ask Alf Palmer about his language.)
-----------
/jinu I]ani gugu/
[ino I]ani koko]
HL LM HH
2SG.GEN what language
'What [is] your language?'
B (AlfPalmer answers.)

/I]aigu wan.ti.]uf
(I]aigo WacOIJO]
:M:MM HML
ISG.GEN language
'My [language is] Warrongo.'
A few aspects of intonation that have been noticed are described below.
Declarative sentences tend to have a sentence-final fall, e.g. (2-126-B) and (2-
127-B).
Special questions (which contain an interrogative word) tend to start with a
high pitch and have a sentence-final fall. Roughly speaking, this applies to (2-126-
A). But this is not always the case; see (2-127-A).
Also, compare (2-128) (which was made up by me; 'TT') and (2-129) (which
was uttered by AlfPalmer; 'AP').
144 Phonology
(2-128) /Ijani jinda IDUJa-llf (TT)
HL MM HL
what-ACC 2SG.ERG eat-NF
'What did you eat'?'
(2-129) /Ijani jinda IDUJa-n/ (AP)
MM MM LH
what-ACC 2SG.ERG eat-NF
'What did you eat?'
(2-128) is an ordinary kind of special question. It starts with H and has a sentence-final
fall. Such special questions seem to be used when, for instance, the speaker does not
know what the addressee ate and he/she requests this information. A similar into-
nation contour is observed in (2-126-A). In this context, probably the speaker did
not know where Alf Palmer had learned that language and wanted to know where.
In contrast, (2-129) starts and continues with M and has a sentence-final rise.
It was uttered by Alf Palmer when he knew what the addressee had eaten, and he
was possibly teasing the addressee. Special questions with such an intonation con-
tour do not seem to really request information.
General (or polar) questions may start with the question marker /waji/ (men-
tioned in (2-50)); see 4.2.1.2. They tend to start with a high pitch and have a sen-
tence-final fall.
(2-130) (AlfPalrner asked me as follows.)

/waji jinda Jlaga-n/ (AP)
[wai: inda Jlagan]
HL MM H L
Q 2SG.NOM see-NF
'Did you see [him]?'
Intonation contour may be involved in formation of subordination (4.18). An
example cited from a text. In this story, a man says to his travel companion as fol-
lows. It seems that he considered it not safe to travel during the daytime, probably
for fear of some other tribe's attack.

(2-131) /gari galijga-n IJali-0 jani-jaV (AP)
[kari kalijga-n IJale jani-jal]
(rise in pitch) (fall in pitch)
sun fall-NF lDU-NOM go-PURP
'When the sun has set, we will go.'
Connected speech 145
/gali]ga-rJ! has a rise in pitch, and (2-131) has the meaning of 'when'. (Another ex-
ample is (3-319). Additional examples are given in 4.18.6.) If /gab:,)ga-n/ has a fall
in pitch, the sentence will probably have a different meaning: 'and' or the like.

(2-132) /gari gali]ga-n IJali-0
(fall in pitch)
sun fall-NF lDU-NOM
'The sun has set, and we will go.'
jani-jall (TT)
(fall in pitch)
go-PURP
Intonation contour for a topic NP is discussed in 4.22.
2.9.2. Other phenomena
'Casual speech' (Labov 1972: 7f}), fast speech and the like exhibit a few differences
from the pronunciation of 'citation forms' (Lass 1991: 295). Selected examples are
g1ven.
[1] Vowel lengthening
(a) The negation word /pawa/ is generally pronounced [pawa] (HL or maybe HM).
AlfPalmer sometimes said [pawa:] (with a long vowel; MHMM). This is probably
in order to emphasize negation.
(b) The interjection /jagaj/ 'Ouch!' is often pronounced [jaka:i] (with a long vowel).
(c) In the texts, the verb /janil yani-0 'go-NF' is sometimes pronounced [jani:]
(with a long vowel). This is probably in order to emphasize the length and continu-
ation of the walk.
(2-133) !IJali gulbila jani/
[IJale kolbila jani:]
lDU-NOM to.south go-NF
'We kept going/walking to the south.'
(d) In a text, a man fancies a woman and says to her as follows. Here the length-
ening of the vowel probably indicates joy.
(2-134) /IJajgu bi.Igo/
[IJaigo bi{g6:]
lSG.GEN wife
'[You are] my wife!'
(e) In a text, a man found a woman, and (with joy?) he called out, [wa:n:Jo]
'Woman!' (lwarqu/ 'woman').
146 Phonology
(f) The interjection that is tentatively phonemicized /guwuj/ 'Hi! Hey!' (a call to
someone in the distance) almost always involves a fairly long vowel (impressionis-
tically the length of two, three or even four syllables): [ko::::i]. The stress is on the
very end of the long vowel [o::::], immediately preceding [i]. See 2.13.1 for more on
this word.
The vowel that is lengthened is the second vowel in (a) to (d), but it is the first
vowel in (e) and (f). The cause for this difference is not known.
There appears to be no example in which a consonant is lengthened.
[2] Addition of [i] or [e]
This is observed in texts. When a person calls out to someone in the distance,
sometimes a high front vowel (generally [i], and sometimes [e]) is added to the
end of the noun that refers to the addressee. This vowel is non-phonemic. The last
vowel of the noun is somewhat long (e.g. [jar.Ja:i] in (2-135)), though not always (e.g.
[kab.]ai] in (2-136)). Examples:
(2-135) /jar.Ja IJaja jaru/
[jar.Ja:i IJaja jaco]
MHHM MM HL
mother 1SG.NOM here
'Mother, I [am] here!'
(2-136) /jaru IJaja gab.]a/
[jaco IJaja kab.]ai]
HM MM HL
here 1SG.NOM J\.18
'I [am] here, Uncle (MB)!'
A similar phenomenon, in which a non-phonemic [i] is added to the end of a
word, occurs in Japanese. (In Wanyjirra ([waJlra]) (Northern Territory, imme-
diately east of Djaru), Maggie Scott sometimes adds the nasal [IJ] to the end of the
name of the language, i.e. [waJllf3icaiJ]. This [IJ] is not phonemic.)
[3] Addition of [o]
[ o] is added to the end of a word in many examples, but its function is not known.
(2-137) /Jlula guji jamba-IJga
[)lola koi:o j ambar.Jg a
3SG.NOM hungry camp-LOC
'She is lying hungry in the camp.'
wuna-n/
wonan]
lie-NF
Careful speech 147
[o] is phonetically identical with the allomorph of the dative suffix for vowel-final
sterns (see (2-61) and [4]-(b) below). But it is unlikely to be the dative suffix in
(2-137). [o] occurs in a song, too. See (1-10-d).
[4] Phoneme deletion
Alf Palmer spoke fairly fast, and in texts, sometimes the end of words was blurred.
Also, in fast and/or casual speech, certain phonern es were dropped.
(a) Dropping of /w/. Compare [cf3awa] !Jawa/ 'mouth' and [d3a:]. /w/ seems to be
dropped between /a/'s only, and in no other environment Dropping of /j/ has not
been observed.
(b) Dropping of /g/. The dative suffix is generally /gu/ [go] following a consonant,
and lwul [o] following a vowel (see Table 3-2). However, there is just one example
in which the dative suffix following a consonant is [o], and not [go]. Compare
[winga.rno] /winga.xgu/ 'fish-DAT' and [winga.J..o]. (In [winga.J..o] a dot is written
between [.J] and [o] in order to indicate that these two phones are pronounced
separately, and not jointly as [.J.o].) /g/ of /gu/ is dropped here. Here, the !.I/-final
stern behaves rather like a vowel-final stern. This is probably because, as seen in
2.5-[4], Ill is a non-syllabic 'vowel'. It is highly sonorous; see also (2-93). (In Djaru
(Tsunoda 1981a: 36) /c/-final, /11-final and /nl-final sterns sometimes behave like
vowel-final sterns, although they behave as consonant-final sterns elsewhere in the
language. These phonemes are all sonorous.)
(c) Dropping of /n-gu/ 'LINK-DAT' See 3.6.5-[5] and 4.9.9.4-[1].
(d) Possibly, omission of the transitive-stem-forming suffix /rja/ and the intransi-
tive-stem-forming suffix fbi/. See 3.1.2-[7].
2.10. Careful speech
In 2.9.2, we looked at casual speech, fast speech and the like. 'Careful speech'
(Labov 1972:79, Lass 1991: 295) by Alf Palmer produced two interesting phe-
nomena.
2.10.1. Deletion of a phoneme
Upon my request, Alf Palmer pronounced the word for 'grey hair' (lbi.u.Jg a/)
slowly and said [b4. (pause) ga]. The nasallrj/ was deleted and the consonant
cluster was reduced. Dixon (1972: 285, 1981: 21-22) reports a number of similar
instances from Dyirbal and Warrgarnay, respectively. In each of them a nasal as the
second member of a tri-consonantal cluster was deleted in careful pronunciation.
It is irn portant to note the following.
148 Phonology
(a) It is only in this 'careful (and deliberate) speech' that the nasal/rj/ of lbi.u]gal
was deleted. It was never deleted in 'non-careful speech'. /bi.nma/ and /bi.rga/ are
not in free variation. It is certain that /Jijg/ and /Jg/ are in contrast, for there is a
near-minimal pair: /bi.u]ga/ 'grey hair' and lb:i.Igu/ 'wife'. If lrj/ of lbi.u]ga/ is deleted
in 'non-careful speech' (i.e. lbirga/), most probably this will not be recognized as
the word for 'grey hair'
(b) Deletion of certain phonemes in fast and/or casual speech is common in
Warrongo (2.9.2-[3]) and other languages of the world. Phenomena such as cluster
reduction and nasal deletion are reported, e.g. Lass (1991: 298, 302) and Zwicky
(19TI: 279, 282, 291). In contrast, such reduction and deletion are unlikely to occur
in careful speech (see Labov 19TI: 70-109). What is striking about [bi.J. (pause)
ga] and similar instances reported by Dixon, is that it is in careful speech, and not
in fast or casual speech, that nasal deletion and cluster reduction occurred.
(c) The deletion of /rj/ from lbinma/ 'grey hair' (which causes cluster reduction) is
the reverse of the insertion of /rj/ (2.6-[2]-(c)). The latter produces tri-consonantal
clusters, e.g. lbaiJgaj-1]-gu/ 'spear-I]-DAT' Note, however, that the insertion of /rj/
occurs at morpheme boundaries, whereas the deletion of /rj/ from /biri]ga/ 'grey
hair' occurred within a root.
In terms of phonological analysis, the deletion of lrj/ from /bi.u]ga/ 'grey hair'
has the following two implications.
(d) Phonemic analysis is generally based on 'very careful speech' (Lass 1991: 30,
295). However, the deletion of /rj/ from /bi.u]ga/ and the instances reported by
Dixon, suggest that this may not be a good method.
(e) The deletion of lrj/ from /biri]ga/ suggests that /rj/ in /biri]ga/ has a 'lower' status
as a phoneme than the other phonemes in the word (lb, i, J, g, a/). This applies to
the instances reported by Dixon. Also, it may apply to all of the tri-consonantal
clusters (Tables 2-18, 2-19), in which the second member is a nasal. It should be
added here that elsewhere /1]1 has a full status as a phoneme, for example, root-
initially and intervocalically. Thus, compare /I]arna/ 'handle of shield' with /gam a/
'dance style', and /mUI]an/ 'mountain' with /mug an/ 'to get (NF)'. The above sug-
gests that phonemes can be classified as follows. (i) Those which always have a
full status as a phoneme, e.g. probably /b, i, J, g, a/, and (ii) those which may have a
lower status as a phoneme in certain environments, e.g. /rj
Most of the Australian languages have only hi-consonantal clusters, e.g. Djaru
(Tsunoda 198la: 38), but a number of Australian languages possess tri-consonantal
clusters as well, e.g. Warrongo. (See Dixon 2002: 656.) It is possible, though by no
means certain, that language such as Warrongo used to have hi-consonantal clusters
only, and that subsequently a nasal (often homorganic with the following stop) was
inserted. (See Dixon (1980: 166, 219, 2002: 657-658) and O'Grady and Fitzgerald
Careful speech 149
(1995).) Warrongo has at least one suggestive pair: /galga/ Vt 'put down' (L-class)
and /gab;J9a/ Vi 'fall down' (Y-class). (See Table 3.9.4.1 for the classes of verbs.)
2.10.2. Isolation of suffixes
People who are not trained in linguistics may recognize words, but they will be
unlikely to recognize affixes. Alf Palmer was an exception. One day I was dis-
cussing reciprocal forms of verbs with him. I cited forms such as !Jajmba-wa-n/
[d3alm.bawan] 'find-RECP-NF', lrn.uga-wa-n/ [m6gawan] 'hold-RECP-NF',
/jil]ga-wa-n/ [cJ9IJgawan] 'punch-RECP-NF', and IJUllda-wa-n/ [d36ndawan]
'kiss-RECP-NF' Upon hearing these, AlfPalmer said:
(2-138) All [ban].
There are striking facts about this.
First, Alf Palmer recognized and isolated a suffix, to be more precise, a se-
quence of two suffixes: /wa/'reciprocal' and /n/'nonfuture'.
Second, AlfPalmer said [ban], not [wan]. This is truly intriguing for a number
of reasons. (i) He always said [wan] (and never [ban]) when he pronounced the
entire reciprocal forms; see the forms cited above. (ii) [wa] is possible root-initially
and word-initially (Table 2-17), e.g. !wanda/ 'news' Consequently, Alf Palmer
could have easily pronounced [wan]. (iii) Elsewhere in the language, /w/ and /b/
contrast, e.g. !wama/ 'mad' and /hamal 'man'. (iv) Elsewhere in the language,
there is no alternation between /w/ and /b/ (although there are other alternations;
see 2.4). (v) The cluster /lw/ is allowed- even within a root ((d) of Table 2-18), e.g.
/walwa/ 'bad'. Therefore, AlfPalmer could have easily pronounced 'All [wan]'
It is worth noting the following points.
(a) Like lw/, /bl occurs suffix-initially (Table 2-17), e.g. /baral 'denizen' (3.7.1-[1]).
(b) This [ban] occurred when Alf Palmer's attention was focused on the element
in question. That is, it occurred in 'careful speech'. Furthermore, it occurred, so to
speak, at a 'meta-language' level.
(c) This [ban] does not contrast with [wan] of /muga-wa-n/ [mogawan] 'hold-
RECP-NF', etc. Nor is it in free variation with this [wan]. The choice is not op-
tional. [ban] was observed in careful speech only.
In terms of phonological analysis, it does not seem adequate to assign this [b]
of [ban] to the phoneme /b/, for the circumstance under which it occurred is un-
usual; see (b) above. Nor does it seem appropriate to allocate it to the phoneme lw/.
Also, it is not suitable to set up an arc hi phoneme /B/ (or /WI) for this [b] and [ w] of
IJajmba-wa-n/ 'find-RECP-NF', etc., for this [b] does not manifest neutralization
150 Phonology
in the usual sense. Then, it may be useful to recognize a level of phonological
analysis as distinct from that of phonemes and that of archiphonemes. Thus, this
[b] may be presented by ///b//1, using a triple slash. (A single slash is used for the
phonemic level. Double slashes are used by some authors for the morphophonemic
level; see Lass 1991: 58).
The suffi.x /ba/ or its variant occurs as a reciprocal marker in neighbouring
languages such as Warrgamay (lba!, Dixon 1981: 49, 74-75), Nyawaygi (/ba/, etc.,
Dixon 1983: 489), and Dyirbal (/(n)bari/, Dixon 1972: 92). Like Warrongo, Gugu-
Badhun has /wa/ 'reciprocal' (Sutton 1973: 144-145). It seems certain that this /wa/
is a reflex of the reciprocal /ba/ Then, it looks as if a diachronic change was real-
ized synchronically in AlfPalmer's idiolect, although he was no doubt unaware of
this diachronic change.
2.11. Women's pronunciation (?)
Alf Palmer often included conversations in texts. When he was assuming the role
of a woman in a conversation, he sometimes changed his pronunciation, in which
pitch seems to be somewhat higher than usual. There is at least one example that
involves non-voicing of a stop in addition to high pitch.
The pronoun /IJajgUJ '1SG.GEN' is generally pronounced by Alf Palmer as
[IJaigo] both in texts and elsewhere. This pronunciation involves the voiced [g].
It does not have high pitch. Examples are in (2-127-B) and (2-134). In contrast,
when he was assuming the role of a woman in a conversation, he often (though
not always?) pronounced it as [IJaiko]. This involves the voiceless [k]. An example
is cited from a text. In this story, a group of men are dancing, and the women are
choosing their husband from among them. One woman says as follows. Here, /Ijajgu/
starts with high pitch.
(2-139) IIJUlla
[IJona
MM
that
IJajgu/
IJaiko]
HML
1SG.GEN
'That [man is my] husband.'
It may (or may not be) significant that this example involves /gu/ [ko]. Recall
that, among the four stop phonemes, /g/ is the most likely to be voiceless (see (2-
4)), and that, among the three vowel phonemes, lui is the most likely to make the
preceding stop voiceless (2.1.3.2.5). That is, this voiceless allophone [k] occurs in
an environment where the voiceless allophone is favoured (2.1.3.2.7).
The use of a voiceless allophone and high pitch in 'woman's speech' does not
seem systematic. Nor is it known whether it was confined to Alf Palmer's texts
or it was common in women's speech when the language was actively spoken. No
Ideophones 151
such phenomenon is observed among the fluent speakers ofDjaru (Tsunoda 1981a:
24-25) and Wanyjirra. In view of this, the phenomenon in question may possibly
be a product of Alf Palmer's ingenuity, and it may not reflect the situation of the
time when the language was thriving.
2.12. Ideophones
There are probably many ideophones. (See Dixon (1972: 328-329, 1977a) on
Dyirbal ideophones.) Examples follow.
The word for 'crow' is /waJal [wat:fa] (HL) or lwaJagan/ [wat:fagan] (HML).
The call of a crow is described by [wa1fa:] (no stress, MMM). The second vowel
is long. (The meaning of /gan/ is not known. It does not seem to be the feminine
suffi..x /gan/ (3.7.1-[17]).) Alf Palmer stated that, in a crow corroboree, people
danced about., saying [ wat:fa:] or [ wat:fa:wat:fa:]. (The word for 'crow' starting with
/wa/ is common across the continent. Djaru has separate words for the name of a
crow (/wangoJal [wango.J.a]) and the word to describe its call ( [wa:kwa:k]).)
The call of brolga (a bird sp.) is [ko.J.or], and its name, too, is /gwur/ [ko{or].
(The word for 'brolga' starting with /gul seems common across the continent.
Djaru has /gurgali/ [korkale] 'brolga'.)
According to Alf Palmer, a bird species with the name /guguwtg1/ [k6koop]
has the call [k6koop]. This word involves two occurrences of [o], and not a long
[o:]. See (2-77). (/guguwup/ means 'brown pigeon' in Jirrbal and Girramay (Dixon
1972: 403), and 'blue pigeon' in Warrgamay (Dixon 1981: 117).)
/dii.)gil-dii]gill [dii.)gilc:fu]gil] 'bell' (mentioned in 2.2.6.2-(b)). This word contains
the only instance of /ld/ cluster in Warrongo. This word contains two occurrences
of [il] (not [ell). They are exceptions to the lowering effect of Ill (2.1.4.3-[2]-(a)).
This my be because this word is an ideophone.
The word /jigir-Jigir/ [d3igi..r<J9gir] 'willy wagtail (bird sp.)' appears to describe
the way the bird wags its tail vertically and repeatedly. This word contains two
occurrences of [ir] (not [er]). They are exceptions to the lowering effect of /r/
(2.1.4.3-[2]-(b)). (The word !Jigir-Jigir/ is also a placename; see Text 1, Line 20.)
Words for 'willy wagtail' seem similar across the continent., e.g. IJirbiji-Jirbi/ of
Djapu of Northern Territory (Morphy 1983: 163), !Jigili-Jig/ of Djaru (Kimberley
Language Resource Centre 1992: 50), and !Jindir-Jindir(ba)/ of Pitjantjatjara/
Yankuntjatjara of central Australia (Goddard 1992: 151). Woiwurrung of
Melbourne (Blake 1991: 87) has qjirri-qjirri (probably !Jiri-Jiri!- TT). Note that
all of these words start with !Ji/, that they involve reduplication (two of them in-
volve partial reduplication, e.g. IJigili-Jig/), that they involve a rhotic (/r/ or Ill),
and that the vowels are /il, except for /a/ of lba/ in !JindirJindir(ba)/.
Note that /dii.Jgil-dii.Jgill and !Jigir-Jigir/ contain exceptions to the respective
lowering effects. This suggests that ideophones might be exempt from the lowering
effect.
152 Phonology
Other onomatopoeic words (a type of ideophones) include !IJa:IJa:/ [IJa:IJa:] for
babies' cries (2.2.3.1-[1]) and /gigigil [kikiki] for possums' squeaks'
2.13. Adaptation offoreign words
2.13.1. Words from other Australian languages
'When the First Fleet arrived at Sydney Cove in 1788 they soon adopted words
from Dharuk, the local language' (Dixon et al. 1990: 1), and these words are now
widely distributed over the continent. (The name of the language is also spelt
'Dharug' (Troy 1994: 5).)
At least two words from Dharuk occur in Warrongo: (i) /guwuj/ [ko::::i] 'Hi!
Hey!' (Dixon et al. 1990: 208) (see 2.9.2-[1]-(e)), and (ii) /gaJlal 'hut' (/ganyi/
'hut' in Dharuk (Dixon et al. 1990: 199); it is spelt 'gunya' by Troy (1994: 44)).
Nowadays [ko::::i] is by white people, too, and it is even listed in a dictionary (The
Australian Pocket Oiford Dictionary 1976: 178). The spelling is 'Cooee' or 'Cooey'
/jacamanl'horse' (e.g. (4-735)) has a wide geographical distribution, and it occurs
in Warrongo, too. According to Dixon et al. (1990: 86-87), its source is unknown.
Alf Palmer said that the 'proper' Warrongo word for 'horse' is not /jacamanl, but
that it is !Jaja.Ii!, e.g. (3-160). The latter's source is not known, either.
How AlfPalmer pronounced the names of the neighbouring languages was dis-
cussed in 1.8.4.1-[2] and 2.6-[4].
There may be many words from other Australian languages which I have been
unable to identify as non-Warrongo. For example, /gambi/ 'clothes' may come
from the Townsville region. It was first recorded at Mt. Black, west of Townsville
(with the meaning of 'possum cloak' (1.9)).
All of the words cited above (and possibly those words that I have been unable
to identify as non-Warrongo) have been well adapted to the Warrongo phonology.
Indeed, /gaJla! 'hut' was used in 2.1.2-[1]-(c) in a presentation of phonemic opposi-
tions in Warrongo.
There is one word that sounded non-Warrongo to my ears: /w3I]glllldilal [W3IJ-
gondila] 'a place on Herbert River, where there is a big cliff (see 2.2.6.1). This
word contains [il] (not [el]), and it is an exception to the lowering effect of /11 (see
2.1.4.3-[2]).
2.13.2. Words from English
Words from English found in Warrongo are of two types: (i) those which have a
wide geographical distribution, occurring in other languages as well, and (ii) those
which do not (or, do not seem to). Native Warrongo words appear to prohibit stops
root-finally and word-finally (Table 2-17). In the followin